Novel Tools for Atomization Simulation Analysis

We present a comprehensive methodology for analyzing high-fidelity atomization simulations using the Atomization Simulation Statistics Extraction Tool (ASSET) to derive novel insights from liquid breakup events. This approach involves extracting statistics in addition to capturing the full geometry of the breakup and the surrounding flow field. These data are then systematically organized into a database using the Weighted Euler Characteristic Transform (WECT). This results in a vectorized representation of the breakup data, making it easily parsable, queryable, and ideal for advanced statistical analysis techniques. The goal of this work is to present the methodology and display the potential utility of easily accessible, vectorized atomization databases.

The present work develops a database from a simulated liquid jet. We apply a simple k-means clustering algorithm to the resultant WECT database. Results show promise for comparing liquid shapes from reduced vectorized data. This methodology was developed with a future reduced-order model in mind, specifically, the vectorized WECT database is particularly well-suited for developing a predictive Lagrangian droplet tracking model. Such a model can predict the breakup behavior of droplets in ongoing simulations by connecting them to the most similar droplets in the WECT database, based on shape, size, and flow field. Key results show unique trends identified from topological and geometric data intrinsic in large-scale atomization data.