Multiscale numerical modeling of a complete spray evolution including breakup of liquid jet injection in gaseous cross flow

Abstract

A direct coupling of the Volume of Fluid method (VOF) and the Lagrangian Particle Tracking (LPT) technique within a Large Eddy Simulation (LES) framework is revisited and assessed by numerically investigating a liquid jet injection into a cross-flowing air (JICF). To close the subgrid scale (SGS) terms appropriate SGS models are considered. In particular, an interfacial SGS term is included into the VOF scalar transport equation in order to achieve a suitable LES and to describe properly the liquid column and the break-up processes (primary and secondary). The dilute spray region including the droplets resulting from the secondary breakup is solved using the LPT approach. Since the latter allows the cell size to be multiple times larger than the droplet diameter, this reduces the computational expenses significantly making it possible to handle the simulation of the whole spray evolution including the complete liquid atomization without complex tunable atomization models. For this purpose, two operating conditions of a JICF existing in the literature are considered. The achieved results are first presented in terms of mesh sensitivity, required computational costs and impact of the interfacial SGS term. Next, comparisons between pure VOF/LES and VOF/LPT/LES for both qualitative features and quantitative properties are provided and discussed. In overall, a good agreement with experiments for the quantitative properties is reported. It is especially found that both the Rosin–Rammler and Gaussian distributions offer nearly a more accurate fit to the particle size distribution except for the range from 0 to 10 μm. Together with the detailed characteristics of vortices evolving and an extended breakup regime map, the spray characteristics including the spray structure are well captured. Finally, even though the interfacial SGS term has an evident influence on the jet and atomization dynamics, its impact on quantitative properties is not so prominent under these operating conditions.