Design Optimization of Continuous Fiber Arrangement Using Lamination Parameters in Material Extrusion-Based Additive Manufacturing

Abstract

Continuous carbon fiber-reinforced polymer (CFRP) can be used in material extrusion-based additive manufacturing (AM). By appropriately arranging and orienting the continuous fibers, lightweight or high-strength mechanical parts and structures with complex deformation behavior and locally modified stiffness can be fabricated. Although many studies have been conducted to optimize the arrangement of continuous fibers fabricated using CFRP in AM, most of them have focused on the mechanical properties of the fabricated object in the lamination plane. This focus is due to the characteristics of AM, in which continuous fibers are placed in a plane and then layered, allowing for optimization at a relatively low computational cost. However, the computational cost of targeting mechanical properties outside the fabrication plane is enormous, making optimization design difficult. Furthermore, if the fiber material is arranged discontinuously, a process to cut the fibers is required during fabrication, resulting in decreased productivity and fabrication accuracy. Therefore, it is necessary to optimize the fiber arrangement, considering the continuity of the fiber material. To address these problems, this study aims to propose an efficient fiber arrangement optimization method, considering the continuity of the fiber material. For efficient stiffness optimization, the stiffness is represented by a combination of lamination parameters that have been used in the lamination design of CFRP sheets. A genetic algorithm was employed as the optimization algorithm. The proposed method using lamination parameters was implemented, and a case study of fiber arrangement optimization was performed on a simple structure. In addition, a full search was performed to evaluate all possible fiber arrangements for the target structure. The results of the proposed method and the full search confirmed the reliability of the proposed method, which achieved results that were equivalent to the best results obtained in the full search. In addition, a conventional method that directly optimizes the fiber arrangement as a design parameter was implemented. This result was compared with that of the proposed method. For a simple structure with a small number of layers, averaged over 20 runs, the conventional method converged faster than the proposed method, but the convergence speed worsened as the number of layers increased. Moreover, the fiber arrangement obtained by the conventional method was less continuous than the result of the proposed method. These results confirm the usefulness of the proposed method.