A challenge remains in achieving adequate surface roughness of SLM fabricated interior channels, which is crucial for fuel delivery in the space industry. This study investigated the surface roughness of interior fine flow channels (1 mm diameter) embedded in SLM fabricated TC4 alloy space components. A machine learning approach identified layer thickness as a significant factor affecting interior channel surface roughness, with an importance score of 1.184, followed by scan speed and laser power with scores of 0.758 and 0.512, respectively. The roughness resulted from thin layer thickness of 20 µm, predominantly formed through powder adherence, while from thicker layer of 50 µm, the roughness was mainly due to the stair step effect. Slow scan speeds increased melt pools solidification time at roof overhangs, causing molten metal to sag under gravity. Higher laser power increased melt pools temperature and led to dross formation at roof overhangs. Smaller hatch spaces increased roughness due to overlapping of melt tracks, while larger hatch spaces reduced surface roughness but led to decreased part density. The surface roughness was recorded at 34 µm for roof areas and 26.15 µm for floor areas. These findings contribute to potential adoption of TC4 alloy components in the space industry.

Geopolymers, as low-carbon inorganic binders with excellent mechanical and thermal properties, offer adjustable rheology and solidification behavior suitable for Direct Ink Writing (DIW). However, DIW of geopolymers remains challenging due to the complex interaction between material behavior and printing parameters. This study proposes a data-driven and machine learning approach for optimizing geopolymer DIW. Metakaolin-based geopolymer inks with varying rheological properties were formulated using sodium silicate (modulus 2.4) at solid-to-liquid ratios from 8:12 to 8:20. Printing performance was evaluated using nozzle diameters of 0.25–1.03 mm and extrusion pressures of 15–75 psi. Results show that high-solid inks maintained shape but suffered clogging, while low-solid inks extruded smoothly but collapsed after deposition. Intermediate-solid inks achieved optimal printability, balancing yield stress and geometric fidelity. Experimental data were further used to develop a modified Herschel–Bulkley model that accurately predicts relationships among pressure, nozzle geometry, and filament diameter. Machine learning models, including Random Forest, Support Vector Machine, and Gradient Boosting, achieved up to 78% classification accuracy. This study demonstrates the strong relationship between composition, rheology, and printability in geopolymer DIW through a combined experimental and data-driven framework.

Geopolymers, as low-carbon inorganic binders with excellent mechanical and thermal properties, offer adjustable rheology and solidification behavior suitable for Direct Ink Writing (DIW). However, DIW of geopolymers remains challenging due to the complex interaction between material behavior and printing parameters. This study proposes a data-driven and machine learning approach for optimizing geopolymer DIW. Metakaolin-based geopolymer inks with varying rheological properties were formulated using sodium silicate (modulus 2.4) at solid-to-liquid ratios from 8:12 to 8:20. Printing performance was evaluated using nozzle diameters of 0.25–1.03 mm and extrusion pressures of 15–75 psi. Results show that high-solid inks maintained shape but suffered clogging, while low-solid inks extruded smoothly but collapsed after deposition. Intermediate-solid inks achieved optimal printability, balancing yield stress and geometric fidelity. Experimental data were further used to develop a modified Herschel–Bulkley model that accurately predicts relationships among pressure, nozzle geometry, and filament diameter. Machine learning models, including Random Forest, Support Vector Machine, and Gradient Boosting, achieved up to 78% classification accuracy. This study demonstrates the strong relationship between composition, rheology, and printability in geopolymer DIW through a combined experimental and data-driven framework.