Sie sind auf Seite 1von 7

ACCURATE LOCATION OF FAULTS ON THREE-TERMINAL LINE

WITH USE OF THREE-END UNSYNCHRONISED MEASUREMENTS


Jan Izykowski Przemyslaw Balcerek Murari Mohan Saha
Eugeniusz Rosolowski, Rafal Molag Marek Fulczyk ABB AB
Wroclaw University of Technology, Wroclaw, Poland ABB Corporate Research Center Västerås, Sweden
jan.izykowski@pwr.wroc.pl Krakow, Poland murari.saha@se.abb.com
eugeniusz.rosolowski@pwr.wroc.pl przemyslaw.balcerek@pl.abb.com
rafal.molag@wp.pl marek.fulczyk@pl.abb.com

Abstract – This paper presents a new fault location Limited availability of measurements for fault
algorithm for three-terminal line utilising unsynchronised location on three-terminal lines was considered in [5]–
measurements of three-end currents and voltages. The [7]. In [5]–[6] ensure complete immunity to saturation
distributed parameter line model is strictly considered for
formulating three subroutines of the algorithm. Efficient of current transformers (CTs), since only voltage signals
procedure for selecting the valid results is introduced. The are utilised for fault location. The method presented in
presented results of the ATP-EMTP evaluation prove the [7] is based on using minimal amount of measurements.
validity of the fault location algorithm and its high In [8] use of three-phase current from all three line
accuracy. terminals and additionally three-phase voltage, acquired
at the terminal at which the fault locator is installed, has
Keywords: three-terminal power line, fault location, proposed. Such specific availability of measurements
three-end unsynchronised measurements, ATP-EMTP has been assumed with the aim of simple supplementing
of the fault location function to current differential
1 INTRODUCTION relays protecting a three-terminal line.
In this paper, an algorithm, which utilises
Accurate location of faults on overhead power lines
unsynchronised measurements of currents and voltages
for the inspection-repair purpose [1]–[9] is of vital
at all three terminals is presented. It is assumed, that
importance for operators and utility staff for expediting
analytical synchronisation of such measurements has to
service restoration, and thus to reduce outage time, be performed without resorting to synchronising with
operating costs and customer complains. the pre-fault measurements. In order to assure high
Different fault location algorithms for three-terminal accuracy of fault location, the distributed parameter line
lines (Fig. 1), have been developed so far [1]–[8]. The model is strictly utilised.
cited algorithms are based on an impedance principle, The paper starts with derivation of the fault location
making use of the fundamental frequency voltages and algorithm. Then, the sample results of extensive ATP-
currents. They differ each other mainly with respect to EMTP [10] evaluation of the algorithm are presented
the applied input signals of the fault locator. and discussed.
In [1] usage of synchronised measurements of
currents and voltages from all three terminals has been 2 FAULT LOCATION ALGORITHM
considered. Distributed parameter models of the line
sections have been utilised there. This assures high 2.1 Basics of fault location algorithm
accuracy of fault location, and the faulted line section is The circuit diagram of a three-terminal line for
also reliably indicated [1]. considering the fault location using three-end
Use of three-end unsynchronised measurements of unsynchronised measurements is presented in Fig. 1.
Different types of symmetrical components of the
currents and voltages has been considered in [2]. The
signals can be processed in the algorithm. In Fig. 1, and
lumped models of the line sections were applied there
also further, the signal type is marked by the subscript i.
and it was also implied that the error resulting from such
In Fig. 1 the measurements from the end B are taken
simplification is minimised due to the redundancy of the as the basis, while the measurements from the buses A
fault location equations. Yet another utilisation of three- and C have to be analytically synchronised, with use of
end unsynchronised measurements has been proposed in
the synchronisation operators e jδ Ai and e jδ Ci ,
[3], where exchanging the minimal amount of
information between the line terminals over a protection respectively. Thus, two additional unknowns: angles δAi
channel was considered. and δCi appear, in comparison to the case of
synchronised measurements [1].
In turn, a demand and importance of developing fault
The fault can occur at any of three line sections
location algorithms for three-terminal lines utilising only
(faults: FA, FB or FC in Fig. 1). Therefore, there are
two-end synchronised measurements of voltages and
three hypotheses regarding placement of the fault.
currents has been stated in [4].

16th PSCC, Glasgow, Scotland, July 14-18, 2008 Page 1


FC
d Ci C
SYSTEM C
jδCi
I Cie
V Ci e jδCi
A FA FB
B
d Ai d Bi
SYSTEM A SYSTEM B
I Ai e jδAi T I Bi
V AiejδAi V Bi

Figure 1: Fault location on three-terminal line using three-end unsynchronised measurements of the i–th symmetrical component of
voltages and currents.

The distance to the particular fault, counted from the basis, while the phasors from the end S are analytically
particular bus (A, B or C) up to the fault point is marked synchronised – with use of the synchronisation
as: dAi, dBi, dCi, respectively. All three hypotheses operator:
regarding indication of the faulted line section are
e jδSi = cos(δSi ) + j sin(δSi ) (1)
considered and after calculating the distances to fault
(dAi, dBi, dCi) by using these three subroutines, the The unknowns of the fault location algorithm are:
judgement on selecting only one solution (the valid d i – distance to fault from bus S (p.u.),
solution), which is consistent with the actual fault, is δSi – synchronisation operator.
performed. All three subroutines of the presented fault
If the lumped line model (Fig. 2) is considered, the
location algorithm are based on the two-end
unsynchronised fault location algorithm presented in [9]. parameter: Z iL – impedance of the line SR, for the i-th
symmetrical component, is utilised. For the distributed
2.2 Fault location algorithm for two-terminal line
parameter model (Fig. 3), the line parameters: Z ci , γ –
The algorithm from [9] is formulated for the i-th i
symmetrical components and considers both the lumped surge impedance and propagation constant of the line
(Fig. 2) and the distributed parameter (Fig. 3) models of SR having the length l , for the i-th symmetrical
a two-terminal power line. component, is considered.
S F R In [9] the following compact formula for the
I Si e jδSi d i Z iL (1 – di ) Z iL IRi
synchronisation angle has been derived (by comparing
the voltage at the fault point, viewed from both ends):
Ai cos( δSi ) + Bi sin( δSi ) = Ci (2)
V Si e jδSi VRi
where:
Figure 2: Lumped model of faulted line for considering ( (
Ai = imag Z *iL (V Ri − Z iL I Ri ) I *Si − V Si I *Ri , ))
B = real(− Z ((V )) ,
unsynchronised two-end fault location. * *
i iL Ri − Z iL I Ri ) I Si + V Si I *Ri
S F R C = imag(− Z ((V *
− Z iL I Ri ) I *Ri − V Si I )) ,
*
I Si e jδSi IRi i iL Ri Si

x* – conjugate of x .
di [p.u.] (1 – di ) [p.u.] There are two solutions of (2) for the synchronisation
V Si e jδSi VRi angle: δSi_1 , δSi_2 , if the reasonable angle range from

Figure 3: Distributed parameter model of faulted line for (–π) to (+π) is considered [9].
considering unsynchronised two-end fault location. After determination of the synchronization angle from
(2), the distance to fault can be calculated as:
The input signals of the two-end unsynchronised fault  V e j δSi _ 1( 2 ) − (V − Z I ) 
location algorithm are: V Si , V Ri , I Si , I Ri – voltage d i _ 1( 2) = real Si Ri iL Ri 
(3)
 Z ( I e
j δSi _ 1( 2 )
+ I 
and current phasors from the terminals S and R,  iL Si Ri ) 
respectively, for the i-th symmetrical component. Digital There are two solutions for the synchronisation angle,
measurements of the phasors from two line ends are and two solutions are for the distance to fault (3).
considered as the unsynchronised. Therefore, the Applying the distributed parameter line model, the
measurements from the terminal R are assumed as the voltage at the fault point F, viewed from the terminals S

16th PSCC, Glasgow, Scotland, July 14-18, 2008 Page 2


and R, are compared, giving: Comparing the transferred voltage (7) (after
V FSi (d i , δSi ) − V FRi (d i ) = 0 (4) synchronising with use of the synchronisation operator
where: ( e jδ CAi ) and the voltage (5) one obtains the following
V FSi (d i , δSi ) = V Si cosh( γ ld i ) − Z ci I Si sinh(γ i ld i ) e jδSi formula for the sought synchronisation operator:
i
V Bi cosh(γ i l LB ) − Z ci LB I Bi sinh(γ i l LB )
V FRi (d i ) = V Ri cosh(γ i l(1 – d i )) − Z ci I Ri sinh(γ i l(1 – d i )) e jδCAi = LB LB
(9)
V Ci cosh(γ i l LC ) − Z ci LC I Ci sinh(γ i l LC )
LC LC
γ , Z ci , l – propagation constant, surge impedance of
i Taking the signals from the end A and the signals
line for the i-th sequence, line length. transferred analytically to the tap point T one complets
In [9] the way of solving (4) with use of the Newton- the required set of the input signals for the subroutine
Raphson iterative calculations has been presented. As SUB_A. Applying the two-end algorithm from [9] (the
the starting point of these calculations, the results signals specified on the left side of the below
obtained according to (2)–(3) are utilised. assignments) for performing the calculations of the
2.3 Subroutine SUB_A for locating faults on section AT subroutine SUB_A (the signals specified on the right
Fig. 4 presents the circuit diagram used for deriving side of the below assignments), we get:
the subroutine SUB_A. This subroutine is designated for V Si e jδSi := V Ai e jδAAi (10)
locating faults FA on the line section AT.
I Si e jδSi := I Ai e jδAAi (11)
jδ CAiC
I Ci e V Ri := V TBi (12)
I Ri := I TBi + I TCi e jδ CAi (13)
jδ CAi
V Ci e
e jδ CAi – synchronization operator determined in (9).
A FA Processing the signals for the subroutine SUB_A,
I TCi e jδ CAi B
I Ai e jδ AAi T I Bi assigned in (10)–(13), one has to take the line
parameters relevant for the line section AT. This
I TBi
d Ai [p.u.] subroutine will yield two solutions for the distance to
V Bi fault dAi_1, dAi_2 and also two solutions for the
V Ai e jδ AAi V TBi = V TCi e jδ CAi jδ jδ
synchronization operator e iA_1 , e iA_2 . Therefore the
Figure 4: Subroutine SUB_A – circuit diagram for the i-th
selection will be required (provided in Section 2.6).
symmetrical component under fault at section AT.
2.4 Subroutine SUB_B for locating faults on section TB
Transferring analytically the signals from the end B Fig. 5 presents the circuit diagram which is used for
towards the tap point T (Fig. 4) one obtains: deriving the subroutine SUB_B, designated for locating
V TBi = V Bi cosh(γ i l LB ) − Z ci LB I Bi sinh(γ i l LB ) (5) faults FB on the line section BT.
LB LB
−V Bi C
I TB i = sinh( γ iLB l LB ) + I Bi cosh( γ iLB l LB ) (6) I Ci e jδ CBi
Z ciLB
where:
V Ci e jδ CBi
Z ci LB , γ – surge impedance and propagation
i LB
A FB
constant of the line section BT for the i-th sequence, B
I Ai e jδ ABi I TCi e jδ CBi I Bi
l LB – length of the line section BT.
Similarly, as in (5)–(6), the signals from the end C I TAi e jδ ABi T
can be transferred analytically towards the tap point T: d Bi [p.u.]
V Ai e jδ ABi V e jδ ABi = V e jδ CBi V Bi
V TCi = V Ci cosh(γ i l LC ) − Z ci LC I Ci sinh(γ i l LC ) (7) TAi TC i
LC LC
−V Ci Figure 5: Subroutine SUB_B– circuit diagram for the i-th
I TC i = sinh( γ iLC l LC ) + I Ci cosh( γ iLC l LC ) (8)
Z ciLC symmetrical component under fault at section BT.
where: Comparing the voltages at the tap point T (Fig. 5),
Z ci LC , γ – surge impedance and propagation one obtains the following relations between the
i LC
constant of the line section CT for the i-th sequence, synchronisation operators:
l LC – length of the line section CT. e jδCBi = (V TAi V TCi )e jδABi (14)
In Fig. 4 the signals (7)–(8) are synchronised V Ai cosh(γ l ) − Z ci LA I Ai sinh(γ l LA )
analytically to the common basis (the measurements e j(δCBi −δ ABi ) = i LA LA i LA

acquired at the bus B are assumed as the basis) with use V Ci cosh(γ l ) − Z ci LC I Ci sinh(γ l LC )
i LC LC i LC
jδ CAi where: (15)
of the synchronisation operator e .

16th PSCC, Glasgow, Scotland, July 14-18, 2008 Page 3


γ , Z ciLA – propagation constant and surge impedance 2.6 Selection of valid results
iLA
The three subroutines (SUB_A, SUB_B, SUB_C) are
of section AT for the i-th sequence, applied in the presented fault location algorithm
l LA – section length. designed for application to a three-terminal line. Taking
Applying the two-end algorithm [9] (the signals a particular type of symmetrical components (denoted
specified on the left side of the below assignments) for by the subscript ‘i’) each subroutine yields two solutions
performing the calculations of the subroutine SUB_B for the distance to fault:
(the signals specified on the right side of the below - SUB_A: (dAi_1 and dAi_2),
assignments), we get: - SUB_B: (dBi_1 and dBi_2),
V Si e jδ Si := V Bi (16) - SUB_C: (dCi_1 and dCi_2).
j δ Si
Each subroutine also yields two solutions for the
I Si e := I Bi (17) synchronisation angle, used for synchronising the
V Ri := V TAi (18) measurements from the ends A and C:
- SUB_A: (δAAi_1 and δAAi_2), (δCAi_1 and δCAi_2),
I Ri := I TAi + I TCi e j(δ CBi −δ ABi ) (19) - SUB_B: (δABi_1 and δABi_2), (δCAi_1 and δCBi_2),
One has to take into account that on the right sides of - SUB_C: (δACi_1 and δACi_2), (δCCi_1 and δCCi_2).
the assignments (18)–(19) there are the signals, which There is a need for selecting:
are synchronised to the line end A, and not to the end B - valid subroutine: SUB_A or SUB_B or SUB_C,
as assumed (Fig. 1). Therefore, the synchronisation - valid solution of the subroutine, which has been
angle yielded by the Subroutine SUB_B is: indicated earlier as the valid one: the first solution
(δ Si ) SUB_B = −δ ABi (20) (marked with ‘1’ at the last position of the
subscript) or the second solution (marked with ‘2’).
2.5 Subroutine SUB_C for locating faults on section CT The following selection indexes are applied:
Fig. 6 presents the circuit diagram which is used for abs(V TBi )
deriving the subroutine SUB_C – designated for SEL LAi = (26)
abs(V TCi )
locating faults on the line section CT.
abs(V TCi )
FC C SEL LBi = (27)
I Ci e jδCCi abs(V TAi )
abs(V TAi )
d Ci [p.u.] SEL LCi = (28)
abs(V TBi )
jδCCi
V Ci e If there is a fault at the section AT, then the
A B magnitudes of the phasors of voltages (for the
I Ai e jδ ACi I TAi e jδACi I TBi I Bi considered i-th symmetrical component) at the tap point
T T, transferred analytically from the line ends B: | V TBi |
and C: | V TBi | , have to be identical (in practice very
V Ai e jδ ACi V TAi e jδACi = V TBi V Bi
close each other). Therefore, for the fault at the section
AT, we have: SEL LA i ≅ 1 , while the selection indexes:
Figure 6: Subroutine SUB_C– circuit diagram for the i-th
symmetrical component under fault at section CT. SEL LBi ≠ 1 , SEL LCi ≠ 1 . Analogously is for faults
occurring in the other line sections. Moreover, when a
Comparing the voltages at the tap point T (Fig. 6), given subroutine yields all results for the distance to
which are obtained by the analytical transfer, one fault (both the first and second solutions) outside its
obtains the following relation range: from 0 to 1 [p.u.], then this subroutine has to be
V Bi cosh(γ l LB ) − Z ci LB I Bi sinh(γ l LB ) rejected.
e jδ ACi = i LB i LB
(21)
V Ai cosh(γ l LA ) − Z ci LA I Ai sinh(γ l LA ) Selection of the valid subroutine (SUB_A or SUB_B
i LA i LA
or SUB_C) and the valid solution (the 1-st or the 2-nd)
Applying the two-end algorithm [9] (the signals can be performed also by repeating the calculations for
specified on the left side of the below assignments) for the other type of symmetrical components (say, the k-th
performing the calculations of the subroutine SUB_C type, where k ≠ i ), and then comparing the results of the
(the signals specified on the right side of the below
distance to fault for these two types (i-th and k-th) of
assignments), we get:
symmetrical components. The estimated distance to fault
V Si e jδ Si := V Ci (22) with using the i-th type, coinciding with the result for the
jδSi k-th type (with some error margin) is selected as the
I Si e := I Ci (23)
valid result – being consistent with the actual fault.
V Ri := V TBi (24) The described selection procedure is explained in
Section 3 in relation to the presented fault example.
I Ri := I TBi + I TAi e jδ ACi (25)

16th PSCC, Glasgow, Scotland, July 14-18, 2008 Page 4


3 ATP-EMTP ALGORITHM EVALUATION The equivalent sources: Z 1SA = (0.65 + j3.69)Ω ,
ATP-EMTP software program [10] was applied to Z 0SA = (1.16 + j6.57) Ω , Z iSB = 2Z iSA , Z iSC = 3Z iSA
evaluate performance of the developed fault location
algorithm. The modelled 110 kV test network includes were also included. Phase angles of the sources were
the line sections – AT: 100 km, BT: 80 km, CT: 50 km, assumed as equal to Æ source A: 0o, source B: –30o and
having the positive- and zero-sequence source C: –15o.
' In Fig. 7–Fig. 10 the fault locator input signals and
impedances: Z 1L = (0.0276 + j0.315) , the fault location results for the example fault are
Z '0L = (0.275 + j1.027) (Ω/km). shown.

4000
a b c
0.8
SIDE A – Phase voltages: vA [105 V]

SIDE A – Phase currents: iA [A]


2000
0.4

a b c
0 0

–0.4
–2000

–0.8

–4000
0 20 40 60 80 100 120 0 20 40 60 80 100 120
Time [ms] Time [ms]

600
a b c a b c
SIDE B – Phase voltages: vB [105 V]

0.8
400
SIDE B – Phase currents: iB [A]

0.4
200

0 0

–200
–0.4

–400
–0.8
–600
0 20 40 60 80 100 120 0 20 40 60 80 100 120
Time [ms] Time [ms]

600
a b c
0.8
SIDE C – Phase voltages: vC [105 V]

SIDE C – Phase currents: iC [A]

400

0.4
200

a b c
0 0

–200
–0.4

–400
–0.8

–600
0 20 40 60 80 100 120 0 20 40 60 80 100 120
Time [ms] Time [ms]

Figure 7: The fault location example – waveforms of fault locator input signals.

16th PSCC, Glasgow, Scotland, July 14-18, 2008 Page 5


POSITIVE-SEQUENCE QUANTITIES – 1st SOLUTION POSITIVE-SEQUENCE QUANTITIES – 2nd SOLUTION
2.5
SUB_A
SUB_B
2 2
d A1_1(30 ÷ 50)ms = 2.1369 p.u.
d B1_1(30 ÷ 50)ms = –0.2384 p.u.

Distance to fault [p.u.]


Distance to fault [p.u.]

1.5 1.5
dC1_1(30 ÷ 50)ms = 0.6848 p.u.
SUB_C
1 1

0.5 SUB_C 0.5


SUB_A
0 0
d A1_2(30 ÷ 50)ms = 0.2009 p.u.
SUB_B dB1_2(30 ÷ 50)ms = 2.0383 p.u.
–0.5 –0.5
dC1_2(30 ÷ 50)ms = 1.0075 p.u.
–1 –1
0 10 20 30 40 50 60 0 10 20 30 40 50 60
Post-fault time [ms] Post-fault time [ms]

Figure 8: The fault location example – the 1st and 2nd solutions for the distances to fault applying positive-sequence quantities.

NEGATIVE-SEQUENCE QUANTITIES – 1st SOLUTION NEGATIVE-SEQUENCE QUANTITIES – 2nd SOLUTION


SUB_B

1.6 SUB_B 6
Distance to fault [p.u.]

d A2_2(30 ÷ 50)ms = –0.5480 p.u.


Distance to fault [p.u.]

dB2_2(30 ÷ 50)ms = 1.6408 p.u.


1.2 4
SUB_C
dC2_2(30 ÷ 50)ms = 7.2297 p.u.

0.8 2
d A2_1(30 ÷ 50)ms = 0.1998 p.u. SUB_C
dB2_1(30 ÷ 50)ms = 1.4909 p.u.
0.4 dC2_1(30 ÷ 50)ms = 0.9997 p.u. 0
SUB_A
SUB_A
0 –2
0 10 20 30 40 50 60 0 10 20 30 40 50 60
Post-fault time [ms] Post-fault time [ms]

Figure 9: The fault location example – the 1st and 2nd solutions for the distances to fault applying negative-sequence quantities.

POSITIVE-SEQUENCE QUANTITIES – 2nd SOLUTION POSITIVE-SEQUENCE QUANTITIES – 2nd SOLUTION


d A1_2(30 ÷ 50)ms = 0.2009 p.u.
40
δCA1_2
0.4
dA1_2
Synchronisation angle [o]
Distance to fault [p.u.]

30
0.3

20 δAA1_2
0.2

0.1 10
δ AA1_2(30 ÷ 50)ms = 17.94 o
δ CA1_2(30 ÷ 50)ms = 35.96 o
0 0
0 10 20 30 40 50 60 0 10 20 30 40 50 60
Post-fault time [ms] Post-fault time [ms]

Figure 10: The fault location example – the results (the 2nd solutions for use of the positive-sequence quantities) for the distance to
fault and for the synchronisation angles of measurements from the ends A and C, selected as the valid results.

The specifications of the fault from the example - fault distance: dA=0.2 p.u.
presented in Fig.7–Fig. 10 are as follows: The current and voltage signals from the line end A
- fault type: a-g, were intentionally delayed by a single sample, while the
- fault resistance: 10 Ω, signals from the end C by two samples. Such delays for
- faulted section: AT, the applied sampling frequency of 1000 Hz correspond

16th PSCC, Glasgow, Scotland, July 14-18, 2008 Page 6


to the case of unsynchronised measurements with the model is strictly taken into account for deriving the
synchronisation angles equal to: δA=18o, δC=36o. algorithm subroutines. The calculations are performed
For evaluating the accuracy of the presented fault for a particular type of symmetrical component, as for
location algorithm itself, the errorless current and example for the positive-sequence.
voltage transformers were modelled. Analogue anti- There is a need for selecting the valid result, which is
aliasing filters with the cut-off frequency of 350 Hz consistent with the actual fault. The selection indexes
were included (1 kHz sampling frequency was used). can be used for that. Also, the selection can be
Full-cycle Fourier orthogonal filters were applied. The performed by additional use of the other type of
obtained continuous time results were averaged within components, as in the presented example.
the post-fault interval: from 30 up to 50 ms. The presented results of the ATP-EMTP evaluation
Input signals for the presented example are shown in prove the validity of the derived algorithm. The obtained
Fig. 7. In Fig. 8 there are the results of a distance to fault accuracy of fault location is comparable with the
calculation using the positive-sequence quantities: the 1st achieved when using synchronised measurements.
solution indicates the fault on the section CT (SUB_C:
dC1_1=0.6848 p.u.), while the 2nd solution the fault on the REFERENCES
section AT (SUB_A: dA1_2=0.2009 p.u.). Using the [1] R.K. Aggarwal, D.V. Coury, A.T. Johns and A.
selection indexes (26)–(28), the subroutine SUB_A Kalam, “A practical approach to accurate fault
undergoes indicated as the valid one, for which the 2nd location on extra high voltage teed feeders”, IEEE
solution yields the result dA1_2 being very close to the Trans. Power Delivery, Vol. 8, pp. 874–883, July
actual distance to fault: dA=0.2 p.u. Also, the selection 1993
can be performed by repeating the calculations for the [2] A.A. Girgis, D.G. Hart and W.L. Peterson, “A new
other type of sequence component, as for example for fault location technique for two-and three-terminal
the negative-sequence. Confronting the results from Fig. lines”, IEEE Trans. Power Delivery, Vol. 7, No.1,
8 with those shown in Fig. 9, one gets that the result: pp. 98–107, January 1992
dA1_2=0.2009 p.u. is confirmed by the result:
[3] D. A. Tziouvaras, J. Roberts and G. Benmmouyal,
dA2_1=0.1998 p.u. (as being the most closer) as the valid
“New multi-ended fault location design for two- or
one. In Fig. 10 the valid results are shown. The errors
three-terminal lines”, Proceedings of 7th
are very low: 0.0009 p.u. (for the fault distance) and
International IEE Conference on Developments in
0.06o (for the angle δA=18o) and 0.04 (for δC=36o). Power System Protection, pp. 395–398, April 2001
Besides the presented example, an extensive
[4] Y. Lin, C. Liu and C. Yu, “A new fault locator for three-
evaluation study has been performed. Different
terminal transmission lines using two-terminal
specifications of faults have been considered in the
synchronized voltage and current phasors”, IEEE Trans.
study. For example, in the synthetic evaluation for the
Power Delivery, Vol. 7, No.3, pp. 452–459, July 2002
population of 108 fault cases:
- 4 fault types (a-g, a-b, a-b-g, a-b-c), [5] J.F. Minambres, I. Zamora, A.J. Mazon, M.A.
- faults on each of 3 sections (AT, BT, CT), Zorrozua and R. Alvarez-Isasi, “A new technique,
- faults at 9 locations (from 0.1 to 0.9 p.u.) based on voltages, for fault location on three-
the following maximal errors in fault distance terminal transmission lines”, Electric Power Systems
calculation were achieved: Research 37, pp 143–151, 1996
- for the lumped line model (starting point for [6] S.M. Brahma, “Fault Location Scheme for a Multi-
iterative calculations with the distributed parameter Terminal Transmission Line Using Synchronized
line model): 1.01% Voltage Measurements”, IEEE Trans. on Power
- for the distributed parameter line model: 0.14%. Delivery, Vol. 20, No. 2, pp. 1325–1331, April 2005
Note: a fault resistance of 10 Ω for single phase faults [7] J. Izykowski, R. Molag, E. Rosolowski and M.M.
and 1 Ω for inter-phase faults was applied. Saha, “Fault location in three-terminal line with use
The achieved accuracy is very high, especially for the of limited measurements”, CD Proceedings of
use of the distributed parameter line model. Reliable PowerTech, St. Petersburg, June 2005
selection of the valid results was obtained as well. Im- [8] J. Izykowski, E. Rosolowski, M.M. Saha, M. Fulczyk
plementing this method to the real fault location device, and P. Balcerek, “A fault location method for
also the errors caused by non-ideal transformation of application with current differential relays of three-
instrument transformers have to be accounted for. terminal lines”, IEEE Trans. on Power Delivery,
Vol. 22, No. 4, pp. 2099–2107, October 2007
4 CONLUSIONS [9] J. Izykowski, R. Molag, E. Rosolowski and M.M. Saha,
“Accurate location of faults on power transmission
New fault location algorithm for three-terminal line lines with use of two-end unsynchronized
utilising unsynchronised measurements of three-end measurements” IEEE Trans. on Power Delivery,
currents and voltages has been derived. The algorithm Vol. 21, No. 2, pp. 627–633, April 2006
consists of three subroutines and the procedure for
[10] H.W. Dommel, Electro-Magnetic Transients
selecting the valid result. The distributed parameter line
Program, BPA, Portland, Oregon, 1986.

16th PSCC, Glasgow, Scotland, July 14-18, 2008 Page 7