You are on page 1of 8

Fault Location Techniques for Simultaneous Faults in

Power Distribution Systems with Distributed


Generations
Wen Fan, Yuan Liao
Department of Electrical and Computer Engineering
University of Kentucky
Lexington, KY, USA
wen.fan@uky.edu, yuan.liao@uky.edu

Abstract— Modern electric power systems are witnessing restoration [2][3]. Although fault location methods have been
deployment of an increasing number of distributed renewable proposed in the past, challenges still exist.
energy resources. Reliability of grid infrastructure plays a key Most of existing methods are impedance based [4]-[8]. The
role in increased use of renewable energies. Fault management is authors of [8] proposed a fault location method for distribution
a key function in reducing outage time and increasing system system based on symmetrical component analysis using
reliability. In the past, various fault location methods have been measurements at the substation and DG sites, and considered
proposed for electric power distribution systems. Most existing uncertainty related to load and fault conditions. Wavelets
fault location methods are applicable to single faults and require analysis technique is used in [9][10] for locating faults.
a number of measurements. This paper proposes new algorithms
Machine learning approach is adopted in [11] that uses
for locating simultaneous, multiple faults for distribution systems
measurements from smart meters across the power grid.
using limited number of data. The method utilizes the voltage
measurements taken from the systems. Bus impedance matrix
Recently, the national grid have witnessed a trend of
techniques is harnessed to establish the connection between the integrating more renewable energies including solar, wind, etc.,
fault locations of multiple faults and the measurements. The into the grid. Such integration has the following impacts.
distinct features of the method include no need of fault type Distribution system is traditionally designed and operated as a
classification, consideration of shunt capacitance of long feeders, radial system. Although feeders may be reconfigured using
applicability to any type of faults, and suitability to systems with sectionalizes or breakers for improved reliability, each feeder is
distributed generations. Evaluation studies have shown still operated as a radial system during normal operation. With
promising results. The methods can be integrated into existing more renewable energy being integrated, many of which is
distribution management system for enhanced fault management distributed, dubbed as Distributed Generation (DG), the
capability for power systems with distributed renewable energy distribution system is no longer radial. The power flow
sources. becomes bi-directional, which means that the power does not
always flow from upstream to downstream, but may flow in
Index Terms-- distributed generation, fault location, power reverse direction. Presence of multiple sources in the system
distribution systems, renewable energy, simultaneous faults will therefore alter the voltage and current profile during a fault
in comparison with presence of only the substation source.
I. INTRODUCTION Another challenge is that DG output may be intermittent and
Electric power distribution systems serve to deliver power rapidly varies over time. Treating it as constant in fault location
to the end customers and faults occurring on any section of the algorithm design will lead to inaccuracy.
distribution systems will interrupt service and cause In addition, multiple faults may occur simultaneously in the
inconvenience or even huge cost to customers. When a fault system, and there is a need for efficient methods to handle this
occurs, protective relays and associated circuits breakers or situation, which can provide reliable fault location estimates
reclosers promptly and selectively open certain parts of the even when the power grid is not equipped with a lot of
circuits according to prescribed design to isolate or de-energize measuring devices.
the faulted component [1]. The rest of the systems will continue This paper aims to propose algorithms to handle multiple
operation. Some loads will be interrupted due to opening of faults for distribution networks with DGs. Section II presents
breakers. To recover power to those interrupted customers, it is the proposed methods. Section III reports evaluation studies,
essential to locate and repair the faulted component first. followed by the conclusion.
Although system reconfiguration may recover partial II. PROPOSED ALGORITHM FOR LOCATING MULTIPLE
interrupted loads, full restoration of power will count on repair FAULTS
of faulted component. Therefore, accurate location of faults
plays a pivotal role in speedy maintenance and fast system
Section A presents the basis for the proposed multiple fault
location methods by examining a single fault, and Sections B,
C and D describe the proposed fault location methods for : the transfer impedance between fault nodes and source
simultaneous, multiple faults. nodes;
A. Basis for fault locaiton : the driving point impedance at the fault nodes;
=[ ] (10)
The fault location algorithm, based on voltages and currents
measured at the sources including local substation and DGs, is
illustrated by examining a fault on a three phase feeder. The = (11)
method will be applicable to single and two phase feeders as
well. No fault type information is required. Without losing The complex power consumed by the fault resistances is
generality, it is assumed that there are three sources, with a bus obtained as
number , , , respectively, and that all sources are three
] {( ) [ ∗
phase. Each source bus consists of three nodes. For example, =[ ( − )] (12)
bus consists of three nodes , , and , corresponding
to phase a, b and c, respectively. It is emphasized that these where, “*” denotes complex conjugate.
assumptions are merely for the sake of easier illustration of the It follows from the reactive power consumed by fault
method and the method will be applicable to a network with resistances being zero that
any number of sources which can be single phase, two phase ( )=0 (13)
or three phase sources.
The current injections to the network include the line where, Imag(.) yields the imaginary part of its argument.
currents at the sources and the fault currents. Therefore, the
voltage during the fault at bus can be written as Given the voltage measurement during the fault at bus and
= − (1) currents at sources, Equ. (13) contains only one unknown
=[ ] (2) variable: the fault location. The bus can be a source bus or
where, any other bus, and most likely be the source bus since the
is the voltage during the fault at bus ; the voltage of each method requires source currents. The Newton-Raphson
node is denoted by , , ; technique can be applied to solve the fault location. The
is the transfer impedance between bus and buses derivation holds for any type of fault with unequal fault
= { , , }; resistances.
is the transfer impedance between bus and fault bus ; Alternatively, by assuming voltage measurements from
is the current injections by sources at buses , consisting of two buses and are available, the voltages during the fault
phase currents at each source bus. For example the source at are given by
bus has current injection , and for phase a, b and = − (14)
c, respectively;
is the fault current at the fault bus flowing away from the = − (15)
network, comprising , and for phase a, b and c,
where, and are voltage measurements at buses
respectively;
and .
T represents matrix and vector transpose.
⋯ It follows from the two equations that
⋯ − =
=
⋯ − (16)
(3)
The fault location is the unknown variable contained in the
= (4) equation, which can be estimated based on Newton-Raphson
technique.
In general, when measurements from more than two buses
=[ ⋯ ] (5) are available, optimal estimation theory can be used to detect
= (6) possible bad measurements and enhance fault location
Equ. (1) is rewritten as estimation accuracy.
= − (7) B. Overview of the proposed methods for locating multiple
Then, the fault current is derived as faults
=( ) [ ( − )] (8) This subsection presents the general ideal of the proposed fault
The voltages at the fault nodes during the fault is given by location algorithms to locate multiple, simultaneous faults on
= − (9) distribution systems. Fig. 1 illustrates a scenario involving two
where, simultaneous faults.
: the voltages at the fault nodes during the fault, comprising
, and ;
, : transfer impedance matrix between bus and
, bus and , respectively;.

Or in more compact form,


= − (22)
Fig. 1. Illustration of multiple faults.

In the figure, two faults occur at point on line segment The fault current vector is obtained as
=
with fault location , and at point on line segment
with fault location , respectively. The line segments ( ) − (23)
can be single phase, two phase or three phase. The faults can be
grounded or ungrounded, and can involve single phase, two
phases or three phases where applicable. Bus impedance matrix
including the fault points will be developed to link the voltage The voltages at the fault points are expressed as
and current measurements at any location with the unknown = − (24)
fault location. The derivation will be based on phase domain so

that any unbalances of the network is automatically considered.
Based on where measurements may be taken, two types of = (25)
algorithms will be presented in the following sections. One type
uses measurements from all source locations and the other type
uses measurements from any selected locations on the network.
= (26)
C. Type 1 algorithm - Fault location algorithm for multiple
faults using measurements from all sources
This algorithm utilizes current measurements captured at local = (27)
substation and DG sites, and voltage measurements captured
at selected locations. Since currents at source buses are needed, The dimensions of the matrices depend on the available
it is natural to use voltages at source buses for fault location. phases of involved circuits. For example, assuming both faults
Source impedances are not needed and bus impedance matrix are three phases, then takes the following expanded
does not include source impedance. form:
The fault currents at point and are denoted as , and
, respectively.
The current injections to the network include the line
currents at the sources and the fault currents. Therefore, the =
voltage during the fault at any bus can be written as
= − − (17)

= −[ ][ ] (18)
(28)
Now assume that the voltages at two buses and are
known. The following two equations are obtained: The following equations are obtained considering that the
= −[ ][ ] (19) reactive power, i.e., the imaginary part of the complex power
consumed by the fault resistances is zero,
= −[ ][ ] (20) ∗
= 0 (29)

or ∗ (30)
=0

= − (21) Then, the two unknown fault locations can be determined
by solving this set of two real equations.
An alternative solution is described as follows when more
Where, measurements are available. Now assume that the
, : the voltage during the fault at bus , , measurements from another two buses and are known.
respectively; The following equations are obtained:
, : transfer impedance matrix between bus and ,
bus and , respectively; = − (31)
, : transfer impedance matrix between bus and
, bus and , respectively;
= , : the voltage preceding the fault at bus , ,
− (32) respectively;
, , , : transfer impedance matrices.

It follows from (23) and (32) that Or in more compact form,


( ) [ ( − )] = − (38)
=( ) [ ( − )](33)
The voltage change due to fault is
∆ =− (39)
This equation contains two unknown fault locations.

Separating the equation into two real equations, from which
The fault current vector is obtained as
the fault location can be obtained using Newton-Raphson
method.
Since source currents are needed, it is natural to use the =− Δ (40)
voltages at the source buses for fault location purposes. The
derivation indicates that voltage measurements from any bus
can be utilized for fault location purposes. Since branch Furthermore, the voltage during the fault at fault buses are
currents can be expressed as voltages of line terminals, branch given by
currents at any location if available, can also be used in fault
location. = − (41)
It is evident that this method is applicable to a radial
network that contains only one source at the local substation. Where
, : the voltage during the fault at fault bus , ,
D. Type 2 algorithm - Fault location algorithm for multiple respectively;
faults using measurements from selected locations , : the voltage preceding the fault at fault bus , ,
This algorithm utilizes voltage measurements from any respectively;
locations that may be other than source locations, and do not , : driving point impedance matrix at , ,
need current measurements from source sites. respectively;
This formulation assumes that the source impedances are , : the transfer impedance matrix between and ,
known and the bus impedance matrix includes the source = .
impedances. DGs are modeled as a voltage source behind an The equation can also be written as
impedance. = − (42)
Based on superimposition theory, the voltage at any bus
during the fault can be expressed as Prefault voltages at the fault bus and can be
= −[ ][ ] (34) expressed in terms of fault location and the prefault voltages at
bus and , bus and . For example, is derived as
Where
follows:
: the voltage at bus during the fault;
: the voltage at bus during preceding the fault; = + (43)
: the transfer impedance between bus and fault bus ; 1−
: the transfer impedance between bus and fault bus . = + (44)
Now assume that the measurements from two buses and 2 (1 − )
are known. The following two equations are obtained:
= −[ ][ ] (35)
where
= −[ ][ ] (36) : total series impedance matrix of the feeder between and
;
: total shunt admittance matrix due to shunt capacitance of
or
the feeder between and ;
: a three by three identity matrix.
= − (37) Where , are prefault voltage at and , and can be
estimated based on substation voltage and network data, or
obtained through network management functions such as state
Where, estimation and SCADA data.
, : the voltage during the fault at bus , , The reactive power consumed by fault resistances at the
respectively; two fault locations are zero, i.e.,
∗ in Matlab simpowersystem. Simulation studies are performed
∗ =0 (45) to obtain required voltage and current measurements by posing
faults with various fault conditions with different fault types,
impedances and locations. The short circuit program is
Solving the equation will yield the two unknown variables corroborated by the Matlab simpowersystem simulation results.
and . After the simulated data are obtained, the proposed algorithms,
which are implemented in Matlab, are applied to find the fault
location.
III. EVALUATION STUDIES The estimation accuracy is measured by the percentage
This section presents the evaluation studies for verifying the error calculated as
proposed fault location methods. Fig. 2 depicts the 21-bus
distribution system used in the study, with phases, loads, feeder |Actual location − Estimated location|
length being labelled, adapted from [1]. The system includes %error = × 100(46)
Total length of the main feeder
three distributed generations. It is noted that there is no
restriction on the number of DGs present in the system for the Evaluation studies have demonstrated that the proposed
proposed methods. Without losing generality, all loads are methods yield highly accurate results for single or multiple fault
assumed to have the same power factor of 0.95. scenarios under diverse fault conditions. The short circuit
program developed by the authors is used to generate simulated
data. Then the voltage at bus 1 is added a 0.2% error and then
the proposed methods are applied to obtain the fault location.
Results under various fault conditions are presented as follows.
Fig. 3. illustrates modeling of fault types and resistances
using a LLLG and LLLG fault. Each LLLG fault include phase
fault resistance, e.g., , , for fault 1, and ground
resistance, e.g., for fault 2. Other types of faults are
modeled using its variant. For example, for AG fault, is set
to zero, and are set to infinite, and is the fault
resistance. For AB fault, and are set to infinite, and the
sum of and is the fault resistance. For ABG fault,
is set to infinite, and , , and are fault resistances.
Other types of faults that involve other phases can be modelled
similarly.

LLLG fault LLLG fault


LLLG-LLLG

Fig. 3. Illustration of modeling of fault type and resistance

TABLE I presents the fault location results obtained by


Fig. 2. A 21-bus system used in the study. Type 1 algorithm using voltage measurements from bus 1 and
To verify the proposed algorithms, simulated data are obtained 20 and currents from all sources. It is shown that highly
by a short circuit program developed by the authors in Matlab accurate results are achieved by the proposed methods.
that is capable of dealing with simultaneous, multiple faults that
occur on any section of the system. The system is also modelled
TABLE I. TYPE 1 ALGORITHM - FAULT LOCATION RESULTS TABLE III presents the fault location results obtained by
USING VOLTAGES FROM BUS 1 AND BUS 20 Type 1 algorithm using voltage measurements from bus 1, 14
and 20 and currents from all sources, for cases involving three
Case Fault Fault section, FL Fault FL simultaneous faults. It is evinced that the proposed methods
type number and fault (p.u.) res. Esti.
type (ohm) err. (%) have yielded quite accurate results.
Single Fault 1 (1-2, AG) 0.6 0.08 TABLE III. TYPE 1 ALGORITHM - FAULT LOCATION RESULTS
50 USING VOLTAGES FROM BUS 1, BUS 14, AND BUS 20
fault
Single Fault 1 (2-5, BC) 0.3 0.5 0.03
fault Case type Fault Fault section, FL Fault FL
Single Fault 1 (2-5, ABG) 0.8 [1,1, 0.04 number and fault (p.u.) res. Esti.
fault 50] type (ohm) err. (%)
Single Fault 1 (1-2, ABC) 0.4 [1, 0.5, 0.01 Multiple Fault 1 (1-2, AG) 0.4 10 0.70
fault 2] faults Fault 2 (8-11, BC) 0.6 0.5 0.10
Single Fault 1 (5-8, ABCG) 0.6 [1,1,1, 0.08 Fault 3 (18-19, AG) 0.5 5 0.00
fault 30] Multiple Fault 1 (1-2, BC) 0.8 1 0.00
Multiple Fault 1 (1-2, AG) 0.3 1 0.01 faults Fault 2 (8-9, BG) 0.2 5 0.07
faults Fault 2 (16-17, CG) 0.6 10 0.12 Fault 3 (18-19, 0.7 [1,1, 0.02
Multiple Fault 1 (1-2, BG) 0.7 20 0.62 ABC) 1]
faults Fault 2 (5-8, AB) 0.3 1 0.06 Multiple Fault 1 (2-5, AG) 0.6 10 0.03
Multiple Fault 1 (2-3, AG) 0.8 10 0.54 faults Fault 2 (8-9, BG) 0.6 15 0.29
faults Fault 2 (18-21, 0.5 [0.5,0.5 0.01 Fault 3 (16-18, AC) 0.5 1 0.32
ABC) 0.5] Multiple Fault 1 (5-7, AB) 0.3 1 0.02
Multiple Fault 1 (2-5, CG) 0.6 5 0.00 faults Fault 2 (8-9, BG) 0.2 5 0.23
faults Fault 2 (8-10, AG) 0.3 10 0.07 Fault 3 (18-19, 0.9 [1,1, 0.25
Multiple Fault 1 (1-2, AG) 0.6 20 0.40 ABG) 20]
faults Fault 2 (11-13, 0.8 [1, 1,1, 0.02 Multiple Fault 1 (1-2, AC) 0.2 0.5 0.01
ABCG) 10] faults Fault 2 (11-13, 0.7 [1,1,2 0.12
ABG) 0]
Fault 3 (18-19, 0.5 [0.5,0. 0.00
It is noted that to locate single faults, voltages from one bus ABCG) 5,0.5,
will suffice. For example, using voltage from bus 1, the fault 20]
location results for the first five cases listed in TABLE I can be TABLE IV presents the fault location results obtained by
obtained, and the estimation errors are 0.14%, 0.02%, 0.03%, Type 2 algorithm using voltage measurements from bus 1 and
0.01% and 0.06%, respectively. 20. Note that no currents are used for Type 2 algorithm.
Similarly accurate results are acquired.
TABLE II. TYPE 1 ALGORITHM - FAULT LOCATION RESULTS TABLE IV. TYPE 2 ALGORITHM - FAULT LOCATION RESULTS
USING VOLTAGES FROM BUS 1, BUS 8, AND BUS 20 USING VOLTAGES FROM BUS 1 AND BUS 20

Case type Fault Fault section, FL Fault FL Case Fault Fault section, FL Fault FL
number and fault (p.u.) res. Esti. type number and fault (p.u.) res. Esti.
type (ohm) err. (%) type (ohm) err. (%)
Multiple Fault 1 (1-2, AG) 0.4 10 0.61 Single Fault 1 (1-2, AG) 0.4 0.06
faults 50
Fault 2 (5-6, BG) 0.6 5 0.37 fault
Fault 3 (8-11, AG) 0.5 5 0.07 Single Fault 1 (8-11, AC) 0.3 1 0.26
Multiple Fault 1 (1-2, BC) 0.8 1 0.01 fault
faults Fault 2 (8-11, AG) 0.2 5 0.11 Single Fault 1 (5-8, BCG) 0.7 [1,1, 0.10
Fault 3 (13-16, 0.7 [1,1, 0.03 fault 50]
ABC) 1] Single Fault 1 (8-9, BG) 0.3 30 0.04
Multiple Fault 1 (2-5, ABCG) 0.6 [0.5,0. 0.02 fault
faults 5,0.5, Single Fault 1 (13-14, 0.6 [1,1,1, 0.01
20] fault ABCG) 30]
Fault 2 (5-8, BCG) 0.6 [1,1,2 0.05 Multiple Fault 1 (1-2, AG) 0.6 1 0.02
0] faults Fault 2 (13-15, BG) 0.5 10 0.12
Fault 3 (18-20, CG) 0.5 10 0.01 Multiple Fault 1 (1-2, BG) 0.6 5 0.04
Multiple Fault 1 (5-7, AB) 0.3 1 0.02 faults Fault 2 (5-8, AB) 0.7 0.5 0.02
faults Fault 2 (8-9, BG) 0.2 5 0.07 Multiple Fault 1 (2-3, BC) 0.4 0.5 0.12
Fault 3 (11-13, 0.9 [1,1, 0.16 faults Fault 2 (5-8, ABC) 0.6 [0.5,0.5 0.03
ABG) 20] 0.5]
Multiple Fault 1 (1-2, AC) 0.2 0.5 0.01 Multiple Fault 1 (2-5, CG) 0.6 20 0.25
faults Fault 2 (5-6, BG) 0.7 5 0.20 faults Fault 2 (11-12, AG) 0.7 30 0.52
Fault 3 (13-15, BG) 0.5 10 0.23 Multiple Fault 1 (8-11, AG) 0.5 50 0.44
TABLE II presents the fault location results obtained by faults Fault 2 (16-18, 0.8 [1,1,1, 0.05
Type 1 algorithm using voltage measurements from bus 1, 8 ABCG) 10]
and 20 and currents from all sources, for cases involving three It is noted that to pinpoint single faults, voltages from one
simultaneous faults. Again, accurate results are obtained. bus will suffice. For example, using voltage from bus 1, the
fault location results for the first five cases listed in TABLE IV
can be obtained, and the estimation errors are 0.01%, 0.08%, matrix singularity occurs and no solution is identified. The fault
0.11%, 0.05% and 0.11%, respectively. type and location are not shown. The last column shows the bus
from which voltage is taken. Type II algorithm does not yield
TABLE V presents the fault location results obtained by solution to these cases either.
Type 2 algorithm using voltage measurements from bus 1, 8
and 20 and currents from all sources, for cases involving three TABLE VII. CASES WITHOUT SOLUTION DUE TO INSUFFICIENT
simultaneous faults. Very accurate results are achieved. NUMBER OF METERS

TABLE V. TYPE 2 ALGORITHM - FAULT LOCATION RESULTS Case Fault Fault section Meter
number number placement bus
USING VOLTAGES FROM BUS 1, BUS 8, AND BUS 20
1 Fault 1 (2-5) 1
Case type Fault Fault section, FL Fault FL Fault 2 (8-11)
number and fault (p.u.) res. Esti. 2 Fault 2 (2-5) 1, 3
type (ohm) err. (%) Fault 3 (8-11)
Multiple Fault 1 (1-2, AG) 0.4 30 0.60 3 Fault 1 (5-7) 1, 6
faults Fault 2 (8-11, BG) 0.6 10 0.44 Fault 2 (8-10)
Fault 3 (18-19, CG) 0.5 50 0.42 4 Fault 1 (1-2) 1, 10, 14
Multiple Fault 1 (1-2, AB) 0.8 0.5 0.02 Fault 2 (11-13)
faults Fault 2 (2-5, CG) 0.2 50 0.22 Fault 3 (18-19)
Fault 3 (8-11, ABC) 0.6 [1,1, 0.00 5 Fault 1 (1-2) 1, 14, 20
1] Fault 2 (2-5)
Multiple Fault 1 (1-2, BG) 0.7 10 0.08 Fault 3 (8-11)
faults Fault 2 (5-8, BCG) 0.6 [1,1,2 0.03
0]
Fault 3 (8-11, CG) 0.3 20 0.02 Further studies related to fault location observability and
Multiple Fault 1 (2-5, AB) 0.3 0.5 0.03
faults Fault 2 (5-8, AG) 0.3 10 0.35
determination of optimal placement of sufficient meters to
Fault 3 (11-13, 0.8 [1,1, 0.01 uniquely locate each fault across the network may be
BCG) 20] performed, following the method presented in [1]. Though the
Multiple Fault 1 (1-2, AC) 0.3 0.5 0.01 study may be more complex since simultaneous faults are
faults Fault 2 (8-11, AG) 0.6 10 0.61 present.
Fault 3 (13-15, BG) 0.6 30 0.14
IV. CONCLUSION
This paper presents two types of algorithms for locating
TABLE VI presents the fault location results obtained by multiple simultaneous faults on distribution systems, which is
Type 2 algorithm using voltage measurements from bus 1, 14 also applicable to single faults. Type 1 algorithm requires
and 20 and currents from all sources, for cases involving three currents from all sources and voltages from selected locations,
simultaneous faults. Again accurate results are obtained. but does not need source impedance. Type 2 algorithm needs
only voltages from selected locations, does not need currents
TABLE VI. TYPE 2 ALGORITHM - FAULT LOCATION RESULTS
from any location, and needs source impedance. Both
USING VOLTAGES FROM BUS 1, BUS 14, AND BUS 20 algorithms are capable of handling unbalances and any type of
faults, and still work for situations where only limited
Case type Fault Fault section, FL Fault FL measurements are available. Both algorithms produce highly
number and fault (p.u.) res. Esti. accurate fault location estimates based on simulation studies.
type (ohm) err. (%) With increasing deployment of distributed generations, the
Multiple Fault 1 (1-2, AG) 0.4 10 0.24 proposed method will speed up maintenance process when
faults Fault 2 (5-8, BG) 0.6 5 0.31 faults occur on the system and thus improve system reliability.
Fault 3 (16-18, CG) 0.5 30 0.00
Multiple Fault 1 (1-2, AB) 0.8 0.5 0.02
faults Fault 2 (2-5, CG) 0.2 50 0.60 REFERENCES
Fault 3 (18-20, 0.6 [1,1, 0.00 [1] Wanjing Xiu and Yuan Liao, “Fault location observability analysis on
ABC) 1] power distribution systems,” Electric Power Components and Systems,
Multiple Fault 1 (2-3, AG) 0.7 10 0.18 vol. 42, no. 16, pp. 1862-1871, 2014.
faults Fault 2 (8-10, AG) 0.6 30 0.55 [2] R. Krishnathevar and EngEngNgu, “Generalized impedance-based fault
Fault 3 (13-16, CG) 0.3 20 0.00 location for distribution systems,” IEEE Transactions on Power
Multiple Fault 1 (5-7, AB) 0.3 0.5 0.07 Delivery, vol. 27, no. 1, pp. 449 - 451, Jan. 2012.
faults Fault 2 (8-11, BG) 0.3 10 0.31 [3] Yuan Liao, “Electric distribution system fault location considering shunt
Fault 3 (18-19, 0.8 [1,1, 0.00 capacitances,” Electric Power Components and Systems, vol. 41, no. 5,
BCG) 20] pp. 519-536, Feb. 2013.
Multiple Fault 1 (1-2, AC) 0.3 0.5 0.00 [4] Gustavo D. Ferreira, Daniel S. Gazzana, Arturo S. Bretas, and Afonso S.
faults Fault 2 (8-11, AG) 0.6 10 0.69 Netto, “A unified impedance-based fault location method for generalized
distribution systems,” IEEE Power and Energy Society General Meeting,
Fault 3 (13-16, BG) 0.6 30 0.07
July 22-26, 2012, San Diego, CA, USA.
It is noted that some multiple faults may not be uniquely [5] Yimai Dong, Ce Zheng, and M. Kezunovic, “Enhancing accuracy while
determined if the number of meters is not enough, or if the reducing computation complexity for voltage-sag-based distribution
meters are not placed in the needed locations. Type 1 algorithm fault location,” IEEE Transactions on Power Delivery, vol. 28, no. 2, pp.
is used to run the example cases shown in TABLE VII, where 1202 - 1212, April 2013.
[6] D.S. Gazzana, G.D. Ferreira, A.S. Bretas, A.L. Bettiol, A. Carniato, combined overhead line and cable distribution networks,” 7th
L.F.N. Passos, A.H. Ferreira, and J.E.M. Silva, “An integrated technique International Conference on Electrical and Electronics Engineering
for fault location and section identification in distribution systems,” (ELECO), December 1-4, 2011, Bursa, Turkey.
Electric Power Systems Research, vol. 115, pp. 65-73, October 2014. [10] M. Goudarzi, B. Vahidi, R.A. Naghizadeh, and S.H. Hosseinian,
[7] Javad Sadeh, Ehsan Bakhshizadeh, and Rasoul Kazemzadeh, “A new “Improved fault location algorithm for radial distribution systems with
fault location algorithm for radial distribution systems using modal discrete and continuous wavelet analysis,” International Journal of
analysis,” International Journal of Electrical Power & Energy Systems, Electrical Power & Energy Systems, vol. 67, pp. 423-430, May 2015.
vol. 45, no. 1, pp. 271-278, February 2013. [11] M. Majidi, M. Etezadi-Amoli, and M. Sami Fadali, “A novel method for
[8] Juan J. Mora-Flórez, Ricardo A. Herrera-Orozco, and Andres F. Bedoya single and simultaneous fault location in distribution networks,” IEEE
Cadena, “Fault location considering load uncertainty and distributed Transactions on Power Systems, vol. 30, no. 6, pp. 3368 - 3376,
generation in power distribution systems,” IET Generation, November 2015.
Transmission & Distribution, vol. 9, no. 3, pp. 287 – 295, February 19,
2015.
[9] Behnam Feizifar, Mahmoud Reza Haghifam, and Soodabeh Soleymani,
“Application of continuous wavelet transform for fault location in

You might also like