Dynamic Contact Angle Modeling Using Bayesian Inference

Abstract While autonomous vehicles (AV) technology is rapidly advancing, achieving fully autonomous driving in all conditions with no human intervention is still a significant challenge. Weather management and sensor cleaning are important considerations for reliability of AV, leading to design of hydrophobic durable sensor coatings to withstand wet conditions. High-fidelity prediction of droplets moving on surfaces with different wettability requires accurate dynamic contact angle modeling, and several empirical dynamic contact angle models have been developed. Such models include unknown parameters to be tuned for different applications, and traditionally require contact angle measurements at droplet terminal velocity limiting the range of conditions that can be explored through experiments. Here, Bayesian inference approach is implemented to learn the probability distribution of each of these parameters over a range of operating conditions relevant to water management in AV. Droplet transient velocity data as well as measurements at terminal states are used to infer dynamic contact angle model parameters. The reinforced contact angle models improve prediction of droplet dynamics through 3D high-fidelity simulations.