Cyber-Physical Simulation @ TU Darmstadt

Are you interested in physics-augmented data-driven material modeling? Then our "MS04 - Data-driven methods for constitutive modeling” at the GACM 2025 will be interesting to you! We aim to bring together young researchers working on different aspects of data-driven material modeling, including - Constitutive modeling based on data-driven-methods such as neural networks and Gaussian processes - Modeling of complex material behavior such as dissipative and multiphysical effects, fracture mechanics, and parametric dependencies - Fulfillment of physical constraints, interpretability, sparsity, and uncertainty - Parameter calibration and material model discovery, e.g., using full-field data - Efficient implementation in numerical schemes such as multiscale simulation and topology optimization We warmly welcome you to submit your abstract by April 1st, and we look forward to interesting presentations and discussions in Braunschweig! You’ll find further information on:  https://lnkd.in/edPg4HQn If you have any questions, feel free to contact us. Dominik K. Klein, Max Rosenkranz, Karl Kalina, Miguel Ángel Moreno-Mateos, Moritz Flaschel

Machine learning in the beautiful city of Paris 🇫🇷 This week, the Cyber-Physical Simulation Group is at the DTE & AICOMAS conference. We’re looking forward to many fruitful discussions and several contributions of our group: Jasper Schommartz: “Symmetric and Parametrized Physics-Augmented Neural Networks for Hyperelastic Constitutive Modeling in Beam Theory” - MS040C, February 20, 9:00 - 11:00 (https://lnkd.in/e9uNCWZg) Dominik K. Klein: “Neural Networks meet Hyperelasticity: On Benefits and Limits of Polyconvexity” - MS040C, February 20, 9:00 - 11:00 (https://lnkd.in/e9uNCWZg) Oliver Weeger: “Physics-Augmented Neural Networks for Hyperelastic Material Modeling with Softening” - MS040D, February 20, 11:30 - 13:30 (https://lnkd.in/evRu8fDW) Fabian Roth: “Physics-Guided Neural Networks for Lyapunov-Stable Dynamic System Identification” - MS045C, February 20, 11:30 - 13:30 (https://lnkd.in/e5eMi63C) Maximilian Kannapinn: “DT4DED: A Digital Twin with Real-Time Slicing for Residual Stress Design of L/DED-W Additive Manufacturing” - MS018C, February 20, 16:15 - 18:15 (https://lnkd.in/ebBu36pB)

Check-out our latest pre-print on "Stable Port-Hamiltonian Neural Networks" 🚀 📃 https://lnkd.in/eqbfvKKF

Meet 𝗙𝗮𝗯𝗶𝗮𝗻 𝗥𝗼𝘁𝗵, who recently joined us as a doctoral researcher 🥳! He will develop 𝘱𝘩𝘺𝘴𝘪𝘤𝘴-𝘨𝘶𝘪𝘥𝘦𝘥 𝘮𝘢𝘤𝘩𝘪𝘯𝘦 𝘭𝘦𝘢𝘳𝘯𝘪𝘯𝘨 𝘮𝘦𝘵𝘩𝘰𝘥𝘴 𝘧𝘰𝘳 𝘮𝘰𝘥𝘦𝘭𝘪𝘯𝘨 𝘥𝘺𝘯𝘢𝘮𝘪𝘤𝘢𝘭 𝘴𝘺𝘴𝘵𝘦𝘮𝘴. 📝 The application of machine learning in modeling dynamic systems has gained significant attention in recent years, particularly for applications such as creating surrogate and reduced-order models for digital twins. However, purely data-driven approaches often face challenges with extrapolation, require large datasets, and can produce physically implausible predictions. Physics-guided machine learning overcomes these limitations by integrating physical laws and domain knowledge into the model architecture. This approach can improve accuracy, robustness, and interpretability while reducing data requirements. 🔬 Continuing with ongoing research, the interplay between machine learning architectures and their dynamic properties, such as stability, will be explored. The focus will be on utilizing these models as surrogates for optimization, control, and seamless integration into engineering products and processes. ❗ Stay tuned to learn more about his exciting research, e.g., for surrogate modeling of DED AM processes (https://lnkd.in/eUw5Ufzv)

Congratulations to Dominik K. Klein for his recent pre-print "Neural networks meet hyperelasticity: A monotonic approach"! 🎉 Thanks to Mokarram Hossain and Antonio J. Gil for making this research in Swansea University possible, as well as to Maximilian Kannapinn and Stephan Rudykh for their contributions to the work. In this work, we applied physics-augmented neural network constitutive models to experimental data of soft elastomers with parametric dependencies on manufacturing parameters. Thereby, we gained interesting insights on the structure of hyperelastic constitutive models.

Congratulations to our doctoral researcher Jasper Schommartz on the publication of his first journal article in CMAME 🥳 Thanks also to Dominik K. Klein and Juan C. Alzate Cobo for their contributions 👏 📝 In this work, we present neural network-based constitutive models for hyperelastic geometrically exact beams. The proposed models are physics-augmented, i.e., formulated to fulfill important mechanical conditions by construction, which improves accuracy and generalization. This enables efficient constitutive surrogate modeling for geometrically exact beams with nonlinear material behavior and cross-sectional deformation, which otherwise would require computationally much more expensive methods. The models are calibrated and tested with data generated for beams with circular and ring-shaped hyperelastic deformable cross-sections at varying inner and outer radii, showing excellent accuracy and generalization. 🌐 You can check out the Open Access paper at: https://lnkd.in/d3aE7Wdi #machinelearning #computationalmechanics

📝🚨 Publication alert: In Iman Valizadeh's latest manuscript, which was just published in "Smart Materials & Methods", we explore the enhancement of energy absorption capabilities and stability in polymeric structures by blending different resin types, aiming to achieve optimal mechanical properties. For this purpose, a elasto-visco-plastic material model is experimentally characterized and applied within 3D finite element simulations. 👉🌐 Check-out the Open Access publication at: https://lnkd.in/eU3By7Ph #3Dprinting #polymers #materialmodeling

Our group showed a strong presence at the ICCE 2024 in Darmstadt this week with contributions by Juan C. Alzate Cobo, Yusuf Elbadry, M.Sc., Jonas Fey, Maximilian Kannapinn, Dominik K. Klein, Fabian Roth, Felix Rutsch, Mohammad shojaee, and Oliver Weeger 📑 It was a fruitful scientific event with inspiring invited presentations, among other by David Keyes 💽 👉 https://lnkd.in/dMyiyK84 #computationalengineering

Congratulations to our former guest researcher Pooyan Parnian and postdoc Mohammad shojaee on the publication of their manuscript "Multiscale modelling and characterisation of fused filament fabricated neat and graphene nanoplatelet reinforced G-polymers" in Progress in Additive Manufacturing! 🎉 📝 👉 In this work, a combined framework of computational multiscale modelling and experimental studies is applied to characterise and model the mechanical behaviour of composite and nanocomposite materials, fabricated using a 3D-printer by depositing G-polymer filaments. 🔍 Check out the manuscript for details (Open Access): https://lnkd.in/excF6bZc

Thanks to our co-authors Jan Niklas Fuhg and Reese Jones 🇺🇸, we extended the application of physics-augmented neural networks to thermo-hyperelastic material modeling. 🌡 💪 📈 📝 Check out the publication in the Journal of the Mechanics and Physics of Solids (https://lnkd.in/ejhnqa9H) or on the arXiv (https://lnkd.in/enqkZmWU) for further details and applications!