1

Traveling Wave Fault Location in Protective

Relays: Design, Testing, and Results
Stephen Marx, Bonneville Power Administration
Brian K. Johnson, University of Idaho
Armando Guzmn, Veselin Skendzic, and Mangapathirao V. Mynam, Schweitzer Engineering Laboratories, Inc.
AbstractFaults in power transmission lines cause transients
that travel at a speed close to the speed of light and propagate
along the line as traveling waves (TWs). This paper shows how
these transients can be measured in a protective relay and used
to enhance its fault locating function. The TW-based fault
locating function in a protective relay takes advantage of the
internal protection elements, the communications channel to the
remote terminal, and Global Positioning System-based time
synchronization. This approach provides accurate fault location
estimation for transmission lines automatically within a couple of
seconds after the fault. The TW fault locator within the relay
uses conventional current transformer measurements and does
not require any additional wiring or special installation
considerations. These relays detect internal line faults and use
TW- and impedance-based algorithms to optimize the estimation
and reporting of the fault location. The TW- and impedance-
based algorithms complement each other to provide accurate
fault location estimation for all internal faults, independent of the
fault incidence angle. This paper provides a tutorial on TW fault
location and describes a TW fault locating algorithm that uses
time-synchronized measurements of the TW currents at the line
terminals to determine the fault location. The proposed
implementation is suitable for line terminals with one or two
breakers. The paper also discusses methods for testing the TW
fault locating algorithm using a playback system and a hardware
model that mimics a transmission line. The paper presents the
details and experiences of a field application of these relays on a
high-voltage transmission line.
I. INTRODUCTION
The Bonneville Power Administration (BPA) has been
using traveling wave (TW) technology to locate faults in their
extra-high voltage (EHV) transmission networks since the
1950s [1]. In the 1960s and 1970s, BPA installed an
automated fault locating system that used microwave
communications to send the TW arrival information to the
remote terminal for fault location estimation [2].
References [3] and [4] describe the development and field
evaluation of the performance of a digital fault locator for
high-voltage direct current (HVDC) lines that uses voltage and
current measurements from one line terminal to estimate the
fault location. Using voltage and current measurements, we
can calculate incident and reflected waves. Applications based
on incident waves are immune to the effects caused by
termination impedances. In 1987, BPA started to use Global
Positioning System (GPS) time information to measure the
arrival of the TWs [5].
Later on, utilities like BPA used dedicated TW-based fault
locating devices that sample the currents or voltages at
sampling rates greater than 1 MHz [6]. Some of these devices
also include fault disturbance recording and are capable of
reporting accurate fault location results [7].
Numerical protective relays include fault location
estimation algorithms based on the line impedance and voltage
and current measurements. In most applications, these relays
only use measurements from the local terminal. Some relays
also use information from the remote terminal to estimate the
fault location. Using information from the local and remote
terminals minimizes errors due to mutual coupling with
adjacent lines, system nonhomogeneity, and fault resistance
[8] [9]. In some applications, such as series-compensated
lines, the impedance-based fault locating methods are
challenged and utilities require more accurate estimation than
traditional relays provide.
This paper discusses the basic principles of TWs, a
protective relay that estimates fault location using TW
information, and the benefits of having TW fault location
estimation within a protective relay. The paper also describes
how to test the TW fault locating system using low-energy
analog signals and an analog transmission line model. We also
share the field experiences from an installation on a BPA
161 kV transmission line.
II. OVERVIEW OF TRAVELING WAVES IN TRANSMISSION LINES
A fault on a transmission line generates TWs that
propagate from the fault location to the line terminals with a
propagation velocity that depends on the inductance and
capacitance of the line. Fig. 1 shows the equivalent circuit of a
segment with length x of a two-conductor transmission line.
The circuit includes the resistance R, inductance L,
conductance G, and capacitance C of the line in per unit of the
total line length [10].
v(x,t)
i(x,t) i(x + x,t)
v(x + x,t)
R x L x
C x G x
x

Fig. 1. Equivalent circuit of a segment of a two-conductor transmission line.
2

We use Kirchhoffs voltage law, shown in (1), and
Kirchhoffs current law, shown in (2), to relate the voltages
and currents at locations x and x+x.

( ) ( )
( )
( )
v x, t R x i x, t
i x, t
L x v x x, t
t
= A
c
+ A + + A
c
(1)

( ) ( )
( )
( )
i x, t G x v x x, t
v x x, t
C x i x x, t
t
= A + A
c + A
+ A + + A
c
(2)
Equations (3) and (4) determine the voltage and current as
a function of x and time for the two-conductor transmission
line in the time domain as the length of the segment x
approaches zero.

( )
( )
( ) v x, t i x, t
R i x, t L
x t
c c
=
c c
(3)

( )
( )
( ) i x, t v x, t
G v x, t C
x t
c c
=
c c
(4)
We differentiate (3) with respect to x and (4) with respect
to t to obtain (5) and (6), respectively.

( ) ( ) ( )
2 2
2
v x, t i x, t i x, t
R L
x x t x
c c c
=
c c c c
(5)

( ) ( ) ( )
2 2
2
i x, t v x, t v x, t
G C
x t t t
c c c
=
c c c c
(6)
We substitute
( ) i x, t
x
c
c
from (4) and
( )
2
i x, t
x t
c
c c
from (6)
into (5) to obtain the voltage wave equation shown in (7).

( ) ( )
( )
( )
( )
2 2
2 2
v x, t v x, t
LC
x t
v x, t
RC GL GR v x, t
t
c c
=
c c
c
+ + +
c
(7)
Similarly, we differentiate (3) with respect to t and (4) with
respect to x to obtain (8) and (9), respectively.

( ) ( ) ( )
2 2
2
v x, t i x, t i x, t
R L
x t t t
c c c
=
c c c c
(8)

( ) ( ) ( )
2 2
2
i x, t v x, t v x, t
G C
x x t x
c c c
=
c c c c
(9)
We substitute
( ) v x, t
x
c
c
from (3) and
( )
2
v x, t
x t
c
c c
from (8)
into (9) to obtain the current wave equation shown in (10).

( ) ( )
( )
( )
( )
2 2
2 2
i x, t i x, t
LC
x t
i x, t
RC GL GR i x, t
t
c c
=
c c
c
+ + +
c
(10)
Equations (7) and (10) determine how the voltage and
current waves propagate along a two-conductor transmission
line.
To analyze the voltage and current TWs in multiphase
transmission lines, we used an Electromagnetic Transients
Program (EMTP) that models transmission lines considering
the changes in the conductor resistance and inductance due to
skin effect [11] [12] [13]. Fig. 2 shows how the current waves
propagate across a 400 kV line in response to a nominal
voltage step change on A-phase and B-phase at the sending
end (the step change occurs at t = 0).
0.2 0.4 0.6 0.8 1 1.2 1.4 1.6
0.2 0.4 0.6 0.8 1 1.2 1.4 1.6
150 km
0.2 0.4 0.6 0.8 1 1.2 1.4 1.6
300 km
450 km
1,000
1,000
0
1,000
1,000
0
1,000
1,000
0
1,000
1,000
0
0.2 0.4 0.6 0.8 1 1.2 1.4 1.6
0.2 0.4 0.6 0.8 1 1.2 1.4 1.6
0.2 0.4 0.6 0.8 1 1.2 1.4 1.6
0.2 0.4 0.6 0.8 1 1.2 1.4 1.6
50 km
150 km
300 km
450 km
Time (milliseconds)

Fig. 2. Current waves at 50, 150, 300, and 450 kilometers traveling on a
400 kV line for a nominal voltage step change at the sending end, where
A-phase is green, B-phase is blue, and C-phase is red.
3

III. TRAVELING WAVE FAULT LOCATION
FOR TWO-TERMINAL LINES
TW-based fault location provides better accuracy relative
to impedance-based fault locating methods. Single end
(Type A) and double end (Type D) are the two most common
methods for computing fault location using TWs [1]. Type A
uses the time difference between the first arrived wave and the
successive reflections from the fault location to compute the
fault location. This method is appealing because it only
depends on local information; therefore, it does not require a
communications channel. However, identifying the reflections
is a major challenge. The reflections can arrive from the fault
location, from the remote terminal, or from behind the local
terminal. Accurately identifying the reflection from the fault
location poses a challenge for single-end TW-based fault
location, especially on ac transmission lines. The double-end
method overcomes the challenge of identifying the reflections
from the fault but requires the TW information from the
remote terminal. This Type D method uses the time difference
between the first arrived TWs captured at both terminals along
with the line length and the wave propagation velocity to
compute the fault location. Fig. 3 shows the waves
propagating to Terminals A and B following a fault condition
on a transmission line.

Fig. 3. TWs propagating to Terminals A and B.
Measurement devices at the line terminals detect the TWs
and accurately time-stamp the arrival of the wave using a
common time reference (e.g., IRIG-B or IEEE 1588). The
typical time-stamping accuracy is better than 1 microsecond.
The TW-based fault location is computed using (11).

( ) LL TwaveA TwaveB c LPVEL
TWFL
2
+
= (11)
where:
TWFL is the TW-based fault location.
LL is the line length.
TwaveA is the TW arrival time recorded at Terminal A.
TwaveB is the TW arrival time recorded at Terminal B.
c is the speed of light.
LPVEL is the propagation velocity of the TW in per unit
of the speed of light.
The TW propagation velocity is a key parameter in the
fault location calculation and is typically obtained from line
parameter estimation programs. We can also estimate
propagation velocity using TW measurements with the
following:
- Local TW information recorded during line or reactor
energization tests.
- Local and remote TW information recorded during
external faults.
Voltage and/or current measurements capture the TWs. The
adequate frequency bandwidth of current transformers (CTs)
makes current TWs better suited for this application than
voltage TWs measured at the secondary terminals of the step-
down transformer of the capacitance coupled voltage
transformer (CCVT) [14]. Typical installations have
communications between the substation and the control center,
where computer-based analysis tools retrieve the TW
information captured at the line terminals and compute the
fault location. In this paper, we discuss installations where
protective relays exchange TW information obtained from the
phase currents and automatically calculate the fault location at
the line terminals within a couple of seconds after the fault.
IV. TW FAULT LOCATION IN PROTECTIVE RELAYS
A. Benefits of TW Fault Location in a Protective Relay
Numerical protective relays have included fault location
estimation based on voltage and current measurements and
line impedance since 1982 [15] [16]. These relays use voltage
and current measurements acquired at the local terminal and
report the fault location estimation results at the substation and
the control center through automessaging right after the
occurrence of a fault. While this approach provides
estimations within 2 percent of the line length, there are cases
where mutual coupling, fault resistance, and system
nonhomogeneity can cause large errors. In these cases, the
impedance-based fault locating methods in protective relays
can be improved using local and remote (double-end)
measurements [8] [9] [17].
Impedance-based fault location estimation requires the
presence of the fault for a couple of cycles to provide accurate
results. While this requirement is not an issue in
subtransmission network applications, it can be an issue in
EHV and ultra-high voltage (UHV) applications where faults
are sometimes cleared in less than two cycles. Furthermore,
these impedance-based estimation methods might not be
applicable to lines with series compensation or lines close to
series compensation due to subsynchronous oscillations,
voltage inversion, and so on. Because of the importance of
locating faults to avoid fault reoccurrences and the high cost
associated with finding line faults, utilities require accurate
fault locating devices for all applications. For this reason,
some utilities have installed dedicated devices that detect the
time of arrival of TWs at the line terminals and estimate the
fault location using this information [6] [18]. While these
devices provide more accurate fault location estimation than
relays that use impedance-based methods to estimate the fault
location, there are cases where they cannot estimate the fault
location when faults occur at the voltage zero crossing.
Protective relays that include both impedance-based and
TW-based fault location have the advantage of providing fault
location even in cases where the TW amplitude is too low for
reliable detection (e.g., faults that occur at voltage zero). In
these cases, the relays estimate the fault location using line
impedance and local and remote voltage and current
measurements. If the remote measurements are not available,
the relay estimates the fault location using local measurements
4

only, thus providing robust response with the best possible
accuracy under all fault conditions.
B. Relay-to-Relay Communications
The relay discussed in this paper uses a 64 kbps channel
that exchanges currents for differential protection purposes.
The relay takes advantage of this bandwidth and includes TW
information within the data packet without affecting the
performance of the differential element. The relays exchange
the times of arrival of the TWs (see Fig. 4) and use this
information to estimate the fault location, make the results
available at the relay location, and send the results to the
control center within a couple of seconds after the occurrence
of the fault.

Fig. 4. Relays exchange the time of arrival of the TW to estimate the fault
location in real time.
C. Traveling Wave Measurements
High-frequency transients created by power system faults
propagate at speeds that are close to the speed of light.
However, high-voltage transmission lines are optimized to
operate at nominal power system frequency with standard
values of 50 or 60 Hz, and some of them are dc lines.
Significant engineering effort is made in reducing the
transmission line losses at these frequencies, with no attempt
to consider their behavior at the high frequencies (0.1 to
1 MHz) that are used by TWs.
Fortunately, the physics associated with the construction of
efficient high-voltage transmission lines aid with the TW
propagation. For various economic, operational, and
environmental reasons, high-voltage transmission lines are
built as regular structures, with uniform distances among
phase conductors, uniform dielectric (air), constant conductor
cross section, and regular transmission tower support, as
illustrated in Fig. 5. Because of all of these factors,
transmission lines can transport signals well into the
megahertz range. This fact has been used for a very long time
and is best exemplified by the power line carrier-based
communications that operate in the 100 to 600 kHz range. In
addition to transmission lines with well-defined losses, TW-
based fault location is further aided by the fact that power
system faults generate significant amounts of energy in the
frequency range of interest. This energy provides sufficient
signal levels at the transmission line ends to indicate the
presence of a fault on the line.

Fig. 5. Typical high-voltage transmission tower.
Arrival of the TWs at the substations with sufficient energy
is only the initial prerequisite for successful TW-based fault
location. Once the waves arrive, they must be measured
(extracted from the current and/or voltage measurements) and
delivered to the fault location estimation algorithm. Currents
and voltages on the transmission line are measured using
standard CTs and voltage transformers (VTs). The responses
of the CTs and VTs have been optimized for nominal
frequency operation. TW signals can be measured using
specialized high-frequency transducers similar to those used in
high-voltage laboratories, but the high cost and custom nature
of these devices make this approach impractical for wide-scale
utility applications.
From the perspective of utilities, it would be ideal if the
TW-based fault locating device could be installed in the
substation control house using conventional wiring practices.
It is also highly desirable that the new device be deployed
without requiring new communications or time-
synchronization infrastructure. While studying this problem,
the authors decided to start by investigating the frequency
response of conventional instrument transformers. We
classified measurement transformers in the following
subcategories.
Current transformers include:
- Conventional iron-core CTs
- Nonconventional low-energy CTs
Optical CTs
Low-power CTs
Rogowski coils
Voltage transformers include:
- Conventional VTs
CCVTs
Inductive VTs
Open magnetic core-based VTs
40.35feet
(12.3meters)
82feet
(25meters)
5

- Nonconventional low-energy VTs
High-voltage capacitive, resistive, and
compensated dividers
Transformer bushing tap-based VTs (capacitive
dividers)
Optical VTs
Comparison of the available technologies shows that the
two most popular choices used for current and voltage
measurements on high-voltage (greater than 69 kV)
transmission lines are conventional CTs (freestanding or
bushing design) and CCVTs.
Conventional CT construction is very simple (a toroidal
core with at least one secondary winding). Conventional CTs
have good high-frequency response [19]. The CT bandwidth
measurement results reported in literature vary, but the
common consensus is that the usable pass band (3 dB point)
easily reaches 100 kHz [19] and may often be closer to the
200 kHz [20] or 500 kHz level [21]. Measurements performed
by the authors confirm these results (300 kHz, in our case),
with the additional finding that the measurement bandwidth is
inversely proportional to the CT ratio. Low-ratio CTs have
fewer turns and lower interwinding capacitance, thus
providing a larger bandwidth. This effect is especially
pronounced in input CTs, which are typically used to measure
currents inside the protective relay.
Fig. 6 illustrates the results obtained by using an
IEC 61000-4-5-compliant test generator to inject a 2 kA (8 by
20 microseconds) pulse into the primary of the tested CT. The
test setup included 70 meters of #10 (5.26 mm
2
) wire that
connects the protective relay to the secondary of the CT.
The results shown in Fig. 6 demonstrate that equipment
that was originally built to operate at 60 Hz works very well at
frequencies well beyond the original design range. The output
current observed at the CT secondary terminals (the middle
trace) matches the primary current measured using a high-
frequency shunt (the top trace).

Fig. 6. A 600:5 CT response measured using a 2 kA transient with 8 by
20 microsecond lightning surge waveform and 70 meters (230 feet) of
secondary cabling.
The bottom trace shows the signal recorded at the
secondary of the input CT (inside the relay). The recorded
waveform is a reasonably faithful representation of the
primary signal. Data obtained from this and other tests were
used to design an input circuit optimized for TW applications,
with the frequency response shown in Fig. 7. The response is
flat up to 1 MHz with controlled high-frequency attenuation
extending well into the radio frequency range.
The current measured by the relay (see Fig. 6) has
additional ripple. Further simulations confirmed that this
ripple was contributed by the secondary cable and that this
cable does not behave as a properly terminated transmission
line. At one end, the cable is connected to the CT (a current
source with an impedance approaching the impedance of an
infinite source), and it is terminated into a short circuit at the
relay end. This configuration creates multiple reflections
resulting in multiple resonances, as described in [22]. The
observed ripple frequency was consistent with the length of
the secondary cable.

Fig. 7. Frequency response of the relay analog circuitry (excluding anti-aliasing filter).
PhaseResponse
MagnitudeResponse
6


Fig. 8. LTspice simulation showing ringing of the current signal at the relay end caused by the secondary wiring of the CT.

Fig. 9. LTspice simulation showing the peaks in the frequency response caused by the secondary wiring of the CT.
Fig. 8 shows the results of an LTspice simulation that
demonstrates the secondary cable ringing effects, while Fig. 9
shows the frequency response peaks caused by the same
effect.
Ringing effects are present in real life, as demonstrated by
the TW transient waveform shown in Fig. 10. This waveform
was recorded in the field during an actual power system fault.
The TW spectral content is inherently limited by the high-
frequency attenuation provided by the power system
transmission line. However, as the TW transient reaches the
substation (the relay location), it is capable of exciting a large
number of local high-frequency resonances, which have no
value for fault location estimation. Therefore, sampling
frequencies found on most TW systems are normally
somewhere in the 0.5 to 5 MHz range.
The TW voltage transients are also suitable for fault
location estimation. Unfortunately, the quality of the
measurements provided by conventional VTs is significantly
inferior to that of their CT counterparts. CCVTs normally
used at transmission levels are tuned to the nominal power
system frequency. Although the high-frequency capacitor
stacks used in these transformers are often usable up to
1 MHz, the tuning reactor eliminates the high-frequency
signals, leaving only the unintended transients caused by
parasitic capacitance, which tends to be different for every
CCVT design [23]. Because the voltage measurements
captured at the secondary terminals of the CCVT step-down
transformer do not properly measure TW transients, some
systems use additional high-frequency transducers mounted in
series with the CCVT stack or nonconventional resistive-
capacitive dividers to capture the voltage TWs.
7

The TW-based fault locator described in this paper does
not require additional CTs or CT wiring because the relay is
able to use the same current measurements for line protection
and fault location estimation.
The TW-based fault locating subsystem is implemented in
parallel with the conventional data acquisition chain, ensuring
no interference with the mission-critical protection functions.
The TW-based fault locating subsystem uses high-speed
analog-to-digital (A/D) converters that sample the phase
currents at 1.56 MHz and creates dedicated COMTRADE
event reports that contain TW information.
250
0
50
100
150
200
320 500 400 520
C
u
r
r
e
n
t

(
A
)
Time (microseconds)
440 460 480 340 380 360 420

Fig. 10. Field-recorded TW waveform showing ringing effects in an actual
substation application.
D. Current Monitoring in Double-Breaker Applications
The protective relay has two sets of current inputs to
measure the currents through the two breakers in breaker-and-
a-half and double-breaker schemes, as Fig. 11 illustrates. This
capability allows the relay to monitor the line currents based
on user-programmable logic variables that select which
currents to use according to the operation requirements of the
power system. The fault locating algorithm uses the selected
TWIA, TWIB, and TWIC currents for fault location
estimation.

Fig. 11. Current source selection for TW fault location in double-breaker
applications.
V. TRAVELING WAVE TESTING
A. Testing Using Low-Energy Analog Signals
To verify the accuracy of the TW-based fault location
estimation, we created a power system model using an EMTP.
The power system model includes frequency-dependent
transmission line models and effective terminal capacitance at
the buses. Table I shows typical terminal capacitances for
generators, transformers, and bus systems [24].
TABLE I
TYPICAL TERMINAL CAPACITANCES IN MICROFARADS (F)
Maximum Minimum Average
Hydrogenerators
(Salient Poles)
0.001 0.0002 0.0006
Generators
(Steam Turbine)
0.001 0.0001 0.0005
Transformers
(Distribution)
0.002 0.0004 0.0010
Transformers
(Power)
0.001 0.0002 0.0005
Bus Systems 0.015 0.0020 0.0050
We simulated phase-to-ground and phase-to-phase faults at
known locations on a 189-mile line. We saved the current
signals from both terminals (sending end and receiving end)
that were sampled at a rate of 3 MHz as comma-separated
value (CSV) files. We used a signal generator that is capable
of playing back megahertz sampling rate data with 16-bit
resolution (see Fig. 12). The signal generator plays back the
CSV files via the low-energy analog (LEA) interface of the
relay. Two playback units supplied the three-phase currents
from each line terminal to the corresponding relays. The
playback units were configured to inject the signals
synchronously to both relays.

Fig. 12. Test setup using LEA signals to test the TW fault locating system.
8

Fig. 13 and Fig. 14 show the phase currents at the sending
and receiving terminals, respectively, for an AB fault at
160.45 miles (from the sending terminal).
P
r
i
m
a
r
y

C
u
r
r
e
n
t

(
A
)

Fig. 13. Phase currents applied to the sending-end terminal for an AB fault.
Fault Initiated
2
1.5
1
0.5
0
0.5
1
1.5
0.04 0.045 0.05 0.055 0.06 0.065
x 10
4
Time (seconds)
0.07

Fig. 14. Phase currents applied to the receiving-end terminal for an AB
fault.
The relays filter the phase currents to capture the high-
frequency content of the signals. The relays exchange the time
stamp of the retrieved high-frequency wave. Fig. 15 and
Fig. 16 show the high-frequency content of the phase currents
that were retrieved from the relays.

Fig. 15. TWs captured at the sending end for an AB fault.

Fig. 16. TWs captured at the receiving end for an AB fault.
We calculated the TW propagation velocity using the TW
information obtained for a line energization event.
The relays recorded the following left terminal (TwaveL)
and right terminal (TwaveR) time stamps for this event:
- TwaveL = 6.773364044 seconds
- TwaveR = 6.772648441 seconds
- LL = 189 miles
- LPVEL = 0.9903
- c = 186282.39705 miles per second
The distance to the fault is 160.51 miles, as shown in (12).
In this case, the fault location estimation error is
0.06 miles.

( )
Fault
Fault
0.9903186, 282.39705 miles/s 6.773364044 6.772648441 189 miles
D
2 2
D 160.51 miles

= +
=
(12)
9

B. Testing Using a Physical Line Model
Additional verification was performed using the physical
single-phase transmission line model shown in Fig. 17.
Although less precise and significantly less versatile than
EMTP simulations, the physical low-energy transmission line
model offered the ability to test the entire system (including
the relay input CTs).

Fig. 17. Physical transmission line model.
The physical transmission line model was implemented
using a total of 500 sections to simulate a 38-mile line. The
custom card developed for this model is shown in Fig. 18. The
model operates at 48 V, 1 A with a maximum fault current
level of 5 A. Faults can be applied at three locations. Fig. 17
shows faults at 20, 60, and 100 percent of the line length.

Fig. 18. Transmission line model section used to simulate a 3.8-mile single-
phase segment. The thyristor-based fault model is visible at the lower right-
hand side. The complete line model uses 10 sections.
VI. FIELD INSTALLATION
A. Fault Location Experience
BPA owns and operates Goshen and Drummond
substations. The Goshen-Drummond line is operated at
161 kV, and according to the BPA system data book, its line
length is 72.77 miles. This transmission line is located in
eastern Idaho close to the Wyoming border (see Fig. 19). The
line shares a right of way with two other 161 kV lines for
approximately 4.75 miles. Then it shares the right of way with
another 161 kV line for the next 17 miles. The line was
originally built for 115 kV and was later upgraded to 161 kV
without changing conductors or insulators.
Targhee Tap
Swan Valley
Targhee
Goshen
Drummond
Idaho Wyoming

Fig. 19. Goshen-Drummond 161 kV line (blue) and neighboring 161 kV
lines (magenta).
After the 161 kV upgrade, the line experienced 40 faults in
the past five years. The most common causes of faults on this
line include the following:
- Galloping conductors clashing because of the wind.
- Farmers spraying fertilizers on the conductors and
insulators.
- Flying projectiles hitting the conductors and
insulators.
In the past, for a permanent fault, a lineman would drive
along the line until the fault was found. For a long line built
over rough terrain, this approach could cause a long outage.
Later, BPA used oscillographic records of faults to estimate
the fault location. BPA personnel calculated the impedance to
the fault from the fault voltages and currents and estimated the
fault location. This information would give the lineman a
starting point to look for the fault. For temporary faults,
linemen inspected every insulator looking for an indication of
where the flash occurred. With the advent of digital relays, the
relay would estimate the fault location based upon the
impedance and the length of the line. If either of these were
not accurate, then the location would not be accurate. Multiple
sections, different tower configurations, fault resistance,
system nonhomogeneity, and zero-sequence mutual coupling
challenge the accuracy of impedance-based fault locating
methods.
10

The Goshen-Drummond line is composed of four different
tower structures, as shown in Fig. 20. The line is spanned
across 18 sections. Table II shows the tower type and line
length associated with each section.
J30
27'GW
24.67'
18.67'
GMD = 27.13'
SC
18.67' 24.67'
GMD = 28.15'
DC
J7
GMD = 25.20'
SC
20' 20'
H33
7'
1
6
.5
5
'
15'
GMD = 15.65'
SC
GW
H18
GMD = 20.16'
SC
16' 16'

Fig. 20. Tower structures on the Goshen-Drummond line.
TABLE II
TOWER TYPE AND LINE SECTION LENGTH
Tower Type Length (miles)
H18 0.12
J30 21.05
J7 0.68
H33 11.63
J7 0.52
J7 0.25
J7 0.94
H33 2.22
J7 4.08
J30 0.79
J30 0.42
J30 0.08
J7 8.86
H33 5.41
J7 0.37
J7 1.27
H33 5.82
J7 8.27
Fig. 21 shows the one-line diagram that includes the
Goshen-Drummond line and relay CT connections. Notice
that the line termination at Drummond is an autotransformer.

Fig. 21. Transmission network that includes the Goshen-Drummond line.
11

B. Traveling Wave Device Installation
On April 4, 2012, BPA installed two relays with TW
locating capability on the Goshen-Drummond 161 kV line.
These relays are capable of exchanging TW information via a
64 kbps channel and estimating fault location in real time, but
the communications channels were not available at
installation. Therefore, we manually retrieved the
COMTRADE event records with TW information and
estimated the fault location after the occurrence of each fault.
The event records include the TW phase currents and time-
stamp information.
C. Propagation Velocity and Line Length
As previously mentioned, double-end TW fault location
relies on the line length and propagation velocity settings
along with the measured time difference between the arrival
times of the TWs captured at both terminals of the
transmission line. Typically, utilities have an estimate of the
line length based on the geographic or linear length or road
miles of the transmission line; some utilities consider line sag
in their estimates. Including the sag in line length estimates
reduces errors in the TW fault location.
We measured the propagation velocity based on the line
length and travel time of the waves. We estimated travel time
using the TW information that we captured during line
energization. We energized the line from Goshen while the
terminal at Drummond was open and captured the event
reports to determine the wave propagation velocity. Fig. 22
shows the phase currents and voltages captured at the Goshen
terminal sampled at 8 kHz.

Fig. 22. Phase voltages and currents captured during line energization from
Goshen.
The phase currents depicted in Fig. 22 show the pole
scatter and the sequence of pole closing: B-phase, C-phase,
and A-phase. Furthermore, we can observe that the B-phase
and A-phase poles closed near the peak of their corresponding
voltages, while the C-phase pole closed near the voltage zero
crossing. Fig. 23 shows the phase currents filtered using an
analog band-pass filter, preserving the high-frequency content
and rejecting the fundamental frequency content; the currents
are sampled at 1.56 MHz.

Fig. 23. TW phase currents captured during line energization.
Fig. 23 shows that there is no detectable transient
corresponding to the C-phase closing as the pole closed at near
voltage zero. The information in Fig. 23 can be used to
determine the pole scatter among the breaker contacts
accurately to hundreds of nanoseconds. This information can
be further used to determine if corrective breaker maintenance
is needed. We used the time stamps corresponding to B-phase
pole closing and the reflected wave from the open terminals to
calculate the propagation velocity. We show the propagation
velocity calculations in (13) and (14), with travel time equal to
790.605 microseconds and line length equal to 72.77 miles.

2 LL 1
LPVEL
Travel time c
= (13)

2 72.77 miles 1
LPVEL
790.605 s 186282.39705 miles/s
0.98821
=

=
(14)
D. Power System Faults and Fault Location Estimates
1) Event 1: C-Phase-to-Ground Fault
The first fault occurred on April 24, 2012. Fig. 24 and
Fig. 25 show the TWs captured at the Goshen and Drummond
terminals for the C-phase-to-ground fault.

Fig. 24. Phase currents at Goshen for the C-phase-to-ground fault at
67.91 miles from Goshen terminal.
A-Phase Closing
B-Phase
Closing
Reflection From
B-Phase Closing
12


Fig. 25. Phase currents at Drummond for the C-phase-to-ground fault at
67.91 miles from Goshen terminal.
The time stamps corresponding to the TW arrival obtained
from the event records include the following:
- Goshen: Twave
Gosh
= 24.089532202 seconds
- Drummond: Twave
Drum
= 24.089186645 seconds
Based on the measured TW arrival times, we estimated
from (11) a fault location of 68.19 miles from the Goshen
terminal. When the line crew patrolled the line, they found a
damaged insulator at 67.91 miles from the Goshen terminal.
Fig. 26 shows the damaged insulator. The line crew reported
that the cause of the insulator damage could be a flashover.

Fig. 26. Damaged insulator at 67.91 miles from Goshen terminal.
2) Event 2: B-Phase-to-Ground Fault
The second fault occurred on May 11, 2012. This
permanent fault was caused by a lead projectile hitting the
B-phase insulators at a high speed. Fig. 27 and Fig. 28 show
the high-frequency components of the phase currents captured
at both terminals.

Fig. 27. Phase currents at Goshen for the B-phase-to-ground fault at
38.16 miles from the Goshen terminal.

Fig. 28. Phase currents at Drummond for the B-phase-to-ground fault at
38.16 miles from the Goshen terminal.
The time stamps corresponding to the TW arrival obtained
from the event records include the following:
- Goshen: Twave
Gosh
= 36.832684476 seconds
- Drummond: Twave
Drum
= 36.832667109 seconds
We estimated a fault location of 37.98 miles from the
Goshen terminal. The line crew found the fault at 38.16 miles
from the Goshen terminal. Fig. 29 shows one of the damaged
insulators in the insulator string.

Fig. 29. Damaged insulator at 38.16 miles from the Goshen terminal.
Damaged
Insulator
13

3) Event 3: B-Phase-to-Ground Fault
The third fault was on May 26, 2012. Fig. 30 and Fig. 31
show the high-frequency components of the phase currents
captured at both terminals.

Fig. 30. Phase currents at Goshen for the B-phase-to-ground fault at
66.86 miles from the Goshen terminal.

Fig. 31. Phase currents at Drummond for the B-phase-to-ground fault at
66.86 miles from the Goshen terminal.
Based on the prestrikes recorded at the Goshen terminal, it
is suspected that the fault was due to lightning. The time
stamps corresponding to the TW arrival obtained from the
event records include the following:
- Goshen: Twave
Gosh
= 32.815358756 seconds
- Drummond: Twave
Drum
= 32.815023378 seconds
We estimated a fault location of 67.25 miles from the
Goshen terminal. The line crew found the fault at 66.86 miles
from Goshen.
4) Summary of Results
Table III provides the fault location reported by the relay
based on TW measurements and the actual fault location
reported by BPA. The errors between the TW-based estimated
distances and the BPA reported distances are attributed to the
nonuniformity of the line sag due to terrain elevation changes
and differences in tower structures. BPA is working on
providing accurate line length estimates to include line sag.
TABLE III
REPORTED FAULT LOCATIONS AND ASSOCIATED ERRORS
Event
Number
Faulted
Phase
TW-Based
Estimated
Distance
BPA
Reported
Distance
Error
1 C 68.19 miles 67.91 miles 0.28 miles
2 B 37.98 miles 38.16 miles 0.18 miles
3 B 67.25 miles 66.86 miles 0.39 miles
VII. CONCLUSION
Adding TW-based fault location to line protective relays
improves their fault locating capability compared with relays
that use only impedance-based fault locating methods.
Relays that use TWs and impedance-based methods to
estimate fault location provide results independent of the fault
incidence angle. These relays provide fault location at the line
terminals within a couple of seconds without the need for an
additional computer and software.
Field cases demonstrate that relays with TW fault locating
capability can locate faults to within a tower span in
applications including lines with mutual coupling.
Furthermore, there is no need for additional secondary wiring
and communications equipment when TW-based fault location
is part of the line differential relay.
VIII. REFERENCES
[1] T. W. Stringfield, D. J. Marihart, and R. F. Stevens, Fault Location
Methods for Overhead Lines, Transactions of the American Institute of
Electrical Engineers Part III: Power Apparatus and Systems, Vol. 76,
Issue 3, April 1957, pp. 518530.
[2] D. J. Marihart and N. W. Haagenson, Automatic Fault Locator for
Bonneville Power Administration, proceedings of the 1972 IEEE
Power and Energy Society Summer Meeting, San Francisco, CA,
July 1972.
[3] M. Ando, E. O. Schweitzer, III, and R. A. Baker, Development and
Field-Data Evaluation of Single-End Fault Locator for Two-Terminal
HVDC Transmission Lines, Part I: Data Collection System and Field
Data, IEEE Transactions on Power Apparatus and Systems,
Vol. PAS104, Issue 12, December 1985, pp. 35243530.
[4] M. Ando, E. O. Schweitzer, III, and R. A. Baker, Development and
Field-Data Evaluation of Single-End Fault Locator for Two-Terminal
HVDC Transmission Lines, Part II: Algorithm and Evaluation, IEEE
Transactions on Power Apparatus and Systems, Vol. PAS104,
Issue 12, December 1985, pp. 35313537.
[5] M. A. Street, Delivery and Application of Precise Timing for a
Traveling Wave Powerline Fault Locator System, proceedings of the
22nd Annual Precise Time and Time Interval (PTTI) Applications and
Planning Meeting, Vienna, VA, December 1990, pp. 355360.
[6] P. F. Gale, Overhead Line Fault Location Based on Travelling Waves
and GPS, proceedings of the Precise Measurements in Power Systems
Conference, Arlington, VA, October 1993.
[7] M. Aurangzeb, P. A. Crossley, and P. Gale, Fault Location on a
Transmission Line Using High Frequency Travelling Waves Measured
at a Single Line End, proceedings of the 2000 IEEE Power Engineering
Society Winter Meeting, Vol. 4, Singapore, January 2000,
pp. 24372442.
14

[8] D. A. Tziouvaras, J. Roberts, and G. Benmouyal, New Multi-Ended
Fault Location Design for Two- or Three-Terminal Lines, proceedings
of the 7th International Conference on Developments in Power System
Protection, Amsterdam, Netherlands, April 2001.
[9] K. Zimmerman and D. Costello, Impedance-Based Fault Location
Experience, proceedings of the 31st Annual Western Protective Relay
Conference, Spokane, WA, October 2004.
[10] B. S. Guru and H. R. Hiziroglu, Electromagnetic Field Theory
Fundamentals. PWS Publishing Company, Boston, MA, June 1997.
[11] P. Moreno, P. Gmez, J. L. Naredo, and J. L. Guardado, Frequency
Domain Transient Analysis of Electrical Networks Including Non-linear
Conditions, International Journal of Electrical Power & Energy
Systems, Vol. 27, Issue 2, February 2005, pp. 139146.
[12] D. E. Hedman, Propagation on Overhead Transmission Lines I: Theory
of Modal Analysis, IEEE Transactions on Power Apparatus and
Systems, Vol. 84, Issue 3, March 1965, pp. 200205.
[13] S. A. Schelkunoff, The Electromagnetic Theory of Coaxial
Transmission Lines and Cylindrical Shields, The Bell System Technical
Journal, Vol. 13, No. 4, October 1934, pp. 532579.
[14] D. Hou and J. Roberts, Capacitive Voltage Transformers: Transient
Overreach Concerns and Solutions for Distance Relaying, proceedings
of the 22nd Annual Western Protective Relay Conference, Spokane,
WA, October 1995.
[15] E. O. Schweitzer, III, Evaluation and Development of Transmission
Line Fault-Locating Techniques Which Use Sinusoidal Steady-State
Information, proceedings of the 9th Annual Western Protective Relay
Conference, Spokane, WA, October 1982.
[16] T. Takagi, Y. Yamakoshi, M. Yamaura, R. Kondow, and
T. Matsushima, Development of a New Type Fault Locator Using the
One-Terminal Voltage and Current Data, IEEE Transactions on Power
Apparatus and Systems, Vol. PAS101, Issue 8, August 1982,
pp. 28922898.
[17] B. Kasztenny, B. Le, and N. Fischer, A New Multiterminal Fault
Location Algorithm Embedded in Line Current Differential Relays,
proceedings of the 11th International Conference on Developments in
Power System Protection, Birmingham, UK, April 2012.
[18] H. Lee, Development of an Accurate Travelling Wave Fault Locator
Using the Global Positioning System Satellites, proceedings of the 20th
Annual Western Protective Relay Conference, Spokane, WA,
October 1993.
[19] D. A. Douglass, Current Transformer Accuracy With Asymmetric and
High Frequency Fault Current, IEEE Transactions on Power Apparatus
and Systems, Vol. 100, Issue 3, March 1981, pp. 10061012.
[20] M. A. Redfern, S. C. Terry, F. V. P. Robinson, and Z. Q. Bo, A
Laboratory Investigation Into the use of MV Current Transformers for
Transient Based Protection, proceedings of the 2003 International
Conference on Power Systems Transients (IPST), New Orleans, LA,
SeptemberOctober 2003.
[21] A. M. Elhaffar, Power Transmission Line Fault Location Based on
Current Traveling Waves, doctoral dissertation, Helsinki University of
Technology, Finland, March 2008.
[22] D. J. Spoor, J. Zhu, and P. Nichols, Filtering Effects of Substation
Secondary Circuits on Power System Traveling Wave Transients,
proceedings of the 8th International Conference on Electrical Machines
and Systems (ICEMS), Vol. 3, September 2005, pp. 23602365.
[23] M. Kezunovic, L. Kojovic, V. Skendzic, C. W. Fromen, D. Sevcik, and
S. L. Nilsson, Digital Models of Coupling Capacitor Voltage
Transformers for Protective Relay Studies, IEEE Transactions on
Power Delivery, Vol. 7, Issue 4, October 1992, pp. 19271935.
[24] L. V. Bewley, Traveling Waves on Transmission Systems. Dover
Publications, Mineola, NY, 1963.
IX. BIOGRAPHIES
Stephen Marx received his BSEE from the University of Utah in 1988. He
joined Bonneville Power Administration (BPA) in 1988. He is presently the
District Engineer in Idaho Falls, Idaho, for BPA. He has over 25 years of
experience in power system protection and metering. He has been a lecturer at
the Hands-On Relay School in Pullman, Washington, since 2007. He is a
registered professional engineer in the state of Oregon. He is a member of
IEEE and has authored and coauthored several technical papers.
Brian K. Johnson received a Ph.D. degree in electrical engineering from the
University of Wisconsin-Madison in 1992. He is currently a professor in the
ECE Department at the University of Idaho, Moscow. His interests include
power electronics, power system protection, and modeling and simulation of
power system transients. Dr. Johnson is a professional engineer in the state of
Idaho.
Armando Guzmn received his BSEE with honors from Guadalajara
Autonomous University (UAG), Mexico. He received a diploma in fiber-
optics engineering from Monterrey Institute of Technology and Advanced
Studies (ITESM), Mexico, and his MSEE and MECE from the University of
Idaho, USA. He served as regional supervisor of the Protection Department in
the Western Transmission Region of the Federal Electricity Commission (the
Mexican electrical utility company) in Guadalajara, Mexico, for 13 years. He
lectured at UAG and the University of Idaho in power system protection and
power system stability. Since 1993, he has been with Schweitzer Engineering
Laboratories, Inc., in Pullman, Washington, where he is a fellow research
engineer. He holds numerous patents in power system protection and
metering. He is a senior member of IEEE.
Veselin Skendzic is a principal research engineer at Schweitzer Engineering
Laboratories, Inc. He earned his BS in electrical engineering from FESB,
University of Split, Croatia; his Masters of Science from ETF, Zagreb,
Croatia; and his Ph.D. from Texas A&M University. He has more than
25 years of experience in electronic circuit design and power system
protection-related problems. He is a senior member of IEEE, has written
multiple technical papers, and is actively contributing to IEEE and IEC
standard development. He is a member of the IEEE Power Engineering
Society (PES) and the IEEE Power System Relaying Committee (PSRC) and
a past chair of the PSRC Relay Communications Subcommittee (H).
Mangapathirao (Venkat) Mynam received his MSEE from the University
of Idaho in 2003 and his BE in electrical and electronics engineering from
Andhra University College of Engineering, India, in 2000. He joined
Schweitzer Engineering Laboratories, Inc. (SEL) in 2003 as an associate
protection engineer in the engineering services division. He is presently
working as a lead research engineer in SEL research and development. He
was selected to participate in the U.S. National Academy of Engineering
(NAE) 15th Annual U.S. Frontiers of Engineering Symposium. He is a senior
member of IEEE.
2013 by Bonneville Power Administration, University of Idaho, and
Schweitzer Engineering Laboratories, Inc.
All rights reserved.
20130309 TP6601-01