Data-Driven Design of Disordered Materials

With increasingly advanced manufacturing techniques, architected materials or metamaterials continue to gain popularity.

Researchers have produced ultrastrong, ultrastiff and ultralight metamaterials, whose anomalous properties emerge upon mechanical actuation. Their vast majority are designed with a periodic and regular lattice structure.

On the other hand, architected disordered materials have received little attention (e.g. earlier studies on foams) despite their robustness and flaw tolerance compared to regular lattice-based materials. This is largely due to their vast design space, which has been inaccessible with standard sampling techniques.

The aim of the project D4M (DEFORM) is the development of a novel rational framework for material design, that systematically exploits disorder, and is completely data-driven, and hence experience-free.

The framework relies on four synergistic elements: i) a unified network-theoretic representation of disordered material architectures, ii) the use of mechanics and complex networks as tools for evaluating design objectives, iii) the development of efficient graph machine learning techniques for executing the design, and iv) the practical implementation and validation of a suite of designs by additive manufacturing and testing.

By focusing on design objectives such as high energy absorption and tailored nonlinear deformation response, the proposed research is expected to have a diverse impact in the design of cellular, granular and fibrous materials with applications in biomechanics (prosthetics, orthotics, bioimplants) and the sports industry (protective equipment, clothing, shoes).

The implications of the proposed research stretch beyond these engineering applications and into the scientific understanding of complex biological systems such as bone and collagen.

This project will constitute a significant next step for the academic reintegration and professional establishment of the researcher in Europe.