Hybrid Predictive Modelling of Surface Roughness in DMLS AlSi10Mg: Integrating Thermo-Mechanical Simulation and Optical Metrology on ISO 10791-7 Components

Abstract

Control of surface quality remains a critical challenge in Direct Metal Laser Sintering (DMLS), particularly for geometrically complex components. While various empirical models exist, there is a lack of predictive frameworks that combine thermo-mechanical simulation with optical metrology validated on standardized industrial geometries. This study establishes a hybrid predictive model for the surface roughness (Sa) of AlSi10Mg components manufactured via DMLS, utilizing the ISO 10791-7 standard to represent complex thermal histories. Thermo-mechanical simulations were conducted using Simufact Additive to model the interaction of laser power (P), scan speed (v), and hatch distance (h) on surface morphology. These were validated against optical focus-variation measurements of nine test artifacts. Results indicate that while direct thermo-mechanical simulation systematically overestimates roughness magnitudes, it accurately captures parameter-driven trends. A 'fused' second-order regression model, integrating simulation-derived structural insights with experimental coefficients, achieved exceptional predictive accuracy. The proposed model demonstrated a coefficient of determination (R2) of 0.99 for simulation data and prediction deviations of less than 1.6 % against physical validation measurements. The study reveals that scan speed and hatch distance are the dominant factors governing surface quality in the stable process window (v approx. 1200 mm/s, h approx. 0.15 mm). This framework offers a robust tool for pre-manufacturing quality assurance, enabling process optimization without costly iterative testing.