Master Thesis Felix Elliger
Approaching the Distributed Simulation of Related Business Processes
© 2011 Felix Elliger
Business process simulation is a powerful tool for the analysis and improvement of business processes based on their models. It is, however, not sufficient to regard a single business process as an isolated unit. Instead, a process is related to other process by sub-process invocations or resource constraints. This thesis addresses the problem of concurrently simulating multiple related business processes along with their resources. Therefore, we develop a formal model of simulation scenarios and provide a mapping onto extended timed colored Petri nets. As such scenarios may grow large, e.g. for the simulation of a whole department or company, we consider the distribution of the simulation as a reasonable measure for achieving a performance improvement. Thus, the generated Petri net must be divided into partitions. Within this thesis, two algorithms for partitioning the gathered Petri nets are proposed. Their effect on the simulation performance is evaluated using a conservative distributed simulation engine. The architecture of the accompanying prototypical implementation is sketched and selected components are described in more detail. In conclusions, we infer on the potential of distributing business process simulations.