This paper describes how past pipeline incidents can be used to test whether the consequences of major hazard events as predicted by computer model are realistic.
Most materials transported by cross-country pipeline are flammable in nature, e.g. crude oil, petroleum products and natural gas. The hazardous consequences of a loss of containment may be modelled by a variety of computer models. One of the key inputs is how the failure may be modelled, in terms of the initial source terms and how the released material behaves subsequently. Modelling appropriate behaviour of the release is incumbent on the modeller / engineer, as well as accurately interpreting the output from the computer model.
There have been a number of high profile pipeline incidents in recent years that have had a devastating effect on the local community. Although one recognises the distressing effects of such incidents, these also provide an opportunity to test the consequences predicted by computer models. One of the key questions is whether it is likely that the adverse effects of an incident are overpredicted by the modelling inputs / technique and therefore whether the outputs from the model present a conservative thermal radiation dose.
This paper presents such a benchmarking exercise, which has been carried out to assess the degree of realism provided by computer modelling and the way in which the modelling is carried out. This exercise was conducted following a number of pipeline risk assessments where it was predicted that in some cases, hundreds of fatalities may occur following rupture of a pipeline transporting natural gas. It was felt that there may be a level of over-conservatism in the modelling, particularly as many incidents that have occurred have not resulted in the predicted level of fatal injuries.