[en] This contribution discusses surrogate models that emulate the solution field(s) in the entire simulation domain. The surrogate uses the most characteristic modes of the solution field(s), in combination with neural networks to emulate the coefficients of each mode. This type of surrogate is well known to rapidly emulate flow simulations, but rather new for simulations of elastoplastic solids. The surrogate avoids the iterative process of constructing and solving the linearized governing equations of rate-independent elastoplasticity, as necessary for direct numerical simulations or (hyper-)reduced-order-models. Instead, the new plastic variables are computed only once per increment, resulting in substantial time savings. The surrogate uses a recurrent neural network to treat the path dependency of rate-independent elastoplasticity within the neural network itself. Because only a few of these surrogates have been developed for elastoplastic simulations, their potential and limitations are not yet well studied. The aim of this contribution is to shed more light on their numerical capabilities in the context of elastoplasticity. Although more widely applicable, the investigation focuses on a representative volume element, because these surrogates have the ability to both emulate the macroscale stress-deformation relation (which drives the multiscale simulation), as well as to recover all microstructural quantities within each representative volume element.
VIJAYARAGHAVAN, Soumianarayanan ; University of Luxembourg > Faculty of Science, Technology and Medicine > Department of Engineering > Team Stéphane BORDAS
Wu, L; University of Liege, Bât. B52/3 Computational & Multiscale Mechanics of Materials, Quartier Polytech 1, allée de la Découverte 9, 4000, Liège, Belgium
Noels, L; University of Liege, Bât. B52/3 Computational & Multiscale Mechanics of Materials, Quartier Polytech 1, allée de la Découverte 9, 4000, Liège, Belgium
BORDAS, Stéphane ; University of Luxembourg > Faculty of Science, Technology and Medicine (FSTM) > Department of Engineering (DoE)
Natarajan, S; Department of Mechanical Engineering, Indian Institute of Technology, Madras, Chennai, 600036, India
BEEX, Lars ; University of Luxembourg > Faculty of Science, Technology and Medicine (FSTM) > Department of Engineering (DoE)
Co-auteurs externes :
yes
Langue du document :
Anglais
Titre :
A data-driven reduced-order surrogate model for entire elastoplastic simulations applied to representative volume elements.
Date de publication/diffusion :
07 août 2023
Titre du périodique :
Scientific Reports
eISSN :
2045-2322
Maison d'édition :
Nature Research, England
Volume/Tome :
13
Fascicule/Saison :
1
Pagination :
12781
Peer reviewed :
Peer reviewed vérifié par ORBi
Focus Area :
Computational Sciences
Objectif de développement durable (ODD) :
9. Industrie, innovation et infrastructure 12. Consommation et production responsables
Multiscale modelling of lightweight metallic materials accounting for variability of geometrical and material properties
N° du Fonds :
INTER/FNRS/15/11019432/En-LightenIt/Bordas
Subventionnement (détails) :
S.V., L.B. and S.B. gratefully acknowledge the support of Fonds National de la Recherche Luxembourg (FNR) grant INTER/FNRS/15/11019432/En-LightenIt/Bordas. This project has received funding from the H2020-EU.1.2.1.-FET Open Programme project MOAMMM under grant No. 862015 and the EU’ s H2020 project DRIVEN under grant No. 811099. Computational resources have been provided by the HPC of the University of Luxembourg and HPC CÉCI (funded by FRS-FNRS).
Liu, Z., Wu, C. & Koishi, M. A deep material network for multiscale topology learning and accelerated nonlinear modeling of heterogeneous materials. Comput. Methods Appl. Mech. Eng. 345, 1138–1168 (2019). DOI: 10.1016/j.cma.2018.09.020
Avdelas, A., Panagiotopoulos, P. & Kortesis, S. Neural networks for computing in the elastoplastic analysis of structures. Meccanica 30, 1–15 (1995). DOI: 10.1007/BF00987122
Stavroulakis, G., Avdelas, A., Abdalla, K. M. & Panagiotopoulos, P. A neural network approach to the modelling, calculation and identification of semi-rigid connections in steel structures. J. Construct. Steel Res. 44(1–2), 91–105 (1997). DOI: 10.1016/S0143-974X(97)00039-4
Daoheng, S., Qiao, H. & Hao, X. A neurocomputing model for the elastoplasticity. Comput. Methods Appl. Mech. Eng. 182, 177–186 (2000). DOI: 10.1016/S0045-7825(99)00091-2
Waszczyszyn, Z. & Ziemiański, L. Neural networks in mechanics of structures and materials-new results and prospects of applications. Comput. Struct. 79, 2261–2276 (2001). DOI: 10.1016/S0045-7949(01)00083-9
Eghbalian, M., Pouragha, M., & Wan, R. A physics-informed deep neural network for surrogate modeling in classical elasto-plasticity. arXiv:2204.12088, (2022)
Huang, D., Fuhg, J., Weissenfels, C. & Wriggers, P. A machine learning based plasticity model using proper orthogonal decomposition. Comput. Methods Appl. Mech. Eng. 365, 113008 (2020). DOI: 10.1016/j.cma.2020.113008
Ghavamian, F. & Simone, A. Accelerating multiscale finite element simulations of history-dependent materials using a recurrent neural network Comput. Methods Appl. Mech. Eng. 357, 112594 (2019). DOI: 10.1016/j.cma.2019.112594
Wu, L., Nguyen, V., Kilingar, N. & Noels, L. A recurrent neural network-accelerated multi-scale model for elasto-plastic heterogeneous materials subjected to random cyclic and non-proportional loading paths. Comput. Methods Appl. Mech. Eng. 369, 113234 (2020). DOI: 10.1016/j.cma.2020.113234
Koeppe, A., Bamer, F. & Markert, B. An intelligent nonlinear meta element for elastoplastic continua: deep learning using a new time-distributed residual U-Net architecture. Comput. Methods Appl. Mech. Eng. 366, 113088 (2020). DOI: 10.1016/j.cma.2020.113088
Heider, Y., Wang, K. & Sun, W. SO(3)-invariance of informed-graph-based deep neural network for anisotropic elastoplastic materials. Comput. Methods Appl. Mech. Eng. 363, 112875 (2020). DOI: 10.1016/j.cma.2020.112875
He, X. & Chen, J. Thermodynamically consistent machine-learned internal state variable approach for data-driven modeling of path-dependent materials. Comput. Methods Appl. Mech. Eng. 402, 115348 (2022). DOI: 10.1016/j.cma.2022.115348
Bonatti, C. & Mohr, D. On the importance of self-consistency in recurrent neural network models representing elasto-plastic solids. J. Mech. Phys. Solids 158, 104697 (2022). DOI: 10.1016/j.jmps.2021.104697
M. Maia, I. Rocha, P. Kerfriden, & F. Van der Meer. Physically recurrent neural networks for path-dependent heterogeneous materials: Embedding constitutive models in a data-driven surrogate. arXiv:2209.07320, (2022)
B. Liu, M. Trautner, A. Stuart, & K. Bhattacharya. Learning macroscopic internal variables and history dependence from microscopic models. arXiv:2210.17443, (2022)
Bhattacharya, K., Liu, B., Stuart, A., & Trautner, M. Learning Markovian Homogenized Models in Viscoelasticity arXiv:2205.14139, (2022).
Wu, L. & Noels, L. Recurrent neural networks (RNNs) with dimensionality reduction and break down in computational mechanics; application to multi-scale localization step. Comput. Methods Appl. Mech. Eng. 390, 114476 (2022). DOI: 10.1016/j.cma.2021.114476
Koronaki, E. et al. Classification of states and model order reduction of large scale chemical vapor deposition processes with solution multiplicity. Comput. Chem. Eng. 121, 148–157 (2018). DOI: 10.1016/j.compchemeng.2018.08.023
Cao, B., Obel, M., Freitag, S., Mark, P. & Meschke, G. Artificial neural network surrogate modelling for real-time predictions and control of building damage during mechanised tunnelling. Adv. Eng. Softw. 149, 102869 (2020). DOI: 10.1016/j.advengsoft.2020.102869
Kapteyn, M., Knezevic, D., Huynh, D., Tran, M. & Willcox, K. Data-driven physics-based digital twins via a library of component-based reduced-order models. Int. J. Numer. Methods Eng. 123(13), 2986–3003 (2020). DOI: 10.1002/nme.6423
Hijazi, S., Stabile, G., Mola, A. & Rozza, G. Data-driven POD-Galerkin reduced order model for turbulent flows. J. Comput. Phys. 416, 109513 (2020). DOI: 10.1016/j.jcp.2020.109513
Kumar, A., Hu, R. & Walsh, S. Development of reduced order hydro-mechanical models of fractured media. Rock Mech. Rock Eng. 55, 235–248 (2022). DOI: 10.1007/s00603-021-02668-9
Torregrosa, A., Gil, A., Quintero, P. & Cremades, A. A reduced order model based on artificial neural networks for nonlinear aeroelastic phenomena and application to composite material beams. Compos. Struct. 295, 115845 (2022). DOI: 10.1016/j.compstruct.2022.115845
Cao, B., Freitag, S. & Meschke, G. A hybrid RNN-GPOD surrogate model for real-time settlement predictions in mechanised tunnelling (Model. Simul. Eng. Sci, Adv, 2016). 10.1186/s40323-016-0057-9. DOI: 10.1186/s40323-016-0057-9
Freitag, S., Cao, B., Ninić, J. & Meschke, G. Recurrent neural networks and proper orthogonal decomposition with interval data for real-time predictions of mechanised tunnelling processes. Comput. Struct. 207, 258–273 (2018). DOI: 10.1016/j.compstruc.2017.03.020
Im, S., Lee, J. & Cho, M. Surrogate modeling of elasto-plastic problems via long short-term memory neural networks and proper orthogonal decomposition. Comput. Methods Appl. Mech. Eng. 385, 114030 (2021). DOI: 10.1016/j.cma.2021.114030
Kerfriden, P., Goury, O., Rabczuk, T. & Bordas, S. A partitioned model order reduction approach to rationalise computational expenses in nonlinear fracture mechanics. Comput. Methods Appl. Mech. Eng. 256, 169–188 (2013). DOI: 10.1016/j.cma.2012.12.004
Casenave, F., Akkari, N., Bordeu, F., Rey, C. & Ryckelynck, D. A nonintrusive distributed reduced-order modeling framework for nonlinear structural mechanics-application to elastoviscoplastic computations. Int. J. Numer. Methods Eng. 121(1), 32–53 (2020). DOI: 10.1002/nme.6187
Doškář, M., Zeman, J., Krysl, P. & Novák, J. Microstructure-informed reduced modes synthesized with Wang tiles and the Generalized Finite Element Method. Comput. Mech. 68, 233–253 (2021). DOI: 10.1007/s00466-021-02028-y
Zhang, Y., Combescure, A. & Gravouil, A. Efficient hyper reduced-order model (HROM) for parametric studies of the 3D thermo-elasto-plastic calculation. Finite Elem. Anal. Des. 102–103, 37–51 (2015). DOI: 10.1016/j.finel.2015.04.009
Daniel, T., Casenave, F., Nissrine, A. & Ryckelynck, D. Model order reduction assisted by deep neural networks (ROM-net) (Model. Simul. Eng. Sci, Adv, 2020). 10.1186/s40323-020-00153-6. DOI: 10.1186/s40323-020-00153-6
Nagel, F., Stascheit, J. & Meschke, G. Process-oriented numerical simulation of shield-supported tunnelling in soft soils. Geomech. Tunn. 3(3), 268–282 (2010). DOI: 10.1002/geot.201000024
Kouznetsova, V., Brekelmans, W. & Geers, M. An approach to micro-macro modeling of heterogeneous materials. Comput. Mech. 27(1), 37–48 (2001). DOI: 10.1007/s004660000212
Geers, M., Kouznetsova, V. G. & Brekelmans, W. Multi-scale computational homogenization: Trends and challenges. J. Comput. Appl. Math. 234(7), 2175–2182 (2010). DOI: 10.1016/j.cam.2009.08.077
Akbari Rahimabadi, A., Kerfriden, P. & Bordas, S. Scale selection in nonlinear fracture mechanics of heterogeneous materials. Philos. Mag. 95(2–30), 3328–3347 (2015). DOI: 10.1080/14786435.2015.1061716
Van-Dung, N., Wu, L. & Noels, L. Unified treatment of microscopic boundary conditions and efficient algorithms for estimating tangent operators of the homogenized behavior in the computational homogenization method. Comput. Mech. 59(3), 483–505 (2017). DOI: 10.1007/s00466-016-1358-z
Ryckelynck, D. A priori hyperreduction method: An adaptive approach. J. Comput. Phys. 202, 346–366 (2005). DOI: 10.1016/j.jcp.2004.07.015
Chaturantabut, S. & Sorensen, D. Nonlinear model reduction via discrete empirical interpolation. SIAM J. Sci. Comput. 32, 2737–2764 (2010). DOI: 10.1137/090766498
Hale, J., Schenone, E., Baroli, D., Beex, L. & Bordas, S. A hyper-reduction method using adaptivity to cut the assembly costs of reduced order models. Comput. Methods Appl. Mech. Eng. 380, 113723 (2021). DOI: 10.1016/j.cma.2021.113723
D. Kingma & J. Lei Ba. Adam: a method for stochastic optimization. arXiv:1412.6980, (2015)
Varrette, S., Bouvry, P., Cartiaux, H. & Georgatos, F. Management of an academic HPC cluster: The UL experience. In HPCS p 959–967 (IEEE, Bologna, 2014).
