Dynamic grid adaptation applied to large eddy simulation turbulence modelling

Abstract

At present a large number of fluid dynamics applications are found in aerospace, civil and automotive engineering, as well as medical related fields. In many applications the flow field is turbulent and the computational modelling of such flows remains a difficult task. To resolve all turbulent flow phenomena for flow problems where turbulence is of key interest is a priori not feasible in a Computational Fluid Dynamics (CFD) investigation with a conventional mesh. The use of a Dynamic Grid Adaptation (DGA) algorithm in a turbulent unsteady flow field is an appealing technique which can reduce the computational costs of a CFD investigation. A refinement of the numerical domain with a DGA algorithm requires reliable criteria for mesh refinement which reflect the complex flow processes. At present not much work has been done to obtain reliable refinement criteria for turbulent unsteady flow. The purpose of the work presented in this thesis is to use both a DGA algorithm and Large Eddy Simulation (LES) turbulence model for predicting turbulent unsteady flow. The criteria for mesh refinement used in this work are derived from the equation for turbulent viscosity in the LES turbulence model. By using a modification to the turbulent viscosity as a refinement variable there is a link between both DGA algorithm and turbulence model. The smaller scale turbulence is modelled via the LES turbulence model, while the larger scales are resolved. In comparison with the simulations using a conventional mesh, substantial reduction in mesh size has been obtained with the use of a DGA algorithm. The reduction in mesh size is obtained without a decay in the quality of the prediction. It is shown that the use of a DGA algorithm in the context of turbulence modelling is a suitable tool which can be used as a next step in an attempt to resolve turbulence more realistically.