An integrated geometric and mechanical digital twin framework for shear buckling assessment of steel beams

Abstract

Accurate assessment of the shear buckling behavior of steel beams is essential for ensuring the safety and reliability of steel structures. This study proposes an integrated geometric and mechanical digital twin (DT) framework for the assessment and prediction of shear buckling behavior in steel beams. Steel beams tested to shear buckling failure are taken as the physical entities, whereas the corresponding high-fidelity finite element (FE) models reconstructed from 3D laser-scanned point cloud data are regarded as the virtual entities. A geometric DT is first established to explicitly capture the as-measured initial geometric imperfections of the web. Based on sensitivity analysis and particle swarm optimization, the measured mid-span deflection, web out-of-plane displacement, and strain responses collected from the physical entity are fused with the virtual entity to establish a mechanical DT, thereby reducing mechanical uncertainties and improving physical-virtual consistency. The developed DT is finally used for shear buckling assessment and prediction. The results demonstrate that the proposed framework is both feasible and effective, significantly enhancing the accuracy and robustness of structural performance assessment.