Hybrid Reynolds-Average Navier-Stokes and Kinetic Eddy Simulation of External and Internal Flows

Abstract

thus alleviating the grid-spacing dependency found in other detached eddy simulation and hybrid Reynoldsaveraged Navier–Stokes and large eddy simulation models. Within the hybrid Reynolds-averaged Navier–Stokes andkinetic eddy simulation model, fourdifferent options (a combination of two different blending functions andthe use of realizability constraints to bound the kinetic eddy simulation model parameters) have been evaluated and studied for flows over airfoils (RAE2822, NACA0015) and a turbine vane configuration. Although all four options showed good correlation with the test data, blendingk-!-SST with kinetic eddy simulation using Menter’sk-!-SST F2 function showed better predictions in separated flows than using the baseline k-!-SST model. OST current computational fluid dynamics (CFD) methods used in the simulation and analysis of external and internal flows are based on Reynolds-averaged Navier–Stokes (RANS) approaches, which depend on turbulence-closure models to provide flowfield turbulent variables. Although RANS approaches yield good predictions for attached flowfields, they fail to accurately predict flow structures in separated-flow regions because they resolve only a portion of the turbulence scales of interest. Of course, the ideal approach would be direct numerical simulation (DNS), because the entire range of spatial and temporal scales of turbulence is resolved, but the computational cost of DNS is prohibitive. Another intermediate technique between DNS and RANS has been proposedtoreplaceRANSinsuchcases;thisapproachiscalledlarge eddy simulation (LES). In LES, the contribution of large energy-containing structures and all scales larger than the grid resolution to momentum and energy transfer is computed, and the effect of subgrid unresolved small