Extension of the multi-scale-network-model for the virtual commissioning of complex material flow systems

• Applicants: Professorin Dr. Simone Göttlich; Professor Dr.-Ing. Alexander Verl
• Subject Area: Production Automation and Assembly Technology
• Term: since 2022
• Project identifier: Deutsche Forschungsgemeinschaft (DFG) - Project number 508338261

In this project, a macroscopic flow model for material flow simulation is being further developed and linked with a physics-based model to form a multi-scale network model for use in virtual commissioning in material intensive production plants.Modern production plants are complex and extensive systems with multiple requirements, which can be developed faster and with higher quality using virtual commissioning. For virtual commissioning of control systems, time-deterministic hardware-in-the-loop simulations are needed, which are connected to the control system instead of the real machines. Here, the material flow defined as the whole movement of piece goods in a factory plant, is a particularly computationally intensive element. In preliminary work, a macroscopic flow model based on a hyperbolic partial differential equation was developed, whose computation time is independent of the number of simulated goods, so that material flow-intensive plants such as those in beverage technology can be simulated efficiently. In the previous project "OptiPlant", this model was accelerated by parallelization and the numerical solution was validated. In addition, the fast model was used for automated throughput optimization of the material flow layout. In this project, the applications of the flow model will be extended so that different piece goods, geometries of the plant layout, sources and sinks can be modeled. The flow model is a non-local equation in a bounded domain in two spatial dimensions that incorporates influences from the local environment. The non-locality in combination with the bounded domain leads to special challenges for the numerical and theoretical treatment of the flow model, which have not been explored yet and will now be implemented. Further, combining the modeling with real data will allow parameter estimation to define the previously unknown parameters of the flow model. The application of the flow model to more complex material flow systems (with inflow and outflow of goods, variation of material and geometry and dynamic layout) and the modeling with parameters specifically optimized for the use case, enables the use of the flow model in the virtual commissioning of material flow-rich plants in industry. For the extension as well as for the parameter estimation new real data with more variations are needed, for which a real demonstrator will be developed together with industry contacts and built. Since specific unit loads cannot be represented in the flow model, it is to be combined with a physics-based model such that it is possible to switch between the two types of models in a multi-scale network model. For this purpose, a prediction method has already been developed in the previous project, which predicts results of the physics-based model for the control cycle. In order to avoid having to model the scenarios more than once, a common initial model is to be developed from which the scenarios in both models can be generated.

• DFG Programme: Research Grants