Large Eddy Simulation of Scalar Mixing

Abstract

The chapter describes the implementation of a passive scalar transport equation into an existing large eddy simulation (LES) code, including assessment/testing of alternative discretization schemes to avoid over/undershoots and to prevent excessive smoothing. The numerical schemes are evaluated using a simple test case involving pure scalar convection in a specified velocity field. Both second order accurate (away from extrema) TVD and higher order accurate dispersion relation preserving (DRP) schemes are assessed for use in solving the scalar conservation equation in order to optimize numerical behavior in terms of both dispersive and diffusive errors. Although the best performance is displayed by a DRP method, this is only true on reasonably fine meshes, it produces similar (or even slightly larger) errors to TVD scheme on coarser meshes, and at present the TVD approach is retained for LES applications. The unsteady scalar mixing performance of the LES code is then validated in turbulent flow against published direct numerical simulation (DNS) data for a slightly heated channel flow. Excellent agreement between current LES predictions and the DNS data is obtained, for both velocity and scalar statistics.