Data-driven multiscale topology optimization of solid-lattice hybrid structures for additive manufacturing

Compared to pure solid or lattice structures, solid-lattice hybrid structures offer enhanced mechanical performance, effectively balancing various design requirements. In this paper, a data-driven multiscale topology optimization method is proposed for designing solid-lattice hybrid structures tailored for additive manufacturing. Implicit modeling techniques are employed to construct microstructure models, enabling both mechanical characterization of microstructure unit cells and geometric reconstruction of optimized designs. A sample database and surrogate model are created to accelerate the optimization process. The solid and lattice structures are represented by two sets of density variables, defined on the solid and lattice layer meshes, respectively. Additionally, the Heaviside function is introduced as a projection function for density filtering, removing pseudo-density elements from the solid layer mesh. Two volume constraints are applied to regulate the structural configuration and performance during the optimization process. After optimization, full-scale solid-lattice hybrid structures are reconstructed by combining topologically optimized solid structures with graded lattice structures using the hybrid level set method (HLSM). Several 2D and 3D numerical examples are provided to validate the effectiveness of the proposed method. Finally, the optimized structures are fabricated via additive manufacturing (AM), and their performance is validated through both finite element analysis (FEA) and experimental testing. The results confirm notable improvements in overall mechanical performance, highlighting the effectiveness of the proposed method in designing lightweight, high-performance structures.