Mesh quality effects on the simulated flow around a square cylinder: a computational study

This Thesis provides a systematic study about the flow past a square cylinder, which is a central field of research in Computational Wind Engineering (CWE). This work consists in compare seven types of mesh changing only the grid quality in order to prove a relationship between the mesh quality and inaccurate solutions on simulation. Fluent is the software used to compute all the results shown in this study. All seven grids are very similar, except on the near region where the boundary layer is present. In addition, all the settings of the model are the same. In this way, the errors that could appear on the solution is only affected by the computational grid. Several results will be shown to prove the discrepancies caused by changing grid quality. The seven meshes involved are named structured orthogonal, structured non orthogonal, structured/unstructured, unstructured, structured downwind non orthogonal, structured upwind non orthogonal and, finally, structured top non orthogonal. These modifications of the grids consist to change the grid quality measure introducing skewness errors in local regions, that are differed in each case by the specific location. There will be shown in this Thesis that a skewness grid leads to diffusive error that are not only local, but affects the results even in regions away from the source of errors. Finally, a direct correlation between skewness measure and solution errors will be computed. This, with other results, will prove the importance of a correct definition of a mesh in simulation.