Application of a genetic algorithm to the fuel reload optimization for a research reactor

This article presents results from an application of a genetic algorithm (GA) to the fuel reload optimization for a research reactor. In this work, we proposed an improved model of the problem and a new coding procedure for the GA to automatically search for optimal fuel loading patterns most suitable for the research reactor. The model consists of an objective function, which maximizes the effective multiplication factor and minimizes the power peaking factor, and operational and safety constraints. The new coding procedure is used to handle a constraint on the limited number of fuel shuffles in a refueling operation. The GA works with an elitist selection based on the elitism strategy and the roulette wheel spin method, a modified one-point crossover and a simple mutation. A computer program was developed in FORTRAN 90 running on a Pentium III personal computer to perform illustrative calculations for a research reactor type TRIGA MARK II. Results from illustrative calculations show that the GA can successfully search for the optimal loading patterns, which can be employed to establish a simple refueling scheme for the reactor with a limited number of fuel shuffles in a practical refueling operation.