Time Series Analysis of Shipper Receiver Differences

In safeguarding reprocessing plants the inspectorate is faced with the problem of analyzing data generated by shipper receiver difference (SRD) at the headend. This paper suggests an approach to the analysis of such data employing time series analysis techniques. An SRD sequence of a facility might contain deterministic trends which reflect on the particular way of operation and the measuring possibilities of SRDs. These trends can, at least in theory, be largely accounted for by analyzing the facility's mode of operation. What is left in the SRD sequence after removing the deterministic trends is a noise-like random fluctuation process, which nevertheless can be modeled as an autoregressiveintegrated moving average (ARIMA) process. Modeling an ARIMA process means that its driving white noise process is to be found. Based on the estimated model parameters including the underlying white noise parameter, an associated process, the socalled standardized innovation process can be constructed. This process then satisfies the requirements of control chart and cumulative sum techniques, which we propose to apply for the SRD safeguards evaluation problem. Two SRD sequences from two reprocessing plants are taken as examples. For both cases the same class of ARIMA process provides the best fit, the process parameters, however, are different. The process parameters appear to be characteristic of a particular reprocessing plant.