Assesment of Modeling and Discretization Error in Finite-Volume Large Eddy Simulations

Abstract

Large eddy simulations of turbulent flows are known to suffer from two separate error sources: the subgrid stress model and the numerical discretization scheme. In general, the two sources of error cannot be separately quantified for finite-difference/finite-volume CFD simulations. The motivation of this paper lies in the desire to determine optimum combinations of currently available subgrid stress models and numerical schemes for use in large eddy simulations of complex flows. Error assessments for large eddy simulation of turbulent fluid flow are presented. These assessments were carried out using pseudospectral simulation techniques in order to isolate finite-differencing and modeling errors by explicitly adding numerical derivative error terms to the simulations and analyzing their effect. Results from several combinations of subgrid stress model and spatial discretization scheme are presented. Simulations were performed for decaying isotropic turbulence on both 323 and 643 grids. Results were compared in terms of spectral energy distributions at succeeding time intervals. For verification, the pseudo-spectral results were compared to LES solutions obtained from a commercially available finite-volume flow solver (FLUENT), and comparisons were made in terms of energy decay rates, numerical versus subgrid stress dissipation levels, and computed energy spectra. The results highlight the interaction between subgrid stress model and discretization scheme. The results also indicate that certain combinations of model and numerical scheme may be more appropriate for finite-volume LES than others.