Role of directional conditional filtering on droplet velocity and size distribution

This study investigates the role of directional filtering in the analysis of spray behaviour using two full cone nozzles (65° and 75°) emitting aqueous sprays under atmospheric conditions. A hybrid computational–experimental framework was applied, using Large Eddy Simulation (LES) with Discrete Phase Model (DPM) and Rosin-Rammler distribution in ANSYS Fluent, validated against Shadowgraph and Particle Image Velocimetry (PIV) measurements. The sensitivity of directional filtering to the effects of spray angle and flow rate (1.75 and 3.5 L/min), were also done. Upward-filtered droplets consistently exhibited greater breakup than both raw and downward-filtered data due to gravitational opposition and longer air exposure. For the 75° nozzle, upward-filtered droplets showed 82.15% reduction, 7% higher than raw (75.17%) and 9% more than downward-filtered (73.26%). For the 65° nozzle, upward-filtered reduction was 58.35%, exceeding raw (52.72%) by 5.6% and downward-filtered (51.12%) by 7.2%. At 1.75 L/min, upward-filtered droplets reduced by 85.21%, which is 10.8% more than raw data. In contrast, downward-filtered trends remained within 2% of raw values in all configurations. Moreover, upward-filtered droplets showed higher central density (0–0.05 m) but reaching 34.52%, 42.46%, and 33.17% of raw counts at 0.05–0.1 m for 75° nozzle, 65° nozzle (3.5 L/min), and 65° nozzle (1.75 L/min), respectively. Furthermore, upward-filtered droplets exhibited significantly greater velocity decay compared to raw data, with losses ranging from 28.9% to 48.3%, while raw ensemble reductions remained between 13.9% and 15.3%. Downward-filtered velocities, in contrast, differed by less than 2–8% from raw data, benefiting from gravitational acceleration and shorter travel paths. The outcomes highlight the need to consider conditionally filtered data in relation to spray atomisation.