Double-Ended Traveling-Wave Fault Locating

Without Relay-to-Relay Communications

David López Cortón and Jorge Vaquero Melado

Red Eléctrica de España

Jesús Cruz, Richard Kirby, Yusuf Zafer Korkmaz, Gianfranco Patti, and Greg Smelich
Schweitzer Engineering Laboratories, Inc.

Presented at the
25th Annual Georgia Tech Fault and Disturbance Analysis Conference
Atlanta, Georgia
May 2–3, 2022

Revised edition released July 2021

Originally presented at the

74th Annual Conference for Protective Relay Engineers, March 2021
 1

Double-Ended Traveling-Wave Fault Locating

Without Relay-to-Relay Communications
David López Cortón and Jorge Vaquero Melado, Red Eléctrica de España
Jesús Cruz, Richard Kirby, Yusuf Zafer Korkmaz, Gianfranco Patti, and Greg Smelich,
Schweitzer Engineering Laboratories, Inc.

Abstract—This paper discusses how Red Eléctrica de España, calculations from time-stamped information contained in the
the transmission system operator in Spain, successfully applied event records and an automated process where the relays send
double-ended traveling-wave-based fault locating (DETWFL) on the time-stamped information to a central device via the DNP3
a pilot project where relay-to-relay communications were
unavailable. The offline DETWFL methodology uses traveling- protocol over Ethernet. The paper also describes the use of
wave arrival time information that is manually collected from event analysis software that is capable of analyzing TWs and
time-synchronized relays at both ends of the line after an internal estimating the fault location using TWs.
fault occurs. The paper also discusses the possibility of using the Red Eléctrica de España (REE) is the transmission grid
DNP3 protocol over Ethernet to retrieve event information, owner and operator for the transmission system in Spain. With
allowing for automatic DETWFL calculations offline. The paper more than 44,000 km of high-voltage transmission lines, REE
includes an analysis of a B-phase-to-ground fault on the line to
illustrate the performance of the applied offline DETWFL is responsible for ensuring the proper operation of the Spanish
method. electrical system, coordinating the generation transport system,
and planning the long-term growth of the transmission network
I. INTRODUCTION to provide secure and continuous power supply to the country.
The double-ended traveling-wave-based fault-locating With the intent to explore and gain experience with
(DETWFL) method using current traveling waves (TWs) was TW-based fault-locating technology (particularly, improved
first introduced in a transmission line protective relay in 2012. accuracy of the results and ability to adaptively control
Since then, the DETWFL method has also been made available autoreclosing on hybrid lines), REE decided to install several
in ultra-high-speed (UHS) relays. TW-based fault locating is UHS relays in their transmission system. In the early stages of
widely popular with transmission system operators, largely due their evaluation, relay-to-relay communications were not yet in
to its field-proven track record with reported errors being within place on one of the transmission lines where the UHS relays
one tower span (300 m; 1,000 ft) on average, regardless of line were installed. However, the relays were connected to high-
accuracy Global Positioning System (GPS) clocks, allowing
length. The DETWFL method incorporated in UHS relays
provides accurate results, but it requires TW data from the two event data to be time-synchronized and enabling the ability to
use the offline DETWFL methods.
line terminals. When relay-to-relay communications are
available, the UHS relay can collect the necessary time stamps A detailed event analysis from a B-phase-to-ground fault on
of the initial TW that arrives at each terminal when a fault the 220 kV overhead line between the Casaquemada and Onuba
occurs, automatically calculate the fault location using the terminals in the REE transmission network serves as an
DETWFL method, and make the result available to the user example to show the application of the different offline
within tens of milliseconds. As described in [1], relay-to-relay DETWFL strategies explained in this paper. A summary of
communications can be achieved by using a dedicated point-to- results from five faults that happened on this line while
point fiber-optic channel (i.e., direct fiber) or an IEEE C37.94 relay-to-relay communications were not available is also
multiplexed channel. However, there are situations where provided. These results highlight the improved accuracy of the
relay-to-relay communications may not be available. In this DETWFL method as compared to the single-ended impedance-
case, accurate fault location results can still be obtained by based fault-locating (SEZFL) method and the single-ended
traveling-wave-based fault-locating (SETWFL) method.
using the offline DETWFL methodology described in this
paper.
II. BACKGROUND
The availability of a precise time reference applied to both
line terminals allows for accurate time-stamping of the wave A. Project Drivers
arrival information and enables calculations using information In 2019, REE launched the Traveling-Wave Fault
from the two line terminals. This time-stamping process is Identification System (referred to as SIFOV) project to test new
summarized in Section III.C and is critical for implementing technologies and developments that could provide
analysis and calculations. The offline DETWFL methodologies improvements in two main areas:
described in Section V require the extraction of TW time • Accurate fault locating
stamps from the UHS relays using one of the methods described • Adaptive autoreclosing for hybrid lines
in Section IV. These extraction methods include manual
 2

Autoreclosing is one of the main functions included in 2. Laboratory tests. Two UHS relays were installed in
REE’s line protection standard. Based on REE’s philosophy, REE’s laboratory to carry out intensive hardware-in-
the autoreclose function is activated in overhead transmission the-loop tests of the equipment. Scenarios like the
lines, in single- or three-pole mode, depending on the voltage pilot field installations were modeled in a real-time
level and the proximity with generation. In underground cables, digital simulator (RTDS) to analyze the performance
the autoreclosing is not in service, as faults tend to be of the equipment in a wide variety of conditions.
permanent. For hybrid lines, which contain both overhead line
B. Transmission Lines Used for DETWFL Evaluation
and underground cable sections, autoreclosing is active for
those cases in which the protection system can determine that This section describes the locations selected for the project.
the fault occurred in an overhead section. 1) N. Valladolid-to-Mudarra Line, 220 kV
Presently, the REE design for adaptive autoreclosing on The system was commissioned in November 2019. The
hybrid lines is based on the operation of available distance circuit is a hybrid line with an overhead section of 21.85 km
protection zones with a reach set to block autoreclosing on (Mudarra) and an underground section of 2.42 km
underground sections or to allow autoreclosing on overhead (N. Valladolid). Line energization tests were performed during
sections. Since the distance protection is single-ended and commissioning, and the measured TW propagation times
operates based on the measured apparent impedance to the fault (TWLPT) for the overhead and underground sections were
location, the usual security margins considered are around determined to be 74.5 µs and 21.8 µs, respectively.
20 percent. Depending on the lengths of the overhead and Relay-to-relay communications have been available on this
underground sections, these security margins may result in line since the time of commissioning, allowing REE to evaluate
large overhead portions in which autoreclosing is blocked the adaptive autoreclose cancel logic described in [2] using the
unnecessarily, only to ensure dependability for faults in the UHS relays. When writing this paper, this feature had not yet
underground section. been implemented. The telecommunications infrastructure
The continuous growth of cities, along with the opposition deployed for this line is described in the Appendix,
by some to electrical facilities, has caused hybrid lines to Section VIII.A.
become a common solution. In Spain, hybrid lines can consist
of several overhead and underground sections, which further 2) Puerto de la Cruz: Spain-to-Morocco 1 Line
challenges the technology required to distinguish between and Spain-to-Morocco 2 Line, 400 kV
In this pilot project, the two transmission lines
sections.
interconnecting Spain and Morocco were selected. Each
To achieve the two objectives of the project, REE needed an
interconnection consists of a hybrid line that has an overhead
accurate fault-locating system able to provide an adequate
section on the Spanish end, an overhead section on the
location in less than 250 ms. This timing requirement was not
Moroccan end, and a submarine cable section between them.
fixed by the fault location indication itself, but rather by the
The Spanish overhead section has a total length of 9.33 km, and
autoreclose blocking logic, since the minimum time for the first
the Moroccan overhead section is 22.21 km. The approximate
autoreclosing attempt in the REE system is 400 ms. Reference
length of the submarine cable is 31.3 km. When writing this
[2] explains the method used to accomplish this with UHS
paper, the UHS relays were installed only on the Spanish side
relays, which provides the ability to allow or inhibit
of each line.
autoreclosing by using adaptive autoreclose cancel logic based
The first circuit was commissioned in March of 2020, and
on the accurate real-time DETWFL result. This function
the second circuit was commissioned in February of 2021. Line
requires relay-to-relay communications.
energization tests were performed during commissioning, and
REE assessed various alternatives, and finally they decided
the measured TWLPT for the Spanish overhead section was
that the UHS relays were the best equipment for the project due
determined to be 31.93 µs. The propagation times of the
to the four fault-locating methods (single- and double-ended
submarine cable and the Moroccan overhead section could not
TW- and impedance-based fault locating: SETWFL,
be deduced from the line energization event records, and they
DETWFL, SEZFL, and DEZFL, respectively) running in
are still under study. The TWLPT values used in the UHS relay
parallel and the promising developments on TWFL included in
settings for these sections were approximated using software
the relay, such as the ability to control the line autoreclosing
simulation tools and may be modified later when additional
relay to cancel reclosing for faults on underground cable
field data becomes available. The TWLPT values for the
sections of a line while still allowing reclosing on overhead
submarine cable section and Moroccan overhead section are
sections.
236.06 µs and 72.5 µs, respectively.
The project was structured in two stages running in parallel:
1. Pilot field installation. In this stage, REE selected 3) Casaquemada-to-Onuba Line, 220 kV
various facilities representative of the applications in The system was commissioned in November 2019. The
which the solution could be applied. The system was circuit is a 61.98 km (38.51 mi) overhead line. Line
commissioned, and its performance was monitored. energization tests were performed during commissioning, and
During this period, telecommunication, maintenance, the measured TWLPT was determined to be 210.5 µs.
and design engineers received training in order to gain Due to various delays in the deployment of the
knowledge about the technology. telecommunications infrastructure, this system operated
 3

without communications for several months. During this LL  ( t L − TWCPTL ) − ( t R − TWCPTR ) 

period, five faults occurred on the line, which led to the offline M
= 1 +  (2)
2  TWLPT 
DETWFL calculations described in this paper. Section VI
shows the offline analysis performed for one of these faults. When relay-to-relay communications are obtained through a
The telecommunications infrastructure that was later deployed direct fiber-optic channel, the UHS relays synchronize their
for this line is described in the Appendix, Section VIII.B. internal clocks to each other, providing a common time
reference and making the operation of the DETWFL method
III. DOUBLE-ENDED TRAVELING-WAVE-BASED independent of an external time source.
FAULT LOCATING When relay-to-relay communications are obtained through a
A. Principle of Operation multiplexed fiber-optic channel, such as when using an
IEEE C37.94-compliant synchronous optical networking
Fig. 1 shows a Bewley diagram, which is a time-spatial chart
(SONET)/synchronous digital hierarchy (SDH) multiplexer,
that shows TWs progressing along the time axis (vertically
each UHS relay must be synchronized to an absolute time
down) and simultaneously progressing along the distance axis
reference by connecting the IRIG-B input to an external
(left to right and right to left), for a fault at location F on a line
IEEE C37.118-compliant clock with submicrosecond accuracy.
of length LL. The fault is M (km or mi) from the local terminal,
Reference [2] provides an accuracy analysis of the
L, and LL – M (km or mi) from the remote terminal, R. Faults
DETWFL method with respect to the LL, TWLPT, and
launch TWs that propagate with a velocity (PV) equal to LL
TWCPT settings, as well as suggestions on how to improve the
divided by TWLPT (i.e., PV = LL/TWLPT) [1].
accuracy of these settings.
B. Mode and Phase Reference Selection
UHS relays use the measured current signals to obtain the
phase TWs (TWA, TWB, and TWC), obtain the modal TWs
(zero, alpha, and beta Clarke components), and apply them
accordingly in various functions. The alpha components are
appropriate for analyzing TWs launched by single-line-to-
ground faults and the beta components for line-to-line faults.
The modal signals are calculated as follows:
Fig. 1. Bewley diagram explaining the DETWFL method. • The ground mode is calculated as
When a fault occurs at time tFAULT = 0, the first TW arrives TW0 = (TWA + TWB + TWC)/3.
at terminal L at time tL = M/PV = M • TWLPT/LL. Similarly, • The alpha aerial mode referenced to phase A is
the first TW arrives at terminal R at time tR = (LL – M)/PV = calculated as TWA – TW0. This mode is typically the
(LL – M) • TWLPT/LL. Solving these two equations for fault highest and most reliable for A-phase-to-ground faults
location, M, we obtain the general equation used for DETWFL on overhead transmission lines. Similar equations are
as (1). used to calculate the phase B and phase C referenced
alpha modes.
LL  tL − tR 
M
= 1 +  (1) • The beta aerial mode referenced to phases A and B is
2  TWLPT  calculated as (TWA – TWB)/√3. This mode is
The TWLPT for the transmission line can be measured typically the highest and most reliable for A-phase-to-
during commissioning by performing a line energization test. B-phase (AB) faults on overhead transmission lines.
When the propagation time through current transformer Similar equations are used to calculate the BC- and
(CT) secondary cables is significantly different at each line CA-referenced beta modes.
terminal, the accuracy of this method can be affected. C. Time-Stamping the Arrival Time of TWs
Assuming the TW propagation velocity for cables is 70 percent
Since the accuracy of the calculated fault location from (1)
of the speed of light in free space, 50 m of difference in cable
or (2) is highly dependent on TW arrival times with
length between line terminals may result in 36 m of fault
submicrosecond resolution, the method the fault locator applies
location error for the DETWFL method on an overhead line [1].
to estimate the TW arrival time is critical to its accuracy.
To increase accuracy, the fault locator compensates for the
As [1] describes, a differentiator-smoother filter provides a
time delay associated with the cables between the CTs and the
simple and robust method for extracting the TWs from the input
relay by backdating the time stamp of the initial TW through
signals sampled at 1 MHz. The differentiator-smoother filter
use of the TW cable propagation time (TWCPT) setting. If the
responds to an ideal step change in the input signal with a
TWCPT setting for each terminal is known (TWCPTL for
triangle-shaped output, and it responds to a ramp transition
Terminal L and TWCPTR for Terminal R), they can be used to
between two levels with a parabola-shaped output, as Fig. 2
compensate for the variation in CT lead length by using (2),
shows.
which is a modification of (1).
 4

A. IEEE COMTRADE Header Files

The UHS relay provides a transient recording functionality
with two types of records:
• Ultra-high-resolution record containing voltages and
currents (megahertz record [MHR], 1 MHz sampling)
• High-resolution record containing voltages and
currents, derived protection quantities, and all digital
bits (time-domain record [TDR], 10 kHz sampling)
Both types of records are stored in IEEE C37.111-2013
COMTRADE format. Per this format, both the MHR and TDR
IEEE COMTRADE records comprise three files: a
configuration (CFG), data (DAT), and header (HDR) file,
where the three-letter abbreviations serve as the file extension
type. The CFG file describes the content of the DAT file, the
DAT file contains the values for each input channel for each
sample in the record, and the HDR file contains relay settings
and event-related analog quantities (such as prefault and fault
voltages and currents, fault type and location, etc.) helpful in
analyzing power system events and relay operation [1].
Fig. 4 shows the portion of the HDR file that contains fault
Fig. 2. Response of a differentiator-smoother filter to (a) an ideal step, and location information. This information is available in the HDR
(b) a ramp.
file for both the MHR and TDR IEEE COMTRADE records. If
To get a better time resolution, which translates to better the UHS relay can estimate the fault location using one of the
fault location accuracy, the TW arrival times can be estimated available methods (SETWFL, DETWFL, SEZFL, and DEZFL,
using interpolation between samples, fitting a parabola to the with as many as four possible alternatives for the SETWFL
output of the differentiator-smoother filter using the least- method), the results are populated in the respective fields.
squared errors method. The time of the peak of the best-fitting
parabola, as shown in Fig. 3, is the TW arrival time and
provides a time-stamping accuracy of approximately 0.1 μs
[1] [3].

Fig. 4. Portion of the IEEE COMTRADE header file that contains fault
location information.

The fields in Fig. 4 are described as follows:

• SE_TW_Locationn (n = 1, 2, 3, 4) are the fault
locations from the SETWFL method.
• DE_TW_Location is the fault location from the
DETWFL method.
• SE_Z-Based_Location is the fault location from the
SEZFL method.
• DE_Z-Based_Location is the fault location from the
DEZFL method.
Fig. 3. Using interpolation to find the time of the peak with an accuracy that • First_TW_Time_Local is the time stamp of the first
is better than one sampling interval. local TW, compensated for by the CT cable delay.
• First_TW_Time_Remote is the time stamp of the first
IV. EXTRACTING TW TIME STAMPS remote TW, compensated for by the CT cable delay.
Section V.A describes how the DETWFL results may be Without relay-to-relay communications, the time stamp of
obtained offline either by using a manual process of obtaining the first remote TW is unavailable to the local relay, and,
the necessary information and performing calculations or by therefore, the result from the DETWFL method cannot be
implementing an automated method for collecting the data and automatically calculated by the relay. Regardless of the
calculating the results in a central device. In either case, the availability of relay-to-relay communications, the arrival time
time stamp of the initial TW that arrives at each terminal of the of the initial TW at the local terminal is First_TW_Time_Local.
line when the fault occurs must be obtained. This section This time stamp is interpolated using the method described in
describes several methods to obtain the TW time stamps. Section III.C. It is also compensated for the CT cable delay [1].
 5

Section V.A explains how to calculate the DETWFL results LL and TWLPT values. Otherwise, the algorithm can exit
offline by obtaining First_TW_Time_Local from the HDR file without returning a fault location result.
in both the local and remote relays and using the time stamps Fig. 5 shows a flowchart depicting the proposed method for
in (1). automating retrieval of TW time stamps via DNP3 and
performing offline DETWFL calculations in a central device.
B. DNP3 LAN/WAN Over Ethernet
It is possible to automate the manual DETWFL calculations
discussed in this paper using DNP3 with an optional remote
terminal unit (RTU) in the substation that polls the respective
UHS relay. This section describes how to obtain the time stamp
of the initial TW using DNP3 LAN/WAN over Ethernet. It also
explains how to use these time stamps in (1) to automatically
calculate the DETWFL results in a central device. Since
information must be available from both terminals of the
transmission line, the scheme must check the communication
status and the availability and quality of data from both
terminals. It must also check the time-synchronization status of
both UHS relays providing input data for the calculation. The
automated method must check that data from both terminals are
for the same fault event and consistent in terms of time stamps.
The following information can be employed in a DNP3
register map in the UHS relays and sent to a client system
through DNP3 communications.
• Analog input objects:
− First local TW arrival time in DNP3 format
truncated to the second (FILTWH, FILTWM,
FILTWL, which are the high, middle, and low
16 bits, respectively)
− First local TW arrival time, milliseconds portion
(FILTWMS)
− First local TW arrival time, microseconds portion
Fig. 5. Flowchart representing proposed offline DETWFL method
(FILTWUS) using DNP3.
− First local TW arrival time, nanoseconds portion
(FILTWNS) The fault location can be calculated using the arrival time
• Binary input object: Relay clock synchronized with difference of the first TW at each end as follows:
high accuracy to the absolute time (TSOK) 1. Construct the date-time object for the local and remote
The analog input objects must be set without any deadband relay using FILTWH, FILTWM, FILTWL
so that any data change is reported as an analog input event. (e.g., TW_TIME_S_L, TW_TIME_S_R).
A DNP3 client can then poll the UHS relays (or optionally These quantities are truncated to the second.
the RTUs which are polling the UHS relays) and collect these 2. Compute the date-time difference in seconds (e.g.,
data objects from both ends of the transmission line. The DNP3 TW_DIFF_S = TW_TIME_S_L – TW_TIME_S_R).
client must also store the LL and TWLPT values of the 3. If the absolute value of the date-time difference is
transmission line as local variables. greater than 1 second, exit without producing a result.
Using this information, the DNP3 client can implement an 4. Otherwise, compute the first TW arrival time for each
algorithm that can be triggered when the analog objects relay in nanoseconds (e.g., TW_TIME_NS_L and
reported by both relays have good data quality, their time TW_TIME_NS_R) as FILTWMS • 106 + FILTWUS •
stamps have changed, and the time stamps are coherent (i.e., the 103 + FILTWNS.
difference between the time of fault reported by each UHS relay 5. Compute the first TW arrival time difference in
is less than one second). nanoseconds, including the date-time difference
Using the binary input object information that represents the (e.g., TW_DIFF_NS = TW_DIFF_S • 109 +
UHS relay clock synchronization status, the algorithm should (TW_TIME_NS_L – TW_TIME_NS_R))
check that both relays are time-synchronized at the time of the 6. If the absolute value of the arrival time difference in
fault event. If this condition is also satisfied, then the time nanoseconds is greater than TWLPT in nanoseconds
difference between the two first wave arrival times can be (e.g., TWLPT_NS), exit without producing a result.
computed. If the computed time difference is smaller than the 7. Otherwise, compute the fault location with respect to
TWLPT value, the algorithm can calculate the fault location the local terminal (e.g., DETWFL_L = LL • 0.5 •
using this time difference and the local variables that store the (1+ TW_DIFF_NS/TWLPT_NS)).
 6

C. Transient Records With 1 MHz Sampling initial TW peak at the local and remote ends, the DETWFL
The UHS relay stores ultra-high-resolution transient records results are automatically displayed in an information pane
(MHR IEEE COMTRADE record, 1 MHz sampling) that within the software tool. Section VI.A provides an example of
contain voltage and current samples with 18 bits of resolution how REE used this method to verify the DETWFL results for
and an effective measurement bandwidth of about 400 kHz. an internal BG fault in their system and includes the Bewley
Event analysis software is available that allows the user to open diagram used for the analysis (see Fig. 10).
the MHR IEEE COMTRADE record and plot the voltage and
current signals. The software makes available the modal TW VI. FIELD EXPERIENCE
signals (ground, alpha, and beta Clarke components), which are In December 2019, the 220 kV, 61.98 km line between the
described in Section III.B. The software also provides time Casaquemada and Onuba terminals (described in
cursors that replicate the interpolation method described in Section II.B.3) did not have relay-to-relay communications in
Section III.C. Therefore, the time stamp of the initial TW at place, making the DETWFL results unavailable for automatic
each terminal may be obtained by opening the corresponding calculation by the UHS relays. However, both UHS relays were
MHR IEEE COMTRADE record in the event analysis connected to a high-accuracy IRIG-B time source.
software, plotting the appropriate modal TW signal for the fault
A. Internal BG Fault, December 8, 2019
type, and sliding the time cursor to line up with the peak of the
initial TW. It is important to note that this time stamp does not For this fault event, the pair of UHS relays detected an
include compensation for the TW CT cable delay. Fig. 6 shows internal B-phase-to-ground fault while monitoring the line.
an example of the signals available in the MHR IEEE Since the relays were synchronized to absolute time, the
COMTRADE record from the UHS relay for an internal BG DETWFL results could be obtained using the offline methods
fault in the REE system, and Fig. 9 shows how the alpha-mode described in Section V. REE calculated the DETWFL result for
B-phase TW signals may be used to obtain the time stamp of the Casaquemada terminal to be 27.045 km. This was
the initial TW that arrived at each end of the line. corroborated by the SEZFL and SETWFL results, which were
24.220 km and 26.728 km, respectively.
V. OFFLINE METHODOLOGY The event report data in Fig. 6 were used to evaluate the
UHS relay line protection, which uses elements and schemes
There are two general approaches to the methodology for
based on incremental quantities and TWs [1]. Fig. 6 shows the
offline calculations of the DETWFL result: manual calculations
voltage and current signals sampled at 1 MHz by the UHS relay
and software tools.
at the Casaquemada terminal. Fig. 6 also shows the
A. Manual Calculations performance of the incremental-quantity directional forward
After the necessary TW time stamps are obtained by using (TD32F) element and incremental-quantity ground distance
one of the methods described in Section IV, they can be applied (TD21G) element, which operated in 1.29 ms and 9.09 ms,
(along with the LL and TWLPT settings) to (1). If time stamps respectively. In the UHS relay at the Onuba terminal, the
are retrieved from the First_TW_Time_Local entry in the IEEE TD32F and TD21G elements operated in 1.29 ms and 9.69 ms,
COMTRADE header file from each end (Section IV.A) or by respectively.
using the DNP3 protocol over Ethernet (Section IV.B), the time As described in Section IV.A, the HDR file of the TDR and
stamps do not need to be compensated for by the CT cable delay MHR IEEE COMTRADE records from the UHS relay contains
because the UHS relay has already applied the compensation. fault location results for all four available methods. Fig. 7
However, if the time stamps are obtained by using the MHR shows the fault location results from the UHS relay at
IEEE COMTRADE records (Section IV.C), the compensation Casaquemada, and Fig. 8 shows the results from the UHS relay
must be applied for the CT cable delay. at Onuba.
REE calculated the DETWFL results manually using data
B. Software Tools: Bewley Diagram available in the event records from both terminals. The
The event analysis software described in Section IV.C also First_TW_Time_Local value in Fig. 7 and Fig. 8 indicates the
provides the ability to plot and analyze Bewley diagrams. arrival time of the initial TW at each end, respectively. These
DETWFL results may be obtained offline by retrieving the arrival times are compensated for by the TWCPT setting in each
MHR IEEE COMTRADE record from each end of the line, relay. Comparing these arrival times shows that the UHS relay
opening them in a single session of the event analysis software, at Casaquemada was the first to detect a TW, which indicates
and plotting the appropriate modal TW signals in the Bewley the fault was closer to that terminal. Furthermore, the first TW
diagram. After the time cursors are manually aligned to the arrived at Casaquemada 26.798 µs before the first TW arrived
at Onuba.
 7

Using (1), REE calculated the DETWFL result for

Casaquemada to be 27.045 km, as shown in (3). Note that tL – tR
is negative in (3) because the fault was closer to Casaquemada.
61.98  −26.798 
=M 1 + 
2  210.50  (3)
M = 27.045 km
Similarly, REE used (1) to calculate the fault location from
the Onuba terminal. Since the fault was closer to Casaquemada,
tL – tR is a positive number when performing the calculation
with respect to Onuba, and the result is 34.935 km.
Fig. 9 shows the B-phase alpha-mode current TWs captured
by the UHS relays (available in the MHR IEEE COMTRADE
record) at the local and remote terminals, Casaquemada (black)
and Onuba (blue), respectively, and it displays the relative
arrival time difference between the first TWs at each end
(–27.037 µs). For this application, the TWCPT setting at
Casaquemada and Onuba is 0.238 µs and 0.477 µs,
respectively. When compensation for TWCPT is applied to the
Fig. 6. Casaquemada UHS relay event record data showing a B-phase fault.
time stamps determined in Fig. 9, the same arrival time
difference used in (3) is obtained: –27.037 – (0.238 – 0.477) µs
= –26.798 µs. Fig. 9 also shows that both TWs are in phase
(i.e., both have negative polarity), as we would expect for an
internal (on-the-line) event or fault.
The Bewley diagram in Fig. 10 shows the time and distance
relationship of the measured B-phase alpha-mode current TWs
for the local and remote terminals, Casaquemada (black) and
Fig. 7. Fault location information in the HDR file from the UHS relay at
Casaquemada. Onuba (blue), respectively, and it shows the first TW that
arrived at each terminal, along with the subsequent reflections.
This view is obtained after aligning the time cursors to the
initial TW peaks for each end (green and red, respectively). The
fault location provided by the software in the information pane
on the left side of Fig. 10 matches the result of the manual
calculations shown in (3).

Fig. 8. Fault location information in the HDR file from the UHS relay at
Onuba.

Fig. 9. B-phase alpha-mode current TWs captured by UHS relays (available in the MHR IEEE COMTRADE record) at the local and remote terminals,
Casaquemada (L, black) and Onuba (R, blue), respectively, including the arrival times of first TWs at each end.
 8

Fig. 10. Bewley diagram showing B-phase alpha-mode current TWs for the local and remote terminals, Casaquemada (L, black) and Onuba (R, blue),
respectively.

The offline DETWFL results calculated by REE differed Table I shows that the offline DETWFL method provided
from the SETWFL results in Fig. 7 and Fig. 8 by 317 m results that were accurate to within one tower span for all five
(1,040 ft) and 180 m (591 ft), respectively. The SETWFL and faults that occurred on this line while relay-to-relay
SEZFL results corroborated the DETWFL results REE communications were unavailable. Having these accurate
obtained from the Bewley diagram and by manual calculation. results allowed line crews to quickly identify the cause of the
Using the results of this analysis, REE’s maintenance team fault and restore the line to service. Fig. 11 shows an example
inspected the circuit and found evidence of the fault at a tower in which evidence of the December 17, 2019 fault was found,
located 26.943 km from Casaquemada (35.009 km from confirming the fault location and showing that bird
Onuba). Identification of the true fault location confirmed that contamination was the likely cause.
the results obtained from the Bewley diagram in Fig. 10 and
from manual calculations were within one tower span from
either end: 102 m from Casaquemada and 74 m from Onuba.
B. Summary of Results
A summary of five fault events, including the event
described in Section VI.A, is provided in Table I. It includes
fault location results on the Casaquemada-to-Onuba line while
relay-to-relay communications were not available. The
tabulated results include the following:
• SEZFL, provided by the distance relays protecting the
line.
• SETWFL, provided by the UHS relays monitoring the
line.
• DETWFL, obtained from UHS relays using the offline
calculation methods explained in this paper.
• The location of the tower where the fault was
Fig. 11. Evidence found of the December 17, 2019, fault, showing bird
confirmed. contamination and arcing across the insulators.
 9

TABLE I
SUMMARY OF FAULT LOCATION RESULTS FOR THE 220 KV, 61.98 KM OVERHEAD TRANSMISSION LINE BETWEEN THE CASAQUEMADA AND ONUBA TERMINALS

Event Fault Terminal SEZFL SETWFL DETWFL Tower

Date Type Location
Result (km) Error (km) Result (km) Error (km) Result (km) Error (km) (km)

Nov 29, Casaquemada 30.573 3.294 33.449 0.418 33.872 0.005 33.867
BG
2019 Onuba 25.332 2.753 28.212 0.127 28.108 0.023 28.085

Dec 08, Casaquemada 24.220 2.723 26.728 0.215 27.045 0.102 26.943
BG
2019 Onuba 32.081 2.928 34.755 0.254 34.935 0.074 35.009

Dec 17, Casaquemada 25.922 3.115 28.758 0.279 29.138 0.101 29.036
BG
2019 Onuba 29.825 3.091 32.714 0.202 32.842 0.074 32.916

Dec 26, Casaquemada 55.398 4.033 33.153 26.278 59.210 0.221 59.431
BG
2019 Onuba 2.154 0.367 2.505 0.016 2.770 0.249 2.521

Jan 01, Casaquemada 30.447 3.420 33.343 0.524 33.908 0.041 33.867
BG
2020 Onuba 25.162 2.923 27.918 0.167 28.072 0.013 28.085

VII. CONCLUSION VIII. APPENDIX

UHS transmission line protective relays include built-in Obtaining successful results from the DETWFL method,
TW-based fault-locating methods that are accurate to within a including when relay-to-relay communications were not
tower span (300 m; 1,000 ft), regardless of the line length. available, has motivated REE to continue evaluating UHS
Having accurate fault location results allows for improved relays based on the project drivers described in Section II.A.
system operations by reducing patrol and repair time, and they The ability of the UHS relay to automatically provide
help minimize costs by aiding in the detection of problematic DETWFL results and allow for adaptive autoreclosing control
areas of the line so reoccurring faults can be reduced. requires that relay-to-relay communications be available. This
Since 2019, REE has been gaining experience with UHS section describes the telecommunications infrastructure that is
relays for TW-based fault locating and adaptive autoreclosing now in place for the N. Valladolid-to-Mudarra line and
on hybrid lines with relay-to-relay communications. This paper Casaquemada-to-Onuba line.
presents REE’s experience with performing offline calculations
A. Telecommunications Infrastructure for the N. Valladolid-
using the DETWFL method when relay-to-relay
to-Mudarra Line, 220 kV, 24.27 km
communications were not yet available on one of the lines
where the new technology was being evaluated. Analysis of the Fig. 12 shows the telecommunications infrastructure
results verifies that the accuracy was within one tower span for deployed for this line. To make an efficient use of the available
this line, providing confidence for REE to continue moving fiber optics and communications equipment, REE decided to
forward with the SIFOV project. use passive optical multiplexers. For the length of this line,
Methods for obtaining DETWFL results using offline passive multiplexers were sufficient to satisfy the distance
calculations and software tools (such as the Bewley diagram limitations of the SDH equipment. These multiplexers receive
available in event analysis software) are described, as well as two signals:
requirements for time-stamping TWs and options for obtaining 1. A 1,550 nm wavelength fiber-optic signal from the
the time stamps using manual and automated protocol-based direct-fiber port of the UHS relay.
methods. 2. A 1,310 nm wavelength fiber-optic signal from the
Fault location data recorded during five faults on an REE SDH equipment connected to a multiplexed-fiber port
transmission line confirmed that DETWFL results could be on the UHS relay.
obtained using an offline methodology when relay-to-relay The passive optical multiplexers introduce both signals into
communications were not available, providing REE with a single fiber-optic channel for transmission to the other
results that were accurate to within a tower span. terminal. The demultiplexers at both sides decouple the
The offline calculated DETWFL result can be used to received signals and distribute them to the appropriate ports of
confirm the SETWFL result(s), help select the correct the UHS relay.
alternative result, or provide a TW-based result when the results
from other fault-locating methods are not available.
 10

Fig. 12. Telecommunication infrastructure between Mudarra and N. Valladolid terminals using passive optical multiplexers/demultiplexers (MUX/DEMUX).

B. Telecommunications Infrastructure for Casaquemada-to- include electronics to amplify the signals before introducing
Onuba Line, 220 kV, 61.98 km them in the fiber-optic channel. The active multiplexers receive
Fig. 13 shows the telecommunications infrastructure two signals of 1,310 nm, one from the direct-fiber port of the
deployed for this line. To accommodate the distance limitations UHS relay, and the second from the SDH equipment connected
of the SDH equipment for the length of this line, REE decided to a multiplexed-fiber port on the UHS relay. The active optical
to apply active optical multiplexers. These active multiplexers multiplexers introduce both signals into a single fiber-optic
use optical principles similar to the passive multiplexers channel for transmission to the other terminal.
employed on the N. Valladolid-to-Mudarra line, but they also

Fig. 13. Telecommunication infrastructure between the Casaquemada and Onuba terminals using active optical multiplexers.
 11

Greg Smelich earned a BS in mathematical science in 2008 and an MS in

IX. REFERENCES electrical engineering in 2011 from Montana Tech of the University of
[1] SEL-T400L Time-Domain Line Protection Instruction Manual. Montana. Greg began his career at Schweitzer Engineering Laboratories, Inc.
(SEL) as a protection application engineer, where he helped customers apply
Available: selinc.com/products/T400L/docs/.
SEL products through training and technical support, presented product
[2] B. Kasztenny, A. Guzmán, M. V. Mynam, and T. Joshi, “Locating Faults demonstrations, worked on application guides and technical papers, and
Before the Breaker Opens – Adaptive Autoreclosing Based on the participated in industry conferences and seminars. In 2016, Greg transitioned
Location of the Fault,” proceedings of the 44th Annual Western to the SEL Research and Development group as a product engineer, where he
Protective Relay Conference, Spokane, WA, October 2017. now helps guide product development, training, and technical support related
[3] A. Guzmán, B. Kasztenny, Y. Tong, and M. V. Mynam, “Accurate and to time-domain technology. He has been a certified SEL University instructor
Economical Traveling-Wave Fault Locating Without since 2011 and an IEEE member since 2010.
Communications,” proceedings of the 44th Annual Western Protective
Relay Conference, Spokane, WA, October 2017.

X. BIOGRAPHIES
David López Cortón received his Master’s Degree in Industrial Engineering,
specialty Electrical Engineering, by the Polytechnique University of Madrid
(UPM), Spain. In 2009, he joined Red Eléctrica de España (REE), the Spanish
TSO, as dispatcher of the national control room. In 2013, he moved to the
System Security Department of REE, where he currently coordinates the
Testing and Calibration Laboratory, being responsible for protection testing,
qualification of new protection equipment, validation of protection and
automation schemes, disturbance root cause analysis, and metrology. His
research interests are protection equipment, metrology, power quality, and
renewable energies integration.

Jorge Vaquero Melado received his Master's Degree in Industrial Engineering

specializing in Electrical Engineering from the University of Valladolid
(UVA), Spain. In 2004 he joined Red Eléctrica de España (REE), Spanish TSO,
as a protection engineer in the System Security Department. In 2016, he became
the head of the department, where he is responsible for the Testing and
Calibration Laboratory, relay configuration, protection of the electrical system,
and incident analysis. His research interests are the protection of electrical
systems, protection and renewable energies, wide-area adaptive protection
systems, and analysis of electrical systems.

Jesús Cruz received his BS degree in electrical engineering in 2010 from the
Polytechnic University of Madrid. His final project, which involved protection
relays applied to transmission lines, developed his interest in protecting power
systems. From 2010 to 2015, he worked for the engineering department in ZIV
Automation, providing technical support, configuration, testing, and training on
protection and control equipment. Since 2015, he has been an application
engineer at Schweitzer Engineering Laboratories (SEL), providing technical
assistance and developing solutions for European customers.

Richard D. Kirby, is a senior product sales manager at Schweitzer Engineering

Laboratories, Inc. (SEL) in Houston, Texas. His current focus is ultra-high-
speed transmission line protection technology. He is a registered Professional
Engineer in Arkansas, Louisiana, Michigan, Oklahoma, and Texas. He has
28 years of diverse electric power engineering experience. He received a BS in
engineering from Oral Roberts University in Tulsa, Oklahoma, in 1992, and in
1995, he earned his master of engineering in electric power from Rensselaer
Polytechnic Institute in Troy, New York. He is an IEEE Power & Energy
Society and Industrial Applications Society senior member.

Yusuf Zafer Korkmaz received his BS in electrical and electronics

engineering from the Middle East Technical University, Ankara, Turkey, in
1995. He has worked on protection and control schemes, substation automation
and telemetry systems, and generator automatic synchronization systems. In
2013, he joined Schweitzer Engineering Laboratories, Inc. (SEL) where he is
presently a senior application engineer. He is a member of the IEEE.

Gianfranco Patti received his BS in electrical engineering from Simón Bolívar

University in 2013 and specialized in distribution and transmission electrical
systems at University of the Andes in 2015. For four years, he worked on
projects in Colombia that involved the design, construction, and operation of
© 2021 by Red Eléctrica de España and
power systems, specifically for oil fields and generation plants. In 2018,
Schweitzer Engineering Laboratories, Inc.
he joined Schweitzer Engineering Laboratories, Inc. (SEL) as an application
All rights reserved.
engineer, which has allowed him to expand his knowledge, particularly in
20210701 • TP7002-01
protection and control systems.