While it is premature to draw firm conclusions about the nature of these failures, some reasonable observations have been made to help narrow the course and scope of this gearbox.
1.
Most of the problems with the current fleet of wind turbine gearboxes are generic in nature, meaning that the problems are not specific to a single gear manufacturer or turbine model. Over the years, most wind turbine gearbox designs have converged to a similar architecture with only a few exceptions. Therefore, there is an opportunity to collaborate among the many stakeholders in the wind turbine gearbox supply chain to find root causes of failures and investigate solutions that may advance the collective understanding of the industry.
2.
The preponderance of gearboxes failures suggests that poor adherence to accepted gear industry practices, or otherwise poor workmanship, is NOT the primary source of failures. Of course, some failures have been directly attributed to quality issues, and further improvements in this area are not precluded from consideration, but we assume that manufacturers are capable of identifying and correcting quality control problems on their own if they choose to do so. Therefore, the target of this project will be the greater problem of identifying and correcting deficiencies in the design process that may be diminishing the life of the fleet.
3.
Most gearbox failures do not begin as gear failures or gear-tooth design deficiencies. The observed failures appear to initiate at several specific bearing locations under certain applications, which may later advance into the gear teeth as bearing debris and excess clearances cause surface wear and misalignments. Anecdotally, field-failure assessments indicate that up to 10% of gearbox failures may be manufacturing anomalies and quality issues that are gear related, but this is not the primary source of the problem.
4.
The majority of wind turbine gearbox failures appear to initiate in the bearings. These failures are occurring in spite of the fact that most gearboxes have been designed and developed using the best bearing-design practices available. Therefore, the initial focus of this project will be on discovering weaknesses in wind turbine gearbox bearing applications and deficiencies in the design process.
5.
Furthermore, we believe that the problems that manifested themselves in the earlier 500-kW to 1000kW sizes five to ten years ago still exist in many of the larger 1 to 2 MW gearboxes being built today with the same architecture. As such, it is likely that lessons learned in solving problems on the smaller scale can be applied directly to future wind turbines at a larger scale, but with less cost.
Using these observations to help bound the problem, we reason that the accepted design practices that are applied successfully throughout other industrial bearing applications must be deficient when applied to wind turbine gearboxes. This characterization is based primarily on anecdotal field-failure data, and the experience of gear and bearing experts who have studied the problems for many years. Unfortunately, the available analytical methods to assess design life in typical gear designs are not accurate enough to shed much light on this problem, so much of the investigation must be conducted empirically.
A major factor contributing to the complexity of the problem is that much of the bearing design-life assessment process is proprietary to the bearing manufacturers. Gearbox designers, working with the bearing manufacturers, initially select the bearing for a particular location and determine the specifications for rating. The bearing manufacturer then conducts a fatigue life rating analysis to determine if the correct bearing has been selected for the specific application and location. Generally, a high degree of faith is required to accept the outcome of this analysis because it is done with little transparency. Even though bearing manufacturers claim adherence to international bearing- rating standards (ISO 281:2007 [8]), each manufacturer uses its internally developed design codes that have the potential to introduce significant differences that can affect actual calculated bearing life without revealing the details to customers. A new code is needed in the public domain that will give the industry a common method for due diligence in bearing design.
Moreover, since the bearing manufacturers do not have broad or intimate knowledge of gearbox system loads and responses that may be contributing to unpredicted bearing behavior beyond the bearing mounting location such as housing deformations, they are not capable of making valid root-cause analyses on their own. A broader collaboration of the various stakeholders, each of whom holds a piece of the answer, is clearly needed.