Authors: Dr P Praks, VSB-Technical University of Ostrava, S Medonos, Petrellus Engineering Ltd. Proceedings of the European Safety and Reliability Conference ESREL 2010.
Statistical data for blowouts from hydrocarbon gas or oil wells are scarce and time-dependent ignition probability is difficult to handle by Event Trees. In this Paper ignition data observed in the industry are statistically analysed to give point estimates and confidence intervals with improved accuracy. The Paper shows that time-dependent probabilities can be simplified without losing accuracy, which improves quantitative risk assessments and their efficiency. The presented method is robust and can be used for a variety of statistical analyses where the number of statistical observations is limited. Numerical examples are presented and confirm high quality of results although the method uses limited number of observed data.
Return to the Publications page.