Recent Advances and Trends in Computational Thermal-hydraulics: A Personal Reflection

Abstract

Computation is poised to fundamentally reshape the future of thermal-hydraulics in nuclear engineering. Over the past two decades, the field has evolved dramatically, particularly in high-end simulation: from relatively modest single-assembly calculations to fully resolved, full-core simulations on exascale supercomputers. Yet the most important advances are still ahead. The future of computational thermal-hydraulics will not be defined simply by ever larger simulations, but by the ability to integrate modeling across scales, combine physics-based methods with artificial intelligence, and translate high-fidelity predictions into practical engineering decisions.
The emergence of exascale computing has enabled simulations that only a few years ago seemed unattainable. For the first time, fully coupled, pin-resolved calculations of entire reactor cores can be performed on machines such as Frontier using thousands of GPUs. Using tools such as NekRS, these simulations provide unprecedented insight into turbulent flow, heat transfer, and multiphysics interactions in advanced reactor systems. They are also revealing new physical phenomena and challenging long-standing assumptions in thermal-fluids modeling. The future of the field lies not in pursuing the largest possible calculation for its own sake, but in developing computational ecosystems in which high-fidelity simulations, reduced-order models, engineering tools, and experimental data work together. Exascale simulations will increasingly serve as “numerical laboratories,” generating the understanding needed to improve lower-order models used for design, licensing, and operation. Artificial intelligence and machine learning will play an essential role in this process through automation, surrogate modeling, uncertainty quantification, and the reconstruction of high-resolution fields from lower-fidelity simulations. This talk reflects on these opportunities, as well as the challenges that must be overcome to realize them.

Bio

Elia Merzari is a professor and current associate department head at the Ken and Mary Alice Lindquist Department of Nuclear Engineering at Pennsylvania State University. He also has an appointment in the Department of Mechanical Engineering and in the Institute for Computational and Data Science. He served in various roles at Argonne National Laboratory between 2009 and 2019. His expertise covers modeling and simulation of advanced reactors including safety analysis for a range of reactor types. Since 2019 he is a member of the faculty at Penn State. He has received several awards related to these efforts on the area of high performance computing (HPC), including the American Nuclear Society (ANS) Landis Young Member Engineering Achievement Award, the American Society of Mechanical Engineers (ASME) George Westinghouse Silver Medal, the ANS Bal-Raj Sehgal Memorial Award, and the ANS Cisler-Untermeyer Reactor Technology Medal. He is a fellow of ASME and ANS. He was a finalist for the Gordon Bell prize for 2023.