IEEE TRANSACTIONS ON POWER DELIVERY, VOL. 19, NO.

1, JANUARY 2004 35

A New Fault Location Algorithm Using Direct Circuit

Analysis for Distribution Systems
Myeon-Song Choi, Member, IEEE, Seung-Jae Lee, Member, IEEE, Duck-Su Lee, and Bo-Gun Jin

Abstract—The unbalanced nature of distribution systems due to Most of the research mentioned above focuses on the fault lo-
single-phase laterals and loads gives difficulty in the fault location. cation of the transmission lines. It is because transmission net-
This paper proposes a new fault location algorithm developed by works are generally under the balanced operation that enables
the direct three-phase circuit analysis for unbalanced distribution
systems, which has not been investigated due to high complexity.
the circuit analysis based on the symmetrical component. Note
The proposed algorithm overcomes the limit of the conventional al- that the sequence component method generates three indepen-
gorithm, which requires the balanced system. It is applicable to any dent sequence networks only in case of the balanced systems
power system, but especially useful for the unbalanced distribution and becomes a powerful tool in the circuit analysis. However, it
systems. Its effectiveness has been proved through many EMTP cannot be applied to the distribution system case, since most
simulations. distribution systems are unbalanced due to the mixed use of
Index Terms—Distribution system operation, fault location, single-phase and three-phase laterals and loads, etc.
power system protection, protective relaying. Although a direct circuit analysis of three-phase network is
another alternative but it has not been considered due to its high
I. INTRODUCTION complexity and difficulty during the course of analysis. This is
why not much research on the fault location problem has been

T HE service continuity is one of the most important con-

cerns to the utility company, but a fault is inevitable often
resulting in power interruption. This is more prominent in the
reported for the distribution system.
Fault location methods for the unbalanced distribution systems
utilize harmonics [5], fundamental component of voltage and cur-
distribution systems, because most of lines are exposed to air rent [6]. Additional calculation burden like recalculation of the
and spread over a wide area. A fast fault location helps re- voltage and current at each node [7], [8] is needed for the com-
ducing the interruption time, but conventional visual inspection pensation of the characteristics unique to the distribution system.
not only takes a long time but also requires high manpower. In In this paper, a fault location algorithm based on the direct
recent years, the digital protective devices are replacing the old circuit analysis is suggested. The fault location equation has
type devices like the electromechanical or static devices. Those been derived by applying matrix inverse lemma and is relatively
digital devices are generally equipped with fault recording func- simple and easy to be applied for any system regardless of a
tion, which enables an automatic fault location. phase balance condition. Therefore, it can handle not only the
During the last decade, there has been much research on the transmission systems but also the distribution systems.
fault location problem that is to calculate the distance from a Section II reviews the symmetrical components and its useful-
protective device to the fault. Most of them deal with trans- ness in fault location problems for the balanced systems. Then de-
mission networks, which are, in general, operated in a balanced tails of the proposed algorithm follow. Test results are also given
manner. These methods can be classified into three groups: the in order to show the effectiveness of the proposed method.
first one is to use traveling waves [1], the second one is to use
harmonics [2], and the last one is to use apparent impedance II. FAULT LOCATION OF BALANCED SYSTEM
based on fundamental components of voltage and current [3].
The last method can be further divided into two subgroups: to A. Sequence Circuit of Balanced Three-Phase System
use one-end information or to use information from both ends Fig. 1 shows a simple balanced three-phase system that has
of the faulted line. Among them, the latter method gives higher two line sections and two three-phase laterals and loads.
accuracy; however, it needs additional devices for communica- Representing the equivalent impedance at bus as ,
tion and data transmission. Therefore, the former one is more which is the combination of two loads and one line, the phase
widely used with some accuracy improving technique adopted voltage at A is given as follows:
[4].
(1)
Manuscript received August 19, 2002. This work was supported by the Min- Applying the sequence transformation of (2) into (1)
istry of Science and Technology of Korea and the Korea Science and Engi-
neering Foundation.
M.-S. Choi, S.-J. Lee, and B.-G. Jin are with the Next-Generation Power (2)
Technology Center and with the Department of Electrical Engineering,
Myongji University, Yongin 449-728, Korea (e-mail: [email protected];
[email protected]; [email protected]).
D. S. Lee is with P&C Tech., Anyang, Korea (e-mail: [email protected]). where
Digital Object Identifier 10.1109/TPWRD.2003.820433

0885-8977/04$20.00 © 2004 IEEE

36 IEEE TRANSACTIONS ON POWER DELIVERY, VOL. 19, NO. 1, JANUARY 2004

Fig. 2. Single line-to-ground fault on balanced system.

Fig. 1. Balanced distribution system.

and the symmetrical component impedance can be

obtained as

(3.1)
(3.2)

It is well known that the sequence impedance matrices, (3)

have only diagonal elements in the balanced system. Therefore,
each sequence voltage and current can be expressed separately Fig. 3. Negative sequence circuit.
using only the corresponding diagonal elements as follows:

(4.1) The negative sequence circuit for the balanced system in

Fig. 2 is shown in Fig. 3. Applying KVL to the route A, the
(4.2) following voltage equation is obtained:
(4.3)
(7)
Since there is no coupling among equations, each sequence
circuit represented by (4) can be handled independently, which where
provides a big convenience in development of the fault location equivalent negative sequence impedances for the
algorithm as will be seen in Section II-B. source and load;
negative sequence currents flowing into the fault
B. Fault Location Based on Symmetrical Component from the source and load side;
Fig. 2 shows a single line-to-ground fault in a balanced three- negative sequence line impedance.
phase system. From (7), distribution factor can be obtained
A-phase voltage at the measuring point is given as
(8)
(5)

where where , , .
a-phase voltage; Then a negative sequence current distribution factor is
a-phase current; obtained using (8)
zero sequence current;
zero and positive sequence impedance of line; (9)
;
fault current; Equation (9) provides a way to obtain negative sequence com-
fault resistance; ponent of the fault current from the source current.
fault distance from the other end.
In (5), all of the impedances except fault resistance are known C. Fault Location Equation
values, and phase voltage and current are also available from
Substituting in (5) by (6) and (9), we obtain
measurement. One of two unknown variables, a fault current can
be obtained using the current distribution factor and another one,
a fault resistance can be eliminated during derivation of fault
location equation, whose details are described in the following.
For a single line-to-ground fault, the following relation holds: (10)
(6)
Rearrangement of (10) yields
and the negative sequence component of the fault current
can be obtained from the distribution factor and
the negative sequence current at the measuring point. (11)
CHOI et al.: A NEW FAULT LOCATION ALGORITHM USING DIRECT CIRCUIT ANALYSIS 37

Equation (11) is rearranged again to form a second order

polynomial with respect to fault distance variable as follows:

(12)

where

Fig. 4. Unbalanced distribution system.

From the imaginary part of (12), the fault resistance is

obtained

(13)

Fig. 5. Single line-to-ground fault on general three-phase system.

Replacing in the real part of (12) by (13)
sequence), are no longer independent of one another. In general,
(14) symmetrical component transformation never provides a con-
venient way of analysis in unbalanced networks. Therefore, the
is obtained. Then the fault distance can be obtained by sequence circuit and distribution factor cannot be obtained and
solving (14). previous fault location equation is not valid for the unbalanced
systems case. This is another factor that makes the fault analysis
D. Unbalanced Three-Phase System in unbalanced distribution network difficult.
Note that above-mentioned method can only be applied to a
balanced system because, not to an unbalanced system. Con- III. DIRECT CIRCUIT ANALYSIS-BASED FAULT
sider the distribution system in Fig. 4 that shows a simple un- LOCATION ALGORITHM
balanced three-phase system due to no line transposition and A. Basic Concept
single-phase lateral and load. Suppose the line impedance and
Fig. 5 shows a single line-to-ground fault in a general three-
the equivalent impedance matrix at bus B be given as shown in
phase system.
the first equation at the bottom of the page.
Three-phase voltage at is given by (15)
If the above phase-impedance matrices are transformed to se-
quence impedance, we obtain the second equation at the bottom (15)
of the page.
Since the sequence impedance matrix has nonzero terms in where
off-diagonal part, the three sequences in symmetrical compo- : phase voltage vector;
nent system (i.e., zero sequence, positive sequence, and negative : phase current vector;
38 IEEE TRANSACTIONS ON POWER DELIVERY, VOL. 19, NO. 1, JANUARY 2004

Now, let us define , , , as follows:

Fig. 6. Two parallel paths at the fault. (20)

: phase voltage vector at fault (21)

point;

Then, applying the lemma, we obtain

: line impedance matrix;

fault current.

Since a-phase voltage at the fault is the same as , a-phase (22)

voltage at A can be expressed as follows:

(16) Note that the first row of (22) is given as

Note that there are two parallel circuits at the fault position as
shown in Fig. 6. One is associated with the fault and the other
with the load. (23)
Note that the fault admittance matrix is given as
Then, the fault current equation (18) can be rewritten as
(17)

and the load admittance matrix can be obtained by taking an (24)

inverse of the impedance of series connection of line and load where represents an element that we are not concerned about.
as follows: The final expression for the fault current becomes

where is the load impedance

matrix. (25)
Then, the fault current can be obtained from the using
the current distribution law of a parallel circuit
Substitution of (25) into the fault current equation of (16)
yields
(18)

where

(26)
The inverse matrix in (18) can
be simplified by the matrix inverse lemma [9] or

(19) (27)
CHOI et al.: A NEW FAULT LOCATION ALGORITHM USING DIRECT CIRCUIT ANALYSIS 39

where have been considered. EMTP simulation has been performed,

in which 1920-Hz sampling frequency is used. Voltage and
current phasors are obtained by the discrete Fourier transform
(DFT) having one cycle data window.
Equation (27) can be rearranged as a second order polynomial The error of the fault location is calculated by the following
with respect to the distance variable equation:

(32)
(28)
Expressing the coefficients of (28) by the complex number where
represented in the rectangular form, that is estimated distance;
real distance;
whole line length (p.u.).

A. Three-Phase Balanced System

The balanced system given in Fig. 1 is used as the test system
Equation (28) becomes and its line impedance data used in simulation are given as
shown in the top equation at the bottom of the page.
(29)
The faults are assumed to have occurred between point and
and its imaginary part gives . A distribution factor-based method (called “DF algorithm” af-
terwards) introduced in Section II gives very satisfactory results
(30) with maximum estimation error of 0.04% when a fault resistance
From its real part, a quadratic polynomial with respect to is is 50 . In the proposed direct circuit analysis based-algorithm
derived by use of (30) as follows: case (called “DCA algorithm” afterwards), the accuracy is of the
samelevelas DFalgorithm,withthemaximumerrorbeing0.05%.
(31) In both cases, the estimation errors are caused by calculation error.
The estimation errors are illustrated in Fig. 7 for DF algorithm and
Finally, the fault distance can be obtained by solving in Fig. 8 for DCA algorithm, respectively.
(31). Note that this fault location equation based on the direct From these results, it can be seen that both algorithms are
circuit analysis can be applied to any system-balanced or unbal- very effective for a line-to-ground fault case in a balanced three-
anced, three-phase, or three/single-phase systems. phase system.

IV. SIMULATION RESULTS B. Three-Phase Unbalanced System

The suggested direct circuit analysis based-algorithm has An unbalanced system case is studied in the same manner as
been applied to a balanced three-phase system and an unbal- in Section IV using the system in Fig. 4. The impedance data
anced system for verification. The results are compared with are assumed to be given as shown in the lower equation at the
those of the conventional method using distribution factor of the bottom of the page
negative sequence current described in the paper. In each case, The single line-to-ground fault is assumed to have occurred
nine different fault distances varying from 0.1 to 0.9 p.u. by between and and the simulation results are illustrated in
0.1-p.u. step and three fault resistances of 0, 10, 30, and 50 Figs. 9 and 10.
40 IEEE TRANSACTIONS ON POWER DELIVERY, VOL. 19, NO. 1, JANUARY 2004

Fig. 7. DF algorithm error in a balanced system. Fig. 10. DCA algorithm error in an unbalanced system.

V. CONCLUSION
A new fault location algorithm based on the direct circuit
analysis is suggested in this paper. Application of the matrix
inverse lemma has greatly simplified the derivation of the fault
location equation that, otherwise, is too complicated to be de-
rived.
The proposed algorithm overcomes the limit of the conven-
tional fault location algorithm based on the sequence circuit
analysis, which is the balanced system requirement. The pro-
posed algorithm is applicable to any power system, but espe-
cially useful for the unbalanced distribution systems. Its effec-
tiveness has been proved through many EMTP simulations.
The fault location algorithm for the phase-to-phase fault in an
unbalanced system needs to be developed and the effect of load
Fig. 8. DCA algorithm error in a balanced system. variation or uncertainty need to be investigated to enhance the
practical power and these works are in progress by the authors.

ACKNOWLEDGMENT
The authors would like to thank the Ministry of Science and
Technology of Korea and Korea Science and Engineering Foun-
dation for their support through the ERC program.

REFERENCES
[1] G. B. Ancell and N. C. Pahalawaththa, “Maximum likelihood estimation
of fault location on transmission lines using travelling waves,” IEEE
Trans. Power Delivery, vol. 9, pp. 680–689, Apr. 1994.
[2] T. Takagi, Y. Yamakoshi, J. Baba, K. Uemura, and T. Sakaguchi, “A new
algorithm of an accurate fault location for EHV/UHV transmission lines
: Part—Fourier transform method,” IEEE Trans. Power App. Syst., vol.
PAS-100, pp. 1316–1323, Mar. 1981.
[3] M. S. Sachdev and R. Agarwal, “A technique for estimating transmission
Fig. 9. DF algorithm error in an unbalanced system. line fault locations form digital impedance relay measurements,” IEEE
Trans. Power Delivery, vol. 3, pp. 121–129, Jan. 1988.
[4] Y.-J. Ahn, M.-S. Choi, S.-H. Kang, and S.-J. Lee, “An accurate fault lo-
A significant accuracy difference can be observed between cation algorithm for double-circuit transmission systems,” in Proc. IEEE
Power Eng. Soc. Summer Meeting, vol. 3, 2000, pp. 1344–1349.
two algorithms. The maximum estimation error is 25% in case [5] M. E. Hami, L. L. Lai, D. J. Daruvala, and A. T. Johns, “A new travel-
of DF algorithm while 0.08% in case of DCA algorithm. Such an ling-wave based scheme for fault detection on overhead power distribu-
unacceptably high error observed in DF algorithm is due to the tion feeders,” IEEE Trans. Power Delivery, vol. 7, pp. 1825–1833, Oct.
1992.
system unbalance. The error in DCA algorithm is contributed [6] A. A. Girgis, C. M. Fallon, and D. L. Lubkeman, “A fault location tech-
by the calculation error, which is very small, showing its effec- nique for rural distribution feeders,” IEEE Trans. Ind. Applicat., vol. 29,
tiveness for the real application. Many other tests using DCA pp. 1170–1175, Nov./Dec. 1993.
[7] R. Das, M. S. Sachdev, and T. S. Sidhu, “A fault locator for radial
algorithm still have shown very accurate and reliable fault loca- sub-transmission and distribution lines,” in Proc. IEEE Power Eng. Soc.
tion results both in a balanced and unbalanced system. Summer Meeting, vol. 1, 2000, pp. 443–448.
CHOI et al.: A NEW FAULT LOCATION ALGORITHM USING DIRECT CIRCUIT ANALYSIS 41

[8] J. Zhu, D. L. Lubkeman, and A. A. Girgis, “Automated fault location and Seung-Jae Lee (S’78–M’88) received the B.S. and M.S. degrees from Seoul
diagnosis on electric power distribution feeders,” IEEE Trans. Power National University, Seoul, Korea, in 1979 and 1981, respectively, and the Ph.D.
Delivery , vol. 12, pp. 801–809, Apr. 1997. degree from the University of Washington, Seattle, in 1988.
[9] K. J. Astrom and B. Wittmenmark, Computer Controlled Systems, 2nd Currently, he is a Professor with the Department of Electrical Engineering
ed. Englewood Cliffs, NJ: Prentice-Hall, 1990. at Myongji University, Yongin, Korea, where he has been since 1988. He is a
Director of Next-Generation Power Technology Center. His main research areas
are protective relaying, distribution automation, and AI applications to power
systems.

Duck-Su Lee received the B.S. and M.S. degrees from Myongji University,
Yongin, Korea, in 2000 and 2002, respectively.
He joined P&C Tech. in 2002, where he is working in the area of protective
relaying.

Myeon-Song Choi (M’96) received the B.S., M.S., and Ph.D. degrees from
Seoul National University, Seoul, Korea, in 1989, 1991, and 1996, respectively. Bo-Gun Jin received the B.S. and M.S. degrees from Myongji University,
Currently, he is an Associate Professor at Myongji University, Yongin, Korea. Yongin, Korea, in 2000 and 2002, respectively. He is currently pursuing the
He was a Visiting Fellow at Pennsylvania State University, University Park, in Ph.D. degree at Myongji University.
1995. His research interests include power system control and protection. His main research interests are power system control and protection.