Abstract

For the past fifty years, electric power systems have rapidly grown. This has
resulted in a large increase of the number of lines in operation and their total
length. These lines experience faults which are caused by storms lightning,
snow, freezing rain, insulation breakdown and short circuits caused by birds and
other external objects . In most cases, electrical faults manifest in mechanical
damage, this must be repaired before returning the line to service. The
restoration can be expedited if the location of the fault is either known or can be
estimated with reasonable accuracy. Speedy and precise fault location plays an
important role in accelerating system restoration, reducing outage time and
significantly improving system reliability. This paper presents an improved fault
location method based on the traveling wave theory of the transmission lines.
Fault transients recorded at one end of a transmission line are processed to
determine the distance from the fault location.

Key words: Fault location method, Distribution power system, System

restoration.
Introduction
Power quality requirements resulting from the deregulated electrical markets
have motivated the improvement of fault location methods in distribution
systems to speed up the restoration process. Faults and outages affect power
quality in terms of service continuity and disturbance propagation. Utilities are
forced to improve quality indexes associated to these phenomena in order to be
competitive in the current electric open market. Faults in power distribution
systems cause supply interruptions being responsible of process disturbances,
information and economic los s and equipment damage among others. Fault
Location includes the determination of the physical location of the fault.
Nowadays, about 80% of interruptions are caused by faults in distribution
networks and the application of fault location algorithms developed for
transmission system is not an easy task due to the topology and operating
principles of the first (i.e., nonhomogeneous feeders, load taps, laterals, radial
operation and the available measuring equipment).

The presence of laterals imp lie s that the distance obtained match to several
fault locations in the power distribution system and causes the multiple
estimation problem. This cause problem to the maintenance team because it is
not easy to determine the real fault location in a high spread system, delaying
The restoration of the power service. To the aforementioned, some strategies for
fault location in distribution systems have been developed. There exists a
variety of approaches for locating faults in power distribution systems. Most of
these methods estimate the relative distance to the fault from data acquisition
provided by the protection devices. The performance of those techniques can be
affected due to some particular characteristics of the respective system, such as
unbalanced system, non-homogeneous conductors, systems with high
ramification degree, etc.
This paper reviews selected fault location techniques proposed for distribution
systems. In general, these methods can be classified in three broad categories,
which are Impedance and Other Fundamental Frequency
Component Based Methods, High frequency components and travelling wave
based methods and knowledge based method. Knowledge-based method may be
divided into three groups: Artificial intelligence and statistical analysis based
methods, Distributed device based methods and Hybrid methods.

Distribution system Topology

The most important aspects considered by fault location methods are directly
related to the characteristics of distribution systems:
(a)Heterogeneity of feeders given by different size and length of cables,
presence of overhead and underground lines, etc.
(b)Unbalances due to the untransposed lines and by the presence of single,
double and three phase loads.
(c)Presence of laterals along the main feeder.
(d)Presence of load taps along the main feeder and laterals.
These aspects introduce errors in the estimation of the fault locations performed
by means of a simplified model. The usual distribution line model (short lines)
is the lumped-parameter model, applying symmetrical components on phasor -
based algorithms. The most common type of faults in distribution systems are
short circuits. These faults including such as single line-to-ground, Line-to line,
double line-to ground, three phase and three phase-to-ground faults with
different values of fault resistance between 0 and 50.
Types of fault location methods
Researchers have done considerable work in the area of fault diagnosis
particular to radial distribution systems. Traditional outage handling methods
were based on the customer trouble calls. Here the geographic location of the
caller and the connectivity of the distribution network have to be overlapped
exactly for the exact location of fault. Also, there might not be any calls during
night-time, which poses a problem for the operator in locating the fault. In
recent years, some techniques have been discussed for the location of faults
particularly in radial distribution systems. These methods us e various
algorithmic approaches, where the fault location is iteratively calculated by
updating the fault current. A brief overview of the algorithmic approaches has
been presented in the following section.

Impedance and Other Fundamental Frequency

Component Based Methods:
The distance of fault from the primary distribution bus to the fault location is
estimated by impedance based method. Volt age and current values measured at
one end or two ends of the line are required in this method. This method used
mathematical equation to estimate the fault location. Most of the method
employs for radial system only and require other information such as circuit
breaker status, fault current waveforms, and fault indicator status for non-radial
system.
In these approaches, the first, the fault types and faulted phases are identified.
Next, the apparent impedance is calculated based on the selected voltage and
selected current. If Load currents at different taps are not considered so they are
sources of error. Girgis presented equations to calculate all kinds of faults
occurring at the main feeder and a single-phase lateral. Loads were considered
as constant impedance loads. It did not consider dynamic nature of the loads.
Performance of the technique in situations where cables were used could also be
an issue. Sahaetal proposed method is devoted for estimating the location of
fault s o n radial MV system, which could include many intermediate load taps.
In this method non homogeneity of the feeder sections was also taken into
account. A distribution utility MV networks were used as an example, but he
supposed the system was balanced. Santosoetal gave equations to calculate all
kinds of faults occurring at the main feeder and a single-phase lateral. He used a
different method to take account of the fault resistance. Dasetal. Suggested a
technique that used the fundamental frequency
Voltages and currents measured at a line terminal before and during the fault.
The fault location technique was described by considering a single-phase-to-
ground fault on a radial system. Nevertheless, he still considered the line were
fully transposed, and was only good for line-to-ground faults Choietal locating
faults by solving a quadratic equation resulting from the direct circuit analysis.
It assumed all load impedance was accurately known. Sengeretal ., proposed
method that it was based on measurements provided by intelligent electronics
Devices (IEDs) with built-in oscillography function, installed only at the
substation level and on a database that stores information about the network
topology and its electrical parameters. In particular on 11kV networks, applying
this method is very difficult, if not impossible, to establish reliable statistical
estimates. Some of this method also time consuming due to iterative process
and need to know the fault type before specific equation could be applied. Liao
presented a fault location algorithm by utilizing synchronized or
unsynchronized pre-fault and fault voltage and current measurements from both
ends of the line without requiring line parameters. This method applied to
power line parameter estimation relies on synchronized phasor measurements
and require different system operating conditions from which to estimate the
parameters.
High Frequency Components and Traveling Wave Based
Methods:
The view of this method was based on the reflection and transmission of the
fault generated travelling waves on the faulted power network. Although in this
technique fault can be located with a high accuracy, the implementation is
complex and more expensive than the implementation of impedance based
techniques. Because it needs too many added equipment, such as the GPS
system, fault transient detectors and diagnostic software. Furthermore, due to
the complex configurations of distribution systems, the configuration or the
sites to install the fault transient detectors become very difficult. Thomas et al
used the cross –correlation function between the incident wave and the reflected
wave to locate faults in distribution systems. He used both the double-ended
meth o d and the single-ended method. The double-ended method could provide
an accurate result if the fault happened at the line where the fault recorders were
installed or a t the main feeder.

The single-ended meth o d did not work well. One of the negative aspects of
these methods is the necessity of measuring devices with a very high sampling
rate (MHz) and the author did not present any idea to overcome the problem
caused by multiple discontinuities in distribution systems.
Bo et al presented another method to locate faults. For distinguishing the
reflected wave from the fault point and that from the remote bus bar, a new fault
locator unit was developed, which captured high frequency voltage signals
between 1 and 10 MHz. In this method the effects of tapped-off loads are found
significant, and caused problems in identifying the fault location. He did not
solve the problem completely.
Tang et al suggested a technique based on the voltage magnitude difference
between the device’s terminal voltages. This method needed to insert some
equipment into to distribution feeders. It could not be done only using the
terminal measurements. Magnago et al offered method based on the high
frequency signals measured at the substation. Based on the use of the wavelet
transform, usually adopting the discrete wave let transform (DWT), due to its
straightforward implementation and the reduced computational time it requires
correlation analysis between transmited and reflected waveform is performed.
To use this method, simulation need to be done performed for each distribution
system so that users know the wavelet coefficients of different fault laterals, and
can identify the faulted lateral based on these coefficients.
Utilities cannot afford to implement a solution on a distribution feeder if there
has to be significant configuration and modelling of each specific feeder. It
never would be able to keep it up-to-date, even if a utility could enter feeder-
specific topology information during initial setup. Borgheti et al., proposed
method based on the continuous-wavelet transform (CWT) for the analysis of
voltage transients due to line faults. Correlation exists between typical
frequencies of the CWT-transformed signals and specific paths in the network
covered by the travelling waves originated by the fault. Shown uncertainty
levels lower than 200 Hz on the estimate of the frequency associated to the
paths covered by the travelling waves. This value provides an indication of the
uncertainty of the estimated fault location. The system that implemented was
balanced and Accuracy of this approach is low too. According to
aforementioned section, in general, the most important drawback of these
methods is the necessity of measuring devices with a very high sampling rate
(MHz)

Knowledge-Based Method:
The third category is knowledge-based method. This method can be divided into
three groups:
A. Artificial intelligence and statistical analysis based methods
B. Distributed device based methods
C. Hybrid methods
A. Artificial Intelligence (AI) and Statistical Analysis Based
Methods:
There are several artificial intelligent methods such as Artificial Neural network
(ANN), Fuzzy Logic (FL), Expert System (ES) and Genetic Algorithm (GA),
etc., with the development of computers emerged. These methods can help
operators or engineers to do mu c h laborious work. By using these methods, the
time factor is substantially reduced and human mistakes are avoided. Therefore,
many researchers used AI based methods in distribution system fault locations.
A multi-way graph partitioning method is employed by Bi et al., based on
weighted minimum degree reordering to partition a large -scale power network
into some sub-networks. The speed of the distributed fault section estimation
system mad e it possible to use it as an on-line system. Al-Shaher et al.,
developed a fault location method for multi-ring distribution systems using
neural network. He used the feeder fault voltage, circuit breaker status, real
power of feeders during the normal condition, and real power of feeders during
short circuit, etc., to train the neural network. W en proposed a method that
constructed a probabilistic causality matrix to represent the probabilistic
relationship between faulted sections and protective device action. He applied
the parsimonious set covering theory to make the faulted section estimation as
an integer-programming problem.
Finally, a refined g en e tic algorithm (RGA) was adopted to solve the problem,
Based on the “natural selection, best survival” theory, the RGA found the most
reasonable hypothesis or hypotheses based on the evaluation result of each
hypothesis evaluated by set covering theory. Thukaram et al.,(2002) offered
method estimated the voltage magnitude and phase angle at all load buses
through state estimation. A threshold was used to detect the fault path. Chen et
al., used a cause-effect network to represent causality between faults and the
actions of protective devices. The cause-effect network’s features of high-speed
inference and ease of implementation made it feasible to implement an on-line
fault section estimation system. Based on the actions of protective devices, the
network could quickly find faulted section candidates. Mora e t a l ., presented
an alternative solution to the problem of power service continuity associated to
fault location. A methodology of statistical nature based on finite mixtures is
proposed. A statistical model was obtained from the extraction of the magnitude
of the voltage sag registered during a fault event, along with the network
parameters and topology. The approach is based in the statistical modelling and
extraction of the sag magnitude from voltage measurements stored in fault data
bases. The determination of groups of well-defined characteristics which was
allows an optimization in the classification of data thus ensuring good model
accuracy. A significant feature is the low investment cost for the
implementation of the fault detection system based in the proposed method.
Martinas et al., proposed an approach used the Eigenvalue and an artificial
neural network based learning algorithm. The neural network was trained to
map the non-linear relationship existing between fault location and
characteristic Eigen value. This approach used the eigenvalue/eigenvector and
an artificial neural based learning algorithm. The main characteristics and
particularities of the proposed method were: Reduced number of input signals
(this was an importance aspect due to the non-us e of voltage detectors);
Recognition of the faults type and identification of faulty line or lines; Location
of the fault, independent of his presence at the moment of the analysis; almost
independent on harmonics influence. Simulation results presented show that the
proposed algorithm was a promising technique for fault location on distribution
power systems. He also proposed a method for locating faults for parallel
double-circuit distribution lines (Sous a et al.). It based on impedance
calculations or on travelling-wave analysis. The Clarke-Concordia
transformation was used to transform line currents into current components.
Based on these components, a data sample correlation matrix was constructed.
The eigenvalues of the correlation matrix had a non -linear relationship with the
fault distance. A neural network was used to extract the unknown relationship.
B. Distributed Device Based Methods:
Another type of knowledge-based technique is distributed device based methods
for locating fault. Wang et al., presented a mathematical approach that located
faults based on installed voltage sensors’ information and the network’s
topological structure. The relation of the voltage sensors with sections was
formulated as a matrix. The other matrix was constructed based on the
topological relation between sections and nodes in an electric network. Through
some matrix operations, all faulted sections could be found. Mokhlis offered the
method that it worked by matching the actual voltage magnitude and phase
angle with the simulated ones from fault analysis that stored in a database. The
method was simple and inexpensive for implementation since its only requires
single measurement at the monitored bus. Any changes such as load variations
or network reconfiguration could be adapted by this method by updating the
database. But it could not improve well to consider the effect of fault resistance.
Kezunovic, proposed another type of knowledge-based technique is by
matching measured data with historical fault data. Whenever fault occurs in the
distribution network, the waveform of voltage sags measured at the substation is
recorded into the database together with the known fault type and location. This
database is updated whenever fault occurs in the system.

When actual fault occurs, the measured voltage sags wave forms a t the
substation are compared to all the voltage sags waveforms in the database. The
most matching waveform in the database will give the type of fault and the
location. The drawback of this method is that it will not work if the actual fault
never happens at particular location, or not recorded in the database.
C. Hybrid Methods:
Almost all of the above methods locate faults based on one algorithm, such as
the fault distance calculation or operated protective device’s status analysis, to
locate faults. Some have investigated the use of hybrid methods that locate
calculation or operated protective device’s status analysis, to locate faults. Some
have investigated the use of hybrid methods that locate faults based on more
than one algorithm to achieve a more accurate estimation of the faulted section.
Zhu et al., proposed a hybrid method that computed the fault distance using
measurements available at the substation. He used post-fault values of current or
voltage to reduce the multiple estimation problems induced by the existence of
multiple fault points in the network with the same impedance.
To identify the actual fault location, a fault diagnosis procedure was applied to
rank the list of multiple potential fault locations. By doing a circuit simulation,
the operation of a particular combination of protective devices and the load
change pattern during different fault scenarios could be obtained. Then by
matching the fault situation to these scenarios, the actual faulted section could
be determined. However, this diagnosis procedure modelled the circuit and
simulated different fault scenarios, which were time-consuming tasks, and the
modelling process needed to be done again for another system. Järventausta et
al., used the fault distance calculation, fault detector information, and
geographical information of the distribution system to locate faults.
Zhong et al., presented a method to locate faults based on fault current
measurements, fault currents calculated from short circuit analysis, and system
operators’ experience. Lee et al., calculated the fault distance first to provide
some fault location candidates. Then, by current pattern matching and
interrupted load analysis, the candidate pool was reduced. Khosravi developed a
framework for fault detection and modelling when uncertainty in the plant was
present. In the proposed method, a fault alarm was fired when an inconsistency
between the behaviours of the system and the model emerges. Afterwards, the
behaviour of the faulty system is modelled using an Adaptive Neuro Fuzzy
Inference System (ANFIS). The identified model can be used for the fault
accommodation task.
Mora et al., proposed a fault location approach based on the current waveforms
measured at the power substation and the knowledge of protective device
setting and ANFIS nets. The ANFIS nets proved the capability to locate the
fault in a specific zone of power distribution system and showed validation
errors lower than 1% to locate faulted zones. This approach did not use the
electrical model of the network and the signal treatment to obtain descriptors is
simple. The fact of not using the electrical network model was an additional
Advantage because it was not always easy to have the electrical power system
parameter values. Fuzzy inference is used in both methods (Khosravi, Mora et
al.,) to deal with uncertainty inherent in these methods. Fan (Chunju et al.,)
proposed a fault location method employing wavelet fuzzy neural network to
use post-fault transient and steady-state measurements. However, this method
was not influenced by the fault resistance and load current but he assumed the
system was balanced and it tacked a lot of time in off line mode the training and
calculation for the fault location in industrial distribution lines. On more than
the algorithm to achieve a more accurate estimation of the faulted section. Zhu
et al., proposed a hybrid method that computed the fault distance using
measurements available at the substation. He used post-fault values of current or
voltage to reduce the multiple estimation problems induced by the existence of
multiple fault points in the network with the same impedance.

To identify the actual fault location, a fault diagnosis procedure was applied to
rank the list of multiple potential fault locations. By doing a circuit simulation,
the operation of a particular combination of protective devices and the load
change pattern during different fault scenarios could be obtained. Then by
matching the fault situation to these scenarios, the actual faulted section could
be determined. However, this diagnosis procedure modelled the circuit and
simulated different fault scenarios, which were time-consuming tasks, and the
modelling process needed to be done again for another system. Järventausta et
al., used the fault distance calculation, fault detector information, and
geographical information of the distribution system to locate faults
.Zhong et al., presented a method to locate faults based on fault current
measurements, fault currents calculated from short circuit analysis, and system
operators experience. Lee et al., calculated the fault distance first to provide
some fault location candidates. Then, by current pattern matching and
interrupted load analysis, the candidate pool was reduced. Khosravi developed a
framework for fault detection and modelling when uncertainty in the plant was
present. In the proposed method, a fault alarm was fired when an inconsistency
between the behaviours of the system and the model emerges. Afterwards, the
behaviour of the faulty system is modelled using an Adaptive Neuro Fuzzy
Inference System (ANFIS). The identified model can be used for the fault
accommodation task.
Mora et al., proposed a fault location approach based on the current waveforms
measured at the power substation and the knowledge of protective device
setting and ANFIS nets. The ANFIS nets proved the capability to locate the
fault in a specific zone of power distribution system and showed validation
errors lower than 1% to locate faulted zones. This approach did not use the
electrical model of the network and the signal treatment to obtain descriptors is
simple. The fact of not using the electric al network model was an additional
advantage because it was not always easy to have the electrical power system
parameter values. Fuzzy inference is used in both methods (Khosravi, Mora et
al.,) to deal with uncertainty inherent in these methods. Fan (Chunju et al.,)
proposed a fault location method employing wavelet fuzzy neural network to
use post-fault transient and steady-state measurements. However, this method
was not influenced by the fault resistance and load current but he assumed the
system was balanced and it tacked a lot of time in off line mode the training and
calculation for the fault location in industrial distribution lines.

Automated distribution system

Today, in distribution system planning, increasing attention is being paid to

improve customer service reliability. In a distribution system large number of
equipment are installed. Power supply covers industrial, commercial, residential
and agricultural consumers. Hence, considerable amount of labour and expense
are needed for monitoring and operation of distribution network. Several studies
on utility experience suggest that customer satisfaction is closely related with
frequency of service interruption and duration of service interruption. On the
other hand, utilities are also facing some difficulties due to which they are not
able to fulfill the requirement of consumers. Distribution automation (DA) is a
tool used to benefit the consumer as well as utility. DA refers to automation of
repetitive tasks on the electric utility distribution system. In DA system,
normally closed sectionalizing switches (SSs) and normally open route switches
(RSs) are used to automate the distribution feeder functions. These switches
restrict the extent of disruption caused by long power interruption when
properly positioned. Current transformers and voltage transformers are mounted
in each section. The status of load break switches, SS, RS and RMS values and
phase angle of section current and voltage are made available from pole top
remote terminal unit to substation remote terminal unit (RTU). Availability of
this information at substation RTU greatly helps in location of faults in
distribution system. Proposed symbolic method is developed using above
information. The study network is an 11 kV distribution network with
automation facility. It has four outgoing feeders as shown in Fig. are four links
connected through route switches (normally opened).
 Fig. 2 Substation under study.

Fig. 2 Automated distribution system network

The statistics of substation is collected. Based upon the actual data analysis,
automation benefits are extracted. The complete analysis is carried out on these
substation feeders. Number of sectionalizing switches is determined so as to
minimize the cost of outage. Sectionalizing switches and route switches are
remotely operated switches, the status of which can be continuously monitored
by sub-station personal computer. Fig. 2 automated system diagram of four
feeders with above facility. It has total 15 nodes and 14 branches. At each node
there is facility to open/close the feeder for reconfiguration as well as supply
restoration. Links are available between nodes 3–11, 4–15, 6–8 and 11–14. Bus
number-1 is 11 kV substation bus assumed as a reference bus.

Symbolic method
The voltage and current signals include large harmonics and dc offset during
the fault period. Discrete Fourier transform (DFT) is an excellent filter
 algorithm that can extinguish harmonics using simple computation.
Estimation of fundamental frequency component is an essential part of
proposed method. Consider a period T and continuous sinusoidal signal x(t).
The DFT is used to evaluate the Fourier co-efficient from N samples of x(t)
taken at times t = 0, Ts, 2Ts, . . ., (N − 1)Ts, where Ts= T/N is the sampling
interval. The complex valued data sequence is obtained by sampling at a finite
number of frequency points. Therefore, the input to the DFT is a sequence of
samples rather than a continuous function of time. The DFT of data sequence
xm, m = 0, 1, 2, . .., (N − 1) is defined as

1
1 2
f e (1)
0

Where Cf is the complex Fourier coefficient. Eq. (1) can be written as

1 2 2
f cos sin (2)
0
or
1
1 1 2 2
( fr fm ) cos sin (3)
2
0

Therefore
1
2 2
fr cos (4)
0

Eqs. (4) and (5) call full-cycle discrete Fourier transform where Cfrand
Cfmare the real and imaginary part of amplitude of fundamental frequency
component of a phasor.

1
2 2
fm sin (5)
0
 So the amplitude C and phase angle φ of fundamental frequency
component of a phasor are given by Eqs. (6) and (7):
2 2 (6)
fr fm

1 fm
tan (7)
fr

As mentioned earlier analog current and voltage signals are made available at
substation RTU. Based upon the samples collected from CTs and PTs the rms
value of voltage and current is determined from above technique. High
frequency components are filtered out from the distorted waveform. Only
fundamental component is considered for further decision-making process.

Short circuit faults

Comparing the magnitude of the current signals after fault in all the sections,
immediately the zone in which fault occurs is identified. In the feeder having no
laterals, it is very easy to locate the fault. The comparison of magnitude of
current in each section shows that the fault is in that section in which starting
bus CT have fault current and ending bus CT have normal current. This
particular condition is read as symbols like digits 1 and 0. This status is
monitored continuously for each section CT. Under normal conditions, when
there is no fault on the system, the status is normal load current read as 0. After
fault the sections of faulted current towards substation will have high fault
current read as 1 and towards far end have normal current read as 0. So the fault
is detected in that section where this status is 1 and 0. Rest of the section has
status either 1 and 1 or 0 and 0. This is illustrated as follows. Consider the
feeder-3, shown in Fig. 2, having sections 1–10, 10–11, 11–12 and 12–13. Let
us assume that the fault occurs in between sections 11 and 12. The status is
determined from each section CT on this feeder. Polling for magnitude of
current is started from substation end CT. Results of this polling is 1, 1, 1, 0 and
0 means fault is in between buses 11 and 12. However, the problem of fault
location in the feeder having laterals is more challenging. Consider the feeder-1
having sections on main feeder as 1–2, 2–5, 5–6 and sections on lateral as 2–3,
3–4. Let us assume that the fault occurs in between sections 2 and 3. The CT
placed on bus 2 is measuring sum of load current at buses 3, 4, 5 and 6. The
status detected by above logic is 1 and 0 for sections 2–3 and 2–5 both. Actually
the fault is in between buses 2 and 3. This problem is solved by updating current
and voltage vectors using static impedance load model [5] as follows.

The data provided by a load flow program can be used to account for laterals
and voltage magnitude at a load bus can be used to obtain equivalent load
admittance. The series impedance for each line section is calculated from
knowledge of construction configuration and is then used to determine the
voltage drop of line section. The voltage and current phasor at each bus are
updated using the equations summarized below. This is illustrated as follows.
Referring to sections 2–3 and 2–5, the load current at the bus may be described
as

[ load ] [ load ][ ] (8)

where I loadi is the load current vector at bus I, Vi the voltage vector at bus i,
and Yloadi is the matrix representing the load admittance at bus i. From the
network topology starting bus, ending bus, impedance of line section and real
and reactive loading of the buses is known. The status of voltage read by PTs
shows that the lateral ending bus voltage collapsed due to fault on that section.
Main feeder voltage is normal. After this first guess for fault location, the
voltage drop in both the section is estimated by following set of equations. The
load current at ending bus does not change while starting end bus current is
updated by adding fault current. KCL at every node is givenby Eq.(9):

[ ]T[ b ] [ load] (9)

In order to accurately calculate the load current at each bus, the updated bus
voltage must be known. The updated voltage vectors are calculated as follows:
[ ] [ line] [ d] (10)
where Vline is the voltage measured at the line terminal, and Vd is the total
voltage drop in the line section up to bus i. The equation for voltage drop in any
zone may be stated as
[ d] [
b][ b ] [ b] [ ][ load] (11)
where Zb is the impedance of branch, and Ib is the current flow in the branch.
The faulted feeder lateral voltage drop is more. Noiterative steps are needed to
solve above set of equation.
Fault location test
The laboratory model of the automated distribution system is developed. It is
rated on the base of 250V and 5A. Normal working load current is kept below
one ampere so as to accommodate fault current within the capacity of the model
components. Two types of signals namely analog input and digital input are
taken into the personal computer through data acquisition card PCL208 and
8255 pc add-on card. These cards are mounted on Pentium motherboard. The
status of SS, RS and load break switches is known by reading

Fig. 3. Results for short circuit fault on feeder-3 between sections 2 and 3.
digital inputs. The signals coming from CTs and PTs form the analog inputs.
Samples from 15 CTs and PTs are read in personal computer. Feeder-1 is tested
for fault in between sections 2 and 3. The waveforms obtained experimentally
are shown in Fig. 3(a)–(e). These figures titled CT1, CT2, CT3, CT4, CT5
shows the current flowing through zones 1–2, 2–3, 3–4, 2–5, and 5–6. The
laboratory test set is initially operated on normal load condition. After 50ms

Fig. 4. EMTP simulation results

fault is created on sections 2–3. The current read before 50 ms is the normal
load current and as soon as fault occurs the magnitude of current in the CTs
located towards substation from fault point increase. It is seen from waveform
that symbols read from CT1 to CT5 is 1, 1, 0, 0, and 0. This feeder have lateral
also. The fault is suspected in zones 2–3 and 2–5. After estimating the voltage
drop in each section fault is confirmed in zone. Analogues results are also
obtained by simulating this feeder using EMTP and are shown in Fig. 4(a)–(e).
Similar test was carried out for open conductor fault on same feeder. It is
observed that Systems Research 75 (2005) 51–55 55 Fig. 4. EMTP simulation
results. Current flowing through zone towards extreme end of feeder is zero
while rest of the zone has normal load current.

Conclusion

This paper has presented an overview of fault location on distribution power

system and summarised various locating fault strategies. Fault locating
strategies using impedance based methods, travelling wave based methods and
knowledge-based methods have been reviewed. Most of the fault location
techniques discussed have some limitations. Some of them are as follows:
a) The iterative fault location algorithm are generally time consuming and
always has the constant risk of running into a diverging solution.
b) Heuristic procedure may take large number of trials and also time in practical
distribution systems before identifying the fault location and restoring the power
supply to healthy part of system.
c) Almost it requires voltage and current measurements from all the nodes and
branches in order to detect the fault location.
According to the advantages and disadvantages described for each method we
can conclude that in comparison between all methods, knowledge based method
seemed to have more accuracy and speed and less cost. Among the different
knowledge based algorithms, artificial intelligence method such as ANN
algorithm are more used with accordance to the success progress in AI in recent
years .When a fault occurs in distribution network power supply to the whole
feeder is interrupted. The proposed technique, use of symbolic method for
detection of short circuit fault and open conductor fault in automated
distribution system correctly identifies the section in which fault occurs. Lateral
faults are also identified effectively. Application of distribution automation
equipment minimizes the area in dark and hence maximum consumers are
benefited. Detection of high impedance gives personal safety. Discrete Fourier
Transform extracts the fundamental frequency for decision-making in location
of faults.