Cahiertechnique

Technical series ..........................................................................
Cahier technique no 0
List of Cahiers Techniques in English
Cahiers Techniques are a collection of documents intended for people in the industry who are looking for information in greater depth, in order to complement that given in display product catalogues. Cahiers Techniques go beyond this stage and allow executives and technicians to know about news techniques in electrical and electronical industries. Each Cahier Technique provides an in-depth study of a precise subject in the fields of electrical networks, protection devices, monitoring and control and industrial automation systems. Some of them were issued some years ago, and their look is oldfashioned, but they keep, in some measure, a real technical interest. The latest publications can be downloaded on Internet from the Schneider server. Code: http://www.schneider-electric.com Section: mastering electrical power Please contact your Schneider representative if you want either a Cahier Technique or the list of available titles. With certain exceptions, Cahiers Techniques are issued in French and English. Some countries, such as Spain, translate CT in their own language. The Cahiers Techniques collection is part of the Groupe Schneiders Collection Technique .
no 0
List of Cahiers Techniques in English
Contents
1 Electrical knowhow 2 Networks 3 Electrical and environmental constraints 4 Dependability: availability, safety 5 Protection devices and monitoring 6 Breaking techniques and switchgear pp. 3 pp. 4 pp. 5 pp. 6 pp. 7 pp. 8
ECT 0 up dated May 1998

Cahier Technique Schneider no 0 / p.1
1 Electrical Knowhow
1.1 General knowledge

c Fault arcs on busbar sets and switchboards. ECT38, G. Bouvier, A. Ducluzaux, 8 pp., 05/83. c Extra losses caused in high current conductors by skin and proximity effects. ECT83, A. Ducluzaux, 12 pp., 12/82. c EMC: Electromagnetic compatibility. ECT149, F. Vaillant, 28 pp., 10/92. c Ferroresonance ECT190, Ph. Ferracci, 28 pp., 03/98.
1.2 Low Voltage 1 kV

c Calculation of short-circuit currents. ECT158, B. de Metz Noblat, G. Thomasset, 24 pp., 03/95. c Electrodynamic forces on busbars in LV systems. ECT162, Ch. Kilindjian, J.P. Thierry, 20 pp., 10/96.
1.3 High Voltage > 1 kV

c Overvoltages and insulation coordination in MV and HV. ECT151, D. Fulchiron, 24 pp., 02/95. c Lightning and H.V. electrical installations. ECT168, B. De Metz Noblat, 24 pp., 06/94. c Dynamic stability of industrial electrical networks. ECT185, B. de Metz Noblat, G. Jeanjean, 24 pp., 06/97.
2 Networks

c Harmonics in industrial networks. ECT152, P. Roccia, N. Quillon, 24 pp., 10/94. c Calculation of short-circuit currents. ECT158, B. de Metz Noblat, G. Thomasset, 24 pp., 03/95. c Automatic transfering of power supplies in HV and LV networks. ECT161, G. Thomasset, 16 pp., 03/96. c Electrical installation dependability studies ECT184, S. Logiaco, 24 pp., 06/97. c Dynamic stability of industrial electrical networks. ECT185, B. de Metz Noblat, G. Jeanjean, 24 pp., 06/97.

c Inverters and harmonics (case studies of non-linear loads). ECT159, J.N. Fiorina, 20 pp., 09/93. c Harmonics upstream of rectifiers in UPS. ECT160, J.N. Fiorina, 20 pp., 12/93. c Earthing systems in LV. ECT172, R. Calvas, B. Lacroix, 24 pp., 09/95. c Earthing systems worldwide and evolutions. ECT173, R. Calvas, B. Lacroix, 24 pp., 09/95. c Active harmonic conditioners and unity power factor rectifiers. ECT183, E. Bettega, J.N. Fiorina, 28 pp., 01/97.

c Neutral earthing in an industrial HV network. ECT62, F. Sautriau, 12 pp., 04/96. c Process generating plants in industrial systems. ECT99, P. Bibollet, 8 pp., 04/80. c Overvoltages and insulation coordination in MV and HV. ECT151, D. Fulchiron, 24 pp., 02/95. c MV public distribution networks throughout the world. ECT155, Ch. Puret, 28 pp., 03/92. c HV industrial network design. ECT169, G. Thomasset, 24 pp., 10/94.
3 Electrical and environmental constraints

c EMC: Electromagnetic compatibility. ECT149, F. Vaillant, 28 pp., 10/92. c Harmonics in industrial networks. ECT152, P. Roccia, N. Quillon, 24 pp., 10/94. c Enclosures and degrees of protection. ECT166, J. Pasteau, 12 pp., 01/93. c Earthquakes and electrical equipment. ECT180, E. Melmoux, 26 pp., 04/97. c Ferroresonance ECT190, Ph. Ferracci, 28 pp., 03/98.

c Thermal study of LV electric switchboards ECT145, C. Kilindjian, 24 pp., 12/97. c Inverters and harmonics (case studies of non-linear loads). ECT159, J.N. Fiorina, 19 pp., 09/93. c Harmonics upstream of rectifiers in UPS. ECT160, J.N. Fiorina, 20 pp., 12/93. c Active harmonic conditioners and unity power factor rectifiers. ECT183, E. Bettega, J.N. Fiorina, 28 pp., 01/97. c Cohabitation of high and low currents ECT187, R. Calvas, J. Delaballe, 24 pp., 08/97.

c Overvoltages and insulation coordination in MV and HV. ECT151, D. Fulchiron, 24 pp., 02/95. c Lightning and HV electrical installations. ECT168, B. De Metz Noblat, 24 pp., 06/94.
4 Dependability: availability, safety

c Introduction to dependability design. ECT144, P. Bonnefoi, 20 pp., 12/90. c Automatic transfering of power supplies in HV and LV networks. ECT161, G. Thomasset, 16 pp., 03/96. c Electrical installation dependability studies. ECT184, S. logiaco, 24 pp., 09/97.

c High availability electrical power distribution. ECT148, G. Gatine, 20 pp., 10/91. c Dependability and L.V. switchboards. ECT156, O. Bouju, 24 pp., 06/97. c Energy-based discrimination for LV protective devices. ECT167, R. Morel, M. Serpinet, 16 pp., 03/98. c Earthing systems in LV. ECT172, R. Calvas, B. Lacroix, 24 pp., 09/95. c Earthing systems worldwide and evolutions. ECT173, R. Calvas, B. Lacroix, 24 pp., 09/95.

c Protection of electrical distribution networks by the logic selectivity system. ECT2, R. Calvas, F. Sautriau, 8 pp., 02/83. c Neutral earthing in an industrial HV network. ECT62, F. Sautriau, 11 pp., 04/96. c Dependability of MV and HV protective devices. ECT175, M. Lemaire, 16 pp., 08/95.
5 Protection devices and monitoring

c Enclosures and degrees of protection. ECT166, J. Pasteau, 12 pp., 01/93.

c Residual current devices. ECT114, R. Calvas, 24 pp., 11/94. c Energy-based discrimination for LV protective devices. ECT167, R. Morel, M. Serpinet, 16 pp., 03/98. c L.V. circuit breakers confronted with harmonic transient and cyclic currents. ECT182, M. Collombet, B. Lacroix, 16 pp., 04/97. c Intelligent LV switchboards. ECT186, A. Jammes, 28 pp., 06/97.

c Protection of electrical distribution networks by the logic selectivity system. ECT2, R. Calvas, F. Sautriau, 8 pp., 02/83. c Switching MV capacitor banks. ECT142, D. Koch, 16 pp., 09/92. c Behaviour of the SF6 MV circuit-breakers Fluarc for switching motor starting currents. ECT143, J. Hennebert, 20 pp., 07/92. c Control, monitoring and protection of HV motors. ECT165, J.Y. Blanc, 28 pp., 05/95. c Protection of industrial and tertiary MV networks. ECT174, A. Sastr, 28 pp., 06/96. c Directional protection equipment. ECT181, P. Bertrand, 20 pp., 03/98.
6 Breaking techniques and switchgear

c SF6, the dielectric gas. ECT78, M. Dubois, 12 pp., 01/83. c From current transformer to hybrid sensors, in HV. ECT170, Ch. Teyssandier, 24 pp., 03/95. c SF6 properties and use in MV and HV switchgear ECT188, D. Koch, 24 pp., 05/97.

c Development of LV circuit-breakers to standard IEC 947-2. ECT150, E. Blanc, 20 pp., 02/98. c LV circuit-breaker breaking techniques. ECT154, R. Morel, 20 pp., 12/94. c LV breaking by current limitation. ECT163, P. Schueller, 8 pp., 11/94.

c The behaviour of SF6 puffer circuitbreakers under exceptionally severe conditions. ECT101, J.C. Henry, G. Perrissin, C. Rollier, 9 pp., 01/83. c The breaking process with a Fluarc SF6 puffer-type circuit breaker. ECT112, J. Hennebert, 12 pp., 03/83. c Design and utilization of M.V. limiting fuses. ECT128, O. Bouilliez, 20 pp., 02/85. c Switching MV capacitors banks. ECT142, D. Koch, 16 pp., 09/92. c Behaviour of the SF6 MV circuit-breakers Fluarc for switching motor starting currents. ECT143, J. Hennebert, 20 pp., 09/92. c Current transformer for HV protection. ECT164, M. Orlhac, 24 pp., 04/95. c Breaking by auto-expansion. ECT171, G. Bernard, 16 pp., 03/95. c Switching and protection of MV capacitor banks. ECT189, D. Koch, 28 pp., 07/98.
Schneider throughout the world
Europe
Austria - Wien: Fax: (43)1 6105454 Belgium - Brussels: Fax: (32)2 375 3858 Bulgaria - Sofia: Fax: (359)297 12572 CIS - Moscow: Fax: (7)502 224 5050 Denmark - Ballerup: Fax: (45)44 68 5255 Finland - Espoo: Fax: (358)95 27 00375
France - Rueil Malmaison: Fax: +33 (0)1 41 29 81 95 Germany - Ratingen: Fax: (49)2102 404256 Greece - Kifissia: Fax: (30)16 29 52 10 Ireland Celbridge - Co. Kildare: Fax: (353)1 627 0859 Italy - Agrate Brianza: Fax: (39)39 609 1510 Norway - Moss: Fax: (47)6925 7871 Netherlands - Haarlem: Fax: (31)23 512 410 Poland - Warszawa: Fax: (48)2 625 1483
Portugal - Linda A Velha: Fax: (351)1 416 5857 Romania - Bucuresti: Tel: (40)1 212 1074 Slovenia - Ljubljana: Fax: (383)61 168 5426 Spain - Barcelona: Fax: (34)3 484 3308 Sweden - Sollentuna: Fax: (46) 8 623 84 80 Switzerland - Ittigen: Fax: (41)31 917 3355 United Kingdom London: Fax: (44)171 747 22 51
Asia
China - Beijing: Fax: (86) 10 65 00 50 13 Hong Kong: Fax: (852) 28 11 10 29 Indonesia - Jakarta: Fax: +62 21 7504415 India - New Dehli: Fax: (91) 116 46 50 80 Malaysia Selangor Darul Ehsan: Fax: (60) 37 05 11 70 Philippines Metro-Manila: Fax: (63) 28 16 00 63 Japan - Tokyo: Fax: (81) 354 74 44 70 Singapore - Singapore: Fax: (65) 484 78 00 South Korea - Seoul: Fax: (82) 26 30 98 00 Taiwan Roc -Taipei: Fax: (886) 27 33 64 10 Thailand - Bangkok: Fax: (66) 22 60 82/83 Vietnam Ho Chi Minh City: Fax: (84) 88 29 60 67
North America
Canada - Toronto: Fax: (1)416 752 4203 Mexico - Mxico: Fax: (52) 56 86 24 09 United States -Palatine: Fax: (1)847 925 7500
France - Grenoble: Fax: +33 (0)4 76 90 49 64
South America
Argentina Caseros (provincia B. Aires): Fax: (54) 17 16 88 77 Brazil - Sao Paulo: Fax: (55) 115 22 51 34 Chili - Santiago: Fax: (56) 26 23 59 76 Colombia - Bogota: Fax: (58) 14 13 90 12 Cuba - La Havane: Fax: (53) 733 12 17 Dominican Republic Santo Domingo: Fax: (1809) 562 82 53 Ecuador - Quito: Fax: (593) 243 49 40 Peru - Lima: Fax: (511) 221 81 84 Venezuela - Caracas: Fax: (58) 22 42 43 30 Uruguay - Montevideo: Fax: (598) 260 79 54
Africa
Algeria - Algiers: Fax: (213) 269 80 02 Egypt - Cairo: Fax: (20) 24 01 66 87 Ivory Coast - Abidjan: Fax: (225) 25 69 86 Morocco - Casablanca: Fax: (212) 299 08 67 to 69 Nigeria - Lagos: Fax: (234) 14 97 45 99 Tunisia - Ariana: Fax: (216) 175 03 42 South Africa Bergviel-Sandton: Fax: (27) 118 04 36 00 Zambia - Kitwe: Fax: (260) 222 83 89 Zimbabwe - Harare: Fax: (263) 475 45 55
Middle East
Iran - Teheran: Fax: (98) 218 71 81 87 Isral - Herzlia Pituach : Fax: (972) 956 57 15 Kuwait - Safat: Fax: (965) 240 75 06 Qatar - Doha: Fax: (974) 32 28 61 Saudi Arabia - Riyadh: Fax: (966) 14 98 38 27 Turkey Gtztepe-Istanbul: Fax: (90) 21 63 86 38 75 United Arab Emirates Dubai: Fax: (971) 431 36 98
Pacific
Australia - Baulkham Hill: Fax : (61 2) 9629 8555 New Zealand - Auckland: Fax: (64) 98 28 22 25
Schneider
Direction Scientifique et Technique, Service Communication Technique F-38050 Grenoble cedex 9 Tlcopie: (33) 04 76 57 98 60
Edition: Schneider - Grenoble Printing: Clerc Fontaine - France - 1000.
63092
05-98
1998 Schneider
Collection T echnique ..........................................................................
Cahier technique no. 114

Residual current devices in LV
R. Calvas
Cahiers Techniques is a collection of documents intended for engineers and technicians, people in the industry who are looking for more in-depth information in order to complement that given in product catalogues. Furthermore, these Cahiers Techniques are often considered as helpful tools for training courses. They provide knowledge on new technical and technological developments in the electrotechnical field and electronics. They also provide better understanding of various phenomena observed in electrical installations, systems and equipments. Each Cahier Technique provides an in-depth study of a precise subject in the fields of electrical networks, protection devices, monitoring and control and industrial automation systems. The latest publications can be downloaded from the Schneider Electric internet web site. Code: http://www.schneider-electric.com Section: Experts' place Please contact your Schneider Electric representative if you want either a Cahier Technique or the list of available titles. The Cahiers Techniques collection is part of the Schneider Electrics Collection technique.
Foreword The author disclaims all responsibility subsequent to incorrect use of information or diagrams reproduced in this document, and cannot be held responsible for any errors or oversights, or for the consequences of using information and diagrams contained in this document. Reproduction of all or part of a Cahier Technique is authorised with the prior consent of the Scientific and Technical Division. The statement Extracted from Schneider Electric Cahier Technique no. ..... (please specify) is compulsory.
no. 114
Roland CALVAS With an engineering degree from Ecole Nationale Suprieure dElectronique et de Radiolectricit de Grenoble (1964) and a Business Administration Institute diploma, he joined Merlin Gerin in 1966. In the course of his career, he has held the position of sales manager, followed by marketing manager for the activity dealing with the protection of people against electrical hazards. He is currently charged with technical communication within Schneider Electric.
ECT 114 updated, February 1999
Lexicon
Cardiac fibrillation: A malfunctioning of the heart corresponding to loss of synchronism of the activity of its walls (diastole and systole). The flow of AC current through the body may be responsible for this due to the periodic excitation that it generates. The ultimate consequence is stoppage of blood flow. Direct contact: Contact of a person with the live parts of electrical devices (normally energised parts and conductors). Earthing system: Standard IEC 60364 stipulates three main official earthing systems which define the possible connections of the neutral of the source and frames to the earth or neutral. The electrical protection devices are then defined for each one. Electrification: Application of voltage between two parts of the body of a living being. Electrocution: Electrification resulting in death. Fault current Id: Current resulting from an insulation fault. Frame: Conductive part likely to be touched and which, although normally insulated from live parts, may be brought to a dangerous voltage further to an insulation fault. Indirect contact: Contact of a person with accidentally energised frames (usually further to an insulation fault). Insulation: Arrangement preventing transmission of voltage (and current flow) between a normally energised element and a frame or the earth. Insulation fault: Insulation rupture causing an earth fault current or a short-circuit via the protection conductor.
Leakage current: Current which, in the absence of an insulation fault, returns to the source via the earth or the protection conductor. Limit safety voltage (UL): Voltage UL below which there is no risk of electrocution. Live conductors: Set of conductors assigned to electrical power transmission, including the neutral in AC and the compensator in DC, with the exception of the PEN conductor whose protection conductor (PE) function takes priority over the neutral function. Operating residual current If: Value of the residual current causing a residual current device to trip. According to construction standards, at 20C and for a threshold set at IDn, low voltage residual current devices must comply with: n < f < n 2 In high voltage, the zero phase-sequence relays have, allowing for operating accuracy, an operating current equal to the threshold displayed in amperes. Protection conductors (PE or PEN): Conductors which, according to specifications, connect the frames of electrical devices and some conductive elements to the earthing connection. Residual current: Rms value of the vector sum of the currents flowing through all live conductors in a circuit at a point of the electrical installation. Residual current device (RCD): Device whose decisive quantity is the residual current. It is normally associated with or incorporated in a breaking device.
Cahier Technique Schneider Electric no. 114 / p.2

Today, the residual current device is recognised the world over as an effective means of guaranteeing protection of people against electrical hazards in low voltage, as a result of indirect or direct contact. Its choice and optimum use require sound knowledge of the electrical installations and in particular of the earthing systems, existing technologies and their possibilities. All these aspects are dealt with in this Cahier Technique, completed by numerous answers provided by Schneider Electrics technical and maintenance departments to the questions which they are frequently asked.
Contents
1 Introduction 1.1 The RCD: its scope 1.2 Residual current protection and Earth leakage protection: two separate notions 1.3 The RCD, a useful protection device 2 The patho-physiological effects of electrical current on people 2.1 Effects according to current strength 2.2 Effects according to exposure time 2.3 Effects according to frequency 3 Insulation fault protection 3.1 3.2 3.3 3.4 3.5 3.6 4.1 4.2 4.3 4.4 4.5 The installation standards The direct contact risk Fire protection The TT earthing system The TN earthing system The IT earthing system Operating principle Sensors Measuring relays and actuators Product manufacturing standards The various devices p. 4 p. 4 p. 5 p. 6 p. 6 p. 8 p. 10 p. 11 p. 11 p. 11 p. 12 p. 12 p. 14 p. 14 p. 17 p. 19 p. 21 p. 22 p. 23 p. 26 p. 27 p. 31 p. 32
4 RCD operating principle and description
5 Optimised use of the RCD
5.1 EMC: manufacturers obligations and what this implies for contractors 5.2 A need: discrimination 5.3 Avoiding known problems 5.4 RCDs for mixed and DC networks
6 Conclusion Bibliography
1 Introduction
1.1 The RCD: its scope

In electrical installations, direct and indirect contacts are always associated with a fault current which does not return to the source via the live conductors. These contacts are dangerous for people and equipment (see Cahiers Techniques no. 172 and 173). For this reason the use of Residual Current Devices (RCD), whose basic function is detection of residual currents, is widespread. Moreover, RCDs monitor insulation of cables and electrical loads, and are therefore frequently used to indicate insulation drops or to reduce the destructive effects of a strong fault current.
Outgoing current Load Source Return current id = ia - ir Fault current Source
i1 i2 i3 in Load
In I3
Id
I1 I2
Fig. 1 : a current leakage results in a residual fault current id.
1.2 Residual current protection and Earth leakage protection: two separate notions
It is important not to confuse these two notions. A residual current device (RCD) is a protection device associated with a toroidal sensor surrounding the live conductors. Its function is detection of current difference or, to be more precise, residual current (see fig. 1 ). Existence of a residual current indicates presence of an insulation fault between a live conductor and a frame or the earth. This current takes an abnormal path, normally the earth, to return to the source. The RCD is normally combined with a breaking device (switch, circuit-breaker, contactor) which automatically de-energises the faulty circuit. Earth leakage protection consists of one or more measuring devices whose function is to detect a difference between the input current and the output current on part of the installation: line, cable, transformer or machine (generator, motor, etc.). This protection is mainly used in medium and high voltage. Earth leakage protection (zero phase-sequence current) for insulation fault protection (see fig. 2 ) and current leakage protection for phase-to-phase fault protection (see fig. 3 ) are both found.
Fig. 2 : earth leakage protection.
Fig. 3 : current leakage protection.
1.3 The RCD, a useful protection device

The first decisive factor in choosing and using RCDs in an installation is the earthing system provided. c In the TT earthing system (directly earthed neutral), protection of people against indirect contact relies on the use of RCDs. c In the IT and TN earthing systems, the medium and low sensitivity (MS and LS) RCDs are used: v to limit the risk of fire, v to prevent the destructive effects of a strong fault current, v to protect people against indirect contact (very long outgoers). c For all earthing systems, the high sensitivity (HS) RCDs provide additional protection against direct contact. They are compulsory in final distribution in a large number of countries. Their efficiency was confirmed at the end of this century by the remarked reduction in the number of people electrocuted. The result of an IEC survey conducted in August 1982 in Japan already proved the efficiency of these devices (see fig. 4 ). The residual current device is generally recognised (throughout the industrialised world) as being the best and most reliable of the protection devices developed to provide protection against indirect contact in the low voltage field. Such were the words of professor C.F. DALZIEL (Berkeley-USA), one of the pioneers of the study of the effects of electrical current on people in the fifth international conference of the AISS (Lucerne 1978).
Annual number of deaths by electrocution 40 Decree of the law making HS-RCDs compulsory
30
20
10
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80 Years
Fig. 4 : graph showing the evolution of deaths by electrocution due to the use of hand-held tools in Japanese companies. This figure begins to drop in 1970, the year after that in which a law was decreed making the use of high sensitivity RCDs compulsory.
2 The patho-physiological effects of electrical current on people
The patho-physiological effects of electrical current on people (tetanisation, external and internal burns, ventricular fibrillation and cardiac arrest) depend on a variety of factors: the physiological characteristics of the person in question, the environment (e.g. dry or wet) and the characteristics of the current passing through the body. As protection of people is the main function of the RCD, it is clear that optimum implementation of these devices requires knowledge of the
sensitivity thresholds of people and of the risks incurred. The International Electrotechnical Committee (IEC) has looked into the problem in order to pool, at international level, a variety of viewpoints reflecting and even often defending national practices, habits and standards. Many scientists have participated in this undertaking and have helped clarify the subject (Dalziell, Kisslev, Osypka, Bielgelmeier, Lee, Koeppen, Tolazzi, etc.).
2.1 Effects according to current strength

The effects of the electrical current passing through the human body depend on the frequency and strength of this current (see fig. 5 ).
Effects (for t < 10s) Slight tingling, perception threshold Painful shock, but no loss of muscular control Non-release threshold Considerable breathing difficulty Respiratory paralysis threshold
Current strength (mA) DC 50/60 Hz 3.5 41 51 60 0.5 6 10 15 30
10 kHz 8 37 50 61
Fig. 5 : effects of weak electrical currents on human beings.
2.2 Effects according to exposure time

The risks of non-release, respiratory arrest or irreversible cardiac fibrillation (see lexicon) increase in proportion to the time during which the human body is exposed to the electrical current (see fig. 6 ). The chart in figure 6 identifies in particular zones 3 and 4 in which danger is real. c Zone 3 (situated between curves b and c1). For people in this situation, there is normally no organic damage. However there is a likelihood of muscular contractions, breathing difficulties and reversible perturbation of the formation of impulses in the heart and of their propagation. All these phenomena increase with current strength and exposure time. c Zone 4 (situated to the right of curve c1) In addition to the effects of zone 3, the likelihood of ventricular fibrillation is: v approximately 5 % between curves c1 and c2, v less than 50 % between curves c2 and c3, v more than 50 % beyond curve c3. Patho-physiological effects such as cardiac arrest, respiratory arrest and serious burns increase with current strength and exposure time. For this reason it is accepted that use of an RCD with instantaneous operation and with a threshold of less than 30 mA, ensures this situation is never reached and such risks never incurred. With a more general approach, IEC 60364 (NF C 15-100 in France) stipulates the operating
times for the Residual Current Devices according to contact voltage. These times are recalled in the two tables in figure 7 . Limit safety voltage (UL) According to environmental conditions and particularly presence or absence of water, limit
safety voltage UL (voltage below which there is no risk for people, according to standard NF C 15-100) is, in AC: v 50 V for dry and wet premises, v 25 V for damp premises, for example for outdoor worksites.
ms 10000 5000 2000 1000 500 200 100 50 20 10 0.1 0.2 0.5 1 1 2 a
Duration of current flow b c1 c2 c3
mA 2 5 10 20 50 100 200 5001000 2000 500010000 Threshold = 30 mA Current flowing through the body
Fig. 6 : duration of current flow in the body as a function of current strength. In this chart, the effects of AC current (15 to 100 Hz) have been divided into four zones (as per IEC 60479-1).
Prospective contact voltage (V)
Maximum breaking time of the protection device (s) AC
DC
c Dry or wet premises or locations: UL i 50 V < 50 5 50 5 75 0.60 90 0.45 120 0.34 150 0.27 220 0.17 280 0.12 350 0.08 500 0.04 c Wet premises or locations: UL i 25 V 25 50 75 90 110 150 220 280 5 0.48 0.30 0.25 0.18 0.10 0.05 0.02
5 5 5 5 5 1 0.40 0.30 0.20 0.10
5 5 2 0.80 0.50 0.25 0.06 0.02
Fig. 7 : maximum duration of contact voltage holding as per standard IEC 60364.
Direct contact Direct contact with normally energised parts is dangerous for voltages in excess of UL. The main protection precautions to be taken are distance and insulation. The RCD can detect a fault current flowing through a person and, as such, is specified, regardless of the earthing system, in final distribution as an additional protection. Its operating threshold, as shown in the table in figure 5, must be less than or equal to 30 mA, and its operation must be instantaneous since the value of the fault current, dependent on the exposure conditions, may exceed 1A. Indirect contact On contact with an accidentally energised frame, the danger threshold is also fixed by the limit safety voltage UL. To ensure there is no danger when network voltage is greater than UL, contact voltage must be less than UL.
In the diagram in figure 8 , when the installation neutral is earthed (TT earthing system) where: RA = earthing resistance of the installation frames, RB = earthing resistance of the neutral, this implies choosing an operating threshold (I n) of the RCD such that: Ud = RA d UL and thus: I n i UL RA
The protection operating time must be chosen according to fault voltage RA Ud = U RA + RB (see fig. 7). Note that if the equipotentiality of the site is not ensured or is badly ensured, contact voltage is equal to fault voltage.
Id
N
RCD
PE RCD
Ud
RB
RA
Fig. 8 : fault voltage generation principle RCD.
2.3 Effects according to frequency

IEC 60479-2 deals with the effects of AC current of a frequency in excess of 100 Hz. Skin impedance decreases in reverse proportion to frequency. The standard states that the frequency factor, which is the ratio of current at the frequency (f) over current at the frequency of 50/60 Hz for the same physiological effect considered, increases with frequency. Moreover, it has been observed that between 10 and 100 kHz the perception threshold increases approximately by 10 mA to 100 mA in rms value. Although standards do not yet stipulate specific operating rules, the major manufacturers, aware of the potential risks of such currents, ensure that the thresholds of the protection devices they propose are below the ventricular fibrillation curve defined in standard IEC 60479-2 (see fig. 9 ).
Id() / Id(50 Hz)

25.00
20.00
15.00
10.00
5.00
0.00 10 Limit A type ID AC type ID Vigirex RH328A 100 50 1 000 10 000 Frequency (Hz)
Fig. 9 : variations in ventricular fibrillation threshold (as per IEC 60479-2) and thresholds of various RCDs set on 30 mA, for frequencies of between 50/60 Hz and 2 kHz (source: Merlin Gerin).
3 Insulation fault protection
3.1 The installation standards

RCDs are used in electrical, domestic and industrial installations. Their use depends on standards and mainly on the IEC 60364 (in France NF C 15-100). This standard officially stipulates three main systems for earthing the electrical network: the earthing systems (see fig. 10 ), used to a greater or lesser extent depending on the country. Furthermore, for each of these systems it defines more precisely the use of the RCDs, as the electrical hazard is greatly influenced by choice of earthing system (see Cahier Technique no. 172). It also describes the basic precautions which, in normal operating conditions, considerably reduce electrical hazards, for example: c distance and obstacles, c insulation: class II devices and safety transformers, c earthing of frames, c equipotentiality. General rules Whatever earthing system is chosen for an installation, the standards require that: c Each application frame be connected to an earthing connection by a protection conductor. c Simultaneously accessible application frames be connected to the same earthing connection. c A breaking device automatically disconnects all parts of the installation where a dangerous contact voltage develops. c The breaking time of this device be less than the maximum time defined (see fig. 7).
Directly earthed neutral (TT) 1 2 3 N PE
Multiple earthed neutral (TN-C) 1 2 3 PEN
RB
RA 1 2 3 N PE
RB Multiple earthed neutral (TN-S) 1 2 3 N PE
Unearthed neutral (IT)
RB : Permanent insulation monitor.
RB
Fig. 10 : the three main earthing systems are the TT, TN and IT systems, defined by IEC 60364-3. The TN may be either TN-C (neutral and PE combined) or TN-S (separate neutral and PE).
3.2 The direct contact risk

This risk is the same for people whatever earthing system is used. The protection measures stipulated by standards are therefore identical and use the possibilities offered by the high sensitivity RCDs. This is because: c As the fault current flows through a person in contact with a live conductor, he or she is exposed to the patho-physiological risks described above. c An RCD placed upstream of the contact point can measure the current flowing through the person and break the dangerous current. Regulations recognise the use of an RCD with high or very high sensitivity (i 30 mA) as an additional protection measure when the risk of direct contact exists due to the environment, the installation or people (article 412.5.1 of IEC 60364). This risk also exists when the protection conductor is likely to be broken or does not exist (hand-held devices). In this case use of a high sensitivity RCD is compulsory. Standard NF C 15-100, paragraph 532-2.6.1, states that RCDs with a threshold at most equal to 30 mA must protect the circuits supplying power outlets when they are: c Placed in damp premises or in temporary installations. c Of rating i 32 A in all the other installation cases. Note Standard IEC 60479 states that the resistance of the human body is greater than or equal to 1000 for 95 % of people exposed to a 230 V contact voltage and thus through whom a 0.23 A current flows. An RCD with a 30 mA threshold does not limit current, but its instantaneous operation ensures safety up to 0.5 A (see fig. 6). Use of an RCD with a sensitivity of 5 or 10 mA therefore does not increase safety. However it makes the risk of nuisance tripping not negligible as a result of capacitive leakage (distributed capacitances of cables and filters).
3.3 Fire protection

Whatever earthing system is used, the electrical installations of premises where risk of fire is present must be equipped with RCDs of a sensitivity In i 500 mA, as it is acknowledged that a 500 mA current can result in incandescence of two metal parts coming into occasional contact.
3.4 The TT earthing system

Protection of people against indirect contact In this system protection relies on use of RCDs. The fault current depends on the resistance of the insulation fault (Rd) and the resistances of the earthing connection. A person in contact with the metal enclosure of a load with an insulation fault (see fig. 8) may be subjected to the voltage developed in the load earthing connection (RA). For example Where U = 230V, RA = RB = 10 and Rd = 0, if the person is not on an equipotential site, he or she is subjected to Uc = Ud = 115 V. Protection must be provided by use of an RCD of medium or low sensitivity which must de-energise the faulty device as soon as the voltage Ud exceeds the limit safety voltage UL. We remind you that their operating threshold must be set at: In i UL . RA Protection of machines and equipment The level of the RCD tripping thresholds necessary for protection of people in the TT earthing system is well below that of the fault currents able to damage the magnetic circuits of machines (motor) or cause fires. The RCDs therefore prevent such electrical destruction.
3.5 The TN earthing system

Reminders c With this earthing system, the current of a full insulation fault is a short-circuit current. c In TN-C, in view of the fact that the neutral and the protection conductor are combined, RCDs cannot be used. The following text therefore mainly concerns the TN-S. Protection of people against indirect contact As the fault current depends on the impedance of the fault loop, protection is normally provided by overcurrent protection devices (calculation/ measurement of loop impedances). If the impedance is too great and does not allow the fault current to trip the overcurrent protection devices (very long cables), one solution is to use a low sensitivity RCD (In u 1 A). Moreover this system cannot be used when, for example, the network is supplied by a transformer whose zero phase-sequence impedance is too great (star-star connection). Protection of electrical devices and circuits In the multiple earthed neutral system, insulation faults are responsible for strong fault currents equivalent to short-circuit currents. The flow of such currents results in serious damage, for example: perforation of the magnetic circuit plates of a motor, requiring replacement instead of rewinding of the motor. Such damage can be greatly limited by use of a low sensitivity RCD (e.g. 3 A) with instantaneous operation, which is thus able to react before the current reaches a high value. Note that the need of protection increases as operating voltage rises, as the energy lost at the fault point is proportional to the voltage square. The economic consequence of such destruction must be estimated as it is a vital criterion in choice of earthing system. Detection of insulation faults between the Neutral and the protection conductor (PE) or building frames This type of fault insidiously transforms the TN-S system into a TN-C system. Part of the neutral current (increased by the sum of 3rd order and multiple of 3 harmonic currents) permanently flows in the PE and in the metal structures of the building with two consequences: c Equipotentiality of the PE is no longer ensured (a few volts may disturb the operation of the digital systems connected by bus and which must have the same potential reference). c Current flow in the structures increases the risk of fire. The RCDS are used to highlight this type of fault. Detection of insulation faults without tripping and protection of equipment In the TN-S earthing system, unlike the IT system, there are no safety rules stipulating insulation monitoring. However, all tripping further to an insulation fault is the cause of operating losses and often of costly repairs prior to re-energisation. For this reason, more and more often operators request prevention devices in order to take action before the insulation loss becomes a short-circuit. The answer to this need is the use in indication, in TN-S, on critical outgoers, of an RCD with a threshold of around 0.5 to a few amperes, which can detect insulation drops (on the phases or neutral) and alert operators. On the other hand, the risk of electrical fire is reduced and destruction of equipment avoided by using an RCD with tripping for In i 500 mA.
3.6 The IT earthing system

Protection of people against indirect contact When the first insulation fault occurs, the fault current is very weak and the fault voltage not dangerous: the standards require that this fault be indicated (by the permanent insulation monitors) and tracked (by the power on fault tracking devices). When the second fault occurs, the installation finds itself in a situation similar to a fault in the TN earthing system. However there are two possibilities: that of a single earthing connection for all the frames and that of multiple earthing connections.
c Case of a single earthing connection In this case protection is usually ensured by the overcurrent protection devices (calculation/ measurement of the loop impedances). c Case of multiple earthing connections When both faults affect devices not connected to the same earthing connection, the fault current may not reach the operating threshold of the overcurrent releases. The standards stipulate use of RCDs on each group of frames interconnected with the same earthing connection. c In all cases, simple or multiple earthing connections If the impedance of a fault loop is too great (very long cables), a simple, practical solution is to use a low sensitivity RCD (1 to 30 A).
Protection of equipment, electrical devices and circuits Although there is no particular danger for equipment when the first fault occurs, a second fault is normally responsible for strong fault currents equivalent to short-circuit currents, as in the TN earthing system. RCDs with medium or low sensitivity can then be provided for the more critical cases (premises with risk of fire, sensitive and expensive machines), bearing in mind that the risk of the second fault is particularly low, especially when tracking of the first faults is systematic. In actual fact, assuming a fault once every three months and that this fault is eliminated the same day, the average time between two double faults is approximately 22 years!
4 RCD operating principle and description
4.1 Operating principle

All residual current devices are made up of at least two components: c The sensor The sensor must be able to supply an electrical signal which is useful when the sum of the currents flowing in the live conductors is different from zero. c The measurement relay The relay compares the electrical signal supplied by the sensor with a setpoint value and sends, with a possible deliberate delay, the opening order to the associated breaking device. The unit controlling the opening of the device (switch or circuit-breaker) placed upstream of the electrical circuit monitored by the RCD is known as the trip unit or actuator. The entire RCD is shown in the diagram in figure 11 .
Toroid
Shaping
Threshold
Time delay relay
Static or relay output
Auxiliary supply source
Fig. 11 : functional diagram showing an electronic RCD with auxiliary supply source.
4.2 Sensors
Two types of sensors are normally used on AC circuits: c The toroidal transformer, which is the most common for measuring leakage currents. c The current transformers, used in HV and MV and sometimes in LV. The toroidal transformer This covers all the live conductors and is thus excited by the residual magnetic field corresponding to the vector sum of the currents flowing through the phases and the neutral. Induction in the toroid and the electrical signal available at the terminals of the secondary winding are thus the image of the residual current. This type of sensor is used to detect residual currents from a few milliamperes to several dozen amperes. The current transformers (CT) To measure the residual current of a threephase electrical circuit without neutral, three current transformers must be installed as shown in figure 12 .
I1
I2
I3
A
Ih
RCD
Fig. 12 : the vector sum of the phase currents yields the residual current.
The three CTs are parallel-connected current generators, causing circulation between A and B of a current which is the vector sum of the three currents and thus the residual current. This circuit, known as the Nicholson circuit, is commonly used in MV and HV when the earth fault current can reach several dozen or even several hundred amperes. During use, care must be taken with the CT accuracy class: with CTs of 5 % class, it is prudent not to set earth protection below 10 % of their nominal current. The HV electrical installation standard NF C 13-200 of December 1989 specifies 10 %. Special cases c High power supply The Nicholson CT circuit, which would be useful in LV when the conductors are large crosssection bars or cables for the transmission of strong currents, does not allow, even with coupled CTs, settings that are compatible with protection of people (threshold In i UL / Ru). There are a number of solutions: v If the problem occurs in a main switchboard downstream of the transformer, the following may be considered: - either installation of a toroid at the supply end of the installation on the earthing connection of the transformer LV neutral (see fig. 13 ). This is because, according to the Kirchhoff node law, the residual current detected by (N) is strictly the same as that detected by (G) for a fault occurring in LV distribution, - or installation of a toroid on each outgoer, all parallel-connected to a single relay (see fig. 14 ). When the measurement relay (normally electronic) only needs a very weak electrical signal to operate, the toroids can be made to operate as current generators. When parallelconnected, they give the image of the vector sums of the primary currents. Although this circuit is laid down in the installation standards, the approval of the RCD manufacturer is preferable. However, for discrimination reasons, it is preferable to use one RCD per outgoer. v If the problem arises with parallel-connected cables which cannot all cross a toroid, a toroid can be placed on each cable (including all the live conductors), and all the toroids can be parallel-connected (see fig. 15 ). However the following must be noted: v That each toroid detects n turns in short-circuit (3 in the figure) which may reduce sensitivity.
HV / LV
G 1 2 3
RCD
RCD
Fig. 13 : toroid N delivers the same information as toroid G.
RCD
Fig. 14 : toroids placed on the outgoers and parallelconnected to a single relay compensate the impossibility of placing a toroid on the incomer.
Fig. 15 : layout of toroids on parallel-connected large diameter single-line cables.
v If the connections represent impedance differences, each toroid will indicate a false zero phase-sequence current. However proper wiring considerably limits these currents. v That this circuit implies for each toroid that the output terminals S1-S2 be marked according to the energy flow direction. This solution calls for the approval of the RCD manufacturer. c High power outgoer To ensure a reliable, linear toroid response, the live conductors must be placed as close as possible to the centre of the toroid so that their magnetic effects are completely compensated in the absence of residual current. In actual fact, the magnetic field developed by a conductor decreases in proportion to distance: thus in figure 16 , phase 3 causes at point A a local magnetic saturation and thus no longer has a proportional effect. The same applies if the toroid is placed near or in a bend of the cables that it surrounds (see fig. 17 ). The appearance of a stray residual induction, for strong currents, will generate on the toroid secondary a signal that may cause nuisance tripping. The risk increases as the RCD threshold drops with respect to phase current, particularly on a short-circuit. In problem cases (Max. Iph. / high In), two solutions can be used to counter the risk of nuisance tripping: v Use a toroid that is far larger than necessary, for example with a diameter that is twice the one just right for conductor insertion. v Place a sleeve in the toroid. This sleeve must be made of magnetic material in order to homogenise the magnetic field (soft iron - magnetic plate), (see fig. 18 ). When all these precautions have been taken: - centring of conductors, - large toroid, - and magnetic sleeve, the ratio max. phase may reach 50,000. n
c Can operate the toroid at higher induction in order to maximise the energy sensed and minimise sensitivity to stray inductions (strong currents).
1 3 2
Fig. 16 : incorrect centring of conductors in the toroid is responsible for its local magnetic saturation at point A, which may be the cause of nuisance tripping.
L u2
Fig. 17 : the toroid must be far enough from the cable bend so as not to be the cause of nuisance tripping.
Using an RCD with built-in toroid It must be pointed out that RCDs with built-in toroids provide contractors and operators with a ready-made solution since it is the manufacturer who studies and works out the technical solutions. This is because he: c Masters the problem of centring the live conductors and, for weak currents, can anticipate and properly distribute several primary turns around the toroid.
Lu2
Fig. 18 : a magnetic sleeve placed around the conductors, in the toroid, reduces the risk of tripping due to the magnetic effects of the current peaks.
4.3 Measurement relays and actuators

The RCDs can be classed in three categories according to their supply mode or their technology. According to their supply mode With own current: in this device the tripping energy is supplied by the fault current. This supply mode is considered by most specialists as the most reliable. In many countries and particularly in Europe, this category of RCD is recommended for domestic and similar installations (standards EN 61008 and 61009). With auxiliary supply source: in this device the tripping energy requires a source of energy that is independent from the fault current. These devices (normally electronic) can therefore only cause tripping if this auxiliary energy source is available when the fault current appears. With own voltage: this is a device with an auxiliary supply source but whose source is the monitored circuit. Thus, when this circuit is energised, the RCD is supplied, and when this circuit is not energised, the RCD is not activated but there is no danger. An additional guarantee is provided by these devices when they are designed to operate correctly with voltage drops of up to 50 V (safety voltage). This is the case of the Vigi modules which are RCDs associated with the Merlin Gerin Compact circuit-breakers. However, as far as power supply is concerned, the RCDs are also classed according to whether or not their operation is of the fail-safe kind. Two types of devices are considered to be failsafe: c Those whose tripping only depends on the fault current: all own current devices are fail-safe devices. c And those, more seldom used, whose tripping does not only depend on the fault current but which are automatically placed in the tripping position (safety position) when the conditions no longer guarantee tripping in the presence of the fault current (e.g. a voltage drop up to 25 V). According to their technology Electromagnetic devices (see fig. 19 ). These modern devices are of the own current type and use the principle of magnetic latching. A very low electrical power (100 VA for some) is sufficient to overcome the latching force and cause the contacts to open by means of a mechanical amplifier. They are very widespread (with the fail-safe function) and particularly suitable for the creation of an RCD with a single sensitivity. Electronic devices These devices are particularly used in industry as electronics ensures: c A very low acquisition power, c Accurate, adjustable thresholds and time delays (thus ensuring optimum tripping discrimination). Due to these two characteristics, these devices are ideal for the creation of: c RCDs with separate toroids, which are associated with high rating circuit-breakers and contactors. c RCDs associated with industrial circuitbreakers up to 630 A. Electronic devices require a certain energy, often very weak, to operate. RCDs with electronic devices are therefore available with the various supply modes described above, either with own voltage or with auxiliary supply source.
Ia
Ir
Fig. 19 : the fault current, via the toroid, supplies energy to an electromagnet whose moving part is stuck down by a permanent magnet. When the operating threshold is reached, the electromagnet destroys the attraction of the permanent magnet and the moving part, drawn by a spring, opens the magnetic circuit and mechanically controls circuitbreaker opening.
Mixed devices (own current) This solution consists of inserting between the toroid and the magnetic latching relay a signal processing device, allowing: c An accurate, precise operating threshold. c Excellent immunity to interference and steep front current transients, while respecting an operating time compatible with safety curves. As an example, Merlin Gerin si type RCDs are mixed devices. c Creation of time-delayed RCDs. A similar principle is used in MV. In point of fact, a few years ago tripping in electrical power supply consumer substations (MV/LV substation) required an accumulator bank which was the source of many problems. The combination of an own current electronic device and an electromechanical trip unit with magnetic latching offers a satisfactory solution with respect to cost and reliability with removal of the battery. Operational requirements IEC 60364, paragraph 531-2-2-2 states the following for non fail-safe devices with auxiliary supply source: Their use is permitted if they are installed in installations operated by experienced and qualified people. Standard NF C 15-100, paragraph 532.2.2 also states that they must not be used in household installations or for similar applications. Proper operating test An RCD is a safety device. Whether it is electromagnetic, electronic or mixed, it is thus essential for it to be equipped with a test device. Although own current devices appear the most reliable, implementation of fail-safe safety with the other own voltage or auxiliary supply source energy sources grant the RCDs an increased degree of safety which does not, however, replace the periodical test. c Recommend periodical RCD testing In practice perfect fail-safe safety, particularly concerning internal faults, does not exist. For this reason, in France, RCDs using an auxiliary supply source are reserved for industrial and large tertiary installations and own current RCDs for domestic and similar installations: an arrangement which is consistent with their inherent possibilities described above. In all cases, periodical testing should be recommended for highlighting internal faults. c The manner in which the test is conducted is important.
This test must allow for the fact that capacitive earth leakage currents are always present in an electrical installation, as are often resistive leakage currents resulting from damaged insulation. The vector addition of all these leakage currents (Id) is detected by the toroidal sensor and may affect test operation, in particular when the test circuit is the one shown in figure 20 . Despite this, this test principle is widespread as it checks the toroid/relay/breaking device assembly. Construction standards limit the test current, which may account for a certain number of RCD operating failures during the test, as shown by the vector addition (see figure 20) of the leakage current (Id) and the test current (test I). For example standards IEC 61008 and 61009 state that the test current must not exceed 2.5 In for an RCD usable in 230 or 400 V, i.e. 1.15 In if it is supplied in 230 V - 20%. The test principle described above is used on earth leakage protection sockets, residual current circuit-breakers and residual current devices. With respect to residual current relays with separate toroid, the same principle is sometimes chosen when the contractor is the person producing the test circuit. However some relays, for example the Merlin Gerin Vigirex, are equipped with the test function, and also permanently monitor continuity of the detection circuit (toroid/relay link and toroid winding).
Test 1 2 3
I test Id
location of If
I test
Id
Ir
Ir = Id + I test Ir u If
Fig. 20 : some test circuits created on installation may not operate in the presence of weak fault currents.
c Verification of the operating threshold Even more so than for the test, it is important to bear in mind when carrying out this verification that leakage currents of the downstream circuit,
whether or not they are natural, may flow through the sensor. For reliable measurement, the downstream circuit will always be disconnected.
4.4 Product manufacturing standards

The main RCD manufacturing standards are listed in the appendix. The IEC has standardised for the RCDs, types, threshold values, sensitivities and operating curves. AC, A and B type RCDs to be chosen according to the current to be detected The current conveyed in electrical networks is increasingly less sinusoidal. Consequently standard IEC 60755 has defined three types of RCD: the AC, A and B types, according to the residual current to be detected (see fig. 21 ). c The AC type, for sinusoidal AC currents. c The A type, for sinusoidal AC currents, pulsed DC currents or pulsed DC currents with a DC component of 0.006 A, with or without phase angle monitoring, whether they are suddenly applied or slowly increase. c The B type, for the same currents as the A type, but in addition for rectifier currents: v with simple halfwave with a capacitive load producing a smoothed DC current, v three-phase with simple or double halfwave.
For RCDs of the type:
Id
AC
Id
A
Id
B
Fig. 21 : fault currents stipulated in the RCD construction standards.
Sensitivities (In) These are standardised by the IEC: c high sensitivity -HS-: 6-10-30 mA, c medium sensitivity -MS-: 100-300 and 500 mA, c low sensitivity -LS-: 1-3-5-10 and 20 A. It is clear that HS is most often used for direct contact protection, whereas the other sensitivities (MS and LS) are used for all other protection needs, such as indirect contact
(TT earthing system), fire hazards and machine destruction protection. Tripping curves These curves take into account the international studies performed on electrical hazards (IEC 60479) and in particular: c the effects of current in the case of direct contact protection, c limit safety voltage in the case of indirect contact protection.
With respect to the domestic and similar sector, standards IEC 61008 (residual current circuitbreakers) and 61009 (residual current devices) define standardised operating time values (see table in figure 22 for the operating curves G and S in figure 23 ): c The G curve for the instantaneous RCDs. c The S curve for the selective RCDs with the lowest time delay level, used in France for incomer circuit-breakers for example. For power residual current devices, they are given in appendix B of standard IEC 60947-2. The above standards define the maximum operating time as a function of the Id/If ratio for inverse response time RCDs (often electromagnetic). Electronic RCDs, mainly used in industry and large tertiary, normally have an adjustable threshold and time delay, and their response time is not dependent on the fault current. IEC 60364 (NF C 15-100) defines maximum breaking times on final circuits for the TN and IT earthing systems (see fig. 24 ). For the TT earthing system, RCD operating time must be
chosen according to fault voltage. In practice G and S type RCDs are suitable on final circuits for i 230/440 V network voltages. The standard also stipulates that a time of 1 second is acceptable in the TT system, for distribution circuits, in order to create the discrimination stages required for continuity of supply. In addition to the above-mentioned characteristics of the residual current function, product standards also stipulate: c impact strength and jarring withstand, c ambient temperature and humidity, c mechanical and electrical durability, c insulation voltage, impulse voltage withstand, c EMC limits. The standards also make provision for type tests and for periodical checking of quality and performance carried out either by the manufacturer or by approved organisations. They thus guarantee users product quality and safety of people. RCDs are also marked for quality, for example NF-USE marking in France.
Type
In
(A)
In
(A) All values > 0.030
Standardised value of operating and non-operating times (in seconds) at: In 2 In 5 In 500 A 0.3 0.5 0.13 0.15 0.2 0.06 0.04 0.15 0.05 0.04 0.15 0.04 Maximum operating time Maximum operating time Minimum nonoperating time
General All (instantaneous) values Selective > 25
Fig. 22 : standardised values of the maximum operating times and non-operating times as per IEC 61008.
t (ms) 500 200 100 50 20 10 1 2 5 10 500 A G S max.
Id / In.
Fig. 23 : maximum operating time curves for S (selective) and G (general purpose) residual current circuitbreaker or device.
Nominal phase-to-earth network voltage (VCA). 120-127 220-230 400 580
Maximum breaking time (s) TN IT IT Non-distributed Distributed neutral neutral 0.8 0.4 0.2 0.1 0.8 0.4 0.2 0.1 5 0.8 0.4 0.2
Fig. 24 : maximum breaking times.
4.5 The various devices

The standards state that technologically different RCDs exist that are suited to the two main sectors: domestic and industry. The RCD must be chosen according to the network earthing system and the protection target (direct contact, indirect contact, load protection, etc.). However it is also necessary to: c Define its type (A, AC or B) from the network characteristics (AC, mixed, etc.), c Analyse operating requirements (discrimination needs, fail-safe safety needs, etc.), in order to determine: v the required threshold level (sensitivity), v the time delay ranges (delay). The table in figure 25 gives a concise presentation of the various devices.
Areas - Types Domestic and similar Extension with earth leakage protection (breaking by built-in contact) Earth leakage protection socket (breaking by built-in contact) Residual current circuit-breaker Residual current device c Incomer
Network earthing system
Sensitivity i 30 mA
Time delay
TT - TN - IT
TT - TN - IT
30 mA
TT - TN - IT
30 - 300 mA
TT
In France
In = 500 mA
is the most common TT TT - (TN and IT in socket circuit protection) TT - (TN and IT in fire, machine and long outgoer protection) TT - (TN and IT in fire and machine protection) TT - (TN and IT in fire, machine and long outgoer protection) 30 - 300 mA 30 - 300 mA
S type as option (disturbed network with or without surge arrester) 0 0
c Final distribution Industry and large tertiary Residual current circuit-breaker Residual current device c Power
30 mA to 30 A
0 to 1 s
c Final distribution Residual current relay with separate toroid
30 - 300 mA 30 mA to 30 A
0 0 to 1 s
Fig. 25 : general presentation of the various RCDs.
5 Optimised use of the RCD
5.1 EMC: manufacturers obligations and what this implies for contractors
EMC (Electro Magnetic Compatibility) is the control of electrical interference and its effects: a device must neither be disturbed nor disturb its environment. All electrical equipment manufacturers must naturally comply with certain EMC standards. RCDs are tested for electromagnetic compatibility (emission and susceptibility) according to the European Directive which specifies compliance with a certain number of standards (for example: EN 61543 for domestic RCDs). However, electrical installations generate or transmit disturbances (see Cahier Technique no. 187), which can be permanent or temporary, alternating or impulse, low or high frequency, as well as conducted or radiated, common or differential mode, internal or external to buildings. Overvoltage is one of the most troublesome disturbances. Overvoltage withstand RCDs can be sensitive to lightning strokes, particularly on overhead networks which are more likely to be affected by atmospheric disturbances. In point of fact, according to the distance of the cause of the disturbance, an LV network can be subjected to: c An overvoltage occurring between the live conductors and the earth: the disturbance flows off to the earth well upstream of the RCDs (see fig. 26a ). c An overcurrent, a part of which flows off in the network downstream of the RCD, particularly via the stray capacitances (see fig. 26b ). c An overcurrent detected by the RCD and which is due to breakdown downstream of this RCD (see fig. 26c ). Technically speaking, solutions are known and normally implemented by RCD manufacturers. Such solutions include: c For electromagnetic relays, installation of a parallel diode on the relay exciting circuit. This solution is used for incomer circuit-breakers. c For electronic relays, use of a low-pass filter at signal shaping level (see fig. 11). Manufacturing standards make provision for RCDs immunised against these stray currents: the S type RCDs (In u 100 mA). However manufacturers also propose devices with high sensitivity and reinforced immunity such as the Merlin Gerin RCDs of the si type (In i 30 mA). Thus, confronted with this problem, installation service quality is only dependent on the device chosen.
t 1,2 s I 50 s b
t 10 s
t 8 s 20s
Fig. 26 : standardised voltage and current waves representative of lightning.
Influence of choices when designing an installation Installation designers and installers, while respecting proper procedures, are also active in this area, particularly when choosing the earthing system for the installation. For example they must know that in the TN system, several currents are responsible for the disturbance due to radiation of sensitive devices: c On an insulation fault, strong currents flow in the PE, the device frames and the structures.
c In the TN-C earthing system, load unbalance currents flow continuously in the metal structures of the buildings. c In the TN-S earthing system, these unbalance currents also appear on an insulation fault between the neutral and the protection conductor. Moreover, this fault, which cannot be detected by the overcurrent protection devices, insidiously changes the TN-S system into a TN-C system.
5.2 A need: discrimination

Ensure that only the faulty outgoer is deenergised by the tripping of the protection device: this is the purpose of discrimination and the aim of protection co-ordination. Vertical discrimination This type of discrimination presides over the operation of two protection devices connected in series on a circuit (see fig. 27 ). In view of RCD operating requirements as well as of their manufacturing standards, discrimination must be both current and time. c Current, as, according to the standards, an RCD must trip at I n and not trip at I n / 2. In practice, a ratio of 3 is required: In (upstream) u 3 In (downstream). c Time, as, in order to react, all mechanisms need a period of time, even the smallest, to which a deliberate time delay or delay must sometimes be added. The double condition of non tripping of Da for a fault downstream of Db is therefore: In (Da) > 2 In (Db). and tr (Da) > tr (Db) + tc (Db) or tr (Da) > tf (Db) where: tr = tripping delay = time of non operation tc = time separating the moment of breaking (including arcing time) from the moment when the breaking order was given by the measurement relay, tf = operating time, from detection of the fault through to complete breaking of the fault current. Time-delayable electronic relays may exhibit a fault memorisation phenomenon by their threshold circuit. It is then necessary to take into account a memory time -tm- (see fig. 28 ) to ensure that they do not trip after opening of the downstream device: tr (Da) > tf (Db) + tm. Note Problems may be encountered when implementing discrimination if it is necessary to combine residual current devices and residual current relays, since: c The residual current device is defined in delay time -tr-.
Da RCD
Db RCD
Fig. 27 : vertical discrimination.

tr (1) tr (2) tc tm tc tm tc
(3)
Fig. 28 : the time delay of an upstream RCD must take account of the breaking time associated with the downstream RCD and of the memory time of the upstream relay.
c The residual current relay is defined in specific operating time or time delayed to a value t, which corresponds to the time elapsing between the occurrence of the fault and the transmission of the opening order to the breaking device (see fig. 29 ).
The successive times tf and tr (or t) must then be calculated (at 2 I n) for each RCD, starting at final distribution and moving back towards the origin of the installation.
RCD Vigicompact tr = 60 ms Vigirex t = 200 ms
RCD
RCD Vigirex RH tr = 15 ms tc = 30 ms tf = 45 ms Vigicompact tr = 60 ms tf < 140 ms
RCD
Fig. 29 : two examples of time discrimination, associating a residual current device of the Vigicompact type and a Vigirex relay (Merlin Gerin). Note that these times are far shorter than the authorised tripping times in figure 24.
Horizontal discrimination Sometimes referred to as circuit selection, stipulated in standard NF C 15-100 paragraph 536.3.2, it means that a residual current device placed in a cubicle at the supply end of the installation is not necessary when all the outgoers in this cubicle are protected by residual current devices. Only the faulty outgoer is deenergised: the residual current devices placed on the other outgoers (parallel to the faulty outgoer) do not detect the fault current (see fig. 30 ). The residual current devices may then have the same tr (or t). In practice, horizontal discrimination may go wrong. Indeed nuisance tripping known as sympathy tripping has been observed, particularly on networks containing very long outgoers (stray capacitances of unbalanced cables) or capacitive filters (computer). Two examples are given below: c Case 1 (see fig. 31 ) The opening of Db placed on the supply circuit of a load R, a powerful overvoltage generator
(e.g. welding machine) causes an overvoltage on the network. This overvoltage causes on outgoer A, protected by Da, the occurrence of a capacitive earthing current which may be due to the stray capacitances of the cables or to a capacitive earthing filter.
RCD
RCD
Fig. 30 : example of horizontal discrimination.
A solution: the RCD of Db may be instantaneous and the RCD of Da must be time-delayed. Note that for this configuration, the time delay of the RCD (Da) is often vital as, when circuit A is energised, the capacitances (stray or otherwise) cause the appearance of a damped oscillating residual current (see fig. 32 ). As a guideline, a measurement taken on a large computer containing an interference filter revealed a current with the following characteristics: v 40 A (first peak), v f = 11.5 kHz, v damping time (66 %): 5 periods. c Case 2 (see fig. 33 ) A full insulation fault on phase 1 of outgoer B places this phase at the potential of the earth.
The capacitive current supplied by outgoer A will cause by sympathy the tripping of the corresponding RCD. This phenomenon exists for all earthing systems, but mainly affects networks using the IT system. Both examples show the need to time delay the RCDs of long outgoers and those containing filters. Use of directional RCDs is another solution to prevent tripping due to the return of capacitive current via the healthy outgoer. This type of RCD detects the fault current, compares its amplitude with the scheduled threshold level and only trips if this current passes through the toroid from upstream to downstream.
I
(A) Da Db (B)
RCD
RCD
Fig. 31 : the presence of a capacitance on outgoer A may cause: c on opening of Db, the tripping of Da , and/or c on energisation of outgoer A, the tripping of Da . The use of time-delayed RCDs is often necessary to protect against the nuisance tripping caused by lightning overvoltages or equipment switchings.
Fig. 32 : transient current wave occurring on closing of a highly capacitive circuit.
Da
(A)
Extended network
1 2 3 Db
RCD
Cp
(B)
RCD
Fig. 33 : in the presence of a fault, Da may open instead of Db. Use of time-delayed RCDs is often necessary to protect against nuisance tripping on healthy outgoers.
5.3 Avoiding known problems

Taking leakage currents into account The last sub-chapter emphasises the attention that must be paid to these currents, often capacitive, which by deceiving the RCDs are able to seriously disturb operation. c 50 Hz - 60 Hz leakage currents As from the design stage of the installation, the lengths of the various outgoers should be evaluated, together with the future equipment containing capacitive earthed devices. It is then necessary to design a distribution system able to reduce the importance of this phenomenon. Consequently, interference filters (compulsory according to the European directive on EMC) placed on the microcomputers and other electronic devices, generate in single-phase permanent leakage currents at 50 Hz of the order of 0.5 to 1.5 mA per device. These leakage currents add up if the devices are connected to the same phase. And if these devices are connected to all three phases, these currents cancel each other out when they are balanced (vector sum). This reflection is all the more true when the RCDs installed have low thresholds. In order to guard against nuisance tripping, the permanent leakage current must not exceed 0.3 In in the TT and TN systems, and 0.17 In in the IT system. c Transient leakage currents These currents appear on energisation of a circuit with a capacitive unbalance (see fig. 33) or on a common mode overvoltage. S type RCDs (In u 300 mA) and si type RCDs (In = 30 mA and 300 mA) prevent nuisance tripping as do also slightly time-delayed RCDs. c High Frequency leakage currents Examples of large EMC polluters are thyristor rectifiers whose filters contain capacitors which generate an HF leakage current able to attain 5% of nominal current. Unlike the 50 Hz - 60 Hz leakage currents whose vector sum is zero, these HF currents are not synchronous over all three phases and their sum constitutes a leakage current. In order to prevent nuisance tripping, RCDs must be protected against these HF currents (equipped with low-pass filters): this is the case for industrial RCDs and for the Merlin Gerin S and si type RCDs. c Lightning currents If the installation is equipped with a surge arrester, the RCD sensor should not be placed on the flow path of the current generated by the lightning (see fig. 34 ). Otherwise, implementation of RCDs immunised against these currents (time-delayed or S type) is the solution. Maintaining the earthing system When replacement sources are provided, protection of people and equipment should be studied in the various configurations of the installation, as the position of the neutral with respect to the earth may be different. The supply, even temporary, of an installation with a generator set requires interconnecting the sets frame with the existing earthing network whatever the earthing system and, in the TT system, earthing of the generator neutral, since otherwise the fault currents would not reach the RCD threshold. When the installation in the TT earthing system contains an Uninterruptible Power Supply (UPS), earthing of the neutral downstream of the UPS is essential for proper operation of the RCDs (K contactor on figure 35 ), but not for protection of people as: c The installation is then in the IT system and the first fault is not dangerous (see standard C 15-402, paragraph 6.2.2.2.). c The likelihood of a second insulation fault occurring during the period of operation limited by back-up time of the UPS batteries is very slight.
A RCD
B RCD
Flow off of current generated by lightning
Surge arrester
Fig. 34 : in an installation containing a surge arrester, according to local obligations, the RCD may be placed differently: in A a time-delayed or S type RCD and in B a standard RCD.
Non backed up outgoers
3L
3L N 3L N 3L Transfer switches N 3L 3L N N (Maintenance) K
3L N
Bypass circuit Power loss detection relay
Transfer switches Backed up equipment
Fault supplied by the self-generating UPS Fault supplied by the mains
Fig. 35 : on detection of mains power loss on the UPS supply, the contactor K reproduces the TT system downstream of the UPS.
5.4 RCDs for mixed and DC networks

An insulation fault with DC current is far less dangerous than with AC current Experiments (see fig. 5) have shown that for weak currents people are approximately 5 times less sensitive to DC current than to 50/60 Hz AC current. The risk of ventricular fibrillation only appears over 300 mA. Installation standards NF C 15-100 and IEC 60479 have chosen a ratio close to 2, taking account of the fact that in practice fault currents are directional but not always smoothed. This is illustrated by figure 36 drawn up using the table in figure 7. Note that a three-phase rectifier supplied by a 400 V AC phase-to-phase voltage generates a direct contact voltage of 270 V DC, which corresponds to a maximum breaking time of 0.3 s.
t (s) 5 2.5
1 0.75 0.5 0.3 0.2 0.1 0.08 0.04 100 50 120 Contact voltage 100 200 230 400 500 (V AC) 200 250 300 400 500 (V DC)
Fig. 36 : curves established from the maximum breaking times of an RCD laid down by NF C 15-100, paragraph 413.1.1.1.
RCD manufacturing standards take into account the existence of non AC currents, and particularly define the standard cases shown in figure 21 and describe the relevant tests. To give an example, residual current circuit-breakers must operate for Id i 1.4 In in all cases corresponding to figure 37 , with or without superimposition of a smoothed DC current of 6 mA: the fault current is applied either suddenly or by slowly increasing from 0 to 1.4 In in 30 s. The RCDs satisfying these tests can be identified by the following symbol on their front face:
k
Real fault currents These currents reflect the voltages existing between the fault point and the neutral of the installation. The waveform of the fault current is seldom the same as that of voltage or applied current, delivered to the load. Fault voltages and currents of the pure DC type (zero ripple factor) are very rare. c In the domestic sector, distribution and rectifier circuits are single-phase, and correspond to the diagrams marked A to F in figure 38 . A type RCDs provide protection of people. However, for diagram B, they only detect fault currents if their occurrence is sudden. Note that circuit E is increasingly common as it is placed at the input of switch mode power supplies that are widespread in electrical household appliances (TV, microwave, etc.) as well as in professional equipment (microcomputers, photocopiers, etc.). c In industry most rectifier circuits are threephase (diagrams G to K in figure 39 , see page 30). Some circuits may generate a DC fault current with a small ripple factor:
v Circuits G and H Circuit G supplies a rectified voltage with a permanent small ripple factor, and consequently fault currents that are hard to detect by the RCD. On the other hand, circuit H generates highly chopped fault currents which are thus visible by the RCD. However this circuit is equivalent to circuit G for full wave conduction. v Circuit J This common circuit type is particularly used for variable speed controllers used in DC motors. The back-electromotive force and reactor of the motors generate smoother fault currents than the G and H circuits described above. However, regardless of the thyristor conduction angle, the RCDs placed upstream of the variable speed controllers must be able to provide protection. Some standard RCDs may be suitable provided their lDn threshold receives a suitable setting. To give an example, figure 40 shows the sensitivity of an RCD, with analogue electronic technology, according to the variable speed controller output voltage applied at the motor. v Circuit K With this circuit type, a fault on the DC circuit does not produce d / dt in the magnetic sensors of the RCDs which are then blinded. This circuit is dangerous unless a transformer is used instead of an autotransformer, as AC and A type RCDs are inoperative. Special case: DC current return Let us now see what happens when a second fault occurs on the AC part of a network (see fig. 41 ) containing a rectifier according to circuit G described above. If the power supply (A) of the rectifier is not monitored by an RCD, or if this RCD has been incorrectly chosen or is inoperative for any reason, the insulation fault existing on the DC part remains.
Sensitivity and 100 % Sinusoidal AC fault On-load motor and 90 Off-load motor 20 % 50 %
and 135
Fig. 37 : waveform of the A type RCD test currents.
0.15
Ud/Udo 100 %
Fig. 40 : evolution of the sensitivity of an electronic RCD placed upstream of a thyristor rectifier.
ph
A/ Soldering iron or two "setting" light dimmer switch
Id
R t N
ph
B/ Television, battery charger, etc. R
Id
t N
ph
C/ Light dimmer, arc welding machine R
Id
t N
ph
D/ Household appliances with motor (universal) M _
Id
t N
E/
Id
ph N t R
F/
Id
ph N t R
Fig. 38 : form of the fault currents detected on the single-phase supply of rectifiers when an insulation fault occurs on their positive output.
(+)
1 2 3 R
G/ Welding machine c electromagnet c electrolysis c etc.
Id
Fault on (+)
(-) (+) H/ Rectifier set for: c industrial DC network c electrophoresis R N NB: The fault current in (+) follows the upper limit of the conduction zones. Likewise, the fault current in (-) follows the lower limit. J/ Variable speed controller for DC motor.
Fault on (-)
Id
Fault on (+)
1 2 3 t
(-)
Fault on (-)
(+)
1 2 3 NB : The fault current is pulsed at low speeds and is very close to pure DC current at high speeds. (+) K/ Stationary battery charger for: c DC auxiliary network c UPS M _
(-) L
Id
+ Fault on (+)
1 2 3 N t
(-)
NB: In this diagram, the smoothing reactor (L) causes conduction (cyclic and in pairs) of the thyristors such that the fault point (+) or (-) is always electrically connected to the neutral, resulting in a virtually pure DC fault current.
Fault on (-)
Fig. 39 : form of fault currents detected on the three-phase supply of rectifiers when an insulation fault occurs on their output.
However, should a fault occur on an AC outgoer B, the current of this fault is equal to i1 + i2, and there is no certainty that the RCD placed on this outgoer, if it is of the AC type, will trip at the displayed threshold. For this reason standard C 15-100, paragraph 532-2-1-4 stipulates: When electrical devices likely to produce DC currents are installed downstream of an RCD, precautions must be taken so that, should an
earth fault occur, the DC currents do not affect operation of the RCDs and do not jeopardise safety. It is thus advisable to: c choose the right RCD placed just upstream of a rectifier system, c if necessary, use A type RCDs in the remainder of the installation.
+ 3 2 1 Db Da (A) N 311 V (B) i1 Ru i2 Ru
Fig. 41 : the current of a latched fault at the rectifier output (non-opening of Da) may blind the RCD placed on B.
6 Conclusion
At a time when electricity, as an energy source, is playing an increasingly dominant role in housing, tertiary and industry, it is useful to point out and quantify the electrical hazard and to further knowledge of Residual Current Devices. These devices, like any others, have their strong and weak points. Not yet fully perfected, they play an increasingly important role in the protection of people and equipment. All industrialised countries make extensive use of RCDs, with a variety of earthing systems, both in industry and housing. Generally speaking, the following information is important for installation standards and practices: c For protection of people against the indirect contact risk, an RCD is: v compulsory in the TT system, v necessary in the IT system if there are several earthing connections,
v to be provided in the case of very long outgoers in the TN and IT systems. c For protection of people against direct contact risk, an RCD is very useful and often stipulated by standards as an additional precaution irrespective of the earthing system. c RCDs also provide protection against: v fires of electrical origin, v destruction of machines in the TN system, v electromagnetic disturbances in the TN-S system (neutral insulation monitoring). Present day RCDs comply with construction standards (see chapter 4) and continue to progress in terms of reliability and immunity to interference phenomena which are not ascribable to insulation faults. The purpose of this study is to further knowledge of residual current devices and thereby contribute to the safety of us all.
Bibliography
Standards From 1997 onwards the new publications, issues, versions and IEC amendments to existing publications have a designation in the 60000 series. We would like to draw the users attention to the fact that the former publications printed before 1997 continue to bear the old numbers on the printed copies, while waiting to be revised. Product standards c IEC 60479: Guide to the effects of current passing through the human body. c IEC 60755: General rules concerning residual current protection devices. c IEC 60947-2: Low voltage switchgear - Part 2: Circuit-breakers. c IEC 61008, NF C 61-150 and 151: Automatic residual current circuit-breakers for domestic and similar purposes. c IEC 61009, NF C 61-440 and 441: Circuitbreakers for domestic and similar purposes. c IEC 61557-6, NF EN 61557-6: Electrical safety in low voltage distribution systems up to 1000 V AC and 1500 V DC - Part 6: Residual current devices (RCD) in TT and TN systems. c UTE C 60-130: Residual current protection devices. c NF C 61-420: Small residual current devices. c NF C 62-411: Connection and similar equipment, residual current devices for first category installation monitoring switchboards. c Draft standard: earth leakage protection socket.
Installation standards c IEC 60364, NF C 15-100: LV electrical installations c UTE C 15-401: practical guide, installation of thermal motor/generator sets c UTE C 15-402: practical guide, static uninterruptible power supplies (UPS). Schneider Electric Cahiers Techniques c Protection of people and uninterruptible power supplies J.-N. FIORINA, Cahier Technique no. 129 c Evolution of LV circuit-breakers with standard IEC 60947-2 E. BLANC, Cahier Technique no. 150 c Earthing layouts in LV B. LACROIX and R. CALVAS, Cahier Technique no. 172 c Earthing systems worldwide and evolutions B. LACROIX and R. CALVAS, Cahier Technique no. 173 c Disturbances of electronic systems and earthing systems R. CALVAS, Cahier Technique no. 177 c Cohabitation of strong and weak currents R. CALVAS and J. DELABALLE, Cahier Technique no. 187 Other publications The Schneider Electric guide to the LV electrical installation Editor: CITEF S.A.
Schneider Electric
Direction Scientifique et Technique, Service Communication Technique F-38050 Grenoble cedex 9 Fax: (33) 04 76 57 98 60
DTP: AXESS - Saint-Pray (07) Edition: Schneider Electric Printing: Imprimerie du Pont de Claix - Claix - France - 1000 - 100 FF 06-99
063139L
1999 Schneider Electric
Collection Technique ..........................................................................

Electrical disturbances in LV
R. Calvas
"Cahiers Techniques" is a collection of documents intended for engineers and technicians, people in the industry who are looking for more in-depth information in order to complement that given in product catalogues. Furthermore, these "Cahiers Techniques" are often considered as helpful "tools" for training courses. They provide knowledge on new technical and technological developments in the electrotechnical field and electronics. They also provide better understanding of various phenomena observed in electrical installations, systems and equipments. Each "Cahier Technique" provides an in-depth study of a precise subject in the fields of electrical networks, protection devices, monitoring and control and industrial automation systems. The latest publications can be downloaded from the Schneider Electric internet web site. Code: http://www.schneider-electric.com Section: Experts' place Please contact your Schneider Electric representative if you want either a "Cahier Technique" or the list of available titles. The "Cahiers Techniques" collection is part of the Schneider Electrics "Collection technique".
Foreword The author disclaims all responsibility subsequent to incorrect use of information or diagrams reproduced in this document, and cannot be held responsible for any errors or oversights, or for the consequences of using information and diagrams contained in this document. Reproduction of all or part of a "Cahier Technique" is authorised with the prior consent of the Scientific and Technical Division. The statement "Extracted from Schneider Electric "Cahier Technique" no. ....." (please specify) is compulsory.
no. 141
Roland CALVAS ENSERG 1964 graduate (Ecole Nationale Suprieure dElectronique et Radiolectricit de Grenoble: Grenoble Higher Institute of Electronics and Radioelectricity) and graduate from the Institut dAdministration des Entreprises (Corporate Administration Institute), he joined Merlin Gerin in 1966. During his career he has been sales manager, then marketing manager of the activity concerned with protection of people against electrical hazards.
ECT 141 first issue, March 2000
Lexicon
Earthing systems: Standard IEC 60364 (in France standard NF C 15-100) stipulates the three main official earthing systems defining the possible connections of the neutral, source and frames to the earth or neutral. The necessary electrical protection devices are defined according to the earthing system chosen for a network. Electrical disturbance: All disturbing phenomena that cause the mains voltage (or current) wave to depart from its nominal characteristics. Electromagnetic disturbance: All disturbing phenomena of an electrical, magnetic or electrostatic nature that may affect the mains and/or the operation of electric and electronic devices. Impulse voltage: Impulse lasting less than 10 ms superimposed on sinusoidal voltage.
Interruption: Absence of voltage for a varying period of time. Overvoltage: Overshooting Un + 10% for a varying period of time. PWM: Pulse Width Modulation. r.m.s. : root mean square. Short overvoltage: Transient voltage wave characterised by a rapid increase followed by a slower decrease, normally of a periodic nature. Very short overvoltage: Transient voltage wave characterised by a rapid increase followed by a slower decrease, normally of a periodic nature. Voltage dip: Voltage drop of more than 10% followed by a return to the normal variation range.
LV networks are highly polluted and subject to a large number of aggressions. Electrotechnical equipment and in particular electronic devices, that are increasingly numerous and process ever lower signals, must withstand a harsh electromagnetic environment. At the same time availability requirements, whether for production, management or commerce, are increasingly great. To improve the quality of the electricity product and avoid malfunctioning and in some cases destruction of the components of the electrical network and/or sensitive loads, it is essential to understand the origin of the disturbances and evaluate their risks. This is the purpose of this Cahier Technique, which is an introduction to the vast field of electromagnetic compatibility and immunity from 50 Hz through to radioelectric frequencies. In this document the author deals with all parameters and their main anomalies used to characterise an electrical supply: c amplitude: variations interruptions dips overvoltages, c frequency: variations, c waveform: harmonics transients carrier currents, c phase: unbalance, c power: short-circuits overloads (effects on voltage).
Contents
1 LV industrial voltages 2 Amplitude disturbances 3 Sine wave disturbances 1.1 Nominal voltages 1.2 Frequency 2.1 Voltage dips and short interruptions 2.2 Overvoltages 3.1 Generators 3.2 Transformers 3.3 Loads 3.4 Switch mode power supplies 3.5 Industrial furnaces 3.6 Carrier currents 3.7 Effects of disturbances 4 Disturbances specific to multiphase systems 5 Electromagnetic disturbances 6 Conclusion Appendix 1: transmission of overvoltages through transformers Appendix 2: ferroresonance Bibliography 4.1 Current and voltage unbalance 4.2 Phase unbalance 5.1 Fields and radiation 5.2 Electrostatic discharges p. 4 p. 4 p. 5 p. 9 p. 16 p. 16 p. 16 p. 17 p. 17 p. 17 p. 17 p. 19 p. 20 p. 21 p. 22 p. 23 p. 24 p. 25 p. 26
1 LV industrial voltages
All networks undergo or create disturbances. However, loads, as well as control/monitoring and protection devices, are sensitive to the quality of the voltage applied to them. Knowledge and consideration of these electrical power supply requirements are thus justified: c manufacturers of electrotechnical and electronic equipment designed for an international market must allow for tolerances that vary from country to country,
c electrical consultants and contractors must also take these phenomena into account right from the design stage of networks, and, according to their customers needs, find technical solutions for distribution diagrams and switchgear. Important reminder: Low Voltage installations must comply with standard IEC 60364 (in France standard NF C 15-100).
1.1 Nominal voltages

The low voltage range concerns the following voltages: c AC: i 1,000 V (frequency i 10 kHz), c DC: i 1,500 V Also known as rated voltages they are defined in publication IEC 60038 of 1983. The most common LV voltage is 230/400 V in France, made official by the decree of 29/05/86. World-wide and up to 2003, various nominal voltages will continue to be found: c European: 220/380 V, 50 Hz, c English: 240/415, 50 Hz, c USA: 120/240 V and 480 V, 60 Hz, c Japan: 100/200 V, 50 and 60 Hz. However, these voltages vary around nominal voltage according to structure and load of the network (conductor length and load fluctuation) and the regulating systems of the public (or private) network. In Europe, standard EN 50160 states that in normal operation, for each period in a week, 95% of r.m.s. values averaged out over 10 minutes must be included in the range Un 10%. For example, in France, the Utility Company (EDF) announces a fluctuation of 10% but accepts that, temporarily, voltages may depart from nominal voltage by more than 20%. Furthermore, these fluctuations must be associated with the notion of time, e.g.: c Permanent Un 10%. c Five minutes Un 20%. In developing countries, fluctuations in excess of 20% can be frequently observed.
1.2 Frequency
With respect to frequency, slow fluctuations around the nominal value of 50 or 60 Hz are normally small. In European countries, whose networks are interconnected, standard EN 50160 specifies 50 Hz 1% for 95% of the week, and [+4%, - 6%] in the event of major disturbances. These fluctuations may be far greater for autonomous networks, for example on an island.
2 Amplitude disturbances
Networks are disturbed not only by load fluctuations in normal operation, but also and above all by random phenomena of accidental origin, for example short-circuits. A LV network is subjected to these two causes of disturbances
whatever their origin (LV, MV or HV). The two main amplitude disturbances are: c voltage dips and short interruptions, c overvoltages.
2.1 Voltage dips and short interruptions

A voltage dip is normally defined as a drop from 10% to nearly 100% of the nominal voltage value for a time lasting between 10 ms and a few seconds (see fig. 1 ). When power is off (100 % voltage drop), for 1 s to 1 min, we talk of a short interruption, and of a very short interruption between 10 ms and 1 s. When duration is i 10 ms, the interruption is normally due to transient or voluntary phenomena, such as use of dimmers. Disturbances due to operation of HV and MV networks c Deliberate modification of network geometry causes short interruptions of up to several seconds. However meshed distribution networks are less sensitive to this phenomenon. c Energisation of large transformers causes developing voltage dips due to inrush current. The table in figure 2 gives, by way of an example, the energisation current of MV/LV transformers. The same applies on energisation of MV capacitors and large motors: for capacitors the inrush current may reach 100 times their nominal current (as per IEC 60831-1) and up to 8 In for a few dozen seconds for motors. c Disturbances due to faults on HV and MV networks (short-circuit and insulation fault). Phenomena such as: equipment insulation fault, lightning stroke, damage to an underground cable, earthing of an overhead conductor, cause short-circuits that generate high voltage drops in an area of varying size around the origin of the fault, prior to disconnection of the faulty device by the protection devices. At the same time users directly serviced by the faulty device have their power supply permanently or temporarily cut off. These phenomena affect all voltage levels, but particularly the MV networks (20 kV). For each fault, the number and type of voltage dips and short interruptions depend on the protection devices and reclosers used.
Voltage amplitude v
v 10 % < v i 100 %
t (ms)
t t > 10 ms
Interruption
Fig. 1 : voltage dip (v) and interruption.
P (kVA) 50 100 160 250 400 630 800 1000 1250 1600 2000
ne =
Ipeak energis. In transformer
Time constant (s)
15 14 12 12 12 11 10 10 9 9 8
0.1 0.15 0.2 0.22 0.25 0.3 0.3 0.35 0.35 0.4 0.45
Fig. 2 : value of the energising currents and time constant of the MV/LV transformers on energisation on the MV side (on energisation on the LV side, ne is twice as high).
In France v On EHV networks (400 and 225 kV), various reclosers are used: - slow single-phase reclosers tripping in event of single-phase faults: breaking time is around 1.5 to 2 s, - slow three-phase reclosers (multiphase faults), whose operation depends on voltage synchronisation check: breaking time is around 5 s. These reclosers are also used on HV networks (90 and 63 kV). v On overhead MV networks (5 to 20 kV) for which the neutral is earthed by an impedance limiting the earth fault current to 300 A in the primary substation, shunt circuit-breakers eliminate intermittent single-phase faults (contacts with branches) without affecting LV users (see fig. 3 ). However, multiphase faults are treated by rapid and slow reclosers that cause voltage dips on LV (see fig. 4 ). Roughly 75% of faults are eliminated after rapid reclosing that corresponds to the conventional short 300 ms interruption.
v On underground MV networks (5 to 20 kV). In theory there are no intermittent faults, thus all faults result in permanent de-energisation of the faulty part of the network. Consequently underground networks (town) are organised either in double shunt (see fig. 5 ) or, in most cases, in an open ring (see fig. 6 ). Automated systems restore supply to fault-free HV/LV substations in less than one minute.
300 A
Shunt circuit-beaker
Fig. 3 : elimination of a single-phase fault by a shunt circuit-breaker which short-circuits the fault, thus quenching the arc.
a - Current inrush by the faulty feeder HV
In
0
t (s)
MV b - Voltage on the faulty feeder Un 0
t (s) Faulty feeder
c - Voltage on the other feeders Un 0 0.1 Origin of fault 0.3 0.5 Rapid reclosing 15 0.5 Slow reclosing
t (s)
Permanent
Circuit-breaker opening
Principle: On an overhead MV network, most faults are intermittent. Thus, the incoming circuit-breaker on each feeder is equipped with an automated system that attempts two reclosings (sometimes three): c the first instantaneously (opening + closing time: 0.3 s), c the second after roughly 15 s.
Fig. 4 : voltage fluctuations on rapid and slow reclosing on a MV distribution network.
HV/MV substation
HV/MV substation
Circuit A Circuit B
U Incomer A Incomer B Network supplied by 1 0 1 0 1 0 A B A t tf tf
tf = 5 or 25 s
Fig. 5 : double shunt MV power supply. In the box, automated system sequence for a Merlin Gerin changeover switch conform to EDF specifications (France).
HV/MV transformers
HV/MV substation
Supply path of all substations except no. 1
Supply path of substation no. 1
MV/LV transformers
MV/LV transformers Ring opening point Substation no. 1
MV/LV transformers
Fig. 6 : open ring MV power supply (main feeder interruption).
Disturbances due to operation of LV networks The voltage dips observed on LV networks, due to their operation, are caused by: c activation of capacitors (possibly placed at the supply end of the network) that causes, just as in MV, a very short voltage dip or a transient phenomenon of the oscillating kind; c direct on line starting of large motors, directly supplied by the main LV board, that causes a voltage dip or whose catching a spinning load (resumption of supply after a short interruption) results in an inrush current that may be high (up to twice the 8 to 10 In of an across the line start). Disturbances due to faults on LV networks Short-circuits on the main LV distribution, between two live conductors or between a live conductor and a frame in the TN earthing system (see Cahier Technique no. 172) are also responsible for voltage dips and interruptions. These short-circuits eliminated by circuitbreakers, selective or otherwise, cause a voltage dip that is particularly deep if the faulty feeder is important for source power, and that can last from 50 to 500 ms. The shape of some voltage dips and interruptions is given in the table in figure 7 . Note that LV networks backed up by rotating sets and source changeover switches (main/ standby) may not be powered for several seconds (see Cahier Technique no. 161): Breaking time = T1 + Td + T, where: T1 = unit starting time delay (0.4 to 10 s), Td = unit starting time (a few seconds), T = changeover time (< 0.4 s). Orders of magnitude of voltage dips according to various switching types are given in the table in figure 8 . Conclusion concerning voltage dips and short interruptions Statistically we observe interruptions and voltage dips, at the supply end of a LV network, at a frequency of once a month to 10 times a day. They are mainly found in medium and high voltage, and in 80% of cases last between 20 and 350 ms. Electrotechnical switchgear is normally unaffected by these phenomena, except for contactors and relays whose nuisance tripping and switching may be a source of major problems in industrial processes. With the growth of distributed intelligence and micro-PLCs, sensors, intelligent pre-actuators, etc. great attention should be paid to equipment immunity to voltage dips and supply interruptions. As a matter of fact, switchgear using electronic components must not give incorrect indications or untimely orders.
c Voluntary source switching Duration: a few milliseconds to several minutes. Un
c Automatic source switching further to an external fault Duration: a few milliseconds to 10 s. Un
c External short-circuit Duration: 10 ms to more than 100 ms. Un
c Energisation of a large unit Duration: depends on network power. Un
c Short interruptions with large motor and catching a spinning load. Un
Fig. 7 : main types of interruptions and voltage dips.
Type of switching Synchronous Idle time LV Application examples c Busbar switching c Substitution of one generator for another c Switching between a UPS and the mains None c Coupler c Synchrocoupler c UPS assembly with static switch Switching must be performed before all power is cut to the main source c Pump supply c Supply of auxiliary circuits of a transformer substation c Supply of hypermarkets c Etc. 0.5 to 10 s c Automatic source changeover switch with circuit-breakers MV c Supply by two switchable MV incomers c Supply by one main source and one replacement source 1 to 30 s c Set of MV cubicles with changeover switch c Catching a spinning load of asynchronous motors Pseudosynchronous
Switching duration Devices used
0.06 to 0.3 s c Rapid HV circuit-breakers associated with an ultrarapid changeover switch and a phase controller Problems in preparing the switching order (residual voltage present)
Observations
The changeover devices mentioned above are normally factory-assembled, cabled and adjusted assemblies (see for example the Schneider Electric catalogue)
Fig. 8 : various types of switching and their duration.
2.2 Overvoltages
Except for accidental voltage rises, for example connection of a 230 V transformer to a 400 V line, the only voltage rises that may exceed the nominal variation range of Un + 10% are observed when the LV network is not on-load. However, the transformers of primary substations are normally equipped with on-load tap changers that enable the electrical utility to maintain MV within the range Un 7%. Impulse overvoltages are mainly due to switching operations on the network and to lightning. While without major risk for conventional LV switchgear built to manufacturing standards, they can affect devices containing electronic components (voltage withstand reduced to 1,500 V). Switching on the MV network c Energisation of MV capacitor banks without a blocking reactor: peak value may reach 2 Un (see fig. 9 ). c De-energisation of MV/LV transformers: peak value may reach 3 Un: it considerably depends on the current chopped by the MV breaking device (see fig. 10 ). These overvoltages are transmitted to the LV by the MV/LV transformer according to its inductive and capacitive ratios (see appendix 2). They must not be ignored as they represent a risk for LV switchgear particularly when the LV circuit is weakly capacitive, i.e. small. This is particularly the case on commissioning of an installation.
Un
1 0 -1 -2 Un
Energisation
t -20 (ms)
Fig. 9 : voltage wave on energisation of a MV capacitor bank.
Un 1 t
-1
Fig. 10 : voltage wave on tripping of a MV/LV transformer.
Lightning Lightning directly affects overhead networks. Many studies deal with this subject. Not all regions are equally exposed: we talk of lightning density level. Likewise, not all installations are equally exposed. Extreme exposure applies to very long overhead lines in regions that are particularly affected by storms. Figure 11 shows the number and level of overvoltages that can be expected to be observed over one year. These are statistical values and there is always the risk of a direct lightning stroke of a far higher value. Standards concerning insulation co-ordination and electromagnetic compatibility (EMC) take account of this natural phenomenon that develops a considerable impulse energy transmitted to LV networks by the MV/LV transformers. Switchgear standards increasingly make provision for withstand to voltage and current waves representative of lightning. In addition to these steep front waves, lightning has a perverse effect due to the spark-gaps and surge arresters placed upstream of MV/LV substations.
The importance of the risk depends on the earthing system (see standard NF C 13-100). Take the case of the TTN earthing system (see fig. 12 ). When lightning strikes the MV line, it causes arcing in the spark-gap or surge arrester, which lets through part of the impulse wave, that may be amplified by the wave reflection phenomenon (see Cahier Technique no. 168). This spark-gap arcing results in a 50 Hz current, from the faulty phase to the Rpn earthing connection. This current, limited by the Z impedance (at 300 A in France) is maintained until the protection devices in the primary substation trip. When it flows through Rpn it results in a potential increase of the entire LV network with respect to the deep earth to which the LV equipment is connected by Ra. There is then a risk of return breakdown of the LV equipment which is concretised when the Current x Rpn product is greater than insulation voltage with respect to the equipment frame. Thus, with a 300 A current and a 2500 V insulation voltage, Rpn must not exceed 8 W, a value that can be exceeded according to the hydraulicity and nature of the land! Note that the problem is the same for an earth fault of an MV device in the substation. In this case, the underground MV networks present the greatest risk as this current is either limited to a higher value (1000 A in France) or not limited as is the case in some countries. The 2 U + 1000 test at 50 Hz for 1 min enables the voltage withstand of LV equipment to be checked.
Number of overvoltages a year (without protection) High exposure 103 102 10 1 10-1 10-2 Low exposure 0.5 1 2 5 10 20 Overvoltage levels (kV) Device arcing Extreme exposure
0.2
Fig. 11 : number and value of overvoltages statistically possible according to the level of exposure to lightning.
Finally, the substation earthing system plays an important role with respect to transmission of the impulse wave to the LV. As a matter of fact, in addition to the overvoltage due to capacitive transmission through the transformer, the lightning current wave develops a voltage Ul in the earthing connection impedance, as high as the earthing connection is inductive (see fig. 12).
Ul
N Z Spark-gap Ul
Rpn Primary substation
Ul
MV/LV substation
Ra Load
Fig. 12 : TTN earthing system of a MV/LV substation (common earthing of the spark-gap, transformer frame and LV neutral), and transmission of the lightning wave to the LV network.
To allow for this phenomenon, pole-top substations in France are installed (see fig. 13 ) with separate substation, neutral and LV load earthing connections (this is known as a separate earthing connection diagram, TTS as per NF C 13-103). This arrangement avoids potential increase of the LV installation and thus prevents the risk of return breakdown of switchgear and LV loads (see fig. 14 ). Their manufacturing standards mean that MV equipment can withstand these overvoltages (125 kV impulse wave withstand for a device with a specified voltage of 24 kV). On an HV/LV breakdown in the transformer, it is the value of Rn or Rpn that prevents return breakdowns in the LV equipment.
Readings, mainly made in Switzerland by Landis & Gyr, and mainly published by the I.E.E.E., show that transient overvoltages in excess of 1 kV are observed several times a year at various points of a LV distribution network (see fig. 15 ). It is relatively exceptional to detect overvoltages greater than 6 or 7 kV as, at these levels, arcing occurs in the air or in the cabling that acts as a spark-gap. Such arcing or dielectric breakdown clips overvoltages to a level that is particularly low for a careless wiring. Such a protection damages the installation and may in time lead to serious safety problems by piercing, bypassing and carbonising of insulators. The solution for exposed sites is to use surge arresters between the live conductors and the earth. For more details refer to Cahier Technique no. 179.
MV/LV transformer
LV pole-top protection
Metering
Individual LV protection
Rp
Rn
Ra
Fig. 13 : MV/LV pole-top substation. TTS earthing system of a MV/LV substation (separate earthing of the sparkgap, transformer frame and LV neutral). This earthing system is ITS if a surge limiter is inserted in the LV neutral earthing connection.
Number of transients a year Home (living room on upper floor) 2,000 1,000 Landis & Gyr plant (laboratory) Bank in Basel (service entrance) Farm supplied by overhead lines
Rp
Rn
Ra
Rp i 30 as the substation equipment (pole-top transformer and circuit-breaker) is overinsulated at 20 kV impulse. Rn i 4 to cover the risk of return breakdown of the consumers LV equipment further to HV/LV breakdown, this value results from the following calculation: Rn i Rn i 2 x U + 1,000 - U 300 A 2 x 220 + 1,000 - 220 =4 300
200 100 Landis & 2 Gyr plant 1 (furnace room) 0.2 Home 0.1 (underground incoming supply) 0.02 0.01 0.1 0.3 0.7 1 0.2 0.5 20 10
Composite curve in USA 120 V network kV 23 5 7 10 20
Fig. 14 : earthing connections of a MV/LV substation supplied by overhead lines.
Fig. 15 : frequencies and peak values of overvoltages due to lightning on LV installations (source: Landis & Gyr).
Switchgear operations and operation of the LV protection devices c Relay and contactor coils have an inductive impedance. Breaking of an inductive current nearly always results in high overvoltages. In the case of a relay (see fig. 16 ), when the current is suddenly broken by the switch K, the energy 1/2 Li2 is converted into energy 1/2 CV2, hence an overvoltage:
L 101 = 0.15 = 3, 000 V C 0.25 x 109 The overvoltages observed may reach nearly 10 kV on a 230 V circuit. Interference must be suppressed on these control auxiliaries by their manufacturers, or a RC network or a varistor, connected in parallel on their coil, enables overvoltage to be absorbed. V = i
breaking devices, means that this phenomenon is fairly frequent. It is even observed on shortcircuit current breaking, when at the end of breaking the current becomes sufficiently small to be chopped. c Breaking a short-circuit current can also present a risk of overvoltage as: v rapid action must be taken, before the current reaches prospective short-circuit current, v and insert in the circuit an arcing voltage that is not always easy to control according to the breaking conditions, as this voltage constitutes an overvoltage of several hundred volts for circuit-breakers and sometimes several kV for fuses (see fig. 18 ).
c On LV and HV networks, no-load or only on minimum load, breaking a low inductive current is not easy and results in creation of overvoltages that are frequently multiple (wave trains). On the single-phase circuit in figure 17, taken as an example, when switch K is opened, current continues to flow in L2 with, at the terminals of C2, the voltage VBN:
VBN = i L2 sin 2 t C2
i L = 100 mH R = 180 Cinterference = 250 pF i = 150 mA Cinterference R
Fig. 16 : equivalent diagram of a network.
The same applies to the upstream network with at the terminals of C1 the voltage:
VAN = i L1 sin1t C1
e
L1
L2
C1 N
C2
On opening, voltages VBN and VAN evolve in opposition, and voltage VAB quickly becomes high enough to cause arc re-arcing (re-arcing between the contacts of breaking device K). The current to be broken is small, the contacts move apart, and the interference between the two voltages VBN and VAN are elements that extinguish the arc. Then the voltages begin to rise again and the phenomenon is repeated Several cycles known as re-arcing can thus occur until the energy lost by the arc and the increasing distance between contacts interrupt the phenomenon. The result is the appearance of wave trains, of high voltage and frequency ( =
L1, C1 = inductance and capacitance of the upstream network L2, C2 = inductance and capacitance of the downstream network
Fig. 17 : equivalent diagram of a single-phase circuit.
C32L circuitbreaker Prospective Isc (kA) 11 Cos U (V) Overvoltage (V) U + U 0.25 380 + 10 % 400 to 650
GL fuse 20A 4 0.3 380 + 10 % 2,100 to 2,650
1 LC
),
that are genuine interference generators. The presence of different no-load circuits with reactor or transformer, and of slow opening
Fig. 18 : overvoltages recorded with two protection devices tested on the same test installation.
With respect to this last point, note that: v the closer the short-circuit occurs to the downstream terminals of the breaking device, the higher and shorter the overvoltage. Thus the results presented in figure 19 were read for a fault observed on a 1.5 mm2 diameter cable downstream of a 10 A fuse: where L = 0
where L = 20 m
U + U 1700 + 220 2 = = 6.4 and T = 50 s U 220 2
U + U = 2 and T = 200 s U v generally, the higher the feeder protection rating (and thus the greater the cross-section of its conductors) with respect to the short-circuit power of the busbar, the longer and lower the overvoltage. Figure 20 shows a graphic summary of several tests.
T Fuse U/2 L
U 10
t ms
Fig. 19 : overvoltage depends on the distance separating the fault from the protection device.
+ 10
10 A
3 Fuse 2 400 A 35 A
1 0.03
t 0.1 0.3 1 3 ms
Fig. 20 : with the same upstream short-circuit power, overvoltage levels and durations on a feeder depend on the rating of its protection device and the cross-section of its conductors.
Ferroresonance This phenomenon may occur on no-load and slightly loaded networks, particularly in high voltage, as well as behind small transformers and in the presence of filters. Readers wishing to know more can consult appendix 3 and read Cahier Technique no. 190. Overvoltage tests and solutions The impulse overvoltages transmitted by conduction, with an inductive or capacitive effect,
are very dangerous in particular for electronic devices. In this area, experience has enabled tests to be defined. c Standardised waves for equipment testing These are mostly (see fig. 21 ) dealt with in the IEC 61000 publications, particularly 61000-4-4, 61000-4-5 and 61000-4-12 (see appendix 3). In addition to waveforms, standards stipulate levels corresponding to conditions of use and to test voltages for equipment manufacturers. For example, figure 22 gives, for the various
Origin Switching
Wave type Damped long wave: 0.25 / 2.5 ms
Test wave a
v 50% t ms
0.25
Recurrent damped oscillating wave: (100 kHz < f < 1 MHz) b
2.5
v U 0.9 U
10
0.1 U 0.5
t ms
Lightning
Recurrent impulse voltage wave: 1.2 / 50 s
50 % t
1.2
Current wave 8 / 20 s d
50
50 % t 8
LV switchgear, (e.g. fuse blowing) 5 / 50 ns wave e v
20
50 % t 5 50 ns
Fig. 21 : the main standardised test waveforms.
levels, the voltages for the impulse voltage wave test (see curve d on figure 21). c Overvoltage solutions These are applied both when equipment is implemented and chosen: v on installation, it is advisable to remove overvoltage sources away from the most sensitive items of equipment and to monitor equipment equipotentiality; v when choosing, it may prove useful to provide: - electrostatic shielding or capacitors (see appendix 3), - screen transformers, - filters, - surge limiters (see fig. 23 ). Low voltage surge limiters are also available designed for installations with an IT earthing system (with unearthed neutral). Defined in France by standard NF C 61-740, they are intended to prevent all voltage rises due to a MV fault: Cahier Technique no. 179 describes their implementation.
Level Test voltage 1 % (kV)
1 2 3 4 x
0,5
1 2 4 Special
Level x is open: it must be defined for special equipment specifications.
Fig. 22 : test voltages according to standard IEC 61000-4-5.
Limited U (kV) Varistors Rectifier diode 14 to 1,200 5 to 400
I max (kA)
(8/20 wave) 10 to 5,000 10 to 1,000
Fig. 23 : characteristics of the various surge limiter types.
3 Sine wave disturbances
Voltages (waves) observed, permanently or temporarily, on networks are never completely sinusoidal. This is because: c generators are more or less perfect,
c transformers are also imperfect, c loads generate or absorb harmonics, c the network conveys various signals (carrier currents).
3.1 Generators
Inexpensive devices, normally low power, are often harmonic-generators (see fig. 24 ).
Generators (presence of a slot noise) Inverters (UPS without filtering)
3.2 Transformers
When they are calculated improperly, they begin to saturate as soon as voltage exceeds their nominal value. They then absorb a nonsinusoidal current (see fig. 25 ).
Fig. 24 : example of a voltage wave delivered by inexpensive generators.
3.3 Loads
Capacitors Harmonic currents are particularly well absorbed by capacitors when their frequency is high. The current absorbed by the capacitors is thus a caricature of the voltage of the network on which they are placed. Associated with inductances, they form network cleaning filters and then absorb a non-sinusoidal current (see Cahier Technique no. 152). Fluorescent tubes Electric arcs absorb non-sinusoidal currents particularly harmonic-loaded. Tandem mounting and compensation do not prevent fluorescent tubes from generating harmonics. Phase control rectifiers In most cases, they are connected to an inductive circuit and successively absorb a rectangular current on each phase. At each semiconductor switching, the sine wave of the supply voltage is briefly interrupted. Figure 26 provides a representation of the voltage and current waves.
Ie
Ue
Fig. 25 : off-load current wave absorbed by a transformer with an insufficiently large power rating.
Fig. 26 : voltage and current waves distorted by a phase control rectifier.
As regards phase controlled heating dimmers, each thyristor firing results in a supply voltage dip due to its inductive component followed by a slight voltage drop linked to the internal resistance of the power supply (see fig. 27 ).
3.4 Switch mode power supplies

Very common (microcomputers, TV sets, electrical household appliances, etc.), they use at the input circuit a diode bridge followed by a filtering capacitor that generates high harmonic currents.
Fig. 27 : voltage and current waves distorted by a heating dimmer.
3.5 Industrial furnaces

These are real network polluters: c arc furnaces produce a genuine electric noise containing all the frequencies, whose amplitude decreases with frequency, c HF induction furnaces are narrow-band interference generators centred on a few dozen kilohertz.
3.6 Carrier currents

These are additional signals that are deliberately injected on the network by the electrical utility, for example to remotely control Day/Night tariff changes, such as the Pulsadix system (175 Hz) used in France. These signals are audible with a car radio when the car passes under an HV line. Their frequency varies from country to country: 110 Hz, 175 Hz, 183 Hz, 217 Hz, 283 Hz, 317 Hz, 600 Hz, 1050 Hz, 1350 Hz. The signals are made up of pulse trains of the chosen frequency, injected according to a certain code. Their voltage is around 10 V. Other signals conveyed by the network can also disturb the mains at a LV consumer: c digital communications, such as those used for Building Management Systems (BMS), c intercom transmissions (a few dozen kilohertz), c remote control orders (50 to 150 kHz).
3.7 Effects of disturbances

All periodic signals, whatever their waveform, can be considered as the sum of alternating signals of a frequency equal to that of the signal, The fundamental, and multiples of that of the signal, The harmonics. The harmonics and the special frequencies injected on the network produce harmful and in some cases dangerous effects. Thus harmonics: c increase iron losses and eddy currents in motors and transformers, c jeopardise capacitor bank service life, whose long time protection has to take harmonic currents into account, c jeopardise service life of filtering capacitors of electronic circuit power supplies, c can cause temperature rise of the neutral conductor (3rd order harmonic currents and multiples generated by switch mode power supplies and fluorescent tubes, are added up and return to the source by the neutral). To avoid these risks, long time or overload overcurrent electronic protection devices (equivalent to electromechanical releases said to be thermal) must contain a circuit measuring the total root mean square value of the current.
Standard EN 50160 indicates, for public distribution networks, that for each period of a week, 95% of r.m.s values of each harmonic voltage averaged out over 10 minutes must not exceed the values stated in the table in figure 28 . Higher voltages for a given harmonic can be due to resonance.
Moreover, the total harmonic distortion provided (including all harmonics up to the 40th order) must not exceed 8%. To reduce harmonic currents and voltages, passive filters or active compensators can be used (see Cahier Technique no. 183).
Odd harmonics Multiples of 3 Order 3 9 15 21 Relative voltage (% Un) 5 1.5 0.5 0.5 Non multiples of 3 Order Relative voltage (% Un) 5 7 11 13 17 19 23 6 5 3.5 3 2 1.5 1.5
Even harmonics
Order 2 4
Relative voltage (% Un) 2 1
6 to 24 0.5
Fig. 28 : values of harmonic voltages at the supply points (source: EN 50160).
4 Disturbances specific to multiphase systems
4.1 Current and voltage unbalance

Electrical networks are normally three-phase. They supply three-phase loads but also a large number of single-phase loads. The currents absorbed on the three phases are thus of different amplitude, which results in voltage unbalances. These voltage unbalances generate negative sequence components (see fig. 29 ) that mainly lead to unwanted braking torques and temperature rises in AC motors. In Europe, standard EN 50160 specifies In normal operating conditions, for each period of a week, 95% of rms values calculated over 10 minutes of the negative sequence component of the supply voltage must fall between 0 and 2% of the direct component. In some regions, unbalance may reach 3%. In France, EDF estimates that for LV networks
Vi ) Vd although fluctuating for the most part, is between 0.5 and 2% on average. The highest ratios are observed on LV networks supplied by a low power transformer with a majority of single-phase connections This is the case of pole-top substations that supply few LV consumers with single-phase. In difficult cases, Zig-Zag connection is a means of attenuating unbalance for the upstream network.
(at the customers), the unbalance ratio (
Direct component V3 Vd3
Negative component Vi2
Zero-sequence component t
=
V2 t V1 Vd2 Vd1
+
Vi3 Vi1
+
Vh
3Vd = V1 + aV2 + a2V3
3Vi = V1 + aV2 + aV3
3Vh = Vi + V2 + V3
Fig. 29 : reminder of the various components, direct, negative and zero-sequence of the three-phase system.
4.2 Phase unbalance

Besides the effects of negative sequence components on motors, unbalance of threephase voltages disturbs the operation of phase control thyristor devices (see fig. 30 ). Furthermore, harmonics, which modify the zero passage of voltages, can interfere with thyristor ignition synchronisation. As concerns protection against electrical hazards, this type of disturbance is also a nuisance as current unbalance limits use of the Nicholson set-up (pooling four current transformers) to detect residual currents due to an insulation fault (see Cahier Technique no. 114).
(+)
Ve V+/-
Vs
(-) t
Fig. 30 : effect of a voltage unbalance on a rectifier in semi-controlled Graetz bridge.
5 Electromagnetic disturbances
5.1 Fields and radiations

The LV network disturbances dealt with in the earlier chapters are mainly of the conduction type, i.e. transmitted by conductors, transformers, stray capacitances, etc. Electrical engineers know that a device placed near a conductor conveying a large 50 Hz current will collect an electromotive force by inductive coupling. This is true for all circuits placed in the magnetic field (see fig. 31 ). They also know that a device placed near an HV network will collect an induced voltage: this is true for all conductors placed in the electric field (see fig. 32 ). A high current is said to radiate a low impedance magnetic field, and a high voltage to radiate a high impedance electric field. In point of fact, field impedance is equal to:
E H In fact, all electromagnetic fields are characterised by the simultaneous presence of electrical and magnetic fields linked in what is known as the Poingting vector: Zc =
I
e H r (S) H=
I
2r
(A/m)
(s) =
d e=(Volt) dt
(s) H n ds (Weber)
Fig. 31 : voltage induced by a current.
P (W / m ) = E (V / m) H( A / m)
2
V (V/m) E= D e = E d (Volt)
Impedance Zc depends on distance and frequency (see fig. 33 ). If for 50 Hz we can talk of a field E or H, high or low impedance, this is because:
C = = 1, 000 km 2 2f Around a conductor through which a high current flows, the magnetic field H decreases according to 1 / r (r = radius) up to / 2. Near an antenna (high impedance) it is field E that decreases as a function of distance in 1 / r3. When distance exceeds / 2, the two fields E and H decrease according to 1 / r and impedance is a constant that, in air, is equal to 377 W. When transmission of interference is, for example, due to fuse blowing, as wave rising edge is 5 ns, the equivalent frequency is 108 Hz. The transition distance is:
D d
Fig. 32 : example of an electric field.
Log Zc
Electric field Z0 = 0
= 377
Flat wave Zc = Z0 Magnetic field /2 Log (r) or (d)
C = = 5 cm , beyond this value the 2 2 f fields E and H decrease in 1 / r; we say we are in a flat wave.
Fig. 33 : example of wave impedance for a bipole.
There is a large diversity of radiated disturbance sources, normally divided into two major families: c those with a narrow spectrum: radio transmitters, radars, induction furnaces, microwaves, etc. c those with a wide spectrum: lightning, electrostatic discharge, breaking devices, commutator motors, welding machines, arc furnaces, etc. Note that some sources such as static converters and lightning are both conducted and radiated interference generators. Electronic circuits are the main victims of electromagnetic disturbances since they process very low voltage signals and have large impedances (they consume less and less). Interference occurs on the victim, mainly through the cables entering and leaving the box (by coupling in common mode differential mode common impedance crosstalk). Also, the paths of printed circuits and components can and must be determined to ensure minimum electromagnetic susceptibility.
The most sensitive functions are: c LF amplifiers (low-pass) c Input circuits (integrators) c Logic circuits (digital). Furthermore, growth of digital communications is responsible for new difficulties as buses create, with the electrical network, loops that are sensitive to the magnetic fields developed by the high electric currents and close lightning strokes. Precautions must therefore be taken for the study and production of installations (see Cahier Technique no. 187). In LV, immunity tests against radiated fields are defined in standard IEC 61000-4-3 and subsequent sections. Thus, the releases of the Compact circuit-breakers of Merlin Gerin brand are tested according to the maximum degree of severity of standard IEC 60947-2 referring to standards IEC 61000. Finally, to better understand electromagnetic compatibility (EMC), readers can refer to Cahier Technique no. 149.
5.2 Electrostatic discharges

This type of disturbance affects and/or destroys electronic switchgear, but has no impact on electrical networks. In point of fact, electronic components are particularly sensitive to these electrostatic discharges as they use increasingly more low level components with low consumption and high input impedance. Also, a part of these switchgear, e.g. indicator lights, potentiometers, is always accessible to touch. People, depending on their footwear and the floor on which they walk, plus air humidity, can be charged with a potential of up to 15 kV (phenomenon modelled and represented on figure 34 ). In this case the discharge current can reach several dozen amperes (see fig. 35 )! It is thus necessary to immunise and test electronic apparatus for these discharges. This subject is described in standard IEC 61000-4-2.
I
%I 90
Where U = 15 kV, which corresponds to class 4 in standard IEC 61000-4, I = 70 A in event of breakdown.
150 2
50
150 pF
2
10 tm
Fig. 34 : modelling the electrostatic load circuit of a person.
Fig. 35 : electrostatic discharge current (tm = 5 ns, T = 30 ms).
6 Conclusion
LV networks are very polluted as they are subjected to numerous aggressions of varying origins: external (sources, MV networks, etc.), operation (elimination of faults, control by carrier current, etc.), applications (furnaces, electronic loads, etc.). At the same time, quality requirements of the electricity product as well as operating dependability of electrical and electronic equipment, whether for production, management or commerce, are increasingly great. c Users, aware of malfunctions and operating losses, are seeking to control their energy supply. They are increasingly measuring conventional values (voltage, current, power, cos ) and recording overvoltages, dips and power cuts, etc. For example they also want to know: v total harmonic distortion, v harmonic level, v peak factor, v voltage unbalance ratio. c Electrical utilities have to guarantee a certain level of quality of the electricity product they deliver (see appendix 3: standard EN 50160). c Electrical and electronic switchgear manufacturers are obliged to comply with the European directive on EMC as from design of their products that they then test according to the
specifications of standards such as IEC 61000 (see appendix 3). However they are increasingly offering new products able to meet users expectations with respect to follow-up of energy quality and distribution reliability, for example: v control units placed in circuit-breakers are used to take all measurements (I, F, P, cos , etc.), v switchboard centres are now used to know instantaneous, maximum and average reactive and apparent energies and powers over a specific period. They can also measure total harmonic distortion, the peak factor and phase unbalances (voltage and current). c Professional electricians (ranging from designers to cable layers) must study the main probable disturbances in order to find the most suitable solutions. For example designers have to provide emergency sources, Uninterruptible Power Supplies (UPS), surge arresters, filters and active harmonic compensators. In this sense, for all people concerned with quality and availability of electrical power throughout the distribution process, this Cahier Technique forms an initial tool in so far as it explains the various disturbances that can affect electrical networks and equipment. Its purpose is to increase knowledge of these disturbances in order to combat them more effectively.
Appendix 1: transmission of overvoltages through transformers
Magnetic transfer For an MV/LV transformer, no-load or on minimum load, lightning and switching overvoltages are transmitted to the secondary according to the ratio to which a corrective coefficient is applied. This coefficient is normally less than 1.3 for the 1.2/50 s lightning wave, and only rarely exceeds 1.8 for switching overvoltages according to the 250 / 2,500 s wave specified by IEC 60060-2. For example in the case of a 20 kV/410 V-Dyn 11 transformer, its ratio has the following value:
20, 000 = 48.8 410 c a 1.2/50 s wave of 125 kV applied on its primary winding gives its secondary a similar waveform of an amplitude equal to:
HV
LV
Fig. 36 : position of the windings on a MV/LV transformer column.
125 kV x
1 x 1.3 = 3.3 kV peak 48.8

Ue
c a 250 / 2,500 s wave of 90 kV gives:

90 kV x 1 x 1.8 = 3.3 kV peak . 48.8
CHV
CLV
Us
Capacitive transfer This depends on the arrangement of the primary and secondary windings with respect to the transformer frame. Normally for each column, the LV winding is close to the core and the MV winding is relatively far away, since it is placed around the LV winding (see fig. 36 ). The overvoltage transmitted to the secondary, by the various stray capacitances of the transformer (see fig. 37 ) is calculated by the formula: CLV / Frame Us = Ue CHV / Frame + CLV / Frame In the worst possible case, according to IEC 60071-2, it can reach 0.7 Ue. In reality, the transmission coefficient is around 0.3 for a MV/LV transformer with no connection
Fig. 37 : diagram of the capacitive transfers of overvoltages in a MV/LV transformer.
to the secondary. The presence of capacitances between the live conductors and the frame at the LV main switchboard considerably reduces these overvoltages. It should be borne in mind that to avoid breakdown in the LV part of the overhead supplied substations, the switchgear installed must have reinforced insulation (10 kV / 50 Hz) and that it is strongly recommended to install surge arresters in the main LV board. Furthermore, if this board contains control and monitoring auxiliaries, they must be supplied by means of a shield LV/LV transformer.
Appendix 2: ferroresonance
In theory The serial (and parallel) resonance phenomenon is well known. When =
1 LC and C are identical, opposed and in absolute value far higher than the voltage e (see fig. 38 ). The serial (and parallel) ferroresonance phenomenon is a special case of the above that occurs when the reactor is iron and thus saturable. If we plot the development of the absolute values of the various voltages for a specific frequency as a function of current (see fig. 39 ), two stable operating points A and B are observed on the curves. To move from A to B, all that is required is for a transient (interference) to temporarily increase the value of e and exceed the voltage limit at point E.
Note that, as long as the curves
and Li intersect, there is a risk of
, the voltages developed by L
ferroresonance, hence: c Ferroresonance can occur in a wide range of frequencies, including 50 Hz and its harmonics. In this case, the voltage wave at 50 Hz is overmodulated by a high voltage at 150 Hz for example, c To be certain of avoiding this,
greater than Lo (Lo = L with low induction) or the circuit must always be loaded (damped). In conclusion LV equipment designers must take the risk of ferroresonance into account. This risk is present on no-load networks with respect to: c the dielectric phase/frame withstand of switchgear (u 8 to 10 Un), c the supply voltage of very low power loads (e.g. own-voltage electronic protection devices). For more details on this phenomenon consult Cahier Technique no. 190.
must be
In practice The voltages developed in L and C are then very high. There is a risk of dielectric breakdown and a high risk of a low power load parallelconnected on C being destroyed.
U i C Li
C
i - Li C E
0 e i/JC
Overvoltage i/JC
e A
Fig. 38 : a resonant circuit.
Fig. 39 : principle of ferroresonance.
Bibliography
Standards c EN 50160: Voltage characteristics of electricity supplied by public distribution systems. This standard specifies in MV and in LV: v Frequency, v Amplitude of the supplied voltage, v Variation in the supplied voltage, v Rapid voltage variation, v Voltage dips, v Short interruptions in the supplied voltage, v Long interruptions in the supplied voltage, v Temporary overvoltages on the network between phase and earth, v Transient overvoltages between phase and earth, v Unbalance of the supplied voltage, v Harmonic voltages, v Interharmonic voltages, v Signalling on the network. c IEC 60038: IEC standard voltages. c IEC 60060-2: High-voltage test techniques Part 2: Measuring systems. c IEC 60071-2, NF C 10-102: Insulation coordination Part 2: Application guide. c IEC 60364, NF C 15-100: Electrical installations of buildings. c IEC 60664: Insulation co-ordination for equipment within low-voltage systems. c IEC 61000: Electromagnetic compatibility (EMC). With hereafter a selection of most interesting parts for a first approach: v 1.1: Fundamental definitions and terms. v 2: Environment: - Electromagnetic environment for low-frequency conducted disturbances and signalling in public power supply systems. - Compatibility levels for low-frequency conducted disturbances and signalling in public low-voltage power supply systems. - Radiated and non-network-frequency-related conducted phenomena. - Low frequency magnetic fields in various environments. v 3: Limits: - Limits for harmonic current emissions (equipment input current 16 A per phase). - Limitation of voltage fluctuations and flicker in low-voltage supply systems for equipment with rated current 16 A. - Signalling on low-voltage electrical installations.
v 4: Testing and measurement techniques: - Electrostatic discharge immunity test. - Radiated, radio-frequency, electromagnetic field immunity test. - Electrical fast transient/burst immunity test. - Surge immunity test. - General guide on harmonics and interharmonics measurements. - Power frequency magnetic field immunity test. Basic EMC publication. - Pulse magnetic field immunity test. - Damped oscillatory magnetic field immunity test. - Voltage dips, short interruptions and voltage variations immunity tests. - Oscillatory waves immunity test. - Test for immunity to conducted, common mode disturbances in the frequency range 0 Hz to 150 kHz. NB: Readers wanting to know more about documents related to the standards may consult the internet sites: c For IEC: http://www.iec.com, c For UTE: http://www.ute-fr.com. Schneider Electric Cahiers Techniques c Residual Current Devices in LV R. CALVAS, Cahier Technique no. 114. c EMC: electromagnetic compatibility J. DELABALLE, Cahier Technique no. 149. c Harmonics in industrial networks. P. ROCCIA and N. QUILLON, Cahier Technique no. 152. c Automatic transferring of power supplies in HV and LV networks, G. THOMASSET, Cahier Technique no. 161. c Lightning and HV electrical installations B. DE METZ-NOBLAT, Cahier Technique no. 168. c Earthing systems in LV B. LACROIX, R. CALVAS, Cahier Technique no. 172. c Flicker ou scintillement des sources lumineuses R. WIERDA, Cahier Technique no. 176. c Perturbations des systmes lectroniques et schmas des liaisons la terre R. CALVAS, Cahier Technique no. 177. c LV surges and surge arresters LV insulation co-ordination C. SERAUDIE, Cahier Technique no. 179.
c Cohabitation of high and low currents. R. CALVAS, J. DELABALLE, Cahier Technique no. 187. c Ferroresonance P. FERRACCI, Cahier Technique no. 190. Other publications c Caractristique de la tension dalimentation BT. Rapport UNIPEDE (Union Internationale des Producteurs et Distributeurs dElectricit) 1981. c Qualit de lalimentation lectrique. EDF, 09-90, Cahiers de lingnierie. c Electromagnetic compatibility A. KOUYOUMDJIAN Schneider Training Institute, 07-96.
Schneider Electric
DTP: AXESS (07) Edition: Schneider Electric Printing: Imprimerie du Pont de Claix - Claix - France - 1000 - 100 FF 03-00
086144
cahiers techniques
introduction to dependability design
P. Bonnefoi
144
Pascal Bonnefoi earned his engineering degree ESE in 1985. After working for a year in Operational Research for the French Navy he started his work as a reliability analyst for Merlin Gerin in 1986, in the Reliability studies for which he developed a series of special software packages. He aslo taught courses in this field in the industrial and academic worlds. He is presently working as a software engineer for HANDEL, a Merlin Gerin subsidiary.
MERLIN GERIN service information 38050 Grenoble Cedex France tl. : 76.57.60.60 E/CT 144 December 1990
MERLIN GERIN
la matrise de l'nergie lectrique
GROUPE SCHNEIDER
introduction to dependability design

P. BonnefoiP=.
Table of contents 1. Importance of dependability In housing In services In industry 2. Dependability characteristics Reliability Failure rate Availability Maintainability Safety 3. Dependability characteristics interdependence 4. Types of defects Interrelated quantities Conflicting requirements Time average related quantities Physical defects Design defects Operating errors 5. From component to system: modeling aspects Data bases for system components FMECA method Reliability block diagram Fault trees analysis State graphs 6. Conclusion 7. References and Standards p. 2 p. 2 p. 2 p. 2 p. 2 p. 3 p. 4 p. 4 p. 5 p. 5 p. 6 p. 7 p. 7 p. 7 p. 8 p. 11 p. 11 p. 14 p. 17 p. 19 p. 20
Equipment failures, unavailability of a power supply, stoppage of automated equipment and accidents are quickly becoming unacceptable events, be it to the ordinary citizen or industrial manufacturers. Dependability and its components: reliability, maintainability, availability and safety, have become a science that no designer can afford to ignore. This technical report presents the basic concepts and an explanation of its basic computational methods. Some examples and several numerical values are given to complement the formulas and references to the various computer tools usually applied in this field .
cahiers techniques Merlin Gerin n 144 / p.1
1. importance of dependability
Prehistoric men had to depend on their arms for survival. Modern man is surrounded by ever more sophisticated tools and systems on which he depends for safety, efficiency and comfort. Ordinary citizen are specially concerned in everyday life by: s the reliability of the TV set, s the availability of the mains supply, s the maintainability of freezers and cars, s the safety of their boiler valves. Bankers and, in general, service industries give a lot of weight to: s computer reliability, s availability of heating, s maintainability of elevators, s fire related safety.
In competitive industries it is not possible to tolerate production losses. This is even more so for complex industrial processes. In these cases one vies to obtain the best: s reliability of command and control systems, s availability of machine tools, s maintainability of production tools, s personnel and invested capital safety. These characteristics, known under the general term of DEPENDABILITY, are related to the concept of reliance, (to depend upon something). They are quantified in relation to a goal, they are computed in terms of a probability and are obtained by the choice of an architecture and its components. They can be verified by suitable tests or by experience.
For over 20 years Merlin Gerin has pioneered work in the DEPENDABILITY field: in the past, with its contribution to the design of nuclear power plants or the high availability of power supplies used at the launching site of the ARIANE space program, nowadays, by its design of products and systems used worldwide.
2. dependability characteristics
reliability
Light bulbs are used by everyone: individuals, bankers and industrial workers. When turned on, a light bulb is expected to work until turned off. Its reliability is the probability that it works until time t and it is a measure of the light bulbs aptitude to function correctly. Definition: The reliability of an item is the probability that this item will be able to perform the function it was designed to accomplish under given conditions during a time interval (t1,t2); it is written R(t1,t2). This definition follows the one given by the IEC (International Electrotechnical Commission)International Electrotechnical Vocabulary, Chapter 191. There are certain basic concepts used by this definition which must be detailed:
Function: the reliability is a characteristic assigned to the systems function. Knowledge of its hardware architecture is usually not enough. Functional analysis methods must be used to determine the reliability. Conditions: the environment has a fundamental role in reliability. This is also true for the operating conditions. Hardware aspects are clearly insufficient. Time interval: we wish to emphasize an interval of time as opposed to a specific instant. Initially, the system is supposed to work. The problem is to determine for how long. In general t1=0 and it is possible to write R(t) for the reliability function.
the probability that it will suddenly burn out in the interval of time (t, t+t), given that it kept working until time t. Failure rates are time rates and, as such, their units are inverse time. Mathematically, the failure rate is written as: (t) = lim
t0
1 R(t) - R (t+t) t R(t)
)
(1)
-1 d R(t) R(t) dt
failure rate
Consider the light bulb example again. Its failure rate at time t, written as (t), gives
For a human being, the failure rate measures the probability of death occurring in the next hour: (20 years)=10-6 per hour. If is represented as function of age, one obtains the curve given in figure 1.
cahiers techniques Merlin Gerin n 144/ p.2
After the high values corresponding to the infant mortality period, reaches the value of adult age during which it becomes constant since causes of death are mainly accidental and thus, independent of age. After 60 years old, old age causes to increase. Experience seems to show that many electronic components follow a similar bathtub curve, from which the same terminology is borrowed: infant mortality, useful life and wearout. During the useful life, is constant and Equation (1) becomes R(t) = exp(-t). This is the exponential distribution and the shape of the reliability function is given in figure 2. The exponential distribution is one among many other possibilities. Mechanical devices which are subject to wearout since the beginning of their operating life can follow other distributions, like Weibulls distribution. In this case the failure rate is time dependent. A curve illustrating the time dependency of is seen in figure 3, in which no plateau, as in figure 1, exists.
(t)
infant mortality
useful life
wearout
fig. 1: bathtub curve
availability
To illustrate the concept of availability consider the case of an automobile. A vehicle must start and run upon demand. Its past history may be of little relevance. The availability is a measure of its aptitude to run properly at a given instant. Definition: The availability of a device is the probability that this device be in such a state so as to perform the function for which it was designed under given conditions and at a given time t, under the assumption that external conditions needed are assured. We will use the symbol A(t). This definition, inspired by the one given by the IEC, mimicks the one for the reliability. However, its time characteristics are basically different since the concept of interest is an instant of time instead of a time length. For a repairable system, functionning at time t does not necessarily imply functionning between [0,t]. This is the main difference between availability and reliability. It is possible to plot the availability curve
R (t) = e - t
0 t fig. 2: exponential reliability
(t)
infant mortality period
fig. 3: wearout reliability curve
as a function of time for a repairable device, having exponential times to failure and to repair, (see figure 4). It can be seen that the availability has a limiting value which, by definition, is the asymptotic availability. This limit is reached after a certain time. The limiting reliability is always zero since, eventually, all devices will fail. (This last point is controversial when dealing with software). Consider again the case of the automobile. Two kinds of cars can have poor availabilities: those with frequent failures and those which do not fail often but instead spend a long time in the garage for repairs. Thus, although the reliability is an important component of the availability, the aptitude to being promptly repaired is also of paramount importance: this is measured by the maintainability.
safety
It is possible to distinguish between dangerous failures and safe ones. The difference does not lie so much in the failures themselves but in their consequences. Switching off the light signals in a train station or suddenly switching them from green to red has an impact (all trains stop) but is not functionally dangerous. The situation is totally different if the lights would accidentally turn all to green. Safety is the probability to avoid dangerous events.
The concept of safety is closely linked to that of risk which, in turn, not only depends on the probability of occurrence but also on the criticality of the event. It is possible to accept a life threatening risk (maximum criticality) if the probability of such an event is minimal. If it is just a matter of having a broken limb the acceptable probability might be greater. The curve on figure 5 illustrates the concept of acceptable risk.
D (t) 1
maintainability
Many designers seek top performance for their products, sometimes neglecting to consider the possibility of failure. When all the effort has been concentrated on having a functionning system, it is difficult to consider what would happen in case of failure. Still, this is a fundamental question to ask. If a system is to have high availability, it should very rarely fail but it should also be possible to quickly repair it. In this context, the repair activity must encompass all the actions leading to system restoration, including logistics. The aptitude of a system to be repaired is therefore measured by its maintainability. Definition: The maintainability of an item is the probability that a given active maintenance operation can be accomplished in a given time interval [t 1,t 2]. It is written as M(t1,t2).This definition also follows closely that of the IECs international vocabulary. It shows that the maintainability is related to repair in a manner similar to that of reliability and failure. The maintainability M(t) is also defined using the same hypotheses as R(t). The repair rate (t) is introduced in a way analogous to the failure rate. When it can be considered constant, the implication is an exponential distribution for: [M(t) = exp(-t)].
D
0 t
fig. 4: availability as a function of time
criticality
unacceptable risk
acceptable risk
probability of occurrence
fig. 5: the level of risk is a function of both, criticality and probability of occurrence.
3. dependability characteristics interdependence
interrelated quantities
The examples given so far have shown that the concept of dependability is a function of four quantifiable characteristics: these are related to each other in the way shown by figure 6. These four quantities must be considered in all dependability studies. The dependability is thus often designated in terms of the initials RAMS. Reliability: probability that the system be failure free in the interval [0,t]. Availability: probability that the system works at time t. Maintainability: probability that the system be repaired in the interval [0,t]. Safety: probability that a catastrophic event is avoided.
one of three states, see figure 7. In addition to the normal functionning state, two further failed states can be considered: a failsafe state and a state of dangerous failure. In order to simplify this description we are including in the failed states all modes of degraded performance, labeled incorrect performance. The time spent before leaving state A is characteristic of the reliability. The time spent on state B, after a safe failure, is characteristic of the maintainability. The
ratio between the time spent on state A and the total time is characteristic of the availability. The aptitude of the system to avoid spending any time on state C is a characteristic of safety. It can be seen that state B is acceptable in terms of safety but is a source of unavailability.
AVAILABILITY
SAFETY
conflicting requirements
Some of the requirements of the dependability can be contradictory. An improved maintainability can bring about some choices which degrade the reliability, (for example, the addition of components to simplify the assemblydisassembly operations). The availability is therefore a compromise between reliability and maintainability. A dependability study allows the analyst to obtain a numerical estimate of this compromise. Similarly, safety and availability might conflict with each other. We have noted that the safety of a system is defined as the probability to avoid a catastrophic event and is often maximum when the system is stopped. In this case, its availability is zero! Such a case arises when a bridge is closed to traffic when there is a risk of collapse. Conversely, to improve the availability of their fleet, certain airlines are known to have neglected their preventive maintenance activities thus diminishing flight safety. In order to ascertain the optimum compromise between safety and availability it is necessary to produce a scientific computation of these characteristics. A system can be described as being in
repair STATE A NORMAL FUNCTIONNING failsafe
RELIABILITY
fig. 6: the components of dependability
MAINTAINABILITY
STATE B INCORRECT PERFORMANCE AND NOT DANGEROUS
STATE C dangerous failure INCORRECT PERFORMANCE AND DANGEROUS
fig. 7: failsafe: availability dangerous failure: safety
time average related quantities

In addition to the previously mentioned probabilities (reliability, availability, maintainability and safety) of occurrence of events, it is common to use mean times before the ocurrence of events in order to describe the dependability. Mean times It is useful to recall here the exact definition of all the mean times as they are often misunderstood. The worst example of abuse is probably the most widely known, the MTBF, which is often confused with lifetime. On the average, in a homogenous population of items following an exponential distribution, about 2/3 of these items will have failed after a time equal to the MTBF. A single system having a constant failure rate will have a 63%
chance of having failed after such a time. The definitions and relative positions of these mean times during the life of a system are given in figure 8. MTTF or MTFF (Mean Time To First Failure): the mean time before the occurrence of the first failure. MTBF (Mean Time Between Failures): mean time between two consecutive failures in a repairable system. MDT (Mean Down Time): mean time between the instant of failure and total restoration of the system. It includes the failure detection time, the repair time and the reset time. MTTR (Mean Time To Repair): mean time to actually restore the system to an operating condition. MUT (Mean Up Time): mean failure free time.
Important relations and numerical values There are many mathematical relations linking the quantities introduced thus far: For an exponential distribution with R(t) = exp(-t) one has MTTF = 1/. In this case, for a non repairable system, we have MTBF = MTTF (in fact, in this case, all failures are first failures). This explains why the classical formula used for electronic components (non repairable) is: MTBF = 1/. The above formula is only valid for exponential distributions (constant failure rates) and, strictly speaking, for non repaired items although it is possible to apply it for repaired systems with very small MDTs. Analogously, when repair times obey an exponential distribution, it is possible to show that MTTR = 1/. One also has: MTBF = MUT + MDT. In general it is also true that MDT = MTTR, except for the logistic delay and restart times. Furthermore:
s
asymptotic availability
MTTF
MTBF
MTBF
This formula illustrates the assertion given

A = lim A t
t +
MDT
MUT
MDT
MUT
MDT
on page 3 concerning the availability (ratio of correct performance time to total time). This quantity MUT corresponds to the MTBF asymptotic value given in figure 4, page 4. s asymptotic unavailability = 1 - asymptotic availability
U = lim 1 - A t
t +
time
failure repair
failure repair
failure repair
failed state
up state
The asymptotic unavailability is usually easier to express numerically than the availability: it is much easier to read 10-6 than 0.999999. For exponential distributions, using the equations MUT = 1/ and MDT = 1/ one obtains:
+ +
fig. 8: diagram for mean times in the case of a system with no interruptions due to preventive maintenance
U=
or A =
is often much smaller than since the repair times are much smaller than the times to failure. It is therefore possible to simplify the denominator and write:
U= = .MTTR
It can be seen that the reliability is degraded when the complexity of the system increases. This corresponds to a well-known rule of dependability design: simplify as much as possible. The concept of mean time is often misunderstood. For example the next two sentences have, for exponential distributions, the same meanings: The MTTF is 100 years and The odds are one in 100 to observe a failure in the first year. Still, the second sentence seems more worrisome for a manufacturer selling 10 000 devices of this type per year. On the average, about 100 units will fail on the first year.
This last formula illustrates, in the case of exponential distributions, the compromise between reliability and maintainability which has to be optimized to improve the availability. The table of figure 9 gives failure rates and mean times to failure for certain devices belonging to the electronic and electrotechnical fields.
To illustrate the impact of redundancy on the unavailability, consider the national power grid. One is concerned with the deliverance of energy to the final user. The unavailability is about 10-3. This corresponds to about 9 hours of downtime per year. For a computer room, having a heavily redundant system of Uninterruptible Power Supplies (UPS), it is possible to reduce this figure between 1000 and 10 000 times.
resistances
microproc.
(/h) MTTF
10-9 1000 centuries
fuses and circuitbreakers, 300 ft. cables, busbars 10-6 10-7 to 10-6 100 years 100 to 1000 years
generator
mains outages
10-5 10 years
10-2 4 days
fig. 9: failure rates and mean times to failure for certain devices belonging to the electronic and electrotechnical fields
4. types of defects
The design of a system with respect to its dependability goals implies the need to identify and take into account the various possible causes of defects. One can suggest the following classification:
operating errors
arising from an incorrect use of the equipment:
s hardware being used in an inappropriate environment, s human operating or maintenance errors, s sabotage.
physical defects
induced by internal causes (breakdown of a component) or external causes, (electromagnetic interferences, vibrations,...).
design defects
comprising hardware and software design errors.
The various techniques discussed in this document concern mostly physical defects. Nevertheless, human and software errors are also very important although the state of the art in these fields is not as advanced as for physical defects. Still, within the scope of this document, we feel the following elements are worth mentioning:
Software aspects s the reliability of a piece of software in which all the inputs are exhaustively tested is equal to 1 forever. Nevertheless, this is unrealistic for real life, complex programs. s having two redundant programs implies development by different software teams using different algorithms. This is the principle behind fault tolerant software in which a majority vote may be implemented. s most software reliability models can be split in two major categories: s complexity models: based upon a measure of the complexity of the code or algorithm, s reliability growth models: based upon previous observed failure history. s the quantitative evaluation of the
different models does not allow yet for a systematic study of software reliability. The best results are obtained in particular cases and for given environments (language, methods). This is the case for the SPIN (Integrated Digital Protection System) software developped by Merlin Gerin for use in nuclear power plants. Merlin Gerin is also an active participant in different working groups dealing with software reliability (see references). The Technical Paper CT 117 gives further details on this subject. The title is Methods for developping dependability related software. Human reliability
Qualitative approaches are predominant in this field. The efforts lie mostly in the modeling of the human operator, task classification and human errors. The most advanced studies belong to the nuclear and aerospace industries. Human behavior is known as much by simulators as by field reports. Both sources can be compared to each other. Some references exist which propose some numerical values. However, these must be used with utmost caution. According to these references it is feasible to assign an error probability depending on the nature of the activity: mechanical, procedure or cognitive action. Some of the recent major catastrophes
have shown that the human factor can have great impact, not only from the operator standpoint but also at the designers stage. The more freedom of action is given to a human operator the more the risks are increased. This also includes management, as the Challenger Space Shuttle accident has shown: it is possible to go all the way up to the designers of the working structure of the designers team! Many disciplines are called upon to tackle the problem of human reliability. Among them psychology and ergonomy.
5. from component to system: modeling aspects
data bases for system components

Electronics Reliability calculations have been widely used in this field for many years. The two best known data bases are the Military Handbook 217 (version E at present) issued in the U.S. and the Recueil de donnes de fiabilit, from CNET (French Telecom Center), see figure 11 for an example. Merlin Gerin participates in its updates. These data bases allow the calculation of the failure rates of electronic components, assumed to be constant. These rates are a function of the application characteristics, environment, load, etc. The type of component is also relevant, e.g., number of gates, value of the resistance, etc. Computation is usually faster with the CNET approach but many specialized computer programs exist to implement either technique with ease. As an example, let us take a 50 k
resistance used in an electronic board and used inside an electric switchboard. It is necessary to consult the table given in figure 11 in order to determine the corresponding correcting values. The environment is au sol (fixed, ground) and therefore, the environment corrective factor is: E = 2.9 The resistance value gives the corresponding multiplying factor: R = 1 This resistance is taken as being non qualified which gives the multiplying quality factor Q = 7.5 The load factor is a characteristic of the application, as opposed to the other factors which are characteristic of the component itself. If the load factor is 0.7 and the environmental temperature for the board is 90C, the diagram gives b = 15 The global failure rate for this resistance
is thus obtained by multiplying all the corrective factors and the base failure rate: = b.R.EQ = 0.33 x 10 -6 / hour If at the design stage the reliability goals have been integrated, then: s better thermal designs will allow a lowering of the environment temperature, s better board designs will lower the load factor . With t = 60C and = 0.2 the diagram gives: b = 1.7 If now a qualified component is selected, we have: Q = 2.5, which gives = 0.012 x 10 -6, that is an improvement factor of 30. Knowledge of the reliability of each component provides a means to obtain the reliability of the boards, (which are repairable or replaceable), and therefore that of whole electronic systems. This is done by using the techniques described in the rest of this report.
Mechanics and electromechanics Data bases in these fields exist although they are not really standards. Some sources are: s RAC, NPRD 3: report by the Reliability Analysis Center (RADC, Griffiss AFB), under contract from the US DoD, dealing with non electronic parts. s IEEE STD 500: field data on reliability of electrical, electronic and mechanical equipment used in nuclear power plants. In France and the US, some reference books exist that deal specifically with mechanical components. As an example of data relevant to our activities, figure 10 gives some information concerning circuit breakers. This comes from RACs NPRD 3-1985. First, there is a failure mode distribution in a pie chart. For example, 34% of all field failures are due to the circuit breaker failing to open
when it should. The table in figure 10 gives a point estimate of the failure rate for the thermal function of circuit breakers. Various information items given are as follows: s environment: GF, Ground Fixed, industrial conditions. s failure rate estimate: 0.335 10-6 h-1 s a 60% confidence interval for the failure rate using the 20% lower and 80% upper bounds. s the number of records used in this calculation, i.e. 2. s the number of observed failures: here 3. s the total number of operating hours: 8.994 106 h . The actual knowledge of the global failure rate and the failure mode distribution allows the calculation of the probability of specific events by using a simple proportionality rule.
For example, for the stuck closed mode, we have a corresponding failure rate of: -6 -7 34 = 1.17.10 0.335.10 x 100 Another approach can sometimes be more relevant: instead of considering the calendar time, the number of makebreak operations can be tallied. Then, a test is planned in which a sample is selected and the reliability is estimated using a more realistic model (e.g. Weibull distribution). Which technique to use is largely a matter of determining the kind of failure one wishes to study: contact wear is related to the number of make and break cycles whereas corrosion is time dependent. Specific use and environment conditions are always important.
15.00 %
8.00 %
noisy
15.00 % 6.00 %
no movement intermittent degraded stuck closed
8.00 % 9.00 %
stuck open out of adjustment others
4.00 %
34.00 %
component APPL part type ENV thermal GF
user code M
point estimate 0.335
60 % upper single-side -
20 % lower internal 0.171
80 % upper internal 0.621
% of recs 2
% of fail 3
operating HRS (E6) 8.944
fig. 10: failure modes and reliability data for circuit breakers
The people interessed in this kind of information can refer to American Standard referenced: MIL HDBK 217 E
fig. 11: example of CNET publications
Failure Modes, Effects and Critically Analysis (FMECA) method

This is a technique to analyse the reliability of a system in terms of the failure modes of its components. The IEC has issued a standard (IEC 812) giving a description of this technique. Each element of the system can, in turn, be analyzed using
one of the relevant data bases. The hardware structure of the system as well as its functional characteristics allow the analyst to inductively assess the effect of each and all of the failure modes corresponding to each element and their effects on the system. An FMECA should also give an estimate of the criticality of each failure mode, see figure 12. This depends on two factors:
the probability of occurence of failure and the seriousness of its consequences. Thus an FMECA is a tool to study the influence of the component failures on the system. The main interest of this technique lies in its exhaustiveness. It is nevertheless incomplete in that the combination of effects must be seraparately considered. This can be accomplished using the methods described in the rest of this chapter.
component circuit-breaker
function switch short circuit prot. current path
failure mode stuck closed unable to close unable to open sudden open heat
cause solder mechanical solder adjustment bad contact
effect no shedding no power no protect no power electronic failure
criticality 2 2 4 3 2
comments
action
fig. 12: example of FMECA table
Reliability Block Diagram (RBD)

The RBD method is a simple tool to represent a system through its (nonrepairable) components. Using the RBD allows the computation of the reliability of systems having series, parallel, bridge and k-out-of-n architectures or any of its combinations. Although it is possible to apply the RBD technique to repairable systems, the implementation is much more difficult. Series-parallel systems Two components are in series, from the reliability standpoint, if both are necessary to perform a given function. They are in parallel when the system works if at least one of the two components works, see figure 13. These considerations are easily generalized to more than two components. Whenever two components are in series and can be considered to be independent, (the failure of one does not modify the probability of failure of the other), the reliability of this sytem can be calculated by multiplying the individual reliabilities together since the first component AND the second must work:
1
series
parallel 1
fig. 13: series/parallel systems
R(t)=R1(t).R2(t). In the case of two independent components in parallel, the system works if one OR the other works. It is easy to calculate the unreliability of the system since it is equal to the product of the two component unreliabilities: the system fails if the first component AND the second component fail: 1 - R(t) =(1 - R1(t)).(1 - R2(t)). Or equivalently: R(t) = R1(t)+R2(t) - R1(t).R2(t). In this case, components 1 and 2 are said to be in active redundancy. The redundancy would be passive if one of the parallel components is turned on only in the case of failure of the first. This is the case of auxiliary power generators.
For the particular case of non repairable components following an exponential distribution of times to failure, one can write: For the series case: R(t) = exp(-1t).exp(-2t) = exp(-(1+2)t). It follows that the systems times to failure also follow an exponential distribution, (constant failure rate), since the reliability function is an exponential with: = 1+ 2 For the parallel case: R(t) = exp(-1t)+exp(-2t)-exp(-(1+2)t). Here, the reliability function is not an exponential. Therefore, it can be concluded that the failure rate is not constant.
All these formulas can be generalized to a system with n non repairable components, mixing series and parallel architectures. k-out-of-n redundancies A k-out-of-n system, or simply K/N, is a ncomponent system in which k or more components are needed for the system to work properly. We will consider only active redundancies here, see figure 14: Let us call Ri(t) the reliability of each one of the n components of the system. In some simple cases the reliability of the system can be computed by adding the favourable combinations:
s
2 K/N
2/3 system:
fig. 14: K/N redundant systems
R=R1.R2+R1.R3+R2.R3
s
series system (n/n):

n i=1
R(t) = R i (t)
s
parallel system (1/n):

n
1 - R(t) = ( 1 - R i (t)
i=1
k/n system of identical components

3
If we write Ri (t) = r (t), then,
R(t) = C n r(t) ( 1 - r(t))

i i=k
n-i
Bridge systems These are systems which cannot be described by simple series-parallel combinations. They can, however, be reduced to series-parallel cases by an iterative procedure, see figure 15. In order to compute the reliability of this system in terms of the five non repairable component reliabilities it is necessary to apply conditional probabilities: R=R3.R(given that 3 works) + (1-R3).R(given that 3 has failed). It is thus possible to derive the system reliability R(t) by decomposing the original bridge system in the two disjoint systems illustrated in figure 16. Example: reliability of an intrusion detection system. The system consists of two sensors, a vibration sensor and a photoelectric cell. Each of these sensors could be connected to its specific alarm, as in figure 17, and we would have two independent branches. However, a bridge system
fig. 15: bridge systems
would result if each sensor is connected to either one of the two alarms, as in figure 18, through a coupler. We will calculate the reliability improvement due to this modification. Let us also suppose that the mission time of this system is three months, i.e., the maximum expected absence during which the system must function. Furthermore, after each mission, the system is thoroughly checked and maintained and can be considered as good as new when reset. During the mission, there are no repairable elements. Let us use the following realistic constant failure rates to obtain the different orders of magnitude: Vibration sensor: Photoelectric cell: 1 = 2.10-4 2 = 10-4
Coupler: 3 = 10-5 Alarms: 4 = 5 = 4.10-4 All these failure rates are given in (hours)-1
s
computation for Diagram A of figure 17. This is a simple case of two parallel branches, each having two components in series: Reliability of Branch 1: R1(t).R4(t) Reliability of Branch 2: R2(t).R5(t) System reliability: RA(t) = R1(t).R4(t)
+ R2(t).R5(t) - R1(t).R4(t).R2(t).R5(t) Using Ri(t)= exp(-it) with t = 3 months = 2190 hours as the mission time one obtains: RA(3 months) = 0.51.
fig. 16: decomposition of a bridge system
computation for Diagram B of figure 18 This is the bridge system. Whenever the coupler is failed we are back to the diagram of figure 17. On the other hand, when it works, we have 1 and 2 in parallel, both in series with 4 and 5, themselves in parallel. The system reliability for figure 18 is then: RB = (1-R3).R+R3.(R1+R2-R1.R2).(R4+R5 -R4.R5) The numerical computation gives RB(3 months) = 0.61. In spite of the excellent reliability of the coupler, the systems reliability is only marginally improved. This numerical example shows, through a simple calculation, that there is not much sense in having a more expensive set-up. Case of repairable elements RBDs cannot be used as systematically as before: s for two components in parallel, the equation relating R(t) to R1(t) and R2(t) is no longer valid. In fact, a working system in the interval [0,t] may correspond to an alternating working condition between 1 and 2, with non repairable components there should be at least one working component in the time interval [0,t] whereas for repairable components both can fail, but not simultaneously. s the equation R(t) = R1(t).R2(t) remains valid for a two reparaible component series system. s in the case of repairable components the main concern is the numerical estimate of the availability. It is possible to use the RBDs with the same formulas as
s
alarm 1 vibration sensor 1 4
((
(( ((
4
alarm 2 photoelectric cell 2 5
((
fig. 17: alarms with no coupling, diagram A
coupler
fig. 18: system with coupler, diagram B
for the reliability calculations: A(t) = A1(t).A2(t) for a series system A(t) = A1(t)+A2(t)-A1(t).A2(t) for parallel systems. These formulas are valid only for simple cases For instance, the formula A(t)= A1(t)+A2(t) -A1(t).A2(t) ceases to be valid if only one
repairman is available, (instead of as many as necessary). This sequential feature, i.e. having a component waiting to be repaired while the other is being serviced, is not possible to model by a simple RBD. In these cases the State Graphs, to be dealt with later, are adapted to this problem.
fault trees analysis

The computation of the systems failure probability is the main goal of this type of analysis. It is based upon a graphical construction representing all the combinations of events, essentially through AND-gates and OR-gates, that may lead to a catastrophic event. Except for extremely simple cases, computer resources must be used to evaluate the probability of the catastrophic event. It is then possible to modify the structure of the systems design to lower this probability. Basic procedure A deep understanding of the system and a clear definition of the catastrophic event are essential to build the fault tree. The catastrophic event, sometimes called the top event, is then analyzed in terms of its immediately preceding causes. Then, each one of these causes is analyzed in terms of their own immediately preceding causes until the basic events are reached. These are supposed to be independent. A simple example is given in figure 19 and its corresponding fault tree in figure 20. This tree only contains OR-gates connecting the intermediate events (rectangles) and the basic events. The basic events are represented by circles. It is convenient to define a cut-set as a simultaneous combination of basic events that, by themselves, produce the top event. The analysis proceeds in two phases: qualitative analysis: the minimal cutsets, or min cuts, are obtained. The min cuts are minimal combinations that include basic events that lead to the top event. The order of a min cut is simply the number of basic events it contains.
s s quantitative analysis: this is performed using the min cuts and the probability of occurrence of the basic events. This gives an approximate value for the probability of the top event. It is also necessary to validate the accuracy of this approximation in a systematic fashion. Then, depending on the objectives of the analysis, different probabilities are used to compute the system reliability or its availability. fuse switch
The top event is: motor unable to start
fig. 19: electrical supply for a motor
motor idling and unable to start
no power
motor failure
immediate causes
no + link
no - link
dead battery
intermediate causes
fuse
open wire
switch
open wire
fig. 20: fault tree for fig. 19 circuit
We can illustrate these ideas by two examples:
s an overhead projector with one lamp inside and one spare. The top event is "no working lamp available", see figure 21.
A single AND-gate is necessary. The chances of this happening is seen to be 2 in two thousand.
s a simple light bulb. The top event is no light, see figure 22. A single OR-gate is necessary. The probability of the top event is seen to be about 0.001, one in a thousand of not having light. The main cause for this event is the burn out of the light bulb.
no light
failure probability: P AND-Gate
In the general case it is often possible to obtain an exact calculation of the probability of the top event using recursivity instead of the min cuts: Boolean probability calculations are performed for each gate in terms of the sub-trees being input to the gate considered. The assumption of independence must be verified but this procedure leads to an exact evaluation of the top event. Thus, the recursive calculation allows a comparison to the min-cut approach. Both methods are complementary. Application of fault tree using mincuts to the availability of a low voltage network. The fault tree corresponding to the network given in figure 23 is shown in figure 24. Power is considerd to be either present or absent. The top event is assumed to be the absence of power at the output, noted E. In building this tree certain assumptions are made:
s only two failure modes are considered for the circuit-breakers: sudden contact break and failure to open upon a shortcircuit. s each transformer line can, by itself, supply voltage to the main network, to which E belongs. s the two mains supplies are coming from two different Medium Voltage sources. This reduces the Common Mode failure to the unavailability of the High Voltage supply. Each event in the Fault Tree will have a certain probability of occurrence associated with it. In this case the probability will be the unavailability. The unavailability associated with the basic events is calculated by the formula: U .MTTR.
P1
1st. light bulb dead
P2
2nd. light bulb dead or missing
one order 2 min-cut
P = P x P = 0 , 0 5 x 0 , 0 4 = 2 . 10 - 3 1 2
fig. 21: fault tree for an overhead projector
no light
failure probability: P
OR -Gate
P1
no mains
P2
light bulb dead
two order 1 min-cuts
1- P = ( 1 - P
) (1 -P
) = ( 1 -1 0
-4
) ( 1 -1 0
-3
) = 0,9989
fig. 22: a fault tree for a light bulb
B Busbar 1
C Busbar 2 E
D Busbar 3 F
is the failure rate corresponding to a particular failure mode of a component. It can be obtained from several sources of field data.
fig. 23: low voltage network
no power in output E G11*
BB 3 failure 2*1*
no power to BB 3 G22*
sudden opening of C.B.E 2*3*
short circuit through F G24*
wire failure 3*1*
sudden opening of C.B. D 3*2*
C.B. F stuck on short circuit 3*4*
short circuit above F 3*5*
BB 1 failure 4*1*
short circuit through C G43*
double line failure G51*
no HV supply 5*2*
short circuit through C G53*
C.B. C stuck on short circuit 5*4*
line A G61*
line B G62*
cable
BB 2
6*3*
6*4*
transfo A 7*1*
C.B. A
transfo B 7*3*
C.B. B
7*2*
7*4*
fig. 24: fault tree corresponding to Fig. 23 network
MTTR is the Mean Time to Repair and it depends on the component being considered as well as the particular installation, technology, geographical location, service contract. In some instances a specific value of a probability is unknown. A worst case situation, or upper bound, is therefore assumed. For example, we have taken the upper bound probability of a shortcircuit above F to be 10-2. The results of the Fault Tree Analysis, shown in figure 25, indicate that the unavailability on output E is 10-5 which corresponds to 5 minutes per year. The min cut approach allows, in addition to the calculation of the probability of the top event, the assessment of the weight each min cut carries in producing the top event. Figure 25 also shows this weight, as a percentage of the total unavailability which is possible to attribute to each min cut. This contribution is one measure of the importance of the min cut. An eyeball examination of the min cuts relative importances shows that the cable linking busbar 1 to busbar 3, (third min cut), is critical. To a lower extent this is also true of the two busbars 1 and 3. If these components were improved, the mains supply then becomes critical. If a further improvement on the overall availability became essential, it would be necessary to incorporate an auxiliary power supply, such as a diesel generator. A detailed study of the availability of an electrical supply is presented in Merlin Gerins Technical paper Suret et distribution lectrique (in French).
transitions correspond to the different events that concern the components of the system. In general, these events are either failures or repairs. As a consequence, the transition rates between states are essentially failure rates or repair rates, eventually weighted by probabilities like that of an equipment refusing to turn on upon demand. The graph on figure 26 shows the behavior of a system with a single repairable component. Assumptions A model is said to be markovian if the following conditions are satisfied: s the evolution of the system depends only on its present state and not on its past history, s the transition rates are constant, i.e. only exponential distributions are considered, s there is a finite number of states, s at any given time there cannot be more than one transition. Equations Under the above hypotheses, the probability of the system being in state Ei at time t+dt can be written as: Pi(t+dt) = P(the system is in state E i and it stays
there) + P(the system comes from another state Ej). For a graph having n states, n differential equations are obtained which can be written as:
d(t) = (t).[A] dt where: (t) = [P1(t), P2(t), , Pn(t)]
[A] is called the transition matrix of the graph. The solution of this equation in matrix form is performed by computer and gives the probabilities Pi(t), that is the probability of the system being in state i as a function of all the transition rates and the initial state. Computation of dependability quantities The availability being the probability of the system being in a working state, it follows: D(T) = P i (t) .[A i where Pi(t) = probability of being in working state Ei.
unavailability: 1.01 E -05, i.e. 1.01 10 -5 list of min cuts and their importance min cuts indicated on the fault tree, percent contribution 1 2 3 4 5 6 7 8 9 10 11 12 13 :2*1* :2*3* :3*1* :3*2* :3*4* , 3*5* :4*1* :5*2* :5*4* , 6*3* :5*4* , 6*4* :7*1* , 7*3* :7*1* , 7*4* :7*2* , 7*3* :7*2* , 7*4* : : : : : : : : : : : : : 9,5 1,6 68 1,6 ,013 9,5 9,9 9,1E - 6 3,2 E - 6 ,00058 1,3 E - 5 1,3 E - 5 2,7 E - 7
state graphs
State graphs, also called Markov graphs, allow a powerful modeling of systems under certain restrictive assumptions. The analysis proceeds from the actual construction of the graph to solving the corresponding equations and, finally to the interpretation of results in terms of reliability and unavailability. Mathematically, a great simplification is obtained by considering only the calculation of time independent quantities. Construction of the graph The graph represents all the possible states of the system as well as the transitions between these states. These
fig. 25: contributions of network components to its unavailability
:failure rate up state down state
: repair rate
fig. 26: elementary state graph
The reliability is the probability of being in a working state without ever having passed through a down state. A graph is constructed by deleting all transitions going from a failed state to a working state. Once the new probabilities Pi(t) are obtained, we have: R(t) = P i (t)
i
UPSs. Each working UPS in state Ei adds its own exit rate towards state Ei+1. These exit rates are 3, 2 and respectively. The up states are 0 and 1. We assume that the repair strategy is such that there can be three repairmen working simultaneously on each UPS. Thus, the transition rates corresponding to the repair activity are proportional to the number of failed UPSs in the state being considered. The numerical values are as follows: = 2.10-5 h-1 ; = 10-1 h-1 Figure 28 gives the computed results corresponding to the time independent
quantities. It can be seen that the MTTF is here 4.17 107 hours whereas the nonredundant case (3/3) has an MTTF equal to 1/3 = 1.67 104 hours. For the asymptotic unavailability the change is from 1.19 10-7 for the redundant system to 6 10-4 for the non redundant case (3/3) system. The comparison of these figures is easily visualized through the graph itself: in the redundant case, the unavailability is calculated by summing the probabilities of the two failed states, i.e., A = P2+P3 while, in the non redundant case, the sum is performed over three failed states: A = P1+P2+P3
There are two other quantities which are very simple to obtain: the meant time of state occupancy: 1 Ti = (rates of departure from state i) s the occupancy frequency corresponding to state i: Pi f i= Ti The characteristic mean times MTTF, MTTR, MUT, MDT, MTBF are calculated using matrix calculus and some of the equations already discussed. For the MTTF, the initial state of the system must be specified in terms of the probabilities of the system being initially in each one of its different states.
s
3 state 0 state 1
2 state 2 2
state 3 3
Application: Uninterruptible Power Supplies (UPS) in parallel A UPS is a device which improves the quality of the electrical supply. It is often used for critical applications such as computers and their peripherals. We will consider a typical configuration (Triple Modular Redundancy), i.e. the UPSs constitute a 2/3 redundant system. The unavailability is not the only quantity of interest: the MTTF gives the mean time before the first black-out. In the construction of the state graph it is here possible to use the fact that the three UPSs are identical and therefore states can be grouped, according to the number of failed UPSs. The failure and repair rates for the UPSs, and respectively, are given in figure. 27 The number associated with each state corresponds to the number of failed
fig. 27: UPS's in parallel
Time independant quantities: Unavailability: MTTF MUT MTBF : : : : 1.199360E-07 4.169167E+07 4.169167E+07 4.169167E+07 Availability MTTR MDT : 9.999999E-01 : 8.333667E+00 : 5.000333E+00
fig. 28: values corresponding to the graph on figure 27
6. conclusion
The dependability is a concept becoming ever more critical for comfort, efficiency and safety. It can be controlled and calculated. It can be designed in, be it for devices, architectures or systems. Dependability characteristics are now frequently included in specifications and
contracts. The existence of computational methods and tools allows the systematic study of the dependability during the design phase and for quality assurance purposes. An intuitive insight, combined with exact or approximate calculations, allow the
comparison of different configurations and thus provide an evaluation of risk associated to a better performance, i.e. performance adapted to clearly specified needs.
7. references and standards
Military Handbook 217E DoD (U.S.A.) October 1986. Recueil de donnes de fiabilit, CNET (Centre National dEtudes des Tlcommunications, France) 1983. IEEE Std. 493 and IEEE Std. 500 (Institute of Electrical and Electronic Engineers) 1980 and 1984. NPRD document 3 Nonelectronics Parts Reliability Data Reliability Analysis Center, (RADC) 1985. A. Pags, M. Gondran: Fiabilit des systmes Eyrolles, France1983.
A. Villemeur: Suret de fonctionnement des systmes industriels Eyrolles, France 1988. International Electrotechnical Vocabulary VEI 191 International Electrotechnical Commission June 1988. Proceedings of the 15th InterRam conference Portland, Oregon June 1988. C. Marcovici, J. C. Ligeron: Techniques de fiabilit en mcanique Pic, France, 1974.
EPRI document 3593 Electrical Power Research Institute Hannaman, Spurgin, 1984. NUREG document 2254 US Nuclear Regulatory Commission Bell, Swain, 1983. Merlin Gerin Technical Report 117 : Mthode de dveloppement dun logiciel de suret A. Jourdil, R. Galera 1982. Merlin GerinTechnical Report 134 : Approche industrielle de la suret de fonctionnement H. Krotoff 1985. Merlin Gerin Technical Report 148 : Suret et distribution lectrique G. Gatine 1990.
IEC Standard 271 List of basic terms, definitions and related mathematics for reliability. IEC Standard 300 Reliability and maintainability management. IEC Standard 362 Guide for the collection of reliability, availability and maintainability data from field performance of electronic items. IEC Standard 409 Guide for the inclusion of reliability clauses into specifications for components (or parts) for electronic equipment.
IEC Standard 605 Equipment Reliability Testing. IEC Standard 706 Guide on maintainability of equipment. IEC Standard 812 Analysis techniques for system reliability - Procedure for failure mode and effects analysis (FMEA). IEC Standard 863 Presentation of reliability, maintainability and availability predictions. IEC Standard 1014 Programmes for reliability growth.
Merlin Gerins dependability experts have published extensively in this field and have presented papers in most international reliability conferences. Merlin Gerin is also an active participant in several national and international committees dealing with dependability: s presidence of the French National Committee for IEC TC 56 activities, (dependability) and expert with IEC Working Group 4, TC 56, (statistical methods), s software dependability with the European Group of EWICS- TC7: computer and critical applications, s french AFCET Working Group on computer systems dependability, s updating contributions to the French CNET Electronic components reliability handbook, s working Group IFIP 10.4 on Dependable Computing.
Cahier technique n 145

Thermal study of LV electric switchboards
C. Kilindjian
s Merlin Gerin s Square D s Telemecanique
Cahiers Techniques are a collection of documents intended for engineers and technicians people in the industry who are looking for information in greater depth in order to complement that given in display product catalogues. These Cahiers Techniques go beyond this stage and constitute pratical training tools. They contain data allowing to design and implement electrical equipement, industrial electronics and electrical transmission and distribution. Each Cahier Technique provides an in-depth study of a precise subject in the fields of electrical networks, protection devices, monitoring and control and industrial automation systems. The latest publications can be downloaded on Internet from the Schneider server. code: http://www.schneiderelectric.com section: mastering electrical power Please contact your Schneider representative if you want either a Cahier Technique or the list of available titles. The Cahiers Techniques collection is part of the Groupe Schneiders Collection Technique .
Foreword The author disclaims all responsibility further to incorrect use of information or diagrams reproduced in this document, and cannot be held responsible for any errors or oversights, or for the consequences of using information and diagrams contained in this document. Reproduction of all or part of a Cahier Technique is authorised with the prior consent of the Scientific and Technical Division. The statement Extracted from Schneider Cahier Technique no..... (please specify) is compulsory.
n 145
C. KILINDJIAN After graduating as an engineer from the Ecole Suprieure d'Energie et des Matriaux of Orlans in 1986, he then joined Merlin Gerin in this same year as part of the Technical Section in the Low Voltage Switchboards unit. Responsible for basic studies, he specialises in problems of heat exchanges and electrodynamic withstand in LV equipment. He is currently working in the Anticipation departent of the Low Voltage Power Compartments SBS as an expert on thermal problems in LV circuit-breaker and equipment development.
E/CT 145 first issued, december 1997

This Cahier Technique aims at furthering the understanding and mastery of the thermal problems encountered in LV electric switchboards. After a brief review of standards and of thermal phenomena: conduction radiation - convection, the author shows how LV cubicles can be modelled using modelling techniques normally reserved for other areas. Modelling naturally leads to software to aid design of electrical cubicles equipped with switchgear. The results are compared with real temperature measurements. Finally, the methods and possibilities of the IEC 890 guide are described.
Contents
1 Introduction 2 Thermal problems in a switchboard 1.1 Controlling thermal phenomena in LV cubicles 2.1 Causes - effects and solutions 2.2 Taking stock of standards 3 Thermal behaviour of a LV electric switchboard 4 Presentation of modelling 3.1 Briefeview of the main thermal phenomena 3.2 Exchanges at switchboard level 4.1 Principle 4.2 Modelling convection 4.3 Application to LV enclosures 5.1 Busbars 5.2 Switchgear devices 6.1 Principle 6.2 Description of the data to be provided and of the results obtained 6.3 Modelled configurations 6.4 Results 6.5 Experimentl results 7 Method proposed by the IEC 890 report 8 Conclusion p. 4 p. 5 p. 6 p. 8 p. 10 p. 11 p. 12 p. 12 p. 14 p. 14 p. 17 p. 17 p. 18 p. 18 p. 21 p. 22 p. 24
5 Behaviour of heat sources and characteristics 6 Method for calculating temperature in envelopes and experimental results
Cahier Technique Schneider n 145 / p.3
1 Introduction
1.1 Controlling thermal phenomena in LV cubicles

The new manufacturing methods developed in industry in recent years (just in time...) have brought a new notion to light: industrial dependability. This concept which covers two different aspects, safety of persons and equipment, and availability of electrical power, shows when it is applied to complex processes, the critical points whose operation must be thoroughly mastered. The electric switchboard is one of these critical points. Note that the problem is similar for major tertiary. Formerly considered as a simple passing point, it has become the genuine nerve centre of electrical installations. The safety of the entire installation and thus of all industrial and tertiary activities relies on its dependability. Mastery of its operation requires knowledge and control not only of the functioning of its components but also of the external influences to which they are subjected. An electric switchboard is the combination of 4 basic elements: c the envelope, c the switchgear, c the connections, c the functions performing indication, control and processing of information. Electric switchboards are increasingly technical and require a certain number of basic studies in order to master, in the design stage, the operating conditions of its components in a specific environment. One area covered by such studies is the thermal aspects which form the subject matter of this Cahier Technique.
2 Thermal problems in a switchboard
Three main reasons make thermal mastery increasingly vital. These reasons are: c The tendency to place electrical equipment in envelopes (for safety purposes) which are increasingly made of insulating material (poor calory dissipation capacity). c Progress of switchgear which includes more and more electronic components of increasingly compact size.
c The tendency to fill switchboards to their limit and an increasing bulk factor (ratio between the nominal current of the switchboard incoming circuit-breaker and the sum of nominal feeder currents. This factor is also known as the diversity factor).
2.1 Causes, effects and solutions

The temperature of an electrical device is the result of: c the Joule effect (P = R I2), i.e. of its withstand to current flow, c ambient temperature. Electrical switchgear is designed in accordance with manufacturing standards which define the maximum temperatures not to exceed to ensure safety of persons: temperature of case and of switching devices, maximum temperature deviation for terminals; this is verified by product certification tests. As devices function in a wide variety of working conditions in switchboards, the causes of excessive temperature are numerous. Table (see Figure 1) shows the main causes, their effects and the possible solutions.
Causes External temperature too high
Effects c Switchboard internal temperature too high c Tripping of thermal releases c Ageing of electronics c Temperature of enclosure walls too high c Tripping of switchboard incoming protection c Switchboard internal temperature too high c Temperature of enclosure walls too high c Damaged conductors c Damaged insulated bar supports c Device conductors destroyed c Conductors destroyed c Abnormal operation (tripping) c Premature ageing
Protection c Alarm c Automatic fan startup
Solutions c Improve ventilation of room and/or switchboard
High diversity factor. Installation possibilities exceeded.
c Load shedding
c Adequately sized switchboard
Can occur in some cases even when designed according to standard practice.
IEC439
Short-circuit or overload
c Safety tripping
c Adequately sized conductors. c Supports with good electrodynamic at high T c Tightness checks. c Temperature rise detection. c Adequately sized conductors. c Review choice of components and/or positioning. c Ventilation. Mounting and maintenance problems Installation design error
IEC634
Loose connections
c Uncertain upstream tripping c None c Tripping or indication
Conductor cross-section too small Device derating error or incorrect positioning
IEC898
Error in choice or use of device
IEC947
fig. 1: thermal problems in terms of cause and effect.
The problem in fact consists of ensuring, on switchboard design, that all its components will operate in temperature conditions that are less restrictive than those laid down by their construction standards. The scheduled current must obviously be able to flow through the connection switchgear (circuit-breakers, contactors, etc...) without any problem. In addition to safety of persons and equipment, two other objectives must be considered: c availability of electrical power (no untimely operation or failure to operate), c lifetime of components. In conclusion, the challenge consists of anticipating with a high degree of certainty the thermal operating state of the switchboard.
Three types of solutions can be identified: c the panel builder's experience, c the real tests for repetitive switchboards, c the use of software which can determine, according to envelope characteristics, the current strength/temperature pair for each heat source (switchgear - conductors) (see paragraph 4), in accordance with their position and with the temperature of the surrounding air. It is obvious that a software validated by experience and tests is of great use as it allows comparative study of the many possible installation configurations and thus optimisation of the future switchboard as regards thermal aspects and... cost.
2.2 Taking stock of standards

Many standards cover the Low Voltage area, for example the NF C 15-100 for France which defines the rules to be complied with for all LV installations. As regards definition and design of LV devices and assemblies, the following can be referred to respectively: c Switchgear standards, e.g. IEC 947. c The IEC 439 standard for LV cubicles (assemblies). The IEC 439 international standard is divided into three parts: v IEC 439.1 which contains the rules for type tested assemblies and for partially type tested assemblies, v IEC 439.2 which defines the rules for prefabricated ducts, v The IEC 439.3 draft standard which covers LV switchgear assemblies installed in places accessible to untrained persons. The part particularly of interest to us for LV switchboards is IEC 439.1 edited in 1985. In the European context, this standard acts as a structural framework for most national standards (British Standard, NF C, VDE...) whose contents are a fairly accurate copy of the text of the IEC standard in which differences correspond rather to the country's specific practices than to questioning of fundamental points of the IEC standard. In France this is the case of the NF C 63-410 standard. The main contribution of this standard has been a more accurate definition of two notions aiming at increased safety. These notions are: c That of Totally Tested Assemblies, TTA (type tested assembly) or of Partially Type Tested Assemblies, PTTA. c The notion of forms (see fig. 2). Without going into detail, we can say that the TTA correspond to products that are completely defined and frozen both as regards their components (exact drawings of each component) and manufacturing (assembly guide...) and which have to meet type tests (temperature rise, short-circuit, continuity of frames...) laid down by the standard. The PTTA correspond to assemblies whose basic structure is a TTA to which one or more modifications have been made which must be validated either by calculation or by a specific test. The notion of forms corresponds to a precise definition of the degrees of separation that can be found in a switchboard and which increase protection of persons by inaccessibility to live parts (busbars...). Four types of forms can be identified ranging from total absence of separation (form 1) to complete partitioning of the various switchboard elements (form 4). It should be noted that these partitions obviously greatly affect the thermal behaviour of these assemblies. The IEC standard also defines the temperature rise test to be verified on assemblies. It stipulates the conditions and temperature rise limits (paragraph 8.2.1. of the standard) that must not be exceeded by the assembly components. c Test conditions: v the assembly must be set out as in normal usage, v the current corresponding to the rated value is distributed in the various devices allowing for a diversity factor (Kd) varying according to the number of main circuits 2 i number of feeders i 3 Kd = 0.9 4 i number of feeders i 5 Kd = 0.8 6 i number of feeders i 9 Kd = 0.7 number of feeders u 10 Kd = 0.6 v thermal stabilisation is reached if the temperature variation does not exceed 1C/h. The cross-sections of the conductors connected to the devices must conform with the standard.
v the T measurements are performed using thermocouples v the reference ambient temperature is 35 C c Temperature rise limits Compared with ambient temperature, the following temperature limits must not be exceeded: v 70 K for terminals connecting external conductors, v 25 K for manual control devices, v 30 K or 40 K for accessible or inaccessible external metal surfaces, v specific values for built-in components and for insulators touching the conductors.
As concerns standardisation, a technical guide for the predetermination of these temperature rises is also available (IEC 890). However it requires validation by a number of tests as it does not have standard status. It provides correct results for simple configurations (envelope with few partitions, evenly distributed hear sources...). A presentation of this method is proposed in paragraph 7 together with a comparison with our cubicle designer approach.
Form 1
Form 2
Form 3
Form 4
fig. 2: various forms as in IEC 439-1/NF C 63-410 standards.
3 Thermal behaviour of a LV electric switchboard
An electrical switchboard is a system made up of a fluid (air) and of solid bodies in which electric current flow is accompanied by energy losses causing the temperature to rise. Progress towards thermal equilibrium involves the transfer of heat from live parts (devices,
conductors....) where it is generated, to the parts in contact with the exterior which in turn transmit this heat to the surrounding atmosphere.
3.1 Brief review of the main thermal phenomena

The thermal behaviour of any system, including an electrical switchboard, can be described in terms of heat exchanges. Three types of phenomena are involved: Conduction: Transfer of heat inside solid bodies (see fig. 3). This phenomenon can be divided up into: c Simple conduction where the body in question is not a source of thermal phenomena, e.g. conduction inside a wall. c Live conduction where heat is created inside the body in question, e.g. a copper bar with an electric current flowing through it. Calculations concerning the transmission of heat by conduction are based on Fourier's law which, for simple geometries, can be resumed by the equation: Ti Tj where d i j : heat flux between two points i and j in W, : thermal conductivity in W/m C, S: area of the heat exchange surface in m2, Ti , Tj : temperatures of the two points in C, d: distance between the two points in m, is characteristic of the conductive medium. Its value depends on temperature but in most cases is considered as constant. e.g. a few values of in W/m C Silver = 420 = 385 Copper = 203 Aluminium = 45 Steel = 0.2 Plastics Concrete = 0.935 = 0.657 Brick = 0.055 Glass wool = 0.026 Air (30 C) Radiation: Transfer of heat between solid bodies separated by a medium of varying transparency (see fig. 4). Such exchanges take place between the surfaces of any bodies facing one another and are represented by fairly complex relationships involving: c The emission of the solid which, if considered to be an ideal black body, depends only on its temperature. c The nature of the surface of the solid, expressed by its emissivity which reflects the relative ability of a surface to radiate energy as compared with that of an ideal black body under the same conditions. c Reflection and absorption phenomena.
i j =
fig. 3: conduction.
c The disposition of these surfaces via form factors. However in the special case where one surface (for example j) completely surrounds another surface (i) such that the ratio Si / S j is small, these expressions are simplified and we obtain: i = i Si Ti4 Tj4 where : heat flux transferred through the surface i in W, i : emissivity of the surface i, : Stefan-Boltzmann constant (5.67032 x 10-8 W m-2 K-4), Si : surface area in m2, Ti , Tj : temperature of opposite surfaces in K,
Convection: The general term of convection in fact covers two different phenomena which are frequently treated together. c Actual convection which corresponds to a transfer of heat between a solid body and a moving fluid. According to the origin of fluid movement, convection can be natural or forced (see fig. 5). These transfers are characterised by exchange coefficients hi : i = hi Si (Tf Ti ) where i : heat flux at the surface Si in W, hi : heat exchange coefficient in W/m2 C, Tf , Ti : temperatures of the fluid and of the surface of the solid body in C, From a physical viewpoint, the problem of heat exchange by convection is closed related to a fluid mechanics problem. However from a practical viewpoint it can be tackled simply using heat exchange coefficients with expressions involving: v parameters describing the type of fluid flow (velocity, etc.), v the physical properties of the fluid (thermal conductivity, dynamic viscosity, thermal capacity, density, etc.). They are often combined in the form of dimensionless numbers or characteristics (Nusselt, Prandtl, Reynolds, Grasshof numbers...). For example: expression of the heat exchange coefficient for natural convection and a simple geometry: flat vertical plate of height L with a uniform temperature distribution Nu h = where Dh Nu : Nusselt number, 0.25 Nu = 0.53 (Gr Pr ) where Gr and Pr are the Grasshof and Prandtl numbers respectively, functions of the physical properties of the fluid and of the temperature difference between the fluid and the heat exchange surface, : thermal conductivity of the fluid (W/m C), Dh: characteristic dimension (m). In most cases Dh corresponds to the largest dimension of the solid body in contact with the moving fluid, in this case L.
fig. 4: radiation.
fig. 5: convection.
fig. 6: convection currents.
NB: Note that the heat exchange coefficient depends on the temperature difference raised to a power of 0.25, hence: h = K( t)
0.25
c convection currents which transfer heat through a fluid by the actual movement of the fluid. This explains, for example, the temperature gradient observed between the top and the bottom of a volume of a closed fluid subjected to heating. The movement of air between two volumes is characterised by mass flowrates which are functions of flow cross-sections and flow velocities (see fig. 6).
Heat transfer is represented by: i j = M cp Ti Tj where i j : heat flux exchanged between i and j in W,
Di j : distance between the two points i and j in m. Moreover, if the fluid in question is assumed to have a perfect gas behaviour, then:
M : mass flowrate in kg/s, cp: heat capacity of the fluid in J/kg C, Ti , Tj : temperature of the fluid in volumes i and j (C). NB: heat transfer is imposed by the direction of flow. Expression of fluid velocity: in the case of natural convection, the fluid is set in motion between points i and j by the variation of its density with temperature. Velocity is thus assumed proportional to these variations, i.e. a function of the difference in temperature between i and j. Vi j = Constant
/ = Ti Tj hence
Vi j = Constant Ti Tj g Di j where =
(Ti + Tj ) / 2
(case of perfect gases)
Ti , Tj : temperature of fluid in K These formulae correspond to ascending or descending fluid volume movements. In the case of fluid movement near a wall, the problem is both thermal and hydraulic and can be solved analytically in some cases (laminar flow along a wall). In this case the fluid velocity along the wall has a similar expression, i.e. it is proportional to a temperature difference (fluid-wall). See page 25 for a review of the definition of C, K and F.
g Di j where
/ : relative variation of density, g: acceleration due to gravity in m/s2,
3.2 Exchanges at switchboard level

The diagram below (see fig. 7) represents the elements making up the system studied: ambient air, enclosure, internal air and the various heat sources. This description of the switchboard thermal state shows that all the exchange phenomena described above must be taken into consideration and are all considerably inter-related. For example: c The internal air temperature results: v from exchanges by convection between the internal air and the surfaces of the various devices, conductors and walls, v from the heat conveyed by the convective movements of air. c For the electrical devices in the switchboard, the heat generated by Joule effect is exchanged: v by convection between their heat exchange surfaces and the internal air, v by conduction with the bars and cables, v by radiation with the enclosure walls and the surfaces of the other devices. The most important phenomena involved in overall behaviour are the convection phenomena.
Ambient air Room walls v v v Internal air Devices v Enclosure v
Conductors, horiz. and vert. busbars
Conduction Radiation Convection Convective movement
fig. 7: thermal behaviour of an enclosure.
w w
v
w w w
4 Presentation of modelling
4.1 Principle
All the solution methods (e.g. Monte-Carlo, finite differences, finite elements) are based on a breakdown of the system to be modelled into elementary modules. The chosen method, nodal analysis, is derived from a finite difference approach. Although conventional, this technique has the advantage of being able to represent thermal behaviour of a complex system while allowing for the interactions between the various parts or components of which it is made. It can be used in a wide variety of applications, for instance to describe the behaviour of an artificial satellite, an electric motor, the climatic conditions inside a transformer substation or a building consisting of several rooms. In theory this method consists of breaking up the system in question into various isothermal volumes known as nodes. Each node has a number of parameters, including a temperature, and, in some cases, a heat input independent of the heat exchanges. We then examine couplings between nodes, i.e. the various exchanges between volumes which will allow us to write our balance equations (conservation of energy and mass in the volume element attached to a specific node). This approach is in fact a spatial discretisation of the system and results in the definition of a thermal network with its nodes, capacities, heat sources and conductances expres-sing the various couplings between nodes (analogy of electrical and thermal phenomena): (see fig. 8). We thus obtain a system of coupled equations, linear or non-linear, which will enable us to define a matrix, the thermal admittance matrix. We then have to specify the numerical values of the elements of this matrix which correspond to the thermal conductances. Expression of conductances per type of exchange: c Conduction: Gi j = i Si j / Di j c Convection: Gi j = hi Si j c Radiation: Gi j = S Fi j Ti + Tj Ti2 + Tj 2 c Convective movement: Gi j = M cp Expression of heat flux equivalent to electric current: 1 I = (U) R i j = Gi j Ti Tj where Gi j : energy flux between nodes i and j, Gi j : conductance between i and j, dependent on the type of exchange considered,
Ti , Tj : temperatures associated with nodes i and j respectively. As an example, let us model a room containing a heat source. This system is broken down into 4 nodes: 1 for the internal air 2 for the walls (internal and external) 4 for the external ambient air Nodal representation (simplified) (see fig. 9). Equations expressing the heat fluxes for this simple system: node 1: Q1 h1.2 S1.2 (T1 T2 ) + M4.1 cp (T4 T1) M1.4 cp (T1 T4 ) = 1 V1 cp1 T 1

Thermal quantities Temperature Thermal resistance Heat flux
= G (T2 T1)
Thermal capacity
Electrical quantities Potential Electrical resistance Current 1 I = (U U1) R 2 Electrical capacitance
fig. 8: Correspondence between thermal and electrical quantities.
1
w
Nod 1 : internal air Nos 2 and 3 : internal and external walls Nod 4 : external ambient air represents exchanges by conduction
)(
represents exchanges by convection represents exchanges by displacement of air represents the input of heat in node 1 represents the heat capacity associated with each
fig. 9: Simplified nodal representation - modelling of a room.
node 2: h1.2 S1.2 (T1 T2 ) = 2 V2 cp2 T 2 node 3:
2 S2.3
d2.3
(T2 T3 )
Therefore they can be ignored when only the steady state with stabilised temperatures is considered. Using these equations we then deduce the system of equations [G] [T] = [R] corresponding to: i j = Gi j Ti Tj
2 S2.3
d2.3
(T2 T3 ) h3.4 S3.4 (T3 T4 )

= 3 V3 cp3 T 3 node 4: h3.4 S3.4 (T3 T4 ) + M1.4 cp (T1 T4 ) M4.1 cp (T4 T1) = 4 V4 cp4 T 4

where: G: is the thermal admittance matrix T : is the vector of unknown temperatures R: is the vector of imposed conditions (heat sources Q1, temperature,...). This type of approach has made it possible to establish calculation codes and regulations relating to thermal problems in buildings.
d Ti NB: the terms Ti correspond to . dt
4.2 Modelling convection

As already mentioned in section 2, convection covers two phenomena which are treated together in most cases (exchanges between solid body and fluid and exchanges in the actual fluid). Modelling of exchanges by convection must therefore by divided into two parts. One part describes the mass flowrates (air movement) and the other the heat exchanges (heat exchange coefficient). The two parts are connected by the mass/thermal transfer dependencies (see fig.10).
w w
w w
w w
w
w
Air movements
Corresponding nodal diagram showing the two aspects of convection
fig. 10: mass and thermal modelling of convection.
4.3 Application to LV enclosures

Two main types of enclosures can be identified for modelling purposes: Non-partitioned enclosures (boxes, cubicles...). In this case the nodal diagram, shown in figure 11, resembles the diagram in figure 10, with integration of the heat sources. Highly partitioned enclosures with or without natural ventilation. There are two possible modelling approaches: c Each switchboard zone can be modelled as above and then these volumes are associated. However this results in overly large matrices bearing in mind that there can be a dozen zones to associate.
w w
w
w w
w
w
w
w
w
c A more global approach can be used without modelling the convection currents inside the various volumes and allowing only for air flows between zones (see fig.12). These approaches have resulted in different software for each enclosure type. These
programs are all structured in the same manner. Before describing in detail how to use the software (section 5), it is first necessary to further our knowledge of heat sources (busbars, devices) in order to determine the real operating currents of the devices installed in a switchboard.
fig.11: Non-partitioned enclosures.
zone A
zone B
fig.12: case of a partitioned enclosure.
ambient air
ambient air
w w
w
w
w w
opening
w w
zone A
zone B
5 Behaviour of heat sources and characteristics
The heat sources considered in modelling are busbars, connection conductors and electrical devices. The latter are considered to be black boxes dissipating calories instead of model modes. In
other words, rather than their operating temperature, we calculate the maximum current that they are able to convey for a given installation configuration so that they do not exceed their maximum operating temperature.
5.1 Busbars
Busbars are designed to satisfy two conditions: c Sufficient capacity to convey the required rated current without inducing a temperature rise in the bars that could damage the insulators supporting them. For example the bars can be sized so that they do not exceed a steady state temperature of 110 C; this value is completely dependent on the type of insulating materials with which they are in contact, for example the supports. The table in figure 13 gives a few busbar temperature values for an ambient temperature of 50 and 65 C. c Capacity to withstand a short-circuit current without serious bar deformation, rupture of insulator supports or excessive temperature rise. The second condition corresponds to a problem of electrodynamic forces and may be studied separately. However the first condition requires knowledge of the total of the currents flowing through the switchboard. The temperature of the air surrounding the bars is of particular importance in order to size the bars accurately and ensure that they do not exceed a critical temperature mainly depending on the type of material used for the supports. Consequently, knowing the air temperature in the various switchboard zones, we can determine, at the end of the program, the temperature of the bars according to their characteristics (dimensions, forms, arrangements...) and thus validate their sizing. NB: as regards calculation of heat flux, we consider that bars mainly dissipate power by convection and radiation with internal air.
Temp. near the bars (C) 50 50 50 50 65 65
Cross-section 1 b 100x5 1 b 100x5 3 b 100x5 3 b 100x5 1 b 100x5 3 b 100x5
Current (A) 1000 1500 1500 3400 1000 1500
Power loss (W) 45 107 10 61 45 11
Bar temperature (C) 79 109 65 110 92 80
fig.13: thermal values of a few busbars for different ambient air temperatures.
5.2 Switchgear devices

In power distribution cubicles, the switchgear devices used are mainly circuit-breakers. Together with the contactors and fusedisconnectors, they dissipate heat when electric current flows through them. The table in figure 14 gives, as a general indication, a few power loss values per phase (per pole). Note that the powers dissipated at a given In are of same order of magnitude for the different devices, although slightly lower for circuitbreakers as compared to fuse-disconnectors and even compared to contactors due to their hard but resistant contacts.
Circuit breakers In (A) Pw - fixed at In - withdrawable Fuse-disconnectors In (A) Pw at In Contactors In (A) Pw at In
250 17.4 23 250 30 265 22
400 25 35 400 44 400 45
630 21 54 630 67 630 48
800 36 58 800 _ 780 60
fig.14: Power loss at ln by conventional switchgear devices.
Electronic Voluntary "derating " Compensated bimetal strip
TN TN: nominal operating temperature TL: limiting operating temperature
TL
Simple bimetal strip Ambient T
fig. 15: typical derating curves of various releases as a function of temperature.
Let us examine thermal problems in greater detail for circuit-breakers: c Power loss is proportional to the square of the I current flowing through them: PW = PN In where PN represents the power loss at rated current I n .
2
Id
1,05 I n
c the rated current ( I n ) of a circuit-breaker corresponds to a specific ambient temperature, for example 40 C, set by the manufacturing standard. In fact, for some circuit-breakers, the ambient temperature corresponding to I n can reach and even exceed 50 C, which provides a certain safety factor in hot countries for example. c the operating current ( I ) can vary as a function of ambient temperature, according to the type of release: simple thermal, compensated thermal, electronic (see fig. 15), which may enable a maximum operational current other than I n to be defined. The parameters used to determine derating take the following into consideration, besides the temperature of the air around the device ( Ti ): c The limiting temperature ( TL ) of the circuitbreaker internal components: v maximum operating temperature of the bimetal strip for a circuit-breaker with a thermal-magnetic release, v temperature of the electronic components for a circuit-breaker with built-in electronic release v temperature not to be exceeded for the plastic parts most exposed in a circuit-breaker with remote electronics (external relay for an air circuit-breaker...). These limiting temperatures are between 100 and 150 C. c The ratio of the release I n and the real tripping current when the latter is placed at the temperature used to define I n
In Id (I rth)
fig. 16: time-current curve of a circuit-breaker.
K1 =
Id (see fig. 16) In
c The cross-sections of the connecting cables or bars which act as a radiator. Their influence is taken into consideration by a coefficient K 2 . NB: the cross-section of the conductors used rarely equals the cross-section used for circuitbreaker certification tests. The derating allowing for these criteria can be expressed in mathematical terms. Derating formula: The circuit-breaker and its connection conductors dissipate heat mainly by convection. This yields the relationship: W1 = h S (TL Ti ) where W1: power loss in W, h: heat exchange coefficient in W/m 2 C, S: heat exchange surface area in m 2, TL : temperature of the hot point in C (e.g. the bimetal strip), Ti : temperature of the internal air around the device in C,
h = Constant S (TL Ti )
0.25
(see 2)
1.25
hence W1 = Constant S (TL Ti )
The final relationship also integrating the effect of the cross-sections (coeff. K2)
When the device is in open air at 40C, the resulting relationship is similar. W2 = Constant S (TL 40) hence T Ti W1 = L W2 TL 40
1.25
I = In K1 K 2
TL Ti TL 40
0.62
1.25
c The data for circuit-breaker behaviour used in this formula are contained in files called by the software when temperatures in the cubicle are calculated.
Moreover, we know that W1 = RI 2 and W2 = RI 2 d T Ti thus I = I d L TL 40 where I is the current flowing through the device and I d = K1 In
0.62
6 Method for calculating temperature in envelopes and experimental results (see p. 21)
The modelling method described above acted as a basis for the development of our calculation method which enables us to determine the real operation of the switchboard (maximum current on each feeder...) and thus to optimise use of the assembly
and master dependability. As is frequently the case in thermal matters, the numerous relationships between parameters call for an iterative approach resulting in the drawing up of a program, the principle of which is presented below.
6.1 Principle
The program uses two overlapping iteration loops in order to determine the operating level of the envelope in steady state. One concerns resolution of the thermal problem, the other the derating coefficients. The calculation diagram is illustrated in figure 17. 1st stage: description of the configuration, i.e. the type of envelope used, the name and position of the devices. The program calls on the device file to retrive the data described above. 2nd stage: the envelope is broken down into isothermal subvolumes (nodal modelling nodes). 3rd stage: start of iteration loops with calculation of: c dissipated power (at the first iteration the derating coefficients are taken equal to 1), c the admittance matrix factors from the balance equations, c internal temperatures (resolution of the thermal problem), c the new derating coefficients, followed by a comparison with the above. If the difference is considered too large (iteration stop test), the new current strengths flowing through each device are calculated, followed by recalculation of dissipated power... 4th stage: the results are issued.
Configuration description studied w Power loss in the enclosure Current strength Derating cfficients
fig. 17: software operating principle.
6.2 Description of the data to be provided and of the results obtained

Data: c type of envelope (enclosure, cubicle, switchboard) and material, c protection index, c ambient temperature around the envelope, c number of rows of devices, c name of devices allowing search in file, c configuration of the switchboard and position of switchgear. Results: c choice of a horizonal and vertical (crosssection) busbar and current strength in these bars, c total thermal power dissipated in the switchboard, c derating coefficient for each device, i.e. currents flowing through, c if applicable, the temperature reached by the bars and its level in the various switchboard areas.
w
Internal temperatures
6.3 Modelled configurations

Naturally not all the installation configurations can be considered by this program. Only the most common ones have been selected, i.e. those which let us meet 90% of needs (see figure 18 which gives an example).
Configuration 1 no incoming device
Configuration 2 incoming device on top
Configuration 3 incoming device at the bottom
fig. 18: modelled configurations.
6.4 Results
This software approach is particularly advantageous as it lets us carry out the studies below: Detailed study of a specific configuration Made to optimise position of a device or choice of busbar, to know the power dissipated by the assembly, to size a suitable air conditioning... The following example concerns a column of a partitioned industrial power switchboard, form 2, containing: c a horizontal busbar supplying an incoming device and an adjacent column, c an 2500 A incoming device c various moulded case circuit-breakers. The program provides: c the derating coefficients Kdecl, c the currents flowing through each device, Ir. Remark concerning coefficient Kdiv: This coefficient enables us to take into account the diversity or bulk factor feeder by feeder, in other words, the operating levels at a specific moment of the various devices: e.g. at a specific moment, 2 feeders for example will be used to their full and the others at only 0.5 of their possibilities, with the resulting consequences on the thermal conditions of the assembly. The results are shown on the calculation sheet in figure 19. Derating table for a specific configuration This software usage possibility, similar to the above usage, lets us group, for a common configuration, the deratings of the various devices allowing for their real position in the switchboard, the conductor cross-sections used, the protective indexes and the external ambient temperature. An example of such a switchboard concerning devices installed in an industrial power switchboard column is shown in figure 20.
Masterbloc + MB 2000 IP = 31 Ambient temperature: 35 C Switchboard with incoming device on top supplied by the hor. busbar. Name of device M25 H C630H/D630 C630H/D630 C401N/D401 C401N/D401 C250N/D250 C250N/D250 Position 1 12 17 21 22 26 27 31 32 36 37 40 41 44 Kdecl .92 .92 .94 .98 .99 1 1 Kdiv 1 1 1 1 1 1 1 Ia(A) 2300 580 592 392 396 250 250 Ir(A) 2300 542 554 367 370 234 234
Hor. busbar: current - 2300 A cross-section - 3b 100x5 Vert. busbar: Cross-section: Cross-section: Cross-section: Cross-section: Cross-section: Cross-section: Cross-section: Cross-section: 4b 4b 3b 2b 1b 1b 1b 1b 80x5 80x5 80x5 80x5 80x5 80x5 80x5 80x5 Length Length Length Length Length Length Length Length 109 C (m): (m): (m): (m): (m): (m): (m): (m): .24 .5 .2 .2 .2 .18 .16 .24 Current: Current: Current: Current: Current: Current: Current: Current: 2300 A 2300 A 1758 A 1204 A 838 A 468 A 234 A 0 A
Hor. busbar temperature:
M 25 C 630 C 630
120 W:
Vert. busbar temperature: 100 C Total power loss: 2015 W devices: 613 W - auxiliaries: 0 W Vert. + tap-off busbars: 1282 W - hor. busbars:
Ambient temperature: 35 C Roof T: 69 C - Hor. busbar T: 74 C Device T: high - 61 C / low - 35 C Auxiliary T: high - 48 C / low - 35 C Vert. + tap-off busbars T: high - 67 C / low - 35 C Connection T: high - 53 C / low - 35 C
C C C C
400 400 250 250
w w w w
fig. 19: calculation result for a specific configuration.
IP 31 T amb M25 M16 M08 35 0.9 0.97 1 40 0.87 0.94 1 45 0.84 0.91 1 50 0.81 0.88 1 55 0.86 1 0.79
w
3b 100x5 3b 100x5 4b 80x5 1b 80x5
M 25 2500 A 3b 100x5 2b 80x5
IP 42/54 T amb M25 M16 M08 35 0.79 0.87 1 40 0.77 0.85 1 45 0.75 0.83 1 50 0.73 0.81 1 55 0.71 0.79 1
M 16 w 1600 A 2b 80x5 1b 63x5
w M 08 800 A 1b 63x5 empty
fig. 20: derating of the above circuit-breakers according to ambient temperature.
The derating coefficients are therefore drawn up, by excess, placing devices in turn on the top of the cubicle or compartment. See for example figure 21. Curves characterising the thermal behaviour of a type of envelope Two types of graphs have been drawn up: c A set of curves used to determine the mean temperature within a specific envelope as a function of the dissipated power and of the external ambient temperature.
See the curves in figure 22 concerning a nonpartitioned distribution cubicle type. c curves used to determine the watts that these envelopes can dissipate for a specific temperature rise, as a function of their dimensional characteristics. For example: ext. ambient T 35 C, required max. temperature rise v cubicle: height 2 m, width 0.9 m, depth 0.4 m dissipable power: 850 W v cubicle: height 2 m, width 0.9 m, depth 0.6 m dissipable power: 1000 W (see fig. 23.)
IP31 Tamb C125N/H C125L C161N/H C161L C250N/H C250L C401N/H 35 0.95 0.94 0.95 0.94 0.94 0.93 0.94 40 0.91 0.90 0.92 0.91 0.90 0.89 0.91 45 0.88 0.86 0.88 0.87 0.87 0.86 0.87 50 0.84 0.83 0.85 0.84 0.83 0.82 0.84 55 0.80 0.79 0.82 0.82 0.80 0.78 0.81
IP 42/54 35 0.82 0.80 0.81 0.79 0.82 0.79 0.79 40 0.79 0.77 0.78 0.76 0.79 0.76 0.76 45 0.76 0.74 0.76 0.73 0.76 0.73 0.74 50 0.72 0.71 0.73 0.70 0.72 0.70 0.72 55 0.69 0.68 0.69 0.67 0.69 0.67 0.69
Masterpact Compact
fig. 21: derating of Compact circuit-breakers placed under the incoming circuit-breaker.
Mean temperature in C 100 90 80 70 60 50 40 30 20 10 Tamb: 60 Tamb: 55 Tamb: 50 Tamb: 45 Tamb: 40 Tamb: 35
C C C C C C
Tamb: 25 C
Enclosure dimensions: height: 2 m width: 0.9 m depth: 0.4 m
100 200 300 400 500 600 700 800 900 10001100 Power loss Watts
fig. 22: mean temperature of air inside an IP2 form 1 metal distribution cubicle.
400 mm deep enclosure Power dissipated in Watts 1600 1400 1200 1000 800 600 400 200 800 900 1000 1100 Width in mm T = 20 C T = 10 C T = 30 C T = 40 C
600 mm deep enclosure Power dissipated in Watts 1600 1400 1200 1000 1000 600 400 200 800 900 1000 1100 Width in mm T = 10 C T = 20 C
T = 40 C T = 30 C
fig. 23: power that can be dissipated by an enclosure for a specific temperature rise according to its width. Curves refer to a metal cubicle, form 1, 2 m high.
6.5 Experimental results

Temperature rise tests have been conducted in the ASEFA Ampre laboratory on various envelope types: metal and plastic enclosures, Prisma cubicle, Masterbloc distribution switchboards. During these tests the following measurements were taken: c Temperatures: v of air in the various envelope areas, v of conductors: busbars and branch-offs, v hot points in devices (bimetal strip, electronic ambient). c Current strength. c Parameters used for modelling, particularly air/ wall heat exchange coefficients. These measurements have enabled both verification of conformity with IEC 439.1 standard of certain values (see temperature rise limits mentioned in paragraph 1.2 on standards) and validation of this model. With respect to air temperatures, the difference between the values measured and the values calculated depends on the type of envelope modelled, since modelling approaches differ according to whether or not the envelopes are partitioned. Out of all the tests carried out on switchboards of various forms (partitioned or not), the maximum differences observed were always less than 6 C. The temperatures calculated for the busbars also show satisfactory agreement with the measurements and enabled us to validate the software. As regards current strengths, differences are on average less than 5%. Consequently, for a recent official approval of a Masterbloc switchboard configuration in temperature rise, the software allowed us to determine the operating level of the switchboard.
8 Method proposed by the IEC 890 report
Not so long ago a large number of electric cubicles were chosen and equipped/filled in the light of experience. This concerns the filling ratio and evaluation of temperature in the cubicle in operation. For example, the maximum external temperature of 30 C and maximum internal temperature of 60 C (switchgear manufacturers give derating up to 60 C). This practice resulted in unoptimised use of the equipment, untimely tripping of the protective devices or the need for operators to operate with open doors. The method proposed by the IEC report, even if this is rather a guide than a standard, thus merits attention. It is described in detail in the report of the IEC 890 or in the appendix of the NF C 63-410. We shall review the basic aspects, show its limits and compare it with the method presented in the Cahier Technique. In theory this method applies to envelopes for which the following assumptions can be made: c even distribution of dissipated power, c switchgear arranged so as not to obstruct air circulation, c no more than 3 horizontal separations. Necessary data: c dimensions of the envelope, c power dissipated in the envelope (switchgear, conductor), c type of installation (insulated envelope or insulated at one end...), (see fig. 25). Calculation: Temperature is calculated only at 2 points of the envelope: at mid-height 0 T0.5 = Ta + T0.5 where T0.5 = d k PW.804 c d is a coefficient taking into account the presence of horizontal separations. v if Ae < 1.25 m2, d = 1 (definition of Ae, see below) v if Ae > 1.25 m2, d = 1 with and without ventilation apertures for 0 separation d = 0.5 with and without ventilation apertures for 1 separation d = 1.10 or 1.15 if ventilation apertures for 2 separations d = 1.15 or 1.30 if ventilation apertures for 3 separations c k is a constant characterising the envelope: its value is determined on charts, (see fig. 24).
Effective cooling surface Ae in m2
Enclosure constant k 0.38 0.36 0.34 0.32 0.30 0.28 0.26 0.24 0.22 0.20 0.18 0.16 0.14 0.12 0.10 0.08
1 1.5 2 2.5 3 4 5 6 7 8 10 12 14
100
200 300 400 500 600 Ventilation apertures section in cm2
fig. 24: Enclosure constant k for enclosure with ventilation opening and an effective cooling surface area of Ae > 1.25 m2.
Temperature distribution factor c 1.65 1.6 1.55 1.5 1.45 1.4 1.35 1.3 1.25 1.2 1.15 1.1 1.05 1 2 3 4 5
1 2 3 4 5 6 7 8 9 10 11 12 13 Factor f Curve/Installation type 1 Separate enclosure, detached on all sides 3 Separate enclosure for wall-mounting 2 First or last enclosure, detached type 3 Central enclosure, detached type 5 Central enclosure, wall-mounting type 4 Central enclosure for wall-mounting and with covered top surface
fig. 25: temperature distribution factor c for enclosures without ventilation openings and with an effective cooling surface Ae > 1.25 m2.
k is a function of the heat exchange surface of the envelope Ae (m2). Ae = A0 b where A0 is the geometric surface of the various envelope walls. b is a constant allowing for the type of wall and type of installation. Values of b: v exposed upper part b = 1.4 v covered upper part b = 0.7 v exposed side surfaces b = 0.9 v covered side surfaces b = 0.5 v side surfaces of central envelopes b = 0.5 v lower part b=0 c Pw power dissipated in watts at the top of the enclosure: T1 = Ta + T1 where T1 = c T0.5 where T0.5 represents the above temperature rise c c is a temperature rise constant determined from charts
Example of a chart, see figure 25 c is function of Ae and of one of the two factors, f or g f = h 1.35 / (L P) if Ae > 1.25 m2 g = h 1.35 / L if Ae < 1.25 m2 Limits: The main limits of this method are that it: c applies only to non-partitioned envelopes of the cubicle and enclosure type and not to highly partitioned power switchboards. c does not take into account the position of the heat sources which in most cases are not distributed evenly. Comparision with our approach We observe that both approaches yield similar results for non-partitioned cubicles with distributed heat sources (see curves in figure 26). As regards highly partitioned envelopes, the location of the heat sources and the exchanges between the various areas considerably affect temperature rise!
Temperature in C 90 80 70 60 50 40 30 20 10 Temperature of ambient air 35 C Enclosure dimensions: height: 2 m width: 0,9 m depth: 0,4 m 100 200 300 400 500 600 700 800 900 1000 1100 Power loss in watts Temperature calculated as in IEC 890 report Temperature calculated with MG software
fig. 26: Air temperature at mid-height of an IP2, form 1 metal distribution cubicle.
8 Conclusion
The importance of electric switchboards in distribution is an established fact. At a time when availability of electrical power and operating dependability are absolutely vital, thermal mastery of electric switchboards is a fundamental goal. Standards concerning envelopes and products specify the thermal limits not to be exceeded. All that was left was for professionals to become "thermal architects" in design of envelopes and electric switchboards. This has now been achieved, even for partitioned switchboards. Reminder: definition of the various temperature scales: c degree Celsius (formerly centigrade) C: relative temperature Reference points : v 0 C: temperature of melting ice v 100 C: temperature of boiling water at normal atmospheric pressure. c degree Fahrenheit F: unit used in English speaking countries: Reference points: v 32 F: temperature of melting ice v 242 F: temperature of boiling water at normal atmospheric pressure Equivalence Conversion 1F = T F = 5 C = 0.55 C 9
T C + 32 0.55 c degrees Kelvin K: international system unit. Absolute temperature scale, since its definition relies on exact phsyical bases. Same graduation as the Celsius scale, but the origin is offset: the temperature of melting ice corresponds to 273 K Conversion: T K = T C + 273
Schneider
Direction Scientifique et Technique, Service Communication Technique F-38050 Grenoble cedex 9 Fax. (33) 04 76 57 98 60
Real : Sodipe (26). Printing : CLERC Fontaine - 1500 - 100 FF-
87473
12-97
1997 Schneider
n 148
high availability electrical power distribution
Guy Gatine
Merlin Gerin engineer since 1982. In 1965, in charge of research on means of testing semi-conductors for the Radiotechnique company. In 1966, he became a research manager in the Merlin Gerin automation branch, and then worked in the SES (Electronic Safety Systems) department. Presently, still in the SES department, and enriched by his long-standing experience in the nuclear and military fields, he directs the consulting and design of high level quality and availability installations.
E/CT 148 first published october 1991
glossary
ASEFA, French Equipment Testing Station Association, part of the RNE: National Testing Network. s LCIE, Central Electronics Industry Laboratory. s MDT, Mean Down Time: mean failure duration. s MTBF, Mean Time Between Failures: mean operation time of a device in between two failures. s MTTF or MTFF, Mean Time to First Failure: mean proper operation time before the first failure occurs. s UPS, Uninterruptible Power Supply: comprising a battery charger, battery and inverter.
s
Cahier Technique Merlin Gerin n 148 / p.2
high availability electrical power distribution
contents
1. Introduction 2. Designing a dependability system Specifying Constructing Demonstrating Distribution circuits Diesel generator set Power source changeover devices "Short-term" back-up Electronic control-monitoring system Operating criteria Search for and identification of weak points Knowing the level of component reliability Choosing technologies Failure tolerance Running the installation Specification Construction Maintenance arrangements Demonstrating specified availability p. 4 p. 4 p. 5 p. 5 p. 6 p. 7 p. 7 p. 8 p. 8 p. 8 p. 8 p. 9 p. 9 p. 10 p. 12 p. 13 p. 16 p. 16 p. 17 p. 17 p. 18 p. 19
The objective of this Cahier Technique is to explain to electrical installation designers how to design electrical power distribution systems that fulfil the objective of continuous voltage supply across the feeder terminals. In other words: how to master power availability so as to achieve: "the operating DEPENDABILITY objective".
3. Description of a "backed-up" installation
4. Solutions for increasing availability
5. Example of increased availability backed-up distribution
6. Conclusion 7. Bibliography
1. introduction
Operating dependability is a fundamental characteristic of all systems, installations and products. It is determined by design and use. Dependability describes the aptitude of a system to operate properly throughout its service life. Proper operation implies: s not breaking down (reliability), s not experiencing dangerous failures (safety), s being in good operating condition as often as possible (availability), s being quickly repairable (maintainability). Whatever the system and the efforts implemented in its design and use, the level of dependability is a concrete reality. It must be:
s taken into account starting at the design phase, s observed a posteriori: by counting the operating hazards that occur during installation operation.
Electricity, a modern source of energy, contributes to the level of dependability through the fact that it is needed for operation. Its availability, or rather its unavailability, has increasingly important consequences on companies competitiveness: s in industry, lack of power causes production losses, s in the service sector, lack of power causes computer blockages and utility shutdowns (lighting, heating, lifts...).
The more complex the systems, the higher the risk that even a brief power failure will have major consequences. Safety and availability have been particularly well developed and mastered in previous years in fields such as nuclear, military and space. Nowadays, energy availability is a definite concern with regard to intelligence, monitoring of the most widely varied systems and, to an increasing extent, with regard to the power supply of those same systems. Electrical power installations, especially those containing sensitive feeders, must be designed so as to limit the occurrence and consequences of failures in the public distribution network (referred hereafter in the booklet as the mains).
2. designing a dependability system
Beginning with a simple, minimal system, the design approach that has been adopted highlights the strong points and weak points of an electrical supply system, sometimes called a power station. The weak points are then reinforced: s increased sturdiness and quality of constituents, s redundancy of equipment (duplication, triplication...). The design is therefore optimized with a view to achieving the required level of dependability: the effort employed in design concern only the weak points of the system. This approach necessitates the use of a rigourous design methodology together with dependability techniques. The design phase (cf. fig. 1) takes place in three stages:
specify
demonstrate
user/designer dialogue
construct
fig. 1: design method.
specifying, designing/constructing, s demonstrating. The dependability of a system, based on specifications, is well illustrated by the very definition of dependability as used by task forces specialized in operating dependability, IFIP at the world level and EWICS at the European level: the Quality of the service supplied is such that the user has justified confidence in it. The design of a dependability system therefore requires that the expected service be specified (knowing the need), that this service be constructed (quality of design), and that it be demonstrated that the solution complies with the dependability specification (justified confidence).
s s
specifying
Specification of dependability constraints enables the target to be identified and the right amount of effort to then be devoted to design. This stage has a decisive effect on the system. Specification can be based on: s the history of malfunctions in similar installations (existing power stations), s standards (e.g. MIL) or recommendations, s economic analyses establishing the cost of installation down time (direct and indirect consequences) as a result of failures,
s identification of the most dreaded events. Dependability is a generic concept encompassing four criteria: s reliability, s safety, s availability, s maintainability. Cahier Technique n 144 Introduction to dependability design gives, among other things, a precise, official definition of these terms. Using these criteria, the specifier establishes the dependability characteristics for his installation, based on these four criteria which are naturally quantifiable. Through dialogue with the customer, the most dreaded events are determined, together with the acceptable probability of the occurrence of such events according to the seriousness of the consequences thereof.
constructing
Once the dependability objectives have been established, the dependability system (how to prevent the occurrence of failures and how to master them) is constructed. The means of doing so are listed below. s quality: a dependability system is above all a quality system (failure avoidance) Quality must be taken into account at two levels: s quality of design, so as to guard against design errors (project team, quality assurance manual, audits...),
quality of constituents, so as to guard against failures (sturdiness, qualification). s surviving failures (failure tolerance) The sturdiness and quality of the system are not sufficient criteria to guarantee its dependability. Certain functions are critical with respect to the assignment to be accomplished: the failure of a single component can bring about a loss of the power supply. The system must therefore be designed so as to respect dependability objectives in spite of the failures that may occur, this being generally achieved through redundancy or the use of special technology (example: failure-oriented logic in electronics). In order to survive failures, it is essential to detect the faulty function. It is then necessary to: s orient the failures so that they do not affect the assignment (technological barriers), and then s mask the failures through the parallel operation of several units (even though only one would suffice), thereby enabling operation to continue with equivalent (standby) equipment. In order to employ the right amount of effort in terms of failure avoidance and/ or failure tolerance, measurements or calculations of the efficiency of such arrangements are carried out to directly evaluate the design and adapt the system architecture to best fit the cost. This approach is constructive: the initial architecture is the simplest possible, minimal one (only relevant functions are taken into account); the architecture is then enriched according
s
to the results of the dependability evaluation so as to attain the target set during the specification stage. Two iterations, implementing the study phases described in figure n 2, are generally needed to design a system to the dependability requirements. s the first iteration consists of: s consolidating the dependability requirements, s establishing by means of a functional analysis method the simplest possible, minimal initial architecture, s evaluating the degree of dependability of this architecture, s proposing a certain number of corrective measures relative to the design so as to comply with the dependability requirements. s the purpose of the second iteration is to: s reassess the level of dependability of the corrected architecture,
conclude (or not conclude, in which case the process needs to be reiterated) upon the validity of the architecture with respect to the dependability objectives.
s
demonstrating
In order to achieve justified confidence, the customer must be given proof that the dependability level complies with the specified objective. This is done by means of two techniques: s elimination of design-related failures: debugging, tests, environmental testing... s prediction of failures so as to measure the risk (probability) incurred during the system operating life. Breakdown prediction involves dependability studies which, through modelling and evaluation, estimate the
simple, minimal architecture
evaluation of dependability the result is...
correct
too good!
slightly insufficient
insufficient
validation of architecture
failure tolerance: reduce redundancy and/or failure avoidance: use less sturdy components
failure avoidance: improve component quality
failure tolerance: choose partial or global redundancy
presence, creation and consequences of failures. Predictive dependability studies are carried out using a set of modelling methods (FMECA - Failure Modes Effects and Critically Analysis method, failure tree, Markov graph). Quantitative evaluation is based on analysis of similar equipment having experienced problems in industrial operation and/or on the results of analyses recorded in reliability reviews (CNET, IEEE...). Dependability studies make it possible to achieve justified confidence in the installation. In the simplest electrical power distribution diagram, with power from the mains (cf. fig. 3), the level of availability of one of the feeders cannot be higher than the network level. Considering that a mains failure incorporates the following criteria: s out-of-range voltage, s phase loss, s harmonic distortion (in the case of power supply for sensitive systems such as electronic systems). The average level of unavailability of the French Electricity Board (EDF) mains is in the vicinity of a cumulative total of 7 to 8 hours per year (i.e. an unavailability rate in the range of 10-2 according to TDF observations), essentially due to the environment (e.g. storms). It is therefore evident that if one wishes (specification) to improve the level of unavailability, to 10-4 for instance, it is necessary to provide for an architecture that is more than a mere radial feeder system, and more like an improvement on the basic diagram as illustrated in figure n 4.
mains
new architecture
fig. 2: approach to employ the right amount of effort to dependability.
fig. 3: energy availability on one feeder cannot be higher than that of the source.
3. description of a backed-up installation
distribution circuits
(cf. fig. 4) These circuits comprise essentially: s in Medium Voltage: s protection for the Medium Voltage (MV) incoming feeder, s MV/LV transformer; s in Low Voltage: s a main circuit breaker that protects the switchboard as a whole and eliminates the risk of inadvertent connection of the diesel generator set to the mains, s equipment for the protection of people and property against insulation faults; s feeder group power circuit breakers that distribute power, these breakers: s opening each time there is a power source changeover, s closing simultaneously if supplied by the mains, s closing in sequence if supplied with back-up power by the generator; s a power source changeover (mains/ generator) controlled by the mains/ standby voltage monitoring relay; s power source changeover that switches to the short time back-up source (UPS), generally a static contactor.
EDF (Electricity Board)
GEN.
insulation monitoring
power feeders UPS priority power feeders
increased availability rate feeders
fig. 4: schematic diagram of backed-up electrical power distribution.
diesel generator set

(cf. fig. 5) This equipment includes: s a diesel power motor suited to the power needs of the application. It is equipped with auxiliary circuits: s a starting circuit including one or two starting chains (cf. chap. 4 choosing technologies); each comprising a starter and a battery with a charger; s a gas oil circuit including: - a so-called daily tank, with a maximum capacity of about 500 litres (depending on the generator power rating), - an outside tank with a capacity calculated according to the maximum required motor autonomy,
gas oil circuit outside tank
cooling circuit
daily tank
regulator lubrication circuit
oil reservoir
diesel engine
alternator
starting circuit
starter
battery
charger
fig. 5: description of a diesel generator set.
- an automatic gas oil pump backed by a hand pump enabling the daily tank to be filled from the outside tank. This pump is not necessary if the daily tank is installed above the motor at a height calculated according to the pressure imposed by the injection circuit; s a pre-lubrication and lubrication circuit fitted with an oil reservoir calculated according to the motor autonomy chosen in order to fulfil special Electricity Board peak consumption compensation tariff requirements; s an air or water cooling circuit depending on the type of motor. In the case of air-cooled generators, the motor is cooled by a fan driven by the motor shaft, either directly or by belts. In the case of water-cooled motors, the inclusion of an exchanger (primary and secondary circuit) and an air cooler entail the use of circulation pumps and a fan. s a power alternator suited to the need, fitted with a voltage regulator. The alternator reactance rates should comply with the type of load (reactive, capacitive, electronic system...). For example, an application comprising 50 % of the load in the form of uncoupling battery rectifier-chargers entails the use of an alternator with a subtransient reactance rate of about 8 % in order to limit voltage distortions.
s detection of a mains anomaly (frequency, wave form, out-of-range rms value). In both cases, temporary coupling to the mains prevents power cuts due to generator takeover time.
not play a direct active role in the system.
operating criteria
A distribution system of this type, with a service life of 20 years for example, should ensure the electrical power supply when the following mains faults occur: s mains loss, s out-of-range mains voltage, s out-of-range phase unbalance. Furthermore, it should also fulfil utility tariff constraints such as: s peak consumption compensation, s additional power beyond subscribed power. The operation of each item of equipment comprising the station is linked to its role in the station and is defined as follows: s out-of-range mains voltage, s the diesel generator set operates as follows: - mains failure : 200 h/y - tariff constraints : 400 h/y - testing : 50 h/y i.e. maximum cumulative operation total : 650 h/y s the low voltage switchboard operates as follows: - standby position : 8 % of the time - mains position : 92 % of the time s the short-term back-up source takes over: s during very brief power cuts (microcuts) that vary in number depending on the power supply network and the environment, s during the phase in which the diesel generator set takes over the power supply, to which must be added low voltage switchboard switching times. 10 minutes autonomy is ordinarily required of batteries at the end of their service life, 5 minutes being the minimum, s during battery test cycles. Very brief in duration, these are negligible in relation to the takeover phases. Maintenance time should be provided for. The selected availability guarantee is also linked to repair time. These times and the related means depend on the chosen level of dependability.
short-term back-up (UPS)

This function, comprising the Uninterrupted Power Supply (UPS), is fulfilled by one or more UPS with unit power of 40 to 800 kVA or more, equipped with a control-monitoring device, a battery and a diagnostic device that communicates via an asynchronous link. These types of UPS can be installed in parallel. Battery autonomy should be sufficient (cf. table in figure n 6) to supply power to the application during the time required for the generator to takeover as the long term back-up source. This takeover sequence includes: s mains actually out : 20 s s generator starting taking into account a startup at the last attempt : 50 s s Mains/Standby changeover (load shedding, then changeover) : 20 s s restoring of priority circuit breakers : 210 s for a total sequence time of 5 minutes (100 s for the highest priority feeder).
power source changeover devices

These devices make it possible to switch to the diesel generator set. Aside from mains losses, switching or coupling can prove to be necessary in the two following cases: s failure of the short-term back-up source with the mains on,
electronic controlmonitoring system

This system is a federation of electronic control or monitoring units (CU), each of which runs one of the main installation constituents (generator, source changeover,...). These CU are combined with one or more supervision units (SU) which enable man-process dialogue but do
battery autonomy (mn) at the end of battery life at the beginning of battery life target unavailability levels for the installation as a whole:
10 15 10-6
10 15 5.10-6
7 10 10-5
5 8 10-4
fig. 6: UPS battery autonomy according to unavailability level.
search for and identification of weak points

The minimal basic architecture is analyzed, taking into account: s feedback on experience from various sources, s failure rates established by manufacturers and standardization bodies such as IEEE, MIL and CNET which allow the weak points of this type of installation to be established. The failure probabilities for the main constituents of the installation, expressed in terms of the number of minutes of failure per year, are for example: s MV mains = 450 mn/y, s LV switchboard = 90 mn/y, s diesel generator set = 360 mn/y, s short-term back-up = 150 mn/y. The weights that the components of each of the above bear on unavailability are as follows:
main low voltage switchboard power source changeover : 65 % s distribution equipment : 25 % s auxiliary and control-monitoring : 10 %
s s
short-term back-up rectifier and frequency converter : s batteries : s auxiliaries :

s s
35 % 55 % 10 % 100 %
100 % diesel generator set s starting chain : s cooling circuit : s fuel circuit (gas oil pump) : s generator load takeover : s generator environment (e.g. temperature) : s auxiliaries + control-monitoring :
s
65 % 8% 7% 6% 6% 8% 100 %
It is easy to observe that the three sensitive constituents are: s the LV switchboard power source changeover, s the generator starting chain, s the short-term back-up battery.
4. solutions for increasing availability
The minimal basic architecture (cf. fig. 4) studied above, produces a maximum unavailability rate of about 5 h per year (i.e. 6.10-4) with minimal back-up battery autonomy of 10 mn, and preventive and corrective maintenance requiring no assistance (cf. fig. 12). The distribution of failure probabilities is expressed in terms of minutes of failure per year. If the targeted unavailability is less than 1 h/y (10-4) on the feeder backed up by the UPS, improvements need to be made to the basic architecture and/or components. This is possible by: s ensuring key component reliability; s choosing the appropriate technologies and techniques;
s having a fine division of operation in the aim of: s enabling stepped operation (modularity), s ensuring operation by only the components required; s redundancy.
As a result, the various mechanical, electrical and electronic constituents must be chosen according to quality and reliability levels, taking into account the thermal, climatic and mechanical environments, this being particularly true for components that bear substantial weight on unavailability. Debugging can be employed to bring out latent defects that are liable to appear in the operating environment, without affecting the quality of the components nor causing wear. When the components are not certified, qualification bodies can be called in, such as the LCIE for electronics or the ASEFA test stations for electrotechnical components.
knowing the level of component reliability

The reliability of a system (mechanical, electrical and electronic) is its aptitude to perform a required function, under given conditions, during a given period of time; it is the probability of system survival (cf. Cahier Technique n 144 Introduction to dependability design).
The table in figure n 7 summarizes the main technical choices influencing availability.
choosing technologies
For each constituent (LV switchboard, Generator, Short-term back-up), the choice of the various technologies plays a major role in reliability and maintainability. Low voltage switchboard (LVSB) Although the equipment comprising the LVSB accounts for only 20 % of system availability, it should be chosen with care. s choice between fuses and circuit breakers: s fuse: this short-circuit protection device is no longer justifiable at present in dependability installations due to the maintainability constraints it imposes. s circuit breaker: apart from customized protection settings, it has a very low MTTR (Mean Time To Repair, actually reclosing time) and should therefore be used whenever a good level of energy availability is required. s choice between contactors and remote control circuit breakers: s contactor a durable control device; the device is closed when its coil is being supplied with power and open when it is not. It is said to be monostable (i.e. a single stable position: open), s remote control circuit breaker: this device is of the bistable type, i.e. it maintains its closed or open position in the event of a voltage drop. Circuit breakers are therefore chosen for high availability stations so that the control position prior to power supply or electronics losses will be maintained.
s protection devices If only the faulty feeder is isolated by the circuit breaker located immediately upstream from the fault, and if the feeder is isolated by that circuit breaker alone, this being the case for all fault values ranging from overloads to shortcircuits, there is said to be discrimination.
Discrimination contributes to continuity of service, and hence to energy availability. Choosing the appropriate discrimination technique is therefore of some consequence. s amperage level discrimination: this technique utilizes instantaneously operating circuit breakers. The scaling of settings according to short-circuit current values can provide partial or total discrimination; s time-based discrimination: this technique involves the scaling of operation times for circuit breakers fitted with tripping devices with adjustable short and long timer settings. Discrimination is total. However the constraints and the destructive effects caused by short circuits during time delays can be considerable and can reduce maintainability; s the SELLIM system (cf. Cahier Technique n 126) combines total discrimination requirements with the advantages of strong short-circuit current limitation. Also to be cited are the Logic Discrimination System used especially in Medium Voltage (cf. Cahier Technique n 2) which provides total discrimination with delay times reduced to a minimum.
s fixed or withdrawable equipment A choice needs to be made between fixed circuit breakers that require switchboard de-energizing in order to be changed and withdrawable breakers which can be replaced with the power on. When choosing a remote control circuit breaker that will have a high rate of operation, it is advisable to select a withdrawable circuit breaker. It should also be ensured that the system can evolve; for example that the addition of control-monitoring auxiliaries would be possible. It is important to seek the most suitable balance between equipment cost and MTTR. For availability levels greater than 10-4, withdrawable equipment is recommended because of the following elements: withdrawal (base + circuit breaker): - MTBF = 100 years, MTTR = 1 hour, - circuit breaker unavailability = 3.4.10-6 fixed: - MTBF = 100 years, MTTR = 24 hours, - circuit breaker unavailability = 2.4.10-5
Diesel generator set s starting system: this is the sensitive point; it can be pneumatic, connected to a compressor, or electric, connected to a rectifier/charger and battery. The elements involved in choosing between a pneumatic or an electric starter are the following (the choices made are shown in the table in figure n 7): s electric starter: Advantages: - simple to supervise, - simple to install for generators with power ratings < 500 kVA,
levels of unavailability required for the installation as a whole: type of starter (P > 1MVA) electric + pneumatic pneumatic only fuel supply by the force of gravity (or two pumps) with a single pump lubrication circuit (depending on the motor) with added oil with prelubrication
10-6
s
5.10 -6
s
10-6
10-4
s s s s s s s s
fig. 7: choices of technologies for a diesel generator set.
- no effect on motor ageing, - simple to maintain; Drawbacks: - monitoring the starter battery is a delicate matter, - inoperative when mechanical starter ring positioning faults occur; - large size for power ratings > 1 MVA, - installation constraint: the battery must be near the motor; it is often of the maintenance-free type and must be capable of sudden discharges. s pneumatic starter: Advantages: - simple to supervise circuit starting, - lower cost and smaller sizes for generator power ratings > 500 kVA; Drawbacks: - supervision of compressor is a delicate matter, - corrective maintenance can be long and delicate. s taking the environment into account The ambient temperature of the generator as well as altitude can reduce generator performances. As an example: s an ambient temperature of 40 C will bring about a declassification of 10 % (rated temperature 25 C), s an altitude of 2,000 m will cause a declassification of 25 % (rated at 100 m). These declassifications are functions that are proportional to the variable and lead to motor oversizing and oversupply. Too low a motor idling temperature (< 15 C) can cause the motor to stall when taking on a load. It is possible to remedy this by installing a preheating circuit on the oil and water circuits for water-cooled motors, or on the oil circuit for air-cooled motors.
It is also possible to stagger the resupply of electricity to the circuits, starting with the highest priorities. Short-term back-up (UPS) This function, fulfilled by an uninterrupted power supply (UPS) largely contributes to the objective of power station availability. Four criteria are to be taken into consideration in establishing the optimal configuration for short-term back-up: s power used in normal operation, s instantaneous load variations (load side), s availability level desired, s autonomy required. The choice of technology includes various elements that enable the UPS to operate properly: s supply-side and load-side protection devices, s connection cabling, s battery supply. Regarding protection devices, particular attention should be paid to the setting of overcurrent devices (magnetic and thermal circuit breaker trip mechanisms) since: s current peaks frequently occur during switch-on, s UPS have reduced short-circuit power. It is therefore necessary to check: I current peaks < I protection limit < Isc. As for the equipment for protecting (people) against insulation faults, unearthed neutral systems should be chosen whenever possible since there is no tripping when the first fault occurs. For batteries it is advisable to: s choose a technology that facilitates maintenance: lead-sealed battery or maintenance-free lead battery;
s provide access enabling quick replacement. The type of operation and short-term back-up configuration should correspond to the level of availability required for the planned application: s n 1: continuous on lineoperation of the UPS is preferable to off line operation and is imperative when the UPS protects against micro-cuts. With off line operation, the UPS only supplies power with the mains off. With on line operation, the mains are back-up for the UPS when overcurrent or a static power supply failure occur. The elements supplied by the UPS are then backed up directly by the mains through the static contactor (SC). s n 2: several static power supplies coupled in parallel, with no redundancy and no use of a back-up network; this configuration allows suitable distribution according to the power required by the backed-up equipment, and stepped operation according to static power supply availability. s n 3: several static power supplies coupled in parallel, with redundancy and without the use of a back-up network; this configuration offers greater availability than the two previously described, availability depending directly on the level of redundancy. s n 4: several static power supplies coupled in parallel, one of which is redundant with the use of a back-up network; this configuration offers greater availability than the previous one for a small additional cost. The table in figure n 8 gives an indication of MTTF values for different configurations.
configurations one static power supply two static power supplies in parallel wothout redoundancy two static power supplies in parallel with 1/2 redundancy
without a back-up network 31,000 15,500 250,000
with a back-up network: mains quality good quality 183,000 261,000 112,000 177,000 411,000 450,000
fig. 8: MTTF values in hours for different configurations (factory-debugged equipment).
Control-monitoring electronics The electronics have the role of managing each function in the power station. So as to obtain the greatest possible level of reliability, it is wise to select the following options: s high integration level, use of highly integrated components such as microcomputers for the supervision function and a micromonitor for the control-monitoring unit; s division of functions at both the control-monitoring and supervision levels, two examples being: on the control unit, separating the interface parts (sensors-actuators) from processing, and on the supervision unit, separating the processing and communication functions; s integration of power supplies into their functional levels (e.g. the control unit has its own power supply implanted in its circuit boards); s low consumption components; s modularity for easy maintainability, if possible without having to interrupt the process. Sensors and actuators Special attention should also be paid to the choice of sensors and actuators: s for sensors, it is very important to take into account their physical and electrical environments since these are key elements providing: s efficient control-monitoring, s corrective maintenance assistance, s a high level of preventive maintenance; s actuators that are directly related to guaranteed power availability must carry out their assignment, regardless of power supply failures or losses of control (problems on the SU or CU). In other words, they must: s retain their ON or OFF status (bistable operation), s allow operation in manual mode. The circuit breaker is an example.
s redundancy techniques (already referred to regarding short-term backup), s the possibility of stepped operation, s the appropriate choice of an earthing system.
failure tolerance
If the techniques and technologies chosen are not sufficient to achieve the desired level of availability, failure tolerance can be used. This tolerance is achieved essentially by:
Redundancy Redundancy should be provided for, as a priority, on the equipment that bears the most weight in the calculation of unavailability for the power station as a whole. Let us examine the choices that are possible/and or to be selected. s diesel generator set It is easy to assume that two generators in redundancy will ensure greater availability, but this is true only if the two generators use separate busbars; otherwise availability is decreased by the reliability of the extra coupling device. s short-term back-up This level, assigned to supply power to the application during the generator takeover phase, plays an essential role in power station availability. To fulfil the assignment, this level cannot be a common mode. A practical solution is to divide the risk by the modularity, s 3 kW (battery rectifier charger) to supply d.c. feeders such as telecommunication equipment, s 3, 40, 80 kVA (UPS) to supply a.c. feeders such as data processing equipment. This modularity allows: s stepped operation, and correction maintenance action without interrupting the power station assignment, s power redundancy according to the level of availability required and the repair times imposed by maintenance logistics. s low voltage switchboard power source changeover This is a common mode which, with its control parts, represents a failure rate in the vicinity of 10-5. The following two types of redundancy allow a greater level of availability to be achieved: s switchboard redundancy which makes at least 50% of the power distributed by both switchboards available during maintenance,
power supply changeover redundancy which is used when an anomaly is detected on the changeover device, taking battery autonomy into account. s automation systems Different types of PLC redundancy can be used. For this type of equipment, we will choose only the following redundancy: two totally asynchronous PLCs that are continuously active in the process, each of them synchronized with process status. The first PLC to enforce an action regarding availability will automatically impose this action on the other PLC. The actuators, by means of their control mechanism cabling, should favour ON status. The faulty PLC will withdraw without resetting its watchdog. s sensors Certain measurements, such as speed, temperature, gas oil level, etc. are fundamental to availability, not to mention equipment safety: the sensors used for these measurements are therefore provided with back-up. The coherency of measurements is assessed by the control-monitoring system in relation to process status and, in the event of an observed incoherence, the system rejects the measurement and declares the sensor to be faulty. s power supply for control-monitoring electronics and auxiliaries So as to enable stepped operation, there should be more than one power source for the various controlmonitoring functions in a dependability system. Each function should have its own power supply, and if some of them use the same power supply, it is necessary to provide a protection device for each function. Earthing systems The three standard earthing systems or diagrams are the TT (earthed), the TN (directly earthed neutral) and the IT (unearthed) systems. s TT earthed system Availability is provided by the choice of residual current circuit breakers with discrimination (amperage level and time-based) which make it possible to
s
isolate only the faulty feeder and to immediately eliminate the danger without altering installation operation on the whole. Fault current is limited by the neutral and feeder earth socket impedance and, as a result, faults will not damage the installation. This system is especially recommended for networks that are liable to be modified, altered by mobile or temporary receivers, or operated by non-specialized personnel. s TN directly earthly neutral system In this system, all insulation faults cause short circuits with current greater than the tripping limit of the short-circuit protection device. Availability depends upon the choice of the discrimination technique and the overcurrent protection devices (cf. chap. 4 LVSB technology choices). It should be noted that the TNS (separate neutral and protective conductor) system, when combined with the use of residual current devices is preferable to the TNC (combined neutral and protective conductor) system in terms of possible installation damage. Waiting for strong fault current to form is synonymous with major damage, particularly in receivers. This
has a definite effect on maintainability and hence on availability. s IT unearthed system Insulation faults do not entail any risks for people and do not require isolation by disconnection of the faulty portion; hence no breaking takes place. It is therefore advisable to track the fault and clear it before a second one occurs since if this happens (as in the TN system), one (or both) of the faulty feeder circuit breakers would open. The current of the first fault is very weak and does not cause any damage. This earthing system should be chosen for the best availability provided that... the first fault is tracked. With this earthing system, reference can be made to fault tolerance. Summary of the choices The choice of techniques related to failure tolerance according to the level of unavailability are summarized in the table on the next page (cf. fig. 9).
Human behaviour is considered as a failure if it reduces, even partially, the system reliability. The following question must be asked: What sort of work sharing is assigned to the Man Machine pair? The use of automatic control-monitoring is based on the following criteria: s reflex perception, decision and action, s complexity and implementation, s repetitive procedures. For example, switching from the main power source to generator power can be assigned to the system. Human intervention is found at two levels: s system control-monitoring (veto regarding functional matters), s taking into account of maintenance with system assistance for the user. Hence: s the division of tasks reduces the effect of human errors since people do no intervene in the normal operating process, s man is considered as an agent who contributes to reliability by checking and he is the last bastion of safety in the event of system malfunction. The electronics are broken down into three levels:
running the installation

The electronics play an active part in the level of dependability by assisting personnel with operation and maintenance tasks, in the aim of compensating for possible failures.
accessible unavailability levels for an installation as a whole according to its architecture: redundancy of CUs for the generator in the LVSB of power source changeovers of sensors (voltage and oil, water, pressure levels...) of UPSs level/total power modularity of the generator of specific distribution for electronic devices earthing system
10-6
s
5.10-6
s s s s
10-5
s s s s
10-4
s s
none
s s
none none none

s
1/6
s
1/6
s
1/8
s
none
s
none
s
none
s
none
s
IT
IT
TT or TNS or IT
TN or TT or IT
It should be note that the generator has no redundancy since the purpose it whould serve does not justify the very high cost it would entail. A reminder regarding erathing systems: IT = continuity of service, TT = reduced damage when insulation faults occur.
fig. 9: technical choices relatives to failure tolerance.
CU for control-monitoring SU for supervision s MU (management unit) for global management (cf. fig. 10). The equipment level has already been discussed at length as well as the control-monitoring (CU) level. The SU and MU levels, while less operational, are just as important. Supervision level (SU) This level provides the user in real time with an indication of process status in the form of:
s s
management MU supervision SU Control-monitoring CU installation constituents
fig. 10: hierarchical levels of electricity technical management.
alarms establishing the nature of the fault together with the type of clearance and repair, s logs providing access to the history of faults and process status changes, s system reports giving process status in real time. This level also enables the user to perform control-monitoring and hence to intervene in the system by means of the Man Machine Relation (MMR) via an operator terminal in the form of: s read-out of system reports, s modification of process operating parameters, s start of testing, s alarm clearance, s time changes, s etc. Management level (MU) When such a level exists, for several stations spread across a geographical area, it is remote from the local system and manages stations with the following functions: s remote supervision, s inventory, s statistics, s remote control with interlocking corresponding to the selected levels of availability. Should a problem occur, the user can be alerted locally by a radio call system. He then connects with the MU that generated the call by means of a telephone equipped for example with
s
a Minitel. Once he is aware of what is happening, he can make the initial arrangements before going to the local control-monitoring station, if necessary. These different levels take part in: s corrective maintenance, by enforcing inspection of all repairs on sub-assemblies that are critical to the power availability assignment. Only a positive test result will clear the alarm at the origin of the request for repairs, s preventive maintenance, by automatically or manually conducting periodic testing according to an electronically controlled schedule. Communication (cf. fig. 11) The reliability of communication (by bus) between the various levels is also very important: s it ensures the exchanges between - installation and CU (by bus if intelligent sensor-actuators are used), - CU and SU, - SU and MU. s it also enables the user to communicate with the system both locally and remotely. Operation, management and archiving data can be: s unidirectional for file transfers and periodic collection of maintenance information, s interactive, of the command/answer type for remote control and remote diagnostic operations.
local
remote
site 1
1234567890 1234567890 1234567890 azertyuiop^$ azertyuiop^$ azertyuiop^$ qsdfghjklm` qsdfghjklm` qsdfghjklm` <wxcvbn,;:= <wxcvbn,;:= <wxcvbn,;:= 1234567890 azertyuiop^$1234567890 azertyuiop^$ qsdfghjklm`qsdfghjklm` <wxcvbn,;:= <wxcvbn,;:=
Minitel
GEN SU
Page 3 Etat 1 Etat 2 896654 859305
1234567890 1234567890 1234567890 azertyuiop^$ azertyuiop^$ azertyuiop^$ qsdfghjklm` qsdfghjklm` qsdfghjklm` <wxcvbn,;:= <wxcvbn,;:= <wxcvbn,;:= 1234567890 1234567890 azertyuiop^$azertyuiop^$ qsdfghjklm`qsdfghjklm` <wxcvbn,;:= <wxcvbn,;:=
LVSB SU
local PC Telephone Switching Network
UPS SU
site 2
1234567890 1234567890 1234567890 azertyuiop^$ azertyuiop^$ azertyuiop^$ qsdfghjklm` qsdfghjklm` qsdfghjklm` <wxcvbn,;:= <wxcvbn,;:= <wxcvbn,;:= 1234567890 1234567890 azertyuiop^$azertyuiop^$ qsdfghjklm`qsdfghjklm` <wxcvbn,;:= <wxcvbn,;:=
Minitel MU
GEN SU
LVSB SU
local PC Telephone Switching Network

UPS SU
site n
fig. 11: architecture for communication with user.
5. example of increased availability backed-up distribution
specification
Unavailability rate: 10-5, i.e. 6 minutes per year (cf. fig. 12 and 13) Repair time: 8 h, for the repair of components liable to eventually comprise the assignment. As an image: the time required to repair the belt when both the belt and suspenders are being used at the same time.
average yearly failure rate: mains = 450 mn/y unavailability rate: -4 mains = 8.5 . 10
MT
transformer = 10 mn/y
BT
generator = 360 mn/y
GEN.
generator = 7 . 10
-4
construction
Based on the diagram in figure 4, the weak points of the installation (cf. chapter 2) should be improved and measures should be taken in terms of maintenance so as to divide the unavailability rate by 60. Action on installation components s diesel generator set s motor oversized by 30 % (full power can only be supplied when the motor is cold) or continuous preheating; s starting chain composed of: - an electric starter up to 600 kVA and a pneumatic one thereabove, - two chargers equipped with a battery, - two speed measurement chains, s gas oil circuit supplying the motor by the force of gravity; s lubrication circuit controlled by two temperature measurements; s two ventilation circuits; s closed circuit water cooling with a lost water cooling circuit as well, connected to the public water system; s two control-monitoring units. s power source changeover device The standby circuit breaker is backed up by a contactor which intervenes when ordered to do so by the control-monitoring unit (CU) in the event of a power source changeover failure. s short-term back-up The calculation shows that it is necessary to provide a minimum power redundancy of 10 %, implying modular equipment with total power exceeding rated power.
changeover = 90 mn/y changeover = 1.7 . 10 U = 2 . 10 UPS = 150 mn/y UPS
-4 -4
U= 6 . 10
-4
fig. 12: standard equipment produces an unavailability rate of 5h/y.
MV LV GEN. unavailability rate after improvements:
U = 0.5 . 10
-4
UPS
U = 10
-5
fig. 13: the inprovement of sensitive points (GEN - M/S and UPS) enables feeders to attain unavailability rates of 6 mn/y.
maintenance arrangements
electronics: a circuit board of each type for SU and CU. s power: a sub-assembly corresponding to each element that is critical to the performance of the assignment, throughout the chain, and which takes part in power supply to feeders with increased availability. Composition of the maintenance package: s preventive maintenance Action is requested by the system following either periodic testing or timedelayed alarms related to the end of operating intervals (e.g. generator discharge). In this case, the user should take action within 48 hours of the time the alarm is generated. s corrective maintenance This refers to repair action taken as a result of alarm generation. All measures should be taken to ensure quick repair. The 10-5 rate corresponds to the proper operation time before the first repair and proceeds from preventive maintenance. If, through negligence, the high availability power supply should enter the corrective maintenance system, the unavailability rate will drop. The mean time to repair will then be added on to the 6 minutes. The composition of the maintenance package and the efficiency of the maintenance department will therefore be determining factors.
s
S battery charger n 1
battery charger n 2
changeover generator starter
~~ ~~
UPS UPS
CU 1 -GEN-
CU 2 -GEN-
CU 1 -changeover-
CU 2 -changeover-
SU
SU
SU
MU
demonstrating specified availability

The detailed calculation is far too complicated to be presented here. By simplifying to a large extent, based on the data in figure n 12: s the probability of a voltage drop in the main LV circuit breaker is 450 mn/y, i.e. U = 10-3, s the probability of a voltage drop downstream from the power source changeover corresponds to the probability of the simultaneous occurrence of a mains failure and s the generator out of operation after 5 minutes time, or
Telephone Switching Network
fig. 14: control-monitoring corresponding to unavailability of 10-5 (6 mn/y).
the power source changeover out of service. This probability is very close to the changeover failure rate which is a common mode (compulsory stage), i.e. in the range of 100 mn/y, equivalent to U = 2.10-4. Backing up the changeover by a contactor will raise this rate to 0.5.10-4. The probability of a voltage drop at the feeder level reaches 10-5 with the UPS
s
and static contactor which prevent micro-cuts, backed up by an electromagnetic contactor. Referring to the table in figure n 8, this solution corresponds to an MTTF of 261,000 h, taking repair time into account. The MTBF for the installation as a whole is therefore in the range of 100,000 h, i.e. an average unavailability rate of 6 minutes per year.
6. conclusion
The spread of process technical management, building utilities and electrical power distribution entails continuous power supply for those systems, at the control-monitoring level and, to an increasing extent, at the power level. Mastering energy availability is nowadays a necessity for electricians. This Cahier Technique shows that this objective can be achieved, provided that: s a global approach is used, including the establishment of: s objectives (needs), s operating criteria, s conditions of use, (training, supervision, maintenance): s and action is taken regarding: s component reliability, s fault tolerance, component redundancy, and, naturally: s information processing, in other words: control-monitoring intelligence.
s
We have seen that to improve availability, efforts must be focused essentially on: s back-up sources close to the feeder, s common mode (compulsory path) equipment circuits, s preventive maintenance. It is currently possible to attain unavailability rates of 10-6 (less than a minute per year) thanks to UPS in particular, for power ratings that can reach several hundreds of kW. With power UPSs, the concept of clean, dependable mains has emerged.
7. bibliography
Publications s Un nouveau systme dalimentation haute disponibilit (A new high availability power supply system). C. Francon and R. Delooze Merlin Gerin. SEE Conference. s The decentralized DC unit in telecommunications equipment energy systems. J.P. Leblanc and D. Marquet, CNET, G. Gatine, Merlin Gerin INTELLEC 1987 Conference. s The operation of the GEODE energy system. J.P. Leblanc and D. Marquet, CNET. J.M. Rollet, Merlin Gerin. INTELLEC 1986 Conference. s A new material and data processing design for the availability target: the GEODE system. J.C. Chigolet, CNET, M.J. Grard Seri, Renault, C. Franco, Merlin Gerin. INTELLEC 1985 Conference. Merlin Gerin's Cahiers Techniques s Protection of electrical distribution networks by the logic selectivity system Cahier Technique n 2 (R. Calvas - F. Sautriau) s La slectivit des protections Cahier Technique n 13 (F. Sautriau) s Low voltage protection system selectivity: SELLIM system Cahier Technique n 126 (C. Albertin) s Industrial approach dependability Cahier Technique n 134 (H. Krotoff) s Introduction to dependability design Cahier Technique n 144 (P. Bonnefoi)
Ral. : Illustration Technique - Photo : IPV IPV - 10/91 - 2500 - Imprimeur : Lostic
n 149
EMC: electromagnetic compatibility
Frdric Vaillant Mr. Vaillant graduated from the Ecole Polytechnique in 1984 (X81) and was awarded a PhD in microelectronics in 1987 (thesis prepared under a CIFRE contract with the Saint Gobin company). His career at Merlin Gerin began in 1987 within the Research Department, where he was in charge of a project concerning static current interruption techniques for medium voltage applications. Since the end of 1988 he has been in charge of Electromagnetic Compatibility within the Electronics Proficiency Centre of the Research and Development Management Division.
CT 149, first issued octobre 1992
EMC: electromagnetic compatibility
summary
1. Introduction Electromagnetic Compatibility - p. 4 EMC - a characteristic and a discipline Today, EMC is indispensable p. 4 EMC theory is complex p. 5 2. The source The importance of identifying p. 5 the source An exemple of a continuous p. 6 source of conducted disturbances in power electronics An example of radiated p. 7 disturbance sources: circuit closing in MT and THT substations 3. Coupling Different coupling modes exist p. 8 Common or differential mode p. 8 field to wire coupling Common impedance coupling p. 10 Differential mode wire to wire p. 11 coupling or crosstalk 4. The susceptor Equipment malfunction p. 12 Solutions to the problem p. 12 5. Installation Installation is an important factor p. 14 in the overall system EMC Design phase p. 14 Installation phase p. 14 Pratical examples p. 14 6. Standards, test facilities and Standards p. 17 tests standards Test facilities p. 17 Tests p. 18 7. Conclusion p. 23 Appendix 1: glossary p. 24 Appendix 2: impedance of a conductor at high frequencies p. 25 Appendix 3: the different parts of a cable p. 26 Appendix 4: tests performed at the Merlin Gerin EMC laboratory p. 27 Appendix 5: bibliography p. 28
For all electrotechnical equipment, EMC must be considered right from the initial design phase and the various principles and rules carried on through to manufacture and installation. This means that all those involved, from the engineers and architects that design a building to the technicians that wire the electrical cabinets, including the specialists that design the various building networks and the crews that install them, must be concerned with EMC - a discipline aimed at achieving the "peaceful" coexistence of equipment sensitive to electromagnetic disturbances alongside equipment emitting such disturbances. This publications is a compilation of more than the years of acquired experience at Merlin Gerin and presents various disturbances encountered and provides some practical remedies.
1. introduction
electromagnetic Compatibility -EMCa characteristic and a discipline

EMC is a characteristic of equipment or systems that mutually withstand their respective electromagnetic emissions. According to the International Electrotechnical Vocabulary IEV161-01-07, EMC is the ability of an equipment or system to function satisfactorily in its electromagnetic environment without introducing intolerable electromagnetic disturbances to anything in that environment. This definition has also been adopted in the NF C 15-100 standard, chapter 33. EMC is now also a discipline aimed at improving the coexistence of equipment or systems which may emit electromagnetic disturbances and/or be sensitive to them.
class high energy
type voltage dips
origin
s power source switching s short circuits s starting of high power motors s systems with power semi-conductors s electric arc furnaces s direct or indirect lightning strikes s switching of control devices s breaking of short-circuit currents by
medium frequency high frequency
harmonics overvoltages
electrostatic discharges
protection devices discharge of static electricity stored in the human body
fig. 1: the most common electric disturbances.
wave interference and interference with control and monitoring systems caused by electromagnetic emissions. In recent years, several trends have together made EMC more important than ever: s disturbances are becoming stronger with increasing voltage and current values, s electronic circuits are becoming increasingly sensitive, s distances between sensitive circuits (often electronic) and disturbing circuits (power circuits) are becoming smaller. In the development of its new products, Merlin Gerin foresaw the necessity of understanding and applying EMC principles. In modern electrical switchgear and controlgear, low and high currents, control and power electronics, electronic protection and electric power devices all reside in close proximity. EMC is therefore a fundamental criterion that must be respected in all phases of product development and manufacture (see fig. 2), as well as during installation and wiring. Moreover, EMC is now included in standards and is becoming a legal requirement.
fig. 2: EMC application example: a mediumvoltage Fluair panel containing a circuit breaker designed to interrupt hundreds of ampere at tens of kilovolts and a SEPAM programmable control, monitoring and protection unit. The complete assembly must remain operational under all circumstances.
today, EMC is indispensable

Equipment or systems are always subjected to and, to some extent, generate electromagnetic disturbances. These disturbances are generated in many ways. However, the main underlying causes are sudden variations in current or voltage. The most common electrical disturbances (see fig.1) in the low voltage electrotechnical field are discussed in Cahiers Techniques Publication no. 141. Cahiers Techniques Publication no. 143 discusses disturbances generated when operating medium voltage switchgear. These disturbances can be propagated by conduction along wires or cables or by radiation in the form of electromagnetic waves. Disturbances cause undesirable phenomena. Two examples are radio
The experience and achievements of Merlin Gerin are not limited to the satisfactory operation of electrical and/ or electronic systems in their usual electromagnetic environment: Merlin Gerin designs and builds equipment capable of withstanding harsher conditions such as electromagnetic radiation generated by high altitude nuclear blasts. The necessary radiation hardening, i.e. improvement of the immunity of systems exposed to electromagnetic pulses from nuclear sources, requires the most advanced EMC techniques.
EMC theory is complex

Any work involving EMC involves the analysis of a three component system: s the disturbance generator or source, s propagation or coupling, s the device or system affected or the susceptor. Strictly speaking, the three entities are not independent but for all practical purposes are assumed to be. Note that installation, described in chapter 5, plays the most important role in the propagation of disturbances. Theoretical analysis is difficult because it must deal with the propagation of
electromagnetic waves described by a set of complex differential equations known as Maxwells equations. Generally speaking, they cannot be solved to yield an analytical solution for real devices and dimensions. Even with powerful computer systems, a close numerical solution is often extremely difficult to obtain. In practice, EMC problems must therefore be dealt with via simplifying assumptions, the use of models and in particular conducting experiments and taking measurements.
2. the source
the importance of identifying the source

The identification and measurement of the source is essential since the type of source will determine which of the following measures must be taken: s limit the disturbances generated (e.g. on a contactor, by installing an interference suppressing RC unit in parallel with the A.C. coil, or a diode on the D.C. coil), s avoid cross-coupling (i.e. physically separate two highly incompatible elements), s desensitize potential susceptors (e.g. using shielding). Main causes Any device or physical/electrical phenomenon that emits an electromagnetic disturbance, either conducted or radiated, qualifies as a source. The main causes of electromagnetic disturbances are electric power distribution, radio waves, electrostatic discharge and lightning.
s in
electric power distribution, a large number of disturbances are created by circuit switching operations. s in the low voltage field, the opening of inductive circuits such as contactor coils, motors, solenoid valves etc. generates very high surge voltages (up to several kV across the coil terminals) that contain high frequency harmonics (ten to hundreds of MHz). s in the medium and high voltage fields, the opening and closing of disconnectors produces waves with a very fast rate of rise (a few nanoseconds). These waves are particularly harmful to microprocessor-based systems. s radio waves emitted by remote monitoring systems, remote controls, radio communications, television sets, walkie-talkies etc. are, for some equipment, sources of disturbances in the range of several volts per meter. All of these disturbance emitters are nowadays increasingly common and susceptible equipment must therefore be provided with increasingly effective protection.
s an
electrically charged human body: for example, a person walking on certain types of carpet in a cold and dry climate can be charged up to more than 25 kV ! Any contact with equipment produces a discharge with a very fast rise time (several nanoseconds) which enters the device by conduction and radiation, generating a major disturbance.
Disturbance characteristics Sources may be intentional (e.g. radio transmitters) or not (e.g. arc welding units). However in general they can be distinguished by the characteristics of the disturbances they produce: s spectrum, s waveform, rise time or envelope of the spectrum, s amplitude, s energy. s the spectrum, i.e. the frequency band covered by the disturbance can be very narrow as for the case of mobile telephones, or very wide, as for electric arc furnaces.
Pulse type disturbances cover a particularly wide spectrum extending up to 100 MHz or more (see fig. 3). To this last category belong almost exclusively sources such as: s electrostatic discharges, s switching of relays, disconnectors, contactors, switches and circuit breakers in the LV, MV and HV range, s lightning, s nuclear electromagnetic pulses (a special domain). Since the degree of coupling is directly proportional to frequency, EMC uses the frequency domain to characterize disturbances. This type of representation, for a periodic signal, is similar to a Fourier series decomposition (like a sum of harmonics). s the waveform describes the characteristics of the disturbance with time and can, for example, be a damped sine wave or double exponential function. It is expressed as a rise time tr, an equivalent frequency 1/.tr or simply the disturbance frequency for a narrow band signal or as a wavelength l related to frequency by l = c/f, where c is the speed of light (3.108 ms-1).
s the
amplitude is the maximum value the signal reaches in terms of voltage (Volts), electric field (Volts/Meter), etc.
s the
energy is the integral of the instantaneous energy over the time the disturbance lasts (Joules).
most problematic of which is, based on experience, the generation of currents flowing through any parasitic capacitances. Taking only the parasitic capacitance Cp into account, the common mode current: Icm = Cp . dV/dT. With the rise times mentioned earlier, a parasitic capacitance of 100 pF is sufficient to generate currents of several hundred milliamperes. This disturbance current will flow through chassis ground (0 V reference
an example of a continuous source of conducted disturbances in power electronics

In power electronics, the principal sources of disturbances are more often voltage rather than current transients. The voltages can vary by hundreds of volts in a matter of a few nanoseconds giving dV/dts in excess of 109 V/s. Pulse Width Modulation (PWM) (see fig.4), for example, used to generate a sine wave voltage from a D.C. voltage, works with voltage changes from 0 to Udc (660 V for rectified three-phase) occurring in a very short time, nano to microseconds depending on the technology used. Rapid voltage changes are the source of various disturbance phenomena, the
(a) U Ucc
Uca t
Radio wave amplitude of disturbance 0 time spectral density narrow band
(b) tr Ucc
tf
0 1/T frequency
e) ve cur wav AC sine of a art
Indirect lightning effect amplitude of disturbance spectral density wide band
(p
tr
0 time
1/ t r
frequency
fig. 3: spectral characteristics of disturbances.
fig. 4: a source of disturbances in power electronics: pulse width modulation. a: principle, b: even with the time scale not well chosen for this type of phenomenon tr 2 to 3 tf (10 ns to 1s) while the sine wave covers 20 ms.
of the unit) of the electronics and can, via coupling, modify signals (information or controls), be superimposed on sensitive measurements and disturb other equipment by injecting the disturbance back into the public distribution network. One way of dealing with this type of phenomenon, i.e. of ensuring EMC, is to increase the voltage rise time. However such a solution would considerably increase the switching losses in the transistors, producing harmful thermal stresses. Another effective way of reducing common mode currents consists of increasing the common mode impedance. For example, when mounting electronic power devices, two methods are commonly used:
s either
leave the heat sinks floating (no electric connection), (see fig. 5), if safety regulations are not violated, s or reduce the parasitic capacitance between the device and the heat sink using an insulator with a low dielectric constant (see fig. 6). In the field of UPSs - Uninterruptible Power Supplies- for instance, the above precautionary measures make the difference between a polluting system and a clean system. For UPSs, note that the low level electronics in the static inverter must be protected against disturbances created by its own power circuits. It is necessary to understand and control the phenomenon at the source to effectively and economically limit conducted emissions. Other less frequent sources of conducted disturbances exist such as lightning and switching surges that can generate large dV/dts and dI/dts. These disturbances also radiate.
insulator heat sink semi-conductor
PP PP @@ @@ ,, ,,
Cp I CM ground
, , , ,
an example of radiated disturbance sources: circuit closing in MT and THT substations

The substation environment, especially in medium and very high voltage applications, can contain very strong pulsed electromagnetic fields. Certain switchgear operations can generate voltages much higher than the rated value in a very short time. For example, when a 24 kV switch is closed, the preignition phenomenon causes voltage variations of tens of kilovolts in a few nanoseconds (10-9 s).
fig. 5: the parasitic capacitance of the heat sink (for cooling of electronic devices) is taken into account in the design of UPS inverter stacks.
This is discussed in greater detail in Cahiers Techniques Publication no. 153: SF6 Fluarc circuit breakers and MV motor protection. Measurements performed at the Merlin Gerin laboratories have shown that during the switching of a 24 kV medium voltage circuit breaker, damped sinusoidal pulsed fields reach peak values of 7.7 kV/m with a frequency of 80 MHz at a distance of one meter from the cubicle. The field strength is enormous when compared to that of a 1 W portable two way radio (walkie-talkie) which generates 3 to 5 V/m measured at a distance of one meter. The transients propagate along conductors, busbars, cables and overhead lines. At the frequencies involved, i.e. the rapidity of the phenomenon, the conductors (especially busbars) behave like antennas and the characteristics of the electromagnetic fields they emit are highly dependent on the design of the metal enclosures (partitioning, cubicles). In metal clad very high voltage substations, the electromagnetic fields are particularly strong. Metal clad SF6-insulated substations have a coaxial shape and therefore display a constant characteristic impedance. Rapid voltage changes inside the tubular metal enclosures generate standing wave phenomena. They are created by reflections occurring at impedance mismatches due to conic outgoing feedthroughs that cross the shielding for example. The magnitude and duration of the phenomenon is also increased by this effect.
insulating washer for TO3 case Mica Plastic Alumina
thickness (mm) 0.1 0.2 2
parasitic capacitance (pF) 160 95 22
fig. 6: dielectric constants for the most common insulators used in mounting electronic devices.
The electronic environment at medium and very high voltages requires in depth electromagnetic compatibility studies for the design and installation of relay systems and control and monitoring devices. This is particularly important because in addition to the radiated disturbances, conducted voltage transients are also generated in substations as discussed at the beginning of this section (see fig. 7).
fig. 7: SEPAM and Masterpact units; MV and HV protection and control and monitoring devices with digital electronics developed by Merlin Gerin and designed taking full advantage of EMC research.
3. coupling
different coupling modes exist

Coupling refers to the linking, transfer or transmission of electromagnetic disturbances from an emitter to a susceptor. Coupling is expressed in terms of a coupling coefficient k, expressed in dB (e.g. -75 dB), which can be seen as the transmission efficiency of the disturbance from the emitter to the potential susceptor (k = 20 log A (received)/A (transmitted), where A is the amplitude of the disturbance).
It is important to define this coefficient for EMC since the lower the coefficient (the larger its absolute value in decibels) the weaker the disturbance voltage received by the susceptor and the better the EMC. This coefficient k is only meaningful when the transfer of electromagnetic disturbances is proportional to frequency, which is often the case in practice. Three well known coupling modes can be distinguished: s common and differential mode field to wire coupling, s common impedance coupling,
s differential
mode wire to wire coupling
or crosstalk.
common or differential mode field to wire coupling

An electromagnetic field can couple into any kind of wire-like structure and generate either common mode (with respect to ground) or differential mode (between wires) voltages or, as is generally the case, both. This type of coupling is called field to wire coupling and is also known as the antenna effect of wiring, printed circuit board traces, etc.
s common
mode coupling generates common mode disturbance voltages or currents.
A conducted common mode voltage disturbance (VCM) is a voltage that affects all active conductors. It is referenced to chassis or earth ground (typically in electrical systems): all common mode isolation tests on low voltage circuit breakers are therefore performed between earth ground and all phases.
A common mode current (ICM) is a current that flows through all active conductors in the same direction (see fig. 8). The current induced in a LV line by a lightning impulse is a common mode current.
s differential
mode coupling involves voltages and currents in the classic sense, for example, between two phases of a circuit breaker or between two wires which transmit sensor data to the electronics.
,,,, ,,,, ,,,, ,,,, ,,,, ,,,, ,,,, ,,,,

Cp disturbance generator
I CM
VCM
PP @@ ,,
PE
I CM
fig. 8: common mode voltage and current between two relays of a low voltage compartment in a medium voltage cubicle.
PP @@ ,,
,,, ,,, ,,, ,,, ,,, ,,, ,,, ,,,

Cp
The equations that govern the coupling between the electromagnetic field (impedance of an arbitrary wave) and a wire-like structure (which can also be arbitrary) are very complex. In most cases they can neither be solved analytically nor numerically. Nonetheless, one of the simpler and most common types of coupling can be expressed analytically: the coupling between the magnetic component of an electromagnetic field and a loop of area A formed by the conductors (see fig. 9). The magnetic component H of the field induces in the loop a series voltage equal to: e = 0 'A' dH/dt, with = the permeability in a vacuum (4 10-7 H/m). For example, in a medium voltage substation, a loop (of wire or cable) covering 100 cm2 placed 1 m from the
E electromagnetic field
surface exposed to electromagnetic field
e = voltage induced by the electromagnetic field
fig. 9: an example of differential mode field to wire coupling.
cubicle (see fig. 10) and exposed to a pulsed field of 5.5 kVrms/m (laboratory measurement) will generate (by induction) a series transient voltage of 15 V. The above equation holds as long as the largest dimension of the loop does
,,, ,,, ,,, ,,, ,,, ,,, PP @@ ,,

0 volt
not exceed a tenth of the wavelength of the disturbance. Note that such a green/yellow wire loop (see fig. 10) is easily created in the relay compartment when the wires are connected in a star configuration to ground.
common impedance coupling

As the name implies, common impedance coupling results from an impedance that is shared by two or more circuits. The common impedance can be the ground connection, the earth ground network, the power distribution network, the return conductor shared by several low power signals etc... An example follows showing the effects of this type of coupling (see fig. 11): A disturbance current in circuit A in the tens of mA range is sufficient to generate disturbance voltages in the volt range in circuit B. If circuit B uses point M as its reference (possibly ground), then the reference can vary over several volts. This certainly influences integrated circuit electronics that work with volta-ges of the same order of magnitude. The example in figure 11 shows that a common impedance can be formed by a wire a few meters in length and which is common to both circuits A and B. The disturbance has a magnitude Uc = Ia . Zc where s Ia is the disturbance current and s Zc is the common impedance (see fig. 12). At low frequencies the common impedance is usually extremely small. For example, safety requirements dictate minimum cross-sectional areas for the PE conductors, i.e. the green/ yellow wires, of grounding networks depending on the prospective shortcircuit current. The impedance at 50 Hz between two points in the network is therefore always much lower than one Ohm.
i1 i2
cubicle ground
,,, ,,, ,,, ,,, ,,, ,,, PP @@ ,,

0 volt
fig. 10: example of a ground loop in a low voltage compartment of a medium voltage cubicle.
supply circuit A +
common Z 0 volt input I supply + I measurement

Z1
Zc Uc
Z2
fig. 11: the quantities measured by the operational amplifier will be incorrect because the disturbance current in circuit A (power supply) is high enough to create a disturbance voltage in circuit B (measurement).
PP @@ ,,
measurement circuit B
E1 supply circuit
la = i1 + i2 E2 measurement circuit
ground of measurement device
fig. 12: common impedance diagram.
But that same impedance can be much larger at the typical frequencies of the disturbances discussed earlier. Impedances can reach several kiloohms or more (see appendix 2).
the distance between wires or pairs of wires and the higher the frequency of the disturbances. For example, using the notation in figure 13, the voltage coupling coefficient (capacitive crosstalk) can be expressed as:
C12
differential mode wire to wire coupling or crosstalk

Crosstalk is a mode of coupling that resembles the field to cable coupling. It is called capacitive or inductive crosstalk, if a change in current or voltage respectively is its cause. A rapid voltage change between a wire and a ground plane or between two wires (see fig. 13) generates a field that can nearby, with some approximations, be considered an electric field only. This field can couple into any other parallel wire-like structure. This is called capacitive crosstalk. Similarly, a current change in a wire or cable generates an electromagnetic field that with the same approximations can be considered a magnetic field only. The field can couple into a pair of wires and induce a disturbance voltage. This is called inductive crosstalk (see fig. 14). Capacitive and inductive crosstalk exists whenever conductors are routed in parallel or reside in close proximity to each other. Crosstalk can occur in cableways and troughs and especially between power cables carrying high frequency disturbances differentially and twisted pairs used by digital networks such as Batibus. The crosstalk will be stronger the longer the parallel paths, the smaller
C 12 j 2 f (C 12 + C 20) VN = V1 C 12 j 2 f R (C 12 + C 20)
where: s V1: voltage source, s VN: disturbance voltage induced by coupling, s C12: coupling capacitance between two wires which is proportional to the wire length and the distance coefficient Log [1 + (h/e)2] where h is the distance between the two wires of the pair and e the distance between pairs, s C20: leakage capacitance between the two wires of the pair creating the disturbance, s R: load impedance of the susceptor pair. To be more specific, consider two pairs with wires of 0.65 mm diameter running 10 meters in parallel; the wires in the pair are 1 cm apart and the pairs 2 cm away from each other and R = 1 k . For a 1 MHz signal, a coupling coefficient of - 22 dB is found, therefore
R h V1
VN
C20
fig. 13: a rapid change in V1 creates a field which at a short distance can be assumed to be purely electric and induces a voltage VN in another wire-like structure which runs in parallel; this mode of coupling is called capacitive crosstalk.
power cable
I H
low power pair of wires
VN = 1 V1 12
In practice, capacitive and inductive coupling of this type is considerably reduced by the use of twisted pairs and shielded cables.
fig. 14: a current change in the cable generates an electromagnetic field which at a short distance can be considered to be purely magnetic and induces a disturbance (voltage) in wires that form a loop; this mode of coupling is called inductive crosstalk.
4. the susceptor
The susceptor is the third participant in the source/coupling/susceptor system and refers to any equipment that may be affected by a disturbance. It is typically equipment containing some electronics which malfunction because of electromagnetic disturbances occurring in an unexpected frequency band.
of a display; others may not be acceptable such as security equipment malfunctions.
solutions to the problem

Numerous solutions in terms of how equipment is to be built exist to provide effective and low-cost immunity to electromagnetic disturbances. Precautionary measures can be taken in: s the design of printed circuit boards (functional partitioning, trace layouts, interconnects), s the choice of electronic devices, s the ground interconnections, s the wiring. The choices involve many different disciplines and should be made during the design phase of a project to avoid additional costs which are always high for modifications after the design is completed or when the product is already on the market. Implementing all of these precautionary measures requires know-how which
goes far beyond the standard filtering and shielding techniques often recommended to increase immunity even if their effectiveness has not been proven. Printed circuit boards The designer of printed circuit boards must follow certain rules that concern functional partitions and layout. Starting with component placement, it is already possible to reduce coupling effects related to proximity. For example, the grouping together of elements that belong to the same circuit category (digital vs - analogue vs - power circuits), according to their susceptibility, reduces interferences. Furthermore, the layout of circuit board traces (routing) has a dramatic effect on susceptibility: the same electrical schematic implemented in different ways can display orders of magnitude different immunity levels. For example, a minimum etch circuit board layout (see fig. 15) reduces radiation effects and sensitivity.
equipment malfunction
Equipment malfunctions are divided into four categories and can be: s permanent and measurable, s random and non-repetitive, appearing when the disturbances appear, s random and non-repetitive, remaining after the disturbances vanish, s permanent equipment failure (components physically destroyed). The above types characterize the duration of the fault but not its severity. The severity of a fault is a matter of functionality or, in other words how critical the equipment is. Certain malfunctions may be acceptable for a limited time such as the temporary loss
0 volt
thin circuit layout
minimum etch layout
layout with ground plane
fig. 15: the circuit layout can reduce the electromagnetic susceptibility of a PCB: either by minimizing impedances (minimum etch), or by reducing the coupling of the electromagnetic field (ground place).
Electronic devices Numerous devices are available to provide effective protection against conducted disturbances. Selection is guided by the power level of the circuit to protect (power supply, control and monitoring, etc.) and the type of disturbance. Consequently, for common mode disturbances in a power circuit, a transformer will be used if the disturbances are at low (< 1 kHz) frequencies and a filter if they are at high frequencies. The table in figure 16 gives a nonexhaustive list of protection devices. All are not equivalent: a filter does not protect against surges, and a surge protector does not protect against high frequency disturbances. Shielding Enclosing sensitive equipment in a conductive shield provides protection against electric fields. To be effective, the thickness of the conductive shield must exceed the skin depth at the frequencies of the disturbance encountered (see fig. 17). The choice of material is of little importance. In some cases a conductive lacquer can be used as a shield. The metal or metal-coated insulator shield constitutes the ground. Ground interconnections When it comes to grounding, good electric conductivity between different parts of the housing is extremely important. They must be carefully and correctly interconnected, for example protecting contact areas from any paint and also by using short, wide wire braids (to reduce impedance to a minimum).
type surge arrester
device example spark gap lightning arrester limiter varistor Zener diode transformer inductors capacitors filters wire grid door braid shielded cables high frequency gaskets current finger
applications power supply, control and monitoring s in installations
s electronic devices
filtering
power supply, control and monitoring (installations and electronic devices)
shielding
data transmission (cabinet in disturbed area)
fig. 16: list of protection devices.
incident wave absorption
conductivity ( /cm2 )
transmission
reflection
shield depth
skin depth
fig. 17: screening effect of a metallic shield.
Wiring The shielding of wires, sometimes called screening, can be seen as an extension of the conductive envelope placed around sensitive systems. It therefore has the shortest possible connection and if possible all around its perimeter to protect against high frequency disturbances. Just as with the coupling between an electromagnetic field and a wire-like structure (see section 3), the theory governing wire shielding is very complex and too vast to be covered in
this paper. References to special literature are given in the bibliography. When all design and manufacturing rules are respected, the system will be sufficiently immune to electromagnetic disturbances in the environment it was built for. Nevertheless, this immunity can only be validated by actual measurements that determine the effectiveness of different shielding techniques. At Merlin Gerin, for example, different prototype models of electronic trip units for circuit breakers are exposed to rigorous tests
representative of the largest disturbances to which they may be expected to be subjected to. The true objective of these tests is to check that the trip unit does not operate inadvertently and that the circuit breaker opens correctly and in the required time. The product standards now include these specifications: a document is currently being discussed at the IEC , representing an EMC appendix to the IEC 947-2 standard concerning industrial circuit breakers.
5. installation
installation is an important factor in the overall system EMC

Evidence of this fact can be found in the NF C 15-100 general LV installation standards which devotes an entire chapter (33) to electromagnetic compatibility. The two previous chapters have shown that installation plays an important role in EMC; this is true for both the design and layout phase and the actual installation phase.
disturbances and a low sensitivity susceptor, s or form a compromise between the above two extremes. The second factor that depends directly on the first concerns the positioning of equipment, already selected with respect to their individual characteristics, to satisfy EMC requirements. It is obvious that this selection must take into account the cost of equipment and of its installation.
installation phase
Electrical and electronic installation work should follow the guidelines already discussed in the previous chapters. In practice, the different coexistent coupling modes must be studied and reduced to satisfy the EMC requirements. Different techniques should be applied: s the circuits and the chassis/earth grounds must be laid out in a grid, s the circuits must be physically separated, s the wiring must be carefully planned.
interconnected electronics sensitive to high frequencies. Common impedance coupling frequently occurs and to avoid it, the best possible equipotential grounding system or to be more precise a ground grid, is essential. This is the first step in providing protection against disturbance problems. In a factory power distribution network, all protection (PE) wires must be joined together and connected to the existing metal structures as specified in NF C 15-100 (see fig. 19). Similarly, within equipment, all grounds and frames must be connected to a grid-like grounding system in the shortest possible way using low impedance (at high frequencies), wide and short electrical connections (wires or braids). The wiring of an electrical cabinet is a typical example: all grounds must be connected together. There is a change to be noted here: the method involving the connection of all grounds to a central point (star configuration), sometimes used for electronic equipment sensitive to 50/ 60 Hz hum, has been replaced by grids which are far more effective in reducing disturbances that affect todays digital systems, protection relays and control and monitoring systems.
design phase
During the design and layout phase two major factors govern EMC: the choice of equipment and their relative locations (see fig. 18). The first factor concerns the choice of both emitters and susceptors: a given piece of equipment can to some extent generate disturbances and/or be susceptible. For example, if two units are to operate close to each other they must: s either combine an emitter that generates low levels of disturbances and an ordinary (i.e. not overly sensitive) susceptor, s or combine an ordinary emitter that generates moderate levels of
practical examples:
Grid layout for circuits and chassis/ earth grounds Today, equipment can be susceptible to very low energy levels. It contains
low voltage for "machines" switchboard with isolation transformer
low voltage for "workshops" distribution switchboard
Laboratory
electric welding sets MV/LV substation main low voltage switchboard
Production
Sales department Computer department low voltage for "offices" distribution switchboard and UPS
fig. 18: example of electrical equipment layout respecting EMC.
P P @ @ , , P @ ,
PE PE
fig. 19: the grids forcircuits and for chassis/earth grounding systems are often combined in electrical cabinets.
Separation of electrical circuits This technique consists of separating the energy sources (usually 50 or 60 Hz). The aim is to avoid interference on a sensitive device caused by conducted disturbances generated by other systems connected to the same power source. The principle is to create two separate power sources isolated by impedances that are high at the frequency of the disturbances. Transformers (not auto-transformers) are effective isolators, especially at low frequencies: MV/LV transformers, isolation transformers and any input transformer for electronics stop conducted disturbances. Sometimes an isolating filter is required to eliminate high frequency disturbances. If the sensitive equipment also requires emergency power, it can be supplied by an uninterruptible power supply (UPS) as long as the UPS contains the required isolation transformer(s). Rational wiring The effects of the three coupling mechanisms discussed earlier can be reduced if the wire and cable routing adheres to the following rules: s in all systems that cannot be separated physically for economic reasons, wires/cables must be grouped together by category. The different categories should be routed separately: in particular, power cables should be on one side and low power cables (telephone, control and monitoring) on the other. If a sufficient number of cableways or troughs are available, power cables carrying more than a few amperes at 220 V should be routed separately from
the low power signal cables. Otherwise, a minimum distance of at least 20 centimeters must be kept between the two. Any element common to these two categories of cables must be avoided. Circuitry using low level signals should have, whenever, possible its own return wire (0 Volts) to avoid common impedance coupling. The majority of systems that communicate over buses require pairs of wires reserved exclusively for data exchange.
s in
s the
cable routing troughs should be, if at all possible, made out of metal. The troughs should be correctly electrically interconnected, e.g. screwed together and connected to the grounding grid (see fig. 20). most sensitive cables (e.g. those used in measurements) should be placed in the corner of the trough where they can benefit from maximum protection against electromagnetic radiation. Their shielding, if any, should be connected to the trough at regular intervals. The use of prefabricated cable trunking assemblies in which the cables are positioned and connected correctly, such as Telemecaniques Canalis system with built-in control wires, are highly recommended. All these cabling techniques, which effectively avoid EMC problems, only increase costs slightly when applied at design or installation time. Later modifications of an existing installation showing excessive electromagnetic coupling are far more expensive.
s the
any case, the overall loop area formed by the conductor and its return must be minimized. In data transmission, twisted pairs reduce the susceptibility to differential mode coupling. The twisted pair is to be preferred over straight wires. s cables used for measurements and low signal level data transmission should be shielded, if possible, and unless specially instructed by the manufacturer, their shield connected to ground at a maximum number of points.
power cables
control wires
shielded cable conducting measurement data, possibility connected to the cable trough at regular intervals
d d = a few centimeters
fig. 20: cable routing example.
6. standards, test facilities and tests standards
standards
Documented standards that regulate electromagnetic compatibility of systems have long been in existence. The first regulations were issued by the International Special Committee on Radio Interference (CISPR). These regulations covered only the maximum acceptable power level that could be emitted by different types of equipment, mainly to protect radio transmission and reception. National Committees and the International Electrotechnical Commission (IEC) have issued documented standards that cover all aspects of EMC emission and susceptibility encountered in the civilian domain. Military standards on EMC have been compiled in the GAM EG 13 series in France and in the MIL-STD series in the United States. The increasing importance of EMC and the forthcoming unification of Europe are changing the landscape of civilian standards. The European Council published a Directive (reference 89/336/EC) in May 1989 on this subject. It relates to unifying the EMC legislation of the member countries. Every member country is committed to include it in its national legislation and make its use and application mandatory. The European Directive not only imposes limits on emitted disturbances but also sets the minimum immunity to electromagnetic disturbances. The Directive makes reference to standards not yet ratified; standards that define maximum acceptable disturbance levels, minimal immunity levels and measurement methods. A Technical Committee, TC 110, has been created for this purpose by the European Committee for Electrotechnical Standardization (CENELEC). Its duty is to bring together the existing standards
that are in accordance with the Directive and to write or rewrite those that are not. Without anticipating the work of TC 110, it seems likely that it will be based on existing standards already in use in the industrial community (see fig. 21). For emission tests, the German standards VDE 0871 and VDE 0875 were used for some time as a reference. The recent European standards EN 55011 and EN 55022 are now replacing them. For immunity tests, the IEC 801 publication is currently used as a reference. It will be included in the IEC 1000 publication which gathers all material on EMC written by the IEC in the framework of Committee 77. Publication 801 contains several parts for different types of disturbances that may affect a system or equipment. The parts are respectively: s 801-1: general introduction, s 801-2: electrostatic discharge requirements, s 801-3: radiated electromagnetic field requirements, s 801-4: electrical fast transient/burst requirements, s 801-5: surge immunity requirements (proposed), s 801-6: current injection (proposed). Parts 801-2, 801-3 and 801-4 relate to typical disturbances encountered in the modern electrotechnical world. They are widely accepted in the international community and Merlin Gerin has decided to adopt them for its products.
The following section describes in more detail the tests that relate to these standards.
test facilities
As mentioned before, to respect regulations, standardized measurements and tests must also be performed. Due to its field of applications, Merlin Gerin made EMC one of its major concerns long ago. Large installations such as Faraday rooms have been in use since the seventies. In 1988 a new dimension was reached with the opening of the EMC laboratory at the DTE Research and Development Centre in Grenoble. This centre makes full use of skills and knowledge and promotes the exchange of information. It also offers measurement services and is involved in special projects, training, and standards work as a recognized expert. As a centre offering services to outside customers, it performs measurements in all EMC fields: electrostatic discharge, conducted and radiated emissions, susceptibility to conduction or radiation. As with any other measurements, electromagnetic compatibility measurements must be reproducible both in time and in space, which means that two measurements performed at two different laboratories must yield the same results. In the EMC discipline, this means large facilities requiring considerable investments and a strict quality policy.
application field s susceptibility example s emission example
french standards NF C 46-02x NF C 46-022 NF C 91-0xx NF C 91-022
original international standards IEC 801 - (x + 1) IEC 801 - 3 EN 55 0xx EN 55 022
fig. 21: table of main standards in use in France and their international counterparts.
The quality program at the Merlin Gerin EMC laboratory is based on a Quality Manual and a set of procedures. These procedures concern calibration and the connection to calibrated standards in addition to each type of measurement itself. The list of tests for standards that can be performed at the laboratory are listed in appendix 4. The fact that the laboratory has been accredited by the Rseau National dEssais (National Testing Network) acknowledges the quality assurance policy.
tests
Electrostatic discharge These tests are designed to check the immunity of circuit boards, equipment and systems to electrostatic discharge. Electrostatic discharges are the result of charge accumulated by a person, for example, walking on a floor covered with an electrically insulating material. When the person touches an
electrically conducting material connected via an impedance to ground, he discharges suddenly through the impedance. Several studies have shown that the waveform is a function of the characteristics of the emitter (the source of the discharge) and of the circuits involved, but also of other parameters such as relative humidity (see fig. 22) or the speed at which the charged body approaches, in our example the hand of the person etc. This research has led to standardized discharge tests. They are performed with an electrostatic gun that simulates a human being in predetermined configurations (see fig. 23). Discharges are performed on all accessible parts of the device under test, in its immediate environment and repeated a sufficient number of times to make sure that the device resists electrostatic discharge. These measurements require an appropriate test bench.
All tests are completely defined by standard IEC 801-2 (revised in 1991) with severity levels shown in the table of figure 24. Conducted electromagnetic susceptibility Susceptibility tests are used to verify the resistance of equipment to disturbances reaching it via external equipment cables (inputs, outputs and power supply). As mentioned before, these disturbances differ depending on the type and installation characteristics of the cable. The electromagnetic signals or pulses used in these tests have characteristic amplitudes, waveforms, frequencies etc. Disturbance measurements performed on numerous sites have led to the selection of two tests. The first test, covered by IEC 801-4, simulates typical disturbances generated by the operation of controlgear. The test uses bursts consisting of a number of fast transients. The burst repetition
voltage (kV) 16 15 14 13 12 11 10 9 8 7 6 5 4 3 2 1 0
synthetics
insulated table wool
anti-static 5 10 20 30 40 50 60 70 80 90 100 relative humidity (%)
,, ,,, ,,, ,,,,,,,,,,,,,,, ,,,,,,,,,,,,,,,, @ P , @ ,, ,,,,,,,,,,,,,,,, , ,P , ,,,, ,,,,,,,, , , ,,,, ,,, ,,,, , ,,, , ,, , , , ,,,,,,,,,,,,,,,,, , ,, ,, , ,
equipment under test (EUT)
conductive surface
conductive surface
470 k resistors
ground reference plane
fig 22: the effect of relative humidity on the electrostatic discharge voltage for three types of floor materials.
, ,, ,, ,,, ,
fig. 23: electrostatic test site as defined by standard IEC 801-2.
P @ , P @ , P @ , P @ , P @ , P @ ,
power supply
insulator
mains
frequency is approx. 3 Hz. Each burst contains approx. 100 transients every 100 s . Each transient rises steeply (5 ns) to an amplitude of several kV, depending on the required severity level (see fig. 25 and 26). All cables can be subjected to fast transients. This type of disturbance couples into wiring very easily e.g. crosstalk (see the chapter on coupling). It takes only one cable generating such disturbances in a cable or wire trough to pollute all other cables running along the same path. The test must therefore involve all cables and
wires: a common mode test is performed on all wires with artificially induced disturbances (cables other than the power supply) and a common and differential mode test on cables connected to the mains. Disturbances are injected into the tested cables either via direct capacitive coupling (power supplies), or via a coupling clamp consisting of two metal plates that enclose the secondary cables (see fig. 27). The equipment under test must not show a malfunction over a predetermined period (1 min). This test is the
most relevant one for device immunity because fast transients are the most frequent ones encountered. The second test is representative of secondary effects created by phenomena such as lightning. It simulates conducted disturbances appearing on LV power lines after lightning strikes (801-5 draft). These disturbances consist of energy that is transformed into: s voltage impulses 1.2/50 s, if the impedance of the tested device is high, with amplitudes that can reach several kV,
severity level according to IEC 801-2 1 2 3 4
tests voltage 10 % (kV) 2 4 8 15
fig. 24: electrostatic discharge voltages that devices must withstand to comply with standard IEC 801-2.
severity applied test voltage ( 10 %) in kV level without malfunctions occurring (open circuit output) according to on power supply on input/output lines IEC 801-4 (signal, data, control) 1 0.5 0.25 2 1 0.5 3 2 1 4 4 2 x special special level x is defined contractually between manufacturer and client.
fig. 26: table of severity levels defined in IEC 801-4.
(a) u
15 ms 300 ms (b) u
5 ns 100 s
fig. 25: shape of the bursts (a) and their fast transients (b).
fig. 27: susceptibility to fast transients, measured on an Isis master control unit (test 801-4) in a Faraday room. This photo shows the disturbance generator being adjusted by an operator, the wooden case containing the coupling clamp and the Isis master control unit connected to the Batibus network.
s current
impulses 8/20 s if the impedance is low, with amplitudes reaching several kA.
The rise time of this type of disturbance is in the order of a thousand times longer, in the microsecond range, than for bursts of fast transients (see fig. 28). Crosstalk type of coupling is therefore less prevalent and this second type of test only applies to cables directly connected to the mains. The common and differential mode tests use capacitive coupling and appropriate levels. The procedure resembles the fast transients test: the equipment under test must not malfunction. Susceptibility to radiated emission The susceptibility tests for radiated emissions were devised to ensure the satisfactory operation of equipment when exposed to electromagnetic fields. Since these tests are particularly environment sensitive, the means deployed and competency levels required to produce reliable and reproducible susceptibility measurements are very high. The surrounding environment must be sufficiently clean and free of waves normally present, since (as discussed in the source chapter) electromagnetic fields with strengths in the several V/m range are frequent (e.g. two-way portable radios) and pulsed electromagnetic fields with even higher levels are common in
industrial environments. These tests must therefore be conducted in Faraday rooms with walls covered by high frequency absorbing materials. The rooms are called anechoic chambers when all walls including the floor are covered and semi-anechoic when the floor is not. In the chambers, the fields are generated by different types of antennae depending on the type of field, the frequency range and polarization. The antennae are driven by a wideband power amplifier controlled by a R.F. generator (see fig. 29). The generated fields are calibrated using broadband isotropic sensors
(field strength monitors). The diagram in figure 30 shows a typical test setup. Standards define the acceptable disturbance levels. In particular, standard 801-3 (currently being revised) recommends tests using frequencies in the range of 27 to 500 MHz at three severity levels. (1.3 and 10 V/m). Note that the test conditions that can be created at the Merlin Gerin laboratories are much more severe: the frequency range that can be covered extends from 10 kHz to 1 GHz. From 27 MHz to 1 GHz devices can be tested against fields reaching 30 V/m and 80 % modulation. Standardized measurements for pulsed electromagnetic fields do not yet exist. In this domain, Merlin Gerin uses its own internal procedures.
severity test open-circuit levels output voltagerr according to (kV) draft IEC 801-5 1 0.5 2 1 3 2 4 4 x special level x is defined contractually between manufacturer and client.
fig. 28: severity levels as defined in project IEC 801-5 (generator impedance = 2 ).
fig. 29: Faraday room: semi-anechoic chamber and a several antennae of the Merlin Gerin EMC laboratory.
Conducted emission Conducted emission measurements quantify the disturbances that the equipment under test reinjects into all cables connected to it. The disturbance strongly depends on the high frequency characteristics of the load connected to it since the equipment under test is the generator in this case (see fig. 31). To obtain reproducible measurement results and especially to avoid problems with the characteristic impedance of the network, the conducted emission measurements are performed with the help of a Line Impedance Stabilizing Network (LISN). A high frequency receiver is connected to the network to measure emission levels at each frequency.
Faraday room semi-anechoic
network antenna equipment under test (victim) broadband amplifier
1 kW
filter
10 kHz to 1 GHz
RF generator
fig. 30: typical test setup in a Faraday room. Measurements are performed in two stages: 1 - calibration of the filed for a given frequency range, without the EUT, 2 - verification of the EUT immunity.
Faraday room semi-anechoic line impedance stabilizing network equipment under test (source) network filter
measurement device
fig. 31: measurement configuration for conducted emissions. The EUT (equipment under test) is the generator, the line impedance stabilizing network is the load.
The level of disturbances reinjected may not exceed the limits defined in the standards. These limits depend on the type of cable and the environment. The graph below (see fig. 32) shows the results of a measurement performed on an uninterruptible power supply and the levels defined in standard EN 55 022 for comparison. Radiated emissions Radiated emission measurements quantify the level of disturbances emitted by a device in the form of electromagnetic waves. Just as with radiated susceptibility tests, radiated emission tests must be performed in the absence of waves normally present such as CB, radio etc. and must not be modified by reflections
from surrounding objects. These two conditions are contradictory and this is the reason for the existence of two test methods. The first method consists of placing the EUT in a field free of obstacles within a given perimeter. The environment is uncontrolled. The second method is implemented in a Faraday room; the reflections from the walls are deliberately attenuated by high frequency absorbing materials (see fig. 29). The environment can be perfectly controlled. The Merlin Gerin laboratory uses the second method. It offers a key advantage in that measurements can be automated and that equipment
handling is minimized, since emission and susceptibility level measurements can be performed at the same site with just few setup changes. As for conducted emissions, the emission levels must be less than the limits set by specifications or standards. Measuring pulsed fields Standardized tests are performed to measure emission levels or test the susceptibility of devices or systems to the most common types of electromagnetic disturbances encountered in an industrial environment. However, the environment for devices developed by Merlin Gerin has certain characteristics not yet covered by standards.
fig. 32: results of measurements performed on a Maxipac SX5000 uninterruptible power supply.
For example, specific EMC test procedures for equipment in medium voltage substations do not yet exist. This is why Merlin Gerin performs a series of measurements to better understand the typical disturbances that exist in the vicinity of the
equipment it manufactures, especially near low, medium and very high voltage switchgear. In a second phase, in-house tests using special test systems have been developed. They allow testing of the electromagnetic compatibility of
devices without having to revert to full scale tests. These tests are easier to reproduce and less costly. They are performed early in the design which minimizes costs for EMC protection.
7. conclusion
The use of electronics in a large number of applications, and especially in electrotechnical equipment, has introduced a new and important requirement: electromagnetic compatibility. Trouble-free operation in disturbed environments and operation without producing disturbances are essential to product quality requirements. To achieve both these goals, the complex phenomena involved in the sources, coupling and susceptors must
be well understood. A certain number of rules must be followed in the design, industrialization and manufacture of products. The site and installation characteristics also play an important role in electromagnetic compatibility. This explains the importance of carefully considering the location and layout of power components, cable routing, shielding etc. right from the intitial design phase. Even if equipment
offers satisfactory EMC, a well designed installation can extend the compatibility safety margins. Only measurements requiring a high level of expertise and sophisticated equipment can produce valid results quantifying the electromagnetic compatibility of equipment. Compliance with standards therefore provides the certainty that equipment will operate satisfactorily in its electromagnetic environment.
appendix 1: glossary
Electromagnetic compatibility, EMC (abbreviation) (IEV 161-01-07) The ability of an equipment or system to function satisfactorily in its electromagnetic environment without introducing intolerable electromagnetic disturbances to anything in that environment. (Electromagnetic) compatibility level (IEV 161-03-10) The specified maximum disturbance level expected to be impressed on a device, equipment or system operated in particular conditions. Note: In practice the electromagnetic compatibility level is not an absolute maximum level but may be exceeded by a small probability. Electromagnetic disturbance (IEV 161-01-05) Any electromagnetic phenomenon which may degrade the performance of a device, equipment or system, or adversely affect living or inert matter. Note: An electromagnetic disturbance may be an electromagnetic noise, an unwanted signal or a change in the propagation medium. Disturbance level (not defined in IEV 161) Level of an electromagnetic disturbance of a given form measured in particular conditions. Limit of disturbance (IEV 161-03-08) The maximum permissible electromagnetic disturbance level, as measured in a specified way. Immunity level (IEV 161-03-14) The maximum level of a given disturbance incident on a particular device, equipment or system for which
it remains capable of operating at a required degree of performance. (Electromagnetic) susceptibility (IEV 161-01-21) The inability of a device, equipment or system to perform without degradation in the presence of an electromagnetic disturbance. Note: Suceptibility is a lack of immunity. Decibel The decibel is a unit of sound pressure that is also used to express amplitude ratios according to X/Xo (dB@) = 20 . log10 X/Xo, with X = measured amplitude, Xo = reference amplitude, @ = mesure unit for X and Xo. A few sample values are given in the table below (see fig. 33).
amplitude ratio X/Xo 1 1.12 1.25 1.41 2 3.2 4 5 10 100 1000
dB
0 1 2 3 6 10 12 14 20 40 60
fig. 33: amplitude ratios expressed in decibels.
appendix 2: impedance of a conductor at high frequencies
The level of EMC in equipment depends on coupling between circuits. Coupling is directly related to the impedance between circuits, especially at high frequencies. To improve EMC, these impedances must be determined and then reduced. A few approximating formulae exist to determine the high frequency impedance of typical conductors. These formulae are cumbersome and their results meaningless if the exact position of all involved elements is unknown. But who knows the exact position of a wire with respect to the others in a cable trough? The answers to this and similar questions come from experience together with basic knowledge of the theory of electrical phenomena. First of all it is important to keep in mind that the impedance of a conductor is mainly a function of its inductance and becomes preponderant starting at a few kilohertz for a standard wire. For a wire assumed to be infinitely long, the inductance per unit length increases logarithmically with the diameter, therefore very slowly: for wires that do not exceed 1/4 of the disturbance wavelength, an inductance of one H/m can be used irrespective of the diameter (see fig. 34). This value is much lower when the wire is correctly run against a conductive plane. It becomes a function of the distance between the wire and the plane and the inductance can easily be decreased by 10 dB. At very high frequencies the wire must be considered as a transmission line with a characteristic impedance of around one hundred ohms. In this light, a common inductance of several H can easily be created, for
example, with a few meters of greenyellow (grounding) wire. This translates into a few ohms at 1 MHz and a few hundred ohms at 100 MHz. Conclusion A conducting metal plate represents the electrical interconnect offering the
lowest impedance, independent of thickness as long as it is greater than the skin depth (415 m at 10 kHz for copper). A copper plate displays an inductance of 0.6 nH (at 10 kHz) and a resistance of 37 per square.
(a)
(b) Z2 Z1
(c)
(d)
Z3
Z4
fig. 34: at equal lenghts, the different impedances: a: wire in air (l 1 H/m), b: cable placed on a metal plane, c: metal grid with electrical contact at each node (e.g. welded concrete rebar), d: metal plane, have a per unit lenght impedance Z1 > Z2 > Z3 > Z4.
appendix 3: the different parts of a cable
The technical terms used to describe different parts of a cable can have slightly different meanings depending on the cables field of application (power transmission, telephone, data or control and monitoring), (see fig. 35). The IEC definitions are in italics. Jacket The jackets most important role is to protect the cable from mechanical damage. That is why it usually contains two helically stranded soft steel sheets (NF C 32-050). For data transmission cables, it also serves as an electrostatic and more often electromagnetic shield. Shield Same as a screen; i.e. device designed to reduce the intensity of electromagnetic radiation penetrating into a certain region. A jacket or screen of a cable, whether for power or data transmission, can form a shield. Screen A device used to reduce the penetration of a field into an assigned region It has multiple functions: s creation of an equipotential surface around the insulator, s protection against the effects of external an internal electrostatic fields,
s draining the capacitive current as well as earth leakage fault currents (zero sequence short-circuits), s protection of life and property in the event of a puncture.
That is why it is generally made of metal and is continuous (lead tubing, braided wire, helically wound bands). For cables carrying data, the screen, more often called a shield, consists of copper or aluminum wire bands or braids, wrapped around to form a
shield against electrostatic or electromagnetic fields. It can be an overall shield, for all conductors in the cable, when the disturbances are external to the cable. It can also be partial, for a limited number of conductors, to protect against disturbances emitted by the other conductors in the cable. Insulator The insulator renders the cable water and/or air tight.
Telephone cable
Medium voltage power transmission cable insulator (PVC)

insulator (PVC) jacket conductor (two steel bands) internal insulation (PVC) metal screen (aluminum) insulator (PVC)
jacket conductor (two steel sheets) cushion (paper) metal screen (copper) conductive ribbon filler insulator (PVC)
core (copper wire)
core (copper wire)
fig. 35.
appendix 4: tests performed at the Merlin Gerin EMC laboratory
The Merlin Gerin EMC laboratory has the required equipment and expertise to perform a large number of tests.
standards compliance tests

Immunity s IEC 801-2 (1984) Electromagnetic compatibility for industrial-process measurement and control equipment - Part 2: Electrostatic discharge requirements. s IEC 801-3 (1984) Electromagnetic compatibility for industrial-process measurement and control equipment - Part 3: Radiated electromagnetic field requirements. s IEC 801-4 (1988) Electromagnetic compatibility for industrial-process measurement and control equipment - Part 4: Electrical fast transient/burst requirements. s NF C 63-850 (October 1982) Programmable controllers 10-2-8-1 and 10-2-8-3: electromagnetic compatibility tests.
Emission s EN 55 011 (to be published) Limits and methods of measurement of electromagnetic disturbance characteristics of Industrial, Scientific and Medical (ISM) radio-frequency equipment. [conducted emission part] s EN 55 022 Limits and methods of measurement of radio interference characteristics of information technology equipment. [conducted emission part] s VDE 0871 (June 1978) Disturbance suppression for Industrial, Scientific and Medical (ISM) or equivalent high frequency equipment. Specific standards
s Telecommunications s
[the parts on immunity of equipment to radiated disturbances, radiated and conducted disturbances created by equipment].
s military s GAMEG13 62C1*, 62C2, 62R1*, 62R2, 62R3**, 63C1, 63C2, 63C3, 63C4, 63R1, 63R2, 63R3**, s
MIL STD 461/462 CE01*, CE03, RE01*, RE02**, CS01, CS02, CS06, RS01, RS02, RS03** * : low frequency limit = 10 kHz ** : high frequency limit = 1 GHz
non-standardized tests
Within the limits of available expertise and facilities, the laboratory can perform tests complying with other standards.
centres
I 12-10 (1988) published by the Committee for Equipment Specifications (CSE), France Telecom. Electromagnetic environment of equipment in a telecommunications centre.
appendix 5: bibliography
Standards s IEC 1000-2-1 s IEC 1000-2-2 s IEC 801-1 to 801-4 s EN 55 011, CISPR 11 s EN 55 022, CISPR 22 s NF C 15-100 Merlin Gerin Cahiers Techniques Publications s CT 141 : les perturbations lectriques en BT - R. CALVAS. s E/CT 143 : Behaviour of the SF6-MV circuit breakers Fluarc for switching motor starting currents J. HENNEBERT and D. GIBBS. Other publications lectromagntique bruits et perturbations radiolectriques P. DEGAUQUE et J. HAMELIN Dunod diteur. s Compatibilit lectromagntique M. IANOVICI et J.-J. MORF Presses Polytechniques Romandes. s RGE no 10 (Novembre 1986) dedicaced to electromagnetic compatibility.
s Compatibilit
Ral. : Illustration Technique Lyon - Photo : IPV
IPV 10-92 - 2500 - Imp. Lostic
n 152
harmonics in industrial networks
Pierre Roccia Obtained an Electrical Engineering degree from the INPG (National Polytechnic Institute of Grenoble) in 1969. Worked as project manager in the industrial equipment and high voltage public distribution sector, before being put in charge of extending the Merlin Gerin range of protection relays and developing a technical approach for the protection of high voltage industrial networks using devices associated with circuit breakers. After three years as a training instructor, he is presently working as an engineer in the "Network Studies" department of the Central R&D organisation. Nol Quillon After joining Merlin Gerin's Low Voltage Equipment Department in 1968, he subsequently took part in the development of LV circuit breakers within the testing laboratory. A graduate engineer from the INPG, he worked in the "Network Studied" department of the Central R&D organisation for eight years where he studied electrical network phenomena and their behaviour in order to establish guidelines to control these phenomena. In 1985, he joined the Training Department. After being in charge of the electrotechnical training programme, he is presently the training correspondent for the UPS division.
E/CT 152 first issued october 1994
glossary
Symbols: C D
f1 far fn fr
n In
j L Lsc n nar nr k p p1 pn P (W) PB q Q Q (var) r R spectrum Ssc T U Vn X X0 Xsc Y0 Yn Z
capacitance or, more generally, the capacitors themselves harmonic distortion loss angle of a capacitor fundamental frequency anti-resonance frequency frequency of the nth harmonic component resonance frequency phase angle of the nth harmonic component when t = 0 rms current of the nth harmonic component complex operator equal to the square root of 1 inductance or, more generally, the reactors, producing the inductance short-circuit inductance of a network, seen from a given point, as defined by Thevenin's theorem the order of a harmonic component (also referred to as the harmonic number) the order of anti-resonance, i.e. the radio of the anti-resonance frequency to the fundamental frequency the order of resonance, i.e. the radio of the resonance frequency to the fundamental frequency a positive integer number of rectifier arms (also referred to as the pulse number) filter losses due only to the fundamental current filter losses due only to the nth harmonic current active power pass-band of a resonant shunt filter quality factor of a reactor quality factor of a filter reactive power resistance resistance (or the real part of the impedance) the distribution, at a given point, of the amplitudes of the various harmonic components expressed relative to the fundamental short-circuit power of a network at a given point period of an alternating quantity phase-to-phase rms voltage rms voltage of the nth harmonic component reactance characteristic inductance or impedance of a filter short-circuit reactance of a network, seen from a given point, as defined by Thevenin's theorem amplitude of the DC component rms value of the nth harmonic component impedance
Abbreviations: CIGRE Confrence Internationale des Grands Rseaux Electriques (International Conference on Large Electrical Networks) IEC International Electrotechnical Commission
summary
1. Introduction: harmonic distortion is a problem that must often be p. 4 dealt with in industrial power distribution networks 2. Harmonic quantities p. 4 3. Principal disturbances caused by Instantaneous effects p. 6 harmonic currents and voltages Long-term effects p. 6 4. Acceptable limits, recommendations Typical limits for distribution p. 7 and standards networks Typical limits for industrial p. 7 networks 5. Harmonics generators Static converters on 3-phase p. 8 networks Arc furnaces p. 8 Lighting p. 9 Saturated reactors p. 9 Rotating machines p. 9 Calculation model p. 9 6. Can capacitors cause a problem on In the absence of capacitor banks p. 10 networks comprising disturbing In the presence of a capacitor p. 10 equipment bank 7. Anti-harmonic reactors p. 13 8. Filters Resonant shunt filters p. 14 Damped filters p. 15 9. Measurement relays required for the Basic protection against device p. 17 protection of reactor-connected failures capacitors and filters Basic protection against abnormal p. 17 stresses on the devices 10. Example of the analysis of a Capacitor bank alone p. 18 simplified network Reactor-connected capacitor bank p. 18 Resonant shunt filter tuned to the p. 19 5th harmonic and a damped filter tuned to the 7th harmonic 11. Conclusion p. 21 12. Bibliography p. 22
1. introduction
harmonic distortion is a problem that must often be dealt with in industrial power distribution networks
Electricity is generally distributed as three voltage waves forming a 3-phase
sinusoidal system. One of the characteristics of such a system is its waveform, which must always remain as close as possible to that of a pure sine wave. If distorted beyond certain limits, as is often the case on networks comprising sources of harmonic currents and
voltages such as arc furnaces, static power converters, lighting systems, etc., the waveform must be corrected. The aim of the present document is to provide a better understanding of these harmonics problems, including their causes and the most commonly used solutions.
2. harmonic quantities
To help the reader follow the discussion, we will first review the definitions of a number of terms related to harmonics phenomena. Readers already familiar with the basic terminology may proceed directly to the next chapter. On AC industrial power supply networks, the variation of current and voltage with time is considerably different from that of a pure sine wave (see fig. 1). The actual waveform is composed of a number of sine waves of different frequencies, including one at the power frequency, referred to as the fundamental component or simply the fundamental. Harmonic component The term harmonic component, or simply harmonic, refers to any one of the above-mentioned sinusoidal components, the frequency of which is a multiple of that of the fundamental. The amplitude of a harmonic is generally a few percent of that of the fundamental. Harmonic order The harmonic order, also referred to as the harmonic number, is the ratio of the frequency fn of a harmonic to that of the fundamental (generally the power frequency, i.e. 50 or 60 Hz):
By definition, the harmonic order of the fundamental f1 is equal to 1. Note that the harmonic of order n is often referred to simply as the nth harmonic. Spectrum The spectrum is the distribution of the amplitudes of the various harmonics as a function of their harmonic number, often illustrated in the form of a histogram (see fig. 2).
Expression of the distorted wave Any periodic phenomenon can be represented by a Fourier series as follows:
y(t) = Y 0 + Y n 2 sin (nt n )

n = 1
n =
where: s Y0 = the amplitude of the DC component, which is generally zero in electrical power distribution;
I phase
fundamental distorted wave
harmonic
n =
fn . f1
fig.1: shape of a distorted wave.
s Yn = the rms value of the nth harmonic component, s n = phase angle of the nth harmonic component when t = 0. Harmonics with an order above 23 are often negligible.
true rms value, by thermal means or by spectrum analysers. Individual harmonic ratio and total harmonic distortion The industrial harmonic ratios and the total harmonic distortion quantify the harmonic disturbances present in a power supply network. s individual harmonic ratio (or harmonic percentage) The harmonic ratio expresses the magnitude of each harmonic with respect to the fundamental (see fig. 2). The nth harmonic ratio is the ratio of the rms value of the nth harmonic to that of the fundamental. For example, the harmonic ratio of In is In/I1 or 100 (In/I1) if expressed as a percentage (note that here In is not the nominal or rated current); s total harmonic distortion (also referred to as THD, the total harmonic factor or simply as distortion D). The total harmonic distortion quantifies the thermal effect of all the harmonics. It is the ratio of the rms value of all the harmonics to that of one of the two following quantities (depending on the definition adopted): the fundamental (CIGRE), which can give a very high value:
n =
the measured rms quantity (IEC 555-1), in which case 0 < D < 1:
n =
D =
n = 2 n = n = 1
2 Yn 2 Yn
Rms value of a distorted wave Harmonic quantities are generally expressed in terms of their rms value since the heating effect depends on this value of the distorted waveform. For a sinusoidal quantity, the rms value is the maximum value divided by the square root of 2. For a distorted quantity, under steadystate conditions, the energy dissipated by the Joule effect is the sum of the energies dissipated by each of the harmonic components:
2 2 R I 2 t = R I1 t + R I 2 t + ... + R I n t 2
Unless otherwise indicated, we will use the definition adopted by CIGRE (see the glossary) which corresponds to the ratio of the rms value of the harmonic content to the undistorted current at power frequency.
100 %
where:
2 2 I 2 = I1 + I 2 + ... + I n 2
i.e. where:
I =
n = n = 1
2 In
if the resistance can be considered to be constant. The rms value of a distorted waveform can be measured either directly by instruments designed to measure the
D =
n = 2
2 Yn
Y1
fig. 2: the amplitude of a harmonic is often expressed with respect to that of the fundamental.
3. principal disturbances caused by harmonic currents and voltages
Harmonic currents and voltages superimposed on the fundamental have combined effects on equipment and devices connected to the power supply network. The detrimental effects of these harmonics depend on the type of load encountered, and include: s instantaneous effects; s long-term effects due to heating.
running, the distance between the two circuits and the harmonic frequencies (coupling increases with frequency).
long-term effects
Over and above mechanical fatigue due to vibrations, the main long-term effect of harmonics is heating. Capacitor heating The losses causing heating are due to two phenomena: conduction and dielectricc hysteresis. As a first approximation, they are proportional to the square of the applied voltage for conduction and to the frequency for hysteresis. Capacitors are therefore sensitive to overloads, whether due to an excessively high fundamental or to the presence of voltage harmonics. These losses are defined by the loss angle of the capacitor, which is the angle whose tangent is the ratio of the losses to the reactive power produced (see fig. 3). Values of around 10-4 may be cited for tan . The heat produced can lead to dielectric breakdown. Heating due to additional losses in machines and transformers s additional losses in the stators (copper and iron) and principally in the rotors (damping windings, magnetic circuits) of machines caused by the considerable differences in speed between the harmonic inducing rotating fields and the rotor. Note that rotor measurements (temperature, induced currents) are difficult if not impossible. s supplementary losses in transformers due to the skin effect (increase in the resistance of copper with frequency), hysteresis and eddy currents (in the magnetic circuit). Heating of cables and equipment Losses are increased in cables carrying harmonic currents, resulting in temperature rise. The causes of the additional losses include: s an increase in the apparent resistance of the core with frequency, due to the skin effect;
instantaneous effects
Harmonic voltages can disturb controllers used in electronic systems. They can, for example, affect thyristor switching conditions by displacing the zero-crossing of the voltage wave (see IEC 146-2 and Merlin Gerin Cahier Technique n 141). Harmonics can cause additional errors in induction-disk electricity meters. For example, the error of a class 2 meter will be increased by 0.3 % by a 5th harmonic ratio of 5 % in current and voltage. Ripple control receivers, such as the relays used by electrical utilities for centralised remote control, can be disturbed by voltage harmonics with frequencies in the neighbourhood of the control frequency. Other sources of disturbances affecting these relays, related to the harmonic impedance of the network, will be discussed further on. Vibrations and noise The electrodynamic forces produced by the instantaneous currents associated with harmonic currents cause vibrations and acoustical noise, especially in electromagnetic devices (transformers, reactors, etc.). Pulsating mechanical torque, due to harmonic rotating fields, can produce vibrations in rotating machines. Interference on communication and control circuits (telephone, control and monitoring) Disturbances are observed when communication or control circuits are run along side power distribution circuits carrying distorted currents. Parameters that must be taken into account include the length of parallel
an increase in dielectric losses in the insulation with frequency, if the cable is subjected to non-negligible voltage distortion; s phenomena related to the proximity of conductors with respect to metal cladding and shielding earthed at both ends of the cable, etc. Calculations can be carried out as described in IEC 287. Generally speaking, all electrical equipment (electrical switchboards) subjected to voltage harmonics or through which harmonic currents flow, exhibit increased energy losses and should be derated if necessary. For example, a capacitor feeder cubicle should be designed for a current equal to 1.3 times the reactive compensation current. This safety factor does not however take into account the increased heating due to the skin effect in the conductors. Harmonic distortion of currents and voltages is measured using spectrum analysers, providing the amplitude of each component. The rms value of the distorted current (or voltage) may be assessed in any of three ways: s measurement using a device designed to give the true rms value, s reconstitution on the basis of the spectrum provided by spectral analysis, s estimation from an oscilloscope display.
s
tan =
p Q
fig. 3: triangle relating to the capacitor powers, (active (P), reactive (Q), apparent (R)).
4. acceptable limits, recommendations and standards
General limits s synchronous machines: permissible stator current distortion = 1.3 to 1.4 %; s asynchronous machines: permissible stator current distortion = 1.5 to 3.5 %; s cables: permissible core-shielding voltage distortion = 10 %; s power capacitors: current distortion = 83 %, corresponding to an overload of 30 % (1.3 times the rated current); overvoltages can reach up to 10 % (see IEC 871-1, 931-1 and HD 525.1S1); s sensitive electronics: 5 % voltage distortion with a maximum individual harmonic percentage of 3 % depending on the equipment.
typical limits for industrial networks

It is generally accepted that industrial network without any sensitive equipment such as regulators,
programmable controllers, etc. can accept up to 5 % voltage distortion. This limit and the limits for the individual harmonic ratios may be different if sensitive equipment is connected to the installation.
harmonic order 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23
low value (%) 1 1.5 0.5 5 0.2 4 < 0.2 0.8 < 0.2 2.5 < 0.2 2 < 0.2 < 0.3 < 0.2 1 < 0.2 0.8 < 0.2 < 0.2 < 0.2 0.5
high value (%) 1.5 2.5 1 6 0.5 5 1.5 3.5 3
typical limits for distribution networks

The French electrical utility, EDF, considers that voltage distortion will not exceed 5 % at the supply terminals as long as each individual subscriber does not exceed the following limits: s 1.6 % voltage distortion; s individual harmonic percentages of: 0.6 % for even voltage harmonics, 1 % for odd voltage harmonics. The table in figure 4 lists typical percentages observed for the various voltage harmonics where: s low value = value likely to be found in the vicinity of large disturbing loads and associated with a low probability of having disturbing effects; s high value = value rarely exceeded in the network, and with a higher probability of having disturbing effects.
2 1.5
fig. 4: individual voltage harmonic percentages measured in high voltage distribution networks.
5. harmonics generators
In industrial applications, the main types equipment that generate harmonics are: s static converters; s arc furnaces; s lighting; s saturated reactors; s other equipment, such as rotating machines which generate slot harmonics (often negligible).
static converters on 3-phase networks

Rectifier bridges and, more generally, static converters (made up of diodes and thyristors) generate harmonics. A Graetz bridge, for instance, requires a rectangular pulsed AC current (see fig. 5) to deliver a perfect DC current. In spite of their different waveforms, the currents upstream and downstream from the delta-star connected transformer have the same characteristic harmonic components. The characteristic harmonic components of the current pulses supplying rectifiers have the following harmonic numbers n, with n = kp 1, where: s k = 1, 2, 3, 4, 5... s p = number of rectifier arms, for example: Graetz bridge p = 6, 6-pulse bridge p = 6, 12-pulse bridge p = 12. Applying the formula, the p = 6 rectifiers cited above generate harmonics 5, 7, 11, 13, 17, 19, 23 and 25, and the p = 12 rectifiers generate harmonics 11, 13, 23 and 25. The characteristic harmonics are all odd-numbered and have, as a first approximation, an amplitude of In = I1/n where I1 is the amplitude of the fundamental. This means that I5 and I7 will have the greatest amplitudes. Note that they can be eliminated by using a 12-pulse bridge (p = 12). In practice, the current spectrum is slightly different. New even and odd harmonics, referred to as non-
characteristic harmonics, of low amplitudes, are created and the amplitudes of the characteristic harmonics are modified by several factors including: s asymmetry; s inaccuracy in thyristor opening times; s switching times; s imperfect filtering. For thyristor bridges, a displacement of the harmonics as a function of the thyristor phase angle may also be observed. Mixed thyristor-diode bridges generate even harmonics. They are used only at low ratings because the 2nd harmonic produces serious disturbances and is very difficult to eliminate. Other power converters such as cycloconverters, dimmers, etc. have richer and more variable spectra than rectifiers. Note that they are increasingly replaced by converters using the PWM (Pulse Width Modulation) technique. These devices
operate at high chopping frequencies (20 to 50 kHz) and are generally designed to generate only low levels of harmonics. The harmonic currents of several converters combine vectorially at the common supply busbars. Their phases are generally unknown except for the case of diode rectifiers. It is therefore possible to attenuate the 5 th and 7 th current harmonics using two equally loaded 6-pulse diode bridges, if the couplings of the two power supply transformers are carefully chosen (see fig. 6).
arc furnaces
Arc furnaces used in the steel industry may be of the AC (see fig. 7) or DC type. AC arc furnaces (see fig. 7) The arc is non-linear, asymmetric and unstable. It generates a spectrum
load
T/6
T/3
T/6 phase current upstream from a delta-star connected transformer supplying the rectifier
rectifier supply phase current
fig. 5: alternating current upstream from a Graetz bridge rectifier delivering a perfect direct current.
including odd and even harmonics as was well as a continuous component (background noise at all frequencies). The spectrum depends on the type of furnace, its power rating and the operation considered (e.g. melting, refining). Measurements are therefore required to determine the exact spectrum (see fig. 8). DC arc furnaces (see fig. 9) The arc is supplied via a rectifier and is more stable than the arc in AC furnaces. The current drawn can be broken down into: s a spectrum similar to that of a rectifier; s a continuous spectrum lower than that of an AC arc furnace.
To a large extent, the harmonic currents drawn by the disturbing equipment are independent of the other loads and the overall network impedance. These currents can therefore be considered to be injected into the network by the disturbing equipment. It is simply necessary to arbitrarily change the sign so that, for calculation purposes, the disturbing equipment can be considered as current sources (see fig. 10). The approximation is somewhat less accurate for arc furnaces. In this case, the current source model must be corrected by adding a carefully selected parallel impedance.
in %
In I1
100 10
100 continuous spectrum 4 3.2 1.3 0.5
1 0.1 1 3 5 7
9 rang
fig. 8: current spectrum for an arc furnace supplied by AC power.
lighting
Lighting systems made up of discharge lamps or fluorescent lamps are generators of harmonic currents. A 3rd harmonic ratio of 25 % is observed in certain cases. The neutral conductor then carries the sum of the 3rd harmonic currents of the three phases, and may consequently be subjected to dangerous overheating if not adequately sized.
I5 and I7 attenuated
HV transformer
I5 and I7
I5 and I7
cable
Dy 11
Yy 0
rectifier
saturated reactors
The impedance of a saturable reactor is varying with the current flowing through it, resulting in considerable current distortion. This is, for instance, the case for transformers at no load, subjected to a continuous overvoltage.
6-pulse diode bridge load equal loads load
6-pulse diode bridge
cable
furnace
rotating machines
Rotating machines generate high order slot harmonics, often of negligible amplitude. However small synchronous machines generate 3rd order voltage harmonics than can have the following detrimental effects: s continuous heating (without faults) of earthing resistors of generator neutrals; s malfunctioning of current relays designed to protect against insulation faults.
fig. 6: attenuation circuit for I5 and I7.

HV transformer
fig. 9: arc furnace supplied by DC power.
cable
calculation model
When calculating disturbances, static converters and arc furnaces are considered to be harmonic current generators.
furnace
fig. 7: arc furnace supplied by AC power.
fig. 10: harmonic current generators are modelled as current sources.
6. can capacitors cause a problem on networks comprising disturbing equipment?
We will consider the two following cases: s networks without power capacitors; s networks with power capacitors.
in the absence of capacitor banks, harmonic disturbances are limited and proportional to the currents of the disturbing equipment.
In principle, in so far as are concerned harmonics, the network remains inductive. Its reactance is proportional to the frequency and, as a first estimate, the effects of loads and resistance are negligible. The impedance of the network, seen from a network node, is therefore limited to the short-circuit reactance Xsc at the node considered. The level of harmonic voltages can be estimated from the power of the disturbing equipment and the shortcircuit power at the node (busbars) to which the disturbing equipment is connected, the short-circuit reactance considered to be proportional to the frequency (see fig. 11). In figure 11: Lsc = the short-circuit inductance of the network, seen from the busbars to which the disturbing equipment is connected, In = currents of the disturbing equipment, Xsc n = Lsc n = Lsc n (2 f1) therefore V n = Xsc n I n = Lsc n (2 f1) I n. The harmonic disturbances generally remain acceptable as long as the disturbing equipment does not exceed a certain power level. However, this must be considered with caution as resonance (see the next section) may be present, caused by a nearby network possessing capacitors and coupled via a transformer.
Note: In reality, the harmonic inductance of network X, without capacitors (essentially a distribution network), represented by Lsc, can only be considered to be proportional to the frequency in a rough approximation. For this reason, the network shortcircuit impedance is generally multiplied by a factor of 2 or 3 for the calculations. Therefore: Xn = k n X1 with k = 2 or 3. The harmonic impedance of a network is made up of different constituents such as the short circuit impedance of the distribution system as well as the impedance of the cables, lines, transformers, distant capacitors, machines and other loads (lighting, heating, etc.).
In
Xsc
Vn
fig. 11: the harmonic voltage Vn is proportional to the current In injected by the disturbing equipment.
node A (busbars)
in the presence of a capacitor bank parallel resonance can result in dangerous harmonic disturbances
Resonance exists between the capacitor bank and the reactance of the network seen from the bank terminals. The result is the amplification, with a varying degree of damping, of the harmonic currents and voltages if the order of the resonance is the same as that of one of the harmonic currents injected by the disturbing equipment. This amplified disturbance can be dangerous to the equipment. This is a serious problem and will be dealt with in below. This phenomenon is referred to as parallel resonance. What is this parallel resonance and how can it cause dangerous harmonic disturbances? In so far as harmonic frequencies are concerned, and for a first approximation, the network may be represented as in figure 12.
Lsc
load
Vn
In
0 a: harmonic electrical representation of a phase.
50 Hz source
Lsc node A (busbar)
load
disturbing equipment
b: single-line diagram.
fig. 12: equivalent diagrams for a circuit subject to harmonic currents and including a capacitor bank.
In this diagram: Lsc = the short-circuit inductance of the network seen from the busbars to which the capacitor bank and the disturbing equipment are connected, C = capacitors, In = currents of the disturbing equipment, load = loads (Joule effect, transmission of mechanical energy). In principle, we consider the shortcircuit harmonic reactance seen from the busbars, i.e. the node (A) to which the capacitors, the loads and the disturbing equipment are connected, giving Vn = ZAO In. The impedance versus frequency curves (see fig.13) show that: s for the resonance frequency far, the inductive effect is compensated for exactly by the capacitive effect; s the reactance of the rejecter circuit: is inductive for low frequencies, including the fundamental frequency, increases with frequency, becoming very high and suddenly capacitive at the resonance frequency far; s the maximum impedance value reached is roughly R = U2/P where P represents the sum of the active power values of the loaded motors, other than those supplied by a static converter. If a harmonic current In of order n , with the same frequency as the parallel resonance frequency far, is injected by the disturbing equipment, the corresponding harmonic voltage can be estimated as Vn = R In with n = n ar = f ar/f1. Estimation of nar The order nar of parallel resonance is the ratio of the resonance frequency far to the fundamental frequency f1 (power frequency). Consider the most elementary industrial network, shown in the equivalent diagram in figure 14, including a capacitor bank C supplied by a transformer with a short-circuit inductance LT, where Lsc represents the short-circuit inductance of the distribution network seen from the upstream terminals of the transformer,
f ar = 2 1 . (Lsc + L T ) C
impedance is seen from point A or point B (e.g. the supply terminals). In general, given the short-circuit power at the capacitor bank terminals,
nar = Ssc Q
and undoubtedly present a danger to the capacitors. s if the parallel resonance order corresponds to the frequency of the carrier-current control equipment of the power distribution utility, there is a risk of disturbing this equipment. To prevent resonance from becoming dangerous, it must be forced outside the injected spectrum and/or damped. The short-circuit impedance of the network is seldom accurately known and, in addition, it can vary to a large extent, thereby resulting in large variations of the parallel resonance frequency. It is therefore necessary to stabilise this frequency at a value that does not correspond to the frequencies of the injected harmonic currents. This is achieved by connecting a reactor in series with the capacitor bank.
where: Ssc = short-circuit power at the capacitor bank terminals, Q = capacitor bank power at the applied voltage. Generally S is expressed in MVA and Q in Mvar. Practical consequences: s if the order of a harmonic current injected by disturbing equipment corresponds to the parallel resonance order, there is a risk of harmonic overvoltages, especially when the network is operating at low loads. The harmonic currents then become intensively high in network constituents
X inductive 0 f (Hz) capacitive far without capacitors X = Lsc 2 f
IZI
~R without capacitors IZI = Lsc 2 f
0 far
f (Hz)
fig. 13: curves showing the impedance due to the loads and due to the resistance of the conductors.
LT
lopp
distributor Lsc C load
The order of the parallel resonance is roughly the same whether the network
fig. 14: the capacitor, together with the sum of the upstream impedances, forms a resonant circuit.
The rejecter circuit thus created is then represented by the diagram in figure 15 where Vn = ZAO In. A series resonance, between L and C, appears. As opposed to this resonance, which gives a minimum impedance, the parallel resonance is often referred to an anti-resonance. The equation giving the frequency of the anti-resonance is:
f ar = 2 1 (Lsc + L) C
s a minimum resistive value r (resistance of the inductance coil) for the resonance frequency fr; s a capacitive reactance below the resonance frequency fr; s an inductive reactance above the resonance frequency fr, where
fr =
1 2 L C
Lsc generally being small compared to L, the equation shows that the presence of reactor L, connected in series with the capacitors, renders the frequency far less sensitive to the variations of the short-circuit inductance Lsc (from the connections points = busbars A). Series resonance The branch made up of reactor L and capacitor C (see fig. 16), form a series resonance system of impedance Z = r + j(L - 1/C) with:
The curves in figure 17 show the shape of the network inductance, including the short-circuit impedance and that of the LC branch, seen from busbars A. The choice of far depends on Lsc, L and C, while that of fr depends only on L and C; far and fr therefore become closer as Lsc becomes small with
respect to L. The level of reactive power compensation, and the voltage applied to the capacitors, depend partly on L and C. The reactor L can be added in two different manners, depending on the position of the series resonance with respect to the spectrum. The two forms of equipment are: s anti-harmonic reactors (for series resonance outside the spectrum lines); s filters (for series resonance on a spectrum line).
X
inductive
ph1 inductive r capacitive 0
X
capacitive
0 f (Hz) fr far
f (Hz)
busbar node, point A
fr L IZI
IZI
L Lsc C 0 Vn
In
C
0
neutral
r f (Hz) fr
far
~r
f (Hz) fr
fig. 15: the reactor, connected in series with the capacitor, forms a rejecter circuit.
fig. 16: impedance of the rejecter circuit.
fig. 17: network impedance at point A.
7. anti-harmonic reactors
An anti-harmonic reactor can be used to protect a capacitor bank against harmonic overloads. Such solutions are often referred to as reactor-connected capacitor installations. The reference diagram is once again figure 15. In this assembly, the choice of L is such that the LC branch (where L is the reactor and C the reactive power compensation capacitors) behaves inductively for the harmonic frequencies, over the spectrum. As a result, the resonance frequency fr of this branch will be below the spectrum of the disturbing equipment. The LC branch and the network (Lsc) are then both inductive over the spectrum and the harmonic currents injected by the disturbing equipment are divided in a manner inversely proportional to the impedance. Harmonic currents are therefore greatly restricted in the LC branch, protecting the capacitors, and the major part of the harmonic currents flow in the rest of the network, especially in the shortcircuit impedance. The shape of the network impedance, seen from the busbars to which the LC branch is connected, is shown in figure 18. There is no anti-resonance inside the current spectrum. The use of an antiharmonic reactor therefore offers two advantages;
s it eliminates the danger of high harmonic currents in the capacitors; s it correlatively eliminates the high distortions of the network voltage, without however lowering them to a specified low value. Certain precautions are necessary: s no other capacitor banks must be present that could induce, through antiresonance, a capacitive behaviour in the initial network inside the spectrum; s care must be taken not to introduce an anti-resonance with a frequency used by the distribution utility for
carrier-current control, since this would place an increased load on the high frequency generators (175 Hz, 188 Hz). The anti-harmonic reactor is tuned to an order of 4.5 to 4.8, giving a value of fr between 225 to 240 Hz for a 50 Hz network, which is very near the ripple control frequency used on many distribution networks; s due to the continuous spectrum, the use of anti-harmonic reactors on arc furnaces requires certain precautions which can only be defined after carrying out special studies.
IzI
theoretical impedance without the LC branch
f (Hz) f1 fr harmonic current spectrum
far
fig. 18: the capacitors are protected when fr is well below the harmonic spectrum.
8. filters
Filters are used when it is necessary to limit harmonic voltages present on a network to a specified low value. Two types of filters may be used to reduce harmonic voltages: s resonant shunt filters, s damped filters.
there are as many anti-resonances as there are filters. These anti-resonances must be tuned to frequencies between the spectrum lines. A careful study must therefore be carried out if it is judged necessary to segment the filter bank. Main characteristics of a resonant shunt The characteristics depend on n r = fr/f1 the order of the filter tuning frequency, with: s fr = tuning frequency; s f1 = fundamental frequency (generally the power frequency, e.g. 50 Hz). These characteristics are: s the reactive power for compensation: Qvar. The resonant shunt, behaving capacitively below its tuning frequency, contributes to the compensation of reactive power at the power frequency. The reactive power produced by the shunt at the connection busbars, for an operating voltage U1, is given by the following equation:
characteristic impedance:
X0 =
s
L ; C
resonant shunt filters

The resonant shunt filter (see fig. 16) is made up of an LC branch with a frequency of fr = 1 2 L C
the quality factor: q = X0/r. An effective filter must have a reactor with a large quality factor q, therefore: r << X0 at frequency fr. Approximate values of q: 75 for air-cored reactors, greater than 75 for iron-core reactors. s the pass-band (see fig. 20), in relative terms:
PB =
s
tuned to the frequency of the voltage harmonic to be eliminated. This approach is therefore fundamentally different than that of reactor-connected capacitors already described. At fr, the resonant shunt presents a low minimum impedance with respect to the resistance r of the reactor. It therefore absorbs nearly all the harmonic currents of frequency fr injected, with low harmonic voltage distortion (since proportional to the product of the resistance r and the current flowing in the filter) at this frequency. In principle, a resonant shunt is installed for each harmonic to be limited. They are connected to the busbars for which harmonic voltage reduction is specified. Together they form a filter bank. Figure 19 shows the harmonic impedance of a network equipped with a set of four filters tuned to the 5th, 7th, 11th and 13th harmonics. Note that
f fr 1 = 2 fr q
r ; X0
Q var =
2 nr 2 n1 n
2 U1 C 2 f1
the resistance of the reactor: r = X0/q. This resistance is defined at frequency fr. It depends on the skin effect. It is also the impedance when the resonant shunt is tuned; s the losses due to the capacitive current at the fundamental frequency:
p1 = Q var q nr
(note that the subscript 1 refers to the fundamental). C is the phase-to-neutral capacitance of one of the 3 branches of the filter bank represented as a star. At first glance, the presence of a reactor would not be expected to increase the reactive power supplied. The reason is the increase in voltage at power frequency f1 caused by the inductance at the capacitor terminals;
IZI
IZI
r
r f fr f (Hz)
11
13
f/f1
fig. 19: impedance of a network equipped with shunt filters.
fig. 20: Z versus f curve for a resonant shunt.
with: Qvar = reactive power for compensation produced by the filter, p1 = filter losses at power frequency in W; s the losses due to the harmonic currents cannot be expressed by simple equations; they are greater than:
pn =
2 Unr
terminals. Figure 21 shows one of the three phases of the filter. The 2nd order damped filter has zero reactance for a frequency fr higher than the frequency f where: f = 2 1 and L C
fr =
1+ Q q 2 q (Q 2 1) L C
in which Unr is the phase-to-phase harmonic voltage of order nr on the busbars after filtering. In practice, the performance of resonant shunt filters is reduced by mis-tuning and special solutions are required as follows: s adjustment possibilities on the reactors for correction of manufacturing tolerances; s a suitable compromise between the q factor and filter performance to reduce the sensitivity to mis-tuning, thereby accepting fluctuations of f1 (network frequency) and fr (caused by the temperature dependence of the capacitance of the capacitors).
damped filters
2nd order damped filter On arc furnaces, the resonant shunt must be damped. This is because the continuous spectrum of an arc furnace increases the probability of an injected current with a frequency equal to the anti-resonance frequency. In this case, it is no longer sufficient to reduce the characteristic harmonic voltages. The anti-resonance must also be diminished by damping. Moreover, the installation of a large number of resonant shunts is often costly, and it is therefore better to use a wide-band filter possessing the following properties: s anti-resonance damping; s reduced harmonic voltages for frequencies greater than or equal to its tuning frequency, leading to the name damped high-pass filter; s fast damping of transients produced when the filter is energised. The 2nd order damped filter is made up of a resonant shunt with a damping resistor R added at the reactor
The filter is designed so that fr coincides with the first characteristic line of the spectrum to be filtered. This line is generally the largest. When Q (or R) take on high values, fr tends towards f, which means that the resonant shunt is a limiting case of the 2nd order damped filter. It is important not to confuse Q, the quality factor, with Qvar, the reactive power of the filter for compensation. The 2nd order damped filter operates as follows: s below fr: the damping resistor contributes to the reduction of the network impedance at anti-resonance, thereby reducing any harmonic voltages; s at fr: the reduction of the harmonic voltage to a specified value is possible since, at this frequency, no resonance can occur between the network and the filter, the latter presenting an impedance of a purely resistive character. However, this impedance being higher than the resistance r of the reactor, the
filtering performance is less than for a resonant shunt; s above fr: the filter presents an inductive reactance of the same type as the network (inductive), which lets it adsorb, to a certain extent, the spectrum lines greater than fr, and in particular any continuous spectrum that may be present. However, antiresonance, if present in the impedance of the network without the filter, due to the existing capacitor banks, reduces the filtering performance. For this reason, existing capacitor banks must be taken into account in the design of the network and, in some cases, must be adapted. The main electrical characteristics of a 2nd order damped filter depend on n r = fr/f1 , the order of the filter tuning frequency, with: s fr = tuning frequency; s f1 = fundamental frequency (generally the power frequency, e.g. 50 Hz). These characteristics are: s the reactive power for compensation: For a 2nd order damped filter at operating voltage U1 (the subscript 1 referring to the fundamental), the reactive power is roughly the same as for a resonant shunt with the same inductance and capacitance, i.e. in practice:
Q var =
2 nr 2 nr 1 2 U1 C 2 f1
C is the phase-to-neutral capacitance of one of the 3 branches of the filter bank represented as a star.
phase X r inductive R capacitive fr 0 f (Hz) L
C neutral
f =
1 2 L C
fig. 21: 2nd order damped filter.
X0 =
s
L ; C
IZI r with resonant shunt Z network L
phase
the quality factor of the reactor: q = X0/r where r is the resistance of the reactor, dependent on the skin effect and defined at frequency fr; s the quality factor of the filter: Q = R/X0. The quality factors Q used are generally between 2 and 10; s the losses due to the fundamental compensation current and to the harmonic currents; these are higher than for a resonant shunt and can only be determined through network analysis. The damped filter is used alone or in a bank including two filters. It may also be used together with a resonant shunt, with the resonant shunt tuned to the lowest lines of the spectrum. Figure 22 compares the impedance of a network with a 2nd order damped filter to that of a network with a resonant shunt. Other types of damped filters Although more rarely used, other damped filters have been derived from the 2nd order filter: s 3rd order damped filter (see fig. 23) Of a more complex design than the 2nd order filter, the 3rd order filter is intended particularly for high compensation powers. The 3rd order filter is derived from a 2nd order filter by adding another capacitor bank C2 in series with the resistor R, thereby reducing the losses due to the fundamental. C2 can be chosen to improve the behaviour of the filter below the tuning frequency as well, which favours the reduction of anti-resonance. The 3rd order filter should be tuned to the lowest frequencies of the spectrum. Given the complexity of the 3rd order filter, and the resulting high cost, a 2nd order filter is often preferred for industrial applications; s type C damped filter (see fig. 24) In this filter, the additional capacitor bank C2 is connected in series with the reactor. This filter offers characteristics roughly the same as those of the 3rd order filter;
with 2nd order damped filter
f (Hz)
C neutral
fig. 22: the impedance, seen from point A, of a network equipped with either a 2nd order damped filter or a resonant shunt.
phase
phase
r R
R
L C2
L C2
C neutral
C neutral
fig. 23: 3rd order damped filter.

s damped double filter (see fig. 25) Made up of two resonant shunts connected by a resistor R, this filter is specially suited to the damping of the anti-resonance between the two tuning frequencies;
fig. 24: type C damped filter.

phase
ra
rb
low q resonant shunt This filter, which behaves like a damped wide-band filter, is designed especially for very small installations not requiring reactive power compensation. The reactor, with a very high resistance (often due to the addition of a series resistor) results in losses which are prohibitive for industrial applications.
s
La R Ca neutral
Lb
Cb neutral
fig. 25: damped double filter.
9. measurement relays required for the protection of reactor-connected capacitors and filters
An anti-harmonic reactor must withstand the 3-phase short-circuit current at the common reactorcapacitor terminals. Furthermore, both anti-harmonic reactors and filters must continuously withstand fundamental and harmonic currents, fundamental and harmonic voltages, switching surges and dielectric stresses. In this chapter, anti-harmonic reactorconnected capacitor assemblies and filters will be referred to collectively as devices.
basic protection against device failures

All the elements of these devices can be subject to insulation faults and shortcircuits, while the capacitor banks are mainly the source of unbalance faults caused by the failure of capacitor elements. s protection of these devices against insulation faults can be provided by residual current relays (or zero phase sequence relays). Note: the neutral is generally not distributed on such devices; for higher sensing accuracy, it is better to use a toroidal type transformer, encircling all the live conductors of the feeder, rather than three step-down current transformers; s protection against short-circuits can be provided by overcurrent relays installed on the filter feeder. This protection must detect 2-phase short-circuits at the common reactorcapacitor terminals, while letting through inrush transients; s detection of unbalance currents in the connections between the neutrals of
the double star connected capacitor banks (see fig. 26). In addition to the damage that can be caused by the resulting unbalanced stresses, the failure of a small number of capacitor elements is detrimental to filter performance. This protection is therefore designed to detect, depending on its sensitivity, the failure of a small number of capacitor elements. Of the single-pole type, this protection must be: insensitive to the harmonics, set to above the natural unbalance current of the double star connected capacitor bank (this unbalance depends on the accuracy of the capacitors), set to below the unbalance current due to the failure of a single capacitor element,
operate on a major fault causing an unbalance. The fluctuation of the supply voltage must be taken into account in the calculation of all these currents.
basic protection against abnormal stresses on the devices

These abnormal stresses are essentially due to overloads. To protect against them, it is necessary to monitor the rms value of the distorted current (fundamental and harmonics) flowing in the filter. It is also necessary to monitor the fundamental voltage of the power supply using an overvoltage relay.
phase 1
phase 2
phase 3
current relay
C/2
C/2
C/2
C/2
C/2
C/2
fig. 26: unbalance detection for a double star connected capacitor bank.
10. example of the analysis of a simplified network
The diagram in figure 27 represents a simplified network comprising a 2,000 kVA six-pulse rectifier, injecting a harmonic current spectrum, and the following equipment which will be considered consecutively in three different calculations: s a single 1,000 kvar capacitor bank; s anti-harmonic reactor-connected capacitor equipment rated 1000 kvar; s a set of two filters comprising a resonant shunt tuned to the 5th harmonic and a 2nd order damped filter tuned to the 7th harmonic. The capacitor bank implemented in this manner is rated 1,000 kvar. Note that: s the 1,000 kvar compensation power is required to bring the power factor to a conventional value; s the harmonic voltages already present on the 20 kV distribution network have been neglected for the sake of simplicity. This example will be used to compare the performance of the three solutions, however the results can obviously not be applied directly to other cases.
reactor-connected capacitor bank

This equipment is arbitrarily tuned to 4.8 f1. Harmonic impedance (see fig. 31)
The network harmonic impedance curve, seen from the node where the harmonic currents are injected, exhibits a maximum of 16 ohms (antiresonance) in the vicinity of harmonic order 4.25. This unfortunately favours the presence of 4th voltage
network 20 kV Isc 12.5 kA 20/5.5 kV 5,000 kVA Usc 7.5 % Pcu 40 kW
2,000 kVA disturbing equipment
5.5/0.4 kV 1,000 kVA Usc 5 % Pcu 12 kW motor load 500 kVA at cos = 0.9 560 kW
capa.
reactor + capa.
resonant shunt and 2nd order damped filter
capacitor bank alone

The network harmonic impedance curve (see fig. 28), seen from the node where the harmonic currents are injected, exhibits a maximum (antiresonance) in the vicinity of the 7th current harmonic. This results in an unacceptable individual harmonic voltage distortion of 11 % for the 7th harmonic (see fig. 29). The following characteristics are also unacceptable: s a total harmonic voltage distortion of 12.8 % for the 5.5 kV network, compared to the maximum permissible value of 5 % (without considering the requirements of special equipment); s a total capacitor load of 1.34 times the rms current rating, exceeding the permissible maximum of 1.3 (see fig. 30). The solution with capacitors alone is therefore unacceptable.
fig. 27: installation with disturbing equipment, capacitors and filters.
Z () 38.2
V (V)
350 11 %
7.75
9 10 11 13 H
fig. 28: harmonic impedance seen from the node where the harmonic currents are injected in a network equipped with a capacitor bank alone.
fig. 29: harmonic voltage spectrum of a 5.5 kV network equipped with a capacitor bank alone.
harmonic. However, the low impedance, of an inductive character, of the 5th harmonic favours the filtering of the 5th harmonic quantities. Voltage distortion (see fig. 32) For the 5.5 kV network, the individual harmonic voltage ratios of 1.58 % (7th harmonic), 1.5 % (11th harmonic) and 1.4 % (13th harmonic) may be too high for certain loads. However in many cases the total harmonic voltage distortion of 2.63 % is acceptable.
For the 20 kV network, the total harmonic distortion is only 0.35 %, an acceptable value for the distribution utility. Capacitor current load (see fig. 33) The total rms current load of the capacitors, including the harmonic currents, is 1.06 times the current rating, i.e. less than the maximum of 1.3. This is the major advantage of reactorconnected capacitors compared to the first solution (capacitors alone).
resonant shunt filter tuned to the 5th harmonic and a damped filter tuned to the 7th harmonic
In this example, the distribution of the reactive power between the two filters is such that the filtered 5th and 7th voltage harmonics have roughly the same value. In reality, this is not required. Harmonic impedance (see fig. 34) The network harmonic impedance curve, seen from the node where the harmonic currents are injected, exhibits a maximum of 9.5 ohms (antiresonance) in the vicinity of harmonic 4.7. For the 5th harmonic, this impedance is reduced to the reactor resistance, favouring the filtering of the 5th harmonic quantities. For the 7th harmonic, the low, purely resistive impedance of the damped filter also reduces the individual harmonic voltage. For harmonics higher than the tuning frequency, the damped filter impedance curve reduces the corresponding harmonic voltages. This equipment therefore offers an improvement over the second solution (reactor-connected capacitors).
I (A)
V (V) 50 1.55 %
- 82
19 0.6 %
48 1.5 % 45 1.4 %
9 10 11 13
11 13 H
fig. 30: spectrum of the harmonic currents flowing in the capacitors for a network equipped with a capacitor bank alone.
fig. 32: harmonic voltage spectrum of a 5.5 kV network equipped with reactorconnected capacitors.
Z ()
I (A)
15.6
Z ()
34 24 %
9.5
4.7
~ 4.25 4.8 H
11 13
fig. 31: harmonic impedance seen from the node where the harmonic currents are injected in a network equipped with reactorconnected capacitors.
fig. 33: spectrum of the harmonic currents flowing in the capacitors for a network equipped with reactor-connected capacitors.
fig. 34: harmonic impedance seen from the node where the harmonic currents are injected in a network equipped with a resonant shunt filter tuned to the 5th harmonic and a damped filter tuned to the 7th harmonic.
Voltage distortion (see fig. 35) For the 5.5 kV network, the individual harmonic voltage ratios of 0.96 %, 0.92 %, 1.05 % and 1 % for the 5th, 7th, 11th and 13th harmonics respectively are acceptable for most sensitive loads. The total harmonic voltage distortion is 1.96 %. For the 20 kV network, the total harmonic distortion is only 0.26 %, an acceptable value for the distribution utility. Capacitor current load The total rms current load of the resonant shunt filter capacitors (see fig. 36) is greater than 1.3 time the current rating. The capacitance must therefore be increased, which will improve the filtering performance,
reducing the 5th harmonic ratio to less than 1 %. The result is of course an increase in the reactive power compensation capacity. To avoid overcompensating, a compromise must be found for the size of these capacitors. The calculation is therefore repeated with this new data. For the damped filter tuned to the 7th harmonic, the total rms current load of the capacitors (see fig. 37) is within the tolerance of 1.3 times their current rating. This example demonstrates an initial approach to the problem. However in practice, over and above the calculations relative to the circuit elements (L, r, C and R), other calculations are required before
proceeding with the implementation of any solution: s the spectra of the currents flowing in the reactors connected to the capacitors; s the total voltage distortion at the capacitor terminals; s reactor manufacturing tolerances and means for adjustment if necessary; s the spectra of the currents flowing in the resistors of the damped filters and their total rms value; s voltage and energy transients affecting the filter elements during energisation. These more difficult calculations, requiring a solid understanding of both the network and the equipment, are used to determine all the electrotechnical information required for the filter manufacturing specifications.
V (V)
I (A)
I (A)
0.96 % 0.91 %
1.05 % 1%
39
22 23 % 10 10 %
11 13
11 13
fig. 35: harmonic voltage spectrum of a 5.5 kV network equipped with a resonant shunt filter tuned to the 5th harmonic and a damped filter tuned to the 7th harmonic.
fig. 36: spectrum of the harmonic currents flowing in the capacitors of a resonant shunt filter tuned to the 5th harmonic on a network equipped with a damped filter tuned to the 7th harmonic.
fig. 37: spectrum of the harmonic currents flowing in the capacitors of a damped filter tuned to the 7th harmonic on a network equipped with a resonant shunt filter tuned to the 5th harmonic.
11. conclusion
Static power converters are increasingly used in industrial distribution. The same is true for arc furnaces in the growing electricpowered steel industry. All these loads produce harmonic disturbances and require compensation of the reactive power they consume, leading to the installation of capacitor banks. Unfortunately these capacitors, in conjunction with the inductances in the network, can cause high frequency
oscillations that amplify harmonic disturbances. Installers and operators of industrial networks are thus often confronted with a complex electrical problem. The main types of harmonic disturbances and the technical means available to limit their extent have been presented in this document. Without offering an exhaustive study of the phenomena involved or relating all
acquired experience, this document should provide the necessary background to, if not solve the problems, at least facilitate discussions with specialists. For further information or assistance, feel free to contact the Network Studies department of the Central R&D organisation of Merlin Gerin, a group of specialised engineers with more than twenty years of experience in this field.
12. bibliography
Standards s IEC 146: Semi-conductor converters. s IEC 287: Calculation of the continuous current rating of cables. s IEC 555-1: Disturbances in supply systems caused by household appliances and similar electrical equipment - Definitions. s IEC 871-1 and HD 525.1-S-T: Shunt capacitors for AC power systems having a rated voltage above 660 V. s NF C 54-100. s HN 53 R01 (May 1981): EDF general orientation report. Particular aspects concerning the supply of electrical power to sensitive electronic equipment and computers. Merlin Gerin's Cahier Technique s Residual current devices Cahier Technique n 114 R. CALVAS s Les perturbations lectriques en BT Cahier Technique n 141 R. CALVAS Other publications s Direct current transmission, volume 1 E. W. KIMBARK published by: J. WILEY and SONS. Le cyclo-convertisseur et ses influences sur les rseaux d'alimentation (The cyclo-converter and its effects on power supply networks T. SALZAM and W. SCHULTZ - AIM Lige CIRED 75.
s
s Perturbations des rseaux industriels et de distribution. Compensation par procds statistiques. Rsonances en prsence des harmoniques crs par les convertisseurs de puissance et les fours arc associs des dispositifs de compensation. (Disturbances on industrial and distribution networks. Compensation by statistical processes. Resonance in the presence of harmonics created by power converters and arc furnaces associated with compensation equipment.) Michel LEMOINE - DER EDF RGE T 87 n 12 12/78. s Perturbations des rseaux industriels et de distribution. Compensation par procds statistiques. Perturbations de tension affectant le fonctionnement des rseaux fluctuations brusques, flicker, dsquilibres et harmoniques. (Disturbances on industrial and distribution networks. Compensation by statistical processes. Voltage disturbances affecting network operation - fluctuations, flicker, unbalances and harmonics. M. CHANAS - SER-DER EDF RGE T 87 n 12 12/78. s Pollution de la tension (Voltage disturbances). P. MEYNAUD - SER-DER EDF RGE T 89 n 9 09/80. s Harmonics, characteristic parameters, methods of study, estimates of existing values in the network. (ELECTRA) CIGRE 07/81. s Courants harmoniques dans les redresseurs triphass commutation force. (Harmonic currents in forced commutation 3-phase rectifiers) W. WARBOWSKI CIRED 81.
Perturbations rciproques des quipements lectroniques de puissance et des rseaux - Quelques aspects de la pollution des rseaux par les distorsions harmoniques de la clientle (Mutual disturbances between power electronics equipment and networks - Several aspects concerning network pollution by harmonic distortion produced by subscribers). Michel LEMOINE - DER EDF RGE T 85 n 3 03/76.
s
Origine et nature des perturbations dans les rseaux industriels et de distribution. (Origin and nature of disturbances in industrial and distribution networks). Guy BONNARD - SER-DER-EDF RGE 1/82. s Problmes particuliers poss par ltude du phnomne de distorsion harmonique dans les rseaux. (Particular problems posed by the study of harmonic distortion phenomena in networks). P. REYMOND CIGRE Study Committee 36 09/82. s Rduction des perturbations lectriques sur le rseau avec le four arc en courant continu (Reduction of electrical network disturbances by DC arc furnaces). G. MAURET, J. DAVENE IRSID SEE LYON 05/83. s Line harmonics of converters with DC motor loads. A. DAVID GRAHAM and EMIL T. SCHONHOLZER. IEEE transactions on industry applications. Volume IA 19 n 1 02/83. s Filtrage dharmoniques et compensation de puissance ractive Optimisation des installations de compensation en prsence d'harmoniques. (Harmonic filtering and reactive power compensation - Optimising compensation installations in the presence of harmonics). P. SGARZI and S. THEOLERE, SEE Seminar RGE n 6 06/88.
s
Ral. : Illustration Technique Lyon -
DTE - 10/94 - 2500 - Imprimeur :
n 152
Pierre Roccia Obtained an Electrical Engineering degree from the INPG (National Polytechnic Institute of Grenoble) in 1969. Worked as project manager in the industrial equipment and high voltage public distribution sector, before being put in charge of extending the Merlin Gerin range of protection relays and developing a technical approach for the protection of high voltage industrial networks using devices associated with circuit breakers. After three years as a training instructor, he is presently working as an engineer in the "Network Studies" department of the Central R&D organisation. Nol Quillon After joining Merlin Gerin's Low Voltage Equipment Department in 1968, he subsequently took part in the development of LV circuit breakers within the testing laboratory. A graduate engineer from the INPG, he worked in the "Network Studied" department of the Central R&D organisation for eight years where he studied electrical network phenomena and their behaviour in order to establish guidelines to control these phenomena. In 1985, he joined the Training Department. After being in charge of the electrotechnical training programme, he is presently the training correspondent for the UPS division.
E/CT 152 first issued october 1994
glossary
Symbols: C D
f1 far fn fr
n In
j L Lsc n nar nr k p p1 pn P (W) PB q Q Q (var) r R spectrum Ssc T U Vn X X0 Xsc Y0 Yn Z
capacitance or, more generally, the capacitors themselves harmonic distortion loss angle of a capacitor fundamental frequency anti-resonance frequency frequency of the nth harmonic component resonance frequency phase angle of the nth harmonic component when t = 0 rms current of the nth harmonic component complex operator equal to the square root of 1 inductance or, more generally, the reactors, producing the inductance short-circuit inductance of a network, seen from a given point, as defined by Thevenin's theorem the order of a harmonic component (also referred to as the harmonic number) the order of anti-resonance, i.e. the radio of the anti-resonance frequency to the fundamental frequency the order of resonance, i.e. the radio of the resonance frequency to the fundamental frequency a positive integer number of rectifier arms (also referred to as the pulse number) filter losses due only to the fundamental current filter losses due only to the nth harmonic current active power pass-band of a resonant shunt filter quality factor of a reactor quality factor of a filter reactive power resistance resistance (or the real part of the impedance) the distribution, at a given point, of the amplitudes of the various harmonic components expressed relative to the fundamental short-circuit power of a network at a given point period of an alternating quantity phase-to-phase rms voltage rms voltage of the nth harmonic component reactance characteristic inductance or impedance of a filter short-circuit reactance of a network, seen from a given point, as defined by Thevenin's theorem amplitude of the DC component rms value of the nth harmonic component impedance
Abbreviations: CIGRE Confrence Internationale des Grands Rseaux Electriques (International Conference on Large Electrical Networks) IEC International Electrotechnical Commission
summary
1. Introduction: harmonic distortion is a problem that must often be p. 4 dealt with in industrial power distribution networks 2. Harmonic quantities p. 4 3. Principal disturbances caused by Instantaneous effects p. 6 harmonic currents and voltages Long-term effects p. 6 4. Acceptable limits, recommendations Typical limits for distribution p. 7 and standards networks Typical limits for industrial p. 7 networks 5. Harmonics generators Static converters on 3-phase p. 8 networks Arc furnaces p. 8 Lighting p. 9 Saturated reactors p. 9 Rotating machines p. 9 Calculation model p. 9 6. Can capacitors cause a problem on In the absence of capacitor banks p. 10 networks comprising disturbing In the presence of a capacitor p. 10 equipment bank 7. Anti-harmonic reactors p. 13 8. Filters Resonant shunt filters p. 14 Damped filters p. 15 9. Measurement relays required for the Basic protection against device p. 17 protection of reactor-connected failures capacitors and filters Basic protection against abnormal p. 17 stresses on the devices 10. Example of the analysis of a Capacitor bank alone p. 18 simplified network Reactor-connected capacitor bank p. 18 Resonant shunt filter tuned to the p. 19 5th harmonic and a damped filter tuned to the 7th harmonic 11. Conclusion p. 21 12. Bibliography p. 22
1. introduction
harmonic distortion is a problem that must often be dealt with in industrial power distribution networks
Electricity is generally distributed as three voltage waves forming a 3-phase
sinusoidal system. One of the characteristics of such a system is its waveform, which must always remain as close as possible to that of a pure sine wave. If distorted beyond certain limits, as is often the case on networks comprising sources of harmonic currents and
voltages such as arc furnaces, static power converters, lighting systems, etc., the waveform must be corrected. The aim of the present document is to provide a better understanding of these harmonics problems, including their causes and the most commonly used solutions.
2. harmonic quantities
To help the reader follow the discussion, we will first review the definitions of a number of terms related to harmonics phenomena. Readers already familiar with the basic terminology may proceed directly to the next chapter. On AC industrial power supply networks, the variation of current and voltage with time is considerably different from that of a pure sine wave (see fig. 1). The actual waveform is composed of a number of sine waves of different frequencies, including one at the power frequency, referred to as the fundamental component or simply the fundamental. Harmonic component The term harmonic component, or simply harmonic, refers to any one of the above-mentioned sinusoidal components, the frequency of which is a multiple of that of the fundamental. The amplitude of a harmonic is generally a few percent of that of the fundamental. Harmonic order The harmonic order, also referred to as the harmonic number, is the ratio of the frequency fn of a harmonic to that of the fundamental (generally the power frequency, i.e. 50 or 60 Hz):
By definition, the harmonic order of the fundamental f1 is equal to 1. Note that the harmonic of order n is often referred to simply as the nth harmonic. Spectrum The spectrum is the distribution of the amplitudes of the various harmonics as a function of their harmonic number, often illustrated in the form of a histogram (see fig. 2).
Expression of the distorted wave Any periodic phenomenon can be represented by a Fourier series as follows:
y(t) = Y 0 + Y n 2 sin (nt n )

n = 1
n =
where: s Y0 = the amplitude of the DC component, which is generally zero in electrical power distribution;
I phase
fundamental distorted wave
harmonic
n =
fn . f1
fig.1: shape of a distorted wave.
s Yn = the rms value of the nth harmonic component, s n = phase angle of the nth harmonic component when t = 0. Harmonics with an order above 23 are often negligible.
true rms value, by thermal means or by spectrum analysers. Individual harmonic ratio and total harmonic distortion The industrial harmonic ratios and the total harmonic distortion quantify the harmonic disturbances present in a power supply network. s individual harmonic ratio (or harmonic percentage) The harmonic ratio expresses the magnitude of each harmonic with respect to the fundamental (see fig. 2). The nth harmonic ratio is the ratio of the rms value of the nth harmonic to that of the fundamental. For example, the harmonic ratio of In is In/I1 or 100 (In/I1) if expressed as a percentage (note that here In is not the nominal or rated current); s total harmonic distortion (also referred to as THD, the total harmonic factor or simply as distortion D). The total harmonic distortion quantifies the thermal effect of all the harmonics. It is the ratio of the rms value of all the harmonics to that of one of the two following quantities (depending on the definition adopted): the fundamental (CIGRE), which can give a very high value:
n =
the measured rms quantity (IEC 555-1), in which case 0 < D < 1:
n =
D =
n = 2 n = n = 1
2 Yn 2 Yn
Rms value of a distorted wave Harmonic quantities are generally expressed in terms of their rms value since the heating effect depends on this value of the distorted waveform. For a sinusoidal quantity, the rms value is the maximum value divided by the square root of 2. For a distorted quantity, under steadystate conditions, the energy dissipated by the Joule effect is the sum of the energies dissipated by each of the harmonic components:
2 2 R I 2 t = R I1 t + R I 2 t + ... + R I n t 2
Unless otherwise indicated, we will use the definition adopted by CIGRE (see the glossary) which corresponds to the ratio of the rms value of the harmonic content to the undistorted current at power frequency.
100 %
where:
2 2 I 2 = I1 + I 2 + ... + I n 2
i.e. where:
I =
n = n = 1
2 In
if the resistance can be considered to be constant. The rms value of a distorted waveform can be measured either directly by instruments designed to measure the
D =
n = 2
2 Yn
Y1
fig. 2: the amplitude of a harmonic is often expressed with respect to that of the fundamental.
3. principal disturbances caused by harmonic currents and voltages
Harmonic currents and voltages superimposed on the fundamental have combined effects on equipment and devices connected to the power supply network. The detrimental effects of these harmonics depend on the type of load encountered, and include: s instantaneous effects; s long-term effects due to heating.
running, the distance between the two circuits and the harmonic frequencies (coupling increases with frequency).
long-term effects
Over and above mechanical fatigue due to vibrations, the main long-term effect of harmonics is heating. Capacitor heating The losses causing heating are due to two phenomena: conduction and dielectricc hysteresis. As a first approximation, they are proportional to the square of the applied voltage for conduction and to the frequency for hysteresis. Capacitors are therefore sensitive to overloads, whether due to an excessively high fundamental or to the presence of voltage harmonics. These losses are defined by the loss angle of the capacitor, which is the angle whose tangent is the ratio of the losses to the reactive power produced (see fig. 3). Values of around 10-4 may be cited for tan . The heat produced can lead to dielectric breakdown. Heating due to additional losses in machines and transformers s additional losses in the stators (copper and iron) and principally in the rotors (damping windings, magnetic circuits) of machines caused by the considerable differences in speed between the harmonic inducing rotating fields and the rotor. Note that rotor measurements (temperature, induced currents) are difficult if not impossible. s supplementary losses in transformers due to the skin effect (increase in the resistance of copper with frequency), hysteresis and eddy currents (in the magnetic circuit). Heating of cables and equipment Losses are increased in cables carrying harmonic currents, resulting in temperature rise. The causes of the additional losses include: s an increase in the apparent resistance of the core with frequency, due to the skin effect;
instantaneous effects
Harmonic voltages can disturb controllers used in electronic systems. They can, for example, affect thyristor switching conditions by displacing the zero-crossing of the voltage wave (see IEC 146-2 and Merlin Gerin Cahier Technique n 141). Harmonics can cause additional errors in induction-disk electricity meters. For example, the error of a class 2 meter will be increased by 0.3 % by a 5th harmonic ratio of 5 % in current and voltage. Ripple control receivers, such as the relays used by electrical utilities for centralised remote control, can be disturbed by voltage harmonics with frequencies in the neighbourhood of the control frequency. Other sources of disturbances affecting these relays, related to the harmonic impedance of the network, will be discussed further on. Vibrations and noise The electrodynamic forces produced by the instantaneous currents associated with harmonic currents cause vibrations and acoustical noise, especially in electromagnetic devices (transformers, reactors, etc.). Pulsating mechanical torque, due to harmonic rotating fields, can produce vibrations in rotating machines. Interference on communication and control circuits (telephone, control and monitoring) Disturbances are observed when communication or control circuits are run along side power distribution circuits carrying distorted currents. Parameters that must be taken into account include the length of parallel
an increase in dielectric losses in the insulation with frequency, if the cable is subjected to non-negligible voltage distortion; s phenomena related to the proximity of conductors with respect to metal cladding and shielding earthed at both ends of the cable, etc. Calculations can be carried out as described in IEC 287. Generally speaking, all electrical equipment (electrical switchboards) subjected to voltage harmonics or through which harmonic currents flow, exhibit increased energy losses and should be derated if necessary. For example, a capacitor feeder cubicle should be designed for a current equal to 1.3 times the reactive compensation current. This safety factor does not however take into account the increased heating due to the skin effect in the conductors. Harmonic distortion of currents and voltages is measured using spectrum analysers, providing the amplitude of each component. The rms value of the distorted current (or voltage) may be assessed in any of three ways: s measurement using a device designed to give the true rms value, s reconstitution on the basis of the spectrum provided by spectral analysis, s estimation from an oscilloscope display.
s
tan =
p Q
fig. 3: triangle relating to the capacitor powers, (active (P), reactive (Q), apparent (R)).
4. acceptable limits, recommendations and standards
General limits s synchronous machines: permissible stator current distortion = 1.3 to 1.4 %; s asynchronous machines: permissible stator current distortion = 1.5 to 3.5 %; s cables: permissible core-shielding voltage distortion = 10 %; s power capacitors: current distortion = 83 %, corresponding to an overload of 30 % (1.3 times the rated current); overvoltages can reach up to 10 % (see IEC 871-1, 931-1 and HD 525.1S1); s sensitive electronics: 5 % voltage distortion with a maximum individual harmonic percentage of 3 % depending on the equipment.
typical limits for industrial networks

It is generally accepted that industrial network without any sensitive equipment such as regulators,
programmable controllers, etc. can accept up to 5 % voltage distortion. This limit and the limits for the individual harmonic ratios may be different if sensitive equipment is connected to the installation.
harmonic order 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23
low value (%) 1 1.5 0.5 5 0.2 4 < 0.2 0.8 < 0.2 2.5 < 0.2 2 < 0.2 < 0.3 < 0.2 1 < 0.2 0.8 < 0.2 < 0.2 < 0.2 0.5
high value (%) 1.5 2.5 1 6 0.5 5 1.5 3.5 3
typical limits for distribution networks

The French electrical utility, EDF, considers that voltage distortion will not exceed 5 % at the supply terminals as long as each individual subscriber does not exceed the following limits: s 1.6 % voltage distortion; s individual harmonic percentages of: 0.6 % for even voltage harmonics, 1 % for odd voltage harmonics. The table in figure 4 lists typical percentages observed for the various voltage harmonics where: s low value = value likely to be found in the vicinity of large disturbing loads and associated with a low probability of having disturbing effects; s high value = value rarely exceeded in the network, and with a higher probability of having disturbing effects.
2 1.5
fig. 4: individual voltage harmonic percentages measured in high voltage distribution networks.
5. harmonics generators
In industrial applications, the main types equipment that generate harmonics are: s static converters; s arc furnaces; s lighting; s saturated reactors; s other equipment, such as rotating machines which generate slot harmonics (often negligible).
static converters on 3-phase networks

Rectifier bridges and, more generally, static converters (made up of diodes and thyristors) generate harmonics. A Graetz bridge, for instance, requires a rectangular pulsed AC current (see fig. 5) to deliver a perfect DC current. In spite of their different waveforms, the currents upstream and downstream from the delta-star connected transformer have the same characteristic harmonic components. The characteristic harmonic components of the current pulses supplying rectifiers have the following harmonic numbers n, with n = kp 1, where: s k = 1, 2, 3, 4, 5... s p = number of rectifier arms, for example: Graetz bridge p = 6, 6-pulse bridge p = 6, 12-pulse bridge p = 12. Applying the formula, the p = 6 rectifiers cited above generate harmonics 5, 7, 11, 13, 17, 19, 23 and 25, and the p = 12 rectifiers generate harmonics 11, 13, 23 and 25. The characteristic harmonics are all odd-numbered and have, as a first approximation, an amplitude of In = I1/n where I1 is the amplitude of the fundamental. This means that I5 and I7 will have the greatest amplitudes. Note that they can be eliminated by using a 12-pulse bridge (p = 12). In practice, the current spectrum is slightly different. New even and odd harmonics, referred to as non-
characteristic harmonics, of low amplitudes, are created and the amplitudes of the characteristic harmonics are modified by several factors including: s asymmetry; s inaccuracy in thyristor opening times; s switching times; s imperfect filtering. For thyristor bridges, a displacement of the harmonics as a function of the thyristor phase angle may also be observed. Mixed thyristor-diode bridges generate even harmonics. They are used only at low ratings because the 2nd harmonic produces serious disturbances and is very difficult to eliminate. Other power converters such as cycloconverters, dimmers, etc. have richer and more variable spectra than rectifiers. Note that they are increasingly replaced by converters using the PWM (Pulse Width Modulation) technique. These devices
operate at high chopping frequencies (20 to 50 kHz) and are generally designed to generate only low levels of harmonics. The harmonic currents of several converters combine vectorially at the common supply busbars. Their phases are generally unknown except for the case of diode rectifiers. It is therefore possible to attenuate the 5 th and 7 th current harmonics using two equally loaded 6-pulse diode bridges, if the couplings of the two power supply transformers are carefully chosen (see fig. 6).
arc furnaces
Arc furnaces used in the steel industry may be of the AC (see fig. 7) or DC type. AC arc furnaces (see fig. 7) The arc is non-linear, asymmetric and unstable. It generates a spectrum
load
T/6
T/3
T/6 phase current upstream from a delta-star connected transformer supplying the rectifier
rectifier supply phase current
fig. 5: alternating current upstream from a Graetz bridge rectifier delivering a perfect direct current.
including odd and even harmonics as was well as a continuous component (background noise at all frequencies). The spectrum depends on the type of furnace, its power rating and the operation considered (e.g. melting, refining). Measurements are therefore required to determine the exact spectrum (see fig. 8). DC arc furnaces (see fig. 9) The arc is supplied via a rectifier and is more stable than the arc in AC furnaces. The current drawn can be broken down into: s a spectrum similar to that of a rectifier; s a continuous spectrum lower than that of an AC arc furnace.
To a large extent, the harmonic currents drawn by the disturbing equipment are independent of the other loads and the overall network impedance. These currents can therefore be considered to be injected into the network by the disturbing equipment. It is simply necessary to arbitrarily change the sign so that, for calculation purposes, the disturbing equipment can be considered as current sources (see fig. 10). The approximation is somewhat less accurate for arc furnaces. In this case, the current source model must be corrected by adding a carefully selected parallel impedance.
in %
In I1
100 10
100 continuous spectrum 4 3.2 1.3 0.5
1 0.1 1 3 5 7
9 rang
fig. 8: current spectrum for an arc furnace supplied by AC power.
lighting
Lighting systems made up of discharge lamps or fluorescent lamps are generators of harmonic currents. A 3rd harmonic ratio of 25 % is observed in certain cases. The neutral conductor then carries the sum of the 3rd harmonic currents of the three phases, and may consequently be subjected to dangerous overheating if not adequately sized.
I5 and I7 attenuated
HV transformer
I5 and I7
I5 and I7
cable
Dy 11
Yy 0
rectifier
saturated reactors
The impedance of a saturable reactor is varying with the current flowing through it, resulting in considerable current distortion. This is, for instance, the case for transformers at no load, subjected to a continuous overvoltage.
6-pulse diode bridge load equal loads load
6-pulse diode bridge
cable
furnace
rotating machines
Rotating machines generate high order slot harmonics, often of negligible amplitude. However small synchronous machines generate 3rd order voltage harmonics than can have the following detrimental effects: s continuous heating (without faults) of earthing resistors of generator neutrals; s malfunctioning of current relays designed to protect against insulation faults.
fig. 6: attenuation circuit for I5 and I7.

HV transformer
fig. 9: arc furnace supplied by DC power.
cable
calculation model
When calculating disturbances, static converters and arc furnaces are considered to be harmonic current generators.
furnace
fig. 7: arc furnace supplied by AC power.
fig. 10: harmonic current generators are modelled as current sources.
6. can capacitors cause a problem on networks comprising disturbing equipment?
We will consider the two following cases: s networks without power capacitors; s networks with power capacitors.
in the absence of capacitor banks, harmonic disturbances are limited and proportional to the currents of the disturbing equipment.
In principle, in so far as are concerned harmonics, the network remains inductive. Its reactance is proportional to the frequency and, as a first estimate, the effects of loads and resistance are negligible. The impedance of the network, seen from a network node, is therefore limited to the short-circuit reactance Xsc at the node considered. The level of harmonic voltages can be estimated from the power of the disturbing equipment and the shortcircuit power at the node (busbars) to which the disturbing equipment is connected, the short-circuit reactance considered to be proportional to the frequency (see fig. 11). In figure 11: Lsc = the short-circuit inductance of the network, seen from the busbars to which the disturbing equipment is connected, In = currents of the disturbing equipment, Xsc n = Lsc n = Lsc n (2 f1) therefore V n = Xsc n I n = Lsc n (2 f1) I n. The harmonic disturbances generally remain acceptable as long as the disturbing equipment does not exceed a certain power level. However, this must be considered with caution as resonance (see the next section) may be present, caused by a nearby network possessing capacitors and coupled via a transformer.
Note: In reality, the harmonic inductance of network X, without capacitors (essentially a distribution network), represented by Lsc, can only be considered to be proportional to the frequency in a rough approximation. For this reason, the network shortcircuit impedance is generally multiplied by a factor of 2 or 3 for the calculations. Therefore: Xn = k n X1 with k = 2 or 3. The harmonic impedance of a network is made up of different constituents such as the short circuit impedance of the distribution system as well as the impedance of the cables, lines, transformers, distant capacitors, machines and other loads (lighting, heating, etc.).
In
Xsc
Vn
fig. 11: the harmonic voltage Vn is proportional to the current In injected by the disturbing equipment.
node A (busbars)
in the presence of a capacitor bank parallel resonance can result in dangerous harmonic disturbances
Resonance exists between the capacitor bank and the reactance of the network seen from the bank terminals. The result is the amplification, with a varying degree of damping, of the harmonic currents and voltages if the order of the resonance is the same as that of one of the harmonic currents injected by the disturbing equipment. This amplified disturbance can be dangerous to the equipment. This is a serious problem and will be dealt with in below. This phenomenon is referred to as parallel resonance. What is this parallel resonance and how can it cause dangerous harmonic disturbances? In so far as harmonic frequencies are concerned, and for a first approximation, the network may be represented as in figure 12.
Lsc
load
Vn
In
0 a: harmonic electrical representation of a phase.
50 Hz source
Lsc node A (busbar)
load
disturbing equipment
b: single-line diagram.
fig. 12: equivalent diagrams for a circuit subject to harmonic currents and including a capacitor bank.
In this diagram: Lsc = the short-circuit inductance of the network seen from the busbars to which the capacitor bank and the disturbing equipment are connected, C = capacitors, In = currents of the disturbing equipment, load = loads (Joule effect, transmission of mechanical energy). In principle, we consider the shortcircuit harmonic reactance seen from the busbars, i.e. the node (A) to which the capacitors, the loads and the disturbing equipment are connected, giving Vn = ZAO In. The impedance versus frequency curves (see fig.13) show that: s for the resonance frequency far, the inductive effect is compensated for exactly by the capacitive effect; s the reactance of the rejecter circuit: is inductive for low frequencies, including the fundamental frequency, increases with frequency, becoming very high and suddenly capacitive at the resonance frequency far; s the maximum impedance value reached is roughly R = U2/P where P represents the sum of the active power values of the loaded motors, other than those supplied by a static converter. If a harmonic current In of order n , with the same frequency as the parallel resonance frequency far, is injected by the disturbing equipment, the corresponding harmonic voltage can be estimated as Vn = R In with n = n ar = f ar/f1. Estimation of nar The order nar of parallel resonance is the ratio of the resonance frequency far to the fundamental frequency f1 (power frequency). Consider the most elementary industrial network, shown in the equivalent diagram in figure 14, including a capacitor bank C supplied by a transformer with a short-circuit inductance LT, where Lsc represents the short-circuit inductance of the distribution network seen from the upstream terminals of the transformer,
f ar = 2 1 . (Lsc + L T ) C
impedance is seen from point A or point B (e.g. the supply terminals). In general, given the short-circuit power at the capacitor bank terminals,
nar = Ssc Q
and undoubtedly present a danger to the capacitors. s if the parallel resonance order corresponds to the frequency of the carrier-current control equipment of the power distribution utility, there is a risk of disturbing this equipment. To prevent resonance from becoming dangerous, it must be forced outside the injected spectrum and/or damped. The short-circuit impedance of the network is seldom accurately known and, in addition, it can vary to a large extent, thereby resulting in large variations of the parallel resonance frequency. It is therefore necessary to stabilise this frequency at a value that does not correspond to the frequencies of the injected harmonic currents. This is achieved by connecting a reactor in series with the capacitor bank.
where: Ssc = short-circuit power at the capacitor bank terminals, Q = capacitor bank power at the applied voltage. Generally S is expressed in MVA and Q in Mvar. Practical consequences: s if the order of a harmonic current injected by disturbing equipment corresponds to the parallel resonance order, there is a risk of harmonic overvoltages, especially when the network is operating at low loads. The harmonic currents then become intensively high in network constituents
X inductive 0 f (Hz) capacitive far without capacitors X = Lsc 2 f
IZI
~R without capacitors IZI = Lsc 2 f
0 far
f (Hz)
fig. 13: curves showing the impedance due to the loads and due to the resistance of the conductors.
LT
lopp
distributor Lsc C load
The order of the parallel resonance is roughly the same whether the network
fig. 14: the capacitor, together with the sum of the upstream impedances, forms a resonant circuit.
The rejecter circuit thus created is then represented by the diagram in figure 15 where Vn = ZAO In. A series resonance, between L and C, appears. As opposed to this resonance, which gives a minimum impedance, the parallel resonance is often referred to an anti-resonance. The equation giving the frequency of the anti-resonance is:
f ar = 2 1 (Lsc + L) C
s a minimum resistive value r (resistance of the inductance coil) for the resonance frequency fr; s a capacitive reactance below the resonance frequency fr; s an inductive reactance above the resonance frequency fr, where
fr =
1 2 L C
Lsc generally being small compared to L, the equation shows that the presence of reactor L, connected in series with the capacitors, renders the frequency far less sensitive to the variations of the short-circuit inductance Lsc (from the connections points = busbars A). Series resonance The branch made up of reactor L and capacitor C (see fig. 16), form a series resonance system of impedance Z = r + j(L - 1/C) with:
The curves in figure 17 show the shape of the network inductance, including the short-circuit impedance and that of the LC branch, seen from busbars A. The choice of far depends on Lsc, L and C, while that of fr depends only on L and C; far and fr therefore become closer as Lsc becomes small with
respect to L. The level of reactive power compensation, and the voltage applied to the capacitors, depend partly on L and C. The reactor L can be added in two different manners, depending on the position of the series resonance with respect to the spectrum. The two forms of equipment are: s anti-harmonic reactors (for series resonance outside the spectrum lines); s filters (for series resonance on a spectrum line).
X
inductive
ph1 inductive r capacitive 0
X
capacitive
0 f (Hz) fr far
f (Hz)
busbar node, point A
fr L IZI
IZI
L Lsc C 0 Vn
In
C
0
neutral
r f (Hz) fr
far
~r
f (Hz) fr
fig. 15: the reactor, connected in series with the capacitor, forms a rejecter circuit.
fig. 16: impedance of the rejecter circuit.
fig. 17: network impedance at point A.
7. anti-harmonic reactors
An anti-harmonic reactor can be used to protect a capacitor bank against harmonic overloads. Such solutions are often referred to as reactor-connected capacitor installations. The reference diagram is once again figure 15. In this assembly, the choice of L is such that the LC branch (where L is the reactor and C the reactive power compensation capacitors) behaves inductively for the harmonic frequencies, over the spectrum. As a result, the resonance frequency fr of this branch will be below the spectrum of the disturbing equipment. The LC branch and the network (Lsc) are then both inductive over the spectrum and the harmonic currents injected by the disturbing equipment are divided in a manner inversely proportional to the impedance. Harmonic currents are therefore greatly restricted in the LC branch, protecting the capacitors, and the major part of the harmonic currents flow in the rest of the network, especially in the shortcircuit impedance. The shape of the network impedance, seen from the busbars to which the LC branch is connected, is shown in figure 18. There is no anti-resonance inside the current spectrum. The use of an antiharmonic reactor therefore offers two advantages;
s it eliminates the danger of high harmonic currents in the capacitors; s it correlatively eliminates the high distortions of the network voltage, without however lowering them to a specified low value. Certain precautions are necessary: s no other capacitor banks must be present that could induce, through antiresonance, a capacitive behaviour in the initial network inside the spectrum; s care must be taken not to introduce an anti-resonance with a frequency used by the distribution utility for
carrier-current control, since this would place an increased load on the high frequency generators (175 Hz, 188 Hz). The anti-harmonic reactor is tuned to an order of 4.5 to 4.8, giving a value of fr between 225 to 240 Hz for a 50 Hz network, which is very near the ripple control frequency used on many distribution networks; s due to the continuous spectrum, the use of anti-harmonic reactors on arc furnaces requires certain precautions which can only be defined after carrying out special studies.
IzI
theoretical impedance without the LC branch
f (Hz) f1 fr harmonic current spectrum
far
fig. 18: the capacitors are protected when fr is well below the harmonic spectrum.
8. filters
Filters are used when it is necessary to limit harmonic voltages present on a network to a specified low value. Two types of filters may be used to reduce harmonic voltages: s resonant shunt filters, s damped filters.
there are as many anti-resonances as there are filters. These anti-resonances must be tuned to frequencies between the spectrum lines. A careful study must therefore be carried out if it is judged necessary to segment the filter bank. Main characteristics of a resonant shunt The characteristics depend on n r = fr/f1 the order of the filter tuning frequency, with: s fr = tuning frequency; s f1 = fundamental frequency (generally the power frequency, e.g. 50 Hz). These characteristics are: s the reactive power for compensation: Qvar. The resonant shunt, behaving capacitively below its tuning frequency, contributes to the compensation of reactive power at the power frequency. The reactive power produced by the shunt at the connection busbars, for an operating voltage U1, is given by the following equation:
X0 =
s
L ; C
resonant shunt filters

The resonant shunt filter (see fig. 16) is made up of an LC branch with a frequency of fr = 1 2 L C
the quality factor: q = X0/r. An effective filter must have a reactor with a large quality factor q, therefore: r << X0 at frequency fr. Approximate values of q: 75 for air-cored reactors, greater than 75 for iron-core reactors. s the pass-band (see fig. 20), in relative terms:
PB =
s
tuned to the frequency of the voltage harmonic to be eliminated. This approach is therefore fundamentally different than that of reactor-connected capacitors already described. At fr, the resonant shunt presents a low minimum impedance with respect to the resistance r of the reactor. It therefore absorbs nearly all the harmonic currents of frequency fr injected, with low harmonic voltage distortion (since proportional to the product of the resistance r and the current flowing in the filter) at this frequency. In principle, a resonant shunt is installed for each harmonic to be limited. They are connected to the busbars for which harmonic voltage reduction is specified. Together they form a filter bank. Figure 19 shows the harmonic impedance of a network equipped with a set of four filters tuned to the 5th, 7th, 11th and 13th harmonics. Note that
f fr 1 = 2 fr q
r ; X0
Q var =
2 nr 2 n1 n
2 U1 C 2 f1
the resistance of the reactor: r = X0/q. This resistance is defined at frequency fr. It depends on the skin effect. It is also the impedance when the resonant shunt is tuned; s the losses due to the capacitive current at the fundamental frequency:
p1 = Q var q nr
(note that the subscript 1 refers to the fundamental). C is the phase-to-neutral capacitance of one of the 3 branches of the filter bank represented as a star. At first glance, the presence of a reactor would not be expected to increase the reactive power supplied. The reason is the increase in voltage at power frequency f1 caused by the inductance at the capacitor terminals;
IZI
IZI
r
r f fr f (Hz)
11
13
f/f1
fig. 19: impedance of a network equipped with shunt filters.
fig. 20: Z versus f curve for a resonant shunt.
with: Qvar = reactive power for compensation produced by the filter, p1 = filter losses at power frequency in W; s the losses due to the harmonic currents cannot be expressed by simple equations; they are greater than:
pn =
2 Unr
terminals. Figure 21 shows one of the three phases of the filter. The 2nd order damped filter has zero reactance for a frequency fr higher than the frequency f where: f = 2 1 and L C
fr =
1+ Q q 2 q (Q 2 1) L C
in which Unr is the phase-to-phase harmonic voltage of order nr on the busbars after filtering. In practice, the performance of resonant shunt filters is reduced by mis-tuning and special solutions are required as follows: s adjustment possibilities on the reactors for correction of manufacturing tolerances; s a suitable compromise between the q factor and filter performance to reduce the sensitivity to mis-tuning, thereby accepting fluctuations of f1 (network frequency) and fr (caused by the temperature dependence of the capacitance of the capacitors).
damped filters
2nd order damped filter On arc furnaces, the resonant shunt must be damped. This is because the continuous spectrum of an arc furnace increases the probability of an injected current with a frequency equal to the anti-resonance frequency. In this case, it is no longer sufficient to reduce the characteristic harmonic voltages. The anti-resonance must also be diminished by damping. Moreover, the installation of a large number of resonant shunts is often costly, and it is therefore better to use a wide-band filter possessing the following properties: s anti-resonance damping; s reduced harmonic voltages for frequencies greater than or equal to its tuning frequency, leading to the name damped high-pass filter; s fast damping of transients produced when the filter is energised. The 2nd order damped filter is made up of a resonant shunt with a damping resistor R added at the reactor
The filter is designed so that fr coincides with the first characteristic line of the spectrum to be filtered. This line is generally the largest. When Q (or R) take on high values, fr tends towards f, which means that the resonant shunt is a limiting case of the 2nd order damped filter. It is important not to confuse Q, the quality factor, with Qvar, the reactive power of the filter for compensation. The 2nd order damped filter operates as follows: s below fr: the damping resistor contributes to the reduction of the network impedance at anti-resonance, thereby reducing any harmonic voltages; s at fr: the reduction of the harmonic voltage to a specified value is possible since, at this frequency, no resonance can occur between the network and the filter, the latter presenting an impedance of a purely resistive character. However, this impedance being higher than the resistance r of the reactor, the
filtering performance is less than for a resonant shunt; s above fr: the filter presents an inductive reactance of the same type as the network (inductive), which lets it adsorb, to a certain extent, the spectrum lines greater than fr, and in particular any continuous spectrum that may be present. However, antiresonance, if present in the impedance of the network without the filter, due to the existing capacitor banks, reduces the filtering performance. For this reason, existing capacitor banks must be taken into account in the design of the network and, in some cases, must be adapted. The main electrical characteristics of a 2nd order damped filter depend on n r = fr/f1 , the order of the filter tuning frequency, with: s fr = tuning frequency; s f1 = fundamental frequency (generally the power frequency, e.g. 50 Hz). These characteristics are: s the reactive power for compensation: For a 2nd order damped filter at operating voltage U1 (the subscript 1 referring to the fundamental), the reactive power is roughly the same as for a resonant shunt with the same inductance and capacitance, i.e. in practice:
Q var =
2 nr 2 nr 1 2 U1 C 2 f1
C is the phase-to-neutral capacitance of one of the 3 branches of the filter bank represented as a star.
phase X r inductive R capacitive fr 0 f (Hz) L
C neutral
f =
1 2 L C
fig. 21: 2nd order damped filter.
X0 =
s
L ; C
IZI r with resonant shunt Z network L
phase
the quality factor of the reactor: q = X0/r where r is the resistance of the reactor, dependent on the skin effect and defined at frequency fr; s the quality factor of the filter: Q = R/X0. The quality factors Q used are generally between 2 and 10; s the losses due to the fundamental compensation current and to the harmonic currents; these are higher than for a resonant shunt and can only be determined through network analysis. The damped filter is used alone or in a bank including two filters. It may also be used together with a resonant shunt, with the resonant shunt tuned to the lowest lines of the spectrum. Figure 22 compares the impedance of a network with a 2nd order damped filter to that of a network with a resonant shunt. Other types of damped filters Although more rarely used, other damped filters have been derived from the 2nd order filter: s 3rd order damped filter (see fig. 23) Of a more complex design than the 2nd order filter, the 3rd order filter is intended particularly for high compensation powers. The 3rd order filter is derived from a 2nd order filter by adding another capacitor bank C2 in series with the resistor R, thereby reducing the losses due to the fundamental. C2 can be chosen to improve the behaviour of the filter below the tuning frequency as well, which favours the reduction of anti-resonance. The 3rd order filter should be tuned to the lowest frequencies of the spectrum. Given the complexity of the 3rd order filter, and the resulting high cost, a 2nd order filter is often preferred for industrial applications; s type C damped filter (see fig. 24) In this filter, the additional capacitor bank C2 is connected in series with the reactor. This filter offers characteristics roughly the same as those of the 3rd order filter;
with 2nd order damped filter
f (Hz)
C neutral
fig. 22: the impedance, seen from point A, of a network equipped with either a 2nd order damped filter or a resonant shunt.
phase
phase
r R
R
L C2
L C2
C neutral
C neutral
fig. 23: 3rd order damped filter.

s damped double filter (see fig. 25) Made up of two resonant shunts connected by a resistor R, this filter is specially suited to the damping of the anti-resonance between the two tuning frequencies;
fig. 24: type C damped filter.

phase
ra
rb
low q resonant shunt This filter, which behaves like a damped wide-band filter, is designed especially for very small installations not requiring reactive power compensation. The reactor, with a very high resistance (often due to the addition of a series resistor) results in losses which are prohibitive for industrial applications.
s
La R Ca neutral
Lb
Cb neutral
fig. 25: damped double filter.
9. measurement relays required for the protection of reactor-connected capacitors and filters
An anti-harmonic reactor must withstand the 3-phase short-circuit current at the common reactorcapacitor terminals. Furthermore, both anti-harmonic reactors and filters must continuously withstand fundamental and harmonic currents, fundamental and harmonic voltages, switching surges and dielectric stresses. In this chapter, anti-harmonic reactorconnected capacitor assemblies and filters will be referred to collectively as devices.
basic protection against device failures

All the elements of these devices can be subject to insulation faults and shortcircuits, while the capacitor banks are mainly the source of unbalance faults caused by the failure of capacitor elements. s protection of these devices against insulation faults can be provided by residual current relays (or zero phase sequence relays). Note: the neutral is generally not distributed on such devices; for higher sensing accuracy, it is better to use a toroidal type transformer, encircling all the live conductors of the feeder, rather than three step-down current transformers; s protection against short-circuits can be provided by overcurrent relays installed on the filter feeder. This protection must detect 2-phase short-circuits at the common reactorcapacitor terminals, while letting through inrush transients; s detection of unbalance currents in the connections between the neutrals of
the double star connected capacitor banks (see fig. 26). In addition to the damage that can be caused by the resulting unbalanced stresses, the failure of a small number of capacitor elements is detrimental to filter performance. This protection is therefore designed to detect, depending on its sensitivity, the failure of a small number of capacitor elements. Of the single-pole type, this protection must be: insensitive to the harmonics, set to above the natural unbalance current of the double star connected capacitor bank (this unbalance depends on the accuracy of the capacitors), set to below the unbalance current due to the failure of a single capacitor element,
operate on a major fault causing an unbalance. The fluctuation of the supply voltage must be taken into account in the calculation of all these currents.
basic protection against abnormal stresses on the devices

These abnormal stresses are essentially due to overloads. To protect against them, it is necessary to monitor the rms value of the distorted current (fundamental and harmonics) flowing in the filter. It is also necessary to monitor the fundamental voltage of the power supply using an overvoltage relay.
phase 1
phase 2
phase 3
current relay
C/2
C/2
C/2
C/2
C/2
C/2
fig. 26: unbalance detection for a double star connected capacitor bank.
10. example of the analysis of a simplified network
The diagram in figure 27 represents a simplified network comprising a 2,000 kVA six-pulse rectifier, injecting a harmonic current spectrum, and the following equipment which will be considered consecutively in three different calculations: s a single 1,000 kvar capacitor bank; s anti-harmonic reactor-connected capacitor equipment rated 1000 kvar; s a set of two filters comprising a resonant shunt tuned to the 5th harmonic and a 2nd order damped filter tuned to the 7th harmonic. The capacitor bank implemented in this manner is rated 1,000 kvar. Note that: s the 1,000 kvar compensation power is required to bring the power factor to a conventional value; s the harmonic voltages already present on the 20 kV distribution network have been neglected for the sake of simplicity. This example will be used to compare the performance of the three solutions, however the results can obviously not be applied directly to other cases.
reactor-connected capacitor bank

This equipment is arbitrarily tuned to 4.8 f1. Harmonic impedance (see fig. 31)
The network harmonic impedance curve, seen from the node where the harmonic currents are injected, exhibits a maximum of 16 ohms (antiresonance) in the vicinity of harmonic order 4.25. This unfortunately favours the presence of 4th voltage
network 20 kV Isc 12.5 kA 20/5.5 kV 5,000 kVA Usc 7.5 % Pcu 40 kW
2,000 kVA disturbing equipment
5.5/0.4 kV 1,000 kVA Usc 5 % Pcu 12 kW motor load 500 kVA at cos = 0.9 560 kW
capa.
reactor + capa.
resonant shunt and 2nd order damped filter
capacitor bank alone

The network harmonic impedance curve (see fig. 28), seen from the node where the harmonic currents are injected, exhibits a maximum (antiresonance) in the vicinity of the 7th current harmonic. This results in an unacceptable individual harmonic voltage distortion of 11 % for the 7th harmonic (see fig. 29). The following characteristics are also unacceptable: s a total harmonic voltage distortion of 12.8 % for the 5.5 kV network, compared to the maximum permissible value of 5 % (without considering the requirements of special equipment); s a total capacitor load of 1.34 times the rms current rating, exceeding the permissible maximum of 1.3 (see fig. 30). The solution with capacitors alone is therefore unacceptable.
fig. 27: installation with disturbing equipment, capacitors and filters.
Z () 38.2
V (V)
350 11 %
7.75
9 10 11 13 H
fig. 28: harmonic impedance seen from the node where the harmonic currents are injected in a network equipped with a capacitor bank alone.
fig. 29: harmonic voltage spectrum of a 5.5 kV network equipped with a capacitor bank alone.
harmonic. However, the low impedance, of an inductive character, of the 5th harmonic favours the filtering of the 5th harmonic quantities. Voltage distortion (see fig. 32) For the 5.5 kV network, the individual harmonic voltage ratios of 1.58 % (7th harmonic), 1.5 % (11th harmonic) and 1.4 % (13th harmonic) may be too high for certain loads. However in many cases the total harmonic voltage distortion of 2.63 % is acceptable.
For the 20 kV network, the total harmonic distortion is only 0.35 %, an acceptable value for the distribution utility. Capacitor current load (see fig. 33) The total rms current load of the capacitors, including the harmonic currents, is 1.06 times the current rating, i.e. less than the maximum of 1.3. This is the major advantage of reactorconnected capacitors compared to the first solution (capacitors alone).
resonant shunt filter tuned to the 5th harmonic and a damped filter tuned to the 7th harmonic
In this example, the distribution of the reactive power between the two filters is such that the filtered 5th and 7th voltage harmonics have roughly the same value. In reality, this is not required. Harmonic impedance (see fig. 34) The network harmonic impedance curve, seen from the node where the harmonic currents are injected, exhibits a maximum of 9.5 ohms (antiresonance) in the vicinity of harmonic 4.7. For the 5th harmonic, this impedance is reduced to the reactor resistance, favouring the filtering of the 5th harmonic quantities. For the 7th harmonic, the low, purely resistive impedance of the damped filter also reduces the individual harmonic voltage. For harmonics higher than the tuning frequency, the damped filter impedance curve reduces the corresponding harmonic voltages. This equipment therefore offers an improvement over the second solution (reactor-connected capacitors).
I (A)
V (V) 50 1.55 %
- 82
19 0.6 %
48 1.5 % 45 1.4 %
9 10 11 13
11 13 H
fig. 30: spectrum of the harmonic currents flowing in the capacitors for a network equipped with a capacitor bank alone.
fig. 32: harmonic voltage spectrum of a 5.5 kV network equipped with reactorconnected capacitors.
Z ()
I (A)
15.6
Z ()
34 24 %
9.5
4.7
~ 4.25 4.8 H
11 13
fig. 31: harmonic impedance seen from the node where the harmonic currents are injected in a network equipped with reactorconnected capacitors.
fig. 33: spectrum of the harmonic currents flowing in the capacitors for a network equipped with reactor-connected capacitors.
fig. 34: harmonic impedance seen from the node where the harmonic currents are injected in a network equipped with a resonant shunt filter tuned to the 5th harmonic and a damped filter tuned to the 7th harmonic.
Voltage distortion (see fig. 35) For the 5.5 kV network, the individual harmonic voltage ratios of 0.96 %, 0.92 %, 1.05 % and 1 % for the 5th, 7th, 11th and 13th harmonics respectively are acceptable for most sensitive loads. The total harmonic voltage distortion is 1.96 %. For the 20 kV network, the total harmonic distortion is only 0.26 %, an acceptable value for the distribution utility. Capacitor current load The total rms current load of the resonant shunt filter capacitors (see fig. 36) is greater than 1.3 time the current rating. The capacitance must therefore be increased, which will improve the filtering performance,
reducing the 5th harmonic ratio to less than 1 %. The result is of course an increase in the reactive power compensation capacity. To avoid overcompensating, a compromise must be found for the size of these capacitors. The calculation is therefore repeated with this new data. For the damped filter tuned to the 7th harmonic, the total rms current load of the capacitors (see fig. 37) is within the tolerance of 1.3 times their current rating. This example demonstrates an initial approach to the problem. However in practice, over and above the calculations relative to the circuit elements (L, r, C and R), other calculations are required before
proceeding with the implementation of any solution: s the spectra of the currents flowing in the reactors connected to the capacitors; s the total voltage distortion at the capacitor terminals; s reactor manufacturing tolerances and means for adjustment if necessary; s the spectra of the currents flowing in the resistors of the damped filters and their total rms value; s voltage and energy transients affecting the filter elements during energisation. These more difficult calculations, requiring a solid understanding of both the network and the equipment, are used to determine all the electrotechnical information required for the filter manufacturing specifications.
V (V)
I (A)
I (A)
0.96 % 0.91 %
1.05 % 1%
39
22 23 % 10 10 %
11 13
11 13
fig. 35: harmonic voltage spectrum of a 5.5 kV network equipped with a resonant shunt filter tuned to the 5th harmonic and a damped filter tuned to the 7th harmonic.
fig. 36: spectrum of the harmonic currents flowing in the capacitors of a resonant shunt filter tuned to the 5th harmonic on a network equipped with a damped filter tuned to the 7th harmonic.
fig. 37: spectrum of the harmonic currents flowing in the capacitors of a damped filter tuned to the 7th harmonic on a network equipped with a resonant shunt filter tuned to the 5th harmonic.
11. conclusion
Static power converters are increasingly used in industrial distribution. The same is true for arc furnaces in the growing electricpowered steel industry. All these loads produce harmonic disturbances and require compensation of the reactive power they consume, leading to the installation of capacitor banks. Unfortunately these capacitors, in conjunction with the inductances in the network, can cause high frequency
oscillations that amplify harmonic disturbances. Installers and operators of industrial networks are thus often confronted with a complex electrical problem. The main types of harmonic disturbances and the technical means available to limit their extent have been presented in this document. Without offering an exhaustive study of the phenomena involved or relating all
acquired experience, this document should provide the necessary background to, if not solve the problems, at least facilitate discussions with specialists. For further information or assistance, feel free to contact the Network Studies department of the Central R&D organisation of Merlin Gerin, a group of specialised engineers with more than twenty years of experience in this field.
12. bibliography
Standards s IEC 146: Semi-conductor converters. s IEC 287: Calculation of the continuous current rating of cables. s IEC 555-1: Disturbances in supply systems caused by household appliances and similar electrical equipment - Definitions. s IEC 871-1 and HD 525.1-S-T: Shunt capacitors for AC power systems having a rated voltage above 660 V. s NF C 54-100. s HN 53 R01 (May 1981): EDF general orientation report. Particular aspects concerning the supply of electrical power to sensitive electronic equipment and computers. Merlin Gerin's Cahier Technique s Residual current devices Cahier Technique n 114 R. CALVAS s Les perturbations lectriques en BT Cahier Technique n 141 R. CALVAS Other publications s Direct current transmission, volume 1 E. W. KIMBARK published by: J. WILEY and SONS. Le cyclo-convertisseur et ses influences sur les rseaux d'alimentation (The cyclo-converter and its effects on power supply networks T. SALZAM and W. SCHULTZ - AIM Lige CIRED 75.
s
s Perturbations des rseaux industriels et de distribution. Compensation par procds statistiques. Rsonances en prsence des harmoniques crs par les convertisseurs de puissance et les fours arc associs des dispositifs de compensation. (Disturbances on industrial and distribution networks. Compensation by statistical processes. Resonance in the presence of harmonics created by power converters and arc furnaces associated with compensation equipment.) Michel LEMOINE - DER EDF RGE T 87 n 12 12/78. s Perturbations des rseaux industriels et de distribution. Compensation par procds statistiques. Perturbations de tension affectant le fonctionnement des rseaux fluctuations brusques, flicker, dsquilibres et harmoniques. (Disturbances on industrial and distribution networks. Compensation by statistical processes. Voltage disturbances affecting network operation - fluctuations, flicker, unbalances and harmonics. M. CHANAS - SER-DER EDF RGE T 87 n 12 12/78. s Pollution de la tension (Voltage disturbances). P. MEYNAUD - SER-DER EDF RGE T 89 n 9 09/80. s Harmonics, characteristic parameters, methods of study, estimates of existing values in the network. (ELECTRA) CIGRE 07/81. s Courants harmoniques dans les redresseurs triphass commutation force. (Harmonic currents in forced commutation 3-phase rectifiers) W. WARBOWSKI CIRED 81.
Perturbations rciproques des quipements lectroniques de puissance et des rseaux - Quelques aspects de la pollution des rseaux par les distorsions harmoniques de la clientle (Mutual disturbances between power electronics equipment and networks - Several aspects concerning network pollution by harmonic distortion produced by subscribers). Michel LEMOINE - DER EDF RGE T 85 n 3 03/76.
s
Origine et nature des perturbations dans les rseaux industriels et de distribution. (Origin and nature of disturbances in industrial and distribution networks). Guy BONNARD - SER-DER-EDF RGE 1/82. s Problmes particuliers poss par ltude du phnomne de distorsion harmonique dans les rseaux. (Particular problems posed by the study of harmonic distortion phenomena in networks). P. REYMOND CIGRE Study Committee 36 09/82. s Rduction des perturbations lectriques sur le rseau avec le four arc en courant continu (Reduction of electrical network disturbances by DC arc furnaces). G. MAURET, J. DAVENE IRSID SEE LYON 05/83. s Line harmonics of converters with DC motor loads. A. DAVID GRAHAM and EMIL T. SCHONHOLZER. IEEE transactions on industry applications. Volume IA 19 n 1 02/83. s Filtrage dharmoniques et compensation de puissance ractive Optimisation des installations de compensation en prsence d'harmoniques. (Harmonic filtering and reactive power compensation - Optimising compensation installations in the presence of harmonics). P. SGARZI and S. THEOLERE, SEE Seminar RGE n 6 06/88.
s
Ral. : Illustration Technique Lyon -
DTE - 10/94 - 2500 - Imprimeur :
n 154
Low Voltage circuit-breaker breaking techniques
Robert Morel .... to my working friends and colleagues. Gratuated with an engineering degree from ENSMM in Besanon and joined Merlin Gerin in 1971. Specialised in designing low-voltage switchgear and participated in designing the Sellim system. In 1980, took over development of Compact Circuit-breakers and Interpact switches. In 1985, became manager of the Low-Voltage Current Interruption design office in the Low-Voltage Power Components division
E/CT 154 first issued, december 1994
glossary
: closing voltage angle. category (of a circuit-breaker): defined by standard IEC 947.2. A = circuit-breaker not delayed on opening in short-circuit conditions; B = circuit-breaker delayed on opening in short-circuit < Icw conditions. E: DC rated voltage, AC peak voltage. : voltage/current phase angle. i, (i0): current at an instant t, (at an instant t0). ia: arcing current at an instant t. c: broken peak current. Icc: short-circuit current. Ics: rated service breaking capacity (expressed in kA or in % of Icu). Icu: rated short-circuit ultimate breaking capacity. Icw: rated acceptable short current. In: rated current in steady state, Ar.m.s. Ip: prospective current. Is: overload current. r: generator impedance. R, L, C: total components of broken circuit. rating (of a circuit-breaker): rated current value of the circuit-breaker, defined by the maximum setting of the trip unit (electronic or thermal overload protection). size (of a circuit-breaker): the highest rating accepted by a circuit-breaker case. It is the rated current of the device. t, (t0 ): time (initial instant). ta: arcing time.
: time constant.
u: voltage at an instant t. ua: arcing voltage at an instant t. Ua: stabilised arcing voltage. UAC: Anode-Cathode voltage of each elementary arc. Ud+, Ud-: regeneration characteristics. Un: rated voltage in AC, V r.m.s. Ur: recovery coltage. Wa: arcing energy. WL0: initial inductive energy for i = i0. : AC pulsation ( = 2f = 2/T).
Low Voltage circuit-breaker breaking techniques
content
1. Introduction 2. The electric arc definition of the currents to break p. 4 its formation conditions its physical properties its electrical properties its extinguishing conditions 3. Using the arc to break 4. Breaking steady-state currents in DC voltage in single-phase AC voltage in three-phase AC voltage 5. Breaking prospective currents (with limitation) definitions breaking with limitation in DC voltage in single-phase AC voltage in three-phase AC voltage the breaking parameters the fuse breaking technique 6. The low voltage circuit-breaker its functions its technologies its performances 7. Conclusion 8. Bibliography p. 5 p. 5 p. 5 p. 6 p. 7 p. 8 p. 8 p. 9 p. 10 p. 11 p. 11 p. 12 p. 12 p. 13 p. 14 p. 15 p. 16 p. 18 p. 20 p. 20
This Cahier Technique aims to: s define the currents to break in LV installations; s explain a phenomenon which occurs whenever a circuit is opened: the electric arc; s theoretically analyse the breaking of various currents by the arc placed in the extinguishing condition; s describe breaking devices and in particular circuit-breakers from a technological standpoint.
1. introduction
The energy sources for electrical installations are AC generators and transformers. All these generators, however perfect, have an internal impedance which has two major effects (see fig. 1). s in normal operating conditions, this impedance causes voltage to drop at no-load condition to bring it to Un when the generator delivers In; s when a short-circuit occurs, this impedance limits the current to a value, given in multiples of In. To take the example of transformers, their short-circuit voltage Ucc (as a %) corresponds to the upper limit of their relative voltage drop under rated current, In. For example, an Icc of In/5 % = 20 In is obtained for a short-circuit voltage of 5%, i.e. a current of 29 kA for a 1000 kVA/400 V transformer. It is not hard to imagine the damage such a current could cause in an installation (temperature rises and electrodynamic forces are proportional to the current square!). Thus, even if all precautions are taken to make such an occurrence unlikely, protective devices are still needed to break the short-circuit currents.
U = U0 - Zi I n U0 Un = Zi I n Un
In
operational current overload current short-circuit current
Icc
fig. 1: incidence of the internal impedance, Zi, of energy sources on voltage delivered and their maximum short-circuit current.
definition of currents to break

Knowing the value of the current to break is not enough to design a suitable breaking device ! Current breaking is dependent on a number of parameters relating to generators (AC generators or transformers), lines and loads: s an electric circuit is always inductive, and thus the very fluctuations in the current to break, generate, as soon as the circuit is opened, negative current feedback which help maintain the current. The value of this backelectromotive voltage of the L di/dt type may be high whatever the value of current i until this current is cancelled;
s the resistive value of the circuit to break is of assistance in breaking as long as the current is high, but ceases to be of any help when current tends to zero, since the ohmic drop is then negligible; s the capacities between live conductors, whether distributed (stray capacities of generators and cables) or additional (capacitor bank in reactive energy compensation or filter), alter breaking conditions; s the frequency of the current to break, since in theory it is easier to break AC currents with periodic zeros than DC currents; s finally, the voltage delivered by the generator. Once the current has been cancelled, the breaking device must dielectrically withstand the mains voltage still present.
depends on the generator characteristics, 3% < Ucc < 7% for example; s it may be smaller: s according to the fullness of the fault s according to the length and crosssection of the upstream lines; s it may be greater if a number of generators are parallel-connected. 2. Overload current The current may overshoot rated value and become inacceptable after a certain time: s during the transient period of load startup or operation; s if the sum of the powers of the loads in operation exceeds the designer's estimates for all or part of the installation. 3. Rated current (or lower) Since a circuit-breaker's function is to break high short-circuit currents and overloads, it can also provide circuit and load control.
s it
In practice, there are three types of currents to break: 1. Short-circuit current In a given point of an installation, this current is not systematically equal to 20 In of the generator:
2. The electric arc
The electric arc is no invention, but appeared to the first physicist who tried to break a circuit through which a current flowed. Circuits, always inductive, supply electrons with sufficient energy to cross the distance in the conductor separation zone. The gas present, normally air, is ionised by these "pioneer" electrons and the resulting plasma will then facilitate current flow. Breaking thus seems somewhat compromised.... unless a better understanding of this phenomenon were to reveal remarkable and even irreplaceable properties. Luckily this is the case !
spot, resulting in arc stagnation which can give rise to metallic vapours. These vapours and the ambient gas will then be ionised, hence: s more free electrons; s the creation of positive ions which drop back on the cathode and ensure its high temperature is maintained; s the creation of negative ions which bombard the anode causing temperature to rise. All this occurs in a high temperature plasma column, from 4 000 to 20 000 K according to column current and confinement.
Anode
ion -
ion +
a: composition of the arc column.
e N ion +
Cathode
its electrical properties

(see fig. 2b) most striking property is appearance of an arcing voltage, which has: s a fixed part, UAC 20 to 40 V which appears on the slightest separation of the contacts (according to the materials used), s a variable part, UL = 50 to 100 V/cm, when the arc is stabilized in elongation in pressure-temperature balanced conditions. i.e. a total value Ua = UAC + UL. Note that: - the sign of Ua changes at the same time as the arcing current sign, - the arcing current value does not have any real effect on arcing voltage. This is because the arc works with a virtually constant current density (j = i/s) (the anode and cathode spot crosssections and the arc column crosssection are proportional to the current, hence by analogy with a resistance: l U = R i = i = l j = constant); s s an arcing energy is produced; Wa =
s if s its
its formation conditions

The arc appears in gaseous atmospheres: s by dielectric breakdown between two electrodes: s beyond an electric field value E/d, dependent on electrode shape and on gas type and density (d = distance between electrodes). s further to moving over ablative materials in the ambient gas. s as soon as an electric circuit opens through which current flows, even if the circuit is purely resistive, a certain distance is called for to prevent dielectric breakdown. Moreover, if the circuit is inductive, the distance required is even greater due to the high L. di/dt (result of rapid fluctuation in current strength).
Anode
UAC
Cathode UAC ~ 30 V on contact separation.
Anode
UL
its physical properties

(see fig. 2a) As soon as two contacts separate, one of them (cathode) transmits electrons and the other one (anode) receives them. Since electronic emission is by its very nature energy generating, the cathode will be hot. With the arc foot thus becoming thermoemissive, the electrons are mostly emitted at the hot
u a i a dt ;
Cathode UL ~ 70 V/cm after arc elongation. Ua = UAC + UL b: arcing voltage.
the arc is placed in a magnetic field, it is subjected to Laplace's law forces, F = B i l sin which causes it to bend if r r B is perpendicular to i , and then to move transversely.
fig. 2: the electric arc, its physical phenomenon (a) and its characteristic or arcing voltage Ua (b).
its extinguishing conditions

The arc is extinguished when the arcing current becomes and remains zero. Thermal aspect When the arcing current is low or drops, below 10 A for example, the heat energy exchanges may exceed internal arcing energy, causing the arc to die of cold. This results in an increase in arcing voltage (see fig. 3a). During this voltage rise, the arc may even suddenly be extinguished if it is short-circuited by stray capacities. This is the case when arcing voltage becomes and remains greater than the charging voltage of the distributed capacities (see fig. 3b). This phenomenon is known as pinching off. However there are exceptions: s if the arcing current stabilises against an insulating wall, its heat exchange area decreases, and the insulating material components, locally very hot, may promote arc conduction and holding. s if the arcing current is high, the arc column is extremely exothermal and only the joint evolutions in arcing voltage and mains voltage can reduce and finally cancel this current.
Dielectric aspect The arcing current is not extinguished merely by reaching zero. In addition, the atmosphere ionised up to that point must be dielectrically regenerated in order to withstand the mains voltage still present ! These regeneration phenomena of recombination of + or - ions and electrons are fortunately very fast ! Thus, in practice, for the arcing current to remain zero, mains voltage must be less than the regeneration characteristic (Ud). If arcing voltage becomes and remains greater than mains voltage (in absolute value for AC voltage), regeneration will begin as the current zero approaches. The number of plasma current loads is adjusted to the strict minimum and becomes zero at the same time as the current. However the arc and the stray capacities have the same voltage until the arcing current is extinguished. Once this happens, this voltage rejoins mains voltage due to free oscillation between these distributed capacities and circuit constants L and R (see fig. 4). This voltage "connection" is known as the Transient Recovery Voltage (TRV). Low capacities mean these oscillations have a very high frequency and are extremely damped. These conditions are present: s in DC voltage: (see fig. 5a); The arcing voltage, Ua, is greater than mains voltage, Ur, when the current is cancelled, and the regeneration
characteristic, Ud, remains greater than Ur with TRV. s in AC voltage: s when the instantaneous mains voltage value still has the same sign as arcing voltage at the time of current zero (see fig. 5b). Final breaking will occur when subsequent change in mains voltage ceases to cross the regeneration characteristics in either positive or negative values. s when the instantaneous mains voltage value has the opposite sign to arcing voltage but has an absolute lower value (see fig. 5c). The arc is permanently extinguished if the TRV does not exceed the regeneration characteristic.
i,u
Ua Ur i
a i,u Ua Ud +
i,u
thin and unstable arc Ud Ua ia Ur
t Ur b Ud -
i,u Ud i,u Ua Ud + i
im t im a: arc extinguishing L,r ic C ia

im ia
Ua Ur TRV t Ur im L,r t c Ud -
b: the stray capacities "short-circuit" the arc.
C i
TRV
fig. 3: extinguishing of an electric arc by pinching off
fig. 4: the transient recovery voltage TRV
fig. 5: arc in the extinguishing condition. a- in DC voltage b- in AC voltage where Ur has the same sign as Ua at the time of current zero, c- in AC voltage where Ur has the opposite sign to Ua.
Otherwise, when the TRV breaks the regeneration curve, a postarcing current of the electroluminescent type may appear. In this case: - if the postarcing current is still of the small type, extinguishing conditions are still present;
- if the postarcing current exceeds a critical value under an equally critical voltage, the arcing current will be recovered and another zero will be required to break.
3. using the arc to break
The current established on closing can be calculated by Ohm's general law: e - R i - L di/dt = 0. Following a transient closing state, the current becomes stable or in steady e state I = . R On the basis of such a law, the current can only be cancelled permanently if the voltage e becomes zero or if R becomes infinite.
In preference to these two extremes which would present too many operational restrictions, an arc was introduced in the circuit for use of its Ua voltage properties and extinguishing conditions. As soon as the circuit opens, the equation becomes: e - R i - L di/dt - ua = 0. The current will thus be forced to zero or will pass through zero, and the arc
extinguishing conditions will break the current. Examination of the two different cases below provides a progressive approach to the breaking theory: s the Ua arcing voltage was introduced in the circuit when the current was in steady state (see chapter 4); s the Ua arcing voltage was introduced in the circuit before the current reached the stabilized value of the prospective current (see chapter 5).
4. breaking steady-state currents
Steady-state currents are rated currents, overload currents and shortcircuit currents which have reached a stable value on circuit opening. Circuit opening may be: s voluntary, controlled by the user, completely separate from current value; s reflex, by the action of a device affected by current value and directly or indirectly controlling circuit opening. For simplicity's sake, breaking conditions are examined : s in DC voltage; s then in AC voltage.
It is easier to interprete this expression by writing: WL0 = (1/2 L i02) and observing the curves (Wa/WL0), and (ta/) as a function of (Ua/E), (see fig. 7). These curves show: s if Ua = E then Wa/WL0 = 2 only! But breaking time is infinite! s if Ua is very big or even infinite then: Wa/WL0 = 1. Arcing energy is equal to initial inductive energy and breaking time is virtually zero: arcing power Wa/ta is very high! s that the "bend" of curve Wa/WL0 is a practical optimum and thus that
Ua < 2.5 is a good compromise; E then Wa 1.2 WL0 and ta . The coefficient 1.2 (read on the bend) is very satisfactory due to its proximity to the 1 minimorum minimum , difficult to reach. 1.5 <
Ua
E, r
L R
ua Ua E
in DC voltage
(see fig. 6) u=E before opening: i0 = E/R after opening: E - R i - L di/dt - ua = 0 When the contacts open, ua moves towards a maximum value Ua. Ohm's general law shows that current can only be forced to 0 if ua becomes greater than E. Otherwise, it will move to i0 = (E - Ua)/R, not zero. For current breaking purposes, it is thus easier and sufficiently clear to consider this arcing voltage as a step function, ua = Ua for t > t0, (t0 = instant when ua = E). The complete calculation then yields:
t E Ua ia = 1 e and R R
t0
Ua i,u
in single-phase AC voltage
u = E sin t i = I0 cos (t + ) where cos = L
i0 ia
Ur
(L ) 2 + R 2 (L )
2
t0
2
ta
and I 0 = E
+R
fig. 6
Wa W L0 ___ 2
ta __ 2
t a = Log
Ua Ua E
1.5
remembering that breaking occurs as soon as the current passes through zero (a negative current due to dominance of Ua compared with Ur has no physical significance). Calculation of the integral:
Wa WL0 ___
1.5
1 0.5 ta __
1 0.5
Wa =
ta
0
0.2
u a i a dt gives
1.5
10
Ua E __
Ua Ua Ua 1 Log Wa = Li0 2 2 1+ 1 2 E Ua E E
optimum
fig. 7: curves Wa/WL0 and ta /.
The arc appears as soon as the contacts separate, and evolution of its voltage in time may appear complex. However ua still has the sign of i and its mean absolute value tends towards Ua (see fig. 8). The mathematical study of ia, ta and Wa, based on Ohm's general law u - R i - L di/dt - ua = 0 is still possible but not so easy. Moreover, since these calculations do not consider the voltage recovery conditions of real AC breaking, the two cases Ua E and Ua << E must be analysed: s if Ua E, (see fig. 9), the arcing voltage helps force the current to 0 and to hold it there, whatever the phase shift of i compared with u; s if Ua << E, breaking is still possible and easier, overall, than in DC because of the natural zeros of i. Successful breaking depends on postarcing phenomena at each zero current. In fact, this condition can be summed up by a race between arc dielectric regeneration and mains voltage. Let us look at two possibilities: s if the symmetry of Ua acquired at a current zero is greater than mains voltage at that instant (see fig. 10a), including TRV, then the dielectric regeneration evolution curve remains greater than mains voltage and breaking occurs;
s if the symmetry of Ua acquired at a current zero is less than mains voltage at that instant (see fig. 10b), including TRV, then the mains voltage may well break the dielectric regeneration curve if this curve is too slow. In this case, arcing may reoccur and no break occurs, at least not at that current zero! In both these cases, the power factor cos of the circuit to break has considerable influence due to the phase shift of the current zeros compared with mains voltage value. In particular, if cos 1, voltage and current are zero at the same time and breaking is easy.
neutral voltage ( 3 2 = 0.86 instead of 1; moreover the slightest dielectric weakness in one pole would cause the other pole to break under phase-tophase voltage).
i,u Ua ia Ud+
t Ua Ur Uda
in three-phase AC voltage
When the neutral wire is distributed, three-phase voltage breaking conditions are the same as for singlephase voltage, reasoning in phase to neutral voltage, phase by phase. When the neutral wire is not distributed, the short-circuit point defines a floating neutral point (see fig. 11). Thus: s the first pole breaking has to withstand a recovery voltage equal to an intermediate voltage since the neutral point moves from N to N' (in fact N' moves to N" in proportion to the arcing voltages on the other two phases). Voltage recovery is thus penalised by a factor of 1.5 (at 3 ); s the two remaining poles are in series to ensure permanent breaking under phase-to-phase voltage. Breaking is made easier if each phase has an arcing voltage Ua. However, it is not really easier than breaking the same current on just one phase in phase to
i,u Ua i Ud+
UaUdb
Ur
i a ,ua Ua ia
fig. 10
t i,u Ur i Ua(t) t Ua t t0 ta
i2 i1 = 0 i2 = i 3 N' and N" = Floating neutral N" N' Un N i3
Ua+
i1
IUaI
Ua
fig. 8
fig. 9
fig. 11
5. Breaking prospective currents (with limitation)
definitions
Prospective current In an installation, this is the current which would flow through a circuit if each connection device pole or the fuse
were replaced by a conductor of negligible impedance (IEC 50). In a switchgear test circuit, it is the calibration current. Remember that: s in DC voltage, current evolution takes the form:
E (1 e ) = P (1 e ) (see fig. 12); R s in single-phase AC voltage:the moment of appearance of the fault or t t
i E R
i=
Ip
E R
the moment of closing, compared with mains voltage value, considerably influences evolution of the transient current. If this moment were characterised by its closing voltage angle , voltage may be written as: u = E sin (t + ), (see fig. 13a) Current evolution takes the shape: R t E i = sin (t + ) sin ( ) e L Z
VOLTA laboratory A3076 90/05/31/001
fig. 12
.005 10 ms
200.0
real assymmetry
channel 1 40 kA I1 channel 4 205 V V1
i "symmetrical" i
channel 2 40 kA I2 channel 5 204 V V2 channel 3 40 kA I3 channel 6 204 V V3
quasi-symmetry
= b
small loop
"asymmetrical" i
t =0 c voltage waves current waves
fig. 13
fig. 14: Oscillograms for test circuit breaking in three-phase AC voltage.
with two components: s an AC one, with a phase shift of /u, s a DC one, tending to 0 when t tends to infinity. Two special cases are defined by: s = known as the symmetrical condition (see fig. 13b) The current has the shape: i= E sin t Z
three limitation conditions (see fig. 15c): s creation of an arcing voltage as early as possible; s increase of this arcing voltage as quickly as possible to obtain Ua = e - R
In short Early, Quickly, High. Such is the slogan for : Using the arc to break prospective currents, with limitation
Right from the start the current has the same curve as in steady state and a peak value of E/Z. s = 0 known as the asymmetrical condition (see fig. 13c) The current has the shape:
di = 0 , which means that dt the current has then reached a maximum value c; s holding this arcing voltage, Ua, at as high a value as possible; di/dt is then negative and the current is forced to 0.
i and thus L
in DC voltage
DC voltage takes the form u(t) = E. s until the circuit is opened, current evolves as in the formula:
E (1 e ) = p (1 e ) R s when the circuit is opened, an arcing voltage appears. If it increases rapidly, its overall evolution may be likened to a step function with a rising voltage defined by ua = E at an instant t0 (see fig. 16). The current, having reached a value i0, then decreases exponentially and disappears after a time ta << . Calculation of arcing energy: t t
i=
Ip
i=
E Z
R t sin (t ) sin e L
broken i
The first peak value of the current is thus a function of the circuit cos . (In the borderline case where R = 0, the current would keep the same sign as the first voltage half-wave)! s in three-phase AC voltage (see fig. 14) The current in each phase may result in the same special cases (symmetrical and asymmetrical) as in single-phase. In any case, whatever the value of , there is nearly always: s a phase in quasi-symmetrical condition, s a phase in quasi-asymmetrical condition, s the last phase is said to be in small loop.
t a
Wa
ta
t0
u a i a dt , gives:
Ua Un R i0 1 R i Log 1+ Ua Un 0
Ua 1 Wa L i 0 2 2 2 R i0
Ua u
Ur Ua
Ip Ip
i0
ic
Breaking with limitation

This expression means that measures are taken to prevent the short-circuit current from reaching the maximum peak value of its prospective current (see fig. 15a). This objective is an important one and in many cases vital if damage to the installation is to be avoided. Arc limitation is only possible if arcing voltage quickly becomes and remains greater than mains voltage (see fig. 15b). In fact, Ohm's law: eRiL di Ua = 0, is used to define dt
i c di = 0 dt di < 0 dt t c
t0 t ua
Ua E
fig. 15
fig. 16
Hence the curve network (Wa/WL0), (see fig. 17), when the limitation ratio k = i0 / ip is introduced. Note that the smaller the ratio k, the lower the arcing energy. This energy is optimal for 1.5 < Ua / E < 2.5, which was the case in steady-state current.
In single-phase AC voltage
In limitation conditions, this current is broken as though it were temporarily a DC voltage break. s in the symmetrical condition, in particular, it is virtually equivalent to consider breaking under prospective current with a mains voltage E = Un 2 (see fig. 18a). s in the asymmetrical condition, limitation is often better since the arcing voltage breaks the mains voltage before the current really has chance to evolve (see fig. 18b). s in all the intermediate cases, with the small loop, breaking with limitation may only occur on the second current half-wave, since the strength of the first half-wave was too low (see fig. 18c). Remark: Efficient limitation on high short-circuit currents is only possible if the arcing voltage appears within a time much less than T/4.
react on a tripping device ensuring very fast opening of all poles. In this case, the arcing voltages develop on all three phases at the same time. It is as though the phase in the quasisymmetrical condition were broken in its phase-to-phase voltage with a double arcing voltage. This opening of all poles must take place within a time < T/4 and will be most efficient for < T/8. The small loop phase will then be broken although little current has flown through it. This breaking behaviour: s occurs on devices with low averall inertia of their moving parts; s is sought on large-size equipment with ultra-fast external operating energy (for example, with Thomson effect with capacitive discharge).
i,u
Ua Un 2 i p SYM. ic Ur
T/2 a
i,u
i p ASYM.
2
Ur ic T/2 b i
k=1 k = 0.9 k = 0.8 k = 0.7
15
Wa WL0 ___
in three-phase AC voltage
Two cases should be considered: 1 case: separate opening of the poles Each phase generates an arcing voltage according to the current flowing through it (see fig. 19). At first sight, it is as though: s one of the phases breaks in the single-phase symmetrical condition, but with voltage recovery in intermediate voltage. s finally, the other two phases ensure two-phase breaking of a current tail. 2 case: simultaneous opening of the poles The current of the phase in the symmetrical condition is the first to
nd st
1.2 1 0.95 0.85 0.75 0.5 0.3 0.2 0 1 1.5 2 2.5
ip
ia
k = 0.6 k = 0.5
t
k = 0.25 3 Ua E __
i p "small loop"
fig. 17
fig. 18
the breaking parameters

The parameters chosen to assess breaking efficiency are: s broken peak current = c (absolute value of the maximum peak current) Knowledge of this value enables definition of the maximum electrodynamic stresses in the circuit: s thermal stress or Joule's integral =
Since the shape of the broken current does not correspond to a simple mathematical function, this integral is calculated step-by-step by computer. This integral expresses both thermal and electrodynamic energy stresses on the circuit elements; s breaking time = ta Total breaking time if the circuit is three-phase; s maximum arcing time = Ua Magnitude: s 250 to 500 V for standard circuitbreakers,
600 to 900 V for limiting circuitbreakers. These overvoltages do not present a risk since they are lower than the test voltages standardised for LV installations;
s s
arcing energy =
a i a dt
dt
This term is the recognised expression.
This integral is also calculated step-bystep by computer and expresses the energy consumed in the arcing zone. Magnitude: 1, 10, 100 kJ according to the device and the currents broken. It conditions device breaking endurance.
VOLTA laboratory A0201 89/02/014
1 ms
i p1
channel 1 20 kA channel 4 200 V
I1
V1
I2
ip2 V2
I3
V3
fig. 19: oscillograms for test circuit breaking in three-phase voltage with separate opening of poles.
fuse breaking technique

A fuse breaks because of the arc. Its relative simplicity lies in the fact that a carefully calculated filament is brought to its melting temperature by the current flowing through it. For high currents, the temperature rise resulting in filament melting is of the adiabatic type; its prearcing energy is defined by the formula: R
appearing in the breaking unit (see fig. 20). This breaking unit can be filled with silica powder which, as it melts, will absorb the arcing energy. Note: the current tail is explained by the preferential path created by the arc in the melted silica. The arc decreases in size against the walls which are still warm. A few remarks about fuses: s their action is restricted to high overload and short-circuit currents; s some types of fuses have strikers for melt indication purposes, as well sometimes as to indirectly act on an extra breaking device to ensure opening of all phases; s following a fuse fault and melt, some survivors may have come close to melting and their characteristics may be altered. They may then melt inopportunely under a current lower than their rating. All fuses must therefore be replaced at the same time.
t pa
i dt = m c Tf
where R = filament resistance, m = filament mass, c = thermal capacity, Tf = melting temperature, tpa = prearcing time. This prearcing thermal energy is separate from mains voltage. The arc quickly assumes the length of the melted filament and the arcing voltage takes on a value in accordance with this length and the pressure
i ip
PPPPP @@@@@ ,,,,, PPPPP @@@@@ ,,,,, PPPPP @@@@@ ,,,,, PPPPP @@@@@ ,,,,, PPPPP @@@@@ ,,,,, PPPPP @@@@@ ,,,,,
fuse link t pa silica insulating enclosure Ur connection device U Ua t
t i
ia
fig. 20: the fuse, its composition and its characteristic breaking curves.
6. the low voltage circuit-breaker
A circuit-breaker (see fig. 21) is a connection device able to close and break a circuit irrespective of current up to its ultimate breaking capacity: Icu (refer to standard IEC 947-2). Although its main function is to break short-circuit and overload currents by self-energized reflex action, it also breaks normal currents and overload currents by voluntary action from external sources. Moreover, after opening, it provides voltage insulation of the broken circuit. The circuit-breaker's design enabling it to house all these functions in the same case has led to the adoption of specific solutions regarding: s closing/opening mechanisms; s trip units; s electrical pole circuits; s breaking elements (contacts, arc chutes...). The purpose of this chapter is to analyse its functions, technologies and performances.
and the possibility of quick opening are to be obtained. Furthermore, if the circuit-breaker is of the discriminated type, it may require a high electrodynamic withstand to accept short-circuit currents during the discrimination time, necessary for the downstream devices to operate. Circuit opening, current breaking s by voluntary action on the mechanism, manual or remote controlled; any current can be broken. s by reflex action on the mechanism by the trip unit due to an overcurrent. The circuit-breaker automatically and permanently opens, even if the
operating device is held in the closed position. s by action of an auxiliary trip unit on the mechanism: Undervoltage, energising, earth leakage current devices... Opening is automatic and permanent: the current can have any value at this time. Isolation When the circuit-breaker is open, a certain isolation level is required between the energized and deenergized parts. This isolation level is confirmed by: s a maximum leakage current test betwen input and output under max Ue;
its functions
Circuit closing Action on the mechanism supplies the load(s) with current. When energized, some loads absorb currents far greater than rated current In (e.g. motor 7 to 8 In for a few seconds). To prevent these overcurrents resulting in dangerous phenomena for the contact zone (erosion by arcs), closing must be prompt for values 100A. If they are to suit all standard cases, circuit-breakers must therefore be able to establish currents 15 to 20 times greater than their rated current. Specific measures must be taken to perform this function, since a circuitbreaker must always be ready to open again in the event of an installation fault, even during or just after it has closed ! Current conducting This passive function requires a number of construction precautions if both an acceptable temperature rise
fig. 21: cross-section of an industrial 400 A LV circuit-breaker.
an impulse voltage (e.g. 12.3 kV instead of the 9.8 kV required for a device of the same type without this function); s a mechanism sturdiness test, known as the welded contact (see Cahier Technique Merlin Gerin n 150).
s
its technologies
The mechanisms The three basic principles are: s mechanism with 2 stable positions (for circuit-breakers with ratings under 100 A); s mechanism with 3 stable positions particularily used in industrial circuitbreakers (see fig. 21). Their operating device enables: s sudden closing of contacts, regardless of how they are operated, s sudden opening of contacts, regardless of how they are operated, s opening by tripping, suddenly and even if the handle is held. Resetting must then precede reclosing, s positive break (the operating device can only be padlocked in the O position if the contacts are really open); s mechanism for high current, more sophisticated circuit-breaker. This type has a device for charging energy storage before closing and opening, thus allowing an O - FO cycle without intermediate resetting. The trip units In view of trip unit diversity, only the basic principles providing the minimum knowledge required to study overcurrent breaking, are reviewed below: s the thermal-magnetic trip units s in overload conditions, significant overheating of a particular current (or temperature in many cases) causes tripping by a thermal-mechanical element, generally a bimetal strip. - Trip unit nominal rating is defined by temperature rise conditions in asymptotic heating. The trip unit can be compensated to prevent it being affected by ambient temperature. - in high overload conditions, temperature rises develop in adiabatic heating. Tripping time thus depends on
the circuit-breaker's preliminary temperature rise. s in short-circuit conditions, as from a certain current threshold, tripping is performed instantaneously by a magnetic circuit which actuates an armature or a core. This threshold is defined on a current impulse of 200 ms. However its action time is extremely reduced (3 to 5 ms) for high currents; s electronic trip units Their prime purpose is evaluation of the current flowing through the circuitbreaker poles to take the appropriate action on a tripping device. Their advantages: s greater precision of target thresholds, s tripping curves which can be adjusted according to use, s local or remote information possibilities. The contacts LV circuit-breaker contacts are made up of conductive element zones pressurised in the same direction as their possible displacement (see fig. 22). Two physical phenomena linked to the materials used and to contact force should be noted: s contact resistance (Rc) This must be as small as possible since it conditions the ohmic power developed at the contact point which must be discharged by conduction. These temperature rises can accentuate oxidation and corrosion phenomena; to guard against them, contacts may be made of copper up to 100 A and must be made of silver at higher values. On high currents, the power produced at the contact point may exceed the power which can be dissipated. The contact zone may then be brought to its melting point. Thus in order to prevent contact welding, an heterogeneous pair of materials is normally provided, for example with tellurium or carbon placed in one of the two contact materials. The single blades contact technique is used up to In = 630A. For higher values, the multi blades contact technique is preferred.
contact striction repulsion Magnetic interaction between the radiating current lines gives rise to a contact repulsion force known as striction repulsion. Its consequences are damaging since while it lasts: s the contacts are eroded pointlessly by arcing energy, s there is a risk of welding or microwelding if the contacts close, s hot spots are created, promoting arc stagnation and thus thermionic emission; arc extinguishing conditions during its regeneration phase may thus be compromised. Note that to increase electrodynamic withstand beyond In = 630 A, striction repulsion also results in use of the multi blades contact technique. In short, choice of materials and of contact force is decisive for contact resistance, repulsion threshold and for other aspects such as their behaviour to erosion, microwelding, etc...
s
Moving contact on high currents over 15 In, the following measures must be taken: s for devices which must stay closed, reinforce electrodynamic withstand by a
contact force (Fc)
fig. 22: the LV circuit-breaker contacts are pressed in the same direction as they are moved.
self-energized compensation effect. There are several possible diagrams: s by mutual attraction: this diagram used in switches prevents opening on high currents (see fig. 23a), s by balanced repulsion device, used in circuit-breakers with high rated current (see fig. 23b). Since these circuitbreakers are the main ones, their tripping is often delayed to obtain discrimination. They must therefore have a high electrodynamic withstand, approaching the 20 In short-circuit values. s for devices needing to open and break quickly, enhance moving contact repulsion conditions in order to obtain arcing voltage as quickly as possible. A few diagrams are possible (see fig. 24). s with simple repulsion loop, s with double repulsion (often created by a double contact), s with extractor, a magnetic core pushes or pulls the moving contact. The repulsion effects can be reinforced by the use of magnetic circuits: s with effects proportional to the current square: - U-shaped swallowing circuit (see fig. 25a), - U-shaped expelling circuit (see fig. 25b), s with effects proportional to the current slope (di/dt), thus particularily effective on high currents (Icc), (see fig. 25c).
The arc chutes Their main function is to maintain arcing voltage at a suitable value and absorb the energy generated by the arc (this energy is sometimes phenomenal: if Ua = 500 V and i = 10 000 A for 2 ms, then Pa = 5 MW and Wa = 10 kJ!). The arc chute must also meet dielectric regeneration conditions sufficient to ensure permanent breaking of the current, despite mains voltage presence. The physical phenomena to be considered for breaking are no longer solely electrical: thermal phenomena (melting, sublimation, evaporation) aerodynamics and radiation also play a role in each instant's energy balances. Basically the arc chute sends the arc against an arc plate stack, arranged at right angles to the main arc column in order to (see fig. 26a): s split the arc up into the same number of elementary arcs as there are intervals, each of them thus generating
a minimum arcing voltage due to the anode/cathode phenomenon and to its elongation. Arcing voltage when splitting occurs is calculated as follows: Ua N x UAC + (L - N e) UL For example: where N = 10, L = 4 cm, e = 2 mm, UAC = 30 V and UL = 75 V/cm Ua = 200 + 150 = 350 V s store, by temperature rise or temporary arc plate melting, the energy produced under high currents in the plasma column. In actual fact, in particular situations, there is an upper current limit beyond which the arc remains in front of the arc plate stack, while continuing to exchange considerable heat with them (see fig. 26b). Although the arc is no longer split, arcing voltage has the same magnitude. The zone before chute This zone is made up of the volume separating the contact separation zone
Fr Fm
,, ,,,
a Fm i i b Fm i
opening
Fm
Fm b
opening
Fm
i Fm
i
Fr i 1/3
(i/2)
Fr Fr
c extractor
,, ,,,, ,
i c opening i
i2 i2
Fr
(i/2)
Fr
Fm
i
Fm
A 2/3 b i i
Fm
fig. 23: reinforcement of contact electrodynamic withstand
fig. 24: contact repulsion principle : a: with simple repulsion loop; b: with double repulsion (often created by a double contact); c: With extractor, a magnetic core pushes or pulls the moving contact.
,,, ,, ,
i i
fig. 25: magnetic contact repulsion devices: a: U-shaped swallowing circuit; b: U-shaped expelling circuit; c : repulsion with high di/dt.
and the beginning of the arc plates making up the arc chute. Specific precautions must be taken to: s prevent the arc stagnating on the contacts. The lower arc runner helps by moving the fixed contact arc foot underneath the chute arc plates; s promote faster and more extensive arc elongation than that caused by simple mechanical opening of the contacts. The magnetic effects, already mentioned for moving contact repulsion, will help by acting on the arcing current. In addition to this magnetic blowing, a real aerodynamic blowing will occur if the energy of the emerging arc vaporises or sublimates ablative materials by generating overpressures and gases which enhance arcing voltage evolution.
Electrical installations require the use of many circuit-breakers (at installation origin, at line cross-section changes, near certain loads,....) with highly varying performances: s rated voltages from 400 to 690 volts in three-phase; s rated currents, In, from a few amps to 3200 A according to where they are placed in the installation; s overload protection devices from 1.3 to 10 In according to the elements protected; s breaking capacities of values often less than 35 kW, but able to reach 150 kA according to installed power; Special features of LV circuitbreakers To meet all the needs of electrical distributions in industry and the service sector, a range of circuit-breakers is thus required. Circuit-breakers whose characteristics are obtained by technical solutions adapted to their functions and sizes. In this way the breaking function, made to suit each level, helps ensure the safety of the entire installation. s protection (of people and equipment), s availability of energy or continuity of service, in particular through circuitbreaker tripping discrimination.
LV uses two main discrimination types: current discrimination and time discrimination. s the former, reinforced by use of energy discrimination (refer to Cahier Technique n 167) is obtained with A category circuit-breakers as in standard IEC 947-2. These circuit-breakers have to break the fault current very quickly and considerably limit short-circuit currents. s the latter is achieved with B category circuit-breakers. These circuit-breakers, normally the main ones, have to withstand the flow of steady-state fault currents and therefore need an excellent electrodynamic withstand. Excellent short-circuit current limitation (also refer to Cahier Technique n 163) This is particularly aimed at for circuitbreakers less than or equal to 630 A. These circuit-breakers develop an arcing voltage of 600 to 900 V in small volumes. It is easier to obtain this voltage using double breaking systems (by combining the diagrams in figures 24b and 25b) and by implementing a rotating type moving contact which has the added advantage of simplifying production of one breaking unit per pole (see fig. 28).
its performances
A circuit-breaker's performances ensure its suitability for use in a given electrical installation and at a specific point in this installation.
e
i
e N e e
moving contact
fixed contact b
fig. 26: the arc plates placed in the arc chutes simplify breaking.
fig. 27: range of Merlin Gerin Low Voltage Circuit-Breakers.
It is thus possible to break 100 kA in 2.5 ms using a 250 A circuit-breaker. Excellent electrodynamic withstand This is aimed at for circuit-breakers equal to or greater than 800 A. This aim requires compensation of electromagnetic forces, which is easier to achieve with simple breaking (diagram 23b) all the more so since the wide opening (distance between contacts) of these larger devices (importance of conductive parts for high current flow) also enables a high arcing voltage (600 to 900 V) to be obtained. A 3200 A circuit-breaker thus breaks "100 kA" in 15 ms (without tripping delay) as well as withstanding 75 kA for 3 s (see fig. 29). Proven performances Circuit-breaker performances are evaluated and guaranteed by the carrying out of standardized tests (refer to IEC 947-2 and NF C 63-120 (Norme Franaise)). Thus, with respect to "breaking", tests are used to verify for example: s endurances under In, s overload endurances (e.g. under 6 In), s breaking capacities by cycles: s O-CO at Icu, the ultimate breaking capacity, s or O-CO-CO at Ics, , the service breaking capacity where Ics Icu.
,

arcing horn temporary contacts Ag based main contacts upper terminal lower terminal current transformer connected to electronic trip unit
,, ,, ,, ,

arc chute enclosure bar
upper connection
ablative part moving contact
fig. 28: breaking unit of a rotating contact LV circuit-breaker (Compact NS - Merlin Gerin).
0 . ,, , && $$ ++ ** , ,,,'' a&& ` 55 $$ 44
,, KK JJ G F E Z Y X W V U T S RI QH 33 P? 22 O= ,9 N +3 M ,,, *2 L'' 1 00 .. -,, ++ ** = ; :
Note: The publication of standard IEC 947-2, dealing with industrial LV circuitbreakers, is the subject of Cahier Technique n 150 which completes the details given above.
, ,, ,,,
,, , ,,

magnetic circuit connection trip unit
arc chute
poles shaft activated by the O-CO mechanism
insulating pole cage

fig. 29: breaking unit of a LV circuit-breaker with excellent electrodynamic withstand (Masterpact - Merlin Gerin)

7. conclusion
The future of the arc?

The electric arc continues to be an excellent means of breaking with current limitation in low voltage. Moreover, low voltage circuit-breakers have been considerably enhanced as a
result of developments in knowhow, materials and use of electronics. For many decades to come, electric circuit protection will therefore continue to require circuit-breakers with arc control.
8. bibliography
Normes
s IEC s IEC s NF
Cahiers Techniques Merlin Gerin

s Development of LV circuit-breakers to standard IEC 947-2. Cahier Technique n 150 E. BLANC s LV breaking by current limitation. Cahier Technique n 163 P. SCHUELLER s Energy based discrimination for low voltage protective devices. Cahier Technique n 167 R. MOREL - M. SERPINET
947-2: Low-voltage switchgear. Part 2: circuit-breakers. 50: International Electrotechnical vocabulary.
C 63-120 (Norme Franaise): Appareillage basse tension - 2me partie : disjoncteurs.
Ral.: Sodipe - Valence - Photo.: IPV - Grenoble Edition: DTE - Grenoble 12-94 - 2500 - Printing.: Clerc Printed in France
n 156
dependability and LV switchboards
Olivier Bouju After graduating in 1989 as an engineer from the Institut National Polytechnique de Grenoble and the Institut Administratif de l'Entreprise, Olivier Bouju joined Schneider in 1990 where he specialised in dependability studies in the Low Voltage Switchboard Department of the LV Power Equipment Division. He is currently responsible for Technical Management of Low Voltage Power Equipment.
E/CT 156, first issued, May 1997
glossary
HV, high voltage. IEC, International Electrotechnical Commission. IP, degree of protection of low voltage switchboards. LV, low voltage. MCC, Motor Control Centre, LV switchboard grouping the control and monitoring elements of a number of motors. MLVS, Main Low Voltage Switchboard. MTBF,Mean Time Between Failures. MTTR, Mean Time To Repair. MV, medium voltage. PE, protective conductor. TTA and PTTA, Type-Tested Assemblies and Partially Type-Tested Assemblies: LV switchgear and controlgear assemblies defined by standards imposing certain service conditions, construction requirements, technical characteristics and tests. UPS, Uninterruptible Power Supply. Bus (serial), a communications network over which all data elements, including those related to monitoring, are transmitted one after another. Communicating component, device such as a circuit breaker or relay that is capable of transmitting a wide range of information such as trip unit settings, currents, overloads, causes of tripping and insulation-resistance values. Intelligent switchboard, an assembly including communicating components (circuit breakers, relays, etc.) and a central unit (processing capacity), connected by a communications network or bus. It can function alone or as part of a supervision system. Protocol, set of rules ensuring cooperation between entities, generally separated by a certain distance, particularly in order to establish and maintain the orderly exchange of information between them. Switchboard Central Unit, a processing unit within the switchboard, used to organise digital information forwarded by the communicating components, automate electrical distribution functions and communicate with the installations supervision system.
dependability and LV switchboards
contents
1. Introduction 2. Switchboard functions p. 4 p. 5 p. 8 p. 9 p. 11 p. 18 p. 20 p. 20 p. 21 p. 21 p. 23 p. 24
The switchboard and its functions The switchboard's functional guarantee 3. Optimum dependability Dependability characteristics Industrial dependability concepts Required dependability levels 4. Future perspectives for switchboards Power management Power management for greater dependability The technology The "intelligent" switchboard 5. Conclusion Appendix: bibliography
1. introduction
This Cahier Technique publication deals with the dependability of commercial and industrial low voltage electrical installations. Its aim is to answer the question "which installation best satisfies our growing needs in terms of electrical power availability?". The subject is dealt with for LV switchboards and focuses on the following problems: c which switchboard functions guard against failure of the LV distribution system?
c how should they be used? c with what components? c in what power system environment (number of sources and loads, type of system earthing)? The reason for this focus is that LV switchboards are vital links in any power distribution system. This document is intended to help operators and designers of electrical installations to: v determine the points which must be considered. These points are related to
the technical choices dealt with in the sub-chapter entitled "industrial dependability concepts" . The discussion is based on reliability levels calculated on concrete cases and yields solutions in terms of equipment type. A summary is given in the subchapter entitled "required dependability levels". v realise the increasing influence of power management systems on LV switchboard dependability.
2. switchboard functions
The switchboard is a key part of any electrical installation. It incorporates devices designed to: v distribute electrical power and protect circuits, v protect persons, v control and monitor the installation. Recent developments in this control and monitoring function have made the switchboard even more vital to the installation. The dependability of the entire installation is largely determined by the dependability of the switchboard. Moreover, the lasting viability of the associated industrial or commercial activity depends on the capacity of the switchboard to keep pace with future needs. Dependability of electrical distribution means: c a very low probability of failure (reliability), c no dangerous failures (safety), c the ability to operate at any given time (availability), c fast repair (maintainability), ... throughout the entire lifetime of the installation. These notions of dependability must be taken into consideration right from the switchboard design phase. Today, dependability requires decentralised management of the installation. For instance, load shedding/reconnection and source changeover automation systems, measurement instruments and protective devices, are placed as close as possible to the application to ensure: v optimum modularity, v greater reliability (a local failure does not paralyse the entire installation), v operating flexibility with local control and monitoring possibilities at switchboard level in addition to centralised supervision. The dialogue between the various distribution levels is considerably simplified by the use of digital communications networks.
As a result of this decentralisation, part of "intelligence" is integrated in the various LV switchboards of the installation which house the main electrical components used between the transformer and the load devices (see figure 1). The result is a switchboard system including: v the Main Low Voltage Switchboard, v the switchboards specific to motor control (MCCs - Motor Control Centres), v the subdistribution switchboards, v the final distribution enclosures.
the switchboard and its functions

The implementation of the functions of a switchboard involves various aspects. c the LV installation architecture, broken down into various switchboards,
enclosures, etc. at various locations, forming the installation layout. The switchboards are further divided into a number of zones for: v components, v busbars, v connection, v auxiliaries. The minimum clearances and safety distances must be satisfied. c the functional units, providing the electrical functions needed by the user. Each unit includes the components designed to cover a given function, for example protection of a feeder or a set of feeders, motor control, incoming protection, etc. c the enclosure, providing: - protection of the electrical equipment against external influences, - protection of persons against electric shocks (direct and indirect contact).
fig. 1: intelligence is now distributed and integrated in LV switchboards. Our example shows a circuit board in an LV assembly (Digibloc board - Schneider).
v protection of electrical equipment inside enclosures against penetration by solid bodies and liquids; v protection of persons provided by: - interconnection of all metal parts (frames, enclosures, including the door) which are earthed using protective conductors (PE), - reduction of openings (ventilation, cable entries, etc.) to prevent access to live parts either directly or via tools (e.g. screwdrivers), - possible use of barriers to avoid contact with live parts when the door is open; v degrees of protection (IP) IEC 529, HD 365 and NF C 20-010 standards define the various degrees of protection of persons (against direct contacts) and equipment (see figure 2) using two numerals and two letters. Impact strength is characterised by a
separate IKx index as in European standard EN 50102; v adaptability The enclosure must be adapted both to the volume of the components to be housed and to the size of the premises and means of access. Connections are made from the top or bottom, front or rear, as required. c internal partitions. For increased dependability, the cubicles can be divided up by partitions and barriers (metal or not). The various equipment items are installed and cabled in the switchboard in such a manner that they do not interfere with each other, for example through electromagnetic fields, vibrations or arcs. Partitioning is a solution for most of these phenomena and suitable ventilation solves the associated thermal problems.
Barriers and partitions also contribute to: v protection against contact with live parts belonging to the adjacent functional units, v limitation of the probability of initiating arc faults, v protection against the passage of solid foreign bodies from one functional unit to another. The corresponding levels of dependability are evaluated further on in this document. These partitions are often related to the switchboard architecture and thus delimit the various zones intended for components, busbars, connections and auxiliaries. The separation of the various switchboard elements and functions (see figure 3) is defined in standards IEC 439-1 paragraph 7.7 and NF C 63-410.
element first characteristic numeral
numerals or letters 0 1 2 3 4 5 6
second characteristic numeral 0 1 2 3 4 5 6 7 8 additional letter (optional) A B C D H M S W
meaning for the protection of equipment against ingress of solid foreign bodies: (non-protected) diameter u 50 mm diameter u 12.5 mm diameter u 2.5 mm diameter u 1.0 mm dust-protected dust-tight against harmful ingress of water: (non-protected) vertically dripping dripping (15 tilted) spraying splashing jetting powerful jetting temporary immersion continuous immersion -
meaning for protection of persons against access to hazardous parts with: (non-protected) back of hand finger tool wire wire wire -
supplementary letter (optional)
supplementary information specific to: high voltage apparatus motion during water test stationary during water test weather conditions
against access to hazardous parts with: back of hand finger tool wire -
fig. 2: elements defining a degree of protection IP as in standards IEC 529, HD 365 and NF C 20-010.
v form 1: no separation, v form 2: separation of busbars from functional units, v form 3: same as form 2 plus separation of all functional units, but not of their terminals for external conductors, from one another, v form 4: same as form 3 plus separation of the terminals for external conductors which are an integral part of the functional unit. c internal electrical connections Consisting of conductors (busbars and cables) within the enclosure, they carry and distribute the current according to the installation diagram. v their cross-sections and number vary according to the nominal currents.
form 1 +
form 2 +
However their characteristics also depend on other parameters, for example the rated short-circuit withstand current of a switchboard, equal to the root mean square of the current that can be withstood by the switchboard for one second (see standard IEC 439-1). v their supports must in turn withstand the corresponding electrodynamic forces and thermal stresses, and also comply with minimum creepage distances. v as regards control circuits, their coexistence with power circuit is achieved by running them separately and using appropriate connections. Likewise, the auxiliaries (for form 3 separation or higher) are isolated from the other units and are thus subjected to a less restrictive environment in thermal and electromagnetic terms. c component connections The way a component is connected or installed influences availability and
maintainability. Component installation methods include fixed, withdrawable or disconnectable. Reminder: v a device is said to be fixed when tools are required to separate it from the main circuit, v a device that is withdrawable from a base or frame (for a heavy device) can be moved to a position for which an isolating distance is achieved between its upstream and downstream connecting elements, v a disconnectable device has a withdrawable upstream connection and a fixed downstream connection. Likewise, these installation methods are linked to switchboard technology which may fixed, drawout (racks) or disconnectable (see figure 4). For example: a withdrawable assembly can be either a switchboard containing fixed devices in drawout units or withdrawable devices (on base or frame) on a fixed panel.
component technology
fixed
withdrawable on base
withdrawable on carriage or frame
disconnectable
switchboard technology
form 3
form 4
fixed
+
drawout
+
elements: component disconnectable subassembly LV switchboard
fig. 3: "the forms" defined by standards IEC 439-1, EN 439-1 and NF C 63-410 delimit the various zones in a switchboard.
fig. 4: the various possible installation methods for components in an LV switchboard.
the switchboards functional guarantee

Switchboard design refers to standards governing the entire low voltage domain, and, more specifically, to standards relating to assemblies (cubicles, desks,...). Compliance with these standards is the minimum guarantee of a level of quality and dependability. Standards IEC 439-1, EN 60 439-1 and NF C 63-410 define the construction requirements, technical characteristics and tests for "type-tested" and "partially type-tested" assemblies. c assemblies manufactured in accordance with established types are known as "Type-tested assemblies" (TTA), c assemblies derived from type-tested arrangements (e.g. by calculation) are known as "Partially type-tested assemblies" (PTTA). The standards are discussed in Cahier Technique n 145 that deals with thermal studies of LV switchboards. Heat exchange must be controlled within a switchboard to avoid overheating the equipment installed inside. This requires proper ventilation and in some cases a careful choice of installed components to ensure a suitable level of reliability. Moreover these thermal studies are part of work currently conducted by the Schneider technical sections and aimed at optimising the technical characteristics of LV switchboards, particularly as concerns: c power connections (definition of a certain number of parameters as a function of currents), c short-circuit mechanical and thermal withstand described above (using computer models),
c control and monitoring installation (using studies and tests), c dependability of low voltage distribution systems through switchboards. In addition to the above work, the switchboards undergo numerous tests (see above-mentioned standards) to validate the theory and guarantee operation of the resulting assembly. These tests include verification of: c temperature-rise limits, c dielectric properties,
c short-circuit withstand strength, c continuity of the protective circuit, c clearances and creepage distances, c mechanical operation, c degree of protection. Likewise, in order to meet customer needs and ensure the durability of the required quality level, the design, industrialisation and manufacture of LV switchboards must comply with the Quality directives, methods and controls (see figure 5).
fig. 5: an LV assembly under test... "to meet customer needs and ensure the durability of the required quality level" (LV Power Equipment division, Schneider).
3. optimum dependability
Reduction in the number of failures and of the resulting shutdown times increases safety and productivity in companies. What is more, users today demand a "tailor-made" level of dependability, i.e. an installation adapted to their needs. The notion of optimisation is thus vital, meaning just the right level of dependability in order to ensure the best price. If this is to be possible, manufacturers, installers and specifiers must master the dependability parameters of their installations.
expressed (for most systems) by: ID = MTTR where represents the reliability and MTTR the maintainability. For a transformer, if 12 hours elapse between the failure and resumption of power, its unavailability is = 6 x 10-7 x 12 = 7.2 x 10-6, which is equivalent to 4 minutes of unavailability a year (i.e. 7.2 x 10-6 x number of minutes in a year). Remember that for a given installation architecture, availability is
characterised by a combination of good reliability and efficient maintenance. c dependability applied to assemblies. To calculate dependability, the failure tree method must be applied to the LV electrical distribution system studied (see Cahier Technique n 144). Analysis Let us consider the availability of electrical power of application U1, shown in figure 6.
dependability characteristics
c dependability parameters The notions involved in dependability (reliability, maintainability, availability and safety) are all linked. Three of these notions can in particular be associated by their representative quantities which are: v for reliability, the failure rate () or its reciprocal (1/), the MTBF (Mean Time Between Failures). The failure rate of a transformer is for example 6 x 10-7 h-1 which corresponds to a mean time between failures of 195 years, or to 1 device out of 195 failing on the average each year. v for maintainability, the value MTTR (Mean Time to Repair) is used. This time covers detection of the failure, the time required to supply the spare parts and the actual repair time. v for availability, the quantification depends on the combined aspects of reliability and maintainability. The opposite of availability, which is obviously unavailability (ID) is
incomer 1
T1
T2 incomer 2
m1 Q1 v1 main busbars B1 B1 m3 Q3 v3 B1 B1 m4 Q4 v4
m2 Q2 v2
B1
B1
Q5
C1 application U1 R1
fig. 6: example of an electrical distribution system.
absence of power on application U1
or
unavailability of incomer (upstream of the main busbars) or
unavailability between main busbars and the application or
unavailability due to shortcircuit on another feeder
"fixed maintainability" unavailability
Q5 repairs -> incomer de-energised HV utility failure fault on busbars B1 Q3 circuitbreaker faulty fault on cable C1
upstream disconnecting contact m3 faulty and
downstream disconnecting contact v3 faulty
load R1 faulty
no voltage on incomer 1
no voltage on incomer 2 or or
or
MV supply failure
circuit-breaker Q1 faulty
MV supply failure
cirucit-breaker Q2 faulty
upstream disconnecting contact m1 faulty
upstream disconnecting contact m2 faulty
short-circuit on circuit-breaker Q4 short-circuit short-circuit on upstream on disconnecting circuit-breaker contact m4 Q5
transformer T1 faulty
transformer T2 faulty
fig. 7: failure tree associated with the diagram in figure 6.
The undesirable event at the top of the failure tree is thus absence of power on U1. This event is broken down into four modules as shown in the chart in figure 7. c unavailability of incomer. Each incomer can alone supply the entire LV distribution system on which the application depends. The two medium voltage (MV) incomers are assumed to be taken from two different substations, which virtually reduces the common failure mode to the unavailability of HV (high voltage transmission). The circuit-breaker failure modes considered for calculation of incomer unavailability are: v spurious tripping, v refusal to close, v internal short-circuit, v temperature rise. The failures of HV system components, MV incomers, transformers and disconnecting contacts have been considered together. c unavailability between the main busbars B1 and the application. This sums the unavailabilities of the elements encountered from the main busbars to point U1. Each failure is broken down as finely as possible and results in different repair times. For example, for the busbars: v loosening of the busbar supports due to strong vibrations may cause the bars to break when they are subjected to a high electrodynamic force. The resulting repair time is several hours (part replacement). v an object falling on these bars when energised, although highly unlikely given the construction arrangements chosen (form, IP,...), often results in arcing and in a repair time of several working days. c unavailability due to short-circuit on another feeder The clearing of a short-circuit occurring upstream of the first protective device on a feeder parallel to the feeder considered results in de-energisation of all the feeders.
We thus have to add up all the shortcircuit probabilities by descending on each parallel feeder up to the first protective device. Downstream, a short-circuit affecting U1 is possible only if there is combination of a short-circuit and failure of a protective device to react, the combined probability of which is negligible. c "fixed maintainability" unavailability. Fixed maintainability is the term used to indicate that the repair time depends on the installation method (fixed or withdrawable) and affects use of the other feeders. Examples (see figure 6): application U1 is affected by repair of Q5 which, as it is fixed, requires shutdown of the incoming supply, whereas repair of Q4, withdrawable, can be carried out with the busbars energised and thus without affecting application U1. The results The following results are those corresponding to the usual reliability and MTTR values encountered for the various system components. Unavailability of the load is 6.4 x 10-5, i.e. 33 minutes a year. An examination of the relative importance of the different aspects gives the following breakdown of unavailability: v incomer 45% v between busbars and application 51% of which: - cable and load 32% - rest upstream 19% v short-circuit on another feeder 1% v fixed maintainability 3% The various points to be examined are derived from this analysis and will be dealt with in the next chapter.
you do not want to end up "pushing a Porsch", the importance of the various technical choices must be evaluated with regard to dependability. These choices include: c the diagram (incomer, final application, system earthing arrangement), c the connections, c the electric arcs, c the switchboard options (form, connection, fixed or withdrawable components, IP...), c the motor feeder units, c the control and monitoring auxiliaries. Dependability in relation to the diagram Two elements are of critical importance to dependability: v the incoming diagram, v the final applications. A third element, the system earthing arrangement, also has great influence. c the incoming diagram As availability of the incomer affects all applications, whether or not they are critical, it is important, if at all possible, to choose an incoming configuration in keeping with the downstream need. The chosen solution will depend on the environment studied. For example: v in isolated regions, it may be hard to obtain an MV line with good availability and even harder to obtain two separate MV lines. In this case the study must consider independent energy production such as by an engine generator set. v some sectors of industry (chemistry, petrochemistry, paper-making) generate energy (often in the form of steam) through their manufacturing process which they use to drive turbogenerators. The public distribution system is then used as a backup source. NB: if, despite this, the availability of the incomer is insufficient, a UPS (Uninterruptible Power Supply) must be placed as close as possible to the critical applications.
industrial dependability concepts

As defined above, the installation must be designed to meet the customers specific requirements. In all systems, just one small element can often jeopardise overall dependability. So if
v calculation of unavailability due to the incomer On the example in figure 8 (2 parallel incomers, 20 motor feeders), unavailability of the application is roughly 1/2 hour a year, 50% of which is due to failure of the incomer. From this we conclude that unavailability of the incomer, although not always preponderant, may nevertheless account for a large part of total unavailability. We shall see later on that the incomer unavailability percentage ranges between 7% and 90% according to the measures taken to ensure reliability of the rest of the system. The incomer has two main critical points, namely: - the high voltage transmission line, - the medium voltage line; Transformers, circuit-breakers and disconnecting contacts are 100 to 1000 times more reliable than these two sources of failure. How can the dependability of the incomer function be improved? There are many possible solutions, and the context is a decisive factor. Greater reliability can be obtained by concentrating on the following points: v redundant incomers Two medium voltage lines from two different source substations, used in parallel, solve the problem of unavailability of the medium voltage lines. The unavailability of the incomer function is now virtually reduced to that of the high voltage system alone which is roughly 17 minutes a year, compared to 10 hours for the MV system. Availability can also be increased by adding one or more generator sets (see Cahier Technique n 148 High availability electrical power distribution). v splitting into priority and non-priority feeders The search for increased availability of electrical power nearly always results (depending on installation size) in dividing applications into two types: - priority, - non-priority. In the event of an overload or failure of the main source, non-priority loads are then shed, while priority loads continue
a) diagram
incomers
20 feeders
load M b) causes of unavailibility on a feeder M M M c) unavailability on a feeder as a function of the system earthing arrangement unavailability (hrs/yr) 50% 1
30%
20%
1/2
0 failure of incomer, of which: s 98% due to public HV failures, s 2% due to MV failures, s roughly due 0% to circuit-breakers. failure of LV distribution and of control devices. failure of the final load devices (cables and motors). TT TN IT
type of system
NB: for IT systems, unavailability is calculed considering repair to be compulsory, on the first fault.
fig. 8: unavailability of an incomer may account for a large part of total unavailability, in this case 50%.
to run on a secondary source (second MV incomer, generator set,...). v source changeover systems If a failure occurs, circuits can be transferred to backup sources not used in normal operation or to the sources of non-priority feeders, with load shedding of the latter. Three types of changeover systems are possible: -synchronous The main source and the replacement source are or have the possibility of synchronising, thus ensuring changeover without loss of load supply. This process is used in installations requiring a high level of dependability. - delayed This is the most common type of source changeover system. With transfer times ranging from 0.4 to 30 seconds, its use is widespread for industrial and commercial applications. - pseudo-synchronous A fast-acting switching device (60 to 300 ms) is used for the source changeover. This system is found, for example, in the following sectors: - chemistry, - petrochemistry, - thermal power plants. c loads Unavailability due to the load devices is illustrated in the diagram in figure 8 and concerns for instance the motors and the cables supplying them from the switchboard. Reliability calculations show for example that when using a motor M, 30% of its down time due to failures is caused either by the cable or by the actual motor. It is thus necessary to clearly define the technical characteristics of the loads as regards conditions of use, as well as the maintenance procedures intended to prevent failures. Most electrical failures in motors are due to phase/earth faults occurring on motor startup. Insulation monitoring before starting a motor, particularly using the Vigilohm SM 20 developed by Schneider, enables:
v preventive maintenance to be programmed, v irreversible motor damage to be avoided. c system earthing The three system earthing arrangements are (see figure 9): v TT system (earthed neutral and earthed protective conductors), v TN system (earthed neutral and protective conductors connected to neutral), v IT system (unearthed neutral and earthed protective conductors). The system earthing arrangement affects availability and maintainability in that the circuit must be broken on a first fault for TN and TT systems but not for IT systems. In addition, the magnitude of the earth fault current depends on the system earthing arrangement and determines the extent of damage caused to the installation and in particular to the loads. The results of a reliability study are shown on the histogram in figure 8. The IT system, with an automatic system for fast locating of the first fault, is the one offering the best availability, as it ensures that: v operation is not interrupted (continuity of the production cycle in progress), v the fault can be repaired when the installation is not in operation, v servicing can be prepared during production, resulting in increased maintainability. The IT system is recommended in the following cases:
v presence of loads sensitive to high fault currents, v high risk of fire, v installations with generator sets (to prevent damage to the generator by an internal fault), v need for a high level of dependability (availability + safety), for example in operating rooms in hospitals. NB: in the IT system, the probability of de-energising due to a second fault (if this fault occurs before the first fault has been located and cleared) is less than in the TN and TT system as the simultaneous presence of the first and second faults is necessary on different phases. We saw earlier that the system earthing arrangement must be selected with great care. Once this choice has been made, the equipment (switchboard and components) can be chosen, and a certain uniformity sought in the reliability of the different links in the chain making up final unavailability. Dependability and connections A switchboard is made up of a large number of connections and it is therefore important to consider the failures they cause. A connection fails when it ceases to convey the electrical power for which it was designed. A local temperature rise then occurs which may cause irremediable damage to the device and/ or the cables. The importance of good connections is illustrated by the example of an installation with two separate incomers supplying 20 feeders.
system action during an insulation fault
TT immediate de-energising
TN immediate de-energising
IT
s continuation of operation s fault tracking s preparation before de-energising several dozen milliamps (1st fault)
magnitude of fault current (determines damage to installation)
several dozen amps
several kiloamps (short-circuit current)
fig. 9: choice of system earthing arrangement directly affects the dependability and reliability of the installation.
The results of the reliability study (see figure 10) show that 88% of total unavailability is due to various failures (incomer, components,...) and 12% to connections. A distinction should be made between factory connections and those made on site, as statistics show that the latter are more prone to failure. In practice, dependability can be considerably increased by: c properly sized contact surfaces (overlapping), c proper surface finish (flat and clean), c a tightening torque suited to the materials. Dependability and arcing c unavailability due to arcing A number of events can result in the creation of arcs in the switchboard, for example intrusion of small animals (rodents or reptiles), objects forgotten in the switchboard during maintenance work, a temperature rise or deposition of conducting dust. The damage caused by electric arcs is frequently serious and leads to shutdown times of up to several hundred hours for an ordinary switchboard, i.e. 11% of its total
unavailability (see figure 10). In comparison, for an "improved" switchboard, this percentage is negligible as its shutdowns are limited to the time required to put the distribution system back into working order (cable tightening, cleaning of carbonised surfaces...), i.e. roughly one hour. To prevent this unavailability, the following three points should be concentrated on: v risk of arc occurrence, v arcing time, v propagation of electric arcs in the switchboard. These actions aim at reducing both repair times and the extent of the damage caused by arcs. c preventing arcing It is better to prevent a problem than to cure it, in other words to take action on the cause of electric arcs. Note that: v arcing due to dielectric breakdown does not occur if: - materials are properly chosen, - creepage distances and clearances are complied with. v introduction of objects or foreign bodies, including conducting dust, and
incomers
20 feeders
12% of unavailability due to connections. 88% of unavailability due to: MV incomer, transformer, components...
11% of unavailability due to arcing. 89% of unavailability due to various failures.
fig. 10: unavailabilities due to arcing and connections account for roughly 20% of causes of system unavailability.
intrusion of small animals, are the cause of numerous electric arcs in LV cubicles. To prevent arcing, considerable care must be taken with enclosure design: - form, - choice of IP, - addition of filter... v when breaking occurs (on a shortcircuit or overload), pressurised ionised gases are given off by the protective device and may cause arcing, for example on a nearby busbar. This risk can be eliminated by a carefully designed architecture and/or barriers. v a faulty connection can often result in creation of an arc. To avoid this, connections must be correctly tightened (see section on "dependability and connections"). c limiting the arcing time Damage caused by arcing can be limited by minimising the duration of the arc. Possible solutions are: v setting the short-time delays (shortcircuit protection) to the minimum value that will still provide discrimination. These short delays, designed to implement time discrimination, delay circuit-breaker tripping on a shortcircuit, thus prolonging the arcing time. Note that when zone selective interlocking can be implemented, it is the best solution as it allows absolute discrimination with minimum delays for all distribution stages (see Cahier Technique n 18). v using limiting devices which quickly break short-circuit currents, thus limiting the fault current. Arcing time is thus reduced and thermal effects limited. v choosing a protective device that takes past transient short-circuits into account The peculiarity of the arc is that it is a somewhat furtive phenomena, for two reasons: - due to switchboard layouts, an arc is quickly extinguished. However the ionised gases that it generates may cause restrikes on other live parts. A number of extinguishing and restrike sequences are therefore possible. - moreover, the impedance of the arc varies according to the speed at which it moves and the obstacles that it comes across.
However, each time arcing occurs, the equipment is subjected to a number of stresses which can be cumulated. The solution to the problem is to provide protection systems which integrate the fault over time, i.e. when a fault appears and disappears (or drops below the trip threshold before the protective device trips), this time and current information must be stored in the protective device to obtain tripping if the fault or brief high current values rapidly reoccur. Thus LV circuitbreakers can be designed to store transient short-circuit information in memory and only gradually return to their initial tripping characteristics (see figure 11). c preventing propagation in the switchboard The laws of physics cause the arc to move quickly away from the source. To limit its consequences, the arc must not be allowed to spread through the entire switchboard. It is essential to control the arc throughout its duration by:
v partitioning the various switchboard zones; insulated bushings and partitions prevent the arc and its ionised gases from spreading; v enhancing arc extinction, for instance by implementing - insulation shrouds around the busbars, - busbar geometries that lengthen the arc. Dependability and the switchboard "options" The form, type of connection (front or rear), device installation method (fixed or withdrawable) and the degree of protection are all possible options when manufacturing and/or purchasing an LV switchboard. The example in figure 12 shows the effect of these choices on availability at feeder level. c form (see figure 3) Consider form 1 with "unsealed openings" compared to form 2 with "cable access openings sealed".
The abbreviated expression "cable access openings sealed" means that the user has passed the cables through a bottom plate equipped with a cable bushing. NB: this arrangement is used for form 2 and above. This example clearly shows that a wise choice of form increases availability (see figure 12), as it affects: - likelihood of fault occurrence (rodent intrusion impossible), - arc propagation (presence of partitions). For good availability, LV switchboards should be partitioned (form 3), including the terminals for external conductor connections (form 4), since, as already pointed out, these connections are the cause of most faults (see paragraph on "dependability and connections"). c front or rear connection The space reserved for electrical equipment when designing premises
unavailability (hrs/yr) 4 3 2 1 0 front conn. rear conn. front conn. rear conn.
form 1
form 2 cable access sealed
unavailability dur to: incomers. fixed components. withdrawable components.
fig. 11: a type B Masterpact circuit-breaker (delayed) equipped with an ST608 control unit temporarily stores short-circuit information in memory (Schneider).
fig. 12: unavailability times depend on switchboard technology and particularly on its connection type (the chart corresponds to the diagram in figure 10).
frequently determines the type of connections used. This constraint has a certain effect on availability. Access to a switchboard with front connections only is often difficult, resulting in lengthy repair times compared with switchboards offering dual accessibility (see figure 13). Note that the unavailability of a switchboard with front connections is even higher if fixed components are used that require tools for dismantling. To increase maintainability of a switchboard with front connections, designed to stand against a wall, a small servicing clearance should be provided at the rear. c fixed or withdrawable Availability can be improved by choosing withdrawable devices (see figure 12). In this way, maintenance is faster and does not affect adjacent feeders. Since withdrawal takes place off-load (with the circuit open) but with power on, breaking is not necessary upstream and interruption of supply to the other feeders parallel to it is not required.
However the withdrawable option may not offer any great advantage for installations subject to a high levels of unavailability elsewhere (unreliable source, single incomer presenting risks,...) or when excellent maintainability does not affect other feeders. However, in the case of a form 2 switchboard with front connections, the advantage of using "withdrawable" circuit-breakers is clear (see figure 12). In this instance unavailability is divided by 3 compared with the "fixed" solution. c degree of protection (see figure 2) Only the first two characteristic digits of the IP (ingress of solid bodies and liquids) are examined in this section. The first numeral gives the maximum size of objects or particles likely to enter the switchboard, thus limiting the size of the access points to live parts. This numeral (1 to 6) increases as size decreases. The second numeral concerns liquids and describes protection obtained by: v canopies, covers or baffles protecting against vertical and/or horizontal liquid splashing and jetting,
v seals and suitable devices protecting the enclosure even in the event of immersion. In conclusion, the higher the first two characteristic numerals of the IP index, the better the protection. However, all electrical devices produce heat and most of them have a thermal limit. Excessive imperviousness is contrary to proper switchboard ventilation and may thus affect operation of the components. Heat extraction and/or a suitable choice of devices is thus necessary. The degree of severity of the environment and the qualification of switchboard operators determine the choice of degree of protection. The necessary protection levels, for each type of premises, are reviewed in figure 14. Dependability and the drawout motor feeder unit MCC drawout type switchboards are often used in process industries (see figure 15). Good continuity of service is normally required for motor control. Drawout units allows quick, easy maintenance: the faulty feeder is immediately replaced by an identical device while power continues to be supplied to the switchboard. A drawout unit corresponding to a motor feeder can be composed (see figure 16) of a fuse, contactor and thermal relay or of a circuit-breaker, contactor and thermal relay. In terms of availability, both configurations are more or less equivalent in normal operation, but differ considerably should the contactor fail. In actual fact some 20% of feeder failures are due to contactors (contacts sticking), with the added disadvantage of having to remove the contactor from the drawout unit. The power circuit then has to be opened, which is possible with the circuit-breaker/contactor combination by opening the circuitbreaker. In the other case (fuse/ contactor combination) power must be switched off upstream, thus making all the other motor feeders unavailable.
small servicing clearance
servicing clearance
a) switchboard with front connections, standing against the wall
b) switchboard with front connections and small servicing clearance
c) switchboard with rear connections and the necessary servicing clearance
fig. 13: a good compromise between maintainability and floor space can be obtained using a switchboard with front connections and a small servicing clearance at the rear.
sectors of use domestic premises technical premises
examples bedroom washroom electrical service air conditioning washer refrigeration chamber fuel storage coal storage boiler plant repair shop washing area offices gymnasium large kitchen alcohol warehouse hen-house fodder storage electroplating shop paperboard manufacturing quarry art gallery hardware shop bakery cabine tmaker
IP degree 20 27 00 24 33 20 50 61 20 25 20 21 35 23 45 60 03 33 65 20 33 50 60
boiler plants and associated premises (power > 70 kW) garages and parking areas (area > 100 m2) buildings for collective use
The consequences of this procedure can be demonstrated for a diagram with 20 motor feeders supplied by 2 separate MV incomers, an example illustrated by the results histogram (see figure 16). Two contactor operating rates can be identified (low and high). The likelihood of failure of a drawout unit are linked to the operating rate of the contactors. It is thus preferable to use a circuit-breaker rather than a fuse as a protective device if intensive use is made of contactors (operating rate and also utilisation categories of loads AC3, AC4, operating voltage...). Dependability and the control and monitoring auxiliaries Using the same example (see figure 16), the influence of the control and monitoring auxiliaries on total availability can be determined. Their associated failures relate to relays, connections or to their power supply. The individual wiring of nonstandardised auxiliaries is a lengthy process and subject to errors by fitters, resulting in potential failures.
farms
industry
commercial and associated premises
fig. 14: examples of minimum degrees of protection (as in NF C 15-100 and practical guide UTE C 15-103).
unavailability (hrs/yr) 1
1/2
high rate
low rate
high rate
low rate
non-standardised auxiliaries
standardised auxiliaries
fuse/contactor combination circuit-breaker/contactor combination auxiliaries
fig. 15: detailed view of an MCC type LV switchboard with drawout units (MB400 model Schneider).
fig. 16: comparison of levels of unavailability for a 20-feeder drawout type switchboard depending on type of components and their operating rate.
To avoid this, Schneider offers standardised products for auxiliary functions (Digibloc, Dialpact,...). These are boards or control and monitoring modules connected by power distribution blocks or by standardised digital links. These elements centralise information and can be used to implement a wide variety of control schemes. Furthermore, these schemes can be easily modified by setting board parameters or by associating new modules, with the following advantages: c time savings on implementation, c increased reliability by eliminating wiring errors, c repair time reduced to the time required to replace the board or module, c open-ended solution. The results of the reliability calculations on the histogram show that these standardisation efforts increase the availability of control auxiliaries depending on the operating rate (from 30% at low rate to 60% at high rate).
required dependability levels

A large number of technical options are available for LV installations, all offering different dependability levels. The right choices depend on the application and on the choices made at other levels. For example use of a form 4 switchboard is advantageous provided that the other major sources of failures on the installation have been overcome. The right approach when designing an LV installation is not therefore to choose and install at random a range of effective, reliable devices in the hope of gaining maximum "peace of mind". In actual fact, each application or sector using LV electrical power requires an appropriate level of dependability,
depending on operating imperatives (see figure 17): c the commercial and service sector includes both of small shops and schools..., as well as supermarkets, shopping centres, large banks, office blocks, hospitals. c industry comprises all types of factories (automobile, aeronautic, textile,...) and has special needs in terms of distribution (power system protection and architecture) and processes (motor control, control system), which are vital in continuous production applications such as petrochemistry, cement works, food processing... In what way do these sectors represent different needs? Accidents such as BHOPAL (December 1984), CHERNOBYL (April 1986) and PASADENA (October 1989) are evidence of the high risks run by people and the environment. Hence the unfailing question "is it dependable?". In fact this question is meaningless. As the possibility of failure is always present, however small, the right question is rather: "is it dependable enough?". For all sectors this means choosing an acceptable level of probability of dangerous failure (in safety terms) and of dependability (in economic terms): c in telecommunications, France Telecom has a probability of unavailability of 1 hr/century for telephone exchanges ( < 10-6 h-1). c in air transport, two dependability conditions are laid down to ensure that: v all "overall catastrophic" failures are extremely unlikely ( <10-9 h-1), v all "critical" failures are extremely rare ( <10-6 h-1). This figure can be compared with the likelihood ( <10-6) of a human being dying within the next hour. c in banks, power failures result in lost entries and recording of erroneous operations. The costs involved in
tracking and recovering these errors provide the necessary reference elements. c in hospitals, safety of persons can be immediately affected by a failure. Operating theatres and reanimation wards are especially designed to ensure a high level of dependability. c in industry, failures also considerably affect continuity of service. An article written by Y. Lafarge and published in "Le Monde" quotes two examples: v for BSN (Danone), a 10 minute shutdown causes a production loss of 20,000 items, v for Peugeot, out of a production of 1,650 vehicles a day, a one hour computer failure means 100 cars are not manufactured, i.e. a loss of profit of 4 million francs. It is thus easy to understand the importance attached by firms to availability of the electrical power on which the entire activity of the company depends. Thus, in the commercial and service sector and industry alike, failures may have economic consequences, cause damage or be a source of major risks. All of which may affect our everyday life in which good service in 99% of cases ( =10-2) would mean: c more than 140 new-born babies would accidentally be dropped by doctors and nurses each year; c no electricity or water for several dozen hours each year; c your telephone and television would be out of order for more than 10 minutes a week; c 400 letters an hour would never reach their destination. These evocative images clearly show the consequences of choice of dependability level. The table in figure 17, although not complete, gives the most important choices for an LV installation and for the various sectors of activity. To specify these choices, the need must be defined and the dependability concepts examined in the previous chapter must be implemented.
sectors of activity commercial and service sectors shops the problem to be solved: types of incoming diagrams
hospitals
industry workshops
plants
process manufacturing
operating imperatives
numerous mobile and portable loads, frequent changes to distribution system, supply by a public power system.
continuity of service for certain sectors, risk of fire, presence of generator sets.
uncertain earth circuits (worksites), supply by a public power system.
continuity of service for certain sectors, presence of backup generator sets.
continuity of service for most of the operation. risk of serious damage by insulation faults (motors, automation). risk of fire
recommended system earthing arrangements
TT
IT
TT numerous auxiliaries (machine-tools), loads with low insulation resistance.
IT
IT
atmosphere and/or loads corresponding to high risk of insulation faults.
TN solutions implemented: component type fixed or disconnectable or withdrawable switchboard type fixed fixed or disconnectable
TN sub-system fixed
form to F1 degree of protection IP (first two numerals) 2 motor control components low rate high rate technology of control and monitoring auxiliaries non-standardised
(individual wiring)
to
or withdrawable fixed with drawout units or with disconnectable subassemblies or with drawout units F4 F4 F4 to to to F2 F2 F2 5 5 5 to to to 2 3 3
F4
fuse/contactor combination circuit-breaker/contactor combination standardised

(boards, modules and connections)
non-standardised
standardised
fig. 17: the sectors of activity and operating imperatives determine the system earthing arrangement and the solutions implemented depend on the form used and on the degrees of protection required.
4. future perspectives for switchboards
Modern switchboard technology has been and continues to be greatly influenced by the development of power management systems. We must therefore look into the implications of power management on dependability. The reader will see that power management provides the installation with greater dependability by integrating information processing electronics in the LV switchboard which thus becomes "intelligent".
Power management
Power management is already used in Building Management systems, which have gradually replaced more centralised systems in industrial, commercial and even domestic applications to supervise, monitor and control the following standard functions and facilities: c heating and air conditioning, c fire protection, c intrusion protection, c access and worktime control, c lifts, lighting..., c energy tariff management. Power management is becoming more and more decentralised for reasons of availability, user convenience and modularity (already mentioned in chapter 1). Over and above the traditional functions performed by electrical equipment (protection, automation, transfer of loads to backup sources), a power management system provides a number of functions in the electrical control and monitoring field. To cite a few examples: c automatic, progressive resumption of feeder supply after a fault, c alignment of consumption to energy supply possibilities at a specific time (load shedding and reconnection, generator startup and shutdown), c optimisation of sources according to consumption to derive maximum advantage from electricity contracts with differentiated tariffs,
c optimisation of capacitor bank operation, c contribution to discrimination (coordination of protective devices). It also enables: c local and remote control and monitoring (indications, alarms, controls and setting modifications,...), c supervision (graphic representation of system status, event logging and installation control). The need for power management increases with the need for availability and, more generally, for dependability. Power management systems have been made possible by the introduction and widespread use of microprocessor technology which at the same time provides an opening towards greater, distributed "intelligence".
v complete information eliminates the risk of error in system management. c maintainability Reliability can be obtained by rigorous design, but product dependability also requires a high level of maintainability. There are two types of maintenance: v preventive maintenance is designed to anticipate problems and thus to limit the risk of shutdown due to a fault (it prevents the fault from occurring), v curative maintenance is designed to quickly restore the system to its "operating" condition (it locates the fault). Preventive maintenance takes priority over curative maintenance as it avoids problems during operation. However it requires sound knowledge of the products at all stages and the capacity to detect potential failures. Experiments and tests on equipment can provide this knowledge and a power management system can use it in an optimum manner: v a preventive maintenance system is established to reduce the number of failures, using the following: - operation counters, - insulation resistance measurement devices... v a curative maintenance system locates the fault in the event of a failure. v two other systems, remote maintenance and/or remote diagnostics, considerably enhance switchboard operation: - remote maintenance ensures surveillance without the need for a control room and a permanent maintenance team on site. Remote transfer of information on failures makes frequent inspections of the various electricity supply points unnecessary. - remote diagnostics enable troubleshooting to be conducted on the basis of quantifiable parameters transmitted via a telecommunications system. The reduction in maintenance time is obvious, particularly when
power management for greater dependability

A power management system relies on two principles when a failure occurs: c the electrical distribution system can remain as it is and is not at risk by failure of a management module. This is simplified by the use of bistable power control devices such as switches, impulse relays and circuit-breakers. c the protection, control and monitoring systems continue to be independently activated, thus making operation in crippled mode possible. This principle ensures the prime objective of dependability even though certain functions of convenience are temporarily lost. Thus even if the supervision system fails, protection functions will continue to fulfil their task and the switchboard central unit will remain operational. Moreover, power management reinforces dependability of the LV installations in terms of: c reliability v the power management system reduces the major risk of failure represented by human intervention,
outside suppliers are responsible for management and servicing of the installation. Remote diagnostics give them the best chance of repairing the failure on their first visit to the site. c in terms of availability Availability is naturally the result both of reliability and maintainability, as well as: v prevention of overloads with the solution of load shedding and reconnection to prevent tripping, v management of sources (switching, coupling and startup of generator sets), v discrimination of the various protection levels which, as explained above, has an important role in installation availability. c in terms of safety v safety of persons is guaranteed by reflex protective devices (placed as close as possible to the fault) which, although part of the management system, can function independently if a fault occurs. v maintenance operations are fewer and can often be scheduled, allowing personnel to work under less stress. v operating staff are guaranteed additional protection by indication of device status in maintenance areas, and by warning of potential failures.
electronic modules or by using components with increased reliability. c buses are responsible for the development of decentralised intelligent systems and form the communication backbone. The serial links making up the buses enable the transfer of data to many points via a single cable (coaxial or twisted pair). Their reliability has recently been upgraded and it is now possible to isolate them from external disturbances of the electromagnetic type and by using protocols including monitoring of information exchanges. This subject is developed in Cahier Technique n 147 "Introduction to digital communications networks". c system dependability also depends on that of the software controlling the
system. In this case, rather than a revolution, we witness a systematic race for rigour at all levels, from design to commissioning (specification and development methods, special tools, highly sophisticated verification and test procedures).
the "intelligent" switchboard

The "intelligent" switchboard includes a large part of the power management system (see figure 18), in particular: v the "intelligent" electrotechnical components, v specific systems (e.g. insulation monitoring), v the switchboard central unit
supervision
MV
UPS
generator set (GE)
LV switchboard central unit
the technology
Control and monitoring "intelligence" must be organised with sufficient care to ensure a good level of dependability. It particularly calls for implementation of: v high-performance electronics, v communication networks using reliable buses, v software of recognised reliability, for overall control. c electronic components and circuits are today increasingly reliable, driven by developments in the aerospace, military, nuclear and general public sectors. The reliability levels are easily controlled, since the statistical reliability laws associated with components are perfectly applicable and reliability calculations for assemblies well controlled. Critical points are backed up by redundancy of all or some parts of the
- intelligent protection and control/monitoring components - classical components (via Input/Output modules)
insulation monitoring system
reactive energy compensation system intelligent switchboard
fig. 18: general diagram showing control and monitoring of an electrical installation and its links (BUS) supervision.
v the digital communications buses. Use of microprocessors means that intelligence is distributed right through to component level (circuit-breakers, switches,...). In addition to their basic function, they process a variety of information and communicate with the switchboard central unit, thus ensuring: c "sequencing" of actions (logic and time sequencing), c capacity to calculate and process many pieces of digital information sent by these devices, sensors and specific systems, c remote transmission by serial bus enabling the control and monitoring system to communicate with the operator and/or the supervision system,
c control and monitoring, both local and remote, as well as supervision (orders are transmitted by bus). The switchboard can now be said to be "intelligent": the intelligence integrated in the power management system will depend on the degree of complexity of the installation to be managed. The distribution of electrical power in small commercial applications may only require the display of measurements and status information on the front panel of the switchboard, whereas in large buildings, remote control functions are required (lighting, source changeover,...). Power management is at present implemented in MV and LV by means
of various components. These components, more and more standardised and convenient to use by electricians, will be available in increasingly wide ranges. The various components in the intelligent switchboard are designed to work together. The consistency, both in terms of hardware and software, guarantees easy implementation and use. An "intelligent" switchboard with appropriate overall design and made up of consistent, carefully designed and manufactured products, opens the way to efficient power management and the mastering of electrical power will greater dependability.
5. conclusion
The distribution of electrical power must meet increasing requirements in terms of: c dependability, c upgradability, c user friendliness. Designers thus aim at producing installations which are "intelligent", independent, communicating, modular, reliable and easy to service. These criteria can all be achieved by decentralisation. The basic functions (protection and control) are performed as close as possible to the application, and only supervision has a "central" position, playing a vital role in distribution management as regards the man/system relationship. Decentralisation is a design feature of each product, both in their connections (defined between them) and in their overall architecture. All these items (high power components, control and monitoring devices and electrical connections) are integrated in the LV switchboard. Its role is thus vital for distribution as a whole, given that it has to guarantee overall dependability. The following points should be borne in mind:
c the incoming diagram and the reliability of the final loads are the points which may most handicap dependability, c the system earthing arrangement affects availability of final loads and must therefore be chosen carefully, c connections seriously affect switchboard reliability, thus calling for careful design and implementation, c switchboard technology, form, degree of protection, connection,... must be adapted to the environment in which the equipment is installed (degree of pollution of premises, qualifications of operators,...), c withdrawable components are used when they provide the added dependability required, c drawout motor feeder units are particularly used in process industries for the flexibility and increased availability that they provide, c auxiliaries with standardised connections and implementation guarantee the reliability of installation control and monitoring. Dependability is everybodys job, including that of the designer (the right choices from the start), the installer (implementation in accordance with the
manufacturers recommendations and proper practices) and the maintenance engineers (surveillance and preventive maintenance of critical points). This Cahier Technique shows how dependability objectives can be achieved and how, by choosing the right options, particularly in terms of technology, the required level of dependability can be obtained. The "intelligent" LV switchboard, associated with power management, meets the criteria of dependability and user convenience particularly well, providing a solution for both present and future needs. The degree of built-in "intelligence" required depends on the complexity of the installation. This intelligent switchboard, designed to ensure maximum standardisation, integrates power, control and monitoring and communication via buses. Should changes be made to the distribution system, switchboard modularity and simple parameter resetting of the control and monitoring system ensure easy upgrading . There is no need to redo studies and tests for each application as the product has already been thoroughly tested.
appendix: bibliography
Schneider Cahiers Techniques c Analyse des rseaux triphass en rgime perturb laide des composantes symtriques, Cahier Technique n 18 B. DE METZ NOBLAT c Mthode de dveloppement dun logiciel de sret, Cahier Technique, n 117 A. JOURDIL, R. GALERA c Introduction to dependability design, Cahier Technique n 144 P. BONNEFOI c Etude thermique des tableaux Cahier Technique n 145 C. KILINDJIAN c Initiation aux rseaux de communication numriques, Cahier Technique n 147 E. KOENIG c High availability electrical power distribution, Cahier Technique n 148 A. LONCHAMPT, G. GATINE
Standards c NF C 12-101: protection of workers in buildings implementing electric currents. c NF C 15-100: rules for LV electrical installations. c IEC 529: classification of degrees of protection provided by enclosures.(NF C 20-010; NF C 20-011; HD 365); c NF C 20-030: electric shock protection, safety rules. c NF C 20-040: creepage distances and clearances in air. c IEC 439-1: low-voltage switchgear and controlgear assemblies.
Various publications c Les automates programmables sontils plus fiables que les relais? Revue J3E - October 1990 F. SAGOT c Experience in critical software development, IEEE Fault Tolerant Computing Symposium, 26-28 June, 1990. Newcastle C. SAYET, E. PILAUD (Schneider) c Risque et scurit dans le domaine du transport, Revue Maintenance - November 1990 J-C LIGERON
Real.: Illustration Technique - Lyon Edition: DTE - Grenoble 05-97 - 1500 - Printing: Clerc
n158
calculation of short-circuit currents
This document was prepared by: Roland Calvas, Benoit De Metz Noblat, Andr Ducluzaux, Georges Thomasset.
E/CT 158 first issued, march 1995
glossary
Abbreviations BC breaking capacity. MLVS main low voltage switchboard. Symbols A cross-sectional area of conductors. angle between the initiation of a fault and zero voltage. voltage factor. power factor. reactance (%) of rotating machines. electromotive force (maximum value). phase angle (current with respect to voltage). instantaneous current. alternating sinusoidal component of the instantaneous current. aperiodic component of the instantaneous current. maximum current value (first peak of the fault current). maximum rms current. short-circuit breaking current (IEC 909). steady-state short-circuit current (Isc3 = three-phase, Isc2 = phase-to-phase, etc.). steady-state short-circuit current (IEC 909). initial short-circuit current (IEC 909). rated current of a generator. service current. factor depending on the saturation inductance of a generator. constants (tables and graphs). equivalent resistance of the upstream network. line resistance per unit length. transformer kVA rating. short-circuit power. minimum dead time for short-circuit development time, often equal to the time delay of a circuit breaker. instantaneous voltage. transformer short-circuit voltage in %. network phase-to-phase voltage with no load. network rated voltage with load. equivalent reactance of the upstream network. line reactance per unit length. positive-sequence, negative-sequence and zero-sequence impedances of a network or an element. link impedance. network upstream impedance for a three-phase fault. equivalent impedance of the upstream network.
c cos e E
i ia idc ip
I Ib Isc Ik I" k Ir Is
k&K Ra RL Sn Ssc tmin u usc U Un Xa XL Z(1), Z(2), Z(0) ZL Zsc Zup
calculation of short-circuit currents
contents
the main types of short-circuits establishing the short-circuit current standardised Isc calculations methods presented in this document basic assumptions 2. Calculation of Isc using the Isc depending on the different impedance method types of short-circuit determining the various short-circuit impedances relationships between impedances at the different voltage levels in an installation calculation example 3. Calculation of Isc values in a radial advantages of this method network using symmetrical symmetrical components components calculation as defined by IEC 909 equations for the various currents calculation example 4. Computerised calculations and conclusion Appendix: bibliography 1. Introduction p. 5 p. 6 p. 9 p. 10 p. 10 p. 11 p. 12 p. 16
In view of sizing an electrical installation and the required equipment, as well as determining the means required for the protection of life and property, short-circuit currents must be calculated for every point in the network. This Cahier Technique publication reviews the calculation methods for short-circuit currents as laid down by standards UTE 15-105 and IEC 909-781. It is intended for radial low-voltage (LV) and high-voltage (HV) circuits. The aim is to provide a further understanding of the calculation methods, essential when determining short-circuit currents, even when computerised methods are employed.
p. 16 p. 19 p. 19 p. 20 p. 21 p. 22 p. 24 p. 24
1. introduction
Electrical installations almost always require protection against short-circuits wherever there is an electrical discontinuity. This most often corresponds to points where there is a change in conductor cross-section. The short-circuit current must be calculated at each level in the installation in view of determining the characteristics of the equipment required to withstand or break the fault current. The flow chart in figure 1 indicates the procedure for determining the various short-circuit currents and the resulting
parameters for the different protection devices. In order to correctly select and adjust the protection devices, two values of the short-circuit current must be evaluated: c the maximum short-circuit current, used to determine: v the breaking capacity of the circuit breakers; v the making capacity of the circuit breakers; v the electrodynamic withstand capacity of the wiring system and switchgear.
The maximum short-circuit current corresponds to a short-circuit in the immediate vicinity of the downstream terminals of the protection device. It must be calculated accurately and used with a safety margin. c the minimum short-circuit current, essential when selecting the timecurrent curve for circuit breakers and fuses, in particular when: v cables are long and/or the source impedance is relatively high (generators, UPSs); v protection of life depends on circuit breaker or fuse operation, essentially
Upstream Ssc
HV/LV transformer rating
Usc (%)
Isc at transformer terminals

power factor, coincidence factor, duty factor, foreseeable expansion factor conductor characteristics s busbars: - length, - width, - thickness. s cables: - type of insulation, - single-core or multicore, - length, - cross-section, s environment : - ambient temperature, - installation method, - number of contiguous circuits. breaking capacity inst. trip setting main circuit breaker
Isc of main LV switchboard outgoers

breaking capacity inst. trip setting main LV switchboard distribution circuit breakers
Isc
at head of secondary switchboards breaking capacity inst. trip setting
secondary distribution circuit breakers
feeder current ratings voltage drops
Isc
at head of final switchboards breaking capacity instan. trip setting final distribution circuit breakers
load rating
Isc
at end of final outgoers
fig. 1: short-circuit (Isc) caculation procedure when designing an electrical installation.
the case for TN and IT electrical systems. Note that the minimum short-circuit current corresponds to a short-circuit at the end of the protected line, generally phase-to-earth for LV and phase-to-phase for HV (neutral not distributed), under the least severe operating conditions (fault at the end of a feeder and not just downstream from a protection device, one transformer in service when two can be connected, etc.). Note also that whatever the case, for whatever type of short-circuit current (minimum or maximum), the protection device must clear the short-circuit within a time tc that is compatible with the thermal stresses that can be withstood by the protected cable:
a1 > a2
5s
I2t = k2A2
Iz1 < Iz2
fig. 2: the I2t characteristics of a conductor depending on the ambient temperature.
dt i k 2 A 2 (see fig. 2, 3 and 4)
t design current cable or I2t characteristic
where A is the cross-sectional area of the conductors and k is a constant calculated on the basis of different correction factors for the cable installation method, contiguous circuits, etc. Further practical information may be found in BS 7671 16th Edition IEE Wiring regulations or Guide de linstallation lectrique published by Merlin Gerin (see the bibliography).
transient overload
circuit breaker time-current curve
the main types of shortcircuits

Various types of short-circuits can occur in electrical installations. Characteristics of short-circuits The primary characteristics are: c duration (self-extinguishing, transient and steady-state); c origin: v mechanical (break in a conductor, accidental electrical contact between two conductors via a foreign conducting body such as a tool or an animal); v internal or atmospheric overvoltages; v insulation breakdown due to heat, humidity or a corrosive environment; c location (inside or outside a machine or an electrical switchboard). Short-circuits can be: c phase-to-earth (80% of faults); c phase-to-phase (15% of faults). This type of fault often degenerates into a three-phase fault;
IB Ir Iz
fig. 3: circuit protection using a circuit breaker.
Isc BC
cable or I2t characteristic
transient overload
fuse time-current curve
IB
Ir
Iz
fig. 4: circuit protection using an aM fuse.
c three-phase (only 5% of initial faults). These different short-circuit currents are presented in figure 5. Consequences of short-circuits The consequences are variable depending on the type and the duration of the fault, the point in the installation where the fault occurs and the shortcircuit power. Consequences include: c at the fault location, the presence of electrical arcs, resulting in: v damage to insulation; v welding of conductors; v fire and danger to life; c on the faulty circuit: v electrodynamic forces, resulting in: - deformation of the busbars; - disconnection of cables; v excessive temperature rise due to an increase in Joule losses, with the risk of damage to insulation; c on other circuits in the network or in near-by networks:
v voltage dips during the time required to clear the fault, ranging from a few milliseconds to a few hundred milliseconds; v shutdown of a part of the network, the extent of that part depending on the design of the network and the discrimination levels offered by the protection devices; v dynamic instability and/or the loss of machine synchronisation; v disturbances in control / monitoring circuits, etc.
establishing the shortcircuit current

A simplified network comprises a source of constant AC power, a switch, an impedance Zsc that represents all the impedances upstream of the switch, and a load impedance Zs (see fig. 6). In a real network, the source impedance is made up of everything
upstream of the short-circuit including the various networks with different voltages (HV, LV) and the seriesconnected wiring systems with different cross-sectional areas (A) and lengths. In figure 6, when the switch is closed, the design current Is flows through the network. When a fault occurs between A and B, the negligible impedance between these points results in a very high short-circuit current Isc that is limited only be impedance Zsc. The current Isc develops under transient conditions depending on the reactances X and the resistances R that make up impedance Zsc:
Zsc = R + X
In power distribution networks, reactance X = L is normally much greater than resistance R and the R / X ratio is between 0.1 and 0.3. The ratio is virtually equals cos sc for low values:
cos sc = R
2 2 R +X
a) L3 L2 L1
b) L3 L2 L1
I"k
I"k
However, the transient conditions prevailing while the short-circuit current develops differ depending on the distance between the fault location and the generator. This distance is not necessarily physical, but means that the generator impedances are less than the link impedance between the generator and the fault location. Fault away from the generator This is the most frequent situation. The transient conditions are those resulting
c) L3 L2 L1
d) L3 L2 L1
I"k I"k
I"k
I"k
A Zsc
short-circuit current, partial short-circuit currents in conductors and earth. In calculations, the various currents ( I " ) are k identified by an index.
a) symmetrical three-phase short-circuit. b) phase-to-phase short-circuit clear of earth. c) phase-to-phase-to-earth short-circuit. d) phase-earth short-circuit.
Zs
fig. 5: different types of short-circuits and their currents. The direction of current is chosen arbitrarily. (See IEC 909).
fig. 6: simplified network diagram.
from the application of a voltage to a reactor-resistance circuit. This voltage is: e = E sin ( t + ) Current i is then the sum of the two components: i = ia + idc. c the first (ia) is alternating and sinusoidal:
ia = I sin ( t + ) where
fault initiation I
ia = I sin ( t + )
idc = -I sin e L
- R t
i = ia + idc
E I = maximum current = , Zsc
= angle characterising the difference

between the initiation of the fault and zero voltage. c the second (idc) is an aperiodic component,
idc = I sin e Its initial value depends on and its decay rate is proportional to R / L.
R t L .
fig. 7: graphical presentation and decomposition of a short-circuit current occuring away from the generator.
a) symmetrical i
I = 2 Ia
The moment the fault occurs or the moment of closing, with respect to the network voltage, is characterised by its closing angle (occurrence of the fault). The voltage can therefore be expressed as: u = E sin ( t + ) The current therefore develops as follows: i= E
At the initiation of the short-circuit, i is equal to zero by definition (the design current Is is negligible), hence: i = ia + idc = 0 Figure 7 shows the graphical composition of i as the algebraic sum of its two components ia et idc. Figure 8 illustrates the two extreme cases for the development of a shortcircuit current, presented, for the sake of simplicity, with a single-phase, alternating voltage. The factor e L is inversely proportional to the aperiodic component damping, determined by the R / L or R / X ratios. The value of ip must therefore be calculated to determine the making capacity of the required circuit breakers and to define the electrodynamic forces that the installation as a whole must be capable of withstanding. Its value may be deduced from the rms value of the symmetrical short-circuit current Ia using the equation: ip = K 2 I a where the coefficient K is indicated by the curve in figure 9, as a function of the ratio R / X or R / L. Fault near the generator When the fault occurs in the immediate vicinity of the generator supplying the circuit, the variation in the impedance of
R t
b) asymmetrical i ip idc
with its two components, one being alternating with a shift equal to with respect to the voltage and the second aperiodic and decaying to zero as t tends to infinity. Hence the two extreme cases defined by: v = /2, said to be symmetrical (or balanced), (see figure a). The fault current can be defined by: E i = sin t which, from the initiation, has Z the same shape as for steady state conditions with a peak value E / Z. v = 0, said to be asymmetrical (or unbalanced), (see figure b). The fault current can be defined by: i= E
sin ( t + ) sin ( ) e Z
R L
Its initial peak value ip therefore depends on , i.e. on the R / X = cos ratio of the circuit.
sin ( t ) sin e Z
R L
fig. 8: graphical presentation of the two extreme cases (symmetrical and asymmetrical) for a short-circuit current.
the generator, in this case the dominant impedance, damps the short-circuit current. The transient current-development conditions are complicated by the variation in the electromotive force resulting from the short-circuit. For simplicity, the electromotive force is assumed to be constant and the internal reactance of the machine variable. The reactance develops in three stages:
K 2.0 1.8 1.6 1.4 1.2 1.0 0 0.2 0.4 0.6 0.8 1.0 R/X 1.2
fig. 9: variation of coefficient K depending on R / X or R / L. (see IEC 909)
15% 0.05s a) 0 t (s)
c subtransient (the first 10 to 20 milliseconds of the fault); c transient (up to 500 milliseconds); c steady-state (or synchronous reactance). Note that in the indicated order, the reactance acquires a higher value at each stage, i.e. the subtransient reactance is less than the transient reactance, itself less than the steadystate reactance. The successive effect of the three reactances leads to a gradual reduction in the short-circuit current which is the sum of four components (see fig. 10): c the three alternating components (subtransient, transient and steadystate); c the aperiodic component resulting from the development of the current in the circuit (inductive). Practically speaking, information on the development of the short-circuit current is not essential: c in a LV installation, due to the speed of the breaking devices, the value of the subtransient short-circuit current, denoted I " , and of the k maximum asymmetrical peak amplitude ip is sufficient when determining the breaking capacities of the protection devices and the electrodynamic forces; c in LV power distribution and in HV applications, however, the transient short-circuit current is often used if breaking occurs before the steadystate stage, in which case it becomes useful to use the short-circuit breaking current, denoted Ib, which determines the breaking capacity of the timedelayed circuit breakers. Ib is the value of the short-circuit current at the moment interruption is effective, i.e. following a time t after the development of the short-circuit, where t = tmin Time tmin (minimum time delay) is the sum of the minimum operating time of a protection relay and the shortest opening time of the associated circuit breaker, i.e. the shortest time between the appearance of the short-circuit current and the initial separation of the pole contacts on the switching device. Figure 11 presents the various
b) 0
t (s)
c) 0
t (s)
d) 0 0.1 0.3 0.5
t (s)
e) 0 subtransient transient steady-state
t (s)
fig. 10: components of the total short-circuit Isc (e). Contribution of each component: a) subtransient of each reactance; b) transient reactance; c) steady-state reactance; d) aperiodic component. Note that the decrease in the generator reactance is faster than that of the aperiodic component. This is a rare situation that can cause saturation of the magnetic circuits and interruption problems because several periods occur before the current passes through zero.
currents of the short-circuits defined above.

i
standardised Isc calculations

The standards propose a number of methods: c application guide C 15-105, which supplements NF C 15-100 (Normes Franaises) (low-voltage AC installations), details four methods: v the impedance method, used to calculate fault currents at any point in an installation with a high degree of accuracy. This method involves adding the various resistances and reactances of the fault loop separately, from (and including) the source to the given point, and then calculating the corresponding impedance. The Isc value is finally obtained by applying Ohm's law:
symmetrical
subtrans.
transient
steady-state
asymmetrical
fig. 11: short-circuit currents near a generator (schematic diagram).
I sc = Un / (Z) .
All the characteristics of the various elements in the fault loop must be known (sources and wiring systems). Note that in the application guide, a number of factors are not taken into account, notably: - the reactances of the circuit breakers and the busbars; - the resistances of rotating machines. The results obtained may be different from those presented in the next chapter, because these factors are taken into account. v the composition method, which may be used when the characteristics of the power supply are not known. The upstream impedance of the given circuit is calculated on the basis of an estimate of the short-circuit current at its origin. The power factor cos sc = R / X is assumed to be identical at the origin of the circuit and the fault location. In other words, it is assumed that the elementary impedances of two successive sections in the installation are sufficiently similar in their characteristics to justify the replacement of vectorial addition of the impedances by algebraic addition. This approximation may be used to calculate the value of the short-circuit current modulus with sufficient accuracy for the addition of a circuit. This very
approximate method should be used only for installations rated up to 800 kVA. v the conventional method, which can be used, when the impedances or the Isc in the installation upstream of the given circuit are not known, to calculate the minimum short-circuit currents and the fault currents at the end of a line. It is based on the assumption that the voltage at the circuit origin is equal to 80% of the rated voltage of the installation during the short-circuit or the fault. This method considers only the resistance of the conductors and applies a coefficient greater than 1 to conductors with large cross-sectional areas to take into account their 2 inductance (1.15 for 150 mm , 1.20 for 2 185 mm , etc.). It is mainly used for final circuits with their origin at a distance that is sufficiently far from the power source (network or power-station unit). v the simplified method (presented in detail in this application guide), which, via tables based on numerous simplifying assumptions, indicates for each conductor cross-sectional area: - the current rating of the overload protection device; - maximum lengths of wiring systems to maintain protection against indirect contact;
permissible lengths in view of line voltage drops. The data in the tables is in fact the result of calculations run using essentially the composition and conventional methods. This method may be used to determine the characteristics of a circuit to be added to an existing installation for which sufficient information is not available. It is directly applicable to LV installations, and can be used with correction coefficients if the voltage is not 230 / 400 V. c standard IEC 909 (VDE 0102) applies to all networks, radial or meshed, up to 230 kV. This method, based on the Thevenin theorem, calculates an equivalent voltage source at the shortcircuit location and then determines the corresponding short-circuit current. All network feeders as well as the synchronous and asynchronous machines are replaced in the calculation by their impedances (positive-sequence, negative-sequence and zero-sequence). All line capacitances and the parallel admittances of non-rotating loads, except those of the zero-sequence system, are neglected. c other methods use the superposition principle and require that the load current first be calculated. Note also the method proposed by standard IEC 865 (VDE 0103) which calculates the
-
thermally equivalent short-circuit current.
basic assumptions
To simplify the short-circuit calculations, a number of assumptions are required. These impose limits for which the calculations are valid but usually provide good approximations, facilitating comprehension of the physical phenomena and consequently the short-circuit current calculations. They nevertheless maintain a fully acceptable level of accuracy, erring systematically on the conservative side. The assumptions used in this document are as follows: c the given network is radial with rated voltages ranging from LV to HV, but not exceeding 230 kV, the limit set by standard IEC 909; c the short-circuit current, during a threephase short-circuit, is assumed to occur simultaneously on all three phases;
methods presented in this document

In this Cahier Technique publication, two methods are presented for the calculation of short-circuit currents in radial networks: c the impedance method, reserved primarily for LV networks, was selected for its high degree of accuracy and its instructive value, given that virtually all characteristics of the circuit are taken into account. c the IEC 909 method, used primarily for HV networks, was selected for its accuracy and its analytical character. More technical in nature, it implements the symmetrical-component principle.
c during the short-circuit, the number of phases involved does not change, i.e. a three-phase fault remains three-phase and a phase-to-earth fault remains phase-to-earth; c for the entire duration of the shortcircuit, the voltages responsible for the flow of the current and the short-circuit impedance do not change significantly; c transformer regulators or tap-changers are assumed to be set to a medium position (if the short-circuit occurs away from the generator, the actual position of the transformer regulator or tap-changers does not need to be taken into account; c arc resistances are not taken into account; c all line capacitances are neglected; c load currents are neglected; c all zero-sequence impedances are taken into account.
2. calculation of Isc using the impedance method
Isc depending on the

different types of short-circuit
Three-phase short-circuit This fault involves all three phases. Short-circuit current Isc3 is equal to: U/ 3 I sc 3 = Zsc where U (phase-to-phase voltage) corresponds to the transformer no-load voltage which is 3 to 5% greater than the on-load voltage across the terminals. For example, in 390 V networks, the phase-to-phase voltage adopted is U = 410, and the phase-toneutral voltage is U / 3 = 237 V. Calculation of the short-circuit current therefore requires only calculation of Zsc, the impedance equal to all the impedances through which Isc flows from the generator to the location of the fault, i.e. the impedances of the power sources and the lines (see fig. 12). This is, in fact, the positive-sequence impedance per phase: Zsc =
neutral voltage V = U / 3 . The short-circuit current Isc1 is: U/ 3 Zsc + ZLn In certain special cases of phase-to-neutral faults, the zerosequence impedance of the source is less than Zsc (for example, at the terminals of a star-zigzag connected transformer or of a generator under subtransient conditions). In this case, the phase-to-neutral fault current may be greater than that of a three-phase fault. I sc1 =
Phase-to-earth fault (one or two phases) This type of fault brings the zerosequence impedance Z(0) into play. Except when rotating machines are involved (reduced zero-sequence impedance), the short-circuit current Isc(0) is less than that of a three-phase fault. Calculation of Isc(0) may be necessary, depending on the neutral system (system earthing arrangement), in view of defining the setting thresholds for the zero-sequence (HV) or earth-fault (LV) protection devices. Figure 12 shows the various shortcircuit currents
three-phase fault
ZL
Zsc
ZL ZL
I sc 3 =
U/ 3 Zsc
R = the sum of series resistances; X = the sum of series reactances.

It is generally considered that threephase faults provoke the highest fault currents. The fault current in an equivalent diagram of a polyphase system is limited by only the impedance of one phase at the phase-to-neutral voltage of the network. Calculation of Isc3 is therefore essential for selection of equipment (maximum current and electrodynamic withstand capability). Phase-to-phase short-circuit clear of earth This is a fault between two phases, supplied with a phase-to-phase voltage U. In this case, the short-circuit current Isc2 is less than that of a three-phase fault: U 3 I sc 2 = = I sc 3 0.86 I sc 3 2 2 Zsc Phase-to-neutral short-circuit clear of earth This is a fault between one phase and the neutral, supplied with a phase-to-
( R) + ( X )
2
where
phase-to-phase fault
ZL
Zsc
U ZL Zsc
I sc 2 =
U 2 Zsc
phase-to-neutral fault
ZL
Zsc
ZLn
V ZLn
I sc1 =
U/ 3 Zsc + ZLn
phase-to-earth fault
ZL
Zsc
V Z(0)
I sc(0) =
Z(0)
U/ 3 Zsc + Z(0)
fig. 12: the various short-circuits currents.
determining the various short-circuit impedances

This method involves determining the short-circuit currents on the basis of the impedance represented by the circuit through which the short-circuit current flows. This impedance may be calculated after separately summing the various resistances and reactances in the fault loop, from (and including) the power source to the fault location. (The circled numbers x may be used to come back to important information while reading the example at the end of this section.) Network impedances c upstream network impedance Generally speaking, points upstream of the power source are not taken into account. Available data on the upstream network is therefore limited to that supplied by the power distributor, i.e. only the short-circuit power Ssc in MVA. The equivalent impedance of the upstream network is: U2 Ssc where U is the no-load phase-to-phase voltage of the network.
1
For public distribution MV/LV transformers, the values of usc have been set by the European Harmonisation document HD 428-1S1 issued in October 1992 (see fig. 13). Note that the accuracy of values has a direct influence on the calculation of Isc in that an error of x% for usc produces an equivalent error (x%) for ZT.
4
The upstream network impedance may be neglected, in which case the new current value is: U ' sc = 3 ZT The relative error is:
sc ' sc sc Zup U2 / Ssc = = = sc sc ZT usc U2 / Sn i.e.:
In general, RT << XT , in the
order of 0.2 XT, and the internal transformer impedance may be considered comparable to reactance XT. For low power levels, however, calculation of ZT is required because the ratio RT / XT is higher. The resistance is calculated using the joule losses (W) in the windings:
W = 3 RT n2 RT =
Notes :
W 3 n2
sc 100 Sn = x sc usc Ssc Figure 14 indicates the level of conservative error in the calculation of Isc, due to the fact that the upstream impedance is neglected. The figure demonstrates clearly that it is possible to neglect the upstream impedance for networks where the short-circuit power Ssc is much higher than the transformer kVA rating Sn. For example, when Ssc / Sn = 300, the error is approximately 5%.
5 v when n identically-rated transformers are connected in parallel, their internal impedance values, as well as the resistance and reactance values, must be divided by n.
c link impedance The link impedance ZL depends on the resistance per unit length, the reactance per unit length and the length of the links. v the resistance per unit length of overhead lines, cables and busbars is calculated as: RL = A where A = cross-sectional area of the conductor; = conductor resistivity, however the value used varies, depending on the calculated short-circuit current (minimum or maximum).
6
Zup =
The upstream resistance and reactance may be deduced from Rup / Zup (for HV) by: Rup / Zup 0.3 at 6 kV; Rup / Zup 0.2 at 20 kV; Rup / Zup 0.1 at 150 kV.
v particular attention must be paid to special transformers, for example, the transformers for rectifier units have usc values of up to 10 to 12% in order to limit short-circuit currents. When the impedance upstream of the transformer and the transformer internal impedance are taken into account, the short-circuit current may be expressed as:
Xup = 0.980 Zup at 20 kV,
sc =
hence the approximation Xup Zup. c internal transformer impedance The impedance may be calculated on the basis of the short-circuit voltage usc expressed as a percentage: U2 where Sn U = no-load phase-to-phase voltage of the transformer;
U 3 (Zup + Z T )
The table in figure 15 provides
Initially, Zup and ZT may be considered comparable to their respective reactances. The short-circuit impedance Zsc is therefore equal to the algebraic sum of the two.
values for each of the above-mentioned cases. Practically speaking, for LV and conductors with cross-sectional areas 2 less than 150 mm , only the resistance is taken into account (RL < 0.15 m / m when A > 150 mm2).
Z T = usc
Rating (kVA) of the HV/LV transformer Short-circuit voltage usc (%)
50 4
100 4
400 4
630
1000
2500 6
Sn = transformer kVA rating; U usc = voltage that must be applied to the primary winding of the transformer for the rated current to flow through the secondary winding, when the LV secondary terminals are short-circuited.
4 or 6 6 depending on the substation
fig. 13: standardised short-circuit voltage usc (HD 428-1S1) for public distribution immersedtype transformers.
v the reactance per unit length of overhead lines, cables and busbars may be calculated as: d X L = L = 15.7 + 144.44 Log r expressed as m / km for a singlephase or three-phase delta cable system, where (in mm): r = radius of the conducting cores; d = average distance between conductors. N.B. Above, Log = decimal logarithm. For overhead lines, the reactance increases slightly in proportion to the distance between conductors d ( Log ), and therefore in proportion r to the operating voltage.
Isc/Isc (%) 12 10 Psc = 500 MVA 5 Psc = 250 MVA
0 500
1000
1500
2000
Pn (KVA)
fig. 14: resultant error in the calculation of the short-circuit current when the upstream network impedance Zup is neglected.
7 the following average values are to be used: X = 0.3 / km (LV lines); X = 0.4 / km (MV or HV lines).
The table in figure 16 shows the various reactance values for conductors in LV applications, depending on the wiring system. The following average values are to be used: - 0.08 m / m for a three-phase cable ( ), and, for HV applications, between 0.1 and 0.15 m / m.
8 - 0.09 m / m for touching, single-conductor cables (flat or
Current
Resistivity (*) 1 = 1.25 20 1 = 1.5 20 1 = 1.25 20
Resistivity value ( mm2/m) Copper Aluminium 0.036 0.043 0.036 0.0225 0.027 0.0225
Concerned conductors PH-N PH-N PH-N (**) PE-PEN PH-N (*) Phase-Neutral PEN-PE if incorporated in same multiconductor cable Separate PE
Maximum short-circuit current Minimum short-circuit current Fault current in TN and IT systems Voltage drop Overcurrent for conductor thermal-stress checks
1 = 1.25 20 1 = 1.5 20
0.0225 0.027
0.036 0.043
triangular
);
1 = 1.25 20
0.0225
0.036
9 - 0.15 m / m as a typical value for busbars ( ) and spaced, single-conductor cables ( ) For sandwiched-phase busbars (e.g. Canalis - Telemecanique), the reactance is considerably lower.
(*) 20 is the resistivity of the conductors at 20 C. 0.018 mm2/m for copper and 0.029 mm2/m for aluminium. (**) N, the cross-sectional area of the neutral conductor, is less than that of the phase conductor
fig. 15: conductor resistivity values to be taken into account depending on the calculated short-circuit current (minimum or maximum). See UTE C 15-105.
wiring system diagram
busbars
three-phase cable
spaced single-core touching single-core 3 touching cables cables (triangle) cables (flat)
3 "d" spaced cables (flat) d = 2r d = 4r

d d r
average reactance per unit length 0.15 values (m/m) extreme reactance per unit length 0.12-0.18 values (m/m)
0.08
0.15
0.085
0.095
0.145
0.19
0.06-0.1
0.1-0.2
0.08-0.09
0.09-0.1
0.14-0.15
0.18-0.20
fig. 16: cables reactance values depending on the wiring system.
Notes: v the impedance of the short links between the distribution point and the HV/LV transformer may be neglected. This assumption gives a conservative error concerning the short-circuit current. The error increases in proportion to the transformer rating. v the cable capacitance with respect to the earth (common mode), which is 10 to 20 times greater than that between the lines, must be taken into account for earth faults. Generally speaking, the capacitance of a HV three-phase cable with a cross-sectional area of 120 mm2 is in the order of 1 F / km, however the capacitive current remains low, in the order of 5 A / km at 20 kV. c the reactance or resistance of the links may be neglected. If one of the values, RL or XL, is low with respect to the other, it may be neglected because the resulting error for impedance ZL is consequently very low. For example, if the ratio between RL and XL is 3, the error in ZL is 5.1%. The curves for RL and XL (see fig. 17) may be used to deduce the cable crosssectional areas for which the impedance may be considered comparable to the resistance or to the reactance. Examples: v first case. Consider a three-phase cable, at 20C, with copper conductors. Their reactance is 0.08 m/m. The RL and XL curves (see fig. 17) indicate that impedance Z L approaches two asymptotes, R L for low cable crosssectional areas and XL = 0.08 m/m for high cable cross-sectional areas. For the low and high cable cross-sectional areas, the impedance Z L curve may be considered identical to the asymptotes. The given cable impedance is therefore considered, with a margin of error less than 5.1%, comparable to: - a resistance for cable cross-sectional 2 areas less than 74 mm ; - a reactance for cable cross-sectional 2 areas greater than 660 mm . v second case. Consider a three-phase cable, at 20C, with aluminium conductors. As above, the impedance ZL curve may be considered identical to the asymptotes, but for cable crosssectional areas less than 120 mm2 and 2 greater than 1,000 mm (curves not shown).
Impedance of rotating machines c synchronous generators The impedances of machines are generally expressed as a percentage, for example: Isc / In = 100/e where e is the equivalent of the transformer usc. Consider: e U2 Z = 100 Sn where U = no-load phase-to-phase voltage of the generator,
10
v 0.03 to 0.1 seconds for very large HV slipring motors (1 000 kW). In the event of a short-circuit, an asynchronous motor is therefore a generator to which an impedance (subtransient only) of 20 to 25% is attributed. Consequently, the large number of LV motors, with low individual outputs, present on industrial sites may be a source of difficulties in that it is not easy to foresee the average number of inservice motors that will contribute to the fault when a short-circuit occurs. Individual calculation of the reverse current for each motor, taking into account the impedance of its link, is therefore a tedious and futile task. Common practice, notably in the United States, is to take into account the combined contribution to the fault current of all the asynchronous LV motors in an installation.
13
Sn = generator VA rating.
11
What is more, given that the
value of R / X is low, in the order of 0.05 to 0.1 for MV and 0.1 to 0.2 for LV, impedance Z may be considered comparable to reactance X. Values for e are given in the table in figure 18 for turbo-generators with smooth rotors and for hydraulic generators with salient poles (low speeds). On reading the table, one may be surprised to note that the steady-state reactance for a short-circuit exceeds 100% (at that point in time, Isc < In) . However, the short-circuit current is essentially inductive and calls on all the reactive power that the field system, even over-excited, can supply, whereas the rated current essentially carries the active power supplied by the turbine (cos from 0.8 to 1). c synchronous compensators and motors The reaction of these machines during a short-circuit is similar to that of generators.
12
They are therefore thought of
as a unique source, capable of supplying to the busbars a current equal to three times the sum of the rated currents of all installed motors. Other impedances c capacitors A shunt capacitor bank located near the fault location discharges, thus increasing the short-circuit current. This damped oscillatory discharge is characterised by a high initial peak value that is superposed on the initial peak of the short-circuit current, even though its frequency is far greater than that of the network. Depending on the coincidence in time between the initiation of the fault and the voltage wave, two extreme cases must be considered: v if the initiation of the fault coincides with zero voltage, the discharge current is equal to zero, whereas the shortcircuit current is asymmetrical, with a maximum initial amplitude peak; v conversely, if the initiation of the fault coincides with maximum voltage, the discharge current superposes itself on the initial peak of the fault current, which, because it is symmetrical, has a low value. It is therefore unlikely, except for very powerful capacitor banks, that superposition will result in an initial
They produce a current in the
network that depends on their reactance in % (see fig. 19). c asynchronous motors When an asynchronous motor is cut from the network, it maintains a voltage across its terminals that disappears within a few hundredths of a second. When a short-circuit occurs across the terminals, the motor supplies a current that disappears even more rapidly, according to time constants in the order of: v 0.02 seconds for single-cage motors up to 100 kW; v 0.03 seconds for double-cage motors and motors above 100 kW;
peak higher than the peak current of an asymmetrical fault. It follows that when calculating the maximum short-circuit current, capacitor banks do not need to be taken into account. However, they must nonetheless be considered when selecting the type of circuit breaker. During opening, capacitor banks significantly reduce the circuit frequency and thus produce an effect on current interruption. c switchgear
14
16
For example, the arc resulting
from a short-circuit between conductors or busbars may reduce the prospective short-circuit current by 20 to 50% and sometimes by even more than 50% for rated voltages under 440 V. However, this phenomenon, highly favourable in the LV field and which occurs for 90% of faults, may not be taken into account when determining the breaking capacity because 10% of faults take place during closing of a device, producing a solid fault without
an arc. This phenomenon should, however, be taken into account for the calculation of the minimum short-circuit current. c various impedances Other elements may add non-negligible impedances. This is the case for harmonics filters and inductors used to limit the short-circuit current. They must, of course, be included in calculations, as well as wound-primary type current transformers for which the impedance values vary depending on the rating and the type of construction.
Certain devices (circuit

m /m 1 0.8
breakers, contactors with blow-out coils, direct thermal relays, etc.) have an impedance that must be taken into account, for the calculation of Isc, when such a device is located upstream of the device intended to break the given short-circuit and remain closed (selective circuit breakers).
15
For LV circuit breakers, for
0.2
ZL
example, a reactance value of 0.15 m is typical, with the resistance negligible. For breaking devices, a distinction must be made depending on the speed of opening: v certain devices open very quickly and thus significantly reduce short-circuit currents. This is the case for fastacting, limiting circuit breakers and the resultant level of electrodynamic forces and thermal stresses, for the part of the installation concerned, remains far below the theoretical maximum; v other devices, such as time-delayed circuit breakers, do not offer this advantage. c fault arc The short-circuit current often flows through an arc at the fault location. The resistance of the arc is considerable and highly variable. The voltage drop over a fault arc can range from 100 to 300 V. For HV applications, this drop is negligible with respect to the network voltage and the arc has no effect on reducing the short-circuit current. For LV applications, however, the actual fault current when an arc occurs is limited to a much lower level than that calculated (bolted, solid fault), because the voltage is much lower.
0.1 0.08 0.05

XL RL
0.02
0.01 10
20
50
100
200
500 1000
Cross-sectional area A (in mm 2)
fig. 17: impedance ZL of a three-phase cable, at 20C, with copper conductors.
subtransient reactance turbo-generator salient-pole generators 10-20 15-25
transient reactance 15-25 25-35
steady-state reactance 150-230 70-120
fig. 18: generator reactance values, in e%.
subtransient reactance high-speed motors low-speed motors compensators 15 35 25
transient reactance 25 50 40
steady-state reactance 80 100 160
fig. 19: synchronous compensator and motor reactance values, in e%.
relationships between impedances at the different voltage levels in an installation

Impedances as a function of the voltage The short-circuit power Ssc at a given point in the network is defined by: U2 Ssc = U 3 = Zsc This means of expressing the shortcircuit power implies that Ssc is invariable at a given point in the network, whatever the voltage. And the equation U implies that all sc 3 = 3 Zsc impedances must be calculated with respect to the voltage at the fault location, which leads to certain complications that often produce errors in calculations for networks with two or more voltage levels. For example, the impedance of a HV line must be multiplied by the square of the reciprocal of the transformation ratio, when calculating a fault on the LV side of the transformer: 2 U 17 ZLV = ZHV LV UHV A simple means of avoiding these difficulties is the relative impedance method proposed by H. Rich. Calculation of the relative impedances This is a calculation method used to establish a relationship between the impedances at the different voltage levels in an electrical installation. This method proposes dividing the impedances (in ohms) by the square of the network line-to-line voltage (in volts) at the point where the impedances exist. The impedances therefore become relative. c for lines and cables, the relative resistances and reactances are defined as: R X RR = 2 and X R = 2 U U where R is in ohms and U in volts. c for transformers, the impedance is expressed on the basis of their shortcircuit voltages usc and their kVA rating Sn: U2 u Z= Sn 100 c for rotating machines, the equation is identical, with e representing the impedance expressed in %.
c for the system as a whole, after having calculated all the relative impedances, the short-circuit power may be expressed as: 1 Ssc = ZR from which it is possible to deduce the fault current Isc at a point with a voltage U: Ssc 1 I sc = = 3U 3 U ZR
calculation example
(with the impedances of the power sources, the upstream network and the power supply transformers as well as those of the electrical links) Problem Consider a 20 kV network that supplies a HV/LV substation via a 2 km overhead line, and a 1 MVA generator that supplies in parallel the busbars of the same substation. Two 1,000 kVA parallel-connected transformers supply
the LV busbars which in turn supply 20 outgoers to 20 motors, including the one supplying motor M. All motors are rated 50 kW, all connection cables are identical and all motors are running when the fault occurs. The Isc value must be calculated at the various fault locations indicated in the network diagram (see fig. 20), that is: c point A on the HV busbars, with a negligible impedance; c point B on the LV busbars, at a distance of 10 meters from the transformers; c point C on the busbars of an LV subdistribution board; c point D at the terminals of motor M. Then the reverse current of the motors must be calculated at C and B, then at D and A. In this example, reactances X and resistances R are calculated with their respective voltages in the installation. The relative impedance method is not used.
upstream network U1 = 20 kV Psc = 500 MVA overhead line 3 cables, 50 mm 2, copper length = 2 km generator 1 MVA subtrans. Z = 15 % 2 transformers 1000 kVA secondary winding 237/410 V e=5% main LV switchboard busbars 3 bars, 400 mm 2/ph, copper length = 10 m link 1 3 single-core cables, 400 mm 2, aluminium spaced, flat length = 80 m LV sub-distribution board link 2 3 three-phase cables, 35 mm 2, copper length = 30 m motor 50 kW e = 25 %
3L
G A
10 m B
3L
3L D M
fig. 20: diagram for calculation of Isc values at points A, B, C and D.
Solution section (the circled numbers 20 kV calculations results
I - Fault at A (HV busbars) Elements concerned: 1, 2, 3. The network + line impedance is parallel to that of the generator, however the latter is much greater and may be neglected: XA = 0.78 + 0.8 1.58
0.78 0.15
indicate where explanations may be found in the preceding text)

X () Zup = (20 x 103)2 / 500 x 106 Xup = 0.98 Zup Rup = 0.2 Xup R ()
1. upstream network
2. overhead line (50 mm2)
3. generator
20 kV
Fault A 4. transformers
410 V
5. circuit-breaker 6. busbars (3 x 400 mm2)
Fault B 7. circuit-breaker 8. cable link 1 (3 x 400 mm2) Xcb = 0.15 Xc1 = 0.15 x 10-3 x 80 Rc1 = 0.036 x Fault C 9. circuit-breaker 10. cable link 2 (35 mm2) Xcb = 0.15 Xc2 = 0.09 x 10-3 x 30 0.15 80 3 x 400 0.15 12
Fault D 11. motor 50 kW Xm = 25 410 x 100 50 x 103

2
RA = 0.15 + 0.72 0.87 Z A = R 2 + X 2 1.80 hence A A 20 x 10 3 6,415 A 3 x 1.80 IA is the steady-state Isc and for the purposes of calculating the peak asymmetrical Isc: RA = 0.55 hence k = 1.2 on the curve XA in figure 9 and therefore Isc is equal to: IA =
Xco = 0.4 x 2
Rc o = 0.018 x 2,000 50
7 6
0.8 0.72
XG =
20 x 103 15 x 100 106
10 11
60 6 X (m) R (m)
RG = 0.1 XG
1.2 x 2 x 6,415 = 10,887 A.

II - Fault at B (main LV switchboard busbars) Elements concerned: (1, 2, 3) + (4, 5, 6).
ZT =
1 5 4102 x x 2 100 106
5
4.2
XT ZT RT = 0.2 XT
0.84
The reactances X and resistances R calculated for the HV section must be recalculated for the LV network via multiplication by the square of the voltage ratio 17 , i.e.: (410 / 20,000)2 = 0.42 x 10-3 hence XB = [(XA 0.42) + 4.2 + 0.15 + 1.5] 10-3 XB = 6.51 m and
Xcb = 0.15 XB = 0.15 x 10-3 x 10 10 RB = 0.0225 x 3 x 400
15 9 6
0.15 1.5 0
RB = [(RA 0.42) + 0.84] 10-3 RB = 1.2 m These calculations make clear, firstly, the low importance of the HV upstream reactance, with respect to the reactances of the two parallel transformers, and secondly, the nonnegligible impedance of the 10 meter long, LV busbars.
2 2 ZB = RB + X B 6.62 410 IB = 35,758 A 3 x 6.62 x 10 3 RB = 0.18 hence k = 1.58 on the curve XB in figure 9 and therefore the peak Isc is equal to:
2.4
2.7 19.2
30 Rc2 = 0.0225 x 35
12
840 168
Rm = 0.2 Xm
1.58 x 2 x 35,758 = 79,900 A. What is more, if the fault arc is taken into account (see c fault arc section 16 ), IB is reduced to a maximum value of 28,606 A and a minimum value of 17,880 A.
III - Fault at C (busbars of LV subdistribution board) Elements concerned: (1, 2, 3) + (4, 5, 6) + (7, 8). The reactances and the resistances of the circuit breaker and the cables must be added to XB and RB. XC = (XB + 0.15 + 12) 10-3 = 18.67 m and -3 RC = (RB + 2.4) 10 = 3.6 m These values make clear the importance of Isc limitation due to the cables.
2 2 Z C = R C + X C 19 m
motor + cable impedance: XM = (840 + 2.7) 10-3 843 m RM = (168 + 19.2) 10 188 m ZM = 863 m hence 410 IM = 274 A 3 x 863 x 10 3 For the 20 motors
-3
being equal to three times their rated current (95 A), i.e. 5,700 A. This figure is very close to the value of IMB (5,400 A). Using the fact that R / X = 0.22 => k = 1.5 and the peak I sc = 1.5 x 2 x 5,400 11,455 A. Consequently, the short-circuit current (subtransient) on the main LV switchboard increases from 35,758 A to 41,158 A and the peak Isc from 79,900 A to 91,355 A. However, as mentioned above, if the fault arc is taken into account, Isc is reduced between 45.6 to 73 kA. c fault at A (HV side) Rather than calculating the equivalent impedances, it is easier to estimate (conservatively) the reverse current of the motors at A by multiplying the value at B by the LV/HV transformation value
17 , i.e.:
IMC = 5,480 A.
Instead of making the above calculations, it is possible (see 13 ) to estimate the current injected by all the motors as being equal to three times their rated current (95 A), i.e. (3 x 95) x 20 = 5,700 A. This figure is very close to the value of IMC (5,480 A). On the basis of R/X = 0.22 => k = 1.5 and the peak
I sc = 1.5 x 2 x 5,480 11,630 A.
IC =
410 12,459 A 3 x 19 x 10 3
RC = 0.19 hence k = 1.55 on the curve XC in figure 9 and therefore the peak Isc is equal to:
1.55 x 2 x 12,459 27,310 A.
IV - Fault at D (LV motor) Elements concerned: (1, 2, 3) + (4, 5, 6) + (7, 8) + (9, 10). The reactances and the resistances of the circuit breaker and the cables must be added to XC and RC. XD = (XC + 0.15 + 2.7) 10-3 = 21.52 m and -3 RD = (RC + 19.2) 10 = 22.9 m
2 2 ZD = RD + XD 31.42 m
Consequently, the short-circuit current (subtransient) on the LV busbars increases from 12,459 A to 17,939 A and Isc from 27,310 A to 38,940 A. c fault at D The impedance to be taken into account is 1/19th of ZM, plus that of the cable. 843 X MD = + 2.7 10 3 47 m 19 187 RMD = + 19.2 10 3 29 m 19 ZMD = 55 m hence 410 4,300 A 3 x 55 x 10 3 giving a total at D of: I MD = 7,534 + 4,300 = 11,834 A rms, and a peak Isc 17,876 A. c fault en B As for the fault at C, the current produced by the motor may be calculated on the basis of the motor + cable impedance: XM = (840 + 2.7 + 12) 10 855 m
-3
410 110 A 20 x 10 3 This figure, compared to the 6,415 A calculated previously, is negligible. 5,400 x Rough calculation of the fault at D This calculation makes use of all the approximations mentioned above (notably 15 and 16).
ID =
410 7,534 A 3 x 31.42 x 10 3
X = 4.2 + 1.5 + 12 + 0.15 X = 17.85 m = X' D R = 2.4 + 19.2 = 21.6 m = R' D

2 2 Z' D = R' D + X' D 28.02 m
RD = 1.06 hence k 1.05 on the curve XD in figure 9 and therefore the peak Isc is equal to:
I' D =
410 8,448 A 3 x 28.02 x 10 3
hence the peak Isc:
1.05 x 2 x 7,534 11,187 A.

As each level in the calculations makes clear, the importance of the circuit breakers is negligible compared to that of the other elements in the network. V - Reverse currents of the motors It is often faster to simply consider the motors as independent generators, injecting into the fault a reverse current that is superimposed on the network fault current. c fault at C The current produced by the motor may be calculated on the basis of the
2 x 8,448 11,945 A.
To find the peak asymmetrical Isc, the above value must be increased by the contribution of the energised motors at the time of the fault 13 i.e. three times their rated current of 95 A: (3 x 95) x 20 = 5,700 A, hence
RM = (168 + 19.2 + 2.4) 10-3 189.6 m ZM = 876 m hence IM = 410 270 A 3 x 876 x 10 3
I sc =11,945 + 3x95 x 2 x 20 = 20,005 A.

These two results are close to those obtained by the full calculation (11,945 in stead of 11,843 and 20,005 instead of 17,876), and err on the side of safety.
[(
) ]
For the 20 motors IMB = 5,400 A. Again, it is possible to estimate the current injected by all the motors as
3. calculation of Isc values in a radial network using symmetrical components

Z(2) Z(1), except for rotating machines, whereas Z(0) varies depending on each element (see fig. 22). For further information on this subject, a detailed presentation of this method for calculating solid and impedance fault currents is contained in the Cahier Technique n 18 (see the appended bibliography).
advantages of this method

Calculation using symmetrical components is particularly useful when a three-phase network is unbalanced, because, due to magnetic phenomena, for example, the traditional cyclical impedances R and X are, normally speaking, no longer useable. This calculation method is also required when: c a voltage and current system is not symmetrical (Fresnel vectors with different moduli and imbalances exceeding 120). This is the case for phase-to-earth or phase-to-phase shortcircuits with or without earth connection; c the network includes rotating machines and/or special transformers (Yyn connection, for example). This method may be used for all types of radial distribution networks at all voltage levels.
This principle, applied to a current system, is confirmed by a graphical representation (see fig. 21). For example, the graphical addition of the vectors produces, for I 2 , the following result:
I 2 = a 2 I1(1) + a I1(2) + I1(0)

Currents I1 and I 3 may be expressed in the same manner, hence the system:
I1 = I1(1) + I1(2) + I1(0)

I 2 = a I1(1) + a I1(2) I 3 = a I1(1) + a 2 I1(2) 2
+ +
I1(0) I1(0)
Elements transformer (seen from secondary winding) no neutral Yyn or Zyn Dyn or YNyn primary D or Y + zn machine synchronous asynchronous line
Z(0)
free flux forced flux 10 to 15 X(1) X(1) 0.1 to 0.2 X(1) 0.5 Z(1) 0 3 Z(1)
symmetrical components
Similar to the Leblanc theorem which states that a rectilinear alternating field with a sinusoidal amplitude is equivalent to two rotating fields turning in the opposite direction, the definition of symmetrical components is based on the equivalence between an unbalanced three-phase system and the sum of three balanced three-phase systems, namely the positivesequence, negative-sequence and zero-sequence (see fig. 21). The superposition principle may then be used to calculate the fault currents. In the description below, the system is defined using current I1 as the rotation reference, where: c I1(1) is the positive-sequence component; c I1(2) is the negative-sequence component; c I1(0) is the zero-sequence component; and by using the following operator

These symmetrical current components are related to the symmetrical voltage components by the corresponding impedances: V(2) V(1) V(0) Z (1) = ,Z = and Z (0) = I (1) (2) I (2) I (0) These impedances may be defined from the characteristics (supplied by the manufacturers) of the various elements in the given electrical network. Among these characteristics, we can note that
fig. 22: zero-sequence characteristic of the various elements in an electrical network.
positive-sequence I3(1)
negative-sequence
I2(2)
zero-sequence I1(0)
I1(1)
+
I3(2) I3 I2
I1(2)
+
I1
I2(0) I3(0)
I2(1)
=
geometric construction of I1
t geometric constructions of I2 I1(1) I1(0) I2 a2 I1(1) a I1(2)
I1 I1(1) I1(2) I1(0)
I1(2)
fig. 21: graphical construction of the sum of three balanced three-phase systems (positivesequence, negative-sequence and zero-sequence).
a=e
2 3
1 3 between I1,I 2 andI 3 = +j 2 2
calculation as defined by IEC 909

Standard IEC 909 defines and presents a method implementing symmetrical components, that may be used by engineers not specialised in the field. The method is applicable to electrical networks with a rated voltage of less than 230 kV and the standard explains the calculation of minimum and maximum short-circuit currents. The former is required in view of calibrating overcurrent protection devices and the latter is used to determine the rated characteristics for the electrical equipment. In view of its application to LV networks, the standard is accompanied by application guide IEC 781. Procedure 1- Calculate the equivalent voltage at the fault location, equal to
c Un / 3 where c is a voltage factor required in the calculation to account for: c voltage variations in space and in time; c possible changes in transformer tappings;
c subtransient behaviour of generators and motors. Depending on the required calculations and the given voltage levels, the standardised voltage levels are indicated in figure 23. 2- Determine and add up the equivalent positive-sequence, negative-sequence and zero-sequence impedances upstream of the fault location. 3- Calculate the initial short-circuit current using the symmetrical components. Practically speaking and depending on the type of fault, the equations required for the calculation of the Isc are indicated in the table in figure 24. 4- Once the Isc ( I" ) value is known, k calculate the other values such as the peak Isc value, the steady-state Isc value and the maximum, steady-state Isc value. Effect of the distance separating the fault from the generator When using this method, two different possibilities must always be considered: c the short-circuit is away from the generator, the situation in networks where the short-circuit currents do not have a damped, alternating component.
This is generally the case in LV networks, except when high-power loads are supplied by special HV substations; c the short-circuit is near the generator (see fig. 11), the situation in networks where the short-circuit currents do have a damped, alternating component. This generally occurs in HV systems, but may occur in LV systems when, for example, an emergency generator supplies priority outgoers. The main differences between these two cases are: c for short-circuits away from the generator: v the initial ( I" ), steady-state (Ik) and k breaking (Ib) short-circuit currents are equal ( I" = Ik = Ib); k
rated voltage Un LV 230 - 400 V others HV 1 to 230 kV 1.1 1 1 1.05 0.95 1 voltage factor c for calculation of Isc max. Isc min.
fig. 23: values for voltage factor c (see IEC 909).
type of short-circuit three-phase (any Ze)
I" k
General situation Dfaut loign des gnrateurs
c Un 3 Z(1)
c Un 3 Z(1)
In both cases, the short-circuit current depends only on Z(1), which is generally replaced by Zk, the short-circuit impedance at the fault location, defined by Zk = Rk 2 + Xk 2 where Rk is the sum of the resistances of one phase, connected in series; Xk is the sum of the reactances of one phase, connected in series. phase-to-phase clear of earth (Ze = )
c Un Z(1) + Z(2)
c Un 3 Z(1) + Z(2) + Z(0)
c Un 2 Z(1)
phase-to-earth
c Un 3 2 Z(1) + Z(0) c Un 3 Z(1) + 2 Z(0)
phase-to-phase-to-earth (Zsc between phases = 0) Symbols used in this table c phase-to-phase rms voltage of the threephase network = U c modulus of the short-circuit current = I" k
c Un 3 Z(2) Z(1) Z(2) + Z(2) Z(0) + Z(1) Z(0)
c symmetrical impedances = Z(1), Z(2), Z(0) c short-circuit impedance = Zsc c earth impedance = Ze
fig. 24: Short-circuit values depending on the positive-sequence, negative-sequence and zero-sequence impedances of the given network (see IEC 909).
v the positive-sequence (Z(1)) and negative-sequence (Z(2)) impedances are equal (Z(1) = Z(2)); c for short-circuits near the generator: v the short-circuit currents are not equal, in fact the relationship is Ik < Ib < I " ; k v the positive-sequence impedance (Z(1)) is not necessarily equal to the negative-sequence impedance (Z(2)). Note however that asynchronous motors may also add to a short-circuit, accounting for up to 30% of the network Isc for the first 30 milliseconds, in which case I" = Ik = Ib no longer holds k true. Conditions to consider when calculating the maximum and minimum short-circuit currents c calculation of the maximum shortcircuit currents must take into account the following points: v application of the correct voltage factor c corresponding to calculation of the maximum short-circuit currents; v among the assumptions and approximations mentioned in this document, only those leading to a conservative error should be used; v the resistances per unit length RL of lines (overhead lines, cables, phase and neutral conductors) should be calculated for a temperature of 20C; c calculation of the minimum shortcircuit currents requires: v applying the voltage factor c corresponding to the minimum permissible voltage on the network; v selecting the network configuration, and in some cases the minimum contribution from sources and network feeders, which result in the lowest shortcircuit current at the fault location: v taking into account the impedance of the busbars, the current transformers, etc.; v neglecting the motors; v considering resistances RL at the highest foreseeable temperature: 0.004 RL = 1+ ( e 20C) x RL20 C where RL20 is the resistance at 20C; e is the permissible temperature (C) for the conductor at the end of the short-circuit. The factor 0.004 / C is valid for copper, aluminium and aluminium alloys.
equations for the various currents

" Initial short-circuit current Ik The different initial short-circuit currents I" are calculated using the equations in k the table in figure 24.
Ib = I" where is a factor defined by k

the minimum time delay tmin and the I" / Ir ratio (see fig. 25) which k expresses the influence of the subtransient and transient reactances with Ir as the rated current of the generator. Steady-state short-circuit current Ik The amplitude of the steady-state short-circuit current Ik depends on generator saturation influences and calculation is therefore less accurate than for the initial symmetrical current I" . The proposed calculation methods k produce a sufficiently accurate estimate of the upper and lower limits, depending on whether the short-circuit is supplied by a generator or a synchronous machine. c the maximum steady-state short-circuit current, with the synchronous generator at its highest excitation, may be calculated by: Ikmax = max Ir c the minimum steady-state shortcircuit current is calculated under noload, constant (minimum) excitation conditions for the synchronous generator and using the equation: Ikmin = min Ir where Ir is the rated current at the generator terminals; is a factor defined by the saturation inductance Xd sat.
Peak value ip of the short-circuit current In no meshed systems, the peak value ip of the short-circuit current may be calculated for all types of faults using the equation: ip = K 2 I" where k
I" is the initial short-circuit current; k

K is a factor depending on the R / X ratio and defined in the graph in figure 9, or using the following approximate calculation: R K = 1.02 + 0.98 e
-3 X
Short-circuit breaking current Ib Calculation of the short-circuit breaking current Ib is required only when the fault is near the generator and protection is ensured by time-delayed circuit breakers. Note that this current is used to determine the breaking capacity of these circuit breakers. This current may be calculated with a fair degree of accuracy using the following equation:
1.0 minimum time delay tmin. 0.9 0.02 s 0.05 s 0.8 0.1 s > 0.25 s
0.7
0.6
0.5
4 5 6 7 8 " three-phase short-circuit I k / Ir
fig. 25: factor used to calculate the short-circuit breaking current Ib (see IEC 909).
The max and min values are indicated in figure 26 for turbo-generators and in figure 27 for machines with salient poles.
calculation example
Problem Consider four networks, three 5 kV networks and one 15 kV network, supplied via a 30 kV network by transformers in substation E (see fig. 28). During construction of line GH, calculation of the breaking capacity of circuit breaker M is requested. The following information is available: c only the secondary windings of the transformers in substation E are earthed; c for a 30 kV line, the reactance value is 0.35 /km (positive-sequence and negative-sequence conditions) and 3 x 0.35 /km (zero-sequence conditions); c the short-circuit reactance is 6% for the transformers in substation E and 8% for the other transformers; c the factor c for U is set to 1;
c all loads connected to points F and G are essentially passive; c all resistances are negligible with respect to the reactances. Solution c on the basis of the positive-sequence and negative-sequence diagrams (see fig. 29), the following may be calculated:
2.8 2.6 2.4 2.2 2.0 1.8 1.6 1.4 1.2 1.0 0.8 0.6 0.4 0.2 0 0 1 2 3 4 5 6 7 8 three-phase short-circuit current I" / Ir k min max Xd sat
1.2 1.4 1.6 1.8 2.0 2.2
a=
U2 302 = j 3.1 Ssc 290 6 302 U2 = x j 5.4 Sn 100 10
b = usc
c1= 0.35 x 40 j 14 c2 = 0.35 x 30 j 10.5 c3 = 0.35 x 20 j 7 c4 = 0.35 x 15 j 5.25

d = usc 8 302 U2 = x j9 Sn 100 8
60 kV network 290 MVA 10 MVA 10 MVA
fig. 26: factors max and min for turbogenerators (see IEC 909).
E
5.5 5.0 4.5 4.0 3.5 3.0 2.5 2.0 1.5 1.0 0.5 1 2 3 4 5 6 7 8 " three-phase short-circuit current Ik / Ir 0 min max Xd sat
0.6
15 km
30 kV 8 MVA
40 km
5 kV 2 MVA cos : 0.8
4 MVA F
15 kV G 6 MVA cos : 0.8 20 km H M 4MVA 5 kV 2 MVA cos : 0.8 30 km
4 MVA
5 kV 2MVA cos : 0.8
0.8 1.0 1.2 1.7 2.0
fig. 27: factors max and min for generators with salient poles (see IEC 909).
fig. 28.
e=
U2 302 x 0.6 = x 0.6 j 90 S 6
a b E d c4 g f F e G c2 H f g c1 f g b
f = usc
8 30 U = x j 18 Sn 100 4
b' E d' c'4 f' F
b'
U2 302 g= x 0.6 = x 0.6 j 270 S 2

c note on the zero-sequence diagram (see fig. 30): v the delta windings of the transformers in substation E block zero-sequence currents, i.e. the network is not affected by them; v similarly, the transformers in the substations F, H and G, due to their delta windings, are not affected by the zero-sequence currents and, therefore, have an infinite impedance for the fault. b = b1 = j 5.4 c1 = 3 x c1 = j 42 c2 = 3 x c2 = j 31.5 c3 = 3 x c3 = j 21 c4 = 3 x c4 = j 15.75 d = f = c calculations may therefore be made using two simplified diagrams: v with line GH open (see fig. 31): Z(1) = Z(2) = j 17.25 Z(0) = j 39.45 I sc 3 = c Un Z (1) 3 1.104 kA
fig. 31.
c'1 G f'
c3
c'3 H f'
c' 2
fig. 29.
fig. 30.
zero-sequence diagram
positive-sequence and negative-sequence diagram j3.1 j5.4 E j9 j5.25 j270 j 18 j7 H F j14 j90 j18 G j5.4 j17.05
j5.4 E
j5.4
j39.45
j270
H Z(1), Z(2)
j15.75 F j21 H Z(0) G
j42
H Z(0)
Z(1), Z(2)
positive-sequence diagram
zero-sequence diagram
I sc1 =
c Un 3 Z (1) + Z (2) + Z (0)
0.773 kA
j3.1 j5.4 E j9 j5.25 j270 j 18 j7 H F j14 j90 j18 G j10.5 j5.4 j13.05 j5.4 E j5.4 j27.2
Note the network is HV, hence coefficient c = 1.1. v with line GH closed (see fig. 32): Z(1) = Z(2) = j 13.05 Z(0) = j 27.2
j270
H Z(1), Z(2)
j15.75 F j21 H G
j42
H Z(0)
Isc3 = 1.460 kA Isc1 = 1.072 kA

Given the highest short-circuit current (Isc3 = 1.460 kA), the line circuit breaker at point M must be sized for: P = U I 3 = 30 x 1.460 x 3 P 76 MVA.
fig. 32.
j31.5
Z(1), Z(2) Z(1) = Z(2) = j13.05
Z(0) Z(0) = j27.2
4. computerised calculations and conclusion
Various methods for the calculation of short-circuit currents have been developed. Some have been included in a number of standards and are consequently included in this Cahier Technique publication as well. Several standardised methods were designed in such a way that shortcircuit currents could be calculated by hand or using a small calculator. When computerised scientific calculations became a possibility in the 1970's, electrical-installation designers devised software for their particular needs. This software was initially run on mainframe computer systems, then on minicomputers, but was difficult to use, and therefore limited to a small number of experts.
This software was finally transferred to the PC microcomputing environment, proving much easier to use. Today, a wide range of software packages are available which comply with the applicable standards defining the calculation of Isc currents in LV applications, for example Ecodial, a program designed and marketed by Merlin Gerin. All computer programs designed to calculate short-circuit currents are predominantly concerned with determining the required breaking and making capacities of switchgear and the electro-mechanical withstand capabilities of equipment. Other software is used by experts specialising in network design, for
example, research on the dynamic behaviour of electrical networks. Such computer programs can be used for precise simulations of electrical phenomena over time and their use is now spreading to include the entire electro-mechanical behaviour of networks and installations. Remember, however, that all software, whatever its degree of sophistication, is only a tool. To ensure correct results, it should be used by qualified professionals who have acquired the relevant knowledge and expertise.
bibliography
Standards c IEC 909: Short-circuit current calculation in three-phase AC systems. c IEC 781: Application guide for calculation of short-circuit currents in low voltage radial systems. c NF C 15-100: Installations lectriques basse tension. c C 15-105: Guide pratique, Dtermination des sections de conducteurs et choix des dispositifs de protection.
Merlin Gerin Cahier Technique publications c Slectivits des protections Cahier Technique n 13 - F. SAUTRIAU. c Analyse des rseaux triphass en rgime perturb l'aide des composantes symtriques, Cahier Technique n 18 B. DE METZ-NOBLAT c Mise la terre du neutre dans des rseaux industriels haute tension Cahier Technique n 62 - F. SAUTRIAU. c Techniques de coupure des disjoncteurs Basse Tension, Cahier Technique n 148 - R. MOREL.
Other publications c Guide de l'installation lectrique (July 1991 edition), Written Merlin Gerin, Edited by: France Impressions Conseils BP 283 38 434 ECHIROLLES CEDEX c Les rseaux d'nergie lectrique (2me partie), R. PELISSIER. Edited by DUNOD.
Ral. : Sodipe - Valence - Photo. : IPV Edition : DTE - Grenoble 04-95 - 2500 - Printing: Clerc Printed in France
n 159
inverters and harmonics
(case studies of non-linear loads)
Jean Nol Fiorina Joined Merlin Gerin in 1968 as a laboratory technician in the ACS (Alimentations et Convertisseurs Statiques) department where he participated in the performance setting up procedures for static converters. In 1977 he obtained his ENSERG Engineering degree following a 3 years evening course and rejoined the ACS department. Starting as development engineer he was soon afterwards entrusted with projects. He became later responsible for design projects in EPS department (Electronic Power System). He is in some ways the originator of medium and high power inverters. At present he is with the UPS Division where, as responsible for Innovations he works on the preparation of new UPS designs of tomorrow.
E/CT 159 first issued september 1993
glossary
UPS IEC CIGREE PWM D Hn n In Uref Vn Zsn Uninterrupted Static Power Supply - Static UPS International Electrotechnical Commission Confrence Internationale des Grands Rseaux Electriques et Electroniques (International conference on hight voltage electric systems) Pulse Width Modulation global distortion rate individual ratio of harmonics of order n phase angle shift of harmonic component at t = 0 effective current of harmonic component of order n reference voltage distortion factor effective voltage of harmonic component of order n output impedance for harmonic of order n
inverters and harmonics

(case studies of non-linear loads)
summary
1. Introduction 2. Characteristics of non-sinusoidal alternating quantities p. 4 Harmonic analysis of a periodic function p. 4 Effective value of a non-sinusoidal alternating quantity p. 4 Distortion rate p. 4 Power factors and cos 1 p. 5 Distortion factor p. 5 Crest factor p. 5 Relation between current distortion and voltage distortion p. 5 3. Impedances of some conventional Impedance of a transformer p. 6 sources Impedance of an alternator p. 7 Output impedance of an inverter p. 7 Impedance of line p. 11 4. Micro and mini-computer loads Description p. 12 Influence of source impedance p. 12 Calculation of source power for supplying RCD type loads p. 13 5. Conclusion p. 16 Appendix 1: influence of line impedances on voltage distortions p. 16 Appendix 2: input filters in computer/micro-processor equipment p. 18 Appendix 3: bibliography p. 19
1. introduction
Static UPS are virtually perfect electric generators. They are highly reliable and, by nature, ensure (within the performance limits of the battery) the uninterrupted availability of electric power. As regards electrical characteristics, the inverter (which constitutes the UPS generator) possesses from the point of
view of frequency stability as well as voltage stability, performances superior to those of the mains. The only doubtful characteristic is, in the opinion of many engineers, its ability to deliver a sinusoidal voltage regardless of the shape of the current drawn by the load. The aim of this Cahier Technique is to clarify this point and to demonstrate
that modern inverters are excellent generators of sinusoidal voltage even when they supply non-linear loads. This is considered quite normal as UPS are designed and very often utilised to supply computer/ microprocessor systems which draw non-sinusoidal currents.
2. characteristics of non-sinusoidal alternating quantities

n: phase shift angle of harmonic component at t = 0. Global rate of distortion (as defined by CIGREE) This parameter represents the ratio of the effective value of harmonics to that of the fundamental alone:
n=
harmonic analysis of a periodic function

As alternating non-sinusoidal currents and voltages are the main topic of this study, it will be worth while to review the electric quantities in the presence of harmonics. Fourier theorem states that any nonsinusoidal periodic function can be represented by a series of terms consisting: s of a sinusoidal term at fundamental frequency, s of sinusoidal terms whose frequencies are whole multiples of the fundamental (harmonics), s and eventually of a continuous component (DC component). The formula denoting the harmonic analysis of a periodic function is as follows:
n=
effective value of a nonsinusoidal alternating quantity

Applying the general formula
Y rms = 1 T
D % = 100
n=2
Yn 2
Y1
y 2 (t) dt
gives with harmonic representation:

n=
Note: when the distortion rate is low, as is most frequently the case for the voltage, the two definitions lead in practice to the same result. For example, if:
n= n=2
Y rms =
n=1
Yn 2
Yn 2 = 10 % de Y 1
distortion rates
Harmonic rates (as defined in IEC dictionary) This parameter, also called harmonic distortion or distortion factor represents the ratio of the effective value of harmonics (n 2) to that of the alternating quantity:
n=
The IEC expression gives: THD = DF = 100 (0.1) 2 1 + (0.1) 2 = 9.95 %
y (t) = Yo +
n=1
Yn 2 sin (nt - n)
Whereas the CIGREE expression gives:
where: Yo: value of continuous component generally equal to zero and considered as such hereafter, Yn: effective value of harmonic of order n, : pulsation of fundamental frequency,
THD % = DF % = 100
n=2 n= n=1
Yn 2 Yn 2
D % = 100 0.1 = 10 % 1 Hereafter we shall retain for the distortion rate, the expression D which corresponds to a more analytical view of the influence of harmonics on a non-deformed wave.
Individual harmonic rate This parameter represents the ratio of the effective value of a harmonic of order n to that of the alternating quantity (according to IEC dictionary) or to that of the fundamental alone (according to CIGREE), s according to definition in IEC dictionary:
crest factor
As defined by IEC, it is the ratio of crest value to the effective value of a periodic quantity.
relation between current distortion and voltage distortion

For a given voltage source, it is always possible to define an output impedance, even if the latter is frequency dependent. To the extent where this impedance is independent of the current value (linear case) it is possible to calculate for each current harmonic a corresponding voltage harmonic and thus to deduce the individual harmonic rate (percentage). The effective value of voltage harmonic of order n equals: Un = Zsn . In where Zsn: output impedance for harmonic n and In: effective current of harmonic n. The individual rate of harmonics of order n for this voltage is equivalent to:
Hn % = 100
Yn
n= n=1
Yn
s and
according to CIGREE definition:
Hn % = 100 Yn Y1 This latter definition will be retained in subsequent reasoning.
power factors and cos 1

According to IEC, the power factor is the ratio of the effective power P to the apparent power S:
=P S This power factor should not be confused with the phase shift angle factor (cos 1) which represents the cosine of angle formed by the phase elements of fundamental components of voltage and current:
1 = cos 1 = P 1 S1 where: P1 = effective power of fundamental component S1 = apparent power of fundamental component.
Hn = Un U1 where U1 = effective value of fundamental voltage. The global distortion rate of voltage is thus obtained by means of expression:
n-
D % = 100
and also:
n=2
Un 2
distortion factor
According to standard specification IEC 146-1-1, this factor enables to define the relation between power factor and cos 1:
U1
n-
D % = 100
n=2
Hn 2
cos 1
Where voltages and currents are perfectly sinusoidal the distortion factor equals 1 and cos 1 is equal to the power factor.
The input impedance for various harmonic frequencies plays therefore an important role in bringing about the onset of voltage distortion. The higher this input impedance, the greater will be the voltage distortion rate for a given non-sinusoidal current.
3. impedances of some conventional sources
Very often the impedance Zs (at 50 Hz) of a generator is given as percentage of nominal impedance of load Zc:
Zs % = 100 Zs Zc
For the nominal current, the voltage drop across this impedance represents, therefore, as percentage in relation to the nominal voltage, the value of this source impedance:
Zs . In % = 100 Zs . In Un Un
where Zc . In = Un
Zs . In % = 100 Zs . In = 100 Zs Un Zc . In Zc
impedance of a transformer
Figure 1 represents an equivalent circuit diagram of a single phase transformer seen from secondary winding. The transformer impedance consists of an inductance L in series with a resistance R. An equivalent value of the relative impedance is given by the transformer short-circuit voltage Ucc. Indeed, by definition, the short-circuit voltage is the voltage that must be applied across a winding in order to induce a nominal current in the other winding also under short-circuit,
Ucc % = 100 Ucc Un
of UccR is of the order of 1 % to several % (this latter value becoming correspondingly smaller as the power rating of transformer increases). In practice, as regards harmonics, since only the inductance impedance is frequency dependent, it is the inductance alone which determines the behaviour/performance of the transformer. s in three phase transformers, it is necessary to take into account the different possible connection types of primary and secondary windings, as these exert an influence on the source impedance for some harmonics (in particular, third harmonic and multiples of 3). In fact, in the case of a transformer which supplies to each of its secondary windings distorted and balanced currents comprising harmonics of order 3 and multiples of 3, say 3 k, and considering that these currents are balanced, it is thus possible to write for each of these phases: I1 3 k = I sin 3 k t
I2 3 k = I sin 3 k ( t - 2 ) 3 I3 3 k = I sin 3 k ( t - 4 ) 3
The behaviour of a transformer towards these harmonics is therefore dependent on the homopolar impedance Zh of the transformer (refer to Cahier Technique n 18 Analyse des rseaux triphass en rgime perturb l'aide des composantes symtriques). Two types of secondary windings are suitable for not amplifying or reducing harmonic distortions: star connected secondary with distributed neutral
s
L e
fig. 1: equivalent circuit diagram of a single phase transformer seen from secondary winding.
or I1 3 k = I sin 3 k t
Ucc
Uccx = L I2n
I2 3 k = I sin (3 k t - k 2) I3 3 k = I sin (3 k t - k 4)
These equations show that the three currents are in phase. It is this phenomenon which leads one to observe in the neutral conductor of some wiring installations (neon tubes for example) the presence of much higher currents than originally anticipated.
Ucc % = 100 Zs . In = 100 Zs = Zs % Un Zc

This short-circuit voltage is made up of two terms: UccR et Uccx (see fig. 2). in distribution transformers or general purpose transformers with ratings superior to 1 kVA, the value of Uccx ranges from 4 - 6 %, whereas the value
s
Ucc R = R I2n A B
I2cc = I2n
fig. 2: Kapp triangle of a transformer (values referred to secondary).
When primary windings are delta or star connected with the neutral point connected to the source neutral (see fig. 3), the harmonic impedances of order 3 and multiples are neither encouraged nor discouraged (Zh = Zd). The transformer behaves as three single phase transformers. ZIGZAG connected secondary These connections ensure minimum distortion in secondary - in fact, in this case, the harmonic currents of order 3 k do not circulate in the transformer primary, and the impedance Zs is no longer dependent on secondary windings. The inductance is thus very low: Uccx 1 %, and the resistance is reduced roughly by half when compared with the resistance of delta star connected transformer of same rating. Figure 4 and the following calculation explain why currents of pulsating frequency 3 k are not found in the transformer primary (homopolar current equal to zero).
s
As a general rule, the other types of connection are to be avoided, in particular those that do not allow the neutral to be distributed in the secondary; in fact for these Zh = .
Principe of an inverter An inverter comprises first of all a converter referred to as mutator i.e. switching device which converts the DC voltage supplied by a rectifier or a DC battery into AC voltage.
impedance of an alternator
An alternator can also be represented by a voltage source in series with an inductance and a resistance. However, this inductance assumes very different values according to the speed of current variations to which it is related. During such current variation, the equivalent reactance passes progressively from a value called subtransient to its synchronous value via a transient value. These different values reflect the variation of the alternator magnetic flux. As regards current harmonics, only the sub-transient reactance is to be considered in any phenomenon lasting less than 10 ms. This reactance, referred to as longitudinal sub-transient reactance is denoted as X''d. For an alternator of current production, this reactance amounts to 15 - 20 %. In traditional machines but of design optimised in this respect, a value of 12 % can be achieved. Finally, in special machines, some constructors claim values decreasing to 6 %. In conclusion, it is worth recalling that, except in very particular cases, the alternator output impedance is considerably greater than that of a transformer; consequently, the same applies to the voltage distortion rate in the presence of distorted currents.
primary A B C secondary a b c n a b c n
A B C N
For a turn ratio N 2 , N1 the current circulating for instance in the primary winding 1 equals: N 2 (i - i ) 1 3 N1 with i 1 = I 1 3 k = I sin 3 k t i 3 = I 3 3 k = I sin 3 k (t - 4 ) 3 = I sin (3 k t - 4) this gives N 2 (i - i ) = 0 1 3 N1 The ZIGZAG connected secondary winding acts therefore as an attenuator to harmonics of order 3 k. This type of transformer is often used as an output transformer for classic inverters of high rating.
fig. 3: winding connections of three-phase transformers which have a homopolar impedance Zh equal to a direct impedance Zd.
N2 N1
(I 1 - I 3 )
I3
N1
N2
N2 I2
N1
N2
N2 I3
N1
N2
N2
output impedance of an inverter

The impedance of an inverter is essentially dependent on the output impedance of its filter and on the type of regulation adopted.
fig. 4: transformer with ZIGZAG connected secondary and attenuation of harmonics of order 3 k.
In a single phase unit, there are two ways of achieving this conversion: s half-bridge converter (see fig. 5a), s full-bridge converter (see fig. 5b). The square wave voltage appearing between A and B is then filtered so as to obtain in the output of the unit a sinusoidal voltage wave with a low distortion rate. In practice, the switching device (mutator) produces several positive and negative pulses (see fig. 6) which makes it possible to reduce the size of the filter and to have a faster acting voltage regulator. By modulating the relative time intervals corresponding to conduction and non-conduction periods, it is possible to spread the voltage during the period in such a way as to make the conduction time of the switching device practically proportional to the instantaneous value of the fundamental. This principle is called PWM (Pulse Width Modulation) - (MLI in french). The filter inserted behind the switching device (mutator) is generally of the L and C type (see fig. 7). The inverter is therefore a voltage source with the filter impedance in series. Voltage V is the voltage measured at no load, and the impedance consisting of L and C elements in parallel is the impedance measured when terminals A and B are short-circuited (obtained by applying Thevenin theorem; see fig. 8). Classic inverters When the commutation frequency is low, regulation can: s cope with variations of current drawn by user equipment, s compensate for voltage variation of DC battery (or rectifier), s have, however, difficulties in dealing permanently with variations of current due to harmonics generated during half cycle. In these inverters, the output impedance is equal to the impedance of their filter. They can, therefore, be described as classic inverters since
+E + E/2 0 B load A
+E I1 I2 0
I1 I2
I3 I4
+E
VA I 3 closed I 4 open T/2 I 3 open I 4 closed T
+E
VA I 1 closed I 2 open T/2 I 1 open I 2 closed T
t +E
VB I 1 closed I 2 open T/2 I 1 open I 2 closed T
V AB + E/2 t - E/2
+E
V AB
t -E
fig. 5a: principle of switching unit (mutator) half-bridge converter.
fig. 5b: principle of mutator full-bridge converter.
L
VAB
A mutator
fundamental
user (receiver)
fig. 7: output filter of an inverter.

t
fig. 6: output voltage of switching unit (mutator) with 5 pulses per half period.
fig. 8: equivalent circuit diagram of an inverter seen from its output.
operationally they function in the same way as the early design units (due to the limited capacity of semi-conductors to operate at high frequencies). The output impedance of these inverters is therefore frequency dependent and can be represented by the diagram used in figure 9. s thus at low frequencies the impedance of the filter is nearly equal to L. s at high frequencies the filter impedance differs little from
s at
1 . C
resonant frequency
Fo =
1 2 LC
the impedance of the filter assumes a high value that can attain, in terms of magnitude, the value of the nominal load impedance of the equipment (Zs = 100 % Zc). In practice, frequency Fo is therefore chosen so as not to correspond to the possible frequency of current harmonic, i.e. 210 Hz (harmonic currents of order 4 are non-existent or are of very small amplitude). This being the case, various ingenious ways have been devised by constructors in an effort to reduce the output impedance. For example: s additional filters, s special connection circuits for the transformer inserted behind the three-phase switching device (mutator). At first sight, classic inverters have a behaviour towards harmonic currents comparable to that of well designed alternators and therefore less satisfactory than that of transformers. Inverters with PWM and appropriate regulation When the switching frequency of the switching unit (mutator) is high (at least several kHz) and the regulation system allows rapid intervention through the modification to pulse widths during the same period, it is naturally possible to maintain the inverter output voltage within its distortion limits even when dealing with highly distorted currents.
The block diagram of such inverter, shown in figure 10 is as follows: The output voltage Vs is constantly compared with a reference voltage Uref which is sinusoidal and has a very low distortion rate (< 1 %). The voltage difference is then processed by a correction circuit of transfer function C (p) whose aim it is to ensure the performances and the stability of control circuit systems. The resulting voltage issued from this correction circuit is then amplified by the switching unit (mutator) itself and its ancillary control circuit with an amplification gain A. The voltage Vm supplied by the switching unit is shaped by the filter consisting of L and C elements before becoming the output voltage Vs. In practice, one should take into account: s the impedance of the transformer, if present in the circuit, in order to obtain the total value of inductance (often the inductance is integrated within the transformer. That is why it does not appear in circuit diagrams), s the output impedance of the switching unit which according to designs, is not necessarily negligible.
In general, it is then useful to show the whole output circuit part (switching unit + filter) in the form of a series impedance Z1 together with a parallel impedance Z2 (see fig. 11). By applying Thevenin theorem, it is possible to transform the circuit diagram into that shown in figure n 12. Vm = voltage measured at no load thus: V'm = Vm . Z2 Z 1 +Z 2
Zs 1 C L
Fo
fig. 9: variation of output impedance in a classic inverter with frequency.
Uref
+ -
C (p) A
Vm
Vs
Zc
fig. 10: block diagram of a PWM inverter.
Z1
Zs
Vm
Z2
Vs
V'm
Vs
fig. 11: equivalent circuit diagram of a switching unit seen from output.
fig. 12: transformed equivalent circuit diagram of switching unit seen from output.
Zs = measured in output by shortcircuiting V'm, thus: Zs = Z 1 . Z 2 Z1 + Z2 Ratio Z2 is the transfer function Z1 + Z2 of the filter, say H (p): thus H (p) = Z2 Z1 + Z2
Vref
(p) H (p)
V1
Zs
Vs
Zc
fig. 13: transformed block diagram of a PWM inverter.
To simplify still further, it is convenient to replace the product C (p) x A by (p) which represents the transfer function of action chain. The block diagram becomes that shown in figure 13, where Zs = output impedance in the absence of regulation as is the case in classic inverters. When a current is drawn by the load, a voltage drop appears at the terminals of the output impedance Zs, such that: V1 - Vs = ZsI Developing still further: V1 = . (p) . H (p) = Vref - Vs V1 = (Vref - Vs) . (p) . H (p) V1 = Vs + ZsI Vs + ZsI = (Vref - Vs) . (p) . H (p) thus: Vs 1 + (p) . H (p) = V ref (p) . H (p) - ZsI thus: Vs = V ref. (p) . H (p) 1 + (p) . H (p) ZsI 1 + (p) . H (p)
Everything happens as if the output impedance of the inverter were divided by 1 + (p) . H (p). To throw further light on this impedance, it is convenient to carry out additional calculations. In the band-pass of regulation, the product (p) . H (p) being 1, calculations are as follows: 1 + (p) . H (p) (p) . H (p)
Zs classic inverter
PWM inverter
Z's
Zs (p) . H (p)
since Zs = Z 1 . Z 2 Z1 + Z2 and H (p) = thus Z2 Z1 + Z2
fig. 14: comparison of output impedances between classic inverter and PWM inverter in function of frequency.
With PWM inverters, the output impedance remains very low up to high frequencies and the output voltage distortion due to circulating currents, even highly distorted currents, is negligible. Limitation of current The semi-conductors utilized in switching units can deliver a maximum current, above which their performance can no longer be guaranteed. It is therefore advisable to limit the current to this maximum value in order to ensure reliability of performance. As soon as the current drawn by the load exceeds the maximum value set for the inverter, the latter becomes a generator of constant current until the current value required by the load drops below the fixed threshold limit. Under these conditions, the output voltage does not follow the shape of the reference voltage and remains distorted as long as the load current exceeds the threshold limit. This voltage distortion is all the more significant, the longer the duration stage above the threshold limit.
Z's Z 1 . Z 2 . 1 . Z 1 + Z 2 Z 1 + Z 2 (p) Z2
thus
Z's Z 1 (p)
This means that in the band-pass of regulation, the output impedance of inverter is equal to the series impedance of the filter for the whole output circuit divided by the amplification gain of the action chain. Beyond the band-pass of regulation, the output impedance becomes again the impedance of filter which by then becomes the impedance of a capacitor offering a low impedance at high frequencies. Hence the shape of the curve of output impedance in function of frequency (see fig. 14).
The first term represents the result obtained for a conventional control system with no disturbance present. Here, the disturbance is introduced by means of current I circulating in the internal impedance Zs. In the absence of regulation, the term denoting the disturbance would have assumed a value of ZsI. With regulation, this disturbance is limited to:
ZsI 1 + (p) . H (p)
Such cases are met essentially when single-phase loads consisting of a capacitor in front of a rectifier giving a high crest factor. The latter is usually of the order of 3 (crest value 3 times the effective value of current) whereas for a pure sine wave it is only 2 . The performance of the PWM inverter for this type of load is examined in chapter 4.
impedance of line
There is always a length of cable of greater or lesser importance between the voltage source and each user installation. Value of line impedance The line impedance consists essentially of an inductance L in series with a resistance R (see fig. 15). The value of the inductance is hardly dependent on the section of conductors and is usually assumed to be 0.1 /km (at 50 Hz) which is roughly equivalent to 0.3 H/m. The value of the resistance is dependent on the section of the cable and is taken as r = 20 /km for 1 mm2 section. For example, a cable of 16 mm2 section exhibits a resistance of 1.25 /km and a reactance of only 0.1 /km. As a first approximation, it will be possible to represent a cable by its resistance only in the case of small and medium size power rating installations where the use of small section conductors is quite common. Note: for harmonic frequencies, it might be necessary to take into account the skin effect. In this respect, one must remember that in a copper conductor, the equivalent conduction thickness, referred to as skin thickness, is given by the formula:
a (mm) = 66 F (HZ)
Influence of line impedance on voltage distortion Since the line impedance is additional to the source impedance, it has the effect of increasing the distortion rate of the voltage in installations drawing distorted currents. Figure 16 shows an example where an user installation U2 draws a highly distorted current. When this occurs, the distortion rate measured at its input terminals is D2; however, because of the impedance divider consisting of Zs and ZL2, a distortion rate D is measured at the output terminals of the source D being smaller than D2. Consequently, to minimise the influence of receiver installations which generate harmonic currents in other receivers, it is recommended to supply the receiver installations through a special line. Readers interested in further details can refer to appendix 1.
In conclusion Figure 17 below shows the variation of output impedances of various sources of same power rating with frequency.
fig. 15: equivalent circuit diagram of line.
Zs e source
ZL 1 U1 ZL 2
D2
U2
fig. 16: supply to polluter receiver (U2) by means of special line.
Zs % 150 Zc
ratio of output impedance to nominal load impedance
classic inverter
100
alternator X"d = 12 %
50
transformer Uccx = 4 %
Thus, at 50 Hz the skin thickness is 9.3 mm, whereas at 1 kHz it is reduced to 2.1mm. The skin effect must therefore be taken into account for large section conductors which generally carry harmonic currents of high order.
PWM inverter 0 50 250 500 750 F (Hz)
fig. 17: output impedance of different sources in function of frequency.
It is clearly apparent that the PWM inverter exhibits by far the lowest output impedance. To better clarify this point, figure 18 shows three sources, each with the same impedance at 150 Hz. It is thus obvious that the impedance of a classic transformer as well as the impedance of the supply line, must both be taken into account when distorted currents are to be supplied to a load. The PWM Inverter is by far the best generator on the market as regards its ability to minimise the voltage harmonic distortion. It is 5 to 6 times better than a transformer of the same rating.
electric cable I = 30 m S = 6 mm
2
transformer
PWM inverter
~
S = 60 kVA S = 12 kVA
fig. 18: sources exhibiting same impedance at 150 Hz.
4. micro and mini-computer loads
description
These single-phase loads, as many other types of electronic equipment, are supplied by means of switched-mode power supplies. Thus, a load of RCD type (Resistances, Capacitors, Diodes) has been retained in Standard Specification NF C 42-810 to characterise inverters of rating below 3 kVA. A load of RCD type consists of a Graetz full-bridge converter and preceded by a capacitor. The latter acts as an energy storage reservoir in order to supply current to the load between two successive peaks of the rectified voltage. The supply source is represented by a voltage e and an output impedance Zs. In the examples cited in this chapter, the time constant of discharge of the capacitor through the resistance is fixed at 125 ms (see fig. 19). Current i starts flowing when voltage e exceeds the DC voltage U and circulates for a relatively short time to recharge the capacitor to its nominal voltage.
Figure 20 shows the voltages and currents obtained with a relatively low source impedance consisting of an inductance and a resistance such that their short-circuit voltages referred to the load power are respectively Uccx = 2 % and UccR = 2 %. It must be pointed out the distortion rate of the voltage v in the rectifier input is already important as it reaches 7.5 % even despite a low source impedance. The current i starts flowing as soon as the voltage e becomes higher than U but its rate of rise is limited by the source inductance. This inductance extends the time of current circulation when voltage e becomes again smaller than v. It is therefore essentially the value of the source inductance which determines the shape of current i. It is apparent that the current is highly distorted compared with a perfect sine wave and, in addition, slightly out of phase with respect to the source voltage. In this example, the power factor is equal to 0.72.
influence of source impedance

In the previous example it is shown that the load cannot be considered as a generator of harmonic current, but on the contrary, that the current is highly dependent on the source impedance. Figure 21 shows the variation of current i and voltage v in the rectifier input when the source impedance changes from Uccx = 0.25 % to Uccx = 8 %
Zs
R U
fig. 19: basic circuit diagram of micro and mini-computer type load.
fig. 20: currents and voltages of a computer type load of 1 kW with source such that: Uccx = 2 % and UccR = 2 %.
0,25 % 0,5 % 1% Uccx 2 % 4% 8%
whilst the resistive part has been arbitrarily fixed at UccR = 2 %. Table in figure 22 brings to light, for these different impedances, the variation of the different characteristic parameters relating to current and voltage; when the source impedance increases, the power factor improves whereas the distortion rate (see page 4) of the voltage in the input of user installation increases. It is the value of the distortion rate which determines the choice of the source. A distortion rate of 5 % is often the limiting value admissible for receiver installa-tions that can be either polluters or polluted. Curves in figure 23 page 14 show the variation or the global distortion rate of voltage in the input of the rectifier in function of two parameters: s when the short-circuit voltage of the source varies from 0 to 8 %, s for 3 values of resistive short-circuit voltage (UccR = 0, UccR = 2 % and UccR = 4 %). They also show that, in practice, it is the inductive short-circuit voltage that determines the voltage distortion rate except when this short-circuit voltage is lower than 1 %.
calculation of source power for supplying RCD type loads

fig. 21: variation of current and voltage at the computer type load input when the short-circuit voltage Uccx changes from 0.25 % to 8 % while the short-circuit voltage UccR remains constant and equal to 2 %.
Knowing the active power absorbed by the rectifier (Pr), it is essential to choose correctly the power source (Ps) that must supply it.
Uccx
crest factor
power factor
current spectrum
Hn % = 100 I N I1
H3 87 85 81 76 69 63 59 H5 64 60 52 42 29 21 17 H7 38 33 24 14 8 8 8 H9 15 11 6 7 8 6 5 H11 1 4 7 6 4 3 3 H13 7 7 6 3 4 3 2
global distortion rate of voltage
% 0.25 0.5 1 2 4 6 8
I crest I rms
2.7 2.63 2.51 2.35 2.19 2.1 2
=P S
0.64 0.65 0.68 0.72 0.75 0.77 0.78
2.8 3.5 5.4 7.5 11.2 14.2 16.8
fig. 22: variation of principle characteristic parameters of current and voltage for a computer type load supplied from a source of impedance UccR constant and equal to 2 % for values of Uccx varying from 0.25 % to 8 %.
In our development hereafter, the impedance of the supply line is neglected (or integrated into Ucc of source). A first indication is provided by the power factor:
=P S This power factor is dependent on the total short-circuit voltage upstream of rectifier but can be given a mean value of the order of 0.7. Having established this first criterion, the power of the source must therefore be at least equal to the active power absorbed by the rectifier multiplied by 1 or 1.43. 0.7
order of 1 % (in accordance with figure 23), s for a distortion rate of 10 %, a short-circuit voltage of the order of 3 % must be retained. For a transformer s if Uccx = 4 %
s
Note: for a transformer, it is often necessary to take a much higher power ratio considering that distortions can already be present in the network. A distortion rate of 3 % due solely to the working of the rectifiers, leads one to retain a inductive short-circuit voltage of 0.45 % (in accordance with figure 23) which amounts to multiplying by 2.2 the power ratings of transformers to obtain a distortion rate of 5 %. For an alternator As distortion rates of 5 % and 10 % lead to inductive short-circuit voltages of 1 % and 3 % respectively, power ratios of alternator to rectifier are therefore equal to respectively:
for D = 5 % a power ratio of:
Ps = 4 % = 4 is sufficient, Pr 1 %
s for D = 10 % the power ratio would be:
Ps = 4 % = 1.33 Pr 3 %
The second criterion is related to a distortion rate that would be acceptable: s if a distortion rate of 5 % is envisaged, it is possible to retain an inductive short-circuit voltage of the
but in this case, a value at least equal to 1.43 would be required by the power factor. s if Uccx = 6 % s for D = 5 % a power ratio of:
Uccx and Uccx . 1% 3%

If Uccx = 18 %, it will be necessary: s for D = 5 % to have a power ratio of:
Ps = 18 . Pr s for D = 10 % to have a power ratio of:
Ps = 6 % = 6 is necessary , Pr 1 %
s for D = 10 %, a power ratio of 2 is required.
voltage distortion rate as %

15
U CCR = 0 U CCR = 2 % UCCR = 4 %
Ps = 18 % = 6. Pr 3%
For an inverter s classic inverter As it was explained in our discussion on source impedances, this type of inverter of single phase mode exhibits an impedance comparable to that of an alternator of good design (with Uccx of the order of 12 %). As generally the output distortion of an inverter must be limited at 5 %, then it is desirable to retain a power ratio of the order of 12. Inverters of the classic type are available today mostly in three phase version. In these, always assuming a distortion rate of 5 %, the power ratio is 7 when operated with a transformer whose secondary is connected in ZIGZAG.
10
Uccx %
fig. 23: variation of voltage distortion rate at input of microprocessor type load with respect to Uccx and several values UccR of the source.
PWM inverter with appropriate regulation (reminder: its impedance is at least five times lower than that of a transformer for which the power rating must be multiplied by 4).
As long as the current drawn by the load exhibits a crest value lower than the limiting threshold value for the equipment, the distortion rate remains very low and inferior to 5 %. As soon as the threshold limit is reached, the voltage supplied by the inverter becomes distorted (sine wave becomes affected by crest flattening) and the voltage distortion rate increases. Experience shows that, in order to avoid a voltage distortion exceeding 5 %, it is necessary to set the threshold limit for current at 1.5 times the crest value of the nominal effective current of the inverter,
thus I limit = 1.5 2 I rms.
The PWM inverter appears therefore to be the ideal source of voltage for supplying not only loads of RCD type but also all receiver equipment which are generators of harmonic currents (non-linear loads). In the preceding section, we have discussed the case of inverters and
single phase loads; the same reasoning can indeed be applied to three phase equipment providing the equipment is fitted with independent regulation in each phase (this is generally the case with this type of equipment).
Urms Irms power factor crest factor distortion rate apparent power active power
: 220 V : 11 A : 0.61 : 3.6 : 2.7 % : 2.4 kVA : 1.5 kW
The corresponding crest factor of current is then equal to

1.5 2 that is 2.12.
Figure 24 shows the variation (evolution) of voltage and current in a 5.2 kVA inverter with a threshold limit set at: 5,000 . 1.5 . 2 = 48 A . 220 A voltage distortion rate of 5 % is reached here for an apparent power of 5.2 kVA, that is slightly greater than 5 kVA which is its design parameter for rating. The power factor of the RCD load is, in this case very close to 0.8 (0.79) and consequently the inverter does not need to be over-dimensioned in order to supply this type of load (except when the distance between inverter and loads is relatively significant, this being however, true for all sources). In the example shown in figure 24, a 5 kVA inverter is capable of supplying a 4 kW rectifier with a distortion rate inferior to 5 %.
Urms
48A
: 220 V : 20 A : 0.69 : 2.4 :3% : 4.4 kVA : 3 kW
Irms power factor crest factor distortion rate apparent power active power
Urms
48A
: 220 V : 24 A : 0.79 :2 :5% : 5.2 kVA : 4.17 kW
Urms
48A
: 220 V : 29 A : 0.82 : 1.64 : 10 % : 6.3 kVA : 5.2 kW
Thus P inverter = PR = 1.25 PR 0.8 It is worth noting that the fact of limiting the current improves the power factor.
In the preceding paragraph dealing with transformer, it was noticed that, with a power factor of rectifier amounting to 0.7, it was necessary, even in the absence of constraints on the distortion rate, to choose a transformer whose power rating was at least equal to 1.43 PR.
fig. 24: variation of output voltage of 5 kVA inverter with threshold limit set at 48 A.
5. conclusion
Static inverters equipped with PWM are nearly perfect sources of voltage. Besides their qualities as regards voltage stability and frequency stability, they are the best generators on the
market for supplying electronic and micro-processor loads. The high speed response of their regulation systems endows them with a very low harmonic impedance; thus enables
them to supply a low distortion voltage to receivers that are generators of harmonic currents (non-linear loads).
appendix 1: influence of line impedances on voltage distortions
The end of paragraph 3 stresses the fact that it is desirable to supply receivers that are generators of harmonic currents by means of special lines. This is true for loads of RCD type, but also for all receivers utilising power electronics such as rectifiers, battery chargers, speed controllers etc. The use of a special line provides isolation of harmonics through impedance (see fig. 25). For a clean receiver The distortion rate D1 is practically identical to D, and this is all the more true as the impedance of line Z1 is small compared with that of receiver Zp. For a non-linear receiver D2 will be all the more lower as the sum Z2 + Zs will remain low, in other words as the non-linear receiver will have a low power rating in relation to its supply. The following example shows more clearly the influence of Z2 on D and D2. Lets consider a set of micro-computer absorbing 10 kW at 230 V that is being supplied by a cable conductor 100 m long connected to a transformer.
s s s s
characteristics of transformer: 50 kVA (with Uccx = 4 % et UccR = 2 %).
It is necessary to calculate the impedances of the inductive shortcircuit and resistive short-circuit of the transformer but referred to the active power of micro-computers, thus:
U'1ccx = U1ccx . PR Ps U'1ccR = U1ccR . PR Ps
first assuming that Z2 = 0 (load very close to transformer). Curves in figure 23 will give D = 4.6 % = D2. s it is necessary now to calculate D et D2 with a line 100 m/10 mm2 (i.e. 100 m long and a section 10 mm2): s thus short-circuit impedances of the line referred to PR:
s
U'2ccx = l . PR . 100 Un 2 U'2ccR = R . PR . 10 Un 2 thus with: l = 0.1 . 100 = 10 m 1,000 r = 20 . 100 . 1 = 0.2 1,000 10
thus
U'1ccx = 4 % . 10 = 0.8 % 50 U'1ccR = 2 % . 10 = 0.4 % 50
Zs Ls e U1ccx U1CCR Z2 D2 I r non-linear receiver Rs Z1 D D1 Zp clean receiver
characteristics of cable: section: 10 mm2, L = 0.1 /km at 50 Hz, r = 20 /km for a 1 mm2 section.
fig. 25: power supply through a specific line a receiver generator of harmonic currents.
4 U'2ccx = 10 . 10 -3 . 10 . 100 (230) 2 = 0.19 %
distortion rate due to Uccx
U CCR = 0 U CCR = 2 % U CCR = 4 %
4 U'2ccR = 0.2 . 10 . 100 = 3.8 % (230) 2

s
10
total short-circuit impedances:
U'ccx = 0.8 % + 0.19 % = 0.99 %

U'ccR = 0.4 % + 3.8 % = 4.2 %
thus
5
U'ccx = U'1ccx + U'2ccx U'ccR = U'1ccR + U'2ccR

s
the voltage distortion rates DL and DR related to impedances of inductive short-circuits and resistive short-circuits. These values are obtained from curves in figure 26a and figure 26b and are respectively DL = 3.9 %, DR = 3.9 %.
0 1 2 3 4
Uccx %
fig. 26a: voltage distortion rates due to Uccx for various values of UccR.
distortion rate at input of personal computers:

s
2=
(3.9 ) + (3.9 %)
2 %)2
= 5.52 %.
distortion rate due to UCCR
Uccx = 0 Uccx = 1 % Uccx = 2 % Uccx = 3 %
s voltage distortion rates DL and DR at the source:
D L = D' L . U'1ccx U'ccx

D R = D' R . U'1ccR U'1ccR
Uccx = 4 %
thus:
D L = 3.9 % . 0.8 = 3.15 % 0.99
2
D R = 3.9 % . 0.4 = 0.37 %. 4.2

s
voltage distortion rate D at the source

1
D = D L2 + D R2 D = (3.15 %) 2 + (0.37 %) 2 = 3.17 %. in this example, the supply line causes D and D2 to change as follows
s
D from 4.6 % to 3.17 %, and D2 from 4.6 % to 5.52 %.
UCCR %
fig. 26b: distortion rates due to UccR for various values of Uccx.
appendix 2: input filters in computer/micro-processor equipment
Their purpose is to stop the propagation of disturbances caused by switched mode power supplies towards other equipment installations that could be adversely affected. Conversely, they help attenuate some disturbances present in the network which are likely to alter the functioning of electronic and data information equipment. The question is to know if these filters attenuate harmonic currents generated by RCD loads. Interference rejection in network Switched mode power supplies operate at high frequencies in an effort to reduce the size and weight of transformers. In figure 27, the load resistance R is the basic circuit shown in figure 19 is replaced by a transformer and its load. In this circuit, the line current remains identical because of the presence of capacitor C. To achieve silent operation, the switching frequency is always high and in any case in excess of 20 kHz. The commutation times of a transistor (change from non-conducting to conducting state and vice versa) are very brief and do not, in some cases, exceed a few tens of nano seconds. These high frequency commutations (switching) do generate HF interference that is propagated by conduction and radiation. This gives rise to the presence of parasitic interference along the line upstream of the switching device, that is in the mains (on this subject, it is recommended to refer to Cahier Technique n 149 Electromagnetic Compatibility). In order to limit the circulation of these HF currents, constructors of data information processing equipment insert filters upstream of the switched mode power supply unit; a typical circuit of such filters is shown in figure 28. These filters reduce disturbances: s of common mode which affect in the same way both conductors with respect to earth,
s of differential mode which are present between the two conductors. Inductance L offers a high impedance to currents of common mode but practically none to those of differential mode as its windings are wound in opposition. Disturbances of common mode are conducted to earth by capacitors C1 and blocked by inductance L. Disturbances of differential mode are attenuated by capacitors CA and CR which, at high frequency, offer a low impedance between the conductors.
In fact, currents of the order of 1 to 2 mA have been measured by UTE. If a line supplies several electronic and data processing equipment, the sum of the leakage currents can trip the highly sensitive differential residual current device (30 mA) inserted in the line. Filtering of harmonics The filters inserted between the mains and the RCD supply operate efficiently in the frequency band-pass ranging from 10 kHz to 100 MHz. Unfortunately, they are of no use against harmonic currents injected into the mains network. This is due to the fact that harmonic currents produced by RCD supplies are of relatively low frequency: 1 kHz corresponds in fact to a harmonic of order 20 in relation to a fundamental at 50 Hz!
Protection of switched mode power supply The filter inserted between the AC mains and the RCD supply ensures a second function: it protects the RCD supply from impulse type over-voltages and from HF interference of differential and common mode which are present in the mains. Leakages to earth The presence of capacitors C1 causes a leakage current at 50 Hz to flow to earth. Design standards generally specify values of leakage current not to be exceeded (a few milliamperes for equipment connected to a mains point). For example, standard specification IEC 950 relating to data processing equipment recommends that these leakage currents should be kept below 3.5 mA for equipment connected to a mains point.
+ 12 V 0V C
fig. 27: basic circuit diagram switched mode power supply to RCD load.
Imc L C1 mains supply CR CA C1
Imd
PP @@ ,,
Imc
RCD supply
Imd
fig. 28: basic circuit diagram of an antiparasitic interference filter.
Normes IEC 146-1-1 Semi-conductor converters. General requirements and line commutated convectors - part 1-1: specifications basic requirements. s IEC 950 Safety of information technology equipment including electrical business equipment. (NF C 77-210, modification 1 incorporated). s NF C 42-810 Alimentations sans interruption, de puissance nominale infrieure 3 kVA.
s
Cahiers Techniques Merlin Gerin Analyse des rseaux triphass en rgime perturb l'aide des composantes symtriques Cahier Technique n 18 - by B. De METZ-NOBLAT. s EMC: electromagnetic compatibility Cahier Technique n 149 - by F. VAILLANT.
s
Real.: Illustration Technique - Lyon Edition: DTE - Grenoble 09-93 - 3000 - Printing: Lostic Printed in France
n 161
photographie
automatic transfering of power supplies in HV and LV networks
Georges THOMASSET An engineering graduate from the IEG in 1971, he proceeded to conduct research into the design of complex industrial networks in Merlin Gerins Technical Management Department. Following his position as manager of the medium voltage public distribution and hydraulics engineering and design department, since 1984 he is head of the technical section of the industrial unit in the Engineering and Contracting Department.
E/CT 161, first issued march 1996
automatic transfering of power supplies in HV and LV networks
contents
1. Introduction 2. Various types of transfer schemes p. 4 Problems presented by switching p. 4 and precautions to be taken Presence of fault on the p. 4 downstream network Characteristics of the replacement p. 5 source Preparing the switching orders p. 5 Voltage loss during nonp. 6 synchronous switching Mechanical interlocking of p. 6 switching devices in LV an HV Dielectric withstand of switching p. 6 devices in HV 3. Synchronous switching Example n 1 p. 7 Example n 2 p. 7 Example n 3 p. 8 4. Interrupted circuit transfer In low voltage p. 9 Example p. 9 In high voltage p. 10 Example n 1 p. 10 Example n 2 p. 10 5. Pseudo-synchronous switching Principle p. 12 Area of application p. 12 Difficulties p. 12 Ultra-rapid switching p. 13 with phase shift monitoring 6. Summarising table p. 15 7. Conclusion p. 15 Appendix: brief description of a phase comparator p. 16
Nowadays even temporary loss of electrical power is a major handicap for firms whose production processes brook no interruption as well as for high-rise buildings whose safety circuits must be operational at all times. Consequently, transfering of main supply sources onto replacement or emergency sources has proved to be a function used with increasing regularity in electrical distribution, public and private alike. This Cahier Technique begins with an examination of the difficulties of implementing such switching devices, together with the possible technical solutions. This section is followed by a presentation of the various switching types backed up by practical examples. Finally, a table summarises the above and gives the main areas of application.
1. introduction
The purpose of installing source switching devices is to guarantee the continuity of power supply of certain priority loads, for example to ensure protection of persons or to maintain production cycles. Such devices are put into operation either by failure of the main power supply or by voluntary operator action. Switching devices are particularly used: c to supply: v computers, v high-rise buildings, v lighting and safety systems: airfield lighting, premises receiving general public, etc, v main auxiliaries of thermal power plants,
v complete manufacturing lines whose process brooks no interruption, even temporary, of any element in the line (iron and steel, petrochemical, etc...); c in medium voltage public distribution for: v switching of HV lines and transformers in source substations, v supply of HVA/LV-HV level A (U i 50 kV)/Low Voltage (U i 1 kV) dual incomer substations. The devices using these power transfers are highly varied. For example, in power circuits, switching devices are electromechnical or static contactors, circuit-breakers and switches, in high and low voltage alike.
These devices can be operated: c manually: such devices are the most elementary and economic. They require intervention of an operator, and the time needed to change from the defective source to the replacement or backup source can be very long (the operator has to move place); c automatically: these devices are the fastest and the most commonly used. Nevertheless the basic configuration can, in most cases, be simply a main power supply, a replacement or backup source and a busbar which is the point common to both sources from which the loads are supplied.
2. various types of transfer schemes
There are three main source switching types, namely: c synchronous: transfer time: zero (e.g. generator coupling); c interrupted circuit transfer transfer time: 0.2 to 30 s (e.g. main/emergency function in LV); c pseudo-synchronous transfer time: 100 to 300 ms (e.g. reacceleration of asynchronous motors). For switching, a certain number of prior conditions must be met, some of which call for special precautions.
problems presented by switching and precautions to be taken

To construct an installation containing source transfers complying with the continuity of service requirements of its users, a number of factors, calling for special precautions, must be examined in the design stage: c presence of a fault on the downstream network; c characteristics of the replacement source; c preparing the switching orders; c voltage loss during switching (for non-synchronous switching);
c mechanical interlocking of switching devices in LV and HV; c dielectric withstand of switching devices (in HV).
presence of a fault on the downstream network

When failure of the main power supply is caused by a fault downstream of the supply transfer location, we recommend you do not switch the supply sources. Rather, the circuit controlling the switching device should be blocked by an order from the downstream network protection system.
characteristics of the replacement source

The rated power, short-circuit power, connection impedances and earthing system of the replacement source may vary considerably from the main source. Thus, for example, the main source may be an 800 kVA, 380 V, 50 Hz, Isc = 20 kA transformer, where as the replacement source is an 200 kVA generator set, having a transient short circuit current of 1 kA. The devices protecting against phaseto-phase and phase-earth faults in the backed-up network may thus cease to function in certain conditions when this network is supplied by the replacement (or backup) source. Extreme care must be taken when choosing and setting the protective devices in order to find a protection system compatible with the electrical characteristics and the operating and maintenance modes of both supply sources. Two cases should be emphasised at this point: the reacceleration of certain motors and the resumption of supply of several loaded step-down transformers: c the network load includes a number of motors. When the replacement source is a low power one, after the main source has switched to the replacement source, the inrush current and the permanent working current must be limited by: v partially shedding loads, v staggering restarting of priority motors in the event of an interruption. If these measures are not taken, in view of the low power replacement source, voltage drops would be extremely serious and motor reacceleration impossible (driving torque less than mechanical load torque). c energising of a number of step-down transformers in the distribution network. When switching takes place in HV, allowance must be made for the inrush currents of HV/LV transformers which are some 10 to 15 times their
rated current. In actual fact, if the replacement source is an LV generator set, its generator cannot supply currents that high at rated voltage and acts as though it were supplying a short-circuit. It thus delivers a very low voltage for the first few moments after switching which does not simplify motor restarting. Consequently, it is preferable to trip all the step-down transformers on the HV side before switching, and then to re-energise them one after the other.
preparing the switching orders

The input for these orders comes from voltage monitoring: c if the replacement source is a generator set: loss of voltage on the main power supply to initiate startup of its motor; c presence of stabilised voltage of replacement source to authorize transfer to the replacement source; c presence of voltage on the main supply source to return to the normal position. The switching orders c to transfer from the main to the replacement source. Loss of or drop in main supply voltage may be: v permanent, due to: - tripping of the upstream protective device, - excessive overloading of the network causing a large voltage drop, - etc, v but also temporary, due to: - operation of the rapid or slow reclosing controllers of the electricity boards overhead lines, - a short-circuit between phases, normally eliminated by the protection devices, - etc. The action of the main source undervoltage detector will thus generally be delayed so as not to switch sources due to temporary drop
in or loss of voltage. Moreover, if the replacement source is a generator set whose starting order is given by loss of main source voltage, the sets voltage must first stabilise before a switching order is given (a few seconds). c to transfer from the replacement to the main source. Resumption of the main supply may be preceded by attempts to re-energise the main line, in order to: v locate a fault, v perform source looping further to an incident, v carry out tests following repair or modification of the main line. The action of the main source voltage presence detector will thus be amply delayed (a few dozen seconds to a few minutes). Note: a) devices used to switch from the main to the replacement source without returning to the main source when it is once more present, are normally known as changeover switches; b) devices automatically returning to the main source are known as main/ emergency changeover switches. The difficulties in detecting loss of main supply voltage c residual distribution network voltage on loss of supply source. Although the supply source has failed, distribution network voltage can be sustained by: v the residual voltage generated by asynchronous motors during deceleration for a period of approximately 0.3 to 1 s, v the voltage induced at the terminals of synchronous motors during deceleration, v the voltage due to discharging of capacitors connected to the network. In the case of rapid source switching, this sustained voltage prevents rapid detection of loss of main supply voltage by ordinary conventional devices such as threshold voltage relays.
c temporary loss of voltage during which the transfer must be blocked. These voltage losses are caused by network controllers such as rapid and/or slow reclosers and switching of HV transformers or lines in HV transformer substations, etc. The same applies to network undervoltage due to excessive voltage drops caused by transient overcurrents (phase-to-phase or phase-to-earth fault eliminated by discriminating network protection devices, startup of large motors, etc...). c choice and cabling of detectors with: v one single-phase relay Normally, only one single-phase detection relay is connected between two phases of the main incomer. In this case, the third phase may fail without being seen by the relay. Switching therefore will not take place, and power supply to the loads remains defective. This system is satisfactory only if the three-phase power supply is unable to operate for any length of time with two phases only. This calls for the use of protective devices such as three-pole circuit-breakers or fuses with striking-pin delivering a multi-pole breaking order. Otherwise, to prevent this problem, either two relays connected between different phases or three relays in a delta connection must be fitted. v three single-phase relays The above connection (three relays in a delta connection) may however prove tricky when relay thresholds are set between 20 and 30 % of rated voltage. In point of fact, if only one phase fails, the two relays with a terminal connected to this phase are then series-connected and are supplied by the two remaining sound phases. The voltage at the terminals of these two relays equals half their rated voltage, a voltage greater than the setting value (0.2 Un). No switching order is then sent. For this reason it is preferable to use three star-connected relays or three delta-connected relays which are set to 60 % of Un or, better still, a rotating field three-phase voltage relay. v a single three-phase voltage relay This type of relay does not allow detection of supply phase failure at busbar level if the consumer network
contains asynchronous three-phase motors, since these motors sustain the voltage of the broken phase at busbar level. A three-phase rotating field overcurrent relay must then be connected to the main incomer. c detector mounting Instantaneous electromagnetic relays are vulnerable to impacts which cause their contacts to bounce and to send false switching orders. This is particularly true when relays are mounted on doors; particular care must thus be taken with this type of mounting to avoid all vibrations likely to interfere with equipment operation.
mechanical interlocking of switching devices in LV and HV

With the exception of synchronous switching devices for which the two switching elements (main and replacement source) may be simultaneously closed, mechanical interlocking of the devices and electrical interlocking preventing simultaneous supply of the two devices control circuits are recommended in all installations and are normally a standard requirement of electricity boards.
voltage loss during nonsynchronous switching

Such voltage loss, although temporary, is normally quite sufficient to de-energise all the contactors whose coils are supplied by the power circuit. Automatic transfer systems may lose their efficiency to a large extent since the loads controlled by these contactors, now open, cease to be supplied. However, manual staggered resumption of motor operation is possible using the on push buttons. In preference to manual resumption, contactor coils can be supplied by a reliable auxiliary source (battery or rotating set with an inertia flywheel), or a slow release relay can be used. Another possibility is a capacitor connected in parallel with the coil and charged by the power supply via a rectifier. In this case, the energy required to keep the contactor in the closed position is provided by the capacitor for the brief instant of voltage loss. However, to avoid stretching the capacity of the buffer capacitor, the undervoltage time must be relatively short (a few hundred milliseconds) and coil consumption low. Note that if these solutions are to be implemented, the replacement source must be able to take up all the loads and particularly all the motors in the reacceleration. Note: When a coil control circuit opens, the high voltages induced at the coil terminals must be withstood by the rectifier and the capacitor.
dielectric withstand of switching devices in HV

The dielectric withstand of the switching device of the replacement source used in such synchronous and pseudosynchronous switching systems must be particularly appropriate. In point of fact, during coupling conditions, the poles of these devices may be submitted, between input and output, to twice the phase-to-neutral voltage of the network (voltage of the two sources to be coupled in opposition of phases).
3. synchronous switching
Both the main source and the replacement source can be synchronized, namely: c their voltage vectors are in phase; c and their frequency and amplitudes are identical. The possibilities offered by this switching type are numerous in that source changeover takes place before voltage loss of the source in operation; a major advantage is that loads experience no loss of power supply. The following examples illustrate this type of transfer.
example n 2
Substitution of a generator in operation by a standby device in a self-powered electricity producing plant consisting of generator sets. These generator sets require frequent periodical servicing; the plant is run with n - 1 sets, the nth being either on standby or undergoing servicing. A set in operation is replaced by the standby set as follows: the generator of the standby set is brought up to synchronous speed and to rated voltage; the circuit-breakers energising order is sent once coupling conditions have been checked (frequency is equal, voltages have same magnitude and are in phase). In order to obtain such conditions for coupling and to maintain them after coupling, the generator and the thermal motor are equipped with a voltage and speed regulator, respectively. The coupling conditions are obtained: c either by intervention of an operator giving, according to readings on the differential voltmeter and frequency
example n 1 (see fig. 1)

Operating an EHV/HV interconnection substation with double busbar. The 2 busbars are supplied by the EHV transmission lines of the interconnection network, the coupling circuit-breaker is open and the two busbars are synchronized. The line and transformer feeders are connected to either busbar A or B. To change the supply of a feeder (changing the busbar), assuming that it is currently supplied from A, simply: c close the bus coupling circuitbreaker 1 without checking coupling conditions as both busbars are synchronized; c close the second disconnecting switch 2b for the feeder in question; c open the first disconnecting switch 2a ; c open the bus coupling circuitbreaker 1 . The feeder is then supplied by the other busbar B. Note: Throughout switching all incomers are connected in parallel on the two coupled busbars; short-circuit power is then high, and the equipments electrical characteristics must be sufficient for this mode of operation if the likelihood of a fault occurring during the transfer is not negligible.
meter and on the synchroscope, the speed orders to the speed regulator, the excitation orders to the voltage regulator and the energising order to the circuit-breaker when coupling conditions are met. In this case a controller, known as a coupler, can be used. This device is designed to check coupling conditions and to give the energising order; regulators are always ajusted manually. c or using a synchrocoupler which is a special controller associated with a voltage regulator. It sends the speed orders to the motor and the energising order to the generator circuit-breaker. The regulator sends the excitation orders to the generator. Coupling is then fully automatic. After coupling, the set to be placed out of operation is discharged (using the speed regulator) and disconnected from the network by opening its coupling circuitbreaker. Substitution is thus completed without disturbing the distribution network and without voltage loss.
incomers
busbar A busbar B 2b 2a 1
bus coupling
feeders
fig. 1: diagram showing an EHV/HV interconnection substation with double busbar.
example n 3 (see fig. 2)

Automatic no break transfer without breaking of an UPS (Uninterruptible Power Supply) on the public network (or mains) using an SS (Static Switch). This requirement is a common one: supply of computers, computer management centres, measurement, process control, etc. The SS is a device enabling the mains to be used to back up the UPS. It stands out by the fact that no voltage loss, even transient, occurs on
disappearance (accidental or deliberate) of UPS voltage. To obtain this result, the UPS permanently ensures its voltage is in phase with the mains voltage. However, switching is only possible if the mains voltage level is correct. The operating sequences are as follows: c the UPS, its voltage regulated, its frequency controlled and in phase with the mains, supplies the load. The SS is open and the mains delivers no power; c the UPS-mains changeover takes place:
UPS main source rectifier inverter
mains replacement source
v on an UPS shutdown: - either due to an internal fault, - or operator-initiated, v on detection of an overload. The switching order always instantaneously closes the static switch. In the case of overload, the two sources run for a short instant in parallel before the UPS is disconnected. c the mains-UPS changeover is operator-initiated. Following UPS startup, the automatic cycle is as follows: v the UPS is synchronised with the mains, v the mains and the UPS are connected in parallel, v the static switch opens, v the UPS is then permanently synchronised on the mains and alone supplies the load. Importance of the static switch The ac SS ensures: c permanent supply of the load, comparable in terms of reliability to the solution with two parallel-connected UPS, one backing up the other, but at lesser cost; c in the event of overload, an ultrarapid device controls the static switch connecting the mains in parallel to the UPS. The mainss short-circuit power can thus be profitably used to eliminate the downstream faults without taking any special precautions other than standard discrimination rules.
static switch
load
fig. 2: diagram showing a circuit supplied by two sources with automatic no break switching using a static switch.
4. interrupted circuit transfer
This type of source switching is the most common, in LV and HV alike, with an area of application extending from industry to the service sector. Switching times normally vary from 0.5 to 30 seconds, while not ruling out lower values for special cases.
example
Diagram The most common diagram for Low Voltage source switching with automatic load shedding is presented in figure 3. Operating principle Neither source can be coupled, and circuit-breakers Jn and Jr are mechanically locked: their electrical control mechanism is interlocked in such a manner that a closing order sent simultaneously to both circuit-breakers will only close one of them. A three-position selector controls the device: position 1 = off, position 2 = automatic, position 3 = voluntary closing of the replacement source circuit breaker and tripping of the coupling circuitbreaker. c position 1 = off All the control circuits are de-energised and all the circuit-breakers are open. c position 2 = automatic v main network voltage is present, and the corresponding circuit-breaker and coupling circuit-breaker are closed, v for a voltage loss exceeding 0.4 second (up to 10 s), the switching controller sends tripping orders to the
in low voltage
Devices must be extremely simple as the electricians operating the LV networks are normally not specialists. Switchgear The type of switching device chosen depends on switching frequency: c large number of switchings: contactor; c small number of switchings (one a week): circuit-breaker. Control circuit The control circuits of switching devices are supplied either by a backed-up auxiliary source (e.g. battery) or directly by the power circuit of the device to be controlled. Power supply The main supply source is normally the Low Voltage public distribution network (or mains) or a private Low Voltage network from an HV/LV transformer supplied on the HV side by the public distribution network. The replacement source may be: c a second LV network separate from the first; c a no-break generator set for rapid resumption of operation; c a generator set with manual or automatic startup on undervoltage at the main source; c an UPS; c etc. These various replacement sources, in most cases far less powerful than the main source, have a limited backup time. When the backed-up network is supplied by the replacement source, it is thus often wise and sometimes even vital to shed some loads and to restart only ultra-priority motors (see chapter 2).
main and coupling circuit-breakers, and a starting order to the generator set, v on receipt of the message gen set voltage OK, the controller sends an energising order to the circuit-breaker of the replacement source; v on resumption of main voltage, and after a time delay of 10 to 180 seconds: - circuit-breaker Jr opens, - circuit-breaker Jn closes, - the set receives a stopping order, - circuit-breaker Jc closes; c position 3 = voluntary energising This allows voluntary tripping of the coupling circuit-breaker, with the backed-up network supplied by the replacement source. A special case: reacceleration of LV motors A controller for restarting motors should be provided when the backed-up network supplies a large proportion of motors which must be reaccelerated as quickly as possible after loss of the main supply. This common need is particularly required by the process and safety of persons and property. In actual fact, on loss of the main source, if no special measures have been taken, all contactors will open. When the supply is resumed, none of the contactor-controlled loads are supplied. On the other hand, if the motors are
replacement source main source Un Jn Ur Jr
generator set
Jc
non-priority loads
priority loads (backed-up network)
fig. 3: diagram for low voltage source switching with automatic load shedding.
protected and controlled by circuitbreakers, they will simultaneously restart when voltage is restored. If the contactor coils of the motor feeders are supplied by the ac network, use of slow release relays (see chapter 1) also enables their operating order to be maintained during voltage loss. Residual voltages of asynchronous motors during deceleration normally present no risk on resumption of power supply, since at that moment (t = + 500 ms), the amplitude of these voltages is less than 20 % of Un, a value tolerated by manufacturers for resumption of supply in phase opposition. However the main (overcurrent) protection device of the distribution network may be tripped by the sum of the motor restarting currents.
and c there are no faults in the installation. Operation The controller has two high impedance inputs: a Main input, connected to a capacitive divider connected between one of the main networks phases and the earth, and an Emergency input connected in like manner to one of the phases of the replacement or emergency network. Voltage loss on the Main input controls a time delay relay , t1 (0.1 s to 1 s) which, at the end of the cycle, sends a temporary opening order to the main network switching device and a temporary closing order to the replacement (emergency) source switching device. If voltage is lost on the replacement (emergency) source, the t1 time delay relay is locked by the Emergency input and switching cannot take place. Note that the RCV420 controller has a second Emergency input in the form of a loop in which the contact of an external voltage relay can be inserted to prevent switching when it is open. If voltage is restored on the Main network, a second time delay relay, t2 (10 to 100 s) is energized and, at the end of the cycle, sends a temporary opening order to the Emergency network switching device and a temporary closing order to the Main network switching device. Faults in the installation must be detected by an external device fitted with a closing contact which prevents switching using the fault input of the controller.
c one electronic controller of the RVH type (see fig. 5). Principle A dual incomer assembly (see fig. 6) can be supplied by either incomer as the system is completely reversible. Detection of voltage presence or loss is identical to the Main-Emergency device. Detection of phase-earth or phase-tophase faults on the load network is performed by the controller informed by toroids current transformers (three toroids per incomer). In normal operation, switching is dependent on certain electrical conditions and is performed after certain manual or automatic operations described below: c manual switching The operator manually opens switch A then closes switch B, having first checked that the following conditions are met: v no voltage on circuit A, v voltage present on circuit B, v no faults in the substation (downstream network). The conditions for resumption of the main supply are: v checking there are no faults in the substation, v manual opening of switch B, v manual closing of switch A. c automatic switching The controller opens switch A, then closes switch B if the following conditions are met: v no voltage on circuit A, v voltage present on circuit B, v no faults in the substation, v presence of this information for 5 or 30 consecutive seconds. The main purpose of these 30 seconds is to wait for the end of the cycles of the automatic reclosers used on overhead networks. Restoration of voltage on circuit A does not automatically cause switching from circuit B to A, although manual tripping by the operator is possible.
in high voltage
Source switching taking place at high voltage involves very high powers and, as such, must offer still greater guarantees as to the overall operational dependability of the device. HV switching is particularly used when there is a large number of priority loads or when these loads are not supplied by a distribution sub-switchboard. The switching device control orders are prepared by standard electronic controllers. The two examples given below illustrate this switching type.
example n 1
Main-Emergency device These devices have one switching device per incomer and an RCV420 type controller (see fig. 4) Principle The controller manages the operations. Its function is to detect voltage loss on the main source and to automatically control switching of the load to a replacement source when the following two conditions have been met: c voltage is present on the replacement source;
example n 2
Dual incomer device This device is extremely popular in France for the dual incomer supply of HV/LV transformer substations directly connected to the HV public distribution network. It consists of: c one switching device per incomer (HV switch);
incomer N incomer R
1 0 1 0 t1 t2
network supplied by
1 0 N R t1 and t2 adjustable N
fig. 4: Main-Emergency RCV420 controller and its operating sequence (Merlin Gerin).
incomer A incomer B
1 0 1 0 t t
network supplied by...
1 0 A B t = 5 or 25 s A
fig. 5: Dual incomer RVH215 controller and its operating sequence (Merlin Gerin).
HV
network A network B A P
HV/LV transformer P = changeover switch
HV/LV transformer
fig. 6: diagram showing dual incomer distribution.
5. pseudo-synchronous switching
principle
This type of source switching lasts roughly 150 ms. The most common diagram is shown in figure 7. In normal operation, the two halfbusbars are supplied by the two incomers respectively: the tie circuitbreaker is open. Failure of either source triggers startup of the rapid switching device which, in under conditions, sends two orders: c a closing order to the tie circuitbreaker; c an opening order to the circuitbreaker of the faulty source. The half busbar of the faulty source is thus re-energised.
area of application
The standard case is that of an installation connected to two HV sources and mainly consisting of asynchronous motors. The operating imperatives of the machines driven by the motors mean the latter cannot afford to stop, even temporarily, or slow down during transfer from the main to the replacement source. This switching type is especially popular in chemical and oil plants and, more generally, in industries whose manufacturing processes brook no interruption, even temporary, of an element in the line. It is also used to supply thermal power plant auxiliaries.
The maximum amplitude of this residual voltage decreases exponentially as a function of time, with a time constant depending on: c motor power; c operating state of the stator connections: v open stator, case of a three-phase supply failure, v short-circuited stator, case of a threephase fault on the supply. On the other hand, the rated supply voltage of the motor only slightly modifies the value of the time constant. The table in figure 8 provides the approximate values of the time constants for extinguishing the residual flow for medium cage asynchronous motors. If no special precautions are taken, rapid resumption of motor power supply during deceleration may lead to coupling in phase opposition between the replacement source and the load
network whose voltage is maintained by the asynchronous motors. Only high voltage epoxy resin insulated motors can support resumption of supply in phase opposition. Note, however, that in this case the peak current equals roughly 3 times motor starting current, i.e. 15 to 20 In, with the result that the entire distribution network is seriously disturbed: c voltage drop and large, repetitive electrodynamic forces; c nuisance tripping of circuit-breakers by full short-circuit protection; c etc. For these reasons, ultra-rapid resumption of motor supply should not be done without a prior comparison of the phases of the source voltage with residual voltage. Using a voltage phase shift comparison device, however, allows ultra-rapid switching. A brief description of a phase comparator is given in the appendix.
source 1
source 2
J1 HV busbar Jc
J2
M
M: asynchronous motors
fig. 7: diagram usually applied for pseudo-synchronous switching.
difficulties
The main problem of this switching type is that a three-phase asynchronous motor sustains at its terminals, during deceleration on loss of power, a threephase ac voltage with decreasing frequency and amplitude induced by the motors residual flux.
motor power time constant, short-circuited stator time constant open stator
10 kW 0.02 s 0.3 s
100 kW 0.03 s 0.4 s
200 kW 0.04 s 0.6 s
400 kW 0.06 s 1.1 s
800 kW 0.1 s 1.5 s
fig. 8: time constants for extinguishing residual flux for medium cage asynchronous motors.
ultra-rapid switching with phase shift monitoring

The ultra-rapid switching operations, which are possible, are shown in the three diagrams in figure 9. Sequence A: The switching order trips separator circuit-breaker J1 or J2; after opening, the phase comparator is put into operation and, when switching conditions are favourable, sends an
energising order to the tie circuit-breaker Jc. Sequence B: The switching order trips separator circuit-breaker J1 or J2 and energizes the phase comparator. When switching conditions are favourable, the phase comparator sends an energising order to the tie circuit-breaker Jc. Sequence C: The switching order energizes the phase comparator. When switching conditions are favourable,
the comparator simultaneously sends an energising order to the tie circuitbreaker Jc and a tripping order to separator circuit-breaker J1 or J2. Note: a) sequence C is the one normally chosen as switching times are shortest. b) certain problems may arise when preparing the switching order, such as: c detection of real loss of main supply voltage in presence of residual voltage; c relay speed; c etc.
sequence A switching order phase comparator coupling circuit-breaker Jc separator circuit-breaker J1 or J2 breaking time reclosing time sequence B switching order phase comparator coupling circuit-breaker Jc separator circuit-breaker J1 or J2 breaking time reclosing time sequence C switching order phase comparator coupling circuit-breaker Jc separator circuit-breaker J1 or J2 pending synchro energising time breaking time reclosing time end of switching pending synchro energising time end of switching pending synchro energising time end of switching
fig. 9: possible ultra-rapid switching operations.
Conditions be met by the network for rapid coupling The first condition to be met when coupling actually takes place is expressed by the inequation:
Us Um < Un + Ur
(see fig. 10)
Us
Us - U m
IU nI + IUr I
Um
Us = voltage of source resupplying motors
coupling meeting this inequation, the speed of all motors is less than their rated speed, meaning motors absorb a current exceeding their rated current. To increase chances for successful resumption of motor supply on rapid switching, the following conditions are thus required: c the speed reached by motors, on resumption of supply, must be as high as possible. Speed depends on: v length of undervoltage time, v inertia of rotating masses, v load torque during deceleration; c the supply network voltage drop must be slight. This drop depends on: v impedance of the electrical circuits, v the current absorbed by the motors, v the number of motors reaccelerated; c the value of the driving torque during resumption must be (far) greater than load torque.
Driving torque depends on: v rated value of driving torque on full voltage, v the form of torque in the speed range between speed at which resumption takes place and rated speed, v the voltage applied at the motor terminals. Note that if motor slip is considerable on resumption of power supply, throughout the restart period, the current absorbed by the motors is constant and approaches, in an initial approximation, starting current (the curve of current absorbed by an asynchronous motor as a function of rotation speed, is relatively flat). Diagram showing the rapid switching device (see fig. 11)
(replacement source).
source 1 Us 1 Du J1 tripping cdp UM UM Du tripping Us 2
source 2
Um = voltage at motor terminals after
separation of their first source (residual voltage).
Un = rated voltage of motor. Ur = acceptable residual voltage of motor. Us Um < Un + Ur
J2
fig. 10: electrical variables and inequation conditioning success of rapid coupling.
HV busbar
energising Jc
As a rule motors can support a coupling in phase opposition, after separation of their first source, provided that residual voltage at their terminals does not exceed the value Ur equal to 25 % Un. This initial condition, although necessary, is not sufficient to ensure successful reacceleration of motors. In point of fact, despite monitored
M
Du: undervoltage detector cdp: phase comparator M: asynchronous motors
fig. 11: diagram showing the rapid switching device to supply to MV networks.
6. summarising table
switching type synchronous application examples - busbar switching, - substitution of one generator by another, - switching an UPS to the mains
switching time devices used (examples of Merlin Gerin equipment) observations
zero - coupler, - synchrocoupler, - UPS unit with static switch (EPS 2000 and 5000) switching must take place before main source voltage is completely lost
interrupted circuit transfer LV from industry to service sector: - supply of pumps, - supply of auxiliary circuits of a transformer substation, - supply of hypermarkets, - etc 0.5 to 10 s - automatic source changeover switch with circuit-breaker (Compact and Masterpact)
pseudo-synchronous HV - supply with 2 switchable HV incomers, - supply by one main source and one replacement source reacceleration of asynchronous motors
1 to 30 s - cubicle assembly with changeover switch (VM6, DDM and NSM)
0.06 to 0.3 s rapid HV circuit-breaker associated with a phase comparator
difficulties in preparing switching orders (presence of residual voltage)
7. conclusion
This description of operating conditions to be met and technical requirements to be considered leads us to the following practical conclusion. Before choosing a switching device, it is preferable and indeed vital to know: c the quality of the main and replacement sources: v amplitude, duration and frequency of voltage drops,
v duration and frequency of voltage loss (whether or not service resumption controllers are present upstream, e.g. rapid or slow reclosers), v available power; c load requirements as regards continuity of supply: v no supply failure tolerated, v voltage loss tolerated (0.3 s, 1 s, 30 s, etc).
appendix: brief description of a phase comparator
The phase shift of the two sources is measured by subtracting the two voltage vectors, Us and Um, i.e.:
Us Um = U
Where
Us = Us sin 1t
Um = Um sin 2 t
When the envelope curve voltage drops below a set threshold, the comparator sends a switching order until the envelope curve voltage once again exceeds the threshold value. If there is a great difference between the two frequencies, the comparator

prevents switching since, in this case, the conditions for ultra-rapid switching are not favourable. However, if residual voltage is less than a preset value (e.g. 0.20 to 0.60 Un), switching takes place despite a large difference in frequency.
(replacement source voltage)
a) U = Us Um
(residual voltage) Note: the writing

Us = Um = Ua
envelope curve
simplifies calculations Using the trigonometric formula

sin sin = 2 cos
+
2
sin
the following can be written:

Us Um
b) rectified U
= 2 Ua cos
1 + 2
2
t sin
1 2
2
t=U
The pulsation voltage envelope 1 + 2

2
has a pulsation beat expressing phase shift
1 + 2
2
c) comparaison between Isc rectified U and threshold voltage
evolution in time between Us and Um (see fig. 12).

The voltage Us Um is rectified and
U rectified, filtered signal
filtered. The instantaneous values of the envelope curve voltage, or beating voltage, thus obtained are used to determine phase shift between the two sinusoidal voltages to be compared: there is a zero phase shift for the minimum instantaneous value.
threshold 2 possible switching order
fig. 12: phase comparator analysis between two voltages: Us and Um.
Real. Illustration Technique - Lyon Edition: DTE - Grenoble 03-96 - 2,500 - Printing: Clerc Printed in France
n 162
electrodynamic forces on busbars in LV systems
Jean-Pierre Thierry Engineering graduate from the CESI (Centre d'Etudes Suprieures Industrielles) and from the CNAM (Conservatoire National des Arts et Mtiers), he was initially employed in the Iron and Steel industry (roll mill automation and fluid monitoring). Following several years dedicated to preparing and developing mechanical vibration testing means, he joined Telemecanique in 1969. He then held in turn the positions of engineering and design department manager and technical Project manager for new products. He is currently in charge of developing prefabricated electric ducts. Christophe Kilindjian After graduating as an engineer from the Ecole Suprieure d'Energie et des Matriaux of Orlans in 1986, he then joined Merlin Gerin in this same year as part of the Technical Section in the Low Voltage Switchboards unit. Responsible for basic studies, he specialises in problems of heat exchanges and electrodynamic withstand in LV equipment.
E/CT 162 first issued, october 1996
electrodynamic forces on busbars in LV systems
contents
1. Introduction 2. Electrodynamic forces between two conductors: origin and calculations Preliminary remarks Origin and calculation methods Calculation for two parallel filiform conductors of infinite length Influence of conductor shape Conductors of reduced length Non-rectilinear conductors Calculation in the case of complex configurations 3. Electrodynamic forces in a three-phase busbar on a two or three-phase fault Reminder on short-circuit current making Maximum force on a three-phase busbar Resonance phenomena p. 4 p. 4 p. 4 p. 6
p. 6 p. 7 p. 7 p. 7 p. 9 p. 10 p. 11
4. Application to LV three-phase busbars
Case of busbars in p. 12 LV switchboards Case of prefabricated ducts p. 16 of the Canalis and Victa Dis type p. 20 p. 20
The importance attached to the concept of industrial dependability (safety of persons and equipment, availability of electrical power, reliability and maintenability of products) increasingly affects the design of the electrical devices used in industry (process...) and tertiary (hospitals). Their operating dependability thus contributes, often to a large context, to the dependability of the installation as a whole. This is the case of low voltage (LV) switchboards and of prefabricated transformerswitchboard connections. This quest for dependability requires studies in order to master, from the design stage, the behaviour of their components in the light of their environment and of possible operating stresses. One of these studies has already been dealt with in a Cahier Technique (thermal behaviour of LV electric switchboards). Withstand of electrodynamic forces is now the subject of a second study. Designers will find in this Cahier Technique the calculations laid down to allow for these forces and in particular to determine LV busbar requirements (prefabricated in ducts for electrical power distribution, and in switchboards). However calculation alone is not sufficient and results need to be validated by a real-life tests. We thus briefly describe the standardised tests.
1. introduction
The problem of withstanding electrodynamic forces arises on the LV power circuits of the installation. Although mainly dependent on the strength of the fault current, it also depends on the shape of the conductors, their mutual setout and securing method. Although this problem can be solved by calculation, only validation by a real-life tests enables provision of a document acknowledging conformity with standard and/or customer requirements. The very high current strengths that may occur during a short-circuit
between the various conductors of a LV installation (solid conductors of the bar, cable type...) generate considerable forces (several thousands of daNm). These forces thus need to be determined in order to mechanically size both the conductors and the structures supporting them so that they can withstand these forces whatever protective devices are placed upstream and downstream (standards stipulate electrodynamic withstand tests of one second). The exact calculation of electrodynamic forces is often complex in view of the
geometry of the conductors and associated structures. However a few approximations yield in most cases valid results on the basis of simple formulae. After a brief reminder of calculation of electrodynamic forces in simple geometries, this Cahier will deal with busbars in switchboards and prefabricated ducts on the basis of these formulae.
2. electrodynamic forces between two conductors: origin and calculations
The problem of conductor withstand to electrodynamic stresses is certainly not new as is shown by the number of publications which have treated this issue. However this problem is still of interest to designers as a result of the application of modern numerical methods which provide a solution for complex conductor configurations. This accounts for the summary presented in this chapter.
However the root mean square values
Irms are used in most cases; in this case Irms must be multipled by a
coefficient defined in chapter 3. c forces are expressed in absolute value without specifying their direction depending on field and current direction. In most cases they are forces per unit of length. c conductors are made of nonmagnetic material and are sufficiently distant from all magnetic elements likely to alter distribution of the magnetic field that they create. c skin effect and proximity phenomena which can considerably alter current distribution in the cross-section of solid conductors are ignored.
preliminary remarks
Application of the formulae call for compliance with the following points: c all the formulae involve the product of the current strengths, I1.I2, flowing in each conductor and inter-reacting. If their values are identical, this product is replaced by the term I2. c the current strengths appearing in the formulae correspond to the peak value of the currents conveyed in each conductor.
origin and calculation methods

The highlighting and understanding a century ago of mutual influences,
whether between two current elements or between a magnetic field and an electric current (work conducted by Oersted, Ampre...) resulted in the construction of a theoretical framework integrating these dynamic phenomena between conductors through which electric current flows. The direction of the electrodynamic forces is known (repulsion if the currents in the conductors flow in opposite directions, otherwise attraction) and their values are obtained by applying the laws of magnetism. There are in fact two main methods for calculating electrodynamic forces. The first method consists of calculating the magnetic field created by an electric current at a point in space, then deducing from it the resulting force exerted on a conductor placed at this point and through which
an electric current flows (possibly different from the first one). To calculate the field it uses (see box, fig. 1) either Biot and Savart's law:
or Ampre's theorem:
B
c
d l = 0 I ,
dl u , (1) d B = 0 i 4 r2
and to calculate electrodynamic force, it uses Laplace's law: (3) d f = i d l B .

The second method is based on calculating the potential energy variation of a circuit and uses Maxwell's theorem : (4) Fx = i
(see box, figure 1).
Biot and Savart's law Each element of a circuit through which a current i flows, of a length d l , produces at a point M a field d B such that:
dB =
0 d l u . i 4 r2
This field is: c perpendicular to the plane defined by the element d l containing point P and point M, c oriented to the left of an observer placed on the element, with the current flowing from his feet to his head and his gaze directed to point M (Ampre's theorem) c modulus d B where u is the directing vector of PM .
Ampre's theorem Deduced from Biot and Savart's formula, it is expressed as follows: Let I be the current strength flowing through a conductor crossing any surface of contour C. Circulation of the magnetic field along C is given by the equation: B d l = 0 I .
c
Laplace's law When a circuit through which a current of strength i flows, is placed in a magnetic field B , each element d l of the circuit is subjected to a force equal to:
d f =i d lB When B has an electric circuit as its origin, the law applied to each one expresses the force exerted between them:
d f = i1 d l B 2 = i2 d l B1 . Maxwell's theorem The work of the electromagnetic forces exerted during displacement of an undeformable conductor through which an invariable current flows, placed in a magnetic field, has the following expression: w = i or is the flow of the magnetic field swept during the displacement. Used in the form of elementary work, it easily obtains the components Fx, Fy and Fz of the resultant F of the electromagnetic forces: dw = i d
= d f d l

dB
dl
M r
= F d l hence
Fx = i
and likewise for Fy and Fz.
fig. 1: reminder of physical laws.
According to the geometry of the conductor system considered and to the calculation difficulty, one of the three approaches (1)+(3), (2)+(3) or (4), can be used. However the results obtained may differ slightly according to the approach used since the assumptions on which these laws are established are not the same.
calculation for two parallel stranded conductors of infinite length

For simple geometries such as filiform rectilinear conductors, application of Biot and Savart's and of Laplace's law results in the classical formula for electrodynamic force between two current lines; F/ l = 2 10-7 I1 I2/d where : F/ l in N/m, I1 and I2 in A, d in m, (The coefficient 2 x 10-7 results from the ratio 0/4 ). As this formula acts as a basis throughout this study, we must specify the assumptions for which this expression is valid. c the conductors are reduced to a current line. Their cross-section is thus reduced to a point. In practice this condition is considered acceptable for conductors of all cross-sections if the distance between the two conductors is considerably larger than the largest transverse dimension of the conductors (e.g. ten times). c the conductors are considered to be rectilinear and infinitely long. In practice this condition may be considered satisfactory if they are at least 15 to 20 times longer than the distance between them. Whenever one of these assumptions is not valid, a corrective factor must be applied.
always valid. In this case the influence of conductor shape may be determined by considering the conductor crosssection as a superimposition of interacting current lines. This approach was made by Dwight for a conductor with a rectangular cross-section. The resulting corrective factor, conventionally denoted k, can be determined by calculation. However as the expression of k is relatively complex, its value is determined in most cases on the S-shaped curves as in figure 2. The equation then has the form: F/ l = 2 10-7 I1 I2 (k/d) where :
F/ l in N/m, I1 and I2 in A, d in m. Examples of forces withstood by two parallel bars on a short-circuit are given in the table in figure 3. Although the same approach can be followed for all conductor shapes, calculations quickly become tiresome. In the above equation the term (k/d) is often replaced by 1/D, where D stands for the distance between the conductors corrected to allow for the influence of their shape. These coefficients are also useful in the case of a set of three-phase conductors
k 1.4 a 1.3 b 0.01...0.2 a 1.2 d 1.1 0.5 1 1 0.9 0.8

5 2
15 20 30 40 50 60 80 10 0
0.7 0.6
0.5
10
b a
0.4 0.3 0.2 2 3 4 5 6 8 10 20 40 60 80 100 200 300 d a
influence of conductor shape

This formula of F/ l only applies to current lines. However for solid conductors this assumption is not
Fig. 2: variation of k as a function of ratios b/a and d/a (Dwight's chart).
containing several conductors per phase. This case is dealt with in chapter 3.
This formula can be used only for values of a and b such that: 1 <
conductors of reduced length

c conductors of identical length When conductors have the same lengt l , 15 to 20 times smaller than their centre distance d, the resulting force is:
F = 2 10 7 I 2
a < 10 b
calculation in the case of complex configurations

The busbar configurations considered up to now in this study were mainly mono-dimensional, or sometimes twodimensional in the case of conductors forming an angle. In these cases, the methods used to calculate electrodynamic forces result in relatively simple formula. However conductors can be arranged in many different ways or be associated with a disturbing environment, such that the above formulae cease to apply. Such arrangements are referred to as complex configurations.
C 1 2 1 0.5 0.2 0
1+
d2
l2
d l
c1
c2
c conductors of unequal length (see fig. 4) In this case the resulting force is:
F = 2 10 7 I 2
fig. 4: drawing showing two conductors of unequal length.
l
D
[C1+ C2]
2
where d c C1= 1+ 1 + 2 l l and d2 c C2 = 1+ 2 + 2 l l

2 2
c12 2
d2 c C = 1+ + 2 l l
c2
d2
l2
c
l2
c22
l2
d2
0.9 0.8 0.7
l2
F
The values of C1 and C2 can be read on the chart in figure 5. If the conductors do not face each other over the entire length, with one passing the other, the formula applies with c1 or c2 negative. NB If c / l = 0, the equation is F in the above paragraph. The value of the expression between square brackets is given directly by reading the relevant curve on the chart in figure 5.
0.6
0.5 0.4 0.3
I a
0.2 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2
fig. 5: calculation and variation of C as a function of the ratios c / l and d / l .

characteristics a b d mm 100 100 forces
non-rectilinear conductors
This is, for example, the case of conductors with a bend (see fig. 6). The branches may inter-act with one another when a strong current passes through them. The conductor b may pivot around point O of the fixed conductor a. Force F has the following value:
F = 2 10-7 I 2 a a b 2 1- cos l + 1+ 2 b b a sin
l
m 1 1
I kA
F daN/m 224 1170
mm mm 5 5 80 80
F b I
0.91 35 0.91 80
a O
fig. 3: characteristics required to calculate the forces F between two conductors of the same length. Examples of forces withstood by two parallel bars during a short-circuit.
fig. 6: drawing showing two part of conductors (a and b) with a bend.
Three types of problems may then arise, either separately or combined: c the conductors facing one another are not all in the same plane: the problem is three-dimensional; c the conductors are close to metal frames which may alter the distribution of the magnetic field surrounding them; c the conductors are arranged so that it may be necessary to allow for skin effect and proximity phenomena which may considerably alter current distribution in the cross-section of solid conductors. Calculation of electrodynamic forces for the three types of problems mentioned above uses the general approach described in the paragraph on origin and calculation method, namely calculation first of the value and distribution of the magnetic field at each point in the system, then of the stresses in the conductors. The problem is thus divided into two to yield a magnetic and a mechanical problem. The basic physical laws used are therefore the same. However the difficulty, compared with the simple cases, lies in performing the calculations, as the three-dimensional aspect requires a numerical approach. Numerous methods have been developed in recent years to
numerically solve the problems described by differential equations. In particular the finite elements method, initially developed for mechanical problems, has been extended to a wide range of sectors and notably that of electromagnetism. In short, to define the calculation scope, this method consists of breaking down the system studied into a certain number of elements constituted and connected with one another by points known as nodes. The quantities which are of interest to us (magnetic field, stresses) are determined numerically at each node by salving the relevant equations (Maxwell and elasticity). Consequently, the value of each quantity studied is not known exactly at all points of the system but only at node level. Hence the importance of ensuring a good correspondence between these nodes and the real system, and of having a sound meshing. In practice, calculation using this method is made up of the following stages: c choice of the analysis type (e.g. magnetism...); c choice of type of elements to describe the system; c definition of the system geometry and of the calculation scope using key points;
c choice of meshing parameters and meshing of the calculation scope with the type of elements chosen; at this stage, the system studied is merely a set of nodes; c definition of boundary conditions to solve the equations; c carrying out the calculation; c using the results. A wide range of calculation software is available, differing by the categories of problems they can solve and the reliability of the results that they yield. For example Merlin Gerin has chosen the ANSYS software and Telemecanique Flux 2D since: c they enable very different problems to be dealt with (thermal, mechanical, electromagnetism...), c they are open-ended; thus their latest versions enable different problems to be paired (magnetic and mechanical or mechanical and thermal...). It is true that these methods may seem cumbersome and call for considerable investment. However with thorough mastery of the problems relating to modelling techniques, they allow rapid evaluation of the behaviour of a system or of one of its parts other than by tests. This is especially appreciable during the design and development phases when you consider the cost of a test campaign.
3. electrodynamic forces in a three-phase busbar on a two or three-phase fault
Consideration of three-phase busbar peculiarities when designing busbars for LV switchboards and prefabricated ducts, and of the peculiarities relating to the establishment and type of fault, is achieved by integrating factors into the formule presented in chapter 2. These peculiarities are: c relative layout of phases (conductors in ribbon, staggered...), c phase shift of currents in each phase with respect to one another, c type of short-circuit (two or threephase), c short-circuit making characteristics (symmetrical or asymmetrical state), c the peak current value, c the alternating aspect of currents, hence the vibrating aspect of the phenomena they generate. In the remainder of this section, the study will consider only busbars in ribbon, where phases 1,2, 3 are set out in the same plane and with the same distance between phases. The aim is to determine, by analysing the change in electrodynamic forces as a function of time and the various parameters above, the maximum value
of these forces and the conductor with the highest mechanical stress. As the electrodynamic forces of the current are proportional to the square of its maximum amplitude, the short-circuit currents need to be studied.
reminder on short-circuit current making

The aim of this paragraph is to review and specify: c the various short-circuit types that can arise in a three-phase system, c the notions of symmetrical and asymmetrical state, c the procedure to follow to determine the expression of short-circuit currents and the parameters on which they depend. The short-circuit types There are four types on a three-phase network. These types are shown in figure 7. Expression of short-circuit currents in the case of a three-phase fault We shall now concentrate only on symmetrical three-phase faults and
isolated two-phase faults which have the advantage, in steady state, of behaving like one or two independent single-phase networks. Let us consider a fault occurring on the single-phase diagram in figure 8 in which R and L are network impedance elements. If we set as the origin of time, the moment when the short-circuit occurs, the e.m.f. (e) of the generator has the value:
e = 2 E sin( t + )
where is the energising angle (see fig. 9) corresponding to the offset in time between a zero of the e.m.f. and the moment when the short-circuit was made. Ohm's law applied to the circuit yields:
di dt If the current is nil before the shortcircuit is made, the solution for this equation is: e = R i+L
i( t ) = 2 I sin( t + - ) + sin( - ) e -t /
a)
L3 L2 L1
c)
L3 L2 L1
x A
I" k
I" k I" k
I" k
Zcc B
Z1
b)
L3 L2 L1
d)
L3 L2 L1
fig. 8: equivalent single-phase diagram on a three-phase fault (see IEC 909).

u u = f(t)
I" k
I" k
a) symmetrical three-phase short-circuit. b) short-circuit between phases, isolated or two-phase.
c) short-circuit between phases, with earthing. d) phase-earth short-circuit
fig. 7: the various short-circuits and their currents. The direction of the arrows showing the currents are random (see IEC 909).
fig. 9: representation of known as the energising angle.
where :
= arctg
L R
(impedance angle)
=
I=
L R
E R +L2 2
2
by side. Thus each conductor undergoes at a time t a force which results from the algebraic addition of its interactions with the two other conductors. These conductors can have only two situations, external or central: c external position, for example phase 1: F1(t) = F21(t) + F31(t) F1(t) = cF [ i1(t) i2(t) + i1(t) i3(t)/2] cF is a function of distance between bars and bars shape. c central position, for example phase 2: F2(t) = F12(t) - F32(t) F2 = cF [i1(t) i2(t) - i2(t) i3(t)] However, as seen in the above paragraph, there are many cases to be considered for current expression according to the value of , , and of the type of short-circuit. In actual fact, only the value of the maximum forces is required to size the busbars: this value is the highest current occurring when = 0. NB: Fa>b = action (force) of the conductor(s) of phase a on the conductor(s) of phase b. Case of a three-phase short-circuit The effects on the conductors take the form: F1 = 0.87 [i1(t) i2(t) + i1(t) i3(t)/2] F2 = 0.87 [i1(t) i2(t) - i2(t) i3(t)] The maximum force on conductors over time is determined by the time values which cancel the derivatives of these expressions with respect to time: dF1/dt = 0 et dF2/dt = 0. Hence, after a few calculations, where Imax,3ph = 2 Irms,3ph the two equations: c F1max ,3ph = 2 10- 7 0.808 ( 2 Irms,3ph )2 1/d (case of one of the conductors external to the three-phase busbar) c F2max ,3ph = 2 10-7 0.866 ( 2 Irms,3ph )2 1/d (case of one of the conductors external to the three-phase busbar) Note: c compared with the reference formula reviewed in chapter 2
F/ l = 2 10-7 I1 I2/d the additional corrective factor which, according to the position of the conductor under consideration, equals 0.808 or 0.866. The maximum force is thus generated on the central conductor. c in practice, the coefficient k takes the circuit characteristics (R and L) into consideration: its value is between 1 and 2 (see fig. 10). Case of a two-phase short-circuit In this case i1 = - i2 and, using the above formulae, we can show that the maximum electrodynamic forces are reached when = 0 (asymmetrical state). F2max ,2ph = 2 10-7 1 ( 2 Irms,2ph )2 1/d Remarks The maximum force is not shown in two-phase, as it is often thought, but in three-phase. In actual fact: F2max,3ph F2max,2ph = 0.866 I 2 rms,3ph
All the factors representing the current variation as a function of time are then grouped in the following equation:
= sin( t + - ) + sin( - ) e-t /
The term can also be calculated using the approximate formula defined by IEC 909:
= 1.02 + 0.98 e
3R L
The difference with the exact value is less than 0.6%. Analysis of this function enables definition of the symmetrical and asymmetrical states of a fault (cf. Cahier Technique n 158). In the case of a three-phase system, the current in each phase takes the form:
i1( t ) = 2 Irms,3ph sin( t + - )
I 2 rms,2ph
+sin( - ) e-t/
which can also be written as:
however in the three-phase distribution state: 3 Irms,2ph = I 2 rms,3ph which yields the ratio:
F2max,3ph F2max,2ph 1.15
i1( t ) = 2 Irms,3ph
where Irms,3ph stands for the symmetrical root mean square current in the three phases in steady state. In view of their relative phase shift: c i2 same as i1 by replacing by + 2/3 c i3 same as i1 by replacing by - 2/3. Finally, the electrodynamic forces thus depend on : c the initial instant of the short-circuit (via the value of ) ; c the characteristics of the circuit (via the value of ) ; c the phase shift between the phases (2/3).
k 2.0 1.8 1.6 1.4 1.2 1.0 0 0.2 0.4 0.6 0.8 1.0 R/X 1.2
maximum force on a threephase busbar

A three-phase busbar normally contains three conductors placed side
fig. 10: variation of factor k as a function of the ratio R/X.
Test organisations often demand two and three-phase tests with currents of identical value. These test conditions do not correspond to real distribution characteristics and result in two-phase forces which are greater than threephase forces.
F( t ) = M
y 2 y y 4 + +E J 4 2 t x t
resonance phenomena
Forces appearing on a short-circuit do not form a static phenomenon, but are vibrating quantities of a frequency twice that of the network or of its multiples. Conductors which have a certain elasticity can then start to vibrate. If the vibration frequency corresponds to a natural frequency for all conductors, resonance phenomena may occur. In this case the resulting stresses in the conductors may be far greater than those created by the forces due to the peak current value. It is thus necessary to determine the ratio between the real and static forces undergone by the conductor. This ratio conventionally denoted V is known as the stress factor. In addition to the mechanical characteristics of the conductors, we must allow for the way in which they are secured in the device housing them (LV switchboard, duct...). We thus need to reason on the busbar structure. There are two standard methods for securing busbars: flush mounting and simple support. However in reality the insulating elements support the conductors, which results in a combination of these two methods (see fig. 11). The large number of parameters to be considered makes a complete study of these phenomena complex. The starting point for such a study is the general equation applied to a conductor assumed to have an elastic behaviour:
where: M = mass of the conductor per unit of length, J = moment of inertia of the crosssection perpendicular to the conductor axis, E = modulus of elasticity, = damping coefficient, y = distance from a point of the conductor with respect to its position of equilibrium or deflection, x = distance from a point of the conductor with respect to a fixed bearing, t = time. where F(t) = Fo sin (2 t) where: Fo = amplitude of the force, = network pulsation ( = 2 f). The solutions take the form: y = cste Fk(t) Gk(x) where the functions Fk(t) and Gk(x) depend on time and on the space variable respectively, as well as on: c the securing methods, c the electrodynamic force relating to the short-circuit state (symmetrical or asymmetrical). The complete study was conducted by Baltensperger and leads to an
expression of conductor natural resonance frequencies: 2 Sk E J ok = 2 M p where Sk = coefficient function of the securing methods, for example for a bar flush mounted at its ends: Sk = (4 k - 1) /2 ; k = rank of resonance frequency; p = distance between the supports. In practice, we observe that natural conductor frequencies, for a specific cross-section, depend on the longitudinal distance between supports. The calculation therefore aims at examining whether the stress factor, resulting from the selected distance between the supports, is acceptable for the natural frequency of the conductor or all the conductors, resulting in a coefficient R, homogenenous with a length:
EJ 103 M 2 The graph in figure 12 shows the stress factor V to be anticipated as a function of the ratio p/R, i.e. of the distance p between the supports. p must be chosen so that the ratio is outside the hatched zone for the accepted factor V. R=4
V 5
- V: stress factor - p: distance between supports EJ 103 - R=4 M 2
3 resonance zone 2
1 p 1 2.24 2.45 3 3.55 4 5.22 6.12 7 R
fig. 11: the various busbar securing methods: by flush mounting (a), by simple support (b) and a combination of both (c).
fig. 12: stress factor V to be anticipated as a function of the ratio p/R.
4. application to LV three-phase busbars
In this chapter the authors define how the above theoretical considerations are taken into account for two LV items of equipment, namely LV switchboards and prefabricated electrical ducts of the Canalis and Victa Dis type.
In practice, sizing requires determination of the distance between the supports and thus the number required, for a specific busbar and support technology. Practical calculation procedure The method to be followed is summarised in the chart below: I. Definition of basic data II. Calculation of forces III. Calculation of the distance between supports based on stresses on the conductor with the greatest stress IV. Calculation of the distance between supports based on stresses on supports. V. Determination of the maximum distance between supports, and verification of the vibration behaviour of the busbar.
The details of each stage are described below for a busbar consisting of several rectangular cross-section bars per phase. I - Basic data for carrying out the calculation c dimension and shape of a conductor (for example for a bar, its thickness a and its width b in m.) c number of conductors per phase: n. c root mean square value of the shortcircuit current: Isc in kA. c type of fault: two or three-phase. c distance between phase centres: dph in m. c conductor securing method in the supports (flush mounting or simple support). This data is taken into account by a coefficient : = 1 for all the conductors of a phase, = 2 for a conductor belonging to one phase, c elastic limit of the conductor: Rp0.2 in N/m2 (Rp0.2 = 125 x 106 N/m2 for 1050 type aluminium and Rp0.2 = 250 x 106 N/m2 for copper). c characteristics of supports: mechanical withstand Rm (in N/m2) according to the type of stress, and cross-section of the stressed support Sm (in m2). II - Calculation of forces Each conductor of a phase is subjected to a force due to the actions between phases and to the actions of the other conductors of the same phase. The maximum force is exerted on the most external conductors of the central phase. This conductor is subjected: c firstly to the force resulting from the other two phases: F1/ l = 0.87 (or 1) 2 10-7 k1 (2.2 Isc)2 1/dph) 0.87 : if the fault is three phase 1 : if the fault is two phase
case of busbars in LV switchboards

The three-phase busbar of a LV electric switchboard is made up of a set of conductors grouped by phase and held in place by supports. It is characterised by: c the shape of the conductors, c the relative layout of the phases, c the arrangement of the conductors in the same phase, c the type of support and the conductor securing method (insulating bars, combs, insulating rods...). The various elements making up the busbar system must be sized to withstand the electrodynamic forces which appear when a short-circuit occurs (see fig. 13).
fig. 13: busbar of a Masterbloc LV switchboard, designed to withstand the effects of a 80 kA short-circuit current (Merlin Gerin).
(force per unit of length of the busbar in N/m). k1 = Dwight's coefficient allowing for the shape of all the conductors of the phase. This coefficient, parameterised by the ratios height (h)/width of a phase (I') and dph/width of a phase, can be calculated or read on charts. Isc = root mean square value of the short-circuit current in kA. dph = distance between phase centres in m. The multiplying factor 2.2 is used to calculate the peak value of the shortcircuit current. c secondly to the force of attraction (current in the same direction) resulting from the other conductors of the phase considered (see fig. 14), if these are mechanically linked:
F2 / l = F21i / l (in N / m)
i
phase 1
phase 2 d' b d ph h
phase 3
d1 2 d1 3
l'
here b = h k1 = f(h, I', dph) k2 = f(a, b, d')
fig. 14: parameters considered to establish the equation for the force of attraction between busbar conductors
configuration; bars of the same phase are flush mounted and the three phases positioned (see fig. 15). Busbar deformation criterion The bar with the greatest stress must not be deformed. However a slight residual deformation is accepted according to a coefficient q defined by the IEC 865 standard. The above formula includes d1. This distance between supports can be determined from a maximum stress level at the conductors which must not be exceeded, such that = q.Rp0.2 (for example q = 1.5). IV - Calculation of the distance between supports based on stresses on supports The supports must therefore withstand the stresses linked to the force F1. Support break criterion:
d2 = Rm Sm F1/ l where = constant whose value depends on the securing method and the number of supports.
Equation of the same form as the one above, but taking into account the following three parameters: d1 i = centre distance from conductor 1 to conductor i in m, n = number of conductors per phase, k2 = Dwight's coefficient for the phase conductor. III - Calculation of the distance between supports based on stresses on the conductor with the greatest stress The conductor with the greatest stress must withstand the stress:
a distance d equal to the smallest value of d1 and d2: d i min (d1,d2). Moreover you must ensure that this distance does not generate resonance phenomena. This calculation procedure complies with the recommendations of the IEC 865 standard (1986) dealing with calculation of the effects of short-circuit currents as regards both the thermal and mechanical aspects. Although these calculations do not replace real-life tests, they are vital for designing new products and for satisfying specific cases.
= 1+ 2
=
1 (F1/ l) d12
8Z
2 (F2 / l) d12
8 Z0
F1/ l and F2/ l = forces in N/m, d1 = distance between two supports in m, Z0 = resistance module of a bar in m3, Z = resistance module of a phase in m3, 1 = 0.73 (simple support coefficient), 2 = 0.5 (flush mounting coefficient). These values are given by way of guidance for a specific busbar
V - Determination of the maximum distance between supports, and verification of the busbar vibration behaviour In order to withstand electrodynamic forces, the supports must be placed at
,,,,, ,, ,,,,, ,,,,, ,,

flush mounted conductors phases in simple support
fig. 15: configuration of a busbar for coefficients 1 = 0.73 (simple support) and 2 = 0.5 (flush mounting).
Calculation example I. definition of basic data c conductors flat copper bars thickness a = 5 mm width b = 100 mm securing: flush mounted bars c each phase is made up of n = 3 bars, with a 5 mm spacing (d' = 10 mm) c distance between phase centres dph = 95 mm c three-phase fault Isc = 80 kA rms c elastic limit of copper Rp0.2 = 250 x 106 N/m2 c mechanical withstand of support Rm = 100 x 106 N/m2 c cross-section of support subjected to tensile stress Sm = 150 x 10-6 m2 II. calculation of forces c between phases F1/ l = 0.87 2 10-7 k1 (2.2 Isc) 2 1/dph k1 : Dwight's coefficient, function of the ratios b/(2n - 1) a and dph/(2n - 1) a k1 (100/5 5, 95/5 5) = 0.873 c between bars of the same phase particularly on the external bars of the central phase F2 / l = F21,i / l
i
F1/ l = 0.87 x 2 x 10-7x 0.873 x (2.2 x 80 x 103) 2 x 1/95 x 10-3 F1/ l = 49 530 N/m = 4 953 daN/m
1 index of the first bar i = 2 and 3 index of the two other bars of the phase F2 1, i / l = 2 10-7 k21,i ( 2.2 Icc/n ) 2 1/d d1i : distance between the axis bars 1 and i k2 1, i: Dwight's coefficient as a function of ratios b/a and d1i /a k2 1, 2( 100/5, 10/5 ) = 0.248 k2 1, 3( 100/5, 20/5 ) = 0.419 III. Calculation of the distance between supports based on stresses on the conductor with the greatest stress (elastic limit of conductor) = 1 (F1/ l ) d12/8 Z + 2 (F2/ l ) d12/8 Z0 = 1.5 Rp0.2 d12 = 1.5 Rp0.2 / [1 (F1/ l )/8 Z + 2 (F2/ l )/8 Z0 ] 1 = 2 = 0.5 Z0 = b a2/6 = 4.2 10-7 m3 Z = n Z0 = 3 Z0 = 1.25 10-6 m3 IV. Calculation of the distance between supports based on stresses on the supports (elastic limit of supports) d2 = Rm Sm/(F1/ l ) = 0,5 V. determination of the maximum distance between supports d = minimum entre d1 et d2
F2/ l = 2 x 10-7x (2.2 x 80 x 103/n) 2 x [0.248/10 x 10-3 + 0.419/20 x 10-3] F2/ l = 31 490 N/m = 3 149 daN/m
d12 = 1.5 x 250 x 106 / [ 0.5 x (49 530)/8 x 1.25 x 10-6 + 0.5 x (31 490)/8 x 4.2 x 10-7] d1 = 0.229 m = 229 mm
d2 = 100 x 106 x 150 x 10-6/(0.5 x 49 530) d2 = 0.604 m = 604 mm d < 229 mm
Standards and tests There are two test categories for LV equipment, namely: c development tests assisting with design, c certification tests. The latter are part of a set of tests known as type tests whose reports are frequently demanded for a product defined as a Type tested assembly (T.T.A.). This designation, which requires tests, thus forms an additional guarantee for users. However, despite this constraint, manufacturers develop products which allow them to valorise their know-how. The type tests defined by the standards IEC 439-1 (1992) and 2 (1987) or NF 63-421 (1991) total 7 (439-1) and 10 (439-2) respectively. As regards short-circuit withstand, which is the subject of this document, these standards specify both the test conditions to be complied with and the standardised value of the coefficient connecting the peak value to the root mean square value of the short-circuit current (see fig. 16). If the system considered varies only slightly from the reference system (T.T.A.), it is known as a Partially type tested assembly (P.T.T.A.) and it can
be qualified by calculation from an T.T.A. structure. With respect to short-circuit current withstand, an extrapolation method for the P.T.T.A. has been defined by the technical report IEC 1117 (1992). Complete certification in short-circuit current withstand requires three tests: c a three-phase short-circuit current withstand test; c a withstand test for a short-circuit current between the neutral and the nearest phase. Note that if the neutral has the same cross-section as the other phases and if the distance between the neutral and the nearest phase is the same as the distance between phases, this test corresponds to a two-phase short-circuit; c a withstand test for a short-circuit between a phase and the protective conductor. For each test the manufacturer must specify the root mean square value of the short-circuit current and its duration, normally 1 s (to verify the thermal constraint linked to the short-circuit current). As regards the value of the short-circuit current for the three-phase test, two values must be identified: the prospective value and the real value. Their difference is due to whether or
root mean square value of the short-ciruit current (kA) Ii 5 5 < I i 10 10 < I i 20 20 < I i 50 50 < I
cos 0.7 0.5 0.3 0.25 0.2
n 1.5 1.7 2 2.1 2.2
fig. 16 : standardised value of the coefficient n connecting the peak value to the root mean square value of the short-circuit current; n correspond to coefficient 2 defined in chapter 3 (ie. IEC 439-1).
not the busbar impedance is taken into account during calculation. In practice: c calculation performed with a voltage equal to the operational voltage, at the entrance to the switchboard presumed value of Isc; c calculation performed, at extra-low voltage, at the end of the busbar at the short-circuit point real value. It is obvious that for the same announced value of the short-circuit current strength, the second case is far more restrictive. The difference may range from 20 to 30% according to the circuit. For the phase-neutral test, the value of the short-circuit current corresponds to 60% of the value of the current (prospective or real) of the three-phase test. Many manufacturers (including Merlin Gerin and Telemecanique) currently tend to perform these tests in real current. Moreover, to ensure that these tests are representative of the most unfavourable tests possible during a short-circuit, the following points must be complied with: c presence of an asymmetrical state at least on one of the three phases; c presence of at least one joint or fishplate on the tested busbar; c creation of a bolted short-circuit; c consideration of vibrating phenomena, while maintaining the fault for at least ten cycles, i.e. 200 ms at 50 Hz; this time is often extended to 1 s to check thermal withstand at the same time (IEC 439-1). The various test stages are: c calibration circuit by short-circuiting the transformer outputs; c connecting the busbar to the platform transformer; c setting up the short-circuit (specific part connecting all the bars) on the busbar; c short test (roughly 10 ms) to determine busbar impedance; s 1 s withstand test on the assembly.
case of prefabricated ducts of the Canalis or Victa Dis type

Construction of prefabricated threephase busbars of the Canalis or Victa Dis type, designed for current transmission and distribution (see fig. 17) complies with proper procedures and with specific standards, the main ones of which are the IEC 439-1 and 2 (international) and UL 857 (United States). Design The techniques implemented vary according to the current ranges considered, especially for large currents exceeding 100 A. There are currently three main duct designs: v standard, v sandwich, v flattened. c standard design (100 to 800 A) The conductors are placed in a metal envelope and maintained at regular intervals by comb shaped insulators (see fig. 18).
The electrodynamic forces developing when short-circuits occur, observe the laws stated above and result in the deflection of conductors between insulators and in an overall vibration. The shape and cross-section of the conductors result from the best possible balance achieved between: v temperature rise of conductors; v acceptable voltage drop; v production cost. With the following vital requirements for mechanical withstand: that conductor deflection continues to be elastic (no permanent deformation after a shortcircuit) and does not abnormally reduce the insulation level (between phases or between phases and earth) during the short-circuit transient period in which the electrodynamic phenomena are created. In practice this is obtained by adjusting the distance between insulators. c sandwiched design (1000 to 5000 A): Beyond a certain current, 1000 A, in order to remain within acceptable heat exchange conditions and dimensions for the duct, the current of the same phase is distributed over several
conductors. Some ducts sandwich up to five conductors per phase. Designers can then either: v leave the conductors of the same phase grouped together, v or insert the elementary phase conductors in an orderly manner (1-2-3) + (1-2-3) + (1-2-3) to obtain the sandwiched configuration (see fig. 18). This type of design is ideal for horizontal current distribution. c flattened design (1000 to 5000 A): In this layout, the rectangular crosssection conductors coated with an insulating sheath are kept in contact all along the duct, just as in a cable (see fig. 18). Conductors are clamped to ensure the necessary heat exchanges. To simplify manufacture, conductors normally have constant thickness, and only their width varies according to nominal busbar current strength (up to roughly 250 mm). For high currents, two or even three conductors per phase, but not sandwiched, are required.
a)
N L3 L2 L1
b)
N L3 L2 L1
c)
N L3 L2 L1
fig. 17: Canalis (Telemecanique) 3000 A electrical distribution prefabricated busbar.
fig. 18: the various prefabricated threephase busbar designs: standard (a), sandwiched (b) and flattened (c).
The electrodynamic forces (distributed loads) when a short-circuit occurs are balanced in these busbars by the reaction of the envelope sheet metal. Its thermal behaviour means that this type of design is ideally suited to transmission of horizontal or vertical current. Distribution of electrodynamic forces This paragraph uses a simple, concrete example to visualise and quantify the
various forces to which conductors are subjected. The structure studied has the following characteristics: c In = 3000 A, c three conductors/phase, i.e. 1000 A/conductor, c conductor cross-section = 90 mm x 6 mm, c material = aluminium or copper, c distance between conductor centres = 18 mm.
The calculations shown in the box (see fig. 19) evaluate the mechanical stress of the elementary conductors (according to current direction) of phases 1 and 2 for a phase 1/phase 2 two-phase short-circuit with correction of the geometric incidence in accordance with Dwight's chart. Partial conclusions: With the standard layout, an increase and large dispersion of the forces applied to the various conductor
1'
1"
2'
2"
3'
3"
standard design
+ + +
d
F = f i2 ,cos with i = I / 3
Fph1= F d
=
1 1 1 1 0.42 + 0.62 0.83 0.87 0.92 2 d 4 d 5 d 6 d
F 0.19 K1= 0.19 d
Fph1' = F d
=
1 1 1 1 0.42 + 0.42 0.75 0.83 0.87 d 3 d 4 d 5 d
F 0.63 K1'= 0.63 d
Fph1" = F d
=
1 2 3 1' 2' 3" 1" 2" 3"
1 1 1 1 0.42 + 0.62 0.62 0.75 0.83 2 d 2 d 3 d 4 d
F 1.49 K1"= 1 .49 d
sandwich design
Fph1= F d
=
1 1 1 1 0.42 + 0.83 0.87 1 1 4 d 5 d 8 d 9 d
F 0.37 K1= 0.37 d

1 1 1 1 1 0.42 0.75 + 0.83 0.97 + 1 3 d 4 d 7 d 8 d
Fph2 = F + d
=+
F 0.36 K2 = 0.36 d
fig. 19: mechanical stress of phase 1 and 2 elementary conductors.
elements are observed, whereas for the sandwich layout, forces remain more or less the same for each conductor element. In this example, the difference in mechanical stress has a ratio of 1 to 5 in favour of the sandwich layout. Moreover, this layout offers another advantage as for as voltage drop is concerned: sandwiching of phases causes a reduction in the magnetic induction resultant and thus in reactance,... i..e. in voltage drop. Branch-offs and splicing Two technologies are normally chosen to sample current at the branch-offs or to conduct it in the splice bars of a prefabricated transmission and distribution line. These are bolted technology and contact technology. c bolted technology The connections are made from special bolted pads provided in the equipment design stage. The above laws are also applied for sizing the pads and insulators.
Application of this technology has the following practical limits: v 1250 A, in branch-off, v up to 6000 A in splicing. NB : Some articulated bends produced in the same plane use the bolted technology principle. c contact technology Current conducted using parallelconnected contact fingers. As a first approach, the current is distributed in proportion to the number of parallel contacts. Each contact point has a static force F (developed by an external spring) whose sizing results from a compromise between the level of the required contact resistance to ensure nominal current flow without abnormal temperature rise, and the friction force withstand during conductor expansion. With this in mind, we should note the advantage of lubricating the elastic contacts or of using, for mounted contacts, silver/graphite type combinations.
This technology is particularly used in the following current ranges: v 16 to 400 A in branch-off, v 40 to 1000 A in splicing. c electrodynamic force withstand Whereas bolted technology imposes on the elements of the structure in question the same electrodynamic forces as for the busbars in LV switchboards, contact technology benefits from these forces. The layout normally chosen for the contact fingers or pawls is illustrated in the drawing in figure 20 on which it is clearly shown that the currents flowing in opposing pawls run in the same direction. The electrodynamic forces (distributed loads) developed along the pawls and calculated using the above methods thus tend towards an attraction. They consequently reinforce the contact force and oppose the repulsion force of contacts which has as its origin striction of the current lines in the vicinity of the contact point (see fig. 21). This is the selfcompensation principle (see fig. 20). Electrodynamic force tests Type tests, specific to ducts, are defined by the IEC 439-2 and NF C 63-411 standards. The main difference compared with LV switchboards lies in the shortcircuit test conditions which specify that the tests must be performed on an installed line no more than six metres long with at least one splice joint and a bend (see sketch in fig. 22).
cross-section BB
l
A
I d A cross-section AA B
k is the shape factor, to be read off the chart in figure 2, for a conductor with a global cross-section a x b. If we write Fa Fr so that compensation is achieved, the result is:
2 3 l u + 1 1 n k d For example, for k = 0.8, the ratio I/d must reach: 4.6 for 1 contact jaw (n = 1) 2.7 for 2 contact jaws (n = 2) 1.4 for 5 contact jaws (n = 5) 0.95 for 10 contact jaws (n = 10). Although it may seem interesting to increase the number of parallel contact jaws n, we are quickly limited by technological considerations as well as by differences in resistance and reactance between adjacent contact jaws which do not allow even current distribution between each other such as is assumed by the calculated value of I/d. We must therefore take a safety margin on the calculated value of I/d, as large as the number of parallel contact jaws is high. In practice, there are applications of up to 2 x 12 parallel contacts which can withstand acceptable short-term currents of the order of 50 kA RMS - 1 s.
B In the calculation, we consider the forces on each half-contact jaws, of cross-section a x b, grouping all the contact elements of jaws. If n is the number of parallel contact jaws, the total repulsion force on the half-contact jaws is: Fr = 2 n 3 10-7 (I2n)2 The force of attraction which has to compensate it is: 2 2 I k l 1+ d d Fa = 2 10 7 2 2 d l l
I R 2r
A repulsion force F is exerted between the two conductors:
(In = natural log and r = contact point radius calculated with R. Holm's formula).
fig. 20: the self-compensation principle.
fig. 21: example of striction of current lines in the vicinity of the contact point between two cylindrical current-carrying elements
test source
source connecting device
sheathed busbar length
maximum imposed length =6m
splice joint (at least one)
sheathed bend (at least one)
short-circuit device
fig. 22: sketch showing a prefabricated busbar line such as defined by the standards for the type tests.
5. conclusion
The high electrodynamic forces occurring on a short-circuit and the material damage that they can cause justify the importance attached to mechanical withstand of busbars. An importance all the more vital as busbar withstand failure requires at the very least replacement of these busbars and thus shutdown of the installation.
It is thus advantageous for installers and/or users to choose equipment presenting a maximum guarantee (T.T.A.) or made up of modified standard elements, mounted in the factory and tested (P.T.T.A.). In both cases, the importance of testing is obvious. However such tests call for considerable investment that only major
manufacturers can support in view of the necessary infrastructure and costs involved. Design modifications from the type tested cases are, however, possible.It is in this respect, to a certain extent, that the calculation approach and the manufacturer's knowhow can take over from the experimental approach.
6. bibliography
Standards c IEC 439-1: Low-voltage switchgear and controlgear assemblies. Typetested and partially type-tested assemblies. c IEC 439-2: Low-voltage switchgear and controlgear assemblies. Particular requirements for busbar trunking systems (busways). (NF C 63-411). c IEC 865: Short-cicuit currentscalculation of effects. c IEC 909: Short-circuit current calculation in three-phase a.c. systems. c IEC 909-1: Short-circuit current calculation in three-phase a.c. systems. Factors for the calculation of shortcircuit currents in three-phase a.c. systems according to IEC 909. c IEC 1117: Method for assessing the short-circuit withstand strength of partially type-tested assemblies (PTTA). Merlin Gerin Cahiers Techniques c Thermal study of LV electric switchboards Cahier Technique n 145 C. KILINDJIAN
c Operating dependability and LV electric switchboards Cahier Technique n 156 - O. BOUJU c Calculation of short-circuit currents Cahier Technique n 158 R. CALVAS, B. DE METZ-NOBLAT, A. DUCLUZAUX, and G. THOMASSET Various publications c Elektodynamische Beanspruchung von parallelen Leitern, or Electrodynamic effects on parallel lines P. BALTENSPERGER Bulletin Schweiz Elektotechn Verein n 25, 1944 c Sectioned busbars in first and second category installations R. ROLS. Four-part paper published in the Revue de l'Aluminium n 212 - 213 - 214 - 215, 1954 c Transmission of high currents in LV and MV ac current - 2nd part P. BEIGBEDER Bulletin Etudes et Ralisations n 43, 1957
c Calculation of three-phase busbar installations in view of withstand of electrodynamic forces R. MASCARIN Revue gnrale de l'Electricit RGE, August 1957. c On the establishment of formulae designed to determine force per unit length in case of a short-circuit affecting a three-phase busbar in a ribbon arrangement. R. MASCARIN. Revue gnrale de l'Electricit RGE, March 1959. c Uber den Einflu von Resonanzerscheingungen auf die mechanische Kurzschlu festigkeit von biegesteifen Stromleitern. P. SIEBER AEG Mitteilungen n 49, 1959 c Mechanical forces on current-carrying conductors. E.D. CHARLES Proceedings IEE, vol. 110, n 9, September 1963 c Electrodynamic forces appearing in electric substations on a short-circuit G. SCHAFFER Revue Brown Boveri, 1970.
Ral.: Sodipe - Valence - Photo.: IPV - Grenoble Edition: DTE - Grenoble 10-96 - 2500 - Printing.: Clerc Printed in France
n163
LV breaking by current limitation
Pierre Schueller
Arts et Mtiers engineer from the Grenoble Electrotechnical Institute, he joined Merlin Gerin in 1967. His first function consisted in design of low voltage limiting circuit-breakers for terminal and then industrial distribution. A member of the Technical Section of the Low Voltage Power Division since 1983, he now directs the engineering and design department with responsibility for sensor and actuator development.
E/CT 163, first issued november 1994
LV breaking by current limitation
content
1. General Definition Why limit? How to limit? Conditions to be respected by Ua for current limitation Special case of miniature circuit-breakers Status change resistor Positive temperature coefficient resistor Variable resistor formed by the actual breaking arc Contact propellers Ultra-fast trip units p. 4 p. 4 p. 4 p. 5 p. 5 p. 6 p. 6 p. 6 p. 6 p. 8 p. 8 p. 8
This Cahier Technique provides a simple introduction to the principles of low voltage current limitation, a technique developed by Merlin Gerin in direct current as early as 1930 and in alternating current in 1954. It simplifies understanding of the advantages gained by using limiting circuit-breakers in electrical installations. The document ends with a detailed bibliography for those wishing to satisfy their scientific curiosity.
2. How to obtain Ua voltage
3. Contact propellers and ultra-fast trip units 4. Conclusion 5. Bibliography
1. general
definition
A device is said to be limiting when the current passing through it during a short-circuit has an amplitude considerably lower than the prospective current (see fig. 1). In the case of limiting circuit-breakers, this reduction in amplitude is accompanied by a reduction in the current flow time T compared with the short-circuit current flow time of a nonlimiting circuit-breaker.
If the same current I flows through both conductors, the formula becomes:
F I2 = 2 107 (MKSA) L d Example: where I = 50 kA and d = 10 cm
F = 5000 N / m L Possible consequence: deformation or rupture of parts. s in all switchgear, separable contacts, held together by springs, tend to open under the effect of an electrodynamic force known as repulsion. These forces must sometimes be balanced by "compensation" systems. For I = 50 kA, this force is 1000 N. Possible consequence: arcing between control device contacts with damage to contacts.
how to limit?
Take a single-phase AC circuit with an apparent power S and voltage E, delivering in a load Z through a protective device A presenting a negligible impedance before it is activated (see fig. 2) with for the group: source + line + fault: R = equivalent resistance L = equivalent inductance. When a short-circuit occurs at the terminals of load Z, before A is activated (thus with negligible ua ), the mains is supplied with an electromotive voltage e such that: e=Ri+L di dt the current is thus established with an initial derivative equal to:
why limit?
to make more cost effective circuitbreakers, especially in low rated current ranges. Only the limited current, far smaller than the prospective current, flows through the limiting device, which then has only to break this limited current. s to minimise fault current effects in electrical installations. What are these effects?
s
Thermal effect During a short-circuit, there is an adiabatic temperature rise of the S cross-section conductors, of up to: =
s
e di = dt o L
This derivative is greatest for shortcircuits occurring when mains voltage is highest. For power factors less than 0.25, this corresponds to a virtually symmetrical prospective current.
Electromagnetic effect At a certain distance from a conductor through which a current of strength l flows, a magnetic induction B is in the air with a value: I (MKSA) d Example: where I = 50 kA and d = 10 cm, B = 0.1 tesla. Possible consequence: disturbance of electronic devices placed close to electric conductors through which a short-circuit flows. B = 2 10 Mechanical effects s if at the distance d of a conductor through which a current I flows, there is another conductor parallel to the first with the same length L and through which a current I' flows, this conductor is subjected to a force F (attraction if the direction of I and I' is identical, repulsion if it is not) which equals per unit of length:
F = B I' L
7
k S
2
T i2 dt
T i2 dt is known as the thermal

i limited current prospective current
stress (given in A2 s); s K is a coefficient dependent on the type of conductors (approximately for copper) A2 s Example: A copper wire with a crosssection of 1.5 mm2, is heated to roughly 110K when a current period of 2000 A r.m.s. at 50 Hz flows through it. Obvious possible consequences: deformation of device and destruction of insulating material with risks of fire and electrocution.
6 10 3
K mm4
fig. 1: representation of prospective and limited short-circuit currents
source S E
R i
A e Z
ua fault
fig. 2: schematic diagram of a faulty circuit
Example: a 400 V 50 Hz three-phase source, phase-to-phase, with an apparent power S = 3200 kVA and a maximum short-circuit current of 100 kA r.m.s. (with a peak which can exceed 200 kA in asymmetrical condition). The maximum initial current derivative is 44 kA/ms. To prevent such currents developing and to guard against their effects, a limiting protective device A must be placed in the circuit. When a short-circuit occurs, this device quickly provokes a voltage drop or back electromotive voltage ua which opposes current increase.
conditions to be respected by ua for current limitation

The equivalent single-phase diagram yields the following relation for a full short-circuit: di E=Ri+L + ua dt A vectorial representation of the equivalent impedance for power factors cos 0.25 (thus > 75) shows that the term L di/dt is far larger than the term R i (see fig. 3). Thus if the latter is not taken into account: di e = L + ua dt then the limited current reaches its peak value, i.e. when di = 0, the electromotive voltage has dt the value ua. We can thus conclude: the highest limited current is reached when voltage ua equals source e (see fig. 4). One of the first consequences is current limitation, which is easiest to obtain when mains voltage e is low. Then, in figure 4 where P is the point of intersection of the development curves of voltage ua and voltage e of the source, the curves show that to obtain correct limitation, the instant of intersection P must occur well before the highest prospective current (thus < 5 ms in 50 Hz). It is thus advantageous for voltage ua to develop as quickly as possible. Finally, the maximum voltage UM introduced by ua must be greater than the maximum voltage EM of the source.
In short, the three conditions to be respected by ua for correct limitation are: s early action ts minimum. However there is a lower limit laid down by the device activation threshold (e.g. maximum setting of a circuitbreaker's instantaneous trip units or non-melting thermal stress for fuses). s prompt action rapid development of voltage ua, achieved for example in a circuit-breaker by high contact acceleration. s high action UM > EM obtained for example by elongation, splitting and cooling the arc in the breaking device. Out of these three conditions, the first two, rapidity and speed, are the most important. As regards the third condition, UM need not overshoot EM by a large amount. Consequently, for a three-phase 420 V r.m.s. network (thus with a peak voltage of 240 2 = 340 V ), a UM voltage of 400 V is sufficient.
highest value, the following can be written: ua = EM - R II which shows that maximum voltage ua can remain less than maximum mains voltage. For example: on a circuit with a phase to neutral voltage V = 240 V r.m.s. (i.e. EM = 340 V), if the prospective short-circuit current Ip is 6 kA with cos = 0.6, knowing that R = V / Ip cos and assuming a limited peak current II of 4 kA, the calculation yields: ua = 243 V, less by nearly 100 V to EM . As regards the three conditions to be met for correct limitation, the next two chapters look into the various physical principles and techniques implemented in the design of limiting devices, fuses and circuit-breakers.
special case of miniature circuit-breakers

In this case, the short-circuit power factors are normally greater than 0.5 The term Ri can no longer be ignored. Thus, when a limited current reaches its
L di dt Ri
fig. 3: vectorial representation of the two components R i and L di/dt
i prospective i limited i 5 ms 0 UM EM P e ua u T 10 ms t
iI
ts
recovery u
Note : ts is the moment of appearance of voltage ua (e.g. contact separation or vaporisation of a fuse link). fig. 4: curves u = f(t) and f(t) development of arcing voltage and its consequence: decrease in short-circuit current.
2. how to obtain voltage ua
A prompt voltage drop ua is generally obtained by inserting a number of devices in series in the circuit. However, it should be pointed out that limiting devices do not always have a breaking function and must then be backed-up by a circuit-breaker when used for protection of electric circuits.
positive temperature coefficient resistor

(but with a limited temperature rise to remain below melting point). Permanently installed resistor In practice its use is restricted to rated currents under 100 A for continuous heating purposes. Parallel-connected resistor, with contacts opening quickly on a fault [1] [2]. Without the continuous heating stress, this system enables higher rated currents to be reached. However, current switching stresses, with a contact in parallel to the PTC resistor, are still present. Moreover, other contacts must always be connected in series to break the limited current.
status change resistor

Its creation is based on two principles: s melting a solid conductive element in an impervious enclosure by overshooting the thermal melting stress. This is the traditional fuse with the disadvantage that the fuse link has to be replaced after use: s in the above case replace the fuse link by a substance easily vaporisable on a high thermal stress (e.g. sodium or potassium), but whose vapours, subjected to high pressure, quickly recondense after the breaking arc has extinguished: this is the self-regenerating fuse. Note that this fuse type is always backed-up by a parallel resistor to prevent overvoltage. Moreover, a circuitbreaker must also be placed in series (with the fuse and its resistor) to break the circuit before regeneration of the fuse link.
variable resistor formed by the actual breaking arc

The breaking arc in a circuit-breaker is in fact a variable resistor with a value which can be increased by cooling. Use of a sufficiently energetic cooling means ensures the required voltage is reached for current limitation.
On limitation resistors, the arc has the added advantage of not generating overvoltages proportional to the current. Whatever the breaking conditions, maximum arcing voltage remains at a virtually constant and controllable value. Furthermore, arc insertion is automatic on separation of two metal contacts through which a high current flows. In practice, in networks of over 1000 V, it is hard to obtain sufficient arcing voltages in small volumes to limit the current (except for low rated current fuses used in MV up to 36 mV). This explains why use of the arc as a limitation resistor is the most common and cost effective process in LV network protection. All these means favour the creation of ua, thus meeting the need to "aim high". However prompt and early action are also necessary (refer to previous chapter). Hence the advantage of contact propellers and ultra-fast trip units for the limiting circuit-breakers presented in the next chapter.
3. Contact propellers and ultra-fast trip units
contact propellers
The main systems proposed for contact separation (thus arc insertion) are classified according to the origin of the energy required for them to work.
Short-circuit current independent systems With an auxiliary energy source which may be: s mechanical s energy stored in a spring,
s s
pneumatic energy, hydraulic energy. Correct limitation requires accelerations several thousand times the acceleration of gravity, to be obtained in very short times (approx. 1 ms). In practice, these
three energy sources cannot reach this objective in acceptable economic conditions. s chemical The chemical energy contained in explosives is able to develop the required accelerations, but its use remains complex. Moreover, the explosive cartridge must be replaced after use. This process has not therefore really been developed [5] [13].
s electrical The necessary energy is stored in a capacitor. This principle is the result of the experiment conducted at the end of the 19th century by Elihu Thomson (see fig. 5). A flat coil B wound in a spiral is magnetically coupled as far as possible with conductive disk D. The sudden discharge of capacitor C in coil B, controlled by an electronic trip unit, creates induced concentric currents of opposite direction in disk D. The result is a repulsion force F on the disk which is both very high and very fast (less than a millisecond after the tripping order), but short (only a few ms). This process is sometimes used to quickly unlatch limiting circuit-breakers [6] [8].
s electromagnetic With a magnetic circuit and thus with occurrence of the saturation phenomenon. s figure 7a shows this device: the solenoid (S) through which a high (shortcircuit) current flows, swallows the moving magnetic core (N) which strikes the moving contact (B) thus causing the circuit to open. This is the standard diagram for miniature circuit-breakers [10]. s figure 7b shows how this principle is used for devices with a high rated current. The device now consists of a magnetic circuit (C), with airgap, through which current I of the circuit to protect flows. A coil (B) around the magnetic circuit, closes on a bar (A) placed in the circuit airgap. A and B form the secondary winding of an airgapped
current transformer with l as the primary current. Interaction of the secondary current in A and of the magnetic field in the airgap generates a force F which propels a moving contact. This device has been used for limiters installed on DC electrical traction networks [1]. Remark Whereas the energy available with an auxiliary source system is separate from the fault current level, the force developed by current-operated devices and its moment of activation are automatically linked to the value of this fault current. This propeller type therefore has a current level below which the system no longer works: contacts are then separated simply by the device's operating mechanism.
trip unit
F D B
charger
fig. 5: diagram showing a contact propeller according to Elihu Thomson's principle

F B
Current-operated devices The energy required to propel the moving contact is taken off the actual fault current. A great number of devices use this principle. These systems are divided into two major families, depending on whether or not magnetic circuits are used (saturable). s electrodynamic (without magnetic circuit, thus not saturable). Natural contact repulsion under the effect of electrodynamic forces is amplified by special configurations, two examples of which are given below: s repulsion between two conductors forming a loop: a fixed one (A) and a moving one (B), rotating around point 0 (see fig. 6a). s repulsion on a moving contact in bridge (B) accentuated by crossing of the fixed contacts A and A' (see fig. 6b).
B O A i
a) simple repulsion
i b) reinforced repulsion
A'
fig. 6: diagram showing contact propellers with self-energized electrodynamic current

F fixed core S N B O A
I C
a) with magnetic core, for miniature circuitbreaker
b) with magnetic circuit in C, for circuitbreaker with high rated current.
fig. 7: diagrams showing self-energized electromagnetic contact propellers
ultra-fast trip units

Their function is to mechanically confirm contact "reflex" separation. Their presence is vital when the contact propeller is self-energized and the moving contact does not latch in the open position. In actual fact, given the mechanical inertia of the moving contact, contact separation must be relieved in less than 10 ms by the opening mechanism, since the repulsion force is lost after the shortcircuit current is broken. Otherwise, the pressure springs close the contacts and restore the short-circuit.
These devices use the same electrical, electrodynamic and electromagnetic principles described in the above paragraph. Thus, to give an example, in some miniature circuit-breakers, the moving magnetic core (N in the diagram in figure 7b) is used not only to accelerate contact separation, but also to quickly unlatch the mechanism holding the moving contacts in the closed position. Likewise, the principle shown in figure 5 has already been used for this unlatching function [6] [8]. Other ultra-fast trip units use the pressure developed by the electric arc in
the arc chute when breaking a high current. As an arc moves through an arc chute, it builds up a pressure of several bars which becomes available as soon as the limited current reaches its highest value (at point P in figure 4). Use of this principle, recently patented by Merlin Gerin, enables the construction of ultra-fast and highly limiting circuitbreakers: via appropriate ducts and valves, this pressure is used to actuate a piston which controls in less than 5 ms the circuit-breaker opening mechanism.
4. conclusion
You will now have realised the importance of research in the creation of high performance circuit-breaker ranges.
Since 1930, Merlin Gerin, along with other manufacturers, has helped increase safety and reliability of electrical power distribution. Recent
patents filed show the promising future of limiting circuit-breakers in electrical power distribution, with their capacity to increase its discrimination and hence availability.
5. bibliographie
[1] Disjoncteurs ultra-rapides pour courant continu P. BRANCHU. Patents n 596.483 (1925) - 629.040 (1927) - 721.451 (1931). [2] Nouvelle disposition de branchement pour limiteurs de courant. K. KESL. RGE, February 1942, p. 85-96. [3] Coupure des courants de lordre de 100 kA en BT. G. BOUVIER. RGE, November 1955, p. 554. [4] Disjoncteurs limiteurs basse tension pour courants alternatifs. A. MOLAS. RGE, May 1958, p. 259-276.
[5] Problmes de coupure et utilisation des limiteurs Is. P. BRUCKNER. ETZ-B, H3, March 21st 1959. [6] Disjoncteurs rapides avec limitation du courant. A. ERK. ETZ-B, H7, April 2nd 1962, P. 169. [7] Laccroissement des courants de court-circuit et leur matrise dans les installations BT. P. BRUCKNER. ETZ-B, H19, September 17th 1962, p. 511. [8] Limitation et coupure du courant avec un disjoncteur Gearapid dans un rseau alternatif. H. FEHLING. ETZ-B, H19, September 17th 1962,
p. 537. [9] Disjoncteur limiteurs de courant. E.B. HEFT. Power, July 1968, p. 55. [10] Exploitation de la limitation des courants de court-circuit. J.R. COCHENNEC. Revue Klckner-Moeller, November 1970. [11] Interrupteurs limiteurs du courant de court-circuit. G. CANTARELLA. L'Elettrotecnica, July 1970. [12] Dveloppement de disjoncteurs sans fusibles limitation de courant. H. SUZUKI. Revue Hitachi, vol. 19, n 12, p. 441. [13] Le systme Pyristor de Ferraz. G. GUEZ. Moniteur de l'Electricit, October 1984, p. 42.
Ral.: Sodipe - Valence Edition: DTE - Grenoble 11-94 - 2500 - Printing: Clerc Printed in France
n 164
current transformers for HV protection
Michel Orlhac Graduated from the Ecole Centrale de Paris in 1977. After one year's specialisation at the university of Stuttgart (Germany), he entered the overseas projects department of Stein Heurtey (iron and steel engineering). In 1980 he joined Merlin Gerin, becoming part of the technical section of the High Voltage Prefabricated Switchgear Department (P.S.H.T.) where he completed a study on current transformers. This Cahier Technique publishes the results of this study. At present he is the marketing manager for France-Transfo, a subsidiary of the Merlin Gerin Group.
current transformers for HV protection
content
1. Theoretical review Hysteresis - Saturation Characterisation of CTs Current transformers Functional CTs Protection relays Technological evolution Experiments - Wiring Testing with symmetrical constant currents Testing with asymmetrical currents Conclusions on CTs delivering on an overcurrent relay
2. General current protection information
3. Response of a CT in saturated state
p. 4 p. 5 p. 6 p. 7 p. 8 p. 8 p. 8 p. 9 p. 10 p. 11 p. 12
Current transformers or CTs take up a lot of space in HV cubicles. Thorough knowledge of how they work makes it possible to: c reduce their dimensions and thus their cost, c use standard CTs in a larger number of configurations. The purpose of this study is to learn more about CT operation in association with protection relays and to lay down a few rules for sizing them properly. After a brief theoretical review of CT operation and current protection devices, the behaviour of the CTprotection relay combination is studied in two particularly important cases in HV: c overcurrent relay supplied by a heavily saturated CT, c protection relay connected to two CTs in parallel in duplex cubicles. This study is completed by experimental results.
4. Parallel cubicle operation 5. General conclusions Appendix: CT standards
p. 14 p. 16 NF C 42-502 (Norme Franaise) p. 17 IEC 185 p. 19
1. theoretical review
Current transformers consist of a magnetic circuit in toroid form. The primary is made up of n1 turns or simply a single conductor crossing the toroid (n1 = 1). The secondary is wound in n2 regular turns around this toroid (see fig. 1 and 2). Ampere's theorem states that the sum of the ampere-turns is equal to the circulation of the magnetic field vector. n1 i1 + n2 i2 = H n dI

e 2 = v 2 + R 2 i2 + l 2
di 2 dt
d d e 1 = n1 and e 2 = n 2 dt dt If all the functions described are sinusoidal of pulsation , the following can be written vectorially:
V1 = E1 + (R1 + jl1 ) I 1 E2 = V2 + R2 I 2

i1
dI n
E1 = jn1 E2 = jn2
H = magnetic field n = tangent unit vector A transformer is said to be perfect when

H Toroid n dI = 0 In the real transformer, this term refers to the error introduced by the magnetic circuit and defines the exciting current ie formed at the secondary by: n1 i1 + n 2 i 2 = n 2 i e If n =

Toroid
I1
n The wiring diagram in figure 3 and equations (1) result in the vectorial representation in figure 4. The exciting current I e is broken down on the axes and E into:
+ I2 = Ie
i2
fig. 1.
I e = Ia + Im
i1
n2 is the winding ratio, the n1 relationship is written as: i1 + i2 = ie n
c where I a represents the part of this current lost in the magnetic circuit (iron losses due to hysteresis and eddy currents). c and I m is the magnetising current which transfers power from one winding to the other by creation of a magnetomotive force which induces the flux .
i2
The transformer can then be represented (see fig. 3) as having two parallel elements: c a perfect transformer of ratio n delivering a current i1/n at the secondary, c an impedance which consumes a current ie. Moreover, each winding, both primary and secondary, creates a slight voltage drop due to the resistance of the winding (R1 and R2) and to the leakage inductances ( l 1 and l 2 ). Since, in the case of the CT, the secondary winding is tight and regular, l 2 need not be considered. If is the flux common to both windings, the following can be written between the emf e1, e2 and the difference in potential v1, v2:
fig. 2.
i1
R1
l1
i1/ n
R2
l2
i2
n = n 2/ n 1 e1 n2 n1 e2 im
ie
V1
V2 ia
perfect transformer
real transformer
v 1 = e 1 + R 1 i1 + l 1
di1 dt
fig. 3: CT schematic diagram.
hysteresis - saturation
Magnetic circuit quality is defined by the relationship it imposes between the induction vector B and the magnetic field vector H. At a given moment and in a fixed point, these two vectors are linked by the relative permeability of the magnetic material r such that: l 1 I1
R1 I1 V1 E1 I 1/ n I1 order of creation of values: I1, V1 I1, Ie
1
B Im

E1
B = o r H
A magnetic circuit is thus characterised by the curve b = f (h) known as the magnetising curve. According to the different material types, the curves in figure 5 are obtained, the results of sinusoidal excitation (primary current). In sinusoidal state, b represents voltage since:
B=

Ia H
Ie
2
V2 E2
I2
E2
n S
E2 = n2 j
V2 R2 I2
I2
V E2
h represents the exciting current since n2 Ie = H n dI
Toroid
N.B.: the real proportions, between the representative vectors of primary and secondary values, are not repected.
l 2 I2
fig. 4: vectorial representation of a CT.
assuming that
H n = H = constant
hypotheses:
magnetising curves b
ie and B as a function of time B, ie
n 2 Ie = L H Perfect transformer Permeability of the medium is assumed infinite

H = 0 hence I e = 0 and I 2 =

perfect transformer
I1
n This hypothesis approaches the real situation with CTs since they normally work far below saturation. I2 is then the mirror image of I1.
linear transformer
Linear transformer Permeability of the medium is constant B = Cste x H hence ie and i2 are sinusoidal functions. Saturable transformer without hysteresis Saturation is the sudden variation of r from a high value to a low value at the point known as the saturation bend. Induction b then increases only slowly and ie deforms to form a peak.
saturable transformer without hysteresis
saturable transformer with hysteresis
h exciting current: ie induction B
fig. 5: magnetising curves and their incidence on ie.
Saturable transformer with hysteresis The magnetising curve is undoubled, thus indicating the resistance of the magnetic circuit to the induction variations. Curve ie then exhibits a characteristic swing. The magnetising curve of a CT can easily be observed using an oscilloscope. A sinusoidal voltage V2 (t) is applied to the secondary (the primary is not charged). The current ie (t) absorbed then represents the exciting current and is proportional to the magnetic field vector H. Integration of voltage V2 represents the flux 2 which is proportional to the magnetic induction vector B (see fig. 6a). Integration of a sinusoidal value causes a rotation of /2 (90). It is thus sufficient on an oscilloscope: c to sweep with ie, c to apply voltage V2 to the vertical amplifier. The magnetising curve of the material is thus obtained (see fig. 6b).
c for measurement CTs The module error I / n I2 M = 1 I1 / n The phase error

= (I1, I2 ) 10
2
Bear in mind that the less the CT is loaded (the more it is below its accuracy level power Y), the greater its accuracy. Its real accuracy level is therefore greater than its rated accuracy level Fp. This point is developed in chapter 3. Admissible short term current Expressed in kA it is the maximum current admissible Ith for one second (the secondary being short-circuited). It represents CT thermal overcurrent withstand. (standard values are given in the standards in the appendix). For times other than 1 second, the heat conservation law I2 t = cste can be applied: for t < 1 sec. the calculation gives I > Ith, thus increasing electrodynamic forces. However, the limit guaranteed value is Idyn = 2.5 Ith.
rd
An accuracy class X is given (generally 0.5 or 1) which expresses limit values of the module error M and of the phase shift error as a function of the load ratio N: I N = 1 (N var ies from 0.1 to 1. 2) I1n for N = 1 M = X (in class 0.5 for I1 = I1n, M = 0.5 %) (for value details refer to the standards in the appendix). c for protection CTs The composite error c
c = 1 I1 / n 1 T
i1 i2 n
dt
V2(t)
characterisation of CTs
CTs are characterised in practice by the following values (according to standards NF C 42-502 and IEC 185). CT voltage This is the operating voltage applied to the CT primary. Note that the primary is at the HV potential level and that one of the terminals of the secondary (which must never be opened) is normally earthed. Just as for all equipment, a maximum 1 min withstand voltage at standard frequency and a maximum impulse withstand voltage are defined (refer to the standards in the appendix). e.g. for a rated voltage of 24 kV, the CT must withstand 50 kV for 1 mn at 50 Hz and 125 kV impulse voltage. Rated winding ratio Normally takes the form: l1/l2. I2 is very generally 5 A or 1 A (for rated values of I1, refer to the standards in the appendix). Rated accuracy class This depends on whether the CT is used for measurement or protection:
Protection CTs are characterised by 3 symbols: Y, P, Fp: Y = error rate (5 or 10), P = protection, Fp = accuracy limit factor which gives the limit values of errors M, and c as a function of the load ratio N. For N = Fp c = Y (in class 10P5 for I1 = 5 l1n: c = 10 %) (for value details, refer to the standards in the appendix). For a CT working at a rated induction Bn, a saturation coefficient Ks such that:
ie(t)
a - scales: ie = 0.25 A per square V2 = 50 V per square.
2 = V2 dt (or B)
Ks =
Bs Bn
where Bs is the saturation induction characterising the core material. In practice K s Fp and they are often treated as the same in calculations. Accuracy level power Expressed in VA, it indicates the power that the secondary can supply while respecting the rated accuracy class Y, P, Fp. It represents the total consumption of the secondary circuit (except for CT), i.e. the power consumed by all the connected devices as well as the connecting wires. (for rated values, refer to the standards in the appendix).
ie (or H)
2 I a = constant
b - scales: ie = 0.25 A per square 2 = 0.077 V.s per square. fig. 6 : oscillographic reading of curves i(t) V2(t) and h(b) of a CT, 50/5, 15 VA, 10P20 where: V2 = 83 V and le = 0.26 A.
2. general current protection information
Protection devices have many functions since they have to: c protect equipment from destruction or damage as a result of faults (shortcircuit, overload...), c ensure normal operation of the installation and its equipment (control, load shedding...), c guarantee safety of personnel.
current transformers
Since relays cannot be connected directly onto the MV network, the information they receive comes from current transformers or CTs (see fig. 7) and from voltage transformers or VTs. When primary current is high, the CTs are of the cross bar type, and when it is low they are of the wound primary type. CTs have a number of roles to play in electrical networks: c supplying at their secondary a current exactly mirroring the one flowing in the HV conductor concerned, c providing galvanic insulation between the HV and the measuring and protection circuits, c protecting the measuring and protection circuits from damage when a fault occurs on the HV network. Using this current image in the HV conductor, the relay generates in turn a tripping order according to the type of protection it provides and the values at which it has been preset [threshold(s), time delay(s)....]. This order is transmitted to one or more breaking devices (circuit-breaker, contactor, switch). CT configurations vary according to the type of protection to be provided. Overcurrent protection (see fig. 8) This directly uses the current information supplied at the CT secondary to detect short-circuit or overload currents or calculate the thermal status of a machine. Note that this configuration type must also contain the protection devices using in addition to VTs: c directional overcurrent protection, c power protection (active or reactive). Earth leakage protection (see fig. 9) This measures the current difference between two CTs, one connected downstream and the other upstream
CT with cross primary winding (cable) 1 secondary - 600/1
Wound type CT with wound primary winding 1 secondary - 200/5
Wound type CT with wound primary winding 2 secondaries - 200/5 and 100/5
fig.7: different types of CTs.
from part of the network to be monitored (a motor, a transformer, a busbar...) to quickly detect and isolate any faults inside that part. Zero sequence protection This monitors the zero sequence component Io of the three-phase current which appears during phaseearth faults. There are two possible configurations: c a toroid transformer encircling the three phase conductors (if possible). This configuration (see fig. 10a) enables detection of small zero sequence currents (1 to 100 A). c three CTs achieving in the neutral connection of their secondary the sum of the three phase currents. This configuration (see fig. 10b) is the only one possible for large and numerous cables or busbar ducts. It is not recommended when the zero sequence current to be detected is 5% less than ln (or even 12% for consumer substations according to standard NF C 13-100 (French Standard)).
I1
I2 = I1/m
relay
fig. 8.
I1 I'1 - I1 relay
I'1
fig. 9.
a)
Io
relay
functional CTs
In HV cubicles, the current transformer function takes on a new dimension as a result of its content and shape. Thus: c a number of CTs can be moulded in the same enclosure: one core for the measurement function, one core for the protection function and sometimes even a third core for earth leakage protection,
b)
Io relay
fig. 10.
c the enclosure is used to ensure insulation between two compartments and plugging-in of the breaking device: the CT is then said to be functional. An application example is given in the metalclad cubicles for withdrawable switchgear (see fig. 11 and 12). Overall dimensions are thus reduced by using one insulating enclosure (the most appropriate), thus also reducing costs.
the protection relays

The equipment currently available is based on the three technologies: electromechanical, analog and digital. The oldest of these is electromechanical technology: relays are simple and specialised (current, voltage, frequency, ... monitoring) but their accuracy is poor as their settings may be altered over time. The last two technologies benefit from the advantages provided by electronics (see fig. 13): c compact dimensions of the device, c low power required for acquisition of current information (a few fractions of VA), c response time not dependent on the current received by the relay, c reliability increased by lack of mechanical parts (no dirt accumulation or corrosion, not affected by impacts), c low cost since they use mass produced non-specific electronic components. Finally, in the nineteen eighties, digital technology made it possible, thanks to microprocessor processing power, to produce information processing units able to: c globally provide the various protections,
c replace relays (automation) in the cubicle, c provide operators with measurement of electrical parameters. These units, with their increased vocation, are: c flexible (protections are chosen simply by programming), c parameterisable (large choice of settings), c reliable (they are fitted with selfmonitoring or with watchdog and self-test), c economic (reduced wiring and implementation time). Their digital communication and powerful algorithms also enable additional functions such as logic discrimination to be performed. This communication capacity means that genuine network operation (similar to technical management of industrial installations) is now possible. Finally, their ability to acquire and process the information provided by sensors allows them to make full use of the performances of the new nonmagnetic sensors.
fig. 11: functional CT for HV metalclad cubicles (Merlin Gerin).
technological evolution
In this current sensor field, sensors with wide measuring bands are being increasingly used instead of current transformers (1 or 5 A). These sensors based on Rogowski's principle (nonmagnetic sensors) are currently on the market and provide distributors with optimised solutions (fewer alternative versions and simplified choice) which are far more efficient (improved response curve linearity) than traditional transformers.
fig. 12: installation example of functional CTs in a Fluair 200 12 kV HV metalclad cubicle (Merlin Gerin).
fig. 13: Vigirack static relays (Merlin Gerin).
3. response of a CT in saturated state
The emergence of static relays leads to revision of protection behaviour as a whole in the case of strong currents: as the CT saturates beyond a certain threshold, the first reaction is often to avoid this by raising the threshold. However, this results in both additional costs (more efficient, larger, more space consuming CT) and in the risk of excessive temperature rise of the relays. On the contrary, saturation plays a useful role for the measurement function since primary current image accuracy is only useful up to the value of the rated current I1n. Beyond this point, the measurement ceases to be of any use and saturation must occur for a low current (2 to 3 I1n) in order to limit the secondary current and protect the measuring instruments. It is thus necessary to know the response of the CT in saturated state to ensure the protection device works properly when the primary current exceeds rated current strength, particularly for the high values which appear if a short-circuit occurs. In theory, induction in the core reaches a plateau at the saturation bend, thus limiting current strength at the secondary. In actual fact the experiment performed will show that current strength at the secondary slightly increases and that protection relay operation is quite satisfactory.
i2 i1 i2 resistance tested CT R measuring oscilloscope relay
standard CT
u2 i1 u1
fig. 14: diagram for checking proper relay operation.
I2 (A) 500 1
100
2 3
10
experiments - wiring
A current i1 is injected in the CT primary, and the current supplied by the secondary in a load Z containing a relay R and a resistance is analysed (see fig. 14). The currents at the secondary l2 are given, according to the current supplied at the primary (represented by the I parameter N = 1 ) for various loads Z I1n and various CTs (see fig. 15).
10 100 500 N = I1/I1n
Fig. 15: I2 = f(N) for 1 CT only (15 VA 10P5 100/5). Load Z at the secondary: 1. relay only, 2. Z = rated Z of CT, i.e. 0.6 and cos = 1, 3. Z = rated Z of CT, i.e. 0.6 and cos = 0.8.
testing with symmetrical constant currents

Testing at resistive rated load The test was carried out using a CT with low performance: 10P5, 50/5 with a rated load Z of 15 VA (at 5 A) made up of an overcurrent relay and a resistance. Two relays were used: v a Vigirack static relay, v an electromechanical relay. As both these relays have a low internal resistance, a resistance was added to reach roughly 0.6 , i.e. 15 VA at 5 A (connecting wiring included). Because the inductance of the electromechanical relay was low (15 H, i.e. cos = 0.95 for the relay only), the load can be considered to be purely resistive in both cases. The test consisted in making current l1 vary in the range I1n = 50 A at I1max = 54 kA I.e. N max = = 54,000 = 1,080 and 50
Nevertheless, the rms current I2 continues to increase as is shown in line 2 in figure 15. As I2 increases, the power supplied at the secondary P2 = Z l2 and the power 2 delivered at each relay Pr = R l2 also 2 increase. This accounts for the tripping of both relay types as from threshold to which they were set right up to Nmax. Testing at rated partly inductive load This test resembles the previous one. However, a choke is placed in the secondary circuit to represent the case of an electromechanical relay connected by itself to the secondary which would consume the rated power of the CT. In practice, these relays never fall below cos = 0.8. In this test, the current I1 explored the range I1n = 50 A to I1 max = 16,400 A, i.e. N max = 328 and = N max = 65.6 Fp
v both relays trip from their threshold right up to Nmax. Testing at reduced load The secondary load only comprises the relay and the connecting wires. Compared with the rated load of 15 VA, this represents a load of roughly 9%. c results Curves i2 (t) (see fig. 16c) and i2 (N) (see line 1 in figure 15) show that the saturation bend is far higher than at rated load. This bend follows the law:
K s (P2 + R 2 I2 ) = constant
2
N max 1,080 = = 216 5 Fp
(the latter value indicates the level of saturation to which the CT was subjected). c results The current i2 (t) collected at the secondary takes the form of a peak above: N = 10 (see fig. 16a).
c results Current i2 (t) assumes the curve given in figure 16b. The presence of a choke spreads out the peak, hence the lower value I2 (see line 3 in figure 15). With respect to testing at pure resistive load: v I2 is multiplied by a factor of 0.65, v the total power supplied at the secondary is multipled by a factor of 0.4,
with P2 = Z l 2 is the total power 2 supplied at the secondary (consumed by the relay and the connecting wires). R2 = internal resistance of the CT secondary winding, Ks = saturation coefficient (real or rated). Thus, in practice, when a CT delivers into a load less than its rated accuracy level power (in VA), saturation occurs at a far higher overcurrent level than the rated saturation coefficient Ks. This phenomenon must be taken into . consideration and calculated for each application since it may generate overcurrents in the secondary which are incompatible with the thermal and dynamic withstands of the relays connected to the CT secondary (for calculation, refer to the conclusions given below).
i2 i2 v2 v2
i2 v2
fig. 16 a - CT 15 VA 10P5 50/5 testing at purely resistive rated load I1 = 16,400 A scale: i2 = 100 A/square; v2 = 100 V/square.
fig. 16 b - CT 15 VA 10P5 50/5 testing at rated load with cos = 0.8 I1 = 16,400 A scale: i2 = 25 A/square; v2 = 50 V/square.
fig. 16 c - CT 15 VA 10P5 50/5 testing at reduced rated load relays + connecting wires - I1 = 14,200 A scale: i2 = 100 A/square; v2 = 5 V/square.
testing with asymmetrical currents

The test was performed using an asymmetrical current, i.e. the sum of a symmetrical sinusoidal current and a DC component with the following characteristics: 2.3 Irms These values are slightly less than those in standard NF C 64-100 for which Irms i.e. 20% of asymmetry at 70 ms. The secondary load is identical to that of the main test at resistive rated load comprising an electromechanical or static relay. c results Both relays correctly respond in a few ms and in the same manner as in symmetrical testing throughout the range explored (up to 1 = 140 kA peak with Irms = 54 kA). Remarks: c the first peak seen at the secondary by the relays is enough to make them trip, if its energy is sufficient: this is the case for Irms greater than 2 kA but below this value (see fig. 17a) the third peak is required; c the CT does not saturate during the first negative peak of the primary current for 1 = 4 k ; c the response delivered by the CT on the first negative peak of the primary (or even secondary) is normally shorter than the responses in steady state (which is reached as from the sixth peak); c the above points show that for higher peak factors (case of off-load energising of transformers with an = 3.7 ), there is a risk of the Irms response at the secondary disappearing during the first peaks. If, in addition, the time constant of the primary current DC component is high (t = 80 ms in the case quoted), this disappearance continues until the primary current crosses the 0 axis. This phenomenon is shown on the curves in figure 17 (tripping time moves to 68 ms). = 205 = 1.8 2
a)
scales: v1 i1 500 A/mm
i2 10 A/mm
no saturation
b)
scales:
v1 i1 1,000 A/mm
i2 10 A/mm
c)
scales:
v1
i1 2,000 A/mm i2 20 A/mm
fig. 17: CT secondary responses on an asymmetrical primary current for: a) lrms. 1.4 kA, b) lrms. for 14 kA and 1 32 kA, c) lrms. 54 kA and 1 140 kA.
conclusions on CTs delivering on an overcurrent relay

The above tests show that for both electromagnetic and static relays, tripping is obtained whatever the
current value, even if the CT is strongly saturated. Thus, the CT saturation coefficient Ks must be calculated not according to the short-circuit current lcc but according to the maximum setting threshold of the associated relay (see fig. 18 and 19).
choice of Ks I1n ITIn I1r = I1n Ith = Icc
Icc
I2n Ir min Ir max I2max I2 max Ithr
of the network
I1n I1r Icc
of the relay
= rated current = I1n setting current = short-circuit current = maximum short-circuit time
Ir min to Ir max = setting range Ithr = admissible short term current (1s)
of the CT
ITIn I2n I2 max
= rated primary rating = rated secondary current = CT response to Icc
fig. 18: characteristics to be considered for defining a CT.
1. the saturation threshold Ks must correspond to the maximum setting value of the relay. 3. this CT must electrodynamically withstand the peak value 2.5 Icc. 2. the CT must thermally withstand the current Icc for a time at least equal to the breaking time of the short-circuit by the circuit-breaker.
4. the secondary circuit must thermally withstand the maximum secondary rms current I2 max created by Icc at a primary for the time .
5. the relay setting range (Ir mini, Ir maxi) must be large enough to cover the CT response at the setting current of network B I1n.
fig. 19: general rules for sizing a CT.
4. parallel cubicle operation
Power supplies with double busbars are frequently used in HV network configurations. There are currently two solutions for most cubicles: c the double busbar cubicle: the circuitbreaker may be connected to either busbar without discontinuity of service. One of the drawbacks often
encountered in this system is highly complex lockings: c cubicles connected in duplex (see fig. 20). Using standard elements, this solution can advantageously replace the double busbar, as it is more reliable. As on the new cubicle generations, the CTs are standard elements used to
provide insulation between compartments and to plug in the breaking device. This arrangement makes it necessary to connect the relays (which are not backed up) on each CT secondary. This has resulted in the study below concerning operation of two identical CTs connected in parallel on the same load.
1455
1300
1455
relay
1800
1060
fig. 20: connection of 2 cubicles in duplex relay
Wiring diagram Connection of two cubicles in duplex, as shown in figure 18, results in the diagram in figure 21 for protection. One of the CTs (said to be live) is supplied at the primary by the HV network; its secondary supplies a current i2 broken down into a current im on the secondary of the other CT (said to be dead) and a current ir on the rated load of 15 VA made up of an electromagnetic or static relay and a pure resistance. The tests were performed on two identical CTs of the same series (15 VA 50/5 10P5 as in the above paragraphs).
standard CT
I1
resistance tested live CT shunt i2 im Im ir shunt R
relay Ir recorder
I2
tested dead CT
fig. 21: wiring diagram for study of a parallel-connected CT.
Results These are given in the curves of figures 22 (currents as a function of time) and figure 23 (root mean square currents and tripping times). The following observations are made: c both relays quickly respond from their tripping threshold to = 72, c the static relay trips in a constant time T 20 ms, whereas the electromagnetic relay reacts as a function of I2 (T 80 ms at the tripping threshold to T a few ms at = 72); c the secondary current I2 continues to increase but two separate zones appear: v before = 10, Ir I2 and lm << lr the secondary current flows entirely into the relays since the dead CT acts as an infinite impedance, v after = 10 Im I2, which means that the secondary current mostly flows into the dead CT but, however, Ir continues to grow, thus causing the relays to trip (do not forget with respect that ir is dephased by 2 to im); c the current lr flowing in the relays during testing with two CTs is lower than for testing with one CT (at N = 300, roughly - 40%). Conclusion The connection of two CTs in parallel presents no problem: c for low currents: hardly any current flows into the dead CT, c for high currents: sufficient current flows into the relay to trip it.
scales: im 9.4 A/mm
ir 9.55 A/mm
i2 19 A/mm
i1 955 A/mm
Vigirack relay
fig. 22: currents at the secondary of 2 parallel-connected CTs. I1 = 12,500 A.
I(A) T(ms) 100
I2 (total secondary current) Im (current flowing in the dead CT)
Ir (current flowing in the relay) T: static relay tripping 10 T: electromagnetic relay tripping
10
100
N = I1/I1n
fig. 23: rms currents and tripping times of the relays connected to 2 CTs in parallel (10P5 50/5 15 VA).
5. general conclusions
The conclusions in chapters 4 and 5 show that: c the relays operate correctly in both cases studied: v high CT saturation, v parallel-connection of two CTs; c static relays give the most reliable response (constant operating time for all currents greater than the setting threshold). Moreover, static relays generally have a very small acquisition time, thus meaning operation is more reliable when the CT is strongly saturated and supplies a very short current impulse. Do not forget, however, that the transient phenomena considered were limited to the asymmetrical current less than: = 2.5 Irms
Nevertheless, CT saturation, as shown in this experiment, should not be considered a handicap: c when a CT supplies one or more measuring instruments, saturation, by limiting rms current at the secondary, protects the devices which, moreover, do not generally need to be very accurate above l1n. c when a CT supplies a protection device, operation is ensured even if saturation occurs. The idea of sizing a CT according to the highest current it may have to withstand at the primary must therefore be rejected. Moreover, this oversizing is risky for the relay and cabling which could be seriously damaged.
appendix: CT standards
NF C 42-502 (French Standard)
Rated insulation levels The insulation levels recommended by the standard are given in table II A presented in figure 24. Normal rated current values c at the primary (in A): 10 - 12.5 - 15 20 - 25 - 30 - 40 - 50 - 60 - 75 and their decimal multiples or submultiples. Preferential values are given in bold. c at the secondary (in A): 1 - 5
to impulse voltage (peak value) (kV)
highest voltage for equipment (kV) 0.6 1.2 2.4 3.6 7.2 12 17.5 23 24 36 52 72.5
withstand voltage 1 minute at standard frequency (rms value) (kV) 3 6 11 16 22 28 38 45 50 70 95 140
Accuracy class c measurement CTs The normal accuracy classes are: 0.1 - 0.2 - 0.5 - 1 - 3 - 5. The rated frequency operating range is 96% to 102% of rated frequency. For transformers of accuracy classes 0.1 - 0.2 - 0.5 and 1, the current error and phase shift in the rated frequency range must not exceed the values in table III (see fig. 25) when the secondary load is between 25% and 100% of accuracy load. For transformers of accuracy classes 3 and 5, the current error in the rated frequency range must not exceed the values in table IV (see fig. 26) when the secondary load is between 50% and 100% of accuracy load. In all cases, the load used must be inductive with a power factor of 0.8, unless the corrresponding power is less than 5 VA, in which case its power factor is the unit. On no account must the load be less than 1 VA.
45 60 75 95 95 125 170 250 325
fig. 24: insulation levels (table II A).

phase shift, for current values given as a percentage of rated current minutes 10 10 20 60 120 20 8 15 45 90 100 120 5 10 30 60 5 10 30 60 centiradians 10 0.30 0.60 1.8 3.6 20 0.24 0.45 1.35 2.7 100 0.15 0.3 0.9 1.8 120 0.15 0.3 0.9 1.8
accuracy class
current error (ratio error) as a percentage, , for current values given as a percentage of rated current 20 0.35 0.75 1.5 100 120 0.1 0.2 0.5 1.0 0.1 0.2 0.5 1.0 0.25 0.20 0.5 1.0 2.0
% I1n 10 0.1 0.2 0.5 1
Note: after agreement between manufacturer and user, guarantees can be provided for accuracy and phase shift, between 120% and 200% of In n.
fig. 25: error limits (table III).
accuracy class
current error (ratio error) as a percentage, , for current values given as a percentage of rated current 120 3 5 3 5
% I1n 50 3 5
There is no phase shift limit for classes 3 and 5.
fig. 26: error limits (table IV).
c Protection CTs The normal accuracy limit factor values are: 5 - 10 - 15 - 20 - 30 - 40. The rated frequency operating range is 90% to 110% of rated frequency. The normal accuracy classes are 5P and 10P. For accuracy level power and in the rated frequency range, the current error, phase shift and composite error must not exceed the values in table V (see fig. 27). To determine the current error and phase shift, the load must be inductive and equal to the accuracy load with a power factor of 0.8, unless the corresponding power is less than 10 VA; in this case the load could be resistive (unit power factor). To determine the composite error, the load power factor may be between 0.8 (inductive circuit) and the unit, the value being set by mutual agreement between manufacturer and user. Accuracy level power The normal accuracy level power values are: 2.5 - 5.0 - 10 - 15 - 30 - 75 100 VA. Admissible peak current and short term current Admissible peak current and short term current (Ith). The short term current (Ith) must be specified for each transformer. Their preferential values are given in paragraph 10.1 (see fig. 28). Notes c for the highest network voltage less than or equal to 36 kV, the admissible short term current value is constructively linked to rated current value. It is thus frequently expressed as a multiple of rated current, for which the preferential values are: 40 - 80 - 100 200 and 300. c if no admissible values as a function of time are given, it is accepted that the transformer can withstand for a time t, expressed in seconds, a current with a root mean square value given by the formula: I' th = I th t2
accuracy class
5P 10P
ratio error for currents between In and 2 In (as a %) 1 3
phase shift for rated current minutes 60 centiradians 1.8
composite error for accuracy limit current (as a %) 5 10
fig. 27: error limits (table V) .
c the admissible times for the admissible short term current are set from the cold state. However, at the user's request, the manufacturer is obliged to indicate, for a given type of device, the admissible short term current based on a state corresponding to operation, the heating current and maximum ambient temperature. However, in the latter case, verification of admissible short term current cannot be made mandatory as an acceptance test. Admissible current peak value (Idyn). The admissible current peak value is 2.5 Ith. However, another value can be accepted provided it is stated on the identification plate.
highest network voltage (kV) 3.6 7.2 12 17.5 23 24 36 72.5 100 245 420
Ith (kA)
10 8 8 8 8 8 8 20 20 20 40
16 12.5 12.5 12.5 12.5 12.5 12.5 25 31.5
25 16 16 16 16 16 16
40 25 25 25 25 25 25
40 40 40 40 40 40
where t2 > t1 bearing in mind that Ith is given for t1 (= 1s).
fig. 28: preferential values of Ith (paragraph 10.1).
IEC 185
This is the reference standard. The NF C 42-502 (Norme Franaise) differs only slightly from it. The differences are as follows: Rated insulation levels The IEC standard gives two tables: c the same table as the NF C standard for European countries, c another table as per USA practice with slightly more stringent values: refer to table II B (see fig. 29). Normal rated current values Same preferential values at the primary. At the secondary possibility of a I2n = 2 A. Accuracy class c measurement CT Current errors in module and phase are the same in class 3 and 5. For classes 0.1 - 0.2 - 0.5 and 1, the errors are the same, except for the 10% of I1n column which is replaced by 5% of I1n with the errors listed in table IV A in figure 30. Moreover, the IEC standard defines two additional classes, 0.2 S and 0.5 S for CTs with special applications (connection with special electrical energy meters). In this table, the module and phase errors are given for I2n = 5 A only. c protection CT The IEC gives the same limit errors. The only difference is that the accuracy limit factor, Fp = 40, does not exist. Accuracy level power The IEC only gives the same normal values up to 30 VA. Beyond this point, power can be chosen to meet needs. Peak current and short term current Unlike the NF C standard, the IEC standard does not define preferential values of Ith for each network voltage. However, application of the law i2 t = Cste to define the Ith is limited to: 0.5 < t < 5 s.
highest voltage for equipment Um (rms value) kV 4.40 13.20 13.97 14.52 26.4 36.5 rated lightning impulse withstand voltage (peak value) network power 500 kVA > 500 kVA kV kV 60 95 75 110 rated short term withstand voltage at standard frequency (rms value) kV 19 34
150 200
50 70
fig. 29: rated insulation voltages set for the U.S.A (table II B).
accuracy class 0.1 0.2 0.5 1 accuracy class
error M for I1 = 5 % of I1n 0.4 0.75 1.5 3 error M for values as a % of rated current I1n 5 20 0.75 0.35 0.2 1.5 0.75 0.5 0.2 0.5 0.2 0.5
error for I1 = 5 % of I1n minutes 15 30 90 180
centiradians 0.45 0.9 2.7 5.4
error for values as a % of rated current I1n minutes 100 120 1 5 30 15 90 45 10 10 30 30 10 30 centiradians 20 100 120 1 5 20 100 120 0.3 0.9 0.3 0.9 0.9 0.45 0.3 2.7 1.35 0.9
% l1n 1 0.2S 0.5S
fig. 30: accuracy class (table IV A).
Ral.: Sodipe - Valence - Photo.: IPV Edition: DTE - Grenoble 03-95 - 2500 - Printing.: Clerc
n 166
photographie
enclosures and degrees of protection
Jean Pasteau Graduated in physics sciences in 1959, he joined Merlin Gerin in 1962 in the air-blast circuitbreaker design-office where he was mainly in charge of the preparation and follow-up of the technical part of contracts. Since 1983 he has been Technical Delegate for Merlin Gerin in standardization bodies (IEC, UTE, CENELEC,...) That led to him becoming in particular: s chairman of the French Committee 70 in charge of the preparation of standards for Degrees of protection provided by enclosures in 1987, s convenor of the BTTF 68-3 in charge of the preparation of the European standard for the IK Code in 1991.
E/CT 166, first issued 1993, january
enclosures and degrees of protection
summary
1. Introduction 2. General p. 4 p. 5 p. 6 p. 7 p. 7 p. 8 p. 9 p. 9 p. 9 p. 10 p. 11 p. 11 p. 12 p. 12 p. 12
Definitions Uses of a code system 3. The IP Code General First numeral Second numeral Additional letter Supplementary letter 4. The IK Code Introduction Degrees of protection 5. Application to the design of electrical equipment 6. Conclusion Appendix 1: correspondence between the IP code and water-tests of IEC 68-2-18 Appendix 2: equivalence between the old third numerals of the french IP code and IK code Appendix 3: bibliography
The protection of persons against direct contacts is required by certain regulations (in France the Decret of the 14th of November 1988), and specified by electrical installation standards, such as IEC 364. On the other hand, types of external stresses likely to influence the behaviour of equipment are described in IEC 721-2. The protection against some of these influences is often specified in product standards. Therefore, to explain the degrees of protection that enclosures can provide, and how standards codify them, constitutes an essential information for all, prescribers, installers, operators and inspecting organizations, the purpose of this Cahier Technique. It replaces Cahier Technique No 6 written in 1982, made obsolete by the revision of IEC 529.
1. introduction
It is not sufficient that a piece of equipment fulfils the functional requirements assigned to it. It also has to be protected against possible adverse external influences, and likewise to ensure that it is not harmful to the user or to the environment. Different means can be used alone or combined to comply with this last requirement. All of them derive from one of the following methods: s placing out of reach vertically or horizontally, for instance by means of an obstacle, s total solid insulation as used particularly for cables, but which is not easily applicable when moving parts are involved, s putting into an enclosure, which is the object of this Cahier Technique. This last method has the advantage of providing an easy solution to the other requirement, i.e. the protection of equipment against certain influences such as:
s the
ingress of foreign bodies which could disturb the mechanical or electrical operation. They comprise not only sand and dust but also small animals and flying or creeping insects, s water and other liquids which could alter the insulation and generate degradation, s mechanical impacts which might deform or break brittle parts, s corrosive gas from the environment, s radiated electromagnetic fields, s various radiations including light. Constituting a supporting structure, the enclosure also allows for the building of assemblies of complementary and coordinated apparatus. Therefore it is the most common method of protection. It is used for electronic or information technology equipment as well as for domestic appliances or for high or low voltage equipment, or for rotating machines. The enclosure may be built into the equipment or manufactured separately and sold empty to an
assembly-maker. It can be made of various materials: metal or synthetic, insulating or conductive. In order to facilitate the relationships between manufacturers, users and legislators, standards define terminology, characteristics and means of checking a product, a service or an installation. As far as protection provided by enclosures is concerned, it is the aim of Publication 529 of the IEC and of prEN 50102 of CENELEC (see corresponding French standards in appendix 3). It must be noted at this point that standards define the protections that enclosures can provide, but not the characteristics of the enclosures themselves. These so-called horizontal standards apply only when referred to by the relevant product standard. By extension, degrees of protection by enclosures are also used to characterize protection provided by barriers.
2. general
definitions
To understand the actual importance of standards, it is absolutely necessary to refer to the definitions of the vocabulary used, definitions which are themselves standardized for a given field. Some terms which are used when protection of electrical equipment by enclosure is referred to are therefore given hereafter. Given the extent of international trading connections, the corresponding French terms are also provided. The reference which is sometimes stated just after, is the index of the term in the International Electrotechnical Vocabulary (IEV). s enclosure (enveloppe) IEV 826-03-12 A part providing protection of equipment against certain external influences and, in any direction, protection against direct contact. IEC 529 adds the following note: This definition taken from the existing IEV needs the following explanations under the scope of this standard: 1) enclosures provide protection of persons or livestock against access to hazardous parts, 2) barriers, shapes of openings or any other means - whether attached to the enclosure or formed by the enclosed equipment - suitable to prevent or limit the penetration of the specified test probes are considered as a part of the enclosure, except when they can be removed without the use of a key or tool. Figures 1 and 2 display this definition. So enclosures provide protection against direct contacts. The way they are made can also participate into the protection against indirect contacts when the continuity of conducting parts is secured. s degree of protection (degr de protection) The extent of protection provided by an enclosure against access to hazardous parts, against ingress of foreign objects and/or against ingress of water and to give additional information in connection with such protection.
,,,,,, , ,,,,, ,
2
fig. 1: a mask in front of an opening prevents the test probe of 1 mm diameter () from penetrating and allows a degree of protection of IP3 XD (according to IEC 529).
, ,,,,,
2,5 < 12,5
s hazardous part (partie dangereuse) A part that is hazardous to approach or touch. It may refer to any kind of hazard: electrical (high or low voltage), mechanical, thermal or other. s adequate clearance for protection against access to hazardous part (distance suffisante pour la protection contre laccs aux parties dangereuses) A distance to prevent contact or approach of an access probe to a hazardous part. Distance is related to the internal voltage of the equipment. s access probe (calibre daccessibilit) A test probe simulating in a conventional manner a part of a person or a
,, ,,,,
1
fig. 2: a fixed panel inside the enclosure prevents the test probe of 2.5 mm and of 1 mm diameter () from penetrating; therefore it is an IP2 XD (according to IEC 529).
tool, or the like, held by a person to verify adequate clearance from hazardous parts. The conventional approach of the representation of a part of the human body is to be noted here. The jointedtest finger does not pretend to represent the longest nor the thinnest phalanges; it is only the name of a test probe showing, typically, a finger. When a more selective protection is looked for, a higher degree should be used. This type of probe is used to check protection of persons. The criterion to pass a test with such a probe is that, if it penetrates partially, the adequate distance is kept.
s object-probe
(calibre-objet) A test probe simulating a solid foreign object to verify the possibility of ingress into an enclosure. The test is satisfactory if the full diameter of the probe does not pass through any opening, as the bore probe.
An assembly enclosed on all sides, top and bottom in such a manner as to provide a specified degree of protection.
uses of a code system

According to the dictionary, a code is A set of rules on any subject, or a set of letters or figures or word groups with arbitrary meanings for brevity or secrecy. Apart from the purpose of secrecy, both meanings apply to code systems dealt with in this Cahier Technique. A code system allows the representation by means of an alphanumerical symbol, of properties according to their definitions and to the way they are tested.
s swichgear
and controlgear (appareillage) IEV 441-11-01 A general term covering switching devices and their combination with associated control, measuring, protective and regulating equipment, also assemblies of such devices and equipment with associated interconnections, accessories, enclosures and supporting structures. s enclosed assembly (ensemble sous enveloppe) IEV 441-12-02
It can be used in different directions: s to specify required levels. It is the case of regulations relevant to the protection of the public or of workers. It is also the case of some installation standards for a particular environment; it can also be the case of a particular users specification. s for a manufacturer, to describe the properties of his equipment. Thus some product standards give a set of degrees of protection to choose in one direction or in an other. As far as the protection provided by enclosures of electrical switchgear is concerned, nowadays the international standardization defines two codes: the IP code and the IK code.
3. the IP code
general
The IP code is described in Publication 529 of the International Electrotechnical Commission. IP means International Protection. This code allows for the description of the degrees of protection provided by enclosures against access to hazardous parts, ingress of solid foreign bodies and against harmful effects of water, by means of the numerals and letters described hereunder. This standardized code is intended to be used in product standards. It can also be used to describe an empty enclosure, but then some difficulties can appear as to the application: s where to locate harmful parts to keep them at adequate distance? s where may water or dust deposit without disturbing the correct operation of the equipment? In fact, degrees of protection will only be required when the enclosure is in service with equipment inside. The assembler is therefore responsible for the final product complying to its standard. But the manufacturer of the enclosure must state in the documents where equipment has to be installed inside the enclosure to maintain the degrees of protection he has assessed. Eventually, people in charge of the installation who will connect the equipment (cable pass), fix it and in some cases adapt auxiliaries (pushbuttons, meters,...) must ensure that the specified degree of protection is maintained. Letters IP of the code are followed by two independent numerals and sometimes by letters. When the degree of protection corresponding to one of the numerals is not stated (be it unnecessary or unknown) it is replaced by an X.
first numeral
As a result of decisions made for previous editions of the standard, decisions which cannot to be reconsidered, the first numeral indicates simultaneously: s protection of persons against access to harmful parts and s protection of equipment against ingress of foreign bodies. To check compliance with the first numeral, two probes must therefore be used (an access-probe and an objectprobe) with the application forces specified in the standard, or the same probe is used with two acceptance criteria. The various degrees correspond to the following meanings: IP 1X: it can be a wire-mesh or an enclosure, the largest opening of which does not allows a ball of 50 mm diameter to ingress. This corresponds approximately to the ingress of a hand (see fig. 3). IP 2X: the protective wire-mesh has smaller holes and the diameter of the objet-probe is 12.5 mm. In addition, the jointed test-finger must stay at adequate distance from harmful parts. IP 3X: the enclosure must not allow ingress of foreign bodies 2.5 mm of
diameter. The test is performed with a steel wire with edges free from burrs, because the use of a 2.5 mm diameter ball would not be convenient. IP 4X: as for the previous degree but with 1 mm instead of 2.5. IP 5X and IP 6X: these two degrees correspond to protection against ingress of dust. IP 5X allows penetration of some dust in places where it is not harmful. IP 6X accepts no ingress of dust at all. The test is performed in a test room where talcum powder is sustained floating by means of an air flow. In addition, the enclosure is depressurised internally except if the relevant equipment standard specifies that it be of category 2: that is, the normal operation of the enclosed assembly cannot generate significant internal pressure reduction. Although the test is performed with talcum powder, the effects that might give any other type of dust must be taken into account here. Dust-test of the IP code is currently to be incorporated into IEC 68-2 as test La2. The indication provided by the first numeral implies that the equipment complies with all lower degrees.
,, ,,,,, ,,,,
12 sphre 12,5
fig. 3: the ball does not enter, but the finger-tip touches the live part; therefore it is only IP 1X (according to IEC 529).
second numeral
The second characteristic numeral of the IP code indicates the degree of protection against harmful effects of water penetration. It is specified that the tests be performed with fresh water (see fig. 4) with no wetting agents. The interpretation of tests for this numeral may be difficult since water penetration into the enclosure is permitted, provided it does not generate harmful effects. The various degrees of the second numeral correspond to the following situations: IP X1: this first degree correspond to the protection against vertical water drops to which indoor equipment can be exposed due to leaks or condensation-drops from the ceiling of the room or on tubes passing above the enclosure. IP X2: this degree corresponds also to water drops, but with a larger flow and with an angle up to 15. It is for instance the case of equipment on ships. IP X3: this degree corresponds to the protection against rain. The maximum spray angle is 60 from vertical. The base of the enclosure may be open. The test may be performed by means of a oscillating arc-tube equipped with nozzles (over 60 from each side of vertical) or of a mobile spray with a mask limiting the incidence of the jets. In both cases, the water flow is specified. IP X4: the water flow of each nozzle is the same for this degree as for the previous one but there are nozzles over 180; in addition, the tube oscillates over 180 so that water is sprayed in every direction.This test simulates heavy rain and splashes. IP X5 and IP X6: these degrees are tested by water jet hose simulating water-jets, heavy sea, etc. Test conditions are more severe for degree 6 than for degree 5: a larger diameter of the nozzle and water flow. IP X7 and IP X8: no longer correspond to water projections but to transient or permanent immersions.
IP X3 : protg contre l'eau en pluie.
,,, ,,,
IP X1 : protg contre les chutes verticales de gouttes d'eau.
,,, ,,,
IP X2 : protg contre les chutes de gouttes d'eau jusqu' 15 de la verticale.
IP X4 : protg contre les projections d'eau.
IP X5 : protg contre les jets d'eau.
IP X6 : protg contre les paquets de mer et projections assimilable.
,,, ,,, ,,,
IP X7 : protg contre les effets d'une immersion temporaire.
IP X8 : matriel submersible (essais sur accord particulier).
fig. 4: test means provided by the standard to assess the second numeral of the IP code.
Therefore enclosures complying with these degrees must have double marking if they also comply with a lower degree, for instance IP X5/X7 (a bucket immersed upside down is IP X8, but not IP X4). Water-tests of the IP code have been recently incorporated into IEC 68-2-18 with the following correspondence (see appendix 1).
additional letter
In some cases, the protection provided by an enclosure against access to harmful parts is better than indicated by the first numeral (which also indicates the protection against ingress of foreign bodies). For instance it is frequently the case of an opening of the enclosure blinded by a staggered joint or a sheetbent. This protection can be characterized by an additional letter added after the two numerals. It allows openings useful for thermal dissipation when keeping the degree of protection required for the protection of persons. It has one of the following meanings: IP XXA has no practical application since the test for the letter A is the same as for the first numeral 1 (see fig. 3). IP XXB means that foreign bodies of diameter larger than 12.5 mm can ingress into the enclosure, but that the test-finger does not penetrate more than 80 mm, i.e. not beyond its 50 x 20 mm
guard and stays at an adequate distance from harmful parts (see fig. 5). IP XXC allows penetration of foreign bodies of diameter larger than 2.5 mm, but a straight steel wire of this diameter and 100 mm long stays at an adequate distance from harmful parts. IP XXD the situation is similar to the previous degree, but for a diameter of 1 mm. The additional letter is also used when only the protection of persons is aimed at.
other characters in order to add particular information. For electrical switchgear, only supplementary letter W is used. It indicates a protection against bad weather checked by other means than those specified for the second characteristic numeral, which are difficult to apply to large equipment. For instance, the spraying equipment designed for dielectric wet tests is used to check the weather proofing of enclosed high voltage switchgear. Letters M and S are used for rotating machines to indicate that they have been tested with the rotor in Movement or Stationary.
supplementary letter
The IP code also comprises some supplementary letters following the
sphre 12,5
,,,,,, ,,,,,
15
fig. 5: the ball penetrates but the test finger stays away from the live-part; it is therefore IP 1XB (according to IEC 529).
4. the IK code
introduction
Some countries had felt the need also to codify the protection provided by enclosures against mechanical impacts. To do that, they added a third characteristic numeral to the IP code (the case of Belgium, Spain, France and Portugal). But now that IEC 529 has been adopted as a European standard, no countries may have a different IP code. Since IEC has so far refused to add this third numeral to the IP code, the only possibility keeping a code system in this area has been to create a different code. This is the object of the European draft standard prEN 50102: IK code. Since the third numeral of the different countries might have different meanings, and to introduce supplementary severity levels in order to cover the main needs of Product Committees, the degrees of the IK code have meanings different from the old third numerals (see appendix 2). In order to limit confusion, each new degree is indicated by a number of two numerals.
degrees of protection
The degrees of protection correspond to impact energy levels, given in Joules. An impact, action of a hammer directly applied to the equipment must be distinguished from a shock transmitted by the support and expressed in vibration terms i.e. in frequency and acceleration. Figure 6 shows table 1 of the standard completed with information relevant to test means. In fact, the degrees of protection against mechanical impacts can be checked with different types of hammers: a pendulum hammer, a spring hammer or a free fall hammer (see fig. 7). Each has a particular scope in energy and in direction of application. In order to get similar severity with impacts of the same energy, certain characteristics of the test means have to be complied with: the radius and the hardness of the striking element. The product standard must specify on which parts the blows are to be applied and what the acceptance criteria are.
IK code energy Joules radius mm (1) IK 01 0.15 10 IK 02 0.2 10 P IK 03 0.35 10 P IK 04 0.5 10 P IK 05 0.7 10 P IK 06 1 10 P IK 07 2 25 S IK 08 5 25 S IK 09 10 50 S IK 10 20 50 S
material (1) P steel = S (2) polyamide = P (3) pendulum hammer spring hammer vertical hammer yes yes no
yes yes no
yes yes no
yes yes no
yes yes no
yes yes no
yes yes yes
yes no yes
yes no yes
yes no yes
(1) of the striking element (2) Fe 490-2 according to ISO 1052, of hardness 50 HR to 58 HR to ISO 6508 (3) of hardness HR 100 according to ISO 2039/2
fig. 6: test requirements for the various IK degrees.

(a) pice de frappe
(b)
,,,,,,,,, ,,,,,,,,, , ,,,,,,,,, ,,,,,,,,, ,,,,,,,,, ,,,,,,,,, ,,,,,,,,, ,

cne de dtente bouton d'armement pivot du pendule support
,,, ,,, ,,,

support de fixation
mcanisme d'accrochage
fig. 7: spring-hammer (a) (according to IEC 68-2-63 and pendulum hammer (b) (according to IEC 68-2-62). Note that a calibration device not represented here is required.
,, ,,
hauteur de chute
spcimen
5. application to the design of electrical equipment
These notions of protection have a large influence on the design of equipment because protection has to be assessed not only by the external enclosure, but also by internal enclosures or parts of internal enclosures (partitions, shutters, etc). Therefore the degree of protection of persons has to be defined also for internal parts which can initiate a direct contact during an operation, for instance when withdrawing a circuitbreaker. In addition, even if an enclosure provides the required degree of protection, it is also necessary that it
cannot be removed partially or totally. The point does not concern pieces of equipment such as motors, transformers, etc, but it is of primary importance for some compartments of assemblies which have to be accessible during servicing of the equipment. Two kinds of compartments are considered in this case: s those which are opened only rarely (bus bars) and for which bolted covers can be considered as satisfactory. Opening them being not a simple operation, it is supposed that adequate
precautions dictated by safety requirements will be taken.

s those
which may have to be opened during normal operation of the equipment. They are generally closed by doors which can be locked or blocked by an additional control system which completes the protection provided by the enclosure. During all these service and maintenance operations the electrical continuity of the enclosure must not be interrupted whatever the position of the equipment is.
6 . conclusion
To be satisfactory, any piece of equipment has to comply with its relevent product standard. But this standard uses horizontal ones, particularly those relevant to the degree of protection.
The manufacturer as well as the user should therefore refer to the corresponding standards after reading this Cahier Technique (see bibliography).
appendix 1: correspondence between water-tests of the IP code and water-tests of IEC 68-2-18
IP code water-tests of IEC 68-2-18
IP X1 & 2 test Ra2
IP X3 & 4 test Rb2
IP X5 & 6 test Rb3
IP X7 test Rc1
appendix 2: equivalence between the old third numerals of the french IP code and IK code
old 3d numerals of the IP code of NF C 20-010 (1986) IK code
IP XX1
IP XX3
IP XX5
IP XX7
IP XX9
IK 02
IK 04
IK 07
IK 08
IK 10
Documents describing degrees of protection s IEC 529 (1989-11): Degrees of protection provided by enclosures (IP code) s European application: EN 60529, s French application: NF C 20-010 (1992). s prEN 50102 (1993): Degrees of protection provided by enclosures for electrical equipment against external mechanical impacts (IK code). Draft. s French application: prNF C 20-015 (1993). Draft.
Documents specifying degrees of protection for switchgear s Decret of the 14th of November 1988 from the French government: Protection of workers against electrical currents. s HN 60-E-01 (1974) EDF: Specification for general rules relevant to plastic materials used in electrical equipment for lowvoltage networks and connecting points. s IEC 298 (1990): A.C.metal-enclosed switchgear and controlgear for rated voltages above 1 kV and up to and including 52 kV.
European application: HD 187, French application: NF C 64-200. s IEC 439-1 (1985): Requirements for type-tested and partially type-tested (low-voltage) assemblies. s European application: EN 60439, s French application: NF C 63-421. s IEC 947-1 (1988): Low-voltage switchgear and controlgear. s European application: EN 60947, s French application: NF C 63-001.
s s
Ral. : Illustration Technique Lyon DTE 01/93 - 2500 - Imp. Lostic
n 167
photographie
energy-based discrimination for low-voltage protective devices
Marc Serpinet Joined Merlin Gerin in 1972 and worked until 1975 in the low-voltage equipment design offices, in charge of designing electrical cabinets for various installation layouts. Since 1975, he has managed research and development testing for low-voltage circuit-breakers. Graduated in 1981 from the ENSIEG engineering school in Grenoble. In 1991, after managing a Compact circuit-breaker project from the preliminary studies on through to production, he was appointed head of the electromechanical design office in charge of anticipating future developments.
Robert Morel Graduated with an engineering degree from ENSMM in Besanon and joined Merlin Gerin in 1971. Specialised in designing low-voltage switchgear and participated in designing the Sellim system. In 1980, took over development of Compact circuit-breakers and Interpact switches. In 1985, became manager of the Low-Voltage Current Interruption design office in the Low-voltage Power Components division.
E/CT 167 first issued March 1994
glossary
Eb Energy let through by the protective device during breaking. This energy is characterised by
2 2 ib . dt I . t b .
Trip-unit rating Corresponds to the maximum current setting of the trip unit. High-set instantaneous release (HIN) Instantaneous release used to limit thermal stress during a short-circuit. Instantaneous release (INS) Release without an intentional time delay system. It trips at a low multiple of In to ensure short-circuit protection. Long-time release (LT) Release with an intentional time delay system (several seconds) for overload protection. Short-time release (ST) Release with an intentional time delay system ranging from ten to several hundred milliseconds. If the time delay is reduced as Ip increases, the system is referred to as dependent short-time (DST). Current-limiting circuit-breaker Circuit-breaker which, when interrupting a short-circuit current, limits the current to a
value considerably less than the prospective current (Ip). Selective circuit-breaker Circuit-breaker with an intentional time delay system (time discrimination). Partial discrimination Discrimination is said to be partial when it is ensured only up to a certain level of the prospective current (Ip). Total discrimination Discrimination is said to be total when it is ensured for all values of the prospective fault current. Circuit-breaker rating Corresponds to the models of the range (ex. 160 A, 250 A, 630 A, 800 A, etc.).
ib Limited short-circuit current actually flowing through the circuit-breaker (the break current is less than Ip).
Ip
Prospective short-circuit current that would develop in the absence of protective devices (rms value).
Ir
Corresponds to the overload protection setting. tb The actual breaking time (arc extinction). UT Electronic processing unit. Actuator Device capable of producing a mechanical action.
energy-based discrimination for low-voltage protective devices
summary
1. Discrimination in low-voltage protective devices Definition Enhanced safety and availability Discrimination zones Current discrimination Time discrimination SELLIM discrimination Zone selective interlocking Combining the different types of discrimination 3. Energy-based discrimination Choice of operating curves Characterisation of a Compact NS circuit-breaker p. 4 p. 4 p. 5 p. 6 p. 6 p. 7 p. 7 p. 7 p. 8 p. 9
2. Discrimination techniques for short-circuits
The purpose of this Cahier Technique publication is to present the new energy-based discrimination technique that ensures tripping discrimination between protective devices during a short-circuit. Both simpler and more effective than standard discrimination techniques, it has been implemented on the Compact NS range of circuit-breakers used in low-voltage power distribution networks. Discrimination is ensured for all prospective fault currents on the condition that upstream and downstream circuit-breakers have different current ratings (ratio 2.5) with a trip-unit rating ratio 1.6. Following a brief review of standard discrimination techniques, the authors examine the behaviour of circuit-breakers and various trip units from the energy standpoint. They then demonstrate that total discrimination is possible up to the circuit-breaker breaking capacity, over several levels, without using time discrimination techniques.
4. Advantages and implementation of energy-based discrimination
Characterisation of the trip units p. 10 Current-limiting circuit-breaker fitted with a pressure trip system p. 12 Discrimination with Compact NS circuit-breaker Combination with traditional protective devices p. 13 p. 14 p. 15 p. 16
5. Conclusion 6. Appendix - indications concerning breaking with current limiting
1. discrimination in low-voltage protective devices
definition
In an electrical installation, loads are connected to sources via a succession of protection, isolation and control devices. This Cahier Technique publication deals essentially with the protection function using circuitbreakers. In a radial feeder layout (see fig. 1), the purpose of discrimination is to disconnect only the faulty load or feeder from the network and no others, thus ensuring maximum continuity of service. If discrimination studies are not or are incorrectly carried out, an electrical fault may cause several protective devices to trip, thus provoking an interruption in the supply of power to a large part of the network. That constitutes an abnormal loss in the availability of electrical power for those parts of the network where no fault occurred. Several types of overcurrents may be encountered in an installation: s overloads, s short-circuits, s inrush currents, as well as: s earth faults, s transient currents due to voltage dips or momentary loss of supply. To ensure maximum continuity of service, there must be coordination between protective devices. Note that voltage dips may provoke unnecessary opening of circuitbreakers by actuating undervoltage releases.
A CB1 If
B CB2
CB3
CB4
If passes through CB1, CB2, CB3, CB4.
fig. 1: several circuit-breakers are concerned by the fault If.
Examples s a high short-circuit current creates an undervoltage and may trip the undervoltage protective device.
s an insulation fault may be interpreted as a zero-phase sequence fault by an earth-leakage protective device and as an overcurrent by the short-circuit protective device (applicable for TN and IT earthing systems). s a high short-circuit current may trip the earth-leakage protective device (in TT earthing systems) due to local saturation of the summation toroid which creates a false zero-phase sequence current.
enhanced safety and availability

A specific type of protective device exists for each type of fault (overloads, short-circuits, earth faults, undervoltages, etc.). However, a fault may simultaneously bring several types of protective devices into play, either directly or indirectly.
For a given network, discrimination studies and the evaluation of the protection system in general are based on the protective device characteristics published by the manufacturers. Studies begin with an analysis of requirements concerning protective devices needed for each type of fault. The next step is an evaluation of coordination possibilities between the protective devices concerned by a given fault. The result is improved continuity of service while still guaranteeing protection of life and property. The following chapter will deal exclusively with discrimination in the event of overcurrents (overloads and short-circuits).
In this context, the existence of discrimination between circuit-breakers is determined quite simply by whether several circuit-breakers open or not (see fig. 2). Total discrimination Discrimination is said to be total if and only if, among the circuit-breakers potentially concerned by a fault, only the most downstream circuit-breaker trips and remains open, for all fault current values. Partial discrimination Discrimination is said to be partial if the above condition is no longer valid for fault currents exceeding a certain level.
CB1
CB1
CB2
CB2
discrimination zones
Two types of overcurrent faults may be encountered in an electrical distribution network: s overloads, s short-circuits. Overcurrents ranging from 1.1 to 10 times the service current are generally considered as overloads. Overcurrents with higher values are short-circuits that must be cleared as rapidly as possible by instantaneous (INS) or short-time (ST) releases on the circuit-breaker. Discrimination studies are different for each type of fault. Overload zone This zone starts at the ILT operating threshold of the long-time (LT) release. The tripping (or time-current) curve tb = f (Ip) is generally of the inverse-time type to remain below the permissible thermal stress curve of the cables. Using the most common method, the curves of the LT releases concerned by the fault are plotted in a system of log-log coordinates (see fig. 3). For a given overcurrent value, discrimination is ensured during an overload if the non-tripping time of the upstream circuit-breaker CB1 is greater than the maximum breaking time (including the arcing time) of circuit-breaker CB2. Practically speaking, this condition is met if the ratio ILT1/ILT2 is greater than 1.6.
a) CB1 and CB2 open. discrimination is not ensured, i.e. power is not available for the feeders where no fault occurred.
b) CB1 opens, CB2 remains closed. discrimination is ensured, i.e. power is available for the feeders where no fault occurred (continuity of service).
fig. 2: circuit-breaker behaviour during a fault.
Short-circuit zone Discrimination is analysed by comparing the curves of the upstream and downstream circuit-breakers. The techniques that make discrimination possible between two circuit-breakers during a short-circuit are based on combinations of circuitbreakers and/or releases of different types or with different settings designed to ensure that the tripping curves never cross. A number of such techniques exist and are presented in the next chapter.
tb
overload discrimination zone CB2 CB1
overloads I LT2 I LT1
short-circuits Ip I i n s2
fig. 3: overload discrimination.
2. discrimination techniques for short-circuits
Several techniques can be used to ensure discrimination between two circuit-breakers during a short-circuit: s current discrimination, s time discrimination, s SELLIM discrimination, s zone selective interlocking, s energy-based discrimination (see chapters 3 and 4).
s the most common involves installing selective circuit-breakers including an intentional time-delay system. s the second applies only to the last distribution stage and involves using current-limiting circuit-breakers.
tb
CB2 CB1
short-circuit discrimination zone
current discrimination
This type of discrimination is the result of the difference between the thresholds of the instantaneous or ST releases of the successive circuit-breakers. Applied primarily in final distribution systems, it is implemented using rapid circuit-breakers not including an intentional tripping time-delay system. It protects against short-circuits and generally results in only partial discrimination. This form of discrimination is all the more effective when the fault currents are different, depending on where they occur in the network, due to the non-negligible resistance of conductors with small cross-sectional areas (see fig. 4). The discrimination zone increases with the difference between the thresholds of the instantaneous releases on circuit-breakers CB1 and CB2 and with the distance of the fault from CB2 (low Isc < Iins of CB1). The minimum ratio between Iins1 and Iins2 must be 1.5 to take into account threshold accuracies.
Use of selective circuit-breakers The term selective means that: s the circuit-breaker trip unit has a fixed or adjustable time-delay system; s the installation and the circuit-breaker can withstand the fault current for the duration of the intentional time delay (sufficient thermal and electrodynamic withstand capacities). A selective circuit-breaker is generally preceded in the network by another selective circuit-breaker that has a longer intentional time delay. Use of this type of circuit-breaker, corresponding to time discrimination solutions, results in total breaking times greater than 20 ms (one period) in the event of a fault. This figure may run up to a few hundred milliseconds (see fig. 5). When the installation (and perhaps even the circuit-breaker) cannot withstand a high short-circuit current (Isc) for the entire time delay, circuit-breaker CB1 must be equipped with a high-set instantaneous release (HIN). In this case, the discrimination zone is limited to the high-set threshold of the upstream circuit-breake (see fig. 5). Use of current-limiting circuit-breakers and pseudo-time discrimination These circuit-breakers have two main characteristics: s they severely limit short-circuit currents due to fast opening times and high arcing voltages. s the higher the prospective short-circuit current, the faster they act. Use of a current-limiting circuit-breaker downstream thus makes it possible to ensure pseudo-time discrimination between two protection levels. This solution, due to the current-limiting effect and rapid clearing of the fault, limits thermal and electrodynamic stresses in the installation (see fig.6).
Ip
short-circuit discrimination limit Iins2 Iins1
fig. 4: current discrimination.

CB2 : rapid CB1 : selective with 1-2-3 ST settings
installation and/or circuitbreaker thermal withstand capacity limit
tb CB2 CB1
3 2 1 Iins1 IDIN1
Ip
Note: use of a high-set instantaneous release determines the discrimination limit.
fig. 5: time discrimination.
tb
CB2 CB1
CB2 : rapid current limiting CB1 : rapid
time discrimination
To ensure total discrimination, the time-current curves of the two circuit-breakers must never cross, whatever the value of the prospective short-circuit current. For high fault currents, total discrimination is guaranteed if the horizontal sections of the curves to the right of Iins1 are not one on top of another. Several solutions may be implemented to achieve total discrimination:
Ip
Note: use of dependent ST releases (dotted line) on CB1 improves discrimination.
fig. 6: pseudo-time discrimination.
SELLIM discrimination
The SELLIM system offers a number of advantages: s discrimination, s cascading, s reduced stresses in the installation. Upstream from a rapid circuit-breaker CB2, the system requires an ultra current-limiting circuit-breaker CB1 fitted with a special release that does not trip during the first half-wave of the fault current (see fig. 7). A major fault at B is detected by both circuit-breakers. CB2, equipped with an instantaneous release, opens as soon as the fault current exceeds its trip threshold and clears the fault in less than a half-period.
CB1 detects only a single current wave and does not trip. The fault current nonetheless causes contact repulsion, thus limiting the current and the resulting stresses. This limiting of the fault current means that downstream circuit-breakers may have breaking capacities less than the prospective fault current. A fault at A causes repulsion of the contacts of the current-limiting circuit-breaker, thus limiting the stresses produced by the fault current. CB1 opens after the second half-wave of limited current.
zone selective interlocking

This technique requires data transmission between the trip units of the circuit-breakers at the various levels in a radial feeder network. The operating principle is simple (see fig. 8):
s each trip unit that detects a current greater than its tripping threshold sends a logic wait order to the next trip unit upstream. s the trip unit of the circuit-breaker located just upstream of the short-circuit does not receive a wait order and reacts immediately. With this system, fault clearing times remain low at all levels in the network. Zone selective interlocking is a technique used with high-amp selective LV circuit-breakers, though its main application remains HV industrial networks. For further information, refer to Cahier Technique Publication Number 2, entitled Protection of electrical distribution networks by the logic selectivity system.
combining the different types of discrimination

The different types of discrimination presented above are generally combined to ensure the highest degree of availability of electrical power. See figure 9 for an example. Discrimination studies are still carried out using the tables supplied by manufacturers. The tables indicate the discrimination limits for each combination of circuit-breakers and for the various trip units. The costs of non-discrimination and of the various devices selected are taken into account. The energy-based discrimination technique presented in the next chapter constitutes a true innovation that will considerably simplify LV discrimination studies and make possible total discrimination over several levels at minimum cost.
fault at B i1 CB1 u1 i2 A CB2 B i3 u3 26 k u2
CB1
logic relay
logic wait order
2.5 ms fault at A i1 u1 i2 u2
CB2
logic relay
fig. 8: zone selective interlocking.
circuits concerned
type of discrimination zone selective time interlocking SELLIM pseudo-time
type of circuit-breaker selective logic selective rapid/current limiting SELLIM rapid rapid/current limiting
i3 u3 3.5 ms
34 k
head of LV network power distribution
12 ms
fig. 7: SELLIM discrimination. (CB1 - Compact C250 L SB CB2 - Compact C125 N).
final distribution
fig. 9: example of uses for different types of discrimination.
3. energy-based discrimination
Energy-based discrimination is an improved and generalised version of the pseudo-time technique described in the preceding chapter. Discrimination is total if, for all values of Ip, the energy that the downstream circuit-breaker lets through is less than that required to actuate the trip unit of the upstream circuit-breaker. The actual implementation of the energy-based discrimination principle is covered by a Merlin Gerin patent and has been incorporated in the design of the new Compact NS range of circuit-breakers. These rapid and highly current-limiting circuit-breakers meet the rapidly evolving criteria of the market concerning: s increases in installed power, which lead to higher short-circuit currents and correspondingly higher breaking capacities; s the need to limit stresses in the installation as well as the level and duration of fault currents. When reasoning in terms of energy and in order to understand energy-based discrimination, the choice of the means of presenting the operating curves is essential and the subject of the next section. Following that discussion is an analysis of the behaviour in terms of energy for current-limiting circuit-breakers and the various trip units.
choice of operating curves

The tb = f (Ip) curves commonly used for discrimination studies are of no use with current-limiting circuit-breakers when currents exceed 25 In (breaking times are less than 10 ms at a frequency of 50 Hz). Discrimination studies may no longer be carried out on the basis of periodic phenomena, but rather require analysis of transient phenomena. An understanding of energy-based discrimination requires that the following elements be characterised:
s the current wave that the circuit-breaker lets through during breaking, which is characterised by its Joule integral i2 . dt (often expressed as I2 . t) , and corresponds to the breaking energy Eb. s the sensitivity of the releases to the energy corresponding to the current pulse. Thus, quite logically, the above characteristics are represented using I2 . t = f (Ip) curves instead of tb = f (Ip) curves (see fig. 10). It should be noted that standard IEC 947-2 specifies characterisation of circuit-breakers using such curves. For practical reasons the I2 . t = f (Ip) curve is presented in a system of log-log coordinates. For discrimination studies, the limits of the breaking I2 . t value (Eb for circuit-breakers) are between 104 and 107 A2 . s for prospective currents ranging from 1 to 100 kA. Three powers of ten are therefore used for Eb and two for the current. Assuming that the half-wave of the interrupted current is equivalent to half of a sine-wave with the same initial slope as the prospective current, the breaking energy Eb may be expressed as a function of Ip using the following expressions (see the appendix on breaking with current limiting): s for t 10 ms (2) Eb = Ip2 . t s for t < 10 ms (3) Eb = 4 . f2 . Ip2 . tvb3 or 3 b (4) Eb = 4 2 . f . Ip
I2 . t (A2 . s)
Ip(A)
10 In
t(s)
LT
ST1 ST2 INS
On the basis of these equations, the I2 . t versus Ip system can be improved, thus providing further information on the virtual breaking time (tvb) and the limited peak current value (b).
Ip(A)
10 In 15 In 30 In
fig. 10: tb = f (Ip) and I2 . t = f (Ip) curves for a circuit-breaker equipped with an electronic trip unit.
Time lines (see fig. 11) A series of lines representing constant breaking times can be included in the log-log representation for a given frequency. For example, when f = 50 Hz, the line for: s t = 20 ms corresponds to the most common breaking time when Ip is greater than the instantaneous thresholds and less than the contact repulsion threshold: (2) Eb = Ip2 . 2 . 10-2. s t = 10 ms is the breaking time at the current-limiting threshold: (2) Eb = Ip2 . 10-2. s t = 9 to 4 ms which indicate circuit-breaker behaviour when current limiting: (3) Eb = Ip2 . tvb3 . 104. Peak-current lines Similarly, on the basis of equation (4) 3 b Eb = 4 2 . f . Ip a series of lines corresponding to constant, limited peak currents can be included in the representation (see fig. 11). It should be noted that this method of representation makes it possible to characterise circuit-breakers and trip units at 50 Hz for three-pole, two-pole and single-pole faults.
40 ms 20 ms 10 ms
7 ms
( A . s) 7 10
5 ms = 40 kA 2.5 ms = 20 kA = 10 kA
2 . t
10
10
= 5 kA 10
4
Ip (kA)
1 3 5 10 30 50 100
fig.11: graph representing energies.
40 ms 20 ms 10 ms
7 ms
(A . s) 7 10
characterisation of a Compact NS circuit-breaker

Display of the breaking t value The I2 . t values that a circuit-breaker lets through are determined by standardised type tests or by computer models run for a given voltage and frequency. The curves presented here correspond to three-phase faults at 400V/50Hz. The same curves may be generated for other voltages and other frequencies. The indicated values are the maximum values obtained irrespective of the moment at which the fault occurs (upper limits) (see fig. 12).
5 ms
2. t
= 40 kA 2.5 ms 10
6
I2.
A (E) C (B) D
= 20 kA
10
= 10 kA
= 5 kA 10
4
Ip (kA)
1 3 (10 n) 5 10 30 50 100
fig. 12: breaking curve for a current-limiting circuit-breaker.
Curve analysis A great deal of information is available from the graph in figure 12 which corresponds to a 250 A Compact NS circuit-breaker, equipped with a dependent ST (DST) electro-mechanical release with a 10 In threshold. The information characterises the different phases in the breaking behaviour of the current-limiting circuitbreaker depending on the value of the prospective short-circuit current Ip. s point A: when the fault current reaches the trip threshold of the release, the breaking time is typically 50 ms for an INS or DST release. s point B: when the fault current is greater than the trip threshold of the release, the breaking time drops and stablises at 20 ms beginning at 16 In. s point C: when the fault current reaches the contact repulsion level, current limiting starts due to the insertion of an arc voltage in the circuit. Current limiting results in the return to in-phase conditions for the voltage and the current and consequently a drop in fault clearing times from 20 ms to 10 ms as Ip increases. s point D: when the fault current reaches approximately 1.7 times the contact repulsion level, the energy is sufficient to totally open the contacts. At that point, the breaking time is typically 10 ms. This reflex-type breaking is autonomous and a trip unit is required only to confirm the tripped status of the circuit-breaker and avoid untimely reclosing of the contacts. s zone E: when the fault currents runs beyond 2 times the contact-repulsion level, current limiting is increasingly effective and results in increasingly short breaking times. s point F: the end of the curve represents the breaking capacity limit of the circuit-breaker. The curve provides a great deal of information: s tripping threshold (I threshold, point A); s breaking I2 . t value as a function of the prospective current; s contact-repulsion level (Ir, point C);
breaking capacity (point F); breaking time (tvb) as a function of the prospective current; s limited peak current (b) as a function of the prospective current; s current value above which tvb < 10 ms (beginning of current limiting).
s s
characterisation of the trip units

Trip units are characterised by their response time to a given current (fullwave, half-wave, etc.). By modifying the duration and the peak value of the current, which corresponds to the various currents limited by a circuit-breaker, a number of tests can be run to obtain a series of points which may be plotted on the previously described graph, thus producing the curve characterising a trip unit. Magnetic trip units s instantaneous release Generally made up of a magnetic U and a blade, it ensures short-circuit protection. The response time is
under 50 ms at its operating threshold (between 5 and 10 times the rated current), then drops rapidly to below 10 ms when the current increases (see fig. 13). s high-set release As indicated in the time discrimination section, the role of high-set releases in time discrimination systems is to limit thermal stresses (see fig. 5) in the installation and the circuit-breaker. The high-set release is an instantaneous unit with a threshold of 15 to 50 In. The release may be either electro-mechanical or electronic. s constant time-delay release This is an instantaneous release fitted with a clock-type time delay system intended to make tripping selective with respect to the downstream circuit-breaker. The time delay may range from 10 to 500 ms and is generally set using notched dials. Figure 13 shows the curve (20 ms setting) for a short-time delay.
40 ms 20 ms 10 ms
7 ms
(A . s) 7 10
5 ms
= 40 kA 2.5 ms = 10 kA
2 . t
10
20 m tim s fixe ed d ela y (S T) high set

in
t ) ST den pen lay (D de e ed tim s (INS) = 20 kA stantaneou
10
= 5 kA 10
4
Ip (kA)
1 3 (10 In)
5
10
30
50 100
fig. 13: curves for various magnetic releases.
If the thermal stess (I2 . t) resulting from a long time delay must be limited, the high-set release enters into play (see fig. 13). s dependent time delay release (function of Ip, dependent short-time - DST). The time delay results from the inertia of a mass and is therefore inversely proportional to Ip (see fig. 13). Electronic trip units The instantaneous thresholds in electronic trip units are sensitive to the rms value or the peak current value. For high fault currents, their I2 . t characteristic is theoretically a straight line (b = constant). In fact, the above is true for current pulse durations greater than the response time of the actuating elements of the trip unit (generally 4 ms). Below this value, the inertia of the mechanical elements of the trip unit produces, for high Ip values, a characteristic similar to that of an instantaneous electro-mechanical release. The trip unit must therefore be characterised by its Eb = f (Ip) curve by carrying out tests identical to those for magnetic trip units. These trip units may be of either the instantaneous or time delay type. It is possible to combine several types of electronic trip units, for example: s 10 to 15 In - ST (40 ms), s 15 to 30 In - ST (10 ms), s > 30In - INS. Figure 14 is an illustration of this example. The curves for this combination should be compared with those in figure 10 for the breaking energies of the circuit-breaker. Trip units with arc detection Generally combined with electronic trip units, arc detectors may be used to provide protection for: s a cubicle: if an arc occurs in a cubicle, the detector orders opening of the incoming circuit-breaker, s a selective circuit-breaker: positioned in the breaking unit, the detector provokes via the electronic trip unit the instantaneous tripping of the circuit-breaker. The circuit-breaker is thus self-protected and can therefore be used up to the limit of its electrodynamic withstand capacity.
Pressure trip system The pressure that develops in the breaking unit of a circuit-breaker is a result of the energy produced by the arc. Above a certain fault current level, this pressure may be used for detection and tripping. This is possible by directing the expanding gases in the unit toward a piston that trips the circuit-breaker (see fig. 15). Pressure trip systems may be used to: s ensure self-protection of a selective circuit-breaker (similar to the arc detector),
s improve breaking and operating reliability of a rapid current-limiting circuit-breaker. If each circuit-breaker is fitted with a correctly designed pressure trip system, discrimination is ensured between circuit-breakers with different ratings for all overcurrents greater than 20 In. It is this energy-based trip system (constant I2 . t value) that makes possible the energy-based discrimination technique employed in the Compact NS current-limiting circuit-breakers.
40 ms 20 ms 10 ms
7 ms
(A . s) 7 10
ela
y(
ST
5 ms
= 40 kA 2.5 ms = 20 kA = 10 kA
2 . t
ed tim 40 10 ms tim
10
ms
ed
(INS)
10
ela
= 5 kA 10
4
Ip (kA)
1 3 (10 In) 5 10 30 50 100
fig. 14: examples of combinations of electronic trip-unit curves.
P1
P2
P3
breaking units
flap valves
fault on phase 1 pressure P1 pressure P2 and P3
piston
fig. 15: operation of the pressure trip system.
4. advantages and implementation of energy-based discrimination

Note that the circuit-breaker trip-unit system, whether electromechanical, electronic or a combination of the two, must offer the following features: s minimum stresses in the installation (limited and I2 . t values), s tripping dependability (safety), s minimum disturbance for correctly functioning circuits (voltage dips), s ease of discrimination studies. Tripping dependability The pressure trip system is a part of the opening mechanism for short-circuits and therefore depends on the current rating of the circuit-breaker. The adjustable DST release, whether electromechanical (see fig. 13) or electronic (see fig. 14), is physically independent of the pressure trip system. Physical independence enhances operating dependability. Voltage dips Voltage dips in an installation can trip undervoltage releases in circuit-breakers and contactors. Unnecessary opening, following a voltage dip caused by a short-circuit, results in reduced continuity of service. Consequently, discrimination studies must also take into account the reactions of undervoltage releases and contactors during voltage dips. A voltage dip in a network lasts until the arc voltage that opposes the source voltage enables interruption of the current. It follows that the voltage dip depends on the type of circuit-breaker and/or trip unit used: s with non-limiting circuit-breakers, the voltage dip is more pronounced
current-limiting circuit-breaker fitted with a pressure trip system

The above requirements may best be met with a pressure trip system, combined with either an electromechanical or electronic trip unit. Figure 16 indicates the energy sensitivity of this combination. The higher the prospective short-circuit current, the shorter the response time, which leads to a virtually constant tripping time at I2 . t. The energy let through by the current-limiting circuit-breaker during a break follows the same curve, but with a slight shift. Stresses in the installation Stresses are limited compared to those observed in current-limiting circuit-breakers of the previous generation. On the basis of the example in figure 16, the figures for a Compact NS 250 A and an Ip of 40 kA are: s 4 ms for the breaking time; s 20 kA for the peak current; s 8 x 105 . A2 . s for the I2 . t.
40 ms 20 ms 10 ms
7 ms
(A . s) 7 10
5 ms = 40 kA 2.5 ms = 20 kA = 10 kA
2 . t
10
pressure trip system

ms 40 0m ST 1 s
10
ST
= 5 kA 10
4
Ip (kA)
1 3 5 10 30 50 100
fig. 16: trip-unit combination curves (electromagnetic and pressure or electronic and pressure).
and can last from 10 to 15 ms (see fig.17), s with current-limiting circuit-breakers, the rapid development of a high arc voltage reduces the voltage dip both in duration and in amplitude (see fig. 17). The voltage dip lasts approximately 5 ms and amounts to 50 % of the rated voltage for currents close to the level required for contact repulsion. The voltage dip amounts to 30 % of the rated voltage for higher currents, but the duration is reduced to 3 to 4 ms. The higher the Isc, the shorter the voltage dip. Any undervoltage releases equipping the circuit-breakers are not affected by such voltage dips.
Discrimination The severely limited energy let through by the circuit-breaker is insufficient to trip the trip unit on the upstream circuit-breaker which remains closed.
discrimination with Compact NS circuit-breakers

Using the energy-based discrimination technique and depending on the ratios between the upstream and downstream circuit-breaker ratings and the trip unit ratings, the Compact NS range (100, 160, 250, 400 and 630 A) offers either partial or total discrimination up to the breaking capacity.
Total discrimination Figure 18 provides an example of total discrimination up to 100 kA over three levels with 100 A, 250 A and 630 A circuit-breakers fitted with various trip units. Using Compact NS circuit-breakers, discrimination is total up to 150 kA. To ensure total discrimination, the energy that a circuit-breaker lets through must be less than that required to trip the upstream circuit-breaker. General rule Discrimination is total and without any restrictions if: s the ratio between the ratings of the successive circuit-breakers is equal to or greater than 2.5, s the ratio between the trip unit ratings is greater than 1.6.
7 ms
i ua
(A . s) 7 10
magnetic 100 A ST 400 ST 500 ST 630
5 ms
ur
40 ms 20 ms 10 ms
2 . t
630 A 630 A 250 A
a) non-limiting circuit-breaker U, I
i t (ms) 10 i ur ua 20
10
ST 160 ST 200 ST 250
breaking = 40 kA non-tripping 2,5 ms 2.5 ms breaking non-tripping = 20 kA
250 A
100 A 10
5
breaking non-tripping
b) highly limiting circuit-breaker U, I
100 A
= 10 kA
ic 5
ua ur 10
10
= 5 kA
4
Ip (kA)
1 3 5 10 30 50 100
t (ms) 20
Note
ST 160, ST 200 and ST 250: electronic trip units for 250 A circuit-breakers. ST 400, ST 500 and ST 630: electronic trip units for 630 A circuit-breakers.
fig. 17: the voltage dip on the network depends on the type of circuit-breaker.
fig.18: total discrimination between 100 A, 160 A and 250 A Compact NS circuit-breakers.
Partial discrimination If the general rule presented above is not respected, discrimination is only partial. Figure 19 indicates that between a 160 A circuit-breaker and a 250 A circuit-breaker fitted with a 250 A trip unit, discrimination is ensured up to a prospective short-circuit current of 4 800 A. This level is higher than that observed, under the same conditions, with standard Compact circuit-breakers. Cascading with the Compact NS Cascading, covered by standard NF C 15-100, enables the upstream circuit-breaker to help the downstream device to break high short-circuit currents. Note that this is detrimental to discrimination (except with the SELLIM system). For the Compact NS, cascading in no way modifies the total and partial discrimination characteristics mentioned above. A Compact NS circuit-breaker can however always assist a downstream circuit-breaker of a different type and with insufficient breaking capacity.
40 ms 20 ms 10 ms
7 ms
(A . s) 7 10
250 A (10 In)
160 A ( 8 In)
5 ms
= 40 kA 2.5 ms breaking 160 A non-tripping 250 A = 20 kA = 10 kA
2 . t
10
discrimination limit 10
5
= 5 kA 10
4
Ip (kA)
1 3 5 10 30 50 100
combination with traditional protective devices

Standard circuit-breakers In an existing installation, the highly limiting Compact NS circuit-breakers may be used for extensions or to replace existing circuit-breakers without reducing the previous discrimination limit. On the contrary, if the new circuit-breaker is installed: s downstream, its current-limiting capacity can only improve the discrimination level, possibly to the point of making discrimination total (see fig. 20), s upstream, the discrimination level is at least equal to the previous level and the high current-limiting capacity of the Compact NS can be used to reinforce cascading. Fuses The I2. t = f (Ip) curves (supplied by manufacturers) concern: s the energy required to blow the fuse (prearcing), s the energy that flows through the fuse during the break. To ensure discrimination between an upstream circuit-breaker and a fuse, the circuit-breaker trip unit must not react to the sum of these two energies.
fig. 19: partial discrimination between two Compact NS circuit-breakers, 160 and 250 A.
40 ms 20 ms 10 ms
7 ms
(A . s) 7 10
magnetic 630 A
H C 250 N C 250 L = 40 kA 2.5 ms non-tripping
10
NS 250
5 ms
2 . t
= 20 kA
10
= 10 kA
= 5 kA 10
4
Ip (kA)
1 3 5 10 30 50 100
fig. 20: replacement of a Compact C250 N, H or L by a Compact NS 250 provides improved discrimination. In this example, discrimination becomes total.
5. conclusion
Using a few simple rules, highly limiting circuit-breakers that operate faster for higher prospective short-circuit currents can be implemented to provide total discrimination over several network levels. They may also implement time-discrimination techniques. This is a major technical innovation that can be used to: s considerably simplify discrimination studies, s minimize electrodynamic forces, thermal stresses and voltage dips resulting from short-circuits. This new discrimination technique, referred to as energy discrimination and based on total control over the energy let through by the circuitbreakers during breaking and on the sensitivity of the trip units to the same energy, is an important contribution to improving the availability of electrical power.
6. appendix - indications concerning breaking with current limiting

Figure 21 shows the currents and voltages for a half-period current-limiting phenomenon. The short-circuit current (ib) obeys the following relationship:
Ur - Ua = r .i + L . di di L. dt dt
ip
Ua di/dto Ur
s at the beginning of the short-circuit, Ua is zero, ib and ip are equal and have identical slopes. s when Ua is equal to the network voltage Ur, ib attains its maximum value (b) because its derivative is equal to zero. s when Ua is greater than Ur, ib declines to zero at tb. The interrupted current wave is equivalent to a sinusoidal half wave with a period equal to twice the virtual breaking time (tvb). With the above information, it is easy to determine the energy dissipated in the impedances of the concerned circuit. Expressed in other terms, the formula for this energy, called the breaking energy, is:
b ir
ib
t 0 tr ta
^
tvb t b
T/2
Ua: arcing voltage Ur: network voltage ip: prospective current ib: break current (limited) b: maximum break current ir: contact repulsion current
^ time corresponding to b t: ta: time at which the arc appears tb: breaking time tr: time at which contact repulsion occurs tvb: virtual breaking time : angular frequency of the interrupted wave
fig. 21: breaking with current limitation.

s tvb < 10 ms The circuit-breaker limits the fault current. ib and ip have the same initial slope, therefore: di = . Ip. 2 = ' . b dt
we obtain:
2.Eb = (t vb . 2. f . Ip . 2 )2 t vb
Eb =
t vb
2 ib . dt
where ib is a sinusoidal function:

1 2 Eb = . b . t vb (1). 2 It is useful to express Eb as a function of Ip and the duration (tvb) of the break: s tvb 10 ms For such a duration, the fault current is low, the circuit-breaker contacts do not repel each other and there is therefore no arcing voltage:
ib = ip and b =
Eb = Ip 2 . t (2).
hence:
Eb = 4. f 2 . Ip 2 . t vb 3 (3).
Again on the basis of (1), but with b in mind:
2 b we obtain:
where t vb .
' =
. Ip. 2 = . b
t vb
hence: b = t vb . 2. f . Ip. 2 or t vb =
t vb =
2.Eb
b 2. f . Ip. 2
2 . Ip;
and formula 1 may be expressed as:
2. f . Ip. 2 If we express equation (1) as: 2.Eb 2 b = t vb
Eb =
3 b
4 2 . f . Ip
(4)
Formulas (3) and (4) can be used to plot the time and peak current curves.
Ral. : ERI Ingnierie - Lyon DTE - 03-94 - 2 500 - Imprimeur : Lostic
n 170
photographie photographie
from current transformers to hybrid sensors, in HV
Christian Teyssandier In 1962 he joined Merlin Gerin, going on to graduate from the Institut National Polytechnique de Grenoble (INPG) in 1967. Working first on magnetism problems for linear motors until 1974, he then held a variety of positions of technical responsibility in the HV, MV and LV power capacitor field. In 1983 he was appointed project manager for the MV instrument transformer activity. In 1992 he joined the Anticipation section of Merlin Gerin's Medium Voltage Division.
glossary
birefringence: materials with a refraction coefficient depending on propagation direction, polarisation status and light wave frequency, are said to be anisotropic or birefringent. EMC: electromagnetic compatibility: this is the capacity of a device to operate correctly in its electromagnetic environment without generating intolerable disturbances for the equipment placed in this environment (see Cahier Technique n 149) hybrid sensor: current or voltage sensor comprising at least one element sensitive to the value to be measured, coupled to an electronic system sending a secondary signal (current or voltage), which reflects the module and phase of the primary value. Remark Voltage levels enter a variety of classifications according to decrees, standards and other more specific specifications such as those of certain energy distributors, e.g. as regards AC voltages above 1 000 V: c French decree of November 14th 1988 defines two voltage ranges: HVA = 1 kV < U i 50 kV, HVB = U > 50 kV. c the CENELEC (European Electrotechnical Standardisation Committee) states in its circular of July 27th 1992: MV = 1 kV < U i 35 kV, HV = U > 35 kV. c publication IEC 71 specifies the highest voltage ranges for equipment: range A = 1 kV < U < 52 kV, range B = 52 kV i U < 300 kV, range C = U u 300 kV. A revised edition, which keeps only two ranges, is scheduled: range l = 1 kV < U i 245 kV, range ll = U > 245 kV. c the French electrical power distributor, EDF, currently uses the classification of the decree quoted above.
from current transformers to hybrid sensors, in HV
content
1. Introduction 2. General Sensor functions Sensor evolution The values to be measured The various types of current sensors Standardisation 3. Current transformers Use Standards Specification of a CT Special applications EMC behaviour A special risk 4. Rogowski coil current sensor Operation Standards Steady and transient state operation p. 15 Specification of Rogowski coil current sensors EMC behaviour 5. Hybrid sensors Faraday effect optical sensors Hall effect current sensors Zero flux current sensors 6. Comparison table, synthesis 7. Conclusion and future Appendix 1: CT accuracy as in IEC 185 Appendix 2: CT classification as in IEC 44-6 Appendix 3: bibliography Present solutions Future solutions p. 4 p. 4 p. 5 p. 5 p. 6 p. 7 p. 9 p. 9 p. 10 p. 12 p. 13 p. 13 p. 14 p. 15
Technical and technological changes in the protection and control/monitoring equipment used in electrical power distribution networks must be mirrored by parallel changes in their information sources, i.e. in current and voltage sensors. This Cahier Technique mainly deals with current sensors for medium voltage applications. Following a few reminders on information needs and on current transformers, this document then presents the new hybrid sensors, with particular emphasis on those based on a Rogowski coil. It points out both the advantages and disadvantages of these solutions according to their field of application.
p. 16 p. 16 p. 17 p. 19 p. 20 p. 21 p. 22 p. 22 p. 23 p. 24 p. 24
1. introduction
Proper operation and safety of electrical power distribution networks, from the electrical power station right through to the point of use, is ensured by protection and control/monitoring equipment. This equipment requires permanent knowledge of the two fundamental electrical values, namely current l and voltage U. The knowledge of these values has many angles: c type of current (AC or DC);
c voltage level (low voltage - LV, high voltage - HVA and HVB-); c transient evolutions of these values linked to status changes occurring naturally or accidentally in operation of electrical networks. A number of physical phenomena can be used to measure AC currents. These methods result in levels of performance which have a varying degree of compatibility with those
required by the various protection, operation and safety levels sought. Evaluation of sensor performance is vital to ensure their best possible specification for installation on a network. This requires knowledge of how the different types of sensors work.
2. general
sensor functions
Sensors have three main functions: c providing a correct image, as accurate as possible, of the electrical value to be measured; c isolating the power networks from the measuring, protection and control/ monitoring networks; c ensuring either interchangeability between the measuring, protection and control/monitoring units or performing one specific function of these units. Providing a correct and accurate image Based on the two characteristic values of all electrical networks, i.e. current and voltage, the measuring, protection and control/monitoring equipment defines a certain number of parameters such as: cos , threshold overshooting, instantaneous power,... Thus, there are a number of reasons (financial, safety, operating dependability) why the signals sent by the sensors supplying this equipment must be correct and accurate: c correct A sensor is correct if it gives, in specified conditions, a signal x2 which
is identical, in a ratio of a measuring factor, to the one to be measured. x1 = k x2 where k = measuring factor. c accurate A sensor is accurate if the measuring factor k is not dependent on time and usage conditions, provided the latter remain within the specified values: if at t1 x1 = k1 x2 and at t2 x1 = k2 x2 and if k1 k2, then the measuring sensor is not accurate. Further on in this Cahier Technique, examples will be given of sensors which are neither correct nor accurate in certain operating conditions, particularly in transient states, which differ from specified conditions. The winding ratio, or more generally the measuring factor, is used to adapt the signal to be measured to the performances of the device measuring, analysing and processing this signal. The measuring, protection and control/ monitoring equipment, which uses low level input values, cannot accept the disturbances existing on the power
networks to which they are connected by means of the measuring sensors. Separating the power network from the measuring, protection and control/monitoring network Electrical networks are affected by strong electrical and electromagnetic disturbances, particularly severe in high voltage substations. These disturbances are the result of switchgear operations (disconnectors, switches, circuit-breakers and contactors), of the atmospheric discharges to which overhead lines are exposed and of the appearance and disappearance of faults on the operating networks. These disturbances are locally and temporarily superimposed on rated current and voltage values, thus causing disturbances. Transmission of these disturbances to the sensor secondary must be compatible with the insulation and input impedance levels of the measuring, protection and control/monitoring equipment. The level of this transmission depends on a galvanic insulation of varying quality between
the sensor primary and secondary circuits. The transmission factor is a function of: c sensor production technologies; c and the physical phenomena chosen (to perform the measurement), some of which result in a virtually zero factor. Galvanic insulation therefore plays a vital role in EMC (Electromagnetic compatibility) behaviour of the various sensor types. The EMC behaviour of each type of sensor presented in this Cahier Technique will be examined. Behaviour also depends on the associated equipment since sensors are used to adapt the various measuring, protection and control/ monitoring equipment to the power network. Interchangeability and satisfaction of specific functions of the devices supplied by the sensors Measuring, protection and control/ monitoring equipment underwent major changes in the nineteen eighties and is still being developed today. As a result, three different technologies can be found in the same electrical switchboard. c electromechanical This is the oldest technology, using the electromagnetic effects of the electrical values. This means that sensors must supply a very high energy, around 15 VA in normal operation and reaching, in certain specific cases, 3 400 VA on occurrence of a fault on the sensor primary circuit. c analog electronics This more recent technology emerged with the intensive development of semiconductors. This equipment requires far less energy, around 1 VA in normal operation and 225 VA on a fault. A number of protection relays can then be connected onto the same sensor output. c digital electronics This microprocessor based technology, which is the most modern, is still being developed. The energy required is very low, around mVA (0.001 VA) in normal operation and 0.25 VA on a fault. Consequently, sensors may generally only have one very low power output sufficient to supply their associated protection and control/monitoring unit.
In certain cases, however, particularly for differential protections (zone, busbar, transformer,....), sensors must have at least two outputs. These developments have not yet been fully taken into consideration by sensor standards, thus meaning that full use cannot be made of all the advantages offered by microprocessor technologies and sensors developed in recent years. However, measuring, protection and control/monitoring units benefiting from these developments are available on the market, particularly in HVA and HVB. This equipment is associated with specific sensors ensuring the best match possible between the power network and the control and monitoring unit. These sensors can only be used with the measuring, protection and control/monitoring equipment for which they were designed.
Standard IEC 71 defines the highest voltage for equipment, Um, as the maximum value of phase to phase voltage that the network can assume. The power frequency short duration withstand, the impulse voltage withstand, 1.2/50 and, where applicable, switching impulse and broken wave withstand are associated with Um both in this publication and in publications on measuring sensors. This publication divides standard Um values into three ranges: c range A: from 1 kV to under 52 kV, c range B: from 52 kV to under 300 kV, c range C: from 300 kV and over. The remainder of this Cahier Technique chiefly concentrates on sensors for range A networks. Value types Management, monitoring, protection and remote control of all electrical network types require use of the two values, voltage and current, characteristic of all electrical circuits. c voltage The rated value of an electrical HVA network varies from a few hundred volts to a few tens of thousand volts. Fault voltages are generally low and often close to zero. c current Rated value varies from a few amperes to a few thousand amperes. Fault currents can reach several tens of kiloamperes. Only current sensors will be presented in this Cahier Technique, as they account for the majority of sensors detecting electrical values in networks. Their economic influence is greatest in HVA... and must therefore be minimised. This requires: c firstly, knowledge of required performances for the application to deal with and then specify these performances as accuratly as possible; c secondly, knowledge of the operation and evaluation of the performances of the various current sensor types. Using the values For network operation, these characteristic values (voltage and current) are used by a variety of equipment. Knowledge of this equipment makes it possible to specify
sensor evolution
Besides the fact that modern equipment is less energy consuming, sensor evolution in recent years is above all related to three requirements: c reliability Motivated by permanent search for continuity of service and limitation of external effects in the case of incident. c correctness and accuracy Evolution of network equipment, particularly in HVB with the emergence of gas insulated devices, and search for continuity of service, have resulted in development of linear or linearised sensors. These sensors ensure protection and control/monitoring systems can take efficient action in transient states. c cost Proper network operation to increase continuity of service, requires knowledge of network characteristic values in a large number of points. This results in an increasingly large number of sensors being used. Sensor cost is thus an important factor.
the values to be measured

Voltage level This is an important characteristic of the network on which the sensors will be installed. This voltage level determines the dielectric stresses used to determine sensor size.
the characteristics of the current sensor secondary. c measuring instruments v switchboard indicators: not very accurate, they are used to read the value of the measured values. Note that pointer type display is increasingly being replaced by digital display built into the control and monitoring unit. These instruments are ammeters, voltmeters, wattmeters, frequency meters, etc... From class 1.5 to 3, their association with a class 1 sensor is generally sufficient for the function to be performed. v meters and recorders: used either for billing by energy distributors or for breakdown of power consumption between workshops belonging to the same consumer. Their accuracy is generally greater than the indicators described above. For billing purposes, the associated sensor is generally class 0.5. For power consumption breakdown, the accuracy required is lower, and a class 1 sensor is generally sufficient. c protection relays Of modular type, each element has a clearly defined function. These elements are often combined to monitor a part of the electrical network (see fig. 1). The best known elementary functions are the following protections: v phase overcurrent (overload or short-circuit), v zero sequence overcurrent, v current directional (phase and zero sequence), v zone differential, v active reverse power, v reactive reverse power. c protection and control/monitoring units These are incorporated, configurable microprocessor based units (see fig. 2). These units combine, in the cubicles (level 1 of an electrical network technical management system), in a very small space, the various functions required to operate electrical networks, namely: v measurements, v protection devices, v automatic controls, v communications to higher levels 2 (substation) or 3 (operating station).
Note that high currents and voltages must be avoided on the input circuits to this equipment. The function of the current and voltage sensors is to adapt signal level to these input circuits (measurement and/or protection).
operating mode, steady state or transient state, for which it is intended. Note that the operating range of a CT is generally far more extensive than its range of accuracy, since it must allow for short-circuit current. Specific sensor - SS - or Rogowski coil sensor Rogowski defined the principle of this sensor in 1912. It differs from previous designs by the fact that it contains no ferromagnetic materials, thus ensuring a perfect linearity in a wide current range, a linearity unaffected by the various frequencies present on HVA and HVB networks. This sensor type combined with a load impedance Z of high value ( 10 k at 50 Hz) is a voltage source (see fig. 3). Hybrid sensors - HS The definition of a hybrid sensor given in the glossary covers several types of sensors. Only the best known types and those most commonly used in HVA and HVB are described below. c optical current sensor Its sensing element is either an optical fiber or an optical crystal. In both cases, Faraday's principle, discovered in 1845 by the physicist of the same name, is used. c zero flux transformers In this type of HS, the sensing element is a CT in which the flux created by the primary is cancelled for each secondary winding by an auxilary winding (see fig. 4). This cancels the distortion caused by saturation but only in a
the various current sensor types

Current sensors fall into three main families: c transformers, c specific sensors, c hybrid sensors. Current transformer - CT This sensor has two electrical circuits, a primary and a secondary, and a magnetic circuit. It supplies a secondary signal of the same type as the primary value to be measured: it is a current source. Although it is not linear and its operating range is limited by magnetic saturation phenomena, it is at present the most commonly used sensor type in HVA and HVB. A CT may contain several secondaries, each used for a precise function, measurement or protection. c measurement secondary It has a very narrow range of accuracy, generally limited to currents equal to or less than the rated primary current. c protection secondary Its range of accuracy is very large, often one to twenty times rated primary current. The design of this secondary varies considerably according to the
fig. 1: modular type protection relay (Vigirack - Merlin Gerin).
fig. 2: protection and control/monitoring unit (SEPAM 2000 - Merlin Gerin).
limited operating range (current and frequency). c Hall effect current sensor Its sensing element is a Hall cell (see fig. 5), which enables both AC and DC currents to be measured. As it generally uses a magnetic circuit to increase its sensing capacity, it is affected by saturation phenomena, just like a CT.
non-ferromagnetic support support radius
I
mean toroid radius
Z secondary winding (fine wiring)

I = current to be measured,
standardization
Existing standards, both national and international, only cover CTs (current transformers). Work is currently in progress to draw up standards for hybrid sensors (HSs). However, there are as yet no plans to deal with specific sensors. National standards In the EEC, national standards of the different countries are currently being harmonised by CENELEC on the basis of the international standards edited by the IEC. International standards The IEC, via the Technical Committee N 38 draws up the standards for current and voltage sensors. Accuracy On the basis of current CT standards, a certain number of generic terms can be defined for application to all sensor types. In this standard they are referred to as accuracy and are used to specify and evaluate the performances and fields of application of current sensors. c theoretical winding ratio Ratio between the rated rms values of the primary and secondary values, generally referred to as Kn. For CTs it is a number without dimension. For SSs and HSs it is often expressed in Amperes (primary) /Volts (secondary). c error All sensors have imperfections which introduce distorsions in the restoration of the secondary signal. There are three error types in AC current: v ratio error: expressed as a percentage, it is calculated from the difference between real and theoretical winding ratios (see appendix 1); v phase error: normally expressed in angular minutes, it gives, to the
output voltage
u=f(
dI dt
fig. 3: schematic diagram of a SS.
I1 = current to be measured, I2 = secondary circuit current, CM = magnetic circuit, Z = load impedance, generally low, A = current amplifier, ES = secondary winding, SD = zero flux detection winding controlling amplifier A.
CM
I1 I2
ES
SD
fig. 4: schematic diagram of a zero flux CT.
I = current to be measured, CM = magnetic circuit, CH = Hall cell, i = cell supply current, Vh = Hall voltage, proportional to i and I, A = voltage or current amplifier.
Vh
+
A
CM
fig. 5: schematic diagram of a Hall effect current sensor.
nearest (CT) or /2 (SS), the phase displacement existing between the vector of the primary value and the vector of the secondary value (see appendix 1); v composite error: expressed as a percentage of the rms value of the primary current, in steady state it is the rms value of the difference between: - the instantaneous value of the primary current,
- and the product of the rated winding ratio by the instantaneous value of the secondary current. c accuracy class The accuracy class defines for current sensors the maximum error limits (ratio and phase errors) in specified conditions. c accuracy load Expressed in ohms, with a specified power factor, it is the value of the
impedance connected to the sensor secondary terminals on which the accuracy conditions are based. c accuracy output power Expressed in VA, it is the apparent power that the sensor can supply at its accuracy load when the rated primary current flows through it.
3. current transformer
Its principle (summarised in chapter 2) gives it properties which, although advantageous, also present problems in some cases for network operation. A technical description of this current transformer, including its behaviour, is given in detail in Cahier Technique N 164 which deals mainly with the problems relating to operation of electrical installations in steady state.
standards
Steady state CTs having to operate in this state must meet international, European and national standards. c international standards IEC 185, second edition of 1987 currently being revised by CE 38: it concerns class P protection and instrument CTs (see appendix 2). The main purpose of this revision is to remodel clauses on dielectric characteristics and to add a certain number of measures concerning only CTs for HVB such as mechanical forces on connections, radioelectric interference. c European standards These standards, edited by CENELEC, are based on IEC documents. In 1993 there were not yet any EN documents for CTs. Only the harmonisation documents (HD) are under discussion. c National standards The different European standards vary considerably from each other. However, in the future they should be more similar since they will be made to conform to the EN standard on CTs for steady state operation. v France Standard NF C 42-502 (February 1974) complies with IEC documents in virtually all aspects, and will comply with EN standards completely except for the ways of marking secondary terminals. Note : Standard NF C 42-502 states that the secondary terminal connected to the earth is always marked S2. It is also the terminal common to all ratios in the case of CTs with several winding ratios obtained by connections to the secondary winding. Moreover, the same standard states that windings used for measurement must have an odd number and windings used for protection an even number.
use
In HVB, emergence of gas insulated metalclad equipment and the search for permanent dynamic stability of networks containing high power generators make it necessary to consider operation when the network changes status (transient state). Saturation and hysteresis, with no major oversizing of CT magnetic cores, mean that the transient state response of this sensor type is neither correct nor accurate. In general, only at the end of the transient state can a correct response be obtained. This delay, in certain operating and fault cases, is not always compatible with safety of equipment and persons. Sometimes the fault needs to be detected during the first period of the transient state which, in certain network types, can last 200 ms (i.e. 10 periods). Correctness and accuracy are also necessary in transient states: c in HVB, for equipment located near high power stations and on busbars of major substations; c in HVA, near sources, when a high power HVB network is supplied either by a transformer with a high winding ratio (e.g. 220 kV/20 or 36 kV) or by generators with a very high unit power. Current standards make it possible, for CTs and in both operating cases (steady and transient states), in specified conditions, to accurately evaluate the performances of this equipment.
v Great Britain Standard BS 3938 (February 1973) closely resembles the IEC standard and will virtually comply with the EN document. Moreover, it contains the class X windings for protection. This type allows a more accurate specification of protection windings. This class may be included in the EN European standards in the near future. v Italy Standard CEI 1008 (October 1987) (Comitato Electronico Italiano) complies with IEC 185. It will also comply with the EN standard when it is published. v USA Standard ANSI/IEEE C57 13 (1978) differs considerably from IEC 185 and European standards: - accuracy classes and powers are not defined in the same terms, - terminal marking is very different, - and the devices frequently take up more space. Transient state Major energy distributors have long since possessed company specifications for CT transient state operation. These specifications were and still are, met by special manufacturing processes and are the subject of direct agreements between manufacturers and users. c international standards Specifications concerning CTs for protection, for a transient state response, are now stipulated at international level by standard IEC 44-6 (first edition 1992-03). c European and national standards These standards do not yet exist. The European standard, currently being drawn up, will closely resemble IEC 44-6. National standards will be edited by the various organisations on the basis of the EN document.
specification of a CT
The various people involved The user, the network designer, the protection system manufacturer and the CT manufacturer all have different roles to play, at different levels, in CT specification. c the network designer, for operating safety reasons, tends to increase the sizing factors linked to the CT primary: v short time thermal and dynamic current withstand represented by the root mean square (rms) and peak values of the short-circuit currents to be withstood for a period generally of 1 second. v length of assymetrical (unbalanced) state by giving overestimated X/R ratio and time constants. c the relay manufacturer who, also for equipment operating safety reasons, tends to specify high secondary performances. v accuracy level power by overestimating the value of the coupled impedances between relay and CT; v accuracy class by asking the CT not to introduce any additional errors into the unit. Whereas it would perhaps be more advantageous to use slightly more accurate equipment and slightly less accurate CTs to obtain an identical overall accuracy. Example: a class 3 measuring instrument with a class 0.5 CT gives an overall accuracy of class 3.5. In certain cases, it is more advantageous (financially) to take a class 2 instrument and a class 1 CT which give an overall accuracy of class 3. This is particularly true in the case of small primary currents and a high short-circuit current. c the CT manufacturer who tries to reconcile the various requirements as well as his own, in order to meet the demand. The Ith / In ratio (short time thermal withstand 1s / rated primary I) gives a good idea of CT feasibility, irrespective of the secondary performances required. For example: v Ith / In i 100: the CT obtained can be considered standard with normal secondary performances. v 100 < Ith / In i 400: the CT meeting
this specification is a CT whose feasibility is studied individually: its secondary performances are reduced. v Ith / In > 400: this CT is not always feasible. When it is, its secondary performances are very poor. The values to be specified A number of values must be specified to make a CT. Some of these values are standardised (refer to the standards quoted in the standards paragraph). For CTs needing a specified accuracy in the transient state, the reference is either standard IEC 44-6 or company specifications. The following list only concerns CTs operating in steady state. c primary v insulation level defined by three voltages, the highest network voltage (Um), the rated time power frequency withstand voltage and the lightning impulse withstand voltage; v the rated short time thermal current (Ith) and its its duration if it differs from 1 s; v the rated dynamic current (Idy) if its peak value differs from 2.5 Ith; v the rated primary current. Rules for the profession stipulate that the rated current of the network on which a CT is installed, be between 40 and 100% the rated primary current of the CT. c secondary The function, measurement or protection, of the secondary must be specified and leads to varying constraints and specifications. In both cases, the rated secondary current must be specified (1 or 5 A). v measurement The rated output power (in VA), the accuracy class and the maximum safety factor (SF), normally between 5 and 10 and very exceptionally less than 5, must be specified. Note: The safety factor is the ratio between the primary current for which the winding ratio error is greater than or equal to 10%, and the rated primary current. The various accuracy classes and the resulting constraints are given in the standards.
For switchboard devices, class 1 is generally more than enough. This secondary is normally referred to as follows: 10 VA Cl1 FS < 10 safety factor < 10 accuracy class = 1 rated output = 10 VA v protection The protection windings can be specified in two ways: - As in IEC 185 and European standards: by specifying the rated output (in VA), the accuracy class (5P or 10P) and the accuracy limit factor (ALF). The accuracy class gives the maximum composite error allowed on the secondary current for a primary current equal to ALF times the rated primary current (5P = 5%, 10P = 10%). The characteristics and constraints associated with the various accuracy classes are given in the various standards. The windings are then referred to as follows: 10 VA 5P 10 accuracy limit factor = 10 accuracy class = 5P accuracy level power = 10 VA - As in BS 3938: by specifying the value in volts of the knee point voltage (Vk), the maximum winding resistance (Rct) and, if necessary, the maximum exciting current (Io) for the voltage Vk. In this case, the windings are referred to as follows. 0,050 150 R 0,50 resistance Rct in Ohms voltage Vk in Volts current Io in Amperes CT imperfections Magnetic imperfections (saturation, remanence, eddy current and hysteresis losses) generate inaccuracies in the CTs: ratio and phase errors, imperfect linearity,
response depending on previous situations.... Other imperfections are linked to the electromagnetic and electrical environment of the CT. c magnetic phenomena Saturation and hysteresis are the two main causes of disturbance: the output signal of a saturated CT. Output sends a signal which is no longer sinusoidal, and its accuracy can no longer be guaranteed (highly amplified ratio and phase errors). These phenomena appear: v in transient state, for example closing of a circuit on a fault with or without DC component: the state of saturation reached depends on the initial magnetic state of the magnetic circuit (degree of residual induction present); v in short-circuit steady state if the value of this circuit is greater than ALF times the rated primary current; v when the value of the load, to which the CT is connected, is greater than its rated burden, as is the case for very long connections or in case of addition of equipment to the load circuit of a secondary winding; v if network frequency is less than rated frequency: use in 50 Hz of CTs with a rated frequency of 60 Hz causes a 20% induction increase; on the other hand, use in 60 Hz of CTs with a rated frequency of 50 Hz presents no risks. Operation in saturated state must not be allowed to continue, as saturation causes abnormal overheating the CT components: v in the magnetic circuit, since eddy current and hysteresis losses increase; v in the secondary winding, since the currents, although highly deformed, are also very high. c external phenomena v positions of the primary conductor and of the adjacent conductors Their respective geometries and positions may have a considerable effect on the accuracy of instrument transformers as a result of the nonlinearity of the ferromagnetic materials. A typical case is of current transformers installed in a loop (see fig. 6) or installed in staggered form in a busbar (see fig. 7). These two assemblies cause a localised increase
in induction, thus introducing an error. v reclosing cycle After a primary short-circuit current has been broken, return to the remanence value of the induction in the CT magnetic core is not instantaneous.
This induction decreases according to an exponential law with a time constant T2. Depending on the secondary circuit, this constant is generally between one and three seconds. When rapid reclosing occurs, a residual induction is
PR
PA
PP @@ ,, PP @@ ,,
I1
CM
BPA AA cross section
BPR BPA

BPR
The return or adjacent phase conductor (PR) creates a disturbing magnetic field in the magnetic circuit (CM); this field is vectorially added to the one created by the current I1 to be measured of the conductor crossing it normally (PA). This vectorial addition results in increased induction in zone A. This induction increase depends on: - the current flowing in the disturbing conductor, - the distance between the magnetic circuit and this disturbing current. It then results in local saturations which increase the value of the exciting current (Ie), thus introducing errors.
increased induction zone A
fig. 6: diagram showing a transverse CT with a looped primary circuit.
TC2 TC1
TC3
I1
I2 I3
increased induction zones
PP @@ ,, PP @@ ,, P @ , P @ ,
TC1 TC2 TC3
fig. 7: diagram showing three transverse CTs installed in staggered form in a busbar
therefore present in the CT magnetic circuit which is vectorially added to the induction created by the current being formed (see fig. 8). If both inductions have the same sign and if the CT has not been designed to guarantee a given accuracy in the transient state, it is highly likely that the secondary signal delivered by this CT will be totally unlike the primary current flowing through it. c precautions to be taken with CTs v in steady state: - CTs must be designed for their intended purpose. - the sum of the input impedances of all the relays and/or measuring instruments, to which the value of the wiring impedance must be added, must be less than or at the most equal to rated burden. This impedance is obtained by dividing the rated output by the square of the rated secondary current. - installation conditions should not cause high local saturation. It shall be doing away with staggered installations (see fig. 7) v in transient state, for protection secondaries only - in the general case of constant time protections, to allow for all or part of the hysteresis phenomena, it is sufficient to check that the value of the intervention setting current (of the protection) divided by the value of the CT rated secondary current, is less by twice the accuracy limit factor of the secondary in question. - for time dependent protection devices (differential, zero sequence,...), ensure that the CT specification complies with the recommendations of the relay manufacturer - if an accurate response is required during this operating period, CTs must be specified and designed in accordance with the various classes defined in IEC 44-6 (see appendix 1). This standard always leads to considerable oversizing of the CTs. The need for a low remanent flux (case of reclosings) results in use of magnetic circuit with airgap. This is how the linearised CTs are obtained (see TPZ in standard IEC 44-6).
(primary)
Ip
(secondary) Is
(induction)
fig. 8: evolution of currents and induction in an unsaturated CT.
special applications
Measurement of residual currents Protection of persons in LV distribution networks is frequently ensured by monitoring residual current value. This protection, generally provided by a device incorporated in the LV circuitbreaker, is often autonomous: its operating energy is supplied by the CT detecting the residual currents. Stipulated CT performances generally call for use of ferromagnetic materials with excellent relative permeability (r) using nickel, which raises their cost. There is a quick method of sizing this CT type (see appendix 3, bibliography: paper in the RGE review n 4).
Measurement of zero sequence current (Io) This is the current resulting from the vectorial sum of the three phase currents of a three-phase circuit. This sum can be achieved in two ways. c by adding up the secondary currents of three CTs (Nicholson assembly). For this, CTs with the same winding ratio must be used, and the primary and secondary connections must respect the polarities (winding direction) of the various primary and secondary windings (see fig. 9). Two phenomena limit the detection thresholds in this method: v in steady state, the differences in
phase and winding ratio error mean that the vectorial sum is not zero. This results in a false zero sequence current which may not be compatible with the required thresholds. Pairing of CTs (in phase and module) enables practical thresholds to be lowered: v in transient state, saturation and hysteresis of magnetic circuits generate the same fault. Oversizing of the CTs delays the moment of appearance of this phenomenon. These solutions (pairing and oversizing) do not generally allow detection of a current Io less than 6% of the phase currents. c by adding up the fluxes To avoid the inaccuracy of this first method and find a way round its constraints, the Io currents can be measured using a single toroid CT or toroid: the three phase currents I1, I2, I3 of the three-phase network (see fig. 10) flow through its magnetic circuit. With a suitable design (ferromagnetic material, dimension and accuracy load) and taking certain toroid installation precautions (grouping and centering of cables, use of a ferromagnetic sleeve if necessary), this method enables measurement of very low Io current values with a high degree of accuracy (module error around 1% and phase error less than 60 angular minutes): a few hundred mA in HVA and a dozen mA in LV. c fault detection In HVA distribution networks, use of fault detectors facilitates rapid fault location, thus minimising the part of the network not supplied and reducing outage time. There are two possible means of signalling the fault current detected by these devices: v using mechanical or electrical indicator lights placed at points easy to reach by operating staff (case of MV/LV substations in underground rural and urban networks). v by remote transmission to the operating centre for fault detectors placed on remote-controlled switches of public distribution networks.
These fault detectors are supplied by CTs for which no standards exist. Only the CT-fault detector combinations are specified by operators.
I1
P1
P2
EMC behaviour
In HVA, the EMC of the CTs can be said to be satisfactory. In HVB the mandatory electric field distribution shields of varying quality may result in unsatisfactory EMC of CTs. The coupling capacity between the primary and secondary CT windings helps transmit disturbances from the primary to the secondary circuit. The value of this capacity depends on the CT insulating voltage, the secondary characteristics and the insulation technology used. Some company specifications, for voltages Um > 123 kV, give a maximum value for the disturbance transmission factor. This value is measured in a standard test described in the specification. The introduction of this notion into international standards is currently being discussed within CE 38.
I2 S1 P1 P2 S2
S1 P1 P2
S2
I3
S1 Io
S2
fig. 9: connection of three CTs to measure zero sequence current (Nicholson assembly).
I1
I2
a special risk
Opening the secondary circuit of a CT is dangerous. The magnetic induction flux flowing in the magnetic core is the sum of two fluxes of opposite signs, one due to the presence of a primary current and the other to the presence of the secondary current. Cancellation of the latter by opening the secondary circuit considerably increases flux in the core, causing a very high voltage rise at the secondary terminals. Peak or instantaneous voltages of over 5 kV can be reached which may be fatal for persons and cause severe equipment damage. Practical conclusions c on no account must the secondary circuit of a current transformer in operation be opened. c before carrying out interventions on the load of a CT in operation, a very high quality short-circuit must be installed between its secondary terminals.
I3
toroid
metal sleeve
conductor grouping and centering sleeve L > 2 toroid
fig. 10: measuring current using a toroid.
4. Rogowski coil current sensor
The principle of this current sensor was defined by ROGOWSKI in 1912. From 1986 onwards this sensor, referred to as SS (specific sensor), has been developed in industry for HVA networks.
operation
Physical principle Application of Ampere's theorem to a Rogowski coil (see fig. 3) shows that the voltage appearing at the terminals of a load Z of very high value is a function of the current I = i(t). The current I to be measured creates locally, at each turn, an induction b = 0 h, where 0 is the permeability of vacuum, the winding support not being made with a ferromagnetic material, and h the magnetic field corresponding to current I. The flux encompassed by the entire sensor is written as:
Electromagnetic components A SS sensor is made up of five parts (see fig. 11). c a primary winding consisting of a single copper conductor, the cross section of which is determined by: v a primary rated continuous thermal current, v a rated short-time thermal current; c a secondary winding support generally toric and made of a nonferromagnetic material; c a secondary winding support generally toric and made of a nonferromagnetic material; c a setting resistance connected to the secondary winding; c a magnetic shielding protecting the winding from any disturbances linked to the magnetic fields outside the sensor. Dielectric components c dielectric insulation Just like current transformers, the primary and secondary of SS sensors are insulated from each other by a solid dielectric resin in HVA.
c dielectric shield In order to improve the system's EMC behaviour, an earthed dielectric shield is placed between the primary and the secondary winding. Modelling It is useful to design and use a model to study SS operation, in the same way as for CTs. The model proposed below only applies to standard frequencies. For high frequencies (several hundred kHz), the distributed capacitances of the secondary winding must be introduced as well as the various primarysecondary, primary-frame and secondary-frame coupling capacities. Equivalent diagram Two equivalent diagrams can be drawn up: c the first one (see fig. 12) is derived from the CT diagram by the presence of an ideal transformer, where: v L = inductance value of the wiring connecting the sensor to its load M,
= turns r 2 b
If all the N turns have identical cross sections and if their centres are placed on the same circle of diameter R which can be considered as very large compared with their own radius r, the following can be written:
primary winding dielectric shield dielectric insulation
=Nr 0 h
2
and by application of Ampere's theorem:

i(t) 2 R The electromotive force developed in the winding is written as:
secondary winding secondary winding support magnetic shielding
=Nr2 0
setting resistance
Nr 2 d 0 di e(t) = = dt 2R dt If i(t) = 2 sin (t + )

di / dt = 2 cos (t + ) and
fig. 11: cross section of a SS sensor for HVA.
thus
i2 = N1 I 1/N2 1 N1 N2 L1 Zc L Lf Rt Ra
Nr 2 0 2 cos (t + ) e(t) = 2R = K 2 cos (t + )
fig. 12: equivalent CT type diagram of a SS sensor.
v Lf is the winding leakage inductance. The secondary winding of the SS sensors has touching turns with excellent distribution on the support. Its leakage inductance is very low and need not thus be considered. v L1 = magnetising impedance of the equivalent current generator, v Rt = sum of the resistances of the winding and connection, v Ra = setting resistance, v Zc = impedance of load M at the considered frequency. c the second diagram (see fig. 13) includes a voltage generator derived from the theoretical study. This is the diagram that will be used hereafter. In this equivalent diagram: E 0 (t) = K 2 cos (t + )
residual flux. Their perfect linearity means they give at the secondary an almost perfect image of steady and transient states of the primary. Manufacturing tolerances on winding support dimensions and on the value of the number of turns (several thousands) are compensated for by a setting resistance (Ra). Equations The vectorial diagram (see fig. 14) is drawn up from the equivalent diagram in figure 13. This diagram yields the following equations: U(t) = E 0 (t) (RA + Rt + jL ) i 2 (t)
avec i 2 (t) = U(t) / Zc
Error analysis c ratio and phase error The perfect image of the primary current for a SS sensor is the vector E with a laggin phase displacement of /2 with respect to current I1, i.e. in phase with E0 in figure 15. The module of this vector is given by E = K1 I1, where K1 is the constant representing the transformation ratio at a given frequency. In the same way as for all sensors, the secondary signal of the SS sensors has an error. This error is defined as the vector representing the differences of vectors E and U: it is vector (nat) on the diagram in figure 15.
v E0(t) is a source of voltage proportional to the primary current. It has a /2 lagging phase displacement with respect to current i(t) where
2 N r 0 , where 2R v the product K represents the transformation ratio and is expressed in Volt per Ampere (V/A).
Lf
Rt
Ra
Eo
Zc
K=
fig. 13: equivalent diagram with voltage generator of a SS sensor.
standards
No national or international standards currently define this type of sensor. Consequently SS sensors on the market today comply with the IEC 185 standard, except for the parameters concerning the secondary signal supplying very specific protection and control/monitoring units. These units with their microprocessor technology, enable, by simple parameterisation, via a keyboard or display, all the functions (protection, measurement, automation and communication) to be performed adapted to each installation. Note: Today these SS sensors and protection and control/monitoring units, SEPAM, are designed and marketed by one manufacturer only (Merlin Gerin).
E0 L i2

i2 U
(Ra + Rt) i2
fig. 14: vectorial diagram of a SS sensor.
(nat) = natural error (manufacturing) (reg) = error after setting.

1
U
= setting range linked to Ra optimum setting = (reg) minimun
E0 E L i(t)
(reg) A
(nat)
steady and transient state operation

As SS sensors have no magnetic circuit they are not subjected to saturation or
i(t)
Ra i(t)
Rt i(t)
fig. 15: vectorial diagram, with error, of the SS sensor.
At the end of manufacturing it is minimised by setting potentiometer Ra to a value giving a minimum error (reg) vector. This sensor setting is adapted to the current inputs of the protection and control/monitoring units for which it was designed. The maximum value of this error vector was set at 1% of the value of the reference vector for all primary currents between 0.2 and 10 times the rated primary current, and at 5% for a current 200 times the rated primary current. With these maximum errors, SS sensors with the same winding ratio are all completely interchangeable and are virtually identical in accuracy terms (see fig. 16) to CTs of class: v 0.5 for measurement, v 5P for protection. c linearity The SS sensor is linear:
c frequency The signal sent by this sensor type depends on the primary current derivative (see the paragraph on physical principle). Freedom from the effects of frequency is obtained by processing the SS sensor signal by an accuracy integrator amplifier. c primary conductor position Ampere's theorem makes no reference to the relative position of the current (primary conductor) and of the closed contour (secondary winding) to which it is applied. This remark indicates that the sensor is theoretically unaffected by the relative positioning of its components. However, imperfections in making of the secondary winding may make it slightly sensitive. Therefore, when installing sensors with nonintegrated primary (LV), fairly precise relative centering and azimuthing between primary and secondary must be performed. If this precaution is not taken, errors of around 3% may occur. c adjacent conductor position An adjacent conductor, through which the current from another phase or return for a loop (see fig. 6) flows, produces a magnetic field which is vectorially added to the one created by the current to be measured, thus modifying sensor response. SS sensors must be protected against this type of disturbance.
specification of Rogowski coil current sensors

SS sensors and protection and control/ monitoring units are supplied by the same manufacturer, thus making SS specification far simpler for the end-user who does not have to specify, as for the CTs, the characteristics of the secondary (secondary current, accuracy level power, accuracy class and accuracy limit factor). He only has to specify: c the insulation level of the sensor defined as for a CT; c the rated thermal short-circuit current (Ith) and the dynamic current (Idyn) established according to the same rules as for the CTs; c the operating range (rated primary current and rated continuous thermal current). For example, there are four operating ranges (30-300, 160-630, 160-1600, 500-2500 A) for the SS sensors produced by Merlin Gerin.
e(t) = K 2 cos (t + ) This linearity gives it many advantages, the main ones being: v the possibility of reducing winding ratio variety and thus of increasing standardisation potential. Winding ratios are imposed by the electronics dynamics of the control and monitoring unit with which the sensor is associated and by its required discrimination level. v an excellent response in transient state. Absence of saturation, hysteresis and residual flux means these sensors respond perfectly in the transient state. Consequently, with no special precautions, this sensor type is installed on networks where protection devices need to take rapid action during transient states, and in particular on networks with long time constant or containing gas insulated metalclad equipment (with a risk of explosion).
External influences The response of these SS sensors, like that of CTs, may be affected by the environment in certain conditions. c modification of the secondary load of this sensor type causes an error variation. To reduce these variations, as a SS is a voltage source, its purely resistive load must be as high as possible (u 10 k).
EMC behaviour
SS sensors have a small primary secondary link capacity ( 20 pF in HVA). The presence of dielectric and magnetic shields, connected to earth, prevents transmission of conducted (from MV primary network) and radiated disturbances. The SS sensor and protection and control/ monitoring unit thus has good EMC behaviour.
60' 30' 0 A class 1 class 0.5 class CSp A = vector (reg) 0 0.5 % 1 %
fig. 16: comparison of accuracy of CT and SS sensors Point A positions the operation of a SS sensor which meets the accuracy requirements of CTs of class 0.5 and of class 1.
5. hybrid sensors
The output signals of the CTs and SSs are directly used by the protection and control/monitoring units. However, the signals of certain other current sensors must be processed electronically before they can be used by these units: these are the hybrid sensors. Their diagram resembles the one in figure 17. Diagram It may contain up to six elements: c primary sensing element Uses the various effects (optical, electronic or electrical) of the materials subjected to a magnetic field created by the current to be measured. c primary converter Converts the effect used by the sensing element into a signal depending on the primary current and adapted to the transmission system. c transmission system Conveys the signal transmitted by the primary converter over a distance of varying length. c secondary converter Converts this signal, representing the primary current, into an electrical signal which can be used by the protection and control/monitoring units. c primary supply Supplies necessary energy to the sensing element, the primary converter and, if necessary, the transmission system. c secondary supply Supplies energy to the secondary
converter and, if necessary, to the transmission system. In certain sensors, these two supplies, primary and secondary, may be the same. The sensing elements Hybrid sensors have undergone major developments in the course of recent years. A number of magnetic field effects have been used in primary sensing elements, in particular: c optical effects Use of the effects of the magnetic field on the properties of light (optical current sensor). Optics may also be used solely as a transmission system from a primary sensing element of any type. Transmission then takes place by optical fibre. The use of devices obeying the laws of light physics (sensing element and transmission system) gives the sensor its perfect galvanic insulation. This advantage has been made use of in many development programmes, some of which resulted in the Faraday effect current sensor. c electronic effects Influence of a magnetic field on a semiconductor (Hall effect current sensor) and on a ferromagnetic material (resistivity variation used in magneto-resistant current sensors). c electrical effects The flux created in a magnetic circuit by the magnetic field coming from the
current to be measured, is cancelled by a magnetic flux generated and regulated by means of an auxiliary current (zero flux current transformer).
Faraday effect optical sensors

The laws of light physics will be briefly reviewed below to help understanding of the following sections. Reminders c polarisation A phenomenon specific to wave propagation, in particular light waves, characterised by their vibration direction in a given plane, known as the propagation plane, containing the propagation direction. When this plane keeps a direction set in time, the light waves have a linear polarisation. If the plane rotates around the propagation direction at constant speed, polarisation is elliptical or, in a very specific case, circular. c birefringence Certain natural bodies exhibit the phenomenon of birefringence. A flat light passing through them is not propagated at the same speed according to whether its polarisation plane is parallel to one or the other of the two perpendicular directions specific to the birefringent body. Birefringence may be intrinsic (anisotropic materials) or induced by a stress: v mechanical stress or photo-elastic effect, v electrical stress or Kerr or Pockels electro-optical effect, v magnetic stress or Faraday magnetooptical effect. Faraday effect In 1845 Michael Faraday discovered that the polarisation plane of polarised light rotates as it passes through a piece of glass placed in a strong magnetic field and propagated parallel to this field. The polarisation rotation angle (F) is proportional to the
primary current
primary sensing element
primary converter
transmission system
secondary converter
secondary signal (i or u)
primary supply
secondary supply
fig. 17: hybrid sensor diagram.
F = V H dL In this equation, V is a characteristic of the optical medium, known as Verdet's constant. Generally small, it has a varying dependence on temperature. As the Faraday effect is divergent, a monochromatic light (with single frequency) must be used.
In practice This effect is used with optical crystals or fibres. In both cases a light source is required and the optical information must be processed so that it can be used by the protection and control/ monitoring units. c light source Frequently a monomode laser diode with a wavelength approaching 780 nanometres: Verdet's constant is greatest in this part of the wavelength spectrum. c optical crystal One or more crystals can be used, surrounding to a greater or lesser extent the conductor in which the current to be measured flows (see fig. 19a). In free field optical configuration, which is the most frequent case with crystals, mechanical-optical alignment problems are particularly great. c optical fibre The controlled optics technique uses as a sensing element a monomode optical fibre which can be wound several times around the primary conductor (see fig. 19b). In this case application of Ampere's theorem gives: F=VNI This technique ensures increased sensitivity.
Optical fiber sensors are not sensitive to external currents (return conductor, other phases, other circuits), whereas optical crystal sensors are, to a greater or lesser extent, depending on their construction technology. On the other hand, the optical characteristics of the sensing element (crystal or fibre) are particularly affected by variations in temperature and mechanical stresses. c converting the optical signal into an electrical signal. This is achieved by comparing the light beams emitted and received, generally using polarising-
y , ~ , P ~ | z y Q ~ z , y Q P ~ | z Q P ~ | z P |
Faraday effect optical medium linear polarisation incident light polarisation plane rotation angle F linear polarisation transmitted light L magnetic field H
circulation of the magnetic field (H) along the optical path L (see fig. 18).
y , P ~ | z y , Q ~ , P | y Q P ~
fig. 18: graphic representation of the Faraday effect.
a - exploded view of an optical crystal sensor
4 crystals making up the sensing element primary conductor PR analyser polariser lens optical fibre connectors
b - diagram of an optical fibre sensor
L L F2 monomode connection F1 L LSR P laser diode P electronics P SSP
L F3
LSR L F1 F2 F3
I
SSP P PR
semi-reflecting blade Selfoc lenses polarisation hold optical fibre sensor fibre multimode optical fibres current to be measured separating/polarising system at 45 photodiodes reference photodiodes
fig. 19 : Faraday effect current sensors.
separating prisms combined with photodiodes which convert the light signal into an analog electrical signal. The latter is then processed and amplified so that it can be used by the protection and control/monitoring units. Accuracy Optical sensors (fibre or crystal) are sensitive to external conditions (temperature, auxiliary energy source), which thus affects their accuracy. c influence of temperature Temperature affects three parameters: v Verdet's constant V v the birefringence of the optical medium, v the wavelength of the light emitted by the laser diode. In order to operate in the conditions encountered on electrical networks, optical sensors must be temperature compensated (see appendix 3, [5]). Compensation can take the form of: v permanent action on the optical sensing element (double twist of the fibre, return journey of light through the fibre, thermal compensation of crystal, etc...), v keeping the laser diode at a temperature compatible with the required accuracy, v allowing for the real temperature of each element in the output signal shaping line. c influence of mechanical constraints. The crystal or optical fibre must have a very low birefringence rate so as not to alter light polarisation in the absence of magnetic field. Mechanical stresses on the crystal or fibre, linked to temperature variations, implementation and operation, must not change this birefringence rate. c influence of signal conversion electronics. The crystal and optical fibre
a - in HVA (Merlin Gerin), with:

c in the foreground an optical sensor with its wound-on optical fibre conductor, c in the background an equivalent conventional, but more space-consuming, CT sensor.
are perfectly linear. However, the processing electronics is limited in its dynamics, for a given accuracy, by: v its pass-band; v its ability to detect to a 2 the rotation angle of the polarisation plane. However, digital signal processing techniques can be used to correct this; v the supply voltages of the components making up the primary and secondary converter(s). Despite these problems, present-day techniques can produce optical current sensors of an accuracy comparable to those of CTs (see fig. 20). EMC behaviour Since galvanic insulation between the circuits (primary and secondary) is perfect (no coupling capacity), the EMC behaviour (conducted disturbances) of this sensor is good. However, this behaviour may be affected by that of its primary and secondary converters sensitive to radiated disturbances (shieldings, relative positioning, etc...). Note that this perfect galvanic insulation is, as regards safety, a major asset for this sensor type in that it does away with the explosion risks existing in HVB with oil insulated CTs.
Hall effect current sensors

Hall effect A semiconductive wafer through which a current i flows, immersed in a magnetic induction field B , develops between two sides a potential difference known as Hall's voltage VH meeting the equation: VH = K i B where K is the sensor's coefficient of sensitivity. This wafer forms the sensing element of the Hall effect current sensor. Principle The Hall effect assumes that in a long wafer (see fig. 21), fitted with wide electrodes injecting current i, all the electrons move uniformly at speed V in the opposite direction to current i. When a magnetic induction field B is applied perpendicularly to one of the large sides of the wafer, the -e charged electrons are deflected to one of the
d i
EH
-e v -e
EH
VH
b - in HVB (Square D), with:

c in the foreground a conventional CT, c in the background an equivalent, but less space-consuming, optical sensor.
fig. 20: optical sensor examples.
fig. 21: theoretical diagram of the Hall effect sensor.
small sides where they accumulate due to the effect of Laplace's law force. F = e V B The load unbalance between the two small sides causes a Hall electric field, EH , to appear, which grows until the force (e EH ) balances that of the induction field. In these conditions, the electrons resume a uniform movement, and Hall's electric field is written as: EH = B j / (N e) where N is the number of charge carriers (-e) and j the current density in the wafer: this results in Hall's voltage: VH = K i B / (N e d) In practice A practical solution for increasing sensor sensitivity is to increase B . This is achieved by placing the Hall generator in the airgap of a magnetic core through which the induction flux flows due to the magnetic field created by the current to be measured (see fig. 5). Current is supplied and the signal processed by means of electronic elements. Accuracy The response of Hall effect sensors is not exactly proportional to B. This is due to three factors: v offset voltage, v linearity error, v fluctuation with temperature. c offset voltage An error voltage linked to the production of the sensing element. For a given temperature range it can be corrected by the secondary converter. c linearity error Presence of a magnetic circuit, even with a relatively high airgap, introduces a nonlinearity resulting from the saturation phenomena. This sensor's dynamics depends on the sizing of the magnetic circuit. c fluctuation with temperature Temperature influences in two ways: v by the coefficient of sensitivity K which varies by roughly 0.01 % per C, v by the mechanical constraints, further to temperature variations undergone by the sensing element. Sensor production must allow for all these influencing factors and aim to compensate them in order to obtain and
guarantee, in specified conditions of use, an accuracy compatible with the intended purpose (measurement or protection or both). This results in the functional electronics diagram necessary for proper operation of this sensor (see fig. 22). The pass-band of these sensors is relatively large and it is possible to measure DC currents and currents with frequencies of around 40 kHz. Passband width, for this sensor type, depends on the technology of the magnetic circuit, the electronic components and the architecture used for signal processing. EMC behaviour Absence of galvanic insulation between the sensor and the electronic elements is a major handicap, particularly in HV. The EMC of the assembly (Hall effect sensor, protection and control/ monitoring unit) may thus not be perfect.
amplifier (A) controlled by probe (SD) voltage proportional to the flux flowing in the magnetic core (MC). The resulting flux in this core is zero and the following can thus be written: I2 = N1 I1 / N2, where N1 = number of primary winding turns, N2 = number of secondary winding turns. Accuracy This system has an excellent accuracy. Current transformer error measuring benches make use of this principle. The module error can be limited to very small values ( 0.02 %), as can also be the phase error which may be less than 0.1 angular minute (but which depends on the flux cancellation electronic circuits). This sensor's performances mainly depend on the performances of the amplifier, both as regards measuring range and accuracy. Note: Zero flux current transformers are used to measure DC currents. EMC behaviour The signal for cancelling flux approaching zero is easily disturbed. This CT type must therefore be placed in a highly protected electromagnetic environment (shields, filtered supplies, etc...).
zero flux current sensor

Principle The sensing element is a magnetic circuit (CM) (see fig. 4) in which the flux created by the current to be measured (I1) is cancelled by a current (I2). The value of this current is adjusted automatically by an electronic power
i probe
DC generator
differential amplifier and stabiliser of the coefficient of sensitivity in temeraturep
residual induction compensation circuit thermistors
powers supplies
amplifier + filter
mechanical force compensation circuit
Uh = K
fig. 22: functional diagram of the electronics of a Hall effect sensor.
6. comparision table, synthesis
c poor c c average c c c good c c c c very good performances: c linearity c exactness c dynamics c accuracy c EMC capacities: c measuring standard c supplying energy to protection and control/monitoring units c supplying the measuring signal to: v analog energy meters v digital energy meters v digital protection and control/ monitoring units relative cost compared with switchgear: c in HVA c in HVB number of sensors installed each year: c present situation c foreseeable evolution
CT conventional current transformer c c c ccc ccc cc cccc cccc cccc cccc
SS Rogowski coil sensor
optical sensors
zero flux current transformer
cccc cccc cccc ccc ccc cc c c ccc cccc
cccc ccc cccc ccc cccc cc c c cc cc
ccc ccc cc cccc cc cccc cc c c c
ccc cc cccc ccc
cccc ccc ccc ccc
cc ccc c cc
c c c c
7. conclusion and future
present solutions
Today most protection and control/ monitoring equipment in operation uses electromagnetic or electronic technologies. This equipment requires sufficiently powerful signals ( 5 to 50 VA) from current sensors which are often at a considerable distance ( 2 to 150 m). In HV equipment, this power is supplied by conventional current transformers. However, in HVA, several hundred digital protection and control/monitoring units are in operation. The majority, able to process signals with low energy levels, are associated with SS sensors. In HVB, experiments are currently underway with Faraday effect sensors using optical crystals or fibres. Zero flux CTs are above all used in test bays and in DC transmission networks.
future solutions
The very fast evolution of protection and control/monitoring systems towards digital technologies has already resulted in major changes to sensor specifications. These specifications give priority to EMC, linearity and operating range of sensors.
c EMC Linked to the increasing use of electronic technologies. In this field, the optical sensor's behaviour is ideal. c linearity Linear CTs (TPZ type) are generally space consuming (airgap present) and their linearity is not perfect. When a fault occurs, they transmit very high current leading to very high thermal stresses to the equipment connected to them. The SS sensor, with its perfect linearity, has the best performance here. c operating range A CT has a very narrow operating range which limits its use to one single application. On the other hand, optical and SS sensors, with wider operating ranges ( 10 times) have greater possibilities of use, only limited by the equipment to which they are connected. Optical and SS sensors thus give the best performances as regards the new technical constraints. A hybrid sensor, using a Rogowski coil as its sensing element and optical fibres as its transmission system, could prove the ideal solution. In HVA, this solution is not yet feasible from an economic point of view.
However, the SS solution, through its cost and associated advantages, is the solution both for now and the future. In HVB the use of these new sensors depends on the development of digital solutions for the protection and control/ monitoring units and on the creation of digital interfaces for existing units. Sensor evolution will take off as soon as this equipment becomes available. This evolution has already begun and systems are already available, either fully optical or as in the ideal solution recommended above. Consequently, conventional CTs, with their limited performances and relatively high costs, are doomed to vanish in the long term in HV, except perhaps those used for metering energy bills.
appendix 1: CT accuracy as in IEC 185
limits of errors for measurement secondaries

Percentage current (ratio)
accuracy class error at percentage of rated current shown below 5 0.1 0.2 0.5 1.0 0.4 0.75 1.5 3.0 20 0.2 0.35 0.75 1.5 100 0.1 0.2 0.5 1.0 120 0.1 0.2 0.5 1.0
phase displacement at percentage of rated current shown below

minutes 5 15 30 90 180 20 8 15 45 90 100 5 10 30 60 120 5 10 30 60 centiradians 5 0.45 0.9 2.7 5.4 20 0.24 0.45 1.35 2.7 100 0.15 0.3 0.9 1.8 120 0.15 0.3 0.9 1.8
Limits of error for protective secondaries

accuracy class current error at rated primary current % 1 3 phase displacement at rated primary current minutes 60 centiradians 1.8 composite error at rated accuracy limit primary current % 5 10
5P 10 P
appendix 2: CT classification as in IEC 44-6
The various classes of current transformers for protection, defined by IEC 44-6 according to their performances, are listed in the table opposite.
class P TPS
performances Accuracy limit defined by the composite error ( c) with steady state symmetrical primary current. No limit for remanent flux. Low leakage flux current transformer for which performance is defined by the secondary excitation characteristics and turns ratio error limits. No limit for remanent flux. Accuracy limit defined by peak instantaneous error ( ) during specified transient duty cycle. No limit for remanent flux. Accuracy limit defined by peak instantaneous error ( ) during specified transient duty cycle. Remanent flux not to exceed 10% of the saturation flux. Accuracy limit defined by peak instantaneous alternating current component error ( ac) during single energization with maximum d.c. offset at specified secondary loop time constant. No requirements for d.c. component error limit. Remanent flux to be practically negligible.
TPX TPY
TPZ
[1] Techniques de l'ingnieur: Transformateurs de mesure. D 4720 12-1990, D 4722 12-1990, D 4724 3-1991, R1016 10-1992. [2] CEI 44-6 : First edition 1992-03 : Transformateurs de mesure. Part 6 : Prescriptions concernant les transformateurs de courant pour protection pour la rponse en rgime transitoire. [3] CEI 185 : Second edition 1987 and its amendment 1 1990-07. Transformateurs de courant.
[4] Mthode rapide de prdtermination des transformateurs de courant, Pierre SCHUELLER (Merlin Gerin) page 41 to 45, RGE n4 april 1990. [5] Techniques de lingnieur: Capteurs de courant fibres optiques. R 1016 10-1992. [6] La CEM: la compatibilit lectromagntique, Cahier Technique Merlin Gerin n 149. F. VAILLANT.
[7] Le transformateur de courant pour la protection en HT, Cahier Technique Merlin Gerin n 164. M. ORLHAC.
Ral.: Sodipe - Valence - Photo.: Merlin Gerin, Square D Edition: DTE - Grenoble 03-95 - 2500 - Printing: Clerc Printed in France
n 172
earthing systems in LV
Bernard Lacroix An ESPCI 74 engineering graduate (from the Ecole Suprieure de Physique et Chimie Industrielle de Paris), he then worked 5 years for Jeumont Schneider, where his activities included development of the TGV chopper. After joining Merlin Gerin in 1981, he was then in turn Sales Engineer for UPS and sales manager for protection of persons. Since 1991 he is in charge of prescription for LV power distribution.
Roland Calvas An ENSERG 1964 engineering graduate (from the Ecole Nationale Suprieure d'Electronique et Radiolectricit de Grenoble) and an Institut d'Administration des Entreprises graduate, he joined Merlin Gerin in 1966. During his professional career, he has been sales manager and then marketing manager for protection of persons. He is currently in charge of technical communication for the groupe Schneider.
E/CT 172, first issued September 1995
glossary
Electric Shock Electrocution EMC In IMD GFLD MV/HV RCD SCPD STD TBM TEM UL Application of a voltage between two parts of the body Electric Shock resulting in death Electro Magnetic Compatibility Operating threshold of a RCD Insulation Monitoring Device Insulation Fault Location Device Medium Voltage: 1 to 35 kV as in CENELEC (circular of the 27.07.92) High Voltage: 1 to 50 kV as in french standard (14.11.88) Residual Current Device Short-Circuit Protection Device (circuit-breakers or fuses) Short Time Delay protection (protection against short-circuit overcurrents by circuit-breaker with rapid trip release) Technical Building Management Technical Electrical Power Distribution Management Conventional limit voltage (maximum acceptable contact voltage) known as the safety voltage
earthing systems in LV
content
1. Introduction Evolution of needs Causes of insulation faults Hazards linked to insulation faults 2. Earthing systems and protection of persons TN system TT system IT system 3. Earthing systems confronted with Fire fire and electrical power unavailabiliy Electrical power unavailability hazards 4. Influences of MV on BV, according to the earthing systems Lighning Operating overvoltages MV-frame disruptive breakdown of the transformer MV-LV disruptive breakdown inside the transformer TN system TT system IT system Neutral protection according to the earthing system Methods for choosing the earthing system Conclusion p. 4 p. 4 p. 4 p. 7 p. 8 p. 9 p. 9 p. 12 p. 12 p. 14 p. 14 p. 14 p. 15 p. 17 p. 17 p. 18 p. 19 p. 21 p. 21 p. 22
This Cahier Technique reviews the hazards that insulation faults represent for safety of persons and propery. It emphasises the influence of earthing systems and the availability of electrical power. It presents the three earthing systems defined in standard IEC 364 and used to varying degrees in all countries. Each earthing system is looked at in terms of dependability (safety, maintenability and availability). There is no such thing as a bad earthing system; they all ensure safety of persons. Each system has its own advantages and disadvantages and the user must therefore be guided according to his needs, with the exception, however, of prescription or of standard or legislative bans. Readers interested in the practices of various countries and in evolution of earthing systems should read Cahier Technique n 173.
5. Switchgear linked to choice of earthing system
6. Choice of eathing system and conclusion 7. Bibliography
1. introduction
evolution of needs
Today the 3 earthing systems such as defined in IEC 364 and French standard NF C 15-100, are: c exposed-conductive parts connected to neutral -TN-; c earthed neutral -TT-; c unearthed (or impedance-earthed) neutral -IT-. The purpose of these three systems is identical as regards protection of persons and property: mastery of insulation fault effects. They are considered to be equivalent with respect to safety of persons against indirect contacts. However, the same is not necessarily true for dependability of the LV electrical installation with respect to: c electrical power availability; c installation maintenance. These quantities, which can be calculated, are subjected to increasingly exacting requirements in factories and tertiary buildings. Moreover, the control and monitoring systems of buildings -TBM- and electrical power distribution management systems -TEM- play an increasingly important role in management and dependability. This evolution in dependability requirements therefore affects the choice of earthing system. It should be borne in mind that the concern with continuity of service (keeping a sound network in public distribution by disconnecting consumers with insulation faults) played a role when earthing systems first emerged.
c use of insulating materials; c distancing, which calls for clearances in gases (e.g. in air) and creepage distances (concerning switchgear, e.g. an insulator flash over path). Insulation is characterised by specified voltages which, in accordance with standards, are applied to new products and equipment: c insulating voltage (highest network voltage); c lightning impulse withstand voltage (1.2; 50 s wave); c power frequency withstand voltage (2 U + 1,000 V/1mn). Example for a LV PRISMA type switchboard: c insulating voltage: 1,000 V; c impulse voltage: 12 kV. When a new installation is commissioned, produced as per proper practices with products manufactured as in standards, the risk of insulation faults is extremely small; as the installation ages, however, this risk increases. In point of fact, the installation is subject to various aggressions which give rise to insulation faults, for example: c during installation: v mechanical damage to a cable insulator; c during operation: v conductive dust, v thermal ageing of insulators due to excessive temperature caused by: - climate, - too many cables in a duct, - a poorly ventilated cubicle, - harmonics, - overcurrents, etc, v the electrodynamic forces developed during a short-circuit which may damage a cable or reduce a clearance, v the operating and lightning overvoltages, v the 50 Hz return overvoltages, resulting from an insulation fault in MV.
It is normally a combination of these primary causes which results in the insulation fault. The latter is: c either of differential mode (between live conductors) and becomes a short-circuit; c or of common mode (between live conductors and frame or earth), a fault current -said to be common mode or zero sequence (MV)- then flows in the protective conductor (PE) and/or in the earth. LV earthing systems are mainly concerned by common mode faults which mainly occur in loads and cables.
hazards linked to insulation faults

An insulation fault, irrespective of its cause, presents hazards for: c human life; c preservation of property; c availability of electrical power; the above all depending on dependability. Electric Shock of persons A person (or animal) subjected to an electrical voltage is electrified. According to the gravity of the Electric Shock, this person may experience: c discomfort; c a muscular contraction; c a burn; c cardiac arrest (this is Electrocution) (see fig. 1). Since protection of persons against the dangerous effects of electric current takes priority, Electric Shock is thus the first hazard to be considered. The current strength I -in value and time-, passing through the human body (in particular the heart) is the dangerous aspect. In LV, the impedance value of the body (an important aspect of which is skin resistance) virtually changes only according to environment (dry and wet premises and damp premises).
causes of insulation faults

In order to ensure protection of persons and continuity of service, conductors and live parts of electrical installations are insulated from the frames connected to the earth. Insulation is achieved by:
In each case, a safety voltage (maximum acceptable contact voltage for at least 5 s) has been defined: it is known as the conventional limit voltage UL in IEC 479. IEC 364 paragraph 413.1.1.1 and NF C 15-100 state that if there is a risk of contact voltage Uc exceeding voltage UL, the application time of the fault voltage must be limited by the use of protection devices (see fig. 2). Fire This hazard, when it occurs, can have dramatic consequences for both persons and property. A large number of fires are caused by important and localised temperature rises or an electric arc generated by an insulation fault. The hazard increases as the fault current rises, and also depends on the risk of fire or explosion occurring in the premises. Unavailability of electrical power It is increasingly vital to master this hazard. In actual fact if the faulty part is automatically disconnected to eliminate the fault, the result is: c a risk for persons, for example: v sudden absence of lighting, v placing out of operation of equipment required for safety purposes; c an economic risk due to production loss. This risk must be mastered in particular in process industries, which are lengthy and costly to restart. Moreover, if the fault current is high: c damage, in the installation or the loads, may be considerable and increase repair costs and times; c circulation of high fault currents in the common mode (between network and earth) may also disturb sensitive equipment, in particular if these are part of a low current system geographically distributed with galvanic links. Finally, on de-energising, the occurrence of overvoltages and/or electromagnetic radiation phenomena may lead to malfunctioning or even damage of sensitive equipment. Direct and indirect contacts Before beginning to study the earthing systems, a review of Electric Shock by direct and indirect contacts will certainly be useful.
time during which the human body is exposed ms

10,000 5,000 2,000 1,000 500
c1
c2 c3
1
200 100 50 20 10 0.1 0.2 0.5 1 2
mA
5 10 20 50 100 200 500 1,000 2,000 5,000 10,000
current passing through the human body zone 1: perception zone 2: considerable discomfort zone 3: muscular contractions zone 4: risk of ventricular fibrillation (cardiac arrest) c1: likelihood 5 % c3: likelihood > 50 %
fig. 1: time/current zones of AC current effects (15 Hz to 100 Hz) on persons as in IEC 479-1.
c dry or wet premises and places: UL i 50 V presumed contact voltage (V) maximum breaking time of the protection device (s) AC DC < 50 5 5 50 5 5 75 0.60 5 90 0.45 5 120 0.34 5 150 0.27 1 220 0.17 0.40 280 0.12 0.30 350 0.08 0.20 500 0.04 0.10 c damp premises and places: UL i 25 V presumed contact voltage (V) maximum breaking time of the protection device (s) AC (a) DC (b) 25 5 5 50 0.48 5 75 0.30 2 90 0.25 0.80 110 0.18 0.50 150 0.10 0.25 220 0.05 0.06 280 0.02 0.02
fig. 2: maximum time for maintenance of contact voltage as in standard IEC 364.
c direct contact and protection measures This is accidental contact of persons with a live conductor (phase or neutral) or a normally live conductive element (see fig. 3a). In cases where the risk is very great, the common solution consists in distributing electricity using a nondangerous voltage, i.e. less than or equal to safety voltage. This is safety by extra-low voltage (SELV or PELV), in LV (230/400 V). Protection measures consist in placing these live parts out of reach or in insulating them by means of insulators, enclosures or barriers. A complementary measure against direct contacts consists in using instantaneous i 30 mA High Sensitivity Residual Current Devices known as HS-RCDs. Note: this measure is necesssary in all circuit supply cases where implementation of the earthing system downstream is not mastered. Consequently, some countries make this measure a requirement: v for sockets of rating i 32 A, v in some types of installations (temporary, worksite, etc.). Treatment of protection against direct contacts is completely independent from the earthing system. c indirect contact, protection and prevention measures Contact of a person with accidentally energised metal frames is known as indirect contact (see fig. 3b). This accidental energising is the result of an insulation fault. A fault current flows and creates a potential rise between the frame and the earth, thus causing a fault voltage to appear which is dangerous if it exceeds voltage UL. As regards this hazard, the installation standards (IEC 364 at international level) have given official status to three earthing systems and defined the corresponding installation and protection rules.
The protection measures against indirect contacts are based on three basic principles: c earthing of the frames of loads and electrical equipment to prevent an insulation fault from resulting in the equivalent of a direct contact; c equipotentiality of simultaneously accessible frames Interconnection of these frames considerably helps to reduce contact voltage. It is performed by the protective conductor (PE) which connects the frames of electrical equipment for entire buildings, completed if required by additional equipotential links (see fig. 4). Reminder: equipotentiality cannot be complete in all points (in particular in single level premises). Consequently, for the study of earthing systems and their associated protection devices, the hypothesis chosen by standard makers Uc = Ud is applied since Uc is at the most equal to Ud. v Ud = fault voltage, with respect to the deep earth, of the frame of an electrical device with an insulation fault, v Uc = contact voltage depending on the potential Uc and the potential reference of the person exposed to the hazard, generally the ground. c managing the electrical hazard v this management is optimised by prevention. For example, by measuring insulation of a device before energising it, or by fault prediction based on live monitoring of insulation evolution of an unearthed installation (IT system), v if an insulation fault occurs, generating a dangerous fault voltage, it must be eliminated by automatically disconnecting the part of the installation where this fault occurred. How the hazard is removed then depends on the earthing system.
a) direct contact ph
Uc
b) indirect contact 3
Id
Uc
fig. 3: direct and indirect contacts.
individual protective conductors (PE) reinforcement meshing main protective conductor
water gas heating
measuring strip
ditch bottom loop
earthing conductor
fig. 4: equipotentiality in a building.
2. earthing systems and protection of persons

This section defines the Electric Shock and Electrocution hazards for the various earthing systems, such as specified by the International Electrotechnical Committee in standard IEC 364. The LV earthing system characterises the earthing mode of the secondary of the MV/LV transformer and the means of earthing the installation frames. Identification of the system types is thus defined by means of 2 letters: c the first one for transformer neutral connection (2 possibilities): v T for connected to the earth, v I for isolated from the earth; c the second one for the type of application frame connection (2 possibilities): v T for directly connected to the earth, v N for connected to the neutral at the origin of the installation, which is connected to the earth (see fig. 5). Combination of these two letters gives three possible configurations: transformer neutral c if T c if I i.e. TT, TN and IT. frame T or N; T;
3 N PEN N PE N PE PE
however several earthing systems may be included in the same installation, see figure 6 as an example. Note 3: In France, as in standard NF C 13-100 concerning delivery substations, in order to prevent hazards originating
in MV, the LV earthing system is expressed by an additional letter according to interconnection of the various earth connections (see fig. 7). Let us now see how to protect persons in each case.
3 N
3 N
T T 3 N I N 3 N
fig. 5: connection mode of the neutral at the origin of the installation and of the frames of the electrical loads.
Note 1: The TN system, as in IEC 364 includes several sub-systems: c TN-C; if the N and PE neutral conductors are one and the same (PEN); c TN-S: if the N and PE neutral conductors are separate; c TN-C-S: use of a TN-S downstream from a TN-C (the opposite is forbidden). Note that the TN-S is compulsory for networks with conductors of a crosssection i 10 mm2 Cu. Note 2: Each earthing system can be applied to an entire LV electrical installation;
TN-C
TN-S
TT
IT
fig. 6: example of the various earthing systems included in the same installation.
additional earthing of the letter MV/LV substation R (connected) c N (of neutral) c S (separated) v (c = interconnected, v = separate) earthing of the LV neutral c c v earthing of the LV application c v v
fig. 7: linking of LV earth connections with that of the MV/LV substation.
TN system
When an insulating fault is present, the fault current Id is only limited by the . impedance of the fault loop cables (see fig. 8):
Id =
Uo . Rph1 + Rd + RPE
For a feeder and as soon as Rd 0:
Id =
0.8 Uo . Rph1 + RPE
In point of fact, when a short-circuit occurs, it is accepted that the impedances upstream from the relevant feeder cause a voltage drop of around 20 % on phase-to-neutral voltage Uo, which is the nominal voltage between phase and earth.
calculation. Certain country standards recommend this. To guarantee this condition, another approach consists in imposing a maximum impedance value on the fault loops according to the type and rating of the SCPDs chosen (see British standard BS 7671). This approach may result in increasing the cross-section of the live and/or protective conductors. Another means of checking that the device will ensure protection of persons is to calculate the maximum length not to be exceeded by each feeder for a given protection threshold Ia. To calculate Id and Lmax, three simple methods can be used (see Cahier Technique n 158): c the impedance method; c the composition method; c the conventional method. The latter gives the following equation:
For the protection device to perform its function properly, Ia must be less than Id, hence the expression of Lmax, the maximum length authorised by the protection device with a threshold Ia:
Lmax = 0.8 Uo Sph
(1 + m) Ia
Id thus induces a fault voltage with

respect to earth: Ud = RPE Id i.e.:
c Lmax: maximum length in m; c Uo: phase-to-neutral voltage 230 V for a three-phase 400 V network; c : resistivity to normal operating temperature; c Ia: automatic breaking current: v for a circuit-breaker Ia = Im (Im operating current of the magnetic or short time delay trip release), v for a fuse, current such that total breaking time of the fuse (prearcing time + arcing time) complies with the standard (see fig. 9), c m =
RPE Ud = 0.8 Uo Rph1 + RPE .

For 230/400 V networks, this voltage of around Uo/2 (if RPE = Rph) is dangerous since it exceeds the limit safety voltage, even in dry atmospheres (UL = 50 V). The installation or part of the installation must then be automatically and promptly de-energised (see fig. 9). As the insulation fault resembles a phase-neutral short-circuit, breaking is achieved by the Short-Circuit Protection Device (SCPD) with a maximum specified breaking time depending on UL. Implementation To be sure that the protection device really is activated, the current Id must be greater than the operating threshold of the protection device Ia (Id > Ia) irrespective of where the fault occurs. This condition must be verified at the installation design stage by calculating the fault currents for all the distribution circuits. If the same path is taken by the protective conductor - PE- and the live conductors, this will simplify the
Sph . SPE
Id =
=
0.8 Uo 0.8 Uo = Z Rph + RPE
0.8 Uo Sph
(1 + m) L
If the line is longer than Lmax, either conductor cross-section must be increased or it must be protected using a Residual Current Device (RCD).
A N D PE
Id
C Rd Ud
Ud
0.8 Uo if RPE = Rph and Rd = 0 2
Id =
Uo 0.8 Uo RAB + Rd + RCD Rph + RPE
fig. 8: fault current and voltage in TN system.
TT system
When an insulation fault occurs, the fault current Id (see fig. 10) is mainly limited by the earth resistances (if the earth connection of the frames and the earth connection of the neutral are not associated). Still assuming that Rd = 0, the fault current is: Uo Id Ra + Rb This fault current induces a fault voltage in the earth resistance of the applications:
Uo Ra Ra + Rb As earth resistances are normally low and of the same magnitude ( 10 ), this voltage of the order of Uo/2 is Ud = Ra Id, or Ud =
dangerous. The part of the installation affected by the fault must therefore be automatically disconnected (see fig. 11). Implementation As the fault current beyond which a risk is present ( Id0 =
UL ) is far lower Ra than the settings of the overcurrent protection devices, at least one RCD must be fitted at the supply end of the installation. In order to increase availability of electrical power, use of several RCDs ensures time and current discrimination on tripping. All these RCDs will have a nominal current threshold In less than Id0. The standard stipulates that de-energising by the RCDs must occur in less than 1 s (this time may be
increased to 5 s if Ra and Rb are interconnected). Note that protection by RCD: c does not depend on cable length; c authorises several separate Ra earth connections (an unsuitable measure since the PE is no longer a unique potential reference for the entire installation). Cahier Technique n 114 gives a detailed description of RCD technology and use.
IT system
The neutral is unearthed, i.e. not connected to the earth. The earth connections of the frames are normally interconnected (just like the TN and TT earthing systems). c in normal operation (without insulation fault), the network is earthed by the network leakage impedance. We remind you that natural earth leakage impedance of a three-phase 1 km long cable is characterised by the standard values: v C = 1 F / km, v R = 1 M / km, which give (in 50 Hz): v Zcf = 1 / j C = 3,200 , v Zrf = Rf = 1 M, therefore Zf Zcf = 3,200 . In order to properly set the potential of a network in IT with respect to the earth, we advise that you place an impedance (Zn 1,500 ) between transformer neutral and the earth.... this is the IT impedance-earthed system.
Uo (volts) phase/neutral voltage 127 230 400 > 400
breaking time (seconds) UL = 50 V 0.8 0.4 0.2 0.1
breaking time (seconds) UL = 25 V 0.35 0.2 0.05 0.02
fig. 9: breaking time in TN system (taken from IEC 364 tables 41 and 48A).
Id
PE
In i
UL 3A 1A 500 mA 300 mA 30 mA
UL Ra
maximum resistance of earth connection 50 V 16 50 100 166 1,660 25 V 8 25 50 83 833
Ud
Rb
Ra
Uo Ra Ud = Uo Ra + Rb Ra + Rb fig. 10: fault current and voltage in TT system.
Id
fig. 11: upper limit of the resistance of the frame earth connection not to be exceeded according to RCD sensitivity and limit voltage UL [In = F (Ra)].
c behaviour on the first fault v unearthed neutral: The fault current is formed as follows (maximum value in the case of a full fault and neutral not distributed) If = Ic1 + Ic2, where: Ic1 = j Cf V1 3 and Ic2 = j Cf V2 3, where: Id = Uo 3 Cf . For 1 km of 230/400V network, the fault voltage will be equal to: Uc = Rb Id, i.e. 0.7 V if Rb = 10 . This voltage is not dangerous and the installation can thus be kept in operation. If the neutral is distributed, the shift of neutral potential with respect to the earth adds a current Icn = Uo Cf and Id = Uo 4 Cf (see fig. 12). v impedance-earthed neutral: First fault current:
U Id = where Zeq 1 1 = + 3j Cf Zeq Zn
If
3 2 1 N PE insulation monitoring device surge limiter
If
If
Cf Cf Cf Cf
IcN Ic1 Ic2

Ud
Rb
If
V1
V2
Ic2
If
V1 3 V2 3
IcN Ic1
The corresponding fault voltage is still low and not dangerous; the installation can be kept in operation. Although risk-free continuity of service is a great advantage, it is necessary: - to know that there is a fault, - to track it and eliminate it promptly, before a second fault occurs. To meet this need: - the fault information is provided by an Insulation Monitoring Device (IMD) monitoring all live conductors, including the neutral, - locating is performed by means of fault trackers. c behaviour on the second fault When a second fault occurs and the first fault has not yet been eliminated, there are three possibilities: v the fault concerns the same live conductor: nothing happens and operation can continue, v the fault concerns two different live conductors: if all the frames are interconnected, the double fault is a shortcircuit (via the PE). The Electric Shock hazard is similar to that encountered with the TN system. The most unfavourable conditions for the SCPDs (smallest Id) are obtained when both faults occur on feeders with the same characteristics (cross-sections and lengths) (see fig. 13).
V3
Ud Rb If
fig. 12: first insulation fault current in IT system.
0,8 Uo
Id
3 2 1 N PE
Id
RPE Rph RPE Rph
Ud
Ud
Rb
Id
0.8 Uo 2 (RPE + Rph)
Ud
0.8 Uo 4
fig. 13: 2nd insulation fault current in IT system (distributed neutral) and relevant feeders with the same cross-section and length.
The SCPDs have to comply with the following relationships: - if the neutral is distributed and one of the two faulty conductors is the neutra:
- non-distributed neutral: Lmax = 3 0.8 Uo Sph . 2 (1 + m) Ia
0.8 Uo , 2Z - or if the neutral is not distributed:
Ia i
Ia i
0.8 Uo 2Z
Note that if one of the two faults is on the neutral, the fault current and fault voltage are twice as low as in the TN system. This has resulted in standard makers authorising longer SCPD operating times (see fig. 14). Just as in the TN earthing system, protection by SCPD only applies to maximum cable lengths: - distributed neutral:
1 0.8 Uo Sph , Lmax = 2 (1 + m) Ia
This is provided that the neutral is protected and its cross-section equal to phase cross-section... This is the main reason why certain country standards advise against distributing the neutral. v case where all frames are not interconnected. For frames earthed individually or in groups, each circuit or group of circuits must be protected by a RCD. In point of fact, should an insulation fault occur in groups connected to two different earth connections, the earthing system's reaction to the insulation fault (Id, Ud) is similar to that of a TT system (the fault current flows through the earth). Protection of persons against indirect contacts is thus ensured in the same manner In i (see table in fig. 11).
UL Ra
Note that in view of the times specified by the standard, horizontal time discrimination can be achieved to give priority to continuity of service on certain feeders. Note: in order to protect LV unearthed networks (IT) against voltage rises (arcing in the MV/LV transformer, accidental contact with a network of higher voltage, lightning on the MV network), French standard NF C 15-100 stipulates that a surge limiter must be installed between the neutral point of the MV/LV transformer and the earth (Rb). Readers wishing to study the IT earthing systems in greater detail should read Cahier Technique n 178. So as to obtain a concise overview of the quantities characterising the various earthing systems, as regards protection of persons, the main formulas are listed in the table in figure 15.
Uo/U (volts) Uo: phase/neutral voltage U: phase to phase voltage 127/220 230/400 400/690 580/1 000
UL = 50 V breaking time (seconds) neutral not distributed neutral distributed 0.8 0.4 0.2 0.1 5 0.8 0.4 0.2
UL = 25 V breaking time (seconds) neutral not distributed 0.4 0.2 0.06 0.02
neutral distributed 1.00 0.5 0.2 0.08
fig. 14: maximum breaking times specified in IT system (as in IEC 364 tables 41B and 48A).
Id
TN TT IT 1st fault double fault with neutral double fault between phases
Ud
0.8 Uo 1+ m
Lmax
continuity of service vertical discrimination vertical discrimination no tripping
(1 + m) L
Uo Ra + Rb
0.8 Uo Sph
0.8 Uo Sph (1 + m) Ia
Uo Ra Ra + Rb
<< UL m 0.8 Uo i 2 1+ m
i m 2 3 0.8 Uo 1+ m
no constraint
<1A
1 0.8 Uo Sph i 2 (1 + m) L
i 3 2
1 2
3 2
0.8 Uo Sph (1 + m) Ia
(1 + m) L
0.8 Uo Sph
(1 + m) Ia
0.8 Uo Sph
vertical discrimination and possibility of horizontal discrimination to the advantage of hich current feeders
Remember that: c = 22 10-6 /mm2/m for Cu (36 for Al);

cm = Sph ; SPE
c PE cross-section, normally equal to phase cross-section, can be equal to half of phase cross-section when the latter exceeds 35 mm2.... thus increasing Ud in TN and IT.
fig. 15: characteristic quantities of earthing systems.
3. earthing systems confronted with fire and electrical power unavailability hazards
fire
It has been proved, then accepted by standard makers, that contact between a conductor and a metal part can cause fire to break out, in particularly vulnerable premises, when the fault current exceeds 500 mA. To give an example: c premises particularly at risk: petrochemical factories, farms; c premises averagely at risks, but where consequences may be very serious: very high buildings receiving the general public... In the unearthed neutral system, the risk of fire: c is very small on the first fault; c is as important as in TN on the second fault. For the TT and TN earthing systems, the fault current is dangerous given the power developed (P = Rd I2): c in TT = 5A < Id < 50 A; c in TN = 1 kA < Id < 100 kA. The power present where the fault has occurred is considerable, particularly in the TN system, and prompt action is vital as from the lowest current levels in order to limit the dissipated energy ( Rd i2 dt). This protection, specified by the IEC and a requirement of French standards (NF C 15-100, paragraph 482-2-10) is provided by an instantaneous RCD with threshold i 500 mA, regardless of the earthing system. When risk of fire is especially high (manufacture/storage of inflammable materials....) it is necessary and indeed compulsory to use an earthing system with earthed frames which naturally minimises this hazard (TT or IT). Note that the TN-C is banned in certain countries when a risk of fire
de-energising on fault restoration of voltage failure status
and/or explosion is present: as the PE and neutral conductors are one and the same, RCDs cannot be used.
electrical power unavailability

This hazard is a major one for operators, since it results in non-production and repair costs which can be high. It varies according to the earthing system chosen. We remind you that availability (D) is a statistical quantity (see fig. 16) equal to the ratio between two periods of time: c time during which the mains is present;
c reference time equal to the time mains present + mains absent. Mean Up Time (MUT) depends on the overall condition of network insulation. However, insulation deteriorates with time as a result of thermal aggressions and electrodynamic forces due in particular to fault currents. Mean Down Time (MDT) also depends on the fault current and in particular on its strength which, according to its value, may cause: c damage of varying degrees to loads, cables...; c fires; c malfunctionings on the low current control and monitoring equipment. Each earthing system must therefore be examined as regards availability of
MDT
MUT
MDT
MUT
MDT
time de-energising on fault restoration of voltage operating status de-energising on fault restoration of voltage
MDT = Mean Down Time D = availability of a system (detection + repair + MUT = Mean Up Time resumption of operation) Mean failure free time
D =
MUT MDT + MUT
fig. 16: availability of electrical power.
electrical power, with special emphasis on the IT earthing system since it is the only one that authorises non-tripping in the presence of a fault. c the IT earthing system In order to retain the advantage of this system, i.e. not interrupting electrical distribution on the first fault, the second fault must be prevented, since this then presents the same high risks as the TN system. The first fault must therefore be eliminated before a second fault occurs. The use of efficient detection and locating methods and the presence of a reactive maintenance team considerably reduces the likelihood of the double fault. Moreover, monitoring devices are currently available which monitor in time the evolution in insulation of the various feeders, perform fault prediction and thus anticipate maintenance of the first fault. This ensures maximum availability with the IT earthing system. c the TN and TT earthing systems These systems use discrimination on tripping. In TN, this is acquired with short-circuit protection devices if the installation protection plan has been properly designed (current discrimination). In TT, it is easy to implement thanks to the RCDs which ensure current and time discrimination. Remember that, in TN system, repair time according to i2 dt, may be longer than in TT system, wich also affects availability. c for all the earthing systems It is always useful to anticipate insulation faults and in particular those
of certain motors before startup. Bear in mind that 20 % of motor failures are due to an insulation fault which occurs on energising. In point of fact, an insulation loss, even small, on a hot motor cooling down in a damp atmosphere (condensation) degenerates into a full fault on restarting, causing both considerable damage to windings and production loss and even major risks if the motor has a safety function (drainage, fire, fan pump motor, etc.). This type of incident can be prevented, whatever the earthing system, by an Insulation Monitoring Device monitoring the load with power off. If a fault occurs, startup is then prevented. To round off this section on the hazard presented by electrical power unavailability it is clear that, regarding proper electrical power availability, the
earthing systems can be listed in the following order of preference: IT, TT, TN. Note: If, to ensure continuity of service, the installation is fitted with a generator set or a UPS (Uninterruptible Power Supply) in off line, there is a risk of failure to operate or of delayed operation of the SCPDs (the short-circuit current is lower) on changeover to the replacement source (lowest Isc see fig.17). In TN and IT, for safety of persons and property, it is thus vital to check that the protection conditions are always met (operating time and threshold), especially for very long feeders. If this is not so, then RCDs must be used.
I rms
subtranscient state
transcient state
3 In
generator with compound excitation or overexcitation
In 0.3 In
occurence of fault 10 to 20 ms 0.1 to 0.3 s
generator with serial excitation
fig. 17: making a short-circuit in a network supplied by a diesel standby generator.
4. influences of MV on LV, according to the earthing systems

LV networks, unless a replacement uninterruptible power supply (with galvanic insulation) or a LV/LV transformer is used, are influenced by MV. This influence takes the form of: c capacitive coupling: transmission of overvoltage from MV windings to LV windings; c galvanic coupling, should disruptive breakdown occur between the MV and LV windings; c common impedance, if the various earth connections are connected and a MV current flows off to earth. This results in LV disturbances, often overvoltages, whose generating phenomena are MV incidents: c lightning; c operating overvoltages; c MV-frame disruptive breakdown inside the transformer; c MV-LV disruptive breakdown inside the transformer. Their most common consequence is destruction of LV insulators with the resulting risks of Electric Shock of persons and destruction of equipment. advisable to install surge limiters (lightning arresters) at the origin of the LV network, whatever earthing system is used (see fig. 18). Likewise, to prevent coupling by common impedance, it is wise never to connect the following to the earth connection of the LV neutral: c MV lightning arresters; c lightning rods placed on the roof of buildings. In point of fact, the lightning current would cause a rise in potential of the PE and/or the LV neutral (risk of disruptive breakdown by return) and loss of earth connection effectiveness by vitrification. transmitted to the LV network by capacitive and magnetic coupling. Just like all differential mode phenomena, operating overvoltages do not interfere, or only very slightly, with any of the earthing systems.
MV-frame disruptive breakdown of the transformer

On MV-frame disruptive breakdown inside the transformer and when the transformer frame and LV installation neutral are connected to the same earth connection, a MV zero sequence currrent (whose strength depends on the MV earthing system) can raise the frame of the transformer and neutral of the LV installation to a dangerous potential. In point of fact, the value of the transformer earth connection directly conditions the contact voltage in the substation Ut i Rp IhMV and the dielectric withstand voltage of the LV equipment in the substation Utp = Rp IhMV (if the LV neutral earth is
operating overvoltages
Some MV switchgear (e.g. vacuum circuit-breakers) cause considerable overvoltages when operated (see Cahier Technique n 143). Unlike lightning which is a common mode disturbance (between network and earth), these overvoltages are, in LV, differential mode disturbances (between live conductors) and are
lighning
If the MV network is an overhead one, the distributor installs ZnO lightning arresters to limit the effects of a direct or an indirect lightning stroke. Placed on the last pylon before the MV/LV substation, these lightning arresters limit overvoltage and cause lightning current to flow off to earth (see Cahiers Techniques n 151 and 168). A lightning wave, however, is transmitted by capacitive effect between the transformer windings, to the LV live conductors and can reach 10 kV peak. Although it is progressively weakened by the stray capacities of the network with respect to earth, it is
i 125 kV i 10 kV
3 N short connections 3 3
fig. 18: limitation and transmission of lighting overvoltages (whether or not the neutral is earthed, there are common mode overvoltages on phases).
separate from the substation one). The earth connections of the substation and of the LV neutral are not generally connected. If however they are, a limit is given to the common earth connection value to prevent a rise in potential of the LV network compared with the deep earth. Figure 19 gives the common earth connection values for the IhMV values of French public networks. Readers interested in this can consult standard IEC 364-4-442 which explains the risks according to LV earthing systems. Still for public networks (except for Australia and the USA where the fault current can be very high), values encountered range from 10 A in Ireland (an impedance compensates the capacitive current) to 1,000 A in France (underground networks) and in Great Britain. MV industrial networks are normally run in impedance-earthed IT and have a zero sequence current IhMV of a few dozens of amps (see Cahier Technique n 62). The maximum value authorised for the earth connection depends on the equipotentiality conditions of the frames of the LV network, i.e. on its earthing system.
c in TT Whereas the load frames are at the potential of the deep earth, the entire LV network is subjected to IhMV RPB
or RB: there is a risk of disruptive breakdown by return of loads if the voltage developed in RPB or RB exceeds their dielectric withstand.
diagrams (1)
maximum resistance of the earth connection of substation frames Rp () no value stipulated but the following values prevent potential rise of the assembly
IhMV (A)
Z
300 1,000
RPAB () 3 to 20 1 to 10
TNR or ITR
RPAB
IhMV (A)
300 1,000
RPB () 3 1
TTN or ITN
RPB
RA
MV-LV disruptive breakdown inside the transformer

To prevent potential with respect to the earth of the LV network from rising to the phase-to-neutral voltage of the MV network on MV-LV disruptive breakdown inside the transformer, the LV network must be earthed. The consequences of this fault are: c in TN The entire LV network, including the PE, is subjected to voltage IhMV RPAB or RAB. If this overvoltage exceeds the dielectric withstand of the LV network (in practice of the order of 1,500 V), LV disruptive breakdowns are possible if the equipotentiality of all the frames, electrical or not, of the building is not complete.
Z
Utp (kV) IhMV (A) 300 1,000
2 4 RP () 4 8 1 3
10 20 10
RP
RB
TTS or ITS
RA
Z: direct earthing in TN and TT impedance-earthed or unearthed in IT with presence of a discharger. IhMV: maximum strength of the first earth single-phase fault current of the high voltage network supplying the substation. Utp: power frequency withstand voltage of the low voltage equipment of the substation. (1) the third letter of the earthing systems means: c all the frames are linked R; c the substation frame is connected to the Neutral frame: N; c the earth connections are Separated S.
fig. 19: maximum resistance of the earth connection of the substation frames according to network earthing system.
c in IT Operation of a discharger/short-circuiter (known as a surge limiter in France), which short-circuits itself as soon as its arcing voltage is reached, then brings the problem to the level of the TN network one (or TT if there are several application earth connections). In all cases, MV/LV disruptive breakdowns give rise to constraints which can be severe, both for the LV installation and loads, if the value of the LV neutral earth connection is not controlled. Interested readers can
consult IEC 364 which explains risks according to the earthing systems. The example of overhead public distribution in France provides a solution to a situation where risks of lightning, operating overvoltage and transformer frame-MV and MV-LV disruptive breakdown are present (see fig. 20). It shows that equipotentiality of the entire distribution (all MV frames, neutrals and application frames connected) is not vital: each risk is dealt with separately.
This section has described the influence of the MV network. Its conclusions are: c the value of using lightning arresters at the origin of the LV installation, whatever the earthing system type, if the MV and particularly the LV supply is overhead; c connection of the earth connection of the substation with the earth connection of the LV neutral or with those of the application frames, imposes variable constraints on the LV network according to the MV earthing system (value of Ih).
Ih i 300 A
N
metering
u 30 m u8m u8m earth trip Rp < 50 RB < 4
lightning arrester RCD PE
RA < 100
fig. 20: rural overhead public distribution in France.
5. switchgear linked to choice of earthing system

Choice of earthing system affects not only dependability (in the largest sense) but also installation, in particular with respect to the switchgear to be implemented.
TN system
In this system the SCPDs (circuitbreaker or fuses) generally provide protection against insulation faults, with automatic tripping according to a specified maximum breaking time (depending on phase-to-neutral voltage Uo: see fig. 9). c with circuit-breaker As soon as the fault current exceeds the threshold of the short-circuit protection trip release (generally instantaneous), opening occurs in a time far shorter than specified maximum breaking time, for example 5 s for distribution circuits and 0.4 s for terminal circuits (see fig. 21). When impedance of the source and cables is high, either low threshold trip releases must be used or RCDs associated with the SCPDs. These RCDs may be separate residual current devices or be combined with circuit-breakers (residual current circuit-breakers) of low sensitivity. Their threshold must be:
0.8 Uo Rph + RPE Use of a RCD has the advantage of making loop impedance checking unnecessary, a fact which is of particular value when the installation is modified or extended. This solution is clearly not applicable with a TN-C type earthing system (the protective conductor being the same as the neutral one).
In <
c with fuses The fuses used for short-circuit protection are of the gG type and their time/current characteristics (see fig. 22) are defined by standards (household fuses: IEC 241, industrial fuses: IEC 269). Checking suitability with the maximum specified breaking time therefore calls for individual validation of the ratings provided for each protection device. If they are not suitable, either fault loop impedance must be reduced (increased crosssections) or the fuse must be replaced by a low threshold or a residual current circuit-breaker.
persons against indirect contacts. RCDs (see fig. 23) need to be used, associated with circuit-breakers or switches (see IEC 364 - paragraph 413.1.4.2). These devices must meet the following standards in particular: c IEC 755: general rules; c IEC 1008: household residual current switches; c IEC 1009: household residual current SCPDs; c IEC 947-2: industrial residual current circuit-breakers. Their implementation must meet the objectives for: c protection of persons, i.e.: v threshold In i UL/RA, v breaking time i 1s; c continuity of service with thresholds and time delays enabling current and time discrimination; c fire protection with In i 500 mA.
TT system
With this system, the small value of the fault currents (see previous section) does not allow the SCPDs to protect
household (EN 60898) industrial (IEC 947-2)
trip release type B C D G (low threshold) D MA (for motor starter)
operating threshold 3 In i Ia i 5 In 5 In i Ia i 10 In 10 In i Ia i 20 In 2 In i Ia i 5 In 5 In i Ia i 10 In 6.3 In i Ia i 12.5 In
fig. 23: Vigi module of a Compact NS.
output time delay
fig. 21: tripping current (magnetic or short time delay) of LV circuit-breakers.
In gG (A)
63 80 100
Imin. 10 s
160 215 290
Imax. 5 s
320 425 580
Imin. 0.1 s
450 610 820
Imax. 0.1 s 820 110 1,450
threshold shaping
fig. 22: example of fuse operating threshold limits (as in IEC 269 paragraph 5-6-3).
I0 fig. 24: fonctional diagram of a RCD.
IT system
Remember that in the event of a double fault, safety of persons is provided by the SCPDs. When the first insulation fault occurs, the calculation proved there was no risk (contact voltage lower than limit safety voltage). Automatic de-energising is therefore not compulsory: this is the main advantage of this system. To retain this advantage, standards recommend (IEC 364 - paragraph 413.1.5.4) or stipulate (NF C 15-100) the use of an Insulation Monitoring Device (IMD) and locating of the first fault. In point of fact, if a second fault occurs, automatic breaking is vital due to the Electric Shock risk: this is then the role of the SCPDs backed up by the RCDs if required. Locating the first fault for repairs (curative maintenance) is considerably simplified by the use of a Ground Fault Location Device (GFLD). Predictive maintenance, based on the monitoring (recording) of variations in insulation impedance of each circuit, is also possible. LV networks, using the IT system, which take their origin at a MV/LV transformer, must be protected against risks of insulation faults between MV and LV by a surge limiter. Finally, to fix the potential of the LV network with respect to the earth (short network supplied by a MV/LV transformer) and avoid the risk of ferromagnetic resonance, an impedance can be installed between the transformer neutral and the earth. Its value in 50 Hz, of the order of 1,500 , is very high in DC and in very low frequency so as not to obstruct insulation measurement and fault locating. c operating principle of the IMDs A fault on a circuit results in a drop in insulation, or more precisely in resistance of the network compared with earth.
In France, the IMDs and GFLDs have to comply with manufacturing standard UTE 63080. The purpose of the IMDs is thus to monitor the value of this resistance. They normally work by injecting an AC or DC current between the network and the earth and by measuring the value of this current (see fig. 25). Injection of a DC current ensures continuous knowledge of network insulation resistance. If this resistance drops below a pre-set threshold, then the IMD reports the fault. Injection of low frequency AC current (F a few hertz) monitors fault resistance but with a distorsion due to the presence of network leakage capacitites. This minor drawback compared with injection frequency, is made up for by an advantage in first fault locating (one single injection device). LF current injection devices are now available which can separately give the networks insulation resistance and reactance. Moreover, they enable locating of the first fault without circuit opening and without the problems due to highly capacitive feeders. c operating principle of the GFLDs The most common solution is to inject an identifiable current (with a frequency other than network one). The generator can be the IMD. Then, by means of magnetic Current Sensors (toroid transformers and/or clamp-on probe) associated with an amplifier tuned to the injected current frequency, to trace its path up to the fault (see fig. 26). Finally, another solution is also used, which consists in comparing, constantly and for each feeder, the value of its resistance with a pre-defined or programmable threshold value. This solution, computerised, enables the following actions, both locally and remotely: v reporting of the first fault (IMD),
v then locating of this fault (GFLD) to put it right (curative maintenance) (see fig. 27), v and knowledge of insulation evolution in time, feeder by feeder, to take action on feeders with abnormal insulation drops (predictive maintenance); c surge limiters: these are connected between a live conductor (neutral or phase) of the installation and the earth. Their arcing voltage Ue must therefore be adapted to the assembly planned: thus there are two models for a 50 Hz 230/400 V network: v 250 V for connection to the neutral (400 V < Ue i 750 V), v 400 V, for connection to a phase (700 V < Ue i 1,100 V). Their purpose is twofold: v limit voltage on the LV network on MV/LV disruptive breakdown in the distribution transformer. In this case, the limiter must flow off to earth the residual current of the MV network, v limit lightning overvoltages. This accounts for their characteristics, for example for the 250 V model: - rate voltage: 250 V,
insulation measuring current (Rd) impedance (100 k at 50 Hz ; low in LF) threshold time delay alarm measuring
LF current generator PE
fig. 25: functional diagram of an Insulation Monitoring Device (IMD).
- disruptive breakdown voltage at 50 Hz: min 400 V, max 750 V, - disruptive breakdown voltage according to the 1.2/50 wave: < 1,570 V, - lightning: 20 times 2,500 A (8/20 ms wave): without short-circuiting, 20,000 A / 0.2s, - 50 Hz: 5,000 A / 5 s, 1,200 A / 2 mn. This 50 Hz: peak current withstand is far greater than the value of the residual current of the MV network since a limiter which has been arced during a very high overvoltage may continue to be short-circuited and must therefore be still able to withstand a LV short-circuit current resulting from a first insulation fault in the protected LV network. The limiters marketed under the Merlin Gerin brand can withstand 40 kA/0.2 s.
earth connection must be provided for each zone/consumer. Figure 28, page 20, shows which types of circuit-breaker should be used for
which earthing system. Note that TT and TN can use the same devices (with an additional residual current module in TT).
LF generator PE PE
neutral protection according to the earthing system

The neutral must be broken by a multipole device: c in TT and TN, if neutral cross-section is less than phase cross-section; c in terminal distribution in view of the Neutral/Phase reversal risk. The neutral must be protected and broken: c in IT for intervention of the protection device on the double fault, with one of the faults possibly on the neutral; c in TT and TN-S if neutral cross-section is less than phase cross-section; c for all earthing systems if the installation generates harmonic currents of rank 3 and multiples (especially if neutral cross-section is reduced). In TN-C the neutral, which is also the PE, cannot be broken which is dangerous as a result of its potential variations, due to load currents and insulation fault currents. To prevent risks, a local equipotentiality and an
fig. 26: locating insulation faults by tracing the path of a low frequency current injected at the origin of the installation.
a
PE PE
"locating voltage" bus

The locating current flowing in the conductors is detected by Current Sensor s (CS). Each load comprising a discriminating amplifier (set to the frequency and phase of the locating current) calculates the resistance and capacity of the circuit (with the voltage and phase whose reference it obtains via a bus) and indicates the presence of the fault.
fig. 27: operating principle of an GFLD with LF impedance measurement.
circuits
diagrams TN-C TN-S yes
TT yes
IT no
single phase circuits single phase circuits with one protected pole no two-pole I> circuit-breaker (1 protected pole, N 2 de-energized poles) phase to neutral circuits with two protected poles no
yes
yes
yes
I>
N
I>
two-pole circuit-breaker (with 2 protected poles)
three-phase circuits without neutral with two-pole protection

1 2 3
yes
yes
yes
yes
I> I> I>

no yes yes no
three-pole circuit-breaker
three-phase circuits with neutral without overcurrent detection on neutral 1 I> four-pole 2 I> circuit-breaker with three 3 I> protected poles N 1 2 3 N with overcurrent detection on neutral 1 2 3 N
I> I> I>

three-pole circuit-breaker
yes
yes
yes
no
no
yes
yes
yes
I> I> I> I>

four-pole circuit-breaker with four protected poles
fig. 28: examples of circuit-breakers according to earthing systems
6. choice of earthing system and conclusion
The three earthing systems internationally used and standardised by IEC 364 have as their common objective the quest for optimum dependability. As regards protection of persons, the 3 systems are equivalent if all installation and operating rules are complied with. In view of the characteristics specific to each system, no one system can be preferred over another. Rather, choice of earthing system must result from a concertation between the network user and designer (engineering firm, contractor, etc.) on: c installation characteristics; c operating conditions and requirements.
methods for choosing the earthing system

c firstly do not forget that the three earthing systems can all be included in the same electrical installation: this guarantees the best possible answer to safety and availability needs; c then check that the choice is not specified or stipulated by standards or legislation (decrees, ministerial decisions); c then dialogue with the user to get to know his requirements and resources: v need for continuity of service,
v whether or not there is a maintenance service, v fire hazard. Generally: v continuity of service and maintenance service: the IT will be chosen, v continuity of service and no maintenance service: no fully satisfactory solution: prefer the TT whose discrimination on tripping is easier to implement and which minimises damage with respect to the TN, The installation of additionnal output is easily achieved without the necessity of further calculations. v continuity of service not essential and compent maintenance service: prefer the TN-S (rapid repairs and extensions performed according to rules), v continuity of service not essential and no maintenance service: prefer the TT, v fire hazard: IT if maintenance service and use of 0.5 A RCD or TT. c allow for the special features of network and loads: v very long network or, even more important, leakage current: prefer the TN-S, v use of replacement or standby power supplies: prefer the TT, v loads sensitive to high fault currents (motors): prefer the TT or IT,
v loads with low natural insulation (furnaces) or with large HF filter (large computers): prefer the TN-S, v supply of control and monitoring systems: perfer the IT (continuity of service) or the TT (enhanced equipotentiality of communicating devices).
conclusion
As there is no ideal choice with a single earthing system, it is thus advisable, in many cases, to implement several earthing systems in the same installation. As a rule, a radial network installation, with a clear distinction between priority and non-priority circuits and using standby sources or uninterruptible power supplies, is preferable to an arborescent monolithic installation. The purpose of this Cahier Technique was to perfect your knowledge of earthing systems; we hope it will enable you to optimise the dependability of your installations. Cahier Technique n 173 which provides an insight into use of earthing systems worldwide and their evolution will usefully complete this first document.
7. bibliography
Standards c IEC 241: Fuses for domestic and similar purposes. c IEC 269: Low voltage fuses. c IEC 364: Electrical installation of buildings. c IEC 479: Effects of currents flowing through the human body. c IEC 755: General rules for residual current devices c IEC 947-2: Low voltage switchgear 2nd part: circuit-breakers. c IEC 1008: Residual current operated circuit-breakers without integral overcurrent protection for household and similar uses (RCCB's) c IEC 1009: Residual current operated circuit-breakers with integral overcurrent protection for household and similar uses (RCBO's) c NF C 15-100: Installations lectriques basse tension. French decree of the 14.11.88.
Merlin Gerin's Cahiers Techniques c Earthing of the neutral in a HV industrial network Cahier Technique n 62, F. SAUTRIAU c Residual current devices Cahier Technique n 114, R. CALVAS c Protections des personnes et alimentations statiques sans coupure, Cahier Technique n 129, J.-N. FIORINA c Les perturbations lectriques en BT, Cahier Technique n 141 R. CALVAS c Introduction to dependability design Cahier Technique n 144, P. BONNEFOI c EMC: Electromagnetic compatibility Cahier Technique n 149, F. VAILLANT c Overvoltages and insulation coordination in MV and HV Cahier Technique n 151, D. FULCHIRON c Lightning and HV electrical installations Cahier Technique n 168, B. DE METZ NOBLAT c Earthing systems worldwide and evolutions Cahier Technique n 173, B. LACROIX and R. CALVAS c Connaissances et emploi du SLT neutre isol, Cahier Technique n 178, E. TISON et I. HERITIER (available: end of 1995)
Other publications c Guide de linstallation electrique Ed. France Impression Conseil 1991. c Guide de lingnierie lectrique Ed. ELECTRA 1986. c Electrical Review Nov. 1991 - Oct. 1992. c La protection diffrentielle Cahier Technique J3E - 02/90.
Real.: Illustration Technique - Lyon Edition: DTE - Grenoble 09-95 - 2500 - Printing: Clerc Printed in France
n 173
earthing systems worldwide and evolutions
Bernard Lacroix An ESPCI 74 engineering graduate (from the Ecole Suprieure de Physique et Chimie Industrielle de Paris), he then worked 5 years for Jeumont Schneider, where his activities included development of the TGV chopper. After joining Merlin Gerin in 1981, he was then in turn Sales Engineer for UPS and sales manager for protection of persons. Since 1991 he is in charge of prescription for LV Power distribution. Roland Calvas An ENSERG 1964 engineering graduate (from the Ecole Nationale Suprieure d'Electronique et Radiolectricit de Grenoble) and an Institut d'Administration des Entreprises graduate, he joined Merlin Gerin in 1966. During his professional career, he has been sales manager and then marketing manager for protection of persons. He is currently in charge of technical communication for the groupe Schneider.
E/CT 173, first issued September 1995
glossary
Electrocution Electro Shock EMC GFLD In MT/HVA PIM RCD SCPD STD TBM TEM TPM UL Electro Shock resulting in death Application of a voltage between two parts of the body Electro Magnetic Compatibility Ground Fault Location Device Operating threshold of a RCD Medium Voltage: 1 to 35 kV as in CENELEC (circular of the 27.07.92) Class A High Voltage: 1 to 50 kV as in French decree of the 14.11.88 Permanent Insulation Monitor Residual Current Device Short-Circuit Protection Device (circuit breakers or fuses) Short Time Delay protection (protection against short-circuit overcurrents by circuit-breaker with rapid trip release) Technical Building Management Technical Electrical Power Distribution Management Technical Process Management (automation of...) Conventional limit voltage (maximum acceptable contact voltage) known as the safety voltage
earthing systems worldwide an evolutions
summary
1. Review of standardised earthing systems History p. 4 Emergence of earthing p. 4 systems Earthing systems of IEC 364 p. 7 General p. 9 Influence of MV earthing systems p. 9 LV earthing systems p. 10 Earthing systems of private p. 11 LV networks in some countries Evolution of electrical p. 15 installations Earthing systems and disturbances p. 15 in electronic systems Evolution of earthing systems p. 17 Choosing the earthing system p. 18 p. 21 p. 22 p. 24
2. Earthing systems worldwide
Following an historical review of the origins of Earthing Systems, this Cahier Technique goes on to provide information on the practices in some countries concerning medium voltage, HV/LV substations, in particular in LV public, industrial and tertiary distribution. Electrical installations are evolving, electronics are everywhere, thus leading us to look afresh at earthing systems used in LV and indeed even to predict an evolution which should bring the TN-S and TT systems closer together. The criteria for the selection of earthing systems has changed. We advise those not very familar with earthing systems standardised by IEC 364 to first read Cahier Technique n 172.
3. Evolutions and choices of earthing systems
4. Conclusion Appendix 1: IEC 364 standard Appendix 2: bibliography
1. review of standardised earthing systems
Electrical power was actually used in 1900. Today electrical installation standards are highly developed and cover all major aspects for a safe installation. In LV, the reference standard is IEC 364 (see appendix no. 1). Standard makers have paid particular attention to the measures to be implemented to guarantee protection of personnel and property (part 4 of the above-mentioned standards). This concern has resulted in the standardisation of three Earthing Systems. Before reminding readers of these three systems, a concise historical review will certainly be of use.
history
Electrical hazard and protection of persons c in the 18th century, the static electricity produced by friction of certain insulating bodies formed a scientific diversion causing experimenters to jump up.... in drawing rooms. A few dangerous experiments showed the electrical nature of lightning. And in 1780: by chance an electrostatic machine made a frogs legs move. Galvani observed the contraction of muscles by electricity. c in 1880: in order to transmit electricity over several kilometres, DC voltage left the 100 V range (required for arc lamp operation) and rose to 1,300 V (1882 exhibition in Munich) (see fig. 1), and then to 3,000 V (Grenoble-Vizille link in 1883). Insulation faults cause leaks and short-circuits. The 100 V DC voltage can allegedly be touched without risk. c in 1886: the first distribution installation in the USA: 12 A/500 V/AC generator and 16 small transformers supply consumers with 100 V AC for the first time;
fig. 1: Mr. Desprez's installation located in Munich Palace during the Munich exposition.
c in 1889: AC and DC current wage war in North America: v Edison defends DC and describes the dangers of AC for personnel. He carries out tests on dogs and horses, v Westinghouse supports AC. Edison challenges Westinghouse to a duel: each will be subjected to identical voltages of 100, 150, 200 V etc. in DC for Edison and in AC for Westinghouse...: prediction: at 200 V AC, death will ensue for Westinghouse! The duel did not come off... a telegraph operator climbing on a pole was electroducted in the very heart of New York. c in 1890: Kremler entered the electric chair and was electrocuted with... AC current! Thus, at the end of the 19th century, it was obvious to the technico-scientific
community that electric current was dangerous for man, and that AC was more dangerous than DC.
emergence of earthing systems

These systems are the result of a lengthy evolution guided by the search for increased personnel protection. Between 1880 and 1920, transmission and distribution of electrical power took place in unearthed neutral. Lines are uninsulated, placed out of reach, supported by insulators; no points of the network are deliberately earthed. In homes, voltage is 100/110 V AC. Throughout this period, fuses blow and persons receive Electric Shocks
(see fig. 2). However, in view of distribution voltage level, few persons are electrocuted. c in UK, in the last quarter of 19 th century, electric arc lighting was developing rapidly. When it was introduced into houses, insurance companies became concerned about danger of fire due to undersized cables, poor jointing and insulation breakdown. Many insurance companies produced sets of rules to minimise their risks. In May 1882, the Council of the Society of Telegraph Engineers and of Electricians (later to become the Institution of Electrical Engineers), appointed a committee to consider rules for the prevention of fire risks due to electic light. These rules were not popular with the insurance companies who continued to publish their own. The IEE had yet to become a recognised authority on the subject. By the third edition of the IEE rules in 1897, there was still strong opposition from insurance companies and it was not until 1916 that the final opposition crumbled and the IEE rules became universally accepted in the UK. In the first edition of the rules, in 1882, two items were concerned with danger to people: no one should be exposed to more than 60 V and the potential between two points in the same room should not exceed 200 V. The earthing of metalwork of appliances working at domestic voltages was first required in the eighth edition in 1924, althought it was soon recognised that an adequate earth was not always easy to obtain. In 1930, the requirement for an earth leakage trip operating at 30 mA or less was introduced (since deleted). c in France in 1923 a standard for electrical installations makes earthing of frames a requirement: v casings of fixed and moving motors, which may be touched in a noninsulated area, in installations with a voltage greater than 150 V, v fixed and portable electrical household appliances with a power greater than 4 kW, v electrical bath heater enclosures installed in bathrooms, v metal parts placed in premises steeped in conductive liquids and which, due to insulation faults, might become live.
1st fault nothing happens
double fault the fuse blows if full fault
Earthing of load frames (1923) to prevent Electric Shock by indirect contacts
fig. 2: the emergence of the unearthed neutral.
The standard provides absolutely no information on earthing conditions or on the value of earth connection resistance, and stipulates no protection device. Although it contains a few rules for fuses these are only for installation conditions. In order to prevent fuses blowing on a double insulation fault, it quickly become obvious that indication of the presence of the first fault was a good idea. For this reason, the first failsafe insulation monitor was installed in industrial installations (see fig. 3).
A lamp going out indicates an insulation fault on the corresponding phase.
fig. 3: lamp insulation monitor in industry.
If a lamp goes out, there is a fault between the corresponding phase and the earth. Thus the first earthing system came into existence: the unearthed neutral. The Permanent Insulation Monitor (PIM), with three lamps (in three-phase) was used up to 1955. In 1951, the first tube PIMs, injecting DC, were installed in mines: insulation of phases and neutral was monitored. In 1962 the first transistor PIMs (Vigilohm TA) were produced, and in 1972 the first PIMs injecting low frequency AC. In 1927 a decree stipulated the earthing of the transformer neutral in public distribution in France (U u 150 V AC). At this time, production of electricity in France was approximately 350 kWh/ inhabitant/year (in 1900 it was 7); a tenth of this production was distributed in LV. Electricity firms supply a number of consumers by transformer. However, in unearthed neutral, two earthing faults occurring at two different consumers, do not always cause fuses to blow and there is a definite risk of fire (the indirect contact risk exists, but is low and not known). Application of the 1927 decree thus stipulates more reliable disconnection of the faulty consumer, thereby ensuring a sound network is maintained. In 1935, the decree on protection of workers and standard C 310, taken up by standard C 11 of 1946, began to mention the risk inherent in insulation faults. It is at this moment that the combination of earthing of loads and automatic breaking devices first appeared. The latter may be fuses, RCDs or voltmeter relays of frame/ earth voltage (see fig. 4). Note that protection devices with a threshold of under 30 A are supposed to guarantee safety! The first residual current connection circuit-breakers were manufactured in 1954. In addition to protection of
persons and disconnection of consumers, they made it possible to combat illegal connections (current stealing between phase and earth when 127 A single-phase moves to 220 V two-phase) (a single current measuring winding in the meter). This is how the earthed neutral came into existence in France. However, it was not until the decree of the 14.11.62 on protection of workers and standard C 15-100 (blue) of the 28.11.62 that fault loop impedance and thus earth connections were defined accurately, according to fuse rating or RCD threshold then set by standard C 62-410 at 450 200 mA. Standard C 15-100 of 1962 thus gave official status to the unearthed neutral and the earthed neutral (measurement B1), as well as to the TN system (measurement B3). It made a clear distinction between direct and indirect contact and lists the primary protection measurements (A)
and how to protect by automatic disconnecting devices (B), without however giving information on operating times. Alongside the standard, the decree of the 14.11.62 legalised the unearthed and the earthed neutrals. In 1973 a decision of the Board of Employment authorised the TN system in France. Between 1962 and 1973 each earthing system had its ardent supporters in France and elsewhere. The TN system has the advantage of a simple principle: the SCPDs de-energise loads (or LV consumers) having an insulation fault. The TN system (exposed-conductive parts connected to neutral) is used in some countries in public distribution (not in France): (see fig. 5). As personnel protection against indirect contact is involved, use of this system requires complete mastery of loop
Ph SCPD or RCD SCPD or RCD
a
N
fig. 4: earthing in public distribution.
Ph
SCPD
SCPD
a
N
fig. 5: TN-C system in public distribution.
impedances (irrespective of where the fault occurs) to ensure operation of the SCPD which will disconnect the faulty part within the specified time. Definition of these times by IEC experts in the nineteen seventies, according to impedance of the human body and pathophysiological effects, has made its use possible. It should be noted that transformation of an insulation fault into a short-circuit increases risk of damage to equipment and of fire. With this in mind, let us remember that protection is based on the rapid evolution of an insulation fault to a full fault between phase and neutral.
TN-C system
PEN
fault
earthing systems of IEC 364

The three earthing systems internationally standardised are currently taken over in many national standards. These three systems have been studied in detail in Cahier Technique n 172 and, for each of them, the hazards and associated protection switchgear have been presented. We shall however briefly review their protection principle. The TN system (see fig. 6) c the transformer neutral is earthed; c the frames of the electrical loads are connected to the neutral. The insulation fault turns into a short-circuit and the faulty part is disconnected by Short-Circuit Protection Devices (SCPD). The fault voltage (deep earth/frame), known as indirect contact is Uo/2 if the impedance of the outgoing circuit is equal to that of the return one. When it exceeds the conventional limit voltage (UL), which is normally 50 V, it requires disconnection, which must be especially quick when Ud is larger than UL. The TT system (see fig. 7) c the transformer neutral is earthed; c the frames of the electrical loads are also connected to an earth connection.
TN-S system
N PE
fault
fig. 6: TN-C and TN-S systems.
N PE
fault
RB
RA
fig. 7: TT-system.
The insulation fault current is limited by the impedance of the earth connections and the faulty part is disconnected by a Residual Current Device (RCD). The fault voltage is:
Uc = Uo RA greater than RB + RA UL . RA
c the frames of the electrical loads are connected to the earth. If an insulation fault occurs, a small current is developed due to the networks stray capacities (see first diagram, fig. 8). The voltage developed in the earth connection of the frames (a few volts at the most) does not present a risk. If a second fault occurs (see second diagram, fig. 8) and the first one has not yet been eliminated, a short-circuit appears and the SCPDs must provide the necessary protection. The frames of the relevant loads are brought to the potential developed by the fault current in their protective conductor (PE).
voltage UL, the RCD comes into action as soon as Id u
The IT system c the transformer neutral is not earthed. It is theoretically unearthed, but in fact is connected to the earth by the stray capacities of the network and/or by a high impedance 1,500 (impedanceearthed neutral).
1st fault
RA
double insulation fault
PE
fig. 8: IT system.
2. earthing systems worldwide
general
In all industrialised countries, LV networks and loads are earthed for safety reasons to guarantee protection against electric current for persons. The objectives are always the same: c fixing the potential of live conductors with respect to the earth in normal operation; c limiting voltage between the frames of electrical equipment and the earth should an insulation fault occur; c implementing protection devices which remove the risk of Electric Shocks or electrocution of personne; c limiting rises in potential due to MV faults.
c increased continuity of service (availability of electrical power) by authorising automatic reconnection on a transient fault, c connection or not of the frames of the MV/LV substation and those of the LV neutral to avoid risk for LV users and equipment. IEC 364-4-442 states that the earthing system in a MV/LV substation must be such that the LV installation is not subjected to an earthing voltage of: c Uo + 250 V: more than 5 s; c Uo + 1,200 V: less than 5 s (Uo 3 in IT). This means that the various devices connected to the LV network must be able to withstand this constraint (see fig. 9a).
The same standard states that if Rp > 1 , the voltage Rp IhMT must be eliminated, for example: c in under 500 ms for 100 V; c in under 100 ms for 500 V. If this is not so, Rp and RN must be separate whatever the LV earthing system. This rule, not always complied with in certain countries, often leads to the separation of the two earth connections (for MV networks with a high zero sequence fault current). If all the earth connections (substationneutral-applications) have been grouped into a single one, a rise in potential of LV frames may be observed which can be dangerous (see fig. 9b).
MV LV
influence of MV earthing systems

While the first three objectives listed above fall into the range of LV earthing systems, the fourth has considerable repercussions on safety of personnel and property in LV. Thus, at MV/LV substation level, a MV phase/ frame fault or a fault between MV and LV windings may present a risk for equipment and users of the LV network. In public and industrial MV, except in certain special cases, the neutral is not distributed and there is no protective conductor (PE) between substations or between the MV load and substation. A phase/earth fault thus results in a single-phase short-circuit current limited by earth connection resistance and the presence of limitation impedances, if any (zero sequence generator). The current tendency, in various countries, is to limit the zero sequence fault currents of MV networks, thus allowing:
HV
MV
IhMT
RP
RB
fig. 9a: if Rp and RB are connected, the fault current causes the potential of the LV network to rise with respect to the earth.
MV
LV
IhMT
RT (RPBA)
fig. 9b: the LV load frames are raised to the potential IhMT RT.
The table in figure 10 gives a few examples for public distribution worldwide. It shows that, in many countries, the earth connections of the substation and neutral must be separate if their resulting value is not less than 1 . Note that the impedance-earthed IT earthing system is the most commonly used in MV industrial networks. The zero sequence generator supplies a resistive current around twice the capacitive current of the network (see Cahier Technique n 62), thus allowing use of a RCD to ensure protection by disconnection of the faulty feeder.
LV earthing systems
The MV/LV transformers used are generally Dy 11 (delta/star). However the use of midpoint single-phase distribution for public distribution in the USA and Japan should be pointed out (see fig. 11). Most countries apply or derive inspiration from standard IEC 364 which defines the TN, IT and TT earthing systems and the protection conditions, both for public and private distribution. In public distribution The most common systems are TT and TN; a few countries, in particular Norway, use the IT system. The table in figure 12 lists some examples for public distribution (LV consumers). This table shows that Anglo-Saxon countries mainly use the TN-C, whereas the TT is used in the rest of the world.
MV earthing system Germany unearthed or compensed 10 and 20 kV Id < 60 A Australia directly earthed 11 and 12 kV Id= a few kA Belgium limitation impedance 6.3 and 11 kV Id < 500 A France limitation 20 kV impedance overhead Id i 300 A underground Id i 1,000 A Great Britain direct or limitation 11 kV impedance Id < 1,000 A Italy unearthed 10-15 and 20 kV Id i 60 A (more in reality) Ireland unearthed on 10 kV 10 and 38 kV compensated on 38 kV Id < 10 A Japan unearthed 6.6 kV Id < 20 A Portugal limitation 10 to 30 kV impedance overhead Id i 300 A underground Id i 1,000 A USA directly earthed 4 to 25 kV or by low impedance Id = a few kA
country
frame connection connected if Id x RT < 250 V separated except if RT < 1 separated d u 15 m separated except if RT <3 <1 separated except if RT < 1 separated
observations Rp < 2 or 5 Rp < 10 Rp < 5
Rp < 30 Rp < 1 Rp < 25 Rp < 20
separated except if RT < 10 connected RT < 65 separated except if RT < 1 connected
stipulations on how to produce Rp Rp < 20
the earths of the source of the MV/LV substation and of the LV neutral are connected
fig. 10: public distribution examples - MV earthing systems.
country Germany 230/400 V Belgium 230/400 V Spain 230/400 V France 230/400 V Great Britain 240/415 V
a) star three-phase
b) midpoint single-phase Italy 230/400 V Japan 100/200 V Norway 230/400 V
Portugal USA 120/240
LV earthing observations system TT and TN-C the TN is the most commonly used; RT must be < 2 ; earth connection at the consumer's, even in TN TT Ru < 100 30 mA RCD for sockets TT Ru < 800 with 30 mA RCD at supply end of the installation TT Ru < 50 , (100 shortly) 30 mA RCD for sockets TT and TN-C - town areas: TN-S and TN-C (New Est installations: 15 %), the earth connection (< 10 ) of the neutral is provided by the distributor - rural areas: TT TT RCD with In as a function of Ru (In < 50/Ru). For consumers without earth connection 30 mA RCD TT Ru < 100 , frequent use of 30 mA RCD, no search for equipotentiality IT premises in insulating materials and poor earth connections account for this choice. homes with signalling 30 mA RCD. tripping of connection circuit breaker if 2 faults. TT Ru < 50 (100 as from 1995). TN-C earthing of neutral at LV consumers (all earth connections are connected to the source substation).
fig. 11: coupling of the secondary windings of MV/LV transformers.
fig. 12: public distribution examples worldwide (LV consumers) - LV earthing systems.
The TN-C requires costly equipotentiality research: c for the distributor: v in the USA, installation of an additional conductor throughout MV and LV distribution with earthing every 400 m, v in Great Britain, numerous earth connections are installed on the town public LV network neutral, if TN-C used, thus meaning consumers may not need their own earth connections, v in Germany, an earth connection is made for the neutral just upstream from the consumers connection point, c for the consumer: as a rule, connection to the protection conductor of the metal structures of the building and all the metal ducts. In industrial and tertiary LV distribution The 3 earthing systems are used to various degrees in all countries: c the TN-C system is particularly used in Anglo-Saxon countries for carefully designed and produced installations (SCPD/loop impedance matching) corresponding to modern blocks of flats where all metal parts are connected to the protective conductor and where explosion and fire risks are extremely low. It is not currently recommended in premises equipped with communicating electronic systems (computer TBM, TPM and CTM -Centralised Technical Management- networks) as currents in the neutral and thus in the PE cause potential references to vary. Do not forget that the TN-C can no longer be used when cross-section of live conductors is < 10 mm2 Cu. c the TN-C system is also used in Anglo-Saxon countries and requires an additional conductor and careful design and production. However its use is more flexible and RCDs are used for personnel protection (for very long cables), for fire protection and for extensions without loop impedance calculation. Nonetheless, the insulation fault currents which are short-circuit currents, may, if the PE is connected, in distribution, to metal structures, generate electromagnetic disturbances preventing electronic equipment from working properly (sum of currents in cable not zero and stray currents).
Finally, as the neutral is not protected in some countries (measure authorised by standard IEC 364), it may be damaged by overcurrents, in particular when loads generating rank 3 harmonic currents and multiples are supplied by the network. In some cases this results in neutral cross-section being doubled (seen in the USA).... In the long term, international standards should specify systematic protection of the neutral and indeed protection (without breaking) of the PEN in TN-C. c the IT system requires as much care as the TN-S one. Permanent insulation monitoring allows fault prediction, currently simplifed by digital systems which monitor insulation evolution for each feeder. This system calls for fault tracking and elimination; thus electrical engineers should preferably be present on site. In many countries, the unearthed neutral is used whenever continuity of service is essential or when human life is at stake (e.g. hospitals), however, in UK, TN-S is used in hospitals. c the TT system is the easiest one to implement; insulation fault currents are 1,000 times smaller than in TN or IT (2nd fault), thus accounting for its value as regards risk of fire, explosion, material damage and electromagnetic disturbances. Its weak point is the risk of disruptive breakdown by return on an insulation fault in the substation on the MV side if the zero sequence fault current is high and if the substation and neutral frames are connected. There are no statistics on use of earthing systems worldwide, but the
TT system is by far the most commonly used. Moreover, its simplicity makes it without doubt the system best suited to developing countries. Following these brief considerations on the three official earthing systems, we shall now see how they are implemented in some countries.
earthing systems of private LV networks in some countries

In the USA All the various earthing systems are used: the TN-S (see fig. 13) is the most common, but the IT and impedanceearthed IT are used in process factories. c implementation of the TN-S has two major special features: v the neutral is not protected and not switched, which may present risks for persons and property: - potential of the neutral compared with the earth may be high on a fault originating at MV level; this presents a risk - rank 3 harmonic currents and multiples of 3 add up in the neutral and may cause unacceptable temperature rise v the protective conductor is often made up by the cable path and the metal tubes conveying the live conductors: - the impedance of this PE is hard to control. Thus NEC paragraph 230-95 (National Electrical Code) considers that SCPDs do not always guarantee safety in the event of an insulation fault,
load
fig. 13: TN-S earthing system in the USA.
- as the PE is not mechanically connected to the faulty live conductor (cables on cable path acting as a PE), the electrodynamic forces due to the high fault current separate the cable from its support (American LV networks have very high prospective fault currents). This causes a transient fault with, as a result, a risk of non-operation of the SCPDs and an increase in fire risk. Note that when the PE is a distributed conductor, earthing of the transformer neutral is sometimes performed through a low impedance, in order to limit the I2t at the fault point (Id i than 1,000 A). c protection devices used in TN-S In addition to the use of SCPDs, remember that to the American way of thinking the main purpose of the earth protection devices used is protection of property and limitation of the fire risk. In this field, the NEC imposes minimum requirements, i.e. use of residual current protection devices on LV installations when the following 3 conditions are met: - neutral directly earthed, - phase-to-neutral voltage greater than 150 V and less than 600 V, - nominal current strength of the device at the supply end greater than 1,000 V, v implementing the RCDs This protection can be performed in three ways: - Residual Sensing (detection of residual current by vectorial addition of currents in live conductors), (see fig.14). This assembly, known as Nicholsons, requires the installation of a current transformer on the neutral; in the USA the neutral is neither switched nor protected. - Source Ground Return (residual current device placed in the neutralearth link) and usable only at the supply end of the installation. It allows parallelconnection of sources (see fig. 15), - Zero sequence (classical RCD). For low current detection, it can be used at various levels of the installation to form discriminating protection (see fig. 16),
earth leakage circuit-breaker N
fig. 14: residual sensing.
fig. 15: source ground return.
fig. 16: zero sequence.
v incoming device settings: - max threshold: 1,200 A, - tripping time: must not exceed 1 s for a fault current of 3,000 A. Note The NEC does not specify a tripping time at 1,200 A, but it is common practice to install protection devices with lower thresholds and as instantaneous as possible. c discrimination of residual current devices. The NEC 230 paragraph 95 only stipulates earth protection for the incoming device. Naturally, this protection must also be installed downstream to prevent the entire installation being put out of operation in the event of one earth fault. Discrimination has then also to be achieved between the various protection devices. This problem can be solved in two ways: v between the earth protection devices with discrimination of the: - time type by a time delay of 0 s to 1 s, - logic type or Zone Selective Interlocking; recommended in the USA, it prevents long time delays (reduction of the passing I2t) and easily achieves discrimination on 3 levels or more, v between the earth and thermalmagnetic protection Discrimination will be determined by comparing the setting threshold of the upstream earth protection with the curve I = f(t) of the thermal-magnetic setting of the downstream protection device (see fig. 17). Use of fewer earth protection devices in subdistribution switchboards leads to higher settings of protection devices, resulting in increased risk of damage. In the Republic of South Africa In the RSA, industrial and tertiary electrical installations comply with IEC standards. All three earthing systems are used, with a preference for the TN-S. A country with a long mining tradition, the RSA uses in its gold mines, for
a)
insulation fault protection R 3 D1 M25 + GFP
R 3
D2 M10 + GFP
short-circuit protection
D3 C161
b)
ts D3 D2 STR38 D1 STR58
T1200 A - 0.2 s T1000 A - 0.1 s
5 0.2 0.1
875 1,000 1,200 2,500
IpA
fig. 17: discrimination between downstream short-circuit protection (D3) and upstream earth protection (GFP) (orange).
example, an earthing system which is half TN-S and half TT (see fig. 18). It uses RCDs for protection. Characteristics of this earthing system: c the protective conductor is distributed; c the load frames are connected to the PE which is earthed at MV/LV substation level; c a resistor placed between the transformer neutral and the earth connection limits the insulation fault current to less than 20 A. This system has both advantages and drawbacks: c advantages: v a low contact voltage despite use of a network voltage of 525/900 V,
Uo RPE RPE + RPh + 27
c additional measure Limitation resistance is monitored by an ohmic relay: v if resistance is broken: the earthing system becomes an IT: operation can continue, but the RCD will open a feeder on a double fault, v if resistance is short-circuited, the earthing system becomes TN-S and the first insulation fault causes the BT circuit-breaker to open, unless, of course, the electrical maintenance service has acted in time. Compared with the classical TT and TN-S systems, this earthing system is to be preferred when Uo voltage is greater than 400 V (as is the case in mines), since it limits contact voltage. The desire to limit insulation fault currents is fairly widespread for a variety of reasons: c high short-circuit power: USA; c uncertain loop impedance: mines, worksites; c limitation of damage and/or fire hazard: process - mines petrochemistry (note that British
Petroleum (BP) produces all its installations worldwide using the impedance-earthed TN-S (see fig. 18) with a resistance of 3 in LV and of 30 in 3.2 kV). In China China is waking up! However, it has long been under the technical influence of the USSR, which is a member of the IEC (Russian is one of the IECs official languages together with English and French). Consequently, all three earthing systems are known and are used to varying degrees. c IT is used when continuity of service is vital and there is a real risk for persons (hospitals); c TT used in public distribution, is also used in industry and the tertiary sector, but increasingly less so, perhaps due to the rare use of time discrimination; c TN-C, which originated in the USSR, has completely gone out of use; c TN-S is increasingly chosen by Design Institutes for large projects.
v a low fault current, thus considerably reducing risk of fire and damage to faulty loads, v discriminating protection by RCD with use of time discrimination. Note that use of RCD is particularly advantageous since the LV network topology is constantly evolving (loop impedance!). c drawbacks: In the event of HV/LV disruptive breakdown in the transformer, there is a risk of rise in potential of the live LV network conductors compared with the earth and frames (IhMT R): this risk can be reduced by use of a surge limiter. Moreover, a residual current device placed on the neutral/earth circuit causes the MV circuit-breaker to open immediately if it detects a fault current greater than 20 A.
525/900 V N
R = 27
fig. 18: earthing system used in RSA (mines).
3. evolutions and choices of earthing systems

evolution of electrical installations
In 1960 the tertiary sector had barely started to develop: plants, normally large in size, were often installed next to source substations. The main concern of companies was operation of processes; boasting a competent electrical service, some of them would be won over by the unearthed neutral. Little by little, the safety guaranteed by this system led it to be stipulated in tertiary installations where dependability was of prime importance: e.g. hospitals. In the 1990s, electrical power is the universal driving force in homes, tertiary and industry. Although public distribution has made enormous headway in terms of availability of electrical power, this availability is still not always sufficient, and generator sets and uninterruptible power supplies are thus used. c the housing sector no longer accepts power cuts; c tertiary is a major computer consumer; c industry has set up in rural areas, is a major automation system consumer and is increasingly using static converters; for example, motors are controlled by a speed controller and functionally linked to a PLC. In all buildings, intelligent devices are increasingly being controlled by technical management systems (process - electrical distribution building utilities). These digital systems, including distributed computing, nowadays require the problem-free joint existence of high and low currents; in other words, electromagnetic compatibility (EMC) is vital. A clash of technical cultures is inevitable: c electrical engineers have problems with the harmonics generated by static converters. These harmonics cause temperature rises in transformers, destruction of capacitors and abnormal currents in the neutral; c electronic engineers place filters upstream of their products, which do not always withstand overvoltages and lower network insulation; c lamp manufacturers are unaware of the problems caused by energising inrush currents, harmonics and high frequencies generated by certain electronic ballasts; c computer engineers (same applies to designers of distributed intelligence systems) are concerned with equipotentiality of frames and conducted and radiated interference. These specialists sometimes have problems understanding one another and do not necessarily all have the same approach. Also, very few of them are familiar with earthing systems and their advantages and drawbacks faced with the evolution in the techniques described above. Readers wishing to improve their knowledge in this area should study the following Cahiers Techniques: c n 149 - EMC: Electromagnetic compatibility; c n 141 - Les perturbations lectriques en BT; c n 177 - Perturbations des systmes lectroniques et schmas des liaisons la terre. This section will only review the most important aspects, without describing earthing system behaviour faced with MV (50 Hz) faults. Faced with harmonics The TN-C should be avoided since rank 3 harmonics and multiples of 3 flow in the PEN (added to neutral current) and prevent the latter from being used as a potential reference for communicating electronic systems (distributed intelligence systems). Moreover, if the PEN is connected to metal structures, both these and the electric cables become sources of electromagnetic disturbance. Note The TNC-S (TN-S downstream from a TN-C should also be avoided even though risks are smaller). Faced with fault currents c short-circuits: avoid separating the live conductors; otherwise the Icc creates an electromagnetic pulse in the resulting loop; c electrical earthing fault: the PE must follow the live conductors as closely as possible, or, better still, be in the same multi-conductor cable. Otherwise, as above, the transmitting loop effect appears. The higher the fault current, the greater this effect. The TT earthing system will thus be preferred, as the TN and IT (2nd fault) can develop currents a 1,000 times greater.
earthing systems and disturbances in electronic systems

Electromagnetic disturbances assume many different forms, namely: c continuous or occasional; c high or low frequency; c conducted or radiated; c common or differential mode; c internal or external to the LV network. Choice of earthing system is not a neutral one as regards: c sensitivity to disturbances; c generation of disturbances; c effects on low current systems.
In TN and IT, do not connect the PE to the metal frames of the building as the return currents may take a variety of paths and turn into transmitting antennae. The same also applies to the power cable, incorporating the PE, in which the sum of currents is no longer zero. With respect to equipotentiality of frames, the TN and IT (on the 2nd fault) are equivalent since frame potential at the fault point suddenly rises to Uo/2 whereas it remains at 0 V at the origin of the installation. This leads to certain specialists specifying in TN and IT the creation of a low current frame circuit separated from the earth circuit (PE), both being connected to the earth connection at the origin of the LV installation. The TT with distributed PE throughout the installation is the best system in this respect (small Id and same potential reference for all the communicating devices), (see fig. 19). Faced with lightning and operating overvoltages These overvoltages, of common or differential mode and with a frequency of 1 kHz to 1 MHz can damage certain electronic devices if they are not fitted with an isolating transformer with a small primary/secondary capacitive coupling. As regards differential mode overvoltages, all the earthing systems are equivalent. The solution consists in: c implementing surge reducing at disturbance source level (e.g. RC on contactor coil); c protecting sensitive equipment by installing a surge limiter (varistor, ZnO lightning arrester) directly at their terminals. As regards common mode overvoltages (lightning), ZnO lightning arresters should be installed at the origin of the LV installation with the shortest possible earth connections. In this case, although the TN and TT earthing systems may seem more suitable than IT but overvoltages are
transmitted on LV phases. In actual fact, at the frequencies considered, the phase/neutral impedance of the LV windings is very high (the phases are as though they were unearthed even if the neutral is earthed). Faced with HF disturbances: All the earthing systems are equivalent. Advice for minimising the effects of HF disturbances: c use the Faraday cage effect for buildings (metal structures and meshed floors), or for certain rooms in the building reserved for sensitive equipment, c separate the frame network (structural and functional frames) from the earth network (PE),
c avoid loops which may be formed by the high and low current circuits of communicating devices or place low current links (frame surfaces - ducts/metal screens accompanying frames) under a reduction effect, c avoid running them too close to power cables and make them cross at 90; c use twisted cables, or, even better, shielded twisted cables. There are still not many standards in this area and they are often prepared (EMC standards) by electronic engineers. Installation standard IEC 364, sections 444 and 548, should provide increasingly more recommendations.
a)
V
PE
digital link
In TN: on an insulation fault, the voltage drop in the PE causes the reference potential of the communicating devices to vary. Uo where as the devices close to the The frames of devices 2, 3..., are at the potential 2 source are at the earth potential. b)
PE
digital link
With a single load frame earth connection: In TT: all the frames are at the same potential, even during a fault; no disturbance of communictions by bus.
fig. 19: equipotentiality of the PE on an insulation fault.
evolution of earthing systems

Evolution of the TN The original aim of this earthing system was simplicity, efficiency and minimum installation cost (see the American TN where the neutral is not even protected); Safety of personnel is guaranteed, but that of property (fire, damage to electrical equipment) is less so. Proliferation of low current power electronics is increasing and will continue to increase complexity of its implementation. Derived from the TT of the nineteen twenties, the TN was a solution for controlling fault current value and ensuring that all insulation faults could be eliminated by a SCPD. It grew up in Ango-Saxon countries where rigour of installation designers and users is excellent. The logical evolution is TN-C TN-C-S TN-S TN-S with fault current limitation to limit fire hazards, damage to loads and malfunctionings due to widespread use of distributed electronics (see fig. 20). A survey carried out in Germany in 1990 showed that 28 % of electrical (electronic) problems were due to EMC. In terms of protection, the TN system often uses fuses; already hindered by an overlong breaking time when limit safety voltage UL is 25 V, they will be further hindered in the long term if LV networks with voltages greater than 230/400 V are developed. The use of RCDs (impedance-earthed TN-S) solves this problem. Evolution of the IT The earliest electrical installations (1920) were produced in IT. However, double faults quickly gave this system a bad name (failure to master loop impedances). Standards gave it official status in the sixties in order to meet continuity of service requirements of process industries and safety requirements in mines.
a) TN-C earthing system

3 MV LV Ph PEN
b) TN-C-S earthing system

3 MV LV Ph N PE (2) (1)
(1) new earth connection preferable if the transformer is at a distance (public distribution); improves local equipotentiality compared with the eath. This solution is used in Germany and is being experimented in France (in DP). (2) in France, the C 15-100 stipulates changing to TN-S when cross-section of conductors is i 10 mm2 Cu. c) TN-S earthing system
3 MV LV Ph N PE
Avoids equipotentiality disturbances due to flow of neutral current and 3K harmonics in the PEN. d) impedance-earthed TN-S
RCD 3 MV LV Ph N PE
Solution used in the USA (Id of the order of 500 A), in RSA (I 20 A); limits fire risks, damage and potential reference problems for distributed electronics. This earthing system is similar to the TT one.
fig. 20: evolution of the TN.
Today, the IT system closely resembles the TN-S as regards installation (an additional surge limiter and insulation monitor). It is the champion of continuity of service and safety on the first fault, if this fault is promptly tracked and eliminated. Following widespread use of the distributed PE throughout the installation (as in TN), this system, in which the second fault current cannot be limited, will not really evolve, except for the rapid fault tracking techniques. As the likelihood of a double fault increases with the number of feeders and size of the installation, its use should be reserved for parts of the network and for control and monitoring circuits with, naturally, use of isolating transformers (see fig. 21). On these small circuits, use of the impedance-earthed IT allows signalling RCDs for fault tracking. Evolution of the TT To begin with, electrical distribution in France was in single-phase 110 V, followed by two-phase 220 V. Earthing of frames, combined with use of RCDs, aimed at de-energising consumers with insulation faults and cheaters. The development of electric household appliances led to protect people against indirect contacts. Protection against indirect contacts by RCD with standardised operating times was made official in the nineteen sixties. Today, the tendency is (as in TN and IT) to distribute the PE throughout the installation and thus to use only one application earth connection (see fig. 22). This tendency should continue with the use of the LV neutral earth connection only (as in TN and IT), but maintaining the advantage (damage, fire, EMC) of a small insulation fault current.
a) at the outset
3 M MV LV Ph N
b) in 1960
MV
LV
Ph N PE CPI
limiter
Limitation of number of earth connections and interconnection of frames or use of RCD to master the double fault. c) in 1990
MV
LV
Ph N PE CPI
limiter
Becoming more similar to the TN-S (PE distributed, calculation of loop impedances). d) 2000
MV
LV
choosing the earthing system

Choice may be determined by normal practice in the country. Choice of earthing system should be influenced by electrical power users and network operators (electrical service). Experience shows however that the choice is mainly made by the engineering firms designing the installation.
TN-S or TT
IT
IT is used mainly on small networks or parts of networks downstream from TN and TT systems.
fig. 21: evolution of the IT.
For users and operators These both demand absolute DEPENDABILITY; electrical power should thus always be available and be completely risk-free, i.e. out of sight, out of mind. The elements making up installation dependability: c safety; c availability; c reliability; c maintenability, must therefore be optimised. In addition, a new requirement, electricity must not disturb the numerous low current devices. These are the criteria used to make the best choice according to: c type of building; c the activity it houses; c whether or not an electrical service is available. In safety terms, the TT is the best, In availability terms, the IT is the most suitable, In maintenability terms, fault tracking is fast in TN (thanks to the SCPD) but repair time is often long. Conversely, in IT, tracking of the first fault may be more difficult, but repairs are quicker and less costly. The TT is a good compromise. In reliability terms, the protection devices used are reliable, but reliability of the installation and loads may be affected: c in TN-C by the fact that the PEN, not protected, may be damaged by harmonic currents; c in TN-C and TN-S; v by insufficient rigour for extensions, v by use of replacement sources with low short-circuit power, v by the effects of electrodynamic forces; c in IT, on a double fault, the risks inherent in TN described above also exist. However if tracking and elimination of the 1st fault are rapid, installation reliability is excellent. c in TT, by disruptive breakdown by return of the loads due to a fault in the HV/LV transformers. However the likelihood of this fault occurring is small and preventive solutions are available, e.g. use of surge arresters between one of the live conductors and the load earth connection. In disturbance terms, the TT is to be preferred to the TN-S whose high fault currents may be the source of disturbance.
a) at the outset
RCD 3 HV LV Ph N
b) in 1960
RCD LV
HV
Ph N PE RCD RCD
Multiple RCDs with time discrimination, local equipotentialites and minimum number of earth connections. c) in 1990
HV
LV
Ph N PE
Same use of RCDs. PE distributed as in TN-S and IT. In some installations, the two earth connections are connected... it is TN-S without impedance calculation as RCDs are used. d) 2000
CDR RCD HV LV
To retain the advantage of the small fault current (damage and EMC), an impedance-earthed TT (r 12 /Id = 20 A) emerges with a single earth connection. This system requires the use of a surge limiter if the MV zero sequence current exceeds 80 A - DDRs are used in the same way (time discrimination)
fig. 22: evolution of the TT.
The table in figure 23 reviews the strong and weak points of each earthing system: For installation designers Designing is simpler in TT, the same for extensions (no calculations). Designing complexity is equivalent in TN-S and IT. As regards costs: c the TN-S is the least costly to install, for example if the neutral is neither protected nor switched. But be warned: the cost of curative maintenance can be high; c the IT is slightly more costly to install (insulation monitoring and insulation fault tracking devices). Search for maximum availability of electrical power requires the presence of an electrical engineer, whose action will minimise curative maintenance; c the TT, if enough discriminating RCDs are installed, is slightly more costly to install than the IT, but fault tracking is simple and curative maintenance less costly than in TN. In terms of complete cost over 10 to 20 years, all three earthing systems are equivalent. The right choice In a certain number of countries, for some buildings or parts of a building, the choice is laid down by legislations or standards, e.g. for hospitals, schools, navy, worksites, mines, etc. In other cases, certain earthing systems are strictly prohibited, for example the TN-C in premises with explosion risks. Apart from these compulsory choices, the DEPENDABILITY objectives (safety, availability, reliability, maintenability and proper operation of low current communicating systems) are those which should determine which earthing system is chosen for a specific building type. The degree of development of the country should also be taken into consideration, as should be national practices, climate.... If we plot an axis from North to South, as regards public distribution, we find the IT earthing system in Norway, TN-C in Germany, TT in France and in most African countries. In temperate, industrialised countries, all three earthing systems are used in private installations. Finally, it should be noted that it is possible and even advisable to mix the earthing systems (see fig. 24).
safety c of persons c fire c explosions availability(further to 1 fault) maintenability reliability of the installation disturbances c radiation transmission EM c equipotentiality of PE (1) : 1st insulation fault. (2) : 2nd fault.
TN-C TN-S + --+ + +
TT + + + +
IT(1) ++ ++ ++ ++
IT(2) +
Observations Uc # 0 on 1st fault (IT) TN-C not recommended TN-C strictly prohibited depends on discrimination of the SCPDs or RCDs (easier to implement) the IT authorises preventive and even predictive maintenance advantage for small Ids (damage- electrodynamic forces) advantage for small Id pay attention to harmonics in TN-C
+ ++
++ ++
--
+ ++
++ +
fig. 23: comparing the earthing system.
a) "series" association of earthing systems
3 HV LV PEN N PE TN-S TT IT
TN-C
b) "parallel" association of earthing systems TN-S - lighting - heating - computing centre
3 HV LV N PE TN-S
TT PE
- machines - communicating systems . automation . office . TBM - premises with fire hazards - safety systems - medical facilities - industrial process
IT
fig. 24: several earthing system included in the same LV installation.
4. conclusion
The three earthing systems (TN - IT - TT) and their implementation are clearly defined in installation standards (IEC 364). Their respective use varies from country to country: c mainly TN in Anglo-Saxon countries; c TT often used in the other countries; c IT used when safety of persons and property, and continuity of service are essential. All three systems are considered to guarantee personnel protection. Two major changes have had a considerable effect on choice of earthing systems: c search for optimum continuity of service;
c proliferation of high current (disturbers) and low current (disturbed) electronic devices, which are increasingly set up in communicating systems. Thus the general tendency for earthing systems, in both MV and LV, is to limit insulation fault currents. At present, the fault currents of traditional LV earthing systems have the following standard values: c IT (1st fault): Id < 1 A; c TT: Id 20 A; c TN: Id 20 kA; c IT (2nd fault): Id 20 kA. Limiting fault currents: c simplifies maintenability of the electrical installation, thus increasing availability;
c minimises the fire hazard; c can reduce contact voltage; c and, for sensitive systems, minimises disturbance due to electromagnetic radiation and common impedance. Moreover, in view of the proliferation of communicating digital systems (computers, video, automation, TBM etc., it is vital that earthing systems provide a potential reference which is not disturbed by high fault currents and harmonics. Consequently, future evolution should favour earthing systems generating fault currents which do not exceed a few dozen amps. TT earthing systems should therefore be increasingly used.
appendix 1: IEC 364 standard
c 364-1 - Electrical installations of buildings (NF C 15-100). c 364-1 - 1992 Part 1: Scope, object and fundamental principles c 364-2-21 - 1993 Part 2: Definitions - Chaper 21 - Guide to general terms c 364-3 - 1993 Part 3: Assessment of general characteristics c 364-4 Part 4: Protection for safety v 364-4-41 - 1992 Chapter 41 : Protection against Electric Shock v 364-4-42- 1980 Chapter 42: Protection against thermal effects v 364-4-43 - 1977 Chapter 43: Protection against overcurrent v 364-4-45 - 1984 Chapter 45: Protection against undervoltage v 364-4-46 - 1981 Chapter 46: Isolation and switching v 364-4-47 - 1981 Chapter 47: Application of protective measures for safety - Section 470: General - Section 471: Measures of protection against Electric Shock v 364-4-442 - 1993 Chapter 44: Protection against overvoltages - Section 442: Protection of low-voltage installations against faults between high-voltage systems and earth v 364-4-443 - 1993 Chapter 44: Protection against overvoltages - Section 443: Protection against overvoltages of atmospheric origin or due to switching
v 364-4-473 - 1977 Chapter 47: Application of protective measures for safety - Section 473: Measures of protection against overcurrent v 364-4-481 - 1993 Chapter 48: Choice of protective measures as a function of external influences - Section 481: Selection of measures for protection against Electric Shock in relation to external influences v 364-4-482 - 1982 Chapter 48: Choice of protective measures as a function of external influences - Section 482: Protection against fire c 364-5 Part 5: Selection and erection of electrical equipment. v 364-5-51 - 1979 Chapter 51: Common rules v 364-5-51 - 1 - 1982 Amendment No. 1 364-5-51 - 1979. v 364-5-51 - 2 - 1993 Amendment No. 2 364-5-51 - 1979. v 364-5-53 - 1986 Chapter 53: Switchgear and controlgear v 364-5-53 - 2 - 1992 Amendment No. 1 364-5-53 - 1986. v 364-5-54 - 1980 Chapter 54: Earthing arrangements and protective conductors v 364-5-54 - 1 - 1982 Amendment No. 1 364-5-54 - 1980. v 364-5-56 - 1980 Chapter 56: Safety services v 364-5-523 - 1983 Chapter 52: Wiring systems - Section 523: Current-carrying capacities
v 364-5-537 - 1981 Chapter 53: Switchgear and controlgear - Section 537: Devices for isolation and switching v 364-5-537 - 1 - 1989 Amendment No. 1 364-5-537 - 1981. c 364-6 Part 6: Verification v 364-6-61 - 1986 Chapter 61: Initial verification v 364-6-61 - 1993 Amendment No. 1 364-6-61 - 1986. c 364-7 Part 7: Requirements for special installations or locations v 364-7-701 - 1984 Section 701: Locations containing a bath tub or shower basin v 364-7-702 - 1983 Section 702: Swimming pools v 364-7-703 - 1984 Section 703: Locations containing sauna heaters v 364-7-704 - 1989 Section 704: Construction and demolition site installations v 364-7-705 - 1984 Section 705: Electrical installations of agricultural and horticultural premises v 364-7-706 - 1983 Section 706: Restrictive conducting locations v 364-7-707 - 1984 Section 707: Earthing requirements for the installation of data processing equipment v 364-7-708 - 1988 Section 708: Electrical installations in caravan parks and caravans
Standards c IEC 241: Fuses for domestic and similar purposes. c IEC 269: Low voltage fuses. c IEC 364: Electrical installations of buildings. c IEC 479: Effects of currents flowing through the human body. c IEC 755: General requirements for residual current operated protective devices. c IEC 947-2: Low voltage switchgear and controlgear. c NF C 15-100 : Installations lectriques basse tension. c NF C 63-080 : Dispositifs de contrle permanent d'isolement et dispositifs de localisation de dfauts associs. c NF C 63-150 : Limiteurs de surtension : rgles. French decree of the 14.11.88
Merlin Gerin's Cahiers Techniques c Earthing of the neutral conductor in High-Voltage networks, Cahier Technique n 62 F. SAUTRIAU c Residual current devices, Cahier Technique n 114 R. CALVAS c Protections des personnes et alimentations statiques sans coupure, Cahier Technique n 129 J.-N. FIORINA c Les perturbations lectriques en BT, Cahier Technique n 141 R. CALVAS c Introduction to dependability design, Cahier Technique n 144 P. BONNEFOI c Lightning and HV electrical installations, Cahier Technique n 168 B. DE METZ NOBLAT c Earthing systems in LV, Cahier Technique n 172 B. LACROIX and R. CALVAS c Perturbations des systmes lectroniques et schmas des liaisons la terre, Cahier Technique n 177 R. CALVAS c Connaissance et emploi du SLT neutre isol, Cahier Technique n 178 E. TISON et I. HERITIER
Other publications c Guide de linstallation lectrique (partie G). Ed. FRANCE IMPRESSION CONSEIL 1991. c Guide de lingnierie lectrique. Ed. ELECTRA 1986. c Electrical Review. Nov. 1991 - Oct. 1992. c La protection diffrentielle. Cahier Technique J3E - 02/90.

The IT earthing system (unearthed neutral) in LV
F. Jullien I. Hritier
Cahiers Techniques is a collection of documents intended for engineers and technicians, people in the industry who are looking for more in-depth information in order to complement that given in product catalogues. Furthermore, these Cahiers Techniques are often considered as helpful tools for training courses. They provide knowledge on new technical and technological developments in the electrotechnical field and electronics. They also provide better understanding of various phenomena observed in electrical installations, systems and equipments. Each Cahier Technique provides an in-depth study of a precise subject in the fields of electrical networks, protection devices, monitoring and control and industrial automation systems. The latest publications can be downloaded from the Schneider Electric internet web site. Code: http://www.schneider-electric.com Section: Experts place Please contact your Schneider Electric representative if you want either a Cahier Technique or the list of available titles. The Cahiers Techniques collection is part of the Schneider Electrics Collection technique.
Foreword The author disclaims all responsibility subsequent to incorrect use of information or diagrams reproduced in this document, and cannot be held responsible for any errors or oversights, or for the consequences of using information and diagrams contained in this document. Reproduction of all or part of a Cahier Technique is authorised with the prior consent of the Scientific and Technical Division. The statement Extracted from Schneider Electric Cahier Technique no. ..... (please specify) is compulsory.
no. 178
Franois JULLIEN Joined Schneider Electrics Low Voltage activity in 1987. A 1996 engineering graduate (from the Conservatoire National des Arts et Mtiers), he was then placed in charge of the electronic technical team in the Low Voltage Power Components activity, with particular responsibility for follow-up of the Vigilohm system range for electrical network insulation monitoring and insulation fault tracking.
Isabelle HERITIER An ENSERG engineering graduate (Ecole Nationale Suprieure dElectronique et de Radiolectricit de Grenoble), she joined Merlin Gerin in 1989. During her professional career, she has been responsible for development of an insulation monitoring system for the National Marine, support engineer for sales forces and product manager for residual current relay, insulation monitor and communicating device ranges. She is currently product manager for LV circuit-breakers from 100 A to 600 A.
ECT 178 first issue, June 1999
Lexicon
BB: busbar. C1 for phase 1, C2 for phase 2, and C3 for phase 3: earthing impedance capacitive components for each phase. CR: network overall capacity (leakage capacities of cables and filters if any). IC: capacitive current. Id: fault current flowing in the earth connection resistance RA of the application frame. Ifu: fuse blowing current within a maximum time stipulated by standards. Im: short time delay (magnetic or electronic) tripping current (threshold) of a circuit-breaker. IN: capacitive current flowing through the earthed neutral connection, in particular through the impedance ZN, when present. L: length of faulty circuits. m: ratio of live conductor/protective conductor cross-section (Sa / Spe). : resistivity of copper. Ra: resistance of the live conductor (phase or neutral) of the circuit where the fault occurred. RA: resistance of the earth connection of the application frames. RB: resistance of the neutral earth connection. RCD: Residual Current Device.
Rd: fault resistance. Rpe: resistance of the PE protective conductor. R1 for phase 1, R2 for phase 2, and R3 for phase 3: earthing impedance resistive components of each phase. Sa: live conductor cross-section. SCPD: Short-Circuit Protection Device. Spe: protective conductor cross-section. UC: contact voltage between the frame of a faulty device and another frame or the earth. U0: phase to neutral voltage. UL: limit safety voltage (24 V) not to be exceeded between the frame of a device and another frame or the earth. Un: nominal voltage or phase-to-phase voltage (U1, U2, U3), equal to e U0 for a three-phase electrical circuit. Ur: network voltage. ZN: additional impedance connected between the neutral point of a network in the IT earthing system and the earth. ZR: overall impedance of a network with respect to the earth, made up of the capacitive components C1, C2, C3 and the resistive components R1, R2, R3.

Although all Earthing Systems offer users the same degree of safety, they do not all have the same operating characteristics. This is why, in certain countries, a specific earthing system is stipulated by legislation or standards according to buildings. For example, in France the IT system is compulsory in hospital operating theatres, and the TN-C is forbidden in premises where there is a risk of explosion. These stipulations apart, dependability objectives (safety, availability, reliability, maintenability and proper operation of low current communication systems) determine which earthing system should be chosen for which installation. The aim of this Cahier Technique is to describe the advantages and areas of application of the IT earthing system. After a brief introduction of the electrical hazard and the various earthing systems, the first fault situations, followed by the double fault specific to the IT system, are studied, and the advantages and disadvantages of this particular earthing system are developed. This Cahier Technique also offers solutions for the surge limiter with the various types of possible overvoltages. Finally, a choice table is provided for all earthing systems, based on criteria for safety, availability, electromagnetic compatibility and operators professional requirements.
Contents
1 Introduction 1.1 Protection of persons against electrical shocks 1.2 The various standardised earthing systems 1.3 Choosing an earthing system 1.4 Type of insulation 1.5 Equivalent system for an unearthed or impedance-earthed neutral network 2.1 Calculating fault currents and contact voltage on the first fault 2.2 Permanent insulation monitors, history and principles 2.3 Tracking the 1st insulation fault 3 The 2nd insulation fault with the IT earthing system 4 Special features of the IT earthing system 3.1 Analysis of the double insulation fault 3.2 Elimination of the double insulation fault 4.1 Overvoltages in the IT system 4.2 Surge limiters 4.3 Why use an impedance? 5 Advantages and disadvantages of the IT earthing system in LV 5.1 Increased availability 5.2 Increased safety against risk of fire 5.3 Less downtime on control and monitoring circuits 5.4 Usage limits and precautions of the IT earthing system 6 Conclusion 6.1 Availability: an increasing need to be satisfied 6.2 The IT earthing system finds its true place 6.3 The added advantage of safety 6.4 In short Bibliography p. 4 p. 4 p. 7 p. 7 p. 8 p. 9 p. 11 p. 13 p. 15 p. 16 p. 18 p. 20 p. 21 p. 22 p. 22 p. 23 p. 23 p. 26 p. 26 p. 26 p. 27 p. 28
2 The 1st insulation fault with the IT earthing system
1 Introduction
1.1 Protection of persons against electric shocks

Use of Safety by Extra Low Voltage (< 25 V) -SELV- is the most drastic solution since it eliminates the electrical hazard. However it is applicable only in low power distribution. Regarding everyday use of electricity, a number of studies have identified the causes of electric shocks and provided specific solutions for each one. Electric shocks have two causes, namely: c direct contact, i.e. a person or an animal touching an exposed live conductor; c indirect contact, i.e. a person touching the metal frame of an electrical load on which an insulation fault has occurred. Protection against direct contact To provide protection against direct contact, insulation and/or distancing measures are taken. These measures can be reinforced in final distribution by additional protection in the form of a high sensitivity Residual Current Device (RCD). Protection against indirect contact With respect to protection against indirect contact, between an accidentally energised frame and the earth, the basic solution is to earth all the load frames via the protective conductors. However, this measure does not rule out the existence of a contact voltage hazardous for persons if it exceeds the conventional limit safety voltage UL defined by standard IEC 60479. This contact voltage depends on the earthing systems defined in the international standard IEC 60364.
1.2 The various standardised earthing systems

The three earthing systems given official status by international standards (IEC 60364) are also stipulated by a large number of national standards: in France by the LV installation standard: NF C 15-100. A brief reminder of the protection principle of these systems will now be given before describing the IT system in greater detail. The TN system c Its principle: v the transformer neutral is earthed; v the electrical load frames are connected to neutral. This type of system has three possibilities: v the same conductor acts as a neutral and a protective conductor: this is the TN-C system; v the neutral and the protective conductor are separate: this is the TN-S system; v and the combination of these two systems, known as TN-C-S when the neutral and the protective conductor are separated downstream of part of the installation in the TN-C system. Note that the TN-S cannot be placed upstream of the TN-C. c Its operation (see fig. 1 ): An insulation fault on a phase becomes a shortcircuit and the faulty part is disconnected by a Short-Circuit Protection Device (SCPD). The TT system c Its principle: v the transformer neutral is earthed; v the electrical load frames are also earthed. c Its operation (see fig. 2 ): The current of an insulation fault is limited by earth connection impedance.
a)
b)
Id
N PEN SCPD PE N
Id
SCPD
Ud RB
Rd RB
Ud
Rd
c)
Id
1 2 3 N PEN PE SCPD PE
Ud RB
Rd
Fig. 1 : insulation fault on a network operated in TN-C [a], TN-S [b] and TN-C-S [c].
RCD
Id
Protection is provided by the Residual Current Devices (RCD): the faulty part is disconnected as soon as the threshold In, of the RCD placed upstream, is overshot by the fault current, so that In RB i UL. The IT system c Its principle: v the transformer neutral is not earthed, but is theoretically unearthed. In actual fact, it is naturally earthed by the stray capacities of the network cables and/or voluntarily by a high impedance of around 1,500 (impedanceearthed neutral); v the electrical load frames are earthed.
Ud RB RA
Rd
Fig. 2 : insulation fault on a network operated in TT.
c Its operation: v should an insulation fault occur, a low current develops as a result of the networks stray capacities (see fig. 3a ). The contact voltage developed in the frame earth connection (no more than a few volts) is not dangerous; v if a second fault occurs on another phase before the first fault has been eliminated (see fig. 3b and 3c ), the frames of the loads in question are brought to the potential developed
by the fault current in the protective conductor (PE) connecting them. The SCPDs (for the frames interconnected by the PE) or the RCDs (for the frames with separate earth connections) provide the necessary protection. This deliberately brief presentation of the various earthing systems clearly cannot cover all the specific installation possibilities. Readers requiring more details can consult Cahiers Techniques no. 114, 172 and 173.
a)
Id
3 2 1 PE
Permanent insulation monitor (PIM) ZN : optional impedance
ZN
Surge limiter
Id Id
C1 C2
C3
IN IC
RB
Ud
IC1 IC2 IC3
IC
b)
Id Id
N 3 2 1 N PE
Permanent insulation monitor (PIM)
Surge limiter
Id
SCPD
SCPD
Id
Ud2 RB c)
Rd2
Ud1
Rd1
Id
N
RCD
Id
3 2 1 N PE
PE Permanent insulation monitor (PIM) Surge limiter SCPD
Id
SCPD
Id
Ud2 RB
Rd2
Ud1 RA
Rd1
Id
Fig. 3 : single [a] and double [b and c] insulation fault on a network operated in IT.
1.3 Choosing an earthing system

Although all three earthing systems offer users the same degree of safety against indirect contact, only the IT system guarantees risk-free continuity of supply in the presence of an insulation fault. This undeniable advantage also has certain drawbacks: for example the need to locate this first fault and the possibility of overvoltages occurring that may affect operation of sensitive loads. However, choice of earthing system for an installation also depends on parameters other than safety of persons and continuity of supply, namely: c the environment (e.g. premises with a risk of fire or sites frequently struck by lightning); c electromagnetic compatibility (EMC) (presence in the installation of harmonics and radiating fields, and sensitivity of equipment to such phenomena); c technicity of installation designers and operators; c maintenance quality and cost; c network size; c etc. Although consideration of the above parameters guarantees choice of the earthing system most suited to the installation, it should be emphasised that the advantage offered by the IT system in terms of availability (2nd fault most unlikely) generates installation and operating costs that should be compared with the downtime costs generated by other earthing systems (operating losses and repair costs caused by the first insulation fault).
1.4 Type of insulation

Common mode impedance All electrical networks have an impedance with respect to earth known as the common mode impedance, the origin of which is insulation of network cables and loads. This impedance consists of the leakage capacity and resistance between each live conductor and the earth. In LV, the leakage resistance of a new cable is around 10 M per kilometer and per phase, whereas its capacity evenly distributed with respect to earth is approximately 0.25 F, i.e. 12.7 k at 50 Hz. It should also be noted that in MV and HV this leakage capacity is even greater and MUST be taken into account when drawing up a protection plan (see Cahier Technique no. 62). Loads also have a natural leakage capacity, usually negligible. Effect of distributed capacity in the IT system In electrical installations, other capacities are added to the network cable ones. This is the case of certain electronic loads that are generators of HF harmonic currents, in particular when they use the chopper principle (e.g. pulse with modulation converters). However, ElectroMagnetic Compatibility (EMC) standards state that these HF currents must be shunted to earth, resulting in the presence of filters and thus capacitors between phases and frame. According to the number of loads, their contribution to the networks leakage capacity can be significant or even important. Measurements taken on a variety of electrical power networks show that capacity varies considerably from network to network and covers a range of a few F to a few dozen F. Excessively high capacities may question the advantage of the IT earthing system: if, on the first fault, the value of network impedance with respect to earth means that contact voltage exceeds 50 V, safety of persons is not guaranteed. This is rare, however, as with a 10 earth connection, the networks earth leakage capacity must exceed 70 F (23 F per phase if the neutral is not distributed). An IT network must therefore have a limited capacity with respect to earth, and the presence of loads equipped with HF filters must be taken into account in the network design stage.
1.5 Equivalent system for a network with unearthed or impedance-earthed neutral

A few definitions and assumptions are given below in order to define the equivalent system for this network (see fig. 4 ): c the neutral point is unearthed or earthed by an impedance (ZN) of high value (normally 1 k to 2 k) whose earth connection is equivalent to a resistance (RB); c the load frames are interconnected either fully or by group. For EMC reasons (see Cahier Technique no. 187), it is advisable to interconnect all the application frames of the same installation and to connect them to the same earth connection (resistance RA); c the earth connections (RA and RB) are interconnected (in most cases), or separate. NB: Two earth connections are considered to be separate if they are more than 8 m apart; c each live conductor has, with respect to earth, an impedance made up of a resistance and a capacity.
3 2 1 N N ZN PE
RN
R1 R2
R3
CN C1
C2
C3
RB
RA
Fig. 4 : equivalent system of a network with unearthed or impedance-earthed neutral.
2 The 1st insulation fault with the IT earthing system
In normal operating conditions, safety of persons is guaranteed when contact voltage is less than 50 V as per standard IEC 60364 (NF C 15-100). When this contact voltage is exceeded, these
standards require automatic opening of the circuit. The following section shows how use of the IT earthing system for network operation prevents tripping on the first insulation fault.
2.1 Calculating fault currents and contact voltage on the first fault
General case (resistive fault) Should a fault with a resistive value Rd occur between phase 3 and the earth, a fault current Id flows in the neutral impedance and in the capacities C1, C2 and C3 (see fig. 3a). Assuming that the phase-to-earth capacities are balanced (C1 = C2 = C3 = C), the fault current has the following value: 1+ 3j C ZN I d = U0 Rd + ZN + 3j C ZN Rd The capacitive current is written as: 3j C ZN I c = U0 Rd + ZN + 3j C ZN Rd and the current in the impedance ZN: U0 IN = Rd + ZN + 3j C ZN Rd The contact voltage UC (contact voltage between the frame of a faulty device and another frame or the earth) is calculated from the fault current Id flowing in the earth connection resistance RA of the application frames if they are not interconnected, else RB (only network earth connection): UC = RA Id. Case of the full fault This paragraph calculates the configuration generating the highest contact voltage (UC): thus for a fault occuring on a frame with earth connection separate from that of ZN. By application of the above formulae, where Rd = 0, we obtain:
U0 ZN + 3j C U0 Uc = RA ZN + 3j C The capacitive current is equal to: IC = +3j C U0 and the current in impedance ZN: U IN = 0 ZN In the various examples below, studied for ZN = (unearthed neutral) and ZN = 1 k
(impedance-earthed neutral), the calculations are made for a network in the IT system, 400 VAC (U0 = 230 V), where: RA, earth connection resistance = 10 Rd , insulation fault value = 0 to 10 k. c Case 1: Low capacity network (e.g. limited to an operating theatre) C1 = C2 = C3 = C = 0.3 F per phase. c Case 2: Power network, where C1 = C2 = C3 = C = 1.6 F per phase. c Case 3: Very long power network, where C1 = C2 = C3 = C = 10 F per phase, i.e. roughly 40 km of cables! The results of all these calculations, grouped in the table in figure 5 , confirm the low fault voltage ( 20 V in the most unfavourable cases), ensuring continuity of operation, without risk for persons, of a network designed using the IT system. They prove that addition of an impedance between the neutral and the earth has very little effect on contact voltage.
Rd (k) Case 1 ZN = CR = 1 F UC (V)
0.5
10
0.72 0.71 0.69 0.22 0.07 0.07 0.07 0.02 1.19 0.21 0.24 0.16 0.12 0.02 0.36 0.28 0.19 0.02 0.43 0.25 0.17 0.02 2.29 0.23 2.17 0.45 0.23 0.02 2.18 0.44 0.23 0.02
Id (A) Id (A)
ZN = 1 k UC (V) 2.41 1.6 Case 2 ZN = CR = 5 F
UC (V) 3.61 2.84 1.94 0.23
Id =
Id (A) Id (A)
ZN = 1 k UC (V) 4.28 2.53 1.68 0.22 Case 3 ZN = CR = 30 F UC (V) 21.7 4.5
Id (A) Id (A)
ZN = 1 k UC (V) 21.8 4.41 2.26 0.23
Fig. 5 : comparison of fault currents and contact voltages on a first fault.
The curves in figure 6 representing these results show the considerable effect of network capacity on the value of UC. In point of fact, regardless of the distributed capacity of the sound network or network on which a first fault has occurred, users can be certain that this voltage will always be less than the conventional safety voltage and thus without risk for persons. Also, the currents of a first full fault are low and thus have minimum destructive or disturbing (EMC) effect.
Effect of distributed capacities, vector chart and neutral potential c Effect of distributed capacities on a sound network The capacities of all 3 phases create an artificial neutral point. In the absence of an insulation fault, if network capacities are balanced, this neutral point is then at earth potential (see fig. 7 ). In the absence of a fault, the phase-to-earth potential is thus equal to phase to neutral voltage for each phase.
Uc (V) where Zn = 1,000 100 50 CR = 70 F CR = 30 F
10 CR = 5 F
CR = 1 F
0.1 1 10 100 500 1,000 (recommended threshold)
Rd () 104
Fig. 6 : contact voltage on a first insulation fault is always less than safety voltage.
Artificial neutral 3 2 3 C C C 3C 2
Fig. 7 : the networks distributed capacities create a connection between neutral and earth.
c Vector chart in presence of a full fault In event of a full fault on phase 1, the potential of phase 1 is at earth potential (see fig. 8 ). The neutral-to-earth potential is thus equal to phase to neutral voltage V1, and that of phases 2 and 3 with respect to earth is equal to phase-to-phase voltage. If the neutral is
distributed, the fault current is arithmetically increased: IC = 4j C V1. However, detection, location and correction of this fault must be immediate in order to reduce the risk of a second simultaneous fault occurring which would result in opening of the faulty circuits.
a)
V1 N V3 T V2 I V1-T I = I V3-T I = I V2-T I
b)
2 N v2 v3 1 T
Id
IC3 IC2
V1-T = 0
Id = IC = IC2 + IC3 IC2 = j C v2 IC3 = j C v3 IC = 3j C V1

I Id I = 3 C I V1 I
V3-T = V1 + V3 V2-T = V1 + V2
Fig. 8 : vector charts of a network in the IT system, without fault [a] and with an earth fault on phase 1 [b].
2.2 Permanent insulation monitors, history and principles

The first LV electrical distribution networks were operated using the IT earthing system. Operators rapidly sought to detect the presence of the first insulation fault in order to prevent the hazards linked to a short-circuit current of varying impedance and the de-energisation of a faulty feeder (with the lowest rating protection) or of the two faulty feeders. The first PIMs These devices used 3 lamps connected between the phases and the earth (see fig. 9 ). On a sound network, the three lamps form a balanced three-phase load, all lit and with the same brilliance. When an insulation fault occurs, one of the three lamps is short-circuited by the fault impedance. Voltage is reduced at the terminals of this lamp, and lamp brilliance decreases. However, voltage at the terminals of the other two lamps increases until phase-tophase voltage is reached. Their luminosity also increases. This system is easy to install and use. However, given that its practical operating threshold is low, attempts were quickly made to try to detect impedant faults in order to anticipate the full fault. For a DC network (supplied by batteries or by DC generator). The technique of the voltmeter balance (see fig. 10 ) was the first to be used, and indeed is still used today.
(1) (2)
(+) (-) R R
(3) The OFF indicator light indicates the faulty phase: in this case no. 3.
The needle indicates the faulty polarity; in this case the (-) polarity.
Fig. 9 : principle of the first PIM.
Fig. 10 : principle of the PIM with voltmeter balance.
The principle consists of measuring and comparing voltages between the (+) polarity and the earth, on the one hand, and the (-) polarity and the earth on the other. This principle makes auxiliary power sources unnecessary, since the network supplies the PIM directly via measurement sensors (resistances). This technique applies to two-phase AC and DC networks and does not allow live fault tracking. For AC networks The most commonly used PIM are those with insulation measurement by DC current injection. Permanent measurement of insulation resistance required use of active systems in place of the previously used passive systems. This resistance can be measured accurately in DC (see fig. 11 ), which is why the first PIMs, placed between the network and the earth, injected a low DC current which flowed through the fault. This simple, reliable technique is still extensively used today, but does not allow live fault tracking. Note that when these PIMs are used on mixed networks (containing rectifiers without galvanic insulation), they may be disturbed or even
blinded if a fault is present on the DC part of the network. These were followed by PIMs with AC current injection at low frequency (< 10 Hz), operating on the same principle. Although these PIMs allow live fault tracking, they can be misled by cable capacities that are seen as insulation faults and disturbed by frequency converters (variable speed controllers). For all AC and DC networks Finally, nowadays, given that networks are frequently of the mixed AC/DC kind as well as variable frequency, the new PIMs are able to monitor insulation on all types of networks. c Some use squared wave pulses at very low frequency ( 1Hz). They allow PIM not to be disturbed by earth leakage capacities, as they are then immediately loaded then unloaded by the next strobe pulse of opposite sign. They are universal in use and easily adapted to modern networks, in particular to those supplying power electronic devices which often deform the AC pulse. However, their response time, depending on the networks earth leakage capacity, may be as much as a few minutes and does not allow the detection of intermittent faults. c In order to compensate the usage restrictions of these PIMs for very long networks and networks with a large number of capacitive filters, the low frequency AC current injection technique has been improved by means of synchronous demodulation (see fig. 12 ): this type of PIM applies a low frequency AC voltage between the network and the earth, measures the current flowing back via network insulation impedance and calculates the voltage-current shift. It is then possible to determine the resistive and capacitive components of this current and thus relate the threshold to the resistive component only. This upgrade, the result of digital technology, combines the advantages of DC current and low frequency AC current injection without their disadvantages.
IPIM
3 2 1 N N PE
RB
IPIM
Fig. 11 : principle of PIM with current injection.
IBF IR-BF IC-BF IC-BF IBF
~
mA
BF
V = UBF RNetwork CNetwork
IR-BF
ZNetwork
UBF
Fig. 12 : the low frequency AC current injection technique has been improved by means of synchronous demodulation, which enables the insulation drop (resistive leakage) to be distinguished from capacitive leakage.
PIM standards c The manufacturing standards Standard IEC 61557-8 is in existence since February 1997. It defines the special specifications governing insulation monitors designed for permanent monitoring, irrespective of the measurement principle, of insulation resistance with respect to earth of unearthed AC and DC IT system networks, and of AC IT system networks containing rectifiers supplied without galvanic separation (transformer with separate windings). It places particular emphasis on three points. v Properly inform specifiers and contractors. The manufacturer must provide the characteristics of the devices he produces and in particular those that are dependent on network capacity (response time and threshold values). v Ensure that these devices are properly integrated in their electrical environment. This requires compliance with the specifications of standards IEC 61326-1 and 61326-10 concerning ElectroMagnetic Compatibility (EMC). v Guarantee operating safety for users. The main stipulations are: device operating testing must be possible without inserting an additional impedance between the monitored network and the earth, settings must be protected to prevent modification by error or by unauthorised users, and impossibility of device disconnection (the need to use a tool for disassembly).
c The operating standards As concerns PIM setting, standard IEC 60364 provides an initial answer: A PIM designed according to is set at a value less than the minimum value of the insulation resistance defined for the installation in question, i.e. greater than or equal to 0.5 M for a circuit with a nominal voltage greater than or equal to 500 V. Guide NF C 15-100 states: set at a value roughly less than 20% of the resistance of the installation as a whole However, a clear distinction must be made between the insulation resistance of the installation, which only takes electrical distribution into account, and the insulation level which is set for overall network monitoring, including the various machines and switchgear connected to it. In the previous chapter we saw that for faults greater than 500 , contact voltage does not exceed 5 V with an earth connection of 10 (see fig. 5). In practice, for a normal industrial installation, it is thus reasonable, without taking risks, to set the lower alarm threshold at a value of between 500 and 1,000 , ensuring effective fault tracking (and thus location of the reported insulation fault). To organise preventive tracking, it is useful to have a first level threshold around 10 k for example. This threshold must be adapted according to installation characteristics and operating requirements. Note that short networks allow a higher prevention threshold.
2.3 Tracking the 1st insulation fault

When tracking a fault, although certain operators merely identify the faulty feeder, accurate determination of the location of this fault is recommended (e.g. damaged cable or insulation fault in a device) in order to put it right as quickly as possible. Tracking by successive de-energisation of feeders This means of fault tracking is quoted for memory only. It consists of opening the feeders one by one, beginning with the main feeders. When the faulty feeder is opened, the current injected by the PIM decreases markedly and drops below the detection threshold. The audible alarm normally controlled by the PIM then stops, enabling remote identification of the faulty feeder. This procedure, which requires interruption of operation on each feeder, is contrary to the operating philosophy of the IT earthing system, which stipulates continuity of supply. Although frequently used in the past, it is gradually disappearing with the development of the new fault tracking systems which allow live tracking (without power breaking). Live tracking c Detecting the fault current As seen above (see fig. 3a), a current Id flows through the first insulation fault at the same frequency as that of the network (50 Hz or 60 Hz), returning to the source via the capacities of the other sound phases and via the neutral impedance if any. An initial live tracking method (without interrupting distribution) consisted of using a clamp-on probe to measure the earth leakage current on each feeder. The faulty feeder was the one on which the highest value was measured. This method has two drawbacks, namely: v It is not reliable for networks with a large number of feeders some of which are highly capacitive (how can the earth current of a short faulty feeder be distinguished from that of a long capacitive feeder?).
v It is not applicable on networks with few capacitive leakages (the fault current is virtually undetectable). In order to improve detection of the fault current path (at power frequency) using a clamp-on probe, two tricks were used. The first consisted of increasing this fault current by temporarily placing a low impedance in parallel on the PIM. The second consisted of distinguishing the capacitive leakage currents from the fault current by the periodic use of the above-quoted impedance by means of a beating relay (approx. 2 Hz). c Detecting an injected current This method uses a low frequency sinusoidal pulse (i 10 Hz) injected by a generator or a PIM. Choice of low frequency for fault tracking ensures no disturbance by network leakage capacities, but this frequency cannot be less than 2.5 Hz, as detection with a magnetic sensor becomes difficult. This method uses devices sensitive to the injected pulse only, that can either be fixed with detection toroids placed on all feeders, or portable with a clamp-on probe tuned to signal frequency in order to locate the exact position of the fault (see fig. 13 ).
When the devices (generator, sensors and load) are fixed, live fault tracking can be automatic on detection of a fault, with transmission of an order by the PIM. c Measuring insulation of each feeder Operators, with their ever-increasing need for continuity of supply, are no longer prepared even to wait for the first fault, but want to be able to programme maintenance work and thus anticipate the next feeder likely to be affected by an insulation fault. It is thus necessary to monitor the changes in insulation of each feeder and to carefully identify the resistive and capacitive insulation components. The synchronous demodulation principle can also be used by measuring, first, the injection current flowing in the feeders (by the toroid sensors) and, second, the injection voltage. Development of this tracking method is encouraged by application of digital techniques to the management of electrical power distribution (see Cahier Technique no. 186): the user can now remotely and continually monitor insulation changes of the various feeders. Use of digital buses enables data to be centralised on a supervisor, displayed and logged, thus allowing intelligent, predictive maintenance.
Fixed load with manual or automatic scanning 12 3N
////
PE
LF generator (G) PIM ( )
RB Manual load PE
Fig. 13 : tracking can take place with fixed or portable devices sensitive to the injected pulse.
3 The 2nd insulation fault with the IT earthing system
As we have already seen in the previous chapter, the advantage of using the IT system in network operation lies in the possibility of continuity of supply even though an insulation fault has occurred on a circuit. This message has been received loud and clear by standard drafters who, in order to maintain a high level of availability, stipulate in installation standards indication and tracking of this first fault so as not to fear a second fault. Protection devices are also provided for this second fault in order to guarantee the same level of safety of persons as for the TN and TT earthing systems.
The two sections below study the fault currents and contact voltage which depend on how the frames are earthed. There are two possibilities, namely: c The load frames are all interconnected by a PE protective conductor: this is the general case. c The frames are not interconnected and are connected to separate earth connections (configuration to be avoided due to EMC: see Cahier Technique no. 187).
3.1 Analysis of the double insulation fault

In this section, fault currents and contact voltage are calculated by considering two full insulation faults on two different live conductors (on one phase and the neutral if the neutral is distributed, or on two different phase conductors if the neutral is not distributed) of two circuits of identical cross-section and length. This assumption, which results in a minimum fault current, is the one normally chosen to calculate the maximum lengths protected by the short-circuit protection devices. Contact voltage and double fault current when the frames are interconnected When a fault current occurs between two faulty frames, a current flows in the phase conductors and the PE protective conductor ensuring interconnection of frames (see fig. 3b). This current is only limited by the impedance of the fault loop equal to the sum of the impedances of the live conductors concerned and the circuit of the equipotential links (PE). There are a number of methods for calculating fault currents for an electrical installation (see Cahier Technique no. 158). In this case, the conventional method has been chosen, as it enables calculation of fault current and contact voltage values without making too many assumptions on installation characteristics. It will thus be used from now on in this Cahier Technique to give an idea of the value of the currents and voltages involved on a double fault in the IT system. It is based on the simplified assumption that considers that, during the duration of the fault, the voltage at the origin of the feeder considered is equal to 80% of installation nominal voltage. This assumes that the impedance of the feeder in question accounts for 80% of total impedance of the faulty loop, and that upstream impedance accounts for 20%. For the following calculations: U = phase to neutral voltage, (= U0 if one of the two faults is on the distributed neutral), or U = phase-to-phase voltage, (= e U0 if the neutral is not distributed). L Ra = = resistance of the live conductor Sa (phase or neutral) of the circuit on which the fault occurred. L = resistance of the circuit protective Rpe = Spe conductor. Sa = cross-section of the live conductor. Spe = cross-section of the protective conductor. L = length of the faulty circuits. S m = a = ratio of live conductor cross-section Spe over protective conductor cross-section (normally i 1). c If we consider that the live and PE conductors of the two faulty feeders have the same crosssection and length and if we ignore their reactance: v if one of the faults is on the neutral
Id =
2 Ra + Rpe
0,8 U0
, i.e. I d = 0,8 U0
Sa , 2 (1 + m) L
v if the double fault concerns two phase Sa conductors I d = 0,8 e U0 . 2 (1 + m) L
c The corresponding contact voltage is UC = Rpe Id , i.e.: v if one of the faults is on the neutral m , or Uc = 0,8 U0 2 (1 + m) v if the double fault concerns two phase m conductors Uc = 0,8 e U0 . 2 (1 + m) NB: this method is not applicable for installations supplied by generator set, as, due to high generator impedance compared with the impedance of the supplied network, voltage at the origin of the network in question is low when a fault occurs (<< 0.8 Un). In this case, irrespective of the earthing system, only complete electro-technical methods, of the impedance method type, can be used.
c Digital example The results presented in the table in figure 14 confirm that a double insulation fault is a risk for safety of persons since contact voltage is greater than limit safety voltage UL. The automatic protection devices must then de-energise the installation. Contact voltage and double fault current when the frames are not interconnected If the two faults occur on two loads connected to two separate earth connections (see fig. 3c), the fault current Id is then closed by the earth and is limited by the earth connection resistances RA and RB. A simple calculation shows that this second insulation fault is just as dangerous (see fig. 15 ), and must therefore be automatically eliminated, and that the threshold of the short-circuit protection devices cannot be reached.
Faults on 2 identical feeders (where m = 1) Double fault c phase - neutral c phase - phase UC = 46 V UC = 79.7 V
Faults on a BB and a feeder (where m = 4) UC = 73.6 V UC = 127.5 V
UC Double fault c phase - neutral c phase - phase 115 V 200 V
Id
11 A 20 A
Fig. 14 : contact voltages on a double fault for a 230/400 V network in the IT earthing system.
Fig. 15 : fault currents and contact voltages on a double fault on two frames with separate earth connections, where RA = RB = 10 .
3.2 Elimination of the double insulation fault

Case of interconnected application frames In view of the importance of the fault current, comparable with a short-circuit current, automatic short-circuit protection devices (SCPD) can be used for tripping if cable lengths are compatible with their operating thresholds. Otherwise residual current devices (RCD) are used. Elimination of the double fault must also satisfy other requirements which apply regardless of the type of SCPD installed (fuse or circuit-breaker): c The contact voltages calculated in the previous chapter, for all SCPD types, leave little time for fault elimination. In order to simplify the network designers task, standard IEC 60364 specifies maximum breaking times as a function of operating voltage (see fig. 16 ). c Multi-pole breaking, including the neutral conductor when distributed. The reasons for this are: v breaking only of the faulty phase conductor of a feeder means that three-phase machines are supplied by the two other phases, v breaking of the neutral exposes to phase-tophase voltage, single-phase loads normally supplied by phase to neutral voltage.
Network Uo (V) Maximum breaking time (s) Non-distributed neutral 127 230 400 >> 400 0.4 0.2 0.06 0.02 Distributed neutral (*) 1 0.5 0.2 0.08
Fig. 16 : maximum breaking time specified for the IT earthing system by installation standards (* for singlephase networks).
c Protection of the neutral conductor when it is distributed. Figure 3b shows that when a double fault occurs, the two SCPDs detect the fault current but each one on a single phase or on the neutral. This situation calls for particular monitoring of SCPD characteristics: this is because if the cables of the two feeders have similar crosssections, the two SCPDs play an equal role in breaking, but if the cross-sections are different, there is a risk of only one SCPD, the one with
the lowest rating, performing breaking. It is thus necessary to verify that its breaking capacity on one phase, thus under e U0, is greater than Id. For this reason, circuit-breaker manufacturers specify the single-phase breaking capacities of their devices according to each nominal voltage, and standard IEC 60947-2 specifies a test sequence for circuit-breakers designed for protection of IT networks. Devices failing to satisfy the requirements of these tests must be marked: IT Protection must also be confirmed for the neutral conductor when its cross-section is less than that of the phase conductors. Note that four-pole circuit-breakers (the fourth pole has a half rating) can be used to protect cables with neutral crosssection half of phase cross-section. It should be stressed that four-pole SCPDs are becoming increasingly necessary, irrespective of the installation earthing system used (TN, TT or IT), due to the proliferation of harmonics in networks, and thus that the neutral can be overloaded by harmonic currents of rank 3 and multiples. c Fuse protection The fuse blowing zone is located between two envelope curves. Using the expression of current Id, defined in the previous chapter, and the condition Ifu < Id, it is possible to determine the maximum length of the protected circuit. v If the neutral conductor is distributed: 0,8 U0 S1 . Lmax = 2 (1+ m) I fu v If the neutral conductor is not distributed: 0,8 e U0 Sph Lmax = . 2 (1+ m) I fu
v with the distributed neutral conductor: Sa Lmax = 0,8 U0 2 (1+ m) Im . v with the non-distributed neutral conductor: Sa Lmax = 0,8 e U0 . 2 (1+ m) Im Note that whether protection is provided by fuse or circuit-breaker, the fact of distributing the neutral in IT divides by e the maximum length protected. c Improvement of tripping conditions. When tripping conditions are not satisfied (lengths greater than maximum lengths protected), the following measures can be taken: v reduce the value of Im of the circuit-breakers: however current discrimination between circuitbreakers may be reduced as a result; v increase PE conductor cross-section. The impedance of the return circuit of the double fault current is thus reduced and enables an increase in maximum length for protection of persons. However, although contact voltage will be reduced, the electrodynamic stresses on the cables will increase. v Increase live conductor cross-section. This is the most expensive solution and also results in an increase in three-phase short-circuit currents. v Finally, there is a simple solution that requires no calculation: use of low sensitivity RCDs on very long feeders. This solution is also possible in IT, as the PE conductor is separate from the neutral conductor which is not the case in TN-C. Case of application frames with separate earth connections When an installation supplies a number of separate buildings at a distance from one another, their application frames are often connected to separate earth connections. The impedance of the path of fault current Id is then increased by the resistance of the two earth connections in question, and the condition necessary for protection of persons (respect of maximum breaking times) can no longer be guaranteed by the short-circuit protection devices. The simplest study and installation solution is to use RCDs. Their settings follow the same rules as in TT. To derive maximum benefit from the continuity of supply offered by the IT system, the RCDs must be prevented from tripping on the first fault by not setting their threshold IDn at too low a level, particularly for circuits with a high leakage capacity, while at the same time respecting the U inequation: In < L . RA The thresholds In of the RCDs normally used for this purpose are between 3 and 30 A.
Ifu corresponds to the fuse blowing current within

a maximum time stipulated by the standards. It should be checked that this time is compatible with protection of persons in event of a double fault. Note that use of fuses in the IT earthing system often clashes with the need for multi-pole breaking, including that of the neutral conductor when distributed. c Circuit-breaker protection Protection of persons is guaranteed when the fault current is greater than the circuit-breakers short time delay protection setting. Just as with fuses, the maximum length of the protected circuit can be determined according to the expression of current Id, defined in the previous chapter and the condition Im < Id. The maximum length of the circuit protected by a circuit-breaker is:
4 Special features of the IT earthing system
4.1 Overvoltages in the IT system

Electrical networks can be subject to overvoltages of varying origins. Some overvoltages, such as differential mode ones (between live conductors) affect all the earthing systems. Readers requiring more information on this subject should consult Cahier Technique no. 179. This section is particularly concerned with common mode overvoltages, which mainly affect the IT system as the network is then unearthed: c overvoltages due to insulation faults, c overvoltages due to internal disruptive breakdown in the MV/LV transformer, c overvoltages due to lightning striking the upstream LV network, c overvoltages due to lightning striking the building in which the installation is housed. These overvoltages are particularly taken into account by standard NF C 15-100 which stipulates installation of a surge limiter downstream of an MV/LV transformer and when there is a risk of lightning (overhead lines). Overvoltages due to insulation faults c When the first insulation fault occurs, the phase-to-earth voltage of the sound phases is permanently brought to the phase-to-phase voltage of the network. LV equipment must thus be designed to withstand a phase-to-earth voltage of U0 e and not the phase to neutral voltage U0 for the time required to track and eliminate the fault. This particularly applies to: v Y capacitive filters fitted on many electronic devices; v installation PIM when installed between phase and earth because the neutral is not accessible. When choosing a PIM, it is thus important to verify the voltage of the network to be monitored declared by its manufacturer. These recommendations are specified in particular in standard IEC 60950. c On occurrence of the first fault, a transient overvoltage appears with a possible peak of 2.7 x r U0 (U0 = phase to neutral voltage of the LV network). On a 230 V/400 V network, this value is 880 V, an overvoltage level that is not dangerous for equipment with an insulation of 1,800 V (voltage constraint at power frequency on the LV side as per IEC 60364-4-442). Note that these overvoltages do not cause permanent short-circuiting of the surge limiter. Overvoltages due to intermittent insulation faults Intermittent faults (according to international electrotechnical vocabulary, or restricting or arcing faults in Anglo-saxon literature) behave like a series of transient faults. Experience and theoretical studies show that intermittent faults can generate overvoltages and thus result in equipment destruction. Such overvoltages are particularly observed on MV networks operated with an earthed connection by a tuned limiting reactance (Petersen coil). These overvoltages are caused by incomplete discharging of zero sequence capacity on re-arcing. The zero sequence voltage therefore increases each time the arc is ignited. Assuming that the arc is ignited at the highest value of the phase-to-earth voltage of the faulty phase, and zero sequence voltage increases each time, overvoltages of 5 to 6 times phase to neutral voltage may be generated. Yet again, in the IT system, protection is provided by the surge limiter, and presence of an impedance between neutral and earth encourages rapid discharging of the zero sequence capacity. Overvoltages due to internal disruptive breakdown of the MV/LV transformer Voltage withstand at power frequency of LV equipment is defined in standard IEC 60364-4-442 which specifies their values and durations (see fig. 17 ). c Internal disruptive breakdown between the MV/LV windings. This kind of overvoltage is at network frequency.
Acceptable AC voltages on LV equipment U0 + 250 V (i.e. 650 V in IT) U0 + 1,200 V (i.e. 1600 V in IT) *
Breaking time (s) >5 i5
(*) For an IT network, the voltage U0 must be replaced by the voltage e U0.
Fig. 17 : acceptable AC voltage constraints on LV installation equipment in the IT system for a 230/400 V network.
These overvoltages are rare and their sudden appearance means that the surge limiter, whose certain arcing voltage is set at least at 2.5 times type voltage (NF C 63-150), i.e. for example 750 V for a limiter placed on the neutral of a 230/400 V network, immediately earths the LV network, preventing it from rising to MV potential. c MV/frame internal disruptive breakdown also known as return disruptive breakdown When the transformer frame and the LV network are connected to the same earth connection (see fig. 18 ) there is a risk of LV equipment disruptive breakdown if the voltage Rp IhMT exceeds equipment dielectric withstand, with Rp (earth connection resistance) and IhMT (zero sequence current due to MV disruptive breakdown). One solution is to connect the LV installation frames to an earth connection that is electrically separate from that of the substation frames. However, in practice, this separation is difficult due to frame meshing in MV/LV substations. Consequently standard IEC 60364-4-442 states that the LV installation frames can be connected to the earth connection of the transformer substation frames if the voltage Rp IhMT is eliminated within the stipulated times. Overvoltages due to lightning striking the upstream MV network When lightning strikes the MV network, a wave is transmitted to the live conductors on the LV side as a result of capacitive coupling between the transformer windings. If the installation is in IT, the surge limiter absorbs the overvoltage occurring on the live conductor to which it is connected (neutral or phase) and is short-circuited if this overvoltage is very high: the network can then be compared to a network in TN-S. Experience and measurements have resulted in the following observations: c Overvoltages of around 2 kV occur at the end of short cables (10 m) irrespective of load and earthing system. c Higher overvoltages occur at the end of cables with open end or which supply loads likely to generate resonance. Even with a resistive load, overvoltages exist (see fig. 19 ), caused by wave propagation and reflection phenomena and by capacitive coupling between conductors. In view of the waveform of these overvoltages, the surge limiter is effective on the conductor to which it is connected. Consequently, regardless of the earthing system, we strongly recommend that surge arresters be installed at the origin of the LV network, between all live conductors and
HV
MV
MV
LV
IhMT
RT (Rp)
Fig. 18 : when the substation frames (MV) and the PE earth connection (LV) are connected to the same earth connection, the LV load frames are brought to the potential IhMT Rp.
(kV) System : c IT c TN-S Ph/Ph Ph/PE Ph/N N/PE PE/deep earth 1.62 1.62
0.38 0.36
4.35 4.82
0.20 0.20
4.30 4.72
Fig. 19 : overvoltages, caused by a lightning shock wave, measured at the end of a 50m cable supplying a resistive load.
the earth, if there is a risk of the upstream MV network being directly struck by lightning (case of overhead lines) and especially if the LV network is also at risk. The surge limiter continues to perform its function for MV/LV disruptive breakdown. Overvoltages due to lightning striking the building housing the installation These overvoltages are caused by lightning current flowing through the buildings earth connection, particularly when lightning strikes a building equipped with a lightning rod. The entire earth network then markedly rises in potential with respect to the deep earth. The LV network, immediately earthed by the surge limiter, changes from the IT to the TN-S system if all the application frames are interconnected. The lightning energy thus flown off can be considerable and require replacement of the limiter. In order to minimise these overvoltages on electrical installations, the buildings horizontal and vertical equipotentiality must be the best possible in low and high frequency. A single earth circuit (PE network) is naturally recommended, and use of metal cable trays with proper electrical connections (braids) is highly advisable for distribution.
4.2 Surge limiters

Nominal voltage of a limiter -Un(V) (NF C 63-150) 250 440 660 Arcing U (V) at power frequency 400 < U < 750 700 < U < 1,100 1100 < U < 1,600 Example : limiter to be chosen for a 230/400 V network if connected between the earth and the neutral. if connected between the earth and the phase.
on 1.2/50 impulse wave < 1,750 < 2,500 < 3,500
Fig. 20 : the nominal voltage of a surge limiter must be adapted to network voltage.
The previous section clearly explains why the surge limiter is an essential accessory of the IT earthing system and thus stipulated by standards. It also protects the PIM against overvoltages. Its clipping thresholds for overvoltages at power frequencies and for common mode impulse overvoltages are defined by standard NF C 63-150 (see fig. 20 ). These thresholds are lower than the specified withstand of equipment used on LV networks (230/400 V). It must be connected as close as possible to the MV/LV transformer between neutral and earth, or between a phase and earth if the transformers secondary connection is of the delta or non-distributed neutral kind. NB: c limiters are not necessary on networks downstream of a LV/LV transformer, c standard IEC 60364 does not specify use of surge limiters, as it considers that occurrence of an MV/LV fault is rare. However, when this fault does occur, its consequences are frequently serious. Operation A surge limiter consists of two conductive components separated by an insulating film (see fig. 21 ). Impulse overvoltages generate arcing between the two conductive components, but do not short-circuit the limiter. Energetic overvoltages melt the insulating film, thus allowing the run-off of a high current to earth. The cartridge must then be replaced: its short-circuiting is reported by the PIM just as an insulation fault. Moreover, it is useful, for live fault tracking, to consider its earth connection in the same way as a feeder, particularly if this connection is normally inaccessible (such as when, for example, the limiter is placed in the transformer cubicle). Important characteristic When all the application frames are properly interconnected, the double fault concerning both the arced surge limiter and an insulation fault on
Insulating film disappears during high power overvoltages Arcing zone during low power overvoltages
Insulating case Connection pad
Fig. 21 : surge limiter principle (Merlin Gerin Cardew type).
a phase, becomes a short-circuit. Limiter withstand must then be sufficient for the time required to eliminate the fault current (For example, 40 kA must be withstood for 0.2 sec for Merlin Gerin Cardew limiters). In the rare case of the second insulation fault occurring upstream of the incoming circuitbreaker, the double fault is eliminated by the MV protection devices (just as for an upstream short-circuit on the main LV switchboard). For this reason, the time delay setting of the transformers MV protection must take the thermal withstand [f (I2t)] of the surge limiter into account. The cross-section of the connection conductor upstream and downstream of the surge limiter must also have the same thermal withstand. Its cross-section is calculated in standard NF C 15-100.
4.3 Why use an impedance?

An impedance can be connected between the network and the earth, normally between the transformer neutral and the earth. Its value is approximately 1,700 at 50 Hz. Its purpose is to reduce variations in potential between network and earth, caused by MV disturbances or fluctuations in potential of the local earth. It is therefore particularly recommended for short networks supplying measurement instruments sensitive to this potential and for networks placed next to communication networks (Bus). A reading of the table in figure 5 shows that when the network is very slightly capacitive (case 1), the neutral impedance ZN causes the fault current to increase, which nevertheless remains very low ( 250 mA in figure 5). This effect is even slighter when the network is highly capacitive (cases 2 and 3). In practice, this impedance effects only very slightly the contact voltage UC which remains less than UL in sound networks. Finally, presence of a resistance in the impedance enables a reduction of the ferromagnetic resonance hazard.
5 Advantages and disadvantages of the IT earthing system in LV
The main advantage of using the IT earthing system for network operation is without doubt the continuity of supply it offers, as there is no need to trip on the first fault (as described in the section below). Another of this systems strong points is guaranteed safety against the fire
hazard and for control and monitoring circuits of machine tools. However, to benefit fully from such advantages, the restrictions of this system must also be considered.
5.1 Increased availability

This advantage can be confirmed by a simple probability calculation. Let us assume that the occurrence of an insulation fault in an electrical installation is one fault every three months (90 days), 1 j i.e. = 90 and the time needed to track and put right the faulty part is one day, i.e. = 1 j. The Markof graph technique gives the representation shown in figure 22 and enables us to calculate that the average time between two double faults is 8,190 days! This corresponds to an average electrical power availability that is 91 times better in IT than in TN or TT. Consequently, preference is frequently given to the IT earthing system for use in: c hospitals, c airport take-off runways,
=
No fault
1 j 90 1 fault
1 j 90 2 faults
= 1j
= 1j
Fig. 22 : a Markof graph shows that average electrical power availability is 91 times better in IT than in TN or TT.
c vessels, c plants with continuous manufacturing processes, c laboratories, c cold storage units, c electrical power plants.
5.2 Increased safety against the fire hazard

Electricity is often the cause of fire. Standards set the threshold for this risk at 500mA on an insulation fault (NF C 15-100, part 482.2.10). This value can be considerably exceeded, particularly with stray currents that flow through building structures when faults occur in the TN system. Also worthy of note is that the IT is the only earthing system that monitors insulation of the neutral conductor, compared with the TNS which can insidiously turn into a TNC on a neutral-PE fault with an increase in the fire hazard. It is because the current of the first fault is particularly low that the IT earthing system has been chosen for use in certain establishments at risk from fire and explosion (see chapter 1). Furthermore the first PIMs were used in firedamp mines.
5.3 Less downtime on control and monitoring circuits

The relay diagram illustrated in figure 23 with the TN earthing system shows three possible insulation faults which, when full, result in immediate downtime, whose material and economic consequences are rarely negligible. These faults have the same consequences with the TT system. In particular, faults c and d cause tripping of the master protection device, and prevent all subsequent operations, such as for example the order to change direction on a transporter bridge! These same first faults which can cause operating malfunctions or even accidents with the TN and TT systems, have no effect with the IT system, except if they occur as the second fault (extremely unlikely, see section 5.1). These examples show that even if safety of persons with respect to the electrical hazard is guaranteed by the various earthing systems, or by use of Safety by Extra Low Voltage (SELV), safety of persons with respect to mechanical hazards may not be guaranteed in certain cases. More care must therefore be taken when wiring such circuits in the TT and TN than in the IT system, as the latter warns the operator of the incident (first insulation fault), thereby guarding against electrical and mechanical hazards. PIMs are increasingly used for just this purpose, to monitor automation networks. An additional solution is often advisable, particularly with relays using electronic devices sensitive to electromagnetic disturbances. The aim is to supply all the control and monitoring circuits separately by means of a LV/LV transformer with separate windings. Despite this, as stated in chapter 2, use of the IT earthing system has its limits which are described in the section below.
3 2 1 N PE
Id
A RB a b c
M d a b
M c
A d
Fault a cannot be detected. Fault b prevents the off function. Faults c and d cause a short-circuit.
Fault a cannot be detected. Faults b, c and d cause a short-circuit.
Fig. 23 : monitoring circuit may be concerned by several types of insulation faults always resulting in downtime with TT and TN system.
5.4 Restrictions and precautions for using the IT earthing system

The restrictions for using the IT system are linked to loads and networks. Limits linked to loads c With a high earth capacitive coupling (presence of filters). A number of devices fitted with capacitive filters (see fig. 24 ) offer the same disadvantage, due to their number, as very long networks when the IT system is used. These capacitive leakages have a particularity, with respect to distributed capacity mainly due to network cables, i.e. they can be unbalanced.
Device Micro-computer UPS Fluorescent tubes (in ramps of 10) Network/earth capacity 20 nF to 40 nF 40 nF 20 nF
Variable speed controlers 70 nF
Fig. 24 : guideline capacitive values for HF filters built into various devices.
Da
(A)
1 2 3 N Db (B) RCD 30 mA CF CF
Fig. 25 : in the IT system, capacitive current flow can cause nuisance tripping of the RCDs by sympathy. In this case, in presence of a fault on feeder B, circuit-breaker Da placed on a highly capacitive feeder (presence of several filters) may open instead of Db .
Office computer equipment: micro-computers, monitors and printers, concentrated on the same single-phase feeder, is an example of this. It should be borne in mind that interference suppression filters (compulsory according to the European Directive for EMC) placed on these devices, generate in single-phase permanent leakage currents at 50 Hz that can reach 3.5 mA per device (see IEC 95); these leakage currents add up if the devices are connected on the same phase. To prevent nuisance tripping (see fig. 25 ), especially when the RCDs installed have low thresholds, the permanent leakage current must not exceed 0.17 In in IT. In practice, the supply by a 30 mA RCD of three micro-computer stations is the maximum recommended. This problem also exists with the TT and TN systems. For memory: v to guarantee safety of persons (UC i UL), the limit not to be exceeded is 3C i 70 F. v for insulation monitoring, PIMs with DC current injection are not affected by these capacities. Note that if the devices are connected on all three phases, these capacitive currents cancel each other out when they are balanced (vector sum). c With a low insulation resistance This particularly applies to induction furnaces and arc welding machines, as well as very old cables. A low insulation resistance is equivalent to a permanent insulation fault: the IT system is transformed into a TN or TT system, with a PIM on permanent alert.
PIM A R
PIM B
PIM C
Fig. 26 : insulation monitoring of the various parts of a network with a replacement source.
Limits due to the physical characteristics of networks High capacitive leakages disturb insulation monitoring using PIMs with AC current injection and tracking of the first fault using a very LF generator (see chapter 2). When an insulation fault occurs, they can also cause flow of residual currents likely to generate nuisance tripping by sympathy of the RCDs placed on very long or highly capacitive feeders (see Cahier Technique no. 114). Use of the IT system is thus advised against for very long networks, containing long feeders, for example for electrical power distribution in a number of buildings at a distance from one another. Case of networks with replacement power supply The fact that a network can be supplied by several sources makes it necessary to detect the first fault and to trip on the second fault, irrespective of the voltage source in operation.
c Permanent monitoring of network insulation, regardless of the supply source, makes it necessary to choose PIM position carefully. In some positions insulation monitoring may be partial (see fig. 26 ). Permanent connection of two PIMs at positions A and B is not acceptable as they would each obstruct the other on coupling. On the other hand, position C could be acceptable, as access to supply sources is reserved for authorised users, but there is the risk on source switching of finding that a fault already exists on the new source. It is thus preferable to provide a PIM on each source [A and B] with a relay [R] preventing simultaneous operation of both devices on the same circuit. There are also new insulation monitoring systems which exchange digital data via bus and automatically adapt to network configuration. These PIMs make use of special, often complicated, relays unnecessary (see fig. 27 ). c Tripping on a double fault, whatever the voltage source, calls for (just like the TN system) verification of SCPD compatibility with the presumed fault currents, particularly when the replacement source is a generator set. This is because the short-circuit current it delivers is far lower than that supplied by an MV/LV transformer supplied by the public distribution networks: the SCPD threshold must be set accordingly. The first solution is to lower the threshold of these protection devices, but the problem is that this also reduces current discrimination ability. A second, simpler, solution, is to provide low sensitivity RCDs. c With an Uninterruptible Power Supply (UPS) The problems experienced are the same as with
a generator set. Only insulation monitoring is more complex as it is linked to the various operating configurations that a UPS can assume. In practice In all these restriction cases, the most appropriate solution is to reserve use of IT for the network only supplying the devices requiring a high degree of electrical power availability. c For existing installations, in order to restore continuity of supply, it is necessary to identify low insulation feeders and supply them separately with, for example, a TN system, and save the network in IT for the more demanding applications. This solution requires use of a new transformer, either LV/LV, or directly MV/LV, according to power requirements. A similar approach can be applied in order to supply machines experiencing operating problems in IT. c For new installations, electrical power distribution in IT must be provided as soon as the need for continuity of supply is apparent. It is preferable, in order to reduce the incidence of network capacity with respect to earth, to limit the size of this network to a building, for example. Finally, if phase to neutral voltage is required, distribution of the neutral conductor should not be a cause for concern as: v neutral insulation is monitored, v use of circuit-breakers with B or G curve and of low sensitivity RCDs simplifies the protection study, and avoids the installation (and thus the additional cost) of a specific transformer or a special line.
Communication XTU interface
Information exchange bus
XM300C
PIM
XM300C
PIM
XM300C
PIM
Fig. 27 : insulation monitoring system of the various parts of a network with several sources (Merlin Gerins Vigilohm System).
6 Conclusion
Evolution of the various earthing systems should mirror the changing needs of electrical power users.
6.1 Availability: an increasing need to be satisfied

The increasing number of computer, automation and control/monitoring equipment has resulted in all major economic entities (industrial, commercial, etc.) calling for greater availability of electrical power. Today, electrical power is considered to be a simple product with which a number of quality criteria, particularly availability, are associated. To ensure that users benefit from this increased availability, this demand, already acknowledged by electricity utilities, must also be incorporated in the design of the new internal and private distribution installations precisely there where the IT earthing system assumes its full importance by indicating the very first fault (not dangerous) and preventing tripping. However, for the network to benefit from all the advantages of the IT system, designers must carefully consider the future operation of the network and have excellent knowledge of the devices to be supplied.
6.2 The IT earthing system finds its true place

Usable in a very large number of electrical installations The IT earthing system can be used in a very large number of electrical installations in industrialised countries, with the exception of applications (e.g. arc furnace, old lighting circuit) and situations (e.g. damp environment, very long network) normally or frequently exhibiting a low insulation level. These countries possess skilled electricians, sufficiently reactive to offer rapid installation servicing (the same day). Moreover, their infrastructures allow use of remote supervision. For adapted distribution circuits Changes in continuity of service requirements and implementation of new machines with specific characteristics, particularly in the field of electromagnetic compatibility (EMC) mean that the electrical power supply sometimes requires specially adapted distribution circuits. This accounts for the emergence of private distribution networks comprising a variety of subnetworks with an appropriate earthing system. In these conditions, the IT system easily guarantees the necessary continuity of supply.
6.3 The added advantage of safety

Installation designers must also identify fire and explosion hazards and satisfy EMC requirements (disturbance of measurements and communications). The IT earthing system offers the most advantages and best meets operators requirements with such specific features as: c better EMC (interconnection of frames and in theory a single earth connection), c minimum fire and explosion hazards (low first fault currents). Moreover, its use is encouraged by the upgrading of equipment (PIM, tracking device, supervisor, etc.) allowing: c anticipation of maintenance (prediction), c quicker tracking of the first insulation faults (automation), or even remote tracking (remote supervision via digital connections), c preparation of troubleshooting (remote diagnosis).
6.4 In short
Our readers now understand the importance of properly listing the requirements relating to equipment used, the environment and the study conditions of the installation and subsequent modifications, before choosing the earthing system for an electrical distribution network. A brief reminder of the advantages and disadvantages inherent in each earthing system is essential at this point: this is the purpose of figure 28 . NB: The installation cost is not included in this table as the possible additional cost of an IT system (PIM, fault tracking system) must be compared with the financial loss generated by unexpected downtime on the first fault this must be evaluated for each activity.
TT Safety of persons (perfect installation) Safety of equipment c against the fire hazard c for machine protection on an insulation fault Availability of electrical power Electromagnetic compatibility For installation and maintenance c skill c availability c c v v c c c excellent c c good v average poor c c c c c c c c c v v v v c c v
TN-C c c c v v v v v c c c c v v
TN-S c c c v v v v v v v c c c c v v
IT c c c c c c c c c c c c c v v c c c c c c
Fig. 28 : summary of the advantages and disadvantages of the various earthing systems.
Bibliography
Standards and decrees c IEC 60364: Electrical installations of buildings. c IEC 60479-1: Effects of current on human beings and livestock. c IEC 60947-2: Low voltage switchgear and controlgear - Part 2: Circuit-breakers. c IEC 60950: Safety of information technology equipment. c IEC 61000: Electromagnetic compatibility. c IEC 61557, NF EN 61557: Electrical safety in low voltage distribution systems up to 1,000 V AC and 1,500 V DC - Equipment for testing, measuring or monitoring of protective measures -. Part 8: Insulation monitoring devices for IT systems. c NF C 15-100: Installations lectriques basse tension. Schneider Electrics Cahiers Techniques c Neutral earthing in an industrial HV network. F. SAUTRIAU, Cahier Technique no. 62 c Residual current devices. R. CALVAS, Cahier Technique no. 114 c EMC: Electromagnetic compatibility. F. VAILLANT, Cahier Technique no. 149 c Harmonics in industrial networks. N. QUILLON, P. ROCCIA, Cahier Technique no. 152 c Calculation of short-circuit currents. B. De METZ-NOBLAT, G. THOMASSET, R. CALVAS and A. DUCLUZAUX, Cahier Technique no. 158 c Earthing systems in LV. R. CALVAS, B. LACROIX, Cahier Technique no. 172 c Earthing systems worldwide and evolutions. R. CALVAS, B. LACROIX, Cahier Technique no. 173 c Perturbations des systmes lectroniques et schmas des liaisons la terre. R. CALVAS, Cahier Technique no. 177 c LV surges and surge arresters - LV insulation co-ordination -. Ch. SERAUDIE, Cahier Technique no. 179 c Intelligent LV switchboards. A. JAMMES , Cahier Technique no. 186 c Cohabitation of high and low currents. R. CALVAS, J. DELABALLE, Cahier Technique no. 187
Schneider Electric
DTP: AXESS - Saint-Pray (07) Edition: Schneider Electric Printing: Imprimerie du Pont de Claix - Claix - France - 1,000 - 100 FF 06-99
28289
n 180
earthquakes and electrical equipment
Eric MELMOUX Graduated in mechanical engineering from the INSA School, Lyon in 1981, and obtained a DEA (equivalent to 1st year of Phd) in vibrations the same year. After ten years spent with a company which specialises in sound and vibrations, he joined Merlin Gerin. He is, at present, head of the shocks and vibrations group at Schneider Electric.
E/CT 180, first issued, April 1997
glossary
Epicenter Frequency appropriation Hypocenter or seismic focus Intensity LOVE Waves Magnitude Meshing MHEL Modal shape Qualification RAYLEIGH Waves Response Spectrum Seismic Activity Single DOF mechanical system Space appropriation SSE Strong part of response spectrum Time-history Zero Period Acceleration ZPA part point at ground level, vertically above the hypocenter. excitation frequency corresponds with resonance frequency of the structure. position of the earthquake in the earths crust. measures the force of the earthquake in terms of the effects produced (MERCALLI scale). correspond to the horizontal component of the surface waves. measures the force of the earthquake in terms energy released at the seismic focus (RICHTER scale). action whereby a complex structure is broken down into its beam gantrys - plates volume. Maximum Historical Earthquake Likelihood for a site. oscillating deformation adopted by an elastic structure during excitation on one of its resonance frequencies. process which consists in establishing the appropriate withstand capabilities for equipment with required or normal stresses. correspond to the vertical component of the surface waves. device which enables a characterization of the earthquake to be effectuated in terms of its effects on a simple structure. violent movement of tectonic plates which produces an earthquake. single degree of freedom structure characterized by mass, spring and damper. excitation forces apply pressure on the antinodes of the modal shape. Safe Shutdown Earthquake (MHEL plus one degree on the MERCALLI scale). corresponds to the frequencies which cause the structure to amplify ground movements. recording of ground acceleration during the earthquake. corresponds, on a response spectrum, to the frequencies which cause the structure to trace ground accelerations without amplification of motion.
earthquakes and electrical equipment
contents
1. Earthquakes Causes - location p. 4 Propagation of seismic waves p. 4 Characteristics of vibrations p. 4 generated at ground level Intensity and magnitude p. 4 Seism characterization p. 6 Defining the seismic severity p. 8 of a site Reading the response spectrum p. 9 applicable to a piece of equipment Brief summary of single degree p. 12 of freedom oscillator Elastic structures (with N p. 12 degrees of freedom) Defining objectives p. 16 Design principles p. 16 Simulation by analysis at p. 18 design stage Introduction p. 20 Combined qualification p. 20 (numerical analysis and experimental tuning) Qualification by real size p. 21 tests preceded by numerical analysis Qualification by test p. 22 p. 25 p. 26
In all countries there is either a zone of significant seismic activity or installations which require high security in order to operate (e.g. nuclear power stations, which generally have low seismic activity). In both cases the electrical and control and monitoring equipments must assure their safety functions correctly. This technical paper aims to facilitate dialogue between operators and specialists. After briefly summarizing the earthquake phenomenas and the way in which they are specified, the author presents the theoretic approach required for seismic withstand capabilities to be taken into account at the design stage. Both design and qualification are, today, increasingly requiring numerical analysis and, as a result, powerful scientific and technical data processing methods.
2. Dynamic behaviour of structures
3. Equipment design
4. Qualification by simulation or test
1. earthquakes
causes - location
The majority of earthquakes occur on the fault lines which demarcate the tectonic plates from the earths crust. Tension accumulates when the plates slowly shift away from each other. The sudden release of distorting energy which has thus accumulated inside the earths crust, or in the underlying layer called the mantle, provokes a local agitation. Some of this energy is then transformed into seismic waves at the surface of the ground. It is the creation of a fault line, or more frequently the slide along an existing fault line, which constitutes the generating mechanism of an earthquake. The place where it occurs is called the seismic focus or the hypocenter and the projection from this point to the ground is called the epicenter (see fig. 1). The depth of the hypocenter varies greatly: from a few kilometres to up to 100 km.
direction of the propagation, which provoke shear and propagate at a speed of 4 to 5 km/s (see fig. 2). Remark: It is the difference in speed between the longitudinal and transversal waves and recordings taken from several seismographs, which enables an earthquakes seismic focus to be located. Surface waves These are generated by the bulk waves which reach the surface and propagate at a speed of 1.5 to 5 km/s. A distinction is made between: c RAYLEIGH waves which cause the ground points to describe ellipses in the vertical plane; they engender compression and shear in the ground, c LOVE waves which cause the ground points to shift at a tangent to the surface, perpendicular to the propagation direction; they engender shear only (see fig. 2).
c duration It is usually between 15 and 30 s (an intense earthquake can last between 60 and 120 seconds). c frequency Broad band random movement produces preponderant energy between 1 and 35 Hz, and provokes the most destructive effects at between 1 and 10 Hz. c level of acceleration There is no correlation between the waves observed in the two different directions: at any given moment the amplitudes and frequencies are independent. Horizontal ground acceleration is generally lower than 0.5 g (exceptionally higher than 1g, or 10 m/s2). Vertical acceleration has a lower amplitude. Observations show that the relationship between the maximum vertical and horizontal amplitudes is in the order of 2 (for frequencies higher than 3.5 Hz).
3
propagation of seismic waves

Earthquakes propagate in waves which, taking the heterogeneity of the ground into account, provoke a complex vibratory movement at the surface which is difficult to predict for a given site. A distinction is made between two types of wave: bulk waves and surface waves. Bulk waves They originate at the seismic focus and propagate inside the earths mantle in two different forms: c longitudinal waves characterised by alternating compressions and dilations which propagate at a speed of 7 to 8 km/s, c transversal waves characterised by a plane distortion perpendicular to the
characteristics of vibrations generated at ground level

In reality things are far more complex; the propagation of a seismic wave in a heterogeneous environment provokes a complex system of refracted and reflected waves for each discontinuity, so that the seismic movement is completely random at ground level. However, vibratory movements brought about at ground level by earthquakes do produce common characteristics, and a certain number of parameters are generally employed to describe them. Characteristics of random vibrations provoked at ground level by an earthquake: c direction The movement is made up of simultaneous independent vertical and horizontal components;
intensity and magnitude

Intensity The scale of an earthquake is generally measured in terms of its intensity at the observation site. This subjective evaluation is established in terms of the effects felt by the population and the damage incurred. Different intensity scales have been defined, which class the seismic effects in order of increasing size, with the help of some conventional descriptions: c the MERCALLI scale describes commonly observed effects on the environment, buildings and man after an earthquake, c the MSK scale (or modified Mercalli scale), more precise than the original, includes an evaluation of the damage,
the type of building and percentage of buildings affected. These estimations are useful for assessing the size of earthquakes if there are no time-histories or spectra available. They do not enable the seismic load of a site to be specified.
Magnitude Another method of determining the scale of an earthquake is to establish its magnitude, the intrinsic characteristic of an earthquake which measures the total energy released. Magnitude, defined in1935 by RICHTER, and the scale which ensued, are used universally. Magnitude is, in practice, determined according to ground movement recordings taken at a number of observation points at various distances from the epicenter. From these observations seismologists calculate the energy E (expressed in ergs) of the earthquake, from which the magnitude M is deduced. The simplified empirical equation: log E = 9.9 + 1.9 M + 0.024 M2 gives an approximate, but rapid calculation. Intensity / maximum ground acceleration / seismic zone correlation The table in figure 3 (see overleaf) establishes a correlation between the various subjective levels of intensity in the modified MERCALLI scale and the maximum ground acceleration level responsible for the damage observed. This table also indicates the type of seismic activity zones prone to such seismic intensity (see fig. 4, page 7 for the division of the worlds seismic activity zones).
,, ,, ,,

epicenter distance from epicenter earth focal distance rock hypocenter or seismic focus fault line
fig.1: geosismic vocabulary and characteristic features.
ground level
vertical waves R Q
horizontal waves
bulk
P: longitudinal waves S: transversal waves
Intensity / magnitude correlation Theoretically, no relationship can exist between intensity and magnitude; intensity is dependent on the distance from the site in question to the seismic focus, on soil type, the type of foundations used, the type of building and the duration of the earthquake. However, an approximate correlation is proposed by the experts (see table in figure 5, page 7).
hypocenter
fig.2: seismic bulk and surface waves.
intensity
modified Mercalli scale
approximate horizontal acceleration values in m/s2
seismic zone
1 2 3 4
Tremor not felt. Tremor felt by those resting or situated on top floors. Hanging objects move. Slight vibrations. Vibrations similar to those provoked by a heavy lorry passing are felt. Window panes and crockery rattle. Stationary vehicles sway. Tremor felt outside buildings. Those sleeping are woken up. Small objects fall down. Pictures are tipped sideways. Tremor felt by everyone. Furniture is moved. Damage: broken glass, objects thrown from shelves, slight cracks appear in plastering. Tremor felt in moving vehicles. Those standing fall off balance, church bells ring. Damage: chimneys and other external architecture break away, plaster falls down, broken furniture, extended cracks in plastering and masonry, some houses in brick collapse. Drivers in moving vehicles experience difficulties. Branches fall from trees. Fissures appear in waterlogged ground. Destruction: water towers, monuments, brick houses. Slight to subtantial damage: brick buildings, prefabricated houses, irrigation works, causeways. Sand craters in sandy and waterlogged ground in urban areas. Landslides. Fissuring in ground. Destruction: non-reinforced brick masonry. Slight to substantial damage: insufficiently reinforced concrete structures, underground piping. Landslides ans major ground destruction. Destruction: bridges, tunnels, certain reinforced concrete strutures. Slight to substantial damage: the majority of buildings, dams, railway lines. Permanent ground deformation. Almost total destruction.
zone 0
5 6
zone 1
zone 2
zones 3 and 4
10
11 12
Seismic zones correspond to the anticipated level of intensity according to observations carried out over a period of 200 years.
fig. 3: Mercalli scale.
seism characterization
Intensity, magnitude or maximum ground accelerations do not suffice when estimating the risks to a building or a piece of equipment. In fact, in order to estimate the response of a given structure, a more detailed knowledge of the duration and frequency of the ground movement is required. There are two methods for characterizing ground movement: c time-history: = f(t);
c response spectrum which characterizes the effects produced by the seism on an elementary structure (1st order linear mechanical system). Time-history Ground acceleration evolution as a function of time (see fig. 6). This type of information, recorded by seismographs according to the three spatial directions, is used to estimate the seismic risk incurred by equipment, when determining the withstand either by test or analysis.
The time-history is the only possible option for determining the chronology of a structures response to seismic excitation, which is required when ascertaining the evolution of the different com-ponents relative displacement over time. However, this rarely figures in specifications sheets either because it is not available or because it does not lend itself to the seismic severity calculations of a site.
seismic zone 0-1 2 3 4
fig. 4: seismic activity zones in the world.
Mercalli (intensity) Richter (magnitude)
1 0-2
2 1-2
3 2-3
4 3-4
5 4-5
6 5-6
7 5-7
8 6-8
9 7-9
10 - 11 - 12 >8
The correlations between the Mercalli and Richter scales are completely indicative because they depend on soil type, distance from the seismic focus (from 5 to 100 km), and earthquake duration.
fig. 5: indicative correlation between Mercalli and Richter scales.
Response spectrum The response spectrum allows the earthquake to be characterized in terms of the effects it produces on equipment. For this, the effect of the time-history (in seismic waves) is calculated on standardized equipment, that is to say, on an array of single degree of freedom oscillators, caracterised by their reso-nance frequency and damping values. A single DOF system is characterized by: a mass M, spring K and damper ; its resonance frequency is:
Fr = 1 2 K . M
m/s2
(max)
10
12
14
16
18
20
22
24
t (s)
fig. 6: sample time history = f(t) record of horizontal north-south ground acceleration, El Centro - California, Mai 18, 1940.
The maximum response of this system to the seismic wave (maximum acceleration of mass) gives one point on the response spectrum (see fig. 7). By varying the resonance frequency (K/M), the curve obtained: max = f (Fr) (see fig. 8) is the response spectrum which characterizes the severity of the seismic wave for a given damping. Figure 9 shows the family of curves obtained when the damping is modified. The response spectrum generally figures in specifications sheets, so that it can be applied in horizontal directions. The vertical response spectrum is deduced by using a coefficient. The response spectrum is the most widely used tool today for determining the seismic severity of a site, because it lends itself naturally to: c severity comparison, c elaboration of severity envelopes for several sites, c simple severity increases, c approximate estimates of seismic effects on equipment (damage potential). Remark: The response spectrum must not be confused with the FOURIER series expansion of a periodic phenomenon, or the FOURIER transform of an aperiodic phenomenon, which are not used in seismic studies.
M
K
M: mass K: stiffness : damping
single DOF systems
max
t
Fr =
1 2
K M
Fr
time-history
spectrum
fig. 7: application of seismic excitation (time-history, see fig. 6) to 1st single DOF, induces accelerations. The maximum value (max ) is by definition one point on the response spectrum of the seismic system.
M1
K1
Mi
Mn
Ki
Kn
defining the seismic severity of a site

Safe Shutdown Earthquake (SSE) Maximum Historical Earthquake Likelihood (MHEL) Defining the seismic severity of a site generally requires the sites geological data and seismic history. In France for example, data from the exceptionally well documented seismic history (100 years) enables the seismic risk of a site to be established. This allows the Maximum Historical Earthquake Likelihood to be defined
response spectrum
Fr1
Fri
Frj
Frn
Fr
fig.8: construction of seismic response spectrum (various K/M with constant).
which is likely to provoke the maximum effect on a given site. For the dimensioning of works or equipment it is the SSE which is taken into account: the SSE is equivalent to the MHEL plus one degree on the MSK scale (modified Mercalli scale). Basic response spectrum Macro seismic data which correspond to the above definitions are not sufficient for the engineer who has to design a building or an equipment. He will also require the representative response spectrum of the site concerned, which is established by using instrumental seismic data. A seismotheque has been created (readings taken in regions of considerable seismic activity), which corresponds to a scale of magnitudes, seismic focus depths and epicentral distances for very diverse geological contexts. This seismotheque allows the form of the response spectrum, or basic response spectrum as it is called, to be established, for a given region, with its amplitude depending on the chosen SSE. This response spectrum defines seismic severity at ground level. The seismic severity for the storey where the equipment will be installed still has to be evaluated. Dimensioning spectrum Seismic withstand specifications are widely presented in the form of a family of response spectra for each storey. These are calculated by taking the buildings transfer impedance into account. An example is given in figure 10.
1 > 2 > 3
Fr
1 :
2 :
3 :
Acceleration / velocity / displacement conversion Response spectra are often represented in an acceleration/ frequency system of coordinates but are sometimes represented in the velocity/frequency system of coordinates. For low damping of equipment studied (i10%), the response spectra measured in terms of velocity and the relative displacement can both be deduced from acceleration spectra by applying the following equations to each frequency:
V max =
fig. 9: family of response spectra obtained for different dampings during the same earthquake.
max ;
2f
D max =
(2 f)2
max
(g)
6 5 4 +7 3 0 2
+ 27
1 0.8 0.6 0.5 0.4 0.3
reading the response spectrum applicable to a piece of equipment

The benefit of the response spectrum is that it visualises the extreme acceleration effects (or displacement effects) provoked by excitation on a single DOF system.
0.2
Fr (Hz) 0.1 1 2 3 5 10 20 30 50
fig. 10: dimensioning spectrum, according to floor levels (in metres) for an industrial site. This is a spectrum for a damping of 2 %.
In fact, everything occurs as if sinusoidal quantities were involved, with
velocity (cm/s) 1000

er at io n (g 20 ) 0
Maximum floor acceleration and displacement Since the energy from the seismic excitation is limited to a frequency of 35 Hz, the points on the spectrum situated above this frequency represent the behaviour of a rigid oscillator (very high K/M), which remains dimensionally stable under seismic excitation. The relative displacement of the mass in relation to the support is therefore zero and its acceleration is equivalent to the supports acceleration (see fig. 12 a). The high frequency asymptotic curve on the response spectrum (Fr u 35 Hz) corresponds therefore to the maximum floor acceleration (see right-hand part of figure 13). Remark: For the right hand part of the spectrum (which corresponds to the infiniterelative frequency), experts use the abbreviation ZPA (Zero Period Acceleration) to establish the acceleration level. In the same way, the lowest frequencies on the spectrum represent the behaviour of an oscillator which remains infinitely supple under seismic excitation. The relative displacement of this type of oscillator equates to the displacement of the support (see fig. 12 b). At low frequency the asymptotic curve on the response spectrum, when represented in log/log scales, corresponds to the ground displacement zone (see left-hand part of figure 13). Maximum acceleration and displacement of oscillator Between 1 and 35 Hz (central part of figure 13) the oscillator accelerations and displacements are generally higher than the floor equivalents.
10 0
100
50
20
5 10 20
2
50
10
5
02 01 0.
0.
05 1 0. 0. 5 1
10
0.
0.
00
0.
2
2 00
0.
0.
0.
20
05
0.
02
01
1
0.
0.
0.
00
0.
10
20
0.5 0.1 0.2 0.5 1 2 5 10
0.
00
20
50 di sp la ce m en t( cm
200
damping (%)
frequency (Hz) 50 100
fig. 11: sample response spectrum which can be read along acceleration, velocity and displacement axes.
a) with a very high K/M ratio the system does not deform (the mass adopts the ground displacement) = sol b) with a very low K/M ratio the system deforms (the mass remains immobile) =0
Fr = "infinite"
Fr = 0
support
fig. 12: response on a single DOF system for maximum values of its resonance frequency.
10 0
V = (t) dt et d = (t) dt With log/log coordinates the response spectrum can be read along acceleration, velocity or displacement axes, (see fig. 11).
500
ac ce l )
Maximum acceleration and displacement, as well as the corresponding resonance frequencies are read directly off the spectrum (see fig. 13 readings along axes and d). Where the dynamic characteristics of an equipment are not known, it is assumed that the equipment comprises a certain number of single DOF oscillators. The majorant values of the maximum displacement and maximum acceleration it will be subjected to during the earthquake, can be ascertained from the response spectrum. Choice of damping for equipment The damping factor taken into account in the response spectrum analysis is supposed to represent the global damping for the equipment concerned. If the equipment comprises components with different dampings, then it is normal practice to work with the smallest, with regard to the choice of response spectrum; this leads to an increase in the stresses. The table in figure 14 shows, indicatively, the values commonly agreed upon in terms of the percentage of the equipments maximum stress. Since seismic specification is generally expressed in terms of a family of response spectrum which corresponds to the different damping (2 %, 5 %, 10 %, etc.); the equipment designer can then effectuate an interpolation. Benefits of the response spectrum The information provided on an examination of the response spectrum is much more useful to an equipment designer than that provided by the temporal representation of an earthquake. In fact, while the timehistory provides the maximum floor acceleration, the response spectrum provides a maximum amount of information, noteably, the following maximum values: c maximum floor acceleration, c maximum floor displacement, c maximum acceleration for a part of the equipment, c maximum displacement for a part of the equipment.
d d = dmax of the "system"
max of the "system"

strong part of spectrum
= max of the ground
d = dmax of the ground
1 Hz
35 Hz
Fr
fig. 13: response spectrum reading (in log/log scale it is possible to read the values which characterize the acceleration and displacement).
type of structure damping in % for 50 % of the maximum yield 2 4 4 2 2
welded steel structures bolted steel structures reinforced concrete structures cabinet rack
for 100 % of the maximum yield 4 7 7 5 5
fig. 14: commonly agreed upon dampings for diverse structures according to the yield of the stress (deflection or traction/compression).
2. dynamic behaviour of structures
Today, the study of the dynamic behaviour of structures consitutes an essential stage in the design of all industrial equipment. At this stage, it is appropriate to present the main concepts which govern a structures response to seismic excitation. Refer to the bibliography for further reading.
elastic structures (with N degrees of freedom)

Resonance frequencies and modal shapes Generally speaking an elastic structure (for example a whip antenna, a beam gantry), is characterized by a multitude of resonance frequencies (theoretically infinite in number) which correspond to the resonance modes or characteristic modes, (these structures have N
brief summary of single degree of freedom oscillator

The single DOF oscillator, constitutes one of the basic principles in dynamic analysis of structures. In fact, the dynamic behaviour of an elastic structure amounts to the behaviour of a certain number of basic oscillators. Furthermore, it is often the case that the inclusion of the first mode of resonance for a given structure suffices for dimensioning; this equates to studying a basic equivalent oscillator (two types are shown in figure 7). The basic oscillator is characterized by its resonance frequency or natural frequency, and by its damping. The resonance frequency corresponds to the free movement of the oscillator, with no external force. In other words, it concerns the frequency of the oscillators displacement when it is distanced from its rest position (free oscillation test) or on impulse. When the oscillator is excited to this frequency, resonance is produced, that is to say, the movement is amplified. This amplification is inversely proportional to the damping of the oscillator. Resonance frequency and damping are sufficient for calculating this systems response under any excitation, and in particular, excitation by support displacement which constitutes the earthquake.
degrees of freedom), (see fig. 15). Each of these resonances is accompanied by a specific deformation of the structure termed the modal shape. For each resonance frequency, the structure deforms and oscillates on both sides from its rest position (the points of the structure evolve in phase and in antiphase and produce nodes and antinodes in the modal shape). The higher the mode, the more complex the corresponding modal
first mode modal shape
rest position
second mode modal shape
third mode
nodes modal shape
antinodes
fig. 15: modal base: primary resonance modes of two basic structures.
shape becomes, with an increasing number of nodes and antinodes. The dynamic behaviour of structures, comprising N basic structures with one degree of freedom, (see fig. 16), is usually determined by using what is known as the modal analysis of the stucture, which consists of tracing the resonance frequencies and modal shapes on the frequency band that corresponds to the earthquake. This establishes a work basis comprising the structures N primary modes, called the modal base, in which the initial problem will be reduced to the study and combination of N single DOF systems (see fig. 16 right). Frequential and spatial adaptation An elastic structures resonance is obtained on two conditions: c that the excitation frequency coincides with the structures resonance frequency. Here it is a matter of frequential appropriation (only condition required for systems with 1 DOF), c that the excitation direction as well as its localization are coherent with the corresponding modal shape. If pontual, the excitation must not act on one of the structures nodes, and is all the
more efficient when it acts on an antinode in a direction parallel to the antinode displacement.
Where multiple excitations occur, they also have to respect the phase relationships of the modal shape (see fig. 17).
"appropriated" excitations excitation
"non-appropriated" excitations
excitation
or excitation
or excitation excitation
excitations in antiphase
excitation
excitations in phase
fig. 17: examples of spatial appropriation for punctual excitations.
structure with N degrees of freedom
in the modal base
M1
M2
m1
Mn
m2
mn
fig. 16: complex structure with N degrees of freedom.
This second condition applies to spatial appropriation, which is characterized by the coefficient known as the modal accommodation ratio: t. Response under support displacement excitation When a structure undergoes floor acceleration (t) (ie. a earthquake) everything happens, supposing that the floor is fixed, as if each element of mass m of the stucture was subject to an inertia load Fi = - mi (t). These Fi loads are in phase according to the above equation; which reveals the importance of the notion of spatial appropriation in the response of each of the structures modes. This particularity of seismic excitation results in primary mode responses being favoured, because these modes show, for a given direction, a minimum of vibration nodes. In other words, most of the points of the structure move in phase and verifie the spatial appropriation (see fig. 18). From an analytical point of view, the structures response takes the form of a linear combination of the modal shape of Di of each mode:
D = n yi (t) Di i = 1
m m m m m m m
- m s - m s - m s - m s - m s - m s - m s
s
- m s - m s - m s - m s - m s - m s - m s - m s - m s - m s - m s - m s - m s - m s - m s - m s - m s - m s - m s - m s - m s
first mode appropriated to 100 %
second mode appropriated to 60 %
third mode appropriated to 30 %
fig. 18: equivalence between support displacement and punctual loads. Example of spatial appropriation of the first three modes of a clamped-free beam.
with response spectra and modal shapes for a structure, the yi coefficients are obtained by: yi = i t i where i = spectrum acceleration for Fi, and t i = modal appropriation ratio which corresponds to the Di mode.
The summation of the Di modal shapes with the yi coefficients gives the maximum displacement (or acceleration) for the different points of the structure during the seism (see fig. 19). However, the response spectrum does not show the exact time at which the
maxima occur for each mode. The arithmetic summation in absolute value leads to a pessimistic evaluation, and is as a result only used when modes present neighbouring frequencies (10 %). A quadratic summation is therefore more appropriate for modes which do not present neighbouring frequencies.
D1 primary modes (frequencies Fri, modal shapes Di)
D2
D3
Fr1 appropriation ratio (ti) t1 = 100 %
Fr2 t2 = 60 % ground horizontal acceleration response spectrum
Fr3 t3 = 30 %
2 1 3
Fr1 Fr2
Fr3
Fr
resulting deformation of the structure : D = t1 1 D1 + t2 2 D2 + t3 3 D3
1 D1
2 D2
3 D3
t1 1 D1
t2 2 D2
t3 3 D3
resulting modal shape D
fig. 19: to obtain the deformation for all the points of a structure, when the appropriation ratios and the corresponding max are known, a geometric sum suffices.
3. equipment design
defining objectives
If the design is to take the seismic stresses fully into account, then the degree of safety it has to assure during and after the earthquake must be known. The demands made on the behaviour of equipment exposed to earthquakes are: c stability (equipment must not become projectile), c integrity (equipment must conserve its initial geometry), c functioning (equipment must assure either complete, partial or degraded functioning). While the designer is only concerned with the equipments mechanical withstand capability as far as the first two demands are concerned, the procedure is more complex for the third demand, since the different functional aspects have to be taken into account. This is usually the case for electrotechnical equipment for which operating conditions, in the eventuality of an earthquake, are usually similar to nominal operating conditions. The characteristics of the structure which transfer the excitations must therefore be defined according to the functional cells fragility threshold. See fig. 20 for vocabulary. Mechanical withstand capability In order to check an equipments mechanical withstand capability, it must be ensured that there is good dimensioning of the ground supports and that the structures safe stresses are not exceeded. These dimensioning criteria are entirely dependant on the position of the first resonance frequencies envisaged for the seismic sollicitations. As far as electrotechnical equipment is concerned, the nature of the transfer of ground seismic sollicitation to functional cells, must be taken into account at the design stage. The fragility threshold of functional devices
(level of acceleration at which the equipment no longer performs its function) has to be determined beforehand. Functional withstand capability The vibratory stresses to which the functional devices will be subjected must be evaluated, and there must be assurance that they would function correctly when put under these stresses, or that their availability is unaffected. There are two possibilities: c the functional device is a protective or monitoring device produced in series: the equipment generally undergoes a vibratory environment qualification, the results of which can be exploited in order for the seismic withstand capabilities to be evaluated. Otherwise, the equipments behaviour on the seismic excitation range (0-40 Hz) must be studied, c the functional device is a special device, in which case an evaluation by test is necessary.
In certain cases, an analysis of the tests carried out on an analogous equipment, can provide the technical elements which will reveal the functional withstand capabilities of an apparatus.
design principles
It is essential that the transfer notion figures in the design. In fact, electrical equipment generally comprise a frame or skeleton (structure) which, in the eventuality of an earthquake, transmits more or less faithfully the floor vibrations towards the functional devices. Response spectrum and dynamic characteristics of the structure The response spectrum, which represents the floor movements, allows: c on the one hand, the excitation severity the structure is likely to undergo to be determined (in terms of acceleration and displacement (see end of first chapter)),
functional device
functional cell
frame
floor
ground
fig. 20: definition of the terms used for an equipment (LV cabinet, HV circuit-breaker).
c and on the other hand, to determine whether or not the structure will amplify the seism, in view of the position of the resonance frequencies with respect to the strong part of the response spectrum. It is therefore imperative that the designer knows the structures primary resonance frequencies; these can be estimated by analysis, tests or by analogy, remembering that a spectrum consists of two zones (see fig. 21): c the right hand side of the spectrum, during which the equipment has the same accelerations as the ground without amplification. When the equipments resonance frequencies are in this zone, the equipments mechanical behaviour is termed static equivalent or pseudo static. Evaluation of maximum stresses is obtained by successively applying the maximum ground acceleration (ZPA or Zero Period Accele-ration) to the masses concerned according to the three spatial directions. c the strong part of the spectrum, during which the structure amplifies the accelerations by its resonances, which leads to increased forces and stresses than was previously the case. In this zone, the equipments mechanical behaviour is dynamic and, as a result, it is necessary to know the frequencies and modal shapes and to combine them in order to evaluate the maximum possible damage that the equipment might incur. The process therefore consists in: v characterizing the natural vibrational modes (Fr, Di), v determining the modal responses, v superimposing the modal responses, v deducing the forces and stresses induced. State of the art seismic design Applying the above procedure will obviously avoid the equipment having resonance frequencies in the strong part of the response spectrum. Where possible, the designer should endeavour to limit the equipments dimensions and in particular its height (small dimensions favour increased frequencies).
This being the case, the most widely used solution is the stiffening of the structure to discard the first resonance frequencies higher than the seismic excitation range, or at the very least higher than the strong part of the response spectrum. In any case, it is sensible to avoid resonance modes situated in the 0-10 Hz, or indeed 0-15 Hz frequency band, in order to increase safety. Putting these basic concepts into practice does however have to fall in line with restrictions imposed by the equipments cost, dimensions and functioning. When an equipment is obviously too fragile and can not be sufficiently stiffened, it is possible to isolate the equipment from the ground by interposing a suspension stage.To be effective, the suspension must however possess the characteristics required by very high-performance damping plugs (suppleness and deflection). So as to achieve accelerations for the equipment of a lower amplitude than the ground accelerations, the suspension should confer very low resonance (suspension) frequencies, (in the order of 1 Hz) to the suspended equipment, and should accept displacements greater than 40 cm.
amplification
Such characteristics cannot of course be obtained with classical plugs and the resulting displacements are not free from inconveniencies for supported equipment (respecting positioning, external connections, dielectric distances). This method is not therefore widely used. On switchgear in cabinets When for example a cabinet houses diverse equipment, it is necessary to do carry out both a mechanical transfer study and an evaluation of the robustness of the equipment in question, with a view to establishing the compatibility between the aptitudes of the cabinet and the limits of the switchgears functional withstand capability to the vibratory environment. To limit the amplifications of the cabinets movements, and as a result the transfer, it is necessary to have rigid cabinet frames (reinforced or braced). The required degree of rigidity is in accordance with the robustness of the switchgear. The usual recommendations include: c cabinet construction: assemblies which are bolted or welded together are preferable to those which are riveted together which can work loose and generate impacts which are harmful to the switchgear,
acceleration of the structure = ground acceleration
ZPA level
FAPN strong part (dynamic modal behaviour) zero period acceleration part (static behaviour)
Fr
fig. 21: resonance frequencies in the strong part of the response spectrum are to be avoided.
c mounting of the cabinet: the ideal solution consists in bolting the cabinet to the ground and the wall, with fixings which must be dimensioned so as to resist the loads resulting from seismic acceleration, c disposition of the cabinet: if the cabinet is only mounted to the ground, it is better for the heavy masses to be arranged at the bottom of the cabinet; the same applies to fragile devices, c mounting of devices: rigid fixings are preferable; otherwise, it is wise to be wary of local resonance modes, and to the different movements during the seism, c circuit boards: avoid boards which are too large or too full, heavy components; provide stiffeners if necessary, c cabling: so as to avoid inertia loads, flange the cable layer as close as possible to the connectors.
The detail of the meshing is not essential in order to access fundamental modes, but care must be taken to respect the distribution of the principle elements of stiffness and mass. However, when the stresses are being calculated the meshing must be sufficient. The computing program then determines, on a seismic excitation range (0-40 Hz), the resonance
frequencies and modal shapes associated to the model, as well as the modal participation factors: this is modal analysis. The table in figure 23 gives the elements of modal analysis for a transformer; these reveal the LV insulator to be a sensitive element, but that the conservator is even more so, because its second resonance mode has a frequency of 11 Hz (which is likely to
simulation by analysis at design stage

Numerical analysis for a structures dynamic behaviour generally uses the finite elements method. This numerical technique allows the mechanical behaviour of a structure, subjected to dynamic sollicitations, similar to those generated by an earthquake, to be predicted. This technique is particularly well suited to the design stage, when the structure only exists in the form of definition or utilisation plan, and can still be modified. It provides essential data for the designers with regard to stresses, anchorage loads and the deformations produced by the seismic excitation. Principle The principle behind this method is to construct a simplified model of the equipment, by using a certain number of finite elements ( beams, plates, volumes) which represent the structure, as well as concentrated masses representing the functional devices. Meshing uses the structures geometric data (profile, thickness, profile intertia) and the equipments physical characteristics (Youngs modulus, density) (see fig. 22 for an example).
Y X
fig. 22: meshing of finite elements of a EHV transformer (2,000 elements and 1,500 nodes).
appear in the strong part of the seismic excitation response spectrum). The next stage consists of simulating the equipments response under seismic excitation and thus obtaining the displacements, stresses and loads on the supports. Two methods can be used according to the position of the structures resonance frequencies with regard to the strong part of the response spectrum: c pseudostatic analysis (The resonance frequencies appear above the strong part of the response spectrum). If this is the case, then the analysis is static, and the model is subjected to maximum ground acceleration for every direction. c superposition method analysis analysis (Certain resonance frequencies are in the strong part of the response spectrum). In this case the models response for a given direction is obtained by multiplying the modal shape for each mode (Di) by its modal participation factor and by the acceleration read at this modes frequency on the response spectrum. The responses of the different modes are then combined in order to obtain a global response for a given direction. It must be recalled that the models response spectrum does, by definition, include damping: it is therefore necessary to use a response spectrum which corresponds to the damping of the structure studied, even a less favourable one, that is to say the weaker one. The last stage involves ensuring that the maximum values of displacements, stresses and loads on connections are acceptable, that is to say compatible with the structures mechanical characteristics. There are combination rules for seismic and static stresses (own weight, snow, wind, pressure, traction loads). Moreover, the spatial distribution of accelerations will make it possible to check the functional withstand
capability of the switchgear and the connected instruments, either by using specific tests or by making comparisons with tests which have already been carried out. Limitations of the finite elements method When test models with finite elements are used, divergence is obviously going to occur between the calculated and actual resonance modes of the installed equipment. This is mainly due to the approximations made as far as the limit conditions (how structure is mounted to the ground), internal connections, non linearities, as well as the different simplifications inherent in test models are concerned. Generally speaking only the primary vibrational mode calculation which is acceptable.
However, given the broad band nature of seismic excitation, the evaluation of stresses calculated by response spectrum tolerates an error as to the resonance frequencies exact position. This explains why it still makes sense to use this method at the design stage. Modal tuning As soon as the equipment prototype is ready, it is possible to correct the finite elements model, by carrying out a tuning according to the measurement data. Different experimental techniques enable the strucures actual dynamic characteristics to be obtained, by using the tuning software, and to make the necessary modifications to the finite elements model, so that it might provide a closer representation of reality.
mode No
frequency (Hz) 8.7 11 12.7 13.2 13.8 15.9 17.2 19 19.3 21.2 22.9 23.1 24.1 24.4 24.6 24.8 24.9 25.5 26 26.5 26.6 29.3 30.3 30.5 31.6
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25
modal participation factor 4 232 14 34 5 24 11 105 51 24 18 7 4 47 42 3 0,2 33 96 6 25 115 354 11 2
element concerned
conservator conservator radiator all auxiliary parts heat exchanger conservator heat exchanger all auxiliary parts all auxiliary parts heat exchanger conservator heat exchanger surge arrester surge arrester surge arrester surge arrester radiator surge arrester radiator heat exchanger radiator LV insulator LV insulator LV insulator LV insulator
fig. 23 : rsultat du calcul modal pour les divers lments du transformateur de la fig. 22.
4. qualification by simulation or test
introduction
To qualify means proving the equipments withstand capability under identified or normalized stresses. There are two ways of realizing seismic qualification: c the first involves effectuating real size tests on the equipment; c the second uses finite element test models which can be combined with a certain amount of experimental data. The latter is becoming more and more important in the qualification process, particularly as far as mechanical withstand capability is concerned. But today it is still tricky to take account of the functional aspect by using a test model. Qualification by test is used: c for equipment with dimensions which lend themselves to the vibration testing machine, c for specific equipment (unitary, small series), c if the functional aspect is determining (complex or high level of safety). Qualification by numerical analysis is used if: c the dimensions of the equipment are incompatible with the testing machines (as is the case for large transformers), c a device has already been tested under other seismic conditions, c the device is a modified version of a qualified device, c the functioning of the equipment is not requisite during the earthquake. Combined qualification by numerical analysis and experimental modal tuning is used: c for large series equipment, c when standards or operators permit this kind of justification (knowledge of functional data). In fact, numerical analysis often precede real size tests. This maximizes chances of correctly
effectuating the qualification tests successfully first time round. We will now: c illustrate by means of two examples: combined qualification and qualification by real size tests preceded by a design test model, c develop the methodology of qualification by test.
combined qualification (numerical analysis and experimental tuning)

The method which combines both analysis and tests involves: c creating a mathematical model, c gathering in the data from partial tests (modal experimental analysis) carried out on the prototype, concerning the devices dynamic behaviour (damping, resonance frequencies, modal shapes), c tuning the mathematical model with the preceding data. The analysis model then allows the mechanical withstand capability to be evaluated under accumulated seismic sollicitations and service stresses. The functional withstand capability involves checking that the equipment is not badly affected by the deformations and accelerations delivered by the analysis. The following example shows the method which combines both tests and calculation used to define seismic resistance of HV circuit-breaker. Seismic qualification of a HV circuitbreaker (see fig. 24) c first stage: numerical analysis of circuit-breaker. The model is made out of finite elements: (beam-gantries, plates and shells for the insulators), the model comprises 2,670 elements and 3,200 nodes,
c second stage: experimental modal analysis (see fig. 25 and 26). This analysis is carried out on the prototype. It consists in acquiring the transfer functions between a point of excitation (generated force) and the response points (measured accelerations), then identifying the actual modes of the structure (resonance frequencies and associated modal shapes), c third stage: tuning of finite element model. Here the parameters of the finite element model are readjusted (fineness of meshing, physical parameters: YOUNGs modulus, density, limit conditions) so that the models dynamic characteristics can be made to reflect reality as closely as possible, c fourth stage: measuring the damping coefficient.
fig. 24: Merlin Gerin circuit-breaker designed for use in HV switchgear equipment.
X Y
X Z
fig. 25: modal experimental analysis.
fig. 26: resulting experimental modal shape (f = 3.8 Hz).
In order to calculate the circuitbreakers response using the response spectrum method, it is necessary to know which damping to apply to the model. This is obtained by subjecting the prototype to a substantial mechanical deformation (free oscillation test) ; damping is then deduced by observing the decline in oscillations, c fifth stage: analysis of loads, stresses, and displacements under seismic charge. The analysis of the response is carried out by using the response spectrum method, it makes it possible to take different earthquakes into account, c sixth stage: verification of the devices integrity and functioning under seismic loads. This verification consists: v on the one hand, in verifying the mechanical withstand capability of the structure in terms of loads on connections and stresses in the
materials when the circuit-breaker is subjected to accumulated seismic stresses and service stresses (weight, internal pressure, static loads on the terminals, wind) (see fig. 27 page 22, stresses in the HV circuit-breaker), v and on the other hand, in assuring that the deformations which result from the seismic sollicitation do not interfere with the functioning of the device; this final verification is carried out statically by imposing the deformation obtained from the analysis on the equipment, and by executing the different operations for which it is intended.
qualification by real size tests preceded by numerical analysis

Even if the material is to undergo qualification tests, it is still of interest, in order to save time and money, to precede building and testing by a test model effectuated from plans.
Qualification of control/monitoring cabinets Control/monitoring cabinets intended for nuclear power stations are considered in the following example. This equipment is subject to strict functioning safety regulations, and, to this end, undergo real size tests for resistance to seismic sollicitations (see fig. 28, p. 23). In order to be able to present an equipment with higher guarantees of good resistance for testing, a certain number of simulations and investigations are carried out at the design stage. The procedure is as follows: c robustness evaluation of main devices installed in the cabinet. For equipment which does not have any historical data, a maximum withstand capability test is carried out on the seismic excitation frequency range. This involves establishing the equipments fragility threshold (if necessary live). This data is then useful for defining the desired limitation of the cabinet transfer,
c evaluation of transfer for the cabinets structure. The response spectrum form, as well as the fragility threshold of the embarked equipments, indicate the desirable characteristics of the cabinet transfer. Making a numerical analysis for the cabinet aims to identify the main resonance modes in the seismic excitation frequency range (0-40 Hz). The cabinet is modelled in beam-gantry and plate elements and the equipment is represented by punctual masses and inertia. Finite element analysis is carried out in order to estimate the levels of vibration the cabinet will communicate to the equipment situated in the interior. According to the position of the primary resonance frequencies, modifications are made (on a plan) to reduce the amplification of ground acceleration. c experimental verification of the cabinets characteristics (figure 29). An experimental measurement of the primary resonance frequencies is carried out on the equipped, wired cabinet in order to ensure that the structures actual characteristics do not differ from those provided by the analysis, c impact of cabinets actual characteristics on the equipment. The impact, on the equipments resistance during seismic tests, of the discrepancy between the cabinets calculated dynamic characteristics and those actually measured, is evaluated by referring to the seismic excitation response spectrum. Modifications are made if the discrepancies engender vibrational amplifications which are incompatible with the equipments characteristics. For example, the mounting devices for fixing the equipment to the ground and/ or fixing other devices will be reinforced.
these tests. In addition to the laboratory performance, the following expenditure has to be taken into account: c cost of transport, c cost of mounting the test equipment, c cost of replacing material if it turns out to be non operational after the test. The procedure for carrying out qualification tests on an equipment usually features in the specifications
sheet (or in a test program), and conforms with the current standards or recommendations (IEC 68-3-3/UTE C 20420, ANSI, ENDESA, IEEE, etc.). Several variations of the qualification procedure are possible, and they run according to: c the information on the geoseismic context of the equipments location, c the equipments complexity,
qualification by test
Qualification by test is not always plausible (equipment is too heavy or the dimensions are too big) and it is often tricky to set up. Powerful test equipment is required (vibrating tables with large displacements powered by hydraulic jacks, sophisticated control systems), and only some specialised laboratories are capable of carrying out
15 25 40 45 50 55
fig. 27: analysis of stresses.
fig. 28: LV cabinet for a nuclear power station during qualification tests.
c the data on the dynamic behaviour, c the representativeness of the equipment in tests with regard to the series, c the degree of safety the equipment has to ensure during an earthquake. Without going into too much detail, we now propose an insight into the criteria which influence the choice of modalities of a qualification by test, according to the IEC standard 68-3-3. Configuration of material being tested Before proceding with seismic qualification test a certain number of modalities have to be fixed. These are notified in a particular specification which mentions, amongst other things, the arrangements to be taken into account concerning: c choice of test specimen. Preliminary analyses are sometimes necessary in order to ensure that the chosen specimen does actually represent the worst case scenario, c fixings and mounting. They have to be identical to those used on site, c service conditions which have to be taken into account (mechanical or electrical),
Z X Y
fig. 29: modal deformation of the cabinet's frame derived from the experimentation.
c functioning observed during the test, as well as the methods to be used in order to check that the desired performance is maintained (measurement of the insulation or the electrical continuity, breaking capacity, animation software, etc.). Acceptance and/or bad functioning criteria are classed according to three severities: c severity 0: equipment shall not show disfunctioning during and after the seismic test, c severity 1: equipment can function badly during the seismic test but shall remain operational after the test, c severity 2: equipment can function badly during the seismic test and requires intervention or an adjustment in order to regain its normal state, without replacement or repair being necessary. Choice of test severity level Two classes have been provided according to the availability and/or the exactitude of the seismic characteristics definition of the environment and the equipment. c general seismic class: in this case the acceleration to which the equipment will be subjected is normalized (several performance levels are usually proposed by the standard), c specific seismic class: this concerns equipment for which the considered seismic movement is the result of a seismological study (MHEL). It also takes the geographical implantation characteristics into account, in addition to those of the building or support structures. This is generally the case
for electrotechnical equipment, particularly for nuclear power stations, where the acceleration to which the equipment is subjected will be specified in the response spectrum. Different types of tests Different methods are used in order to recreate the seismic environment on a vibrating table. They differ in the number of directions in which are excited simultaneously (single axis, multiaxis), and in the way in which the seismic waves are recreated by test methods. c single axis, multi-axis test. The seismic ground movement is produced simultaneously in all directions and, in order to simulate this exactly, triaxial tables are required, whereas normal testing permits at best biaxial tests (two excited simultaneously axes). Where no accuracy is available concerning the equipments seismic behaviour, the biaxial test is recommended. It requires an installation which will permit simultaneous excitation with independant waves according to two of the equipments preferred axes; the test response spectrum must be at least equivalent to the required response spectrum. In order to effectuate testing according along the 3rd axis, it is sufficient to turn the equipment through 90 around a vertical axis and effectuate the same test again. Single axis testing, executed according to the three preferred axes, is justified in the following circumstances: v if the equipment is only subjected to an excitation which could be
considered as monoaxial, on account of its on site mounting conditions, v if there is no coupling (or weak coupling) between the equipments three preferred test axes, taken two by two. c monofrequency and multifrequency waves. In general the test wave used must: v produce a Test Response Spectrum (TRS) which is superior or equivalent to the Required Response Spectrum (RRS) notified in the particular specification, v have a peak maximum acceleration amplitude value equivalent or superior to that of the RRS ground acceleration (Zero Period Acceleration (ZPA)), v reproduce the effects of a specified earthquake with a safety margin, v not include frequencies higher than 35 Hz, v be of a duration which is at least equivalent to that of the strong part of the earthquake; in general this is between 5 and 10 s. The required response spectrum is generally broad band, and the use of multifrequency waves is recommended. Modern testing machines enable this kind of wave to be generated, or to be more precise, to pilot the test machine, either from a time-history provided directly by the test specification, or by synthesizing a time-history from a required response spectrum (RRS), for a duration of 20 seconds. In certain transfer cases such as those found in buildings, the response spectrum is narrow band; testing can as a result be effectuated in mono frequency waves.
5. conclusion
The high degree of continuity of service required for electrical distribution and control/monitoring demands that all types of environmental restrictions (mechanical, climatic, electromagnetic) are taken into account. Seismic stresses,which are particularly destructive are included here and they must be specified at the design stage of the equipment. In order to do this it is necessary to know the severity of the maximum historical earthquake likelihood in the form of a dimensioning spectrum, or, for mass-produced equipment, to
choose the level of severity of the normalised seismic class. Today, an equipments mechanical withstand capability can be understood with a high degree of precision thanks to numerical analysis and finite element calcuation of the stresses Proving that the equipment remained operational during or after the earthquake, is more difficult and generally requires the numerical analysis to be combined with tests on the operational elements. The examples of qualification by numerical analysis and/or tests presented in
chapters 3 and 4, reveal the knowhow of a company which, for many years has been providing countries exposed to earthquakes with equipment for nuclear power stations and other electrical equipment. As for quality or electromagnetic compatibility, seismic withstand must be mastered at the design stage; if this is neglected then it is often difficult and more expensive to correct problems at a later stage. As a result, numerical analysis and powerful calculation methods are used widely in the anti-seismic design of electrical and electronic equipment.
6. bibliography
Standards c IEC 68-3-3 (UTE C20 420): Guidance. Seismic test methods for equipment. c IEC 1166: HV circuit-breakers: guide to seismic qualification. Merlin Gerlin Cahiers Techniques c CT 85 (1977): Seismic withstand capabilities of electrical equipment by P. PY, J.-Y. BERTHONNIER (presents technological solutions for HV circuit-breakers).
Miscellaneous publications c AFPS90 recommendations for drafting rules relative to works and installations to be realised in regions subject to seismic volumes 1 and 2. c Dynamic calculations of structures in a seismic zone by Alain CAPRA and Victor DAVIDOVICI. c Revue des laboratoires dessais, ASTE publication, n 9, 31, 36, 39.
n 182
LV circuit-breakers confronted with harmonic, transient and cyclic currents
Michel Collombet A graduate engineer from the INPG in 1968, he joined Groupe Schneider in 1975 as manager of digital electronic developments in SES (Electronic Safety Systems). He is currently in charge of electronic developments in the Low Voltage Power Components Business Unit (DBTP). Bernard Lacroix A graduate engineer from ESPCI in 1974, he joined Groupe Schneider in 1981 as a sales engineer in the UPS division and later as sales manager in the people protection products division. In 1991 he was appointed as head of specfication in the LV Power distribution business.
E/CT 182, first issued april 1997
glossary
ASIC IIR filter GFP Application Specific Integrated Circuit. Infinite Impulse Response. Ground Fault Protection. instantaneous magnetic protection tripping threshold. ultimate (maximum) breaking capacity of a circuit-breaker. magnetic protection setting or Short Time (IST). thermal protection setting or Long Time (ILT). time delay setting of a magnetic trip unit or Short time protection (tST). setting (if required) of the thermal protection delay or of the time delay of the Long Time protection (tLT).
I Icu Im Ir
tm tr
Other current quantities are defined in the IEC 364 installation standard as in the figure below.
t
tr (tLT)
IEC 947-2
tm (tST)
Ir (ILT)
Im (IST)
I
(Iinst)
Icu
In IZ IB
for a circuit-breaker
rated current of protection acceptable current of line circuit operational current NF C 15-100 433.2
IB i IZ i In = Ir
LV circuit-breakers confronted with harmonic, transient and cyclic currents
contents
1. Review of the LV circuit-breaker Role of a circuit-breaker Technology / organisation of an LV circuit-breaker Current measurement Theoretical review of harmonic currents Harmonic current generators Management of harmonic currents by LV circuit-breaker electronic control units Inrush current examples Management of transient currents by LV circuit-breaker electronic control units Cyclic current loads Management of cyclic currents by LV circuit-breaker electronic control units Electronic circuit-breakers settings The advantages of digital technology for dependability Communication by BUS LV circuit-breaker standards Electronic circuit-breakers: new possibilities available p. 4 p. 4 p. 5 p. 8 p. 8
2. Harmonic currents
3. Transient and cyclic currents
p. 10 p. 11
Development of loads, a result of the technological breakthroughs of the last decade, has led to an increased number of constraints in electrical power distribution. Protection devices have had to adapt accordingly, particularly with regard to three phenomena: c high harmonic currents due to multiplication of non-linear loads using power electronics (rectifiers, switch mode power supplies,...); c transient currents caused by energising loads with a high inrush current such as capacitive loads, LV/LV transformers; c cyclic currents resulting from a marked increase in load automatically in repetitive cycles (welding robots, wave train heating). The purpose of this Cahier Technique is to show how electronic control units take these new requirements into account and tend to replace thermalmagnetic trip units. It also shows how the possibilities of digital technology have turned these control units into intelligent, communicating sensors/ actuators.
p. 11 p. 12
p. 13 p. 14 p. 14 p. 15 p. 15 p. 15 p. 16 p. 16
4. Elecronic circuit-breakers: a wide range of possibilities
1. review of the LV circuit-breaker
role of a circuit-breaker
The main role of a circuit-breaker is to protect the electrical installation, and the conductors placed downstream, against abnormal operating conditions such as overloads and short-circuits. In order to perform this function effectively, the circuit-breakers trip unit must take load development into account. This development is characterised by: c increased harmonic pollution Development of power electronics and thus of non-linear loads (data processing machines, rectifiers, dimmers, choppers) and the progress made in load technology (discharge lamps, fluorescent lamps,...) have increased the strength of harmonic currents in distribution power networks. c more frequent transient currents due to standard and new loads generating high inrush currents: v capacitors for compensation of the cos (whose reference value has risen), LV/LV transformers, v but rectifiers with front end capacitors are also becoming increasingly common (lamps with electronic starter, computers...). c loads controlled in cycles. Ever increasing automation results in greater repetition of operations of loads such as process motors, production robots, heat regulation by wave trains... This development is accompanied by a demand for increased continuity of service. Consequently: c to avoid undervoltage and improve continuity of service, replacement sources such as generator sets need to be installed. These sets have specific features that the protection device has to incorporate, for example a higher source impedance which increases disturbances due to harmonic currents and reduces the value of the fault currents, thereby modifying the setting value of the protection devices. c to avoid untimely tripping to satisfy safety and comfort requirements in the
service sector, and in view of the cost of power failure in industry, it is vital to trip only when the risk is real.
technology / organisation of an LV circuit-breaker

Circuit-breakers from 1 to 6300 A are extensively used in LV installations. The trip units for these circuit-breakers are produced using two technologies: c thermal-magnetic trip units. Mainly used for domestic and industrial ranges with small ratings. On modular type devices the trip unit is built into the circuit-breaker. c electronic control units. Formerly reserved solely for high current rating circuit-breakers, there is a marked trend (see fig. 1) towards this type of trip unit as it is a solution offering a high degree of flexibility and is becoming increasingly affordable. In point of fact, the use of digital technologies and in particular use of simple Application Specific Integrated Circuits (ASIC) enables: c universal trip units to be produced providing greater setting facilities, c more data to be processed, c the communication required for installation control and monitoring to be performed.
Thermal-magnetic trip unit These trip units contain bimetal strips and an electromagnet coil, normally mounted in series with the circuit requiring protection. The higher the overload, the quicker the reaction of the bimetal strip. The coil reacts almost instantaneously to high overcurrents according to the electromagnet principle. Control unit These trip units contain sensors, processing and control electronics and an actuator (see fig. 2).
circuit-breaker rating
100
1000
10 000 80 85 90 95 years
fig. 1: development of control units.
actuator
sensors (current transformers)
power supply
processing control
communication
L1
L2
L3
fig. 2: functional modules of a control unit.
c sensors. Current sensors first process the current image to be measured and then, to ensure optimum dependability, supply the trip unit with electrical power (self powered trip units). As these sensors have to perform this dual function (trip unit measuring and power supply), they use a magnetic circuit ( iron CT). c data processing. Processing contains the following functional features (see fig. 3): v function 1 : digitise the signal supplied by the sensor using an analog/ digital converter to monitor current evolution in real time, v function 2 : compensate for CT saturation, if any. In point of fact magnetic core CTs can be saturated for high current values, a phenomenon which is amplified in the event of a temperature rise (see fig. 4), v function 3 : calculate the RMS value of the primary current, v function 4 : compare the RMS value with the thresholds previously set by the user. According to the value, this comparison is made with or without a time delay. If overshooting occurs the electronics send an electrical order to an actuator which will then convert it into a mechanical action to unlatch the circuit-breaker. c actuator. The actuator is confronted with the problem of having to produce a major force instantaneously... without, however, consuming too much electrical current. Its action is equivalent to making an effort of several Newtons over a few millimetres, in other words producing several joules for a few
milliseconds, i.e. several hundred Watts! Actuator efficiency must therefore be outstanding, a fact which rules out the use of electromagnets (coils) and calls for systems with potential energy. This ensures that ultra-rapid unlatching of the circuit-breaker is possible in all circumstances.
current measurement
Thermal-magnetic trip unit In this type of trip unit the bimetal strip does not supply current values but thermally and mechanically reacts to its effects. c thermal. Heating of a bimetal strip uses the same principle as for conductors, i.e. the temperature rise is proportional to the energy supply (J = RI2t) generated
by the flow of a constant current (I) for a given time (t). Tripping thus occurs for an energy J0, thus defining a relationship t = f(I). The standard (IEC 947-2) characterises this relationship by a specific point (I,t) corresponding to a 30 % overload for a period of 2 hours. This parameter determines the bimetal strip and thus the rating of the thermal trip unit. In practice, a bimetal strip is: v directly heated for small circuitbreakers: the current flowing through the circuit-breaker pole crosses the bimetal strip in its entirety. In this case the bimetal strip time constant is low and can be compared with that of the protected cables, v indirectly heated for large circuitbreakers: a wire through which the current to be monitored flows, is placed in the immediate vicinity of the strip.
a) sampling and digitising the signal
b) saturation correction
primary I
primary I
secondary I
secondary I
corrected secondary I
fig. 4: sampling and correction of CT saturation (if any).
primary I
2 correction device secondary I (digital) corrected secondary I
4 comparison with thresholds
A/D secondary I (analog)
integration
actuator tripping order
IRMS
fig. 3: functional diagram.
Cahier Technique Merlin Gerin n182 / p.5
The distance (resulting in a thermal impedance) introduces a delay in heating the strip thus increasing its time constant. However, this time constant is normally far lower than that of the protected cables. In both cases, the thermal time constant of the bimetal strip cannot be adjusted. c magnetic If a short-circuit occurs, the fault current flowing through the coil creates a magnetic field sufficient to cause the displacement of a moving blade which then releases the spring latching mechanism and opens the contacts with a sudden movement. The threshold is set at the air gap. Once the set threshold is exceeded, operating time is more or less constant (a few milliseconds to 50 milliseconds) (see fig. 5). Thermal-magnetic trip units do not measure the value of the monitored current. Electronic trip unit c thermal protection (Long Time protection). The electronic trip unit uses the heating and cooling model of a conductor. In actual, fact it models the temperature of the conductor by calculating its heating in real time according to its thermal equation. A good approximation considers the heating and cooling of a cable between t and t + dt takes place in accordance with the following physical principles: v heating: results from supply of calories mainly by joule effect: A i2dt. A = constant function of resistance, mass, specific heat of the conductor, v cooling: results from losses due to conduction, convection and radiation. These losses are practically proportional to the difference in temperature between the conductor and ambient temperature, i.e. to heating i.e. in all - dt, where depends on the physical and geometrical characteristics of cable installation. The thermal equation of the cable is thus: d = A i2 dt - dt or
d + dt
= Ai2 (1)
where = 1/ the thermal time constant of the conductor. If we proceed by digital sampling at the frequency f such that dt = 1/ f, a digital equation equivalent to (1) is obtained:
t + dt = [1 ] t + i2 where d = t + dt t , = 1/f and = A dt = A/f 2 i.e. k + 1 = [1 ] k + I k (2)

if the measurement at instant t represents the kth measurement and at instant t + dt the k + 1th. Resolution of this digital equation then allows exact modelling of conductor heating. In fact equation (2) representing the temperature calculated by calibration is none other than the digital transfer function of a first order low pass filter 2 (see fig. 6) to which IK is applied as an input signal, ie:
Sk +1 = Sk + = 1
2 I k Sk
section and heat inertia of the cables it has to protect, - real time processing of the equation means this calculation is not dependent on power network frequency. This electronic control, used to monitor evolution of conductor heating/cooling, is also known as the thermal memory of the control unit. c making the thermal protection: Long Time (LT) and thermal memory. The thermal behaviour of a cable defined by equation (2) also corresponds to the long delay function of the electronic circuit-breaker. It protects cables and loads against overloads. The temperature value or its current equivalent, Irms, calculated by digital filtering, is compared with the setting value of the long delay threshold, linked to the acceptable limit m (see fig. 7), thus performing the digital equivalent of the bimetal strip function.
2 ] Sk + I k
t thermal cable limit curve magnetic
Sk+1 represents the RMS value, Irms, of the current after k + 1 samples. Consequently the installation of this filter simply gives the cable temperature by its equivalent in RMS value of the current. v advantages of digital technology: 2 - greatly simplifies calculation of IK , - calculates the RMS value, Irms, of the current, and thus the heating, over a period of time compatible with conductor time constants (several minutes to several hours as the heat inertia () of conductors vary according to their size). This feature is built into the IIR (Infinite Impulse Response) filter of the calculation algorithm; the time constant is defined according to circuit-breaker sizing as this depends on the cross-
Ir
Irm
Icu
Ir = thermal setting Irm = magnetic setting Icu = ultimate breaking capacity

fig. 5: thermal-magnetic circuit-breaker tripping curve.
+
2 Ik
+ +
Sk z
-1
Sk + 1
fig. 6: filtering algorithm (R) of the low pass digital filter.
Permanent knowledge of the cable temperature equivalent ensures not only that m is not overshot but also offers new possibilities for load and cable protection such as: v overload information and protected feeder management. Overload information is available when the setting current or certain preset thresholds have been overshot, thus enabling feeder load monitoring. The operator can thus anticipate tripping due to overload by shedding a nonpriority feeder. This type of information starts opening out towards distribution automation systems, v storage of heating after power shutdown. If an overload causes the circuitbreaker to trip or if the electrical power supply fails upstream (e.g. transfer to a replacement source), cooling continues to be monitored. However management
is now different since the electronics is no longer supplied. Immediately after breaking, conductor temperature is modelled by the discharge voltage of a circuit with a very high time constant (similar to the conductor thermal constant). When power is restored either by switching the circuitbreaker back on or by energising the load, the trip unit electronics will recover the residual voltage value. This value will then be used as a new initial temperature for conductors in the filtering algorithm modelling conductor heating. c short-circuit protection (Short Time - ST). The short time function protects the power network against high overcurrents (in distribution normally around 10 In, adjustable by the user). Just as for long time protection, short time protection is achieved by filtering,
however in this case the RMS value of the current is processed over a period of time (a few ms) compatible with the speed of intervention required for this function. An intentional delay that the user can adjust is also incorporated in the data processing function (see fig. 8). c instantaneous protection. This provides protection against full short-circuits. The unfiltered peak value is the processed value and has no time delay. The long time, short time and instantaneous protection make up the tripping curve of a circuit-breaker with an electronic control unit (see fig. 9). We shall now see how disturbed currents and special applications are treated by a circuit-breaker with an electronic control unit.
(time constant)
calculated I2rms sampled I2 1st order low pass filter (corresponding to calculated )
t LT
D long time cable limit curve
tripping order (if any)
I
fig. 7: making the Long Time (LT) function.
2threshold
(corresponding to max
of cable)
short time ST
TR
IR +
saturation
IIR T = LTTC
TM
Ir ILT
normal overzone load zone
Irm IST
Iinst
Icu
IR Ie (K Te)
X2
IIR t x 20 ms
+ +
short-circuit zone
IIR TM time delay
TRIP
IM
IIR TM time delay
fig. 9: electronic circuit-breaker tripping curves.
IIR t = STTC
TM
IM
I2ton
fig. 8: structure of the long and short delay filter assembly.
2. harmonic currents
Development of loads and in particular the increasing use of static converters, means that the current encountered on distribution power networks are no longer perfect 50 Hz sinusoidal ones... far from it! These currents have an adverse influence on measurement, especially in normal operation for monitoring of thermal effects: for fault currents in excess of 10 In their effect is virtually zero.
Y1 = RMS value of the 50 Hz sinusoidal component (or fundamental), = fundamental frequency, 1 = fundamental phase shift Yn (for n > 1) = RMS value of the order n harmonic component, n = pulsation of harmonic n, n = phase shift of harmonic n. Ohms law In the presence of non-linear loads, Ohms law is applied only between harmonic current and voltage of the same order n with an impedance value calculated for a pulsation equal to n times that of the fundamental: Un = Z(n) x In. Deformation of the current waveform can be said to cause an equivalent deformation of voltage for each harmonic order, whose amplitude and phase depend on the value of the impedance for each harmonic frequency. There is no longer a simple relationship between the RMS values of both waves considered globally. Distorted RMS current Irms = I1 + I2 + I3 + In + ... which can also be expressed by the different current total harmonic distortions hni: In/ I1:
2 2 2 2
theoretical review of harmonic currents

Non linear and harmonic loads Ohms law formulates a proportionality (linearity) between sinusoidal current and voltage at mains frequency. Some loads (said to be non-linear) deform the current sine wave and thus also the voltage sine wave. This deformation is analysed using Fouriers serial decomposition which reveals harmonic currents which are superimposed on the initial sine wave (the fundamental) causing its deformation. (Fouriers) harmonic decomposition All currents and voltages in electrical power networks can be represented by the superimposition of a DC component, a sinusoidal component at mains frequency and a certain number of sinusoidal (harmonic) components with a multiple frequency of mains frequency. These quantities are formulated by Fouriers development of the current or voltage function y(t):
y(t) = Yo + Y1 +
n= n=2
v if the peak factor is greater than 2 there is a risk of untimely tripping if the protection deduces Irms from Ipeak. v if the peak factor is less than 2 there is a risk of overheating due to failure of the protection devices to trip; c heating. For a given load, with a nominal current In at 50 Hz, the RMS current in the conductors will be higher as it is multiplied by 1 + Di2 resulting in: v additional losses and thus overheating of transformers, cables and generators, v magnetic losses, heating and stray torque in rotating machines; c circulation of high currents in the neutral in the presence of order 3 harmonics and their multiples. Consequently thermal trip units must take the RMS value into account to protect conductors. Readers particularly interested in harmonics are invited to read Cahier Technique n 152.
harmonic current generators

Rectifiers Three-phase Gratz bridge type rectifiers create harmonic distortions. This rectifier type which is relatively widespread in all industrial devices due to its economic nature, is however a source of disturbance for power networks. These rectifiers are used in a wide variety of industrial applications such as variable speed drives, UPS and computer power supplies. They are also common in the service sector as a result of the proliferation of switch mode power supplies for office automation applications and fluorescent lamps with electronic ballast. The shape of the input current greatly depends on the presence of a
Irms = I1 1 + h2i2 + h3i2 + + hni2

= I1 1 + Di2
which reveals the current total harmonic distortion Di. Current peak factor for the fundamental, Peak I1: 2 I1 and the peak factor is 2 . For the total distorted current, peak I = K Irms = K . I1 1 + Di2 Harmonic effects c effect of the peak factor:
2 sin (t 2 sin (nt
1) n)
Yn
Yo = DC component amplitude,
smoothing reactor which most rectifiers do not have. c example 1: (see fig. 10) Input current of a three-phase rectifier (unmonitored Gratz bridge). Harmonic composition corresponding to the current in figure 10 is (as a percentage of fundamental amplitude and with phase shift with respect to the latter): = 100 % h1 (50 Hz) h5 (250 Hz) = 33 % (180) h7 (350 Hz) = 2.7 % h11 (550 Hz) = 7.3 % (180) h13 (650 Hz) = 1.6 % h17 (850 Hz) = 2.6 % (180 %) total harmonic distortion:
D = h3
2
% 150
100
50 20 ms 0
- 50
+ h5
+ h7
+ h17
= 6 %
- 100
Irms = 106 % of Ih1. Imax/ V2 = 78 %.

This means that a control unit based on an RMS measurement using peak current would measure an RMS value of 78 instead of 106. The installation would be underprotected in this case. c example 2: (see fig. 11) Input current of the three-phase rectifier of a variable speed drive for asynchronous motor. Harmonic composition corresponding to the current in figure 11 is (as a percentage of fundamental amplitude and with phase shift with respect to the latter): = 100 % h1 (50 Hz) h5 (250 Hz) = 85 % (180 %) h7 (250 Hz) = 72 % h11 (550 Hz) = 41 % (180 %) h13 (650 Hz) = 27 % 8 % (180 %) h17 (750 Hz) = total harmonic distortion in this case:
- 150
fig. 10: example 1: a rectifier.
% 300
200
100
20ms 0
D =
h3 2 + h5 2 + h7 2 + h15 2
- 100
= 58 %
Imax/ V2 = 203 %.
This means that a control unit based on an RMS measurement using the peak current would measure an RMS value of 203 instead of 158. The installation would be overprotected in this case.
- 200
- 300
fig. 11: example 2: a variable speed drive with variable frequency.
Standard fluorescent lighting The presence of harmonic currents is observed in steady state (see fig. 12). Harmonic composition for the phase current gives: = 100 % h1 (50 Hz) 35 % h3 (150 Hz) = h5 (250 Hz) = 27 % (180) h7 (350 Hz) = 16.1 % (180) h9 (450 Hz) = 2.2 % (180) h11 (550 Hz) = 3.4 % h13 (650 Hz) = 1.1 % D = 42.6 % Irms = 199 % of Ih1 = 39 A. Note that as these are well-distributed single-phase loads, the RMS current in the neutral is 33 A as a result of the third harmonic and its multiples, whereas it ought to be zero.
of a digital filter has simplified calculation of the RMS value and enables the conductors thermal equation to be modelled (see para. 1). The problem is then to define sampling frequency to obtain an accurate RMS value. In order to calculate the real RMS value of a signal of fundamental frequency f loaded with harmonics up to
order n, Shannons theorem states that this signal must be sampled at frequency 2n f. In practice the sampling chosen for electronic control units is 1600 Hz, thus enabling harmonic currents up to order 16 to be considered. The above examples showed that harmonic currents over order 16 can be completely ignored.
management of harmonic currents by LV circuitbreaker electronic control units

The first electronic control units used an analog technology. This solution consisted of implementing a simple RC filter in the measuring circuit behind a double wave rectifier. A technique which satisfied needs relatively well as long as harmonic phenomena remained marginal. Technological progress and in particular ASIC type solutions integrating a large number of components, made rapid, ultra-fine signal sampling possible. Current use
0 20 ms
fig. 12: fluorescent lighting.
3. transient and cyclic currents
For some feeders the main problem is to clearly distinguish the normal energising current from the fault current. The loads presenting this type of problem are mainly LV/LV transformers, motors, tungsten and fluorescent lamps,... Another problem to solve is how to properly protect the cables feeding cyclic current loads. This chapter will deal with both these problems.
15 In
1st peak
fig 13: transformer energising.
inrush current examples

LV/LV transformers A transformer primary is a choke with a magnetic circuit. On energising a dual phenomenon may occur: c first the creation of the load current (transient state) of an LR circuit (with the characteristics in steady state of the transformer primary); c second, due to the presence of the saturable magnetic circuit, a high current peak, according to the time of energising, due to saturation of the magnetic circuit. The result is an inrush curve of the type shown in figure 13, composed of a series of peaks absorbing one another in accordance with an exponential law. The first current peak frequently reaches 10 to 15 times transformer rated current and even more than 20 times nominal current for the smaller ratings (i 10 kVA). The inrush current is quickly damped with a time constant of a few dozen ms. To give an example: for a 50 kVA LV/LV transformer, the peak is around 15 In and the time constant for the phenomenon 20 ms (see fig. 13).
Motors Asynchronous motors account for 90 % of motors used. This motor type presents a current inrush on startup whose envelope curve has the form of figure 14 (for direct starting). An excitation peak (8 to 12 In) followed by a starting current (from 5 to 8 In) (see fig. 14). Fluorescent lighting Fluorescent lamps also absorb a very high thermal current on energising. Switch mode power supplies This power supply type, for example on computer load input, presents peaks of around 10 In during energising (capacitor load through a rectifier). It should also be pointed out that the inrush current of many loads after a short power cut is greater than the initial energising current: the standard example is the capacitor bank which remains charged. Digital electronics has enabled short time protection to be suitably adapted to distinguish transient currents from short-circuit currents (see fig. 5).
Id (5 to 8 In)
In t 10 to 20 ms td (1 to 10 s)
fig. 14: motor energising.
management of transient currents by LV circuitbreaker electronic control units

Let us now see how an electronic control unit manages transient currents greater than the short time threshold.
If the current exceeds the threshold Im, the control unit uses IIR filtering to calculate over a very short period (a few ms) the RMS value of the current, thus smoothing this overload. This kind of time delay depends on the energy of the transient current. c in the case of a normal transient current supplying considerable power in a very short time, then rapidly decreasing, the tripping threshold is not reached. Consideration of the energy of the peak (and not of its peak value) lets this transient pass even if it lasts several periods, whereas a magnetic trip unit would have tripped (see fig. 15). c if the transient turns out to be a persistent fault (see fig. 16), the short time delay function filter continues to increment very quickly, thus causing rapid tripping once the threshold is overshot. This technique also allows special fault currents to be monitored, for example the current resulting from the stalled rotor of a motor.
0.005
0.01
0.015
0.02
0.025
0.03
0.035
0.04
0.045
0.05
t(s)
I
filter 1 filter 2 ST threshold (10 Ir)
fig 15: transient current solved by smoothing the IIR filter.
cyclic current loads

Intermittent operation of a motor or a load causes rapid heating particularly if the energising currents are high. Supply cables, like loads, undergo the same current stresses but not necessarily the same heating (different thermal time constants). Some loads have specific protection. The overcurrent value that can be withstood by a cable depends on initial heating and on the cooling period elapsing between two consecutive overcurrents. Thus a cable can be sized for the energy it conveys if the cycle is less than its thermal time constant. Using the thermal equation model of a cable, cyclic heating of a cable can be represented by a curve of the type shown in figure 17. This is the case for example of welding machines, static switches with wave trains or motors with cyclic starting.
0.005
0.01
0.015
0.02
0.025
0.03
0.035
0.04
0.045
0.05
t(s)
I
filter 1 filter 2 trip unit ST threshold
fig 16: persitent fault.
management of cyclic currents by LV circuitbreaker electronic control units

Digital control units take cooling status into account and accurately control thermal stress on conductors, as explained at the end of the first chapter. Whatever the characteristics of the cyclic current I/In, period, cyclic ratio, the cable is protected if the rated current of the protection (Ir) is correct. However to derive maximum benefit from cable possibilities without tripping
the circuit-breaker and to limit stresses if a fault occurs, the extensive setting possibilities of the Long and Short Time protection must be used. For example, figure 18 illustrates the extreme cases: c high I/Ir but of short duration; c I/Ir close to 1 but of long duration. In this case, as for the inrush currents, we can observe the superiority of electronic control units as a result of the presence of the short time delay function, but above all due to the enhanced performance of its thermal function compared with bimetal strips.
TmaxLT
cable temperature (in Irms equivalent)
TminST
ILT max av
t T TmaxLT
ImaxST
cable temperature (in Irms equivalent)
Imax Irms
IminST
Iinst
fig. 17: cyclic heating of a conductor with continous load plus cyclic load.
fig. 18: Long and Short Time settings for strong, short or meak, long cyclic loads (steady state).
4. electronic circuit-breakers: a wide range of possibilities
electronic circuit-breaker settings

Confronted with distorted, inconsistant currents, electronic control units simplify the task of installation designers and operators, by their ability to take real RMS currents into account, by managing overcurrents and by the flexibility and range of their settings. Harmonic currents Precise consideration of the effects of harmonic currents means that no special settings (linked to these currents) are required in the control unit. The possibility of knowing the RMS current value in real time by electronic measurement enables fine adjustment of Ir if required. Moreover, digital technology allows this information to be easily transmitted to an ammeter (local or remote mode) or to a bargraph. Transient currents The possibility offered by electronics to distinguish transients from fault currents provides increased cable protection and avoids untimely tripping. Cyclic currents Adjustment of current Ir for precise conductor sizing is completely compatible with an installation presenting normal overloads by using the short and long delay protection settings. Tripping curves One of the advantages of electronic circuit-breakers is that they provide universal protection. With the same control unit, all operator needs can be taken into consideration thanks to the flexibility and wide range of the settings. Electronics allow settings to be made in an extensive range for both time delay and thresholds.
Not only does this solve the problem of inrush and cyclic currents, but it is a definite advantage when implementing time discrimination. It also allows a transformer, cable or generator to be protected using the same device (see figure 19 for an example).
In addition this technology provides greater reliability and immunity (electromagnetic compatibility) than discrete technologies. A host of functions Besides the functions for protecting live conductors against overcurrents and short-circuits, other functions are or can be integrated, for example: c ground fault protection (GFP): this function is often requested for installations in the USA, c load monitoring by calculating Irms/ ILT which informs the user of the load level of the feeder in question, c actual short-circuit current, c number of operations (useful for maintenance management).
the advantages of digital technology for dependability

Digital technology with extensive use of ASICs, makes it possible to perform a host of measuring, protection, control/ monitoring and communication functions.
generator thermal cable thermal limit limit
protection setting generator tLT (mini notch)
protection setting cable tLT (maxi notch) 3 Ir
IR = In
IZ
fig. 19: long time tripping curves of an electronic circuit-breaker protecting a cable or generator.
Reliability Circuit-breakers must have a very high level of reliability. For this reason the integrated electronics are equipped with self-monitoring functions to indicate malfunctioning of the microprocessor or abnormal rise in temperature. It must also satisfy the tests defined in standards IEC 801 and IEC 1000, specifying the EMC withstand of devices and in particular immunity to magnetic fields.
communication by BUS
Digital technology and the position of the circuit-breaker in the electrical installation means that all relevant parameters required for proper power network operation are easily available on the Bus. Integrated digital electronics enable communication to
electrical distribution management and supervision systems. The data transmitted provide information on the circuit-breaker environment: c position of the setting switches, c phase and neutral current values, c overshooting of the load monitoring threshold, c overload alarm, c cause of tripping. This information is collected in data logs to help the system operator and/or manager improve the management of his installation.
and particularly of ElectroMagnetic Compatibility (EMC) has led standards authorities to include recommendations concerning these phenomena in circuit-breaker standards (see fig. 20).
electronic circuit-breakers: new possibilities available

The guarantee supplied by the standard Compliance with standard IEC 947-2 and in particular appendix F, combined with suitable design, ensure the reliability of electronic circuit-breakers. Moreover the tests stipulated by IEC 947-2 guarantee installation designers and users perfect adaptation of the protection function (see Cahier Technique n 150 for more details).
LV circuit-breaker standards
Industrial circuit-breakers meet standard IEC 947-2. The increasing importance of environmental problems
test F.4.1
disturbances non-sinusoidal currents.
tests performed 3 tests with peak factor 2 H3 80 % ; H5 50 % and H3 u 60 % + H5 u 14 % + H7 u 7 %. current reduced by 30 % ; 60 % ; 100 % for 0.5 to 50 periods. circuit-breaker frequency range. 1 Hz steps. 5/50 ns wave (Fr: 2, 5 kHz) level 4 kV, 1.2/50 s wave - 6 kV and 8/20 s - 3 kA. on 8 kV contact discharge. from 26 to 1000 MHz. 10 V/m. amplitude modulation 80 % 1 MHz.
F.4.2 F.4.3 F.5 F.5.2.2.1 F.5.2.2.2 F.6 F.7
sags and breaks. frequency variations. conducted transients and HF disturbances: IEC 1000-4-4 rapid transients. IEC 1000-4-5 shock waves. electrostatic disturbances IEC 1000-4-2. electromagnetic field disturbances IEC 1000-4-3
fig. 20: table showing EMC tests as in appendix F of standard IEC 947-2.
5. conclusion
LV circuit-breakers u 250 A, with electronic control units, are perfectly suited to the various constraints encountered in installations. The current calculation power of the ASICs are responsible for considerable progress: thus: c despite the increase in harmonic currents, long time protection takes the real RMS value into consideration, c the thermal memory, more effective than indirectly heated bimetal strips,
allows improved monitoring of cable temperature evolution particularly for loads with cyclic operation, c the short time protection settings ensure better management of energising currents than magnetic trip units, c the wide range of the various settings allows adaptation with cables of varying cross-sections and with generators. In addition to these protection functions, digital electronics enable the circuit-breaker to transmit measure-
ments, states, etc... by Bus, to have access to remote setting and naturally to be remote controlled. Present-day circuit-breakers have thus become intelligent sensors/actuators which, as part of Electrical Power Management (EPM), play a large role in simplifying operation of power networks and in increasing continuity of service. One regret however... electronics are still too expensive to be used in circuitbreakers below 250 A ratings.
6. bibliography
Standards c IEC 947-2: Low voltage switchgear and controlgear - part 2: circuitbreakers. c NF C 63-120: appareillage basse tension - 2me partie: disjoncteurs. c IEC 364/NF C 15-100: Electrical installation of buildings. c IEC 801: Electromagnetic compatibility for measuring and control equipment in industrial processes. c IEC 1000: Electromagnetic compatibility (EMC). c IEC 50: General index of electrotechnical vocabulary. Merlin Gerin Cahiers Techniques c Les perturbations lectriques en BT, Cahier Technique n 141 R. CALVAS. c Development of LV circuit-breakers to standard IEC 947-2, Cahier Technique n 150 E. BLANC.
c LV circuit-breaker breaking techniques, Cahier Technique n 154 R. MOREL. c LV breaking bu current limitation, Cahier Technique n 163 P. SCHUELLER. c Energy-based discrimination for lowvoltage protective devices, Cahier Technique n 167 R. MOREL, M. SERPINET. Other Merlin Gerin documents c Guide de linstallation lectrique 07/91. c Les filtres IIR et FIR - E. SUPITZ. c La distribution lectrique de qualit D. FRAISSE. c Llectronique dans les disjoncteurs BT - D. FRAISSE. Other external documents c Guide de lingnirie lectrique. c J3E n 619. c Le contact lectrique - M. RIVAL.
Ral. : Illustration Technique - Lyon DTE - 04-97 - 1500 - Printing. : Clerc - Fontaine

Active harmonic conditioners and unity power factor rectifiers
E. Bettega J-N. Fiorina
no. 183
Eric BETTEGA After joining Merlin Gerin in 1983 as a laboratory technician in ABTs electronics engineering and design service, he became part of the Scientific and Technical Division in 1986. In 1991 he obtained his Engineering degree from the CNAM (Conservatoire National des Arts et Mtiers), and is currently a senior Corporate Research engineer.
Jean Nol FIORINA He first joined Merlin Gerin in 1968 as a laboratory technician in the ACS (Static Converter Power Supplies) department where he participated in the development of static converters. In 1977 he obtained his ENSERG (Ecole Nationale Suprieure dElectronique et de Radiolectricit de Grenoble) Engineering degree and rejoined ACS. Starting as development engineer, he was soon afterwards entrusted with projects. He became later responsible for design and development for MGE UPS Systems. He is in some ways the originator of medium and high power inverters.
ECT 183 first issued, June 1999

Electric loads are becoming increasingly non-linear in the industrial, tertiary and even household sectors. These loads absorb non-sinusoidal currents which, under the effect of circuit impedance, distort the purely sinusoidal voltage waveform. This is what is known as harmonic disturbance of power networks, currently a cause for concern as it gives rise to serious problems. We recommend that harmonic non-specialist readers begin by reading the appendix where they will find the basic concepts required to understand the various standard and new solutions to limit or combat harmonics. Not only the characteristic quantities, but also the non linear equipment, influence of the sources and disturbing effects of harmonics need to be known. Last but not least, standards lay down levels of compatibility, i.e. the maximum permissible levels. The purpose of this Cahier Technique is to describe the active harmonic conditioners. This attractive, flexible and self-adaptive solution can be used in a wide variety of cases to complete or replace other solutions. However chapter 1 of this Cahier Technique will review other traditional solutions which should also be taken into consideration.
Contents
1 The traditional solutions 1.1 Reducing harmonic currents of non linear loads 1.2 Lowering harmonic impedance of the source 1.3 Carefully choosing the installation structure 1.4 Harmonic isolation 1.5 Using detuning reactors 1.6 Passive harmonic filters 2 Unity PF rectifiers and active harmonic conditioners 2.1 Introduction 2.2 Unity PF rectifiers 2.3 The shunt type active harmonic conditioner 3.1 The parallel/series hybrid structure 3.2 The series/parallel hybrid structure 3.3 Parallel combination of passive filters and active harmonic conditioner 3.4 The performances of hybrid structures 4 Implementing a shunt type active harmonic conditioner 4.1 Objective and context 4.2 The insertion point of a shunt-type active harmonic conditioner 4.3 Sizing a shunt-type active harmonic conditioner 4.4 Application examples 5 Conclusion Appendix: review of harmonic phenomena Definition and characteristic quantities Origin and transmission Deforming loads Harmful effets of harmonics Standard and recommendations p. 4 p. 4 p. 4 p. 5 p. 5 p. 6 p. 7 p. 8 p. 11 p. 17 p. 18 p. 19 p. 19 p. 22 p. 23 p. 24 p. 25 p. 27 p. 28 p. 29 p. 30 p. 31 p. 32
3 Hybrid active harmonic conditioners
1 The traditional solutions
Electricians need to be familiar with these solutions in order to take the right measures when installing polluting equipment or to take all factors into account when designing new installations. The solutions described hereafter depend on the objective sought and on the non linear/sensitive equipment installed.
They use passive components: reactors, capacitors, transformers and/or carefully choose the installation diagram. In most cases the aim is to reduce voltage total harmonic distortion at a load multi-connection point (in a distribution switchboard).
1.1 Reducing harmonic currents of non linear loads

Besides the obvious solution which consists of choosing non-disturbing equipment, the harmonic currents of some converters can be limited by inserting a smoothing reactor between their connection point and their input. This solution is particularly employed with rectifiers with front end capacitors: the reactor may even be proposed as an option by manufacturers. A word of warning however! Although this solution reduces voltage total harmonic distortion upstream of the reactor, it increases it at the terminals of the non-linear load.
1.2 Lowering harmonic impedance of the source

In concrete terms this consists of connecting the disturbing equipment directly to the most powerful transformer possible, or of choosing a generator with a low harmonic impedance (see appendix and fig. 1 ). Note that it is advantageous on the source side to use several parallel-connected cables of smaller cross-section rather than a single cable. If these conductors are far enough apart, apparent source impedance is divided by the number of parallel-connected cables.
E ZS ZL
THD
Non-linear load
Fig. 1: addition of a downstream reactor or reduction in upstream source impedance reduces voltage THD at the point considered.
1.3 Carefully choosing the installation structure

Sensitive loads should not be parallel-connected with non-linear loads (see fig. 2 ). Very powerful non-linear loads should preferably be supplied by another MV/LV transformer.
a) Solution to avoid
b) Solution to recommended
Non-linear equipment
Non-linear equipment supply
Sensitive equipment
Clean power network
Fig. 2: a Y-shaped distribution enables decoupling by natural and/or additional impedances.
1.4 Harmonic isolation

The aim is to limit circulation of harmonic currents to as small a part as possible of the installation using suitable coupling transformers. Use of Y-connected primary transformers (without neutral!) with zig-zag secondary is an interesting solution as it ensures minimum distortion at the secondary. In this case 3 k order harmonic currents do not flow at the transformer primary, and the impedance Zs depends only on the secondary windings. The inductive part of the impedance is very low: Uccx 1%, and resistance is practically halved compared with a Y transformer of identical power. Figure 3 and the following calculation show why 3 k angular frequencies are not present at the transformer primary (zero sequence current is nil). Current circulating for example in the primary winding 1 equals: The calculation shows that the 6 k 1 order harmonics where k is odd are removed from the transformer primary. The first harmonics removed, which are also the highest in amplitude, are for k = 1, harmonics 5 and 7. The first harmonics present are then 11 and 13. This property can be generalised by increasing the number of rectifiers and the number of transformer secondaries or the number of transformers by choosing the appropriate phase displacement for each secondary. This solution is commonly employed in the case of very high power rectifiers where current distribution in the various bridges presents no problems. It is frequently used by electrolytic rectifiers (up to 72 phases!). Parallel-connected uninterruptible power supplies (UPS) are of special interest, as the inverters share the output currents and the rectifiers supplying them absorb identical currents.
N2 (i1 - i2 ) N1
where i1 = 1 (3k) = sin (3k t)
4 i3 = 3 (3k) = sin 3k t 3
i3 = sin (3k t) = i1
N2 N1 (i1 - i3) N1 i3 N2 i1 N2
hence
i2 N1 N2 N2
N2 (i1 - i3 ) = 0 N1
As regards three-phase loads, some harmonic orders can be removed by using transformers or autotransformers with a number of displaced secondaries, a solution particularly adopted for powerful rectifiers. The best known of these circuit assemblies is the rectifier consisting of two serial or parallel-connected bridges, supplied by a transformer with two secondaries, one Y and the other delta connected. This assembly produces a 30 degree phase displacement between the volta-ges of the two secondaries.
i3 N1 N2 N2
Fig. 3: zig-zag secondary transformer and attenuation of 3 k order harmonics.
1.5 Using detuning reactors

This solution consists of protecting the capacitors, designed to improve the displacement power factor by installing a series reactor. This reactor is calculated so that resonance frequency matches none of the harmonics present. Typical tuning frequencies are for a 50 Hz fundamental: 135 Hz (order 2.7), 190 Hz (order 3.8) and 255 Hz (order 4.5). Thus for the fundamental, the battery can perform its displacement power factor improvement function, while the high impedance of the reactor limits amplitude of the harmonic currents. The switched-steps capacitors must allow for the priority of certain resonance frequencies.
1.6 Passive harmonic filters

This case differs from the above in that a capacitor is used in series with a reactor in order to obtain tuning on a harmonic of a given frequency. This assembly placed in parallel on the installation has a very low impedance for its tuning frequency, and acts as a short-circuit for the harmonic in question. A number of assemblies tuned on different frequencies can be used simultaneously in order to remove several harmonic orders. Passive filters contribute to reactve energy compensation of the installation. This apparently simple principle nevertheless calls for thorough study of the installation since, although the filter acts as a short-circuit for the required frequency, there is a possibility of resonance risks with other power network reactors on other frequencies and thus of increased previously non-troublesome harmonic levels prior to its installation (see Cahier Technique no. 152).
2 Unity PF rectifiers and active harmonic conditioners
2.1 Introduction
The previous chapter reviewed the techniques and corresponding passive systems used to reduce harmonic disturbances. These systems all modify impedances, impedance ratios or cause the opposition of certain harmonic currents. Other impedance monitoring means are available (which we shall not dare to term intelligent!), which use static converters of ever increasing effectiveness due to the steady increase in semiconductor power component possibilities (see table fig. 4 ). IGBTs made possible the industrial development of power converters able to guarantee nondisturbance at the point of common coupling (unity power factor rectification), and harmonic compensation of power networks (active harmonic compensation). c Unity PF rectification is a technique enabling static converters to absorb a current very similar to a sinusoidal waveform with, in addition, a displacement power factor close to the unit: this highly interesting technique should be used with increasing frequency. c Active harmonic compensation An active harmonic conditioner is a device using at least one static converter to meet the harmonic compensation function. This generic term thus actually covers a wide range of systems, distinguished by: v the number of converters used and their association mode, v their type (voltage source, current source), v the global control modes (current or voltage compensation), v possible association with passive components (or even passive filters). The only common feature between these active systems is that they all generate currents or voltages which oppose the harmonics created by non-linear loads. The most instinctive achievement is the one shown in figure 5 which is normally known as shunt (or parallel) topology. It will be studied in detail in paragraph 3. The serial type active harmonic conditioner (see fig. 6 ) will be mentioned merely as a
Technology Transistor MOS Bipolar IGBT Thyristor GTO
V 500 1,200 1,200 4,500
A 50 600 600 2,500
F (kHz) 50 2 10 1
Fig. 4: typical characteristics of use of power semiconductors in static converters.
Load(s) Power network
Active harmonic conditioner
Fig. 5: shunt-type active harmonic conditioner generates an harmonic current which cancels current harmonics on the power network side.
Active harmonic conditioner Power network
Sensitive load(s)
Fig. 6: series type active harmonic conditioner generates an harmonic voltage which guarantees a sinusoidal voltage on the load terminals.
reminder as it is seldom used. Its function is to enable connection of a sensitive load on a disturbed power network by blocking the upstream harmonic voltage sources. However, in actual practice, this upstream harmonic compensation technique is of little interest since: v the quality of the energy at the point of common coupling is satisfactory in the majority of cases, v insertion of a component in the serial mode is not easy (for example short-circuit current withstand), v it is more useful to examine the actual causes of voltage distortion within a power network (the harmonic current sources). Out of the numerous hybrid alternatives we shall concentrate on the serial/parallel type combining active and passive filtering (see fig. 7 ) which is a very effective solution for harmonic cancellation close to high power converters. However this Cahier Technique does not aim to be comprehensive and deliberately chooses not to treat many topologies. This is because all the other systems are merely variations on a same theme and because the basic solutions are described in this document. Before going on to describe unity PF rectifiers and active harmonic conditioners in detail, it should be noted that there is a certain technological resemblance between these two devices, namely: c when the control strategy of a rectifier bridge (integrating, for example, a BOOST stage) imposes circulation of a current reduced merely to its fundamental, this is called unity PF rectification and the rectifier is said to be clean, c when the current reference applied to this control is (for example) equal to the harmonic content of the current absorbed by a third-party non linear load, the rectifier cancels all the harmonics at the point of common coupling: this
Active harmonic conditioner Power network
Non-linear load(s)
Passive filter(s)
Fig. 7: series/parallel type hybrid filter.
a) Unity PF rectifier Power network

Converter Control processing
Load
b) Active harmonic conditioner Power network
Non-linear load
Converter
Control processing
Fig. 8: unity PF rectifier and active harmonic conditioner.
is known as active harmonic conditioner. Thus the same power topology is able to meet the two separate needs which are nondisturbance and harmonic compensation. Only the control strategy differs (see fig. 8 ).
2.2 Unity PF rectifiers

Whether for rectifiers, battery chargers, variable speed drives for DC motors or frequency converters, the device directly connected with the power network is always a rectifier. This component, and more generally the input stage (power and control) determines the harmonic behaviour of the complete system. Unity PF rectification principle (in single-phase) This consists of forcing the absorbed current to be sinusoidal. Unity PF rectifiers normally use the PWM (Pulse Width Modulation) switching technique. Two main categories are identified according to whether the rectifier acts as a voltage source (most common case) or a current source. c Voltage source converter In this case the converter acts as a backelectromotive force (a sinusoidal voltage generator) on the power network (see fig. 9 ), and the sinusoidal current is obtained by inserting a reactor between the power network and the voltage source. Voltage is modulated by means of a control loop designed to maintain current as close as possible to the required sinusoidal voltage waveform. Even if other non-linear loads raise the power networks voltage total harmonic distortion, regulation can be used to draw a sinusoidal current. The frequency of low residual harmonic currents is the frequency of modulation and of its multiples. Frequency depends on the possibilities of the semiconductors used (see fig. 4).
c Current source converter This converter acts as a chopped current generator. A fairly large passive filter is required to restore a sinusoidal current on the mains side (see fig. 10 ). This type of converter is used in specific applications, for example to supply an extremely well regulated DC current.
Voltage converter implementation principle Its ease of implementation means that the diagram in figure 11 is the one most often chosen (as for certain MGE UPS systems). This diagram uses the voltage generator principle.
iL
iL
Power network
B.e.m.f.
Power network
B.e.m.f. +E iL 0 -E t
0 -I +I iL t
Fig. 9: single-phase diagram equivalent to a voltage PWM converter.

iL L
Fig. 10: single-phase diagram equivalent to a current PWM rectifier.

D i
i1 Power network T v e1 u Vs
Control loop, iL, Vs e1 0

From the source viewpoint, the converter must act like a resistance: (sinusoidal) i1 and in phase with e1 (DPF = 1). By controlling transistor T, the controller forces iL to follow a sinusoidal type current reference with double wave rectification. The shape of i1 is thus necessarily sinusoidal and in phase with e1. Moreover, to keep voltage Vs at its nominal value at the output, the controller adjusts the mean value of iL.
i1 0
v 0 iL 0 u Vs 0
t
Transistor T (normally using MOS technology) and diode D make up the voltage modulator. The voltage (u) thus moves from 0 to Vs according to whether transistor T is in the on or off state. When transistor T is conductive, the current in reactor L can only increase as voltage v is positive and u = 0. The relationship is then: di e = 0. dt L When transistor T is off, the current in L decreases, provided that Vs is greater than v, so that: di e - Vs = 0. dt L For this condition to be fulfilled, voltage Vs must be greater than the peak voltage of v, i.e. the rms value of the ac voltage, multiplied by r. If this condition is fulfilled, the current in L can be increased or decreased at any time. The time evolution of current in L can be forced by monitoring the respective on and off times of transistor T. Figure 12 shows the evolution of current iL with respect to a reference value. The closer are the switching moments of T (i.e. switching frequency is high), the smaller the errors of iL compared with the reference sine wave will be. In this case the current iL is very close to the rectified sinusoidal current, and the line current i1 is necessarily sinusoidal.
Figure 13 shows the time curve and the harmonic spectrum of the current drawn by a unity PF rectifier of a 2.5 kVA UPS. In this case the transistor is a MOS, and the switching frequency equals 20 kHz.
t Reference i iL
Fig. 12: evolution of current iL compared with the reference i.
a) Time form
b) Spectral breakdown Order Max. contrib. as a% of I1 as in IEC 1000-3-2 3 5 7 9 11 13 14.65% 7.26% 4.90% 2.55% 2.10% 1.34% Typical values without unity PF rectifier (Uccx = 1%) 81% 52% 24% 6% 7% 6% Measured value 8.03% 2.94% 3.15% 1.65% 1.09% 1.07%
Fig. 13: current upstream of a clean single-phase rectifier (2.5 kVA UPS - PULSAR-PSX type).
The harmonics of the current absorbed are highly attenuated compared with a switch mode power supply which does not use the unity PF rectification control strategy, and their level is below standard requirements. Filtering of u 20 kHz orders is easy and inexpensive. Three-phase circuits The basic circuit arrangement is shown in figure 14.
We recognise the arrangement in figure 11 with reactors placed upstream of the rectifiers. The operating principle is the same. The monitoring system controls each power arm, and forces the current absorbed on each phase to follow the sinusoidal reference. There are currently no three-phase unity PF rectifiers on the market as additional cost is high. Changes in standards may however stipulate their use.
Power network
Vs
2.3 The shunt-type active harmonic conditioner

Operating principle The shunt-type active harmonic conditioner concept can be illustrated by means of an electro-acoustic analogy (see fig. 15 ). The observer will no longer hear the noise source S if a secondary noise source S generates a counter-noise. The pressure waves generated by the loudspeaker have the same amplitude and are in opposition of phases with those of the source: this is the destructive interference phenomenon. This technique is known as ANR (Active Noise Reduction).
Control microphone
Error microphone
Primary noise source S
Secondary source S'
Controller
This analogy is a perfect illustration of the shunt-type active harmonic conditioner: the aim is to limit or even remove the current (or voltage) harmonics at the point of common coupling by injecting an appropriate current (or voltage) (see fig. 16 ). Provided that the device is able to inject at any time a current where each harmonic current has the same amplitude as that of the current in the load and is in opposition of phases, then Kirchoffs law at point A guarantees that the current supplied by the source is purely sinusoidal. The combination of non linear loads + active harmonic conditioner forms a linear load (in which current and voltage are linked by a factor k). This kind of device is particularly suited for harmonic compensation of LV networks irrespective of the chosen point of coupling and of the type of load (the device is self-adaptive). The following functions are thus performed according to the level of insertion: c local harmonics compensation: if the active harmonic conditioner is associated with a single non linear load, c global harmonics compensation: if the connection is made (for example) in the MLVS (Main Low Voltage Switchboard) of the installation. The shunt-type active harmonic conditioner thus forms a current source independent of
power network impedance, and with the following intrinsic characteristics: c its band-width is sufficient to guarantee removal of most harmonic components (in statistical terms) from the load current. We normally consider the range H2 - H23 to be satisfactory, as the higher the order, the lower the harmonic level. c its response time is such that harmonic compensation is effective not only in steady state but also in slow transient state (a few dozen ms), c its power enables the set harmonic compensation objectives to be met. However this does not necessarily mean total, permanent compensation of the harmonics generated by the loads. Provided that these three objectives are simultaneously achieved, the shunt-type active harmonic conditioner forms an excellent solution as it is self-adaptive and there is no risk of interaction with power network impedance. It should also be noted that the primary aim of this device is not to rephase the fundamental U and I components: insertion of an active harmonic conditioner has no effect on the displacement power factor. Nevertheless, if the load treated is of the multiphase rectifier kind, then the global power factor is indeed considerably improved as the distortion factor is closer to the unit and the
Source current iF
Load current iF + iH
iF A Source iH Compensator current iH
iF + iH
Non-linear load to be compensated
Harmonic active harmonic conditioner
Fig. 16: principle of compensation of harmonic components by shunt-type active harmonic conditioner.
displacement power factor of a rectifier (not controlled) is close to the unit. However this is more a secondary effect than an actual objective! Although the main pupose of the device is harmonic depollution, it can also compensate power factor. In that case the reactive current may be proportionally high and should be accounted for in the product current rating. Structure of the shunt-type active harmonic conditioner This device is broken down into the following two subassemblies (see fig. 17 ). c Power: input filter, reversible inverter, storage components, c Control: reference processing, U/I controls, converter low level control. The main difference between a converter and a unity PF rectifier, described in the previous chapter, lies in the control and monitoring (as the setpoint is no longer a 50 Hz sine wave). If the storage component is a capacitor or battery, the converter has a similar structure to that of the
input stage of the converter with unity PF rectifier (see fig. 18 ). A reactor can also be used (see fig. 19 ). MGE UPS Systems chose Voltage Source Inverter -VSI- for its SINEWAVE range because
Source
Load i measurement To load Input filter Filter i measurement Control and monitoring
Reversible inverter
References
Vcapa monitoring
Fig. 17: schematic diagram showing the structure of the shunt-type active harmonic conditioner.
Fig. 18: diagram showing the shunt type active harmonic conditioner with VSI (voltage source inverter).
Fig. 19: diagram showing the shunt type active harmonic conditioner with CSI (current source inverter).
of its added value in technical and economic terms: wider pass-band, simpler input filter. Moreover the VSI structure technically resembles inverter structure. Control and monitoring electronics Its main function is to control the power semiconductors. As such it must: c control capacitor load (c) on energising, c regulate voltage at the terminals of c, c generate rectifier on/off patterns when it has an inverter function so that the active harmonic condi-tioner permanently supplies a current compensa-ting the non linear harmonic currents (see fig. 16). There are 2 signal processing methods, namely: c the real time method, which is particularly suitable for loads with ultra-fast variations in their harmonic spectrum. It can use the synchronous detection method or use Clark transformations;
c the non real time method, used for loads where the harmonic content of the current absorbed varies slightly in 0.1 s. This method uses the frequency analysis principle and is based on the Fast Fourier Transform (FFT). It enables global or selective treatment of harmonic orders. Examples of performances obtained using non-linear loads In these examples the loads do not operate on full load, as the THD (I) is at its lowest on full load. In the example below, the THD (I) is 30% on full load, whereas it is 80% with a 20% load. c Case of a UPS A shunt-type active harmonic conditioner is parallel-connected on a three-phase uninterruptible power supply of a power of 120 kVA. The current time waveforms are shown in figure 20 . The spectrum of the current absorbed by the load is given in figure 21 and corresponds to an
a) Load current (THD = 80% Irms = 44 A)
b) Source current (THD = 4,6% Irms = 35 A)
Fig. 20: shunt-type active harmonic conditioner associated with a UPS - time waveforms of currents (20% load).
Relative value (% of fundamental)
80 70 60 50 40 30 20 10 7 9 11 3 5 c Source I without active harmonic conditioner c Source I with active harmonic conditioner 0 13 15 17 Harmonic order
Fig. 21: shunt type active harmonic conditioner on UPS - source currents spectrum.
harmonic distortion of 80%. Use of the shunt type active harmonic conditioner considerably attenuates the THD (I) which drops from 80% to 4.6%. The rms current drops by nearly 20%, and the power factor increases by 30% (see fig. 21 and 22 ). c Case of a VSD (frequency converter type) An active harmonic conditioner is parallel-
connected to a variable speed drive for asynchronous motor of a power of 37 kW operating on half-load. The current time waveforms are shown in figure 23 and correspond to an harmonic distortion of 163% for the load current. Figure 24 shows the harmonic spectrum of the source and load currents.
Current characteristics
Without active harmonic conditioner 44.1 1.96 80.8 0.65 0.84 27.7
With active harmonic conditioner 35.2 1.52 4.6 0.86 0.86 1.6
Irms (A)
Peak factor THD (I) as a % Power factor DPF Harmonic Irms (A)
Fig. 22: shunt-type active harmonic conditioner on UPS: measured values.
a) Load current (THD = 163%, Irms = 25 A)
b) Source current (THD = 22,4%, Irms = 15.2 A)
Fig. 23: shunt-type active harmonic conditioner on variable speed drive - waveforms of currents on half-load.
Relative value (% of fundamental)
40
30
20
10
0 3 5 7 9 11 13 15 17 19 c Source I without active harmonic conditioner c Source I with active harmonic conditioner Harmonic order
Fig. 24: shunt-type active harmonic conditioner associated with a variable speed drive - harmonic spectrum of the source current.
Use of the shunt-type active harmonic conditioner considerably attenuates the THD (I) which drops to 22.4%. The rms current drops by nearly 40% (see fig. 24 and 25 ). Performance is lower than in the first case (UPS) since line current fluctuations are much faster. In this cases addition of a 0.3 mH smoothing reactor is recommended. The table in figure 26 illustrates the resulting increase in effectiveness. We can conclude that the shunt type active harmonic conditioner is an excellent means for removing harmonics on a feeder or non-linear load. However:
c removal of all disturbances, even if it is possible, is not necessarily the aim, c it is not suited to voltage power networks exceeding 500 V, c it has no effect on disturbances upstream of the current sensor, c technical and economic considerations may require use combined with a passive component; for example a reactor (see fig. 26 ) or a passive filter to remove the 3rd or 5th harmonic (considerable decrease in shunt type active harmonic conditioner power rating).
Characteristics of current on half-load
Without active harmonic conditioner 25.9 3.78 163 21.7
With active harmonic conditioner 15.2 1.95 22.4 3.3
Characteristics of current on full load
With active harmonic conditioner and smoothing reactor 57.6 1.46 3.4 2
Irms (A)
Peak factor THD (I) as a % Harmonic Irms (A)
Irms (A)
Peak factor THD (I) as a % Harmonic Irms (A)
Fig. 25: shunt-type active harmonic conditioner associated with a variable speed drive - current characteristics.
Fig. 26: shunt-type active harmonic conditioner associated with a variable speed drive with smoothing reactor - current characteristics.
3 Hybrid active harmonic conditioners
Harmonic compensation needs are many and varied, since the requirement may be to guarantee: c non-disturbance of a clean power network by a disturbing load, c or proper operation of a sensitive load (or power network) in a disturbed environment, c or both these objectives simultaneously! The problem of harmonic compensation can thus be handled at two levels (exclusive or combined): c parallel compensation by current source downstream of the point in question: this is the
shunt-type solution described in the previous chapter, c serial compensation by implementing an upstream voltage source. The structures that we shall refer to as hybrid hereafter in this document are those which simultaneously implement both solutions, as shown for example in figure 27 . They use passive filters and active harmonic conditioners. We have chosen to describe three of the many alternatives available.
VA.H.C Zs
Vs(h1) vL Vs(hn) Zf
iL(h1)
iL(hn)
Load
Passive filter
Fig. 27: active/passive hybrid conditioners - example.
3.1 The parallel/series hybrid structure

The diagram in figure 28 illustrates the main subassemblies of this structure, namely: c one (or more) bank(s) of resonant passive filters (Fi), parallel-connected with the disturbing load(s), c an active harmonic conditioner, made up of: v a magnetic coupler (Tr), the primary of which is inserted in series with the passive filter(s), v an inverter (MUT) connected to the secondary of the magnetic coupler. The active harmonic conditioner is controlled so that: Vfa = K SH where: Vfa: voltage at the magnetic coupler terminals, K: value in ohm fixed for each order, ISH: harmonic current from the source.
Load
Source
Is
Passive filter Fi
MUT.
Tr
Vfa
Fig. 28: parallel/series type hybrid conditioners - one line diagram.
In this configuration the active harmonic conditioner only acts on the harmonic currents and increases the effectiveness of the passive filters: v it prevents amplification of upstream harmonic voltages at the anti-resonance frequencies of the passive filters, v it considerably attenuates harmonic currents between load and source by lowering global impedance (passive filters plus active harmonic conditioner). Since not all the power network current flows through the active harmonic conditioner, the components of the latter can be downsized (and in particular the magnetic coupler).
This structure is thus ideal for treating high voltage and power networks, while at the same time ensuring rephasing of fundamental components. Its main drawback is that the passive filters depend on the type of load, thus requiring a preliminary study. Finally, virtually all the pre-existing harmonic voltages (on the source) are present on the load side. This configuration can therefore be compared with the shunt type active harmonic conditioner.
3.2 The series/parallel hybrid structure

The diagram in figure 29 shows that this structure contains the main subassemblies of the previous structure, the only difference being in the connection point of the coupler primary (in series between source and load). The active harmonic conditioner control law is the same: its aim is for the active harmonic conditioner to develop a voltage which opposes circulation of harmonic currents to the source. It therefore acts as an impedance (of value K fixed for each order) for harmonic frequencies. Passive filtering is thus more efficient (as the presence of this serial impedance forces circulation of the harmonic load currents to the passive filters). Moreover, the serial filter isolates the load of the harmonic components already existing on the source and prevents passive filter overload. This topology is thus most often referred to as an harmonic isolator since, in some respects, it isolates the source of a disturbing load, and, reversely, prevents overload of a passive filter by upstream disturbance. It should be noted that this topology generates sizing and protection problems for the magnetic coupler, since: c total load current flows through this coupler, c a very high current wave is applied in the event of a short-circuit. A possible solution to both problems may be to use a transformer with an additional secondary winding (see fig. 30 ). Compensation then takes place magnetically by directly acting on the flow.
Source
Is
Vfa
Load
Source
Load
Is
Vs
Tr. MUT. Active harmonic conditioner
Vc
Passive filter Fi
Fig. 29: series/parallel type hybrid conditioners.
Fig. 30: hybrid conditioner with injection by transformer.
3.3 Parallel combination of passive filters and active harmonic conditioner

The principle consists of parallel connection of one (or more) tuned passive filter(s) and a shunt type active harmonic conditioner (see fig. 31 ). In this case also, the active harmonic conditioner and the passive filter prove the ideal combination. It may prove useful to limit (by the FFT technique), the action of the active harmonic conditioner to the orders not treated by the passive filters. This structure is used (as applicable) to: c improve the harmonic cancellation obtained using only passive filters, c limit the number of orders of passive filters, c improve the effectiveness of the active harmonic conditioner only (for the same power effectiveness of the active harmonic conditioner). Nevertheless, this combination does not prevent passive filter overloads or the effects of antiresonance with power network impedance. In short These hybrid structures do not possess the universal character of the shunt type active harmonic conditioner as passive filters need to be chosen (in terms of type, number of orders and tuning frequencies) according to the kind of harmonic currents generated by the load. The presence of the active harmonic conditioner downsizes the passive filters and reinforces their
Load
FP1
FP2
Fig. 31: parallel connection of active harmonic conditioner and passive filters - principle.
effect. Vice versa, the addition of an active harmonic conditioner of reduced power to an existing installation increases the efficiency of existing passive filters.
3.4 The performances of hybrid structures

Prototypes have been designed, produced and tested in partnership with Electricit de France (which is the main power utility in France). These models comprised two banks of resonant passive filters tuned on orders 5 and 11 (harmonic compensation of a UPS-type load) or 5 and 7 (variable speed drive load). The result of the tests combining two types of hybrid filters with a frequency converter (variable speed drive for asynchronous motor) is given below. Parallel/series configuration (see fig. 28) The test circuit characteristics are defined in the table in figure 32 . Comments: this topology is not suitable for treating power networks with a high upstream voltage THD. However, its current
Circuit characteristics Source 400 V, three-phase, 600 kVA, 5%, THD (Vs) < 1.5% 130 kW, 70% load, 0.15 mH smoothing reactor. 35% 9% 2%
Load
Measurements taken THD (IL) THD (Is) THD (VL)
Fig. 32: parallel/series type conditioner characteristics and results.
TDH(VL) 3
TDH(IS)
Is (in %)
2
VL (in %)
40 30 20
1 10 0 Without filter Passive filters only Passive filter and active harmonic conditioner
Fig. 33: parallel/series type hybrid conditioner associated with a variable speed drive - evolution of THD (VL) and THD (IS).
performances are totally respectable (the THD (I) is reduced from over 35% to 9%) (see fig. 33 ). It is thus particularly well suited for treating power networks with low upstream THD, or for which serial insertion of a device is particularly problematic. Series/parallel typeconfiguration (see fig. 29) The test circuit characteristics are defined in the table in figure 34 . Comments: the performances are in this case also totally satisfactory even if the quality of the source voltage (THD (u) very low) does not allow to appraise performance in terms of isolation. The source current THD is however reduced from more than 35% to 11% (see fig. 35 ).
Circuit characteristics Source 400 V, three-phase 600 kVA, 5 %, THD (Vs) < 1.5 % 130 kW, 70% load, 0.15 mH smoothing reactor.
Load
Measurements taken THD (IL) THD (IS) THD (VL) 35 % 11 % 2.1 %
Fig. 34: series/parallel type hybrid conditioner: characteristics and result.
TDH(VL)
TDH(IS)
5 4 3 2 1 0
Is (in %)
40 30 20
VL (in %)
10
Without filter
Passive filters only
Passive filter and active harmonic conditioner
Fig. 35: series/parallel type hybrid conditioner associated with a variable speed drive - reading of TH (V L) and THD (IS).
Passive filter current remains constant and is thus representative of isolation from the source. Additional tests proved that for very high upstream distortion (THD (V) = 11%), voltage quality at the load terminals continued to be good (THD (VL) = 4.7%). Characteristics of the active solutions We have now dealt with series and parallel type active harmonic conditioners and with hybrid structures.
To round off this chapter, we propose to summarise the qualities of these various active solutions used to combat harmonic disturbance. The table in figure 36 shows that, except for a few special cases, the shunt type active harmonic conditioner and the parallel-connected structure are the solutions to be preferred in low voltage.
Type of filter Series Criterion Schematic diagram

Sup. A.H.C Load
Shunt hybrid
Parallel hybrid
Parallel/series hybrid
Load
Sup. P.F. Load
Series/parallel type
Sup. A.H.C
Load
Sup. A.H.C P.F.
Sup.
A.H.C P.F.
Load
A.H.C
A.H.C.: Active Harmonic Conditioner Action on Performance Active harmonic conditioner sizing Short-circuit impact Insertion Improvement of DPF Open-endedness Resonance risk Uh/source +++ fund + harm great difficult no no NA (not applicable)
P.F.: Passive Filter Ih/load +++ harm. none easy possible yes NA (not applicable) Ih/load +++ harm. none easy yes yes yes Ih/load ++ harm. none easy yes no no Ih/load, Uh/source ++ fund + harm great difficult yes no no
Fig. 36: summary of the various active solutions to combat harmonic disturbance.
4 Implementing a shunt type active harmonic conditioner
We would first like to emphasise that our aim is not to act as a selection guide between the various types of harmonic compensation techniques (both active and passive), but rather to present the criteria used to size and insert the active harmonic conditioners. Furthermore, a selection guide would imply that the various solutions given are available in product form. At present, given that both the traditional
solutions and the hybrid solutions require indepth study and suitable solutions, only the shunt type active harmonic conditioners are available on the market (they require merely a simple study). We shall thus concentrate on identifying the main parameters that potential active harmonic conditioner users need to know in order to make the right choice.
4.1 Objective and context

Knowing the mechanisms The main problem of harmonic phenomena is undeniably linked to their very weak visibility. Although it is usually easy to observe deterioration in wave quality (voltage and/or current) at one or more points, the combinational function between the various sources (self-sufficient or not), loads and topology of the power network is no simple matter! Moreover, the association between harmonic phenomena (often overlooked) and the malfunctions observed in the power network components (often random) is not instinctive. Knowing the power network and its topology The first preliminary requirement thus concerns the power network environment: implementation of an harmonic compensation technique requires knowledge of the entire power network (sources, loads, lines, capacitors) and not just a fragmented view limited merely to the zone concerned. This single-line diagram is in some respects the first component of our tool box. Carrying out an inventory We have first placed an harmonic distortion analyser in this tool box, vital for quantifying disturbance at various points of an existing installation. Identifying and characterising disturbing equipment We need to identify the main disturbing equipment(s) and their respective spectra. The latter can be obtained either by measurements or by consulting the technical specifications provided by each manufacturer. Defining the harmonic compensation objective The second preliminary requirement concerns the actual objective of the action considered: the method used differs considerably according to whether you wish to correct malfunctioning observed, or to ensure compliance with the specifications of power utility or a non-linear load manufacturer. Short term power network changes must also be taken into consideration. For example this stage must enable identification of at least: c the type of compensation (global or local), c the power rating at the node considered, c the type of correction required (on voltage and/ or current distortions), c the reactive energy compensation need. Once the above analyses are complete, the most advantageous technical and economic solution must be chosen. The same objective often has several technical possibilities, and the problem is in most cases to make a choice according to the individual difficulties of each electrical installation. For example, isolation or decoupling by impedance of disturbing loads is easily carried out on new installations provided it is considered in the design phase. However it frequently generates unacceptable difficulties on existing power networks. It is thus obvious that no active solutions (regardless of the type) can be systematically chosen, but that an analytical approach is required in which active harmonic conditioner cost alone is not necessarily the most important factor. Although Active harmonic conditioners have undeniable advantages over passive filters, they are not necessarily preferred particularly for existing installations already equipped with passive filters. The insertion of a serial or parallel type active harmonic conditioner, after study, is a good solution. We shall now use experience acquired on site to describe the implementation of a shunt-type active harmonic conditioner which is the simplest solution.
4.2 The insertion point of a shunt-type active harmonic conditioner

The connection principle of a shunt type active harmonic conditioner is shown in figure 37. In our example it is inserted in parallel mode in the LV switchboard of an installation, and the only interaction with the power network to be treated, is the insertion of the current sensors.
LV distribution switchboard
Supply network
Load to be compensated Harmonic compensation current Harmonic current to be compensated Active harmonic conditioner
Fig. 37: connection of a shunt type active harmonic conditioner: principle.
As regards insertion of the active harmonic conditioner, harmonic compensation can be considered at each level of the tree structure shown in figure 38. The compensation mode may be termed global (position A), semi-global (position B) or local (position C) according to the point of action chosen. Although it is hard to lay down strict rules, it is obvious that if disturbance is caused by a large number of small loads, the mode preferred will be global, whereas if it is caused by a single high power disturbing load, the best result will be achieved using the local mode. Local harmonic compensation The shunt type active harmonic conditioner is directly connected to the load terminals. This mode is the most efficient provided that the number of loads is limited and that the power of each load is significant compared with global power. In other terms, the loads treated must be the main generators of the harmonic disturbances. Circulation of harmonic current in the power network is avoided, thus reducing losses by Joule effect in upstream cables and components (no oversizing of cables and transformers) as well as reducing disturbances of sensitive loads. It is worth pointing out, however, that the shunt type active harmonic conditioner lowers source impedance at the connection point, and thus slightly increases current total harmonic distortion between the connection point and the load.
MV
LV Main LV switchboard A Active harmonic conditioner
Subdistribution switchboard B Active harmonic conditioner
Terminal switchboard
C Active harmonic conditioner
Fig. 38: the various insertion points of a shunt-type active harmonic conditioner: principle.
Semi-global compensation The active harmonic conditioner, connected to the input of the LV subdistribution switchboard, treats several sets of loads. The harmonic currents then flow between the MLVS and the loads of each feeder. This type of compensation is ideal for multiple disturbing loads with low unitary power, e.g. on floors in service sector buildings (office equipment and lighting systems). It also makes it possible to benefit from nonalgebraic summing between loads, at the cost of a slight increase in losses by Joule effect on each feeder treated. NB: this type of compensation can also be applied to a single feeder, thus limiting harmonic compensation to a single type of load (see fig. 37).
Global compensation This form of compensation contributes rather to compliance with the point of common coupling according to power utility requirements, than to the reduction of internal disturbances in the customers power network. Only the power transformer(s) actually derive direct advantage from harmonic compensation. Nevertheless, this form has a serious advantage for operation in autonomous production mode as a result of the numerous interactions between disturbing loads and generator sets with high harmonic impedance. However, compared with local compensation, this compensation technique results in a reduction in active harmonic conditioner power rating which benefits from the non-algebraic summing of the disturbing loads throughout the power network.
4.3 Sizing a shunt-type active harmonic conditioner

The main factor to consider when sizing a shunt type active harmonic conditioner is its power rating (or more precisely its rms current): the rms curren ICA RMS is the current that can be permanently generated by the active harmonic conditioner. Other characteristic active harmonic conditioner factors are its bandwith and its dynamic capacity: c the active harmonic conditioner bandwith is defined by nmin and nmax, the (minimum and maximum) action orders of the active harmonic conditioner. The following can be written: Provided that the above condition is met, the new current total harmonic distortion (upstream) can be calculated once the active harmonic conditioner is put into operation:
n = nmin 2 CH(n) + n
THD (%) =
n=2
n = nmax + 1
2 CH(n)
CH
(1)
This formula is used to determine whether the maximum theoretical performance of the active harmonic conditioner is compatible with the target objective. It can be simplified still further, if we consider the specific case of MGE UPS Systems products for which nmin = 2 and nmax = 23:
n 2 CH(n)
CA RMS (A ) =
n = nmax n = nmin
2 CA (n )
THD (%) =
n = 24
CH
c the active harmonic conditioner current di ) is tracking dynamic capacity (expressed by dt the capacity of the active harmonic conditioner to track rapidly varying references. NB: these last two factors are not considered to affect size, since they form a characteristic inherent in the active harmonic conditioner and not an adjustable parameter. Choosing nominal rating: Provided that the spectrum of the current to be treated ICH is known, the nominal current of the active harmonic conditioner IN CA RMS, can be determined such that:
(1)
N CA RMS (A )
n = nmax n = nmin
2 CH(n)
Furthermore, the above nominal rating selection rule must be weighted by the following practical considerations: c the harmonic spectrum of most loads is significant only in the band h3 to h13, c the purpose of inserting the active harmonic conditioner is not to cancel the THD (I) but to limit it below a predefined level (e.g. 8%), c an active harmonic conditioner can be chosen with a rating lower than IN CA RMS, and then operate in permanent saturation (by permanent, automatic limitation of its rms current). Finally, parallel-connection of a number of active harmonic conditioners at the same insertion point is technically feasible, a solution which may prove of interest for upgrading of a pre-equipped network.
4.4 Application examples

Reduction of line distortions As regards high rise buildings or buildings occupying a large ground surface area, the main problem concerns the lengths of the lines between the point of common coupling (MV/LV transformer) and the loads. This is because, irrespective of voltage wave quality at the origin of the installation and of the precautions taken for the lines (choice of cable diameter, splitting,...), voltage total harmonic distortion increases at the same time as altitude or distance! As from a specific point, therefore, voltage distortion can be considered unacceptable in permanent mode, and the shunt-type active harmonic conditioner provides an interesting alternative to traditional solutions (e.g. isolation by suitable LV/LV coupling transformers). Let us consider the example of a three-phase UPS supplying a set of computer loads at the end of a 60 m line. We then observe a voltage distortion of 10.44% (phase to phase) and of 15.84% (phase to neutral) at load level. This deterioration is the result of a combination of two factors, namely: c UPS sensitivity (with non-PWM control) to the non-linear characteristic of the downstream current, c the mainly inductive characteristic of the line which amplifies distortions. The proposed solution is illustrated in figure 39 and is based on insertion of a shunt-type active harmonic conditioner as close as possible to loads. Performances are then totally satisfactory with respect to the objective: the THD (U) drops to 4.9% phase to phase and to 7.2% phase to neutral. Combination of shunt-type active harmonic conditioner and passive components Effect on tariffs A pumping station is designed to maintain constant water pressure on the drinking water distribution network (see fig. 40 ). The motordriven pump P1 is thus controlled by a variable speed drive with frequency converter. In this particular instance, the main objective was compliance of the source current spectrum with the power utilitys requirements. With no filtering device, the authorised harmonic emission level was: c greatly exceeded on order 5, c more or less reached on orders 7 and 11.
L Starter Active harmonic conditioner
Non PWM UPS sn = 200 kVA
Connecting cable: 60 m/50 mm2
n computers & similar loads
Fig. 39: using an active harmonic conditioner to treat voltage total harmonic distortion at the end of a 60 m cable.
Variable speed drive
P2
P1
Fig. 40: pumping station diagram (main low voltage switchboard).
The choice made is a combination of smoothing reactors and a shunt type active harmonic conditioner: performances are shown in figure 41 . c All the harmonic orders are well below authorised emission limits. c The current total harmonic distortion is reduced by 89%.
An advantage particularly appreciated by the customer is the reduction of his contracted power (in kVA). This example also shows that the combination of an active harmonic conditioner + smoothing reactor is particularly suitable in view of the high degree of disturbance.
25
20
15
10
2 3 4 5 6 c authorised I c I without compensation
7 9 11 13 c I with compensation
15
17
Fig. 41: pumping station - spectral representation of harmonic currents.
5 Conclusion
The profusion of non-linear loads makes harmonic distortion of power networks a phenomenon of increasing amplitude, the effects of which cannot be ignored since almost all the power network components are in practice affected. Up to now the most popular solution was passive filtering. However, an attractive alternative to this complex and non risk-free solution is now commercially available in the form of active harmonic conditioners. These devices use a structure of the static power converter type. Consequently, semiconductor
progress means that converters, which are normally harmonic disturbers, now form efficient, self-adaptive harmonic compensation devices . The easy to use, self-adaptive shunt- type active harmonic conditioner, which requires virtually no preliminary studies prior to use, is the ideal solution for harmonic compensation on a non-linear load or LV distribution switchboard. However it does not necessarily replace passive filters with which it can be combined advantangeously in some cases.
Appendix: review of harmonic phenomena
Definition and characteristic quantities

Joseph FOURIER proved that all non-sinusoidal periodic functions can be represented by a sum of sinusoidal terms, the first one of which, at the recurrence frequency of the function, is said to be fundamental, and the others, at multiple frequencies of the fundamental, are said to be harmonic. A DC component may complete these purely sinusoidal terms. FOURIERs formula:
y ( t) = Yo +
n= n=1
Individual harmonic ratio This quantity represents the ratio of the value of an harmonic over the value of the fundamental (Y1), according to the standard definition or over the value of the alternating quantity (Yrms). Y Hn (%) = n Y1 (Frequency) spectrum Representation of harmonic amplitude as a function of their order: harmonics value is normally expressed as a percentage of the fundamental. Power factor (PF) and Displacement Power Factor (DPF) It is important not to confuse these two terms when harmonics are present, as they are equivalent only when currents and voltages are completely sinusoidal. c The power factor () is the ratio between active power P and apparent power S: P = S c The displacement power factor (cos 1) relates to fundamental quantities, thus: P cos 1 = 1 S1 In pure sinusoidal waveform: cos 1 = cos = Distortion factor The IEC 146-1-1 defines this factor as the ratio between the power factor and the displacement power factor cos 1 :
Yn 2 sin(n t - n )
where: - Yo: DC component value, generally nil and considered hereafter to be nil, - Yn: rms value of the nth harmonic component, - o: angular frequency of the fundamental, - n: displacement of the nth harmonic component. The notion of harmonics applies to all periodic phenomena irrespective of their nature, and particularly to AC current. rms value of a non-sinusoidal alternating quantity There is similarity between the normal expression of this rms value calculated from the time evolution of the alternating quantity (y(t)) and the expression calculated using its harmonic content:
Yrms =
1T 2 y (t)dt = T0
n=
n=1
Yn2
Note that when harmonics are present, the measuring instruments must have a wide bandwidth (> 1 kHz). Total harmonic distortion Total harmonic distortion is a parameter globally defining distortion of the alternating quantity:
n=
THD (%) = 100
n=2
Yn2
Y1
cos 1 It is always less than or equal to 1.

Peak factor The ratio of peak value over rms value of a periodic quantity.
cos 1 : =
There is another definition which replaces the fundamental Y1 with the total rms value Yrms. This definition is used by some measuring instruments.
Fc =
Ypeak Yrms
Origin and transmission

Linear and non-linear loads A load is said to be linear when there is a linear relationship (linear differential equation with constant factors) between current and voltage. In simpler terms, a linear load absorbs a sinusoidal current when it is supplied by a sinusoidal voltage: this current may be displaced by an angle j compared with voltage. When this linear relationship is not verified, the load is termed non-linear. It absorbs a nonsinusoidal current and thus harmonic currents, even when it is supplied by a purely sinusoidal voltage (see fig. 42 ). Voltage and current total harmonic distortion A non-linear load generates harmonic voltage drops in the circuits supplying it. In actual fact all upstream impedances need to be taken into consideration right through to the sinusoidal voltage source. Consequently a load absorbing harmonic currents always has a non-sinusoidal voltage at its terminals. This is characterized by the voltage total harmonic distortion:
n=
The greater the non-linearity of the load, the larger the voltage distortion and the higher the order of the harmonic currents (inductive source impedance 2 f1 n L). Remember that current total harmonic distortion is:
n=
100
n=2
n2
1
In order to illustrate the main types of behaviour of the main sources, figure 43 shows the evolution of their impedances as a function of frequency. For further details, readers can consult Cahier Technique no. 159. Do not forget that large diameter cables are mainly inductive and that small diameter cables have a non-negligible resistance.
I
U
THD (%) = 100
n=2
(Z n n )
U1
1 F
where Zn is the total source impedance at the frequency of harmonic n, and In the rms value of harmonic n.
Fig. 42: current absorbed by a non-linear load.
Zs % Zc
Ratio of output impedance over nominal load impedance
100
AC generator X"d = 12 %
50 Transformer Uccx = 4 %
PWM inverter 0 50 250 500 750 F (Hz)
Fig. 43: output impedance of the various sources as a function of frequency.
Deforming loads
Most deforming loads are static converters. They may be powerful and few in number, or low-power and plentiful. Some examples are: c fluorescent lamps, dimmers, c computers, c electrical household appliances (television sets, microwaves, induction plates). Nowadays the proliferation of low power devices is chiefly responsible for increased voltage harmonic distortion in power networks. Figure 44 illustrates the current absorbed by a few loads, and figure 45 the matching harmonic spectra (typical values).
Type of converter 1: Light dimmer or heating regulator
Diagram
Current (& voltage) waveform
i e i e R
0
= /2
2: Switch mode power supply rectifier, for example: c computer c electrical household appliances
i
i
3: Three-phase rectifier with front end capacitor, for example: variable speed drive for asynchronous motors
e1 e2 e3
i1
i1 i2 i3
e1
4: Three-phase rectifier with DC filtering reactor, for example: battery charger.
Lc
e1
i1 e1 e2 e3 i2 i3 C R
i1
5: Three-phase rectifier with AC smoothing reactor, for example: high power UPS
i1 e1 e2 e3 i2 i3 C R
e1
i1
Fig. 44: curve of the current absorbed by some non-linear loads.
N 1 2 3 4 5
H3 54 75 0 0 0
H5 18 45 80 25 33
H7 18 15 75 7 3
H9 11 7 0 0 0
H11 11 6 40 9 7
H13 8 3 35 4 2
H15 8 3 0 0 0
H17 6 3 10 5 3
H19 6 2 5 3 2
Fig. 45: example of the harmonic spectrum of currents absorbed by the loads in figure 44.
Harmful effects of harmonics

Effects on low current appliances and systems Harmonic distortion may cause: c malfunctioning of certain appliances which use voltage as a reference to generate semiconductor controls or as a time base to synchronise certain systems; c Disturbances by creating electromagnetic fields. Thus when data transmission lines circulate in the vicinity of power lines through which harmonic currents flow, they may be subjected to induced currents able to cause malfunctioning of the equipment to which they are connected; c finally circulation of harmonic currents in the neutral provokes a voltage drop in this conductor: thus in the case of the TN-C earthing system, the frames of the various devices are no longer at the same potential, which may well interfere with information exchange between intelligent loads. Moreover current circulates in the metallic structures of the building and creates disturbing electromagnetic fields. Effects on capacitors Capacitor impedance decreases as frequency increases. Consequently if voltage is distorted, relatively strong harmonic currents flow in these capacitors whose aim is to improve the DPF. Furthermore the presence of reactors in the different parts of the installation reveals a risk of resonance with the capacitors which may considerably increase the amplitude of an harmonic in the capacitors. In practice, capacitors should never be connected on installations with a voltage total harmonic distortion greater than 8%. Effects on transformers Harmonics generate additional losses in the transformers: c losses due to Joule effect in the windings, accentuated by the skin effect, c losses by hysteresis and eddy current in the magnetic circuits. To take these losses into consideration, a standardised empirical formula (NFC 52-114) is used to calculate the derating factor k to be applied to a transformer.
k = 1
n=
where Hn =
1 + 0.1
n =2
2 Hn n1.6
n 1
For example where H5 = 25% ; H7 = 14% ; H11 = 9% ; H13 = 8%, the factor k is 0.91. Effect on ac generators In the same way as for the transformer, the harmonics create additional losses in the windings and magnetic circuit. Furthermore the harmonics create pulsating torques which generate vibrations and additional overheating in the damping windings. Finally as the subtransient reactance is relatively large, the voltage total harmonic distortion quickly rises with the increase in harmonic currents. In practice, limitation of the current total harmonic distortion to a value less than 20% is accepted, with a limit of 5% for each harmonic order. Beyond these values, manufacturers must be consulted as to the spectrum of current really absorbed by the loads. Effect on power lines and in particular on the neutral conductor Harmonic currents create additional losses in conductors accentuated by the skin effect. Losses are even more serious when single-phase loads absorb harmonic currents 3 and multiples of 3. These currents are in phase and are added together in the neutral conductor. With, for example, an harmonic 3 of 75%, the current flowing in the neutral is 2.25 times the fundamental. The current in each phase is only
1 + 0.752 = 1.25 times the fundamental.
Special attention must thus be paid to the sizing of the neutral conductor when non linear loads are present. The TN-C earthing system is strongly advised against.
Standards and recommendations

Electricity is today regarded as a product, especially in Europe with the directive of July 25th 1985. The EN 50160 standard defines the main characteristics at the customers point of common coupling for a low voltage public supply network, and in particular the harmonic voltage levels (class 2 levels in the table in figure 47 ). These are the levels of compatibility in terms of electromagnetic compatibility (see fig. 46 ). In addition to this European standard, the maximum levels of the various harmonic orders are defined in IEC 61000. c For low voltage public supply networks: IEC 61000-2-2 and CIGRE recommendations. c For medium and high voltage public supply networks: IEC draft standard for medium voltage and CIGRE recommendations. c For low voltage and medium voltage industrial installations: IEC 61000-2-4. By way of illustration, the table taken from this standard gives the harmonic levels of compatibility in three standard situations (classes) (see fig. 47 ). To ensure these levels are not reached, limits must be set for the disturbances emitted (emission level) by devices either considered separately, or for a group of devices as regards their point of connection to the power network. IEC 61000-3-2 deals with low voltage and devices absorbing current of less than 16 A, and the IEC 61000-3-4 draft guide deals with devices absorbing current greater than 16 A. Although there is no standard for industrial applications, there is a sort of consensus concerning the notion of stages for authorisation of connection to the public supply network: stage 1 means automatic acceptance for low powers with respect to contracted power, stage 2 means acceptance with reservations (a single consumer must not exceed a level representing half the level of compatibility) and stage 3 means exceptional and provisional acceptance when the previous level is exceeded. Finally, to guarantee proper operation of devices, these devices must be able to withstand levels of disturbance greater than the levels of compatibility given in figure 47 should these levels be overshot (permissible temporarily): this is their level of immunity.
Level of disturbance
Level of susceptibility: level from which a device or system starts malfunctioning. Level of immunity: level of a disturbance withstood by a device or system. Level of compatibility: maximum specified disturbance level that can be expected in a given environment. Level of emission: maximum level authorised for a consumer on the public supply network or for a device.
Fig. 46: the various levels of disturbance for compatibility of non linear/sensitive equipment.
Harmonic order 2 3 4 5 6 7 8 9 10 11 12 13 THD
Class 1 (sensitive devices and systems) 2 3 1 3 0.5 3 0.5 1.5 0.5 3 0.2 3 5%
Class 2 (public and industrial supply networks) 2 5 1 6 0.5 5 0.5 1.5 0.5 3.5 0.2 3 8%
Class 3 (for connection of large non-linear loads) 3 6 1.5 8 1 7 1 2.5 1 5 1 4.5 10%
Fig. 47: ratio (as a %) of acceptable harmonic voltages (compatibility).
n184
electrical installation dependability studies
Sylvie LOGIACO An ISTG Engineer (Institute Scientifique et Technique de Grenoble) graduating in 1987, she was initially involved in risk analysis in the chemicals industry ; at Pechiney, then at Atochem. With Schneider since 1991, she is part of the centre of excellence for Dependability for which she has carried out a large number of dependability studies on electrical and monitoring and control installations.
E/CT 184 first issued September 1997
glossary
Availability: Percentage of time for which the system functions correctly. Dependability: Generic term which combines the independent variables of reliability, availability, maintainability and safety. Hasard analysis: Based on the functional analysis, this is the analysis of the failures that can occur in a system (in practice it is a synonym for dependability study). Feedback of experience: Operational reliability data collected during failure of equipment in operation. Maintainability: The aptitude of a system to be repaired quickly. Reliability: The aptitude of a system to function correctly for as long as possible. Safety: The aptitude of a system to not put people in danger.
electrical installation dependability studies
contents
1. Introduction 2. How a study is carried out General Dependability studies The chronological order of phases Expression and analysis of the requirements Functional analysis of a system Failure mode analysis Reliability data Modelling Dependability criteria assessment calculations Comparing two electrical network configurations in a factory The interest of remote monitoring and control in an EHV substation Dependability study packages Modelling tools p. 4 p. 6 p. 9 p. 9 p. 10 p. 11 p. 11 p. 12 p. 14
In both the industrial and tertiary sectors, the quality of power supply is of increasing importance. The quality of the electricity product, besides imperfections such as variations in voltage and harmonic distortion, is basically characterised by the availability of the electrical energy. Loss of power is always annoying, but it can be particularly penalising for information systems (computing monitoring and control); they can even have disastrous consequences for certain processes and in certain cases endanger people's lives. Dependability studies enable electrical availability requirements to be incorporated in the choice of electrical network to be installed. They also enable two different installation configurations to be compared ... the most expensive one is not always better ... The aim of this Cahier Technique is show how dependability studies are performed: methodology and tools. Two examples of studies are given: that of an electrical network in a factory and that of the monitoring and control system of a high voltage substation. These studies are becoming increasingly easy to perform thanks to computing tools.
3. Examples of studies
p. 15
4. Dependability tools 5. Conclusion 6. Bibliography
p. 17 p. 20 p. 20 p. 21 p. 22
1. introduction
general
The voltage at the terminals of a current consumer can be affected by phenomena originating in the distributor's network, the electrical installation of a subscriber connected to that network or the electrical installation of the user who is suffering the disturbance. Disturbances in the electricity product c the main characteristics of the voltage supplied by both the MV and LV public distribution network are those defined by European standard EN 50160. This details the tolerances that must be guaranteed for both voltage and frequency as well as the disturbance levels that are usually encountered; e.g. harmonic distortion. The table in figure 1 details the standard's recommended values. At any moment in time therefore, the quality of the energy supplied to the current consumers in an installation can be affected by various disturbances, either imposed by the external supply network or self-generated within the internal distribution network. The operations of current consumers that are either independent or federated in systems are affected by these disturbances. c malfunctions and the type and cost of the damage incurred depend both on the type of current consumers and on the installation's criticality level. Thus, a momentary break in supply to a critical current consumer can have serious consequences on the installation's operation without it being intrinsically affected. c in all cases, a study detailing the effects of the suspected disturbances must be performed. Measures must be taken to limit their consequences. c the table in figure 2 shows the disturbances that can usually be found in electrical networks, their causes and the solutions that are possible to reduce their effects. c the reduction of the effects of harmonics, Flicker, voltage imbalances,
standard EN 50160 frequency voltage amplitude rapid voltage variations voltage drops (indicative values) low voltage supply 50 Hz 1% for 95% of a week 50 Hz + 4%, - 6% for 100% of a week for each one week period, 95% of the average effective values over 10 minutes must be within the range of Un 10% from 5% to 10% of Un (4 to 6% in medium voltage) c amplitude: between 10% and 99% of Un most voltage drops are <60% of Un c duration: between 10 ms and 1 minute, most voltage drops < 1s c number: several tens to 1 thousand per year c amplitude: 100% of Un c duration: up to 3 minutes, 70% are less than 1s c number: several tens to several hundreds per year c amplitude: 100% of Un c duration: over 3 minutes, c number: between 10 and 50 per year
brief loss of power (indicative values)
long loss of power (indicative values)
fig. 1: network disturbances : values found in the standard.
frequency variations and overvoltages is achieved by installing equipment suited to each case. The design and choice of connection points for such equipment are decided by performing detailed studies that are outside the scope of this document (see bibliography). Loss of power For industrial processes and low voltage systems this is increasingly difficult to tolerate since it generates unacceptable costs. c immunity to loss of power requires dedicated equipment such as uninterruptible power supplies and independent generator sets. This equipment is generally not enough to resolve all the problems. The network configuration, automatic power restoring control devices, the level of reliability of the equipment, the selectivity of protection devices as well as the maintenance policy, all play an important role in reducing and eliminating down time. Minimising loss of power requires reliability / dependability studies to be performed that take account of all these factors as well as the frequency and
duration of the loss of power that is acceptable to the installation. These studies enable the best suited configuration and equipment to be determined for the operator's requirements. They generally require the current consumers or systems to be classed according to their sensitivity level as well as distinguishing between: v current consumers that accept prolonged stoppage: 1 or more hours (non priority), v current consumers accepting stoppages of several minutes long (priority), v current consumers which must have the power restored to them after several seconds (essential) v current consumers not accepting any break in supply (vital), As an example, figure 3 shows the simplified diagram of a network in which this distinction has been performed: v vital current consumers not tolerating the slightest break in supply, are supplied power through an uninterruptible power supply. v power is restored to essential current consumers a few seconds after network
disturbance frequency variations rapid voltage variations
possible causes c supply network c grouped operation of independent generators c supply network c arc furnace c welding machine c load surge c supply network c high load demands c external and internal faults
main effects c variation in speed of motors c malfunctioning of electronic equipment c lighting flicker c variation in speed of motors
possible solutions c UPS
c increasing short circuit power c modifying the installation configuration
voltage drops
c extinction of discharge lamps c malfunctioning of regulators c speed variation, stopping of motors c switching of contactors c disturbances in digital electronics systems c malfunctioning of power electronics c equipment stoppage c installation stoppage c loss of production c switching of contactors c various malfunctions c overheating of motors and alternators
c UPS c increasing short circuit power c modifying the installation configuration
brief and long loss of power
c supply network c reclosing operations c internal faults c source switching c supply network c many single phase loads c lightening c switchgear operation c insulation fault c supply network c a lot of non linear current consumers
c UPS c independent generators c modifying the network configuration c setting up of a maintenance policy c increasing short circuit power c modifying the network configuration c balancing of single phase loads c rebalancing devices c lightening arrestors c choice of insulation level c control of the resistance of earthing electrodes c increasing short circuit power c modifying the installation configuration c filtering
voltage imbalances
overvoltages
c breakdown of equipment
harmonics
c overheating and damaging of equipment, mainly motors and capacitors c malfunctioning of power electronics
fig. 2: disturbances in networks, causes, effects and solutions.
failure, as soon as the voltage and frequency of the generator set have stabilised, v power is restored to priority current consumers once the essential current consumers have been started up, v power is not restored to non-priority current consumers, which accept a long break in power supply, until the external network is functioning again. By appropriate selection of the configuration and the automatic source switching control devices (see. Cahier Technique n161) we can optimise the positioning and dimensioning of back up and replacement supplies in order to meet operating constraints. It should be remembered that the choice of neutral earthing arrangement is an important factor; it is clearly established that for current consumers requiring a high level of availability, it is highly advisable to use the isolated neutral arrangement since it enables continuity of supply on the occurrence
grid supply
G TR no back-up source changeover switch
independent source
load restorable
load shedding
backed-up UPS
down time: several hours types: non-priority
several secs. essential
none (high quality) vital
several mins. priority
fig. 3: a reliable power supply. Simplified network diagram.
Cahier Technique Merlin Gerin n184 /p.5
of an initial insulation fault (see. Cahiers Techniques n172 and n173).
dependability studies
Before looking at electrical network dependability studies, it would be useful to recap on several
definitions of the terms used by reliability specialists, even if everyone knows the definitions of the words: reliable, available, maintenance and safety (see fig. 4 and 5). It is also necessary to detail the applicational scope and general features of such studies.
Scope and features of studies c applicational scope: studies are carried out on all types of electrical networks, from low voltage to high voltage, and on their monitoring and control systems. The networks may be: v single feeder
dependability: Dependability is a generic idea that measures the quality of service, provided by a system, so that the user can have justified confidence in it. Justified confidence is achieved through quantitative and qualitative analyses of the various features of the service provided by the system. These features are based on the probability parameters defined below. reliability: the probability that a unit is able accomplish a required function, under given conditions, during a given interval of time [t1,t2] ; written R(t1, t2). availability: the probability that a unit is in a state that enables it to accomplish a required function under given conditions and at a given moment in time, t ; written D(t). maintainability: the probability that a given maintenance action can be carried out within a given time [t1, t2].
safety: the probability of avoiding an event with hazardous consequences within a given time [t1,t2]. failure rate: the probability that a unit loses its ability to accomplish a function during the interval [t, t+dt], knowing that it has not failed between [0, t] ; it is written (t). the equivalent failure rate: the value or a constant failure rate: eq for which the reliability of the system in time t is equal to R(t) - eq is a constant approximation of (t). MTTF (Mean Time to Failure): the average time of correct operation before the first failure. MTBF (Mean Time Between Failure): the average time between two failures in a repairable system. MUT (Mean Up Time): the average time of correct operation between two failures in a repairable system. MTTR (Mean Time To Repair): the average time required to repair.
MDT (Mean Down Time): the average time for which the system is not available. It includes the time to detect the fault, the time for the maintenance service team to get to the fault, the time to get the replacement parts for the equipment to be repaired and the time to repair. the repair rate: the inverse of the mean time to repair. FMEA (Failure Modes and Effects Analysis): this enables the effects to be analysed of the failure of components on the system. model: graphical representation of the combination of failures found during an FMEA and of their maintenance process. undesirable event: system failure that must be analysed in order for the user to feel justified confidence in the system. This system failure gauges the quality of service,
fig. 4: definitions.
initial failure 0 correct operation waiting
start of work repairing MTTR
restarting correct operation
second failure time
MTTF
MDT MTBF
MUT
fig. 5: relations between the various values that characterise the reliability, maintainability and availability of a machine.
turbo-alternator G G
a. double line feeder
b. double busbars
c. loop
fig. 6: network configurations.
preliminary study accuracy of assumptions accuracy of modelling 1 single failure type (1 frequency, 1 average duration) the consequences of the failures are combined into major families
detailed study failures divided into families according to their effect on the system. the consequences of the failures are analysed in depth.
simple and minimal configuration
fig. 7: differences between a preliminary study and an in-depth study.
v double feeder (see fig. 6a), v double busbar (see fig. 6b), v loop (see fig 6c), v and with or without reconfiguration or load shedding. c characteristic features of studies: studies are tailor made to take account of the expressed requirements. They are set up in terms of: - accuracy of study, - type of analysis, - type of dependability criteria, v the accuracy of study (see fig. 7): - brief or preliminary study:
dependability criteria
this is a pessimistic study generally used to quickly make technical choices - very detailed studies that take account of as many factors as possible. Taking account of all the operating modes, detailed analysis of possible failure modes and their consequences, modelling the malfunctioning behaviour of the system. v the types of analysis - design assistance in assessing the dependability criteria (see fig. 8),
satisfying of dependability criteria
no
new configuration
yes
fig. 8: design assistance.
- comparison of configurations (see fig. 9), - qualitative configuration analysis. v the types of criteria to be quantified (see fig. 10) - the average number of hours that the system functions correctly (MUT): - the average number of hours that the system functions before not supplying certain current consumers for the first time (MTTF) - the probability of no longer supplying power to certain current consumers (availability), - the average number of failures per hour during a year (eq), - an average repair time (1/eq), - the optimal frequency of preventive maintenance, - the calculation of replacement part kits. These criteria enable an assessment to be made of the system's performance level and thus to determine the configuration that meets dependability criteria without forgetting economic constraints.
configuration A
configuration B
dependability criteria
configuration A's dependability criteria are nearer the objectives configuration B
no
chose configuration B
yes chose configuration A
fig. 9: configuration comparison.
public network
generator set GE
we can calculate: - the probability that GE does not start when power supply is lost. - the optimum preventive maintenance frequency for the GE. - the number of minutes a year that current consumer n1 is not supplied. - the average number of hours before current consumer n2 will no longer have power.
UPS
SC*
current consumer n1
current consumer n2
* SC= Static contactor
fig. 10 : illustration of dependability assessment criteria.
2. how a study is carried out
the chronological order of phases

Whatever the requirement expressed by the designer or operator of an electrical installation, a dependability study will include the following phases (see fig. 11): c expression and analysis of the requirements, c functional analysis of the system, c failure modes analysis, c modelling, c calculation or assessment of dependability criteria. In most cases, this procedure must be performed several times: c twice if we want to compare two layouts, c n times if we want to determine a configuration that is adapted to requirements taking account of technical and economic constraints.
analysis of the customer requirement criteria to be network points where the assessed criteria wil be assessed system specifications
functional analysis of the system functions and components involved
failure mode analysis possible system failures
search for data - failure rate of components - time to repair - functional frequencies
data modelling key assessment of dependability criteria required qualitative analysis to perform the phase study phase phase conclusions data required to perform the next phase
expression and analysis of requirements

As discussed at the end of the previous chapter, the study initiator must detail (see fig. 12): c what the study involves ; e.g: substation monitoring and control, and provide the design elements that he has in his possession. c the type of request (type of analysis required), e.g.: v a demonstration of the confidence level that we can assign to a power supply for a critical process (orienting us towards a type of study, types of criteria to be assessed), v the search for objective criteria making technical and economic analysis possible, v the determination of the most suitable configuration to meet requirements (orienting us towards a type of analysis), v support for a specific equipment design. These points can be combined ; in other words a company can search for the configuration best suited to its needs to supply power to a critical process as part of a technical and economic analysis.
fig. 11: the chronological order of phases in performing a dependability study.
expression of requirement expressed in terms of
determination and classification of the representative and/or strategic current consumers in the system to target the network points at which will be calculated the dependability criteria
type of analysis
study criteria to be accuracy assessed
leading to the carrying out of a customised study
fig. 12: information required for the study.
the various current consumers in consideration
office heating
tank cooling stoppage possible for 1 hour max. beyond this: loss of tank contents disastrous event
oven n1 stoppage possible for 3 mins max. beyond this: loss of tank contents disastrous event
oven n2 stoppage possible for 3 mins max. beyond this: loss of tank contents critical event
minor event
computers no break greater than 20 ms beyond this: critical event
process
fig. 13: the availability specifications to be provided by the customer.
Faced with the questions: where, to supply what, what is the criticality and the admissible down time, the customer must think in terms of, for example, safety and loss of production or information. c the risk: for an insurance company the idea of risk corresponds to the seriousness (moral, social or economic) of the event in question. In terms of loss of production, the estimate of the risk is quite simple: the product of the probability of the occurrence and the cost of an event gives a good idea of criticity. The assessment of the risk enables the price to pay to be calculated: loss of profits, insurance or optimised electrical installation. c formalising of requirements (see fig. 13): a means of expressing the requirements involves classifying the various current consumers according to the down time that they can withstand (none, a few seconds, a few minutes, a few hours).
monitoring and control
power supply B stand-by
power supply A normal supply
GE stand-by
MV
MV electrical supply
fig. 14: functional block diagrams.
functional analysis of the system

Functional analysis describes in both graphical and written terms the role of the network and/or its constituent parts.
This analysis leads to two complementary descriptions of the network: c one description using functional block diagrams (see fig. 14), whose aim is to present the system configuration and the functional links between the various parts of the system.
c a behavioural description (see fig. 15) whose aim is to describe the sequence of the various possible states. The main purpose is to identify the events that induce the system to change. The model developed during the course of this analysis notably
highlights the various reconfiguration points in the system. This second analysis enables account to be taken of functional aspects when they interact with malfunctional aspects.
? loss of power supply A
failure mode analysis

This analysis aims to provide: c the list of possible failure modes for each of the items identified in the functional analysis, c their causes of occurrence (one cause is enough), c the consequences of these failures on the system (also called single events), c the failure rate associated with each failure mode as part of a quantitative study. The results are presented in table form (see fig. 16). This analysis can be considered as the first stage of modelling.
! switch to power supply B
? loss of power supply B ! start-up of GE
This Petri network expresses the fact that in the case of loss of power supply A, the network is switched to the power supply B. If power supply B fails, then the GE start-up.
fig. 15: behavioural description in the form of a Petri network.
functions
failure mode loss of normal mode
causes c power supply failure c transformer failure c spurious circuit breaker tripping c failure of GE in operation c spurious circuit breaker tripping c transformer failure
effects on the system switch over to stand-by
reliability data
As part of a quantitative analysis, it is necessary to have probability data for the failure of the system's equipment. The probability characteristics are then combined with the failure modes. These are: c the failure rate and how it is broken down according to the failure mode, c the average time to repair associated with the frequency of preventive maintenance (see fig. 17). Failure rates Remember that this involves quantifying the probability that there will be a failure between [t, t+dt], knowing that there has been no failure before t. Schneider Electric has a database built up from several sources: c internal studies and analysis of equipment returned to the factory after failure, c failure statistics observed by utilities companies and other manufacturers, c reliability data, v for non-electronic components: - IEEE 500 (feedback of experience from nuclear power stations in the USA),
power supply feeder
stand-by
loss of stand-by mode in operation
loss of electricity supply
normal mode failure and c GE does not start up stand-by mode is not c circuit breaker is available blocked open
fig. 16: failure mode analysis.
equipment
failure mode
failure rate
time to repair, frequency of preventive maintenance
circuit breaker generator set transformer
c blocked closed c spuriously open c failure in operation c failure on start-up c failure in operation
1 2 4 5 6
1, 2, 6 months 3, X months
fig. 17: fictive summary of the reliability data required for a study.
- NPRD91 (feedback of experience from military and non-military systems in the USA), v for electronic equipment, the data is generally obtained by calculation from the Military Handbook 217 (F) or from reliability data from the National Centre for Telecommunications Studies. Some components have a constant failure rate () over time throughout a period of their life. In other words the probability of failure is independent of time ; this is the case of electronic components (exponential law). Electrotechnical components age, or in other words their failure rate is not constant over time. Data most commonly available supposes constant rates. The use of data that is non-specific to the system being studied and of constant failure rates even if certain components age is, nonetheless, very attractive since it enables valid comparisons to be made between systems. The fact of finding a configuration that is 10 times more reliable than another, 10 times more available ... means that this configuration is ten times better for the criteria in question ; the relative value is often more important than the absolute value. Schneider Electric's reliability database has the aim of gathering together as much data as possible. Validity criteria have been set up to guarantee the quality of the data included. We ensure: c the way that feedback of data was carried out, on what data it was based, c the preliminary calculation method used, the conditions of use, temperature, environment, etc In time, this work provides us with reliability data to study any installation or to be able to obtain them by extrapolation. The mean times to repair depend mainly on the company's maintenance policy. Decisive choices notably include the possibility of inventory, the times when the repair
teams are present, the frequency of preventive maintenance, the type of maintenance contract with suppliers, etc. The impact of the variation of these maintenance policy-based parameters on the system's performance level can be dealt with by a specific analysis. Functional reconfigurations In the instance of functional reconfigurations, when functional aspects interact with those that are malfunctional (e.g. in the case of tariff based load shedding) the frequency of these reconfigurations is included in the modelling process.
modelling
The malfunctioning of the network is represented by a model. The model is a graphical representation of the
combinations of events determined by the analysis of failure modes which, for example, contribute to the loss of electrical supply for certain current consumers and to their repair procedures. This model enables: c discussions with the customer to validate our understanding of the way the network malfunctions, c qualitative analysis of the network performance levels, by the search for common modes, of the simplest combinations which contribute to the event in question, c assessment of the network's performance levels by calculating dependability criteria. Various techniques are available according to the configuration of the system being studied, the undesirable events in question, the criteria to be
peak event
loss of electrical supply
D: circuit breaker T: transformer PS: power supply
AND
intermediary events
GE failure
loss power supply
OR
OR
base event
GE
PS
fig. 18: fault tree modelling. The fact that the GE failure only makes sense if there is loss of EDF supply does not initially appear.
assessed and the assumptions taken account of in the model(s). The main modelling techniques c combination: the combining of single events. This is the case of fault tree diagrams (see fig. 18). A fault tree diagram breaks down events into single events. The immediate causes of the loss of electrical supply are therefore sought, these are intermediary events. The causes of these intermediary events are then sought in turn. The break-down continues in this way until it becomes impossible or until it serves no further purpose. Terminal events are called basic events. The breaking down of an event into causal events is performed using logical operators or so called gates (the AND gate, the OR gate). The fault tree diagram in figure 18 expresses the fact that in the given network, there is a total loss of electrical power if there is a loss of power supply and loss of the generator sets. In this type of modelling exercise, no account is taken of the fact that the failure of the generator sets is only significant if there is loss of power supply beforehand. Networks with reconfiguration possibilities and complex maintenance strategies are difficult to model using the fault tree method. c combination and sequential: the combining of single events whilst taking account of the moment in time when the events occur. v Markovian type (see fig. 19). This type of model is commonly represented by a graph that shows the various possible states in the system. The arrows that link the system states are quantified by rates, which can be the failure rates, the repair rates or the representative operating mode frequencies. These rates represent the probability that the system changes status
a: frequency at which there is loss power supply b: frequency of GE failure : frequency of repair
failure of power supply a
failure of GE's b
status 1 correct operation
status 2 power supply failure/start-up of GE's repair 3
status 4 loss of power supply
repair 1
repair 2 c loss of power supply, GE's do not start up
status 3 failure of power supply GE's do not start up
fig. 19: modelling of malfunctions in the form of a Markov diagram a, b and are assumed to be constant.
between time t and t+dt. The graph in figure 19 shows the fact that the system can have any one of four operating status: - status 1: correct operation, - status 2: failure of power supply and the generator sets have started up, - status 3: failure of power supply and the generator sets have not started up, - status 4: when the generator sets have failed in operation with the power supply also failed. Markovian modelling assumes that the frequencies (or rates), which enable passage from one status to another, are constant. The calculating algorithms associated
with this modelling technique can only be applied under these conditions. This limit leads us to make assumptions and therefore to represent the reality of the situation to a greater or lesser extent. v other types. The procedure used is generally that of a Petri network. The network is represented by places, transitions and tokens. The crossing of a transition via a token corresponds to a possible functional or malfunctional system event. These transitions can be associated with any type of probability law. Simulation is the only way to enable calculations to be resolved.
The Ptri network in figure 20 represents the various failure modes that the system may be subject to: - associated with transition n1 is the probability of power supply failure, - associated with transition n2 is the probability of the starting up and the non starting up of the generator sets, - associated with transition n3 is the probability of failure of the generator sets in operation. Criteria used to select a modelling type: figure 21 shows these modelling techniques in table form.
the measurement estimate is related to rare events. The modelling of a system using a Petri network is the model most closely
representing the actual operation of the system being studied. But in view of the limits related to simulation, this technique is not systematically used.
time to repair power supply failure time to repair 1 time to repair
dependability criteria assessment calculations

Two different techniques can be used. c analytical resolution: v for states graphs, v for fault tree diagrams. When systems are large or complex, analytical resolution may be impossible. c system component behaviour simulation (operation or failure): v for Petri networks, v for fault tree diagrams. To achieve accuracy, a large number of simulations must be carried out and the calculation time may be prohibitive if
1P=probability of start-up of GE's
P=probability of GE's not starting up
failure of GE's in operation
fig. 20: modelling using a Petri network.
selection criteria interaction of the operating mode in modelling numerical dispersion of data the timing of the failures is important estimate involving rare events estimate of: c MUT c MTTF c D(t) c D() c average D at t c MTTR c eq
fault tree diagram resolution impossible or false depending on the algorithms used resolution impossible or false depending on the algorithms used; no problem if simulation no simulation time which can be prohibitive, no problem in the case of analytical resolution suitable suitable (analytical calc.) suitable (simulation) -
Markov graph under certain conditions, resolution can be impossible or false depending on the algorithm used under certain conditions, resolution can be impossible or false depending on the algorithm used suitable suitable
Ptri network suitable
suitable
suitable simulation time which can be prohibitive
suitable suitable suitable suitable suitable suitable
suitable suitable suitable suitable suitable suitable
fig. 21: criteria used to select the type of modelling used.
3. examples of studies
comparing two electrical network configurations in a factory

c presentation of the factory: A drinking water production plant supplies 100 000 m3/day at off peak times, 300 jours/year and 200 000 m3/day at peak times, 65 days a year. Production of water is performed using 4 production plants, each capable of supplying 100 000 m3/day. Each plant is associated with six types of current consumers : R1, R2, R3, R4, R5 and R6 (pumps, disinfectors, etc.) which must work simultaneously in order to ensure production (ref. fig. 22). The current consumers ensuring the operation of each plant can be distinguished in the following manner: R1a,, R6a for plant n 1
R1b,, R6b for plant n 2 R1c,, R6c for plant n 3 R1d,, R6d for plant n 4 To ensure operation at off-peak times, only one plant is required ; during peak times, two plants are required. c analysis of the enquiry, the requirements After two meetings with the customer, the specialists have determined: v that the following was required: - a proposal for a new LV electrical network configuration, the configuration of the MV (5.5 kV) should change very little, - proof that the proposed layout was at least as good as the old in terms of certain dependability criteria. v that the dependability criteria to be assessed were: - the probability of simultaneously losing electrical supply to current consumers n1 and n6,
- the probability of losing electrical supply to current consumers n3. c functional analysis General considerations concerning the network operation: v the site is supplied by two independent power supply feeders. Two generator sets are there on standby in the case of failure of the power supply feeders, v in normal operation, the site is supplied by the power supply feeder A. If this feeder fails, then the system switches to power supply feeder B. If both power supply feeders fail, the generator sets start up, v there are two levels of production, offpeak and peak. In the case of peak production, the power of the generator sets is not sufficient to ensure production. v the power supply split between current consumers is such that the availability is
power supply A
G1
G2
power supply B
MV switchboard 1
MV switchboard 2
20 kV T1 5,5 kV
20 kV T2 5,5 kV
5,5 kV T3 400 V JDB3 R6a......R6d R5a......R5d
5,5 kV T6 400 V JDB6
fig. 22: functional diagram of the original installation. Certain penalising common modes are not shown.
not very good. Thus, for example a fault on the busbar JDB3 knocks out all current consumers of type R5 and R6. Production is stopped and no other plant can operate, v the existing electrical network was such that that there were common mode busbars. A short circuit across these busbars made any reconfigurations inoperative. The probability of a busbar short circuit is low, but since the system had high reconfiguring possibilities this failure mode becomes preponderant. The common modes occur at coupling level, during reconfigurations involving transformer T4. v for the new network, the current consumers have been separated so that it is possible to supply off-peak demand with the current consumers associated with one single transformer and peak demand with the current consumers associated with two transformers. Figure 23 presents the proposed network layout and figure 24 its functional analysis. c results analysis The improvement in results obtained with the proposed network is highlighted by giving the improvement ratios, with the existing network being used as a reference. Besides the probabilities of the simultaneous loss of current consumers 1 and 6 and the loss of current consumers 3, we have assessed their frequency of loss. Also calculated : the optimum maintenance frequency for the generator sets and the contribution of the MV and LV networks to the undesirable events. Here are the obtained improvements : v on the relative probability of simultaneously losing supply to current consumers n 1 and 6: - off-peak operation 110 - peak operation 55 - overall 105 v on the relative probability of losing supply to current consumers n 3: - off-peak operation 99 - peak operation 54 - overall 97 Overall, the probability of losing electrical supply to current consumers n1 and 6 is 100 times greater with the old network than with the proposed network. Indeed,
in the old network, there are failures that contribute directly to the loss of electrical power supply. If this system only worked in off-peak conditions, this ratio would be virtually the same since the system works mostly in off-peak conditions, thus explaining its preponderance in the overall result. If the system only works in peak conditions, it is around 50 times more probable to lose electrical supply to current consumers 1 and 6 with the old network. In the case of peak conditions, it is MV related faults that predominate and this part of the network cannot be changed very much. v on the frequency of loss of supply to current consumers n1 and 6: - off-peak operation 22 - peak operation 21 - overall 21 v on the frequency of loss of supply to current consumers n3: - off-peak operation 18 - peak operation 21 - overall 20 The improved performance of the new network is less pronounced in terms of frequency of failure. Calculation parameters used in a probability of unavailability are the frequency of failure and the time to repair. The failures which directly contribute to the loss of power have an effect on the unavailability time which is proportional to their time to repair. The new network makes it possible for background preventive maintenance to be carried, in other words without interrupting the plant's normal operation. c additional assessments It seemed interesting to assess some additional criteria to compare the two configurations. v the relative probability of losing peak operating conditions In the case of peak operating conditions, the economic stakes are high. The criteria assessed above do not measure this risk. It is entirely feasible to lose peak production even with current consumers 1, 6 and 3 supplied power. The probability of losing peak operation is four times less with the new network. The improvement is less pronounced than the other calculated criteria, since the main failures occur in the MV part which remains unmodified.
power supply A
GE1
GE2
power supply B
MV switchboard 1 20 kV T1 5,5 kV 5,5 kV 20 kV
MV switchboard 2 T2
5,5 kV 400 V plant n1 R1a R2a...................R6a
5,5 kV 400 V plant n2 R1b R2b...................R6b
5,5 kV 400 V plant n3 R1c R2c...................R6c
5,5 kV 400 V plant n4 R1d R2d...................R6d
fig. 23: functional diagram of the new configuration.
power supply A
generator sets
power supply B
transformer 20 kV/5.5 kV
transformer 20 kV/5.5 kV
current consumers n7
protection devices
coupling
protection devices
T5
T3
T6
T4
current consumers n6a, 5a, 4a 3a, 2a, 1a
current consumers n6b, 5b, 4b 3b, 2b, 1b
current consumers n6c, 5c, 4c 3c, 2c, 1c
current consumers n6d, 5d, 4d 3d, 2d, 1d
fig. 24: functional analysis of the new configuration.
v the optimum preventive maintenance frequency The graph resulting from the calculations (see fig. 25) shows the impact of preventive maintenance frequency of the generator sets on the probability that they will be available when they are required. For a maintenance frequency of 6 months, the graph shows that there is a trough in the probability of unavailability in the event of a loss of power. v contribution of the MV and LV parts to the undesirable events for the proposed network. The MV part has a much higher contribution than the LV part (around 99.9% compared with 0.1%). Since the MV network has not been modified, its preventive maintenance must be optimised ; moreover, it would be greatly preferable to have a quality monitoring and control system on the MV network.
P(failure when power down)
1 month 6 months 1 year 2 years 5 years fig. 25: unavailability of a generator set when power is down as a function of the periodicity of maintenance.
the interest of remote monitoring and control in an EHV substation

c analysis of the enquiry, the requirements Schneider is carrying out preliminary dependability studies as part of an export monitoring and control offer for an EHV substation. The following
preliminary study had to determine which configuration would enable the achievement of dependability objectives that had been set by the customer in terms of the loss of monitoring and control. The objective was to determine whether it was necessary to incorporate a stand-by remote work station.
work station
console level 2 communication
network
communication
4 PLC's*
communication
sorties TOR on/off outputs central unit centrale processor entres TOR on/off inputs central processor
on/off outputs on/off inputs
level 1
entres analogiques analogical inputs automate PLC* (*Programmable Logic Controller) PLC*
analogical inputs
fig. 26: base monitoring and control configuration of an EHV substation.

to level 3 work station link central processor console level 2
communication
communication
network
communication
4 PLC's*
communication
sorties TOR on/off outputs central unit centrale processor entres TOR on/off inputs central processor
on/off outputs on/off inputs
level 1
entres analogiques analogical inputs automate PLC* (*Programmable Logic Controller) PLC*
analogical inputs
fig. 27: monitoring and control configuration of an EHV substation with access and remote work station.
c functional analysis The monitoring and control of the EHV substation in question includes: v four plc's which acquire datas on the status of the substation's electrotechnical equipment and which pass on the control orders (level 1), v a work station enabling visualisation of the substation's status and the sending of control orders (level 2), v and possibly, a remote work station. The remote workstation is therefore on stand-by with respect to the substation's own work station. Figure 26 shows the basic solution corresponding to one single level 2 work station. Figure 27 shows the solution with a link to a remote work station. c results analysis the selected monitoring and control system must have a probability of unavailability of less than 10-4 at t=1200 hours. This means the equipment must be non-functional for less than 7 minutes over 1200 hours of operation (50 days). The undesirable event - loss of monitoring and control - has been broken down into three undesirable events corresponding to calculations of: v the probability of totally losing monitoring and control (common mode), v the probability of losing at least on/off information, v the probability of losing at least analogic information, or in other words three calculations for each configuration. The equipment is divided into two classes: v equipment for which we have replacement parts (n), v equipment for which we do not have replacement parts (s). Two calculations are performed for each undesirable event, to show the impact of the choice of the average repair time on the end result (see fig. 28). Hypothesis 2 is the most realistic one; it is these times that will be taken into account to choose between the two solutions.
MDT for equipment of type n hypothesis 1 hypothesis 2 1 hour 4 hours
MDT for equipment of type s 3 hours 12 hours
fig. 28: two hypotheses for the average time to repair.
objective base solution: analogic inputs
on/off inputs/outputs
common modes
solution with link to level 3: analogic inputs
hypothesis 1 on/off inputs/outputs hypothesis 2
common modes
1E-4
2E-4
3E-4
4E-4
5E-4
6E-4
unavailability at t = 1200 h
fig. 29: bar chart of the unavailability of the two solutions (the orange line corresponds to the objective to be achieved, it is achieved if the bar chart is to the left of the line).
As figure 29 shows; v the solution without the remote station does not enable the required availability to be achieved, v the solution with remote monitoring and control enables the objectives to be
achieved for on/off inputs and the common modes ; but the probability of losing at least one analogic input remains greater than the target objective.
4. dependability tools
dependability study packages

Certain tools automatically generate a dependability study from the functional analysis of the system and the failure modes of its components. A model is generated which enables evaluation of the dependability criteria. These tools are useful for complex systems and/or repetitive systems. They enable a database to be created on the failure modes of the components and on the consequences of these failures when possible. (see. fig. 30).
tool name Adlia
modelling technique fault tree
calculation resolution technique c analytical c simulation
main features tool developed by Schneider Electric to model the malfunctions in electrical network components. It integrates an electrical network malfunction database. The description of the network and of the undesirable event automatically generates the corresponding fault tree. tool developed by SGTE Sofrenten. Automatic generation of a fault tree and associated FMEA, by a functional description of the system and the creation of an associated malfunction database.
Sofia
fault tree
c analytical (logical polynomial)
modelling tools
Two types of tools (see. fig. 31): c simulation tools, c analytical calculation tools. Comment: a Markov graph can very easily be transformed into a Ptri network and be used for simulation. As opposed to a Ptri network, a graph can be associated, but in this case the transitions must be Markovised: in a Markov graph the frequency of occurrence of the transitions are necessarily constant. Schneider currently has a PC graphical interface called PCDM which automatically generates the file defining the Markov graph or the Ptri network that has been drawn (ref. fig. 32). This provides time savings and improved reliability of data entry. The Petri network modelling technique is that which currently best approaches the real behaviour of the system. The acceleration of the simulation of Petri networks, notably using MOCA-RP software, is the subject of a thesis whose initial findings have been presented at the ESREL 96 congress (European Safety Reliability). In the near future, the use of Petri networks will no longer be limited by the simulation time they require.
fig. 30 : two types of tools regularly used by Schneider Electri for dependability study.
modelling type Markov graph Petri network
simulation tool
analytical calculation tool super Cab developed by ELF AQUITAINE
MOCA-RD developed by Microcab
fig. 31: the modelling tools.
fig. 32: PCDM graphical interface - Markov graph.
5. conclusion
c of the monitoring and control system, c of the maintenance policy. They enable solutions to be chosen that are tailored to achieve the required availability threshold as cost effectively as possible. Often the study can be limited to a key point in the installation, responsible for the major part of the global unavailability. In many cases, it is interesting to call in a specialist. The advice they may provide could prove decisive.
Dependability can be broken down in terms of: c safety, c reliability, c availability, c maintainability. Safety requirements initially imposed dependability studies on hazardous applications : rail and air transport, nuclear power stations, etc. If we add the requirements of reliability, availability and maintainability, many other sectors are concerned.
Requirements have progressed in terms of quality of electricity supply. Taking account of the considerable improvements that have been made in methods and equipment, users are entitled today to demand a high level of availability. To achieve this objective with justified confidence, dependability studies are necessary. They enable optimisation of c the configuration of the electrical network,
bibliography
Standards c IEC 50: International electrotechnical vocabulary - general index. c IEC 271/UTE C 20310: List of basic terms, definitions and related mathematics for reliability. c IEC 300: Reliability and maintainability management. c IEC 305/UTE C20321 327: Characteristics of string insulator units of the cap and pin type. c IEC 362: Guide for the collection of reliability, availability and maintainability data from field performance of electronic items. c IEC 671: Periodic tests and monitoring of the protection system of nuclear reactors. c IEC 706/X 60310 and 60312: Guide on maintainability of equipment. c IEC 812/X 60510: Analysis techniques for system reliability Procedure for failure mode and effects analysis (FMEA). c CEI 863/X 60520 : Prvision des caractristiques de fiabilit, maintenabilit et disponibilit. c IEC 880: Software for computers in the safety systems of nuclear power stations. c IEC 987: Programmed digital computers important to safety for nuclear power stations. c IEC 1165: Application of Markov techniques. c IEC 1226: Nuclear power plants ; instrumentation and control systems important for safety ; classification. c NF C 71-011 : Sret de fonctionnement des logiciels Gnralit c NF C 71-012 : Sret de fonctionnement des logiciels Contraintes sur le logiciel. c NF C 71-013 : Sret de fonctionnement des logiciels Mthodes appropries aux analyses de scurit.
Miscellaneous publications [1] Fiabilit des systmes A. Pags et M. Gondran Eyrolles 1983 [2] Sret de fonctionnement des systmes industriels A. Villemeur Eyrolles 1988 [3] Recueils de donnes de fiabilit : c Military Handbook 217 F Department of Defense (US) c Recueil de donnes de fiabilit du CNET(Centre National dEtudes des Tlcommunications) 1993 c IEEE 493 et 500 (Institute of Electrical and Electronics Engineers) 1980 et 1984 c IEEE Guide to the collection and presentation of electronic sensing component, and mechanical equipment reliability data for nuclear-power generating stations c Document NPRD91 (Nonelectronics Parts Reliability Data) du Reliability Analysis Center (Department of Defense US) 1991 [4] Recommandations: c MIL-STD 882 B c MIL-STD 1623 Merlin Gerin Cahier Techniques c Mthode de dveloppement dun logiciel de sret Cahier Technique n 117 A. JOURDIL et R. GALERA 1982 c Introduction to dependability design, Cahier Technique n 144 P. BONNEFOI c Sret et distribution lectrique Cahier Technique n 148 - G. GATINE c High availability and LV switchboards, Cahier Technique n 156 O. BOUJU
c Automatic transfering of power supplies in HV and LV networks Cahier Technique n 161 G. THOMASSET c Energy-based discrimination for LV protective devices, Cahier Technique n CT167 R. MOREL, M. SERPINET c Dependability of MV and HV protective devices, Cahier Technique n 175 M. LEMAIRE Schneider's participation in various working groups c statistical group of IEC committee 56 (reliability standards), c software dependability with the European EWICS-T7 group : computer and critical applications, c groupe de travail de lAFCET sur la sret de fonctionnement des systmes informatiques, c participation la mise jour du recueil de fiabilit du CNET, c IFIP working group 10.4. Dependable computing.
n 186
Intelligent LV switchboards
Antoine Jammes Received his engineering degree from ENSEM (Ecole Nationale Suprieure d'Electricit et de Mcanique de Nancy) in 1979. Joined Schneider (Merlin Gerin) in 1980 and participated in the development of protection software in the Dependability Systems and Electronics Department (SES). In 1991, he moved to the Low Voltage Power Distribution SBU where he has played a major role in the development of intelligent LV switchboards.
E/CT 186 first issued, june 1997
Glossary
Application: Set of functions executed using hardware and software. Architecture: Type of organisation for hardware and/or software components in a system. Also the manner in which functions and processing are distributed. Bus: Link used to exchange digital information between the various connected elements. Communications network: Synonymous with communications bus. Connected device: Electronic device connected to the bus via a communications interface. Sometimes also referred to as a station. CSMA (Carrier Sense Multiple Access): Method providing random access to the communications network. CTM (Centralised Technical Management): System grouping all the applications in a given installation, for example process control, power management or building management. Decentralised processing: System in which a part of information processing is carried out as close as possible to the load. Dependability: Concept encompassing reliability (of control and monitoring systems), availability (of devices, machines), maintainability (of production resources) and the safety of life and property. Distributed processing: Particular type of decentralised-processing system in which processing is carried out in several processing units, each having a certain degree of autonomy and capable of communicating with the other units.
Gateway: A product enabling data exchange between two different communications networks, without local processing. Intelligent: Designates a system with its own processing power and a certain level of operating autonomy with respect to the computer system to which it is connected. LV: Low voltage. Master / slave: A master regularly polls its slaves and gives them orders. MCC (Motor Control Centre): LV switchboard grouping the control and monitoring devices for several motors, valves, etc. Power management: Form of technical management specifically intended for electrical distribution systems. The intelligent LV switchboard is a major element in a power-management system. Protocol: Sequence of rules that must be followed to establish and maintain data exchange between devices connected to a bus. Real time: Designates a control and monitoring system with response times compatible with the requirements of the given process. Stations: Information processing devices connected to the bus. Switchboard central unit: Unit which centralises all the information available in the switchboard and the direct environment, processes the information and communicates with a supervision system, thus making the LV switchboard intelligent.
Intelligent LV switchboards
In all buildings, regardless of the activity carried out inside, the distribution of electrical power must today satisfy ever-increasing needs for dependability and efficiency. Energy must be available not only to ensure the comfort and safety of users, but also to avoid the costs incurred by power failures. Electrical installations must therefore be monitored and be capable of reacting automatically to optimise power distribution. Information processing makes this possible. Already used in medium-voltage industrial and public-distribution applications, digital control and monitoring is now becoming a reality for low-voltage installations as well. Starting with an analysis of needs, this Cahier Technique takes a close look at how LV power distribution can be managed. Particular emphasis is placed on decentralising and distributing intelligence in and around the LV switchboard. Several examples of such installations are also provided.
Contents
1 Control and monitoring needs 1.1 Introduction 1.2 Needs 1.3 Functions 2.1 Currently used solutions 2.2 Advantages and disavantages of these solutions 3 Intelligent switchboards 3.1 Dfinitions - decentralised architecture and distributed intelligence 3.2 Decentralisation of functions in an electrical installation 3.3 Advantages of decentralised architecture and distributed processing 3.4 Conclusion on decentralised processing in a LV switchboard 3.5 A switchboard bus suited to electrical applications 4 Implementation examples 5 Conclusion and prospects for the future Bibliography 4.1 Computer centre 4.2 Hospital p. 4 p. 4 p. 7 p. 9 p. 10 p. 12 p. 13 p. 17 p. 18 p. 20 p. 23 p. 24 p. 27 p. 28
2 Current solutions
1 Control and monitoring needs
1.1 Introduction
For whatever type of application, whether in office buildings, banks, hospitals, supermarkets, airports, tunnels or industrial sites, the need to monitor and control electrical installations is increasingly prevalent, to ensure the following: c safety, c availability of power, c optimisation of energy consumption and costs (depending on the energy supplier's tariff schedules), c reduction in operating and maintenance costs; c ease of operation, c maintainability and upgradeability of the electrical installation. Power management can today be implemented by a Digital Control System (DCS) designed to meet all the above needs. Power management may be combined with the management of other facilities: c building management (access control, airconditioning and heating, anti-intrusion systems, lighting ), c digital control and monitoring of industrial processes. Due to the wide range of needs and significant technological progress over the last few years, a number of solutions are today available when designing systems to monitor and control electrical installations. It is now possible to arrive at a judicious balance between needs and the corresponding solutions through the use of digital communications buses and the integration of microprocessors in electrical equipment. Cahier Technique n 156 explains how to design the power section of an electrical switchboard so that it satisfies needs concerning dependability. The goal of this document is to discuss the optimised design of power-management systems in LV electrical installations. The first step is to review the needs expressed by users and operators.
1.2 Needs
The needs of users and operators of electrical installations are different, depending on whether the building is intended for commercial, industrial or infrastructural purposes. A hierarchy of needs may be established (see fig. 1 ). For example, in a small office building, the cost of energy and ease of use of systems by nonspecialists are the foremost criteria. On the other hand, in a hospital or a factory implementing an industrial process, the most important need is continuity of service. Safety of life and property An electrical installation must distribute electrical power while ensuring the safety of life and property. A power-management system does not replace the primary protective functions (reflex-type devices).
Examples:
Comfort
Source management
Costs
Tariff schedule management
Availability
Automatic source changeover
Safety
Protection of persons
fig. 1: hierarchy of needs in LV electrical.
Insulation resistance
Periodic drop in insulation, steadily worsening.
Intervention
First fault alarm threshold
Time
fig. 2: insulation monitoring for an outgoing circuit (IT systems).
Given its capacity to communicate as well as store and process data, it simply assists the operator by indicating the settings of protective devices, the type of fault that caused a device to trip and the status of the installation prior to the incident, etc. Power management can, however, include overall protective functions. For example, on IT systems, insulation-monitoring may be implemented to warn the operator of a first fault. It is then possible to identify and clear the fault without any break in the continuity of service (see fig. 2 ). Technological advances have made it possible for operators to reduce the duration of a fault in an installation, thus reducing the probability of a second fault occurring. Operators can check at any time the insulation measurements at different points in the installation and even the evolution of the insulation measurements over time. Preventive maintenance therefore becomes a real possibility. The insulation monitoring function is autonomous and may be considered a decentralised function in the framework of a power-management system. Availability Each field of activity has its own requirements concerning continuity of service: c in hospitals, operating rooms and reanimation centres are designed to provide a high level of dependability, c in commercial buildings, the widespread use of computer systems has led many people to use uninterruptible power supplies (UPS) installed either locally for individual machines or more centrally for the supply of entire installations with high-quality power,
c in industry, power failures result in production losses. For example, a ten-minute power outage in a Danone factory results in a production loss of 20 000 cups of yoghurt. The need to ensure the availability of power has led to a number of technological choices for equipment (withdrawable or disconnectable devices or switchboard units, switchboard forms, etc.) and to the distinction in electrical installations between uninterruptible, high priority and low priority circuits, with different choices for the system earthing arrangement. In this context, the job of a LV electrical switchboard is to manage the sources. To be effective, action taken when a problem occurs must be automatic and immediate. Managing power failures is one function of power-management systems. Energy costs A constant concern for all companies is the need to reduce the cost of energy. Reductions may be achieved by working on two different factors, the level of consumption and the pricing system of the energy supplier. To that end, in-depth knowledge is required on daily and seasonal fluctuations, as well as on power and consumption levels. A measuring system providing digital data for use on a supervision screen is required to monitor and analyse the above elements. It is then possible to: v undertake action to improve the situation, v check the effects of the action taken, v determine energy costs per workshop, department, etc.
c Reduce consumption There are numerous possibilities, depending on the type of application: v turn off lighting and reduce heating in unoccupied rooms, v use motors equipped with variable-speed drives for industrial applications, v use conditioners and/or filters to reduce losses due to harmonic and capacitive currents in cables and transformers. c Reduce costs related to the terms of the contract with the energy supplier v use capacitor banks to avoid being billed for reactive power, v smoothing peaks in consumption to reduce the subscribed power and avoid overrun penalties. An intelligent system is capable of making optimum use of the subscribed power by shedding certain loads, smoothing peaks and alternating the supply of power to high-inertia loads. v Select the best available contract and program production cycles requiring particularly high quantities of electrical power for periods when the cost of power is low. These periods may be a part of the day, the season or the year. For example, certain contracts offer attractive prices if the subscriber accepts to reduce his consumption on a certain number of peak days per year. v Use replacement sources. This solution makes its possible not only to have a backup source of power in the event of a failure, but also to smooth peaks in consumption and to avoid moments when the power costs are highest. Managing consumption and energy costs is another function of power-management systems. Operating ease Certain installations are managed remotely, either from a control and monitoring station inside the building or from a centre covering several sites (remote supervision).
Centralisation of management functions is a means to optimise human resources and improve the working conditions for personnel through the use of ergonomic computerised systems and automatic execution of repetitive tasks (programmed operating times for air-conditioning or heating of offices, etc.). Another consideration is the fact that in office buildings and on small industrial sites, the personnel in charge of the installation is increasingly a non-specialist. The electrical switchboard is commonly under the responsibility of the building watchman or a receptionist. To ensure effective operation as well as for safety reasons, the information presented to these persons must be in the form of a man/switchboard interface that is as ergonomic and simple as possible. Operating ease is achieved by an electrical installation that is as autonomous as possible (self managed). Maintainability The primary mission of the electrical maintenance department in a company is to keep the electrical installation up and running. There are two types of servicing: c corrective action following an operating fault; c periodic preventive action. Maintenance may be enhanced in two ways: c by stressing preventive rather than corrective action to avoid breaks in the continuity of service, c for preventive level, by stressing conditional maintenance, i.e. action taking into account monitored data, rather than systematic maintenance. The more maintenance is preventive and based on monitored information, the higher the availability of the installation (see fig. 3 ). Depending on the type of application, the time required to begin servicing and the duration may be very different. They may be very short for industrial processes if there is on-site
Availability Conditional preventive Systematic preventive
Corrective
Type of maintenance
fig. 3: operational availability as a function of the type of maintenance.
maintenance personnel and a stock of spare parts. On the other hand, they may be much longer in office buildings if an outside company must be called in and the spare parts ordered. The time required to service an installation always depends on the information available: c when troubleshooting, precise and rapidly available information on the problem and data on installation operating parameters prior to the fault are critical to making the right analysis and preparing the subsequent work (new parts), c when undertaking preventive maintenance, indepth information on the installation status makes it possible to intelligently select those elements most requiring servicing. To carry out effective maintenance, personnel must have relevant information concerning the installation status. Providing information for maintenance is one of the advantages of power-management systems.
Upgradeability The points in an electrical installation that are most subject to change are those closest to the final loads. In a factory, the electrical switchboard may be upgraded to keep pace with changes in production facilities. In office buildings, changes in how rooms are used, increasing use of microcomputers, installation of air-conditioning, etc., all result in modifications to the electrical installation. Improvements in availability and reductions in the cost of the power consumed are also reasons for modifying electrical installations. To correctly manage these changes, in-depth knowledge of the installation and operating parameters is required. Power-management systems contribute to easy and effective installation upgrades.
1.3 Functions
Satisfaction of all the above needs by a powermanagement system requires that a number of devices be installed in the electrical system, generally speaking in order to: c carry out a number of automatic actions, c provide the operator, either locally or remotely, with the information needed to plan ahead and carry out the required work on the installation. These devices provide a number of functions, not all of which are required in a given installation. Automatic-control functions c Source management. Loads are supplied depending on the availability of power on the different incomers (source changeover systems, normal and replacement sources, enginegenerator sets, etc.). c Load shedding. Only priority loads are supplied with power when demand exceeds the available level of power on the incomers (for example, when power is supplied by an engine-generator set). c Time management. To reduce consumption. c Tariff schedule management. Installation operation is organised to respect the terms of the contract signed with the power distributor (smoothing of peak power levels, special tariffs, etc.). c Protection of the electrical distribution system. In large industrial installations, system disturbances (transient voltage drops) may, due to the presence of large motors, provoke transient instability phenomena. This function ensures the necessary load shedding to avoid collapse of the entire electrical distribution system. c Power-factor correction. This function manages the switching of capacitor banks. c Switchboard safety (over-temperature, internal arcing, etc.). c Insulation monitoring and fault locating for IT systems. Information to plan and take action The purpose of the functions presented above is to make the main LV switchboard autonomous. It is then capable of reacting to various situations to ensure continuity of service and optimal operating conditions. The second major type of function in an intelligent switchboard is the capacity to communicate information for planning and taking action. Information includes: c the status of breaking devices (open or closed), c measurements (U, I, P, cos ), c the settings of protective devices. These functions require links to: c a power-management system at a higher level, in charge of managing the entire LV or MV installation, c a local or remote control and monitoring station, c where applicable, secondary switchboards, c where applicable, a process-control system. Before the operator can be informed and take action (manually reconfigure the distribution system, maintenance, comfort), the electrical switchboard must first communicate with a higher-level system that can be consulted by the electrician and the person in charge of monitoring operation of facilities in the building or factory. During normal operation, an intelligent (i.e. communicating) LV switchboard is useful in piloting and managing the electrical installation, but it is all the more so when planning and action are required in a fault situation.
This is because corrective maintenance is more effective if each of the persons involved is rapidly provided with the relevant information. Below is an example of a circuit breaker, communicating via the intelligent LV switchboard, in the event of a fault: c on the circuit breaker, the information is provided by an mechanical indicator. c near the circuit breaker, a red light identifies the device that has tripped. c on a screen at the head of the switchboard, a message in clear text is displayed 10:32:23 outgoer to Lift 2 - section B - position 12b tripping due to short-circuit. c on the supervisor screen of the electrical manager, the same message.
c on the main supervisor screen (e.g. in the security room), a message in clear text is displayed 10:32:23 - Lift 2 out of order due to an electrical fault - call the electrical department on line 347. Note that planning and taking action also relate to preventive maintenance if the following are available: c information on the protection and control switchgear in the LV switchboards. This information may be provided by a counter for the number of times a device has opened or closed, a maintenance indicator derived from data such as the sum total of currents interrupted, etc c information on the electrical installation, for example, the number of hours the supplied loads operate, drops in insulation
2 Current solutions
2.1 Currently used solutions

The functions presented in chapter 1 are already available, either in whole or in part, using a number of different technical solutions: c in industry, by adding electrical management to the facilities already implemented for industrial-process control, c in commercial buildings, by including electricaldistribution management systems in the existing building-management systems. Consider the solutions implemented for a motor control centre (MCC) or a main low-voltage switchboard (MLVS). Solution implementing PLCs c PLCs and wired connections The first step toward an intelligent switchboard involved the use of industrial PLCs (Programmable Logic Controllers) near the switchboard. The PLCs serve as interfaces between the switchboard and the technical management system and are capable of carrying out certain automatic-control functions. Made up of racks filled with input/output boards, the PLCs are wired to the various sensors and actuators of an electrical switchboard. A specialist is required to program the PLCs and each application is the result of a specific development. This type of solution is subject to the following limits and constraints: v great quantities of control wires between the switchboard and the PLC, with the following disadvantages: - very high wiring costs; - a large number of terminal blocks which increase the volume and notably the footprint of the switchboard; - high risk of latent defects due to the many connection points; - risk of malfunction due to the very strong magnetic field created by a short-circuit on an outgoer; v significantly reduced capacity for installation upgrading, due to the very specific nature of the PLC programming which can rarely be modified by the in-house electrical department, v a data-processing system poorly suited to the given applications in that the main task of a PLC is to continuously poll the status of devices which in this case often remain in the same position year round. c PLCs and remote input/outputs In the past few years, PLC manufacturers have taken advantage of dropping costs in microelectronics and communications buses and developed remote input/output modules, thus making it possible to reduce the quantity and cost of wiring. This solution has been put to very little use in the field of electrical switchboards because it is poorly suited to the constraints inherent in the field, notably the thermal environment, electromagnetic disturbances, the need to control switchgear locally, etc. Solution implementing automated switchboards In the 1980's, a number of offerings were developed by the major panel builders for applications in continuous-process industries or in large commercial buildings. These offerings differ from the solutions presented above in two aspects: c the development of specialised modules wired to the switchgear components and communicating via a parallel link or a serial bus with a PLC installed at the head of the switchboard. These modules, designed for use exclusively in switchboards produced by specific panel builders, are installed on the front panel of the switchboard and include built-in local control and status-indication functions. c the development of repetitive functions for the electrical automatic-control systems. For example, source changeover systems with load shedding and reconnection of outgoers. These systems are characterised by decentralised data processing in the switchboard and the fact that the functions can be handled by electricians. What is more, they contribute to the massive reduction in the quantity of wiring inside the switchboard. The limited success of this type of system is due to the fact that these modules were specific to the different panel builders. Communicating components Microprocessors are now used by manufacturers of electrical equipment to: c improve the performance of their products. A good example is the widespread use of electronic trip units in circuit breakers. The latter are increasingly capable of communicating the data they process via digital buses. c enhance their offering with new functions, for example, power and energy measurements at a
given point in an installation, with the capacity to communicate the data. In parallel, automatic-control functions such as source changeovers or reactive power regulation continue with their own developments. The increasing use of such products and modules in electrical switchboards has resulted in a considerable increase in the quantity of information that can be transmitted to a
centralised supervision system. In large-scale applications, the engineering firm in charge of the entire installation is still obliged to implement a complex communications architecture with intermediate levels fulfilling a dual mission: c sorting and analysing the available information, c providing communications gateways between different buses.
2.2 Advantages and disadvantages of these solutions

The three types of solution presented above were developed over the last ten years to satisfy some of the needs listed in chapter 1. Their advantages and disadvantages are summed up in figure 4. In conclusion, the following may be observed concerning the currently implemented solutions: c a tendency, well underway, toward decentralisation of automatic-control functions and data processing for electrical switchboards, c a hierarchical structure for data flow, c the need for specific development work and, consequently, for specialists. In the electrical-switchboard field, decentralisation of data processing has been made possible by digital communications buses. This is the case for: c industrial-process control and monitoring. PLCs with hundreds of input/outputs first gave way to PLCs with remote input/outputs and are now gradually being replaced by networks of PLCs and micro-PLCs positioned as close as possible to the controlled sensors or actuators. The near future will see networks comprising intelligent sensors and actuators, c building management. Functions have now been standardised and are carried out by specific products well suited to needs. Solutions providing industrial-process management, building management and electrical-distribution management are now dedicated systems built around decentralised architectures that increasingly incorporate distributed intelligence (see fig. 5 ). A notable aspect of low-voltage electrical switchboards is their great diversity and the wide range of functions that they must provide. On the basis of existing solutions described in this chapter and that have already been put to use, it is today possible to determine a number of principles that define and specify what is understood by the term intelligent switchboard and the corresponding control and monitoring system. c An intelligent switchboard is defined by its capacity to autonomously carry out the functions assigned to it and to fit into the control system of an electrical installation. c To handle the wide variety of installations, the design of the switchboard must be based on the following principles:
PLC-based solutions Weak points
Automated switchboards
Communicating components Single-function solutions Increase in the number of modules Requires electricians specialised in communications
Specific to each application, Solution available only little capacity for upgrading to large panel builders Large quantity of wiring Requires PLC specialists Requires PLC specialists
Strong points
Reliable equipment, used on large industrial sites Solutions perfectly tailored to the initial needs of each customer
Implementation of the decentralisation concept
Reliable field-bus technology
Functions may be mastered Industrial products by an electrician capable of withstanding EMC constraints in switchboards Functions standardised with progression of projects
fig. 4: advantages and disadvantages of traditional solutions in meeting control and monitoring needs.
v design must be modular in terms of both hardware and software, v decentralisation of processing must be optimised, v functions and products must be standardised, designed for installation and implementation by an electrician and capable of handling the severe environmental conditions prevalent in an electrical switchboard.
c For users confronted with automation needs and the necessity of obtaining relevant information, an intelligent switchboard provides: v an enhanced level of dependability and upgradeability, v standardised solutions, even for large sites, v cost reductions and simplified implementation, v easier operation and maintenance.
Company management
Building-management station
Electrical-distribution management station
Industrial-process management station
Intrusion
MV distribution system
Workshop
Fire protection Personnel presence
LV switchboard
MCC
Enginegenerator sets UPSs
Machines
Air-conditioning
Variablespeed drive, starter PLCs
Processing units
Processing units
fig. 5: decentralised architecture for building management, electrical-distribution management and industrial-process management systems.
3 Intelligent switchboards
Intelligent switchboards, as defined in the preceding chapter, are based on the concept of decentralised architecture with distributed intelligence. After defining these terms, we will go into how the various functions of an electrical installation may be decentralised and distributed in the best
possible manner, whether for an entire installation (power management), for a low-voltage switchboard or for a given outgoing circuit. Then the criteria determining the selection of an internal communications bus for the LV switchboard, suited to the given needs, will be examined.
3.1 Definitions - decentralised architecture and distributed intelligence

c An analogy based on how companies are organised may be useful in understanding these two terms. In centralised organisations, all decisions are made by the boss. Subordinates provide him with all information and wait for orders. In an effectively decentralised organisation, a majority of decisions are delegated by the boss to the subordinates. Each person, within the limits of the delegated powers, acts autonomously and reports only the necessary information back to the boss. Only those functions concerning the entire company are centralised, for example, the payroll. Finally, a given function may be distributed between a number of subordinates. This form of organisation implies information exchange and a certain degree of autonomy for the team in charge of the function. c Figure 6 shows how a function may be: v totally decentralised, v partially decentralised, whereby execution of the function is decentralised, but parameter settings remain centralised and common to a number of functions, v distributed among equipment on the same hierarchical level.
Solution 1. Centralised system (acquisition and processing).
F1 F3 F4
F2
Solution 2. Centralised processing, decentralised acquisition.
F1 F3 F4
F2
F5
F5
Solution 3. Function F3 and F4 decentralised.
F1
F2
Solution 4. Distributed processing of functions F2 and F5.
F1
F5
F2
F3 F4 F3 F4 F5
fig. 6: possibilities ranging from a fully centralised to a decentralised system with distributed intelligence.
c Decentralisation as implemented in a company may be applied in a similar manner to the control and monitoring system of an electrical installation. The Centralised TechnicalManagement (CTM) concept is now giving way to the decentralised power-management concept with distributed processing.
Note that high-power distribution systems (architecture and protection) follows the same principles, thus ensuring coherence between high- and low-current systems (see fig. 7 ). Below are examples of these concepts applied to different electrical functions.
Power-supply substation
Supervision station
Operator can monitor, issue orders and set parameters
LV switchboard VAR MLVS central unit Switchboard control and monitoring
Configuration console
Autonomous functions for measurements and control and monitoring of outgoers
VAR
fig. 7: small to mid-sized installation with a control and monitoring system (power management), of which the major part is located in the MLVS.
3.2 Decentralisation of functions in an electrical installation

Energy contract management This type of function requires an overall view of the installation. In many cases (small and mid-sized installations), the LV switchboard is the central element in the installation. If this is the case, the contractmanagement function is handled by the LV switchboard central unit, with either local or remote (from a supervision station) parameter settings. On a large site (medium-voltage distribution system), switchboards receive operating orders from a higher-level system. Time management of outgoers In centralised systems, this function is traditionally assigned to the supervision station which can be
used to set operating-time parameters for outgoers and to issue opening and closing orders for devices. On the other hand, in a decentralised powermanagement system, these commands are executed at the level of the switchboard central unit or even at each device. A device must simply receive the operating set-points and be equipped with an internal clock that is regularly synchronised by the supervisor.
In figure 8, the information flow is shown for a traditional centralised solution and a decentralised solution. It is clear that the permanent information flow is reduced as decentralisation is increased. On the other hand, new data exchanges, much more limited in scope, are required to periodically synchronise the various internal clocks and transmit new operating set-points.
Centralised system Set operating-time parameters Store parameters Issue orders
Power-management system Operating-time parameters set by operator
Transmit orders and acknowledgements
Execute orders Q1 Q2 MLVS 1 Non-intelligent outgoers Outgoer 1 Outgoer 2
Transmit parameters
Store parameters Issue and execute orders
Q3
Q4 MLVS 2
Outgoer 3
Outgoer 4
Intelligent outgoers (equipped with internal clock and memory), capable of handling part of the function
fig. 8: time management of outgoers, in a centralised system (CTM) and in a decentralised and distributed system (power management).
Source management This function opens or closes the incoming circuit breakers in the switchboard, depending on the data processed either in the switchboard or in the immediate proximity. It is therefore perfectly logical that the operations required for this function be carried out in the electrical switchboard. Note that in relay-based systems, the relay sub-assemblies were installed in the switchboard and the diagrams were drawn up by the panel builders. It was only when a new technique arrived, one that most panel builders could not handle, that this processing was remoted to a centralised PLC. If the incoming diagram is simple, for example with a normal and replacement source, this function is totally decentralised and is carried out by an autonomous standard product. If the incoming diagram is more complex or requires programmable shedding of outgoers, the function is located at the level of the switchboard central unit: c if the replacement source supplies the main LV switchboard alone, a switchboard central unit will carry out the function autonomously (see fig. 9 ), c on the other hand, if the replacement source supplies the MV system and/or several MLVSs,
this function is distributed between the MV switchboard central unit and the central units of the various LV switchboards. Reactive power regulation Power-factor correction using capacitor banks is an independent automatic-control function built into a product called a reactive power regulator. This type of regulator must operate autonomously in over 90% of all installations. A communicating reactive power regulator can be built into a power-management system to provide the following additional useful functions: c setting of parameters from a supervision station, c action on alarms processed by the switchboard central unit, c action on maintenance information in the framework of overall switchboard maintenance, c coordination of the reactive power regulation function with other switchboard functions. For example, during operation on an enginegenerator set, the capacitors must be disconnected. This can be carried out by opening a circuit breaker upstream from the capacitor banks or by transmitting a shutdown order to the regulator if it is connected by bus to
GE
Generator-start order
Presence Un Presence Ur
main LV switchboard central unit
fig. 9: example of source management. With this solution, the switchboard central unit manages the outgoers. Priority outgoers are progressively reconnected during operation on an engine-generator set. Definition of outgoers as priority or non-priority is adjustable. Note that it is not necessary to separate the busbars into two parts, thus eliminating the coupling device. Finally, this solution makes it possible to handle multiple-incomer diagrams with great ease.
the switchboard central unit managing or monitoring source changeovers. Threshold-initiated load shedding In certain situations (voltage drop due to a problem on the distribution system, failure of a source, demand exceeding the available power from a source supplying the switchboard, etc.), it may be necessary to rapidly shed a group of non-priority outgoers, for example to avoid transient-stability problems. Figure 10 shows how shedding of non-priority outgoers is processed in a decentralised manner, following an overload on the main LV switchboard. This example shows that the amount of information exchanged is very small. The central unit receives a signal, issues an order via the bus and the concerned circuit breakers carry out the order.
Management of an incomer or an outgoer Management of an incomer (or an outgoer) may include some or all of the following functions: c control and monitoring (control of the device and monitoring of its status), c measurements (currents, power levels, energy drawn, etc. ), c local or remote operator interface, c communication with the switchboard central unit. By distributing these functions among different modules (see fig. 11 ), it is possible to solve certain problems: c not all the outgoers require all the functions listed above, c the operator interface can be remoted, c the interface must be adaptable to the various users (language, level of competence, etc.).
Parameter setting of threshold and priority (P) non-priority (NP) status for each outgoer
Switchboard central unit
Store threshold parameter Decide: if measured power > threshold, send load-shedding order
Measure power
Q1
Q2
Q3
Store P/NP parameter Decide: if parameter = NP, open outgoer on receiving load-shedding order Control the circuit breakers
fig. 10: example of decentralised processing of a load-shedding order.
Control & Monitoring M1 Local display M3 Measurement M2
I1 = 125 A
Outgoer 3 closed E = 327 kVAh M4
Module M1 controls the device and reads its status conditions. Module M2 measures the currents and voltages and prepares the power and energy information. Module M3 displays the status and measurement information for the outgoer on the front panel of the switchboard. It can also control the outgoer. Module M4, identical to module M3, is a display remoted to a point outside the electrical room. The information may be displayed in a manner different than on M3
fig. 11: example of distributed processing for outgoer management.
3.3 Advantages of decentralised architecture and distributed processing

Mastering complexity A complex problem can often be broken down into a set of simple basic problems. Similarly, controlling an electrical installation may prove very complicated given the size of the installation and the number of functions that must be processed. By decentralising a majority of the functions, most of the processing can be carried out by smaller units. The processing is then easier to handle and can be standardised. The concept of type-tested assemblies, already used for the power part of electrical switchboards, can now be expanded to include power-management functions. The load on the higher-level processing unit is significantly lightened and it can devote its processing power to the tasks for which it is specifically intended. Technical and economical constraints As already mentioned in the preceding chapter, the considerable increase in the quantity of information to be transmitted has led to the development of hierarchical architectures. Just as in large installations where there are levels in the power-distribution structure (main and secondary LV switchboards, final distribution enclosures, etc.), the creation of levels for the processing of information is the best solution: c constraints (response times, environment, throughputs, etc.) are not the same inside a switchboard and throughout an entire installation, c all the information that is useful for a function at a given level is not necessarily relevant on a
higher level, for example, all the information available locally on each device is not necessarily of great use to the operator: v some information is intended strictly for maintenance purposes, v other information must be summarised to avoid submerging the operator (see fig. 12 ), v the cost of programming is reduced by using standard codes for a vast majority of the functions. Continuity of service In a centralised system, a breakdown results in an interruption of service for the entire installation. In a decentralised system, however, the same breakdown can be limited to the single
subsystem where it occurred, thus enabling the rest of the installation to continue operating, though perhaps in a downgraded operating mode. For example, if maintenance is required on the switchboard central unit, local functions inside the switchboard remain operational, given the decentralised nature of the installation. Maintainability A decentralised system implements a large number of processing units, however, their failure rate is not cumulative. The limited number of connection points reduces the number of breakdowns. The self-test system on the digital products and the communications buses can detect nearly 100% of possible breakdowns. Flexible implementation
Information
User Maintenance Supervision room station Device position X X Faulty outgoer X X Energy measurement X Outgoer not available (summary) X v Disconnected/locked out X v Not supplied X Trip unit setting X Load shedding in progress X X
fig. 12: example of sorting information to be made available to different users.
c Setting up a new site often takes place over relatively long periods. It is not uncommon that for budgetary reasons, the remote supervision station is installed one or two years after commissioning of the switchboards. The latter can, nonetheless, operate autonomously over such long periods if decentralised processing is carried out locally. c When existing installations must be renovated, upgrading can be spread out over several years. Decentralisation makes the replacement of a switchboard simpler. The new switchboard can be factory tested and a single serial link is all that is required to connect the new switchboard to the control system.
3.4 Conclusion on decentralised processing in a LV switchboard

The examples presented in chapter 3 show that the functions managed by an intelligent LV switchboard can be distributed to varying degrees among different processing units. c Certain functions are handled by the switchboard central unit when: v the processing is complex and cannot be carried out by a standard autonomous module. For example, source management when there are multiple incomers, v the functions call on processing that is common to other functions. For example, a source changeover can be caused by the failure of the normal source or by an order issued by a contract-management function, v the functions must be coordinated with other equipment. For example, when management of replacement sources brings MV equipment into play. c Certain autonomous functions can be carried out by dedicated products that have been optimised for the given function. This is the case for reactive power regulators and source changeover units. When these autonomous products are incorporated in an intelligent switchboard, they can be connected via a bus to the switchboard central unit which provides additional functions such as: v setting of parameters for the products by a more user-friendly device that is common to all the functions carried out in the switchboard, v minimum management during downgraded operating modes, v incorporation of the products in predictive- and corrective-maintenance functions. The various functions mentioned in this document may be included in the architecture presented in figure 13: c The switchboard central unit is in charge of: v processing the general switchboard functions and the interdependent functions, v coordinating the functions managed by lowerlevel modules, v integrating in a higher-level control system, v communicating with a terminal intended for the electrician in charge of implementation and maintenance operations. During servicing, this terminal is connected locally in front of the
switchboard. In no way does it replace the centralised supervision station from which the operator manages the installation. c Certain modules are in charge of processing autonomous functions (reactive power regulation, insulation monitoring, etc.). c Other modules are in charge of managing an incomer or an outgoer. Modularity makes it possible to: v integrate control and monitoring in the concept of tested, standardised functional units, v standardise connections between the module and the switchgear, thus reducing the risks of breakdowns due to faulty connections, v take action on a given outgoer without shutting down other elements in the switchboard, in the event of a breakdown or installation upgrading.
The modules and the switchboard central unit are connected via a digital bus. Note that use of this type of bus offers a wide range of advantages: c massive reductions in the quantity of control wires in the switchboard and consequently in the cost of wiring and the space required, c reduced risks of breakdowns due to faulty connections c less design and wiring time for the panel builder, c greater installation upgradeability, for example, the addition of outgoers or functions in an existing installation. The next chapter describes the types of buses best suited to power management applications.
Installation supervision station
Configuration
Central functions
Source management with load shedding / reconnection
Autonomous functions
Generator management Reactive power regulation UPSs Insulation monitoring
Local functions
Outgoer control and monitoring and measurements
Secondary switchboards
fig. 13 : distribution of functions in the architecture of a switchboard.
3.5 A switchboard bus suited to electrical applications

Before selecting a suitable field bus, it is essential to fully comprehend the constraints weighing on an electrical application, notably the information flow, response times, the environment, etc. Characteristics of an electrical application c A naturally stable and continuous application The purpose of an electrical installation is to distribute power to each load. The purpose of an LV switchboard is therefore to permanently maintain the operational status of outgoers. Opening of a device may be related to one of the following events: v reaction of a protective device to an electrical fault, v operator intervention to isolate a circuit (for servicing, to turn off the lights on a floor at the end of the day, etc.), v intervention of an automatic-control function to shed non-priority loads, for example, following the failure of the normal source. The change in status of a device is therefore an exceptional event. An electrical switchboard is continuously in a naturally stable state. Note that circuit breakers are by nature bistable devices. c Certain situations cause an information avalanche On the other hand, certain situations may result in an avalanche of information over very short time periods. For example, following the failure of the source supplying the switchboard, monostable devices such as contactors simultaneously open and the automatic source changeover and load-shedding functions issue orders to the circuit breakers. c Limited real-time constraints In an electrical installation, the response time of the system to an event depends on the nature of the event: v when the operator issues an order from the supervision station, the system must respond within an time delay that is acceptable to the operator, i.e. one or two seconds between confirmation of the order by the operator and the change in status of the device displayed on the screen, v for source-changeover automatic-control functions, no specific constraints concerning the response time weigh on the application. The goal is simply to reduce to the strict minimum the time that the loads are not supplied with power. Response times of several hundred milliseconds are perfectly reasonable v if, during operation on an engine-generator set, the rated output of the set is overrun, certain non-priority loads must be shed. The authorised overload time is indicated by the manufacturer of the engine-generator set and depends on the level of the overload. In complex installations where local power generation facilities are coupled with the power supplied by the utility, in the event of a utility failure, certain loads must be shed in a fraction of a second, before the engine-generator set protective functions can react. c Data flow capacity sized for the number of measurements Electrical measurements may result is a constant flow of information on the switchboard bus. The most common measurements concern voltages, currents, power levels and quantities of energy. Sizing of the bus therefore depends not only on the quantity of information that must be transmitted, but above all on how often the information must be transmitted: v measurement values for currents or power levels may be used by the operator to monitor the distribution system in real time and the values may therefore have to be transmitted every few seconds, v values concerning the quantity of energy consumed are required only every few minutes, at most, i.e. the frequency of transmission for these values is very low. c Implementation constraints in an electrical switchboard Installation of a bus inside an electrical switchboard must take into account the following constraints: v the bus must not be sensitive to the major electromagnetic disturbances that exist in a lowvoltage switchboard, v it must be easy to install during wiring of the switchboard and be easily modified during switchboard upgrades, v the cost of each connection point, which is a decisive element in selecting a bus in that a lowvoltage switchboard comprises great numbers of connection points. Master/slave protocols are inadequate For the solutions discussed in chapter 2, master/ slave protocols are commonly used. An example is ModBus (for further information, see Cahier Technique n 147). For a basic automated switchboard, i.e. one that manages only orders and acknowledgements, a master/slave protocol is sufficient to satisfy the required functions. For example, given a switchboard with 50 incomers and outgoers and a polling time of 20 milliseconds for each one, approximately one second is required to poll all the incomers and outgoers. When an event occurs (order from the supervisor or intervention of an automatic-control function in the switchboard central unit), polling of the status of each incomer or outgoer can be interrupted to send the necessary orders. But when the system functions require the transmission of measurement values, the
weaknesses of master/slave protocols, i.e. the increased time required for each polling cycle, rapidly become apparent. What is more, when a device-status change occurs following tripping, the information is made available to the switchboard central unit only during the next polling cycle. Finally, this type of protocol is inadequate for distributed processing because the central unit can act as the master only if all the information runs through it. CSMA protocols Contrary to protocols using the master/slave access method, CSMA (Carrier Sense Multiple Access) protocols allow the various stations connected to the bus to spontaneously transmit data only when there is a real need. c CSMA constraints Random access to the bus creates three constraints that do not exist in master/slave systems which are, by definition, centralised. Solutions for these difficulties are easy to implement. v Risk of collision. Several connected stations may transmit data simultaneously. Rules are set up to avoid collision between the different messages. Two different solutions exist: - CSMA-CD (Collision Detection). Using this technique, the stations detect the interference on
the bus due to the two simultaneous messages and stop transmitting. Each station will then attempt to retransmit its message as some later time. This is the solution used by Ethernet - CSMA-CA (Collision Analysis). Using this technique, the station transmitting the message with the lowest priority level stops, thus allowing the higher-priority level message continue. Management of priorities is based the coding of the frames transmitted. This is the solution used by BatiBus (see fig. 14 ). v Non-deterministic response times. Depending on the information load on the bus, the transmission time for a frame is not constant. It is therefore not possible to guarantee a maximum transmission time using a protocol implementing this type of access to the bus. However, a certain number of devices and design rules make it possible to ensure maximum transmission times that are nearly 100% certain. For example, in the BatiBus system, commands are priority messages. This is a means to avoid the response-time constraint. v Detection of faulty stations. In a system using the master/slave method of access to the bus, the breakdown of a slave is detected by the master during the next polling cycle. But for protocols in which messages are transmitted only when necessary, a faulty module is not detected. Each application must therefore develop and implement the necessary monitoring devices required to periodically check the status of each module.
Station 1 Station 2 15 V Line status 0V Line monitoring by station 1 Line monitoring by station 2
1 1
0 0
1 0
1 Stop transmitting
OK _
OK _
OK _
OK OK
OK OK
OK Error
OK
OK
Station 1 leaves the line at rest. Station 1 pinches the line and short-circuits it.
Stations 1 and 2 leave the line at rest.
Station 1 did not detect a collision and continues transmitting its message.
Stations 1 and 2 pinches the line.
Station 1 pinches the line. Station 2 leaves the line at rest. It detects the short-circuit on the line, analyses a collision and stops transmitting.
fig. 14: collision analysis in the BatiBus system.
c Advantages of a CSMA bus for an electrical installation In the preceding paragraphs, it was shown how the constraints specific to CSMA buses may be avoided. The main advantages that may be gained from this type of bus are listed below: v Optimised data exchange: a CSMA protocol optimises data exchange because the bus is not clogged with continuous polling operations. Consequently, for a stable application, which is generally the case for an electrical installation, with the same transmission speeds as a master/ slave protocol, the quantity of useful information transmitted is significantly increased and response times can even be shorter. v Reduced costs: the greater the transmission speed on the bus, the greater the system constraints concerning protection against electromagnetic disturbances and, consequently, the greater the costs. A CSMA protocol makes it possible to select slower transmission speeds and thus reduce transmission-related costs. v Decentralised processing: a protocol offering this type of access to the bus makes for optimised processing of decentralised and/or distributed operations. The example in figure 15 (decentralisation) shows how data exchange can be simplified (opening orders to non-priority circuits) with respect to a centralised master/ slave system. Note that for distributed processing, it is the measurement module which directly transmits the opening order to the non-priority loads. Consequently, even if the switchboard central unit has failed, load shedding remains possible. Using FIP for MCC applications Certain industrial applications impose very severe demands in terms of continuity of service and performance levels. For example, a guaranteed response time (deterministic) may be required for an order issued by an automaticcontrol function managing the industrial process. This is the case for certain MCC (Motor Control Centre) switchboards. Contrary to a main LV switchboard, opening and closing orders for devices are by no means exceptional. In this case, the performance levels offered by buses implementing master/slave protocols are not sufficient, unless very high transmission speeds are used, with the corresponding high costs. Buses implementing random-access protocols are not up to the job either. It was for this type of application that the FIP bus was designed by industrial companies and manufacturers. It is not within the scope of this document to present the FIP bus in detail, however, it should be noted that it combines the advantages of both master/slave and randomaccess protocols: c access to the bus is controlled by a bus manager located in the switchboard central unit (LV switchboards),
0 ms
Status Commands
10 ms
Status
c Status information is sent as periodic variables every 10 milliseconds. c Commands are sent as periodic messages every 20 milliseconds. c The remaining available time (shown in grey) can be used to transmit parameters, measurements, diagnostics information, etc., as non-periodic messages.
20 ms
Status Commands
30 ms
Status
40 ms
Status Commands
Time
fig. 15: time diagram for an MCC implementing FIP.
c data may be periodically transmitted over the bus (orders and status information, for example), c stations may request permission from the bus manager to transmit information, as needed, for example, in the event of a significant change in the value of a measurement, etc, c the data issued by one station may be used by one or several other stations, for example, for distributed processing, c finally, the protocol has a number of built-in systems that make it possible to guarantee a very high level of transmission dependability. The FIP bus thus combines the advantages of: c master/slave protocols (deterministic and guaranteed response time), c random-access protocols (transmission of useful information or following an event). The FIP bus offers a high level of performance and meets very severe dependability constraints.
4 Implementation examples
4.1 Computer centre

Needs In a computer centre operating 24 hours per day, the primary concern of the electrical manager is to ensure continuous availability of power and fast response when maintenance is required. In addition to these basic needs, the customer may want to reduce his energy bill by: c improving the power factor; c taking advantage of special tariffs with the utility, for example by agreeing to sharply reduce consumption on peak-demand days upon reception of a special signal. Through these two modifications, the return on investment for the installation drops to less than three years due to the sharp decrease in electricity bills. Implemented solution c Electrical installation The electrical installation is supplied by a 20 kV medium-voltage loop. The MV loop supplies a 1 000 kVA transformer which in turn supplies a main LV switchboard (see fig. 16 ). The main LV switchboard is made up of
MV cubicle
Surveillance room 1 000 kVA 20 kV / 230 - 400 V F 550 kVA G D
C VAR MLVS E C A A B
VAR UPSs
fig. 16: solution implemented for a computer centre.
withdrawable units. It supplies 23 outgoers, including two reserve outgoers. The high-power outgoers are equipped with motor mechanisms for remote-control purposes. The computers are protected by two UPSs set up in a redundant configuration. A 550 kVA engine-generator set can replace the power supply for all the electrical equipment in the computer centre. c Required functions v Source changeover. In the event of a voltage loss downstream of the MV/LV transformer (or if a load reduction signal is received from the utility), the switchboard is automatically supplied by the replacement source. The cabinet managing the engine-generator set receives the shutdown and start orders from the main LV switchboard and autonomously manages the engine-generator set. When the load is transferred to the enginegenerator set, the high-power outgoers are shed to reduce the load step change during switching, and are then reconnected one after the other according to an adjustable individual time delay. When utility power returns (or the special utility signal is discontinued), the switchboard automatically transfers the load back to the normal source and requests shutdown of the engine-generator set. v Contract management. For special tariff contracts, the power utility sends a load reduction signal to the customer 30 minutes before actually shifting to the special mode. The signal, transmitted via the field bus, is decoded by a specific relay. On reception of the signal, the switchboard transfers the load to the engine-generator set, exactly as if a power failure had occurred. When the signal is discontinued, the load is automatically transferred back. v Reactive power regulation. A 100 kVAR capacitor bank used to compensate for reactive
energy consumption is managed by a reactive power regulator. v Remote surveillance. In the event of an incident in the electrical installation, the watchperson is immediately informed via a supervision console which transmits any alarms issued by the main LV switchboard. c The control and monitoring system v Each incomer and outgoer in the main LV switchboard is managed by a module (marked A in the diagram), which: - acquires the position of the device (open, closed, tripped, withdrawn, etc.); - displays this status locally; - for the remote-controlled incomers and outgoers, orders opening, closing or resetting. These orders may be given locally or sent via the switchboard bus; - dialogues with the switchboard central unit via the digital communications bus. v The switchboard control unit (marked B in the diagram) located inside the main LV switchboard is in charge of: - managing the control and monitoring functions for the incomers and outgoers via the modules marked A in the diagram; - directly acquiring two elements of information, namely the presence of utility voltage or enginegenerator set voltage (via voltage relays marked C in the diagram), and passage to the special utility mode (special utility relay marked E in the diagram); - transmitting a start order to the cabinet (marked D in the diagram) managing the enginegenerator set; - processing the source changeovers caused by a failure in utility power or reception of the special utility signal; - generating and transmitting any alarms to the supervision console (marked F in the diagram) which then displays them in the appropriate manner.
4.2 Hospital
Needs In a hospital, the continuity of electrical power service is critical. The example below deals with a mid-sized hospital. To provide optimum management of the electrical distribution system and in compliance with the expressed wishes of the operator: c Outgoers are divided into three categories, backed-up (by a generator set), priority (protected by a UPS) and no-break (protected by a UPS and a generator set). Each incomer and outgoer is monitored and may be remotely controlled from the supervisor; c The entire installation is remotely supervised. Implemented solution c Electrical installation v The electrical installation is supplied by a 20 kV medium-voltage loop. The MV loop supplies three 1 000 kVA transformers which in turn supply an LV distribution switchboard. v Two 400 kVA engine-generator sets can step in to provide back-up power to certain electrical equipment in the hospital. v Two UPSs supply the no-break and priority outgoers. v The outgoers are grouped in three LV switchboards. The diagram in figure 17 makes clear the supply system for each outgoer in LV switchboard 1. c Organisation of power management v A supervision station (supervisor) may be used by the operator to monitor the installation, issue orders and set parameters.
v The switchboard central unit in the LV distribution switchboard: - provides control and monitoring of the circuit breakers on incomers and outgoers; - checks operation of the reactive power relays (reactive-energy compensation) and stops compensation if utility voltage is absent (operation on the engine-generator sets); - dialogues with the UPS processing units; - supplies the transformers ON information. v The switchboard central unit in the enginegenerator set LV switchboard: - provides control and monitoring of the circuit breakers;
- dialogues with the control cabinets of the engine-generator sets for monitoring and transmission of ON and OFF orders; - dialogues with the LV switchboards (1, 2, 3 and no-break) which issue a starting order for the engine-generator sets and receive a set-point indicating the maximum power available depending on the operating engine-generator sets and the LV switchboards supplied with power (all are not necessarily supplied during maintenance operations). v The central units in the LV switchboards 1, 2 and 3: - control and monitor the circuit breakers;
2 x 400 kVA GE MV cubicle Gen. MLVS GE
20 kV line 3 x 1 000 kVA 20 kV / 230 - 400 V transformer Distribution MLVS MLVS 2 MLVS 3
No-break MLVS
MLVS 2 MLVS 3 UPSs with automatic bypass
MLVS 2 MLVS 3
MLVS 2 MLVS 3
MLVS 1
Backed-up outgoers (mains / generators)
Priority outgoers (mains / UPSs / mains)
No-break outgoers (mains / UPSs / generators)
fig. 17: solution implemented for a hospital.
- provide the automatic source changeover function for the backed-up outgoers on the basis of the information supplied by the voltage relays, i.e.: . load shedding of the major outgoers; . transmission of a start request for the enginegenerator sets; . closing of the engine-generator set incoming circuit breaker; - provide load regulation. Depending on the priority rating of the backed-up outgoers, they are shed and reconnected according to the power supplied by one or both engine-generator sets or by the transformers (1 and/or 2 and/or 3); - dialogue with the insulation monitor for the no-break outgoers. v The central unit in the no-break LV switchboard: - controls and monitors the circuit breakers; - controls source changeovers, after requesting starting of an engine-generator set if the UPS signals a problem.
In this example, not all the available powermanagement functions are implemented (they never are). c Time management is not implemented because a hospital operates 24 hours per day. c Contract management (smoothing of peaks, special utility modes) was not applicable, only reactive-energy compensation was implemented. c The power-management system set up is entirely dedicated to ensuring maximum availability of electrical power. c Each switchboard received local and autonomous processing capacity to carry out its assigned functions. c Very little event information circulates on the bus (automatic source changeover, enginegenerator sets, load regulation) and no measurement values except for metering values in the MV switchboard. Status checks are run periodically.
5 Conclusion and prospects for the future
Intelligent switchboards, a critical element in electrical distribution systems, provide solutions meeting the needs of managers and operators of electrical installations, notably: c energy savings; c dependability; c remote control of the installation (with possible extensions to building management for commercial applications and process management in industry); c installation maintainability and upgradeability; c gradual evolution of the installation over time toward greater intelligence. The construction of switchboards with integrated management functions, but with decentralised and distributed intelligence, is today made much easier due to the existence of standardised modules, equipment and software that will remain available over long periods. In this sense, control and monitoring can now be implemented using concepts similar to those of type-tested assemblies and shows sharp differences with respect to automatic-control functions for industrial processes.
Integration of intelligence in switchboards has made it possible to: c simplify switchboard and electrical-installation architecture, during the initial design process and later during upgrades (distributed electrical distribution, elimination of half busbar arrangements, discrimination, knowledge of switchboard reserves, management of operating conditions at switchboard maximum limits (temperature, overloads, etc.)): c manage the switchboard over time (black-box function, up-to-date diagram file, etc.); c combine communications functions (low currents) with power functions (high currents). In the near future, communication and processing will continue even further downstream to the individual devices, sensors and actuators. This will make the distribution of intelligence easier and thus further reduce centralisation. Considerable advances may be expected in the fields of design, wiring, installation, operation, dependability and upgradeability.
Bibliography
Schneider Cahier Technique publications c Introduction to dependability design Cahier Technique n 144 P. BONNEFOI c Initiation aux rseaux de communication numriques. Cahier Technique n 147 E. KOENIG c High availability electrical power distribution Cahier Technique n 148 A. LONGCHAMPT - G. GATINE c Dependability and LV Switchboards Cahier Technique n 156 O. BOUJU

Cohabitation of high and low currents
R. Calvas J. Delaballe
s Merlin Gerin s Square D s Telemecanique
n 187
Roland CALVAS Graduated from the ENSERG engineering school in 1964 and also holds a degree in Business Administration. He started at Merlin Gerin in 1966 and is currently in charge of technical communications at Schneider Electric Technical Management. In this position, he has already provided coordination for a number of publications and seminars on EMC.
Jacques DELABALLE Ph.D University of Limoges in 1980, started at Merin Gerin in 1986, after seven years at Thomson. Since 1991, he has headed the EMC laboratory of the Schneider Electric Corporate Research Organisation. He is the secretary of Technical Sub-committee 77B (HF phenomena) of the International Electrotechnical Committee (IEC).
E/CT 187 first issued april 1997

CT 187 / pp.1
Glossary
Bus: A low current link conveying digital signals. Disturbed device: Load sensitive to electromagnetic disturbances. See disturbing device. Disturbing device: Source of electromagnetic disturbances. See disturbed device. Earth (ground) loop: Loop made up of two conductors, one of which is a live conductor (phase and neutral) or a low current link, and the other one of the conductors of an EBS loop. Electromagnetic compatibility: The ability of an equipment or system to function satisfactorily in its electromagnetic environment without introducing intolerable electromagnetic disturbances to anything in that environment. Equipotential Bonding System loop (EBS loop): Loop made up of two conductors which may be the PE, a parallel earthing conductor, a shield, an additional equipotential bonding, a conductive part (not electrical). Equipotential Bonding System (IEV 195) (EBS): Interconnection of conductive parts providing equipotential bonding between those parts. Exposed conductive part: From the electrical engineers viewpoint, VEI 195, NF C 15-100 para. 232-9: conductive part of an electrical device at risk of being touched and which, although not normally live, may become so as a result of an insulation fault. Extraneous-conductive-part: Conductive part not forming part of the electric installation and liable to introduce an electric potential, generally the electric potential of a local earth. Functional earth: Point of a system or device which must be earthed for reasons other than safety of persons. High currents: Currents and links able to convey power and mainly links supplying electrical power from the
distribution mains, including the protective conductor. Loop: Surface delimited by two conductors, irrespective of their functions, likely to disturb by electromagnetic induction a sensitive and/or communicating load. Low currents: Typically buses and Voice-Data-Image signals. All electrical signals conveying information and not power. Parallel earthing conductor: Conductive device accompanying a low current link in order to minimise the surface of the loop formed by the low current link and the EBS: it may be a conductor, a trunking or a shield. The parallel earthing conductor is also a functional equipotential bonding. Pig tail: Connection in loose wire wound in corkscrew form. Absolutely to be banned for earthing of shields. High HF impedance. Protective conductor (PE) NF C 15-100 para. 241-1: Conductor specified in certain protection measures against electrical shocks and designed to electrically connect some of the following parts: (electrical) exposed conductive parts, main earthing terminal, point of the supply connected to the earth or to the artificial neutral point. Pulse Width Modulation (P.W.M.): One of its functions is to monitor the mean value of the output signal by converting the input signal into pulses of modulable width. Screen (IEV 195): Device intended to reduce the penetration of an electric, magnetic or electromagnetic field into a given region, or to enclose or separate electrical circuits. Shield (IEV 195): Barrier or enclosure provided for mechanical protection ( screen). This term is still commonly used to mean screen.
CT 187 / pp.2
In our society today, the rapid development of digital systems using low current links (bus) has given rise to the critical problem formed by cohabitation of high and low currents. The real problem is how to reconcile electrical safety and electromagnetic compatibility. The following issues thus arise for which solutions must be found: c how to treat the problem of exposed conductive parts, c which earthing system should be chosen, c which shields, mitigating planes, Faraday cages should be chosen and for what purpose, c how to organise routing of high and low current circuits, c and many others. Although written with electrical engineers in mind, this Cahier Technique will certainly also be of interest to low current specialists since it mainly deals with low frequency disturbances of < 1 MHz.
Contents
1 Introduction 1.1 EMC: a discipline covering many professions 1.2 Review of disturbances and coupling mechanisms 1.3 Distinction between high and low currents 2 Earth and exposed conductive parts 2.1 Earth electrode 2.2 Equipotential bonding system 2.3 Mesh monting between protective ESB and other ESB 3 EMC in non-communicating electronic devices 3.1 Self-induced disturbances 3.2 Exposure to radiated fields 3.3 Conducted disturbances 4.1 Example of a disturbance caused by common impedance 4.2 Example of a disturbance caused by radiation 5 Conclusion Appendix: bibliography pp.4 pp.4 pp.7 pp.8 pp.10 pp.12 pp.13 pp.14 pp.15 pp.16 pp.18 pp.20 pp.22
4 EMC in communicating electronic devices
CT 187 / pp.3
1 Introduction
1.1. EMC: a discipline covering many professions

Electromagnetic compatibility (EMC), a rapidly developing field, came into being with the start of radio broadcasting. One of the first applications, a half-century ago, was the shields of sparkplugs of internal combustion engines to avoid interference with radios receivers. Today, EMC is the subject of numerous studies carried out by experts, particularly in the fields of electronics and telecommunications. Many standards and even an european directive have been published to ensure the cohabitation of disturbing and disturbed devices and systems. For this, they set electromagnetic disturbance emission limits and immunity levels. In spite of all efforts, there remains a complex field in which not all experts agree: the cohabitation of high-current and lowcurrent systems. This concerns non-communicating electronic devices because they are supplied by the electrical distribution system and are sensitive to LF and HF radiated fields as well as to conducted disturbances. However, the subject is of even greater importance to communicating systems making use of distributed intelligence. In this case, disturbances may affect the dependability of systems designed for information technology, building and energy management (BEMs), power management, automatic process control, etc. All these systems are increasingly prevalent today and make essentially use of digital communications via low-current media. In the design and installation of such systems, people working in computing, electrical engineering, automatic control and electronics frequently encounter problems in communicating with each other and in coordinating their respective tasks. A common understanding of the phenomena involved must be achieved to avoid design and installation errors that can lead to malfunctions or even serious damage to equipment. The purpose of this Cahier Technique is to contribute to that common understanding which is becoming all the more important with the rapid evolution of techniques, because: c electrical distribution systems are increasingly powerful, carry more and more harmonic currents and use system earthing arrangements that may have a negative impact, c digital communication networks are expanding rapidly, with ever lower electrical levels (a few volts) and ever higher bit rate (measured in Mbit). Electricians and electronics experts must work together to optimise EMC conditions in buildings and improve the cohabitation of high and low currents. In this Cahier Technique we will focus more on LF rather than HF phenomena: the limit between LF and HF is set at 1 MHz.
1.2 Review of disturbances and coupling mechanisms

Disturbances A wide range of electromagnetic disturbances exist and they may be classed according to a number of criteria. They may be continuous or transient, sinusoidal or pulsating, low or high frequency (> 1 MHz), conducted or radiated, common or differential mode, and have their cause either inside or outside the building. The disturbing phenomena studied for EMC are many. This Cahier Technique will concentrate mainly on the following: c harmonic currents and high fault currents, c switching surges, c voltage and current impulses resulting from lightning.
Log(Z) High impedance source (E predominates) Low impedance source (H predominates) /2 Log(d)
377
fig. 1: wave impedance Z = f(d).
High voltages and currents generate electromagnetic fields.
CT 187 / pp.4
Emitter Mains (1 kA, 1 ph) Lightning Drying oven 20 kV switch FM radio Long wave radio Walkie-Talkie UHF television Mobile telephone Radar Microwave oven
Frequency (f) 50 Hz 30 kHz at 3 MHz 27 MHz 75 MHz 100 MHz 200 kHz 450 MHz 600 MHz 900 MHz 1 GHz 2.45 GHz
Wave length () 6000 km 10 km 11 m 4m 3m 1500 m 66 cm 50 cm 33 cm 30 cm 12 cm
Field 20 A/m at 10 m 10 A/m at 500 m 1.5 V/m at 10 m 5 kV/m at 1 m 1 V/m at 500 m 30 V/m at 500 m 10 V/m at 1 m 0.5 V/m at 500 m 20 V/m at 1 m 40 V/m at 500 m 1.5 V/m at 1 m
fig. 2: some emitters of electromagnetic disturbances.
Note that all electromagnetic fields are made up of a magnetic field H and an electric field E. Near a LF current, the magnetic field is dominant up to a distance of approximately /2 i.e. 1 000 km for a 50 Hz field. Near a high impedance HF source, the electrical field is dominant. This is often the case for switching surges on electrical distribution systems. Beyond /2 (for sources with small dimensions compared to ), the ratio between E and H is constant and determines the wave impedance:
Z0 = E H =
(average value) and the fields that they are capable of radiating, given their respective powers. The coupling mechanisms Note that depending on the nature of the disturbance, the coupling between the disturbing phenomenon and the disturbed device may be one of the following: v common impedance (conducted disturbance), v electrical (stray capacitance and radiation), v magnetic (mutual inductance and radiation). c Common-impedance coupling This is the result of a circuit that is common to two or more devices. The common circuit may be the electrical distribution system, equipotential-bonding system, protective EBS, etc. Figure 3 illustrates this type of coupling.
= 377 in the air
(see. fig. 1 ). The table in figure 2 provides several examples of disturbance emitters with their frequency
a - Differential mode coupling (DM)
Electrical distribution U system
ZMD1 ZMD2 ZMD = ZMD1+ ZMD2 Disturbed device U+ZMD IMD
IMD
IMD
Harmonic transients
Disturbing load
b - Common mode coupling (CM)

Electrical distribution system U ZMC1 ZMC = ZMC1+ ZMC2 U - ZMC IMC Electronic device (disturbed device) Disturbing load
IMC
ZMC2 PE
fig. 3: common-impedance coupling. The disturbances generated by a disturbing load have a negative effect on a sensitive load when the power supplies are not decoupled ( a ) or the EBS are not decoupled (PE) ( b ).
CT 187 / pp.5
c Electrical coupling This type of coupling is due to the presence of either an external electric field, or a stray capacitance between two circuits or two conductors (see fig. 4 ).
Cable (power) H
C12
C10
Pair of wires (low-level)

R V1 H
VN
C20
fig. 5: a variation in the current in a cable generates an electromagnetic field which, over a short distance, may be considered exclusively magnetic, in which case it induces a disturbing voltage in wires forming a loop. This phenomenon is called inductive crosstalk.
fig. 4: a sudden variation in voltage V1 between two wires generates a field which, over a short distance, may be considered primarily electric and may induce a voltage VN in another parallel set of wires. This phenomenon is called capacitive crosstalk.
c Magnetic coupling This type of coupling is due to the effect of magnetic field, either external or generated by a cable, on an inductive loop (generated by 2 cables located in the same circuit) (see figure 5 ). In many cases, several of these coupling mechanisms may simultaneous exist. Particularly concerning electrical power distribution: v short-circuit currents (or insulation fault currents in TN systems) generate very strong magnetic fields. According to Amperes theorem: , where d is the distance between the 2d conductor and the disturbed device, v switching of switchgear: switches, contactors, circuit-breakers and fuses all generate a variety of strong radiated and conducted disturbances which need to be
H =
controlled by the manufacturers of the electronic equipment incorporated in electrical switchboards. These waves have been characterised by standard-makers (see fig. 6 ), v lightning when it falls on or near a building or on an electrical line is a highly disturbing pulsating phenomenon which may generate a magnetic field, an electrical field, a conducted wave or a difference in potential between two electrical points of the installation.
CEI 1000-4-2 CEI 1000-4-3 CEI 1000-4-4 CEI 1000-4-5 CEI 1000-4-6 CEI 1000-4-8 CEI 1000-4-9 CEI 1000-4-10 CEI 1000-4-12 CEI 1000-4-13
Electrostatic discharges Radiated disturbances Rapid transient bursts Shock waves (lightning type) Conducted disturbances f > 9 kHz Magnetic fields at power system frequency Pulsating magnetic fields Damped oscillating magnetic fields Damped oscillating waves Harmonics
fig. 6: a few disturbance test standards.
CT 187 / pp.6
1.3 Distinction between high and low currents

The term high currents is applied to normal phenomena such as flow of a 50 Hz electrical current or abnormal phenomena such as lightning currents. Figure 7 classifies the various electrical links into four groups. c Group 1 for high currents. c Group 2 for insensitive, non-disturbing and normally short links. c Group 3 for digital links (bus) is rapidly growing both in tertiary and industry; mutiple routings between communicating devices mark this group out for cohabitation with the high current links. c Group 4 for high impedance analog links which seldom cohabitate with links from the other groups due to their sensitivity: links are rare and short. A link belonging to one of the low current groups 2 or 3 may disturb a more sensitive link.
Power links (power supplies)
4 separate groups: c Group 1: not very sensitive but disturbs the groups below.
Digital I.O. System
Relay links
c Group 2: not very sensitive but disturbs the group below.
Digital links (bus)
c Group 3: sensitive to pulses, disturbs the groupe below.
Analog links Interface sensor c Group 4: very sensitive.
fig. 7: classification of electrical link types in order of increasing sensitivity.
CT 187 / pp.7
2 Earth and exposed conductive parts
2.1 Earth electrode

The prime function of an earth electrode is the protection of persons. It is a fact that we live on Earth ! And it is vital to earth exposed metal parts of electrical equipment to avoid electrocution by indirect contact should an insulation fault occur. This measure has been stipulated in the standards since 1923 (IEC 364; NF C 15-100). The fault current varies in strength according to the earthing system used, and measures are taken to ensure that contact voltage does not exceed conventional safety voltage for a stipulated time: UL (50 V in a.c.) (see Cahiers Techniques 172 and 173). The exposed metal parts of electrical equipment are connected to the protective conductors (PE) in turn connected to the earth, thus forming the earthing arrangement. The second function of an earth electrode is to minimise common mode disturbances external to the LV installation. An example is 50/60 Hz overvoltage in the event of MV/LV transformer breakdown (see fig. 8 ) or overvoltage due to lightning (see fig. 9 ). With reference to the above, NF C 13-100 has laid down limit earth electrode values in France. Lightning, MV/LV faults and safety of persons call for use of low impedance earth electrodes (IhMV can reach 1000 A and the insulating voltage of sensitive devices is 1500 V !). This problem particularly needs to be managed in TT earthing systems. Naturally multiple earth electrodes should be avoided unless they are interconnected. The earth electrode may be one or more spikes pressed into the ground or a foundation ditch loop, or a combination of both. For a spike: Rp =
MV
LV 3
Ih
TN TT RN RU
fig. 8: MV/LV breakdown; the power system potential rises with respect to the earth: U = RN Ih resulting in risk for the equipment in TT system or for persons in TN system if the building is not completely equipotentially bound.
OR MV LV 3
fig. 9: lightning surge: the power system is subjected to pulsating overvoltage on all live conductors, resulting in a high EMC risk. Requires use of lightning arresters whatever the earthing system used.
L where L is the length of the spike or the perimeter of the loop. The earth electrode must be a solid copper or stainless steel conductor so as to limit oxidation. Ground resistivity () is an important parameter, varying considerably with ground humidity and nature from 1 to 5000 .m. It is vital to place the right earth at the bottom of the foundation ditch around the loop conductor (see fig. 10 ).
For the foundation ditch loop: RFF = 2
The impedance of an earth electrode varies only slightly between 50 Hz and 500 kHz. If the building is equipped with lightning rods, the rod downcomers must be connected to earth electrodes in triangular crossbracing. All conductors which might be required to convey lightning currents must be flat conductors in order to reduce the self-induction factor and skin effect and consequently the drop in linear voltage (see fig. 11 ).
CT 187 / pp.8
Top soil Tinned copper flat bar 30 x 2
fig. 10: earth electrode: creating a loop in the foundation ditch.
c Continuous resistance of a conductor is RC = l / s , i.e. RC = 1.7 m for a 10 m long cylindrical copper conductor with a 100 mm2 cross-section. As frequency increases, the skin effect strengthens this resistance. This is because HF currents are concentrated at the edges of the conductor over a depth of = ( f )-1/2 which decreases with frequency. is known as the skin depth. For copper,
yy ,, yy ,, yy ,, yy ,, QQQQ PPPP @@@@ ,,,, yy ,, QQQQ PPPP @@@@ ,,,, QQQQ PPPP @@@@ ,,,, QQQQQ ,,,,,
External wall
Backfill
Foundation
Clean concrete
This formula only applies if the radius r of the conductor is greater than the skin depth . Our copper conductor is such that: v at 50 Hz: RHF = RC = 1.7 m, v at 1 MHz: RHF = 43.1 RC = 73 m, v at 10 MHz: RHF = 135 RC = 230 m. c The self-inductance L (H) of a conductor of a length l is: v for a cylindrical conductor:
(50 Hz) = 9.3 mm, (1 MHz) = 65.8 m, (10 MHz) = 21 m.
As a result the real cross-section of the conductor decreases. The ratio between RHF and RC is given for a cylindrical conductor with a radius r by:
4l 1+ 100r 0.2l 2.303 log10 d v for a conductor with rectangular crosssection:

2l w+e + 0.5 + 0.2235 0.2l 2.303 log10 w+e l l, d(diameter), , w (width), e (thickness) are expressed in metres.
RHF 1 r 6 = + + 0.178 RC 4 2
Dimensions d (mm) 11.28 10 50 100 500 w (mm)
1 6
0.25 +
r 2
Inductance Impedance Z e (mm) 10 2 1 0.2 L(H) 16.2* 14.8 12.9 11.6 8.4 at 50 Hz () 0.0068 0.00635 0.00575 0.00535 0.00435 at 1 MHz () 90 93 81 73 53 at 10 MHz () 900 930 810 730 530
Self-inductance and impedance of a 10 m long copper conductor with a 100 mm2 cross-section depending on its geometry. (*) in the table the self-inductance of the cylindrical conductor is given at 50 Hz. In HF the term 100 r. becomes negligible, and L 14.35 H becomes, like the rectangular cross-section conductor, independant of frequency. Finally note that in HF the impedance Z = 2.f.L becomes preponderant compared with the resistance RHF.
fig. 11: impedance of conductors depending on their geometry and frequency.
CT 187 / pp.9
It is important for these conductors to be as short as possible. Let us take the example of an LV lightning arrester designed to limit common mode voltage to 1.5 kV. If it is connected between the protected phase and the earth strip by a 1 m long conductor with circular crosssection, a 5 kA current and a 8 s rising edge, it will develop the following voltage: L
connect the application directly to the terminals of the lightning arrester (see fig. 12 ).
di 5 x 103 = 1 106 = 625 V dt 8 x 106

fig. 12: optimum connection of a lightning arrester.
hence a total overvoltage of 2.1 kV which is dangerous as many devices have an impulse voltage withstand of 1.5 kV. The solution is to
2.2 Equipotential bonding system

Types of conductive parts A building contains a variety of metal conductive parts, for example: c the metal casings of electrical loads and electronic equipment, c the metal structures of buildings, c the water or gas pipes and facilities, c the functional bonding conductors of the signal transmission electronic equipment (0 volt), c the shield and Faraday cage type exposed conductive parts whose function is to block electromagnetic fields. Like the earth, the EBS have two functions: First function: protection of persons Dangerous potentials may be present between metal casings, gas or water pipes and the metal structures of buildings. To ensure protection of persons, all simultaneously accessible exposed conductive parts must therefore be interconnected and the building must be made equipotential. It is with this in mind that installation standards stipulate that all the abovementioned exposed conductive parts must be connected to the main equipotential bonding regardless of the earthing system (see fig. 13 ). The earthing of load exposed conductive parts forms a star-shaped protection equipotentialbonding system, with tree-structured distribution of the protective conductors (PE) as they are in the same cables as the live conductors. Second function: dependability of electronic systems Electronic systems are more sensitive than people to differences in potential and electromagnetic radiation. In addition to conducted disturbance blocking devices, they require ground planes, shields and Faraday cages to block the electromagnetic fields, as well as equipotential-bonding system, particularly in the case of devices communicating via data transmission bus. In this case equipotential bonding must be optimum in the building since communicating devices, whether used for control/monitoring or computer purposes, may be located geographically at some distance from each other on the same floor of a building or even on different floors.
PE
Gas
Water
Building earth electrode
Structures
fig. 13: connection of all exposed conductive parts to the main equipotential bonding.
CT 187 / pp.10
Solution The solution is a meshed equipotential bonding system. A number of reasons justify this choice: c the fight against lightning electromagnetic fields. Lightning may directly strike the building. If this happens, if only one lightning rod downcomer conductor is used, the lightning current will result in: v appearance of a very strong magnetic field in the building, v a pulsating electrical field due to the very high voltage developed in the downcoming conductor
(v = l L
di . ) dt
fig. 14: horizontal and vertical mesh bonding of the building. It forms a Faraday cage.
The solution is vertical mesh bonding with a downcoming conductor every 10 m for example.
a - Mean spectral amplitude density of the electromagnetic fields of lightning flashes (PIERCE curve) Peak amplitude of the field (V/m) 10 Km from the flash 106 105 10
4
The return arc can be likened to a vertical aerial several kilometres high through which a peak current of several dozen kiloamps flows. This aerial radiates an electrical field and a magnetic field which decrease by 1/D beyond a distance:
c = 2 2f d = 50 m at 1 MHz.
d =
The return arc is a pulse current with an extremely rich frequency content (broad frequency spectrum) and varying considerable from one lightning stroke to another. Figure 15a shows the graph for the average spectrum characterising lightning flashes, obtained from many field measurements at several points of the world. IEC standard-makers have chosen for induced voltages on electrical distribution systems, a 1.2/50 s wave whose spectrum is given in figure 15b. It is very similar to Pierces curve.
103 102 102
103
104
105
106 107 Frequency (Hz)
b - Frequency spectrum of the standardised lightning wave (1.2 / 50 s) Relative amplitude 1 10-1 10-2 10-3 10-4 102
103
104
105
106 107 Frequency (Hz)
fig. 15: frequency spectrum of the lightning electromagnetic phenomenon.
The advantage is the division of currents and hence of magnetic fields, and the selfattenuation of these fields inside the building due to their mutual opposition. Lightning may fall near a building. If this occurs, protection of the installations inside this building
requires creation of a Faraday cage and thus addition of a horizontal mesh bonding to the vertical mesh bonding (see fig. 14 ). If we consider that the frequency spectrum of lightning (see fig. 15 ) is essentially, from an energy point of view, less than 1 MHz and that
CT 187 / pp.11
the cage is effective for a pitch of /30, the mesh must be
c 3 x 108 = = 10 m 30f 3 x 107
c Locally, reduction of electromagnetic fields as a result of the ground plane effect. If a sensitive device or communication bus is placed on a conductive surface, it is less exposed to electromagnetic fields as this surface develops a field which opposes the disturbing field.
This is why computer rooms have meshed floors and why low current cables are placed on metal trunkings. c Minimisation of bonding impedances between any two points. The impedance of a copper conductor rises with the frequency of the current that it conveys (inductance and skin effect). Thus at 1 MHz, Z is of the order of 10 a metre. Equipotential-bonding is considerably improved if the disturbing current is able to choose between a large number of routes.
2.3 Mesh bonding between protective ESB and other ESB

We have seen above that earthing arrangements dedicated to the protection of persons are star structured (tree-structured for the protective conductor) and that a single meshed EBS is required for dependability of electronic systems. In theory these circuits can be separated in the building even if they are connected to the same earth connection. Even if standards define several types of potential references (see fig. 16 ), in practice few electrical and electronic devices and systems make such distinctions. For example the notion of a noise-less exposed conductive part is questionable and rapidly going out of use in view of the development of communicating systems and the large number of interconnections. In high frequency, stray capacitances make these distinctions even more illusory. The equipotential protection system (electrical exposed conductive parts) and the equipotential functional system (other exposed conductive parts) thus need to be connected in the new buildings to form one single system of
Earth
Portective conductor
Noiseless earth
Exposed conductive part
fig. 16: examples of earth and exposed conductive part symbols as in NF C 03-202.
equipotential bondings. This system must guarantee the integrity of the protective links (PE) to ensure protection of persons. There is no need to oppose high current star-shaped systems and the meshed systems required for low currents. In existing buildings it is advisable to ensure interconnection of exposed conductive parts between sensitive devices (if they communicate), to increase electrical continuity of trunkings and create meshed ground planes if required.
CT 187 / pp.12
3 EMC in non-communicating electronic devices
The term non-communicating means that no low current circuits enter or leave the device. The adjective isolated is not applicable because the device is supplied with power from the electrical distribution system, is in contact with its electrical environment via stray capacitances and is influenced by electromagnetic fields. One example is a personal computer (without a printer). This would appear to be a simple case, because manufacturers must observe standards defining emission and immunity levels.
The question is whether, in spite of compliance with standards, the device still risks being disturbed and whether precautionary measures should be taken when the electromagnetic environment is particularly severe or when very sensitive loads are placed nearby. Such measures are indispensable when there is a risk of self-induced disturbances or the presence of radiated fields or conducted disturbances due to high currents .
3.1 Self-induced disturbances

A wide range of devices today use power electronics. An example is switch-mode power supplies which are now commonly used in the residential, commercial and industrial sectors. Figure 17 shows the general layout of this type of device and the disturbances that it generates. These devices use power transistors operating at frequencies of several dozen kHz and even much higher for low-output devices, with surprisingly high di/dt and dV/dt values (e.g. a few hundred amperes per microsecond). Inherent EMC is required to ensure correct operation and is generally taken into account account right from the design stage: circuit board
3 1 1 1 2 230 V electrical distribution system 2 1 Differential mode 2 Common mode 3 Radiation Application
Heat sink
fig. 17: different disturbances produced by a switch mode power supply.
CT 187 / pp.13
routing (see fig. 18 ), limitation of stray capacitances, shields, HF filters on the mains side, minimum distances from disturbing cables, etc, with tests to ensure compliance. In spite of all the above, the risk of self-induced
disturbances remains, due in particular to devices frequently being implemented differently from standard test implementation: for example insufficient earthing links which are too long and suitable for common-impedance coupling, etc.
fig. 18: inductive couplings due to loop presence: the lefthand line shows two large loops to supply the two circuits; the righthand line is preferable.
3.2 Exposure to radiated fields

Electric fields Surrounding electric fields can disturb a device and its supply circuits (50/60 Hz) via the antenna effect. Counter-measures include: c minimising the antenna effect (for example, no conductors with free ends, conductors near EBS conductive parts), c using HF filters on the power supply cable to stop incoming disturbances, c shielding the building by limiting the size of openings. Magnetic fields The attenuation resulting from shields depends mainly on absorption which is reversely proportional to skin depth (). For steel this is 0.07 mm at 30 kHz and 1.7 mm at 50 Hz, whereas for copper it is 10 mm at 50 Hz! (attentuation is 8.7 dB where shielding thickness e = ). The above makes clear that effective shields in low frequency is very difficult and costly. What is more, some openings or slots allow penetration of induction by magnetic dipole effect (see fig. 19 ).
H radiation
An opening of length l in a shield acts behaves like a dipole of the same length (Babinets principle) and re-radiates part of the incident field from the other side of the shield. Opening transparency becomes total when l /2. Beneath its resonance frequency, i.e. when l < /2 (a condition encountered in low frequency phenomena) a slot can be likened to an inductance which equals roughly 1 nH/cm. A surface current J broken by such a slot generates a difference in potential between its edges and lets through some magnetic field. It then radiates an electric field E at right angles and a magnetic field H (dominant) parallel to its length.
fig. 19: slot coupling.
CT 187 / pp.14
The magnetic field penetrating the system and/or the magnetic fields generated in the cubicle by high currents, induce common mode and differential mode voltages in the loops. Note that attenuation of LF magnetic disturbances arriving via the power supply cable, using standard filters (termed HF ), is very limited. In buildings, radiated magnetic fields may have a number of sources, notably lightning, transformer leakage flux, incoming and outgoing circuits of static converters as well as the protective conductors of TN-C systems. This case is illustrated in figure 20. Take the case of a power cable running in a building void. The system is TN-C. The load imbalance current flows continuously through the protective conductor (PE), which is also the neutral conductor (N), along with the third order harmonic currents and their multiples. Use of non-linear loads such as static converters and in particular switch mode power supplies is responsible for the increasing importance of these harmonic currents. Due to the current drawn off to the metallic structure of the building, the vectorial sum of the currents in the cable is not equal to zero. This differential current, which may amount to several dozen amperes, generates, over a distance d, a disturbing magnetic field H = /2d. For example a 63 A current generates, over 10 metres, a 1 A/m field which is sufficient to disturb not only cathode ray tubes but also certain magnetic components, given that HF shields do not block LF magnetic fields. Note that the magnetic field developed in the
Metal structures of building
PEN Phases
VISU
I 0
fig. 20: magnetic radiation due to neutral or fault current in TN-C system.
event of an insulation fault, still in the TN system, is very strong (Id = Icc). For low frequency magnetic fields, the countermeasures are: c distance, c power cables incorporating the phase, the neutral and the protective conductor. Single-core cables must be avoided wherever possible, c a suitable system earthing arrangement, c metal trunking (mitigating plane), c use of metal enclosures, preferably ferromagnetic.
3.3 Conducted disturbances

Above 1 MHz, the impedance of cables rapidly limits the level of disturbances (L = 1 H/m). Troublesome disturbances are therefore made up essentially of harmonics, overvoltages due to switching and lightning voltages and currents. c Harmonic voltages are, generally speaking, not troublesome for non-communicating devices. They may disturb controlled rectifiers (variation of zero crossing point). c Harmonic currents may generate, through inductive crosstalk, conducted disturbances in low current circuits. c Overvoltages due to switching or lightning may cause malfunctions or even the destruction of components. The counter-measure is the use of surge limiters (lightning arresters) or wave absorbers combining filters, lightning arresters and a shielded transformer. The latter attenuate HF disturbances and, above all, block commonmode LF voltages.
CT 187 / pp.15
4 EMC in communicating electronic devices
Communicating systems are more sensitive to disturbances than non-communicating devices because digital links involve very low voltage levels and are more exposed to disturbances generated by common impedance and radiation. Digital signals are particularly sensitive to transient phenomena. Their susceptibility is increased by the increasing lengths of connections and number of sources of disturbances.
For disturbances with frequencies less than 1 MHz, the phenomena that must be taken into account are, once again, primarily the 50 Hz mains disturbances and harmonics, overvoltages due to switching and the effects of lightning. Switching overvoltages are generally differential mode and are best blocked where emitted using surge limiters (RC circuits-ZnO limiters, etc). This subject will not be dealt with in this Cahier Technique (see CT 179).
L PEN
Id PEN
( 20 kA)
Id In
V Device 1
Device 1
Digital link
V0
Digital link
Device 2
Device 2
LV
LV MV
MV
Conductive parts of building
fig. 21: TN-C system; the neutral current develops differences in potential which disturb the voltage references of the digital links; the fault current (several kA) develops a destructive V.
fig. 22: TN-C system with mesh bonding: the permanent V (unbalance, harmonic multiples of 3) and temporary V (fault) is attenuated, but there are stray currents : risk of fire and disturbing magnetic fields, including at power cable level (see fig. 20).
CT 187 / pp.16
4.1 Example of a disturbance caused by common impedance

Consider two communicating devices that are part of a distributed intelligence system and supplied by the same electrical distribution system. What happens when an insulation fault occurs (see fig. 21 ): If there is no mesh bonding in the TN-C system, the insulation fault generates a short-circuit current Id. The voltage drop d in the electrical equipotential bonding conductor may exceed half phase voltage (PE cross-section < phase crosssection). This voltage is also present between the exposed conductive parts of the two devices, which disturbs communication and may even seriously damage the transmitters/receivers of the digital link if appropriate protection is not provided. This is because in cases where the low current cable is coaxial, V will be applied in differential mode. The figures 22, 23 and 24 show what happens if the conductive parts are meshed in the various earthing system possibilities (on the second fault in the IT system the risks are the same as those present in the TN-S). Note that if the mesh bonding of all conductive parts is very dense, the problem of the V ceases to exist in the TN system, but the insulation fault current is now greater than 3Ph Icc (very small impedance of the neutral return circuit). What conclusions must be drawn for existing and new buildings where conductive part mesh bonding is insufficient?
Id PE ( 20 kA)
Id
( 20 A)
PE
Device 1 Digital link

V0
Device 1 Digital link
V0
Device 2
Device 2
LV
LV
MV
RU Conductive parts of building RN
MV
Conductive parts of building
fig. 23: TN-S diagram, temporary V; the fault current in the protection cable temporarily destroys equipotential-bonding. As the PE is connected to the structures, division of currents and magnetic fields occurs as in figure 20 .
fig. 24: TT system; the low current in the PE does not disturb the equipotential-bonding of the potential references of the communicating electronic devices (RPE is low compared with RU and RN and Id small; the magnetic fields and V are 1000 times smaller than in TN-C or TN-S.
CT 187 / pp.17
c Avoid system earthing arrangements which, in the event of an insulation fault, generate a shortcircuit current (see fig. 24 ). c Divide the fault currents by making numerous connections between the exposed conductive parts and the metal structure of the building. The V is reduced to a minimum however, the radiation of the power cables, including
the PE conductor, must be controlled (see the section on radiated fields in the preceding section). c Use twisted shielded pairs where possible because the current flowing in the shield has limited effects in view of the small transfer impedance.
4.2 Example of a disturbance caused by radiation

For two computers interconnected by a network (see fig. 25a ), what happens when lightning strikes? Consider a lightning strike 200 m from the building with di/dt = 25 x 109 A/s ( = 25 kA, tm = 1 s). If the loop, formed by the 50 Hz distribution system and the digital links (see fig. 25a ) forms a surface area of 50 m2 for the impulse field, the resulting EMF is equal to:
a - Loop formed by the electrical distribution system and the digital link
Digital link
e= e=
d dH 0S di = 0S = dt dt 2d dt
PE Power supply
4 107 50 25 109 = 1 .25 kV 2 200 That is a dangerously high level for the digital transmitters/receivers. If the loop is closed, it is the resulting current that causes the damage.
What counter-measures can be used? Minimisation of loop surface, power cables, low current cables: the problem is that if the loop is open, a voltage which is dangerous for electronic devices is developed and, if it is closed, the induced current (transfer impedance) will disturb the signal or even destroy the transmitterreceiver circuits. Figure 25a shows that the loop may be a large one. A parallel earthing conductor, trunking or metal tube (see fig. 25b ) can be used to minimise loop surface. A word of warning, however, as this creates an EBS loop. The conductive link between the two communicating devices must therefore be a low impedance one to prevent developing induced voltage between the exposed conductive parts of the communicating devices (avoid pig tails). Note that if this link impedance is small, a large part of the fault current will flow through it should an insulation fault arise. The solution is once again to mesh the exposed conductive parts as densely as possible in order to divide the currents and/or an earthing system with low fault current.
b - The trunking, connected to the exposed conductive parts of both devices, minimises the induced current effect with the digital link
Reduced earth (ground) loop
EBS loop
c - Using parallel trunkings for the power supply and digital link: reduced loops, mitigating planes and effects attenuated by mesh bonding
Earth loops and EBS loops reduced
fig. 25: disturbance of a digital link by inductive loop and the solution.
CT 187 / pp.18
Whatever the circumstances we recommend that you apply the solution shown in figure 25c in which the high/low current loop is removed as is also the EBS loop. Three conductors are connected in parallel: the two trunkings and the protective conductor (PE). The trunkings should preferably be separate, follow the same route and run close to one another: for example a riser. On a floor of a building (high currents of reduced strength) the power and signal cables can run on the same trunking (see fig. 26 ). The distance between power and signal cables on a metal trunking depends on a number of factors: c the level of immunity of the communicating devices, c the capacity of the communication protocols to manage disturbed messages, c the distance over which the cables run together, c the current flowing in the power cables, c whether the cable contains the PE or singlepole conductors,
Analog Digital
Power dV/dt dI/dt
Power Analog Digital
10 cm
fig. 26: separation of conductors of different types in the same trunking.
c the electrical characteristics of the signals (electrical level, frequency, impedance). Communicating equipment manufacturers recommend a minimum distance, normally 30 cm. However the distance to be used in a given case should always be validated by an expert.
CT 187 / pp.19
5 Conclusion
This Cahier Technique has shown that a variety of professions are concerned by the cohabitation of high and low currents. We have not dealt with communication softwares for which designers must find solutions for the disturances which could affect them. Electronic, electrical, automatic control and software experts must join forces and work together to design and produce installations. In France Sub-Committee 15D of the UTE is currently working on the guide designed to meet this need. Electronic engineers know that preference should be given to shielded twisted cables, that there should be no conductors with free ends, that cables should be flattened against an equipotential metal surface, that pig tails should be avoided and that great care must be taken with choosing and implementing cable glands. However they are not always aware that it is highly advisable for low current cables to follow the paths of high current ones, that copper links between two buildings are dangerous even if their respective earth electrodes are connected by a 35 mm 2 conductor (L = 1 H/m) and that exposed conductive part mesh bonding, if required, may cause problems (for example risk of fire in TN-C). Electrical engineers are disturbed by the distinction made between protective EBS
and EBS. They do not always understand the importance of total mesh bonding and do not always grasp why interconnection of exposed conductive parts and continuity of trunkings need to be ensured with such care. We have shown that non-communicating devices, even when correctly designed and in compliance with emission and immunity standards, are nonetheless affected by the disturbances and coupling mechanisms found in installations and not always perfectly elucidated. Communicating systems, increasingly numerous and wide-spread, clearly pose the problem of how to make use of high currents and low currents in a specific building. Inductive loops must be avoided or shunted by a parallel earthing equipotential bonding conductor. The best possible equipotential bonding must be achieved for both high and low frequencies, not only on each floor, but also between floors to form a complete equipotential bonding mesh. In all cases correct use of cableways, trunkings and metal skirting boards is vital. Figure 27 reflects the electrical engineers vision, that of the EMC expert, and offers a minimum, inexpensive solution for existing buildings.
CT 187 / pp.20
Electrical safety
a - Equipotential protection system with a bus (between floors) and star (on each floor) topology. The same applies to electrical power distribution. c Purpose: safety of persons/indirect contact c Attributes: - no EBS loops, - possible calculation of fault current, - monitored fault current routing, - low impedance at 50 Hz but high in HF, - no common impedance coupling (except in TN at the riser).
Metal structures
PE
Pipes
EMC equipotential-bonding
b - Meshed equipotential bonding system c Purpose: - equipotential-bonding in HF, - low impedance in LF and HF. c Attributes: - random, multiple routing of 50 Hz fault currents and neutral currents (TN-C) as earths in PE, - small earth loops (HF), - common impedance coupling, - increased fault current in TN. NB: to ensure efficiency, particularly in HF, mesh bonding must be very dense. Type 3 links, if made, create EBS loops which will affect devices located on different floors. Thus, if links of this type are made, the equipotential bondings between floors must be multiplied in order to minimise the surface of the EBS loops.
2 1 2 3 2 3 3
(1) Equipotential-bonding conductor encircling the floor of a building. (2) Shield or parallel earthing conductor, metal trunking. (3) Additional equipotential bonding conductor. Safety and EMC
c - Minimum protection and equipotential-bonding system. Optimum use of trunkings. c Purpose: same as in figures 22a and 22b. c Attributes: - does away with high current/low current loops, earth loops and EBS loops, - HF equipotential-bonding, - LF equipotential-bonding (if low Id), - no stray currents, - mitigating plane effect/EM fields. NB: - can be completed by local meshed ground planes, - does not prevent direct links between two loads close to each other (with parallel earthing conductor), - at floor level, a single trunking may be enough, - the riser in the centre limits external influences. c On the upper floors, type 3 links must be provided if the electrical equipment is less than one metre away from the metal structures outside the building.
parallel earthing conductors high current trunkings
fig. 27: searching for equipotential bonding in LF and HF.
CT 187 / pp.21
Appendix: bibliography
Standards c IEC 1000-4: sections 2 to 13: Electromagnetic compatibility (EMC), part 4: testing and measurement techniques. (see fig. 6 ) c European Directive CEM 89/336/CEE c UTE C 90-490: recommendations for cabling intelligent buildings, October 1995 Schneider Cahiers Techniques c Les perturbations lectriques en BT, Cahier Technique n141, April 1991 R. CALVAS c EMC: electromagnetic compatibility, Cahier Technique n149, August 1996 F. VAILLANT c Perturbations des systmes lectroniques et schmas des liaisons la terre Cahier Technique n177, September 1995 R. CALVAS Various publications c La compatibilit lectromagntique Editions Schneider 07/96 MD1CEM1F c Manuel didactique CEM Editions Schneider 01/96 ART 62920 c REE April 96 (Revue de llectricit et de llectronique) c REE November 95
CT 187 / pp.22
Cahier technique no 191

Fuzzy logic
F. Chevrie F. Guly
no 191
Fuzzy logic
Franois CHEVRIE After joining Telemecanique in 1987, he joint the Advanced Automation Laboratory of the Research Division in 1993. A CNAM Industrial Automation engineering graduate, his dissertation was based on the integration of fuzzy logic in Schneider programmable controllers. He played an active part in the preparation of the fuzzy logic product offer for the Micro/Premium PC range, and helped implement this technique, particularly in the car and food industries.
Franois GUELY After graduating from the Ecole Centrale de Paris in 1988, he joined Telemecanique in Japan in 1990 and was awarded his PhD in fuzzy logic based automatic control in 1994. He has been in charge of Schneiders Advanced Automatic Department since 1995 where he has helped prepare the extension to fuzzy logic of the IEC language standard for programmable controllers.
ECT 191 first issued, december 1998

Cahier Technique Schneider no 191 / pp.1
Lexicon
Activation: See degree of truth. Conclusion: A rule conclusion is a statement combining a linguistic variable and a linguistic term written after the then of the rule. A conclusion can be made up of a combination of several statements. Condition: See predicate. Data merge: Data merge consists of extracting, from several pieces of data, one or more items of information which may be different kinds. For example: from variables R, V and B giving the colour of a biscuit, the cooking state of the biscuit can be deduced. The term Sensor merge is also used. Defuzzification: Conversion, after inference, of a fuzzy set of a linguistic output variable into a numerical value. Degree of activation: See degree of truth. Degree of membership: An element x belongs to a fuzzy set A with a degree of membership between 0 and 1, given by the membership function A (x). Degree of truth: The degree of truth, or degree of activation, of a rule is a value y between 0 and 1 deduced from the degrees of membership of the rule predicates. It directly affects the value of the conclusions of this rule. The rule is also said to be active at y. Fuzzification: Conversion of a numerical value into a fuzzy degree of membership by evaluating a membership function. Fuzzy set: In the classical set theory, the characteristic function defines the set: this function only takes
the two discrete values 0 (the element does not belong...) or 1 (...belongs to the set). A fuzzy set is defined by a membership function which can take any real values between 0 and 1. Inference: Calculation of the degrees of activation of all the rules in the base as well as of all the fuzzy sets of the linguistic variables contained in the conclusions of these rules. Knowledge base: Set of membership functions and rules of a fuzzy system containing expertise, knowledge of the operator, expert, etc. Linguistic term: Term associated with a membership function characterising a linguistic variable. Linguistic variable: Numerical variable with a name (pressure, temperature to which are associated inguistic terms. Membership function: Function A (x) associating to any input value x its degree of membership to the set A. This gradual value belongs to the [0; 1] interval. Predicate: Also known as premise or condition, a rule predicate is a statement combining a linguistic variable and a linguistic term written between the if and the then of the rule. A predicate can be made up of a combination of several statements linked by AND, OR, NOT operators. Premise: See Predicate. Sensor merge: See Data merge. Singleton: Membership function A (x), equals to zero for all x, except at a singular point x0.
Cahier Technique Schneider n o 191 / pp.2
Fuzzy logic
Initially a theory, today fuzzy logic has become an operational technique. Used alongside other advanced control techniques, it is making a discrete but appreciated appearance in industrial control automation systems. Fuzzy logic does not necessarily replace conventional control systems. Rather it completes such systems. Its advantages stem from its ability to: c formalise and simulate the expertise of an operator or designer in process control and tuning, c provide a simple answer for processes which are difficult to model, c continually take into account cases or exceptions of different kinds, and progressively incorporate them into the expertise, c take into account several variables and perform weighted merging of influencing into variables. How does this technique contribute to industrial process control? What is the effect on product quality and manufacturing cost? Following a few basic theoretical notions, this Cahier Technique answers the questions asked by automatic control engineers and potential users by means of industrial examples, in terms of implementation and competitive advantages.
Contents
1 Introduction 1.1 Fuzzy logic today 1.2 The history of fuzzy logic 1.3 Value and use of fuzzy logic for control 2 Theory of fuzzy sets 2.1 Notion of partial membership 2.2 Membership functions 2.3 Fuzzy logic operators 2.4 Fuzzy rules 3 A teaching application example 3.1 Introduction 3.2 Presentation of the example 3.3 Linguistic variables and terms 3.4 Rules and outputs 4 Implementation 4.1 when can fuzzy rule bases be used? 4.2 4.3 4.4 4.5 5 Fuzzy application 6 Conclusion Appendix Bibliography Designing an application Using an application Choosing the implementation technology Standards pp. 4 pp. 4 pp. 5 pp. 6 pp. 6 pp. 8 pp. 9 pp. 14 pp. 14 pp. 15 pp. 15 pp. 16 pp. 16 pp. 17 pp. 17 pp. 18 pp. 19 pp. 20 pp. 24 pp. 26 pp. 28
5.1 Application types 5.2 Examples of industrial achievements
1 Introduction
1.1 Fuzzy logic today

In the majority of present-day applications, fuzzy logic allows many kinds of designer and operator qualitative knowledge in system automation to be taken into account. Fuzzy logic began to interest the media at the beginning of the nineties. The numerous applications in electrical and electronic household appliances, particularly in Japan, were mainly responsible for such interest. Washing machines not requiring adjustment, camcorders with Steadyshot (TM) image stabilization and many other innovations brought the term fuzzy logic to the attention of a wide public. In the car industry, automatic gear changes, injection and anti-rattle controls and air conditioning can be optimized thanks to fuzzy logic. In continuous and batch production processes, as well as in automation systems (which is the subject of this Cahier Technique), applications have also increased. Fuzzy logic has developed in this area as it is an essentially pragmatic, effective and generic approach. It allows systematisation of empirical knowledge and which is thus hard to control. The theory of fuzzy sets offers a suitable method that is easy to implement in real time applications, and enables knowledge of designers and operators to be transcribed into dynamic control systems. This makes fuzzy logic able to tackle automation of procedures such as startup and setting of parameters, for which few approaches were previously available. This Cahier Technique describes fuzzy logic and its application to production processes.
1.2 The history of fuzzy logic

Appearance of fuzzy logic The term fuzzy set first appeared in 1965 when professor Lotfi A. Zadeh from the university of Berkeley, USA, published a paper entitled Fuzzy sets. Since then he has achieved many major theoretical breakthroughs in this field and has been quickly joined by numerous research workers developing theoretical works. Initial applications At the same time, some researchers turned their attention to the resolution by fuzzy logic of problems considered to be difficult. In 1975 professor Mamdani from London developed a strategy for process control and published the encouraging results he had obtained for the control of a steam motor. In 1978 the Danish company, F.L. Smidth, achieved the control of a cement kiln. This was the first genuine industrial application of fuzzy logic. Boom Fuzzy logic experienced a veritable boom in Japan where research was not only theoretical but also highly application oriented. At the end of the eighties fuzzy logic had taken off in a big way, and consumer products such as washing machines, cameras and camcorders with the mention fuzzy logic were too numerous to be counted. Industrial applications such as treatment of water, harbour container cranes, undergrounds and ventilation/air conditioning systems began to use fuzzy logic too. Finally, applications developed in such other fields such as finance and medical diagnosis. From 1990 onwards, many applications began to emerge in large numbers in Germany, as well as, to a lesser extent, in the USA.
1.3 Value and use of fuzzy logic for control

Value Fuzzy logic stems from several observations, namely: c The knowledge that a human being has of any situation is generally imperfect, v it can be uncertain (he doubts its validity), v or imprecise. c Human beings often solve complex problems with approximate data: accuracy of data is often useless; for example, in order to choose an apartment he may take into account surface area, proximity of shops, distance from the workplace and rent without, however, needing a very precise value for each piece of information. c In industry and technology, operators frequently solve complex problems in a relatively simple manner without needing to model the system. Likewise, it is common knowledge that a mathematical model is not required to drive a car, and yet a car is a highly complex system. c The more complex a system, the more difficult it is to make precise assertions on its behaviour. The following are naturally deduced from these observations: c rather than modelling the system, it is often more useful to model the behaviour of a human operator used to control the system; c rather than using equations, operation can be described by qualitatively with an appropriate quantitative translation. Use for control purposes Fuzzy logic is well known by automatic control engineers for its applications in process control and monitoring, then commonly referred to as fuzzy control. Just like a conventional controller, the fuzzy controller is incorporated in the control loop and computes the control to be applied to the process according to one or more setpoints and one or more measurements taken on the process. Fuzzy rule bases are advantageous in control as they allow: c consideration of existing qualitative expertise, c consideration of variables the effect of which would be difficult to model with traditional means, but is known in a qualitive way, c improvement of conventional controller operation by: v self-tuning of controller gains off line or on line, v modification of their output (feed forward) according to events that cannot be taken into account using a conventional technique. Using knowhow to its best advantage A vital condition for the use of fuzzy rules is the existence of human expertise and knowhow. Fuzzy rule bases cannot provide a solution when no-one knows how the system operates or people are unable to manually control it. When such knowhow exists and can be transcribed in the form of fuzzy rules, fuzzy logic simplifies its implementation, and operation is then easily understood by the user. Fuzzy logic also enables maximum benefit to be derived from practical knowhow, often sought for in order to prevent loss of knowhow or to share this knowhow with other people in the company. When collecting expertise, unconscious omission of information, the difficulty to explain and the fear to disclose knowhow are obstacles that are frequently encountered. This stage must therefore be prepared and conducted with care, taking into account the human factor. If human expertise exists, then fuzzy rules can be used, particularly when system knowledge is tainted by imperfections, when the system is complex and hard to model and when the method used requires a global view of some of its aspects. Fuzzy rules do not replace conventional automatic control methods, rather they complete these methods.
2 Theory of fuzzy sets
2.1 Notion of partial membership

In the sets theory, an element either belongs or does not belong to a set. The notion of a set is used in many mathematical theories. This essential notion, however, does not take into account situations which are yet both simple and common. Speaking of fruits, it is easy to define the set of apples. However, it is harder to define the set of ripe apples. We understand that an apple ripens progressively... the notion of a ripe apple is thus a gradual one. The notion of a fuzzy set was created in order to take situations of this kind into account. The theory of fuzzy sets is based on the notion of partial membership: each element belongs partially or gradually to the fuzzy sets that have been defined. The outlines of each fuzzy set (see fig.1 ) are not crisp, but fuzzy or gradual.
B
t z x
Fuzzy or gradual outline
Crisp outline A: conventional set B: fuzzy set
x belongs neither to A nor B y belongs completely to A z belongs completely to B t belongs partially to B
Fig. 1 : comparison of a conventional set and a fuzzy set.
2.2 Membership functions

A fuzzy set is defined by its membership function which corresponds to the notion of a characteristic function in classical logic. Let us assume that we want to define the set of people of medium height. In classical logic, we would agree for example that people of medium height are those between 1.60 m and 1.80 m tall. The characteristic function of the set (see fig. 2 ) gives 0 for heights outside the range [1.60 m; 1.80 m] and 1 for heights in that range. The fuzzy set of people of medium height will be defined by a membership function which differs from a characteristic function in that it can assume any value in the range [0;1]. Each possible height will be assigned a degree of membership to the fuzzy set of medium heights (see fig. 3 ) between 0 and 1.
Degree of membership 1 Characteristic function medium height 0 1m60 1m80 Variable: height
Degree of membership 1 Characteristic function medium height 0 1m72 Variable: height
Fig. 2 : characteristic function.
Fig. 3 : membership function.
A number of fuzzy sets can be defined on the same variable, for example the sets small height, medium height and tall height, each notion being explained by a membership function (see fig. 4 ).
c they are simple, c they contain points allowing definition of areas where the notion is true and areas where it is false, thereby simplifying the gathering of expertise. We have chosen to use membership functions of this kind in the rest of this document.
Small 1 0.7 Medium Tall
0.3 0 1.60 1.80 2 Height (m)
In some cases, membership functions may be equal to 1 for a single value of the variable, and equal to 0 elsewhere. They are then known as singleton membership functions. A fuzzy singleton (see fig. 6 ) defined on a real variable (height) is the expression in the fuzzy field of a specific value (Pauls height) of this variable (see appendix).
Fig. 4 : membership function, variable and linguistic term.

1
This example shows the graduality that enables fuzzy logic to be introduced. A 1.80 m tall person belongs to the tall set with a degree of 0.3, and to the set medium height with a degree of 0.7. In classical logic, the change from average to tall would be sudden. A 1.80 m person would then be of medium height, whereas a 1.81 m person would be tall, an assertion which shocks intuition. The variable (for example: height) as well as the terms (for example: medium, tall) defined by the membership functions, are known as linguistic variable and linguistic term respectively. As we shall see further on, both linguistic variables and terms can be used directly in rules. Membership functions can assume any shape. However they are often defined by straight segments and said to be piece-wise linear (see fig. 5 ). Piece-wise linear membership functions are frequently used as:
0 1.78 m Paul's height
Fig. 6 : singleton membership function.
Fuzzification - Degree of membership Fuzzification enables a real value to be converted into a fuzzy one. It consists of determining the degree of membership of a value (measured by example) to a fuzzy set. For example (see fig. 7 ), if the current value of the input variable is 2, the degree of membership to the low input membership function is equal to 0.4 which is the result of the fuzzification. We can also say that the low input proposal is true at 0.4. We then talk of degree of truth of the proposal. Degree of membership and degree of truth are therefore similar notions.
Totally medium height
Small Medium Tall 1 Height Small not at all medium Tall not at all medium 0.4 0 2 Input Low
Fig. 5 : piece-wise linear membership functions.
Fig. 7 : fuzzification.
2.3. Fuzzy logic operators

These operators are used to write logic combinations between fuzzy notions, i.e. to perform computations on degrees of truth. Just as for classical logic, AND, OR and NOT operators can be defined. For example: Interesting Apartment = Reasonable Rent AND Sufficient Surface Area. Choice of operators These operators have many variants (see appendix). However the most common are the Zadeh operators described below. The degree of truth of a proposal A will be noted (A). Intersection The logic operator corresponding to the intersection of sets is AND. The degree of truth of the proposal A AND B is the minimum value of the degrees of truth of A and B: NB: this fuzzy OR is compatible with classical logic: 0 OR 1 yields 1. Complement The logic operator corresponding to the complement of a set is the negation.
(NOT A) = 1 - (A)
For example: Low Temperature is true at 0.7 NOT Low Temperature that we will normally write as Temperature NOT Low is therefore true at 0.3. NB: the negation operator is compatible with classical logic: NOT(0) yields 1 and NOT(1) yields 0. Fuzzy ladder Ladder language or contact language is commonly used by automatic control engineers to write logic combinations, as it enables their graphic representation. Schneider has introduced the use of ladder representation to describe fuzzy logic combinations. Below is an example dealing with the comfort of ambient air: hot, damp air is uncomfortable (excessive perspiration); likewise breathing is difficult in air that is cold and too dry. The most comfortable thermal situations are those in which air is hot and dry, or cold and damp. This can be transcribed by the fuzzy ladder in figure 8 corresponding to the following combination: Good comfort = (Low Temperature AND High Humidity) OR (High Temperature AND Low Humidity). It represents a possible definition of the sensation of comfort felt by a person in a thermal environment in which air does not move.
(A AND B) = MIN((A),(B))
For example: Low Temperature is true at 0.7 Low Pressure is true at 0.5 Low Temperature AND Low Pressure is therefore true at 0.5 = MIN(0.7; 0.5). NB: this fuzzy AND is compatible with classical logic 0 and 1, yelds 0. Union The logic operator corresponding to the union of sets is OR. The degree of truth of the proposal A OR B is the maximum value of the degrees of truth of A and B:
(A OR B) = MAX((A),(B))
For example: Low Temperature is true at 0.7 Low Pressure is true at 0.5 Low Temperature OR Low Pressure is therefore true at 0.7.
Low High
Low High Low temperature High humidity Good comfort
10
20 Temperature
30
50 Humidity
100
High temperature
Low humidity
Fig. 8 : fuzzy ladder.
Fuzzy classification Classification normally consists of two steps: c preparation: determining the classes to be considered, c on line: assigning the elements to classes. The notions of class and set are identical theoretically. There are three types of assignment methods according to the result produced: c boolean: the elements either belong or do not belong to the classes, c probabilistic: the elements have a probability of belonging to boolean classes, such as for example the probability that a patient has measles given the symptoms that he shows (diagnosis), c gradual: the elements have a degree of membership to the sets; for example a lettuce
belongs to a varying degree to the class of fresh lettuces. Classification methods, whether they produce a gradual, boolean or probabilistic result, can be developed from: c an experiment (case of fuzzy ladder mentioned above), c examples used for learning purposes (e.g. for neuron network classifiers), c mathematical or physical knowledge of a problem (for example, the comfort of a thermal situation can be evaluated from thermal balance equations). Gradual (or fuzzy) classification methods can be used in control loops. This is the case of the industrial cooking example for biscuits described later on.
2.4. Fuzzy rules

Fuzzy logic and artificial intelligence The purpose of fuzzy rule bases is to formalise and implement a human beings method of reasoning. As such it can be classed in the field of artifical intelligence. The tool most commonly used in fuzzy logic applications is the fuzzy rule base. A fuzzy rule base is made of rules which are normally used in parallel but which can also be concatenated in some applications. A rule is of the type: IF predicate THEN conclusion. For example: IF high temperature and high pressure THEN strong ventilation and wide open valve. Fuzzy rule bases, just like conventional expert systems, rely on a knowledge base derived from human expertise. Nevertheless, there are major differences in the characteristics and processing of this knowledge (see fig. 9 ). A fuzzy rule comprises three unctional parts summarised in figure 10 .
Inputs Outputs
Fuzzification
Inferences
Defuzzification
Numerical values
Fuzzy area
Numerical values
Fig. 10 : fuzzy processing.
Predicate A predicate (also known as a premise or condition) is a combination of proposals by AND, OR, NOT operators. The high temperature and high pressure proposals in the previous example are combined by the AND operator to form the predicate of the rule.
Fuzzy rule base Few rules Gradual processing Concatenation possible but scarcely used
Conventional rule base (expert system) Many rules Boolean processing Concatenated rules A OR B C, C D, D AND A E Rules used one by one, sequentially No interpolation, no contradiction
Rules processed in parallel Interpolation between rules that may contradict one another
Fig. 9 : fuzzy rule base and conventional rule base.
Inference The most commonly used inference mechanism is the Mamdani one. It represents a simplification of the more general mechanism based on fuzzy implication and the generalised modus ponens. These concepts are explained in the appendix. Only the Mamdani rule bases are used below. Conclusion The conclusion of a fuzzy rule is a combination of proposals linked by AND operators. In the previous example, strong ventilation and wide open valve are the conclusion of the rule. OR clauses are not used in conclusions as they would introduce an uncertainty into the knowledge (the expertise would not make it possible to determine which decision should be made). This uncertainty is not taken into account by the Mamdani inference mechanism which only manages imprecisions. Therefore the Mamdani fuzzy rules are not in theory suitable for a diagnosis of the medical kind for which
conclusions are uncertain. The theory of possibilities, invented by Lotfi Zadeh, offers an appropriate methodology in such cases. Likewise, negation is not used in conclusions for Mamdani rules. This is because if a rule were to have the conclusion Then ventilation not average, it would be impossible to say whether this means weak ventilation or strong ventilation. This would be yet another case of uncertainty. Mamdani inference mechanism c Principle A Mamdani fuzzy rule base therefore contains linguistic rules using membership functions to describe the concepts used (see fig. 11 ). The inference mechanism is made up of the following steps: c Fuzzification Fuzzification consists of evaluating the membership functions used in rule predicates, as is illustrated in figure 12 :
IF high pressure
AND
high temp.
THEN
valve wide open
High
High
Wide
Pressure
Temperature
Valve opening
IF average pressure
AND high temp.
THEN
average valve opening
Average
High
Average
Pressure
Temperature
Valve opening
Fig. 11 : implication.
IF high pressure
AND
high temp.
THEN
valve wide open
0.5
High 0.3 2.5 bar Pressure
High
Wide
17C Temperature
Valve opening
Fig. 12 : fuzzification.
c Degree of activation The degree of activation of a rule is the evaluation of the predicate of each rule by logic combination of the predicate proposals
(see section 2.3.), as shown in figure 13 . The AND is performed by realising the minimum between the degrees of truth of the proposals.
IF high pressure
AND
high temp.
THEN
valve wide open
0.5 0.3 2.5 bar Pressure
Min
} = 0.3
17C Temperature
High
Valve opening
Fig. 13 : activation.
c Implication The degree of activation of the rule is used to determine the conclusion of the rule: this operation is called the implication. There are several implication operators (see appendix), but the most common is the minimum operator. The conclusion fuzzy set is built by realising the minimum between the degree of activation and the membership function, a sort of clipping of the conclusion membership function (see fig. 14 ).
IF high pressure
AND
high temp.
THEN
valve wide open
Wide
0.5 0.3 2.5 bar Pressure
} = 0.3
17C Temperature Valve opening
Min
Fig. 14 : implication.
c Aggregation The output global fuzzy set is built by aggregation of the fuzzy sets obtained by each rule concerning this output. The example below shows the case when two rules act on an
output. The rules are considered to be linked by a logic OR, and we therefore calculate the maximum value between the resulting membership functions for each rule (see fig. 15 ).
IF high pressure
AND
high temp.
THEN
valve wide open
High
High
Wide
0.3 2.5 bar Pressure 17C Temperature Valve opening
IF average pressure
AND
high temp.
THEN
valve wide open
Average
High
Average
2.5 bar Pressure
17C Temperature
Valve opening
Aggregation: MAXIMUM
Valve opening
Fig. 15 : aggregation of rules.
Defuzzification At the end of inference, the output fuzzy set is determined, but cannot be directly used to provide the operator with precise information or control an actuator. We need to move from the fuzzy world to the real world: this is known as defuzzification. A number of methods can be used, the most common of which is calculation of the centre of gravity of the fuzzy set (see fig. 16 ).
Free and able rules Fuzzy rule bases, in their general case, use membership functions on system variables, and rules that can be written textually. Each rule uses its own inputs and outputs, as shown by the example below: R1: R2: IF high temperature THEN high output IF average temperature AND low pressure THEN average output IF average temperature AND high pressure THEN low output IF low temperature AND high pressure THEN very low output
x(x)dx (x)dx
35.6 Valve opening
R3:
R4:
Fig. 16 : defuzzification by centre of gravity.
In diagram form, the areas of action of the rules and their overlapping can be represented in the table in figure 17 .
Pressure
High
Very low output
Low output High output Average output
it does not interest us. It is best to verify it as this may be an omission; c the first rule only takes temperature into account: this situation is normal in that it reflects the existing expertise. However, many applications define rule tables. In this context, the space is gridded and each box in the grid is assigned a rule. This approach has the advantage of being systematic, but: c it does not always allow simple expression (in a minimum number of rules) of the existing expertise, c it can be applied only for two or three inputs, whereas free rule bases can be built with a large number of variables. Remarks c The behaviour of a fuzzy rule base is static and non-linear with respect to its inputs. c Fuzzy rule bases are not themselves dynamic, although they often use as inputs variables expressing system dynamics (derivatives, integrals, etc. ) or time. c The main advantage of the fuzzy PID controller, often presented as a teaching example to give an idea of fuzzy logic, is to make a non-linear PID, which rarely justifies its use in the place of a conventional PID. Moreover it would be hard to incorporate an existing expertise in this case.
Low
Low
Average
High
Temp.
Fig. 17 : implication represented in a table.
We can make the following observations: c not all the space is necessarily covered: the combination low temperature and low pressure is not taken into account in this case. The explanation is for example that this combination is not physically possible for this machine or that
3 A teaching application example
3.1 Introduction
Most fuzzy logic achievements require preliminary specialist knowledge of the application area. In order to be easily understood by the reader, the following example is based on a fictitious application and is designed to illustrate the procedure for creating a fuzzy rule base.
3.2 Presentation of the example

The example concerns a process for washing lettuce for the production of prepacked lettuce in the fresh produce counters of supermarkets. The lettuce is cut, washed and packed. The purpose of washing is to remove earth from the lettuce as well as any micro-organisms which could proliferate during product shelf-life. The manufacturer wishes to automate the washing process. Washing is a continuous process. The lettuce leaves are placed in drums which move through a tunnel fitted with nozzles spraying chlorinated water. The water removes the earth, whereas the chlorine kills the micro-organisms (see fig. 18 ). The following priorities were formulated by the marketing department and listed in the order of their importance: c With respect to the customer v Guarantee quality - Very clean lettuce (appearance) - No taste of chlorine. v Guarantee safety - Acceptable level of micro-organisms c With respect to profitability v Maximise production v Save water v Save chlorine. The operators manually controlling the process usually inspect the dirty water at the end of the tunnel washing. If the water is clear, they deduce by experience that the lettuce will have a clean appearance. The decision is thus made to install an optic turbidity sensor designed to determine the degree of transparency of the water. Moreover, operators use once an hour a report based on analysis conducted in the factory which gives the ratio of micro-organisms and residual chlorine found in washed and prewashed lettuce at the end of the line. The aim is therefore to use the above information to improve control of: c lettuce conveyor belt speed (in order to increase production output), c the amount of chlorine sprayed, c the amount of water sprayed. Limits are imposed: c on conveyor belt speed, by the mechanism, c on water flow to prevent damaging the lettuce leaves.
Water flow
Chlorine flow Tunnel
Drum Measurement off line of: - chlorine ratio - micro-organism ratio
Belt speed
Waste water after washing
Turbidity measurement
Fig. 18 : lettuce washing process.
3.3. Linguistic variables and terms

The following variables will therefore be chosen: c inputs: v micro-organism ratio: Micro_ratio v residual chlorine ratio: Cl_ratio v turbidity of water: Turbidity v conveyor belt speed: Speed v water flow: Water_f c v v v outputs: modification of water flow: Water_f_var modification of chlorine flow: Cl_f_var modification of speed: Speed-var
A session with an experienced operator, a microbiology specialist and a lettuce taster produces the following membership functions (see fig. 19 ):
Negative Acceptable High Cl_ratio
Positive Positive big Water_f_var
Negative Low High Turbidity
Positive Positive big CI_f_var
Low
Acceptable High TMicro_ratio
Negative Positive
Speed_var
Not high High Water_f
Not high High Speed
Fig. 19 : piece-wise linear membership functions.
3.4. Rules and outputs

Writing fuzzy rules A meeting with operators enables the seven rules below to be determined, each corresponding to a specific case: c Lettuce badly washed IF Turbidity IS High AND Water_f IS NOT High THEN Water_f_var IS Positive big. c Lettuce badly washed but high conveyor belt speed IF Turbidity IS High AND Water_f IS High THEN Speed_var IS Negative. c Too many micro-organisms IF Micro_ratio IS High THEN Cl_f_var IS Positive big. c Everything is fine and production can be increased IF Turbidity IS Low and Micro_ratio IS NOT High AND Speed IS NOT High and CL_ratio IS Acceptable AND Water_f IS NOT High THEN Speed_var IS Positive AND Cl_f_var IS Positive AND Water_f_var IS Positive. c Lettuce tastes of chlorine, but there are no micro-organisms IF Cl_ratio IS High AND Micro_ratio IS NOT High THEN Cl_f_var IS Negative. c Everything is fine and production is maximum: save water. IF Speed IS High AND Cl_ratio IS Acceptable AND Turbidity IS Low THEN Water_f_var IS Negative. c No micro-organisms: save chlorine If Micro_ratio IS Low THEN Cl_f_var IS Negative. Defuzzification Insofar as the aim is progressive behaviour of the rule base in all cases and an interpolation between the rules, the centre of gravity is chosen as the defuzzification operator.
4 Implementation
4.1 When can fuzzy rule bases be used?

Fuzzy rule bases can be chosen to solve application problems when the following conditions are satisfied: c it is possible to act on the process (controllability), c an expertise or knowhow exists, c the variables (inputs and outputs) can be measured or observed, (measurability), c qualitative expertise (if it is mathematical, conventional automatic control should be preferred), c gradual expertise (if it is boolean, expert systems are more suitable).
4.2 Designing an application

Choice of operators In most applications, Mamdani rule bases are used. This choice is suitable except if the expertise contains indeterminations. In most cases, the choice is also made to use trapezoidal membership functions as they are easier to implement and simplify the gathering of expertise. Output membership functions are often singletons, except when rules are concatenated. A triangular output membership function in fact implies an uncertainty on the output to be applied, and does not have much effect on interpolation between the rules. Finally, defuzzification takes place using the centre of gravity for control (all active rules are taken into account): the use of the average of maxima for decision-making problems enables a decision to be made when rules are conflicting and avoids intermediate decisions. Methodology Designing a fuzzy rule base is an interactive process. The largest portion of the task consists of collecting knowledge. One of the advantages of fuzzy logic is the possibility of having the rule base validated by the people who provided the expertise before testing it on a real system. Figure 20 illustrates the procedure used. Collecting knowledge This is a three-step process: c listing the variables to be taken into account: they will become the linguistic variables of the rule base; c listing the qualitative quantities to be taken into account and specifying when they are true and false: these quantities will become the linguistic terms of the rule base; c formulate how these concepts are manipulated: which cases should be considered, how they are characterised, how should you act in each case.
Professional expertise level: - Expert - Operator - Designer
Gathering knowledge Validation of principle Validation of operation
Programming level: - Automatic control engineer - Ladder / Grafcet
Interpretation in form of rules and membership functions Open loop tests
Implementation
Fig. 20 : design methodology.
Transcription in fuzzy rule form is then straight forward. However as few membership functions and rules as possible should be written in order to limit the number of parameters which will have to be tuned later on and to ensure legibility of the base. We observe that it is easier to add rules in order to take new situations into account than to remove them. Validating the knowledge base This takes place in a number of steps: c presentation of the rule base to the experts who helped collect knowledge, and discussion. The aim of the discussion is to identify points that have not been covered and to ensure that the rules are understood by everyone; c open loop simulation: the experts compare the behaviour of the rule base to the behaviour that they expect on cases chosen beforehand;
c if the process can be simulated, closed loop simulations can also be performed. Tuning The rule bases written in this manner often give satisfaction right away. However the rule base may need to be modified or tuned. The following principles will act as a guideline in searching for the probable cause of the deviation observed: c if the behaviour of the closed loop controller is the opposite to what you expected, some rules have most likely been incorrectly written; c if you wish to optimise performance, it is normally preferable to properly tune the membership functions; c if the system is not robust and works in some cases but not all the time, it is likely that not all cases have been taken into account and that rules must be added.
4.3 Using an application

The function of the operators The degree of involvement of operators controlling an application using fuzzy logic varies considerably. The following cases can be observed: c completely autonomous system: the end-user is not familiar with fuzzy logic and is not aware of its use, c fuzzy logic is a black box which can be disconnected or changed to manual mode by the operator, c the operator is able to modify (tune) the membership functions according to the situation, and he does this for a production change (for example); c the operator is able to read the rules (e.g. their degree of activation): he understands and is able to interpret the actions of the rule base. For example he can control the rule base in exceptional situations; c the operator is the main designer of the base: he has been given the means to record his own knowhow and to validate the resulting behaviour. Production changes During an application, the rule base must be able to be adapted to changes in the production system and the products manufactured. These changes can be of various kinds: c objectives have changed (cooking temperature, etc.), for example due to a change in product manufactured. The setpoints or rule input membership functions must then be modified; c system dimensions have changed; the membership functions must then be modified; c the type of system has changed (e.g. portage of the rule base from one machine to another); the rules and membership functions must then be modified. The most common changes are of the first type. They can then be managed by qualified operators.
4.4 Choosing the implementation technology

Most of todays applications run on standard hardware platforms (micro-controller, microprocessor, programmable controller, microcomputer, etc.). Many software programs designed to help develop fuzzy rule bases and aimed at microcontrollers, programmable controllers and microcomputers (to name but a few), enable rapid implementation of fuzzy rule bases without programming. Fuzzy inferences can be directly programmed (assembler, C language, etc.). The disadvantage of this solution is that it is slower in the prototype phase and requires programming skills and command of fuzzy logic algorithms.
For applications with exacting response time demands or in order to obtain very low mass production cost prices, use of fuzzy logic ICs is advantageous. Use of such electronic chips is increasing as: c the operations required to produce fuzzy inferences are elementary and feasible in integers, c some operations can be carried out in parallel, c the calculation takes place in successive steps, thereby enabling simple pipeline architectures to be made. In particular, numerous ASIC components designed for specific markets exist (car, electrical household appliances, etc.). They are
now commonly integrated inside microcontrollers, even low cost ones, where they are used to accelerate fuzzy inferences. Figure 21 shows as an example the applicational needs that can be encountered in number of rules (complexity of the application) and cycle time (rapidity) as well as the possible technologies (1993 figures). The rules considered have one predicate and one conclusion. The necessary technical-economic choice is thus a compromise between the flexibility provided by software solutions, scale economy and the performance of dedicated hardware solutions.
Cycle time (s) 10 10

-7
-6
10-5 10 10 10
-4
RISC 32 bits 16 bits 8 bits 4 bits
Image processing Control system, car
-3
-2
10-1 1 10 102 10
3
Cameras Control Washing machines Financial analyses Medical diagnosis Number of rules
10
100
1 000
10 000
Micro-programming technology
ASIC technology
Analog technology
Fig. 21 : performance of components and application areas.
4.5 Standards
Components Absence of standards is one of the main problems holding up the use of fuzzy logic chips. This is because these components are not compatible with one another as each one is the result of choices made by manufacturers. Software Regarding software, lack of portability has also slowed down widespread use of fuzzy logic in industry. Today, a work group in which Schneider plays an active part, has incorporated the fuzzy logic language standard into the language standard of programmable controllers (first official draft of standard IEC 61131-7 available in 1997). Other initatives in the field of fuzzy logic standardisation should spring from this.
5 Fuzzy applications
5.1 Application types

Functions performed The following table shows the functions most often performed in industry by means of fuzzy systems (X means possible use, XX that the technique is suitable for this type of problem). Rule bases excel in cases when interpolation and action are required, whereas classification methods are suitable for evaluation and diagnosis tasks normally performed upstream. Applications sometimes combine several of these functions, while retaining the graduality of the information.
Rule bases Regulation, control Automatic parameter setting Decision-making help Diagnosis Quality control XX XX Classification algorithms Neuron network Data Conventional automatic control Model
Fuzzy logic
Theory of possibilities
Probabilities
Imprecision and graduality
Uncertainty
Expertise Fuzzy rules
XX X
X XX XX
Fig. 22 : comparing fuzzy logic with other control techniques.
fuzzy logic may be preferred for the ease with which it is understood by operators. Hybridation of techniques Fuzzy logic is often used in combination with other techniques. These combinations are advantageous when each approach make use of its own strong points. c Learning fuzzy rules or neurofuzzy Fuzzy rule bases can be modified using learning methods. The first methods known as self-organizing controller were developed as early as 1974 and aimed at heuristically modifying the content of fuzzy rules belonging to a rules table. The actual expertise is modified by the learning, but the membership functions remain the same. A second approach, consists of modifying parameters representative of the membership functions. Unlike the first method, the rules and structure of the expertise are not altered. The membership function parameters are modified using optimisation methods, for example gradient methods, or global optimisation methods such as genetic algorithms or simulated annealing. This approach is often referred to as
Fuzzy logic and other techniques Fuzzy logic is above all an extension and a generalisation of boolean logic. It enables graduality to be introduced into notions which were previously either true or false. Probabilities, without challenging the binary nature of events (either true or false) enable the uncertainty of these events to be managed. On the boundary between these two approaches, the theory of possibilities (invented by Lotfi Zadeh) enables both graduality and uncertainty to be taken into account (see fig. 22 ). Fuzzy base rules are often compared for control/ regulation applications to neuron networks and conventional automatic control. These three approaches require respectively, in order to be applied, an expertise, data for learning purposes, and a dynamic model of the process. These approaches can only be compared when all three are available at the same time, which is often the case in theoretical studies but rare in practice. If all three are available, practical considerations often take priority. In particular,
neurofuzzy, in particular when the gradient is used. Use of the gradient to optimise these parameters is likened to retropropagation used in neuron networks known as multi-layer perceptrons in order to optimise weights between neuron network layers. A third approach (that can be qualified as structural optimisation of the rule base) aims at simultaneously determining rules and membership functions by learning. The learning process then normally takes place without referring to an expertise. The resulting rules can then theoretically be used to help build an expertise.
c Using fuzzy logic in association with automatic control A fuzzy rule base is sometimes part of a controller. Use of fuzzy logic to simulate a proportional term allows all kinds of nonlinearities. Specific cases of downgraded operation such as overloads, maintenance or partial failures are easily integrated. A fuzzy rule base is used to greater advantage outside the control loop, to supervise a controller. It then replaces an operator in order to tune controller parameters according to control system operating conditions.
5.2 Examples of industrial achievements

Today fuzzy logic is accepted as being one of the methods commonly used to control industrial processes. Although PID controllers still suffice for most applications, fuzzy logic is increasingly recognised and used for its differentiating advantages, particularly for controlling quality of production and costs. Due to the competitive advantages offered by fuzzy logic in some applications, the integrator or end-user do not normally wish to mention the subject. These applications benefit from extensive acquisition of knowhow or use of a crafty technical short-cut. Confidentiality is then essential. This explains why it was not possible to describe in a detailed way all the examples given below. Sewage plant Most modern sewage plants use biological processes (development of bacteria in ventilated tanks) to purify sewage water before discharging it into the natural environment. The organic matter contained in the waste water is used by the bacteria to create its cellular components. The bacteria discharges carbon dioxide (CO2) and nitrogen (N2). Air is blown into the tanks. The energy used for ventilation purposes frequently accounts for more than half the global energy consumed by the plant. In order to ensure correct development of bacteria and sewage, the NH4 and O2 concentrations in the ventilation tanks must be carefully controlled, all the more so since in order to reduce energy costs, air flow is kept to a minimum compatible with the biological process. Added to these requirements is the consideration of some specific operating cases, such as for example a very high upstream flow, which is an extreme circumstance under which parameters are seriously modified and sewage capacity affected. Although partial mathematical models of plants are available, there are no complete models, and the overall control strategy must often be heuristically developed. Use of fuzzy logic is relatively common nowadays in sewage plants. The plant shown in figure 23, based in Germany, has been in operation since 1994. Fuzzy logic was produced on a Schneider Modicon programmable controller by means of its standard fuzzy control functional modules. The designer highlights the advantage of using fuzzy logic in control: exceptions, i.e. situations when sewage capacity is partially downgraded, are treated simply and without discontinuity. The method chosen to introduce these exceptional states into a control loop is described below: A proportional term which must adapt to the exceptional circumstances is identified in the control loop: this term is first transcribed in fuzzy logic, then this fuzzy logic element is inserted in the control loop.
Blower
Precipitant tanks for phosphates Recirculation
Control station and operating building
Grid building 10 11 4 7 6 5 3 2 1
8 9 Outlet
1 - Sewage water supply 2 - Inlet mechanism lifting 3 - Ventilated sediment removal basin 4 - Venturi drain 5 - Excess sludge 6 - Recycled sludge 7 - Sludge scraper 8 - Final purification I 9 - Final purification II 10 - Nitrification channels 11 - Denitrification basin
Fig. 23 : block diagram of the sewage plant.
Once the membership functions have been suitably tuned, two rules are sufficient to describe the proportional controller: IF low input THEN low output. IF high input THEN high output. A third rule is added at the operators request as they find it improves their understanding of the operation:
IF average input THEN average output (see fig. 24 ). Once the proportional term has been simulated, the exceptions are introduced in the form of other rules depending on other input variable combinations. A simple example of this possibility is illustrated in figure 25 .
Low input Average input High input
Controlled output z Area corresponding to the proportional controller Exception influencing area
Low output Average output High output Input variable x Exception y
Fig. 24 : simulating a controller proportional term.
Fig. 25 : introducing an exception into a proportional term.
The table in figure 26 lists the rules for recirculation. The proportional term is created from the input variable NOx content. The two
input variables nitri O2 content and denitri O2 content define an exceptional situation in the first rule.
IF nitro O2 content Not low
AND denitri O2 content Greater than 0
AND NOx content
THEN recirculation quantity Low
Low Normal High
Low Normal High
Fig. 26 : recirculation function rules table.
Below is another treatment using fuzzy logic: part of the sludge deposited in the downstream basin is recycled and re-injected upstream. The table in figure 27 lists the rules for sludge recycling. The first rule expresses an exception due to an excessively high upstream flow. In these conditions, a high degree of recycling would result in increased overload of the installation.
The exceptional condition is detected by the strong turbidity, as sludge deposits minimum sediment due to the excessively high flow. For information, other installation functions use fuzzy logic: c air injection, c management of excess sludge.
IF turbity of discharged water High
AND drained off quantity AND sludge level of recycled sludge Low Normal High Low Normal High Low Normal Low Low Normal Normal Normal High High
THEN quantity
Low Low Normal High Normal High Normal High
Fig. 27 : sludge recycling function rules table.
Food produce Automation of industrial oven production lines used for cooking biscuits interests biscuit manufacturers both in France and Germany. For this control type, a conventional solution is not satisfactory due to the non-linearities, multiplicity and heterogeneity of sensitive parameters. Modelling of the cooking process is both complex and uncomplete. However, experienced operators are perfectly able to control cooking using their empirical knowledge.
The chosen example is an aperitive biscuit production line. A French group contacted Schneider who then, in co-operation with ENSIA (French Higher Institute of Agricultural and Food Industries) worked out an automated solution. The main characteristics that can be measured in a biscuit are its colour, humidity and dimensions. These characteristics can be affected by variations in quality of pastry
ingredients, environmental conditions and the time the biscuit remains in the oven... These influences must be compensated by oven setting and conveyor belt speed. Control of production quality of this kind of food process can be broken down into the following functional steps: c conditioning and merging of data, c evaluation of subjective quantities (linked to quality), c diagnosis of quality deviations, c decision-making, c subjective evaluation Fuzzy logic enables qualitative variables to be taken into consideration and existing professional expertise to be used. Fuzzy rule bases have been used associated with other techniques (see. fig. 28 ).
c Subjective evaluation Most quality defining notions depend on a number of variables. One of the factors for evaluating quality is colour which is threedimensional: hence the interest of defining membership functions upon several variables. Classification algorithms, based on the input variables perform a gradual evaluation of such qualitative variables (top of biscuit well cooked, over cooked,...). c Diagnosis The fuzzy ladder was used to diagnose quality deviations observed on biscuits (see fig. 29 ). The oven has 3 sections. The overall operating evaluation is satisfactory. Other examples c Automation systems G.P.C.s (Global Predictive Controllers) are extremely effective, but require the setting of 4 parameters: N1, N2, Nu, l (2 control horizons, prediction horizon, weighting factor). Such setting is both lengthy and difficult and normally requires an expert. Schneiders NUM subsidiary is currently developing numerical controls and would like to use G.P.C.s in future productions.
Functions Sensor melting Subjective evaluation Diagnosis Decision making
Associated techniques
Fuzzy classification Fuzzy ladder Fuzzy rule bases
Fig. 28 : functions and associated techniques.
High biscuit humidity
Bottom of biscuit well cooked
Top of biscuit well cooked
Section 1 temperature too low
Bottom of biscuit a little over cooked
Top of biscuit a little over cooked
Bottom of biscuit far too cooked
Top of biscuit far too cooked
Bottom of biscuit undercooked
Top of biscuit undercooked
Fig. 29 : fuzzy ladder for quality deviation diagnosis.
Schneider has thus developed for NUM a method for automatically setting the parameters for such controllers by means of fuzzy rule bases. Some twenty rules suffice for rapid, reliable parameter setting. Moreover, the presence of a monitoring and control specialist, hard to find in numerical control installations, is no longer necessary. c Car industry Renault and Peugeot (PSA) have announced an automatic gear box which uses fuzzy logic to adapt to the type of driving of the person behind the wheel. c Cement plants The first industrial application of fuzzy logic, then copied by other manufacturers, was produced by
the Danish company, F.L. Smidth Automation, to control cement kilns. This process takes many variables into account, and in particular the climatic influences on the kiln which is several dozen metres high. c General public electrical and electronic household appliances A large number of applications are now available to the general public, especially in Japan. For example, compact size digital camcorders are highly sensitive to movement. Fuzzy logic controls the stadyshop image stabilization of these devices.
6 Conclusion
c Classed as an artificial intelligence technique, fuzzy logic is used to model and replace process control expertise and designer/operator expertise. c A tool for enhancing quality and increasing productivity, fuzzy logic offers competitive advantages to industrial firms seeking technicaleconomic optimisation. c This Cahier Technique specifies the areas in which this interesting approach can be used to advantage. c Thanks to suitable programmable controllers and user-friendly tools, fuzzy logic is now accessible to all automatic control engineers wishing to increase the scope of their skills and the performance of their achievements. These tools are available in the development environment of some programmable controllers
(see fig. 30 ), and offer simple evaluation possibilities. c Evaluation limited to competition with the other conventional control tools is not productive as such tools (e.g. PID controllers) continue to be useful in most application areas. c Fuzzy logic has its own special areas in which it works wonders: these are areas involving expertise, nuanced decision-making, consideration of non-linear phenomena and subjective parameters, not to mention contradictory decision-making factors. Contact with Schneider specialists will enable users and designers to find a suitable answer to their perfectly understandable question: What decisive advantages can fuzzy logic offer me in my application?.
a - configuration of the fuzzy logic module
c - writing of rules
b - definition of membership functions
d - simulation - validation
Fig. 30 : for fuzzy logic, the Schneider programmable controllers are equipped with user-friendly development tools on PC.
Appendix
Operators between fuzzy sets The table in figure 31 shows the ZADEH operators.
ZADEH operator AB A Intersection B
Logic operation
A AB = MIN (A, B)
AND
AB
AB A Union B
A AB = MAX (A, B)
OR
AB
_ A
A
Negation A A
_ A
_ A
= 1 - A
NOT
Fig. 31 : operators between fuzzy sets.
Singleton output membership functions Singleton membership functions are often used as output membership functions for fuzzy rules. This is because they allow the same interpolation effect between rules as for triangular membership functions (for example) for far simpler calculations. There is no need to calculate the maximum of output membership functions (aggregation), and the centre of gravity is also simplified. Figure 32 illustrates this calculation.
1
Low Average High Output Action
Fig. 32 : defuzzification of singleton membership functions.
Fuzzy inferences: fuzzy implication and Generalised Modus Ponens As shown in figure 33 , the conventional forward inference mechanism from the front or modus ponens consists of using rules, also known as implications, and a deduction mechanism (the modus ponens) to deduce conclusions from observed facts. The implication A B is considered to be true as long as it is not invalidated (A true and B false): see figure 34 . With knowledge whether the implication is true or false, the modus ponens enables a conclusion B to be deduced from an observation A. The same theoretical principle can be generalised in fuzzy logic. The general diagram is given in figure 35 .
Rules (implications)
Facts observed
Modus Ponens
Conclusions
Fig. 33 : principle of inference from the front.
A AB B 0 1 0 1 1 1 0 1 AB B' 0 0 1 0 0
A' 1 0 1
Implication
Modus Ponens
Fig. 34 : principle of implication and Modus Ponens.
Rules (fuzzy implications)
Inputs
Fuzzification
Generalised Modus Ponens
Defuzzification
Outputs
Fig. 35 : principle of fuzzy inferences.
The mechanism generalising the implication is known as the fuzzy implication. There are several fuzzy implication operators, including those mentioned below: MAMDANI: A B = MIN (A, B) LARSEN: A B = A . B LUKASIEWICZ: A B = MIN (1,1 - A + B ) The fuzzy implication works like a conventional implication, where A and B are fuzzy sets. The mechanism generalising the modus ponens is known as the generalised modus ponens. It obeys the following formula, and is used to determine a B conclusion fuzzy set. In most cases the operator T used is the Minimum (known as the Zadeh operator). B (y) = MAXx (T(A(x), AB (x,y)) ) where T: modus ponens operator (t - standard),
The Lukeziewicz operator behaves like the conventional implication when we limit ourselves to boolean values. This is not the case for the Larsen and Mamdani operators used in the Mamdani rule bases. These operators are the most extensively used as: c they offer a high degree of robustness in applications. c calculations are considerably simplified and allow simple graphic interpretation (see section 2.4.). Calculations on input x and output y are decoupled, as the formula below shows:
B(y) = MAXx (Min (A (x), A(x), B(y)) ) = Min (B(y), MAXx (Min (A(x), A(x)) )
Bibliography
Standards IEC 61131-7: Programmable Controllers part 7 Fuzzy Control Programming. Miscellaneous works c Fuzzy models for pattern recognition. IEEE Press, 1992. James C. BEZDEK & Sanker K. PAL. c Fuzzy sets and systems: Theory and applications. Academic Press 1980, Mathematics in Sciences and Engineering vol. 144. D. DUBOIS, H. PRADE. c Evaluation subjective ; mthodes, applications et enjeux. Les cahiers des clubs CRIN, club CRIN logique floue. c A.I. and expert system myths, legends and facts. IEEE Expert 02/90, pp 8-20, 29 rf. M.S. FOX. c La logique floue et ses applications. Addison-Wesley, 1995. Bernadette BOUCHON-MEUNIER. Internet c http://www-isis.ecs.soton.ac.uk/research/nfinfo/ fuzzy.html c http://www.ortech-engr.com/fuzzy/ reservoir.html c http://www-cgi.cs.cmu.edu/afs/cs.cmu.edu/ project/ai-repository/ai/areas/fuzzy/0.html
Schneider UK - Guide to forms of separation
This menu is only visible while using a java enabled browser. Please click on Help in your browser for more information on how to enable Java.
Introduction Switchboards and Separation Forms of Separation Equipment Protection Form Subdivisions Subdivisions - The Implications Subdivisions - The New and the Old Beyond the Standards
INTRODUCTION
Today, things are moving so fast in our industry that it's hard to keep up. In particular, it's becoming increasingly difficult for engineers and technicians to find up-to-date, reliable and, above all, practical literature to guide them in their work. Yesterday's publications just won't do. Standards and requirements change so rapidly that the guidance given in old literature may be worse than useless - it may actually be dangerous. Recognising the need for timely and accurate information, Electrical Equipment, the industry's foremost practical magazine, has joined forces with Schneider, a leading world and UK expert in electrical distribution and industrial control. The result of the collaboration is a series of supplements, of which this is the first, dealing in a totally practical way with some of today's hottest topics. We intend that these supplements will grow to become a valuable and accessible reference library containing information which is of day-to-day use to engineers and technicians. This first supplement deals with switchboard construction, an area where there is currently considerable confusion, and where recent changes have made most of the available literature obsolete. Schneider's experts have studied the subject in great depth, and they present their conclusions here for you in a clear and concise form. Your comments on this supplement are welcome, as are your suggestions for topics which you would like to see covered in future issues. The scope of this publication This supplement principally explains Schneider's interpretation of the National Annex NC to BS EN 60439-1, the most recent standard covering the construction and testing of electrical switchboards. All information presented here is offered in good faith, and has been carefully checked. Readers are reminded, however, that the final authorities on all matters, and particularly those relating to safety, are the relevant regulations and standards, to which reference should always be made.
http://www.schneider.co.uk/booklets/no.3/index.htm [25.10.2000 5:13:26 PM]
Introduction Switchboards and Separation
SWITCHBOARDS AND SEPARATION
What is separation? In recent years, separation has become one of the most important factors to be Forms of Separation considered in the design and specification of switchboards. It is based on a very simple idea, the need to separate people from live parts and, under some Equipment circumstances, to separate different groups of live parts from one another. Just Protection putting the electrical elements of the system in an enclosure provides a degree of separation, since it separates users from the live parts and prevents accidental Form Subdivisions contact with them - at least while the enclosure door is shut! Subdivisions - The In many applications, however, there are definite benefits to using rather more Implications elaborate types of separation. For example, dividing the enclosure into separate Subdivisions - The compartments provides better opportunities for cable termination while the New and the Old equipment is live. It may also allow emergency repairs to be carried out on one section of a switchboard, without the need to shut down the complete system. Beyond the Standards While the idea of separation is simple, its implications for safety and reliability of equipment are important enough for it to need a formal definition which will provide a common basis for agreement between switchboard manufacturers and users. The current British Standard offers the following definition for separation: q protection (of personnel) against contact with live parts belonging to adjacent functional units q limitation of the probability of initiating arc faults q protection against the passage of solid foreign bodies from one unit of an assembly to an adjacent unit It is vitally important to realise that only one of these three criteria needs to be met for separation to be claimed. It is also worth noting that the reference to arc faults is concerned only with limiting the probability of initiating a fault. A further noteworthy point relating to arcs is that, in practice, these divide into two categories - normal arcs which occur during switching operations, and abnormal arcs which are produced by, for example, conductive foreign bodies bridging live conductors. The performance of a switchboard under normal arc conditions is usually addressed as part of the type-testing procedure. Abnormal arcs are not explicitly covered by the standards, except that the good engineering practice which the standards recommend will help to limit the probability of such arcs occurring. From what has already been said, it will be clear that many different levels and types of separation are possible. Most of the remainder of this supplement will deal with these so-called forms of separation, and their implications for the manufacturers and users of switchboards. The Standards In 1994, BS EN 60439-1, a new British Standard governing LV switchgear and controlgear assemblies, superseded BS 5486:Part 1:1990, an earlier standard covering the same subjects. Another important document is the EIEMA Guide to Forms of Separation which details subdivisions for the forms of separation described in BS 5486. Although it has no formal status, this guide has proved invaluable, and has come to be accepted as an extension to that standard. The subdivisions in the forms of separation which are detailed in the EIEMA guide have no equivalents in the main body of BS EN 60439-1. To address this omission, a National Annex was added to the new standard in March 1995. The
http://www.schneider.co.uk/booklets/no.3/switch.htm (1 of 2) [25.10.2000 5:14:37 PM]
old standard, in conjunction with the EIEMA guide, and the new National Annex define various forms of separation, and they use very similar names to identify the forms. There are, however, detailed differences which mean that, in many instances, the old and new definitions are not directly equivalent. Unsurprisingly, this is generating considerable confusion among both manufacturers and users of switchboards. The detailed definitions of the various forms, and the differences between the old and new standards are, however, dealt with in later sections of this supplement. Summary Separation increases the safety and maintainability of switchboards by enclosing and segregating functional parts. Different levels, or forms, of separation are possible, and these are detailed in the National Annex to BS EN 60439-1. This standard supersedes BS 5486-1 and the EIEMA guide associated with it, and introduces changes in the definitions for the forms of separation.
http://www.schneider.co.uk/booklets/no.3/switch.htm (2 of 2) [25.10.2000 5:14:37 PM]
FORMS OF SEPARATION
The four basic forms The latest standard, in common with its predecessor, defines four basic forms of Forms of Separation separation: Form 1An assembly so enclosed as to provide protection against contact with Equipment internal live parts, but where no internal separation is provided. Protection Form Subdivisions Subdivisions - The Implications Subdivisions - The New and the Old Beyond the Standards Form 2As Form 1, but with internal separation of the busbars from the functional units. Form 3As Form 2, but with separation of all functional units from one another. Form 4As Form 3, but the incoming and outgoing terminals are also required to be separated from the busbar system, and from one another. All of these forms, with the exception of Form 1, are further subdivided, and it is in these subdivisions that the old and new standards differ. Before examining these details, however, it is worth looking at how the basic forms relate to real equipment and applications. Why not Form 1 for all? Form 1 requirements are satisfied by virtually any enclosed electrical system. Since Form 1 provides full protection against accidental contact with live parts, it might easily be seen as completely adequate for the majority of applications. In a perfect world, this might be true but, in practice, Form 1 construction has a number of serious limitations. From the safety point of view, it is only adequate when there is never any requirement to work on the equipment while it is live. While no manufacturer of switchgear - Schneider included - would recommend or even support live working, there are certain circumstances where it is virtually unavoidable. For example, it would hardly be acceptable to cut off the supplies to a large section of a hospital simply to replace a fuse in one of the outgoing circuits. All references to live working in this supplement assume that the work is carried out in strict accordance with the Electricity at Work Act, particularly Regulation 14 which states: No person shall be engaged in any work activity on or so near any live conductor (other than one suitably covered with insulating material so as to prevent danger) that danger may arise unless (a) it is unreasonable in all the circumstances for it to be dead; and (b) it is reasonable in all the circumstances for him to be at work on or near it while it is live; and (c) suitable precautions (including where necessary the provision of suitable protective equipment) are taken to prevent injury. A basic Form 1 construction must never be used where live working is even remotely possible. In these instances, the choice lies principally between Forms 3 and 4. Form 2 may be worth considering but, in its most basic version, it provides very few advantages over Form 1 from the point of view of safe live working. Form 1 construction has further limitations. For example, there is no protection against solid objects, such as washers and nuts removed to replace a fuse cartridge, subsequently falling out of one item of equipment into another.
http://www.schneider.co.uk/booklets/no.3/forms.htm (1 of 2) [25.10.2000 5:14:57 PM]
None of this, however, means that Form 1 construction is without a role to play. It is the least expensive form and, providing its limitations are clearly understood, its economy may be put to good use in undemanding applications. The higher forms As already mentioned, the most basic version of Form 2 construction offers few benefits over Form 1, since there is no requirement for incoming and outgoing terminals to be separated from the busbars. This basic version is, however, rarely offered. The more commonly encountered versions of Form 2 construction are slightly more tolerant of live working than Form 1, in that they permit the busbars to remain live while other equipment is being worked on, and may, therefore, eliminate the need for complete removal of power from the system. It is only when Form 3 is reached, however, that individual sections of the equipment may be worked on in relative safety with other sections live. Form 3 is also the lowest form of protection to restrict the movement of solid foreign bodies from one item of equipment to another. In many applications, Form 3 adequately satisfies all operational and safety requirements without the expense which is usually involved in full Form 4 designs. Form 4 does, however, provide the greatest flexibility, allowing the best opportunity for external circuits to be connected or disconnected without the need for adjacent items of equipment to be isolated. When specifying switchboards, however, it is worth bearing in mind that this may be a relatively infrequent requirement. It must also be clearly understood that even the highest forms of separation do not guarantee safe live working. This requires consideration not only of the switchboard design, but also the circumstances surrounding the work, in order to ensure that the obligations of the Electricity at Work Act can be fully satisfied. In fact, BS EN 60439-1 clause 7.4 specifically states that live working is an issue which requires a separate agreement between the manufacturer and user of a switchboard.
http://www.schneider.co.uk/booklets/no.3/forms.htm (2 of 2) [25.10.2000 5:14:57 PM]
EQUIPMENT PROTECTION
Forms of separation for switchboards are most often thought of in terms of the personnel protection which they provide, but they also have an important effect on Forms of Separation the amount and type of damage caused to equipment under various conditions of malfunction. Two of the main areas of consideration relate to mechanical Equipment protection. Protection The first is concerned principally with the movement of foreign bodies between Form Subdivisions compartments. The IP2X rating specified by the standard for separation does Subdivisions - The potentially allow items such as washers and small spanners to pass between compartments. The panel specifier may, however, insist that there are no holes in Implications the horizontal partitions between compartments. Subdivisions - The A second and more subjective issue is the ruggedness of the materials used for New and the Old separation. It is partially addressed by the footnote to the National Annex, which Beyond the refers to clause 7.4.2.1. This requires material to be "capable of durably Standards withstanding the mechanical, electrical and thermal stresses to which it may be subjected in service". As no specific performance requirements are defined, judgements must be made, e.g., will access only be required in order to carry out delicate adjustments; is there going to be a need to terminate large cables without total isolation; can taping or wrapping of busbars and terminals provide adequate mechanical protection for the application being considered? Choosing the right form As the cost of a switchboard usually increases with increased levels of separation, automatically choosing the highest level is not always the best solution. Instead, the decision should be made after a careful consideration of the following factors: q the maintenance needs of the switchgear q the probability of requiring access to the cable terminals of a circuit with adjacent terminals live and in service q the skill level of personnel having access to the equipment q the difficulty and inconvenience of isolating the complete switchboard q the costs and benefits of the different forms of construction The following table gives an indication of the types of decision which are involved in the selection process.
Will it be necessary to - Answer work on the board live? No Yes
Consider using Form 1 3 or 4 (possibly 2, but see comments in text) 1, 2 or 3 4
cable one circuit with the rest live? carry out maintenance while the board is live?
No Yes
No 1 Yes - tighten busbar No form is suitable connections 3 or 4 Yes - replace contactor contacts
http://www.schneider.co.uk/booklets/no.3/equip.htm (1 of 2) [25.10.2000 5:15:13 PM]
change fuses while the board is live?
No Yes
1 1 (subject to supplementary separation ensuring safe access to fuse links)
Summary The major levels of separation are designated Forms 1 to 4. Form 1 may be adequate if no live working is contemplated; Form 2 provides a little extra flexibility but, for the majority of applications, Form 3 is a better choice. Form 4 is appropriate where maximum opportunity for live working is essential.
http://www.schneider.co.uk/booklets/no.3/equip.htm (2 of 2) [25.10.2000 5:15:13 PM]
Introduction
FORM SUBDIVISIONS
Switchboards The National Annex to BS EN 60439-1 recognises the following subdivisions in the forms of separation: and Form 1 No subdivisions Separation Form 2 Subdivided into Form 2, Form 2 type 1, and Form 2 type 2 (see note below) Forms of Separation Equipment Protection Form 3 Subdivided into Form 3a, Form 3b type 1, and Form 3b type 2 Form 4 Subdivided into seven types, designated Form 4 type 1 to Form 4 type 7
(Note: Form 2 subdivisions may be renamed in the near future as Form 2a, Form 2b type 1, and Form 2b Form type 2.) Subdivisions The differences between the subdivisions may, in some cases, seem relatively minor but, depending upon Subdivisions the application, they may have important implications. - The Implications FORM 2 SUBDIVISIONS Form 2 Busbars are separated from the functional units, but not from the terminals for external connections Subdivisions - The New Form 2 type 1 Busbars are separated from the functional units, and from the terminals for external and the Old connections. Separation of the busbars is achieved by means of insulated busbar covering, such as sleeving, wrapping or coating. Beyond the Standards Form 2 type 2 As Form 2 type 1, but busbar separation is provided by the use of rigid metallic or non-metallic barriers. FORM 3 SUBDIVISIONS Form 3a Busbars are separated from the functional units, and the functional units are separated from each other. Terminals are separated from the functional units, but not from each other. Terminals are not separated from the busbars. Form 3b type 1 As Form 3a, but terminals are separated from the busbars. Busbar separation is provided by means of insulated coverings. Form 3b type 2 As Form 3b type 1, but busbar separation is provided by rigid metallic or non-metallic barriers. FORM 4 SUBDIVISIONS Form 4 type 1 Busbars are separated from the functional units, and the functional units are separated from each other, including the terminals which are an integral part of the functional unit. Terminals are in the same compartment as the associated functional unit, but cables may be glanded elsewhere. Busbar separation is achieved by means of insulated coverings. Form 4 type 2 As Form 4 type 1, but busbar separation is provided by rigid metallic or non-metallic barriers. Form 4 type 3 As Form 4, type 2, but each functional unit has its own integral glanding facility. Form 4 type 4 Busbars are separated from the functional units, and the functional units are separated from each other. Terminals are separated from the busbars, the functional units, and from each other, but cables may be glanded elsewhere. Busbar separation is achieved by means of insulated coverings. Form 4 type 5 As Form 4 type 4, but busbar separation is provided by rigid metallic or non-metallic barriers. Terminals may be separated by insulated coverings, and may be glanded in common cabling chambers. Form 4 type 6 As Form 4 type 4, but all separation requirements are provided by rigid metallic or
http://www.schneider.co.uk/booklets/no.3/subdiv.htm (1 of 3) [25.10.2000 5:20:01 PM]
non-metallic barriers. Cables are glanded in common cabling chambers. Form 4 type 7 As Form 4 type 4, but all separation requirements are provided by rigid metallic or non-metallic barriers. Each functional unit has its own integral glanding facility. The diagrams on the following pages illustrates the requirements for all form subdivisions.
Key to forms of separation diagrams
Form 1
Form 2
Form 2 type 1
Form 2 type 2
Form 3a
Form 3b type 1
Form 3b type 2 Form 4 type 1
Form 4 type 2
Form 4 type 3
Form 4 type 4
Form 4 type 5
Form 4 type 6
Form 4 type 7
Introduction
SUBDIVISIONS - The implications
Switchboards With so many form subdivisions available, it's difficult to work out quickly the implications of each. The following chart, however, provides a snap overview of the forms of construction with subdivisions. and Separation Forms of Separation Equipment Protection Form Subdivisions Subdivisions - The Implications Subdivisions - The New and the Old Beyond the Standards
Separation - BS EN 60439 - (with National Annex NC)
SUBDIVISIONS - The new and the old

As mentioned earlier, the subdivisions in forms of construction detailed in the National Annex to BS EN 60439-1 differ from those used in the earlier standard, BS 5486:Part 1:1990, and the associated EIEMA guide. Much existing equipment will have been designed, however, to the old standard. This makes it useful, particularly when contemplating modifications or extensions, to know how the new and the old standards compare. Major differences are summarised in the following table. BS5486 EIEMA BS EN 60439-1 Guide Differences between BS EN 60439-1 National Annex and EIEMA guide forms 1 2 1 none 2a 2b none 1 2 2 type 1 2 type 2 3a none not included none none not included
http://www.schneider.co.uk/booklets/no.3/implic.htm (1 of 2) [25.10.2000 5:21:14 PM]
3a 3b 4a 4b 4c 4a 4b 4b 4c
3b type 1 3b type 2 4 type 1 4 type 2 4 type 3 4 type 4 4 type 5 4 type 6 4 type 7
outgoing terminals separated from busbars none none terminals in same compartment as functional units terminals in same compartment as functional units terminals separated from functional units, located in individual compartments, and shrouded to IP2X none rigid barriers over terminations none
Comparison of BS EN 60439-1 forms of construction with their nearest equivalent in BS5486 and the EIEMA guide Summary The latest standard defines 14 subdivisions in the forms of separation, most of which are not directly equivalent to the subdivisions in the EIEMA guide used with the older standard. Some subdivisions relate to the separation between terminals and functional equipment, while others deal with the method of separation. Subdivisions have an important bearing on personnel safety during live-board working.
http://www.schneider.co.uk/booklets/no.3/implic.htm (2 of 2) [25.10.2000 5:21:14 PM]
BEYOND THE STANDARDS
The new standard for switchboards, together with the National Annex, provides valuable guidance for specifiers, users and manufacturers of switchboards. Forms of Separation Adoption of one or other of the forms of construction which it describes will provide the safest, most practical and most economic solution for virtually every Equipment application. The standard should, however, never be seen as restricting alternative Protection approaches which may be needed to cope with special circumstances or requirements. The standard is not intended to be, and cannot be, all-embracing; the Form Subdivisions final details of the method of construction adopted for every project must always Subdivisions - The be the subject of a specific agreement between the manufacturer and the supplier. Implications Subdivisions - The New and the Old Beyond the Standards
http://www.schneider.co.uk/booklets/no.3/beyond.htm [25.10.2000 5:21:30 PM]

Current transformers: specification errors and solutions
P. Fonti
no. 195
Paola FONTI INPG engineer (Institut National Polytechnique de Grenoble). Graduated in 1970. Joined Merlin Gerin in 1981 as consultant and head of the Medium Voltage Export engineering and design department. She is currently responsible for the MV project completion and tender support group for Schneider Electric.
ECT 195 first issue, February 2000

Lexicon
If: maximum through current crossing a protected area. Is: current threshold setting.
kn: nominal accuracy limit factor (ALF) of a CT (associated with its accuracy load). kr: real ALF of a CT associated with its real load.
2 Pi: (= Rct In ). Internal losses of the CT at In. 2 Pn: (= Rn In ). Accuracy power of the CT.
2 Pr: (= Rr In ). Real load consumption of the CT at In.
RL: wiring resistance. Rp: protection relay resistance. ALF: accuracy limit factor. CT: current transformer. Overrating of a CT: selection of a CT whose primary In is greater than the In immediately greater than the load In. Matching, auxiliary or interposing CT: low voltage CTs installed at the secondary of the main CTs for correcting a ratio and/or the current phase shift. SF: security factor.
Rct: CT secondary winding resistance.
After a reminder of current transformers (CTs), the author highlights the errors most often encountered when defining current transformers, an essential and little known link between the electrical network and the protection relays. It explains how to find a way out of difficult situations: CTs that cannot be manufactured, delays, additional costs, malfunctions, etc. This Cahier Technique should be useful for electricians designing installations, for protection specialists, panel builders and all CT manufacturers. It is in the best interest of all to exchange all information required for the safety and optimisation of CTs. This Cahier Technique is an operational addition to Cahier Technique no. 194 Current Transformers: how to specify them.
Contents
1 Specifying current transformers properly 2 Examples of specification errors 1.1 Introduction 1.2 Reminder of CTs 2.1 Optimisation and safety 2.2 When CTs do not seem to be suitable 2.3 The most frequent errors 2.4 And if the CT cannot be manufactured? 3 Equivalence of the various possible definitions of the same CT 3.1 How to move from Pn1-5Pk1 to Pn2-5Pk2 3.2 How to move from Pn1-5Pk1 to Pn2-10Pk2 3.3 What is the Vk of a CT: Pn-XPk 3.4 How to move from class X (Vk, Rct) to a class 5P: Pn-5Pk 4 Conclusion 5 Bibliography p. 4 p. 4 p. 7 p. 8 p. 9 p. 12 p. 15 p. 15 p. 16 p. 16 p. 17 p. 18
1 Specifying current transformers properly
1.1 Introduction
Traditional current transformers (see Cahier Technique no. 164) and hybrid current transformers (see Cahier Technique no. 170) form an essential link within the protection chain of electrical networks. Their specification, even if it is handled by specialists, often includes errors and is insufficiently optimised. This often leads to technological impossibilities, operating delays, extra costs, incorrect operation of protections and can even jeopardize the safety of installations and people. Proper specification of CTs (see Cahier Technique no. 194) requires sound knowledge of: c the electrical installation diagram, c the electrical data (voltage, nominal current, short-circuit current, etc.), c the associated protections, c the overall network protections (protection plan, the load that they represent for the CTs, as well as wiring and their settings. Often, due to lack of data or even ignorance of how a CT shall be used, a CT manufacturer says these features are not feasible, while a standard CT may be suitable. Although this Cahier Technique emphasises optimisation, it particularly stresses the equivalence between the different definitions of the same current transformer. You should bear in mind that power, class and accuracy limit factor are interdependent values, which have no significance if taken individually. This knowledge is a means of finding a way out of many nearly impossible situations. Before entering the heart of the matter, a few reminders of CT characteristics are given in the following sections.
1.2 Reminder of CTs

According to IEC standards (among others), CTs can be defined by: a - Their ratio, example: 2000/5 A. b - Their power, example: 15 VA. c - Their class, example: v 5P, 10P for a protection winding, v class 0.5, 1, etc. for a metering winding. d - The characteristics linked to their saturation: d-1 Accuracy Limit Factor (ALF) for a protection winding, d-2 Safety Factor (SF) for a metering winding. e - Other characteristics: v thermal withstand, example 50 kA - 1 sec., v insulation voltage, v etc. In the reminder of this section, we shall be concerned only with characteristics a, b, c, and d and their consequences (see fig. 1 ).
Protection
2500/1 A
15 VA
5P
20
(a) Metering 500/5 A
(b) 20 VA
(c) cl 0.5
(d-1) FS3
(a)
(b)
(c)
(d-2)
Fig. 1 : main characteristic values of the CT.
A 15 VA-5P20 CT has a guaranteed error of less than 5 % when it is subjected to 20 times its nominal current and delivers into its nominal load (15 VA to In). Each of the characteristics b, c, d is a function of the two others. The same CT can be affected a different power, a different accuracy class and a different ALF. However, a given CT only has one magnetising curve and only one secondary winding resistance (at a given temperature). When these last two elements are known (curve and resistance), we can identify all the necessary correspondences between the various values a, b, and c to be assigned to the CT or rather between the various triplet combinations: (b1, c1, d1) (b2, c2, d2) (bi, ci, di) All the equivalences are deduced from the simple laws of electricity, in particular Ohms law. CT equivalent diagram (see fig. 2 ). c CT ratio: In / In .
1 2
It is the magnetising current Im which generates a metering error. If the CT were perfect, then Im = 0. The CT magnetisation curve represents the magnetising current as a function of voltage Vs developed at the CT secondary. It can be divided into 3 zones (see fig. 3 ): 1 - non-saturated zone, 2 - intermediate zone, 3 - saturated zone. In zone 1, current Im is low and voltage Vs increases almost proportionally to the primary current. Zone 2 is a vague zone between the nonsaturated zone and the saturated zone. There is no real break in the magnetisation curve. It is hard to locate a precise point on the curve corresponding to the saturation voltage. In zone 3, the curve (Vs Im) becomes almost horizontal. The error is considerable on the ratio and the secondary current distorted by the saturation. A certain number of characteristic voltages are highlighted for a CT: they correspond to zone 2 ; knowledge of these voltages is necessary when another definition is given to a particular CT.
c Lm: CT equivalent inductive magnetisation (saturable). c Im: magnetising current. c I1: primary current. c I2: secondary current corresponding to a perfect CT, i.e. I2 =
I1
In
2 1
In
Vs 3 2
c Is: secondary current effectively crossing the CT secondary: I 2 = I s + I m .

I1
I2
Is
Rct
In1 In2
Im
Lm Vs
Rp
1 - Non-satured zone 2 - Intermediate zone 3 - Satured zone
Im
Fig. 2 : a CT equivalent circuit.
Fig. 3 : magnetisation curve (excitation) Vs = f(Im) of a CT.
Characteristic voltages linked to a CT c Knee point voltage defined by the BS 3938 standard: Vk for class X (PX in IEC 60044-1). Vk is determined by the point on the curve Vs (Im) from which a 10% increase in voltage Vs leads to a 50% increase in the magnetising current. c Voltage linked to the accuracy limit of the class 5P CTs: V(5P) = Vs1 . c Voltage linked to the accuracy limit of the class 10P CTs: V(10P) = Vs2 . c Voltage linked to the safety factor Fs V(Fs ) = Vs2 , since the safety factor is linked to an accuracy limit of 10 % just like the class 10P CT. These various voltages Vk < V(5P) < V(10P) are each linked to an induction level. With the materials commonly used for CT manufacturing, for example: c Vk corresponds to 1.4 tesla, c V(5P) = Vs1 corresponds to 1.6 tesla, c V(10P) = Vs2 corresponds to 1.9 tesla, c V(Fs ) = Vs2 corresponds to 1.9 tesla. The following ratios can be deduced:
Vs1 Vk V 1.4 1.4 1.6 = = ; k = ; ; etc. Vs1 1.6 Vs2 1.9 Vs2 1.9
CT accuracy is guaranteed better than 5% up to Is = 15 In. From this point on, it is sufficient to refer to the CT equivalent circuit and to Ohms law to obtain the value of V(5P) or Vs1 (see fig. 4 ). The result is simply: Vs1 = (Rct + Rp ) I s , i.e. Vs1 = (Rct + Rp ) 15 In . This relation shows that knowledge of the internal resistance of the CT secondary winding is absolutely necessary to correlate the various possible definitions of the CT. We shall say that a good CT definition must include, whatever the case, the value of Rct. In our case, let us assume that Rct = 0.6 , where Rn =
10 = 0.4 , the calculation yields: 52
Vs1 = 15 x 5 (0.6 + 0.4) = 75 volts.

With the induction values mentioned above for the 10P classes and class X, Vs 2 and Vk can be calculated.
I1
I2
Rct
Is = 15 In
If one of these voltages is known, it is simple to deduce the others. How to calculate the characteristic values from a CT defined in the 5P or 10P class. c Let us take an example: Let us assume a 10 VA-5P15 CT with a ratio of 2000/5. 10 VA-5P15 means that when there is a load equal to its nominal load Rp =
Pn
2 In
In1 In2
Ie
Vs
Rp =
Pn
2 In
Vs1 = (Rct + Rp) Is
Fig. 4 : calculating the characteristic voltage of a CT.
2 Examples of specification errors
2.1 Optimisation and safety

Examination of a protection plan shows that it is designed, for the transformer feeders at switchboard level, to fit CTs of different ratios 50/5 and 1000/5 with the same definition 15 VA-5P20. These CTs are associated to the same overcurrent protection relays with settings of 15 In (of CTs) for 50 A feeders, and 12 In (of CTs) for 1000 A feeders. Same load (0.05 VA), same wiring (1.25 VA) and relay settings not too much different: it seems logical that the two sets of CTs have the same definition. However a quick calculation of the necessary power shows that this is not the case c For the 1000/5 The necessary ALF (see Cahier Technique no. 194) is More serious still, in event of a trip failure of the transformer feeder, the short-circuit current (Ith of the 40 kA /1 s switchboard) will cause an rms current greater than: 172 x 5 A = 860 A to flow at the CT secondary. (Without saturation 40000 A /50 A x 5 A = 4 kA). The relay and the wiring will be destroyed as well as the CT. A 5 VA CT is thus more than enough (kr = 67 > 30). A 2.5 VA CT is also suitable: it is less expensive, takes up less space and, most important, it can be manufactured. c Conclusion The power of low ratio CTs must be calculated as their naturally low Rct induces a risk of dangerous oversizing. Figure 5 sheds some light on the interactions between kr, Pp, Pi.
2 Ir , i.e. k = 24. r
In
Knowing that k r =
k n (Pi + Pn ) , where: Pi + Pp
v an internal resistance Rct = 0.6 , v kn = 20, v CT internal losses at In = Pi

2 Pi = Rct x In = 0.6 x 52 = 15 VA, v a real CT load at In = Pn = 1.3 VA, we obtain: k r = 36.8 > 24 . The CT is thus more than suitable.
kr = real ALF 220.0 200.0 180.0 160.0 140.0 120.0 100.0 80.0 60.0 40.0 20.0 0.0 0 2 4 6 8 10 (Pn) 12 Pp (VA) Pi = 2 VA Pi = 5 VA Pi = 0.5 VA
c For the 50/5 The necessary ALF is 15 In x 2, i.e. kr = 30 . v First of all, such a CT is theoretically not feasible. Its internal resistance would be around 0.02 . But we can show that it is oversized in power. v Furthermore, with Rct = 0.02 , Pi = 0.5 VA; and with kn = 20 and Pp = 1.3 VA we obtain kr = 172 u 30.
Fig. 5 : accuracy limit factor behaviour of three CTs (with different Rct), of 10 VA-5P20 as a function of the real load connected to the secondary Pp.
2.2 When CTs do not seem to be suitable

This paragraph traces the difficulties encountered further to the manufacturing of CTs for a 33 kV public distribution switchboard consisting of 8 line feeders, without taking into account final needs. The entire problem came up on site a short while before energisation, when the CTs were already installed in the MV panels (see fig. 6 ). c The needs formulated when the contract was awarded. The aim was to have a double primary 300-600 and three 1 A secondary windings: v a 5 VA-5P20, to supply a Sepam 2000 protection unit, v a 15 VA classe 0.5, for remote metering, v a class X, for distance protection. c The equipment delivered. As it was impossible to manufacture three windings in the same functional CT with these characteristics, additional CTs were required. They were dedicated to class X and installed downstream of the double secondary CT. c The error observed on site. CTs operating with distance protection must be installed as close as possible to the circuitbreaker to ensure the greatest possible protected zone. However the class X CTs were not close enough. c What should we do? The proposal was thus made to replace all the CTs, i.e. 2 CTs by phase, 6 CTs by feeder, i.e. a total of 48 CTs for 8 feeders! Disassembly, manufacture of new CTs and reassembly led to additional delays and costs c Was there another solution? v An initial possibility was to consider use of additional CTs (class X) to supply the Sepam 2000. The supplier confirmed this possibility: the CT dedicated to distance protection (class X) corresponds to 10 VA-5P20 on ratio 300 and to 20 VA-5P20 on ratio 600/1. It was thus possible to associate the additional CTs with the Sepam 2000. What a relief! already 24 CTs less to replace.
Functional CTs
300-600/1 A 5 VA-5P20 300-600/1 A 15 VA-0.5
Sepam Metering
Distance protection Metering
Additional CT
300-600/1 A class X
Distance protection
Sepam
Delivered arrangement
Chosen solution
Fig. 6 : a 33 kV switchboard CT arrangement: the unsatisfactory diagram and the new solution proposed.
v The second possibility concerned the CTs with two windings: could the 5VA-5P20 windings be suitable for the distance protections? The fact is that, voluntarily, all the CTs were identical and corresponded to the worst possible case, while the lines were of different lengths (from 2 km to 38 km). The short lines had a cross-section of 50 mm2, the others 150 mm2. To come back to the real needs of distance protections, it was observed that for 6 feeders, class 5 VA-5P20 corresponded to a suitable class X whatever CT primary was chosen (300
or 600). For the two other feeders, the resulting class X was satisfactory only on the 600/1 ratio! A solution that the customer accepted for the two 150 mm2 feeders. c We can draw the following conclusion from this contract, which reveals all the consequences of unsatisfactory CT manufacture: the idea of replacing all components is often the first considered, but with the help of specialists it is possible to avoid pointlessly wasting time and money.
2.3 The most frequent errors

These errors normally lead to oversizing of the CTs which increases costs and can be dangerous. Many CT definition errors stem from ignorance of their operation and from the unknown or incomplete characteristics of the network component to be protected and of the associated protections. The better informed the CT manufacturer, the fewer errors and the more the CT will be optimised. Protections and conventional CTs For these protections which do not require class X defined CTs, the most frequent errors are: c Using two CTs or one CT with two secondaries for two protective relays the manufacturers of which recommend different ALF or different accuracy classes. As the CT manufacturers can translate 10P to 5P (according to the corresponding induction levels), and can move from one ALF to another by adjusting power, they can find a CT matching the needs of both relays. c Taking into account the wiring resistance although the protection manufacturer has already integrated it into the needs formulated for the CT. Let us take an example of two relays whose technical data indicate for 1 A CTs: CT1 for relay 1: 5 VA-10P15 (assuming 2 RL < 1.5 ), CT2 for relay 2: 10 VA-5P15 (assuming 2 RL < 2 ). A single CT may be suitable for both relays: in theory a 10 VA-5P15. You need to: v Avoid adding up the powers (5 + 10 VA) required for each relay. In point of fact, for the CT2, relay 1 only represents a load (just like wiring) and vice versa. v Check, in this case for the CT2, that: 2 RL + Rp1 i 2 ; and if CT1 was selected, that: 2 RL + Rp2 i 1.5 . If this was not verified, the relay supplier can suggest that x VA be added per additional ohm. Adding the specified power for several protections linked to an application results in CTs that are often impossible to manufacture or that jeopardise safety during short-circuits. Use of multifunction numerical relays avoids such errors. You only need to size the CT for the most restrictive protection (see Cahier Technique no. 194). c Changing the required characteristics without verifying the consequences. v A CT manufacturer cannot make a low ratio CT and suggests increasing this ratio; let us take an example: - Requested: 30/1 CT - 2.5 VA-5P20, - CT manufacturers proposal: 60/1, - With motor In = 16 A and minimum thermal protection setting: 40 % of CT In, i.e. 60 x 0.4 = 24 A. The protection setting at 16 A (normal thermal protection setting at motor In), is then impossible. The solution is to increase the rating and lower the ALF requirement: 40/1 - 2.5 VA-5P10. This CT, feasible, allows the required setting (40 x 0.4 = 16 A). v A buyer accepts a thermal withstand of 0.1 s proposed by the CT manufacturer instead of 1 s. What is likely to happen is that, on a short-circuit, if the real fault duration exceeds 0.1 s, thermal and probably electrodynamic withstand will be insufficient and may result in CT destruction. c Due to lack of information on real requirements. Let us take the following case, relatively educational: a CT with two primaries and three secondaries (200-1000/1-1-1) is requested with: - the first secondary: 1 A, class X (given Vk ), - the second secondary: 1 A-15 VA class 0.5 for metering, - the third secondary: 1 A-10 VA-5P20.
The supplier can propose a CT with three magnetic cores and secondary tappings to meet the 200 A or 1000 A need at the primary. However, such a CT is hard to manufacture because to obtain 15 VA-class 0.5 and 10 VA-5P20 on 200/1 ratios, you need 5 x 15 VA-class 0.5 and 5 x 10 VA-5P20 on 1000/1 ratios! Moreover, the supplier must comply with class X for both ratios! In point of fact, class X concerns only the 1000/1 ratio (for busbar differential protection). The 200/1 ratios concern metering and the traditional protections (see fig. 7 ). The CT to be manufactured is then easier, less bulky, cheaper and definitely feasible. This example shows that the lack of information shared between those involved is a source of errors and of non-optimisation. A consultation that does not begin properly may result in a CT that cannot be manufactured. c Taking into account the relay impedance Rh for calculation of real load (see fig. 8 ) in the CT calculation for overcurrent or in the calculation for CTs in class X. A word of warning: Rh is only considered when calculating CTs for zero sequence current lh (see Cahier Technique no. 194). For high impedance differential protections, in the calculation of Vk given by: 2 If (Rct + 2 RL + Ra), where Ra = other loads, Rh must not intervene. This is the load of one phase (we assume that no current flows through the neutral).
Rct
Rwiring
RO/C relay
Rct
Rwiring
RO/C relay
Rct
Rwiring
RO/C relay
Rh
Fig. 8 : internal and load impedances of a CT.
Vk is indeed calculated for relay stability conditions, i.e. no phase or earth fault, in the protected zone, no incorrect unbalance, therefore, in the differential connection I = 0 and the voltage of that connection = 0. Differential protections and class X For these applications, the most usual errors are: c Asking the CT manufacturer to supply CTs with the greatest Vk that he can build using a standard mould. This occurs when the differential protection relay (make, type) is not defined. There are three consequences: v overcost, v possibility of high overvoltages and overcurrents at the CT secondary which can lead to destruction of the circuit and the relay, v with no requirements for the CT Rct, it is not certain that the Vk expression corresponding to the relay used, will be complied with. To illustrate this case, let us take the example of a high impedance busbar differential protection. The CT supplied is a 2000/5 where Vk = 400 V and Rct = 2.5 . For the relay used, the expression to be satisfied is: Vk u 200 Rct + 20, i.e. 520 V. The Vk = 400 V is not sufficient! More serious still, the requirement of too high a Vk may lead to the manufacturing of a nonstandard CT (see the first two consequences above) requiring a specially designed stabilising resistance and an overvoltage limitor as well as the use of a deeper panel! c Error on the through current This error is very common. Let us take the example of a high impedance differential protection where the switchboard Isc is taken into
a - The CT manufacturer's understanding
200/1 1000/1 Cl. X
200/1 1000/1 15 VA Cl. 0.5
200/1 1000/1 10 VA-5P20
b - The real need
1000/1 Cl. X
200/1 15 VA Cl. 0.5
200/1 10 VA-5P20
Fig. 7 : example of poor understanding between the customer and the CT manufacturer.
account instead of the maximum through current. The aim is to protect a motor, the CTs have a ratio of 100/1. v Result obtained with the through current (7 I n of CT): Vk u 14 (Rct + 2 RL). v Result obtained with switchboard Isc (Isc = 40 kA): Vk u 800 (Rct + 2 RL) It is not necessary to go into too much detail to understand the importance of choosing the right parameter! The table in figure 9 gives the through current values to be taken into account when the through current is the CT calculation base (see Cahier Technique no. 194). c With line differential protections, taking into account the pilot wires in the calculation of Rwiring.
In point of fact, RL is given by the wiring linking the CTs to the relay located on the same side (end) of the line (see fig. 10 ). You must not take into account the length of the pilot wires which run from one end to the other of the protected line. Reminders With respect to high impedance differential protections: c For the calculation of min. Vk, take account of the through current (see fig. 9). c Calculation of the stabilising resistance Rst is a function of min. Vk and of the relay setting current. c Calculation of peak voltage (Vp) is based on the internal Isc of the protected zone and on the real Vk of the CT.
Applications BB differential protection Motor differential protection Generator differential protection
Through Imax Switchboard real Isc
Excess through
Comments Take real Isc if no increase possible. Else take Ith If you know neither the starting I nor the motor In, take 7 x In (CT) X = generator subtransient reactance as a %. If unknown, we assume X % u 15 i.e. 100/15 = 6.67 (7 is taken by excess) Pa = upstream shortcircuit power and Pt = power limited by transformer Zsc % = transformer short-circuit impedance Switchboard Ith by default
Imax
Switchboard Ith
Motor starting I
7 In (motor) otherwise 7 x In (CT) 7 In (generator) otherwise 7 x In (CT)
Generator Isc contribution only, i.e. In (100 / X)
Restricted earth fault differential protection
Isc seen at the CT primary for a fault at the transformer secondary, i.e. Isc = Psc /Ue Psc = (Pt .Pa) / (Pt + Pa) Isc at 80 % of line
If Pa unknown, we take Psc = Pt Pt = Pn (100 / Zsc)
Line differential protection
Switchboard Ith by default
Fig. 9 : determining the through current properly.
RL
Pilot wires (several kilometers)
RL
Fig. 10 : RL is given by the wiring between the CT and the relays located on the same side of the line.
2.4 And if the CT cannot be manufactured?

When a CT manufacturer says that he is unable to manufacture the requested CT, nine times out of ten this is because the CT has been incorrectly specified. To eliminate all the cumulated safety margins taken by all the people involved, the CT must be redefined on the basis of real needs: c real currents in the installation, c types of protection, required power, c discrimination study and protection plan (settings). This approach must be adopted whenever the specification leads to a non standard CT. Costs, lead times and safety are the factors at stake. Let us take an example: We have calculated the class X of a 1000/5 CT for a generator differential protection, assuming that X = 15 %. Not knowing the exact characteristics of the generator, we have assumed that the generator In equals the CT In .
100 In of the CT 15 i.e. Vk u 2 x 6.7 x 5 (Rct + 2 RL)
If the CT is declared impossible to manufacture, a solution, i.e. a compromise, must be found between all those involved. There is always a way out, which can be found with the help of specialists. As an example, here are a few leads: c play on the equivalences between CTs (see next section), c reduce the safety coefficient (for instance 2 to 1.5 for an overcurrent protection), c change the secondary from 5 to 1 A (see fig. 11 ), c increase wiring cross-section, c overrate the CTs (primary In), c move the relay with respect to the CT, c use matching CTs with low consumption, c and so on. The overrating of a CT can solve a manufacturing problem Let us take two examples: c A 100/1 CT with a load of 2.5 VA requires an ALF of 25 for an overcurrent protection. The standard CTs proposed are 2.5 VA-5P20. If a CT with a ratio of 150/1 - 2.5 VA-5P20 is proposed, the ALF need will be reduced in the CT primary ratio, i.e. necessary ALF = 25 x (100/150) = 16.7. An ALF of 20 is thus sufficient! c If the class X requested for a CT is proportional to a through current or a primary Isc, these values are multiplied by the CT ratio; thus, the required knee point voltage will be less for an overrated CT, unless its increasing resistance Rct starts to neutralise the ratio benefit. In all cases, it will be possible to create a higher knee point voltage than with a CT of lower ratio, as it is proportional to the number of secondary turns. Globally, the chance of obtaining workable characteristics will be greater. The same reasoning can be made for a 1 A secondary CT compared with a 5 A CT. However, the factor gain of 5 obtained on the formula by the CT ratio is often completely erased, if not reversed, by a far greater increase of secondary winding resistance.
This results in: I f =
We have rounded off 6.7 as 7. We assumed: 2 RL = 300 m of 2.5 mm2, i.e. 2.4 hence: Vk u 70 Rct + 168. Since this CT requires two other windings, this value could not be achieved using the standard mould. The solution was found by using 4 mm2 connections and by requesting the generator characteristics. Then: 2 RL = 1.5 Generator In = 830 A. X = 25 %, hence: Vk u 2 x
830 x 100 5 x (Rct +1.5) 25 1000
Vk = 33.2 Rct + 50 The difference is marked and shows the importance of obtaining the right information and of knowing the safety margins.
Length (m) Wiring losses (VA) for: In = 1 A In = 5 A
5 0.04 1
10 0.08 2
20 0.16 4
50 0.4 10
100 0.8 20
200 1.6 40
400 3.2 80
Fig. 11 : losses in wiring for a 2.5 mm2 cross-section (8 /km at 20 C). With 1 A, losses are 25 times less.
Indeed, the space required for the number of turns x 5 results in reduction in wiring crosssection, thus naturally increasing its linear resistance. The new resistance can thus be amply multiplied by 10 with respect to the 5 A CT. c If you are tempted to impose a CT overrating, you must check the repercussions of the change in ratio. For example: v If the CT supplies a pilot wire differential protection, you must ensure that the corresponding CT at the other end of the line has also the same ratio change. v In the case of a restricted earth protection, you must ensure that: - the CT installed on the neutral point is also modified, - the earth fault detection is not compromised by the overrating. v For all protection types, you must check that setting of the protection is still possible. Optimisation of the differential protection CTs Let us take the example of a transformer differential protection (see fig. 12 ). c Calculating the through current. The transformer impedance limits the through
Isc = 31.5 kA (Psc 600 MVA)
300/5
RL1
11 kV Distance 500 m 5 MVA Zsc = 8 % Rr 3.3 kV 1000/5 RL2 87T
RL3
5 x 100 = 62.5 MVA) . 8 v Short-circuit power becomes:

current to: (Psct =
600 x 62.5 = 56.6 MVA . 600 + 62.5 v The through current at the secondary is: - 11 kV side: Psc =
Fig. 12 : transformer differential protection.
I f1 =
5 56.6 x 106 x = 49.5 A , 300 11 3 x 103
- 3.3 kV side:
I f2 =
5 56.6 x 106 x = 49.5 A , 1000 3.3 3 x 103
c Formulas to be applied for Vk (standard protection): v Calculating the matching CTs with a ratio of: Vka mini = 5 5/ 3
4 I f1 Rsr + 3 RL3 + Rp 3 v Calculating main CTs - 11 kV side: 300/5
)]
5 5 3
Vk p1 min = 4If1 (Rct + RL1 + Rsp) + Vka mini - 3.3 kV side: 1000/5
Vk p2 min = 4 I f2 (Rct + RL2 + Rr )
c Optimisation approach. Let us examine the case of the 300/5 CT placed in the 11 kV switchboard. v First hypothesis 5 is the one proposed as The matching CT 5/ 3 standard by the relay manufacturer. It is located with the relay on the 3.3 kV side. Wiring is 2.5 mm2 throughout. RL1 = 4 RL2 = 0.08 RL3 = 0.024 Rsr = 0.25 , secondary winding resistance of the matching CT, Rsp = 0.15 , primary winding resistance of the matching CT, Rp = 0.02 , relay resistance. We find: - Vka mini = 43.7 V (standard Vka= 58 V), - Vkp1 mini = 198 Rct + 847
v Second hypothesis The same as the first, except that RL1 wiring is in 10 mm2, hence RL1 = 1 The result is: - Vkp1 mini = 198 Rct + 243 v Third hypothesis The matching CT is on the 11 kV side as well as the relay: RL1 = 0.08 - Vkp1 mini = 198 Rct + 61 v Fourth hypothesis Same as the third hypothesis, except for the matching CT which is not standard, but which is imposed on the CT manufacturer where: Rs i 0.1 , Rp i 0.1 , which results in: - Vka mini = 26.5 V - Vkp1 mini = 198 Rct + 41
We observe that by modifying the wiring cross-section, the position and the characteristics of the matching CT, the gain on the minimum needed Vk of the 300/5 CT is around 800 V. The same approach adopted for the 1000/5 CT, placed on the 3.3 kV side, yields results that are fairly similar concerning Vk. However, in view of the fact that a 1000/5 CT is easier to manufacture than a 300/5 CT, it is more advantageous to place the relay and the matching CT on the 11 kV side. If a 1 A CT is used, the same hypotheses as above enable a move from: - Vkp1 = 39.6 Rct + 249 to Vkp1 = 39.6 Rct + 17 The 1 A CTs may be easier to manufacture than the 5 A CTs, but all depends on the relative weight of the Rct and the wiring in the Vk expression.
3 Equivalence of the various possible definitions of the same CT
In many cases, you need to know how to juggle with the various CT characteristics; ratio, power, class, ALF. The reason for this is not only in order to get out of a tricky position, but also to be able to use standard CTs that are available, less costly and tested.
This section thus aims to show how CT characteristics can be manipulated. First, however, it should be pointed out that the only CT constants are its magnetising curve and resistance and, naturally, its ratio.
3.1 How to move from Pn1-5Pk1 to Pn2-5Pk2

Vs1 and Rct are fixed.
c if Pn2 is imposed, we shall obtain:

k2 = (Pi + Pn1 ) (Pi + Pn2 ) k1 or k 2 =
2 (Rct In + Pn1 ) 2 (Rct In + Pn2 )
Pn Pn Vs1 = Rct + 21 k1 In = Rct + 22 k 2 In In In Pn = Rct + 2i k i In In

2 Knowing that Pi = Rct In (internal ohmic losses of the CT), we obtain: (Pi + Pn1 ) k1 = (Pi + Pn2 ) k 2 = (Pi + Pn3 ) k 3 .
k1
c if k2 is imposed, we shall obtain:
Pn2 =
k k1 Pn1 + 1 - 1 Pi k2 k2
k k1 2 Pn1 + 1 - 1 Rct In k2 k2
or else:
Pn2 =
Sometimes, some people ignore Pi: this is a serious error as Pi can be roughly of the same value, if not higher, than Pn.
3.2 How to move from Pn1-5Pk1 to Pn2-10Pk2

We have: c if k2 is imposed, we shall have:
Pn2 = 1.9 k1 1.6 k 2 1.9 k 1 - 1 R I2 Pn1 + ct n 1.6 k 2
Pn Vs1 = Rct + 21 k1 In In Pn Vs2 = Rct + 22 k 2 In In

But:
Vs1 = 1.6 Vs2 1.9
If you wish to move from a 10P to a 5P definition, the above expressions apply: just reverse the induction ratio.
c if Pn2 is imposed:
k2 =
2 1.9 (Rct In + Pn1 ) 2 1.6 (Rct In + Pn ) 2
k1 or
k2 =
1.9 (Pi + Pn1 ) 1.6 (Pi + Pn )

2
k1
3.3 What is the Vk of a CT: Pn-XPk

How to move from Pn-5Pk to Vk We have seen that: Pn Vs1 = Rct + 21 k In In and Vk = 1.4 1.6 Vs1 How to move from Pn-10Pk to Vk We also have:
Vk =
hence
1.4 Vs 1.9 2
hence Vk =
1.4 Pn Rct + 2 k In 1.6 In
Vk =
1.4 Pn Rct + 2 k In In 1.9
3.4 How to move from a class X (Vk, Rct) to a class 5P: Pn-5Pk
c Assuming that k is imposed, we shall obtain: c Assuming that Pn is imposed, in this case:
k = Vk In 1.6 2 1.4 Pn + Rct In
Pn =
1.6 Vk In 2 - Rct In 1.4 k
If the result is negative, this means that the necessary ALF cannot be obtained with this CT as its internal losses are too great.
Note: In this section, the induction levels: 1.4 - 1.6 - 1.9 are given as examples since they vary from one manufacturer to another.
4 Conclusion
This Cahier Technique completes Cahier Technique no. 194 in its aim to increase awareness of all those involved in the process from design of an electrical network to implementation of protections, of the time wasted and the financial losses resulting from incorrect specification of CTs. In particular, it gives some examples of errors not to make and leads for solutions when the original specification is not satisfactory or when you apparently come up against a dead end. It stresses that although communication with the CT manufacturer and particularly knowledge of induction levels enables a solution to be found by playing with equivalences, it is by identifying the exact needs at all stages in the process, that the optimised solution can be found. We hope that you will find this document useful.
Bibliography
Standards c IEC 60185: Current transformers Characteristics. c IEC 60044-1: Instrument transformers - Part 1: Current transformers (replaces the IEC 185). c IEC 60044-8: Instrument transformers - Part 8: Electronic current transformers. c NF C 42-502: Measuring instruments. Current transformers. Characteristics. c BS 3938 (replace by BS 7626): Specification for current transformers. Schneider Electrics Cahiers Techniques c Protection des machines et des rseaux industriels HT. P. ROCCIA, Cahier Technique no. 113. c Current transformer for HV protection. M. ORLHAC, Cahier Technique no. 164. c Protection of industrial and tertiary MV networks. A. SASTRE, Cahier Technique no. 174. c Directional protection equipment. P. BERTRAND, Cahier Technique no. 181. c Dynamic stability of industrial electrical networks. B. DE METZ NOBLAT and G. JEANJEAN, Cahier Technique no. 185. c Current transformers: How to specify them. P. FONTI, Cahier Technique no. 194. Various works c Guide de lingnierie lectrique ELECTRA - 07.86 c Protection des rseaux lectriques Ch. PREVE - Ed. Hermes - 06.98

Ferroresonance
Ph. Ferracci
s Merlin Gerin s Modicon s Square D s Telemecanique
n 190
Ferroresonance
Philippe FERRACCI Graduated from the Ecole Suprieure dElectricit in 1991, where he defended a PhD on the resonant earthed system in cooperation with EDF - Direction des Etudes et Recherches. He joined the Groupe Schneider in 1996, and is currently conducting advanced studies in the electrotechnical and electrical power system fields.
ECT190, first issued March 1998
Lexicon
CVT: Capacitor Voltage Transformer (IEC 186) Voltage transformer comprising a capacitor divider unit and an electromagnetic unit designed and interconnected so that the secondary voltage of the electromagnetic unit is substantially proportional to and in phase with the primary voltage applied to the capacitor divider unit. Irrational number (not fractional) Number which cannot be expressed as the quotient of two integers ( 2 , 3 , ...). PIM: Permanent Insulation Monitor A device designed to indicate (by an audible or visual signal) the appearance of the first earth fault of a live part. VT: Voltage transformer (IEC 50) Instrument transformer in which the secondary voltage, in normal conditions of use, is substantially proportional to the primary voltage and differs in phase from it by an angle which is approximately zero for an appropriate direction of the connections.
Ferroresonance
Ferroresonance is a non-linear resonance phenomenon that can affect power networks. The abnormal rates of harmonics and transient or steady state overvoltages and overcurrents that it causes are often dangerous for electrical equipment. Some unexplained breakdowns can be ascribed to this rare, non-linear phenomenon. The purpose of this Cahier Technique is to help readers understand ferroresonance. The methods described ensure credible prediction and evaluation of the risk of ferroresonance in existing and future installations. Practical solutions designed to avoid or eliminate ferroresonance are also given.
Contents
1 Introduction 2 Understanding ferroresonance 3 Identifying ferroresonance 2.1 Resonance 2.2 Ferroresonance 3.1 Diagnosis elements 3.2 Examples of electrical power system situations favourable to ferroresonance 4 Preventing or damping ferroresonance 5 Studies for predicting and understanding 4.1 Practical solutions 4.2 Practical evaluation criteria 5.1 A study example 5.2 Modelling, a mathematical approach 5.3 Summary 6 Conclusion Appendix 1: Bibliography Appendix 2: Selection guide for VT load resistances p. 4 p. 5 p. 5 p. 10 p. 10 p. 16 p. 19 p. 20 p. 23 p. 25 p. 26 p. 27 p. 28
1 Introduction
The term Ferro-rsonance , which appeared in the literature for the first time in 1920, refers to all oscillating phenomena occurring in an electric circuit which must contain at least: c a non-linear inductance (ferromagnetic and saturable), c a capacitor, c a voltage source (generally sinusoidal), c low losses. Power networks are made up of a large number of saturable inductances (power transformers, voltage measurement inductive transformers (VT), shunt reactors), as well as capacitors (cables, long lines, capacitor voltage transformers, series or shunt capacitor banks, voltage grading capacitors in circuit-breakers, metalclad substations). They thus present scenarios under which ferroresonance can occur. The main feature of this phenomenon is that more than one stable steady state response is possible for the same set of the network parameters. Transients, lightning overvoltages, energizing or deenergizing transformers or loads, occurrence or removal of faults, live works, etc...may initiate ferroresonance. The response can suddenly jump from one normal steady state response (sinusoidal at the same frequency as the source) to an another ferroresonant steady state response characterised by high overvoltages and harmonic levels which can lead to serious damage to the equipment. A practical example of such behaviour (surprising for the uninitiated) is the deenergization of a voltage transformer by the opening of a circuit-breaker. As the transformer is still fed through grading capacitors accross the circuit-breaker, this may lead either to zero voltage at the transformer terminals or to
permanent highly distorted voltage of an amplitude well over normal voltage. To prevent the consequences of ferroresonance (untimely tripping of protection devices, destruction of equipment such as power transformers or voltage transformers, production losses,...), it is necessary to: c understand the phenomenon, c predict it, c identify it and c avoid or eliminate it. Little is known about this complex phenomenon as it is rare and cannot be analysed or predicted by the computation methods (based on linear approximation) normally used by electrical engineers. This lack of knowledge means that it is readily considered responsible for a number of unexplained destructions or malfunctionings of equipment. A distinction drawn between resonance and ferroresonance will highlight the specific and sometimes disconcerting characteristics of ferroresonance. Practical examples of electrical power system configurations at risk from ferroresonance are used to identify and emphasise the variety of potentially dangerous configurations. Well-informed system designers avoid putting themselves in such risky situations. A predictive study should be undertaken by specialists should doubts persist concerning limit, inevitable configurations. Appropriate numerical analysis tools enable prediction and evaluation of the risk of ferroresonance in a power system for all possible values of this systems parameters in normal and downgraded conditions. Practical solutions are available to prevent or provide protection against ferroresonance.
2 Understanding ferroresonance
2.1 Resonance
Resonance is a phenomenon encountered on electrical power systems of all voltage levels. It can be observed for example in a resonant earthed system (Petersen coil) used to minimise MV single phase to earth fault currents, or it can be responsible for dielectric or thermal destruction or premature ageing of electrical equipment due to overvoltage or overcurrent (harmonic resonance,...). As regards series resonance and in sinusoidal conditions ( U = E cos(nt) ), the vectorial relationship between voltages can be written as: r r r r U = UR + UL + UC (see fig. 1 ). In the specific case of resonance, the voltages at the terminals of the capacitor and the inductance are compensated and the circuit is said to be in a resonance situation. The pulsation n for which this resonance appears is such that L C n2 = 1 . The amplitude of current I is then equal to: E I = R This current can be very high. Voltage amplitude at the capacitor (and inductance) terminals is equal to k.E . The quality factor k is expressed as: L n 1 k = = R R C n According to the value of k, voltage amplitude UL ( = UC ) can be less than or greater than amplitude E of excitation voltage U.
UR= U UC UL UR
I
U
UL
UC
I
Fig.1: series resonance.
n coincides with an harmonic pulsation n 0 (0 is the system pulsation) generated by
Harmonic resonance occurs when the pulsation
certain machines (variable speed drives, static rectifiers...). Harmonic resonance can also have harmful effects on electrical equipment and, as such, needs to be controlled [6], [7].
2.2 Ferroresonance
The main differences between a ferroresonant circuit and a linear resonant circuit ( 2.1) are for a given : c its resonance possibility in a wide range of values of C, c the frequency of the voltage and current waves which may be different from that of the sinusoidal voltage source, c the existence of several stable steady state responses for a given configuration and values of parameters. One of these states is the expected normal state (in the linear assumption), whereas the other unexpected abnormal states are often dangerous for equipment. Initial conditions (initial charge on capacitors, remanent flux in the core of transformers, switching instant) determine which steady state response will result.
Physical approach A study of the free oscillations of the circuit in figure 2a illustrates this specific behaviour. Losses are assumed negligible and the simplified magnetization curve (i) of the ironcore coil is that represented in figure 2b . Despite these simplifying assumptions, the corresponding waveforms (see fig. 2c ) are typical of a periodic ferroresonance. Originally, voltage at the capacitance terminals is assumed equal to V0. c At the instant t0 switch K closes. A current i is created, and oscillates at the pulsation 1 = 1/ LC The flux in the coil, and voltage V at the capacitor terminals are then expressed as: = (V0 / 1)sin1t ; v = V0 cos1t .
a - Schematic diagram b - Simplified characteristic (i)
K C v R
max sat
L
LS
Imax
-sat
c - Voltage v, current i and flux as a function of time
v V0 V2
c If V0 / 1 > sat , at the end of time t1, the flux reaches the saturation flux sat , voltage v is equal to V1 and the inductance of the saturated coil becomes Ls. As Ls is very small compared with L, the capacitor suddenly discharges across the coil in the form of an oscillation of pulsation 2 = 1/ L s C . The current and flux peak when the electromagnetic energy stored in the coil is equivalent to the electrostatic energy 1/ 2 CV12 restored by the capacitor. c At instant t2 the flux returns to sat , the inductance reassumes the value L and, since the losses have been ignored, voltage v, which has been reversed, is equal to -V1. c At instant t3 the flux reaches - sat and voltage v is equal to -V2. As 1 is in practice very small we can consider V2 V1 V0 . Consequently period T of the oscillation is included between 2 LC in the non-saturated case, and 2 L s C + 2( t 3 t 2 ) in the saturated case (where t 3 t 2 2 sat / V0 ). The corresponding frequency f (f = 1/T) is thus 1 1 such that: 2 LC < f < 2 L C
s
t0
t1
t2
t3
-V1 i
Imax
sat
max
Fig. 2: free oscillations of a series ferroresonant circuit
This initial frequency depends on sat i.e. on the non-linearity and the initial condition V0. In practice, due to the losses R i2 in the resistance R (mainly at each voltage reversal), the amplitude of voltage V decreases (V2 < V1 < V0). Since the flux variation during the non-saturated period (t3 - t2) is such that = 2 sat =
t 2 vdt , decrease of v
t3
the frequency of the oscillations, as it decreases, can lock at the frequency of the source (if the initial frequency is greater than the power frequency) or even at a sub-multiple frequency of the source frequency (if the initial frequency is smaller than the power frequency). This shows that unlike linear resonance ( 2.1), for a given inductance, resonance can occur no longer for a specific value of C but for a wide range of values of C.
results in a reduction in frequency. If the energy losses (joule losses, core losses, ...) are supplied by the voltage source in the system,
c The main differences between a ferroresonant circuit and a linear resonant circuit are: v possibility of resonance in a wide range of parameter values, v the frequency of the voltage and current waves may vary from the frequency of the sinusoidal source, v the existence of several stable steady states responses for a given configuration and parameter values. c Ferroresonance can be single-phase (as above). It is said to be three-phase if there is non-linear magnetic coupling between phases, and three-single phase if there is no coupling between the three non-linearities. Ferroresonance can be series or parallel. Main characteristics Thanks to the appropriate methods defined hereafter in this document, a study of the circuit in figure 3a enables the curves shown in figures 3b and 3c to be plotted. These curves illustrate the characteristics of ferroresonance: c Sensitivity to system parameter values, jump phenomenon The curve in figure 3b describes the peak voltage VL at the terminals of the non-linear inductance as a function of peak amplitude E of the sinusoidal voltage source. By gradually building up peak amplitude E from zero, the curve in figure 3b shows that there are three possible different types of behaviour according to the value or E, as well as the jump phenomenon: v For E = E1, the solution (M1n) is unique, and corresponds to the normal state (obtained in the linear assumption), v For E = E2, there are three solutions (M2n, M2i, M2f), two of which are stable (M2n and M2f). M2n corresponds to the normal state, whereas M2f corresponds to the ferroresonant state. The dotted part of the curve (which cannot be obtained in practice) corresponds to unstable states. v For E = E2, voltage VL suddenly moves from the point M2 to the point M2 (the jump phenomenon). The point M2 is known as a limit point, v For E = E3, only the ferroresonant state (M3f) is possible. v When the value of E decreases from E3, the solution suddenly moves from the point M1 (second limit point) to the point M1. The jump phenomenon, characteristic of ferroresonance, can also be obtained by considering another system parameter (for example, resistance R or capacitance C). A small variation in the value of a system parameter or a transient can cause a sudden jump between two very different stable steady states.
a - Basic series ferroresonance circuit
VC
R E
C VL
b - Sensitivity to system parameters and the jump phenomenon
VL M2f M1 M2i M1n M'1 E1 E"2 M2n M2 M'2 M3f
E2
E'2
E3
c - Sensitivity to initial conditions

VC
M2f
C M2n t
M02
M01
Fig. 3: illustration of ferroresonance characteristics.
c Sensitivity to initial conditions Whether M2n or M2f is obtained depends on the initial conditions. Figure 3c illustrates the trajectories of the transient of pairs (,Vc) as a function of time for different initial conditions (M01 and M02). Curve C describes a boundary. If the initial conditions (residual flux, voltage at capacitor terminals) are on one side of the boundary, the solution converges to M2n. If the initial conditions are on the other side, the solution converges to M2f. As the point M2i
a - Fundamental mode v(t) V(f) v
Ferroresonant mode (1 point) t Normal mode
f0
3f0
nf0
b - Subharmonic mode v(t) V(f) v
(n points) t
nT f0/n f0/3 f0 f i
c - Quasi-periodic mode
v(t) V(f) v (Closed curve)
f2-f1
f1 f2 3f1-f2
nf1+mf2 f
d - Chaotic mode
v(t)
V(f)
t Strange attractor
Fig. 4: illustration of ferroresonance characteristics.
belongs to the boundary, the steady state effectively reached around this point is extremely sensitive to the initial conditions. Classification of ferroresonant modes Experience of waveforms appearing on power systems, experiments conducted on reduced system models, together with numerical simulations, enable classification of ferroresonance states into four different types. This classification corresponds to the steady state condition, i.e. once the transient state is over, as it is difficult for a ferroresonant circuit to distinguish the normal transient state from ferroresonant transient states. However, this in no way implies that transient ferroresonance phenomena do not present a risk for electrical equipment. Dangerous transient overvoltages can occur several system periods after an event (for example following energizing of an unloaded transformer) and persist for several power system cycles. The four different ferroresonance types are: c fundamental mode, c subharmonic mode, c quasi-periodic mode, c chaotic mode. The type of ferroresonance [13] can be identified: c either by the spectrum of the current and voltage signals, c or by a stroboscopic image obtained by measuring current i and voltage v at a given point of the system and by plotting in plane v, i the instantaneous values at instants separated by a system period. The characteristics of each type of ferroresonance are defined below. c Fundamental mode (see fig. 4a ) Voltages and currents are periodic with a period T equal to the system period, and can contain a varying rate of harmonics. The signal spectrum is a discontinuous spectrum made up of the fundamental f0 of the power system and of its harmonics (2f0, 3f0 ...). The stroboscopic image is reduced to a point far removed from the point representing the normal state.
c Subharmonic mode (see fig. 4b ) The signals are periodic with a period nT which is a multiple of the source period. This state is known as subharmonic n or harmonic 1/n. Subharmonic ferroresonant states are normally of odd order. The spectrum presents a fundamental equal to f0/n (where f0 is the source frequency and n is an integer) and its harmonics (frequency f0 is thus part of the spectrum). A stroboscopic plotted line reveals n points. c Quasi-periodic mode (see fig. 4c ) This mode (also called pseudo-periodic) is not periodic. The spectrum is a discontinuous spectrum whose frequencies are expressed in the form: nf1+mf2 (where n and m are integers and f1/f2 an irrational real number). The stroboscopic image shows a closed curve. c Chaotic mode (see fig. 4d ) The corresponding spectrum is continuous, i.e. it is not cancelled for any frequency. The stroboscopic image is made up of completely separate points occupying an area in plane v, i known as the strange attractor. To conclude: c Ferroresonance is a complex phenomenon in which: v there is several steady states for a given circuit, v the appearance of these states is highly sensitive to system parameter values, v the appearance of these states is highly sensitive to initial conditions. c Small variations in the value of a system parameter or a transient may cause a sudden jump between two very different steady states and initiate one of the four ferroresonance types. The modes most commonly encountered are the fundamental and subharmonic ones. c Abnormal rates of harmonics, overvoltages/ currents, either as stable oscillation or as transients caused by ferroresonance, often represent a risk for electrical equipment. c Steady state ferroresonance is sustained by the energy supplied by the power system voltage.
3 Identifying ferroresonance
3.1 Diagnosis elements

Ferroresonance is frequently accompanied by some of the symptoms described below: c high permanent overvoltages of differential mode (phase-to-phase) and/or common mode (phase-to-earth), c high permanent overcurrents, c high permanent distortions of voltage and current waveforms, c displacement of the neutral point voltage, c transformer heating (in no-load operation), c continuous, excessively loud noise in transformers and reactances, c damage of electrical equipment (capacitor banks, VT, CVT,...) due to thermal effect or insulation breakdown. A characteristic symptom of VT destruction by ferroresonance is a destroyed primary winding and an intact secondary winding. c apparent untimely tripping of protection devices. Some of these symptoms are not specific to the ferroresonance phenomenon. For example permanent displacement of the neutral point of an unearthed neutral system may be the consequence of a single phase-to-earth fault. An initial diagnosis is simplified by comparing the curves of any recordings taken with typical ferroresonance waveforms, specified in the above paragraph (see fig. 4 ). Faced with the difficulty of diagnosis (no recordings, a number of possible interpretations of the symptoms) the first reflex is to analyse system configuration while the symptoms are present, together with the events preceding them (transformer energizing, industrial process specific operating phase, load rejection...) which might initiate the phenomenon. The next step is to determine whether the three conditions necessary (but not sufficient) for ferroresonance to be present, are united: c simultaneous presence of capacitances with non-linear inductances, c existence in the system of at least one point whose potential is not fixed (isolated neutral, single fuse blowing, single phase switching....), c lightly loaded system components (unloaded power or instrument voltage transformers...) or low short-circuit power sources (generators). If any one of these conditions is not verified, ferroresonance is highly unlikely. Otherwise more extensive investigations are required. A predictive study may be carried out by specialists, which will require implementation of methods defined later on in this document. A comparison with examples of typical power system situations favourable to ferroresonance may simplify identification of configurations at risk.
3.2 Examples of electrical power system situations favourable to ferroresonance

Due to the multitude of various sources of capacitances and non linear inductances in a real power network and the wide range of operating conditions, configurations under which ferroresonance can occur are endless. Experience has, however, made it possible to list the main typical configurations that may lead to ferroresonance. A few standard examples are given below. Voltage transformer energized through grading capacitance of one (or more) open circuit-breaker(s). In EHV, certain switching operations (padlocking a bus coupler or switched busbar circuit-breaker, removal of a fault on a busbar section...) can drive voltage transformers (VT) connected between phases and earth into ferroresonance.
These configurations can be illustrated by the circuit in figure 5 . Opening of circuit-breaker D initiates the phenomenon by causing capacitance C to discharge through the VT which is then driven into saturation [11]. The source delivers enough energy through the circuit-breaker grading capacitance Cd to maintain the oscillation. Capacitance C corresponds to all the capacitances to earth of the VT and the
connection supplied by means of the grading capacitances of the open circuit-breaker(s). Ferroresonance is of the subharmonic type. Voltage transformers (VT) connected to an isolated neutral system This earthing system can be chosen, can result from the coupling of an isolated neutral emergency source or from a loss of system earthing. Transient overvoltages or overcurrents due to switching operations on the power system (load rejection, fault-clearing ...) or to an earth fault, can initiate the phenomenon by driving into saturation the iron core of one or two of the VTs of the parallel ferroresonant circuit in figure 6 . Ferroresonance is then observed both on the phase-to-earth voltages and on the neutral point voltage (VN). The neutral point is displaced and the potential of one or two phases rises with respect to earth, which may give the impression of a single phase-to-earth fault in the system. Overvoltage values may exceed normal phaseto-phase voltage under steady state condition, and cause dielectric destruction of the electrical equipment. Depending on the relative values of the magnetizing inductance of the VT and the capacitance C0, ferroresonance is fundamental, subharmonic or quasi-periodic.
Cd
Cd Open circuit Circuit-breaker D
En
VT
Fig. 5: : ferroresonance of a voltage transformer connected in series with an open circuit-breaker.
Co VT
Co VT
Co VT
Co : system zero-sequence capacitance
Fig. 6: ferroresonance of a VT between phase and earth in an isolated neutral system.
Transformer accidentally energized in only one or two phases A few examples of configurations at risk are given in figure 7 . These configurations can occur when one or two of the source phases are lost while the transformer is unloaded or lightly loaded, as a result of a fuse blowing on an MV power system, of conductor rupture or of live works, for example when commissioning a remote controlled breaking cubicle (ACT). The capacitances can be in the form of capacitance of underground cable or an overhead line supplying a transformer whose primary windings are wye connected with
isolated or earthed neutral, or delta connected. For example the series ferroresonant circuit is made up of the connection in series of the phase to earth capacitance (between circuit breaker and transformer) of the open phase and the magnetizing impedance of the transformer. The modes are fundamental, subharmonic or chaotic. The phase-to-phase and phase-to-earth capacitances, the primary and secondary windings connections, the core configuration (three singlephase, free flux or forced flux), the voltage source system neutral earthing (solidly earthed, earthed,
Fig. 7: examples of unbalanced systems at risk from ferroresonance.
isolated) and the supply mode (one or two phases energized) are all factors involved in the establishment of a given state. Isolated primary neutral is more susceptible to ferroresonance. To avoid such risks, use of multi-pole breaking switchgear is recommended. Voltage transformers and HV/MV transformers with isolated neutral Ferroresonance may occur when the HV and MV neutrals are isolated, and unloaded VTs are connected on the MV side between phase and earth (see fig. 8a ).
When an earth fault occurs on the HV side upstream from the substation transformer, the HV neutral rises to a high potential. By capacitive effect between the primary and secondary, overvoltages appear on the MV side, and may trigger ferroresonance of the circuit made up of the voltage source E0, the capacitances Ce and C0 and the magnetizing inductance of a VT (see fig. 8b ). Once the HV fault has been removed, the voltage of the HV neutral due to a natural unbalance of the system, may be enough to sustain the phenomenon. Ferroresonance is fundamental.
a - Faulty system
HV
MV
Ce
Co
Co
Co VT VT VT
b - Equivalent diagram
Ce
Eo Co Co Co
E0 : zero-sequence voltage on the HV side Ce : capacitance between HV and MV winding C0 : zero-sequence capacitance of the MV power system
Fig. 8: ferroresonance of a VT between phase and earth with an isolated neutrals source transformer.
Power system earthed through a reactor The two configurations in figure 9 are susceptible to ferroresonance between an inductance connected between neutral and earth, and the capacitances to earth of the network. c As regards LV power systems with isolated neutral, standards recommend (IEC 364) or stipulate (NF C 15-100) implementation of a Permanent Insulation Monitor (PIM). Some PIMs measure insulation impedance of a power system by injecting direct current between the system and the earth. Their impedance is mainly inductive (low impedance for the direct current, and high impedance at power frequency). They can be a factor contributing to ferroresonance. Overvoltages may cause sufficient potential rise of the neutral point to initiate ferroresonance between the inductance of the PIM and the capacitances to earth of the network (see fig. 9a ). c In MV, in order to limit earth-fault currents and help the fault to self-extinghish, a coil of inductance L (such that 3 L C0 02 = 1 where C0 is the zero-sequence capacitance of the MV power system and 0 the power pulsation) is connected between the MV neutral of an HV/MV transformer and earth (Petersen coil). Excitation and start of resonance of the circuit consisting of series connection of inductance L and capacitance 3 C0 may occur in the following cases:
v HV neutral of the HV/MV transformer earthed, and HV fault flowing through the earth conductor of the substation, v iron core saturation of the HV/MV transformer, v transformer design dissymmetry, v natural dissymmetry of the capacitances (C1, C2, C3 in figure 9b ) to earth. This may result in saturation of the iron coil, thus initiating or sustaining ferroresonance. Transformer supplied by a highly capacitive power system with low short-circuit power Ferroresonance may occur when an unloaded power transformer is suddenly connected to a low short-circuit power source compared with transformer rated power through an underground cable or a long overhead line. This is the case, for example, on return to service in an MV (underground cable) urban or industrial power networks, but also in very extended rural public MV power networks (see fig. 10 ) or where underground cables are increasingly used (reliability and aesthetics). This parallel ferroresonance (capacitance parallel-connected on the transformers magnetizing inductance) is normally threephase, of the fundamental or of the quasiperiodic type. In short: c Configurations under which ferroresonance can occur are endless. c There are many different types of ferroresonance: single-phase, three-phase, common mode, differential mode.
a - Internal inductance of a PIM of an impedance between neutral and earth
b - Resonant earthing system
PIM
C1
C2
C3
C1
C2
C3
PIM = permanent insulation monitor
Fig. 9: ferroresonance in the case of a system earthed through a reactor.
c Experience, however, makes it possible to identify some configurations at risk which command some attention. These are: v voltage transformer connected between phase and earth on an isolated neutral system, v transformer fed through long and/or capacitive lines, v fuse protection, blowing of which results in non-multi pole breaking,
v unloaded or lightly loaded power or voltage instrument transformer. c The phenomena most likely to trigger ferroresonance are: v switching operation of capacitor banks and unloaded lines, v insulation faults, v lightning, v switching operation of unloaded transformers.
Source
Capacitive connection (long line or cable)
Unloaded power transformer
Fig. 10: equivalent diagram of unloaded power transformer supplied by a capacitive system.
4 Preventing or damping ferroresonance
A number of practical measures can be taken to prevent ferroresonance, whose overvoltages, overcurrents and distortions wave forms result in thermal and dielectric stresses which may be dangerous for electrical equipment (failure, reduction in performance and lifetime of insulators...). The various methods used are based on the following principles: c avoid, by proper design and/or switching operations, configurations susceptible to ferroresonance. This may involve prohibiting certain system configurations, and/or certain power system switching operations and /or certain switchgear; c ensure that system parameter values are not included (even temporarily) in an area at risk and if possible provide a safety margin with respect to danger areas;
c ensure that the energy supplied by the source is not sufficient to sustain the phenomenon. This technique normally consists of introducing losses which damp out ferroresonance when it occurs. Publication 71 of the IEC [2] states that temporary ferroresonance (and resonance) overvoltages shall be prevented or limited (by one of the above means). They shall not normally be considered as the basis for the surge arrester rated voltage, or for the insulation design unless these remedial measures are not sufficient. This means that the insulation co-ordination procedure does not normally take into account the overvoltage levels due to ferroresonance, and that, consequently, surge arresters (whose residual voltage is usually higher than the overvoltages due to ferroresonance) do not theoretically provide protection against ferroresonance.
4.1 Practical solutions

Application of these principles results in the recommendation of practical solutions, some of which are defined below for a few typical configurations susceptible to ferroresonance quoted in paragraph 3.2. c In well-designed VTs and CVTs, suitable design measures are taken in order to neutralise the phenonemon. The case of (single pole) VTs connected between phase and earth on an isolated neutral system is considered, in practice, as the case most favourable to ferroresonance (caused, for example, by overvoltages between sound phases and earth as a result of a single phase-to-earth fault). A fact which justifies in this case the implementation of special ferroresonance protection measures. The case of (two poles) VTs connected between phases can also be the source of ferroresonance phenomena when one of these VTs is likely to be supplied, even momentarily, on a single phase. This can occur, for example, during live work, during non-simultaneous operations on all three phases, on non-multi-pole breaking by fuse blowing or conductor rupture on one phase. The practical solutions are: v in isolated neutral systems, avoid wyeconnection of VT primaries with earthed (primary) neutral either by leaving the neutral of the VT primaries unearthed or using deltaconnection for the VTs, v if wye-connection of primaries with earthed neutral is used (for example to measure zerosequence voltage) in an isolated neutral system or on a system whose earthing system cannot be anticipated: - use design measures to make magnetic core work at lower induction value (around 0.4 to 0.7 T) so that overvoltages are unable to initiate ferroresonance, with at least a ratio of 2 between the saturation bend voltage and rated voltage, - introduce losses by means of one or more load resistances whose value is sufficiently low to effectively damp the phenomenon, while yet ensuring that total power consumption complies with required precision conditions. The following method can be used to compute load resistances values. It should be applied to each case individually: v VTs with one secondary winding: A damping resistor R is connected to the
secondary of each VT (see fig. 11 ), if consumption downstream of the VT is not sufficient. In this case the resistors continuously absorb power as soon as the VTs are energized. The recommended minimum values for the resistance R and power PR of this resistance are: R = U2 s , k Pt Pm PR = U2 s R
1 2 3 A A A
where: Us : rated secondary voltage (V), k : factor between 0.25 and 1 such that errors and service conditions remain within the limits specified by standard IEC 186 [1] (k Pt is for example around 30 W for a 50 VA rated output). Pt : VT rated output (VA), Pm : power required for measurement (VA). v In the case of VTs with two secondary windings (one secondary winding for measurement, and one residual voltage secondary winding also known as a tertiary winding), it is advisable to connect a resistance to the terminals of the open delta connected tertiary windings of the three transformers (see fig. 12 ). The advantage of this damping device is that it does not affect measurement accuracy or introduce losses in normal (balanced) operating conditions, but only in unbalanced conditions in order to damp the phenomenon. The recommended minimum values for the resistance R and power PR of this resistance are:
n Secondary
a R R R
Fig. 11: protection device against the risk of ferroresonance for VTs with one secondary.
1 2 3 A A A
( 3 Us ) 3 3 Us , PR = Pe R where: Us : rated voltage of the VT secondary, connected to the resistance (V) Pe : rated thermal burden of the VT secondary winding concerned by the resistance (VA).
2 2
R =
n Secondary
The rated thermal burden (in VA) is the apparent power that can be supplied by the VT to the secondary without exceeding the limits of normal temperature rise, without precision requirements. Resistance R must be chosen to ensure permanent dissipation of power PR. For example:
TT = 10000 : 3 - 100 : 3 - 100 : 3 V , Pe = 100 VA (Us = 100/3)
a da
a da
a da Residual voltage secondary
R =
3 3 (100/3) 100
dn
dn R
dn R : Damping resistor
= 57.7 ,
2
PR = (3 100/3) /57.7 = 173 W (standardised value immediately above 57.7 : 2 120 // , 2 140 W )
Fig. 12: protection device against the risk of ferroresonance for VTs with two secondaries.
c To prevent ferroresonance occurring on a transformer accidentally energized in only one or two phases (see fig. 7 ), practical solutions consist of: v lowering the value of the capacitance between the circuit breaker and transformer below its critical value by using, for example, a circuit breaker cubicle closer to the transformer or placing circuit breakers just upstream of the transformers and closing them only when voltage has been restored to all three phases, v avoiding use of transformers delivering an active power which is lower by 10% than its rated apparent power, v avoiding no-load energizing, v prohibiting single-phase operations or fuse protection, blowing of which results in single-pole breaking, v prohibiting live work on a cable-transformer assembly when the cable length exceeds a certain critical length, v resistance-earthing of the supply substation neutral, v solidly earthing of the neutral (permanently or only during energizing and deenergizing operations) of a transformer whose primary is wye-connected (available neutral). c Case of isolated neutral systems To avoid risk of ferroresonance with over-inductive PIMs, an impedance can be installed between the transformer neutral and earth. This is the impedant neutral solution. An impedance whose purely resistive value
at 50 Hz is around 1500 is recommended for a short LV power system supplied by an MV/LV substation [4]. In MV, DC injection PIMs are connected, according to voltage, either to a VT loaded by a resistance (see fig. 13 ) or to a resistance connected in series with the PIM (Un < 5.5 kV). The star point of the primaries of all other wye-connected VTs connected to the same isolated neutral system must also be earthed by means of a capacitance (P1 plate). This measure is often overlooked in extensions and sub-switchboards. c Case of MV power systems earthed through a reactor (see fig. 9b ) The following measures can be taken in resonant earthed systems: v overcompensate the power frequency capacitance component of the earth-fault current by detuning the neutral earthing reactance, v add a resistive component to increase earth coil losses. The measure taken must not, however, affect self-extinguishing of earth faults which is one of the aims of the resonant earthing system. c As regards a transformer fed through a capacitive power system (see fig. 10 ), the best solution consists of avoiding risky configurations when active power delivery is less than 10 % of the transformer rated power. This risk is considerable during low load periods (holidays, night time).
a - Available neutral
b - Unavailable neutral
VT R C
VT R C
P1 1R PIM
P1 3R PIM
Fig. 13: PIM auxiliaries.
4.2 Practical evaluation criteria

The equation systems describing the behaviour of ferroresonant circuits cannot generally be solved in analytical manner, but require use of numerical methods. However, as regards the series ferroresonant circuit, it is possible to analytically predict the existence of periodic, fundamental (at the power pulsation 0 and n-subharmonic ferroresonance (of pulsation 0 /n where n is an integer). L is the value of the non-linear inductance in linear unsaturated state. Ls corresponds to the saturated state. Periodic ferroresonance is impossible if one of the criteria below is verified: L 0 n > c n C 0 where n is the order of the subharmonic. (n equals one in the case of fundamental ferroresonance) L n < s 0 c n C 0 Both criteria can be illustrated by figure 14 : the magnetization curve enables knowledge of L and Ls. v The value of Ls can be supplied by the manufacturer, v The value of the magnetizing inductance of a VT or of a single-phase power transformer is: Un 1 L = (mH) 0 2 P U 2 I0 ( 0 n) Where: Un : rated voltage (kV), I0 : no-load current under Un (A), P0 : iron losses under Un (kW). v Typical capacitance values The magnitudes of the zero-sequence capacitances of synthetic insulated, screened underground cables and of overhead lines are given in figure 15 . As regards cables, we advise the reader to consult the analytical formulae provided by cable manufacturers or the value tables given in standard NF C 33-220. I
Fig. 14: values of C incompatible with periodic ferroresonance.
n > Lo Co n VL (I) Lso
n < Lso n Co Lo
v Finally, for ferroresonance to be sustained, source coupling must be able to compensate circuit losses [10].
Synthetic insulated screened cables Rated voltage Uo/U(Um) (kV) Crosssection (mm2) 16 240 8.7/15 (17.5) 12/20 (24) 25 240 35 240 PE EPR
Overhead lines 20 kV 90 kV 150 kV 220 kV 400 kV double circuit line + earth cable
5.8/10 (12)
0.17 0.43 0.16 0.34 0.15 0.28
0.21 0.52 0.19 0.41 0.18 0.35
5 x 10-3
4.8 x 10-3 5.6 x 10-3
5.5 x 10-3 7.1 x 10-3
Fig. 15: zero-sequence capacitances (in F/km) of cables and lines (typical values).
5 Studies for predicting and understanding
The ideal solution is to predict the risk of ferroresonance for all possible power system parameter values in normal and exceptional operating conditions, taking account of future modifications to the installation. A safety margin can thus be provided with respect to the dangerous
area, and countermeasures taken. A reliable and credible solution requires implementation of numerical methods suited to the study of certain types of ferroresonant states.
5.1 A study example

The procedure, together with implementation and practical application of these methods are illustrated below. The case studied In order to reduce breaking times due to malfunctions, MV rural public power system operators operate switch-disconnectors by remote control. Figure 16 shows the example of a remote controlled breaking cubicle (ACT), free standing and connected to an MV overhead power system by an over-underground connection. A voltage transformer (VT) connected between two phases (phase 1 and phase 3) provides the independent LV power supply of the SRI type remote control box (SRI = Switch Remote control Interface). This study is motivated by explosive failure of VTs in various installations of the same kind while energized for the carrying out of live work (live installation of the jumpers connecting the overhead line to one of the two overunderground connections). The explosion of the VT was observed 5 to 55 minutes (depending on the case) after installing the jumper for phase 1 of the pole 1 (switch-disconnector closed and pole 2 jumper not installed).
ACT SRI HV/MV substation Another feeder from the same substation or feeder from a different substation
VT (between two phases) Jumpers
Jumpers
1 Cable Cable
20 km
15 m
15 m
Fig. 16: remote controlled breaking cubicle (ACT), free standing, connected to an overhead MV power system.
Parameters of the power system studied: TT : 20000/230 V, 100 VA, HV/MV substation: 63 kV/21 kV, 10 MVA, Neutral earthing resistance: 40 , 20 km overhead line between the HV/MV substation and pole 1 , ACT-pole connections: 15 m cable (radial field), 150 mm2, Aluminium. At working frequencies (50 Hz and subharmonics), the overhead line can be modelled by its longitudinal impedance. The equivalent diagram is then the one shown in figure 17a . Where: e(t): sinusoidal voltage source e(t) = E cos(100 t) where E = 21000
2 / 3 = 17000 .
e(t) R L
a - Equivalent single-phase diagram

VT Rp Lp Co Co R2
b - Simplified diagram
R2 e(t) R1 Lp Co
R: neutral earthing resistance + resistance of the HV/MV transformer + longitudinal resistance of the overhead line. L : self-inductance of the HV/MV transformer + longitudinal self-inductance of the overhead line. C0 : zero-sequence capacitance of the 30 m cable (C0 = 6.7 nF). Lp : (non-linear) magnetizing inductance of the VT seen from the primary. Its characteristic is determined from voltage-current measurements (magnetization curve) taken on the no-load VT. Rp : primary winding resistance. R2 : resistance equivalent to iron losses and to hysteresis losses. R2 is assumed to be constant and independent of rms voltage and peak flux. This circuit can be simplified as the diagram shown on figure 17b . This circuit is a series ferroresonance circuit and thus favourable to ferroresonance. Application of appropriate methods allows the possibility of ferroresonance to be studied between the VT and the capacitance to earth of the 30 m length of cable connected to the free phase (not connected to the source) of the VT. Determining the areas at risk
Fig. 17: installing the jumper for phase 1 of the pole

^ (Wb) 140 120 100 80 60 Mn 40 20 17 kV 0 0 5 10 15 20 25 30 35 Mf
M1
E(kV)
Fig. 18: bifurcation diagram.
C0 (nF) 10 9 8 7 6 5 4 3 2 1 0 0 5 10 15 20 25 30 35 17 kV E (kV)
The bifurcation diagram (E) in figure 18 shows that for power system parameter values, ferroresonance (point Mf) may occur for nominal phase to neutral system voltage (17 kV). The area at risk is located above the green bifurcation line shown in figure 19 in the plane (C,E) (position of points M1 in the bifurcation diagram in figure 18 ). The capacitance value of 6.7 nF is located amply in the area at risk.
Fig. 19: bifurcation lines.
This condition is not sufficient to conclude that the phenomenon can really exist. Time domain digital simulations The question is then to know whether the possible initial condition values are responsible for initiating ferroresonance. A time domain digital simulation of the three-phase circuit will provide an answer. In the case in hand, the cable capacitance is discharged and the initial conditions are thus mainly determined by the energizing conditions of a phase of the over-underground connection. These energizing conditions are directly dependent on the operator, and cannot be compared with the conditions obtained by energizing by switch. The waveforms obtained by simulations and illustrated on figures 20a and 20b show that the live work procedure described above can result in a sustained ferroresonance. Equipment withstand to stresses We must then determine whether this state is dangerous for equipment and whether it can account for VT failure. As the overvoltages computed at the terminals of the VT in the ferroresonance steady state are lower than VT rated power-frequency short-duration withstand voltage (50 kVrms/1 min, i.e. 70 kV peak), the possibility of thermal and/or mechanical destruction must be investigated.
As the rated short-circuit voltage of the VT is 2.5%, it is designed to withstand without damage at least the mechanical and thermal effects of an external short-circuit current of 40 In for the duration of 1 sec (IEC 186). This enables computation of the thermal withstand time tmin under Irms:
(40 In )2 1 s
hence t =
(40 100 20000)2 (0.13)2
= (Irms ) tmin
2
= 2.4 s
Consequently, there is at least one ferroresonant state resulting in VT failure by thermal effect, and special precautions must be taken. Solutions In this precise case, the following methods for increased ferroresonance protection can be suggested: c load the VT secondary windings: appropriate numerical methods will enable you to determine the value of this load, c perform switchings when the equipment is deenergized, c modify the energizing procedure. First install the three jumpers of pole 2 : the switchdisconnector is open. The switch is then closed on the three phases simultaneously supplying the two phases of the VT. The three jumpers of pole 1 can then be installed. The methods used in this study (and others) are described in the following sections.
a - VT primary current A 0,30 0,20 0,10 0 -0,10 -0,20 -0,3 1,8
b - VT primary voltage kV 75 50 25 0 -25 -50
1,85
1,9
1,95
2 s
-75 0 0,2 0,4 0,6 0,8 1 s
Fig. 20: time domain numerical simulation.
5.2 Modelling, a mathematical approach

The following means are used to study transient electromagnetic problems and ferroresonance: c Analog simulation Based on a reduced model representation of power system components, this method has the advantage of obtaining real time results, but the disadvantage of finding it hard to represent a real case with any accuracy. c Time domain digital simulation in transient state Resolution of the system of mathematical equations describing power system behaviour requires use of computer tools. As transient ferroresonant states are normally long, simulation times are lengthy and studies costly. Since ferroresonance is extremely sensitive to parameter values and to initial conditions of which we have little practical knowledge, a study must be conducted for each possible combination. This is not realistic. The methods described above are therefore not suited to the search for an overall view of a power system behaviour. To guard against the drawbacks of such methods, mathematicians have developed the following methods: c Methods for direct computation of steady state These methods enable solutions to be computed in steady state without requiring computation of the transient state which is normally lengthy in the case of ferroresonance. c Continuation method The mathematical framework most suited to the general study of dynamic systems behaviour is the bifurcation theory, the main tool of which is the continuation method. Used jointly with the methods for direct computation of steady state, it determines the areas at risk. Note that time domain digital simulation has undeniable advantages: v it confirms the results of another method for a given configuration and numerical parameter values, v it specifies waveforms and their corresponding overvoltage/current levels, thanks to fine modelling of power system components. Numerical methods for computation of steady state Mathematicians have developed frequency and temporal methods enabling computation of solutions in steady state without requiring computation of the transient state. The following methods enable study of periodic ferroresonance (fundamental, subharmonic). The main frequency method is the Galerkine method. The main temporal methods are the Poincar map fixed point method and the perturbation method. c Galerkine method This method consists of finding an approximate solution for the system of non-linear differential equations describing power system behaviour. The solution is broken down in the form of a Fourier series limited to the order k in order to find periodic solutions. The unknown quantity is replaced by its expression. This results in a system of 2k+1 equations with 2k+1 unknown quantities (the Fourier components). c Poincar map fixed point method As the solution is periodic with a period nT, the method consists of iteratively searching for an invariant solution by numerical simulation of the system over a period nT. c Perturbation method This method consists of simplifying the system of equations describing circuit behaviour by cancelling certain parameters such as the losses and/or amplitude of the voltage source. Once the resulting equation has been solved, the cancelled terms are introduced by a limited development around the solution previously obtained. This method is of particular interest when combined with the continuation method. Continuation method This numerical method based on an iterative process allows the study of the influence of a parameter (e.g. voltage source amplitude) on the solutions (e.g. the flux in a transformer core) of the equations describing the system behaviour (e.g. the electrical power system). The points on the solution curve are obtained step-by-step from a known solution: With knowledge of a solution x0 corresponding to the value E0 of the chosen parameter, the neighbouring solution x1 is obtained by initialising system resolution by the solution x0 and the neighbouring value E1 of the parameter. The resulting step-by-step curve is known as a bifurcation diagram. The resolution method used for ferroresonant circuits is one of the steady state direct computation methods, thus ensuring independence from initial conditions.
c Bifurcation diagrams As regards the series ferroresonant circuit in figure 21a , examples of bifurcation diagrams according to amplitude E of the voltage source for two values (Ra and Rb) of parallel resistance R2 are plotted on figure 21b . For a given value of R2, these curves reveal two specific points (M1 and M2 for R2 equal to Ra) known as limit points for which there is a change in stability (see 2.2.). c Bifurcation lines For each point, M1 and M2, the corresponding value of E known as the critical value can be determined (E1 for point M1, and E2 for point M2). The line plotted in the plane of two parameters, for example E and R2, of pairs (E1, R2) and (E2, R2) corresponding to singular points for different values of R2, results in curves (see fig. 21c ) representing the boundary between two different operating conditions, namely normal conditions and ferroresonant conditions. The resulting curves are known as bifurcation lines. Bifurcation lines can also be plotted in the plane of various parameter pairs such as (R2, C) or (R1, E). c Isolats The bifurcation diagrams in figure 21b illustrate the case of fundamental ferroresonance. These curves in fact pass through the obvious solution (0,0), the point from which the user can also initialise the continuation method. However, continuation of isolats, which are closed curves, is more tricky, as it is necessary to know a solution belonging to this isolated curve (isolat) in order to initialise the continuation method. The perturbation method can be used to obtain this solution. The resulting curves correspond to a frequency different from that of the voltage source, such as for example the subharmonic isolat shown in figure 22 . For a voltage E between the ends of the isolat, the slightest perturbation may cause the system to move from the green curve to the black curve and vice versa. c Practical use of the bifurcation diagrams With given circuit parameter values (and in particular R2 equal to Ra), figure 21b shows that as long as E remains less than E1, a fundamental ferroresonance cannot be sustained under steady state condition. c Practical use of the bifurcation lines If in all power system operating conditions, the amplitude E of the voltage source is less than En, figure 21c shows that a resistance value of R2 less than Rn guarantees that a fundamental ferroresonance cannot occur.
a - Series ferroresonant circuit
R1 E
C R2
b - Bifurcation diagram according to E
Stable state Unstable state M1 M'1
R2 = Ra R2 = Rb (< Ra)
M2
M'2
E1
E'1
E2
E'2
c - Bifurcation lines in the plane R, E
R2
Rn
Point of M1 , M'1 ... Point of M2 , M'2 ...
Ra Rb
M1 M'1
M2 M'2
En
E1
E'1
E2
E'2
Fig. 21: bifurcation diagrams and lines (fundamental ferroresonance).
VL
Stable state Unstable state Subharmonic mode (isolat) Fundamental mode
Fig. 22: isolat of a subharmonic mode.
Given the inaccuracy of the numerical values of the power system parameters, the practical solution is normally to provide a safety margin with respect to the areas at risk. It should be stressed that not all the states located in the areas at risk, whose boundaries
are formed by the bifurcation lines, are necessarily reached in practice as special initial conditions have to be verified. Time domain numerical simulations are used to determine whether these initial conditions are possible in reality.
5.3 Summary
Below is a plan for a ferroresonance study conducted using the methods described above. The various steps are listed in chronological order in the following table.
1 - Identifying a configuration at risk
2 - Simplifying the circuit

a
3 - Determining the power system parameters - Non-linear characteristic of the inductance - Variations and tolerances of R and C.
4 - Determining the areas at risk (overview) - Continuation method - Steady state computation method 5 - In practice, do initial conditions result in an area at risk? Time domain digital simulations. 6 - Consequences on thermal, mechanical and dielectric withstand of equipment.
7 - Proposed solutions.
6 Conclusion
The risk of ferroresonance must be taken into consideration as early as the design stage of an electrical installation. Vigilance is also called for when servicing or extending a power system. Basically, risk control requires knowledge of dangerous configurations and of the conditions in which the phenomenon can exist. If a configuration considered critical is inevitable, only a detailed study will enable an assessment of the risks and an evaluation of the efficiency of the solutions to be provided. This document has provided LV and HV power system designers and installers with an insight into the precautions to be taken to prevent this odd and often dangerous phenomenon, and should facilitate their dialogue with specialists. Operators will have found a few elements for diagnosis enabling them to rightly suspect that ferroresonance is present. They now understand that not all unexplainable breakdowns can be ascribed to this phenomenon! It seems a good idea at this point to briefly remind readers of the events initiating ferroresonance and of the configurations at risk: c A few examples of phenomena likely to cause ferroresonance: v capacitor switching, v insulation faults, v lightning, v transformer switching. c A few configurations at risk deserving particular attention: v voltage transformer (VT) between phase and earth of an isolated neutral power system, v long and/or capacitive cables or lines supplying a transformer, v fuse protection where blowing results in a nonmulti pole breaking, v unloaded or lightly loaded voltage or power transformer, v voltage transformer working at saturation limit, v over-powerful voltage transformer. Readers seeking additional information or interested in case studies are invited to consult the extensive bibliography on this subject.
Appendix 1: Bibliography
Standards [1] IEC 186: Voltage transformers [2] IEC 71: Insulation co-ordination Cahiers Techniques [3] Neutral earthing in an industrial HV network, F. SAUTRIAU, Cahier Technique Merlin Gerin n62. [4] Earthing systems in LV, B. LACROIX, R. CALVAS, Cahier Technique Merlin Gerin n172. [5] Overvoltages and insulation coordination in MV and HV, D. FULCHIRON, Cahier Technique Merlin Gerin n151. [6] Harmonics in industrial networks, P. ROCCIA, N. QUILLON, Cahier Technique Merlin Gerin n 152. [7] Active harmonic conditioners and unity power factor rectifiers, E. BETTEGA, J.N. FIORINA, Cahier Technique Merlin Gerin n 183. Various works [8] Revue des phnomnes de ferrorsonance dans les rseaux haute tension et prsentation dun modle de transformateur de tension pour leur prdtermination, N. GERMAY, S. MASTERO, J. VROMAN, CIGRE Session de 1974-21-29 aot. [9] Ferroresonance in a transformer switched with an EHV line, E.J. DOLAN, D.A. GILLIES, E.W. KIMBARK, IEEE Power Apparatus and Systems, 1972. [10] Contribution thorique et exprimentale ltude des phnomnes de ferrorsonance monophase, P. MAHY, SRBE, mars 1972. [11] Transformateurs de mesure-Gnralits. Thorie. Fonctionnement, J.P. DUPRAZ, D 4720, Techniques de lIngnieur, trait Gnie lectrique. [12] Ferroresonance study using Galerkin Method with pseudo-arclength continuation method, G. KIENY, G. LE ROY, A. SBAI, IEEE PWD, Vol. 6, No.4, October 1991. [13] Ferrorsonance dans les rseaux, C. KIENY, A. SBA, D 4745, Techniques de lIngnieur, trait Gnie lectrique.
Appendix 2: Selection guide for VT load resistances
Residual voltage windings connected in open delta, closed on a resistance. Rated secondary voltage Rated thermal burden of the secondary winding concerned by the resistance (VA) 50 100 200 50 100 200 50 100 200 50 100 200 Minimum computed resistance () Practical choice (in the 140 W series) Standardised resistance () 120 2 x 120 in // 2 x 100 in // 150 2 x 150 in // 2 x 100 in // 390 2 x 390 in // 2 x 220 in // 3 x 390 in // 470 2 x 470 in // 2 x 390 in // Resistance power (W) 83 2 x 83 2 x 100 80.7 2 x 80.7 2 x 121 77 2 x 77 2 x 136 3 x 77 77 2 x 77 2 x 93
100 3 110 3
100 3
115.5 57.7 28.9 139.7 69.9 34.9 346.4 173.2 86.6 419.2 209.6 140.8
110 3

Part 1 of 3
Vacuum switching
P. Picot
no. 198
Vacuum switching
Philippe PICOT Ingnieur civil des mines (ENSMP) Hired in 1982 by Merlin Gerin, he has held different positions, primarily in the technical field of the Medium Voltage division. Since 1995, he has been participating in the development of Schneider-Electrics vacuum interrupter range. He is presently in charge of technological anticipation of MV switchgear.
ECT 198 first issue, March 2000
Vacuum switching
This "Cahier Technique" constitutes a general presentation of basic notions relative to the functioning and use of vacuum switching devices. The first section, entitled Theory and Use of vacuum switching, is a brief description of the physical phenomena that are associated with vacuum switching, and of their use. It also includes a presentation of the different technological options that are available to vacuum interrupter designers. The second section is dedicated to the interaction between vacuum switching devices and the electrical network, in inductive circuits for which vacuum switching may cause overvoltages, and to overvoltage protection means. In the third section, the author explains how vacuum switching characteristics, which have been presented in the two preceding sections, determine the application fields best suited to this technique, depending on voltage levels and switchgear types. This "Cahier Technique" is completed with an extensive bibliography of works and other documents which the reader can consult if he wishes to acquire more in-depth information on a particular point.
Contents
1 Introduction: use of vacuum as a breaking medium in the electrical switchgear industry 2 Theory and use of vacuum switching 2.1 The dielectric properties of vacuum 2.2 Electrical arcing in vacuum 2.3 Phenomena associated with breaking at current zero 2.4 The practical design of vacuum interrupters 3 Breaking in vacuum and overvoltages during switching of inductive circuits 4 The main fields of application for vacuum switching 4.1 Vacuum switching applications in Medium Voltage 4.2 Vacuum switching applications in Low Voltage 4.3 Vacuum switching applications in High Voltage 5 Conclusion Bibliography 3.1 Overvoltage generating phenomena 3.2 Means of protection against overvoltages p. 4
p. 5 p. 7 p. 9 p. 12 p. 19 p. 22 p. 24 p. 25 p. 28 p. 29 p. 30 p. 31
1 Introduction: use of vacuum as a breaking medium in the electrical switchgear industry
SF6 and vacuum are the two most modern breaking techniques in the fields of Medium Voltage (from 1 to 52 kV) and High Voltage (> 72.5 kV). They appeared in the 1960s and rapidly developed as of the 1970s. Today they have replaced the former air and oil breaking techniques (see fig. 1 ).
% 100 80 60 40 20 0 1980 82 Air
84
86 Oil
88
90
92
94
96 SF6
Vacuum
Fig. 1: evolution of MV circuit-breaker markets in Europe.
Whereas SF6 is used in all of the medium voltage and high voltage ranges, vacuum has primarily developed in the medium voltage field, with limited incursions in low voltage and in high voltage: the two techniques only compete with each other in the medium voltage field. This notion of rivalry between the two techniques is now in fact out of date: even if at some time
there was commercial competition between manufacturers that opted for one or the other of these techniques, today all large-scale manufacturers offer both techniques so as to be able to satisfy as best as possible their clients needs. Indeed, each technique has its strong points and its weak points. Even if each is highly multi-functional and can offer a reliable and competitive solution for most medium voltage interruption problems, users want to be able to choose for themselves in function of their applications, operation and maintenance policies, priorities and of course habits! In the past, the vacuum switching technique was first developed by American and English manufacturers (the pioneers were General Electric and VIL), followed by the Japanese and the Germans: these countries have the common feature of using networks with relatively low voltage ratings (from 7.2 to 15 kV) for medium voltage electrical energy distribution. However, in countries like France and Italy which distribute electricity with voltage levels near 24 kV, manufacturers opted for the SF6 breaking technique. It is remarkable to note, 30 years later, the appropriateness of these technological choices in regard to the foreseen application. In fact still today, a global technical-financial evaluation of both techniques shows an equivalence when using voltages between 12 and 24 kV, with a relative advantage for SF6 above this voltage level, and for vacuum below this level. However, the difference in cost remains low, which explains how the two offers, vacuum and SF6, can coexist, for all medium voltage levels from 7.2 to 36 kV.

Part 2 of 3
Vacuum switching
P. Picot
2 Theory and use of vacuum switching
2.1 The dielectric properties of vacuum

Any breaking medium must first be a good insulator for it is to stop current from flowing through it. Vacuum is not an exception to the rule: it has interesting yet particular dielectric properties in comparison to other insulating gases that are commonly used under pressure that is higher than or equal to 1 bar. Vacuum, that is qualified as being high (pressure range from 10-1 to 10-5 Pa, i.e. 10-3 to 10-7 mbar) of vacuum switch interrupters (see fig. 2 ) is in fact a low pressure gas: typically 10-6 mbar in a new interrupter. effect) responsible for the breakdown: the electrons must acquire between two collisions V ) to ionise sufficient energy (proportional to pd the gas molecules and thus create other electrons.
V (V) 10
6
105 10 10 10
4 3
10-5
10-4
10-3 10-2
10-1
1 10 p d (bar cm)
Fig. 3: change in dielectric strength of the air in function of the pressure (Paschen curve)
Fig. 2: a 17.5 kV vacuum interrupter by Schneider Electric.
At this pressure, a 1 mm3 volume still contains 27.106 gas molecules, but their interactions are negligible since their mean free path between two collisions is of the order of a hundred meters: the term "vacuum" is thus appropriate since each molecule behaves as if it were practically alone. Reminder concerning the dielectric behaviour of gases At normal pressure levels (atmospheric pressure and above) the dielectric behaviour of gases is represented by the right branch of the Paschen curve (see fig. 3 ): the breakdown voltage V is a growing function of the p d product (p = pressure, d = distance between the electrodes). This relation characterises the chain ionisation mechanism (Townsend avalanche
At low pressure values, this mechanism no longer functions. In fact, the electrons can acquire a lot of energy during their mean free path, but the probability that they encounter molecules to be ionised before reaching the electrode becomes weak: the electron avalanche and multiplication process of the charged particles cannot take place and the dielectric withstand is improved. This is what the Paschen curve shows: a minimum dielectric withstand for a p d product in the region of 1 Pa m for nitrogen. Below this value, the dielectric withstand rapidly improves (left branch of the Paschen curve) up to a level of values for p d that are lower than 10-2 Pa m. This level characterises the dielectric behaviour in vacuum interrupters (pressure lower than 10-3 mbar, i.e. 10-1 Pa, distances in the region of 1 to 10 cm). It corresponds to a high withstand level that is comparable to that of SF6 gas which is at roughly 2 bars for intervals in the region of one cm. In this field, it is no longer the residual gas ionisation mechanisms that limit the dielectric withstand but rather phenomena linked to the surface condition of electrodes, such as field electron emission and the presence of detachable particles. c Field emission Electron emission consists in extracting electrons from the metal of electrodes. This can
be done by sufficiently raising the temperature of the metal: its the thermionic emission that is produced at the heated cathode level of the electron tubes. Another means is to apply a sufficiently strong electric field to the metal surface. This last phenomenon, field emission, is likely to be encountered in vacuum interrupters. It is controlled by the Fowler-Nordheim equation that, in a simplified form, is written: je = B 1.5 AE2 , where exp E
Breakdown voltage (kV) 180 160 140 120 100 80 60 40 20 0 100 200 300 400 500 Number of discharges
je is the electronic current density in Am-2 A = 1.54 x 10-6 AJV-2 E is the electric field in Vm-1 is the work function in eV (4.5 eV for copper) B = 6.83 x 109 VJ-1.5 m-1 As can be seen from the values indicated above, field emission only becomes appreciable for field values on the surface of metals that are included between a few 109 Vm-1 and 1010 Vm-1. Very high values are being dealt with here; values that are significantly higher than the macroscopic field values for typical vacuum interrupters (in the region of 107 Vm-1 = 100 kV/cm). Even so field emission has been acknowledged in vacuum interrupters: it must therefore be concluded that locally, at the microscopic site level, the electric field is reinforced by an enhancement factor in the region of a few 102 or 103. The phenomena that could explain these high values have not yet been completely elucidated by researchers, who in general favour the microscopic point effect, or the inclusion of insulating particles at the surface of metals. v Voltage conditioning The existence of active microscopic emission sites in general results in poor dielectric withstand of new interrupters (a few 10 kV/cm); however, it has been experimentally noted that repeated dielectric breakdowns destroy these sites or at least considerably reduce the value of the enhancement factor that characterises them. A satisfactory dielectric withstand (in regard to assigned values) can thus be obtained only once the voltage conditioning process has been completed. It consists in applying a high voltage (around the expected withstand value) for a few minutes: the multiple breakdowns that occur, progressively raise the withstand between electrodes. This phenomenon is illustrated in figure 4 which shows the change over time of the breakdown voltage as discharges pass: an upper limit for dielectric withstand improvement appears near 108 Vm-1, which again corresponds to an "irreducible" microscopic of about 100. v Breakdown mechanisms Dielectric breakdowns that originate in electronic current emission implement additional mechanisms: in fact, stable electronic currents (for maximum values of a few mA) do not
Fig. 4: improvement of the breakdown voltage between two electrodes in vacuum as a function of the number of discharges.
necessarily degenerate into a breakdown if the applied voltage is not increased, they may even diminish by themselves through the conditioning effect. Breakdown, itself, is linked to the creation of localised plasma (ionised gas), which is sufficiently dense for the electron avalanche phenomenon characteristic of gaseous discharges to be produced. The plasma may be produced on the cathode side through the explosion of the microscopic emissive site caused by the intense overheating due to the current density which is locally very high (Joule effect): breakdown is produced in the metal vapour that was generated by the destruction of the emissive site. The plasma may also be produced on the anode side which is bombarded by a beam of highly energetic electrons (which also results in the emission of X-rays). This localised flow of energy causes the desorption of gases absorbed on the surface and the vaporisation of anode metal: the gas produced from this is thus ionised by beam electrons, and the breakdown occurs. c The influence of detachable particles A second factor is likely to cause dielectric breakdowns in vacuum: detachable particles present on the surface of the vacuum interrupter walls. Set free, either by a shock, or by the effect of electrostatic forces, these charged particles acquire energy by going through the interelectrode gap. At the moment when they impact with an electrode that attracts them, they are likely to trigger a breakdown in two ways, which may be complementary: v through a local rise in the gas density due to the desorption of absorbed gas molecules; v by triggering the field emission phenomenon and the partial vaporisation of the particle or of the electrode under the effect of the beam that bombards them.
Confirmation of the practical importance of the particles is the experimental observation that the dielectric withstand in vacuum between two electrodes increases approximately in proportion to the square root of the distance which separates them. This relation can be explained by the hypothesis that the particles must reach sufficient energy (proportional to V2/d) to be able to cause a breakdown. For this same reason, large particles, that can carry a higher electrical charge, are more troublesome than small ones. From the unfavourable influence that detachable particles have on the dielectric withstand of vacuum interrupters, two consequences are to be noted: c it is difficult to reach very high withstands, even with a great amount of space between electrodes (see fig. 5 ), c the dielectric withstand of a vacuum interrupter presents a random character: a delayed breakdown can occur in regard to voltage application and for a voltage of less than that which was tolerated right before without a breakdown. Resume c Vacuum shows interesting dielectric properties if applied voltages are limited to a region of 100 to 200 kV, which corresponds to an insulation level required for voltage ratings of i 36 kV for which distances of a few centimetres between electrodes suffice. Above this level,
Breakdown voltage (kV) 1000
100
10
10 100 1000 Distance between electrodes (mm)
Fig. 5: accessible withstands for very large distances between electrodes.
reaching the necessary dielectric withstand level becomes laborious and less efficient than with SF6 gas insulation. c The dielectric withstand of a vacuum switching device evolves over time. Indeed, mechanical operations and the effect of electrical arcing modify the contact surface condition and generate particles: the withstand level reached after voltage conditioning therefore cannot be considered as permanently acquired. Vacuum is thus not the ideal insulating medium when the reliability of dielectric withstand is essential, for example for a disconnector application.
2.2 Electrical arcing in vacuum

Even though, as described in the above section, vacuum may be an excellent dielectric, an arc can very well "live" in the "vacuum". In fact, the arc voltages in vacuum are in general considerably lower than those of arcs that develop in other mediums, which constitutes an advantage in regard to the energy that is dissipated in the arc. Arcs in vacuum occur, by voluntarily simplifying, in two main forms: the diffuse mode and the constricted mode. A diffuse mode, characteristic of the "vacuum" medium The diffuse mode is specific to arcing under vacuum: it shows remarkable particularities which clearly differentiate it from arcings in gaseous mediums. It is the mode which a vacuum arc naturally adopts for a current range covering a few amps to a few kA. The main characteristics of the diffuse mode are as follows: c the cathode emits into the inter-electrode gap, via one or several cathode spots, a globally neutral plasma made up of electrons and of high speed ions whose velocity is primarily directed perpendicularly to the surface of the cathode; c the anode, with its entire surface immersed by this plasma, reacts as a passive charge collector. The cathode spots and the plasma are specificities of the arc in the diffuse mode. c The cathode spot The cathode spot is a very small sized zone (radius in the region of 5 to 10 m), capable of emitting a current that can reach some hundred amps. Extreme temperature and electric field conditions rule at the cathode spot level (typically 5000 K and 5 109 V/m). These conditions allow for electronic emission by combining thermionic and field emission mechanisms into thermo-field emission which is capable of producing very high current densities (between 1011 to 1012 A/m2). Above 100 A, this spot subdivides itself and several spots coexist on the cathode, in sufficient number to transit the current at the rate of some hundred amps each. They mutually
drive each other back, which led their movement to be qualified as retrograde for it is contradictory to the normal effect of electromagnetic forces. Thus arcing in the diffuse mode tends to occupy the entire available surface on the cathode (even if at any given moment the emissive sites only represent a very small fraction of the cathode). c The plasma At the macroscopic level, the cathode spot (crater and close-range plasma that is associated with it) seems to be the production point of a low density plasma coming from the spot and which fills the inter-electrode gap. This globally neutral plasma (equal densities of + and - charges), is made up of electrons and ions which are typically double charged (for arcing on electrodes with a Cu base). One of the characteristics of this plasma is the great speed of the ions which have an energy that is higher than the arc voltage (which testifies to the highly energetic phenomena that are produced in the zone of the cathode spot). It is therefore not difficult for these ions, which emanate from the spot with a distribution of speed approximately in cos (angle/normal) to reach the anode and create an ionic current in the opposite direction to the main electronic current which typically represents 10 % of the arc current. The directed velocity of these ions is in the region of 104 m/s, higher than their thermal agitation speed. One of the significant consequences of the high speed of the ions created by the cathode spots is their low transit time through the interelectrode gap (typically in the region of 1 s). The plasma, created by a cathode spot, is made up of highly mobile particles (rapid electrons and ions, virtually no neutral particles) and thus disappears very rapidly when the spot stops functioning (around current zero). The anode is immersed in the plasma that emanates from the cathode spots. It behaves like a passive electrode that collects charges and extracts the current that is imposed by the circuit by adjusting its voltage: it is negative with respect to the plasma as long as the current is lower than the one that corresponds to the impacts linked to the thermal agitation of electrons. The distribution of voltages in the arc is as follows: v a cathode voltage drop in the region of 20 V in the immediate area of the cathode; v a voltage drop of a few volts in the plasma which increases with the distance and the current (positive characteristic allowing for the coexistence of several parallel arcs, contrary to arcs in gas); v a negative anode drop in the case considered above (moderate current absorbed by the anode).
In this mode, there is little cathode erosion: it corresponds to the ion flow leaving the cathode, i.e. roughly 40 g/C. A significant number of these ions place themselves on the anode which, in alternating current, means that net erosion is much lower: approximately divided by a factor of 10 for contactors that operate in this mode with limited currents and electrodes with little spacing. A constricted mode similar to the one of an arc in a gaseous medium When the current increases, the previously described situation tends to evolve first of all on the anode side. Several phenomena converge towards this evolution. c First a contraction of the plasma column generally explained by the Hall effect (charge deviation by the azimuthal magnetic field created by the other current lines, from which the appearance of a radial component tends to confine the current lines towards the axis): the current is concentrated on a more limited area of the anode. c Furthermore the anode attracts more and more electrons, and the neutrality of the plasma is no longer ensured: positive ions are lacking to balance the space charge of electrons near the anode. This leads to the formation of a positive anode voltage drop which is needed to attract electrons despite the space charge. The energy received by the anode increases and tends to be concentrated on a reduced area: the anode heats up and starts to emit neutral particles that are ionised by the incident electrons. Near the anode, a secondary plasma, made up of secondary electrons and ions that are less energetic than those emitted by the cathode spots, appears. These phenomena result in the appearance of a luminous anode spot, considerably larger (in the region of a cm2) than the cathode spots, made of molton metal which spills considerable amounts of vapour, which becomes ionised in the flow coming from the cathode, into the inter-electrode gap. This contraction effect on the anode side also leads to a contraction on the cathode side since a preferential path is created thanks to the plasma generated by the anode: a cathode spot corresponding to the anode spot is established and the arc takes up the constricted mode that is characteristic of arcs in a gaseous medium. Here, we are dealing with an arc in an atmosphere of dense metallic vapours, for which operating mechanisms now rely on the ionisation of the gaseous medium. This arc in the constricted mode is thus characterised by a plasma made up of electrons (most of which are secondary), of neutral particles and ions the energy of which is near that of the neutral particles, thus relatively slow.
2.3 Phenomena associated with breaking at current zero

General breaking principles All medium voltage circuit-breakers take advantage of the natural passage of alternating current through zero (twice per period, i.e. every 10 ms for a 50 Hz current) to interrupt the current. c The inevitable arc phase Once a fault current has been established in a circuit, the separation of circuit-breaker contacts does not have an immediate repercussion on current flow. At the level of the last contact points, the current density becomes very high, which causes a local fusion and the appearance of a liquid metal bridge. The contacts continue to move away from each other, this bridge is heated up by the current and becomes unstable and its rupture results in the appearance of a constricted arc in the metal vapours originating from the liquid bridge explosion. The arc voltage that appears is, in the case of a vacuum, low in comparison to the electromotive forces of LV or HV network generators: the current flowing in the circuit is thus not considerably affected, nor limited, by this arc voltage. This arc will adopt the diffuse mode or the constricted mode described in the preceding section, possibly evolve from one to the other, and will be maintained up to current zero. c The recovery phase after current zero If the plasma, which up to now allowed for the current to flow through, takes advantage of this break to dissipate itself very rapidly, the current may be prevented from establishing itself for the following half-cycle. A transient recovery voltage (TRV), imposed by the circuit, then appears at the terminals of the element that has switched from a conductive state to an insulating state. In the case of a short-circuit, this TRV is caused by the oscillations between the local capacitances and the network inductances. In its initial phase, it approximately presents a (1-cosinus) shape with a natural frequency in the order of a few tens of kHz in MV network and reaches a peak value that is greater than the normal network voltage, which corresponds to average rates of rise of a few kV/s. If the newly insulating medium tolerates the dielectric stress which is then applied to it, the current is successfully interrupted (see fig. 6 ).
Laboratoire VOLTA C2531 98/12/04/009
1 ms
I
10.00 kA
U 5.00 kV
UE 40 V/V 0.50 V
Fig. 6: a successful current interruption (source Merlin Gerin).
Case of vacuum switching To determine the conditions for successful current interruption, it is necessary to study the phenomena that intervene near current zero in the vacuum arc plasma. c Post-arc current Near the end of the half-cycle, the current decreases at a rate which is proportional to the peak current value and to network frequency (di/dt = ). The vacuum arc returns to the diffuse mode and, near current zero, only a single cathode spot remains. However, the intercontact gap is still filled with a residual, globally neutral plasma, that is made up of electrons, ions and neutral particles which come from the preceding arc. At the time of current zero, the last cathode spot extinguishes itself because the arc voltage disappears. Thus, the emissive site, which created charged particles (electrons and ions) needed to transport the electric current, no longer exists. From this moment on, a voltage with an inverse polarity to that of the preceding arc voltage (the TRV), starts to appear between the two contacts: the ex-anode becomes negative in regard to the ex-cathode and drives back the electrons. The current that flows in the circuit is now only made up of ionic current that the ex-anode extracts
from the residual plasma that becomes scarce: this current with an inverse polarity to that of the arc current is called post-arc current. The ex-anode is thus no longer in contact with the neutral plasma which is still present in the inter-contact gap: it is separated from it by a sheath from which the electrons, driven back by the negative voltage of the ex-anode, are absent. Only positive ions cross the neutral plasma boarder into the sheath and are then accelerated towards the ex-anode. The voltage that appears between the ex-cathode and the ex-anode is thus applied only to the thickness of the sheath that separates the neutral plasma from the exanode. Moreover, the presence of positive space charges in this sheath reinforces the electric field on the surface of the ex-anode which is higher than the average field that corresponds to the TRV value divided by the sheath thickness (see fig. 7 ). The thickness of the sheath that surrounds the ex-anode is proportional to the voltage applied between the neutral plasma and the electrode and inversely proportional to the density of the positive ions: it thus increases according to the change in TRV and all the more rapidly as the plasma rarefies. When the limit of the sheath reaches the ex-cathode, the residual plasma has disappeared, since all of its charges have
Electric field
2 3
Series 1: E on the ex-anode surface Series 2: UTRV / sheath thickness, average field in the sheath Series 3: UTRV / inter-contact gap
Fig. 7: electric field on the surface of the ex-anode and the corresponding average field between the electrodes.
been used by the post-arc current which becomes nil. These phenomena take place on a very reduced time scale: the total length of post-arc current is typically 1 to 10 s (see fig. 8 ). c Causes of interruption failure So that current can be maintained, mechanisms that create electrical charges must replace the cathode spots that have extinguished on the excathode. The first possible mechanism is the ionisation of the neutral metal vapour that is present in the inter-contact gap. This ionisation is all the more easy that the density of neutral particles is higher. If the vapour density is very high (very hot zones on the contacts produce a great amount of metal vapours), the current does not interrupt at all: there is no increase in the TRV, this is called "thermal non-breaking". If the density of neutral particles is sufficiently high so that the dielectric withstand of the vacuum can be reduced (approaching the minimum value of the PASCHEN curve), the current can be interrupted, but the inter-contact gap cannot tolerate the applied TRV and a breakdown occurs during the rise in the TRV, here we have "dielectric non-breaking". A second possible mechanism is the appearance of cathode spots on the ex-anode. For this, electronic emission conditions must be locally reunited on the surface of the ex-anode:
v thermionic emission if very hot points remain, this is the case when the anode contains refractory metal (W); v field emission or combined T.F. emission if the electrical field applied to the surface is significant at certain sites with a high enhancement factor . We previously saw that the electric field applied to the surface of the ex-anode appears with high values as of the start of TRV application since the sheath is thin; the higher the ion density, the thinner the sheath is. Furthermore, the ex-anode is bombarded by ions that have been accelerated in the sheath by the TRV, which causes localised overheating. The probability of cathode spots appearing on the ex-anode is thus greater if the density of ions in the residual plasma is high, which goes hand in hand with a high density of neutral particles which slow down, through collision, the rapid ions emitted by the cathode spots, thermalize them (average energy near the temperature of the plasma) and slow their diffusion at the time of current zero. If plasma density is sufficiently low at the time of current zero, the conditions for successful breaking have probably been satisfied: the current is interrupted and the inter-contact gap withstands the recovery voltage up to its peak value. In the case of vacuum circuit-breakers, success is not however entirely guaranteed once this stage has been completed. In fact, for a few
0.01 ms
I3
P-ARC 1A
V3 5.00 kV
Fig. 8: post-arc current with a particularly long length of roughly 40 s, test at the breaking capacity limit of the interrupter.
milliseconds after the break, the situation inside the interrupter can still change and dielectric breakdowns can occur: v particles generated during the arcing phase can detach themselves from the walls under the effect of vibrations and/or electrostatic forces; v molten areas on contacts can emit droplets under the effect of electrostatic forces; v solidification of the liquid metal can modify the surface of the contact or free dissolved gas. When a vacuum interrupter is tested at the limit of its breaking capacity, after breaking that appears to be successful, it is not rare to see late occurring dielectric breakdowns (see fig. 9 ) which may be:
v either transient (duration of a few s) for the interrupter is able to break the HF current that follows the discharge. If these transient breakdowns occur more than a quarter of the industrial frequency period after current zero, they are considered as non-sustained disruptive discharges (NSDD) and interpreted as a sign of device weakness (for this reason the maximum number of NSDD that is tolerated is three for a complete series of breaking tests on a circuitbreaker as in IEC 60056); v or complete and, in this case, the power current reappears after a more or less long interruption period (in the region of 0.1 to 1 ms).
0.1 ms V1 5.00 kV
V2 5.00 kV
V3 5.00 kV
Fig. 9: example of late occurring dielectric breakdowns.
2.4 The practical design of vacuum interrupters

Choice of the breaking technique The preceding section highlighted the conditions that must be satisfied for successful breaking. These conditions are almost always satisfied when an arc remains in the diffuse mode, that is to say when currents to be interrupted do not exceed a few kA. It is the case for switches and contactors that can therefore use very simple butt contacts. When an arc passes into the constricted mode, the energy is dissipated onto a reduced electrode surface, and it causes localised overheating and considerable vaporisation. If this arc remains immobile, breaking is no longer guaranteed. Two methods are used to overcome the difficulties that are produced by the passage of an arc into the constricted mode. v The first consists in causing a rapid circular movement of the constricted arc so that the energy is distributed onto a large part of the contact and overheating is limited at all points: this is obtained through the application of a radial magnetic field Br in the arc zone.
v The second consists in preventing the passage into the constricted mode through the application of an axial magnetic field: when the field reaches a sufficient value, the arc is stabilised in a mode qualified as a diffuse column and does not concentrate itself ; even though it is immobile the arc uses most of the contacts surface and overheating therefore remains limited in this case as well. c Radial magnetic field technique Br The constricted arc can be compared to a conductor through which a current flows, the direction of which is parallel to the axis of the contacts. If a radial magnetic field (RMF) is applied to this conductor, the resulting electromagnetic force will have an azimuthal direction and cause rotation of the arc around the axis of the contacts. The Br field is caused by the path imposed on the current in the contacts. Two types of contact structures are used to obtain this result (see fig. 10 ): v contacts of the spiral type, v contacts of the "cup" or "contrate" type. Correct functioning of RMF interrupters is linked to obtaining a satisfactory compromise at the contact geometry level and in particular of the slot width for contacts of the spiral type:

v if the width is too large, the arc has a hard time "jumping" from one part of the contact to the other, which may make it stationary at the end of the track and thus overheat part of the contact (since the arc is in the constricted mode); v if the width is too small, the slot may be easily filled by the fusion of contact material, and the current path, thus modified, leads to the disappearance of the RMF and immobilisation of the arc. Even though mobile, the rotating arc remains constricted and therefore exerts energetic force on the part of the electrode which carries it, the high pressure of the arcs roots expulses the molton contact material in the form of droplets. This process is an efficient means of limiting the overheating of the rest of the electrode (or to facilitate its cooling), for the energy brought by the arc is taken away with the expulsed material which has condensed on the surrounding walls; in return, it leads to relatively high contact erosion. c Axial magnetic field technique Ba When an arc plasma is submitted to a sufficient axial magnetic field (AMF), the electrons are obliged to follow trajectories that are parallel to the field lines which are helical-shaped lines, the axis of which is parallel to the contact axis since Ba is combined with the azimuthal field produced by the current itself.
I
I
F Br
Br
I
a - Spiral
I
b - Contrate
Fig. 10: contact structures used to create the RMF (spiral and "contrate").
The much heavier positive ions in the plasma are not controlled as efficiently by the field, but are retained by the electrostatic force developed by the negative space charge of the electrons trapped by the AMF: these electrostatic forces ensure that the plasma has a tendency to remain globally neutral. Consequently, electron confinement results in the confinement of all of the plasma in a column that corresponds to the field tube intercepted by the cathode: if this tube is parallel to the electrode axis, most of the plasma produced by the cathode arrives at the anode. The arc, in these conditions, conserves most of the diffuse mode characteristics although with a current density level that is considerably higher: v the arc voltage remains moderate since the plasma conserves its neutrality up to near the anode (no ion "starvation" phenomenon); v the tendency of the arc to concentrate on the anode side (the Hall effect) is interfered with by the AMF which forces electrons to maintain a trajectory that is essentially parallel to the axis; v if the surface of electrodes, in particular of the anode, which is intersected by the arc column, is sufficient for the current, then the energy density and thus overheating remain limited. The vaporisation of contact material is sufficiently reduced so that the nature of the plasma is not modified by the ionisation of neutral particles. Two main conditions need to be satisfied so that the arc remains in this diffuse column mode that is favourable to current interruption: v Ba must be sufficiently high. The critical AMF needed to prevent the formation of an anode spot is given in the experimental formula: Bacrit = 3.9 (Ip - 10) (Ba in mT, with Ip peak current value in kA),
v the surface of the electrode must be sufficient for a given current value: the current density not to be exceeded is in the region of 17 A/mm2 (RENTZ formula). In fact, this current density limit is only valid as a first approximation and the breaking capacity of interrupters with AMF does not change in direct proportion to the surface of the contacts. In fact, the initial constricted arc that was produced at contact separation and the time needed for it to occupy the entire available electrode surface must be taken into account: the breaking capacity, as a function of contact diameter, approximately follows a variation of d1.4 . The arc in an AMF interrupter is much less mobile than in a RMF interrupter. Even if the current density is sufficiently high to cause the fusion of anode material, projections remain limited. Contact erosion is therefore slighter than in a RMF, however the molton material remains in place and delays the cooling of the electrode surface. Due to this, even though in principle the use of the available contact surface appears more efficient in an AMF then in a RMF this is not always verified. In particular, for high currents and low voltages, in a RMF higher breaking capacities can be reached for a given surface, the price to pay however being significant erosion. Diverse solutions can be used to obtain the AMF between contacts by using the current being interrupted: v coils integrated behind the contacts (see fig. 11 ); v a magnetic circuit that channels the azimuthal field created by the power leads and which straightens it into an AMF in the inter-contact zone;
Incoming current
4 coil elements Contact disks (the slots prevent the flow of induced currents which oppose the coil currents) 4 coil elements
Outgoing current
I
Fig. 11: example of axial magnetic field contacts .

Part 3 of 3
Vacuum switching
P. Picot
Choice of the architecture c Vacuum interrupter components. A vacuum interrupter is made up of few components (see fig. 14 ). v Two electric contact assemblies of the butt type (since, under vacuum, sliding contacts would weld with each other); one is fixed, the other mobile. Each assembly includes a cylindrical electrode that conducts the current to the contact disks. v A gas-tight enclosure including an insulator that ensures electric insulation between fixed and mobile contacts. v A shield that protects the internal side of the insulator against condensation of metal vapour produced by the arc. v A metal bellows which allows for the mobile contact to move while maintaining the enclosure tightness. These are the basic components that are included in all interrupters. Moreover, in circuitbreaker interrupters, there are devices that generate magnetic fields (radial or axial) needed for breaking the power arc. Variations are mainly possible at the shield level and on devices that produce magnetic fields. c Shield configurations The main choices at the shield level deal with: v its fixing mode which determines its voltage: the voltage is fixed (is the same as that of the fixed electrode) if the shield is connected to this end of the interrupter, it is floating if the shield is fixed to an intermediary point on the insulator without an electric connection with one of the contacts.
Fig. 12: axial magnetic field interrupter with external coil.
v an external coil that surrounds the inter contact zone (see fig. 12 ). In general, the path imposed on the current, to create a sufficient AMF in the inter-contact gap, is longer than the one needed to create a local RMF. For a given volume, contact resistance is thus lower with the RMF technique, which is advantageous for circuit-breakers with high continuous current ratings. However, the shapes needed for RMF contacts are more angular than those of AMF contacts and therefore less favourable on the dielectric level: the AMF is thus advantageous for high voltages. The engineer thus chooses one of the two techniques in function of their respective advantages and depending on the foreseen application (see fig. 13 ).
Secondary shield Insulating enclosure Main shield Fixed contact
Mobile contact
Capacity: High continuous current High voltage rating Electrical endurance Breaking capacity legend: +++ = very good RMF technique +++ + + ++ ++ = good AMF technique + +++ +++ ++ + = average
Metal bellows Secondary shields
Fig. 13: comparison table for the two breaking techniques
Fig. 14: vacuum interrupter components.
v its position which can be inside or outside the enclosure, in the latter case the shield is part of the enclosure and must be gas-tight. By combining these different options, four configurations are possible, they are all used in function of desired characteristics. As a general rule: v a shield with fixed voltage is chosen when low cost is desired and a shield with floating voltage when high performance is sought. v an external shield is chosen for compacity in diameter and an internal shield because it is simple to make. c Devices that generate radial or axial magnetic fields Devices that produce the RMF needed to rotate the arc must be positioned as close as possible to the arc: they are therefore built-into the same structure as the contacts inside the interrupter. The two most common geometries were described in the preceding chapter: "spiral" contacts and contacts of the cup or contrate type. The choice of one solution over the other does not modify the general architecture of the interrupter. However, there are two possible architectural choices for AMF interrupters. In fact, the device that generates the AMF (most often elements of circular coils with an axis parallel to that of the interrupter) can be housed in the internal contact structure as with RMF interrupters, or outside of the interrupter. In the last case, there is a coil that surrounds the contact separation zone. The coil is in series with the fixed contact and the circuit current flows through it. Figure 15 shows a realisation of this type of configuration: it can be noted that, to reduce the dissipated power in the device, the coil is made up of three parallel elements. One of the disadvantages of this architecture is the path length imposed on the current to create a sufficient AMF in a significant volume. This therefore leads to greater losses that however do not necessarily result in greater temperature rise, the coils in the air being more efficiently cooled (by convection) than those that are integrated into the contacts on the interrupter. Moreover, the presence of a coil with the same voltage as the fixed contact, around the contacts, practically imposes the choice of a shield with a fixed voltage for this type of interrupter. One might think that the presence of an external coil presents a disadvantage in regard to interrupter dimensions by increasing its external diameter. In fact, the possibility of using the entire contact surface that is subjected to the relatively uniform AMF created by the external coils (which is not the case for contacts that have integrated coils) compensates this disadvantage and dimensions are comparable. The main advantage of AMF architecture with an external
I
Coil made up of 3 overlapping elements
Interrupter
Fig. 15: example of a coil, surrounding the contact separation zone, made up of three parallel elements.
coil is the possibility of making a compact interrupter, simple thus economical. The disadvantages on the heat dissipation and dielectric levels (because of the fixed shield design) make the architectures with devices integrated into the contacts (AMF or RMF) more attractive for high voltage levels (u 24 kV) or for high current ratings (u 3150 A). Choice of materials and manufacturing technologies For vacuum interrupters, the choice of materials and manufacturing technologies are guided by the need to: v guarantee the preservation of high-vacuum (< 10-3 mbar) needed to operate the interrupter for its life span (30 years), v ensure the rated performances and in particular the breaking capacity. c Choice relative to the requirements for vacuum quality All vacuum chambers are subjected to deterioration of the vacuum level that is linked to degassing phenomena which appear when pressure reaches sufficiently low values. Degassing is first of all a surface phenomenon that corresponds to the detachment of gas molecules absorbed on the walls. This gas is rather easily and rapidly eliminated by relatively moderate heating (in the region of 200 C) of the walls of the enclosure during pumping.
Then volume degassing, which corresponds to the diffusion, towards the surface of metal materials, of dissolved gases such as hydrogen, appears. To prevent degassing, mainly coming from massive parts, from progressively degrading the vacuum level of the interrupter, it is important to: v use materials with as low a gas content as possible (for example copper Cu-OFE oxygen free); v proceed with high degassing of materials by conducting long-term pumping of the interrupter at a sufficiently high temperature (typically for some ten hours at a temperature in the region of 500 C). Gases bound to the metals (in the shape of chemical compounds) are not sensitive to degassing, however they can be freed under the arc effect. Therefore, the materials used for arc contacts must be elaborated in a vacuum environment so as to have the lowest possible gas content. The enclosure of the interrupter must be perfectly tight, which implies the absence of leakages and permeation in service conditions. That is why enclosures are made of metal and ceramic materials: insulators made of alumina ceramic have replaced glass for they can tolerate much higher temperatures and thus allow for better degassing. Bonding between the metal parts of the enclosure are welded or brazed. Ceramic and metal are also brazed, either using reactive brazing which enables a direct bond with the ceramic, or using traditional brazing by coating the ceramic with metal beforehand (Mo-Mn + Ni). Final brazing operations are conducted in a furnace, under a vacuum to ensure the degassing of materials. More and more often, sealing of the interrupter is conducted during the brazing under vacuum process as well, which allows for the pumping operation to be eliminated. Taking into account the tightness level needed to allow the mobile contact to move, the metal bellows is the only solution used. It is generally made of thin austenitic stainless steel (typically 0.1 to 0.2 mm). Its design and that of brazing joints with the rest of the enclosure must be carefully studied so as to ensure high mechanical endurance despite the unfavourable effect of thermal cycles imposed by brazing. Lastly, materials that are used in small quantities, but which play an important role in obtaining and maintaining high-vacuum over time must be mentioned. Getters are based on very chemically active metals (barium, zirconium, titanium, etc.) with most of the gases that are likely to be found in vacuum enclosures. The getters are activated, under high-vacuum, through heating at a sufficient temperature to cause the diffusion of the passivated superficial
layer into the bulk, and the regeneration of an active metal surface capable of absorbing the gas molecules that are in the interrupter. This activation operation is conducted during pumping or when the interrupter is sealed using brazing under vacuum: it is in particular due to getters materials that this last procedure, more industrial than pumping, while ensuring a satisfactory quality of vacuum, was able to be developed. c Choice of contact material. Good contact material for a vacuum interrupter must meet a certain number of requirements: v be a good electrical conductor, so as to offer reduced contact resistance; v present good mechanical resistance to repeated shock which the contacts undergo when closing; v must not form solid welding upon on-load or short-circuit closings so that the opening mechanism can separate the contacts and so that the break of the welded zone does not create excessive damage to their surfaces; v produce little metal vapour during the arc phase so as to enable rapid dielectric recovery of the inter-contact gap after breaking, which implies: - low vapour pressure, - reduced droplet production during the material fusion phase; v present good dielectric characteristics during the TRV application phase, which implies: - a sufficiently smooth surface, without any notable roughness (low ), - no overheated points emitting by thermionic effect (case of refractory materials with reduced thermal conductibility), - no likelihood of forming easily detachable particles; v allow the existence of stable cathode spots up to low current values so as to minimise the chopped current and overvoltages associated with this phenomenon, which in particular implies a sufficiently high vapour pressure. It turns out that these numerous required qualities are sometimes contradictory. It is thus necessary to find an acceptable compromise for the foreseen application in function of privileged properties which are: v for circuit-breakers, dielectric recovery after the high current arc phase (good breaking capacity); v for contactors, low erosion and minimum chopped current (electrical endurance and reduction of overvoltages); v for switches, resistance to welding and dielectric withstand under high voltages (absence of restrikes). Presently, the best compromises have been obtained with composite materials and the three material families that are the most often used are: v CuCr for circuit-breaker applications; v AgWC for contactor applications;
v WCu for switch applications and in particular those designed for the control of high voltage capacitors. CuCr have been proven to be the best materials for circuit-breaker applications and do not appear to be able to be dethroned in the shortrun, even if changes cannot be excluded. Proportions used vary between 80 and 50 % for Cu, the remaining percentage for Cr. A high proportion of Cu is favourable for the electric conductivity (low contact resistance) and thermal conductivity (good evacuation of arc energy). A high proportion of Cr is favourable for withstanding welding and dielectric withstand under high voltage. The gas content of material must be as low as possible since, when it is fused or vaporised,
these gases are freed into the inter-contact gap and are harmful to breaking. The long-term effect on the vacuum level is less disturbing than could be imagined since Cr condensed on interrupter walls plays the role of getter and reabsorbs these gases. Lastly it must be noted that the arc modifies the superficial layer of the material and improves its qualities by: v eliminating included gases and surface oxides, v obtaining very fine granulometry (precipitation of Cr melted in the copper matrix), v homogenising material. This effect is sometimes qualified as "current conditioning" (through analogy with voltage conditioning): in general the behaviour of contacts and the breaking performance improve after a few breakings.
3 Breaking in vacuum and overvoltages during switching of inductive circuits
Vacuum switching devices (contactors, circuitbreakers, switches) are likely to generate overvoltages when interrupting current in inductive circuits (no-load transformer, non-charged motor or motor in the start-up phase). Due to the special properties of vacuum, these overvoltages can be of a different nature than those generated in the same conditions by
switchgear that uses another type of medium (air, SF6, oil, etc.). In general these overvoltages do not pose a problem and do not need any special device. However in the case of sensitive loads (for example motors) it is recommended to install overvoltage limiting equipment.
3.1 Overvoltage generating phenomena

Overvoltage associated with an ideal breaking Even in the theoretically perfect breaking case using an ideal circuit-breaker, a certain overvoltage level is inherent to the interruption of current in an inductive circuit. Indeed, voltage values at the terminals of different circuit elements must reach a new steady state that corresponds to the open state. The transition in regards to the closed state preceding the breaking moment (current zero) leads to oscillations around the new steady state and produces overvoltages in comparison to normal maximum network voltage (see fig. 16 ). In the case of a three-phase interruption, the fact that the interruption is not simultaneous on the three phases also introduces a transient state which generates overvoltages. As an example, in the case of the breaking of a short-circuit current in a system without a directly earthed neutral, the recovery voltage at the terminals of the first pole to clear reaches roughly 2.1 to 2.2 p.u. (IEC standardised TRV) and 2.5 p.u. for the breaking of a capacitor bank with isolated neutral. Current chopping The best known and most widespread phenomenon, for it deals with all breaking techniques, is current chopping: premature interruption of the alternating current before its natural passage through zero. This phenomenon above all concerns circuit-breakers, that are sized for breaking short-circuit currents, when they interrupt small currents. If Ia is the chopped current value, current that flows in the load inductance L immediately before breaking, the electromagnetic energy that is stored in the load is transferred in the form of electrostatic energy in capacitance C located at the terminals of the load (1/2 L Ia2 = 1/2 C V2). A voltage increase on the load side appears which accentuates the gap in regard to the "open circuit" steady state and amplifies the
t=0 E i Vx Vy L
Vx Vy Vx t=0 t i Vy
(Vx -Vy)
2,0 p.u.
1 p.u. = maximum nominal phase to earth voltage Un 2 = 3
Fig. 16: overvoltages in comparison to maximum normal network voltage during the breaking of an inductive circuit.
TRV (Ia = 0) TRV (Ia > 0)
Ia
Vs
Vs : Suppression voltage = Ia
L C
Fig. 17: overvoltages associated with the breaking of a circuit with current chopping.
overvoltages associated with the break (see fig. 17 ). These overvoltages are therefore proportional to the chopped current and the characteristic impedance (surge impedance)
L of the load. C
Material Cu CuCr AgWC
Ichopped ave.
15 4 0.5
Ichopped max.
21 8 1.1
In the case of vacuum switching, current chopping corresponds to the premature extinguishing of the last cathode spot due to its instability at low current values: this characteristic primarily depends on the nature of contact material. The average chopped current values for a few common materials are given in the following table (see fig. 18 ). In practise, chopped current values of a few amps, characteristic of the CuCr material, do not pose a problem. However values obtained using pure copper are excessive and explain, with other considerations, that this material cannot be used as such. Multiple pre-strikings and re-ignitions There is striking between the contacts when the applied voltage is higher than the dielectric withstand of the interval. This phenomenon is inevitable when this interval is very short (at the end of closing and at the beginning of opening). Pre-striking upon closing thus systematically occurs when the operation is conducted under voltage: the time interval between the prestriking and the moment when the contacts touch each other (pre-arcing time) depends on the closing speed and the voltage value applied at the moment when the contacts move closer to each other. Re-ignition upon opening only occurs if the arcing time (time interval between contact separation and current break) is low: in this case the contact gap is not sufficient enough to tolerate the TRV and there is another dielectric breakdown.
Fig. 18: average chopped current values for a few common materials (Cu, CuCr, AgWC).
During pre-striking or re-ignition, the oscillating discharge of local capacitances results in an HF current (some ten kHz) that flows between the contacts superimposed on the power frequency current that progressively establishes itself (as it is nil before ignition). These inevitable phenomena concern all types of switchgear. The particularity of vacuum switchgear is their ability to interrupt HF current following striking whereas other breaking techniques are in general incapable of this due to high di/dt at the time this current passes through zero. The breaking of HF current generates a new applied TRV between the contacts the gap of which has only slightly varied, for these phenomena occur on a small time scale in comparison to the contact movement time, which thus leads to new striking and repetition of the same phenomena (see fig. 19 ). There is a succession of multiple strikings associated with variable amplitude voltage waves depending on the change in the contact gap: v upon closing the amplitude of the overvoltage train linearly decreases until the contacts touch each other, v upon opening amplitudes increase until the gap between the contacts is finally sufficient enough to withstand the recovery voltage which, due to voltage escalation, is still higher than the voltage that corresponds to normal breaking. Overvoltage trains with steep fronts, generated by these multiple striking phenomena, are
ud(t) dielectric strength of the inter-contact gap
Fig. 19: succession of multiple strikings associated with voltage waves with varying amplitudes.
Laboratoire VOLTA C2581 99/04/22/035 0.01 ms
V1 AVAL 25 kV
V2 AVAL 25 kV
V3 AVAL 25 kV
Fig. 20: multiple striking phenomena during contact separation and the breaking of small inductive current.
therefore still limited by the inter-contact gap that is maintained and which plays the role of sparkgap. However this limitation is only truly efficient upon closing; upon opening, the values reached can be high (see fig. 20 ). The characteristics of these two types of similar phenomena are summarised in the table in figure 21.
Type of multiple striking
Occurrence Amplitude of overvoltages Low High
Pre-striking upon closing Systematic Re-ignition upon opening Occasional
Fig. 21: characteristics of overvoltages linked to multiple striking phenomena.
The disadvantage of these overvoltage trains are due more to their steep front than to their amplitude. In fact, these voltage waves with low rise times (in the region of 0.2 to 0.5 s) are not distributed in a uniform manner in transformer and motor windings, rather they mainly stress the first turns (see fig. 22 ). They can therefore cause deterioration and accelerated ageing of the insulation between these turns.
Fraction of the overvoltage applied to the 1st turn of the winding 1.0 0.7 0.4 0.2 0.1 0.1
0.2
0.4
0.7 1
4 7 10 Rise time (s)
Fig. 22: percentage of the overvoltage applied to the first coil in the winding depending on the rise time.
Virtual current chopping In special configurations (rarely encountered in practice) that are characterised by strong capacitive / inductive coupling between circuit phases, the multiple re-ignition phenomena on the first phase that attempts to break, lead not only to significant HF current oscillations in the phase dealt with, but also in the neighbouring phases in which a notable current still flows, for they are far from their natural zero. If induced HF currents reach an amplitude exceeding that of the power frequency current, current zeros ("artificial" but nonetheless real, and not virtual) are produced. The device can take advantage of it to break the current well before its natural zero. In such cases the chopped currents can be tens, or even hundreds of amps and the associated overvoltages are very high. A possible solution is to open one of the device poles in advance so that during the time interval when multiple re-ignitions are likely to occur, the two other phases remain closed and thus insensitive to induced disturbances. In practice, this solution has not been applied due to the problems that it poses (stress non-uniformly distributed between the poles during the breaking of a short-circuit current) and due to the exceptional character of the phenomenon.
3.2 Means of protection against overvoltages

"Soft" contact materials Contact materials (ex: AgWC, CuBi) that have a very low chopped current value were developed for the contactor application. This performance was reached by combining low thermal conductivity with high vapour pressure so as to obtain stable cathode spots up to very small current values. These characteristics go against the breaking capacity: that which is acceptable for a contactor application is not acceptable for a circuit-breaker application. Furthermore, the use of these materials is only efficient in reducing overvoltages linked to chopped current, which does not pose a problem in practice if it does not exceed a few amps (case of CuCr). "Soft" contact materials do not bring an improvement when compared with traditional "hard" materials (CuCr) on the multiple striking level. In fact these materials are also capable of breaking currents with high di/dt and are characterised by a slower dielectric recovery rate after contact separation (see fig. 23 ): consequently overvoltage trains with steep fronts are not eliminated but, on the contrary, have a tendency to remain longer than with better performing material for breaking.
Breakdown voltage (kV) CuCr 14 12 10 8 6 4 2 0 0 100 200 300 400 500 Time after contact separation (s) AgWC
Fig. 23: change in the dielectric withstand between contacts from the moment of their separation depending on their materials.
Synchronised breaking A theoretical solution to eliminate these multiple re-ignition phenomena would be to control the moment when contacts are opened in regard to the current wave so as to prevent short arc times. In practice, it poses complex reliability problems concerning the response time of the control mechanism; it is therefore only used in
the High Voltage field where mastering switching overvoltages can justify the cost difference at the switchgear level. In the Medium Voltage field it is more economical to call upon overvoltage protection devices when a load is to be protected. Protection devices providing overvoltage limitation As mentioned above, the worst phenomenon is that of multiple strikings which calls primarily upon the first turns of transformer or motor windings. These two types of load must be taken into consideration separately. Indeed, transformers are designed to tolerate dielectric stress generated by lightning impulses which are overvoltages with steep fronts, they thus have a good level of insulation of the first turns. Moreover the inductive currents to be switched are small (no-load transformer) and
associated overvoltages remain limited. As a general rule, it is not necessary to provide special protection for transformers that are operated by vacuum switchgear, except possibly for solid insulated transformers that are more sensitive than those insulated in oil. Motors have a dielectric withstand lower than that of transformers, whereas the currents to be interrupted may be high (breaking in start-up phase or stalled rotor) and thus overvoltages are severe. As a general rule, it is recommended to place protective devices at terminals of a motor whatever its control device may be, contactor or circuit-breaker, and whatever the contact material used may be. These devices may be capacitors which reduce the rise time of overvoltages, or RC circuits (typically C in the region of 0.1 to 0.5 F and R of 10 to 50 ) and/or ZnO surge arresters.
4 The main fields of application for vacuum switching
Vacuum properties as a breaking medium for electrical switchgear are summarized in the table in figure 24.
Field Breaking capacity
Characteristics Very rapid dielectric recovery
Strong points Breaking of fault currents with severe di/dt and TRV.
Weak points Breaking of HF currentsfollowing restrikes: overvoltages are generated, protection devices necessary in certain networks. No current limiting effect in LV.
Low arc voltage (energy). Ability to break even without contact movement.
High electrical endurance. Current interruption in case of striking between open contacts (partly compensates for the lack of reliability of the dielectric withstand).
Dielectric withstand
Influenced by the surface condition of electrodes and the presence of particles. Influenced by the arc phase that immediately preceded.
Intrinsic dielectric withstand limited in HV and may change over time. Random post-break dielectric withstand: risk of re-striking after capacitive breaking if the interrupter is not adapted. High contact pressure needed to prevent "popping" by electromagnetic force. Constant contact resistance (no oxidation and no deterioration upon breaking). Tends to weld upon closing.
Current flow
Non-compensated contacts of the butt type. Contacts in vacuum.
Same contacts for continuous current flow and breaking. Breaking environment Vacuum < 10-3 mbar. No decomposition products and no effects on the environment.
High contact resistance: significant thermal dissipation for high ratings. Permanent monitoring of the vacuum level is impossible: periodic dielectric checks make shutdown necessary.
Fig. 24: vacuum properties as a breaking medium.
These strong and weak points of the vacuum switching technique have thus led to its use being favoured in certain fields of application for electrical switchgear. In the presentation that follows, the different fields of application are segmented in the following manner: c by voltage level; c then by function, or type of switchgear; c lastly, depending on the type of load to be switched.
This chapter successively reviews the Medium Voltage (MV: 1< U < 52 kV), Low Voltage (LV: U < 1 kV) and High Voltage (HV: U u 52 kV) fields. The section that is the most developed is dedicated to MV which is the primary field of application for the vacuum switching technique. The LV and HV fields are only briefly described for the intrinsic limitations of vacuum switching only allow for this technique to occupy a marginal position: dominating techniques are breaking in air for LV and breaking in SF6 for HV.
4.1 Vacuum switching applications in Medium Voltage

Medium Voltage is primarily used for electrical energy distribution, between the transmission over long distances that is carried out using High Voltage (HV) and use that is mainly carried out in Low Voltage (LV). The lower voltage levels of the MV field are also used to supply loads of unit power that is too high for LV. In MV, the main types of switchgear that are used are switches, disconnectors, circuitbreakers and contactors (see fig. 25 ). Switches are simple and relatively economical devices that are used in normal operation of electrical networks: they are operated upon an order coming from an operator and allow the current to be established or interrupted in a network element. They are capable of breaking the normal load current of the circuit in which they have been inserted, and to establish the fault current caused by a short-circuit located downstream from their position in regard to the supply of electrical energy. General purpose switches that are designed for MV distribution networks, upon which the switching frequency is low, have, through their design, an electrical and mechanical endurance that is relatively limited, typically: c some hundred breaks at In; c some thousand mechanical operations. For special applications, certain types of switches must be able to counter more severe stress, for example: c switches for arc furnaces operate frequently with high currents; c switches for back-to-back capacitor banks that operate rather frequently and must establish inrush currents (with high frequency and amplitude). Disconnectors are not strictly speaking breaking devices for they operate without a load (they must however be able to interrupt the residual capacitive currents of open circuits). They are used to isolate a circuit from the rest of the network and allow for safe intervention on the circuit. To that effect, they must have a high dielectric withstand between contacts and must respect the construction measures that aim at preventing the crossing over of the isolating distance even in the case of overvoltage on the network. Despite these measures, the safety of persons intervening in the system is not fully guaranteed unless the network element that was isolated by a disconnector is earthed in an efficient manner as well. Disconnectors are often combined with switchgear that does not satisfy the disconnection function, in general circuitbreakers and contactors. Switches are, however, most of the time also able to fulfil the disconnection function: they are then referred to as switch-disconnectors. Circuit-breakers are safety devices that protect the network by automatically separating the faulty sections of the network: they are able to interrupt maximum short-circuit current likely to occur at the place where they are installed. Circuit-breakers can therefore be considered as high-performance switches that are capable of operating upon an order from an operator or
Type of switchgear Switch
IEC definition A switching device capable of making, carrying and breaking currents under normal circuit conditions which may include specified operating overload conditions and also carrying for a specified time currents under specified abnormal circuit conditions such as those of short circuit. (IEV 60050-441-14-10). A switching device which provides, in the open position, an isolating distance in accordance with specified requirements. (IEV 60050-441-14-05). A switching device capable of making, carrying and breaking currents under normal circuit conditions and also making, carrying for a specified time and breaking currents under specified abnormal circuit conditions such as those of short circuit. (IEV 60050-441-14-20). A switching device having only one position of rest, operated otherwise than by hand, capable of making, carrying and breaking currents under normal circuit conditions including operating overload conditions. (IEV 60050-441-14-33).
Applicable standard for MV IEC 60265-1
Disconnector
IEC 60129
Circuit-breaker
IEC 60056
Contactor
IEC 60470
Fig. 25: standardised definitions of the main types of switchgear.
from an automatic protection device that detects fault situations. These devices must be highly reliable since safety and network availability depend on their correct operation. Circuit-breakers require higher electrical and mechanical endurance than switches, typically: c from 10 to 100 short-circuit current breaks; c from 2000 to 10000 mechanical switching operations and breaks at In. Contactors are control devices for loads that function in an intermittent manner, notably electric motors. They are switches with high operating rates that must be able to break overload currents that are higher than nominal current (ex: starting motor or stalled rotor currents) but not short-circuit currents which are eliminated by a combined protection device (circuit-breaker or fuse). Their high mechanical and electrical endurance generally amounts to several hundred thousand operations. The graph in figure 26 enables the respective positions of the four types of switchgear described above to be visualised. One of the strong points of the vacuum switching technique is its ability to obtain a high breaking capacity and electrical endurance: that is why this technique is primarily used for circuitbreakers and contactors.
Circuit-breaker application in MV A high breaking capacity is required for a circuitbreaker application. Vacuum interrupters used for this application either call upon RMF technology, or on AMF technology. Both can reach the highest breaking capacities required in MV (up to 63 kA); they are thus used in function of their respective advantages (see fig. 13 ). As with SF6, vacuum offers for this application the advantages of an enclosed break with no external manifestations and a maintenance free design with high electrical endurance. The very rapid dielectric recovery of the vacuum can be an advantage in comparison with SF6 in special applications for which the rate of rise of the TRV is faster than that required by the IEC 56 and ANSI C37-06 standards (ex: case of a circuit-breaker directly connected to the secondary of a high power transformer). In such cases, not very frequent for standardised TRVs cover the great majority of applications, vacuum circuit-breakers need less derating than SF6 circuit-breakers. Since vacuum switching is conducted without an external energy supply, vacuum circuit-breakers require less operating energy than SF6 circuitbreakers of the puffer type. For that which deals with SF6 circuit-breakers with rotating arc or with self-expansion, the gap is less significant.
Operating cycles 100000
10000
1000
100
10
Disconnector
Switch
Contactor
Circuit-breaker
1 Fuse
Isolate
I = 0 Load current I = In
switching
Motor starting
I = 10In
Short- I = 100In circuits
Fig. 26: respective positions of the four types of switchgear in terms of the current to be broken and of the number of operations to be conducted.
This advantage is however counterbalanced by the inherent disadvantages of the vacuum technique which can only use butt contacts. These contacts need high contact pressure to prevent repulsion and contact welding upon closing on fault: contact pressure needed per pole is in the region of 200 daN for a 25 kA circuit-breaker and of 600 daN for a 50 kA circuit-breaker. This requirement leads to a rise in the operating energy for closing and to reinforced pole structure that must tolerate these permanent stresses in the closed position. Furthermore, despite high contact pressure, the use of butt contacts made of CuCr material does not allow for contact resistance as low as with silver-plated multiple contacts to be obtained: vacuum circuit-breakers thus have a handicap in comparison with SF6 circuit-breakers through higher thermal dissipation for high nominal currents (2500 A and above). Contacts in vacuum tubes, sheltered from oxidation, are not limited in overheating, unlike contacts of other circuit-breakers, but the interrupters environment must evacuate the calories generated by it while respecting the admissible temperature limits on the connections and contacts; that is why vacuum circuit-breakers with high ratings are characterised by particularly large-sized connections and cooling fins. Aside from their environment, vacuum interrupters are limited in overheating as well, not through the nature of the materials of which they are made or through their manufacturing process (high temperature brazing), but rather due to the properties of gas permeation through metal walls (in particular of the metal bellows) which become significant concerning atmospheric hydrogen as of 200-250 C. In conclusion, the vacuum switching technique is well adapted for general purpose circuit-breaker applications in MV and covers all of the normally required performances for voltage, nominal current and breaking capacity. For special applications such as the control of inductive or capacitive loads, special precautions must be taken, or other technologies may be better suited. Contactor application in MV This application is especially well adapted to the vacuum switching technique, which has acquired a dominating position in this segment. In fact, currents to be broken are located in the range of currents that are easily interrupted by diffuse vacuum arcing, with contacts that have simple shapes and low contact material wear, thence excellent electrical endurance. Contact pressure can be low, since nominal currents are modest and fault current is limited through the use of combined fuses, and even more so since the contact materials used have a very reduced tendency to welding and thus tolerate a certain degree of repulsion.
Supply voltages for MV motors located in the bottom of the MV range (in general i 7.2 kV) authorise a small contact gap (in the region of 4 mm) and the realisation of compact interrupters which have high mechanical endurance and are especially well adapted to electro-magnet operating mechanisms. All these advantages explain the success of the vacuum switching technique for the MV contactor application. However, the risk of overvoltages during the switching of inductive circuits, which is specific to vacuum technique, must not be overlooked (motor in the start-up phase, no-load transformer) and the need for adapted protection devices (see chapter 3). This problem, which concerns all types of vacuum switchgear, must be especially taken into account in the case of motor switching, motors being loads that are sensitive to overvoltages. Switch and disconnector applications in MV The vacuum switching technique which allows for MV circuit-breakers and contactors to be made, can also, a fortiori, fulfil the more modest requirements of switches. It has however encountered limited success for this application. In fact, low performances can, in general, be obtained in a more economical manner by using breaking techniques in air or in SF6. But above all, this function is often combined with the disconnector function, which is easily attainable using the air or SF6 technique, but not using vacuum. Combining a vacuum switch with a conventional disconnector makes this solution non-competitive. The impossibility of ensuring disconnection with a vacuum interrupter is due to the voltage deconditioning phenomenon which is caused by the deterioration of the contact surface condition caused by mechanical and electrical switching operations. This deconditioning does not allow for the dielectric withstand that was obtained without any particular difficulty at the end of the voltage conditioning procedure on a new interrupter, to be guaranteed. Furthermore, it is impossible to continuously monitor the integrity of the dielectric medium in a vacuum interrupter which also limits its use as a disconnector. With a switch, closing upon a short-circuit is particularly penalising for the dielectric withstand between contacts, for it is not followed by a fault current interruption that is capable of eroding the roughness caused by the break of the contact weld due to the pre-striking. To prevent significant deterioration of their dielectric withstand, during consecutive closings upon short-circuit, switch contacts are made of materials that do not easily weld such as WCu, instead of CuCr which is used for circuitbreakers. For special applications which require high electrical endurance (ex.: switches for arc furnaces), the vacuum switching technique is
well suited and is widely used, even if overvoltage problems due to vacuum can, in certain cases, privilege the use of SF6 technology despite its lower endurance. Another special application is that of the back-to back capacitor bank switch, which can be ensured by using a standard SF6 circuit-breaker, but which, using the vacuum technique, requires a special interrupter. In fact, the electrical charge of the capacitor induces a recovery voltage, applied to the terminals of the switching device, that is especially high. The post-break dielectric withstand of a vacuum interrupter is not its strong point due to the possibility of breakdown caused by the particles generated during the arcing period (see chapter 2). In the case of capacitor banks in parallel that are separately closed, the risk of re-striking is
accentuated by the effect of the high frequency inrush current due to the discharge of neighbouring capacitors in the one which is energised: this inrush current imposes the use of contact materials of the WCu type which do not easily weld and which is incompatible with the vacuum circuit-breaker application. Furthermore, to prevent attempts to interrupt HF inrush current during the pre-striking phase, which result in overvoltages that are harmful to capacitor banks, measures must be taken: the addition of surge inductances reduces the inrush current frequency, raising the closing speed reduces the pre-striking time. In brief: the vacuum switching technique is not to be excluded for controlling capacitive loads, but other techniques, in particular the SF6 technique, are better suited.
4.2 Vacuum switching applications in Low Voltage

The vacuum switching technique, widely used in MV for the circuit-breaker and contactor functions, can also fulfil the same functions in LV. It is however rarely used at this voltage level. In fact, on the one hand, it competes with the air breaking technique which is simpler, more economical and better adapted, and on the other hand, the disadvantages that have been noted in MV use are more disturbing in LV. The main shortcoming of the vacuum switching technique in LV for the circuit-breaker function is due to the low arc voltage which cannot reach or exceed the network voltage like in an air circuitbreaker: it therefore cannot limit the fault current to a notably lower value than the prospective short-circuit current. This limiting effect is particularly useful for it avoids intense electrodynamic forces, that would be produced by prospective short-circuit currents that are often high in LV (up to 100 kA and above). This limiting effect also facilitates the natural selectivity between circuit-breakers for it is all the more accentuated, the smaller the circuit-breaker rating. Furthermore, the disadvantages of vacuum interrupters, mentioned above for MV, that are linked to the use of butt contacts (high contact pressure and relatively high contact resistance), are more disturbing in LV power circuits which are characterized by high values of short-circuit currents (non-limited) and need higher continuous current ratings than in MV. Lastly the high breaking capacity needed in LV imposes penalising dimensions for vacuum interrupters in comparison to air solutions for circuit-breakers with small current ratings. For these different reasons, the use of vacuum interrupters in LV circuit-breakers is limited to a restrained section that corresponds to the following performances: c breaking capacity i 75 kA, c ratings between 800 and 2500 A. In this context, even though vacuum switching is not cheaper than in air, it is worth considering for the following reasons: c enclosed breaking with no external manifestations, c use in polluted and explosive atmospheres, c higher electrical endurance. In LV contactor use, the disadvantages of vacuum interrupters, mentioned above for use in circuit-breakers are no longer to be taken into consideration. The main factors that slow down the development of this technique in this field are: c first of all, the cost advantage in favour of classical air solutions; c then, the specificities of vacuum concerning overvoltages generated during the interruption already explained in chapter 3. In brief, in low voltage, the vacuum switching technique is not really able to compete with air breaking, except in special cases where enclosed breaking is significantly advantageous.
4.3 Vacuum switching applications in High Voltage

In the field of HV, the vacuum switching technique can be considered for use in the circuit-breaker function: diverse attempts have been made, without convincing success to date. In fact, it seems that the characteristics of vacuum switching do not allow it to truly rival the SF6 breaking technique in High Voltage. One of the main difficulties to be overcome is the production of vacuum interrupters with a sufficiently high unit voltage rating. If vacuum interrupters capable of breaking under 36 kV are commonly made, already as of 52 kV it is often necessary to use two interrupters in series. Today, the highest voltage level at which a circuit-breaker equipped with a single interrupter per pole is available on the market is 72.5 kV. Interrupters designed for use in applications at 123-145 kV are still, at present, in the prototype stage whereas SF6 breaking chambers up to a unit voltage rating of 420 kV are available. The solution which consists in placing a large number of interrupters in series to reach high voltages, above and beyond the technical problems that it poses (voltage distribution, reliability, etc.), can obviously not financially rival with the SF6 breaking technique. The main obstacle for obtaining a vacuum interrupter with a high unit voltage rating is the ceiling value of the dielectric performance for high voltages that is around 500 kV (see fig. 5), which corresponds to the lightning impulse voltage level to be reached for 123-145 kV devices. Presently, no technological solution is foreseeable. Furthermore the use of vacuum at high unit voltage poses the problem of X rays that are likely to be emitted by interrupters subjected to supply voltages in the region of a hundred kV. Here we are dealing with voltage levels applied to MV interrupters to condition them: this operation is carried out in shielded enclosures so as to protect the operators against X-ray emission. Using interrupters that have a dielectric design which is adapted to HV and already conditioned, the emitted radiation level (in the open position) should remain acceptable, but since interrupter operation can lead to partial deconditioning, this concern cannot be totally eliminated. For physical limitation reasons, vacuum switching cannot therefore even come close to rivalling SF6 except for the lowest voltage levels in HV and only in unfavourable economic conditions. For very special applications, the combination of the two techniques, vacuum and SF6, can be foreseen, as was done for a 250 kV direct current circuit-breaker that uses a vacuum interrupter in series with an SF6 breaking chamber. This solution combines the qualities of vacuum, for breaking with high di/dt and initial TRV rate of rise, with those of SF6 which relays it to ensure withstand at the end of the TRV rise. For common HV circuit-breaker applications, it is not certain that hybrid solutions can rival on a financial level, with solutions that are 100 % SF6, even if on the technical level such solutions are attractive because they allow for the qualities of each breaking technique to be combined.
5 Conclusion
To conclude this overview, the vacuum switching technique appears, because of its good breaking capacity and electrical endurance performance, to be in general well adapted to circuit-breaker and contactor applications in medium voltage. However the SF6 breaking technique is often better suited when privileged characteristics are dielectric withstand, low level of switching overvoltages or the ability to deal with high continuous currents. Even though it has matured, the vacuum switching technique still presents notable potential for progress to be made concerning its performances, in particular using the relatively recent AMF technology. And so, the trend to reduction of the circuit-breaker interrupter size should be maintained. For this, progress is to be made in the optimisation of the use of contact surfaces and in the increase of permissible current densities. With these objectives, current research is primarily focused on: c modelling of the arc and its interactions with the axial magnetic field; c the mechanisms for diffusing and distributing the arc energy on the surface of contacts; c improving contact material characteristics. To widen the fields of application of the vacuum switching technique, and better use its qualities, switchgear manufacturers also foresee new solutions and notably its combination with other techniques, in particular with the SF6 technique, so as to combine their respective advantages. This approach is already used for certain medium voltage cubicles with gas insulation that unite the qualities of vacuum switching with those of insulation in SF6. Another possibility, as of yet little explored, is the realisation of hybrid circuit-breakers that combine two breaking techniques, vacuum and SF6. A priori more expensive, it could however prove to be interesting in certain fields of application if it can efficiently conciliate the best of the two technologies.
Bibliography
Schneider Electric Cahiers Techniques c MV breaking techniques, Cahier Technique no.193 - 1999 S. THEOLEYRE Other publications c Reference works v Circuit interruption - Theory and techniques, 1984 THOMAS E. BROWNE Marcel Dekker, Inc. v Vacuum Switchgear, 1994 ALLAN GREENWOOD IEEE (Institute of Electrical and Electronics Engineers) v Handbook of vacuum arc science and technology - Fundamentals and applications, 1995 RAYMOND L. BOXMAN, PHILIP J. MARTIN, DAVID M. SANDERS Noyes publications c General Info - Breaking v Circuit-breaker application (in Circuit interruption), 1984 CHARLES L. WAGNER Marcel Dekker, Inc. c Vacuum insulation v The effect of contact geometry and HV conditioning on the field intensification factor of vacuum interrupter contacts, Elektryka Nr 38 -1989 R. P. P. SMEETS, H. F. REIJNDERS, W. M. C. VAN DEN HEUVEL, Y. H. FU v Isolation sous vide, D2540 et D2541 - 1991 ANNE ZEITOUN-FAKIRIS Techniques de l'ingnieur c Vacuum arc v Sheath growth in a low pressure plasma, February 1971 J. G. ANDREWS, R.H. VAREY The physics of fluids v Interruption in vacuum (in Circuit interruption), 1984 CLIVE W. KIMBLIN, PAUL G. SLADE, ROY E. VOSHALL Marcel Dekker, Inc. v Physique de la coupure du courant de courtcircuit dans les disjoncteurs vide, Revue ABB - 5/1993 EDGAR DULLNI
v Measurements and modelling in the current zero region of vacuum circuit-breakers for high current interruption, 1996 JOHANNES KAUMANNS IEEE 17th ISDEIV-Berkeley v Post-arc dielectric breakdown in vacuum circuit-breakers, August 1999 STEPHEN W. ROWE IEE High voltage engineering symposium, 22-27 c Axial magnetic field v Interaction between a vacuum arc and an axial magnetic field, 1978 H. C. W. GUNDLACH IEEE 8th ISDEIV-Albuquerque v Axial magnetic field type vacuum circuitbreakers based on exterior coils and horse shoes, 1986 H. SCHELLEKENS, K. LENSTRA, J. HILDERINK, J. TER HENNEPE, J. KAMANS IEEE 12th ISDEIV-Shoresh v The current distribution and the magnetic pressure profile in a vacuum arc subject to an axial magnetic field, 1986 I. IZRAELI, R. L. BOXMAN, S. GOLDSMITH IEEE 12th ISDEIV-Shoresh v Effect of an axial magnetic field upon the development of the vacuum arc between opening electric contacts, March 1993 B. SCHULMAN, PAUL G. SLADE, J. V. R. HEBERLEIN IEEE transactions on components, hybrids, and manufacturing technology v Evaluation of AC axial magnetic field needed to prevent anode spots in vacuum arcs between opening contacts, March 1994 BRUCE SCHULMAN, JOHN A. BINDAS IEEE transactions on components, packaging, and manufacturing technology v Technological progress of axial magnetic field vacuum interrupters, 1996 K. WATANABE, E. KANEKO, S. YANABU IEEE 17th ISDEIV-Berkeley v Arc behaviour in axial magnetic field vacuum interrupters equipped with an external coil, 1998 HANS SCHELLEKENS IEEE 18th ISDEIV-Eindhoven
c Contact materials v Contact materials for vacuum switching devices, 1992 F. HEITZINGER, H. KIPPENBERG, K. E. SAEGER, K.-H. SCHRDER IEEE 15th ISDEIV-Darmstadt v Advances in material development for high power, vacuum interrupter contacts, March 1994 PAUL G. SLADE IEEE transactions on components, packaging, and manufacturing technology c Overvoltages v Vacuum circuit-breaker application and switching surge protection (in Circuit interruption), 1984 JOHN F. PERKINS Marcel Dekker, inc. v Types of reignition following high frequency current zero in vacuum interrupters with two types of contact material, 1992 R. P. P. SMEETS, T. FUNAHASHI, E. KANEKO, I. OHSHIMA IEEE 15th ISDEIV-Darmstadt v Mathematical modelling of the high frequency behavior of vacuum interrupters and comparison with measured transients in power systems, 1996 J. HELMER, M. LINDMAYER IEEE 17th ISDEIV-Berkeley

Current transformers: how to specify them
P. Fonti
no. 194
Paola FONTI INPG engineer (Institut National Polytechnique de Grenoble). Graduate in 1970. Joined Merlin Gerin in 1981 as consultant and head of the Medium Voltage Export engineering and design department. She is currently responsible for the MV project completion and tender support group for Schneider Electric.
ECT 194 first issue, February 2000

Lexicon
If: maximum through current crossing a protected area. Is: current threshold setting.
kn: nominal accuracy limit factor (ALF) of a CT (associated with its accuracy load). kr: real ALF of a CT associated with its real load. Pi: (=Rct In2). Internal losses of the CT at In. Pn: (=Rn In2). Accuracy power of the CT. Pr: (=Rr In2). Real load consumption of the CT at In. RL: wiring resistance.
Rp: protection relay resistance. ALF: accuracy limit factor. CT: current transformer. Overrating of a CT: selection of a CT whose primary In is greater than the In immediately greater than the load In. Matching, auxiliary or interposing CT: low voltage CTs installed at the secondary of the main CTs for correcting a ratio and/or the current phase shift. SF: security factor.

Electrical power management requires implementation of data processing units able to monitor networks or equipment and, as applicable, to initiate the appropriate actions... Data sent by current transformers are processed by protection, control and monitoring units that send signals to operate switchgear and/or information to a supervisory unit or to a central control room. The task, consisting of identifying and rating the current transformers and associating them with the protection and/or metering units, has always given rise to problems, both for electrical engineers (oversizing the characteristics) and for the manufacturer (random feasibility, excessive sizing, high costs). This document does not cover the technical demonstrations amply referred to in the literature (see Cahiers Techniques no. 164 and 170). Its purpose is to remind users of a few simple rules enabling the best possible definition of the secondary characteristics of a current transformer (CT) according to the protections and applications concerned. Practically, it offers constructive assistance for technicians who have reached a dead end: c either because they do not possess the necessary information, c or because the results of their design have led to current transformers that cannot be manufactured by the potential suppliers.
Contents
1 Introduction 2 Network disturbances and protections 3 Current transformers 2.1 Disturbances 2.2 Protections 3.1 Reminder of ferromagnetic transformers 3.2 Non-magnetic transformers 3.3 CT manufacturing and implementation 4 Choosing CTs according to protections and applications 4.1 Choosing CT ALF according to protections 4.2 Characterising CTs according to applications 4.3 Special case of differential protection 4.4 Distance protections 5 CT specification examples 5.1 Motor feeder protections 5.2 Transformer feeder protections 5.3 Transformers differential protection 5.4 Differential protection for busbars (87B) 6 Conclusion Bibliography p. 4 p. 7 p. 7 p. 9 p. 13 p. 13 p. 15 p. 17 p. 21 p. 26 p. 27 p. 27 p. 28 p. 28 p. 31 p. 32
1 Introduction
The design of MV and HV electrical networks is a complex undertaking that must take into account the needs to be satisfied, i.e.: c safety of people and equipment, c continuity of supply, c installation and operating costs. The designer uses the load points, their simultaneity coefficient and the above-mentioned
criteria to draw up the single-line diagram for the installation (see fig. 1 ). He must then select the earthing system, define the busways, calculate the fault currents, define the protection system (discrimination, choice of protections, see Cahier Technique no. 174). Figure 2 shows an example of protections chosen for the standby supply in figure 1.
63 kV Stanby supply
G
10 MVA 2 MVA 10 MVA
20 kV
Bank 5 kV Processing Mills
Crushing
LV Shaft no.1, level 300
LV
5 kV
Shaft no. 2, level 500
Fig. 1 : example of a general single-wire diagram for a mine.
Permanent insulation monitor
49
Thermal image Negative sequence Voltage restrained O/C Reverse active power Field loss (max. reactive power) Undervoltage Overvoltage Over and under frequency Zero sequence voltage
a
87T Differential
46 51V 32P 32Q
27 59 81 59N
Fig. 2 : protections of a MV unit module.
The protection plan must specify the operating or non-operating conditions for all the protections during a fault and during normal operation (transients). It must indicate the protection settings. However, the plan rarely indicates the characteristics of the protection input circuit and other data necessary to specify the current transformers (CT). This is because it is often very hard for the designer to collect all the necessary data. The consequences on industrial start-up can be serious: unsuitability, overrating, non-standard specifications and high costs, addition of matching CTs, last minute CT changes, postponement of delivery, commissioning, production times, etc. More serious still, incorrect definition can lead to malfunctions in the protection channel causing destruction of equipment or, worse still, danger for the operator.
A few examples: c Overestimation of the short-circuit current can lead to feasibility problems, overrating and high CT costs. c On the other hand, underestimation of the short-circuit current can lead to failure to detect the fault, thus destroying the equipment, placing the operator in danger and generating operating downtime. c An output power or accuracy error can result in a malfunction or in failure to trip of the protection devices, thus destroying the equipment, placing the operator in danger and generating operating downtime. c An error in defining the accuracy class of a metering winding will lead to incorrect energy billing and thus a loss of income for the electrical utility or the customer. c Etc.
The aim of this document is to provide assistance with defining current transformers. Before entering the heart of the subject,
a reminder is given of the information necessary to define a CT (see the table in figure 3 ).
Necessary information Insulation level Nominal short-circuit current Duration (1 to 3 seconds) Nominal primary current Number of secondary windings (1 to 3) For each secondary winding: c what type c associated protection or metering and setting c output power (relay and wiring consumption) c accuracy factor v protection v metering c Nominal secondary current (1 or 5 A)
Abbreviations Un
Units kV kA s A
Isc
t
Ip
Poutput
VA
ALF SF
Is
Fig. 3 : information necessary to specify a CT with a single primary.
2 Network disturbances and protections
2.1 Disturbances
A MV or HV electrical network is disturbed: c exceptionally by lightning overvoltages, by temperature rises further to overloads or following violent short-circuits between phases or phase-to-earth, c more frequently, and more naturally, by switching overvoltages (e.g. capacitor energisation) or natural transient conditions (e.g. motor starting or power transformer switching) resulting in high but temporary overcurrents. The consequences of disturbances Major disturbances, such as short-circuit currents, can result in serious damage: c fatigue or deterioration of network components, c danger for people, c loss of supply and production, etc. It is thus necessary to provide the relevant protection devices with the right information to ensure prompt action, as the greater the damage, the longer and more costly the repairs and the heavier the losses. However, transient and normal disturbances are a necessary evil and installations must be designed to withstand them. Furthermore, the current transformer/protection pair(s) must not cause nuisance tripping. Elimination of faults Permanent monitoring of network electrical values by reliable and properly rated current transformers supplying protection relays allows rapid isolation of the faulty area. These relays must ignore transient and normal disturbances but systematically trip when a destructive fault has to be eliminated.
2.2 Protections
Dfinition of protections c Function of protections The protection functions of a network are intended to monitor one or more parameters of the installation, for example: currents, voltage, temperature, frequency, etc. These values are permanently measured and compared with setpoints or thresholds beyond which the situation is defined as abnormal and dangerous. When a fault occurs, the protection device issues a tripping signal. Then, in order to durably isolate the faulty part, it prevents reclosing until the device has been repaired. It can also generate an alarm to inform maintenance personnel and enable them to take the necessary action. c The technologies With the particularly rapid growth of electronics technology, protection relays, originally electromechanical, have become static devices: analogue electronic then digital electronic thanks to microprocessors. These devices perform increasingly sophisticated functions and are more and more often referred to as processing units. Digital technology is becoming widespread for all applications (network components or protected loads). This technology has an ample data processing capacity, which allows control and monitoring linked to protection functions and communication to a supervisory unit or a centralised control system. These units are normally supplied with a standard protection, control and indication programme, thus enabling them to be used without extra studies or programming. Only parameters have to be set on commissioning (e.g. the protection settings). They are designed to fully meet application needs. All you have to do is select the version corresponding to the functions available for each application. Some examples of applications are: v transformers, v generators, v capacitors, v motors, v substations, v etc.
These units incorporate meters such as ammeters, voltmeters, wattmeters, kilowatt hour meters, etc. with fewer wiring and compact space requirements (see fig. 4 ). Furthermore, their reduced power requirements (less demanding than those of electromagnetic relays), requiring less powerful CTs, make them more economic. When combined with protection functions and control logic, they also display alarm and operating messages. c The current transformers The CT characteristics are defined according to the technology selected for the processing unit and the scheduled functions (protection, metering, control and monitoring, indication). c Implementation Choice of protection settings is very tricky. It calls for thorough knowledge of parameter levels
according to whether the disturbance is normal or transient or caused by a fault that must be eliminated. Furthermore, it is common knowledge that a short-circuit at one point of the network may be detected right up to the source. The protection co-ordination study ensures that only the faulty part of the network is deenergised. The protections relating to short-circuits either phase-to-phase or phase-to-earth use one of the following discrimination types, as applicable: v overcurrent, v time, v logic, v differential, v directional, (see Cahiers Techniques no. 174 and 113).
Electromagnetic technique
Multifunctional digital system
Fig. 4 : simplification and savings provided by a multifunctional digital system (protection - automation - metering) compared with the former electromagnetic technique.
3 Current transformers
3.1 Reminders of ferromagnetic transformers

Instrument and protection CTs Current transformers are used to supply information to the protective relays and/or current, power and energy metering instruments. For this purpose they must supply a secondary current proportional to the primary current flowing through them and must be adapted to network characteristics: voltage, frequency and current. They are defined by their ratio, power and accuracy class. Their class (accuracy as a function of CT load and of overcurrent) is chosen according to the application. c A protection CT must saturate sufficiently high to allow a relatively accurate measurement of the fault current by the protection whose operating threshold can be very high. Current transformers are thus expected to have an Accuracy Limit Factor (ALF) that is usually fairly high. Note that the associated relay must be able to withstand high overcurrents. c An instrument CT requires good accuracy around the nominal current value. The metering instruments do not need to withstand currents as high as the protection relays. This is why the instrument CTs, unlike the protection CTs, have the lowest possible Safety Factor (SF) in order to protect these instruments through earlier saturation. c Some CTs have secondary windings dedicated to protection and metering. These instrument and protection CTs are governed by standard IEC 60044-1 (in France NF C 42-502). The matching of CTs with protection relays calls for a thorough knowledge of CTs. The following section gives a few reminders of CTs corresponding to this use. Characterisation of CTs c An example of a protection CT: v rated primary current: 200 A, v rated secondary current: 5 A.
15 VA 5P 10 accuracy limit factor = 10 accuracy class = 5P accuracy power = 15 VA
Its accuracy load: Pn = 15 VA Its accuracy limit factor is ALF = 10 For I = ALF. In, its accuracy is 5% (5P), (see fig. 5 ) To simplify, for the protection CT given in example, the ratio error is less than 5% at 10 In , if the real load consumes 15 VA at In. However these data are not sufficient. Also, it is useful to know the standard values.
CT serial number with year of manufacture
Network voltage characteristics Rated insulation voltage: 17.5 kV Power frequency withstand voltage: 38 kV 1 mn 50Hz Impulse withstand voltage: 95 kV peak CT type Applicable CT standard
transformateur de courant - current transformer
Network current characteristic Ith : 25 kA/1 s Idyn : 62.5 kA peak
n 9191671 17,5/38/95 Ith 25

rapport ratio
type
bornes terminals
kV kA
50 Hz 1s
norme standard
Idyn VA
RCF 2 / B CEI - 185 62,5 kA ext. classe

class
Safety factor (SF)
FS ou FLP
150/5 150/5
1S1 - 1S2 2S1 - 2S2
15 15
0,5 5P
7 10
2 221 625
Accuracy limit factor (ALF)
Ratio 1 primary circuit 1 secondary circuit 1S1 - 1S2 1 secondary circuit 2S1 - 2S2 Accuracy power Accuracy class
Fig. 5 : example of the nameplate of a current transformer with two secondaries.
c A few definitions v Rated (nominal) primary current I1 Defined by standards, it is chosen from the discrete values: 10 - 12.5 - 15 - 20 - 25 - 30 40 - 50 - 60 - 75 A and their decimal multiples. v Rated (nominal) secondary current I2 Equals 1 or 5 A. v Ratio (I1 / I2) The primary and secondary currents are standard, thus these values are discrete. v Accuracy load Load value on which the accuracy conditions are based. v Rated (nominal) accuracy power Pn Expressed in VA, it is the apparent power supplied to the secondary circuit for the nominal (rated) secondary current and the accuracy load. The standard values are: 1 - 2.5 - 5 - 10 - 15 - 30 VA. v Real power Pr In this Cahier Technique, it is the power corresponding to the real load consumption of the CT at In. v Accuracy class This class defines the error limits guaranteed on the ratio and on the phase shift in specified power and current conditions. For the nominal 5P and 10P classes, the table in figure 6 defines these limits. v Special accuracy class Class X is a class defined by British standard BS 3938. It must also be defined in the future standard IEC 60044-1 under the name of class PX. This class specifies the minimum value of the knee point voltage Vk of the CT. It also imposes a maximum value of Rct (CT secondary winding resistance). Sometimes, it specifies the maximum value of the magnetising current Io at knee point voltage. If we consider the magnetising curve V(Io) of the CT, the knee point voltage Vk is defined as the point on this curve from which a 10% increase in
voltage causes a 50% increase in the magnetising current Io. Class X corresponds to a better metering accuracy than classes 5P and even more so 10P (see fig. 7 ). It is always possible to find an equivalence between a CT defined in class X and a 5P CT or in some cases even a 10P CT (refer to Cahier Technique no. 195 which deals with equivalences). v Real accuracy factor (Fp or Kr) This is the ratio between the overcurrent corresponding to the nominal error and the rated current of the CT when the real load is different from the nominal load. v Accuracy limit factor (ALF or Kn) This is the ratio between the nominal overcurrent (e.g. 10 In) and the rated current (In). v Short time withstand current Expressed in kA, this is the maximum current Ith that can be withstood for one second (when the secondary is short-circuited). It represents the
VS
10P VS2 5P X VS1 Vk
IS
Io
Fig. 7 : voltages corresponding to different CT classes.
Accuracy class 5P 10P
Current error for the nominal current as a % 1 3
Phase shift for the nominal current Minutes Centiradians 60 1.8
Composite error for the accuracy limit current as a% 5 10
Fig. 6 : errors on the module and the phase at nominal current according to standard IEC 60044-1.
thermal withstand of the CT to overcurrents (the standard values are given by the standards mentioned in the appendix). v CT rated voltage This is the rated voltage to which the CT primary is subjected. It is important to remember that the primary is at HV potential and that one of the terminals of the secondary (which must never be opened) is normally earthed. Just as for any devices, a maximum withstand voltage for one minute at power frequency and a maximum impulse voltage withstand are also defined. Their values are defined by the standards. For example: for a rated voltage of 24 kV, the CT must withstand 50 kV for 1 minute at 50 Hz and 125 kV at the impulse voltage. c CT with several secondaries Some CTs may have several secondaries dedicated to protection or to metering. The most typical cases are CTs with 2 secondaries, more rarely with 3 secondaries. Physically, these CTs group in the same mould the equivalent of 2 or 3 separate CTs that can have different classes and ratios (see fig. 8 ). Influence of the load on the accuracy limit factor Remember that the equivalent simplified diagram of the magnetic current transformer is shown in figure 9 .
I1
IS Io
V
Rct
I2
Fig. 9 : CT equivalent diagram.
Applied to this diagram, Ohms law lets us write: V = I2 (Rct + R), where: Rct : CT secondary winding resistance R: load resistance including wiring, v if I2 = kn In and R = Rn = Pn / In2 , Vn = kn In (Rct + Rn) (1) ( kn = nominal ALF) v if I2 = kn In and R = Rp = Pr / In2 , Vr = kn In (Rct + Rp) On figure 10 we can see that although Rp is far smaller than Rn, the CT saturation knee point is far from being reached at the declared accuracy limit factor kn. The real accuracy limit factor corresponding to the real load (protection + wiring) can be calculated. This is ALFr = kr for which the saturation knee point voltage Vn is reached: Vn = kr In (Rct + Rp) (2) If Rp is less than Rn, kr is greater than kn (ALFr > ALF)
V (volts)
I1
Vn Operating point of the CT when it is loaded at Pn and supplied by a current kn In Operating point of the CT when it is loaded at Pr and supplied by the same current kn In
Vr
S1
S2
S3
Fig. 8 : manufacturing principle of a CT with 3 secondaries (with 3 windings in the same mould).
Ior
Ion
Io
Fig. 10 : operating points of the CT according to its load.
By combining the equations (1) and (2), we find the following formula:
kr = kn P +P Rct + Rn , or k r = k n i n Rct + Rp Pi + Pr
where: Pi = Rct In2 = internal losses of the CT at In Pn = Rn In2 = CT accuracy power Pr = Rp In2 = real load consumption of the CT at In. It is obvious that proper operation of a protection relay is linked to the behaviour of the associated CT and to its real load and not to the behaviour of the CT associated with a theoretical nominal load.
Real needs enable us to determine the minimum accuracy power to be chosen. Using a CT with a load Pr < Pn increases the ALF. Likewise, the ALF increases more if the Rct (internal losses Pi) is low (see fig. 11 ). Calculation of the real ALF (kr) of a CT, associated with its real load, ensures that the right CT is chosen in all traditional cases. Note: for very demanding protections (e.g. differential protections), current transformers are most often defined in class X. This class is always defined according to the real load of the CT and to its own internal losses.
kr
80
Pi = 2 VA
70
60
50
Pi = 5 VA
40
30
20
10
0 0
5 10 15 20 Pr
Fig. 11 : behaviour of the accuracy limit factor kr = f(Pr) of two CTs of 10 VA-5P20 with different internal losses (Rct) according to the real load connected to the secondary.
3.2 Non-magnetic transformers

The output signal, delivered by the non-magnetic transformers (also known as ROGOWSKI coils) is a voltage proportional to the derivative of the primary current.
d ) dt They do not saturate and their response is linear. Consequently, they can be used over wide current ranges: the only limitation is the dynamics and the linearity of the input circuit of the associated protection. The technology of the protection, control and monitoring units connected to these nonmagnetic transformers is of the digital microprocessor type. This technology is able to process signals of very low amplitude. For a given non-magnetic transformer, in view of the linearity of the output signal, the nominal primary current is replaced by a wide range, for example 30 to 300 A. In addition to the advantage of linearity, the use of non-magnetic CTs reduces: c risks of error when choosing primary current at the design stage of the installation, c the number of models to be managed. It also minimises the delivery times. Today these transformers are seldom used. A standard (IEC 60044-8) should define them. Schneider Electric has been using these transformers (see fig. 12 ) in association with the Sepam protection, control, monitoring and metering units since 1986.
(Lenz law: e = n To specify them, all you have to do is indicate:
1 5 2 3 6 7
1 - Primary winding 2 - Dielectric screen 3 - Dielectric insulation 4 - Setting resistance
5 - Secondary winding 6 - Secondary winding support 7 - Magnetic shielding
Fig. 12 : cross-section of a non-magnetic transformer used in MV.
c the CT insulation level, defined just as for a traditional CT, c the rated thermal short-circuit current (Ith) and the dynamic current (Idyn) set according to the same rules as for the CTs, c the utilisation range (rated primary current and the thermal current).
3.3 CT manufacturing and implementation

CTs are industrial products designed according to standards. They are mass produced, thereby reducing costs and guaranteeing their characteristics. Their live part is duplicate moulded in order to comply with insulation, temperature rise and electrodynamic withstand requirements. The number of moulds corresponding to a standard range is necessarily limited. Furthermore, the switchgear and CTs are most often installed in panels that have been optimised, standardised and subjected to qualification tests. In this case, use of standard moulds is compulsory as CTs perform other functions such as the bushings between the cable and the circuitbreaker compartments (see fig. 13 overleaf). Consequently, any modification in volume or shape of a CT results in major study, production and test investments.
To solve special cases, without overcosts or additional technical risks, it is therefore always advisable to look for solutions in order to fit the mould of standard CTs. Panel structure is thus maintained intact. These solutions are:
c good rating of electrical characteristics: for example avoid overrating in power and ALF, c use of CTs with two or three windings, c use of relays performing several protection functions with the same current information.
Fig. 13 : cross-sectional view of the panel and multifunctional CT.
4 Choosing CTs according to protections and applications
Thorough knowledge of CTs, their possibilities and their limits is useful only when they are associated with a specific protection relay whose characteristics and scope of action regarding the monitored current range are known. The protection relays installed on an electrical network are defined in the protection plan. This plan specifies the position and setting of the selected protections. It also defines the position of the CTs, their ratio and, more rarely, their power, accuracy and ALF. In point of fact, complete specification of CTs also requires knowledge of: c the protection input impedance,
c the wiring impedance, c the protection operating thresholds (normally taken into account in the protection co-ordination study). Today, most protections are of the digital technology kind and are highly accurate. CT accuracy is thus a decisive factor. The type of protection also affects the required CT accuracy: c an overcurrent protection only takes the current value into account, c a differential protection compares two currents, c an earth fault protection treats the sum of the three phase currents.
4.1 Choosing CT ALF according to protections

When choosing a CT out of the standard CTs, a reminder is necessary of the relationship linking the nominal ALF (linked to Rn) and the real ALF linked to the real load Rp: kn = kr Rct + Rp Rct + Rn R + Rn or kr = k n ct Rct + Rp
V
Operating point at kr In 2Vs Operating point at 2 Is max
A CT can supply several different protections either separate or grouped in a multi-protection system (e.g. the Sepam). This leads us to examine the protections determining CT sizing. Definite time overcurrent protection The threshold Is (protection setting) can be set, for example, from 2 to 10 In of the CT if the CT In is the application In. To ensure that the CT will not affect the operating accuracy of the protection, it is necessary to have no saturation up to the setting point. But it is usual to take a safety coefficient of 2 (see fig. 14 ). Thus the ALFr (kr) at real load will be: kr u 2
Vs
Operating point at Is max
Im
Fig. 14 : operating points of the CT at maximum threshold.
I s of O/C 160 = 8 = 6.4 In of CT 200

the minimum recommended ALFr (kr) is thus: kr u 2 x 6.4 = 12.8. If the CT load and its internal resistance are known with, for example: Rct + Rp =
1 (Rct + Rn) 2
Is if Is = 10 In kr u 20 In
Example: 200/5 CT - 10 VA-5P10, Load In: 160 A Is = 8 In of load Let us verify if the proposed CT is suitable:
we obtain: kr = kn x 2 = 20 higher than the minimum value required. Thus the CT is suitable.
Inverse time overcurrent protection If the aim is correct accuracy over the entire relay inverse curve, it is necessary to know at what point it becomes a definite time curve. For most relays it occurs at the minimum value between 20 Is and 24 In (CT). So, assuming Iscmax is the maximum short-circuit current, the same reasoning as above, using a safety coefficient of 1.5, gives: krmin = minimum value between:
30
Is I max , 36, and 1.5 sc In (CT) In (CT)
Ih >
Fig. 15 : the vector sum of the phase currents gives the earth fault current.
Directional current protection The rules, unless otherwise specified (refer to Cahier Technique no. 181) are the same as for the overcurrent protections. Note that for the three current protections described above: c If several current protections are supplied by the same CT, it is the one with the lowest curve (the shortest time delay) for high currents that determines the sizing. c In difficult cases, the safety coefficient of 2 can be lowered to 1.5. Earth fault protection As the protection is supplied by the vector sum of the secondary currents of 3 CTs connected as per the Nicholson arrangement (see fig. 15 ), it is preferable to use identical CTs produced by the same manufacturer. However, if there is a DC component (energisation of a transformer) or when a high current occurs, this arrangement (parallel-connection of the secondary of the 3 CTs) will deliver a false earth fault current which may cause nuisance tripping of the protection. To give an example, with the 5P10 CTs, a protection threshold of 10 % of CT In is a limit below which there is a risk of nuisance tripping for definite time protections. The CT accuracy limit factor is given by the expression: krh > X Ihs .
This corresponds to the fact that the CT (associated with the phase with the earth fault) must be able to develop a voltage Vh = X Ihs (Rct + 2 RL + Rh). Note: c If a CT also supplies an overcurrent relay, Rh must be replaced by Rh + Rp. c If the CTs are initially designed for overcurrent protection, we recommend you check that they are suitable for supplying earth fault protection too. Thus, the krh of a 100/1 CT - 10 VA-5P10 is given by the expression:
k rh =
2 Rct + Pn /In kn . Rct + 2 RL + Rp + Rh
Bearing in mind that the impedance of the relay used depends on the setting of Ih (in this case 0.1 A), the digital application gives: Rh = 1VA
(0.1A)2
= 100 ,
3 +10 k rh = 10 = 1.2 3 +1+ 4 +100
In
a value to be compared with the expression of the needed krh:

k rh = 6 0.1 = 0.6 , the CT is suitable. 1
The safety coefficient (X) is normally equal to 6 (given by the relay manufacturers).
If the short-circuit current is very high and if the protection time delay is short, to avoid nuisance tripping it may be necessary to add a stabilising resistance in series with the earth fault relay. Its purpose will be examined in paragraph 4.3. In order to avoid difficulties that may arise with the 3 CT arrangement, it is preferable, whenever possible, to use a toroidal CT around the 3 phases (see fig. 16 ). Note that the 3 phases must be placed in the centre of the toroid to avoid local saturation of the magnetic material. Use of a toroidal CT allows very low operating thresholds (a few amps) to be chosen. Differential protections Increasingly used for the protection of transformers, rotating machines and busbars, they have the advantage of being fast and independent from the other protections in terms of discrimination. These relays often operate during the transient fault current period. Just as for the earth fault protections, an aperiodic DC component may result in transient saturation of the CTs and thus generate a false differential current. Since differential protections are tricky to implement, manufacturers normally supply the necessary information to design and install the CTs.
1 2 3 2
Fig. 16 : the toroid placed in (1) or (2) gives the same information but the toroid placed in (1) also monitors the faults located upstream of the toroid (2).
Conclusions c It is the high setting of the overcurrent protections that determines the minimum kr (real ALF) to be complied with. c It is the stability of the relay up against transient phenomena that determines the ALF or class X for earth fault and differential protections.
4.2 Characterising CTs according to applications

In this context the term applications means the components of the electrical network to be protected: line incomers and feeders, transformers, busbars, generators, motors, capacitor banks, etc. Each component requires the use of several protections. An example is given by the diagram in figure 17 (overleaf) in which the protections are identified by their ANSI code (American National Standard Institute). However, to optimise choice of a CT, you need to know which protections it supplies, as well as their settings, real impedance, short-circuit current, etc. The difficulty lies in the fact that these values are rarely known by the CT specifier when his choice has to be finalised. For this reason this section proposes to study the various applications where values can be defined by excess. They can therefore be chosen without risk as they will necessarily cover needs. However, they will lead to a less restrictive definition which is more realistic than the one proposed by relay suppliers in their catalogues. In point of fact, each supplier gives some guidelines for choice covering all the setting ranges of his relay associated with CTs assumed to have high internal losses and also maximum wiring impedances. For the various applications, we shall list the protection functions commonly used. The most unfavourable protection will be chosen to size the current transformers. Note: Initially, the earth fault and differential protections will not be taken into account.
150/1 50 51 51G 150/6 kV 22.5/30 MVA Zsc = 11 % (30 MVA) YD 11 87T 27I 27 51G 27 27R 3000/1 3000/5 51 50t 27 87
300/1
300/5
40 46
81
49 51 51V
300/1
66 66 100/5 600/5 46 46 49 51N 49 50 51N 6 kV/400/230 V 2500 kVA Zsc = 7.5 % DY11 51N 300/5 50 51 200/5 51 100/5 50t 100/5
51N
400 kVA Zsc = 4 %
51G M 710 kW M 4500 kW In = 491 A Id = 5.8 In Td = 5 sec 51 50t
2000 kVA In = 192.45 A
In = 82.3 A Id / In = 6
Td = 5.2 sec
315 A
49 132 kW In = 250 A Id / In = 5 Td = 2 sec
Fig. 17 : example of a protection plan.
Applications with typical protections Examination of the protections normally implemented for the various applications (see fig. 18 ), shows that it is always the high threshold overcurrent protection which is the CT sizing factor. The exception to this rule is, however, the motors protected by fuses for which the locked rotor protection is the most restrictive. To optimise the CTs, the settings of these protections need to be taken into account. c Line incomers and feeders The high threshold overcurrent protection is normally set between 3 i Is i 10 In, hence a krm u 2 I s with at maximum a krm = 20. In If an inverse time protection is used: krmin = minimum value between:
c Generator incomer There is no reason to select a setting > 7 In, given the relatively low short-circuit current of a generator, hence a krm u 14. c Transformer incomer If we take the example of the protections downstream of a HV/MV transformer, the setting threshold must be lower, for example, than 70 % of the Isc at the secondary side, i.e. as an initial approximation:
Isc = 0.7 In2
100 . Zsc
If we apply the rule: krm u 2 I s ,
In
it becomes: krm u 1.4
100 . Zsc
I max 30 , 36, and 1.5 sc In (CT) In (CT)
Is
This krm is maximised; in reality it is the discrimination study of the downstream network that sets the Is and thus determines the krm.
ANSI Typical protections code
Applications Line Generator incomer incomer and feeder Transformer incomer Transformer feeder Capacitor Motor feeder with feeder circuitcontactor breaker + fuses c c c c (c) c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c
37 46 49 50 50N 51N
Undercurrent Unbalance Thermal image Instantaneous high threshold overcurrent Instantaneous earth fault Time delayed earth fault c
51N1 Star-star capacitor unbalance 51LR Locked rotor / too long start 51V 51-1 51-2 66 67 67N Voltage restrained overcurrent Time delayed low threshold overcurrent Time delayed high threshold overcurrent Number of starts Directional overcurrent Directional earth fault
Fig. 18 : protections implemented by application, in the most common cases.
c Transformer feeder The operating current of the high threshold overcurrent protection (Is), placed upstream of the transformer, must be higher than the current observed at the primary (Isct) on a short-circuit at the transformer secondary terminals. As a first approximation we can state that:
Transformer rating (MVA) 0.5 0.63 0.8 1 2.5 5 10 20 30 40 80 160
Zsc (%) 4 4 4 5 5 6 8 10 12 13 18 20
Isct i In1 100

where
Zsc Pn = nominal primary current, 3 Un1
In1 =
Pn = nominal power of the power transformer, Un1 = primary voltage. In fact, the upstream source impedance also helps to limit Isct. We can thus be certain that: Is i In1 100 Zsc If the general rule krm u 2 I s applies,
Fig. 19 : typical short-circuit impedances for power transformers.
In
we obtain: krm u 2
In1 100 . In Zsc
protection is still possible. Otherwise you must plan this protection downstream of the transformer. c Capacitor feeder The time delayed high threshold recommended by capacitor suppliers is 3 In (0.3 s). If we take Ismax = 5 In ; krm u 10. c Motor feeder The high threshold must be set above the starting current, which is always i 7 In. If we take Ismax = 8 In ; krm u 16. If the short-circuit protection is provided by fuses, then the locked rotor / too long start protection is the most restrictive. Its setting is between 2.5 and 4 In ; krm u 8. Note: From the above, we have assumed that CT In = application In. If this is not the case, the resulting krm must be multiplied by the ratio:
The typical values for short-circuit impedances of power transformers can range from 4 % for small transformers to 20 % for the largest ones (see fig. 19 ). This would result in requiring minimum krm
I ranging from 10 n1 for large transformers to In In1 50 for the smallest ones. In
These values may be too difficult to obtain for low rated CTs particularly when their thermal withstand is high (e.g. 40 kA, 1 s). Therefore, in difficult cases, the problem of feasibility could be solved by overrating the CT primary or by using a reduced coefficient (1.5 instead of 2) as described in paragraph 4.1, which gives: krm u 1.5 In1 100 . In Zsc If you choose to overrate a CT, you must ensure that the setting of the transformer thermal
In (application) . In (CT)
The table in figure 20 summarises the minimum ALF to be complied with according to the applications.
kr mini for definite time overcurrent c Is is known, all applications kr = 2 c Is is unknown Transformer feeders kr = 2
kr mini for inverse time overcurrent
Is In
between: 30
Is I max , 36, and 1.5 sc In (CT) In (CT)
In1 100 In Zsc
Not concerned
Transformer incomers, Line incomers and feeders Capacitor feeders Circuit-breaker motor feeders Contactor with fuse motor feeders Generator incomers In this table:
kr = 20 (by excess)
between: 30
Is I max , , 36, and 1.5 sc In (CT) In (CT)
if no definite time on a second threshold kr = 10 (by excess) kr = 16 kr = 8 kr = 14* kr = 14* Not concerned Not concerned
Is is the setting current of the overcurrent threshold for which the response time is
shortest for the high currents,
In is the CT nominal primary current, In1 is the power transformer nominal current,
(*) = general cases.
Fig. 20 : real ALF (kr) required for overcurrent according to the applications.
4.3 Special case of differential protection

Although the manufacturers of relays dedicated to differential protection impose the necessary CT secondary characteristics required for proper operation, it is useful, for understanding and avoiding errors, to possess minimum knowledge of this type of protection. A differential protection monitors an area limited by CTs which measure incoming and outgoing currents. If the outgoing currents are not consistent with the incoming currents, this is normally because a fault has occurred in the protected area. We shall examine in turn high impedance, pilot wire, percentage differential and low impedance differential protections with their respective CT requirements. According to the type of protection and its use, relay manufacturers have had to apply a variety of principles varying in complexity in order to guarantee the stability of their relays against transient phenomena likely to cause nuisance tripping of this protection. High impedance differential protection This type of protection is normally used for protection of motors, generators, busbars as well as for restricted earth fault protection of transformers. c General This type of protection is used to protect an area with the same voltage level.
In healthy conditions, the input current ie is identical to the output current is and thus the differential current id = 0 (see fig. 21 ) A high fault current can flow through the monitored area and cause CT saturation, hence the risk of nuisance tripping of the protection (non stability). The stability of the relay is obtained by connecting it in series with a stabilising resistance Rst. This resistance is calculated in such a way that the derived current in the differential circuit (Rst + Rp) cannot reach the relay setting threshold when the maximum through current saturates a CT with its DC component. This results in: (1) (Rst + Rp) u (Rct + 2 RL)
Ie
A Application (monitored area) i'e i 's id
Is
B
Rst Protection (Rp)
Fig. 21 : high impedance differential principle.
I ssc Ir
where Issc = maximum through current observed at the CT secondary, Ir = relay secondary setting current. Rst may vary between a few ohms and a few hundred ohms (exceptionally it may be greater than 1000 ohms). For the relay to operate properly at Ir if a fault occurs in the area, the knee point voltage Vk must be greater than: 2 (Rst + Rp + Rct + 2 RL) Ir . As a rule Rct + 2 RL are negligible compared with Rst + Rp thus: (2) Vk u 2 Ir (Rst + Rp). By combining the equations (1) and (2) we find: (3) Vk u 2 Issc (Rct + 2 RL). These equations show that Rst (expression 1) and Vk have greater values when Rct is high. A high value stabilising resistance generates high overvoltages at the CT secondary. Therefore when overvoltages in excess of 3000 V are anticipated, a protection by a nonlinear resistance (ZnO) is added. The result of these observations is that the CTs are optimised if Rct and Vk are as low as possible and if the through current (seen from the CT secondary, i.e. Iscc) is defined without excess. Whatever the application where the high impedance differential is used, all the CTs must have: v the same ratio, v the same magnetising curve (same minimum Vk), v the same maximum Rct, and comply with expression (3) ; for Vk, since the relevant CTs are not at the same distance from the relay, take the maximum RL. For this protection, a maximum value of the magnetising current Io must also be defined at Vk /2 according to the required sensitivity.
For the relay to detect a current Ir, a voltage Vs = Vk /2 must be developed at the terminals of the parallel-connected CTs; for this purpose, the minimum primary current Irms really detected by the relay will be Irms = n(Ir + Io), where n = CT ratio and = number of parallel-connected CTs (there may be many of them on a busbar protection!). c Application to the motor differential protection The maximum through current for which the motor must remain insensitive is in this case the motor starting current:
Issc = Ist (seen at the secondary). If you do not know this current Ist, you know that: Ist < 7 In motor
c Application to the generator differential protection The maximum through current is in this case the short-circuit current supplied by this generator only. If you know the generator subtransient reactance X %, the following will be taken:
100 ; X If this value is not known, X % = 15 will be taken Note: the peak voltage at the CT secondary must be calculated using: Issc maxi = Igenerator + Isc network.
I ssc = In
c Application to the busbar differential protection (see fig. 22 ). In this case, the through current is equal to switchboard Isc : Issc = switchboard Isc seen from the CT secondary. c Application to the restricted earth fault differential protection of transformers (REF) v In the case of figure 23a , this protection detects the insulation faults at the transformer secondary windings and up to the CTs located downstream.
Rst 87B 1 2 3
v In the case of figure 23b , this protection detects the insulation faults at the transformer primary and advantageously improves the typical earth fault protection which is sensitive to transformer violent inrush currents and to the through currents resulting from a downstream asymmetrical short-circuit. In this case also, we shall calculate Rst and Vk based on the maximum through current in the CTs for a fault outside the protected area. As a first approximation, we can say that this current is less than the current limited by the transformer impedance, i.e. I through = where Psct = Pn
Fig. 22 : high impedance busbar differential protection.

a
Psct , Un 3
100 , (transformer short-circuit Zsc
power); if we know the upstream short-circuit power (Pu), a more accurate value can be calculated replacing Psct with
Rst
(Psct + Pu )
(Psct Pu )
The through fault current must then be converted into Issc seen at the CT secondary.
Rp
b Rp
Rst
Line or cable differential protection with pilot wires (see fig. 24 ). A relay of this type is installed at each end of the cable or the line. On the pilot wires, each relay reproduces a voltage which is the image of the sum: a I1 + b I2 + c I3 + d Ih . If the two voltages are different, both relays trip. NB: Note that the coefficients a, b, c and d are different to ensure that all fault types correspond to a sum other than zero. Consequently, the operating threshold of a two-phase fault or of a phase-to-earth fault is slightly different according to the faulty phase.
Fig. 23 : restricted earth fault protection of the secondary [a] or primary [b] windings of a transformer.
In this case also, class X CTs are required, and each manufacturer gives an empirical formula for the minimum knee point voltage Vk.
RL Pilot wires Line differential 87L Line differential 87L
RL
Fig. 24 : line or cable differential protection with pilot wires.
An example of the minimum knee point voltage required: Vk mini = 0,5 N kt In (Rct + X RL) where N, kt and X are constants associated with the relay response time, its sensitivity and its type of connection. Another example: Vk mini =
50
In
+ If (Rct + 2 RL)
where: In is the CT nominal secondary current (1 or 5 A), If is the through short-circuit current, seen at the CT secondary. The stability of this relay is achieved both by complying with the required knee point voltage and by a percentage differential operating threshold which will increase with the through current by use of retaining windings. The CTs at the ends of the line must have the same ratio and must comply with the minimum Vk and maximum Io specified by the manufacturer. However, their magnetising curves and Rct do not need to be identical. Percentage biased differential protection for transformers The term percentage differential stems from the fact that the operating threshold increases with the through current. Simple comparison of the currents in each upstream phase with the currents in the same downstream phases is not suitable for transformer differential protection. This is because: v the upstream and downstream currents of a power transformer do not have the same amplitude or the same phase angle, v when the transformer is energised, its magnetising current is only seen upstream, v the presence of an earthing generator in the protected area (e.g. earthing the transformer neutral) can trip the protection, while the fault is, for example, located on a downstream feeder. c Precautions to be taken to solve these problems: The aim is to ensure that the relay sees upstream and downstream currents of the same amplitude and in phase during normal operating conditions. This can be achieved by intelligent choice of CT ratio and connections.
Matching CTs are used for this purpose and often help to eliminate the zero-sequence current that could trip the protection when an earth fault occurs outside the protected area. However, most of the new digital relays are able to perform internally, by parameter setting, the adjustments necessary to reset the currents, thus considerably simplifying their implementation. Furthermore, it must be noted that all the transformer differential relays are immunised to the 2nd order harmonic blocking their operation when the transformer is energised. c CT voltage Vk In 99 % of cases, a class X is requested. The minimum knee point voltage is imposed and depends on the resistance of the secondary winding Rct of the CT and of its real load Rr. More complex specifications are sometimes mentioned, which include the X/R ratio of the network or the magnetising current of the power transformer. However, faced with the problems that users have in obtaining all these parameters, relay suppliers sometimes provide simplified empirical formulas which lead to a slight oversizing. Examples of minimum knee point voltage imposed for the Sepam 2000 D02 (Schneider Electric): Vk mini = A Ib (Rct + 2 RL) where: 2 RL = total resistance of the secondary wiring, Rct = CT secondary winding resistance, Ib = power transformer nominal current seen at the CT secondary, A = constant depending on transformer power: Some suppliers take into account the through current, for example: 4 If Rct + 3 RL + Rp 3 on the star side of the power transformer, and Vk u 4 If (Rct + 2 RL + Rp) on the delta side of the power transformer. The through current will be defined in the same way as for the restricted earth fault protection. Note: Use of matching CTs leads to different expressions of the knee point voltage for the main CTs which must take into account the extra load that they represent. Vk u
))
To conclude, the stability of this protection is ensured by: c the threshold which increases with the through current (restraint system), c the right choice of CT knee point voltage Vk , c a system ensuring immunity to 2nd order harmonics generated by inrush currents, c the most sophisticated relays are also immune to 5th order harmonics which occur during power transformer overexcitation (saturation). Low impedance differential protection This protection is used for busbar differential protection. It is very costly and spaceconsuming, as it requires a large number of
modules and matching CTs which need one or more cubicles according to switchboard size (see fig. 25 ). In the case of a double busbar switchboard, the protection must be continually informed on the position of the transfer switches in order to direct the currents of each feeder and incomer to the relay associated with monitoring of the busbar on which this feeder or incomer is connected. The CTs associated with this sophisticated protection may have different ratios. Their secondaries are also defined as class X in most cases. However, as saturation can be tolerated, knee point voltage requirements are less severe than for high impedance differential protection.
B1
C1.2
B2
B3 C3.4
B4
C1.3 D1 D2
C2.4 D3 D4
CMT D1 1
CMT D2
CM C1.3
CM C1.2/C3.4
CM C2.4
CMT D3
CMT D4
B1 B3 Global
B2 B4
87B B1
87B B3
87B Global
87B B2
87B B4
CMT = current metering and transfer (for feeders D1, D2, D3 and D4) CM = current metering (for crossing and in-line couplers C1.2, C1.3, C2.4 and C3.4) 1 = transfer and metering 2 = protection
Fig. 25 : example of a low impedance differential protection for a double busbar.
4.4 Distance protections

These protections, extremely common in high voltage, are increasingly used for very long medium voltage lines, as pilot wires do not need to be installed (see fig. 26 ). The formula generally used for CTs defined as class X is as follows: X Vk u If 1 + Rp + Rct + 2 RL R Besides the usual terms already defined, the following parameters can also be found: In many cases, information regarding the line to be protected (cross-section, length) is nonexistent or impossible to obtain before the switchboard is delivered. However, the calculation example in figure 27 shows the considerable difference between CT characteristics according to line length. Between 1 and 12 km, there is a ratio of 10 between the characteristics. This type of relay is always used for very long lines. It would not be reasonable, without information, to resign ourselves to using for If the value Isc at the head of the line. The example in figure 27 shows that the short-circuit current drops from 26.2 kA to 13.4 kA for a 2 km line only, then to 3.8 kA for a 12 km line. Knowledge, even approximate, of line length is an important factor in optimising CTs.
X : reactance/resistance ratio between the R source and a three-phase short-circuit occurring at the end of the protected area. If is in this case equal to the three-phase shortcircuit current at the end of the monitored area, seen at the secondary side of the upstream CT. Rp : relay resistance.
L 21 RL Monitored area
RL Monitored area 21
Fig. 26 : distance protections: at each end of the line a relay monitors 80 % of the line with instantaneous operation.
Source
Isc
26.2 26.2 26.2 26.2 26.2 26.2
Rs 0.015 0.015 0.015 0.015 0.015 0.015
Xs 0.727 0.727 0.727 0.727 0.727 0.727
Line L Rl (km) 1 2 5 12 24 40 0.12 0.2396 0.599 1.4376 2.8752 4.792
Xl 0.338 0.675 1.688 4.051 8.102 13.5
Calculations X t / Rt Zt 7.925 5.518 3.936 3.29 3.055 2.961 1.073 1.425 2.492 4.994 9.29 15.02
Isc
(kA) 17.75 13.37 7.646 3.815 2.051 1.268
If (A)
29.59 22.28 12.74 6.358 3.418 2.114
Vk (V) (264.1 x Rct) + 108.28 (145.23 x Rct) + 59.54 (62.90 x Rct) + 25.79 (27.28 x Rct) + 11.18 (13.86 x Rct) + 5.68 (8.37 x Rct) + 3.43
U = 33 kV CT primary side: 600 CT secondary side: 1 Rp = 0.36 2 RL = 0.05
Line cross-section = 150 mm2 Rl = 0.12 /km Xl = 0.388 /km Rs , Xs = source resistance and reactance
Xt = Xs + Xl Rt = Rs + Rl
X Vk u If 1 + t Rp + Rct + 2 RL Rt
Fig. 27 : calculation of CT knee point voltage Vk for distance protection relays, for various line lengths, showing the advantage of using the Isc as a reference at the end of the line to define these CTs.
5 CT specification examples
We shall deal in an incomplete but educational manner with two examples of specifications
concerning conventional protections and two examples concerning differential protections.
5.1 Motor feeder protections

For this application the functions are, for example: v overcurrent, v thermal image, v unbalance. With electromagnetic relays, serial-connected in the CT secondary, the minimum specification often encountered is 20 VA-5P30. With multifunction digital relays, the specification is often 5 VA-5P20 it is superabundant. The minimum ALF is 2
8 In (motor) , In (CT)
But taking account of a motor In (200 A) for a 300/1A CT: 16 x(300/200) = 12. The relay consumption is for example 0.025 VA (Sepam 2000) and 0.05 VA for wiring (6 m in 2.52); the 5 VA-5P20 CT has internal losses of 2 VA. Let us calculate kr :
2 + 5 = 67.5 2 + 0.075 a value far greater than 12! A 2.5 VA-5P10 CT (where Pi = 1.5 VA) is more than sufficient. Its kr is: k r = 20 k r = 10 1.5 + 2.5 = 25 1.5 + 0.075
i.e. kr u16 if In (motor) = In (CT).
5.2 Transformer feeder protections

This is the high threshold overcurrent protection which sizes the CT (refer to paragraph 4.2): Taking into account that the thermal withstand of the requested CT is 50 kA-1s the CT cannot be manufactured. In actual fact the problems start as soon as Ith / In > 500, and in this case 50000/30 = 1666! Faced with this kind of problem, the CT primary can be overrated. In view of their characteristics, figure 28 shows the overratings of CTs able to match the required ALF and satisfy CT feasibility.
In1 100 In Zsc where In1 = Inominal of the transformer primary and In = Inominal of the CT primary.
kr u 1.5 Let us take the example of a 1 MVA transformer; Zsc = 5 %; Uprimary = 22 kV, hence In1 = 26.2 A. This gives, when In = 30 A, a minimum kr of 26.
Transformer U = 22 kV Power (MVA) 0.5 0.63 0.8 1 2.5 5 10 20 30 40 80 Zsc (%) 4 4 4 5 5 6 8 10 12 13 16
CT characteristics
Isc max
(kA) 0.3 0.4 0.5 0.5 1.3 2.2 3.3 5.2 6.6 8.1 13.1
In TFO
(A) 13 17 21 26 66 131 262 525 787 1050 2099
In
(A) 40 40 40 50 100 200 300 600 1000 1500 2500
ALF required (Isc / In x 1.5) 12.3 15.5 19.7 15.7 19.7 16.4 16.4 13.1 9.8 8.1 7.9
Fig. 28 : standard CT for a 22 kV transformer feeder.
In the example: choose In1 = 50 instead of 30 A. The minimum kr moves from 26 to 15.7. The ratio Ith / In becomes 50000/50 = 1000. It is still higher than 500, but such a CT is now feasible.
Schneider Electric, manufacturer of protection CTs and MV panels, offers standard 2.5 VA-5P20 CTs suitable for electronic and digital protection and which consume less than 0.5 VA with a wiring resistance of 2 RL < 0.1 .
5.3 Transformer differential protection

In this example, the relay used is a Sepam 2000 D02 (Schneider Electric). This relay does not need an interposing CT (see fig. 29 ). The minimum knee point voltage Vk required is given by the formula: Vk = A Ib (Rct + 2 RL) where: Ib = power transformer nominal current at the CT secondary, Rct = CT secondary winding resistance, RL = resistance of a conductor linking the CT and the relay, A = constant depending on transformer power: c 30 for 2 MVA < Pn < 14 MVA, c 24 for 15 MVA < Pn < 39 MVA, c 16 for 40 MVA < Pn < 70 MVA. Let us take an example Pn = 50 MVA hence: A = 16, I1 = 600 A U1 = 63 kV I2 = 3000 A U2 = 11 kV I1 / In1 I2 / In2
87T
Fig. 29 : transformer differential protection principle.
In2 = nominal current of the secondary windings

of the CT located at the secondary side of the power transformer. Calculating CTs We assume that the relay is located in the downstream switchboard, resulting in a wiring 2 RL of 1000 m for upstream CTs and 10 m for downstream CTs. c CTs at the primary side of the power transformer If the wiring is 2.5 mm2 (i.e. 8 per km): 2 RL = 8 x 1000/1000 = 8 . This gives: Vk > 16 x 0.764 (Rct + 8) Vk > 12.2 Rct + 98 c CTs at the secondary side of the power transformer If the wiring is 2.5 mm2 (i.e. 8 per km): 2 RL = 8 x 10/1000 = 0.08 . This gives: Vk > 16 x 0.875 (Rct + 0.08) Vk > 14 Rct + 1.12
In1 = 1 A, In2 = 1 A,
Ib1
( (
In1 = 0.764 A, 3 U2 I1 Pn
Ib2 =

3 U2 Pn
I n2 I = 0.875 A, 2
where: I1 = nominal current of the CT at the primary side of the power transformer, I2 = nominal current of the CT at the secondary side of the power transformer, In1 = nominal current of the secondary windings of the CT located at the primary side of the power transformer,
5.4 Differential protection for busbars (87B)

For 87B busbar differential protection (see fig. 30 ), the relay used is a Sepam 100 LD (Schneider Electric). The minimum knee point voltage Vk required for this relay is given by: Vk u 2 If (Rct + 2 RL) where: If = maximum through current at the CT secondary, Rct = CT secondary winding resistance, 2 RL = resistance of the wiring loop between the CT and the relay.
Calculating 2 RL 2 RL = (2 L/S) Loop length: 2 L = 45 m, Wiring cross-section: S = 2,5 mm2, where = 1.8 x 10-8, 2 RL = 0.324 Calculating If
We choose:
Ir = 5 % I2n = 0.05 A
In this case we can consider that Rp = 0 Rst = 2530 Calculating Vpeak Vp = 2 2 Vkr (Vs Vkr ) Vs = (Rs + Rp) Issc where: Issc = maximum fault current seen at the CT secondary: in this case Issc = Ir Vkr = real value of the knee point voltage of the CT (270 V), Vs = 50600 V, we find: Vp = 10426 V. Vp > 3000 V, a surge limiter is necessary. Calculating the fault current Id really detected
I f = I sc
I 2n I1n
I1n I2n Isc I1n I2n Isc If
= CT nominal primary current, = CT nominal secondary current, = short-circuit current at switchboard level, = 1250 A =1A = 25 kA = 20 A
Now Vk can be determined: Vk > 2 x 20 x (Rt + 0.32), i.e. Vk > 40 Rct + 13 After consultation, the proposed CT has an Rct of 6 and a Vkr equal to 270 V. It is suitable as: 40 x 6 + 13 = 252.96 V < 270 V. Calculating Rst
Id = Ir + Io m,
where: Io = magnetising current at Vk / 2 (data given by the CT manufacturer), m = number of CTs per phase used for busbar protection, in this case = 5, Io = 0.006 A, We find: Id = 0.08 A, i.e. 100 A on the primary side. We thus observe that earth faults will also be detected in a satisfactory manner since in this network the earth fault current is limited to 300 A.
Rst =
Vk Rp 2 Ir
Ir = setting current,
Rp = relay resistance, Vk = minimum knee point voltage required.
Rst 87B
Fig. 30 : busbar differential diagram.
This section has given only a few examples of CT specification (ALF or Vk) according to the applications. Their complete and optimised specification
requires active and co-ordinated participation of many contributors. The main information required for each protection is given in the table in figure 31 .
Information to be provided
Typical Differential protections protections High impedance differential protection (see note) Overcurrent BB 51 + 51 N 87B Motor 87M c c c Gen. 87G c c c Earth 87N c c c % Pilot differen- wire tial Transfo. Line 87T 87L c c c c c c
Contributors
Switchboard real Isc Maximum through current Maximum earth fault current Ihl if earth fault detection Transformer rating Transformer Zsc Vector group of the power transformer CT ratio c c
c c c
Network designer
c c
c c c
Protection plan designer
Relay manufacturer c and type Relay setting current Ir Motor starting current Generator subtransient short- circuit reactance Distance between c CT and relay Cross-section of wiring used (or value of RL) c c
c c
c c c
c c
c c
Motor manufacturer c Generator manufacturer
c c
c c
c c
c c
c c
c c
Switchboard designer
Note: When consulting the class X CT suppliers, you MUST ask for all the values of minimum Vk, maximum Rct and maximum Io as these are essential for completing the study. Besides the minimum Vk, the value of real Vk is also necessary to calculate peak voltage, when high impedance differential relays are used.
Fig. 31 : information to be provided by each contributor for defining CTs.
6 Conclusion
Current transformers are essential links between electrical busways and the devices protecting the MV and HV network components. Their precise definition and optimisation are not simple and require thorough understanding of their operation and close co-operation of many contributors. As a rule the choice of a CT is easier when it is associated with conventional protections. However, when differential protections (class X) are chosen, they must be examined with care and require good communication with the CT manufacturer. However, as we have shown in this document, it is possible to specify CTs by excess according to protections and applications. This solution is a means of avoiding many problems relating to safety, costs and lead times.
CT technology must also be taken into account as it can offer advantages, for example: c standard CTs are optimised and available, c CTs with several secondaries offer space and cost savings, c multifunctional CTs, used in standard panels, allow cost savings. If, despite all the precautions taken, problems arise, there is always a solution. This is the purpose of Cahier Technique no. 195 which highlights the traps (most common errors) and the possible solutions. Finally, this Cahier Technique demonstrates the extreme care that installation designers must take when defining CTs, in particular when drawing up the protection plan and the discrimination study.
Bibliography
Standards c IEC 60185: Current transformers Characteristics c IEC 60044-1: Instrument transformers - Part 1: Current transformers (replaces IEC 185). c IEC 60044-8: Instrument transformers - Part 8: Electronic current transformers. c NF C 42-502: Measuring instruments. Current transformers. Characteristics. c BS 3938 (replaced by BS 7626): Specification for current transformers. Schneider Electrics Cahiers Techniques c Protection des machines et des rseaux industriels HT. P. ROCCIA, Cahier Technique no. 113. c Current transformer for HV protection. M. ORLHAC, Cahier Technique no. 164. c Protection of industrial and tertiary MV networks. A. SASTRE, Cahier Technique no. 174. c Directional protection equipment. P. BERTRAND, Cahier Technique no. 181. c Dynamic stability of industrial electrical networks. B. DE METZ NOBLAT and G. JEANJEAN, Cahier Technique no. 185. c Current Transformers: specification errors and solutions. P. FONTI, Cahier Technique no. 195. Various works c Guide de lingnierie lectrique ELECTRA - 07.86 c Protection des rseaux lectriques Ch. PREVE - Ed. Hermes - 06.98

Integration of local power generation in industrial sites and commercial buildings
T. Hazel
no. 196
Terence HAZEL Terry Hazel received his BSc in Electrical Engineering from the University of Manitoba Canada in 1970. He then worked in Perth Australia for a year as power coordination engineer, and in Frankfurt Germany as a consulting engineer until he joined Merlin Gerin in 1980. For 15 years was the technical team leader for several major international projects involving process control and power distribution. He has since been with the tendering section of the industrial projects department and often meets with clients during the front end engineering stage to discuss and compare the various possible electrical distribution systems. He is an active member of IEEE and has presented papers dealing with electrical power distribution at Industry Applications Society conferences.
ECT 196 (e) first issue, January 2000
Lexicon
Black start: The capability of starting generator sets without the presence of a utility supply. Damage curve: A current-versus-time curve showing the allowable limit without permanent damage to equipment. Equipment commissioning: Performance of the testing and adjustment at site leading up to and including the energization of a piece of equipment. An example would be the operation of one generator set. Frequency droop: The absolute change in frequency between steady state no load and steady state full load, typically 4%. An increase in power output results in a decrease of frequency for generator sets operating alone in this mode. Isochronous speed governing: Governing with steady-state speed regulation of essentially zero magnitude. Load sharing: Centralized elaboration and sending of set points for generator set loading. This ensures that all sets will share the load in an equal manner proportional to their power rating. Load shedding: Voluntary disconnection of low priority loads when the available power is insufficient to supply the total plant load. Residual voltage: The voltage on a busbar after disconnection from the supply. This voltage is generated by rotating machines which remain connected to the busbar. Spinning reserve: The difference between the total available capacity of all generating sets already coupled to the system and their actual loading. Static switch: A fast acting switch normally consisting of a power electronics device which will transfer the load from the power conversion module of a UPS to another supply without delay or unacceptable transients. Synchronism-check relay: A verification relay whose function is to operate when two input voltage phasors are within predetermined limits. Synchroscope: An instrument embodying a
continuously rotatable element whose position is a measure of the instantaneous phase difference between the voltages across a circuit-breaker. System commissioning: Performance of additional testing and adjustments at site of equipment which have been commissioned to ensure correct operation of the system comprised of the equipment. An example would be the parallel operation of several generating sets including synchronizing, and load shedding features. System stability: A system is considered stable if bounded input disturbances result in bounded output disturbances. For an electrical distribution system, changes in load, faults, switching operation, etc. will not cause wide fluctuations in voltage or frequency if it is stable. Unit substations: A substation containing the electrical distribution equipment necessary for supplying the loads of a particular plant production unit. It typically contains medium voltage switchgear, power and distribution transformers, low voltage switchgear and MCC. Voltage restrained overcurrent relay: An overcurrent protection relay having a voltage input which opposes the typical response of the relay to the current inputs. This is used for generators since they deliver much lower shortcircuit currents than utility connections having the same capacity. Voltage waveform distortion: The difference between the actual voltage waveform and a pure sinusoidal waveform, often expressed as total harmonic distortion,
THD = 2 Uh U1
where Uh is the harmonic voltage and U1 is the fundamental of the voltage waveform. X/R ratio: The ratio of the electrical distribution system inductance to the resistance. This ratio determines the time constant of the d.c. component of the short-circuit current which is an important factor in defining the rating of highvoltage circuit-breakers.

Engine driven alternating current generator sets are often used in remote industrial sites as a prime source of electrical energy. They are also extensively used in both industry and commercial buildings as a source of back-up power. This cahier technique discusses most of the subjects which have to be handled when implementing engine driven alternating current generator sets having rated powers up to 20 MW.
Contents
1 Types of engine generator sets 2 Rated power for generator set applications 3 Typical applications 4 Operation of generator sets 3.1 Stand-by generator sets 3.2 Production generator sets 4.1 Starting and stopping of generator sets 4.2 Stand alone operation 4.3 Parallel operation with utility supply 4.4 Parallel operation with other generator sets 5 Transfer schemes and synchronization 5.1 Automatic transfer on loss of supply 5.2 Maintenance transfer 5.3 Synchronization of generator circuit-breaker 5.4 Synchronization of bus-tie, bus coupler, or utility incoming circuit-breakers 6 Generator set protection 6.1 General protection philosophy 6.2 Electrical protection 6.3 Machine protection 7 Connection of generators to electrical network 8 Load shedding 9 Interfacing generator with electrical distribution system 9.1 Typical split of supply between generator set manufacturer and switchgear manufacturer 9.2 Information to be exchanged 9.3 Integration of generator set into electrical distribution supervisory system 10 Installation of engine generator sets 10.1 Location 10.2 Air intake and exhaust 10.3 Compliance with local regulations 10.4 Special tools and spare parts 11 Conclusion Bibliography 7.1 Connection to generator circuit-breaker 7.2 Connection of generator neutral point p. 4 p. 5 p.7 p. 9 p. 11 p. 12 p. 12 p. 12 p. 14 p. 14 p. 14 p. 15 p. 16 p. 17 p. 18 p. 19 p. 19 p. 20 p. 21 p. 21 p. 22 p. 23 p. 23 p. 23 p. 24 p. 24 p. 24
1 Types of engine generator sets
The main types of prime movers used in engine driven generator sets for industrial sites and commercial buildings are Diesel engines, gas turbines, and steam turbines. Turbines are used mainly for production sets whereas Diesel engines can be used for both production and standby sets. Most of the topics covered in this cahier technique are not dependant on the type of
prime mover used, and therefore the general term generator set will be used. The choice of the prime mover is determined by such considerations as the availability and type of fuel and is not covered in this cahier technique. Since Diesel engines are very often used some specific information about Diesel generator sets will be given.
Above is an example of a combined oil treatment and power plant. It incorporates two gas turbine generator sets with an output of approx. 100 MW.
(Courtesy of GE Energy Products France S.A.).
In most industrial plants, however, power generation is not the main purpose. The plant may have one or several Diesel generator units to produce the necessary electrical power, mainly for stand-by, and possibly for local consumption requirements. The picture shows a 1 MW Diesel generator unit.
(Courtesy of Houvenaghel/Hennequin S.A.).
Fig. 1: different configurations of local generation
2 Rated power for generator set applications
The power output requirement for the generator set is probably the most important criterion to be defined. The output of a generator set is typically
defined on the active/reactive power graph as represented in figure 2.
Active power axis Under-excited machine Over-excited machine Normal operation point, cos = 0.8 Stator current limit Rated engine power
Pn
Excitation current limit Stability limit
Reactive power axis
Qn
Fig. 2: active/reactive power graph showing operating limits
The active power output depends on the type of fuel used, and on site conditions including ambient temperature, cooling medium temperature, altitude, and relative humidity. It also depends on load characteristics such as possible overloading and load variations over time. The ISO 3046-1 standard for Diesel engines defines three different types of power ratings, and a standard definition of overload capability. The different power ratings are: c continuous power rating: The engine can supply 100% rated power for an unlimited time. This rating is normally used for production sets. c prime power rating: The engine can supply a base load for an unlimited time, and 100% rated power for a limited time. The base load and acceptable time for 100% rated power are different for each manufacturer. Typical values are a base load of 70% of the rated power, and 100% rated power during 500 hours per year.
c standby power rating: This is the maximum power that the engine can deliver and is limited in time, typically less than 500 hours per year. This rating should only be applied to generator sets which are used exclusively for emergency power. Since the engine is incapable of supplying more power, a security factor of at least 10% should be used when defining the standby power rating. The standard overload capacity is defined as 10% more power during 1 hour for every 12 hours of operation. There is no overload capacity with a standby power rating. Most manufacturers allow the standard overload capacity with the continuous power rating and the prime power rating, but since there are exceptions, the overload capacity should always be specified together with the type of power rating used. A typical example is a Diesel engine having a continuous power rating of 1550 kW, a prime power rating of 1760 kW, and a standby power rating of 1880 kW.
When generator sets are used as a prime source of electrical energy the following points should be considered: c provide for parallel operation with other sets and/or with utility, c allow for long maintenance periods (overhaul), c ensure black-start capabilities, c use low speed equipment for long life (maximum 750 rpm for Diesel engines). When used as a standby source: c ensure quick and reliable start-up and loading, c implement reliable load shedding to avoid overloading or stalling, c allow for periodic testing under load, c provide for parallel operation with utility if set is used during peak loads, c supply magnetizing current for distribution transformers. One common application for standby generators is to supply UPS (uninterrupted power supply) equipment during power outages. Since the
generator has a relatively high impedance as compared to a utility supply, voltage waveform distortion can occur due to harmonic currents generated by the UPS. Generator manufacturers normally derate their machines by up to 60% to ensure correct voltage waveforms when loads are UPS equipment without harmonic current filtering. The engine must also be able to supply the power absorbed by the UPS which is determined by P= UPS output kW + battery recharge kW + auxil. load UPS efficiency
For preliminary generator set sizing where detailed UPS information is unavailable, the battery charger kW can be estimated to be 25% of the UPS output kW, and the UPS efficiency can be estimated to be 90%. Final determination of the generator set should be based on specified values of acceptable voltage distortion, and the actual UPS data such as efficiency, and harmonic currents.
3 Typical applications
3.1 Standby generator sets

The typical supply of essential loads for commercial buildings, small industrial sites or for emergency power to unit substations in a larger site, is shown in figure 3. Under normal operating conditions the essential load is supplied from the utility supply. Upon loss of this supply the bus-tie circuit-breaker Q3 is tripped, the generator set is started, and then load is supplied by the standby generator set by closing the generator circuit-breaker Q2. Critical loads which cannot accept any power outage are supplied by the UPS. The UPS is equipped with a static switch which will immediately bypass the rectifier/inverter module in case of an internal fault and thus ensure a continuous supply of electrical power. Typical generator set sizes for this scheme are 250 kVA to 800 kVA. The advantage of this scheme is its simplicity and clarity. All essential loads are connected to the same busbar as the generator set and therefore no load shedding is required. UPS backup time can normally be limited to 10 minutes since the UPS will be supplied by the emergency supply. Both the normal and the backup supply to the UPS should be taken from the essential busbar.
Emergency supply Utility supply
G
Q1 Q3 Q2
Normal loads
~~
Static switch UPS
Emergency loads
Critical loads
Fig. 3: typical emergency supply for small industrial sites
For large industrial sites a centralized emergency power supply system as shown in figure 4 is often used. The main emergency switchboard is normally supplied from the utility, although in some sites
one of the generator sets may be in constant operation. The emergency switchboard is designed to allow generator sets to operate in parallel and also to be connected to the utility supply.
Utility incoming feeders
33 kV
G
6 kV
Essential loads Emergency switcboard main substation
Essential loads
To essential busbar of other unit substations ATS 6 kV
Normal loads
Essential loads
Typical unit substation - medium voltage switchboard (ATS: Automatic transfer system)
Fig. 4: typical emergency supply for large industrial sites
The automatic transfer from the utility to the emergency supply is performed in each unit substation. Since the emergency switchboard is normally energized, fast transfers (described in section 5.1) without loss of plant load can be used. The use of a centralized emergency supply has the following advantages:
c fewer generator sets for the site (normally maximum of 2), c permanently energized emergency supply allowing fast transfer schemes to be used, c no loss of emergency supply due to maintenance of one generator set. Generator sets for such systems are normally in the 1-4 MW range.
3.2 Production generator sets

For remote sites having no utility supply, several generator sets are used. A typical distribution system is shown in figure 5. The number of sets N will depend on the power required, but since generator sets require periodic maintenance, plant power should be able to be supplied by N - 1 sets without any load shedding. The generator set size should be such that they are loaded at least 50%. A poor load factor can be detrimental to the sets. For example Diesel engines loaded at less than 30% will not achieve a good operating temperature resulting in poor combustion and degrading of lubrication oil. Plant operation at N - 2 sets should also be considered, this case occurring when one set is being maintained and there is a loss of an additional set. The highest initial load factor F that can be used with N installed generators such that load
Generator under maintenance
To other substations
To other substations
Earthing transformer
Typical unit substation
Fig. 5: industrial site without utility supply
shedding is not required for N - 2 operation can be determined from:

F= N 2 N1
For example the highest load factor for N = 6 will be 80%. Bus-tie circuit-breakers are often used for maintenance purposes. During normal plant operation all bus-tie circuit-breakers are normally closed. Short-circuit calculations should always take operation with N generators into account since it is normal to connect standby sets prior to switching off sets for maintenance. A power supply using local generation is generally much weaker than a utility supply and therefore it is probable that load shedding will be required to maintain system stability during fault conditions. Determination of how much load must be shed requires dynamic simulation of the network for different fault conditions such as a loss of a generator or a short-circuit. Prior to the study it is
necessary to determine which operating configurations are to be considered. Operating conditions with the bus-tie circuit-breaker both in the open and the closed positions will greatly increase the complexity of the load shedding system since each busbar can be operated independently and will require specific load shedding criteria. For most plants it is recommended that only the standard operating configuration be used for the dynamic simulations and definition of the load shedding strategy. Figure 5 shows each generator having its own transformer. The use of generator transformers has several advantages: c provides flexibility in the choice of generator voltage, c reduces peak short-circuit current at main board, c allows use of high impedance generator grounding (reduces possible damage to generator).
4 Operation of generator sets
4.1 Starting and stopping of generator sets

Since Diesel generator sets are often used for emergency power, it is necessary that steps be taken to ensure that the set will start correctly and quickly when required. An example of measures to be taken is lubrication, and heating of the cooling water when the set is not operating. The Diesel generator set manufacturer should list all such measures and the design should take into account the availability of all auxiliary supplies necessary during times when set is not operating. A starting time of 15 seconds from the start order to the closing of the generator circuit-breaker can be guaranteed by manufacturers. Specifying shorter starting times should be avoided since the decrease in starting time will be small and could increase the cost of the set. Critical equipment must be supplied by an UPS in any case. Two techniques are commonly used for starting. These are compressed air and battery, compressed air generally being used for larger sets. The starting equipment should be designed for a minimum of 3 consecutive starts. It should be carefully monitored in order to enable preventive maintenance to be carried out prior to a failure during an attempted start. Failure to start is most often due to a problem with the starting battery. Where reliable starting is essential, consideration should be given to using compressed air. When a generator is operating in parallel with another source, it will be synchronized as described in section 5.3 hereafter, and gradually loaded. When a generator set is operating alone, the load will be applied in one or more steps. The variation in frequency and voltage will depend upon the size of the step loads. As an example, step loads of 90% can be applied to a Diesel generator set without the frequency varying more than 10% and the voltage more than 15%. Should specific limits on frequency and voltage variations be required, they should be specified together with the type of load which is to be connected. This information should include motor starting characteristics such as the starting current, and the type of starting (direct-on-line, wye-delta). Several steps may be required should the frequency and voltage tolerance be small. When stopping a generator set, the power output should be reduced to zero by transferring the load to other sources, and the circuit-breaker then tripped. The generator set should be run for several minutes to allow it to cool down prior to shutdown. In some cases the cooling system should continue to operate after shutdown in order to remove latent heat from the machine. Manufacturers recommendations for shutdown should be followed. Generator set start and stop sequences should be handled by the generator set control equipment. Generator sets should be operated periodically. For installations where short power outages are not critical, opening the normal incoming circuitbreaker will cause the set to start and automatically pick up the emergency load. After the required minimum operating time, the generator circuit-breaker can be tripped and the normal source circuit-breaker closed. For plants where power outages mean unacceptable production losses, it must be possible to test generator sets without first switching off the supply. This is normally done by using a maintenance transfer. The generator set is started, and after it is ready to take load, it is synchronized to the incoming supply (see section 5.3 below). The generator circuit-breaker (or bus-tie circuitbreaker depending on the scheme) will then be closed and the generator will thus be paralleled with the incoming supply. The closing of the circuit-breaker will cause tripping of the incoming supply and the loads will be supplied by the generator. The transfer to the normal incoming supply is done in the same manner without power interruption. Since the supplies are paralleled only for a few hundred milliseconds, it is not necessary to dimension the switchboard for the combined short-circuit power of both the normal incoming supply and the generator. Where equipment has been designed to operate in parallel on a permanent basis, it is not necessary to trip the incoming supply after connection the generator to the load. For this case, however, the switchboard must be designed for the combined short-circuit power of the incoming supply and the generator.
4.2 Stand alone operation

Generator sets are often designed to operate independently (isochronous mode). In such cases the system frequency will be controlled by the engine governor. Overloads exceeding the maximum power output (standby power rating for Diesel engines as described in 2) of the set will cause the system frequency to decrease and this can be used for initiating load shedding. The generator voltage regulator will determine the system voltage. Generators can normally operate at a power factor of 0.8 and therefore supply most industrial loads without additional power factor compensation equipment.
4.3 Parallel operation with utility supply

In some cases permanent operation of the generator set in parallel with the utility supply is required. Since the utility supply is much stronger it will determine the system frequency and the system voltage. The governor will therefore be used to control the active power output of the engine, and the voltage regulator will control the reactive power output of the generator. The generator set must know in which configuration it is operating in order to be able to switch the governor and voltage regulator operation from frequency and voltage control (isochronous operation) to active and reactive power control (parallel operation). Auxiliary contacts from the switchboard are normally used to provide the necessary information to the generator sets.
4.4 Parallel operation with other generator sets

In this case generator sets are operated in parallel with other generator sets of approximately the same size. There are three basic schemes used. a) All generator sets but one have fixed active and reactive power output settings. One generator set is in the iscochronous mode and will provide the active and reactive power necessary to keep the system frequency and voltage within the allowable limits. Any synchronizing instructions for frequency or voltage changes will be sent to the generator set in the isochronous mode. Since all power fluctuations will be absorbed only by this generator set, this scheme cannot be easily used where there are large variations in load. b) All generator sets operate in the droop mode. The active and reactive power is then shared equally among the sets or in proportion to their rated power if sets with different ratings are used. Variations in load will cause voltage and speed fluctuations due to the droop characteristic which is normally 4% from zero to 100% load. Since synchronizing of the sets with another source can only be done by adjusting the droop setting, this scheme is normally not used when parallel operation with another source is required. c) All generator sets are interfaced in order to share the active and reactive power. An example of how this is done is shown in figure 6. Each engine governor receives the active power set point from the active load dispatcher which also provides frequency regulation. Similarly each excitation regulator receives the reactive power set point from the reactive power dispatcher which also provides voltage regulation. This scheme allows for large load variation without changes in frequency or voltage.
Governer kW sharing and frequency regulation kvar sharing and voltage regulation G G Excitation regulator G
Load
Fig. 6: parallel operation using a load dispatcher
5 Transfer schemes and synchronization
5.1 Automatic transfer on loss of supply

An automatic transfer normally occurs when there is a loss of the normal supply and the load is to be supplied from the back-up supply with a minimum outage time. The transfer is blocked should the reason for the loss of supply be a fault on the busbar. Closing the emergency supply circuit-breaker onto a busbar fault will result in loss of the emergency supply and could result in damage to the equipment. Two techniques for transferring are generally used, their choice being based on whether or not the plant can accept a brief loss of supply. Residual voltage transfer This is the most commonly used automatic transfer scheme and has the following basic steps: c trip the incoming breaker to isolate the load from the supply c start the generator set c shed any loads which cannot be supplied from the generator set c close the generator circuit-breaker after the generator set is able to be loaded, and the residual voltage on the busbar is less than 30%. Fast transfer A fast transfer scheme is used when the process cannot accept any power outages. Such a system requires that the backup supply be permanently available and that the load is transferred to the backup supply before drives have had time to slow down. The time window for such switching is about 150 ms. In order to avoid the mechanical stresses and large currents due to out-of-phase switching, it is necessary to give the closing order to the emergency supply circuit-breaker such that the voltage generated by the decelerating motors is close to being in phase with the emergency system voltage when the circuit-breaker closes. Control gear for such transfer systems take into account the closing time of the circuit-breaker in order to anticipate the correct switching moment. If switching does not occur during the 150 ms time window, the fast transfer is blocked and a residual voltage transfer is made including any required load shedding.
5.2 Maintenance transfer

After the normal supply has returned, the load should be transferred from the emergency supply back to the normal supply. This is normally initiated manually as described at the end of section 4.1 above.
5.3 Synchronization of generator circuit-breaker

Any time parallel operation of a generator set is required, it is necessary to be able to synchronize it to the system. Synchronization basically consists in adjusting the generator frequency and voltage to values close to the system values. Since the system frequency and voltage can vary within a few percent, it is necessary that both the engine speed and the generator voltage be able to be adjusted for synchronization purposes. The engine speed and generator voltage are controlled by the governor and voltage regulator. Adjustments in the frequency and voltage are normally achieved by momentarily closing contacts connected to the governor and voltage regulator. When the generator voltage is almost in phase with the system voltage a closing order is given to the generator circuit-breaker. Synchronization is normally done automatically by means of relays which measure generator and line voltages, frequencies, and phase angles. The relay automatically adjusts the speed and voltage of the generator set and closes the circuit-breaker when the phase angle between the generator and line voltages is sufficiently small. One set of automatic synchronization equipment can be used for several generators by selecting the appropriate voltage transformers and sending the voltage, speed as well as the closing order to the selected circuit-breaker. Manual synchronizing should be provided in all cases, either as a back up to the automatic synchronizing system, or for use in applications where synchronization would only rarely occur. For manual synchronization the operator uses
push buttons to provide the voltage and speed adjustment signals. A synchroscope will let the operator know when the line and generator voltages are sufficiently in phase to close the circuit-breaker. For manual synchronization use of a synchronism check protection relay is recommended which will
inhibit closing of the circuit-breaker unless all conditions of frequency, voltage, and phase angle have been satisfied. Synchronization across the generator circuitbreaker is often included as a standard feature in generator set control equipment.
5.4 Synchronization of bus-tie, bus-coupler, or utility incoming circuit-breakers

When several generator sets are used, they are often connected to different busbars in order to facilitate maintenance. It is therefore possible at times to have generator sets supplying loads on busbars which are not connected together. In order to have all busbars connected it will be necessary to synchronize groups of generator sets across bus-tie or bus-coupler circuitbreakers. Specific synchronization equipment is normally required for such applications since the generator set normally allows synchronizing across the generator circuit-breaker only. A similar situation can occur when plant load is being supplied by generator sets and it is necessary to connect the loads to the utility. Synchronization across the utility circuit-breaker will be necessary. Synchronization requires voltage and speed adjustments. As described in section 4.4 above, synchronization of a group of generator sets is possible when one set is in the isochronous mode, or when a load dispatcher is used which will change the power output (and therefore speed) of all sets. When a set is in the isochronous mode, the voltage and speed adjustment signals will be sent to that set and the others will follow according to their droop characteristic. When a load dispatcher is used, the frequency signal will be sent to the load dispatcher which then sends appropriate signals to the individual governors. The voltage regulators used in such cases are sometimes connected to the voltage transformer of the busbar to which they are to be synchronized and can therefore adjust their excitation accordingly without receiving a separate voltage signal. For both schemes, once the required frequency, voltage, and phase angle have been achieved, the circuit-breaker can be closed. Some manufacturers of load dispatching systems offer adjustment of the voltage in addition to adjustment of the speed. Specifications for synchronization equipment should therefore clearly specify all the functional requirements thereby allowing suppliers to choose the best solution.
6 Generator set protection
6.1 General protection philosophy

Since generators are a source of electrical power, the overcurrent protection relays should be connected to current transformers on the neutral side of the stator windings in order to cover faults occurring in the windings. Additional protection relays are required at the generator circuit-breaker only for applications where generator sets will be operating in parallel with other generator sets or with the utility, and will pick up faults on the line side of the generator. The current transformers for these protection relays are installed at the generator circuit-breaker in order to cover the whole connection to the generator. Reverse active and reverse reactive power relays are normally connected to current transformers on the neutral side of the generator as shown in figure 7. They can also be connected to the current transformers associated with the circuit-breaker. The location will depend on the split of works as described in chapter 9.1.
25
51
67
67N 27 87G 59 59N E 81 49T
a
64F
46
49
51
32P
32Q
51V
Generator star point 51G
Neutral earthing resistor
Fig. 7: recommended generator protection
6.2 Electrical protection

The recommended protection functions are shown in figure 7. Function reference numbers are the following: c protection functions connected to generator neutral current transformers: v 32P : reverse active power v 32Q : reverse reactive power serving as loss of field (for generators above 1 MVA) v 46 : negative sequence (for generators above 1 MVA) v 49 : thermal image v 51 : overcurrent v 51G : earth fault v 51V : voltage restrained overcurrent v 87G : generator differential protection (for generators above 2 MVA) (Note: 46,49, 32P and 32Q can also be connected to the line-side current transformers) c protection functions connected to voltage transformers: v 25 : synchronism-check (for parallel operation only) v 27 : undervoltage v 59 : overvoltage v 81 : overfrequency and underfrequency c protection functions connected to line-side current transformers (for parallel operation only): v 67 : directional overcurrent (not required if 87G is used) v 67N : directional earth fault (on core balance CT for better sensitivity) c generator mechanical protection functions connected to sensors v 49T : stator temperature (recommended for generators above 2 MVA) v 49T : bearing temperature (recommended for generators above 8 MVA) v 64F : rotor earth fault protection The following table ( see fig. 8 ) gives typical settings for each protection function, and what action should be taken. This information should be verified with the generator set manufacturer for each application. A general shutdown means tripping and locking out the generator circuitbreaker, switching off the excitation, and closing the fuel supply to the engine.
Function 27 32P 32Q 46 49
Typical setting 0.75 Un, T 3 s T > longest time of 51, 51V, 67 1-5 % for turbine, 5-20 % for Diesel, T = 2 s 0.3 Sn, T = 2 s 0.15 In, inverse time curve 80% thermal capacity = alarm 120% thermal capacity = trip time constant 20 min operating time constant 40 min standstill 1.5 In, 2 s 10 A, 1 s 1.5 In, T= 2.5 s 1.1 Un, 2 s Overfrequency: 1.05 Fn, 2 s Underfrequency: 0.95 Fn, 2 s 5 % In In, 0.5 s Is0 10 % of earth-fault current, 0.5 s Frequency < 1 Hz, Voltage < 5 %, Phase angle <10 120 C 10 A, 0.1 s
Action General shut-down General shut-down General shut-down General shut-down Trip breaker only, overload may be temporary
51 51G 51V 59 81 87G 67 67N 25 49T 64F Mechanical protection
General shut-down General shut-down General shut-down General shut-down General shut-down General shut-down General shut-down General shut-down Inhibit closing during synchronization Trip breaker only, overload may be temporary General shut-down General shut-down without lockout
Fig. 8: recommended relay settings and action
Particularities of generator short-circuit currents As shown in the above table, it is the duty of the generator circuit-breaker to effectively isolate the generator from the network. Due to the low values of transient and permanent short-circuit currents, care must be taken in the choice and setting of the protection relays. In addition, in order to reduce losses in the generator, generator stator resistance is normally kept low by the manufacturers. This will result in high X/R ratios which cause generator short-circuit currents to have a d.c. component with a long time constant. The IEC 60056 defines test conditions for medium-voltage circuit-breakers. The test conditions are based on short-circuit currents having a d.c. component with a time constant of 45 ms. Since generator short-circuit currents may have time constants greatly exceeding this value, the circuit-breaker manufacturer must
choose the adequate circuit-breaker and demonstrate that it is suitable for the application. Possible delaying of circuit-breakers In addition to the significant d.c. component, the generator short-circuit current can also have zero-axis crossings which occur only after several periods resulting in unsuccessful interruption of the short-circuit current as shown in figure 9. This is due to the alternating component of the short-circuit current decreasing much more rapidly than the d.c. component. Since medium-voltage circuit-breakers require natural zero-axis crossing of the short-circuit current for successful interruption, it may be necessary to delay operation of the circuitbreaker until such time as zero-axis crossings do occur. Such delays must be taken into account in the protection relay coordination study and can also reduce the system stability.
Voltage prior to fault
Fault current
Three-phase shortcircuit occurs
Circuit-breaker contact separation
Fig. 9: generator short-circuit current with delayed zero-axis crossing on phases 1 & 3 (phase 2 interrupts correctly since short-circuit occurs here when voltage is at its peak on this phase, consequently short-circuit current, with 90 lag, starts at zero, without dc component).
6.3 Machine protection

The generator set will have mechanical protection related to the prime mover. This typically includes oil level, oil temperature, water level, water temperature, and exhaust temperature. Often the rotor earth fault protection is provided as an integral part of the set since it requires injection of a d.c. current between the rotor and earth. A signal should be sent to trip the generator circuitbreaker without lockout should mechanical protection require a shutdown.
7 Connection of generators to electrical distribution network
7.1 Connection to generator circuit-breaker

Generators have a limited capacity to withstand voltage impulses. When it is possible to operate medium-voltage generators in parallel with the utility supply it is recommended to provide surge protection at the generator incoming terminals. This normally consists in connecting surge capacitors (typical value of 0.3 F) and lightning arresters between phase and ground in the generator line-connection box. Such precautions are not required for low-voltage generators since they are shielded from impulses by the upstream step-down transformers. When the generator line-connection box has been designed for surge protection devices, it is recommended to install the generator voltage transformers in it as well. The voltage transformer can however be easily integrated into the downstream switchgear should the generator connection box not be sufficiently large. Current transformers should be installed in the generator neutral point connection box. When generator differential protection excludes the generator line-side connection cable (or busduct), current transformers are installed in the generator line-side connection box. When generator differential protection includes the lineside connection cable (or busduct), the current transformers are installed in the downstream switchboard.
7.2 Connection of generator neutral point

Stand-alone generator set A generator which does not operate in parallel with any other source should be earthed by means of a resistor connected between the star point and earth. The generator manufacturer can provide a damage curve showing the allowable earth fault current as a function of time. The earthing resistor and protection relay settings should be determined based on this curve. In general earth fault current for medium voltage generators should be kept less than 30 A in order to prevent any damage to the stator itself. Operation in parallel with utility or other sets When several sets operate in parallel or together with the utility it is difficult to keep the earth fault current within acceptable limits. The maximum earth fault current will be the sum of the earth fault current in all sources and this can easily exceed the value given on the damage curve mentioned previously. Reducing this maximum value by limiting the earth fault current to a small value for each source will result in earth fault current being too small when only one or two sets are in operation. It is recommended to keep the star points unearthed and to provide earthing transformers for each busbar as shown in figure 5. When busbars are operated with the bus-tie circuit-breakers closed, only one earthing transformer should be connected. When the bus-tie circuit-breakers are open, one earthing transformer should be connected to each busbar section. This will permit a constant value of earth fault current independent of the type and number of sources used, and greatly simplify the earthfault protection system. Should a fault occur in the earthing transformer, it should be tripped but the generator sets connected to the busbar should be kept in operation. There is no immediate danger to the sets when operated on a temporarily unearthed system. The maintenance personnel should determine the subsequent operation of the system.
8 Load shedding
Load shedding is often required in order to ensure that the essential parts of the process are supplied with electrical energy during high load conditions, or when system disturbances occur. Since the only additional energy available in an electrical distribution system is the spinning reserve of rotating machinery, sites supplied by generator sets only have very limited reserves and are very susceptible to instability due to disturbances such as faults in the electrical distribution system. Three different scenarios requiring load shedding can be considered: c gradual increase in load c loss of a generator c electrical faults To ensure a reliable electrical supply to essential process equipment, each of the above cases must be studied to ensure that correct load shedding is implemented. In general the load shedding system must continually check the balance between the load and the available power in order to switch off non-essential loads required to maintain system stability. The effects and remedial measures for each scenario are described below. Gradual increase in load It is possible during certain periods that the total load exceeds the rated power of the generator sets. Due to the overload capacity of 10% for one hour normally provided with production sets, and the gradual increase of load, the load shedding system can perform all calculations in real time and generate load shedding signals to trip non-essential loads. The operators can switch the non-essential loads back on after the peak period has passed.
Loss of a generator The loss of a generator can suddenly result in the available power being much less than the load. It is necessary to shed non-essential loads immediately in order to ensure the stability of the electrical distribution system. If this is not done, other generators will be tripped due to overload, undervoltage or underfrequency and the whole electrical supply could be lost. The load shedding system normally prepares load shedding tables based on the scenario of loss of a generator so that when such an incident does occur, it can immediately send the trip signals. Load shedding can be achieved in less than 200 ms which is normally sufficient to prevent loss of system stability which could lead to a complete loss of the distribution system. Electrical faults When an electrical fault occurs, protection relays will detect the fault and circuit-breakers will isolate the faulty equipment. During the time required to eliminate the fault, the voltage at the fault can be very close to zero which can cause all the motors in the plant to decelerate. After the fault has been cleared, the motors will draw more current since they must be brought back up to speed. This can further reduce the voltage in certain portions of the network causing a snowball effect which can lead to tripping of circuit-breakers supplying healthy portions of the distribution system. In order to prevent such a loss of stability, load shedding based on voltage and/or frequency should be implemented. In order to determine how much load should be shed, and at what value of voltage or frequency, a stability study of the electrical distribution system is required. This study will modelize the dynamic response of the system to disturbances and enable the load shedding strategy to be prepared.
9 Interfacing generator with electrical distribution system
9.1 Typical split of supply between generator set manufacturer and switchgear manufacturer
It is very common for the generator set to be supplied by a different company than the company which supplied the switchgear to which it is connected. It is therefore beneficial for all parties to reduce the interfaces between the equipment to a minimum. A coordination meeting between the switchgear and generator set suppliers should be held prior to any detailed engineering. During this meeting the split of works, interfaces, information to be exchanged, and schedule should be determined. Correct definition should allow each supplier to do engineering, manufacturing, erection, testing, and equipment commissioning at site in an independent manner. System commissioning can then be done by both parties after all interfaces have been made. Keeping interfaces simple also enables each manufacturers responsibility to be clearly defined. Each supplier should be responsible for the installation of all equipment in his supply. Installing components supplied by one manufacturer in equipment supplied by the other should be avoided. A typical example is the generator excitation module which should be installed in a panel supplied by the generator manufacturer, and not in the switchgear. When generators can operate in parallel it is necessary to install protection gear in the switchgear for eliminating faults occurring between the generator and the switchgear. This protection gear should be in the switchgear manufacturers scope. Protection gear for the generator itself can be supplied either by the generator set manufacturer, or the switchgear manufacturer. Either solution is acceptable, and both require exchanges of information since equipment data for setting the relays will come from the generator set manufacturer, whereas information for the integration into the overall plant protection scheme will come from the switchgear manufacturer. When generator differential protection is used, it is quite common for the line-side current transformer to be installed in the switchgear and the neutral side current transformer to be installed in the generator neutral connection box. The supplier of the differential protection relay should define the characteristics of the line and neutral current transformers and each manufacturer should supply the current transformer to be installed in his equipment. It is not necessary nor for reasons mentioned previously is it desirable that one manufacturer supply the current transformers to be installed in the other manufacturers equipment. The auxiliary supplies for the generator set should be independent of those of the switchgear. The generator set should have its own battery backed d.c. supply.
9.2 Information to be exchanged

The information to be exchanged between the generator set and the switchgear should be kept to a minimum. The information should be exchanged by means of potential-free contacts, and 4-20 mA analog signals. The meaning of each signal (eg. close to actuate, closed for circuit-breaker open position) and the minimum duration of each signal (eg. closing signal duration: 500 ms) should be clearly stated on the interface documentation. Fail-safe circuits should be used. Such circuits use contacts which close to actuate, and normally open contacts which are maintained closed for authorization. These circuits are called fail-safe since a broken wire will not result in undesired actuation or authorization. The voltage to be applied to the potential-free contacts, and the contact loading should also be stated in order to ensure that the correct devices have been chosen.
This type of interfacing enables each supplier to design, manufacture, and test his equipment independently. Data exchanged directly via serial links should be avoided since this is much more difficult to define, commission, and trouble shoot. The amount of information to be exchanged does not justify this type of interface. The information typically exchanged is: c information from generator set: v ready to start (information) v ready for loading (information) v trip on fault (order)
v general alarm (information) v generator voltage (from voltage transformer, for synchronizing) c information to generator set: v start (order) v circuit-breaker on/off status (information) v busbar voltage (from voltage transformer, for synchronizing) v stand-alone operation, or parallel operation (information) v type of fault (information)
9.3 Integration of generator set into electrical distribution supervisory system

In order to prevent loss of supply preventative maintenance is required. Preventative maintenance can be very effective provided that the information needed to trigger it is available, thus ensuring that the maintenance will be made prior to the fault occurring. The required information can be collected and displayed to the operator by a power management system. Such information can include running hours of generator sets, temperature measurements of generator windings or bearings, and power consumption of particular loads. The power management system can also be used to supply the information required by the load shedding system described in section 8 above to perform the power balance calculations. The operator can also reconfigure the power distribution system from the power management system console. This is very convenient should an incident have occurred and switching be required to reenergize equipment.
10 Installation and maintenance of generator sets
The installation of generator sets requires close cooperation among several disciplines such as electrical, construction, process, and mechanical.
The following information should be considered when designing the installation of the sets.
10.1 Location
The location should be chosen close to the load center to reduce voltage drop and losses in the connections. Due to the relatively large size of the equipment, adequate space must be allowed for the transportation to and from the location. The building housing the equipment must have adequate space to allow maintenance including overhauling, and be provided with the necessary overhead cranes. The generator set manufacturer should provide all information concerning space and access requirements on civil works guide drawings. In many locations noise emission will be a problem. The solution consists in sound proofing the generator set, the building, or a combination of both. Sound proofing will have a significant impact on cost and therefore must be defined prior to placing an order for equipment. Care must also be taken to avoid noise transmission via the generator set base.
10.2 Air intake and exhaust

In the definition of the rated power of generator sets, the length and configuration of the air intake ducts and the exhaust piping is important. In certain cases generator sets will be located in areas where long ducting and piping is required, and this is to be taken into account in the definition of the rated power of the engine. Care must also be taken to ensure that the air intake is remote from the exhaust. Generator sets used for emergency power must be able to operate in all site conditions. In desert areas this can include sand storms. Special sand filters are required at the air intake and can increase the foot print and cost of the generator set.
10.3 Compliance with local regulations

In many countries there are local regulations that must be met. In addition to requirements related to emissions, environmental considerations often dictate the design of the fuel system. This can include the maximum capacity of day tanks and the type of buried storage tanks (double walled, etc.). Local regulations must also be respected for the fire detection and protection equipment. Fire detection should be installed in all locations where generator sets are located. Automatic fire protection equipment should also be provided where possible. Fire protection is normally achieved by flooding the building with inert gas. This type of system requires automatic shutting of ventilation openings, air intake openings, and doors. Local regulations cover many aspects such as the number and location of warning signs, the location of the fire control panel, and the type of inert gas which can be used. The assistance of a local company familiar with such regulations to get all required approvals is very useful and often indispensable.
10.4 Special tools and spare parts

Generator sets require periodic maintenance and also overhauls after a certain number of years of operation. Special tools are normally required for periodic maintenance, and additional special tools are required for overhauls. The definition and supply of tools should be made with the generator set manufacturer based on the type of maintenance to be performed. The list of special tools should be checked with the maintenance manuals in order to ensure that all have been provided. Spare parts for the first overhaul should be provided in addition to those required for normal operation.
11 Conclusion
Engine driven alternating current generating sets are often installed in industrial sites and commercial buildings as main sources of electrical energy or for supplying essential loads in case of loss of the utility supply. A good understanding of the electrical and mechanical characteristics of the generator sets and the standards which define them is important for correct choice of the equipment. The integration of the generator sets into the electrical distribution system has a large impact on most of the electrical equipment. The generators will contribute to the maximum available short-circuit current which must be taken into account in dimensioning the switchgear. The plant electrical protection
system must take into account the particularities of generators in order to ensure correct protection of persons and equipment but at the same time avoid nuisance tripping which results in loss of the supply of electrical power. The control system must enable the electrical distribution system to be operated in different configurations required for ensuring a reliable supply of power. The engineer responsible for the correct design of the complete electrical distribution system is confronted with many different types of problems to solve. Being aware of the problems and knowing typical solutions to them is the first step in ensuring that the final electrical distribution system will meet the requirements of the application.
Bibliography
Standards c IEC 60056: High voltage alternating current circuit breakers. c IEC 60255: Electrical relays. c IEC 60298: A.C. metal enclosed switchgear and controlgear for rated voltages above 1 kV and up to and including 52 kV. c IEC 60439-1: Low-voltage switchgear and controlgear assemblies. c ISO 3046: Reciprocating internal combustion engines. c ISO 8528: Reciprocating internal combustion engine driven alternating current generating sets.
Schneider Electric Cahiers Techniques c Electrical disturbances in LV Cahier Technique no. 141 R. CALVAS c Active harmonic conditioners and unity power factor rectifiers Cahier Technique no. 183 E. BETTEGA, J-N. FIORINA c Disjoncteurs au SF6 Fluarc et protection des moteurs MT Cahier Technique n 143 J. HENNEBERT et D. GIBBSo
Collection Technique ...................................................................................

Power Quality
Ph. Ferracci
"Cahiers Techniques" is a collection of documents intended for engineers and technicians, people in the industry who are looking for more in-depth information in order to complement that given in product catalogues. Furthermore, these "Cahiers Techniques" are often considered as helpful "tools" for training courses. They provide knowledge on new technical and technological developments in the electrotechnical field and electronics. They also provide better understanding of various phenomena observed in electrical installations, systems and equipment. Each "Cahier Technique" provides an in-depth study of a precise subject in the fields of electrical networks, protection devices, monitoring and control and industrial automation systems. The latest publications can be downloaded from the Schneider Electric internet web site. Code: http://www.schneider-electric.com Section: The expert's place Please contact your Schneider Electric representative if you want either a "Cahier Technique" or the list of available titles. The "Cahiers Techniques" collection is part of Schneider Electrics "Collection technique".
Foreword The author disclaims all responsibility subsequent to incorrect use of information or diagrams reproduced in this document, and cannot be held responsible for any errors or oversights, or for the consequences of using information and diagrams contained in this document. Reproduction of all or part of a "Cahier Technique" is authorised with the prior consent of the Scientific and Technical Division. The statement "Extracted from Schneider Electric "Cahier Technique" no. .." (please specify) is compulsory.
no. 199
Power Quality
Philippe FERRACCI Graduating from the "cole Suprieure dlectricit" in 1991, he defended a PhD on the resonant earthed neutral system in cooperation with EDF-Direction des Etudes et Recherches. He joined Schneider Electric in 1996, where he now conducts advanced research into the area of electrotechnical and electrical power system.
ECT 199 first issued September 2000
Power Quality
One of the properties of electricity is that some of its characteristics depend not only on the electricity producer/distributor but also on the equipment manufacturers and the customer. The large number of players combined with the use of terminology and definitions which may sometimes be imprecise partly explain why this subject area is so complex. This "Cahier Technique" aims to facilitate exchanges on this topic between specialists and non-specialists, as well as customers, manufacturers, installers, designers and distributors. The clear terminology used should help avoid confusion. It describes the main phenomena causing degradation in Power Quality (PQ), their origins, the consequences for equipment and the main solutions. It offers a methodology for measuring the PQ in accordance with differing aims. Illustrated with practical examples for the implementation of solutions, it shows that only by observing best practice and by applying strict methodology (diagnostics, research, solutions, implementation and preventive maintenance) can users obtain the right quality of power supply for their requirements.
Contents
1 Introduction 2 Degradation of PQ: origins characteristics - definitions 1.1 Context 1.2 Objectives of Power Quality measurement 2.1 General 2.2 Voltage dips and interruptions 2.3 Harmonics and interharmonics 2.4 Overvoltages 2.5 Voltage variations and fluctuations 2.6 Unbalance 2.7 Summary 3 Effects of disturbance on loads and processes 3.1 Voltage dips and interruptions 3.2 Harmonics 3.3 Overvoltages 3.4 Voltage variations and fluctuations 3.5 Unbalance 3.6 Summary 4 Level of power quality 5 Solutions for improving PQ 4.1 Evaluation methodology 4.2 EMC and planning levels 5.1 Voltage dips and interruptions 5.2 Harmonics 5.3 Overvoltages 5.4 Voltage fluctuations 5.5 Unbalance 5.6 Summary 6 Case studies 6.1 Hybrid filtering 6.2 Real time reactive compensation 6.3 Protection against lightning 7 Conclusion p.4 p.5 p.6 p.6 p.8 p.10 p.10 p.11 p.11 p.12 p.13 p.15 p.15 p.15 p.15 p.16 p.18 p.19 p.23 p.25 p.26 p.26 p.26 p.27 p.28 p.30 p.31 p.32
Bibliography
1 Introduction
1.1 Context
The quality of electricity has become a strategic issue for electricity companies, the operating, maintenance and management personnel of service sector and industrial sites, as well as for equipment manufacturers, for the following main reasons: c the economic necessity for businesses to increase their competitiveness, c the widespread use of equipment which is sensitive to voltage disturbance and/or generates disturbance itself, c the opening up of the electricity market. The economic necessity for businesses to increase their competitiveness c Reduction of costs linked to loss of supply continuity and problems of non-quality The cost of disturbance (interruptions, voltage dips, harmonics, lightning overvoltages, etc.) is substantial. These costs must take into account losses in production and raw materials, restarting of production facilities, non-quality of production and delivery delays. The malfunction or shutdown of vital equipment such as computers, lighting and safety systems may put lives at risk (e.g. in hospitals, airport lighting systems, public and high-rise buildings, etc.). Costs also include high quality, targeted preventive maintenance measures for anticipating possible problems. There is an increasing transfer of responsibility from the industrial user to the equipment manufacturer for the provision of site maintenance; manufacturers are now becoming electricity suppliers. c Reduction of costs linked to oversized installations and energy bills Other less obvious consequences of PQ degradation are: v A reduction of installation energy efficiency, leading to higher energy bills v Overloading of the installation, causing premature ageing and increasing the risk of breakdown, leading in turn to oversizing of distribution equipment This is why professional users of electricity are keen to optimise the operation of their electrical installations. The widespread use of equipment which is sensitive to voltage disturbance and/or generates disturbance itself As a consequence of their numerous advantages (flexible operation, excellent efficiency, high performance levels, etc.), we have seen the development and widespread use of automated systems and adjustable speed drives in industry, information systems, and fluocompact lighting in the service and domestic sectors. These types of equipment are both sensitive to voltage disturbance and generate disturbance themselves. Their multiple use within individual processes requires an electrical power supply which can provide ever increasing performance in terms of continuity and quality. The temporary shutdown of just one element in the chain may interrupt the whole production facilities (manufacture of semiconductors, cement works, water treatment, materials handling, printing, steelworks, petrochemicals, etc.) or services (data processing centres, banks, telecommunications, etc.). Consequently, the work of the IEC on electromagnetic compatibility (EMC) has led to stricter and stricter standards and recommendations (limitations on disturbances emission levels, etc.). The opening up of the electricity market The rules governing the electricity sector are undergoing radical change: electricity production has opened up to competition, production is decentralised, and (large) electricity consumers now have the opportunity to choose their supplier. In 1985, the Commission of the European Communities states (directive 85/374) that electricity is to be considered a product and as a consequence made it necessary to define its essential characteristics clearly. In addition, in the context of liberalising energy markets, the search for competitiveness by electricity companies now means that quality has become a differentiating factor. A guarantee of quality is a potential criterion of choice for industrial users when looking for an energy supplier.
1.2 Objectives of Power Quality measurement

The measurement parameters and accuracy may differ depending on the application. Contractual application Within the context of a deregulated market, contractual relations may exist not only between the electricity supplier and the end user, but also between the power production company and transmission company or between the transmission company and distribution company. A contractual arrangement requires that terms are defined jointly and mutually agreed upon by all parties. The parameters for measuring quality must therefore be defined and the values compared with predefined, i.e. contractual limits. This arrangement frequently requires the processing of significant quantities of data. Corrective maintenance Even where best practice is observed (singleline diagram, choice of protective devices and neutral point connection, application of appropriate solutions) right from the design phase, malfunctions may occur during operation: c Disturbances may have been ignored or under-estimated. c The installation may have changed (new loads and/or modification). Troubleshooting is generally required as a consequence of problems of this nature. The aim is frequently to get results as quickly as possible, which may lead to premature or unfounded conclusions. Portable measurement systems (for limited periods) or fixed apparatus (for continuous monitoring) make it easier to carry out installation diagnostics (detection and archiving of disturbances and triggering of alarms). Optimising the operation of electrical installations To achieve productivity gains (operational economies and/or reduction of operating costs) correct operation of processes and sound energy management are required, both of which are factors dependent on PQ. Operating, maintenance and management personnel of service sector and industrial sites all aim for an PQ which matches their requirements. Complementary software tools to ensure control-command and continuous monitoring of the installation are thus required. Statistical surveys Such research requires a statistical approach on the basis of wide-ranging results from surveys generally carried out by the operators of transmission and distribution power systems. c Benchmark the general performances of a power system These can be used, for example, to: v Plan and target preventive actions by mapping disturbance levels on a network. This helps reduce operating costs and improve control of disturbance. An abnormal situation with respect to an average level can be detected and correlated with the addition of new loads. Research can also be carried out into seasonal trends or excessive demand. v Compare the PQ of various distribution companies in different geographical areas. Potential customers may request details of the reliability of the electricity supply before installing a new plant. c Benchmark performances at individual points on the power system These can be used to: v Determine the electromagnetic environment in which a future installation or a new piece of equipment may have to operate. Preventive measures may then be taken to improve the distribution power system and/or desensitise the customer power system. v Specify and verify the performance levels undertaken by the electricity supplier as part of the contract. This information on the electricity quality are of particular strategic importance for electricity companies who are seeking to improve competitiveness, satisfaction of needs and customer loyalty in the context of liberalising energy markets.
2 Degradation of PQ: origins - characteristics - definitions
2.1 General
Electromagnetic disturbances which is likely to disturb the correct operation of industrial equipment and processes is generally ranked in various classes relating to conducted and radiated disturbance: c low frequency (< 9 kHz), c high frequency (u 9kHz), c electrostatic discharge. Measurement of PQ usually involves characterising low frequency conducted electromagnetic disturbances (the range is widened to include transient overvoltages and transmission of signals on a power system): c voltage dips and interruptions, c harmonics and interharmonics, c temporary power frequency overvoltages, c swell, c transient overvoltages, c voltage fluctuations, c voltage unbalance, c power-frequency variations, c DC in AC networks, c signalling voltages. It is not generally necessary to measure each type of disturbance. The types can be placed in four categories, affecting the magnitude, waveform, frequency and symmetry of the voltage. Several of these characteristics may be modified simultaneously by any one type of disturbance. Disturbances can also be classified according to its permanent, semi-permanent or random nature (lightning, short-circuit, switching operations, etc.).
2.2 Voltage dips and interruptions

Definitions A voltage dip is a sudden reduction of the voltage at a point in an electrical power system followed by voltage recovery after a short period of time from a few cycles to a few seconds (IEC 61050-161 ). A voltage dip is normally detected and characterised by the calculation of the root mean square value "rms (1/2)" over one cycle every half-cycle -each period overlap the prior period by one half-cycle- (see fig. 1). There is a dip to x % if the rms (1/2) value falls below the dip threshold x % of the reference value Uref. The threshold x is typiclly set below 90 (CENELEC EN 50160, IEEE 1159). The reference voltage Uref is generally the nominal voltage for LV power systems and the declared voltage for MV and HV power systems. A sliding reference voltage, equal to the voltage before the beginning of the disturbance is useful to study transference factor between different voltage systems. A voltage dip is characterised by a pair of data (see fig. 1b for x equal to 90): c depth: U (or its magnitude U), c duration T. In case of a non-rectangular envelope, the duration is dependent on the selected dip threshold value (set by the user according to the objective). The duration is typically defined as the time interval during which the rms (1/2) is lower than 90 %. The shape of the envelope (for example in case of complex multi-step and not simple one step dip) may be assessed using several dip thresholds set and/or wave form capture. Time aggregation techniques may define an equivalent dip characterised by the smallest rms (1/2) value measured during the dip and the total duration of the dip. For three-phase systems phase aggregation techniques (mainly used for contractual applications) may define a single phase equivalent dip (characterised for example by the greatest depth on the three phases and the total duration). Interruptions are a special type of voltage dip to a few percentage of Uref (typically within the range 1-10 %). They are characterised by one parameter only: the duration. Short interruptions last less than one minute (extended to three minutes depending on network operating conditions) and often result from tripping and automatic reclosure of a circuit breaker designed
to avoid long interruptions which have longer duration. Short and long interruptions differ in both their origins and the solutions required to prevent or reduce their occurrence. Voltage disturbances lasting less than a halfcycle T (T < T/2) is regarded as transient. Different terms are used in the USA depending on the length of the dips (sags) and interruptions: c instantaneous (T/2 > T > 30 T), c momentary (30 T > T > 3 s), c temporary (3 s > T > 1 min), c sustained interruption and undervoltage (T > 1 min). Depending on the context, the measured voltages may be between live conductors (between phases or between phase and neutral), between live conductors and earth (Ph/ earth or neutral/earth), or between live conductors and the protective conductor. In a 3-phase system, the characteristics U and T in general differ for each of the three phases. This is why a voltage dip must be detected and characterised separately on each phase. A voltage dip is regarded as occurring on a 3-phase system if at least one phase is affected by the disturbance. Origins c Voltage dips and short interruptions are mainly caused by phenomena leading to high currents, which in turn cause a voltage drop across the network impedances with a magnitude which decreases in proportion to the electrical distance of the observation point from the source of the disturbance. Voltage dips and short interruptions have various causes: v Faults on the transmission (HV) or distribution (LV and MV) networks or on the installation itself The occurrence of faults causes voltage dips for all users. The duration of a dip is usually conditioned by the operating time of the protective devices. The isolation of faults by protective devices (circuit breakers, fuses) will produce interruptions (long or short) for users feeded by faulty section of the power system. Although the power source is no longer present, network voltage may be maintained by the residual voltage provided by asynchronous or synchronous motors as they slow down (0.3 to 1s) or voltage due to the discharge of capacitor banks connected to the power system. Short interruptions are often the result of the operation of automated systems on the network such as fast and/or slow automatic reclosers, or changeover of transformers or lines. Users are
a V(p.u.) 1
0,5 t
-0,5
-1 b rms (1/2) (%) 110 100 90 70 T = 140 ms (duration) 10 0 0 50 100 150 200 250
U = 30 % (depth)
U (magnitude) t (ms) 300
Fig. 1: Characteristic parameters of a voltage dip [a] waveform [b] rms (1/2).
subjected to a succession of voltage dips and/or short interruptions caused by intermittent arc faults, sequence of automatic reclosing (on overhead or mixed radial networks) intended to extinguish transient and semi-permanent faults or voltage feedback intended to locate the fault. v Switching of large loads (asynchronous motors, arc furnaces, welding machines, boilers, etc.) compared to the short-circuit power. c Long interruptions are the result of the definitive isolation of a permanent fault (requiring to repair or to replace any component before re-energising) by means of protective devices or by the intentional or unintentional opening of a device. Voltage dips and interruptions are propagated to lower voltage levels via transformers. The number of phases affected and the depth of the voltage dips depend on the type of fault and the transformer coupling.
Overhead networks, which are exposed to bad weather, are subject to more voltage dips and interruptions than underground networks. However, an underground feeder connected to the same busbar system as overhead or mixed networks will suffer voltage dips which are due to the faults affecting overhead lines.
c Transients ( T < T/2) are caused, for example, by the energisation of capacitor banks, the isolation of a fault by a fuse or a fast LV circuit breaker, or by commutation notches from polyphase converters.
2.3 Harmonics and interharmonics

Summary: All periodic functions (of frequency f) can be broken down into a sum of sinusoidal waves of frequency h x f (h is an integer). h is the harmonic order (h > 1). The first order component is the fundamental component. harmonic frequencies thus has a vital role in limiting the voltage distortion. Note that if the source impedance is low (Scc is high), voltage distortion is low. Main sources of harmonics These are loads which can be distinguished according to their domain, i.e. industrial or domestic. c Industrial loads v Power electronic equipment: drives, rectifiers (diode or thyristor), inverters or switching mode power supplies; v Loads using electric arcs: arc furnaces, welding machines, lighting (discharge lamps, fluorescent tubes). Starting motors using electronic starters and power transformers energisation also generates (temporary) harmonics. Note that because of its multiple advantages (operating flexibility, excellent energy efficiency, high performance levels, etc.), the use of power electronic equipment is becoming more widespread. c Domestic loads with power inverters or switching mode power supplies such as television sets, microwave ovens, induction hotplates, computers, printers, photocopiers, dimmer switches, electrodomestic equipment, fluorescent lamps.
y(t) = Y0 +
h=1
Yh 2 sin(2 h f + h )
The rms is:

2 2 2 2 Yeff = Y0 + Y1 + Y2 + Yh + ...
The THD (Total Harmonic Distortion) factor measures the signal distortion:
2 Yh 40
THD =
h= 2
Y1
Harmonics are mainly produced by non-linear loads which draw current of a different wave form from the supply voltage (see fig. 2). The spectrum of the harmonics depends on the nature of the load. Harmonic voltages occur across network impedances resulting distorted voltages which can disturb the operation of other users connected to the same supply. The value of the supply impedance at different
Other loads Voltage source E Z U = E - ZI
Harmonics generator
Fig. 2: Degradation of network voltage caused by a non-linear load.
Although their individual power ratings are much less than for industrial loads, the combination of large numbers and simultaneous use over long periods creates significant sources of harmonic distortion. Note that the use of this type of equipment is increasing, as in some cases is the power rating. Harmonic levels These generally vary according to the operating mode of the device, the time of day and the season (heating and air conditioning). The sources usually generate odd harmonic components (see fig. 3). Power transformer energisation, polarised loads (half-wave rectifiers) and arc furnaces all generate even harmonics in addition to odd harmonics components.
Interharmonics are sinusoid components with frequencies which are not integer multiples of the fundamental component (they are located between harmonics). They are due to periodic or random variations in the power drawn by various devices such as arc furnaces, welding machines and frequency inverters (drives, cycloconverters). The remote control frequencies used by the power distributor are also interharmonics. The spectrum may be discrete or continuous and vary randomly (arc furnaces) or intermittently (welding machines). To study the short, medium and long term effects, the various parameters must be measured at time intervals which are compatible with the thermal time constant of the devices.
Non-linear loads Adjustable speed drive
Current waveform
A
Spectrum
% 100 t 50 0 1 5 7 11 13 17 19 23 25 h % 100 t 50 0 1 5 7 11 13 17 19 h % 100 t 50 0 1 3 5 7 9 11 13 h % 100 t 50 0 1 3 5 7 9 11 13 h
THDI
44%
Rectifier/charger
28%
Data processing load
115%
Fluorescent lighting
53%
Fig. 3: Characteristics of certain harmonics generators.
2.4 Overvoltages
Where voltage is applied to a device and the peak value exceeds the limits defined in a standard or specification, this is an overvoltage (see "Cahiers Techniques" nos. 141, 151 and 179). Overvoltages are of three types: c temporary power frequency, c switching, c lightning. They can appear: c in differential mode (between live conductors: ph/ph ph/neutral), c in common mode (between live conductors and the exposed-conductive-part or earth). Power frequency overvoltages By definition, these occur at power frequency (50/60 Hz). They have various origins: c An insulation fault When an insulation fault occurs between phase and earth in an isolated neutral system or impedance earthed neutral system, the voltage of the heathy phases to earth may reach the phase to phase voltage. Overvoltages on LV installations may come from HV installations via the earth of the HV/LV station. c Ferroresonance This is a rare non-linear oscillatory phenomenon which can often be dangerous for equipment and which is produced in a circuit containing a capacitor and a saturable inductance. Ferroresonance is often the apparent cause of malfunctions or the destruction of devices (see "Cahier Technique" no. 190). c Break of the neutral conductor Devices powered by the phase with the least load witness an increase in voltage (sometimes up to the phase to phase voltage). c Faults on alternator regulators or tap changer transformer c Overcompensation of reactive power Shunt capacitors produce an increase in voltage from the source to their location. This voltage is especially high during periods of low load. Switching overvoltages These are produced by rapid modifications in the network structure (opening of protective devices, etc.). The following distinctions are made: c switching overvoltages at normal load, c overvoltages produced by the switching on and off of low inductive currents, c overvoltages produced by the switching of capacitive circuits (no-load lines or cables, capacitor banks). For example, the energisation of a capacitor bank produces a transient overvoltage in which the first peak may reach 2r times the rms value of the nominal voltage and a transient overcurrent with a peak value of up to 100 times the rated current of the capacitor (see "Cahier Technique" no. 142). Lightning overvoltages Lightning is a natural phenomenon occurring during storms. A distinction is made between direct lightning strike (on a line or structure) and the indirect effects of lightning (induced overvoltages and increase in earth potential) (see "Cahiers Techniques" nos. 151 and 179).
2.5 Voltage variations and fluctuations

Voltage variations are variations in the rms value or the peak value with an amplitude of less than 10% of the nominal voltage. Voltage fluctuations are a series of voltage changes or cyclical or random variations in the voltage envelope which are characterised by the frequency of variation and the magnitude. c Slow voltage variations are caused by the slow variation of loads connected to the network. c Voltage fluctuations are mainly due to rapidly varying industrial loads such as welding machines, arc furnaces or rolling mills.
2.6 Unbalance
A 3-phase system is unbalanced if the rms value of the phase voltages or the phase angles between consecutive phases are not equal. The degree of unbalance is defined using the Fortescue components, comparing the negative sequence component (U1i) (or zero sequence component (U1o)) of the fundamental to the positive sequence component (U1d) of the fundamental.
Ui =
U1i U1d
and Uo =
U1o U1d
The following approximate formula can also be used: Ui = maxi
Vi Vavg Vavg
where Vi = phase voltage i and

Vavg = V1 + V2 + V3 3
The negative sequence (or zero sequence) voltage is produced by voltage drops along the network impedances due to negative sequence
(or zero sequence) currents produced by unbalanced loads leading to non-identical currents on the three phases (LV loads connected between phase and neutral, or singlephase or 2-phase MV loads such as welding machines and induction furnaces). Single-phase or 2-phase faults produce unbalance until tripping of the protective devices.
2.7 Summary
Disturbances
Voltage dips
Overvoltages
Harmonics
Unbalance
Voltage fluctuations
Characteristic waveforms
Origin of disturbance c Power system v Isulation fault v Switching c Equipment v Asynchronous motor v Synchronous motor v Welding machine v Arc furnace v Converter v Data processing loads v Lighting v Inverter v Capacitor bank : Occasional phenomenon : Frequent phenomenon
3 Effects of disturbance on loads and processes
Generally speaking, the effects of all disturbances can be classified in two ways: c Instantaneous effects: unwanted operation of contactors or protective devices, incorrect operation or shutdown of a machine. The financial impact of the disturbance can be costed directly.
c Deferred effects: energy losses, accelerated ageing of equipment due to overheating and additional electro-dynamic stress caused by the disturbance. The financial impact (e.g. on productivity) is more difficult to quantify.

Voltage dips and interruptions disturb many types of devices connected to the network. They are the most frequent cause of Power Quality problems. A voltage dip or interruption of a few hundred milliseconds may have damaging consequences for several hours. The most sensitive applications are: c complete continuous production lines where the process cannot tolerate any temporary shutdown of any element in the chain (printing, steelworks, paper mills, petrochemicals, etc.), c lighting and safety systems (hospitals, airport lighting systems, public and high-rise buildings, etc.), c computer equipment (data processing centres, banks, telecommunications, etc.), c essential auxiliary plant for power stations. The paragraphs below cover the main consequences of voltage dips and interruptions on equipment used in the industrial, service and domestic sectors. Asynchronous motors When a voltage dip occurs, the torque of an asynchronous motor (proportional to the square of the voltage) drops suddenly which slowdown the motor. This slowdown depends on the magnitude and duration of the dip, the inertia of the rotating masses and the torquespeed characteristics of the driven load. If the torque developed by the motor drops below the resistant torque, the motor stops (stalls). Following an interruption, at the time of voltage recovery, the motor tends to re-accelerate and absorb current whose value is nearly its starting current, the duration of which depends on the duration of the interruption. Where there are several motors in an installation, the simultaneous restarting may produce a voltage drop in the upstream impedances on the network which will increase the duration of the dip and may make restarting difficult (long restarts with overheating) or even impossible (motor torque lower than the resistive torque). Rapidly reconnecting (~ 150 ms) the power to an asynchronous motor which is slowing down without precautionary measures may lead to reclosing in opposition to the phase between the source and the residual voltage in synchronous motors. In this case the first current peak may reach three times the startup current (15 to 20 In) (see "Cahier Technique" no. 161). The overcurrents and consequent voltage drops have consequences for the motor (excessive overheating and electro-dynamic force in the coils, which may cause insulation failures and torque shocks with abnormal mechanical stress on the couplings and reducers, leading to premature wear or even breakage) as well as other equipment such as contactors (wear or even fusion of the contacts). Overcurrents may cause tripping of the main general protective devices of the installation causing the process to shutdown. Synchronous motors The effects are almost identical to those for asynchronous motors. Synchronous motors can however withstand deeper voltage dips (around 50%) without stalling, owing to their generally greater inertia, the possibilities of overexcitation and the fact that their torque is proportional to the voltage. In the event of stalling, the motor stops and the entire complex start-up process must be repeated. Actuators The control devices (contactors, circuit breakers with voltage loss coils) powered directly from the network are sensitive to voltage dips whose depth exceeds 25% of Un. Indeed, for a standard contactor, there is a minimum voltage value which must be observed (known as the drop-out voltage), otherwise the poles will separate and transform a voltage dip (lasting a few tens of milliseconds) or a short interruption into a long interruption (lasting several hours).
Computer equipment Computer equipment (computers, measurement apparatus) today occupy a dominant position in the monitoring and control-command of installations, management and production. All of this equipment is sensitive to voltage dips with depth greater than 10% Un. The ITIC (Information Technology Industry Council) curve formerly CBEMA curve shows on a duration-amplitude scale, the typical withstand of computer equipment to voltage dips, interruptions and overvoltages (see fig. 4). Operation outside these limits leads to loss of data, incorrect commands, and shutdown or malfunction of equipment. The consequences of
the loss of equipment functions depend in particular on the restart conditions when voltage is restored. Certain equipment, for example, has its own voltage dip detection devices which enable data to be backed up and ensure safety by interrupting calculation processes and any incorrect commands. Adjustable speed machines The problems of voltage dips applied to variable speed drives are: c It is not possible to supply sufficient voltage to the motor (loss of torque, slowdown). c The control circuits supplied directly by the network cannot function. c There is overcurrent when voltage recovers (the drive filter capacitor is recharged). c There is overcurrent and unbalanced current in the event of voltage dips on a single phase. c There is loss of control of DC drives functioning as inverters (regenerative braking). Adjustable speed drives usually trip out when a voltage dip deeper than 15% occurs. Lighting Voltage dips cause premature ageing of incandescent lamps and fluorescent tubes. Voltage dips deeper than or equal to 50% with a duration of around 50ms will extinguish discharge lamps. The lamp must then be left off for several minutes to cool the bulb before it is turned on again.
U (%) 500
200 140 120 100 80 70 0 0 0,01T 10-3 3.10-3 0,02 0,5 110 90
10 T (s)
Fig. 4: Typical withstand as defined by the ITIC curve.
3.2 Harmonics
The consequences of harmonics are linked to the increase in peak values (dielectric breakdown), rms values (excessive overheating) and to the frequency spectrum (vibration and mechanical stress) of voltages and currents. The effects always have an economic impact resulting from the additional costs linked to: c degradation in the energy efficiency of the installation (energy loss), c oversizing of equipment, c loss of productivity (accelerated ageing of equipment, unwanted tripping). Malfunctions are probable with a harmonic distortion factor of greater than 8% of the voltage. Between 5 and 8%, malfunctions are possible. c Instantaneous or short term effects v Unwanted operation of protective devices: harmonics have a harmful influence mainly on thermal control devices. Indeed, when protective devices of this type calculate the rms value of the current from the peak value, there is a risk of error and unwanted operation even during normal operation with no overload. v Disturbances induced by low current systems (remote control, telecommunications, hi-fi systems, computer screens, television sets). v Abnormal vibrations and acoustic noise (LV switchboards, motors, transformers). v Destruction of capacitors by thermal overload If the actual frequency of the upstream capacitor-network system is similar to a harmonic order, this causes resonance and amplification of the corresponding harmonic. v Loss of accuracy of measurement instruments A class 2 induction energy meter will produce in current and voltage, a 0.3% additional error in the presence of 5% of harmonic 5. c Long term effects Current overload produces excessive overheating and leads to premature ageing of equipment: v Overheating of sources: transformers, alternators (through increased joule and iron losses).
v Mechanical stress (pulse torque in asynchronous machines). v Overheating of equipment: phase and neutral conductors through increased joule and dielectric losses. Capacitors are especially sensitive to harmonics as their impedance decreases in proportion to the harmonic order. v Destruction of equipment (capacitors, circuit breakers, etc.). Overload and excessive overheating of the neutral conductor may result from the presence of third harmonic (and multiples of 3) currents in the phase conductors which add in the neutral. The TNC neutral earthing system uses the same conductor for neutral and protection purposes. This conductor interconnects the
installation earth, including the metal structures of the building. Third harmonic (and multiples of 3) currents will flow through these circuits and produce variations in potential with the following results: v corrosion of metal parts, v overcurrent in the telecommunication links between the exposed-conductive-part of two devices (for example, printer and computer), v electromagnetic radiation causing screen disturbance (computers, laboratory apparatus). The table in figure 5 summarises the main effects of harmonics and the normal permitted levels. Interharmonics affect remotely-controlled devices and produce a phenomenon known as flicker (see "Cahier Technique" no. 176).
Equipment Power capacitors
Effects Overheating, premature ageing (breakdown), resonance.
Limits
I < 1.3 In (THD I < 83%),

or U < 1.1 Un for 12 hrs/days at MRV or 8 hrs/days at LV
Motors
Losses and excessive overheating. Reduction of capacity for use at full load. Pulse torque (vibrations, mechanical stress) Noise pollution. Losses (ohmic-iron) and excessive overheating. Mechanical vibrations. Noise pollution. Unwanted tripping (exceeding voltage peak values, etc.). Additional dielectric and ohmic losses (especially in the neutral conductor if third harmonic currents present). Operating problems. Problems related to waveform (commutation, synchronisation).
h (harmonic variation factor)
HVF i 2%
Transformers Circuit breakers Cables
Uh / U1 i 6 to 12% THD i 10% Uh / U1 i 7% Uh / U1 i 5%
Computers Power electronics

FVH =
2 Uh 13
h= 2
Fig. 5: Effects of harmonics and normal permitted levels.
3.3 Overvoltages
The consequences are extremely varied according to the period of application, repetitivity, magnitude, mode (common or differential), gradient and frequency: c Dielectric breakdown, causing significant permanent damage to equipment (electronic components, etc.). c Degradation of equipment through ageing (repetitive rather than destructive overvoltages). c Long interruptions caused by the destruction of equipment (loss of sales for distribution company, loss of production for industrial companies). c Disturbance in control system and low current communication circuits (see "Cahier Technique" no. 187). c Electrodynamic and thermal stress (fire) caused by: v Lightning (usually) Overhead networks are most vulnerable to lightning, but installations supplied by underground networks may also be affected by
stress due to high voltage if lightning strikes close to the site. v Switching overvoltages: these are repetitive and their probability of occurrence is
considerably higher than that of lightning, with a longer duration. They can lead to degradation as serious as that caused by lightning.
3.4 Voltage variations and fluctuations

As fluctuations have a magnitude no greater than 10%, most equipment is not affected. The main effect of voltage fluctuations is a fluctuation in the luminance of lamps (flicker). The physiological strain (visual and nervous fatigue) depends on the magnitude of the fluctuations, the repetition rate of the variations, the composition of the spectrum and the duration of the disturbance. There is however a perceptibility threshold (the amplitude as a function of the variation frequency) defined by the IEC below which flicker is no longer visible.
3.5 Unbalance
The main effect is the overheating of 3-phase asynchronous machines. In fact, the zero sequence reactance of an asynchronous machine is equivalent to its reactance during the start-up phase. The current unbalance factor will thus be several times that of the supply voltage. Phase currents can thus differ considerably. This increases the overheating of the phase(s) which the highest current flows through and reduces the operating life of the machine. In practice, a voltage unbalance factor of 1% over a long period, and 1.5% over a few minutes is acceptable.
3.6 Summary
Equipment Sensitivity to disturbance Voltage dips Overvoltages < 0.5 s > 0.5 s
Harmonics
Unbalance
Voltage fluctuation
c Asynchronous motor c Synchronous motor c Actuator c Speed drive c Data processing load, numerical control c Induction furnace c Lighting c Capacitor bank c Transformer c Inverter c Circuit breaker c Cable
4 Level of Power Quality
4.1 Evaluation methodology

Contractual application The contract must state: c Its duration. c The parameters to be measured. c The contractual values. c The measurement point(s). c The voltages measured: These voltages (between phases and/or between phase and neutral) must be the equipment supply voltages. c For each parameter measured the choice of measurement method, the time interval, the measurement period (e.g. 10 minutes and 1 year for the voltage amplitude) and the reference values; for voltage dips and interruptions, for example, the reference voltage, detection thresholds and the distinction between long and short interruptions must be defined. c The measurement accuracy. c The method of determining penalties in the event of one party failing to honour the terms of the contract. c Clauses in the event of disagreement concerning the interpretation of the measurements (intervention of third parties, etc.). c Data access and confidentiality. Corrective maintenance This is generally the consequence of incidents or malfunctions during operation requiring troubleshooting in order to apply corrective measures. The usual steps are: c Data collection This involves the collection of information such as the type of load, the age of the network components and the single-line diagram. c Search for symptoms This involves identifying and locating the equipment subject to disturbance, determining the time and date (fixed or random) when the problem occurred, any correlation with particular meteorological conditions (strong winds, rain, storm) or recent modification of the installation (installation of new machines, modification of the power system). c Examination of the installation This phase is sometimes sufficient for quickly determining the origin of the malfunction. Environmental conditions such as humidity, dust and temperature must not be overlooked. The installation, especially the wiring, circuit breakers and fuses, have to be checked. c Monitor the installation This step consists in equipping the site with measurement apparatus to detect and record the event where the problem originated. It may be necessary to place instruments at several points in the installation, especially (where possible) close to the equipment subject to disturbance. The apparatus detects events when the thresholds of the parameters used to measure the Power Quality are exceeded, and records the data characterising the event (for example date, time, depth of voltage dip, THD). The waveforms just before, during and after the disturbance can also be recorded. The threshold settings must match the sensitivity of the equipment. When using portable apparatus, the duration of the measurements must be representative of the operating cycle of the factory in question (e.g. one week). It must always be assumed that the disturbance will recur. Fixed apparatus can be used for continuous monitoring of the installation. If the apparatus settings are correct, it will carry out prevention and detection by recording each occurrence of disturbance. The data can be displayed locally or remotely via an Intranet or Internet connection. This can be used to diagnose events as well as to anticipate problems (preventive maintenance). This is the case with apparatus in the Power Logic System range (Circuit Monitor - Power Meter), Digipact and the latest generation of Masterpact circuit breakers fitted with Micrologic P trip release (see fig. 6). Records of disturbance from the distributors power system which have caused damage (destruction of equipment, production losses, etc.) may also prove useful when negotiating compensation claims.
c Identification of origin The signature (waveform, profile of rms value) of the disturbance can in general be used by experts to locate and identify the source of the problem (fault, motor starting, capacitor bank energisation, etc.). The simultaneous recognition of the signature for the voltage and the current can be used to determine if the disturbance is sourced upstream or downstream of the measurement point. The disturbance may come from either the installation or the distribution power system. c Definition and choice of mitigation solutions A list of solutions and costings is prepared. The choice of solution is often made by comparing the cost with the potential lost earnings in the event of disturbance. After implementing a solution, it is important to verify, via measurement, that it is effective. Optimising the operation of electrical installations The operation of electrical installations can be optimised through three complementary actions: c Saving energy and reducing energy bills: v making users aware of costs, v assigning costs internally (by site, department or product line),
v locating potential economies, v managing peaks in consumption (load shedding, standalone sources), v optimising the power contract (reduction in subscribed power demand), v improving the power factor (reduction in reactive power). c Ensuring the Power Quality: v displaying and monitoring the measurement parameters for Power Quality, v detecting problems in advance (monitoring of harmonics and neutral current, etc.) for preventive maintenance purposes. c Ensuring continuity of service: v optimising maintenance and operation, v becoming acquainted with the network in real time, v monitoring the protection plan, v diagnosing faults, v reconfiguring a network following a fault, v ensuring an automatic source transfer. Software tools are used for the control-command and monitoring of the installation. They can be used for example to detect and archive events, monitor circuit breakers and protection relays in real time, control circuit breakers remotely, and generally make use of the possibilities of communicating devices (see fig. 6).
Circuit Monitor measurement and control device
Digipact power meter
Sepam
Digipact DC150 data concentrator
Compact NS circuit breaker
Masterpact circuit breaker
Fig. 6: Some communicating products (Merlin Gerin brand).
4.2 EMC and planning levels

Electromagnetic compatibility (EMC) Electromagnetic compatibility is the ability of an equipement or system to fonction satisfactorily in its electromagnetic environment without introducing intolerable electromagnetic disturbances to anything in that environment (IEC 60050-161). The aim of electromagnetic compatibility is to ensure that: c The emission of disturbances from each separate source is such that the combined emission from all sources does not exceed the expected levels of disturbance in the environment. c The immunity level of the equipment gives the appropriate level of performance for the expected disturbance in three classes of environment (see fig. 7). Note that the environment is also determined by the characteristics specific to the user installation (single-line diagram, types of load, etc.) and by the characteristics of the supply voltage. One way of ensuring compatibility levels is to specify the emission limits of user installations with a sufficient margin below the compatibility level. In practice this is possible for large installations (IEC 61000-3-6, IEC 61000-3-7). For other installations (e.g. LV) the "product" standards specify emission limits for families of equipment (e.g. standard IEC 61000-3-2 imposes emission limits on current harmonics for loads under 16 A). In certain cases, technical solutions must be applied to keep the emission levels below the prescribed levels. Voltage characteristics The method used to evaluate the actual voltage characteristics at a given point on the network and to compare them to the predefined limits is based on a statistical calculation over a given measurement period. For example, for the harmonic voltage the measurement period is one week: 95% of the rms values calculated over successive periods of 10 minutes must not exceed the specified limits. Planning levels These are the internal quality objectives specified by the network operator which are used to evaluate the impact of all disturbanceproducing loads on the network. They are usually equal to or below the compatibility levels. Summary Figure 8 summarises the relations between the various levels of disturbance.
Level of disturbance Susceptibility of equipment Immunity level (specified test value) Voltage characteristic Compatibility level (conventional value) Planning level Emission level Statistical distribution
Fig. 8: Relations between the various levels of disturbance.
Disturbances Voltage variations U/UN Voltage dips(1)
Class 1 8% 10% to 100% 1 none 2% 1%
Class 2 10% 10% to 100% 1 to 300 2% 1%
Class 3 +10% -15% 10% to 100% 1 to 300 i 60 3% 2%
U / UN T (number of half-cycle)
Short interruption (s) Voltage unbalance Ui / Ud Frequency variations f / fN
(1) These values are not compatibility levels: they are given for indicative purposes only.
Fig. 7: Compatibility levels according to IEC 61000-2-4.
5 Solutions for improving PQ
A degradation of quality may lead to a change in behaviour, performance or even the destruction of equipment and dependent processes with possible consequences for the safety of personnel and additional economic costs. This assumes three elements: c one or more generators of disturbance, c one or more loads sensitive to the disturbance, c a route for the disturbance to be propagated between them. The solutions consist in taking action with regard to all or part of the three elements, either globally (the installation) or locally (one or more loads). The solutions can be implemented to: c correct a malfunction in an installation, c take preventive action when polluting loads are to be connected, c ensure the installation conforms to a standard or to the power distributors recommendations, c reduce energy bills (reduction of subscribed power in kVA, reduction in consumption). Loads are not sensitive to the same disturbance and have different levels of sensitivity, the solution adopted, as well as being the best from a technical and economic point of view, must ensure an appropriate level of PQ which meets actual requirements. It is vital that specialists carry out a prior diagnosis to determine the nature of the
disturbance to be prevented (e.g. remedies may differ depending on the duration of an interruption). This determines the effectiveness of the chosen solution. The definition, choice, implementation and maintenance (to ensure long-term effectiveness) of solutions must also be carried out by specialists. The value of the choice and implementation of a solution depends on: c The required level of performance Malfunction is not permitted if it would put lives at risk (e.g. in hospitals, airport lighting systems, lighting and safety systems in public buildings, auxiliary plant for power stations, etc.). c The financial consequences of malfunction Any unprogrammed stop, even when very short, of certain processes (manufacture of semiconductors, steelworks, petrochemicals, etc.) results in loss or non-quality of production or even restarting of production facilities. c The time required for a return on the investment This is the ratio of financial losses (raw materials, production losses, etc.) caused by the non-quality of electrical power and the cost (research, implementation, operation, maintenance) of the solution. Other criteria such as practices, regulation and the limits on disturbance imposed by the distributor must also be taken into account.

The network architecture, automated power restart systems, the reliability of equipment, the presence of a control-command system and maintenance policy all play an important role in the reduction and elimination of interruptions. Correct diagnosis is vital before choosing an effective solution. For example, at the point of common coupling (the customers electricity input), it is important to determine whether the voltage dip is coming from the customers installation (with a corresponding increase in current) or from the distribution power system (no increase in current). Different types of solution exist. Reducing the number of voltage dips and interruptions Distributors can take certain measures such as making their infrastructure more reliable (targeted preventive maintenance, modernisation, underground installation) or restructuring power systems (shortening feeders). For impedance earthed neutral power systems, they can also replace auto-reclosing circuit breakers with shunt circuit breakers which present the major advantage of not causing interruptions on a damaged feeder in the event of a transient earth fault (reducing the number of short interruptions).
These circuit breakers allow the extinction of transient earth faults by cancelling the voltage to the fault terminals for at least 300 ms by earthing the single faulty phase at the substation busbars. This does not alter the voltage between phases supplying the customer. Reducing the duration and depth of voltage dips c At power system level v Increasing the possibilities of ring connections (new substations, ring closing switch) v Improving the performance of electrical protective devices (selectivity, automatic power restart, remote control devices on the network, remote management, replacement of spark gaps with surge arresters, etc.) v Increasing the network short-circuit power c At equipment level Decrease the power consumed by the switched large loads with real time reactive compensators and soft starters which limit current peaks (and mechanical stress). Increasing immunity of industrial and service installations The general principle for ensuring that equipment is immune to voltage dips and interruptions is to compensate for a lack of power with an energy storage device between the distribution power system and the installation. The availability of the storage device has to be greater than the duration of the disturbances to which the system has to be immune to. The information required when choosing mitigation solutions is: c the quality of the source (maximum level of existing disturbances), c the load requirements (voltage sag ridethrough capability in the duration-depth scale). Only by careful analysis of the process and of the technical and financial consequences of disturbances can these two elements be reconciled. There are various possible solutions to provide immunity depending on the power required by the installation and the duration of the voltage dip or interruption. It may well be helpful to study solutions by making a distinction between power supplies for control systems and
regulation systems and those for motors and large power consumers. Indeed, a voltage dip or interruption (even of short duration) may be sufficient to open all of the contactors whose coils are supplied by the power circuit. Loads controlled by the contactors are thus no longer supplied when the voltage is restored. Increasing immunity of the control system The increase of immunity of a process is in general based on providing immunity to the control system. In general, the control system is not of high power and is thus extremely sensitive to disturbances. It is therefore often more economical to immunise only the control system rather than the equipment power supply. Maintaining control of machines assumes: c There will be no risk to the safety of personnel or equipment when the voltage is restored. c The loads and processes tolerate a short interruption in the power circuit (high inertia or slowdown is tolerated) and can be restarted on the fly when the voltage is restored. c The source can ensure that all of the equipment can be supplied simultaneously (in the case of a replacement source) and provide the inrush current caused by the simultaneous restart of several motors. The solutions consist in powering all of the contactor coils from a reliable auxiliary source (battery or rotating set with flywheel), or in using an off-delay relay, or in using a rectifier and a capacitor connected in parallel with the coil. Increasing immunity of the equipment power supply Certain loads cannot withstand the expected disturbance levels, i.e. neither voltage dips nor interruptions. This is the case for "priority" loads such as computers, lighting and safety systems (hospitals, airport lighting systems, public buildings) and continuous production lines (manufacture of semi-conductors, data processing centres, cement works, water treatment, materials handling, paper industry, steelworks, petrochemicals, etc.). The following different technical solutions are possible depending on the power required by the
installation and the duration of the voltage dip or interruption. c Solid state uninterruptible power supply (UPS) A UPS consists of three main elements: v a rectifier-charger, powered from the main supply, to convert AC voltage to DC, v a flywheel and/or battery (kept charged) which on interruption provide the necessary power for the load via the inverter, v an AC-DC inverter. Two technologies are currently in use: on-line and off-line. v On-line technology During normal operation, power is supplied continuously via the inverter without drawing on the battery. This, for example, is the case for MGE-UPS brand Comet and Galaxy UPS units. They ensure continuity (no changeover delays) and quality (voltage and frequency regulation) of the power supply for sensitive loads ranging from a few hundred to several thousand kVA. Several UPS can be connected in parallel to obtain more power or to provide redundancy. In the event of overload, power is provided by the static contactor (see fig. 9) from network 2 (which may be combined with network 1). Power is maintained without interruption via a maintenance by-pass.
v Off-line (or stand-by) technology This is used for applications of no more than a few kVA. During normal operation, power is supplied from the network. In the event of loss of network power or if the voltage exceeds the prescribed tolerances, use is transferred to the UPS. The changeover causes an interruption of 2 to 10 ms. c Sources transfer A device is used to control transfer between the main source and a replacement source (and vice versa) for supply to priority loads and if necessary orders the shedding of non-priority loads. There are three types of transfer depending on the duration of transfer (t): v synchronous (t = 0), v delayed (t = 0.2 to 30 s), v pseudo-synchronous (0.1 s < t < 0.3 s). The devices require special precautions (see "Cahier Technique" no. 161). For example, if there are several motors in the installation, simultaneous restart may produce a voltage drop which could prevent restart or lead to excessively long restarts (with the risk of overheating). It is therefore prudent to install a PLC which will restart the priority motors at intervals, especially with a replacement (backup) source with a low short-circuit power.
Manual maintenance by-pass (NO) Static contactor Network 2 Supply network feeders Network 1 Switch or Circuit breaker (NC) Battery circuit breaker (NC) Battery Switch (NC) NO : normally open NC : normally closed Switch (NC)
Rectifier / charger
Inverter
Sensitive equipment
Fig. 9: Schematic diagram of an on-line uninterruptible power supply (UPS).
This solution is selected where the installation cannot withstand a long interruption of more than a few minutes, and/or requires a large amount of power. It can also be used in conjunction with a UPS. c Zero-time set In certain installations, the autonomy required in the event of interruption makes it necessary to install a generating set (large batteries would be too expensive, or cause technical or installation problems). Here, in the event of any loss of power supply, the battery or flywheel is used to provide sufficient time for starting and running up the stand-by engin generator, load shedding (if necessary) and interruption-free coupling by means of an automatic source changeover. c Electronic conditioners These are modern electronic devices to compensate voltage dips and interruptions to a certain extent with a short response time: for
example the real time reactive compensator compensates the reactive power in real time and is especially well suited to loads with rapid, large variations (welding machines, lifts, presses, crushers, motor starting, etc.). Clean stop If a stoppage is acceptable, it is especially advisable to prevent uncontrolled restarting if an unwanted restart would present a risk for the machine operator (circular saws, rotating electrical machines) or for the equipment (compression chambers while still under pressure, staggered restarts of air-conditioning compressors, heating pumps or refrigeration units) or for the application (necessity of controlling production restart). The process may be automatically restarted by a PLC using a predetermined restart sequence when conditions return to normal. Summary (see table below)
Installation power
Duration (indicative values) and technical requirements 0 to 100 ms 100 ms 400 ms 1 s 1 min > 3 min to 400 ms to 1 s to 1 min to 3 min
Immunisation solution
A few VA contacteurs.
Time-delayed contactors DC power with capacitor storage
< 500 kVA . < 1 MVA < 300 kVA < 500 kVA < 500 kVA < 1 MVA (up to 4800 kVA with several UPS in parallel) Between 15 minutes and several hours depending on battery capacity Transfer time to a backup source may cause a short interruption Between 15 minutes and several hours depending on battery capacity Between 10 minutes (standard) and several hours UPS depending on battery capacity
Rotating set with flywheel Transfer source with diesel set DC power with battery storage Rotating set with flywheel and thermal motor or backup source DC motor connected to battery and alternator
Effective mitigation solution Ineffective mitigation solution
5.2 Harmonics
There are three possible ways of suppressing or at least reducing the influence of harmonics. One section will examine the question of protective devices. c Reducing generated harmonic currents v Line choke A 3-phase choke is connected in series with the power supply (or integrated into the DC bus for frequency inverters). It reduces the line current harmonics (especially high number harmonics) and therefore the rms value of the current consumption and the distortion at the inverter connection point. It is possible to install the choke without affecting the harmonics generator and to use chokes for several drives. v Using 12-phase rectifiers Here, by combining currents, low-order harmonics such as 5 and 7 are eliminated upstream (these often cause the most disturbance owing to their large amplitude). This solution requires a transformer with two secondary windings (star and delta), and only generates harmonics numbered 12k 1. v Sinewave input current devices (see "Cahier Technique" no. 183) This method consists in using static converters where the rectifier uses PWM switching to absorb a sinusoidal current. c Modifying the installation v Immunise sensitive loads with filters v Increase the short-circuit power of the installation v Derate equipment v Contain polluting loads As a first step, the sensitive equipment must be connected as close as possible to the power supply source. Next, the polluting loads must be identified and separated from the sensitive loads, for example by powering them from separate sources or from dedicated transformers. These solutions involve work on the structure of the installation and are, of course, usually difficult and costly. v Protective devices and oversizing of capacitors The choice of solution depends on the installation characteristics. A simple rule is used to choose the type of equipment where Gh is the apparent power of all generators of harmonics supplied from the same busbar system as the capacitors, and Sn is the apparent power of the upstream transformer(s): - If Gh/Sn i 15%, standard equipment is suitable - If Gh/Sn > 15%, there are two possible solutions. 1 - For polluted networks (15% < Gh/Sn i 25%): the current rating of the switchgear and in-series links must be oversized, as must the voltage rating of the capacitors. 2 - For very polluted networks (25% < Gh/Sn i 60%): anti-harmonic chokes must be connected to the capacitors and set to a frequency lower than the frequency of the lowest harmonic (for example, 215 Hz for a 50 Hz network) (see fig. 10). This eliminates any risk of resonance and helps to reduce harmonics.
Z () Network only with capacitor with anti-harmonic choke
fr
f (Hz) far Zone where harmonics are present
Fig 10: Effects of an anti-harmonic choke on network impedance
c Filtering Where Gh/Sn > 60%, specialists must calculate and install the harmonics filter (see fig. 11). v Passive filtering (see "Cahier Technique" no. 152) This involves connecting a low impedance bypass to the frequencies to be attenuated using passive components (inductor, capacitor, resistor). Several passive filters connected in parallel may be necessary to eliminate several components. Careful attention must be paid to the sizing of harmonic filters: a poorly designed passive filter may lead to resonance and amplify frequencies which did not cause disturbance before installation of the filter.
v Active filtering (see "Cahier Technique" no. 183) This consists in neutralising the harmonics emitted by the load through analysis of the harmonics consumed by the load, and subsequently restoring the same harmonic current with the appropriate phase. It is possible to connect several active filters in parallel. An active filter may for example be connected to a UPS to reduce harmonics which have been injected upstream. v Hybrid filtering This consists of an active filter and a passive filter set to the order of the dominant harmonic (e.g. 5) which supplies the necessary reactive power.
Filter
Principle
Characteristics c No limits in harmonic current. c Compensation of reactive power. c Elimination of one or more harmonic orders (generally 5, 7, 11). One filter for one or two orders to be compensated. c Risk of amplification of harmonics in the event of network modification. c Risk of overload caused by external pollution. c "Network" filter (global). c Case by case engineering study.
Passive By-pass series LC circuit tuned to each harmonic frequency to be eliminated.
Network Passive filter(s)
Polluting load(s)
Active
Generation of current cancelling out all harmonics created by the load.
c Solution particularly suited to "machine" filtering (local). c Filtering on a wide frequency band (elimination of harmonic orders 2 to 25). c Self-adapting: v network modification has no effect, v adapts to all variations in load and harmonic spectrum, v open-ended, flexible solution for each type of load. c Simple engineering study.
Load(s) Network
Active filter
Hybrid Network Load(s) Active filter Passive filter
Offers the advantages of passive and active filtering solutions and covers a wide range of power and performance: c filtering on a wide frequency band (elimination of harmonics numbered 2 to 25), c compensation of reactive power, c large capacity for current filtering, c good technical-economic solution for "network" filtering.
Hybrid filter
Fig. 11: Principles and characteristics of passive, active, and hybrid filtering.
c Special case: circuit breakers (see "Cahier Technique" no. 182) Harmonics may cause unwanted tripping of protective devices: care must be taken when choosing protective devices to avoid this. Circuit breakers can be fitted with two types of trip device, thermal-magnetic or electronic. The heat sensors of thermal-magnetic circuit breakers are particularly sensitive to harmonics and can identify the actual load on the conductors caused by the presence of harmonics. They are thus well suited to use on low current circuits, essentially in domestic or industrial applications. The method used by electronic circuit breakers to
calculate the current being carried may present a risk of unwanted tripping and care must therefore be taken when choosing these devices that the true rms value of the current is measured. These devices have the advantage of being better able to track changes in the temperature of cables, particularly in the case of cyclical loads, as their thermal memory is superior to that of indirectly heated bimetallic strips. c Derating This solution is applicable to some equipment and is a simple and frequently adequate response to disturbance caused by harmonics.
5.3 Overvoltages
Correct insulation co-ordination involves ensuring the protection of personnel and equipment against overvoltages, with the best balance between technical and economic considerations. This requires (see "Cahier Technique" no. 151): c knowledge of the level and energy of the overvoltages which may occur on the network, c selection of the level of overvoltage withstand of the power system components to meet constraints, c use of protective devices where necessary, in fact, the appropriate solutions depend on the type of overvoltage encountered. Power frequency overvoltages c Switch off all or some of the capacitors during periods of low load. c Avoid configurations susceptible to ferroresonance or introduce losses (reducing resistors) to damp the phenomenon (see "Cahier Technique" no. 190). Switching overvoltages c Limit the capacitors energisation transients by installing a fixed reactor and pre-insertion resistors. Static automatic reactive compensators which control closing instant are especially suitable for LV applications which cannot withstand transient overvoltages (PLCs, computer systems). c Connect line chokes upstream of the frequency inverters to limit the effects of transient overvoltages. c Use main residual current circuit breakers of discriminatory type (type "S") for LV and circuit breakers of type "si" (In = 30 mA and 300 mA). Their use avoids unwanted tripping due to transient leakage currents (lightning and switching overvoltages, energisation of circuits with a high capacitance to earth (capacitive filters connected to earth, extended cable networks, etc.) which flow through the network downstream of the RCD (residual current device) via the network capacitance to earth. Lightning overvoltages c Primary protection This protects the building and its structure from direct lightning strikes (lightning conductors, Faraday cages, overhead earth wire/earthing wire). c Secondary protection This protects equipment against the overvoltages which follow lightning. Surge arresters (spark gaps are now used less and less) are installed on the particularly exposed points of HV and MV networks and at the input to MV/LV installations (see "Cahier Technique" no. 151). On LV installations, they are installed as far upstream as possible (to offer maximum protection) and as close as possible to the load. It is sometimes necessary to cascade surge arresters: one at the head of the installation, and one close to the load (see "Cahier Technique" no. 179). An LV surge arrester is always connected to a disconnection device. In addition, the use of main residual current circuit breakers of discriminatory type on LV installations avoids any current flow to earth via the surge arrester tripping the circuit breaker at the head of the installation, which would be incompatible with some equipment (freezers, controllers, etc.). Note that overvoltages can be propagated to the equipment by other routes than the electrical power supply: telephone lines (telephone, fax), coaxial cables (computer links, TV aerials). Suitable protective devices are commercially available.
5.4 Voltage fluctuations

Fluctuations produced by industrial loads may affect a large number of consumers supplied from the same source. The fluctuation magnitude depends on the ratio between the impedance of the device generating the disturbance and the impedance of the power supply. The solutions are: c Changing the type of lighting Fluorescent lamps are less sensitive than incandescent lamps. c Installing an uninterruptible power supply This may be a cost-effective solution if users subject to disturbance are identified and grouped together. c Modify the device generating the disturbance Changing the starting mode of motors which have to start frequently, for example, can reduce overcurrents. c Modify the network v Increase the short-circuit power by connecting lighting circuits to the nearest power supply point. v Increase the "electrical distance" between the disturbance-generating load and lighting circuits by powering the disturbance-generating load from an independent transformer. c Use a reactive compensator This device provides real time reactive compensation for each phase. Flicker can be reduced from 25% to 50%. c Connect a reactance in series By reducing the inrush current, a reactance downstream from the connection point of an arc furnace can reduce flicker by 30%.
5.5 Unbalance
The solutions are: c balancing single phase loads on all three phases, c reducing the power system impedance upstream of the devices causing the unbalance by increasing the transformer rated power and the cable cross-section, c fitting the appropriate protective device for the machines, c using carefully connected LC loads (Steinmetz connection).
5.6 Summary
Type of disturbance Voltage variations and fluctuations Voltage dips Origins Consequences Examples of mitigation solutions (special equipment and modifications) Electromechanical reactive power compensator, real time reactive compensator, series electronic conditioner, tap changer. UPS, real time reactive compensator, dynamic electronic voltage regulator, soft starter, series electronic conditioner. Increase the short-circuit power (Scc). Modify the discrimination of protective devices. UPS, mechanical source transfer, static transfer swtich, zero-time set, shunt circuit breaker, remote management. Anti-harmonic choke, passive or active filter, hybrid filter, line choke. Increase the Scc. Contain polluting loads. Derate the equipment. Series reactance.
Large load variations (welding Fluctuation in the luminance of lamps machines, arc furnaces, etc.). (flicker).
Short-circuit, switching of large loads (motor starting, etc.).
Disturbance or shutdown of process: loss of data, incorrect data, opening of contactors, locking of drives, slowdown or stalling of motors, extinguishing of discharge lamps.
Interruptions Short-circuit, overloads, maintenance, unwanted tripping. Harmonics Non-linear loads (adjustable speed drives, arc furnaces, welding machines, discharge lamps, fluorescent tubes, etc.). Fluctuating loads (arc furnaces, welding machines, etc.), frequency inverters. Operation of switchgear and Transient overvoltages capacitors, lightning. In terharmonics Voltage unbalance Unbalanced loads (large single-phase loads, etc.).
Overloads (of neutral conductor, sources, etc.), unwanted tripping, accelerated ageing, degradation of energy efficiency, loss of productivity. Interruption of metering signals, flicker.
Locking of drives, unwanted tripping, destruction of switchgear, fire, operating losses. Inverse motor torque (vibration) and overheating of asynchronous machines.
Surge arrester, surge diverter, controlled switching, pre-insertion resistor, line chokes, static automatic compensator. Balance the loads. Shunt electronic compensator, dynamic electronic voltage regulator. Increase the Scc.
6 Case studies
6.1 Hybrid filtering

Description of the installation Ski-lifts are powered by an MV/LV transformer (800 kVA). The connected loads are the chair lifts together with other loads such as payment booths, skipass validation systems, the official timing system for competitions and a telephone network. Problems encountered When the chair lifts are running, the low voltage network powered by the MV/LV transformer is subject to disturbance. The measures taken at the site pinpointed a high pre-existing harmonic distortion factor in the voltage (THDU 9%) from the MV power system as well as harmonic pollution from the chair-lift feeder. The resulting distortion of the supply voltage (THDU 12%) disturbed sensitive equipment (payment booths, timing system, etc.). Solutions The aim of the device is to ensure the simultaneous reactive compensation when harmonics are present and neutralisation of harmonics likely to disturb the installation. The solution chosen (see fig. 12) was to install a hybrid filter (see fig. 13) consisting of a passive filter tuned to the order of the dominant harmonic (H5) which provides the required reactive power (188 kvar), and an active filter rated at 20 A is dedicated to the elimination of all other harmonics. After commissioning, measurements show that the device reduces the magnitude of the harmonics over a wide frequency spectrum in both current and voltage (see fig. 14) and reduced the voltage distortion factor from 12.6% to 4.47%. It also increased the power factor of the installation from 0.67 to 0.87. This solution solved all of the problems as no malfunction was subsequently detected.
Fig. 13: Rectiphase hybrid filter device (Merlin Gerin brand).

a 14 12 10 8 6 4 2 % No filter H5 filter Hybrid filter
SCC
0 b 40 35 30 25 20 15 10 5 0 THDI H5 H7 H11 THDU % H5 H7 H11 H17 H23
MV/LV transformer
No filter H5 Filter Hybrid filter
5th order passive filter Harmonics generator Linear loads Active filter Hybrid filter
H17
H23
Fig. 12: Implementation of the solution.
Fig 14: Spectrums showing the effectiveness of a hybrid filter: [a] in voltage [b] in current.
6.2 Real time reactive compensation

Description of the installation A car equipment manufacturers plant in Concord (Ontario - Canada) is supplied by a transformer rated at 2000 kVA - 27.6 kV / 600 V - Yy Ucc = 5.23%. It manufactures exhaust assemblies from steel plate using spot welders and seam welders. Problems encountered c Visual and nervous fatigue in personnel, due to the fluctuation in brightness of lamps (flicker) when welding equipment was in operation. c Noise pollution and premature mechanical ageing of equipment caused by vibrations mainly in the transformer and the main switchgear when welding equipment was in operation. c Inability to add equipment for fear that the installation would be overloaded (peak currents when welders were fired were greater that the nominal current of the main circuit breaker). Expansion of the installation would thus require substantial investment, either to upgrade the existing installation or to build a new power supply facility. c Annual penalties of 5k Euro for exceeding the reactive power consumption limit (0.75 power factor). c Defective parts caused by welding faults appeared at the end of the manufacturing process when the tubes are bent into shape. All these factors reduced company productivity. Solutions The measures taken during the operation of the welding equipment showed a nominal voltage of 584 V, voltage dips of 5.8%, current peaks of 2000 A, and reactive power peaks of 1200 kvar (see fig. 15). The problems clearly stemmed from voltage fluctuations caused by the operation of welders with loads which vary rapidly and frequently and which consume significant reactive power. A voltage dip of 6% produces a reduction of 12% (1-0.942) in the power available for welding. This was the reason for the large number of defective welds. Standard devices for reactive power compensation use electromechanical contactors which cannot achieve the required response times; the operation of capacitor steps is deliberately time delayed to reduce the number of operations and avoid reducing the service life of the contactors through premature wear, as well as to enable the capacitors to discharge. The solution chosen was to connect a real time reactive compensator (see fig. 16). This innovative device offers: c ultra-rapid compensation of the variations in reactive power within one cycle (16.6 ms at 60 Hz), which is especially suitable for loads with rapid, large variations (welding machines, lifts, presses, crushers, motor starting, etc.); c transient-free switch through controlled switching, which is especially useful with loads which cannot withstand transient overvoltages (PLCs, computer systems, etc.); c increased service life of capacitors and contactors owing to the absence of moving mechanical parts and overvoltages With compensation of 1200 kvar it would be possible to minimise voltage dips, but 800 kvar was deemed sufficient to maintain the voltage at
a L1 L2 L3
Before Voltage (V) Voltage dip c Depth (%): c Duration (cycle) Current c Average c Peak Reactive power (kvar) Power factor THDI (%) 584 5.8 20 to 25 1000 2000 0.75 5 12
After 599 3.2 10 to 15 550 1250 > 0.92 4 12
600 1200 0 300
Fig. 15: Improvements due to the real time reactive compensator.
Fig. 16: Real time reactive compensator [a] principle, [b] practical implementation.
an acceptable level for all processes in the plant under all load conditions. The results of implementing the solution are (see fig. 17): c a reduction in current peaks to 1250 A and the addition of loads without modification of the installation, with improved installation efficiency through reduction of joule losses; c a reduction in reactive power peaks to 300 kvar and an increase in the power factor
to over 0.92, thus avoiding power factor penalties; c an increase in the nominal voltage to 599 V and a reduction in voltage dips to 3.2% (see fig. 16). This is a consequence of the increase in the power factor and reduction in the current amplitude (see fig. 18). Visual and nervous fatigue in personnel due to the flicker was also eliminated. Welding quality improved, as did productivity.
a 340 Voltage (V) 335 330
1500 Current (A) 1000 1000 500
kvar
d=3s b Voltage (V) Current (A) kvar 350 340 1250 750 500 0 d = 1.5 s Phase 1 Phase 2 Phase 3 Average
Fig. 17: Measurement of current, voltage and reactive power: [a] without compensation [b] with compensation.
V V
VS VL
VS VL VS Reactive compensator Welder VL
With
Without
Fig. 18: Reduction in voltage drop obtained using a real time reactive compensator.
6.3 Protection against lightning

Description of the installation The site consists of offices (computer hardware, lighting and heating unit), a security post (fire alarm, burglar alarm, access control, video surveillance) and three buildings for the manufacturing process on 10 hectares in the Avignon region of France (probability of lightning is 2 strikes per km2 per year). There are trees and metal structures (pylons) in the vicinity of the site. All of the buildings are fitted with lightning conductors. The MV and LV power supplies are underground. Problems encountered A storm struck the site, destroying the LV installation in the security post and causing 36.5k Euro of operating losses. The presence of lightning conductors prevented the structure from catching fire, but the electrical equipment which was destroyed was not protected by surge arresters, contrary to the recommendation in standards UTE C-15443 and IEC 61024. Solutions After analysing equipotentiality and earthing of the power system, followed by verification of the installation of lightning conductors and checking of the values of the earth electrodes, the decision was taken to install surge arresters. Surge arresters were installed at the head of the installation (main LV distribution board) and in cascade in each manufacturing building (see fig. 19). As the neutral point connection was TNC, protection would only be provided in common mode (between phases and PEN). In conformity with guide UTE C-15443 regarding operation in the presence of lightning conductors, the characteristics of the Merlin Gerin PF65 and PF8 surge arresters (see fig. 20) are as follows: c At the head of the installation
Main LV distribution board Building 1

3L N
Main LV distribution board Building 2

3L N
PF65
PF65
L N
L N
Secondary distribution board Building 1
PF8
Secondary distribution board Building 1
PF8
Fig. 19: Installation diagram for several surge arresters in cascade.
In = 20 kA Imax = 65 kA Up = 2 kV
c In cascade (at least 10 m apart) In = 2 kA Imax = 8 kA Up = 1.5 kV In cascade, good protection is provided for the secondary distribution boards (offices and security post). As the neutral point connection was converted to TNS, protection had to be provided in common mode (between phase and PE) and differential mode (between phases and neutral). The disconnection devices in this case are circuit breakers with a breaking capacity of 22 kA.
PF65
PF8
Fig. 20: Low voltage surge arresters (Merlin Gerin PF65 and PF8).
7 Conclusion
Electrical disturbance may originate in the distribution power system, in the installation of the user who is subject to disturbance or in the installation of a nearby user. The consequences of the disturbance vary according to the economic context and the area of application: from inconvenience to shutdown of production facilities - it can even put lives at risk. The search to improve company competitiveness and the deregulation of the electricity market
mean that the quality of electricity has become a strategic issue for electricity companies, the operating, maintenance and management personnel of service sector and industrial sites, as well as for equipment manufacturers. However, problems of disturbance should not be regarded as insurmountable, as solutions do exist. Employing specialists to define, implement and maintain these solutions while observing best practice will provide users with the right quality of power supply for their requirements.
Bibliography
Standards c IEC 61000-X-X Electromagnetic compatibility (EMC): v 2-1: Description of the environment. v 2-2: Compatibility levels (public low-voltage power supply systems). v 2-4: Compatibility levels in industrial plants for lowfrequency conducted disturbances. v 2-5: Classification of electromagnetic environments. v 3-2: Limits for harmonic current emissions (equipment input current i 16A per phase). v 3-3: Limitation of voltage fluctuations and flicker in low-voltage supply systems for equipment with rated current i 16 A. v 3-5: Limitation of voltage fluctuations and flicker in low-voltage supply systems for equipment with rated current > 16 A. v 3-6: Assessment of emission limits for distorting loads in MV and HV power systems. v 3-7: Assessment of emission limits for fluctuating loads in MV and HV power systems. v 4-7: Harmonics and interharmonics measurements v 4-11: Voltage dips, short interruptions and voltage variations immunity tests. v 4-12: Oscillatory waves immunity test. v 4-15: Flickermeter. c Other standards and laws v European Union "Council Directive 85/374 on the approximation of the laws of the Member States relating to the liability for defective products", Official Journal (07.08.1985). v EN 50160 Characteristics of electricity supplied by public distribution systems (07-1994). v Application Guide to the European Standard EN 50160 - July 1995 - UNIPEDE. v IEEE Std 1159-1995: Recommended Practice for Monitoring Electric Power Quality. v IEEE Std 1000-1992: IEEE Recommended Practice for Powering and Grounding Sensitive Electronic Equipment. v IEC 60071-1: Insulation coordination. v IEC 60050-161: International Electrotechnical Vocabulary. Cahiers Techniques Schneider Electric c Residual current devices. R. CALVAS, Cahier Technique no. 114. c Electrical disturbances in LV. R. CALVAS, Cahier Technique no. 141. c EMC: Electromagnetic compatibility. F. VAILLANT, Cahier Technique no. 149. c Overvoltages and insulation coordination in MV and HV. D. FULCHIRON, Cahier Technique no. 151. c Harmonic disturbances in networks, and their treatment. C. COLLOMBET, J.-M. LUPIN, J. SCHONEK, Cahier Technique no. 152.
c Inverters and harmonics (case studies of non-linear loads). J.-N. FIORINA, Cahier Technique no. 159. c Harmonics upstream of rectifiers in UPS. J.-N. FIORINA, Cahier Technique no. 160. c Automatic transferring of power supplies in HV and LV networks. G. THOMASSET, Cahier Technique no. 161. c HV industrial network design. G. THOMASSET, Cahier Technique no. 169. c Earthing systems in LV B. LACROIX, R. CALVAS, Cahier Technique no. 172. c Earthing systems worldwide and evolutions B. LACROIX, R. CALVAS, Cahier Technique no. 173. c The IT earthing system (unearthed neutral). F. JULLIEN, I. HERITIER, Cahier Technique no. 178. c LV surges and surge arresters. LV insulation co-ordination. Ch. SERAUDIE, Cahier Technique no. 179. c LV circuit-breakers confronted with harmonic, transient and cyclic currents. M. COLLOMBET, B. LACROIX, Cahier Technique no. 182. c Active harmonic conditioners and unity power factor rectifiers. E. BETTEGA, J.-N. FIORINA, Cahier Technique no. 183. c Cohabitation of high and low currents. R. CALVAS, J. DELABALLE, Cahier Technique no. 187. c Switching MV capacitor banks. D. KOCH, Cahier Technique no. 142. c Ferroresonance. Ph. FERRACCI, Cahier Technique no. 190. Other publications c Guide de lingnierie lectrique des rseaux internes dusines - Collection ELECTRA. c Method of symmetrical co-ordinates applied to the solution of polyphase networks - Trans. Amer. Inst. Electr. Engrs, June, 1918 - C.L. FORTESCUE. c Guide to quality of electrical supply for industrial installations Part 2: voltage dips and short interruptions Working Group UIE Power Quality 1996. c Supply Quality Issues at the Interphase between Power System and Industrial Consumers - PQA 1998, A. ROBERT. c Real time reactive compensation systems for welding applications - PQ 1998, R. WODRICH. c Low voltage hybrid harmonic filters, technical & economic analysis - PQ 1999, J. SCHONEK.
Schneider Electric
Direction Scientifique et Technique, Service Communication Technique F-38050 Grenoble cedex 9 Fax : 33 (0)4 76 57 98 60
Real: LAI Ltd - Tarporley - Cheshire -GB. Edition: Schneider Electric Printing: Imprimerie du Pont de Claix - Claix - 1000. - 100 FF09-00
94836

Discrimination with LV power circuit-breakers
JP. Nereau

LV protection devices and variable speed drives (frequency converters)
J. Schonek Y. Nebon
"Cahiers Techniques" is a collection of documents intended for engineers and technicians, people in the industry who are looking for more in-depth information in order to complement that given in product catalogues. Furthermore, these "Cahiers Techniques" are often considered as helpful "tools" for training courses. They provide knowledge on new technical and technological developments in the electrotechnical field and electronics. They also provide better understanding of various phenomena observed in electrical installations, systems and equipments. Each "Cahier Technique" provides an in-depth study of a precise subject in the fields of electrical networks, protection devices, monitoring and control and industrial automation systems. The latest publications can be downloaded from the Schneider Electric internet web site. Code: http://www.schneider-electric.com Section: Experts' place Please contact your Schneider Electric representative if you want either a "Cahier Technique" or the list of available titles. The "Cahiers Techniques" collection is part of the Schneider Electrics "Collection technique".
Foreword The author disclaims all responsibility subsequent to incorrect use of information or diagrams reproduced in this document, and cannot be held responsible for any errors or oversights, or for the consequences of using information and diagrams contained in this document. Reproduction of all or part of a "Cahier Technique" is authorised with the prior consent of the Scientific and Technical Division. The statement "Extracted from Schneider Electric "Cahier Technique" no. ....." (please specify) is compulsory.
no. 204
LV protection devices and variable speed drives
(frequency converters)
Jacques SCHONEK Graduate engineer from ENSEEIHT with a doctorate in Engineering from the University of Toulouse, he was involved in designing variable speed drives for the Telemecanique brand from 1980 to 1995. He then became manager of the Harmonic Filtering group. He is currently responsible for Electrotechnical Applications and Networks in the Advanced Design Office of Schneider Electrics electrical distribution management.
Yves NEBON He joined Merlin Gerin in 1969 and worked for 14 years in the low voltage design offices while continuing his professional training, gaining several diplomas and achieving the title of engineer. He then held a number of different positions in the Low Voltage Division. Since 1995 he has been responsible for the marketing, management and development of the Merlin Gerin brand LV electrical distribution product ranges within Schneider Electric.
CT 204(e) first issue, december 2001
LV protection devices and variable speed drives (frequency converters)

The purpose of this Cahier Technique is to explain the specific phenomena observed in LV installations when there is an overload or electrical fault in circuits equipped with variable speed drives. Various recommendations are given to ensure that persons and property are protected, and to provide optimum continuity of service.
Contents
1 Frequency converter type variable speed drives for asynchronous motors 2 Devices providing protection against overcurrents 3 Protection of persons 1.1 Description 1.2 Need for appropriate protection devices 2.1 Protection devices integrated in drives 2.2 Protection devices external to drives 3.1 Risks connected with insulation faults 3.2 Summary of earthing systems 3.3 Use of RCDs according to the earthing system 3.4 Earth fault protection devices integrated in drives 3.5 Insulation faults and variable speed drives 4 Protection devices to be used with drives 5 Special phenomena (summary table) 5.1 High-frequency leakage currents 5.2 Leakage currents on power-up 5.3 Fault at the drive output with a TT or TN system 5.4 Fault at the drive output with an IT system 5.5 Fault current with a DC component 6 Selection and installation recommendations 6.1 Selection of RCDs 6.2 Selection of PIMs 6.3 Prevention of malfunctions Bibliography p. 4 p. 7 p. 8 p. 9 p. 11 p. 12 p. 14 p. 14 p. 15 p. 19 p. 20 p. 22 p. 23 p. 25 p. 26 p. 28 p. 28 p. 29 p. 30
1 Frequency converter type variable speed drives for asynchronous motors
1.1 Description
Purpose The purpose of frequency converter type variable speed drives is to supply 3-phase asynchronous motors in such a way as to obtain operating characteristics which are radically different from those obtained in normal use (motors supplied directly from the power supply), with constant amplitude and frequency. The table in figure 1 lists the advantages of these devices. Principle This consists of supplying the motor with a voltage wave with variable amplitude and frequency, while keeping the voltage/frequency ratio more or less constant. This voltage wave is generated by an electronic power device whose simplified schematic is shown in figure 4.
Asynchronous motor in normal use Starting current Very high, around 6 to 8 times the nominal current in rms value and 15 to 20 times in peak value High and not controlled, around 2 to 3 times the nominal torque Tn
with variable speed drive Limited in the motor (in general: around 1.5 times the nominal current) Around 1.5 times the nominal torque Tn and controlled for the whole of the acceleration
Starting torque Ts
Starting
Sudden: its duration only depends on the Smooth, gradual and controlled characteristics of the motor and the driven (for example, linear speed ramp) load (resistive torque, inertia) Varies slightly according to the load (close to the synchronous speed Ns) High, around 2 to 3 times the nominal torque Tn Relatively complex, requiring protection devices and a special schematic Easy only after the motor has stopped Yes, with overtorque (resistive torque > Tm), or if voltage drops See fig. 2 Variation possible from zero up to a value greater than the synchronous speed Ns High, available across the whole speed range (around 1.5 times the nominal torque) Easy Easy No See fig. 3
Speed
Maximum torque Tm
Electrical braking Reversing Risk of stalling Motor operation in the torque-speed plane
Fig. 1 : comparison of the operating characteristics, showing the advantages of frequency converter type variable speed drives.
T Tm
Rectifier
Inverter
Motor
M
Ts Tn
Fig. 4 : simplified schematic of a frequency converter.

0 1 (/s)
Fig. 2 : speed-torque diagram for a motor supplied directly. The operating zone of the motor in the torquespeed plane is limited to the green part of the curve.
T Tm
Tn
reference, etc) and the measurement of the electrical values (line voltage, motor current). The PWM principle used in the inverter bridge consists of applying a series of voltage pulses to the motor windings, whose amplitude is equal to the DC voltage supplied by the rectifier. The width of the pulses is modulated so that a variable amplitude AC voltage is created. The curves shown in figure 5 are examples of phase-to-phase voltage and of current in one of the machine windings (assuming delta-connected windings).
(/s)
Fig. 3 : speed-torque diagram for a motor supplied via a frequency converter. The operating zone of the motor in the torque-speed plane is shown in green.
I load V load
The converter consists of: c a single-phase or 3-phase diode rectifier bridge combined with a capacitor, forming a DC voltage source (DC Bus), c an inverter bridge, generally with IGBTs (Insulated Gate Bipolar Transistors), supplied with a DC voltage, which generates an AC voltage wave with variable amplitude and frequency using the Pulse Width Modulation (PWM) technique, c a control unit providing the conduction commands to the IGBTs according to instructions given by the operator (run command, direction of operation, speed
t (s)
0.005
0.01
0.015
0.02
Fig. 5 : voltage with PWM and current in a machine winding.
Characteristics of the current drawn by the drive c Waveform v In the case of a 3-phase power supply The rectifier bridge combined with the filter capacitor draws a non-sinusoidal current from the line supply. The shape of this current is shown in figure 6 and its harmonic spectrum is shown in figure 7 . The typical harmonic distortion value THD is 40%. It should be noted that this distortion is obtained by adding line chokes that cause a voltage drop of between 3 and 5%. In the absence of these
line chokes, the current distortion is higher: it can reach 80% if there is no inductance in the drive. v In the case of a single-phase power supply The current drawn is shown in figure 8 and its spectrum in figure 9. The typical harmonic distortion value THD is 80%. It should be noted that this distortion is again obtained by adding a line choke that causes a voltage drop of between 3 and 5%. In the absence of this line choke, the current distortion is higher: it can exceed 100% if there is no inductance in the drive.
Line voltage Line current
Line voltage Line current
t (s)
t (s)
0.02
0.04
0.02
0.04
Fig. 6 : current drawn and phase voltage of the line supply (3-phase supply).
Fig. 8 : current drawn and phase voltage of the line supply (single-phase power supply).
% 100 90 80 70 60 50 40 30 20 10 0 1 3 5 7 9 11 13 15 17 19 21 23 25 Harmonic order 100 90 80 70 60 50 40 30 20 10 0
9 11 13 15 17 19 21 23 25 Harmonic order
Fig. 7 : spectrum of the current drawn from the line supply (3-phase supply).
Fig. 9 : spectrum of the current drawn from the line supply (single-phase supply).
c Variation of the line current according to the motor operating point Since the fundamental current drawn by the drive is practically in phase with the voltage, the current is proportional to the electric power drawn from the line supply. If we disregard the efficiency, this current is therefore proportional to the mechanical power supplied by the motor. The mechanical power is equal to the product of the torque times the speed. Thus, at reduced speed the mechanical power is low. As a result, the current drawn from the line supply by the drive is low when the motor operates at low speed, even if the motor produces a high torque and takes a high current.
Standard references Two standards are particularly concerned with the design of variable speed drives: c IEC 61800-3 Adjustable speed electrical power drive systems - Part 3: EMC product standard. c EN 50178 Electronic equipment for use in power installations. Conformity to this standard authorizes CE marking under the European Low Voltage directive. It should be noted that this standard also gives details for installing these products.
1.2 Need for appropriate protection devices

The characteristics described above, both technological (electronic power circuits) and those relating to the operation of the motor combined with a drive, demonstrate the need for appropriate protection devices in order to make the best use of such equipment. The creation of drives which incorporate electronic technology enables a number of these protection devices to be integrated at a very reasonable cost. Of course, these protection devices do not replace those that are required under existing regulations to be installed at the start of each circuit and which are thus external to the drives. Operation of all the protection devices used is described in the following sections.
2 Devices providing protection against overcurrents
The protection devices commonly used against overcurrents (circuit-breakers or fuses) are mainly provided to deal with two situations: c To protect an installation against any risk of short-circuit. c To avoid the risks resulting from a circuit overload or an operating current that exceeds
the capacities of the conductors (busbars and cables) and the control and protection equipment Variable speed drive technology enables some of these functions to be provided electronically.
2.1 Protection devices integrated in drives

Motor overload protection Modern variable speed drives protect the motor against overloads: c by instantaneous limiting of the rms current to around 1.5 times the nominal current, c by continuously calculating the I2t, taking account of the speed (as most motors are self-ventilated, cooling is less efficient at low speed). It should be noted that when a start circuit only supplies one motor and its drive, this motor overload protection simultaneously protects all the switchgear and wiring against overloads. Protection against motor or line shortcircuits downstream of the drive If there is a short-circuit between phases at the drive output (at the motor terminals or anywhere on the line between the drive and the motor), the overcurrent is detected in the drive and a blocking order is sent very quickly to the IGBTs. The short-circuit current (see fig. 10 ) is broken within a few microseconds, which protects the drive. This very fleeting current is mainly provided by the filter capacitor used with the rectifier, and is therefore imperceptible in the power supply line. Other protection devices integrated in drives Drives have other self-protection functions against: c Overheating of their electronic components that could result in their destruction. A sensor placed on the heatsink stops the drive when the temperature exceeds a certain threshold. c Line voltage dips: This protection is necessary to avoid any malfunction of the control circuits
Rectifier
Inverter
Motor
Fig. 10 : short-circuit downstream of the drive.
and the motor, as well as any dangerous overcurrent when the line voltage returns to its normal value. c Overvoltages at the line supply power frequency: This avoids possible destruction of their components. c Loss of a phase (for 3-phase drives): because the single-phase supply which replaces the 3-phase supply triggers an increase of the current drawn. Operation of the integrated protection devices If there is a fault, all these devices lock the drive and bring the motor to a freewheel stop. The power supply is then cut by the line contactor (its opening is controlled by a relay integrated in the drive).
2.2 Protection devices external to drives

In addition to the requirements described at the beginning of this section, these devices for protection against overcurrents are also designed to operate if there is an internal fault in the drive (destruction of the rectifier bridge for example): the line protection device breaks the fault current. Note: Although this device cannot normally protect the drive components, its automatic opening limits the consequences of such faults. Location of the devices The most common circuit layout for which these protection devices have been defined appears in figure 11 : c with, at the start of the circuit, individual protection against overcurrents often associated with a contactor, c without a breaker device downstream of the drive. The functions of the various devices (circuitbreaker, contactor and drive) are shown on this diagram. Type 2 coordination means that if there is a short-circuit: c no damage or loss of adjustment is permitted, c the insulation must be maintained, c the motor combination must be able to operate after the short-circuit has been removed, c the risk of contact welding on the contactor is permitted if these contacts can be separated easily. If there is a short-circuit risk upstream of the drive, in order to provide type 2 coordination, it is necessary to refer to the coordination tables provided by the manufacturers of the protection devices placed upstream. Note: With a drive, there is no current peak on power-up, and therefore no particular stress placed on the protection device. Calculation of the circuit-breaker and contactor rating This is determined according to the line current drawn by the drive. It is calculated based on: c the motor nominal mechanical power, c the nominal supply voltage, c the efficiency of the motor and the drive, c a permissible continuous overload of 1.1 Tn at constant torque and 1.05 Tn at variable torque, c harmonics, since the current is not sinusoidal. The rms value of the current, depending on the harmonic distortion, is obtained using the formula:
Circuit breaker: c short-circuit protection, c isolation for maintenance. Contactor: c automatic on-off, c power supply breaking in the event of a fault. Drive: c soft start, c variable speed control, c motor protection, c overload protection (wiring and switchgear).
Irms = I1 1+ THD2
thus, when THD = 40 %, Irms = 1.08 I1 . Since the fundamental current I1 is practically in phase with the voltage, the typical value of the current drawn by the drive, when it supplies a motor operating at its nominal point (constant torque application), is calculated using the formula:
Motor
Fig. 11 : recommended diagram for overcurrent protection.
Irms = 1.08 I1 1.08 x 1.1
Pmot 1 1 3U mot drive
These combinations (circuit-breaker, contactor and drive) offered by manufacturers are called motor combinations. Due to the protection devices integrated in the drives, these combinations naturally provide type 2 coordination if there is a short-circuit downstream of the drive.
Where: Pmot: motor nominal power U: phase-to-phase voltage mot: motor efficiency drive: drive efficiency
c Example: Motor rating: Line voltage: mot: drive: giving:
15 kW 400 V 0.95 0.97 Irms = 27.9 A
M
Circuit-breaker Contactor
Two special cases c Motors supplied in parallel In this case the overload protection integrated in the drive cannot protect each motor. This means that one of the motors could have an overload, although the current drawn by all the motors together does not exceed the drive nominal current. The motors must therefore be protected individually by a thermal overload relay (see fig. 12 ). It is however recommended that the overload protection integrated in the drive should remain active, in order to protect the cables upstream. c Integral drive overload protection disabled For certain applications for which continuity of operation is essential, the drive overload protection can be disabled. The cables and switchgear, which must be protected upstream, must therefore have an overload protection associated with the motor combination circuit-breaker (see fig. 13 ). Oversizing of the cable and the switchgear by 20% is recommended in this situation.
Fig. 12 : overload protection of a number of motors supplied by the same drive.
Circuit-breaker
Contactor
Fig. 13 : integral drive overload protection disabled.
3 Protection of persons
3.1 Risks connected with insulation faults

An insulation fault, whatever its cause, presents risks for: c the safety of persons (risk of electric shock), c the safety of property (risk of fire or explosion due to excessive localized temperature rise), c the availability of the electrical power (disconnection of part of an installation in order to eliminate the fault). The standards and regulations concerning the protection of persons identify two types of dangerous contact and stipulate the corresponding protection measures. Direct contact Persons coming into contact with live conductors (phase or neutral) or conductive parts that are normally live (see fig. 14 ). Protection against this risk is usually provided by insulating the live parts using barriers, screens or enclosures (in accordance with IEC 60364-4-41). These devices are of a preventive nature and are not fool proof. To alleviate any risk, an additional, automatic break protection measure is used, consisting of the detection of any earth leakage current which may circulate through a person, and which does not loop back to the source via the live conductors. Its trip threshold is set at 30 mA for AC current (IEC 60364-4-41) and 60 mA for DC current. Indirect contact Persons coming into contact with conductive metal frames, which are normally volt-free, and accidentally become live. This energizing is the result of failure of the insulation of a device or a conductor leading to an insulation fault (see fig. 15 ). This electrical risk depends on the contact voltage that develops between the frame of the faulty equipment and the earth or other conductive metal frames located nearby. Depending on the electrical connections between the live conductors, frames and earth, the standards provide different installation diagrams for defining the protection devices to be used. For further explanations see Cahier Technique no. 172.
If
N PE
Uf
If
RB
RA
N PE
In the event of contact with a frame which is accidentally energized (Uf) the danger threshold is fixed by the safety limit voltage UL . Thus where: RA = earthing resistance of the installation frames, RB = earthing resistance of the neutral, the operating threshold (In) of the protection device must be such that: Uf = RA .In i UL and then In i UL / RA. (For more details, please refer to Cahier Technique no. 114)
RB
RA
Fig. 14 : direct contact.
Fig. 15 : indirect contact.
3.2 Summary of earthing systems

There are three types of earthing system, commonly known as neutral point connection, for LV networks. They differ according to whether or not the neutral point of the voltage source is earthed and also the method used to connect the frames (see fig. 16 ). The choice of the neutral point connection depends on the installation characteristics and the operating conditions and requirements. For further details see also the following Cahiers Techniques: c No. 173 - Earthing systems worldwide and evolutions, c No. 178 - The IT earthing system (unearthed neutral) in LV. TT system In this type of system, known as directly earthed neutral: c the source neutral is connected to a separate earth connection from that of the frames, c all frames protected by one breaking device must be connected to the same earth connection. This is typically the case with public distribution in France. The TT system requires immediate breaking since any insulation fault may present a risk of electrocution. TN system The principle of this system, known as neutral connection, is that any insulation fault will trigger a single-phase short-circuit between phase and neutral. Immediate breaking is also essential and this system allows the use of the usual overcurrent protection devices to protect against insulation faults. In this type of system: c the LV neutral point of each source is connected directly to earth, c all frames in the installation are earthed (and thus connected to neutral) via a protective earth conductor : v PE separate from the neutral conductor; this is the TN-S system, v or common PEN with the neutral conductor; this is the TN-C system. Note: The TN-C system is not recommended for supplying electronic devices due to the possible circulation of harmonic currents in the neutral conductor that is also the protective earth. IT system In this type of system known as unearthed neutral: c The transformer neutral is: v either isolated from the earth (unearthed neutral), v or earthed via a high impedance (impedance-earthed neutral),
Directly earthed neutral (TT) 1 2 3 N PE
Neutral connection (TN-C) 1 2 3 PEN
RB
RA
RB
Unearthed neutral (IT) 1 2 3 N PE
Neutral connection (TN-S) 1 2 3 N PE
RB : Permanent insulation monitor.
RB
Fig. 16 : the three standard types of earthing system.
c all frames in the installation are connected to one another and earthed. In the IT system, the first insulation fault does not require breaking to take place, which enables the installation to continue to operate normally. However, this fault must be detected, signaled and then repaired before a second insulation fault occurs on another live conductor, which would then require an immediate trip. This rule gives the IT system the best continuity of service (see also Cahier Technique no. 178). Need for special detection methods The value of the insulation fault current between phase and earth (in common mode) depends on the earthing system. Its value is often too low to be detected and eliminated by conventional overcurrent protection devices (thermal or magnetic protection of a circuit-breaker), as is the case with the TT and IT systems. Two devices are designed particularly for protecting persons: Residual Current Devices RCDs - and Permanent Insulation Monitors PIMs -. c Residual Current Devices v Principle of RCDs This is illustrated in figure 17. If there is no insulation fault, the algebraic sum of the currents in the live conductors is zero and the toroid is not subjected to any magnetomotive force. If there is an insulation fault, this sum is no longer zero and the fault current creates a magnetomotive force in the toroid that generates a current in its coil. If the current exceeds a fixed threshold for a period longer than the optional time delay, an opening command is transmitted to the breaking device. For further details, see Cahier Technique no. 114: Residual current devices in LV.
v Types of RCD Standard IEC 60755 distinguishes three types of residual current protection device: AC: For sinusoidal AC currents A: For AC currents with a DC component (see fig. 18 ). These devices are suitable for detection of rectified single-phase currents. B: For DC currents. These devices are suitable for all types of current and are necessary, in particular, for rectified 3-phase currents. c PIMs The principle of a PIM consists of injecting an AC or DC voltage between the line supply and the earth (see fig. 19 ). The measurement of the current crossing the monitor is used to calculate the insulation resistance when this is a DC current, and that of the line supply/earth impedance when this is an AC current. These measurements associated with threshold devices authorize different alarms, for example that for the gradual reduction of insulation for the purpose of planned maintenance, or for an earth fault requiring rapid intervention (before the second fault).
6 mA
= 90
= 135
Fig. 18 : waveforms which characterize type A RCDs.
Actuator Time delay Threshold
Measurement current Threshold time delay alarm Installation insulation (Rf)
Measurement Measurement
I0
Wave shaping
a current generator
PE
Fig. 17 : principle of the RCD.
Fig. 19 : principle of a PIM.
3.3 Use of RCDs according to the earthing system

Cahier Technique no. 172: Earthing systems in LV covers this subject in greater depth. For all earthing systems RCDs are used as additional protection against the risks of direct contact. In certain countries they are even compulsory upstream of i 32 A power sockets with a i 30 mA trip threshold (for example, in France, according to standard NF C 15-100 532.26). RCDs with a sensitivity less than or equal to 500 mA (IEC 60364, section 482) are also recommended for monitoring electrical supply circuits in areas where there is a risk of fire. TT system The use of RCDs is the only method of detecting low fault currents. In fact, the impedance of a fault is not known with certainty and may be high (the load frames have separate earth connections and are not always connected to one another). TN-S or TN-CS system Using RCDs avoids the need to check the value of the current if there is a fault. They also control breaking of the circuits when the fault current, limited by a very long cable, is insufficient to activate the overcurrent protection devices. The trip threshold of the RCDs may then be high (low sensitivity), from a few amps to tens of amps. Note: According to IEC 60364: c a residual current device should not be used in a TN-C system, c when a residual current device is used in a TN-C-S system, a PEN conductor should not be used downstream. IT system In an IT system, RCDs are used in the following two situations. c If the short-circuit current (at the 2nd fault) may not be sufficient to activate protection against faults between phases, for example on feeders supplying loads which are a long distance away. c For groups of receivers connected to earth individually (groups of frames not connected to one another).
3.4 Earth fault protection devices integrated in drives

If the line supply has a TN system, a fault between a drive output and earth causes a significant overcurrent due to the interconnection of the frames (see fig. 20 ). As with a short-circuit between phases, this overcurrent is detected and a blocking command is sent to the IGBTs. However, in this situation the fault current circulates in the power supply line for a very short time (a few hundred microseconds). The integrated protection device then intervenes. The drive is electrically isolated by the opening of the line contactor. This device does not protect persons against indirect contact under all
Rectifier
Inverter
Motor
M
N
If
RB PE
Fig. 20 : earth fault at the output.
circumstances. In fact, the impedance of the fault may limit the current to a value that is below the drives protection threshold. This is generally the case with the TT system. A 300 mA RCD is therefore necessary.
In the case of an IT system, the first fault does not cause current to circulate and the drive continues to operate normally.
3.5 Insulation faults and variable speed drives

Direct contact There are a number of possible direct contact situations on circuits containing variable speed drives (see fig. 21 hereafter and fig. 22 next page). c TT and TN-S systems An additional protection device, in the event of failure of other protection measures against contact or carelessness by users, can be provided upstream of the drive using an RCD with a 30 mA threshold.
Risk of direct contact Upstream of the drive

A
Contact voltage Phase - neutral voltage
Additional protection 30 mA RCD
On the DC bus
B
Rectified line voltage
Downstream of the drive

C
Drive output voltage
Fig. 21 : voltages present in the event of direct contact with TT and TN-S systems.
c IT system with a first insulation fault present on the line supply
Risk of direct contact

A
Contact voltage Rectified line voltage
Additional protection No automatic break protection possible
Drive output voltage
Fig. 22 : various situations specific to a line supply with an IT system which has an insulation fault and a person in direct contact.
In the three situations specific to the IT system shown in figure 22, no automatic break protection can be applied: the fault current cannot be distinguished from the normal
operating current. This underlines the importance that installers must give to the wiring of these machines in order to ensure the basic precaution, i.e. insulation of live parts.
Notes: v The same is found for each of these situations when the two faults (insulation and direct contact) are swapped. v When a number of drives are powered via the same line supply, the DC buses of the different drives can be considered as being at the same voltage. Faults on different drives therefore have the same consequences as if they were located on one drive.
Indirect contact c TT and TN-S systems The indirect contact situations are shown in figure 23 . c IT system With this system the presence of two simultaneous faults must be taken into account. The table in figure 24 gives the various possible faults and their consequences.
Situation
Effect (risk)
Relevant protection Relevant with a TT system protection with a TN-S system Overcurrent protection placed upstream
Overcurrent RCD upstream of the drive
Overcurrent across a diode of the rectifier bridge. The fault current is a rectified current. (Risk of irreparable damage)
RCD immune to passage of DC current
Overcurrent across Internal in the drive a diode of the or rectifier bridge and RCD an IGBT. (Risk of irreparable damage depending on the type of short-circuit protection integrated in the drive)
Internal in the drive or overcurrent protection placed upstream, but a very long cable downstream of the drive may mask the fault, and an RCD is therefore recommended
Fig. 23 : various insulation faults with TT and TN-S systems.
Situation
A
Effect (risk) Overcurrent upstream of the drive
Relevant protection Overcurrent protection placed upstream
Overcurrent across a diode of the rectifier bridge (Risk of irreparable damage)
Overcurrent protection placed upstream
Overcurrent across a diode of the rectifier bridge and an IGBT (Risk of irreparable damage depending on the type of short-circuit protection integrated in the drive) Short-circuit on the DC bus (Risk of irreparable damage to the rectifier bridge)
Internal in the drive or overcurrent protection placed upstream, but a very long cable downstream of the drive may mask the fault. Overcurrent protection placed upstream
Short-circuit at the terminals of an IGBT (Risk of irreparable damage to the opposite IGBT depending on the type of short-circuit protection integrated in the drive)
Internal in the drive or overcurrent protection placed upstream
Overcurrent at the drive output Internal in the drive
Fig. 24 : various situations with two dead shorts or low impedance faults on an IT system.
c Special case of motors supplied via two separate circuits each containing a drive, where the frames have separate earth connections. When an installation supplies a number of devices that are a long way away from one another, the load frames are often connected with separate earth connections. The impedance of the circuit through which the fault current runs is thus increased by the resistance of the two earth connections concerned. The condition required for the protection of persons (compliance with maximum breaking times) can no longer be met by the short-circuit protection devices. The usual solution, which is very simple both to design and install, is to use RCDs (see Cahier Technique no. 178) placed at the start of each
circuit containing a separate earth connection (see fig. 25 ). These RCDs must not be disturbed by the passage of a DC component.
PIM
M
Fig. 25 : positioning of RCDs in an IT system to protect two circuits with separate earth connections.
4 Protection devices to be used with drives (summary table)
The following table summarizes the preceding sections, listing the protection functions provided by the drive, and if necessary the addition of
devices external to the drive such as circuitbreakers, thermal overload relays, and RCDs.
Protection to be provided Cable overload Motor overload Drive overheating Overvoltage Undervoltage Phase break Upstream short-circuit Internal fault
Protection generally provided by the drive Yes = (1) Yes = (2) Yes Yes Yes Yes
External protection Unnecessary if (1) Unnecessary if (2)
Downstream short-circuit Yes
Circuit-breaker (short-circuit trip) Circuit-breaker (short-circuit and overload trip) Self-protection RCD u 300 mA RCD i 30 mA
Downstream earth fault (indirect contact) Direct contact fault
5 Special phenomena
The purpose of this section is to analyze the special phenomena connected with the operation
of frequency converters, which affect RCD and PIM protection devices.
5.1 High-frequency leakage currents

The voltage waveform generated by the drive, and in particular the presence of transient voltage fronts generated by switching of the IGBTs, causes high-frequency leakage currents to circulate in the power supply cables. Flows These voltage fronts are applied to the different capacitances in the circuit (see fig. 26 ): Cc: capacitance of the IGBT components between conductors and enclosure connected to earth, Cm: capacitance between the motor windings and earth (depending on the motor rating), Cr: capacitance between the line supply and earth (short-circuited if the neutral is connected to earth), Cs: capacitance between the output conductors and earth (depending on the type and length of the cables), Cy: interference suppression capacitance at the drive input. Currents therefore circulate across these capacitances. The most significant are shown in figure 27 .
M
Cy Cs
Fig. 27 : circulation of high-frequency leakage currents.
Drive
M
Cr Cy Cc Cs Cm
Fig. 26 : capacitances in the circuit.
Characteristics These currents may reach an instantaneous value of several amps and an rms value of several dozen or several hundred milliamps. The spectrum and amplitude of these currents depend both on the PWM frequency (between 1 and 20 kHz) and the installation characteristics: c power supply: line impedance, earthing system, c type and length of the motor cable (shielded, not shielded, protective earth), c motor rating, The shape and spectrum of the HF currents at the drive input (when there is no fault), for a PWM at 4 kHz, are shown in figures 28 and 29 .
(A) 0.3
0.2
0.1
- 0.1
- 0.2
- 0.3
0.0
0.01
0.02
(s)
Fig. 28 : high-frequency leakage currents.
(A) 0.06
0.05
0.04
0.03
0.02
0.01
10
15
(kHz)
Fig. 29 : spectrum of the leakage current.
Their effect: risk of disturbance of RCDs c Undesired tripping These currents may cause malfunctioning of the residual current devices (RCD) when they flow through conductors surrounded by the measurement toroid. The measurement of the residual current may therefore be disturbed, in particular when the drive motor connection cable is very long and/or the capacitances between phases and earth are high (see fig. 30 ).
Solution The measurement device must therefore include a filtering circuit so that only the low-frequency component of the signal is taken into account (see fig. 31 ). Additional precautions, given at the end of this Cahier Technique, may be necessary in certain extreme situations.
Modular type Distribution board type
Fig. 30 : disturbance of an RCD by high-frequency leakage currents.
Fig. 31 : RCD integrating filtering of HF currents (Vigirex RH99M and RH99P Merlin Gerin brand).
5.2 Leakage currents on power-up

Origin Capacitors are generally placed at the input of frequency converters to provide them with immunity from the HF interference present on the line supply and reduce their HF emissions. Their capacitance is around 10 to 100 nF. These capacitors are responsible for residual currents (see fig. 32 ) at power-up and during normal operation. Their effect: risk of undesired tripping For a device that is operating normally, these currents are low (from 0.5 to 3.5 mA). However, in an industrial device that contains a number of drives, they can cause undesired tripping of RCDs. Solution This is the responsibility of the equipment manufacturer or the installer. It consists of limiting the number of drives supplied by the same RCD.
Fig. 32 : leakage current flowing through the input capacitors of devices (dotted lines).
Device with capacitive filter
5.3 Fault at the drive output with a TT or TN system

Risk of electrocution The fault current contains a component at the PWM frequency and HF currents created by the oscillations of the stray capacitances, but the dangers of HF currents are not widely known. Document IEC 60479-2 provides information in particular on the variation of the threshold of cardiac fibrillation. This curve (see fig. 33 ) shows that the frequency factor, which is the ratio of the current at frequency f to the current at 50/60 Hz frequency when considering the same physiological effect, increases with the frequency. It therefore permits a higher trip threshold for current frequencies higher than 50 Hz. This threshold variation is achieved technically by filtering. If the impedance of this insulation fault is high, the overcurrent protection threshold may not be reached, and an RCD must therefore be fitted to provide this protection. As already explained, correct operation of an RCD depends on the fault currents passing through its toroidal sensor, thus in this example these currents are not perfectly sinusoidal. The waveform of the zero-sequence fault current can be analyzed by examining the simplified equivalent circuit diagram in figure 34 .
V2
M
V1
Frequency factor F
15
10
Fig. 34 : fault voltage.
5 f (Hz) 100 300 1,000
Voltages V1 and V2 are responsible for circulating any fault current that may occur, as shown in figure 35 .
1 50/60
V1
V2
Fig. 33 : variation of the cardiac fibrillation threshold as a function of the frequency (according to IEC 60479-2).
Shape of the fault current If there is a dead short to earth at the drive output, with a TN system, the overcurrent trips the internal drive protection or the overcurrent protection devices placed upstream.
Fig. 35 : fault current.
The fundamental frequency of voltage V1, between the neutral of the 3-phase supply and the central point of the rectifier, is 150 Hz (see fig. 36 ).
Its shape is shown in figure 38 . This fault current also contains the HF currents described in the preceding sections, but not included here in order to simplify the illustrations.
V 100 80 60 40 20 0 -20 -40 -60 -80 -100 0 0.01 0.02 0.03 0.04 t (s)
A 0.5 0.4 0.3 0.2 0.1 0 -0.1 -0.2 -0.3 -0.4 -0.5 0 0.01 0.02 0.03 0.04 t (s)
Fig. 36 : voltage of the rectifier neutral point with a 3-phase supply.
Fig. 38 : fault current with a 3-phase supply.
Voltage V2 (see fig. 37 ), between the central point of the rectifier and one output phase is the result of PWM. It therefore contains a lowfrequency component equal to the drive output frequency (40 Hz in this example) and a component at the PWM frequency (1 kHz in this example).
As shown in figure 39 , the amplitude of the various components changes as a function of the motor operating frequency: c The total rms value of the current remains constant, as does the 150 Hz component. c Components at the motor supply frequency and the PWM frequency vary in opposite ways.
V 400 300 200 100 0 -100 -200 -300 -400 0 0.01 0.02 0.03 0.04
100 50 0 0 10 20 30 40 50 60 Total rms value Hz mA 300 250
t (s)
200 150
Fig. 37 : output voltage of the inverter stage.
This results in a fault current containing all of the following components: c 150 Hz, c drive output frequency, c modulation frequency, and their harmonics.
Motor frequency component 150 Hz component 1 kHz component
Fig. 39 : evolution of the fault current components.
The shape of the fault current for a single-phase power supply is shown in figure 40 . Note that there is a 50 Hz component and not a 150 Hz component as with a 3-phase supply.
Solution The complex shape of the fault current requires the use of a type A RCD (see fig. 41 ).
0.3 0.2 0.1 0 -0.1 -0.2 -0.3 0 0.01 0.02 t

(s)
Fig. 40 : fault current with single-phase power supply.
Fig. 41 : example of a type A RCD suitable for use with an LV circuit-breaker (C60-300 mA Vigi unit Merlin Gerin).
5.4 Fault at the drive output with an IT system

Rapid fluctuation of the line voltage with respect to earth In an IT system, an earth fault at the drive output does not necessitate tripping, but will cause a rapid fluctuation in the line voltage relative to earth. Unlike the TN system, the line voltage relative to earth is not actually fixed, and will follow the fluctuations set by the PWM. This is shown in the diagram in figure 42 . Any load connected to the line supply is therefore subject to the same fluctuations, including significant voltage gradients (see fig. 43 ). These gradients may result in damage to the capacitive filters connected between the line supply and earth. Solutions The use of EMC filters to improve Electromagnetic Compatibility is not recommended on supplies with IT systems (see standard IEC 61800-3). When it is essential that HF emissions are reduced, a suitable solution is to place an EMC filter with no earth connection at the drive input. To eliminate the phenomenon of rapid voltage fluctuation, it is advisable to install a sinus filter at the drive output. This eliminates any high voltage gradient applied to the motor and the power supply cable.
V2
M
V1 Z
Vz
Fig. 42 : earth fault in IT system.
V 400 300 200 100 0 -100 -200 -300 -400 0 0.01 0.02 0.03 0.04 t (s)
Fig. 43 : fluctuation of the line supply voltage.
5.5 Fault current with DC component

Description Conventional protection devices are suitable for measuring AC fault currents. However, insulation faults on the DC bus of the drive or on the braking energy dissipation circuit (function performed by a resistor which is usually external to the drives) cause the circulation (see fig. 44 ) of a current with a DC component (see fig. 45 with a 3-phase supply, and see fig. 46 with a single-phase supply).
A 0.5 0.4 0.3 0.2 0.1 0 -0.1 t (s) 0 0.02 0.04
M
Rb
Fig. 46 : current in the event of a fault on the braking resistor, for a single-phase supply and a fault resistance of 1 k.
Solution
Fig. 44 : fault between the braking resistor and earth.
A 0.5 0.4 0.3 0.2 0.1 0 -0.1 0 0.01 0.02 0.03 t (s) 0.04
The protection devices must remain operational despite this DC component. If an insulation fault is possible on the DC bus, or on the braking resistor circuit, a type B RCD must be used when the drive has a 3-phase supply. When the drive has a single-phase supply, a type A RCD should be used. Practical rule for using RCDs c In the first situation, with an IT system, illustrated by figure 47 , the fault current has a DC component. The RCDa used to provide protection against direct contact must therefore be sensitive to this type of current. c In the second situation (see fig. 48 ), two RCDs are connected in cascade. If there is a fault on the DC bus, the fault current may not be sufficiently high to trip RCD2. Conversely, this current, which has a DC component, may be sufficiently high to saturate the measurement toroid of RCD1, preventing it from tripping if there is a fault on another feeder.
Fig. 45 : current in the event of a fault on the braking resistor, for a 3-phase supply and a fault resistance of 1 k.
RCDb
RCD1
RCD2
RCDa
Fig. 47 : risk of blinding the RCDa.
Fig. 48 : risk of blinding the RCD1.
c The following rule is therefore used: If the fault current may have a DC component, a type A or B RCD is necessary, depending on the type of power supply. Thus all the RCDs in which this current may circulate should be identical type A or type B RCDs. This condition
is essential in particular in the following situations: v when these RCDs are installed in series, v in IT systems, since the RCDs may be affected by double faults occurring on different feeders.
6 Selection and installation recommendations
Based on the following principles: c for the use of overcurrent protection devices (short-circuit and overload) covered in section 2, c for the protection of persons, covered in section 3, c then the special phenomena described in the previous section,
this section gives practical recommendations that answer the question: how can I provide correct protection for a circuit containing variable speed drives?
6.1 Selection of RCDs (see fig. 49 )

Protection against indirect contact Power supply Hardware and installation characteristics 3-phase No double insulation of the DC bus Double insulation of the DC bus against direct contact Single-phase 3-phase Single-phase If an additional protection measure is necessary, in the event of failure of other measures providing protection against contact or carelessness of users (see installation standards) Type B 30 mA Type A 30 mA
Earthing system TT Type B, low (or IT with frames not sensitivity connected together) (u 300 mA) Earthing system IT
Type A, low sensitivity (u 300 mA)
Earthing system TN-S Type A, low sensitivity (u 300 mA) [*] [*] The insulation fault is similar to a short-circuit. Tripping must normally be performed by the short-circuit protection device, but the use of an RCD is recommended if there is a risk of overcurrent protection devices not tripping.
Fig. 49 : type of RCD depending on the earthing system and the required protection.
Special recommendations: c only connect one drive per RCD, c provide an RCD as an additional protection
measure against direct contact when the braking resistor is accessible.
6.2 Selection of PIMs

DC injection PIMs can be misled by a fault causing a DC voltage between the line supply and earth. Depending on the polarity of this voltage, the insulation level will be falsely increased or decreased. Only AC injection PIMs can therefore be used on supplies without any electrical isolation used to power devices containing DC buses, such as variable speed drives. However, if there is a fault at a frequency converter output, the insulation measurement may be distorted (see fig. 50 ). The converter actually behaves like a voltage source with variable magnitude and frequency. This voltage is added to the measurement voltage injected by the PIM. If the frequency of this voltage is close to the measurement frequency, the measurement is distorted.
M
PIM
Fig. 50 : disturbance of measurement of the PIM.
6.3 Prevention of malfunctions

The disturbance described in the preceding sections may cause undesired tripping of the protection devices. For good continuity of service it is recommended that the following instructions are followed. Precautions concerning RCDs c Choose an appropriate model, which has: v filtering of HF currents, v a time delay (prevents any tripping due to the charging current of the stray capacitance on power-up). It is not possible to have a time delay for 30 mA devices. In this case, choose devices that are immune to undesired tripping, for example high-immunity RCDs in the s.i. range (Merlin Gerin brand). c If possible raise the trip threshold, while keeping to the limit values set for the protection of persons. These precautions are in addition to the rules described in section 5.5 on using RCDs. Precautions concerning PIMs Choose an appropriate model: c AC injection type or c coded pulse type, which makes the drive output frequency irrelevant. Precautions concerning installation Converters must be used in accordance with standards EN 50178 and IEC 61800-3. Certain additional precautions may also be necessary. c Reduce the capacitances to earth as much as possible. To do this: v avoid the use of shielded cables when the operating environment permits, v reduce the length of cable between the drive and the motor, v ensure that the wiring is carried out in line with accepted practice, v avoid the use of EMC filters or use filters with low capacitances (especially in IT systems). c Reduce the PWM frequency (reduction of the number of switching operations per second and therefore reduction of the rms value of the HF currents). c Divide the drives over a number of RCDs (to avoid bringing the leakage currents together). c Place sinus filter at the drive output (elimination of voltage gradients applied to the cables). c Use an isolation transformer and place the RCD upstream (separation of the circuit which is disturbed by the drive from its power supply).
Bibliography
Products standards c IEC 60479: Guide to the effects of current passing through the human body. c IEC 60755: General requirements for residual current operated protective devices. c IEC 60947-2: Low voltage switchgear - Part 2: Circuit-breakers. c IEC 61008: Residual current operated circuit-breakers for household and similar use. c IEC 61009: Residual current operated circuit-breakers with integral overcurrent protection for household and similar use. c IEC 61800-3: Adjustable speed electrical power drive systems - Part 3: EMC product standard. c EN 50178: Electronic equipment used in power installations. c UTE C 60-130: Dispositifs de protection courant diffrentiel rsiduel. c NF C 61-420: Petits disjoncteurs diffrentiels. c NF C 62-411: Matriel de branchement et analogues, disjoncteurs diffrentiels pour tableaux de contrle des installations de premire catgorie.
Installation standards c IEC 60364: Electrical installations of buildings. Schneider Electric Cahiers Techniques c Residual current devices in LV. R. CALVAS, Cahier Technique no. 114. c Protection of people and uninterruptible power supplies. J-N. FIORINA, Cahier Technique no. 129. c Earthing systems in LV. B. LACROIX and R. CALVAS, Cahier Technique no. 172. c Earthing systems worldwide and evolutions. B. LACROIX and R. CALVAS, Cahier Technique no. 173. c Disturbances of electronic systems and earthing systems. R. CALVAS, Cahier Technique no. 177. c The IT earthing system (unearthed neutral) in LV. F. JULLIEN and I. HERITIER, Cahier Technique no. 178. c Cohabitation of high and low currents. R. CALVAS and J. DELABALLE, Cahier Technique no. 187.
Schneider Electric
Direction Scientifique et Technique, Service Communication Technique F-38050 Grenoble cedex 9 Tlcopie : 33 (0)4 76 57 98 60 E-mail : fr-tech-com@mail.schneider.fr
DTP: AXESS - Valence (26). Edition: Schneider Electric Printing: Imprimerie du Pont de Claix - Claix - France - 1500. - 20 -
XXXXX
12-01
Cahiers Techniques is a collection of documents intended for engineers and technicians, people in the industry who are looking for more in-depth information in order to complement that given in product catalogues. Furthermore, Cahiers Techniques are often considered as helpful tools for training courses. They provide knowledge on new technical and technological developments in the electrotechnical field and electronics. They also provide better understanding of various phenomena observed in electrical installations, systems and equipment. Each Cahier Technique provides an in-depth study of a precise subject in the fields of electrical networks, protection devices, monitoring and control and industrial automation systems. The latest publications can be downloaded from the Schneider Electric Internet web site. Code: http://www.schneider-electric.com Section: The experts place Please contact your Schneider Electric representative if you want either a Cahier Technique or the list of available titles. The Cahiers Techniques collection is part of Schneider Electrics Collection Technique.
Foreword The author disclaims all responsibility subsequent to incorrect use of information or diagrams reproduced in this document, and cannot be held responsible for any errors or oversights, or for the consequences of using information or diagrams contained in this document. Reproduction of all or part of a Cahier Technique is authorized with the prior consent of the Scientific and Technical Division. The statement Extracted from Schneider Electric Cahier Technique no. .... (please specify) is compulsory.
no. 201
Jean-Pierre NEREAU An IEG graduate engineer, he joined Merlin Gerin in 1981; during his time in the design office of the Low Voltage Switchgear division, he was involved in developing a number of products, initially the range of moulded case circuit-breakers, then the range of power circuit-breakers. He is currently the Manager of Schneider Electrics Advanced Design Office for this division.
ECT 201 first issued April 2001
Lexicon
Breaking capacity: This is the usual name for the ultimate breaking capacity (Icu). Icu is the highest short-circuit current intensity which the circuit-breaker is able to interrupt. It is defined for a given rated operating voltage Ue. Cascading: Using the limiting capacity of the upstream circuit-breaker to increase the actual breaking capacity of the unit downstream. Enables use of circuit-breakers with a lower breaking capacity than the prospective short-circuit current downstream of a current-limiting circuit-breaker. Current limiting circuit-breaker: Circuit-breaker which, when interrupting a short-circuit current, limits the current to a value considerably less than the prospective current. DIN: Dclencheur INstantan: Instantaneous self-protection release. By assimilation, the corresponding threshold. DINF (or MCR): Dclencheur INstantan la Fermeture or Making Current Release, instantaneous release intended for self-protection of the breaker during the closing operation. Electrodynamic withstand (EDW): Capacity of a unit to tolerate, by nature of its construction, the electrodynamic effects of a short-circuit current, in particular without repulsion of its main or plug-in contacts. Isc: Short-circuit current, given as a peak value, actually crossing the circuit-breaker, taking account of any limitation. Icw: Short-time withstand current. This is the maximum short-circuit current (as an rms value), which the circuit-breaker can withstand for a defined period (0.5 or 1 or 3 s) without alteration of its characteristics.
IDMTL: (Inverse Definite Minimum Time Lag) This refers to long-time delay curves where the slope can take different values (see section on IDMTL trip units). In: Nominal current of the device. Ip: Prospective short-circuit current which would develop in the absence of protective devices (rms value). Ir: Current (as an rms value) corresponding to the overload protection setting. Generally varies from 0.4 to 1 times In. Instantaneous release : Release without intentional time delay (short-circuit protection). Long-time delay release (LT): Release with intentional time delay lasting several seconds (overload protection). This delay is generally dependent on the current. Partial discrimination: Discrimination is said to be partial when it is ensured only up to a current value lower than the prospective short-circuit current. Rating: Current (= In) corresponding to the maximum trip unit setting. Selective circuit-breaker: Circuit-breaker with high Icw (capable of withstanding a short-circuit current for several hundred milliseconds). Sellim: Discrimination principle which allows both discrimination and current limitation. Short-time delay release (ST): Release with an intentional time delay from tens to hundreds of milliseconds. tc: Actual breaking time (arc suppression). Total discrimination: Discrimination is said to be total when it is ensured up to the prospective short-circuit current.

The purpose of this Cahier Technique is to set out the discrimination techniques which apply specifically to low voltage power circuit-breakers. These devices are characterized by their high rating (800 A to 6300 A), and their location at the head of the LV installation, generally directly downstream of an MV/LV transformer. This location justifies the strict discrimination requirements which apply to them. This article begins with a resum of discrimination techniques, followed by an explanation of the links between discrimination and general circuit-breaker characteristics. Finally, some practical examples will be provided on selection of devices to be installed.
Contents
1 LV discrimination 2 Discrimination techniques for short-circuits 1.1 Introduction 1.2 Discrimination according to the type of fault 2.1 Current discrimination 2.2 Time discrimination 2.3 Pseudo-time discrimination 2.4 "SELLIM" or energy-based discrimination 2.5 Zone selective interlocking 2.6 Combining the different types of discrimination 3 Discrimination with power circuit-breakers 3.1 Circuit-breaker characteristics 3.2 Trip unit characteristics 3.3 Discrimination on closing 4 Examples of circuit-breaker selections for an LV installation 4.1 Presentation of the installation concerned 4.2 Dimensioning the protective equipment 4.3 Selecting breakers to ensure discrimination 4.4 Variant with zone selective interlocking 4.5 Variant with two more powerful incoming lines Bibliography p. 4 p. 4 p. 6 p. 6 p. 7 p. 7 p. 7 p. 8 p. 9 p. 12 p. 16 p. 18 p. 19 p. 19 p. 22 p. 23 p. 26
1 LV discrimination
1.1 Introduction
In a radial feeder layout (see fig. 1) the purpose of discrimination is to disconnect only the faulty load or feeder from the network and no others, while keeping as much as possible of the installation powered up. Safety can thus be combined with continuity of service, and the fault easily located. It is an especially important concept for high-power equipment, since this is generally located at the head of the installation and therefore has even greater consequences in the event of false tripping. Discrimination is said to be total if it is assured irrespective of the value of the fault current, up to the maximum value available in the installation. If this is not the case, it is said to be partial. The faults encountered in an installation are of different types: c overload c short-circuit as well as: c earth fault c voltage dip or momentary loss of supply
D1
D3
D2
Fig. 1: Discrimination
For each type of fault there is a specific corresponding protective device (protection against overload, short-circuit or earth fault currents, or against loss of voltage, etc). Each of these faults can cause a loss of discrimination if coordination of the protective devices has not been taken into account.
1.2 Discrimination according to the type of fault

The techniques for using discrimination have to be adapted to the phenomena involved, and therefore differ according to the type of fault. Overloads These are currents between 1 and 10 times the duty current. They should be eliminated within a period which is compatible with the thermal withstand of the conductors concerned. The trip time is generally inversely proportional to the square of the current (this is known as inverse time tripping). Circuit-breaker discrimination works by comparing the time/current curves for the long-time delay releases affected by the fault (see fig. 2) It is effective if, for any overload current value, the time during which the upstream circuit-breaker D1 does not trip is greater than the maximum breaking time for the downstream circuit-breaker D2 (including the arc suppression time). In practice, this condition is achieved if the ratio Ir1/Ir2 is greater than 1.6.
Overloads Short-circuits
tc
Overload discrimination zone D2 D1
I r2
I r1
I ins2
Ip
Fig. 2: Discrimination in the overload zone
Short-circuits Because of the magnitude of short-circuit currents, and especially the presence of electrical arcs which generally accompany them, the circuits concerned should be interrupted almost instantly, in less than a few hundred milliseconds. Discrimination can work, to some extent, by comparing the time/current curves, provided that time tc is at least thirty or forty milliseconds. Below that time, these curves are not sufficiently precise to reach a verdict with certainty. Moreover, the time and current are not then the only discriminating criteria. Depending on the situation, it may be necessary to take account of the peak current, limiting, or a combination of time and current (for example, i2 dt ). It is then necessary to refer to the discrimination tables published by the relevant circuit-breaker manufacturer. Various techniques can be used to achieve discrimination in the event of a short-circuit between 2 circuit-breakers, and these are outlined in the following section. Earth leakage currents Here too, discrimination has to be taken into account so as to prevent an insulation fault at some point in the installation leading to tripping of the main devices. There are 2 major protection families with regard to leakage currents. For low or very low current values (typically between 30 mA and 30 A), a sensor is used which surrounds all the live conductors. This sensor naturally adds up the total current, and provides a signal which is proportional to the fault current. The presence of an earth (or ground) fault current causes the sum of I1+I2+I3+In to be other than zero. This system is generally known as residual protection or vigi. For higher leakage current values, above 20% of the nominal current, one sensor is used per live conductor.
The system, which is called simply ground fault protection, adds up all the signals provided by each of these sensors. In both cases, discrimination works by differentiating between thresholds and time delays. It can be controlled by time/current curves (see fig. 3). Voltage dips or loss of supply These phenomena can be generated by a short-circuit in the installation, or by a fault upstream of it, and lead to tripping of the main devices if they are equipped with an undervoltage trip unit. The solution consists of using time-delayed undervoltage trip units, with a reaction time which is longer than the short-circuit trip time of the equipment located downstream. Even without a time delay, undervoltage trip units should offer immunity against undervoltages lasting approximately ten milliseconds, in order that they are not affected during short-circuits eliminated by equipment located near the loads.
tc D2 D1
t2 t1
I s2
I s2
Ip
Fig. 3: D1 is selective with regard to D2.
2 Discrimination techniques for short-circuits
Improving discrimination generally comes down to restraining tripping by the circuit-breaker concerned as opposed to the circuit-breakers located downstream in the installation. This goal can be achieved by: c Creating a difference between the trip thresholds, which is current discrimination c Delaying - by a few tens or hundreds of milliseconds - tripping of the upstream circuit-breaker, which is time discrimination
c Using a more sophisticated discrimination criterion, for example detection of the number of current waves, or the form of these waves ( idt , 2 i dt , etc), which is Sellim or energybased discrimination c Communicating threshold overshoot information from one circuit-breaker to the other, which is zone selective interlocking
2.1 Current discrimination

This results from the difference between the thresholds of the instantaneous or short-time delay releases of circuit-breakers in series in a circuit. It is applied in the event of short-circuit faults and generally leads, unless associated with another type of discrimination (time, Sellim or energy-based), to partial discrimination limited to the intervention threshold of the upstream device (see fig. 4). Discrimination is ensured if the maximum threshold of the trip unit for the downstream device is less than the minimum threshold of that for the upstream device, including all tolerances.
tc
D2
D1 Discrimination zone for short-circuits
Iins2 Iins1
Discrimination limit for short-circuits
Fig. 4: Current discrimination
Ip
2.2 Time discrimination

To ensure discrimination above the short-time threshold (ICR1) of the upstream device, it is possible to use a time delay, which may or may not be adjustable, on the trip unit for the upstream device D1 (see fig. 5). This solution can only be used if the device can withstand the short-circuit current during this time delay. It therefore only applies to devices with high electrodynamic withstand, which are also called selective. On two circuit-breakers in series, the different time bands, when they exist, are arranged so that they discriminate between one another. The maximum operating time of one band, including the breaking time, should be less than the minimum detection time of the following band.
D1 : Circuit-breaker with short-time bands 0-1-2-3, selective with bands 1, 2 and 3 with D2 D2 : Instantaneous circuitbreaker with Iins2 threshold : Time delay obtained with band 1
tc D2 D1
3 2 1 0
Iins2 ICR1
Fig. 5: Time discrimination
Ip
2.3 Pseudo-time discrimination

If a limiting circuit-breaker is being used downstream, the magnitude and duration of the actual short-circuit current is significantly reduced, especially if the prospective current is high. The trip unit on the upstream device therefore detects a much weaker current than if there is no downstream circuit-breaker. This can be shown on the time/current trip curve for the downstream device by an equivalent time, which diminishes considerably when the prospective short-circuit current increases. The comparison with the detection curve for the device highlights the discrimination between the two devices. It is called pseudo-time, since it does not use an intentional time delay (see fig. 6). This solution, with its limiting effect and the speed with which the fault is eliminated, can also be used to limit the thermal and electrodynamic stresses in the installation.
tc D2 D1 D2 : Current limiting
Ip
Note: If a short-time dependent release (dotted line) is used on D1, discrimination will be much improved.
Fig. 6: Pseudo-time discrimination
2.4 SELLIM or energy-based discrimination

These principles, developed by Schneider Electric, are particularly useful for medium power equipment (100 to 630 A), where current limiting is a necessity. This type of equipment, with very active electrodynamic repulsion, cannot even withstand a delay of a few hundred milliseconds. Time discrimination relating to downstream equipment is therefore unsuitable, or limited to a very low current value. The solution consists of using more sophisticated trip criteria than just the value of the current or time, generally a combination of both these values, for example i2 dt . The type of criterion, and the threshold value, are adapted very precisely to the upstream/downstream combination of equipment under consideration. They can be used to ensure discrimination over several stages, while limiting considerably the thermal and electrodynamic stresses on the installation. This discrimination is used in Merlin Gerins Compact NS circuit-breakers. For a more detailed explanation, the reader may like to refer to Cahier Technique no. 167 entitled Energy-based discrimination for low voltage protective devices.
2.5 Zone selective interlocking

This technique requires data transmission between the trip units of the circuit-breakers at the various levels in the feeder network. The operating principle is simple (see fig. 7): c A trip unit that detects a current greater than its trip threshold sends a logic wait instruction to the trip unit for the circuit-breaker which is next upstream. The time delay will be that displayed on the trip unit. c The trip unit of the circuit-breaker located immediately upstream of the short-circuit does not receive a wait instruction and reacts immediately, regardless of the time delay displayed.
D2 D1 Trip unit Logic wait instruction
Trip unit
Fig. 7: Zone selective interlocking
Zone selective interlocking is a technique used in addition to time discrimination. It is used to reduce fault clearing times, which reduces the stress on the installation. It is applied to high-amp selective LV circuit-breakers, but it is
also used on HV industrial networks. It requires the trip units to be compatible with one another. For further details, see Cahier Technique no. 2 entitled Protection of electrical distribution networks by the logic selectivity system.
2.6 Combining the different types of discrimination

The choice of a type of discrimination in an electrical feeder network depends on the type of device and their location in the installation. Different techniques can be combined between two devices in order to obtain the best availability of electrical energy; see example in figure 8. Current discrimination is, without exception, the first link in the discrimination chain.
Circuit concerned Head of installation Power distribution Final distribution
Type of discrimination Current Time + zone selective int.
Time
Pseudo time
Type of Sellim and circuitenergy-based breaker Selective
Limiting Limiting
Fig. 8: Example of uses for different types of discrimination
3 Discrimination with power circuit-breakers
LV power circuit-breakers, due to their position at the head of the installation, are especially concerned by discrimination requirements. Their natural robust nature means that time discrimination is mainly used in the event of a short-circuit. This does not exclude the additional use of pseudo-time discrimination (current limiting circuit-breaker downstream of a selective circuit-breaker), and zone selective interlocking
(logical wiring between the various levels in the feeder network). In this section we will examine the characteristics which have an influence on this discrimination, first considering those of the circuit-breaker itself, and then those of the trip unit fitted on it. The special case of discrimination on circuit-breaker closing is then analyzed, along with the characteristics which determine it.
3.1 Circuit-breaker characteristics

Short-time withstand current (Icw) The short-time withstand current (Icw) characterizes the capacity of devices to withstand short-circuit currents, which may be very high, for a sufficient period for them to be eliminated by circuit-breakers or protective devices located downstream. It is therefore an essential characteristic for power circuit-breakers which are always found at the head of the installation. The higher the Icw, the higher the usage limit for time discrimination. This is why devices with high Icw are often known as selective devices. It is, of course, essential that the switchboard where the device is installed, and all the conductors located upstream, are capable of withstanding such currents. c Constraints Short-circuit currents generate 2 types of phenomenon: v Electrodynamic forces between the various parts of the circuit conducting the current: These forces may be either repulsion or attraction depending on the respective direction of the currents; they appear instantly, and the resistance of the device to these forces, called electrodynamic withstand (marked EDW) will therefore be characterized by the maximum instantaneous value of the current it can withstand, measured in peak kA. Above this value, parts may be irreversibly deformed, or electrical arcs may be produced which could damage the parts concerned. v Temperature rise in the parts conducting the current: This temperature rise is not a function of the instantaneous value of the current, but of its rms value and its duration; the device withstand can therefore be expressed in kArms and in seconds. The short-time withstand current is defined by a number of standards, including IEC 60947-2 which has allocated it the symbol Icw. The associated test can be used to test the behavior of the device both from the electrodynamic point of view, when the short-circuit occurs, and from the thermal point of view, since the current is maintained for a predefined period (usually 0.5 s, 1 s or 3 s). Since the maximum peak current is fixed by the standard as a function of the rms current, if this is known, the Icw can be defined. It is clear that the Icw is limited by the most severe phenomena, whether electrodynamic or thermal, and its value therefore often diminishes when the associated time increases: an Icw lasting 3 s is thermally 9 times more restrictive than an Icw lasting 1 s. The Icw value to be taken into account for discrimination is that which corresponds to the maximum time setting for the short-time delay release, generally 0.5 s. As this value is usually determined directly by the electrodynamic withstand, the thermal stress is easily controlled. Values at 1 s, or even 3 s, are only an indication of extra robustness in this case. c Construction recommendations to obtain a good Icw All these requirements specify: v Robust and rigid device construction, which holds the current-carrying parts firmly in place; compared to the old construction techniques based on metal parts which were cut, bent and then assembled, the use today of thermosetting polyester moulded cases offers a notable improvement in the structural rigidity of circuit-breakers.
v Excellent rigidity of the mechanism to keep the contacts in the closed position v Special arrangement of the moving contacts and disconnecting contact fingers (see fig. 9) to ensure automatic compensation of the repulsion forces generated between the contact points: - The disconnecting contact fingers are located on either side of the conductors to be linked; the parallel currents circulating in these fingers create an attraction force Fm which compensates the repulsion forces Fr generated at the contacts (fig. 9a). - The moving contacts incorporate a hinge pin located approximately one-third of the distance between the incoming conductors. Therefore, the result of the repulsion forces Fm produced by the current loop creates a torque on the contacts which compensates that generated by the repulsion Fr at the contact points (fig. 9b). Compensation of these forces does however have the effect of increasing the forces transmitted to the mechanism, which constitutes a restriction for the manufacturer. v Generous dimensioning of the power circuit cross-section, so as to avoid reaching an excessive temperature when the time delay on the trip unit is set to maximum v Use of thermosetting moulded materials (with no melting point), or special thermoplastics with a high melting point, near the power circuit Breaking capacity In order to use a circuit-breaker on a given circuit, its ultimate breaking capacity (Icu) must be higher than the prospective short-circuit capacity of this circuit at the point under consideration. Usually, on low voltage power circuit-breakers, this breaking capacity equals the Icw at 0.5 s. In this case, time discrimination can be used up to the breaking capacity, since the device is capable of withstanding these currents for the corresponding time. Total discrimination is therefore achieved. However, the values of Icw obtained, even with the best types of construction, are at present typically limited to around 85 kArms, which therefore limits the breaking capacities. Yet an increasing number of installations can generate short-circuit currents above this value, reaching 150 kA in some cases, or even more. This is particularly the case for installations incorporating several high-power transformers in parallel, or networks looped with multiple generators. There is now a response to this need, with circuit-breakers which have a breaking capacity higher than the Icw.
a Fr i Fm Fr (i/2) Fr (i/2) Fr
b i 1/3 A 2/3 i Fm Fr
Fig. 9: Construction recommendations to ensure compensation of repulsion forces in a circuit-breaker
a: Asymmetrical energization
250 200 150 100 50 0 -50 -100 -150 0 10 20 30 40 50
Iasym Isym Irms
b: Symmetrical energization
200 150 100 50 0 -50 -100 -150 -200 0 10 20 30 40 50
Isym Irms
Fig. 10: Timing diagram of the current in the event of asymmetrical or symmetrical energization
c Devices with breaking capacity higher than the Icw, construction requirements For their own protection, circuit-breakers with a breaking capacity higher than the Icw/0.5 s require instantaneous tripping (DIN) as soon as the current exceeds their electrodynamic withstand, as they cannot withstand arcs of such intensity produced on contact repulsion for several hundred milliseconds. However, this condition is not sufficient of itself, and controlling the breaking capacity of these devices requires the expertise of specialists in interruption of electrical arcs. In particular, as these devices are not limiting, the current on breaking powerful short-circuits can reach approximately 2.3 times the rms value of the prospective current in asymmetrical operation, ie. 230 kApeak in the case of a prospective short-circuit of 100 kArms (see fig. 10). The electrodynamic constraints are therefore significant on the device itself, with the consequences being amplified by effective opening of the device at the very moment when these stresses are strongest. These considerations limit the maximum breaking capacity that can be obtained with devices with high Icw, and only an extremely robust construction combined with an exceptional ability to control phenomena associated with breaking high currents enables values higher than 100 kA rms to be obtained. Merlin Gerins type H3 Masterpact NW devices, which offer a breaking capacity of 150 kA at 440 V, for an Icw of 65 kA/3 s are an excellent illustration of this expertise. Note that in this case, the withstand of the switchboard and the installation also requires very robust construction of the busbar sets and their supports. The use of factory-produced LV switchboards, tested to standard IEC 439, ensures the reliability of this construction (see Cahier Technique no. 162). c Current limiting devices When, on devices with high Icw, the maximum breaking capacity indicated by the manufacturer is insufficient, the only remaining option is to use current limiting circuit-breakers, which usually have breaking capacities of as much as 150 kA at 400 V. By their very nature, these devices limit the maximum value reached by the current, and provide a high breaking capacity, while reducing the effects of the short-circuit on the installation and the device itself. High-rated current limiting circuit-breakers do however suffer from a handicap, with regard to discrimination with devices located downstream, as their EDW is always relatively low. In fact,
current limiting is usually obtained by using an electrodynamic contact repulsion effect, which conflicts directly with the EDW. The threshold for the instantaneous self-protection release (DIN) should therefore be set very low, which restricts discrimination with downstream equipment to low values, unless more sophisticated trip criteria are used (see Cahier Technique no. 167, Energy-based discrimination for low voltage protective devices). Here too, it is the clever design of current limiting power circuit-breakers that enables manufacturers to offer high breaking capacity and effective current limiting, while still ensuring good EDW. This is particularly the case with Merlin Gerins Masterpact NW current limiting devices, which have an EDW of as high as 37 kArms! This EDW would never be as high, however, as a non-limiting device. Thus, the maximum breaking capacity of devices with high Icw, by avoiding the need to use current limiting devices at the head of the installation, is a fundamental element of discrimination. Current limiting The instantaneous value of a sinusoidal alternating current, in steady state, oscillates between + r and - r times its rms value. During energization, this instantaneous value can reach approximately 2.3 times the rms value on the first wave, due to the asymmetry of the current. The actual value depends on the circuit inductance; in practice, it is also related to the level of short-circuit in question, and increases with it. If the incoming circuit-breaker is fitted with an instantaneous self-protection release (DIN), because its breaking capacity is higher than its Icw, discrimination with the downstream device is limited by the presence of this instantaneous
2.5 2 1.7 1.5 1 0.5 0 6 to 10 10 to 20 20 to 50 2.04 2.16
2.29
> 50
Prospective Isc in kA
Fig. 11: Asymmetry coefficients as a function of the prospective rms current acc. to standard IEC 60947-1
release. If the value of its threshold is known (in kApeak), this value simply has to be divided by the asymmetry coefficient (see fig. 11) to find out the discrimination limit (in kArms). However, if the device located downstream of the circuit-breaker in question is a current limiting type, and if the short-circuit occurs downstream of this current limiting device, the maximum instantaneous value mentioned earlier will not be achieved. In this case, the discrimination limit obtained is increased, even more so if the circuit-breaker downstream has a high current limiting capacity (pseudo-time discrimination). In extreme circumstances, if the maximum current limited by the downstream circuit-breaker is less than the instantaneous threshold of the upstream device, there is total discrimination between the two devices (see fig. 12).
Limited current
Upstream device threshold
Prospective Isc Discrimination limit with non-limiting device Discrimination limit with limiting device TOTAL discrimination with very limiting device
Fig. 12: Partial or total discrimination obtained between two devices, depending on the current limiting capacity of the downstream device
3.2 Trip unit characteristics

The discrimination potential of a device can only be fully exploited by using an appropriate trip unit.
t
Types of trip unit On high-rated equipment, trip units are exclusively electronic nowadays. There are a number of different types, with different setting options (see fig. 13). c Simple trip units These usually offer an inverse time curve with an adjustable threshold, for overload protection, and an instantaneous trip threshold (< 10 ms), also adjustable, for short-circuit protection. This instantaneous threshold generally has a maximum value of 10 to 12 In. It is this maximum value which limits the actual discrimination which can be obtained using this trip unit. c Selective trip units These offer, in addition to the overload and short-circuit protection described above, a trip threshold with time delay, where both the threshold and the time delay can be adjusted (from 0 to 500 ms), and an instantaneous
Instantaneous 0.5 s Long time Short time
20 ms Long-time threshold DIN Short-time Instant. threshold threshold threshold
Fig. 13: Trip curve for a circuit-breaker, illustrating the setting parameters
release which can be adjusted up to the maximum permissible value for the breaker.
If the circuit-breaker Icw equals its breaking capacity, this maximum value can be infinite, equivalent to the Off position: instantaneous tripping will never occur - see section on Breaking capacity). In this case, discrimination is then total, otherwise the actual discrimination is limited by the value of the instantaneous threshold set as for a simple trip unit. If the Icw is less than the breaking capacity, this instantaneous threshold can nonetheless be very high (much greater than 12 In) when the EDW is high (see section on Breaking capacity). Discrimination is then partial, up to the rms current corresponding to this instantaneous threshold, or even total if the downstream protective device is sufficiently limiting for this value never to be reached (see section on Current limiting). Below this threshold, time discrimination has to be used, for example: a 3rd level device is time-delayed by 100 ms, a level 2 device by 200 ms, and a level 1 device by 300 ms. c Trip units with zone selective interlocking. A hard-wired link connects the circuit-breaker trip units on a single circuit. A trip unit which detects a short-circuit sends a time delay command to the upstream trip unit. This trips instantaneously above its short-time threshold (whatever its time delay setting), if it has not received a wait command from downstream. This function does not modify the rules for discrimination, but it reduces the stresses on the installation since the circuit-breaker immediately upstream of the fault will always trip instantaneously. c Trip units with IDMTL curves In a very different area from the previous considerations, which concern circuit-breaker discrimination in short-circuit situations, some top-of-the-range trip units offer trip curves known as IDMTL, as defined by standard IEC 60255-3. These curves can be used to improve circuit-breaker discrimination in the area of overloads, where discrimination can easily be studied by comparing the trip curves for the upstream and downstream protective devices (see fig. 14). With these trip units it is possible to set not only the threshold and time delay for the long-time delay release, but also the slope of the trip time as a function of the current. As standard, this slope is at I2 t = constant (the trip time is inversely proportional to the square of the current) and it offers constant thermal stress protection.
IDMTL curves permit different trip times, as required by the user: v Constant (t = constant; DT = Definite Time) v Inversely proportional to the current (I t = constant; VIT = Very Inverse Time) v Inversely proportional to the square of the current (I2 t = constant; EIT = Extremely Inverse Time) v Inversely proportional to power 4 of the current (I4 t = constant; HVF = High Voltage Fuse) This offers improved discrimination, especially with medium-voltage circuit-breakers located upstream, which often have constant trip times, or with medium-voltage fuses, which have a slope higher than I2t (see section 4.3). The self-protection release function As we saw earlier (see section on Current limiting devices), a circuit-breaker whose breaking capacity is higher than the Icw needs to have an instantaneous release (DIN) for its own protection. c Standard DIN The DIN threshold should be selected so that, even in the least favorable withstand conditions, it is still less than the ultimate circuit-breaker withstand. In particular, the tolerance of the current measurement system should be taken into account. If there is a wide degree of tolerance, the nominal threshold must be reduced by the same amount. However, if this system is very precise,
I4t constant I2t constant It constant

t constant
tr Short time
Ir
6Ir
Fig. 14: Circuit-breaker IDMTL trip curve
the nominal threshold can be set nearer the limit withstand value for the device (see fig. 15).
With a trip unit which is sensitive to the current differential coefficient, the trip command can be given instantaneously, although still with a very short time delay in order to avoid false tripping due to interference. v Effect on discrimination This type of self-protection release does, however, behave in a particular way in terms of discrimination. In fact, even a very limiting device placed downstream of this device has no immediate effect on the differential coefficient for the current of a fault that it detects: some time, however minimal, is necessary for its contacts to open and for the arc voltage generated to slow down the current rise, before stopping it altogether. In this case, discrimination will therefore be limited by the threshold value of the current differential coefficient, irrespective of the downstream protective device. It is therefore vital for discrimination that the manufacturer sets this threshold at the highest possible value, compatible with the desired current limiting and the device electrodynamic withstand. In the previous example, if the threshold is set at a value of 44.3 kA/ms, corresponding to a prospective current of 100 kArms at 50 Hz, current limiting only comes into effect above this prospective current value and then discrimination will occur with downstream devices up to this same value. v For 60 Hz: di/dtmax = 0.531 I rms so a threshold set at 44.3 kA/ms corresponds to a limit of 83 kArms (instead of 100 kArms at 50 Hz). c Contact separation detector One way of completely eliminating current sensor inaccuracy is not to use them. Using photoelectric light sensors to detect, between the
Variation of EDW
Prospective current
Variation of EDW Precise DIN
Imprecise DIN
Discrimination zone
Discrimination zone
Fig. 15: Effect of the accuracy of a circuit-breakers current measurement system on discrimination
c DIN with di/dt To improve breaking performance, and obtain a certain degree of short-circuit current limiting on non-limiting devices, a self-protection release can be used which is not based on the instantaneous current value, but on its differential coefficient (di/dt). v Principle At known frequency, the maximum differential coefficient of the current is in fact directly linked to its rms value by the equation: di/dtmax = Irms 2 2 f, where f is the network frequency, which gives di/dtmax = 0.443 I rms at 50 Hz (differential coefficient in kA/ms if Irms is in kA) di/dtmax = 0.531 I rms at 60 Hz The least favorable case in terms of speed of establishing the short-circuit current consists of a symmetrical wave, which develops in the form of a sinusoidal equation: Irms 2 sin(2 f t) (see fig. 16) To limit the maximum current produced by this type of wave, it is essential to act extremely quickly. The current differential coefficient offers this opportunity, since its maximum value, in this case, is reached as soon as the short-circuit is initiated, while the value of the current instantaneous threshold may only be reached a few milliseconds later. Thus, for a short-circuit of 100 kArms at 50 Hz, the symmetrical wave will generate a maximum current of 140 kApeak at the end of 5 ms (see fig. 16). With a self-protection release based on an instantaneous threshold value of 100 kApeak, it is necessary to wait approximately 2.5 ms before reaching the threshold. Too little time then remains to limit the current in any significant fashion.
Current
140
100
Limited wave di/dt 0 2.5 5 7.5 10 Time
Fig. 16: Detection principle for a short-circuit, based on the current instantaneous differential coefficient and limiting obtained
contacts of each of the poles, the appearance of an arc when the electrodynamic withstand is exceeded, is one technique for commanding the circuit-breaker to open. In this case, discrimination is no longer affected by the tolerance of the current sensor, but only by that of the electrodynamic withstand itself. In practice, these complex devices can only be justified to remedy the inaccuracy of conventional magnetic sensors on high currents. Current sensors The accuracy of the measurement system depends on the sensor accuracy. Two major families of sensors are used on circuit-breakers: v Magnetic circuit current transformers v Current transformers with non-magnetic toroids c Magnetic circuit current transformers v Their technology This is the oldest type and offers satisfactory accuracy for equipment with a low EDW. These transformers are fitted with a secondary winding with n turns around a core of magnetic material, with the passage of the main conductor across the magnetic circuit constituting the primary (see fig. 17).
Secondary current Is
Saturation
Tolerance on Is
Tolerance on Ip
Primary current Ip
Fig. 18: Variation in the secondary current of a transformer as a function of the primary current (effect of saturation of the magnetic circuit)
Is = Ip / n
Moreover, when the circuit-breaker closes due to a short-circuit, the response of the sensor on the first current wave depends to a large extent on the magnetic state (remanent induction) in which it was left by the previous current. If the primary current on closing is in the same direction as the previous current, the secondary current Is is noticeably attenuated on the first wave; if it is in the opposite direction, Is is increased. Consequently, the measurement system may be marred by a significant error, the DIN threshold therefore needs to be set well below the EDW value by the manufacturer. c Current transformers with non-magnetic toroids v A new technology (see fig. 19)
n turns
us = k dip / dt
Ip
Magnetic circuit
Fig. 17: Schematic diagram of a current transformer with magnetic circuit
This transformer draws a current (Is) from the secondary equal to the primary current (Ip) divided by the number of turns (n) in the secondary. v Their accuracy Accuracy is satisfactory as long as the magnetic circuit is not saturated, ie. up to 5 to 10 times the nominal current. Thereafter, the secondary current is significantly weaker than Ip/n (see fig. 18).
ip
plastic support
Fig. 19: Schematic diagram of a current transformer with non-magnetic toroid
These current transformers, or Rogowski toroids, consist of a secondary winding wound onto non-magnetic material surrounding the primary circuit. They provide the secondary with a voltage proportional to the variation of the primary current. The integration of this voltage by the electronic circuits gives an image of the primary current. v Their accuracy The absence of a magnetic circuit gives these sensors perfect linearity for all current values. They enable optimum usage of circuit-breakers. The accuracy of these sensors means that the trip threshold can be set as close as possible to the limit value of the device electrodynamic withstand (EDW), and the discrimination limit is correspondingly increased.
The downside of this characteristic is: - the low value of the voltage delivered - the fact that this sensor delivers no power - the sensitivity of the secondary signal to the toroid dimensions. These various points can be solved respectively by: - careful signal processing - installation of a second, magnetic, sensor which delivers the power required to supply the trip unit electronics - controlling the toroid dimensions by the use of suitable materials, ie. which are stable, not sensitive to temperature and reproducible
3.3 Discrimination on closing

Risks associated with switch-on-to-fault When a device closes, the mechanism should supply the necessary power for contact operation, and in particular for compression of the springs which provide the bearing force for the moving contacts on the fixed contacts. It is this force which ensures that the current flows correctly through the mechanism, without causing an excessive temperature rise. When the device closes on a normal or overload current, the above conditions are not significantly altered. However, when the device closes on a short-circuit current, considerable electrodynamic forces are generated between the contacts even before the mechanism closes completely, and may lead to closing being prevented, and then unwarranted reopening. This situation should be avoided, otherwise the device may be rapidly destroyed by an uninterrupted series of make and break attempts, without intervention by the trip unit. Necessity of distinguishing between instances of closing on a normal current, or a short-circuit current There is therefore a clear difference between the current which the device can withstand when it is closed (electrodynamic withstand), and the current which the device can fully make (make capacity), also known as close & latch. It is possible to control the value of the current which the device is capable of fully making, by controlling the power of the control mechanism. By increasing this power, the limit current is also increased. However, as this additional power is not consumed to overcome electrodynamic forces during operations where there is no current or with normal currents, it is dissipated by jolts in the mechanism. This power cannot therefore be increased with impunity without compromising the endurance of the mechanism, a value which is essential to the user as it determines the service life of the device. The 2-step release solution There is a solution which enables a device to be used on circuits where the current can reach values higher than its make capacity. It consists of tripping the device if the current exceeds this capacity on closing the circuit. Opening then occurs in controlled conditions which do not lead to any particular difficulties. Of course, since this make capacity is less than its electrodynamic withstand, it is not desirable to simply have an instantaneous release with a threshold lower than this capacity: the whole point of high electrodynamic withstand would then be lost. It is therefore necessary to have a two-step instantaneous release; one low step, which is only active on closing (called DINF), the other high, which is active when the device is completely closed (DIN). This system can be used in two ways: c The first solution, which is widely used, consists of activating the low threshold for thirty or forty milliseconds after the trip unit detects a current. This solution is easy to use, as it only concerns the trip unit, and can therefore be implemented entirely electronically. It does, however, have a major disadvantage: it is not possible to distinguish between an open device which is closing, and a device which, having been closed with no current or with a very weak current, is suddenly affected by a short-circuit current. This is what happens with a closed incoming circuit-breaker, with no current, when one of the downstream feeder circuit-breakers is
closed due to a short-circuit. In this case, the DINF of the first circuit-breaker is activated unnecessarily and adversely affects discrimination, whereas the device would have been sufficiently protected by the DIN threshold. c A second more satisfactory solution consists of detecting a closure movement by the device, delaying this information for as long as necessary to ensure that the device has closed completely, and to use this information in the form of an electrical contact to switch the trip unit from the DINF state to the DIN state. This solution ensures that the low threshold is only activated at an appropriate moment, and does not reduce discrimination unnecessarily for a device which is already closed.
Advantage of discrimination in the event of closing on short-circuit Finally, remember that when a circuit-breaker closes, the loss of discrimination caused by DINF protection is of limited consequence, since the device tripping is not likely to switch off part of the installation which would have been supplied with power previously. Nonetheless, discrimination is still useful since it makes it possible, at least up to the DINF threshold, to close the upstream device, and to allow the downstream device affected by the fault to trip, thus making it easier to locate the short-circuit.
4 Examples of circuit-breaker selections for an LV installation
4.1 Presentation of the installation

The LV installation which forms the subject of this study is shown in figure 20. This study includes coordination of protection between the LV equipment and the protection located upstream of each MV/LV power supply transformer. The selections refer to Merlin Gerin products. The installation includes 2 medium voltage 20 kV incoming lines protected by a fuse, each equipped with an MV/LV transformer with characteristics 20 kV/410 V, 1600 kVA, and an incoming LV circuit-breaker (A) or (B). A section switch (C) can be used to operate both parts of the installation together or separately, in order to optimize availability of power in the event of failure of one of the two incoming lines.
20 kV
80 A MV fuse
80 A MV fuse
20 kV / 410 V 1600 kVA In 2253 A Isc 36 kA Main LV switchboard A NW25H1
4'
20 kV / 410 V 1600 kVA In 2253 A Isc 36 kA B NW25H1
Isc 72 kA
D NT08L1 700 A Cable
C NW25H1
3'
Isc 72 kA
E
2'
NW10H2 750 A
1'
Cable
Isc 50 kA
F NS250H 185 A
Isc 50 kA
G NS400H 330 A
Fig. 20: 1st installation example (with 1600 kVA MV/LV transformers) with indication of the order in which discrimination is examined
4.2 Dimensioning the protective equipment

Rating of devices (A) and (B) installed on LV incoming lines Determining the nominal current for the LV incoming lines: 1600 kVA at 410 V corresponds to a nominal current of 1,600,000 / 410 x 3 = 2253 A. Incoming devices with a rating of 2500 A are therefore chosen. When the section switch is closed, downstream of devices (A) and (B) and ignoring the busbar impedances, the short-circuit current is 2 x 36 = 72 kArms . Given the cable impedances, the short-circuit current crossing the circuit-breakers located at (F) and (G) is no more than approximately 50 kA. c Selecting breaking capacity The breaking capacity required for each device must be determined according to the short-circuit current values at different points in the installation. Circuit-breakers (D) and (E) must have a breaking capacity higher than 72 kA, while for circuit-breakers (A), (B) and (C), a breaking capacity higher than 36 kA is adequate. Circuit-breakers (F) and (G) must have a breaking capacity of 50 kA minimum.
Rating of fuses installed on MV incoming lines The nominal current for the MV incoming lines is:
In = 1,600,000 / 20 000 x 3 = 46 A
Based on the manufacturers selection tables, fuses with a rating of 80 A are therefore chosen (in order to take account of inrush and overload currents, while providing thermal protection for the transformer). Breaking capacity for the various devices c Determining the short-circuit currents at different points in the installation Each transformer has a short-circuit current Isc equal to 36 kA (current linked to the transformer power and short-circuit voltage).
4.3 Selecting devices to ensure discrimination

Principle Discrimination is determined by comparing the characteristics of each circuit-breaker with those of the protective device (circuit-breaker or fuse) located immediately upstream. The circuit-breakers located furthest downstream in the installation should be selected and set in order to trip as quickly as possible, so as to limit the stresses on the installation in the event of an overload. Once the characteristics of these circuit-breakers have been established, one can work back up the installation, ensuring discrimination between circuit-breakers 2 by 2 (downstream circuit-breaker/upstream circuit-breaker). Discrimination between circuit-breakers (F) and (D) 1 c At F: In = 185 A; Isc = 50 kA A circuit-breaker with a rating of 250 A is suitable, for example a Compact NS 250 H (breaking capacity 70 kA at 415 V). c At D: In = 700 A; Isc = 72 kA A circuit-breaker with a rating of 800 A is suitable, for example a Compact NS 800 L or a Masterpact NT 08 L1 (breaking capacity 150 kA at 415 V). c Discrimination mechanism Device (F) is very limiting (the maximum current which can cross it is 22 kApeak for a prospective short-circuit of 50 kArms), and this circuit-breaker therefore allows pseudo-time discrimination with circuit-breaker (D). This discrimination is improved by applying the SELLIM principle to circuit-breaker (D). This device, which is also limiting (with low EDW to ensure very good current limiting), enables total discrimination between (F) and (D) since, according to the SELLIM principle of discrimination, the device on (D) does not trip on the 1st current wave. Note: The SELLIM function is systematically included in Micrologic - Merlin Gerin trip units, and automatically activated on the devices concerned. Discrimination between circuit-breakers (G) and (E) 1 c At G: In = 330 A; Isc = 50 kA A circuit-breaker with a rating of 400 A is suitable, for example a Compact NS 400 H (breaking capacity 70 kA at 415 V). c At E: In = 750 A; Isc = 72 kA The same circuit-breaker (current limiting) can be used as for (D), but since the NS 400 H current limiting is weaker than on the NS 250 H, this combination will not be totally discriminating. To achieve this discrimination, a selective circuit-breaker must be selected, for example Masterpact NW 10 H2 (In 1000 A, breaking capacity 100 kA at 415 V, Icw = 85 kArms /1 s).
In addition, the current limiting power of device (G) enables, if necessary, pseudo-time discrimination. c Discrimination mechanism Since the Icw (85 kA) is less than the breaking capacity (100 kA), this device has an instantaneous self-protection release (DIN) with a threshold of 170 kApeak. With an Isc = 72 kArms, the maximum current at (E) is 72 x 2.3 = 165 kApeak. Since the DIN threshold is therefore never reached, no trip will be generated which would hinder discrimination. Moreover, in the event of a short-circuit at (G), the maximum current, which corresponds to an Isc of 50 kA, will be limited for (G) to 30 kApeak! Discrimination will therefore be total, as long as device (E) is fitted with a trip unit with an instantaneous threshold higher than 30 kApeak, say 30/r = 21 kArms = 21 In, and that the short-time release delay is set on the 0.1 s band. c Variant A current limiting device can also be used at (E), with a better EDW than (D), for example an NW 10 L1 (In 1000 A, breaking capacity 150 kA at 415 V, Icw 30 kA/1s). Because of its current limiting (125 kApeak to 72 kArms, against 165 kApeak with no current limiting), the choice of this type of circuit-breaker considerably reduces the electrodynamic stresses on the cables between (E) and (G). This circuit-breaker is fitted with an instantaneous self-protection release 80 kApeak, which is never therefore called on if there is a fault downstream of (G) (Isc limited to 30 kApeak). This also provides total discrimination, of the pseudo-time type, due to the current limiting of the device downstream. Note: A non-limiting device at (G) would allow a peak current of 50 kA x 2.3 = 115 kApeak to pass in the event of a short-circuit, which would cause circuit-breaker (E) to trip. Discrimination between circuit-breakers (E) and (C) 2 This discrimination is not essential if both incoming lines are operational (since opening of the section switch does not interrupt the power supply via (A) and (B)). Conversely, it is essential if incoming line (B) is non-operational. c Value of the nominal current In at (C): To offer the maximum flexibility, the section switch devices have identical dimensions to the incoming devices, ie. In = 2500 A. As Isc = 36 kA, a selective circuit-breaker placed at (C) allows time discrimination with (E) and even more with (D) which is current limiting, for example a Masterpact NW 25 H1 (In 2500 A, breaking capacity 65 kA at 415 V, Icw 65 kA/1 s).
c Reason for this selection Since the Icw for the device equals the breaking capacity, it does not incorporate an instantaneous self-protection release; time discrimination can therefore be applied without restriction up to the breaking capacity. The circuit-breaker (C) must therefore be fitted with a selective trip unit, with its instantaneous release set to the Off position, and the short-time delay on the 0.2 s band (since the short-time release delay of circuit-breaker (E) is set on the 0.1 s band). Discrimination between circuit-breakers (D) and (C) 2 The solution chosen for discrimination between (E) and (C) is also suitable between (D) and (C) since (C) is totally discriminating up to its breaking capacity. Discrimination between circuit-breakers (C) and (B) or circuit-breakers (C) and (A) 3 3 (A) and (B) are selective devices, without a self-protection instantaneous release. Here too, time discrimination applies up to the breaking capacity, with for (A) and (B): their instantaneous release set to the Off position and their short-time delay set on the 0.3 s band (since the short-time release delay of circuit-breaker (C) is set on the 0.2 s band). Discrimination between circuit-breakers (A) or (B) and MV fuses 4 4 To analyze this discrimination, we need to compare the trip curves for LV circuit-breakers and MV fuses. To do this, transpose the MV fuse curve to LV, by multiplying the current scale by the transformer ratio, or here 20,000/410 = 48.8 (see fig. 21). Discrimination is considered with 2 types of trip unit: a standard selective trip unit, and a trip unit with IDMTL curves. c Settings for standard selective trip units v Long-time threshold No problem, the non-tripping limit current for the fuse is well above the limit current for circuit-breaker tripping; the long-time threshold can therefore be set to maximum (ie. Ir = In). v Long-time delay and short-time threshold The blowing characteristic for MV fuses has a much steeper slope than that of the long-time delay release (LT) tripping, with a slope of I2t (see fig. 21). To avoid the curves intersecting, the long-time delay (tr), or short-time threshold (Isd), must be set to sufficiently low values. A good compromise in this example consists of setting tr = 12 s (at 6 Ir, in a range generally going from 1 to 24 s), and Isd = 4 Ir (in a range from 1.5 to 10 Ir).
Td (s)
10000
HVF curve tr = 2 s 1000
I2t curve
100 tr = 12 s
MV 80 A fuse corrected to LV
Max. Isc 36 kA 10
Short-time threshold = 4 Ir 1
Short-time band 0.3 s
0.1
0.01 0.1
10
100 I/In
Fig. 21: Analysis of discrimination between an LV circuit-breaker and MV fuses - applied to the example of the installation concerned
These values allow the passage of peak currents at switch-on or starting currents for loads located downstream without false tripping; a detailed study needs to be undertaken on these loads. With a higher short-time delay threshold, 5 Ir for example, tr should be reduced to 4 s. v Short-time delay As the short-time delay is set on the 0.3 s band, to ensure discrimination with the devices downstream as indicated above, the fuse and circuit-breaker curves cross at around 10 In (see fig. 21). Discrimination between the circuit-breaker and the MV fuse is therefore limited to approx. 25 kArms, for a maximum short-circuit current Isc of 36 kArms.
c Settings for trip units with IDMTL curves (see section Trip units with IDMTL curves) With these trip units, it is possible to select the slope of the long-time curve. In this case, we can opt for the HVF (High Voltage Fuse) slope, which is the closest to that for the fuse (slope of I4t). With a delay at 6 Ir of 2s, better immunity to high transient currents (peak currents at switch-on or starting) is possible, in the zone for currents between 5 and 10 Ir, since the short-time threshold can be set at any desired value up to 10 Ir (see fig. 21).
4.4 Variant with zone selective interlocking

This variant requires, for the relevant circuitbreakers, trip units with this function (type Micrologic 5.0 A - Merlin Gerin). Description The principle and operation of zone selective interlocking are explained in a Cahier Technique dedicated to this type of discrimination. Remember that each trip unit has four terminals: c 2 input terminals, for connection to downstream equipment c 2 output terminals, for connection to upstream equipment When a trip unit detects a fault above its short-time threshold, it short-circuits both its output terminals. When a trip unit has a short-circuit on both its input terminals, it activates the short-time delay. Otherwise, it trips instantly. Implementation in this example c The first devices (D) and (E) have a permanent short-circuit on their inputs, so that their short-time delay is activated. This ensures discrimination with the stage below (Compact NS circuit-breakers). c Next the wiring is effected and the short-time delays set according to figure 22. Direct wiring between (E) and (B) on the one hand, (D) and (A) on the other hand, is a means of ensuring discrimination between these devices when the section switch (C) is open. In this case the diodes ensure the independence of both halves of the installation: they avoid (D) acting on (B) and (E) acting on (A). Operation c In the event of a fault downstream of (G) (see fig. 20): v (G) trips instantly. v (E) is delayed for 100 ms and does not therefore trip, but sends a signal to (C). v (C) is then delayed for 200 ms and does not therefore trip, but retransmits a signal to (A) and (B) which are then delayed like (C). v Hence only (G) will trip. c In the event of a fault between (G) and (E): v (E) trips after 100 ms and sends a signal to (B) and (C) which are then delayed for 200 ms and therefore do not trip. v (C) retransmits the signal to (A) which is then delayed like (C). If (C) is open, it does not send a signal to (A), which is of no importance since the short-circuit supplied by (B) does not affect (A). c In the event of a fault between (E) and (C): v If (C) is closed, power is supplied to the fault by both incoming lines in parallel.
OUT A ST band 2 IN
OUT B ST band 2 IN
OUT C ST band 2 IN
OUT Other feeders D ST band 1 IN
OUT E ST band 1 IN Other feeders
Fig. 22: Use of zone selective interlocking - illustration of time delay settings and trip unit wiring
- (C) trips instantly and send a signal to (A) and (B) which are then delayed by 200 ms and therefore do not trip. - (C) then interrupts the current supplied by the incoming line (A). - (A) stays closed and maintains the power supply to the part of the installation downstream of its busbar. - (B) interrupts the fault current after 200 ms. v If (C) is open, it does not send a signal and (B) trips instantly. c In the event of a fault between (C) and (B): (B) trips instantly. NB: A fault between (E) and (C) or between (C) and (B) is extremely unlikely, as these devices are generally located in the same switchboard.
Zone selective interlocking, by reducing or even eliminating the always considerable time delays at the head of the installation, can thus be used to limit stresses on the installation even more effectively when the feeders are close to the network. Therefore, using this technique in this installation, a fault immediately downstream of (A) or (B) is eliminated within tens of milliseconds instead of taking longer than 300 ms. In addition, since device tripping is instantaneous in nearly all instances, discrimination with the MV fuses is total, whereas using time discrimination it was limited to 25 kArms due to the 300 ms short-time delay of devices (A) and (B).
4.5 Variant with two more powerful incoming lines

This is the same diagram as before, except for the following: c The transformer power has been increased to 2500 kVA, and the current on the outgoing circuit at (E) increased to 2200 A. c MV protection is provided by MV circuitbreakers (see fig. 23). The nominal current at (A) and (B) becomes 3520 A and the Isc 54 kA. On the main busbar, Isc becomes 108 kA.
20 kV
MV circuit-breaker
MV circuit-breaker
20 kV / 410 V 2500 kVA In 3520 A Isc 54 kA Main LV switchboard A NW40H1
4'
20 kV / 410 V 2500 kVA In 3520 A Isc 54 kA B NW40H1
Isc 108 kA
D NT08L1 700 A Cable
C NW40H1
3'
Isc 108 kA
E
2'
NW25H3 2200 A
1'
Cable
Isc 78 kA
F NS250H 185 A
Isc 95 kA
G NS400H 330 A
Fig. 23: 2nd installation example (MV/LV transformer power 2500 kVA)
Selecting the MV circuit-breaker For an MV voltage of 20 kV, it is possible to use a Merlin Gerin MC-Set circuit-breaker, with a transformer type Sepam protective relay. This relay has two trip thresholds (see fig. 24): c The first provides protection in the event of a short-circuit between the transformer and the LV circuit-breaker, or in the event of failure of the LV protection. c The second provides protection in the event of a short-circuit upstream of the transformer. Selecting the LV circuit-breakers c Circuit-breaker (E) Since the Isc is higher than 100 kA, it is not possible to use an NW 25H2 (breaking capacity 100 kA). Nor is it possible to use a current limiting circuit-breaker, since the nominal current does not exceed 2000 A (NW 20L1). The solution lies in selecting a selective circuit-breaker with high breaking capacity, such as the NW 25H3, which offers a breaking capacity of 150 kA, with an Icw of 65 kA/3 s.
c Circuit-breakers (A), (B) and (C) For In = 3520 A, NW 40H1 type circuit-breakers (In 4000 A, breaking capacity 65 kA, Icw 65 kA/1 s) is chosen. Setting devices to ensure discrimination On the basis of the rules defined in section 4.3, the only modifications to be made are for: 1 the delay for device (E) to be set on the 0.1 s band 2 device (C) on the 0.2 s band 3 3 devices (A) and (B) on the 0.3 s band If zone selective interlocking is being used, the diagram in figure 22 is still valid. Setting the protective relay for the MV circuit-breaker 4 4 The first threshold should be lower than the short-circuit current downstream of the transformer, ie. 54 kA on the LV side, equivalent to 1100 A on the MV side. It should discriminate against the short-time threshold of circuitbreakers (A) or (B). If this threshold is set at 5 Ir,
Td (s) 10000
1000
MV circuit-breaker low threshold 600 A corrected to the LV = 29 kA
Max. LV Isc 54 kA
I2t curve
100 tr = 12 s
Max. MV Isc
10
Short-time threshold = 5 Ir
MV circuit-breaker high threshold 1400 A corrected to the LV = 68 kA
0.1
0.01 0.1
10
100 I/In
Fig. 24: Analysis of discrimination between LV circuit-breakers and MV protection of the transformer
the corresponding maximum value will equal Ir x threshold x tolerance, or 4000 x 5 x 1.1 = 22 kA, corresponding to 450 A for the MV incoming line. The first MV threshold can therefore be set at 600 A. To avoid interference with the short-time delay of 0.3 s, the delay associated with this first threshold is set for example at 0.6 s. The second threshold should be higher than the above short-circuit current (1100 A), and lower than the short-circuit current upstream of the transformer. Assuming that the network has short-circuit power of 150 MVA, the corresponding current is 4 kA on the MV side. The second threshold can therefore be set at 1400 A.
Bibliography
Standards c IEC 60255-3 Electrical relays - Part 3: Single input energizing quantity measuring relays with dependent or independent time Schneider Electric Cahiers Techniques c Protection of electrical distribution networks by the logic selectivity system Cahier Technique no. 2 R. CALVAS - F. SAUTRIAU c Behaviour of the SF6 MV circuit-breakers Fluarc for switching motor starting currents Cahier Technique no. 143 D. GIBBS - J. HENNEBERT c Development of LV circuit-breakers to standard IEC 947-2 Cahier Technique no. 150 E. BLANC c Electrodynamic forces on busbars in LV systems Cahier Technique no. 162 JP. THIERRY - C. KILINDJIAN c Energy-based discrimination for LV protective devices Cahier Technique no. 167 R. MOREL - M. SERPINET
Schneider Electric
Transl: Lloyd International - Tarporley - Cheshire - GB. Edition: Schneider Electric. Printing: Imprimerie du Pont de claix - Claix - 1000. - 100 FF04-2001
11658

Cahiertechnique

Uploaded by

Document Information

Original Description:

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Cahiertechnique

Uploaded by

Copyright:

Available Formats

Technical series ..........................................................................

ECT 0 up dated May 1998

1.1 General knowledge

1.2 Low Voltage 1 kV

1.3 High Voltage > 1 kV

Cahier Technique Schneider no 0 / p.3

2.1 General knowledge

2.2 Low Voltage 1 kV

2.3 High Voltage > 1 kV

Cahier Technique Schneider no 0 / p.4

3 Electrical and environmental constraints

3.1 General knowledge

3.2 Low Voltage 1 kV

3.3 High Voltage > 1 kV

Cahier Technique Schneider no 0 / p.5

4 Dependability: availability, safety

4.1 General knowledge

4.2 Low Voltage 1 kV

4.3 High Voltage > 1 kV

Cahier Technique Schneider no 0 / p.6

5 Protection devices and monitoring

5.1 General knowledge

5.2 Low Voltage 1 kV

5.3 High Voltage > 1 kV

Cahier Technique Schneider no 0 / p.7

6 Breaking techniques and switchgear

6.1 General knowledge

6.2 Low Voltage 1 kV

6.3 High Voltage > 1 kV

Cahier Technique Schneider no 0 / p.8

Schneider throughout the world

France - Grenoble: Fax: +33 (0)4 76 90 49 64

Edition: Schneider - Grenoble Printing: Clerc Fontaine - France - 1000.

Collection T echnique ..........................................................................

Cahier technique no. 114

ECT 114 updated, February 1999

Cahier Technique Schneider Electric no. 114 / p.2

Residual current devices in LV

4 RCD operating principle and description

5 Optimised use of the RCD

Cahier Technique Schneider Electric no. 114 / p.3

1.1 The RCD: its scope

Outgoing current Load Source Return current id = ia - ir Fault current Source

Fig. 1 : a current leakage results in a residual fault current id.

Fig. 2 : earth leakage protection.

Fig. 3 : current leakage protection.

Cahier Technique Schneider Electric no. 114 / p.4

1.3 The RCD, a useful protection device

Cahier Technique Schneider Electric no. 114 / p.5

2 The patho-physiological effects of electrical current on people

2.1 Effects according to current strength

Current strength (mA) DC 50/60 Hz 3.5 41 51 60 0.5 6 10 15 30

Fig. 5 : effects of weak electrical currents on human beings.

2.2 Effects according to exposure time

Cahier Technique Schneider Electric no. 114 / p.6

Duration of current flow b c1 c2 c3

Prospective contact voltage (V)

Maximum breaking time of the protection device (s) AC

5 5 5 5 5 1 0.40 0.30 0.20 0.10

5 5 2 0.80 0.50 0.25 0.06 0.02

Cahier Technique Schneider Electric no. 114 / p.7

Fig. 8 : fault voltage generation principle RCD.