STUDY OF DIFFERENT SUB-GRID SCALE MODELS IN LES TO UNDERSTAND THEIR IMPACT ON TURBULENCE PREDICTION

In the recent years part of the industry has focused on making improvements in the aerodynamic performance of their designs. In F1 for example, where an optimal shape is synonym of tenths of a second or in aeronautics, with new approaches like the morphing wings. To do that, very accurate CFD simulations must be done and this is why LES models play an important role. This dissertation is about studying different LES models (Smagorinsky, WALE and Sigma). However, running good simulations is much more than simply generating a mesh and waiting several hours until the solution is obtained. The main idea is to generate and validate a CFD methodology (studying a 2D cylinder). Specific aspects in the mesh generation will be considered (y+, aspect ratio, Kolmogorov scale, etc.). After that, OpenFOAM (the open source software used to run the simulations) will be set up properly. Once the solution is completed, paraview (another open source program) will help with the post process, so the results will be compared with experimental data. This will be done for Smagorinsky and WALE because sigma is not implemented in OpenFOAM yet, so it will be coded using C++ before. After validating the methodology, the best model will be used in a more realistic case (a 2D bluff body with a diffuser). In summary, the project highlights the steps required to run and post process LES simulations properly in order to obtain accurate results in every kind of geometries studied