Decarbonization, cost increases, pace of development, pressure to innovate — without a comprehensive product-related knowledge base, it is impossible to overcome these challenges, some of which are conflicting goals. This is particularly true for technology-intensive products that must meet high requirements in terms of safety, reliability, and service life, and it applies equally to cast components.
The diverse quality characteristics of cast components result from a complex interaction between the properties of the raw material and the process parameters. If a parameter or alloy component changes, or if damage occurs during component use, the cause must be investigated and the balance between materials, process, and component properties must be readjusted. Where to start is often based on a mix of experience and available data. Trial and error loops are often the method of choice. However, they reach their limits when the CO2 footprint of the component, the verification of the materials in components, and the remaining service life of the component are required, especially with a large product portfolio and limited time.
What is needed then is, on the one hand, systematic traceability, i.e., knowledge of the many factors that have led to specific materials and component conditions, and, on the other hand, the quantitative relationships between process conditions and parameters, the resulting microstructure of the cast materials, and ultimately the resulting component properties. In order to represent the complexity of the casting process, research scientists at Fraunhofer IWM have developed a knowledge graph that describes the complex process chain in detail. In addition, the process data was analyzed using machine learning to draw new insights into process optimization and prediction of casting qualities from the data.
To "breathe life" into the digital twin, it is filled with data from the machine control system and process sensors. The twin is also fed with materials properties and microstructure data. Queries can now be made in this data space, for example, on the die casting history of a specific cast component, on process windows that lead to increased porosity and oxide contamination in highly stressed component areas, as well as on the batches that were used in a rejected component and the shot parameters used.
In addition, the knowledge graph is an ideal starting point for the so-called digital product passport (DPP), which is part of the EU regulation on the environmentally friendly design of sustainable products and in which information on sustainability and circular economy of products, components, and materials is stored and digitally accessible.
In addition to queries for specific information, cause-and-effect relationships can be described mathematically with digital die casting twins. For example, the research project described the relationship between process parameters such as casting temperature and pressure on oxide contamination and porosity, and the relationship between outside temperature and humidity and casting pressure. The ultimate goal is to be able to make predictions that eliminate time-consuming analysis steps in quality assurance and replace trial-and-error loops in development and production with systematic process parameter studies.
Technologically, the "transparent die casting process" is based on a knowledge graph that maps all relevant steps in the process chain — from melt production to the casting process and heat treatment to testing and component use — in a standardized, ontology-based form. It is machine-readable, expandable, versionable, and structured in such a way that it supports both classic evaluations and AI methods. This makes the knowledge graph the central semantic data hub for the digital twin.
Every die casting process is different, and so are the specific information requirements of foundries and users of cast components. The configuration of the digital twin therefore begins with a comprehensive inventory of the process steps and the necessary and available data. For the process description, the existing knowledge graph modules are adapted or expanded. Once the data structure has been created, the queries are programmed so that the digital knowledge base can lead to greater value creation and quality.
The "transparent die casting process" was developed in the Fraunhofer Future Car Production flagship project to enable sustainable vehicle construction and empower development process companies to make robust and holistic decisions about components and technologies. Data from Fraunhofer IFAM and Fraunhofer EZRT was used in this process. Fraunhofer IWM will be presenting the demonstrator for the digital die casting process at the EUROGUSS trade show in Nuremberg from January 13 to 15, 2026, in Hall 4a, Booth 233.
Fraunhofer Institute for Mechanics of Materials IWM
