Authentication of reprocessing plant safeguards data through correlation analysis

This report investigates the feasibility and benefits of two new approaches to the analysis of safeguards data from reprocessing plants. Both approaches involve some level of plant modeling. All models involve some form of mass balance, either applied in the usual way that leads to material balances for individual process vessels at discrete times or applied by accounting for pipe flow rates that leads to material balances for individual process vessels at continuous times. In the first case, material balances are computed after each tank-to-tank transfer. In the second case, material balances can be computed at any desired time. The two approaches can be described as follows. The first approach considers the application of a new multivariate sequential test. The test statistic is a scalar, but the monitored residual is a vector. The second approach considers the application of recent nonlinear time series methods for the purpose of empirically building a model for the expected magnitude of a material balance or other scalar variable. Although the report restricts attention to monitoring scalar time series, the methodology can be extended to vector time series.