Modelling high consequence, low probability scenarios: Fid-Bau Portal

Modelling high consequence, low probability scenarios

Cleaver, R.P. / Halford, A.R. / Humphreys, C.E.

By their very nature, major accidents in the gas and oil or petrochemical industries have a low frequency but potentially high consequences. However, assigning quantitative measures to the risks of such an accident is difficult, as the frequency has a high degree of uncertainty attached to it and the consequences may be sensitive to the assumptions made in representing the scenario within the analysis. It follows that the assessment of such accidents requires a robust methodology to be used. There may be cases in which a worst-case or single representative analysis may distort the analysis compared with a full evaluation. Advantica has developed a methodology that allows the different realisations of scenarios to be analysed. taking into account relevant parameters such as release and wind directions. This paper provides details about the basis of this method. It is shown that the calculations allow FN curves to be built up from the frequency and number of casualties for each realisation. This allows the contributions from different realisations to be analysed and those contributing most to be identified. It is also noted that using this method may reveal that some preconceived ideas about what is the 'worst case' are not always correct. As well as being of use when carrying out qualitative risk assessments for Safety Reports for example, it is noted that these methods are of great help in cost benefit analysis. Using worst-case assumptions can often mask the benefits of protective measures, as they appear to have little effect on the worst-case realisation. However, using the methodology allows a correct identification of suitable measures to be taken to reduce risks, as the impact of the measure on all the different realisations is assessed.