A statistical analysis in the BWR fuel lattice design

• A statistical analysis in the neutronic BWR fuel lattice design is shown.
• The analysis was carried out using a sample space of 2700 fuel lattices.
• Several heuristic rules and three meta-heuristics algorithms for sample space were used.
• In fuel design statistically it is convenient to choose an initial distribution for materials.

In this work, a statistical analysis in the neutronic fuel lattice design process is shown. It presents a study where different ways to choose rods with uranium and uranium-gadolinia are considered, and their allocation into the fuel lattice was analyzed. The analysis was carried out studying three different frequency distributions for fuel materials, with a fixed uranium stock, a fixed gadolinia stock, as well as several average uranium enrichment and gadolinia concentrations. Three heuristic techniques to design fuel lattices with the above specifications: ant colony systems (ACSs), a Recurrent Neural Network (NN) and a hybrid with scatter search and path relinking (SS+PR) were used. These heuristic techniques minimized the Local Power Peaking Factor while the infinite multiplication factor was kept in a reactivity interval. Both nuclear parameters were calculated with CASMO-4 code. The results show that statistically it is important to choose an appropriate frequency distribution for fuel materials, and taking some rules into account to allocate both uranium and gadolinia rods, in order to obtain suitable fuel lattice designs.