Predictive Modeling and Multi-Objective Optimization of SLM 316L Stainless Steel via Response Surface Methodology and Grey Relational Analysis

This study presents a methodology for balancing surface integrity, densification, and hardness in selective laser melting (SLM) of 316L stainless steel. While SLM offers unparalleled geometric freedom, achieving optimal performance is often hindered by trade-offs among process parameters. To address this, a high-fidelity predictive and optimization framework was established by integrating response surface methodology (RSM) with Taguchi-based grey relational analysis (GRA). A Taguchi L9 orthogonal array with analysis of variance (ANOVA) was implemented to identify the discrete influences of laser power, scanning speed, hatch spacing, and scanning pattern. Reduced quadratic RSM models were then developed to capture nonlinear interactions. Statistical validation yielded coefficients of determination (R2) of 85.6% for surface roughness and 80.7% for hardness, with a mean absolute error within 2.5% of average response values. Interactive response surface analysis revealed that surface roughness improves monotonically with energy density, whereas hardness exhibits a convex behavior governed by the synergy between laser power and scanning speed. Multi-objective optimization through GRA identified the optimal parameter set as 180 W laser power, 500 mm/s scanning speed, 0.08 mm hatch spacing, and a spiral scanning pattern. Validation experiments produced a balanced profile of 8.45 μm surface roughness, 0.79% porosity, and 207.8 HV hardness, closely matching analytical predictions. Compared to single-objective optimization, the integrated GRA approach effectively reconciled contradictory optimization trajectories, providing a robust strategy for fabricating high-performance 316L components for demanding industrial and biomedical applications.