Professional Documents
Culture Documents
Received 24 June 2005; received in revised form 2 May 2006; accepted 24 May 2006
Available online 24 July 2006
Abstract
The application of the distance protection for distribution systems is increasingly more frequent. However, the distance relaying algorithms
employed are designed for single lines without considering mutual coupling between parallel lines. Mutual coupling affects the accuracy of the
results. In this work, the behavior of the conventional differential equation algorithm (DEA) is analyzed for its accuracy in the estimation of the
fault distance. The response time, the variation of the line parameters, the fault inception angle, the presence of erroneous samples and the mutual
coupling are considered in the study. These results are compared with those obtained using the Fourier algorithm. In both algorithms, the errors are
significant when mutual coupling is present, reaching 12.59% for fault at the end of the line. In this paper, a new distance relaying called DEA with
mutual coupling (DEAMC) has been presented. The new algorithm is an improvement to the DEA algorithm that takes into account the coupling
effect in the analytical equations. The DEAMC response has been compared with the conventional algorithms results. The results show that the
new algorithm improves the accuracy when mutual coupling is present. The DEAMC algorithm reduces the error to 1.93% for fault at the end of
the line.
© 2006 Elsevier B.V. All rights reserved.
0378-7796/$ – see front matter © 2006 Elsevier B.V. All rights reserved.
doi:10.1016/j.epsr.2006.05.011
M. Garcı́a-Gracia et al. / Electric Power Systems Research 77 (2007) 566–573 567
dia dia
u a = R1 · i a + R m · i a + L 1 + Lm + R m · ib
dt dt
dib dic
+ Lm + ·Rm · ic + Lm (5)
dt dt
Fig. 1. Model of transmission line considered in the differential equation algo- Rearranging
rithm.
dia di0
lel lines with mutual coupling. To solve this problem, the DEA ua = R1 · ia + L1 + 3 · Lm + 3 · R m · i0
dt dt
algorithm for parallel lines considering mutual coupling is pre-
dia di0
sented. The behavior of the new algorithm is analyzed based on = [R1 ia + (R0 − R1 )i0 ] + L1 + (L0 − L1 ) (6)
the response time, the fault inception angle, the variation of the dt dt
line parameters, the presence of erroneous samples in the cur-
rent data, and the accuracy in the estimate of the impedance. The where the zero sequence current is given by
DEA with mutual coupling (DEAMC) improves the results when
mutual coupling is present. The method presented in this paper 1
i0 = (ia + ib + ic ) (7)
needs the current on the extreme of the two lines located in the 3
same substation, then communication channel are not needed.
Then, the general equation of the line in matrix form is
2. The differential equation algorithm expressed in function of the three-phase voltage as
⎡ ⎤ ⎤
⎡ ⎡ ⎤ ⎡ ⎤
The DEA algorithm is based on the differential equation ua ia ia i0
⎢ ⎥ ⎢ ⎥ d ⎢ ⎥ ⎢ ⎥
which describes the circuit. It is assumed that it can be rep- ⎣ ub ⎦ = R1 · ⎣ ib ⎦ + L1 · ⎣ ib ⎦ + (R0 − R1 ) · ⎣ i0 ⎦
dt
resented by an impedance of lumped parameters, as is shown in uc ic ic i0
Fig. 1. ⎡ ⎤
The voltage drop across the line impedance of phase-a is i0
d ⎢ ⎥
given by + (L0 − L1 ) · ⎣ i0 ⎦ (8)
dt
ua = Za · ia + Zab · ib + Zac · ic (1) i0
where [U] represents the matrix of the voltage samples, [K] the
matrix of the constant values kr and k1 and [A] represents the
coefficient matrix.
The kr and kl values from (10) have been obtained using
matrices division. For this case, the expression used is:
[U]
[K] = (11)
[A]
where the matrices [A] and [U] are known.
Fig. 2. Power system modeled for the study.
2.2. Double phase-to-ground fault
ua 1 −
1
−
1
uα ⎢ ⎥ 2 ⎢ ⎥
= M · ⎣ ub ⎦ with M = ·⎢ √2 √2 ⎥ The power system has been studied by means of the EMTDC
uβ 3 ⎣ 3 3⎦ simulation [19]. Current and voltage relaying signals has been
uc 0 −
2 2 sampled with the frequency rate f = 10 kHz (200 samples per
(15) cycle).
The three-phase quantities can be condensed into two new 4. Testing and results
orthogonal quantities, which are
uα iα d iα The behavior of the DEA algorithm is analyzed under differ-
= kr · R 1 · + kl · L · (16) ent situations and the results are compared with those obtained
uβ iβ dt iβ
using the Fourier algorithm. In this study, the accuracy of the
M. Garcı́a-Gracia et al. / Electric Power Systems Research 77 (2007) 566–573 569
Table 1 Table 3
Errors in the estimate of the impedance for different fault points Response times and accuracy of the algorithms for different inception angles of
fault
Fault distance DEA algorithm Fourier algorithm
(%) Angle in fault DEA algorithm Fourier algorithm
Error (%) Distance (p.u.) Error (%) Distance (p.u.) (degrees)
Time (ms) Error (%) Time (ms) Error (%)
25 1.95 0.255 2.16 0.255
50 1.95 0.510 1.94 0.505 0◦ 13 1.95 64 1.94
75 1.95 0.765 2.12 0.765 30◦ 10 1.95 64 1.94
100 1.93 1.020 2.02 1.020 60◦ 4.5 1.95 64 1.94
90◦ 4.5 1.95 64 1.94
Table 2
Response time of the DEA and Fourier algorithms time is 76 ms. Therefore, the DEA is faster than the Fourier
Fault distance (%) DEA algorithm time (ms) Fourier algorithm time (ms) algorithm.
25 4.5 76
50 4.5 76
4.3. Influence of the fault inception angle in the response of
75 5.0 76 the algorithms
100 8.0 76
Table 3 shows the errors obtained in the calculation of
the impedance for different fault inception angles. This angle
estimation of the impedance, the influence due to the parame- expresses the state of the voltage wave when the fault happens.
ters variations of the line [20], the fault inception angle and the The fault angles considered are 0◦ , 30◦ , 60◦ and 90◦ of the volt-
mutual coupling are analyzed below. age wave for phase-a. The distance of fault is 50% of the total
length. In this table, it is observed that the accuracy of the algo-
4.1. Accuracy of the algorithms rithms is independent of the fault inception angle. Furthermore,
the response time diminishes for faults occurring in the instant
Table 1 shows the errors obtained in the estimate of the of maximum value of the voltage wave.
impedance by the DEA and Fourier algorithms for different fault
points, with an inception angle of 60◦ . The errors obtained in the 4.4. Influence of parameter variations in the accuracy of
DEA algorithm are less than 1.95% while in the Fourier algo- the algorithms
rithm the maximum error reached is 2.16%. Therefore, from this
point of view, both algorithms are quite similar. The settings of the distance relays are usually adjusted with
parameters calculated during the line design. However, these
4.2. Response times of the algorithms parameters change appreciably when the line is installed. This
is due to conductor height variations and ground resistivity,
In Table 2, the response times of the DEA and Fourier algo- amongst others. Table 4 shows the influence in the accuracy
rithms are shown in function of the fault distance. The response of the algorithms for variations in the ground resistivity from
time is selected when the measured impedance reaches its sta- 50 to 100 m, for a single phase-to-ground fault at 100% of
ble value. The value is considered stable when the maximum the line. Likewise, it is observed that the change in the conduc-
variation is within a range of 0.01 , as shown in Fig. 3. It is tor height does not affect the accuracy of neither the DEA nor
observed that in the worst case the DEA algorithm reaches a the Fourier algorithm in an appreciable way; however, when the
stable value within 8 ms. In the Fourier algorithm the maximum ground resistivity diminishes the error diminishes.
Table 4
Errors due to the parameters variation of lines
Height of the tower (m) DEA algorithm Fourier algorithm
50 m 100 m 50 m 100 m
Error (%) Distance (p.u.) Error (%) Distance (p.u.) Error (%) Distance (p.u.) Error (%) Distance (p.u.)
Table 5 currents of the phase-a, i01 and i02 the zero sequence currents in
Errors in the estimation of the impedance considering the mutual coupling the lines 1 and 2, Z1 , Z2 and Z0m are the positive and negative
between parallel lines
sequence impedances of the line 1 and zero sequence mutual
Fault distance DEA algorithm without Fourier algorithm [17] impedance between lines 1 and 2.
(%) mutual coupling The voltage ua of the phase-a in the point of location of the
Error (%) Distance (p.u.) Error (%) Distance (p.u.) relay is the sum of their positive, negative and zero sequence
25 3.75 0.259 3.57 0.257 values
50 6.18 0.530 6.16 0.530
75 9.08 0.817 9.22 0.817 ua = u1 + u2 + u0 (23)
100 12.59 1.125 12.91 1.300
Substituting (20), (21) and (22) in (23), the voltage ua
becomes
reach is produced. As it is observed in the table, the mutual
coupling affects both DEA and Fourier algorithms significantly. ua = Z1 · i1 + Z2 · i2 + Z0 · i01 + Z0m · i02 (24)
Considering Z1 = Z2
5. DEA with mutual coupling algorithm (DEAMC)
ua = Z1 · (i1 + i2 + i01 ) − Z1 · i01 + Z0 · i01 + Z0m · i02
To correct the previous mentioned effects, an improvement in
the formulation of the DEA algorithm is presented considering (25)
the effects of the magnetic coupling.
In the DEA algorithms shown in Eq. (9), the term (Z0m ·i02 ) is
To obtain the analytical equations of the algorithm, with mag-
not taken into account; however, the consideration of this con-
netic coupling, the circuit of parallel lines shown in Fig. 4 is
cept improves the accuracy when mutual coupling is present.
considered. This shows a power system fed at both ends by their
The DEA considering the mutual coupling is the DEAMC algo-
Thévenin equivalents, represented by Zth1 and Zth2 and their
rithm.
sources. The impedances of both lines are ZL1 and ZL2 , while
Similarly, the current of the phase-a is the sum of their
ZmL indicates the mutual impedance between the lines. The relay
sequence values
is located in bus A of line 1.
The symmetrical component is a usual method for the anal- ia = i1 + i2 + i01 (26)
ysis of unbalanced faults in power systems [21,22]. Using this
method, the sequence networks of the system are shown in Fig. 5, Now, considering (26), Eq. (25) can be expressed in terms of
therefore the electrical parameters of the line, resistance and inductance,
u0 = i01 · Z0 + i02 · Z0m (20)
u2 = i2 · Z2 (21)
u1 = i1 · Z1 (22)
where u0 , u1 and u2 are zero, positive and negative sequence volt-
ages of the phase-a, i1 and i2 the positive and negative sequence
Fig. 4. Power system of parallel lines used in the study. Fig. 5. Sequence networks of the power system of parallel lines.
M. Garcı́a-Gracia et al. / Electric Power Systems Research 77 (2007) 566–573 571
R0m = 1.335 ; L0m = 0.02362 H
uan B11 B12 kr
= · (29) In Eqs. (23), (26)–(29), corresponding to the proposed
uan+1 B21 B22 kl
DEAMC algorithm, the zero sequence of the mutual coupling
where the coefficients are given by impedance appears as an addend. In this section, the influence
of this impedance in the estimation of the fault distance is deter-
B11 = R1 · ian + (R0 − R1 ) · i01n + R0m · i02n mined.
dian di01 di02
B12 = L1 + (L0 − L1 ) + L0m 6.1. Response times of the algorithm
dt dt dt
B21 = R1 · ian+1 + (R0 − R1 ) · i01n+1 + R0m · i02n+1 Table 6 shows the results of the response times of the DEAMC
algorithm. The variation is only significant for faults at the end
dian+1 di01n+1 di02n+1 of the line.
B22 = L1 + (L0 − L1 ) + L0m
dt dt dt
and i02 is the zero sequence current in the line 2, R0m the resistive 6.2. Influence of the fault inception angle in the response of
component of zero sequence mutual impedance between the the algorithm
two lines and L0m is the inductive component of zero sequence
mutual impedance between the two lines. In this section, faults at 50% of the line are applied with
Again the solutions kr and kl are obtained using matrices different inception angles. As it can be observed in Table 7, the
division. variation of the angle does not affect the accuracy of the DEAMC
The developed algorithm implicitly assumes that zero algorithm, in a similar way in the case of the conventional DEA
sequence currents from both lines are available at the relay algorithm.
location. The availability of this current depends on the sys-
tem configuration. Fig. 6 shows the possible configurations. The
DEAMC algorithm ameliorates the accuracy in comparison to Table 6
the DEA algorithm in both ends for the CONFIG 1 and in the Response times of the DEAMC algorithm
Fault distance (%) Time (ms)
25 3.5
50 3.5
75 4.1
100 9.0
Table 7
Response times and accuracy of the algorithm for different inception angles of
fault
Angle in fault (degrees) Time (ms) Error (%)
0◦ 3.5 1.91
30◦ 3.5 1.91
60◦ 3.5 1.91
90◦ 3.5 1.91
Fig. 6. Mutual coupling configurations.
572 M. Garcı́a-Gracia et al. / Electric Power Systems Research 77 (2007) 566–573
Table 8
Errors due to the parameters variation of lines with the DEAMC algorithm
Height of the 50 m 100 m
tower (m)
Error (%) Distance (p.u.) Error (%) Distance (p.u.)
Table 9
Errors for different fault points with the DEAMC algorithm distance, for a system with mutual coupling. The error of the
DEAMC algorithm is independent of the fault distance and it is
Fault distance (%) Error (%) Distance (p.u.)
lower than those obtained from the conventional algorithm.
25 1.93 0.254 It can be observed in Fig. 8 that the problem of under-reach
50 1.91 0.509 associated to the coupling effect of the conventional algorithm is
75 1.89 0.764
100 1.86 1.018
appreciably corrected by the proposed algorithm. As it has been
seen in Section 4.5, for a fault at 100% of the line, the conven-
tional DEA locates the fault outside the protection zone with a
6.3. Influence of the variation of the parameters on the high error (12.59%). With the proposed DEAMC algorithm the
accuracy of the DEAMC algorithm fault distance is calculated with a lower error (1.86%).
Shunt capacitance is not considered in the DEA algorithm. In
The variations of the parameters considered in the conven- short transmission lines its effect is not significant and therefore
tional DEA are used again to evaluate the DEAMC. In Table 8, it its error is neglected. However, in long lines the accuracy of the
can be observed that the algorithm is insensitive to the different distance relaying can be affected. In those cases, to reduce the
tower heights. However, when the ground resistivity diminishes error the methods presented in [23] can be used.
the error diminishes as in the conventional algorithm.
7. Conclusion
6.4. Influence of the mutual coupling on the accuracy of the
algorithm The variations of the conductor height or the fault inception
angle, and the presence of erroneous samples do not significantly
In this study, the power system of Fig. 2 is used considering affect the accuracy of DEA, Fourier and DEAMC algorithms.
the mutual coupling between lines 1 and 2. Table 9 shows the These algorithms are more sensitive to ground resistivity varia-
results of the errors obtained for faults simulated in line 1 at 25, tions.
50, 75 and 100% of the line length. Furthermore, in power systems where the mutual coupling
The results show that the DEAMC algorithm is accurate for is present, the accuracy of the DEA or Fourier algorithms is
the different faults simulated. The error is practically the same negatively affected producing under-reach.
in all the points and it is not higher than 1.93%. Therefore, when The algorithm developed in this work considers the mutual
coupling is considered the error is greatly reduced by using the coupling in distribution system, which improves the accuracy.
DEAMC algorithm. A good capability of the DEAMC algorithm In this algorithm, the zero sequence current of the parallel line is
to compensate the effect of the mutual coupling is observed. required for its correct operation. The error obtained in compar-
Fig. 7 compares the behavior of the errors in the conven- ison with the conventional DEA is reduced from 12.59 to 1.93%
tional and proposed DEAMC algorithm in function of the fault for the example analyzed. The selectivity and the under-reach
problems are minimized, while the response time is not affected.
[K] matrix of the constant values kr and kl [7] Y. Hu, D. Novosel, M.M. Saha, V. Leitloff, An adaptive scheme for
[M] transformation matrix from abc to the α–β components parallel-line distance protection, IEEE Trans. Power Delivery 17 (1) (2002)
105–110.
u0 , u1 , u2 zero, positive and negative sequence voltage of the
[8] M.I. Gilany, Q.P. Malik, G.S. Hope, A laboratory investigation of a digital
phase-a protection technique for parallel transmission lines, IEEE Trans. Power
ua , ub , uc voltage of the a, b and c phases, respectively Delivery 10 (1) (1995) 187–193.
uα , uβ α and β components of the voltage [9] M. Mir, Adaptive vs conventional reach setting of digital distance relays,
[U] matrix of the voltage samples Electric Power Res. 43 (1997) 105–111.
[10] R. Venkatesan, B. Balamurugan, A real-time hardware fault detector using
Z0 , L0 , R0 zero sequence impedance, inductance and resistance
an artificial neural network for distance protection, IEEE Trans. Power
of the line 1 Delivery 16 (1) (2001) 75–82.
Z0m , L0m , R0m zero sequence mutual impedance, inductance [11] M. Akke, J.T. Thorp, Some improvements in the three-phase differential
and resistance between the two lines equation algorithm for fast transmission line protection, IEEE Trans. Power
Z1 , L1 , R1 direct sequence component of impedance, induc- Delivery 13 (1) (1998) 66–72.
[12] E. Rosoxllowskia, J. Iżykowskia, B. Kasztennya, M.M. Sahab, A new
tance and resistance of the line 1
distance relaying algorithm based on complex differential equation for
Z1 , Z2 , Z0m positive and negative sequence impedances of the symmetrical components, Electric Power Systems Res. 40 (3) (1997) 175–
line 1 and zero sequence mutual impedance between 180.
lines 1 and 2 [13] C.E. de Morais Pereira, L.C. Zanetta, Fault location in transmission lines
Za , Zb , Zc impedance of the a, b and c phases, respectively using one-terminal postfault voltage data, IEEE Trans. Power Delivery 19
(4) (2004) 570–575.
Zab , Zac mutual impedance between phases
[14] T. Tagaki, Y. Yamakoshi, M. Yamaura, R. Kondow, T. Matsushima, Devel-
Zm , Lm , Rm mutual impedance, inductance and resistance, opment of a new type fault locator using the one-terminal voltage and
respectively current locator, IEEE Trans. Power Apparatus Syst. 101 (8) (1982) 2892–
Zs , Ls , Rs surge impedance, inductance and resistance of the 2898.
line, respectively [15] A.G. Phadke, J.S. Thorp, Computer Relaying for Power Systems, first ed.,
Research Studies Press Ltd., Exeter, United Kingdom, 1994.
[16] L. Wag, Frequency responses of phasor-based microprocessor relaying
References algorithms, IEEE Trans. Power Delivery 14 (1) (1999) 98–109.
[17] G. Song, J. Sounan, Q. Xu, P. Chen, Y. Ge, Parallel transmission lines
[1] D.L. Waikar, A.C. Liew, S. Elangovan, Design, implementation and per- location algorithm based on differential component net, IEEE Trans. Power
formance evaluation of to new digital distance relaying algorithm, IEEE Delivery 20 (10) (2005) 2396–2406.
Trans. Power Syst. 11 (2) (1996) 445–448. [18] A.T. Johns, S.K. Salman, Digital Protection for Power Systems, Peter Pere-
[2] Y. Liao, S. Elangovan, Improved symmetrical component-based fault dis- grinus Ltd. on behalf of IEE, London, 1995.
tance estimation digital for distance protection, IEE Proc. Gen. Transm. [19] The EMTDC Users Manual, Manitoba HVDC Research Center Inc., Man-
Distribution 145 (11) (1998) 739–745. itoba, 1988.
[3] M.M. Eissa, M. Masoud, To beginner digital distance relaying technique [20] T. Kawady, J. Stenzel, Investigation of practical problems digital for fault
for transmission line protection, IEEE Trans. Power Delivery 16 (7) (2001) location algorithms based on emtps simulation, IEEE Trans. Distribution
380–384. Conf. Exhibit. 1 (10) (2002) 118–123.
[4] H. Lee, Development of an accurate traveling wave fault locator using [21] A.G. Phadke, M. Ibrahim, T. Hlibka, Fundamental basics for distance relay-
global positioning satellites, Spring Meeting of the Canadian Electrical ing with symmetrical components, IEEE Trans. Power Apparatus Syst. 96
Association, Montreal, Quebec, 1993. (3) (1977) 635–643.
[5] D. Chanda, N.K. Kishore, A.K. Sinka, A wavelet multiresolution analysis [22] A.G. Jongepier, L. Van der Sluis, Adaptive distance protection of double-
for location of faults on transmission lines, Int. J. Electrical Power Energy circuit, IEEE Trans. Power Delivery 9 (7) (1994) 1289–1297.
Syst. 25 (1) (2003) 59–69. [23] D. Novosel, D.G. Hart, E. Udren, J. Garitty, Unsynchronized two-terminal
[6] O.A.S. Youssef, To fundamental digital approach to impedance relays, fault location estimation, IEEE Trans. Power Delivery 11 (1) (1996)
IEEE Trans. Power Delivery 7 (10) (1992) 1861–1870. 130–138.