Triply periodic minimal surfaces (TPMS) are a class of metamaterials with a variety of applications and well-known primitives. We present a new method for discovering novel microscale TPMS structures with exceptional energy-dissipation capabilities, achieving double the energy absorption of the best existing TPMS primitive structure. Our approach employs a parametric representation, allowing seamless interpolation between structures and representing a rich TPMS design space. We show that simulations are intractable for optimizing microscale hyperelastic structures, and instead propose a sample-efficient computational strategy for rapidly discovering structures with extreme energy dissipation using limited amounts of empirical data from 3D-printed and tested microscale metamaterials. This strategy ensures high-fidelity results but involves time-consuming 3D printing and testing. To address this, we leverage an uncertainty-aware Deep Ensembles model to predict microstructure behaviors and identify which structures to 3D-print and test next. We iteratively refine our model through batch Bayesian optimization, selecting structures for fabrication that maximize exploration of the performance space and exploitation of our energy-dissipation objective. Using our method, we produce the first open-source dataset of hyperelastic microscale TPMS structures, including a set of novel structures that demonstrate extreme energy dissipation capabilities. We show several potential applications of these structures in protective equipment and bone implants.
View on arXiv