LES-Lagrangian particle method for turbulent reactive flows based on the approximate deconvolution model and mixing model

We propose a numerical method for turbulent reactive flows using a large eddy simulation (LES) based on the approximate deconvolution model (ADM). LES based on the ADM is combined with a Lagrangian notional particle (LP) method for computing reactive flows without using models for chemical source terms. In the LP method, values of scalars are assigned to each particle. The evolutions of Lagrangian particles in physical and scalar composition spaces are modeled by using the mixing model for molecular diffusion and the resolved velocity field of LES. We also propose a particles-interaction mixing model using a mixing volume concept, in which the mixing timescale is determined by relating the decay of scalar variance in the mixing volume to the scalar dissipation rate. The LES-LP method based on the ADM and the mixing model is applied to a planar jet with a second-order reaction for testing the numerical method. The statistics obtained by the LES-LP method are compared with the direct numerical simulation data. The results show that the evolutions of Lagrangian particles are well modeled in the LES-LP method by using the resolved velocity and the mixing model, and this method can accurately predict the statistical properties of reactive scalars. The mixing timescale depends on the distance among the Lagrangian particles. It is also shown that the present mixing model can implicitly take into account the effect of distance among the particles by adjusting the mixing timescale without using any model parameters.