Optimization and mechanical characterization of 3D printed PLA/carbon nanotube composites using artificial neural networks

In this study, nanocomposite filaments of polylactic acid (PLA) reinforced with 5 wt% carbon nanotubes (CNTs) were successfully prepared via single-screw extrusion ensuring homogeneous CNT dispersion and consistent filament diameter suitable for Fused Deposition Modeling (FDM) 3D printing. Specimens were fabricated using a Pratham FDM printer under systematically varied process parameters, layer height (0.1, 0.2, 0.3 mm), infill density (50, 75, 100%), infill patterns (line, cubic, trihexagonal), and raster orientations (0°, 45°, 90°), following Taguchi L27 orthogonal array design. Mechanical characterization comprising tensile, compressive, and flexural tests per ASTM standards revealed substantial enhancements with CNT reinforcement: tensile strength increased up to 109.83 MPa (97% improvement), compressive strength reached 80.07 MPa (16% improvement), and flexural strength improved to 72.67 MPa (8% improvement) compared to pure PLA controls. ANN-based Levenberg-Marquardt backpropagation models accurately predicted these mechanical properties with regression coefficients above 0.88 for all tests, confirming strong correlations between experimental and modeled results. Microstructural analysis verified improved interlayer bonding and reduced porosity due to CNT addition, underpinning the observed mechanical enhancements. The integrated experimental and modeling approach establishes effective process structure-property relationships critical for optimizing 3D printing of PLA nanocomposites with superior performance.