A comparison between two indirect exponentiation methods and O.D.E. integration method for RAMS calculations based on homogeneous Markovian models

Abstract : Even if solving Markovian systems is usual for dependability engineers, the choice of an adequate method to compute these systems is an actual problem for them. The object of this technical note is to contribute to the practician's information by presenting a comparison between the performances of the two most frequently used computation methods: indirect exponentiation and O.D.E. integration approaches. The retained comparison criteria and desired accuracy, computer time and computer storage required are calculated separately, although the two former are often interconnected. The O.D.E. integration method appears to be a good choice in the context of this limited study, but the optimized indirect exponentiation methods are interesting for systems with one hundred or more states—because of their good control of accuracy.