European Wind Energy Conference

1
Improving Wind Turbine Gearbox Reliability

• Walt Musial
• National Renewable Energy Laboratory
• Golden, Colorado USA
• walter_musial_at_nrel.gov
• Coauthors Sandy Butterfield and Brian McNiff

2
Observations on the basic problems

• Actual life is below expected design life.
• Problems are generic in nature.
• Poor quality is not the primary cause.
• Most failures initiate in the bearings.
• Problems are more dependent on bearing
  configuration than size.

3
Bearing Failure Observations

• General adherence to design standards
• ISO 2812007
• Proprietary codes prevent design transparency.
• Bearing manufacturers are not equipped to solve
  the problem on their own.
• No single, simple solution is expected.
• Collaborative approach is needed.
• Weaknesses in the design process are suspected.

4
Possible Design Process Weaknesses

• Missing Load Cases
• Irregular or unanticipated bearing responses
• Excessive flexibility of gearbox mount.
• Non-uniform safety applied to gearbox
  subcomponents.

5
Typical Wind Turbine Architecture
6
Typical Gearbox Mounting
Hub
Main Low Speed Shaft Bearing
Gearbox
Trunnion Mounts
Mainshaft
Bedplate
7
Trouble Spots

1. Planet bearings
2. Intermediate shaft-locating bearings
3. High-speed locating bearings

3
2
1
8
Test Articles (phase I)

• Test platform between 600-kW and 900-kW.
• 2 gearboxes with identical instrumentation.
• Upgrade both units to state-of-the art.
• Cooling, filtration, gear finish, lubrication,
  and bearing types.
• Measure External and internal loads and
  displacements.
• Thermal measurements
• Condition monitoring
• Expert failure analysis and forensics

9
Three Point Approach
10
NREL Dynamometer Specifications

• 2.5 MW power delivery
• Full power regeneration at 480/575/690 or 4160
  volts
• Torque input range 0 - 1.62 million N-m.
• Speed range from 0 - 2250 RPM
• 500 kN non-torque shaft loads capacity.
• SCADA and automated torque/speed controls

11
Dynamometer Testing

• Measure bearing responses to controlled load
  cases.
• Increase load complexity and build confidence.
• Develop non-torque load capability and simulate
  actual operating conditions.
• Establish transfer functions between shaft
  loading and bearing responses.

12
Field Testing

• Ponnequin Windfarm in Northern Colorado USA
• Extensive measurements on a single turbine.
• Characterize load events
• Correlate loads with component internal gearbox
  responses.
• Site-wide failures and statistics.

13
Drivetrain Analysis

• Multi-body dynamic analysis of test article.
• Codes FAST, Simpack, LVR
• Model bearing response under various load
  conditions measured in Dyno and Field.
• Model drivetrain solutions with tuned model.

14
Summary

• Bearing failures are contributing to fleet-wide
  reductions in wind turbine gearbox life.
• A comprehensive three-part program to identify
  and fix weaknesses in the design process was
  initiated at NREL.
• A collaborative long-term approach is required
  involving all stakeholders.

15
Panel Questions
16
Panel Questions 1

1. What is different with the Wind Turbine
   application compared to other industries
2. What efforts (technical) are happening with your
   part of the industry to improve reliability
   (changes, research, design tools etc)
3. Is there something in the design process that is
   lacking or needs more emphasis (loads
   description, rating methods, design tools

17
Panel Questions 2

1. Where do uncertainties exist in the loads,
   design, rating and system integration?
2. For prototype validation what should be tested/
   validated/ verified to improve confidence in
   designs
3. What future research needs to be done to improve
   reliability