Reynolds Averaged Navier Stokes (RANS) turbulence models have been shown to be insensitive to the effects of boundary layer relaminarlsatiou present in highly accelerated flows. Further, downstream separation of these quasi-laminar boundary layers suffer from incorrect flow reattachment prediction. This results from RANS poor performance at sufficiently transitioning the flow in the separated shear layer, whereby modelled turbulence production is often insufficient to correctly reattach the flow. The present work describes laminarisation and transition modifications for the one equation Spalart-Allmaras (SA) turbulence model applied to a a wall mounted hump test case. This is coupled with a zonal RANS-Dettached Eddy Simulation (DES) method for efficient high fidelity simulations. The standard SA model is shown to over predict separation bubble size by 44%, as well as over prediction of peak C_p by 7%. The laminarisation modification is able to improve surface pressure predictions over the accelerated flow region of the hump, bringing peak C_p to within 5%. The transition modification is then able to boost turbulence production post separation to correctly reattach the flow. These highlight a major concern for standard RANS models ability to predict separated flows well without modification. While the transition modification is a calibrated process, a switch to DES in this separated aone enables an effective alternative to this, giving very good separation bubble size predictions to within 5% of experimental measurements in Its standard form. DES also gives the added benefit of resolving higher order moments such as Reynolds stresses, which are also predicted well.
展开▼