Risk assessment of process system considering dependencies

Abstract

Risk assessment is conducted in process systems to identify potential accident scenarios and estimate their likelihood and associated consequences. The bow-tie (BT) technique is most frequently used to conduct the risk assessment. It is a simple, comprehensive and straightforward technique; however, it considers independence among the causation factors (initiating events) of an accident scenario and the safety barriers in place to minimize the impact of the accident scenario. This is a serious limitation and can lead to erroneous results. This paper presents a simple yet robust approach to revise the Bow-tie technique considering interdependence. It employs copula functions to model the joint probability distributions of causations in the BT model of the accident scenario. This paper also analyzes the impact of dependence on two common logic gates used to represent the potential accident scenario. The probability of a potential accident scenario in a hexane distillation unit using both the traditional BT technique and the revised approach is compared. Results confirm that the revised approach is reliable and robust.