Dependency modelling using fault-tree and cause-consequence analysis
The technique of fault tree analysis is commonly used to assess the probability of failure of industrial systems. During the analysis of the fault tree the component failures are assumed to occur independently. When this condition is not satisfied alternative approaches such as the Markov method can be used. Constructing the Markov representation of a system is not such as intuitive process for engineers as fault tree construction since the state-transition diagram does not readily document the failure logic. In addition to this the size of the Markov diagram increases rapidly as the number of components in the system increases. This thesis presents the development of a new model which uses a combination of conventional fault tree methods with those of Markov methods to solve systems containing sequential or standby failures. New gates were developed in order to incorporate the dependent failures on the fault tree structure. The new assessment method was shown to efficiently solve these systems. With theses extended fault tree capabilities in place the technique was embedded within an optimisation framework to obtain the best system performance for systems containing standby failures. Sequential failures can be represented on a fault tree by using the Priority-And gate, however they can also be represented on a Cause-Consequence diagram. As with the fault tree analysis method, the Cause-Consequence Diagram method documents the failure logic of the system. In addition to this the Cause-Consequence Diagram produces the exact failure probability in a very efficient calculation procedure and has significant implications in terms of efficiency for static systems. Construction and analysis rules were devised for a cause-consequence diagram and used on systems containing independent and dependent failures.