Beta Distributed Preferences in the Comparison of Failure Modes

Abstract

The Composition of Probabilistic Preferences (CPP) is employed here to combine evaluations of a set of factors determined by adding human reliability criteria to the traditional FMEA criteria. The risks according to each of nine criteria are evaluated for ten failure modes by a set of seven evaluators. The probabilistic transformation of CPP employs a beta distribution, whose parameters are estimated on the basis of the means and modes of the samples of seven evaluation values. In the case studied, higher discrimination is provided by the use of the sample means.