Abstract
This study proposes a subgrid-scale (SGS) dispersion approach for large-eddy simulation (LES) of spray. Rather than associating the turbulence correlation timescale used in the dispersion approach with the computational grid, we put forward an approximation of this timescale proportional to the Kolmogorov timescale. The Kolmogorov scale kinetic energy is added to the variance of the stochastic part of the SGS dispersion velocity to account for the effects of micro-scale flows on droplets. The proposed approach supports the notion of grid-independent LES and is illustrated for two benchmark Engine Combustion Network's Spray-A cases, including a vaporizing spray case and a non-vaporizing spray case. We investigate the performance of the approach in LESs of two cases by comparing it with other dispersion approaches. In particular, the investigation mainly covers spray structures, vapor and liquid penetrations, and mass fraction profiles. Results obtained using the proposed approach shows the statistics of the simulated spray cases that exhibit both good agreement with measurements and a noticeable improvement over simulations based on other tested approaches. The approach also yields more consistent results under different grid resolutions, which shows its ability to achieve LESs in non-uniform grids.