Accurate yet computationally efficient models of turbine fatigue response are essential for wind farm layout design and flow-control optimization. Wake effects dominate turbine fatigue loading, and fatigue behavior varies across turbine designs and control strategies, requiring costly wind-farm-level simulations in most existing data-driven surrogate approaches. This work demonstrates that location-agnostic fatigue load surrogates can be trained using only single-turbine simulations. A large library of inflow slices is generated from dynamic simulations of a single wake and reused to drive aero-servo-elastic simulations of different turbine models of comparable scale. The methodology is demonstrated for three distinct turbines and validated against full dynamic simulations of a six-turbine wind farm under varying wake impingement, including wake-steering scenarios. The results show that the proposed approach accurately captures relative variations in turbine fatigue loads across the wind farm, both with and without wake steering, enabling efficient fatigue-aware wind farm optimization.
Link to paper: https://zenodo.org/records/22045776