CIRED 18th International Conference on Electricity Distribution
___________________________________________________________________________________________________________
FAULT LOCATION FOR RADIAL DISTRIBUTION SYSTEMS USING FAULT GENERATED
HIGH-FREQUENCY TRANSIENTS AND WAVELET ANALYSIS
Davood JALALI
Niroo Research Institute (NRI) – Iran
djalali@nri.ac.ir
INTRODUCTION
Abstract: Fault location in radial distribution Systems can
be divided into two important steps. The first is to identify
the fault branch and the second one is to determine the
location of the fault in the faulty branch.
It is rarely less papers in the field of faulty branch
identification. In this paper a new method based on
previous works is presented for identifying the faulted
branch in distribution system. The proposed method takes
advantage of the special properties of wavelet transform to
differentiate between faults occurring along different
laterals of the same main feeder, with equal distance away
form the main substation. The advantages of the method
comparing with the previous ones are; a) JMarti line
model simulation improvement, b) using a specified rang
for s factor instead of a fixed number for a branch of
network depend on it’s topology c) generalizing the method
for untransposed lines and deriving modal matrix for them.
The algorithm of this method has been tested for a test
radial distribution network. Simulation results obtained by
using ATP/EMTP for the sample distribution system are
included for demonstration of the proposed method.
Faults in distribution systems effect power system
reliability, security and quality. Accurate fault location
minimize the time needs to restore power and will reduce
the costs. The application of traditional fault location
techniques that use fundamental voltages and currents at
line terminals for distribution lines with tapped load is
difficult. In recent years they have been used in conjunction
with the concept of traveling ware and wavelet transform
[8]
In distribution systems, fault location method based on the
information provided by DFR is not so common. Usually
the fault location is estimated based on the information
provided by the customer and is verified by the experienced
dispatcher who sends the crew to the suspected location. If
the initial estimation is incorrect, the dispatcher will have
to make another guess and redirect the crew to the newly
suspected location. This procedure may be repeated
several times before successful location of the fault, for
systems with several laterals emanating from the same
feeder.
Several new techniques have been proposed in the recent
years to address this problem. In [1] a fault location
technique for radial transmission lines with multiple loads
taps data available only at the source side of the line is
presented. A technique based on the high frequency
measurement is proposed in [2]. Reference [3] proposed a
CIRED2005
Session No 3
Turin, 6-9 June 2005
Niki MOSLEMI
nmoslemi@nri.ac.ir
fault location technique for rural distribution systems
based on the apparent impedance method approach.
Another technique based on the concept of super imposed
components of the voltages and currents rather than total
quantities are presented in [4]. Nevertheless, the case of a
feeder having multilateral, with measurement normally
available only at the substation where multiple possibilities
of fault location exist for a given recording has not been
fully resolved. The essential problem of such a case is that
a number of possible locations by the same electrical
distance from the substation can be found for a specified
recorded signal at the substation end.
Some attempts to address the problem of multiple solutions
have been made using knowledge based approaches [5].
They are based on not only the information provided by the
measurements but also that information presented in [7]
integrated the information available from a substation
DFR with the known feeder configuration as well as the
protective coordination scheme used for the feeder. This
data may not be accurate or even not available at all at the
substations.
PROPOSED METHOD
According to the traveling wave theory, fault transients will
have different signature at the substation terminals
depending in network junctions which they pass through
from them as travel toward the substation.
These signatures can only show up at the high frequency
study of voltage signals. Fig.1 shows that the fault path
between the substation and the actual fault location
contains different number of network junctions depending
on which lateral is actually faulted. For instance, for a
fault at NQ or at EG, there will be 2 or 3 junctions
respectively between fault and the Node C (substation bus).
Here “Junction” is used to refer to a network node with
more than 2 incident branches. A load may also appear as
a branch, even though most loads being inductive, will be
have like open circuits at high frequencies. Although the
traveling wave signals due to faults at EG and NQ may
have the same frequency spectrum, their amplitude will be
different due to the different attenuation factors resulting
from different number of junctions in their paths
connecting to the substation [8].
In order to determine fault branch in a radial distribution
network, the three-phase voltage of substation supplying
bus must be recorded in the sampling rate of at least 100
KHz.
CIRED 18th International Conference on Electricity Distribution Turin, 6-9 June 2005
___________________________________________________________________________________________________________

Simulation result

This algorithm is tested for the network shown in Fig 1.

This network is part of a real distribution network. The
lines are untransposed by which the EMTP line constant
module is used for extracting J-Marti model.
Sampling rate is 100 KHZ and desired scale the frequency
boundary is 12.5-25KHz. A database is formed for this
network.
At the next step the fault is modeled in six nodes as NQ,
EK, EG and QS, KM, GI in each time. In the Fig 2.
The S factor for the database and for each of six cases as
mentioned above is shown in Fig 3. It is clear that this
algorithm can locate the. Faulted by good estimation.
Fig.1: This diagram used for database generation

The procedure of fault location can be described as

follows: first, the phase voltages are transformed in to the
modal domain, then the modal signals are decomposed into
the wavelet coefficients (WTC), in the next step, the energy
of the signal which is in the bound of 12.5-25 KHz is
described. The amount of this energy is compared with the
database and finally. The path and the distance to the fault
are estimated using the information provided by the WTC
[8, 9].

Calculation routine:

The most important step in this algorithm is the calculation Fig.2: This diagram used for testing the method
of modal matrix. In untransposed line, this matrix is
extracted by “line constant” module of EMTP for desired
frequency range. This matrix is used to transform the
Phase signal to modal signal.[9]
Smodal = T. Sphase (1)
In Simple case, when the line is fully transposed above
formulation is simplified to Clark transformation. (2)
Mode 2 of Voltage signal which is used for wavelet
transformation,after decomposition to 4 detail levels by
db4 will be used for branch identification. The s factor for
this mode is as below.
T
S= [ WTCi u WTCi ] (3)
Where WTC is the wavelet transformation coefficients of
the mode 2 in desired scale.(i is the scale number)
This factor is now ready to compare with the database,
which is provided for each faulty branch of the network.
The amount of S factor is in the specific range between two
nodes S factor, it will show the branch of fault in the
network.
It must be noted that in constraint with the usual method Fig3-1:three phase current signal for fault in EG ,EK,NQ
for modeling J-Marti which a typical model it as a
reference model and is repeated in all the simulation, for
extracting the S factor in forming data base each branch
must be modeled by a unique J-Marti alone.
(Because the virtual nodes of each J-Marty typical model
should not add the nodes of the network)

CIRED2005

Session No 3
CIRED 18th International Conference on Electricity Distribution
___________________________________________________________________________________________________________
Fig3-2:three phase voltage signal for fault in EG ,EK,NQ
Fig3-3: S factors bar chart for comparing database and
simulated data
Fig5: compare S factor for different test (test1:three
phase fault, test2:single phase fault, test3fault with
different incident angle , test4:fault with different load):
Conclusion
CIRED2005
Session No 3
Turin, 6-9 June 2005
A fault location thechnique based on high frequency
signals is developed .The method identifies the faulted
lateral. The identification method is described in detail
with examples .The method is based on the wavelet
decomposition of the transient signals . After the initial
identification stage ,using the post fault steady state
phasors along with the simplified model of the system ,the
exact fault location can be calculated .The identification of
the faulted lateral is based on the voltage signal
decomposition in a frequency spectrum from 12.5 to 25
KHz ,therefore voltage transducer with a bandwidth of 50
KHz are needed
In this paper a powerful method for fault location in
distribution network was proposed. Besides of the good
estimation of this method for determining the path and
distance of faulted point, this method is insensitive to the
initial phase angle of Voltage, and the network loading.The
result is shown in Fig 4.
REFERENCE
[1] Nouri, H., Wang, Ch. and Davies, T., 2001, "An
Accurate Fault Location Technique for Distribution Lines
Tapped Loads Using Wavelet Transform", IEEE Conf. on
Power Tech, Proto, Vol. 3, PP. 44-47
[2] Magnago, F. H. and Abur, A., 1998, "Fault Location
Using Wavelet", IEEE Trans. on Power Delivery, Vol. 13,
No. 4, PP. 1475-1480
[3] Girgis, A. A., Fallon, Ch. M. and Lubkeman, D. L.,
1993, "A Fault Location Technique for Rural Distribution
Feeder", IEEE Trans. on Industry Application, Vol. 29, No.
6, PP. 1170-1175
[4] Agarwal, R. K., Coury, D. V., Johns, A. T. and Kalam,
A., 1993, "A Practical Approach to Accurate Fault
Location on Extra High Voltage Teed Feeders", IEEE
Transactions on Power Delivery, Vol. 8, No. 3, PP. 874-
883
[5] Johns, A. T., Lai, L. L., EL-Hami, M. and Daruvala, D.
J., 1991, "New Approach to Directional Fault Location For
Over Head Power Distribution Feeders", IEE Proceedings-
Gener. Trans. and Distrib., Vol. 138, No. 4, PP. 351-357
[6] Liu, W. H. E.and Zhong, W., 1996, "A Fuzzy Set
Method for Fault Location Identification in Power
Distribution Systems" IEEE Conf. on Decision and
Control, Kobe Japan, Vol. 2, PP. 2208-2212
[7] Zhu, J., Lubkeman, D. L. and Girgis, A. A., 1997,
"Automated Fault Location and Diagnosis on Electric
CIRED 18th International Conference on Electricity Distribution Turin, 6-9 June 2005
___________________________________________________________________________________________________________

Power Distribution Feeders", IEEE Trans. on Power

Delivery, Vol. 12, No. 2, PP. 801-809

[8] Magnago, F. H. and Abur, A., 1999, "A New Fault

Location Technique for Radial Distribution Systems Based
on High Frequency Signals", IEEE Power Eng. Society
Summer Meeting, Vol. 1, PP. 426-431

[9] Niki Moslemi , 2002, "A study and proposal of suitable

method for fault location in M.V. distribution
system",M.S.C. Thesis of Isfahan University of
Thechnology

CIRED2005

Session No 3