Multi-response optimization of flexural performance in additively manufactured carbon fiber-reinforced PLA composites

Abstract Additively manufactured carbon fiber-reinforced polylactic acid (CF-PLA) composites are increasingly used for lightweight components that require improved flexural performance. However, their mechanical response is strongly governed by the combined effects of printing parameters, making systematic modeling and multi-response optimization essential. In this study, the flexural behavior of CF-PLA composites was investigated using a central composite design (CCD) based on four material extrusion parameters: printing speed (PS), printing temperature (PT), layer thickness (LT), and number of outer shells (OS). Eight flexural responses were analyzed to describe stiffness-, strength-, strain-, and energy-related performance indicators. The results showed that PT, LT, and OS had dominant but response-dependent effects, whereas PS exhibited a comparatively lower influence. Flexural stress at break was significantly affected by interaction terms, indicating strong process-parameter interdependencies, while flexural strain at break showed a more independent response, with no significant interaction effects. The developed response surface models showed high predictive accuracy, with coefficients of determination (R 2 ) ranging from 93.06 to 96.64%. To account for the unequal importance of the flexural responses, MEREC and CRITIC objective weighting methods were integrated with Grey Wolf Optimizer (GWO) and Desirability Function Approach (DFA). Among the investigated strategies, GWO-MEREC-DFA achieved the highest composite desirability value (0.9899), outperforming GWO-CRITIC-DFA (0.9756) and equal-weight optimization (0.9751). The optimized condition also produced superior flexural performance and the most compact, least disrupted damage morphology. These findings demonstrate the effectiveness of hybrid GWO-based approaches for optimizing MEX-printed CF-PLA composites.