Efficient Plant Nov./Dec 2019 - 22

feature | predictive maintenance

Predictive strategies
indicate a bearing
problem far enough
in advance to allow
maintenance personnel
to plan for repair or
replacement rather than
react to a failure.

tions and component
replacements occur at a
specific time. Maintenance crews perform
extensive checks on
specific assemblies
and components
during a scheduled outage and
another set of tests
during another
maintenance routine. Inefficiencies
are inherent with this
process because of the
need for increased staff
or because of the unneeded
replacement of parts.
While both methods can work,
rising costs and the complexity of modern
equipment have pushed companies to consider a
predictive-maintenance (PdM) strategy. Because of
the need for reliable operation and cost containment,
PdM strategies that monitor key assets and provide
early detection of bearing problems help managers
plan for outages and have the required resources and
components on hand, when needed. This approach
maximizes the useful life of assets, eliminates unnecessary maintenance that occurs with PM strategies, and allows workers to perform maintenance
in less-stressful conditions, thus increasing safety.
This approach also plays a major role in significantly
increasing overall operational reliability.

HOW DOES PDM WORK?
Predictive maintenance creates the intelligence needed to predict bearing failures that would otherwise
lead to 70% to 85% of downtime for equipment. Rather than repair equipment after a failure or conduct
tests because of a schedule, PdM compares actual
operating conditions against a baseline consisting of
average bearing operating conditions.
Many PdM systems include libraries of detailed
information about thousands of bearings and assemblies and provide methods for finding the brands/
models used in a particular plant. The component

| EFFICIENTPLANTMAG.COM

information housed within a library includes vendor
specifications, known failure causes, and the type of
inspection technology used to identify a problem.
The historical trends for the components become
part of the comparative analysis of actual and average
conditions.
Sensors and transducers attached to the x- and
y-axes of bearings compare changes in vibration
frequencies that occur during real-time operation
against the baseline operating conditions. Intelligent
sensors can record real-time vibration data, establish
baseline information, and create alarm thresholds. If
vibration exceeds set limits, notifications inform the
reliability team that a bearing failure is likely in the
near future.
Advanced PdM solutions use machine-learning
algorithms to determine the cause of increased vibration by interpreting raw vibration data. Some systems
go beyond simple notification and prescribe potential
resolutions.

PDM TOOLS
When working with bearings, it's understood that
certain operating conditions produce common types
of failures. For example, poor lubrication of rolling
elements can cause different types of bearing damage,
such as spalling, while vibration can result in fretting
or damage to retainers. Given the range of failures
and causes that exist, predictive maintenance relies on
a variety of analysis tools.
Vibration monitoring and measurement is the classic starting point for bearing PdM strategies. Using a
motor operating at a constant rotational velocity as
an example, the rotational speed of the inner race
establishes the angular velocity of the bearing elements. When a fault appears on the external race,
movement of a ball past the fault causes an impact
that, in turn, generates a mechanical vibration peak
and a unique, identifiable frequency.
Vibration monitoring detects the higher frequencies emitted by a defective bearing through sensors
located close to the bearing. When considered as
part of a trend, the frequencies indicate a deteriorating condition. Spalling, for example, generates a
narrow-band frequency spectrum that features an
increase in the amplitude of defect frequencies and

NOV/DEC 2019

http://www.EFFICIENTPLANTMAG.COM

Efficient Plant Nov./Dec 2019

Table of Contents for the Digital Edition of Efficient Plant Nov./Dec 2019