A FAULT LOCATION METHOD FOR APPLICATION WITH CURRENT DIFFERENTIAL PROTECTIVE RELAYS OF SERIES-COMPENSATED TRANSMISSION LINE

M.M. Saha*, J. Izykowski, E. Rosolowski

*ABB AB, Sweden, murari.saha@se.abb.com Wroclaw University of Technology, Poland, jan.izykowski@pwr.wroc.pl Wroclaw University of Technology, Poland, eugeniusz.rosolowski@pwr.wroc.pl

Keywords: transmission line, series compensation, fault location, current differential protective relay.

Abstract
This paper presents an accurate algorithm for locating faults on a series-compensated line. It applies two-end currents and one-end voltage and thus can be applied with current differential protective relays. The algorithm applies two subroutines and the procedure for selecting the valid subroutine. The algorithm has been evaluated using the fault data of ATP-EMTP versatile simulations of faults on a seriescompensated transmission line. The presented example shows the validity of the presented algorithm and its high accuracy.

Varistors) for overvoltage protection (Figure 1). The MOVs are protected from overheating by a thermal protection (TP). The fault locator (FLA) is supplied from CTsA, VTsA measuring three-phase current and voltage at the bus A. Additionally, the three-phase phasors of currents {IB} measured at the remote bus B are provided via the communication channel of the differential relays.
A
SYSTEM A

SCs

dA [p.u.]
CTsA

dB [p.u.]

B
FB CTsB
SYSTEM B

~
VTsA

FA

Y
MOVs AIR-GAPs

~
{iB}

{iA}
DIFF REL A

TP

1 Introduction
Series-compensated lines [1], [5][9] usually are spreading over few hundreds of kilometres and are vital links between the energy production and consumption centres. Fault location on such lines for the inspection-repair purpose [3] with high accuracy is of vital importance. The method based on an impedance principle is the most popular for fault location. In particular, in [5, 8], the algorithms utilising one-end current and voltage measurements have been presented. In [9] use of two-end currents and voltages, measured synchronously [4] with the aid of PMUs, has been considered for fault location. In turn, application of the two-end unsynchronised measurements have been considered in [7]. Also the knowledge-based approaches have been proposed for the fault location purpose, as for example as in [1]. In this paper a new fault location algorithm for a seriescompensated line is presented. Innovative contribution relies on using two-end currents, while voltage from only one-end of the line as the fault locator input signals. This corresponds to including the fault location function into a current differential protective relay Figure 1: the fault locator FLA incorporated into the relay DIFF RELA. The presented fault location algorithm is designated for application to a line compensated with one three-phase bank of fixed series capacitors equipped with MOVs (Metal Oxide

{IB} d, RF

{IA}

{vA}

DIFF REL B

FLA

Figure 1: Schematic diagram for fault locator FLA associated with current differential relays of series-compensated line.

2 Fault location algorithm

A fault is of a random nature and therefore one needs to consider the faults appearing at both sides of the three-phase capacitor compensating bank Figure 1: faults FA and FB. In consequence, two subroutines: SUB_A, SUB_B are utilised for locating these hypothetical faults. In addition, the selection procedure is applied for indicating the valid subroutine yielding the result consistent with the actual fault. The compensating bank divides the line of the length l [km] into two line segments having the length: dSC [p.u.] and

(1 d SC ) [p.u.], as shown in Figures 2 and 3. The subroutines SUB_A (Figure 2), SUB_B (Figure 3) yield the per unit distance to fault: dFA, dFB, each related to the length of the particular line segment: ( d SC l ) or (1 d SC )l . Finally, one recalculates the relative distances dFA, dFB into the distances: dA, dB which also are expressed in [p.u.], but related to the whole line length: d A = d FA dSC (1 ) d B = d SC + (1 d FB ) (1 dSC )
(2 )

2.1 Fault location subroutine SUB_A The subroutine SUB_A is here derived under neglecting the line shunt capacitances effect, however, it can be accounted for in the next step.

I FAi = I Ai + I Xi

(7)

where: i=1positive-, i=2negative-, i=0zero-sequence component I Ai ith sequence component of current at the bus A;

A
FA I Ai

B_A V SU Vi

B
I Bi

X C Y
I Xi MOV

I Xi ith sequence component of current at the point X. When neglecting the line shunt capacitances it is equal to the current I Bi , and thus one obtains: I FAi = I Ai + I Bi
(8)

I Yi

V Ai dFA [p.u.] (1dFA) [p.u.] dSC [p.u.] (1 d SC ) [p.u.]

Figure 2: Subroutine SUB_A scheme of series-compensated line for ith symmetrical component under fault FA. The subroutine SUB_A for fault FA in the section AX is based on the following generalized fault loop model [6]: V Ap d FA Z 1LA I Ap RFA I FA = 0 (3) where: d FA unknown distance to fault [p.u.] on the section AX;

In order to assure possibly high accuracy of fault location, the following priority for usage of particular sequence components, as in Table 2, was proposed: for phase-to-earth and phase-to-phase faults: use of negative-sequence components; for phase-to-phase-to-earth faults: use of negative- and zero-sequence components; for three phase symmetrical faults: use of superimposed positive-sequence components.
Fault type aE bE
cE
ab, abE abc, abcE bc, bcE ca, caE

a1
1
0.5 j0.5 3

a2
1
0.5 + j0.5 3
0.5 j0.5 3 1.5 j0.5 3 j 3 1.5 j0.5 3

a0
1
1
1 0 0 0

RFA unknown fault resistance; V Ap , I Ap fault loop voltage and current; I FA total fault current (fault path current); Z 1LA positive-sequence impedance of the line section AX;
Note: Z 1LA = d SC Z 1L , where: impedance of the whole line.

0.5 + j0.5 3 1.5 + j0.5 3

j 3

1.5 + j0.5 3

Z 1L

positive-sequence

Table 1: Coefficients for composing signals (4)(5).

Fault type a E bE cE ab bc ca abE bcE caE

Fault loop voltage and current, composed accordingly to the fault type, are expressed as follows [6]:

a F1
0 0 0 0 0 0 0 0 0

a F2
3

a F0
0 0 0 0 0 0 j 3 j 3 j 3 0

V Ap = a1V A1 + a 2 V A2 + a 0 V A0 I Ap Z = a1 I A1 + a 2 I A2 + a 0 0LA I A0 Z 1LA

(4) (5)

1.5 + j1.5 3
1.5 j1.5 3 1.5 j0.5 3 j 3 1.5 j0.5 3

where a1 , a 2 , a 0 weighting coefficients (Table 1);

V A1 , V A2 , V A0 symmetrical components (positive-,

negative- and zero-sequence) of voltage from side A; I A1 , I A2 , I A0 symmetrical components of side A currents;

3 j 3
j2 3

Z 0LA zero-sequence impedance of the line section AX.

It is proposed to determine the total fault current in (3) using the following generalized fault model: I FA = a F1 I FA1 + a F2 I FA2 + a F0 I FA0 (6) where: a F1 , a F2 , a F0 share coefficients (Table 2). The ith sequence component of the total fault current is determined as a sum of the ith sequence components of currents from both ends of the faulted section AX:

*)

3+ j 3 abc 1.5 + j0.5 3 1.5 j0.5 3 *) abcE a F2 0 and there is no negative-sequence component

Table 2: Share coefficients used in fault model (6). In the case of three-phase balanced faults the total fault current is determined taking the superimposed positivesequence currents from both line ends A, B, respectively: superimp. superimp. I FA1 = I A1 + I B1 (9)

where the superimposed (superscript: superimp.) positivesequence currents are calculated by subtracting the pre-fault quantity (the superscript: pre) from the fault quantity:

I FA1 = ( I A1 I A1 ) + ( I B1 I B1 )
pre pre

(10)

In this way, an accurate calculation of the total fault current is assured since the positive-sequence components, for which the shunt capacitance effect is the most distinct, are excluded for all fault types (Table 2). After resolving (3) into the real and imaginary parts, and eliminating the unknown fault resistance ( RFA ), the sought fault distance ( d FA ) is determined as follows:
d FA = real(V Ap ) imag( I FA ) imag(V Ap ) real( I FA ) real( Z 1LA I Ap ) imag( I FA ) imag( Z 1LA I Ap ) real( I FA )
(11)

In contrast, at both sides of the compensating bank there is a different voltage due to presence of voltage drops across the SC&MOV in particular phases. These voltage drops can be calculated with use of the fundamental frequency equivalent impedance ZV, consisting of the resistance RV and reactance XV connected in series [7], as follows:

V V _ j = Z V_j (| I X_j |) I X_j

SUB_B

(15)

where: I X_j phasors of the currents flowing through the SC&MOV in phase j (phases are marked as: j=a, b or c);

X denotes the amplitude of the phasor X.

The ith sequence of voltage at the point Y is determined as:

V Yi = V Xi V Vi
where:
SUB_B

SUB_B

(16)

Having the fault distance calculated (11), the fault resistance RFA can be also determined.

V Vi voltage drop across the compensating bank for the ith sequence, obtained from the drops in phases: j=a, b, c (15).
In the considered case of the subroutine SUB_B, the generalised model describes the fault loop seen from the point Y towards the fault point FB (Figure 3): V FBp (d FB ) RFB I FB (d FB ) = 0 (17) where: V FBp (d FB ) fault loop voltage, composed accordingly to the fault type, obtained after the analytic transfer of this voltage from the point Y to the fault point FB; RFB unknown fault path resistance;

2.2 Fault location subroutine SUB_B

Transferring the voltage from the bus A towards the compensating bank, up to the point X (Figure 3), gives: V Xi = V Ai cosh( i d SC l) Z ci I Ai sinh( i d SC l) (12) where:

Z ci = Z iL Y iL , i = Z iL Y iL
' '

'

'

surge impedance and

propagation constant of the line for the ith sequence;

Z iL , Y iL impedance (/km) and admittance (S/km) of the line for the ith sequence.
'
'

I FB (d FB ) total fault current (fault path current);

The fault loop voltage transferred from Y to FB can be composed as the following weighted sum of the respective symmetrical components of voltage: V FBp (d FB ) = a1V FB1 (d FB ) + a 2 V FB2 ( d FB ) + a 0V FB0 (d FB ) (18)

A
I Ai

B_B V SU Vi

B
FB I Bi
I Yi
V Yi

X
I Xi
V Xi

C MOV

where: a 1 , a 2 , a 0 weighting coefficients, as in Table 1. Applying the distributed parameter line model, the ith symmetrical component of voltages from (18) are determined as follows: V FBi = V Yi cosh( i (1 d SC )l(1 d FB )) (19) Z ci I Yi sinh( i (1 d SC )l(1 d FB )) where: V Yi symmetrical components of voltage at the point Y, determined in (16); I Yi symmetrical components of current at the point Y, determined in (13)(14). Analogously as in (6) for the subroutine SUB_A, the total fault current ( I FB ) can be expressed as the following combination of its symmetrical components:

V Ai

(1dFB) [p.u.] dSC [p.u.]

dFB [p.u.]

(1 dSC ) [p.u.]

Figure 3: Subroutine SUB_B scheme of series-compensated line for the ith symmetrical component under fault FB. Transferring the ith symmetrical sequence current from the beginning of the line section (bus A) to the end point (X) of the un-faulted section AX gives: sinh ( i d SC l) V Ai I Xi = + cos( i d SC l) I Ai (13) Z ci If there is no internal fault in the compensating bank, then at both sides of the bank we have identical currents (Figure 3): I Yi = I Xi (14)

I FB (d FB ) = a F1 I FB1 (d FB ) + a F2 I FB2 (d FB ) + a F0 I FB0 (d FB ) (20) where:

' ' '

a F1 , a F2 , a F0 share coefficients. The recommended set of

these coefficients is delivered in Table 3 and differs from that for the subroutine SUB_A presented in Table 2.
Fault type a E bE cE ab bc ca abE bcE caE abc, abcE
*)

'

'

'

2.3 Selection procedure

First, the subroutine yielding the distance to fault outside the section range and/or the fault resistance of negative value is rejected. In vast majority of the cases this allows to select the valid subroutine. If this is not so, then one has to proceed with further selection. For this purpose, the circuit diagrams of the network for the negative-sequence, relevant for both subroutines are considered. In the case of three-phase balanced faults the superimposed positive-sequence components are considered. The particular subroutine is selected as the valid one if the determined remote source impedance has an RL character and its value is close to the actual source impedance: Z 1SB .

a F1 0 0 0 0 0 0 1.5 + j0.5 3

'

a F2
3

'

a F0
0 0 0 0 0 0 0 0 0

'

1.5 + j1.5 3 1.5 j1.5 3

1.5 j0.5 3 j 3

1.5 j0.5 3
1.5 j0.5 3 j 3

j 3 1.5 + j0.5 3

1.5 j0.5 3

3 ATP-EMTP evaluation
The 300-km, 400-kV transmission line compensated in the middle (dSC=0.5 p.u.), at the degree of 70%, was modelled using ATP-EMTP [2]. The line impedances were assumed as: Z1L=(8.28+j94.5) , Z0L=(82.5+j307.9) , and the line capacitances: C1L=13 nF/km, C0L=8.5 nF/km. The supplying systems were modelled with the impedances: Z1SA=Z1SB=(1.31+j15), Z0SA=Z0SB=(2.33+j22.5) ; the e.m.f. of the system B delayed by 10o with respect to the system A.

0 1.5 + j0.5 3 1.5 j0.5 3 *) there is no negative-sequence component under these faults and the coefficient can be assumed as equal to zero

Table 3: Share coefficients used for composing total fault current (20). Taking into account that the zero-sequence is eliminated (Table 3: a F0 = 0 for all fault types), one obtains the total fault current [8] in the form:
'

I FB (d FB ) =
where:

a F1 M 1 + a F2 M 2 cosh( 1 (1 d SC )ld FB )
' '

The MOVs with the common approximation i = P(v VREF ) q were modelled taking: P=1 kA, VREF=150 kV, q=23. The (21) model includes the Capacitive Voltage Transformers (CVTs) and the Current Transformers (CTs). The analogue filters with 350 Hz cut-off frequency were included. The sampling frequency of 1000 Hz was applied and the phasors were determined using the DFT algorithm.

Figure 4 presents the results for the example fault fault type: aE; fault distance: dB=0.5+ p.u. (just behind the Substitution of the total fault current (21) into the generalised compensating bank, as seen from the bus A); fault resistance: fault loop model (17) gives: RFB=25 . The continuous results were averaged within the ' ' fault period ( 30 50 ) ms. Both subroutines indicate the fault a F1 M 1 + a F2 M 2 V FBp (d FB ) RFB =0 (22) as occurring within their line sections: cosh( 1 (1 dSC )ld FB ) SUB_A: d A = 0.3017 p.u. < d SC p.u. (section AX); where: SUB_B: d B = 0.5026 p.u. > d SC p.u. (section YB). V FBp (d FB ) defined in (18)(19); Also the fault resistance values are positive (Figure 4b): M 1 , M 2 quantities defined in (21); SUB_A: RFA=47.92 ; ' ' SUB_B: RFB=24.45 . a F1 , a F2 share coefficients (Table 3). Continuing the selection, the remote source impedance values The derived fault location formula (22) is compact and covers SUB_A SUB_B have been according to both subroutines: Z 1SB , Z 1SB different fault types, what requires setting the appropriate fault type coefficients provided in Tables 1 and 3. There are calculated (Table 4). two unknowns in (22): distance to fault dFB and fault Remote source impedance Valid SUB_A SUB_B resistance RFB. After resolving (22) into the real and Subroutine Z 1SB Z 1SB Z 1SB imaginary parts, one of the known numeric procedures for 1.31+j15 41.60 +j57.08 2.30+j16.45 SUB_B solving nonlinear equations can be applied. It has been checked that the Newton-Raphson iterative method is a good Table 4: Fault location example selection of valid subroutine on base of estimated remote source impedance. choice for that.

V M i = I Bi + I Yi cosh( 1 (1 d SC )l) Yi sinh( 1 (1 dSC )l) ; Z c1 and where: i=1: positive-sequence or i=2: negative-sequence.

a)
Distance to fault (p.u.)

0.8

(dA)aver.= 0.3017 p.u. (dB)aver.= 0.5026 p.u.

fault measurements. Thus, the main limitations of the one-end fault location algorithms are in simple way overcome. The other important advantage of the developed algorithm relies on limiting the need for representing the compensating bank only for one subroutine (SUB_B). However, one can add the fault location function to the relays at both line ends and then the distance to fault will be calculated without representing the bank at all. Efficient procedure for selecting the valid subroutine has been presented as well. It allows reliable indication of the results, which are consistent with the actual fault. The presented fault location algorithm has been thoroughly tested using signals taken from ATP-EMTP versatile simulations of faults on a series-compensated transmission line. The presented example and the carried out evaluation with use of large number of the simulations show the validity and high accuracy of the developed fault location algorithm.

0.6

SUB_B: dB SUB_A: dA

0.4

0.2

0 0

10

20

30

40

50

60

Fault time (ms)

b)
Fault resistance ()

60 50 40 30 20

SUB_A: RFA

SUB_B: RFB

References
10 0 0 10

(RFA)aver.= 47.92 (RFB)aver.= 24.45

20 30 40 F a u l t ti m e ( m s ) 50 60

Figure 4: Fault location example: a) distance to fault, b) fault resistance. On the base of the impedances from Table 4, the subroutine SUB_B remains selected as the valid one. It yields estimation of the fault distance with around 0.25% error. Quantitative evaluation of the presented fault location algorithm has been performed for very wide range of fault conditions. For faults involving earth the fault path resistance up to 30 has been taken into account. Average fault location error was at the level of 0.5%, not exceeding 1%.

Conclusions
Use of two-end current and one-end voltage as the input signals for fault location on a series-compensated line has been considered. Such set of the fault locator input signals is of practical importance since it corresponds to use of two-end synchronised measurements of currents accomplished by current differential protective relays, with additional incorporation of the locally (i.e. at the bus where the fault location function is incorporated into the relay) measured three-phase voltage. Thus, the fault locator is designed as utilising the communication infrastructure of the relays, thus not demanding additional communication links. As a result of incorporating the fault location function into the differential relay its functionality is greatly increased. The subroutines of the algorithm have been formulated with use of the generalised fault loop model, leading to the compact formulae. It is important that the distance to fault calculations do not involve source impedances and the pre-

[1] W.J. Cheong, R.K. Aggarwal, A novel fault location technique based on current signals only for thyristor controlled series compensated transmission lines using wavelet analysis and self organising map neural networks, Proc. of 8-th IEE International Conference on Developments in Power System Protection, vol. 1, pp. 224227, (2004). [2] H. Dommel, ElectroMagnetic Transients Program, BPA, Portland, Oregon, (1986). [3] IEEE Std C37.114: IEEE guide for determining fault location on AC transmission and distribution lines, IEEE PES Publ., pp. 142, (2005). [4] IEEE Std. C37.118: IEEE standard for synchrophasors for power systems, IEEE PES Publ., pp. 165, (2006). [5] J. Sadeh, N. Hadjsaid, A.M. Ranjbar, R. Feuillet, Accurate fault location algorithm for series compensated transmission lines, IEEE Trans. on Power Delivery, vol. 15, No. 3, pp.10271033, (2000). [6] M.M. Saha, J. Izykowski, E. Rosolowski, Fault Location on Power Networks, Springer, London, (2010). [7] M.M. Saha, J. Izykowski, E. Rosolowski, P. Balcerek, M. Fulczyk, Accurate location of faults on seriescompensated lines with use of two-end unsynchronised measurements, Proc. of the 9th Int. Conference on Developments in Power System Protection, Glasgow, pp. 338343, (2008). [8] M.M. Saha, J. Izykowski, E. Rosolowski, B. Kasztenny, A new accurate fault locating algorithm for series compensated lines, IEEE Trans. on Power Delivery, vol. 14, No. 3, pp. 789797, (1999). [9] C.-S. Yu, C.-W. Liu, S.-L. Yu, J.-A. Jiang, A new PMU-based fault location algorithm for series compensated lines, IEEE Trans. on Power Delivery, vol. 17, No. 1, pp. 3346, (2002).