| Summary: | This thesis addresses the important problem of shock wave--boundary layer interaction (SBLI) flows for aerospace engineering applications. Moreover, the work emphasizes the need for high fidelity simulations for the appropriate treatment of such flows. In this context, RANS solvers appear as a cost effective CFD approach. Therefore, the present work conducts studies in such a way to identify and to understand limitations, strengths and capabilities of RANS simulations for SBLI flows. Since turbulence modeling is an important issue on the accuracy of such simulations, the efforts here are concentrated on assessing the capabilities of several models that range from linear eddy-viscosity models (EVM) to Reynolds-stress closures (RSM). It would be expected that a RSM-type model could provide better solutions for a 3-D turbulent boundary layer under the action of high adverse pressure gradients, once such models allow for anisotropy between the Reynolds stress components. In order to achieve such goals, the configurations presented at the 2010 AIAA SBLI Workshop are chosen as the current test cases. Such test cases deal with high speed flows and very complex phenomena, including boundary layer separation. Meshes, composed of hexahedral and wedge elements, have been built. Mesh refinement and grid convergence studies are performed in order to identify a grid with a good compromise between accuracy and computational cost. In any event, even using the baseline grids, the present work has found that the computations are considerably expensive. Several simulations are presented for the test cases. Although no turbulence model has remarkably shown an outstanding performance over the others, the present work indicates that the SST and SA closures are the ones providing the best results for the test cases of interest here. Nonetheless, the two closures still present shortcomings in the simulation of SBLI flows. The overall simulation results using the RSM closure for the present SBLI test cases are not better than the SA and SST results. One must observe that the latter are much simpler turbulence models. Additional studies shall be focused on providing more robustness to the simulations with the 7-equation RSM turbulence model.
|
|---|
