Fuzzy multiphase Markov chains to handle uncertainties in safety systems performance assessment

In this article, we address the problem of imprecision in assessing the performance of safety instrumented systems (SIS) using fuzzy multiphase Markov chains. The elementary probabilities usually considered in Markov chains are replaced by fuzzy numbers. It allows experts to express their uncertainty concerning the basic parameters of systems and to evaluate the impact of this uncertainty on the SIS performance. We show how the imprecision induces significant changes on the Safety Integrity Level of the SIS. The proposed method ensures the relevance of the results. This is validated by a comparison with the results of an enhanced Markov Analysis.