Study of scalar transport in turbulent flows using direct numerical simulations

We investigate the properties of scalar transport central to the phenomena of turbulent mixing, with emphasis on both physical understanding and model development. Analyses are carried out using data obtained from high resolution direct numerical simulations (DNS) of incompressible isotropic turbulence. A parallel code is used to analyze data for ensemble averaged Taylor-scale Reynolds numbers (R_λ) up to about 240 (with 5123 grid points) and Schmidt numbers up to unity.; We study the structure of the dissipation field of scalar fluctuations and use conditional averages to quantify its relationship with fluctuations in the energy dissipation rate. The dominant mechanisms governing the evolution of scalar dissipation rate conditioned on the energy dissipation are identified as (a) nonlinear amplification due to strain rate fluctuations and (b) destruction due to molecular diffusion. Fluctuating scalar gradients are preferentially aligned along the principal axis corresponding to the most compressive strain rate eigenvalue, especially in regions of high energy dissipation rate. New insights gained through these studies are used to improve the Lagrangian Spectral Relaxation (LSR) model for single-scalar mixing by providing closures for the conditional scalar variance and conditional scalar dissipation rate equations. Extensions of these ideas to the case of multiple scalars in differential diffusion are also made. The effects of differential diffusion are illustrated via the behavior of “conditional correlation coefficients”, corresponding to the joint scalar dissipation rate and the scalar covariance. The statistics (especially conditional averages based on energy dissipation rate) of the joint scalar dissipation rate and terms governing its evolution are similar to their single scalar analogs. Reasonable agreement is demonstrated between DNS and a revised LSR model for differential diffusion.