Experimental and Machine Learning Characterization of Milled Carbon Fiber-Reinforced PLA in Fused Deposition Modeling

Abstract Over the past decade, additive manufacturing of fiber-reinforced polymer composites (FRPCs) has gained attention due to its efficiency and flexibility compared to traditional methods. This study investigates the effects of varying milled carbon fiber (MCF) content in polylactic acid (PLA) for fused deposition modeling on mechanical, thermal, and water absorption properties. This study uniquely integrates experimental analysis with machine learning predictions to determine the optimal MCF content, bridging the gap between theoretical modeling and experimental validation. The filaments were fabricated in-house, followed by tensile and flexural testing. Results showed that MCF content below 5 wt% improved tensile and flexural strength, while higher content led to a decrease. However, stiffness increased with higher MCF content. Scanning electron microscopy revealed carbon fiber distribution and fracture mechanisms, with brittle fracture for neat PLA and MCF/PLA. However, increasing the carbon fiber content results in a more brittle fracture behavior. Increased carbon fiber content also affected water absorption and thermal conductivity. A Gaussian process model predicted an optimal MCF content of approximately 3.1 wt% for overall performance, aligning with experimental findings, which showed the best results at 1 wt%.