ABSTRACT
An integrated model for risk in a real-time environment for the hydrocarbon industry based on the CATS model for commercial aviation safety has been further developed. The approach described in earlier papers required Bayesian Belief Nets (BBN) to be developed for each process unit separately. A much more efficient method for developing the continuous BBNs is based on standard building blocks consisting of nodes for a piece of equipment and nodes for potential safety features. Elements of the BBN can be switched on or off as required. From the equipment inventory of a chemical plant a complete BBN can be built automatically. Humans are represented by multiple instances of the same BBN structure, but with different parameters. The resulting computer program calculates the frequency of Loss of Containment Accidents for an existing or a proposed plant including the effects of human behaviour and intervention, including rigorous handling of the uncertainties.
