Analysis of simplification in Markov-based models for performance assessment of Safety Instrumented System

Abstract

One of the challenges in the use of Markov model in reliability analysis is the exponential growth of number of states and transition rates when the system gets complex. In reliability analysis of a complex safety-instrumented system a simplified model is always desired for reducing the computational cost. In practice, the complex models can be simplified to a much simpler model, which can give almost similar results with some degree of deviation. Depending on the need for the level of accuracy of results, the deviation can be partly compensated with introducing correction factors. The approximated result from the simplified model may or may not influence decision making depending on the degree of deviation and expected accuracy from the analysis. This paper shows how a complex model is simplified through an analytical approach. Comparison of results from both complex and simplified models is presented to discuss the extent in which, this variation influences decision making.