Computational discovery of emergent mechanical function in architected materials
OA Version
Citation
Abstract
Natural and engineered systems often derive their emergent mechanical functions from local spatial heterogeneity. However, the relationship between local geometry and global emergent mechanical behavior is still opaque. To explore the (geometric) structure-(mechanical) function relationship, computational methods and simulations are commonly favored over physical experiments due to the vast geometric parameter space available. Nevertheless, there is a lack of robust open-source computational tools available to numerically simulate and investigate this relationship. Motivated by these challenges, this dissertation aims to investigate non-trivial structure-function relationships by 1) developing open-source computational tools to aid numerical simulation and 2) use developed computational tools to uncover complex structure-function relationships in architected materials. The first part of this dissertation develops open-source computational tools for inference of structure-function relationships in architected materials with complex and irregular microstructure. We start by introducing FEniCS-arclength, a numerical continuation method equipped to deal with problems in solid mechanics exhibiting structural instabilities. Additionally, aside from the arc-length solver, we adopted data-driven methods in point-cloud processing to directly predict the buckling direction of columns with random geometry. Both the datasets and code in this part have been disseminated under open-source license to encourage researchers to improve and modify these methods to other problems in mechanics. The second part of the dissertation applies computational tools to explore the structure-function relationship of architectured materials. We first investigate how the distance-weighted graph shortest paths can predict the mechanical function of random fiber networks such as the emergence of load paths and the strain-stiffening effect. We end by investigating the role of geometry in mechanical information processing. By interpreting an elastic solid as an information encoder, we are able to quantify the amount of information transmitted from the applied load to discrete points in the solid. This work has potential to be extended to the design of architected materials endowed with “mechanical intelligence,” or the ability to sense and interact with its surroundings, a rapidly growing research direction in the mechanics community. Overall, this dissertation serves as a computational foundation for probing and engineering emergent mechanical function in complex architected materials.
Description
2026
License
Attribution-ShareAlike 4.0 International