New Methods in Power Line Carrier

Monitoring and Analysis

Real-World Examples and Implications for Protection System Reliability
2021 74th Conference for Protective Relay Engineers (CPRE) | 978-1-7281-4196-1/20/$31.00 ©2021 IEEE | DOI: 10.1109/CPRE48231.2021.9429829

Craig Palmer, Alan Jayson Jeffrey E. Brown

PowerComm Solutions Georgia Transmission Corporation
Flemington, NJ USA Tucker, GA USA
[email protected]

Abstract—Continuous monitoring devices for power line “Equipment failures…may be hard to find” [6]. PLC-related
carrier protection channels provide robust data about the misoperations “require significant time and effort in testing the
performance of PLC systems. This paper summarizes the PLC channel, as the requirement for investigations on NERC-
significance of the monitoring device data and cites recent cases reportable misoperations are quite stringent” [7]. “Often the
in its application and analysis. Field examples include remotely result of the testing yields the undesirable result of ‘no problem
identifying failed PLC system components before a misoperation, found’” [8].
various effects of capacitor banks, line trap monitoring, and
more. A new method of event-based transient monitoring is also Continuous monitoring data of PLC channels, sampled at
proposed. The potential for this type of data to improve PLC- 20 MHz wideband, with frequency-selective channels sampled
based protection reliability is analyzed. at 20 kHz, makes things a lot easier. See Appendix A for a
further description of how and where the data is captured.
Keywords—power line carrier; PLC; line-trap; transient;
carrier hole; misoperation; checkback; spark gap; gas discharge
tube; reliability; controlled switching; noise III. NEW METHODS

A. Frequency-Selective Carrier “Oscillography”

I. INTRODUCTION
Like a transient recorder for the AC system, the subject
Power line carrier (PLC) technology has protected the US device stores a “capture” of event-driven data (400 ms per
transmission grid for at least 94 years [1]. Yet to this day, event) about the voltage on the coax during normal operation
utilities find that event records of these analog channels are not and during transient events. The capture is taken in a 10 kHz
always robust enough to allow for full event analysis. The lack bandwidth around the channel frequency. Figure 1 shows one
of good data has been a pain point when looking to reduce of these records: three transients in the carrier band due to the
misoperations on the bulk electric system [2][3][4]. New controlled closing of a 115-kV shunt capacitor bank breaker.
methods of PLC monitoring and analysis are demonstrated here
which reduce utility effort in maintenance and troubleshooting, Clearly visible, noise from the switching of all 3 phases
give clarity on the causes of system events, and provide found its way onto the coax in this single-phase-to-ground-
advanced warning of imminent component failure. coupled system. The normal signal is altered after the event due
to the impedance change. The bursts are separated by 2.78 ms,
or 60° at 60 Hz, a clear signature of controlled switching [9].
II. BACKGROUND
The third burst is the most disruptive and triggered the event in
Today’s typical PLC systems monitor events on the the device. It is likely that the third pole to switch was the pole
channel using the most intelligent devices in the system – the on the carrier-coupled phase.
transmitters/receivers [5]. When the system is healthy it
appears that the devices are doing a good job – no alarms, all is
well. But when alarms or events occur and no issue is found
straight away, available records can appear as a jumble of
timestamped event points, toggling high and low at mystical
intervals that may or may not mean something. This binary trail
leaves much to be desired when analyzing these complex
analog systems.
To describe the challenges, a few quotes from recent papers
on PLC issues are presented: “Power Line Carrier applications
provide the most cost-effective installations but have limited
self-monitoring and failure detection” [2]. As a result, Fig. 1 – Sample Time-domain Voltage Capture Showing Cap Bank Switching

Authorized licensed use limited to: Tencent. Downloaded on June 30,2021 at 01:23:48 UTC from IEEE Xplore. Restrictions apply.
 Also visible is a very brief loss of signal occurring in the A close look at the math demonstrates the limitations of
third burst – in this case, an air gap firing due to the transient reflected power. The complex impedance, ZTERM, and reflected
event. The result in the capture is strikingly similar to others power, RP%, are both related to the same complex number, the
that have been measured in the field at the same circuit location “reflection coefficient” or  (“rho”):
[10]. Carrier holes and loss of signal events much worse than
this one will be examined later.

B. Spectrum Analysis  (1)

Dedicated software can perform a Fast Fourier Transform
where ZTERM is the line-side impedance looking into the coax
(FFT) on the time-domain data. The result is a spectrum plot
input of the tuner, and ZSOURCE is the station-side impedance
with frequency and amplitude in the 10 kHz bandwidth of the
looking back into the control house. Both are complex
channel. For example, a channel with a 100 kHz center
impedances with magnitude and phase angle. For this paper,
frequency can find signals from about 95 kHz to 105 kHz.
ZSOURCE is assumed to be 50  / 0° (by design in the US) [13].
There are many applications for this analysis, from
Now consider reflected power percent. It is given by:
overlaying one capture over another to compare the
presence/absence of frequencies in the band as well as their
amplitude, to observing the “noise floor” level, to analyzing
(2)
very fast events and their effect on the signals. Figure 2 shows
two transmit and three receive signals – levels, noise floor, and
any stray or unknown signals are all immediately apparent. Although  appears in the equation, squaring it effectively
reduces its complexity. Other impedance matching figures like
VSWR or return loss in dB use the absolute value of  which
has an effect like squaring. Basically, once the impedance
matching is expressed in reflected power percent, the
terminating impedance ZTERM that caused it cannot be reliably
known. See Appendix B for the reasoning behind this analysis.
Figure 3 illustrates the relationship between ZTERM and
reflected power for a ZSOURCE equal to 50  / 0° (the chart is
generated using (1) and (2), above). It shows the reflected
power calculated by an impedance meter (or PLC transmitter)
connected to a line tuner input for different values of ZTERM.
Plotting the values in this way demonstrates that RP% values
form a closed loop – so that for a 20% reflected power reading
the actual ZTERM magnitude could be 20 , 130 , 50 , or
anywhere in-between, with varying phase angle.

Fig. 2 – Sample spectrum analysis (FFT) showing five carrier signals

C. Impedance Monitoring
Good matching of source impedance (control house
equipment: transmitters, receivers, hybrids, etc) to load
impedance (generally the high-voltage transmission line) is one
of the critical factors influencing the reliability of a given PLC
channel. To meet NERC’s channel monitoring requirements, a
typical PLC transmitter today provides “reflected power”
expressed in percent [11]. Continuous monitoring devices do
calculate the reflected power, but primarily they monitor the
complex impedance at the input to the line tuner.
What’s the difference, and is one more accurate? The short
answer is, reflected power is an effective go / no-go number,
but gives little clue about the source or nature of the “no-go”
state [12]. Fig. 3 – Reflected power by termination impedance, for a source impedance
of 50 ohms / 0 degrees

Authorized licensed use limited to: Tencent. Downloaded on June 30,2021 at 01:23:48 UTC from IEEE Xplore. Restrictions apply.
D. Trending Using the continuous monitoring records, checkback test
One of the simplest new methods is the long-term trending sequences can be evaluated to determine if the monitoring
of instantaneous analog PLC values. Levels, impedance device saw a good sequence – in other words, is there a
magnitude, and impedance phase angle are recorded at regular problem with the analog signal or not. In many cases a pass/fail
intervals and the data may be plotted over time. Samples are result can be provided.
typically taken at 1-hour intervals.
B. Capacitor bank effect on impedance matching
E. RF Transient Detection
TABLE 1 – EFFECT OF UNTRAPPED CAPACITOR BANK ON
In addition to frequency-selective measurements (10 kHz
IMPEDANCE PHASE ANGLE IN A 115-KV SYSTEM
and narrow bands) the device also monitors the wideband
voltage and current on the coax. By setting high-level threshold Cap Bank Out Cap Bank In
detectors the device can indicate the presence of a possible Station Station Station Station
transient – voltage, current, or both. The levels are typically set A B A B
at about 300 V-pk and 2 A-pk; at least 5 consecutive samples at Mag () 42.8 73.8 55.3 68.7
20 MHz must be seen (or 250 ns). To prevent chatter in the Phase (deg) +4.8 -3.0 -35.1 +54.8
event log, the transient flag is held for 1 second after the RP (%) 0.8 3.8 10.2 29.2
condition is declared. These levels were selected based on
Transmit (dBm) +38.9 +40.8 +40.4 +41.1
hardware, prior research, and after review of continuous
monitoring data from many real-world installations [10][14]. Receive (dBm) +31.1 +31.9 +7.8 +25.4
On a 115 kV transmission line with un-trapped capacitor
IV. REAL-WORLD EXAMPLES banks tapped off the line, the carrier system experienced
receive level and reflected power alarms that were correlated
The following examples are from real-world installations
with the capacitor banks being switched “in”. Un-trapped
of PLC monitoring devices on eleven different transmission capacitor banks are known to cause trouble for carrier systems
lines around the USA. Some examples are informative, since they appear as a low impedance path to carrier-frequency
showing signatures of carrier system events. Other examples energy [6]. At this site, monitoring devices were installed to
show how the new methods helped utilities diagnose problems prove the theory and gain insight.
and take action to address them.
The data is presented in Table 1, above. With the capacitor
A. Checkback failure bank OUT, both terminals measured reflected power less than
4%, generally accepted as very good tuning. The cap bank IN
Checkback issues can make for difficult troubleshooting if a state causes tuning issues. Notably, the effect of the insertion of
problem is not found straight away. Show up on site and the the capacitor banks is more severe on impedance phase angle
tests run clean. The question lingers: is there an undiagnosed than magnitude. Further data on capacitor banks and their
issue, waiting to cause more trouble? Figure 4 is a capture from effect on PLC channels are presented later.
such a case, observed at the initiating end in the middle of a
failed test sequence. A healthy signal (in this case) should look
C. Impedance phase angle and line tuner short/open
like part of a clean sine wave, a smooth line. Instead, there is
clear disturbance visible on the signal. In this example, a utility was getting intermittent loss of
signal alarms from FSK receivers at both ends of a
transmission line. The alarms were mostly fleeting and would
clear by the time personnel were on-site. The device event
record and trending log captured the first look at the trouble
state, with an impedance magnitude of about 120  at -85°
(Figure 5).

Fig. 4 – Slow degradation of transmit level during failed checkback sequence

Checkback tests passed again shortly after the issue

occurred, and levels / signals were observed normal. Prior to
reviewing the monitoring data, a different conclusion had been
drawn about the source of the issue, so a quick review of the
checkback sequence prevented unnecessary work. Monitoring
continues and if the issue re-occurs, it will be caught. Note that
not all checkback issues look like this one. Fig. 5 – Trending data shows high impedance, very negative phase angle

Authorized licensed use limited to: Tencent. Downloaded on June 30,2021 at 01:23:48 UTC from IEEE Xplore. Restrictions apply.
 The testing and the data reveal that, for extreme conditions
like a short or open in the coupling circuit, the impedance
phase angle gives more reliable data about the nature of the
trouble condition than impedance magnitude, and certainly
more than reflected power. Counterintuitively, shorts at the
tuner and CCVT grounds showed an increase in the impedance
magnitude at the input of the line tuner, not a decrease.

D. Power System Events in PLC Channels

This section will examine the carrier-band signatures and
the normal and abnormal effects of events on the power
system. Analysis and identification of the effects of these
Fig. 6 – Time-domain captures show discrete changes of state, chatter
A = healthy, B = trouble
transient events in carrier systems require context [15]. For
that, transient detection, time-domain data and FFT analysis are
taken together to build a picture of these very fast events.
Background
Many types of regular power system switching events
generate transient energy. Faults, breaker switching (and re-
striking, re-ignition, pre-striking), line switching, capacitor
bank switching, reactor switching, lightning, arcing and more
can generate these transient signals [16][17][18]. Years of
wisdom gained from operating PLC channels in these harsh
conditions has made the carrier channel extremely robust and
reliable [5].
Lacking a widely-deployable monitoring device for carrier
systems, researchers in the past have had to put forth much
Fig. 7 – FFT Overlay of Healthy (Green) vs Trouble (Yellow) States effort to probe the carrier channel effects of power system
Rx (left) drops > 40 dB; Tx (right) jumps ~5 dB transients. While the tests all produced significant results and
Event captures showed signs of connection issues in the insights, some of them were limited to one installation and
coupling circuit, like the Checkback example above. Here, sometimes under no-load or lab conditions [8][19][20][21][22].
however, the change occurs quickly, repeatedly, and tends to Considering the tight operating limits of transmission
“chatter” at times (Figure 6). FFT with overlay analysis shows systems these days, it is difficult to imagine getting the
the trouble state caused more than 40 dB attenuation on the Rx clearance to re-run some of these tests – like BPA’s 1955
signal, while the transmit level increased by 5 dB (Figure 7). classic on the Grand Coulee–Olympia line, where an arrow
The utility took the initiative to do some testing and trailing a metallic wire was shot from a crossbow into target
simulation to see if they could replicate the impedance rings on the transmission line, thusly generating various faults
magnitude and phase angle of the failed state. Given the and noise conditions for carrier tests [19]. But there may be no
increase in impedance in the trouble state, an OPEN condition need for such tests anymore. Continuous monitoring
was suspected somewhere in the coupling circuit. Various short capabilities demonstrate that detailed carrier analysis can be
and open conditions were simulated at different points in the real-time and ongoing.
coupling circuit. Two of the states showed a complex The data are bringing a new focus to insights from the past,
impedance like the trouble state (see Table 2). With the data while some assumptions are being understood in a new light.
informing the investigation, an OPEN connection was
eventually found – on the back of the tuner protection unit.
Lightning
This section does not discuss direct lightning strikes to the
TABLE 2 – IMPEDANCE TESTING SHOWS CORRELATION transmission line or carrier equipment necessarily. Rather, it
BETWEEN SHORT AND PHASE ANGLE discusses the effects of lightning energy coupled by various
Mag Phase RP means onto the power system and the coax [14][18]. Historical
() (deg) (%) lightning data is freely available and extremely helpful for
Baseline 49.8 0.0 0.0 verifying storm activity [23]. But is it important? Beyond
Tuner input shorted 24.6 +84.3 85.2 lightning’s ability to cause faults, noise, and loss of signal, it
Tuner ground sw. shorted 173.8 +86.5 93.4 turns out that it can have a lot to say about the health of line
traps, as discussed in a later section.
CCVT ground sw. shorted 179 +85.9 92.9
Tuner input open 117.9 -88.8 96.9 Lightning is distinguished in time-domain captures by an
Tuner prot. unit open 120.3 -88.6 96.5 especially fast transient, one or a few at random intervals, but
Line disconnect open 88 -86.1 89.2 not many [14][18]. Low-level white noise is also visible, its

Authorized licensed use limited to: Tencent. Downloaded on June 30,2021 at 01:23:48 UTC from IEEE Xplore. Restrictions apply.
fair-weather level increased due to the storm activity [24][25].
On-Off / DCB systems with a low receive level are especially
sensitive to lightning activity. Data from such sites during
lightning storms will pick up small disturbances as the storm
approaches, which then grow in amplitude and occur more
frequently, then less frequently as the amplitude decreases and
the storm passes. The correlation between the transient activity,
captures, and historical weather data is remarkable.
Figures 8 and 9 are great examples of the possible severe
effects of lightning on the carrier channel, with at least 4
significant loss of signal events visible among the two. These
were captured during a storm that sat over the transmission line Fig. 10 – Effect of breaker operation in an on-off system
for at least 8 hours; the captures were recorded about 6 hours
apart. Note that the transient bursts occur randomly with no
evidence of a 60-Hz time base, and that the noise-floor increase
is visible as a “buzz” riding on the carrier.

Fig. 11 – Effect of breaker operation in an FSK system

Disconnect Switching
Fig. 8 – Transients due to multiple lightning strikes, showing up on the coax If a breaker operation looks like a single lightning stroke, a
disconnect switch operation looks like a steady stream of
lightning. As with lightning and breaker switching, a typical
PLC system encounters disconnect switching noise as a fact of
life.
In continuous monitoring data, a disconnect operation is
characterized by a cluster of chattery events in the log, with
voltage and current transients typically detected at some point
in the operation. Time-domain captures may show a fast onset
of the disturbance, or the level may increase gradually with
each subsequent peak before leveling out then clearing.
Periods of approximately 8.33 ms are almost always visible
Fig. 9 – Lightning in the same FSK system, 6 hours later between the bursts, especially when averaged over 10 periods.
The 8.33 ms signature represents a frequency of 120 Hz, due to
Breaker Switching the 60 Hz energy as it reaches its positive and negative peaks
[22]. Note that a period of 8.33 ms indicates the noise is likely
A typical breaker operation in a healthy PLC channel looks coming from the coupled phase only.
like a single stroke of lightning – a single transient burst.
Unlike lightning, the signal around the breaker operation is Figure 12 shows the effect on an On-Off carrier channel
clean and without noise. The disturbance from a breaker due to a line switch operation. Note how each burst is clearly
operation, however, tends to last longer, on the order of one to distinguishable from the next and is damped quickly (less than
several milliseconds. Note that disturbances as long as 9 ms 1 ms, see detail). The largest burst, in the middle of the capture,
have been tied to breaker operations in trouble systems. is the one that triggered the event. Observed in an On-Off
carrier channel with no carrier energy present, Figure 14 looks
Figures 10 and 11 below show two breaker operations from remarkably like other field recordings of the high-frequency
two different systems. Figure 10 is from an On-Off system, so components of disconnect switching noise [26].
the transient energy is the only energy on the coax (this makes
On-Off channels good observation posts). Figure 11 shows the Alternatively, the FSK system in Figure 13 shows a
effect of a breaker operation on an FSK channel – note the somewhat steady 8.33 ms period but the bursts last several
signal clearly shorting from its clean carrier wave to a very milliseconds. The regularity of the interval deteriorates as the
noisy 0 V. In each of these installations, several component bursts grow longer. Eventually they run into each other and
issues were discovered. create a loss-of-signal event measured at 17 ms. This is not a

Authorized licensed use limited to: Tencent. Downloaded on June 30,2021 at 01:23:48 UTC from IEEE Xplore. Restrictions apply.
healthy response to a line switch operation. Note the trend of Event records from continuous monitoring have sometimes
the bursts around 0 V, while the AC carrier signal moves above seemed in conflict with this conclusion. While the analysis is
and below 0 V. This is clear evidence of a series of flashovers. logical, it might assume very good timing of switching, and
ideal health of components such as spark gaps, gas tubes, line
traps, arresters, coaxial cable, etc. It should also be noted that
in 2004 access to monitoring devices like the one discussed
here was limited.

Fig. 12 – Noise from a line switch operation in an on-off system, showing

characteristic 8.33 ms between bursts

Fig. 14 – Suspected bad timing on 3rd pole of cap bank switching

Fig. 13 – Line switching in an FSK system with too much noise, Fig 15 – The same cap switch transient event in an On-Off channel
8.33 ms period is clear at first but appears more random as arcing increases
Figures 14 and 15 shows the carrier-band effect of a
Controlled (Synchronous) Switching and Cap Banks suspected mis-timed controlled switching event. Because the
monitoring device was configured for both FSK and On-Off
Other cases illustrate noise in carrier systems that may be channels, the effect of the transient in both systems can be
generated by controlled switching. These events are typically observed. The first two transients in the captures are separated
identified by three distinct bursts separated by 2.78 ms, or 60 by 2.78 ms and tend to ride on the signal, suggesting good
degrees in a 60 Hz system. If the switch timing is precisely synchronous switching. The third transient occurs
synchronous, very little transient noise is generated. [9] approximately 3.4 ms after the second, a timing error of
Shunt capacitor bank switching is a common controlled approximately 0.62 ms or 13 degrees. This is apparently
switching operation in the transmission grid [9]. Field data enough error to cause a transient-induced flashover and loss of
from continuous monitoring devices have turned up hundreds signal, visible in the third transient burst in Figure 14.
of records showing a 3-phase, 2.78 ms controlled switching The FSK and DCB channels shown in Figures 14 and 15
characteristic, sometimes showing of evidence of transient- are roughly 80 kHz apart, with both falling between 100-200
induced loss of signal. These events are seen most frequently kHz. This demonstrates the wideband nature of transients in
where there are line trap issues, dielectric breakdown issues, or switching events and the carrier events that they cause
in the presence of un-trapped components in the system, such [14][17]. Counter-intuitively, the FSK channel in Figure 14
as the un-trapped capacitor bank in Figure 1. registered a Loss-of-Guard while the On-Off channel in Figure
This is an interesting finding, because carrier engineering 15 recorded a Block Rx. This seeming contradiction
guidelines suggest that shunt capacitor bank switching does not demonstrates an important insight: the same types of events
pose a problem to the carrier channel. The reasoning holds that that may cause a loss of guard in an FSK receiver may cause a
the dominant frequencies in these events are well below the DCB receiver to “pick up” if no carrier is present; may cause a
carrier frequency band, and that the overvoltages caused by DCB receiver to “drop out” when carrier energy is present;
capacitor switching are sufficiently low enough to be “of no may cause an FSK receiver to see Trip [21][25][27].
concern for the PLC equipment” [6]. The event captured in Figure 16 was recorded by the
monitoring device and it too implicates shunt capacitor banks.
Significantly, it also caused a loss of guard in the

Authorized licensed use limited to: Tencent. Downloaded on June 30,2021 at 01:23:48 UTC from IEEE Xplore. Restrictions apply.
transmitter/receiver (the receiver had an output wired to a (Figure 19). In this state it may not interfere with the carrier
monitoring device input to coordinate records and key receivers. Occasionally, the noise can increase and evolve into
captures). Interestingly, the monitoring device recorded a very sharp peaks with a seemingly random pattern, and here it
correct blocking operation about 800 ms before this event. can cause alarms – loss of guard in FSK, Block Rx in On-Off
(Figure 18). These events tend to evolve slowly and represent a
Investigation is ongoing, but current wisdom holds that this worsening of the initial condition. A close inspection of Figure
event was a capacitor bank being switched back into the system 19, with the noise in a lower-intensity state, reveals some clues
after reclosing. The way that the bursts are successively about the source of the noise.
damped gives away a characteristic of capacitor bank
switching, especially back-to-back switching [9][28][29]. Note In Figure 19, the period between bursts is measured at 5.56
that this was likely not a healthy switching operation. ms, or a frequency of 180 Hz. Also, the bursts are relatively
similar in amplitude, though each 3rd peak is relatively higher
Other abnormal breaker switching operations such as re- than its 2 neighbors. This suggests the 3rd peak represents noise
strike or pre-strike are likely problematic to the carrier channel, on the carrier-coupled phase. These facts and the changing
if experience is any guide. A probable re-ignition event – when amplitude over time all suggest corona discharge as the source
the contacts across a breaker pole re-conduct within 90° of of the noise. The 5.56 ms period comes from the discharge
interruption – is pictured in Figure 17 [9][30]. The time occurring on the negative half-cycles of the 60 Hz phase
between the first 3 bursts is 2.78 ms, with the smaller 4th burst voltages [15]. The presence of noise from all 3 phases indicates
occurring 3.84 ms later, or 83°. the efficiency with which carrier energy couples across the
three phase conductors of a transmission line [25].

Fig. 16 – Transients from suspected cap bank switching cause loss of guard
Fig. 18 – Seemingly random transient events look like a line switch operation,
but 8.33 ms period between bursts is hard to find

Fig 17 – Suspected re-strike in controlled switching operation

Fig. 19 –Regular 5.56 ms period (120°) between background noise bursts,

Three-phase noise on coax similar amplitides indicate 3-phase noise source (corona discharge)
Just as a series of noise bursts or noise with 8.33 ms period
is the signature of 60 Hz noise from one phase, a period Carrier Holes
measured at 5.56 ms suggests the 120°phase separation of the “Carrier holes have been a topic of discussion for eons and
3-phase AC system. Events involving noise with a 3-phase tend to be a mystery.” A carrier hole event may be defined as
signature are rare in carrier monitoring data, but they do occur. simply “the lack of a signal where one should appear”. The
The events in Figures 18 and 19 were captured at one station term also implies that the “lack of signal” is caused by the
but are almost identical to those at an entirely different station flashover of overvoltage protective devices due to transient
and system. energy from the power system [6]. Continuous monitoring data
Where 3-phase noise is visible the in-band noise can be can capture these carrier holes as they happen, and detailed
high, with 1 Vp-p at the upper range of background noise analysis of their characteristics can be performed.

Authorized licensed use limited to: Tencent. Downloaded on June 30,2021 at 01:23:48 UTC from IEEE Xplore. Restrictions apply.
 In the data, a carrier hole event is recognized first by an Using FFT with overlay analysis, the 5 ms carrier hole was
event capture recorded in-time with a voltage transient, current compared against 5 ms of healthy channel data. Shown in
transient, or both. This is the first clue. Time-domain analysis Figure 22, the detail in the comparison is striking. It shows that
can provide context about the amplitude duration, and source the receive signal is down 20 dB during this time, beyond the
of the event – was it lightning, a breaker operation, disconnect, 15 dB margin used by most receivers. It also shows that no
cap bank switching, etc. FFT analysis will also reveal whether Trip energy was dominant during the event.
the event really was a carrier hole – did a “lack of signal”
occur, and was there evidence of flashover? An IEEE report on carrier considerations states, “Power line
carrier is a very robust communication medium. It is not
While carrier holes are traditionally discussed in terms of normal for the signal to be interrupted during fault conditions.
On-Off systems, the term is equally valid for FSK systems. An Observation and classification of carrier holes followed by an
example of a carrier hole event in an FSK system (this one investigation will result in a more reliable protective relaying
lasted about 5 ms) is shown in Figures 20-22. In Figure 20, the system” [6]. Others have echoed this sentiment [2][3][5][7].
healthy transmit and receive carrier signals are disrupted and
pulled to a very noisy 0 V, like the disconnect operation in The suggested methods for observation and classification of
Figure 13. Also visible is the noise generated by the arcing in carrier hole events include visual inspection of gaps and tubes,
the protective device [31] – tell-tale signs of a carrier hole. BIL tests, as well as review of any block-extend timers going
Figure 21 shows the detail of the carrier hole itself. active in the relay logic during faults – effective but rather
tedious, time-consuming, and without any context about the
nature of the analog energy that might have caused the issue
[5][6][7][22][25]. Not on the list: review of the time-domain
characteristics of the event; what kind of transient energy (if
any) was present at the time of the event; FFT analysis to check
the frequency content of the event. When added to inspection
and testing techniques, these new methods sharpen a utility’s
ability to observe, classify, mitigate, and verify fixes for carrier
hole events on their systems.

Directional Comparison Blocking (DCB) Operations

DCB with power line carrier is widely used for transmission
line protection in the US [7]. Recent efforts to reduce
Fig. 20 – 5 ms loss of signal on transmit and receive protection misoperations have identified these schemes as still
especially vulnerable to over-tripping due to loss of carrier
signal (carrier holes) [2][3]. The ability to review the carrier
portion of these protection scheme operations in granular detail
is thus of great value.
While no mis-operations due to carrier holes have yet been
observed, many successful blocking operations have been
recorded and have provided good insights. They have also
demonstrated the ease with which carrier holes are identified
when they are recorded. Interestingly, noise due to fault
inception is also regularly visible in these captures; in
agreement with previous observation, it tends to dissipate very
quickly and seems to pose little risk to the carrier channel
Fig. 21 – 5 ms loss of signal on transmit and receive (detail) [14][20][21]. Note that fault inception effects are separate from
the effects of a sustained fault [14][20].

Fig. 22 –FFT overlay analysis of loss of signal (green) vs. normal (yellow) Fig 23 – Fault & block operation, showing carrier hole in the middle
Left vertical bar shows DTT Trip, right vertical bar shows Guard (the carrier hole in this case did not cause an over-trip)

Authorized licensed use limited to: Tencent. Downloaded on June 30,2021 at 01:23:48 UTC from IEEE Xplore. Restrictions apply.
 Fig 24 – Fault & block operation showing and START/STOP logic issue
Fig. 26 – FFT overlay analyis of DTT mis-op (green) vs normal (yellow)
Figure 23 shows the characteristics of a typical blocking Left vertical bar shows TRIP freq; Right vertical bar shows GUARD freq
operation. The onset of the fault sends out traveling waves that
Figure 25 shows the time-domain capture of the disturbance
contain broad RF transient energy, which is visible here
that caused the mis-operation. The disruption appears to last for
because it has made its way onto the coax (A). Soon after, a
4 ms, but careful FFT analysis shows that trip energy was
Block is sent out from the local transmitter (B). A significant
dominant for as long as 7 ms continuous. As carrier
disturbance is visible on the carrier about 85 ms after the onset
disturbances go, this one was not particularly remarkable in
of the fault – a carrier hole lasting approximately 1 ms (C). The
length or amplitude. In this case the receiver mis-operated
block-hold logic of the receiver was able to ride through the
because the transient noise had just the right TRIP profile,
disturbance because there was no over-trip.
demonstrated in Figure 26. Compare this disturbance to the
Figure 24 shows a blocking operation in which both the carrier hole event in Figure 22.
local and remote Block signals can be seen in the FFT analysis.
Interestingly, this was not one of the more severe transients
The fault inception occurs first (A). Then the local transmitter
in the record. The one that caused a misoperation was due to a
turns on (B), with the Block signal received from the remote
faraway switching event with just the right energy profile.
end a few milliseconds later (C). After about 20 ms the remote
Misoperations in DTT systems tend to occur in these “sweet
carrier turns off due to a STOP input going active (D).
spots” given the various parameters of the receiver and logic
Midway through the local Block transmit signal in Figure that must be satisfied [21].
24, there is a 1 ms disturbance that looks a lot like a carrier
hole (E). FFT analysis (not shown) revealed that this was the Line Trap Monitoring
remote carrier turning back ON. Correlation with relay I/O data In several installations, continuous monitoring data has
showed that this was an anomaly in the logic that occurred been used to identify failing or failed tuning packs in line traps.
when START and STOP were de-asserted simultaneously. Trending, time-domain, and FFT overlay analysis have been
Direct Transfer Trip (DTT) Misoperation used to identify the trap issues. A typical analysis: trending
data shows intermittent changes in impedance (mostly phase
In one case, an unwanted DTT operation was reported at a angle) in the range of 30 to 60 degrees in the trouble state
345 kV station. No monitoring device was installed at the site (Figure 27); FFT overlay analysis shows receive signals
that misoperated, but there was one at the remote station. It significantly down, transmit signals relatively unchanged
captured what looked like a clear breaker switching event at the (Figure 28); even a small event can trigger a capture, though no
time of the DTT misoperation. Initially no breaker operations time-domain disturbance is visible (Figure 29).
were noted nearby but the search was broadened and a breaker
opening event was identified over 60 miles from the breaker
that mis-operated.

Fig. 27 – Trending shows intermittent impedance changes

(magnitude and phase angle)
Fig. 25 – Small 4 ms transient noise (faraway breaker) causes DTT misop

Authorized licensed use limited to: Tencent. Downloaded on June 30,2021 at 01:23:48 UTC from IEEE Xplore. Restrictions apply.
 Line Traps and Transients
Review of field data where failed line trap tuning packs
were found demonstrated some interesting relationships
between line trap tuning and transient events seen on the coax:
x Sites with failed line trap tuning packs were noted to
have detected a relatively high number of events,
especially transients
x The transient events were more likely to cause an event
capture – to cause a disturbance in the tuned carrier
band

Fig. 28 – Part of overlay shows receive levels (left, middle) are approx. 10 dB x The captures show characteristics of switching and
lower during the trouble state, transmit level (right) is relatively unchanged lightning, some with surprisingly low-amplitude
disturbances
One key here is to remember that an important function of a
line trap is to isolate the carrier system from transient energy
generated by switching on the substation bus [5].
Transient events on the substation bus, by their nature, have
a broad frequency spectrum [18]. Energy from these events
regularly passes through the line trap and onto the transmission
line. If the line trap is healthy, transient energy in the carrier
band is significantly attenuated. So long as the out-of-band
energy does not cause a dielectric breakdown somewhere,
resulting in a carrier hole, the receivers are likely to be
unaffected by these normal power system events.
The trap is a bidirectional device, so in the same way it
blocks carrier energy from entering the local station bus, it
Fig. 29 – Overlay shows example of noise floor “blips” due to very small
disturbances while the trap tuning pack was in its trouble state
blocks carrier-band transient energy from leaving the local
station bus and getting onto the transmission line. Once carrier-
Using continuous monitoring data in this manner, four line band transient energy gets onto the line it has a tuned path
traps have been identified as having failed tuning packs (other straight to the PLC receivers at all terminals.
sites are still waiting for tuning pack replacement before the ID
can be confirmed). Where a device is installed at both ends of a Air gaps / gas tubes and transients
line, the relative amplitudes of disturbances can be compared
Another significant pattern seen in continuous monitoring
for context about where the source is. The methods get more
data relates to overvoltage protective devices (generally gas
reliable and accurate with each positive ID.
tubes and spark gaps though other types may still be found in-
service) [22]. When these devices have been found
V. RISK-BASED MONITORING – TRAP, GAP, TRANSIENTS compromised, the following characteristics are seen in the data:
The fact that a carrier system lives on the power system can x Transient events more frequently cause carrier-band
make some components especially difficult to access and test. disturbances and flashover
Line traps and overvoltage protective devices, when they may
need attention, require a hard-to-get outage before work can x When flashover occurs the carrier hole is “bigger” – the
begin [25][32]. Failure or end-of-life of these same period of arc conduction tends to last longer than in a
components may also increase the misoperation risk of a healthy device
particular channel, so the ability to monitor and indicate
imminent failure would be a good application of the continuous Again, these are not remarkably new insights: “Every time
monitoring technology presented in this paper. there is a transient…the protective gap will likely flash over,
which in turn builds up carbon, thereby decreasing the flash-
An approach to this end is being developed and is showing over voltage of the gap. This carbon build-up may cause the
promising results. More field data must be gathered and gap to flash-over sooner than expected the next time and not
analyzed but the basics appear relatively simple, measurable, seal-off in time, thereby creating a (severe) carrier hole” [5].
and effective. The primary components considered in the Gas tubes are de-rated in a similar manner as the electrodes
monitoring approach are the line trap and any overvoltage sputter and coat the walls of tube [31][33].
protective devices (or other types of dielectric degradation, for
instance on the coax). The tendency of these conditions to When combined, these two risk factors – trap tuning pack
worsen the influence of transients on carrier systems is failure, and the condition of protective gaps and component
exploited. dielectric strength – may put the system at severe risk of

10

Authorized licensed use limited to: Tencent. Downloaded on June 30,2021 at 01:23:48 UTC from IEEE Xplore. Restrictions apply.
misoperation. Here is the nightmare scenario, and probably the nominal weight, and a voltage- and current-transient event gets
failure mode of more than one misoperation from the past: the highest weighting. If a channel event is recorded within a
pre-defined window of time around the transient, it is counted
x A tuning pack fails in a line trap, allowing more carrier- as a transient-induced event. The event is similarly weighted
band transient energy from the substation bus onto the based on intensity and counted in another index.
transmission line
These numbers are summarized by day and tracked over
x If the tuning pack failure results in reflected power time. Interestingly, review of continuous monitoring data from
alarms, perhaps the tuner is ‘re-tuned’ into the bad trap carrier installations reveals that these systems tend to have a
to clear the alarm (not advisable) [32] transient profile that follows a 7-day schedule. Most transients
x The transient energy disturbs local and remote receivers tend to occur during the weekdays, when more regular
switching is happening on the station bus and on nearby
x The situation may be compounded (sooner or later) by a systems. With this fact in mind, a 7-day moving average is
de-rated spark gap, gas tube, or other dielectric used to track transient activity and transient-induced alarms.
breakdown issue
A sample analysis is shown in Figure 30. The continuous
x Frequent arcing accelerates the de-rating of the air gap line represents the moving average index. This value should
or gas tube never go over 1 – if it does go over, the system enters the high-
risk zone. The vertical bars in Figure 30 represent an answer to
If this is an FSK system, loss of guard or unblock events may
the question, how affected is the carrier band by normal power
be occurring. If it is an On-Off system, the Block Rx may
system transients? Each data point in the graph represents one
chatter at seemingly random times. Checkback tests are
day on the system.
unlikely to detect the issues. Given the difficulty in getting
outages, the protective device in the tuner might be inspected
but not tested, and the CCVT and trap will have to wait.
Meanwhile the transient activity continues.
Eventually, a misoperation occurs.
Now assume similar conditions to the nightmare scenario
above, but with a continuous monitoring device capable of
detecting an increase in transient activity and measuring its
relationship to alarms on the carrier channel:
x As transient activity starts to exceed the system’s
normal transient profile and alarms increase, an alert –
“maintain-me-soon, transient risk high” – is sent out
before the next maintenance cycle
x Event data and overall system statistics can be reviewed
for insight into the issues, informing the decision to Fig. 30 – Transient Profile shows significantly improved transient resilience
after on-site work (replace failed tuning pack and fix issue with gas tube)
request an outage
The data in Figure 30 come from an installation in which a
x Review of data also informs the work to be done on-site
NERC-reportable misoperation occurred near the middle of the
when an outage is granted
graph. Work was done on-site between the green arrows.
x Whatever steps are taken can be immediately monitored During this time, a failed tuning pack was found in the line trap
to confirm the issue has been mitigated, or to inform and a significant issue with a gas tube was found which would
next steps cause it to arc over at a lower voltage than specified by its
design. After the tuning pack was replaced and the issue with
If the line trap is implicated, trending data, captures and the gas tube was corrected, the line was placed back in service.
FFT analysis will reveal the issue, and the line trap is requested Transient activity was still regularly detected but only one
to come down for maintenance at the next outage. The urgency transient event caused a channel disturbance, and that was a
of the outage may be qualified by what is found in the data. routine breaker operation.
If the air gap or gas tube is implicated, captures and FFT Going forward, more field data will be evaluated in this
analysis will show the deterioration as frequent, extended manner and the technique will continue to be refined. Emphasis
carrier holes occurring frequently. Other dielectric failures may is placed on reporting a single figure of merit which is not
show different characteristics than the air gaps and gas tubes. prone to chatter – it must represent the long-term trend and not
“blip” into the alarm region during a day or two of intense
Quantifying transient risk lightning or station switching. Like the other new methods and
The approach, as stated, is rather simple. Whenever a applications presented in this paper, this technique has
transient is detected it is recorded and counted, weighted potentially big implications for improvements in carrier system
according to its intensity. A voltage-only transient is assigned reliability and reduction in carrier-related outages and
the lowest weight, a current-only transient is assigned a maintenance costs.

11

Authorized licensed use limited to: Tencent. Downloaded on June 30,2021 at 01:23:48 UTC from IEEE Xplore. Restrictions apply.
 VI. IMPLICATIONS FOR RELIABILITY ACKNOWLEDGEMENTS
In 2013 the North American Electric Reliability The authors would like to acknowledge Danny Worrell of
Corporation (NERC) published the findings of a task force Georgia Power for his contributions and insights.
which analyzed transmission system protection misoperations
in the US. Regional entities like SERC, MRO, and WECC (to REFERENCES
name a few) provided historical system data to the task force.
The goals: research data from past misoperations; find the top 3 [1] A. S. Fitzgerald, “A Carrier-Current Pilot System of Transmission Line
Protection” Transactions AIEE, September 1927
causes of misoperations and develop sub-causes for each; and
[2] NERC Protection System Misoperations Task Force, “Misoperations
make suggestions for reducing future misoperations [2]. Report,” April 2013
The findings relevant to PLC systems were significant: 17% [3] NERC, “Lessons Learned – Initiatives to Address and Reduce
of transmission system misoperations were caused by the Misoperations 20181201,” December 2018
communications channel (396 out of 2279). Further, 12% of [4] NERC, “2020 State of Reliability Report – An Assessment of 2019 Bulk
Power System Performance,” July 2020
misoperations were reported as having “unexplainable” cause
[5] M. P. Sanders, “A Reliable Power-Line Carrier-based Relay System –
(273 out of 2279) [2]. NERC’s most-recent “State of Avoiding Mistakes that Cause PLC Systems to Misoperate or Fail to
Reliability” report from 2020 confirms that “communication Operate,” Texas A&M Protective Relay Conference, March 2012
failures” have consistently been a top 3 cause of misoperations [6] IEEE Power System Relay Committee, Working Group H9, “Special
since data collection started in 2011 [4]. considerations in applying power line carrier for protective relaying,”
2004
Reading in-between the lines for clues about which
[7] Z. P. Campbell, Y. Xue, R. C. Dixon, “Power Line Carrier Directional
communication technologies were involved, an estimate is Comparison Blocking Misoperation Event Analyses & Avaoidance
made here from the 2013 NERC report’s data: for the total Techniques,” 67th Georgia Tech Protective Relay Conference, 2013
number of transmission system misoperations caused by [8] T. Bell, “Mitigating Carrier Holes in Power Line Carrier,” Georgia Tech
communications, 30% involved PLC systems. That’s roughly Protective Relay Conference, 2018
5% of all transmission system misoperations documented in the [9] D. Goldsworthy, T. Roseburg, D. Tziouvaras, J. Pope, “Controlled
report. [2] Appendix C details how these numbers were Switching of HVAC Circuit Breakers: Application Examples and
determined. Benefits,” 34th Western Protective Relay Conference, 2007
[10] G. A. Franklin, S. Hsu, “Power Line Carrier Interference Caused by DC
Part of the reason for the high share of misoperations by Electric Arc Furnaces,” 2003 IEEE Power Engineering Society General
PLC systems is undoubtedly their wide installed base. Another Meeting
major reason that cannot be ignored is simply a lack of good [11] NERC PRC-005-6, 2016
data available to utility engineers and technicians. As the 2013 [12] R. Ray, A. Jayson, R. Fella, J.E. Brown, N. Stone, R. Baldwin, “The
NERC report notes, “Improvements in data, while not directly Advantages of Continuous Monitoring of PLC Channels Applied to
Protection Systems”, Texas A&M Relay Conference, 2017
reducing misoperations, help entities determine areas to
improve by identifying misoperation causes and proper [13] IEEE C93.5 – Standard for Power Line Carrier Trnasmitter/Receiver
Equipment used to Transfer Discrete Teleprotection Signals
mitigation steps. Without this analysis, entities will not be able
[14] T. Udo, M. Kawai, “Fault Generated Impulse Noise Voltage in a
to fix root causes that, potentially, could prevent additional Transmission Line,” IEEE Transaction on Power Apparatus and
misoperations” [2]. Systems, Vol. PAS-86, No. 6, June 1967
[15] R. C. Cheek, J. D. Moynihan, “A Study of Carrier-Frequency Noise on
CONCLUSIONS Power Lines – Part I – Theoretical Considerations and Measuring
Techniques,” Transactions of the American Institute of Electrical
Continuous monitoring for power line carrier represents a Engineers, Vol. 70, Issue 2, 1951
step forward in the analysis and operation of these historically [16] W. C. Kotheimer, “The Source and Nature of Transient Surges,” IEEE
reliable protection channels [5]. The ability to identify and Transactions on Industry Applications, Vol. IA-13, No. 6,
November/December 1977
analyze events in sharp detail is incredibly valuable. Further,
[17] W. C. Kotheimer, L.L. Mankoff, “Electromagnetic Interference and
the ability to prioritize maintenance and troubleshooting Solid-State Protective Relays,” IEEE Transactions on Power Apparatus
activities based on predictive analytics related to transients has and Systems, Vol. PAS-96, no. 4, July/August 1977
the potential to dramatically improve the operation and [18] A. Greenwood, “Electrical transients in power systems, 2nd ed.,”
performance of these channels. Taken together, the Renselear Polytechnic Institute, Electric Power Engineering Department,
1991
capabilities ultimately reduce the effort, uncertainty, and costs
[19] A. W. Adams, “Field Tests and Operating Experience with Carrier-
involved in carrier maintenance and event analysis. Transfer Trip Relaying for Line Protection,” Transactions of the
As one relay engineer put it, “To only review…data after American Institute of Electrical Engineers, Part III: Power Apparatus
and Systems, Vol. 76, Issue 3, August 1957
obvious misoperations is analogous to a doctor ignoring your
[20] D. E. Jones, “Staged Fault Tests with Power-Line Carrier Transferred-
reports of anxiety, tightness in chest, nausea, and shortness of Trip Relaying for Line Protection,” Transactions of the American
breath and only treating you for a heart attack if you actually Institude of Electrical Engineers, Part III: Power Apparatus and Systems,
experience cardiac arrest.” [30] With enhanced ability to Vol. 78, Issue 3, August 1959
monitor the vital signs of a carrier system and to hear its [21] B. Bozoki, “Effects of Noise on Transfer-Trip Carrier Relaying,” IEEE
Transactions on Power Apparatus and Systems, Vol. PAS-87, No.1,
complaints, continuous monitoring helps mitigate severe January 1968
events by identifying those sites most in need of attention so
that they can be treated before suffering an acute crisis.

12

Authorized licensed use limited to: Tencent. Downloaded on June 30,2021 at 01:23:48 UTC from IEEE Xplore. Restrictions apply.
[22] J. J. Meinardi, M. P. Sanders, “Investigation and Analysis into the Mis-
Operation due to Carrier Holes,” AMETEK Power Instruments &
Florida Power & Light, 2008
[23] https://www.lightningmaps.org/
[24] R. C. Cheek, J. D. Moynihan, “A Study of Carrier-Frequency Noise on
Power Lines – Part IV – Conclusions of Field Measurements,”
Transactions of the American Institute of Electrical Engineers, Vol. 74,
Issue 3, 1955
[25] IEEE 643 – Guide for Power Line Carrier Applications, 2005
[26] E. O. Schweitzer III, V. Skendzic, A. Guzmán, M. V. Mynam, J. L.
Eternod, Y. Gong, “Mystery Solved: Five Surprises Discovered With
Megahertz Sampling and Traveling-Wave Data Analysis,” 45th Annual
Western Protective Relay Conference, October 2018
[27] P. J. Tatro, K. A. Adamsom, M. A. Eitzmann, and M. Smead, “Power
line carrier interference from HVDC converter terminals,” IEEE
Transactions on Power Delivery, Vol. 8, No. 3, July 1993
[28] M. Beanland, T. Speas, J. Rostron, “Pre-insertion resistors in high-
voltage capacitor bank switching,” Western Protective Relay
Conference, 2004
[29] ABB, “Capacitor switching comparison: the supremacy of diode
technology,” 2015
[30] R. W. Patterson, “The Importance of Power System Event Analysis,” 8th
Annual Fault and Disturbance Analysis Conference, April 2005
[31] IEEE-487 Recommended practice for the protection of wire-line
communication facilities serving electric supply locations,” 2007 Fig. A2 – Monitoring device internal detail (example)
[32] J. E. Brown, “Power line carriers: simplified – a field engineer’s
perspective,” PLCExperts.net, 2018
To obtain an equation that can calculate the termination
[33] Bourns (T. Ardley), “First principles of a gas discharge tube primary
impedance ZTERM from reflected power, equations (1) and (2)
protector,” 2008 are solved for ZTERM in terms of reflected power percent:

APPENDIX A – MEASUREMENT TECHNIQUES

Figure A1 shows the ideal place in the PLC circuit for
(B1)
continuous monitoring devices. Installation at the tuner input is
important because this is the most accurate point to measure
impedance matching. Also, it is a convenient measurement Table B1 shows the resulting ZTERM value calculated from
location because all signals entering and leaving the station are reflected power. Note that each impedance magnitude result in
available at this point in the circuit. the table has a phase angle of zero. Table B2 shows the
percent error introduced when using this method to calculate
the line tuner input termination impedance ZTERM.

APPENDIX C – NERC MISOPERATION DATA

Section VI makes a claim that data in a 2013 NERC report
suggest 30% of all communications-related misops were
caused or sub-caused in some way by a PLC system. The
reasoning and numbers behind that estimate are described
here.
There were a total of 396 misoperations attributed in one
way or another to “communication failures”. Descriptions of
each of the sub-cause groups contained further data about the
Fig. A1 – Monitoring device installation location cases in those groups. For two of the sub-groups, a minimum
number of cases with PLC communications could be
PLC signal data is processed and recorded at several determined
different stages of the device: wide-band, 10-kHz intermediate
frequency, and narrow-band (300, 600, 1200 Hz typical) – see At least 34 were from sub-cause “Modulator Problems”:
Figure A2. Each stage provides different context about the “In 34 cases the radios were replaced without further
signal and/or events. investigation.” Radio is understood to mean a PLC
transmitter/receiver. 34 out of 130 total in this group is a rate
APPENDIX B – MEASUREMENT TECHNIQUES of 26%.
This appendix examines a method for calculating
impedance from reflected power percent and analyzes the
accuracy of this method.

13

Authorized licensed use limited to: Tencent. Downloaded on June 30,2021 at 01:23:48 UTC from IEEE Xplore. Restrictions apply.
 TABLE B1 – TERMINATION IMPEDANCE ZTERM CALCULATED x 22 from sub-cause “Insufficient Information” (111 total
FROM REFLECTED POWER USING EQUATION (B3) * 20%)
ZtermPhaseAngle(deg) x 12 from sub-cause “Communications Medium” (63
40 30 20 10 0 10 20 30 40 total * 20%)
25 146 124 110 103 100 103 110 124 146
x 5 from sub-cause “Incorrect Logic Settings Issued” (24
30 128 108 94 86 83 86 94 108 128 total * 20%)
35 118 97 83 75 71 75 83 97 118
40 111 91 76 66 63 66 76 91 111 x The remaining sub-cause “Human Error” had 12 total
misops. While it did contain a PLC-related misop in the
45 108 88 73 61 56 61 73 88 108
description, it mentioned that this misop has already
Ztermmagnitude(ohms)

50 107 87 71 60 50 60 71 87 107
been counted in another sub-cause category.
55 108 87 72 61 55 61 72 87 108
60 110 89 75 64 60 64 75 89 110
65 113 92 78 69 65 69 78 92 113
34 + 35 + 22 + 12 + 5 = 108 PLC-related events
70 117 96 82 73 70 73 82 96 117
108 PLC events ÷ 396 total events = 27% caused by PLC
75 121 100 86 78 75 78 86 100 121
(round up to 30%)
80 125 105 91 83 80 83 91 105 125
85 130 109 96 88 85 88 96 109 130 108 PLC events ÷ 2279 total events (NERC Misop Report,
90 135 114 100 93 90 93 100 114 135 Figure 2 – NERC-wide Misoperations by Cause Code) =
95 141 119 105 98 95 98 105 119 141 4.74% (round up to 5%)
100 146 124 110 103 100 103 110 124 146

TABLE B2 – PERCENT-ERROR INTRODUCED WHEN

CALCUALTING ZTERM FROM REFLECTED POWER PERCENT
ZtermPhaseAngle(deg)
Craig M. Palmer received his Bachelor of Applied Science in
Electronics and Computer Engineering Technology at the New Jersey
40 30 20 10 0 10 20 30 40
Institute of Technology, Newark, NJ in 2011. He has worked on
25 484% 397% 341% 310% 300% 310% 341% 397% 484% power system communications and pilot protection channels since
30 328% 259% 213% 187% 178% 187% 213% 259% 328% then, first with RFL Electronics and now with PowerComm
35 236% 178% 138% 113% 104% 113% 138% 178% 236% Solutions. He is an active participant in the IEEE Power System
40 178% 127% 91% 66% 56% 66% 91% 127% 178% Relaying and Control Committee (PSRC), a member of the IEEE
45 140% 95% 61% 36% 23% 36% 61% 95% 140% Power System Communications and Cybersecurity Committee
Ztermmagnitude(ohms)

(PSCC), and current chair of the PLC subcommittee in the PSCC

50 114% 73% 43% 19% 0% 19% 43% 73% 114%
(2021).
55 96% 59% 32% 11% 0% 11% 32% 59% 96%
60 83% 49% 25% 7% 0% 7% 25% 49% 83% Alan Jayson has spent the last 35 years designing various audio tone
65 74% 42% 20% 6% 0% 6% 20% 42% 74% and PLC communication and instrumentation products for the electric
utility system protection industry. He started his career in 1986 with
70 67% 37% 17% 5% 0% 5% 17% 37% 67%
INIVEN, where he helped design protection communication systems
75 61% 33% 15% 4% 0% 4% 15% 33% 61% that are still in use today. In 2012 he joined PowerComm Solutions as
80 57% 31% 14% 3% 0% 3% 14% 31% 57% lead design engineer. Prior to that, he worked for Signalcrafters for
85 53% 28% 12% 3% 0% 3% 12% 28% 53% 14 years. He is also a patent holder.
90 50% 27% 12% 3% 0% 3% 12% 27% 50%
Jeffrey E. Brown graduated in 1997 with a Bachelor of Science in
95 48% 25% 11% 3% 0% 3% 11% 25% 48% Engineering from Georgia Tech University. After graduation, he
100 46% 24% 10% 3% 0% 3% 10% 24% 46% worked for Georgia Power for 23 years and finished as Team Leader
for Power Line Carrier covering the state of Georgia. He presently
works for Georgia Transmission Corporation as Principal Engineer
At least 35 were from sub-cause “Station Signal Path Transmission and Power Line Carrier Support. He is a member of the
Failure”: “Of 69 records in this sub-cause, the majority can be North American Transmission Forum (NATF) and a member of the
attributed to power line coupling applications.” A majority Power Line Carrier subcommittee in the IEEE PSCC. He is the
means at least more than half of 69, giving 35. 35 out of 69 is a author of “Power Line Carriers: Simplified.”
rate of 51%.
For the four remaining groups, an estimated PLC rate of
20% was used to get a number of misoperations. The
estimated rate is considered conservative as it is less than both
of those groups above where the minimum percentage could
be determined. This estimate is also informed by field
experience.

14

Authorized licensed use limited to: Tencent. Downloaded on June 30,2021 at 01:23:48 UTC from IEEE Xplore. Restrictions apply.