Approximate deconvolution and velocity estimation modeling of decaying Burgers turbulence

The paper presents Burgers turbulence simulated using Large Eddy Simulations (LES). Two types of subfilter models are applied: Approximate Deconvolution Model (ADM) and Velocity Estimation Concept (VEC). In the ADM approach, two different filter types were applied indicating the influence of the filter on the deconvolution convergence via comparison with a direct inversion method. In both approaches the ADM and VEC the deconvolved and estimated fields are shown. It is stressed that ADM introduces flow structures characterized by subfilter scales down to the mesh cut-off length scale. In the case of the VEC only the kinematic step was applied for the subgrid velocity field estimation. We observe that both ADM and VEC provide predictions that are closely related to direct numerical simulations (DNS) provided the spatial resolution is adequate. Moreover, results based on the top-hat filter as the basis for LES are in general agreement with results obtained on the basis of a sixth-order Padé filter. The VEC model provides further opportunities to include much finer scales that evolve under simplified approximate dynamics, thereby enabling additional fine-tuning of the predictions of higher-order statistical quantities.