Study on multi-objective optimization for nuclear reactor radiation shield design coupling genetic algorithm with paralleling discrete ordinate code

An optimization method combining a genetic algorithm and the paralleling discrete ordinate code is developed applied to shielding design of Savannah reactor. Several approaches are studied cope with multi-objective problem, such as transforming into single-objective approach, non-dominated sorting approach adding constraints on sub-objectives. Comparing current methods, in this study shows better performance since calculations free statistical fluctuations. The sub-objectives seems have best performance. More optimized individuals satisfying can be obtained, objectives show improvements comparing initial design.