Algebraic Hierarchical Graph Neural Networks for Forming Simulation of Thermoplastic Composite Materials

Abstract:

Stamp forming of fiber-reinforced thermoplastic composite materials is governed by large deformations, anisotropic and rate-dependent material behavior, and frictional multi-body contact, making high-fidelity finite element simulations expensive and often impractical for rapid design studies and process optimization. We leverage recent advancements in Machine Learning-based simulations and tailor Algebraic-hierarchical Message Passing Networks (AMPNs) to stamp forming simulation of composite materials. To efficiently handle multi body contact during forming, we model the laminates by a multi-layer graph with explicit ply–ply and tool–ply contact and extend AMPNs by local component-wise contact edges. Using a multiscale graph hierarchy, the method captures local wrinkling effects, global material draw-in, and contact-driven deformation across the full laminate. Trained on high-fidelity data from state-of-the-art Finite Element Method (FEM) simulations, the surrogate accurately simulates the stamp forming process for unseen process settings, while reducing simulation times from hours to seconds, enabling approximately real time simulation of large, complex geometries.

• Periodical: Key Engineering Materials (Volume 1050)
• Pages: 31-42
• DOI: https://doi.org/10.4028/p-kaH2TN
• Online since: April 2026
• Authors: Tobias Würth*, Niklas Freymuth, Johannes Mitsch, Philipp Dahlinger, Tai Hoang, Gerhard Neumann, Luise Kärger
• Keywords: Algebraic Multi-Grid Coarsening, Data-Driven Simulation, Forming, Graph Network Simulators

References

[1] Dibakar Bhattacharyya. Composite sheet forming, volume 11. Elsevier, 1997.

[2] Peng Wang, Nahiène Hamila, and Philippe Boisse. Thermoforming simulation of multilayer composites with continuous fibres and thermoplastic matrix. Composites Part B: Engineering, 52:127–136, 2013.

DOI: 10.1016/j.compositesb.2013.03.045

[3] Adil Wazeer, Apurba Das, Chamil Abeykoon, Arijit Sinha, and Amit Karmakar. Composites for electric vehicles and automotive sector: A review. Green Energy and Intelligent Transportation, 2(1), 2023.

DOI: 10.1016/j.geits.2022.100043

[4] Mats G.Larsonand Fredrik Bengzon. The Finite Element Method: Theory, Implementation, and Applications, volume 10 of Texts in Computational Science and Engineering. Springer, Berlin, Heidelberg, 2013. ISBN 978-3-642-33286-9 978-3-642-33287-6.

[5] Philippe Boisse, Remko Akkerman, Pierpaolo Carlone, Luise Kärger, Stepan V. Lomov, and James A. Sherwood. Advances in composite forming through 25 years of ESAFORM. In ternational Journal of Material Forming, 15(3):99, January 2022. ISSN 1960-6214.

DOI: 10.1007/s12289-022-01682-8

[6] Philippe Boisse, Jin Huang, and Eduardo Guzman-Maldonado. Analysis and modeling of wrin kling in composite forming. Journal of Composites Science, 5(3):81, 2021.

[7] Xiaoxiao Guo, Wei Li, and Francesco Iorio. Convolutional Neural Networks for Steady Flow Approximation. In Proceedings of the 22nd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, pages 481–490, San Francisco California USA, August 2016. ACM. ISBN978-1-4503-4232-2.

DOI: 10.1145/2939672.2939738

[8] Tobias Pfaff, Meire Fortunato, Alvaro Sanchez-Gonzalez, and Peter Battaglia. Learning Mesh Based Simulation with Graph Networks. In International Conference on Learning Representa tions, October 2020.

[9] Alvaro Sanchez-Gonzalez, Jonathan Godwin, Tobias Pfaff, Rex Ying, Jure Leskovec, and Peter Battaglia. Learning to Simulate Complex Physics with Graph Networks. In Proceedings of the 37th International Conference on Machine Learning, pages 8459–8468. PMLR, November 2020.

[10] M. Raissi, P. Perdikaris, and G.E. Karniadakis. Physics-informed neural networks: A deep learning framework for solving forward and inverse problems involving nonlinear partial differential equations. Journal of Computational Physics, 378:686–707, February 2019. ISSN 00219991.

DOI: 10.1016/j.jcp.2018.10.045

[11] Tobias Würth, Constantin Krauß, Clemens Zimmerling, and Luise Kärger. Physics-informed neural networks for data-free surrogate modelling and engineering optimization– An example from composite manufacturing. Materials & Design, 231:112034, July 2023. ISSN 02641275.

DOI: 10.1016/j.matdes.2023.112034

[12] Han Gao, Luning Sun, and Jian-Xun Wang. PhyGeoNet: Physics-informed geometry-adaptive convolutional neural networks for solving parameterized steady-state PDEs on irregular domain. Journal of Computational Physics, 428:110079, March 2021. ISSN 00219991. doi: 10.1016/j. jcp.2020.110079.

DOI: 10.1016/j.jcp.2020.110079

[13] Dule Shu, Zijie Li, and Amir Barati Farimani. A physics-informed diffusion model for high fidelity flow field reconstruction. Journal of Computational Physics, 478:111972, April 2023. ISSN 0021-9991.

DOI: 10.1016/j.jcp.2023.111972

[14] Han Gao, Matthew J. Zahr, and Jian-Xun Wang. Physics-informed graph neural Galerkin net works:Aunified framework for solving PDE-governed forward and inverse problems. Computer Methods in Applied Mechanics and Engineering, 390:114502, February 2022. ISSN 0045-7825.

DOI: 10.1016/j.cma.2021.114502

[15] Tobias Würth, Niklas Freymuth, Clemens Zimmerling, Gerhard Neumann, and Luise Kärger. Physics-informed MeshGraphNets (PI-MGNs): Neural finite element solvers for non-stationary and nonlinear simulations on arbitrary meshes. Computer Methods in Applied Mechanics and Engineering, 429:117102, September 2024. ISSN 0045-7825.

DOI: 10.1016/j.cma.2024.117102

[16] Zongyi Li, Nikola Borislavov Kovachki, Kamyar Azizzadenesheli, Burigede Liu, Kaushik Bhattacharya, Andrew Stuart, and Anima Anandkumar. Fourier Neural Operator for Parametric Partial Differential Equations. In International Conference on Learning Representations, October 2020.

[17] Johannes Brandstetter, Daniel E. Worrall, and Max Welling. Message Passing Neural PDE Solvers. In International Conference on Learning Representations, October 2021.

[18] Jonas Linkerhägner, Niklas Freymuth, Paul Maria Scheikl, Franziska Mathis-Ullrich, and Ger hard Neumann. Grounding graph network simulators using physical sensor observations. In International Conference on Learning Representations, 2023.

[19] Phillip Lippe, Bas Veeling, Paris Perdikaris, Richard Turner, and Johannes Brandstetter. PDE Refiner: Achieving Accurate Long Rollouts with Neural PDE Solvers. Advances in Neural Information Processing Systems, 36:67398–67433, December 2023.

DOI: 10.52202/075280-2946

[20] David Ruhe, Jonathan Heek, Tim Salimans, and Emiel Hoogeboom. Rolling Diffusion Models. In Forty-First International Conference on Machine Learning, June 2024.

[21] Tobias Würth, Niklas Freymuth, Gerhard Neumann, and Luise Kärger. Diffusion-based hierar chical graph neural networks for simulating nonlinear solid mechanics. In Advances in Neural Information Processing Systems, 2025.

[22] Marta Garnelo, Dan Rosenbaum, Christopher Maddison, Tiago Ramalho, David Saxton, Murray Shanahan, Yee Whye Teh,Danilo Rezende, and S. M. Ali Eslami. Conditional Neural Processes. In Proceedings of the 35th International Conference on Machine Learning, pages 1704–1713. PMLR, 2018. URL https://proceedings.mlr.press/v80/garnelo18a.html.

[23] Philipp Dahlinger, Tai Hoang, Denis Blessing, Niklas Freymuth, and Gerhard Neumann. Mango—adaptable graph network simulators via meta-learning. In The Thirty-ninth Annual Conference on Neural Information Processing Systems, 2025.

[24] Philipp Dahlinger, Niklas Freymuth, Tai Hoang, Tobias Würth, Michael Volpp, Luise Kärger, and Gerhard Neumann. Context-aware learned mesh-based simulation via trajectory-level meta learning. arXiv preprint arXiv:2511.05234, 2025.

[25] Steeven Janny, Aurélien Bénéteau, Madiha Nadri, Julie Digne, Nicolas Thome, and Christian Wolf. EAGLE: Large-scale Learning of Turbulent Fluid Dynamics with Mesh Transformers. In The Eleventh International Conference on Learning Representations, September 2022.

[26] Youn-Yeol Yu, Jeongwhan Choi, Woojin Cho, Kookjin Lee, Nayong Kim, Kiseok Chang, ChangSeung Woo, Ilho Kim, SeokWoo Lee, Joon Young Yang, Sooyoung Yoon, and Noseong Park. Learning Flexible Body Collision Dynamics with Hierarchical Contact Mesh Transformer. In International Conference on Learning Representations, October 2023.

[27] Kelsey Allen, Tatiana Lopez-Guevara, Kimberly L. Stachenfeld, Alvaro Sanchez Gonzalez, Pe ter Battaglia, Jessica B. Hamrick, and Tobias Pfaff. Inverse Design for Fluid-Structure Interac tions using Graph Network Simulators. Advances in Neural Information Processing Systems, 35:13759–13774, December 2022.

DOI: 10.52202/068431-1000

[28] Clemens Zimmerling, Christian Poppe, Oliver Stein, and Luise Kärger. Optimisation of manu facturing process parameters for variable component geometries using reinforcement learning. Materials & Design, 214:110423, February 2022. ISSN 02641275.

DOI: 10.1016/j.matdes.2022.110423

[29] J. W. Ruge and K. Stüben. 4. Algebraic Multigrid. In Stephen F. McCormick, editor, Multigrid Methods, pages 73–130. Society for Industrial and Applied Mathematics, January 1987. ISBN 978-1-61197-188-0 978-1-61197-105-7.

DOI: 10.1137/1.9781611971057.ch4

[30] Petr Vanek, Jan Mandel, and Marian Brezina. Algebraic multigrid by smoothed aggregation for second and fourth order elliptic problems. Computing, 56(3):179–196, 1996.

DOI: 10.1007/bf02238511

[31] Dominik Dörr, Fabian J. Schirmaier, Frank Henning, and Luise Kärger. A viscoelastic approach for modeling bending behavior in finite element forming simulation of continuously fiber re inforced composites. Composites Part A: Applied Science and Manufacturing, 94:113–123, January 2017. ISSN 1359835X.

DOI: 10.1016/j.compositesa.2016.11.027

[32] Dominik Dörr, Markus Faisst, Tobias Joppich, Christian Poppe, Frank Henning, and Luise Kärger. Modelling Approach for Anisotropic Inter-Ply Slippage in Finite Element Forming Sim ulation of Thermoplastic UD-Tapes. In AIP Proceedings of the 21th International ESAFORM Conference on Material Forming, January 2018.

DOI: 10.1063/1.5034806

[33] Christian Poppe, Tobias Joppich, Dominik Dörr, Luise Kärger, and Frank Henning. Modeling and validation of gripper induced membrane forces in finite element forming simulation of con tinuously reinforced composites. In AIP Conference Proceedings, volume 1896, page 030002, October 2017.

DOI: 10.1063/1.5007989

[34] M. A. Khan, T. Mabrouki, E. Vidal-Sallé, and P. Boisse. Numerical and experimental analyses of woven composite reinforcement forming using a hypoelastic behaviour. Application to the double dome benchmark. Journal of Materials Processing Technology, 210 (2): 378–388, January 2010. ISSN 0924-0136.

DOI: 10.1016/j.jmatprotec.2009.09.027

[35] P. Harrison, R. Gomes, and N. Curado-Correia. Press forming a 0/90 cross-ply advanced ther moplastic composite using the double-dome benchmark geometry. Composites Part A: Applied Science and Manufacturing, 54:56–69, January 2013. ISSN 1359835X.

DOI: 10.1016/j.compositesa.2013.06.014

[36] Johannes Mitsch, Bastian Schäfer, and Luise Kärger. Rate-dependent 3D forming simulation of thermoplastic composite materials using visco-hyperelastic material modeling and 3D hexahe dral solid-shell elements. Composites Part A: Applied Science and Manufacturing, 200:109306, January 2026. ISSN 1359-835X.

DOI: 10.1016/j.compositesa.2025.109306

[37] Nathan Bell, Luke N. Olson, and Jacob Schroder. PyAMG: Algebraic multigrid solvers in python. Journal of Open Source Software, 7(LA-UR-23-26551), 2022.

DOI: 10.21105/joss.04142

[38] Jonathan Ho, Ajay Jain, and Pieter Abbeel. Denoising diffusion probabilistic models. Advances in neural information processing systems, 33:6840–6851, 2020.

[39] Peter W. Battaglia, Jessica B. Hamrick, Victor Bapst, Alvaro Sanchez-Gonzalez, Vinicius Zam baldi, Mateusz Malinowski, Andrea Tacchetti, David Raposo, Adam Santoro, Ryan Faulkner, Caglar Gulcehre, Francis Song, Andrew Ballard, Justin Gilmer, George Dahl, Ashish Vaswani, Kelsey Allen, Charles Nash,Victoria Langston, Chris Dyer, Nicolas Heess, DaanWierstra, Push meet Kohli, Matt Botvinick, Oriol Vinyals, Yujia Li, and Razvan Pascanu. Relational inductive biases, deep learning, and graph networks, October 2018.

[40] Mario Lino Valencia, Tobias Pfaff, and Nils Thuerey. Learning Distributions of Complex Fluid Simulations with Diffusion Graph Networks. In The Thirteenth International Conference on Learning Representations, October 2024.

[41] Stefan Elfwing, Eiji Uchibe, and Kenji Doya. Sigmoid-weighted linear units for neural network function approximation in reinforcement learning. Neural Networks, 107:3–11, November 2018. ISSN 08936080.

DOI: 10.1016/j.neunet.2017.12.012

[42] Diederik P. Kingmaand Jimmy Ba. Adam:A Method for Stochastic Optimization, January2017.

[43] A. Paszke. Pytorch: An imperative style, high-performance deep learning library. arXiv preprint arXiv:1912.01703, 2019.