Analysis of the mechanism of aerodynamic noise source formation in wind turbine blades under dynamic icing conditions

Ice accretion on wind turbine blades degrades aerodynamic performance and alters noise characteristics, yet the mechanisms governing noise-source formation under realistic icing conditions remain poorly understood. This study investigates the coupled effects of icing duration and rotational speed on aerodynamic noise through dynamic icing experiments on a 600 W horizontal-axis wind turbine with controlled liquid water content, droplet size, and temperature. Noise sources were characterised using a 112-channel circular microphone array with eigenvalue-optimised beamforming and Doppler-shift compensation. Ice geometries reconstructed via 3D scanning were incorporated into large-eddy simulations coupled with the Ffowcs Williams–Hawkings analogy to analyse flow evolution, vortex dynamics, and far-field noise. Results reveal that centrifugal effects produce a nonlinear root-to-tip increase in ice thickness, concentrated on the leading edge and windward surface. As icing intensifies, the dominant noise source in the 4000 Hz band shifts radially from 0.7R to 0.8R, with SPL increasing by 3–11 dBA; the 5/6R region is identified as the primary contributor. Simulations show premature flow separation shifting towards the leading edge, enhanced trailing-edge and tip vortex shedding, intensified pressure fluctuations, and a 55% reduction in lift-to-drag ratio in the critical region. Although the findings are limited to small-scale blades with similar aerofoil profiles under simplified flow conditions, they elucidate the relationships among icing parameters, flow separation, vortex dynamics, and noise generation, offering a foundation for acoustic-based icing diagnostics in cold-region wind farms.