Reliability modelling for systems with stress related dependencies

Abstract

Many of the more efficient systems reliability modelling methods require the components which comprise the system to fail independently. When this condition is true fault trees can be developed for the system failure model and the analysis can be performed taking advantage of the binary decision diagram (BDD) technique. The advantages of such an approach are well documented. When dependencies exist techniques such as simulation or Markov methods can be employed with the associated demands on computational resources. Many system features can give rise to dependencies. One such situation is where a stress-related dependency is encountered. In this case the system has a parallel structure and failures of the parallel streams place higher stresses (with increased failure rate) on those streams still functioning. This paper describes an approach for modelling systems with stress related dependencies. It utilises the BDD method to retain efficiency and forms appropriate integrals to yield the failure probability of each failure mode. An example of where this approach can be applied is used to demonstrate the methodology.