The fabrication of long carbon fiber reinforced polylactic acid composites via fused deposition modelling: Experimental analysis and machine learning

As a promising technology to revolutionize traditional manufacturing processes, additive manufacturing has received great attention by virtue of its cost savings, minimum material waste, and tool-less production of complex geometries. The development of fiber-reinforced composites via this technique that exhibits superior strength is thus becoming a hot spot in recent years. This paper focused on the 3 D printing of polylactic acid (PLA) composites with the incorporation of chopped long carbon fiber (CF) with an average length of 4.6 mm via FDM fabrication. By varying its loading quantity, the effect of CF contents on the mechanical, thermal, and morphological properties of these 3 D printed composites was thoroughly investigated. The results showed that with the increase of CF contents, all assessed properties of CF/PLA composites including tensile properties, flexural properties, hardness, and thermal conductivity were effectively improved compared to the neat PLA. Their performance exhibited the same upward-downward-upward trend with the addition of CF. It can be attributed to the mutual influence generated from inter-/intra filament porosities and the high stiffness of CF. Meanwhile, a machine learning technique, Gaussian Process modeling was also introduced in this study for the property prediction of the composites. In comparison with the experimental analysis, the optimal CF content of 6.7 wt% with the best overall performance was predicted using this model, which was very close to the best experimental results at 5 wt% CF.

Automated summary

Additive manufacturing is a promising technology for revolutionizing traditional manufacturing processes, such as the development of fiber-reinforced composites with superior strength. This study focused on 3D printing PLA composites with chopped long carbon fiber, showing improved mechanical and thermal properties by varying the CF contents. The use of Gaussian Process modeling successfully predicted the optimal CF content for the composites, which closely matched the experimental results.