ABSTRACT
This chapter shows how Bayesian network (BN) can be applied to model and simulate several cases, including scenarios when critical systems can fail coincidentally. It presents a case study for hazard assessment of a Floating Roof tank using BN. The risk profile of the tank at the pre-accident situation of IOC Jaipur fire is described. This is done to see the predictability of the model. The Jaipur tank farm accident was chosen because details of the investigation committee report were available. By changing the parameters of the nodes of BN suitably, the precursor status at the Jaipur tank farm can be represented to a certain measure. Simulation of the various situations, including the coincidental failures at the tank farm, brings out the high probability of an accident at the site, as compared with published failure data. Such simulations are normally not done during traditional QRA.
