Development of Genetic Algorithm-Based Methodology for Optimizing Material Balance Period Combinations in Near Real Time Accountancy System for Bulk Handling Facility

In bulk handling facilities, the effective implementation of a Near Real Time Accountancy (NRTA) system depends on optimally determining Material Balance Periods (MBPs). Shorter MBPs (i.e., shorter material balance period intervals) enhance detection capabilities but increase inventory measurement frequency, resulting in higher operational burdens and safeguards costs. To address this trade-off, a novel methodology employing a Genetic Algorithm (GA) was developed to derive optimal MBP combinations across various facility designs. Inspired by natural selection, the Genetic Algorithm is a robust optimization tool capable of handling complex, multi-variable problems without requiring prior knowledge of search space continuity, differentiability, or unimodality. This flexibility makes the Genetic Algorithm particularly suitable for optimizing MBP settings in NRTA systems, where multiple factors must be considered. Before applying the Genetic Algorithm for optimization, the objective function and constraints were clearly defined. The objective function minimizes Material Balance Evaluations (MBEs) and inventory measurements while maximizing detection probability, subject to a 95% minimum detection requirement for abrupt diversion scenarios. Through iterative evolution, the GA-based approach finds an optimal balance between operational efficiency and detection performance. Comparative analyses indicate that this method surpasses the manual determination of candidate measurement points, especially under complex conditions. When applied to a hypothetical pyroprocessing facility processing 400 tons annually, the GA-based results matched or exceeded those of manual approaches. Notably, to maintain 95% detection for an 8 kg plutonium abrupt loss within a single MBP, only three MBEs per year were needed. These findings demonstrate the potential of the Genetic Algorithm as an effective tool for NRTA systems, offering a balance between stringent safeguards and practical operations. The adaptability of the Genetic Algorithm to various facility configurations and the ability to handle intricate optimization landscapes suggest broader applications in nuclear material management and other safeguards domains requiring sophisticated solutions.