Wind farm predictive maintenance considering component level repairs and economic dependency

Existing studies on wind farm predictive maintenance mostly incorporated repairs on the level of wind turbine (WT) subassemblies, e.g. drive train, gearbox and generator, including major repairs and minor repairs. The economic dependency among subassemblies was also considered. But assumptions needed to be made on major and minor repairs regarding their costs and impact on system functions. Besides, existing studies considered only fixed costs on the WT and farm levels without differentiating fixed costs that apply to a subset of components, such as crane costs, resulting in over-simplification compared to wind farm practices. In this work, a wind farm predictive maintenance approach is developed considering component level repairs and economic dependency. There are multiple components in a subassembly, e.g. the generator consists of components like generator rotor, generator bearings, contactor, etc. These components' failures lead to major or minor repairs, downtime, and demand on repair resources, particularly cranes. With the proposed wind farm predictive maintenance approach, component level major and minor repairs and their costs can be modeled explicitly in a more realistic and accurate way. Economic dependency can also be modeled more accurately by incorporating the downtime, i.e. revenue loss, and repair resource requirement caused by different component failures. A predictive maintenance optimization model is developed to find the optimal maintenance policy. The number of WTs eligible for preventive maintenance is introduced as a new decision variable, which also captures the economic dependency among WTs. A simulation-based method is developed for maintenance cost evaluation. Examples are used to demonstrate the proposed approach.