Development of a Statistical Model For Planning Random Interim Inspections Under Integrated Safeguards

The advent of integrated safeguards has accentuated the need for, and importance of, random inspections for contributing to the achievement of the IAEA’s safeguards objectives. This in turn has posed challenges in the area of statistical approaches to planning and evaluation of random inspections. The number of such inspections needed to achieve target detection probabilities, with the limitations of available verification methods, is usually of particular interest. Random interim inspections (RIIs) have now become a key feature of integrated safeguards approaches applied by the Agency, and are gradually replacing conventional interim inventory verifications (IIVs). Like IIVs, RIIs can serve various purposes including the achievement of timeliness goals and verification of inventory changes. Because of their randomness and short notice characteristics, RIIs may also be used to cover borrowing and misuse of facilities. These characteristics distinguish RIIs as effective and efficient means to contribute to achieving the Agency’s safeguards objectives. Implementation of integrated safeguards particularly gives consideration to the broader use of statistical techniques in optimizing effectiveness and efficiency of inspections. To this end, this paper describes a comprehensive statistical model for estimating the number of RIIs required to achieve a target level of detection probability for timeliness purpose. The model takes into consideration the availability of other safeguards measures at a nuclear material location, with the aim of assigning credit to them in terms of deterring or detecting the diversion of nuclear material, and thereby optimizing the required number of RIIs. Other factors included in the model are the timeliness goal, verification level achieved during the RIIs, the probability that a RII will take place within the timeliness period and the randomization scheme. The randomization scheme is a concept of grouping a material balance period into blocks of consecutive days in order to enhance the probability that a RII will take place within the timeliness period. The model can be applied to a single material balance area (MBA), a group of MBAs (as in sector or site approaches) or to a State-level integrated safeguards approach. The optimality of the randomization scheme is studied in relation to the timeliness goal. In addition, the relationships between the various factors of the model and development of a method of distributing the RIIs over sectors or MBAs, when required, are discussed. Numerical examples are given to demonstrate the use and versatility of the model for planning RIIs.