Year
2006
Abstract
Nuclear material accounting (NMA) includes periodic evaluation of material balances (MB) as a check for whether material is missing. In the case of large-throughput facilities, the MB measurement uncertainty will typically be too large to meet statistical loss detection goals. Nonetheless, many believe that large facilities can be effectively safeguarded with an appropriate combination of containment/surveillance (C/S) measures and MB accounting, provided that the most likely diversion paths are properly considered. Solution monitoring (SM) is a form of process monitoring, which can be considered to provide a C/S capability. Experience to date has shown that SM is a challenging but useful safeguards measure that contributes to both NMA and C/S. Part of the challenge involves choosing effective evaluations of SM data that avoid data indigestion, enable anomaly detection and/or resolution, and also do not burden the operator or the inspector with too many investigations. This paper illustrates potential benefits of SM, including abrupt loss detection while controlling for multiple tests, anomaly resolution, measurement error model validation, and data authentication. Data authentication results from the many internal consistency checks that arise when relating level, density, and temperature readings; this makes it very difficult to alter data without being detected. We also briefly describe multivariate time series analysis methods, arguing for good implementation of basic methods in wellwritten custom software as the appropriate starting point.