Predicting flexural strength of additively manufactured continuous carbon fiber- reinforced polymer composites using machine learning

Carbon fiber-reinforced polymer (CFRP) composites have been used extensively in the aerospace and automotive industries due to their high strength-to-weight and stiffness-toweight ratios. Compared with conventional manufacturing processes for CFRP, additive manufacturing (AM) can facilitate the fabrication of CFRP components with complex structures. While AM offers significant advantages over conventional processes, establishing the structure-property relationships in additively manufactured CFRP remains a challenge because the mechanical properties of additively manufactured CFRP depend on many design parameters. To address this issue, we introduce a data-driven modeling approach that predicts the flexural strength of continuous carbon fiber-reinforced polymers (CCFRP) fabricated by fused deposition modeling (FDM). The predictive model of flexural strength is trained using machine learning and validated on experimental data. The relationship between three structural design factors, including the number of fiber layers, the number of fiber rings as well as polymer infill patterns, and the flexural strength of the CCFRP specimens is quantified.