Safety and Security of Process Plants: A Fuzzy-Based Bayesian Network Approach

Abstract

Chemical process industries handling hazardous chemicals are potential targets for deliberate actions by criminals and disgruntled employees. It is therefore imperative to have efficient risk management programs to identify prospective threats and to develop suitable mitigation strategy. In this study, a probabilistic security risk assessment approach based on dedicated Bayesian networks is presented. Bayesian networks are capable of capturing interdependencies between different factors involved in site security and are found to be very effective in performing analysis even in conditions of uncertainty due to unavailable data by making use of expert opinion. The human subjectivity associated with expert judgment is overcome by incorporating concepts of fuzzy logic. The effectiveness of this approach is demonstrated in an illustrative case study. The site-specific Bayesian network developed for the case study is found to be effective in analyzing the security status of the facility and the performance of security systems installed.