You are on page 1of 7

Paper accepted for presentation at 2003 IEEE Bologna Power Tech Conference, June 23th-26th, Bologna, Italy

Application of Distribution Line Carrier-based protection to prevent DG islanding: an investigating procedure


Roberto Benato, Roberto Caldon, Francesco Cesena
It has been studied the dependence of signal attenuation on the type of line, the number of lines entering a substation and on the presence of compensation capacitor banks installed at the primary substation bus. Simulations allow us to predict standing wave patterns and the possibility of assessing, with the presence of more receivers, an "optimal mismatching" for the HF transmission. II. ANTI-ISLANDING DLCBASED PROTECTION The scheme of fig.1 represents the general idea of using the DLC for preventing DG islanding. The superimposed signal is injected at MV busbar (e.g. 20 kV) by a signal generator (a transmitter Tx) through a suitable coupling device (CD). The signal propagates all over the MV network and can be usefully received at receiver location (Rx) through a coupling device. The coupling device has a bandwidth tuned on transmitter frequency. The presence of HV/MV and MV/LV transformers has a blocking effect of DLC signal (they act as a low-pass filter): the DLC propagation will be confined to the MV section. The transformer impedance as seen at MV bus slightly depends upon load variations and always remains very high [3] (between 4 and 50 kHz it is namely greater than 10 k). Consequently the transformers determine a kind of "unbundling" between MV and HV-LV networks. This allows us to consider only the presence of MV overhead and cable lines in computing the DLC propagation being the transmitter and receiver located at MV circuit. A guard frequency is injected by the transmitter before the line breaker and is detected, under normal conditions, by the receiver: when the breaker is opened for any reason there is a loss of guard frequency for the DG receiver.

Abstract The paper deals with the possibility of using the distribution line carrier (DLC) in order to prevent dispersed generation islanding. The feasibility of this technology applied to MV networks has been investigated from a theoretical as well as experimental point of view. A multiconductor matrix procedure based on bus admittance matrix allows us to predict the high frequency behaviour of MV lines (overhead and cable): the signal strength can be verified at any section of MV networks. Moreover two measurement campaigns have been performed on real MV networks in order to validate the theoretical results.

Index Terms Dispersed Generation, Islanding, Distribution Line Carrier, MV networks.

I. INTRODUCTION

ISPERSED generation connected to distribution system can supply unintentional system islands, isolated from the remainder of the network. Since these events pose an actual risk to safety and equipment they must be quickly detected and eliminated. Unfortunately, passive protections based on traditional measures (under/over-voltage, frequency) demonstrate, in particular system-operating condition, the possibility of failing the island detection (non-detection zones) [1]. In this paper the feasibility of a relay protection based on DLC signal is investigated for MV distribution systems. The proposal consists in the continuous injection, through a suitable coupling device, of a phase-to-phase superimposed carrier high frequency (a kind of guard signal) at the busbar of HV/MV substation [2]. In this way the signal can be transmitted at any section of the feeder so that any line tripping event will thus be immediately detected by each dispersed generator at the common coupled point, independently of the actual power flowing in the feeder before the loss of main supply. It has been implemented the high frequency models for overhead lines with ground return and cable lines (Carson's and Wedephol's theories) and has been simulated the transmission signal in different system configurations.

HV network

HV/MV

MV Lines
Line Breaker

MV/LV MV/LV LV loads

Rx

DG

Tx

Capacitor Bank MV/LV

Rx
DG
MV/LV

Overhead Line Cable Line

R. Benato, R. Caldon are with Department of Electrical Engineering, University of Padova, Italy Via Gradenigo, 6/A 35131. (e-mail: roberto.benato@unipd.it roberto.caldon@unipd.it) F. Cesena is with SELTA S.p.A. Strada Zappellazzo, 27/a 29010 Roveleto di Cadeo (PC), Italy. (e-mail: f.cesena@selta.it)

Tx Rx

Transmitter Location Receiver Location

Fig. 1. DLC-based protection scheme

0-7803-7967-5/03/$17.00 2003 IEEE

Another possibility is the location of the transmitting device after the line breaker: in this case every line tripping sends a trip signal to the receiver. This second arrangement allows the DG to return a confirmation signal of its disconnection from the network so that, after 400 ms, the automatic recloser can operate the line reclosure (to check whether the feeder is then clear of fault) without drawbacks for the DG. The arrangement may be also usefully exploited in the hypothesis of operating distribution networks in an isolated mode: the carrier line signal might ensure a continuous communication channel, for control data exchange, between Distribution Control Centre and peripheral microgrids [4]. III. MV NETWORK MODELLING FOR DLC PROPAGATION ANALYSIS With reference to the comprehensive modelling of multiconductor systems using elementary matrix cascades at rated frequency (50 or 60 Hz) [5], in this paper the procedure has been suitably generalized at DLC frequency. Let us consider a multiconductor system composed of n conductors. It may be represented as a cascade of m elementary cells of length L (ranging between 1 to 10 m), modelled by a lumped -circuit (see fig. 2) where the voltage column vectors uS, uR and the current column vectors iS, iSL, iST, iR, iRL, iRT may be considered.
1 2

iSL = iRL iL

1 ZL 1 - ZL

1 - ZL 1 ZL

uS uR u
(2)

YL (2n2n)

The shunt vector at sending-end iST and at receivingend iRT are


iST = iRT iT YTR Y T (2n2n) YTS uS uR u (3)

An adequate computation of YTS and YTR (nn) depends on the self and mutual shunt capacitive admittances of n conductors and is differently derived for the OHL and CL. The superimposition of (2) and (3) yields:
iS = YL + YT iR i Y (2n2n) uS uR u (4)

iS

iSL iST

1 2 3 ... n

3 ... n

iRL

iR iRT

uS

1 2 3 ... n

1 2 3 ... n

uR

that completely represents the steady-state regime of the elementary cell (of length ). A. Elementary cell cascades and line matrix Yl Once Y is computed, the line matrix Yl must be derived by means of cell cascades. If the line is rather long, the number of cells could be very high: the cell cascades could give computation drawbacks. The problem has been cleverly overcome using binary code. If each cell is represented by the same Y it is possible to compute the whole matrix Yl with a very low number of operations even if the cell number m is very high. For instance, if m = 1000, the "cascade procedure" [5] yields at first Y2 Y4 Y8 Y16 Y32 Y64 Y128 Y256 Y512. The representation of m in binary code shows that Yl might be evaluated as a cascade of Y512 Y256 Y128 Y64 Y32 Y8; totally 14 cascade operations are needed. Yl matrix synthesizes the multiconductor line behaviour as seen at endsections, and might be introduced in a matricial modelling of any interconnected systems. Alternatively the elementary cell cascades have been compared with modal analysis i.e. using eigenvalues and eigenvectors [8] giving a really good concordance.

TS

TR

Fig. 2 Line elementary cell

Being L sufficiently small, it is possible to lump shunt admittances at the ends of the cell (block TS and TR) allowing the separate study of longitudinal elements (block L, where iRL iSL). Self and mutual longitudinal impedances, which account for the earth return currents, can be obtained applying the Carson's theory [6] for OHL and Wedephol's theory [7] for cable line and form the matrix ZL (nn) that characterises the block L: uS-uR= ZL iSL , (1)

from which, considering that iRL iSL, and being ZL nonsingular 1 1 ZL uS- Z L uR=iSL
1 1 uS + Z L uR = iRL . - ZL

The two above algebraic equations can be synthesized in the subsequent matrix relation:

B. The bus admittance matrix Y Once each line has been represented by its own line matrix Yl , the bus admittance matrix Y is achieved by means of the generalized primitive Yprim and connection C matrices. Y= C t Yprim C .

The effectiveness of signal level is strictly linked to the media attenuation. The distribution system has many components and therefore is rather complex. Nevertheless, a general view of DLC problems is possible. The attenuation in MV networks is due to:
overhead and cable line attenuation; capacitor banks; different line segments; branching; mismatching and standing wave patterns. A. Overhead and cable line attenuation In order to highlight the different attenuation of overhead and cable lines, it is important to state beforehand some considerations on matching. For phase-to-phase transmission, the matching is done by adding the ZM impedance between the two phases. By considering a 3 km length EPR-insulated cable line (fig. 4) and by changing the location of the receiver, fig. 5 shows the attenuation with two different values of matching impedance namely ZM=3000 and ZM=100 . When the line is terminated in a very high (3000 ) impedance a severe standing wave conditions results.
x 3 km EPR-insulated cable S=185 mm2

C. Capacitor bank and Receivers The presence of shunt components such as capacitor banks (at MV busbar) or signal receivers can be inserted in the bus admittance matrix Y in the suitable position. Once the shunt components has been represented by its admittance matrix YCB or YRIC (33), the block-diagonal highly sparse matrix YS (of the same order of Y and with non-zeros only in the blocks where the shunt components are located) can be summed to Y giving: YTOT = Y + YS . D. High frequency regime The situation for a phase-to-phase transmitter is shown * is in fig. 3. Section 1 is the transmitter location whereas u11 the HF voltage signal between phase 1 and 2; ZT is the impedance matching coupling device. The bus impedance matrix ZTOT can be obtained by inversion of YTOT.
Active Sec. 1 1 ZT + u* 11 u*
12=

Passive

Tx
72 kHz phase-to-phase

Rx ZM

i11 2 0 3 i12

MV network
ZTOT=inv(YTOT)

Fig. 4 Cable line Configuration

50 40

ZM=3000 ZM=100

f= 72 kHz

Fig. 3. Phase to phase transmitter

By means of the matrix generalization of equivalent generator the injected current vector ia is completely defined by:
ia = (ZT + Z aa )1 u* a

Attenuation [db]

30 20 10 0

(10)

where Zaa is the partitioned matrix deriving from Z11. So the bus current vector i is obtained and, by means of u = ZTOT i , the bus voltages can be derived. IV. FEASIBILITY OF DLC PROTECTION The possibility of applying the proposed protection depends on the strength of signal at the receivers. A similar idea has been presented [9] using sub-harmonic signal on LV systems for PV power generation.

0.5

Receiver Location x [km]

1.5

2.5

Fig. 5 Attenuation as a function of receiver location

Otherwise when receiver location is fixed and the MV line terminates with a variable matching impedance ZM, the attenuation results as reported in fig. 6; it highlights the sensitivity of signal strength to the matching impedance ZM for overhead and cable lines.

14

Line lenght 2 [km]


CL paper - insulated S=185 mm
2

2 km
2

2 km
EPR-cable

12

Tx

EPR-cable

CL EPR - insulated S=185 mm OHL ACSR S=150 mm


2

72 kHz phase-to-phase

10

Cable Lines

R1 ZMR1

R2 ZMR2 = 460

Attenuation [dB]

Fig. 7 Single line diagram of two cable lines with two receivers

6 5 4 3 2 1 0 0 100 460 500 1000 1500 2000 2500 3000

OverHead Line

Impedance Matching ZM []

B. System compensation The presence of lumped devices such as capacitor banks determines a decreasing of signal strength: if the length of the cable line connecting the capacitor banks to the MV busbar is too short there is a dramatic worsening of the signal strength [2].
24 22

Fig. 6 Attenuation as a function of receiver impedance ZM

The following qualitative considerations can be drawn: a MV overhead line gives an attenuation less than 0.5 [dB/km]: (changing in the section, in the kind of conductor, in the height of the tower has a small influence on the attenuation); for ACSR (Aluminium Conductors Steel Reinforced) attenuation results a bit higher but always slight; a cable line gives an attenuation of about 1.54 [dB/km]. The least attenuation is due to polymeric insulated cable (EPR, HEPR) whereas the paper-impregnated cables give higher attenuation owing to the higher dielectric constant. An analogue behaviour can be observed by increasing of the cross-section S: it gives rise to higher attenuation even if the phenomena is less marked with respect to that of dielectric constant. Each kind of line shows a minimum of attenuation when the receiver transfers an impedance value next to the characteristic impedance (not equal because of phase-to-phase transmission): it should be noted that, even in the presence of slight mismatching, the attenuation is not so great. When more receivers are present in MV network (e.g. a widespread DG), the optimal matching for a receiver could be a lack of signal strength for the others. So the concept of an optimal mismatching zone (OMZ) can be considered: fig. 7 shows a clarifying example. The matching impedance of R2 remains constant (ZMR2=460) whereas ZMR1 changes its magnitude. The best matching for R1 causes a high attenuation in R2 (about 15 dB) as illustrated by fig. 8, being the major level signal derived by R1. It is possible however to single out an optimal mismatching zone ranging between 300 and 600 . If the same case is tackled for overhead lines, the OMZ ranges between 2500 and 3500 . If, as in real MV networks, there is a composite buildup of overhead and cable lines, a specific evaluation must be carried out.
Attenuation [dB]

20 18 16 14 12 10 8 6 4 0 100 200 300 400 500 600 ZMR1 [] 700 800 900 1000 OMZ ZMR2 =460 R2 15 dB

R1

Fig. 8 Signal attenuation for R1 and for R2 as a function of ZMR1

V. FIELD TESTS OF SIGNAL TRANSMISSION


A. Coupling devices. In order to transmit/receive high frequency signals to/from the Medium Voltage network, a special capacitive coupling device has been adopted. The constructive form and the relevant electric circuit are shown in fig. 9 and 10 respectively. This device should have a suitable matching impedance; appropriate values seem to be as follows:

i) ii)

For primary substation ZM = 46 ; For secondary substation: - if only cable lines are present, ZM = 460 ; - if only overhead lines are present, ZM = 3000 ; - if there is a combination of cable lines and overhead lines, then ZM = 1500 .

(transmitted signal 0 dB on 75 ) according to IEC Publication 481 on coupling devices for PLC systems.
B. Tests carried out on different network configuration

Fig. 9 Capacitive coupling device

In order to verify the feasibility of the proposed protection two measurement campaigns have been carried out on a 20 kV radial system. The composition of the network is generally rather complex so that two opposite situations have been chosen for experimental validation of the DLC-based protection. In the former the set of experiments was conducted over an approximately three-kilometre cable line (see fig. 11) with a dispersed generator at the receiving side. The same MV busbar supplies other six feeders with a mix of cable and overhead lines; in addition there is 95 m long cable line supplying a 3.6 Mvar capacitor bank. This network configuration is typical of urban feeder with a predominant presence of cable lines.

The value of the coupling capacitance depends upon the chosen bandwidth (BW). As an example, if BW = 10 kHz @ 3 dB, the value of coupling capacitance for a primary substation is given by: BW 10 10 3 Cc = = 6700 pF 2 f 2 Z M 2 72 10 3 2 46

Tx CD HV/MV
95 m

6 lines 2528 m
EPR-insulated cable S=185 mm2 3.6 Mvar Paper-insulated cable S=240 mm2

CD Rx
310 m

MV/LV

phase R Coupling Side 20 kV

Tr1

PG

CR Tr2

X
S
C: L: Tr1: PG: X: Tr2: S: Coupling Capacitance; Tuning Inductance; Insulating Transformer; Protective Gaps; Compensation Reactance; Impedance Trasformer; Protection Screen.

Transmitter/ Receiving Side

Fig. 11 Measurement Campaign I: urban feeder

phase S

The signal transmitter has been installed on the main busbar by means of a phase-to-phase capacitive coupling device. In the MV/LV substation, the signal receiver has been installed before the interface breaker; moreover, an event recorder has been inserted in order to detect any possible loss of guard frequency (as shown in fig. 12).

MV LINE
TRANSMITTER

Fig. 10 Electric circuit of the coupling device

MV LINE
RECEIVER

The components forming the coupling side are designed to assure the necessary insulation from the MV network and to determine the suitable bandwidth. Due to its high impedance at rated frequency, the coupling capacitance C (2000 8000 pF) represents a block for the rated 20 kV (maximum 24 kV) voltages. The tuning inductance L, together with the leakage inductance of TR1, must provide a resonant circuit at the chosen frequency f = 72 kHz. The transformer TR1 gives the galvanic insulation between the two sides. The components forming the transmitter/receiver side are galvanically unconnected from MV side and have a voltage reference at substation earthing grid. The surge arrester PG protects the components forming the transmitter/receiver side against the residual over voltages coming from MV side. The X capacitive reactance ensures the compensation of insulating transformer inductance. The signal attenuation at the resonant frequency is less than 1.5 dB

MV LINE
COUPLING DEVICE
COUPLING DEVICE MV LINE

RECORDER

Fig. 12 Coupling device arrangements in HV/MV and MV/LV substations

The loss of guard frequency must only result from a substation line tripping and not from a high attenuation level. In order to ensure a high degree of reliability of the protective relay, it is vital to avoid untimely trips. The event recorder has been active in DG location for two months: in this period of time four line trippings have been correctly recorded. In addition, a frequency response of the MV feeder has been performed by means of a pair of frequency selective voltmeters. The voltmeter in HV/MV substation has been

operated both as a signal generator (and the MV/LV substation one as a signal receiver) and vice versa. The monitored data are reported in fig.13 for frequency ranging from 50 and 110 kHz. It is worth noting that: there is an almost perfect bi-directionality; the least attenuation, according to the coupling device Bandwidth, is in the range 70 80 kHz.
50 45 40

The variation band for computed values considering relative dielectric constant r between 3.2 and 4 (typical range for paper-impregnated cables), is reported in fig. 16. It can be seen that the measured values are within the two-computed limit curves of dielectric constants confirming both the validity of line model and the uncertainty in the distribution line data [10].
33

r = 3 .2 4 M easu red v alu es

from HV/MV to DG sub. from DG to HV/MV sub.

31 29 27

Attenuation [dB]

35 30 25 20 15 50

Attenuation [dB]
60 70

25 23 21 19 17 15

Frequency [kHz]

80

90

100

110

13 11 9 7 70
71 72 73 74 75

Fig. 13 Frequency response of the urban feeder

Fig. 15 shows a registration of the sustained whitenoise between 70 and 90 kHz: the noise values range between 80 and 100 dB. However, the measurements have shown very low attenuation: for a 2,838 km long cable line, at 72 kHz the attenuation is about 20 dB (the transmitted signal is 14 dB and the received one is 6 dB). Fig. 16 shows a comparison between measured and computed values in a frequency range very close to CD bandwidth. This is of interest since the matrix algorithm considers the CD as ideal components. The agreement between measured and computed values is rather good considering the uncertainty in the distribution line data that are mainly the insulating material characteristics: dielectric constant () and loss factor (tan).

F requency [kH z]

Fig. 16 Comparison between computed and measured values

The second measurement campaign was conducted over a MV rural network (see fig. 17): the distance between the primary substation and the secondary substation is about 8 km. The MV/LV transformer is located at the end of a 450 m long lateral spur. The main characteristics of the other feeders of the MV busbar are reported in fig. 17: the lengths of the lines are extremely long, up to 50 km. Moreover, there is 50 m long cable line supplying a 3.6 Mvar capacitor bank.

Tx CD HV/MV

#1 #2 #4 #5 #3 50 m 4 lines 7542 m

MV/LV CD Rx
450 m 5667 m

0 -20 -40 Noise [dB] -60 -80 -100 -120

70

75

Frequency [kHz] 80

85

90

3.6 Mvar

# Line #1 #2 #3 #4 #5

Total Lenght Km 47.7 5.8 48.6 26.4 4.8

OHL lenght Km 45 1.7 44.7 23.8 0.4

CL lenght Km 2.7 4.1 3.9 2.6 4.4

Fig. 15 Registration of noise

Fig. 17 Measurement Campaign II: rural radial feeder

90 80 70 60 50 40 30

30

40

50

60 70 80 Frequency [kHz]

90

100

110

120

of applying the protection, a suitable analysis procedure has been developed. The good agreement of test results, performed on two different typologies of distribution systems, with the theoretical calculations demonstrates the accuracy of the procedure. In particular the matrix algorithm represents a powerful tool for investigating the signal strength in any location of a MV network. For each situation, it is able to foresee the performance of a DLC-based protection, identifying possible drawbacks. It is so possible to state the distance limit between Transmitter and Receiver (that is about 10 15 km for a typical radial branched MV feeder). If longer distances are needed, the use of repeaters must be taken into consideration in order to provide acceptable signal performance. VII. ACKNOWLEDGEMENTS Authors are grateful to Ing. S. Rogai and Ing. A. Cerretti (Central Direction - Engineering Dept.) and Ing. L. Cardin (Regional Responsible) of ENEL Distribuzione Spa, for their permission to perform tests on some distribution networks in Veneto Region. Besides authors thank R. Zago and E. Boldrin (ENEL personnel) for their helps in performing tests monitoring.

Fig. 18 Frequency response of the rural feeder

Fig. 18 shows the frequency response of the MV rural feeder. It clearly appears that this situation differs considerably from the former one and represents a high attenuation location. Consequently, this measurement campaign seems to be representative of the DLC transmission length limit without the use of repeater devices. In fact at 72 kHz the attenuation measured is about 69 dB (as fig. 19 clearly shows). The transmitted guard signal is +20 dB (on 75 ) and the received strength is 49.37 dB. Again we find a good agreement between theory and measurements.
IMP 150

Attenuation [dB]

VIII. REFERENCES
[1] R.Caldon, A.Scala, R.Turri: "Grid connected Dispersed Generation: investigation on anti-islanding protections behaviour", Proc. of 1st Int. Symposium on Distributed Generation, June 11-13 2001, Royal Institute of Technology, Stockholm. [2] R.Benato, R.Caldon, F.Cesena: "Carrier Signal-Based Protection To Prevent Dispersed Generation Islanding On MV Systems", Proceedings of 17th International Conference on Electricity Distribution, CIRED2003, 12-15 May 2003, Barcelona, Spain. [3] R.Hemminger, L.J.Gale, F.Amoura, B. ONeal: "The effect of Distribution Transformers on Distribution Line Carrier Signals", IEEE Trans. on Power Delivery, Vol. PWRD-2, No. 1, January 1987, pp. 3640. [4] S. Massucco, B. Delfino, F. Silvestro, A. Bertani, C. Bossi, T. Mayer, I. Wasiak, E. Metten, N. Lewald: Electrical energy distribution networks: actual situation and perspective for distributed generation, 17th CIRED, Barcelona, 12-15 may 2003. [5] R. Benato, R. Caldon, A. Paolucci, R. Turri: "Resonance Phenomena On LineToGround Fault Current Harmonics In Mv Networks", Proc. of 8th International Conference on Harmonics and Quality of Power ICHQP, Athens GREECE, 1416 October 1998, Vol. I, ICHQP 98 254, pp. 359364. [6] CCITT: Directives concerning the protection of telecommunication lines against harmful effects from electric power and electrified railway lines. Geneva, 1989. [7] L.M. Wedepohl, D.J. Wilcox, "Transient analysis of underground power transmission systems", Proc. IEE, Vol.120, No.2, February 1973, pp. 253-260. [8] J. Arrilaga, D.A. Bradley, P.S.Bodger, Power system harmonics, John Wiley & Sons, 1985. [9] M.E. Ropp, K. Aaker, J. Haigh, N. Sabbath: "Using Line Carrier Communication to Prevent Islanding", Proceedings of 28th IEEE Photovoltaic Specialists Conference, 15-22 Sept. 2000 Anchorage, AK, USA. [10] Sarosh N. Talukdar, J.Carlos Dangelo "Uncertainty in Distribution PLC attenuation models", IEEE Trans. On Power Apparatus and Systems, Vol. PAS-99, No. 1 Jan./Feb. 1980, pp. 328-334.

0 -10
Signal Strength [dB]

BWTH: 25 Hz START: 70.000 kHz STOP : 78.000 kHz

Cursor -49.37 dB / 72.000 kHz

-20 -30 -40 -50 -60 -70 -80 70 71 72 73 2 khZ/Div 74 75 76 Frequency [kHz] 77 78

Fig. 19 Signal strength registered at the receiving site of the rural feeder

VI. CONCLUSIONS A novel protection method to prevent DG islanding has been proposed. This is based on a line carrier high frequency signal injected in the MV radial system and appear to be less expensive compared to other methods based on telephone cable signals. In order to assess for a given system configuration the level attenuation of the signal and the consequent possibility

You might also like