Designing Beam Shape of Lattice Structure for Enhancing Mechanical Characteristics Based on Multi-objective Bayesian Optimization

Lattice structures, a kind of metamaterial, with high strength and low density have gained significant attention due to the ever-increasing need for designing lightweight strong parts. Most existing studies examined a variety of lattice structures made of uniform cross-section beams, which does not encompass a vast design space related to the beam shape. Moreover, very sharp corners formed at the cross-sections of the beams serve as the source of stress concentration that significantly hampers the strength of the lattice structures. In this research, we explore the beam shape thoroughly as a new design space in a variety of lattice structures based on smooth Bezier curve generations and search for the ultimate performance of the lattice structures. The elastic modulus of the lattice structures with the free-form beam shape is evaluated via finite element method, and the optimal beam shape for the ultra-stiffness and low density is designed through multi-objective Bayesian optimization (MBO). The efficient search of the optimal shape via MBO enables us to search for the optimal beam shapes for many different lattice structures. The lattice structures with optimal shapes are manufactured using 3D printing, and significant improvement in specific stiffness is confirmed in comparison with existing designs.