Top

Predicting Lattice Crushing Response with Contact State Gated MeshGraphNets

Authors

Files

PDF

Abstract

Lattice crushing involves evolving load paths, irreversible deformation, and intermittent contact between members that are distant in the undeformed mesh. A surrogate restricted to structural connectivity may therefore miss the interactions that govern compaction. This paper proposes a contact state gated MeshGraphNets model for predicting nodal motion and platen reaction during displacement-controlled compression. The formulation combines permanent mesh edges with dynamically updated surface contact edges, using geometric gaps, relative velocities, and edge persistence to distinguish approaching, sustained, and separating interactions. A short rollout objective couples state prediction with reaction consistency and penetration control. The evaluation design considers 480 trajectories spanning three lattice families, with geometry separated training, validation, and test partitions. Across the comparison, median normalized displacement error decreases from 8.4% for a mesh only baseline to 3.2% for the proposed model, while reaction curve error decreases from 12.6% to 4.8%. Maximum penetration decreases from 0.31 to 0.06 mm, and the proposed rollout requires 18 s against a 720 s reference solve. The evaluation tracks both local geometric admissibility and the global force response throughout compaction. The proposed protocol distinguishes contact representation from processor capacity through controlled ablations and evaluates compaction, generalization, and computational cost separately. The formulation establishes a testable route toward contact sensitive surrogates, with separate criteria for prediction accuracy, contact admissibility, and computational efficiency. The present scope is monotonic compression of intact elastoplastic lattices; fracture, unloading, and impact conditions require separate validation.

PDF

References

Bieler, S., & Weinberg, K. (2025). Energy absorption of sustainable lattice structures under static compression. Meccanica, 60(8), 2533–2547. https://doi.org/10.1007/s11012-025-02003-4

Yu, Z., Savinov, R., Matura, M., Zhang, P., & Shi, J. (2025). Current research status on advanced lattice structures for impact and energy absorption applications: A systematic review. Thin-Walled Structures, 215, Article 113490. https://doi.org/10.1016/j.tws.2025.113490

Ding, J., Ma, Q., Li, X., Zhang, L., Yang, H., Qu, S., Wang, M. Y., Zhai, W., Gao, H., & Song, X. (2024). Imperfection-enabled strengthening of ultra-lightweight lattice materials. Advanced Science, 11(41), Article 2402727. https://doi.org/10.1002/advs.202402727

Daynes, S., & Feih, S. (2025). Functionally graded lattice structures with tailored stiffness and energy absorption. International Journal of Mechanical Sciences, 285, Article 109861. https://doi.org/10.1016/j.ijmecsci.2024.109861

Maddio, P. D., Sinatra, R., & Cammarata, A. (2026). A novel 2D penalty formulation for frictionless contact. Multibody System Dynamics, 67(1), 221–245. https://doi.org/10.1007/s11044-025-10119-4

Guo, Z., Yang, F., Li, L., & Wu, J. (2024). Bio-inspired curved-elliptical lattice structures for enhanced mechanical performance and deformation stability. Materials, 17(17), Article 4191. https://doi.org/10.3390/ma17174191

Scheunemann, L., & Faust, E. (2026). A manifold learning approach to nonlinear model order reduction of quasi-static problems in solid mechanics: Proof of concept and parameter study. Computational Mechanics, 77(4), 983–1023. https://doi.org/10.1007/s00466-025-02696-0

Storm, J., Rocha, I. B. C. M., & van der Meer, F. P. (2024). A microstructure-based graph neural network for accelerating multiscale simulations. Computer Methods in Applied Mechanics and Engineering, 427, Article 117001. https://doi.org/10.1016/j.cma.2024.117001

Dubey, V. K., Haese, C. E., Gültekin, O., Dalton, D., Rausch, M. K., & Fuhg, J. (2026). Graph neural network surrogates for contacting deformable bodies with necessary and sufficient contact detection. Computer Methods in Applied Mechanics and Engineering, 448, Article 118413. https://doi.org/10.1016/j.cma.2025.118413

Prakash, S. K. A., & Tucker, C. (2024). Probabilistic graph networks for learning physics simulations. Journal of Computational Physics, 513, Article 113137. https://doi.org/10.1016/j.jcp.2024.113137

Gongora, A. E., Friedman, C., Newton, D. K., Yee, T. D., Doorenbos, Z., Giera, B., Duoss, E. B., Han, T. Y.-J., Sullivan, K., & Rodriguez, J. N. (2024). Accelerating the design of lattice structures using machine learning. Scientific Reports, 14(1), Article 13703. https://doi.org/10.1038/s41598-024-63204-7

Vlachas, K., Garland, A., Quinn, D. D., & Chatzi, E. (2024). Parametric reduced-order modeling for component-oriented treatment and localized nonlinear feature inclusion. Nonlinear Dynamics, 112(5), 3399–3420. https://doi.org/10.1007/s11071-023-09213-z

Kneifl, J., Fehr, J., Brunton, S. L., & Kutz, J. N. (2025). Multi-hierarchical surrogate learning for explicit structural dynamical systems using graph convolutional neural networks. Computational Mechanics, 75(3), 1115–1135. https://doi.org/10.1007/s00466-024-02553-6

Würth, T., Freymuth, N., Zimmerling, C., Neumann, G., & Kärger, L. (2024). 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, Article 117102. https://doi.org/10.1016/j.cma.2024.117102

Gladstone, R. J., Rahmani, H., Suryakumar, V., Meidani, H., D’Elia, M., & Zareei, A. (2024). Mesh-based GNN surrogates for time-independent PDEs. Scientific Reports, 14(1), Article 3394. https://doi.org/10.1038/s41598-024-53185-y

Perera, R., & Agrawal, V. (2024). Multiscale graph neural networks with adaptive mesh refinement for accelerating mesh-based simulations. Computer Methods in Applied Mechanics and Engineering, 429, Article 117152. https://doi.org/10.1016/j.cma.2024.117152

Pegolotti, L., Pfaller, M. R., Rubio, N. L., Ding, K., Brugarolas Brufau, R., Darve, E., & Marsden, A. L. (2024). Learning reduced-order models for cardiovascular simulations with graph neural networks. Computers in Biology and Medicine, 168, Article 107676. https://doi.org/10.1016/j.compbiomed.2023.107676

Dornheim, J., Morand, L., Nallani, H. J., & Helm, D. (2024). Neural networks for constitutive modeling: From universal function approximators to advanced models and the integration of physics. Archives of Computational Methods in Engineering, 31(2), 1097–1127. https://doi.org/10.1007/s11831-023-10009-y

Bernard, A. R., & ElSayed, M. S. A. (2024). Design, manufacturing, and analysis of periodic three-dimensional cellular materials for energy absorption applications: A critical review. Materials, 17(10), Article 2181. https://doi.org/10.3390/ma17102181

Li, L., Yang, F., Guo, Z., Wu, J., Lu, G., & Fan, H. (2024). Truss-plate hybrid lattice metamaterials with broadband vibration attenuation and enhanced energy absorption. Virtual and Physical Prototyping, 19(1), Article e2345386. https://doi.org/10.1080/17452759.2024.2345386

Dai, X., Kövecses, J., & Teichmann, M. (2024). Dynamic simulation and collision detection for flexible mechanical systems with contact using the floating frame of reference formulation. Journal of Computational and Nonlinear Dynamics, 19(11), Article 111005. https://doi.org/10.1115/1.4066329

Fatoretto, M. B., Demétrio, C. G. B., Brophy, C., & de Andrade Moral, R. (2026). Bootstrap confidence intervals for generalized linear mixed models in dose-response analysis. Journal of Statistical Computation and Simulation, 96(14), 3630–3653. https://doi.org/10.1080/00949655.2026.2677023

Bozkurt, I. (2024). Determination of deformation behaviors and energy absorption of lightweight composite lattice cylinders with different structures. International Journal of Polymer Science, 2024(1), Article 2280726. https://doi.org/10.1155/2024/2280726

Du, Y., Li, Y., Coros, S., & Thomaszewski, B. (2024). Robust and artefact-free deformable contact with smooth surface representations. Computer Graphics Forum, 43(8), Article e15187. https://doi.org/10.1111/cgf.15187

Raymundo, C., Pérez-Zuñiga, G., & Yupanqui, I. (2026). Phase-dependent nonlinear MPC for stair climbing: Edge-pivot control with a double-pendulum model. Results in Control and Optimization, 22, Article 100675. https://doi.org/10.1016/j.rico.2026.100675

Kängsepp, M., Valk, K., & Kull, M. (2025). On the usefulness of the fit-on-test view on evaluating calibration of classifiers. Machine Learning, 114(4), Article 105. https://doi.org/10.1007/s10994-024-06652-6

Gao, J., & McShane, G. J. (2026). Using heterogeneous fold patterns to manipulate the compressive collapse mechanisms of ductile origami-inspired metamaterials. International Journal of Mechanical Sciences, 314, Article 111377. https://doi.org/10.1016/j.ijmecsci.2026.111377

Ma, J. (2024). A survey of out-of-distribution generalization for graph machine learning from a causal view. AI Magazine, 45(4), 537–548. https://doi.org/10.1002/aaai.12202

Phan, Q. M., & Luong, N. H. (2024). Parameter-less Pareto local search for multi-objective neural architecture search with the Interleaved Multi-start Scheme. Swarm and Evolutionary Computation, 87, Article 101573. https://doi.org/10.1016/j.swevo.2024.101573

Geiß, A., Hovi, T., Calotoiu, A., & Wolf, F. (2026). Validating the performance of GPU ports using differential performance models. Future Generation Computer Systems, 174, Article 108018. https://doi.org/10.1016/j.future.2025.108018

Details