Proactive PQ for Improved Grid Performance

Authors and Presenters:

William Howe
Electric Power Research Institute (EPRI), USA

Robin Pittwood
Powerco Limited, New Zealand

EEA2023 Conference, 27 – 29 June 2023, Christchurch

1
Abstract: Power Quality (PQ) in the electric utility industry has historically focused on
retroactive understanding and analysis of events after their occurrence – we typically wait for
thresholds to be exceeded, equipment to malfunction, and/or end-use customers to complain of
poor service before bringing our PQ expertise seriously to bear, by which time the problems have
already become expensive. We believe the future of PQ is through proactive and continuous
analysis that enables early detection, prediction and, when possible, preventative action. This
paper will highlight how the Electric Power Research Institute (EPRI) and its member utilities
have collaborated to create and deploy proactive PQ tools and techniques to significantly
increase the value of incoming PQ data streams through detection of grid and operational
changes that may pre-sage costly failures and offering the prospect for significant improvements
in key operational parameters and economic performance. The paper will describe initial
applications of Proactive PQ and illustrative case studies showing early results and economic
returns.

2
Introduction
Power quality (PQ) has long been an essential core competency for electric utilities, but its role
is undergoing an important change. PQ in the electric utility industry has historically focused on
retroactive understanding and analysis of events after their occurrence – we typically wait for
thresholds to be exceeded, equipment to malfunction, and/or end-use customers to complain of
poor service before bringing our PQ expertise seriously to bear, by which time the problems have
already become expensive. What if we changed that paradigm by actively using incoming PQ
data streams to search for indications of unexpected change or malfunction before they exceed
thresholds, negatively impacted operations, or otherwise became expensive?
Using quality proactively has a strong precedent in other industries. Manufacturers of consumer
products are well known to closely monitor their procedures and processes to detect shifts in
quality before non-conforming products are produced. The automobile industry, in particular, is
famous for having emulated/adapted Japan’s Toyota Production System developed in the 1960s
to dramatically improve the quality of automobiles around the world. It is likely that if one were
to ask a quality manager at a major automobile manufacturer about the quality of their products,
you would learn about their investment in proactive quality processes, and the resulting high-
quality products that approach produces. Many of these existing tools and techniques (and
emerging new ones) can be adapted to also improve utility electrical power quality and resulting
in many of the same cost-saving and performance benefits.
The EPRI Power Quality research program, in concert with participating utilities such as
Powerco Limited, has established the broad goal of identifying essential assets where:
1. Indications of declining health or incipient failure are manifested in measurable voltage
and/or current signals
2. Enough understanding of those signals exists (or can be created) to identify these signals
with a high degree of confidence
3. The signals can be detected far enough in advance to allow staff to intervene before total
or expensive failure
Encouragingly, quite a number of common utility assets have been found to meet these broad
criteria, including:
• Substation transformers
• Load-tap Changers
• Measurement devices such as CCVT, MVT, PT, etc.
• Switch gear
• Capacitor banks
Research on applying proactive PQ to health assessment of these and other assets is on-going,
but significant and valuable successes are already achievable.

3
The Case for Proactive PQ
It is the opinion of the authors that the best future for power quality is for the practice to become
increasingly proactive, similar to how quality is pursued in other industries and for other
consumer products and services. While the activities necessary for the overall function of an
electric utility are very broad – ranging from management of large assets and fuel prices to
disaster planning and customer billing – PQ deals solely with the essential qualities of the
electric utility’s core product – electric power.
Historically, utilities produced all electric energy using large, centralized, and well-managed
power plants. As such, electric power was, for all practical purposes, perfect when first created,
with billions of identical voltage sinusoids all equivalent in amplitude, frequency, and phase.
Only when loads were connected (or something unusual happened), did the quality of electric
power begin to deviate significantly from this perfect start and, so, it has been customary to
assume that the quality of electric power was adequate until some issue indicated otherwise.
Turning to a proactive PQ approach requires significantly more up-front effort than a reactive
approach, but the results are often spectacular. Through application of Statistical Process Control
(SPC) techniques, waveform recognition and clustering, automated performance benchmarking
and, more recently, AI/ML techniques, proactive PQ is now a reality.
Early Successful Proactive PQ Application: Capacitor Bank Health
One of the first applications of proactive PQ techniques was for transmission-level capacitor
banks operated by Tennessee Valley Authority (TVA) in the United States. TVA has over 300
transmission level capacitor banks – these assets are large and expensive, and essential to
effective grid operation. Initial proactive PQ efforts looked at a wide array of performance
metrics, as shown in Table 1. It is very important to note that while Table 1 contains technical
analytics, such as I2t, voltage change, and resonance, it also contains many otherwise mundane
monitoring metrics such as timing, loss of signal, etc. As such, it is noteworthy that benefits from
proactive monitoring need not focus only on complicated analytics, but can also include common
status checks, statuses, etc.

4
Table 1: Roster of performance metrics identified and monitored for proactive monitoring of
transmission-level capacitor banks at TVA
In addition to collective monitoring of performance metrics, applied analytics can also be
important for proactive monitoring of capacitor banks as shown in Figure 1. Common
performance issues such as restrike can be mathematically modeled (as was the case in this
example) or, more recently, AI/ML algorithms can be trained to recognize these conditions. This
latter approach is now the focus of recent research.

5
Figure 1: Waveform recognition and cause identification from EPRI’s Capacitor Switching
Assessment Module as applied at TVA
Prior to implementation of a proactive PQ strategy, previous practice was essentially “operate to
failure” wherein malfunctions or performance issues were left untreated until unambiguous
failure occurred. Subsequent repair of failed capacitor banks was also expensive, typically
costing US$250K - $400K per bank.
TVA began by choosing ~20 of their transmission capacitor banks for an initial test of proactive
PQ techniques, representing less than 10% of their overall fleet. The subject banks were selected
because they already had local PQ monitoring in place. Surprisingly, and despite the relatively
small initial sample, TVA quickly identified 3 banks that were in urgent need of repair. The
causes varied widely (from previously undetected capacitor can failure, to switchgear operational
issues, to restrike), but the response was the same: rather than running to failure, the banks were
proactively removed from service and repaired, accruing estimated cost savings of over US$1
million. This approach also avoided other potential negative outcomes such as grid outages, and
possible safety and environmental issues that could have arisen. Perhaps the most ringing
endorsement was the company’s response to this success – based on this initial success, TVA’s
Grid Operations and Planning department quickly budgeted US$5 million to equip all ~300
transmission capacitor banks with dedicated PQ monitoring.
Since this initial success, TVA reports >US$10 million in additional saving via implementation
of proactive PQ techniques. In addition to further capacitor bank detection, the company has also
found:
• Multiple CCVT sensors nearing end of life. These have been particularly beneficial as
these devices have often failed catastrophically. Estimate cost savings: >US$250K per
detection in avoided repair and cleanup costs.
• Multiple detections of non-functional or mis-operating PQ monitors
• Multiple detections of mis-operating or mis-configured 3-phase revenue meters. The
detection and correction of these has produced cost recovery of many US$ millions.

6
PQ Application at Powerco Limited
Powerco installed a few PQMs 2010-2012, but as the installations lacked an ‘owner’, and with
communications and data management platforms being poorly managed, within a few years the
system fell into a state of disrepair, eventually being shut down.
In 2017 Powerco started again installing PQMs at its zone substations, but this time with an
‘owner’, and with better communications with tight security, and data management. We decided
to start by including a PQM in each incomer as protection and switchgear replacement projects
were being carried out.
Currently Powerco has 36 PQMs installed and operating at 18 zone substations. The PQMs are
polled at 20-minute intervals by cloud based database software. The data received is stored in a
PostgreSQL database and can be accessed by all Powerco’s staff. Three types of data are
collected: continuous time series, event logs, and disturbance records.
In addition to measuring and recording all typical power quality metrics to Class A standard
under AS/NZS 61000.4.30, Powerco has also used the PQM’s trigger inputs to capture detailed
records of tap changer operations and protection events, and to record some extra information,
e.g.: sequence voltages.
Some examples of how this extra data have been useful are:
• Identifying several tap changer faults,
• Allowing a more detailed understanding of fault events,
• Highlighting current transformer ratio/wiring errors, and
• Missing voltage transformer earthing.
Two examples are shown below.

Figure 2: A triggered tap change record showing unnecessary additional tap steps. This was
being caused by tap position indication contacts in poor condition.

7
Figure 3: A difference in zero sequence voltages between CB1 and CB2 PQMs was noticed.
While at the same time they showed near identical phase to phase voltage measurements. This
error was caused by an open VT star point on CB1.
Presently Powerco views its PQ data manually, screen by screen, colloquially known as
‘eyeballing’, and while we have learnt valuable lessons from this data, the increasing number of
meters and the various types of disturbances and anomalies recorded, it is becoming increasingly
difficult to analyze this way. Powerco is keen to learn of methods to automate analysis for both
power quality purposes and asset health.
Powerco joined EPRI’s power quality program in 2020 to ramp up its PQ capabilities. Powerco
has enjoyed sharing and learning in EPRI’s online discussions and demonstrations, access to
EPRI’s reports on selected topics, and conversations with other members of the program. This
year Powerco is particularly interested in the projects around data visualization and reporting,
and a guide for enterprise level implementation.

Conclusion
Electricity distributors are facing significant growth of inverter based loads and generators and
proactive PQ monitoring as recommended by Elphick, Smith and Watson in their 2016 paper [1]
is likely underway at many NZ distributors. Powerco has started work in this area.
This paper described several examples where further analysis of PQ data offers additional high
value insights into assets health issues. Power quality measurements along with statistical
analysis of trends, artificial intelligence and machine learning techniques, can provide potential
indications of incipient conditions not otherwise visible. The authors believe that if these
‘discoveries’ can be acted upon early, significant economic return, in both monetary and
intangible terms will follow.

8
References
[1] S. Elphick, V. Smith, N. Watson, “Proactive PQ Monitoring in New Zealand Electricity
Distribution Networks”, EEA Annual Conference, 2016.