Optimization of Pressurized Water Reactor Shuffling by Simulated Annealing with Heuristics

TLDR

The study explores incorporating heuristics into simulated annealing for reactor core loading pattern optimization. The authors implement a simulated‑annealing optimizer (SIMAN) that generates candidate loading patterns using design heuristics, evaluates them with the SIMULATE‑3 nodal method and CASMO‑3 cross‑section data, and automatically assesses successive cycles to capture multicycle effects. Heuristics enhance consistency of simulated‑annealing results without sacrificing performance, enabling SIMAN to efficiently identify loading pattern families that meet design constraints and achieve strong objective performance within realistic runtimes.

Abstract

AbstractSimulated-annealing optimization of reactor core loading patterns is implemented with support for design heuristics during candidate pattern generation. The SIMAN optimization module uses the advanced nodal method of SIMULATE-3 and the full cross-section detail of CASMO-3 to evaluate accurately the neutronic performance of each candidate, resulting in high-quality patterns. The use of heuristics within simulated annealing is explored. Heuristics improve the consistency of optimization results for both fast- and slow-annealing runs with no penalty from the exclusion of unusual candidates. Thus, the heuristic application of designer judgment during automated pattern generation is shown to be effective. The capability of the SIMAN module to find and evaluate families of loading patterns that satisfy design constraints and have good objective performance within practical run times is demonstrated. The use of automatic evaluations of successive cycles to explore multicycle effects of design decisions is discussed.