Systems reliability modelling for phased missions with maintenance-free operating periods

In 1996, a concept was proposed by the UK Ministry of Defence with the intention of making the field of reliability more useful to the end user, particularly within the field of military aerospace. This idea was the Maintenance Free Operating Period (MFOP), a duration of time in which the overall system can complete all of its required missions without the need to undergo emergency repairs or maintenance, with a defined probability of success. The system can encounter component or subsystem failures, but these must be carried with no effect to the overall mission, until such time as repair takes place. It is thought that advanced technologies such as redundant systems, prognostics and diagnostics will play a major role in the successful use of MFOP in practical applications. Many types of system operate missions that are made up of several sequential phases. For a mission to be successful, the system must satisfactorily complete each of the objectives in each of the phases. If the system fails or cannot complete its goals in any one phase, the mission has failed. Each phase will require the system to use different items, and so the failure logic changes from phase to phase. Mission unreliability is defined as the probability that the system fails to function successfully during at least one phase of the mission. An important problem is the efficient calculation of the value of mission unreliability. This thesis investigates the creation of a modelling method to consider as many features of systems undergoing both MFOPs and phased missions as possible. This uses Petri nets, a type of digraph allowing storage and transit of tokens which represent system states. A simple model is presented, following which, a more complex model is developed and explained, encompassing those ideas which are believed to be important in delivering a long MFOP with a high degree of confidence. A demonstration of the process by which the modelling method could be used to improve the reliability performance of a large system is then shown. The complex model is employed in the form of a Monte-Carlo simulation program, which is applied to a life-size system such as may be encountered in the real world. Improvements are suggested and results from their implementation analysed.