Liquid Structure Classification Towards Topology Change Modeling

Abstract

We present a method that, in combination with a two-plane liquid-gas interface reconstruction in each computational cell, identifies thin features in Eulerian simulations of spray atomization that should undergo imminent breakup. Rapid identification of these features, which include gas sheets between colliding droplets, liquid sheets generated in bag breakup, and ligaments that breakup through Rayleigh-Plateau instabilities, allows for the on-the-fly application of subgridscale (SGS) breakup models for each type of thin feature in order to accurately predict the resulting topology change behavior. Without the feature identification abilities of the proposed method, thin features would experience numerical breakup when their characteristic size falls below that of the computational mesh. Such topology change is mesh-dependent, and therefore, unphysical. The proposed method incorporates the interface normal and centroid into any underlying connected-component labeling (CCL) algorithm to distinguish thin features from other fluid structures. We apply the method on data from simulations of offset binary droplet collision and turbulent air-blast atomization to verify its ability to identify individual thin features and demonstrate its usefulness for calculating relevant properties of the identified features.