Wind turbine aeroacoustics driven by dynamic mode decomposition-based aerodynamics

This paper presents an aeroacoustic analysis of the NREL 5 MW rotor-nacelle assembly by the permeable-surface Ffowcs Williams–Hawkings (FWH-P) equation coupled with a Large Eddy Simulation (LES) aerodynamics. By assuming a straight, laminar inflow to the rotor disk, near and far field noise features are analyzed in the time-domain and by the Overall Sound Pressure Level (OASPL) directivity at ground and in the rotor plane, comparing the acoustic contributions from linear (pressure-based) and nonlinear (velocity-based) terms, and highlighting the role of tonal and broadband noise components. Throughout the paper, a Dynamic Mode Decomposition (DMD) of the LES dataset is used to identify key aerodynamic modes in the pressure and velocity fields, whereas the Sparsity-Promoting SPDMD algorithm is applied to isolate the dynamically relevant modes governing noise sources evolution. These modal information enable to establish correlations between radiated noise and sources of sound, driving in turns, the numerical assessment of a mildly-nonlinear FWH-P approximation, whose spectrum contains only those frequencies identified by the (SP)DMD analysis. Numerical results show the link between modal wake structures and radiated noise, proving the effectiveness of the mildly-nonlinear FWH-P solution in capturing the dominant noise field features, while excluding higher-frequency effects associated with acoustic nonlinearities.