Exploring Stochastic Surface and Structural Characteristics in 3D-Printed Polymer Parts Using a Digital Twin Approach

This study presents a digital twin-based framework that integrates computational modeling, stochastic optimization, and experimental validation to enhance both the surface quality and structural performance of 3D-printed polymer components. A MATLAB-CAD interface was developed for automated geometry generation and STL export, enabling parameterized modeling of sinusoidal surface profiles and internal lattice structures. Surface optimization was achieved using a genetic algorithm (GA), which reduced arithmetic mean roughness (Ra) by nearly 40–60% compared to baseline designs, with profilometric scans validating the uniformity of the GA-tuned surfaces. In parallel, fused filament fabrication (FFF) was employed to fabricate lattice specimens with different infill patterns, and finite element analysis (FEA) was conducted to capture stochastic variations in dimensional accuracy and mechanical behavior. Results confirmed that hexagonal infills consistently achieved higher stiffness-to-weight ratios and more uniform stress distributions compared to rectangular counterparts. The DT model further predicted surface Ra with a mean error below 5% and achieved high accuracy in structural performance with [Formula: see text] > 0.95, while optimized printing conditions reduced stochastic variability by nearly 20%. Overall, the findings highlight the effectiveness of combining GA-driven surface optimization, infill design strategies, and DT modeling to improve the reliability, defect prediction, and lightweight potential of polymer additive manufacturing parts.