Tutorial: Petri nets as a graphical description medium for many reliability scenarios

One of the basic problems of dependability modeling is the adequate abstraction of real-world technological problems to the principle terms of reliability/safety scenarios. This concerns primarily the definition of components and of faults on different system levels. Given these terms the rest of any modeling needs basic logical operations, mostly AND and OR, and delays, and often some counting (of time or events). All of these basic operations are offered by a fairly simple kind of Petri net (PN), i.e., timed stochastic Petri nets, allowing also for the modeling of immediate activities and of such with a deterministic delay. In this half-tutorial paper it is shown how such Petri nets modeling, i.e., the construction of the relevant nets, works in practice. No math will be needed for that. Several typical engineering virtues are needed; primarily imagination as to how to (i) find simple solutions, since often nonelegant solutions can be correct too, (ii) compose larger PN from elementary building blocks, and (iii) the ability to model the real world by interpreting the so-called tokens of a PN intelligently in different places of one and the same PN. In the appendix it is shown how the PN can also help in the analytical analysis of nonrepairable systems. In that context they are superior to state graphs since they show state durations explicitly.