Efficient Plant October 2017 - 13
feature | reliability & profitability
Peter G. Martin, Ph.D., Schneider Electric
for more than 100 years. The primary objective has always been to
safely increase plant production. Originally, single-loop feedback
control was the preferred method, but it has been replaced by stateof-the-art process control in the past fifty years. Today, coordinated
multiple-variable approaches, coupled with dynamic-process models,
have enabled some very sophisticated predictive-control strategies.
These advancements in process control have enabled manufacturers to continually increase operational throughput. However, there
is an inherent risk in doing so. As industrial assets are pushed to
deliver more, they move closer and closer to their reliability and
safety thresholds. As a result, today's assets are under continuous
strain that is degrading their reliability and affecting overall
To counter that risk and alleviate the strain, industrial-maintenance tools and practices, intended to improve
asset reliability, have progressed and evolved over the past
two decades. Classic break-fix models, otherwise known as
reactive maintenance, have been expanded first to preventive
maintenance, then to predictive maintenance, and finally to
prescriptive maintenance. Each of these advancements led to
a corresponding increase in asset reliability. But manufacturers were soon stuck in a cycle because, even as advanced tools
and techniques were being applied to improve asset reliability,
process control became more sophisticated, fighting reliability
improvements every step of the way.
It turns out that more advanced technology isn't what we need.
What we really need
to do is rethink how we
address this age-old issue,
and that begins with how we
measure asset reliability in the
Empowering today's workforce with
profitability data and
information will turn
them into business-performance managers.
In the past, measuring the reliability of
industrial assets was limited to analyzing
historical performance. A much more
effective approach is to directly measure
how likely it is that a reliability incident
will occur, i.e., determine reliability risk.
Data-science advancements and the
proliferation of condition and process
measurements in industrial operations make it practical to perform
direct, real-time measurement of asset reliability. Such measurement,
in turn, allows real-time asset-reliability control. Based on extensive
laboratory testing and actual in-plant experience, there is already
considerable information on reliability at the equipment asset level.
For example, accurate reliability curves, coupled with condition and
process measurement, make it possible to precisely measure asset-reliability risk (See Fig. 1 on p. 14).
Once the real-time reliability risk of equipment assets is measured,
it's a small leap to measuring the reliability risk of higher-level units