Predictive Maintenance
Any machine with moving parts will eventually break
down and require service. If a pipe bursts or a con-
veyor belt jams, production lines stop, and immediate
repairs are required. These are examples of condition
maintenance. Fortunately, many system parts have
duty cycle ratings, and their replacements can be
planned and scheduled. We refer to this as preventa-
tive maintenance. It is fair to say that planned /
scheduled downtime for preventative maintenance is
preferable to unplanned downtime for condition
maintenance repairs.
Furthermore if a proactive replacement or repair can
happen simultaneously with another condition repair,
it is possible for field service teams to help avoid fur-
ther unplanned downtime and larger, more costly
repairs down the road.
On the other hand, changing parts just for the sake of
changing them on a predetermined schedule may
actually be a waste of time and capital. A part may be
rated at 10,000 hours but in practice have a life of
20,000 hours. Others may fail well ahead of schedule.
So what’s to be done? Is it worth changing that part
on a schedule, or should you wait? How expensive is
that part? What’s the lead time for procuring a
replacement? Do I need to keep one in inventory, just
in case, and pay for it as an insurance policy? How do
you win at this game, keeping your line running at top
efficiency while managing the financial risk?
These are difficult questions to answer, and they are
the raison d’ etre for Predictive Maintenance, which
seeks a data-driven, best-of-both-worlds approach to
optimally keeping the line running at reasonable cost.
Many systems and machines (although not all) have
long been equipped with sensors that track telemetry
or performance data. That data is typically raw and
without context. Some companies (although not all)
had the foresight to capture that raw data in a special
database, called a data historian. More often than
not, the data historian sits, in some cases for years, in
an un- or underutilized fashion.
Going big doesn’t mean everything
goes to the cloud.
FALL 2017 | THE DOPPLER | 7