See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/229008877

Fault Location Techniques for Distribution Systems

Conference Paper · May 2003

CITATIONS READS
2 2,001

4 authors, including:

Ratan Das Pekka Verho

GE Energy Consulting Tampere University
43 PUBLICATIONS   760 CITATIONS    60 PUBLICATIONS   572 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Centralized Protection and Control within a Substation View project

All content following this page was uploaded by Ratan Das on 29 April 2014.

The user has requested enhancement of the downloaded file.

CIRED 17th International Conference on Electricity Distribution Barcelona, 12-15 May 2003

FAULT LOCATION TECHNIQUES FOR DISTRIBUTION SYSTEMS

Ratan DAS Murari Mohan SAHA Pekka VERHO Damir NOVOSEL

ABB-U.S.A. ABB-Sweden ABB-Finland ABB-Switzerland
ratan.das@us.abb.com murari.saha@se.abb.com pekka.verho@fi.abb.com damir.novosel@ch.abb.com

Summary measured at one or both terminals of a line. The methods can

be divided into three categories: methods that are based on
Faulted lines must be repaired and returned to service in the traveling waves, methods that use high frequency components
shortest possible time to provide reliable service to the of currents and voltages and, methods that use the
customers. Fault locators provide estimate for both sustained fundamental frequency voltages and currents measured at the
and transient faults. Transient faults cause minor damage terminals of a line. The last method, also classified as
that is not easily visible on inspection. Fault locators help impedance-based method, consists of calculating line
identify those locations for early repairs to prevent impedances as seen from the line terminals and estimating
recurrence and consequent major damages. This paper distances of the faults. Impedance-based methods are more
reviews some of the fault location techniques for distribution popular among utilities, because of their ease of
systems, experienced in different types of networks. There can implementation. The methods in this category can be further
be two approaches for solution: local/device solution and classified into two sub-categories: methods that use
system solution. The former uses measurements available in measurements from one terminal of the transmission line [2-
one location. The later uses network modeling and topology 3] and methods that use measurements taken from both
as well as data available from local level (one or more) and terminals.
results in the most plausible fault location(s).
Methods have also been proposed in the past for estimating
INTRODUCTION fault locations on radial transmission lines [4]. These methods
when used for distribution lines are prone to errors because of
Electric power systems have grown rapidly over the past fifty non-homogeneity of lines, presence of 'laterals' and load taps.
years. This has resulted in a large increase of the number of A method using fundamental frequency voltages and currents
lines in operation and their total length. These lines has been proposed for rural distribution feeders [5]. It does
experience faults which are caused by storms, lightning, not consider dynamic nature of the loads. Performance of the
snow, freezing rain, insulation breakdown and, short circuits technique in situations where cables are used could also be an
caused by birds and other external objects. In most cases, issue.
electrical faults manifest in mechanical damage, which must
be repaired before returning the line to service. The This paper reviews selected fault location techniques
restoration can be expedited if the location of the fault is proposed for distribution systems.
either known or can be estimated with reasonable accuracy.
DMS BASED FAULT LOCATION
Fault locators provide estimate for both sustained and
transient faults. Generally, transient faults cause minor Reference [6] presents a fault location method, which has
damage that is not easily visible on inspection. Fault locators been developed as part of Distribution Management System
help identify those locations for early repairs to prevent (DMS) development. The basic idea of DMS has been to
recurrence and consequent major damages. develop a new information system based on integration of
network information system and distribution automation.
The subject of fault location has been of considerable interest From fault location point of view, the basic idea means that
to electric power utility engineers and researchers for over only the existing device and data should be used instead of
twenty years. Most of the research done till date, has been developing a method that requires more accurate network
aimed at finding the locations of transmission line faults. This modeling and special devices in the network level. This basis
is mainly because of the impact of transmission line faults on is of great importance from practical point of view because
the power systems and the time required to physically check the method can be taken into use without remarkable
the lines is much larger than the faults in the subtransmission investments.
and distribution systems. Of late, the location of faults on
subtransmission and distribution systems has started receiving The principle of distance based fault location is very simple:
some attention as many utilities are operating in a deregulated finding the similarity of calculated and measured fault
environment and are competing with each other to increase current. The database of network information system includes
the availability of power supply to the customers. the data needed for fault current analysis of the distribution
network, which originally carried out as part of off-line
The primitive method of fault location consisted of visual engineering analysis. This network modeling and fault current
inspection [1]. Methods, proposed in literature or calculation method are also part of DMS and together with
implemented in practice, for estimating the location of real time topology information it provides the basis for fault
transmission line faults consist of using voltages and currents location. The measured fault current can be obtained for

ABB_Das_A1 Session 3 Paper No 49 -1-

CIRED 17th International Conference on Electricity Distribution Barcelona, 12-15 May 2003

microprocessor-based relays, which are quite common V sf If (2)

Z meas = = mZ L1 + R f
nowadays. Even one relay with such capability is sufficient if I sf I sf
it is located in the incoming bay of the substation. As a result
the algorithm finds one or several possible faulted line
From this equation one can obtain the quadratic relation for
sections based on distance.
distance to fault [7]:
The DMS provides excellent environment for further
m 2 − mk1 + k 2 − k 3 R f = 0 (3)
processing, because the distance based fault location result is
not explicit. The information of possible fault detectors and
even the terrain and weather conditions can be taken into Vsf Z Load
where, k1 = + +1 (4)
account. In the further processing fuzzy logic is applied. An I sf Z L1 Z L1
important part of fault location is the user interface of the
V sf  Z Load 
DMS, which provides geographic view to the network and k2 =  + 1 (5)
fault location results. In addition to the fault location the I sf Z L1  Z L1 
DMS provides a whole concept of fault management, ∆I s  Z S + Z Load 
k3 =  + 1 . (6)
including for example restoration and fault reporting. I sf Z L1  Z L1 

The method has been used in several utilities for many years
Equation 3 has two unknowns m and R f . By separating this
and the experience is very good. The accuracy is very good
and by using the method the mean outage time of customers equation into real and imaginary parts, value of m can be
has been reduced significantly. The method is now part of obtained after elimination of R f .
commercial DMS product and is so far limited to short circuit
faults. The fault type is considered by including the adequate
voltages and currents. Unlike other methods using only local
METHOD OF NOVOSEL et al. data [1, 2], this method is not affected if the fault current at a
fault locator is not in phase with current at the fault. In
The method is based on the idea of fault location applied for conclusion, immunity to effects of load current and fault
short transmission lines [3], with all loads, including tapped resistance is achieved. Compensation for tapped loads enables
lines, represented by a lumped-parameter impedance model this method to provide accurate results, although, for heavily
placed behind the fault. This way of compensating for tapped tapped feeders, the accuracy may degrade toward the end of
loads is accurate as tapped load impedances are much larger the feeder.
than feeder impedance.
The method was tested on an EMTP model of a typical
ES A B
ZS mZL1 (1-m)ZL1 distribution tapped network (Duke Power, 12.47 kV, three-
~ Isf phase, 4-wire) with approximately 200 fault cases by varying
If fault locations (main feeder and taps), fault resistances and
RF ZLoad
fault inception angle. Sample results are shown in Table 1,
Vsf
where m is the calculated fault location in percentage of
feeder length.

TABLE 1 - Test results.

FIGURE 1 - Equivalent scheme of the distribution feeder.

The method is based on calculating both source and load Fault resistance Fault at 50% Fault at 90%
impedances based on pre-fault and fault voltages and currents Rf = 10 ohms m=49.8% m=88%
measured at the substation. From Figure 1, the load
impedance and the impedance behind the fault locator are: Rf = 50 ohms m=49.2% m=82%

Vp s ∆V Additional means are required to distinguish if the fault is on

Z load = − Z L1 and Zs = − s (1) the tap or on the feeder. Accuracy improvements to this
Ips ∆ Is
method will be discussed in next sections.
Vps and Ips are the pre-fault voltage and current measured at
the substation and ∆Vs=Vsf - Vps, ∆Is=Isf - Ips. To avoid TECHNIQUE OF DAS et al.
inaccurate calculation of source impedance for a small
difference between pre-fault and fault values, a negative Reference [8] suggested a technique that uses the fundamental
sequence network is used for unbalanced faults. The basic frequency voltages and currents measured at a line terminal
relation is obtained from the following equation for measured before and during the fault.
fault loop impedance: The fault location technique is described by considering a
single-phase-to-ground fault on a radial system, shown in
Figure 2. The selected system consists of an equivalent source

ABB_Das_A1 Session 3 Paper No 49 -2-

CIRED 17th International Conference on Electricity Distribution Barcelona, 12-15 May 2003

G , the line between nodes M and N and laterals. Loads are The sequence voltages and currents at node F during the
tapped at several nodes and conductors of different types are fault are estimated by assuming that all loads beyond node x
used on this circuit. The fault location technique consists of are consolidated into a single load at N , as in Figure 3.
six steps.
I xf I fx I fn In
R x F y W N
0 0 0 0 0 0
M R x −1 x F x + 1( = y) N − 1(= W) N Vf Vn
Vx L
G o o o o o o o o

LATERAL If O
Po A
D
L L L L L L L L L
O O O O O O O O O
A A A A A A A A A P.U. Distance s 1− s
D D D D D D D D D
LATERAL J
L o o oK

FIGURE 3 - Fault voltages and currents at nodes F and N .

FIGURE 2 - A radial line experiencing a fault at F .

The voltages and currents at nodes F and x are related by

A. Apparent Faulted Section
V f   1 − sB xy   Vx 
A preliminary estimate of the location of the fault is made,  =  I 
(8)
say between nodes x and x + 1(= y ) . Line parameters, the  I fx   sC xy −1   xf 

type of fault and phasors of the sequence voltages and where, s is the per unit distance to node F from node x .
currents are used to obtain this estimate.
The sequence voltages and currents at nodes N and F
B. Equivalent Radial System during the fault are related by the following equation
All laterals between node M and the apparent location of the  Vn   D e − Be   1 − (1 − s ) B xy   V f 
fault are ignored and the loads on a lateral are considered to  − I  =  C − A   − (1 − s )C  , (9)
 n   e   1   I fn 
e  xy
be present at the node to which the lateral is connected.

C. Load Modeling where, Ae , Be , Ce and De are the equivalent constants of the

cascaded sections between nodes x + 1(= y ) and N .
The effects of the loads are considered by compensating for The currents at node F are related by:
their currents. Static response type models are used for all
I fn = − I fx − I f . (10)
loads up to node x and also for a consolidated load at the
remote end. For a load at node, say R , this model is
described by The following equation is obtained by further substitutions
from Equation 8 and Equation 10, and from an equation
 n −2 n −2
involving pre-fault voltages and currents in Equation 9 and,
Y r =  G r V r p + jB r V r q (7)
 neglecting the second and higher order terms in s .
 

Vn   K m + sK n sK p   V 
where, Vr is the voltage at node R , Yr is the load I  =
1
  x  (11)
 f  K +
v sK w  K q + sK r K v + sKu   I xf 
 
admittance, Gr and Br are constants proportional to the
conductance and susceptance respectively and n p and nq are where, K' s are complex parameters and are computed using
the response constants for the active and reactive components Yn , B xy , C xy , Ae , Be , C e and De respectively. The sequence
of the load. voltages at nodes F and N and sequence currents at node F
The constants Gr and Br are estimated from the pre-fault are obtained by using Equations 8 and 11 and various
parameters.
load voltages and currents and, appropriate values of
n p and nq . These constants and voltages are used to estimate
E. Estimating the Location of the Fault
load admittances and sequence currents during the fault.
The distance to the fault node F from node x , s , expressed
as a fraction of the length from node x and node x +1(= y ) , is
estimated from the voltage-current relationships at the fault
and the resistive nature of the fault impedance. For an A-
D. Voltages and Currents at the Fault and Remote End phase-to-ground fault,

ABB_Das_A1 Session 3 Paper No 49 -3-

CIRED 17th International Conference on Electricity Distribution Barcelona, 12-15 May 2003

Vaf V0 f + V1 f + V2 f for a 5.0 ohm fault resistance, maximum error by the

= = Zf (12) proposed technique is less than 1.7%. For a 50 ohm fault
I af I 0 f + I1 f + I 2 f
resistance, the maximum error is less than 2.2%.

where, V0 f , V1 f , V2 f and I 0 f , I1 f , I 2 f are zero, positive and ALGORITHM OF SAHA et al.

negative sequence voltage and current phasors at fault F and,
Z f is the fault impedance. Reference [9] suggested an algorithm that uses the
fundamental frequency voltages and currents measured at a
line terminal before and during the fault. Current can be also
Equating imaginary parts of both sides of Equation 12, measured at a supplying transformer if only one centralized
substituting the sequence voltages and currents at fault from type of DFR is installed at the substation. A distance to fault
Equation 8 and Equation 11, neglecting the second and higher is estimated based on the topology principle. The proposed
order terms in s and rationalizing, the following equation is method is devoted for estimating the location of faults on
obtained. radial MV system, which can include many intermediate load
taps. In the method non-homogeneity of the feeder sections is
 K + sK 
Im A B  = 0, (13) also taken into account. A distribution utility MV networks
 K + sK 
 C D  are used as an example.
where, K A - K D are complex parameters. A. Calculating Fault Impedance

The complex parameters K A - K D are expressed into real and In the proposed method the calculation of fault-location
imaginary parts as K A = K AR + jK AI and so on and, substituted consists of two steps. First, the fault-loop impedance is
in Equation 13. Rationalizing the resulting equation, calculated by utilising the measured voltages and currents
neglecting higher order terms in s and rearranging the obtained before and during the fault. Second, the impedance
following equation is obtained. along the feeder is calculated by assuming the faults at each
successive section. By comparing the measured impedance
K AR KCI − K AI KCR with the calculated feeder impedance, an indication of the
s= (14) fault-location can be obtained. As far as only one-end
( KCR K BI − KCI K BR ) + ( K DR K AI − K DI K AR )
supplied radial networks are considered, the positive
sequence fault-loop impedance is calculated according to
An iterative solution of s is obtained using the pre-fault well-known equations depending on the type of fault.
admittance of the consolidated load at node N and some of
the above-mentioned equations. Measurements at Substation Level. Let us consider a radial
network, where a faulty feeder, as for example, k has the pre-
Two more estimates of the distance are obtained by fault equivalent impedance Z Lk . The remaining parallel
considering that the fault is either located between node x − 1 connected feeders are represented by an equivalent branch
and node x or is located between nodes x + 1 and x + 2 with the impedance Z L . Both Z Lk and Z L are assumed to be
respectively. Most plausible solution is selected and the
the positive sequence impedances. The aim of the analysis is
location of the fault from the relay location, node M , is
to determine the post-fault positive sequence impedance Z k
estimated.
under assumption that the equivalent impedance Z L remains
F. Converting Multiple Estimates to Single Estimate unchanged during a fault. The following equation is valid for
the pre-fault state:
The fault location technique could provide multiple estimates
if the line has 'laterals'. The number of estimates, for a fault, V pre Z L Z Lk
depends on the system configuration and the fault location. Z pre = = (15)
I pre Z L + Z Lk
Software-based fault indicators, like those commercially
available, are developed for this purpose. They detect
downstream faults irrespective of their location. Information where, V pre, I pre are pre-fault voltages and currents.
from the fault indicators is combined with multiple estimates, Two fault cases are considered:
to arrive at a single estimate for the location of a fault.
Phase-to-phase Fault. The positive sequence impedance
Test Results seen from the substation is obtained from the equation:

The fault location technique described above was tested using V pp ZLZk
simulated fault data on a 37 km long 25 kV radial circuit of Z= = (16)
I pp ZL + Zk
SaskPower. .Simulations were performed using the
PSCAD/EMTDC. The fault locations were estimated by the
proposed technique for single-phase-to-ground faults, using where, V pp - phase-phase fault-loop voltage taken at the
fault resistances of 5.0 and 50.0 ohms. Results indicate that
substation.

ABB_Das_A1 Session 3 Paper No 49 -4-

CIRED 17th International Conference on Electricity Distribution Barcelona, 12-15 May 2003

Z pi ( Z f i −1 − Z si −1 )
Combining Equation 15 and Equation 16, Z fi = = Rfi + X fi (23)
Z pi − Z f i −1 + Z si −1

Z Z pre
Zk = (17) Zs1 Zs2
Z pre − Z (1 − k zk ) 1 2 3 k

Z pre S Lk Zf1 Zp2 Zf2 Zp3 Zf3 Zpk Zfk

where, k zk = = , (18)
Z Lk SΣ
S Lk - the pre-fault power in the faulty line and
S Σ - the sum of pre-fault power in each line.
FIGURE 4 - Equivalent positive-sequence diagram of the
faulty cable.
The coefficient k zk for each line is estimated on the basis of
the pre-fault steady-state conditions. From Equation 17 one In the above equation Z si −1 represents the cable segment
can calculate the fault-loop impedance using the impedance while Z pi relates to the load impedance and/or
measurements from the substation, such as
equivalent impedance of the branches connected to the node.
V
Value of this impedance is estimated from the steady-state
Zk =
pp (19) condition of the network. One can see that impedance
V
I pp − (1 − k zk )
pp obtained in the following steps tends toward zero
Z pre

Z f i −1 > Z fi
(24)
Phase-to-ground Fault. In the case of a phase-to-ground
fault (e.g. a phase-to-ground fault), the positive sequence
fault-loop impedance is calculated in a classical way. One can
observe that as only a single phase-to-ground fault is C. EMTP/ATP Simulations and Staged Fault Tests
considered, the zero sequence current measured in the
A 10 kV substation is supplied from 150 kV system. The
substation contains the faulty feeder current I kN and zero-
network contains of rings and sub-rings, containing several
sequence current flows through capacitance of the healthy 10/0.4 kV transformer-houses. Example of the analyzed
feeders. Final expression takes the form network is presented in Figure 5.

Z g Z pre
Zk = (20) equivalent a equivalent b
V
Z pre − Z g (1 − k zk )(1 − 0 )
V ph 5 6 7

1 2 3 4
V ph 9 8
where, Z g = and (21) 10
16 17 18 19
I ph + k kN I kN 12
11
20
V 0 = (V A + V B + V C ) / 3 . (22) 14
13 21
15

equivalent c equivalent d equivalent e

The above equations define fault-loop impedance for phase-

to-ground fault in terms of positive-sequence impedance. FIGURE 5 - Feeder model representation, dotted lines are for
grounding system connection.
B. Estimating Distance to Fault
For verification of the proposed algorithm, a series of field
Based on the measured fault-loop impedance and the cable tests have been made in the considered network. The DFRs
parameters it is possible to estimate the distance to a were installed at the substation and in the faulty feeder. For
fault.Phase-to-phase Fault. Let us consider the equivalent example, a double phase fault was initiated during field tests
positive-sequence scheme of the fault-loop. The shunt at node 20 shown in Figure 5. The estimated distance to fault
elements represent loads at successive nodes while the cable was obtained at a distance 227 m from node 18 (for
impedance is represented by the series elements. Defining an measurement in the feeder) and 64 m from node 18 (for
equivalent fault-loop impedance as seen from i th node to the measurement at the substation). From the EMTP/ATP
fault point, shown in Figure 4, one obtains the following simulation the estimated distance to fault for same fault was
recursive form. obtained at a distance 266 m from node 18. The actual fault
position is 308 m from node 18.

ABB_Das_A1 Session 3 Paper No 49 -5-

 CIRED 17th International Conference on Electricity Distribution Barcelona, 12-15 May 2003

CONCLUSIONS [3] L. Eriksson, M.M. Saha and G.D. Rockefeller, 1985,

“An Accurate Fault Locator with Compensation for
Some experiences with fault location in distribution networks Apparent Reactance in the Fault Resistance from
have been discussed. Proposed methods vary, depending on Remote-end Infeed”, IEEE Transactions on Power
available measurements and network information, as well as Apparatus and Systems, PAS-104, No. 2, February,
different types of networks and applications. The fault 424-436.
location problem can be divided into sub-problems related to [4] K. Srinivasan and A. St-Jacques, 1989, “A New Fault
both network type and fault type. There can be two Location Algorithm for Radial Transmission Line with
approaches for solution: local/device solution and system Loads”, IEEE Transactions on Power Delivery, Vol. 4,
solution. The former uses measurements available in one No. 3, July, 1676-1682.
location. The later uses network modeling and topology as
well as data available from local level (one or more) and [5] A. A. Girgis, C. M. Fallon and D.L Lubkeman, 1993, “A
results in the most plausible fault location(s). Fault Location Technique for Rural Distribution
Feeders”, IEEE Transactions on Industry Applications,
As shown in this paper, by having more data and a better Vol. 29, No. 6, November/December, 1170-1175.
network model, more accurate fault locating is possible. For
example, the more elaborate methods in sections IV and V [6] P. Järventausta, P. Verho, J. Partanen, 1994, “Using
(requiring more information about the network) will be more Fuzzy Sets to Model the Uncertainty in the Fault
accurate than the method in section III. In conclusion, users Location of Distribution Feeders”, IEEE Transactions on
have a choice to select methods that best suit their needs and Power Delivery, Vol. 9, No. 2, April, 954-960.
infrastructure. [7] D. Novosel, D. Hart, Y. Hu, and J. Myllymaki, 1998,
“System for locating faults and estimating fault
REFERENCES resistance in distribution networks with tapped loads”,
US Patent No. 5,839,093 , November 17.
[1] T.W. Stringfield, D.J. Marihart and R.F. Stevens, 1957,
“Fault Location Methods for Overhead Lines”, [8] R. Das, M.S. Sachdev and T.S. Sidhu, 2000, “A Fault
Transactions of the AIEE, Part III, Power Apparatus and Locator for Radial Subtransmission and Distribution
Systems, Vol. 76, August, 518-530. Lines”, Proceedings of the IEEE Power Engineering
[2] T. Takagi, Y. Yamakoshi, M. Yamaura, R. Kondow and Society Summer Meeting, Seattle, Washington, USA,
T. Matsushima, 1982, “Development of a New Type July 16 - 20, Paper No. 0-7803-6423-6/00.
Fault locator Using the One-terminal Voltage and [9] M.M. Saha, F. Provoost and E. Rosolowski, 2001, “Fault
Current Data”, IEEE Transactions on Power Apparatus Location method for MV Cable Network”, Proceedings
and Systems, Vol. PAS-101, No 8, August, 2892-2898. of the Development in Power System Protection
Conference, Amsterdam, The Netherlands, 9-12 April,
323-326.

ABB_Das_A1 Session 3 Paper No 49 -6-

View publication stats