Skip navigation to main content.
NREL - National Renewable Energy Laboratory
About NRELEnergy AnalysisScience and TechnologyTechnology TransferTechnology DeploymentEnergy Systems Integration

Gearbox Reliability Collaborative Research and Development

Since its inception, the Gearbox Reliability Collaborative (GRC) has collected an extraordinary amount of experimental testing data. This data, along with its associated geometric and material properties and solid body models for modeling offer a tremendous resource for the wind turbine industry. Up to 125 channels of data, gleaned from two identical 750-kW gearboxes and their models, represent hundreds of hours in field and dynamometer testing. Categories of data include drive train speed, steady state and transient torque, steady state and dynamic non-torque loads, internal loads and deflections, and condition monitoring data, such as oil quality and debris, gearbox vibration, acoustic emission, and generator electrical signatures.

GRC Key Research Results

GRC Research
NREL completed Phase 1 and Phase 2 research in the GRC project and published the "Gearbox Reliability Collaborative Project Report: Findings from Phase 1 and Phase 2 Testing."

Key Results
The GRC has identified shortcomings in the design, testing, and operation of wind turbines that contribute to excessive loads and deflections within the gearbox and, by extension, contribute to unintended gearbox failures. In the first two project phases, testing of the GRC gearboxes under real operating conditions, along with subsequent modeling, provided important data and conclusions such as the importance of non-torque loads, dynamic torque events, planetary flexibility, clearances, and tolerances.

Potential Impacts
The work of the GRC will enable designers, developers and manufacturers to improve gearbox modeling tools, designs and testing procedures resulting in increased gearbox reliability and an overall reduction in the cost of energy from wind.

Significant GRC research of interest across the wind turbine industry, from wind turbine manufacturers to parts suppliers and wind turbine operators, are in the following areas:

Gearbox Design and Modeling

  • Importance of non-torque loads: Non-torque loads (NTL), such as shaft thrust and bending, have historically been left out of the modeling and dynamometer test validation process. GRC results indicate that NTL affects tooth contact patterns in the low-speed stage resulting in increased gear tooth edge loading and unequal load sharing among planet gears that can shorten gearbox life
  • Importance of bearing clearance: GRC results also show unequal load sharing between upwind and downwind support bearings due to differences in bearing and planet pin fits, especially at lower torque values. Increased loading on bearings may be a contributing factor to shorter bearing life than expected.
  • Influence of assembly variations: Gearbox reliability is heavily dependent on specified tolerances for easy, consistent assembly. Many components are influenced by bearing clearances, shaft interference fits, and other assembly measures. Comparison of GRC data for two nominally identical gearboxes showed a difference in the deformation of the planetary carrier, the subsequent effect on predicted bearing life, and a difference in loading and deflections due to bearing clearances and stiffness. Conversely, modeling has shown that optimized clearances can significantly reduce time-varying contact stresses and likely will result in longer gearbox life.
  • Importance of model fidelity: The GRC has researched the level of model fidelity needed to predict the responses of gearbox components, as captured in testing. A trade-off exists between time (cost) and accuracy. The gearbox housing should be modeled as a flexible element (instead of a rigid body) due to its influence on shaft bore and ring gear misalignments. The flexibility of the carrier also is necessary due to its complex geometry and its effect on planet-ring gear mesh alignment. Tests of various non-torque loading conditions provided evidence that a flexible ring gear is needed to capture the effects of the shifting gear face load distribution and the hoop strain changes caused by planet passes. In summary, the ability to model flexibility in the planetary carrier, support structure, gears, and bearings is important. Load sharing in the planetary section varies within each revolution; thus, each planetary element should be modeled in parallel to capture these effects.

Gearbox Testing and Standard Acceptance

Photo shows a representative gearbox being tested in the dynamometer at the National Wind Technology Center.
  • Run-in procedure: A controlled run-in procedure, usually a factory acceptance test consisting of 30–60 minute load stages, has been a requirement in the wind turbine gear industry since 2002 when the ANSI/AGMA/AWEA 6006 wind turbine gear standard was published. The GRC suggests that this procedure be modified by preheating the lubricant and gearbox to operating temperature, using oil that does not have extreme pressure (EP) additives, and monitoring oil debris count/cleanliness and using it as the controlling factor to determine the length of each load stage.
  • Reproduction of field bending and torque in dynamometer: Dynamic non-torque and torque variations in the field are not reproduced in typical dynamometer testing. Accurately reproducing these loads is important because the non-torque loads have an appreciable effect on the load distribution across the ring gear teeth and on the tilt of the planet carrier. These variations can be reproduced in the dynamometer, but limitations in the test equipment must be considered.

Gearbox In-service Operation

  • Reduction in torque spikes: Field testing uncovered both negative and positive torque spikes during generator speed transitions that greatly exceeded expected values, despite the presence of a soft-starter (used to mitigate these spikes). The magnitude of the torque spikes, and damage from them, could be greatly reduced with controller tuning. Note that this may require adding sensors not typically included on commercial machines, including low-speed torque measured at a rate of least 20 times the operating speed.