The negative effect of leading-edge erosion on wind turbine production is well known. The repetitive impact of rain droplets on the fast-moving wind turbine blades gradually erodes the surface, causing a loss in aerodynamic efficiency. Without timely repairs, the blade’s structural integrity can eventually become threatened. However, predicting leading-edge erosion can be challenging due to the complex influence of wind turbine operation, blade properties, and site-specific weather conditions. Despite this, Visbech et al. (2023) have recently developed a machine-learning approach for doing just that.

Visbech et al. (2023) train hundreds of simple neural networks with data from 678 blade inspections, wind speed, and precipitation time series. An ensemble of models allows for estimating the expected erosion damage and distribution, indicating the prediction uncertainty. After training and validating the model, Visbech et al. (2023) demonstrate its application by predicting blade defects for 99 sites across Nothern Europe. And suggest using the model to support the planning and scheduling future blade repairs.

Reference
Visbech, Jens, Tuhfe Göçmen, Charlotte Bay Hasager, Hristo Shkalov, Morten Handberg, and Kristian Pagh Nielsen. 2023. “Introducing a Data-Driven Approach to Predict Site-Specific Leading Edge Erosion.” Wind Energ. Sci. 8: 173–91. https://doi.org/10.5194/wes-2022-55.