It is common for modern engineering systems to feature dependency relationships between its components. The existence of these dependencies render the fault tree analysis (FTA) and its efficient implementation, the Binary Decision Diagram (BDD) approach, inappropriate in predicting the system failure probability. Whilst the Markov method provides an alternative means of analysis of systems of this nature, it is susceptible to state space explosion problems for large, or even moderate sized systems. Within this thesis, a process is proposed to improve the applicability of the Markov analysis. With this process, the smallest independent sections (modules) which contain each dependency type are identified in a fault tree and analysed by the most efficient method. Thus, BDD and the Markov analysis are applied in a combined way to improve the analysis efficiency. The BDD method is applied to modules which contain no dependency, and the Markov analysis applied to modules in which dependencies exist. Different types of dependency which can arise in an engineering system assessment are identified. Algorithms for establishing a Markov model have also been developed for each type of dependency. Three types of system are investigated in this thesis in the context of dependency modelling: the continuously-operating system, the active-on-demand system and the phased-mission system. Different quantification techniques have been developed for each type of system to obtain the system failure probability and other useful predictive measures. Investigation is also carried out into the use of BDD in assessing non-repairable systems involving dependencies. General processes have been established to enable the quantification.
Identifer | oai:union.ndltd.org:bl.uk/oai:ethos.bl.uk:479449 |
Date | January 2007 |
Creators | Sun, Huiling |
Publisher | Loughborough University |
Source Sets | Ethos UK |
Detected Language | English |
Type | Electronic Thesis or Dissertation |
Source | https://dspace.lboro.ac.uk/2134/34918 |
Page generated in 0.0133 seconds