Statistical modeling of electrowetting-induced droplet coalescence for condensation applications

Coalescence of water droplets strongly influences dropwise condensation on hydrophobic surfaces. This work reports a study on experimental data-based statistical modeling to predict the coalescence dynamics of an ensemble of water droplets under the influence of an electrowetting (EW) field. Previous related studies have primarily used high speed visualization to characterize coalescence. However, this study uses statistical modeling to analyze the parameter space associated with EW-induced coalescence, noting that physics-based modeling of EW-induced coalescence is challenging. The objective of this study is to quantify the influence of the applied voltage, frequency of the AC waveform and the geometry of the EW device on two parameters related to droplet coalescence (droplet radius enhancement and reduction in wetted area). Multiple supervised learning techniques are used to identify dominant variables and statistically model the influence of these variables on coalescence. Data for the statistical models is obtained via image analysis from coalescence experiments. The statistical models lead to a reference tool to predict droplet coalescence-related parameters versus the applied voltage and electrode geometry. Importantly, data analysis shows that droplet coalescence is independent of the AC frequency; this conclusion would be challenging to infer from conventional analysis. It is also seen that an EW field significantly narrows the droplet size distribution. Overall, this study leads to a detailed understanding of the factors that impact EW-induced coalescence and provides a tool (which matches experimental data) to predict the change in droplet size distribution. These findings are key to quantifying the influence of EW on condensation rates and heat transfer. This work leverages the large amount of data from experiments to develop statistical analysis-based predictive models. This approach can be utilized for predictive modeling of other data-rich but complex physical phenomena.