Petri-net based safety analysis of process systems

Safety analysis is very important to prevent events that may cause catastrophic accidents in process industries. Although traditional safety assessment methods including fault tree (FT) have been used widely in this field, they still lack in dynamic structure representation and uncertainty processing, which are important for the process safety analysis. In this study, a safety analysis approach based on the probabilistic Petri-net (PPN), which extends the basic Petri-net, is proposed. Probabilities are assigned to the tokens in the Petri-net, and the probability dependencies of events can then be modeled by a PPN. The enabling rule of a transition and the execution rule of an enabled transition are provided, through which the dynamic evolution process of the system can be revealed by the flowing of tokens, and the probabilities corresponding to the occurrence of events can be updated. A case that vapor cloud may overflow from a tank is studied to illustrate the proposed approach. The result is validated by comparing with that of the fault tree analysis.