Assessment of alternatives to correct inventory difference statistical treatment deficiencies

Abstract

This document presents an analysis of alternatives to correct deficiencies in the statistical treatment of inventory differences in the NRC guidance documents and licensee practice. Pacific Northwest Laboratory•s objective for this study was to assess alternatives developed by the NRC and a panel of safeguards statistical experts. Criteria were developed for the evaluation and the assessment was made considering the criteria. The results of this assessment are PNL recommendations, which are intended to provide NRC decision makers with a logical and statistically sound basis for correcting the deficiencies.; ; i ACKNOWLEDGMENTSThe authors wish to acknowledge their indebtedness and gratitude to the me.mbers of the panel~ who assisted in the development of the alternatives a~.sessed in this study, provided an initial assessment of them, and also SE!rved as peer reviewers of the report:Dr. Robert Easterling In November of 1980, a paper was prepared by several Nuclear Regulatory Commission (NRC) staff members on the validity of the statistical basis for treating inventory differences {ID) in licensee facilities (SECY-B0-514, NRC 19BO). The paper described 13 statistical deficiencies in current NRC guidance and in licensee practices in the areas of ID evaluation and modeling, determination of 10 variability, and statistical terminology. The paper also described possible changes that could be made to remedy the deficiencies. The purpose of this study is to develop additional alternatives and perform an assessment of the safeguards statistical implications of the alternatives and the costs to licensees in the event of their implementation. This assessment was performed for eleven of the deficiencies .. The other two are being ~.tudied by other NRC projects.The assessment was performed in the following steps:• A 10-member panel of experts in the development and application of statistical methods to the problems of nuclear material safeguards was convened to help develop the alternatives for correcting each statistical deficiency. Two statistical experts in safeguards from Pacific Northwest Laboratory {PNL) were members of the panel.• A report of the results of the panel discussion (Appendix A) was prepared by PNL staff members.• An assessment of the proposed alternatives for each deficiency was made by each panel member with respect to five criteria related to the objectives of nuclear safeguards.• The panel assessment, combined with the notes from the panel discussions, was used to assist the PNL staff in assessing the alternatives.• Finally, implementation and operating costs to a typical licensee for each alternative were estimated by PNL staff members.Remedying the statistical deficiencies identified in the NRC staff paper will require coordinated actions on three front.s. PNL 1 s conclusions and recommendations are structured in terms of these three fundamental areas for improvement, listed below in order of importance:1. Proper modeling and analysis of inventory differences 2. Effective implementation of st...

Statistical Methods for Nuclear Material Management

“…Var(x) = ^2 2,0,2 + Sr(«rVbr) (Sj(r)Mf(r)) (14.32) Equation 14.32 is valid only under the assumption of identical distributions for the effects j/j(r) within each source of measurement variability with variances given by Equation 14.29. Assuming, as in the previous section, that E(|3') = 2 E(|3j') = 0 and that E(j7j(r)) = 0 for all r, it follows from Equation 14.31 that E(x) = /*• These assumptions are consistent with the recommendation in a study of deficiencies in the statistical analysis of inventory differences (Byers et al, 1983) that insofar as the measurement process is concerned, "an attempt should be made to estimate and correct for all potentially significant biases. To the extent that this is accomplished, the estimator of inventory difference will be free of 'fixed' errors or biases.…”