Representing complex systems within discrete event simulation for reliability assessment

During design of a high-fidelity discrete event simulation (DES) of aircraft reliability and maintenance under realistic operational scenario, modelling of complex system functionality and control of the associated state-space explosion, was a major concern. Hierarchical system decomposition limited this explosion but correct and efficient analysis of the aircraft systems remained a concern. This paper reviews Markov, Petri net, fault tree, event tree and RBD (reliability block diagram), illustrating their respective computer implementations. Computing requirements of each are compared. Pathand cut-sets, derived from fault tree, event tree or RBD, were identified to be most efficient computer representations but with the disadvantage that only 'static' systems could be modelled. A static system is one in which order of component failure or other conditional system property, does not determine functionality. The alternate is a 'dynamic' system. However, it is also noted that 'dynamic' system functionality is always a conditional sub-set of an equivalent 'static' path-set. The paper proposes a method that integrates DES with path-sets to allow 'dynamic' system modelling. For large systems, path-set representation rapidly becomes more efficient in storage requirements, even when sparse matrix techniques are used with the alternatives.