Part-scale build orientation optimization for minimizing residual stress and support volume for metal additive manufacturing: Theory and experimental validation

• Published in: Computer aided design Vol. 113; pp. 1 - 23
• Main Authors: Cheng, Lin, To, Albert
• Format: Journal Article
• Language: English
• Published: Amsterdam Elsevier Ltd 01.08.2019; Elsevier BV
• Subjects: Additive manufacturing; Anchors; Asymptotic methods; Complexity; Cracks; Deformation; Design for manufacturability; Finite element method; Inherent strain method; Laser cooling; Lattice structure; Manufacturability; Mesh generation; Metal additive manufacturing; Modelling; Multiple objective analysis; Optimization; Orientation; Particle swarm optimization; Powder beds; Residual stress; Stress concentration; Stress distribution; Substrates; Voxel-based mesh generation; Warpage; Inherent strain method; Lattice structure; Particle swarm optimization; Metal additive manufacturing; Voxel-based mesh generation
• ISSN: 0010-4485; 1879-2685
• DOI: 10.1016/j.cad.2019.03.004

Laser powder bed metal additive manufacturing (AM) has been widely accepted by the industry to manufacture end-use components with complex geometry to achieve desirable performance (i.e. conformal cooling). However, residual stress and large deformation introduced in the laser AM process lead to severe issues, such as cracks, delamination, and large deformation. These issues result in the stoppage of powder spreading and warpage of the component. To overcome these issues, a novel optimization framework based on fast process modeling is proposed to find the optimal build orientation by minimizing the maximum residual stress and support structure volume. For support generation, a voxel-based methodology is proposed to systematically capture support surfaces from STL file, form support structure, and generate Cartesian mesh for fast process modeling. Instead of using conformal mesh, the voxel-based fictitious domain method is used to calculate the stress distribution in the design domain including the support structure, which is represented by the homogenized model. This can circumvent time-consuming mesh generation for geometrically complex geometry and its support structure during the optimization iterations, thus making it possible to minimize residual stress through orientation optimization based on process modeling. Due to its self-supporting and open-cell nature, lattice structure is employed as the support structure to anchor the overhangs to the substrate to prevent distortion resulting from residual stress. Asymptotic homogenization (AH) method is employed to compute the effective properties of lattice structure, while a multiscale model is proposed to compute the yield strength. In particular, the multi-objective optimization including both the residual stress and support volume is discussed and investigated in this work. Experimental validation is conducted on a realistic component with some geometric complexity. By comparing the component and support structure without build orientation optimization, it is found that the proposed framework can significantly reduce the influence of the residual stress on the printed part, ensure the manufacturability of the design, and decrease the material consumption for the sacrificial support structure simultaneously. •A new voxel-based overhang detection method is proposed for AM simulation.•A fictitious domain method is developed to simulate residual stress in metal AM.•Residual stress and support volume are minimized through the optimization method.•Experiments are conducted to investigate the efficiency of the proposed framework.