Spatiotemporal-dependent structural reliability analysis of wind turbines using a reduced order finite element-informed surrogate model

The expansion of wind turbine deployment necessitates the development of systems capable of withstanding complex and highly dynamic load conditions. However, existing wind turbine design standards typically evaluate stress, fatigue, torsion, and other load-related parameters based on various simplifying assumptions. These assessments primarily utilize static random variables, thereby limiting the capacity to accurately represent the dynamic behavior encountered in real operational environments. To address this limitation, a spatiotemporal-dependent structural reliability analysis method is proposed in this study, offering the following key contributions: (1) the formulation of a dynamic wind load uncertainty model that captures time-dependent random external loads derived from empirical data, and (2) the integration of expansion op timal linear estimation with a reduced-order finite element-informed surrogate model. This methodology is applied to perform a comprehensive structural reliability analysis of a full-scale wind turbine. The results reveal a minimal global response error of only 0.035 percent, achieved with just one percent of the computational time required by conventional Monte Carlo simulation. Additionally, the global reliability assessment identifies potential structural vulnerabilities in the tower, the reby providing valuable insights for enhancing the overall reliability of wind power systems.