PROPAGATION OF VARIANCE SOFTWARE FOR NUCLEAR MATERIAL ACCOUNTING

This is a discussion of desired and required features of software for doing propagation of variance (POV) in the context of estimating the uncertainty sID of a nuclear materials inventory or inventory difference (ID). The following features are included, for example, in one prototype POV tool at Los Alamos National Laboratory: (1) Estimate the random and systematic error standard deviations (sR and sS) of measurements for each measurement category. Once the number of measurement categories is chosen, the software must recognize the measurement category for each measured value and then use appropriate sR and sS values; (2) In cases where differences between inventory materials and calibration standards affect measured values, “itemspecific” systematic errors should be included in the measurement error model, leading to a requirement to have an alternate assay method in order to generate “working standards”; (3) Combine error variance estimates from each measurement category into sID estimates (via POV) to enable inventory difference evaluation; (4) Allow flexible POV analysis of sub-MBA (material balance area) process units; (5) Allow flexible material balance periods, and (6) Partition sID to identify the major error sources. In particular, those measurement categories that contribute most to sID, become highest priority for developing improved assay methods.