Lifecycle-parallel and uncertainty-based determination of defect distributions for better lifetime predictions and higher reliability of wind turbine gears

ongoing

The CO2 reduction targets of European Union require the massive expansion of wind energy in Europe over the next decades. Typical mechanical drive trains of wind turbines are using gearboxes for speed and torque conversion between turbine rotor and generator. In the course of optimizing the installation space and weight of wind turbine gearboxes, the gear power density of the gearboxes has been steadily increased in the past by optimizing the gears in the near surface area. Due to this, gears are becoming more and more subject to subsurface fatigue, meaning the crack initiation at nonmetallic inclusions. With the increasing risk of subsurface fatigue, there is an increasing demand for statistical calculation models able to predict the failure probability depending on real part defect distributions in gears. In order to offer solutions for this demand, the main objective of WINCLUSION project is the development of a methodology for lifecycle parallel and uncertainty-based determination of defect distributions for large, highly utilized component volumes leading to improved lifetime predictions and less downtime of wind turbine gearboxes. The defect distribution is one of the most important input parameters for statistical calculation models, but difficult to determine. In a first step a simulation study will be performed to find gear designs with a maximized highly stressed volume for small modulus “research” and large modulus “industry” applications. The raw materials and also finished parts will be investigated regarding their defect distributions in detail. UT technology will be enhanced and UT standards for wind turbine gears will be proposed. Finally, experimental validation will be done by fatigue testing in small and large-scale environment. Parallel to these steps the methodology for lifecycle-parallel and uncertainty-based determination of defect distributions will be developed and validated.