Page 32 - Nuclear Plant Journal - March April 2013
Basic HTML Version
On-line
Monitoring
By Richard Rusaw, Electric Power
Research Institute.
1.
How will the online monitoring
system help predict equipment/structure
failure in advance before it becomes a
problem?
Online
equipment
condition
monitoring can provide early warning of
potential failure by detecting incipient
indicators of equipment degradation via
advanced pattern recognition and signal
processing technologies. Today’s on-line
monitoring (OLM) systems/programs are
built around pattern recognition software
that models normal system or component
behaviors by evaluating real-time
operational conditions against the model’s
predicted behavior with a high degree
of sensitivity. Any significant deviation
from predicted behavior is identified
well in advance of any traditional form
of condition monitoring. This capability
allows plant operators to take actions
before degrading conditions result in
damaged or failed systems, structures, or
components.
Toprovideearlywarning,commercial
OLM products rely on a continuous input
of well-correlated plant data, typically
available only for active assets. After
the initial warning, an investigative
review can identify the actual failure
mode and cause and suggest corrective
actions. The investigative review can
involve plant staff as well as experts in
predictive maintenance. In these cases,
the diagnostic process can be manually
intensive.
For utilities using OLM, a more
automated diagnostic and prognostic
capability could directly identify
equipment condition from the “signature”
of the initial warning. This capability
would not only reduce manual resources
required for diagnosing failures, but also
improve the effectiveness of condition
corrections. A more timely response
can, in some cases, significantly improve
safety and prevent collateral damage.
EPRI is currently developing
diagnostics and prognostics capabilities to
address these gaps in plant asset condition
assessment. A Fleet-wide Prognostics
and Health Management (FW-PHM)
software suite currently in Beta testing
will provide an industry-wide database of
failure mode indicators for a wide variety
of power plant assets. The software
includes a “reasoner” that uses this fault
signature database for assessing current
plant information to determine the likely
cause for any anomalies detected in the
plant data. The diagnostic advice included
in the software can reduce the time and
resources required for troubleshooting and
diagnosing impending failures, thereby
increasing the amount of time available to
respond with a corrective action. Further,
the suite’s prognostic models can provide
a sound estimate of the remaining time
available before a degrading asset reaches
a critical condition. This information can
be used to prioritize and plan responsive
actions and can assist plant personnel in
more frequently assessing the potential
impact of a developing problem.
2
.
How will the online monitoring
system help reduce human errors?
Monitoring programs provide a
means for accurately assessing asset
conditions, potentially enabling users
to reduce operating and maintenance
costs by moving from a time-based
preventative maintenance program to an
asset condition-based program. Industry
experience has shown that time-based
programs can result in equipment being
serviced well in advance of condition
requirements. Condition-based programs
can avoid unnecessary maintenance,
reducing opportunities for human errors
during maintenance of plant assets.
An additional means of error reduction
results from the application of monitoring
programs to quickly identify the proper
equipment failure modes and causes of
active component failures. Traditional
forms of diagnostics on equipment
faults provide a higher probability
environment to introduce human errors.
The elimination of potential root causes
with automated assessment or diagnostics
reduces the potential for human error
during corrective maintenance activities.
3.
What are the main technologies that
have been used in the online monitoring
system?
Advanced Pattern Recognition
Software
- Centralized online monitoring
is a highly automated condition analysis
Overview of the PHM Suite
32
Nuclear Plant Journal, March-April 2013
Response to questions by Newal
Agnihotri, Editor of Nuclear Plant
Journal.
