BubbleMask: Autonomous visualization of digital flow bubbles for predicting critical heat flux

Abstract

Flow boiling utilizes the latent heat of the fluid to provide an efficient thermal management solution through bubble-induced advective transport. In microgravity environments, flow boiling becomes even more advantageous as macroscale flow helps efficiently remove bubbles from the heated wall, resulting in enhancing heat transfer as well as critical heat flux (CHF). However, connecting flow boiling physics and bubble information is a challenging task due to the complexity and high dimensionality of bubble dynamics. To overcome this challenge, the advances in computer vision techniques and models such as VISION-iT can be leveraged to autonomously extract physically meaningful features related to spatial statistics, interfacial characteristics, and bubble dynamics by digitalizing flow bubble information. In this study, 30,000 flow boiling images under microgravity conditions are used to compute ten different features of 155,000 individual bubbles. The extracted bubble information is then used to predict CHF by using a classical flow boiling model, the Interfacial Lift-off Model. This vision-based approach suggested here has the potential to revolutionize the study of such thermofluidic topics by providing visual insights that agree well with experimental data.