Search

Digital Twins (DTs) of Cyber-Physical Production Systems (CPPSs) enable the smart automation of production processes, collection of data, and can thus reduce manual efforts for supervising and controlling CPPSs. Realizing DTs is challenging and requires significant efforts for their conception and integration with the represented CPPS. To mitigate this, we present an approach to systematically engineering DTs for injection molding that supports domain-specific customizations and automation of essential development activities based on a model-driven reference architecture. In this approach, reactive CPPS behavior is defined in terms of a Domain-Specific Language (DSL) for specifying events that occur in the physical system. The reference architecture connects to the CPPS through a novel DSL for representing OPC-UA bindings. We have evaluated this approach with a DT of an injection molding machine that controls the machine to optimize the Design of Experiment (DoE) parameters between experiment cycles before the products are molded. Through this, our reference implementation of the DT facilitates the time-consuming setup of a DT and the subsequent injection molding activities. Overall, this facilitates to systematically engineer digital twins with reactive behavior that help to optimize machine use.

Mold filling is an injection molding stage of great significance, because many defects of the plastic components (e.g. weld lines, burrs or insufficient filling) can occur during this process step. Therefore, it plays an important role in determining the quality of the produced parts. Our goal is the temporal refinement in the vicinity of the evolving melt front, in the context of 4D simplex-type space-time grids [1, 2]. This novel discretization method has an inherent flexibility to employ completely unstructured meshes with varying levels of resolution both in spatial dimensions and in the time dimension, thus allowing the use of local time-stepping during the simulations. This can lead to a higher simulation precision, while preserving calculation efficiency. A 3D benchmark case, which concerns the filling of a plate-shaped geometry, is used for verifying our numerical approach [3]. The simulation results obtained with the fully unstructured space-time discretization are compared to those obtained with the standard space-time method and to Moldflow simulation results. This example also serves for providing reliable timing measurements and the efficiency aspects of the filling simulation of complex 3D molds while applying adaptive temporal refinement.