Enhancing tensile and flexural strength of ABS/graphene nanocomposites through FFF parameter optimization

Additive manufacturing, specifically fused filament fabrication (FFF), has transformed the manufacture of complex thermoplastic components; nonetheless, weak interlayer bonding and anisotropic behaviour remain major limitations on mechanical performance. Graphene-reinforced polymers and Taguchi-based FFF parameter optimization have each been studied extensively, but their combined effect on ABS composites — specifically how graphene nanoplatelet (GNP) loading interacts with layer thickness, build orientation, infill pattern, and infill density to govern tensile and flexural performance — has not been systematically quantified; this study addresses that gap. Melt blending and single-screw extrusion were used to prepare ABS/GNP composite filaments at 0.5, 1, and 1.5 wt% GNP loadings, and for each composition a Taguchi L27 orthogonal array was used to evaluate four FFF process parameters (layer thickness, build orientation, infill pattern, and infill density). Specimens were tested per ASTM D638 (tensile) and ASTM D790 (flexural). Among the GNP loadings, ABS/0.5 wt% GNP produced the highest mean tensile (21.52 MPa) and flexural (39.42 MPa) strengths, indicating that the lower GNP content achieved better dispersion under the melt-blending conditions used, while the 1 and 1.5 wt% loadings showed evidence of agglomeration-related reductions in strength. Within the ABS/0.5 wt% GNP composition, the parameter combination of 0.19 mm layer thickness, on-long-edge orientation, grid infill pattern, and 90% infill density produced the best confirmed tensile and flexural strengths of 40.1 MPa and 56.9 MPa, respectively, with a prediction error below 3%. A correctly parameterized ANOVA (two degrees of freedom per three-level factor) identified build orientation as the dominant parameter, contributing 81.8% to tensile strength and 50.7% to flexural strength, followed by infill density (10.8%/29.1%) and infill pattern (3.1%/16.7%), while layer thickness was not statistically significant.