SFB/ TRR 402: Model reduction and formalisation to increase the efficiency of integrative simulation chains

News

27.05.2025 | Events
Enthusiasm for technology – A successful Mechanical Engineering Day in Aachen

On Saturday, 24 May 2025, the Faculty of Mechanical Engineering at RWTH Aachen University opened its doors for the Mechanical Engineering Day. The event on the Melaten campus offered visitors a comprehensive insight into the diversity and innovative strength

read more

Teilprojekt C03

The objective of sub-project C03 is to investigate the efficiency of integrative simulation chains in the context of component and process optimisation and to identify methods for reducing their complexity and efforts. These will serve as a basis for the computation of gradings in material properties of composite material structures investigated within the overall project.

Thermoplastic (and to a certain extent thermoset and ceramic) materials with glass or carbon fibers as reinforcing phases are investigated. These composite types are illustrated by graded wallpaper laminates (A02, B05), realigned discontinuous fibers (B03) and different grades of reinforcement in injection-molded structures (B01, C01, C02).

SFB TRR 402 - C03 — Conventional methodology for determining local material properties using the example of an in-house designed component in contrast to the planned data-driven approach.

The project is investigating methods for data-driven prediction of local material properties such as stiffness or thermal conductivity. Supported by machine learning methods, the procedures use both previously acquired experimental data and synthetic data from established analytical relationships and numerical models as a database. The materials to be examined are categorised and parameterised into morphology classes according to the material components and their morphology resulting from the manufacturing process. Based on this, suitable numerical models are developed, selected and transferred into a modularised Knowledge Discovery in Databases (KDD) procedure for the description of homogenised composite material properties. They are intended to enable efficient surrogate modeling to reduce computing time.

To validate the data-driven models, the generated results are compared with multiscale simulations outside the KDD data set. Morphology modelling, data collection and storage between the relevant sub-projects will be standardised and combined in a central database. The result of the subproject is a method with which data-driven local mechanical and thermal property predictions can be made within the almost infinitely finely graded combinatorial parameter space of morphology and constituent parameters of the materials investigated in the overall project. The iEP system should then have the entire spectrum of graded materials represented in the overall project available for component and process optimisation without computing time restrictions.