Expert‐in‐the‐loop design of integral nuclear data experiments
Abstract Nuclear data are fundamental inputs to radiation transport codes used for reactor design and criticality safety. The design of experiments to reduce nuclear data uncertainty has been a challenge for many years, but advances in the sensitivity calculations of radiation transport codes within the last two decades have made optimal experimental design possible. The design of integral nuclear experiments poses numerous challenges not emphasized in classical optimal design, in particular, constrained design spaces (in both a statistical and engineering sense), severely under‐determined systems, and optimality uncertainty. We present a design pipeline to optimize critical experiments that uses constrained Bayesian optimization within an iterative expert‐in‐the‐loop framework. We show a successfully completed experiment campaign designed with this framework that involved two critical configurations and multiple measurements that targeted compensating errors in 239 Pu nuclear data.
- Sponsoring Organization:
- USDOE
- OSTI ID:
- 2332898
- Alternate ID(s):
- OSTI ID: 2350639
- Journal Information:
- Statistical Analysis and Data Mining, Journal Name: Statistical Analysis and Data Mining Journal Issue: 2 Vol. 17; ISSN 1932-1864
- Publisher:
- Wiley Blackwell (John Wiley & Sons)Copyright Statement
- Country of Publication:
- United States
- Language:
- English
Similar Records
Constrained Bayesian Optimization of Criticality Experiments [Slides]
Constrained Bayesian Optimization of Criticality Experiments at LLNL [Slides]