Institut National Polytechnique de Lorraine
Dependability analysis is crucial to control the risks resulting from failures in modern industrial systems whose complexity increases by leaps and bounds. This paper proposes a modeling approach to construct dynamic models of Fault-Tolerant (FT) systems based on Stochastic Activity Networks (SANs). This approach allows the systematic inclusion of the diagnosis performances to make the dependability analysis. This SAN model is used jointly with the Monte Carlo simulation to make a study of the impact of diagnosis' performances on the availability of a FT system when various redundancy and maintenance policies are employed.