SAMPLAN — A new tool to support effective SG sampling plans

Nuclear safeguards experts use advanced statistical algorithms to calculate verification sampling plans. Sampling plans must meet or exceed safeguards performance targets with minimal use of safeguards resources. For example, a plan will call for a certain number of resource-sparing non-destructive assay (NDA) verifications and high-precision destructive analysis (DA) samples in order to achieve an overall detection probability with minimal “cost” of inspection time, shipping fees, laboratory burden, facility disruption, etc. In recent years, experts in safeguards statistical and probabilistic methodologies have proposed algorithmic advances to further improve sampling plans’ efficiency while improving field usability, adapting to revised safeguards objectives, and maintaining a high level of effectiveness. These algorithms are implemented in the SAMPLAN software toolkit. Today, SAMPLAN is notable for both its current capabilities and its development strategy: evolutionary prototyping is used to rapidly refine algorithms, user interfaces, test cases, etc., which can then be implemented in production software for long-term use. This paper documents the prototyping strategy and SAMPLAN’s resulting capabilities