Computational simulation of the flow around rectangular cylinders: Effects of grid quality at wall

Starting from the past decades, many engineering domains require an accurate knowledge of Fluid Dynamics problems. Possible examples are car and airfoil designing in the automotive and aeronautics fields, blood circulation through the cardiovascular system in bioengineering or the evaluation of wind turbines efficiency in the energetic field. The first historical way to better understand a physical phenomena was to perform as much as possible experiments and to take decisions based on their results. However, sometimes the studied systems are characterized by big length or time scales and, as a consequence, experiments can be very expensive or even not possible to make. In addition to that measurement errors, if not correctly detected, can affect significantly the results. In the sixties, in concomitance with the first calculators, a new field called Computational Fluid Dynamics (CFD) started to work side by side with experiments. The latter are not supposed to be substituted by numerical simulations, that need to be tested with experimental results, but their number can be reduced in order to decreases the costs. Even if CFD has more than half a century of history, some of its topics are still arguments of actual research. The Computational Wind Engineering (CWE) is the branch of CFD, with Civil Engineering applications, that studies the wind effect on buildings. An important test-case in the CWE community is the characterisation of the flow around a rectangular, and in particular square, cylinders. As a matter of fact an important benchmark is the "Benchmark on the Aerodynamics of a Rectangular 5:1 Cylinder (BARC)". Despite being the easiest bluff body that can be thought, it is interesting due to the phenomenological complexity of the turbulent flow around it. Moreover this problem has an application into the study of wind effects on skyscrapers cross-section in Civil Engineering. CWE computations are based on three aspects that are intrinsically linked to each other: the turbulence modelling, the interpolations schemes used for the equations discretization and the grid generation. The latter has been often neglected in literature and more focus has been addressed on the former two. For this reason, the aim of the thesis is to understand how the quality of the near wall mesh can effect the simulated results in CWE computations. This first chapter focuses on the CWE benchmark of flows around rectangular cylinders. For this reason, the first part is devoted to the flux phenomenology around cylinders. The second part analyses some articles that are useful to understand the aim of the thesis. The next chapter discuss the equations used in CWE and their discretization. In the first part of the third chapter the main features of the meshes are explained. Their quality and the errors that they induce in the solution are presented. In the second a brief review on quality indices to measure the grid quality is discussed. Then the setup of the four simulations is presented. The four near wall grids are illustrated and their differences discussed with a particular focus on their quality. The numerical results are compared looking at different flow fields and quantities in different regions like the wall and the wake. The differences between simulations are highlighted and justified in terms of grid features. Finally a brief conclusion is presented.