Schram C., Tournour M.

The increasing public awareness and concerns with respect to aerodynamically generated noise has led acoustic and Computational Fluid Dynamics (CFD) communities to develop joint aeroacoustic prediction tools. This has led to the so-called hybrid approach, where the flow description obtained by CFD is used to synthesize equivalent aeroacoustic sources that are injected into an acoustic propagation code. We propose an implementation of the hybrid approach based on Curle’s analogy, which formally describes the mechanism of sound generated by body-turbulence interaction. The acoustic propagation is carried out using an acoustic code based on the Boundary Element Method (BEM), in which scattering by arbitrary geometries can be computed, including impedance effects where needed. This was proved to give reliable results for cases where the body dimensions are fairly acoustically compact, and/or when the CFD model is accurate enough to capture the near-field acoustical information. Difficulties arise however when the spatial extent of the turbulence-body interaction region becomes comparable to, or larger than the acoustical wavelength. A previous study has shown that a straightforward application of Curle's analogy to non-compact geometries, and in which acoustic source terms (in particular surface source terms) do not account for compressibility effects, can yield a poor estimation of the acoustical far-field. An innovative procedure is therefore developed in the present work, where the dual effects of the body as source and scattering entity are separately handled by the flow and acoustic solvers. This requires a novel numerical implementation of Curle's analogy, in particular considering the handling of the singularities of the Green's functions kernel applied to the CFD data. A special attention is brought to the robustness of the proposed method with respect to inaccuracies inherent to low-order CFD data, considering the numerical dissipation and dispersion of acoustical information.