This presentation describes the value of equipment health anomaly detection and diagnostic systems, such as MaintenanceOpt®, in alerting plant personnel to equipment problems and expediting problem diagnosis and remediation. Several examples are provided including high ID fan motor temperature, low oil on a fan bearing, turbine bearing vibration, and condenser air evacuation issues.
1. Great Catches:
How Equipment Health
Detection & Diagnostic
Systems are Improving
Plant Availability
Coal Gen 2010
Ray Johnson
NeuCo, Inc
2. Optimization Software & Availability
Optimization helps improve availability/
reliability by:
Using predictive analytics to identify potential
equipment health issues
Providing early warnings for impending failures
Minimizing unnecessary outages
Sponsoring better use of planned down-time
Reducing tube leak outages
Proprietary and Confidential
3. MaintenanceOpt®
What it does:
Automatically alerts users to symptoms of potential equipment health
issues (efficiency, capacity or reliability based) and allows quick easy
drill down to uncover actionable responses to diagnose and remedy
the underlying problems.
Using:
Neural Networks
Expert Rules
Other Optimizer Alerts
To Improve:
Unit up-time
Speed of issue resolution
Efficiency
Proprietary and Confidential
4. Equipment Types
MaintenanceOpt may be used to detect problems in a fossil
generating power plant associated with the following types of
equipment:
Air heaters
Auxiliary equipment (ESP, FGD, SCR, crusher, etc.)
Boiler
Condenser
Fans and drives
Feedwater heaters
General (flows, heat rate, etc.)
Pulverizers
Pumps
Turbine
Proprietary and Confidential
5. Examples of Potential Problems
The types of problems that can be detected are
dependent upon the type of equipment. For example, for
air heaters, the following provides a list of potential
problems that may be identified:
Inaccurate leakage calculation due to stratification or instrument error
g
Basket damage
Ash or salt plugging
Bearing inadequate oil cooling
Shaft misalignment
Bearing failure
Proprietary and Confidential
6. Indicators of Potential Problems
The following process values associated with an air
heater are monitored as part of an early warning system:
High AH air or gas side pressure drop
High AH cold end average temperature
Low AH cold end average temperature
Air Heater leakage increasing
Increasing APH coil differential pressure
Bearing temperature increasing
Proprietary and Confidential
7. Problem Identification
Model is used to predict equipment performance
Model output is compared to actual performance
If difference is larger than a threshold, an anomaly is
detected
Proprietary and Confidential
8. Diagnosis Flow Chart
A flow chart for diagnosing
a potential problem is
shown
Proprietary and Confidential
10. NeuCo Customer Support
Answer any questions related to the products
Provide software updates
Update system configuration
Provide weekly reports
Monitor for high urgency alerts
g g y
Monitor performance of the optimizers
Provide additional training
and more…
Proprietary and Confidential
11. Automated Alerts Management
ProcessLink at Site Customer Support
Customer
Email Email
Trigger Email Support
Server Server
Service Composer Interface
at Site at NeuCo
Composite
Historical Trigger
Data Database
Proprietary and Confidential
12. Automated Reporting Helps
Extract Maximum Value
KPIs Reflect Your Asset Management Metrics
EFOR/EFOR-D
Heat Rate (rolling average, vs. targets, etc.)
Emissions (cumulative tons, rolling average, vs. targets, etc.)
Revenue (vs. targets, historical benchmark, etc.)
Profitability (gross & net margins, vs. target, etc.)
Built-In Benchmarking
B ilt I B h ki
What benefits have been achieved?
Additional benefits that could be achieved
Changes in performance over time and implications
Demystification Sheds Light on Key Processes
What optimizers are doing and why
Where the most important trade-offs are occurring
How are constraints limiting benefits and what can be done?
Proprietary and Confidential
14. Example Recent Catches
High ID fan motor temperature
Low oil on a fan bearing
Condenser air evacuation problem
Increasing turbine bearing vibration
g g
Proprietary and Confidential
15. MaintenanceOpt: ID Fan Motor Temperature
Unit: PRPA Rawhide Unit #1, 300 MWs.
Usage: The following problem was identified
during a routine review of MaintenanceOpt
alerts.
Incident: The temperature of an ID Motor
Stator Temperature had increased over a
p
several week period. This issue was
recognized by Customer Support. Early
warnings were sent to the client about this
issue. Based on the warnings, the client
cleaned the filter and performance returned to
normal.
16. ID Fan Problem
ID Fan problem is shown on the MaintenanceOpt
Home page.
Proprietary and Confidential
17. Issues to Screen
The ID Fan problem is also shown on the MaintenanceOpt
Issues to Screen page along with the context data.
Proprietary and Confidential
18. Alert Data
Reviewing the alert data, the problem had started three
weeks earlier and had been getting progressively worse.
The client had been notified about this issue.
Proprietary and Confidential
19. Diagnosing the Problem
By clicking “show causes”, we are able to investigate the
potential causes of the problem. The rule for Dirty Filter is
shown below. Variables associated with this rule are highlighted
including data for the other ID stator motor.
Proprietary and Confidential
20. Identifying the Problem
The temperature of both ID fan motor stators are analyzed. The
temperature on ID Fan 102 is increasing while the temperature
on ID Fan 101 is normal. This is consistent with the rule for a
dirty filter.
Proprietary and Confidential
22. Verifying the Problem
An operations crew inspected the filter. It was found
to be dirty and is thus subsequently replaced. The
temperature returned to normal.
Proprietary and Confidential
24. Low Oil Example
Unit: NRG Limestone Unit 2, 913 MW
Users: Maintenance Supervisor and Performance
Engineer.
Usage: Both work 4/10 shift Monday to Thursday.
They check MaintenanceOpt each morning when
they are at site.
Incident: Low oil on a primary air fan bearing
system was found on Monday during routine login to
MaintenanceOpt.
Fix: Oil was added to the bearing system several
hours after the problem was identified. The
temperature returned to normal.
25. Problem Alerted by MaintenanceOpt
In a routine login to MaintenanceOpt, the users find an alert for a
bearing on a primary air fan. The bearing is at 174 degrees while
it is expected to be at 150 degrees.
26. Checking Alert
Chart shows the actual temperatures for all bearings on this fan.
Also shows expected value for the bearing of interest. Only the
temperature of bearing associated with the alert looks abnormal.
27. Actual vs. Expected
Actual and expected (neural network) value of applicable bearing
temperature shown. Incident starts Friday afternoon at around 1pm with no
maintenance personnel on site. Value stayed under the points DCS alarm
trigger value. This view is from Monday morning during a routine check.
28. Diagnosing the Problem
One likely cause is inadequate lubrication or cooling. Rule clauses are
highlighted below. Temperature is increasing but vibrations are normal.
29. Fixing the Problem
Customer fixes the low oil problem within hours of finding it. Temperature
returns to normal.
30. Closing the Alert
The alert is closed and a comment is posted in MaintenanceOpt’s database.
User
Comment
31. Turbine Bearing Vibration
Unit: APS Four Corners Unit #4, 800 MW
Usage: Routine checks of MaintenanceOpt made
daily
Incident: During a routine check, it was observed
that a turbine vibration was increasing. No
immediate action was necessary but problem needed
to be
t b monitored d il
it d daily.
Fix: Problem was fixed at the next outage
32. Vibration Alert
Vibration on bearing #7 has triggered. Note that alert had been
triggering >50% of the time over the past week and month.
Proprietary and Confidential
33. Increasing Vibration
Turbine bearing vibration increasing for past month (green). Neural model
adapting to the change but not as fast as vibration increase (blue).
Proprietary and Confidential
34. Monitoring the Problem
User has flagged problem so that it stays on the Issues to Screen tab. The
user also entered a comment. Monitoring continued until next outage.
Flagged
gg
item
User
Comment
Proprietary and Confidential
35. Benefits of ProcessLink® Integration
PerformanceOpt indicates equipment degradation through high-fidelity
model of thermal processes
MaintenanceOpt indication of equipment degradation through empirical
models of on-line signals as well as potential problems detected through
manual inspection
Some problems that would escape notice through one or the other
methodology become obvious when viewed through both
MaintenanceOpt also surfaces problems detected by NeuCo’s closed-
loop optimizers
PerformanceOpt informs closed-loop optimizers though virtual on-line
analyzers indicating important boiler performance parameters not directly
measured
Coal quality, boiler efficiency and heat rate for CombustionOpt
Boiler cleanliness and heat rate for SootOpt
Proprietary and Confidential
36. Condenser Problem
Unit: APS Four Corners Unit #2, 190 MW
Usage: During a check of PerformanceOpt and
MaintenanceOpt by NeuCo, we noticed a large
change in condenser performance. Plant was notified
via email of the problem.
Incident: During routine maintenance on the
condenser, th maintenance crew h d i d t tl
d the i t had inadvertently
disconnected both vacuum pumps causing a large
increase in backpressure
Fix: The maintenance crew was notified of the
problem and then properly connected a vacuum
pump. Th.e performance returned to normal
37. M’Opt and P’Opt In Action
Routine review of triggers show alerts on condenser backpressure and
cleanliness have triggered. Backpressure is at 3.05 when expected to be
2.2, and cleanliness is at 0.55 when expected at 0.88.
Backpressure
Alert from M’Opt
Cleanliness
Alert from P’Opt
Proprietary and Confidential
40. Diagnosis
Low condenser cleanliness and high condenser backpressure
indicates inadequate air removal equipment performance.
Proprietary and Confidential