Stochastic Coloured Petri Nets as a modelling language for complex Event Trees

Abstract

This paper is summarizing the outputs of the first stage of the research focused on an evaluation of suitability of Stochastic Coloured Petri Nets (SCPN) for modelling and simulation of complex Event Tree/Fault Tree models. The suggested approach is based on the basic idea to create only one complex Petri Net model of an Event Tree. This model consists of not only the studied Event Tree (ET) but also of all Fault Trees describing Pivotal Events of the ET. SCPN seem to be a very promising platform for this purpose, because they offer coloured tokens and advanced conditioning of a firing of Petri Net transitions (which represent events of trees). These features, together with a suitable software tool, which is able to work with hierarchical nets, brings stochastic Petri Net (PN) model which is easier to read (less arcs and elements than in normal PN model). A SCPN model can be processed only by simulations, not by solving any exact equations. This is caused by a complexity of relations in a model. Using simulations (e.g. Monte-Carlo) has interesting side effect in wider modelling abilities. This paper also presents examples of the usage of the proposed methodology (i.e. Event Tree for a collision between offshore supply vessels and offshore installations) including a confrontation with classical approaches. For the methodology demonstration, the TimeNet software from TU Ilmenau is used.