Nuclear Material Accountancy and Data verification (The MUF-D-Problem)

According to general agreement international nuclear material safeguards is organized in such a way that the plant operators generate all data necessary for the establishment of a material balance, that the inspectors verify the operator's data with the help of independent measurements and that - if there are no significant differences between the operator's data and the inspector's findings - the material balance is established with the help of the operator's data. This procedure implies two tests of significance: one difference test (D-test) for the comparison of the operator's and the inspector's data and one for the material balance establishment (MUFtest) . In this paper these two test procedures as well as their combination, under the boundary condition of a given overall false alarm probability, are analyzed. This bivariate test is compared to the optimum test in the sense of the probability of detection. The results are illustrated with the help of a concrete example.