Procedures for Using Expert Judgment in Risk Analysis: Fid-Bau Portal

Procedures for Using Expert Judgment in Risk Analysis

Cooke, R. M.

Abstract This chapter describes step by step procedures for the development of expert judgment data for probabilistic risk analysis. Many models for combining expert judgments can be found in the literature (for a survey see Cooke, 1991). However, consistent with recommendations to the Dutch Ministery of Housing, Physical Planning and Environment, and to the European Space Agency, only the so called Classical Model is recommended for implementation at this time. The development of expert judgment data is broken into nine phases, which are addressed in the corresponding sections of this chapter: 1. The problem identification phase 2. The expert identification phase 3. The expert choice phase 4. The question formulation phase 5. The calibration variable selection phase 6. The elicitation phase 7. The combination phase 8. The discrepancy analysis / feedback phase 9. The documentation phase Broadly speaking, a probabilistic risk assessment can be broken into two parts. One part, which may be denoted accident prediction, concerns the assessment of the occurrence rates of undesired events (sometimes called Top Events). The dominant methodology in this phase is fault tree analysis, and the input data typically concerns occurrence rates of basic events. Beyond the fault tree itself, the physical modelling in this part is generally confined to the determination of life distributions for components.