The impact of unmeasured process inventories on near real time accountancy

Sequential statistical decision tests for detecting clandestine removal of nuclear material from a running reprocessing plant using material accountancy are investigated. Part of the process material is assumed to be unmeasurable and the moments (mean and variance) of the unmeasured inventory are estimated with the same set of sequential material balance data as is used for the decision test. For uncorrelated, normally distributed process fluctuations detection times comparable to those obtained with a priori known mean and variance of the unmeasured inventory are obtained. (orig.)Available from TIB Hannover: RA 831(2772) / FIZ - Fachinformationszzentrum Karlsruhe / TIB - Technische InformationsbibliothekSIGLEDEGerman