A Feasibility Study on In-Core Fuel Management via Quantum Particle Swarm Optimization

Fuel loading operations in a nuclear reactor core have been calling for adequate methods to determine fuel assemblies configurations able to optimize required features. The current paper leads an investigation over already existing optimization algorithms, aiming to target and further develop a promising method to tackle in-core fuel management optimization problems. To this aim, Genetic Algorithms (GA) and Particle Swarm Algorithm (PS) were selected, tested and eventually compared using an analytical multi-variable continuous test function. Results obtained showed Particle Swarm to have better performances with respect to both accuracy and machine time. The successive analyses were focused on Particle Swarm (PS) and its updated version, Quantum Particle Swarm (QPS). Particle Swarm and a modified version of the original Quantum Particle Swarm algorithm, here proposed by the present authors, were tested with respect to their efficiencies and performances on a reactor physics case study. The adopted case study regards an extremely simplified PWR core. The goal was to optimize fuel assemblies disposition in order to obtain the highest value of the multiplication factor, keff. The calculations were performed coupling Particle Swarm based algorithms with Serpent Monte Carlo code. Obtained results showed a clear superiority in terms of calculation time of the Quantum Particle Swarm based algorithm proposed by the authors. This speaks in favour of its application to more complex in-core fuel management optimization cases.