ABSTRACT
This contribution shows how system reliability changes when certain components do not only fail due to their intrinsic causes, but also because one or several neighbouring components failed before that. More precisely, small systems with one-sided failure propagation (called domino effects) are studied. In this approach the number of components interacting in this way is restricted to two, but the system analysed can have more components if those components are themselves statistically independent and do not interact (as to failure and faults) with the two domino components just mentioned. The results are further examples of the modelling of statistical dependency in systems without restoration. The results are not restricted in ways typical of the Markov modelling.
