A practical approach for evaluating the strength of knowledge supporting risk assessment models. (April 2020)

• Record Type: Journal Article
• Authors: Bani-Mustafa, Tasneem; Zeng, Zhiguo; Zio, Enrico; Vasseur, Dominique
• Is Part Of: Safety science. Volume 124(2020)
• Publication Date: 2020-04
• Subjects: Strength of Knowledge (SoK) -- Probabilistic Risk Assessment (PRA) -- Risk-Informed Decision Making (RIDM) -- Multi-Hazards Risk Aggregation (MHRA) -- Event Tree (ET) -- Nuclear Power Plant (NPP); Industrial accidents -- Periodicals; Accident Prevention -- Periodicals; Safety -- Periodicals; Travail -- Accidents -- Périodiques; 363.11
• DOI: 10.1016/j.ssci.2019.104596
• Languages: English
• ISSNs: 0925-7535

Abstract

Highlights: The importance of communicating the strength of knowledge for RIDM is demonstrated. A framework is developed to assess the strength of knowledge of PRA models. A reduced-order model is constructed to limit the complexity of the implementation. An application to a case study is provided to show the feasibility of the framework. Abstract: In this paper, we develop a new quantitative method to assess the Strength of Knowledge (SoK) of a risk assessment. A hierarchical framework is first developed to conceptually represent the SoK in terms of three attributes (assumptions, data, phenomenological understanding), which are further broken down in sub-attributes and "leaf" attributes to facilitate their assessment in practice. The hierarchical framework, is, then, quantified in a top-down, bottom-up fashion for assessing the SoK. In the top-down phase, a reduced-order risk model is constructed to limit the complexity and number of basic elements considered in the SoK assessment. In the bottom-up phase, the SoK of each basic element in the reduced-order risk model is assessed based on predefined scoring guidelines and, then, aggregated using a weighted average of "leaf" attributes, where the weights are determined based on the Analytical Hierarchical Process (AHP). The strength of knowledge of the basic events is in turn, aggregated using a weighted average to obtain the SoK for the whole risk assessment model. The developed methods are applied to a real-world case study, where the SoK of the Probabilistic Risk Assessment (PRA) models of a Nuclear Power Plants (NPP) is assessed for two hazards groups, i.e., external flooding and internal events.