A STOKES NUMBER-BASED STOCHASTIC IMPROVEMENT FOR DISPERSION MODEL FOR LARGE EDDY SIMULATION

Abstract

To improve the fidelity of large eddy simulation (LES) of spray jet dispersion, a dynamic subgrid dispersion model is proposed based on the Langevin-type stochastic framework to quantify the effective contribution of the stochastic component of the force as a function of the Stokes number related to the subgrid time scale, which is easily accessed by the LES closure model. The proposed model has two coefficients that require calibration, which were obtained following a rigorous calibration procedure based on forward uncertainty quantification algorithms. The performance of the model is assessed by comparison against a reference direct numerical simulation (DNS) test case. The comparisons for the spray analysis include averages of the number of droplets, mass source term, and droplet diameters conditioned on the vapor mass fraction, together with their Eulerian average at different axial locations. The results showed improved prediction of the particle clustering behavior near the nozzle exit observed in the DNS simulations.