Yoonjin Won, associate professor of mechanical & aerospace engineering, has won a $6 million multidisciplinary university research initiative from the U.S. Department of Defense for her project to improve the efficiency of thermal management systems for Navy power and energy applications. The UC Irvine-led project, “Fundamentals of Machine Learning for Phase Change Heat Transfer,” aims to develop an intelligent framework for liquid-vapor phase change physics that integrates advanced metrology with computer vision and machine learning.

Phase change heat transfer, involving boiling or condensation, has been used for over 100 years in energy and power systems. It is an essential element of modern building systems, transportation, refrigeration and power generation. The physics of phase change heat transfer are incredibly complicated due to their chaotic nature and the multiple physical processes present in these systems. The extreme complexity of phase change heat transfer makes it nearly impossible to perform modeling and simulation that can predict phase change flows and be used to design phase change systems.

“With recent advances in computer vision and machine learning, we have an opportunity to overcome these challenges,” Won said. Her team includes Aparna Chandramowlishwaran, UC Irvine associate professor of electrical engineering & computer science; Jiajun Wu from Stanford University; Nenad Miljkovic and William King from University of Illinois Urbana Champaign; and Matteo Bucci from MIT.

“This project will build new approaches that use physics-based machine learning methods that can interpret and predict such flows, and enable the design of new systems that are currently impossible to develop using conventional engineering methods,” said Won.