Efficient two‐scale simulations of microstructured materials using deep material networks

Abstract

Deep material networks (DMN) are a promising piece of technology for accelerating concurrent multiscale simulations. DMNs are identified by linear elastic pre-computations on representative volume elements, and serve as high-fidelity surrogates for full-field simulations on microstructures with inelastic constituents. The offline training phase is independent of the online evaluation, such that a pre-trained DMN may be applied for varying material behavior of the constituents. In this contribution, we investigate a two-scale component simulation of industrial complexity accelerated by DMNs. To this end, a DMN is solved implicitly at every Gauss point to include the microstructure information into the macro simulation.

Material‐informed training of viscoelastic deep material networks

“…Solving the (implicit) DMN in every Gauss point of a macroscopic simulation gives rise to the FE-DMN method [7,8], which allows for incorporating microstructure information into a macroscopic simulation. Most notably, DMNs predict the effective properties for arbitrarily nonlinear constituents and complex three-dimensional microstructures with high accuracy and provide speed-ups of up to five orders of magnitude [8,9], allowing for conducting DMN-accelerated large-scale component simulations on commodity hardware. Extensions to treat thermomechanically coupled problems [1] and problems with fluctuating microstructure characteristics [8,10] are possible.…”