Sensitivity analysis of reverse engineering approaches for 3D-scanned welds with complex geometries

Abstract High stress concentrations are primarily caused by variations in the weld surface geometry and can lead to fatigue failure. 3D scanning technology combined with reverse engineering is becoming more widely used to capture detailed surface geometries and to build numerical models for further fatigue analysis. However, studies on the influence of different scanning systems, pointing spacing, and modeling approaches are limited. For this reason, a sensitivity analysis is needed to evaluate the reliability of the reverse engineering approach. Two scanning systems, fringe light projection (FLP) and laser line sensor (LLS), with different resolutions were applied to welded specimens with complex geometries. The stress concentration factors (SCFs) were determined using finite element analysis (FEA) based on the surface scans (point clouds). Two different modeling methods, connecting parts with tie connections or through sub modeling were compared. In addition, different element size and type, shape functions were used in the FEA. To cover the effect of near surface variation of weld quality and stress concentrations, critical distance theory (CDT) approach was applied. The results show that SCFs are more sensitive to the reverse engineering approach than CDT are.