Generative design strategies for additive manufacturing of lattice structures: A review
Fuyuan Liu, Yitian Shao, Min Chen
Abstract
Additive manufacturing (AM) enables the creation of architected lattices with unprecedented geometric freedom, effectively reshaping the modern design paradigm. This review positions generative design (GD) as a process-aware and human-in-the-loop framework for exploring high-dimensional design spaces under coupled functional requirements and manufacturable constraints. We define a unified architecture that integrates geometry generation, performance evaluation, and cross-scale optimization, spanning from unit-cell synthesis to the integration of conformal and graded lattices. The review highlights two complementary directions: i) physical model-based methods, which offer interpretable, constraint-faithful guidance from unit-cell archetypes (strut-, plate-, and TPMS-based) to graded or conformal layouts through topology optimization, stress-driven grading, and conformal mapping; and ii) data-driven methods, which extend design coverage and accelerate iteration by learning latent geometry-performance relationships for inverse and multi-objective design. We specifically examine how AI-enabled generators, when combined with physics-informed evaluators, can reduce design iteration time by orders of magnitude while simultaneously preserving printability and structural credibility. Importantly, this review synthesizes AM process-specific generative strategies, emphasizing the explicit embedding of process physics and manufacturing constraints for modalities such as powder bed fusion and vat photopolymerization. Finally, emerging directions in scalable multiscale modeling, uncertainty-aware design, and hybrid physics-informed data-driven frameworks are outlined, pointing toward verifiable and industry-ready generative design methodologies.