About this Abstract |
Meeting |
2025 Annual International Solid Freeform Fabrication Symposium (SFF Symp 2025)
|
Symposium
|
2025 Annual International Solid Freeform Fabrication Symposium (SFF Symp 2025)
|
Presentation Title |
Inverse Design of Multi-Material Lattice Structures Using Graph Neural Networks |
Author(s) |
Xiaochen Yu, Ajit Panesar |
On-Site Speaker (Planned) |
Xiaochen Yu |
Abstract Scope |
Multi-material lattice structures offer unprecedented opportunities in designing multi-functional engineering applications. Using heat exchanger as an example, stainless steel provides structural strength while the incorporation of copper fins enables efficient heat dissipation. In this work, we propose a graph neural network (GNN) based inverse design framework to optimise multi-material truss-based lattices for competing mechanical and thermal objectives. Compared to voxel representation, a graph naturally fits truss-based lattice as geometric parameterisation and multi-material assignment could be effectively captured via node and edge features. The trained GNN models can serve either as surrogate models for the homogenised constitutive matrix, or directly inverse design the micro-scale unit cell based on target properties. Pareto of thermal-mechanical objectives will be provided, and case studies will be benchmarked against single-material solutions to illustrate the value-added of multi-material. |
Proceedings Inclusion? |
Planned: Post-meeting proceedings |