Dynamic availability assessment of safety critical systems using a dynamic Bayesian network

Availability analysis of safety critical systems is an integral part of ensuring safety both in onshore and offshore process operations. However, the availability assessment of these systems is complex due to their multistate failure scenarios (dormant failure and failure on demand) and multistate functionality (operational failure). In the present study, a dynamic Bayesian network (DBN)-based dynamic availability assessment technique is proposed. This approach offers much flexibility in representing different failure scenarios and the interdependence of failure causes. Sensitivity and importance analyses have also been performed to identify the most influential failure causes. This helps to design better management strategies and provides more realistic reliance on safety critical systems. Applications of the proposed methodology are demonstrated on two safety critical systems: a fire alarm system and a hazard scenario in a steam generation system. This methodology will offer a pivotal step forward in dynamic safety analysis.