Development of a data driven method for the process design of thermally aided forming processes with additional component evaluation for hot stamping

• Applicants: Dr.-Ing. Dirk Helm; Professorin Dr.-Ing. Marion Merklein
• Subject Area: Primary Shaping and Reshaping Technology, Additive Manufacturing ; Computer-Aided Design of Materials and Simulation of Materials Behaviour from Atomic to Microscopic Scale ; Mechanical Properties of Metallic Materials and their Microstructural Origins
• Term: since 2025
• Project identifier: Deutsche Forschungsgemeinschaft (DFG) - Project number 567215412

Thermally assisted forming processes, such as hot stamping, are technologically particularly demanding production processes with a large number of possible process parameter settings. The effects on the material properties in terms of process-structure-property relationships are not fully known. Process control based on classic methods for estimating material behavior with the aim of producing components with tailored material properties is only possible to a limited extent. The research approach is therefore based on the hypothesis that the synthesis of testing and characterization, modelling and simulation as well as new methods from the field of machine learning make it possible to establish new types of process-structure-property relationships. These in turn make it possible to generate efficient process models for process design and component evaluation in thermally assisted forming processes. The objectives of the project are to develop cause-effect relationships and to develop, apply and evaluate machine learning methods for the improved description of hot stamping processes with the following focal points: 1. a plant-related process model, which is used within the project for process design. The process model can be used to efficiently predict the microstructure and mechanical properties of the material (strength and hardness) at the end of the process for specified process parameters (thermal load path). The plant-related process model is therefore suitable for use in optimizing the process parameters with regard to the desired microstructure and associated properties in the process. 2. a hybrid material model for simulation-supported process design, which allows efficient FEM simulation of the entire hot forming process. The aim of the hybrid material model is to use the advantages of machine learning in material modelling where classical, physically motivated modelling has major challenges. Microstructural processes (such as austenitisation with subsequent quenching with different microstructural states, diffusive processes, phase transformations, etc.), which are very complicated from a physical modelling perspective, are leartned using experimental data. The methodology is developed and validated using the hot stamping process. The goal is to transfer the developed approaches to other materials, alloys and processes.

• DFG Programme: Research Grants