PAPERBACK
\$21.00

• Appendix C: Glossary 77-80

The approach taken in this review is to highlight the PRA method of quantification, comment on applying PRA to weapon performance assessment, discuss possible links and differences between QMU as currently used and PRA, and to identify possible PRA enhancements of QMU. The QMU approach itself is covered elsewhere in this report.

THE PRA APPROACH TO QUANTIFICATION

The PRA approach highlighted is based on the framework of the triplet definition of risk (Kaplan and Garrick, 1981):

where R denotes the risk attendant on the system or activity of interest. On the right, Si denotes the ith risk scenario (a description of something that can go wrong). Li denotes the likelihood of that scenario happening and Xi denotes the consequences of that scenario if it does happen. The angle brackets < > enclose the risk triplets, the curly brackets { } are mathspeak for “the set of,” and the subscript c denotes “complete,” meaning that all of the scenarios, or at least all of the important ones, must be included in the set. The body of methods used to identify the scenarios (Si) constitutes the “Theory of Scenario Structuring.” Quantifying the Li and the Xi is based on the available evidence using Bayes’ theorem, illustrated later.

In accordance with this set of triplets definition of risk, the actual quantification of risk consists of answering the following three questions:

1. What can go wrong? (Si)

2. How likely is that to happen? (Li)

3. What are the consequences if it does happen? (Xi )

The first question is answered by describing a structured, organized, and complete set of possible risk scenarios. As above, we denote these scenarios by Si. The second question requires us to calculate the “likelihoods,” Li, of each of the scenarios, Si. Each such likelihood, Li, is expressed as a “frequency,” a “probability,” or a “probability of frequency” curve (more about this later).

The third question is answered by describing the “damage states” or “end states” (denoted Xi ) resulting from these risk scenarios. These damage states are also, in general, uncertain. Therefore these uncertainties must also be quantified, as part of the quantitative risk assessment process. Indeed, it is part of the quantitative risk assessment philoso-

The National Academies | 500 Fifth St. N.W. | Washington, D.C. 20001