2023 Annual International Solid Freeform Fabrication Symposium (SFF Symp 2023): Applications: Lattices and Cellular III
Program Organizers: Joseph Beaman, University of Texas at Austin
Tuesday 1:40 PM
August 15, 2023
Room: Salon B
Location: Hilton Austin
Session Chair: Vinayak Krishnamurthy, Texas A&M University
1:40 PM
Unweaving the Macrostructure of the Venus Flower Basket: From Bio-morphism to Biological Insight: Yash Mistry1; Dhruv Bhate1; 1Arizona State University
The Venus flower basket has been the subject of significant study due to its hierarchical, lattice-like design. While various hypotheses have been proposed for the basis of the macrostructure of the basket, most have not been validated. In this work, we examined the literature on the structure-function relationships proposed for the basket’s complex network of lattice members, including our own studies with x-ray microtomography. Three key design principles were abstracted into cylindrical designs similar to the actual biological and used to fabricate specimens using Selective Laser Sintering. These specimens were tested under three loading conditions: compression, bending, and torsion, and compared to a baseline design. Results show that each of the three design principles represents a significant, and sequential improvement in performance, mapping to growth stages in the Venus flower basket. This work argues for the use of additive manufacturing in generating analogs of natural structures to isolate structure-function relationships.
2:00 PM
Machine Learning Assisted Mechanical Metamaterial Design for Additive Manufacturing: Jier Wang1; Ajit Panesar1; 1Imperial College London
Metamaterials, widely studied for its counterintuitive property, are recognised to provide foundation for superior multiscale structural designs. However, current mechanical metamaterial design methods usually relay on performing sizing optimisations on predefined topology or implementing time-consuming inverse homogenisation methods. Machine Learning (ML), as a powerful self-learning tool, is considered to have the potential of discovering metamaterial topology and extending its property bounds. This work considers the use of conditional Generative Adversarial Networks (cGANs) to speed up the generation of new topologies for metamaterials. The generator in cGANs is trained to output metamaterial microstructural topologies based on the input condition, which is the desired property. Meanwhile, changing the noise input of cGANs is expected to produce different topologies, which will consequently lead to higher design diversity in metamaterial structural design. This work highlights the potential of data-driven approaches in Design for Additive Manufacturing (DfAM) as an alternative to the time-consuming, conventional methods.
2:20 PM
Generative Design of 2D Cellular Structures through Stochastic Modification of 2-Honeycombs: Matthew Ebert1; Vinayak Krishnamurthy1; 1Texas A&M University
We present a methodology for designing cellular structures through modification of a 2-honeycomb using probability density functions. A 2-honeycomb is a tessellation of the plane composed of a single space-filling prototile which can be created using Voronoi tessellation of sites created using Wallpaper symmetries. While this offers an intuitive method for obtaining interesting mechanical behaviors, the design space is limited by the uniformity of 2-honeycombs. Our method introduces an additional step to this process of structure generation using point-density functions which allow for removal of Voronoi sites resulting in explicit control of local stiffness. Since the site removal is randomized, we end up with completely different structures each time the removal is applied. Through FEA simulations, we show that generated structures created with the same parameters have similar mechanical properties. These structures can be used to modify the infill structure for a 3D printed part to minimize the material usage.
2:40 PM
Design of Three-dimensional Complex Truss Metamaterials with Graph Neural Networks: Marco Maurizi1; Desheng Yao1; Xiaoyu (Rayne) Zheng1; 1University of California Berkeley
The rapid development of additive manufacturing technologies has enabled the fabrication of truss metamaterials, i.e., a novel class of lightweight-yet-strong materials with engineered complex hierarchical structures. Manipulating the architecture over chemical composition dramatically expands the achievable materials design space, allowing to largely control the mechanical response of metamaterials. Despite the great advances made in this area, designing three-dimensional (3D) truss metamaterials under complex or extreme conditions with programmable response is still a challenge. Here we propose a paradigm to design 3D truss metamaterials with complex programmable mechanical responses both under quasi-static and dynamic loading based on graph neural networks (GNNs). By leveraging the ability of our GNN-based model to accurately predict the mechanical response across multiple orders of magnitude, we inverse design truss metamaterials for compressive loading up to 50 % of strain and dynamic transmissibility with desired band gaps, opening the way for full materials design freedom.
3:00 PM
Design of Spatially Varying Orientation Lattice Structures using Triply Periodic Minimal Surfaces: Chongyi Wei1; Douglas Smith1; 1Baylor University
Interest continues to grow for lattice structures produced by additive manufacturing methods that are described by triply periodic minimal surface (TPMS). Tunable parameters that define the TPMS provide unique design flexibility where prior research has focused on designing hybrid or functionally graded TPMS structures. In this paper, a new strategy is proposed to include an orientation angle and volume fraction of each lattice cell simultaneously when defining structures derived from TPMS. The algorithm iteratively solves an underlying partial differential equation with the finite difference method to obtain a smooth, continuous lattice structure with a spatially varying orientation angle. The resulting lattice structure can be combined with other types of TPMS models using Gaussian radial basis or sigmoid functions to achieve multi-TPMS lattice designs. The spatially varying lattice structures can also take the advantage of the directional effective modulus of TPMS to improve the strength the performance of the lattice design.
3:20 PM
Improving the Mechanical Response of the IWP Exo-skeletal Lattice Through Shape Optimization: Joseph Fisher1; Simon Miller1; Joseph Bartolai1; Timothy Simpson1; 1Pennsylvania State University
Triply Periodic Minimal Surfaces have been identified as good candidates for the generation of lattice structures produced with additive manufacturing. These TPMS-based lattice structures avoid sharp features that are characteristic of strut-based lattice structures because of their constant zero mean curvature. Although studies have explored part-scale optimization using TPMS-based lattice structures, they have only varied the volume fraction by changing the level set in the approximate surface equations. By defining new parameterizations in the approximate surface equation, we can redistribute volume within the lattice structure at any volume fraction. In this paper, we introduce an approach for optimization of this new parameterization of TPMS equations using the Borg multi-objective evolutionary algorithm. We demonstrate this framework on the IWP exo-skeletal lattice under uniaxial compression. A relationship between the new parameters and the level set is derived for designs on the Pareto frontier of the optimized IWP TPxS. The performance of the Pareto optimal designs and the efficacy of the optimization approach are shown by comparing to the standard IWP lattice and four other lattices that share the same topology. The optimized designs are implemented and shared in custom nTopology blocks.
3:40 PM
Hybrid Geometry/Property Autoencoders for Multi-Lattice Transitions: Martha Baldwin1; Nicholas Meisel2; Christopher McComb1; 1Carnegie Mellon University; 2The Pennsylvania State University
Additive manufacturing has revolutionized structural optimization. This may be achieved through multi-lattice structures, which have emerged as promising candidates for impact reduction and other applications. However, the performance of these structures relies on the details of mesostructural elements. Many current approaches use data-driven approaches to generate multi-lattice transition regions but focus largely on the geometry of the mesostructure. Therefore, it remains unclear whether the integration of functional performance factors in the generation of multi-lattice interpolations is beneficial beyond geometry alone. To address this issue, this work implements and evaluates a VAE to create a hybrid geometric/functional latent space for generating multi-lattice transition regions. This approach allows us to test the performance of the proposed method against geometry and property defined latent spaces. Our research aims to determine whether it is necessary to incorporate physical properties into the geometrically defined latent spaces to achieve robust multi-lattice structures.
4:00 PM
Tailoring Mechanical Behaviors through Programmable Lattice Structures: Yi Sheng Liao1; Xinyi Xiao1; 1Miami University
Lattice structures are cellular structures composed of periodically arranged unit cells with a high strength-to-weight ratio and remarkable properties like lightweight, energy absorption, and vibration resistance. Owing to their superior mechanical performance and properties. With the development of additive manufacturing technology, more complicated shapes and architectures of prototypes can be made for end-user products.The paper discusses the advantages and disadvantages of lattice structures and suggests ways to overcome limitations. The study uses seven structures, including BCC, Octahedron, and Cross-cube, as well as three structures composed of two of them and one structure composed of three, to analyze the relationship among the structures. The compression strength and physic-based finite element simulations were performed to analyze the quantitative relationship between the structures and the mechanical response. The results can help in the design and improvement of products made with lattice structures.