Day-to-day condition monitoring for a large fleet of wind turbines
Post on 16-Jul-2015
501 Views
Preview:
Transcript
Day-to-day Condition Monitoring For a Large Fleet
of Wind Turbines
Before We Start q This webinar will be available at
www.windpowerengineering.com & email
q Q&A at the end of the presentation
q Hashtag for this webinar: #WindWebinar
Moderator Presenters
Paul Dvorak Windpower Engineering
& Development
Sam Wharton Romax Technology
John Coultate Romax Technology
Day-to-day condition
monitoring for a large fleet
of wind turbines
Dr John Coultate, Head of Monitoring and O&M Consultancy Dr Samuel Wharton, Condition Monitoring Engineer February 19th 2015
Contents 1. Introduction to Romax Technology 2. ‘Condition Monitoring 101’ 3. Challenges faced monitoring a large fleet of wind
turbines 4. Practical examples - main bearing and gearbox fault
detection and workflow
• Gearbox and drivetrain specialists
• Established in 1989
• Approx. 250 employees globally, 120 in UK, 12 offices worldwide
• Work in a range of industries
o Automotive, Off-road, Marine, Aerospace
o Wind energy
Romax Technology
Track record in condition monitoring • Romax has assessed the performance and health of over 5GW of wind turbines
globally • Romax provides a monitoring service for turbines worldwide, including over 40%
of the UK offshore fleet
Monitoring Service Example Project Centrica • Three UK offshore wind farms: Lincs, Lynn and Inner Dowsing
• 129 x 3.6 MW turbines
• Vibration monitoring service delivered by Romax using Fleet Monitor software
• Regular health assessment incorporating SCADA analysis
InSight
Condition monitoring
Condition monitoring 101 • Why install CMS? • The business case is complex with four main sources of return:
1. Catastrophic failures can be avoided • CMS catches faults developing and enables more up-tower repairs. • E.g. High speed shaft and generator bearing faults are reliably detected
before a critical failure occurs. Damaged components are replaced up-tower without a large crane.
2. Crane costs are minimised by combining operations • Early detection of faults means that crane operations can be combined
for multiple turbines rather than reacting to one-off failures.
Condition monitoring 101 • Why install CMS?
3. Downtime reduced • Pro-active maintenance - spare parts and crane are
ordered before a failure occurs.
4. Improved annual energy production • Early detection using CMS means turbines with faults
can be de-rated and run through high wind periods before scheduled repair.
Example return from CMS – main bearing
replacement • Significant benefits to predicting main bearing failure and scheduling multiple simultaneous
repairs:
Nov Dec Jan Feb Mar Apr May Jun Jul Aug Sep Oct
• Main bearing fault detected on one 1.5 MW turbine
• Continue running turbine during windy season.
• Calculate optimal de-‐‑rating if necessary
• Main bearing fault detected on another turbine
• Continue running both turbines while repairs are scheduled
• Crane and spare parts are ordered for 2x turbines
• Repair both turbines simultaneously during low wind season
Total cost saving for this
single operation ~ $310k
on two turbines Main sources of
ROI:
1. Reduced crane cost
2. Reduced downtime/
increased power production
Challenges faced monitoring a large fleet
of wind turbines
CMS (Bently Nevada, Commtest, SMP, etc.)
Wind farm 1
(e.g. Siemens, Vestas, etc.)
Challenge #1 – Too many different type of wind turbine and CMS
Wind farm 2 (e.g.
GE, Gamesa, Clipper, etc.)
CMS-‐‑specific database and software
CMS (Gram & Juhl TCM, B&K Vibro, etc.)
CMS-‐‑specific database and software
• Monitoring engineers can be overwhelmed by different pieces of software and data
• Difficult to make consistent decisions
• Often lots of staff required Monitoring
engineer(s)
CMS (Bently Nevada, Commtest, SMP, etc.)
Wind farm 1
(e.g. Siemens, Vestas, etc.)
Romax server
‘Hardware independent’ condition monitoring architecture
Romax monitoring service
Fleet MonitorTM software
Wind farm 2 (e.g.
GE, Gamesa, Clipper, etc.)
Database or site server
Database or site server
CMS (Gram & Juhl TCM, B&K Vibro, etc.)
Challenge #2 – Keeping track of faults and alarms from 100s/1000s of turbines
• ‘Workflow’ is a major concern. • Above a certain fleet size, keeping track of faults and alarms is difficult. Too many
alarms is a problem. • Concise routine reports with top findings, red/amber/green classifications and
recommendations:
Challenge #3 – Effectively incorporating SCADA analysis and other data
• SCADA analysis is well established for power production reporting; power curve analysis; wind resource assessment, etc.
• SCADA reporting generally well implemented for NERC compliance • Not well utilized for reliability analysis
Condition monitoring tools • Good condition monitoring software
should be able to: o Handle data from multiple CMS vendors. o Easily switch between different
configurations (multiple gearbox variants, turbine types, power ratings)
o Provide useful alarms that accurately indicate fault progression.
o Ideally: Be a portal for allowing operators and monitoring engineers to keep track of data from multiple sources to aid fault diagnosis and maintenance planning.
Insight Fleet MonitorTM Software
Condition monitoring tools
Raw Vibration Data
Time
Processing (FFT..etc)
Component Specific Alarm
Turbine Drivetrain
Operating Conditions
Drivetrain Info +
Experience
Condition monitoring tools – Gearbox Information
• Condition monitoring software needs to have built-in information on every gearbox/drivetrain variant operating in the fleet.
Gearbox1
Gearbox2
Raw Vibration Data
Time (sec)
Time (sec)
Frequency (Hz)
Frequency (Hz)
Frequency Transform
(FFT)
Frequency Transform
(FFT)
Frequency Spectra
Gearbox1 frequency table
Gearbox2 frequency table
Condition monitoring tools – Operating Conditions
• Condition monitoring software needs to have access to the operating conditions of a wind turbine at the point of time the measurement was taken (i.e. Active Power and Shaft Speeds)
HSS Shaft at 20 RPM
HSS Shaft at 25 RPM
All power classes Near rated power
Peak Amplitude Trending
Amplitu
de
• Fault peak tracking is a very useful technique for detecting the onset of faults, but can often be poor indicators of advanced damage
• In Fleet Monitor we can easily combine multiple measurements and trends into one more powerful indicator of fault progression – The Romax Health Index.
Condition monitoring tools- Trending
Combine multiple parameters
in Fleet Monitor Romax Health Index
Peak Amplitude Trending
Condition monitoring tools - Alarm Setting
• Two types of alarm threshold: o Manual – Alarm thresholds are chosen
based on guidelines or experience by monitoring engineer.
• Don’t require historical data • Sometimes are not very sensitive
o Automatic – Alarm thresholds are set based on a fitted distribution to the data.
• Require historical data • Can be very sensitive
Condition monitoring tools – SCADA data
Vibration
SCADA-‐‑based Temperature
Time Fleet Monitor
Practical examples
CMS case study 1 – main bearing faults
• Typical main bearing failure modes detected by CMS:
Severe outer race macropiaing and cracking
Roller macropiaing
Severe roller damage Inner race macropiaing
CMS case study 1 – main bearing faults
• Typical main bearing fault development over a long time period:
First Romax Alarm
8.5 months
This turbine had a damaged front main bearing. Indentation marks recorded on rollers and inner race.
Bearing Replaced
Main Bearing Health Index
Date
CMS case study 1 – main bearing faults • Main bearing fault that developed in 30 days:
This turbine had a damaged front main bearing. There were indentation marks on the inner ring.
Romax Alarm
Increased grease flushing regime to
prolong life
Bearing Replaced
Over 4 months power production after first
alarm
>4 months
Main Bearing Health Index
CMS case study 2 – gear tooth faults • Typical gear failure modes detected by CMS:
Root bending overstress
Tooth fatigue crack
CMS case study 2 – gear tooth faults Gear Health Index
1st CMS Alarm
Turbine Stopped
Turbine started without replacement
Replacement of High Speed Shaft
4 Hours
2nd CMS Alarm
CMS case study 3 – planetary stage faults
• Typical planetary stage failure modes detected by CMS:
Tooth fatigue crack Severe roller macropiaing
Planet bearing inner race macropiaing
CMS case study 3 – planetary stage faults
• Analysis of historical data:
• Planetary gear stage failed catastrophically
• OEM did not detect the fault
• Romax analysis detected the fault over 3 months before failure
Health
inde
x Romax Alarm
>3 months
Turbine with 2nd stage ring gear fault
Healthy turbine
CMS software case study – HSS bearing fault
• HSS Bearing Yellow (warning) Alarm triggered by Romax Bearing Health Index trend.
Alarm triggered
CMS software case study – HSS bearing fault
• HSS Bearing Yellow (warning) Alarm triggered by Romax Bearing Health Index trend.
• Alarm is investigated by monitoring engineer using vibration analysis toolbox.
• Clear fault frequencies associated with specific HSS bearing fault.
• Report sent to farm operator recommending inspection and oil sample analysis in next six months plus continued monitoring.
Alarm triggered HSS Bearing Fault Frequency Harmonic 1
HSS Bearing Fault Frequency Harmonic 2
Fault frequencies
CMS software case study – HSS bearing fault
• Health index trend continues to increase.
• Inspection carried out by Romax engineers confirms damage to bearing.
• Oil analysis shows high Fe content. • Operator keeps track of reports. • Operator stores oil analysis results. • Replacement scheduled.
High Fe
CMS software case study – HSS bearing fault
• Bearing health index triggers red (critical) alarm.
• Exception report sent to operator.
CMS software case study – HSS bearing fault
• Bearing health index triggers red (critical) alarm.
• Exception report sent to operator. • Replacement carried out • Maintenance record updated in
Fleet Monitor. • Post-replacement Health Index
trend drops to new baseline level. • Alarm threshold to be lowered.
Remaining useful life
3y+ 2y 1y Event
Con
ditio
n
Life Model
Inspect
Vibration
What is a remaining useful life model?
Predictive life models • For many years, predictive life models have been used for maintenance scheduling:
o Aerospace; power production; helicopters; etc. • Some pitfalls to avoid:
o You can’t just simply use a model from a different industry for a wind turbine o You can’t rely on a computer simulation to mimic complex wind turbine failures
• Romax are pioneering a predictive life model approach for wind turbines.
Life Model Benefits • Life models allow effective long term maintenance planning by:
o Ranking components for wear levels over time o Working in conjunction with existing systems and processes
(CMS, particle counters, inspections) • A predictive maintenance strategy can greatly reduce future
operating costs
Summary and Conclusions
Summary and conclusions • Scaling up a condition monitoring operation poses
some difficult challenges: o Hardware independent monitoring o Building an expert team o ‘Workflow’ – keeping track of faults and alarms o Predicting failures
• Romax deliver specialist software and services to solve these problems.
Questions? Paul Dvorak Windpower Engineering & Development pdvorak@wtwhmedia.com Twitter: @windpower_eng
Sam Wharton Romax Technology Samuel.wharton@romaxtech.com
John Coultate Romax Technology john.coultate@romaxtech.com
Thank You q This webinar will be available at
www.windpowerengineering.com & email
q Tweet with hashtag #WindWebinar
q Connect with Windpower Engineering & Development
q Discuss this on the EngineeringExchange.com
top related