Timothy Franck, VT Graduate Student, will give a talk entitled "Development of Physics-Informed Machine Learning Methodologies Using RAPID & the Jožef Stefan Institute (JSI) TRIGA Mark-II research reactor"

September 12, 2025

@ 10:10 am, 6-051 VTRC, Arlington, VA (in-person), 440 Goodwin Hall, Blacksburg, VA
For remote access, register here!

Abstract
The rapid growth of artificial intelligence (AI) in recent years is evident by its widespread integration across multiple industries and the corresponding expansion of data centers built to support its intensive energy demands. With commitments to decarbonize by 2030, hyperscalers are considering nuclear reactors because they provide reliable base-load power, have high power density, and produce zero carbon emissions. Small modular reactors (SMRs) are receiving the most attention, but so far, no design has been commercialized. If the nuclear industry aims to position itself as a major contributor to future energy production, it must successfully commercialize SMRS, alongside other advanced reactor designs, in the coming years.

In support of this objective, fast and accurate high-fidelity simulations are necessary for design, safety analysis, licensing, and operation of next generation reactors. However, the most common methodologies for high-fidelity simulations of nuclear reactors are slow and computationally expensive. Machine learning (ML) may offer a solution, as it enables computers to learn from data, allowing well-trained models to produce results quickly and accurately. Yet ML applications within neutronics remain limited, as they are often trained on simplified 2-D or coarser 3-D problems.

Over the past 15 years, the Virginia Tech Transport Theory Group (VT3G), led by Dr. Alireza Haghighat, has been developing the RAPID code system. RAPID enables very fast and accurate high-fidelity simulations of nuclear reactor systems, making it ideal for developing complete, physics-informed datasets for ML. This seminar presents results from my thesis, which developed physics-informed ML methodologies using RAPID to predict eigenvalue and 3-D fission distributions as a function of control rod positions for the Jožef Stefan Institute (JSI) TRIGA Mark-II research reactor.

Bio
Timothy Franck received a B.S. in Chemical Engineering with a minor in Nuclear Engineering from Virginia Tech. He is nearing completion of his M.S. in Nuclear Engineering from Virginia Tech under the advising of Dr. Alireza Haghighat and plans to continue pursuing a Ph.D. under his guidance. Timothy’s research involves developing physics-informed machine learning methodologies using RAPID and applying it for the Jožef Stefan Institute (JSI) TRIGA Mark-II research reactor. His work also involves performing neutronic simulations for the design of a novel research and education reactor with the Virginia Innovative Nuclear Hub (VIN Hub).