This study establishes a multi‐hazard probabilistic assessment framework for assessing the integrity of monopile offshore wind turbines (OWT) under the stochastic coupled effect of wind, wave and earthquake loading. The procedure deals with the entire operational range of inflow wind speed (i.e., 3–25 m/s), for which the probability of failure under multi‐hazard excitations is found to be non‐negligible. Numerical analysis is performed by implementing nonlinear finite‐element models of the OWT developed in OpenSees. The dynamic response of the OWT system under wind‐ and wave‐load combinations is individually validated against those obtained from the aero‐hydro‐servo‐elastic simulator OpenFAST. Following the Latin‐hypercube approach, a cloud‐based assessment procedure is then performed with an ensemble of 300 earthquake ground motions, from which the multi‐hazard performance of the OWT regarding the serviceability limit state (SLS) and the ultimate limit state (ULS) can be evaluated. The epistemic uncertainty associated with various loads, structural properties, and soil conditions is also accounted for. Based on this probabilistic assessment framework, the sensitivity of the resulting OWT fragility surfaces to different statistical regression methods and wind—ground motion intensity measure pairs (IM‐pairs) is further scrutinised. Regression methods are comparatively evaluated. The efficiency, practicality, proficiency and sufficiency of various IM‐pairs are examined for the purpose of assessing operating OWT multi‐hazard fragility functions. The optimum IM‐pair is then employed in a trained Gaussian Process Regression (GPR) scheme for cloud data regression to assess the multi‐hazard fragility of the system. The derived multi‐hazard fragility function shows that the contribution of seismic forces in structural demand for a design‐level earthquake is comparable to those caused by operational‐level wind and wave loads.

No abstract