Teaching solid mechanics to artificial intelligence—a fast solver for heterogeneous materials
نویسندگان
چکیده
Abstract We propose a deep neural network (DNN) as fast surrogate model for local stress calculations in inhomogeneous non-linear materials. show that the DNN predicts stresses with 3.8% mean absolute percentage error (MAPE) case of heterogeneous elastic media and mechanical contrast up to factor 1.5 among neighboring domains, while performing 103 times faster than spectral solvers. The proves suited reproducing distribution geometries different from those used training. In elasto-plastic materials 4 yield adjacent regions, trained simulates micromechanics MAPE 6.4% one single forward evaluation network, without any iteration. results reveal an efficient approach solve problems, acceleration 8300 elastic-plastic compared typical
منابع مشابه
A dynamic lattice model for heterogeneous materials
In this paper, the mechanical behavior of three-phase inhomogeneous materials is modeled using the meso-scale model with lattice beams for static and dynamic analyses. The Timoshenko beam theory is applied instead of the classical Euler-Bernoulli beam theory and the mechanical properties of lattice beam connection are derived based on the continuum medium using the non-local continuum theory. T...
متن کاملRepresentation of Adsorption Data for the Case of Energetically Heterogeneous Solid Surfaces Using Artificial Neural Network
متن کامل
An iterative Riemann solver for systems of hyperbolic conservation laws, with application to hyperelastic solid mechanics
In this paper, we present a general iterative method for the solution of the Riemann problem for hyperbolic systems of PDEs. The method is based on the multiple shooting method for free boundary value problems. We demonstrate the method by solving one-dimensional Riemann problems for hyperelastic solid mechanics. Even for conditions representative of routine laboratory conditions and military b...
متن کاملa dynamic lattice model for heterogeneous materials
in this paper, the mechanical behavior of three-phase inhomogeneous materials is modeled using the meso-scale model with lattice beams for static and dynamic analyses. the timoshenko beam theory is applied instead of the classical euler-bernoulli beam theory and the mechanical properties of lattice beam connection are derived based on the continuum medium using the non-local continuum theory. t...
متن کاملAn Enhanced Viscoplastic Constitutive Model for Semi-Solid Materials to Analyze Shear Localization
Semi-solid materials undergo strain localization and shear band formation as a result of granular nature of semi-solid deformation. In the present study, to analyze the shear localization, a unified viscoplastic constitutive model was developed for the homogeneous flow. Then, a linearized analysis of the stability performed by examining the necessary condition for the perturbation growth. For t...
متن کاملذخیره در منابع من
با ذخیره ی این منبع در منابع من، دسترسی به آن را برای استفاده های بعدی آسان تر کنید
ژورنال
عنوان ژورنال: npj computational materials
سال: 2021
ISSN: ['2057-3960']
DOI: https://doi.org/10.1038/s41524-021-00571-z