MATERIAL DECLARATION VALIDATION USING PROCESS MONITORING FOR A LIQUID TRANSFER FACILITY

This paper presents the development and application of process monitoring techniques to the Oak Ridge National Laboratory (ORNL) mock feed and withdrawal facility for safeguards monitoring. Some safety critical facilities and systems, such as nuclear power plants and aircraft, incorporate process monitoring techniques to detect and identify changes and improve reliability and safety. Several of these techniques can also be applied to international and domestic safeguards. The ORNL mock feed and withdrawal facility is a small scale process laboratory that mimics a general continuous-batch operational facility, such as a uranium enrichment facility. During facility operations, the operator records the initial and ending weight of each material storage container to determine the amount of material introduced or withdrawn from the process. This data is recorded on a material declaration form. The material declaration form serves as the official record of material processing. However, the facility is instrumented with several weight sensors sampled at 1 Hz. This research focused on the development of general categorization algorithms based on the facility instrumentation to identify when and how often material is processed. Additionally, pattern recognition algorithms ensured that the relationships present in the recorded data were similar to historic relationships. These algorithms were used to validate the material declaration sheets and alert the user to any deviations, providing an additional assurance of declared operations. Finally, a graphical user interface was developed to enable the user to quickly analyze the recorded data and investigate any differences between the recorded data and the material declaration sheets. The user would then decide if the differences were due to simple recording errors, abnormal processes, a change in the control scheme, or something of a more serious nature, such as material diversion or undeclared material processing. This paper discusses the steps involved in developing the algorithms for comparing process data with material declaration forms and their performance when evaluating data from the ORNL mock feed and withdrawal facility.