Lightweight design of two-level supports for extrusion-based additive manufacturing based on metaheuristic algorithms

• Published in: Rapid prototyping journal 2023-04, Vol.29 (4), p.850-866
• Main Authors: Feng, Ruiliang, Jiang, Jingchao, Thakur, Atul, Wei, Xiangzhi
• Format: Article
• Language: English
• Subjects: 3-D printers; Additive manufacturing; Design; Extrusion; Genetic algorithms; Heuristic methods; Lightweight; Manufacturing; Optimization; Particle swarm optimization; Path planning; Rapid prototyping; Searching; Source programs; Strategy; Three dimensional printing; Topology
• ISSN: 1355-2546; 1758-7670; 1355-2546
• DOI: 10.1108/RPJ-01-2022-0038

Purpose Two-level support with Level 1 consisting of a set of beams and Level 2 consisting of a tree-like structure is an efficient support structure for extrusion-based additive manufacturing (EBAM). However, the literature for finding a slim two-level support is rare. The purpose of this paper is to design a lightweight two-level support structure for EBAM. Design/methodology/approach To efficiently solve the problem, the lightweight design problem is split into two subproblems: finding a slim Level 1 support and a slim Level 2 support. To solve these two subproblems, this paper develops three efficient metaheuristic algorithms, i.e. genetic algorithm (GA), genetic programming (GP) and particle swarm optimization (PSO). They are problem-independent and are powerful in global search. For the first subproblem, considering the path direction is a critical factor influencing the layout of Level 1 support, this paper solves it by splitting the overhang region into a set of subregions, and determining the path direction (vertical or horizontal) in each subregion using GA. For the second subproblem, a hybrid of two metaheuristic algorithms is proposed: the GP manipulates the topologies of the tree support, while the PSO optimizes the position of nodes and the diameter of tree branches. In particular, each chromosome is encoded as a single virtual tree for GP to make it easy to manipulate Crossover and Mutation. Furthermore, a local strategy of geometric search is designed to help the hybrid algorithm reach a better result. Findings Simulation results show that the proposed method is preferred over the existing method: it saves the materials of the two-level support up to 26.34%, the materials of the Level 1 support up to 6.62% and the materials of the Level 2 support up to 37.93%. The proposed local strategy of geometric search can further improve the hybrid algorithm, saving up to 17.88% of Level 2 support materials. Research limitations/implications The proposed approach for sliming Level 1 support requires the overhanging region to be a rectilinear polygon and the path direction in a subregion to be vertical or horizontal. This limitation limits the further material savings of the Level 1 support. In future research, the proposed approach can be extended to handle an arbitrary overhang region, each with several choices of path directions. Practical implications The details of how to integrate the proposed algorithm into the open-source program CuraEngine 4.1