Study on Ant Colony Optimization for Fuel Loading Pattern Problem

• Published in: Nihon Genshiryoku Gakkai wabun ronbunshi = Transactions of the Atomic Energy Society of Japan Vol. 12; no. 1; pp. 103 - 112
• Main Authors: KISHI, Hironori, KITADA, Takanori
• Format: Journal Article
• Language: English; Japanese
• Published: Atomic Energy Society of Japan 2013
• Subjects: Ant colony optimization; fuel management; Fuels; genetic algorithm; Genetic algorithms; Heuristic; loading pattern; Mathematical models; Mutations; optimization; PWR-type reactor; reactor fueling; Searching; Simulated annealing
• ISSN: 1347-2879; 2186-2931
• DOI: 10.3327/taesj.J12.018

Modified ant colony optimization (ACO) was applied to the in-core fuel loading pattern (LP) optimization problem to minimize the power peaking factor (PPF) in the modeled 1/4 symmetry PWR core. Loading order was found to be important in ACO. Three different loading orders with and without the adjacent effect between fuel assemblies (FAs) were compared, and it was found that the loading order from the central core is preferable because many selections of FAs to be inserted are available in the core center region. LPs were determined from pheromone trail and heuristic information, which is a priori knowledge based on the feature of the problem. Three types of heuristic information were compared to obtain the desirable performance of searching LPs with low PPF. Moreover, mutation operation, such as the genetic algorithm (GA), was introduced into the ACO algorithm to avoid searching similar LPs because heuristic information used in ACO tends to localize the searching space in the LP problem. The performance of ACO with some improvement was compared with those of simulated annealing and GA. In conclusion, good performance can be achieved by setting proper heuristic information and mutation operation parameter in ACO.