Development of Quantitative Risk Assessment Methodology for High-risk Scenarios and Accidents

Abstract

Risk modeling and estimation in the process industries is very often carried out by the application of the event tree method. However, the significant drawback of event trees is that it is a steady state method, thus most suitable for modeling of static processes. As the time of occurrence of events is an important factor in most industrial processes, steady state methods may not correctly identify the most important contributions to the risk. For this reason, innovative methods capable of representing dynamic processes should be preferred. These methods are the main focus of this thesis.

In this work, multiple risk assessment methods are presented and discussed with emphasis on dynamic ones. Two accidents were used for model construction - hydrocarbon leak ignition on an offshore production platform and runway incursion scenario. Both accidents are thoroughly described. Using selected risk assessment methods, mathematical models of these accidents are presented and used for estimation of the risk of possible negative consequences, such as fatality of personnel due to fire occurrence on an offshore platform, aircraft collision during takeoff, etc. Comparisons between steady state and dynamic models were made, mainly focused on the Petri Nets approach.

Following methods (and their extensions) were analyzed, applied and confronted in this thesis: -Event Tree -Fault Tree -Monte Carlo -Stochastic Petri Nets