Optimization of a dry, mixed nuclear fuel storage array for nuclear criticality safety

A dry storage array of used nuclear fuel at an U.S. facility contains a mixture of more than 20 different types of research and test reactor fuels. Neutronic interaction occur between fuel canisters within this large storage arrangement of 636 fuel storage canisters. Nuclear criticality safety is maintained primarily by moderation control and through (spatial) control of neutron interaction and leakage provided by the established geometry. Over 40+ years of operation the collected fuels are now arranged in a configuration that does not necessarily minimize the system neutron multiplication factor (k_eff). This creates a concern since the entire remaining facility storage capacity cannot be utilized without exceeding the minimum subcritical (safe) state (ksafe) for ensuring nuclear criticality safety. This research determined a more optimal arrangement for the stored fuel canisters with a goal to not only minimize the system keff, but to also minimize the potential fuel canister relocation movements. The ultimate goal is to establish a solution for this multiple-objective optimization problem by finding the Pareto Front (PF) for maximum keff reduction versus the number of canister relocations. The process required analyses of the array by MCNP to gain a greater understanding of the interaction characteristics of the disparate fuel types. Some stored fuels are composed of primarily metallic and oxide constituents that provide very little inherent moderation, while other fuel types contain graphite or zirconium-hydride which can contribute a significant amount of moderation. These factors cause a complex relationship within a fuel arrangement with respect to thermal utilization and axial leakage probability. The new MCNP analysis was utilized to develop a simplified spreadsheet-based empirical model which simulates the array's keff result. This empirical model was benchmarked against additional MCNP calculations for various arrangements and then employed in the optimization analysis. A Tabu metaheuristic and stochastic approximation optimization was applied to the empirical model with an optimization function to minimize the array's keff. The optimization algorithm interrogated the empirical model to determine the optimal rearrangement for up to 60 canister movements. The PF was determined, which shows that a limited number of canister relocation movements could allow the existing 31 empty array positions to be filled with common research reactor fuels while also yielding a reduction in the overall array keff, relative to the current arrangement. Result will improve facility utilization while enhancing the safety margin. Lessons learnt with respect to potential mixed fuel type assemblies will described. (author)