Bereiche | Tage | Auswahl | Suche | Aktualisierungen | Downloads | Hilfe

CPP: Fachverband Chemische Physik und Polymerphysik

CPP 24: French-German Session: Simulation Methods and Modeling of Soft Matter IV

CPP 24.2: Vortrag

Dienstag, 10. März 2026, 11:45–12:00, ZEU/0255

A data-driven decoupled multiscale scheme for anisotropic finite strain magneto-elasticity — •Heinrich T. Roth, Philipp Gebhart, Karl A. Kalina, Thomas Wallmersperger, and Markus Kästner — TU Dresden, Dresden, Germany

Structured magnetorheological elastomers (MREs) are composite materials exhibiting magneto-mechanical coupling effects, such as the magnetostrictive and magnetorheological effect. They consist of magnetizable particles arranged in chain-like structures within a soft elastomer matrix. As explicitly resolving their microstructure in real-world samples is infeasible, a multiscale modeling approach is required.

In this work, we present a framework for the macroscale modeling of structured MREs using physics-augmented neural networks (PANNs) [1,2]. The framework begins with data generation, where a representative volume element (RVE) undergoes macroscopic magneto-mechanical loadings in Finite Element (FE) simulations. The resulting homogenized microscale variables form a macroscale dataset for the training and testing of the PANN macromodel, which satisfies key physical principles [1]. Finally, the trained PANN model is used in a decoupled multiscale scheme as the material model for a macroscale FE simulation to examine the magnetostriction of a spherical sample.

We acknowledge support by the German Research Foundation DFG through Research Unit FOR 5599 on structured magnetic elastomers.

[1] H.T. Roth et al., arXiv:2510.24197, 2025. [2] K.A. Kalina et al., CMAME 421, 2024.

Keywords: Magnetorheological elastomers; Finite strain magneto-elasticity; Physics-augmented neural networks; Constitutive modeling; Finite Element Method