Dynamic Risk Analysis to Improve Process Safety Using Bayesian Network

The chemical process industries often suffer from abnormal events of varying magnitudes, which may lead to different consequences, including incipient faults, near-miss, incidents, and accidents. Typically, when an abnormal event occurs, various safety systems, such as alarm systems and safety instrumented system, come into play to prevent the event from propagating. To improve process safety, estimation of the failure probabilities of safety systems and probabilities of corresponding consequences is extremely important and should receive more attention. Risk analysis techniques such as event tree have been used popularly for this purpose. However they suffer limitations of static structures and uncertainty handling, which are of great significance in safety system analyses. In this paper, Bayesian network (BN) for dynamic risk analysis is proposed. Bayesian networks are a probabilistic approach to model and represent influences between safety systems. First, a mapping algorithm which translates an event tree to a Bayesian network is introduced. In BN model, causal relationships reflect the dependencies among safety systems which are represented as nodes of BN. After building the BN causal structure, we estimate the parameters of the network, namely the failure probabilities of safety systems and probabilities of corresponding consequences, using accident precursor data. Then reasoning under uncertainty in BN is performed, including predictive and updating analysis. In addition, other advantages of BN are discussed. First, with incomplete accident precursor data, BN can handle to estimate parameters effectively. Second, when modify the structure of BN because of functions required, it is easy to extend the model. Finally dynamic risk analysis using Bayesian networks is applied to a chemical process system with four safety systems to show its performance.