Metaheuristic optimization in shielding design for neutrons and gamma rays reducing dose equivalent as much as possible

To obtain the optimum solution of the shielding design fast and accurate, 14 well known metaheuristic algorithms were studied in this study. Following the common practice, the algorithms were compared using several benchmark functions, and same number of function evaluations (NFE) was used in the comparison. An algorithm was tested by 30 independent runs in each function. Average of the NFE required to reach the global optimum and the success percentage were used to evaluate the performance of the various algorithms. The results showed that, the Genetic Algorithm (GA), the Differential Evolution algorithm (DE), the Shuffled Complex Evolution algorithm (SCE) and the Teaching-Learning-based Optimization algorithm (TLBO) have better performance than the others. Furthermore, the four algorithms were applied to optimize the shielding material, and 3 cases of shielding design were presented. As a typical case, the fission energy spectrum of 235U was used in the shielding design. After 3 independent runs for each case, and comparing the mean best function values, it was found that the algorithm of SCE performs best. It implies that the algorithm of SCE is a better choice among the algorithms used in this study to optimize the shield. In addition, it was found that the composite multilayer is a better arrangement than the simple multilayer.