Optimization of moderator materials by NSGA II based on macroscopic cross-sections: applications in accelerator neutron sources

Abstract

Abstract In recent years, genetic algorithms have been applied in nuclear technology design, which have been shown to produce optimized results more efficiently than traditional enumeration methods. This advancement in optimization techniques is particularly useful in the field of nuclear technology design, where complexity is high and decision-making time is critical. It can be used to optimize moderator materials for ANS to find composite materials that provide high neutron beam quality. At present, the direct combination of Monte Carlo method and genetic algorithm requires a lot of computing resources and time. And the weights of different optimization objectives are controversial. Thus, we propose a two-step method based on NSGA II, which uses macroscopic section as the intermediate parameters for optimization. It can greatly reduce the time of genetic algorithm optimization. The method is applied to the PAFA project of Sun Yat-sen University, the computational speed has been increased by 50 times based on a 50-generation optimization. And the results of the genetic algorithm show that the neutron beam obtained by using composite materials as moderator is 30.8% better than that obtained by using only MgF2 as moderator. The two-step genetic algorithm optimization has shown its great potential in the optimization problem of moderator materials.