Application of dynamic Bayesian network to performance assessment of fire protection systems during domino effects

Abstract

The propagation of fire in chemical plants – also known as fire domino effects - largely depends on the performance of add-on passive and active protection systems such as sprinkler systems, water deluge systems, emergency shut down and emergency blow down systems, fireproofing, and emergency response. Although such safety barriers are widely employed to prevent or delay the initiation or escalation of fire domino effects, their inclusion in the modeling and risk assessment of fire domino effects has hardly been taken into account. In the present study, the dynamic evolution of fire protection systems has been investigated qualitatively using event tree analysis. To quantify the temporal changes and their impact on the escalation of fire domino effects, a dynamic Bayesian network methodology has been developed. The application of the methodology has been demonstrated using an illustrative case study, considering a variety of fire scenarios, target installations, and firefighting systems.