AGENT-BASED SIMULATION OF DISCRETE-EVENT SYSTEMS
G.Allen Pugh · 2006
Manufacturing facilities can often be modeled as a collection of queues. However, once queueing system complexity rises beyond the trivial, analytical methods become intractable. In these cases, discrete-event simulation is essentially the only tool capable of modeling such systems. In this paper, two approaches to modeling manufacturing queueing systems are explored. While there is a plethora of simulation software tools available, consideration here will be constrained to those most suitable for (and available to) education. The following criteria were used to narrow the selection. • The software must be inexpensive, or be available to to students at a substantial discount. The rationale for this criterion is that students can complete the course with their own copy of the tool. • The interface must be graphical so that models can be readily constructed. Less time consumed in model construction translates into greater time for analysis and fine tuning. • The software should allow for (optional) animation and interaction while running. These features, while computationally expensive, encourage experimentation. • Numerous example and help files should be available. • Computational demands should be reasonable. The software should run on the types of machines students are likely to have access to. 2. WORLD VIEWS Two very different types of world views are examined. The first is the traditional discrete-event simulation modeling environment exemplified by Extend