Dynamic fault tree models: techniques for analysis of advanced fault tolerant computer systems

There is a need for the development of methods for evaluating the vulnerability to failure of goods or systems produced using advanced technology. In particular, the systems for which this evaluation is most critical tend to be complex fault tolerant systems intended for applications where a catastrophic failure can mean loss of life. We contribute to this development of evaluation methods by extending the traditionally combinatorial fault tree evaluation method in such a way that it becomes capable of modeling the full range of system behavior that can be expressed with Markov chains for non-repairable systems. The resulting new modeling technique is called dynamic fault trees and combines the best characteristics of both the fault tree and Markov chain modeling methods. This modeling method requires a two-step procedure that is usually needed for analytical modeling methods: model generation followed by model solution. To further extend the dynamic fault tree method, we develop a one-step algorithm in which the model can be solved as it is generated. This helps ease the use of certain approximation methods for reducing model size and helps optimize the use of computation resources.