Detecting Non-Ergodic Simulation Models of Logistics Networks

Simulation is a frequently applied method when analysing logistics networks. Also within the Collaborative Research Center 559 "Modelling of Large Logistics Networks" simulation is broadly applied and process chains are used as a mutual basis for model development and description. Previous research activities exposed non-ergodicity of models as one of the typical application-specific problems which are difficult to discover by simulation. In order to detect non-ergodic models the problem has been reduced to its core employing the more analysis oriented modelling formalism of Petri nets. With the help of the Petri net formalism we developed an efficient method for the detection of non-ergodic models. Since Petri nets is not the common modelling paradigm for logisticians, this method had to be made available in the process chain modelling world of the logistics area, additionally supported by an appropriate tool. This paper describes our corresponding approach and also demonstrates the process of identifying a problem class in an application area, reducing it to its core, establishing a solution in an analysis-oriented formalism and making corresponding techniques available in the application-oriented modelling world and thus also available for the end-user.