Numerical Study of Resolution Effects on the Statistics of SGS Scalar Mixing Models for LES

Abstract

This paper investigates modeling approaches for subgrid-scale (SGS) scalar transport in large-eddy simulations (LES), by applying different models in LES of mixing in axisymmetric turbulent jets. The LES uses a spherical coordinate geometry, and the flow field computation uses the dynamic Smagorinsky model for the SGS stress, where the dynamic coefficient is determined using the Lagrangian averaging procedure. The models applied for the SGS scalar transport are the eddy-diffusivity (ED) model, a mixed scalesimilarity/eddy-diffusivity model, the dynamic structure (DS) model, and a multifractal (MFR) model. We consider three grid resolutions at jet Reynolds number Red =5,000. Mean scalar field results are very similar for the ED, mixed and MFR models, with the DS model being the significant outlier. The DS model also shows the greatest sensitivity to spatial resolution. The performance of the models in reproducing mean scalar fields appears to be more dependent on the characteristics of scalar energy transfer across the LES grid scale than on the model formulations themselves.