Accountability measurement precision and accuracy values: How good is good enough.

摘要

The Department of Energy (DOE) Order 5633.3A requires that the desired levels of precision and accuracy be established for accountability measurements, that the magnitude of these uncertainties be minimized for major contributors to the limit of error for inventory differences (LEID), and that methods be selected, validated, and qualified that are capable of providing the desired levels. In an effort to consistently determine the desired precision and accuracy levels for measurements within each of the nuclear material processing facilities at the Savannah River Site (SRS), a series of sensitivity studies were performed. To validate the current uncertainties as the goals, variance-propagated LEID models were used to determine the sensitivity of the LEID to each uncertainty value, using a nominal increase in the LEID as a figure-of-merit. These sensitivity studies provided the threshold values that each uncertainty needs to be held below. Engineering judgment and operational experiences were combined to qualitatively determine the need for improvement for each uncertainty. This paper describes the methodology of the sensitivity study, gives examples of the threshold values generated, and discusses the benefits of this approach in the approval process for proposed method changes.