Stress-driven generative design and numerical assessment of customized additive manufactured lattice structures

The rise of additive manufacturing (AM) has positioned lattice infilling as a pivotal strategy for creating lightweight, customized engineering components. This study presents a generative method that enables the conformal design and stiffness prediction of complex gradient strut-node lattice structures. A stress-driven Multi-Agent System (MAS) is introduced for the parametric optimization of lattice material distribution, incorporating geometric limitations, stress factors, and AM constraints. A beam element model simplifies the numerical analysis of the structures' linear stiffness. By applying the Response Surface Method (RSM), we develop a numerical model that evaluates the sensitivity of MAS design variables and predicts mechanical performance. By applying the Response Surface Method (RSM), a numerical model is established by the Response Surface Method (RSM), not only conducting a quantitative analysis on the sensitivity of MAS's design variables but predicting mechanical performance. This method is validated by designing a supporting component, demonstrating that the optimized lattice design can achieve a linear stiffness 1.4 times greater than that of conventional uniform lattice infills for the same mass. This research provides a comprehensive framework for the efficient design and analysis of irregular lattice structures at a macroscopic scale.