Interaction of the filter and spatial discretisation operators for Large-Eddy Simulation using the Approximate Deconvolution Model

Abstract

The Approximate Deconvolution Model (ADM) for Large–Eddy Simulation is an approach for the computation of turbulent flows. The main idea of ADM is to approximate the unfiltered data in the filtered Navier–Stokes equations by deconvolution of filtered values. This is achieved via repeated filtering. Thereby problems may arise for cases where the order of the spatial discretisation scheme is too low or lower than the order of the filter operation. The interaction of the discretisation and the filtering was investigated for one–dimensional forced Burgers turbulence. For this study, both the order of the filter operator and the order of the spatial discretisation of the derivatives were varied independently. It was found that the use of ADM with low–order discretisation schemes can not be recommended.