An introduction to

predictive maintenance
for high voltage assets.
Contents
05 Introduction
06 Predictive maintenance (pdM):
the starting point

08 Investing in predictive maintenance

09 Why predictive maintenance works

11 Predictive maintenance for HV assets

13 What does predictive maintenance
for HV assets look like?

14 Predictive maintenance for HV

assets in action 16 Conclusion
Introduction
Predictive maintenance uses data-driven, proactive maintenance methods to analyse
the condition of equipment, the aim being to predict when maintenance should be
performed. Today, that means using data science and predictive analytics to assess
when equipment is likely to fail, so that problems can be rectified before they reach a
critical point and result in unscheduled downtime.

Predictive maintenance began to gain ground in the 1990s, technologies available, it’s hard to disagree with the statement
when monitoring the condition of mechanical equipment that running equipment to failure is an outdated approach.
through techniques such as vibration analysis started to
be adopted. Other factors, such as heat, pressure, noise Essentially, predictive maintenance is all about getting to
and lubricant condition, are also factors for machinery with the heart of the root cause of any potential issues, predicting
moving parts. But what about power systems? time to failure and maximising uptime. The clear advantage
of predictive maintenance for high-voltage power systems
is that catastrophic failure can be avoided, avoiding costly
While voltages, currents, power, temperature and
disruption, emergency maintenance and even loss of life.
electromagnetic quantities have always been monitored,
combined with regular surface inspection and material quality In this concise ebook, you’ll gain an understanding of
testing, the limitation of this approach is the inability to look predictive maintenance as a strategy for protecting high-
forward into the future.1 Ultimately, that means that the first voltage assets and how techniques and software have
time a problem becomes evident is often when the power evolved to offer new actionable insights that can save time,
system goes down. In the 2020s, with a raft of diagnostic money and, potentially, lives.

1 Condition-based predictive maintenance of industrial power systems, Mohammad Azam; Fang Tu; Krishna R.
Pattipati, 2002, published in SPIE Proceedings Vol. 4733: Component and Systems Diagnostics, Prognostics, and
Health Management II Peter K. Willett; Thiagalingam Kirubarajan, Editor(s)

5
 Predictive
“Periodic testing is certainly better than no testing, but it isn’t
perfect. Serious loss can occur if an undetected problem

maintenance (PdM):
reaches catastrophic failure in the period between surveys.
Obviously, if you test more frequently, the chance of failure

the starting point

goes down. However, at some point, that becomes too labour
intensive and financially onerous.2”

While preventive maintenance brings many benefits over In contrast, predictive maintenance is entirely directed.
To quickly breakdown approaches to maintenance, these can be divided into one of four categories: reactive, proactive, reactive maintenance and moves a step further on from It uses a combination of equipment monitoring and data
preventive and predictive – in effect, a hierarchy of maintenance. The traditional, now outdated, reactive approach sits a proactive approach, it is essentially inefficient, with the analysis tools and techniques to detect problems in
at the bottom of the pyramid, and has been largely superseded by more insightful methods in many sectors. risk of personnel becoming ‘busy fools’ through arduous processes and defects in equipment so that these can be
maintenance schedules. Stating the obvious, maintaining put right in good time – before a catastrophic failure. To do
a safe, uninterrupted power supply is the critical priority for so involves combining data from connected devices and
the energy sector. When the focus is purely on preventive machines with remote monitoring and predictive analysis,
maintenance, the reliance on carrying out regular checks and taking into account any automated maintenance orders.
maintenance loses sight of that goal. Activity is focused on
Use data from sensors to predict when
PREDICTIVE maintenance is needed scheduled, calendar-based inspections that may be neither Not all predictive maintenance has to be fully automated.
necessary nor timely. Moreover, these planned inspections For machinery, predictive maintenance can also be carried
may miss critical warning signs at the point when there is out manually through equipment checks using handheld
Schedule maintenance: time estimates time to address them with minimal disruption. vibration or audio sensors.
PREVENTIVE on parts/OEM recommendations

Identify and address the problem

PROACTIVE causing issues

Run equipment to failure, fix the issue

REACTIVE when it occurs
2 Continuous monitoring is on the rise: can you adapt, William Higinbotham, EA Technology LLC, NETAWORLD, Fall
2018Prognostics, and Health Management II Peter K. Willett; Thiagalingam Kirubarajan, Editor(s)

6 7
 Investing in predictive Why predictive
maintenance maintenance works
To gain a true picture of the cost-benefit gains to be made Predictive maintenance removes the guesswork. It’s data-based and relies on continuous
monitoring, which provides an accurate view of exactly what’s taking place.

25-30%
through the real-time monitoring of assets’ condition and
data using technology and software, there are several key
factors to take into account:
Creating a historic record of power quality to spot trends. Continuous monitoring of Partial Discharge is essential.
Time spent maintaining equipment This could be anything from a surge in the amount or This serves to ensure detection of degradation to insulation
Production hours lost to maintenance REDUCTION IN MAINTENANCE seriousness of dips or a growth in harmonics. The resulting that could cause premature failure of equipment such as
The cost of spare parts and supplies maintenance averts unscheduled shutdowns, which can switchgear and cables.

70-75%
Collateral costs in the event of catastrophic failure bring significant financial benefits in keeping operations
running. Understanding what went wrong. Continuous monitoring
It’s fair to say that the entire cost of installing a predictive means that you have the data to unpick the causes of power
maintenance system would quite possibly be less than the Using data to plan ahead. Providing data to determine the disruptions. Knowing how and where problems happened is
costs incurred by a catastrophic failure event. In fact, data loading of existing transformers, backup supplies and for the first step in understanding how to manage the outcome
from the U.S. Department of Energy indicates that predictive DECREASE IN BREAKDOWNS planning future expansion. All of this is incredibly helpful and avoid a repetition.4
maintenance can generate an impressive potential return in assessing the appropriateness of budget allocation for
return on investment. It’s not just about preventing catastrophic failure. Using

35-45%
on investment of roughly ten times the cost, with a 25-30%
reduction in maintenance costs, a decrease in breakdowns predictive maintenance technologies, power utilities and
of 70-75% and a 35-45% reduction in downtime.3 Optimal identification of problems. The right people can private network operators can discover which assets are
receive instant alerts as soon as a power problem occurs. underperforming. Operating staff can then understand
Immediate action can isolate the issue and stop a knock-on the factors leading to abnormal operations and plan in
In short, weighing up all the permutations of what could
maintenance to address the issues. Predictive maintenance
go wrong with any given HV asset, as well as the ability REDUCTION IN DOWNTIME effect that could put an entire facility or network at risk.
is extremely efficient and enables resources to be directed to
to replace equipment at the right time, the justification exactly where they can bring the most benefit.
for preventive maintenance is not hard to demonstrate.
3 Operations & Maintenance Best Practices: A Guide to Achieving Operational Efficiency, G. P. Sullivan, R. Pugh, A. 4 Voltage Dips: Predictive Maintenance – The Key to Power Quality, European Copper Institute (ECI), David Bradley
P. Melendez, W. D. Hunt, August 2010, Prepared by Pacific Northwest National Laboratory for the Federal Energy Rhopoint Systems Ltd March 2001 (Version 0b November 2001)
Management Program U.S. Department of Energy

8 9
Predictive maintenance
for HV assets
So how does all of this apply to high-voltage assets?
The consequences of failing to prevent high-voltage equipment malfunctions can be dire. Of course, replacing the equipment
that has failed can be extremely costly. But the bigger picture is even more crucial: the domino effect of power outages in terms
of downtime, inconvenience and hazards impacting the public and the environment can add up to an exponentially costly error
if essential repairs are missed. The associated reputational damage, while hard to quantify, can be long lasting.

In an ageing network, directing maintenance in a very focused way can help to avoid the challenges of time-based preventive
maintenance that relies on the availability of technicians. Collecting and entering data manually is limiting and lacks the speed
needed to monitor asset health in real time.

11
 What does predictive maintenance for
HV assets look like?
A high voltage monitoring system can utilise remote monitoring of HV assets. Breaking it
down simply, these are some of the main components involved in predictive maintenance
for high voltage assets:

Partial discharge (PD) monitoring HV system component monitoring

Left unaddressed, partial discharge can erode solid insulation, Whether monitoring is continuous or scheduled,
and the eventual outcome can be breakdown and failure. components under scrutiny should comprise:
PD monitoring assesses the health of your insulation,
replacing periodic PD surveys. Cable systems, including cables, joints and terminations,
underground cables and overhead lines. Monitoring
Signal analysis of the partial discharge activity measured should look for partial discharge in the cable network.
includes localisation, trending and interpretation of
suspicious spikes. In partial discharge testing, sensors Switchgear, open and closed cells, air- or gas-insulated
pinpoint voltage stresses caused by the system voltage on switchgear (AIS and GIS) installations.
deteriorated insulation components. Deteriorated insulation
produces signals with partial discharge characteristics. Current and voltage transformers, connections
and bushings.
PD sensors pick up on these small electrical signals or
discharges in or on the surface of the electrical asset’s In all cases, forewarning via a predictive system allows
insulation. the time to plan the most efficient repair shutdown,
ensuring the safety of both staff and physical assets.

EA Technology Astute HV Monitoring®: Learn more

12 13
 Predictive maintenance for HV assets in action
Online monitoring of high voltage apparatus is a vital feature of a comprehensive predictive maintenance program. When a
predictive maintenance approach is fully implemented, the care of connected assets – whether that be wind farms, substations
or pumps, for example – is managed by the computerised maintenance management system (CMMS).

While it’s possible to create piecemeal solutions by monitoring individual assets, it’s important to remember that there are far
greater advantages to taking a holistic approach to creating an integrated system.

Condition monitoring based maintenance “The greatest benefits of

As predictive maintenance evolves, condition monitoring based maintenance (CDM) offers even deeper insights,
predictive maintenance are
dealing with the entire system as a complete entity. This moves predictive maintenance away from a ‘mix-and-
match’ approach, allowing the full power of an integrated system to uncover risks that, left to human identification,
might otherwise go undetected.7
realised when multiple monitoring
instruments are placed at various
critical locations on the electrical
infrastructure to continuously
monitor power.6”
7 Maintenance Management of Electrical Equipment (Condition Monitoring Based), Ashok Parikh, August 2014, 6 D. Bradley, “Applying predictive maintenance to power quality,” 2001 Twenty-Third International Telecommunications
Electrical Engineering Portal Energy Conference INTELEC 2001, Edinburgh, UK, 2001, pp. 229-237.

14 15
 Conclusion
Embedded predictive maintenance solutions secure and
improve end users’ electricity supply by enhancing the
reliability of high voltage assets, including substations. Using
a combination of mathematical, logical and statistical analysis,
it’s possible to monitor site performance remotely, detect faults
and alert users about required performance improvements.
The result: minimised downtime and optimal repairs through
preempting asset failure, increased safety and the avoidance
of catastrophic failures. Whether you are in the early stages
of implementing predictive maintenance in a private network
or exploring the whole-life costs of an integrated system, the
investment involved in predictive maintenance for HV assets
is more than offset by the many gains.

16 17
Safer, Stronger, Smarter Networks
EA Technology Pty Ltd t +61 (0) 7 3256 0534
381 MacArthur Avenue, Hamilton e [email protected]
QLD, 4007, Australia www.eatechnology.com/australia