STATISTICAL METHODS FOR DETECTING LOSS OF MATERIALS USING NRTA DATA

We selected eight sequential statistical tests and studied their performance on near-real-time-accounting (NRTA) data that is nominally what is expected from the proposed Rokkasho Reprocessing Plant in Japan. The effort divided into three main activities: (1) use process-flow information to determine process vessel inventories and transfers at the time of material balance closure, (2) use variance propagation methods to estimate the variance-covariance matrix of a sequence of material balances, and (3) study the performance of eight sequential tests on a variety of loss scenarios. This paper describes the results of these three activities