AI Model Reveals Which 3D Printer Settings Waste the Most Energy
Researchers have built an AI system that predicts exactly how printing speed, layer height, and infill density affect the environmental cost of 3D-printed parts. The tool achieves 97% accuracy and works across different materials and designs—giving manufacturers a practical way to cut electricity use and plastic waste before hitting print.
Originaltitel: Explainable AI analysis of printing parameter effects on life cycle assessment for sustainable material-extrusion additive manufacturing
As material extrusion additive manufacturing (MEX) moves into industrial use, its electricity and filament consumption translate into non-negligible life cycle assessment (LCA) impacts. We propose a unified explainable AI (XAI)–LCA framework that links four printing parameters (print speed, layer height, infill density, raster angle) to four ReCiPe 2016 Midpoint (H) indicators at the printing stage. Slicer-generated datasets are combined with random forest models, Shapley additive explanations (SHAP), response surfaces and partial dependence plots (PDPs) to quantify how parameters affect impacts through energy- and material-driven pathways. The resulting decomposition model achieves coefficients of determination 𝑅2>0.97 and mean absolute percentage errors (MAPE) below 7% in ten repeated splits; blind validation on fifteen unseen configurations yields MAPE <10%. XAI analyses show that layer height and print speed dominate energy-related indicators, while infill density governs material-related ones. We demonstrate transferability to other materials (ABS, PETG) and geometries (cube, cylinder, L-shape) with MAPE <10%. The framework integrates with standard slicing workflows, estimates cost per part, and is compatible with emerging policy instruments such as the EU Ecodesign for Sustainable Products Regulation and Digital Product Passport schemes, providing transparent decision support for sustainable MEX.