Validation of the GAMMA-PC methodology for dry cask loading optimization

Abstract

This paper evaluates the performance of a new greedy multiobjective memetic algorithm, GAMMA-PC, that was developed to find dry cask loading configurations with fewer casks, lower heat loads, and earlier transportation schedules. GAMMA-PC was applied to the dry cask loading problem for Zion Nuclear Power Station. This site represents a unique opportunity to validate algorithmic performance as its fuel has been completely transferred to dry storage using a high-quality loading strategy. The optimization of Zion was performed here using the same transfer timeframe as the vendor in charge of decommissioning and using an extended ten-year transfer timeline. The results of these scenarios showed that GAMMA-PC produced comparable solutions to the real Zion strategy. In the first scenario, GAMMA-PC produced a solution that dominated the real loading configuration, achieving a lower average cask initial heat load. Under the extended timeline, the real loading configuration dominated the GAMMA-PC solution, which had a higher average initial heat load. The values for the other objectives showed no difference in quality. The analysis of the individual cask characteristics also highlighted the fact that the dry cask loading paradigm does not directly address balance among the casks. Future work might incorporate this as an objective. This validation has shown that while the algorithm and problem objectives have some room for improvement, GAMMA-PC performs well and is a promising tool for the optimization of the dry cask loading problem.