Lattice Metamaterials with Mesoscale Motifs: Exploration of Property Charts by Bayesian Optimization

Roman Kulagin, Patrick Reiser, Kyryl Truskovskyi, Arnd Koeppe, Yan Beygelzimer, Yuri Estrin, Pascal Friederich, Peter Gumbsch

Research output: Contribution to journalArticlepeer-review

1 Citation (Scopus)

Abstract

Through the current work, the usefulness of the concept of architectured rod lattices based on unit cell motifs designed at mesoscale is demonstrated. Specifically, 2D triangular lattices with unit cells containing different numbers of rods are considered. Combinations of rods of two different types provide the lattices explored with a greater complexity and versatility. For mesocells with a large number of variable parameters, it is virtually impossible to calculate the entire set of the points mapping the material onto its property space, as the volume of calculations would be gigantic. The number of possible motifs increases exponentially with the number of rods. Herein, the lattice metamaterials with mesoscale motifs are investigated with the focus on their elastic properties by combining machine learning techniques (specifically, Bayesian optimization) with finite element computations. The proposed approach made it possible to construct property charts illustrating the evolution of the boundary of the elastic compliance tensor of lattice metamaterials with an increase in the number of rods of the mesocell when a full-factor experiment would not be possible.
Original languageEnglish
Article number2300048
JournalAdvanced Engineering Materials
Volume25
Issue number13
Early online date28 Mar 2023
DOIs
Publication statusPublished - Jul 2023

Fingerprint

Dive into the research topics of 'Lattice Metamaterials with Mesoscale Motifs: Exploration of Property Charts by Bayesian Optimization'. Together they form a unique fingerprint.

Cite this