Dynamic probabilistic risk assessment using event sequence diagrams

Dynamic reliability/PRA is the probabilistic study of man-machine-software systems affected by an underlying physical process. Dynamic reliability/Probabilistic Risk Assessment methodologies have become fairly established in academia and their theoretical superiority over the classical Event Tree/Fault Tree approach to risk assessment has been demonstrated. Despite this, dynamic PRA methods have not enjoyed the support of the industry. One of the major reasons for this lack of support is the absence of a framework for qualitatively representing dynamic accident scenarios.; This dissertation provides a framework for qualitatively representing dynamic accident scenarios using Event Sequence Diagrams (ESD). The concept of Event Sequence Diagrams is extended to capture dynamic scenarios. The mathematical equations governing the various building blocks of the Event Sequence Diagram framework are provided. From these equations it can be seen that ESDs can be studied using the theory of semi-Markov processes.; All previous attempts to analytically solve dynamic reliability problems were involved with solving the Chapman-Kolmogorov system of hyperbolic partial differential equations and this could be done for only very small systems. The equations governing the ESD framework are such that they reduce the solution of dynamic reliability problems to the evaluation of a series of multidimensional integrals, thus facilitating analytical solutions by reducing the dimensionality of the problem. One of the drawbacks of the ESD approach, compared to full fledged dynamic methodologies, would arise from the analyst dependent nature of the approach. Indeed, since the scenarios are constructed by an analyst, it is possible that some scenarios would be neglected due to the use of simplifying assumptions. An approach to identify missing scenarios by combining ESDs with probabilistic dynamics is also given in this dissertation.