Adaptive Importance Sampling for Enhancing Offshore Wind Turbine Reliability

In this paper, we continue the reliability study initiated in [1] on the extreme values of key mechanical characteristics -- pitch, surge, and heave -- of a floating offshore wind turbine (FOWT). Utilizing a comprehensive list of wind and wave patterns that cause anomalously large deviations in FOWT characteristics, originally revealed via brute-force Markov Chain Monte Carlo (MCMC) simulations, we have developed an efficient Adaptive Importance Sampling (AIS) MCMC. This new approach enables us to bootstrap and uncover the tails of the probability distributions associated with even higher and potentially more damaging values of pitch, surge, and heave which are not accessible through standard MCMC. Enhanced modeling of fluctuations in the large-scale wind component has allowed us to identify and examine both previously known and new rare but dangerous regimes. Notably, using AIS-MCMC, we pinpoint and analyze a surge anomaly driven by rare coherent wind patterns with relatively low mean values and wave interactions that interfere with wind turbine control.

[1] Yihan Liu and Michael Chertkov. Anomalous Response of Floating Offshore Wind Turbine to Wind and Waves, February

2024. URL: https://wes.copernicus.org/preprints/wes-2024-14/, doi:10.5194/wes-2024-14