A P-DNS framework for unsteady aerodynamics of full-scale wind turbines

This work presents a novel numerical framework for the simulation of unsteady wind turbine aerodynamics at full scale, enabling the simultaneous resolution of blade geometry and rotor motions while remaining computationally tractable for systematic studies. The approach couples geometry-resolved rotating components with a multiscale strategy based on Pseudo-Direct Numerical Simulation (P-DNS), in which fine-scale turbulent effects are embedded through surrogate models. This formulation preserves unsteady aerodynamic behaviour while allowing transient calculations on meshes substantially coarser than those required by wall-resolved methods. Rotor and component motions are treated using an overset mesh formulation that enables large relative displacements between blades, hub, and tower, and incorporates flow-driven rotation through a coupled aerodynamic–resistive torque balance, capturing load–motion feedback without full aeroelastic coupling. The methodology is demonstrated on a utility-scale horizontal-axis wind turbine based on the Siemens 2.3 MW reference design through progressively increasing geometric, environmental, and dynamic complexity. This strategy enables the isolated assessment of blade–blade interaction, rotor–tower interference, inflow shear representative of onshore and offshore conditions, and the transition from prescribed to aerodynamically driven rotation. The results show that the framework captures unsteady load modulation and tower-passage spectral features while maintaining accurate integral performance at industrial scale. Compared with URANS simulations using the Spalart–Allmaras model, which underpredict rotor power by about 20% on coarse wall-modelled meshes, the proposed P-DNS framework shows close agreement with the manufacturer power curve over the full operating range, providing a computationally feasible pathway towards geometry-resolved high-fidelity wind turbine CFD.