MURI Fundamentals of Machine Learning for Phase Change Heat Transfer (23-000005233)
Abstract
Two-phase heat transfer forms the foundation for modern energy generation and thermal management systems in DoD applications. Whilethe fundamentals of two-phase heat transfer have been studied for over a century, key scientific questions remain regarding the fundamental mechanisms governing this process. Furthermore, challenges remain related to the interpretation of experimental data as well as the prediction of two-phase phenomena. Two-phase heat transfer features multiple physical processes with coupled nonlinear behavior across various length and time scales. The generation and growth of bubbles, droplets, and films on solid surfaces are dynamic surface phenomena that are coupled to the macroscale flows. The evolution of the liquid-vapor interface is mutually interdependent with mass transport, leading to nonlinear behavior where small changes in one parameter can result in significant and unexpected effects on other parameters. The multimodal, multidimensional, and transient nature of these processes leads to challenges for investigating and understanding the fundamental mechanisms governing two-phase heat transfer. We propose a Multidisciplinary University Research Initiative (MURI) that will pursue a radical new approach for studying two-phase heat transfer leveraging recent advances in artificial intelligence. The research will establish an intelligent framework for studying the fundamentals of two-phase heat transfer,integrating advanced metrology, computer vision, and machine learning. The research devotes special attention to collecting rich and high-fidelity datasets from imaging and physical sensors and fusing these datasets with modern architectures for computer vision and machine learning under a new paradigm. The research is organized into six thrusts. Thrust 0 will provide the foundations for metrology and design for collecting datasets to be used across the project. Thrust 1 will study two-phase nucleation behavior and its dependance on various surface phenomena. Thrust 2 will interpret and evaluate mass transfer near phase boundaries. Thrust 3 will studywall heat transfer by predicting temperature and heat flux from phase field data. Thrust 4 will generate accurate, high-resolution,and transient four-dimensional flow fields of two-phase flows from multi-view information. Thrust 5 will study two-phase flow instabilities in the presence of new geometries. We have assembled an interdisciplinary team of researchers who are committed to the project. The project will include the training of approximately 12 graduate students who will comprise the future workforce with expertise across two-phase heat transfer and computer science. We will collaborate with DoD laboratories and personnel, DoD contractors, and technology companies to ensure that the research products are relevant and impactful. The proposed research will accelerate new phenomenological discoveries and result in new fundamental understanding of two-phase heat transfer. The research will provide new capabilities for understanding, designing, and operating efficient thermal management systems for Navy power and energy applications. Approved for Public Release.
Document Details
- Document Type
- DoD Grant Award
- Publication Date
- Nov 08, 2024
- Source ID
- N000142412575
Entities
People
- Yoonjin Won
Organizations
- Office of Naval Research
- United States Navy
- University of California, Irvine