Structured Approach to SNM Anomalies Resolution

Nuclear facilities are responsible for resolving nuclear materials safeguards anomalies indicating that special nuclear material (SNM) may be missing. These anomalies can have many causes, including simple bookkeeping or measurement errors or malevolent acts by insiders. Determining the likely cause of an anomaly is often complex, time consuming, and costly. In most cases no specific cause can be conclusively proven. Therefore, it is particularly important that investigations of SNM safeguards anomalies be carried out systematically and that all conclusions from the investigation be thoroughly documented and logically justified. Currently there are no standardized methods available for guiding an investigation and drawing conclusions from it. This paper presents a method we are developing which is based on decision analysis techniques and Bayesian probability analysis. The method helps in developing a set of hypotheses about the possible causes of an anomaly and uses the evidence gathered during an investigation to evaluate the likelihood of each hypothesis. The uncertainty in measurements and evidence collection are accounted for in an explicit and quantitative manner. The method suggests the most effective sequence of steps for the investigation and provides guidance on when to terminate the investigation. The method applies to both item and bulk inventory differences, but the evaluation of bulk differences requires a more complex set of procedures than those presented in this report. A simple computer program and training materials are also being developed. The software is being designed to facilitate quick resolution of an anomaly, help guide an investigation team, and document the team's conclusions.