Advancing composite 3D printing: deep learning-optimized rheology-modified polymers with continuous carbon fiber reinforcement

The optimal design of glass bead-filled polyamide 12-based composites for improved performance in 3D printing is addressed in this research. Orientation-dependent brittleness or the strain-dependent tensile strength variations in polyamides have been problematic issues to date. Optimizing formulations based on material models using deep learning and Elk Herd Optimization (EHO) leads to formulating balanced automotive and industrial parts in terms of strength and flexibility. The mechanical properties are predicted using a Multi-Layer Perceptron (MLP) model, after which the optimized formulations are extruded to form filaments reinforced with continuous carbon fibers with very minimal voids and excellent dimensional accuracy. Tests run on the mechanical tests reveal quite significant improvements: whereas polyamide reaches a peak tensile strength of about 50 MPa at a 10% strain for the polyamide-glass bead composite, this reached its peak at 33 MPa at 4–5% strain. Orientation-dependent testing indicated X-printed composites preserved, to maximum deflection without rupture. Such results indicate the potential of advanced computational techniques applied for optimization of polymer composites toward 3D printing and potential being interesting in industrial applications requiring further high performance and reliability.