Characterization and Propagation of Uncertainties Associated with Holdup Measurements at Rocky Flats

EG&G Rocky Flats, Inc. has an ongoing campaign to measure plutonium holdup in the glovebox exhaust ducts. Up to the present time, the total Pu assay of a duct has been assigned a relative uncertainty of 100%. In practice this has been accomplished by multiplying the integral of the point assays along the duct by a factor of two. While it is generally agreed that this approach is conservative, it is also conceded that the 100% uncertainty is arbitrary, and that this approach provides no basis for deducing realistic confidence limits. A more satisfactory and statistically defensible approach is to determine the uncertainties associated with each point assay and then to propagate these uncertainties as the point assays are integrated for a duct total. This total can then be reported with realistic confidence limits. The difficulty is that the sources of uncertainty in a holdup measurement are numerous and difficult to characterize. This study characterized the significant sources of holdup measurement uncertainty at Rocky Flats, and using standard statistical methods, combined them to estimate the uncertainty for each measurement point. A strategy for propagating the point assay uncertainties into an overall uncertainty for the total duct holdup was also developed.