Title:

Towards data-driven fuel performance modeling for accelerated research and development of nuclear fuels.

Speaker:

Dr. Yifeng Che

Affiliation:

Idaho Falls, Idaho

When:

Tuesday, February 27, 2024 at 11:00:00 AM   Add to Calendar

Where:

MRDC Building, Room 4211

Host:

Dr. Dan Kotlyar
dan.kotlyar@me.gatech.edu

Abstract

The development of nuclear reactors with enhanced safety, better economics and modern design have largely motivated the study of nuclear fuels. As the heart of a nuclear reactor, the nuclear fuel undergoes a series of complicated thermo-mechanical-chemical degradation, which poses safety constraints to reactor operation and design optimization. Multiple challenges remain in nuclear fuels research despite that significant progress have taken place in advanced modeling and simulation capabilities over the past decades. The primary challenge in nuclear fuels relates to the integration of modeling and simulation with experiments. Empirical models used to be fitted according to scarce experimental data, while the transition to mechanistic models requires efficient methodologies to integrate experimental data into the physics-based modeling and simulation. Moreover, simulation of essential fuel failure modes requires high-fidelity fuel performance modeling which is accompanied by high computational cost. As a result, the current design optimization of reactor cores is void of detailed fuel performance but augmented with simplified fuel performance feedback. In this seminar, Dr. Che will introduce a novel Bayesian inference framework that efficiently integrates experimental measurements into expensive computational tools. She will also discuss the power of machine learning techniques to address the issue of high computational cost. The seminar will be concluded by an outlook of future nuclear fuels research.

Biography

Dr. Yifeng Che is a Computational Scientist at Idaho National Laboratory (INL). She is a member of the development team for Bison, an advanced fuel performance modeling code based on MOOSE. Her primary research area of expertise is multi-physics modeling and simulation of advanced nuclear fuel concepts. She has worked on improving the reliability of fuel performance modeling through integrated validation/verification, inverse/forward uncertainty quantification, and Bayesian optimized experimental design. She has also developed reduced-order models for fuel performance predictions using AI/ML techniques and coupled high-fidelity fuel performance models with neutronics and thermal-hydraulics under the MOOSE framework. Dr. Che received her Ph.D. in Nuclear Science and Engineering from Massachusetts Institute of Technology in 2021 and worked at INL as a Russel L. Heath distinguished postdoctoral associate between 2021 and 2023.

Notes

Refreshments will be served.