A deep material network for multiscale topology learning and accelerated nonlinear modeling of heterogeneous materials

Abstract

In this paper, a new data-driven multiscale material modeling method, which we refer to as deep material network, is developed based on mechanistic homogenization theory of representative volume element (RVE) and advanced machine learning techniques. We propose to use a collection of connected mechanistic building blocks with analytical homogenization solutions which avoids the loss of essential physics in generic neural networks, and this concept is demonstrated for 2-dimensional RVE problems and network depth up to 7. Based on linear elastic RVE data from offline direct numerical simulations, the material network can be effectively trained using stochastic gradient descent with backpropagation algorithm, enhanced by model compression methods. Importantly, the trained network is valid for any local material laws without the need for additional calibration or micromechanics assumption. Its extrapolations to unknown material and loading spaces for a wide range of problems are validated through numerical experiments, including linear elasticity with high contrast of phase properties, nonlinear history-dependent plasticity and finite-strain hyperelasticity under large deformations. By discovering a proper topological representation of RVE with fewer degrees of freedom, this intelligent material model is believed to open new possibilities of high-fidelity efficient concurrent simulations for a large-scale heterogeneous structure. It also provides a mechanistic understanding of structure-property relations across material length scales and enables the development of parameterized microstructural database for material design and manufacturing.

33 pages, 19 figures

References in corpus (4)

• A numerical method for computing the overall response of nonlinear composites with complex microstructure
• The Loss Surfaces of Multilayer Networks
• A partitioned model order reduction approach to rationalise computational expenses in multiscale fracture mechanics
• Finite strain FFT-based non-linear solvers made simple

Cited by in corpus (35)

• Geometric deep learning for computational mechanics Part I: Anisotropic Hyperelasticity
• Exploring the 3D architectures of deep material network in data-driven multiscale mechanics
• A machine learning based plasticity model using proper orthogonal decomposition
• Learning Constitutive Relations using Symmetric Positive Definite Neural Networks
• Deep Learning in Deterministic Computational Mechanics
• Learning Deep Implicit Fourier Neural Operators (IFNOs) with Applications to Heterogeneous Material Modeling
• FE${}^{\text{ANN}}$ $-$ An efficient data-driven multiscale approach based on physics-constrained neural networks and automated data mining
• A hybrid approach to simulate the homogenized irreversible elastic-plastic deformations and damage of foams by neural networks
• FEA-Net: A Physics-guided Data-driven Model for Efficient Mechanical Response Prediction
• Physically recurrent neural networks for path-dependent heterogeneous materials: embedding constitutive models in a data-driven surrogate
• An FE-DMN method for the multiscale analysis of fiber reinforced plastic components
• A comparative study on different neural network architectures to model inelasticity
• Deep autoencoders for physics-constrained data-driven nonlinear materials modeling
• Accelerated multiscale mechanics modeling in a deep learning framework
• Model-data-driven constitutive responses: application to a multiscale computational framework
• Deep material network with cohesive layers: Multi-stage training and interfacial failure analysis
• Evolution TANN and the identification of internal variables and evolution equations in solid mechanics
• On-the-fly construction of surrogate constitutive models for concurrent multiscale mechanical analysis through probabilistic machine learning
• Finite Element Network Analysis: A Machine Learning based Computational Framework for the Simulation of Physical Systems
• LS-DYNA Machine Learning-based Multiscale Method for Nonlinear Modeling of Short Fiber-Reinforced Composites
• Two-stage 2D-to-3D reconstruction of realistic microstructures: Implementation and numerical validation by effective properties
• Cell division in deep material networks applied to multiscale strain localization modeling
• A probabilistic virtual process chain to quantify process-induced uncertainties in Sheet Molding Compounds
• Intelligent multiscale simulation based on process-guided composite database
• Enhancing Mechanical Metamodels with a Generative Model-Based Augmented Training Dataset
• Machine learning for metal additive manufacturing: Predicting temperature and melt pool fluid dynamics using physics-informed neural networks
• Cooperative data-driven modeling
• Micromechanics-Informed Parametric Deep Material Network for Physics Behavior Prediction of Heterogeneous Materials with a Varying Morphology
• Multiscale Reduced-Order Modeling of a Titanium Skin Panel Subjected to Thermo-Mechanical Loading
• Emergent properties and the multiscale characterization challenge in condensed matter, from crystals to complex materials: a Review
• A learning-based multiscale method and its application to inelastic impact problems
• Automatically Differentiable Model Updating (ADiMU): conventional, hybrid, and neural network material model discovery including history-dependency
• How machine learning can help the design and analysis of composite materials and structures?
• Scalable Gaussian Processes for Data-Driven Design using Big Data with Categorical Factors
• Empirical Hyper Element Integration Method (EHEIM) with Unified Integration Criteria for Efficient Hyper Reduced FE${}^2$ Simulations