02-08-2010, 05:04 PM
[attachment=4061]
PRESENTED BY
P.RAMAMOHAN T.MUNI PRAKASH
RAJIV GANDHI MEMORIAL COLLEGE OF
ENGINEERING AND TECHNOLOGY
.
NANDYAL .
KURNOOL (DIST).
ABSTRACT
In this paper is presented a new method of AUTOMATED FAULT LOCATION
SYSTEM in PRIMARY DISTRIBUTION NETWORKS based on measurements
provided by intelligent electronic devices (IEDs) with built in oscillography functions,
installed only at the substation level .This system can identify the most probable fault
locations in a fast and accurate way using a database that stores information about the
network topology and its electrical parameters.
The method is based on an algorithm which combines the information about
both the fault type and load rejection for better estimation of fault location .This
algorithm is robust in nature and is also least influenced by errors in load distribution.
This is because of the fact that magnitude of load current is much smaller than that of
the fault current.
This procedure excels the traditional ones which assume that measurements
from local terminals of the faulted distribution lines are available. It is a powerful tool
to operate and maintain the distribution network efficiently and locate the faults ver y
far from the substation.
When the fault resistance is greater, the error in estimation of the fault rises to a
higher level. A part of my contribution : a novel approach is proposed to minimize the
error in such cases. The approach increases the accuracy of estimation and also
improves the stability of the system.
INTRODUCTION :-
A Power system is said to be well designed if it gives a good quality and
reliable supply. Good quality means maintaining flat voltage profile. Reliable supply
means having stability. The reliability of a system is affected because of short circuit
fault occurrences which represent one of the most extreme conditions of the power
systems. The causes of the faults must be determined and eliminated. The process can
be expedited only if f ault location can be determined quickly and reliably. Since the
deregulation process has started , most of the research on power distribution systems
is focused on delivering power in an efficient way , i.e. ,in terms of quality , reliability
and end user price.
PROBLEM DEFINED:-
An accurate estimation of fault location in power systems presents a
challenging task. Several authors have researched on this topic and have proposed
valuable methods. Most of the proposed methods assume that measurements from the
local terminals of faulted line are available. As most of the power systems are
equipped by sparsely located instrumentation, the number of measured quantities is
rather small. The problem is compounded by the presence of measuring noise.
Despite the advances and wide spread use of intelligent electronic devices and
need for performance improvements, the fault location process employed now-a-days
1
is still based on trouble calls from the affected customers. Besides, when a transient
fault occurs, the operation centre will not receive any phone call and prevention of
these transient faults from becoming permanent ones is difficult.
FAULT LOCATION ALGORITHMS IN VOGUE :-
Most of the proposed fault location algorithms were developed for power
transmission systems. A few methods were proposed for distribution networks due to
the following reasons:-
i. Variety of conductors and structures
ii. Lateral branches
iii. Load distributed along the feeder
iv. Modifications in feeder configurations
Most of the proposed algorithms are based on fault detectors installed along the
feeders and can only be applied to balanced networks. As a result, these methods will
estimate a fault location far from the actual one.
AUTOMATED FAULT LOCATION SYSTEM :-
The automated fault location system proposed in this paper combines
information provided by IEDs located only at the substations with knowledge of the
distribution systemâ„¢s topology and its electrical parameters. Whenever an over current
event occurs, the system automatically provides the most probable fault locations to
the operators, at the operation center .
Let us examine the substation ar rangement depicted in fig.1. An IED, connected
to each medium-voltage feeder, is responsible for monitoring voltage and current
signals .It also records transient data whenever an over current event occurs.
A computer, located at the substation, is connected to the array of IEDs and to
the computer at the operation centre via a communication channel such as dial phone
line, dedicated line, radio link, and tie to corporate local area network (LAN).
This fault location system consists of eight software modules. These software
modularities enable future software upgrades with out the need for changing the
whole system. A part of these modules is installed at the substations whereas the rest
are installed at the operation centre as illustrated in fig.2. The decision of installing
2
the fault location modules at the substations and at the operation centre was based on
the following reasons.
D
U
:
The fault location system uses information stored in
ATABASE
PDATE
database, which must be periodically updated in order to reflect possible
modifications in the distribution networks. Therefore, this database and
software modules that have access to it should be located at the operation
centre computers.
C
C
O
:
Transferring all transient data
COMMUNICATION
CHANNEL
OVERFLOW
recorded by the IEDs to operation centre would probably overflow the
communication channel that connects the substationsâ„¢ computers to the
operation centre computer. Therefore it is convenient to preprocess the
transient data and extract only the information located by the fault location
system, which is transmitted to the operation centre.
S
S
M
:
SUBSTATION
SOFTWARE
MODULES
M
M
:
Responsible for scheduling and storing data acquisition . It
AIN
ODULE
also coordinates and monitors all data transfers among other modules.
IED I
M
:
Converts the transient data recorded by the IEDs
NTERFACE
ODULE
into COMTRADE format. This format is desirable to make the fault location
system independent of the recording equipment in such a way that future IEDs
changes will not imply in major software upgrades. The information sets
included in the COMTRADE file are as follows:
i. Detailed oscillography of the fault:
phase voltages and line currents;
ii.
Additional measurements:
Prefault and post fault measurements (active and
reactive powers, voltages and currents),circuit-breaker operation time, etc.
D
S
P
(DSP) M
:
Preprocesses the transient
IGITAL
IGNAL
ROCESSING
ODULE
data stored in COMTRADE file. This software module performs the tasks
described below and stores the results in an ASCII file, which is less than 1
KB and much smaller than the COMTRADE file.
Determination of the fault occurrence instant :
The algorithm
3
developed to perform this task is based on digital signal processing
techniques, which can identif y signal transition instants .
Estimation of pre-fault and f ault phasor quantities :-
The DSP
module places two data windows(one at the present fault region and
other at the fault region) after the fault occurrence instant is determined
and estimates the phasor quantities using the discrete fourier transform
method (DFT).
Fault type and phases involved
:- The algorithm developed to perform
this task is based on the analysis of the super imposed sequence
components of the currents. The DSP module compares the
magnitudes and phases of the positive, negative and zero sequence
components to determine whether the fault is single line to ground
(AN, BN, or CN), line-line faults (AB, BC, or CA) double line to
ground (ABN, BCN, or CAN), or three phase.
Estimation of load rejection
: The DSP module estimates the pre fault
and post fault active power to determine the amount of load rejection.
This information is used to classify the fault.
Fault classification
: Using the information about the circuit breaker
status, after the fault clearance , and the amount of load rejection , the
algorithm can classify the fault as :
Permanent faults isolated by breaker operation
: When the
circuit breaker remains closed after the over current event
Permanent faults isolated by the fuse operation:
When the
circuit breaker remains closed after the overcurrent event and
there is load rejection (the prefault active power is bigger than
the active power measured after the fault clearance).
Transient faults
: When the circuit breaker remains closed after
the event and there is no load rejection.
C
(COMM)
:
Responsible for automatically sending
OMMUNICATION
MODULE
the ASCII file produced by the DSP module to the operation centre.
O
C
M
:
PERATION
ENTRE
ODULES
M
:
Responsible for scheduling and storing data acquisition data
AIN MODULE
transfer among the other modules.
COMM M
:
Responsible for receiving the data sent by the sub station
ODULE
computer and scheduling its processing priority.
WEB B
I
:
Provides graphical results of the fault location
ASED
NTERFACE
procedure. The decision of using this kind of Interface was based on its
flexibility and widespread use of web based tools. Besides, the results
4
provided by the system can not only be accessed f rom computers connected to
the utilityâ„¢s intranet, but also from the computers at the operation center.
Fault location module
:
Performs the fault location procedure, based on the
algorithm detailed in the next item, and provides the results to the web based
interface.
D
ATABASE
The fault location system has access to the database that stores the information
about the topology and electrical parameters of the feeders. This data base is
obtained and periodically updated from the electricity utilityâ„¢s corporate database
and contains the following information.
T
:
The distribution feeders are described using the universal
OPOLOGY
transverse mercator coor dinates (UTM) . Therefore, it is possible to integrate
information provided by the fault location system to any geographic
information systems ( GIS) system ;
E
:
Cable types and feeder geometry such as overhead,
LECTRICAL PARAMETERS
spacer, twisted, and underground, are used to calculate the line impedances,
nominal power of the distribution transformers and connection schemes, etc.
FAULT LOCATION ALGORITHM
Primary distribution feeders are radial networks with several lateral branches.
This means that faults at different locations may result in the same voltage and current
signals recorded at the substation. Therefore, the algorithm should investigate all line
sections in order to determine the possible fault locations.
Consider the feeder illustrated in figure
3, where a starting and ending node identify
each line section. The procedure adopted to
determine whether the
k
th line section,
generically delimited by nodes
p
and
q
, has a
possible fault location, consists of estimating
the fault distance(D) from node
p ,
where a
fault would produce the same voltage and
current signals recorded at the substation. If the estimated distance is less than the
sectionâ„¢s length (
L
), the
k
th line section has a possible fault location.
K
The methodology used to estimate the fault distance is described later. It is
based on a set of equations that depends on the fault type, and on the following phasor
quantities:
5
V
I
p
pq
a
a
V
=
and I
=
P
pq
(1)
V
I
p
pq
b
b
V
I
p
pq
c
c
V
I
are the phasor quantities at node
p
, during the fault, and the
Where
p and
pq
currents at line section
k
,during the fault.
Assuming that the fault current (I
fault
) and the load currents during the fault ( I
j
,
j =1 to n) are known, V
and I
can be calculated using a three phase power flow
P
pq
algorithm.
Since it is impossible to correctly calculate the load currents during the fault using
data available only at the substation, a procedure was developed to estimate them.
This procedure consists of estimating the pre-fault complex power at each node , then
using it to calculate the pre-fault voltage and current phasor quantities(using a three
phase power flow algorithm), and finally calculating the complex power at each node
during the fault by means of modeling its behaviour according to the voltage
variation.
Symbols used in the block diagram
and in following equations are:
Ëœmâ„¢ total number of nodes at the feeder
ËœS
â„¢ Prefault complex power at the
ps
sbstn.
ËœS
â„¢ Nominal power of transformers
j n
connected to node j
ËœS
â„¢ prefault complex power at node j
pj
ËœS
â„¢ complex power at node j during
j
the fault
ËœV
â„¢ pre-fault voltage at node j
pj
ËœV
â„¢ voltage at node j during the fault
j
ËœI
â„¢ load current at node j during fault
j
ËœV
â„¢ voltage at sbstn. during fault
s
ËœI
â„¢ current at sbstn. during fault
s
ËœI
â„¢ fault current phasor
m
Ëœnâ„¢ load model used by the algorithm
Ëœz
â„¢ line and mutual impedances( )
/
km
i j
Fault location algorithm logic diagram
6
Fault Types
The fault location algorithm is recursive, which means that a new fault
distance D is calculated at each step .With this distance, the voltage and current
phasor quantities during the fault are recalculated (considering the fault to have
occurred at the new distance). The block diagram depicted in fig.4, presents the
procedure developed to locate the possible fault points.
E
O
T
P
C
P
STIMATION
F
HE
REFAULT
OMPLEX
OWER
The first methodology used to determine the prefault complex power at each
node consists of aggregating the representative load curves of all consumers
connected to each node of the feeder. Preliminary tests indicate that minor errors in
the load estimation have not considerably affected the algorithmâ„¢s accuracy and, since
this methodology uses a large amount of data to determine the complex power at each
bus, a simplified methodology was developed.
It consists of assuming that all distribution transformers connected to the
feeder operate proportionally to their nominal apparent power. Consequently, the total
complex power measured at the substation is distributed at each node according to
S
their nominal power as in S
= S
x
(2)
t
pj
ps
m
S
t
j
1
P
P
F
REFAULT
OWER
LOW
The prefault load currents and voltages at each node are estimated using a
three phase power flow .The implemented power flow algorithm explains the fact that
distribution networks are almost always radial .Therefore, it does not use the
admittance matrix.
I
A
L
S
:
NVESTIGATION OF
LL
INE
ECTIONS
After estimating the prefault voltage and current phasor quantities, the
algorithm starts investigating all line sections in order to determine all possible fault
points. First, the algorithm assumes that the load currents and node voltages, during
the fault are equal to the prefault ones, and the f ault distances are zero (i.e., the fault
occurred at the beginning of the investigated line section).
E
STIMATION OF THE FAULT CURRENT
The fault cur rent is estimated by subtracting the fault current phasor quantities
(measured at the substation) from the load currents at the feeder, during the fault, as in
m
I
= I
-
I
(3)
f
s
j
j
1
7
P
OWER FLOW DURING THE FAULT
In this item the load currents are calculated using the following procedure.
First the complex load power at each node during every fault is calculated assuming
that the loads connected to each node depend on voltage.
n
V
S
= S
(4)
j
j
pj
V
pj
Then the load currents at each node can be calculated using the complex
power previously calculated, and the voltage phasor quantities at each node are
calculated.
S
I
=
(5)
j
j
V
j
n
2
V
From 4 and 5, I
= (S
) *
* V
(6)
j
j
pj
j
n
V
pj
Once the load currents during the fault, are known, the voltages at each node
can be calculated.
D
:
ISTANCE CALCULATION
The algorithm presented in this paper is based on solution of (7) that describes
the fault condition.Fig.5 and table 1 illustrate all possible fault types occurring in the
line section delimited by nodes p and q.
TABLE I
RESISTANCE VALUES FOR ALL FAULT TYPES
Fault Type R
R
R
R
af
bf
cf
f
AN 0
Unknown
BN
0
Unknown
CN
0 Unknown
AB Unknown Unknown
BC
Unknown Unknown
CA Unknown
Unknown
ABN Unknown Unknown
Unknown
BCN
Unknown Unknown Unknown
CAN Unknown
Unknown Unknown
ABC Unknown Unknown Unknown
ABCN Unknown Unknown Unknown Unknown
Z
Z
Z
V
I
R
R
R
R
I
p
pq
f
f
+
aa
ab
ac
a
a
a
f
f
f
a
= D .
Z
Z
Z
.
x
(7)
V
I
R
R
R
R
I
p
pq
f
f
ba
bb
bc
b
b
f
b
f
f
b
Z
Z
Z
V
I
R
R
R
R
I
p
pq
f
f
ca
cb
cc
c
c
f
f
c
f
c
Based on fault type and phases involved equations are written. Using (7) as
example consider ABN fault now equation (7) can be written resulting in 8 and 9 as
+
+
follows
= D .
Z
.
I
.
(
+
) (8)
V
pq
R
I
R
I
I
p
f
f
f
f
ai
i
a
a
a
f
a
b
i
a
,
b
,
c
8
+
+
= D .
Z
.
I
.
(
+
) (9)
pq
V
R
I
R
I
I
p
f
f
f
f
bi
i
b
b
b
f
a
b
i
a
,
b
,
c
The fault distance (D) and fault resistances can be calculated by separating 9
and 8 into real and imaginary parts and then solving the resulting linear system
R
A
P
F
L
:
ANKING
LL
OSSIBLE
AULT
OCATIONS
The fault location algorithm provides more than one possible fault point. In
order to exactly estimate fault location, the algorithm combines information about the
fault type and load rejection (provided by the DSP module) with the feederâ„¢s topology
and electrical parameter s (stored in data base) to rank them by the most probable
ones. Basically, the algorithm follows the rules described below:
i> When the fault is permanent, isolated by breaker operation, the algorithm
verifies which located points are protected by fuses and ranks them as the less
probable ones.
ii> When the fault is permanent isolated by fuse operation, the algorithm verifies
which located points are not protected by fuses and ranks them as the less probable
ones. Among those who are protected by fuses the algorithm ranks as the most
probable one the point that is protected by a fuse whose opening would cause an
amount of load rejection comparable to the load rejection measured by the IEDs.
INFLUENCE OF LOAD DISTRIBUTION
The fault location algorithm estimates the load currents by proportionally
distributing the total apparent power measured at the substation to each node,
according to their nominal apparent power. However, the load distribution may not be
proportional to nominal installed load. The error caused due to this estimation does
not influence significantly the algorithmâ„¢s accuracy due to the fact that in most cases
the magnitude of load currents is much smaller than that of the fault current.
ADDITIONAL SIMPLIFICATION TO THE GIVEN ALGORITHM
An additional simplification reducing the algorithmâ„¢s computational burden
consists of replacing the power “flow algorithm by a simple equation removing the
iterative process. This is done by considering the voltage phasor quantities at each
node, during the fault, equal to the voltage phasor quantities measured at the
substation also during the fault. The error level is based on the fact that, the bigger the
fault resistance becomes, the smaller becomes the magnitude of the fault current.
When the magnitude of fault current is comparable to that of the load current , the
error level increases greatly. However, large fault resistances (>20 ohms) will not
trigger the digital meter oscillography function resulting in large errors while
estimating the fault.
M
C
T
T
P
:
Y
ONTRIBUTION
O
HE
APER
As mentioned in this presentation, when the fault resistance is greater than 20
ohms, the error in the estimation of fault rises to a greater value. As a part of my
contribution, a novel approach is proposed to minimize the error in such cases.
Considering the operation of the circuit shown in fig. 6, when a fault occurs, the
pulse transformer through the rectifiers forces the thyristors to turn ON(due to high
fault cur rents).Therefore, a shunt path across the load with effective resistance (<20
ohms) is obtained. This decreases the error in estimation. As it can be observed, the
circuit is activated only when a fault occurs.
9
(1) Pulse Transformer
(2) Rectifiers with Filters
(3) Thyristor Controller
(4) Shunt Resistance (< 1ohm)
(5) Fault Resistance
Fig. 6
T
A
O
T
A
A
:-
HE
DVANTAGES
F
HIS
PPROACH
RE
i. Accurate estimation of fault by considerable minimization of error. This accurate
estimation helps especially in the case of underground cables.
ii. When a fault occurs, a considerable part of the fault current is shunted through the
circuit and hence the effect of fault on other loads is reduced. Consequently, stability
of the system increases.
CONCLUSION
The most important benefits provided by the fault location system for primary
distribution networks are as follows:
Downtime reduction: Decrease in time spent by maintenance crews to locate
faults.
Operation costs get reduced.
No. of maintenance crews on standby is reduced.
Higher profits and increase in electricity supply
Consumer satisfaction due to faster system restoration
Identification of, transient faults which do not cause permanent breaker
Operation or fuse blowing.
Identification of areas with high no. of transient f aults resulting in reduced
maintenance.
BIBLOGRAPHY
Eduardo Cesar Senger ,Giovanni Manassero, Clovis Goldemberg ,Automated
fault location system for primary distribution networks, IEEE Trans.Power
Del., vol20,No.2,April2005.
Power system analysis by Hadi Sadat
Computer analysis of power systems by J.Arrilage and C.P.Arnold.
10
