Dependability Analysis Using Temporal Fault Trees and Monte Carlo Simulation

Abstract

The safety and reliability of high-consequence systems is an issue of utmost importance to engineers because such systems can have catastrophic effects if they fail. Fault Tree Analysis (FTA) is a well-known probabilistic technique for assessing the reliability of safety-critical systems. Standard FTA approaches are primarily static analysis techniques and as such cannot effectively model systems with dynamic behaviours, such as those with standby components or multiple modes of operation. There have been several efforts to address this limitation, one of which is Pandora, a temporal fault tree approach. Pandora uses three temporal gates—Priority-AND, Simultaneous-AND, and Priority-OR—to model the effects of sequences of events. Hitherto, Pandora was unable to perform a holistic evaluation of a full system that is repairable, taking account of useful system operating environment variables (such as time of operation, flow rate, etc.) or system data such as repair state and preventive maintenance. This paper aims to address these limitations. Algorithms to evaluate different system configurations have been generated and techniques for modelling and analyzing different system data in a simulation platform have been proposed. This paper extends the capabilities of Pandora so that it is capable of analyzing a modern system that features different failure modes, has diverse component failure distributions, considers the system’s operation environment data, and models different system configurations. The outcome of such analysis enables analysts to understand the operation and dynamics of a system holistically and aids in the implementation of appropriate risk mitigating strategies.