"Progressive collapse" occurs when a localised failure of a structural component triggers propagation of failure to adjoining elements, eventually resulting in collapse of the entire structure or a major part of it. This phenomenon in steel structures is mitigated primarily through beam-to-column connections. In particular, simple shear connections are critical as they are the most common beam-to-column connections in steel structures and experience different load combinations than those considered in conventional design. They experience extreme rotation and axial tension due to catenary action developed in the frame after the removal of an adjoining column. With increasing experimental evidence and detailed numerical models, prediction of the behaviour and failure of shear connections is increasingly feasible. However, uncertainties in such predictions are unavoidable due to the uncertainties in the material and geometric properties. This paper aims to investigate the reliability of the connection capacity predictions due to variability in the connection parameters. As such, high-fidelity finite-element-based models including the connection components (plate, bolts, beams) have been analysed to examine the performance and behaviour of the shear connections under different geometric variables. Models include a single 6 m span beam with shear tab connections at both ends and application of a concentrated load at the removed column. Uncertainties in the plate dimensions such as end distance, plate thickness, and hole diameter have been considered for the reliability and sensitivity assessments. Explicit quasi-static analysis using ABAQUS is employed to overcome the convergence difficulties related to contact, geometric and material nonlinearities, large deformation, and fracture simulation.
展开▼