A Methodology for Analyzing and Updating Detection Element Probablitites from Performance Testing Data

Facilities that perform vulnerability analyses are required to input probabilities of detection for various elements of its safeguards and security system. In addition, DOE 5630.16, Safeguards and Security Performance Test Program, requires the performance test plan to include the description of any statistical models or mathematical formulas to determine probabilities and confidence intervals as well as methods to be used for data analyses. Performance testing data can be evaluated using Bayesian statistical methods. The results can be used to update and revise detection probabilities as well as the confidence intervals for the detection elements. This paper presents a methodology that involves using professional judgement to define the average value and the bounds of a detection probability. Then, a series of performance tests are conducted, and the results applied to generate a new probability distribution. Examples of the application of this method are shown. A general procedure is then presented that allows the methodology to be applied at any facility. A potential extension of this methodology to results of inspections and audits that use performance testing during their review is also discussed.