The Poisson Simulation Approach to Combined Simulation

This paper modifies the foundations of combined Discrete Event Simulation (DES) and Continuous Systems Simulation (CSS) by extending the CSS part to the stochastic domain through including discrete (although aggregated) entities. The traditional combined DES/CSS simulation, where the modeller selects between a DES and a CSS description for components of the system under study, lacks a solid theoretical foundation. Model behaviour, results and conclusions may then depend on whether a DES or a CSS description is utilised. By introducing Poisson Simulation (PoS) as an extension to CSS, several advantages are obtained. First, a theoretically sound foundation is achieved in which aggregation can be performed safely. Second, combined type problems can, in principle, be performed exclusively in PoS, which is often considerably simpler than including DES. The computational complexity is reduced when a large number of discrete entities can be modelled as flows into and out of a relatively small number of state variables. Third, even when it is practical to include DES, the combined DES/PoS approach is more powerful and flexible than a combination of DES and CSS.