Page 39 - ESAM-1-1

Engineering Science in
Additive Manufacturing Gen-AI for lattice structure design

A

B

C

D

Figure 2. Gen-AI for lattice structure generation: an example using diffusion models. Reproduced with permission from Jadhav et al. 35
Abbreviation: Gen-AI: Generative artificial intelligence.

review articles that specifically explore the role of Gen-AI 2. Generative AI for inverse design of lattice
in lattice design. To address this gap, the present study structures
aims to provide a systematic and in-depth discussion on
the application of Gen-AI techniques in lattice structure Inverse design is a critical approach in lattice structure
design for AM. Key focus areas include (1) inverse development, aiming to derive structural configurations
design of lattice structures, exploring how Gen-AI can that precisely meet predefined mechanical performance
49
enable the automated generation of lattice configurations targets. Traditional inverse design approaches rely heavily
tailored to meet specific functional requirements or target on iterative processes, often leading to high computational
50
properties; (2) performance optimization, investigating costs and limited exploration of design spaces. Recent
the role of Gen-AI in optimizing lattice geometries for advancements in Gen-AI offer innovative solutions
mechanical, thermal, and other functional performances, to these challenges, enabling automated, efficient, and
ensuring the balance between material efficiency and precise inverse design processes. However, automatically
structural integrity; (3) process-aware design, addressing generating efficient and reliable structural designs under
how Gen-AI can account for AM process constraints, complex performance constraints remains a significant
such as build orientation, support strategies, and residual challenge in this field.
stress mitigation, to ensure manufacturability and reduce To address the complexities of inverse design in frame
post-processing efforts; and (4) simulation acceleration, and lattice structures, Yang et al. proposed GraphDGM,
51
discussing how Gen-AI can accelerate the simulation a graph-based diffusion-generative multiobjective design
processes for predicting mechanical behaviors, failure approach for the inverse design of frame and lattice
modes, or thermal responses of complex lattice structures, structures under multiple constraints. The method
significantly reducing computational costs and design integrates a graph-based generative model with denoising
cycles. By critically reviewing these aspects, this study diffusion probabilistic models (DDPM) and an attention
seeks to offer valuable insights into current advancements, mechanism to achieve efficient inverse design. They
challenges, and future directions for leveraging Gen-AI in constructed datasets using the FEM for various structures,
the design and optimization of lattice structures. including vehicle skeletons and beams, to train the model.

Volume 1 Issue 1 (2025) 4 doi: 10.36922/ESAM025110006