• Probabilistic Assessment
• Major accidents
• Cascading events
• Protection systems

The quantitative risk assessment of domino effect together with the development of dedicated tools and methodological approaches emerged as a critical topic to be addressed within the safety assessment of chemical and process plants. Clear evidence of this need is given by past accident data analysis, which evidences the high destructive potential of domino accidents, and by the high risk perception and growing public concern caused by high-impact low-probability events. The present study was focused on the development of methods for the probabilistic vulnerability assessment of chemical and process facilities with respect to cascading events resulting from both internal and external causes (i.e. domino effects due to failures, NaTech scenarios and intentional attacks). Different type of protection systems and barriers were investigated in order to assess their role and performance in preventing and/or mitigating escalation, privileging quantitative approaches suitable for the implementation in Quantitative Risk Assessment (QRA).
First, the attention was devoted to the analysis of domino effect triggered by fire. The study aimed to the quantification of the performance of protection systems as well as procedural and emergency measures in the prevention of escalation triggered by fire. Starting from the recomposition of performance assessment results, an innovative methodology for the probabilistic assessment of escalation scenarios triggered by fire was developed. Considering the application in the framework of a conventional Quantitative Risk Assessment (QRA), a modified Event Tree Analysis (ETA) was selected to account for the actual performance of the safety barriers in the estimation of domino effect frequency. Next, the analysis was extended to the characterization of the likelihood contributions to security risks, specifically focusing on the aspects determining the attractiveness and the vulnerability of an industrial site. In particular, a method to the vulnerability assessment of industrial plants according to a Probabilistic Risk Analysis (PRA) approach was proposed starting from the functional analysis of Physical Protection Systems. Quantitative data gathered through an expert judgement study on the performance of physical protection systems (PPSs) adopted to secure chemical plant against external attacks were used to quantify a Bayesian Network-based model aimed at supporting the quantitative vulnerability assessment. Moreover, the BN-based model allowed representing the variables, influencing factors and interdependencies that may affect the performance of PPS in the accomplishment of the protection function. The modelling approach could be further extended to the analysis of security-related domino effects.