A surrogate modeling approach for fatigue damage assessment of floating wind turbines

Abstract

Fatigue analysis for floating wind turbines poses a novel challenge to calculation workflows if a probabilistic load environment is to be considered. The increased complexity of the structure itself as well as its interaction with the environment require a coupled and more detailed analysis with respect to resolution of environmental conditions compared to fixed bottom systems. Different approaches to address the computing challenge for floating turbines are possible to support engineering judgement and have been investigated in the past, with conservative binning on the one end of the accuracy scale and computation intensive Monte Carlo simulations on the other end. This study investigates the feasibility of regression based surrogate models based on radial basis functions. The investigation performed here is aligned with work performed in the H2020 project LIFES50+. Consequently, the considered system is the DTU 10MW Reference Wind Turbine installed on the LIFES50+ OO-Star Wind Floater Semi 10MW. The site under investigation is the LIFES50+ Site B (Gulf of Maine) medium severity representative site. Results show a similar convergence of lifetime fatigue load prediction as with Monte Carlo simulations indicating that this technique may be an alternative if a response model of the considered system is of interest. This may be interesting if damage loading is to be calculated at a different site and if a classification of met-ocean conditions is available.