Traditionally, the assessment of the lifetime of wind turbines relies on the application of safety factors on loads predictions by physically based aeroelastic tools that incorporate the level of uncertainty of its design parameters.

The aim of the present research project is to develop and test, innovative probabilistic fatigue analysis methods which will be used for the assessment of the lifetime of wind turbines based on consistent levels of reliability for all components.

The focus of the work is on the assessment of the remaining lifetime of operating wind turbines, within specified levels of uncertainty for the design parameters (environmental, structural, aerodynamic etc.), with the objective to provide reliable estimates for lifetime extension. For the prediction of the load envelope of the wind turbine, both physically based but also data driven, machine learning models will be employed.

Furthermore, for assessing the inflow conditions experienced by the turbine within the wind farm (a crucial factor for assessing fatigue), data assimilation methods will be employed that leverage high fidelity CFD results to enrich a low cost engineering wake model.

The new wake model will be able to better account for complex wind farm effects such us multiple wake interactions and meandering. The validation of the advances will be done with a prototype integrating data and knowledge resources identified in the other WPs of the project.

A new innovative probabilistic framework for the assessment of wind turbines’ fatigue. Various surrogate data-driven models for the assessment of wind turbines’ fatigue loads that will be trained by existing physically based models. Application of innovative data assimilation techniques to the development of a new wind farm wake models.

One industrial secondment (6 months) at iWind (Dr. P. Chaviaropoulos, M18-M23). for training on machine learning techniques for fatigue loads predictions and one academic secondment (3 months) at DTU (Dr. Ju Feng M28-M30).