High-Fidelity Simulations of Air-Blast Atomization with Predictive Droplet Statistics

Air-blast atomization is a process in which a high-speed gas stream shears and fragments a slower-moving liquid into many small droplets. Accurately predicting the resulting droplet statistics is critical for a wide range of engineering applications, including fuel combustion, agricultural spraying, and spray coating. However, such predictions remain computationally challenging due to the broad range of length and time scales involved, particularly during the formation and breakup of thin liquid structures such as ligaments and bags. In this work, we employ a multi-scale simulation framework by dividing the computational domain into three regions: nozzle injection, liquid atomization, and spray dispersion. The internal gas flow within the nozzle is computed using a cost-efficient single-phase solver, which provides inflow conditions to the atomization region. There, the liquid–gas interface is captured using a geometric volume-of-fluid (VOF) method with Reconstruction with Two Planes (R2P), enabling the representation of the sub-grid scale films that occur during bag break-up. These thin structures are then converted into Lagrangian droplets using a physics-based break-up model. The resulting Lagrangian particles are tracked and advected downstream in the dispersion domain, where droplet statistics are extracted. We compare the predicted droplet statistics against experimental measurements and demonstrate good agreement enabled by this coupled modeling strategy.