Hierarchical simulation of complex production systems by coupling of models

The simulation of production systems usually requires rather detailed data, concerning the duration of the modelled activities, which determine the quality and reliability of simulation results. Application in industry has shown that these data are usually available for manufacturing. However, for non-manufacturing tasks, only rough data are available as expert guesses. Therefore, in order to simulate complex production systems which include manufacturing functions as well as pre- and post-manufacturing functions, it is often necessary to combine simulation models with different levels of detail. Therefore an adequate approach is needed in order to avoid inconsistencies in results. Such inconsistencies may be connected with different levels of detail and occur if e.g. a highly detailed manufacturing model is combined with a more global simulation model for the winding-up of customer orders. The solution until now has been to define a simulation model with only one unique level of detail. If highly specific results are needed, the user is forced to build a very detailed model of the whole production system. Often, this turns out to be either impossible or connected with too much effort. If more global models are used instead, the obtained results may be insufficient to render an answer to the user's questions. At the ifab-Institute of the University of Karlsruhe, a new simulation tool O Sim, which allows for the integration of variously detailed submodels into an overall global model of the investigated production system, is under development. The consistency of different models is guaranteed by a specific method of coupling the models. In this hierarchical way, the global simulation model can be combined with various submodels which contain more detailed processes. These detailed processes are then synchronised with the processes of the global model. For this purpose, the global model initiates an underlying process which then delivers its result to the global process. This combination of variously detailed models can be extended hierarchically. Following the description of this new simulation concept, a practical example for application will be given. This example is derived from an aluminium factory. It concerns the simulation of the overall winding-up of customer orders with their underlying manufacturing processes. The consistency of this simulation approach will be demonstrated by evaluating logistical key data at various levels of detail.