Measurement and qualification of welded I-column failure mode using laser scanning technique

Correct measurement and qualification of failure mode for steel members are important for the development of member design methods, and have the potential to effectively identify the failure mechanisms of structures after disaster. Modern laser based geometric measurement technique can easily obtain 3D point clouds of member. However, rare studies have been reported on the deformation measurement from the geometric model reconstructed from 3D point clouds of an extensively deformed member. This study focuses on the automatic 3D geometric model reconstruction of steel column in failure state using the laser scanning technique and the qualification of the deformation. First, tests were carried out on steel I-columns subjected to axial compression to obtain two typical failure modes which were then transferred to 3D point clouds using the laser scanning technique. Afterwards, a rectangular coordinate system was built, and 3D point clouds were sliced into several cross-sections based on the plane-section assumption. The flanges and web were filtered out by applying thresholds to cross-correlation coefficient between the points' normal and the unit vectors along the axes. A curve segmentation method using the image-based cross-correlation matching was proposed to accurately detect the corner points and weld regions in cross-section. The 3D geometric model was then reconstructed by assembling the cross-sections along the longitudinal axis of column. Finally, a cross-sectional constrained finite strip method (cFSM) was proposed based on the cross-sectional pure buckling modes generated using the classical cFSM method, which is more efficiency and robust. Both the proposed method and the classical cFSM method were used to qualify the deformation obtained from reconstructed 3D geometric model. Comparison of the qualification results between these two methods verified the validity of the proposed method.

Automated summary

The study focuses on automatic reconstruction of 3D geometric model of a steel column in failure state using laser scanning technique and qualification of the deformation. A curve segmentation method based on image-based cross-correlation matching was proposed to accurately detect the corner points and weld regions in the cross-section. The efficiency and robustness of the proposed method were verified through comparison with the classical cross-sectional constrained finite strip method.