Stiffness-to-weight optimization of parametric lattices via adaptive voxelization and genetic algorithms

This paper presents a fully automated and intelligent methodology for designing, modeling, and optimizing lattice-based materials tailored for MEX additive manufacturing. The proposed approach is based on a parametric design of a stiffened rhomboidal lattice derived from a modified octahedral unit cell, engineered to enhance mechanical performance under uniaxial compression while ensuring support-free printability. The geometry is governed by a compact set of parameters and a manufacturability constraint (α). A fast and efficient analytical model, rooted in beam theory and the stiffness method, replaces conventional FEA to predict mechanical behavior with high accuracy. This model feeds a genetic algorithm that optimizes the stiffness-to-weight ratio within the internal volume of the part. A voxelization system, adaptive to local stress fields, enables intelligent geometry adjustment, avoiding manual modeling or costly topology optimization. The methodology was validated through simulations and ASTM D695-15 compression tests, with relative errors below 5%. Results revealed the influence of lattice density and aspect ratio on failure modes, ranging from brittle delamination to ductile deformation. A real-world case study confirmed the approach’s effectiveness, achieving material savings while maintaining structural integrity. This methodology applies to any unidirectional-element lattice typology, eliminating the need for expensive commercial software