An increasing number of systems operate over a number of consecutive time periods, in which their reliability structure and the consequences of failure differ, in order to perform some overall operation. Each distinct time period is known as a phase and the overall operation is known as a phased mission. Generally, a phased mission fails immediately if the system fails at any point and is considered a success only if all phases are completed without failure. The work presented in this thesis provides efficient methods for the prediction and optimisation of phased mission reliability.
A number of techniques and methods for the analysis of phased mission reliability have been previously developed. Due to the component and system failure time dependencies introduced by the phases, the computational expense of these methods is high and this limits the size of the systems that can be analysed in reasonable time frames on modern computers. Two importance measures, which provide an index of the influence of each component on the system reliability, have also been previously developed. This is useful for the optimisation of the reliability of a phased mission, however a much larger number have been developed for non-phased missions and the different perspectives and functions they provide are advantageous.
This thesis introduces new methods as well as improvements and extensions to existing methods for the analysis of both non-repairable and repairable systems with an emphasis on improved efficiency in the derivation of phase and mission reliability. New importance measures for phased missions are also presented, including interpretations of those currently available for non-phased missions. These provide a number of interpretations of component importance, allowing those most suitable in a given context to be employed and thus aiding in the optimisation of mission reliability. In addition, an extensive computer code has been produced that implements and tests the majority of the newly developed techniques and methods.
A Doctoral Thesis. Submitted in partial fulfillment of the requirements for the award of Doctor of Philosophy of Loughborough University.