Functionally Graded Lattice Infill and Cooling Channel Design Optimization for Additive Manufacturing

To realize the full potential of additive manufacturing (AM) techniques, a lattice structure design optimization methodology is proposed to design functionally graded lattice structures, in order to achieve optimal performance while satisfying manufacturing constraints. A lattice structure topology optimization (LSTO) method is first proposed and the framework includes three key steps: (1) Asymptotic homogenization (AH) is developed to calculate effective properties of 3D printed lattice materials, such as elastic modulus, yield strength, thermal conductivity and forced convection heat transfer coefficient; (2) Density-based topology optimization methodology is employed to compute the density distribution of lattice structures by using the material interpolation from AH procedure; (3) A reconstruction method is developed to transform an optimal density profile into variable-density lattice structure for practical fabrications. The proposed LSTO method is extensively studied for various problems ranging from, structural (minimum compliance problem and constraint stress problem), dynamic (natural frequency maximization), and heat and mass transfer. Validation of the LSTO method conducted on practical components is able to significantly improve the physical performance of the component with lightweight design. On the other hand, the LSTO method cannot handle functionally movable features optimization, e.g. cooling channels and bolt holes in components, which are non-designable and remains solid during optimization. To explore the potential benefits, the LSTO method is extended to the concurrent optimization of lattice infill and movable features optimization. A unified scheme for the combination of density-based topology optimization with level set topology optimization is thus proposed and derived. In addition to theory development, the LSTO method is further developed to solve a critical issue regarding build failures induced by residual stress inherent in the metal AM process. Specifically, a voxel-based methodology is proposed to efficiently generate Cartesian mesh for a solid part and its support structure. A build orientation optimization method and a LSTO-based support structure optimization method is developed to minimize the volume of sacrificial support structure under allowable stress constraint. Experiments have proved that the proposed framework can significantly reduce the residual stress, guarantee the manufacturability of the designs, and make it easy for trapped powder removal