This study applies proper orthogonal decomposition (POD) as a data-driven modal decomposition technique to analyze the wake dynamics of a full-scale horizontal-axis wind turbine (HAWT). Using high-fidelity CFD simulation data, the work investigates how large blade-tip deflections alter the downstream flow field and complicate the inflow for subsequent turbines in a wind-farm environment. 

The research focuses on the NREL 5-MW onshore wind turbine, a canonical model involving complex multiphysics interactions, including turbulence, mesh motion, and fluid–structure interaction (FSI). The modal analysis shows that both rigid and flexible blades exhibit comparable dominant structures in the near-wake region. However, the far wake is characterized by pronounced flow features such as local vortices, fluctuating velocity fields, and coherent structures originating from the tower–blade interaction. These structures tend to converge toward the centerline and merge with the nacelle wake, forming an integrated vortical pattern approximately 2.5D downstream of the rotor. The study also demonstrates that excluding the tower from the analysis results in a substantial loss of information about the wake structures, particularly in the far-wake region.

Through this work, we highlight how data-driven reduced-order modeling can offer insights into large-scale wind-energy systems, supporting future developments in surrogate modeling, SciML workflows, and physics-informed analysis.