ON THE PREDICTION OF SHEAR-LAYER FLOWS WITH RANS AND SRS MODELS

This study evaluates the ability of Reynolds-Averaged Navier-Stokes (RANS) and Scale-Resolving Simulations (SRS) models to predict turbulent shear-layer predominant (blunt body) flows. The selected cases are the flows around a circular cylinder at Re = 3, 900 and 140, 000, and past a rounded square prism at Re = 100,000 and incidence angles of 0 and 45 degrees. These cases exhibit complex features making numerical predictions a challenge, in particular, for turbulence modelling: shear-layers (free, boundary and wake), laminar-turbulent transition, low to moderate Reynolds numbers, flow separation and unsteadiness. In this paper, the aforementioned cases are simulated employing isotropic and anisotropic RANS, Delayed Detached-Eddy Simulation (DDES), eXtra Large-Eddy Simulation (XLES), and Partially Averaged Navier-Stokes (PANS) equations. The outcome confirms that traditional isotropic RANS are unable to accurately predict such flows, whereas SRS models can significantly reduce modelling errors. Furthermore, the results show that anisotropic RANS models are an interesting engineering option owing to its compromise between accuracy and cost. Nonetheless, an improvement of the modelling accuracy by both anisotropic RANS and SRS models is inevitably coupled with an increase of the numerical demands.