Turbine Structural Fatigue Damage and Reliability Evaluation Based on Excitation Process Spectral Analysis

ABSTRACT Time‐domain fatigue assessment under stochastic wind loading requires long‐duration stress histories and repeated simulations, resulting in substantial computational cost when reliability evaluation is considered. To improve computational efficiency while maintaining predictive accuracy, this study develops an integrated frequency‐domain framework for fatigue damage and reliability assessment of wind turbine tower structure based on excitation‐process spectral analysis. The novelty of the proposed approach lies in extending the Dirlik‐based spectral fatigue formulation from a standalone damage‐estimation tool to a complete damage–reliability assessment procedure. Specifically, stochastic wind‐speed simulation, finite element stress‐response extraction, Goodman mean‐stress correction, non‐Gaussian correction, broadband fatigue damage estimation, and probabilistic reliability evaluation are systematically coupled within a unified framework. The stress power spectral density of the critical region is derived from the dynamic response, corrected for nonzero mean stress and non‐Gaussian response characteristics, and then used for Dirlik‐based fatigue damage estimation. A wind turbine tower is investigated as an engineering case. The proposed frequency‐domain method differs from the time‐domain benchmark by 6.46%, whereas the Rayleigh‐based method exhibits a deviation of 16.98%. The 20‐year reliability of the critical region is 30.18% under the main Davenport–Weibull case, demonstrating the effectiveness and engineering applicability of the proposed framework.