Seismic performance of bolted column splice connections in steel moment frames

Abstract

Column splices are common in multi-storey steel construction. When buildings are designed for ultimate limit state, discontinuity in column stiffness due to splice stiffness properties is not often taken into account. Unlike welded splices, bolted connections display some degree of movement before they reach their ultimate strength. Although this phenomenon has received a lot of attention in beam-column connections, the stiffness and ductility characteristics of column splices and their importance on seismic response are less understood. These characteristics are important and should be incorporated in frame seismic analysis to prevent the possibility of an undesirable local or global failure.

A series of cyclic moment and shear tests were performed on bolted and welded column splice connections to understand and quantify their mechanical behaviour. The bolted connections covered a variety of construction methods. Specimens were tested quasi-statically. Flexural and shear performance, with a focus on strength and stiffness properties, were separately investigated via hysteresis loop measurements of rotation and translational shear displacement at the splice. Simple modelling approaches were proposed to predict the backbone behaviour the specimens, and were verified by experimental results.

Non-linear time history analysis of two generic moment frames were conducted to explore the effect of splice shear and rotational flexibility on frame dynamic response. The frames were analyzed for two suites of earthquakes representing Design Basis Earthquake and Maximum Credible Earthquake hazard levels. Rotational properties of column splices were modelled with gap material, while shear-deformation was modeled as bilinear, both consistent with observed experimental results. Parametric studies were undertaken to assess variation with shear elastic stiffness, post-slip rotational stiffness, and column splice location, and drift profiles were compared against the frame with rigid column splices.

It was observed in the experiments that all splice connections developed greater strength than expected in design, but showed widely varying rotational stiffnesses. Non-linear time history analyses showed some increase in drift response of the stories containing column splices. The response was also found to be sensitive to splice distribution through the frame stories. While bolted column splices may alter dynamic frame response, they do not appear to cause global failure mechanisms in the frames studied, provided they have sufficient strength. In the end, the key findings of this research were translated to implications on local and global design/analysis procedure for moment frames.