From virtual commissioning to simulation-based digital twins (VC2DT)

• Applicant: Professor Dr. Andreas Wortmann
• Subject Area: Production Automation and Assembly Technology
• Term: since 2025
• Project identifier: Deutsche Forschungsgemeinschaft (DFG) - Project number 560217479

The digital twin as an exact replica of a represented cyber-physical production system (CPPS) is an idealistic vision on which research has produced a continuum of digital twin concepts. In production automation, digital twins promise, for example, more efficient process planning, higher reliability and better utilisation of production systems or cost reductions. However, the term digital twin is not uniformly defined. At one end of the continuum of digital twins, the model end, simulation models of machine classes, i.e. with the claim to be valid for all instances of this type of machine, are modelled, simulated an analysed. These models are therefore also called digital twins of a system "as-designed". At the system end of the continuum, one considers CPPS in operation, i.e., "as-operated". These are mostly software systems that represent and manipulate an already existing system. It is essential that "as-operated" digital twins can access the domain expertise encoded in the simulation models, e.g. to interpret sensor data appropriately and to predict system behaviour. To do this, however, the "as-designed" simulation models must be adapted to the reality of the concrete CPPS instances, whose behaviour increasingly diverges from the idealised assumptions of these simulation models due to tolerances in the structure, environmental influences or wear. As a result, these increasingly lose their predictive capabilities and analyses of this can lead to erroneous conclusions. In the context of commissioning, the "as-designed" class models of the systems must therefore be transformed into "as-operated" instance models of a concrete, physically constructed system to continue to enable useful analyses. In the context of the proposed project, the extent to which a model-based methodology can contribute to the (partially) automatic transformation of the class models of CPPS into precise instance models during commissioning is to be investigated. This will build on the applicant's previous work in the area of digital twins by integrating simulation models derived from virtual commissioning into the existing framework. By identifying dominant causes of the discrepancy between simulation results and measured values of the real CPPS, discrepancy models are derived that minimise the discrepancy. These discrepancy models, which are adapted to the system, are inserted into the simulation models at the appropriate point using a tagging modelling language. At the end of the project, the automatic adaptation of the simulation models will be validated on a demonstrator.

• DFG Programme: Research Grants
• Co-Investigator: Professor Dr.-Ing. Alexander Verl