MS05 - Data-Driven Design of Materials

Keywords: Data-driven; Materials design; Inverse design; Computational homogenization; Microstructure characterization and reconstruction; Process-structure-property linkages; Multiscale simulation

Organizers:
Alexander Raßloff (1) – alexander.rassloff@tu-dresden.de
Alexandra Otto (1) – alexandra.otto@tu-dresden.de
Jan-Hendrik Bastek (2) – jbastek@ethz.ch
Li Zheng (2) – li.zheng@mavt.ethz.ch

Affiliations:
(1) Institute of Solid Mechanics, TU Dresden, Germany
(2) Mechanics and Materials Laboratory, ETH Zurich, Switzerland

Abstract:
The development of innovative, optimized materials, such as metals, polymers, composites, and architected materials, is crucial for scientific and industrial progress. Recent advances in machine learning and manufacturing enable the customization of material properties through inverse design, tailoring the chemical composition and the microstructure of classical materials as well as the internal structure of architected materials for a targeted property. The increasing availability of data, fueled by progress in digitization and high-throughput experiments, raises the interest and demand for data-driven techniques facilitating in-silico materials design.

Data-driven design of materials comprises a broad research field that spans from micro- to macroscale. Appropriate geometries that can be used, e.g., in computational homogenization approaches, are available through the characterization and reconstruction of real microstructures, enhancing the understanding and furthermore prediction of the interplay between effective properties and microstructural features of complex materials. The inverse design of suitable microstructures, tailored to exhibit specific target properties, represents a promising application for machine learning methods. Inverse problems and inverse design focus on minimizing various objectives subject to specific constraints, typically employing gradient-based optimization techniques. Machine learning similarly relies on gradient-based optimization, but it differs in its goal of generalizing beyond the specific problems it is trained on. Recent efforts have sought to exploit connections between these two methodologies across a variety of inverse problems in structural mechanics, including applications in meta- and architected materials as well as structural optimization.
Topics of interest covered within this mini-symposium include but are not limited to:
inverse design and optimization approaches for metals, polymers, composites, and architected materials,

  • microstructure characterization and reconstruction, e.g., 2D and 3D image-based methods,

  • data-driven multiscale simulations,

  • numerical and experimental analysis of new materials across scales,

  • techniques for exploration and inversion of process-structure-property linkages or part of it,

  • design approaches that account for crucial manufacturing constraints.