Study on loss detection algorithms using tank monitoring data

Suzuki, Mitsutoshi; Hori, Masato; Nagaoka, Shinichi ; Kimura, Takashi 

An application of solution monitoring to material balance has been investigated with an actual data of the Solution Monitoring Management System (SMMS) in the Tokai Reprocessing Plant (TRP). Loss detection capabilities of the proposed multivariate statistical methods are examined numerically for a simulated loss corresponding to parameters of a significant quantity in the wait mode. Multi-scale statistical analysis as well as multivariate cumulative sum analysis are successfully used to demonstrate the protracted-loss detection. Since the actual tank data is composed of both records in the wait and the transfer mode, the tank-to-tank transferring data is extracted from the sequence of monitoring data and the effectiveness of error model is evaluated in comparison with calculations. Although the data taken from the SMMS does not describe in the entire solution process, an extended application of solution monitoring to the nuclear material accounting is advanced using the plant data.