IEEE Power Engineering Society

i~~~ -

Power System Relaying

Committee
424 IEEE Transactions on Power Apparatus and Systems, Vol. PAS-104, No. 2, February 1985

AN ACCURATE FAULT LOCATOR WITH COMPENSATION FOR APPARENT REACTANCE

IN THE FAULT RESISTANCE RESULTING FROM REMOTE-END INFEED

Leif Eriksson, Murari Mohan Saha, Member, IEEE G. D. Rockefeller, Fellow, IEEE
ASEA AB, Vasteras, Sweden ASEA AB, Vasteras, Sweden Rockefeller Associates, Inc.
Morris Plains, New Jersey
This novel approach is accomplished by using a complete
network model, where the infeed from the network beyond
the remote end point is rigorously taken into consideration.
Abstract - A microprocessor based fault locator is
described, which uses novel compensation techniques to Pre-fault load-current samples are stored and used for
improve accuracy. It displays the distance to the fault in compensation to eliminate a substantial effect on accuracy.
percent of transmission line length, for facilitating repair Also, representative values for the source impedances are
and restoration following a permanent fault. Also, it stored to compensate for variations in impedance angles.
pinpoints weak spots following transient faults. This novel approach is described as well as the system
design and performance.
This new method for fault location on electric power
transmission lines uses recorded phase currents and voltages PURPOSE FOR FAULT LOCATORS
at the near end. The main feature of the method is that it
considers the influence of the remote-end infeed of the Even where helicopters are immediately available for
transmission line by using a complete network model. patrol following unsuccessful reclosing, fault locators
perform a valuable service. Trouble cannot always be found
with a routine patrol with no indication of where the fault
A microprocessor filters the ac currents and voltages occurred. For example, tree growth could reduce
from the protective relaying instrument transformers to clearances, resulting in a flashover during severe conductor
extract the fundamental components of the signals. It then sagging. By the time the patrol arrives, the conductors have
computes the distance to the fault point, compensating for cooled, increasing the clearance to the tree. The weak spot
the apparent reactance in the fault resistance resulting is not obvious.
from load current and the variations in impedance angles in
the power-system network. The importance of fault locators is more obvious
where foot patrols are relied upon, particularly on long
The design has undergone field tests and evaluation. lines, in rough terrain. Also, locators can help where
The outcome of the field tests is presented and has maintenance jurisdiction is divided between different
confirmed the validity of the concept of the fault locator, companies or divisions within a company.
showing an accurate display of the distance to fault.
Fault locators are valuable even where the line has
INTRODUCTION been restored either automatically or non-automatically. In
this category are faults caused by cranes swinging into the
Distance relays for transmission-line protection line, brushfires, damaged insulators and vandalism. The
provide some indication of the general area where a fault locator allows rapid arrival at the site before the evidence
occurred, but they are not designed to pinpoint the location. is removed or the "trail becomes cold"?. Also, the knowledge
Since pRower circuit breakers are only installed at the that repeat faults are occurring in the same area can be
terminals, it is immaterial where the line faulted for valuable in detecting the cause. Weak spots that are not
isolating the flashover. However, with immediate obvious may be found because a more thorough inspection
knowledge of the location, the nature (type and measuring can be focussed in the limited area defined by the fault
data) of the fault can be determined quickly, facilitating locator.
repair and restoration. A locator is also useful for transient
faults, pointing to a weak spot that is threatening further FAULT LOCATION FUNDAMENTALS
trouble.
The fault location computations determine the
Various methods were developed during the recent apparent fault impedance with novel compensation for the
years to detect the location of fault on a transmission line. fault resistance drop, eliminating the errors inherent in
They are mostly based on analog techniquest 2]. Some fault conventional reactance-type measurements. For a fault-
locator designs are in service which are reliable in detecting protection relay these errors are tolerable because a safety
permanent faults. New methods and systems of fault margin is inserted in its setting. However, for the fault
location based on both analog and digital techniques seem to locator a more precise measurement is quite desirable. The
offer good prospects to obtain a precise location of Appendix derives the equation for the apparent impedance
transient faults also. The method reported in [ii is an seen by a reactance-type measurement. Figs. Al and A2
approximate method. The method of fault location illustrate the apparent reactance effect in the fault-
technique described here seems more advantageous, since it resistance term.
takes into consideration the effects of both ends of the line.
Fig. 1 shows the connection of the fault locator at
84 SM 624-3 A paper recomnended and approved station A. Using the ac quantities available at station A, it
by the IEEE Power System Relaying Committee of the is not possible to determine the total fault current IF unless
IEEE Pow'er Engineering Society for presentation at p and ZSB are known; conventional devices tolerate the
the IEEE/PES 1984 Summer Meeting, Seattle, errors resulting from the infeed current IB flowing through
Washington, July 15 - 20, 1984. Manuscript sub- the fault resistance RF, out of phase with respect to IA.
mitted August 30, 1984; made available for print-
ing May 4, 1984. In the method reported in Reference 1, the current
distribution factors for the parts of the network, located on
either side of the fault point, are assumed to have the same
arguments, or else it is assumed that the difference between
the arguments is known and constant along the line segment
0018-9510/85/0002-0424$01.00(1985 IEEE
 425

L1 UAE (UA)L
L- ( bIL UFE
1I
L

'LD
IF
UA
77+7 (UlA), UF
UA.E UB3E
Fig. 1 Power system one-line diagram with fault.

in question. This simplified assumption can give rise to

substantial errors in the calculated fault distance. The
difference between the two arguments mentioned generally
varies with the distance to the fault.
The fault locator program described here utilizes
representative values of the source impedances to
determine a correct description of the network. The value Fig. 2 Prefault conditions.
of RF is unknown; however, it is not needed - only the
angle of IFRF, the fault point voltage, is used.
The other key aspect of this locator's algorithm is its (UlA)F PZIL (1-P)ZlL
use of pre-fault current memory to determine the change in
line current caused by the fault: the actual fault current
minus the prefault value. Eq. (A9) from the appendix,
repeated here as Eq. (1) states that the station A voltage is
the sum of the drops in the line to the fault point plus the
fault point voltage:

UlA = '1A P ZlL + I1FRF (1)

Eq. (1) is written for a 3-phase fault, using the positive
sequence notations. Eq. (2) expresses the same concept in
general terms applicable to any fault:

UA = IA P ZL + IFRF (2)
To consider load current effects, the actual voltages
and currents can be considered to be composed of the
prefault values plus the changes caused by the fault. The
appendix discusses this concept. The load currents are
generated in Fig. 2 by the angular difference between the
source voltages. Voltage UF exists at the fault point prior to
the fault and is the driving voltage in Fig. 3 producing the
changes caused by the fault.
From Fig. 3, the change in positive-sequence current
in the line at A, AI1A, is related to the total positive-
sequence current by the current distribution factor DiA:
Fig. 3 Faulted network (Thevenin equivalent).
AI1A = DlA IlF (3)
Writing a similar, but general expression: processing of the zero-sequence components to obtain rF.
Fig. 4 shows the current relations in the fault for an SLG
IFA = DAIF (4) fault:

F = I2F = IOF
where IFA is the current change produced by the fault, 3
3
=
(5)
equal to the actual fault current less the prefault current.
The expression for IFA varies with the fault type; Table I
defines these. Changes in phase currents are used except Eliminating the zero-sequence current IOF
for single-line-ground faults. The latter use the change in
faulted phase current less the zero sequence current IOA. IF = 3/2 ( IlF + 12F)
The zero-sequence current is extracted because the zero- AI'A + AI2A
sequence distribution factor DOA is not known as reliably DiA (6)
as the positive-sequence factor DiA. The 3/2 factor in
Table I provides heavier weighting to compensate for the
I-amoval of the zero-sequence current.
The distribution factors for the positive- and negative-
The IFA expression for single-line-ground (SLG) fault
a sequence currents may be assumed, with great accuracy, to
will now be derived with the objective of eliminating the be equal (i.e., DIA = D2A).
426

The complex expression of Eq. (14) contains the unknowns p

IF IF = 3/2(I1F + 12F) and RF. However, Eq. (14) can be separated into two
simultaneous equations, one real and one imaginary. By
eliminating RF, a single expression results with the single
unknown p. This is solved by the program, using the peak
r, Ip 'F1/2(11F + I2F) values and their phase position, taken' from the Fourier
analysis routine which yields the fundamental components of
the signals.
-I-. IIF +
r2F
Fig. 5 provides an alternative view of how the fault locator
determines p without requiring the scalar value of the fault-
point voltage. Fig. 5 is a triangulation problem to determine
I2F = 3F the intersection point F, knowing UA and IA by measurement
and computing aD and aL. While the computer is not
specifically using this algorithm, nevertheless the conclusions
reached using the concept of Fig. 5 are valid: viz. that the
magnitude of IFRF is not required, but only its angle aD.
i1 F = IF
3r

Intersection computed
Fig. 4 Current relations for a single-line-to-ground fault.

At station A in the line:

,s_
4
AIA =A IIA + A12A + AIOA
(7)
or AI1A+ AI2A=AIA-AIOA
(8)
IA
Neglecting pre-fault zero-sequence current:
AIA-IOA
Measured: IA; UA
AIlA+AI2A =

(9) Computed: aD; aL

Eq. (9) in (6):

~ Fig. Fault-resistance-drop compensation.
T3/2
32
I
IA IOA 5

F IA (10)
Parallel Line Cases
Let IFA =3/2 (A4lA -IOA)
(11) The locator can be finished with a modified algotithm
for a parallel-line application. The positive-sequence
The general expression of Eq. (2) can be rewritten as: network is completely described by re-defining equation 13
I as:
UA IA P ZL + jA) RF
(12) DA (1-P) (ZSA + ZSB + ZL) + ZSB (13A)

2 ZSA + 2 ZSB + ZL
From Fig. 1:
Eq. (13A) assumes an identical parallel line. Use of
DA- (1-P)ZL+ZS the equation will improve accuracy with the parallel line in
ZSA +ZL + ZSB'
(13) service.
Zero-sequence mutual coupling can be compensated
Substituting Eq. (13) in (12) and rearranging, yields: for by interconnecting two fault locators . With a fault
locator on the parallel line its residual current reading can
p- p K1 + K2 K3 RF = 0 be input to its companion locator via the local printer loop
(14) circuit; the mutual resistance and reactance are input as
additional setting parameters.
where:
ALGORITHMS

KI _ UA +1+ ZSB (15) The fault location algorithm uses the pre-fault and
K2 IAZL ZL fault currents and voltages at thei near end of the
transmission line to determine the distance to fault.
K2-UA
IAZL 'Z
ZSB
(16) Current and voltage samples are continuously measured.
Following a signal from the line protection at the instant it
initiates breaker tripping, current and voltage' samples for
K3=
IFA ZSA + ZSB (17) six cycles are frozen until the completion of the distance-
IAZL ZL +1)
to-fault computation.
 427

TABLE I QUANTITIES PROCESSED FOR

VARIOUS FAULT TYPES
Type
of
Fault UA IA IPA

RN URA IRA + KN X INA 3/2 (AIRA - I0A)

SN USA ISA + KN X INA 3/2 (AISA -IOA)

TN UTA IrA + KN X INA 3/2 (AITA - IOA)

RST
RS URA - USA IRA - ISA A IRSA
RSN

ST USA - UTA ISA - ITA AISTA

STrN
TR UTA - URA ITA - IRA AITRA
TRN

The program then determines the loop of analog data The program selects 24 samples from the two periods
(currents and voltages) on which to base the computation. spanned by SL1 to SF1 in Fig. 6, where SFI is the third
For example, the RS loop is selected if the R and S phase sample following the fault point. For the fault data, 24
selectors operate. Table I shows the normal loop selection. samples immediately following SF1 in Fig. 6 are selected.
For double-line-ground faults only, a phase-to-ground loop
may be selected instead of the normal phase-phase loop, if The pre-fault and fault currents and voltages on all
desired. Table II shows the loop options. The program next phases are filtered by Fourier analysis to yield the
determines the fault inception point within the 6 cycle data, fundamental component. The selected loop quantities are
looking at the selected loop aC quantities, choosing the then processed for the distance-to-fault determination,
current(s) first. If necessary, the voltage(s) is also while the complete set is available for printing for user
processed to find the fault inception point. analysis.
Reading 24 samples apart are compared for a The fundamental components are determined by
significant change, starting with oldest samples. The multiplying each sample by the appropriate instantaneous.
required threshold for the current change is adaptive, sine value and integrating over a full period, then repeating
depending on the prefault current level. The voltage the process by multiplying by cosine values. The result
threshold is fixed. If no change is found, a one sample provides the scalar value and argument for each ac quantity.
advance occurs and the procedure is repeated. The Fourier filtering effectively attenuates the dc offset
component and power system harmonics plus CCVT
transients and CT saturation distortion.
TABLE II LOOP SELECTION FOR DOUBLE-
The program uses the peak and angle outputs from the
LINE-TO-GROUND FAULTS filtering routine to compute the distance to the fault.

Phase LOOP Selected

Input Normal Cyclic Acyclic

RSN RS SN RN
STN ST TN TN
TRN TR RN RN

If unable to find the fault-inception point, the program

reverts to a "'slow start" algorithm, which eliminates the
load-current compensation. This would occur for a switch-
into-the-fault case or a time delay trip for an end-zone Pre-fault Relay time CB time
fault, where in such cases no prefault samples are available. Memory capacity
In both cases, there would be no load flow to compensate,
because at the time of tripping the far end of the line would
be open. Faulted phase information

When a two-phase loop is selected (e.g. R and S for an

RSN selector input), two different fault inception points will Fig. 6 Samples are selected from two periods
generally occur; the program selects the latter of the two
fault points. Fp
before and after fault inception point
428

initiating test programs to the memory and for resetting the

HARDWARE LED indicator.
Fig. 7 shows the main elements of the hardware. The Table III shows the parameters to be set by the user.
digital inputs consist of the start signal from the line See GLOSSARY OF SYMBOLS. Parameters 1 to 4 are the
protection relay breaker-trip output and the phases R, S, T line constants; 5 to 8 are the source constants. The source
(i.e., A, B, C) and ground phase selection outputs from the impedances will change with system conditions, so a
line protection or from the integral phase underimpedance representative value is selected. The inclusion of the source
and ground overcurrent phase-selector units. The start impedances in the fault location algorithm is a novel
input initiates a dista-nce-to-fault computation. approach, which allows compensation for different
impedance angles in the power system network.
Input sinals from: The parameters are set with the help of thumbwheels.
A pushbutton operation initiates an EPROM (electrically
Line protection: programmable read-only memory, 22 Kbytes used) burn-in.
Measuring transformers: The digits of parameter No. 9 determine four different
Phase items listed in Table II.
Currents Voltages
Start selection
TABLE m: SETTING & PROGRAMMING PARAMETERS

1. RlL 7. RlSB
2. X1L 8. XlSB
3. ROL 9. - Type of phase selection
(Normal, cyclic, acyclic)
4. XOL - CT polarity
- Printout (A,B,C)
5. RlSA - Line number
6. XlSA

DESIGN
Fig. 8 shows the fault locator assembly with phase
selector and printer, which is suitable for 19" rack
mounting. The assembly consists of test unit, power supply
Fig. 7 Hardware configuration. unit, transformer unit and measuring shunt unit, which are
directly mounted to and held together by two apparatus
bars. The remaining plug-in units are connected to a
mother-board, fed from the shunt unit. The bottom half of
The 3-phase currents and voltages enter through input the assembly consists of the fault-type selection
transformers which provide galvanic isolation from the underimpedance and ground overcurrent units and the
instrument transformers, as well as transforming the signals printer.
to a level suitable to the electronics. A screen between the
windings minimizes common-mode surge coupling. The The result of the distance-to-fault calculation is
analog signals feed through low-pass filter for signal shown on the LED indicator. Fig. 9 shows an example of
conditioning, using a 500 Hz cut-off frequency. The filter output results from the printer. The relative distance to
outputs are switched in sequence by the multiplexer and fed fault is expressed in percentage of the line length. "Phase =
into the hold circuit in preparation for conversion to a RN" indicates A and ground inputs from the phase selectors
digital value proportional to the instantaneous value of the (i.e., a phase R to N fault). "LOOP = RN" indicates that the
ac wave. Both the digitized signals and the relay input computer processed phase A to ground voltage and phase A
status are stored in a 6-cycle circular file in the memory and residual currents to determine the fault location. The
with the aid of the microprocessor (MC 6803 computed r.m.s. values of the fundamental component of ac
microprocessor, 8 bit, 814 ns memory cycle). signals in polar form prior to and during the fault are also
shown in Fig. 9.
The microprocessor processes the measuring values
according to the fault location algorithm and presents the DESIGN TESTS
distance to fault in percentage of the line length on the LED
indicator or remote connected equipment (printer, Model-line tests were performed on six fault locator
telemeter, etc.). The microprocessor continuously executes units at the factory under dynamic conditions. A 100 km,
a monitoring routine. If the computer fails to periodically 170 kV line was modelled using CT's and PT's or a CCVT
output an "'all's well" signal, a peripheral circuit operates an modeL The line impedances referred to 380' V were: ZJL =
alarm relay. During normal service, a pushbutton is used for 2.2 + j 17.0 and ZOL = 16.4 + j 68.0 ohms. The ohmic values
 429

Line number 1

Relative distance to
fault p = 75%
Phase = RN, Loop = RN

IR AMPL. = 002,213 A
=
ARG. = 028,4 DEG.
IS = AMPL. = 000,292 A
ARG. = 308,7 DEG
IT = AMPL. = 000,831 A
ARG. = 212,0 DEG
IRS = AMPL. = 000,519 A
ARG. = 103,1 DEG
IS0 = AMPL. = 000,514 A > Pre-fault
ARG. = 343,3
ITO = AMPL. = 000,524DEG.
A
ARG. = 223,1 DEG.
Fig. 8 Fault locator assembly with IN = AMPL. = 001,475 A
phase selector and printer. ARG. = 015,1 DEG.
UR = AMPL. = 052,157 V
ARG. = 093,0 DEG.
stated below are also referred to 380 V, the primary of the US = AMPL. = 062,611 V
ARG. = 332,9
UT = AMPL. = 063,192DEG.
system simulator. Faults were applied with varying
incidence angles. Three series of tests were conducted. V
ARG. = 210,5 DEG.
Series No. 1 tests were run with zero fault resistance URO= AMPL. = 063,510 V
ARG. = 092,0
USO = AMPL. = 063,672DEG.
and rated load current, exported power from station A; the
errors were less than 2 % of the line setting. Series No. 2 V Pre-fault
were run with 50 % load, with fault resistance variations ARG. = 331,7 DEG.
from 5 to 40 ohms; the errors were less than 4 % of the line UTO = AMPL. = 063,774 V
setting. Series No. 3 tests checked for errors resulting from ARG. = 211,9 DEG.
variations in source impedance from set values; errors were
less than 2 % of the line setting. Source A was varied from
2 to 32 ohms, B from 6 to 23 ohms. All ohmic values are on
a 1 A rated base. Fig. 9 Printer output
The locator was also run through the normal test
series applied to protective relays, including disturbance 2) A staged fault test was conducted jointly by
(surge) tests: a 4-8 kV fast transient test per SEN361503 (a Texas Power & Light Co. and Texas -New
fast rise test simulating inductive interruptions in the Mexico
control circuits) and the IEC 255-4 1 MHz, 2.5 kV test. Power Co. on the described fault locator. A fault
locator was installed at Northwest Carrollton sub as
FIELD RESULTS shown in Fig. 10.

1) Two prototype units were installed for

evaluation in Swedish utilities. No misoperations or NW TI
component failures have been experienced One of the Carrollton (Lewisville) Highlands
prototypes was installed on the 30th of June 1982, on a 138kV
76 km, 130 kV transmission line between Nybro and
Vaxjo, on the Sydkraft transmission system in southern
Sweden.
On July 2nd 1982 , a disturbance was detected on
the line between Nybro and Vaxjo. The fault locator
display showed 89 % (corresponding to 67.6 km). The
type of fault that was printed out was a single-phase Source: 34 aQ Source:
ground fault on phase T. An aerial inspection of the 0.84 aa 11.3 6.9 Z9
transmission line was made on July 6th 1982 and the 1.4 /82 /82 19.8 ZI.
point of disturbance was found in phase T at a distance
of 67.0 km from Nybro. The flashover occurred from
conductor to arcing horns, across a tower insulator Imp. in % on 100 MVA base
string. The distance to fault, calculated by the fault
locator deviated from the actual distance to fault by
less than 1 %. Load current was 238A and the
calculated fault resistance was 7.2 ohms. The Fig. 10 Staged faults were applied at TI (Lewisville).
parameter settings matched those existing in the
network at the time of the fault
The other prototype was installed on a 400 kV
There is mutual coupling over 59 % of the
Northwest Carrollton - TI (Lewisville) line with the
Swedish State Power transmission line. However, no Northwest Carrollton - Highlands line. It was the
fault has occurred on that line. intention to estimate the performance of the fault
430

locator on that line without any compensation for the (a)

mutual impedance.
Five faults, two phase-to-phase and three phase-
ItuD UA;E
to-ground faults were placed at the Lewisville
substation during the early morning of October 25th
1983. The faults were applied at the TI (Lewisville)
substation where both lines loop into the substation.
The actual distance to fault was 55 % of the line. The UF
load current was about 100A. Estimated fault
resistance was 1.6 ohms for the ground faults.
l1FDlA'
The locator functioned properly for all five
faults. The highest deviation of the results between
the distance to fault shown on the fault locator display (b)
and the actual distance to fault was 3 %.
12 F- II F ) .

ACCURACY ANALYSIS 12FD1A

Results of overall accuracy during dynamic model-line 12FD1l A
tests are described under DESIGN TESTS. This section will
discuss the accuracy of this locator in comparison with
those using other approaches. IOF
O
The fault locator described above computes the IA
distance to the fault point by compensating for the apparent
reactance in the fault resistance resulting from load current [OFDOA
and variations in the impedance angles in the power system ,, Y
network. However, minor errors (see Fig. 11) can occur if ILJ D
the settings are not adequate. These errors should be
moderate for practical situations. For one, the value of the
fault locators increases with line length; for these (c)
applications the line impedance tends to mask the effect of
source impedance variations. Again for longer lines, the
fault resistance, particularly for single-phase-ground faults, IFRF
becofnes a smaller percent of the line impedance.

I UA
- I oo,)

-I---
'.rror IF RF

U i
IA
IAPZL /

IA
iF

Fig. 11: Location error resulting froin error in comnpuating

arguinent of iFRF. Fig. 12: System quantities for a single-line-ground fault with
load export from station A.

The next series of illustrations shows qualitatively the resistance. The total phase current at A is developed in Fig.
fault-resistance infeed effects on two alternative methods 12(b) and is the sum of the currents resulting from the fault
of determining fault location: the reactance and the zero- and driven by UF, and the pre-fault current ILD. Fig. 12(c)
sequence current-reference techniques. Two examples are extracts from Figs. 12(a) and (b) the essentials for
shown: one for load export; the second, import. developing the error diagrams of Figs. 13 and 14.
Fig. 12(a) shows the station A line currents out of The reactance type of Fig. 13 uses the same basis of
phase with the total fault current IF. Stating it another way, measurement as does the conventional reactance relay. The
source B produces out-of-phase infeeds in the fault angular shift of IFRF produces an error which can become
 431

IFRF- - 1 _ Error
-T (a) UF
LD)
.? V.5
;<Iq J,A

(b)
1FD1A
IA I2FPD1A

Y-IL IOFDOA
I2FD1A
Fig. 13: Reactance-type error resulting from conditions in
Fig. 12.
ILD

Fig. 13 1 IFjRFj Error

Error _ UA

IF RF

IA
UA

3IOFDOA 3IOFDOA

Fig. 14 Zero-sequence-current-reference-type
error resulting from conditions in Fig. 12.
(c)

significant for the fault location function where greater IA

precision is needed than for a protective relay.
The type shown in Fig. 14 differs from the reactance
type in its use of residual current, 31OFDOA = 3I0A as the
reference rather than using 'IA. For the example of Fig. 12,
the two currents 3IOFDOA and IA are nearly in the same
phase position, so the error in Fig. 14 is almost as much as IF
the Fig. 13 error.

In Fig. 15(a), power is being imported to Station A so Fig. 15: System quantities for a single-phase-ground fault
ILD is about 1800 from its position in Fig. 12. The fault with load import into station A.
currents here are identical to those of Fig. 12; however, the
superposition of load current results in a different
magnitude and angle for IA. Compare 16 and 17: the latter then faulted as specified in rable IV. Method 1 has zero
now has a larger error because its reference current, error because the source impedance values were input,
3IOFDOA, leads the IA position. describing the complete system. Method 4 shows substantial
errors, although smaller than both methods 2 and 3.
The load compensation method [1) will experience
errors generally less than those shown in Figs. 14 and 17, Case 3 in Table IV yields greater errors than cases 1
depending upon the amount of out-of-phase infeed into the and 2 because the zero-sequence line impedance argument
fault from the far end. This can be seen in the following of 76.7 degrees is lower with respect to the zero-sequence
numerical examples. source impedance argument. The corresponding angle for
cases 1 and 2 is 81.7 degrees. Thus, the zero-sequence out-
Single-Line Examples of-phase infeed is much greater in case 3, causing greater
errors for methods 2, 3 and 4.
Table IV shows the location errors for three cases,
comparing four computational methods, using an off-line With smaller fault resistances the errors in Table IV
computer simulatation. Table V defines the power system will be correspondingly smaller. On the other hand, lower
conditions for two different transmission lines. These were voltage lines with lower arguments will produce larger
432

Error
- - - Eror
UA
uA
UA -..;0Fig. 16
Error

Fig. 16: Reactance-type error resulting from conditiorns in F iY. 1.7 Zero-sequence-current-reference-type
Fig. 15. error resulting from conditions in Fig. 15.

TABLE IV COMPARISON OF FAULT LOCATOR ERRORS FOR THE APPLICATIONS IN TABLE V

CALCULATED RESULTS
Power Fault Fault RF Method 1 Method 2 Method 3 Method 4
Case System Locator Point Ohms p Error p Error p Error p Error
At:
1 Ashe-Marion Ashe 0.9 50 0.9 0 0.87 -3.1 0.62 -28.4 0.88 -1.9
2 Ashe-Marion Ashe 0.9 100 0.9 0 0.86 -4.5 0.54 -36.7 0.87 -2.7
3 Stenkullen- Sten- 0.8 125 0.8 0 0.72 -7.9 0.39 -41.0 0.76 -4.2
Borgvik kullen

Method 1 As described in this paper.

Method 2 Zero-sequence-current-reference type (see Fig. 14)

KN as defined in the Glossary
Method 3 Reactance type (see Fig. 13)
KN as defined in the Glossary.
Method 4 Load compensated 1].
Errors are in percentage of total length of line.

errors where the remote source argument is higher, such as 0.1-20 times rated; voltage, 0.01-1.5 rated. It is available
would be the case near a generating station. with 1, 2 and 5 A ratings, withstanding 3 times rated current
continuously.
Parallel-Line Examples
It is designed for use with the relaying instrument
Table VI shows computer-simulation results for two transformers and introduces a burden of 1 VA per phase at
parallel-line cases. In both instances the parallel line is rated voltage or current.
identical to its companion listed in Table V. The errors in
Table VI for method 1 would occur if parallel-line The unit is designed and tested just as if it were a
compensation were not included. With both the positive - protective relay.
and zero-sequence effects included in the- algorithms, the
computed errors were zero. The uncompensated errors for Operating temperature range is 0 to 550 C.
the Lieto-Forsa application are greater because of the lower
line arguments compared to the Ashe-Marion lines. The It requires 1.5 cycles of fault dturation for an accurate
parallel-line effect uipon accuracy will be the greatest for measurement.
far-end faults, because these produce . the maximum
parallel-line current. It is not necessary to apply one locator at each line
terminal. Four units can share one optional printer.
APPLICATION CONSIDERATIONS The distance to the fault can be remotely logged using
the built-in telemetering outputs from the locator.
The unit is suitable for use with lines or sources with
secondary impedances in the range of 0-1000 ohms based on
a A CT (0-200 ohms for 5 A CT's). Signal current range is
 433

TABLE V SYSTEM DESCRIPTIONS FOR TABLE VI FAULT LOCATOR ERRORS (METHOD 1)

EXAMPLES IN TABLES IV AND VI F-OR PARALLEL LINE APPLICATIONS
WITHOUT COMPENSATION
Ashe- Stenkullen- Lieto-
Marion Borgvik Forsa Fault locator installed at Ashe and Lieto.
(USA) (SWEDEN) (FINLAND)
Fault R - Ashe Lieto
Location O§ms Marion Forsa
Voltage - kV 525 400 110
Z1SA mag. 17.6 21.3 5.8
ZOM - - 332. 1 75.8
arg. 90 90 85 mag
ZOSA 'nag.
arg.
6.2
90
24.4
90
70
85
arg. - - 73.6 73.5
ZlSBJB mag. 42. 2 25.6 15 ILD mag - - 1165 240
arg. 90 90 85 arg. _ - -37.8 4.3
ZOSB 'mag.
arg.
29. 5
90
27.8
- 90
130
80
ZiL mag. 118.5 53.5 27.6 Percent Error:
arg. 87.0 86.4 73.7 0.3 50 0.1
ZOL mag. 463.0 247.6 108.7 0.3 100 0.1
arg. 81.7 76.7 74.3 0.4 4.3 0.4
UAE mag. 320.4 225.2 63.5 0.4 20 0.4
arg. -17.0 O 0 3.2
UBE mag. 334.9 240.2 63.5 0.8 50
-67.0 -29.4 -15 0.8 100 2.9
arg. 0.8 4.3 15.8
ILD mag.
arg.
1555
-37. 0
1184
-5.7
Impedances in ohms REFERENCES
Voltages in kV
Currents in A 1) T. Takagi, Y. Yamakoshi. M. Yamaura, R.
Arguments in degrees Kondow, T. Matsushima: "Development of a new
type fault locator using the one-terminal voltage
and current datat", IEEE Trans., VOL. PAS-101,
No. 8, August 1982, pp 2892-2898.
SUMMARY 2) M. Souillard, Ph. Sarquiz, L. Mouton:
"Development of measurement principles and of
1 The fault locator displays the percent the technology of protection systems and fault
distance to the fault, using novel compensation location systems for three-phase transmission
techniques for the errors resulting from remote- lines", CIGRE No. 34-02, August 1974.
end source infeed into the fault resistance. This
is accomplished by using a. complete network
modeL. GLOSSARY OF SYMBOLS
2 Compensation for fault-resistance voltage aD angle of fault resistance drop, IFRF, with reference
drop utilizes prefault current and representative to UA
values for source resistance and reactance. The
inclusion of source impedances is a novel aE angle between generated voltages UAE and UBE
approach which provides improved accurancy.
3 The microprocessor-based system is aF angle of total fault current, with reference to UF
designed and built to protective-relaying aL argument of line impedance
standards, including all surge withstand
requirements. aR angular error in computing IFRF
4 It is suitable for use with the relaying DA general expression for current distribution factor at
instrument transformers, introducing negligible station A.
burden.
DiA, D2A, DOA
5 Extensive software filtering minimizes the positive-, negative- and zero-sequence current
effects of power-system and instrument- distribution factor in line at station A
transformer transients.
IA, IB line current on faulted phase(s) at station A and B,
6 The microprocessor monitors itself and respectively. General, i.e. not referring to a
built-in functional-test facilities can pinpoint a specific symmetrical component sequence
particular chip or analog-system failure.
11A total positive-sequence current in line at station A.
7 The locator can function either with the
line-relay phase selectors or with the optional AII AA11B
built-in relay units. positive-sequence current change as a result of
fault, in line at station A and B, respectively
8 Only one end of the line needs to be
equipped. IF total fault current
434

IFA change in line current at station A, as a result of ZOSA, ZOSB zero-

fault. General - see Table I for specific definition sequence source impedance at station A and B,
respectively
depending upon fault type
IlF, I2FY IOF ZlL, ZOL positive-
positive-, negative- and zero-sequence currents in and zero-sequence line impedance
fault
ZSA, ZSB source
ILD load current in line impedance behind station A and B, respectively.
General, i.e. not referring to a specific
symmetrical component sequence
INA residual current (3 IOA) in the line at station A
ZL total line impedance. General, ie. not referring
'OA zero-sequence line current at station A to a specific symmetrical component sequence
IRA, ISA, ITA ZOM mutual impedance of line for the zero-sequence
current in line at station A, phases R, S and T, network
respectively
APPENDIX
A IRA, etc.-total current minus pre-fault current, in
the line at station A, phase R Apparent Impedance for Three Phase Fault
KN Z3L ZlL The apparent impedance seen by a conventional
reactance relay or fault locator will be derivei for a 3-
phase fault, including load current effects.
N ground, when referring to fault type
Load current flow in the line will be developed by
p proportion of the total line from station A (fault phase angle' aE between two sources, per Fig. 2. Shunt
locator location) to the fault capacitances'and reactances will be neglected.
R, S, T power system phases (alternative: The derivation will use the superposition principle with
A,B ,C) Thevenin's theorem, where the total voltage and current at
any point in the network Is the sum of the prefault values
RlL, ROL
resistances
positive-and zero-sequence line plus those produced by short circuiting the voltage sources
and inserting UF, the open-circuit voltage at the fault point,
as shown 'in Fig. 3. The polarity of' UF in the equivalent
RlSA, RlSB positive-sequence source resist- circuit of Fig. 3 is such as to cancel the pre-fault voltage
ance at station A and B, respectively except for the fault resistance drop. For the 3-phase fault
case, the negative- and zero-sequence voltage drops are
RF fault resistance zero.
UA, UB voltage on faulted phase(s) at station The positive direction of the load current ILD is
A and B, respectively. General, i.e. not arbitrarily chosen as shown in Fig. 2.-
referring to a specific symmetrical component
The positive-sequence prefault voltage at station A is:
UAE, UBE source voltage behind equivalent (UIA) L = UAE - ILD ZiSA
impedance at station A and B, respectively
- UF + ILD p ZlL (Al)
UlA actual positive-sequence voltage at station A
(U1A)F The positive-sequence change in voltage at A
positive-sequence change in voltage at station A produced by the fault is:
resulting from fault
(UlA)L, (U1B)L (UlA) F = - DiA IlF ZJSA
positive-sequence voltage prior to fault at sta- =
DIA 11F P ZlL + '1F RF - UF
tion A and B, respectively (A2)
UF voltage at fault point prior to fault The actual voltage at A is:

URA, USA, UTA UJA = (UJA) L + (UlA) F

voltage to ground at station A, phases R, S and = UF + ILD P ZJL + DlA I'F P ZlL
T, respectively
+ IlF RF - UF
XlL, XOL
positive-and zero-sequence line reactance
ILD P ZiL + (DiA P ZiL + RF) IlF
XlSA, XtSJ3 (A3)
positive-sequence source reactance at station A
and B, respectively The actual current in the line at A is:

ZA apparent impedance on the line at station A IJA = ILD + AIJA

ZJSA, ZlSB A positive-sequence source impedance = ILD + DlA IJF (A4)
at station and B, respectively
 435

From (A3) and (A4):

jXA UIAA ZA = RA + jXA
ZA = UlA/IlA
B
P ZlL (ILD + DlA IlF) + RF 11F
ILD + DiA IlF
= P ZlL + RF ( ILIIF)A l

=PZlL+
RF =LDO
DlA + ~fl
(A5) _%.. Increasing RF
IJF

From Fig. 3: ILDVO

UF
IJF RF + DiA (P ZlL + ZiSA) pZlL
(A6)

=
(1 P) ZlL + ZlSB RA
DlA
where
ZlSA + ZlL + ZlSB
(A?) A

Eq. (A6) in (A5):

RF
ZA = P ZlL +
Fig. Al: Appai-kriit i,npedaiee ZA.
DlA + ILD RF + DlA (P ZlL + ZlSA)
UF

(A8) ,pparent
eactance
Eq. (A8) is a quadratic in p, the unknown fault rop
location. RF is also unknown. N

Note that the argument for the RF term is influenced

by both the load current angle and by the impedances.
For a long line and a fault near station B, DlA in Eq.
(A?) approaches: 'lA
ZiSB
Di = --

Assuming that ZlSB has a larger argument than ZlL,

DlA will have positive argument. Fig. A2: Phasers showing apparent reactance effect in fault
Neglecting load current, Eq. (A8) reduces to:
point potentiaL
RF
ZA =
P ZJL + DA
import condition at station A (UAE lags UBE), the error
will be positive.
This is plotted in Fig. Al with the locus for varying
RF; the curve with the load effect included is also plotted.
The apparent-reactance effect can also be seen on a
phasor diagram such as Fig. A2, with the current plotted
horizontally as the reference.
Fig. A2 can be derived by rearranging Eq.(A3):
UiA = P ZIL (ILD + DA IlF) + IlF RF
= P ZiL (11A) + IlF RF (A9)
where 11A = actual line current at station A.
Thus, a conventional reactance distance relay or
fault locator using the same principle is subject to an
error because of fault resistance, load current and
differences in system-impedance arguments. The error
shown in Figs. Al and A2 is negative; however, for an
436

Discussion A computer simulation on a 400 kV, 150 km line with a shunt load
J. M. Crockett (Westinghouse Canada Inc., Hamilton, Ontario, Canada): applied at Station A of the proportions stated by Mr. Crockett produc-
The authors have outlined an interesting concept in fault locator design ed these results:
but some additional information would be useful. Source
The derivation of the equivalent source impedances Z1SA and Z1SN
could be described in greater detail. It would appear that they are not Fault Rf Impedange Error
actual source impedances but based on a system reduction to a two line Case Point Ohms Scalar Angle wo
1 0.5 20 23 90 1-
machine equivalent and then the delta star conversion of the equivalent 2 0.5 20 22.9 85.2 0
parallel line and source impedances. If the resulting mutual branch is 3 0.5 100 23 90 3
eliminated, positive sequence source impedances result which yield cor- 4 0.5 100 22.9 85.2 0
rect distribution factors for all fault locations. Is this the approach used? 5 0.8 100 23 90 4
It is not exact in the zero sequence network if mutual coupling is involved. 6 0.8 100 22.9 85.2 0
The positive sequence equivalent is not precisely rigorous if shunt loads
at the line terminals are ignored or variable, and the degree of precision All six faults were phase to ground. Cases 2, 4 and 6 have zero error,
as regards fault locator accuracy may be very important even though because the load impedance has been rigorously accommodated. Cases
the resulting source impedance argument variation is less than 3 degrees. 3 and 5 represent extreme conditions, with 100 ohms of fault resistance,
As an example of errors, consider source to line impedance ratio in particularly case 5 where the fault is near the far end. By choosing a
the order of 0.5, shunt loads in the order of 12 or more times the source representative value for the load impedances variations in this value will
impedance, a substantial phase shift between sources, a high resistance have a limited error effect.
fault to ground such as a tree, and a fault locator at the power receiving The parameters for this example are:
line terminal. I believe substantial changes in measured fault location magnitude angle-deg.
would be observed as the shunt load, fault location, and fault impedance
vary. Would the authors agree? ZISA 23 90
The ability to locate a high impedance fault to foliage would indeed ZOSA 23 90
be useful but has this capability been demonstated? Slow relaying would ZSH 276 0
be expected for such a fault and hence apparent current and voltage devia- ZISB 20 90
tions could be zero when the line was tripped. ZOSB 40 90
Finally, I believe that the power-sending terminal is the preferred fault ZIL 46.5 84.4
locator position. Would the authors agree? ZOL 156.7 77.5
UAE 230.9 0
Manuscript received August 2, 1984. ZBE 230.9 -38.5
ILD 2312 -14.3
ZSp 22.9 85.2
L. Eriksson, M. M. Saha, and G. D. Rockefeller: The authors thank Impedances are in ohms, voltages in kV, currents in amperes and angles
Mr. Crockett for his interest in this paper and for his relevant comments in degrees. ZSF is the positive-sequence equivalent of the paralleled source
and questions. His discussion can be segregated into four areas: ZSA and load impedance ZSH. Note that the effect of the load impedance
1. derivation of the current distribution factor for the parallel line case at Station A is to shift the actual source angle from 90 deg. to 85.2 deg.
2. effect of shunt loads for the paralleled equivalent of the source and load impedances. The
3. availability of prefault current information for a high-impedance load has a negligible effect on the magnitude of this equivalent.
fault 3. If the normal high-speed proteciton fails to detect a high-resistance
4. preferred spot for the locator. fault, such as a mid-span flashover to foliage, two possibilities exist for
These item numbers will be used below. the locator:
1. The current distribution factor as stated in Eq. 13A is rigorous a) The fault-locator terminal opens following the remote terminal.
assuming an identical parallel line. It is independent of any zero-sequence In this case, there is no load current flow during the fault period used
effect, including mutual impedance, because only the positive-sequence by the locator to compute. Thus, no error occurs because of load current.
factor is needed to derive the total fault current If. b) The fault-locator terminal opens first. The prefault data are not
Measured voltage will contain any mutual induction from a parallel available in memory if the line protection trip is delayed. This
line-this can be compensated for by inputting the zero-sequence mutual situation will be recognized by the program and an alternative
impedance to the locator and providing the program with the measured "slow trip" routine is processed. Errors will occur because of
parallel-line zero-sequence current via the printer loop, from the com- the load current component.
panion locator. 4. The direction of power flow is not of concern to the fault locator,
The Eq. 13A expression is derived using a delta-star conversion, where since its algorithm eliminates load flow effects. That is, there is no bias
the source impedances are merged into the equivalent impedances. towards installing the locator at the "power sending terminal" unless
However, the actual source impedances are input just as for the single this describes the stronger source of fault current. The stronger source
line algorithm which uses Eq. 13 for the distribution factor. installation reduces errors because the weaker end produces less of an
2. Shunt impedance effects at the two ends of the line can be rigously infeed effect.
canceled by including these impedances in the source impedances. This 5. The following additional field results are noteworthy. Six phase-
is the case for either the single line or the parallel line algorithm. The ground faults occurred during very high winds on a 400 kV, 135 km line.
load and source impedances are merely paralleled. Mr. Crockett is cor- The precise location of these temporary faults is not known. The locator
rect that changes in these, coupled with fault resistance, will introduce operated in each case, indicating in the range of 93 to 990/.
errors. The magnitude of the load impedance will have little effect, but
its angular difference compared to the other impedances will have an Manuscript received September 4, 1984.
effect as the impedance varies from the nominal value used to determine
the source impedance input. The following example illustrates this effect.