Improving the predictive capability of thermal-hydraulic modeling and simulation tools with physics-informed machine learning
Abstract: Thermal-Hydraulic (T-H) Modeling and Simulation (M&S) tools play an important role in the design, licensing, and safety analysis of nuclear reactors. In this talk, two approaches based on physics-informed machine learning will be discussed as potential methods to improve the predictive capability of T-H M&S tools. The first is a Bayesian inference procedure for the uncertainty quantification and reduction of M&S tools. The second is the development of non-parametric closure relations informed by important physical features through deep learning. Applications based on Multiphase-CFD solver for these two approaches will be demonstrated and existing limitations will be discussed.
Bio: Dr. Yang Liu is currently a postdoctoral appointee in the Nuclear Engineering Division at Argonne National Laboratory (ANL). He obtained his Ph.D. degree in Nuclear Engineering at NC State University in 2018. Prior to joining ANL in 2019, he was a postdoctoral research fellow at The University of Michigan, Department of Nuclear Engineering and Radiological Sciences. He is currently a code developer for System Analysis Module (SAM) at ANL. His other areas of technical expertise include model validation, physics-informed machine learning, and uncertainty quantification and reduction for thermal-hydraulic modeling and simulation tools. He is the author of a book chapter, 7 journal papers, more than 10 conference proceedings, and multiple technical reports.
