Uncertainty propagation and importance measure assessment

Abstract

The authors investigate the effects that different representations of epistemic uncertainty have on practical risk assessment problems. Two different application problems are considered:

• the estimation of component importance measures in the presence of epistemic uncertainties;
• the propagation of uncertainties through a risk flooding model.

The focus is on the epistemic uncertainty affecting the parameters of the models that describe the components’ failures due to incomplete knowledge of their values. This epistemic uncertainty is represented using probability distributions when sufficient data is available for statistical analysis,and by possibility distributions when the information available to define the parameters’ values comes from experts, in the form of imprecise quantitative statements or judgments.

Three case studies of increasing complexity are presented:

• a pedagogical example of importance measure assessment on a three-component system from the literature;
• assessment of importance measures for the auxiliary feed water system of a nuclear pressurized water reactor;
• an application in environmental modelling, with an analysis of uncertainty propagation in a hydraulic model for the risk-based design of a flood protection dike.