ABSTRACT
Monte Carlo simulation is widely used to compute dependability and performance measures of complex systems, in particular for dynamic systems. It is both simple to implement and able to take into account easily non markovian models. Its main drawback is the fact that it may take a very long simulation time to estimate the system reliability/availability with reasonable accuracy because of the rareness of the system failure. We present in this paper a method able to speed up the simulation while remaining simple to implement, and we discuss its performances on the basis of two case studies: the first one is about a non markovian system, and the second one is a model of emergency power supplies of a nuclear power plant.
