The impact of dynamic models for applications to LES of compressible flows is assessed in the framework of a numerical model based on high order discontinuous finite elements. The projections onto lower dimensional subspaces associated with lower degree basis functions are used as LES filter, along the lines proposed in Variational Multiscale templates. Comparisons with DNS results available in the literature for plane and constricted channel flows at Mach numbers 0.2, 0.7 and 1.5 show clearly that the dynamic models are able to improve the prediction of most key features of the flow with respect to the Smagorinsky models employed so far in a VMS-DG context

No abstract

The definition of the characteristic filter size to be used for subgrid scales models in Large Eddy Simulation using irregular grids, is still an unclosed problem. We investigate some different approaches to the definition of the filter length for anisotropic subgrid scale models and we propose a tensorial formulation based on the inertial ellipsoid of the grid element. The results demonstrate an improvement in the prediction of several key features of the flow when the anisotropicity of the grid is explicitly taken into account with the tensorial filter size

The Hybrid RANS/LES method proposed is based on the hybrid filter approach introduced by Germano in 2004. In this work, instead of using two explicit models respectively for LES and RANS, we use only a model for LES and we reconstruct the RANS field exploiting the properties of hybrid filter. The reconstruction is obtained from the resolved velocity and the LES subgrid stress tensor. The model has been implemented using a variational approach in a DG-FEM framework and tested for the turbulent channel flow test case. Different configurations for hybrid terms have been compared with DNS data and with pure LES results

No abstract