Micromechanics-based surrogate models for the response of composites: A critical comparison between a classical mesoscale constitutive model, hyper-reduction and neural networks

Although being a popular approach for the modeling of laminated composites, mesoscale constitutive models often struggle to represent material response for arbitrary load cases. A better alternative in terms of accuracy is to use the FE2 technique to upscale microscopic material behavior without loss of generality, but the associated computational effort can be extreme. It is therefore interesting to explore alternative surrogate modeling strategies that maintain as much of the fidelity of FE2 as possible while still being computationally efficient. In this work, three surrogate modeling approaches are compared in terms of accuracy, efficiency and calibration effort: the state-of-the-art mesoscopic plasticity model by Vogler et al. (Vogler et al., 2013), regularized feed-forward neural networks and hyper-reduced-order models obtained by combining the Proper Orthogonal Decomposition (POD) and Empirical Cubature Method (ECM) techniques. Training datasets are obtained from a Representative Volume Element (RVE) model of the composite microstructure with a number of randomly-distributed linear-elastic fibers surrounded by a matrix with pressure-dependent plasticity. The approaches are evaluated with a comprehensive set of numerical tests comprising pure stress cases and three different stress combinations relevant in the design of laminated composites. The models are assessed on their ability to accurately reproduce the training cases as well as on how well they are able to predict unseen stress combinations. Gains in execution time are compared by using the trained surrogates in the FE2 model of an interlaminar shear test.

[1]  P Kerfriden,et al.  Bridging Proper Orthogonal Decomposition methods and augmented Newton-Krylov algorithms: an adaptive model order reduction for highly nonlinear mechanical problems. , 2011, Computer methods in applied mechanics and engineering.

[2]  Pascal Vincent,et al.  Representation Learning: A Review and New Perspectives , 2012, IEEE Transactions on Pattern Analysis and Machine Intelligence.

[3]  Radford M. Neal Pattern Recognition and Machine Learning , 2007, Technometrics.

[4]  A. Simone,et al.  POD-DEIM model order reduction for strain softening viscoplasticity , 2017 .

[5]  Abedin I. Gagani,et al.  Micromechanical modeling of anisotropic water diffusion in glass fiber epoxy reinforced composites , 2018 .

[6]  B. A. Le,et al.  Computational homogenization of nonlinear elastic materials using neural networks , 2015 .

[7]  S. A. Grammatikos,et al.  On the response to hygrothermal aging of pultruded FRPs used in the civil engineering sector , 2016 .

[8]  Vinh Phu Nguyen,et al.  Computational homogenization for multiscale crack modeling. Implementational and computational aspects , 2012 .

[9]  V. G. Kouznetsova,et al.  Multi-scale computational homogenization: Trends and challenges , 2010, J. Comput. Appl. Math..

[10]  M. Caicedo,et al.  Dimensional hyper-reduction of nonlinear finite element models via empirical cubature , 2017 .

[11]  George Cybenko,et al.  Approximation by superpositions of a sigmoidal function , 1989, Math. Control. Signals Syst..

[13]  L. Sluys,et al.  Continuum Models for the Analysis of Progressive Failure in Composite Laminates , 2009 .

[14]  Fpt Frank Baaijens,et al.  An approach to micro-macro modeling of heterogeneous materials , 2001 .

[15]  Leo Breiman,et al.  Bagging Predictors , 1996, Machine Learning.

[16]  Bernhard A. Schrefler,et al.  Artificial Neural Networks in numerical modelling of composites , 2009 .

[17]  P. Budrugeac,et al.  Accelerated thermal aging of glass-reinforced epoxy resin under oxygen pressure , 1991 .

[18]  Qingda Yang,et al.  In Quest of Virtual Tests for Structural Composites , 2006, Science.

[19]  L. J. Sluys,et al.  A combined experimental/numerical investigation on hygrothermal aging of fiber-reinforced composites , 2019, European Journal of Mechanics - A/Solids.

[20]  Jimmy Ba,et al.  Adam: A Method for Stochastic Optimization , 2014, ICLR.

[21]  Nitish Srivastava,et al.  Dropout: a simple way to prevent neural networks from overfitting , 2014, J. Mach. Learn. Res..

[22]  António R. Melro,et al.  Micromechanical analysis of polymer composites reinforced by unidirectional fibres: Part I – Constitutive modelling , 2013 .

[23]  A. Nouy,et al.  Model order reduction based on proper generalized decomposition for the propagation of uncertainties in structural dynamics , 2012 .

[24]  Amir F. Atiya,et al.  Comprehensive Review of Neural Network-Based Prediction Intervals and New Advances , 2011, IEEE Transactions on Neural Networks.

[25]  J. Schröder,et al.  Computational micro-macro transitions and overall moduli in the analysis of polycrystals at large strains , 1999 .

[26]  F. Ghavamian,et al.  Accelerating multiscale finite element simulations of history-dependent materials using a recurrent neural network , 2019 .

[27]  Yoshua Bengio,et al.  Understanding the difficulty of training deep feedforward neural networks , 2010, AISTATS.

[28]  James H. Garrett,et al.  Knowledge-Based Modeling of Material Behavior with Neural Networks , 1992 .

[29]  Jamshid Ghaboussi,et al.  Autoprogressive training of neural network constitutive models , 1998 .

[30]  Zeliang Liu,et al.  A Deep Material Network for Multiscale Topology Learning and Accelerated Nonlinear Modeling of Heterogeneous Materials , 2018, Computer Methods in Applied Mechanics and Engineering.

[31]  Renaud Gutkin,et al.  A micromechanical model for kink-band formation: Part I — experimental study and numerical modelling , 2009 .

[32]  Frans P. van der Meer,et al.  Micromechanical validation of a mesomodel for plasticity in composites , 2016 .

[33]  Robert K. Goldberg,et al.  Finite-strain elasto-viscoplastic behavior of an epoxy resin: Experiments and modeling in the glassy regime , 2014 .

[34]  Marc G D Geers,et al.  Integration efficiency for model reduction in micro-mechanical analyses , 2017, Computational Mechanics.

[35]  Danny C. Sorensen,et al.  Nonlinear Model Reduction via Discrete Empirical Interpolation , 2010, SIAM J. Sci. Comput..

[36]  Julien Yvonnet,et al.  A data-driven computational homogenization method based on neural networks for the nonlinear anisotropic electrical response of graphene/polymer nanocomposites , 2018, Computational Mechanics.

[37]  Neil Genzlinger A. and Q , 2006 .

[38]  Issam Doghri,et al.  A thermodynamically-based constitutive model for thermoplastic polymers coupling viscoelasticity, viscoplasticity and ductile damage , 2014 .

[39]  David Amsallem,et al.  Automatised selection of load paths to construct reduced-order models in computational damage micromechanics: from dissipation-driven random selection to Bayesian optimization , 2016, Computational Mechanics.

[40]  T. Westphal,et al.  Micro-mechanical fatigue modelling of unidirectional glass fibre reinforced polymer composites , 2013 .

[41]  I.B.C.M. Rocha,et al.  Efficient micromechanical analysis of fiber-reinforced composites subjected to cyclic loading through time homogenization and reduced-order modeling , 2019, Computer Methods in Applied Mechanics and Engineering.

[42]  Pedro P. Camanho,et al.  A damage model for the simulation of delamination in advanced composites under variable-mode loading , 2006 .

[43]  F. Naya,et al.  Computational micromechanics of the transverse and shear behavior of unidirectional fiber reinforced polymers including environmental effects , 2017 .

[44]  Raimund Rolfes,et al.  Modeling the inelastic deformation and fracture of polymer composites – Part I: Plasticity model , 2013 .

[45]  Giulio Alfano,et al.  Combining interface damage and friction in a cohesive‐zone model , 2006 .