Intelligent Malfunction Prognostics From equipment condition monitoring to optimal asset management...
-
Upload
marlee-tuckett -
Category
Documents
-
view
215 -
download
0
Transcript of Intelligent Malfunction Prognostics From equipment condition monitoring to optimal asset management...
Intelligent Malfunction PrognosticsFrom equipment condition monitoring to optimal asset management
EWEA Annual Conference, Brussels, Belgium, March 14-17, 2011
Copyright © 2011 by Cassantec Ltd. 2
There are many CMS for WT on the market, differing in their functional scope, WT component focus, learning capabilities and life cycle stage
Solution Profile
Scope Monitoring (Predictive) Diagnostics Prognostics
Learning Manual Automated, unit-level Automated, fleet-wide
Stage R&D Validation 100sInstallation 10s 1000s
Focus Pitch Gearbox ConverterDrive Generator Yaw
Intelligent Malfunction Prognostics
Copyright © 2011 by Cassantec Ltd. 3
Intelligent malfunction prognostics can be provided through reliability reports, supporting critical decisions on maintenance scope and schedule
a
b c
d
e
f
g
h
Reliability Report
a View condition diagnostics
b View malfunction diagnostics
c View malfunction prognostics
d Aggregate prognostics
e Cross-check maintenance plan
f Extend condition data sources
g Extend malfunction modes
h Extend prognostic horizon
Intelligent Malfunction Prognostics
Copyright © 2011 by Cassantec Ltd. 4
Monitor power output and technicalcondition of each unit in the fleet
Review diagnostic insights for units in marginal or critical condition
Use prognostic foresight to optimize fleet maintenance process
► Identify and avoid unnecessary preventive measures and costs ► Anticipate malfunctions before failure, damage, foregone output► Realize a commercially optimal fleet maintenance schedule
Unit View
Reliability reports aggregate to a fleet level, with navigation functions, consolidating diagnostic insight and prognostic foresight for several units
Unit vs. Fleet ViewFleet View
Intelligent Malfunction Prognostics
Copyright © 2011 by Cassantec Ltd. 5
Crucial condition data is captured through vibration and lubricant sensors, and directly uploaded into the WT controller via standard protocols
Nacelle
Rotor hub
Slow rotating shaft
Fast rotating shaft
Rotor bearing
Bearing
Tower
Generator
Gearbox
Blade
Pitch
WT Server
Foundation
R2R1
E1
SCADA controller
Sensor controller
T,R,E
Ethernet Switch
T2
T1
T4 T5T8T6 T7
Wind Turbine
Yaw drive
V,L
V1
V2V4 V5
V8
V7
V6
Hardware Package
V1
L1
Very low frequency accelerometer High sensitivity & accuracy Latest-generation technology Armored integral cable
V,L
Versatile Profibus terminal Easy plug-in installation Straightforward configuration Meeting OEM standards
Inline twin laser particle counter Latest-generation technology Integrated humidity sensor Stainless steel block
V8…
Intelligent Malfunction Prognostics
T3
V3
L1
Copyright © 2011 by Cassantec Ltd. 6
• Hosting Network Mgt and PLC Controller• Consolidation of SCADA data from all WT• Consolidation of additional sensor data• Forwarding of consolidated data batches to Cassantec
Wind Park
• Gather malfunction and failure statistics
• Inform suppliers of components affected
• Improve quality of WT components affected
• Ascertain state-of-the-art prognostic solution
• Report Review • Maintenance &
Service Scheduling
• Report Review• Asset Mgt. Decisions• Spare Part Mgt.• Capacity Forecasts
Further Wind Park • Data Mgt & Archiving• Condition Monitoring• Malfunction Diagnostics• Failure Prognostics• Intelligent Reporting
Reliability reports are updated with new condition and process data in periodical intervals, and delivered to the operators on-line via reliability portal
1
2
3
4
5
6
7
8
Etc.
ALAN / Ethernet or similar
B
Router, Firewall
Internet
Cassantec Server
WT Manufacturer
WP Operator
WP Insurer
Download batches of condition and process data
(V,L,T,R,E) for all WT in regular intervals
Upload WP Reliability Reportsin corresponding intervals
WP Service Providers
ISDN, ADSL, or similar
Further Wind Park Etc.
WP Server
WT Server
Fleet Server
WP = Wind Park, WT = Wind Turbine
Intelligent Malfunction Prognostics
Data Transfer
Copyright © 2011 by Cassantec Ltd. 7
We have calibrated and validated our reliability reporting solution with off-line and on-line data from several wind farms predominantly in the U.S.
Intelligent Malfunction Prognostics
Example
Wind Farm: Buffalo Ridge near Alta, IA, U.S.A.
WF Capacity: 150 x 750 kW = 112.5 MW
WT Models: Zond Z-46 (now GE)
Sampling period: 2006 – 2010 (on- & off-line)
Sampling intervals: continuous to 6 months
Malfunction modes: e.g. Gearbox LS wheel wear
Causes: e.g. Micro pitting, contributed by water ingress
Impact: e.g. Bearing life reduces by factor 3
Learnings: ► Upgrade sensor hardware► Monitor condition dynamics► Exploit fleet intelligence
Field Validation of SolutionIllustrative
Map source: www.google.comLogo source: www.altaiowa.com
Copyright © 2011 by Cassantec Ltd. 8
We achieve malfunction and failure prognostics over an explicit time horizon exceeding the best “predictive diagnostic” approaches on the market so far
Prognostic Horizon
Prognostic horizon[Days after last update]
00 1 10 1.000100
Value addedby reliability report
$
$$
$$$Our currentcapability
Potentialfuture
capability
EquipmentProcurement
& Replacement
Work OrderScheduling
MaintenanceCycle
Scheduling
UnscheduledOutage
Coordination
RoutineMonitoring
Competitor capabilities
Intelligent Malfunction Prognostics
Copyright © 2011 by Cassantec Ltd. 9
Benefits of reliability reports have been confirmed by operators beyond wind power – these benefits increase over time through machine learning
Prognostic Accuracy
► In retrospect, 99% of predictable malfunctions were accurately predicted, with a horizon of up to 5 years (!)
► Operator knowledge was exceeded by 20%, with several surprises (e.g. cartridge sealing)
► Diagnostics und prognostics are enhanced over time through machine learning
June 2010 July 2010March 2009August 2008 April 2007
Cartridge sealsMech. seals
Cartridge sealsMech. seals
Cartridge seals CouplingAlignment
CouplingAlignmentMech. seal
Nov.OK
Intelligent Malfunction Prognostics
Copyright © 2011 by Cassantec Ltd. 10
Example for learning value bands
Machine Learning
Value bands must be continuously „learned“ from the empirical condition data: Even with constant equipment utilization, value bands may shift over time!
Collective learning process for equipment of same type
(flagging before adjusting)
Intelligent Malfunction Prognostics
„Normal“ value bands
shift and evolve
Static value
bands not useful
Copyright © 2011 by Cassantec Ltd. 11
This learning process is initialized at different parameter value levels – gearbox oil has fluctuating initial levels of cleanliness, mostly within tolerance intervals
Gearbox oil is rarely “clean” to begin with:units start
up with varying levels of
initial contami-
nation
Learning Process Initialization
Learning process initialization for equipment of same type
Intelligent Malfunction Prognostics
21
18
16
Example for flexibleinitialization
Copyright © 2011 by Cassantec Ltd. 12
In summary, we are targeting new features allowing commercially optimal fleet maintenance schedules, cutting costs of failure, damage and lost power output
State-of-the-art sensor hardware
► High-end specialized sensors for wind power applications
► Integration of latest technologies (e.g. twin laser particle counters)
► Full utilization (and no duplication) of existing data and infrastructure (SCADA)
Intelligent diagnostics
► Comprehensive expertise on model-specific malfunction and failure sources and risk
► Automated learning from ongoing monitoring of the entire fleet
► Reference data from other WT, fleets, applications
Advanced prognostics
► Extended prognostic horizon through computational stochastic model
► Full utilization of recorded and archived condition and process data histories per WT
► Prognostic accuracy exceeding capabilities of any competing product on the market
Cost-effective advice on optimal fleet asset management
► Reduction of risk and costs for WT malfunction, failure, damage and foregone power output
► Reduction of risk and costs of unnecessary preventive measures and foregone power output
► Realization of a commercially optimal condition-based fleet maintenance schedule
State-of-the-art
Enhanced
Enhanced
New
Technical & Commercial Target Benefits
Intelligent Malfunction Prognostics
Copyright © 2011 by Cassantec Ltd. 13
Further Information
For further information, please review our brochure on-line, and contact us by e-mail or telephone
Cassantec team behind this presentation
Frank KirschnickZurich, Switzerland
Heinz GiovanelliMunich & Zurich
Gary EllisCleveland, Ohio
Shuang YuanZurich, Switzerland
Mart GrasmederCleveland, Ohio
Katerina StamouZurich, Switzerland
Mila VodovozovaZurich, Switzerland
► To obtain more information, please download our brochure at
www.cassantec.com/wind.pdf
► Or send an e-mail to
Intelligent Malfunction Prognostics
Copyright © 2011 by Cassantec Ltd. 14
Intelligent Malfunction Prognostics
Appendix
Copyright © 2011 by Cassantec Ltd. 15
Cassantec is an independent provider of integrated, automated prognostic services for critical power plant equipment with a unique, protected technology
Meaning: Cassantec = Cassandra Technologies
Position: Independent provider of integrated, automated equipment condition diagnostics and malfunction prognostics
Technology: Novel combination of best practice techniques from Operations Research, Artificial Intelligence and Data Mining
Comprehensive condition data reference base (since 1993): 500k data sets of 20 equipment types, 2000 models
Offering: Online Condition Monitoring Systems and Reliability Reports on a subscription basis for equipment operators worldwide
References: Chemical and Power industries (U.S.A. and Europe) including nuclear and fossil-fired power plants and wind farms
Promoters: Power corporations, private investors, Swiss government (CTI)
Industry Partner: Leading independent U.S. lubricant lab (Insight Services)
Academic Partner: EPFL, ETHZ, Stanford University
Cassandra prophet of critical future
events in the Greek mythology
Profile of Cassantec Ltd.
Intelligent Malfunction Prognostics
Copyright © 2011 by Cassantec Ltd. 16
Our prognostic services have been successfully applied to a wide range of power equipment, with operators in different regions and industry segments
Cassantec References (Excerpt)
Wind Fossil Nuclear Chemical Steel
Intelligent Malfunction Prognostics