ISS: Predicting condensation and freezing behavior via a machine learning model

Frost and ice formation negatively impact many industries including aviation, refrigeration and air-conditioning systems, consumer devices, clean water, and agriculture. The global deicing market for commercial aircrafts reached one billion dollars in 2019, and condensate freezing on external heat exchangers in air-source heat pumps can reduce performance by 35-60%. The main objective of this project is to better understand frost and ice formation to enable the engineering of surfaces and processes to prevent the negative effects of frost formation. The microgravity environment of the International Space Station will slow the growth of ice crystals, which will enable the visualization and study of frost and ice formation mechanisms. Videos of the surfaces will be analyzed using machine learning to automatically track and quantify condensation (i.e., the formation of liquid droplets) and frost characteristics. Another objective of the project is to share results with the public through a local science museum as well as summer camps for middle school and high school girls. The goal of this project is to use a microgravity environment to gain a deeper understanding of condensation, frost, and ice formation and use machine learning to develop models for predicting condensation and freezing behavior. The project has three main scientific objectives. The first objective is to investigate condensation, including droplet nucleation and droplet dynamics, in microgravity and gravity environments and use machine learning to understand droplet growth rates, and frequency of coalescence events. The suppression of natural convection in microgravity will enable the quantification of convection and radiation during condensation from moist air. The second objective is to investigate freezing front propagation in microgravity and gravity environments. This would provide fundamental new insights into the different mechanisms that propagate frost formation. Machine learning will enable the development of mechanistic models to predict frost propagation rates. The third objective is to investigate ice crystal growth from the initial ice nuclei in microgravity and gravity environments. Machine learning will be utilized to characterize crystal growth and create models to understand factors that impact crystal growth and dendritic formations. This would provide fundamental new insights into the role of water vapor dynamics and heat transfer on frost growth enabling the development of mechanistic models to optimize surface design for controlling ice formation. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.