You are on page 1of 743

A Series of Reference Books and ~ ~ ~ ~ o o k s

F

O

~ EDITOR ~ ~

G

~arlin 0. Thurston Department of Electrical Engineering "he Ohio State University
Columbus,Ohio

1. Rational Fault Analysis, edited by ~ i c h a r d Saeks andS. R. ~ i b e ~ y 2 . onp parametric Methods in Communications, edited by P. Pa~antoniKazakos and Dimitri Kazakos 3. Interactive Pattern Recogni~on, Yi-tzuu C ~ i e n 4. Solid-state Electronics, ~ a ~ r e n E. ce ~urr 5. Electronic, Magnetic, and Thermal Properties of Solid Materials, K/aus Sc~roder 6. Magnetic-Bubble Memory Technology,NW Chang 7. Transformer and Inductor Design Handbook, C o / o ~ e /Wm. T: ~c~yman 8. Electromagnetics: Classical and Modern Theory and Applications, Samue/ Seeiy and A/exa#derD. Pou/a~kas 9. One-Dimensional Digital Signal Processing,Chi- song Chen 10. Interconnected Dynamical Systems,~ a y m o n d A. DeCarlo and ~ i c h a ~ Saeks 11. Modern Digital Control Systems, ~aymo~ G.d Jac~uot 12. Hybrid Circuit Design and ~anufacture, Roydn D. Jones 13.MagneticCoreSelectionforTransformersandInductors: A User's Guide to Practice and Speci~cation, Colonel W ~T .* ~c~yman 14. Static and Rotating Electromagnetic Devices, ~ichard N. €~ge/mann 15. Energy-E~cientElectricMotors:SelectionandApplication, ~ohn C. Andreas 16. Electromagnetic C o ~ p o s s i b i l iNeinz ~, M. Sch/icke 17. Electronics: Models, Analysis, and Systems, James G. Go~/ing 18. Digital Filter Design Handbook,Fred J, Tay/or 19. Multivariable Control: An Introduction,P. K. Sinha 20. Flexible Circuits: Design and Applications, Steve Gu~ey, wit^ cont~~ution by sCarl A. €dstrom, Jr., Ray D. s en way, and W i / / i a P. ~ Ke//y 21. Circuit Interruption: Theory and Techniques, T ~ o m a€ s.Browne, J r . 22. witch Mode Power Conversion: Basic Theory and Design, K. Kit Sum 23. Pattern Recognition: Applications to Large Data-Set Problems, SingTze Bow

24. Custom-Speci~c Integrated Circuits: Design and Fabrication,Sfan/ey L.

25. Digital Circuits: Logic and Design, Rona/d C. €mey

H U ~ f

26.Large-scaleControlSystems:TheoriesandTechniques, ~ a g d S. i n, and ~ o h a m e d G. D a ~ i s h ct Management, €/i 7". fafhi and Cedffc sored by Onfaffo Centre for ~icroe/ecfronics~ omagnetic Design,~ ~ c h aP. e Peny / 29. Multidimens tems: Techniques and Applications, edifed by S~yros G. 7"zafesfas 30.ACMotorsforHigh-PerformanceApplications:AnalysisandControl, Sakae Ya~amura 31. Ceramic Motors for Electronics: Processing, Properties, and Applications, edifed by Relva C. Buchanan 32. ~icrocomputerBusStructuresandBusInterfaceDesign, A~hur L. Dexfer 33. EndUser'sGuidetoInnovativeFlexibleCircuitPackaging, Jay J. ~iniet 34. Reliabili~ Engineering for Electronic Design,orm man B. F u ~ u a 35. Design Fundamentals for Low-Voltage Distribution and Control, Frank W. Kussy and Jack L. ~ a r ~ e n 36. Encapsulation of Electronic Devices and Components, ~ d w a r dR. Sa/~on 37. Protective Relaying: Principles and Applications, J. Lewis B/ack~urn 38. Testing Active and Passive Electronic Components, Richard f .Powe// 39. Adaptive Control Systems: Techniques and Applications, V. V. Cha/am 40. Computer-Aided Analysis of Power Electronic Systems, Venkafachaff Rajagopa/an 41. Integrated Circuit Quality andReliabili~, Eugene R. ~ n a f e k 42. Systolic Signal Processing Systems, edifed by €ad E. S w a r t ~ a n d e Jr. ~ 43. Adaptive Digital Filters and Signal Analysis,~ a u ~ G. ce Be//anger 44. Electronic Ceramics: Properties, Configuration, and Applications, edifed byLiane/ ~. Levinson 45. Computer Systems Engineering Management,Robert S. A/ford 46. Systems Modeling and Computer Simulation, edife~ by ai^ A. Kheir 47. Rigid-Flex Printed Wiring Design for Production Readiness, ~ a / f eS. r Rig/ing 48. Analog Methods for Computer~Aided Circuit Analysis and Diagnosis, edifed by Taka0 Ozawa 49. Transformer and Inductor Design Handbook: Second Edition, Revised and Expanded, Co/one/Wm. ~ c L y m a n 50. Power System Grounding and Transients: An Introduction,A. P. Sakis ~e/iopou/os 51. Signal Processing Handbook,edifed by C. H. Chen 52, Electronic Product esign for Automated Manufactu~ng,H. Richard Sf~//we// 53. ~ynamic Models and Discrete Event Simulation, ~ i / / i a m ~ e / a and ney
7":

and Application: An Introduction, Edwin S. Oxner

and Related Technologies, G e ~ / gins d be^ plies: Power Conditioners for Critical

:Analysis, Design, and Application, Pau/ L.
ry Technology Handbook, edited byH. A. ~ i e h n e orkModeling,Simulation,andAnalysis, edited by Ricardo F. arzia and ~ a R. ~ Garzia o

trol and Decision Making,e~ited by Hiroyuki

65. Industrial Power Distribution andi~luminatin~ Systems, Kao Chen istri~utedComputer Control for Industrial Automation, ~ob~vo~e ~ o ~ o vand i c Vijay P. ~hafkar . Computer-Aided Analysis of Active Circuits, Adrian /oinovici .~ ~ s i ~ with n iAnalog n ~ Switches, Steve ~ o o r e . Cont~mination Effects on Electronic Products, Carl J. Tautscher 70. Computer-Operated Systems Control,~ a gS,~~ ia h ~ o u d 71. ~ntegrated Microwave Circuits,edited by Yoshih~rooni is hi 72. CeramicMaterialsforElectronics:Processing,Properties,andApSecond Edition, Revised and Expanded, edife~ by Relva C.
gnetic Compatibili~:PrinciplesandApplications,
75.

David A.

switch in^

Robotic Systems,edifed by SpyrosG. Tzafestas Phenomena in High-Volta~eCircuit Breakers, edited by ~ ~ n~akanishi i o Signal Processing, edifed by Sadaoki Furui and nd Image Preprocessing,Sing-Tze Bow "E~cientE~ectricMotors:SelectionandApp~ication,Second neeringSystems, edited by S ~ y r o s G. ?za~estas and ~ e i g o ~ a t a n a b e ~igita Filters, ~ ~~-Shet ff ug nalysis and Design of Switc ronic Modules, edifed by ~ i c h a ePecht / d Control: Current Trends and Modern Methodologies, edited by puter-Aided Design of

cessing: Theory and Applications,

G/eff~

87. Neural Networks and Simulation Methods, Jian-~ang Wu 88. Power Distribution ngineering: Fundamentals and Applications, James J.~ u ~ e 89. Modern Digital Control Systems: Second Edition,Raymond G.Jac~uot 90. Adaptive IlR Filtering in Signal Processing and Control, Phil/i~ A. Regalia 91. Integrated Circuit Quality and Reliability: Second dition, Revised and Expanded, 92. Handbook ,Richard edited by H. €ngelmann and William H. 93. Power-Switching Converters,Simon S. Ang 94. Systems Modeling and Computer Simulation: Second Kheir 95. EM1 Filter Design, Richard Lee Ozenbaugh 96. Power Hybrid Circuit Design and Manufacture,Haim Taraseiskey 97.RobustControlSystemDesign:AdvancedStateSpaceTechniques, Chia-Chi Tsui 98. Spatial Electric Load Forecasting, H. Lee Willis 99. Permanent Magnet Motor Technology: Design and Applications, Jacek F. Gieras and itch ell ~ i n g 100. HighVoltageCircuitBreakers:DesignandApplications, Ruben Garzon 101. Integrating Electrical Heating Elements in Appliance Design, ~ h o r Hegbom 102. MagneticCoreSelectionforTransformersandInductors: A User’s Guide to Practice and Specification, Second Edition, Colonel ~cLyman 103. Statistical Methods in Control and Signal Processing, edited by ~ o h r u Katayama and Sueo Sugimoto 104.. Radio Receiver Design,Robert C. Dixon 105. ElectricalContacts:PrinciplesandApplications, edited by Paul G. Slade 106.HandbookofElectricalEngineeringCalculations, edited by A r ~ n G. Phadke 107. Reliability Control for Electronic Systems, Donald J. LaCombe 108. Embedded Systems Design with 8051 Microcontrollers: Hardware and Software, ZdravkoKarakehayov,Knud Smed Ch~stensen,andOle W~nt~er 109. Pilot Protective Relaying,edited by WalterA. Elmore 110. High-Voltage Engineering: Theory and Practice, Second Revised and Expanded, ~ a z e Abdel-sa n la^, Hussein Anis, Ahdab €l~orshedy, and Roshdy Radwan

~ddi~ional V o l u ~in e ~reparation

lectromagnetic ~ o m p a t i ~ i Principles li~: and~p~lications, Second dition, Revised and Expanded,~ a v Wesfon j ~

MI Filter Design: Second Edition, Revised and Expanded, ~ j c ~ a ~ Lee ~ z e n ~ a ~ ~ ~

MARCEL

MARCEL DEKKER, INC.
D E K K E R

NEWYORK BASEL
e

First edition published as High-Voltage Engineering: Theory and Practice, edited by Magdi E. Khalifa.

This book is printed on acid-free paper.
ea~quarters Marcel Dekker, Inc. 270 Madison Avenue, New York, NY 100 16 tel: 212-696-9000; fax: 212-685-4540

Marcel Dekker AG Hutgasse 4, Postfach 812, CH-4001 Basel, Switzerland tel: 41-61-261-8482; fax: 41-61-261-8896

The publisher offers discounts on this book when ordered in bulk quantities. For more information, write to Special Sales/ProfessionalMarketing at the headquarters address above.

Neither this book nor any part m a y be reproduced or transmitted in any form or by any means, electronic or mechanical,including photocopying, ~ i c r o ~ l m i n and g, recording, or by any information storage and retrieval system, without permission in writing from the publisher. Current printing (last digit): 1 0 9 8 7 6 5 4 3 2 1

'7

.. .and

say '0 my Lord increase me in knowledge. .I'
The Sublime Quran (20: 114)

This Page Intentionally Left Blank

Preface to the

Not long after the first edition of this book saw light, Professor Mohamed Khalifa-editor of that edition-passed away after a long and gallant battle with cancer. Professor Khalifa assumed the leadership of high-voltage engineering research in Egypt and the region for nearly three decades. He rose to a position of respect among the international and national scientific communities, where he wasknown to be an extremely devoted researcher and a gifted teacher. The authors of this book will always remember him as a distinguished scholar, a great mentor,-and a beloved colleague. The need for electrical and electronics engineers to be well trained in high-voltage techniques and to be familiar with related basic theories and concepts is still being feltmore keenly today, world wide, than ever before. This book deals with a variety of topics which largely fall under the heading of high-voltage engineering.Such topics as electrical insulation and its breakdown under various stresseshave direct applications in industries where electrical and electronic equipment is manufactured. Knowledge of the techniques for generating high-voltage DC, impulse, and AC low and high frequencies is essentialfor engineers in charge of the design and manufacture of high-voltage power supplies. Such equipment has a wide range of applications, extending from aerospace electronics, radio-transmitting stations, and particle accelerators used in the medical and nuclear physics fields, to television receivers. The subject of gas discharges isanother example, as they are encountered not only as noisy corona on HV and EHV overhead power lines and as troublesome arcs in switches, but also in electrostatic precipitators, ozone production plants, gas-discharge lamps, arc furnaces, plasma torches, and ion-implantation equipment, to give just a few examples. Selected areas of this vast field of high-voltage engineering represent essential ingredients in the information and training that should be acquired by all electrical and electronics engineers.Included are engineers responsible for the design and operation of power transmission and distribution systems at all voltage levels, those responsible for the design and manufacture of

vi

reface

electrical and electronic equipment, those in charge of the design and installation of industrial plants and radio transmitting stations, and engineers and scientists involved in the design and installation of particle accelerators in medical and nuclear research. Not to be overlooked are the staff of highvoltage research centers and university faculties active in research, education, and training in fields of high-voltage engineering. The book requires no prerequisites other than the physics and mathematics courses normally taught to undergraduates in electrical and electronics engineering. The first edition of this book was published over a decade ago. Over this relatively short period oftime high-voltage engineering and related areas have witnessed marked development, manifested by a rapid growth in manufactured high-voltage apparatus combined with the vast a c c u ~ u l a tion of diverse knowledge and skills. This development assures for highvoltage engineering a brilliant future, in which its topics are slowly becoming a favored subject for research by graduate students. It also points to the need for a well-revised version ofthis book. The second edition of the book, in endeavoring to describe the present state of theart, is substantially enlarged and deals with more diverse areas. This edition of the book has undergone many changes and additions. The material included in the first edition has been thoroughly updated by citing numerous new references. New material has been added, while older material bas been carefully revised.To further enhance the role of the book as a textbook, all chapters now include a set of exercisesproblems or solved examples. The book now caters further to the needs of practicing engineers by introducing two new chapters on industrial applications of high-voltage engineering and related issues of hazards and safety. The large amount of information now available on high-voltage engineering and the time required to organize it led to the decision to divide the book into three parts. The first part, High-Voltage Engineering Theory, outlines the theories and models of insulating materials and dielectric phenomena. It comprises Chapters 2 through 8. The second part, High-Voltage Engineering Practice, on the other hand, addresses numerous practical issues and deals with common high-voltage equipment, some important design problems, and high-voltage testing techniques, which are indispensable to the practicing engineer. It comprises Chapters 9 through 18. The third part, High-Voltage Engineering Applications, comprising Chapters 19 and 20, and is devoted to timely applications of high-voltage engineering not only in the electric power utility but in industry at large. Chapter 2 is a rigorous but clear discussion of methods of field computation andmapping. New in this edition is the versatile boundary-element method of field computation. In Chapters 3 and 4 the physical phenomena

reface

ii

acting in different types of gas discharge are reviewed. A practical account of long air-gap characteristics is added in this edition. In Chapters 5 and 6 corona and arc discharges are described in some detail. Liquid and solid insulating materials are surveyed briefly from the point of view of electrical engineers in Chapters 7 and 8. In this edition more comprehensive accounts of streamers in oils and tracking and treeing in solids are offered. In the first chapter of Part 1 1 (Chapter 9), the principles of design of high-voltage busbars are discussed, together with their insulation and ampacity. Metal-clad gas-insulated switchgear, which is now widely used at the H V and EHV levels, is described in Chapter 10, which covers modern diagnostic practices. The various types of circuit breakers and cables are discussed in Chapters 11 and 12, including an account of solid-state breakers and superconducting cables. In this edition basic circuit interruptions are elaborated, and controlled switching is added. Chapters 13 through 15 are assigned to a treatment of power system grounding, external and internalovervoltagesimposed on system insulation, and techniques adopted for insulation coordination. This edition describes modern computer-based methods ofovervoltage evaluation as well as modern distribution grounding systems. The role of insulator contamination in insulation coordination is discussed in this edition at some length. Chapters 16 through 18 focus on the area of insulation testing, covering the topics of high-voltage generation, measurements, and standard specifications. The remarkable developmentin modern measurement tools over the past decade is reflected in this new edition. Two new chapters are introduced in this edition of the book, namely Chapters 19 and 20, which are devoted to present-day applications. Chapter 19 describes some of the industrial applications of high-voltage engineering outside the area of electric utilities, Chapter 20 deals with the timely issues of electrostatic hazards and safety in some industries. It is hoped that electrical and electronics engineers reading this book will findit rewarding and stimulating, and that it will enrich their knowledge of several segments ofthe field of high-voltage engineering,which continues to grow in importance and scope. Many persons have contributed to this book in various ways, and their efforts are most sincerely appreciated, We-the authors-are indebted to our colleagues for their keen interest, helpful suggestions, constructive criticism, and wholehearted cooperation. We appreciate the valuable have received from our numerous graduate studentsover the years. appreciate the courtesy of permission, freely granted by firms and organizations, to reproduce many illustrations included in this enthusiastic and sincere cooperation from the team at

Mazen A b d ~ l - S a l a ~ ~ussein Anis Ahdab ~ l - ~ o r s h ~ ~oshdy ~adw~

This Page Intentionally Left Blank .

Ab~el-sa la^ 9 81 115 5 The Corona Discharge 6 The Arc Discharge M. Anis 235 ix . Abdel-sa la^ 149 185 207 7 Insulating Liquids M. Abdel-sa la^ 3 Ionization and Deionization Processes in Gases M. Khal~a and H. Anis 8 Solid Insulating Materials M. ~ h a l and ~ a M. Khal~a and H. Abdel-sa la^ 4Electrical Breakdown of Gases M. Abdel-sa la^ M.Preface l Introduction M. Anis Part L 2 Electric Fields M. Khal~a and M. Khal~a and H.

El-~orshedy 14 Overvoltages on PowerSystems H. A~del-Salam 1’7 Hi~h-Voltage~easurements H. Anis 415 447 48 1 519 567 18 Testing Techniques R. El-Morshedy 19 Applications of High-Voltage Engineering in Industry M. Radwan and A. Radwan and A. Anis 15 Insulation Coordination H. Anis R.rt 267 A. Radwan 29 1 335 373 12 High-Voltage Cables M. El-Morshedy 10 as-~nsulated Switchgear H. A~del-Sala~ 20 Safety and Electrostatic Hazards R. Abdel-sa la^ 13 Grounding Systems A. Anis 16 High-Voltage Generation M.El-Morshedy Index 619 645 68 1 715 .

Giza. lightning discharges transfer positive charge upward to restore the system’s dynamic balance. Lightning discharges are the only known “natural” form of high voltage. The presence of man-made high voltage falls into three main purpose-oriented modes in which the interdisciplinary nature of high-voltage technology is fully apparent from the numerous and varied industrial processes in which it appears. and electrostatic precipitators . particle accelerators. a higher voltage requires smaller current and. all other forms of high voltage are man-made or “synthetic” to fulfill specific goals. Egypt High voltage is present around us in a variety of forms. smaller conductor cross-sectional area and lower conductor costs. therefore. Moreover. The second mode in which highvoltage is utilized is based on the fact that bodies chargedunderhigh voltage develop an electrostatic force. The regular current flow between the positively charged ionosphere and the negatively charged earth is thus controlled by global thunderstorm activity.Cairo Uni~ersity. Applications of this property may be found in cathode-ray tubes. spray painting. xerography. On the other hand. The first-and. the most well known-mode is the use of high voltage in electric power transmission to avoid excessive line currents which wouldrender the transmission system uneconomical. For a given power. the current-dependent copper losses are reduced. In the global system. by far.

and transient recording. It isnow quite common to find1000 MW being generated in a single power station (Fig. design. high-voltage.6 BTOEin 2020. 1999). measurement techniques. and many other high-voltage phenomena. nearly doubling present electricity demand (Fig. Electric power is expected to be the fastest-growingsource of end-use energy supply thro~ghout the world over the next two decades. and UHF overhead lines. the world global energy demand is to increase from 9. In terms of fuel. e. ignition in internal combustion engines. conventional fuel. EHV. 1. The third mode of high-voltage presence makes use of the ability of high voltage to initiate ionization in dielectric materials where energy is subsequently released in controlled quantities. in which the physical and chemical phenomena governing the electrical properties of insulating materials are of fundamental significance to high-voltage engineers. various industrial. and nuclearfuel. circuit analysis. 1986). and other equipment.halifa and Anis which remove vast quantities of dust from the flue gases of factories and power stations which would otherwise pollute the atmosphere. Physics and chemistry also play prominent roles in high-voltage technology. medical.g. To meet the ever-increasing demand for electricity. The extension of high-voltage technologyfrom energy transmission to encompass many other industrial applications has involved various areas of electrical engineering and science.e. which represents an average annual growth rate of about 1. and their insulation coordination. 1. The field of high-voltage engineering covers all these areas. and other applications of gas discharge. and testing problems for HV. and from about 1920 several higher-voltage systems were developed in various parts of the .2) (Chrobak and Pollak. 1997).gas-discharge lamps.. Other areas of high-voltage engineering include the design of highvoltagepowersupplies for various aerospace and ground applications. To this mode several applications belong.5 BTOE (billiontons oil equivalent) in 1996 to 13. high-frequency sources for radio transmission. larger power stations are being built for efficient utilization of water power.. Demand for electricity is projected to grow to 19 trillion kilowatt-hours in 2015..g. operation.4% (Hammons et al. cables.1) (US Department of Energy. Around the year 1910 voltages up to 100 kV were used. and ozone generators that eliminate the unpleasant smells from urban waste as it passes through sewage-treatment plants.

1970 1975 1980 1985 1990 1995 2000 2005 2010 2015 ure 1 .An ultra-high-voltage (UHV) transmission is currently operating in Russia at 11 50 kVwith a power capacity of nearly 10. With the advent of the transformer. DC systems were quickly replaced by AC systems in which transmission at much highervoltageswaseconomically more attractive. in recentyearshigh-voltage direct current (HV transmission has regained some of its earlier popularity. Some ofthe earliest electricity supply systems used direct current ( for the transmission and distribution of electric power. first at a voltage of 550kV and then at 765 kV or 800 kV. In 1954 a voltage of 380kV was adopted as an international standard and during the 1960s powers upto 1000MW were transmitted over distances of 1000km. This is attributed to the fact that HVDC has no transmission distance limitation. Figure l . The disadvantages of HVDC are basically the difficulty to tap off power at points .000 MW. makes DC transmission sometimes preferable to AC. During the period from 1930 to 1950 many hydroelectric systems were constructed throughout the world which could deliver 250 MW for 400 km at voltages between 200kV and 300 kV. however. The high reliability of modern solid-state converters. However. 1 World electricity c o n s u ~ p t i o n . no stability or reactive power problems. along with other technical and economic advantages.3 shows the historical development of high voltage levels. world. and produces many economical advantages when used in combination with AC transmission. Powers of 50 MW or so were transmitted over distances of approximately 50 km.

) (Courtesy of Asea Brown Boveri AG. 1987). there is a critical distance at which DC transmission becomes more economical to adopt than AC. The extra-high-voltage (EHV) grid covering Western Europe operates at 400 kV AC and is interconnected with that of Eastern Europe at substations in Austria (Putz.Nuclear power station. tielines connect power stations intoan integrated network over an entire country or a group of neighboring countries. As a result. The highest DC transmission voltages are now found in Brazil at zt600 kV and in Russia at A750 kV. along the line. These interconnections greatly improve the reliability and economy of the system and make available the benefits of diversity among the peaks of the component loads. and the excessive station costs. The United States and Canadaare interconnected at voltagelevelsof about 750 kV ACand . Rather than having every power station or small group of power stations look after its own loads. Switzerland.

Every national transmission system has a minimum and maximum voltage with. Equipment insulation is stressed both by continuous normal voltage and by relatively slow . Figure 1. usually. Table 1. 1985). ” I 8 0 0 1600 1400 1200 Russia I DCl “ I I 100 8 00 600 400 200 U. Figure 1.3 Historical developmentof transmission AC and DC voltages. 1983) defineshigh voltage as being greater than 1OOOV AC. The partial schematic of a typical power networkin Figure 1. f 4 0 0 kV DC. and the science of materials has contributed greatly to the development of improved insulation systems for high-voltage apparatus. . or greater than 1200V DC.S. The standard levels in a country are decided in view of the country’s operating limits.V. The weak link in the chain of reliability of equipment is still the insulation.A.1 lists the classes and the most common AC voltages in each class. SIFIC The International Electrotechnical Commission (IEC Standards. a permissible difference of about 200% between these extremes.5 shows a partial view (Bethmann and Boehle.4 shows the main components of an outdoor 500 kV substation. .

5 A 500 kVoutdoor substation.) . ure 1. (Courtesy of South Wales Switchgear. and feeders.A large power station Schematic of a power network showing busbars. transformers. breakers. generators.

4 1 1 5 287 500 (UHV) dynamic and fast transient over-voltages. coordinated. 1976). for the sake of system economy. adequate information needs to be acquired.5 2. regardless of the size. Additional necessary information includes the physicsof breakdown of . We should also be able to translate these voltage magnitudes and durations to electrical stresses across the vulnerable parts of equipment insulation and.l~tro~~ctio~ le 1 . in some cases. to dimension them. The lightning-impulse test voltage is about 10 times the rated AC voltage for low-voltage equipment.47 23 34. Thetest voltages have to esceed the normal rated voltages by some factor of safety. This involves computingand. This applies to equipment in power as well as electronics systems. It applies equally to components and to devices. We should be able to calculate and measure system over-voltages. accordingly. To achieve an economically and technically sound design for equipment insulation. The corresponding ratio is only about 3 for UHV equipment (IEC Standards.9 12. For insulation to withstand the various stresses. 1 Standard Operating Voltages in Europeand the USA Normal linevoltage Europe USA (50 Hz) 120 220/240 3801415 650 1000 kV (60 Hz) (l-ph) 208 600 Voltage class Low (LV) kV Medium (MV) 5 1 1 22 33 66 110 132 156 220 275 380 400 800 1150 69. it bas to be carefully designed. and properly tested. measuring electric fields.0 High (HV) 138 161 230 Extra (EHV) 345 765 high Ultra 6.

liquid. 1987. Electra 114 (October):21-24. Putz W.1999. Chrobak E. 1983. service performance. Brown Boveri Rev 73:224-230. and testing of insulating components. IEC Standards. J . Voltage Standards. Abaza M. Publication 38. Publication 7 l . Pollak M. and the design. Insulation Coordination. Browne J. 72:464-471. and gaseous insulating materials. . Richardson B. US Department of Energy (DOE). IEEE Power Hammons T Eng Rev 19(5): 3-8. 1986. IEEE Power Engineering Rev 17(1l): 16-18. IEC Standards. Bethmann J. 1997. Brown Boveri Rev. Geneva: International Electrotechnical Commission (IEC). Boehle B.solid. Geneva: International Electrotechnical Commission (IEC). Lindahl G. Lay KL. 1976. 1985.

machine windings.Assiut ~ ~ i ~ ~ rAssiut. the electric field problem is complex because of the sophisticated boundary conditions. in others. In some gap geometries. In the presence of a space charge. nowledge of electric fields is necessary in numerous applications in the design and operation of electrical and electronic equipment. and cables as well as within electronic components 2. To name just a few of these. it is necessary: I. the electric fields can simply be expressed analytically in a closed-form solution. we seek solutions to Laplace’s equation that satisfy the boundary conditions. including media with different . For space-charge-free fields. a solution to Poisson’s equation is obtained. s i t y Egypt . In industrial applications such as electrostatic filters and xerograPhY In this chapter we survey how the electric field is determined for various gap geometries. In the design of UHV substations and the effects of electric fields in their vicinity 4. For the design of insulation and for assessing electrical stresses in high-voltage sources. In the study ofgas discharges 3.

pe~ittivities and conductivities. maximum field Emax. the field at any point space x from the tip of the inner paraboloid along its axis is given by where f and F are the focal lengths of the inner and outer paraboloids. For a needle-plane gap with needle-tip radius r < <h. For example. is the electrical potential.field of these configurations. For other cases with electrode surfaces that can be expressed by analytical relations. the gap length E(x) = 1n(4h/r) h(2x r) (x)' 2V + h . the field calculation can be s~mplified by the well-known transformation of coordinates. Laplace's equation has a closed-form solutionfor geometrieswith great symmetry and with one-dimensional fields.l contains formulas for the potential E. respectively. 1983). Configurations consisting of spheres or cylinders occur frequently in highvoltage equipment and the electric field E can easily be calculated for many 4 . as explained in detail in books on electromagnetic fields (Shen and Kong. with two confocal paraboloidal electrodes. Table 2. In such caseswe must resort to experimental modeling or numerical techniques. The general governing equationfor space-charge-free fieldsisLaplace's equation where rl. and field efficiency factor q: where Eav= V / d (d is the electrode spacing and V is the applied voltage).

Electric Fields 11 6 T- 0I Q U- A IQ 31% R n b ~~ A /I 2 U : ut II 2 X II 2 U : II P U : n W ll ut L . - n k H: l - c v) .

1). . y) due to conductor 1 and its image (Fig..1.. 2. . This guarantees that the potential at every point on the plane is zero.. Where the conductor separations and heights above ground are much larger than their \ ure 2. The potential at an arbitrary point M(x.1) is given by where alM and blMare the distances shown in Figure 2. 2.1 ission-line ~ o n d ~ c t oto rs Consider a multiconductor transmission line of very long conductors with potentials Vi and charges qi per unit length (i = l.n) strung parallel to the ground plane (Fig. The effect of ground is usually simulated by mirror images of the charges qi. hi and ri are the height above ground and radius of the ith conductor.1 Transmission-line conductors over the ground plane.

+-In(2.. equation (2.6) takes the form 41 2hl Vl = -1n2nxo rl 92 +-1n+-In-b12 2 2 ~ ~a 01 2 93 b13 a 1 3 2~x0 4n + ..9) bln 2JtEo a1. .10) . the resultant field at M has the components If the point M is placed on the first conductor. al" = rl.diameters. a useful approximation is to consider that the line charge is located at the axis of each conductor. Thus the resultant potential at M due to all the transmission-line conductors is given by Subsequently.9) may be written in the form (2. The field at point M due to conductor 1 and its image isthe gradient of that is. then @M = V . and equation (2. For brevity. blM = 2h1.

2.2. the separation d between the subconductors of a bundle. PI2.14) The potential coef~cientsPI*.9) reduces to (2.10) make up the first set of Maxwell's equations. The magnitude and direction of the maximum conductor surface field can be evaluated using the line conductor charges andpotential coefficients.13) The effects of the image charges can safely be neglected when evaluating the fields at the conductor surfaces. For example. and P22 depend ontheconductor arrangement in the bundle (Table 2.2. as expected. 2.12) = r ln(?h/r) The field strength at the conductor surface is decided primarily by the conductor radius andwhether the conductors arebundled. The field at the conductor surface reaches its maximum value at the point facing the ground. For bipolar DC lines and for three-phase AC lines (Fig.where P. 2. in case of a monopolar bundle 2. and their height h above ground would comply with the following relation: r d< <D and h (2. equation (2.7) and (2. = (1/27teO) l n ( ~ . T/ (2. Equations (2. and can be easily evaluated. = Pji and that they are positive quantities. DC line of voltage V .2) the spatial field variations can be calculated using equations (2.2). the separation D between phases. The maximum field strength Emax at either subconductor of the bundle is expressed as (2. be evaluated in terns of the line voltage and geometry. 2.2 Transmission Line with Bundled Con~u~tors In practical designs.8).1 Trans~ission Line with SingleConductor For a monopolarDC line witha single conductor. / ~ . equation(2. It is evident that P.15) + . the line conductor radius r. These can in turn.11) It can easily be shown that the equipotentials around the conductor take the f o m of circles. )is the potential coefficient in meters per farad. Knowledge of the ~ a x i m u m surface field for different conductor bundles is necessary for estimating the corona discharge phenomena on HV and EHV transmiss~onlines.2.10) reduces to ' v= Pll4l+ P12q2 = P21 41 p2242 (2. independent of the sign of charge and potential.

Comber andZaffanella. three.14) for a Bundle of Two Subconductors when Arranged Horizontally and Vertically Horizontal bundle 1 Pll =-In2m0 2h 2( c - o o + ) Vertical bundle-2 1 Pll =-In 2 Z E O r = P22 2(h + d/2) > >d r l 2h Pi2 =-In"-= P.2.Electric Fields le 2.. the maximum surface field can be similarly calculated for bundles of two..2 Potential Coefficientsof Equation (2.g. 2. some expressions were proposed by previous workers (e. Timascheff. They suggestedthe follow- Three-phase transmission line with a sketch instant when the center phase voltage is maximum. 1974).3 Approxi~ations For quick estimation of the maximum surface field for overhead line conductors.. 19755. 19'7'7). of field lines at the . and four subconductors (El-Arabaty et al.2. if h 2 ~ ~d 0 P22= - - " 2h In ifh>>d 2 ~ ~ d 0 1 2(h d/2) In r 2 Z E O = 42 4 2 = (P11pi 2)V PIP22 e 2 For three-phase AC or bipolar DC lines.

ing relation for the angular variation of the surface field E around the subconductor of a multiconductor bundle: E* = Ea 1 +"(M [~ 1)cosO 1 (2.16) where Ea is the average field obtained from the well-known relation 4 Ea = 2n1Or (2. a high-voltage bushing and a machine insulation) comprises more than one i ~ s ~ l a t i n g material. the higher the I value of the inserted solid dielectric slab. On the contrary.17) with the parameters defined as given in Figure 2. .g. The constancy of the electric flux density at the interface of a multidielectric system makes it possible to increase the field uniformity by an appropriate choice of dielectric permittivities. means that partial replacement of the gas insulation with a solid material does not improve the dielectric strength of the whole. Usually a low-voltage or a high-voltage insulation system (e. it becomes more heavily stressed. The electric field distribution in a multidie~ectric insulation system caneasilybe explained as a parallel-plate capacitor withtwolayers of different pemittivities 11 and 12. The electric displacement is the same in both layers. A typical example is the grading of insulation in coaxial cables (see Section 12.. hence the electricfield intensities El and E 2 are related as Et/E2 = E ~ / E ~This .3.12). Phase geometry of a bundle of n subconductors.

l 4 2 __ E1 (2. For assessing the electric field distributions in such complex arrangements. their sizes are comparable to their separations.5. and the dielectric of the much higher E behaves almost like a conductor as E " + 00. There. and machines. More is said about suchcasesin Section 2.1 Refraction Law of Electric Field For AC voltage applications.) and refraction (e2) are related as follows: tan81 E=tl/-EI. Then the angles of incidence (6. and more than one dielectric medium is involved. smooth cylindrical conductors parallel to each other and to groundwith separations much larger than their diameters represent overhead transmission lines. The former is the subject of this section and the latter is covered in Section 2. experiFenta1 analogs and numerical techniques. analytical methods are quite unsuitable. " " _ .3. The potential distribution in conduotive media in current equilibrium conditions satisfies Laplace's equation. Equation (2. Two other approaches are in use.4.2. gas-insulated switchgear. The conductors and insulation arrangements can be represented using an electrolytic tank. conductors take various shapes. free charges are absent at the interface between dielectrics and the polarization charges define the boundary conditions. Theyare only special simple cases compared to such conductor arrangements as in multicore cables. a sheet of semiconducting paper. in DC voltage applications. transformers. . the same as the electricfields in space-charge-free regions. accumulation of free charges at the interface takes place because of the differing conductivities of the materials (interfacial polarization). This fact makes it possible to obtain solutions to many difficult electrostatic field problems by constructing an analogous potential distribution in a conductive medium where the potential and field distributions can bemeasured directly.18) points out that the electricdisplacementfluxlines penetrating from a dielectric of a high E into one of a much lower E are forced to leave the material nearly perpendicular to its surface. The infinitely long.2. sometimes in three dimensions.18) tan62 E=t2/&2 4 1 E2 On the other hand. or by a mesh of resistors. This means that the equipotential surfaces in the lower-permittivity dielectric are forced to be nearly parallel to the interface.

Equipotential boundaries are represented in the tank by specially formed sheets of metal. care being taken to minimize the errors.5). .4 Electrolytic tank model of athree-corecablerepresentedat instant when one core is at zero voltage. introduce an errorarising from the finite mesh size. 2.4. For example. Replacement of the continuous field by discrete resistors does. among other things. and the contact resistance to the electrodes. the . the same as the sheath.6). In this analog. the continuous field is replaced by a discrete set of points as depicted by a mesh of resistors (Fig. This error may be reduced by reducing the mesh size. ~ifferent permittivities are represented by electrolytes of different conductivities separated by special partitions. a single dielectric problem such as a three-core cable may be represented using a flat tank asshown in Figure 2. The conductance of the entire model is a scale model of the capacitance of the system being represented. Its surface resistivity is of the order of 1ksZ per square (Fig. Otherwise. its dependence on the ambient humidity. A somewhat less accurate but attractively simple alternative to the electrolyte is semiconductive paper. the tank base can be specially shaped. 2.This method has been widely usedfor decades. Errors in this method result. of course. from the nonhomogeneity of the paper resistivity.

5 Field plot between two spheres with skanks.19) or + v2 + v 3 + v4 4vo = IoR Thus the resistive network analog can approximately represent Laplacian fields (Io = 0) as well as Poissonian fields (Iof 0). For the mesh point 0 of Figure 2. . the continuity equation of current demands that v 10 0 v 3 -v 0 v 4v 0 + = I o -v + -+ v2 -v R R R R v 1 (2.' \ ure 2. H 0 / .40% \ \ \ \ \ \ \ \ \ \ " " \ \ " " " " \ " \ " " . ' 70% 1 ' \ I . as plotted by a semic ducting-paper model.6. c // c 0 ' ' 1 / / 8 0 30 7 0 0 0 0 0 ' / // / / \ \ \ . " " " " " " c " " " 20 Y o 0. For Io = 0.

an interpolation between those one step away from it. in fact. the same as in numerical calculations using finite differences (Section 2. Time dependence mayalso be represented using impedance networks containing capacitors and inductors. simulation of electrostatic fields inthe vicinity of curved surfaces of the electrodes is bound to be of reduced accuracy. the analog can be used in three-dimensional field problems by extending the mesh grid in three dimensions. ue to discretization.4.Resistive mesh analog of the field pattern between two electrodes.4). . Some techniques may complement others. Each potential value represents. They are discussed briefly below. (2.20) Thus the potential at each mesh point is determined by the adjacent point potentials. which are in turn determined by others. Several numerical techniques have recently been reported in the literature for solving Laplace’sand Poisson’s equations forthe fields between complex electrode arrangements. Nevertheless. Each numerical technique has advantages and drawbacks.

It isnoticed that each of the n cylindrical conductors has (n l) images. 2. The imaging technique cannot be extendedto multidielectric media or to three-dimensional fields. The sphere maybe represented by a point charge as an image equal to -Qr/D. The field outside the cylinder can be computed in terms of the charges.1. a next-imaging process is performed. The . q2..1 Successive Imaging among Cylinders A system of parallel cylindrical conductors. their respective potentials are VI. to reduce errors in the results.1). The Computations would not yield an equipotential surface exactly coinciding on each cylinder with its respective voltage. At any step in the imaging process. located a distance 6 = r2/D away from its center. Of course. .21). each cylinder presents an equal and opposite image for each of the charges existing outside it (Fig. sizable. parallel to its axis and displaced from it by a distance S="- r2 D (2. as discussed above. In two cases only is it possible to implement the successive imagingtechnique (Khalifa and AbdelSalam...V2.4. Therefore.The successive imagingtechnique is basedon the concept of imaginary point or line charges located outside the region of field evaluation.21) The two charges on the wire and cylinder are equal and opposite. 2. conducting cylinder. These line charges are used to compute the field in the space outside the cylinders. V4 and their surface charges are 41.. The image locations are calculated according to equation (2. For the system of conducting cylinders shown in Figure 2. is used as an example to explain how the image charges are determined.q4. each conductor will also have its image (see Fig. the number ofimagesinside each cylinderwillbe(2n I). but chosen such that their field within this region is identical to that of the induced charges on the electrode boundaries of the region. Now. . If the ground is considered. . if they are placed near the ground or its equivalent. 2. . 1974a) when the region of field evaluation is a single-permittivity medium. thin wire with a charge q (Cjm) and a parallel long. The second case is that of a field between a point charge Q and a charged spherical conductor. The usual practice is to represent the cylinder by a line charge as an image within the cylinder. each of the images in a conductor obtained by the preceding step is resolved into (2n 1) new images.7b). as in gas-insulated switchgear (GIs). The first case isthe field between a long. which in turn are determined by the voltage applied to the cylinder.7a.

largest distance of each of these (2n 1) new images from the image it replaces can beused as a criterion to determine whether the successive imaging process should be continued. of charges q2. (c) locating higher-order images. . large distance of this type indicates that further imaging is necessary to improve accuracy.. and q4 inside cylinder1).. A.(b) first-order images (i.(a) Group of conductors to be modeled. g.e.

7 ~ . but only if the spheres are far from ground. to be the N points of known potentials.23) where [G] is an M x N matrix evaluated during the imaging process.The system of images starts to form a branching network in each conductor. However. the following equation needs to be solved for the unknown charge density vector [q'] : (2. the sameimagingprocess asthat described above for parallel cylinders can easily be adopted here with n = 2.1.24) i= 1 (2. the M line charges representing the system are linearly related to the original N line charges by (2. The advantages characterizing the successive imaging technique are the well-defined coordinates of the image charges.Suppose that the number of line charges found to represent the system after termination of the imaging process is M. . 2. and the accuracy. the number of images in any conductor after the mth imaging process will be (2n l)m.2 Successive Imaging between Spheres Given two spheres located near. The electric potential and field may be calculated at any point outside the conductor boundaries as " M (2. one on each conductor surface. 2. each other.22) where [U is an N x 1 known vector of potentials of the N conductors. A s a practical application. and [q'] is an M x 1 vector of all line charges used in the system.8).25) where ria is the distance from q : to the point a and Uia is a unit vector along the direction from qi to a. Let N line charges represent a system of N conductors ( N = 2n) at various potentials. [P'] is an N x M known vector of the potential-coefficient matrix.If the imaging process is continued uniformly for all the conductors. thesuccessive imaging technique has been used to calculate equipotential contours for an overhead transmission line with two circuits and two ground wires (Fig. as suggested in Figure 2 . Choosing N points. with potentials V I and Vz. which can be improved by increasing the number ofsuccessiveimages.4. and isolated from.

55 cm B LC B v1 L A L C .4. each of them with (212l) image charges. To establish this model for a bundle of yt subconductors. 1989).198'7). homopolar. The dipole method was developed by Thanassoulis and Comsa in 19'71 to calculate the electric fieldin the vicinity of bundled conductors.3. - E" Q1 C 3c UI . 20 l I J 30 1 40 1 distance in meters Equipotential contours plotted for a double-circuit overhead test line using asuccessive imaging technique. .24 cm Groundwire radius = 0. the effect of earth (ground) is included.Conductor radius = 1. successive imaging has been combined withthe charge-simulation technique to compute the electric field of monopolar. Details are given in Section 2. Recently. TI) c 20 2 * 1 0 0 l 0 10 X. It is an extension of the successiveimaging technique described above (AbdelSalam and El-Mohandes. and bipolar conductor bundles (Abdel-Salam and ~1-Mohandes.

2. They are of the same polarity as the subconductor itself. ring charges. The magnitudes of these discrete charges are to be such that. For . the unknown charges qj can be evaluated. the larger the number of charges.3. 2. 1974).10) Thus. The number of charges and their coordinates are arbitrary. However..The technique is discretization of the integralequations technique explained in Section 2. The charges simulating a given electrode canbechosen as point charges. oras finite. corresponding to the other members of the bundle. that is.1 Single-Dielectric Arrangements In this case the distributed charges on the surfaces of the stressed electrodes are replaced by a number (n) of fictitious discrete charges arranged inside the electrodes. The accuracy of the method is satisfactory for practical separations exceeding 10 times the subconductor radius. The technique is described for single.9): (12. and their displacements from the subconductor axis are significant.under their integrated effect. that is. the higher the possible accuracy of Computation. the yz images inside each subconductor caused by the effect of ground can safely be assumed concentrated at the subconductor axis. A dipole is composed of each of the negative charges together with one of the positive charges at the subconductor axis.49.each electrode has its assigned potential as checked at every contour point on its surface.21). Pjqj = j= l v (2. outside the space in which the field is to be computed (Singer et al.26) where qj is the discrete charge and Pj is the associated potential coefficient. semi-infinite. Checking could be at a number m of contour points selected equal to or exceeding the number of discrete charges (n) and used for setting a group of simultaneous equations (Fig.According to equation(2. The charge-simulation technique can be employed successfully for the computation of an electric field between electrodes in a medium where one or more dielectrics are involved.The dipoles with the remaining axial charge can beusedincomputing the field distribution around the bundle.4. The other images. This choiceshould suit the shape of the electrode being simulated. or infinite line charges. will have the opposite polarity.and twodielectric arrangements.

the electric field at any point in space around the electrodes can be determined: (2. Finite and infinite line charges have commonlybeenused to simulate the charge on the surfaces of cylindrical conductors. ring charges centered on the axis are an effective method of simulation. . 1983).1977). the magnitudes of the charges qj are chosen such that thecalculated potentials at a large n u m ~ e of r points m >y1 deviate only slightly from the actual potential V. Therefore. (a) hemi- example. effort should be devoted to choosing the proper number and locationof simulation charges. have been proposed. 1974). axihyperbola.charge charge X X Contour points Ring Finite line Arrangement of simulation charges and contour points: (b) sphere with cylindrical shank. spherically capped rod. In terms of the calculated charges qj. spherical electrodes can easilybe simulated by point charges. Shell and annularplate charges have also been suggested (Mukherjee andRoy. such as oblate spheroid. The calculated potential cannot easily equal V at every point on the electrode surface under the effect of the simulation charges. prolate spheroid. Other charge shapes. For fields with axial symmetry.. and elliptic cylinder.27) The pote~tialand field c o ~ f ~ c i e nfor t s the different discrete charges are reported elsewhere (Singer et al. For sufficient accuracy. The deviations are minimized by the least-squares techniques (~bdel-Salam and Tbrahim.

the choice m >n represents a step forward in improving the conventional char~e-simulationtechnique. 2.4. 1975). Figure 2. The method has been applied (Foo and King.10). 2. there are y1 contour points with IZ discrete charges inside the surface of dielectric I. This version of the charge-simulation technique has the advantage of yielding surface fields directly with simple computations.This was found to improve considerably the stability of point matching.. for example. leaving a net surface charge on the boundary between the dielectrics. nl) points on the interface between the electrode and dielectric 11. In total. surface field values on the electrodes are first evaluated from theknown voltages (V) of these electrodes by solving a set of simultaneous equations analogous to (2. there are (ne n) contour points and (ne 2n) discrete charges (Fig. even with a smaller number of simulation charges (Khalifa et al. 1976) successfully to symmetrical and asym~etrical overhead bundled transmission lines. instead of the electrode charges. This surface charge can be simulated by discrete charges on both sides of the dielectric boundary.~ i e l ~ c t r A ic rran~e~~nts In this case dipoles are aligned by the electric field and compensate for each other in the bulk of each dielectric. The system of simultaneous equations required for d e t e ~ n i n g the unknown discrete charges is formulated according to the following boundary conditions: + + 1.28) j=l Similarly. contributing to the potential and fieldin dielectric 11. At the electrode. The potential at the nl contour points on the electrode/dielectricI interface must be equal to the applied voltage ? ? ne j=l n (2. The values of charges on the electrode surface are calculated from the field values there. with nl points on the interface between the electrode and dielectric I and (n. Indeed. At the dielectric boundary. there are ne charges and contour points.3. The improved technique showed better accuracy than the conventional one. The ne charges contribute to the potential and field in both dielectrics.10). and vice versa. the potential at the (ne nl) contour points onthe electrode/dielectricI 1 interface must be equal to V.10 shows. the cross-sectionof a metal electrode with an applied voltage (V) meeting dielectrics I and 11. . In another version of the charge-si~ulation technique.2 T w o .

The potential at each of the yt contour points on the interface between the two dielectrics is the same whether seen from either side. The displacement at the interface between the twodielectrics is continuous. an electrode and at a dielectric 2.ne 1 ne * * Electrode I I n Simulation charges X X Contour points Discrete simulation charges at boundary.29) j= 1 3. respec- .30) where tively. n j= 1 n (2. and q I are the permittivities of dielectrics1 and 11. Then the continuity is expressed as (2. that is. Let the coefficient$ be defined as the contribution of the charge qj to the component of the electric field normal to the dielectric interface.

three-dimensionalfields with and without symmetry can be handled without excessive computation (Fig.~ o h a n d e s(1987)haveproposed an efficient charge-simulation tech (ECST) for calculating electric fields around conductor bundles of transmission lines.On the other hand. 1987). 2. Moreover. the technique islimited to field calculations inarrangements with only one or two dielectrics.Abdel-Salam and Stanek. are dependent on the number of subconductors of the bundle and how they are arranged in space.The simulation technique was also used for field calculation in arrangements including free potential electrodes. With an increased number of dielectrics. and 4 monopolar. the number of simulation charges involved with the ECSTcould thus be much reduced. Hereexperienceplays an important role. In the ECST. 1987).. Singer. using the ECST. with a considerable saving of the computational time. Another major advantage is the simple simulation of curved interfaces between dielectrics and conductors.and bipolar DC transmission lines and showedhighaccuracy for the potential and fieldvalues calculated comparedwith results using the conventional technique. 1982). * + . the number of charges could be reduced from 16 to 3 w accuracy was improved from 0. The simulation technique was also attempted forcomputing electric fields.. The charge-simulation technique has been applied successfully for calculating the electric fieldin andaround practical suspension insulators (Khan and Alexander.Thus (ne 2n) linear equations are formulated for the calculation of the same number of unknown discrete charges. The ECST was applied for bundles 2. and around overhead transmission lines ( Arabaty et al. rather.1983). the contour points were chosen large compared with the number of discrete charges to achieve better matching of the boundary conditions (Abdel-Salam and Stanek. in such as pollution layers on insulation surfaces (Takuma et al.homopolar.including surface and volume resistance of the insulation. 1981). In a two-dielectric arrangement. The values of unknown charges satisfy the boundary conditions when their number and locations are chosen judiciously. the number E. 1987b).11) (Chen and Pearmain. 1987a. 1977). such as grading foils in condenser bushings and in pot heads of HV cables or suspende~ particles in gas-insulated systems (Abdel-Salam. 3. Therefore. 1981b. for a bundle 2 of a monopolar DC line. As an example. Abdel-Salam and E l . solution of field problems using the charge-simulation technique becomes more and more complex.1 ? 4 to % (Abde~-Salam and Mohandes. of simulation charges and their coordinates are no longer arbitrary but.

12 shows asquare gridwith its sides parallel to the X or Y axis.Thus we replace the derivatives describing Laplace’s equation with “divided-difference” approximations obtained as functionsof the nodal values.All the elementshave the same depthinthe Z direction. If the series is terminated by neglecting the terns containing third and higher derivatives of the potential. a two-dimensional field is treated here for simplicity. [From Okubo and Metz (1978). However. The finite-difference technique is applicable for three-dimensional fields. Among the nodes of the grid. such as node 0. Figure 2. where nodes 1 to 4 surround node 0. onlynodes 0 to 4 willbeof immediate interest.l1 Surface field calculations alongthe HV electrode and the dielectric spacer within a gas-insulated switch gear using the charge-simulation and finiteelement techniques. and &. then . As the potential within the field region is continuous. using the well-known Taylor’s series. 4b4.-Charge simulation technique Finite element technique o Contour polnts Olstance ( L ) measured along HV electrode .] The basis of this technique is the replacement of a contin~ous domain representing the entire space surrounding the high-voltage electrodes with a rectangular orpolar grid of discrete “nodes” at which the value of unknown potential is to be computed. it ispossible to expand the potential at any point. Suppose that the potentials at these nodes are 41 # . Cp3.

The next step is to apply equation (2.31 (2. indicating the node . at the nodes located on the high-voltage electrode surface. This completes the set of equations relating the node potentials [ ( P ] to the boundary coefficients [B]: Y ure 2.3 1) As Laplace’s equation in Cartesian coordinates states that for a square grid.32)in order to evaluate the potential at every node in the space between the electrodes. Of course.12 numbers.31) yields (PO 1 = q((P1 + (P2 (P3 + 44) (2. Regular grid for finite-difference technique. equation (2.32) Equation (2. (P = V.32) isthe finite-difference f o m of Laplace’sequation for a twodi~ensional field.

3. assume the nodes 2.13 shows the computed field within a gas-insulated switchgear.33) can be solved by the Gaussian elimination method for problems with a relatively small number of nodes (James et al. 0. The technique is well suited for use in field regions that are finite in space. For complex cases..32) takes the form (2.33) where [P] is an n x n matrix of the coefficients for n-unknown node potentials. the resultant field may be composed by superposition of two Laplacian fields. There is no reason why this technique cannot be used for computing the field in multidielectric arrangements. [ ( P ] is an n x l column matrix of the unknown potentials. and 4 to be at the interface betweentwodielectricsof permittivities (containing node 1 ) and c2 (containing node 3).4. On the other hand.(2. the potential and displacement are continuous. and [B] is an n x 1column matrix of the sum of known potentials of the bounding electrodes. Therefore. Some difficulties are encountered in solving many problems using the finite-difference technique. Figure 2. the number of nodes becomes very large and solution of equation (2. Referring to Figure 2.33) may be difficult even for machine computation. and equation (2.12.34) The finite-difference technique has thus been applied successfully for field calculation in switchgear design and development (Ryan et al. otherwise. At the interface. 1985).. Ul As noted in Section 2. and therefore its efficiency is considerably limited. Therefore. otential values between the nodes are obtained by interpolation. A regular gridis not suitable for curved boundaries or interfaces because they intersect grid lines at points other than nodes. the grid must be densernear the curved boundaries. Equation (2. the charge-simulation technique (CS) iswell suited for computing fields in the vicinity of curved electrodes placed in open (unbounded) space with one or two dielectrics. The electric field components at any node are easily expressedin terms of the potential values at the nodes surrounding it. 1983). the finite-difference technique (FD) is applicable to bounded spacewith .iterative schemes are most eEficient in combination with successive overrelaxation methods. The areas of field concentrations received detailed computations with finergrids.

In the CS domain. Also.A combination of CS and FD can be utilized in some cases to obtain the advantages of both methods (Abdel-Salam and El1989).unknown discrete charges with a set of contour points on the electrode surfaces. The coupling between the two domains is based on thecondition of continuity of potential anddisplacement at the q-. the surface charges on the electrodes are represented by y t .33) applies to the FD domain. coupling points along the CS-FD boundary. To apply the combined method. (1974).] multidielectrics.100 'l0 LO'Io 30 "10 1 \ 20 "lo l0 " I . the entire field is divided into two domainswhereeach technique isused separately (Fig.14). (b) plot of equipotentials at lower end of bushing. each of the 8 b coupling points on the boundary between the two domains hasits own potential to be . The FD domain is replaced by a rectangular grid whose ng nodes are of unknown potentials. equation(2. showing area enlarged in (b). ur Field computed bythefinite-differencetechniquefor a 300 kV gasinsulated system: (a) part of single-phase layout. While analyzing the CS domain. the FD domain is represented by a number ytb of discrete charges located inside it.10) applies to the CS domain and equation (2. Naturally. [From Scott et al. 2.

at the different nodes are unknown variables and. the total number of unknowns in the entire space is (n. the space in which the electric field is to be calculated is divided up intotriangular elements as shown inFigure 2. equations (2. Segerlind. known fromvariational calculus. that Laplace's equation is satisfied when the total energy functional is minimal. Entire field in air and a dielectric divided into studied.)2dA 2 (2. All the elements havethe same depth in the Z direction. as in the FD technique (Zienkiewicz.3 electrode Simu~ati~ charges Coupling points CS -domain- ' x x * e Contour points Nodes CS and FD domains. . If the permittivity of the medium is E. 2nb ng). 1984).35) where A represents the area scanned by the triangular elements. Solution ofelectricfieldproblemsby the finite-element method is based on the fact. the electrostatic energy functional F for a flat twofield dimensional t F= 1 -€(grad cf. to be evaluated by simultaneous solution of n.10) for the CS domain. + + According to this technique. ng equations (2.15.33) for the F D domain. Therefore. The potentials cf. only oneishalf evaluated in addition to the potentials of the ng grid nodes. 1977. and 2nb equations for the equality of potentials and continuity of displacements at the fib coupling points. for minimum energy functionals. Being symmetrica~.

15need to be considered. 1983).36) for all the potentials at the nodes (Silvester and Ferrari. Ass~mi~ a linear g variation of the potential @ as a function of x and y within the triangle. whenisvariedwhileall other potentials remain constant. It is obvious that. only the energy in the adjoining triangles is affected. In triangle 1. . in the partial derivative a F / ~ = @0. and a r r a n ~ e m e n tof the (2.Y 1 X Gridforthefinite-elementtechnique elements. Therefore. ~ only the energies in the six triangles number l to 6 in Figure 2. For the energy functional in the six triangles.

3 (2. The coefficients a. (P2.19. With a typical triangular element (Fig. For Ax = Ay.38a) and the functional F1 becomes.39) For the other triangles.40) Thus which is the same as the finite-difference solution (Section 2. and c in equation (2. similar equations can be obtained.41) Any increase in accuracy by including higher-order terms is outweighed by the increase in computation time and storage requirements. 2.3 (2. according to equation (2. Thus the actual continuous potential distribution over the XY plane is replaced by a piecewise-planar approximation.4).35). the potential is approximated by the polynomial ( P= a + bx + cy + a1x2 + blxy + cly2 + (2. 2 and the partial derivative takes the form (2. b. and (P3 at the three verticesof coordinates shown in Figure 2.15: (2.43) m=l . Most of the computations are satisfactorily based on a first-order approximation.15) (2.4.41) may be found from the three independent simultaneous equations obtained by constraining the potential to assume vertex values (Pl. the total derivative becomes (Fig. 2.42) That is.

that is. a is a linear function of x and y. 2.lectric Fields where . triangle. For brevity.46a) 4A where AYi and AX. as indicated by the arrow in Fig. a solution is also possible if E changes from element to element. (where i = 1. define the elements of the asymmetrical matrix [SIe for the triangular element (Fig.16) are expressed in terms of the coordinates of element nodes .16) (2.43) as l (2. as it contains the sensitivity of the energy functional with respect to the potentials.2.44) (2. 2.3 are the local node numbers which must be counterclockwise. However.45) where the dielectric within the element is assumedisotropic and the permittivity E is constant.) are the x and y coordinates of the vertices. $ =(A YiAYJ + AXiAXj) A =(A Y2AX3 A Y3AX2) 1 2 (2. It can easily be a are interpolatory on the three vertices of the verified that the variables . y.46) The matrix [ S I e is the well-known “stiffness matrix” of the element under consideration. easily evaluated. The elements of [ S I e are determined as 1 (2. The potential gradient may be found from equation (2.46~) . where (xn.

48) Any composite assembly of triangular elements may be built up one triangle at a time. The m ~ ~ m i z a t i o of n the energy functional for the specific element is expressed as (2. For minimum energy functionals. expressed as (2. _ I '2 3 must be Note that hence 3 i= 1 j= 1 Equations (2.46) yield the energy of the element. by the arrow.35) and (2. l 6 Typical triangular element: the local node nu~berin1 g 2 " c o u n t e r c l ~ c ~ w i as ~ eindicated . Example (1 l) explains how the individual element stiffness matrices are assembled to obtain the global stiffness matrix.3 Y .45) and (2. equations (2. It is therefore sufficient to consider how continuity is enforced whenone triangular element isadded to analready existing assembly.48) yield . and with properly arranged nodes.47) where [#l is the column vector of vertexpotential values of the element and "tr" denotes transposition. This ends up with the formation of the global stiffness matrix [q.

Some alternatives to this technique have recently been reported (Cendes and Hamann. On the other hand. tobe located far enough from theregion of interest as to minimize the resulting errors.. Theprice paid for doing this is that Laplace's equation must be solved for the entire larger region. By proper numbering of the nodes. however. The precision of field calculation using the finite-element method increases significantly with the nwnber of elements.(2. the matrix [ T I takes the form of a narrow diagonal bandof symmetrical structure. l1 shows a comparisonof the finite-element(FE) andchargesimulation (CS) techniques for calculating the electric field at the contacts within a gas-insulated switchgear. This adds to the computation time. A s in thefinite-difference (FD) technique. The small deviation between the field values obtained by the two techniques is attributed to two reasons. which can signi~cantly decrease the computation time (Jennings. However. We should add that the FE method needs more computer storage and time than does the CS technique. 1977)).50) The solution of the Laplace equation takes the form of a set of nodal potential values as obtained from equation (2. a portion of the boundaryis at infinity). This difficulty is usually surmounted by assuming an artificial boundary.e. The order of [ T Ican reach several thousands to cover an entire two-dimensional field. its bandwidth is about 5-15. Figure 2.49) where [+]b are the bound (known) potentials of the nodes lying on the electrode surfaces.50). are only partly bounded (i. 1981). theFE technique is suitable for bounded regions. Similarly tothe FD technique. . and [+lf are the free (unknown) potentials of the other nodes. the electric field components at any node can be expressed in terms of the potential values at the nodes surrounding it. the accuracy of calculation is improved with finer el~me~ts. with an assumed potential. Fortunately. This equation can be rearranged so that the potential at every point in the space betweenthe electrodes can be evaluated in terms of the electrode voltages. techniques have been developed for efficiently solving such sparse matrix equations. The first is that with the CS technique the electrode surface is not exactly equipotential. Many physical problems of interest. The secondreason is the discretization of the field space into finite triangular elements when using the FE technique. Thus (2.

For large problems.e. it involves tedious preparation of the input data. adaptive grid refinement algorithms. On the other hand. Plane triangular elements have been used (~c~hirte and r ravec. 1979). In general. 1985). Some of these difficulties have recently been alleviated by the development of grid generation algorithms. 1979). To apply the combined method. Andjelic. its set of equations usually has a symmetric positive-definite matrix of coefficients. the axisymmetric three-dimensional field of a high-voltage bushing was computed (Fig.5) (Abdel-Salam and El-~ohandes. 1984) to achieve realistic field simulation by a variable charge distribution on these elements. Tan and Steinbigler.. Steinbigler. the space is divided into two domains-the FE and CS domains-which are treated separately.1979. .17). The fundamental idea of the boundary-element method is to divide the electrode or insulator surfaces into a large number of surface elements. the time for grid preparation and editing becomes prohibitive. Elements can have various shapes and can easily be adapted to any shape of boundary and interface geometries. For curved surfaces.The flexibilityof the FE technique is its greatest advantage with respect to the traditional FD technique. and grid optimization (Shephard and Gallagher.. 2. The boundary-element method has recently been developedfor the analysis of two. The coupling of the FE and CS domains is based on the fact that the potential and field strength must be continuous along the boundary between the domains (i. A third advantage of the FE technique is claimed to be simpler programming because of the easier introduction of boundary conditions. at all the boundary points). A s an application of the combined technique. This property is not ensured in the FD technique. Further. Okon. it is more powerful for a wide range of problems with complicated geometries. A combination of FE and CS can yield the advantages of both techniques.and three-dimensional electric fields between electrodes in multidielectric media. 1985.. A comparison of the finite-element (FE) and charge-simulation (CS) techniques shows that each has advantages and disadvantages (Okubo et al. which speaks for itself.most of which is a description of the element grid topology. 1989). similar to the case of coupling CS and FD (Section 2. 1982. 1982.4. curved surface triangles in parametric form were attempted (Misaki et al. The figure shows the configuration of the bushing and the computed equipotential surfaces.

l7 Calculation of three-dimensionalaxisymmetricalfield of a highvolume bushing using a combined charge-simulation/finite-element technique. ] [From Okubo et al. (1 . 982) ./ I l Transformer .

The potential 4 . .2. The field components (Ex. zp) of the point P. I. EZ) due to theboundary-e~ement charges are obtained by differentiat~ng the potential expressed by e~uation (2..51) where KO. . The charge density distribution in each is expressed as CT = KO + + +K ~ X Y Klx K2y (2.. and 01 to u4 are thecharge density values at the nodes of the kthelement. Iy. at any point P(xp. yz). y 4 ) are the x-y coordinates of the nodes belonging to the kth element. Ixy are integralvalues expressedin terxns of the coordinatesof the nodes of the kth element and the coordinates(xp.yp.1 PlaneRectangularElements An electrode or insulator surface in the X-Y plane at z = z1 is divided into Ne elements.4. (x2.18 is The potential 4 at the point P due to all the elements is 02(xlY2Jc Y2Ix X l I y + Ixy) $ -0 3 ( X l Y l I c YlIx XlIy Ixy) 04(X2YlIc YlIx X2Iy Ixy)] (2*52) where k stands for the element number. and ( x 4 .8.52): c X Plane re~~angular element. zp)resulting from the charge distribution within one of the elements of Figure 2. K2. IC. ICl. and K 3 are coefficients related to the charge density values at the nodes of the element (Abdel-Salam et al. 1997). ( ~ 3 y3).Ey. (XI.yp. yl).

lectric Fields Y2Ixx x 2 I y x +I x y x ) 02(xlY2Icx Y2Ixx X l I y x +Ixyx) +03(. conical.and .Icy.2 Ixyx7 Ixyy.and Iyz are respectively 2 ax ay 3.53) where Icx. spherical. and torodial elements are reported in the literature ( ~ u t ~ e i s c1989). h. Iyy . . cylindrical.xlYlIcx YlIxx x d y x +I x y x ) 0 4 ( x 2 Y 1 Icx YlIxx x 2 I y x + Ixyx)l bl(x2Y2Icx Y2Ixy X2Jyy + Ixyy) ~2(XlY2Icy Y2Ixy X l I y y + Ixyy) +03(XlYlJcy Ydxy X l I y y + Ixyy) 0 4 ( x 2 Y 1 Icy Yl Ixy x 2 I y y + Ixyy)l [@2Y2Jcy (2.and Icz are respectively and a x ’ay a2 aIx arx V X Ixx7 Ixy . and 8IY Iyx.lxyz ar aIxy Xxy and are respectively 3 32 a x ay The potential and field equations for other elements such as plane triangular.and Ixz are respectively ax ay and a2 arc arc 34 ar arY.

8.54b) where E z d and Eza are the zcomponents of the electric field calculated at any point on the insulator surface when seen from the insulator and surrounding-gas sides.8.. the continuity of the normal electric flux density (in the 2 1 direction) is satisfied.4. boundary points are chosen on the ele~trode/insulator surfaces equal innumber to the unknown charge density values 17 at the element nodes.54a) and (2. 2. andthe plane part of the disc .4. i. where 0 is the value of the surface charge density at the point and is evaluated by equation (2. (2.4. The H V conductors.2 Boundary Conditions The boundaryconditions are theDirichlet condition at the electrode surface and the Neumann condition at the insulator surface.54a) At the surface of the insulator.8.4 2.4 Determination o f the Unknowns With the aid of equations (2. and the rotationally symmetric insulator ribs were simulated by ring-shaped surface charges according to the charge simulation technique. 2.4. This results in a set of simultaneous equations whose solution determines the values of the unknowns.19a. The insulator material has arelative permittivity of 3.52) and (2. The calculated potential is equal to the applied voltage Y at the electrode surface $=V (2. A ribis placedaround each HV conductor in order to prolong the creepage path. the grounded outer tube.5 Application Example The example deals with a disc insulator of a three-phase CIS and is shownin Figure 2.53) for the potential and field. respectively.51).8.3 Boundary To satisfy the boundary conditions.54'0) are satisfied at the boundary points chosen on the electrode and insulator respectively. the boundary conditions expressed by equations (2.e. Points 2.8.

The potential at any contour point on A I is equal to the sumof potential contributed by all surface charges. A I and 3 . A23. that is. One equation results from the condition that the potential must equal V on the conductor boundaries. fi represents the location of the segmental charge.19~ shows equipotential contours plotted in the plane z = 10 mm. Allthe surfaces of the problem under study are subdivided into a number of segments.. 1994).19d equipotential and field lines in a sectional plane through the points A and are shown. 013 (a). The charge density is (a). Laplace’s equation can beexpressed in surface integral form subject to the boundary conditions (a) a(a) + = (p2 = V on the interface A12 @l = @3 = V on A I 3 4 2 = 43 on A23 (63 = 0 on the infinite boundary of the third region L ! ! ! ! = 0 on the boundary surface A I (conducting surface) The solution of Laplace’sequation is obtained by formulating the following two equations.19b. In this Figure 2. As each dielectric is assumed ization charge. and a23 on the subsurfaces A I 2 . Details of the different components and the discretization are illustrated in Figure 2. By usingGreen’s theorem. The integral-equations technique is applicable for evaluating the electric potential and field in a space composed of conducting and dielectric regions (Fig.20). but of an unknown magnitude. The charge density is assumed uniform on each of these segments.4 insulator was represented by quadrangular and triangular boundary elements (Gut~eisch et al. Figure 2. Hence . by polarization charge on the interface between dielectrics and by both free and polarization charges on the conductor-dielectric interhomogeneous. assumed a12 respectively. 2.0 kV. Aik). Symmetry with respect to the plane z = 0 was taken into account by image charges. especially at the conductor with the potential 1. The unit vectors uni are inward normal to the boundary surface Ai of the ith region (Ai = A. there is no volume polarfaces. The influence of ribs appears in an increase of the field strength.

L a) Longitudinal seetion (dimensions in mm) c) Equipotential lines on the ribbed insulator 1 C*LOkV l l I I area of simulation by means of ringshapedsurfacecharges d) Field distribution in the sectional plane through the points A and B b) Discretization of the surface .

thus requiring considerably less computer time.56) is reduced to a sum of terms. The integral-equations technique is appealing because the potential must be computed only at the required points. F is the location of the contour point.fi) dA = V F on A. Equations (2. A23 becomes the collection of all dielectric-dielectric interfaces (Singer. and A23.52) and (2. G(P. It may be argued thattheequations describing the charge-si~ulationtechnique could be obtained by discretizing the integral equations explained above. un3 @ dietectric &3 F i ~ ~ 2.56) Therefore.55) and (2. each integral in equations (2. Another attractive quality is that it involves fewer unknown quantities than do the FD and FE techniques. Oncethe charge densities are known.7 Known M PC A .56) can be applied to regions composed of several conductors and several dielectrics. the potential and field at any point can be determined (Daffe and Olsen. By applying equations (2. That is . fi) = S ( j fi). 19'79). 1984). The second equation results from the conditionthat the displacement is continuous at the interface A23 between dielectrics. a set of linear equations are obtained that can easily be solved by standard techniques. on A23 (2.56) at several contour points. In this case AI becomes the collection of all conductor-dielectric interfaces. fi) = 1/4@ a)is a generating function that satisfies -V:G(F. a(fi)G(F. 1984). that is. (2. each being a constant depending on the system geometry multiplied by an unknown charge density a(@.55) and (2. where S is the Dirac delta.55) where A is the collection of subsurfaces AQ A13.20 r e System consistingof a conductor and two dielectrics. The constants for any problem can be evaluated using analytical and numerical integration schemes (Singer.

+ x ) . Let the space between the conductor and the duct be ruled off in a Cartesian grid.57) Each Vn will equal I/ or 0. Then the potential at P is defined as Vp = lim n+w y1 (2. . so many random walks are necessary if a fine grid is chosen and reasonable accuracy Y Random walks in the Monte Carlo technique. Laplace’s equation for the electric potential is solved here by numerically simulating random walksof a fictitious particle beginning from a point chosen in the space for which the potential distribution is to be evaluated. choose at random one of the four possible directions and move the particle ahead one more step. Again. but it would result in a larger number of linear equations tosolve than the integral-equations solution. + / ‘ G . in which the sources are treated as surface charges. 2. as for the finite-difference technique. this walk would reach either the conductor or the duct wall and would thus be terminated. and move the particle one step from the chosen point P in the direction thus identified (say. a conductor at a potential of +V centered in a rectangular duct at zero potential will be considered (Fig.21). Eventually.18). depending on whether the corresponding walk terminates on the conductor or on the duct wall (Fig.21).true. Choose at random either of the letters x or y and either of the signs + or . . The rate of convergenceof Vp to the correct valueis proportional to . The path traced out by this sequence of random directions is called a “random walk” (Fig. This random walkof the particle is repeated a fairly large number of times (n). To illustrate the application of this technique. 2. 2.

However. 1989). The first is Poisson’s equation.or two-dimensional problems because the time needed to calculate the potential at each point is not small. In high-voltage engineering. however.Abdel-Salam and Shamloul. unipolar DC lines). nordo a large number of simultaneous equations need to be solved. Analytical calculation of electric fields with space charges created by DC corona is possible for simple configurations such as concentric spheres and coaxial cylinders.This increases the difficulty of field calculation (Abdel-Salam et al. they are very appealing for solving three-dimensional problems. Monte Carlo techniques have been of only limited use for one. the second defines the relationship betwee and E. The most interesting characteristic of Monte Carlosolutions is that the potential can be computed for each point at a time.is required.58) where E and P are the electric field and current density vectors at any point in space. Methods for generating these numbers are described elsewhere. On the other hand.Again. where they tend to supersede other techniques (Anis and Abd-Allah. there are techniques that can be used to apply the Monte Carlo technique to unbounded regions. In more complex electrode geometries such as wire-plane gaps. it is assumed that the region between boundaries is circumscribed. the space charge is constrained in the vicinity of the stressed electrodes and oscillates with periodic reversal of the electric field. At alternating voltages. and the . 1994).g. analytical calulcation of the electric field is extremely difficult because of the nonlinear nature of space-charge generation by corona. Despite this advantage.. Abdel-Salam and Abdel-Aziz. The sequence of instructions that is usedto govern the motion of the fictitious particle is derived from a sequence of random numbers. The equations in this case are as follows: (2. 1984.1988. space charge is mainly developed by corona discharges at highly stressed electrodes. Neither doesa large array of potentials need to be stored. A relatively simple case is that of unipolar space charge (e.. the space charges created by DC corona fill the entire inter-electrode space. Aswas true for the FD and FE techniques. the third is the continuity equation for the ions.

58) and (2. at a value to be determined experimentally. lement T e c ~ n i ~ u e By using some additional assumptions. The two most important such assumptions are: 1. they postulated a constant space-charge density around the wire periphery. Also. Takuma et al.fourth is the electric field in terms of the potential @. 1973) and measured. Abdel-Salam et al. bipolar DC ionizedfields are described by the following e~uations: where the subscripts + and correspond to positive and negative ions. 1982). Instead of assumption 2 above. Both assumptions compensate for thelack of adequate boundary conditions. p is the ion charge density and k is the ion mobility. Computation may follow the finite-element or charge-simulation technique either separately or combined with the method of characteristics or the method of residues. Similarly. not the direction of the electricfield. Iterative methodshave been used for the conditions tobe closely fulfilled at the electrodes. It was found to decreasesignificantly with corona intensity. This added some empiricism to the analysis. Deutsch's assumption continued to be a basis until Abdel-Salam and Khalifa (1974) showed that its use was not a necessity. The surface field at the coronating electrode remains constant at its value corresponding to corona onset. the surface field of conductors in corona could be calculated (Khalifa and Abdel-Salam. and thus equations(2. This assumption wasoriginallysuggested by Deutsch in 1933. as shown below. 2. They based their computation of the resultant local field on a physical understandi~g of the problem as a superposition of the applied electric fieldand thespace charge.59) can be solved along a number of flux lines of the imposed electric field. Most attempts in theliteratureresort to simplifying assumptions (Khalifa and Abdel-Salam. 197413. The space charge affects only the magnitude. . 6 denotes the recombination coefficient for ions and e is the electronic charge.. (198la) succeeded in applying the ~nite-element method to a field problem with space charge in the case of wire-plane gaps.

To solve equations (2.61a) in their general form (2. thus forming a grid.64) is usedinstead of equation (2. .61) Abdel-Salam et al. and the elemental potential components. The potential was computed for the contour points by the charge. and according to the finite-element technique. . the inter-electrode space is an unbounded region and is usually defined by an artificial boundary as explained in Section 2. . Abdel. to find & substitute equation (2.4.63). Now. (1983. In wire-plane gaps. They thus could compute the field in the interelectrode space without empiricism.58). They were able to eliminate both Deutsch's assumption (l) and that of constant surface field intensity (2).61a). an energy functional given in equation (2.35) of the space-charge-free fields.More recently. = # .and imposing the minimization condition ~ F /= 0 ~ for ~ the ~ c $ . that is.62) into equations (2.61a).1993.62) The elemental potential 4scis bound to be zero at both electrodes. simulation (CS) technique as explained in Section 2.60a) and (2.64) Representing the iariation of (bsc over each element in terns of the nodal values of # .61) to get Qi = 4 s + rbsc (2. From equation (2.60a) and (2.1992. we get a set of simultaneous equations for c .(1983) considered the potential SP as the sumoftwo . (2.1995a) applied the finite-element method for analyzing the electric field withmonopolar space charge in wire-plane gaps. and p is equal to p and zero in equations (2.60) (2.61a) or. )0 * (2. F= (2. . the applied electrostatic value # caused by the space charge (psc. where Q. respectively.63) where y is equal to I and kp in equations (2. V ( Y V ~ s c= ) B (2.60a) (2.5. at the entire set of elements.60) and (2. and the potential yjsc within each element was approximated as a linear function of space coordinates. in a general form.alam et al.4.1994.3.60a) and (2. we get V ( k p ~ Q= .The inter-electrode space is divided into triangular elements. respectively.

the partial differential equations (2.35) is obtained from the general equation (2. The method of analysis was estended to bipolar ionized fields (Abdel- A self-consistentdescription of the charge density p and potential dD structures of a monopolar G ionized field was proposed by avis and Hoburg (1986) on the basis of simultaneous computation of both p and @. Use ofthe finite-element method to determine the distribution of ( Pfor a known charge density p in a region with permittivity E is based on minimizing the energy functional [?E(grad l p(P]d. Once the @ distribution is determined.. Iterations are used until the values of both sfScl and sfsc2 agree within an acceptable tolerance (Abdel-Salam et al. is noted that the energy functional given by equation (2. is an acceleration factor chosen equal to 0. Gombination of equations (2. According to the method of characteristics proposed by Davis and oburg (1986). Solution of this set of equations gives the nodal values of #sc: [QlScl] and [qJsc2]. corresponding to equations (2.58) that describe the effect of a known field pattern on the spatial distribution of p can be converted into an ordinarydifferential equation for p along specific space-time trajectories. and the method of characteristics to compute the charge density distribution. the electric field l ? distribution is obtained simply.61a). The technique employs iterative useof the finite-element method to compute potential and electric field distributions.60a) and (2.66) to obtain the solution of Poisson’s equation (V2@= --p/&) It .5. The relation for correcting the charge density pito pnis (2. 1983).nodes for a given p distribution. respectively.67) Along the “characteristic lines” defined by .58) leads to a nonlinear partial differential equation governing the evolution of p: (2.66) by setting p = 0 for space-charge-free fields.65) where f a . (2.

Since these equipotential shells do not contribute to the surface field of the wire.4.22 for a wireplane geometry. He concentrated them at some discretepoints inside the region under study. The contribution of these charges to the totalelectric field should satisfy the condition of current continuity.58). Then.67) becomes (2. as depicted schematically in Figure 2. Their images with respect to ground are also set. time t is determined. This field represents conditions just prior to corona onset.70) to compute p along the characteristic line. These boundary conditions are given partially in terms of known electrode potentials and are obtained partially by assuming that the magnitude of the electric field at the coronating electrodes remains constant at the onset value or varies in the manner cornputed by Khalifa and Abdel-Salam (1973).69) and yields the solution (2.3). To solve a given unipolar DC ionized field. 1984) to include corona space charges. The M selected contour points are also chosen spread about the equipotential contour as shown in . where p For a given starting point at the surface of the coronating electrode.68).68) equation (2. At intersections with triangle boundaries. equipotential charge shells enclosing the wire and represented by discrete line charges are added to the volume. then used in equation (2. approximate In the application of the CS technique to space-charge-free solutions to Laplace's equation are obtained by placing fictitious discrete charges outside the region in which the solution is sought (Section 2.= kiE dF dt - (2.. is the charge density at the starting pointof the characteristic line. The equipotential chargeshell nearest the wireis first located and simulated by M line charges uniformly distributed around its contour. the field remains at its onset level. fields. for a given value of corona current i. This technique was adapted by Hornstein (1980. the characteristic lineis traced out by numerically integrating equation (2. Both sets of discrete charges have to be chosen to satisfy the boundary conditions of Poisson's equation (2.70) . first a space-charge-free field solution is found.

rj (2. M equations are formulated as M j=l 1 &In = constant at each T.22 and located midway between the line charges.7 l) where 7. as shown in Figure 2. respectively. 2m0 r . Figure 2.72) where drj and Alj are distances defining an elemental volume withunit axial length around qj.22. Under the constant levels of the equipotential contours. The quantity (qj/dr~Alj) is the . and Tj are. The values of qj should also satisfy the discretized form of the current continuity equation M j= 1 ArjAlj ‘j kEAlj = ic (2. discrete space charge. and ~ q u i p o t e n t i a contours ~ for a wire-plane gap (not drawn to scale).Schematic representation of trajectories. the location of the contour point and line charge qj.

The simulation charges l ‘ [ of the electrodes are simply expressed as (2. Equations (2. (2. The method was extended to the ionized fields around monopolar bundle conductors (Abdel-Salam and Abdel-Sattar. such as in point-plane gaps.73) The potential at any point in space is then calculated as n j=1 n. the number of simulatio~ charges of the electrode and the number of discrete space charges. The electricfieldis calculated similarly. The concept above was used for simulating electron avalanches growing in the ionization zone around stressed electrodes. The applied voltage necessaryto sustain ic is equal to the onset voltage augmented by the potential contribution of each shell plus the potential of the images calculated at the wire surface. The potentials of contour points selected on the electrode surface are expressed as (2.and ic is the corona current per unit length of the wire with the applied voltage V . This defines the charge-density distribution over the space surrounding the coronating wire. After the first equipotential charge shell and its “image” charges are found.74) j=l where yz and y z .10) where [P. The I-V characteristics of corona computed by the method proposed by Hornstein (1980. 1998).] is the potential-coefficient matrix of the [qs]space charges and [ vl is the applied voltage.equivalent local space charge density p . ~bdel-Salam and Mufti. are.72) formulate M equations with M unknowns. ecently.71) and (2. 1984) agreed satisfactorily with those measured by experiment. respectively. The ion cloud may beapproximated by point or ring charges (Zeitoun et al. is of known magnitude and travels in an inter-electrode space. 1976). 1982. Sometimes the space charge. their electric field contributions are included in integrating outward to the next shell. it is only necessaryto calculate the charges simulating the electrodes. As the space charges are known in magnitude. 1984).. 1989. such as a cloud of ions developed by an avalanche. This was the basis for calculating the development of corona Trichel pulses in negative corona and burst pulses and onset-streamer pulses in positive corona (Ibrahim and Singer. and so on until the entire space is filled with shells of known discrete simulation charges. the concept . to be solved simultaneous~y.

The CS technique combined with this method of weightedresiduals has been used successfully in solving monopolar and bipolar DC ionized fields (Oin et al. The space charge was represented by discrete values at the nodes of a grid that was superposed on the inter-electrode space.was used to calculate the corona pulses from positive surges propagating along overhead transmission lines (Abdel-Salam and Stanek.. high-voltage bushings. The compocomponent g. The charge-simulation technique wasextended to the calculation of the electric field vector l? directly from Poisson’s equation. where l? is obtained by numerical differentiation of @. the applied electrostatic due to the space charge. of the current continuity at each node is determined as R.. 1986). 1986). In the presence of corona. For this purpose.4. The region of interest being defined by an artificial surface is subdivided into small triangular elements. This method is more accurate in the calculation of l ? than would be possible from an FE solution of Poisson’s equation for space potential @ . Section 2. the space charges p at the grid nodes are computed knowing l?. and the deviation is used to increase or decrease p appropriately to speed up the convergence.. = V kpE * (2..75) Setting the integral of R. The values ofthe electric fieldat the wire surface in each iteration cycle are compared with the onset value. and the Component gSc nent l?s is computed by the CS technique.3. Measures have been suggested and employed with the aim of optimizing the stress .. component ESc To complete the iteration cycle. the electricfield E consists of two components..A weighting function for each element can be included in the integration of R. over the region of interest to zero leads to a set of algebraic equations that can be solved for nodal values of p.. The updated estimates of the space-charge values are obtained from the continuity equations for current density. Field stresses are controlled in much high-voltage equipment such as cable teminations. 1987a). The applied voltages are defined at ground and wire surfaces in wire-plane gaps. The can be easily computed from the point form of Gauss’s law. the residual error R. as explained in. and potential transformers. using a weighted residual method (Oin et al. These chargedensities are utilized for thecalculation of g in the next iteration cycle.

2. Another example is a high-voltage line gripped for suspension through bushing without floating screens. a tank of a high-voltage transformer.. the highest stress would be at the lowest r. the stress-controlledcapacitor bushing will be discussed briefly to shed somelight on theproblems involved. Consider a cylindrical conductor of radius a with a cylindrical bushing of length I and thickness t (Fig. thus creating a uniform radial field. the clamp holdingthe bushing is grounded. then the required thickness of the bushing is obtained as or (2.77) Grounded clamp v=ov Bushing 3 High-voltage line gripped for suspension through schematic view and cross-section. The electric stress in the bushing at radius r is + E= r + a)/a] (2. A capacitor bushing is used to bring a high-voltage conductor through the ground case.23). Thisisachieved in practice by the introduction of floating screens separating dielectric layers thatare interleaved at equal distances between the conductorand the grounded case. at r = a. For example. If E b d is the breakdown strength of the bushing matkrial.e.7 throughout every part of the insulation of high-voltage equipment in order to achieve the most economical designs.e. The bushing of a uniform dielectric should have a hyperbolic surface profile. i. for example..76) Clearly. the voltage at r = a is V and at r = a t it is zero. bushing- . without excessive electric fields between the conductor and the edgeof the holein the tank. Usually.i.

Therefore. High-voltage insulators usually serve as supports and spacers of HV electrodes with respect to earthed frames. The concept is to make the stress across each capacitor the same so that the bushing material is properly utilized.79) Thus the bushing of a uniform dielectric should have a hyperbolic surface profile (Fig 2. 1996) rz = constant (2. the tangential field should be kept below the critical level Cylindrical shell of thickness length z z a l Conceptual radius F h r =Constant b 1Actual . The capacitance in each section of thickness dr is ~2nrz c=2-constant dr (2. 2. the bushing material is properly utilized and the bushing thickness is equal to that of the flat bushing.24). The highest stress isclose to the conductor and the bushing is not stressed to the limit elsewhere. A s the capacitors are connected in series.24). each of thickness h . In graded bushing.78) This implies that the bushing length must satisfy the curve for the profile (Hoole and Hoole. Flashover takes place along the insulator surface if the potential field is high enough to sustain a discharge. for the same stress across each capacitor. the stress is uniformly distributed. Also. This isachievedby shaping the profileof the bushing material so that the length of the bushing is 2 2 at radius r (Fig. the voltagedifferenceisthesame for all bushing capacitors. The bushing is considered as a series of parallel-plate capacitors. which in turn implies the same capacitance. they all carry the same charge. but also the stress is not uniform within the bushing materials. the same as for bushings with floating screens.which is much larger than the thickness of a flat bushing whose thickness is Not only is the cylindrical bushing costly. as in gas-insulated systems or the ground plane in air (see Chapters 9 and 10).

27). Recently. Other optimization techniques for electrodes are based on an ‘40ptimized”simulation charge obtained by minimizing an objective function (Moller and Yousef. 1983). At any point on the insulator surface the tangential field is derived from the potential difference between two test points close to the point under consideration.. If the tangentialfield exceedsthe set level.80) where S S‘ = [r2+ I )~ 2r D C O S ~ I ~ = [b2+ r2 2rb cos 6 1 4 .26). ~ o Z ~ t iLet o ~ the : image charge q’ be located at a distance b from the sphere center to satisfy the zero potential at r = R (Fig. Locate the image of the charge q inside the sphere and determine its value. The potential @ at any point P outside the sphere is (2. thus raising the withstand voltage of a given gap (e. 1984). (1) A point charge q is located at a distance I) from the center of a conducting sphere of radius R at zero potential. 1987b). a method developed to calculate the tangential field component was applied to mathematical expressions of the profile to be corrected. Optimizatio~of an HV insulator design is achieved by correcting its profile in elemental steps. The method wasbasedon a modified charge-simulation technique whereby the magnitudes of the simulation charges were determined by minimizing a novel error function formulated over the HV electrode and insulator surface (Fig. either the potential difference is decreasedor the distance between the test points is increased (Gronwald. tominimize the cost of material and installation. 2. Shaping of high-voltage electrodes within insulator systems is aimed at bringing the surface field stress down to the minimum possible. Such electrode optimization is based on an interactive process of charge simulation in which the contour points are shifted after each computation of field values (Fig.g. 2.lectric Fields everywhere along the insulator surface while keeping the overall length of the insulator at the minimum for agiven voltage rating. 2.25)(AbdelSalam and Stanek. a limited space inside a highvoltage laboratory).

e. [From Abdel-Salam andStanek (1987b).] At r = R.2 L . the potential @ must be zero.81) must be true for all values of B..8 1) Since equation (2. (2.82) Eliminating q and q1yields a quadratic equation in b: . q2 b = q12D + q22RbCOS B = qf22RD cos B. 6 Dtstance 2 (cm) 8 Tangential-field distribution along a profile for optimized and nonoptimized insulators. so that q and q' must be of opposite polarity. q2(b2 > R ' + = qf2(D2 R2) i. (2.

optimized contour: field value values = 1.l7 kVlcm. is (2.igur Optimization of arod-planegap.07= 1. Bordaprofile:fieldvalues = 1.25 kVlcm.83) whose solution.25-1. b2-bD l + [ (3'1 - +R2=0 (2.84) P 7 potential. hemispherical cap: field 1.52 kV/cm. Image of a point charge q outside a conducting sphere at zero .

86) Substitute in equation (3): q ' = --qR/D (2) A line charge h is located at a distance a from the axis of a conducting cylinder ofradius R at a given potential. ~ o Z ~ t i The o ~ : image charge -h is located inside the cylinder (Zahn. y).85) (2.87) + + For the cylinder to be equipotential (x .88) which is the equation of a cylinder whose radius is equal to R Y . l 2 Y 2 = K1 (xa)2 y2 + + + (2.28). the potential @ due to the charge h and its image h R2 @ =-ln27t~o R1 h = -In 2 ~ (x+a)2+y2 ~ ( x o a)2 y2 (2. At a point P(x.el-Sala b = R2/D (2. Locate the image ofthe charge h inside the cylinder and determine its value. 1979) at a distance 2a from the charge h (Fig. 2.

.91) (2.93) igure 2.89) and (2.92) (3) Calculate the electric field at a point along the axis of a planar ring charge of radius b and charge density p l .3 (2.e. ai- a(K1 K 11 + D = 2aKl K1 1 __ Solve equations (2. The potential @ at an arbitrary point P along the 2 axis is (2.29 Electric fieldof a planar ring charge on the X-Y plane."--2D The image charge is spaced a distance b from the center (2. solutio^^ Let the ring be located in the X-Y plane (Fig.29).90) +l i.90) simultaneously to obtain D2 R2 a=-. 2.89) with center at x(=D-a)=a"----K1 K 11 (2..

95) where qr is the value of the ring charge (= 2nbpJ. Thus. assuming that the charge -4 is uniformly distributed over the surface of the sphere. This is utilized in the power industry to repair high-voltage lineswithoutever switching off the line. but one can surmise that the components other. D) (Fig.94) implies that 8 may have up to three components.30). 2-30). the field dueto the ring charge has one component along the Z axis [see example (3)]. The ring charge is (q/4)za2)2na2sin 0 d0 (= q sin 0 d0/2). A conductive cage suspended from pulley-likerollers travels down line using the latter as rails. (4) What is the electric field caused by a charged shell (Faraday's cage) of radius a and charge q at a distance D from its center? S o Z ~ t i oSlice ~ : the shell into a ring at an angle 8 and of thickness ad0 (Fig. The radius of the ring is a sin 8 and its surface area is 27ra sin 8.98) The latter equation is a very important result with practical a~plications. is (2. = -V@ (2.96) Take the positive root only. 0.The electric field generated by the ring is related to the potential @ by ii. At the point P(0.. the z component of? ! i . than the z component of l ! ? are zero at all points on the Z axis. (2. and workers repair and service the line from inside.94) Equation (2. 2. the fact that they are in a conductive cage prevents a voltage difference (i.e. Although they are dealing with high voltages. It means that there is no electric field in the shell (Faraday's cage). since this charge distribution is symmetric about that axis. * (2. an electric . ad0 = 2na2 sin 0 de.

Calculate the maximum field at the sphere surface.00 128 0. 7 1 8 Q ' = 4. the charge deposited on the conducting body of the car raises it to a high voltage..25 " 0 (Fig. Field at surface .067 -~ l [ 2ZEo (0. A sphere gap consists of two spheres with R = 0.31.25)2 (0.25 0.~ D-a cos0 Y Building up rings into acharged shell field) from developing across their bodies with the associated health hazards.75 0. The gap between their surfaces is 0.0048 0.067)2 +.l = 1 8 . The same result explains why people in a car are unaffected when a bolt of lightning hits the car. 2JtEo (0.3). Solution: The charge of the stressed sphere (sphere 1) Ql= 4m... However.01795 0. so that it is dangerous for someone inside to step out to the ground. + + [ " + c! 0. E (i.065)2(0.067)2 (0.. Distance between spheres centers is 1 m.067)2 + +.75 0.25 + 0.067)2 (0.75 0. Calculate the charges and their locations to make the sphere potentials l and 0.25 m each. at point P) 0.RV = m 2.5m..6795V/m per volt 4nro . Table 2. since now each foot of the passenger would be at a different potential.e.

(c) the average maximum surface voltage gradient of the bundle under (b)(ii).45 = 0.32.088 74 9 ~[ . 24 400 x lo3 = In----" In2x80 0. = 40.0175 x 0.088 74x11 A s the applied voltage is 400 kV where H = bundle height and d =pole-to-pole spacing. one is stressed by unit voltage and the other is grounded. minimum. and average surface fields on the subconductors (i) omitting the charges of the second pole and image subconductors. (b) the maximum. (ii) considering the charges of the secondpole but omitting the charge on the image subconductor. Calculate: (a) the charge on each bundle and on each subconductor. ~ol~tion~ (a) Replace each bundle by an equivalent conductor of radius re.i"---s"----J I I I Successive imaging between two spheres. (6) The dimensions of a f400 kV bipolar DC line are shown in Figure 2.

25)2 = 0.25Q.067 _ I 7 Q : = 0.67 =10.25)2 10.25 1-0.017 95 Q1 X 0. l .25Q1 (0.067Q1 (0. Q2 = QlR/s = -0.45 = 2607 kV/m + Maximum surface field = = 2412 kV/m Average surface field=.067 Q. = -0.25 10. : charge per bundle Q = 4.067 .067 =10.925 x IO3 ( 7 0 0175 0.0048 Q1 X 0. = -0.067 = 0.067 Q. R2 = (0. : Maximum field = 43.001 28 Q1X 0. EOl (field at outer point of subconductor ##l) .067 10.25)2 X 0.067 =10.0625 =l 0.25 (0.'= (0.067 Q .25)2 10.25)2 10. 2.88 pC/m charge per subconductor q = Q/2 == 2.l = 2510kV/rn 2nrO r (ii) Consider the two subconductors on the left (Fig.25)2/1 S Q{ = 0.32).67 = 0." (0.25 Sc.44 pC/m (b) (i) Maximum surface field = where r = subconductor radius and R = subconductor-to-subconductor spacing.0625 = 0.7 Charges in Example 5 Charges inside sphere 1 Magnitude Distance Magnitude fron sphere center Q1 Charges inside sphere 2 Distance from sphere center S1 = 0.

(b) the line after replacing each pole by an equivalent conductor.I (a) A bipolar DC line with two subconductors per pole.43 kV/m . E o 2 (field at outer point of subconductor #2) 1 d .R -dr.r + 1 = 2617.El1 (field at inner point of subconductor ##l) = 2422. = 2598 kV/m .6 kV/m .

8 kV/m This is almost equal to the average gradient obtained by omitting the charges of the other pole.18 x 10gV/m This example serves for field assessment in a wire-duct electrostatic precipitator whose wire has a charge q per unit length. but are of opposite polarity. A line charge of lC/m is placed at x = 0. (7) Two infinite conducting sheets lie on the planes x = -1 and x = 1.33. This in turn has an image of 1 C/m in the left sheet of -1C/m at x = 6. ] 7 2 E E O q 5 l l +-[-+9+*.39 kV/m The average maximum gradient is defined as the arithmetic average of the two maxima = (2598 + 2617)/2=2607. the charges on the other phases are passing 50% of their peak values.] . . ]+ 4 [ 1 + 5 + 9 . This has an even smaller effectthan what has been shownfor the bipolar DC line. it results in an infinite sequence of images. For a three-phase AC line.] 2nEo --[-i7+. The duct is represented by the two conducting sheets.2n&o q 3 l l = 27. as shown in Figure 2. when the charge on the conductor of one phase is at peak value. By superposition. . Determine the electric field induced at x = 1. In general. Solution: The line charge q(= lC/m) has an image of lC/m in the right sheet at x = 2. the effect of charges on other phases can usually be ignored because. where the charge of the second pole is equal and opposite to the charge of the conductor under consideration... . [-+-+ . .E12(field at inner point of subconductor ##2) = 2422. the total electric field is 1 1 E = " 2nEO 2nEO 3 1 1 ..

95 -[ + Therefore. The inner core is at a potential oflOOV and the outer sheath.The strength of the bushing material is 50 kV/cm.68cm Thus. the volume of regular design V2 = n[(2 14.33 Images of the line charge causing successive images. has to be 20 cm long outside. re 2.4cm2 + The volume of graded design V I = 2(2nr)z dr = 4n x 62.34a. (9) Find the potential at the grid nodes of the rectangular cable of Figure 2. Theinner core and the sheath are 2 x 2cm2 and 6 x 6cm2 in cross-section. as obtained form equation(2. I I I 1 I I -4 I t 6 -6 -4 -2 -1 0 1 2 1 .24 2) = 62. respectively.24 cm The bushing profile follows the curve (Fig.76 cm The thickness of regular design.4(6&4 2) = 3324. is at zero potential. ~ o l ~ t The ~ o thickness ~ : of graded design = thickness of flat bushing = V/Ebd = 1501/2/50 = 4. for mechanical reasons. 2.77) t2 e150Jz/2x50 l] =TI 14.24) zr = lO(4. This saving in insulating material is added to the proper utilization of the material where the stress is uniformly distributed.(8) Design a graded cylindrical bushing that.68)2(2)2] = 8614. The conductorinside the bushing has a radius of 2 cm and is stressed by an AC voltage of 150 kV. the volume of the graded design isabout 38% of that of the regular design. . being grounded. Calculate the voltime of insulator required for the graded design and compare it with the volume for regular design of the bushing.

i. = (Pi and equation (2. the potential is specified and equal to zero.100) For the other nodes (2.101) (2. an equation is necessary since the potential is unknown. equation (2. 6.103) .4). A fictitious node 2' is introduced as shown in Fig.32) takes the form (2. Thegrid points are numberedcolumn by co1umn from top to bottom.e. on the boundary norqal to which the potential does not vary. 13). i. Therefore.102) (2. A fictitious node 4' is introduced as shown in Fig. i. For nodes like (1) on the boundary normal to which the potential does not vary. equation (2.32) is reduced to Similarly. 2.. 15). 12. in the X direction Such nodes are surrounded by only three others (nodes 4.34b. the Y direction.e. 2. a@/aYlnode 1 = 0 Such nodes are surrounded by only three others (nodes 2.3..lectric Fields 71 ~ o Z ~ t iA o~ symmetric : quarter of the cable is considered and a square grid ismapped..e. the potentia1 is specified and equal to 100' v .34~. for nodes like ( 9 . For the grid nodes (6-12) located at the surface of the grounded sheath.32) takes the f o m (P4 $ 5 1 = Z((P12 $ -2(P4 + (P151 (2. For the grid nodes (1 3-15) located at the core surface.

103) which are solved simultaneously to obtain the unknown potentials at the grid nodes It is satisfying to observe the symmetry in the calculated potentials where $1 = $5 and $2 = $4. the set (2. 12 we 2.33) is expressed by equations (2. calculate the potential and field around a hemispherically capped cylindrical rod stressed witha voltage . (b) and (c) fictitious nodes introduced to satisfy boundary conditions. Thus.99) through (2.7 2cm I- I lcm I I Core at #=l00 i 4 " " e .34 (a) Finite-difference grid for a quarter of cable. (10) Using the charge simulation technique.

. The potential Sti at any point (r. images of the simulation charges with respect to this plane are considered. OkV with respect to a ground plane (Fig.j = 1. The charge on the rod surface is simulated by a point charge (Qp) located at the center of the hemispherical cap having coordinates (0.7 17 = l . H R) in addition to a series of N semi-infinite line charges ( Q j . ~oZ~tio Choose ~: a cylindrical coordinate system with the origin located at the intersection of the rod axiswith theground plane (Fig. N (AbouSeada and Nasser. selected on the cylindrical portion. j= 1. z) is + + + (2.. Ai). z) = - 1 J m -J m 1 1 1 1 = -1n 47tE0 Aj' + z f + J r f 2 + (A.. i.z ' . z.. zf + -4 where r f . the number of unknown simulation charges is N 1. is equal to the applied voltage.35)...2. The rodradius R is 1cm and the gap spacing H is l m. 2.. . r . . H. + . respectively. and Aj normalized to the radius of the rod.104) j=l where the potential coefficients P(r. Thus.35). rf = r / R and so on.+ z ' ) ~ A. To maintain the Z = 0 plane at zero potential.e. N) along the axis of the cylindrical portion starting at points of coordinates (0. The boundary conditions at the rod surface are described as follows: (i) the potential at (N 2) points of the z coordinate designated zi 2 H R. Aj 2 R H. and Aj are. 2. 1968).2. H'.

(2.z~)= V.Ai).105a) at the ( N 2) selected points on the rod surface and equations (2. evaluated at the rod tip"(@ =O ) .2.. i.. Satisfaction of equations (2.P4o(O).105d) where P26(0).104). H) = v (2.105b) To satisfy the above condition at all the hemispherical portion defined by the angle 0 1 : 4 2 . fourth. )of 6. Zi>R+H. (0) with respect to 0. @(R. the even derivatives (second. . i = 1. With the d e t e ~ i n a t i o nof the unknown charges. are set equal to zero. and 1.T46(0.... and the field is obtained as + + .105d) at the rod tip results in ( N 1) equations whose s i ~ u l t a n ~ o u solution s dete~ines the unknown (N 1) simulation charges.. @ ( O .105a) (ii) the potentialat the tipof the rod on thehe~ispherical portion is also equal to the applied voltage.e.105b) through (2. U26(0..35 ~emispherical~y-capped cylindrical rod against grounded plane.Aj) have been expressed by Abou-Seada and Nasser (1968).10%) (2.. N-2 (2. the potential at any point is obtained using equation (2.

1 A = "(0.7 + Hence.0857 1 AY2 = 1.3.54 0. Usingthe finiteelement method. The potential and field along the gap axis are shown in Figures 2.7 AX3 = 0.4.38b.4 AY2 = 0.4571 0. ~ o Z ~ t i The o ~ : element stiffness matrices are obtained using equations (2.4571 -0.3667 0.0857 -0.9.236 -0. AX3 0.16) = 0.91 0. 0.consisting of nodes 1-2-4 corresponding to local numbering 1-2-3 as in Figure 2.38b.46).36 and 2.828 -0. AY3 = 0. AX1 = -0.3714 Similarly.3. (1 1)Consider the two-element mesh shown in Figure 2. for element 2 consisting of nodes 2-3-4 corresponding to local numbering 1-2-3 as in Figure 2.525 L AY3 = 0.Electric Fields (2.4571 0.7786 0. AY1 = -1.4571 -0.1 = 0. respectively.667 In the lightof the fact that nodes 2 and 4 are offree (unknown) potentials whereas nodes l and 3 are of bound (known) potentials.6.14) = 0.46a) gives + [Sf''= 0. For element 1.38a. AX1 = -0. [ 1.4667 0. AX2 = 0.6929 0. 1 A = T(0. .2.9.6 AX2 -0. determine the potentials within the mesh.2.7786 -0.106) where iir and tiz are unit vectors along the r and z directions.35 2 Substituting into equation (2.5571 -0.37. AY1 = -0.

o 0.1 1 .4 0. .9 0.2 0.49) E q f [(PI. L 0. the submatrices stiffness matrix [qE and [qfb are expressedin terns of the global According to equation (2.1 Axial f i e l d "[Sljb[(P~b 20 40 60 80 z l " " 100 R Field along the axis of the rod-plane gap.6 0.3 t0.7 A x ial potential 0.7 0. S 0.9 0.5 0.6 0. 6 0 70 80 90 1oQ110 z C Potential along the axis of the rod-plane gap.2 0.7 0.3 0. x 100 kV/m 1.8 0.8 0.

25 S 2 4 =S y 3 + SE) = 0.8) (l. $ 3 = 10 .l) ( 1 2.0143 = S42 S 4 4 = Si:) + Sg) = 0.Z.lectric Node: 1 2 f x.7786 S 2 3 = SE) -0.3714 + 0.G) 3 (2.l.0857 0.G. yf (0.7) L 4 L 3 I 2 2 (b) (b) local and globalnumbering of the (a) Two-elementmesh.6929 + 0.1 = -0. 1994) as S 2 2 + Sf = 0.4667 = 0.8 1. element.4571 S 4 3 $1 = S:) = -0.3667 = 0.2.8381 = S!$ S 2 1 = St: = -0.4571 S 4 1 = Si:) = -0.1.49a) The elements of the global stiffness matrix are obtained (Sadiku. (2.5571 = 1.

Abdel-Salam M. Pittsburgh. El-Mohandes ThM. 1987b. Abdel-Salam M. Shamloul D. El-Mohandes ThM. J Phys D: Appl Phys 30:1017. 1974. 1989. IEEE Trans El-26:669. Al-Hamouz Z. Abdel-Salam M. Abdel-Salam M. Abdel-Salam M. 1998. 1983. J Phys D: Appl Phys 26:2202. TN. 1993. ~bdel-Salam M. 1994. PAS-100: 1806. J Phys D: Appl Phys 25:1318.. TEE Proc-A 141:369. Abdel-Salam M. S.IEEE Trans.014 One obtains (B2 0. Mufti A. Simulation. 1995a. Ibrahim A. 1. Abdel-Sattar S. Cendes ZJ. Proc IEEE 56:813. 1994. Abdel-Salam M.3667 1 = 3. 199513. IEEE Trans lA-31:484. 1984. pp 116-121. the potential at any point within the mesh can be determined using equation (2. p 1677. Abdel-Salam M. Abd-Allah M. Khalifa M.8381 -0. Ahmed A. p 492. Abdel-Salam M. Abdel-Salam M. 1992. Acta Phys 36:201. 1989.014 and ][ "1 [ #4 Cb4 = -0. Elect Power Sys Res 44:145. IEEE Trans EI-18:110. . J Phys D:Appl Phys 25: 13 18. Stanek EK.4571 -0. Anis H. September/October 1989.708 = 4. Proceedings of 4th International Symposium on Gaseous Dielectrics. IEEE-PS paper A-77-131-6. Abdel-Aziz E. 1987. Athens. J Phys D:Appl Phys 20:629. Abdel-Salarn M. 1982. 1987a. Al-Hamouz Z. Abdel-Salam M. 1988. Abdel-Sattar. 1992. Nasser E. Abdel-Sattar S. Proceedings of International AMSE Conference on ~ o d e l l i n g and p 299.7786 -0. Farghally M. Abdel-Salam M. Abdel-Salam et al. Abdel-Salam M. West Germany. 1977. IEEE Trans 1A-31:447.1987. 1997. Abdel-Salam M. Paper 33. Proceedings of IEEE-IAS Annual Meeting. J Phys D:Appl Phys 27:2570. Stanek EK. Al-Hamouz Z. Abdel-Salam M. Abdel-Salam M. IEEE Trans PAS-101:4079.P ~ubstitutingin equation (2. Vol 21 .25 -0. Al-Hamouz Z.49a). IEEE Trans IA-25. Abdel-Salam M. IEEE Trans EI-22:47. Farghally M. Singer H. Al-~amouz Z. Proceedings of International Symposium on High Voltage Engineering. PA. Proceedings of Middle East Power System Conference. 1997eada M.1 1. IEEE Trans IA-23:481. Abdel-Salam M. Cairo. Abdel-Sattar. 198 1.4571 -0. ~raunschweig. Andjelic Z. 1968. Shamloul D. Knoxville. Abdel-Salam M. M-Hamouz Z. S. 1984.438 Once the values of the potentials at the nodes are known. Hamann JR. Farghally M.43).

ETZ-Archiv 6:143. Abdel-Salam M. Sheng JN. Gronwald H. Abdel-Salam M. Stevenage. IEEE Trans PAS-101:4039. PA. IEEE Trans Power Deliv. Roy CK. Hornstein MN. Paper 11. James ML. Abdel-Salam M. Abdel-Salam M. 1982. Athens. Berechnung elektrischer Felder durch Nachbildung der Grenszschichten mit ausga~ahlten Flachenelementen. Aly F.01. Proceedings of 3rd Inte~ational Voltage Engineering. Ibrahim AA. Philadelp~a. Olsen RG. Misaki T. Proc IEE 123:702. Mukherjee PK. JenningsA. 1985. Applied Numerical Methods for Digital Computation. . New York: Harper & Row. Gomollon JA. Ibrahim AA. Milan. Abdel-Salam M. Proc IEE 120:1574. Singer H. Int J Num Math Eng 21:197. IEEE Trans EI-17:325. Hamburg. 1979. Archiv fuer Electrotechnik 60:27. Moller K. 1986. Technical University Hamburg-Harburg. 1977. Hara T. 1983. Gas Discharges and their Application. Proceedings of 4th International Symposium on High Voltage Engineering. Khalifa M.07. 1985. IEEE papers 86T and D-513-6. Oxford: Oxford University Press. Singer H. 1996. 1979. Symposium on High McWhirter JH. Alexander PH. 1982. IEEE-PES paper A-77-236-3. Foo PY. Singer H. Okubo H. Mansour E. Athens. TN. IEEE Trans PAS-93:1699. 1980. Smith GM. 1986. Oravec JJ. Proceedings of 4th International Symposium on High Voltage Engineering. 1975. Davis JL. Deutsch W. 1983. IEEE Trans PAS-93531. Wolford JC. Ikeda M. Daffe J. Proceedings of IEEE-IAS Annual Meeting. Proceedingsof 4th International Symposium on High Voltage Engineering. Khalifa M. p 1081. Yan Z. Forger K. Honda M. Metz D. Gela G. Tanari T. Yousef F. 1976. 1994. Khalifa M.14. Itaka K. Athens. Khan MJ. 1984. 1933. Paper 11. p 106. Comber M. 1982. Okubo H. Abou-Seada M.lectric Fields 7 Chen HS. 1989. 1982. 1983. IEEE Trans PAS-101:627. 1978. El-Arabaty A. Hoburg JF. Khalifa M. IEEE-PES paper A-75-563-7. A Modern Short Course in Engineering Electromagnetics (UK). Okon EE. Paper 12. 1973. Zaffanella L. Tsuboi H. King SY. In: IEE Conference Publication. New York: John Wiley & Sons. Hornstein MN. Oin BL.09. Hoole PRP. 1984. 1974. 1984. Knoxville. 9:743. Gutfleisch F. Proceedings of 4th International Symposium on Gaseous Dialectrics. Gutfleisch F. p 128. Paper 11. Matrix Computation for Engineers and Scientists. Doctoral thesis. 1974a. IEEE Trans PAS-98: 1609. 1977. J Electrost 18: 1-22. IEEE Trans lA-20:1607. IEEE Trans PAS-93:720. Hoole SRH. 1974b. Ann Phys 5:589. UK: Institution of Electrical Engineers. Pearmain AJ.

Steinbigler H. IEEE-PES paper A-76-418-4. Maidenhead. Thannassoulis P. UK: McGraw-Hill. Zeitoun A. 1983. 1974. 1971. Orlando. IEEE Trans PAS-94:104. 1983. 1994. Takuma T. UK: Cambridge University Press. Tan KX. IEEE Trans PAS-90:145. IEEE Trans PAS-93:1660. Steinbigler H. Finite Element Grid Optimization.12. Zienkiewicz OC. 1979. New York. Proceedings of 4th International Symposium on High Voltage Engineering. Proceedings of 3rd International Symposium on High Voltage Engineering. USA: J Wiley & Sons. Zahn M. Scott M. Athens. Singer H.11. Elements of Electromagnetics. Paper 11. Steinbigler H. Kawamoto T. Pacific Grove. Ali SMG. Takuma T. Comsa RP. Singer H. 4909. Kawamoto T. Fujinami H. Ferrari RL. 1985. . 1979. New York: American Society of Mechanical Engineers. COMPEL. IEEE Trans PAS-l00:2802. Sadiku MNO. Archiv fuer Electrotechnik 67:309.The Finite Element Method. Kong JA. Mattingley J. Finite Elements for Electrical Engineers. Ikeda T. 1981a. 1979. USA: J. Singer H. El-Ragheb M. 1975. Electromagnetic Field Theory. 1976. Applied Finite Element Analysis. Ryan H. 1984. FL: Saunders College Publishing (USA). Milan. Cambridge. Segerlind L. Powell CW.Ryan HM. CA: Brooks/Cole. Timascheff A. Abdel-Salam M. ETZ-Archiv 3:265. New York. Silvester PP. 1981b. 1984. IEEE Trans PAS-100:4665. 1977. Applied Electromagnetism. 1983. IEEE Trans EI-9:18. Shen LC. 1981. Paper 12. Weiss P. Shephard MS. 1974. Gallagher RH. Wiley & Sons.

Assiut. Deionization by recombination and by diffusion Also. an electron produces a negative ion.Ionization and eionization Processes in ases Assiut University. Egypt In the process of partial or complete breakdown of a gas gap there are several mechanisms of electron and ion generation or annihilation in operation. in becoming attached to a neutral particle. In the present chapter we summarize briefly the more significant mechanismsfor ionization and deionization in a gas discharge. and electron detachment 3. Before discussing these processes. interaction with metastable atoms. Ionization by cosmic rays. Ion generation by electron impact. either singly or in combination. The processes considered include: l . photoionization. thermal ionization.This calls for a reviewofsomebasicprinciplesof the kinetic theory ofgases as they pertain to gaseous ionization and deionization processes. 1 . and nuclear radiation 2. it would be best first to describe the behavior of gaseous dielectrics in an electric field. x-rays.

The impulse exterted on the molecule during a collision is equal to 2 ~ ~ v If i xan . . it has a velocity component vix in the X direction (Fig. which treats the gas molecules as tiny balls with no need to specify their exactsize. iinetic lnter~retation of Gas Let us consider a gas composed of moving molecules.2 Many of the properties of gases can be well represented by kinetic theory. Consider the ith molecule ofthe gas in a cubic container. each with mass mo. Assume that collision withthe wall reversesthe velocity component vix. The cumulative effect of many molecules striking against a wall leads to a more-or-less steady force (pressure) on the wall. and energies. temperature. shape. or internal construction.1 Motion of a gas molecule in acubiccont~iner. distribution of molecular speeds. then ure 3. 3. depending on the number of molecules hitting the wall per second. These properties includegas pressure. so that the time Ati between two successive collisions of the ith molecule with the right-hand wall is Atj = Vix 2 D where D is the container-side length.1). average force Fixis exerted bythe molecule on the wall during time Ati.

the general gas law applies for the pressure p: m PG = RT M where T is the absolute gas temperature and R is the universal gas constant [ = 8314 J/(kg mol K)].The total force F ’ on thewall due to ND3 molecules in the container will be the sum of the average force due to each. N is the number of molecules per unit volume.1) and (3. -2 -2 ’ W -2 ‘ U = vy = v. is the molecularweight M dividedby Avogadro’s number No. The gas pressure p then becomes If a mass m of a gas of molecular weightM is confined to a volume G.2) and realizing that m = ~ G m o we . and k Boltzmann’s is constant [ = l . = 3 -2 where G2 is the average of the squared velocities of the molecules. the absolute temperature of the gas is a measure of the squared velocity ofthe molecules within it.38 x J/K]. Thus. have a where the mass ofmolecule m. Thusthe temperature T is directly proportional to the average translational kinetic energy ofthe molecules in a gas at that temperature . Combining equations (3. ND3 i= 1 where : G is the mean square of the x-component velocities of the molecules. Since the number of molecules is very large and there is no preferred direction in the container.

3. that is. The speed probability distribution function Pr(v) is the percentage of gas molecules having speeds within the range v fdv/2. The speed variation among the molecules is usually expressed by plotting the number Nu of moleculeshavingspeedswithin a unit interval (AV= 1m/s) centered represented about the speed v (Fig. . The total number Nt of molecules is by the area under the curve. Very often interest is directed to how the gas moleculesare distributed with regard to their energies. The expression for Pr(v) 1s Note that Pr(v) + 0 only as v + 0 and as v + 00. in arbitrary units. it is convenient to represent the z > Speed v (m/ sec ) Number of molecules N. Since P. versus speedv at 460 and 1700 K.(v) dv is the fraction of molecules having speeds in the range dv centered on v.2).Not all molecules in a gas travel with equal speed. Therefore. all speeds are possible for the molecules. The vertical axis is linear.

3)]. The question is: What will happen if the gas is to exist in a region where its potential energy varies from place to place? The MaxwellBoltzmann distribution law tells us how the number of moleculesN changes with the total energy V. the fraction of molecules with translational kinetic energies inthe range dU centered on U is given by Pr(U)dU = 1 f i e " 'lkTd U fi(kT)3/2 (3. where the most probable energy occurs for U = (1/2)kT. as U =2 0v2 and dU = movdv Therefore. is related to the number iV2 in state 2 by the relation Since .then Z2 = 3kT/mo. The number of molecuels NI that will be found in state 1.3. as deduced before [equation (3. Pr( U)dU.Ionizationand ~eionization in Gases 5 equivalent kinetic energy distribution. The average value of the kinetic energy is given by 00 v UPr(CJ)dU = qkT = (1/2)moZ2. say.5) This distribution is shown in Figure 3.

” When some electrons are at higher energy levelsbut do not leave the atom. This state of the atom is known as a “metastable state. These photons have definite magnitudes corresponding to the specific energy levelsthe electrons can acquire in the atoms ( ~ c ~ a n i e1964). The duration of a metastable state is on the orderof 0.”It ordinarily gives up itsenergy to other atoms in the gas.~ o l t z m distribution. its molecules have a strong affinity for electron attachment). The radiated photon has a wavelength h ch We where c is the velocity of light and h is Planck’s constant. Whenever an electron drops or rises from one energy level to another.. Under the action of an applied field. Negative ions have a larger probability of occurrence if the gas is electronegative (i. The electron and the ion resulting from the ionization move independently.1 S. The energy required is known as the “excitation energy” We. l. Some atoms have certain energy levels to which an internal electron may be raised but from which it cannot revert easily to its normal energy level. they become accelerated and acquire kinetic energies. it gets separated from its atom by ionization. If an electron acquires enough energy. The electrons in the atom have a tendency to revert to their lowest possible energy levels. or to the walls of the container. The internal energy of an atom is associated with the energylevelsof its electrons. Whenall the electrons occupy their permitted energy levels.which is known as the ~ a ~ w e l l . Electrons moving in a gas under the action of an electric field are bound to make numerous collisions with the gas molecules.lasting ’~ for a very short duration (lo-*S).e.The atom has to release this energy to return to its normal state. have much less internal energy than do positive ions. the atom has a “normal state. It will release the ionization energy only when an electron recombineswith itto form a noma1 atom. The electrons are accelerated and their energies rise between collisions by amounts h=- . The energy would be acquired from or given up to other atoms of the gas. ann The atoms of a gas can absorb. formed by electrons attached to neutral atoms or molecules. transport. Negative ions. and release energy. It maybe noted that the ion possesses both ionization energy and kinetic energy. What remains of the atom after ionization is a positive ion. to the electrodes. it is in an “excited ~ t a t e . a quantum of energy is released or absorbed as a photon.

They are thus independent of the applied field. Excited atoms and metastable atoms are not influenced by the electric field because they are neutral. 3. The electron velocity component along the field direction taken over several free paths per unit field intensity is termed the electron mobility. This process becomes significant only when densities of electrons are high. Ifno excitation or ionization results. as the ion and the gas molecule have closely comparable masses. whereas an electron only needs a kinetic energy equal to the ionization energy.1.For every gas there exists an optimal electron energy that gives a significant ionization probability. ions have very large masses and small speeds. A s they give up their energy to atoms or molecules. l . Ionization by electron impact is probably the most important ionization process for a gas discharge. molecules. When the electron collideswith an atom or molecule. they lose a considerable part of their energy (about 50% on the average). Therefore. An ion needs to have a kinetic energytwice the ionization energy in order to ionize the atom it collides with. Similarly.4. they vanish.the atoms. When these positive ions collide with gas molecules. Ionization Cross-section In some models.Very fast electrons are also poor ionizers.’’ Ifit is inelastic. Its effectiveness depends on the electron energy. the positive ions receive energy from the field.the gas atom or molecule becomes excited or ionized by the energy acquired from the incident electron. An excited atom or molecule may become ionized by a subsequent collision with another slow-moving electron. and electrons are assumed to be rigid solid-material balls with the aim of deriving very useful expressions such as .since the period of interaction between the electron and the atom becomes too short andtherefore the amount of energy transferred from the electron to the atom decreases significantly. Photons bombarding a cathode give up their energies to help release electrons from the cathode surface. They are therefore responsible for producing space-charge clouds in the interelectrode region.in eioni~ation Gases 7 depending on their free paths and the applied electric field intensity. kinetic energyis exchanged. Thus the positive ions are less capable of ionizing the gas molecules. the collision is called “elastic.This means that a portion of the kinetic energy of the electron prior to impact has been converted to potential energyof the atom or molecule. their energies are lower than those of electrons in similar situations. Photons have neither mass nor charge. In general.

Thus. + r.1).) of their radii. If a gas has N molecules per unit volume. resulting from collision with electrons of energies between 10 and 1000eV.l2 Electron energy (eV Ionization cross-sections for various gases due to electron impact.] . are given in Figure 3. the number of collisions per unit length of the electron path is equal to ON and the mean free path h is (3.4. The collision cross-section + (3. a collision must have taken place (Nasser. It varies from one gas to another (Table 3. 1971). Increased gas pressure p or decreased gas temperature cr = z(r.9) The values of the ionization cross-sections for various gases.8 A~del-~alam the concept of collision cross-section and free path.when the distance between an electron and an atom is equal to thesum (re r. [From Rapp andEnglander-Golden (1965).10) Themeanfree path is a statistical average and indicates only the average distance betweensuccessive collisions.

” isthe number of ionizing collisions per unit length of the path. It is equal to the number of free paths (= l / X . n of them will reach a distance x without collision.1 Gas jL(10-8m) Values of the Mean Free Path 5 .76 N2 O2 Ar CO2 Kr Xe ‘T=288Kandp = 101. 1970). It can be expressed as .77 He 18. 2 - E (3. Note that n is related to no through the probability f . 3. (Kuffel and Abdullah.12) gives the distribution of free paths where the ratio n/no decays exponentially with distance x.28 6. The free-path distribution function can be derived by considering a group of no molecules just after a collision. ) times the probability of a free path being more than the ionizing length hi..12 3.62 13.2 lists values of some vapors.of some Gas Moleculesa H2 Ne 11.it gains an energy = @Eh.22 6. the mean free path j . proportionately increases its density and hence N. is inversely proportional to p at constant temperature. Ionization Coefficient When an electron travels a distance equal to its free path h. When these molecules are moving in a given direction.E 2 eVi or Vi h.2.19 5. Therefore. its gain in energy should be at least equal to the ionization potential Viof the gas: eh.11) where x is the distance traveled without a collision. The ionization coefficient a. in the direction of the field E.66 4.3kPa.For the electron to ionize.12) Equation (3.13) Vifor the molecules of some gases and the atoms of Table 3.79 6. The probability fx that an electron will have a collision while traveling a distance x can be given by a relationship of the form fx = e x p ( 7 ) (3. Source: ~cDaniel(1964).1. known as “Townsend’s first ionization coefficient.4. : n =n d x= no e x p ( 7 ) (3.Ionization and ~eionization in Gases Table 3.

9 19.1 15.2 17. where l? is a constant that depends on 7'. The corresponding values of B = ( W i ) are 273 840.kPa.7.15) as found experimentally.0.6 21.3 7.14) The mean free path he.1 13. and H2.14) a ! = rexp("_) -r vs = rexp(T) "BP = =(~) P (3. a = "1 exp(2) = ~l e x p ( ~ ) he (3.3 m-'.2 6.3 4.kPa. Hence.Excitation and Ionization Potentials of some Gases and Metal Vapors First excitation potential Ionization potential 27.62 (metastable) 21. S"'.8 (metastable) 40.49 (metastable) 11. respectively.4 14.6 14.66 (metastable) 54.Typical valuesfor the constant lr of equation (3.2 10.15) are l1 253. 256 584.2 8.6 Xe 12.4 15.5 12. Typical values ofalp of the technically interesting .9 16.s" for air.1 20. is proportional to the reciprocal ofgas density. 1/ %e = rp.28 (metastable) 9.8 19.4 (metastable) 11.9 Source: Meekand Craggs (1953). and 104 134 V-m-' . NZ. From equation (3.3 12.56 11.7 3.53 (metastable) 16. and 3826.1 15.8 (eV) Double ionization on gle (ev) Material Ar He Ne 24. 9003.5 11.

. x-rays. for example. or to ignite the combustible mixture in internal combustion engines.gases (air.5. or nuclear radiations. and then decreases gradually with further height.and SF6) versus E / p at standard temperature and pressure (STP) are given in Figure 3.5 alp as a function of E / p for different gases.] (1966). The fact that electrons and ions are always available in air has great significance and consequences in many applications. ~easurements have shown that the intensity of cosmicrays increases sharply with altitude to reach 20 times the sea-level value at about 16 km. Ionization by cosmic rays is a continuous process that produces ions and electrons everywhere since these rays are capable of penetrating most conventional walls. NZ. It should be pointed out that photoionization does not occur simply due to radiation emitted from the gasitselfwhenexcited atoms return to their ground state but also takes place as a result of external radiations such as cosmic rays. Its rate of ionization also increases steadily from a very small valueat sea level to about 80 x lo6 ion pairs per cubic meter per second at an altitude of P ( kVI m * k P a ) Figure 3. Without free electrons. it would not be easy to produce a spark. [From Brown Bahalla and Craggs (1962).

it may still . with scattering of x-rays with longer wavelengths. and e" is the ejected photoelectron. If the photon energy is less than eV. A high-energy x-ray photon may knock out an electron from an inner shell of the atom. forming an avalanche-like ionization. This electron may well ionize more atoms by collision. Because of their highly energetic photons. 1971). The excess of the quanum hv over eVi may be imparted to the released electron as kinetic energy. 2. This process is accompanied by the release of energy in the form of the emission of another x-ray of lower energy than the primary x-ray quantum. When a gas atom absorbs an x-ray photon.an electron.10 km (Nasser. The excess energy of the photon. If A represents a neutral atom in the gas. The high energy absorbed may be adequate togive the electron enough momentum to leave the atom. In the ionized atom.. It must be kept in mind that the insulation of highvoltage systems at high altitudes is subject to reduced air density and to increased ionization by cosmic rays. This new x-ray photon can ionize additional atoms. just as in the preceding case. will be converted to kinetic energy for the electrons. v is the photon frequency (velocity of light clwavelength I). a loosely bound valence electron may be liberated. which is usually very high. as follows: 1 . KE the kinetic energyof the electron. 3.. The photoionization process may be described by the reaction A+hv-+A'+e" (3. the atom radiates a photon that may in turn ionize another molecule whoseionization potential (eVJ is equal to orless than the photon energy. X-rays and y-rays ionize gases because of their high energies. Their energies range from a few electron volts to 100 MeV or more.17) where hu is the photon energy. the reaction describing the excitation of the atom is A+e"+KE"+A*+e" (3.16) On recovering from the excited state in about lo-' S. and A* the excited atom of the gas. their ionizing processes are bound to differ from those of radiations with relatively lower energies such as ultraviolet. using part of the quantum energy. The atom does not completely absorb the incident x-ray quantum and the so-called Compton effect occurs. e. the missing electron is substituted by another from the next-higher shell.

Such a state of affairs is known as "resonance" radiation and is encountered frequently in ionized gases.20) where Io is the intensity of photons at their origin. that is.4. (3.. exciting it. Being radiation.~oni~ation and Deionizationin Gases be absorbed by the atom.20). Then where p is the absorption coefficient of the gas. Integrating.22) . If V" of an atom A" exceeds the ionization Y. and the concept of free path may be extended to photons.ho is the photon number at their origin. ionization may result according to the reaction A"+B+A+B'+e" (3.3 loni~ation y Interactionwith For the atoms of inert gases and of some elements in group I1 ofthe periodic table. Ionization by a beam of photons can be analyzed in the same way as that produced by a beam of electrons. These states are represented by A" and associated with energies V". the lifetime of their metastable states extends to seconds.e. Thus the photon mean free path i p h is expressed as (3.18) and the photon number nph decreases with the distance x traveled. The same thing may continue to happen within a gas until the photon is lost to the boundaries.19) where n.21) Figure 3. 3. on collision. photons get reduced inintensity by d l as they travel a distance dx at x from their origin. we get I = Ioe"IcLx (3. a photoexcitation).6 shows p as a function of the photon wavelength for different gases.19) and (3. we get p = " 1 hph (3. of another atom B. which gets excited only to a higher-energy level (i. A photon emitted from an atom returning to its ground state may be absorbed by another of the same gas. then. d l is proportional to dx and I at the distance x. Comparing equations (3.

This process is known as step ionization: Am+hv”+AS+e(3. an atom in the metastable state may be ionized by absorbing another photon before returning to its ground state. Another ~ossibility €or ionization by metastables is when 2Vm for Am is greater than Vi for A. ionization by metastable interactions comes into operation long after excitation. wavelength at ta e ~ p e ~ a t uof re For V” Vi. It has been shown by some research work that these reactions are responsible for the long time lags of breakdown observed in certain gases ( ~ a r t m a n n and Callimberti..-”+A+AS+e(3. .24) As mentioned before. the reaction would produce only excitation of the atom B. l l IO too 1 Absorption coefficient for x-rays in different gases as a function of 0°C and a pressure of 133 Pa. A. ~ometimes.23) where is a diatomic molecule in the excited state. the reaction may proceed as follows: 2Am+A”+Ajr+A A. 1975).1O2 0.

eion~~ation in Gases Nuclear radiations include a-and P-particles. The most important particles are the proton andthe a-particle.7). as well as photons of y-rays.3). They proceed in almost straight lines.023 0.035 0.107 P-particle 3. When theyare injected into a gas theyundergo collision withthe gas atoms. The ionization effects of particles heavier than electrons can be considered by applying the principles ofelastic and inelastic collisions.2. thus having a longer interaction with the gas molecules. The distance traveled by the particle until it is stopped or “absorbed” is a well-defined range that depends on the particle’s energy and mass and on the density of the absorbing medium (Table 3.4. The range of the heavy particles (a-particles and protons) increases very rapidly with energy (Table 3. of course. 3. 1978). Also. most of its molecules get dissociated into atoms (Fig.3 STP Absorption Ranges of a-particles. thus losing their kinetic energy to the latter. The particle range in gases is. and P-particles in Air at Range (m) Proton 0. the proton has a greater range because of its lower mass. Beta particles are fast electrons having energies on the same order of magnitude. leaving ionized atoms along their wakes until they finally lose their kinetic energies and come to relative rest (Fikry et al.14 30. Both types of particle make collisions withthe gas atoms andcause ionization by collision. The energies of a-particles are on the order of MeV. much greater than in solids. The following are the possibilities for thermal ionization: le 3. as in the case of P-rays. Protons. Because oftheir heavy massestheir direction of travel is not greately affectedby the collisions.OO . It is interesting to note that. If a gas isheated to a sufficiently high temperature.34 1.17 Energy (MeV) 1 5 10 Source: Nasser (1971). Gamma rays were discussed in Section 3.00 41 .. a-particle 0.005 0.ionization of its atoms and molecules may result.3). For equal energies. the number of ionizations produced by the particle is much greater toward the end of its path when it has slowed down appreciably.

Cas discharges under impulse voltages may be initiated by electrons . 1 . Loeb (1965) carried out experiments on electron detachment in oxygen and found that it occurs for E / p = 68V/m*Pa. The process is described by the equation A+W. This process.Gas temperature (*K ure 3 .eA++e(3. Photoionization resulting from thermal emission by the hot gas. however. 3. Because of the high temperature. In fact. A' a singly ionizedatom.. detachment has been considered as one of the main processes involved in breakdown phenomena over a wide range of gas pressures (Maller and Naidu. In compressed electronegative gases suchas SF6. Thermal ionization is the chief source of ionization in flames and highpressure arcs (see Chapter 6). Abdel-Salam et al. Ionization by collision with high-energyelectrons produced by the preceding two processes. and ?Vi the ionization energy. 1978). Under certain conditions in highelectricfields. 1976). e. the velocities and kinetic energies are very high and can cause ionization. 7 Proportion o f ionized gas molecules as afunction o f gas temperature. requires a large concentration of negative ions. the ionization coefficient a increases with temperature (Abdel-Salam. 1981. Ionization by collision of the gas atoms with each other.25) where A represents a neutral atom.the electron removed from the atom. 2. electrons may become detached from negative ions.

2. 2At-e(3..on and Ionization in Gases 7 becomingdetached from negative ions available in the gas (Allen et al.31) In this expression B" may be an electron or a negative ion. leading to a vibrational excited state of the negative ion and subsequent loss of the excessive electron A. (3.27) 3.5 Whenever positively and negatively charged particles are present. there is a chance for recombination to take place. Sommerville et al.+B*cT. 1981. Detachment may be ascribed to the following events: 1. associative detachment leading to theloss of the excessive electron A-+Bt-. 1988). Recombinationof negative and positive atomic ions to form a diatomic molecule A"+B+t-.28) 4. (3. 1984.30) process at pressures above a few hundred e" 3.are the numbers of positive and negative ions per unit volume. dt dt (3. The rate of recombination is directly proportional to the concentrations of both positive ions and negative ions or electrons. then dndn. If n+ and n. This is a very likely pascal." = -Cn+n_. that is. Collision of molecular negative ions with fast excited atoms. (3. Photodetachment A"+hvt-. The potential energy and the relative kinetic energy of the recombiningelectron-ion or ion-ion pair is released as a quantum of radiation: A++B-+AB+~v (3. Collision and subsequent association with neutral atoms.26) +B+e- 5. Abdel-Salam and Turkey.. Collision of ions with a fast atom A-+AcT.32) .29) by absorption of radiation by a negative ion (3.

33) If recombination is allowedto take place from t = 0 to t = t and the density of charged particles at t = 0 is no. such as oxygen. where it is evident that recombination is particularly important at high pressure. From experiments in gases in which negativeions can form very easily.where {is a constant known as the recombination coefficient. {does not show a significant increase with temperature. This is true because the two species of ions have almost equal masses and average velocities and the time they remain in close vicinity is therefore long. The variation of {with pressure is shown in Figure 3.no doubt increase with the gas pressure. -12 3 X 10 t i m - 2 -12 1x10 $ E " * 7 5 4 U 3 2 1x10 -13 4 1 2 40 120 400 1 2 0 0 Pressure (kPa 1 COa. 1971).8. by integrating between these limitswe get which indicates that the density of particles decreases hyperbolically with time. In general. such as in the rare gases (Nasser. Effect of gas pressure on recombination coefficients (c) for air and . On the other hand.= n. The densities n+ and n. n+ = n. and we have dn 2 ={ B dt (3. it was found that the recombination of positive ions with negative ions takes place much faster than does the recombination of a positive ion with a free electron in a gas where negative ions cannot form.

Even their presence as a small impurity can change the behavior of an ionized gas drastically. This can be expressed as e"t. Such a process is reversibleand can be expressed as e"+A-A"thv attachmen.35) detachment ~ttach~ through ~ n t three-body collision^ takes place when the excess energyreleased upon electron attachment is acquired by a third particle duringthe collision process as kinetic energy (U).When an electron becomes attached to a neutral molecule or atom. Such a process is reversible and is expressed as e" +XY % (?X)"* % X. Negative-ion formation is very useful in quenching arcs in power circuit breakers in high-voltage generators and accelerators.SF6.+Y (3.A+B"+A"+(B+tJ) (3. whichis stable because it possesses the lowest energy level of all the possible states. it produces a negative ion. a general criterion for negative-ion stability can be inferred from consideration of the neutral atom. its total energy must be lower than that of the atom in its ground state. Thiscoefficientis defined by analogy with the ionization coefficient a as the number of attachments produced in the pathof a single electron traveling a unit distance in the direction of the field.and air. The difference in total energy betweenthe negative ion and the unexcited neutral atom is termed the "electron affinity" of the atom. Hence. Negative ions play an important role in the breakdownof technically interesting gases such as 02.37) The suppression of electrons from an ionized gas byattachment isexpressedby theattachment coefficient q. for a negative ion to exist and be stable for sometime.36) ~issociative attach~ takes e n t : place in molecular gases when the excess energy is not radiated but is used to separate the two atoms into a neutral particle and an atomic negative ion. In electronegative gases. This energy is released as a photon or as kinetic energy upon attachment. (3. NZ. The different modes of negative-ion formation by attachment are: ~ a d i a t i v attachment^ e takes place when the excess energy is releasedas a quantum upon attachment. Typical valuesof q/p .

38) where D = i j j1/3 m2/s is the diffusion coefficient. The rate of ion flow is J= -DVn (3. whenever there is a nonuniform concentration of ions there will be diffusion of ions from regions of higher to regions of lower concentration.9 q/pas afunction of Elpfor air and SF6. This diffusion will cause a deionizingeffect in the regions of higher concentration.] . ‘E v c ‘E v : . As the concentration gradient varies with time. n c5 c I U c . [FromMaller and Naidu (1981).9 for SF6 and air (Maller and Naidu. electron and ion diffusion is one of the most active deionizing agents. 4 r c n n v 3 L! F4: f 14 12 1 2 0 1 0 0 IO B 6 80 60 4 40 2 20 40 60 2 0 80 100 E/P(kVlm kPa) 120 140 160 la0 200 Figure 3. the rate of change of ion density is c. eioniz~tion by ~ i ~ u s ~ o n During the extinction of arcs in circuit breakers. The faster the arc of the circuit breakers can bedeionized. In electrical discharges.100 ~bdel-sal~rn versus E / p (field-to-pressure ratio) at STP are given in Figure 3. The flowof ions along a concentration gradient constitutes a drift velocitysimilar to that under electricfields. X C . Bahallaand Craggs (1962). the higher their arc-interrupting capacity. 1981).

are the normal values (i. Its solution wouldgive the concentration of ions at any time and at any point in space. and ! can be evaluated (Abdel-Salam.023(~) 1230 P (3. ture of twocomponents is equal to the sum ofthe two individual coefficients (al and a2)multiplied by their respective partial pressure ratios: P2 a.2AttachmentCoefficient q Measured values of ~jlin air (Brown. the value of the ionization coefficient.43) .42) P P Humid air can be considered as a mixture of dry air and water vapor.1Ionization Coe~icient a alp (m-'kPa") Curve fitting of the data measured in air results in accurate expressions of as a function of E / p (V.a l p is related to E/p by (Boyd and Crichton.6. 1971) a ! = 0.16 x lom8(:) 2 (3.735 0.541 x (f) + 1.40) This equation is of greatest interest when we are dealing with discharges in cylindrical channels comprising electrons and positive ions. 3. the average square of displacement will be given by (Nasser.39) whichis known as the continuity equation. =P1 a 1 +--a2 (3.m" kPa-l) (Abdel-Salam and Abdellah.3kPa) (Abdel-Salam. In SF6. the greater average velocities C 3.41) a for a mixIn a gas mixture. For the case of diffusion from a cylindrical concentration.where To and p. D is three orders of magnitude higher for electrons than for ions. 1966) made it possible to express q/p (m-' kPa") as a function of E / p (Vm" kPa") as P 2 = 9. To = 293 K and p. 1976).e. In the same gas. 1978).6. 1985). 1971) (fl2 = 4Dt (3. = 101.15) simply as alp = aT/pOTOand E / p = ET/poTo.. owing to . The thus the effect of humidity on a dependence of a ! on gas temperature T is obtained from equation (3.tion and Ionization in Gases 101 (3.

at STP.000 V-m-I kPa-' for E P (3. (m/s) as a function of E / p (V-m-' kPa") in air by Badaloni and Gallimberti (1972) and in nitrogen and SF6 by Abdel-Salarn and Abdellah (1 978): for. Measured values of ve (Naidu andPrasad.500V. 1975) as EA. are the same as those on .150.1 E (ve)air = >75..33(3 (ve)SF6 245.4 0.m"' kPa" P 104(~)0'62/39.127.). It equals 330 for air. ~ using the electron mobility in the mixture.004(~) (3.47) 233. Its dependence on temperature is similar to that of a. 250 for nitrogen.000V.33~) for 82.45) = POP .). The mobility is .n1-~kPa" P E for 127.(^") of any gas at pressure p (kPa) is (3. = (k.500V. . 1971) as P 2 = 1170. a ~sor~tio~ ~o pe f f ~ c i e ~ t The photon absorption coefficient expressed (Khalifa et al.500 I_< .E.46) (3. The absorption coefficient pm@)for a gas mixture can be evaluated using a relation similar to (3.42). 1972). the electron drift velocity ( v ~is)estimated.44) The effects oftemperature and of gasmixtures on the attachment coefficient q .500 I_< . is the value of EA.1 In SF6.m" kPa-' E P -= (3. 1972) made it possible to express v. as (v.48) In a gas mixture. and where p 600 for SF6 (Alexandrov. q/p is related to E / p (Boyd and Crichton.I_< 75.

8 ~ g ( ~ ) 0 ' 4 9 for E >225OV. was expressed (Badaloni and Gallimberti. The dependence ofthe electron velocity .49)wasusedby Abdel-Salam (1985) to determine (ke).51) 0 . . q. and p . V = 516. and humidity on zm are similar to their effects on a.49) simply by replacing E/p by ET/Po To. k is Boltzmann's constant. can be determined from equations (3.7 o ~ ~ m2/s2 ) is substituted by 32 as the atomic weight of oxygen gas is 16 and oxygen is a diatomic gas (see physical constants in Appendix 3.*.1). 1972) as zm= 0.49) Equation (3.mol.m"' kPa"' P (3. on the average. zrn[(l/2)~o~21 (3.K)) ) 3 ( ~ ~ = 267.95 IX/S . is the ratio between the energy of thermal agitation and the mean thermal energy of the gas molecules. temperature. z.m"kPa" P The effects of pressure. 155 ( ~ ) 0 ' 7 1 for E 5 2250V. A fundamental relation was developed by Loeb (1965) correlating the mobility k.expressed in terms of the components (ke)' and (ke)2and the partial pressure ratios as (3.235. to the diffusion coefficient De: " -e e keD. (1) How fast is an oxygen molecule moving. in air at 27"C? ~o~~tion: From equation (3. in v on the gas temperature humid air.3): V2 = 3 ~ ( = ~ 3(3814J/(kg. kT.46) to (3.50) where kTe is the energy of thermal agitation at the effective absolute temperature of the gas. and z.

2): G = (2. considering re = ra = r . the air molecules have speeds of about 500 m/s. 10) G = lm3.2) to obtain P = 220 28 x ""* x 8314 x 300 = 0.01 x 1 0 ~ 1 ~ 2 Total translational kinetic energy = NG($rnoC2) = x 1. ' 7 x 10"o)2(0.mol K). For M = 28 mol". 3 5 We make use of equation (3.rnoC2 = .01 x lo5 x 1533.1): 1 3 .01 x lo5 N/m2) 32 mol-' + = 1233. (molecular radius) where the collision takes place between nitrogen molecules.01 x lo5N/m2 and 25" C = 298 K From equation (3. lo).8 X N 2 220 X 10"8 kg.ala^ So. Use equation (3.7 A molecular radius.8 X 10" 474 l . The mass of a nitrogen molecule is 4. (4) At what temperature would the average kinetic energy of helium atoms (He) in gas become 1.3 J (3) In a certain experiment. it is necessary for the mean free qath to be at least 10 cm. Assuming the gas to be nitrogen with 1.0 x kg (83 14 J/(kg mol K))(298 K)/( 1.196N/m2 since T = 300 K. and R = 8314 J/(kg .p/N L L = -x 1. Solution: N= 1 = 46.15 X 10"6m3 where 1 atm = 1.8 x kg.0g of O2 gas at standard pressure and temperature? Solution: From equation (3.10 Abdel. (2) What is the total translational kinetic energy in 2.15 x = 232. m = Nmo = 4. find the maximum pressure in the chamber at300K.0 eV? .

(b) five times the mean free path.016.01 x lo5 A beam of ions is injectedinto a gas.77 X lo4 K At 0" C and 760 torr. find the volume of 1.0 kg of helium. The beam has an initial density no ions/cm3.6 x J = 1.2): 1 x 8314 x 273 = 11. the density of helium is 178. Use equation (3.37no (b) n = no en5 = 0.Ionizationand Deionizationin Gases 105 Solution: 1 2 Use equation (3.01 x lo5 = 0..0 g/m3. Molecular weight M of H2 = 2.3): But M/mo = N A and KNA = R where i V is the gas density = 178 x kg/m3 3 x 1.016 x 1.ij2 178 x 10-3 i j = 1305m/s - . Solution: Solution: Use equation (3.2) and (3. Find the mean square velocity of the helium atoms. Solution: Use equations (3.6 x J 2 (1.38 x 10-23 = 0. Find the density of the remaining ions at a distance equal to (a) the mean free path.19): (a) n = no e"' = 0.0 x 1.3): T= --moij2= 1.15m3 G= 2.0170 x lo8m2/s2 .6 x 10-l~) 3 x 1.0067 no At normal atmospheric pressure.

2 13.l el-Sala (8) Find the energy of a free electron in eV at 25" C in atmospheric air and in a discharge tube of pressure equal totorr. solution^ M = l for atomic hydrogen.24 x cm3 Average separation between atoms = = 2.075 = 0.15 x x 3 x 10' = 62. 7- A hc 4.075 g/cm3.677 x 1. The photon gives all of its energy to the bound electron. Number of atoms/cm3 = iVA x 0.@: 2 .6 X 1. so 1 gram of H consists of ATA atoms.79eV which is independent of pressure.6 = 48.454 x = 2. So~utio~~ Use equation (3. thus releasing it from the atom (binding energy = 13. Find the kinetic energy in eV and the velocity of the photoelectron. 48.6 x 10-19 x 293 = 606.38 x 1 2 3 3 2 2 606.0224 x x 0.6 eV .2e'V Photon energy = hu = = l 200 x Kinetic energy of the photoelectron = 62.075 = 6.6 eV).6 x 1 J = -m.find the average separation of the atoms and the average volume occupied by one atom.454 X Average volume occupied by one atom = 1 0. ' . (9) If the density of solid atomic hydrogen is 0.677 x J = 3.3): -moG2 = -kT = x 1.8 X 10"'cm (10) A photon of wavelength 200 is incident upon a hydrogen atom at rest.

Solution: Assume N = total number of particles before ionization Ni = number of ions after ionization Ne = number of electrons after ionization Nn = number of neutral particles after ionization N=Nn+Ni=Nn+Ne Use equation (3. Find the liquid photon absorption coefficient for this wavelength. A beamof radiation has a photon energyof0.20): p = --InX 1 I I o l 1 = --h. Sol~tion: Use equation (3.and pn for electrons.15 x 1000 x 1 0 " O 10-l~ = 12.13 x 106m/s L .2MeV.10 4. find the values of pe/pn and pi/p in terms of a.= 0. If a is the degree of ionization and p is the total pressure. What is the binding energy We of the gas? or we = 3 x lo8 x 4.2): .When it is incident on the surface of a liquid its intensity is reduced to 1/6 at 20 cm from the surface. pi. and neutral particles.0896cm-l 20 6 The longest wavelength of radiation io cause the dissociation of a gas was found experimentally to be 1000 A. respectively. positive ions.45eV A partially ionized gashas partial pressures pe.

77 x 10"" m ) " ' . Solution: As derived in example (13): p = NkT N =pIkT = 1.527 x x1 = 2.38 x 760 = 3. Then because re < 85 x lo2 ra = (n x 3.10): <ra. Determine the corresponding value of the diameter of the argon atom.108 Abdei-Salam Total pressure p =pn +pi +pe where E" p " 4 -" Ni Ni Nn+Ni+ N N +e Ne - IN a l+NJN"l+a " (14) An electron was found to make an average of 85 collisions per cm in argon at p = 1 torr and t = 0" C.527 x atoms/m3 x 273) Use equation (3.01 x lo5 l(1.

Calculate the mobility of electrons.38): = k+n+E .O x 1017m-3. The area of the electrodes is 8 cm2 and the voltage across the electrodesis 2OV. assuming that the current is mainly carried by electrons. determine the distribution of the ions along the axis of the tube. Disregard the field of the ions. The electron density in the tube was uniform and had a value of 1..8 m. i. Jdiff = JE Using equation (3.Ionization and Deionization in Gases 109 (l 5) The current flow through a gas dischare tube was found to be 3 A. Solution: Electron current density = nee(-$) k.e. Solution: The ion-flow upwards dueto diffusion should be equal to the ion-flow downwards due to the electric fieldE in order to achieve a steady-state concentration gradient. The spacing d betweenelectrodesis 0. If ions exist in the tube at a steady-state concentration gradient.e 8x 3 = 9375m2/s*V 0. or k .8 1 2 0 x loi7 x 1. = -4 --A Vn.6 x (16) In a cylindrical tube. there is a uniform field along the axial direction.

= Z __.6 D + t a n d t = .i. determine its diameter after drifting a distance of 0. determine the ion density 0. F =6 7 k + . z kT z kT .k+E Before drift: 6yz After drift: .3 x m.2 . If the initial diameter of the cloud is 0.02 m away in both directions at 25" C.05 m at 25" C.11 el-Sal . : In-n+ ___ -k+Ez n+0 D+ (17) If the electric fieldE in the above problem is 5 V/m and the'ion density at a certain location is 10" ions/m3. (3. one can write for 3-dimensional drift v+ k+E From equation (3.53) (18) A spherical cloud of ions is allowed to drift under the effect of a uniform field of 250V/m.53): . solutio^: Similar to equation (3. z Then.40).

Proceedings of IEE Gas Discharges Conference.__ .3 X x 250 X lom6 = 0.38 x .3 IkPa). pp 398-401.584/9003 = 28. 7 =: F + 0.05 = 29.5 V B r Abdel-Salam M.1 1.6 x x 293 x 0.42 x (19) (a) Calculate the mean free path of electrons in nitrogen at p = 500 pascals (1 atm = '760 torr = 101. Turkey A.09 X m + 29. 1985. solution^ (a) Reference is made to equations (3. Hahn N. 1976. Abdellah M. 1972.kPa" h= and B = 256. . Abdel-Salam M. ( b )Calculate the ionization potential of nitrogen. Abdel-Salam M.14) and (3. 1978. Proc IEE 128A565. Abdel-Salam M. J Phys D: Appl Phys 9:L-148. Abdel-Salam M. Allen NL. Alexandrov GP?. Ali Kh. Dring D.584V/makPa - l 9003 x 0.~ i o ~ i z a tin io Gases ~ 6 x 1. Radwan R.15): " a P re-vi/E - h=Ire-BPIE I ' = 9003 m".39 X loM6 F 2= 5. 1978.London.5 = 0.3 X = 29. IEEE Trans TA-24: 1031.2222 x 1 0 . 1988. IEEE-PES paper A-78-601-7. IEEE Trans 119-21:35. Berger G. 1981.~ m (b) Vi = = 256. IEEE Trans IA-14:516.

El-Debeiky S. J Chem Phys 43:1464-1470. McDaniel EW. J Phys D: Appl Phys 8:670. pp 137-144. Proc IEE 118: 1872. Khalifa M. 1965. Basic Data of Plasma Physics. Naidu MS. High Voltage Engineering. Englander-Golden P. 1984. Craggs JD. Tedford DJ. pp 343-347. Prasad P. Abdel-Salam M. 1975. (Lond) 80:151. J Phys D: Appl Phys 5:1090. Hartmann G. Zeitoun A. Kuffel E. . Craggs JD. Knoxville. Abdel-Salam M. Gallimberti I. Electrical Breakdown of Gases. 1966. 1971. Berkeley. Proceedings of International High Voltage Symposium.Badaloni S. MA: MIT Press. 1953.Proc Phys SOC Boyd HA. Farish 0. Meek JM. Fundamentals of Gaseous Ionization and Plasma Electronics. Oxford: Pergamon Press. Goher M. 1978. Zurich. Oxford: Clarendon Press. Oxford: Pergamon Press. Switzerland. Gallimberti I. 1970. New York: John Wiley & Sons. 1972. 1964. 1965. Abdullah M. Proceedings Symposium on Casous Dielectrics. IEEE Trans IA-14:510. Collision Phenomena in Ionized Gases. 1962. of 46th International Sommerville IC. CA: University of California Press. 1972. Nasser E. Crichton GC. Loeb LB. TN. Rapp D. Italy: Universiti di Padova. New York: John Wiley & Sons. Advancesin HV Insulation and Arc Interruption in SF6 and Vacuum. Electrical Coronas: Their Basic Physical Mechanism. 1981. Cambridge. Basic Data of Air Discharges. Fikry L.Padova. Bahalla MS. Naidu MS. 1975. Maller VA. 1971. Brown SC.

hw in A.7588 x IO” 1.314 x IO3 2.l ~ I .mol.a72 X 1.99793 x IO8 6.685 x 3.0224 x 8.eV .85 x 4~ X 1 0 ” ~ 12.602 x IO”* 9. ineV torr N/m2 kg/m2 F/ m Him A .K mo~ecu~es/m~ Permittivity of free space EO Permeability of free space P O Pr2duct of photon wavelength.602 x IO”’ 6.108 x I .3806 X 10-23 2.0336 x 1O4 8.15 X 1 0 .398 Universal gas constant R Gas density at NTP NN ( t = 0” C and p = 760 torr) Gas density at N t = 20”Candp = 1 torr Normal atmospheric pressure Pat coulomb (c) kg kg clkg joule (J) J*s eV*s J/K m/s molecules per kg molecular weight J/kg.56 x 760 1.013 x IO5 1.6257 x 4.Name Dimension Value Electron charge Electron rest mass Proton rest mass Electron charge to massratio Electron volt Planck’s constant Boltzmann’s constant Velocityof light in vacuum Avogadro’s number k C NA -1.and energy.

This Page Intentionally Left Blank .

Egypt Air has been the insulating ambient most commonly used in electrical installations. under DC. or extinguished. and impulse voltages. as in arc furnaces. Depending on the circumstances. Among its greatest assets. the spark may be either extinguished or replaced by a highly conductive conducting arc. In this chapter we present a brief discussion ofthe breakdown mechanisms in uniform and n o n u n i f o ~ fields.es Assiut ~~iversity. as in circuit breakers. Assiut. Depending on the source and the gas-gap conditions. The complete breakdown of an entire gap initially takes the f o m of a spark requiring a high voltage. 1 1 . is its selfrestoring capability after breakdown. of air and SF6 at various pressures and temperatures. Liquid and solid insulants in use often contain gas voids that are liable to break down. and through it a relatively small current flows. AC. the arc may be maintained. An electrical discharge in a gas gap can be either a partial breakdown (corona) over the limited part of the gap where the electrical stress is highest or a complete breakdown. in addition to its abundance. Therefore. the subject of electrical breakdown of gases is indisp~nsable for designers and operatorsof electrical equipment.

n is where Pr(n) the average size. As mentioned in Chapter 3. is where a is Townsend’s first ionization coefficient. However. In a uniform field (i.. together with the primary electron. They could also be produced at a later stage by photons from the discharge itself.. as will be shown later. The transit time -re for the electrons of the avalanche to cross the gap is of the order of nanoseconds. n = cad. the latter are virtually stationary during the time required for the electrons to reach the anode. as noted in Chapter 3. the number of electrons will thus have increased to n. it can ionize a gas molecule by collision. Consider a uniform fieldbetween electrodes immersed in a gas. Electrons may. It aether. 1964) that the size of an electron avalanche from a single starting electron follows the exponential distribution is the probability of occurrence ofan avalanche with size n. originate from the cathode as a result of ultraviolet radiation.) and ions (v+). and d is the gap length.2) It is well to emphasize at this stage the statistical ~uctuating nature of the impact ionization process and the fact that a is onlyan average valuefor the number of ionizations per unit length of electron drift along the field. the electrons present willbeaccelerated toward the anode. constant a) with an initial number of electrons no emitted from the cathode. If an electric field E is applied across a gap. their number at x will be n.1). if the electron acquires enough energy. gaining energy with distance of free travel. this process alone is not sufficient to produce breakdown. for example. 4. At a distance x from the cathode. due to the diffusion of the drifting electron swarm. ecause of the difference in drift velocities of electrons (v. The head of the avalanche is built up of electrons while its long tail is populated by positive ions (Fig. or from the gas volume by ionization of neutral molecules. The track is wedge-shaped. the new electron.e. . proceed along the field and repeat the process.Ionization by electron impact is probably the most important process in the breakdown of gases. = no exp(ax) (4. Leaving a positive ion behind. A further increase dn.

If. The “primary avalanche” process as described above is completed when the ions have entered the cathode. Figure 4. . the probability of additional electrons being liber__ ___ __ . E2 E. however. E. Electron current waveform when secondary electrons are produced by photons received by the cathode. .2 shows the components of the current pulse produced by an avalanche started by one electron leaving the cathode.2 Current due to electrons and positive ionsof an avalanche. _ _ Ideal electron current waveform ( . E. Its initial part is carried by electrons moving rapidly toward the anode.1 Distribution of charge carriers in an avalanche and their contribution to the applied uniform electric field. . l 1) Actual electron current waveform Avalanche positive current I& It( 1) 1) A A ure 4. > E . .17 r c 4. the amplification of the avalanche (ead) isincreased. > E.Anode J ure 4. whereas the latter part is carried by positive ions drifting slowly toward the cathode.

and secondary electrons are liberated from the -2 IO 1 -6 10 E/p( V G ’ . The dashed line shows the idealized case. excited atoms. the radiation intensity is proportional to eavet. The number of secondary electrons produced at the cathode during the life of the first avalanche is given by These new electrons start the second generation of avalanches. and these electrons can initiatenew avalanches. It is a function of E / p in the same manner as a !but has a much smallermagnitude (compare Figures 4.3 and 3. y is called Townsend’s second ionization coefficient. metastables (Section 3. kPci’ ) gases. In fact. which assumes that all of the radiation producing the photo affect at the cathode is produced at the moment the electrons enter the anode.5) (Naidu and Kamaraju. 4.4)-are presented ~uantitativelyby a coefficient defined as the average number of secondary electrons produced at the cathode corresponding to one ionizing collisionin the gap. however.ated in the gap by other (secondary) mechanisms is increased. 1996). The secondary ionizing agents-positive ions. photons. The generation interval after which the succeeding avalanche starts depends on the secondary process.2). It is common in air discharges that thesecondary process is ~hotoe1ectric and the new avalanches start after most re = d/ve (Fig. Secondary ionization coefficienty as a function of E / p for different .

Townsend observed that the current through a unifom-field air gap at first increased proportionately with the applied voltage in the region @-Vl) and remained nearly constant at a plateauvalue . .lal corresponds to the photoelectric current produced at the cathodeby external irradiation. the current rises above Iol at a rate that increases rapidly with the voltage until a sparkresults. leading to breakdown. Theavalanche process described above is basic for both mechanisms of breakdown. Such generations of rapidly succeeding avalanches canproducea space charge of slow positive ions in the gap. each under its own most favorable conditions. Two typical gas breakdown mechanisms have been known: the Townsend mechanism and the streamer (or channel) mechanism. Averagecurrentgrowth to breakdown as a function of theapplied .lol (Fig. If the illumination level at the Voltoge ur voltage.2. It is now widely accepted that both mechanisms operate. 4. These space charges enhance the electric field somewhere in the interelectrode space. which results in the solid line of Figure 4. with a subse~uent rapid current growth.cathode during the transittime of the avalanche electrons.4). At voltages higher than V2. For several decades there has been controversy as to which of these mechanisms governed spark breakdown.

I = I y(ead 1 ) 80 The number of electrons arriving at the anode is tt ad nd = nOe so that nd = noead 1y(ead 1) . let n6 = number of secondary electrons produced at the cathode n : = total number of electrons leaving the cathode and so n: = no Each electron leaving the cathode produces (on the average) (ead l) collisions in the gap. the number of ionizing collisions in the gap equals n:(ead 1). In the steady state. the plateau Io rises proportionately but the voltage Vs at which sparking occurs remains unaltered. 43. i. provided that there is no spacecharge distortion for the electric field between the electrodes.. Therefore.l Current Growth Equations It has been proved in Section 4. nb = Yno(e tt + nb (4.cathode is increased.7a) . the circuit current I willbegivenby I= roead ly(ead 1) In the absence of Townsend’s secondary mechanism. r == roead (4.1 that each electron leaving the cathode produces on the average (ead l) new electrons (and positive ions) while traversing the distance d.2. whereas the secondary (y) process accounts for the sharper increase of current in the region V3-Vs and eventual spark breakdown of the gap. . y = 0.5) ad __ 1) thus n.1 . The increase of current in the region VZ--V3 is ascribed to ionization by electron impact. By definition.e. Having no electrons produced at the cathode by the external source of radiation.

Both coefficients are functions of the electric field strength E. At lower values of d. The spark breakdown voltage corresponding to each value of E can easily be calculated (Abdel-Salam and Stanek. For a given photoelectric current Io and applied field El. yead < < 1 in equation (4. breakdown occurs when y and a acquire the corresponding critical values. so that I c41 roead. ead > > I.1 Equation (4.7). I approaches infinity and sparking occurs at d = d. with slope a... and the expression becomes simply yead = 1.5. This condition is defined as “breakdown” and its “Townsend criterion’’ is thus Normally. ea‘ and yead increase until yeffd approaches unity and I approaches infinity [Equation (4. A s Y increases. . and the plot of ln(I/Io) versus d is linear.7) and as shown in Figure 4. Gap spocing. A maximum length dlsis reached when the criterion (4. spark breakdown occurs and the current I rises sharply).7)].As d increases. 1988). say. At low field strengths V/d. yead increases and upcurving occurs until. for a given gap d.e. In this case the current will be limited only by the resistance of the power supply and the conducting gas.5 Townsend-type plot of versus d. Thus.ead -+ l . the circuit current I varies with the gap d according to equation (4.8) is fulfilled (i.7) describes the growth of the average current in the gap before spark breakdown occurs. d ure 4. so that I = roeadin the region Vz-Y3. when yead c41 1.

equation (4. [Courtesy of F. The breakdown voltage Vs is the same for a given value of the product pd which is the well-known Paschen's law. Llewellyn-Jones (1957) and Oxford UniversityPress.8) can be rewritten as (4. The relation between Vs andpd is plotted in Figure 4. To explain the shape of the curve. Since each collision entails some loss of energy.6. thus a =~ ~ ~ ( E / p ) .] .8a) gives the breakdown voltage V implicitly in terms of the product of gas pressure p and electrode separation d. an electron makes fewer collisions with gas moleculesas it travels toward the anode.8a) quat ti on (4. y = ~2(E/p). consider a gap of fixed spacing. A s the pressure decreases from a point to the right of the minimum.em em be ring from Chapter3 that coefficientsa and y are bothfunctions of the gas pressure p and the electricfield E = V / d . Consequently. the gas density decreases and the electron free path increases. 2 0 Breakdown voltage . V versus the product pd for different gases. it follows that a lower electric field would still furnish the electrons with kinetic energies sufficient for ionizing collisions.

When the minimum of the curve is reached. where p is the gas density. Empirical relations have been suggested by many workers to express the breakdown voltage of uniform-field air gaps at atmospheric pressure (Alston. the density is low and there are relatively few collisions. breakdown can occur only if the chance ofionizingisincreasedby an increase in voltage. 1976). even if the energy of the electron exceeds the ionization energy. Further increase in temperature ultimately results in failure of Pascben’s law because of significant thermal ionization above ~ 0 0 0 K. which depends on its energy (Chapter 3). which takes into account the effectof temperature (AbdelSalam. The validity of Paschen’s lawhas been confirmed experimentally up to 1100°C (Alston. (min) (Pa+m) 0.678 0. le 4.665 0.891 0. (min) (V) 327 137 273 156 420 25 l 418 450 457 414 p d a t V. The electron has a finite chance of ionizing. more general statement of Paschen’s law is therefore V = F ( p d ) . 1968).530 5. .931 0.1 ~inimum Breakdown Voltages for Various Gases Gas V .10) where d is the gap length in meters. A. If the density and hence the number of collisions decreases. It is necessary now to take into account the fact that an electron does not necessarily ionize a molecule on colliding with it. which is shown to the left of the minimum. The minimum brea~downvoltages and the corresponding pd values are givenin Table 4. 1996). as shown in Chapter 3. Vs = 2440d + 61& kV (4.439 0. The discussion so far has ignored temperature variations.798 Source: Naidu and Karnaraju (1996).1 (Naidu and Kamaraju. 1968).197 1.320 0.754 1.

Also. and breakdown occurs through its channel. the electrons are swept into the anode.4. When the avalanche has crossed the gap. 4. the streamer theory of the spark was proposed by Loeb (1955) and Meek for the positive streamer. spark discharge develops directly from a single avalanche which the space charge transforms into a plasma streamer. 4.7a). as evident from Figure 4.el-Salam The very short time lags of spark breakdown. .7 Cathode-directed streamer as envisaged by Meek and Loeb:(a) first avalanche crossing the gap. arenot consistent with the Townsend mechanism. Thus the conductivity grows rapidly. (b) streamer extending from the anode. and anticipating the experimental results of Raether (1964).1 Version by Loeb and Meek The positive streamer from the anode toward the cathode in uniform-field gaps is explained as follows.1. A highlylocalized ure 4. which is based on the generation of a series of successive avalanches. and independently by Raether for the negative streamer. the positive ions remaining in a coneshaped volume extending across the gap (Fig. (c)streamer crossing the gap.3. measured when high overvoltages are applied to uniform field gaps. According to both versions. The space charge produces a distortion of the field in the gap. A s a consequence. The principal features of both versions are the photoionization of gas molecules in the space ahead of the streamer and the local enhancement of the electric field by the space charge at its tip. it wasdifficult to envisagehow the Townsend mechanism would apply for long gaps where the sparks are observed to branch and to have an irregular character of growth.

thus enhanced. Er= I C E where K EX (4. Ultimately.7b). the streamer mechanism plays the dominant role in explaining breakdown phenomena.2 Raether's Version A slightly different criterion was proposed independently by Raether for negative anode-direced streamers. However. At higher pd values. and p the gas pressure in pascals. in the gas surrounding the avalanche.5OO~/p)'/~ with a in m-'. but elsewhere in the gap the ion density is low. which are directed by both the space-charge field and theexternally applied field. 4. a conducting filament ofhighly ionized gas bridges the whole gap between the electrodes (Fig. photoelectrons are produced by photons emitted from the densely ionized gas constituting the avalanche stem. S S) 1 and Er= 5. would promote secondary anode-directed avalanches aheadof the initial one. He postulated that streamers would develop whenthe initial avalanche produces a sufficient number ofelectrons (e") such that its space-charge field E. These secondary avalanches are initiated by photoelectrons in the space ahead of the streamer (Fig. The process thus develops a selfpropagating streamer. lengthen.that is. 4. becomes equal to E. Experiments for breakdown in air proved the validity of the Townsend mechanism in uniform fields with values of pd at least up to about15 kPa. forming a negative streamer. The transition from an electron avalanche into a streamer is considered to occur when the radial field Erproduced by the positive ions at the head of the avalance is of the order of the externally applied field E. Evidently.7~). 1968). x in meters.4.space-charge field is produced near the anode.3. . Equation (4.1 1) expresses the criterion of breakdown by Loeb and Meek. The total field. The breakdown voltage is that at which the avalanche grows to the limit that its field E.m (Alston. 4.27 x 10-7aexp(ax) V/m (S 3. the greatest multiplication in these auxiliary avalanches will occur along the axis of the main avalanche where the spacecharge field supplements the applied field. and intensify the space charge of the mainavalanche towardthecathode.8). is comparable to the applied field E. which effectively extends the anode toward the cathode (Fig. These photoelectrons initiate auxiliary avalanches. Positive ions left behind by these avalanches effectively branch. 4.

Details of breakdown in nonuniform field gaps are discussed in Sections 4. such as in point-plane or point-point gaps.7. In long gaps. bird droppings. Depending on the value of the instantaneous voltage gradient at the stressed electrode and the leader channel length. As a result. dust or leaf particles.6 and 4. filamentary branches called leader streamers exist where most of the ionization activity takes place (Fig. or reaches the plane electrode. develops at the electrode and propagates toward the other electode. a highly ionized and luminous filamentary channel. and other nonmetallic materials.Cathode Anode-directed streamer as envisaged by Raether l The streamer mechanism has been widely used to explain breakdown in nonuniform fields. For such . the leader streamer either stops after having crossed a part of the gap. . if the voltage gradient at the stressed electrode exceeds the corona onset level. the leader channel tip advances very rapidly. the ionizing wave advances toward the leader channel tip and leads to the finaljump.9). causing a return ionizing waveto develop there. Similar but more practical situations are encountered in cases of imperfections on high-voltage conductors such as a nick of a strand or from airborne substances such as insects. called the leader channel.1 . In the latter case. The last three decades have witnessed intense research into electronegative gases such as SF6 (sulfur hexa~uoride) and CC12F2 (Arcton-12). Atthis stage. The electrons produced due to ionization in the leader streamers feed through the leader channel into the stressed electrode. At the tip of the leader channel. 4. the ionization activity in the gap increases. bridging the entire gap and causing a complete breakdown.

With an increase in gas electronegativity (i.13) can safely reduce to a = q.r e a k d o of ~~ Gases Leader channel igure 4. As y is usually much smaller than unity ( y N equation (4.3 V/m-Pa for CC12F2. With the attachment coefficient being quite significant in such gases. Thus a high insulation.13) l+Y This is a condition that depends only on E / p .10).. 4.3 V/m-Pa for SF6 and 89. this criterion is approximated by a=- (4. is modified to For q >a. with an increase inq ) . and sets a limit for E / p below which no discharges should be possible. It is widely used .of d. whatever the value . gases the electron attachment coefficient q is significant and their dielectric strength is considerably higher than that of air. Townsend's criterion for breakdown. and for large gaps.compared to 23. Those studies have led to the development ofmodern gas-insulated systems(GIs). This is why the breakdown voltage measured in electronegative gases such as SF6is substantially higher than that measured in air (Fig. the limiting value of E / p for breakdown increases.9 Leader channeland leader streamers.6 V/m+Pa for air. Townsend's equation can be modified to include the effects of both ionization and attachment processes.e.8). strength can be achieved without severely raising the gas static pressure (see Chapter 10). The limiting values of E / p belowwhich no discharges are possible are 86. equation (4.

75 m diameterquasi 2.11). The high breakdown voltage at Pm is attributed to the so-called “corona stabilization” (i. Breakdown voltages in gases normally show anomalies in their increase with gas pressure. 1978). The decrease in breakdown voltage at pressures above Pm is attributed to the enhanced corona-streamer propagation across the gap as a result of the effectiveness of photoionization at higher gas densities.10 Breakdownvoltages of air and SF6 for 0. Above Pc.e.1 ure 4. 1977). The pressure Pm decreases with an increase in gap spacing. breakdown occurs without any preceding corona and the breakdown voltage resumes a slow increase with pressure (AbdelSalam et al. 4. the breakdown voltage drops with further increase of pressure until a critical value Pc is reached. field emission from the cathode becomes the significant factor for breakdown to occur unpreceded by corona (Khalifa et al.. the breakdown voltage of a positive point-plane gap increases with gas pressure. reaching a maximum value at a pressure Pm (Fig.Above Pm. uniform-field electrodes having as an insulating as well as arc-quenching medium in high-voltage equipment such as circuit breakers and metal-clad switchgear (CIS).1 cm gap spacing. In SF6. a field modification arising from space-charge effects).. At pressures above Pc. as afunction of pressure. The dielectric strength of highly electronegative gases is drastically ..

Point discharges that exist at the irregularities of a rough surface cause a considerable pre-breakdown current and result in a decrease in positive breakdown voltage. In many gases. and at pressures around the atmospheric level. Therefore. The breakdown voltage decreases only slightly at higher humidities. Paschen’s law ceases to apply beyond a certain pressure).6). foreign conducting particles. . At high pressures.11 Positive DC breakdown and corona-onset voltages in tions of pressure. g. as funcigure 4. 1988).e. as discussed in Thebreakdown voltage of a givengapdepends on the gasparameters (a.Pressure p (kpa 1 SF6. p . surface cleanliness and smoothness considerably improve the breakdown characteristics (Chapterlo). provided that the surface remains clean and smooth.. ) . temperature. the static pressure. With successive .which in turn are functions of the electric field and of such factors as the gas pressure. Therefore. Field emission from irregularities on metallic cathodes contributes significantly to the pre-breakdown current (Khalifa et al. the breakdown versus pressure characteristics show saturation (i.. the breakdown voltage of a given gas gap decreases at higher temperatures and increases with. however. The combined factor is the relative air density (Abdel-Salam and Stanek.. 1977). and humidity (Section 3. little difference has been observed between the breakdown voltages obtained with electrodes of different materials. reduced by the presenceof Chapter 10..

13 illustrates schematically the dependence of the voltage on the gap length and sphere diameter. Figure 4. but if the sphere is negative. p) is evaluated from uniform-field data at thesame gas pressure P. In more divergent fields. 1976). Meek’s criterion for the spark is obtained from equation (4. Most practical gas gaps have nonuniform fields. In an o n u n i f o ~ field gap. Once coronastarts. 4. At certain voltages belowbreakdown level. however. ionization may be maintained locally at the highly stressed electrode (Chapter 5). although computations and explanations for streamer growth have been reported (AbdelSalam et al.where the applied field and the first Townsendcoefficient a vary across thegap. the space charge acts as a screen that decreases the field in its vicinity and thustends to raise the breakdown voltage (Fig.11) after modification as a(x) exp ( ~x 0 a dx) = G(x.sparking. p) (4. Electron multiplication is thus governed by the integral of a over the path. characteristics obtained with the pointed electrode positive are usually more important for practical applications. and here the dependence of the breakdown voltage on the electrode configuration is much more complex than the dependence of the corona onset voltage. microscopic protrusionsonthe electrode surfaces tend to get burned off and breakdown voltage rises to a higher plateau.the space charge acts as an extension of the electrode.12).15) where G(x. In sphere-plane and point-to-point gaps. a different phenomenon sets in (Waters.the electric field becomes distorted by space charge. Because of these polarity effects. Examples include wire-toplane gaps and coaxial cylinders. if the stressed electrode is positive.. 1978). if the maximum fieldoccurring at the highly stressed electrode isless than fivetimes the averagefield. Townsend’s criterion for the spark takes the form d a d x = ln(l+:) (4. There are three main regions: . No criterion has yet been well established for the advance of streamers in nonuniform fields. the discharge phenomena will be similar to those in a uniform field.14) where d is the gap length.

01 c. W _"""" " " " " " > -Breakdown 0 "-Corona onset Gap spacing./ / point 0 l 2 3 4 5 6 Gap spacing d(cm) 7 8 . l 2 Breakdown voltage of a point-to-plane gap in atmospheric air with DC voltage of both polarities. d Breakdown and corona-onsetvoltagesversusgapspacing spheres of different diameters. c m . Breakdown voltage with 50 Hz voltage is included for comparision. for .

and therefore the corona onset voltage. The effect of the sphere diameter on the field magnitude becomes more significant as the gap increases. If the applied AC voltage magnitude is such that at the voltage peak the discharge onset conditions are reached.. For gaps exceeding twice the sphere diameter. the field is almost uniform. Therefore. They represent a principal factor in the design of equipment insulation (Chapter 14). depending on the sphere diameter. For moderate gap lengths. whereas the breakdown voltage depends mainly on the gap length. if it occurs. At small gaps. Impulse overvoltages arise in power systems due to lightning or switching surges. Under impulse. this distance is about 1 m in atmospheric air under power frequency.13 Region I. a time lag isobservedbetween the instant the applied voltage is sufficient to cause breakdown and the actual event of breakdown. The maximum gap spacing (Lmax) in which this is possible is the distance the ions move under such conditions. the field shows significant nonuniformity. Region III. Region II. electron avalanches willbe produced in the same way as under DC. The only difference is that the ions in the gas will be subjected to a slowly alternating field. Therefore. 1984). and the breakdown voltage depends mainly on the gap length. the breakdown voltageincreaseswith the sphere diameter as well as with the gap length. This represents an extremely small fraction of halfa cycle ofthe power frequency. . It is therefore important to appreciate the fact that the breakdown mechanim under impulse voltages is different from cases of steady DC or low-frequency AC. breakdown is preceded by corona. Spacings and encountered in high-voltage power transmission systems exceed Lmax the ions are constrained to the vicinity of the conductors. The maximum field. In uniform fields. The space charges produced willhave ample time to leave the gap before the field reverses polarity. are influenced by the sphere diameter. The two basic relevant phenomena are the appearance of electrons for initiating the avalanches and their ensuing temporal growth. the mechanism of breakdown is essentially the same as under DC. The breakdown process gets completed in an interval on the order of to 10"' S. recombination occurs among the outgoing and returning ions (Abdel-Salam et al.

4. where the attachment coefficient exceeds the ionization coefficient. The probability that a negative ion appears in the critical volume to give birth to an electron expresses the distribution of the statistical time lag t. if the seed electron is too close to the point it develops an avalanche of insufficient sizefor streamers to develop. if a gap has then the average statistical time lag t. 1988). streamers cannot form if the seed electron is too far from the anode. For breakdown to occur. of the gap because of its statistical nature (Berger. = A V clearly depends on shown in Figure 4.under an impulse voltage of short duration ( ! x S). that elapses between the application of a voltage greater than V.. or the time to breakdown (TBD). natural resources may not be sufficient to provide the initiating electron at the appropriate site in time for the breakdown to occur.r e a k ~ of o Gases ~ ~ 133 In the caseof steady or slowlyvaryingfields. and for overvoltage protective devices provide the basis of insulation coordination. Also..b.14. shown in Figure 4. 1981). for internal and external insulation of equipment. Breakdown on the front of an applied voltagewaveis V. + + tatistical Time La It is interesting to examine some ofthe factors controlling t. which at the static breakdown voltage Vsshould theoretically be reduced to a point on the axis of the electrode system.The sum ( t f t..g. the gap’s static breakdown voltage. The overvoltage V (t.15. 1973).) is the total time lag.14a). The differences among the volt-time characteristics for different airgap geometries.. For example. cosmic rays or detachment from negative ions) (Allen et al. depending on the gap volume. However. The time t. in a positive point-to-plane gap. Pr1 the probability of an electron appearing in the critical volume whereit can lead to a spark. the subsequent time t f required for the breakdown of the gap to materialize is known as the formative time lag. then the probability that it will break . survived breakdown for a period t. and the appearance of a suitably placed initiatory (seed) electron is calledthe statistical time lag t.. equals 1/(8Pr1Pr2). 4. andPr2 the probability that such an electron will lead to a spark.14a) (Abdel-Salamand Turkey. there isusually no difficulty in findingan initiatory electron from natural sources (e. t f ) and on the rate of rise of the applied voltage. the applied impulsevoltage V must be greater than V. The probability of breakdown increases from zero to 100%over a suitable voltage range. This important subject is discussed in Chapter 15. Thus a critical volume is defined(Fig. If 8 is the rate at which electrons are produced in the gap by external irradiation. Also. After such a seed electron appears. The critical volume grows axiallyand laterally with the increase in applied voltage (Fig.

(b) distribution of time lag (t.)for the initiation of positive corona. .] Vottage / a ” “ impulse “ 4 Breakdown on the front of the applied impulse voltage wave.1 L 1 1 i !l l l I l .l4 Initiation of positive corona under surge voltages: (a) variation of critical volume with applied voltage. [From Goldman and Goldman (1978).1 - .

again raising the breakdown voltage. With much longer fronts. The formative time lag tf consumed in these processes was measured experimentally by Kohrman and others (Nasser. With shorter wavefronts.Pr1Pr2is independent of L. YO for airgaps. At higher overvoltages tf was of the order of 1 p . 4. From Figure 4. (1973) have given an empirical formula for the critical front duration: T ‘ = 40 + 35dp. which supported the streamer mechanism. positive breakdown voltages of non-unifo~-field gaps were minimum (Fig. the breakdown voltage depends upon the rate of rise of voltage in an impulse and on theelectrode spacing. eventually culminating in the sparkover of the gap.down in the next time interval dt is 6Pr1Pr2 dt. relatively long time lags 1971). under the applied voltage. For overvoltages of less than 1 tf(> 20 PS) were observed. Because of the time and expense involvedin high-voltage tests on large structures. corona will develop and produce a large space charge in the gap. the breakdown characteristics will be affected by the rate at which the voltage rises. Thus.s (4. with wavefronts of 50 to 300 p . the space charge of the corona gradually fills the zone near the stressed electrode and reduces the voltage gradient there. which delays the advancement of the leader stroke andleads to anincrease in the breakdown voltage. and leaders. any method of extrapolating known results is economically attractive and important for the design of EHV systems. which led to the belief that at such small overvoltages a Townsend mechanism may be at work. the breakdown voltage for any other gap length can be estimated by . Given the breakdown voltage of a rod-plane gap. The probability of breakdown with a time lag t is exp( t/ ts) . it will. start an electron avalanche with the subsequent processes.consequently.16 it is noticed that the critical front duration corresponding to the minimum breakdown voltage increases with gap length.16) where the gap length d is in meters. the space charge resulting from prebreakdo~n corona causes a marked distortion of the applied electrostatic field. These dependencies are manifest in the existence of the so-called ““U-curves”. such as secondary avalanches. The establishment of such space charge takes a finitetime. Harada et al. Once an initiatoryelectron is made available in the gas gap. In gaps with a highly n o n u n i f o ~ field. It has been observed that. This will be independent of t if 6. streamers.16).

the voltage gradient . [From Bazelyan et al. Paris et al. 1973).. Subsequently. Resemblance between the leader and the electrical arc with a conductance that varies exponentially with the lifetime of the leader (time constant z = 50 p).Dependence of the breakdown voltageof a rod-gap on the voltage front duration for various gap lengths.Axial propogation of the leader. 1979.g. (1961).5 kV (4. to 1.] vs+= 500Kgd0. Constant velocity of leader propagation (v = 1.5cm/ps).3 for rod-rod gaps. based on the breakdown voltage of a rod-plane gap of the same length.17) where Kgis the gap factor for the configuration.Veryrecently. 4. Constant charge injection during propagation (41 = 45 pC/m). 2. and up to 1. 1978).15 for conductor-plane gaps. Waters. 3. ranging from 1. The gap factor method has also been applied to lightning impulses (Paris et al. Semiempirical models have been proposed by researchers to calculate the breakdown voltage of long air gaps of up to 50m(e. The model is based on assuming the presence of the following factors: 1.9 for large conductor-to-rod gaps. (1973) deduced the gap factors for various electrode configurations. Alexandrov and Podporkyn. This is a rule-of-thumb used for estimating the breakdown voltage of gaps of various geometries..Rizk(1989)developed a mathematical model for calculating the continuous leader inception and breakdown voltagesof long air gaps under positiveswitchingimpulses with critical time-to-crest.

breakdown starts with an avalanche process.18) which predicts breakdown voltages that agree reasonably well with findings of the Renardier group (Paris et al..within the leader varies from an initial value Ei(= 400 kV/m) to an ultimate value Em (= 50 kV/m). The model dealt with rod-.89/d + 78 kV for d >4rn (4. and conductor-plane gaps. For d >Lmax. including the height of the final jump. let the field in the gap be expressed as ( V / d )sin 27tft. The agreement between the values predicted by the model. In other words. This accumulating space chargewill no doubt distort the field in the gap. As the frequency f of the applied field increases to very high levels. the discharge behavior starts to differ. for a given d the critical frequencyf.20) At frequenciesf >fc the cloud of positive-ion space charge will oscillate between the electrodes while new avalanches grow and add to its density and size until instability and breakdown occur. 1973). V being the applied peak voltage and d the gap length.Em. most positive ions will not be able to reach the cathode before the applied voltage reverses sign. Constant voltage gradient through the leader corona streamer (Es = 400 kV/m). In uniform-field gaps. breakdown . leader voltage drop. as described above for the cases of DC and low-frequency AC. and 50%breakdown voltage. and Es.19) k+ being the positive ion mobility. at which all positive ions can just be cleared from the gap during one half-cycle is L=-V k+ xd2 (4. v. sphere-. Ei. For rod-plane gaps. 5. and those measured experimentallyis excellentin light of the aforementioned assumptions and the unique values of 41. Rizk(1989)developed an expression for the breakdown voltage V : + 50d v = 1556 1 + 3. To explain this point. Then the maximum distance Lmax that any positive ion can travel during one half-cycle is (4. Therefore.

In analogy withf. depends on the electron mobility k. fee. for ions.] . They will oscillate in the gap and collide with the gas molecules.17). This critical frequency. 1966).21) . 4. - + Mobi Iitycontrolled breakdown . The foregoing analysis of ion motion canbe extended to the electrons oscillating between the electrodes. When the field is adequately high. [From Ganger (1953).. fce = ke V' (4.1 Xlsd o f pasitiw? 0 ion oscillation . they will produce more and more electrons until breakdown is completed with no participation from the electrodes ( ~ a c D o n a l d .occurs at lower field strengths for high-frequency AC than under DC (Fig. and the magnitude of the applied voltage V'.2 1.there will be a critical frequency above which electrons would have no time to reach the opposite electrode. At frequenciesf c fc the breakdown conditions are quite similar to those of static fields.9 1 1 l I Atmospheric air " I I I 7 I I I 1 oL 3000 F"" 300 -30 3Wav~en~th( m) Ratio of high-frequency breakdown voltage to static breakdown voltage as afunction of frequency for a uniform air gap. the electrode spacing d.

The shape of the curve in Figure 4. (1) In an experiment to measure a for a given gas. the electrons are lost by diffusion and the breakdown mechanism is known as the diffusion-controlled breakdown mechanism. electrons are lost by virtue of their mobility to the electrodes and the breakdown mechanism is called the mobility-controlled mechanism. Nonuniform-field gaps also show differences between their high. the electrons in the gap oscillate at increased frequencies and some of them would fail to reach the anode during the half-cycle in which they were created. With the spacing increased to 0. the breakdown voltage drops sharply. Such nonuniform fields are experienced in cases of transmitter antennas and their insulation.1’7) (Nasser. as at low frequency..01m. For point-to-plane gaps. (a) Calculate a. the current increases to 1 2 . When f fco.1’7 can be explained as follows.Since electron drift velocities are two orders of magnitude higher than those of positive ions under the same conditions. with the electrodes playing no role. . the magnitude off. At extremely high frequencies.7 x A at a voltage of l0 kV and a spacing of 0. 7x A for the same electricfield between the electrodes. is about two orders of magnitude higher thanf. 19’71). Whenf >fc0.005 m between the plane electrodes. At a certain frequency fc0 the breakdown mechanism is controlled by diffusion. electrons and ions oscillate in the gap. however. it was found that the steady-state current is 2. They would thus remain in the gap and partly neutralize the positive-ion space charge and the distortion in the field-hence the parabolic part of the curve. giving riseto higher currents whose phase relationship to the applied voltage is controlled by the rate of electron-ion recombination relative to the frequency. (b) Calculate the number of electrons emitted from the cathode per second. the highfrequency breakdown voltages increase with gap length. (Fig. All this discussion has been confined to uniform-field gaps. The breakdown field strength can be determined by relating the field-dependent ionization rate to that of charge loss by diffusion ( ~ a c ~ o n a l1966). At the higher frequencies. The decrease in breakdown voltage in the lower part of the frequency range above f c is caused by the distortion of the electric field in the gap by the positive-ion space charges accumulating in the gap. 4. The breakdown voltage even exceeds its static level at the higher frequencies wherethe electrodes do not contribute to the ionization process as they do under DC or power-frequency AC. d.and lowfrequency performances. but are different in magnitude.

01 0.7a): 11 = roead'. calculate the size of the developed avalanche at a distance of 0.82 3 Avalanche size = e8. If an electron starts its motion at the cathode.72/460.b ~ " / ~ ] 2 = a(0.687 x lolo electrons/s 1.0005) -b(0.0005 adx a dx = (a bfi)dx = [ax . = roeadz I2 = InI1 a = .5 = 0.0005 m.005 2. determine the electrode spacing that would result in an electron ~ultiplicationof lo9.6 x 10-19 (2) At the conditions of example (l). b = 15 x lo5.2) for nonuniform fields: n = noe 0.7 X = 460.045111 (3) In a nonuniform field near a cathode.72 and d = 20.0005)3/2 = 8.82 = 6768 [ 1 0 . a is expressed as a =a bfim-l where a = 4 x lo4.ad = 20.'.5m-1 no = Io = 1. " -1 d2 d l 1 1n 2*7x 0. solution^ Rewrite equation (4.140 Solution: (a) Using equation (4. Solution: lo9 = ea' . and x is measured -from the cathode surface in meters.

2) for nonuniform fields: x (a-bx)dx= . Use the a and b values givenin example (3). determine the rninimum distance measured from the cathode at which an electron may start an avalanche having a size of 10~~.001 ax-^]^ =91n10 0.1 4 1 1 (4) If an electron starts at a distance of 0.001 A quadratic equation in x.5 x 1 0 % ~ 4 x 1 0 ~43.0011m or x = 0. Photoionization .001 m from a cathode in a field where a = a bx m " ' .2) for nonuniform fields: or 7.so two values of x are expected x = 0.0518 m away from the cathode (to be disregarded where a becomes negative for x >0.001l m (6) Derive an expression for the total number of electrons n reaching the anode when photoionization is produced uniformly by x-rays in the gas betweentwo parallel plates spaced d meters apart. Solution: Rewrite equation (4. solution^ Rewrite equation (4.75 ~ = 0.027111) or x = 0.0 + x = 0.find the distance it must travel to produce an avalanche of lo9 electrons. (5) In example (4).001 54m away from the cathode.0522 m (to be disregarded) Minimum distance = 0.

5 x lo6v/rn 0. Calculate the total secondary coefficient of ioniza.1). (7) The breakdown of air between two parallel plates spaced by a distance of 0. Using equation (3. n = 0.15) for evaluating a. 4.19.kPa". Assume the area of the plates is equal to l m2. The plates are stressed by a voltage enough to cause ionization by collision with a electrons produced per electron per unit length along the field direction between the plates. which means that there are no initiatory electrons for electron-multiplication by collision. dn = na dx + nph dx = (na + nph) dx [ln(na a 1 + nph)]: =d n =-(e c u d __ a 1) If nph = 0. dn = number of electrons produced in a slab dx at a distance x measured from the cathode (Fig. __ P are-B~lE - E=--9 x io3 4.7m-7.Use I' and B values listed followingequation (3.l occurs at a rate of nph ionization/m3per second.84O~/m. B = 273. ~olution~ Let n = rate of electron production at distance x.253.002 r = 11.kPa .002 m is 9 kV. ~olution.

(b) the secondary ionization coefficient.7975/0.22 = 0.7a). ~ol~tion~ (a) Using equation (4. 1.005 = 39. * .5988/0.199/0.5~106 = 2396.r e a k ~ of o Cases ~ ~ p = 1 atm = 101.22 = 0.253. is y(ead 1) = I .and 2.97 m" At d3.019 = 41. Calculate: (a) the first ionization coefficient.3/4. = I0effld1 = 1.25 m-' Breakdown criterion equation (4.008 36 (8) Three measurements of the current between two parallel plates were 1.015 04.3 x 11.*.82 = 0.22.7e-27.199 a 1 = 0.019 m.221~ . E / p and p were maintain~dconstant during the ~easurements.3kPa a = 101.8).015 04= 39.82. (b) At d3: . y = 0.005. 0. respectively. a2d2= In 1.8 m " ' Similarly. aldl = In 1.380~101. the y-mechanism must be acting since a should not change as long as E / p is constant.5988 a2 = 0.8 m " ' and a3d3= ln2. and 0.7975 a3 = 0.22 of the initiating photocurrent lo at distances 0.

001 m.2.133 kPa .22 = 1 ___ y(e399.8x0.66V (10) Calculate the breakdown voltage using (a) Raether's criterion and (b) Meek and Loeb's criterion for a uniform atmospheric air gap having a spacing d = 0.019 1) .8 [ I n ( " + 1 1)] = 0.085111 As Efp = 12.*.8x0. Sol~tion: (a) Using Raether S criterion: The radial component of the space-charge field at distance r from the avalanche head where the size is e'lx Er= r2 = 3Det Y eeux ~ X E r2 O (4. determine the distance and voltage at which transition to a self-sustained discharge takes place.*.8) y(ead 1) = 1 or Distance at the transition to a self-sustained discharge =1 0. y = 0.000 V/m.00OV/m+kPa and p = 0. Sol~tion~ At the transition to a self-sustained discharge (equation 4. E = 1596V/m Voltage at the transition to a self-sustained discharge = 1596 X 0.019 __ e39.15) for evaluating a.085 = 135.0354 (9) Keeping E/p and p constant at 12.kPa and 0. 1964) .133 kPa respectively in example (8). Use l? and B values listed following equation (3.22) (as assumed by Raether.0354 39.

001 m. V was considered equal to 1.29 x lo4 V/m The breakdown voltage Vs = Ed = 595. equation (4. e + The numerical value of 1964).22) Er= eeaXE 4 7 ~(2 0V X ) At breakdown.81 x 101.3kPa. alp = 106.7 lnx. + For uniform fields.840V/m.001 = 5.23) = 17.253.81 V/m. .3).a = 10. (4.kPa E = 58764.1 l). Er= E and x = x . the electron energy eV is equal to (Raether.7m-'kPa" and B = 273. 1964) .50) De kT " Ke e r2 = 3-- kl" x e E kT Following equation (3.21~" As p = 1 atm = 101.792.According to equation (3.54m"'kPa" + ' = 11.23) takes the form a d = 17.29 x lo4 x 0..3 = 595.95 kV (b) Using Meek and Loeb's criterion: Using equation (4.5V (Raether. . r2 = -2vx E Substitute in equation (4. = lnIn x .7 lnd AS d = 0. 8 ~ tV ~ . a x .kPa Use I As alp = E/p = 58764. o ea& e 8 ~V ~ 0 a x .

27 x 10-7@/d)1/2aead = E at breakdown p = l atm = 101.determine the travel time of positive ions from one electrode to the other. If the current reaches a peak and declines abruptly in 0.2 X lom6 = 0.24) and (4.05 m spacing.25) iteratively.68 kV.2ps = 0. obtain a at atmospheric pressure (a= pI'e-Bp/?) and substitute at the right-hand-side of equation (4.25) Solve equations (4.25 X 10-6m/~ Time constant = a= ~ a% 35 x 10-gS 10-6 1 35 x 10-9 x 0.4 x m2/s*V.001 = 4.25 x = 114.3111-l (12) An alternating voltage of 200 kV (rms) is applied to a 10 cmuniform gap in atmospheric air. If the exponential rise ofthe current has a time constant of 35 ns.3kPa (4.05/0. The breakdown voltage Vs= 468 x lo4 x 0. (b) What is the maximum frequency that can be applied and still just pemits the clearing of all positive ions? .25) are satisfied simultaneously at E = 468 x lo4V/ m.2 x S Electron drift velocity V. (1l) An oscillogram was recorded for the electron current of an avalanche in a uniform-field gap of 0. For each assumed value of E.24) (4.24) and (4.2 vs after the release of the initiating electrons at the cathode. change the assumed value of E and repeat the procedure until equations (4. determine the electron drift velocity.24) to check its equality to E. solution^ Transit time of electrons z = 0. = d/t = 0.14 = 5. (a) If the frequency ofthe voltage is50 Hz and the mobility ofpositive ions is1. calculate the first Townsend's ionization coefficient. If not.

TN. Proceedingsof 4th International Symposium on Gaseous Dielectrics. High 'Voltage Technology. Knoxville. pp 507-5 12. Dring D. Brago EN.27) k+Ea f= sin ut 1. Bazelyan EM. Proceedings of IEEE-IAS Annual Meeting. Shamloul D. Berger G.14t) 1. 1953.079/314 = 0. .4 x 10-~ x 2828. IEEE Trans IA-24:1025. 1976.2 HZ 2n x 0. Universiti de Paris-Sud. Sov Phys-Dokl 52:101. Farghally M. 1961.26 2n x 50 314t = 0.26): 2nx The maximum time available before the voltage reverses isat ut = n/2 and x = d (4. Der elektrische Durschlag von Gasen. 1988. IEEE Trans IA-24:1031. IEEE Trans PAS-98:597. Chicago. Abdel-Salam M.1 \ x=o. IEEE-PES paper A-78-601-7. 1979. Radwan R.26) 0. 1968.4 x lop4 x 2828.1 = 1. 1973.7 ~al~t~an~ (a) Let the alternating field E be described by E.4 x lo3V/rn (4. -cos ut d(ut) = u u Ea = 2oo x x lo3= 2828. cos ut t Ea k+Ea sin ut k. J Phys D: Appl Phys 9:L-149. Abdel-Salam M. Turkey A.4 x lo3 = 630.1 Abdel-Salam M.079 rad t = 0.253 XIS (b) From equation (4. Stekolnikov IS. IL. Hashem A. Hahn R. Alston LL. Ali K h . Podporkyn GV. PhD thesis.4 x lo3 sin(2n x 50t) = sin(3. Proc IEEE 128:565. Khalifa M. 1984. Allen NL. Abdel-Sattar S. Abdel-Salam M. 198 1. Abdel-Salam M. 1988. Paris. Berger G. 1978. Ganger B. 1976. Berlin: Springer-Verlag. Oxford: Oxford University Press. Abdel-Salam M. Alexandrov GN. Stanek EK.

1977. eds. High Voltage Engineering. Raether H. Weck KH.CA:Universityof California Press. Khalifa M. 195’7. 1955. Schneider KH. New York: John Wiley & Sons. Aoshima Y. Waters RT. Tashchini A.Goldman M. Llewellyn-Jones F. 1973. New Pelhi: Tata IMcGraw-Hill. Nasser E. Ionization and Breakdown of Cases. . Loeb LB. 1989. Electra 29:29. IEEE Trans PAS-96:886. Craggs J P .Coldman A. New York: John Wiley & Sons. Aihara Y. Ali Kh. Radwan R. Corona discharges. 1971. Naidu MS. In: Hirsh MN. New York: Academic Press. pp 219-290. Fundamentals of Gaseous Ionization and Plasma Electronics. Electron Avalanches and Breakdown in Gases. BasicProcesses of Gaseous Electronics. 1978.Berkeley. New York: John Wiley & Sons. 1973. 1996. Rizk F. 1978. Abdel-Salam M. eds. Oskam HJ. pp 385532. Microwave Breakdown in Gases. Paris L. 1966. Spark breakdown in nonuniform fields. London: Butterworth. Kamaraju V. IEEE Trans PP-4:596-606. In: Meek JIM. Gaseous Electronics. IEEE Trans PAS-921085. Electrical Breakdown of Gases. London: Methuen. Harada T. 1964. MacPonald AD.

radio interference.The Corona University. The chemical reactions that accompany corona in air produce the smell of ozone and nitrogen oxides. Assiut. electric current. suchas visible light. This partial breakdown of air is quite distinct in nature and appearance from the complete breakdown of air gaps between electrodes. On the other hand. such as at the surface of a pointed or cylindrical electrode opposite to and at some distance from another. Egypt Assiut University. “Corona has long been a main concern for power transmission engineers because ofthe power loss it causes on the lines and the noise it causes in radio and TV reception. mechanical vibrations. Giza. energy loss. audible noise. The corona is also distinct from the discharges that take place inside gas bubbles withinsolid and liquid insulation. Egypt ““Corona” literally means the disk of light that appears around the sun. The same applies for other gases. as in Van de Graaff generators. The corona discharge is accompanied by a number of observable effects. corona does have several beneficial applications. electrostatic preci1 . The term was borrowed by physicists and electrical engineers to describe generally the partial discharges that develop in zones of highlyconcentrated electric fields. and chemical reactions. although the underlying phenomena of gas discharges are the same.

Landham et al. 5. 1972. 5. ozone production. At the onset level.1 and 5. 1974a). this is known as Trichel pulse corona.. and the discharge takes the form of a 6 “ g l ~ covering ~9’ a sig~ificantpart of the V conductor surface (Fig. electrostatic deposition. 1979).3 presents the range for each type of corona discharge with positive DC voltage applied across a point-to-plane gap (Nasser. 1967.2). Therefore. 5. overlap in space and time.pitators. it produces an electron avalanche (Chapter 3). and slightly above.2b). it will be discussed first for each polarity under DC. and ionization counting (Berg and Hauffe. If conditions are favorable. 5. corresponding to irregular. the corona at the cathode hasa rapidly and steadily pulsating mode. This is in contrast with current pulses corresponding to onset streamers (Fig. The positive current pulse corresponds to a succession of generations of electron avalanches taking place in the ionization zone at the anode (Khalifa and Abdel-Salam. electrostatic printing. tiv At the onset level and slightly higher. breakdown eventually occurs across the air gap. Each . 1965). This mode of corona consists of what are called onset streamers. 5. The cloud of positive ions produced at the avalanche head near the anode forms an eventual extension to the anode. Secondary generations of avalanches get directed to the anode and to these denseclouds of positiveions (Fig. The discharge process depends on the polarity of the applied voltage.2) (Ciao and Jordan. Figure 5. Ifwe continue to raise the voltage. 1971).2). high-am~litude current pulses (Figs. They are short in length. 5. 1) near the anode surface such that the onset-type streamers becomevery numerous. Seelentag. The corresponding current through the HV circuit becomes a quasi-steady current (Fig. 1987. Khalifa. At still higher voltages the clouds of negative ions at the anode can no longer maintain their stability and become ruptured by violent pre-breakdown streamers. 1979). the high field space at the anode may suit the formation of streamers extending tangentially onto the anode. these are called “burst-pulse streamers’’ (Loeb. At slightly higher voltages a cloud of negativeions may form (Fig. When a free electron is driven by the field toward the anode. as described in detail in Chapter 19.1). there exists a small volume of space at the anode where the field strength is high enough to cause ionization by collision.

* GLOW * J PRE. The photoelectrons thus produced can startsubsidiary avalanches that aredirected from the cathode. 5. 1974b. In this case the ionization zone extends from the cathode surface outward and as far as the point where the fieldbecomes too weak for ionization by collision to compensate for electron attachment.1 Development of the first and subsequent generations of avalanches in positive corona discharges.BREAK 5. Zeitoun et al.. current pulse corresponds to one main electron avalanche occurring in the ionization zone (Fig.4). more and more of the avalanche electrons get attached to gas molecules and form negative ions which continue to drift very slowly away from the cathode. Beyond such a point. During the process of avalanche growth. 1976). 5.4) (Khalifa and Abdel-Salam. some photons radiate from the avalanche core in all directions (Fig. The motion of the electrons and negative ions away .

] of different . [From Giao andJordan (1967).l ur Photographs and corresponding current oscillograms modes of positive corona.

1 0 0 +50 5 1 0 1 5 GAP d 20 25 3 0 (cm) ure 5.] from the cathode and that ofpositive ions toward it correspond to the corona current pulses flowing through the high-voltage circuit. Nasser (1971) and Wiley-lnterscience. and its influence on the residual space charge left over from the discharge during preceding half-cycles(Fig.5a. At still higher voltages pre-breakdown streamers appear. 5. With an increase in applied voltage.5~).4 and 5.4 and 5. Thus positive onset streamers and burst-pulse streamers may appear only over an extremely small range of voltage at onset.150 4 . eventually causing a complete breakdown of the gap (Figs.6). 5. 5. followed by a posi- . and can easily be computed.3 Onset voltages of various positive coronamodes and sparkover voltage as functions of point-to-plane gap spacing. the Trichel pulses increase in a repetitive rate up to a critical level at which the negative corona gets into the steady “negative glow” mode (Figs.5b).(kV) + 250 + 200 $ . oron The basic difference betweenAC and DC coronas is the periodic change in direction of the applied field under AC. as shown in Figure 5. [Courtesy of E.

Under AC a very large number of traces get superimposed on each other. 0th negative Trichel pulses and negative glowcan be observedin an AC corona.V. GL Development of electron avalanches in negative corona discharges.. . . Under impulse voltages. electron avalanches and streamers extend over significant distances. Therefore. both positive and negativeglows and streamer coronas can be observed in each cycle.7). Such a figure cannot be produced by DC because of the choking effect of accumulating clouds of space charges. depending on the electrode geometry. then. The onset streamers produced and their branches can easily form traces on photographic films in contact with the anode or cathode.H. in what is known as Lichtenberg figures (Fig. the corona starts inan air gap almost clear of any space charge. If the applied voltage has a suitable magnitude. 5.

] .Photographs and corresponding current oscillograms of various modes of negative corona. [From Giao and Jordan (1967).

Note that when the voltage Va+ is below the corona-onset level V.Period (1)-(2) Period (2) (3) P Period (3) (4) Period (4) (5) AC corona current for a symmetrical gapbetween two parallel conductors a and b.. . the corona current corresponds to the motion and recombination of residual ions in the gap between the high-voltage conductors during the periods shown in ’ the oscillogram.

(right) negative polarity.( M ) Photographic traces of corona discharge under impulse voltage (Lichtenberg figures): (left) positive polarity. setup. (b) test .

and A. the relative air density. correspondingly. At any point of microroughness on a practical conductor surface the field will be highly concentrated and the .3.9 and 1.1. r being the conductor radius in centimeters.3/&) kVpeak/cm. 1977).2) E o = 306 kVPeak/cm A. this is different from practical conditions. + This can easily be calculated using Eo or E* once the conductor arrangement and dimensions are known. It should berealized that field calculations would normally be basedon the assumption of perfectlyclean. ~xperimentally measured Eo values fit the following relations: For AC: + (5.94~ 273 8 with 6 ranging between 0. " .2. 11. It has been well established by experiment and by computation that corona discharge starts at the surfaces of HVelectrodes and conductors when their surface voltage gradients reach a critical value E o (Abdel-Salam and halifa.9 to 7.are in the respective ranges 31 to 39. The magnitude of Eo depends on the voltage polarity and on the pressure and temherature of the ambient gas. B. a slightly higher field strength Ev corresponds to corona beingclearlyvisible on the conductor surface and can beexpressed as Ev = 306(1 0.are. Humidity has a minor effect.4 and 9.4 to 40.8 to 8. as shown by experiment and computation (Chapter 14). Under AC. and B. Thus a = . Methods of field calculations were discussed in Chapter 2. the current corresponding to corona discharges under traveling extra-high-voltage surges has the beneficial effect of reducing the surge peak and front steepness. smooth conductors. Thus their stresses on the power system insulation are relieved. referred to STP.3. The air pressurep (kPa) and temperature 8 ("C) are usually combined into one factor 6.On long HV transmission lines.8 and 29.

the number of electrons NI in the first avalanche is equal to the number N2 in the ~ Positive spdce charge zNv\r photons Development of the primary avalanche and its successors in the vicinity of a positive stressed electrode. Due to the photo-ionization process.3.6 for new rough-stranded conductors and about 0.8).2.1 Case of Positive Polarity The first avalanche proceeds toward the anode and ends at its surface. 5. 5. 1974a). . At coronaonset voltage.1 critical field strength for coronaonset will be reached there. This factor is usually taken as about 0. and photons are emitted from its core. They can also be computed according to an algorithm based on the ionization and deionization process acting in a coronadischarge of either polarity (Khalifa and Abdel-Salam. It develops a cloud of positive ions. while the average field strength over the entire surface remains considerably lower.85 for weathered conductors. these photons generate photoelectrons to startsuccessor avalanches (Fig. A corresponding surface factor should be taken into accountwhile estimating the corona onset voltage Yo for the conductor arrangement.3). Onset voltages of DC corona canbe estimated using the relation (5.

With the growth of the avalanche. the preceding avalanche should somehow provide an initiating electron at the cathode surface.2 Case of Negative Polarity When the applied electric field strength near the cathode surface reaches the threshold value for ionization ofgasmoleculesby electron collision. This instability results in the formation of luminous filamentary streamers (Loeb.9).successor avalanches.2.v w v u photons Deve~oPment Of an avalanche in the vicinity of a negative stressed electrode. .4) The onset voltage does not appear explicitly in the relation (5. more electrons are developed at its head. 1965. pos- 4 - 0 Positive space charge electrons f . and more positive ions are left in the avalanche wake. more photons areemitted in all directions.4) (Khalifa and Abdel-Salam. 5. 1971).4). Therefore. 1974a). it is implied that the level of ionization in the successor avalanches is such that sufficient photons and hence photoelectrons will be generated to make the ionization self-sustainedwhen N2 2 N I (5. an electron avalanche startsto develop along the direction awayfrom the cathode (Fig.3. and the first avalanche becomes unstable. 1985). and the onset voltage is the critical value which fulfillsthe equality (5. Nasser. For a successoravalanche to be started. 5. The computed onset voltages are in good agreement with experiment (Abdel-Salam.

sibly by photoemission. 1974b). metastable action. where Nepb is the number of electrons photoemitted from the cathode. .5) (Khalifa and Abdel-Salam. only the first mechanism (electron emission from the cathode by photons) was considered in the mathematical formulation of the onset criterion. the voltage level corresponding to positive corona onset can be calculated according to the method described above for a monopolar corona.. The streamer is built up by a number of successive generations of electron avalanches which take place in the ionization zone around the positive HV electrode (Khalifa and Abdel-Salam. Therefore. positive ion impact. At the negative conductor. 1998). Abdel-Salam and Stanek. the ionization zone expands and its contour changes. or 5. and the onset voltage is the critical value which fulfillsthe equality (5. It is known that the onset voltage gradient for a positive corona is slightly lower than that for a negative corona. 1977. 1988). 1971). there are positive ions drifting toward it and coming from the positive corona discharge at the positive conductor. The ionization zone around the HV electrode is defined as the space where the resultant field strength is so high that the first Townsend’s coefficient of ionization is greater than the coefficient of electron attachment. Electron emission by positive ion impact is more than two orders of magnitude less probable than photoemission (Loeb.9.1 Case of Positive Polarity Positive pulsating corona discharges occurnear the onset level as onset streamers and at much higher voltages as pre-breakdown streamers (Nasser. 1974b. Field emission is possibleonly at field strengths exceeding 5 x lo7 V/m (Rein et al. These ions enhance the field intensity at the negative conductor and cause a corona to start at a voltage slightly lower than the value calculated for the monopolar case. 1974a). The computed onset voltages are in good agreement with experiment (Khalifa and Abdel-Salam. A1-Hamouz et al. or field emission. 1977). Therefore.. The onset voltage does not appear explicitly in the relation (5. which has not yet reached its corona onset level. i..3. Withsuccessionsof avalanches and accumulation of ion space charges.3. where at least one photoelectron is emitted by the photons of the first avalanche to keep the discharge self-sustaining. 1971). This slight difference can also be computed (Abdel-Salam and Khalifa. Metastables have been reported to have an effect approximately equal to that of positive ion impact (Nasser. 1965).e.

The primary avalanche is initiated by an electron 'driven by the electric field outwards from a point of microroughness on the conductor surface.2 Case of Negative Polarity Different from the positive case. which is of the order 1 ns.e. 5.theinstantaneouscurrent of a given ge~erationof avalanches is calculated as the sum of the components corresponding to the individual avalanches. They become involved with air molecules in exciting and ionization collisions to produce an avalanche. . and the current is averaged over the life of each generation. the negative corona pulse corresponds to one primary avalanche accompanied by a seriesof auxiliary avalanches which develop in the discharge zone around the H V electrode during the life of the primary avalanche (Khalifa and Abdel-Salam. Most of the new avalanches of the succeeding generation get started almost at the ins temination of the previous avalanche when its population of electrons is The corona current throughthe H V circuit corresponds to the motion of electrons and ions in thespace between the HV conductors. The photonsreaching the conductor emit new electrons of number Np which start the successor avalanches.3. 1974. Photons are emitted by the avalanche in various directions. hest st.b). when the first electron leaves the conductor) the first successor avalanche gets launched. Each photoelectron produced within the ionization zone will be accelerated by the prevailing field from its point of origin and start one avalanche of the second generation. The calculations are continuedover successive generations until the magnitude of the current falls below 1% of the maximum value reached for any generation. It grows in the same way as the primary avalanche and contributes to the irradiation of the conductor surface. and a cloud of positive ions are produced in the primary avalanche. The motion of ions can be safely neglected as compared to that of electrons during the corona pulse. the capacitance of the H V and measuring circuits smooths the variationof the current from instantto instant during the life of each generation. photons. and so on.The electrons are driven to the electrode by the resultantof the field of the applied voltage and the field of the space charge. By ionizing and exciting collisions withthe air molecules. In experiment. When Npreaches unity (i..3. Photoelectrons are produced in various locations from which they start a new generation of avalanches. Thus. The current pulse is then considered to have ended. leading to the emission of more and more photoelectrons which start new successor avalanches. more and more electrons.

To calculate the pulse repetition rates.9% of the electrons in the avalanches have becomeattached to air molecules and formed negative ions.As the electrons drift away from the conductor in the decreasing field. 5.3 Pulse Amplitudes and Repetition Rates The a~plitudes of the positive and negative corona pulse are evaluated during the process of calculations as the maximum values obtained in the computations of the current pulse. The discharge-zone boundary is defined as the place where 99.the new pulse willnot startuntil both positive and negative ion space charges have left the discharge zone. 1974b). % 0. Different from positive pulses. the favorable condition is taken as the positive ion clouds being swept so far from the high-voltage electrode that their residual field at the boundary of the ionization zone becomes negligible. The bigger the charge content of the calculated pulses.3. i..1of the applied field strength. The calculated pulse repetition rates are closer to those measured. Also.The positive ions are swept to the H V electrode. 1974b).3. 1974b). In the case of a positive pulse. one has to define the conditions to be fulfilled after the termination of a pulse in order that the subsequent pulse can be initiated (Khalifa and Abdel-Salam. This makesit more difficult for electron avalanches .1% of the field strength produced there by the applied voltage. The shapes and durations of the calculatedpositive and negative pulses agree quite closely with experiment. charge contents of the pulses are easily evaluated by integrating the current waves. For negative pulses. the calculated negative pulse amplitudes are almost equal to those measured (Khalifa and Abdel-Salam. The negative ions move so far from the electrode that their residualfield strength at the boundary of the discharge zone becomes negligible. ssi Because sulfur hexafluoride is an electronegative gas. The magnitudeof the corona pulse current at any instant is calculated as the sum of the current componentscorresponding to the individual avalanches. i. The positive and negative ion clouds grow with time as more avalanches are formed. the longer the time they take to disappear from the ionization zone and allow for the initiation of the next pulse. The space where the field strength is high shrinks and eventually no more avalanches can develop. where they are neutralized. it has a high affinity for electron attachment.e. and the smaller the repetition rate (Khalifa and Abdel-Salam.. more and more of them get attached to air molecules and form negative ions. 0.e.

6) where f is the frequency. In properly designed transmission lines. 1986). Y the line voltage. 1979). 1977.1 to grow.4.1 AC Lines Empirical formulas were suggested early in this century by Peek and Peterson for estimating P. Peek's formula was superseded by that of Peterson. Abdel-Salam.7 to 17 kW/conductor*km for 700 kV lines (Electrical Power Research Institute.. Peterson's formula can be modifiedby including the capacitance geometric meanradius of the bundle instead of the single-conductor radius. K is a factor depending on the ratio of the operating voltage Y to the corona onset line voltage Yo (Fig.4). Thus corona and sparkover occur at voltages considerably higher than those in air. Above a certain critical gas pressure sparkover occurs across the gas gap without any preceding corona (Section 4. 1978).73K x kWlconductor km + (5. There is an optimal separation between subconductors in the bundle that corresponds to minimum corona loss. (1984) on the basis of the physical phenomena of corona discharge (Shamloul. ~ ~ eo fc ~to n d u c t o ~ dZj~g. . the corona loss in fair weather is usually insignificant.km for 500 kV lines and from 0. the separation between subconductors in thebundle has aneffect on the amount of corona loss.7 kW/ conductor. 5. Thus electrons produced at the cathode by photoemission and field emission would have to contribute more substantially to the discharge in order to maintain its stability.1. which takes the form P. and D and r the phase conductor separation and radius. Taking both these processesinto account has made possible the computation of breakdown voltagesof gaps in compressed air and SF6 (Khalifa et al. 5. For transmission lines with bundled conductors. 1989). Because of several flaws.. = -p2 (o/u>2 3.3 to 1. Empirical formulas have been suggestedfor evaluating corona losses on AC lines and on both monopolar and bipolar DC lines. A much more recent and more scientific approach was the computer program developed byAbdel-Salam et al. At such highpressures the coefficientof ionization by collisionbecomeslower than the coefficientof electron attachment: a (Chapter 3). Typical values measured range from 0.the fair-weather corona losses ofoverhead transmission lines (Begamudre. Naturally.10).

o K 1. This particular phenomenon accounts for the corona loss on bipolar DC lines being much higher than on monopolar lines. The same goes for humidity unless it approaches 100%o. wind. 5. (1982).41 -2 DC Lines In unipolar DC lines. depending on the rate of rainfall (Electrical Power Research Institute. as the coronating points are more numerous in the former case. as computed by Abdel-Salam et al. 4 6 8 30 1 2 1 4 1 6 1 8 10 1 . the corona losses increase 10-fold or more. the fair-weather corona loss on a bipolar line was found to be about twice that on a three-phase AC line. On . snow. pressure. and between the conductors and ground. With condensation. rain.10 Factor K t o be used in Peterson’scorona loss formula as afunction of the per-unit operating voltage referred to the corona-onset voltage. we have ions of only one polarity in the space between conductors. With bipolar lines.100 K V IV . The increase in corona loss due to rain and dust on conductors of large diameter is much greater than with smaller conductors. In comparison with AC lines having equal per-unit effective voltages with respect to the corona onset levels.1 PER UNIT VOLTAGE V/ V. The air temperature and pressure are included in the factor S. humidity. and evenhigher than on AC lines. and dust. we have both positive and negative ions in the interconductor space and there is a high probability of ion recombination. and much more seriously with rain. however. Figure 5. The principal weather parameters are air temperature. No perceptible effect of wind on the AC corona can be noticed.0 0. ather. 19’79).

6.4 7.4. _ . the emissionis regulated in such a way that any increase in the space charge due to a new emission causesa reduction of the surface field and therefore a slowing down of emission. " . at 3333 kV and at different wind a r r a n ~ e ~ eis nt indicated.2. .increases in bipolar line corona loss with voltageand in foul weather are not as rapid as in the case of AC lines. Abdel-Salam et al. 5. . _ I _ _ _ _ _ _ _ 15 mls Appearance of bipolar DC corona on two conductors. 1972) and theory (Khalifa and Abdel-Salam. Therefore.5 mm in diaspeeds. 1969. The corona loss of monopolar lines can increase by about 20% if humidity increases from 60% to approaching 100%.the other hand.5 . ... . and charges of the same sign as that of the conductor potential areemitted into space and move away from the conductorin the form of shells.1 AC Lines As the AC voltage applied to a single conductor reaches a critical value V**. The effects of atmospheric humidity and wind can be measured and computed (Khalifa and Abdel-Salam. 1973) that the electric field at the 0 4. thus absor~ing a part of the applied voltage.. 5. 1984). while reducing the field at the coronating conductor surface by provokingan increase of the induced opposite charge there (Clade et al. . The charge in a given shell raises the field in the space outside the shell.11). It has been realized by experiment ~ a t e r et s al. The Conductor 45 mm apart. . 1974b). . the electric field at the conductor surface reaches the corona onset value Eok. A much more significant increase occurs if wind blows across the conductors of bipolar lines (Fig. . Corona startswhen the electric field exceedsthe onset value.

~xperimentshowed that the corona terminates after the voltage peak of the positive and negative half-cycles at V&. In the succeeding half-cycle. corona starts at voltages Vi* Vok due to the effect of the residual space charge. Vi& = V o . Chargedevelopmentduringthe first half-cycle: r = conductor radius. located. The voltage Vo* is the onset voltage at which corona starts in the first positive or negative half-cycle. The conductoris simulated by a set of unknown line charges uniformlydistributed inside the conductor and coaxial. Rf = radius of a fictitious cylinder where the simulation charges are H = conductor height above ground plane.determined by a simple relation += Of course. and . whose area expresses the corona loss per cycle. Consider a single-phase conductorto-plane arrangement as shown inFigure 5.12a. Many experiments have been done to study the performance of AC corona.surface of a coronating conductor assumes a value less than the onset value. for the first positive or negative half-cycles. Vi*are termed the ionization onset voltages. 1992). It has beenassumed(Clade et al. The magnitudesof these line charges are determined to satisfy the boundary conditions (the potential calculated at the conductor surface es. isthe most important outcomeof these experiments to evaluate the AC corona loss (Abdel-Salam and Shamloul. 1969) that the ionization during each cycle t e ~ i n a t e s at the voltage peak ( V .. ) and starts at voltage Vi*. The corona starting and ending voltages are prerequisites for drawing the VQ curve. The V Q relation.

j = 1. some linecharges disappear totally due to many successive recombination processes...The size of the volume element changes from time to time as the line charge departs away from or returns back to the conductor. the same as explained before for the first positive half-cycle. Sometimes. Corona loss depends on both the amount of charges liberated in space and displacement of these charges through the prevailing field Efin the wire neighborhood.2... . u2. are determined at the points PI. They start their journey at the points u t . .(Fig.. When the applied voltage reaches the positive onset value VO+ for the first time. are Qoj+. ion emission starts from the = 1. P2. 5. The space linecharges continue to move in opposite directions and will undergo further recombination processes Once they meet the opposite charges. The ion emission.2.12~). the magnitudes of the unknown charges. 1969) as a block moving away from the wire or returning back to it. each time interval the conductor generates an infinitely thin cylindrical shell of charges. .12b). The liberated negative space charges move away from the stressed conductor... The magnitudes of the unknown charges are Qj+.. . Some of the positive ions are neutralized through recombination. 1 2 ~ and d show the volume elements abcd and dcef represented by the mth line charge and the nth line charge of the first and second emission respectively. . on the conductor surface under the prevailing field (Fig.. For the first negative half-cycle. . Each shell was treated (Clade et al. ion emission starts at Vi-. dictates that each space line charge represents the charge over a volume element per unit length of the wire. being assumeddiscrete.j = 1. the magnitudes of the emitted space charge Qsj+ are assumed(Abdel-Salam and Shamloul. and terminates at V+. . 1992) to be equal to the differencebetween the line charges Qj+ and the corresponding onset values It is quite reasonable to model the ion emission from the wire in a discrete form. denoted as the positive onset values. At the same time. Figures 5 . 5.. .the positive spacecharge returns to the conductor and meets the outgoing negative ones. conductor. The energy dissipation through a complete cycle is calculated as eoj+. When the applied voltage exceeds Yo+. All the procedures described above for emission during the first cycle are repeated for the succeeding cycles until a steady-state ion emission is achieved.j On the assumption of constant conductor surface charge.. . the space linecharges simulating the shell are not concentric with the conductor..2. Even the charges Qj+.must equal the applied voltage) at a number of boundary points equal to the number of unknowns. Due to the presence of the ground plane.

This results in a system of simultaneous linear algebraic equations whose solution evaluates the simulation line charges ofeach phase. method of calculating the corona power loss on three-phase transmission lines is similarto thatfor single-phase lines. The corona power loss is simply the energy dissipation W per cycle times the frequency of the applied voltage. The calculated values of corona power loss agreed satisfactorily with those measured (Abdel-Salam and Shamloul. The charges simulating each phase conductor surface charge are obtained by setting the potential at each boundary point equal to the respective phase voltage.and C the line-charge trajectory (since its emission from the conductor surface) over a cycle. the positive and negative onset field values are defined for each phase conduetor. 5. recombination betweenpositive and negativespaceline charges is taken into account. Also.4. The emissionlawissimilar tothat in the single-phasecase in which. The onset charge for each phase conductor will be the summation of those simulation charges located inside it (Abdel-Salam and Shamloul. Space linecharges are displaced under the electric fielddue to the conductor as well as the space line charges for the three phases. the com- . 1992). More detailed calculations of three. 12000). Basedon the conductor radius.where the summation range A is over all positive and negative space line charges QSi. Abdel-Salam and Abdel-Aziz. 1994b). Corona energy loss is calculated by Corona power loss is calculated by multiplying the corona energy loss by the system frequency. 1994b). However.2 DC Lines The corona power loss on monopolar and bipolar DC lines was computed using finite-element and charge-simulation techniques."whenever the magnitude of the conductor charge ofany phase exceeds the onset value. AbdelSalam and Abdel-Aziz.2. Each phase conductor of the three-phase transmission line is simulated by a set of infinite line charges. ~ ~ ~ e e -~ ~ ~~ a s ~a e s~ ~ ~s ~ ~ The ~ o proposed e ~. 1994. the surplus charge is emitted as space line charges.and six-phase corona powerloss are reported elsewhere(Abdel-Salam and Abdel-Aziz. 1994. The Calculated corona power-loss values agreed with those measured experimentally (Abdel-Salam and Shamloul.

voltages below the onset value. where the solution is initialized and the electric field is updated untila self-consistentsolution of the electric fieldand thespace line charges is obtained. ions are produced. 1984. Abdel-Salam and Abdel-Sattar. ~ultiplying the corona current by the applied voltage determines the corona power loss.3. the same as the surface charges on the conductors asdescribed in Section 2. Deutsche’s assumption.6. the locations of the simulation charges are located in the space outside the coronating conductor(s) where the field is to be calculated. . Satisfaction of the boundary conditions at the boundary points results in a system of nonlinear equations whose solution determines the unknown simulation charges. The technique to solvethese nonlinear equations is iterative in nature. the locations of the simulation charges are singularity points in evaluating the potential and the electric field there. and the current density is calculated using equation (2. whichmeans that the Laplacian and Poissonian fields have the same direction. At voltages above the onset value. The volume charge density is calculated by dividing each discrete line charge by the volume it occupies. 1986. To accommodate the boundary conditions. boundary points arechosen on the coronating conductors.the space charge causes a deviation between the two patterns. such as parallel plates or coaxial cylinders.4. An attempt was made to evaluate the potential and field at the singularity points in a parallel-plate arrangement (Abdel-Sal~m and Abdel-Aziz. For the Poissonian field around DC lines. the space charge is simulated by a set of fictitious discrete charges. such as transmission-line arrangements.6 is no longer valid. 1989. Therefore.putation of corona powerloss on AClineswaslimited simulation technique. In symmetrical arrangements. Images of these charges with respect to the ground plane are considered. Thus.58). 1994). As described in Section 2. the two sets of discrete charges have to be chosen to satisfy the boundary conditions of Poisson’s equation (2. the ions move along the Laplacian flux lines. Elmoursi and Castle.58). 1992). The corona current is obtained by integrating the current density over the periphery of the coronating conductor. filling the inter-electrode spacing. depending on the spatial distribution of ions in the inter-electrode spacing. Abdel-Salam and Abdel-Aziz.forming a drifting space charge of the same polarity as that of the coronatingelectrode. The ions flow towards the other electrode. Elmoursi and Speck. stated in Section 2. 1990. to the charge- i ~ ~ e At . However. where the fieldis Poissonian. the field in the inter-electrode spacing is Laplacian.3 (Horenstein. In asymmetrical arrangements.

Corona undoubtedly interferes also with carrier signals transmitted along EHV lines. Summing up for all the elements of the grid results in a set of simultaneous equations whose solution determines the unknown potentials at the nodes. equation (2.58). 1999) and bipolar lines (Abdel-Salam and Al-Hamouz. 1995a. duration.71 The calculated corona currents agreed with those measured on lines with single conductors (Abdel-Salam and Abdel-Aziz. 1997.e l e ~ e ~ t The~first e c step ~ ~ in j ~ the ~ finitee . element formulation of Poisson’s equation is to divide the area of interest (around transmission-line conductors) into triangular elements form in^ a grid (Aboelsaad et al. The electric field can be d e t e ~ i n e dsee . this has been shown by both measurement and computation. 1998). Al-Hamouz and Abdel-Salam. As the pulses are random in amplitude. Two field lines define a flux tube as field lines never cross. 1994a) and on lines with bundled conductors (Abdel-Salam and Mufti. Therefore. The continuity equation can therefore be extended to all the flux tubes. audible noise is experienced near EHV lines and substations. andrepetition rate. the current density is calculated using equation (2. The calculated corona current agreed with those measured for monopolar (Abdel-Salam and Al-Hamouz. as their amplitudes are much higher than those of the negative Trichel pulses. . caused by corona. 1994. Abdel-Salam et al. and other wireless reception. Corona noise includes interference with radio. d e t e ~ i n a t i o n of the electric field in the presence of space charge must satisfy the continuity of the current density. 1989. 1995b. e~ ~ ~ j t e . their noise is felt over a continuous spectrum. The mainsource of corona radio noise is from positive streamers (Fig. the ionized field is divided into flux (stream) tubes. Poisson’s equation is then transformed into a set oflinear equations by minimizingthe energy functional expressed by equation (2. 1999). 5. Also.58).1992). television. The coronacurrent is obtained by integrating the current density over the periphery of the coronating conductor(s).58). equation (2. The finite-element formulation of Poisson’s equation and the continuity equation of current density are solved iterativelyfor the electric potential and the space charge-density.2).1993. ~ultiplying the corona current by the applied voltage determines the corona power loss. The noise level decreases at higher frequencies. As the space-charge density is known at the coronating-conductor surface... However. Abdel-Salamand Al-Hamouz.. Al-Hamouz et al.35) over an element.

field strength Emax at the HV conductor surface) according to the relation RI = C(Emax)" dB (5. (5. in both AC and DC lines. positive corona tends to take the form of a steady glow. on the other hand. One of the plates is stressed with a voltage V with respect to the other plate which isgrounded. Whatwould be the number of electrons in the avalanche growing between the plates? Neglect the effect of self-space charge of the avalanche on its growth. the RI level was found to increase with the conductor radius according to the. Therefore.41) and (3. In the case of AC corona.172 ~halif~ and A~d~l-Salam In measurements of the radio noise level and its lateral profile near bipolar HV DC lines. (aV)/P = r :A(E/p) + C. positive streamer pulses occur during the positivehalf-cycles andare also the main source of noise.44).10) C being a constant.5 in rain. At voltages slightly above the onset level. the lateral decay of radio noise is less steep than for DC lines. HV DC lines are usually more noisy than AC lines at voltage gradients slightly above the onset levels. particularly in fair weather.e. ffect of Line ~ o n d u c ~ Size or Radio interference (RI) measurements under both AC and DC EHV lines have shown that the RI levelriseswith the voltage gradient (i. the highest levelwas recorded under the positive conductor whereas the noise contributed by the negative conductor was rather insignificant. For DC lines.the exponent n is about 7 to 8 in both fair and foul weather. For the same conductor voltage gradient. assisted by the negative ions produced during the preceding negative half-cycles. the exponent n has a value of 5 to 7 in fair weather and 1. E == V/d . Sol~tion~ According to equations (3..5 to 3. For AC transmission lines. relation RI = Clr2dB This relation was found to be independent of conductor bundling.11) (1) Two parallel plates are spaced a distance d and are placed in SF6 gas. ForAC lines.

42 = 0.44).13) is reduced to a = v. Vs = 9005. (II = 1. According to the criteria expressed by equations (5.4 (av) (5.32 x lo4 18.5). the breakdown voltage will be the same irrespective of the polarity of the stressed plate.12) =AE + Cp q)d] Avalanche size = exp[(a = exp[A V Cpd] + (5. Vs. . Solution: (a) As y is muchsmaller than unity.13) (2) In problem (l).027 and C = -2400.= 9.13).5) Vs+ = 9.32 x lo4 x Vs = 9689.41) and (3.13) 10' = exp[A V where Cp = -2400.3 kPa.1 occurs when (5. According to equation (5.70 kV +~ p d ] (c)As the avalanche self-space charge isneglected. (b) corresponding to an avalanche size of lo'.4) and (5. According to equations (3.4 where E = Vs/d A s p = 101.14)O.4) and (5. (c) according to the criteria expressed by equations (5.3 = -24.O27Vs 24. consider d = 1 mm and the gas pressure p = 1atm. equation (4.2kV .4 kV.3 kPa.4 x 101.9V = 9 kV (b) p = 101.6V = 9.O27(E/p) = 2400.17 where A = 0. Calculate the breakdown voltage Vs (a) according to equation (4.*.

14). (4) Two concentric spheres have radii a and b.71 kV at p = 5 atm (C) According to the criteria expressed by equations (5.(3) Repeat problem (2) for gas pressures of 3 and 5 atmospheres and comment on the results obtained.5 kV. Vs+ = 27. The interelectrode SF6 gas.42 = 0. The inner sphere is stressed with a voltage spacing is filled with V with respect to the outer sphere.= 0 where E = E i .027V = 27kV a t p = 3atm Vs = 2400. Vs = 2400.3 X 10-3/0.7 kV at p = 3 atm 24.3 x Vs = 27.027~ = 45.atp = 3 atm Vs-= 45. which is grounded. With reference to Table 2.5).13).12).004 x 3 x 101. At the ionization zone boundary a .2kV7 Vs. IO* = exp[A V +~ p 4 18. According to equation (5.3 x low3 18.4 X 5 X 101.4 X 3 X 101. Neglectthe effect of self-space charge of the avalanche on its growth.73 kV.4p~/0. b >a.027Vs Vs = 45.1.4) and (5.3 X 10-3/0.5 kV at = 5 atm It is quite clear that the increase of gas pressure improves the dielectric strength of the gassince the breakdown voltage increases with gas pressure. Determine the thickness of the ionization zone surrounding the inner sphere and the ma~imum size of the avalanche growing in this zone. E i = --Cp/A when a . solutio^^ .= 27. S~~ution~ (a) According to equation (5.027Vs 24.42 = 0.004 x 5 x 101.027 = 2400. Vs+ = 45.03 kV at p = 5 atm (b) According to equation (5.= 0.

. Calculate the corona onset voltage (a) corresponding to avalanche size of lo'.15) into equation (5. the latter depends also on V.16) Substituting E from equation (5.9.027.1 cm and the gas pressure p = l atm.15) -cpp = vab r. p = 101. (b) according to the criteria expressed in equations (5. b = 2.4 m. solve Given a = 0. (5.(b a) where ri is the radius defining the ionization zone boundary.1 x m. Maximum avalanche size = exp [ ~ (a!q)dr] (5. solution^ (a) Avalanche size is expressed by equation (5.16). (b) According to the criteria expressed in equations (5.17) iteratively to get Vb (corresponding to an avalanche size of 10') = 13.1 x lom2 equations (5.12) relating to (av). consider a = 0. a-q= r2(b a) Avab " cp The maximum avalanche size occurs when the avalanche grows over the whole thickness of the ionization zone.12).4) and (5. A 0.1 1 7 (5. as expressed byequation (5.16) and (5. C = -2400.1 cm and b = 2.5).17) which depends on the applied voltage V and the corresponding thickness ri of the ionization zone. In equation (5.17) (5) In problem (4).3kPa.92 kV.4) and (5.

solve equations (5. 5.13d.9. 5. (e) bundle-4 conductor with the four subconductors placed at the vertices of a diamond-form square.= 35.17) itera.9 kV. Fig.3kPa. 1977). Calculate the corona onset voltage.4) and (5.13~. V0. The maximum field Em values at the subconductor surface of the middle-phase of the line arrangements of Fig. 5.13a.1 Vo+ = 13.13e. where the effect of ground is disregarded (con>phase-to-phase spacing D). Soluti~n~ It is evident that the surface field strength is higher at the middle phase than at the outer phases. Fig. (c) bundle-3 conductor with the three subconductors placed at the vertices of an upright or inverted triangle. phase voltage = V .16) and (5. solve equations (5. (7) Different flat arrangements of a three-phase transmission line are shown in Fig. Fig. phase-to-phase spacing = D. (d) bundle-4 conductor with the four subconductors placed at the vertices of a square. (b) bundle-2 conductor comprisingtwosubconductors arranged in horizontal or vertical configuration.16) and (5. Each phase of the line has (a) single conductor. subconductor-to-subconductor spacing = S.13b.= 13.3kPa. Fig. The corona ductor height above ground > onset voltage is determined by the onset field Eo.17) iteratively to get V.7 kV (6)Repeatproblem(5) for gas pressures of 3 and 5 atmospheres and comment on the results obtained.13. 5. .6 kV.8 kV.1kVand Vb. Fig. Conductor radius = r. solution^ (a) For p = 3 x 101. 5.13 have already been expressed (Jha.5 kVat p = 5 atm It is quite clear that the increase of gas pressure results in increasing the corona onset voltage. Yo+ = 53. The difference is normally about 7% for practical line dimensions. For p = 5 x 101. tively to get V (b) According to the criteria in equations (5. = 54 kV. Yo+ = 34. 5. (corresponding to an avalanche size of lo8) = 34. 5.1 kVat p = 3 atm To-= 54.

0 0 0 . (c) using bundle-3 conductor arranged at vertices of an upright or inverted triangle. using bundle4 (d) using bundle-4 conductor arranged at vertices of a square. . (e) conductor arranged at vertices of a diamond-form square. (b) using bundle-2conductor arrangedhorizontallyorvertically. l 3 Three-phase transmission-line arrangements: (a) using singleconductor.

= EOr ln(D/r) (5. Corona onset voltage v0= 2 ~In ~ r (5.: Corona onset voltage V .*.*.9) Em = 3r In D 4 J m . Corona onset voltage Yo = 3E0rln S (5.*. Corona onset voltage V . = 4E0rIn (5.18a) .18~) v(1. Corona onset voltage V .18b) .18e) .18d) .*.17 . = 4E0rln (5.

32 X 39 = 342. the positive and negative corona onset voltages are calculated as: (a) V0 = 6. 5 m. = 12.16 X 39 = 474.(8) For the arrangements of problem (7). SOlUt~O~~ ~ccording to equation (5.2 X 39 241.3kV It is quite clear that the onset voltage increases with the increase of the number of subconductors per bundle.Use the empirical formula reported in Section 5.3.79 X 39 = 459.ak = 325.2kV (d) V0 = 11.24 kVp. Eo+ = 39 kvpeak/cm Using equations (5. SolUt~on~ According to the empirical formula E o 306(1 + 0.The radius of conductors r = 1cm. and phase-to-phase spacing D = 5 m.1sa) .18).8 kVpeak = 171kV (b) V0 = 8.1 kV (e) V . 6 = 1. subconductor-to-subconductor spacing S = 40 cm.48 kVpeak = 242. Calculate the corona onset voltage at standard temperature and pressure.& = 335.81 kVp. and 8.3/&) kVpeak/Cm where r is the conductor radius in cm. (9) The conductors of a three-phase transmission line are placed at the vertices of an upright triangle of sides 5 m. calculate the positive and negative corona onset voltages at natural temperature and pressure if the subconductor radius r = l cm. At standard temperature and pressure.66 m.

3 / m ) 0 . = mean geometric distance between the conductors =2/5~5~8. Atmospheric pressure and temperature are respectively equal to 75 cm Hg and 35" C .95(1 + 0 .66=6m The corona onset field Eo = 306(1 0.6). 9 2 X 0.10) = 0.4 kV(= 305. a= 3. 3 + V0 = 39 In600 = 249.3/&) At 6 = 1 at standard temperature and pressure. According to Peterson's formula (5.. 5.5 kVpeak = 176.37 kV(255. calculate corona power loss and the corona current when the line is operating at 50 Hz.25 1. Assume smoothness m1 and weather m2 coefficients are equal to 0..58 h 600 208. respectively.25 kV line-to-line) Yo = 275/255. = Eorln(D.92 x 75 = 0.95 = 32.5 kVline-to-line) (10) For the three-phase transmission line of problem (S).58 kVpeak/Cm V0 + 0.92~ Relative air density 6 = 273 t + where t is the temperature in "C and p is the pressure in cm Hg. 275 kV.95.4kVpeak = 147. Solution: 3.V .078 The corresponding value of the K factor (Fig.3/&)mlm2 0.95 273 35 + The corona onset field Eo = 306(1 = 30 X = 32./r) D.92 and 0. .05.

u. of a 525 kV line is strung 13m above ground.5 times rated voltage).km) x 3. When considering the effective radius of the corona envelope re. one may assume a smooth surface of the envelope.e. 2. A trial and error solution yields re = 5 cm. Calculate (a) the corona onset voltage and (b) the effective radius of the corona envelope. voltage = 2. = Eor ln(2hlr) = 426 kV = 426& = 737.1 mA/km (l l) A single conductor.5 x 525 = 1312. The corona envelope around the conductor is considered an extension to its radius.1 Corona powerloss Pc = 3'73 x 0*05j77&x Peq/r)2 kW/(cond. 13 l2. there is no corona present.175 cm radius.51Jzl& kVpeakperpbase = Eore 1n(2H/re) = 30(1 0 . which is higher than the onset voltage. Consider smoothness m1 and weather m2 factors are both equal to 0.8 kV (line-to-line) At 525 kV line-to-line operating voltage. (i. 3.u. Relative air-density factor = 1. so corona is present on the conductor. 3 / m r e ln(2600/re) + where re is the envelope radius in cm.5kV (line-to-line). This represents an increase in conductor radius r by about 57%.5p. at a voltage of 2.km) = 19.05 x 50 x (275 x - ( 600)2 = 6.6kW/km Corona current = Pc/Vph = 6.9.5 p. .73 x 0. (b) 2.53 kW/(cond..53 x 103/(275X lo3/&) A = 41. Solution: V .

Shamloul D. 1974a. Abdel-Salam M. Abdel-Salam M. 1986. New York: John Wiley & Sons. Berlin: Walter de Gruyter & Company. Berlin. 345 kV and Above. Abdel-Salam M. ProceedingsofIEEE-IAS Annual Conference. Ciao T. Mouton Publishers.Farghally M. Gary CH. Al-Hamouz Z . Houston. Abdel-Sattar S. Abdel-Salam M. IEEE Trans IA-21:35-40. IEEE Trans IA-34:301-309. Extra HighVoltage AC Transmission Engineering. 1992. 1982. Electrostatics 39:129-144. Part I. pp 1627-1 631. J Phys D: Appl Phys 27:2570-2579. Abdel-Sattar S. alam M. 1994. pp . Lefevre CA. Proc IEE-A 136:33-40. Castle GSP. Delhi-Jullundur: Dhanpat Rai & oceedings of 3rd International Symposium on . Proceedings of 4th International Symposium on Gaseous Dielectrics. Abdel-Salam M. Paper 53-05. . IEEE Trans IA-24:1025-1030. Abdel-Salam M. 1999. A Course in High Voltage Engineering. 1992. 2000 (to appear).Abdel~Salam M. Abdel-Aziz EZ. TX. 1984. 1994a. Shamloul D.Knoxville. IEEE Trans IA-22:80-85. 1977.ETZ-Archiv 99(5):271-275. Shamloul D. Abdel-Aziz EZ. Rashwan MM. IEEE paper 82WM-212-9.Stanek EK. Electrical Power Research Institute. 1989. 1985. 1986. 1969. . Proc TEE 120: 1574-1 575. Aboelsaad MM. Elmoursi AA. 1972.Current Problems in Electrophotography. Mufti AH. Abdel-Salam M. 12:143-156. IEEE Trans IA-26:384-392. 1994b. Abdel-Sattar S. 1967. J. Al-Hamouz Z . J Phys D: Appl Phys 25: 1551-1555. IEEE Trans EI-24:669-679. . Jha Rs. pp 5-15. IEEE Trans IA-20:1607-1612. Project IJHV. Abdel-Aziz EZ. Abdel-Salam M. CA: Electrical PowerResearch Institute. 1988. 1998. -Hamouz 2 gamudre RD. pp 435-436. IEEE Trans EI-27:352-361. IEEE Trans PAS-93:720-726. 1984. Abdel-Salam M. heno~ena in Ionized Gases. 1989. Clade JJ. 1990. 1978. . . Mufti AH. l-Salam M. TN. Abdel-Salam M. Elmoursi AA. 1979. 1973. 1997. Electric Power Sys Res 50. Horenstein MH.Milan. Palo Alto. Shafai L. IEEE Trans PAS-88:695-703. IEEE publication 31-C-44. Farghally M. Speck CE. Abdel-Salam M. J Phys D: Appl Phys 275407-817. IEEE Trans IA-35:380-386. Jordan J. 1992. Abdel-Salam M. Transmission Line ReferenceBook. COMPEL.Al-Hamouz Z . Abdel-Aziz EZ.

PlulleW. Arnesen A. Stark WB. 1987. RT. IEEE Trans PAS-96:945-954. Electrical Coronas-Their Basic Physical Mechanisms. GA. Shamloul D. 1974b. Loeb L. In: Watson W. Dubard J. . O’Brien M. Abdel-Salam M. Seelentag W. Atlanta. Fundamentals of Gaseous Ionization and Plasma Electronics. Abdel-Salam M. 1976. Lindsey C. 1989. Electrostatic imaging. 1977. Abdel-Salam M. ed. IEEE Medical Electronics Monographs 28-33. 1965. Richard TE. 1971. Egypt. IEEE Trans PAS-93:1693-1699. Rein A. 1972. CA: University of California Press.Khalifa M. El-Ragheb M. Assiut University. Waters. 1977. Ali Kh. NewYork: Wiley-Interscience. Radwan R. IEEE Trans PAS-96:88~895. Berkeley. Khalifa M. UK: Peter Peregrinus.Proceedings of IEEE Annual Meeting on Electrostatic Processes. Johansen I. Nasser E. Stevenage. Landham E. Assiut. 1979. PhD thesis. Proc IEE 119:717-723. IEEE paper A-76-418-4. Zeitoun A. Paper 8-2.

This Page Intentionally Left Blank .

the contacterosion they cause. Whenever a circuit breaker or a load-break switch is opened whilecarrying a current. Assiut. Therefore. the contacts move apart andthe contact area decreases rapidly until finally the contacts are physically separated. an arc strikes between its seperating contacts. Egypt Assiut University. and protection of such high-voltage equipment. Giza. and processes in the manufacture of pure metals and some electronic devices. operation. Applications of the electric arc and itsplasma include arc-discharge lamps. or cable would eventually involve an arc. I T When opening a switch or breaker. For a circular spot of contact with radius r. the contactresistance increases considerably while the flowing current becomeshighlyconcentrated. and the factors conducive to their extinction is essential for the proper design. machine. some furnaces. information aboutthe characteristics of arcs.Cairo University. the contact resistance 1 . Egypt Arc discharge isencountered in the everydayuseofpowerequipment. A persistent fault in a transformer. When the contact area decreases to a very small spot.

whenthe breaker or switch is closed under voltage. 6.equals p/2r. This occurs in vacuum and other types of circuit breaker. In the space between the surface of the cathode spot and the cloud of positive ions there is a high electric field (Fig. 6. and it burns in composites of sulfur and fluorine inSF6 circuit breakers.152. a balance is struck between power input and losses. An additional important source of electrons at the cathode is ionized metal vapor. electrons can be supplied by field emission from points of microroughness where the electricfieldishighly concentrated. the mediuminwhich it is burning. In air circuit breakers the arc burnsmainly in an atmosphereof air. The metal vapor filling the space between the parting contact spots would thus furnish the conducting medium for the circuit current to flow in the form of an arc. In a steady arc. If the contacts are made of pure copper. . the current density reaches about 3 x 106A. it burns mainly in hydrogen in oil circuit breakers. carbon. or molybdenum). the molten metal bridging the microscopic gap between the contact spots will soon be broken as the spots part further (Fig. The very shortarcthat ensues soon gets extinguished. Therefore.Also. tungsten.1). Under allthese conditions the arc is composed of three principal regions: the cathode and anode regions and the arc column (Fig. The current is also carried partially by positive ions drifting slowly to the cathode from the plasma of the arc column. no matter what the total arc length is. Through all three regions the current is carried by electrons and ions. p being the resistivityof the contact material.. If the electric circuit is suchthat no stable arc can exist. a significant cathode voltage drop builds up over the cathode region. With cathodes of low melting point. Thus. a spark will occur because the ~ontactsgetveryclose. Its magnitude and the widthof the region depend on the arc current. It burns mainly inan atmosphere ofmetal vapor in the caseofvacuumcircuit breakers. The corresponding power loss at such a microscopic spot would suffice for melting and even evaporating the hemisphere of metal at the contact spot within a period on the order of 1 ps.2).g. for a circular spot of radius 10pm and a current of10 A.cm"2. Electrons emitted from the cathode spot can be produced mainly by thermionic emission if the cathode is made of a high-melting-point metal (e.2). and in ~ercury-vapor lamps. 6. the contact resistance can exceed0. and the cathode material.

5 mm. [Courtesy of F.] . Steps: (a).~ e v e l o p m e nof t a molten metal bridge between separating ironcon(a) is tacts while carrying 40A. (c). (d). (b). Contact separation in step 0. Llewe~lyn-Jones (1957) and Oxford University Press.

6. They thus liberate more electrons from the gas and metal vapor in the arc column. however. 6.pressure p.2). Because of the different distributions of space charges at the cathode and anode. Many of their atoms and those of the metal vapor present will be ionized (Chapter 3). Space-charge distribution near the anode produces a nonlinear field in the area and an anode voltage drop (Fig. in the arc and its longitudinal voltage io The electrons from the arcplasma bombarding the anode spot anddelivering all their energies keepthe anode spot at a very hightemperature.2). and ionization potential K according to the well-known Saha relation . the gas molecules are mostly dissociated. covers a sufficient number of free paths forthe electrons leaving the cathode toreach the level of ionization by collision. the cathode drop takes place across a considerably shorter distance than the anode drop (Fig. The degree of thermal ionization of a gas depends on its temperature T. The cathode region.Arc Column Plasma Electrons and positive ions distribution. At an arc’s very high temperature. Positive ions are produced at the anode by thermal ionization of the gas and of any metal vapor near the anode spot. These ions drift slowly away from the anode into the arc plasma. on the orderof lo4 K.

Therefore. The magnitude of the constant A depends on the units used. PERCENT OF ARC RADIUS Figure 6.g. No doubt the extreme temperatures and high conductivities are confined to the core of the arc column.] dependent on the ambient . NZ. It has a significant electrical conductivity. Both decrease sharply at some radius beyondwhich there is no current conduction to speak of. The temperature distributions over the cross-sectionsof 80 A arcs burning in 02.3 Radial temperature distribution over the cross-section of an arc as gas. the arc plasma exerts no electrostatic field. exp-"evi 1°C kT where e is the electronic charge and k is Boltzmann's constant. N 2 . SF6) are insignificant up to about 5000K. [From Frind (1960). 19175). and SF6 are shown schematically in Figure 6. This effective radius of the arc column is a function of the arc current and the ambient gas and its pressure. Inside the column the densities of positive and negative charge carriers are equal on the average and are comparable to that of neutral gas molecules..3. The interaction between the arc column and its surrounding ambient takes the form of diffusion of charge barriers and heat transfer. Measurements have indicated that the conductivities ofseveralgases(e. H2. rise steadily to about 30 San-' at lo4 K.C2 p = A T25 y . and are about 80 San-"' at 2 x lo4K (Flurscheim.

heat conduction into the cathode volume. it can be described as a steady arc. p 2 where G is a geometrical factor. 6. The cathode spot receives both electrical and thermal power inputs. = GB& (6. This input is consumed as heat conduction. + Vi) Tc) (6. The sizes of the anode and cathode spots are functions of the arc current. The power input per unit length of the arc column is E(I. It also loses heat by radiation and convection. in oil circuit breakers. + . These two inputs are balanced by four components of power loss from the cathode spot: electron emission.3) The input to the anode spot isthe sum of the potential energy of the electrons falling through the anode drop. and the ambient gas pressure. E being the voltage gradient along the arc. metal evaporation.If the arc carries a DC current of constant or slowly varyingmagnitude and is not subjected to turbulence in its envjronment. The densities and mobilities of these charge carriers are functions of the arc column temperature. P 1 and p 2* P 1 =d i ( V .2). Let us now look at the power balance for each region of this arc. The components of heat dissipation depend on the ambient. the electrode material. is the cathode drop (Fig. Vithe gas ionization. and le and I i the components of the arc current carried by electrons and ions. and the factor a represents the fraction of the ions’ energy deliveredto the cathode spot and not to the arc plasma. and evaporation of metal from the anode spot. B the coefficient of heat transfer. The current is mostlyelectronic since ion mobilities are aboutthree orders of magnitude lower than that of electrons. li). and Tgand Tc are the temperature of the arc column near the cathode and that of the cathode spot. Here the input is purely electrical. respectively. 4 the current component delivered by ions. their kinetic energy delivered on impact.potential. radiation. and radiation from the cathode spot. and the heat input to the anode spot.2) where V . The arc column loses some of its charge carriers by diffusion from its surface. For example.

25 and 0. The arc column radius r and temperature 7' are such that the heat loss isminimal. on the other hand. its heat loss to the surroundings would increase. An approximate relation between the arc column voltage gradient E and its current I can be obtained as follows: . dissociating some ofits molecules to produce hydrogen and hydrocarbon gases. for a given arc current I . and radius. 19'75). with rz ranging between 0. and ambient. it was noted that. the V-I characteristics follow inverse relations of the form shown in Figure 6. expanding the gas bubble. electrode material. the power input maintains the ionization in the arc plasma. The relation between the arc voltage and current depends on the arc length. In air-blast breakers it is volumetric cooling. For arcs offixed length in atmospheric air. Along the arc column. Because of the different ionization potentials and thermal conductivities and other thermal parameters. whereas for a stationary arc cooling is effected at its column surface. The arc always adjusts itself so that the system ine~uilibrium has minimum entropy. The current density varies along the axis of the arc column. for the same current in the range 10 to 1000 A. It can beshown that there is an interdependence between the arc current. As is evident from the previous discussion. and thus E would increase. This in turn can lead to ejection of molten metal and vapor from the electrode spots. Thus dE = 0 and dr In other words. the arc column gradient E in hydrogen was about 10 times its value in oxygen when the arc was contained in a tube 2cm in diameter. If.its temperature would decrease and so would its conductivity. the arc is by no means a simple element.6. and compensates for the losses at the periphery and for any changes in the heat stored in the column. column tennperature. the cross-section were to increase.4. Thus for a given arc current the column voltage gradient E is also a minimum consistent with the power balance mentioned above. and dissociating and ionizing some of the gas molecules. This induces gas flow and static pressure gradients along the arc. its conductance would decreaseand the voltage gradient E would increase. The vacuum arc voltage was shown to depend principally on the cathode metal (Reece. if the arc column cross-section were to decrease. Experimental tests have shown that the arc column radius r is proportional to I". depending on the type of cooling for thearc.energy isconsumed in boiling someoil.

we get where C is a constant.] The electron and ion densities Ne and Ni and their respective mobilities ke and ki are functions of temperature. [From Rieder (1967). b = 22 V/cm c = 20 W. the constants of this relation have the following values: a = 17 V. .4) can be rewritten as Under steady conditions. and d = 180W / C ~ .7) For arc currents up to20 A and lengths E of about 5 cm. another function of temperature F2(T). Thus equation (6. Many experimental results on DC arcs in air fit relations between the total arc voltage V and current ' I of the form V =a -+bl + (c + d1)I-I (6. the input power is balanced by heat dissipation from the arc column surface. Thus Eliminating r and differentiating.l alif DC voltage-current characteristics for arcs of different lengths burning in air between copper electrodes.

The pressure caneasilybe calculated as being equal to 12/nr2. (Courtesy of E. We thus get the well-known V-Icharacteristic shown in Figure 6. The arc is influenced by two magnetic fields. As the arc current is reduced to a very small value.l n t e r s c i ~ n c e . the arc voltage increases with ambient pressure above atmospheric. The circumferential field produced by the arc’s current does exert a pressure to squeeze the arc column. a limit is reached when the arc input power is no longer sufficient to maintain the column temperature. ) . the arc becomes unstable and turns into a glow discharge. This is experienced in some gasdischarge lamps and tubes. The electron and ion densities in the arc plasma decrease so drastically that there is not enough conductivity to carrythe current through the arc. At the anode and cathode spots the arc radius ismuch smaller than anywhere along the column. Nasser and ~ i l e y . 1971).5 (Nasser.For the same current and length. and it is extinguished. its own field and that of its feeding circuit. Therefore. V I I I I l I I K- (Volt 1 80C 600 400 200 Typical voltage-current characteristic for thevarious discharges a in gaseous gap. and with a high enough source voltage. In a rarefied atmosphere of about 10 however. The gap is 50 cm between disk electrodes 2 cm in diameter.Neon is at a pressure of 130 Pa.

voltage. and temperature for a 50 Hz. 0 K .7 Time variationsof current. (b) medium frequency.6 Voltage-current characteristic of a DC arc witha superposed current modulation of amplitude 1.: (a) low frequency.in air between copper electrodes.10A arc.1 V I W@ 6. 3 mm long. (C) high frequency. .

the arc gap has to break down again under a sufficiently high voltage. Asthe current approaches zero. Such phenomena are usually exploited in the design of circuit breakers for accelerating arc extinction (Chapter 11). . After some time. 6.7). Of course. 6. as current-carry in^ circuits under their own magnetic fields tend to increase their self-inductance. 1972).. When the voltage across the arc is not high enough. a higher and higher voltage is neededto maintain the arc. a sudden rise in the magnitude of an otherwise steady arc current will have to be accompanied by an initial rise in the arc voltage so as to furnish the extra energy needed for building up the column ionization to the level corresponding to the new current magnitude. the arc voltage at the beginning of the half-cycle is considerably higher than that at its end. termed the thermal time constant 8 of the arc. Its thermal capacity prevents its temperature from varying at any significant fraction of such frequencies.l Near the peak value ofthe arc currentthe voltage necessary to maintain it is relatively low(Fig. The loop shape depends on the frequencyf of the current modulation. at extremely high frequencies the arc behaves like a linear resistance. Therefore.7). For the current to flowin the opposite direction in the followinghalf-cycle.dQ dt where H is the heat loss per second and Q is the heat stored in the thermal capacity. This can be visualized by thinking of the power balance for a unit length of the arc column. the arc will bow outward into a larger loop. the input power P = E 1 = H + . There. it will be unstable and will become extinguished even before the zero point of the sinusoid (Fig.the corresponding voltage variation will be such that the operating point will follow a loop (Fig. On the other hand. If the current variation is an AC modulation superposed on the DC current magnitude lo. the opposite will occur if the arc current suddenly drops.6). Also. 6..?. At extremely low frequencies the variation will almost follow the static characteristic. Because of the thermal capacity of the arc column.there is a high axialgradient of that pressure which can set up jets of plasma and metal vapor from these spots with velocities reaching 1 km/s ( et al. . the arc voltage settles down to a steady value according to the V-1 characteristic.

across the arc and the circuit (the arc is stable at the p point .There is also a period At of effectively zero current around the virtual zero point of the sinusoid (Fig. The arc column temperature also varies during the AC cycle. Arc as an element in a DC circuit: (a) arc in the circuit. the ratio V / I has a positive magnitude (the same as for a metal resistance). as in faulty power systems or when arc furnaces are included. 6. a stabilizing impedance must be included in serieswith the arcinbothACand DC circuits (Fig. This applies to both AC and DC circuits.(b) voltages but not at a). Electric circuits subjected to analysis sometimes contain arcs. and both consume power.8). According to DC arc characteristics such as the ones in Figure 6.7). Therefore. The arc is by no means a simple circuit element. Accurate analysis requires a truly representative circuit to take the place of the arc. the peak temperature lags behind the peak current because of the arc’s thermal capacity.4. However. the arc equivalent resistance varies with the current and dV/dI isnegative. 6.

If the current happens to swing below point a it will become extinguished. as will be described in Chapter l 1. Thus the voltage required to maintain it would increase rapidly. Such a circuit (Fig.7). the arc is driven rapidly by its thermal buoyancy and magnetic field to estend its length and to subdivide into a number of partial arcs. . (b) case of AC during the extinction period (Atin Figure 6.9 Equivalent circuitof the arc: (a) cases of DC and AC during the burning period(R.7). 6. the arc can be represented by a high resistance that increases with time. In the DC case the arc current is forciblybrought down to zero. its current I . During periods of effective current zero. PT1 The situation with DC is basically different from that under AC. it can easily be proved that stable operating conditions are represented by point p (Fig.9a) would apply for both DC and AC circuits during periods when the arc is burning. which can be represented approximately. which means that the arc can be represented by a fixed back voltage opposing the source (Fig. the voltage does not vary significantly over a wide current range. The time for arc interruption can be estimated given the I ure 6. To do that. and the current’s rate of change with timeI ’ . At (Fig. 6. The equivalent circuit also comprises a resistance and an inductance that arefunctions of the arc length E. 6. For arcs of very small length.9).8b). and € are variables).L. The nonlinear and irregularly time-varying characteristics of arcs produce current harmonics and voltage ripples in the feeding network. however.Looking at the arc’s V-I characteristic. The magnitudes of these parameters can be obtained from experimental tests. 6.

It should be noted that the circuit inductance acts to delay the arc interruption. having constant resistivity. Some time will be needed for the electrode spots and the gases in the postarc gap to cool down and for the electrons and ions of the arc plasma to recombine and/or diffuse. 6. formulated two differential equations based on rather different concepts of the physical nature of the arc column. Cassie assumed an arc column within which the temperature is fixed and uniform in space and time. These are abridged as follows.10. 1972).9. working independently. Note the overvoltage caused by the circuit inductance and the capacitance across the arc.electric circuit and breaker characte~stics. power loss. 1 32 ms - I Voltage and current variations during arc interruption. and a cross-sectional area variable with current and time.2. ue to its thermal capacity. The conductance of the gap drops during the first few microsecondsafter arcextinction. This requires a constant voltage in steady state given by V = Eo. where V is the arc voltage. On the other hand. The voltage and current oscillograms would look like those represented in Figure 6. and energy content. the conductance of the postarc gap cannot vanish instantly at current zero.. The behavior of the postarc gap was originally represented by Cassie and Mayr and later improved by Browne and others (Thomas et al.1 Cassie and Mayr Models Cassie (1932) and Mayr (1934). a resistance or capacitance across the arc would accelerate its interruption. and the gap's dielectric strength builds up in the subsequent tens of microseconds (Buttler and Whittaker. as it would draw an additional current through the circuit resista~ce. 1995). .

2.10) where 0 represents the ratio of a characteristic amount of stored energy to the power loss for both models and is called the arc time constant. A comparative study betweenCassie andMayr models has been reported (Nakanishi. with its steady-state arc voltage Eo. 6. 1948. The Mayrmodel. 1978) Mayr’s equation can be written in the form (Browne.4). He also assumed the arc behavior to begovernedby a characteristic energy quantity leading to a constant powerlossgivenby IV = No.2 Cassie-Mayr Composite Model It has been observed (Browne.. Urbanek. 6. 1975.e. Therefore. Cassie’s equation can be written in the form (Browne. It cannot actually be defined as a time constant in the rigid mathematical sense.9. Mayr’s modelrepresents the arc around the current zero and indicates how the postarc columnresistance continues to increase. This tends to corroborate the shape of the V-I charactristic of DC arcs (Fig. to use Mayr’s equation . 1978): (6. once we realize how complex the arc phenomena are. its VI = constant). where I is the arc current. Leeds et al. 1978). Thus. The constant 0 wasshown to depend on the magnitude of the arc current.fits best the voltage and current waveforms for the arc during the period prior to current zero. 1978). This is because during the time before current zero the energy loss is mainly caused by turbulence-enhanced thermal conduction (Browne. even during interruption of a large alternating current.1 Mayr assumed a cylindrical arc column of constant cross-sectional area within which the temperature varies with the arc’s radial dimension and withtime. Mayr’s equation becomes a better approximation only when the arc temperature falls to a level just above that at which electricalconductance due to thermal ionization practically disappears and the energy is transferred mainly by radial heat conduction (Browne. for a steady arc.Some further refinements for thismodelwere introduced by Rizk(1985) and others (Buttler and Whittaker.. 1992).10) (i. 197 1)that the Cassie model. its equivalent d(l/R)/dt = 0 in equation (6. describes bestthe characteristics of a low-current arc. such as existsduring the critical few microseconds just after current zero. 1972). with its constant loss No.

The value of the characteristic constant of Mayr’s model No can be shown to be given by (Browne. I = the rms value of the interrupted current in amperes. The model has shown especially good agreement with SF6 interrupter data. 1977) have validated this composite model by further comparisons between experimental data and computed results. the arc is modeled by Cassie’s differential equation again but with a new steady-state voltage constant Eo. as well as most of the models for arc interruption.12) Only two parameters. 1948) N o = EOId (6. 1977) proposed the use of Cassie’s equation again. and W = the angular frequency of the source waveform in radians/second.is valid onlyfor a brief period of time corresponding to the region of a few microsecondsaround current zero. 6.Browne (1978) and Frost (1 976.2. The transition time from Mayr’s to Cassie’s equation occurs when the arc resistance computed by Mayr’s equation reaches a maximum value R .alifa a alone to represent the whole interaction period of the arc may be an avoidable approximation. The exact time when the transition from Cassie to Mayr models occurs is relatively unimportant so long as it occurs near enough to current zero that the arc volt-ampere product issmall compared to the continuing powerloss from the arc The transition time is normally taken as the time when the arc resistance reaches the value given by (Browne. The ratio Eo/0determines the failure or the success of interruption. rowne (1978) and Frost (1976. . given by (Browne. The composite model of Cassie-Mayr. Eo and 0. 1978) (6. after Mayr’s equation. need to be estimated in order to implement the composite model. the composite model in the case of reignition is a Cassie-MayrCassie model. = the value of the arc resistance at aboutcurrent zero. Hence. 1978) R 0 E O =I d (6. to model arc reignition.9.13) To summarize. the composite model of Cassi follows: .11) wher R.3 Cassie-Mayr-Cassie CompositeModel ~dditionally. Thereafter.

given by equation (6.11). The effect of the shunt resistance on the arc behavior at successive current zeros was investigated. i. the air blast or gas flowin the arc interruption chamber of the circuit breaker is to push the ionized medium away from the breaker contacts.13). If the arc resistance computed by Mayr’s equation starts to decrease. Before current zero. 2. 1997) which reduces the current to be extinguished. the arc resistance at about current zero: &h = PRO (6.12). The shunt resistance changes its value from one half-cycle to the next half-cycle to simulate the continuously decreasing current that flows between the breaker contacts. The shunt resistance across the breaker contacts. power. Use Cassie’s differential equation (6. including arc resistance. etc. a satisfactory agreement was found between calculated and measured temporal variations of the arc voltage and current for the same short circuit current. useCassie’s different equation (6. denoted as Rsh. .2.. thus reducing the current to be extinguished. 3. Asiswellknown.10) to model the arc behavior with Mayr’s constant No.1. Use Mayr’s differential equation (6. The arc parameters were found to be inversely proportionalto the arc time constantand differaccording to the arcinterrupting medium. 6. the kind of arc chamber. and dielectric strength.4 ArcExtinction To study the arc extinction in a circuit breaker. Therefore.14) The multiplier .e. The temporal variation of the arc parameters of a circuit breaker. indicating resignation. with a new value for Cassie’s constant Eo given by equation (6. was studied (Ahmed.9) from the point the arc resistance has reached a maximum value.was taken as equal to p times Ro.9) to model the arc behavior from a few arc time constants 0 before current zero until the arc resistance reaches the value given by equation (6. 8changes from one half-cycle to the next and the rate of its changeisgovernedby the speedof the breakerin interrupting the arc. the larger is the current to be bypassed awayfrom the contacts.9. one needs detailed information about the thermal energy of the arc. 1997) followingthe Cassie-Mayr and Cassie-Mayr-Cassie models.confirming that arc extinction in SF6 gas is faster than in air for the same short-circuit current. the effectof the air blast or gas flow may be simulated by a shunt resistance across the breaker contacts (Ahmed. The smaller the value of the shunt resistance.

[From Rizk (1963). the initial conditions of the arc were considered to be nearly unchanged from half-cycle to next half-cycleaccording to the measurements The temporal variationof the arc characteristics during its extinction was studied by Ahmed (1997) following the Cassie-Mayr and Cassie-MayrCassie models. the total arc current drops rapidly and prematurely to zero (Fig.] . 6. As the power-frequency current through the breaker arc falls toward zero. The arc extinction following the Cassie-Mayr-Cassie model was found faster than that of the Cassie-Mayr model. It induces excessive overvoltages in the circuit. Such oscillations could be initiated by any sudden drop in the arc voltage due to any change in its conditions. 6. as discussed in Chapters 1l and 14. With the~gh-frequency component superposed on that at the power frequency.2.Note the superposed oscillations inthe current and its chopping before its natural zero. This phenomenon is called current chopping.5 CurrentChopping The circuit to be interrupted by the circuit breaker inevitably has distributed capacitance and inductance. the arccolumn diameter shrinks andexhibits a negative V 1 characteristic. .However.9.l1 Current and voltage variations duringAC circuit interruption by an air-blast circuit breaker.1 1). This effectively negative resistance of the arc would accentuate negatively damped oscillations of the local LC circuit at a frequency on the order of 10kHz.

the current will continue to flow for another half-cycle. The applied electric field will also accelerate electrons. It will be governed by the applied voltage exceeding the dielectric strength of the gap. Some droplets of molten metal may also be removed from the electrode spots by high differential gas pressures in the gap and by excessive electric fields. The current wave shape also has an effect on the rate of arc erosion (Fig. giving them energies that would probably be high enough forionization by collision. reignition is governedby the energy balance.2. greater latent heats of evaporation. after some microseconds the cloud of positive ions will be so diffuse that its influence on the electric field in the postarc gap will become much less pronounced. During asubsequent period arc reignition will be dielectric rather than thermal. the circuit breaker or switch is considered to have opened the circuit successfully. 6. the rate of contact erosion is lower for contact metals with high melting points. If it does. Arc reignition is decided by a race against time between the voltage appearing across the postarc gap and the deionization of its gases. Ionization will continue to be partly thermal because of the residual heat in the new cathode spot and in the gases in the gap. a dense cloud of positive ions remains near the new cathode.6. and the dielectric strength of the postarc gap will be relatively low. If this exceeds the heat loss. If it does not. If the applied voltage per unit length of the postarcgap is E’. During thefirst few microsecondsafter the virtual current zero. as explained above. However. is the deciding factor. This cloud willseverely distort the field near the cathode. For otherwise similar conditions. and higher thermal conductivities. As the electrodes have reversed polarity. Thus the breakdown voltage of the gap rises fast initially. but afterward slowly approaches its ultimate value. . as explained in Chapter 3. the input power perunit length of the gap is E ‘i. Tests and analyses of erosion phenomena have revealed thatthe energy of the arc. Erosion of the contacts by arcing is caused partly by evaporation of metal from electrode spots during the arc. the arc will reignite.9.12).6 Arc ~ ~ i ~ n i ~ i o n The voltage that appears across the postarc gap after itsextinction may or may not reignite it. rather than its charge. and this causes a current i to flow as a result of the gap’s residual conductance.

(c) (f) cathode of W-Cu sintered mixture. (e) anode and The applications of electric arcs include arc-discharge lamps. particularly when electric power is available economically.Current peak ~ I u (A) e Rates of arc erosion from electrodes of different materials. Arc furnaces are widely used in metal-producing plants. under vacuum (Shankar. Glow discharge has been used in a plasma spray for applying a metallic protective coating to important machine parts. Arcs are also used in the production of high-quality metal strips and for ion implantation on silicon wafers in the production of very large scale integrated circuits. 1987). An effective method employs a plasma beam of high-energy positive oxygen ions (Os) (Dettmer. Also. For smelting aluminum. DC arc charge. . arc furnaces. for instance. arc length. such as turbine blades. arc welders and cutters. 73 mC.furnaces fed from three-phase AC are used for melting and purifying metals. l mm: (a) anode and (b) cathode of molybdenum. Modern techniques for the production of very large scale integrated circuits involve isolating islands of p-type silicon with submicrometer layers of SiOz. 1987). This method was proved superior to the con~entional method involving successive rolling and heat treatments (Millar. anode and (d) cathode of steel. DC arc furnaces are preferred. 1981). It has also beenused more recently for excessive heating of strips of metal powder in a rarefied atmosphere to produce high-quality stainless steelribbons. and plasma torches.

Barrault M. MSc thesis. 1978. (ii) an AC source. 1948. explain why a drooping characteristic of the supply voltageisessential for maintaining a steady arcand describehow this characteristic is obtained in case of (i) a DC source. . electric welders. (2) Sketch the voltage distribution along a stable arc in a DC circuit and comment how the arc is sustained in the cathode and anode regions as well as in the arc column. Arc rupture and circuit severity: A new theory. 1976. Satyanarayana P. Dettmer R. (5) Make a comparative study between the different models of an AC arc: Cassie. 1975. 1987. Stevenage. Browne Jr TE. IEEE Trans PAS-97:478-484. TX. Why iscurrent chopping not a serious problem with vacuum circuit breakers? (4) In arc welding. Conference International des Grands Reseaux Electriques a Haute Tension. (7) Sketch and comment on the waveform of voltageand current in a short arc (a) between electrodes of high-meltingpoint. IEEE International Conference on Plasma Science. 1972. Power Circuit Breakers-Theory and Design. 1972. Proc IEE 119:1295-1300. AIEE Trans Pt I11 67-I:141-153. Whittaker D. (3) Explain how to reduce the current level at which chopping takes place in circuit breakers. Dynamic circuit breaker test analysis and its use in the generalized specification of short line fault performance. and composite Cassie-MayrCassie models. Edels H. 2Angew Phys 42515. Frind G. Frost LS. Mackburn T. Browne Jr TE. Buttler T. IEE conference publication 90. Egypt. Mayr. composite Cassie-Mayr. IEEE Conference Record-Abstract. Paris. Ahmed MAM. 1960. UK: Peter Peregrinus. (b) between electrodes of low melting point. France. Austin. Cassie AM. Flurscheim CH (ed). (8) Describe the principle of magnetic blow-out in arcs. Restriking voltage as influenced by arc modeling in gas blast circuit breakers. IEE Electron Power 33(4):273-277. (6) Compare the problem of interrupting 1000A in a 120V AC circuit with that of interrupting 100 A in a 1200V AC circuit. 1932. pp 221-223. and fuses. Assiut University. circuit breakers.(1) Explain how the arc is initiated in lightning flashover of transmissionline insulators. 1997.

198 1. r IEE Power Eng J 1(5):257-265. Mayr 0. 1975. Stevenage. Leeds WM. Germany: Friedr Vieweg & Sohn. Reece M. New "York: Wiley-Interscience. J Metals 33(10): 13-20.. 197 1. India: Khanna. ed. Sweden. Goteborg. 1992. Koenig D. IEEE Proc 59:502-508. ~lewellyn-JonesF. Braunschweig. Fundamentals of Gaseous Ionization and Plasma Electronics. Browne Jr TE. Beitrag zur Theorie der statischen und der dynamischenLichtbogens. Christopoulos C. 1985. ~ i l l a R. 1957. Pereira FT. Dynamic arc analysis of short line fault tests for circuit breaker specification. Strom AP. Chalmers Technical University. IEEE Trans PAS-104:948-955. 1934. 1971. Archiv Elektrotech (Berlin) 37:588-608. Oxford: Oxford University Press. IEEE Trans. . 1992. UK: Peter Peregrinus. In: Flurscheim CH. 1967. Shankar S. Dardi L. Thomas DWP. Power Circuit Breakers-Theory and Design. New York: Marcel Dekker. Rizk F. 1963. Rizk F. Plasma und Lichtbogen. AIEE Trans 76-III:906-909. Howe AT. Ch 2. Power Deliv 10:1829-1835.Frost LS. Rieder W. Mexico. Rao S. Urbanek J. The Physics of Electrical Contacts. 1977. Delhi-6. Switching Phenomena in High Voltage Circuit Breakers. Nasser E. 1987.1995. PhD thesis. 1975. IEEE Power Engineering Society Summer Meeting. Switchgear and Protection. MexicoCity. Nakanishi E.

and synthetic esters. Molecules of esters contain atoms of carbon. Giza. Synthetic liquids with a wide spectrum of properties started tobe developed by the petrochemical industry in about 1960. Organic oils started in use late in the nineteenth century. Natural esters are produced by the chemical reaction between a vegetable acid and an alcohol. and circuit breakers. hydrogen. The reaction is helped by a catalyst such as sulfuric acid. transformers. Egypt Since the turn of the century. Insulating liquids are broadly classified as organic. They include synthetic hydrocarbons. mineral or synthetic. about 20% (Breuer and Hegemann. This group of liquids includes vegetable oils. 7 . 1987). silicones. oils have been in use for insulating cables.Cairo University. rosin oils. whereas mineral oils were introduced in about 1910 with the development of petroleum refineries. An inert water entrainer is also included to help in removing the water formed in the reaction. halogenated hydrocarbons. and a considerable number of oxygen atoms. and esters.

aromatic hydrocarbons available as byproducts in some petrochemical industries. Their properties can. be tailored to match each specific application. we now have several synthetic liquids covering a wide range of properties. 1987.hosphate esters are manufactured from such raw material as coal tar “phenol” in a chemical reaction with phosphorus oxychloride. and the aromatics by the formula C.1 Askareis The synthetic oils commercially known as askarels are manufactured by chlorinating poly~hlorobiph~nyls (PCBs). the distillation and purification processes. including non~ammability.3. Thanks to intensive research and developments in the petrochemical industries. 1987). A s naphthenic crudes represent only 5 to 10% of the world’s total production. Synthetic liquids have proved to be the answer.+2. Chan. aromatic... For example. naphthenic. Marinho et al. in a sense.the naphthenes by the formula C. their poor oxidation stability limits their use to special applications. which.H2.2. The paraffin base is characterized by the chemical formula CnH2. is inevitably declining. With the inevitable depletion of mineral reserves in the world becoming more of a concern. olefins. 1987. alternative liquids have been explored by research laboratories. Petroleum is known to comprise a variety of molecular species. several attempts have been made recently to use organic liquids in transformers and other applications where mineral oil has been unparalleled and where synthetic oils were being introduced during the last 20 years (~ankaralingam and Krishnaswamy. or intermediate group of molecules. Available are askarels. silicon oils. because of its good gas-absorbing properties. phosphate esters. . 7. hexachloro- . . The number n of monomers per molecule varies over the range of tens to a few hundreds depending on the type of crude. It is therefore broadly classified as of a paraffinic. thus: phenol + phosphorus oxychloride ” + triphenol phosphate + hydrochloric acid osphate esters have very good fire resistance compared to mineral ever. and its subsequent treatment and life in service. and other liquids. Oil of the naphthenic group has traditionally been favored for impregnating paper to be used in insulating H V and EHV cables.

Sierato and Rungis. A s an example.2. The following properties relate to routine and/or acceptance testing of dielectric (insulating) oils (Claiborne and Pearce.H Cl- c / / \ c " . a molecule of poly(dimethy1 siloxane) is represented in Figure 7. biphenyl isproduced from benzene and has an atomic arrangement of molecules. I CH3 1 ure 7. 1989.1 Example of the atomic arrangement in a molecule of hexachlorobiphenyl (askarel).2.2.2 Example of the atomic arrangement in a molecule of poly(dimethy1 siloxane) (silicon oil).3. such as the one shown in Figure 7.2 SiliconOils The molecules of silicon oils contain silicon and oxygen atoms in addition to their many carbon and hydrogen atoms. with the number yz varying from 10 to over 1000. in recent years we have witnessed the development of organofluoric liquids of compositions such as (C4F9)3N and ( C 4 F g ) 2 0 . . Typically.3 Other Synthetic Liquids In addition to the phosphate esters mentioned above. n is about 50 for silicon oils used in transformers.1.3. 7. 7. c i ' C c =c c- c1 H / c \ C l C ure 7. These have very high chemical stability. ncould 10 be to 1000. C HC 3H 13 C~-Si-tO-Si+O 1 : i t I t cW 3 -Si l'. 1995).

Z~ter~aci~Z tension. it may impede charge generation in the oil. and solvents. The test is normally accepted for both vendor qualification and delivery acceptance purposes. being most pronounced for aromatic and least for paraffinic hydrocarbons. oar ~ o i n is t a measure of the ability of the oil to flow at low temperature.5. varnishes. This makes the test a useful screening method for new oils exposed in transport to soaps. ssing t ~ n ~ e n c is y evaluated in new oils to satisfy specification requirements. A high value of interfacial tension for new oil.cases a good oil oxidation stability may be achieved by a suitable content of the naturally occurring aromatics. but it is not usually used as a prime indicator of oil characteristics. Changes in viscosity with temperature depend on theoil constitution. new oils the ma~imum acceptable color value (color number) is 0. It is an important parameter that always has to be considered when low ambient temperatures of working equipment are expected. acids. ~i~a~ion st~~ In iZ many i t y . not less than 0. However. Specific gravity is important when there is a concern about water in oilwhichmayfreeze and rise to float ontop of the oil.Zor and a ~ ~ e a r f f n cFor e . Oils with a moisture content below 10ppm are considered to be dry and otherwise require drying before use. . However. a nitrogen and sulphur compound-free oil is required. ens it^ is not very significant in d e t e ~ i n i n gthe quality of insulatingoils but maybeuseful for type identification and evaluation of an oil's suitability for use. tercontent increases electric conductivity and dissipation factor and worsens electric strength. ~e~tralization n u ~ ~ e r ( a c iis ~a i tmeasure y~ of the trace amount of acidic or alkaline c o n t a ~ n a n t s in the oil. sh ~ o i n is t an indication of the flammability of oil. discharges as well as high thermal stresses and disruptive discharge conditions. indicates the absence of undesirable polar contaminants. and a bright and clear appearance is required. Zn~i~itor content is a parameter used for oil vendor qualification. Some standards recommend the measu~eme~t of the total gas A dissolved gas analysis can identify low energy content of new oil.040N/m. ~ i s c o ~is ~important ty for the dissipation of heat.

The electric strength of oil flowing in electrically stressed regions is a function of flow rate. but italso reflects the oil compositionand canbe lowered.e at powerfrequencyis the most often parameter describing the oil’s function as an insulant. the magnitude of discharges.5 mm gap when measured wtih discelectrodes. such as inception voltage. It can be used to determine the percentage of the aromatic content and the hydrocarbon composition proportions. In general. electric field. osi~io ~~ ~ ~ Z y Slight s~s. still remainsa matter of debate. as well as the presence of inhibitors. changes in the proportions of any of the oil compounds may affect oil properties. . i ~ a c~o t ~ ~affects e ~ t many oil properties. The term “stability to electrical stress” has been used to relate to the gassing properties of oils under discharge conditions. the significanceof partial discharge parameters.whenthe aromatic hydrocarbon content is increased. Dry and cleanoils exhibit highbreakdown voltages. and the dynamics of discharge characteristics in insulating oils. describes the propertyof an oil to become charged when in relative motion against a solid surface. rges. Infrared (IR) spectroscopy has proved useful over the years for the analysis of insulating oils. strength is strongly dependent on the amount of contamination in the oil and on the oil constituents.The preparation of the oil sample may greatly influence the breakdown characteristics of the oil. High resolution mass spectrometry provides a detailed analysis of hydrocarbons and detects low levels of oxygen and nitrogen compounds. is rarely considered in the various oil test programs. Most of the breakdown voltage specifications require a mini~um of 30 kV with a 2. It is sensitive to the amount of specific hydrocarbons andincreases considerably for oils with high aromatic content. which can be reduced dramatically when solid particles and dissolved water are present.This property is sensitiveto bothpolarity and electrode geometry. ~ ~ cis tanoimportant ~ parameter describing the oil’s function as a dielectric. The dissipation factor of an oil at power frequency provides the same information as the oil resistivity and is affected by the same factors and test conditions. The degree of frictional electrification determines whether or not this property can lead to electrical discharges and dielectric breakdown in a transformer. tem- .

I I . I. an olefin. 1982a).(3) silicon oil for transformers. (4) trixylenyl phosphate ester. except at excessive temperatures. and chlorinated hydrocarbon-are depicted in Figure 7. In the oil velocity range of 5 cm/s to 200 cm/s. it is noted that the viscosity of hexachlorobiphenyl is considerably higher than those of other synthetic and mineral oils. . 11. but it decreases at higher velocities. Ill: mineral oils graded by theInternational Electrotechnical Commission.3.3 Variation of viscosity with temperature for (1) low-viscosity polybutene (olefin).perature.3. The variations in the viscosities of some typical mineral oils-a silicon oil. Temperature re 7. (2) hexachlorobiphenyl (askarel). and 111) ofmineraloil prescribed by the International Electrotechnical Commission for use in transformers and similar equipment (IEC. In the figure are shown three grades (I. the mean breakdown strength may increase. andparticulate contaminants. The viscosities of silicon oil and olefin are distinctly less sensitive to temperature variation. an ester. In Figure 7. moisture content.

In liquids such as askarels. Unless extremecare is taken in their handling. 1987) which. water. which act as a catalyst. Oxidation is accelerated in the presence of metals such as copper.1YO. The wax deposits in the form of sludge.Some of the oxidation products inhibit the reaction.In applications where the fire hazard is a concern.4. particularly the dielectric strength. and wax. becoming larger than 10 pm and settling to the bottom of the container. Esters are preferred because of their low toxicity and the fact that they are biodegradable. about 50ppm. Some synthetic chemicals have recently been developed (Cookson. organofluoric liquids are stable up to temperatures as high as 500°C and are ideal for cooling electronic equipment of the sealed-off types that have large power ratings. inhibit the oxidation reaction and thus tend to extend the oil’s life in service. Water can be absorbed by the oil from the ambient. Water is molecularly soluble in small quantities in mineral oils. Because of their low boiling points. which thus stabilizes with time.4. The chemical reactions involved may produce lighter molecules (scission) or heavier molecules (polymerization). where the solubility exceeds 600ppm. The water-absorbing property of phosphate esters presents a serious problem for their users. mineral oils are usually replaced by askarels or phosphateesters because of their remarkable fire resistance. or can be produced by oxidation processes of the oil itself. Under thermal and electrical stresses oils become oxidized in the presence of oxygen. Examples of such reactions are given in Figure7. Its solubility depends on the molecular composition and on temperature. water would collect at the surface. acids. If the water content of the mineral oil exceeds about 400 ppm the droplets tend to flocculate. which have specific gravities exceeding unity. depending on the type of oil and the reaction conditions. Above the saturation level extra water becomes emulsifiedin the form of finely dispersed droplets of diameter 1 to 10 pm suspended in the oil. Water is more soluble in silicon oils and much more in esters. with deleterious effects on their properties. Also. Among the products of oxidation are hydrogen. The gas can beeither in contact with the oil or actually dissolved in it. they are used in evaporative cooling of power equipment. their water content can reach levels as high as 0. when added in small quantities to the oil. as explained in Section 7. . particularly phosphate esters.

xamples of chemical reactions involved in the oxidation of mineral oils: (a) p o l y ~ e r tion; i ~ (b) formation of water.

oils, natural esters, askarels, and some olefins xidation, olefins produce acids and ~olymerize, hand, silicon oils at temperatures below 150" C , however, they oxidize, forming water, carbon oxides, and acids. Traces of acid in the oil could be produced by oxidation reactions. cids could result from the refining process or be resent not be completely eliminated during distillation result of their hydrolysis and thermal decomposi may chemically attack the solid i~sulation and/or enamel or paint of the immersed e ~ u i ~ m e.n t

lectrical dischar~e in oil- immerse^ e ~ u i p ~ ecan n t be either a com~lete or a partial brea~down (i.e., a silent discharge). In the case of silent discharges there is usually a gas bubble, with enhancement of applied electric fields at undaries. Under high electric fields some oil molecules at the surface of ciated, releasing hydrogen, drogen gas produced by the di arge would be chemichemical reactions enhanced gas discharges may on, with hydrogen, oxygen, nitroge~accepted in some ~nsaturated molecules of aromatic oils. ch. degas sin^ action

would no doubt help to maintain the stability of the oil and extend its life under silent electrical discharges. The following is an example: electrical discharge he activity of the chemical reaction and its products depend heavily on the type of oil, the gases present, the temperature and pressure, and thepresence of catalysts. In the case of complete,,breakdown of an oil gapand of immersed solid insulation, an arc is involved. In such cases gases such as acetylene, ethylene, and ethane get evolved, resulting from the chemical decomposition of some oil molecules. Also, carbon monoxide results from decomposition of the solid insulation. Under severe electrical or thermal stresses, askarels evolve acidic gases, mainly hydrogen chloride. The hydrochloric acid thus formed causes severe electrolytic corrosion of insulation and metals immersed in the askarel. Also, as is known, askarels are nonbiodegradable and are thus ecologically sters do not suffer from these disadvantages.

Electrical properties include the electrical conductivity of the oil, its permittivity, dissipation factor, and dielectric strength.

It has been noted that when liquids known to be insulants are subjected to direct voltage, they actually conduct a current, however small. Even a simple insulating liquid such as hexane conducts a current that varies with the electric stress applied. It has been suggested that under low electric the conduction is due mainly to positive and negative ions belongi dissociated molecules. Dissociation and recombination are in dynamic equilibrium, thus: dissociation A+ + recombination However, when an electric field exceeding l.kV/cm is applied, the rate of dissociation exceeds that of recombination and increases at higher fields (Nelson and Lee, 198'7). This state of affairs is basically different from the case of conduction through gases, where below the corona onset level the number of charge carriers is limited and independent of the applied field strength (Chapter 4). This explains why the saturation current is constant in

gas gaps, whereas it increases in oil gaps with applied voltage (Cross and Jaksts, 1987). At still higher fields the current grows at a rapidly increasing rate. This growth is accounted for bothby field emission ofelectrons from the cathode and field-aided dissociation of liquid molecules. As the ions drift through the liquid, they gain energy from the field and lose some of it in collisions with the liquid molecules. The energy thus gained by the liquid molecules accounts for extra vibrations. The vibrations of C-H and C-C bonds among the carbon andhydrogen atoms result in some of them being broken, thus producing extra ions. Under impulse voltages conduction currents reach much higher values because the choking effectof space charges accumulating at the electrodes is insignificant compared to the DC case, as in gas discharges. In liquids of commercialpurity, additional conduction currents will be caused by the impurities. These may include a wide variety of particles: droplets of water and acids, resins, and cellulose fibers. Their different conductivities, permittivities, and affinities for ionization will have widely varying effects on conduction through the insulating liquid. Conductivity rises sharply with temperature as a result of both the increased dissociation of the liquid molecules and its decreasing viscosity. At a temperature T the number of n of dissociated molecules per unit volume is related to the total number per unit volume N in terms of W , the dissociation energy, as
n = N exp(- W/lzT)

(7.1)

k being Boltzmann’s constant. Thus the conductivity of the liquid due to dissociated molecules is

Here e is the electronic charge and p is the ion mobility. It is evident that the conductivity depends exponentially on temperature, as observed experimentally (Fig. 7.5). Under AC the liquid conductivity represents a considerable part of its dielectric losses. The other part of the loss under AC is due to hysteresis in the polarization of the liquid molecules. The losses are expressed as a dissipation factor(tan S) where 6 is the loss angle. The dissipation factor increases only slightly with applied electric stress at moderate levels. Under very high stresses, however,the dissipation factor increases considerably with stress, at an ever-increasing rate, until the liquid breaks down (Denat et al., 1983).

Temperature

igure 7.5 Electrical conductivities of some insulating liquids as functions of temperature: (a) PCB (askarel), (b) commercial transformer oil, (c) purified transformer oil,(d) highly purified transformer oil.

The dielectric constants (i.e., relative permittivities) of insulating liquids range from about 2.2 for mineral oils to about 5.3 for askarels. Askarels have therefore been favored for applications in capacitors. The dielectric constants of mineral and most synthetic oils vary only slightly with temperature.

Unlike the caseofgases, there is no single theory that hasbeen unanimously accepted for the breakdown of insulating liquids. There are two main reasons for this. One is that the liquid phase is much less amenable to theoretical analysis than the gas phase. Second, insulating liquids are almost never absolutely pure, if not becauseof the ingress of contaminants, then due to traces left in the liquid by the processes of manufacture and puri~cation.Each of these ingredients plays a role in the breakdown of the bulk of the liquid. Thus it is an accepted fact that the breakdown strength of a sample of insulating oildepends on its impurity content rather than its molecular composition. The three principal theories that havebeenproposed for the breakdown of insulating liquids are (a) the electronic breakdown theory, (b) the bubble theory, and (c) the suspended particle theory.

ccording to this theory, electrons are emitted from the cathode (Kuffel and 1, 1984). On their way to the anode, the electrons collide withatoms of molecules.If enough energyis transferred during such collisions, some electrons are knocked off their atoms and drift with the original electrons toward the anode. Thus electron avalanches such as those occurring in gas discharges develop in the liquid and ultimately lead to breakvidence supporting this theory, even under fast impulse voltages, is rather scarce.

Leakage currents arenot distributed evenlyover electrode surfaces. As directed by the applied electric field, they tend to concentrate at points of ~icroroughness where the field lines converge(Fig. 7.6a). Near such points, the highly concentrated leakage currents, with corresponding Joule heating of a microscopic volume of liquid, will cause a rapid rise in temperature to high above the boiling point. ~ i t very h limited heat convection and conduction, such a temperature rise could occur over short periods on the order of 1ps. Thus a bubble of vapor forms at a point of microroughness on the surface (Fig. 7.6a). ree other alternatives were proposed to account for formation of obejnikov et al., 1983): release of occluded gases from ode surface layers; cavitation caused by mechanical strain of the liquid under the highly concentrated electric field, with corresponding electrostrictive pressure differential; and electrochemical dissociation of some liquid molecules, with the release of gases. F u r t h e ~ o r ein , the case ofoil-impregnated paper, as in transformer windings, cables,and capacitors, repeated expansion and contraction under cyclic loads would cause cavitation and form gas bubbles. Under each alternative, the concentrated electric fieldat points of microroughness on electrode surfaces or somewhere along the gap would play a basic role in the production of gas bubbles. Tests on oil gaps under high-voltage pulses of nanosecond duration could provide the data on development of gas bubblesat the electrodes and their initiati~g b akdown (~orobejnikov et al., 1983; Lesaint and Tobaz~on,1987). cause of the difference between the pe~ittivitiesof the gas and the liquid, the electric field insidethe bubble would be considerablyhigher than that in the liquid. When the bubble reaches a critical volume, and under high enough fields, the gas in the bubble would become ionized and electron avalanches would develop into streamers. The space charges formed at the ends of the bubble would act to ilefomit under the electric fieldby ~ o u l o forces. ~ b Also, the electrons and ions accelerated by

branch~d

~ t r ~ awith m e ruitimat~ s break~own.

the field inside the bubble would impinge against areas of'its walls, helpin to decompose some of the liquid molecules and toevaporate others. bubble growslongitudinally (Fig.7.6a), which would lead eventually to complete brea~down of the entire liquid gap (Fig. 7.6b). The energy su~plied to the gas b ~ b b l e is C O ~ S U additional liquid (on the order of lo6Jim3), ionizin bubble (105-108 J/m3), and decomposing some oil simpler compoun~s (1o7j/m3). he speed of'expa measured in the range lOOm/s to l0 km/s (Felici, 1987).

Negative streamers start from sharppointson the cathode. Their growth is aided by electrons bombarding the gas-oil interface at the streamer’s tip. The positive streamer propagates from the anode and electrons fall into its tip from the oil molecules located ahead of it, which get dissociated by the local intense electric field (Chadband and SadeghzadehAraghi, 1987). With the growth of the streamer, its space-charge field increases exponentially. It may reach theorder of lOmV/cm,exceeding greatly that of the applied Laplacian field at sites away from the electrodes, which explains lateral branching of these streamers (Fig. 7.6b). The bubble theory is supported by the observed dependence of breakdown strength on the applied static pressure. If the liquid is degassed, its dielectric strength becomes less dependent on the static pressure. Also, gasabsorbing additives cause a rise in the breakdown strength.

The particle could well be a fiber, probably soaked with moisture, or it may be even a droplet of water. Under an applied electric field, and because their dielectric constant is much higher than that of oil gro, fiber particles get polarized and move along converging fields (Fig. 7.7). Assuming hemispherical tips for the fiber particles, with radius r, the charge ztq at either of its ends would be
k q = fXr2(Erf E,,)E~E

(7.3)

ure 7.7 Steps of collection of fiber or other particles in suspension toeventually bridge the gap. oil They becomepolarized and directed by the field.

where EO is the permittivity of free space and E is the field strength at either end of the fiber, assumedequal. If the field in the oil gap is nonuniform, as is usually the case, the resultant force will move the fiber (Fig. 7.7). For a short fiber of length equal to radius r, and with e,f > >E ~ the ~ driving , force will be

A s this particle moves with velocity v in the oil of viscosity U, there will bean impeding force with a magnitude of 67trqv. When a particle reaches either electrode, its outward tip will act as anextension to the electrode and attract more fibers (Fig. 7.7). It is noted from equation (7.4) that the driving force on the fiber increases in relation to its size and with the degree of n o n u n i f o ~ i t yof the field. Thus in a sphere gap they will migrate toward the higher field at its axis and gradually line up and bridge the gap (Fig. 7.8) (Kind, 1978). Such a relatively conducting path would in effect short-circuit the gap, resulting in breakdown. Figure 7.9 shows how a water drop gets distorted under the electric field in an oil gap and ultimately leads to its breakdown.

ure 7.8 Fiber particles bridging a gap under high voltage. [Courtesy of D. Kind (1978) and F. Vieweg &Sohn.]

Water drop being gradually elongated byan electric field and ultirnately causing breakdown. The wateris evaporated by the discharge.

The evidence in support of this theory includes the increased time required to reach breakdown of the liquid with increased viscosity. ~vidently, this phenomeno~ could occur under DC and power-frequency AC, but not under high-frequency or fast impulse voltages.
lali

The study of pre-breakdown and breakdown phenomena in liquid dielectrics has been improvedby the use of fast digitizing oscilloscopes for voltage and current measurements, fast optical detecting systems for measurements of density gradients using shadowgraph and Schlieren photography coupled with an image converter camera, electric field distributions using electric field-induced birefringence of the Kerr effect, and optical spectroscopy of emitted light (Beroual et al., 1998). It is therefore possible to follow the differe~t stages leading to breakdown and to establish that the breakdown of liquids isgenerallyprecededbysome events called “streamers.” The optical refractive index of streamers is different from that of the surrounding liquid and the structure is similar to that observed in the breakdown of gas and solid insulation (Saker and Atten, 1996). Streamers in dielectric oils are characterized by different structures according to the experimental conditions. They produce typical shapes of current transients or emitted light signals. They are accompanied by shock waves and their conductivity depends on the mechanisms involved in their propagation. The streamer stops when the electric field become producing a string of microbubbles that dissolve in the liquid. measurements of the electric fieldfound thatstreamers were highly conducting, with a voltage drop across the streamers that was less than 10% of the total voltage across the electrodes (Jayaram, 1996). Streamers are generally classified as slow and “bushy” for s t r e a ~ e r s emanating from the negative electrode, or fast and “filamentary’’ for streamers emanating from the positive electrode. Positive streamers are often about ten times faster than negative streamers, although ~ r a n s f o ~ oil e r is

3

an exception, with positive and negative streamer velocities being in the same range. Streamers can be classifiedin different modes depending on their velocity and the polarity of the voltage. Each propagation mode corresponds to a given inception electric field. Therefore, the electrode geometry and the amplitude of the applied voltage have a significant influence on the structure, the velocity, and the mode of propagation of the streamers. However, such a classification versus the polarity is inconclusive when considering liquids of specific molecular structure or containing selective additives. The characteristics of streamers depend on the chemical composition and physical properties of the liquid (pure, or containing a small concentration of selected additives); pressure and temperature; the electrode geometry; the voltage magnitude, polarity, and shape; and contaminants of air, moisture, particles, and other trace impurities. Figure 7.10 depicts the propagation of positive streamers in a nonuniform oil gap (Torshin, 1995). First, positive primary streamers are formed, with intensified branching at about 2 MV/cm. Streaqers then propagate at 2-3 km/s. Under higher voltages, a positive secondary streamer is formed as an extension of the primary structure beyond 12 MV/cm field level. Their velocitiesmayreach 32 km/s. At the streamer front, shock waveswere recorded.
7.7

The dielectric strength of insulating liquids has been observed to depend on the liquid temperature, applied static pressure, sizeof the gap, electrode material and surface conditions, and impurities contained in the oil.

The effect of temperature on the dielectric strength of an insulating liquid depends on its type and degree of purity (Calderwood and Corcoran, 1987). For example, the dielectric strength of dry transformer oil is insensitive to temperature except slightly below the boiling point. There, the dielectric strength decreases drastically, probably because of the formation of vapor bubbles and their growth, which are aided by the decrease at such temperatures of the oil's viscosityand surface tension. The dielectric strengths of oils that have a trace of moisture are sensitive to temperature variations over the full range from about -20" C up to their boiling point of about 250" C. The dielectric strength of an insulating liquid under DC and powerfrequency AC increases significantly with applied static pressure. Raising the

Propagation of fast positive streamers in a point-plane geometry under 0.5185 ps impulse voltage: (a) stillphotograph; (b) streak photograph; (c) Schlieren photograph (Torshin, 1995).

pressure from atmospheric to 10 times higher increases the dielectric strength by about 50%, depending on the type of liquid. Another effect of pressure is the suppression of pre-breakdown discharges. These observations support the bubble theory of liquid breakdown.Undervery fast impulse voltages of duration less than 0.05 PS, breakdown voltage is insensitive to both pressure and temperature, as observed by Korobejnikov et al. (1983).

The breakdown voltage of an oil gap depends on its width as well as the electrode shape and material. For gaps with highly nonuniform fields, for example, suchas that of a point-to-sphere gap, there is a polarity effect. The negative DC breakdown voltage is lower than the positive voltage up to a critical gap length abovewhich the relation reverses (Yehia, 1984). This critical gap length depends on the liquid and the electrode material, There seems to be no simple explanation for these phenomena.However, the material of the cathodesurface layer determines the electric stress necessary for electron emission. These electrons play a decisive role in the conduction and breakdown processes. The size and shape of electrodes determine the volume of liquid subjected to high electric stress and the degreeoffield nonuniformity. The bigger this volume, the higher the probability of its containing impurity particles. The more of these particles that are present, the lower would be the breakdown voltage of the liquid gap. The effect of moisture content is shown in Figure 7.1 1. The sensitivity of liquid breakdown to these factors is logically higher under DCand power-frequency AC than under fast impulse voltages (Yehia, 1984; Krueger, 1987). Thus, the impulse ratios of highly nonuniform gaps of contaminated or technically pure liquids can reach about 7, much higher than the gas gaps of similiar geometries. It has also been shown that stressing the oil gap under highvoltage for a long time, and repeated sparks of limited energy, tend to raise the breakdown voltage of the gap (Dawoud, 1986). This is called conditioning the oil gap. Particles in suspension collect at zones of field concentration. Points of microroughnesson the electrodes geteroded by concentrated discharge currents. A filmof discharge byproducts gradually covers the discharge areas of both electrodes. In the caseofsiliconoil, repeated breakdowns tend to cover the electrodes with a filmofgel and soliddecomposition products (Krueger, 1987). If a high-energy arc is allowed to take place in the liquid gap, the arc products cause the liquid properties to deteriorate.

kV

40
30
20

Moisture Content (P,P.m. ure 7 . 1 1 content.

Dielectricstrength

of mineral oil as a function of itsmoisture

Impurities include solid particles of carbon and wax, by-poducts of aging and discharges, cellulose fibers, residues of filtration processes, water, acids, and gases. Impurities usually cause a reduction in the dielectric strength of an insulating liquid, the largest effect being that of the simultaneous presence of moisture and fibers. Cellulose fibersare known to be hygroscopic. Thus, floating moist fibers tend to bridge the oil gap, as explained earlier. Under both AC and DC the effect of a trace of moisture is drastic on meticulously dried liquids, much greater than thatof commercial liquids (Fig. 7.11). The effect of moisture is less pronounced in the case of oil gaps with strongly nonuniform fields and with liquids containing no fibers. Because water solubility is considerably higher in silicon oil and phosphate esters than in mineral oil, they need to be muchmore carefully dried and kept. Metal particles may be present in dielectric liquids, particularly those used in transformers and circuit breakers. Their presence reducesthe dielectric strength of oil by as much as 70%. Figure 7.12 illustrates this phenomenon, in which longer and thinner particles contribute more to the reduction of the oil’s breakdown strength.

The behavior of transformer oil and other dielectricfluidsused for the cooling and insulation of power systemequipment is significantly influenced

80
70
E

60

0 0

l

2 3 Particle length (mm)

4

5

6

Influence of particle dimensions on the mean breakdown fields for free steel particles under AC voltage.

by motion enforced by the action of circulating pumps. Two important factors affect the situation. First,charges generated by streaming electrification in critical parts of the hydraulic circuit having high: velocity and/or turbulence can accumulate to distort the electric field in positions where dielectric integrity is prejudiced. Also, the dielectric strength of the fluid is altered by the actions of the flow (Nelson, 1994). Charge separation at interfaces between a moving fluid and a solid boundary can giverise to the generation of substantial electric fields. Either alone or in combination with the existing electric fields imposed by the energization of the equipment, these can give rise to insulation failure. The initial response of apparatus manufacturers has been to reduce designvelocities and curtail the operation of pumps. In apparent contrast, during standard oil testing, the continuous flow of oil was found to increase the mean dielectric strength, as seen in Figure '7.13 ( ~ a n i ~ a1990). s , The increase depends on the electrode material and is larger with steel electrodes than with brass. The increase of dielectric strength can be explained by assuming either that the oil flow impedes the entry of impurities into the gap, or that the oil motion delays the establishment of particle bridges between the electrodes. The change in dielectric strength was significant with an oil velocity of 3 cm/s, although a much higher velocity is normally needed to have such an effect. To further complicate the picture, excessive increase inoilvelocity causes the flow to become turbulent, where gas bubbles may then be created which lead to a reduction in dielectric strength.

1990).some impurity particles whichinevitably are present gradually . those of the naphthenic groups are considerably more stable than the paraffinics and aromatics. and cables. solid circles for staticoil) (Danikas.92cm2 (points plotted as triangles and squares for flowing and static oil. As low-viscosity. While the insulating liquid is in service or being stored in contact with oxygen. while resistivity and dielectric strength drop. switchgear. with more oxygen available and in the presence of a catalyst such as copper. its properties degrade.theybecomeoxidized faster than do oilsoflower grades and higherviscosities. highly refined oils have higher tendencies to dissolve air and oxygen.13 Gap spacing effect for large steel electrodes. respectively). The amount of moisture content and the acidity level rise. Oxidation of the insulating liquid is accelerated at higher temperatures. The plotted points are for electrode areas of 61. and therefore have dominated applications with transfo~ers. When an insulating liquid has been kept at elevated temperatures and/or under electric stress and exposed to oxygen. Among petroleum oils.56cm2(open circles for flowing oil.ure 7. and 82.

passivators prevent the formation of naphthanates of copper and iron which are the usual catalysts for the oxidation reaction. Second. acids. to minimize the rate of oxidation. There are also passivators that react with metal salts whichotherwisewould act as catalysts for the oxidation reaction. Oxidation inhibitors react with the free radicals and peroxides produced by the oxidation process and thus break their chain reaction mechanism. corroded iron. and polymerized liquid settle as sludges. Wax often forms by polymerization of oil at the walls of gasbubbles when the bubbles become ionized. and copper immersed in the liquid. and electrical properties of the insulating liquid deteriorate.1%. action can be taken along two fronts. To prolong the life of an insulating liquid in service. and the evolution of hydrogen. The aggressive acidic products of oxidation and discharges attack the solid insulation. the amount of oxygen dissolved in. resins. its important physical. anthra- . the actual process of aging-which is mainly oxidation-should be inhibited. Disruptive discharges such as arcs and intensive localizedheating of the liquid and solid insulation produce particles of carbon. chemical. quinones. the oxidation products should be treated so as to minimize their deleterious effects on the liquid properties. established additives include di-tert-butyl-para-cresol (DBPC). and acetylene. To startwith. and electrical properties have to be regularly checked. dimethylaniline (DMA). iron. thermal. Silent discharges of the corona type and concentrated leakage currents help the formation of water. 1986).Solid particles of carbon. For example. 1987). which would otherwise give momentum to the oxidation process (Dawoud. First. Otherwise. and gases such as carbon monoxide. chemical. Thus the physical.flocculate. carbon dioxide. incompletely cured varnishes on oilimmersedwindingsdissolvein the oil and polymerize.An example is the caseofsealed equipment with a nitrogen cushion above the oil. wax. By reacting with peroxides ofradicals. To maintain the qualities of an insulating liquid in service. Beforethe qualities of the insulating liquid change beyond permissible levels. oxidation inhibitors are dissolved in the liquid. The amount of salt added to the liquid is on the order of 0. special measures should be taken to reclaim the liquid. and phenyls. or in contact with. for petroleum oils. Also. These are scavengers that react with oxidation products and thus break the oxidation chain reaction. anthracenes. Thus particle complexes grow in size. the liquid should be minimized. although the exact amount for optimal results depends on the salt and base liquid composition (Kamath and Murthy. whereas for askarels.

quinone acts as a scavenger for the hydrogen chloride, which is the most chemically aggressive decomposition product.

To ensure the qualities of insulating liquids and their compatibility with the equipment in which they are to be used, numerous tests for these liquids have been prescribed. The International Electrotechnical Commission has issued publications describing testing methods and the criteria to be used in accepting insulating liquids (e.g., IEC, 1978, 1982a,b). The electrical tests include measurement, under controlled conditions, of the dielectric strength, dielectric dissipation factor, resistivity, and permittivity. Tests for the liquid’s physical characteristics include measurement of its viscosity, pour point, flash point, and moisture content. Chemical tests include measuring its degree of acidity, oxidation, stability, and gassing characteristics (IEC 1963,1974a,b). The moisture content of an insulating liquid is measured byheating a sample to a set temperature in a rarefied atmosphere. The water vapor developed in the test vessel is a function of the water content in the liquid, Also, the degree of acidity of the oil can be measured by noting the amount of potassium hydroxide that just neutrali~es the acids in the sample. It has been observedthat the acidity of oil affects its dissipation factor much more sensitively than it does its dielectric strength (Krueger, 1987). There are other tests for measuring and identifying the gases dissolved in the liquid. Equipment has been developed, based on the principle of fuel cells, for on-line monitoring of the amount of hydrogen dissolved in transformer oil while in service (Webb, 1987). Identifying the gases dissolved in the insulating liquid of equipment and measuring their relative quantities can assist greatly in monitoring the performance of equipment and in diagnosing any incipient fault at an early stage. Thus a major e~uipment fault could be prevented and the corresponding capital loss avoided.

Insulating liquids normally remain in service as long as their qualities have not deteriorated beyond levels set by standard specifications and general experience. For instance, their breakdown voltage, across the specified test gap of 2 mm, should not decrease below 30-55 kV; the magnitude depends on whether the liquid is used in low-voltage or high-voltage equipment. For the latter, insulation is usually designed to undergo higher stresses. The corresponding limits for the resistivity of the liquid are 3 and 10GGbm, and for the dissipation factor they are 0.2 and 0.05 at 90°C. In more highly

stressed equipment, such as cables, the limit for the dissipation factor is set as low as 0.001. The level of acidity neutralization factor of insulating liquids can be measured as the amount of potassium hydroxide sufficient to neutralize the acids in 1 g of the liquid. The acceptable level is 0.5 mgKOH/g. The acceptable level of water content is 15-30 mg/kg; the lower figure is for power equipment. The level is set even lower (0.1 mg/kg) for EHV cables, as their working electrical stresses are very high (IEC, 1982a). A s the level set for any of the foregoing characteristics is approached, measures for reconditioning the liquid should be taken. There is portable equipment now available for reconditioning insulating oils while in service (Fig. 7.14). A method well known and in use is that of filteringand vacuum drying. While under a vacuum of about 1 kPa, the liquid is heated to about 30-60°C, well above the boiling point of water at such a reduced pressure. With large oil surfaces exposed to the vacuum, it becomes freed from its moisture content and dissolved gases. In the process it also gets freed from acids and particulate matter. The process could include replenishing inhibitors and otheradditives in the liquid. Reconditioned oils usually haveabout half the life of new ones,

lnhibitor Mixing Tank

Setup intowhichinsulating oilis drawn,filtered,driedunder it is returned to the vacuum, degassed, and its inhibitors replenished before equipment.

In an experiment for determining the breakdown strength of transformer oil the following observations were made. Derive a mathematical formula to relate the gap spacing and the applied voltage for the oil. Use the result to determine the breakdown strength of oil according to standard specifications. Gas spacing (mm) Voltage at breakdown (kV)
3 68 6 140 9 190

12 255

hat are the main typesof insulating oils? Outline the merits and isadvantages of each type. List the basic properties of insulating oils and comment on the practical levance of each property. efine-analytically-the electrical conductivity of insulating oil. How is this conductivity affected by temperature? umerate the various modesof breakdown in insulating liquids. scribe the development of streamers in oils, showing their similarities to streamers in gases. Explain how the dielectric strength of insulating oils is affected by the presence of metallic particles, describing the effect of particle dimensions. Explainhow the motion ofoilinsideelectric apparatus affects the dielectric properties of the oil.

Beroual A, Zahn M, Badent A, Kist K, Schwabe A J , Yamashita H, Yamazawa K, Danikas M, Chadband WC, Torshin Y. IEEE Electr Insul M a g 1 4 % 1 7 , 1998, Breuer W, Hegemann G. Proceedings of CIGRE Symposium on New and Improved Materials for Electrotechnology, 'Vienna, 1987, Report 500.09. Calderwood J, Corcoran P. Proceedings of 9th International Conference on Conduction and Breakdown in Dielectric Liquids, Salford, UK, 1987, pp 124128. Carraz F, Rain P, Tobazhon R. IEEE Trans DEI-2:1052-1063, 1995. Chadband W, Sadeghzadeh-AraghiM. Proceedings of 9th International Conference on Conduction and Breakdown in Dielectric Liquids, Salford, UK, 1987, pp 325-330. Chan J. IEEE Trans EI-3(3):10-11, 1987. Claiborne CC, Pearce HA. IEEE Electr Insul Mag 5:16-19, 1989.

Cookson AH. Proceedings of CIGRE Symposium on New and Improved Materials for Electrotechnology, Vienna, 1987, Report 500.00. Cross J, Jaksts A.Proceedings of 9th International Conference on Conduction and Breakdown in Dielectric Liquids, Salford, UK, 1987, pp 271-279. Danikas MG. IEEE Electr Insul Mag 6:27-34, 1990. Dawoud R. MS thesis, University of Alexandria, Alexandria, Egypt, 1986. Denat A, Cosse B, Cosse J. Proceedings of 4th International Symposium on High Voltage Engineering, Athens, 1983, Paper 24.02. International Conference on Conduction and Felici N. Proceedings of 9th Breakdown in Dielectric Liquids, Salford, UK, 1987, pp 30-35. IEC. Method for the Determination of the Electrical Strength of Insulating Oil. Publication 156. Geneva: International Electrotechnical Commission, 1963. IEC. Methods for Assessing the Oxidation Stability of Insulating Liquids. Publication 474. Geneva: International Electrotechnical Cornmission, 1974a. IEC. Methods for Sampling Liquid Dielectrics. Publication 475. Geneva: International Electrotechnical Commission, 1974b. IEC. Measurement of Relative Permittivity, Dissipation Factor and D.C.Resistivity of Insulating Liquids. Publication 247. Geneva: International Electrotechnical Commission,1978. IEC. Specification for Unused Mineral Insulating Oils for Transformers and Switchgear. Publication 296. Geneva: International Electrotechnical Commission, 1982a. IEC. Determination of Water in Insulating Oil, Oil-Impregnated Paper and Pressboard. Publication 733. Geneva: International Electrotechnical Commission,1982b. Jayaram S. IEEE Trans DEI-3:410-416, 1996. Kamath K, Murthy T. Proceedings of CIGRE Symposium on New and Improved Materials for Electrotechnology, Vienna, 1987, Report 500.07. Kind D. An Introduction to High Voltage Experimental Technique. Braunschweig: Friedr Vieweg & Sohn, 1978. Korobejnikov S, Yanshin K, Yanshin E. Proceedings of 4th International Symposium on High Voltage Engineering, Amhens, 1983, Paper 24.09. Krueger M. Proceedings of 9thInternational Conference on Conduction and Breakdown in Dielectric Liquids, Salford, UK, 1987, pp 487-501. Kuffel E, Zaengl W. High Voltage Engineering Fundamentals. Oxford: Pergarnon, 1984. of 9th InternationalConference on Conduction Lesaint 0, Tobazkon R. Proceedings and Breakdown in Dielectric Liquids, Salford, UK, 1987, pp 343-347. Marinho A, Sampaio E, Monteiro M. Proceedings of CIGRE Symposium on New . and ImprovedMaterials for Electrotechnology,Vienna,1987, Report 500.06. Nelson JK.IEEE Electr Insul Mag 10:16-27, 1994. Nelson J, Lee M. Proceedings of 8th International Conference on Conduction and Breakdown in Dielectric Liquids, Salford, UK, 1987, pp 298-302. Saker A, Atten P. IEEE Trans DEI-6:784-791, 1996.

Sankaralin~am S, Krishnaswamy K, Proceedings of CICRE Symposium on New and Improved Materials for Electrotechnology, Vienna, 1987, Report 500.01. Sierato A, Rungis J. IEEE Electr Insul Mag 11%-19 1995. Torshin YuV. IEEE Trans DEI-2: 167-1 79, 1995. Webb N. TEE Power Eng J 1(5):295-298, 1987. Yehia S. PhD thesis, University of Alexandria, Alexandria, Egypt, 1984.

olid Insulating Materials
Cairo ~ n i v e r ~ i tGiza, y , Egypt

Solid insulating materials are encountered in every electric and electronic device and pieceof equipment, both large and small. They are vital for isolating conductors and for their proper performance. Solid insulants cover a very wide range, including organic and inorganic, natural and synthetic, and simple, bonded, and impregnated materials. Their common feature is that they are compounds-not pure chemical elements. They have extremelyhighelectricalresistivities and highdielectric strengths below certain temperatures. They differ in their electrical, physical, and chemical properties, which are used as a guide in selecting the material appropriate for each application. After a discussion ofthe important electrical, physical, and chemical properties common for solid insulating materials, and the theories of their electrical breakdown, some of the most widely used will be briefly described.

These electrical properties are essentially the dielectric constant, electrical resistivity, and dielectric strength of the material. The dielectric constant

h a l i f a and

Anis

(i.e., the relative permittivity) is determined by the phenomenon of polarization that takes place inside the material when under an electric field.The polarizability of a dielectric is an intrinsic property that depends on its molecular composition. The electrical resistivity of the material, together with any hysteresis that might take place under AC fields, determines its dielectriclosses, expressed by power engineers as the loss factor or dissipation factor, and by elecronics engineers in terms of the quality factor of the dielectric. The dielectric strength of an insulator depends on its material, its shape and size, its ambient, the type and duration of the applied voltage, the presence of field concentration, and other factors as discussed in Section 8.4. Solid dielectricsare compounds, as a rule; their properties may acquire magnitudes depending on the direction in which they are measured. In the special cases wherethis happens the material is nonisotropic. This phenomenon of nonisotropicity appears insome crystalline materials, including quartz.

Most dielectrics are chemical compounds ofpositive and negative ions. Thus, under electric fields, each ion tends to migrate toward the electrode of opposite polarity, as is wellknown (e.g., Seelyand Poularikas, 19’79).This polarization induces bound charges on the electrodes (Fig. 8.1). The electric flux density D is related to the field strength E by the well-known relation D = EE = EOErE, where EO and E are, respectively, the permittivities of free space and the dielectric. The relative permittivity (i.e.? the dielectric constant) of the material Er = & / E O . Thus

where P is the partial charge density on the dielectric surface resulting from its polarization. P is also defined as the dipole moment per unit volume of the dielectric. From equation (8.1) it is evident that

x being the susceptibility of the dielectric.
It can be easily shown that when a dielectric is charged, as in a capacitor, its energy stored per unit volume is ( 1 / 2 ) ~ o ~ rTherefore, E~. dielectrics with higher permittivities are favored by capacitor manufacturers. Typical values of the dielectric constants of insulating materials in common use are listed in Table 8.1.

7

+

E = V/d

(b)

Time

t

igure 8.1 (a) Polarization of a dielectric block under DC electric field; (b) current components fed toa chapter as they vary with time from the instant it is connected to a DC source.

) (c) field caused by surface charge on a spherical cavity within the dielectric block of (a).

. l Electrical Properties of Some Insulating Materials tan8 at 25"C, pv at 25°C 1MHz (!2*cm) (average)
5X 0.01 0.001-0.1 0.04 0.003 0.0005 0.1 0.003

Material Quartz glass Lead glass Soda-lime glass Glass ceramic Porcelain Zirconium ceramic Alumina ceramic Barium titanate ceramic Micanite

Er

Breakdown stress (kVlmm)

3.8 9 7 5-7 5 7-12 10 2000-8000 5-11

1 017
1ol*-l 0 j 4 1012-1015

10-40 20-80 20-30 20-30
50 (for 1mm specimen) 100-200 (for 0.1 mm specimen)

1 014-1 0j6

PVC PE Polystyrene PTFE Bakelite Vulcanized natural rubber me thy^ methacrylate

6 2.3 2.6 2 4.5 3.5 3.6

0.1 0.0001 0.0001 0.0002 0.1 0.05 0.01

40 30-40 25

8.2.1.1 The Field insidea Cavity

The insulator body may contain particles of impurities or gas voids (in, for example, plastics, glass, or paper). In such a case the field strength inside the void is bound to differ from that outside, depending on the relative permittivities of the dielectric and the void. Take, for example, the enlarged spherical gas void inside the dielectric shown in Figure 8.1. The field inside it exceeds that outside it ( E )by the component Ec produced by the charges on its surface. The density of this surface charge would vary with G according to a cosine curve, with a minimum equal to P. Thus

Substituting for P from equation ( 8 . 3 , the total field in the void is

In the case that the void has a different shape, this expression would differ. For example, if the void approaches acapillary tube along thefield extending to the electrodes, Et = E. On the other hand, if the void approaches a, thin disk normal to the field, Et= &,E, as can easily be shown.

The majority of dielectrics are nonpolar; that is, the centers of charges of the positive and negative ions in the molecule coincide. In a polar dielectric such as poly(viny1 chloride) (PVC), C2H3C1, the size and charge of the chlorine atom arequite different from those of hydrogen atoms (Fig. 8.2). Thus anet dipole forms, which is different from the case of polyethylene (C2H4) where the four hydrogen atoms of the monomer are balanced. Under an electric field, the dipoles of a polardielectric tend to become oriented along the field. In additionto this orientational polarization, the applied electric field exerts a force on each positive charge (be it a positive ion or even a nucleus of an atom) and on each negative charge (negative ions and electrons) and thus slightly displaces them along the field. Displacement polarization results in all dielectrics, both polar and nonpolar (Fig. 8.3). Also, impurity particles with permittivities and conductivities that differ from those of the parent material become charged and tend to migrate along the applied field (migrational polarization).

P.V.C. monomer

Quartz Polyethy~ene monomer

unit

cell

(b)
Molecular representation of some polar (a) and nonpolar (b) dielectrics.

Thus, in a unitvolume o f the dielectric, there are Ne atoms in Nm molecules, each with a dipole moment pe and pm, respectively. They add up to where a is the polarizability of each under the local field Et. From equations (8.4) and (8.5) it can easily be shown that

which is the Clausius-Mosotti equation. (Hummel, 1985). It is obvious that each component of polarization needs time to materialize fully (Figs. 8.1 and 8.3). The fastest is electronic polarization, with a relaxation time of the order of S whereas the slowest is migrational

"
a

c1

Displacement polarization of an atom (a) and a molecule (b), and orientational polarizationof a molecule (c). Arrows indicate field direction.

polarization for which the relaxation time extends to seconds, minutes, hours, or in some cases even longer. The relaxation time for orientational polarization reaches the order of lo3S for glass. The relaxation time for polarization manifests itself in the fact that, if a step-function voltage is applied across a capacitor with a solid dielectric, its polarization will take some time, howeversmall, to build up, as shown in Figure 8.4. In studies concerned with periods much longer than the electronic and ionic relaxation times, and if the other polarizations are assumed to have one equivalent relaxation time t , the polarization of the dielectric under study varies with time according to a relation of the form
P(t) = P , (P, Po) exp -

" -

(

a

" " " m

" "

0

t

Buildup of polarization P in a dielectric under the effect of a step-function-imposed electric field E. P,, initial polarization; P , , polarization value after infinite time.

If the initial and ultimate magnitudes of polarization Po and P, are significantly different? this means that the relative permittivity of the material has two correspondingly different values, according to equation (8.2): and .,,E Now, if a capacitor with such a dielectric is subjected to a sinusoidally alternating voltage with angular frequency CO,the polarizaton will vary with the voltage but will lag slightly behind it, depending on the difference (P, Po) and on o as compared to l / t . The polarization and permittivity would thus be complex quantities, that is,
P

P -jP”
f I I = Er JE,
*

(8 8)
*

and
E,

It can easily be shown that (8.10) and (8.11) These twocomponents depend on thedielectric material, its relaxation time, and the frequency ofthe applied electric field.Both are shown schematically in Figure 8.5. Under DC, and also at very low frequencies, E:--+& and : E almost vanishes, as noted from Figure 8.5 and equations (8.10) and (8.1 1). With increase in frequency? the inertia of the dipoles will cause the lag of the orientational polarization behind the field alteration to become more and more significant9until this type of polarization ceases to contribute to the whole. Thus E: drops. It can easilybe shown that at wz = 1, ds:/dw is maximum and E: reaches a peak. This explains the variations of the two curves in Figure 8.5 at the first and subsequent resonant frequencies. The variation of E : withfrequencyisusually temed the “dispersion” curve, : ’ is known as the “resonance absorption characwhereas the variation of E teristic,” as E: represents dielectric loss (i.e., energy absorption). The dispersion of dielectric materials can be measured by tests described by national and international standard specifications (IEC, 1973). A. capacitor containing the dielectric is charged from a DC source with a voltage V. and then momentarily short-circuited and its residual open-circuit voltage VI is measured; the dispersion is expressed as
Y,/(Vo

-Vl>*

IS

,

Schematic showing thecomponents E: and E; of the complex permittivity of a dielectric as they vary with the frequency of the applied electric field: at a temperature 7 , ; , at a higher temperature T2. [From 980). ] Tareev (1

The effect oftemperature on the permittivity and polarization of nonpolar dielectrics is negligible. In polar materials, however, the random thermal motion of the dipoles is heavily influenced by temperature. The higher the temperature, the lower will be the relaxation time z. tudes of E/, will be lower and the resonant frequencies of : E will be higher (Fig. 8.5).

Practical insulating materials have admittedly high resistivitiesbut these are, nonetheless, finite.For wood, marble, and asbestos the volume resistivitypv is in the range lo6--lo8 s-2 . m, whereas for polystyrene and polyethylene it is in the range 1014-1016s-2.m. Quartz crystals are nonisotropic; p, is about 10l252 m along its optical axis and about 1014s-2 m in directions normal to it. The volume and surface resistancesof an insulator are two paths in parallel for the leakage currents I, and I, it draws when connected to a power source (Fig. 8.6). Their relative valuescan vary drastically, depending on the insulator material and its surface conditions. Electrical conduction through dielectrics isundertaken by ions, and is basically different from the process in conductors where electrons are responsible for the conductivity. The reason is that in dielectrics the energy necessary for dislodging ions from their positions in the atomic lattice is on the order of that of their thermal vibrations at normal room temperature.
a

Earth Insulating bus support in half-section showing thecomponents two of its leakage current: I . withincreased temperature. the energy required to remove an ion from its lattice position is 0. Also. creeping on its surface. for insulating materials the volumeresistivities decrease sensitively with increased temperature.There is additional energy loss in the process of polarizing the material. their mobilities increase with temperature ex~onentially. more ions could be dislodged from their positions in the atomic lattice and could contribute to the electrical conduction.Therefore. the energy neededto liberate electrons in a dielectric isat least one order of magnitude higher. On the other hand. Practical insulating materials draw leakage currents. The latter . through itsvolume. . in a simple dielectricsuch as rock salt (NaCl). It isevident that. For example.8 eV. whereas electrons need at least 6 eV for their liberation. however small. andI . .

loss is significant under AC with frequencies approaching the resonantfrequencies shown in Figure 8. It causes a considerable rise in the loss factor above the discharge-onset voltage (Fig.4. heat is removed primarily by convection. a trace of moisture accentuates the effect of temperature. Measuring the loss angle 6 is one ofthe important tests to be carried out on test samples of highvoltage cables (Chapter 18). in a heterogeneous material. and some electronic devices. Distinguished examples are plastics and glass. on the other hand. Evaporative cooling is sometimes used to cope with hot spots. Thus. Second. for example. at higher temperatures). in a homogeneous dielectric the conductivity contributed by the jumping of ions from one site to another under the applied electric field becomes easier with higher vibrational energies(i. the quality factor of the dielectric. such as paper. 3% by weight of moisture causes the loss factor to rise from 0. The corresponding power lossis analogous to the corona loss (Chapter 5). Power engineers refer to the factor (tans) as the loss factor. In transformers and machines. under AC the current drawn by a practical capacitor leads the applied voltage by an angle slightly less than 90". First. Heat is conducted through insulating materials by phonons. for instance.. A well-known example is the ionization that occurs in gas voids within the material. the resistivity drops andthe loss factor rises about 10-fold for a temperature rise from 80°C to 120°C.5% at 20°C to about 30% at 99°C. capacitors.2.7).e.2 Effect of Electric Stress The loss factor for an insulating material is not much affected by the magnitude of the applied electric stress except at levels high enough for secondary phenomena to set in. With PVC. 8. Thermal conduction of insulating materials is the one important factor determining heat transfer in cables. 8. At higher voltages the gases in additional voids get ionized.5. Thevolumeresistivitiesof insulating materials therefore drop considerably with temperature.1 Effect of Temperature There are two factors to be considered here. 8.2. whereas electronics engineers prefer to use cot S. A well-designed and well-made cable should have the minimum number of gas voids (Chapter 12).4. the difference S is termed the loss angle. For oil-impregnated paper. . The thermal vibrations in a crystal or a solid continuum can be represented as elastic waves propagating through the solid with the velocity of sound.

capacitors. The curves represent the radial field distributions at different applied voltages V . For a l u ~ i n a and magnesia it is as high as 35 W/m. V Thermal conductivities of plastics range between 0.2. varnish.K. or other insulating liquid. ~ and natural silk. when not impregnated with oil. which recommends both materials for making special ceramics. and cables have been classifiedaccording to the ma~imum safe temperature each material or combination of materials can endure for a very long time under normal service conditions (IEC.K. >. The code letter and temperature limit assigned to each class are given in Table 8. 1984). This class also includes vulcanized natural rubber andthermoplastics limited by their softening temperature to .I ur Cross-section of a coaxial paper-insulated cable with gas voids scattered in its insulation.2 to 1.3 W/m. ' :~paper. ~ Z T . Eois the gas discharge onset field strength. V >. transformers. Insulating materials commonly used in machines. whereas for porcelain and glass it is about 1. .K.7W/m. s cotton. Examples of materials belonging to each class are given below.15 and 0.

m a a.0 N v) 0 CV N cj 0 0 0 2 A N x U CO Lu a .’ 0 0 v) U) Ts nf . - 0 a > . E: .

insulating materials inadmissibly degrade in their electrical and mechanical properties. alkyd. s : inorganic fibrous and flexible materials (e. This class also includes some organic enamels. nce inside the cable and under electric stress. ceramics. 1984) into three subclasses. such as silicon resin. and thermoplastics with suitable softening temperatures. decompose. most plastics have a measurable permeability for water. It is realized that insulating materials in general can withstand temperatures significantly higher than the limit assigned to their class but only for a very short duration-a fraction of a second for a short circuit cleared rapidly by a circuit breaker. The cellulosethey contain.is be within this class. such as plastics and cellulosic materials. asbestos) when bonded or impregnated with an organic binder such as shellac. epoxy enamels. r resin. like hydroxyl. being polymers. materials of class but bonded or impregnated with suitable inorganic resin. and polyethylene terephthalate fibers). coded 200. these water droplets aim at . has an affinity for water. This class also includes polytetrafluoroethylene (PTFE).g. In 1984 this class was split (IEC. Their electrical conductivities and losses increase.. cross-linked polyethylene (XLPE). varnishes. corresponding to their safe operating temperatures.mica. :materials of class B but bonded or impregnated with resins of a1 endurance (e. craze. and varnishes of suitable thermal endurance (e. A: paper. carboxyl. are apt to melt. fused quartz) when used without binders or with binders suitable for the required thermal endurance. Their mechanical strength and elasticitylimitdecrease considerably. mica. phenolformaldehyde. epoxy and silicon alkyd). and 250. Also. or rosin with mineral fillers. char. or even burn at elevated temperatures.g. and natural silk when impregnated with oil. At too high a temperature.glass. and amine groups. ss C: inorganic materials (e. Moisture has been found to permeate through micropores in plastic sheaths of underground cables. Some insulating materials. 220. melamine-formaldehyde... and poly(viny1 chloride) (PVC).g.. Paper and other organic fibrous materials are bound to contain at least traces of moisture. This class includes silicon rubber. cotton. such as polyethylene (PE).g. ss E: laminated and molded plastics with thermosetting binders. shrink.

Each occurs under a certain set of conditions conducive to it. Thermal endurance tests are by no means straightforward. nor should there be . However. for example. three primary processes of dielectric breakdown have been discerned: intrinsic. 1986. a thermal endurance test that has been widely adopted is to measure the loss in weight of a specimen after exposure to a certain hightemperature for several hundred hours (Steffens.1983. extremelydifficult to identify in practice by eliminating all secondary causes of breakdown. At their new sites they will highly distort the field. Therefore. because the degradation of an insulating material due to exposure to a high temperature cannot be expressed by a single variable. and other such easily ionizable ingredients if dissolved in water. no single theory fully explains the process of breakdown and predicts the breakdown stress of a given insulator. acids. Although solid insulating materials have been the subject of numerous investigations. A trace of 3% moisture absorbed in drypaperreduces its resistivityby about six orders of magnitude. 1981.sites of higher field concentration by electrophoresis. meticulous care is taken during the manufacture of paper and similar cellulosic materials when designed for use as electrical insulation. thermal. It is an ideal. Care should be taken not to allow breakdown due to thermal instability of the dielectric with temperature rise caused by leakage currents.They are all discussed briefly in this section. The deleterious effects of moisture are worsened by salts. nor should the dielectric fail under electromechanical stresses. andambient conditions. However. as discussed in Section IV. become ionized. It is therefore sometimes called electronic breakdown. Intrinsic breakdown occurs purely due to the electronic behavior of the dielectric. One of the principal reasons for this state of affairs is the dependence of the breakdown on numerous factors. IEC. voltage duration. including the material’s temperature. and electrochemical. with no effectofambient or temperaturerise. and start water treeing.198’7). The process of breakdown of solid dielectrics is much more complex than that of gases. which may eventually cause breakdown of the cable. Tests have been prescribed for ensuring various qualities of insulating materials.

A detailed discussion of his theory is beyond the scope of this section. Wi is the ionization energy for the atoms in the lattice: . A s his crystal constants are not measurable with any accuracy for simple crystalline dielectrics. It is thus evident that intrinsic breakdown would be possible for idealized crystalline materials. The energies of these electrons differ according to Boltzmann’s distribution (Chapter 3). In 1947. >E2 E. such as ~ o t a s s i u chloride. ~ relating the breakdown stress to six constants of the crystal lattice. and 3 represent the rates of energy gain by an electron from fields with different strengths E l >E2 >-E3.. whereas curve S represents the energy loss. Frohlich proposed a criterion for electronic breakdown as “the electric stress at which the dielectric acquires a field-enhanced conductivity.Thesewouldbeaccelerated by the applied electric field and would collide with some lattice sites on their way. its energy gain from the electric field and loss during the interaction with the lattice site it collides with vary as shown sche~atically in Figure 8. Curves 1. energy. The limit Wi is Rates of energy gain by electrons from the applied electric field the rate of their energy loss S to the crystal lattice as functionsof the electron E.2.” He developed a theory for crystalline dielectrics. a brief qualitative review will suffice. to say nothing of practical amorphous materials.impurities or gas dischargesto initiate failure due toelectrochemicaldecomposition.8. Cosmic and other radiations wouldset a few electrons free in the crystalline dielectric. Depending on the energy of each individ~al electron.

In such a case the temperature starts torise monotonically instead of levelingoff at a steady value (Fig. 8. Thermal breakdown results from disruption of thermal equilibrium in the dielectric: the rate of heat generation within the material exceeds that of cooling. for example.k the thermal conductivity of the dielectric. a runaway process sets in.9a).) . known to decrease almost exponentially with temperature. was 10 MV/cm and for polyethene it was 8 MV/cm. Attemps have been made to measure intrinsic breakdown voltages of dielectric specimens in the laboratory under carefully controlled conditions. we get (8. This would result in the field-enhanced conductivity suggested by Frohlich. as shown in Figure 8. The heat balance of the elemental volume (AAAx) at a distance x from the center plane is E2(x)AAAx = -khAAxPV(X> d2T dx2 (8. the volumeresistivity of the insulating material decreases and the dielectric losses increase. Care was taken to eliminate or greatly suppress secondary causes of breakdown.the energy sufficient for the electron to ionize the atom it interacts with at the lattice site.Thus the critical field strength is E2. obtained with extreme care to prevent secondary effects from causing breakdown at much lower stresses. consider the linear heat flow to the ambient together with the dielectric losses in a capacitor of a large area A under DC (Fig. and pv its volume resistivity.10). equation (9. leading eventually to breakdown. with a monotonic temperature rise. Realizing that the density of leakage current through the dielectric i = E(x)/pv(x)is independent of x. the energy gain should just exceed the loss.13) Integrating both sides over the distance 0-x.8.14.12) where E(x) is the electric fieldstrength at x. The highest breakdown stress reached for mica. With temperature rise. For intrinsic breakdown to occur.12) can be rearranged to -k- d2T = iE(x) dx2 (8. 8. Attempts at mathematical analysis of this situation were made by Whitehead in 1951 (Kuffel and Zaengl. 1984). Thus. For simplicity.

> V. Thus. Thus.the critical temperature at which the thermal runaway process sets in.Time (a) Temporal increase of temperature of dielectric in a capacitor when subjected to different step voltages: V. ) over the range from the . (b) Schematic showing the variationsof the volume resistivity pv of a dielectric and of the product of its k. V. Tcr is the critical temperature at volume resistivityand its thermal conductivity which the dielectric fails. according to Whitehead’s theory. we get ” V2 8 (8. The critical conditions are reached when the temperature at the center of the dielectric I. the center plane of the dielectric block. = Tcr.14)..15) where V is the total voltage across the dielectric block. rearranging and integrating equation (8. as d~T/dx = I : V ( 0 ) = 0 at x = 0. the thermal breakdown voltage of a dielectric block is independent of its thickness and can be obtained by integrating the product ( k p .

Experimental measurements. The heat is caused by loss under voltage. Co~putations of this area were made for numerious insulating materials and their voltages for thermal breakdown ranged between l and 5 MV (Alston. The relation (8. The discrepancy is explained at least partially by the nonhomogeneity of most dielectric materials and inaccurate values of their physical characteristics at higher materials.10 Heat flow tothe ambientfrom adielecric block. no matter how thick the insulation is made. A theoretical account can be developedif the surface temperature of the specimen is allowed to exceed that of the ambient. however. In this case the heat balance equation per unit area over the entire thickness of the specimen (Fig. One other important point to note here is that there is a limit for the thermal breakdown voltage.15) eventually does not hold for very small thicknesses. 1968).) d (8. This is to be kept in mind when designing high-voltage machines and EHV cables.10) takes the form or V2 = HpvF(7.9b). The maximum temperatureTi occurs at the centhe dielectric Ta is the ambient temperature. and the product HF(Ti) gives the rate of heat . ter plane. which is more realistic. whereas ambient to the material's critical temperature (Fig.16) The function F(Ti) relates the surface temperature to the highest temperature inside the specimen. yielded voltages up to only 4QO kV. 8.Figure 8. 8.

Hosokawa etal. This would cause extension and branching of the tubule in a treelike shape in steps of microfractures of the material (Fig. the phenomenon is termed water treeing. Although the discussionabovewasconcernedwith a DC voltage applied across the dielectricspecimen. 1987. some cables have actually failed in service as a result of treeing (Deschampset al.. with its soluble ions.. There maybe metallic particles. Water.. the sameresultcouldbereached had we considered instead an AC voltage. When water is involved. . Thus. the right~hand side of equation (8. was seen to lead to slow. In that case the dielectric loss would be expressed as (oCV2tan 6). 8. 1984. Treeing begins at microscopic sites of imperfections in the molecular structure of polymers. Although specimensofpolyethylene and cross-linkedpolyethylene tested underAC of short duration break down atnot less than 700" 800 kV/mm.16) can be considered as a constant for a given critical temperature at which the dielectric failsthermally. yet complete. 1987). would migrate by electrophoresis under the electric field to sites of higher intensity. 1984. Filippini et al.Such cracks mightbe 10 nm in width in the amorphous regions between crystalline lamellae of semicrystalline polyers (Sletbak and Ildstad.dissipation to the electrodes and ambient. breakdown. combined with hydrostatic pressure caused by water inside it being heated by concentrated leakage currents and chemical decomposition. this simple approximate approachyields the relation between the thermal breakdown voltage and the small thickness d of a dielectric specimen as (8.17) c being a constant. cavities. Treeing beganto receive wideattention in the mid-1970s as polymers started replacingoil-impregnated paperto insulate high-voltagecables. l 1). would result in pulsating mechanical forces at the tip of the tubule. and the macroscopic working stresses in cables with such an insulation range from aboutonly 3 kV/mm (in low-voltage cables)to 20 kV/ mm (for 250 kVcables). all of which are extremely difficult to eliminate. Marton. pulsating under AC. As the dielectric resistivity pvis another function of temperature. water can diffuse through microscopic pores and cracks in the insulation and through sheaths ofpolymer-insulatedcables.Electric treeing initiated by gasdischarges at microscopicsites. A concentrated electric fieldat the tip of a tubule. Also. or thermo-oxidated or deteriorated particles of material. if not inhibited. 1982).

carboxylate ions are dominant. Xu. Rasmy and0. although on a very different time scale (Meyer and Filippini. Sletbak. 1979. Among carbonyl species. The water tree region contains a wide range of chemical species. which results in the condensation of water molecules dispersed in the polymer matrix to form liquid water in the track.. . A s a consequenceof the electro-oxidation. 1979. 1996).. Fan and Yoshirnura. with various forms of carbonyl and metal ions and water being dominant. 1994. polymer chains are broken and a “tree path” is formed. The water tree region contains a wide range of chemical species.) an The oxidation theories of water treeing describe the growth of water trees as driven by electro-oxidation of the polymer. Thus a track becomesselfpropagating in a manner similar to the self-propagation of an electrical tree or a gas breakdown channel.Fischeret al.l1 Tree developed at the tip of a needle electrode inserted into epoxy dielectric. which takes place in the direction of the local electric field and in a polar amorphous region of the polymer. The local electro-oxidation of the polymer along this track converts the track region from hydrophobic to hydrophilic. 1987. followed by trace amounts of esters and ketones. Gouda. with various forms of carbonyl and metal ions and water being dominant. The liquid water in a track promotes the transport of ions that mediate further the electro-oxidation of the polymer at the tip of the track. (Courtesy of A.

The concentration of watersoluble ions on the surface of semiconducting material can bereduced through simple water extraction.3 Role of Semiconducting Compounds in Water Treeing of XLPE Cable Insulation Semiconductor shields are used in power cables. The initiation of watertrees at a semiconductor-polyethy~eneinterface requires ions which are generally present in and on thesemiconducting material. their breakdown strength depends on the aging conditions. hydrophobic)have little influence on their growth of water trees (Fan and Yoshimura. . in turn. can greatly inhibit the growth of water trees.2 Influence of Crystalline Morphology on the Growth of Water Trees in Polyethylene (PE) The growth of water trees is faster in low-density polyethylene samples in which spherulites are larger and the number of the spherulites is smaller. The water treeing processes determine the properties of the resulting water trees.8. 1993).3. with the newer technology products which incorporate the cleanest carbon black showing a trend toward reduced tendency for the growth of vented water trees oggs and Mashikian.3. It has been shown that water trees grown under different conditions can be quite different entities. On the contrary. The type of water tree growth mechanism varies considerably with the applied aging conditions.1 Effect of Aging Conditions on the Type of Water Treeing Thecomposition of water trees and. The characteristics of water trees differ from those of unaged polyethylene. 1996). they take the form of small rounded bodies with radiate fibrous structure. 1994). Spherulites are one of the main constituents of PE. Ground water willgenerally carry substantial ionic contamination.4.3. while other two kinds of boundary region (weak electrolyte. The growth of water trees does not depend solely on ions originating from the semiconducting shields. The tree length and size ~istributiondiffers significantly for the semiconducting compounds tested. having different properties (Ross. and such reduction has a substantialeffect on the number and length of water trees. water trees grow more slowly in low-density polyethylene samples in which spherulites are smaller and thenumber of the spherulites is higher.4. whose character is hydrophilic. The spherulite boundary regions. 8.4. 8.

. due to the manufacturing processes and/or the inability to carry out post-manufacture annealing.is preferred for having less initiation and slower growth. The relationships between treeing phenomena and specific mechanical properties can be masked by the simultaneous dependence on more. The electrical tree forms at the end of the water tree. The main reason for the degradation in mechanical properties is that the material passes through its glass transition point within the range of test temperatures. strain-free insulation may be impracticable. produced by plasticization. the dielectric constant. _ _ . The treeing resistance displays a similar fall as the temperature exceeds this critical value.8. -In hogogeneous ___ __ __ resin. The material's resistance to crack propagation therefore becomes lower with the application of heat. . this outer region that an electrical tree is initiated. For example. An inverse relationship exists between the size of the water trees causing the . modulus of elasticity. however. The true dependence of tree growth on fracture toughness can be observed only under conditions of low internal strain (Auckland et al. which are filled with water whenthe tree is water saturated. " .4 Mechanical Aspects of Electrical Treeing in Solid Insulation The mechanical properties of insulation affect the initiation and growth of electrical trees. The ensuing reduction in tensile strength. but the structure of the outerregions is a collection of voids.5 Water Tree-based Electrical Trees The water tree structure consists of a large number of voids. 8. The electrical field associated with a water tree depends on the shape of the tree. 1993). Strain-free material.4.3. . in which case plasticized insulation may limit initial strain and enhance tree resistance in a material. is detrimental from the point ofviewof tree inhibition.6 Length Distribution of Water Trees The length distribution of watertrees-rather than the breakdownvoltage-is inmanycasesused to assess the degradation in insulation. or even beyond. plasticization effects on fracture toughness are masked by the reduction in initial strain with decreased tree growth.4. and the outcome will depend on the dominance of one parameter over others under a given physical condition. 8.3.4. the tensile stength. than one mechanical property. Channels may be present within the inner regions of the water tree. and fracture toughness all show a pronounced decrease in value as the temperature is increased. and fracture toughness. in a specimen containing initial strain.. and conductivity within the tree. in which case plasticization is to be avoided.3. In reality. modulus of elasticity. It is in. contrary to expectations.

rubbers. dust. strength. With water-treed low-density polyethylene ( L ~ samples ~ ~ ) aged in the laboratory. 1993). showed a loss peak atto low3 H z when moisture was retained in the samples.measurements of transient currents. with subsequent transformationintothe frequency domain. Moreover. e Growth using Transient In a study by Li et al. polymeric materials contain carbon atoms in their molecular structure and are therefore prone to tracking. as time goes on and contaminants such as salt.4. and polyesters for reasons of economy. (1995). andease offabrication to good tolerances. distorting the stress . 8. etc. As the leakage current flows between live parts. the growth rate of water trees can be determined. a higher dielectric loss is observed in water-treed samples than in virgin samples. epoxies. The relaxation behavior may serve as an indicator of water treeing in polymeric cable insulation. on support structures. The results of both transient measurements were said to be affected significantly by the treatment conditionsof the samples. a time window for transient currentmeasurements was chosen so that reliable low-frequency dielectric data could be obtained. This is particularly alarming with the increasing use of dielectric organic materials such as polyethylene. lectrieal l~sulatio Insulation surfaces may be capable of meeting the overvoltage criteria when the equipment is installed. However.” Solidlair interfaces-which are subject to tracking-exist in bushings or terminations. the standarddeviation of tree lengths. or in the end-windings of a rotating machine. will conduct leakage current with a magnitude determined by the type and extent of the pollution.breakdown and the breakdown stress level for cablesaged at moderate stresses (Ross. the integrity and reliability may be jeopardized under humid conditions by a form of surface failure called “tracking. the conductive filmwill heat nonuniformly and begin to evaporate nonuniformly. When the durationof aging is also taken into account. Even with free water removed. whether formed by condensation or precipitation.1 Failure Mechanism A moisture film on a polluted surface. the density of trees. Variables reflecting the length distribution can be: the largest water trees per unit volume. and atmospheric chemical agents collect on these surfaces. or from a live part to ground.4.

to specify track-resistant plastic insulating materials in the first place.. 198’7). Similar surface failures have been observed in other circumstances. and even to modify such materials. support panels. is highly recommended.4. “dry-bands” will form. In most cases the performance of a material such as glass-reinforced polyester resin can be greatly improved by the addition of materials such as hydrated alumina. such as on structures in oil-filled transformers. Preventive maintenance in the form of frequent and thorough cleaning of insulator surfaces. In particular. the temperature of the arc may be sufficient to decompose the material locally. The “track”on the surface usually has a characteristic branched appearance. Dry-banding. In some cases a temporary solution may beachievedby cleaning or filingaway the carbonized matter and applying a good grade of electrical varnish such as an alkyd. Such maintenance is obviously very costly. machine end-windings. 1982). arcing. It is far better to replace with a better material as soon as possible (Kurtz.12 shows a model of the tracking breakdown process (Yoshimura et al. It thus makes sense to spend more money. Figure 8. 8. and bysevere transient overvoltages. or on the surface of polymeric (filled epoxy) spacers in gas-insulated gear. and the like. resulting in regions of very high resistivity between the edges of the remaining wet film.4.2 ~ i t i ~ a t i of on Tracking Tracking may be prevented by keeping the surfaces in question clean and dry. Nearly the total surface voltage will appear across this dry-band. where possible. or at least bridges a sufficient portion of the surface. When this occurs on the surface of an organic material. causing flashoverof the gap. Some polymers are inherently more resistant to tracking than others. during repairs after a tracking failure. releasing free carbon. resulting in flashover. where this is possible. The arcing will continue until the gap widens sufficiently byfurther evaporation to the point where the arc can no longer be maintained. if necessary. Most can be improved by the addition of inert fillers as well as hydrated alumina. . the tracks on GIs spacers are more frequently associated with damage done by internal flashovers caused by the presenceof particles. which acts to oxidize free carbon as it is formed.distribution over the surface. and the formation offree conducting carbon will proceed in a relatively random manner until a continuous conducting path forms between the live parts. Some epoxies are better than others. The installation of heaters in outdoor metalclad switchgear compartments has solved many problems. Whereas tracking on headboards or barriers in oil-filled transformers is also usually associated with the presence of excessivemoisture.

8. and ethylene vinyl acetate (EVA). the tests rank materials in roughly similar order. A number of very effectivetest methods havebeendeveloped to quantify andrank insulating materials in terms of their resistance to failure by surface tracking. contamination. so that virtually any of them may be used to select a “super- .4.Provided Energy: High Provided Oxygen: Low 1 Provided Oxygen :Low I Provided Energy:High 1 Arc or Glow Discharge Formed Carbon A model of tracking breakdown. and moisture.3 Tracking Tests It is important to evaluate the tracking resistance of such materials as high temperature vulcanized (HTV) and room temperature vulcanized (RTV) silicone rubbers. In general.4. the test procedures expose samples ofthe (organic) insulating material to an environment representing an extreme exposure to voltage stress. In most cases.

4 Environ~ental Effects on Tracking The environmental effects on tracking are considered. they have to be specially treated and impreg- . The materials evaluated included various formulations of HTV silicone rubber and polyolefin polymers.13 (Kamosawa et al. A laboratory test for evaluation of tracking and erosion resistance of high voltage insulating materials has been described (Corur et al. The lifetime of insulators is affected by exposure to UV radiation. another material-polyethylene terephthalate-showed a decrease in tracking resistance with decreased atmospheric pressure (Du et al. ) I T I Traditional materials include cellulosics such as paper and organic fabrics.some tests may be better suited to distinguish between materials at one end of the “severity of exposure range” than at the other. 8.4. to be usable as good insulating materials. 1995). Therefore. 1988). They also include mica.4. Of those effects the following are considered to be relevant: UZtraviQZetrays: Outdoor insulators are subjected to ultraviolet (IJV) radiation from the sun. Cellulosicmaterials suffer from hygroscopicity and are bound to contain gas voids. The test was usedfor screening materials and obtaining a relative ranking of the tracking and erosion resistance ofvarious materials.. Ultraviolet rays are one of the most severe aging factors for polymer insulators (Du et al. 1996).and ceramics. however.1 ior” material. It is claimed that materials that pass the 8-hour test without initiation of tracking or significant erosion under conditions that are more severe than field conditions will be suitable for outdoor HV insulation applications. This test is based on combining some features of the dust and fog test and the inclined plane test.. In contrast. and the discharge activity was monitored with a high-speed camera. Other environmental factors that may have effectson tracking resistance of polymers include exposure to gamma radiation and to acid rain . glass. because of the increasing use of organic insulating materials in areas that are heavily contaminated. Low a t ~ o ~ ~ pressure: ~ e r i c The tracking resistance of polycarbonates was reported to increase with the decrease in atmospheric pressures such as that encountered at higher altitudes... Measurements of the leakage current were made via a computerized data acquisition system. and/or in highlyhumid regions. 199’7). as seen in Figure 8. at high altitudes.

l r ~ d i ~ tTime i ~ n (h) . enabling the mica to withstand much higher temperatures. synthetic mica has been developed in which the OH groups are replaced by fluorine atoms. 1984).1987. particularly suitable for insulators in high-voltagehighfrequency equipment. or electric stress. since the 1960s much lighter insulators with excellent electricaland mechanical characteristics have been made of epoxy resins reinforced with glass fibers. Their molecular arrangements are either linear or cross- . nated with oil or varnish. htlmidity. However.24 CZI. The insulating characteristics can be improved drastically by combining paper with polypropylene (PPP) (Samm. they can be manufactured for use as insulators with almost any application. as magnets. Depending on the composition of ceramics.. and DC machine commutators). Mica powder mixed with finely ground glass in a hot-pressing process yields a dense machinable material with excellent electrical qualities. in vacuum tubes. Schaible. Several newly developed insulating materials fall within the classification as polymers. heaters. 20 0 1 2 5 10 25 50 100 U V . Mark. Recently. l 3 Effect of ultraviolet radiation on the time to tracking breakdown. 1987. Porcelain and glass havetraditionally been usedin insulating overhead power lines and busbars (Chapter 9). Mica has traditionally been used where high temperatures and/or surface discharges are experienced(e. or as semiconductors with resistivities sensitively varying with temperature.g.

is stronger than that of carbon. As an exception to the rule. Because of the high binding energies of fluorine atoms. polytetrafluoroethylene (PTFE). polysiloxanes (silicon resins) can be either elastomers or thermosets. Si-0. which accounts for the considerably higher thermal endurance of silicon polymers. there are dielectrics that have other applications. Crosslinking has considerably improved the electrical qualities of PE. and polyamides (nylons). At temperatures ashigh as 4OO0C. The organic radical R could be a methyl (CH3). In additionto the passive dielectrics employed solely as electrical insulators. is not flexible nor does it yield and melt at high temperatures. the greater the yield. It enjoys thermal and mechanical enduranceand chemical stability. the polymer is about 95% crystalline. The latterhave a molecular arrangement such as thatrepresented in Figure 8. Depending on their production process.it decomposes. The link of silicon with oxygen. 1985). whichis a linear polymer. poly(viny1 chloride) (PVC). Thus XLPE has been favored as an insulantin both low-voltage and medium-voltage cables. ethyl (C2Hs).R I l I I R 1 R l R I R R R R l 0 R I R I 0 R R R Silicone Resin Representation of molecular arrangements of a silicon resin. or phenyl (C6HS). linked. the former including elastomers and thermoplastics. C-C.14. In such “active” dielectrics there are mutual effectsbetweenimposedelectricfields and mechanical .They include polyethylene (PE). and are generally very flexible and yield with time under a given mechanical load. much higher than that of PE and PVC (Margolis. The higher the temperature.

stresses. PA. A coaxial cylindrical capacitor is to be designed withan effectivelength of 2Ocm. They are used in piezoelectric and pyroelectric transducers. andif the breakdown strength of air can be taken as 30 kVPeak/ cm. shown in the figure. vidicons. 1987). electric light shutters. (1) A solid specimen of dielectric has a dielectric constant of 4. The specimen is 1cm thick and is subjected to a voltage of 80 kV (rms). nonlinear capacitors. USA. 1993. Pocono Manor. has an internal void of thickness 1mm. 1968. Michel R. and tan 6 = 0.CIGRE report 21-08. and electrets.ProceedingsofConference on Electrical Insulation & Dielectric Phenomena (CEIDP).2. Lepers J . ferroelectric memory devices. IEEE Electr Insul Mag 10:23-27. whose dielectric constant is 2. Deschamps J. If the void is filled with air. Auckland DW. electro-optic voltage sensors.VarlowBR. find the voltage at which an internal discharge can occur. Oxford. 1984. .3 and which breaks down at 50 kV/cm. calculate the heat generated in the specimen due to the dielectric losses. (2) A solid dielectric specimenof dielectric constant 4. If it is subjected to an alternating field of 50 kV/cm. heat influx or penetrating light. High Voltage Technology. estimate suitable dimensions for the capacitor’s electrodes. Mashikian MS. If high density polyethylene is selected as the capacitor’s insulating material. Alston L. The capacitor is expected to have a capacitance of 1000p F and to operate at 15 kV. pp 636-641. Taha A. 1987.0. UK: Oxford University Press. CIGRE.001 at a frequency of 50 Hz. Proceedings of CIGRE Symposium on New and Improved Materials for Electrotechnology. oggs SA. Their far-reaching applications are extending rapidly in many areas of engineering (CIGRE. 1994. Vienna.

1979. Kamosawa T. Filippini J. 1985. 1993. 1987. IEC. Geneva: International Electrotechnical Commission. Proceedings of IEE International Conference on Dielectric Materials. Geneva: International Electrotechnical Commission. Hosokawa K. Kojma K. 1985. 1983. Toshida N. 1981. Fouracre RA. Recommendations for the Classification of Materials for the Insulation of Electrical Machines and Apparatus in Relation to their Thermal Stability in Service. New York: Marcel Dekker. . Proceedings of CIGRE Symposium on New and Improved Materials for Electrotechnology. Proceedings of CIGRE Symposium on New and Improved Materials for Electrotechnology. 1984. Kishi K. Kobayashi S.Du B. Methods of Tests. Kobayashi S. Yoshimura N. Li HM. Mark R. Varadadesikan L. Simmons S. Fan ZH. Filippini J. Publication 371-2. Frohlich H. Suzuki A. Proceedings of the Royal Physics Society A188:521-532.04. Suzuki A. Shah M. 1947. Specifications for Individual Materials. Geneva: International Electrotechnical Commission. High Voltage Engineering Fundamentals. 1995. Noto F. Vienna. IEEE Electr Insul Mag 9:7-16. Specificationsof Insulating Materials Based on Mica-Definitions and General Requirements. Report 620. IEEE Trans DEI4:767-774. IEEE Electr Insul Mag 3:12-14. Nishida M. PoggiY. Zaengl Pergamon. Kurtz M. Meyer C. IEC. CIGRE report 21-09. ed. Handbook of Physical and Mechanical Testing Paper and Paperboard. Proceedings of CIGRE Symposium on New and Improved Materials for Electrotechnology. 1992. New York: Marcel Dekker. Manchester. Schroth R. de Bellet J. pp 484-487. Peschke E. 1987. UK: Kuffel E. Montesions J. IEEE Trans DEI-2:866-874. Report 620. Yoshimura N. 1996. Gouda 0. Publication 261-1. Measurements and Applications. Farkas A. Margolis J. 1987. Trans IEE Japan 108A:397-404.1988. Kasahara T. Matey G. Rahamarimalala V. 1996. IEC. Publication 85. Trans IEE Japan 115A:1284--1293. 1987. Masui M. Electronic Properties of Materials.Kaneko R. 1984. IEC. 1984. Vienna. Polymers 20: 1186-1 187. 1982. Trans IEE Japan 116A:170-176.06. Report 620. Crichton BH. Hummel R. Fischer P.08.1987. Vienna. IEEE Electr Insul Mag 3349-858. IEC. Marton K.1997. W. Engineering Thermoplastics: Properties and Applications. Geneva: International ElectrotechnicalCommission. Publication 171-3. Gorur RS. Du B. Publication 37 11. Rasmy A. 1995. Berlin: Springer-Verlag. Ross R. Guide for the Determination of Thermal Endurance Properties of Electrical Insulating Materials-General Guidelines for Ageing Procedures and Evaluation of Test Results. Oxford.

1984. 1987. 1980. 1987. Steffens H. Noto F. Moscow: Izdatelstvo Mir. Sletbak J. . Sletbak J. 1994. IEEE Electr Insul Mag 5(10):29-36. Seely S. 1982. Proc Inst Electrostat Japan 6:72-79. IEEE Electr Insul Mag 3(1):8-12. Montreal. IEEE Electr Insul Mag 3(4):41-42. Yos~mura N. Poularikas A. ed. X u J. New York: Marcel Dekker.Samm R. 1979. Electricaland Radio Engineering Materials. 1979. IEEE Electr Insul Mag 2(6):39-40. Record of IEEE International Symposium on Electrical Insulation. Nishida M. Schaible M. Tareev B. Ildstad E. Boggs S. Electromagnetics-Classical and Modern Theory and Applications. pp 29-32. 1986. IEEE Trans PAS-98: 1358-1365.

subtransmission feeders. The substation design aims to achieve a high degree of continuity. Egypt The substation or switching station functions as connection a and switching point for transmission lines. Substation busbars (or buses) are amost important part of the station structure since they carry high amounts of energy in a confined space and their failure wouldhavevery drastic re~ercussionson the continuity of power supply. 11 5. and 765 kV. and to meet these objects with the highest possible economy. The substations used in distribution systems operate at voltage levels up to 69 kV. 230. 500. These are referred to as ultrahigh voltages (UHV). Voltage classes used for major substations include 69. and transformers. In this chapter we deal with the conventional types of substation where atmospheric air is the insulating medium.Higher voltage classes. 161. Giza. maximum reliability.considered as extrahigh voltages (EHV). and flexibility. generating units. are in various stages of planning or construction. Therefore. the bussystem must be built to be electrically 7 . and 345.ig h-Voltage Cairo University. and 287 kV. including l100 and 1500 kV. considered as high voltage. 138. Transmission substations serving bulk power sources operate at voltages up to 765 kV. Cas-insulated switchgear (CIS) is discussed in Chapter 10.

\ ~ j s ~ o n n e c t i n gSwitch US \ re 9. location of connecting lines. 9. (b) double bus withdouble breaker.1 Single busbar arrangement. simplicityof relaying. The single bus is the one in common use and has the simplest design. first cost. and (Q breaker-and-a-half with two main buses. (d) main and transfer bus. The single-bus design (Fig. reliability. Most substations conform to one or the other of the following basic arrangements as a result of different factors: (a) single bus. It must have adequate capacity to carry all loads. (c)doublebuswithsingle breaker. The choice of the arrangement to be used depends on the relative importance assigned to such items as safety. and provision for expansion. ease of maintenance. available ground area.1) is not normally used for major substations. flexibilityof operation. It is generally used in small outdoor stations having relatively few outgoing or incoming feeders and lines. (e) ring bus. .flexible and reliable enough to give continuous service. ease of rearrangement. Dependence on one bus can cause a serious outage in the event of bus failure. and robust construction to withstand foreseeable abnormal electromechanical forces.

2 Double busbarswith double breakers. Normally. 9. 9. it allows transfer from one busto the otherwithout deenergizing the feeder circuit. only the bus disconnecting switches need then be operated. It requires two circuit breakers for each feeder circuit. and each circuit includes two disconnecting switches.The double bus with double breaker design (Fig. A bus-tie circuit breaker is connected to the main buses. This scheme (Fig. The main-and-transfer bus (Fig. When closed. which presents a high order of reliability.4) is used for most distribution work. 9. One circuit breaker serves each circuit. . The transfer bus is a standby for emer- 1 1 BUS2 ure 9.3) uses two main buses.2) is used with large switching stations. each circuit is connected to both buses.

I Double busbars with single breaker. I ine line line US l ur Main and transfer busbars. .

The advantageof the ring bus scheme (Fig.5) is that it requires the use of only one circuit breaker per circuit. Two circuits are connected between the three breakers. whereas the double busbar with single breaker is more commonly used above 70 kV. Under normaloperatingconditions all breakers are closed and the two main buses are energized. To trip a circuit. The single busbar arrangementpredominates up to '70 kV. the two associated circuit breakers must be opened.7 rY It has been found that the two commonest busbar arrangements at all voltage levels are the single busbarandthedoublebusbar withsingle breaker. Also. 9. . This system is economical in cost. This scheme (Fig. and very flexible. 9. hence the term breaker-and-a-half. a bus-tie circuit breaker is provided to tie the main and transfer buses together when the need arises.gency use. reliable.6) has three breakers in series between the main buses. each outgoing circuit has two sources of supply. In addition. .

Usually. using copper or aluminum bars or tubings as the phase conductors. The ring bus and the breaker-and-a-half arrangements have been gaining popularity among designers for voltages above 160 kV.Breaker-and-a-half with two main buses. Those of higher voltages use strain insulators and stranded aluminum (ACSR) or copper conductors. Substations of the EHV class are usually of the strain busdesign. low-voltage and medium-voltage busbars are of the rigid bus type. with pedestal-type insulator supports. edium-voltage substations normally use the rigid-bus approach and .

As aluminum has several advantagesover copper. Factors to be considered when making this choice are voltage drop. 3. 2. 4. ice and wind loads. up to 765 kV. They combine high strength and good conductivity.enjoy the advantage of low station profile and ease of maintenance and operation. It usually needs more supports and insulators.whichmaylead to possible damage. . It usually requires more ground space than the strain bus. These factors are determined largely by the type of installation and the size of electrical load to be carried. are mostwidely used in HV and EHV outdoor substations. the aluminumbuswouldhave a 33% larger cross-section than the equivalent copper bus. Several factors determine the proper choice of busbar conductor material. Heattreatable aluminum alloys. Combinationsofrigid and strain bus constructions are sometimes employedin conventional arrangements. The rigid conductors are not under constant strain. 4. which provides good visibility of the conductors and apparatus. It is comparatively expensive. The busbar materials in general use are aluminum and copper. 2. The advantages of the rigid bus compared with the strain bus design are as follows: 1. The rigid bus has a low profile. due to the higher cost of tubing and connections. The disadvantages of the rigid bus design are: 1. Aluminum is one-third the weight of copper for a specified length. current-carrying capacity. and conductor corrosion. low-level structures. It is moresensitive to structural deflections. The resulting deflection for the copper bus is about 31YO greater than for the aluminum bus. power loss. SO as to benefit from both types. short-circuit loads. The pedestal-type bus supportsare usually moreaccessible for cleaning. For agiven current rating and for equal temperaturerise. most rigid businstallations use tubings of aluminum or its alloys. Aluminum and its alloys require little maintenance. especially in tubular shapes. The rigid bus design employs less steel and simple. 3.

by e~periment.8). which in the case of clearances between phases are usually different from the case of clearances to ground. These relations can be used to estimate the 50% breakdown voltages of such air gaps under switching and lightning impulses. between parallel conductor bundles. There the minimum phase-to-phase clearances are usually taken about 15% longer than those to grounded metal frames. Based . Moreover. the stressbetween the phases is greater than the stress to ground. In particular. In fact. on the other hand. For rod-to-plane and similar gaps. se- The i~sulationlevels for systems and equipment withvoltages rated at 245 kV and above are combinations of two components: the rated switching impulse withstand voltage and the rated lightning impulse withstand voltage (Section 18. the behavior of air gaps under switching impulse is affected by electrode shapes. between conductors and planes. and between conductors and portal frames. In certain cases larger phase-to-phase clearances have been preferred. The ~ n i m u m phase-to-ground and phase-to-phase air clearances in substations are prescribed to guarantee the withstand levels of switching and lightning impulses considered necessary by the system designer (see Chapter 14). very different characteristics(see Chapters 4 and 14). 1976).3). as dictated merelyby the requirements of equipment maintenance. The results have been expressed by empirical relations and can also be predicted by a generalized formula (see Section 4. positive impulse flashover voltages are lower than those with negative polarity. air insulation is predominantly dictated by lightning overvoltages. Numerous tests have been carried out in high-voltage testing stations in many countries in dry air and under simulated rain conditions for air gaps between rods and planes. to be about 6% for switching surges and 3% for lightning surges. The standard deviations 0 of these breakdown voltages were observed. Rain has no significant influence on positive impulse flashover voltages (Rizk.For safety and reliability it is essentialthat adequate clearances to live parts be provided. The air clearances between phase conductors and between them and grounded metal frames can be estimated by knowing the dielectric strengths of such air gaps under AC and impulse voltages. The dielectric stresses in the air gaps between phases and to due to switching overvoltages have. in substations up to 200 kV.

83 3.60 2. the phase-to-phase overvoltage peaks are sub9. Dependent on the Rated Switching and Lightningl ~ p u l s e ~ i t h s t a n Voltages d hing Rated ulse and withstand Air impulse voltage ( W 650 750 850 950 1050 1175 1300 1425 1550 (m) 1.35 1. On the other hand.83 4. three main categories of phase-toground clearances are present: 1.14 2.87 ( W 750 850 950 1050 1175 1300 1425 1550 1800 1950 2100 2400 (m) 1. to carry out maintenance work while neighboring conductors are alive.1 ~hase-to.on the evaluated 50% impulse breakdown voltages of air gaps. in some cases.round Air clearance.37 2.15 1.38 .28 3. Distances between conductors and portals istances between live parts of apparatus and portals istances between conductors and ground Table 9.1gives the air clearances according to the recommended rated impulse ~ i t h s t a n d voltages.55 3. In substations. their impulse withstand voltage Vw can be estimated by the formula This withstand voltage represents the level at which the probability of flashover is less than 10%.16 2.45 1.56 3.73 1.92 2.79 2.55 1.64 4. The distances between different live conductors are determined not only by their voltages but also by the necessity.07 3. Lightning surge stresses between phases will n o ~ a l l y be within the magnitude of stresses to ground.24 4.

Figure 9. These clearances were calculated from the 50%-interphase flashover voltages. Clearances between conductors 2. . rod-to-plane gaps. depending on the wave shapes of the two phase-to-ground components.8 and 15. Therefore. only this method allows testing of phase-to-phase insulation under actual geometrical substation conditions without causing undue flashover on phase-to-ground insulation. the degree of inhomogeneity of the electric field varies with the voltage level under consideration. The air clearances in substations may be generally identified as: l.2 gives the recommended minimumclearances between phases as represented by the rod-rod and ring-ring configurations.7 shows the test results for the 50% flashover voltages using two synchronous ( A t = 0) switching impulses with equal crest values U1and U2. Clearances between conductors and apparatus 3. It has beenrecommended that phase-to-phase insulation tests be carried out with two equal switching impulses of opposite polarities. and the timedelaybetween their crest values representing expectedfield conditions. For a rod-plane gap. In the EHV range.7 shows that the lower values are obtained with the positive polarity at the rod. no polarity effect is observed. where the field concentration is highest.2). possibly having opposite polarities. equal crest values. The crest value of each impulse should be half the rated switching impulse withstand voltage between phases. A s mentioned above. Figure 9. the detemination of the necessary interphase clearance is more related to the results for the rod-rod configuration in the lower voltage range. Besides being representative of the real stresses. Rod gaps. the ring-ring configuration becomes more and more representative because of the sizes of the hardware used at such voltage levels. Since the electrical field distribution is symmetric for ring and rod gaps. Clearances between poles of the same phase Most electrode configurations are characterized by highly inhomogeneous and sometimes symmetric fielddistributions. and on the ratioand timedelaybetweenthesetwo components (El-Morshedy. and gaps between ring-shaped electrodes have generally been accepted as representing practical geometrical conditions typical for the high-voltage and extra-high-voltage (EHV) ranges. Table 9. 1982). The time to crest for the positive component has been found equal to that value which gives the lowest flashover voltage (Sections 4. The 50% flashover voltage ofphase-to-phase air clearances isdependent on the wave shapes used in laboratories for the two phase-to-ground components.jected to considerable variations. their times to crest. and synchronous peaks. however. Only for nonsymmetric electrode configurations is the influence of the polarity remarkable.

3 (IEEE Working Croup. Above this temperature oxidation increases rapidly and would give rise to cumulative and excessive heating at joints and contacts. and joints. under two simultaneous equal and opposite switching surges U.7 Fifty percent flashover voltages of a gapd between conductorsof the shape shown. Some trends in the design practices of North American 500 kV and 750 kV AC substations are indicated by the clearances in Table 9. such as the type of material used and the bus's size and shape . bus connections. Buses are generally rated on the basis of the temperature rise that can be permitted without danger of overheating equipment terminals. Many factors influence the heating of the bus.1 and 9. In the design of buses the temperature rise of conductors above ambient while carrying current is very important.2. NEMA. 1970. 1983). The permissible average temperture rise for plain copper and aluminum buses while carrying their normal loads is usuallylimited to 30°C above an ambient temperature of 40°C. It is noted that the minimum phase-toground and phase-to-phase clearances are greater than those recommended in Tables 9. This value is the accepted standard for IEEE. strain.ure 9. 1969. This is the overall temperature rise and a maximum or hot-spot temperature rise of 35°C is permissible. and U2(Af = 0). or a combination of both bus constructions are in use. and that in EHV substations rigid. and ANSI.

.

E I g c c 01 T - Tk? O m b .a" .

For a circular conductor. and A. . watts (9.Themaximum continuous current-carrying capacity is important in selecting the proper conductor material.e. especially for AC use. R the conductorAC resistance (Qfm). With the arrangement of multiple bars. Figure 9.(Conway. each with a cross-sectional area of25.8cm2(Sirnpson and Greenfield. size. The ampacity of a given busbar can be estimated using the following simple relation: r2R= (wC + WJA.8 shows comparative ampacity ratings at power frequency for various conductor arrangements. 1971). the thick single bar is impracticable for high currents. For large currents it is better to provide several parallel on-edge flat bars than to use a single bar of equivalent total area and the same height. This means that fora single shapethere are limits to the current-carrying capacity withrespect to its dimensions. current-carrying capacity) of a busbar is determined by its ratioof surface area (for heat dissipation) to itscross-sectional area. Pcheat dissipated by convection (W/cm2). the proximity and skin effectsshouldbe considered in estimating their AC losses (Weiss and Csendes..2) where I is the conductor current (A). surface area (cm2 per meter of busbar length). the skin effect causes its AC resistance to exceed its DC resistance. Evidently. ? V r heat dissipated by radiation (W/cm2). their ratio being Comparative ampacity ratings for various arrangements and shapes of busbar conductors. and shape of cross-section. 1982). r-r The ampacity (i.1979).

Evidently. aluminum conductors are affected less by skin effect than are copper conductors of similar cross-sections. tubes have lower AC resistance than solid or Bat conductors of the same net cross-section. Tests have shown that aluminum conductors dissipate heat at about the same rate as copper conductors of the same outside diameter when the temperature rise is the same. Because of the higher resistivity of aluminum.2).73 x 1 0 . This method is generally applicable to both copper and aluminum conductors.. pr the relative permeability. 1982).3).5 for average oxidized copper For other conductor shapes.where l z is a constant listed in tables (Weiss and Csendes. . d is the outside diameter of an equivalent circular conductor of the same surface area. The heat dissipated by convection WCand radiation ? V r can be estimated from the following relationships WC= 5. it is then possible to determine I from equation (9. and p resistivity (Q/m2). . A more elaborate method isfollowed for computing the proximity effect.4) Wr = where P v Ta = average absolute temperature of conductor and ambient air (K) To = absolute temperature of the ambient (K) = surrounding air velocity (m/s) p= l . as seen from equation (9. 1984).~ . f is the frqeuency (Hz). using the same basic physical phenomena (Salama and Hackam.0 for black body and about 0. By calculating (Wc + W r ) A. and R. p a 0 T:*123& W/cm2 (9. Tubes with thin walls are affected the least by skin effect.0 for atmospheric pressure) = pressure in atmospheres ( d = outside diameter of cylindrical conductor (cm) T = absolute temperature of the conductor (K) a 0 = temperature rise of conductor above ambient (T To) (K) E = relative emissivity of conductor surface and aluminum -1.

For AC the value of If obtained should be multiplied by the ratio (RAc/RDC)-1’2. its maximum allowable temperature. If the differential term is set to zero. The energy stored in the busbar equals the net heat transfer into the bus plus the heat generated due to Joulean heating. An adequate conductor sizeisnecessary for carrying the fault current caused by an external short circuit without its temperature exceeding the safe limitduring the period from the instant of initiation of the fault until its interruption. 1971). For the case of copper busbars. the resulting algebraic equation governs the steady state ampacity of the busbar. and A the sectional area of the busbar (mm2).45& A bus conductor carrying a short-circuit current is subjected to electromagneticforces that tend to pull adjacent bars to each other if they carry currents in the same direction and push them apart if the currents are in opposite directions (Fig. 9. t the time (S) from initiation to clearing of the fault.. the following equation was suggested If = A where If is the DC fault current. IcA 6. The temperature changes gradually as the metal stores energy due to its thermal capacitance.9). This is becausethe bus temperature will not change instantaneously when the current changes. applying the law of conservation of energy. In fact. the differential equation is used to determine the relationship between the busbar temperature and the current it carries. This force increases quadratically with the .The current-carrying ampacity of a busbar element is limited by the maximum operating temperature. the value of coefficient K is about 70% higher. A differential equation relating the current to temperature was derived by Coneybeer et al. 1994). depending on the conductor material. The net energy into the bus is the absorbed solar energy minus the convective loss and the energy radiated to the surroundings. (1994). transient ratings better reflect the thermal performance of a busbar than steady-state values. K is a coefficient ranging between 600 and 750. For both copper and aluminum bus connectors carrying DC. Neglecting the internal thermal resistance of the busbar. and the ambient temperature (Simpson and Greenfield. Thus the transient ampacity of the busbar is obtained (Coneybeer et al.

These forces are far greater than their ordinary loading caused by the conductors' weights and wind. The electromagnetic force on either of the two conductors carrying currents of instantaneous values 1 . As a result of the electromagnetic forces between the bus conductors there will be a mechanical reaction on the bus supports. The insulators should have enough mechanical strength and resilience. including mounting structure and insulators (Craig and Ford. 9. 1 (Fig. and 12 and separated by a distance d can easily be evaluated as Whether both forces are those of attraction or repulsion depends on the relative directions of the currents 1 . Awad and 1980).~agneticfields and electrodyna~icforc~s acting on two parallel conductors carrying currents in the same direction (a) and when the currents are in oppositedirections (b).9). . and . 1980. current magnitude and depends on the shape and arrangement of conductors and the natural frequency of the complete assembly.

and the corresponding electrodynamic force acting on the busbars. one phase current gets interrupted first and thus a three-phase fault turns into a double-line fault.10). if the X / R ratio is 20. depending on the instant the short circuit occurs with respect to the AG voltage wave and the X / R ratio of the faulted circuit. The maxi mu^ i n s t a n t a n ~ o ~ s short-circuit lateral forces acting on busconductors in various arrangements are listed in Table 9.183 times the sustained rms fault current. during the operationof a circuit breaker clearing a short circuit. For example. the component at power frequency dominates for the first few cycles. the first current peak will be about 2. 9. Phase-to-ground and double-phase-toground faults are more likely.The short-circuit current may reach excessive peak values. A bus installation should be designed so that its natural frequency of oscillation. This causes an offset in the current wave (Fig. to evade resonant pulsations that might otherwise endanger the mechanical structure. Also.but afterward the dominant component is at double the power frequency. is not close to the power frequency or its multiples.10 produce force pulsations. Force Asymmetrical short-circuit current in a circuit with aX/R high ratio. The electromagnetic forces set up between parallel conductorsin which current varies as in Figure 9. yet it should be considered in the design of busbars. .4. itu There are considerable differences between short-circuit forces. or thatof any of its parts or supporting structures. depending on the type of fault and the bus conductor arrangement. The probability of a short circuit occurring on all three phases is very remote.

T- II T- II U U .

With the developments in polymeric materials.. The in~uence of the magnetic field constraint on the optimal interphase spacing decreases as the cost of land increases (Anders et al. the magnetic fieldconstraint is negligible.. 1979). The abnormal loadings include electrodynamic forces occurring during short circuits on the busbars. the more important the magnetic field constraint becomes. Also. and normal and abnormal loadings. With a permitted magnetic field strength of 15 T. as described above. Acceptable electrical and other qualities of insulators are usually set by national and inter~ational standard specifications (see. These insulators have traditionally been made of special porcelain or toughened glass.ecently. if a strain type. (1994). e. The effectof magnetic field constraint on the optimal designof a 500 kV rigid bus substation was examined by Anders et al. traditional insulators have for severaldecadesbeenexperiencing serious competition from insulators made of epoxy reinforced with glass fibers. Insulators for outdoor use should pass h~gh-voltage tests under simulated rain conditions. . insulators to be employed in sites of significant atmospheric pollution should be designed and tested accordingly. suspended by insulators.g. The inclusion of the magnetic field constraint in the station design procedure will produce a tendency to decrease interp~ase spacing in comparison with the standard design. lymeric insulators usuallyweighless than half or even one-fourth of ir corresponding porcelain insulators for the same mechanical and elecengths (Karady and Limmer. as discussed below. sbar insulators should have adequate electrical and mechanical strength under the expected ambient and operating conditions of temperature variations. 1979). regulatory guidelines have been set in several countries limiting the maximum level of magnetic field near transmission substations to which the public can be exposed. The larger the magnitude of the current flowing in the buses. rain. 1994). IEC. Air-insulated busbars are supported if rigid. pollution. The optimal interphase spacing decreases by a factor of 2 when the current magnitude increases bythe same factor. or. Several regulatory authorities have imposed a limit of 10T on the allowable magnitude of magnetic fields at the edge of rights of way or substation property. using a mathematical model.

The voltage distribution over the chain comes to be far from uniform (Aoshima et al. Measures usually taken to improve the insulation strength in polluted atmospheres include using “anti-fog’’ insulators with extendedleakage pathsand insulators with semiconducting glaze. a conducting path. the wet contaminated areas of the insulator surfaces dry up gradually and the partial arcs become extinguish~d. units.” “Long rod” insulators can replace chains of short ““cap-and-pin’. With nonuniformities in pollution and leakage currents naturally occurring over insulator surfaces. getsestablished on the insulator surface all the way from the high-voltage conductor to the grounded metal frame.~ o n g . Rain does not always provide the cleaning needed to combat pollution. To help improve system performance. Houlgate et al. thus maintaining power supply co~tinuity. In such a case. 1982. hot spots form. 1982).. particularly under conditions of heavy marine or industrial pollution or when rain is not regular enough. 1988. with its drastically increased resistance. takes a major fraction of the total insulator voltage. pollution monitors have been developed so that the insulator cleaning operation would be carried out only when necessary ( ~ h a l i f a et al. Alternatively. however resistive. 1988). Combating pollution of high-voltage insulators has also been achieved successfully by regular washing of the insulators while under high voltage. 1981). Thus leakage currents flow. Such a dry band. greasing their surfaces prevents the formation of continuous conducting films. more insulator units are stacked or connected in a chain.. Intensive research has been going on in many countries for several decades (Khalifa et al. When pollution accumulates on ins~lator surfaces and its soluble ingredients dissolve in the moisture accumulated from fog or dew.. With partial arcs burning on one unit of an insulator chain.. A partial arc can strike across the band. whereas for greater mechanical strength more insulator chains are bolstered together “in parallel.For the higher operating voltages. Pei-Zhong and ~ h e n g . 1982).. A localized spot dried up by concentrated leakage currents pushes the current paths sideways and thus extends laterally to form a dry band. desert pollution is a case in point (ElKoshairy et al. This may eventually develop into a complete asho over of the whole insulator chain under normal operational voltage. additional voltage stresses appear over other units. . the current through the partical arcs maybe too small to maintain them.

MB. Proceedings of the American Power Conference. El-Morshedy AMK. 1979. RAC/RDC this fault be sustained while the temperature rises to 300°C above 50°C ambient? Assume k = 700 for this alloy. Nosseir A. Limmer H P . IEEE Working Group.4 cm2. Publication 168. 1979. Black WZ. CIGRE paper 33-01.102513-520. IEEE Trans PAS-89: 1521-1 524.pp 11201125. Aoshima Y. 1970. Craig B. Awad M. 1994. Geneva: International Electrotechnical Commission. IEEE Trans PAS-98:1384-1402. Awad. 1982. Ford GL. 1981. Anders GJ. Houlgate RG. . Karady G. Khalifa M. IEEE Trans PAS. Conway BJ. the surrounding air velocity is l0 m/s. Proc IEE 135C:24-30. 1982. IEC. 1983. IEEE Committee Report.08. Huestis HW. El-Sharkawi E. Horrocks DJ. (3) It is required to select a bar of a certain alloy that will withstand 100 kA fault current if the top temperature limit is 250°C and ambient is 70°C and maximum time is 40 cycles. Zarzoura H.(1) Give a single line diagram of connections showing the layout of the busbar sections in a power station.1979. CIGRE paper 33-09. 1980. Keiji K. Ford GL. IEEE Committee Report. 1980. IEEE Trans PAS-88:854-861. IEEE Trans PAS-99:434"442. Bush RA. 1982. IEEE Trans PAS-100:948-955. Habib SEP. El-Morshedy A. the conductor temperature is 130°C and the temperature of the surroundings is 40°C. area 35. IEEE Trans PWRP-9:1822-1829. (2) A 15 cmx 15 cmx 0. Roberts WJ. Proc IEE 129:199-205. Tatsuya H. is 1.08and k = 650 for this alloy. El-Koshairy MAB. (4) What are the advantages and disadvantages of the rigid bus design? (5) Calculate the heat dissipated by convection and radiation fora cylindrical busbar of 5 cm outside diameter. IEEE Trans PAS-99:480487. assuming R A C / R D C is 1.6 cm square tube of a certain alloy. 1994. (6) Describe the process leading to pollution flashover. Include the necessary isolators and circuit breakers for at least two alternators and two feeders. 1988. Tests onIndoor and Outdoor Post Insulators for Systemswith Nominal Voltages Greater than 1000 V.8. Coneybeer RT. IEEE Trans PWRP-9:1384"1390. Gouda QE. The relative emissivity is 0. Khalifa M.Lambeth PJ. The pressure is 1 atm. 1969. Discuss the methods which may be usedto reduce the probability of pollution flashover in polluted locations. How long can is to withstand 150 kA fault current.

Weiss J. New York: The Aluminum Association. 1982. Csendes Z. Salama M.Electrical Conductor Handbook.Pei-Zhong H. Rizk F. GreenfieldEW. IEEE Trans PAS-103:1493-1501. Simpson TW. Cheng-Dong X. IEEE Trans PAS-101:3796-3803. 1971. CIGRE paper 33-07. 1982. . IEEE Trans PAS-95:1892-1900. 1976. 1984. Hackam R.

This Page Intentionally Left Blank .

It possesses a high degree of operational reliability and safety to personnel. 4. or desert climates. and this ratio increases wtih increasing pressure. 5 . on unstable ground or in seismically active areas). t y . 3 . At 3 atm. The space occupied by the switchgear is greatly reduced.. as it offers the following advantages: 1. In addition to . It is easier to install in difficult site conditions (e. 6 At atmospheric pressure. Egypt Gas-insulated switchgear (GIs) that uses compressed sulfur hexail~oride (SF6) gasovercomes many of the limitations of the conventional opentype HV switchgear. 2. It is totally unaffected by atmospheric conditions such as polluted or saline air in industrial and coastal areas.~ _ _ _ _ _ _ _ Cairo ~ ~ i v @ r s ~ Giza. the dielectric value is about the same as that of transformer oil. SF6 has the advantages of being nontoxic and nonilammable. SF6 has a dielectric strength two to three times that of air.g. In addition to having a dielectric strength much greater than that of air.

143 cal/g 5. It has a high partial vapor pressure at both normal and low temperatures.05 -50. It is noncorrosive. 5.0"C 0. SF6 gas exhibits excellent properties for arc quenching and is therefore also used as an interrupting medium in circuit breakers instead ofoil or air. It is nontoxic. Only when SF6 gas is heated above 500°C will it decompose.63 dyn/cm 3. Table 10. SF6 has the following basic physical and chemical properties: 1. It does not react with other materials.164 g/l 45.98 g/ml 1. however. 1981). and then its decomposition products react with hydrogen and other materials. Its high molecular weight and low viscosity enable it to transfer heat by convection more effectively than do common gases. The useofSF6 in circuit breakers isdiscussed in Chapter 11.329 g/ml 6.62 bar Source: Maller and Naidu (1981). cause suffocation if it displaces oxygenfrom an area because of its much higher density than that of air. Furthermore.8"C -63. 3.755 g/ml 11. With special filters.9"C 1. it is possible to alleviate this effect (Maller and Naidu. as it is inert.557 atm 0.10 10. having almost the chernical properties of a noble gas. It is nonflammable. .2 Anis this superior dielectric property. however.36 (cal/s)/(cm*K) 1. It has excellent heat transfer characteristics. 2.1 Physical Properties of SFs - Molecular weight Melting point Sublimation temperature Density (liquid) at 50°C at 25°C Density (gas at1 bar and 20'C) Critical temperature Critical pressure Critical density Surface tension (at -50°C) Thermal conductivity (x104) Viscosity (gasat 2 5 0 x ~ lo4) Boiling point Specific heat (at 30°C) Relative density (air = 1) Vapor pressure (at 20°C) 146.6"C 36. it inhibits surface erosion and oxidation. 4.1 summarizes some of the basic physical properties le 10. It can.61 poise -63. Furthermore.

The advantages of the three-phase common enclosure design are: 1. This division enables the isolation of one compartment for maintenance or repair purposes while the other compartments remain pressurized.2 and 10. 4. owing to ionization of the gap. or of the single-phase type. The absenceof complicated tie rods and linkages between poles for the circuit breaker. In the case of a three-phase common enclosure. .3 show sectional views of the general arrangements of two different gas-insulated switchgear bays belonging to the ranges 145kV and 220-800 kV. 2. grounded casing for the entire switchgear. Consequently. as in Figure 10. The implementation of the foregoing arrangement into a real CIS depends on the voltage level. and grounding switch drives simplifies the drive system. as in Figure 10. or air increases the dielectric strength of the latter substantially.l a the single-line diagram of a double-busbar arrangement is shown. clearances between phasesand phase-to-ground) the resultant field stress in a common three-phase enclosure is approximately 30%less than that in a single-phase enclosure and hencelesslikely to cause failure. For the same parameters (voltage level. Mixing a small amount of SF6 gas with a relatively inexpensive gas such as nitrogen. within a few milliseconds. and at the same time the phase-to-ground arc will extinguish.3. Figures 10.3 of SF6. evolve into a phase-to-phase fault between conductors. In GIS all live partsare enclosed in a compressed-gassystemwhichis divided into a number of compartments.2. an enclosure burn-through is not possible. earthing switches. carbon dioxide. In Figure 10. A smaller number of enclosures is required per feeder (one-third). an arc between phase and ground will. and current and voltage transformers. disconnectors. hydrogen. conductor size. Basic components that make up any one CIS bay are the circuit breaker. 3. In Figure 10. respectively. busbars. The GIs enclosure forms an electrically integrated.l b the diagram is redrawn in the form of a typical gas CIS circuit breaker bay. isolator. The GIs enclosure can be either of the three-phase type.

8. the use of steel for single-phase enclosure GIS is. for example.nis B . 0th aluminum and steel are used for SF6 GIs enclosures. In the case of a three-phase or single-phase enclosure high-pressure GIs. Both materials fulfill specific purposes. However. insulation requirements necessitate the use of single-phase enclosure types. However. limited to lower current densities because of the problem of electrical losses and heating in the enclosure resulting from induced circulating and eddy currents. DISCONNECTOR la) ure l ~ . . Above that level. three-phase enclosures are usedonly for voltagesbelow about 200 kV. steel has themajor disadvantage that it cannot be shaped and welded into the same homogeneous form as can be achieved with aluminum castings to ensure optimum field stress distribution for a given gas pressure and conductor configuration. (a) l Single-line diagramof afeeder bayof a double-busbar system.

ure 10. high-speed grounding switch.3. maintenance grounding switch. gas connector.5. density monitor. current transforrner.10. circuit breaker. voltage transformer.7. 6.1 ( c o ~ t i ~ u @ (b) ~ )diagram of a typical GIS feeder bay showing gasfilled compartments. 9. disconnector.busbar i. 4. 2. 1. 8. busbar ii. .

While for a rate voltage of 72. A higher SF6 gas pressure results in higher dielectric strength. the dielectric strengh increases at a lower rate than thegas pressure. ith respect to the SF6 insulating gas pressure. 1. which operates at 1. 6. c o ~ b i n e d switch. On the other hand.Sectional viewof a typical SFs gas-insulated. for a given voltage level the required conductor spacing is reduced. high-speed grounding former. 145 kV circuit breaker bay using one enclosure for the three phases. 4. at higher voltage levels the advantages of high-pressure GIS become dominant. voltage transdisconnector/grounding switch.2. resulting in a more compact design. 3. busbar with combined disconnector grounding switch. current transformer. The reason behind this is the increasing sensitivity ofSF6 insulating gas to the roughness of conductor and enclosure surfaces and to contamination at higher gas pressures. 7. 5. there are two signs: high-pressure CIS operating at about 4 bar (405k pressure GIs. which in turn can reduce the reliability and increase the require- . cable-end unit.2bar absolute. circuit breaker.5 kV and below the low-pressure CIS is the more suitable design. ~onsequently.

busbar disconnector. 4. voltage transformer. 3. 5. 6. SF6/air brushing. busbar. current transformer. 8. so are the requirements for a homogeneous electric field distribution if the higher pressure is to be fully utilized. for voltages of 110 kV and above there is also a limit to the SF6 insulat- . Consequently.9.Insulation - . for higher SF6 gas pressures it becomes increasingly necessary to ensure careful homogeneous shaping of conductors. 7 . high-speed grounding switch. By the same reasoning. 1. 2. circuit breaker. disconnector. and enclosures. A s the gas pressure is increased. components. From the above it canbededuced that there is a limit to the SF6 insulating gas pressure above which the economies of a design decrease as the disadvantages of the higher gas pressure dominate. ment for servicing and maintenance.3 Sectional view of a typical SF6 gas-insulated EHV (220-800 kV) bay using single-phase enclosures. maintenance grounding.

Thepresence of a solid insulating material in the compressed SF6 medium distorts the field distribution in the GIS duct (Stone et al. They take various shapes. On the other hand. . as detailed in Section 10. its dielectric strength) because of a number of factors. 2.3). a severe gas leakage with a pressure drop down to 1 bar will result in a reduction of the dielectric strength.nis inggas pressure belowwhich the increasing production cost for larger enclosures andgreater material expenditure as well as larger buildings make the design uneconomical.e. in the case of low-pressure GI§ with a rated insulating gas pressure of 1.3. The conductors of a CIS normally consist of aluminum tubes.pressure is the behavior of the GIS when a leak occurs. and mechanical stresses that arise during normal service and during short-circuit conditions. thesame leak will result in only a 15% deterioration in dielectric strength. together with supporting insulators.7. cers Solid insulators are used in gas-insulated apparatus for physical support of high-voltage conductors and for mechanical operations of the switchgear. must be properly designed to withstand the electrical. thermal.6. such as annular disks. due to aging of the solid material. In the case of high-pressure CIS with a rated insulating gas pressure or approximately 4 bar. Spacers can limit the operating gradient (in kV/m) of the system. and post supports. truncated cones.2bar. 1987). Another important consideration when selectingthe SF6 insulating gas .. Spacers can affect theshort-term behavior of theGIS (i. Several problems are introduced by the use of spacers in GIs: 1. Spring-loaded copper contact fingers constitute thefemale contacts andcopper plugs the male contacts. The contact surfaces are silver plated and the contacts are welded to the aluminum conductors. and therefore of the BIL. by about 75% of the ratedvalue. the diameter and wall thickness of which depend on voltage and rated current (§ections 10.6 and 10. The conductor system. A leakage resulting in a drop in the SF6 insulating gas pressure automatically means a reduction in the dielectric strength andtherefore the integrity of the switchgear..

GIS disconnector: (1) enclosure. bushings and cable-end boxes. (5) moving contact.3 and in detail in Figure 10. . surge arresters. (4) shielding for fixed contact. voltage and current transformers. and gas density monitors.(3) fixed contact. electrical interlocking against incorrect operation.4 They are equipped with copper contacts that are spring loaded to give the disconnector high electrical efficiency and high mechanical reliability. The main features are motorized or manual operation. asshown in Figures 10. otherwise a flashover to earth may occur. Disconnectors are made up from insulators. and conductors of different geometrical shapes to give an optimum layout. (9) driving motor.1 In additionto circuit breakers (discussed in Chapter 1l).2 and 10. enclosures. a GIS bay includes disconnectors.(6) rack-and-pinion drive. The operating mechanisms of the disconnectors and earthingswitches are of the same design for most GIs. (2) barrier insulator. and mechanically lockable end positions. (7) contact support.(8) insulated drive shaft. Disconnectors must be carefully designed and tested to be able to break small charging currents without generating too-high overvoltages.

5). 10.2 and 10. The fast-closing operation is achieved by means of a spring-closing device. three voltage transformers are placed in one enclosure. It is also possible to design a voltage transformer consisting of a low-capacitive voltage divider connected to an electronic amplifier. 10. The capacitance between the inner conductor and a concentric SPACER HIGH-VOLYAGE CONNECTION SHIELD W HIGH-V~~YAG EI N Q I ~ G SECONDARY ~ I N Q I N G CORE ENCLOSURE ure 10. the slow-operating earthing switch and the fast-closing (high-speed) earthing switch (see Figs. In three-phase enclosed GIS designs.3). The most commonly used voltage transformer is of the inductive type (Fig. but are operated only when it is certain that the high-voltage system is not energized.3 nis Two different types ofearthing switch are normally used. Slow-operating earthing switches are used for protection purposes when work isbeing done in the substation. The fast-closing earthing switch can close against full voltage and short-circuit power.5 G I s voltage transformer .

. In the single-phase enclosed CIS.(2) enclosure. 10. It is therefore possible to use arresters with lower sparkover voltages. (7) short-circuiting bar. Figure 10.6 GIS current transformer:(1)core unit. and the entire arrester is insulated with dry compressed gas. (5) secondaryconnection. varistors andspark gaps) but have very compact designs. which creates a highly consistent performance within tolerances. (3) cover.2). This designis suitable only for the highestsystemvoltages (Reisinger et al. In the three-phase enclosed CIS design the cores of the current transformers are normally located inside the enclosure (Fig. They are preferably placed outside on the SF6 bushings or on cables. (4) HV conductor. 1987). (6) insulation layer. The return current in the enclosure is broken by an insulating layer. SF6 gas-insulated arresters are based on the same active parts as conventional arresters (i. thus ensuring a completely undisturbed electrical field between the enclosure and the conductor.6 shows the single-phase enclosed arrangement. The spark gap elements are sealed off from the atmosphere.. which provides a closer margin ure 10..e.01 measuring electrode near the enclosure is then used as the high-voltage capacitor. the core of a current transformer is located outside the enclosure.

(7) insulating tube. exploiting the high insulation strength of compressed gases. Therefore. (4) bursting plate connection.In arresters with higher rated voltages the hood ' " l GIs-enclosed 120 kV arrester: (1)support and barrier insulator. necessary for linearizing the field distribution along the arrester elements. (6) metal cladding. (5) hood for field control.by including a metallic hood) (Fig. can be determined by fieldcomputations (Chapter 2).7 showsan arrester with a rated voltage 120kV.. This isachieved in their design(e.g. Thus capacitances to earth of live parts are far greater than in conventional arresters.7). The shape and dimensions of such a hood. In SF6 gas-insulated arresters the metallic earthed parts are much closer to live parts. (1 . (IO)ground connection. 10.nis for protecting the system insulation. (9) insula1) N2gas connection. The reduction can reach 10% as compared to conventional arresters. tion. extra precaution is neededto compensate the resulting nonlinearities in voltage distribution along the components of gasinsulated arresters. (2) SFs gas connections. (8) manometer. Figure 10. (3) HV connection.

High-voltage cables of various types are connected to the SF6 switchgear via cable-end boxes (Fig. andthebarrierinsulator withfemale plug contacts. be integrated into the GIs in any desired position. Bushings used for the direct connection of a CIS to a transformer may also be of the oilfilled type. The dielectric strength of the switchgear insulated with SF6gas and the breaking capacity of the SF6 circuit breaker depend on the density of the gas. the enclosure. SF6 gas-insulated arresters can. of course. lo. Overhead lines and all-air insulated components are connected to the SF6 GIS by air/SF6 gas-filled bushings. These bushings (Fig. Following are some of those features. which consist of the cable-end bushing with connecting flange.3) employ capacitive grading and are divided into two independent gas compartments by a barrier insulator. 10. During the past yearssome improvements havebeen introduced in the design and structure of GIs. For this. The space surrounded by the porcelain insulator is filled with SF6 gas to slightly above atmospheric pressure. The gas space on the switchgear side of the barrier insulator has the same SF6 gas pressure as the switchgear. a density monitor is employed. Since the pressure varies with temperature. . Oilfilled condenser bushings can also be used for high voltages. 10.lb) is monitored by its own density monitor. If the porcelain is damaged. this reduces the risk to a minimum.3 is supplemented with a number of grading metal rings fitted at certain points along the arrester’s active parts. it is the gas density that is monitored. depending on theprotection requirements.The pressure-tight bushing separates the SF6 gas compartment from the insulating medium of the cable. A completely dry cable terminationfor connecting XLPE cables to GIS has the advantages of smaller dimensions and better thermal properties.?. Each of the gas compartments separated by barrier insulators (Fig.

8 shows the layout of a GIs housing a gas-insulated transformer and gas-insulated shunt reactors. Recently. however.1 Gas-Insulated Transformers As shown earlier. gas-insulated transformers have been manufactured (Toda et al. This measures the current with the highest precision and without saturation across the entire operating range. They fall into two broad categories: 1. Measures taken to improve the cooling capability include raising the SF6 gas pressure. Gas-insulated transformers in the capacity range below 60 MVA which do notgenerate much heat and are therefore cooled by SF6 gas circulating. are being replaced by optical or electrical sensors.2 Voltage and Current Transformers Inductive instrument transformers which are heavy and large. cleanliness. Gas-insulated transformers enjoy all the advantages of GIS such as safety.4.nis 10. non-inflammability. The measured signals are processed after being digitized and then sent. and compactness. CIS substations normally do not encompass power trans- formers in the compressed gas complex. These are normally liquid-cooled gas-insulated transformers using perfluorocarbons (C8F160 orCgFlS). 1995). which are incombustible and have a high cooling capacity as a cooling medium. a Rogowski coil (an air-core ring coil) is used (Hogg et al. optimizing the layout of gas ducts in the windings to produce large gas flow. 10. which avoids any ferromagnetic resonance.. 2. The sensors are normally located in the outgoing circuit of . via the optical processbus (IEC 1375) to the bay protection and control level. The function of the traditional high-voltage current and voltage instrument transformers is performed by an advanced generation of current and voltage sensors combining both the functions in a single component. Figure 10. They are referred to as the “gas cooling type.9. For the current measurement. Such transformers with a rating of up to 300 MVA at 275 kV are now available (Toda et al. 1998).4. This has been mainly because the cooling capacity of the gas is lower than that of oil in oil-immersed transformers. Large-capacity transformers with a capacity above 60MVA..9. The voltage is measuredbymeansof a metal-enclosed capacitor voltage divider..” Transformers of this type normally serve medium-voltage systems (up to 77 kV). in serial form. 1995). and employinghighly thermal-resistant insulating materials to raise the temperature-rise limit of the windings.

1998). 10. (Courtesy of Toshiba.3 Drive Motors The trend now is for drive motors fordisconnectors and earthingswitches to be controlled electronically.\ Gas-insulated transformer Layout of a gas-insulated substation housing a gas-insulated transformer. and various measurements for monitoring are also realized with modern sensor technology.) the breaker and meet all the demands made on control andmeasurement as well as on state-of-the-art protection and revenue metering. The function of the auxiliary switchis performed by a sensor mounted directly on the operating rod. . 10. For this connection to the control and protection sytem (control cabinet). This version can be connected to numerical or digital protection systems.4 Control and Protection Control and protection signals are transmitted in digital form over fiber optic links..4. The measured analog signals are digitized before serial transmission over the optical process bus to the bay level (Hogg et al.9.4. density measurement. Drive control.9. there is a plug-ended cable which incorporates the fiber optic (bus) cable for analog and binary signal exchange as well as the auxiliary power supply for the drive mechanisms.

and the damping time constant of the resistor voltage.9. Thewideuseof sensors makes it possible to implement a large range of monitoringfeatures. disconnectors. Therefore. the energy consumed by the resistor. thus generating very fast transient (VFT) overvoltages. The disconnector was said to withstand the duty of suppressing very fast transient overvoltages due to restrikings at the time of switching charging currents. gas-insulated combination of circuit breaker. there is a need to suppress VFTs in such systems by employing resistor-fitted disconnectors.9(Tlamagata et al.The disconnector connects a resistor in series with the circuit in the event of restriking. contact displacement curve. the GIs disconnector causes repeated restriking. currentand voltage sensors. without needing mechanical contacts to connect the resistor. trend analysis of the gas density. and bushingmaybe located in a common gas compartment ( H o g et al. 1998). Although overhead transmission still represents the most economical solution to bulk power transmission.5 Integrated Compartments A metal-enclosed.6 Resistor-fitted Disconnectors Upon switching a charging current.9. and circuit-breaker conditions (pump running time. energy needs.is Conventional toroidal core current or voltage transformers are required for the electromechanical and solid-state protection relays. 10. A disconnector catering for the above needs was manufactured as shown in Figure 10. SF6 compressed gas-insulated cables present possibilities for underground transmission of high power with rated voltage above . With UHV G I s systems underway the trend is towards reducing the lightning-impulse withstand level.e. Such overvoltages can cause ground faults between disconnector contacts or at a bus contaminated by metallic particles (Ziomek and Kuffel. 199'7).. It met the limiting requirements of the rising speed of the voltage on the resistor. e. remaining lifetime. operations.). it is becoming more difficult to construct overhead lines in densely populated areas. earthing switch. 10. Interference with a low-voltage circuit control systemmay also develop. to only 2250 kV for 1000 kV apparatus.g. etc... such as self-checking.4. and underground cables are then required (Chapter 12). i.. Thehigh-voltage switchgear is limited to the minimum amount of equipment really necessary to guarantee the functionality of the bay or the substation for all typical configurations. and should be able to toso using movableelectrodes.4. 1996).

which also influences their range of application. " Tank 1 Stalionary electrode Resistor / t i S Insulating acer 1 (gad side) Typical resistor-filled disconnector in GIS. .Operaling ~echanls~ Insulating spacer 2 (power supply side) ~ovable electrode Shield on movable side Interelectrode discharge Resistor S h ieid r 1 . willbenecessarily increased. Additional active losses occur in cables and compressed gas-insulated cable systems due toloss currents in the metallic enclosure and to dielectric losses (Chapter 12). For higher voltages it will become increasingly more difficult to dissipate the heat losses since the insulation thickness. 1 The active lossesof a transmission linedepend primarily on the crosssection of the conductor. Conventional cables are adequate for rated voltages up to about l50 kV over short distances and withoutextremelyhigh rated currents.4"can hardly be contained within acceptable limits. Also. a 123kV. The application range of these cables can be expanded only by forced cooling. The rangeof application of compressedgas-insulated cables extends from short links to transmission of energy from cavern-type power stations over long distances. and hence the internal thermal resistance. The cross-sections of individual transmission media that canbe utilized inpractice are determined largelyby their active resistance. the dielectric loss increase with voltage-raised to an exponent of about 1. The compressed-gas cable is an attractive alternative in this case.

the transmission properties of a compressed gas-insulated cable correspond to those of three to four overhead lines operating in parallel (Chapter 14). Overhead lines are generally aluminum-stranded conductors with cross-sections of240"500mm2.10 shows a comparision of cross-sections that can be produced in practice. This relatively small range is. the total of reactive losses is also superior to those of compressed gas-insulated cables. a compressed gas-insulated cable changes from capacitive to inductive losses as loading is increased.nis 0 C 0 Comparative conductor cross-sections as functions of rated voltage: (a) overhead lines. Figure 10.. A point is passed where the ideal operating condition is reached (i. (c) conventional cables. increased by using bundled conductors for higher voltages. however. Therefore. Whereas conventional cables show heavy capacitive losses that only slightly change withloading. at zero reactive losses).e. which isparticularly important when the transmission distance is long.(b) compressed-gas cables. . The surge impedance load of compressed gas-insulated cables is three to four times smaller than that of overhead lines. Furthermore.

Forthe current example.3)give r = 48cm and R = 13. where particle contamination and surface roughness have adverse effects. a gas density of 60 g/l at -20°C).11.1) where V is the applied phase voltage. several formulas for the breakdown threshold have been published.3) kV GIs whose basic impulse insulation level ( For a 420 1425kV and which operates at 0. the maximum allowable value for E'). the electric field at the surface of the conductor is (10.g. As explained in Chapter 2. the field is maintained below 430 kV/cm for a typical system BIL of 1425 k'v if the conductor and inner duct radii are larger than 3. equations (10. corresponds to a maximum allowable withstand electric field. As shown in Chapters 2 and 12.11 (e. 0. Another important factorin the dimensioning of GIs ducts is that condensation of the gas should be avoided. Values obtained in this way are somewhat less than what should be used in practice.2 cm. the enclosure dimensions will be determined.4 MPa (10. According to Figure 10. for a givenminimumtemperature there exists an upperlimitof gas density abovewhichcondensationis sure to occur (e.3 and 9. Er is minimum (i..1)-(10.1 p 0.. in turn. This maximum allowable gas density.0cm7respectively.e.e. typically at the lowest ambient temperature.7ionization is least likely to develop) when m e m 2. 430 kV/cm at 60 g/l).2) If this relation is combined with knowledge of the breakdown threshold of the gas (i.35 MPa. One formula sets the threshold under positive impulse voltages at Etb = 8 1 .72 r R (10. For coaxial cylinders. 2 ~ 10 + kV/cm.g. Finally. l).. the electric field in the gas can be ensured not to exceed this maximum only if a lower limit is imposed on the duct dimensions according to equation (10. as shown in Figure 10.The dimensions of the enclosure are generally decided by considering the duct tobe a coaxial cylinder arrangement with a conductor of outer radius r and an enclosure of inner radius R. .

the material of which solid supports are usually made. and hence breakdown may. the Solid support insulators and spacers represent a parallel insulation. . in general. materialize in either insulation. The surface flashover strength of cast resin. in contrast. to the compressed gas insulation. and condensation temperature . At higher gas pressures. is about 30 kV/cm (ms). a flashover across an insulatoris likely sincethe dielectric strength of the gas is much increased while that of a solid insulator isnearly unaffected by increasing gas pressure. The relation between the applied phase voltage and the maximum field on a support in^ . T E of SF6. Underlow gas pressures a flashover across a solid insulator is highly unlikely to occur prior to a breakdown in the gas and therefore poses no design problem.31 nis ” ” ” ” - I 1 Relations between the impulse breakdown strength gas densityp.

. Several factors are known to have an effect on the dielectric strength of SF6 when used in an actual GIs. the maximum field on the insulator is related to the phase voltage by (10.7cm If the optimal relation between the inner and outer radii. the duct diameter may be required to be larger than that determined on an insulation basis to limit the temperature rise in the duct. 1987).2). the rated current increases much less than linearly withvoltage. R-r=9. the minimum dimensions of the system can be determined as r = 5. then. equation (10. The uniform field breakdown strength is larger .4) If Ernis equated to the reported dielectric strength of 30 kV/cm. the material of the electrodes in an SF6 gap begins to have an influence over breakdown. Under very high pressures and when the breakdown strength exceeds about 200 kV/cm. At lower system voltages. is considered. therefore. In the following sections the effects of those factors are discussed. more current can be passed in the system without excessive temperature rise.3 cm (10.5) While the GIs duct diameter computed on an insulation basis increases more or less linearly with rated system voltage.2 of its average valuealong the insulator (Stone et al. Efforts have been made to reduce the maximum fieldto only 1.6 cm R = 15.insulator depends largely on the insulator shape. for a 420 kV GIs. The heat generated by the “copper” losses in both the conductor and the duct wall will be dissipated to the atmosphere by radiation and convecy increasing the thicknesses of both the hollow conductor and the duct wall. Therefore. The level of dielectric strength of compressed SF6 outlined in the foregoing sections may not be fully attainable in reality.

Studies of the effect of small gas gaps between the spacer and conductors on breakdown performance showed that the reduction in breakdown strength is more pronouncedat higher gas pressures. This phenomenonis explained by the conditioning of the electrodes. Thedependenceofbreakdown strength on electrode material is especially evident in uniform and quasiuniform field gaps. The presence of some degree of moisture adversely affects the dielectric strength of the spacer/gas interface. The breakdown strength is also reduced by the roughness of the electrode surface.8 kPa. Particularly at high SF6 gas pressures. The breakdown strength of an SF6 gap no longer has a single value at a given gas pressure. The dispersion is larger at higher pressures.withsteel electrodes than with copper. times higher than the average. Narrow gaps may exist between electrodes and spacers due to imperfect casting and/or mechanical stresses at the solid/gas interface. repeated breakdowns affecting the electrode surfaces have their influence on the strength of thegap. the protrusion causes a decrease in the bre oltage. Thus a microdischarge is then sure to occur in the gap. The dielectric strength increases with the number of previous breakdowns until it levelsoff at an ultimate value. It was reported that the breakdown voltage dropped by more than . However. It has been found that the reduction in the breakdown voltage due to electrode roughness depends on the type of gas insulation as well as the product of the protrusionheight Rmaxand gas pressure p. Several factors contribute to the reduction of dielectric strength of spacers: 1. The fluctuation in the breakdown strength is thus better expressed by its statistical distribution. I ~ p e r f e c tsolid conductor adhesion.112. The reduction in the maximum withstand stress E related to the gas pressure p as a function of the product pRmaxis shown in Figure 10. The effect of conditioning on breakdown in SF6 was reported to depend on the gas pressure and the electrode area (Anis and Ward. As explained in Chapters 2 and 8.cm. When this product exceeds 0. 1988). 2. a small gas gap stressed in series with insulation of a dielectric constant Er would undergo an electric stress E. the effect is negligible if theproduct is less than 0. ~ o i s t u r econtent.

as shownin Figure 10. .8 0. A s mentioned in Section 10. 3 I o2 I o3 . 3. 1982)..6 0. . Trinh et al. 1984)..7 0. thebreakdown in a high-pressure CIS is more likely along the gas/spacer interface. The effect of metallic particles on the SF6 breakdown voltage is more pronounced at highgaspressures. causing local field enhan~ement.9 0.2.4mm diameter wire contaminants.4 0 .6. unlike low-pressure CIS where breakdown is purelygaseous.0. The presence of particle contaminants in gas-insulated systems isby far the most significant factor responsible for the deterioration of insulation integrity.5 0. Efforts have been made to optimize the design of post-type spacers basedon minimizing the field enhancement in the gas at the spacer boundary while accounting for resultant electrodynamic forces on the conductors (Stone et al.. R P ( Pa cm ) - I 0" I C l Effect of surface protrusions on the strength of SFs. ting Contaminating particles mayeventually adhere to the solid spacerby electrostatic forces. In the following section the subject of particle contamination in CIS is discussed in more detail. 1987. .13 for a 150/ 250mm coaxial system and 0. ~ u ~ t f l ~ i n f l particles. 50% when the humidity was increased from zero to 4000 ppm (Masetti et al.

] In the case of freeparticles.7 0 i 500 PRESSURE kPa IO00 I I I500 2 x) 10.13 Effects of conducting wire particles of different lengths on the strength of an SFGgap at different pressures. the particle motion depends largely on the type of applied voltage. free conducting particles become charged and oscillate in the interelectrode gap. for a wire particle of given radius. [From Cookson and Farish (1973). Under AC voltage. the particles oscillatebetween the electrodes. lateral particle movement along the central conductor is possible. Under the influence of the applied field. In addition.500 I I I CLEAN WIRE LENGTH 400 300 200 IO0 2. and wire particles may exhibit intense “firefly” activity. Under DC voltage. the activity increases with particle length since the particle chargeto-mass ratio at lifting increases with length. As a charge particle .

This must be achieved without risking damage to the system from flashovers. or by providing designated low-field areas in the system in the form of “particle traps” where the particles can be safely trapped and contained.14. The trap must have positive retention characteristics under AC. the breakdown voltage profile is as illustrated in Figure 10. fora given particle-contaminated CIS system at some particle positions in the gap. the problem of particle contamination should be overcome. and transient voltages as well as mechanical vibrations. gas pressure. 1978).approaches either electrode. It has been reported that. An attempt was made to establish an analytical “breakdown voltage profile” that relates the instantaneous breakdown voltage magnitudes to the instantaneous particle position in the interelectrode gap (Anis and Srivastava. DC. Particle-initiated breakdown in compressed gas insulation generally depends. . among other parameters. An electrostatic particle trap can be designed such that a region of low or zero electric field is provided at the outer enclosure by a slightly elevatedmetal shield. Particles can enter the low-field regioncreated by the trap through slots in the shield. and the general effects of particle size. There are several important factors to be considered for particle control with such traps: 1. 1981a). Particle contaminants have to be moved through the CIS system to the location of the traps. The breakdown voltage profile helped to explain the significant difference between the breakdown voltages of fixed and free particles. 2. The trap must provide rapid capture of the particles without any restriction on particle size and shape. The profile explains the experimentally discovered existence of a critical particle position at which the insulation displays minimum dielectric withstand. This microdischarge may well trigger the breakdown of the main gap (Cooke. the breakdown voltage is lower than for others (Cooke et al. For a 1mm radius spherical particle in a parallel-plane SF6 gap at a gas pressure of 5 atm. 1977).. and electrode configuration. For systems to be reliable and economical. Contamination control in CIS can be achieved either by designingthe system with a certain degree ofimmunity against the presence of particles. on the position of the particle in the gap. 3. This “critical” breakdown voltage may be taken as the insulation withstand limit under transient as well as steady-state applied voltages. it may lose its charge to the electrode through a microdischarge in the gas.

showing the effect of a 2 mm spherical particle position along the locus of breakdown voltage between a particle and the right-hand elecrode. (b) locus of breakdown voltage between a particle-caused protrusion on the righthand electrode and the left-hand electrode. Conductors in a gas-insulated system may be coated with a dielectric material to restore some of the dielectric strength of the compressed gas that is lost as a result of surface roughness and contamination with 'conducting .nis 3 l VC 200 1 0 0 0 Breakdown voltage profile of a particle-contaminated SF6 2cm gap: (a) gap.

2.7:7. can be attributed to the following effects: l . due to coating. if any. it can be shownthat the lifting field Eland the corresponding lift-off charge q1 of a filamentary particle in a coaxial CIS under AC configurations are (10. dielectriccoating can be of further benefit in two main ways: a. Coating reduces the degreeof surface roughness on conductors. A particle resting on the bottom of a CIS duct acquires the charge necessary for lifting either by conduction through the dielectric coating or by means of microdischarges in the gas initiated at the particle surface. The charge accumulatedon the particle should reach a value at which the electrostatic lifting force on the particle overcomes the gravitational force (Anis and Ward. due to the coating (Parekh et al. The electric fieldnecessary to lift a particle resting on the bottom of a CIS enclosure is much increased.1988).7) . thus increasing the breakdown voltage. it may collide with either conductor. If the conductor is coated.particles.. 10.6) and (10. and therefore a conductive charging current will pass through. b.. 1983).1 Particle Charging through Coating Layer Conductance The dielectric coating layer has a finite nonzero conductance. 3. The improvement in dielectric strength of the system. In the presence of metallic contaminating particles. Thus therisk of a breakdown initiated by a discharge is reduced significantly. Once a particle begins to move in the gasgapunder the applied voltage. The high resistance of the coating dielectric impedes the development of pre-discharges in the gas. thus decreasing the high local electric fields (Endo et al. Using this mechanism. the particle will acquire a drastically reduced charge. 1979).

31 where 1Cl(44 = COS (9 A= . k is approximately 0. and the electron attachment coefficient. andW is the angular frquency. Ifthe fieid near the particle is highenough. 10. which is a function of both field and space. and g is the gravitational acceleration.8) 1 and x2 are the avalanche limits in space. a the ionization coeffiwhere x cient. thus charging the particle..7./(l a(@) = (cos + Cd/Cg)2/G2 + 1/w2C.2 Particle Charging through ~ i ~ r o d i s c h a r g ~ s The other possible mechanism through which charges can be transferred to a contaminating particle from a coated conductor in a CIS is that of microdischarges. to account for the effectofimage charges on the electrostatic force. discharges mayoccur between the particle and the electrode through the dielectric coating. less than unity. This mechanism was used to compute lifting fields of spherical particles (Parekh et al. e being the electronic . 4cos -COS[(^^ 2(9)/3] sin[(2n +)/3] k/u + 2x-(9 e) sin 3 3 where Cg and C d are. G is the conductance of the particle-to-outerelectrode path.r and R are the inner and outer radii of the CIS system. the capacitance between the particle and the inner and outerelectrodes. electron avalanches will be initiated at a point on the particle surface and propagate along a field line toward the coating layer. and nearly unity for long vertical wire particles (Anis and Srivastava.7. Having M. The size of an electron avalanche started by one electron is given in Chapter 4 as (10. the charge transferred from the conductor surface to the particle is en. respectively. 19’79). When the particle is in contact with the dielectric coating and the field is sufficiently high.. k is a factor. 1981 b). The point on the particle surface at which avalanches will develop is that which ensures maximum avalanche size.82 for spherical and horizontal wire particles..

For thick coatings. A comparison between the possibleeffectiveness of the twomechanisms for lift-off is shown in Figure 10. Equation (10. l 5 Comparative lifting fields for conducting-wire particles off coated electrodes in compressed-gas ducts. the horizontal and the vertical positions. Very fast transient overvoltages (VFT) are generated in GIs during the operation of disconnectors. The controlling mechanism causing the particle to lift off at a lower applied field will be either conduction or microdischarges.9 times the operating .15.L I F T I N G OF V E R T I C A L WIRE BY C O NDUCTION 25 *" 0 WIRE By L IFTING 0 F ~ORIZONTAL PARTIALDISCHARGE L1FTING OF VERTICAL WIRE NDER PARTIALDISCHARGE 20 J U 4! L L f: IO L I F TING OFHORIZONTAL W IRE UNOER CONDUCTION .8) can be applied by first computing the field distribution in space. which is then related to (aq ) (Chapter 2). These surges are steep-front oscillatory overvoltages with a maximum magnitude of the order of 2. charge. Two initial positions for wire particles are shown in each case. a particle will lift off from a horizontal position. It is clear from the figure that particle lift-off for all coating layer thicknesses beyond 100 pm is not possible through a microdischarge mechanism. Conduction throughthe coating or some other mechanism hasto be responsible for lift-off.

and remained constant down to -50°C.4 Particle flashover under very fast transients. independent of the environmental temperature. given the appropriate pressure.5 mm and extending to 7mm.43 x 1019cm-3 (equivalent to pressures of 105 and 624 kPa. . For particle lengths in the range 3-1Omm it was reported that the flashover voltage decreased as the particle length increased. the V-t characteristics for VFT and lightning impulse werenearly identical for each of the particle lengths considered and. Figure 10. respectively.2 0. To verify this.4 0. The insulation characteristics of spacer surfaces for VFT under particle contamination conditions were investigated by Okabe et al.16 depicts the resulting voltagetime characteristics. for a VFT frequency inthe range 1.596 to 16. The experiment was conducted for molecular densities ranging from 2. The electric strength of CIS was believed to be determined by the gas density.voltage and a fundamental frequency of several megahertz. The decrease appeared at a temperature threshold between -25 and -3O"C. 1995). It was concluded that the dielectric strength along a spacer surface for VFT is at least equivalent to that for lightning impulse.7-1 8 MHz. Outdoor equipmentmaybeexposed to temperatures dipping down to -5O"C. No definite trend with regard to a temperature dependence was traced under high field conditions. (1996). it was found that lowering the environmental temperature brought about a 10% reduction in the withstand capability of the system. no dependence on frequency was observed. at 24°C) and for gap lengths starting at 0. a temperature sufficiently low to cause liquefaction of the insulating medium (SF6). Despite a constant molecular density and under preferably uniform-field conditions. which suggested that the variance 1000 particle length A A 0 0 0 A3 m m 10 m m 5 mm 0.8 1 1.6 Time (PS) 0.2 1.6 1.. the low-temperature DC breakdown of an SF6 gas-insulated system wasinvestigated experimentallyfor temperatures ranging from -50 to 24°C (Frkchette et al. However.

thus approaching EMC compatibility. Other noise sources include lightning discharges. and short circuits (~eppeling and Remde. Several means may be used to inhibit the coupling paths. earthing systems. Also to be taken into account are external sources such as local communication systems and power transmitter stations. noise current. it is therefore necessary to make the electronic equipmentcompatiblewith its electromagnetic surroundings. lo%. 1986). the findings set the order of magnitude for the withstand drop below -25"C. electromagnetic noise fields. the conclusion maybe that the equipment design should take this into account because of the implicit reduction in the operational safety factor. Electromagnetic compatibility is achieved when the disturbance variables active at the noise sink are smaller than the immunity to disturbance of the sink. In this regard. voltage and current transformers. Some of those means are: Meshed earthing system Use of control cables and connectors with low transfer impedance Cable shields earthed at both ends Secondary circuits of instrument transformers earthed only once Coaxial entry of secondary cable shields in cubicles Filtering of the mains supply . As the complexity of the electronics grows in a GIS substation. Since the breakdown behavior of GIs was shown to depend on the environmental temperature. The disturbance variables are transmitted over a coupling path to the noise sink (electronic equipment or secondary units). The noise sources themselves are characterized by such variables as noise voltage. and circuit breakers. and electromagnetic field coupling into control cubicles. To ensure high reliability for the installation. where they are reduced by the coupling mechanism. earth faults. so does its susceptibility to increasing electromagnetic interference in the environment. and noise energy.. i.observed under highly uniform field conditions may be related to a triggered disturbance (field related). The electromagnetic environment in GIS substations depends on the noise sources. Typical coupling paths are secondary cables. which provides a basis for deciding what measures to take. which are essentiallyswitching operations carried out with disconnect switches. earthing switches.e. linked to a modification in the state of the cathode surface.

2. Of disadvantage in such cases is the high reactive power required for extensive installations. a powerfrequency test should be carried out wherever possible. sealed in a gastight housing. and then transported. measurement of partial discharges may be carried out.Experienceshows that a final value is reached after this period. tested. At the siteof installation the following tests are performed during commissioning: Every flangedjoint made on site is tested for leaks. 3. and the high test voltages.3 nis Limitation of transient overvoltages Use of fiber-optic cables for communication and data transmission In contrast to conventional switchgear installations. and earthing switches. 4.1 Power-FrequencyVoltage Because it corresponds to the continuous stress met within service. evacuated. The amount of air in the SF6 is measured. If the design of the voltage source and ambient conditions permit. CIS installations require systematic assembly according to a plan laid down at the manufacturing stage. and also in the settings of the supervisory instruments. The moisture content of the SF6 gas is measured between twoand three weeks after the first filling.9. and filled with dry nitrogen at 150kPa. 10.1 . they are evacuated to about l00 Pa toremove any moisture that may have entered during the assembly phase. The earthing system is checked. The high-voltage equipment is assembled at the works to form large units. As soon as the components forming a common gas chamber are assembled. The state of the insulation can be judged more accu- . 5. The maximum permissible values per joint are set so that the entire installation cannot lose more than 1% of its gas per year. isolators. High-voltage tests are performed onthe installation to detect damage or contamination that may have occurred in transit or during assembly. are carried out as soon as the appropriate stage of completion is reached. 7. Functional checks on the breakers. 6. where the switchgear need not be installed according to any given sequence.

however. It is.2 Impulse Voltage Applying impulse voltages to complete installations is not an entirely suitable way of checkingthe quality of erection because it indicates only shortcomingsin the geometricarrangementwhereas contamination remains undetected. 10. is performed for a maximum of 1minute for AC or DC voltage or up to five shots with oscillating switching surges. or with switching surges up to a maximum of three shots. it may be sufficient for the test to be carried out with the more critical polarity. although this stress rarely occurs in operation. testing of the installation can be limited to checking the dielectric strength between the live parts and the enclosure.rately if frequent partial discharge tests are carried out together with the AC test.3 DC Voltage In cases whereAC and impulse testing cannot be performed. owing to the abnormally high ambientnoise level.9. This distorts the shape of the wave. testing with oscillatory switching surges. Certain drawbacks in conventional testing equipment can be avoided by the use of resonant circuits. However. Since the switchgear contact gaps are already factory tested.1. in which the largest possible number of installation sections are coupled together. it is possible to test the switchgear with DC voltage instead. These sections are determined by the position of the isolators and breakers and are normally tested with AC or DC voltage for a period of 10 S . but although the distortion does not much affect the test result it must be taken into consideration when measuring the voltage level. For stresses varying with polarity.9. . expedient to establish single testing sections in order to localize faults during testing. 10. The final test. In mostcases. A DC voltage test can still be carried out at 80% of the rms value of the rated power-frequency test voltage. however. the test results obtainedonsite cannot be conclusive. In a few cases a power-frequency test can be carried out by energizing the voltage transformers on their low-voltage side. is sometimes used. which correspond to the standard switching surge when flashover occurs. When the manufacturing tests have been carefully supervised. the test on the completed installation may be considered a repetition. The voltage amplitude can be limited to 80% of the rated test voltage without overlooking assembly errors.1.

The circuit breaker can normally withstand 20 interruptions at its rated short-circuit breaking current and 2000 interruptions at full-load current. and microdischarges associated with poor contacts orwith the transport of conducting particles. and the section is filled with SF6 gas to the required density. Discharges in voids. therefore. The particle-based discharges in a CIS and the subsequent acoustic and high-frequency emission have been widely investigated (Runde et al. the very fast rise time of the PD pulse causes electromagnetic energy to be coupled into the GIS chamber.1995). lO. both in the laboratory and on site during commissioning (Pearson et al. During this process the gas moisture content is checked. Various diagnostic techniques have been developed.is The CIS is virtually maintenance free and is designed to avoid any opening of the enclosures during its lifetime of at least 30 years. A special gas-handling chart is normally used for this purpose. and so on. 1994.. and the energy dissipated in the discharge is replaced through a pulse of current in the EHV supply circuit. the discharge takes the form of corona streamers which give rise to current pulses with very short rise times (< 1ns). Nitrogen is used to dry out the internal partsof the CIS. In microdischarges and intense coronas thedischarge is followed by rapid expansion of the ionized gas channel. 1997). Ziomek et al. The PD. are also characterized by high rates of change of current. has many . Finally. The operating devices require minor maintenance a few times during the CIS lifetime. and by the creation of chemical breakdown products. In all cases. 1997. A. Sellars et al. floating shield. Charging the CIS with compressed SF6 gas is effected by evacuating each gas compartment (Fig. moisture absorbers are mounted in circuit breakers and each section is filled with SF6 gas to the working pressure. to identify it as a particle. even the circuit breakers need not normally be opened for maintenance. For surface defects such as snlall protrusions.. partial disclxarge (PD) is the localized breakdown of gas over a distance of usually less than 1mm.. The main objectives of GTS diagnostics are to detect whether there is any defect in the CIS. and to locate it so that it can be corrected.lb). and anacoustic pressure wave is artial discharges are also accompanied by the emission of light from excited molecules.. Therefore.

and much longer times would be needed. An example of an acoustic signal is shown in Figure 10. The latter is the only instance of a diagnostic signal not coming from a P (although of course the particle generates PD as well). and in principle any of them could be used to detect the presence of the discharge. and since this is absorbed strongly by both glass and SF6 it is necessary to use quartz lenses and a reasonably short path length. The signals in GIs have abroad bandwidth. The main decompositionproduct of SF6 is SF4.Those originating at the chamber wall propagate as flexural waves. 1992). It reacts further. at velocities which increase with the square of the signal frequency to a maximum of 3000m/s. typically with traces of water vapor. chemical. This is based on the fact that the chemical decomposition isnot affected by the electrical interference which is inevitably present in the GIS. and electrical.. and with any steady discharge the concentration of the diagnostic gas should rise in time to a level where it can be detected.17 (Runde et al.. because a photomultiplier can detect the emission of even a single photon. 1997). in a GIs the diagnostic gases would be greatly diluted by the large volume of SF6 in which they occur. It therefore appearsthat the chemical approach is too insensitive to be considered for PD monitoring. which is a highly reactive gas. Although this is a powerful laboratory tool for finding the onset of activity from a known corona point. Propagation through the gas is at the muchlowervelocityof . there are many difficulties in using it to detect a discharge that might be anywhere in a GIs. to form the more stable compounds thionyl fluoride (SOFZ) and sulfuryl fluoride (SOZFZ). However. typically after about 50 h. Detecting the light output from a discharge is probably the most sensitive of all diagnostic techniques. and by using a gas chromatograph and mass spectrometer they may be detected with sensitivities down to 1 ppmv. These are the twomostcommon diagnostic gases. In smal~-vol~~m laboratory tests a reasonably small discharge of 10 to 15p c can be detected. physical. The radiation is primarily in the IJV band. Acoustic signals are generated both from pressure waves caused by partial discharges (PD) andfrom free particles bouncing on the CIS enclosure. and travel from the source to the detector by multiple paths (Lungaard et al.effects.

It also has other features. give rise to a complex signal pattern. One advantage of acoustic measurements is that they are made nonintrusively. Filled epoxide barriers also attenuate the signal markedly. Other sources of discharge may be identified in a similar way from their own characteristics. and the reflections it experiences at boundaries between them. The acoustic signal from a particle bouncing on the chamber floor is characterized by a signal not correlated with the power frequency cycle. from which the particle shape and its movement pattern can be inferred.3 Anis I I I l I . Because of the rather high attenuation of the signals. and the higher frequencies in the signal are absorbed quite strongly. This can be picked up by accelerometers or acoustic emission sensors attached to the outside of the chamber. using external sensors which may be moved from place to place on the GIs. and the ratio of the lift-off/fall-do~n voltages. such as the crest factor (ratio of the peak rms value). l 7 Acoustic signal from a bouncing particle and the corresponding 50 Hz voltage trace. 150m/s. the sensors . the impact rate. The different propagation velocities of the wave as it passes through various materials.

the total capacitance of a GIs is high. which because of the low losses in these chambers may persist for 3 ps (Hampton et al. The current along the PD path rises in lns and contains frequency components which extend from DC to more than 1000MHz. The resonances are indicative of PD activity. A GIs partial discharge test circuit is described in IEC Publication 270. it is sufficient to fit couplers at intervals of 20m along the chambers. Since the UHF signals propagate throu~houtthe GIs with relatively little attenuation and even low levels of corona can be detected readily.S-lnsul 7 should preferably be on the chamber containing the source. because too many sensors would be needed. and the GIS appears to the external circuit as a lumped capacitor with a depleted charge.. The pulses are attenuated and undergo multiple reflections. 1996). After l ps the pulses die away. and propagates as a traveling wave in each direction along the chambers. In that circuit the charge flowing through a coupling capacitor fitted in parallel with the GIS is measured using a quadrupole and a detector. but do not appear in the external circuit. Also. which is possible for a test assembly but maybe inconvenie~twhen testing a complete GIs. In addition. Over the past years much experience has been gained using UHF at frequencies from 300 to 1500 MHz. there is no means of locating the discharge. They excite the GIS chambers into various modes of electrical resonance. 1990. From then a replacement charge flows into the GIs. and since a cou~ling capacitor is not normally provided in a GIS the technique cannot be used for in-service measuremen~s. but a more accurate position can be found within 1cmby using a second sensor and a time-of-flight method. and it must be divided into sections for test. This in itself gives the approximate location of the defect.. This can be undertaken at VHF frequencies - . Judd et al. The P current pulse at the defect probably has a duration of l0 ns. and if they are picked up by couplers installed in the CIS they may be displayed on a spectrum analyzer. UHF couplers maybefitted to the insideof hatch cover plates onGIS equipment and UHF monitoring systems are permanently installed. The acoustic technique is not suited to a pemanently installed monitor. To obtain the maximumsensitivityof measurement a completely shielded test arrangement is required. and is measured by the detector.

. . . . . . . . . . . . . . A study compared experimentally the above different detection techniques (GIGRE V V G 15. . . . . . . . . the acoustic measurements were nonintrusive and thus could be made on any GIs. . . The UHF components of the resonant signal give it a very fast risetime. .. . . . . . . . . . . . . .. . .nis from 30 to 300 MHz and at UHF. . . . . . . . . . . . . . . . . . .03. and the source of the PI) can be located to within 10 cmfrom the time interval between the PI) pulse wave front arriving at couplers on either side of the defect. Also. . .The coupler signals can be phase resolved to show the point on the AC wave where the discharges occur. . . . . . . 18 (Sellars et al. . . . .. . . . . . . the ability to locate discharges accurately by time-of-~ight measurements. . . . . . . . . . UHF couplers can be either internal or external. . . .19 depicts the results from the various techniques expressed as signal-to-noise ratios. . . . . . . . . . . The principal advantages of the UHF method are its high sensitivity. . . . . . . . . . . . . Since the intensity of air corona falls rapidly with increasing frequency. and that it can beusedreadilyin a continuous and remotely operated monitoring system. . . . say at more than 500 MHz.. Point-on-AC wave for a particle in GIs.. but this can be easily be recognized. as seen in Figure SO. . . . it is preferable to make measurements in the UHF range. . . . . .but when the entry is by overhead line then interference from air corona can be fed into the chambers. . . . . . . . Even then communications interference maybepicked up. . . . . . . . . It has been found that the background noise level in cable-fed CIS usually is very low. .. . . . . . . Figure 10. . l. . . . . . . . . and UHF techniques all showed good sensitivity. . . . . . . . . . . conventional PI> (IEC 270). 1994). For a needle attached to the busbar. . but the attenuation of the signal across barriers and along the chambers was high. . . . . . . conventional PI) measurements needed an external 1 I . . . . . . .1992). . . . . . It was reported that although the acoustic. . . . . . . . . . . . for example from portable telephones. 1 . . . . . . . . .

reclaiming. In a more recent study nosignificant difference between patterns obtained from the different methods was observed (Meijer et al..50 100 150 200 U kV 250 Comparative partial-discharge signals from different detection methods. (c) acoustic. (b) UHF. The world production of SF6 hasbeen steadily increasing. (a)conventional. 1998). Without disposal methods that actually destroy SF6. 1997). testing.. and its release into the environment by the electric power industry comes from normal equipment leakage.5% per year.7%. maintenance. and could not beused on GIS inservice. handling. This release of SF6 into the environment cannot bereducedsignificantlysince there are no . In many industrial applications SF6 is not recoverable. coupling capacitor. The UHF technique was suitable for in-service monitoring. resulting in increased concentration of SF6 in the atmosphere (~hristophorouet al. it can be expected that all of the SF6 that has ever been or will ever be produced will eventually enter the atmosphere (within the next few centuries). This is so even if the present SF6 leak rate from enclosed power system e q u i p ~ e nhas t improved to only 0.The amount of SF6 in the atmosphere has been increasing at an annual rate of 8. etc. The atmospheric concentration of SF6 could reach 10 parts in 10” by volume by the year 2010. but for maximum sensitivity couplers needed to be fitted inside the GIs.

These efforts include improved methods to quantify and stop leakages. it absorbs a portion of the infrared radiation emitted by the earth and returns it to earth by emitting it back. efforts have recently been undertaken by the electric power industry to better monitor the gas pressure in sF6-insulated equipment and the amount ofSF6 released into the environment. i. and eventual disposal. be too small to have signifi~antenvironmental consequences. Because this gas is already widely used. and implementation of a sound overall policy of using. Sulphur hexafluoride isa potent greenhouse gas.e. It has been suggestedthat impure used SF6 in storage containers can be destroyed by thermal decomposition in industrial waste treatment furnaces at elevated temp~ratures (T > lloooc).” The greenhouse effect shifts the balance between incoming and outgoing radiation..2%. One considered solution is the alternative use of gases that are benign 2 which is nonelectronegative and has and environmentally ideal. manufacturing tighter and more compact equipment9development of sealed-forlifeelectrical apparatus. in the near term. causing “global warming. handling.. It is also largely immune to chemical and photolytic degradation and therefore its contribution to global warming is virtually permanent. setting of standards for recycli~g. electrical.5 pm. and other industries have expressed concerns over the possible long-term environmental impact of SF6. chemical.” Sulfur hexafluoride absorbs infrared radiation at wavelengths near 10.nis currently accepted and economicallyfeasible methods for controlling or destroying SF6 as itleaks from enclosures. Government and enviro n ~ e n t aprotection l agencies. decreasing the rate of SF6leakage and increasing the level of recycling are high priorities since they both will curtail production needs of SF6 and thus will reduce the quantities of SF6 that are eventually released into the environment. there are obvious economic implications in any attempts to regulate or control its production. use.. Indeed. The relative contribution of SF6 to nonnatural global warming is nearly 0. the amount of SF6 in the atmosphere should. gradual replacement of older equipment which normally leaks at higher rates. However. better pumping and storage procedures. The effective trapping of infrared radiation by the gr~enhouse gases and its re-radiation back to earth results in an increase in the average temperature of the earth’s atmosphere in what is called the ‘‘greenhouse effect. While the potency of SF6 as a greenhouse gas is extremely high. and tracing SF6. such as N lower dielectric strength (three times lower than SFG) and lacks the funda- .

such environmentally friendly gases might be used by themselves at higher pressures.4 MPa compressed gas-insulated feeder is to be designed. The basic insulation level (BIL) of such a system is 1050 kV. 1995). The positive breakdown field of SF6 is nearly a linear function of the gas pressure and approximately equals 81. A study of radio interference from a CIS revealed that they were either below the background levels(i. transient field levels diminish rapidly with distance from GIS stations. and between 10 and 15MHz for sites outside the station. Oscillatory transients rose to their peak values in an interval between 0.e. as the main component in mixtures with electronegative gases. and decayed to below 50% ofthose values in less than 1 ps and to below 10% in less than 3 ps.. arc quenching.5 MHz for both horizontal and vertical polarizations. Dominant frequencies were between 20 and 25 MHz for sites within the station. The surface discharge field of CIS spacers is 3. Considerations are made for the use of high-pressure N2 and mixtures of NZwith SF6 for insulation..5 ps. which suggestedthat the station building attenuated the higher-frequency Components more than the lower-frequency components. It was found that for such uses the SF6/ N2 mixtures with 50% SF6 are efficient arc interrupting media. In brief. . or at comparatively lower pressures. In the same study. other mixtures in use include SFG/CFd and SF6/He. or contributed little to the background (Harvey et al. Nonetheless.mental requirements for use by itself in circuit breakers. as well as the useof high-pressure nitrogen for gas-insulated transmission.05 MHz. and current interruption.3 (kV/m)/kPa. transient peak fields of up to 580V/m were measured inside the station building and dropped to background values of 10V/ m at 120m from the station. and by 26 dB at 73. N2.2 MV/m. its coaxial dimensions evaluated.5 and 5 MHz..e. 0. Mixtures of N2 and SF6 are being used in circuit breakers under severe weatherconditions (T K -40°C) where SF6 usedunder pressure in circuit breakers may liquefyand thus lose part of its current interruption properties. Measured broadcasting signal levels FM and TV stations werehigher than the RI levels in these bands. i.2 and 0.levels due to all RI sources including power lines but excluding the substation). Peak levelswerehigher than average levelsby about 13dB at 0. (1) A 380 kV. while peak levels were higher than quasi-peak levels by 4 d 0. including SF6.

Prove any relation you may use. Anis H. Ward S. 1981b. without calculation. and impulse voltages. the gas density rather an the gas pressure is measured.nis esign the duct system on bothgaseous and solid insulation withstand bases.2p + 10kV/cm where p is the gas pressure (multiples of atmospheric pressure). (4) Compare.5 kV/mm. Anis H. Assume minimum conductor surface field. (2) the factors that may affect the withstand strength of com-gas-insulated switchgear. calculate the conductor and enclosure radii. The system’s solid spacers are designed in such a way as to have the maximum field 25% higher than its average field. . AC. Comment on the results.35 MPa. Ottawa. (6) What are themain advantages and disadvantages of SF6 when used in are CIS usually sectionalized? pare low. scribe the “breakdown voltage profile” in GIS switchgear. pp 312-317. based on gas and solid breakdown considerations. Anis H. showing your method of calculation. (3) A compressed gas-insulated coaxial system operates at 380 kV at a pressure of 3 atm. IEEE Trans ET-16:327-338.Choose acceptable dimensions forthe system. (l 1) What are the factors governing the conductor and enclosure diameters in CIS? (12) Explain why. The BIL peak AC design voltage ratio is2.and high-pressure GIs systems. are conductors in GIS systems sometimes coated with dielectric (10) Write down the equations governing the movement of solid conducting particles in GIs systems under DC. the surface discharge withstand is 2. 1988. hat are the main troubles caused by solid spacers used in GIs? (14) For a 240 kV GIS system whose basic insulation level is 1050 kV and which operates at 0. the modes of breakdown if the GIs is erated at pressures of 1 atm and 6 atm. in monitoring SF6 in GIs systems. Srivastava KT>.1. and comment on the results. showing the various forces acting on them. The equation governing the impulse breakdown strength and pressure for SF6 is given by -lZBD 81. IEEE Trans PAS-100:3694-3702. Proceedings of a Conferenceon Electrical Insulation and Dielectric Phenomena. Srivastava KD. 1981a.

Larocque 2171. Report 400. 1998. Van Brunt RJ. Hampton B. Meijer S. Vienna. Ikeda M. 1986. Kurtz M. Aurud T. CIGRE paper 15/33-01. Naidu MS. Masetti C. 1977. . Endo F. Hampton BF.05. Boggs S. 1990. Advances in High Voltage Insulation and Arc Interruption in SF6 and Vacuum. Lungaard LE. Skyberg B. IEEE Trans Dielectr Electr Insul 5:830--842. Faugstad K. Pigini A. Proceedings of1st International Symposium on Gaseous Dielectrics. Proceedings of 4thInternational Symposium on Gaseous Dielectrics. Endo F. Geneva: International Electrotechnical Commission. TN. Srivastava KD. Brambilla R. Irwin T. 1979. 1996. 1996. Wong PS. Athens. Koto M. ABB Rev 2: 12-20. Runde M. Report 1010-02. Muhr M. Farish 0. WelchIM. 1992. 1984. Meppeling J. Diessner A. Smit J. IEEE Trans Dielectr Electr Insul 2:925951.1997. Ishikawa T. Judd MD.Christophorou LG. 1982. Trans IEEE PAS-96:768-775. IEEE Trans Power Deliv 10:357-363. 1978. Teranishi T. IEC. Lightle D. Cooke CM. Hogg P. Proceedings of 4th BEAMA International Conference on Electrical Insulation. Harvey SM. IEEE TransPower Deliv 5: 17511759. Pearson JS. Ljdcelsray K. Balma PM.Templeton D. Schemer H. 1998. 1983. Oxford: Pergamon Press.1992. Farish 0. Austria. Paper 32-05. Van Heeswijk RG. IEEE Electr Insul Mag 13:20-23. IEEE Trans Power Deliv 11:210-216. Skyberg D. IEEE Trans PAS-922371-876. IEEE Trans PAS-98:748-755. 198 1. Proceedings of 4thInternational Symposium on High Voltage Engineering. Judd MD. Vienna. Wootton RE. Kobayashi K. Remde H. IEEE Trans Dielectr Electr Insul 1:323-33 1. Cooke CM. Roberge D. IEEE Trans Dielectr Electr Insul 2:895-905. Yanari T. AH. 1995. Tangen G. Braun J. Farish 0. Cookson. Hampton l3F. 1987. Ozawa J. Toda K. CIGRE Working Group 15-03 CIGRE paper 15/23-01. IEEE Trans Dielectr Electr Insul 3213-228. 1995. 1981. Frkchette MP. 1994. 1973. Kara A. Hampton BF. Paper 2296. Maller VN. Trinh NG. pp 119-123. Publication 270. Proceedings of CIGRE Symposium on New and Improved Materials for Electrotechnology. pp 162-189. Farish 0. Knoxville TN. Inoue T. Proceedings of CIGRE Symposium on New and Improved Materials for Electrotechnology. Brown Boveri Rev 73(9):498-502. 1995. Nskleby JE. Ninth International Symposium on High Voltage Engineering (ISH'95). Mitchel GR. Reisinger F. IEEE TransPower Deliv 12:714-721. Olthofe JK. Parekh H. Bargigia A. Yamagiwa T. Sellars AG. Partial Discharge Measurements. Cookson AH. Brighton. Graz. pp 335-341. 1987. Stone G. 1997. Pryor BM. Vincent C. Knoxville.1995. Gulski E. Okabe S. Fuchsle D.

IEEE Trans Power Deliv 11:872-879. 1997.4 nis Yama~ata Y. Nishiwaki S. Tanaka K. Takahashi N. Ziomek W. . Kuffel E. Imai K. 1996. Miwa I. Komukai T. Kokumai T. IEEE Trans Dielectr Electr Insul 439-43.

of which the circuit breaker arcing occupies one-half to one cycle. and capacitive currents. external elements are added to the breaker such as . but are also capable of making and breaking the circuit under the severest system conditions. however. Egypt Circuit breakers differ from switches in that they not only manually make and break the circuit while carrying their normal currents. Under short-circuit conditions. Fault clearance time has been immensely reduced during the last 50 years due to the high technology adopted in circuit breaker design and the use of static relays. Fault clearing time on the order of two to three cycles have been achieved. These voltages may reach 5 times. Giza. Breaking or making the circuit under load conditions represents no real problem for a circuit breaker since the interrupted current is relatively low and the power factor is high. the normal circuit voltage and damage of circuit insulation or restrike in the circuit breaker may occur.1. small inductive currents.Cairo ~niver~ity. high overvoltages appear in the switched circuit. To reduce these voltages. When switching short-circuit currents. It is the duty of a circuit breaker to interruptsuch currents as soon as possible to avoid equipment damage. the current may reachtens of thousands of amperes at a power factor as low as 0. or even greater. Loss of system stability is a consequence of slow fault clearance.

these gases and the metal vapor will be highly ionized.switchingresistors. are usually made of highly conducting material. These methods are to cool the arc. The heat produced within such a small area causes the metal to melt and even evaporate (Chapter 6). 9% methane. In addition to the breaker’s rated voltage and frequency. inductors. and split it into a number of arcs in series. the heat generated within the arcdecomposes some oiland generates gases. the contact area decreases to a very small value just before contact separation. and electronic circuits for controlled switching. All arc interruption methods are aimed at disturbing the energy balance of the arc. The rated circuit breaker current is the rms value of the current that it can carry continuously without the temperature riseof its components exceeding the specified limits. In air circuit breakers.1 shows the short- . increase its length. Figure 11. there are other rated quantities that are important for its operation and selection. These quantities are discussed briefly in. arresters. The circuit breaker contacts. and 8% others. the following paragraphs. For oil circuit breakers. it maybewell to define the basic rated quantities of circuit breakers and discuss the currents they are called upon to interrupt in AC circuits. Before describing in detail the different types of circuit breaker. forming an electric arc. with adequate contact areaensured by a suitable contact pressure. The amount of gasesgenerated is a function of the arc energy and the gases are composed of about 66% hydrogen. metal vapor and hot air bridge the gap between the contacts. fixed and moving. When the contacts part to break the circuit. The rated breaking current is the rms value of the current that it can break under specified conditions of recovery voltage. They thus provide a conducting path between the circuit breaker’s contacts and the circuit current will continue to flow through the arc. 1’7% acetylene. as happens in multi-break circuit breakers and breakers equipped with splitter plates. At such extreme temperatures.

- -Symmetrical S . 7 Time E- Asymmetrical Current S .1) The breaking current value above is the value of current at the instant of short circuit. C. However.tnstant of S. This value can be calculated as follows. current ure 11.1 Symmetrical and asymmetrical short-circuit currents. The value of the rated breaking current may include the DC current component. The rrns AC component is The rms asymmetrical breaking current is thus (11. circuit current flowing in an inductive circuit with negligible resistance containing an AC source and a circuit breaker with a fault applied to its terminals. c. The maximum value of the DC current component is where Vp is the source phase voltage and X is the total circuit inductive reactance per phase. 1 Instant of contact separation . C . in which case it is termed the ‘‘‘a~ymmetrical~~ breaking current. the circuit breaker contacts part a few milliseconds .

The rated making current is given by Im = maximum value of DC component component + peak value of AC (1 1. the rated breaking current is the rms value of the alternating component only at the instant of contact separation and in this case is termed the symmetrical breaking capacity. operating mechanism. The rated short-time current is the maximum current that the circuit breaker can carry for l S without damage to its conductors.is the rated m s breaking current in kA. = (11. the duty imposed on the circuit breaker during this process is very severe. The rated making current is defined as the peak value of the current that a circuit breaker can make when closed onto a short circuit. Obviously. the asymmetrical breaking current is taken by designers as 1.2) and in this case is temed the “asymmetrical” breaking capacity. due to the high mechanical stresses produced within the circuit breaker.8. According to British practice. increases the breaking capacity by a factor of 1. the factor 2 (doubling-effect factor) is replaced by 1. To allow for the slight drop in current. insulation. or tank.2) The first current peak occurs after about one-fourth of a cycle from the instant of short circuit.3.6 (Vp/X>.6 (Section 11.1). Normally.adwan later and the interrupted current thus drops below this value (Fig. 11. . In American practice the rated breaking current includes the DC component. The breaking capacity of a three-phase circuit breaker (in MVA) is given by M[VAI. which.3) where Y is the rated voltage in kV and I I .

2b). t is the time between operations (3 min and 0. 1980). Under such circumstances the breakers and dielectric medium suffer severe mechanicaland electrical stresses respectively. respectively. determines the shape of the transient part of the recovery voltage(Fig. C is a closing operation. and capacitive currents.1 and 18.3 S for circuit breakers without and with auto-reclosures respectively). represented by C. 1 In addition to normal load currents. circuit breakers are called upon to interrupt short-circuit currents. Circuit breakers should satisfy one of the following tests e-t-CO-T-CO or CO-t”-CO where 0 is an opening operation. Circuit breakers. The severity of the circuit . The interruption of each of these currents and circuits is presented below. The stray capacitance of the circuit breaker bushing and other connections. The rated insulation level of a circuit breaker is the withstand power frequency and impulse voltage of its insulation (see Tables 18.2a. CO is a closing operation immediatelyfollowed by an opening operation. This is calledthe restriking voltage and builds up across the circuit breaker contacts. although interphase reactions cannot always be neglected(Guile and Paterson. small inductive currents. e- Symmetrical three-phase-to-ground faults at the circuit breaker terminals can be represented by an equivalent single-phase circuit as shown in Figure 11. Examples for small inductive and capacitive currents are the no-load current of a power transformer and the charging current of an extensive unloaded cable network. particularly those equipped with auto-reclosures.3.1.2 in Chapter 18). The duty imposed on a circuit breaker is strongly affected by the magnitude of the interrupted current and also by the circuit parameters and how far the fault is from the breaker terminals. are subjected to frequent and successive operations. 11. T is 3 min. and t ‘ is 15S for circuit breakers not to be used with rapid auto-reclosures.

4) The solution of equation (1 1. . breaker duty is determinedby the value of the short-circuit current together with the shape and amplitude of the restriking voltage. . In case of a fault at point F (Fig. 112 ) and the breaker interrupting it at a normal current zero. .3 +. the restriking voltage can be obtained as follows. The circuit voltage equation at the instant of contact separation is given by (1 1.(b) single-frequency transient recoveryvoltage. . .5) . -i L F Source instant of Contact Separation Current i I Instant ot Arc Interruption I \ 1 System Voltage LL- Sir l ~ I b) Transient Recovery Voltage 1 .4) gives an expression for the recovery voltage across the circuit breaker contacts in the form cos wt exp -cos 21. . mot (11. . Let the supply voltage be in the form e = E cos wt. . 2 (a) Single-phaseequivalentcircuitof asymmetrical three-phase-toground shortcircuit.

The source-sidefrequency& is equal to 1 / 2 n .and positive-sequence impedances. . In actual circuits the effects of resistance and system lossesare considered and the value of the maximum transient voltage is lowerthan 2E. The short circuit current is given by m. the recovery voltage across the first phase to clear will reach three times the maximum phase voltage.. where its value depends on the ratio between the zero.2a with a source voltage e = &E cos wt. The recovery voltage for such case can be obtained as follows.The transient recovery voltage will oscillate at a natural angular frequency wo = 1 / m If . the transient recovery voltage will havemore than one frequency component. 11. the recovery voltage can be approximated to the form e. 11. If the system neutral is isolated or the fault is not to ground. After 90" from the first phase to clear. Each part willproduce an oscillatory voltage. Neglecting damping in the circuit.5.The maximumrate of rise ofthe transient recovery voltage (de. simultaneous interruption in the other twophases occurs.The equivalent single-phase circuit representing this caseis shown in Figure 11./dt).3. This circuit comprises two parts: the source side (1) and the lineside (2). provided that the three phases are balanced.6) and it is equal to E/JZF at a time (n/2) Since the currents in a three-phase system are displaced by 120" from each other.4). The first-phase-to-clear factor is 1.6) The recovery voltage as expressedby equation (1 1S ) is a singlefrequency transient where a high-frequency oscillatory voltage is superimposed on thesupply normal-frequency voltage during thetransient period (Fig. / m while the line-side frequency fi is 1 / 2 ~ t . interruption of the arc may take place not at current zero but near to it at an angle 0. Under certain conditions. When a three-phase fault occurs away from the circuit breaker terminals. It may be less than this value for an earth fault in a system with earthed neutral.2b). while the other two phases are still arcing. = E(1 cos coot) (11. / m . the effect of circuit resistance R is neglected and the natural frequencyof oscillation ismuch greater than the supply frequency. the maximum value of the transient recovery voltage will be 2E and it occurs after a time n m from the instantof arc interruption.. The transient recovery voltage across the breaker terminals is the vector difference of both voltages and is of a double-frequency nature (Fig. can be easily derived from equation (1 1. the arc in one of the phases will be extinguished at a normal current zero. such as relatively low currents and effective interrupting means. Consider the circuit given in Figure 11.

.Source Equivalent single-phase circuit representation fault away from the circuit breaker. of a three-phase im Double-frequency transient recovery voltage.

This current is called the cancellation current and is driven by the voltage e.(b) main current and cancellation current waveforms.7) The process of current interruption in a circuit can be simulated by driving an equal current in the opposite direction (-i).kc. The recovery voltage in the Laplace form is given by e&) = -i(s).Z(s) (11. which is the voltage between the open contacts when the source voltage isshort circuited (Fig. to the original current (i).5a). . between the circuit breaker contacts.8) we ll.fi(a) Equivalent electric circuit for determining the cancellation current. * =cos wt jwL & E (l 1. S 1.

1lSb). = l / L C . Symmetrical short circuits on power systems are not very common.10) Equation (1 1. The power-frequency recovery voltage under asymmetrical faults to ground is equal to the maximum phase voltage when the power system is solidly earthed. and the recovery voltage in the time domain will be (11. Single line-to-ground faults represent more than 90% of the total number of faults on high-voltage and extra-high-voltage systems.where -i(s) is the fictitious driven current at the instant of current interruption and Z(s) is the circuit impedance.9) The recovery voltage is given by 1 . and hence the current at the instant of interruption is i -ic + Imwt (11. Double line faults are rare but they do occur on medium-voltage and low-voltage cables. Line-to-line faults produce relatively lower power-frequency voltages on the faulted phases (&/2 x maximum phase voltage). otherwise the equation is the same as forinterruption at current zero. The sine-wavecurrent near zero can be represented by a straight line (Fig. since the .. SLlS(S2 +&) + L s2LI( S LC ut m.wsin mot). 10) shows that interruptionat non-current zero introduces the first term (-i.

Referring to the circuit in Figure 11. 1983. the arc energy is adequate to keep the arc path highly ionized and conducting u p to the instant of natural current zero when the arc gets interrupted.1/2)i2L2. these high voltages cause arc restrikes and.a circiut breaker rated on the symmetrical fault level will not be capable of interrupting a single line-to-ground fault. The arc current thus gets forcibly extinguished before the natural currentzero (Fig.. . The value of the voltage appearing across the circuit breaker terminals. However. in the case of small inductive currents such as no-load currents of transformers. If the zero-sequence impedance is less than the positive-sequence impedance (X0 XI).6). whenchopping inductive currents.3 with an arccurrent i flowing between the breaker contacts. Having excessive rates of rise. especially the terminal parts of the transformer winding. canbe estimated by considering the energy balance just before and after the instant of arc interruption. Protection schemes against such voltage transients are usually provided. ofenergybetween the electromagnetic and electrostatic fields. Overvoltages produced by asymmetrical faults havebeencoveredin the literature (e.. However. in the case of double line-to-ground faults the duty on a circuit breaker is only So%. When the circuit breaker is interrupting a short-circuit current. Fakheri et al. This “current chopping. Circuit breakers with rated capacity basedon a symmetrical short circuit are not necessarily capable of interrupting asymmetrical faults.” with a veryhigh rate of change. and thus the rating of the breaker is 7’5% of the symmetrical duty and rating. Flurscheim. or by the use of switching resistors. theyhave serious effects on the system insulation. the electromagnetic energy stored in the inductance L2 When the arc is extinguished there will be a successive transfer is ( . capacitive and resistive shunts. on two poles. . and vice versa. 11. more importantly.twophasesbreak the circuit simultaneously. which may be in the form of surge arresters. induces veryhigh voltage transients L(di/dt). the ionized gases get blown off violently and rapidly. 1975).g.Assuming that the arc currentis chopped at a magnitude i. This can be proved by the theory of symmetrical components to determine the short circuit levels for different system faults.11) . the energy balance equation will be in the form (11. and with powerful circuit breakers.

The process of interrupting capacitive currents can be explained by considering the equivalent circuit shown in Figure 11.Transient Voltage Chopping Due to Current Circuit . Current I I I System p\ II ' .! for power transformers. l ' Vo'^--I I lnstant o f Contact Separation I I Voltage Low inductive current chopping. . _./m When a circuit breaker interrupts a purely capacitive current such as that of a capacitor bank or a large network of unloaded cables. The maximum voltage across the breaker contacts immediately after arc interruption is given by (11. . The terms (l/2)i2L2 and ( 1/2)C2vzare the energies stored in the inductance L2 and capacitance C2 at the instant just before arc interruption. respectively./m.12) It is thus clear that the value of the restriking voltage is a function of the chopped current magnitude and The value of depends on the type of load connected to the circuit breaker and may reach 1055. where Cl repre- . .. a high recovery voltage appears across its contacts. This can lead to several restrikes before complete arc interruption. v2 is the voltage across C2 just before the instant of arc interruption. The current flowing during restriking has a high frequency and its interruption may lead to voltage escalation.7b.

L its inductance. This is reasonably true for air- . (c) recovery voltage across circuit breaker contacts when interrupting its capacitive current. The voltage across the circuit breaker contacts half a cycle later will thus reach twicethe supply voltagepeak value (Fig. the voltage on the source side Vs is at its peak value. When the breaker interrupts the capacitive current at its normal zero. The rate of rise ofthe recovery voltage is relatively low and the arc may be interrupted at the voltage peak. The capacitance C2 is charged to the same value and will remain constant if there is no leakage. and C2 the cable network capacitance. sents the stray capacitance of the system on the source side.Generator Unloaded Cable (a) L CB lnstant of Contact Separation 1 I instant of Arc lnterruption I I I 1 Arc Voltage r - “ T I I I ‘ \ J / ’ I Capacitor I i / Voltage VC ure 11.7~). 11.7 (a) Representation of unloaded cable. (b) its equivalent circuit.

In other types. Consequently. If the inductance and capacitance of the line are considered as distributed parameters. Whena fault occurs at a distance of 1-8 km from the breaker terminals. The fault current is inversely proportional to the impedance between the source and the fault. terminal faults produce the maximum fault currents that a breaker has to interrupt.The source-side voltage is a relatively slow rising voltage with a frequency in the range 500-5000 Hz. The restriking transient voltage across the circuit breaker contacts is the vector difference between the two voltage components (Fig. the period of the line-side transient voltage is four times the timeof travel between the breaker and the fault (Chapter 14) and is given by . The line-side voltage is a high-frequency sawtoothed transient. the severity of this duty is less than if the fault is a few kilometers away. the recovery voltage will reach four times the supply voltage.$). 11.blast circuit breakers.13) If this current is interrupted at its first zero. In practical circuits. deionization of the arc column is relatively slow when interrupting small currents and the arc may strike. However. Again. the voltage across the circuit breaker contacts will jump to three times the supply voltage. such as oil-circuit breakers. where the gas blast powerfully deionizes the arc column within half a cycle.8). The frequency of the line-side voltage is in the range 30-100 kHz and is given approximately by S =2X45c5 (11. damping limits voltage escalation and the arc may become finally extinguished after the first restrike.14) where L2 and C. l l . the restriking voltage will be a doublefrequency transient (Fig. are the equivalent lumped series inductance and shunt capacitance of the linefrom the breaker to the fault. At this instant an oscillatory current flows through the circuit with a frequency (11. Theoretically. if restriking occurs once more with arc interruption at the first current zero. this process may continue indefinitely and the recovery voltage becomes escalated.

and h="j V (11. In the case ofa cable the surge velocity is reducedby the factor E.15) where v is the surge velocity and I the length of the line or cable from the breaker to the fault.ShortLine F Pawer Source (a) Source.Both the magnitude and the rate of rise of this voltage will affect the arc interruption process. The surge velocity is equal to l / m .Side Voltage Normal Circuit Voltage 0 .. The previous analysis of the various currents and circuits has shown the nature of the transient recovery voltage across the breaker contacts immediately after arc interruption. 0 0 CB contacts L > Line-Side Voltage (b) Representation of (a) short-line and (b) voltage waveforms. If the voltage buildup is slow. the relative permittivity of its insulation. At the instant of arc interruption the dielectric medium starts to regain its strength while the voltage across the contacts builds up. e.there will be no . where L and C are the line inductance and capacitance per unit length.

In controlled switching the poles open or close at predetermined instants for optimum reduction of overvol- . modern technology always seeks more and more effective alternatives. This can be carried out by an external isolator since the current is relatively low and its power factor is high. the high-frequency voltage oscillation can be critically damped or overdamped. as mentioned in Section 11. If. The value of the switching resistance for damping the voltage oscillations is given by the relation (1 1.. neutral grounding of the system. and surge arresters. the arc will restrike. current chopping may occur with a consequent abnormal voltage rise. electronic circuits are used tocontrol the instants of circuit breakers’ poles opening or closing with respect to reference signals. In HV and EHV air-blast circuit breakers. However. due to the mechanical system a tolerance of f 2 m s is usually observed. closing reactors. shunt reactors. The basic idea of resistance switching is to interrupt the current in two steps.43. the maximum RRV is a function of the gas pressure and properties and the design features of the breaker. In uncontrolled switching the poles of a circuit breaker open or close simultaneously at random. The rate of rise of restriking voltage (RRRV) depends on the interrupted current and its power factor.16) Switching operations produce overvoltages which may damage the circuit breaker and associated equipment insulation. These voltages can be reduced to acceptable levels by the use of conventional methods such as opening and closing resistors. and the presence of any leaking across the interrupted circuit. Rajotte et al. a resistance is connected across the breaker contacts as soon as they part. If the dielectric recoveryis too slow. thus the arc between them gets relatively easily interrupted and the circuit current drops to a magnitude decided by the resistance value. the circuit natural frequency. In controlled switching.riskof arc reignition. on the other hand. First. However. 1996). 1990. In gas-blast breakers. These methods have proved to be effective and reliable since their first use.. the dielectricregains its strength very quickly. transmission lines. and transformers has drawn the attention of many researchers and manufacturers (Holm et al. Controlled switchingof electrical equipmentsuch as capacitors. The second step is to interrupt the current completely. If the resistance is properly selected. the high-frequency transient recovery voltage can be considerably damped by resistance switching. such as in vacuum or air-blast interrupters.

and (Q solid-state. they may replace oil circuit breakers of the same rating in systems where fire hazards exist.l p. while the arcing time is reduced from 10. They are commonly used in low-voltage systems with a normal current up to 3000A. 1995). 1996).07. Also. due to oil contamination. the overvoltage producedisreduced from 2.4 to 1 . by thermal buoyancy and magnetic effects. However. (e) vacuum. deion champers. air circuit breakers are superseding oil breakers. magnetic blasts.1999). 11. These . to the arcing contacts and then to the arc runners (Fig.u. and vacuum. Field tests on circuit breakers equipped for controlled switching showed good results. the maximum value of the inrush current of a transformer is reduced from 3.. They are also used extensively with electric furnaces. (d) SF.5p. for opening reactors. as mentioned before. special magnetic coils are fitted. With the rapid advancement in the fieldof solid-state devices. In heavy industries having large electric motors with frequent starting. The different types of circuit breaker are (a) air. (b) oil.u. of its rated peak current. when energizing a capacitor bank with controlled and uncontrolled closing operation. To help interrupt the arc. the source voltage is monitored by the switching controller and the closing comrriand is issued randomly withrespect to the phase angle of a reference signal (CIGRE Working Group 13. solid-state breakers have a promising future.95 to 2. The main switching media currently in common use are air.39p. When air circuit breakers open. The same reference showed that. under certain conditions (CIGRE Study Committee 13. The arcestablished between the contactsof air breakers is interrupted in atmospheric air.3 to 1. Due to their simple construction and maintenance. The construction and principle of operation of each type are outlined in the following sections.tage or inrush current. sulfur he~afluoride (SF6).9). For controlled closing.8 to 4. and splitter plates are used. there are many conservations on their reliability and economy (CIGRE Study Committee 13. Circuit breakers are classified according to the switching medium in which their contacts part. Incontrolled opening the arcing time for each pole is controlled by setting the instant of contact separation with respect to the current waveform. (c) airblast. To speed the arc movement. It was found that. the overvoltage can be reduced from 2. air chutes.5ms. oil.u. the arc is drawn between the main contacts and then moves upward.

the higher willbe the pressure of the gas bubble produced.7. The greater the arc energy. Oil current breakers are classified as the bulk oil or minimum oiltype. where the arc is freely interrupted in oil.] [From Lythall coils carry the arc current only duringthe arcing period and areconnected in such a way that their magnetic field will force the arc upward. Here oil provides an excellent medium for arc interruption.” By its excessive heat it evaporates and decomposes some of its surrounding oil. The proper designofoil breakers fully exploits these high pressures in fast arc interruption. . The contacts of bulk oilbreakers may be ofthe plain-break type. 11. The latter have higher ratings. (1986).1 Bulk Oil Circuit Breakers Here oil serves both to insulate the live parts and to interrupt the arc. In oil breakers the arc canbedescribed as “self-extinguishing. The methods of arc control and interruption are different from one type to the other.2.3 Arc chute Arcing contacts across hinge Main partsof an air circuit breaker with closed and open contacts. or enclosed within arc controllers.

Plain-break circuit breakers are evidently limited to the low-voltage range. and interruption may occur even before the moving contact has left the pot (Figs.) . Bulk-oil circuit breakers with arc control devices are used in the medium. Minimum oil circuit breakers work on the same principles of arc control as those usedin bulk-oil breakers. The rushing gases and oil will disturb the arc.2.10. Three-phase dead-tankbulk-oil 13. are made of porcelain or other insulating material. mainly hydrogen. the oilwill deteriorate and may lead to circuit breaker failure. 11.11). produced by oil decomposition.7.and high-voltage ranges. 11. (Courtesy of South Wales Switchgear. 11. When they part.The head of oil abovethe arc should be sufficient to cool the gases. The explosion pot encloses the fixed and moving contacts. with their associated handling and storage problems. however. the arc is drawn between them inside the pot.8 kVcircuit breaker in its close position. They consist mainly of a large volume of oil contained in a metallic tank. Arc interruption here depends on the head of oil above the contacts and the speed of contact separation. Their containers. The gases so produced in the confined space will cause highturbulence of the oil and rush outside the pot.2 ~inimum Oil Circuit Breakers Bulk oil circuit breakers have the disadvantage of using large quantities of oil. In this type of breaker the arcis confined and interrupted in “explosion pots” rather than in the open volume of oil.With frequent breaking and making of heavy currents.

(b) chemicalstability and inertness of air. Fresh and dry air will thus rapidly replace the ionized hot gases withinthe arc zone and the arc length is considerably increased. Equalizing resistors or capacitors are connected in shunt with each interrupter unit to ensure uniform voltage distribution across the breaks (Soderberg. (b) cross jet. arcing is confined to a much reduced volume of oil inside an explosion pot.re 11. (c) great . multibreak breakers are constructed from a number of modules in series. which is contrary to other circuit breakers. The reason for this is the high turbulence caused during large current interruption. Consequently. and vice versa. Single-break minimum oil breakers are available in the voltage range 33-132kV with breaking capacities from 1500 to 5000 MVA. Oil circuit breakers are characterized by their long times of interrupting small currents and short times for large currents. In this type of breaker. shows one pole of a minimum oil circuit breaker. the arc may be interrupted at the first current zero. For higher voltages and ratings.12. The principle of arc interruption in air-blast circuit breakers is to direct a high-pressure blast of air longitudinal~y or perpendicularly at the arc. The merits of air-blast breakers are (a) cheapness and availability of the interrupting medium. The lower chamber contains the operating mechanism and the upper chamber contains the moving and fixed contacts together with the arc-control device. 1978). Figure l 3.11 Explosion pot arc controllers: (a) simple type.

However. 4. Uponarcinterruption. (d) high-speed opreation. 3.lower main terminal. (f) operation infire hazard locations. 5. 1. air-blast breakers . 6. (e) short arcing time. arc controlchamber. the useof air compressors and high-pressure vessels increases the production cost. Switzerland. contact roller. crank housing. 10. vent. air chamber. $.5 ure 11. (g) reduction of maintenance frequency. oil level of Asea Brown Boveri AG. upper main terminal. contact rod. 9. tulip contact. 2. (Courtesy reduction in the erosion of contacts from frequent switching operations.12 One pole of a 12 kV minimum-oil circuit breaker. 7.) observation glass. and (h) consistent breaking time resulting from use of the interrupting medium pressure to open the contacts.

having eight interrupter heads per phaseand equipped withswitching resistors and equalizing of Asea Brown Boveri AG. The contact separation should be enough to interrupt the arc and withstand the recovery voltage under high pressure.13). but it rapidly builds up to separate the contacts by piston action. Designs with voltages up to 110kV are equipped with series isolators to open the circuit while the interrupter headis pressurized. air-blast breakers should be equipped with silencers to reduce the noise to an acceptable level. 11. When the air supply is stopped. The number of breaks per pole or phase depends on the voltage level and it is Air-blast circuit breaker rated 10. contact separation is achieved by admitting highpressure air fromthe air receiver. the contacts reclose under spring action. Prior to contact separationthe air pressure in the interrupter head is atmospheric.000 MVA at 300 kV. (Courtesy . odern designsusepermanently pressurized interrupter heads and airflow pipes (Fig. The relative positions of the interrupter heads and the admission valve ensure simultaneous operation of the breaks. In residential areas.) capacitors. In many designs. Switzerland.3 produce high-level noise when discharging to open atmosphere.

and F. The velocity and pressure of the air blast are independent of the interrupted current. it decomposes to SF4. respectively. when it interrupts a small inductive current. and rubber sealings in the presence of moisture. F2. 11. where arc interruption relies on the turbulences produced in the oil. bushings. At apressure as high as 14 . However. It has been used extensively during the past 30 years in circuit breakers. The gas should therefore be meticulously dried. Thearcing time of airblast breakers does not vary considerably with the interrupted current. at temperatures of the order of1000°C. Most of the decomposition products canbe absorbed by a mixture of soda lime (NaOH + CaO) and activated alumina placed in the arcing chamber.4. In SF6 breakers the decomposition products attack the contacts. SF6 ischemically inert and does notattack metals or glassunder normal conditions. Thus. S. as explained in Section 11. A s with EHV minimum oil breakers.3. The decomposition products recombine shortly after arc extinction (within about lps).4. and gas-insulated cables (Chapter 10). as in electric arcs. S2.customary to use fourand eight breaks forthe voltage levels230 and 750 kV. One of the major problems associated with the use of SF6 is its condensation at high pressures and low tempertures.1 Properties of SF6 Sulfur hexafluoride has a high thermal conductivity and its thermal time constant is about 1000 times shorter than that of air. The velocity of soundin SF6 is about 40% of that in air. chemical. a great advantage in arc quenching. metal parts. 1981). The resistors help to interrupt the arc andlimit transient overvoltages. switching resistors and equalizing capacitors are usually fitted across the interrupters. gas-insulated switchgear (CIS). there is a chance of its chopping. The high dielectric strength and good arc-quenching properties of SF6 are ascribed primarily to its high affinity for electron attachment (Chapter 3). The highly toxic S2F10 has never been traced in the arcing products. SF2. high-voltage capacitors. and therefore the required gas flow for arc interruption can be produced during an opening operation without the need for a gas compression plant (Schumann and Evans. The physical. Sulfur hexafluoride is an excellent insulating and arc-quenching medium. The capacitors are used to equalize the voltage across the open breaks.i'. which is different from the case of oil circuit breakers.and electrical properties of SF6 are superior to many of the other media.

As with air-blast circuit breakers. The coil is dis- .2 Double-Pressure SF6 Circuit Breakers This is the early design of the SF6circuit breaker.7. the circuit breaker contacts part while the high-pressure gas is released from its reservoir to the interrupter compartment. SF6 liquefies at 0°C. by piston (10) action. are used to overcome some of these problems. the contacts separate and an arc is drawn between them. Both have the same gas pressure. the gas is pumped back to its reservoir. while interrupting smaller currents. This force makes the arc rotate around the periphery of the contacts. large and small. and the currents. about 5 atm. 11. For the current-breaking operation. Its operation is similar in principle to that of air-blast circuit breakers. A third compartment is incorporated to augment the gas pressure. A coil (2) surrounding the main contacts speeds up arcinterruption by exerting a magnetic force on the arc. It comprises mainly a highpressure metal reservoir. The gas blasts from the arc compartment to the other compartment (7). The construction and principle of operation of each type are presented in the following sections. such as nitrogen. 11. SF6 is also highly sensitive to strong localized fields. where most of the SF6 is kept. moisture. breakers of this type have been outmoded by simple designs of the selfextinguishing and puffer types. and their complicated design and construction. get interrupted at their natural zero (Schumann and Evans.7. while the breaker is closed. By this arrangement the arcing time isindependent of the current magnitude. The heat generated in the arc heats the gas in the arc compartment (6) and rapidly increases its pressure.4.6-765 kV. and an i~terrupter compartment containing the breaker contacts. EHV SF6 breakers comprise a number of “interrupter compartment” modules in each phase. Because of its need for various auxiliaries. 11.14). where it blows out the arc. thus cooling it and avoiding spitting of the contacts.4. filters. control devices. SF6 breakers to be used in areas where such low ambient temperatures are encountered may be equipped with specialheaters for the gas. In ~edium-voltage circuit breakers the interrupter compartment is a grounded metal “dead tank. 1981).3 S~~f-extinguish in^ SF6 Circuit Breakers The interrupting chamber isdivided into two main compartments (Fig. SF6 circuit breakers cover voltage levels in the range 6. After the current interruption.” whereas in EHV breakers the “live tank” type is more suitable. Mixtures of SF6 with other gases. such as gas compressors. monitors. and foreign solid particles (Chapter 10). This rapid expansion cools the arc column and extinguishes it at a current zero. When the breaker is being opened.3 bar.

1 Mainc ~ n n ~ t i o n s 2 Cylindrical coil 3 Loadcurrentcontact 4 Fixed arcing contact 5 Moving contact 6 Breaking chamber 7 Pressure equalizrng chamber Operatrng shaftwlth rotary seal 8 Operating lever 10 Auxiliary compression piston l ? Transmission casing 9 Section through ase~f-extinguishing SF.) . Switzerland.circuit breaker. (Courtesy of Asea Brown Boveri AG.

Their main shortcoming is that a failure in their hard vacuum cannot be easily detected while in service.16. Their principle of arc interruptionis by compressingthe SF6 gas during contact separation. lS. up to 765 kV. the power needed to close or open it is much less than for other types of breaker.l4). 11. In vacuum breakers a contact separation of about 1 cmis adequate. 11. the current flowsbetween the current (1) via the main contacts: the fixed (2) and the moving contact (3). Modern vacuum breakers successfully interrupt capacitive currents and small inductive currents and short line faults without producing excessive transient overvoleing compact. rated current up to 2350 A. where two modules per phase are sufficient in a 362 kV breaker. with their simple mechanism.4 Puffer-Type SF6 Circuit Breakers These are also sometimes called “single-pressure” or “impulse”-type breakers. including compressed gases and oil. the moving cylinders of (3) and (4) act as pistons and pump compressed SF6 gas. Breakers of this type are normally used in medium-voltage (MV) networks with rated voltage up to 24 kV. The advantages of vacuumas an insulant and arcinterrupting medium have Pa. through the insulating nozzle (7). which flows axially and blows out the arc.7. the contact travel. Its operation can easily be visualized by reference to Figure I l . Since it isso compact. theydo not need much maintenance. on the order of lom4 dielectric strength and an arc-interrupting ability superior to those of other media. a number of modules are arranged in series on insulating supports as shown in Figure 11.17a). A hard vacuum.connected from the main circuit during normal operation and is in series with it immediately when the arc is established between the contacts (Fig. In a vacuum circuit breaker each phase consists of an evacuated interrupter compartment and an external operating mechanism (Fig. the arc is dra een the arcing contacts (4) and (S). and rated breaking current of 3 1. The rate of dielectric recovery of vaccum after arc interruption is about one order of magnitude faster than in air-blast breakers. . has a been known for many years. with the movingcontact acting as a piston.4. l l . ile the breaker isclosed. When the breaker is being opened upon contact separation.S kA. For the IEHV range. Thus the gas pressure in the interrupter compartment builds up rapidly to levelshigh above its steady value of about 3-5 atm.

The wall of the interrupter compartment is made either totally or partly of an insulating material (e. moving. sheath seal. 1. 1982). with spiral segments so that the arc current produces an axial magnetic field (Fig. . 12.) The contacts are of large surface area. arcing contact.. 8. insulating cylindrical enclosure. 7. Switzerland. Moving the arc spots over the contacts minimizes metal evaporation and thus arc erosion (Chapter 6). 11. arcing contact. operating level. fixed.. fixed. exhaust compartment.g. 4. (Courtesy of Asea Brown Boveri AG. glass).. moving as a piston. compression cylinder. transmission link. 10. 3. 1983). insulating nozzle. 1l . main contact.13. lever fulcrum. 6. 17a) make possible the movement of the moving contact while maintaining the hard vacuum (Sunada et al.5 Section through a puffer-type SF6 circuit breaker.17b) to help move the arc over the contact surfaces and interrupt it rapidly (Yanabo et al. 11. The metal bellows (Fig. 5. main contact. current terminals. 9. 2.

The metal vapor swept radially outward from the arc column is bound to condense on the walls of the interrupter compartment. the arc iswhirled magnetically (Fig. Therefore. When its contacts part. 17a. Therefore. intense ionization confines the arc within an intense core. this part of the wall is already metallic.) In a vacuum breaker the process of arc extinction differs from that in other types of breaker. (Courtesy of GEC Switchgear. other parts take care of the insulation strength. their last points to separate get heated up to the metal’s boiling point. At low currents thearc is characterized by a diffuse appearance (Lafferty. In the breaker illustrated in Figure l l. 1980). Therefore. 11. Here the arc is burning almost exclusively in metal vapor. withlarge currents of kiloamperes. insulating parts of the wall facing the arc region should be shielded from the metal vapor to keep their insulation strength.362kV two-~reakthree-po~e puffer SFG circuit breaker. . whereas. the metal or alloy of which the contacts are made directly affects the characteristics of the vacuum breaker.17b) in order to minimize contact erosion. causing excessive heating and localmeltingof its spots on the contacts. The ionized metal vapor thus evolved provides the arc medium (Chapter 6).

igure 11. 2.17 (a) Construction of 24 kV vacuum breaker. contact pressure spring. lower breaker termcoupler. 1. inal. insulating (Courtesy of Siemens. 6. 5. axial magnetic field . 8. interrupter body. metal bellows. fixed contact: 3. moving contact: 4. 7. upper breaker terminal.) (b) Details of the contacts for the current I to produce an B in the vacuum.

triacs. 1sa). ll . . it will still interrupt thefault current within only about one half-cycle Oscillograms of currents handled by the solid-state circuit breaker: (a. Solid-state breakers can clear a fault within only about one half-cycle(Fig. or power transistors.(b) current flowing when the breaker closes afaulty circuit. top waveform) current interrupted within about half a cycle from the instant it exceeds the overcurrent setting. If the breaker closesa circuit that happens to be faulty.This type of circuit breaker uses solid-state devices such as thyristors. They do not suffer from the shortcomings of electromechanical devices.

(Fig. Switching transient overvoltages can also be eliminated. 19. The principle of operation of solid-state breakers is illustrated by the simple circuit shown in Figure l l .c" (c and dl continued. It has no moving contacts that get erodedby the electric arc.18b) (Khalifa et al.. the current sensor and the gate Simplified circuit with solid-state circuit breaker. A s the instantaneous circuit current exceeds the set level. (d) input current when the instantof circuit closing is properly synchronized.b). 11. 1979a. (c)inputcurrent in acircuitwherethecurrent has avery pronounced DC component. Each thyristor will allow the load current to flow during one complete half-cycle when properly triggered. . and thus does not need to bemaintained or replaced. The instant of making a circuit with a solid-state breaker can be synchronized such that the doubling of inrush currents into inductive circuits and the accompanying electromechanical overstresses on the circuit conductors and supports canbe eliminated. In this figure two power thyristors are connected antiparallel.

Before the main circuit breaker is opened. 11.18a). The self-saturated reactor L2 and capacitance C help to reduce the rate of decrease of the circuit current and the rate of increase of the transient recovery voltage. Not all types ofcircuit breaker used in HV AC systems can be installed HV DC systems unless equipped with additional circuits to bring the full C current smoothly to zero for arc interruption (Yanabo et al. The basic principle of such circuits is shown in Figure 1120.. . solid-state breakers have been usedin few power system applications in the low-voltage range. In some low-voltage and low-current power systems applications. The main disadvantage of solid-state breakers is their relatively high power losses. 1982). solid-state breakers for higher voltages and currents can be constructed.3 control device will cut off the gate signals. They are installed in some traction systems. respectively. * R Saturable C Principle of commutation circuit for DC interruption. Air-break circuit breakers are usually used to interrupt low DC currents at low voltages. electrochemical plants. a triac may replace the power thyristors. They are point-to-point transmission systems. High-voltage DC lines are now commonly used to transmit large bulks of electrical energy for long distances. switch S is closed to discharge the precharged capacitor C1 inthe opposite direction of the circuit current. This will turn off the thyristors and the circuit current willbe broken at the ensuing zero crossing (Fig. This situation may be much improved with the development of new semiconductor materials. Up until now. and similar applications. By series-parallel connections of the thyristors. They cannot be tapped or paralleled because of the lack of HV C breakers with high interrupting capacity. This will force it to zero with a few oscillations and the arc will thus be interrupted at a current zero.

Four breaker modules are connected in series (Fig.u. impedance of the system up to the 66 kV busbar is 3 p. As soon as this current zereo is produced. switch SIis closed and this causes a current diversion into C. The breaker is a modified SF6 puffer type. If a three-phase fault occurs at the 66 kV bus.8 kV. due to arc lengthening and cooling. a current zero is produced in the interrupter. 11. The p.21 Schematic of a 500 kV WV DC SF6 puffer-type circuit breaker. When its magnitude is large enough. (1) A powersystemwith1000MVA generating capacity feeds a heavy industrial area at 66 kV. the arc is interrupted and the full circuit current is diverted to C. [From Lee et al..] Lee et al. The circuit energy isabsorbed by the arresters and the fault current is cleared in a few milliseconds. an arc is drawn between them. As the contacts continue to open. 1. (1985). When the breaker contacts start to open to break the circuit current. (1985). Both breakers included commutating circuits and energy-absorbing elements. Air-blast and SF6 puffer 400 kV DC circuit breakers have been developed by Vithayathile et al. The circuit inductive . it will conduct and stop any further increase of voltage across the contacts. When the voltage across C. The current diversion results in a current oscillation that grows. the arc voltage increases..21). reaches the clipping voltage of the ZnO arrester. (2) A three-phase synchronous generator has a line voltage of 13. (1985) described a 500 kV DC circuit breaker to be used for switching load and fault currents up to 2200A.c S2 S2 ure 11. After the buildup of enough arc voltage. calculate the symmetrical and asymmetrical breaking and making currents of circuit breakers located at the point. The generator is connected to a circuit breaker. The speedof operation of these breakers is comparable to that of AC breakers of similar types.u.

Calculate also the value of resistance to be connected across its contacts for critical damping of the restriking voltage. The voltage at the capacitor bank side is 220 kV. Also prove that the first-phase-to-clear factor is 1. Determine the following: 1. (6) In a system of 220 kV. calculate the time taken for arc interruption from the instant of contacts opening. .015 yF respectively.. sin wt.5 cycle If a fault occurs on the feeder. If arc interruption occurs at the second current zero. Calculate the value of a resistance to be connected across the circuit breaker contacts to prevent voltage transient when switching the charging current of the bank.. is connected to a capacitor bank through a circuit breaker.1 I. 0. Average rate of rise of restriking voltage 4. Time to peak restriking voltage and to maximum rate of rise of restriking voltage (3) A generator/transformer unit of 200 MVA. the circuit inductance is 5 H and its capacitance to ground is 0. Frequency of restriking voltage transient 3. The mainparticulars of the relay and breaker are as follows: Relay operating time CB contact speed time Arcing 60 ms 10 cm 10 m/s 0.01 pF. rove that the firstphase to clear in a three-pole circuit breaker interrupts its current 90" ahead of the other two phases. (7) An oil circuit breaker is installed to protect an l1 kV feeder with an overcurrent relay. (4) A 60 Hz circuit breaker interrupts a current in the form: i = I.45 per unit impedance.reactance and capacitance upto the circuit breaker are 4 Q and 0.Calculate the voltage appearing across the pole of a circuit breaker if a current of 8 A is chopped at its peak value. Maximum rate of rise of restriking voltage 5.5. The circuit breaker's contacts open at a current of 0. Assume the system frequency is 60 Hz. The rating of the condenser bank is 20 MVA. Peak restriking voltage across the circuit breaker contacts 2.' after the peak value. calculate the total time needed for fault clearing in seconds and cyles.

The alternator series inductance is 15mH and its stray capacitance is 6 pF. assuming full asymmetry 3.In a 60 Hz circuit breaker. and the contact’s stroke is 6 cm.007 78pF/km respectively. The natural frequency of the circuit 3. deduce the lineside natural frequency and show by what factor its exceeds the value obtained by considering the line inductance and capacitance to be lumped. in seconds and cycles. the time taken by the first phase to clear as well as the other two phases to interrupt the arc.012 pF/km.187 pF/km respectively. Calculate. star-connected. 2. 13.5 km away from a circuit breaker on a transmission line having an inductance of 1. If a short circuit occurs l . Comment on the results.2mH/km and a capacitance of 0.155 mH/kmand 0. calculate the samevalues in 1.8 kV synchronousalternator has its neutral solidly earthed. The m s voltage of the testing source z transmission line has an inductance and capacitance per phase of 0. The first phase to clear interrupts the arc 1Oms after reaching full stroke. The value of the maximum RRRV 5. the following readings were obtained: Time to reach the peak restriking voltage Peak restriking voltage Calculate: 1.If a terminal three-phase fault occurs on the circuit breaker. calculate: 1. The alternator is connected to a power system through a circuit breaker. the speed of the moving contact is 3 m/s. The average rate of rise of the restriking voltage 90 ps 130kV 2. The steady-state interruptedcurrent 2. and 3 above. The interrupted current. The time at which maximum RRRV occurs 4. The maximum rate of rise of the restriking voltage If a cable of the same length.933mH/km and 0.isused instead of the transmission line. obtain an expression for the restriking . L A three-phase. 60 Hz. In a short circuit test on acircuit breaker. having an inductance and capacitance per phase of 0. If a symmetrical 3-phase fault occurs 6km awayfrom the circuit breaker protecting the line.

The alternator neutral is earthed through a 1 st reactance and at the instant of fault it has a positive.voltage across its contacts if arc interruption occurs 8" ahead of current zero. and 3 st respectively. and single line to ground faults.01 5 pF/ km. (14) A 60 Hz generator is connected to a transmission line 1. star-connected. If a short circuit occurs on the circuit breaker terminals. prove that the frequency of oscillation ofthe transient recovery voltage is given by r C ICB / F Equivalent circuit for Problem 13. The source inductance. stray capacitance.22). 11kfl. 11. Calculate the source-side and transmission lineside frequencies ofthe recovery voltage appearing on the circuit breaker poles when clearing a threephase fault occurring at the far end of the transmission line. (15) A 3-phase. calculate the initial rate of rise of the line voltage side. double line. .5 km long having an inductance of 1.01 pF. 60 H z alternator is protected by a circuit breaker. R. If the short circuit current at the far end of the line is 3000 A. negative. Obtain and compare the ratings of the circuit breaker based on three-phase./ phase and a stray capacitance to ground of 0. Obtain also the voltage of the neutral point above ground under a single line to ground fault.C. and r respectively. The generator side circuit has an inductive reactance of 5 st. and switching resistance are L. and zero sequence impedance of 10. (13) A circuit breaker is installed to protect a generating source as shown in the single-phase equivalent circuit (Fig. resistance. 7.2 mH/km and a capacitance of 0.

1996. p49. 1979a. Akesson U. Paterson W. p 32. New York: John Wiley. London: Butterworth. pp. Evans GJ. 1986. 1978. Rajotte J. Schumann R. Electra 183. UK: Pergamon. Arifur-Rahman S. Slade PG. Alvinsson R. 1975. CIGRE Working Group 13. pp 103-106. Arifur-Rahman S. Ware B. Proc IEE 126:75-76. Enamul-Haque S. Hung H.Theory and Applications. Stevenage. Power Circuit Breakers Theory and Design. Charpentier C. Guile AE. Khalifa M. IEE Monograph 17. Soderberg G. Proceedings of Conferenceof the ElectricitySupply Engineers Association of New South Wales. Brault S. Vithayathile J. IEEE Power Eng Rev 10:32. 2. Holm A. Seattle. 1980. The stray capacitance to ground is 8 nF and the magnetizing inductance is 50 H. Sydney. CIGRE Session 1990. Electra 164. WA.07. 1985.1983. 51:3-5. Report 13-201. Lythall RT. hatt N. Flurscheim CH. Electric Power Systems. Vacuum Arcs. UK: Peter Peregrinus. Paper 13-301.A circuit breaker is connected to the high voltage side of a 30 132/11kV transformer. Calculate the voltage appearing across the circuit breaker contacts due to chopping the magnetizing current of the transformer at its peak value. 1981. Lafferty JM. Vol. Proceedings of IEEE Industrial Commercial Power System Technical Conference. 1999. Lee A. Enamul-Haque S. Khalifa M. Desmarais J. Belanger. Porter S. The J & P Switchgear Book. What is the value of a switchingresistance required for critical damping of the transient recovery voltage? What is the differencebetween the symmetrical and asymmetrical rated breaking and making currents of a circuit breaker? Why does the arcing time in oilcircuit breakers decrease withincreasing arc current? Does this occur in air-blast breakers? Why are oil circuit breakers not recommended for installation in power systems with a high frequency of load interruption? Explain the principle of magnetic blowout in air-break circuit breakers. CIGRE Session 1996. Sybille G. IEEE Trans PAS-102:33153328. 1980. 12-1-12-12. Karlen 0. 1979b. CIGRE Study Committee 13. Oxford. Yoon ISH. J. ASEA J. .

Koide T. IEEE Trans PAS-102:1395-1402. Yanabo S.Kanai Y. IEEE Power Eng Rev. Vithayathile JJ. Proceedings of CIGRE Conference on Large HighVoltageElectricSystems. Tsuneo H. . Yanabo S.Sunada Y. Hingorani NG. Awaji H. 10:29. Tarnagawa T. Shoichi I. 1982. 1985. Nilsson S.101: 19581965. Poryer JW. 1983. Okumura H. Ito N. Courts AL.1982. Yanabo S. Report 13-04. Kaneko H. IEEE Trans PAS. Tohru T. Shozo T. Peterson WG.

No interruption of supply. insulator flashover. underground cables are installed in densely populated regions even when their cost is much higher than that of overhead lines. No wayleave troubles that may occur with short circuits of overhead lines crossing main roads 4. Assiut.Hig h-Voltage Cables Assiut University. underground cables have been used in power distribution networks in cities and densely populated areas. The last few decades have witnessed great development in cable technology.. . Cables are usually more expensive than overhead lines at all supply voltages. Underground cables can now supersede overhead lines for electric power transmission over short distances. No liability of accident to the public 3. even under severe weather conditions such as thunderstorms (on overhead lines. result in supply interruptions) 2. No objectionable effect on the aesthetics of the environment Therefore. Egypt Since Ferranti’s first cable in 188l . Cables have the following advantages compared with overhead lines: 1. short circuits. etc.

8. Therefore. Under certain circumstances. 1982).95%). three. 1982). When there is more than one layer of strands. The cable has an overall sheath.2 03 0. For all types of power cable.4 */e ( Impurity ) . and 2 at 400. The conductivity of aluminum is only 60% that of copper. Any cable used in the power industry usually comprises one.Readers interested in DC cables are invited to consult specialized books on cables eedy. King and Halfter. 12. . DC transmission has many advantages over AC. withparticular emphasis on fundamentals.l Effect of impurities on the conductivity of copper. Each core is a metallic conductor surroundedby insulation. a larger aluminum cross-sectional area is required for the same current-carryi~g capacity.with a cost ratio of about 20. asin the case of long submarine transmission. 1988. but now the difference has diminished. alternate layers are spiraled in opposite directions. The conductorsare usually strandedto secure ~ e ~ i b i l i t Aluminum y. armoring is addedfor mechanical protection. DC cables have their recommended applications. copper and aluminum are in common use. The metal purity is very important (>99.1). each core has a metallic screen. was considerably cheaper than copper. In HV and EHV multicore cables. In this chapter we present the many aspects of AC high-voltage cables. respectively (King and Halfter. Impurities seriously reduce the conductivity (Fig.1 0. 100 I I I l U 0 60 0 0. 132. Both solid and stranded aluminum conductors are presently used in cable construction. For underground cables. or four cores. and l1 kV.

igh dielectric strength 2. The increase in conductor resistance with stranding is significant only in the case of aluminum. Sodium is abundant in seawater and is easily and cheaply manufactured by electrolysis of NaCl. Easy handling. Also. Now polymersare in common use. Insulating materials for power cables are generally classified as (a) impregnated paper. Low relative pemittivity (E. Economy Cable insulation used to be oil-impregnated paper. sodium has virtually no mechanical strength. 1987). In Table 12. It is therefore impregnated with oil to improve its electrical properties. and installation 8. Typical values of 50 Hz breakdown .. Sodium conductors would be considerably larger in diameter than equivalent conventional ones. evolving hydrogen. Sodium has been proposed for use as acable conductor (Humphrey et al. 1966). due to increased strand length. Sodium possesses larger values of specific heat and latent heat of fusion. where the oxide layer prevents effective contact between the strands. and (c) compressed gases. when sodium comes in contact with moisture a violentchemical reaction takes place. ~ r e ~ ~ aPaper ted Paper is well known to be hygroscopic.1 are given the ranges of voltage ratings in use for cables insulated with oil-impregnated paper and synthetic materials of the oil-filled. Thus sodium cables have good stable performance under short-circuit conditions. manufacture. with oil under high pressure in HV and EHV cables. The main characteristics that cable insulating material should possess are: 1 . On the other hand.Flexibility obtained by spiraling of the strands is at the expense of a small increase in cable resistance. as the conductivity of sodium is about 30% of that of copper. Reasonably long life 5 . Sufficiently low thermal resistivity 4.) and loss tangent (tan 6) whenused in AC cables 6. Crosslinked polyethylene (XLPE) is used in LV and HV cables. Chemical stability over a wide range of temperature 7. (b) synthetic materials. Attempts to use EPR in HV cables have been carried out (Ross. High insulation resistance 3. and compressed-gas types.gasfilled. which is highly flammable. poly(viny1chloride) (PVC) and ethylene propylene rubber (IEPR) are limited to LV cables.

.

armoring is required.2. and 50-60 MV peak per meter after impregnation. Freon (CCl2F2). Experimental and theoretical investigations have revealed that conductor stranding can increase the maximum electric field in the cable insulation by about 20-30%. More detail is given in Chapter 8. a metallic sheath is essential for mechanical protection against handling and to prevent ingress of moisture to cable insulation. lead alloys containing small amounts of tin. aluminum sheaths have been introduced.5. and better economy. SF6 has several advantagesas an insulating medium. thereby minimizing the possibility of surface discharges due to tangential stresses. Screening the insulation around each core of a multicore cable confines the electric field and makes it symmetrically radial. 12. Sheathlosses are discussed in Section 12. orin For cables subject to mechanical stresses. Cables insulated with compressed SF6 are discussed in Chapter 10. 12.improved mechanical properties. and nitrogen (N2) are used for insulation. conductor screens in the form of aluminum foil or semiconductor carbon paper tape are lapped over the stranded conductor. Laminated insulation usually has higher dielectric strength under normal than under tangential field stresses.3 Compressed Gases In some EHVcable installations. The drawbacks of aluminum sheaths are that they are less chemically inert and have a higher sheath loss than that of lead. A bedding of hessian iswound or a PVC sheath is extruded over the metal sheath to . Inthe past. To mitigate this effect. 1986). The main advantage claimed for aluminum-sheathed cables are about a 50%savinginweight.2. XLPE. metallic sheaths used to be manufactured from lead. A s explained in Chapters 4 and 10. or cadmium were commonly used for cable sheaths. compressed gases such as sulfur hexafluoride (SFG). High-purity lead is not only expensive but is mechanically weak and would crack under vibration.stress are 6-8 MV peak per meter when tested in thin layers before impregnation. For paper-insulated cables. antimony. EPR.2 Synthetic Materials Polymerssuch as PVC. and polyethylene (PE)showsome technical advantages over impregnated paper when used to insulate underground cables in domestic and power distribution networks (Olsson and Hjalmarsson. Recently.2. Therefore.2.

belted cables are restricted to voltages of less than 33 kV where the tangential-field component throughout the insulation is fairly insignificant (Weedy. either by oil in oil-filled cables or by gas in gas-filled and compressed-gas-insulated cables.3. The electric field is no longer radial to the conductor as in the case of single-core cables. Solid-type cablesinwhich the pressure within their insulation is atmospheric 2. as explained in Sections 12.3. heat is generated which may eventually lead to breakdown (Chapter 8). . as in H-type cables. For voltages exceeding 220 kV. 1988. The cable is therefore equivalent to three single-core cables with a radial field in each. Then extra insulation is applied as a circumferential belt over the three cores to provide sufficient insulation to withstand the phase voltage between each conductor and the sheath. The conductor screens are in electrical contact with each other and with the overall metal sheath. The cable size bcomes less bulky and more economical. The armoring consists of steel tapes or wires. For this reason. An outer serving of hessian or PVC sheathing is then applied over the armoring to protect it against corrosion (Ross. three-phase cable insulation is more efficiently employed if the cable design takes the form of three single cores rather than a three-core belted cable.5. 1987). Power cables can be classified into two distinct categories: 1.provide a mechanical cushion and chemical insulation between it and the armoring.Pressurizedcables in which the pressure isalways maintained above atmospheric. 1982). to help raise the insulation strength. Sometimes. Three-core belted cables use impregnated paper wherein each of the conductors is insulated for half of the line voltage. Subsequently. in belted cables there is a tendency for leakage currents to flow along the layers under the tangentialcomponent of the field. King and Halfter. Tangential field components in beltedcables can be eliminated by screening each core separately. nonmagnetic wires (bronze) are used to increase the magnetic reluctance and thus minimize losses. Therefore. A serious difficulty that occurswithbeltedcablesis due to the electric field distribution throughout the insulation.4 and 12. Paper insulation is weakerunder a tangential electric fieldthan under a radialfield.

and XLPE-insulated cables installed in wet environments indicate the deleterious effects of moisture. The reliability of paper-insulated cables has been so good that the introduction of thermoplastic materials in this field has not been rapid.3.or the gas is compressed by applying static pressure on the paper insulation in “gas-filled” cables. 12. Other advantages of plastic-insulated cables include lighter weight. as explained in Section 12. With cyclic loading.2 shows the constructionof a typical plastic-insulated cable. will stand higher electric stresses. This is discussed in more detail in Chapter 8.Another serious problem with paper-insulated cables results from the inelastic property of the lead or aluminum sheath. Reports on the performance of PE. the long-time breakdown strength of an oil-filledcableis approximately twice that of the equivalent paper-insulated solid cable.3). With time.beingcornpressed. 1983). King and Halfter. two advantages of plastic-insulated cables have accelerated their use: the resistance of plastic sheath to the ingress of corrosive moisture. Oil-filled cables are paper insulated and always completely filled with oil under pressure (3-4 atm) from oil reservoirs along the route.3. A gradient in theprofile will produce arise .3. 1982). This includes a decrease in breakdown strength and insulation resistance and an increase in dielectric loss due to water treeing (Bottger et al. Their contained gases. andcolored cores for instant identification. Moreover. and the omission of the lead or aluminum sheath whichisessential forpaperinsulated cables (Weedy. the oil acts as a coolant (Fig. the insulation and sheath expand with different rates and do not return to their original dimensions upon cooling. voids are formed within the volume of insulation. as explained in Section 12. For this reason. Following the successful application of PVC cable designs and with the advent of t h e ~ o $ e t t i n g materials. reducedliability to damage during installation.4. The design of the hydraulic system for oil-filled cablesdepends mainly on thecable route and its profile. Void formation seems to be a main cause of service failures of cables rated at 66 kV and above. Nevertheless.. The oil pressure inhibits the formation of voids. as explained in Chapter 8. 1988. 1987).and threecore cables were introduced for l llcV and higher voltages (Brown. XLPE-insulated single. Figure 12. For this reason the gas in the voids is replacedby oil under pressure in “oil-filled” cables. These voids cause losses and ultimate breakdown of the cable insulation.

3 Cross-sectional views of typical oil-filled cables. c .tandard construction of extruded p . bedding > ” single wire armour - extruded p . outer sheath Construction of a low-voltage PVC insulated cable. v . v . . c . oil duct \ \ core ure 12.

The reservoir consists of a number of atmospheric gas-filled sealed cells within a tank completely filled with cable oil and connected to the cable. the breakdown strength of the cable dielectric is greatly influenced by the time span during which the voltage is applied. the gas acts as a coolant and the steel pipe provides mechanical protection for the cable. There are. short-term values of breakdown stresses have to exceed the long-term operating stress by a suitable safety margin. the excess oil flows along the oil ducts in the cable to the pressure reservoir. A pressure reservoir at one cable end is applied to short lengths of cable routes. During cable heating. but drops to less than 20 MV/m for durations of 50-100 h. their choice being dependent on the length of the cable. Different oil-filling methods have been used in practice. . Moreover. thus putting the insulation under pressure. The voids that may form in the cable insulation are filled with the gas at a pressure sufficient to suppress ionization. where it compresses the sealed cells. usually nitrogen. For comparatively long cable routes a feeding reservoir is added.in static oil pressure. When the cable cools at light loads. balanced pressure reservoirs are used to make it possible to adjust the pressure-volume characteristics of the reservoir to suit the profile of the cable. the oil in the cable contracts but the cells in the pressure reservoir reexpand and force oil back into thecable along the cable ducts and into the insulation. Thebreakdown stress of impregnatedpaper reaches a valueof about 50MV/m if the stress duration is a fraction of a second. The formation of voids in the insulation is thus mitigated. The disadvantages of oil filling are the higher initial cost and complication at joints. Oil leaks can be a considerable nuisance and maintenance costs are high (Ross. 1987). Therefore. in some EHV substations. Gas-filling approximately doubles the working voltage of a cable of the samesize. Gas-filled cables are paper-insulated cables but are placed in steelpipes containing compressed gas. A s explained in Chapter 8. In contrast to conventional pressure reservoirs. These are described in Chapter 10. short lengths of cables insulated with compressed SF6.

In extruded insulation the erosion is often in the form of a propagation channel and discharge paths take the general shape of trees.3 The losses occurring in a cable dielectric include leakage loss and hysteresis loss. These erode the surrounding solid insulation andcan cause breakdown. giving discharges known as partial discharges. the dielectric losses reach about 80% of the three-phase conductor losses (King and Halfter. typically an average . to avoid discharges and “treeing.cables the working stress is made low. This is evident in the case of solid-type cables. For instance. This characteristic results from the combined effects of temperature on both the resistivity of paper and the viscosity of oil. under both DC and AC voltages. In paper-oil cables the working AC stress is made considerably below the inception stress for discharges.01. For cable insulation in general the loss factor increases withapplied electric stress. much more than those with pressurized insulation. The additional dielectric loss under AC is that caused by the hysteresis involved inthe process of dielectric polarization.value of about 3 kV/mm. In low-voltagecables this loss is negligible. but this value increases rapidly with temperature above 80°C in oil-filled cables. asexplained by the loss angle in Chapter8. The former occurs since the cable insulation has a finite volume resistivity. 1982). the rateof increase of the loss factor with applied voltage becomes rather high above the onset level for ionization of these gas voids. where C is the capacitance to neutral per metre and Y is the phase voltage. depending on the temperature and thetype of cable insulation. It therefore draws a leakage current. whether they are of the solid. With extruded polyethylene. Tan 6 assumes values of 0. Breakdown of the gas occurs regularly in the AC cycle and randomly on DC.” . In extruded insulation small cavitiesor voids existalong with impurities as a result of the manufacturingprocess. If the insulation contains gas voids. oil-filled. however small.but it is very appreciable in cables rated at 275 kV and above. This point is explained in Chapter 8. The dielectric loss = wCV2tan 6.002 and 0. For paper-oil cables tan 6 lies between 0. A minimum value for the loss factor occurs at about 40-80”C. The totaldielectric loss under AC is expressed 6 = 90” Qzd. or gas-filled type.003. where Q z d is the dielectric power-factor angle. Several tests have showna marked dependence on temperature of the loss factor (tan6) of oil-impregnated paper insulation of cables. in a 400 kV oil-filled cable.002-0.

and Ish3 = I s h L240 @ be the respective sheath currents. respectively. For three-phase power transmission by three single-core cables. the arrangement and spacing of the cores. 2.Ish2 = Ish L120 @ . the cables are symmetrically disposed at the vertices of an equilateral triangle. and W the angular frequency. r & the mean sheath radius.Experience over many years on samples and real cables has indicated that the lifeof a cable at constant temperature is governed by the empirical E = constant. Due to the symmetrical arrangement of the cables. and I3 = IL240" be the line currents in phases 1. the power frequency. Let Il = ILO". and R & be the resistances of each cable conductor and sheath. 1984). In addition to these circulating-current losses in single-corecables. AC currents flowing along cable conductors produce alternating magnetic fields. eddy currents are induced by the alternating magnetic fields linking every part of the metallic sheaths. Life under service conditions is obtained by extrapolating the straight line resulting from the plot of logE against log t . This law is tested by maintaining constant stress equation t ' on the dielectric and measuring time to failure. M31)between the conductor of a cable and the sheath of another cable are equal: where D is the cable-to-cable spacing. and the sheath resistance (El-Kadi. the mutual inductances ( M 1 2 . These eddy currents flow in both single-core and multi-core cables. and 3 and Ish1 Ish L @ . as is well known. Let R. p the permeability of free space. The EMF induced along the sheath of every core of single-core cables drives a current if these sheaths are bonded. The sheath voltage drop per meter of cable length E & is . M 2 3 . I2 = IL120". Their magnitudes depend on the core currents. say. These link the metallic sheaths of the cable and induce electromotive forces in them. where @ is the phase shift betweenthe conductor and sheath currents. Both types of sheath losses are discussed in the following paragraphs. 30 years or more. This assumes that the law which has shown to be valid over relatively short test times holds for a value of.

2).Skin effect. thus: (12. which is directly proportional to the conductor current (I).3) It is obvious that. The 12Racloss of the conductor respresents the largest heat source. This may result in arcing between sheaths. When all cable sheaths are bonded at each end. equation (12. and (12. The alternating-current resistance of the conductor . Equation (12.2) gives (1 2. the sheath loss is small with respect to the ohmic conductor loss (Table 12. When all cable sheaths are open-circuited at one or both ends of the cable system.4) and @ = 90" tan-' &h OM (12. The sheath loss as a percentage of the core loss per phase can be easily evaluated. with considerable damage. the sheath voltage to neutral may reach dangerous values.5) It is obvious from equation (12.Rac is the direct-current resistance modified to account for the skin and proximityeffects.The leakage reactance that accounts for the flux between the sheath and conductor of any cable is so small that its corresponding voltage drop is neglected.4) that cross-bonding of the sheaths along the route of the cable does not affect the value of the sheath current (Ish).3) shows that the voltage between sheaths increases with an increase in the spacing between cables. even at power frequency.6) Usually. is significant and increases with conductor . particularly under shortcircuit conditions. with long runs of heavily loaded cables.

due to this cause is of the order of 20%.g.Dielectric and SheathLosses Relative to the Conductor Losses in Three-phase Cables of the Oil-Filled Type with Different Rated Voltagesa Voltage (kV) 66 132 275 400 Conductor 1.is quite small.29 0.o 1.This is illustrated by the equivalent circuit of Figure 12. .o Relative losses (%) Dielectric 0.) Equivalent circuit formed conductor by and bonded sheaths.o 1 .13 0. The voltage drop per metre A V = I(& +jwL.o 1.40 aConductor area = 2000 mm2. to Zef. the increase in R. e.36 0.04 0.15 0. The cables were buried at a depth of 1 m in a soil of thermal resistivity = 1 KlmW. It can easily be shown that bonding the three sheaths of the cable effectively changes the impedance per phase of the three-phase balanced cable from 2. where the conductor and sheath form a coupled circuit. The secondary reactance per phase is practically equal to WM. Its primary reactance per phase X..80 Sheath 0. Source: King and Halfter (1982).4. cross-section..) + Ishj u M + IjwM and in the sheath 0 = Ish(& +jwL. A three-phase cable systemwith sheaths bonded at both endsis equivalent to a three-phase 1:l air-cored transformer.With large conductors.12 0. 2000 mm2. Consider a metre length of the circuit.

or at an i n t e ~ e d i a t e point.. no point on any sheath is equidistant from all three current-carryingconductors. resulting in a so-called sheath eddy loss. laid in trefoil.3 From these two equations (12. “cross-bonding” is resorted to for limiting sheath voltages (Kuffel and Poltz. 1981). In case of bonding. each of the outer phases will suffer a higher induced EMF than that of the middle phase. When the three single-core cables are disposed in a flat arrangement. Subsequently. This cross-bonding of individual core sheaths is carried out in addition to their transposition (Fig. 1 y-~~rre Losses nt In three-phase cable sytems. in which the continuity of the sheaths and insulationscreens is .5). Therefore. and to accept the total induced voltages at the remote end. These voltages to ground. 12. 1 nding in T h r e e . should be limited to a safe value (i. the outer phases will also have higher sheath losses. Consider a system consisting of three equallengths of cable per phase.and hence the EMF induced in the sheath will vary from point to point on its circumference.100 V in protected positions and about 25 V in exposed positions).7) Thus the phase conductor resistance is effectively increased while its inductance is reduced. These seldom exceed 2% of the conductor loss and are usually neglected.~ ~ a s e The simplest form of sheath bonding for three-phase single-core cables is to bond and earth the three sheaths only at one end. however. with cable lengths exceeding a few hundred meters.e. an eddy current will flow circumferentially.

) broken at each joint. 2. In three-phase single-core cable systems. the total voltage at the end of each sheath circuit will be zero for a balanced loading of the cables. Losses due to the magnetic field produced by the cable conductor current itself and by currents in neighboring cables if unscreened. With DC. the armor losses are usually divided into: 1.5). The derivation of armor losses has beendealt with thoroughly in the literature (e. At 50 . 1986) (Fig. the three sheaths can be bondedand earthedwithout any sheath current flowing (Pirelli.. At the ends.Sheaths salidly bonded and e a r t h e d / .g. Ametani. the hysteresis lossesare often of sucha magnitude that they preclude the use of ferrous amoring materials. therefore. The amoring of a cable may be considered as a supplementary metallic sheath. In single-core cables. If the sheaths are cross-bonded. 1979).5 Cross-bondingof cable sheaths. (Courtesy of Pirelli. Losses due to circulating and eddy currents in the amoring. the current’s ownmagneticfieldeffectively drives it toward the skin of the conductor. This “skin effect’’ causes the AC resistance of a conductor to exceed its DC resistance. These combined magnetic fields can produce significant hysteresis losses. WithAC. the current gets uniformly distributed over the conductor crosssection. 12. additional losses due tomagnetic hysteresis willtake place. If the cable amoring is magnetic.however. with corresponding losses.

It isusually determined by bridge measurements. known as the “proximity effect. the increase in resistance is 2. the cable length is termed critical. A similar effect is that of neighboring conductors carrying AC currents. The duct inside the tubular conductor is very useful for cooling in oil-filled cables (Fig. respectively. and its accurate calculation is much more difficult. 12. the permittivity of freespace.5% for conductor diameters of 2. A further current reduction is caused by the appreciable magnitude of dielectric losses at higher voltages.8) where a is the conductor radius. The effect is much more significant at higher frequencies.3 ~el-~alam example. a severe decrease in the value of load current transmittable (derating) occurs if the thermal rating is not to be exceeded:in the higher voltage range. and E. For three-core belted cables. the corresponding MVAr requirement is about 10. accurate calculation of the capacitance per phase is very difficult. There are capacitances between conductors and a capacitance of each conductor to the sheath. 12.5 has a critical length of 42 km.3a). which depends on: . b the inner radius of the sheath. When the charging current becomes equal to the rated current. It is much easier to determine it by bridge measurements. 1988). the cable inductance is much smaller. A 345 kV cable with approximately 25. The determination of capacitance is important for d e t e ~ i n i n g the cable charging current. the relative permittivity of the cable insulation. The capacitance per unit length of such a cable can easily be shown to have the value (12.8 cm.Resort is made to numerical computation (Malik and Al-Arainy.3 mmthickness of insulation and a relative permittivity of 3.’’ The skin effect is evidently lowerfor tubular conductors than for solid cylindricalconductors. The inductance per phase of a cable is decided by the magnetic flux linkage. Compared to overhead lines. it is therefore a favored design for conductors in EHV cables. As the charging current flows inthe cable conductor.5% and 7. lengths of the order of 30 k m create a need for drastic derating.3).6 MVAr per km.5 and 3. Consider the cross-section of a single-core cable with a metallic sheath (Fig.

The amount of screening afforded by the metal sheaths 2. Malilc and Al-Arainy. a the conductor radius.e.12. The stress. 12. This is a suitable design for oil-filled cables (Fig. This increase is overcome by the use of taped screens of metallized paper or carbon-black paper. an extruded screen of semiconducting plastic.9) where V is the applied voltage. In practice such a ratio is far from economical. for a given sheath size.2). is further increased by the effects of conductor stranding by an amount dependent on the strand diameter. but in the range 15-25%. and Eav) in any cable means that the insulation is being inefficiently utilized.1) and is V = aln(b/a) (12. 1982. and b the inner radius of the sheath. .of the cable conductors (King and Halfter.12. or. 1987).. The field distributions in such cases are evaluated by measurements on models or by numerical computations (Chapter 2). One is to use multilayer insulation. Some suggestions have been made for reducing the ratio Emax/Eav. In three-core cableswithunscreenedcores. an enlarged conductor diameter for the same net cross-section is achieved with tubular conductors.3). in the case of extruded cables. 12. the inner parts are much more highly stressed than those near the sheath (Fig. Its maximum value occurs at the surface of the inner conductor (Table 2. the outer layerhaving a lower value of E. The proximity of the cable to other conductors and ferrous objects As explained in Chapter 2. the maximum electric stress varieswith the conductor size and has a minimumvalue when a = b/2.6). Emax The big difference betweenthe maximum and average fields (i. (see Section 12. Thus.1). The presence of armoring and whether or not it is ferrous 3. Both suggestions have been faced with numerous practical difficulties in both manufacture and operation.3 1.718 (see Example (2)). However. Some empirical formulas have been developed for estimating the field intensities at some points on the surface . the electric field intensity in single-core coaxial cables can be calculated. Another suggestion is to subdivide the cable’s homogeneous insulation by intermediate metallic sheaths to be energized at suitable fractions of the cablevoltage(seeSection12. which is a maximum at the surface of the conduetor. the electricfieldshave tangential components at some points.

This. depends on the thermal resistances of the different parts of the cable and of the soil in which it is buried. Thus the rated current for a cable depends on the rates of heat generation within it and dissipation to its surrounding.6 insulation. It depends also on whether the cable is buried directly in the soil or run through troughs orducts. the range of earth temperature to be expected depends upon the depth and geographical location. being one of a group of cables. or is laid near gas or water pipes. Take the simple case of the ..3 . El-Kadi et al. whether it is buried by itself. 1984).3). in turn. and should be considered (King and Halfter. Recommended steady-state operating temperaturesof the conductors of paper~o~l cables are as follows: Armored cables buried direct in soil: 65°C Oil-filled or gas pressure cables direct in soil: 85°C The above values assume an earth temperature of 15°C. The rated currentof a cable can be evaluated by setting up the equivlant thermal circuit of the cable-soil system. Electric field distribution in a single-core cable with one uniform The current-carrying capacity of a cable is defined as the maximum current it can carry continuously without the temperature at any point in its insulation exceeding the limit prescribed for it according to its thermal class (Section 8. All these factors contribute to heating or cooling the cable under study. 1982.

so they are included at the appropriate nodes in the equivalent circuit (Fig. 12. The losses and War in the sheath and armoring also contribute to the cable temperature rise. Thus the thermal form of Ohm's law gives the relation As the dielectric loss flows through one-half of the internal thermal resistance of the cable. h . above 0. the temperature rise A@. and the soil G. those of the bedding Gb. ofwhich the temperature is ea. The heat developed by the conductor power loss WC= 12R flows through the thermal resistance G d of the insulation.7 Equivalent thermal circuit of a single-core buried cable.7). There the maximum temperature of the insulation QC OCCUTS at the conductor surface. . the ratio of the sheath loss and armor loss to the conductor loss. are. The dielectricloss w d is assumed concentrated halfway through the insulation. respectively.single-core cable represented in Figure 12. serving GSer. to the ambient.7. is due only to the dielectric loss w d and (12. .11) Substitution yields wsh &(I2&) and War = har(12R) where h& and h .

m/W). What is leftfor calculating the current-carrying capacity of the cable is the determination of its own thermal resistance and the maximum permissible temperature. All the heat evolved is assumed to be stored in the cable..e. The external thermal resistance depends on the typeofsoil. the conductor radius a.12) does not take into account correction factors for the skin and proximityeffects(Section12. the amount of moisture present. 1983. 1984). This results in a lower temperature rise than when permanently on full load. and for conductor areas from 64.. For directly buried cablesof conductor area less than 64.09. can be evaluated (IEC. Short-circuit.13) For a three-core belt-type cable. Prime et al. together with typical ratios Ash and h. which is allowed for in practice by the use of cyclic rating factors. Cyclic loading.. 1978).. Calculations are often based on an idealized daily load curve of full-load current for 8 h and no-load for 16 h.1974). and the sheath radius b. Both effects. i. It is expressed as (12. 13 times the rated full-load current can be passed on cyclic loading.7).. 1980.5 mm2 this factor is 1. Calculation of the thermal resistanceof a single-core cable is easy and is determined in terns of the cable radii and the thermal resistivity of the insulation. A temperature limit of 120°C is often assumed. . 1988): l. and computer programs are best used (Glicksman et al. however. Equation (12. Burghardt et al.5 to 645 mm2 it is taken as 1. The following regimes of transient heating occur (Weedy. the calculation of thermal resistance is more complex.. The thermal resistance of a single-core cable Gd is determined by the thermal resistivity ofthe dielectric material g("C. 2. l . and how the cable cores are disposed in the earth. Theoretical and experimental studies for evaluating the internal and external thermal resistances and the temperature rise for different cooling systems of generalburied cables have been reported in the literature (AbdelAziz and Riege.After simplification. due to the arrangement of the thermal field.13.

all H V cables are produced and laid in a number of limited lengths that have to be joined together on site and terminated at required positions.e.8). The conductor and sheath temperatures may be obtained from a simple model (Fig. Bm2.C3= effective thermal capacitance of soil.9. 12.8 the thermal capacity of the pipe oil. = thermal resistanceof insulation. The simplest analysis results from a single lumped constant representation. There are four main methods of installing cables. = effecof soil. Reasonable results over a few hours of transient are obtained for pipe-type cables by making C 3 in Figure 12. C2= thermal capacitance of sheath.Simple model for transient thermal response of buried cable:C. a single R-C network. The response of such a network to a timevarying load is illustrated in Figure 12. R.. = thermal capacitance of conductor.t ~ emergency ~e loading. shipping. tive thermal resistance 3. Becauseofmanufacturing. the sheath is isothermal and the thermal circuit is an R-C ladder network. i. and installation limitations. i. . Direct in the soil: the cable is laid in a trench which is refilled with a backfill consisting of either the original soil or imported material of lower or more stable thermal resistivity. The number of nodes depends on the number of annular cylinders into which the dielectric is divided. in which t is the thermal time constant (R x C) and Bml. and Bm3 are steady-state temperatures. R. An overload may be sustained for perhaps several hours without excessive temperatures due to the long thermal time-constant of the cable.. as follows: 1.e. For a cable a uniformly radial heat flow is assumed. S h o r t . A lumped-constant thermal network may be used for transient calculations by the connection of appropriate thermal capacitances between the nodes of the steady-state network and the ambient or reference line.

and insulation materials. In circular ducts or pipes through which cables are drawn.g. In ducts or troughs. cable joints areclassified as straight-through joints. 3. Although cables are manufact~redunder carefully controlled conditions. 4. 1982. branch or T-joints. 2. the joints and terminations cannot always be made under similar circumstances. Pirelli. In air. There are two main methods by which a more u n i f o ~ distribution of stress in the cable insulation maybeachieved: by the introduction of intersheaths. its conductors.when completed.l I I I ' I t Temperature rise-time curves under varying load. to eliminate any risk of supply interruptions (King and Halfter. and by the use of layers of insulating material with different permittivities. trifurcating joints. Thetechniques of jointing and te~nating cables depend on long experience withthe specific type of cable. they should be as reliable asthe rest of the cable. e.. installed in tunnels built for other purposes. usually of earthenware or concrete. stop joints of oil-filled cables. and outdoor sealing ends for terminatingcables outdoors. . In general. This has the advantage of further cables being installed without excavation. Extreme care is taken to ensure the high current-carryingcapacity and high insulation strength of the joints and sealing ends. Nevertheless. 1986).

10.v* v . Equating the maximum stresses. the ~ n i m u m stresses are the maximum stresses divided by a.v1 Wrl /a) and r ln(r1l a ) The m a ~ m u m stress is E l= Similarly. Thus. by taking r l / a = r2/r1 = b/r2 = a So. Thus. A1 = v . instead of curve C2 which represents the stress without intersheaths. It is possible to choose rl and r2 so that the stress varies between the same maximum and minimum in the three layers. The conductor and the sheath have radii a and b. the maximum stress between the first and second intersheaths is r11&2II"1) whilst the maximum stress between the second intersheath andthe sheath is E3max J52max = v 1 -v 2 v 2 = r2 ln(blr2) By choice of V1 and V2 the maximum stresses can be made equal.Suppose that intersheaths of radii rl and r2 are inserted into the dielectric and maintained at voltages Vl and 'v2. one obtains .1). The stress between any two metallic cylinders varies inversely as thedistance from theaxis (Table 2. thestress at a point distantr from the conductor (a r e r l ) E l = A1/r l from the conductor where A l is a constant which is found by integrating E to the intersheath as follows: where V is the conductor voltage (applied voltage). and the stress distribution is like that shown by curve Cl of Figure 12.

Without the use of intersheaths. the maximum stress is SO that the maximum stress has been reduced due to theuse of intersheaths in the ratio . v 2 = 1 + + - V l / a l@’ v 1 = V(1 l/a) 1 l/a l/a2 + + + + The maximum stress is then ” v- v 1 aha V (1 + a a2)aIna (1 + a + 3v a2)aln(b/a) since ln(b/a) = In a3 = 3 In a.3 th 0 0 l 2 3 Distance f r o m A x i s ( c m 1 w e 12.10 Effect of intersheaths on stress distribution in single-core cables.

1 1 : ? ( l +a+a2) The use of intersheaths has not been general practice because of the complications involved. These methods take two main foms: . A2 = In(rl /a) E1 1 1 +In(b/rl) E2 Y The maximum value of Elis Y and Y In recent years progress has been made in schemes for reducing the external thermal resistance of cables by artificial cooling.*. the permittivities being el and c2. The sheaths must be supplied with the requisite potentials and must carry quite large charging currents. Jointing is made very difficult . Suppose that the cable dielectricconsists of two layers with a dividing radius r l . The stress in the inner layer is = A2/(Elr) whilst in the outer layer E 1 E 2 = A2/@2r> Then .

et Low-resistance faults can be located by connecting a DC bridge to the fault and a sound conductor at each end of the cable.1. say. The amplitude and polarity of the reflected pulse depend on the resistance of the fault (Fig. The distance X can be measured either by the “pulse” method or by using a bridge. Flashing (se1f-healing) faults The faulty conductor and the type of fault can easily be identified by simple tests at both ends of the cable. the velocity of propagation of electromagneticwaves along thecable. The error in the result depends on errorsin v and the ~easured tx. Greatly increased currents areachievable by the use of large pipes of diameter. These two conductors are . as noted below. and then to locate the fault more accurately. A DC bridgeis suitable for lowresistance faults.11. Normal practice is to identify the zone in which the fault has occurred by an approximate measurement of its distance X from either end of the cable. 2. This pulse will travel along the cable up to the fault. whereas AC bridges are appropriate for faults of broken conductors.or low-resistance earth faults involving one or more of the conductors 2. TI Faults in multiconductor cables maybe divided according to their nature as follows: igh. corresponds to the pulse traveling from the sending end of the cable to the fault and back. Knowing v. Open-circuit faults with low or high resistance at the break 3. Using such a trace of the pulses on a cathoderay osci1loscope. From an engineering standpoint this scheme is much simpler than thefirst. 12. where it gets reflected. the distance X to the fault canbe calculated. 150mm. see also Chapter 14). A voltage pulse from a suitable source is applied between the faulty conductor and a sound conductor or the metal sheath. These include measurin~ the insulation resistances by a megger of suitably high voltage. ~ x t e r cooling: ~ ~ Z water is pumped through plastic pipes laid beside the cables. ~ntern~Z cooling: the pumping of water or insulating oil along a central duct formed in the cable conductor. the time t.

l 1 Pulse method for locating the faulty zone of a cable: (a) low-resistance fault. tolerances in the bridge elements. Too high a resistance of the fault itself would no doubt affect the sensitivity of the bridge. G is either earphones.. . or oscilloscope. The furtherburning of the faultis usuallycarried out using a special powerful energy source. A self-healing fault.(b) high-resistance fault. for instance. would need to be burned further in order tobe located by the bridge measurement with fair accuracy.12). Inaccuracy in measuring the distance X by this method results fromerrors suchasinsensitivity of the galvanometer. If the faultis a break in one of the cable conductors. meter. Thus the balance conditions of the bridge are used in calculating the fault distance.and the contactresistances at both ends and at the jumper. it can be located using an AC bridge (Fig. connected at the farend by a jumperof sizable cross-section forrning a loop. The distance X of the fault is proportional AC bridge method for locating a break in of one the cable conductors. 12.

The possibility of producing very strong magnetic fields by very high currents without resistance losses and with no F for continuously driving it would make possible electromagnetic levitation of fast trains.Its zero per~eabilityresults in its repulsion to magnetic fields. There will be a detectable signal all along the cable up to the point of the fault. superconductors would mean loss-free transmission of very large powers. are correlated as follows: . tolerances in the bridge elements. by headphones and a search coil) along the cable route near the suspected fault location. yttrium. and lanthanum. a train of high-frequency pulses is sent along the cable while their magnetic fieldis detected (e. including oxides ofcopper.g.1 E x ~ e r ~ ~ eEvidence nta~ A superconducting material manifests zero electricalresistivity and zero magnetic permeability as long as it is kept at a temperature lower than Tc. Even as late as 1987 some ceramics were discovered. a drastic reduction in their sizes for large ratings would be possible. For a more precise location of the fault. In machines. barium. Superconductors are materials whose electrical resistivities become immeasurably small or actually zero below critical temperatures Tcunless exposed to magnetic fields exceeding critical strengths He. The proportionality can be evaluated by measuring the entire length of the same cable by the bridge. The critical temperature Tc and the critical magnetic field He of strength lower than H. beyond which it ceases. which accounts for the great possibility of magnetic levitation. 1215. and impedance between the two parts of the broken. If employedin cables. Errors in measuring the distance X by such a bridge result from insensitivity of the detector G. conductor. 1987).. and many other applications.to the capacitance Cx.1. In electronics and computers it would mean an increase in switching rates and memory sizes. These ceramic superconductors have critical temperatures approaching 90K. magnetohydrodynamic propulsion of marine vessels. For such materials the cost of refrigeration would be about fourorders of magnitude lower than the case ofconventional metallic superconductors (Anon.. Over the years intensive research has been carried out to discover more superconducting metals and alloys with the highest possible critical temperatures.

1 (12. it is drawn in the form of several filaments of the superconductive alloy 1060 pm in diameter. Their energy I i Absolute temperature *K H t . The group of filaments is embedded inside a conducting wire of copper or copper-nickel with an overall diameter of about 1mm. This relation is diagrammatically shown in Figure 12.14) states that the critical field strength Hc(T) at the absolute temperature T is lower than the critical field strength H.1. above which the material's superconductivity is destroyed at absolute zero temperature. Equation (12. The stable rated current of such a wire (Fig. Superconducting coils haveactually been manufactured for producing very strong magneticfieldsin cyclotrons and other particle accelerators (Maix et al.1985).l3 Graphs illustrating the states of superconducting metals.14) under magnetic fields of about 2 T is about 300 A..14) where g is a constant. positive ions located in the crystal lattice along the path of the electrons through the superconductor help to cancel out some of the repulsive forces between pairs of electrons having mutually suitable energies. g was considered equal to unity by King and Halfter (1982). based on the concept of a perfect pair of free electrons (Hummel. 1987). 12. in which the current is carried by a stream of individual elecrons. 12.2 Theory A theory to explain superconductivity was proposed in 1957 by Bardeen. According to this theory. the current is carried through a superconductor by pairs of electrons. To ensure stability of superconductor performance. They have equal and opposite movements. as explained very briefly below. .15. Different from norrnal conductors.13 for different metals. Cooper. and Schrieffer.

Bobo et al.c up er conducting filamentarywires clad in acopper wire. The perturbation of the atomic lattice of normal conductorscaused by the flowing electrons accounts for their electrical resistivities. Thus. Later. overall wire diameter = 0. Cryogenic cooling of a suitable metal or alloy to near 0 K damps out vibrations of its lattice atoms and furnishes the condition suitable for free electron pairs.7 mm.activestudieswereprogressing for developing the least expensivedesign for superconducti~gcableswith suitable core material and size. discovered recently. (Courtesy of BBC. Filament diameter = 57 pm. Insulating ~ a t e r i a l s . as each electron pair drifts through the superconductor. A displaced ion reverts back into its original position with a m o ~ e n t a r y oscillation. Some researchers have also s t u ~ i e d the suitability of liquid airimpregnated paperas an insulating material.) levelslie near free states. 1987). which causes cracking. A phonon acts as an elastic coupling betweenthe two electrons as they pass together through the atomic lattice with no net perturbation. cannot be used because of the unacceptable loss tangent. have been considered (Kofler and Telser. 1987. together with solid dielectric spacers. Until 1987. In the early days of superconducting cables. Synthetic materials are not permitted for use because of their differential contraction. Other materials. such as glassepoxy tapes and enamels. one of the two electrons perturbs its atomic lattice. The dielectric loss in superconducti~gcables should be minimal as it imposes an additional load on the refrigeration system. liquid nitrogen and liquid hydrogen have been introduced. a relatively low-loss dielectric such as helium or vacuum was usedfor cable installation.. This theory no doubt would need ~ o d i ~ c a t i o to na c c o m o date superconducting ceramics. . including paper.

e.. is the relative permittivity of the dielectric. and the charge on the conductor be q/cm length. Therefore. the maximum stress in the inner dielectric (i. the electric stress at any radius r is (12. They are brittle and chemically unstable (Bogner et al. ~ o Z ~ t i Let o~: r be the outer radius of the inner insulating material. and acceptance tests on cables are laid down in the appropriatestandards (Chapter 18). On the other hand. routine. TI Detailed descriptions of the procedures for the various type. the sheath diameter over the insulation being 10 cm. Then. at the surface of the conductor) is The maximum stress in the outer dielectric (at radius r) is 4 E2 = ~ E x E 4~x r . (1) The inner conductor of a coaxial cable has a diameter of 4cm.1987).15) where is the permittivity of free space.However. many problems have vanished. with the discovery of ceramicsthat exhibit superconductivity at temperatures approaching normal room temperature. The cable is insulated with two materials having permittivities of 6 and 4 respectively with corresponding safe operating stressesof 40 kV/cm and 25 kV/cm. the next millennium may well witness the solution of these various technical problems and the commercial availability of normal-temperature superconducting cables and wires for numerous power and electronics applications. if is the permittivity of the inner dielectric and the permittivity of the outer dielectric. Calculate the radial thicknessofeach insulating layer and the safe operating voltage of the cable.. the ceramic superconductors available so far have unsuitable mechanical and physical properties.03 Now. and E.

04 4.4cm outer = 5 4. 5 V2(peak) = 25 x 4.16) where Emis the maximum stress in the insulation for the inner dielectric.8cm 25 Therefore. Calculate the optimum value of r if V is 100kV and the maximum permissible gradient is 55 kV/cm. 4. " " E. the radial thicknesses of the dielectric are inner = 4.8 In.8 V I(peak) = 40 x 2 In = 70.94 Safe operating voltage (rms) = = 53 kV y/z For a single-core cable. determine the value of conductor radius r which gives the minimum voltage gradient for fixed values of applied voltage V and sheathradius R.2cm The voltage drop across a dielectricbetween radii rl and r2 is given by the expression r2 V = Emrln-rl (12.9 kV 4.8 cm Therefore. 40 r=-x3=4.9 = 74. rl = 4.Therefore. . r2 = 5 cm Therefore. Em = 40 kV/cm.94 kV + + 74. r2 = 4.8 = 0.8 cm.04 kV 2 For the outer dielectric. Em = 25 kV/cm.8 2 = 2.= 4. rl = 2 cm.8 Peak voltage of cable = Vl(peak) V2(peak) = 70. 4r r E2 12 -3 As El= 40 kV/cm and E2= 25 kV/cm.

=O l? In.= 1. which i s always where the insulation is most heavily stressed. Under these circumstances the voltage gradient at the conductor surface is inversely proportional to the core radius.17) If V and R are fixed and r is variable. If V = 100 kV and the voltage gradient at the conductor surface must not exceed 55 kVjcm.~oZ~ti The o ~ highest : value of the voltage gradient occurs at the surface of the conductor and is given by Em = V r In R/r (12. is a minimum. minimum value of Em is that which makes dr dr For Emto be a minimum l nR -+rL(s) r R r2 i. then for a given core radius the voltage gradient at the surface of the conductor. r " R r -e r = R/e If the value of r is substituted in equation (12.57 cm It is assumed here that the value given for V is rms and that for the voltage gradient is peak value.e.17). then the optimum value of r is given by 100a 55="----r r = 2.. the minimum value of Em becomes Em = V/r The significance ofthese results is that if the overall and core radii are in the ratioof e. . the value of r which gives the dEm equal to zero. R In-r 1 = 0.

5 = 2 p == 2n x lo4 x 0. Prove the formula used. 50 H z frequency supply. Insulation resistance through the element = ~ Pdr 2nd (12. Calculate the resistivity of the insulating material.5 x 106/1n2 = 4. R / r = 3/1. in-core and core-to-sheath capacitances in three-phase. ~ o l ~ t i oConsider n: an element of the insulating material at a radius r from the conductor center and having a thickness dr in a length l of the cable. Let p be the resistivity of the material in Ma-m. B core cables. when it is connected to a lO. the insulation resistance = 0.5 p F / h measured between two of the cores. three- .53 x lo4 M Q . lead-sheathed cable. Sol~tion: Figure 12. The core diameter is 3 cm and the sheath diameter of the cable is 6 cm.5Ma.15 shows the arrangement of capacitors which is equivalent to the capacitances in a 3-phase.(3) A single-core cable 10 km long has an insulation resistance of 0. ~ (4) Calculate the KVA. C1 representing the intercore capacitances and C2 the capacitance of each core to the sheath.18) Total insulation resistance = In the problem given. taken by 10 km of such a cable which has a capacitance of 0.OOOV.5 x lo6 52 l = 10 X lo3 == 104m.

then c 4 =g c z + C3)w Therefore. the cable may also be represented by the system of capacitors shown in Figure 12.Now it is obvious that the star point N and the sheath will be equipotential.15 may be replaced by the equivalent diagram in which the star capacitances C3 take the place of the delta capacitances Cl. For the two circuits to be equivalent. c 2 4-c 3 = 2c4 Line charging current = Vphw2C4 If Y is the line voltage.16 = Yph(C2 + c3)w If C4 is the capacitance measured between A and B with C insulated.three- . Hence. capacitances inthree-phase. Line charging current = 2 Ywc4 ~ 4 3 (12.Figure 12.5 pF/km X 10 km = 5 pF Neutral & sh eath c .19) In this case C4 = 0. 12. l 6 Equivalentcore-to-neutral core cables. Line charging current for the star-connected system of Fig. C3 = 3C1.16.

3-phase.000 x 271. + C3) = 0.9 pF/km. the capacitance between the three cores bunched together and the lead sheath is 0. c 4 C3 = 0.65 pF/km Hence.45 pF/km = capacitance per km between any two cores .16: 3C2 = 0.16 kVA.75 pF/km and that between two of the cores connected together and the third core is 0.. . & x 109 (5) In a 3-core.75 pF/km If B and C are joined together the capacitance between A and B conC3) and (C2 C3) in series. 12.15 and 12. sists of two capacitances of2(C2 Therefore. required to keep 10 km of this cable charged when connected to a 33 kV.6 pF/km. (b) the kVA. : = 1 p. lead-sheathed cable. capacitance between A and B + + C2 + C3 = 0. Solution: (a) Using the notation of Figs. x 50 x 5 = 314. 3-phase.Line charging current IC= 2 x lo4 X 2n X & 50 X 5 X A Charging kVA = &VIC x l 0"3 & x 10.000 x 2 x 10. Calculate: (a) the capacitance per km measured between any two of the cores. 50 Hz supply..

Theinternal and external radii of the lead sheath are 2. The cables are located at vertices of an equilateral triangle. 85km in length and comprising three single-core cables each of conductor radius 1cm.5 109 & = 3079.0 cm respectively.2 x S2 cm.004).52) ~ lo5 = 22.5 and 3. and in this case the conductor-to-sheath mutualinductive reactance Xm=~M=2nx50x21n (~) x 10-7nlm = (12.5 pF Charging kVA. x 2V#C4 =~V"----- 2(33 X 1 0 ~ 1 x ~ 50 2~ x 4. (6) Determine the effective electrical parameters of a three-phase.9 Q + Resistance of sheath &h 23. 50Hz underground circuit. = 0.20) where rsh = mean radius of sheath and D (= 8cm).105 X 85 = 8. 8 cm each side.(b) Capacitance of 10 km between 2 cores = 4.83 Q In a three-phase system the equivalent phase-to-neutral reactance is considered. Therefore. of 10 km of cable = &VI.1 kVA. cable-to-cable spacing .105 S2/ km(conductor operating tempertureassumed to be65°C and temperature coefficient 0.0875 Q/km and the resistivity of lead at the operating temperature may be assumed to be 23. 66 kV. The sheaths are bonded to ground at several points. The conductor AC resistance per kilometer at 15°C is 0.0875(1 0. Solution: Conductor resistance = 0. For the whole length.2 x IOM6 x 85 x ~ (3 2.1 LVA.004 x 50) = 0. R.. to keep 10 km of the cable charged requires 3079.

R : h +X i ” 5. Effective reactance Xef = 11.= W x 21n . I x 85.04s2 22.832 _ _ .x (3 J.y2 + Sheath loss fish X : Conductor loss R.15 For a current of 400 A.83 8. + 5. The current is 250A at 50 Hz.72 22. for 85 kmlength Effective AC resistance of conductor = 10.1 S?.28 kV (7) Find the induced sheath voltage per km of a symmetrical three-phase cable systemwith conductor spacing 15cm and sheath diameters 5.Xm=2n:x50x21n x 10-~ x 85000 = 5.000 = 11. .7 S?. the emf induced without bonding @er sheath) = .9 22. = reactance with sheaths open-circuit = wL.y2 = 0.5 cm.1 = 11.832 5.24Q Effective reactance per cable where X. H m= 2280vT = 2.

the maximum stress is &lax = (1 + 1.384)23.65 cm.3 kV/cm = 1 1n(2.65/1) 53.3/2 = a3.65 1n(2.8 23. = 1 + 1/1.8 With the intersheaths. b / a = 5.8 kV/cm (Emin)o = 2.9 = 41.3W2)= 38.384 Thus. and the voltage on the intersheaths. rl = 1. find the best positions. the voltage between them at the other end is x 26.92 cmare the radii of the intersheaths.6 = 46V/km (8) A single-core 66kV cable has a conductor radius of 1cm and a sheath of inside radius 2.384 + 1.385c m and r2 = 1. the maximum stress is and the minimum stress is = 20. Find the maximum stress. If two intersheaths are used.7 kV/cm 3 x 55. The peak voltage on the conductor 66 53. a = 1.384 + 1/1.E =I X = ~ IOM = 1w2 In x 10"~ =-1n271' 47t x pa1 rsh L) D rsh - x 314 x 250 15 27t = 26.6V/km InV/m 2.38'12VI = (1 + 1/1.1 kV 53'8 = 55.3 .75 If the sheaths are bonded at one end.65/1) Without sheaths..9kV v. the maximum stress.

= 4.2kV and V2 = 23 kV The maximum stress Emax 47.6) In 1. solution^ b = 2.7/a = 27. stress has been reduced by the ratio 1: (9) Suppose that in the previous example the intersheaths are spaced at equal distances from each other. Elmax rl = 2cm.65.55 cm and r2 1 -(b a) = 0. Find their voltage for the same maximurn stresses in the layers. Find the ma~imumstress in the rubber and in the paper.6.43.3253= 51. The maximum 1. a = 1 cm.8kV 1 1 1 1 12 + (4. .65cm. 53.10 showsthe stress distribution in both cases.6.55 cm 3 = 2. and find the maximum stress. V1 = 45. V = 53.5 kV/cm .5 and the rest is impregnated paperof relative permittivity 3.8 b = 2. Solution: a =km. &.96 kV/cm Figure 12.5/3. the conductor and the sheath.1 cm As EI max E2max = E3 max = Emax.5.: rl = 1.5 kV (10) Suppose that thecable of the last two examples has an inner layer 1cm thick of rubber of relative permittivity 4.2 = 3.while the minimum stress is 38.

Schneider J. 1984. El-Kadi M. Lambrecht D. IEE Power Eng J 151. Harlow.New York: John Wiley. Pirelli.5 kV/cm.18 kV/cm = 2[(3. i. Valdes J. SteeveE. Abdel-Aziz M. 1988. 1982. 1987. Malik N. Addis G. 55.and E2max 53. IEC. Electronic Properties of Materials. Poitevin J. Electra 114:96-107. Rauch J. Halfter NA.6/4. Report 100-04. El-Kadi M. Glicksman L. 1988.Hess R. Macias C. Benz H. p. the maximum stress has been reduced from 55. Bottger 0. 1984. 1986. Thermal Design of Underground Systems. Geneva: International Electrotechnical Commission. 1984. Telser E. Vecsey G. Maix R. Radhakrishna H. Hjalmarsson G. Al-Arainy A. Robsenow W. 1987. IEEE Trans PAS-103:2735-2740. a better quality paper is put near the conductor than near the sheath. Chu F. 5923.3 to 51. IEEE Trans PAS-102:373-381.1983. Berlin: Springer-Verlag. Report 100-02. 1979. Porven A. Proceedings of CIGRE Symposium on New and Improved Materials for Electrotechnology. Prime J. Purnhagen D. and Poltz J. 1980. Int J Elect Eng Educ 25(1 ):27-32. Ruprecht A. Vienna. 1987. 1966. 1987. Bogner G.8 = 32. Publication 287-2. Bobo J. Kofler H. Matthysse I. IEEE Trans PAS-100:369-374. Ware P. This method of grading isquite practical. IEEE Trans PAS-98:902-910. Zichy J. Ross A. IEEE Trans PAS-102:21332144. Anon. Malik N. 1987. Underground Power Cables.e. UK: Longman. Humphrey L. IEEE Trans PAS-103:27942798. Nithart H. Report 620-07. and in practice the only grading used is for strength. Milan: Pirelli. Proceedings of CIGRE Symposium on New and Improved Materials for Electrotechnology. Kuffel E. King SY. WeedyBM. . Saure M. SabriC J. IEEE Trans PAS-97:134"139. Al-Arainy A. 1988. 1987. IEEE Trans PAS-99:2386-2392. Brown M. High-Tech Mater 4(9):3. 1987. Brown Boveri Rev 74:40-48. Burghardt R. Calculation of the Continuous Current Rating of Cables (100% Load Factor). Proceedingsof CIGRE Symposium on New and Improved Materials for Electrotechnology. The reduction is hardly worthwhile. Self-contained Oil-Filled Cables. ASEA J. 1983.. IEEE Trans PWRD-2:589-595. 1978. IEEE Trans PAS-103:2043-2050.5) In 2 In 1. Scoran E. Vienna. Ametani A. Olsson J. 1985. IEEE Spectrum 11:73. Hummel R. Jakob B. 1987. Mathews H. 1974. Vienna. Riege H. Sanders J.3251 + Thus. 1981. Engelhardt J. Golz W.

This Page Intentionally Left Blank .

surge arresters are often used which are provided with a low earth resistance connection to enable the large currents . 1972). Grounding is of major importance in our efforts to increase the reliability of the supply service. in the event of fault.Cairo ~ ~ i v e ~ ~ Giza. which rapidly isolates the faulty circuit. it is its metallic frame or grounding conductor. The use of electricity brings with it an electric shock hazard for humans and animals. 13. Thus. The potential of this ground conductor may be quite different from zero (i. with small apparatus. preventing excessive voltage peaks during disturbances. i t yEgypt .e. but most often. particularly in the case of defective electricalapparatus. as it helps to provide stability of voltage conditions.. sufficient current will flow through and operate the protective system. The t e r n “ g r ~ u nis d ’therefore ~ used throug~out this chapter. It is therefore required that the connectioq to ground be of sufficiently lowresistance. Grounding is also a measure of protection against lightning. The proper reference ground may sometimes be the earth itself. For protection of power and substations from lightning strokes. In electricity supply systemsit is therefore a common practice to connect the system to ground at suitable points. that of the earth itself) (IEEE. It is essential to mention here that the terms “ground” and “earth” cannot quite be used interchangeably.

encountered to be effectively discharged to the general mass of the earth. Different salts have different effects on the soil resistivity. The soil resistivity depends on the type of soil (Table 13. the ground is termed either a power system ground or a safety ground.1). Sunde.000 Type of soi I Loam.. the soil resistivity increases as the temperature is decreased. A s water resistivity has a large temperature coefficient. garden soil Cl ay Sand and gravel Sandstone Rocks . anapparent resistivityisdefined for an equivalent homogeneous soil. 13. A small quantity of dissolved salts can reduce the resistivity very remarkably. Depending on its main purpose. The value of resistance to ground of an electrode system is the resistance between the electrode system and another “infinitely large” electrode in the ground at infinite spacing.l). m) 5-50 8-50 60-1 00 10-500 200-10. For nonhomogeneoussoils. the resistivity varies accordingly. 13. In most cases there are several layers of different soils. There areeffects of grain size and its distribution. The moisture content of the soil reduces its resistivity (Fig. The soil resistivity is a deterministic factor in evaluating the ground resistance. and dissolved salts. its moisture content. The grounding system should therefore be installed nearest to the permanent water level. Typical Values of Resistivity of Some Soils Resistivity (ft. It is an electrophysical property. to minimize the effect of seasonal variations on soil resistivity. and effects of temperature and pressure. 1982. Homogeneous soil is seldom met. The resistivity of soil depends on the amount of salts dissolved in its moisture. 1981.2). which explains why the resistivities of apparently similar soils from different locations vary considerably (Fig. Nahman and Salamon.1984. 1968). representing the prevailing resistivity values from a certain depth downward (Sverak et al. particularly when large areas are involved. Komaragiri and Mukhedkar. if possible. A s the moisture content varies with the seasons. with a discontinuity at the freezing point.

If a currentI flows into the ground through this hemispherical electrode.0 I Variation of soil resistivity with its moisture content. the total resistance R up to a large radius rl would be (13.3). the lower will bethe resistivity. it will flow away uniformly in all directions through a series of concentric hemispherical shells. The finerthe grading. There is not much experimental work on the effect of pressure. The ground resistance of this electrode is made up of the sum of the resistances of an infinite number of thin hemispheric shells ofsoil. Considering each individual shell with a radius x and a thickness dx. The simplest possible electrode is the hemisphere (Fig.1) . The distribution of grain size has an effect on the manner in whichthe moisture is held. 13. but it is reasonable to assume that higher pressures resulting in a more compact body of earth will result in lower values of resistivity.

3 " " " 4 .tz l I 0.I I I I 1 1 1 I l I I 1 0.1 l Resistivities of different solutions. . " " 5 ' \/ / \ " Current entering ground through a he~ispherica l electrode. il I ' ' ' 3 2 1 ure 13. 0.

in general. the analytical relation for R takes the form P 4 R = -1n- 27tl P 21 R = -1n- 2x1 d (13. This relation is applicable to any shape of electrode. Evaluating their average potential and total charge. however. They should be driven as far apart as possible so as to minimize the overlap among their areas of influence.where p is the earth resistivity.P (13.2 gives approximate formulas for the resistance of electrodes with various shapes. As rl = OQ R . In practice. so it becomes necessary to determine the net reduction in the total resistance by connecting rods in parallel. then (S 3.5) If.2) R=. The resistance of a single rod is. = -P 27tr The general equation for electrode resistance is (S 3. the rod is assumed to be carrying current uniformly along its length. Its ground resistance can be calculated if its shape is approximated to that of an ellipsoid ofrevolution having a major axis equal to twice the rod's length and a minor axis equal to its diameter d.3) 27tc where c is the electrostatic capacitance of the electrode combined with its image above the surface of the earth.4) d If the rod is taken as cylindrical with a hemispherical end.and it is necessary to use a number of rods connected in parallel. The driven rod is one of the simplest and most economical forms of electrode. One of the approximate methods is to replace a rod by a hemispherical electrode having the same resistance. the formula becomes R=P 2nl [ln(~) S] (13. this is very difficult.6) Table 13. the capacitance and hence the resistance of the system can be calculated. . not sufficiently low. The method assumes that each equivalent hemisphere carries the same charge.

V I : 0 t-U) . n T- I I I Ev C V + + l Ql% II Q II g: U c S L A r/.orshedy S: 1 4 .

- c: 0 I 5 c +“I .1 II Q: ‘1x Q II Q: Q13 + 4- a . CL: W N . ( d CL W c: 3 2 ( d +J c .

8) where L is the total length of all conductors.4 The resistance ofn rods in parallel is thus found to exceed (l/n) of that of a single rod because oftheir mutual screening. 1987) (€3. and K1 and K2 the factors presented graphically as functions of length-to-width ratio of . 13. Tm is the maximum allowable temperature. and Ira is the ambient temperature. A the total area of the grid. Such a grid not only effectively grounds the equipment.7) where a is the copper cross-section (circular mils). spaced ina grid pattern of about 3-10 m. the conductors are securely bonded together. A typical grid system for a substantion would comprise 410 bare solid copper conductors buried at a depth of 30-60 cm. The side of grid conductors required to avoid fusing under the fault current I is estimated as (IEEE. The following equation for grid resistance may be used: (13. d the grid conductor diameter. Ground rods m a y be connected to the grid for further reduction in the ground resistance when the upper layer of soil is of much higher resistivity than that of the soil underneath (Garrett and Holley.2.3. At each junction. 1984). The screening coefficient q for n electrodes in parallel is defined as resistance of one electrode = (resistance of n electrodes in parallel) It is difficult to determine the value ofq for complicated systems and usually it is listed in tables obtained by calculations and measurements (Dawalibi and Blattner. 1 r o ~ n d i Grids n~ A common method for obtaining a low ground resistance at high-voltage substations is to use interconnected ground grids. 1980). h is the depth of burial of the grid. but has the added advantage of controlling the voltage gradients at the surface of the earth to values safe for human contact.1 Resistance to Ground and Mesh Voltages of Grounding Grids The resistance to ground determines the maximum potential riseof the grounding system during a ground fault. t is the fault duration ( S ) .

. Using such a model. The surface potential is givenas a percentage of applied grid voltage and the horizontal axis isin centimeters measured from the center of the grid. Mesh voltages of ground-grounding grids are calculated using a relation of the form (IEEE Committee. diameter d .5 (ElMorshedy et al. various approximate formulas based on the similarity of a grid and a round plate of equal area have been proposed (Table 13.. the effect of changing the grounding mesh parameters and adding grounding rods of different depths at different locations can easily be estimated. By measuring the voltage applied to the model and the current flowing through the electrolyte between the model grid and the return electrode. 1982) (1 3.2.172n Scale Models of Grounding Grids Scale model tests with an electrolytic tank are very useful for determining the ground resistance and surface potential distributions during ground faults in complex grounding arrangements where accurate analytical calculations are hardly possible (El-Morshedy et al. . 1986). the effective grid resistance can be evaluated.3.the area.4 for the normal profile. Surface potential profiles for a 4 x 4 mesh grid 10 cm x 10 cm. their spacing S. (13. It is the potential difference between the grid conductor and a point at the ground surface above the center of the grid mesh.9) where I is the current flowing into the ground and Km is a coefficient that takes into account the effectof number YI of the grid conductors.2). Sample results are given in Figure 13.10) Ki is an irregularity correction factor to allow for nonuniformity of ground current flow from different parts of the grid: (13. The mesh voltage represents the maximum touch voltage to which a person can beexposed at the substation. at a depth of 1cm are shown in Figure 13. The maximum and minimum potentials throughout the grid can be determined. For practical design purposes.65 13. 1986). 1mm conductor diameter.11) Ki 0.2 + 0. and depth of burial h.

A I I l 5 2.5 0 ~i~fancefro^ grid center fine t / h . 10 cm x10 cm grid.4 Profiles of surface potential for a 4 x 4 mesh. Conductor diameter = 1 mm. its depth below electrolyte surface = 1 cm. . ~ ’ / / cm .

The numerical calculation results are given in Figures 13. -l Equally spaced grounding grids have some disadvantages. conductor diameter = 1 mm. In addition. This reference gives a computer program based on the surface element method which calculates the emanating current density. the effects of rectangular ground grids-with and without groundrods-on the performance of a grounding .6 shows two rectangular grounding grids with the same grid area and number of buried conductors.8. and resistance of any configuration of grounding grids buried in uniform soil.0 1 2 3 4 5 6 7 8 c m Length o f ground rods w e 13. Grid size = 20 x 20 cm. such as fringing effect. Unequally spaced grids have some advantages in uniform soils over equally spaced grids. 1998). depthbelow electrolyte surface = 1 cm.. 1995). The program gives the optimum spacing between grid conductors. Current density emanating from grid conductors can be more uniform using unequal spacing. resulting in touch voltages on the corner of the grid much higher than those in the center.7 and 13. unequal spacing can greatly save groundinggrid material. This leads to morecurrents emanating from the peripheral conductor of a grid. In a recent work (Lee et al.5 Maximum touch voltage and groundresistance fora model of grounding grids of different mesh as functions of the lengthof ground rods. Figure 13. Unequal spacing can also decrease potential gradients on the earth’s surface and enhance the level of safety for people and equipment.. 1995). potential distribution.. resulting in a more than 30% decrease in grid cost (Huang et al. An unequal spacing technique has been proposedto overcome the drawbacks of equally spaced grids (Huang et al.

Soil resistivity is 250 52 m.rsl 120 21 0 distance (meters) 60 120 . The study reveals that: In uniform or two-layer soils with a low-resistivity top layer.5 10 Y U 3 10.8. C. two-layer.5 0 60 distance (me.7 and 13.5 g 9.. The study involves uniform.6 m.5 5 15 25 3 5 45 55 65 75 85 95 105115125 ‘ 5 9 i 5 15 25 35 45 55 65 75 85 95 105115‘ 5 distance (rnetitrs) distance (meters) Potential distribution.5 10 9.n L: B 0 10 20 30 40 so M) 70 80 00 lMlll0120 A (meters) distance Grounding grids with area of 120m x 80 m and 22 conductors.7 Current density distribution. system have been investigated. and multilayer soils. . unequally spaced grids having denser conductors at the edges provide the most efficient design. 6 . burial depth is 0. (b) unequally spaced grid 11.5 (a) equally spaced grid 11 r_ 11 5: 10. (A. D) refer to the potential curve positions of Figures 13. I (a) equally spaced grid (b) unequally spaced grid J conductor 4.

198’7) C= 18 ln(4Zld) x 10-9 F (13. Its magneticfield is also represented.(b) ~quivalentcircuitfor a rod under impulse. its the soil has a dielectric constant E.9a the current I travels down the rod electrode and from there diffuses into the ground. there will be a conductive current in addition to a capacitive current..12) ure 13. . The capacitance of the ground electrode is (IEEE. Ground rods are effective only if they penetrate a low-resistivity bottom layer to a significant depth.7 Equally spaced grids provide the most efficient designsin soil structures with a high-resistivity top layer. are not in contact with the grid. Knowledge of the impulse impedance of a grounding system is necessary for determining its performance while discharging impulse currents to ground.9 (a) Impulsecurrent spreading from adriven rod. ~ultilayer soils appear to behave like uniform or two-layer soils derived from the multilayer soils by averaging the resistivities of adjoining layers which. In addition to resistivity. as in the case of lightning and transient ground faults. When the electrode voltage changes with time. In Figure 13.

its effective inductance is reduced to one-third the steady-state value.13) At high frequencies of about 1MHz the ground will draw a considerable capacitive current in addition to its conductive current. In the case of a grounding grid.m (13. This is highest wherethe current is most concentrated (i.The current in the electrode and ground produces a magnetic field. however. The total current flows through the self-inductance of the electrode. for lightning protection. 1980). The circuit parameters can be estimated.14a) For a wire buried very near the surface: Q. To account for the impulse current. ance of C ~ o u n ~Cri in~ Ground wires of considerable length are usually buried as counterpoises along transmission lines or near high-voltage substations. flowing off each conductor along its length.. whereas at extremely high frequenciesthe ground would be represented by a distributed network. its equivalent circuit is analyzedin response to the applied impulse current. Naturally.14b) n d The impulse impedance of a buried ground wire in operational form (Verma and Mukhedkar.x d H (13. At low frequencies the inductance and capacitance can be safely neglected. Figure 13. 1981) is expressed as z=J- (13. but decreases with time to the steady-state value at a rate depending on the circuit and wave parameters.m (13. Its distributed resistance to ground is: For a deeply buried wire: R = -1np 42 2n d R = -1np 42 a. knowing the dimensions of the grid. at the top). the impulse impedance is initially higher than the powerfrequency impedance. A buried wire can be represented by the usual transmission-line circuit of distributed constants. The inductance of the grid is the governing factor contributing to its impulse impedance.e. The inductance of such a rod is 4 2 L= 2 2 ln. .9b represents the equivalent circuit (Cupta and Thapar.15) and L d is the where r is the metallic resistance of the wire per meter (a) distributed inductance per meter (H).

and ensures safety to operating personnel. It is also necessary to adjust the total length of buried conductors. where ISs is a coefficient that takes into account the effect of number.17) . The ground-potential rise depends on fault-current magnitude andthe resistanceof the grounding system.7). spacing S. the nature of the soil. The size ofground conductor is given by equation (1 3. Equation (13. as it provides the ground connection for the system neutral. (13. The practical design of a grid requires inspection of the layout plan of equipment and structures. taking into account future expansion of the connected power system. The size of the grid and the number and length of driven rods depend on the substation size.9) gives the value of the mesh voltage. 13. In such cases.42 13. the various ground cables or buses in the yard and in the substation building should be bonded together by heavy multiple connections and tied into the main station ground. and the ground resistance desired. 13. Estep.10) (IEEE. Low-resistance substation grounds are difficult to obtain in desert and rocky areas. the value of the step is given by the formula (Fig. The grid system usually extends over the entire substation yard and sometimesseveral meters beyond. and depth h of burial of the ground conductors. to be at least equal to those required to keep local potential differences within acceptable limits. including crossconnections and rods. The ground conductor should have low impedance and should carry prospective fault currents without fusing or getting damaged. the discharge path for surge arresters. the use of grids willprovide the most convenient means of obtaining a suitable ground connection. The ground of a substation is very important. Many utilities add ground rods for further reduction of the resistance. To equalize all ground potentials around the station. It also provides a low-resistance path to ground to minimize the rise in ground potential. 1987) voltage.

21) . is (IEEE. is (IEEE. touch. The tolerable touch voltage.25~1. The number of terns within the parentheses is equal to the number of parallel conductors in the basic grid. excluding cross-connections. in ohm-meters. 1987) Etouch = 165 0 . which is the voltage between any point on the ground where a person may stand and any point that can be touched simultaneously by either hand. is the resistivity of ground beneath the feet. The tolerable step voltage with duration t.neutral grounded at otentialriseabove Step.19) Equating the value of KmK@l 165 Emesh to the maximum value of Etouch yields (13.18) where p . Etouch. 1987) (13. and transferred voltages near agrounded structure. taking its surface treatment into account.20) " + 0. 2 5 ~ ~ + & (13. which is the voltage between any two points on the ground surface that can be touched simultaneously by the feet. Estep. & L The approximate length of buried conductor required to keep voltage within safe limits is thus (13.

some contingencies may developferroresonance. the ratings of switchgear needed to cope with them. the equipment insulation.. This type of neutral grounding. transformers. They also include tuned reactance. however. For example. if not suitably designed. with the same method of neutral grounding. It also directly affects the levels ofshortcircuit currents in the power network and. Each ofthese methods has advantages and limitations. excessive overvoltages appear on the system in the case of line-to-ground (L-G) faults. accordingly.1 UN Grounding of the neutral points of generators. solid grounding. would be very vulnerable. The ‘%ealthy” phases acquire transient overvoltages several times higher than the normal peak phase voltage (Chapter 14). as ithas a considerable bearing on the levels of transient and dynamic overvoltages stressing the equipment insulation. and transmission schemes isan important item in the design of power systems. Gulachenski and Courville. with isolated or high-impedance grounding. 1984). Also. and grounding through a high-impedance such as that of a potential transformer (Fig.11 Examples of power system neutral grounding. causing high power-frequencyovervoltages. is favored only in some LV and MV isolated networks where the needof supply continuity is extremely pressing and the equipment insulation is adequate. In both cases. c . the magnitudes of L-G fault currents would be so low that only special protective gear could detect them. The methods of system neutral grounding include resistance and low reactance for effective grounding. Neutral grounding through a reactor tuned to match the system capacitance C to ground (= 3mC) neutralizes the system L-G fault currents l c ure 13.11) (Brown et al. 19’78. Further. 13.

otherwise. On the other hand. 13. 1982). as it helps to maintain their stability by the power consumed during L-G faults. Combination of these letters gives three possible configurations: TT. Exposed conductive parts connected to neutral (TN) 2.almost completely. Ungrounded neutral (IT) The purpose of these three systems is to protect persons and property. and IT. By grounding solidly or through a small reactance. the generators might race out of step (Ershevich et al. The LV grounding system characterizes the grounding mode of the secondaryof the MV/LV transformer and the meansofgrounding the installation frames. the system overvoltages are limited to their possible minimum (Chapter 14). It helps maintain the continuity of the supply without endangering the system insulation. Grounded neutral (TT) 3. For resistance grounding. TN. . T for directly connected to the ground and N for connection to the neutral at the origin of the installation. 1980): 1. This method has been in common use in some high-voltage networks in Europe. whichisconnected to the ground (Fig. This technique is suitable for generators. Identification of the systemtypesis thus definedby means of two letters: the first one for transformer neutral connection. 13. T for neutral connected to groundand I for neutral isolated from ground (Fig. the L-G short circuit currents will be excessive unless the grounding reactor is designed with a suitable magnitude.12a). The different methods of grounding low voltage systems are (IEC. Thusthefaultarc becomes unstable and easilygets extinguished.12b). the second one for the type of application frame connection. a resistance of a suitable design is connected between the system neutral point and the grounding electrode group as an additionto the system ground resistance.. They arenot identical in dependability of the LV electrical installation with respect to electrical power availability and installation maintenance.

several grounding systems may be included in the same installation. 13.13. TN-C-S: use of a TN-S downstream from a TN-C. TN-S: if the N and PE conductors are separate. The main difference between gas-insulated substations (GIs) and conventional substations is their metallic enclosures.Connection modes of the neutral of the installation and of the frames of the electrical loads. . Under faults. as shown in Fig. Each grounding system can be applied to anentire LV electrical installation. these enclosures carry induced currents ofsignificantmagnitude. The TN system includes several subsystems: TN-C: if the neutral conductor (N) and protective conductor (PE) are one and the same (PEN). The use of this systemiscompulsory for networkswithconductors of a crosssection less than or equal to 10 mm2 Cu. The opposite is forbidden.These currents mustbe confined to specific paths (Chapter 10). However.

All metallic enclosures should normally operate at ground voltage level.13 installation. reinforced-concrete foundations may cause irregularities in a current discharge path. No significant voltage differences should exist between individual enclosure sections.13.19) can be used with ps= 0 for a metal-tometal contact. Dangerous touch and step voltages within the GIs area are drastically reduced by complete bonding and grounding of the CIS enclosures. 1982). 4. GIS should be properly designed and adequately The enclosures of grounded so as to limit the potential difference between individual sections within the allowable limit of 65-130V during faults (Sverak et al. The use of a simple monolithic slab reinforced by steel serves as an auxiliary grounding device. 2. The supporting structures and any part of the grounding system must not be influenced by the flow of induced currents. 3. and by using grounded conductive platforms connected to the GIs structures.. As GIS substations have limited space. For this. 5. Example of the various grounding systems included in the same The following requirements should bemet effects caused by circulating currents: to minimize the undesirable 1 . Precautions should be taken to prevent excessive currents being induced into adjacent frames and structures. The analysis of GIs grounding includes estimation of the permissible touch voltage. The reinforcing bars in the foundations can act as additional ground electrodes. . equation (13.

or from the voltage rise .. Dangerous voltages to earthmay result from insulation failure. The influence ofthe grounding impedance connected to the bushing enclosure and the gas circuit breaker enclosure is negligible. charges due to electrostatic induction. circuit with an inductance and a resistance of 1 p Overvoltages in the CIS due to lightning surges were analyzed using this grounding impedance.1 The transient impedance of a grounding grid for a commercial 550 kV GI§ was measuredon site using steep front currents with rise timefrom 100 nsto 2 ps (Karaki et al. counterpoises. The method used to reduce the equivalent footing resistance and the degree to which this is carried out is a matter of economics. It was noticed that: The radius of the grid area in which the grounding potential is raised by the injected steep current is very small (1. and by earthing tower bodies and foundations. is more economical than adding extra insulation. Their design is discussed in detail elsewhere (IEEE Committee.5 m). it is necessary to have a suitable groundingsystem. The degree to which low tower footing resistance can be met depends on local soil conditions. The grounding impedance of the CIS enclosure can be simulated by attaching the groundingimpedanceobtained to eachgrounding point of the CIS enclosure. Experience indicates that some means of reducing the footing resistance to an equivalent of 10 5-2. The isolation of an arrester enclosure end and the grounding impedance of the surge arrester reduce the protecting performance of the surge arrester in the same manner. flowof currents through the sheath. With underground cables the situation is different. Ground wires are so located as to shield the line conductors adequately from direct lightning strokes. 1975. improved grounding cannot be economically effective in the event of direct lightning strokes to phase conductors. 1995). 1 To design and operate a transmission system with a low-outage rating and safety to the maintenance personnel as well as the public. as measured with the ground wires removed. 1978). it was found that the transient impedance of the grid for those currents is simulated by a series H and 3 5-2 respectively. Unfortunately.This can be furnished by overhead ground wires. The importance of having low towerfooting resistance in this case is in avoiding a high rate of back flashovers and thereby improving the conditions for successful fault suppression by ground-fault neutralizers. From the measured results.

There isno simple relation between the resistance of the ground system as a whole and the maximum shock current towhich a person might be exposed. isolation transformers.magnitude. one’sbodyshouldunder no circumstances be a part of an electric circuit. 1981.. stoppage of the heart. 198l ) that for the same effect the current magnitude I varies with its duration t according to a relation of the form I t”1/2. The effects ofan electric current passing through the vital parts of a human body depend on its duration.. and employing high-frequency and direct current. uncontrollable muscular contractions can make breathing difficult.” although unpleasant tosustain. 1983). It hasbeenobserved verak et al. The human body can tolerate slightly larger currents at 25 Hz and approximately five times larger direct currents andstill larger currents at 3-10 kHz (IEEECommittee. and frequency.The first four methods are normally adopted in high-voltage systems. Safety of operating personnel requires grounding of all exposedmetal parts of power equipment. isolation. or inhibition of respiration mayoccur and cause severe injury or death. insulation and double insulation. Using. There are several means of reducing the hazard of electric shock. Currents of 1-6mA. 1982). often termed “let-go currents. Safety ground is meant to ensure that persons working withelectrical equipment will not be exposed to the danger of electric shocks. In the range of 100 mA ventricular fibrillation. as the risk of electric shock will thus be greatly mitigated. High-speed clearing of ground faults is evidently crucial. it is possible to calculate the tolerable potential difference . El-Kady and Vainberg. shock limitation. do notimpair a person’s ability to controlhis or her muscles. To avoiddanger from electric shock. For still higher currents. the magnitude of the tolerable body current and the appropriate circuit constants. including grounding. Human beings are most vulnerable to currents with frequencies of 50-60 Hz. guarding. Thus a low station groundresistance is not in itself a guarantee of safety (Sverak et al. Currents of 9-25 mA may be quite painful and can make it hard to release energized objects grasped by the hand.during discharging of faults to the station ground system to which the sheaths are connected.

especiallywhen a patient in a hospital is actually connected into the circuit. 1983).1 Decreasing the ground resistance of a grounding system in high-resistivity soilis often a formidable task. The shock voltage in this case may be equal to the full voltage rise of the ground grid under fault conditions (ElKadyand Vainberg. Increased application of electric instrumentation has greatly increased the risk of electric shock. When a person standing within the station area touches a conductor grounded at a remote point. Some medicalauthorities believe that a current as small as 20 p4 at 50-60 Hz applied directly to the heart can produce ventricular fibrillation. 13. 1999). filling the holes with lowresistivity materials (LRM)under pressure (Meng et al.23) where C is a factor depending on the soil homogeneity and equals l for uniform soil. This method requires 3 steps: drilling deep holes in the ground. 13. Recently.between possible points of contact. Most of the . he or sheis subjected to a transferred potential. 13.10). The hazard increaseswhen probes or needles are inserted into the body. developing cracks in the soil means by of explosions in the holes.somenew methods havebeen proposed to decrease ground resistance. The hazard is increased if pacemakers are used. Sometimes these are placed directly on or inside a heart chamber..22) (13. Three methods are listed in the following section. or when a person standing at a remote point touches a conductor connected to the station ground grid.10) increases the closer one gets to the site of a ground fault or to the grounding point. The step voltage (Fig. If the enclosure of grounded equipment in which an earth fault has occurred is touched by a person. 13. The tolerable step and touch voltageswith durations t for persons weighing 50-70 kg can be estimated as (1 3. he or she will be subjected to a potential termed a “touch voltage’’ (Fig.

5-2 times the length of the vertical conductor. Chemical rodsare electrodes withholes along their length. They provide stable protection for many years.15). and acomplex network of low-resistivity tree-like cracks linked to the substation grid is formed (Fig. ~ r o u n ~ i system ng with explosion and intrinsic cracks. dry deserts. The rod absorbsmoisture from both air andsoil. . This method is effectivein reducing ground resistances in rocky areas. Continuous conditioning of a large area insures an ultra-low-resistance ground which is more effective than a conventional electrode. The specially formulated mineral salts are evenly distributed along the entirelength of the electrode (Fig. They may be used in rocky soils. If the conductive salts are runninglow. Fieldtests show that the optimum span between vertical conductors is in the range of 1. the lower the ultimate grounding resistance. filledwith mineral salts. These rods are available in vertical and horizontal configurations.cracks around the vertical conductors will be filled with LRM.1 times the length of the rod. 13. About 95% of the grounding resistance of a given electrode is determined by the character of the soil within a hemisphere whose radius is 1. It is obvious that replacing all or part of that soil with a highly conductive backfill will facilitate the achievement of a low-resistance ground connection. 13. The greater the percentage of soil replaced. or tropical rain forests. the rod can be recharged with a refill kit.14). freezingclimates.

while the other end is not practically available and is called“remote soil”.The resistance ismeasured using alternating current to avoid possible polarization effects under direct current (IEEE. 1983.) Ground resistance measurement consists of measuring the resistance of a body of earth surrounding a grounding electrode system. about 98% of the total resistanceis contained within a finite distance from the grounding system. Dawalibi and Blattner. 1984). (Courtesyof LEG. The power-frequency reactances of the grounding system can safely be neglected unless its ohmic resistance is extremely low.C~~mis ~ easily o d inspected and recharged through cap at topof electrode t Chemical rod. The practical and reliable method for measuring the resistance of a ground is that of “fall of potential. Inc. however. In practice. One end of the resistance is definitely available at the grounding system itself.” .

It involves passing through the grounding system G a current I to return from another electrode C (Fig. Vx is measured by a potential probe P. The simplest form of the fall-ofpotential method is obtained when G. (b) resistance plots for the electrode separation tooshort (i) and adequate(ii).Since the performance and cost of any grounding electrode or ground grid is directly related to soil resistivity. curves similar to those shown in Figure l3.being (a) circuit connection. accurate evaluation is a matter of considerable importance. and C are ona straight line and P is located between G and C.16b are produced. If the distance D is large enough with respect to the grounding system dimensions. This method has several variations and is applicable to all types of ground resistance measurements. The passage of this current produces at a distance x from G a voltage drop Vx in the soil. The most common method of measuring soil resistivity employs four electrodes driven into the soil. When Vx/lis plotted as a function of the potential probe distance x.Ground resistance measurement by the fall-of-potential method: D. P. Soilresistivity tests should be carried out beforedesigning a grounding system. the center part of the fall-of-potential curve tends to be flat. It is usually accepted that the flat section of the curve gives the correct magnitude of the resistance RG.16a). The theoretical basis of this method has been . 13.

25) It is important to note that in the relation (13. which is a function of soil resistivityand the electrode geometry.24) where R is the apparent resistance (Q) and p is the soil resistivity (S2 m). When a > >b.26) reduce to (13.. .24) and (13. 1987). The ratio between the observed potential and the injected current is referred to as the apparent resistance. The apparent soil resistivity is computed by multiplying the apparent resistance by a geometric multiplier.25).24) reduces to p = 27raR (l 3. It relaxes the requirement that rod electrodes have a large spacing compared to their buried length.25) the electrodes are considered as “points”buried at a depth b. and making electrical contact with the earth only at the bottom.)2 where b is the electrode driven length in meters. P= 2 ln[(2 + E)/(1 + F)]+ 2F Ea/b 47rRb (13. Small electrodes are inserted into foursmall holes in the earth all to a depth of b meters and spaced along a straight line at intervals of a meters. The following new formula based on the field theory relating soil resistivity to measured resistance has been developed (Baishiki et al. The apparent soil resistivity down to a depth equal to a is (13. A test current I isinjected through the earth between the two outer electrodes and the potential E between the two inner electrodes is measured with a potentiometer or high-impedance voltmeter. it can be easily shown that equation (13. equations (13.26) with F= JI + (. For spacing a much greater than depth b .derived.

For measuring the resistanceof a ure 13. This “ohmmeter” compares the measured potential with the main circuit current I. Figure 13.17 showsa schematic of the circuit in common use. The circuit shown in Figure 13. where a hand-driven DC generator C is coupled to a mechanical inverter CR and to a mechanical rectifier PR.17 is for measuring the soil resistivity according to the four-point method.Any instrument for measuring the resistance of a grounding point and the resistivity of soil mainly comprises an ACsource of an adjustable frequency and a measuring circuit. any pickup would cause errors in the measurements. The inverter inverts the DC current I to AC (of a frequency adjusted by the speed of cranking the generator) before it is fed into the ground. The frequency to be used should differ from that of any neighboring network.17 Schematic connection of a ground resistancemeter used for measuring the ground resistivity by the four-electrode method. DC current would not be suitable for such tests becauseof the entailed electrolysis and polarization. . introducing serious errors. otherwise. The rectifier PR converts the AC potential picked up between the electrodes P1 and P2 back to DCbefore it is fedto its respective coil in the ohmmeter CC.

and its diameter is 6cm. Comment on your results.18. If the earth resistivity is 80 S?.grounding electrode already installed. the length of the driven rod is 8 m.17) combined.000 S 2 and 500 S2 respectively. In the above problem calculate the current passing through a man in both positions.20.1 H 10 m Figure 13. driven vertically at the points shown in Figure 13. A grounding system in a sandy clay soil is composed of five similar driven electrodes of 8 m length and 5cm diameter. (2) Repeat the above problem for the grounding system shown in Figure 13. Discuss the results. If the short circuit current is1500 A. it would be connected to the ground resistance meter at the two electrodes C l and P1 (Fig. given that the earth resistive p = 100 S2 m.19. calculate the actual earth resistance and the overlapping coefficient. Calculate also the touch voltage for a man with arm length 60cm.1 4 (1) Calculate the ground resistance for the grounding system shown in Figure 13. 0 0 0 0 7 7m A rod of radius 5cm is driven in a clay soil of resistivity 90 S2 . Calculate its required length to have a ground resistance of 4 S2. 13. m.m. if his resistance is 10. - l0m l0m 7m 7m Figure 13. and include electromechanical vibrators or solid-state devices for the current inversion and rectification. + . 13. calculate the step voltage for a man standing 7 m from the rod whose step length is 80cm. Some variations of the ground resistance meter use a battery instead of the generator C.

1995. Courville EW. IEEE Trans PAS -97:683-694. E1-Mor~hedy Ershevich VV. Sheimovich VD. Murase H. IEEE Guide for Safety in Alternating Current Substation Grounding. IEEE Trans PAS-103:374"382. Ghourab MM. Brown P.4 24 40 1. 1983. IEEE Trans PAS-101:4006-4023. IEEE Trans PAS-103:2572-2578. 1975. IEC. CIGRE paper 32-09. 1980. IEEE Trans PWRD-10:716--722.10 m _ 1 l--"l 10 m (6) The earth resistivity was measured by the four-electrode method. 1972. IEEE Trans PWRD-10:723-731. IEEE Guide for Measuring Ground Resistance and Potential Gradients in the Earth. Chen X. Vainberg M. Garett D. 1978. Yokota T. R a (m) (a) 12 1. 1984. Yan H. Blattner C. The resistance R in ohms varied with the probe separation a as follows: . IEEE Trans PAS-102:3080"3087. Gulachenski EM. Publication 80. 1983. IEEE. Holley H. Zeitoun AG. IEEE. Publication 81. 1986. Nojima K. IEEE Committee. 1978.08 Calculate the average resistivity down to depths of 30. 1983. IEEE Trans PAS-100:2993-3001. 1981. Komaragiri K. 1980. Karaki S. NewYork: Institute of Electrical& Electronics Engineers. Takahashi H. Dawalibi F. Dawalibi F. El-Kady M. Thapar B.5 100 0. Baishiki RS. 1980. A.Publication 142. Neklepaev BN. 50. Krivwshkin LF. IEEE Trans PWRD-2:64"71. New York: Institute of Electrical & Electronics Engineers. IEEE Trans PAS-97:2243-2252. IEEE Recommended Practice for Grounding of Industrial and Commercial Power Systems. . 1982. 1987. 1995. New York: Institute of Electrical & Electronics Engineers. 1982. IEEE Trans PAS-94: 1241-1247. Osterberg CK. IEEE Committee. IEEE Trans PAS-99:2357-2362. IEEE. Slavin GA. Yamazaki Y. Gupta B.1 0. Kojima S. Stevenson J. Proc IEE 13363287-292. IEEE Trans PAS-99:2008-2011. Mukhedkar D. Publication 364. IEEE committee.1987. Huang L. and 80 m. Johnson I.

Salamon P. Kim JH. IEEE Trans PAS101:4006-4023. 1982. IEEE Trans PAS-100. Podds TH. Pawalibi FP. 1984. Vema R. Sunde ED. 1981. He J. Keil RP. 1998. Sverak JG. Sverak JG. IEEE Trans PWRP-13:’745-751. New York: Macmillan. Earth Conduction Effects in Transmission Systems. Verma R. 1999. Podds TH. Pate1 SC.4281-4290. Ma J. Benson RU. IEEE Trans PAS-103:880-885. Idzkowski JE. Mukhedkar D. Heppe RH. Garret DL. Pawalibi FP. . Zukerman LC. 1968. Ma J. Meng Q. 1981. IEEE Trans PAS-100:1023-1030. Smith GE. Nahman J. IEEE Trans PWRD-14:911-916. Dick WK. Ragan ME. Dick WK.Lee HS.

This Page Intentionally Left Blank .

Giza. they remain the initial factor that determines its dimensions.and frequency of occurrence. The voltage stresses on transmission network insulation are found to have a variety of origins. However.Egypt The examination of overvoltages on the power system includes a study of their magnitudes. This study shouldbeperformed not only at the point where an overvoltage originates but also at all other points along the transmission network to which the surges may travel. lightning is the most common and the most severe. In normal operation AC (or DC) voltages do not stress the insulation severely. Externalovervoltages: generated by atmospheric disturbances. Of these disturbances.vervoltages on ower Systems CairoUniversity. durations. shapes. Overvoltages stressing a power system can generally be classified into two main types: 1. .

the development stages of a lightning flash are depicted. This is the start of the return stroke. which progresses upward like a travelingwaveon a transmission line. The lightning discharge through air occurs as one of the forms of streamer breakdown of long air gaps explained in Chapter 4.1. or a discharge between the negative charge mass in the lowercloud and the positive charge pocket below it. The upward leader joins the downward one at a point referred to as the striking point.positive and negative charges are separated by the movements of air currents forming ice crystals in the upper layer of a cloud and rain in the lower part. According to theories generally accepted. The channel to earth is first established by a stepped discharge called a leader stroke. During thunderstorms. during which positive charges are transferred upward in the form of lightning (Lewis. As the separation of charge proceeds in the cloud. Internal overvoltages: generated by changes in the operating conditions of the network. which is responsible for the known damage of lightning. The leader is generally initiated by a breakdown between polarized water droplets at the cloudbasecaused by the high electric field. Internal overvoltages can be divided into (a) switching overvoltages and (b) temporary overvoltages. This is then counteracted by thunderstorms.Only a part of the total charge-several hundred coulombs-is released to earth by lightning. lightning is produced in an attempt by nature to maintain a dynamic balance between the positively charged ionosphere and thenegatively charged earth (Marshall. the rest is consumed in intercloud discharges. an upward leader begins to proceed from earth before the former reaches earth. The total potential difference betweenthe two main charge centers may vary from 100 to 1000MV. At the earthing point a heavy impulse current reaching the order of tens of kiloamperes occurs. As the downward leader approaches the earth. The height of the thundercloud dipole above earth may reach 5 km in tropical regions.2. 1965). 1973). The cloud becomes negatively charged and has a larger layer of positive charge at its top. Over fair-weather areas there is a downward transfer of positive charges through the global air-earth current. The velocity of progression of the . In Figure 14. the potential difference between the concentrations of charges increases and thevertical electric field alongthe cloud also increases.

for a typical overhead line surge impedance Zo of 300 52.1) v =E() 2 The lightning current magnitude israrely less than 10 kA (Bergeret al. 1 ++++- ++++ t ure 14. The corresponding current heats its path totemperatures up to 2O. In practice. The lightning stroke injects its current into a termination impedance Z . indeed. t. however. Therefore.1 Developmental stages of a lightning flash and the corresponding current surge. the lightning surge voltage will probably have a magnitude in excess of 1500 kV. Equation (14. return stoke is very high and may reach half the speed of light.+ + + + + + + + i T I * t. Equation (14. 19'75) and thus.00O0C. which in this case is halfthe line surge impedance Zo since the current will flow in both directions as shown in Figure 14.2.1) also indicates that the lightning voltage surge willhave approximately the sameshape characteristics. the .1) assumesthat the impedance ofthe lightning channel itself is much larger than hZo. The current pulse rises to its crest in a few microseconds and decays over a period of tens or hundreds of microseconds (Ragaller. it is believed to range from 100 to 3000 52. the voltage surge magnitude at the striking point is 1 (14.causing the explosive air expansion that is heard as thunder. 1980). The most severe lightning stroke is that which strikes a phase conductor on the transmission line as it produces the highest overvoltage for a given stroke current..

Lightning strokes represent true danger to life. Overvoltages produced on transmission lines by lightning strokes are only slightly dependent on the power system voltages.e. Lightning is always a major source of damage to powersystemswhereequipment insulation maybreakdown under the resulting overvoltage and the subsequent high-energy discharge..6). power systems. switching surges have become the governing factor in thedesign of insulation for EHV and UHV systems.Anis u=120/2 l Development of lightning overvoltage. . and communication networks. their magnitudes relative to the system peak voltage decrease as the latter is increased. As a result. laboratory-simulated switching surges) (IEC. atructures. There aretwo fundamental reasons for this shift in relative importance from lightning to switching surges as higher transmission voltages are called for: 1. 2. With the steady increase in transmission voltages needed to fulfill the required increase in transmitted powers. which is typical for switching surges. lightning overvoltages come as a secondary factor in these networks. According to the International Electrotechnical Commission (IEC) recommendations. As shown in Chapter 15. shapes and magnitudes of lightning surge waves get modified by their reflections at points of discontinuity as they travel along transmission lines (Section 14. all equipmentdesigned for operating voltages above 300 kV should be tested under switching impulses (i. In the meantime. external insulation has its lowest breakdown strength under surges whose fronts fall in the range 50500 PS. 1973).

the sinusoidal supply voltage is suddenlyapplied to it as represented by the single-phase circuit of Figure 14. 3. The following specific switching operations are some of the most common in this category: a. The resistance R includes all series resistances of the line and transformer. Switching of high-voltage reactors b. e n switching t. as it represents the voltage at the open-circuit end of the line. however. Switchingof transformers thatare loaded by a reactor on their tertiary winding c. The voltage across the capacitor C is the one under study here. ~ne~gizution of trans~is~ion lines and cables. This means the energization of a transmission line carrying charges trapped by previous line interruptions when high-speed reclosures are used. The switching operations of greatest relevance to insulation design can be classified as follows: 1.There is a great variety of events that would initiate a switching surge in a power network. are the following operations: a. Of particular importance. This may also be followed by opening the line at the sending end in what is called a line dropping operation. Energization of a line that is open circuited at the far end b. Fault initiation andclearing. The tramformer is represented here by its leakage inductance and the line by its inductance and capacitance to ground. The circuit performance after switching may be expressed by the following differential equation: . ~eenergization of a line. The switching operation is effected at an instant T seconds beyond that of zero voltage. operations invol4. Energization of a line that is terminated by an unloaded transfomer c. Energization of a line through the low-voltage side of a transformer 2. When an unloaded transmission line is switched on. Switching of a transformer at no load 5. This is effected by a circuit breaker opening at the far end of the line.3. witching on and off of e q ~ ~ ~ All ving an element ofthe transmission network will produce a switching surge. Load rejection.

2) (14.3 Energization switching transient. us(t) = Ri(t) 1 +L-di(t) + dt C (14.2) and (14.laf sin(q t + p) (14. by using operational calculus) and the expression for the voltage across the line capacitance takes the form sin(wt uc(t) = yC where 0 = tan" +WT 0) + ..4) .g.45 nis ure 14.3) The supply voltage us(t) beyond the switching instant is given by us(t) = V' sin(ot + wT) Equations (14.3) can easily be manipulated (e.@CJ -R W Ll/WC V S .

4. sustained overvoltages) differ from transient switching overvoltages in that they last for longer durations. in effect.1 Example A typical 132/500kV.."--2 z m = 184 HZ This.015 pF/km. Temporary overvoltages (i.e. letting a = 0). which is about 1/4h. the frequency of oscillation of the transient voltage component as given in equation (14. typically from a few cycles to a few seconds. the total series inductance L of Figure 14..5) ~ m = ? l - = 2Xfl 14. a 500 kV transmission line mayhave a seriesinductance of about 1 mH/km and a capacitance to ground of about 0.3 would be 500 mH and thecapacitance C would be 1. .e.4) would be h=-.453 A = -vc sin(wT 0) sin p p = tan" a=- us sin(oT 0) ocos(wT 0) a sin(wT 0) + R oo = W1 2L 1 (14.200 "VA transformer would have an inductance (referred to its 500 kV side) of about 400mH.2. is Lest l io6 = 4 x 400 ~ = 1360 ps In Figure 14.3 the expected switching surge is sketched for theprevious case when energization is made at T = 0.5 pF. For a transmission line of 100 km. Neglecting all series resistances (i. They take the form of undamped or slightly damped oscillations at a frequency equal or close to the power frequency. mean:S that the time to crest of the switching surge.

and Xc the equivalent capacitive input reactance of the system.4. by V=EX C X C X S (14. which is assumed constant over the subtransient period and equal to its valuebefore the incident.nis The classification of temporary overvoltages as distinct fromtransient switching overvoltages is due mainly tothe fact that the responses of power network insulation and surge arresters to their wave shapes are different.7) where Vr and Vs are the receiving-end and sending-end voltages.The amplitude of the overvoltage can be evaluated approximately. When a t r a n s ~ s s i o n line or a large inductive load that is fed from a power station is suddenly switched off. Xs the transientreactance of the generator in series withthe transformer reactance. The Ferranti effect of an uncompensated transmission line is given by (14. respectively. Bois the phase shift constant of -the XS Equivalent circuit during load rejection .6) to be where E is the voltage behind the transientreactance. the generatorwill speed up and the busbar voltage will rise. and t is the line length (km). as illustrated in Figure 14. Someof the most important eventsleading to the generation of temporary overvoltages are discussed briefly below.

will only permit phase-to-ground overvoltages well below the line-to-line value. A single line-to-ground fault will causethe voltages to ground of the healthy phases to rise. Bohas a value of about 6" per 100km at normal power frequency (Diesendorf. 1 It m. Therefore. in general. 19'74). of the line per unit length.An earth fault factor is defined as the ratio of the higher of the two sound phase voltages to the line-to-neutral voltage at the same point in the system with the fault removed. a transmission linemaybe looked upon as being made up of a chain of lumped components such as that shown in Figure 14. on the other hand. armonic oscillations in power systems are initiated by system nonlinearities whose primary source is that of the saturated magnetizing characteristics of transformers and shunt reactors. in which R = line resistance (a/m) L= C= G= dx = line inductance (H/m) line capacitance-to-ground (F/m) line leakage conductance (S/m) length of a line element (m) . In the case of a line-to-ground fault. The magnetizing current of these components increases rapidly and contains a high percentage of harmonics for voltages above the rated voltage. It is equal to the imaginary part of where Z and Y are the impedance and admittance of the lineper unit length.5. Solidly grounded systems. systems with neutrals isolated or grounded through a high impedance may develop overvoltages on healthy phases higherthan normal line-to-line voltages. 1 Any overvoltage surge appearing on a transmission line due to internal or external disturbances will propagate in the form of a traveling wave toward the ends of the line. During its travel the overvoltage surge will. and corona discharges. line losses.line per unit length. saturated transformers inject large harmonic currents into the system. For a &= o m where L and C are the inductance and capacitance lossless line . To be able to develop the time equation of the traveling wave. experience attenuation and distortion inflicted upon it by earth resistance.

we easily obtain 3% ~ (14. The magnetic flux linkage produced by the wave’s current is given by 4 = IL dx Consequently. Equations (14.Therefore.8) +Li)id.x (14. 11) and (14.13) . The solution to equations (14.1 1) and (14. the total current change takes the form -di = ( G + C$)udx y differentiation.nis L R Transmission-line equivalent circuit.12) are the known transmission-line traveling-wave equations.9) The total change in wave current is caused by the leakage component uGdx and the current used to charge the line capacitance to ground.11) and (14.12) has the general form S + + (14. the total voltage drop over an element dx is given by -du = ( R (14.12) being the differential operator. and (G CS) is the admittance operator Y . in which the term (R Ls) is the impedance operator Z .10) ax2 = [(R: Ls)(G + +C S ) ] ~ (14.

19) . The exact solution to equations (14.12) depends on the nature of I" (i. according to Taylor's expansion. The operator I" can easily be shown to take the form (14.15) The time functions Fl(t)and F2(t) are integration constants with respect to x only.and (14.. Then the operator I" reduces to r=s. the voltage and current waves along a lossless line take the general forms (14. Remembering that.14) where the operator I" is defined as r = m (14.e.1 1) and (14.LE=Y S (14.16) in which y has the dimensions of velocity (m/s) and is the attenuation constant R G is the wavelength constant An ideal transmission line with no losses should haveboth R = 0 and G = 0. on the line constants) and on the boundary conditions of the wave.18) (14.17) while both the attenuation and wavelength constants vanish.

as it is expressedin terns of ( t zt: x/y). temed the line’s surge impedance Z0. .indicates that the wave is traveling along x at a velocity y while maintaining its shape such that fora given value of ( t zt: x/y) the wave’s instantaneous value is unchanged.nls This solution.18) and (14. 14. The voltage (or current) wave is also seen in equations (14.19) to be made up of a component F2 traveling forward along x and a component Fl traveling backward (Fig. The magnitudes of the voltage and current waves are related as follows: For the forward component For the backward component This means that the voltage and current waves traveling along the line have a fixed ratio. Voltage and current traveling-wave components.6).

. At that speed a surge will travel 1 km in about 3.9) that (14. For a lossless transmission line. Therefore.20) The surge impedance is clearly independent of the line length.. an expression for that velocity may easily be deduced from equations (14. (14. respectively. of the insulation is larger than unity whereas the permeability remains very much equal to '0.21) Similarly. the dielectric constant E.22) as y=- m H/rn 1 (14. the capacitance C per unit length of the same conductor arrangeme~t is given by (14. the velocity of wave propagation is given by y"- l d m (14.21) and (14.26) which is equal to the speed of light and is independent of line geometry.24) Also.=g 14. In the case of cables.23) As we remember. (14. it can easily be shown from equation (14.22) Realizing that dx/dtis the velocity at which the surge propagates.25) where p 0 and E' are the permeability and p e ~ i t t i v i t y of free space. it is about 300-400 S2 for overhead transmission lines and about 30-80 SZ for underground cables. the inductance L per unit length of a conductor with radius r and height h above ground is approximately given by L =-In"2h J G r (14. In practice.3 vs.

efraction of raveli in When the traveling wave on a transmission line reaches a point beyond which the line constants are different (e.4 1 6 nis Therefore. = i2 I i I I I -1 i l I ure 14. the reflected and refracted waves will have voltages v\ and v2 and currents ii and i2. A surge that is traveling along line 1 has its voltage and current related by v1 =2 1 (14.3 1) and il + i. applying Kirchhoff s laws at the junction yields v1 + v: = v2 (14.28) and 1 2 =- . the velocity of wave propagation should be expected to be smaller than the speed of light by a factor of &. a part of the wave is reflected back along the line (the reflected component) and another part is passed on tothe new section (the refracted component).30) (14. . Consider a junction betweentwolineswhose surge impedances are Z1and Z2.Thesevoltages and currents are related by (14..29) 2 2 Simply. respectively. such as that shown in Figure 14.g.27) il At the junction.7 Wave reflection at a junction.7. a cable connectedto an overhead line or where equipment or more lines are connected). v2 (14.

461 Therefore. The time-space diagram following the issuing of a voltage surge at the sending end of 1p.1963)-is applied in Figure 14. (14.voltage [rT t -= (r + 2 ) a= (1 a)(l a + a2 a3 + .8 to the simple case of a lossless transmission line with surge impedance Zo connected at the sending end to an infinite bus (i..) is termed the reflection coefficient.E. A method-normally called the lattice diagram (Bewley. + + The successive reflection of a traveling wave at the sending and receiving ends of a transmission line may lead to a buildup in the voltage or current at some points along the line. the voltage at the receiving end (R. ) (14. Incident waves at the receiving end are reflected with a reflection coefficient of Subsequent reflections at the sending end occur with a reflection coefficient of -1..E.e.35) After a sufficiently long time. the receiving-end voltagein this case settles at a final value of . is illustrated. The time taken by the surge to cross the entire line is constant and denoted by T.33) (14.34) The quantity (Z2ZI)/(Z2 2. For example.8 can be evaluated to take the form indicated for the case: R.one that has a zero internal impedance) and terminated at the receiving end by a load impedance 2. .32) The reflected and refracted wave voltages can finally be put in the form (14.) of the system shown inFigure 14. and the quantity 2Z2/(Z2 Zl) is the refraction (transmission) coefficient.u. The voltage buildup at any chosen point along the line can be estimated by the algebraic summation of the voltage surge magnitudes after each surge reflection.

5 0. . " " " " " " / i Q.nis a =0.

Corona discharge has the effectof retarding the portion of the wavefront above corona inception voltage. The voltage can be looked upon as made up of line components and ground components. For steep wavefronts. 1983). The latter way is cornrnonly known as overvoltage control.36) The discussion above assumed that the line is a lossless single conductor passing over a perfectly conducting ground. The techniques employed to control switching surges and temporary overvoltages are outlined briefly below. The depthof the current return pathdepends on frequency and soil resistivity. each having different velocities of pro~agationand different rates of attenuation. Following are some of the techniques currently in use to control themagnitudes of switching surges.E. With finite earth conductivity the conductor’s imagegoes wellbelow the surface and the velocity of propagation is reduced.8. ~ 1-a 1+a p.6. 1981.1981). whereas the return currentis well belowthe surface. 1987).dueto energyloss andcorona effects.3 R. skin effect in conductors and ground return paths produce some attenuation and distortion (Sendyen.u. Distortion in this case results from the different velocities of the two voltage components (Ametani and Schinzinger. 14. Nakagawa. For a perfectly conducting earth the return current flows at the ground surface at a velocity equal to the speed of light. (14. In reality. Cristina and &Amore. the electrostatic charges induced in the ground are near the surface. ! . ~ulticonductor lines further distort atraveling wave. voltage (final) = as shown in Figure 14. For ground components. 1976. The adverse effects of overvoltages on power networks can be reduced in two ways: by using protective devices-chiefly surge arresters-or by reducing their magnitudes wherever the surge originates. as discussed in Chapter 5 (Abdel-~alarnand Stanek. a traveling wave suffers from attenuationanddistortion.

it can be seen that the amplitude of the energization surge depends on the switching phase angle w T . When the resistor is shorted at the end of the preinsertion period. 14. They have the additional advantage of reducing energization surge magnitudes.3. the resulting switching overvoltage can be greatly reduced. + 14. Phase-controlled switching should be carried out successively for the three poles to accomplish a reduction in the initial voltages on all three phases. another surge willdevelop. By properly timing of the closing of the circuit breaker poles. in the simple system ofFigure 14. An optimal value of R would normally be afraction of Zo.1. It is effected by initially applying the supply voltage to the line through a resistor.3 Use of Shunt Reactors Shunt reactors are used on many high-voltage transmission lines as a means of shunt compensation to improve the performance of the line. After a suitable period of time. and depends on transmission-line length. as explained in Chapter 11.7.4).4 Drainage of Trapped Charges Charges are trapped onthe capacitance to ground of transmission lines after their sudden reenergization.1.1. This is accomplished mainly by the reduction in temporary overvoltages. the magnitude of the energization surge is usually much reduced by the effect of system damping. usually by means of automatic reclosures. which would otherwise draw large capacitive currents from the supply. control of the first surge becomes ineffective.7.The initial amplitude of the energization surge when a preinsertion resistor of value R is used will be only Zo/(R Zo) of that reached in the absence of the resistor.7.1. 14. This is extremely difficult with conventional circuit breakers but is quite possiblewith solid-state circuit breakers.2 Phase-Controlled Closure Referring to equation (14.which describes the voltage surge waveform. as will be seen in the next section. This effect is evident from equation (14. where Zo is the surge impedance of the line.4). If R is too small. normally between onethird and one-half of acycle. the second surge becomes dangerous. these charges may cause an increase in the resulting surge. if it is too large.1 Resistor Switching This is one of the most common methods for reducing energization overvoltages.nis 14. If the line is reenergized soon after. If. allowing the full supply voltage to be applied to the line. the preinserted resistor is short circuited. the capacitance C has .7. By the end of the preinsertion period.

6). the second harmonic component of temporary overvoltages can be successfully suppressed. shapes. trapped charges maybe partially drained through the switching resistors incorporated in circuit breakers. Randomness created inherently in the overvoltage-producingprocess Examples of the latter category are outlined in the following sections. the surge voltage . All types of overvoltages on a power system are subject to statistical fluctuation in their magnitudes.4 and its related overvoltage equation (S4. S 967). it is evident that by increasing the capacitive input reactance of the transmission line Xc the magnitude of the temporary overvoltage Y is reduced. .6). or even eliminated. properly designed varistor would conduct in such a way as to provide large losses at the frequency in question. Magnetic-type potential transformers also draintrapped charges via a low-frequency oscillation which is highly damped by the effect of magnetic saturation.37) In equation (14. the goal of increasing Xc is obviously achieved by means of decreasing Xr(down from its infinite value in the absence ofthe reactor). will increase the overvoltage on the line (Bickford and Doepel. A. which. The sources of randomness can generally be grouped into two main categories: 1. If a shunt reactor of reactance X r isadded to the transmission line. as mentioned earlier. Randomness produced simply bythe existence of a large variety of overvoltage-producing events 2. In practice. Furthermore. and durations.an initial voltage Vc(0) = Yo caused by trapped charges.37). thus reducing the overvoltage magnitude according to equation (14. by the use of surge arresters with nonlinear resistor (varistor) characteristics (Ragaller. if the same polarity as the surge’s will include an extra component V peak voltage. 1980). the equivalent input reactance of that line will be increased from Xc to (14. Referring to Figure 14.

9 99 Qfe 90 50 to 1 0. It has been reported that. The distribution of thunderclouds in the area.. in turn.] Probabilitydistribution of lightning peakcurrent.1 (1 975) . . wind. Since all switching surges are products of circuit breaker actions. temperature. which will.nis Many factors contribute to fluctuations in the lightning stroke current. The marked statistical fluctuation in lightning surges emphasizes the importance of taking this fluctuation into account when dimensioning the insulation and evaluating the risk of insulation failure. The following factors contribute to the randomness in switching angles: 9 9. Random angles of switching are mainly responsible for this variation. the weather conditions (e. etc. the random variation in switching surges is attributed primarily to the performance of thesedevices.g. reflect on the consequent overvoltages. [From Berger et al. in general.). Figure 14. higher currents are associated with longer fronts. the size and charge distribution within each cloud.9 depicts a cumulative frequency distribution of lightning current magnitudes. as well as to the network circumstances at the time of switching. and the distribution of projected objects in the area are among those factors. pressure.

[From Diesendorf (1974). The distributions are highly skewed with very restricted variations-almost deterministic-at lower vol9 9. 1 5. Figure 14. 1 and l’.0 S ~ I T ~ H I NSURGE G (p .10 Switching surge magnitudes. uncontrolled surges.9 99 90 80 60 50 20 10 l O*l 0.0 ure 14. Arc interruption. 3. over a long period of time. u . Circuit closuremay be prematurely effectedfollowing a breakdown between the breaker’s contacts. whosetimingisrelevant to the production of recovery transient voltages. 1 Temporary overvoltages are not subject to randomness in the same sense that switching surges are. while proper control measures can reduce this fluctuation to only about 12%. Any change in temporary overvoltage magnitude would haveto be related to changes in networkparameters.] . a statistical distribution of overvoltage magnitudes may be produced.1. The mechanicalmovement of circuit breaker contacts produces fluctuations about the aiming angle of interruption or closure.many of which are inherently random. 2. however.01 0 1. as describedin Chapter 6. If these changes are considered.10 showstypical distributions of switchingsurgemagnitudes. ~ncontrolledsurges may fluctuate in m a ~ i t u d e at a coefficient of variation (standard deviation relative to the mean) of about 25%. involves a number of physical processes. controlled surges.0 2’0 3.2.

25p. The Electromagnetic Transients Program (EMTP) is a computer program for simulating electromagnetic. . etc. branch currents or voltages. at highervoltage magnitudes (Ragaller. thereby providing multiphase load flow capability.) as functions of time. and control system transients on electric power systems (EMTP Development Coordination Group.u. The implicit trapezoidal rule of integration is used in the discretization of the equations of most elements which are modeled by ordinary differential equations. be determined automatically by the program. These equations are written in nodal-admittance form (with new unknown voltages as variables). can be manipulated internally without additional approximations or assumptions. Nonlinear reactances can either be linearized during steady state or fully modeled to include harmonic distortion effects. The measured response of a power transformer can be used to create frequency-dependent transformer models using the High Frequency Transformer model. such as the internal transformer representation used by transformer manufacturers. as well as the generation of more complex linear and nonlinear models for use in EMTR simulations. and are solvedby ordered triangular factorization. 1996).g.. machine torques or speeds. Injections of the electric network may also be specified in terms of power and voltage magnitude. algebraic equations which is solved at each time-step using advanced sparsity techniques. 1980).nis tage magnitudes within 1. simultaneous. electromechanical. Large coupled RLC networks. Initial conditions for differential equations of the various components can. The calculation of initial conditions is normally limited to linear elements. The result is a set of real. with more than a 15% coefficientof variation. Support programs provide additional capabilities such as the calculation of overhead line and cable parameters. and a fairly wide variation. Nonlinear resistances are always ignored during the steady-state solution. for most casesof practical interest. Program output-both printed and plotted-consists of component variables (e. The EMTP is used to solve the ordinary differential and/or algebraic equations associated with an “arbitrary” interconnection of different electrical (power system) and control system components.

ligh~ning surges. Pi-equivalents are used to represent multiphase. The simplest traveling-wave models can be defined by surge impedances. length of time to be simulated. Modeling of multiphase.Nonlinear elements are specified by current-and-voltage points for resistors and current-and-~u~-linkage points for inductors. and series capacitor protection. phase conductor transposition. asymmetrical fault current evaluation. resistance per unit length. coupled RLC circuits. motorstarting. grounding. and capacitance in )-1Fat power frequency. and output requests. A multi-platform graphical user environment. carrier frequency propagation. series and shunt resonance. general steadystate analysis of unbalanced systems. Series RLC branches are used to represent single-phase resistances. The input file contains the calculation time step. branch voltages. coupled RL circuits is applied for power frequency transformers. EMTP View. has recently been used whereby data is entered using a circuit schematic diagram and free-format data entry forms. ground wire losses. harmonics. . and control system dynamics. islanding or otherdisturbance events. general control systems. They include: switching surges. machine variables. A cascaded-pi input is used for untransposed transmission lines. capacitor bank switching. branch currents. static VAR compensation. wave velocity.and line length for positive and zero sequence. The lumped branches are defined by resistance in Q. as well as the model data. insulation coordination. high-voltage DC (HVDC). branch energy dissipation. out-of-phase synchronization. and control system variables. Other representations are dynamic synchronous machines (3-phase balanced design only). inductance in m H or in Q at power frequency. shaft torsional stress. Frequency-dependent representations of multiphase distributedparameter transmission lines are used in which propagation time of the line is considered. and capacitances on the basis of their respective voltage-current time relationships. ferroresonance. inductances.Numerous studies can be carried out using the EMTP. unconventional rotating electromechanical energy converters. The primary output from a transient simulation takes the form of plotted bus voltages.

inductance. [c] is split in two.2. and capacitance matrices.2. Representation of a pi-section. 14. Associated with thesebranches are matrices [L] and [R] having performance equations Matrices [R] and [L] are assumed to be symmetric. 14. .12) This is to represent lumped-element. 14.9.1 1) This class of branches provides for the representation of lumped-element resistance. 14.1 The Pi-Circuit Branch(Fig.2 Mutually Coupled RL Branches (Fig.9. For N conductors. the associated differential equations are All matrices are symmetric.nis Two examples are shown in which the representation model is associated with the governing differential equations. with half of the total oneach end of the branch. mutually coupled RL branches.

The number of transients required to fail the cable increases as the applied surge magnitude is decreased. 14. multiphase transformer cores. It was stated that the most accurate assessment of underground distribution systemswitching transients could be made by making field . The finite sections approach. parallel pipes. The material given in the following sections has been reported as various extensions of the EMTP program.3.9. submarine and pipetype cables.. The approach was termed the method of finite sections to reflect its intermediate position between pi-section modeling and finite-element analysis. which amounts to a simplification of a finite-element approach where symmetry isobvious.E & k l End " " " " m 1 Vk Vm 2 I 2 Representation of coupled RL elements. is capable of accurately representing transient electro~agnetic wave propagation in conductive materials. and transformer tank walls. It was applied to coaxial cables. Another study showed thatthe soliddielectric insulation used in underground distribution cables is subject to failure under the cumulative effectofswitching voltage transients which therefore~ontrary to standards-should be considered (Walling et al.1 Cable witching Transients Most approaches using EMTP start with steady-state impedance scans of the line or cable of interest and ignore wave reflections at the centers of conductors and at the sheath-to-earth transition. The steady-state starting points forthe existing methods were said to fail to recognize the existence of transient conductor surface impedances. It was concluded that most of the complexity and uncertainties of present frequency-dependentmodeling techniques would be eliminated simply by recognizing and modeling wave propagation in the conductorsas well as along the dielectrics (Meredith. 1995). as seen in Figure 14.13. 199'7).

nis Multi-windin Ren Ren .

They are mainly produced in gas-insulated substations (CIS). As it does so. The rise time ofthe voltage collapse during one strike is given by (Povh et al. 1996) 0 2 100 Figure 14. 14.measurements in actual systems.. from 100kHz up to 50 MHz. Theworkshowed that restriking does not lead to voltage escalation whenload-breakelbows and disconnects are operated because the resulting arc does not extinguish at transient current zeroes. Figure 14. during disconnect switch operation. can predict the propagation of switching surges in the cable system with reasonable accuracy.14 Cable end voltage measured during elbow closing energizing 925 ft cable section.14 was an example. due to the slow operating. a number of prestrikes or restrikes occur due to the relatively slow speedof the moving contact. This first strike will almost inevitably occur at the crest of the power frequency voltage. the complex behavior of air-break switching devices cannot be predicted by electrical circuit simulation. This traps a voltage on the switched cable which is essentially the same as the instantaneous source voltage. i..9. the potential difference across the contacts falls and the spark will eventually extinguish. Fasttransients can also accompany switching of vacuumbreakers. The current will flow through the spark and charge the capacitive load to the source voltage. Although numerical simulation techniques such as the EMTP.e. a . the electric field between them will rise until sparking occurs.2 Assessment of Very Fast Transients(VFT) Very fast transients (VFTs) belong to the highest frequency range.3. but continues until the transient oscillations are damped. As the contacts close down. In CIS.

Typically.3. skin effects can be neglected due to the aluminum enclosure.0 p.U.Normally the highest overvoltage stress is reachedat the open end of the load side. For thesecases the resulting overvoltages are in the range of 1. resulting in a most unfavorable voltage 1. the basic frequencycomponent of the VFT in the M (b). that these values have been gained by calculations using unrealistic simpli~ed simulation models.7P.9.15 where one prestrike of a disconnector switching is oscillographed showing. measured in an actual CIS. have been reported.u. is given in Figure 14. for the load and the supply side.7 ns.5 P.3 ~ o n u n i f o r ~ T r a n s ~ i sLine sion Modeling It is sometimes required to perform transient analysis of power network elements representing nonuniform transmission lines. and the overall transients and the steady-state condition (c.u.5 P. 14.2. the steep voltage transients (a). the rise time isabout 7.U. caused by the oscillation of the whole system consisting of the CIS and the transformer. An exampleis the . In some cases extreme high values of more than 3 p.nis where ( ~ / P is ) the ~dielectric strength of SF6 which is equal to 860 kV/(cm.d) al.U. the trapped charge remaining on the load side of the disconnector must be taken into consideration. and reach 2. In some cases of a highspeed disconnector the maximum trapped charge could be 1. An example of these transient phenomena. It can be shown. traveling waves will be produced. For the calculation of the VFT stress. and h is a field utilization factor. P is the pressure in MFa.5P. and the highest overvoltages can reach values up to 2.u. 1996). for very specific cases. Fa). As these timesare smaller than the transit time of the CIS components.5 MPa and h = 0... The main portion of the damping of the VFT occurs by outcoupling at the transition to the overhead line. overall transients with frequencies in the range of 20-100 kHz can be observed. In the case of power transformers feeding the CIS.0 p. The internal damping of the VFT influencing the highest frequency components is determined by the spark resistance.u. r P = 0. For a normal disconnector with a slow speed the ma~imumtrapped charge collapse of reaches 0. The main frequencies ofthe VFTs depend on the length of the CIS section affected by the disconnector switch operation and are in the range of one MHz up to 4OMHz for the basic component and even higher for the superimposed frequencies. The VFTs generated depend on the CIS configuration and wave multiple reflections. however.

15 Transients on the source and load of side a GIS due to disconnector (b) basic frequency componentof the VFT switching: (a) steep voltage transient. A nonuniform transmission line model suitable to interface with the E l ~ t r o m a g n e tTransients i~ Program ( E ~ T P was ) proposed by Correia de Barros et al. in the MHz range. lightning surge response of transmission line towers where the surge impedance of a transmission tower variesas thelightning surge travels along it. (1996).Source sidg side Load t -f 0 t 0 Figure 14. (c) overall transients in the KHz range: (d) low frequency transients and steady-state condition. The model was based on a ~nite-differe~ces algorithm .

The input filter. A method to test the transient performance of a sampled-datarelaywas reported by Wilson et al. has been proposed as an effective method of significantly increasing the transmission capacity of the power corridor up to 200% of the total power-carrying capacity of the corridor. Computer modeling ofthe digital relaywas done within the Modelsversionof the Transient Analysisof Control System (TACS) subsection of the EMTP Program.3.9. Digital or sampled-data relays operate in a similar manner to the PC.9. The capacitance coefficients of multiphase transmission lines exhibiting corona were carefully identified. 14. 1995)..3. and the relaymeasuring principle of one digital microprocessor-basedrelay were modeled.9.5 Transients ina Hybrid AC/DC Transmission System Adding a DC line to an existing AC transmission circuit. A model of one measuring unit of a digital relay wascreated on a personal computer. The capacitance coefficientswere expressed in terms of space-charge influence factors that took into account the space-charge distribution around each conductor. which was determined by the local electric field distribution. . or converting one circuit of a double circuit AC transmission system to DC.3.6 Corona Effects on Transient Conditions Corona is the main factor affecting the attenuation and distortion of travelling surges on transmission lines by modifying the value of the transversal line parameters. The program had the capability of modeling transmissionlines for coupling effects as well as the capability to represent HVDC converters in full detail and to represent transformer saturation nonlinearities.evenwhen representing them as frequency-constant parameters. (1993). 14. (1995). Electromechanical relays are less sensitive to power system transient effects. 14. Surge propagation on multiphase transmission lines exhibiting corona was studied by Correia de Barros (1995) and Barros et al. analog-to-digital converter. Fourier fundamental frequency detector.4 EMTP Modeling of Relay Performance The modeling and testing of relays for transient effects was not a critical issue until the introduction of electronic and microprocessor-based relays. An electromagnetic transient simulation program (EMTDC) was used to investigate the overvoltages on a hybrid AC-DC transmission (Verdolin et al.nis to solve the wave equations for any space variation of the line parameters. The model included the transmission line losses.

6Ng x 10-1 where h is the mean height of the line.3. 1993). and Ng is the density of lightning strokes per km2 per year. and finallybyreverse transformation from the frequency to the time domain.. For certain system conditions. transforming the incident wave from the time to the frequency domain. (1997). 1992) by .3. if unknown. according to the electrogeornetrical model. Power transformers as terminations of transmission lineswith traveling waves were studied by Hasman (1997). Ngmay be given by Ng = 0. The algorithm was independent of the shape of the voltage signal to be synchronized on by means of pattern recognition.9. calculating the reflection and transmission coefficients of the transformer. modification of every harmonic of the incident wave by the reflection and transmission coefficients.. is n = 28 ho. pre-insertion inductors may be ineffectivefor mitigating remote overvoltages due to capacitor energization (Blnargava et al. The expected number of lightning strokes n (per 100 km) to an overhead line. The model was based on meas~ring the input and transmission impedance of the transformer in a suitable frequency range. 14. The density of lightning strokes to ground was related to the record of “lightning flash counters” (LFC) Nc (discharges per year) (Popolansky et al.9.8 Reflections on Transformers The peak values of transient overvoltages may becomeparticularly dangerous to insulation owing to the processofreflection and transmission of waves at discontinuities of surge impedance of transmission lines.04 Ti’25 where T b is the isokeraunic level in thunderstorms-days per year.77 14.7 ClosingTransients Control algorithms for point-on-wave closing or auto-reclosing of shunt reactor-compensated transmission lines for reduction ofclosingovervoltages were reported by Froehlich et al.

To predict the lightning surge overvoltage that occurs in an electric power system using the EMTP. (1995). L = 1. The lightning surge response characteristics of a UHV transmission tower were measured by Yamada et al. If the line is rated for 220 kV and has the parameters R = 0. find the surge impedance and the velocity of propagation. Calculate the transmitted . find the voltage building up at the junction. what is the time taken for the surge to travel to the other end of the line? (2) A transmission line of surgeimpedance 500 S2 is connected to a cable of surge impedance 60 S2 at the other end. transmission towers need to be simulated on the basis of their surge response characteristics. IF a surge of 150 kV and an infinitely long tail strikes at one end of the line.189 mH/km and of capacitance 0.009 pF/km. the traveling wave propagation velocity. the surge impedance. (1) A 3-phase single circuit transmission line is 400 km long.nis iatio The measurements and analysis of electromagnetic field radiation from a lightning discharge are important. The response waveform corresponding to a 1/70ps ramp-wave current was subsequently evaluated using Laplace transforms.009 pF/km. The model realistically assumed that the lightning current in the case of tall structures is initiated near the top of the structure. If a surge of 500 kV travels along the line to the junction point. A velocity of propagation of 300m/ps and a tower surge impedance of 120 S2 were reported for the multistory tower. and G = 0. The proposed model consisted of the lightning current propagation anddistribution model and the electromagnetic field radiation model. neglecting the resistance ofthe line. (1996).1 S2/km.26 mH/ km. namely. and the attenuation and defomation of the wave. (3) An underground cable of inductance 0. The relationship between electromagnetic field radiation and return stroke current was studied and a calculation model of electromagnetic field radiation proposed by Motoyama et al. C = 0.since the electromagneticfields depend on the propagation and distribution of the return stroke current along the lightning path. It also accounted for the propagation and distribution of the lightning current inside the lightning discharge path and the tall structure on which electromagnetic fields depend. The surge impedance for a UHV tower was appro~imately130 $2.26 mH/km and capacitance 0.3 pF/km is connected to an overhead line having an inductance of 1.

OmH/km and a capacitance of 0. C = 0. andG = 0. find the rise in voltage at the other end if the line length is 400 km. London: Butter~orth. 1963. Schinzinger R. If the line isa 3-phase line and is charged from one end at a line voltage of230 kV. IEEE Trans IA-23:481-489. Azevedo RM. 1995. Bickford JP. Froehlich K.IEEE Trans PAS-102:1685-1693. 1976. 1987.47 and reflected voltage and current waves at the junction if a surge of 200 kV travels to the junction. along the cable.01 pF/km is connected at “j” to a 100 m long underground cable whose total inductance and capacitance are 0. The time (in p )which the surge takes to reach the substation. Proc IEE 114:465-477. Kroninger H. Electromagnetic Transients Program (EMTP96). Traveling Waves on Transmission Systems. EMTP Development Coordination Croup. . The transmitted and reflected voltages at j 2. Barros H. 1975.03 pF. IEEE Trans PowerDeliv10:1443-1449. Anderson RB. and feeds a substation. Bhargava B. and along the overhead line. IEEE Trans Power Deliv 8:1226-1232. Avent BL. 1993. (5) An overhead powerlinewith an inductanace of l . Correia de Barros M. Alrneida ME. IEEE Trans Power Deliv 10: 1642-1648. sub-station Abdel-Salam M. (4) A transmission line has the followingline constants: R = 0. Berger K. 1997. 1996. Bewley LV.02. Imece A. 1996.m H and 0. 1983. respectively.009 pF/km. Palo Alto. 1995. Peelo DF. Carneiro Jr S.1974. Hoelzel C. IEEE Trans Power Deliv 11:1082-1087. CA: Electric Power Research Institute (EPRI). New York: Dover Publications. Dipietro J. Cristina S. A 500kV surgewas initiated at a distance of one km from the junction j. Estimate: 1. IEEE Trans PAS-95:773-781. Stanek M. 1967. Diesendorf W. H o ~ a u eW. Doepel PS. r Hoegg P. Sawada JH. Khan A. Arnetani A. Electra 41:23-37. The transmitted and reflected currents at j 3.1 St/km. d ’ h o r e M. IEEE Trans Power Deliv 12:734-740. Stanek E. Correia de Barros M. Insulation Coordination in High Voltage Electric Power Systems. Carvalho AC.

. Marshall JL. Nakagawa M. 1997.2. IEEE Trans Power Deliv 10:1514-1522. Yamada T. Walling RA. IEEE Trans PAS-100:848--856. McDermott BA. IEEE Trans Power Deliv 11:1624-1632. 1993. IEEE Trans Power Deliv 12489-496. IEEE Trans Power Deliv 11:2028-2035. Ishii M. IEEE Trans PAS-100:3626-3633. Rusan RM. ed. 1973. Verdolin R.New York: Dover Publications. Bisewski B. Motoyama H. HusseinAM. 1965. Kuffel E. CIGRE report 23/33-04. SpLicek Z. IEC. Sawada J. 1992. Volcker 0. Publication 60:1. IEEE Trans Power Deliv 12:1684-1689. Dvofak N. Melchior RD. IEEE Trans Power Deliv 1:393-402. 1996. 1995. Ragaller K. Geneva: International Electrotechnical Commission. 1980. IEEE Trans Power Deliv 8:984"989. New York: Plenum Press. 1981. Lewis ~. Chang JS. Witzxnann R. High Voltage Test Techniques. 1995. Imece AF Iravani R. Wilson RE. 1997. Popolansky F. Sarshar A. Martinez JA. 3. Nordstrom JM. Meredith RJ. Keri A. Mochizuki A. Povh D. Amettani A. Lightning Protection. Diseko N. IEEE Trans Power Deliv 10534539. Semlyen A. Gole AM. Schmitt H. New York: Wiley-Interscience. Chewdhuri P. Janischewskyim W. 1996. 1973. Surges in High-Voltage Networks.nis Hasman T. The Protection of Transmission Systems against Lightning. Kato S. Chisholm WA. 1981. Zaima E.1995.

’’ Two stages of study must precedea successful insulation coordination. In the light of the outcome of those studies. namely. so as to reduce to aneconomically and operationally acceptable level the probability that the resulting voltage stresseswill cause damage to equipment insulation or affect the continuity of service. (1976). and location of surge arresters and other possible protective spark gaps. “insulation coordinationcomprises the selection of the electric strength of equipment and its application in relation to the voltages which can appear on the system for which the equipment is intended. and taking into account the characteristics of available protective devices. a subject dealt with in Chapter 14. number. ’ l .Egypt According to the International Electrotechnical Commission. The leakage distance of external insulators 4.Giza. and the general assessment ofinsulation performance under the stress of those overvoltages. The phase-to-ground and phase-to-phase clearances 3. type. The ratings. The electrical strength of the insulation of all system components 2. the insulation coordination effort aims at specifying the following: 1.Insulation o o ~ d i n ~ t i o n CairoUniversity. an examination of the different types of overvoltage that thepower system encounters.

asin cables.1. switching overvoltages. compressed-gas insulation. Insulating oils. and dielectric performance. according to location. lends itself to probabilistic treatment where the evaluation of the insulation's risk of failure is economically justified. They are. ~e~"restoring insulation. The last three classes of overvoltage are discussed in Chapter 14. . 2. temporary overvoltages. The coordinationof self-restoringinsulation. Some insulating films in capacitors produce nopermanent leakage path aftertheir breakdown and are therefore called "self-healing" insulation. transformers. and solid insulation of cables. insulation may be either self-restoring or on-self-restoring. and machines. 1 . ~nsulationis classified. andelectric machines belongto this category. liquid.The relative durations of these voltages are shown in Figure 15. Power system insulation is subjected to four classesof dielectric stresses: power-freq~ency normalvoltage. belongsto this type of insulation. it is usefulto classify the insulation of a power network according to location. External insuzation: clearances and insulator surfaces in the open air. transformers. Here the response of system insulation to these stresses and the likelihood of failure are outlined. unlike non-self-restoring insulation.Anis When applying insulation coordination. 1. From the dieelectric performance viewpoint. Solid insulation. External insulation can be located outdoors or indoors. as external or internal. or gaseous components of equipment insulation and therfore not exposed to atmospheric conditions. Inte~nal insuzation: internal solid. influenced by such atmospheric conditions as pollution and humidity. 2. and lightning overvoltages. ~on-se~-restoring insulation^ loses its insulating property after a disruptive discharge and must therefore be replaced. therefore. regains dielectric integrit~following the occurrence of a breakdown. In the latter case it isless affected by the ambient weather conditions.

Atmospheric pollution is the primary cause of breakdown of external insulation under normal voltages (Khalifa et al.Asdiscussedin Section 9. and a sudden breakdown may then occur under operating voltage. Thermal breakdown may develop under the continuous stress and the consequent dielectric losses. b on tam in at ion on the insulation surface can initiate a mechanism that may lead to total breakdown. 3. 4.l Superposition of V 4 characteristics on overvoltage distri bution pat- ter. leakage currents on polluted insulators. These voltages-from an insulation flashover point of view-are character- . if heavy enough.9.. Power-line and stationinsulators are subject to a variet of transient voltages. 2. Normal aging of the insulation may be manifested by the gradual reduction in its withstand capability over a long period of time.W # c3 . Following are some of frequencyvoltages: the causes of insulation failure under power- 1. 1988). may cause eventual flashover at service voltage. Voids and flaws in insulation may develop corona discharges which would leadto prematur~ aging of insulators and their eventual failure. The flashover voltage of a polluted insulator decreases considerably as the conductivity of the pollution layer increases.

2 showsthe pronounced influence of ground proximity and that of voltage polarity on the breakdown voltage of an air gap (Electrical PowerResearch Institute. The shape of an idealized impulse is made up of its time to crest (or front) and the rate of its decay beyond the crest.while others.3.The IEC standards choose 250 ps to represent the front of a switching surge where the air insulation displays the least strength (IEC.2. rod-rod gaps have a gap factor of about 1.7. 15. the term “gap factor” (K’) has been introduced to express the breakdown voltage of a given gap relative to a positive rodto-plane gap of the same length.it takes values between 100and 500 p .2.8.1 Front Duration Air gaps and insulators shorter than about 1m have a flashover voltagethat is independent of the impulse front. 1979). as explained in Chapter4 in detail. and are thus relevant to insulation dimensioning. The gap factor is therefore always larger than. 15. depending on the strike distance and configuration. the larger will be the difference between positive and negative breakdown voltages. For example. unity. The breakdown voltage of a uniform field gap can be as high as six times that of a rod-plane gap. Some impulse application may produce flashover. which is customarily identified by the impulse tail.Positive breakdown voltages (Le.2. 15. 1976)..2. and. for practical purposes.The impulse front corresponding to least dielectric strength is commonly known as the critical impulse front. or equal to. stillwith the same amplitude. The more nonuniform the field in the gap becomes.3. may be withstood by the insulation. are discussed. the outcome of all applications may not be the same.3 Breakdown Probability Distribution If an impulse voltage of a certain shape is applied repeatedly to self-restoring insulation. In this section the various factors that influence the flashover voltage of external insulation.is . as discussed in Section 4.2.2 Gap Geometry and Polarity The physical process governingbreakdown is greatly influenced by the field distribution in the gap (Chapter 4).2. Lightning impulses are known to have fronts on the order of a few microseconds.484 Anis ized primarily by their magnitudes and wave shape. those when the more highly curved conductor is positive) are lower than negative voltages. Therefore.. Figure 15. whereas switching surges have fronts onthe order of a few hundred microseconds (Chapter 14). a conductor-to-tower clearance normally has Kg = 1. This phenomenon. For longer air gaps and insulators the impulse front has a definite effect on the flashover voltage. which is known to have the least breakdown voltage.

at very high transmission voltages this property is used to set an economical compromise between the insulation cost and that of the risk of insulation failure. This probabilistic property is very essential to the process of insulation dimensioning. above which all shots produce breakdowns and below which no shots would cause breakdown. Also. The probability of breakdown (i. In fact.. particularly true for sufficiently large gaps (larger than about a 50cm rodplane gap in air at atmospheric pressure). the number of breakdowns relative to the total number of applied shots) may be plotted against the applied .Insulation ~oordination 485 Figure 15.e.2 Effect of polarity and ground proximity on air gap flashover. it occurs over a finite range of impulse voltage amplitudes.

(b) on a Gaussian scale (a=standard deviation).3a. Experimental investigations have revealed that formost external insulation the breakdown probability function. .1) 0 0 IO I (b) . may be expressed by the well-known cumulative Gaussian distribution function (15.4 nis voltage as shown in Figure 15.3 Breakdown probability of external insulation: (a) on linear coordinates.

particularly when an overvoltage protection device is installed.3b. CFO c7 = standard deviation of the representative Gaussian distribution. CT describes the degree of scatter in breakdown voltages about their mean value V50 If the postulate of equation (15.2. two physically oriented as assumptions are adopted: first. if any. depends on the gap configuration. the amplitude. second.e. are those of the quantity (V 15. a breakdown.7 This function is characterized by two parameters: Vs0 = 50% breakdown voltage. only one gap at a time is liable to breakdown. . referred to as the time to breakdown. The time to breakdown of insulation is essentialfor insulation coordination when twoor more insulation components are subjected simultaneously to the same impulse stress. The linear units on the ordinate in this case Y50)/0 rather tha P(V).2. subjectedto the same normal and abnormal voltages).. Furthermore. depending on the station voltage and capacity. the time to breakdown is a random quantity that may vary markedly from one impulse application to another (Chapter 4). In a substation. also called the critical flashover voltage. In the study of parallel insulation. that whenever more than one gap tends to break down simultaneously. P(V ) can alternatively be plotted on a normal probability ordinate scale as seen in Figure 15.4 Time to Breakdown Upon the application of an impulse voltage. This delay. L A transmission system comprises a large number of insulation components “in parallel’’ (i. only the one with the shortest “time to breakdown” would actually do so. the number of insulation strings connecting the same phase to ground can be as large as a few thousands. that upon the incidence of an overvoltage. and the shape of the voltage impulse. l e t h a t the breakdown probability fits a Gaussian cumulative function-is accepted. in which case P(V ) appears linear. the number of line-to-ground insulation structures may be as high as 100. On transmission lines. would take place after a certain time lag following the initiation of the impulse. The number of bus insulators per phase could vary from 100 to 1000.

nis When the probabilistic properties of the two gaps are given separately.is a function of the probability density distributions f i ( t ) and f2(t) of the times to breakdown of the two gaps.2) The first tern of equation (15.2 + P2.. the new breakdown probability of the first gap during the parallel combination would be P. the probability of withstand for rz identical gaps in parallel is then icy..3) Since the time to breakdown of one gap can only beeither shorter or longer than that of the other gap.l (15.l = 1 (l 5.multiplied by the probability P1.respectively. similarly. the quantity hence P2. the behavior of a Combination ofthe two gaps may be predictedunder a known stress. p. Such a function can be derived to be (15. If the probability of withstand at a given voltage levelfor a single air gap is W. = P2(1 PI) + PlP2P2. and S2. P1.2 (15. then (15. This is complemented by the probability that the two gaps break down simultaneously (PIP2).5) In the special case whenthe two density distributions are normal (Gaussian) with mean valuesTI and T2 and standard deviations S. then . The breakdown probability of the second gap is.2that in this case gap 1 is faster than gap 2 to develop a breakdown. = PIU P2) + PIP2P1. +P! = P1 + P2 P2P2 (15.8) . If the individual breakdown probabilities of the two gaps before the combination were PI and P2.7) which is characteristically independent of the times to breakdown.4) Obviously.2) is the probability that gap 1 flashes over whereas gap 2 does not.6) The overall breakdown probability of the system of two gaps is given by Pc = P. = J v n (l 5..

8) and (15. the attenuation of the traveling overvoltage wave should first be considered in evaluating the line's flashover probability. a system of 20 parallel insulators will have a withstand probability of only 82%. Testing the system of parallel gaps for the determination of its TBD distribution can be impractical in view of the varying number of components of the system. To illustrate this relation. The TBD of a system ofparallel gaps can. The time to breakdown (TBD) of a system of parallel gaps (or insulators) is also a random quantity (Chapter 4). can no longer be Gaussian. Experimental investigations have shown that the relationships given previously hold accurately.or P" = 1 (1 P)" 5.fi(t) andf2(t)under a given impulse stress. be related to the characteristics of the individual components which make up that system.9) (1 where P is the breakdown probability of one gap and P. If the string of insulators in successive towers are roughly considered to be in parallel.10) . The knowledge of the distribution of this time is vital to insulation coordination and theselection of a protective device. each being weighted by a corresponding probability of occurrence of a breakdown. Given two parallel gaps with breakdown probabilities PI and P2 and the TBD distributions . For example.9) that theprobability of withstand of a transmission system decreases as the number of insulators in parallel increases. is that of n gaps. if the distribution of the breakdown probability of a single gap P is accepted to be cumulative Gaussian. with a probability of withstand of a single insulator of 99%. It is concluded fromequations (15. The probability density distribution of the collective TBD is composed of the TBD probability densities of the individual components. alternatively. the TBD distribution of the combination is (15. the caseoftwogenerally unidentical gapsis discussed. the distribution of P. Statistically.

7). etc. twos.12) where F.14) It is worth mentioning that whenever the individual breakdown probabilities are very small. different combinations of gaps will have to be considered (e. Considering the binomial combinations of breakdowns in the n gaps. and F 2 are the cumulative distributions of f i and h.fi2(t) is the density distribution of the minimum TBD of the two gaps since only the gap with the shorter TBD will break down at a time. It can beshown that fiZ(t) is related to the individ~al TBD values by (15.). in view of equation (15. 4 = Pl(1 P2lfi(t>+ p 2 0 PlYm + PlPdI2(0 fc(t) dt = 1 and the denominator is introduced to fulfill the constraint Applying the same constraint-that a density distribution integrated over all times equals unity-to the constituents of the denominator gives where Pc = P 1 P2PIP 2 is the breakdown probability of the combination.nis where. respectively. threes. which represents the contribution of a simultaneous breakdown in the two gaps. To extend the analysis above to a large number of equal gaps n.g.singles.13) The overall TBD of the n-gap system can now be deduced in a manner similar to that of two gaps [equation (15. In the third tern. the overall distribution of the TBD of a system of n equal gaps will be (15. as the number of gaps in parallel is increased and/or the individual breakdown . all combinational contributions tofc(t) can be neglected and fc(t) becomes equal tof(t) of any one gap. On the other hand. as was shown in equation (15.(t) are the contributions of the individual TBDs of the two gaps whenever one gap breaks down alone. The first two terms off. The distribution of the minimum TBD of only r gaps (out of a total of n gaps) can be shown to be given by + = [l w)l'-lm ~')(t) (15.2).1l)]..

1974). P(Vm. overvoltages whose magnitudes fall within an interval dVm. Depending on whether the coordinated insulation is self-restoring or nonself-restoring. i. In the statistical approach a calculated risk of insulation failure is taken. which can be shown to be The procedure leading to well-coordinated system insulation begins by evaluating the stresses to which insulation is subjected. exhibited a distinct random nature wherein they varied markedly in magnitude and form. a statistical or conventional approach is adopted. t). overvoltages.e.16) t)dVmdt] t)V(Vm? ==C P(Vm? The overall risk of insulation failure can thus be written as (Anis et al.17) . The insulation dimensions are then decided in viewof existing protective devices such that a predetermined level of safety against insulation failure is ensured... as explained earlier. should produce a risk of insulation failure given by dR (15. whereas in the conventional approach the coordinator is set to elimiaate the possibility of system failure (Diesendorf. t ~ t i s tlns~~ation i ~ ~ ~ Coor With the advent of extra-high-voltage transmission and in anticipation of ultra-high voltage.. 1978) (15. At the same time. Second. Therefore. two phenomena came into existence and justified the use of a statistical approach to insulation coordination.e. and with times to crest falling within dt. long insulators and air clearances displayed widely scattered probabilities of breakdown and associated times to breakdown. First. the breakdown probability of insulation may also be expressed as a function of the same two quantities. f(Vm9 t)]. an ultimate distribution of the system’s T approached. The overvoltages that stress the system insulation may be expressedby a joint statistical distribution of the magnitude Vm and time to create t [i. particularly those of internal origin.probability increases.

49 Anis Due to the complexity ofequation (1 5. however.17)and the difficulty in acquiring the necessary bivariate data. that is (15.18).The following quantities are defined for this purpose (IEC. it is customary to consider only the overvoltage magnitude when calculating the risk of failure. .4 gives a graphical interpretation of equation (15.18) in which P(Vm)is the breakdown probability distribution. It is therefore necessary to define a reasonable safety factor and design the insulation accordingly.4 Computation of the risk of insulation failure. a probability of withstand equal to 90%. This attitude. Figure 15. under specified conditions. Insulation can theoretically be made sizable enough to eliminate any risk of insulation failure. VOLTAGE ( V ) we 15. as discussed earlier in this chapter. This factor sets a relation between the insulation’s withstand threshold as obtained from P( Vm)and a representative overvoltage given byf( Vm). is economically unwise since the excessive cost of insulation may soon exceed the cost of system interruption due to insulation failure. 1976): Statistical withstandvoltage (SWV): peak valueof a switching (or lightning) impulse test voltage at which the insulation exhibits.

taking into account the statistical distributions of withstand voltages and overvoltages.Statistical overvoltage (SOV) : switching (lightning) overvoltage applied to the equipment as a result of an event of one specific type on the system. I O -l -2 1 0 -4 I O -5 I O Figure 15.19) In Figure 15. of the appropriate statistical switching (lightning) impulse withstand voltage and the statistical overvoltage. 1976). It appears that the relationship above is almost invariant for all EHV transmission networks-a fact that permits the use of SSF as a reliable design factor. the peak value of which has a probability of being exceeded of 2%.5 Effect of the statistical safety factor on the of risk insulation failure. for a given type of event. Statistical safetyfactor (SSF): ratio. (15. . established on the basis of a given risk offailure.5 the effect of the statistical safety factor on the consequent risk of insulation failure is demonstrated (IEC.

the most common temporary overvoltage (i. . usuallya surge arrester.U. Figure 15.u. SOV (from system studies) = 1255kV (peak) = 2.u. With non-self-restoring insulation the risk of insulation failure should at all times be avoided. The conventional .e. whereas the latter refers to a 250/2500 ps impulse.24 7X 7X R 1300 1425 1550 2.. The former expresses the insulation with st an^ under a standard impulse of 12/50 PS.00 kV. SSF 1.1 Example The following procedural example for a disconnecting switch illustrates the application of statistical insulation coordination: ~ i g h e svoltage t for equipment (line voltage) = 765 kV (ms) Phase-to-earth voltage Vph = 442 kV (rms) = 625 kV (peak) = l. ~aximum acceptable risk of insulation failure to ground R = lom4 Choice of insulation’s SWV: S ~ V kV D. in the vicinity ofthe equi~ment to be protected. This is normally achieved by placing an overvoltage protective device.08 I X IO-^ > 10-~ > 10-~ CQncZ~~iQn.1.The associated rated lightning withstand voltage is 24.0p.insulation coordination approach seeks the impulse voltage level at which the equipment insulation will not show any disruptive discharge.0p..6 demonstrates the steps leading to the determination of the insulation’s basic impulseinsulation level ( IL) and basic switching impulse insulation level (BSL).04 1.14 1.u. A rated switching withstand voltage for equipment insulation of 1550kV is selected. This procedure is done in view of the characteristics of existing surge arresters. Ifthe peak value ofthe conductor-to-earth voltage in a three-phase system is defined as 1 P.4.n i a 15.

u.4 for an earthed system.6. Thesymbols are explained in text. which isalso referred to as the basic impulse insulation level (BIL).8 p.the voltage of a healthy phase under single line-to-ground fault conditions) is equal to Ce. whereas the switching impulse insulation level (BSL) is about 2. the . For lightning impulses. This “earth fault factor” is approximately l . Maximum residual voltage at specified discharge current 3. for switching impulses it is taken directly as the maximum in Figure 15. Maximum sparkover voltage 2. protection level is taken as the maximum of the following three values: 1. Vs. the sparkover voltage. The protection level of a protective device is defined as the highest permissible surge voltage that may appear at the terminals of the equipment to be protected. is approximately 3.u. ~ a x i m u m sparkover voltage in the front. for a transmission system with a rated voltage Urngreater than 200 kV.4p. The arrester will limit an overvoltage wave to a certain protection level Vpunder normal operating and temporary overvoltage conditions.l5 4 I 0 Determination of BIL and BSL. divided by l . The surge withstand voltage of the system. however.

u.u. on the other.2.15 Ci 1. take Ce = 1. in accordance with IEC recommendations.u...u.nis The residual voltage Vr is built up in the arrester after sparkover by subsequent flow of current.1 Example The following procedural example(for a transformer) demonstrates the application of the conventional approach to insulation coordination. . Computed (or measured) temporary overvoltage to ground kV Vt = 605 (ms) = 855 kV (peak) Ce = 1.25 Minimum C Characteristics of surge arrester Rated voltage (available rating immediately above the temporary overvoltage level of 605 kV) = 612 kV (rms) impulse sparkover voltage Maximum switching = 1230 kV (peak) Maximum lightning impulse sparkover voltage = 1400 kV (peak) Maximum residual voltage at specified discharge current = 1400 kV (peak) Maximum sparkover voltage in the front = 1660 kV (peak) Protection level VD: to switching impulse = 1230 kV (i. it can be larger or smaller than Vs. Highestvoltage for equipment (linevoltage) Phase-to-earth voltage V.25 15.15 i = 1.4p.e.op.) Minimum CS= 1.4. I = 1. protection factor) = 2.u.97p. these margins must be at least equal to C.depending on that current. protection factor) = 1. (If temporary overvoltages are not known. to lightning impulse = 1443 k V (i.k = 765 kV ( m s ) = (rms) 442 kV = 625 kV (peak) = l.3p.e. on the one hand and the arrester protection level V.37 p. Safety margins ( C iand CS)exist between the insulation levels BIL and BSI.

one being connected to the high voltage.28 p. and one that is still used to some extent. The main drawbacks are the time lag that occurs before the gap sparks over. A s a rule it consists of two metal electrodes pointing toward one another. 1800 kV The duty of overvoltage protection is to ensure that transients arriving at an insta1lation from the power lines or originating in the installation itself are reduced to a level bearing a definite relationship to the rated voltage of the network.7 ~inimum switching impulse voltage (CsVp) Rated switching impulse voltage BSL (nearest higher standard) ~inimum lightning impulse voltage (CiVp) Rated lightning impulse voltage BIL (nearest higher standard) = 1415kV = 1425 kV = 2. thereby affording high operational safety. is the sparkgap.g.u. the propagation of undesired voltages in the network can be prevented. Its follow current. Frequently.u. = 1800 kV = = 2. which have short durations . A major requirement of surge arresters is that they must be able to discharge high energy without changes in their protective levels or damage to themselves or adjacent equipment. tie The surge arrester does not exhibit the disadvantages described previously. This property is decided bythe arrester’s thermal capacity. Over the widerangeof overvoltages met in practice. resulting from the service voltage. Unlike lightning strokes. the spark gap is not really a very efficient means of protection against overvoltages. the other to ground. arcing horns). and reliability. the breakdown voltage depends on the gap width and weather conditions (IEC. One of the oldest known means of providing protection against overvoltages. the variation of the sparkover voltage with the polarity and surroundingconditions. the surge arrester possesses an almost unchanging sparkover voltage characteristic. it continues even after the overvoltage has disappeared. Evidently. causing a line-to-ground short circuit on the network.88p. By allowing the gap to flash over. the spark gap is referred to by a descriptive name (e. and the fact that once an arc has started to burn. is interrupted automatically. 1973). consistency. However..

can be used to make resistors with a much higher degree of nonlinearity (I = cVa. bismuth. which are illustrated in Figure 15. The most important property of these resistors is their currentvoltage characteristics. nonlinear resistance elements R. where a >20) over a large current range. which changes only slowly. a situation that simple arresters with “unassisted” spark gaps cannot handle. The main components (i. creates a strong magnetic field which extends throughout the stack of spark gaps. only heavy-duty surge arresters with magnetically blown spark gaps are recommended to operate on switching surges. The electrical properties of the material make it possible to dispense with the series-connected spark gaps and thereby to produce solidstate arresters suitable for system protection up to the highest voltages. magnetically blown arrester unit. and the spark gaps E) are mounted in an airtight porcelain housing.8 the currentlvoltage characteristics of the arrester are shown. a follow current flows. ensures that the voltage is distributed almost evenly among the arrester elements. To enhance the energy absorption capacity it is also possible to connect disks in parallel.9 for a disk of 80mm diameter and 32mm thickness. now flowing through the coil.. The arrester is then required to discharge a considerable amount of energy. The blowout coil presents no great impedance to the follow current. Figure 15. based on zinc. switching surges generally have long durations (thousands of microseconds). 1979). With such resistors one can design arresters having voltage/current characteristics close to idea.e. initially determined by the system voltage and the value of the discharge resistors. The voltage arising for a standard lightning current surge of 10 kA (Sj20 ks wave) is just over twice the reference DC voltage (Knecht and Menth. prevents local overheating or serious pitting of the electrodes. The electrodes of the tongue-shaped spark gaps are embedded in the disk-shaped chamber K.so the current commutates from the resistor Rb to the blowout coil. This arc voltage helps to extinguish the follow current even before its natural zero. The movement of the point of origin of the arc. When the discharge dies down with a rapid change in the current. The current. . This tends to draw the arc L and extend it. In normal operating conditions the control current i o flowing through the grading resistors R.7 shows schematically the arrangement of a modern. A recently developed ceramic material. A high arc voltage uL thus builds up. which is influenced by the shape of the spark gaps. In Figure 15. and cobalt oxides. Therefore. .nis (hundreds of microseconds).

which is a series-connected stack of such disks. =sparkover voltage. and R .7 Operation of a magnetically blown arrester: (a) arrester voltage and current. = bypass resistor. U = service voltage at . (d) passage of follow current. U . [From Ragaller (1980) . U UL = arc voltage during quenching. operates in a very simple fashion. = surge voltage. U .iA = surge current. ] The arrester. arrester assembly. U .4 Rb Figure 15. . = voltage drop across R . i N = follow current. The resistive losses in the . N . = residual voltage during diversion. resistors during quenching. (c) passage of surge current. is = control ~ u r r e n tRb . = guaranteed protection level. = grading resistor. It is dimensioned so that the peak value of the phase-to-earth voltage in normal operation never exceeds the sum of the reference voltages of the series-connected disks. (b) and (e) normal operation.

in contrast to the conventional arrester. No breakdown occurs but. Characteristics of metal-oxide arresterdisks. At the end of the voltage transient the current is reduced.nis Arrester current-voltage characteristics.e. the current will rise with the wavefront according to the characteristicsof Figure 15. a continuous transition to the conducting state. When an overvoltage occurs. there is no follow current). closely following the I-V curve (i.9 without delay. . rather.. arrester in normal operation aretherefore very small.

so these apprehensions are almost eliminated.. very stable resistors have been developed. Meanwhile.These varistors are widelyutilizedin protecting not only electric power linesbut also electronic components against dangerous voltage surges (Eda. An extremely stable I-V characteristic would guarantee constant losses for the entire life of the arrester as well as a constant protection level (Ieda et al. 1989). The resistivity of a ZnO varistor is very high (more than 1010S2-cm) below a certain threshold voltage (Vt. Furthermore. the absence of spark gaps makes the voltage grading system unnecessary.they are voltage-dependent switching devices. theoretically. However. The zinc oxide arrester is inherently self-regulating in that the current flow of 0.5-1 mA at normal supply voltage leads to reliable operation even in polluted conditions. Zinc oxide (ZnO) varistors are semiconducting ceramics having highly nonohmic current-voltage characteristics. . the absence of spark gaps results in a continuous flowof currentthroughthe device so that. whereas it is very low (less than 1-200 -100 100 200 Voltage (V) -2 (a) Current-voltage characteristics of a typical ZnO varistor.).10. Inother words. 1987).A considerable advantage of the zinc oxide arrester is its very simple construction.. the danger of thermal runaway is present. which originate at the grain boundaries. (b) schematic l-V curves for different nonohmic exponentsa. as shown in Figure 15..

the I-V characteristics are usually expressed logarithmically. The degree of nonohmic property is usually expressedby a n o n o h ~ i c exponent.When a = 1.Inthe regionbetween the threshold voltage and a voltage at a current density of about 100A/cm2. Inthe region above 100 A/cm2. ascan be seen from Figure 15. a. the simple equation I = CV@ (15. The nonohmic property comes from grain boundween semiconducting ZnO grains.Typical a values of ZnO varistors are from 30 to 100. the nonohmic property is not so prominent and ishighly dependent ontemperature. Hence. ZnO varistors are very useful devices as elements of surge arresters for protecting electric power lines-as outlined in the preceding section-or as surge absorbers to protect electronic components against voltage surges. The waveform of the impulse current has an 8 vs rise time and 20 ps decay time up to one-half the peak value. defined by the following equation: a = (V/I)(dI/dV) (1 5. The I-V characteristics measured by the impulse currents show voltages higher than those measured using DC. ZnO varistors have highly nonohmic current-voltage characteristics above a threshold voltage.10(b). This waveform is used as a standard impulse current to test surge arresters. and applications are rapidly found in protecting electrical equipment and electronic components such as transistors and ICs against voltage surges. The I-V characteristics of Z n 0 varistorsare classified into three regions. ZnO varistors are fabricated by sintering ZnO powders with small amounts of additives such as Bi2O3. Various additives to improve the electrical characteristics are sometimes usedto optimize the processing conditions. and Sb2O3. Themicrostructures and the physical properties of the grain boundaries are now well identified. it is an ideal varistor.cm) above the threshold voltage. and when a =0 0 . The I-V characteristics below 100mA/cm2 are usually measured using C electric source. ZnO varistors are characterized by the magnitude of the a values and the width of the range where the highly nonohmic property is exhibited. whereas those above 1 A/cm2 are measured by an impulse current source to avoid heat generation and thermal breakdown. where c is a constant.nis several Q. no. . it is an ohmic resistor. a values of conventional varistors such as Sic varistors do not exceed 10. In the region below the threshold voltage (typically corresponding to 1 pA/cm2.21) is used. the nonohmic property gradually decays. On the contrary. Since the range showing the highly nonohmic property is wide. thenonohmic property is very prominent andalmost independent of temperature.20) ~mpirically.

in the case even if a part of the GIS could be replaced comparatively quickly ite. and high gas pressure. The former case is characterized by a weak field nonuniformity. The dielectric constant of ZnO is 8. In contrast. .l.openair insulation may exhibit strong field nonuniformity. long gaps. the withsta~d of the GIs against temporary overvoltages is satisfactory. elow the threshold voltage. are tected against excessive overvoltage stresses. short clearances. This causes the need to protect also the distance entrance to an open circuit breaker. Good overvoltage protection is thus needed for all gas-insulated parts of the substation. and it is always at atmospheric pressure. While the withstand of air gaps drops to a minimum with increasing impulse front time before gaining strength under longer-fronted impulses. being the most valuable equipment and difficult to replace. The dielectric properties are mainly caused by thin depletion layers (mlOOOA) at the grain boundaries. It is concludedthat lightning overvoltages are more significant for GIs insulation dimensioning. ZnO varistors are highly capacitive. GIS insulation shows a steadily decreasing strength with increasing impulse time. In Figure 15. These differences produce sharply distinct voltage-time characteristics. as the insulation must be designed for the much higher lightning impulse. Due the rather flat insulation withstand dependence of the GIs on the front time of overvoltage. In contrast. Another important electrical characteristic of ZnO varistors is the dielectric property.The discrepancy is usually 10-20% and is caused by the delay of electrical response in the ZnO varistor. The response delayis caused by electron trapping and hole creation at the grain boundaries. The temporal development of breakdown in compressed SF6 in a GIs is significantl~ different from that in open-air insulation. as seen in Figure 15. as well as the strong electronegativityof the gas. whereas an apparentdielectric constant of a Z n 0 varistor is typically1000. a failure would lead to a great disturbance in the operation of the system. which in turns demands installation of lightning arresters at the line entrance where overhead lines are connected.l the V-t characteristics of insulation are superimposed on the frequency-of-occurrence diagram for the various types of overvoltage. whereas conductor spacings and other open-air V and UHV systems are basically determined by switching to a lesser extent by lightning overvoltages.5. For a conventional substation it is normally deemed sufficient if the transformers.

22) where n is the number of reflections. The wave impedance of an overhead line is considerably greater than that of GIS (typically 75 Q). When the wave travels along the line.(2n 1) t T . As explained in Chapter 14. however. respectively. the voltage at the entrance to the GIS is given as a function of time by v = Sat (15.2n t T The voltage at the open end at time t will be n m= 1 2(1 X)””“[t (2m 1)T] where 2(n l) . ~ i t finite h surge steepness S. As n increases. meaning the doubling of amplitude. The structures comprising towers and overhead lines have. It is therefore understood that the steepness of generated surges can be up to 7000 k V / p at the stroke origin. its peak and front steepness will be reduced because of corona discharges (see Chapter 5). U/Uo approaches 2. when reaching the GIS entrance. and thus reflection will occur at the entrance to the GIs.nis Design values of stroke current amplitude and steepness are about 70 kA and 70 kA/ys.24) where 2(n 1) . when a surge enters. At an open gap (disconnector or circuit breaker) the surge will be totally reflected. one part a is transa ) is reflected back along mitted into the substation and the other part (1the line. 1978) =2 (2 a ) ( la)” U 0 U (15. The entered wave is reflected at the open end and.23) before the first reflection. it will take the form (15.at least during the first microsecondafter a stroke has hit the line. a wave impedance of about 100 Q. The growth of the overvoltage will follow the equation (Eriksson and Holmborn. After n reflections. will be reflectedagain.

(Chapter 14) a With the presence of a surge arrester. Enclosed arrester at line entrance. in case of surges with steeperfronts and with longer CIS buses. and length of substation L. " . resulting in a higher spark-over voltage of the arrester. transmission coefficient . T the wave traveling time in a CIS. 15. KV 1500 - > frw S > 1000- v") 0 500 1 . Meanwhile. whereas the voltage buildup reaches a relatively lower level (Fig. On the other hand. a negative waveoriginates at the arrester and travels to the open bus. and n the number of reflections.and where S is the steepness of the wave on the overhead line. it will take a further travel time T between arrester and the open bus end. Therefore.11). the voltage has reached the sparkover voltage of the arrester. Thus the total voltage is reduced. 2000 kVlps incoming surge. I IO00 1500 2000 K V ure 15. the voltage buildup on the system is greater.11 Overvoltage versus sparkover voltage of GIs. for shorter bus lengths. L the length of the CIS from entrance to the open circuit. the sequence of events differs. T = L/300 PS. The voltage increase is dependent on three parameters: steepness S. the voltage increase by repeated refiections becomes rapid.

6times the normal ~hase-to-groundvoltage (IEC. .the following arrester characteristics can be obtained: A~PLITUDE \ RATE OF RISE 2000 kV/ur' 3000 KV(800 KV SYSTE 2000 t((5ZjO ** ** 1500 M ( M 2 t t 91 ) 3 ' (362 KV) 100 LENGTH 0 20 40 60 80 n Rate of rise of voltage across arrester versus GIS length..). the GI§. A choice of a reseal voltage equal to the temporary overvoltage results in reseal voltages 1.12 the average rate of rise ofthe voltage across an arrester at the line entrance has been plotted versus the distance from entrance to an open gap in.3 to 1.(reseal voltage). the surge steepnessacross the arrester will decrease. it is necessary also to have a low surge arrester sparkover voltage for fronttimes less than 1 p . Due to theflat SF6 V-f characteristics. With a longer bus length.. In Figure 15. The first importa~t decision to make is to select the surge arrester rated voltage V. 1976).nis The arrester sparkover voltage for steep surges is of great interest for the insulation coordination of CIS since the increase of the GI§ withstand level is limited for steep surges. Based on the rated voltage (V.

or drizzle.6.u. and are caused by contamination of the insulator surfaces.) BIL of CIS kV= 1300 If external insulation is designed on the basis of switching surge and lightning requirements. = 755 kV (peak) System c~aracteristics Maximum lightning overvoltage in the substation (computed at open end of asGIS explained) V& = 2.1 Vr if unknown) . (determined as stated above) = 2. These flashovers usually take place in wet weather conditions.2/50 ps sparkover voltage ( . there is a great variety of contaminants that can reduce the insulation strength. such as dew.4 V ~ ~ (The factor Ci is usually based on experience. 15. Rated system voltage (line voltage) = 420 kV (rms) Rated phase voltage Vph = 234 kV (ms) = 344kV (peak) Temporary overvoltage factor = 1.u. These particles maybe of natural origin or generated by artificial pollution which ismostly a result of industrial activities.2p. even in wet conditions.u. the strength for powerfrequencyvoltages and clean insulators.7p. A common natural deposit is sea-salt whichcauses a severe contamination of insulators of transmission systems in coastal areas. Arrester s p e c ~ c a t i o ~ ~ Arrester rated voltage Ce x v p h = kV 340 (ms) Protection level V . fog. In industrial areas. power frequency flashovers may occur on transmission systems without any switching or lightning surges.2. However.9Vr if unknown) Residual voltage at 10 kA ( the largest of which determines the protection level of the arrester.3. Contamination flashovers on transmission systems are initiated by the deposition of airborne particles on the insulators.Front of wave sparkover voltage ( . = 929kV Safety margin Ci = ~ ~ ~ = / 1.1 Example The following procedural example demonstrates the determination of the BIL of CIS equipment.4p. .2. is normally very high.1.3 Vr if unknown) 1.

so the voltage distribution on the string can be regarded to be the same as that on a dry and clean insulator string. Scintillations may not only depart from the pin but also from any place on the outer skirts.13. The equivalent circuit can be represented by a 0 5 10 15 20 25 30 35 Leakage distance measured from ground cap-cm . so the voltage is distributed over the wetted area. When this dry band cannot withstand the voltage applied to the unit. whose width can be a few millimeters. Initially. This fact is seen in Figure 15. . The flashover is determined by the flashover strength of the contamination layer which holds the whole voltage at the final voltage distribution just before flashover. The lossof strength is caused only by the combination of two factors: contaminants and moisture. The mechanism of long string flashover consists of different phases. as has been frequently observed. is produced by the leakage current. The surface impedance of a contaminated insulatorunit changes as a function of the wetting rate created by fog and the drying rate created by the leakage current. scintillation activity begins. The increase of the capacitance is a result of the increase of the conductive area on the insulator surface.With the progress of wetting. the impedance changes from capacitive to resistive. The voltage distribution on the unit also changes as a function of the time. The dry band.13. most of the applied voltage is impressed across a very narrow dry band formed around the pin. as seen in Figure 15. With the advance of wetting.nis these deposits themselves do not decrease the insulation strength when they are dry. The dry band is bridged by partial discharges. for different values of surface impedance.l3 Voltage distribution measured from grounded cap before onset of scintillation. the Contaminated insulator surface is completely dry.

acid exhaust. Soon this activity develops upward. phosphate. bulldozing dust. The insulator electricaldesign has to have sufficientlength to prevent an air breakdown at the appropriate voltage and have sufficient creepage to withstand the degree of pollution prevailing at the site wherethe insulators will be located in service.The table shows that the use of fogtype insulators may permit operation at one higher contamination class. Finally. this trend is accelerated. therefore. The value of the resistance is influenced by the drying effect created by leakage current. A wet zone is usually formed at the midsection of the string. the units on the bottom section cannot withstand the voltage stress and theyflash over. sodium plant. which is a function of the voltage across the unit. resistances of different units are not equal to each other. The surface resistance of the units under high voltage stress is higher than that of units under low voltage. The insulation size isrelated to the severity of contamination (General Electric company. mainly as a result of the capacitance between insulator units and ground. Thisis observed when the partial discharge at the bottom section bridges several units. The distribution is usually nonuniform. The most common contaminants causing contamination flashover are: seasalt. However. wood fiber. This activity appears as surge currents. sulfur. drying the contamination layer in this region until it forms a dry zone on the string. bird droppings. potash. gypsum. cement. and extinguish automatically due to their own linearization. usually having peak values ranging from 500 to 700mA (rms). Table 15. copper and nickel salt. road salt. paint industry. where the voltage stress is the lowest. fly ash.1 servesas a guideline for insulator design under contaminating conditions. carbon steel works. this heavy activity does not make the insulator surfaces in the wet zone as dry as those of the dry zone of the string. chemicals. and rock crusher. so the whole voltage distribution remains nonuniform. zinc industry. papermill. The bridging action over the bottom section results in an overvoltage for the insulators in the other sections and produces heavier activity along the string. The high voltage stress usually is concentrated at the bottom section of the insulator string. sulfate. the surface impedance becomes a combined capacitance and resistance. carbide residue. linearize the voltage distribution along the region not bridged by the scintillation arcs. 1979). . fertilizer. limestone. aluminum plant. These surge currents dry up the insulator surfaces in the wet zone. As wetting progresses. As the wetting condition progresses further. 1968. The surface temperature of these insulators is much higher than those on the remaining section of the string.network of capacitances only.

u.3 gx 2 U tu 8 .

11 Figure 15. Sur~ace cond~ctivity. K-2. the influence of the amount of insoluble material is significant but the differences in the physical properties of the insoluble materials (K-1. The surface conductance is defined as the ratio of the power-frequency current flowingover a sample insulator to the applied voltage. Contamination severity for a site can be judged from the behavior of insulators in the area or by measurement of the contamination. Several indices are known to quantify the severity of contamination: Equivalent salt ( ~ ~ deposit ~ 1 density ) (ESDD). The results are expressed in flashover voltage per unit axial insulator length. K-3) do not affect the flashover performance markedly. tonoko (K-2). . A s shown in the figure.14gives the flashover strength of contaminated standard insulators (type A-11) (General Electric Company.01 Q. the performance of insulators is examined for contaminant mixtures having three different insoluble materials. 0. The performance of V-suspension strings is about 25-30% higher than that of tangent strings. and Fuller's earth (K-3). This is expressed in mg of equivalent NaCl salt per cm2 of insulator surface.14 Performance of contaminated insulators with four kinds taminant mixtures. A s shown in the figure.1 Salt deposit density (rng/cm2) 1 of con- Figure 15. namely kaolin (K-l). 1979).

7.4.nis Surge count. 15.3 Clean-Fog Tests The test uses either a “slow” or a “quick” wetting process.4.15 compares a fog-type insulator to a standard .7.5. The degree of contamination is defined by the amount of salt in the sprayed solution. or it uses a “heavy mixture. 15. yet only in the case of slow wetting is the surface impedance capacitive-resistive. The insulator surface is purely resistive during the test. A combination of more than onemeasure sometimes applies.5. Highest current.1 Salt-Fog Test The contamination andwetting are applied simultaneously by spraying salt water. 15. 15. The insulator surface changes from dry to wet. 15. Fogtype insulators have a leakage distance up to 150% of that of the standard type of insulator.7.1 Over-insulation This is the most common countermeasure for all contamination conditions. The most common ways of combating the problem of insulator contamination are summarizedbelow. Figure 15. FZashover stress.4.7. Contamination test methods can be roughly classified into three categories.” in which case Contamination is defined by the amount of dry contaminant. The number of leakage current pulses above a certain amplitude measured over the insulator at working voltage. whose duration is one hour. The highest peak leakage current recorded during a given period over an insulator energized at the working voltage.” in which case contamination is defined bysurface conductivity. The power-frequencyflashover voltage divided by the overall insulator length. and is achieved by increasing the number of insulators.7.2 et-Conta~inant Tests The test either uses a “light mixture.2 Use of Fog-type Insulators In heavily polluted areas fog-type insulators are sometimes employed. In the former casethe layer has a resistive-capacitive impedance.

which can be either of hydrocarbon or ofsilicone materials. The greases. because of their temperature stability. has a deep skirt at the outer rim of the bottom surface.(b) fog-type insulator unit.7. 15. Another fog-type designisgenerallylabeled as “deep rib type” or “high leakage distance insulator. This countermeasure is extensively used in many countries. Today there are two basic types of insulator design.” and is similar to a standard type in appearance. and is particularly effective for “spot contamination” where maintenance is possible. It should be emphasized that leakage distance is not the only factor affecting contamination flashover of insulators. melt at the site of any discharge. Silicone greases.TYPE XIV 159 x 425 mm Leakage: 356 mm M&E Ratmg:178 k N TYPE XI 171 x 356 mm Leakage: 566 mm M&E Rating: 187 kN ure 15.3 Application of Si1icone Grease Silicone grease limits the ability of the insulator surface to retain moisture and produce leakage currents. can be used at higher temperatures. The grease is applied about 3 mm . and envelop the pollution. envelop the pollution and thus prevent the occurrence of discharges. and there is a variety of fogtype insulators whose performance is not satisfactory. This technique is also applicable to sea-salt contamination. considered as more traditional. Hydrocarbon greases soften on heating. One design.5. Thetendency to soften on heating restricts their use to temperate climates. unit.15 (a) Standard insulator unit.

This is attributed to the continuousdiffusion of the silicone fluidfrom the bulk and on to the surface of the pollution deposits. The most commonly used materials for outdoor insulators are glass and ceramics. and facilitate line compaction.5. It has been shown that if too thin a layer of grease is applied glaze damage can result. to a certain extent.5... Polymer insulatorsare lightweight (due to the high strength-to-weight ratio of the central fiber glass rod). The heat produced by the resistive coating keeps the surface dry and. Insulators coated with silicone rubber may withstand exposure to severe weather conditions for 3 years without damage. 15.Polymers are nowbeingincreasinglyused as replacement for porcelain and glass for outdoor high-voltage insulation. Line insulators can be cleaned either manually or by the use of water jets to remove pollution deposits. 15.4 lnsulator~ashing Washing and cleaning of deenergized insulators has been used regularly in the past. The removal of the grease at the end of its useful life can be rather time consuming. 1994..5. 1996. Corur et al.7. using a fixed spray installation or amanually controlled portable jet.5 SemiconductiveGlazed Insulators This provides the insulator surface with a resistive coating. Washing techniques can be carried out with the insulators energized. surge arrester housings. 1996). Xu et al. provides for a linear potential distribution. etc. are easy to install.Anis thick either by hand or by dipping.6 ~oom-Temperaturevulcani~ing In order to suppress the leakage current. Silicone greasesare morecostly than hydrocarbongreases. such as line and stationinsulators.7. terminationsforunderground cables. evenwhen a layer of contaminationhas built up on the surface. The frequency of washing depends on the site condition and should be adequate to maintain the insulator in a clean condition. Live line washing is sometimes employed.. The RTV coating initially provides water repellency (hydrophobicity) which prevents the formation of continuous water filming on the surface. phase spacers.7. Polymeric cable terminations . 1996). thus limiting leakage currents. 15. This is due to several significant advantages theyoffer (Matsuoka et al. possess good vandal resistance. and thus suppresses the leakage current and flashover. a low-energy surface is created by coating porcelain and glass with room temperature vulcanizing (RTV) silicone rubber (Deng et al.

the material is subject to aging and will degrade with time. the aging mechanisms responsible for field failures are not fully understood.6. 1994). act to accelerate the aging. Other methods are based on the fact that leakage current monitoring may bean unreliable indicator of contamination level except near pre-flashover levels. may not be fully expected when the hydrophobic surface of siliconerubber insulators is covered with a large amount of contaminant and/or water. heat. Agingisresponsible for twotypesof failure observed in service.. they can contain violent explosions that can occur with porcelain within the housing.rubber insulators may not besignificantly different from that of porcelain insulators. But these require a detailed understanding of mechanisms responsible for aging. the contamination withstand voltage of silicone. owing to their inorganic nature.. This is because leakage current is a function of both the water present on the surface and the amount of con- . Leakage current leading to dry band arcing. thereby minimizing injury to personnel during arrester failure. 1986). polymers can also provide higher resistance to leakage current. Due to their hydrophobicity or water-beading property.it is said that hydrophobicity on the surface is maintained for a long period because of migration of low molecular silicone from the bulk of the silicone rubber to the surface. chemicals. Other factors such as ultraviolet (UV) in sunlight. is mainly responsible for aging of polymers. however. etc. Knowledge of life expectancy and ability to evaluate the long-term performance in the laboratory within a reasonable time are of paramount importance to utilities. Hydrophobicity.1 Aging in Polymer Insulation With polymers. In the case of polymerinsulators with siliconerubber (SiR) sheds. 15.have provided users with highly simplifiedapplication techniques that have made them more reliable and inexpensive than porcelain. Presently. promoted in the presence of moisture and contamination.7. and flashover at the operating voltage under contaminated conditions. For surge arrester housings. Many methods have been established to monitor the accumulation of contamination. Some of those methods are based on monitoring the leakage current (Habib and Khalifa. namely. material degradation in the form of tracking and erosion. which improves the performance of insulating systems under contaminated conditions. Under such contamination and/ or wetting conditions. with the result that the electrical and mechanical functions are interdependent to a greater degree (Karady et al.

without wavelength scanning it lacked the ability to distinguish the different types of contaminant. Estimate the breakdown probability under 145 kV impulses. Derive the timeto-breakdown of the compound system. Estimate the 50% breakdown voltage and the corresponding standard deviation.51 Anis tamination. If the overvoltages on the substation are statistically distributed by a normal distribution whose mean is 180 kV and standard deviation 40 kV. The probability rose to 75% when the voltage was raised to 135kV. However. (3)Twodissimilar insulators are connected in parallel. A contamination monitor was developed which used ellipsometry to provide valuable information about the optical parameters of the contamination which determine its thickness (Mahmoud and Azzam. the statistical overvoltage. A single-wavelength ellipsometer was suggested for anticipating the dominant type of Contamination in a certain area. Estimate the overall breakdown probability for the combination. (1) An external insulation when tested in the laboratory gave a flashover probability of 5% under 120 kV impulsevoltage. Estimate the overall breakdown probability under 130 kV. a contamination liquid water sensor (LWS) was shown to be a successful surrogate indicator of the buildup of contamination on insulators (Richards and Renowden. estimate the statistical withstand voltage. The work proposed the useof spectrophotometry to measure the thickness and refractive index. The sensor measures both the amount of liquid water and the levelof contamination on an insulator surrogate. . the statistical safety factor. (4) The insulator system in a substation is characterized by a breakdown probability distribution having a 50% FOV of 220 kV and a standard deviation of 10 kV. 1997). This method provided an alternative to the laborious ESDD measurements for quantifying insulator surface contamination. The LWS is effectivefor the determination of contamination levels when the relative humidity is greater than 65%. As the amount of liquid water increases. one of the parameters of the LWS changes. The distribution of their times-to-flashover could be assumed uniformly distributed between120 and 140PS for the first insulator and between130 and 150 ps for the second. Alternatively. 1997). Under a certain impulse voltage the breakdown probabilities of the two insulators were 15% and 25%. (2) Three insulators of the type described above were connected in parallel. and the resultant risk of failure. This would deliver important information about the existence of certain contaminants by showing their characteristic absorption lines.

Menth A. Khalifa M. IEEE Trans PWD-ll:1895-1900. Brown Boveri Rev 66(11):739-742. 1978. Habib S. Matsuoka R. Mimno Y. Diesendorf W. Publication 60. Transmission Line ReferenceBook-345 kV and Above. 1989. Geneva: International Electrotechnical Commission. Radwan R.01. Mizutani T. 1988. Orbeck T. Publication 71. 1996. El-Morshedy A. 1979. 1968. Gouda 0. Richards CN. Hackam R. Karady GG. Palo Alto. Ohki A. Knoxville. Hervig HC. KondoK. Habib S. Renowden JD. IEEE Trans DEI-4:33-38. Khalifa M. CIGRE. General Electric Company. CA: Electrical Power Research Institute. IEC. Anis H. 1996. Proceedings of IEEE Winter Power Meeting. pp 314337. Holmborn H. 15-107. IEEE Electr Insul Mag 5:28-41. 1994. Deng H. Gorur RS. Terada T. Surges in High-Voltage Networks. Knecht B. 1973. El-Morshedy A. IEC. Schneider HM. 1978. Proc IEE 135C:2430. 1980. Palo Alto. McGrath PB. . Eriksson R. 1974. CA: Electrical Power Research Institute (EPRI). Ragaller K. Report 300. Consult surge arrester manufactureres’ specifications and make any necessary assumptions. New York: Plenum Press. 1996. Proceedingsof CIGRE Symposium. Fujimara T. Champion T. Ieda M. Electrical Power Research Institute. Mahmoud F. Naito K. Shinokubo H.17 (5) Repeat the conventional insulation coordination example given in the text for a 550 kV line subjected to temporary overvoltages of magnitude 445 kV (rms). ed. Project UHV. Transmission Line Reference Book-345kV and Above. CIGRE. 1976. 1986. Eda K. Insulation Coordination. IEE Proc 133:105-108. 1997.Bernstein BS. 1997. Azzam RMA. New York. London: Butterworth. Cherney EA. High Voltage Test Techniques. Rizk FAM. Ch 7. IEEE Trans DEI-3:289-298. 1979. 1978. Suzuki Y. General Electric Company. 33-103. Proceedings of 1st International Symposium on Gaseous Dielectrics. IEEE Trans PWD-12:389-397. IEEE Trans PWD-l1:431-443. pp 25 1-28 1. Xu G. Geneva: International Electrotechnical Commission. Ch 10. Paper A-78-154-7. Insulation Coordination in High Voltage Electric Power Systems. New York: Edison Electric Institute. TN. 1987. Vienna. EHV Transmission Line Reference Book. 1994.

This Page Intentionally Left Blank .

in high-voltage testing. high voltages (DC. To name just a few examples: 1. the currents are limited to a small value up to about 1 A under AC o r DC voltages and a few amperes under impulse voltages. currents . require high DC voltages of several kilovolts and megavolts. Assiut. depending on their normal operating voltages. AC. In fact. Electron microscopes and x-ray units require high DC voltages on the order of 100 k'v or more. Normally. and lightning impulse voltages. particle accelerators. 3. one main concern of high-voltage engineers is for insulation testing of various power-system components under power-frequency AC. Testing the insulation of power apparatus requires high AC voltages of up to millions of volts. and so on. and impulse) are required for several applications. Simulation of overvoltages that occur in power systems requires high impulse voltages of very short and longer durations. DC. switching. as discussed in Chapter 18. 2. 4. Electrostatic precipitators. Egypt In the field of electrical engineeringand applied physics.Assiut University. during testing of surge arresters or short-circuit testingofswitchgear. However.

a single transformer can be used. the excitation windings at the lower stages evidently carry higher loading than those of the upper stageE. respectively. The equivalent impedances of testing transformers are usually within about 5%. The other terminal of the high-voltage winding would be insulated to the full output voltage. 16. and erection problems connected with large testing transformers become prohibitive. The highvoltage winding is arranged in layers that are carefully insulated and the potential distribution over it is carefully controlled.construction of a single transformer would entail undue insulation problems.the power carried by the second and first stagesare. however. Cascaded transformers were first used by Petersen. Single-unit testing transformers do notdiffer from single-phase powertransformers with regard to the design of core and windings in relation to the kVA output. Considerable economy is achieved. Dessauer. 1978).if the center point rather than one terminal of the high-voltage winding isearthed.several orders of magnitude higher are required. For higher voltages. Also. 2P and 3P. For generating AC test voltages of less than a few hundred kV.and low-voltage windings. The L and E windings are therefore rated at currents higher than those of the H windings. However. The total short-circuit impedance of a cascade . If the power carried by the third stage of Figure 16. is usually maintained at earth potential. Thus each terminal of the high-voltage winding will be insulated to only halfthe output voltage. These drawbacks are avoided by cascading several transformers of relatively small size with their high-voltage windings effectively connected in series. transportation.1) (Kind. particular attention is given to heavy insulation of the high-voltage winding and to low magnetizing currents for minimum distortion of the output voltage waveform. Methods of generating high voltages and high impulse currents are discussed in this chapter. expenses.1 is P. The iron core. and Welter (Fig. as well as one terminal of the low-voltage and highvoltage windings. A compensating winding is installed close to the core to reduce the high leakage reactance between the high. The low-voltage winding (L) of each upper stage is fed from an excitation winding (E) in the stage immediately belowit.

F. C. excitation winding.1). 1. iron core.1 Cascaded transformerswithcompensatingwindings and extra insulation. low-voltagewinding. To reduce the size and cost of the insulation. . high-voltagewinding.ure 16. L. transformers with center-tapped high-voltage winding are cascaded as shown in Figure 16. The ironcore and the container (if metallic) of each stage other than the first are accordingly insulated from earth. 16. The tankof the first stage is kept at V/2. while those of the second and third stages are at 3 V / 2 and 5 V / 2 (Fig. transformer can easily be related to the impedance of the individual stages (Kind. H. 1978).1. This scheme is more expensive and requires more space than that of Fig. E. If the outputvoltage of the first stage is V. 16. the full output voltage is its multiple according to the number of stages. compensating winding.

2). Total height about 12 m. 1200 kV cascaded transformer with insulating containers. (Courtesy of ~ e s s w a n d l e r . 16. beingdriven by a full-rated voltage regulator. are referred to as “straight circuits.Cascade transformers of ratings up to 10MVA and voltages up to 2. These transformers.’’ as distinguis~ed from resonant circuits.) u .25 M V are available for bothindoorandoutdoor testing(Fig.6 a AG.

3. v 2 = 1 1 -@2LeC (16.The test circuits discussed in Section 16.2 consist of a test transformer (single-unit or cascade) and a voltage regulator.3c). 16. thus improving the power factor.3. Pure sinusoidal output waveforms. the risk of self-excitation of the alternator is eliminated. The shuntcapacitance C usually represents the high-voltage bushing and the test object.1 Parallel Resonance The additionof parallel reactors either in theprimary low-voltage circuit or in the secondary high-voltage circuit may partially or completely neutralize the capacitive load current. Since R.2. andfilters. harmonics are heavily attenuated.Thus there is no longer a fixed ratio of primary to secondary voltage. reactors. . as the resonance is heavily disturbed by the resulting short circuit. either by accentuation of harmonics present in the incoming supply or generated in the transformer itself due to its nonlinear magnetic performance.2. By resonating the circuit through a series reactor L (Fig. Cascading is also possible for producing very high voltages. thus reducing drastically the cost of the regulator. A disadvantage of this arrangement is that harmonics can appear in the test voltage. 16. 16.3a) at the test frequency (50 or 60 Hz). Figure l6. such simple straight test circuits would not be entirely suitable for certain tests.2. and the volthe equivalent circuit of the test transformer. If a motoralternator is used as the supply source. 2.2 Series Resonance An alternative system is the series resonance circuit. Without filtering out these harmonics.3b shows <@L. Therefore.1) A s resonance is approached. < tage Y2 is almost in phase with V\ (Fig. No high-power arcing or heavy current surges occur if the test object fails. An alternative method that is more economical and sometimes technically superior is offered by resonant circuits. such as those for partial discharges. 3. V2 > > Y \ . Less power requirements from the mains (5-10%of straight circuit requirements). Input power reductions of 10:l are feasible. 4. 16. both rated to take full power from the supply. The main advantages of series resonant transformers are: 1. the secondary voltage itself should be measured accurately in the tests.

5.c v2 l 1 . Series resonance sets are usually brought into resonance by mechanical adjustment ofan air gap .S (a) Simplified circuit of a resonant transformer. The main disadvantage is that the additional variable reactors should withstand the full-test voltage and full-current rating. (b) equivalent circuit: (c) phasor diagram. Simple and compact test setup.

1). The RF = ure 16.4) and for testing installed gas-insulated systems in the field. (Courtesy of . 16. and V(av).25 in the iron core of the reactor.3. The “ripple factor. is the value of V(av) at no load. Besides the Van de Graaff generators. The output DC has a trace of AC (ripple).4 800 kV/8000 kVA Hypotronics. is equal to V(av) V(av)o (16.” by definition. Silicon rectifiers are commonly used.7. which increases with the current drawn by the load.2) V(av)o where V(av) is the average value of the rectified output voltage.) series-resonant test system. with their special applications (Section 16. These sets are designed and commonly used for partial discharge testing(Fig. high-voltage AC and rectifiers are usuallyused for generating HVDC.

For ' reasonably small ripple voltage A V . In half-wave and full-wave rectifier circuits (e. z should be at least 10 times the period Single-phase full-wave rectifier with smoothing capacitor: (a) circuit diagram. Different transfo~er-rectifier circuits for generating high DC voltages are discussed below.. Fig.g. . High frequency is recommended when a ripple ofverysmall magnitude is required without use of expensive filters to suppress it. During the other periods.AC supply to the rectifiers may be of power frequency or may be ofa higher frequency.5). (b) steady-state voltages and currents with load R. the capacitor is discharged into the load with a time constant z = CR.the maximum AC secondary voltage of the V transformer during the conduction periods. 16. the capacitor C is gradually charged to Vs(max).

of the AC supply. provided that the t r a n s f o ~ e r has an adequate insulation. a charge q is supplied from the transformer to the capacitor within the short conduction period tl of the rectifier element and is transferred to the load R during the long nonconduction period t2. rectifiers 1 and 2 conduct during alternate half-cycles and charge the capacitor C. The product fC is animportant design factor. thus producing an unsmoothed unidirectional voltage.and the t r a n s f o ~ e in r series. A V is easily found from the charge q transferred to the load: (1 6. an additionalresistance is connected in series with the secondary of the transformer. and during one period T(= l/f) of the AC voltage. . This voltagedoubler circuit may be earthed at any point. where tl < Neglecting the voltage drops within the transformer andrectifiers during conduction. The rectifier element should have a peak inverse voltage of at least 2Vs(max).5) In full-wave rectifier circuits (Fig. so that (16. In Figure 16. The output voltage has a peak value of 3Vs(max) at no load.3) <t2 z T = 1/f and IL is the average load current. With the load R connected.4) This equation correlates the ripple voltage to the load current and the circuit parameters f and C. The ripple voltage according to equation (16. According to equation (16.4) is halved. C2. the voltage ripple factor for half-wave rectification is (16.2).5).6a two half-wave rectifier circuits are connected in opposition. With half-wave rectification. the capacitors discharge throughit with acorresponding voltage drop. 16. To limit the charging current. The source is effectively Cl. The HV transformer requires a center-tapped secondary with a voltage rating of 2Vs.

. (b and c) output voltage contains onlya small ripple.Connection diagrams and steady-state voltage waveforms:(a) output voltage containsa considerable AC component.

Rectifier 1 and capacitor Clact as a clamp so that the voltage at and is clamped at O V with a peak valueof node A ispositivegoing 2 Vs(max). the voltage ripple factor is RF = 6V AV12 2 Vs(max) + (16. obtained from PSpice. Also. Thus.6) C 2 f At the same time. with a small ripple under load. 6 ~ (Price. The capacitor C2 provides the load current until the next peak of voltage occurs at node A. it is discussed in more detail. It is quite clear that the doubler circuit takes a few cycles of the input for the output VR to reach the maximum value. rectifier 2 stops conducting as the voltage on C2 is now greater than the voltage on its anode. let AV be the ripple voltage.Then II 2q / t z . the main voltage drop below 2Vs(max) is (6V average output voltage Vav = 2Vs(max) 6V 2 + A V/2) and the (16.8) . where t2 is the time during which rectifier 2 does not conduct.7) AV According to equation (16. Rectifier 2 conducts and charges C2to 2 Vs(max) whenthe voltage at node A is at its peak value. Let a charge q be transferred from C2 to the load per cycle when the average load current is Il. t2 is much longer than t l . As the doubler of Figure 1 6 .In the circuit of Figure 16. The output and input waveforms. a charge q is transferred from Clto C2 per cycle. 6 ~ represents the basic unit of the commonly used Cockcroft-Walton voltage-multiplier circuits. 1997). the conduction time for rectifier 2. the voltage at the output reaches 2Vs(max).2).6b. then or (16. are shown in Figure 1 6 . As the voltage at A falls below 2Vs(max). Since q = CzA V and tl < <t2 N T = l/f. so AV=6V=I1 1 1 c 1f Therefore. each of the two capacitors is charged to only &@ax) in alternate half-cycles and the total output voltage is 2Vs(max) at no-load.

7 Voltage-triplercircuit: obtained from PSpice.A voltage tripler is an extension to the Cockcroft-~alton voltage-doubler circuit. The output and input waveforms obtained from PSpice are shown in Figure 16. that is. On the negative half-cycle. rectifier3 conducts and voltage whichis applied to C3 is the sum ofVs(max) from the source and 2Vs(max) from C2 less the voltage on C . The output voltage is obtained between nodes A and B and has a value of 3 Vs(max). (b) voltagewaveform . the capacitor C2 is charged to 2V8(max). as shown in Figure 16. During the next negative half-cycle.7a. .7b where the output reaches 3Vs(max)after a few cycles of input. Clcharges through rectifier 1 to Vs(max). On the positive half-cycle. rectifier 2 conducts and the peak voltage is Vs(max) from the source and Vs(max) from Cl. C3 is charged to 2Vs(max). (a) circuitdiagram. . that is. The operation is as follows.

10) where f is the supply frequency. AV3 = qlC3 (16. Cltransfers a charge q to the load and q to C2 (Fig. so the ripple voltage at C 1 is AV1 = 2q/c1 (16.8a.14) According to equation (16.12) Because of the charge q transferred fromC1to C2per cycle.7a).9) (16.2).15) The voltage-multiplier circuit using the Cockcroft-Walton principle is shown in Figure 16. the total voltage drop below 3Vs(max) is and the average output voltage = 3Vs(max). The ripple voltage at capacitor C3 is ~imultaneously. The first stage. the voltage ripple factor is RF = 6V AV12 3 Vs(max) + (16. 4 =Ill. the ripple voltage will be (16. composed of rectifiers 1 and 2.The same principle can be extended to higher-order voltage multipliers.11) Hence. 16. Let a charge q be transferred through C1 and C3 to the load per cycle when the average load is Il.the voltage drop &V= q/c2 (16. the .q (16.13) Therefore.

? 1. 6Vs(rna ' ! ure 16.8 Voltage-multipliercircuit according to Cockcroft-Walton: (a) circuit diagram: (b) tracing of load voltage-drop6Vand ripple AV. .

1 Ripple in ~ u l t i p l i eCircuits r Referring to Figure 16. the total ripple voltage will be (16..2 is Similarly. 16. However. as a transient overvoltage would overstress the smaller capacitors. and the corresponding total ripple voltage is (16.g. 16. C2n-4 transfers a charge q to the load. For higher output voltages of 4.3. Therefore. which is charged to Vs(max) only. But the voltage across any individual capacitor or rectifier is 2Vs(max). is identical to the voltage . Thus the total output voltage reaches 2nVs(max) above the earth potential (Fig. equal capacitances (C) are usually provided. the input voltage Vs(max). the circuit is repeated with a cascade connection. 'the ripple voltage at C2 is (16. q to C2n-39 and q to C2n-l with a ripple voltage at C 2 n-4 : (16.19) Hence. and the HV transformer.4. let q be the charge transferred from C2n to the load per cycle with a ripple voltage at capacitor C2n: (16.18) Proceeding in the same way.20) It is quite evident that the capacitors near the ground terminal (e.so the ripple voltage at capacitor C2n-. 6.. Figure 16.8a.capacitors Cland C2. C2) are the major contributors tothe ripple voltage Iftheir capacitances are chosen suitably large. except for Cl.6~. this is not practical.21) . the total ripple voltage can be reduced significantly.2n of doubler shown in. .16) Simultaneously?C2n-2transfers a charge q to the load and q to C2n-1.8a)..

This reaches an optimum value beyond which it . C2n(Fig. C3 can be charged only up to a maximum voltage of [2Vs(max) 2nq/C] instead of 2Vs(max). For fast response to load changes and supply variations. 16. Cland C2 charge up to [ Vs(max) nq/CJ and [2 Vs(max) n q / q instead of [ Vs(max)] and [2Vs(max)]. . n >5).8a): 4 6V2 = -n C 4 6V4 = -[2n C 4 SV.. An increase of supply frequency is. VR(peak) can be increased with the number of stages n. capacitor C2 loses nq during each cycle. more economical than an increase in the capacitance values.An upper limit of 500-1000 H z is set bythe high voltage appearing across the circuit inductances and by the high capacitive currents to be drawn. ~imilarly. so capacitor C. has to replenish it.8b) 4 x2 n3 VR(peak)= 2n Vs(max) fC 3 (16. the first term in equation (16.e. = -[2n C - + (n -l)] + 2(n -1) + (n 2)] Adding all the n voltage drops gives the total voltage drop on load: (16. there is a voltage 6V which is the difference betweenthe theoretical no-load and on-load voltages.Voltage Drop on Load In addition to the ripple voltage A V. which might be an essential design factor for special breakdown investigations of dielectrics. Small valuesof C provide a C supply with limited stored energy.23) The output voltage VR(peak) can be increased by increasing either C or the frequencyf.8a.22) In the case that the generator comprises a large number of stages (i.. 16. in general.. As seen in Figure 16.C4? C6. respectively. Thus one can easily derivethe general rule for the total voltage drop at the smoothing column C2.8. With reference to Figure 16. high supply frequency and small stored energy are prerequisites.22) will dominate the others and the load voltage will be approximately (Fig.. Therefore. for a given load.

24) Thus. the ripple factor is expressed as RF = 26 Vtotal + A Vtotal 4n Vs(max) (1 6. that is. there are limitations with regard to the number of stages. The very small ripple factor. fast response. For given values of1 1 . the optimum number of stages nop N 12.8b).9). V.~for a multiplier circuit withf = 50 Hz. The transformer in each stage comprises a tertiary low-voltage winding which excites the primary winding of the next upper stage. high stability. and C.drops if n is chosen too large.21) and (16. as the lower transformers have to supply the energy for the upper ones. C = O.22) (Fig. when (16.lpF. This disavantage can be overcome by cascading voltage doublers without a need for additional isolating transformers for every newstage. the mean output voltage A Vtotal VR(av)= 2n Vs(max) 6 ‘Vtotal 2 ~ (l 6. low ripple and voltage drops can be achieved when load currents are high. Vs(max). However. 16. Using equations (16. This can be achievedif the different stages are energized by specially designed cascaded transformers. the output of the circuit is limited to about 1 “ vand a few milliamperes and needs excessive insulation for the higher stages.9. the highest value of VR(peak) is achieved with the “optimum” number of stages. A very sophisticated cascade transformer HV DC voltage multiplier circuit is shown in Figure 16. and I1= 8 mA.26) nsf 1 The multiple charge transfer with the voltage multiplier of CockcroftWalton type demonstrated the limitation in the DC output concerning the ripple voltage and voltage drop. and small stored energies are the main advantages of this circuit. These .25) Therefore.~. In this way. 16.(max) = 200 kV. However. The circuit consists primarily of a cascade connection of t r a n s f o ~ e r s having no iron cores and fed from a high-frequency source (Fig.

transformers are coupled by series capacitors CS to compensate for their leakage inductances. . The voltage remains nearly constant over the cascaded transformers but with a small phase shift with respect to the input voltage. Also. 16. A s the frequency is high.oscillator (50-100 kHz 1 Deltatron circuit operating principle.9). shunt capacitors C. The string of cascaded transformers is loaded by a terminating resistor Rt(Fig. the capacitors C of the cascades can be made very small. 16. The small ripple voltage characterizing this circuit is causednot only by the small storage capacitors. compensate for the transformers' magnetizing inductances.9). The usual Cockcroft-Walton cascade circuit is connected to each stage (Fig. but also by the phase shift between the input voltages of the different stages provided by their transformers. with a correspondingly low energy stored and fast circuit response to load changes or supply variations.

the ions of either polarity produced by corona at pointed electrodes are carried mechanically by a very highly insulating belt to the top of the apparatus.29) . where they are conducted to a voluminous high-voltage terminal. In Van de Graaff generator. and vice versa.28) If dC/dt is negative. dVl/dt = 0 and the power input will be (16. the generator volltage builds up a shigh as millions of volts.7.1 Van de Graaff Generators Van de Graaff generators have been developed to produce very high voltages of 5-6 MV with output currents of microamperes.Compressed gas and/or grading rings are used to raise the output voltage of the generator.16. into the circuit at any instant P= dV1 2dC vlr = cv.dt dt The power intput where C is the capacitance being charged to a voltage VI. A s more and more charges accumulate.2’7) r=c-+v. which can easily be reached with precision and flexibility. The main advantages of these generators are their ripple-free high DC voltages. the belt isreplaced by an insulating cylinder surrounding a gas discharge chamber and rotating at a high speed. forming a variable capacitor that operates in vacuum. A modification of Vande Graaffs generator wasmade byFelici (1953) to eliminate its mechanical shortcomings as regards the belt and its vibrations. mechanical energy is converted into electrical energy.2 Variable-CapacitanceGenerators These generators are designed to convert mechanical energy into electrical energy usingthe variable-capacitanceprinciple. The chamberis subjected to a high DC field and ions of both polarities are drawn from it by a pair of electrodes connected to the load. With the capacitor charged from a DC voltage. It is limited only by the dielectricstrength of the terminal’sinsulation to ground. They consist of a statorwith interleaved rotor vanes.-+ vl dt dt (1 6. The current through a variable capacitor is given by dV1 dC (16. The output current is limited to a few a microamperes (see example 5). In Felici’s generator. 16.3.3. They are useful for energizing particle accelerators. Such a source can produce an output of a few hundred kilovolts at a few milliamperes.7.

The regulators are generally of a series or a shunt type 1986).30) is greater than unity and indicates how many times better V is in stability compared wtih V s .2m maximum and 0. As the rotor rotates. C voltage regulator consists of a detector that senses the voltage from the desired value. A generator of this type. 50 rotor plates of 1. and a controller actuated by the detector in such a manner asto correct the deviations (Naidu an araju. It is necessary to keep the change in voltage between f O . 1% and f0.hen the rotor-to-stator capacitance is maximum (Cm). and further rotation of the rotor eventually causes the current to flow from the generator to the load. . 1982).001%. If A V is the change in the DC outputvoltage V as a result of a change A V1 in the AC supply voltage Vl. with an output voltage of l MV and a field gradient of 1 MV/cm in a hard vacuum and having 16 rotor poles. To maintain a constant voltage at the load terminals.6 m minimum diameter. that is. the charge between the rotor and the stator is therefore Q. could develop a maximumpowerof 7 MW. This vacuuminsulated generator was first discussed by Trump in 1947 and recently investigated by Philp (1977). depending on the applicatio~s.31) is the effective internal resistance of the regulator as seen from its output terminals. = CmVl. AV/V @ r =nl/r (16. it is essential to have a regulator circuit. and a speed of 4000rpm. The “regulation” @r of a DC voltage regulator is defined as the fractional change in V caused by a fractional change in the output current I . and R is the resistance of the load. the stabilization ratio is defined as (16. the capacitance C decreases and the voltage across C(= C m VI /C) increases. The output of a DC generator changes with the load current as well as with the supply voltage.

.2 ps and a wavetail duration of 50 ps.21 (IEC. the tolerance allowed in the peak value is 53%. lightning impulse wave has a front duration of l . 1973).10. However.32) (1 The graph ofthisexpressionisshown in Figure 16.Standardized lightning impulse waves equation e(t) = A[exp(--al t ) exp(-a2t)] are represented by the general 6. The standard wave front and tail are defined in Figure 16. resp~ctively. and is described as a l. +A ernax -A Impulse wave and its components as a double exponential. The tolerances allowed in the front and tail durations are 530% and 520%.2/50 ps wave.

b. The other circuits (b-d) are simply special forms ofcircuit a. This is achieved if R2 is much larger than R1 (Fig. and d of Figure 16.34) .1. 16.1 Circuit Analysis With reference to circuit a of Figure 16. the faster is the rate by which the output voltage e(t) approaches its peak value.11.11 b and c. second. c. The circuits commonly usedfor impulse generators are those of Figure 16. Cl&. so the voltage efficiency (16. However. being capacitive. a long wavetail calls for a slow discharge.1l a 4 or by a series R-L-C circuit under overdamped conditions (Fig.1lb-d).respectively.).Theimpulseenergy consumed is + + + which is an important characteristic parameter of the impulse generator. 16.11. 16.The double-exponential waveformexpressed by equation (16. C1(RI Rz). il = i2 + io (16.1le. and Cl(R1 R2) in circuits a. which isdifferent from the circuit of Figure 16. the test objects admittedly form a part of C2. The peak value emaxcannot exceed the value determined by the distribution of the initially available charge VC1 between C1 and C2.32)maybe obtained in the laboratory by any connection of the R-C circuit as exernplified in Figure 16. Analysis of the inductive circuit e is that of a simple R-L-C circuit. being a general one. the capacitor Cl discharges through the circuit while its voltage decreases from its initial magnitude Vo. The exponential decay of the impulse voltage on the tail occurs with a time constant of about CI(R11 R. The smaller the time constant of the circuit. The magnitudes of the different circuit elements together determine the required rapid charging of C2 to the peak value emax.1l . The advantages of these circuits are that the wavefront and wavetail are independently controlled by changing R 1 and R2 separately. on breakdown of the gap C.1le). Clshould be chosen much larger than C2.4.33) For high efficiency. It is sufficient to analyze circuit a of Figure 16.

11 Circuits for producing impulse voltage waves. .1 G R G L igure 16.

1982) (16. the impulsewaverepresentedby equation (16. the following relation for e is reached: d2e de Al-+A2-+A3e=0 dt dt2 with the initial conditions el and t=O (16.37) Using simple mathemat~calmanipulation.39).38) =o Hence The peak value emax and the corresponding period tmaxcan be evaluated by differentia tin^ equation (16.36) Hence (16.35) (16.32) can be written as (Auget and Lanovici.40) "!a a]= 0 8 (16.54 (16.42) and . In a normalized form.41) a 2= a+G-3 (1 6.

the choice of the resistances to control the wavefront andwavetail durations is not entirely independent but depends on the ratio C1/C2. 1982). The ratios T1/T2 and T2/6 versus a are shown in the nomograms of Figures 16. Table 16.2 Dimensioning of Circuit Elements For a given wave shape.14. as originally proposedin1923by Marx. . The lengthening of the wavefront by increasing the circuit inductance provides a convenient method for generating long-front (switching) impulses.4. For circuit b of Figure 16. and in circuit d. v ing elements of the circuits (Kind.43) (1 (16. A bank of capacitors is charged in parallel and then discharged in series. 1982).3 gives the corresponding limiting values of C1/C2. Evidently. 16. It is noted that. 1 This is the best way to generate voltage impulses of very high amplitude (millions of volts) using a DC source of a moderate output.45) .6 = JCIC2(R11R12 + RllR2 + R12R2) R 6. Kuffel and Zaengl. Table 16.12 and 16. in circuit b. and a and the correspondTable 16.However. On the other hand.13 for lightning and switching impulses (Auget and Lanovici. with increasing series inductance. 1978. 16. an inductance across the load would drastically change the waveform. RI1= 0 and R12= R I . various modifications havebeen implemented.2 gives the relationship between the time constants l/al 1/a2and their ratios to the durations TIand T2 for typical impulse waves including lightning and switching impulses. in circuit c.11 there is an upper limit whereas for circuit c there is a lower and limit. 1984.1. there is a limitation on this ratio. R12= 0 and RI1 = RI.1. R12= Rll = R I .1 givesthe values of the parameters 0. as depicted in Figure 16. Augetand Lanovici. Since then.4. but the magnitude of the peak value is affected only slightly. the wavefront sensitively increases.3 Effect of Circuit Inductance Any residual inductance in the circuit elements would produce some oscillations in the wavefront and wavetail.

n W G S" n v G d "Cn IG n W II I [ T 4 % B c fx a c c L) N L) m n U .

0 185.0 0. l/a2.2/5 1.157 6. T1/T2 and T2/@ versus a for lightning and switching impulses.2/200 0.2/5 1 . R2 is also divided into n parts connected across the stage capacitors.06 3.025 0.37 M i n (Cl /c21 Max(C1/C2) - 150 12s 100 75 50 / T 2 25 0 (1 <a < 12).15. The resistances RI and R2 are incorporated in the individual stages.2 284. 1 1 ~ Figurel6.llb Value measured 1.2/50 1.21200 250/2500 3.44 0.and aTfor Different Standard Waves .2/50 1.73 0. A typical modification of a Marx circuit is shown in Figure 16.15 0.405 0.0 1 . .7 Limiting Values of C1/C2for Different Standard Waves Circuit Figure 1 6 .0054 40.49 3. These modifications save the space and cost of the generator and result in smaller resistors and higher voltage efficiency.l/al.57 T1/ T2(PS) 1.48 68.8 1.19 250/2500 0.381 1. Rl/nis connected in series with the gap C of each stage.

tiuu I40 0 -2 4 6 8 10 14 12 16 18 20 22 24 26 eft) emax 0. .5 emax " " R1LARGE R1 SMALL ure 16.14 Effect of varying the inductance across the output terminals on the generated impulse voltage wave shape e(t).

6. external part of front resistors. 1 . M. regulating transformer. R. R l / n front . 2 . earthing device. 3.R resistor. triggermer. ing unit. HV transfor5. 4.R . motor for control of sphere gaps. measuring resistive divider for thecharging voltage../n. l 5 Multistage impulse generator incorporating the wave-shaping resistors within the generator proper.Test 3bject . . rectifier. discharging resistor. tail resistors.

16.4. trigatron gaps have been in common use.16. but.. The trigatron gap comprises a high-voltage spherical electrode and an earthed main hemispherical electrode. .The gap spacing is chosen such that the breakdown voltage of the gaps C is slightly higherthan the charging voltage V . The trigatron is included in a pulse circuit such as the one shown in Figure 16. Closing the switch S produces. When the impulse generator is to be discharged. as proposed by White in 1947. the gaps G are made to spark over simultaneously by some external means.16). A trigger electrode with the shape of a metal rod is located inside and insulated from the main electrode by an annular clearance (Fig. 16. and the stored gross energy (HylthCavallius.. the voltage across them being only slightly lower than their breakdown voltage. thus no significant discharge takes place through the charging resistors R.2. a voltage 4 + main gap 1st stage Earthed main electrode Impulse generator tripping circuit using a trigatron spark gap.1 Control of Multistage Impulse Generators In multistage impulse generators. The discharge time constant ClRl is very small (microseconds) compared to the charging time constant nCIR. Thus all the capacitors are charged to the voltage V . Thus the impulse is generated at the instant a triggering signal is received by the lowest gap. the spark gaps are arranged such that sparking of the lowest gap induces sparking of the other gaps. with the main gap between them. Thus all the capacitors get connected in series and discharge through the wave-shaping resistors Rl and R2 into the test object. Up tothe 1 9 5 0 ~ the ~ lowest gap was a three-electrode gap. Multistage impulse generators are usually specifiedby their total output voltage. 1986). between the trigger and the main electrode. since then. the number of stages n.

a 7408 AND gate and a power-transistor amplifier (Fig. They may contain oscillatory components with a frequency ranging from a few hertz to a few kilohertz (Chapter 14). 16. an electronic circuit was developed. (ii) a triac to control the switch. Switching surges are accompanied by energies much larger than those of lightning impulses. This breakdown initiates the breakdown of gaps G of the other stages of the multistage impulse generator (Fig.h-Voltage Generation pulse sufficient to break down the annular clearance. When the contacts are remotely separated. to conrol the firing of the auxiliary generator and therefore the main generator. The latter is controlled electronically by a triac circuit (Abdel-Salam and ElMohandes. an inductance L is connected in series with the conventional impulse gen- . The circuit consisted of: (i) a controlled switch connected in series with the spark gap of the auxiliary generator. The ionization of the discharge and the disturbance of the field inthe main gap causes its complete breakdown. A trigatron gap connected across the load can be similarly triggered at the appropriate subsequent moment to produce chopped impulse waves. The switch S can be replaced bya synchronizing circuit so as to generate the impulse at the required instant with respectto the power-frequency AC. thus enabling AC and impulse high voltages to be superimposed on a test specimen as required.).17).C.17). Thus the instant of generating the impulse can be controlled. the tripping pulse for the trigger electrode of the trigatron gap is obtained from an auxiliary impulse generator (Fig. including a 555 duty-cycle oscillator.15). Theyhave front risetimes upto severalmilliseconds and considerably longer tails. In a recent work. Impulse generator circuits are modified to givewave shapes of longer duration by a proper choice of the front and tail resistors. In order to produce unidirectional damped oscillations. The switchingwaveformsexperiencedinpowersystems are not unique. They are grouped as follows: 1. 1991). Several circuits havebeen proposed in the literature for generating switching impulses. the energy stored in the circuit induc\tance is released in the spark. 16. and (iii) a triggering circuit for the triac. 16. which initiates the breakdown of the main gap as mentioned before.D. a crossing-detector (Z. Thus. A modification ofthe trigatron gap was suggested in which the triggering spark is produced by breaking the current in a low-voltage inductive circuit at a pair of contacts embedded in the earthed main electrode.

These oscillations may have a frequency to l-lOkHz. . depending on the circuit parameters. the same as with lightning impulses.C jV CIZV I c= i J . 16. The main drawback of these circuits is that the efficiency of the generator gets reduced to 50% or even less.18). erator circuit (Fig. A sphere gap maybe included in parallel with the testobject for producing chopped waves.

uneconomical. Furthermore. The disadvantages of this technique are associated with the considerable amount of high-frequency distortion in the output waveform. The high-voltage winding is connected in parallel to a load capacitance C2 and the test object. in most cases.19 is charged to a moderate DC voltage (20-25 kV) and then discharged into the low-voltage windingof a power transformer. Variations of the front of the generated impulse were achieved by adding inductances of different magnitudes in series withCl. A preferable method is to energize the transformer from an AC supply for less than half a cycle.Circuit for producing a switching impulse voltage and its test object. switching impulses of proper waveform can be generated across the test object. The capacitorClof Figure 16. A testing transformer is excited by a voltage impulse to give longduration or oscillatory waves inits secondary. the size of the capacitor C l for producing a reasonable output voltage may be large. waveform. output 2. The energization of the transformer is initiated via a thyristor or an ignitron to be triggered at the required phase angle. Cx. Throughautotransformer action. Such a method proved tedious and. The transformer remains ener- .

e. circuit. The switch SIwas replaced by a triac for better control of the firing instant (Fig. Ls) (Fig.t Circuit for producing a switching impulse voltage using transformer energized from a charged capacitor. 16.. several transformer units were arranged in cascade (Anis et al.20b). which consists of an air-cored transformer. Theadvantages inherent in this method are mainly its small high-frequency distortion in the output voltage and the elimination of the energy storage capacitors Cl(Fig. 16. Then the energy stored in the transformer is discharged through the test object. . On an oscillatory current will flow in the LV closing the switch SI. For producing damped oscillations. 3. For generating a much higher output voltage. when LICl = L2C2).19). inducing high oscillatory voltage in the HV circuit of the Tesla coil. a test gized until the supply current passes through a zero value.20a). 1975). C2 is the equivalent capacitance of the high-voltage winding and the test object. Its LV side is connected to the DC or AC supply circuit through a capacitor Cland a series element (Rs.The generator abovehasbeenmodifiedby Sandhaus (19’76). 16. The frequency of the induced oscillations can be made the same by tuning the two circuits (i.. the source of high voltage is a Tesla coil.

According to the IEC (1973).21).0 (a) Teslacoilequivalentcircuit replaced by a triac. switch). The impulse wave shape is defined according to the IEC recommendations. They are also used in electric-arc and electric-plasma studies. generation of impulse currents of the order ofseveralhundredsofkiloamperesfinds application in testing lightning arresters. The waveshapes in common use are the double exponential and rectangular waves (Fig. standard waves are 4/10 and 8/20 p . (b) switch S1 is Lightning strokes involve both high-voltage and high-current impulses on transmission systems. 16. T I is the front duration and T2 is the time to half peak. (S1. . Therefore.

16. the current i can be shown to vary with time according to the relation V i(t) = exp(.y t ) sin ut OL (16.22).1 1 and 16.1 Impulse current ~ouble-exponential (a) and rectangular (b) wave- forms.22is normally used. a circuit such as the one shown in Figure 16.15) is that here the capacitances are much larger and the resistors are much smaller in magnitude. If the capacitor C is charged to a voltage V and discharged when the gap G is triggered (Fig. To produce one of the standard current waveforms. 16. The basicdifference between this circuit and those generating impulsevoltages(Figs.46) .

252/Lc Time to half-peak value T2 22 . If the line is charged to a DCvoltage Y and discharged through the test object of resistance R. where y = R/2L and W = J1/LC R2/4L2.48) where Vris the residual voltage of the surge arrester.vr m (16. A pulse voltage of R Y / ( R Zo) is developedacross the test object. the following approximate espressions are used for determining the waveforms of the impulse current (Haefely.49) (16. The shown in Figure 16. The time taken for the current i to rise from zero to the first peak is T~= -sin" W 1 02/Lc (16-47) If the test object is an ideal surge arrester. The duration Tgo% of the current wave is estimated as + + .22 Circuit for producing double-exponential current impulses.R C L R (test object 1 - ure 16. and C elements into the test object through a sphere gap.the current pulse is given byI = V / ( R Zo). The duration Tlo%c 1. 1970). 1982): Peak current = - v .50) The rectangular impulse wave as defined by the IEC recommendations is (IEC.The equivalent resistance of the test object should be chosen as R 2 4 7 7 7 for underdamped oscillation in the currentwave. 5 m (16. Front time TI2 1.5TgO% rectangular current impulses are generated by discharging an artificial transmission line with lumped L.21. where Zo is the surge impedance of the line.

find: (a) ripple voltage. The supply voltage Vs is 100 kV at a frequency of 50 Hz. So~ution: 4 (a) Ripple voltage A V = C 2 f ( b )Voltage drop = 6V +A V/2 5 x 10-~ = 2000 V = 2.51) where n is the number of L-C stages (Modrusan. If the load current to be supplied is 5mA. 0.7. and (d) ripple factor.10 pF. .05 pF.8kV (d) RF = 3. (b) voltage drop. The supply voltage Vsis 100 kVat a frequency of 50 Hz. respectively. respectively. (1) A Cockcroft-Walton-type voltage doubler has capacitances Cl and C2 of 0.0 kV 0. (c) average output voltage. find: (a) ripple voltage. 1977).5 b~el-~alam T C ) O % 22 - n-l n JLIC (16.05 pF.and C3 of 0.05 x x 50 5x - 50 ” + (0.01 lo6 lo6 2 x 0. If the load current to be supplied is 5mA.05 (c) Vav = 2Vs(max) (6V A V/2) =2~100~--11=271. C2. and 0.01pF and 0. (c)average output voltage.89% + 2 Vs(max) A Cockcroft-Walton-type voltage tripler has capacitances Cl. . ( b ) voltage drop. and (d) ripple factor.01 pF.

01 50 l x lo6 + " -0. (b) voltage drop. If the load current to be supplied is 5mA.01 2 lo6 0. and (e) optimum number of stages for minimum voltage drop.500V = 12. I 1 n(n 1) (a) Ripple voltage A V = -~ fC 2 5 x 10-~ 50 x 0. (d) ripple factor. Solution.76 kV (d) R F = 6V+ AV/2 = 2. The supply voltage Vs is 200 kV at a frequency of 50 Hz.find: (a) ripple voltage.15 x + 12 x 13 2 = 52 X 103V = 52kV .95% 3 Vs(max) + A Cockcroft-Walton-type voltage multiplier has twelvestageswith capacitances all equal to 0.5 = 41 1.05 0.15 pF.1 - x 5x IO-~ x lo6 x 125 50 = 12. (c)average output voltage.5kV (c) Vav = 3 V'(max) (6V A V/2) = 2 X loo& 12.ooov = 21 k'v (b) Voltage drop = 6 V + A V/2 lo6 lo6 " + + 0.7 Solution: (a) Ripple voltage A V = - 2 x lo6 0. l 5 x 10-~ x lo6 x 210 = 21.

Assume 2% for the resistance of the used inductor and the connecting leads.5% resistance (based on transformer rating).59 = 20 stages (4) A 100kVA. Neglect dielectric loss of the cable as well as the no-load current of the transformer.0 = 1 2 ~ ~ ~ 2 Leakage reactance of the transformer -xc .15 x Vs(max) 1 .15 x [2 y 1 +AV / 2 +-. What will be the input voltage and power to the transformer? ~ o l ~ t i oReactance n: of the cable = vc/rc = 500 X 103/4.(b) Voltage drop = 6 V " . = 5 x 10-3 ~ = 20."-+p n2 n 2 6 n(n+lj 4 X [~ 12 12144 1728+---+2 x 13 6 4 1 = 840 x lo3V = 840 kV (c) Vav = 2nVs(max) voltage drop = 2 X 12 X 200& 840 = 5948 kV (e) noptimum ~ fC 200& x lo3 x 50 x 0. 220 V/250 kV testing transformer has 8% leakage reactance and 1. - fC3 __ rl - 5x 10-~ 50 x 0. A long cable has to be tested at 500 kV using this transformer as a resonant transformer operating at 50 Hz. If the charging current of the cable at 500 kV is 4. find the inductance of the required series inductor.0 A.

the belt carries a charge of density 0. the additional series reactance = = I : 125 50 = 75 kS2 Inductance of the required series inductor 75 x io3 = 238.1 m X ps= 0. the charging current is W I = psuw C/m2.5 x I = 0. ~ o Z u t ~Let o ~ .ds : be the surface charge density on the belt.75 x io3 x 220 = 7. XL = Xc. W = 0.875 kS2 base The maximum currentIthat can be supplied by thetransformer = 100/250 = 0.5pC/m2 at a velocity of 10 m/s. 16. The width of the belt is 10 cm. Fig.4A The exciting voltage on the transformer secondary = IR = 0. calculate the charging currentandthepotential difference between the dome and the base.At resonance.7v 250 x 103 The input power to the transformer 7*7x loo 3.5 X 10-6A = 0.1 = 0.5 X 10 X 0.4 x 21.5pA .8 H 2n x 50 Total circuit resistance x lo3 = 21.875 x lo3V = 8. and U and respectively the speed and width of the belt. If the leakage paths between the dome and base have a resistance of 1014 S2. U = lOrn/s. Hence.75 kV The input voltage to the transformer primary = 8.5 220 In a Van de Graaff generator.23.

125 pF.125 x 10"' 20q 2 x 4. R2 = 544 S2.9 .11 for producing lightning impulse voltages are: Cl= 0. shown produ cing positive high voltage.5 x 1 0 . Solution: With reference to Table 16.Principle of operation of theVan de Graaff generator. 0= d w = J0.~ x 1014= 0.1. the charging current is equal to the leakage current and the potential difference between the dome and base is Y = mleakage = 0.95 pS = 0. the excess in the charge collected on the dome is just equal to the charge lost through the leakage paths. Thus.95 x 0.125 x x 1 x lov9 x 360 x 544 24. C2 = 1 nF. Rl= 360 Q . At steady state.987 a=------" - R2C1 1 x 544 x 0. What is the wave generated? Derive an expression for the generated wave if the charging voltage Yo is 100 kV.5 x IO*v = ~ O M V The elements of circuit (c)of Figure 16.987 x 10-4 26.

a = 6.95 x 10-4 6.0088 PS-'.0 15 PS" = 2. .9 x 100 (e-0. T2/T1 = 45.1 . a V0 e(t) = v"----EexP(-alt) exPc"-2 m 01 = 0. a 1 = 0.*. one obtains: T2/0 = 16.l l? The circuit elements are the same as those of example (6).95 ps From Figure 16. a 1 = aa + - m6.*. ~ o Z ~ t iWith o ~ : reference to Table 16. from Figure 16. 0=J ctl4.9 4 0 0 a 2= 4.41) and (16.12.25.25 X 4.02 PS Thus the generated lightning impulse is a 1. one can obtain.015t __ e-2. . a 2 = 4.987 X % m F " T 6.067 PS A s calculated in example (6). TI = 80.9.9+4.1. The charging V .1 X 4.95 x = 0. TI = 50/45 = 1.95 = 80.43 /AS .77t) which is the equation of the waveform of the generated impulse. (7) What is the wave generated by circuit (b) of Figure 16.02/50 ps wave.431120 = 1.12. T2 = 10.42). T 2 = 16. From equations (16.40). T2/T1 = 120.77 PS-' According to equation (16.95 = 5 0 ~ s .1 Corresponding to T2/0 = 10. is the same (= 100 kV).*.62 PS" .*.

125 pF and C .11ismoreefficient than circuit (b).62t) This confirms that circuit (c)of Figure 16.2/50 ps lightning impulse wave.'.125 x lom6 loF9 = 26.1 kSZ + = 1/[1 + (1+ ~ 1 / ~ 2 ) C 2 / C l I = 1/[1 (l 43.3 x 103sz = 53.11 for producing an impulse voltage are: C1 = 0.1)(10-9/0. Find the values of the wave-shaping resistors needed to generate a 1.12.0088t 4 11.1. = 1 nF.1 x lo3 S2 = 26.45 x 100 (e-0.0 = 2500/6 = 416.7 x 10-9 (l-di=TG) S 2 = ~ C l +c 2 a0 (l+Ji=T= 4 x 416. What is the circuit efficiency q? Solution: T ' / 7 7 1 = 2500/250 = 10 From Figure 16.16 1-11. . The capacitive elements of circuit (c) of Figure 16. Determine the wave-shaping resistors RI and S2 required to generate a 250/2500 ps switching impulse. for circuit (c): x=-l+" l ( a2 S 1 2) = 1 4x4 4 = 0. each stage having a capacitor rated 0.What is the maximum output voltage of the generator if .125 = 0. one obtains values of a and T2/0 corresponding to T2/T1 = 10: a = 4.452 1 __ e-4.063 =---(1-4i=-T)= = 53.E(t) = 0.3 ~ S Z c 2 a0 x 416.3/21. The load capacitor is l nF.6 X 11.976 + + X (9) An impulse generator has eight stages. and 125 kV.7 x (l+diTiE3) 0.7 p~ T2/0 = 6 From Table 16.

12.001 x m ) = 440512 R2 = ~ = 3185 st c 1 +c 2 (l+"X)= 6.4 T2/6 = 9.2 41. = 50/1..95 ~ a ~ i output ~ u m voltage = = 0.26 x (l+J1"0 0. i.5 = 5.4 + 1o . the shaping resistors have values of Rl/n and R2/n.67: a 6.15 follows the circuit (c) of Table 16.the charging voltage is120kV? generator? What is the energy rating of the C2 = 1nF = 0.4 x 5.4 x 5.67 From Figure 16.( l .1: a2 ae ' 2 ) ae 1 -6.021 x Per stage.4 x 6.26 x 0. Rl/n = 55s t .e. V 0 R2/n = 398 st The DC charging voltage for eight stages is 8 X 120 = 960 kV- = 0.5 6' = 50/9. = .026 R .02 x (1 - = 0.95 X 960 = 912kV .26 PS The impulse generator of Figure 16.001 pF T2/T. one obtains values of a and T2/6' corresponding to Tz/ TI = 41.~ 0.d ~ ) = c 2 6.

.784 x 6.R1 ==: 70 X 12 = 840 Q .82.15 followsthe circuit (b) of Table 16. front and tail resistances of 70 Q and 400 Q.R2 = 400 X 12 = 4800 Q The generator of Figure 16. "- R2C1 4800 x 0.*.77 PS .001 pF Rl/n = 70.*.*. 0= d w = JO. each having a capacitance of 0.45/25 = 1.02 X 10-6(960)2 = 9216 J = 9. Determine the equation of the generated impulse current if the time to the first peak is 8 ps.'. T 2 =C= Thus the front and tail times of the generated impulse are 1.77 ps and 44.1. from Figure 16.001 x x 840 x 4800 = 6.12. nC1 = 0.'.125/12 N 0. find the front and tail times of the lightning impulse produced.C 1 L== 0. = 4. R2/n = 400. .01 x lom6x 0. respectively. .125. n = 12.Energy rating = & v. one can obtain.125 pF.82 Corresponding to T2/0 = 7. T2/TI L== 25.The charging voltage is 25 kV.22 kJ (10)Animpulse generator has 12stages.45 ps.01 x 2 ~ 0 2 x 0. T I L== 44.35 ps a = . 7 X 6.01 pF C2 = 1OOOpF = 0.35 X low6S = 6. = 4 X 0. respectively. If the load capacitance is lOOOpF.35 = 44.36 x a = 4.45 PS . (l 1) A double-exponential impulse-current generator has a total capacitance of 8 pF and a total inductance of 8 pH.

Vol XXII.3 x 1 0 ~ ~ " V I(t)=exp(yt) sin wt UL - 25 x lo3 exp(-12.3 x 104t)sin(0. . Dorf RC. Basel. MA: Addison-Wesley. Geneva: International Electrotechnical Commission. Reading. Publication BE-701. Haefely. 1970. Switzerland: E. Switzerland: E. 1953. Auget M.02 x lo6S-l 0. Lanovici M.8 x lom6=-sin 'cox 8 x U Solve the above equation iteratively to get 0 = 0.97 s-2 8x R2 x 8 x low6 4 x 8 x 8 x 1 - In equation (I 6. IEC. IEEE Trans Pas-94:187. Meas Sci Techno1 2: l 113. High Voltage Laboratory Planning. Hylt~n-CavalliusN.02x lo6$) 0.Direct Current 1:122. 1982. Haefely Manual. 1982.25 x 103exp(-12.3 x 104t)sin(0. 1986. Switzerland: Presses Polytechniques Romandes.47): 1 . Modern Control Systems. 1991. Nonlinear Resistor Type Arrester for AC Systems.46). Basel. El-Mohandes M. Lausanne. Haute Tension.97/(2 X 8 X = 12. Haefely. 1975.02x 106t)A Abdel-Salam M.02 x lo6 = R = 1.~ o Z ~ t iThe o~: time to the first peak is expressed by equation (16. Haefely. Publication 99-1.1986. Train D. Anis H. Giao Trinh N.02 x lo6 x 8 x = 156. y = R/2L = 1. Felici NJ.

Geneva: International Electrotechnical Commission. Sandhaus S. Bull SEV/VSE 68: 1304. IEEE Trans EI-12:130. Revue mensuelle suisse d’odontostomatologie 86:1312. Publication 60-2. 11-Test Procedures. Philp SF. 1976. Analog Electronics (an Integrated PSpice Approach). 1982. High Voltage Engineering Fundamentals. 1997. 1978. Price TE. Zaengl WS. 1973. West Germany: Friedr Vieweg & Sohn. New Delhi: Tata McGraw-Hill. High Voltage Test Techniques. 1977. Oxford: Pergamon Press. 1977. 1984. High Voltage Engineering.el-Sala IEC. Modrusan M. An Introduction to High Voltage Experimental Techniques. Kamaraju V. . USA: London/New York Prentice-Hall. Kind D. Kuffel E. Naidu MS. Nouveaux aspects de l’ergonomie.

one of the spheres is earthedand the other isconnected to high voltage. for a particular test circuit. determined by sparkover between the spheres. together with their modes of usage.Egypt This chapter describes various methods for measuring high voltages and high transient currents.Giza. a relation between the peak voltage. The procedure is to establish. The sphere gap arrangement is shown in Figure 17. 7 . 1973). A horizontal sphere gap arrangement is sometimes preferred at lowermeasured voltages.1. When sphere gaps are used in high-voltage measurement. and the reading of a voltmeter on the primary or input side of the high-voltage source. Table 17. Cathode-ray oscilloscopes are not included in the table. Also described are means of measuring electric fields nomally associated with high voltages. as their versatility with their probes and accessories make them suitable for almost all kinds of measurements. This relation should be within 3% (IEC.High-Voltage Measurements CairoUniversity.1 lists the most common methods of measuring high voltage.

0 I . m 5 a .nis x x x x x x x X % X m x e a. C X % * ” C 2 2 5 cn m Y- O W E 5 a .

1960).Figure 17. 1984). Standard values of sphere diameter are 6.l) where k is a factor related to the relative air density (RAD) 6 described in Chapter 5. 25. Table 17. havea significant influence on breakdown voltages.25. 75. The breakdown voltage Vsis related to itsvalue under normal atmospheric conditions by Vs = kVn (l?. 50. and 200 cm.5. Temperature and pressure. 12. 150. however. stipulated by international standards (IEG.2 gives the clearance limits in sphere gap arrangements.1 Vertical sphere gap.3 gives the relation between the RAD and the correction factor k. 100. . The effect of humidity is to increase the breakdown voltage of sphere gaps by up to 3% (Kuffel and Zaengl. Table 17.

90 0.09 1. The arithmetic mean value of the rectified current as indicated by the instrument is given by (17. 1960). Table 17.80 0.25 10-1 5 25 50 75 100 150 200 7D 6D 5D 4D 4D 3.10 0. except for the bracketed figures. where the accuracy is about f 5 % .76 0.2 and is made up of two diodes. 17.70 k 1. The accuracy of the values of Table 17.3 Correction Factor for Atmospheric Pressureand Temperature 1.95 0.72 1 .2) T If the voltage is symmetric about zero and its peak value is V .Anis le 17.85 S0.95 0.90 0. and a milliammeter that can be recalibrated for peak voltage.05 Under impulse voltages.0 0.5D 3D 30 of A 9D 8D 7D 6D 6D 50 40 40 8s 8s 7s 6s 6s le 17.2 Clearances around the Sphere Gap (Fig.05 1.4 is generally within 3%.O 1.81 0. a properly chosen capacitor.75 0. the voltage at which there is a 50% breakdown probability is recognized as the breakdown level.4 lists the standardized disruptive discharge voltages from the results of a large number of international experiments (IEC.1) of B 14s 12s 1 o s Sphere diameter Minimum value Maximum value Minimum value d @m> of A Up to 6.86 0. then . The measuring circuit of a peak voltmeter is shown in Figure 17.

A peak voltmeter may be of either the analog or the digital type (Malewski and Dechamplain.I C Figure 17. 1980). r = 2fCV Therefore. (17.3) (17.) The current waveform should ideally be monitored on an oscilloscope to determine whether there will be more than one zero crossing in a half-cycle. .2 Peakvoltmeter circuit.4.

.nis N W N N W N W e 0 1 T- CO $1 N 0 0 0 W CV CO m P- o ) a . . Q UjCOO *v)b T- Y n ?tt: cd " . T- o ) v) N CO c - v! b v) 8 52 v! $1 b c n 0 C O C O 0 N v ) W O N N N N N : CO00m0 COoT-cum V " " T-NNNN .

W P .T .T .P 0 0 0 0 0 0)T-r-Q)P-h- v) v) m PS .O b NcObQ) n m b b S W PV n P .7 - P- W Q) 0 0 0 0 0 P-wCVmPT-mv)v)w 0 0 0 0 0 W N W T . 0 0 0 0 0 .N W T-NC9d-b 0 0 0 0 0 0 m PT-T-T-T-F 0 b v)moP-* Q) 3 m 0 0 v) m 00 m -wwww N T .m N l-wr-v) n m * b V S P- v)ov)ov) CVmmbb 0 o o v ) o v)wr-P-a.T .0 * v) 0 0 T .S .T .V T - 0 W Q) T- S-m*v)v) T- wcOQ)Q)o ? " C v) m 0 0 v) " " V 0 v) Q) T-T-FFT- -c-\nnn 0 0 0 0 0 Pn n n C 4 0 Q ) w m o wwP-Q3Q) 0 0 0 0 0 r .

td CV* 1 : 1 0 U) r( c I ' I : I 0 0 T rrI 1 : 1 m b ( c I cl 0 cn l ( 0 I : r - 13 rd 13 n " ! a rd . c CV 1 c _ 0 0 r d m u CV* S E CVU -.f ) l za 0 0 .nis 1 o c .

frequency df. the error in the measured voltage dV is a function of errors in the current di. It can be proven that dV=di+df+dC (17. The force is created by the process in which a change in stored electrostatic energy is converted into mechanical work. and capacitance dC. it appears that. Oneisfixed and the other has a very small movable part that is restrained by a spring. the electrostatic voltmeter is seen to be made up of two parallel plates.4).575 Referring to equation (17.5) Impulse peakvoltmeters are different in that they contain active circuits and a memorydevice and possibly a storage oscilloscope (Hylt6n-Cavallius. Referring to Figure 17. The force of attraction F(t) created by the applied voltage causes the movable part-to which a mirror is attached-to assume Figure 17. 1988). .in addition to the instrument error.3 Electrostaticvoltmeter. An electrostatic voltmeter utilizes the force existing between two opposite plates.3.

if any.7) The actual force is the arithmetic mean of expression (17. . The voltage on the secondary side is closely proportional in amplitude to and almost in. the stored electrostatic energy W in the system will be W(t) = (1/2)CV2(t).6) where d S is the change in S.8) where T is the period of variation. the separation between the plates (Fig. the factor dC/dS can be simply evaluated by recalling that (17.8) relates the force to the m s value of the applied voltage V: (17.7). d W(t) = -F(t) d S (17. Voltage transformers are used to measure high AC voltage accurately. phase with the voltage on the primary side.1l). Assuming the capacitance between plates to be C. A change d W(t)in stored energy will be faithfully converted into mechanical work. Equation (17. An incident light beam will therefore bereflected toward a scale calibrated to read the applied voltage magnitude.10) where A is the area of the plate and Therefore.11) If V is in volts and A and S2are in m2 the force is in newtons. the factor Ac0/S2is used to control the range of measurement. Therefore.76 nis a position at which a balance of forces takes place. The attraction force as a function of time is thus (17. Two alternatives are. (17.9) In the most common attracted disk type of electrostatic voltmeter. 17.3). the force will be given by is the permittivity offree space. According to equation (17. given by (17.

the voltage drop across C .f)~(C. preferred at very high voltages. C. A capacitive voltage transformer consists of a capacitive divider with Cl and C2 as the high and low voltage arms. namely.77 however. + High v3 Voltage . By adjust~ent of the inductance. to equal 1/(2~. respectively used in conjunction with a conventional auxiliary transformer which further stepsdown the divider output voltage. which may consist wholly or partly of leakage inductance of the auxiliary transformer. typically about 10 kV. capacitive and cascaded voltage transformers (IEEE Committee. 1981). to the desired secondary value. due to the current drained from the divider is largely Compensated so that the overall ratio is nearly independent of burden and is the product of the divider and t r a n s f o ~ e r ratios.).

A resistive potential divider is usually employed for the measurement of direct voltages.57 Anis In cascaded voltage transformers the primary winding is distributed over a seriesofcoilswound on separate cores which are mountedinoil-filled porcelain containers. Z1 and Z2. dependingon the type of voltage to be measured. stacked inseries. An impedance can be used in series with a microammeter or a milliammeter for the measurement of high voltages. if the impedance is a resistor. A potential divider consists of two impedances. the current is in phase with and faithfully represents the voltage. 1978). the current through the instrument will be proportional to the applied voltage. the components that constitute the impedances are referred to as the high. Neglecting the impedance of the instrument. however. The instrument maybe replaced by an oscilloscope if the voltage waveform is to be recorded. either inseries or in parallel. each of which neednot be insulated from its core for more than 100kV. The power-frequency voltage is thus uniformly distributed over a number of coils. but sometimes a combination of resistors and capacitors.and low-voltage arms of the divider. to which the voltage to be measured is applied. Evidently. connected in series. .Thesecondarycoiliswoundon the lowest core. Wire-wound or thin-film resistors may be used. High-voltage dividers generally consist of either resistors or capacitors. Wire-wound resistors are usually preferred for their superior stability in service and their extremely small temperaturecoefficient. is used. a resistive/capacitive potential divider will then be more suitable (Kind. If. the ripples on the DC voltage are to be recorded as well. O 1 %/K. Freedom from corona discharge and improved cooling may be achieved by immersing the resistor in insulating oil. Connection between the low-voltage arm and the measuring instrument must be made through a shielded coaxial cable to avoid the adverse effects of stray capacitance between that connection and the highvoltage a m .which reaches f O .

The influence of earth capacitance Ce on the effective impedance can be derived by considering the general case of a stack earthed at one end and having inherent series resistance Z = R 1 and admittance to earth Y =jmCe. both assumed uniformly distributed along its length (Fig. according to (17. Therefore. amounts to ensuring that (17.which generally possess resistiveand reactive components.13) Therefore.16) I where .14) The high-voltage arm will consist of many resistor elements stacked in series. As the length of the stack increases.j& Under AC conditions the divider ratio.The divider resistive arms should-in the case of alternating voltage-be looked upon as impedances Zi and Z2. this phase shift should vanish. is (17.15) (17. This.13)..5).12) The output voltage V2will be shifted in phase fromthe measured voltage Vl by (17. 17. so does its capacitance to surrounding conductors. and Z 2 = R2 +. The currents io and il at the earthed and high-voltage ends of the stack. in magnitude. respectively. for the output voltage V2 to represent VI faithfully. due to an applied voltage V are then (17.

the screen being one piece (a) or subdivided (b). .5 (A) Unscreened high-voltage resistance.WV f' Supply V i. (B) screened high-voltage resistance. V Seen from supply side HV il the Seen from the earth side HV ii Figure 17.

19) When In(L/r) is greater than about 3.15) and (17.17) When this resistance is used as the high-voltage a m of a resistive divider whose low-voltage-am resistance is R2. It is basically accomplished by surrounding the high-voltage resistor by a conducting screen maintained at the mean potential of the resistor (Fig.2. 17.of stray capacitances. 17.1 The effective impedance Z. the effective divider ratio is thus (17.7. Ce = 17. resistivedividers arenot &itable for high-frequency alternating voltages. 2 ln(L/r) PF (17. if earth is looked upon as a coaxial cylinder of radius L (m). it is also not much affected by the shape and disposition of the earthed surface since. the capacitance depends primarily on L and not on r.7. that is. 1754).18) that the divider ratio is sensitive to frequency.5I3). Capacitive currents will flow between the screen and the resistor in one direction within the upper portion of the resistor and in .16).2.2 1111. its capacitance is only slightly different. Therefore. giving at the earthed end: Z1 EX RI at the high-voltage end: ZI EX RI (17.( V/ioor V / i l ) of the stack may be derived by expanding the hyperbolic functions in equations (17. high-voltage arms of resistive dividersmay be considered as a vertical column of length L (m) and diameter 2r standing on a grounded plane (Fig.1 PF (17.18) It appears from equation (1'7.20) Screened Resistors Screeningis the processofcanceling or neutralizing the effect .1 Unscreened Resistors Most high-voltage resistors-also. For this configuration the equivalent stray capacitance to earth is given approximately by Ce = 111L 2 ln(L/r) 1.

Ce 20 pF/m in practice. with Ce not greater than Cl.7. The two currents io and i l . if the series capacitance of the stack is Cl and the stray capacitance to earth is Ce. then. mica. lS). 17. or polystyrene dielectric). It is preferred. It will. In this case. viewed from either end of the resistor. normally take the form of a stack of identical cylindrical units.which may either be introduced deliberately to avoid the accumulation of a random charge on C2 or be inherent in the measuring instrument.1 Unscreened Capacitors Unscreened capacitors. power losses and stray capacitance to earth.nis the opposite direction within the lower portion. . Resistive potential dividers suffer from two main drawbacks.3.17). substituting CJC1 for g2 in equations similar to (17.19). The low-voltage capacitor Cz will normally consist of a fixed unit of low loss angle (air. to divide the resistor into identical smaller units. . At 50 or 60 Hz the practicable voltage limit for a high-precision screened resistor is about 100 kV. High-voltage capacitors Cl can be either screened or unscreened. This action poses less risk of flashover between the resistor and the screen extremities. which may be used as the high-voltage component of a capacitive divider. 17. the approximate effect of Ce is to decrease the effective capacitance at the earthed end by Ce/6 and to increase it at the other end by Ce/3. will see an effective impedance 2 1 of value z 1 2(17.21) Rl(1 -jWR1Cs/12) which isintermediate between the two values ofZl in the case of unscreened resistors given by equation (17.5I3). be shunted by a high resistance R2. particularly at high voltages and high frequencies. They are limited only by their internal inductances or the dielectriclosses of their components (Fig. each contained in a screenthat is maintained at the mean potential of the unit (Fig. Capacitive potential dividers are therefore more suitable to use with AC voltage.6a). According to equation (17. 17.17) and (17. in general. These factors limit their use to voltages below100kV at50Hz and evenlower voltages at higher frequencies.which also applies here. so that the correction due to it in any particular case can be estimated. however.

6b.V.(b) typical screened capacitor.- L. may be used as the high-voltage component.7. capacitors with parallel-plate electrodes can be used and screened by being enclosed in a metal case.6 (a) Capacitive voltage divider. At higher voltages a capacitor with coaxial cylindrical electrodes. electr High Voltage Figure 17.3. 17. When filled with carefully dried gas they have almost zero power factor and a highly stable value of capacitance.2 Screened Capacitors For voltages up to about 30 kV (peak). asthat shown in Figure 17. The outer cylinder is flared at the .

Higher voltage gradients can be attained by increasing the gas pressure (compressed-gas capacitors). With impulse voltages. and r2 the radius of the inner cylinder (m). If a unit step voltage that is defined by } (17. VI(S) and U($).24) (17. the capacitor can function with voltage gradients at the inner electrode surface of about 14kV/cm peak without partial discharges.718). The output with any other input voltage vl(t) can be found accordingly. The capacitance Cl is calculable from the electrode dimensions: (17.25) In the time domain the output voltage would be Equation (1'7. respectively. rl the radius of the outer cylinder (m). the Laplace transform of the output voltage may be given by V26) = Vl(S) W ) S (17. Instead of exploring the frequency characteristics of a divider.23) is applied to a divider.the complete waveform isto be recorded. a potential divider for impulse voltagesshould have a good impulse fidelity.nis ends to avoid discharges. Otherwise.25) can be solvedanalytically whenever u(t) takes an analytical f o m .22) where I. Therefore.26) . it is more common to examine its response to a step voltage applied at its high-voltage terminals. The ~ a x i m u m stress on the dielectric occurs at the surface ofthe inner electrode and. At atmospheric pressure and with smooth clean electrode surfaces. isminimal when the ratio rl/r2 is equal to e (= 2. 0th the divider and the connection leads must be considered when this impulse fidelity is to be assessed. fora given voltageand size of outer cylinder. numerical methods can be invoked to evaluate v2(t) on the basis of the superposition theorem: v2(t) = u(O)vl(t) + vl(Qu'(t h) dh (17. is the effective length of the LV electrode (m). its response (output) is u(t). If the Laplace transform of the input voltage v1(t)and of the step response u(t) are.

On C I U c Y- 0 ? t % ure 17. . (b) for resis. the wave shape of the response u(t) should be as close as possible to the applied step voltage vm1(t). (c) for capacitive divider with inductive high-voltage arm. 17.tive divider with inductive low-voltage arm. A reasonable measure of the faithfulness ofthe response isthe total area enclosed between the real and the ideal responses (Fig.For faithful reproduction of q ( t ) . This area will have the dimen.7).. sions of time and is thus referred to as the response time of the divider.7 Possibleresponsesof impulsedividers: (a)for resistivedividerwith inductive high-voltage arm.

16) still hold.15) and (17. of frequency (17. With an applied voltage step of amplitude V . where 2 = RI and Y = sCe ( S being the operator d/dt).7. results. the considerations in Section 17. two criteria are used simultaneously to evaluate the response faithfulness: 1. These are identical with the time constants derived from equation (17. Unless the leakage resistances are ensured to be inversely proportional to the capaci- + + . In a capacitive divider the high. equations (17. its time constant is L2/R2.17). The response should settle down to the correct value in a time much shorter than the rise time of the voltage wave to be measured. respectively.2 relating to the effect of stray capacitance to earth (Ce) of the high-voltage arm of a resistor divider applywhere impulses are concerned.and low-voltage arms can be assumed to have capacitances C1 and C2 in series with lead resistances rl and r2 and inductances L 1 and L2 in series. and also leakage resistances R I and R2 across C1 and C2. thus reducing the total response time to zero.27) which is damped out at a time constant of (L1 L2)/(r1 r2). The response of the divider to a step voltage will be an oscillation. however. Only a small reduction in the final settlement time. about unity.8 shows the case where L2/R2 = CeR1/6.58 nis the basis of this representation. esistive Dividers With unscreened resistive dividers. Curve c of Figure 17. 2. The total response time (net enclosed area) mustbe as small as possible. the response time of the low-voltage arm should be made equal to that of the high-voltage arm. When a much shorter divider response is desired. Since the low-voltage arm is mainly inductive. The calculated values of the response times of the resistor are (CeR1/6) and (-CeR1/3) at the earthed and high-voltage ends. A typical divider response is shown in Figure 17. An upper limit on the value of RI should be imposed such that the response time of the high-voltage arm is made much shorter than the rise time of the measured impulse voltage.8 together with its voltage constituents.

A total series resistance value is aimed at which critically damps out the oscillation.e. (c) divider output v when L2/R2 r=t RjCe/6. and + + . R 1 C1 = R2C2). after some time. (b) current at high-voltage end i. it should be nearly 2 d(Ll L2)(CE C2)/C1 C2. and under lowfrequency voltages as a capacitive divider. to a value of R2C2/RICI.'T Response of resistive divider to a unit step voltage:(a) current at earthed end io. The resistances should be well distributed along the divider. Mixed capacitive/resistive potential dividers are also used where the introduced resistors. are connected across the capacitors C .. tances (i. Such a mixed capacitive/resistive divider acts under high-frequency voltages as a resistive divider.the eventual settlement output would drop. R Iand R2.. Additional damping resistances are usually connected in the lead on the high-voltage side of a capacitive potential divider.

Let the input voltage be where k is a constant slope.26) is (17. respectively. Their response is determined initially by the capacitances is made equal to R2C2.26) can be used to predict the output.28) which may be expanded to The difference betweenthe ideal and actual normalized responses v1 and v2. . equation (17. the difference above becomes a constant proportional to k. onse Time of the The response timemaybedefined. as that time beyond which the difference in amplitude between the input and output voltages remains constant. If RICl response will be ideal except for an initial effect of the inherent inductance. The output voltage according to equation (17. (17.the overall and ultimately by the resistors.nis C2. Therefore. is then v&) 1 v2(t) = kf[1 u(h)]dh 0 For all t above a certain value. respectively. in view of linearlyrisingvoltages (ramps). If a linearly rising voltage is applied to the measuring system.29) as defined earlier.

Matching is achieved by making F ~17. this expressiongets simplified to the form . the cable should be terminated at one end.. = 1. The cable is thus essentially compatible with a resistor divider. When the electromagnetic waves are required to travel along an actual cable./m. VN- 3 x lo8 mls where Er is the relative permittivity of the insulation surrounding the conductor and is the relative permeability of the conductor material. p .which for low-loss cables and very high frequencies is a pure resistance. Generally. le Termination To avoid errors due to reflections at its ends.0 for most of the conductors used in cables (IEC.9 ~ ~ e a~ surin rg cable ~ connectionto a resistive divider. to which it maybe connected as shown in Figure 17. For a loss-free cable. 1962).9. the velocity v of propagation of the where L and C are the inductance electromagnetic wave is equal to and capacitance per unit length of the cable. by a resistance equal in value to its characteristic impedance ZQ.5 The output of a potential divider is connected to the oscilloscope via a coaxial cable. . or preferably both ends.

the response of the divider will be practically the same as if R I and R2 were absent. It the mixed divider has its resistances and capacitances in series. the cable arrangement can be such that the low-voltage arm is essentially placed Cland at the output of the delay cable (Fig. with a unit function voltage applied to the divider.nis Normally. The low-voltage-arm resistance -R2 isnowshuntedby a resistance R3 in serieswith the input impedance to the cable Zo. the cable input resistance S 3 is chosen witha value = Zo. its far end is not shunted. The time constants Rl Z0(C2 Cc) should be made equal to ensure equal initial and final response values. This is justified by the fact that the initial portion of the divider's response is controlled by the capacitive components. When Cl C2 = C3 Cc. and is absorbed without significant reflection when it returns to the input end. the cable capacitance Ce is considered and its far end is loaded by Zo and C3 in series.and low-voltage capacitances Cland C2 are shunted by resistances R 1 and R2 and the time constants RICI and R2(C2 C3 Cc) are adjusted to be equal and also large compared with the cable delay time. the divider ratio becomes (17. a voltage CI/2(C1 C2)is initially injected into the cable. the circuit + + + + + + + . Thus the voltage at the open end jumps to C. is doubled by reflection at the open end. R2 is much smaller than RI. Therefore. according to (17.31) where which. When this condition is not valid. Therefore./(C. In cases where the divider capacitance is intentionally made small.30)' becomes (17. C2). Thus.Thearrangementisgenerally suitable whenever the cable capacitance is much smaller than C2. If the high.32) With a capacitive divider. the same cable connection is used. 17.loa). the initial and final values of the response at the oscilloscope end are equal and a flat-topped overshoot occurs. With a divider consisting of resistance and capacitance in parallel.

of Figure 17. the contents of the memory can be transferred to a computer for analysis or be displayed repetitively on the oscilloscope with the aid of a digital-to-analog converter (DAC). To minimize the sampling error. The larger the number of bits. which represent the magnitude of the input at these particular instants of time. Digital instruments have significant advantages over the older analog technology.10 Measuring cable connection to a capacitive/resistive divider. The number of bits defines the maximum amplitude resolution of the scope. 16. The sampling rate defines how many times per second the input signal amplitude will be converted to a digital code by the ADC. The upper limits of the performance ofa digital scope are specified bythe sampling rate and number of ADC bits. the sampling rate should be as high . The samples are then converted into digital codes. high-speed 8.10b can be used.1 Figure 17. and 32 bit digital scopes are available. in which matching is ensured at the cable's sending end by making that is. These codes are stored into memory locations proportional to the time at which the sampleswere taken. Once recorded. A digital recorder has an analog-to-digital converter (ADC) where the input signal is sampled at discrete-time intervals as defined by a clock. the greater will be the digital scope's resolution or precision.

as well as statistical quantities from repeat measurements. This record can be manipulated in a c o ~ p u t e to r calculate rise times. Modern digital scopes have ma~imum sampling rates up to a few CHz. In addition. over the frequency band con- " " Y ' Y X 1 Resistive shunt with the shape of a coaxialtube. having the data in a digital file enables digital signal processing applications such as spectral analysis. where R and L are the effective values ofits resistive and inductive components. Thus. ismade. 3. 17. which can be displayed repetitively. The following advantages of using digital-rather than analogrecording can be outlined: 1. ost digital scopes contain two or more independent ADCs. low-ohmicshunts. Two methods are generally used to measure transient currents: the use of resistive. A permanent record of the transient.. . With a suitable sampling speed and number of bits.11) into which current I isfedvia current terminals X and X' and from which a voltage V is tapped via potential terminals Y and Y'.t e ~ i n a ldevice (Fig.Anis as possible. The shunt should therefore be designed so that.g. and the measurement of currents by their inductive effects. digital recording ismuch more accurate than that obtained with an analog oscilloscope. The impedance V / I under AG conditions can be expressed as R +jwL. current and voltage) can be recorded simultaneously without chopping or alternating on repeat events.low-ohmic shunt is designed as a f o u r . A resistive. two or more signals(e.

an incident step current I is assumed. for which it can be assumed that p . Coaxial tube shunts may have resistances as small as microohms and are used mainly for measuring large currents (Hebner et al.12a.33 mmfor constantan and manganin. = 1. = d m . These current sensors have recently been superseded by Hall-effect transducers (Berkebile et al. The former is used in the secondaries of current transformers and other applications where the transient current is relatively limited in magnitude. Thetime-varyingmagnetic field induces a voltage at the terminals of the coil which is proportional to the rate of change of current with time: v=M- di dt (17. + +R + Z e-t/T (17. depending on whether it is a wire or a ribbon. The response of the shunt is sketched in Figure 17. The skin depth dsis calculable from the frequency f. 1977). respectively. 1981. The use of resistive shunts to record transient currents can be represented by the equivalent circuit shown in Figure 17. The voltage appearing across the impedance 2 is what a recording oscilloscope will show..h-Voltage M e a s ~ r e m ~ ~ t s ~~~ cerned in any measurements in whichit is involved. The impedanceZ is the input impedance of the coaxial cable and its matching resistances.. Two f o m s of shunt are in common use: the coiledbifilar and the coaxial tube shunt. To evaluate the adverse effectof the inductive component of the shunt. R is substantially equal to the DC resistance and o L / R < < 1.34) In this method a toroidal coil known as the Rogowski coil surrounds the path of the current to be measured. The timefunction of the voltage can be shown to take the form v(t) = IZ(- R R+Z where T = L/(R 2) is the time constant. 1982). and relative permeability p . Tokayo et al. It amounts at 1 MHz to 0. Coiled bifilar shunts normally have resistances in the order of a few ohms.35) where M is the mutual inductance between the sensing coil and the measured current path. of the material by theequation d. to be less than the nominal depth to which alternating current at the upper limit of' the frequency band will penetrate a slab of the same material. .35 and 0. which are available in wire or sheet from and are nonmagnetic.12b.Both these conditions require the diameter or thickness of the resistive element.resistivity p ..

which is known as the Hall voltage. 7. 1988). For a constant magnetic field.Anis i: 'OSC ure 17. ffect ~ ~ r r eTrans nt If an electric current flows from left to right along a thin strip of metal or semiconductor and if a magnetic field acts vertically upward. In the Hall-effect current transducer the current-carrying conductor is surrounded by an iron-cored magnetic circuit that produces a magnetic field into the air gap. the Hall voltage is proportional to the current. Current probes based on the Hall effect are manufactured to have a bandwidth extending from DC to 50 MHz and more (Holt.11 Field measurements play an importantrole in the construction and problem solving of equipment using high voltages or subjected to high electric fields . The Hall voltage is amplified and measured or recorded on an oscilloscope. the electrons will be deflected toward its back edge. (b) typical response. A potential difference is produced betweeen opposite points on the edges. This property canbeused to measure high DC and AC and even transient currents.12 (a) Equivalent circuitof a resistiveshunt.

1983). For a probe with an effective head area of A and with an electric flux of coulombs normal to the probe surface. The electrode version shown in Figure 17.36) The capacitance probe is essentially a small disk mounted flush with the surface of an electrode or a screen. as seen in Figure 17. In either case the device measures the displacement charge induced on the probe surface due to the electric flux density.& PROBE ELECTRODE F i ~ ~ 17. Most electricfieldmeasuringdevices are based on the principle that the electric field is proportional to the electric flux density D. the electric field is given by + (17.13.13 can be used to determine the field at the ground plane of a rod-plane gap. . By measuring the probe current id. The .37) where is the permittivity of free space. a displacementcurrent results.5 (IEEE Working Croup.13 r e Capacitance probe.When this changes with time due to natural phenomena or by mechanical movement. the field can be obtainedif the measured current is integrated with respect to time. The integration process can be performed automatically by inserting a capacitor Cm between the probe and ground. The displacement current density is given by Jd - =- ar5 at (17.

its charge is expelled (Stark.39) =EOEA assuming a static uniform field.59 Anis voltage Y across the measuring capacitor is proportional to the electric field. An electric field mill consists of a grounded multivane rotor turning in front of a multisegment sensor plate. which can be calculated from (17. 17. where the effective radius can be approximated by the probe radius and one-half the ring width. When each sensor plate is shielded. " ELECTROSTATIC FIELD E ure 17.14).14 Schematicdiagram of asinglefield mill. The grounded rotor alternately exposes the sensor plates to the ambient electrostatic field and shields them (Fig. 1979). . The induced charge from the displacement current is given by (1 7. The arithmetic mean value of the current for the sensor plates to be exposed and shielded once is then SEGMENT SENSOR -PLATE 4 .38) The effective area A is a function of the width of the annular ring (g).

the peak voltage is approximated as (1 7.43) eoEA(1 + The impedance Z consists normally of a resistance R and a capacitance C in parallel.. the restriction for equation (17. Therefore. If approximated as EO EA $ L (1 7.g.-Volta 3 where f is the frequency of exposure and shielding.40) The function describing the area exposed is between a triangular and sinusoidal waveform and can be approximated by 2 w being the angular velocity. 17.44) and (17.44) > > 1. 100nF) and if the output voltage V is fed into a high-impedance amplifier (e. The principleof operation can be shown by examining the probe with two hemispheres. sometimes called electric dipole sensors. The charge on the sensor is then. This current is the maximum current and can be expressed in terms of capacitance. If < < 1.42) 2 The instantaneous voltage across the impedance Z (see Fig. The charge induced on one hemisphere of radius r is given by . R = 10 MSZ). Free body probes.45) 2c where the field is not a function of frequency. 4= a= A( 1 + sin w t ) (17.45). the results obtained are essentially identical to those of equations (17. If the sinusoidal assumption is relaxed and a pure triangular waveform is used.g. are selfcontained and normally have no electrical connection to ground.45) is satisfied at power frequency. These types of probe generally consist of two electrodes of a specific geometry separated by a small insulating gap. the peak voltage is where the fieldis a function of frequency. if the impedance 2 is a capacitance (e.14) is d4 v(t) = zdt (1 7. where (17.41) sin w t ) (1 7..

The electrical isolation of the measurement system from the system which is being measured. Increased precision and accuracy over conventional electricalmeasurements. 3.. or Faraday. by 3nr2so L. The advantages of using optical techniques in high-voltage measurements are: 1. L.48) This means that the effective distance between the poles of the dipole. SUR Electro-optical and magneto-optical devices are developed to measure electric and magnetic fields.49) The open-circuit voltage across the probe is equal to the product of the electricfield and the effective distance L. 2. The effective area of the sphere. therefore. becomes A = 3nr2 (17. Theeffectsof electromagnetic interference on the measurement systemcanbeavoidedbyphysically separating the measuring system from the higher power system which is the source of the interference.. 4. optical measurement techniques are based on one of three electro-optical effects: Pockels. = ca (17. Ca. The . The stray capacitance CS acts with C2 as a capacitive divider. Ground current effects can be minimized because optical coupling eliminates the wires connecting the two systems. is related to the capacitance of the hemispheres. Normally.47) (17. current.46) from which it follows that the short circuit current between the two hemispheres is i(t) = for P L C fields.nis Q= (1 7. voltage.The parameter to be measured changes the optical transmission characteristics of the sensor. Kerr. and space charge.

h is the wavelength. a commonly used fluid inKerr effect systems. 1986) np n. then equation (17. The Kerr effect at low frequency arises because the molecules in the optically active material tend to align with the applied electric field. is independent of frequency up to about l GHz. is the component perpendicular to the field. and Pockels effects relate to specific interactions between the field to be sensed and the optically active material according to the general expression An=a+/3F+yF2+. if F denotes a magnetic field and a and are the only nonzero coefficients. F is an . Faraday. a light beampolarized in the direction of the applied field and a beam polarized perpendicular to thatdirection will propagate through the material with different velocities. The Kerr effect isthe most relevant effect for high-voltage and dielectric measurements.50) describes the Pockels effect. This field interacts with the optically active material to change the light beam. then equation (17. 7. The basic equation describing the electro-optic Kerr effect is (Thompson and Luessen.51) where np is the component of the index of refraction parallel to the applied field and n.50) (1 where An is a change in the index of refraction of the material. An electro-optical system to measure electrical quantities is shown in Figure 17.. Finally.15 Typical electro-optical measurement system. and E is the electric field. If F denotes an electric fieldand a and y are the only non-zero coefficients. ~I~~~ Polarizer . This alignment produces an anisotropy in the index of refraction. olarizer i Light Source ___ ure 17. and a electric field and a and /3 are the only non-zero coefficients.50) describes the Faraday effect. p.voltage or the current to be measured produces an electric or a magnetic field. then equation (17. B is the Kerr coefficient.15. = A B E ~ (17.50) describes the Kerr effect. Thus. The relative permittivity of nitrobenzene.. . The Kerr.and hence the Kerr coefficient. and y are constants. If F denotes an electric or a magnetic field.

nis In measuring electric fields. For the parallel electrode system used..ely)of incident light transmitting through the dielectric travel at different velocities. the electric field magnitude and direction are constant along the optic path in the 2 direction.. An optical system using linearly polarized incident light is shown in Figure 17. Furthermore. is not measured directly.17. = n/4 from the vertical X axis) with its x and y electric field components having the same amplitude and the same phase. The two fieldcomponents (elx. resulting in a phase shift between the two electric field components (ezx. The applied fields are of higher intensity and may be provided by an external electrode system for the lower frequencies or by a high-power laser for the higher frequencies. and computer image processing techniques. The complete system is less than 50cm long and can be used up to 100kV. 1986) for the measurement of high-voltage pulses is shown in Figure 17. The phase shift @ between the perpendicular and parallel components of the light beam can be shown to be given by @ == ~ Z B E ~ L (17.) of light after passing through the liquid dielectric. highly sensitive cameras. The sensing field is the lower intensity electric field associated with the light beam which is usedto detect the difference between the indices of refraction. The light beam from a laser source is expanded in diameter from 2. Ep.51). as given earlier. Then it is transmitted through the electric field-stressed liquid dielectric in the test cellin the 2 direction. the phase shift is measured betweenthe electric fieldcomponent of the light beam polarizedin the direction of the orienting field.52) where L is the effective optical path length that also includes the effect of fringing fields from the electrode edges. high-quality optical devices. This expanded light beam is transmitted through the polarizer to become linearly polarized light (polarized plane at an angle @ . The dynamic behavior of two-dimensional electricfield distributions can be observed by the useof a two-dimensional lock-in amplifier. Rather. a CT (computer tomography) technique has . the difference between the indices of refraction. According to equation (17.0mm to 50mm by a beam expander. np and n. A portable Kerr system (Thompson and Luessen.16. and the component polarized in an orthogonal direction.e21. E. Measurement sensitivities have been greatly improved by the use of electric field modulation and elliptically polarized incident light modulation. distinction is drawn between the applied electricfield and the sensing electric field. This optic phase shift A0 islinearly proportional to the squared electric field magnitude E2.

1996) distributions of the electrical fieldin dielectric liquids. Accumulated space charge in oil leads to field concentration and may induce breakdown. 1997) and three-dimensional (Shimizu et al. . 17..17 Principle of operation of an electro-optic measurement device.. For rapid measurements? computer-controlled electronic cameras can be ure 17.To ~ r o u n ure 17. been recently used to measure three-dimensional electric field distributions (Takada et al.16 Block diagram of a portable Kerr system used to measure highvoltage pulses. 1996). ~easurements based on the electro-optical Kerr effect are particularly effective in those cases. attempts were made to develop mea~urement techniques for the t~o-dimensional (Zhu and Takada. Based on the electrooptical Kerr effect.1 Information about the electricalfield distribution and the accumulated space charge distribution in oil is vital to both apparatus design and problem diagnosis.

The influence of system nonuniformity could be eliminated by the use of both optical and electrical modulations. The cavity is represented by a capacitance c.18. is reached. this charge transfer is equal to 41 2(b + c)AV (17. Measurement of the space charge field (larger than 80 kV/cm) in transformer oil was undertaken by Mahajan and Sudarshan(1994). therefore. Also. partial discharges within the insula- tion are usually a prelude to insulation failure. as will usually be the case. partial discharges cause patterns of current impulses in the leads of the sample. Vl. The magnitudes of space charges with DC excitation varied from 60pC/cm3 to 10nC/cm3. If the sample is large compared with the cavity. The dielectric in series with the cavity is represented by a capacitance b. Thecharge 41 transferred within the cavity is taken as a measure of the partial discharge. Their detection. When an insulation sample is subjected to high voltage. a diagnostic image lock-in technique was proposed. When the breakdown voltage of the cavity. The magnitudes of space charges in transformer oil with AC applied voltages varied from 2 to 50nC/cm3. As seen in many sections of this book. If the circuit is energized by AC voltage.nis used for optical detection. a Kerr electro-optic measurement system was used to measure the electricfield in a transformer oil/solid composite insulation system (Okubo et al. which is shunted by a breakdown path. the insulation breaks down. Transformer oil was usedas a Kerr medium in these measurements. 1997). A s the collapsing voltage reaches its extinction value Vz the discharge stops and c recharges. and so on. a recurrent discharge occurs where c is charged.53) . The behavior of internal discharges under AC voltage canbe described using the circuit of Figure 17. The results illustrated the limitations of the electro-optic technique to measure interfacial electric field (gas-solid) with transformer oil as the Kerr medium. and the sound part of the dielectric is represented by capacitance a.. To decrease the effect of the noise component resulting from the electrohydrodynamic motion of the dielectricliquid. Spacecharges resulting from charge injection have been found in transformer oil at room temperature under electric stresses in excess of 150 kV/cm with AC excitation and 90 kV/cm under DC excitation. is a vital task.

+ V2) If V2 is neglected. Instead.55) rti All circuits used to detect partial discharge impulses can be reduced to the basic diagram of Figure 1'7. then P"cxAVxV. (17. the "apparentcharge" that flows through the leads of the test object isused to identify the discharge. be measured directly with a discharge detector.18 Analog circuitof partial discharge.19 which is made up of: . P = . The charge q1 cannot. Most discharge detectors are capable of determiningq. c ( + (17.CAV[b/(b + C)] Vi y neglecting b in comparison with c. Therefore. v: -V.) AV X X (V.54) =c . where A V= VIV2 is the voltage drop within the cavity. But VI = b/(bf C ) K where Vi is the external inception voltage at which the sample starts to discharge. The apparent charge q can be related to the energy P of the discharge as follows: P =.I l l high voltage AC I I ure 17. in practice. This chargeis given by q=bAV It causes a voltage drop [(bAV)/(a b)] in the test object.

High voltage k ure Basic diagram for partial discharge detection. namely.13. An impedance Z .2. or an oscillatory RLC circuit. C is the stray capacitance shunting R. of a shape given by (17.processes. the noise increases. A discharge-free high-voltage source 2. Both methods are equal electrically. a noninductive resistor R shunted by a stray capacity C. displays.13.2. as in the figure. or placed in series with the coupling capacitor. I .2 Optimal Amplifier Bandwidth The noise of an amplifier is proportional to the square root of the bandwidth. Optimal sen- . and stores the detected signal I( 17. The tested insulation a 3. and (1 7.Twoforms of impedance are commonly used. as the bandwidth is increased. if the sample is large. On the other hand. However. In the RC circuit the impulse will be unidirectional. A coupling capacitance k which facilitates the passage of the highfrequency current impulses 5. the response factor of the amplifier increases with bandwidth. Z is often placed in series with k so that the large charging current of a does not pass through the impedance. q == b A V .1 TheDetectionImpedance The impedanceZ can be connectedeither in series withthe sample. across which voltage impulses are caused by the discharge impulses in the test object 4. A measurement circuit whichamplifies. in practice.56) where q is the magnitude of the discharge causing the impulse.57) 17.

The impedance is either resistive or an attenuated resonance circuit with a relatively wide band. namely. A peak voltmeter serves as a picocoulomb meter and is used to detect the highest impulse.19 can be used in practice in two different ways. 10. 6.20 showsa modern direct partial discharge detection circuit.1DirectDetectionCircuit Figure 17. The amplifier inthe PD detector has a bandwidth wider than that of the detection impedance. + + The basic detection circuit shown in Figure 17. The test object is connected between high voltage and earth to avoid the need for special connections at the object’s earth terminal. . 8.~ i ~ h . 7. A low-voltage filter in the power supply is needed to suppress interference from the mains. Such a power-supply filter is a heavy and expensive piece of equipment. 4. The measurement impedance is shunted by an overvoltage protection device. 2. 5.V o Meas~rements it~~e sitivity is obtained when the time constant of the amplifier ismade equal to the time constant of the detection circuit. A bandwidth of about 50-100kHzissufficient for satisfactory resolution. in direct detection and bridge detection circuits. A high-voltage filter further suppresses interference or possible small isc charges from the transformer. The latter is given by R[ak/(a k ) c ] It is concluded that optimal amplifier sensitivity is obtained when the coupling capacitance k is of the same order of magnitude as the sample a in order to overcome the effect of the stray capacitance C. 17.133. The lead connecting the high-voltage source to the test object must be free of corona. The coupling capacitor is discharge freeand is usuallyconnected to the measurement impedance as shown. The test circuit is placed in a Faraday cage which is earthed at one single point. The impulses are displayed on an elliptical time base. 1989). but wider bandwidths up to 500 kHz are also in use. 3. Also. It has the following features: 1. The power supply of the detector is filtered.3. the sensitivity decreases linearly with the test object capacitance (Kreuger. 9.

$ 2 .nis E r ) L!-l 1 ' I . e . - rc. I I l I l l - " U .

21).(a) 1 Principle of calibrating partia.13. .3.3. Although bridge detection is more complicated than straight detection. The relation between the discharges in picocoulombs and the detector output is thus determined. discharge detection can be performed in the presence of external H igure 17. a known charge is injected into the test object. another identical test object (or the calibration capacitor) isused together with the first to form a differential bridge. 17. The calibratoris directly connected to the terminals of the object (Figure 1’7.13. Balance is obtained by varying the impedances Z1 and Z2.3 Bridge Detection Circuit Asseen in Figure 17.22 Bridge detection circuit. 17. it has the inherent advantageof suppressing external discharges. Therefore.2 CircuitCalibration To calibrate the detection circuit. This takes the form of a generated step voltage in series with a small capacitor.I discharge detection circuits. Discharges in the objects are detected while external discharges are s u p pressed.22.

nis _"""" i i : I P L > X nf . E L U) L 2 0 . c. nf n L 0 ( D n 2 =I . - c. L ( D CL a ._.

24a-d.discharges or other interference. Figure 17.23 displays a complete practical bridge detection arrangement. 1 Oscilloscopes are still in general use to observe the impulses of partial discharges.24a. The ellipse is situated in such a way that top and bottom coincide with the plus and minus crests of the high-voltage sinewave and the ends coincide with the zero crossings. Currently. processed. 1988). and redisplayed. as seen in Figure 17. The discharge impulses are usually displayed on an elliptical timebase as shown in Figure 17.4. The location and shape of the discharge spikes can be used to identify the source of partial discharge. unlike the latter. .1 Discharges in an Internal Solid Dielectric Cavity The discharges occur ahead of the voltage peaks on both the positive and negative halves of the waveform. Internal discharges can be easily distinguished from esternal ones since. the display is interfaced with a computerwhere it is stored.13. 17. the former are hardly altered by changing the balance. Theyare nearly of 4 Partial discharge interpretation displays. The following typical cases demonstrate this exercise (Nattrass.

4. The inception voltage is above the minimum detectable discharge. Thedischarges occur in advance of the voltage peaks and are often equal on both halves of the waveform. Unequal magnitude and number on the two half-cycles are usually in the ratio of 3: l .4. as seen in Figure 17.24d.4.13.withsome randomness with time.3 Cavity Formingbetween Metal or Carbonon One Side and Dielectric Material the on Other Side The discharge phenomenon is in advance of the voltage peaks and is asymmetrical. and the discharge extinction voltage is equal to or slightly belowthe discharge inception voltage.24~. There is little or no variation in magnitude with increases in voltage. There is little or novariation in magnitude with a fairly rapid increase and decrease in voltage. The form of the time dependence of the electrical response carries information about the charge or polarization dis- . 17. as seen in Figure 17. This is particularly evident in case of high-voltage cables. The thermal pulse method consists of applying a thermal pulse to one surface of the dielectric by means of a light flash and measuring the electrical response generated by the sample as a function of time while the thermal transient diffuses across the sample.61 nis the same amplitude and number on both sidesof the ellipse.13. as seen in Figure 17. 17. Several techniques have been developed to directly measure the space charge distribution.1 The accumulation of space charges in insulation materials has the adverse effect of lowering the breakdown field of the insulation. although differences of 3: 1 are normal. Discharges occur in advance of the voltage peaks and are often similar in number and magnitude.24b. 17. 17. and the phenomenon appears similar to the preceding case.4 Voids in Cast Resin Insulation Systems due to Faulty Processing The formation of electrically conductingproducts can result from the action of discharge upon the resin at these sites.2 Internal Discharges at Fissures in Solids in the Direction of Field This phenomenon is typical of some cable problems.13. but approximately 10:1 is needed to clearly distinguish it from the first case above.

This imposes limitations on the usefulness of the method and necessitates that results should be interpreted appropriately. surface charges are induced on the transducer. 1994..25. A number of methods utilizing pressure pulses have been developed to study charge and polarization profiles in the thickness direction of dielectrics (Li and Takada.tribution. The temperature waves are attenuated as they progress through the sample and are also retarded in phase. The opposite is also true: when pressure is applied. when an electric field is applied to the first transducer a pulsed acoustic wave p(t) is generated. usually a thin film. Such distribution can be obtained by a deconvolution process. each surface of the sample. The interaction of the fluctuating temperature and the spatially distributed polarization and space charge produces a sinusoidal pyroelectric current. When this wave is propagated through the second transducer. In those methods piezoelectric transducers are generally used. Thus a nonuniformly distributed thermal force acts on the sample. a mathematical deconvolution technique is required to compute the polarization and space-charge distributions from the current-frequency data. . When a space charge is accumulated in a polymeric material. When a piezoelectric transducer is exposed to anelectric field.isexposed to a laser beamwhichis intensity modulatedin a sinusoidal fashion by an acousto-optic modulator or light chopper. In Figure 17. The sinusoidal modulation of the laser beam causes a sinusoidal fluctuation in temperature of the front electrode. resulting in propagation of temperature waves into the sample. A s with the thermal pulse methods. 1994). Polymer insulating materials possess neligible piezo strain and stress characteristics. Ahmed and Srinivas. acoustic waves are generated in the transducer. 1997). This method utilizes sinusoidally modulated surface heating of dielectric samples to produce spatially nonuniform temperature distributions through the thickness. In the LIMM technique. surface charges on the transducer are induced so that an electric current flows through the outer circuit. the piezo strain and stress characteristics of the charged material become greatly enhanced (Li et al. The laser beam is absorbed by the front of the electrode of the sample. This current is a unique function of the modulation frequency and the polarization and charge distributions.

(t d t -2) dz q(t) = k h(z=ut)p(t --c )dt [ transducer I transducer I1 " (b) Pressure-WavePropagation Method W ) p(t) = k g(z=ut)e. p (z==ut)p(t-z)dc t ~ laser pulse (d) Pulsed ElectroacousticMethod i(t) p(t) = k j p(z=wr)ep(t -z)dz t t sample I transducer 1 1 I ure 17.(t --c) dz t i (t) = kc.Anis (a) Piezo 'I Piezo Transducer i(t> p(t) = k g(z=ut)e. .25 Space charge measurementby pressure pulse methods. p (z=vz)p(t -z)dt t (c) Laser-Induced Pressure Pulse Method i0) i (t) = ks.

A thermoelastically generated. The acoustic wave generated by this method is large.3 PulsedElectroacousticMethod When a pulsed electric field is applied to a dielectric with accumulated space charge.3. The acoustic wave p(t) is proportional to the space charge distribution p(z).ments 13 17. pulsed force is generated from which a pulsed acoustic wave p ( t ) is launched. and coaxial samples. 17.4.25a is replaced bya dielectricwith space charge accumulation (see Figure S7. the reaction of the evaporation of the electrode material generates the pulse acoustic wave p(t). .4. the signal level being about lOOmV.2 Laser-Induced Pressure Wave Propagation As seen in Figure 17.1 Pressure Wave Propagation Method When the second transducer of Figure 17.3. Therefore. then the duration of the applied pulse voltage should be about 1. if 5% space resolution is desired. the space charge distribution canbe determined. A space charge distribution of thickness 50-200pm can be obtained directly. It employs a stress wave generated thermoelastically by absorption of a sub-nanosecond laser pulse.25-5. when a laser pulse of about 500 PS is irradiated to the target electrode of the sample. 17. This acoustic wave is detected by the second transducer so that the induced charge q(t) on the surfaces of the transducer is proportional to the space charge distribution p(z). The current through the outer circuit is proportional to the accumulated space charge distribution. This methodisusedinfilmsamples (thickness of about 50 pm). Since the acoustic velocity in a solidpolymeric dielectric is about 2000m/s.0mm). The current in the outer circuit is proportional to the accumulated space charge if the duration of the acoustic wave p ( t ) is much narrower than that of the transit time of the acoustic wave across the sample.4.3. plate samples (thickness of about 5. a current flows through the outer circuit which dependson the sample’s space charge.25c.0 ns.25d. with the condition that the duration of the pulsed electric field is the same as in the pressure wave method. One weak point is that the laser pulse generation setup is large and expensive. laser-induced pressure pulse technique is also used for measuring space charge in dielectric films. as shown in Figure S7.25b).

two types of error are known to exist in high-voltage testing: those caused by electromagnetic (EM) interference.4. Power-line RF filters may be used with oscilloscopes. The mechanism of these methods isbased on the interaction of polarized light. 17.1 ~hotoconductivity Method This method is based on the absorption of a narrow light beam in a thin photoconductive layer. information can be obtained about the distribution of the space charge.4. Electromagnetic noisemay infiltrate into the measuringsystem and be superimposed on the measured signal. rapidly changing voltages and currents. Signalsshouldbe transmitted viashielded and matched coaxial cables. In addition to human and instrument errors.14. By the detection of thephotocurrent produced. . but are also used on polymers.nis 17. This method is nondestructive only for short illumination times.4.14. Thefollowingmeasurescanbe taken to reduce EM noise: 1. This noiseiscaused by transient potentials and by strong electromagnetic fields associated with the high. Variations in the field through a sample are observed by the spatial resolution of the spectroscopic signal inthe detection system. and those produced by improper earthing.14. 17.2 Electro-Optic Field and Space Charge Mapping Electro-optical methods consist in the application of the Kerr or Fockel effect. Electro-optical methodshavebeenusedmostly on transparent dielectric liquids. with the field of the space charges.3 Spectroscopic Measurement The splitting or shift of spectral lines in an electric field is used to determine the magnitude of the field. Very weakly absorbed visible monochromatic light liberates carriers that move under the fieldof the space charges. narrow or expanded. 2.

2.choose a proper value for the internal capacitance of the . Impulse generators should be connected to very low-resistance ground composed of a grid with rods driven into the ground close to the generator (Chapter 13).0 59 2.5 99 4. Spacing (cm) Voltage (kV) 2. 4. Multiple grounding of the high-voltage test circuit should be avoided. The power supplies to the DC source of impulse generators and to the oscilloscope should be made via isolating transformers. rapidly changing voltages and currents in a high-voltage laboratory produce transient currents in the ground connections.0 86 3. Withvery steep surge measurements.3. thus raising the impedance of that loop. h-Voltage L ~ b ~ r a t o r i e s High. If necessary. use theattached table to predict the expected sphere gap separation. Indicate the voltage reading obtained by: (i) 25 cm sphere gap (ii) a peak voltmeter V (iii) an electrostatic voltmeter If k = 20 x lo6V/s and the wave frequency is 50 Hz. Following are some of the measures that should be taken to avoid these adverse effects: 1. multiple shielding maybe used. 17. 3. 4.5 73 3.5 125 If a peak voltmeter uses a milliammeter whose full-scale deflection is 500ml2. The ground current loop can be interrupted by winding the cable on a ferrite core.16. the oscilloscope may be located away from the source of noise in a shielded chamber.0 112 4. the temperature and pressure during the measurement are respectively 30°C and 99 kPa. (1) A high-voltage periodic wave may be approximated by the triangular function shown.

5 MQ. 1988. and the high-voltage arm capacitance is l000pF. Berkebile L.The corresponding maximum detectable phase retardation angle is 40". Nilsson S. IEEE Trans PAS-100: 1498-1504. A high-voltage. k i Cassidy EC.2m high and 0. Ahmed NH. Holt P. 1981. Electra 121:69-75. Basel. justifying your answers. M. and the divider is 1. The cell uses a dielectric liquid whose Kerr constant is m/V2 and uses a uniform field gap of 5mm. A capacitive/resistive potential divider is used to measure AC high voltages. The division ratio is to be 1: 1000. 60Hz AC signal of amplitude x kV with a superimposedheavy DC component of y kV is to be measured by a peak voltmeter and an electrostatic voltmeter.15 m in diameter. HighVoltage Laboratory Planning. If the tolerated error in phase shift between the divider's input and output is only S". What is the maximum phase retardation that can be practically detected? Estimate the a ~ p l i t u d eerror in an AC unscreened 2000: 1. 100 MS2 resistive divider which is 120 cm high and l0 cm in diameter. Hebner RE. 1988. E. 1997. ~stimate appropriate values for the low-voltage arm components. ~ a l e ~ sRA. An unscreened 500 kV resistive divider is used to measure AC voltages at a division ratio of 1500. A Kerr cell is used to measure high votlages of up to 20 kV. whose magnitude lies between 0 and 5 kV. Estimate the reading of each instrument. . 1977.Switzerland: Hylt~n-Cav~llius Haefely. If the high-voltagearrnresistanceis 2. Sun S. estimate the error in the maximum measurable voltage. IEEE Trans"Die1ectrElectr Lnsul 5:644"656. Srinivas NN.Anis voltmeter in order to measure the voltage wave shown. It is now required to replace the cell m/V2and which has a liquidwith another whose Kerr constant is discharge threshold of 800 kV/cm. estimate the corresponding tolerance in the proposed low-voltage arrn resistance.the high-voltage arm resistance is600 kQ. Proc IEEE 65(11):1524-1548.

Shimim R. Geneva: International Electrotechnical Commission. 1989. Okubo H. Tokayo. Maeno T. Sudarshan TS. 1996. IEEE Trans PAS-99:636-649. Kato K. IEEE Electr Insul Mag 4:lO-23. 1960. Proceedings of 3rd International Symposium on High Voltage Engineering. Takada T. Thompson JE. IEEE Trans PAS-101:3967-3976. Germany: Friedr Vieweg & Sohn. Harumoto Y. IEEE Trans Dielectr Electr Insul 5:748-757.1997. Paper 44-08. IEC. Ida Y. Yamamoto H. IEEE Trans PAS-lOO:4811-4814. Okubo H. Kind D. . IEEE Committee. Boston: NATO AS1 Series. Kuffel E. Geneva: International Electrotechnical Commission. Li Y. 1979. Braunschweig. 1983. Milan. Publication 96. Recommendations for Voltage Measurement by Means of Sphere-Gaps (One Sphere Earthed). Matsuoka M.1. 1994. Yoshida Y. 1994. Publication 62. Hayakawa N. Kreuger FH. Dechamplain A. Malewski R. High Voltage Test Techniques. IEEE Trans PAS-102:3549-3557. 1996. 1982. Zaengl WS. Kato K. An Introduction to High Voltage Experimental Technique. IEEE Electr Insul Mag 10:16-28. Mahajan SM. 1984. IEEE Working Croup. 1973. Oxford: Pergamon Press. 1986. Nattrass DA. London: Butterworths. Yasuda M. Mukae H. IEEE Trans Dielectr Electr Insul 2:63-70. Geneva: International Electrotechnical Commission. Takada T. Fast Electrical and Optical Measurements. Luessen LH. 1978. Li Y. IEEE Electr Insul Mag 2:8-20. Takada T. IEEE Trans Dielectr Electr Insul2188-195. Radio FrequencyCables: General Requirements and Measuring Methods. Zhu Y. 1988. Takada T. High Voltage Engineering Fundamentals. Stark WB. Publication 52. IEEE Trans Dielectr Electr Insul 4: 191-196.IEC. Shimizu R. Partial Discharge Detection inHigh-Voltage Equipment. 1994. IEEE Trans Dielectr Electr Insul 5:64-70. Zhu Y. 1997. 1962. Hikita M. IEC. 1980. Shimada M. 1981. Sawada A. Hayakawa N.

This Page Intentionally Left Blank

Techniques
Cairo University, Giza, Egypt

TI

This chapter presents some of the techniques and theory involved in the testing of high-voltage equipment either in the laboratory or in the field. Themainconcernwith electric equipmentis that its insulation should withstand its operating voltage and the occasional overvoltage transients expected in the system. A further requirement for the satisfactory performance of the insulation is the minimi~ationof internal discharges in voids present within the dielectric, which may cause its deterioration and eventual breakdown. High-voltage dielectric loss and partial discharge testing can, to some extent, help in the selection and designof insulation for electrical equipment. Equipment insulation may be classified as self-restoring and non-selfrestoring. Self-restoringinsulation completely recoversits insulating properties after a disruptive discharge. The method of sampling and test techniques differ from one type of insulation to the other. International and national specifications for testing are outlined to meet the users’ and manufacturers’ requirements. These specifications describe in detail the tests to be carried out on a specific piece of equipment, their procedure, and acceptable limits of test results. Usually, the specifications are not limited to electrical tests but include other tests, such as mechanical and thermal tests.

20

A new equipment maypass a high-voltage test but may fail in the same test after some time in service. This is due to contamination of the insulators’ surfaces, which can lower their dielectric strength while gradual chemical deterioration due to the reaction of the insulation with air is taking place. Corona discharges in voids in the insulation can gradually cause deterioration (Chapter 8). Voltage transients can initiate tracking or carbonization, and caneven puncture the insulation and lead to early failure. This shows the importance of periodic maintenance and in-service tests. The following sections are devoted to various types of tests and voltages to be applied to circuit breakers, cables, transformers, high-voltage insulators, and surge arresters.

Some of the tests are performed during the early stages of development and production, others after production and installation. Type tests are performed on each type of equipment before their supply on a general commercial scale so as to demonstrate performance characteristics meeting the intended application. These tests are of such a nature that they need not be repeated unless changesare madein the design of the product. Routinetests are made by the manufacturer on every finished piece of product to make sure that it fulfills the specifications. Acceptance and commissioning tests are made by the purchaser and are self-explanatory. ~ a i n t e n a n c e tests are usually carried out after maintenance or repair of the equipment.
1
TU

The conventional forms of test voltages in use can be divided into three main groups: (a) direct voltages, (b) power-frequency alternating voltages, and (c)impulse voltages, which are divided into lightning and switching impulses. Also, in the case of testing machine insulation, alternating voltages of low frequency are sometimes used. Methods of generating these voltages are described in Chapter 16, In this chapter attention is focused on the levels of test voltages and the techniques of voltage application. Also, the test circuits and precautions to be taken in high-voltage laboratories with regard to earthing, clearances, and interference are discussed. Tables 18.1 and 18.2 list the recommended test voltages adoptedfor testing equipment for rated AC voltages ranging between l and 765 kV (IEC, 1976b). For equipment with rated voltages of 1 to 300 kV, performance under power-frequency operating voltage, temporary overvoltages, and switching overvoltages is generally checked by a short-duration power-frequency test.

1

Table 18.1

Recommended Test Voltages for Rated Voltages Less than 300 kV Rated power-frequency Rated lightning-impulse short-duration withstand (peak) ( W 2Ob 40 60' 75d 40b 60 60 75 95c 1IOd 75b 95 95b 125 150 145b 170 200 250 325 450 550 450 550 650 550 650 750 650 750 850 950 1050 (rms) (kW
10 19

ge voltage withstand Rated voltage (rms) (kV) 3.6 4.4a 7.2
12a

20 28 34

13.2a 13.97a 14.52a 17.5
24 26.qa 36 36.5a 52a 72.5a 123" 123 145" 145" 145 170e 170e 170 245" 245" 245e 24tie 245

38 50 50 70 70 95 140 185 230 185 230 275 230 275 325 275 325 360 395 460

aSpecifications for dielectric tests for the United States and Canada. bFor effectively earthed neutral with additional overvoltage protection. 'For transformers with,rating of 500 kVA and below. dFor transformers with rating above 500 kVA. eReduced insulation permissible onlyfor systems with effectively grounded neutral.

Recommended Test Voltages for Rated Voltages Above 300 kV Rated switching-impulse Rated withstand withstand voltage voltage voltage (rms) 300 (peak) 750 850 362 850 950 420 950 1050 Rated lightning-impulse

(kV)
850 950 950 1050 950 1050 1050 1175 1050 1175 1175 1300 1425 1175 1300 1425 1300 1425 1550 1425 1550 1800 1550 1800 2100 1800 1950 2400

525

1050

1175

765

1300

1425

1550

But aging of internal insulation and contamination of external insulation require long-duration power freqency tests. The performance under lightning overvoltage is checked by a lightning-impulsetest. For equipment with rated voltages 2 300kV, the performance under switching overvoltages is checked by switching impulse tests. Equipment in systems with effectively earthed neutrals can safely have reduced insulation. Therefore Table 18.2 lists more than one test voltage for each rated AC voltage of the equipment.

test in^ T @ c ~ n i q ~ e s

Direct voltage is used mainly to test equipment used in high-voltage DC transmission systems. It is additionally used in insulation testing of arrangements with highcapacitance, such as capacitors and cables. It is also used in fundamental investigations in discharge physics and dielectric behavior. The value of the test voltage is defined byits arithmetic mean, The test voltage, as applied to the test object, should not conain AC components corresponding to a ripple factor of more than 5% when normal current is drawn. Duringthe test it is required that the rate of voltage rise above 75% of its estimated final value should be about 2% per second (IEC, 1973). The requirements of the test are generally satisfied if no disruptive discharge occurs on the test object whenunder the test voltage for the specified duration.

The voltage used in this test generally has a frequency in the range 40-60 Hz and a sinusoidal shape. The ratio of its peak to its rms value is equal to 2/2:f 5%. Partial discharges should not reduce the test voltage. This is usually achieved if the total HV circuit capacitance is within 1OOOpF and the circuit current with the test object short circuited is at least 1A(IEC, 1973). For dry tests on small samples of solid insulation or insulating liquids, a short-circuit current on the order of 0.1A rms may suffice. For tests under artificial pollution, the required short-circuit current depends on the ratioof series resistance R, to the steady-state reactance X , of the voltage source, including the generator or supply network, at the test frequency. It should be at least 6 A for &/X, 0. l , and at least 1A for &/Xs = 0.25 (IEC, 1973). The value of the test voltage is defined byits peak divided by A.The peakvaluesof voltages can bemeasuredwith a sphere gap or a peak voltmeter (Chapter 17). The rated withstand voltage is determined by the same method as that in the direct voltage test.

A standard lightning impulse voltage hasbeen accepted as an aperiodic impulse that reaches its peak value in l .2 ps and then decreases slowly in about 50 ps to half its peak value. Switching impulses are characterized by having much longer fronts and total durations (Chapters 14 and 16).

4

18.3.3.1

Rated Impulse Withstand Tests

For tests on non-self-restoring insulation, three impulses are applied at the rated withstand voltage level of the specified polarity. The requirements of the tests are satisfied if no failure occurs. For withstand tests on self-restoring insulation, two procedures are in common use: 1. Fifteen impulses of the rated withstand voltage with the specified shape and polarity are applied. The requirements of the test are satisfied if not more than two disruptive discharges occur (IEC, 1973). 2. The test procedure for deternining the 50% disruptive discharge voltage is applied. The test requirements are satisfied if the determined voltage is not les than l/( l 1.30) times the rated impulse withstand voltage, where B is the per-unit standard deviation of the disruptive discharge voltage (IEC, 1973). The values of the 50% disruptive discharge voltage Vso and its standard deviation B can be calculated using statistical methods. The two test procedures are the multiple-level (or probit) method and the up-and-down method (IEC, 1973).
18.3.3.2 ~UItip~e-Level Method

At least 10 impulses are applied at each test voltage level.The voltage interval between levels is approximately 3% of the expected 50% disruptive discharge voltage. The value of50% disruptive discharge voltage is obtained from a curve of disruptive discharge probability versus prospective test voltage. The accuracy of determination increases with the number of voltage applications at each level. The values of Vso and the standard deviation S can be calculated in terns of the set ofmeasured voltages Vi and their relative deviations Zj = (Vi Ve)/oe,where Ve and oe are the values initially estimated and are to be corrected by reiteration. Thus
I _

vs0 =

V(z")--z:.zV
(-22)

-

( q

2

(18.1)

in terms of the average values of V and 2. Some weighting coefficients can be applied to the readings at each voltage level and the confidence limits can be calculated, as is well known (ANSI, 1968).

S

18.3.3.3

Up-~nd-~ow Method n

A voltage V is chosen that is approximately equal to the expected 50% disruptive discharge level. A V is the voltage interval and is approximately equal to 3% of V . One impulse is applied at the level V . If this does not cause a disruptive discharge, the next impulse should have the level V A V . If a disruptive discharge occurs at the level V , the next impulse should have the level V A V . This procedure is continued until a sufficient number of observations has been recorded. To calculate and o-, the voltage readings corresponding to either the withstands or discharges can be used. If the total number of either type of event is N , with ni shots (I: 20) at each voltage level Vi, the lowest level being VOand the highest being Vk,then

+

(18.2)

S = 1.62AV
where
k

NB A2
N2

+ 0.029
ini,
'

(18.3)

k

and

B=

k
i=O

i=O

i=O

Evidently, the same statistical manipulation of the test results applies to all similar tests.
18.3.3.4 Tests with Switching Impulse

The standard switching impulse is an impulse having a time to crest TIof 250 ps and a time to half-value T2 of 2500 ps. It is described as a 250/2500 impulse. The test procedures for switchingimpulses are, in general, the same as those for lightning impulse testing, and similar statistical considerations apply. With switching impulses, disruptive discharges may occur at random times before the crest. In presenting the results of disruptive discharge tests, the relationship of discharge probability to voltage is generally expressed in terms of the prospective crest value.

Transient currents of large amplitude are experienced during the discharge of energy-storing devices, lightning strokes, and some short circuits. If these

currents have a definite shape, they are referred to as impulse currents. Impulse currents for testing normally take either of two shapes, the double-exponential or the rectangular shape, as definedin Chapter 16.Two standard shapes are in use for impulse currents: the double exponential of 8/20 or 4/10 vs, and the rectangular, with a virtual duration of500 vs, 1000vs, or 2000 vs.

Extreme caution and safety awarenessare essential items of high-voltage test procedures. The following safety features are considered for the safety ofthe operating personnel, installations, and apparatus: essential that the high-voltage equipment be properly 1. It is designed and manufactured to permit testing without unnecessary danger. 1 2 . The actual danger zone of the high-voltage circuitmust be clearly marked and protected from unintentional entry by walls or metallic fences. 3. All doors should be interlocked to remove high voltage automatically when opened. 4. Before touching the high-voltage elements after testing, visible metallic connection with earth must be established. 5. All metallic parts of the setup that do not carry potential during normal service must be grounded reliably. 6. It is preferred that the region of the high-voltage apparatus be matted by a closely meshed copper grid. The earth terminals of the apparatus are connected to it noninductively using widecopper bands. 7. All measuring and control cables and earth connections must be laid avoiding large loops. 8. The measuring signal istransferred to the measuring device (e.g., oscilloscope) via coaxial cables. 9. Shielding of a high-voltage setup by a Faraday cage is necessary for complete elimination of external interference, as sensitive measurements are often required in high-voltage experiments. 10. The clearance between test object and extraneous structure should be at least 1.5S, whereS is the flashover distance between the electrodes of the test object. In this case the effect of such structures on the test results will be negligibly small.

A typical test arrangement of a high-voltage laboratory containing an impulse generator, a high-voltage transformer, and a potential divider is shown in Figure 18.1.

Nondestructive electricaltests are usually carried out on the equipment insulation to ensure that its electrical characteristics comply with the specifications without destroying it. These tests include partial discharges, radio interference, dielectric loss angle, and insulation resistance measurements (Kind, 1978).

Corona discharges may occur on the surface of an insulator or in voids within its volume. The presence of corona may be detected by several nonelectrical and electrical methods (Nattrass, 1988). The electrical discharge detection methods make useof the current impulses accompanying discharges in the cavity. Before we discuss these methods, it is appropriate to explain.what goes on in a void within an insulating material subjected to an alternating voltage stress. A specimen of an insulating material containing a gas void can be represented as shown in Figure 18.2a and its equivalent circuit in Figure 18.2b. The capacitance CVrepresents the void and Cbthe dielectric above and below it. C, represents the rest of the dielectric. When an alternating voltage Va in excess of that corresponding to the breakdown threshold of the gas in the void is applied to the dielectric a partial discharge (PI)) will start in it. The process of partial discharge in the void isillustrated in Figure 18.3. The voltage appearing across the void if there were no discharge is Vv and is given by the expression

vv = 1 + ( d lVa 4-

(1 8.4)

W E ,

where d and dl are the thicknesses of the insulating specimen and the gas void, respectively, and E, is the relative p e r ~ t t i v i t of y the dielectric. Partial discharge in the void will start at a voltage V' on the positive half-cycle and approximately -F on the negative half-cycle. &V, is the voltage at which the discharge stops. The discharge in the void will be accompanied by a sharp current pulse, as indicated in Figure 18.3. This maybe repeated several times on the increasing part of the positive half-cycle. At point m the voltage across the void reverses its polarity, since at this instant Vv is decreasing, and discharge will continue with almost regular negative current

.l

Layout of a high-voltagelaboratory. (Courtesy of IREQ.)

H.V

VOLTAGE SOURCE

T

Void representation;(b) its equivalent circuit.

pulses. The electrical partial discharge detection methods are classified as straight or balanced methods.
18.6.1.1

Straight DetectionMethods

In straight detection methods measuring instruments measure the charge released within the discharging sites of the test specimen. The simplest circuit used for suchmeasurement contains a seriesimpedanceconnected between the test object and ground, as illustrated in Figure 18.4. Partial discharge current measurement is difficult, and precautions taken to improve the fidelityof display of the current waveshapehave been discussed in detail (Nattrass, 1988). Partial discharge measuring instruments available measure the apparent charge (i.e., the charge transfer that takes place at the terminals of the specimen). The amplifieddischarge pulses are displayed oscillographically,superimposed on a power-frequency elliptic timebase, as shown in Figure 18.5. The discharge shown in Figure 18.5a is

ure 18.3

Voltage and current traces of partial discharge in a void.

Partial discharge detection circuit (a) and apparatus (b). (Courtesy of Hypotechnics.)

for a gas void and is characterized by its symmetry around both voltage peaks, while that for a point-to-plane gap is characterized by its regular pattern around the negative voltage peak (Fig. 18.5b). The dependence of the apparent charge Q, on the actual charge Q, in the test specimencan be obtained by considering the discharge model shown in Figure 18.2. When breakdown occurs in the void, charge transfer amounts to (18.5)

+

Partial discharge display: (a) void; (b) point to plane.

where $Vv is the voltage drop in the void at breakdown (Fig. 18.3). The capacitance Cais usually much greater than c b and aproximately expressed as (18.6) After the void breakdown the system restores the voltage across the capacitor c b to its original value, and this requires a charge eato be suppli~d to c b . This charge is the apparent charge and is given by
(18.7)

From equations (18.6) and (18.7) the apparent and actual charges are related by the approximate expression
R

~ommercial partial discharge measuring instruments are usually calibrated in terns of apparent charge. Figure 18.6 shows a simplified circuit for detection, employing a pass RC differentiator to sense the highequency signals generated by P pulses. The RC network acts as a single-pole high-pass filter whose function is ana1ogous to that of the powerfrequency separation filter in conventional PD detection equipment. With

the apparatus sensitivity could reach 0. nearby thyristor-controlled machines. complete rejection of external interference is possible. 1987.6.g. This implies discharge-free high-voltage sources or the provision of filters.2 BalancedDetectionMethods alanced detection methods are much more sensitive than straight detection methods. If the separation capacitor and test samples are of equal capacitance. noise may be pickedup from a variety of possible sources having nothing to do with the test setup (e.01 p c ( et al. should be eliminated. ultrasonic generators. In addition.1. and arcing contacts). The low-voltage arms contain balancing resistors and capacitors.. The high-voltage arms of the bridge contain the test sample and a separation capacitor. Otherwise. although they can be recognized. Screening ofthe low-voltage arms and good earthing will no doubtsubstamtially decrease external interference. Kreuger. on the high-voltage leads or loose earth connections.A simplified circuit of a PD measuring system that utilizes a passive RC differentiator as the sensing element. low-noise amplifiers. In the straight PD detection methods discharges in any part of the test circuit and not within the test sample may be detected and displayed along with the discharge impulses in the sample. . 18. The detection of PI)with bridges similar to that of Schering has been adopted for more than 50 years.. 1989). the bridge would be balanced only at one frequency. disturbances are thus considerably reduced. resulting in less effective rejection of external interferences. The output of the bridge is supplied to and displayed on an oscilloscope through a filter and an a ~ p l i f i eExternal ~.

3. has since been widely used to measure the capacitance and loss angle of high-voltage insulators.Corona discharges are known to produce radio noise over a considerable portion of the radio-frequency spectrum (Chapter 5).10) tan S = uC4R4 18. Proper grounding of the circuit is essential for safety and accuracy. capacitors.6.1 Schering Bridge The Schering bridge.9) (18.6 escept that a radio-noise meter is used instead of the PI) amplifier and oscilloscope. Each of the LV arms is shunted by an overvoltage protective device. which can be utilized for nondestructive high-voltage testing (Kind. devised byScheringin 19 19.6.2 Bridge Incorporating Wagner Earth (18. polarization. This capacitor should have no significant losses over the full working range. 1978).3. which operates at a few tens of volts in case either of the high-voltage arms fails. Losses always occurin dielectrics due to conduction. Measurement of radio interference from HV equipment can be carried out by a circuit similar to that shown in Figure 18. Both the LV armsand the null detector are shielded from the high-voltage circuit to eliminate any errors in the measurementscaused by the effectof stray capacitances.11) The Schering bridge is earthed at the low-voltage end of the high-voltage source. and ionization (Chapters 7 and 8). Dielectric losses cause certain electrical effects. 18. insulated with compressed gas. The capacitance of the detector’s screened leads as well as the stray capacitances of branchesAB and AD affect the balance conditions. The HV arms are the test piece and a standard capacitor. To overcome this drawback. The bridge comprises twohigh-voltage (HV) arms and two low-voltage (LV) arms. and cable samples.7) can easily be evaluated: (18. 18. The LV arms are an adjustable precision capacitor and resistors. auxiliary arms are used for maintaining points . The capacitance C2 and loss angle 6 of the test piece (Fig. A suitable design employs a smooth electrodes with corona shields.

It is evident that if the resistance or theloss factor of the insulation in a certain machine. and whether it is AC or DC (Chapters 7 and 8). When the measured insulation resistance starts to decrease considerably below the reading of the previous test. or cable is measured periodically.V. The stray capacitance of the high-voltage terminal to earth is represented by a capacitance M. 1984). In Figure 18. The ratioM/N must be balanced for phase as well as magnitude with the main bridge arms.7 Schering bridgeincorporating Wagner earth. t The magnitude of the resistance of a solid or liquid insulation varies with its temperature. The arrangement becomes equivalent to a sixarm bridge.7 an additional armN is connected between the low-voltageterminal and earth(Kuffel and Zaengl. The capacitances between the leads and screens are in parallel with the impedance N and thus do not affect the balance conditions. under almost identical service conditions.Its insulation should be more carefully tested and overhauled to avoid sudden failure during operation. ." " " I _ " " " " " " 7 I I t i . the results will indicate its degree of aging. transformer. transformer. the machine. and D at earth potential under balance. At balance the terminals of the detector are at earth potential. or cable should betaken outof service. the applied voltage. An electronic device for automatic balancing of Wagner earth could be developed. its moisture content. .

recorded.The insulation resistance test can be performed using an adjustable C source (IEEE Committee. and compared with the previous readings during similar tests. 198 1). However. Dielectric tests are carried outto make sure that the circuit breaker withstands the overvoltage expected within the power system with a reasonable safety factor. withvoltage magnitudes as given in the relevant standard specifications (Tables 18. lubrication of the mechanical parts is allowed according to the manufacturer’s instructions.Power-frequency .The DC voltage applied during the test should be raised very slowlyup to nomore than 80% of the acceptance test values. After several seconds. with no voltage or current in its main circuit. or minutes. The maxi mu^ observed temperature rise is then compared with the stipulated limits. The circuit breaker is subjected to 1000 operating cycles. The temperature rise test is carried out with normal rated current flowing in the main circuit and the circuit breaker mounted as under normal service conditions. Standard specifications giveand describe in detail the tests to be carried out on circuit breakers and how they are performed (IEC. Mechanical failures represent more than 80% of the total failures in circuit breakers. During the test. This last component is the quantity to be measured. which corresponds to conduction through the insulation. After the test a complete check on the circuit breaker is performed to make sure that all mechanical linkages and contacts are in good condition. The initial current component corresponds to charging the circuit capacitance. Testing of circuit breakers requires highly equipped laboratories and sophisticated testing procedures. These limits depend on each individual part and the circuit breaker type.2).1 and 18. the circuit current decreases to its steady component. During operation the temperature of any part of the circuit breaker should not exceed the specified limits of temperature rise. The main dielectric tests are impulse and powerfrequencyvoltagetests. 1972). mechanical adjustment or replacement of any part is not pemissible. The impulsevoltagewaveshapeusuallyusedis the standard wave 12/50 vs.

and transient and power-frequency recovery voltages. Short-time current tests 4. Capacitor charging-current breaking tests 9.e.8. while the voltage source injects a high-voltage transient across the circuit breaker contacts at the moment it interrupts the current. Single-phase short-circuit tests 6. The synthetic method of circuit breaker testing provides an alternative solution. The voltage source could be connected either across or in series with the current circuit (Lythall. under simulated rain).The current source injects the currentintothe circuit breaker under test at a relatively reduced voltage. Operating sequence tests 5. In the basic form it consists of two independent voltage and current sources (Fig. . in addition to a backup circuit breaker for protecting the sources in caseof failure of the breakerunder test. A circuit breaker should be capable of making and breaking the circuit under short-circuit conditions without appreciable deterioration of its components or change in its performance. The high-voltage source usually contains a large capacitor bank.8).voltage tests are applied for 1 minute to indoor and outdoor circuit breakers when dryor wet (i. respectively. Actual synthetic testing circuits are much more elaborate than the simplified one shown in Figure 18. Out-of-phase switching tests 8. 1986). Short-circuit tests are usually carried out according to test duties that specify the test current. Short-line fault tests 7. reaking current tests 2. Small inductive current breaking tests The short-circuit tests need high-powertesting plants to meet the everincreasing circuit breaker ratings. These tests are numerous and it isdifficult to discuss them all in detail within the space of this section. They would include elaborate control and instrumentation schemes.. Suchplants are extremely expensive. with currents reaching the order of 10 kA or more at voltages ranging up to 750 kV. Making current tests 3. 18. they are classified as follows: 1 . percentage of DC component.The breakderunder test shouldwithstand the specified voltage tests without flashover or puncture. However.

03 x lom3 for aluminum. + . (2) master breaker. These mechanical stresses may cause insulation cracking. 1978). and Rm is the measured resistance in ohms. Cables are subjected to electrical and thermal stresses while in service. the power-frequency test voltage is (2.Testi 37 Synthetic testing circuit:(1) breaker under test.5uo 2) kV for cables of rated voltages (Uo) up to andincluding 6 kV. 1 The high-voltage test is carried out under power-frequency alternating voltage or direct voltage.5u0kV for cables of higher-rated voltages. and hence the leakage current increases. and 0 the ambient temperature in "C.12) where the temperature coefficient a20 is 3. According to IEC 502 (IEC.93 x lom3for copper and 4. leading to thermal breakdown of the cable insulation. (3) synchronizing switch. They also undergo mechanical stresses during their installation and repair. and 2. Because of the much lower power needed under DC. 1 The conductor resistance of a complete length or a specimen of cable is measured and the result is corrected to a temperature of 20°C and 1km length in accordance with the formula (18.(4) overvoltage protector. IC the sample length in kilometers. it is used for commissioning and maintenance tests of complete cable networks.

In this section. For equipment rated voltage less than 3OOkV the reference voltage for external clearance is the lightning impulse .25 times the rated voltage.1 and 2 (IEC. A failure in the non-self-restoring internal insulation would be catastrophic and normally leads to the transformer being withdrawn from service for a long period. 1. 1976c. Measurement of winding resistance b. and 40 PC forPVC cables(IEC. Single-core and multicore cables are tested by applying a voltage of 1. and XLPE. The magnitude of the PI) impulses should not exceed 20pC for cables insulated wth butyl rubber. Type tests a.1976a.Partial discharge tests are carried out to ensure that the levelof the discharge does not exceed a specified limit. Measurement of voltage ratioand check of voltage vector relationship c.d). only type tests are isc cussed. and 76-3-1 (IEC. The recommended clearances among the various terminals and windings are referred to the ratedwithstand voltage of the internal insulationof the transformer (IEC. 1987b). Also. such as a short-circuit test. and measurement of acoustic sound level. Tests ontap changers 2. using a Schering bridge.l lns~~ation Levels and ~ielectric Tests It is essential to specify higher test voltagesfor the internal insulation of the transformer than for the external insulation of the system. EPR. Temperature risetests Some special tests may be required.25 times the rated voltage between the conductorandthe metallicscreen. and load loss d. Measurement of no-load loss and cu~rent e. other tests are given in the standard specifications. ~easurement of impedance voltage. as recommended by IEC 76. 1987a).the loss factor tan6 is measured at different voltages up to 1. measurement of zero-sequence impedance on three-phase transformers. The following tests are required on most power transformers. short-circuit impedance. Dielectric tests b. Routine tests a. 1987a). .

and classof long-duration discharge (IEC. Surge arresters are identified by their rated voltage. to avoid sparkover of its series gaps. Before each test. and oil of transformers designed for operation at normal altitudes are listed in tables according to the cooling medium (IEC. Arresters with rated voltages exceeding 100kV are additionally tested under switching impulses. rated frequency. the gaps fail to spark over once only. Subsequently. The permissibletime during which the applied voltagemay exceed the rated voltage of the arrester is in the range 2-5 S. The power-frequency sparkover voltage is the average of five test results. the samples must be at approximately the ambient temperature. Each sample is subjected to two current impulses of the standard shape with peak values depending on the arrester class.voltage. 1 The temperature rise values for the windings. The voltage initially applied to the arrester should be of a low value. nominal discharge current. 1970). cores. For oil-immersed transformers. the power-frequency . the temperature rise tests include the determination of top oil temperature rise and of winding temperature rise. Some of the type tests made on nonlinear resistor-type arresters for AC systems are summarized below. Five positive and five negative impulses of the standard shape are applied to the test sample. and the series gaps of the arrester must spark over on every impulse. The rated voltage of these tests must be in the range 3-6 kV. If. 1976d). in either series of five impulses. These testsare made on the same samplesas those used for power-frequency sparkover tests.5 S . After sparkover the testvoltageisswitchedoff by automatic tripping within 0. whereas for equipment of higher rating the switching impulse voltage is the one to be employed. Dry and wet tests are made on complete arresters. an additional 10 impulses of that polarity are applied and the gaps must spark over on all these impulses.

This test is applicable only to insulators for indoor use. The 50% flashover procedure is normally used. These tests are intended to provide information on the behavior of outdoor insulators under conditions representativeof pollution inservice. The pollution tests involve application of the pollution and simultaneous or subsequent application of voltage. The characteristics of artificial rain are given in IEC 168 (IEC. Insulators used on overhead transmission lines and in substations are subjected to type tests. The insulator is tested under positive and negative 250/2500 ps impulse voltage waves.2/50ps impulse voltage waves. 2. The applied test voltage is the specified dry powerfrequency withstand voltage adjusted for atmospheric conditions. 1979). The pollution tests fall into two categories: the saline fog method and the pre-deposited pollution method (IEC. 1975). Two test procedures are in common use for the lightning impulse withstand test: the withstand procedure with 15 impulses and the 50% flashover voltage procedure (EEC. The main type tests include the following: 1 . Wet power-frequency withstand test.1979).sparkover voltage is determined. 1970). 4. ests The sample tests on insulators include the following two principal tests: . and routine tests. no evidence of puncture or flashover of the nonlinear resistors or significant damage to the series gaps or grading circuit should occur. This test is applicable only to insulators for outdoor use. D r y lightning i ~ p u l s withstand e test. D r y switching i ~ p u l swithstand e test.The pollution tests maybe made either to determine the maximumdegreeof pollution withstood by the insulator under a given voltage or to determine its withstand voltage for a specified degree of pollution. The dry insulator is tested under positive and negative 1. ArtiYficial pollution tests. 3. The wet power-frequency withstand voltage is determined by the same procedure as thatdescribed in determining the dry power-frequency withstandvoltage. sample tests. D r y p o w e r ~ ~ e q u withstand e~cy test. It should not change by more than 10% (IEC. 5. The applied test voltage is the specifiedwet power-frequencywithstand voltage adjusted for atmospheric conditions. Also.

The ideas behind these tests. m being the mass of the insulator in kilograms. + From the discussion above. The insulator is left in each bath for (15 0. it is evident that each item or type of equipment has its specific tests. (3) A specimen of solid insulating material 1 cm thick has a cylindrical air void 0. the test procedure is slightly different (IEC. No puncture may occur below the specified minimum puncture voltage (IEC. (2) (a) What is the effect of partial discharge on power cables? (b) Oil-impregnated paper cables are used at voltages less than 33 k'v. The test voltage is applied on the insulator and then raised rapidly to the minimum puncture voltage specified. The unit is cyclically immersed in hot. and the general procedures.and cold-water baths maintained with a temperature difference of 50 K. after havingbeen cleaned and dried. 1979). Calculate the voltage to be applied to the specimen . instead of AC sources. 2. are essentially similar to those described above. 1979). The insulator. whereas at higher voltages high-pressure oil-filled cables are used.7m) minutes. with a maximum of 30 minutes.1. (b) Explain briefly why portable voltage sources. For insulators made of annealed glass. (c) What is meant by nondestructive testing of electrical equipment? Mention three of these tests and explain briefly why they are carried out on electrical equipment. (1) (a) Differentiate between type and routine tests. is completely immersed in a tank containing a suitable insulating medium (usually oil) to prevent surface discharges. (c) What is meant by the apparent charge in the process of partial discharge in voids in solid insulating materials? Deduce an expression relating this charge to the actual charge. The insulator is then examined. This test is performed on individual insulator units made of ceramic or toughened glass. are used to test cables on site.5 mm thick. Tests on pieces of equipment other than those mentioned above are described in detail in their relevant standard specifications.f r e ~ upu~cture e ~ c y test. ~emperature cycle test. Explain this statement. ~ o ~ e r . This heating-cooling cycle is repeated three times in succession. It should suffer no cracks or damage and should withstand a subsequent dry flashover test.

The cable has a capacitance of 0. Calculate the loss tangent of the tube and its dielectric loss per meter whenused in a 20kV system. Shea JJ. (7) Why are synthetic testing circuits now commonly used for short-circuit tests of circuit breakers? (8) What are the various tests to be carried out on power transformers? (9) What electrical tests are applicable to indoor and outdoor insulators? ketch the two common voltage waveshapesused for high-voltage testing of electrical e~uipment against lightning and switching surges. after cable maintenance. Standard Techniques for Dielectric Tests.75 F/km.00125 PE. R 4 = :3180 S2 (5) What are the sources of error affecting the balance condition in a Schering bridge? Show how these sources are eliminated.7. Lightning Arresters. given that the dielectric constant of the material is 4. Publication 68. Calculate the charging current and the kVA drawn from the testing source. ANSI. Neglect the cable inductance and losses. Part 1. Fitzpatrick GJ.56 x 1013S2an and its dielectric is 3. Non-linear Resistor Type Arresters for A.25 cm respectively. Also.5. IEEE Electr Insul Mag 3:4. The bridge has the following elementsin its branches under the balance condition: R 3 = 620 S2 C1 = lOOpF C 4 = 0. etemine the series resistanceand capacitance and the loss tangent of a cable test piece using the Schering bridge in Figure 18. calculate the ratio between the apparent and actual charges.C. Geneva: International Electrotechnical Commission. (1 1) A tube of solid insulating material has inner and outer diameters of 3 and 3. 1987. 1968.1970. Bilodeau TM.1. (12) A specimen of 10 m long of a high-voltage cable is testedat 100 kV. IEC. .to start discharge in the void. Systems. The resistivity of the tube material is 1. (6) Explain why. Publication 99-1. New York: American National Standards Institute. Sarjeant WJ. the insulation resistance measurement test is performed before the high-voltage test.

Braunsch~eig. IEEE Electr Insul Mag 4(3):10-23. Artificial Pollution Tests for HV Insulators for AC Systems. Power Transformers. 1972.Geneva: International Electrotechnical Commission. Application Guide. Geneva: International Electrotechnical Commission.2. Publication 60-1. Publication 507. Geneva: International Electrotechnical Com~ission. Part 2. IEC. Kuffel E. 1988. Power Transformers.1976a. 1987b. IEC.1. 1981. Introduction to HighVoltage Experimental Technique. . Lythall RT. The J & P Switchgear Book. Publication 502. Publication 71-1. Discharge Detection in High Voltage Equipment. Definitions. p 52. IEC. High Voltage Alternating Current Circuit Breakers.Terms. Tests on Indoor and Outdoor Post Insulators of Ceramic Material or Glass for Systems with Nominal Voltages Greater than 1OOOV. Zaengl WS. 1976c. Geneva: International Electrotechnical C o ~ ~ i s s i o1987a.Committee. Principles and Rules. Publication 76-2. Publication 885-I.1979. IEC. Insulation Co-ordination. IEC. IEC. Power Transformers. 1978. IEC. Extruded Solid Dielectric Insulated Power Cables for Rated Voltages from 1 kV upto 30kV. Publication 168. Publication 76-3-1. IEEE Trans ~AS-100:3292-3300. IEC. IEC. Publication 71-2. Part 1. Publication 76. Geneva: International Electrotechnical Commission. Publication 56-4.IEC. High-Voltage EngineeringFundamentals. 1984. IEC. Geneva: International Electrotechnical Commission. 1975. 1973. Partial Discharge Testsfor Cables.Geneva: International Electrotechnical commission. London: Butterworth. Insulation Co-ordination. Kreuger F. Geneva: International Electrotechnical Commission. 1989. Oxford: Pergamon Press. High VoltageTestTechniques. Geneva: International Electrotechnical Co~mission. Geneva: International Electrotechnical Commission. Geneva: International Electrotechnical Cornmission. West Gemany: Friedr Vieweg & Sohn.1976d. IEEE. 1976b. Nattrass D. II. 1986. Kind D. London: Butterworth. n.

This Page Intentionally Left Blank .

electrostatic separation. 1 Electrostatic precipitation is essentially the charging of dust particles in a gas and their subsequent separation under the effectof the electric field (Gordon and Peisalchov. the unique properties of high-voltage electrostatic fields and forces are utilized to collect. electrostatic pumping. transport of light materials. The principles ofapplication of electrostatics are discussed for the following selected industries: electrostatic precipitation. paper manufacture. air cleaning from gaseous pollutants. ozone generation. deposit. 1972. Egypt TI Electrostatics is currently found at the basisof many major industries related to environment preservation. 19’78). processing ofmineral ore resources. communications. or select very small or lightweight particles.in Industry ig h-Voltage ngineeri~g Assiut University. Oglesbyand Nichols. and biomedical applications. Assiut. electrostatic spraying of pesticides in orchards. and so on.These processes . electrostatic painting. separate. direct. In the majority of these industries. electrostatic spinning. smoke detection. electrostatic imaging. electrostatic propulsion. electret transducers. electrostatic printing.

may occur within a single zone or be distributed over two zones. and through collision with ions which participate in the continuous thermal motion of molecules without the aid of an electric field. .2 displays schematically the charging of dust particles in a precipitator.1 Pipe-type electrostatic precipitator. 19. deposited by the action of the electric field. where the first zone-the charging zone-is intended to charge the particles. which then experiences a force towards the collecting electrode where it is held by electrostatic forces until it is removed by mechanical rapping (Fig.or duct-type precipitators). Two systems of electrodes are used in electrostatic precipitators to obtain an inhomogeneous electricfield: a wire conductor enclosed in a cylindricalpipe(pipe-type precipitators). ure 19. for particles larger than 1 pm. The electrodes around which a corona discharge is formed are termed the discharge electrodes.for particles in the sub-micron range. These ions interact with the particulates entrained in the gas and impart charge to the dust. or a row of wire conductors located between plates (plate. and the second zone-the collecting zone-is designed to settle the particles. are the collecting electrodes. Figure 19. whereas the electrodes receiving the charged dust particles.1). The stream of ions charging the dust particles is produced by means of a corona discharge (Chapter 5) in an inhomogeneous electric field.The particles are charged as a result of bombardment by ions energized in the electric field.

ecauseof a differencein the values of relative permittivities of dust and gas. This is in addition to a reduction in the sparkover voltage. usually superimposed on a DC level value. The outlet of each channel acts then as a point discharge on the collecting electrode. This phenomenon is termed back corona (Abdel-Salam and Singer. In this way the voltage and current can be decoupled by changing the marklspace ratio of the applied pulses. the mean collector current in this case depends not only on the peak voltage but also on the pulse duty cycle.7 Schematic diagram illustrating the charging of dust particles in an electrostatic precipitator.3). provides a high peak electric fieldfor particle charging. 199l). However. When the voltage is high. the remaining part being channels (cracks)filledwithgas. and high charging fields can be established without back discharge. A dust cake on a collecting electrode consists of only 10-50% dust. so that dust collection is adversely affected and current intensity"in the precipitator greatly increases. depending on the size of the particles. 19. and the gas inside the channels is ionized. the lines of electric field concentrate inside the channels (Fig. High resistivity particles cause back corona to be formed in the precipitated layer and this has a number of deleterious effects. The most significant is the release of ions which move countercurrent with respect to the dust particles and partly neutralize their charges. which consequently reduces the precipitator's performance. Pulse charging can produce a uniform . A train of pulses. an electric breakdown takes place across the dust cake. Pulse charging was proposed as an alternative solution to back discharge.

including the trielectrode charger.Schematic diagram illustrating the arrangement of field lines on formation of back corona. whereas the collecting stage is similar for all prechargers and comprises only parallel plates so that no back corona can be formed. ~recipitatorefficiency can reach or even exceed 99%. In the charging stage. 1987). different techniques are used to eliminate or reduce the effectsofback corona. gas temperature and velocity. distribution of ionic current on both the discharge and collecting electrodes (Masuda and Hosokawa. 198 1). average size and resistivity of the particulates. This results in a low charging rate which impairs the performance even if back discharge is avoided. including the dimensions and geometry of the gas duct. To overcome this shortcoming.. 1988). an external . boxer charger. prechargers have been proposed. and cold-pipe charger (Masuda. The disadvantage of pulse charging is the ion shortage that occurs when the current must be lowered extremely to cope with high dust resistivity. and corona discharge intensity (Landham et al. high-intensity charger. depending on several designparameters. Electrostatic separation is the selective sorting of solid species by means of utilizing forces acting on these species in an electric field. The main items of the separator are a charging mechanism in the charging zone. Different types of precharger have been developed and tested at pilot-scale stage. The fundamental aim of all precharger systems is to separate the charging and collectingprocesses into two separate stages.

1994. The dielectric or poorly conducting particles lose their charge slowly and are thus held to the surface of the rotor by the image force associated with their surface charge. insulating particles fall down by gravity.. 1..4a). A grounded rotor is located close to a positive drum (Fig. Although there are many ways (Lawver and Dyrenforth. 19. The charged particles rapidly share their charge with the grounded rotor and are thrown from the rotor in a trajectory determined by centrifugal force. but the magnitude of the electric charge is signi~cantly different. the most c o m o n mechanisms are as follows. A coal beneficiation system has been developed on the basis of electrostatic separation (Nlasuda. gravity. Phosphate particles enter the separating zone with a net positive charge while the quartz particles bear a net negative charge. in which solid particles pass through a corona discharge (Chapter 5) from a fine wire or a series of needlepoints positioned parallel to a grounded rotor of the separator. or (b) only one type of particle bears an electric charge. is vibrated on its way from the hopper to the forming chute (Fig. and air resistance. The particles are charged by bombardment with the corona ions. 1977/1978).4~). Sadiku.4b). Charging by contact and frictional electrification is the mechanism most frequently used to selectively charge and electrostatically separate two species of different materials such as phosphate and quartz. 1981). and a feeding and product collection system. However. where the external field in the separating zone deflects the coal-rich and ash-rich particles in opposite directions towards the collecting-plate com~artments. This is a kind of electrostatic precipitation of powder or liquid paint on the surface of an object to be coated (or painted) (Inculet. or (c) particles bear the same signof charge. 1998.electric field in the separating zone. 19. Iuga et al. Hoferer et al. Charging by ion or electron bombardment is used. 19. When conductive particles coming from the hopper pass over the rotor they become negatively charged and attracted toward the positive drum. composed of small particles of quartz andphosphate. 1973. The ore. Charging by conductive induction is a charging mechanism suitable for separating well conducting particles from well insulating particles. 1999) to charge solid particles. The charging oftwo different species entering the separating zone results in: (a) particles bear electric charges of opposite sign. The well conducting particles are thrown free of the rotor by a combination of centrifugal force and gravity (Fig. . 2. 3.

. 19. notonly on the front side but also on the back side of the object (Fig.(b) Electrostatic separator based on charging by ion or electron bombardment. a liquid jet issues from thereservoir and extends along the axis of a concentric charging cylinder where a potential V ’ is applied between the jet and the cylinder.5). The jet charges and breaks up into charged droplets while it is in the cylinder. The electric field plus spacecharge effects between the jet and the grounded object deposit the paint droplets on the surface of the object to be painted.(a) Electrostatic separator based on charging by contact and frictional electrification. In liquid paint.(c) Electrostatic seljarator based on conductive charging.

after being charged. is sprayed into the space above . . * P Electrostatic powder painting.6). the paint particles are charged by bombardment with corona ions (Chapter 5) moving under the influence of the prevailing electric field between the corona electrode and the object being grounded (Fig. In its simplest form. 1991) which results in craters that impair the quality of finished coat. The object to be coated is grounded and supported so that its surface can be approached without obstruction. 19. The coating material. an electrostatic coating operation is visualized as taking place in the following manner.l + v i ? In powder painting. The paint precipitation stops at a certain thickness called the “limiting thickness” because of back discharge (Abdel-Salam and Singer.

The major problem with this method. The spray loss and accordingly the deposition efficiency can be attributed to both wind drift out of the target area (exodrift) and deposition to thesoil during the process (endodrift).which will increase not only the pesticide deposition efficiency but the biological efficacy as well. Law. as a result. they move toward and accumulate on the surface to form the coating. agriculturalpesticides are applied in the form of waterbased sprays using a hydraulic-atomizingnozzle. is that a large portion of the spray is lost and the target deposition efficiency is less than 25% (Abdel-Salam and Singer. etc.the surface in the form of finely divided particles. 1989). The mutual repulsion of the particles and their coulomb attraction to the leaves (electrically grounded through the tree itself) ensure a superior coverage on both sides of the leaves (Law and Cooper. increase the coulomb force acting on the droplet to a value which is high enough’to overcome the mass-dependent gravitational and inertial forces (Law. in addition to being essential for increasing the biological efficacy. IC I Electrostatic imagingis a process by which an ordered arrangement of electric charges is deposited on a surface as a latent electrostatic image . furniture. There is an attraction of the particles to the surface and.as discussed for electrostatic painting. This will. In pesticide spraying.. The reason behind this is the small droplet sizes that can be achieved in electrostaticsbasedsystems. That is. however. in effect. Electrostatic painting is widely used in the continuouscoating lines of automobiles. the generation of small droplets will increase the droplet’s charge-to-mass ratio. 1993 and 1995. 19. Y I As has been established. 1989) (Fig. a nozzle has been developed to generate a concentrated jet which forms an electrified cloud inside the tree foliage. and so’on. 1991. As with industrial coating techniques.. the means by which they are charged. The various electrostatic coating applications are somewhat sophisticated modifications of this simple situation. They differ from one another in the manner in which the particles are forced. the applicationof a sufficientlyhigh-voltage field at the surface of a liquid corning from a nozzle will lead to the formation of charged droplets in the #form of aspray. Conventionally. electrostatic pesticide spraying technology offers an attractive alternative. electric appliances. 1995). AbdelSalam et al.7).with a subseq~ent increase of the deposition efficiency of pesticides.

High de Electrostatic spraying of trees. Masuda.8a. 19. the negative charges can get no farther than the bottom of the photoconductor. a “latent” positivecharge electrostatic image isproduced on the surface of the photoconductor (Fig.8b). Since the photoconductor is an insulator in the dark. Initially the entire surface of the photoconductor is charged uniformly by a positive charge. and the negative charges on the lower surface can combine with those of positive polarity on the upper surface to neutralize each other. as shown in Figure 19. The negative charge on the lower surface comes from the metal plate below and is drawn by attraction to the upper positive charges. Development of the electrostatic image is the process of supplying powder (toner) to the charged areas to provide visibility of the image. 19. Consider a system of a metal plate (aluminum) covered by a very thin layer of insulating material (alumina). 1992).8d). the positive surface charge in dark areas is maintained on the photoconductor (Fig. If a sheet of paper is placed over the development image and sprayed with a . However. 19. 198l. Schein.8~). The charge on the top surface comes from the positive ions produced by a corona discharge (Chapter 5) on a wire traveling above the photoconductor. which in turn is coveredby a layer of photoconducting material (selenium). When light from the document to be copied is focused on the photoconductor it becomes a conductor. 1977/78. and subsequently “developed” to convey visual information to an observer (Inculet. The toner particles are charged negatively and attracted to the positive imageby reason of the external electric fieldgenerated by this image (Fig. Thus.

t I C D + a>+ o+ Q+ o+ o+ a+ o+ .

This is why the ink-jet printer is known as the chargemodulation typeof ink printer. CarnahanandHou. maintained at ground potential.8e). Electrostatic printers are classified into ink-jet and ink-spray printers (Swatik. Later. a visible image is instantly produced. This technique is viable on standard forms (sheets of paper). upon contact. 19. a high-voltage. i. At the point where droplets are forming. The ink-jet printing technique produces instantly visibleimages on standard forms by the electrostatic deflection of charged ink droplets into electrically programmable dot matrix patterns.field controlled by a computer is applied to the jetto give the newly formed droplets an electric charge related to the signal (Fig. A second sheet ofpaper can then be placed overthe image and the transfer process is repeated. the electric field penetrating the sheet will act on the negative toner and cause it to transfer to the paper (Fig.9). Electrostatic printers are usually known as nonimpact printers. characters are cast on metal type. 1977. In the ink-spray printing technique.Buehner et al.external to the form (print surface).1977.. the toner is thermally fixed to form a permanent image (Fig. is attracted by the electric field of the gate.e. Many typewriters and computer line printers use so-called impact printing. 19. creating an image. 19. a well-toned image is still present. New requirements for performance and speed exceed the capabilities of most impact printing technologies. Electrostatic printing is an alternative method. which makes an impact on an ink ribbon to produce prints on paper. This deflection causes them to strike the print surface (usually a piece of paper) at different points.positive charge. Whenthe sheet is stripped away from the imaging surface.Ashley. The printing ink is formed into macroscopic droplets which are imaged by electric field. 1977). and no image development or fixing is required. The form serves only as a receptor for the imaged ink droplets and. The electrically conducting ink is forced through a nozzle to form a thin jet. The conductive ink. a reservoir containing the ink is pressurized at a low level (a few centimeters of water) sufficient to form a convex meniscus of ink at the opening of a vibrating nozzle but not high enough to cause an outflowof ink. The drops then move into a deflection region where a steady transverse electric field deflects them by an amount depending on their acquired charge. 1973..80. When the . which then breaks up into droplets under the influence of surface tension and the mechanical vibrations in the nozzle. This principle is analogous to the deflection of electrons in a cathode ray tube.

This is accomplished by electronically controlling the voltage applied to the deflection electrodes (Fig. the droplets are produced when the potential between the gate electrode and the meniscus is v >~ I ) ( y / & ~ o . programmable deflection is obtained byvarying the magnitude of the deflection field..e* . .. 1977/78). ' 'I 1 * . Since the droplets are very small they can be quickly accelerated and are able to produce high-quality hard copymuch faster than electronic (impact) printers. droplets are produced.The droplets are accelerated and then imaged into the desired dot-matrix fomat by electrostatic deflection. and I) = distance between orifice and gate (Inculet. 19. to allow for printing on stationary forns.e .e . Expressed in terns of potential V . ir electrostatic attraction force exceeds the surface tension of the meniscus. * n . electrostatic printers have a definite advantage as the level of noise is only that generated by the hardly visible droplets of ink landing on the paper. Since the droplets are produced with a nearly uniform specific charge. There are commercially available electrostatic printers with capabilities of printing in excess of 30. Two sets of electrodes are employed to obtain both horizontal and vertical deflection. d = nozzleorifice diameter.10). With the current concern for noise from impact printers.. E = ink pernittivity. s where y = ink surface tension. This is why the ink-spray printer is known as the field-modulation type of ink printer.000 lines/minute.

is metallized with a thin layer of foil less than 1 pm thick and backed with a porous metal plate (Fig. as shown schematically in . the higher is the electric field Z b (see Example 5) and the better is the performance of the microphone. Such a change of the charge on the plate is used as an output signal after being amplified. the electric field-Eb and hence the surface charge on the metal plate changes with the distance b. The excellent frequency response and reliability have resulted in an extension of the use of this device to headsets for telephone operators (Crowley. the handling and the phenomena associated with very thin films have also attracted industrial interest. Sound waves strike the electret film. an electretfilm. 1986). thus changing the distance b depending on the loudness of the sound. The word “electret” indicates that this is the electric counterpart of “magnet.10 Ink-sprayprinter. only a few micrometers thick. An electret isa dielectric material with a permanent dipole moment or charge separation.” In the electret microphone. The higher the voltage applied to the metal plate.1 l Figure 19.1l).11. A good example is the electret transducer (microphone) shown in Figure 19. A s the surface charge density psis constant (permanent charge density). 1 The feasibility of electrostatic transport of light materials which triboelectrify themselves on a stationary “conveyer” belt. 19. 1 While the majority of electrostatic applications in industry involve fineparticles. moving it with respect to the metal plate.

. etc. 1977/78).12. silicon carbide.l1 Electret microphone. Any particle which strikes an area already fully covered by other particles falls back and is recovered for reprocessing as surplus abrasive material (Inculet. 1977/78). with a unifor~ ~istribution. . As theyarrive in the strong electric field between two parallel electrodes stressed by a voltage up to 100 kV. as shown in Figure 19. The electric charge which they acquire from the se~iconducting belt causes them to be propelled and implanted in the adhesive. thus. diamond. Figure 19. they line up along the field lines with sharp points facing the electrodes. are poured onto a continuous conveyor or belt. Abrasive particles of quartz. Another example of what electrostatics can do for industry is evidenced in the lining up and gluing of billions of particles in the right position. The charged particles willbe transported along the field. The three-phase high-voltage supply U-V-W generates a traveling field at the surface of the conveyor.13. The electrostatic conveying is independent of the sign of the charge. appears very promising (Inculet. both positively and negatively charged particles will transport in the same direction.

A sudden increase in the number of smoke particles in the ambient will be followed by an immediate increase in the particle density in the sensing Abrasive . 1977/78).ehi e elw Electrostatic conveying of tribo-electrified materials. in contrast to the low-response chamber (Inculet. the radioactive material may be replaced by a corona source of ions (Chapter 5) for sensitive detection of smoke particles.14. The construction is shown schematically in Figure 19. During the normal surveillance periods. two continuous and equal ion currents il and iz flow through the air of the two chambers made conductive by radiation from a radioactive material (Chapter 3). The design of such detectors is based on the fact that electrically charged particles of smoke have substantially lower mobility than ions. Figure 19. The current il flowsin the sensing chamber whose walls have so many holes that smo