Use of Hybrid RANS–LES for Acoustic Source Predictions
Paul Batten, Philippe R. Spalart, Marc Terracol · Cambridge University Press eBooks · 2007
Introduction to hybrid RANS–LES methods The title of this book implies the current or impending feasibility of large-eddy simulation (LES) for use in noise prediction problems. The potential accuracy of LES is generally well regarded; however, the cost of such calculations requires a specifically focused algorithmic and computational effort in order to make LES affordable for practical engineering problems. At high Reynolds numbers, such problems can defeat LES by raising the range of scales beyond affordability owing to the extremely small size of the eddies in the viscous sublayer or even the size of the dominant eddies in the bulk of the boundary layer (Spalart et al. 1997). This chapter outlines a variety of recently developed hybrid methodologies that specifically address the issue of computational cost and make the simulation of large-scale, sound-generating flow structures tractable with existing computer resources. The hybrid nature of the methods discussed here involves the simultaneous use of (or blending between) statistical Reynolds-averaged Navier–Stokes (RANS) and traditional LES within the noise-source region. This hybrid character is distinct from the methods used in traditional acoustic analogies for far-field propagation of sound. Those issues remain here; however, this chapter specifically considers methods offering an affordable prediction of both the mean flow and pressure fluctuations in the near field.