Reconstructed Sub-Grid Methods for Acoustics Predictions at all Reynolds Numbers

This paper considers the concept and practical application of an artificially−generated, or synthetic, unsteady, turbulent flow field, generated using statistics obtained from a quasi−steady, Reynolds−averaged Navier−Stokes (RANS) solution. Two closely−related applications of this synthesis to the extraction of noise from flow fields of arbitrarily−high Reynolds number are considered. The first application is that of hybrid RANS/Large Eddy Simulation (LES), in which the reconstruction enables an energy transfer from the statistical RANS field into the resolved, large−scale (LES) components, thereby automatically providing appropriate boundary conditions for any embedded LES regions. The second application is to a non−linear acoustics solver, in which the reconstruction plays the role of a sub−grid closure, providing the volumetric sources for the unresolved, short−wavelength fluctuations. The synthetic, or artificially− generated, turbulence is constructed so as to mimic certain essential features of the statistical (RANS− represented) turbulence, including spatial and temporal correlations, turbulence energy and anisotropy. This paper considers developments relating to applications in which a limited portion of the turbulence energy spectrum has to be synthesized, and in which the underlying statistical field is both inhomogeneous and time−dependent.