Data-Driven Stochastic Fatigue Load Modeling for Fatigue Failure Simulation of Onshore Wind Turbine Foundations

Abstract

Assessing fatigue damage in onshore wind turbine foundations is crucial for ensuring the safety of the entire wind turbine system. While indirect simulations of fatigue damage based on upper structure wind loads have been explored, direct modeling using real data has been previously unaddressed. This study introduces a novel approach to model the fatigue load of wind turbine foundations directly from real measurements. Recognizing the time-varying nature of the upper wind turbine structure, which complicates the accurate assessment of wind load to fatigue load transition, the research employs the augmented Dickey–Fuller test to treat the foundation fatigue load as a weakly stationary stochastic process. A data-driven stochastic fatigue load model is developed using the stochastic harmonic function method, leveraging a substantial data set of real monitoring data. This model allows for the conversion of random amplitude fatigue loads into equivalent constant amplitude loads, facilitating a deeper investigation into foundation fatigue failure. The study concludes with a fatigue damage analysis of a 2.0-MW onshore wind turbine foundation in Ruyuan County, China, revealing that damage is predominantly concentrated in the concrete near the anchor cage. The research findings indicate that as the turbine’s service time extends, the concrete fatigue damage accumulates, potentially culminating in concrete failure near the anchor cage. This work provides critical insights for the design and maintenance of wind turbine foundations to mitigate fatigue-related failures.