The FORM/SORM method (alternatively referred to here as the most likely failure point method) has been used to perform uncertainty analysis for a range of hazard consequence calculations, namely: chlorine release (toxic dose); BLEVE fireball (thermal dose); BLEVE blast (overpressure) and VCE blast (overpressure). Realistic semi-empirical models, such as those which would be employed in a practical hazard/risk analysis, were used in all cases. The method was used to calculate the probability Pf that the estimate of hazard consequence, calcu-lated on the basis of best-estimate data to be below an appropriate safety limit, would actually exceed that limit when data uncertainties are taken into account. In each case, the method produced its estimate of Pf within roughly 10N-20N evaluations of the con-sequence model, where N is the number of input data items represented as uncertain. Values of Pf were within 10%-20% of values obtained using a benchmark Monte-Carlo calculation, which was much less efficient. Numerical experiments were carried out to investigate the accuracy of the sensitivity estimates pro-duced by the method, ie the sensitivity of Pf to a change in median value of each variable, the sensitivity of Pf to a change in standard deviation of each variable and the effect on Pf of replacing uncertain vari-ables by fixed values. The estimates were shown to be accurate within a factor of about 2. This was concluded to be more than adequate for the purposes of screening out unimportant variables and of identifying areas where calculated values of Pf (which in many cases can be identified with risk) could be reduced by improved data accuracy. A preliminary investigation was carried out into the representation of quantities used to characterise weather conditions (wind speed, wind direction and Pasquill stability category) by continuous distribu-tions rather than the discrete distributions of conventional quantified risk analysis. For the chlorine re-lease calculations, the method correctly identified that calm conditions, though relatively rare, can strongly influence risk because of the high doses they give rise to. This report and the work it describes were funded by the Health and Safety Executive. Its contents, including any opinions and/or conclusions expressed, are those of the authors alone and do not neces-sarily reflect HSE policy.
25 декабря 2013
09 января 2014