CIGRÉ- 324 2013 CIGRÉ Canada Conference

21, rue d’Artois, F-75008 PARIS The Westin Calgary

http : //www.cigre.org Calgary, Alberta, September 9 - 11, 2013

Accurate Fault Location Estimation in Transmission Lines under High dc-offset

and Sub-harmonic Conditions

U. S. ANNAKKAGE, N. PERERA, V. LIYANAGE

ERLPhase Technologies
CANADA

SUMMARY

Transmission lines are considered as very important parts of the power system which provide a
reliable service from the power plant to the consumers. To maintain an economical power supply,
identifying fault locations is very important [1]. In most of the industrial distance relays, the fault distance
estimation is done using the Takagi algorithm [2]. However this task can be difficult under high dc offset
conditions and high resistive fault situations [3]. In addition, sub harmonic conditions that occur in series
compensated lines have made the estimation of fault location more challenging [4]. In some practical
conditions, faults are cleared within two cycles and the fault distance estimation with such a short data
window is difficult [5]. In literature, several attempts have been made to overcome the aforementioned
problems [6-9]. These methods involved the use of different filtering approaches. Although such filtering
methods are being used in most of the distance relays, their behavior is not satisfactory under all practical
conditions described above [10].

This paper investigates the development of an enhanced single-ended impedance based fault location
algorithm for numerical distance relays, suitable under high dc off-set and sub-harmonic conditions. The
proposed solution includes the use of filters to eliminate the effects of the dc off-set and sub-harmonics.
The modified Takagi fault location algorithm was used as the basis. Performance of the proposed
algorithm was evaluated using faulty waveforms generated from an electromagnetic transient (EMT) type
simulation program. In this study, two transmission systems were considered (i.e. a mutually coupled
double circuit transmission line and a series compensated single circuit transmission line). Different types
of faults (phase-to-ground, phase-to-phase, three-phase, etc.) simulated with different fault inception
angles were considered.

Results obtained from this study show that the algorithm is capable of estimating the fault location in
uncompensated double circuit transmission line with a low maximum percent error of 1.62%. The
maximum error observed in the series compensated line under high sub-harmonic conditions was 2.1%.
In addition, performance of the proposed algorithm was also evaluated for short duration faults. For this
study, faults generated from the uncompensated double circuit transmission line were considered. The
results showed errors less than 2% and 3% for the 2 cycle and 1.5 cycle duration faults respectively.
A laboratory prototype is being implemented and more results will be presented in the full paper.

KEYWORDS

fault location, distance relay, dc off-set, sub-harmonics

1 INTRODUCTION

Transmission lines are considered as very important parts of the power system which provide a reliable
service from the power plant to the consumers. To maintain an economical power supply, identifying
fault locations is very important [1]. In most of the industrial distance relays, the fault distance estimation
is done using the Takagi algorithm [2]. However this task can be difficult under high dc offset conditions
and high resistive fault situations [3]. In addition, sub harmonic conditions that occur in series
compensated lines have made the estimation of fault location more challenging [12]. In some practical
conditions, faults are cleared within two cycles and the fault distance estimation with such a short data
window is difficult [5]. In literature, several attempts have been made to overcome the aforementioned
problems [6-9]. These methods involved the use of different filtering approaches.
Although such filtering methods are being used in most of the distance relays, their behavior is not
satisfactory under all practical conditions described above [10]. This research investigates the
development of an accurate and efficient algorithm that is capable of identifying the fault distance using
two cycle information under high dc off-set and sub-harmonic conditions.
The rest of the paper is organized as following. Section II gives a brief description on the enhanced
fault location method proposed in this research. Results obtained
using detailed simulation studies are presented in Section III. Va Vb Vc Ia Ib Ic
Finally, the conclusions are given in Section IV.
High- Pass Filter
a) dc offset removal
2 PROPOSED METHOD b) sub-harmonic removal
2.1 Algorithm
Fig-1 shows the calculation steps involved in the proposed Phasor Estimation
algorithm. First, voltage and current signals are passed through a
high-pass filter to eliminate the dc offset and sub-harmonics.
Next, phasor quantities are estimated using the recursive discrete Phase Selector
furrier transform. The phase selector calculations are performed
subsequently to identify the faulted phase. Finally, the modified Fault Location Estimation
Takagi fault location algorithm [2] is applied to estimate the fault (Modified Takagi method)
distance. More details on each of these computational elements
are given below.
Fault location
2.2 DC Offset Removal Filter
Fig. 1. Steps involved in fault
In order to remove dc offsets in the measurement waveforms
location estimation
with less computations and minimal delay time, an Infinite
Impulse Response (IIR) filter was selected. The IIR filter used in
this paper has two forward sections and one feedback section.
Therefore, the dc component can be eliminated by finding the
difference in the two forward sections while keeping these
coefficients equal in magnitude. The digital implementation of
the proposed IIR filter used for dc offset removal is given in (1).

I Filt  0.8  I (i )  0.8 * I (i  1)0.6 *I Filt (i  1) 1

The output response of the dc removal filter is shown in Fig.2.

2.3 Sub-Harmonic Removal Filter

Fig. 2. DC offset removal from a

current waveform
 Sub-Harmonics make fault location estimation challenging especially for short duration faults. A
high pass filter to remove these sub-Harmonics is designed so that the pass band has a constant gain with
th
a sharp edge yet the time delay is not too large. To accomplish this, a 5 order Butterworth high pass
filter with a cutoff of frequency of 45Hz was
used. The output response of selected filter is Magnitude (dB) and Phase Responses
shown in Fig.3.
0 7.1735
2.4 Phasor Estimation
-10 6.0861
The recursive discrete furrier transform
(RDFT) was used to estimate the voltage and

Phase (radians)
Magnitude (dB)
-20 4.9986
current phasors.
-30 3.9112

-40 2.8238
2.5 Phase Selection
-50 1.7363
Phase selector determines the type of the fault
(AG, BG, CG, AB, BC, etc.) which is required -60 0.6489
for accurate fault location estimation.
0 20 40 60 80 100 120 140 160 180
Frequency (Hz)
2.6 Fault Location Calculation
Fig. 3. Sub-harmonic removal filter response
The single-ended modified Takagi algorithm
for fault location was used. This method does not TABLE I
CONDITIONS TESTED
use the source impedance which is practical since the
CONDITION STATE OF LINE 2
source impedance is not always available. Pre-fault 1 ACTIVE PARALLEL LINE
current is required along with line impedances and 2 OPEN BUT NOT GROUNDED
system voltage and current measurements to accurately 3 OPEN AND GROUNDED
estimate the fault location.
TABLE II
CONDITION 1 FAULT DETECTION RESULTS
2.7 Adjustments for Mutual Compensation FAULT ACTUAL CALCULATED % ERROR
TYPE LOCATION LOCATION
The use of parallel lines on transmission towers is (KM) (KM)
increasingly more common as the electricity AG 12.52 0.21
consumption and demands increase. The introduction of BG 12.53 0.26
parallel lines can cause mutual coupling effects. These CG 12.53 0.26
12.5
mutual coupling effects provide inaccurate impedance ABG 12.52 0.13
BCG 12.52 0.12
measurements and fault distance estimations in distance CAG 12.52 0.12
relays. Therefore, in this algorithm mutual line AG 50.03 0.06
compensation is taken into consideration by making BG 50.03 0.05
necessary adjustments to the original equation [1]-[2]. CG
50
50.03 0.06
ABG 50.15 0.29
3 SIMULATION BCG 50.03 0.06
CAG 50.15 0.30
AG 125.06 0.05
BG 125.04 0.03
Network-1 Line-1 Network-2 CG
125
125.06 0.05
ABG 125.02 0.02
BCG 125.02 0.02
Line-2 CAG 125.03 0.02
AG 200.36 0.18
BG 200.34 0.17
CG 200.35 0.17
200
ABG 200.23 0.11
Fig. 4. Single-line diagram of the double circuit BCG 200.23 0.11
transmission system under analysis CAG 200.25 0.12
 TABLE IV
The cases that were analyzed are practical scenarios that CONDITION 3 FAULT DETECTION RESULTS
were simulated on a Real Time Digital Simulator (RTDS) FAULT ACTUAL CALCULATED % ERROR
to allow for a real-time generation. Fig. 4 shows the TYPE LOCATION LOCATION
(KM) (KM)
400kV, 250 km double circuit transmission system under
AG 12.56 0.45
analysis. In this simulation, the transmission line is BG 12.55 0.42
simulated with mutual effects. The system parameters are CG 12.56 0.45
12.5
given in the Appendix-A. ABG 12.52 0.19
BCG 12.52 0.19
3.1 Fault Location Results CAG 12.53 0.21
In this study, for active line-1 three different conditions AG 50.02 0.05
were considered for line-2, as given in Table 1. For all the BG 50.03 0.06
conditions, the power flow is from Bus A to Bus B CG 50.02 0.05
50
sending 100MVAr and 500MW across the line. ABG 50.10 0.20
BCG 50.06 0.11
CAG 50.05 0.10
TABLE III
AG 124.38 0.50
CONDITION 2 FAULT DETECTION RESULTS
BG 124.38 0.49
FAULT ACTUAL CALCULATED % ERROR
CG 124.39 0.49
TYPE LOCATION LOCATION 125
ABG 125.14 0.11
(KM) (KM)
BCG 125.11 0.09
AG 12.54 0.29 CAG 125.14 0.11
BG 12.53 0.27 AG 197.89 1.05
CG 12.53 0.26 BG 197.93 1.04
12.5
ABG 12.52 0.13 CG 197.92 1.04
BCG 12.52 0.15 200
ABG 200.63 0.32
CAG 12.53 0.25 BCG 200.66 0.33
AG 50.06 0.12 CAG 200.67 0.33
BG 50.06 0.13
CG 50.06 0.11 TABLE V
50
ABG 50.06 0.12 EFFECT OF FAULT DURATION
BCG 50.06 0.13
CAG 50.06 0.12 2 CYCLE 1.5 CYCLE 1 CYCLE
LOCATION (KM)

AG 125.44 0.35 CALCULATION CALCULATIO CALCULATION

BG 125.44 0.35
ACTUAL

S NS S
CG 125.45 0.36
FAULT TYPE

125
LOCATION

LOCATION

LOCATION
% ERROR

% ERROR

% ERROR
ABG 125.90 0.72
BCG 125.57 0.45
CAG 125.57 0.46
AG 202.01 1.00
BG 202.02 1.01 AG 50.43 0.8 49.52 0.9 61.33 22.6
CG 202.02 1.01 BG 50.07 0.1 49.69 0.6 73.14 46.2
200
ABG 202.44 1.22 CG 50.32 0.6 49.48 1.0 54.41 8.8
BCG 203.23 1.62 50
ABG 49.40 1.1 50.58 1.1 56.75 13.5
CAG 202.42 1.21 BCG 50.25 0.5 50.85 1.7 62.79 25.5
CAG 50.40 0.8 48.29 3.4 47.22 5.5
For each condition, six different fault types were AG 126.44 1.1 123.2 1.4 133.8 7.0
evaluated which include the faults occurring at 5%, BG 125.40 0.3 125.9 0.7 160.7 28.5
20%, 50% CG 125.24 0.1 123.5 1.2 119.7 4.2
125
ABG 126.25 1.0 124.3 0.5 138.6 10.8
and 80% of the line. Tables II, III and IV show BCG 126.68 1.3 125.5 0.4 141.5 13.2
the fault location results for the various line CAG 124.44 0.4 121.3 2.9 143.3 14.6
locations corresponding to each condition. AG 202.59 1.2 196.8 1.6 217.4 8.7
BG 200.70 0.3 198.1 0.9 244.5 22.2
CG 200.89 0.4 197.8 1.1 194.6 2.7
200
3.2 Effect of Fault Duration ABG 200.00 0.0 202.2 1.1 208.7 4.3
BCG 201.98 0.9 196.5 1.7 228.6 14.3
CAG 200.6 0.3 200.1 0.1 218.8 9.4
 Determining fault location can be difficult if the fault is recognized and cleared within a short duration
of time such as within two cycles. A good fault location method should be able to give accurate results for
faults with short durations. In order to evaluate the performance of the proposed method for short duration
faults, two cycle, one and a half cycle, and one cycle faults were analyzed. These waveforms were
obtained by using the same set of waveforms as in Section III-A, but disregarding values after the desired
number of cycles. In this study, the Condition-1(i.e. active parallel line) faults were considered only. The
fault location results are summarized in Table V. As it can be seen from these results, the proposed
method is capable of estimating the fault location within 2% limit based on two cycle data.
3.3 Effect of Sub-harmonics
In order to evaluate the performance of the proposed fault location algorithm under sub-harmonic
conditions, the single circuit, series compensated transmission line shown in Fig.5 was simulated in
RTDS. As shown in Fig.5, the capacitor bank was simulated with Metal Oxide Varistors (MOVs) capable
of by-passing the capacitors under high fault current conditions. Different types of faults (AG, BC, BCG,
etc.) were simulated at different distances of the transmission line with different fault inception angles.
Fig. 6 shows the current and voltage signals observed with sub harmonics during a BC fault at 75% of the
line. The fault estimation results are summarized in Table-VI. As seen from the results, proposed
algorithm was able to determine the fault location with a maximum error of 2.1%.
MOV
Network-1 % distance
Network-2

Line-1

Relay F
Line-2
Line-3
Bus-1 Bus-2
Fig. 5. Single-line diagram of the single circuit series-compensated transmission system under analysis

TABLE VI
EFFECT OF SUB-HARMONIC CONDITIONS
FAULT TYPE ACTUAL CALCULATED %
LOCATION(KM) ERROR
Current (A,sec.)

DETECTED LOCATION
(KM)
AG 4.7 0.3
BC 4.71 4.7 0.3
BCG 4.7 0.3
ABC 4.8 1.8
Voltage (V,sec.)

AG 58.7 0.4
BC 58.92 59.1 0.3
BCG 59.5 0.9
ABC 59 0.1
AG 118.2 0.3
BC 117.85 116.4 1.2
Fig. 6. Currents and voltages observed during the fault
BCG 116.4 1.2
ABC 116.3 1.3
4 PROTOTYPE IMPLEMENTATION AND TESTING AG 178.6 1.0
BC 176.77 177.9 0.6
BCG 177.2 0.2
4.1 Prototype Implementation ABC 180.5 2.1
In order to further evaluate the performance of the proposed
fault location method under real-time operating conditions, a
laboratory prototype was implemented on a numerical
distance relay environment [10].
4.2 Test Setup
Fig. 7. Test System
Fig.7 shows the test setup used for this evaluation. The
recorded waveforms generated using the RTDS were played back using a Doble transient playback unit
[12]. User interfacing for the relay and the amplifier was provided using personal computers (PCs).
4.3 Test System
Fig.8 shows the single line diagram of 400kV, 250 km transmission line model used to evaluate the
performance of the relay. The transmission line used here is a double circuit, series compensated line.
System parameters of the transmission line are given in the Appendix. Transmission lines were modeled
using frequency dependent line characteristics with mutual effect. The series capacitor bank was modeled
with a parallel MOV protection arrangement. Current transformers (CTs) and potential transformers (PTs)
were also included in the simulation. The CT and PT ratios used are 1000:1 and 40000:110 respectively.
A power flow of 750 MW and 250 MVar, flowing form Station-A to Station-B was assumed. As the
system used here is a double circuit transmission line, current signals from the mutual line were used to
compensate the mutual effects during ground faults. Quadrilateral impedance characteristics were
assumed for both ground element and phase element settings of the distance relay.
4.4 Results
In this study, different types of
faults (AG, BC, BCG and ABC) are
simulated on Line-1 at different
distances form Relay-1 as shown in
Fig.8. Recorded waveforms were
played back to the relay through the Fig. 8. Single-line diagram of the double circuit
amplifier. Fig.9 shows the variation of series-compensated transmission system under analysis
voltage and current signals observed
by the relay during an ABC fault simulated at 62.5 km.
Fig.9 clearly shows the presence of sub harmonics in current waveforms generated due to the effect of
the series capacitor. The variations of phase impedances (Zab, Zbc and Zca) calculated by the relay
during the fault are shown in Fig.10. As it can be seen from Fig.9 and Fig.10, the relay shows a correct
operation during this fault. The fault location estimated by the relay was 62.4 km.

Fig. 9. Observations during ABC fault at 62.5 km

 TABLE VII
RESULTS FROM PROTOTYPE
FAULT TYPE ACTUAL CALCULATED %
DETECTED LOCATION(KM) LOCATION ERROR
(KM)
AG 24.7 1.2
BC 25 25.2 0.8
BCG 25 0.0
ABC 24.8 0.8
AG 61.7 1.3
BC 62.5 63.6 1.8
BCG 62.9 0.6
ABC 62.4 0.2
AG 126.6 1.3
BC 125 127.0 1.6
BCG 126.2 1. 0
ABC 125.9 0.7
AG 191 1.9
Fig. 10. Variation of phase impedances during the ABC BC 187.5 190.7 1.7
fault BCG 190.7 1.7
ABC 190.7 1.7

Testing was repeated for several fault cases and a selected set of results are summarized in Table-VII.
As it can be seen from the results, the maximum fault location error observed from prototype testing was
1.9%.

5 CONCLUSION
In this paper, applicability of different filtering methods to eliminate the effects of the sub-harmonic
conditions and the dc-offset conditions was investigated for accurate and fast fault distance estimation in
transmission lines. An offline program of the proposed fault location algorithm was developed in
MATLAB. Performance of the off-line program was evaluated using recorded waveforms obtained from
an electromagnetic transient (EMT) type simulation program. Results obtained from the off-line analysis
showed a maximum fault location estimation error of 2.1% based on the two-cycle fault information
under high dc-offset conditions and sub harmonic conditions. Finally, a laboratory prototype was
implemented and tested using the simulated waveforms played back via a real-time playback system. The
maximum fault location error observed from the prototype was 1.9%.
Results presented in this paper show the potential for developing an accurate fault location method for
distance relays using the proposed filters.
APPENDICES
A. Parameters of double circuit transmission line
Table-VIII shows the transmission line parameters for the double circuit transmission line. In this
simulation, a source to line impedance ratio (SIR) of 1.0 was assumed.
TABLE VIII
LINE PARAMETERS: DOUBLE CIRCUIT LINE
Line
Positive 77.2<84.60
ohms
Zero 263.7<75.70
ohms
Mutual 165.6<69.80
ohms
Length 250 km
B. Parameters of single circuit series compensated transmission line
Parameters of the series compensated line are shown in Table IX. In this simulation modal, a SIR of 0.2
was assumed. The amount of the compensation used was 40%.
TABLE IX
LINE PARAMETERS: SERIES COMPENSATED LINE
Line-1 Line-2 Line-3
0
Positive 72.7<84.6 30.9<84.6 30.9<84.60
ohms ohms ohms
Zero 248.7 <75.7 105.8<75.70 105.8 <75.70
ohms ohms ohms
Length 235.7 km 100 km 100 km
C. Parameters of double circuit series compensated transmission line
TABLE X
LINE PARAMETERS: DOUBLE CIRCUIT LINE
Line 1&2 Line 3 Line 4
Positive 77.2<84.60 ohms 30.9<84.60 ohms 46.4<84.60
ohms
Zero 263.7<75.70 105.8<75.70 158.7<75.70
ohms ohms ohms
Mutual 165.6<69.80 N/A N/A
ohms
Length 250 km 100 km 150 km
Table-X shows the transmission line parameters for the double circuit series compensated transmission
line. In this simulation, a source to line impedance ratio (SIR) of 1.0 was assumed. The amount of the
compensation used was 40%.

REFERENCES
[1] IEEE Standard PC37.114, “Draft Guide For Determining Fault Location on AC Transmission and Distribution
Lines,” 2004.
[2] T. Takagi, Y. Yamakoshi, M. Yamaura, R. Kondou, and T. Matsushima, “Development of a New Type Fault
Locator Using the One-Terminal Voltage and Current Data,” IEEE Transactions on Power Apparatus and
Systems, vol. PAS-101, no. 8, August 1982, pp.2892-2898.
[3] E. O. Schweitzer, III, “A Review of Impedance-Based Fault Locating Experience, Proceedings of the 15th
Annual Western Protective Relay Conference, Spokane, WA, October 24-27, 1988.
[4] Peterson, M., Midence, P. E. R., Perez, P. E. J., Mulawarman, P. E. A.; “Application of a subharmonic
protection relay,” 11th International Conference of Developments in Power Systems Protection, 2012.
[5] J.V. Calvano, V.C. Alves, M. Lubaszewski, “Fault detection methodology and BIST method for 2nd order
Butterworth, Chebyshev and Bessel filter approximations,” 18 th IEEE in VLSI Test Symposium, 2000.
[6] I. Sakagami, H. Yanna, M. Mohemaiti, A. Tokunou, “On a transmission-line Butterworth lowpass filter using
radial stubs”, IEEE International Symposium on Circuits and Systems, 2002.
[7] J.C. Gu, S.L. Yu, “Removal of DC offset in current and voltage signals using a novel Fourier filter algorithm,”
IEEE Transactions on Power Delivery, vol. 15, no. 01, pp. 73 – 79, Jan 2000.
[8] D.A. Tziouvaras, J.B. Roberts, G. Benmouyal, “New Multi-Ended Fault Location Design for Two or Three-
Terminal Lines,” CIGRE Conference, 1999.
[9] A.A. Gigris, D.G. Hart, and W.L. Peterson, “A New Fault Location Technique for Two- and Three-Terminal
Lines,” IEEE Transactions on Power Delivery, vol. 7, no. 1, January 1992, pp. 98-107.
[10] Transmission Line Protection Relay L-PRO 4000, ERLPhase Power Technologies, Winnipeg, MB, Canada,
2012.
[11] N. Perera, K. Narendra, D. Fedirchuk, R. Midence, V.K. Sood, “Performance Evaluation of a Sub-Harmonic
Protection Relay Using Practical Waveforms,” IEEE EPEC 2012, London, Ontario, Oct. 10-12, 2012.
[12] Real Time Power System Simulation using F6000 Instruments, Application Note, Doble Eng