Experimental and numerical investigation of the progressive collapse of steel frames
MetadataShow full item record
The purpose of this dissertation is to experimentally evaluate the sequence of failures in steel buildings leading to progressive collapse and, within a multi-hazard design perspective, investigate whether earthquake-resistant design details can enhance the resistance to collapse. While various numerical studies have examined this issue, experimental collapse studies have been limited to single connections, without development of a structure-level load redistribution mechanism, or to miniature frames that cannot adequately capture the post-yield, large-deformation behavior of the structural elements. In this context, a multi-hazard approach is adopted, where a seismically designed structure is subjected to an extreme event (e.g. blast or impact), leading to the instant removal of a primary structural element, such as a base column. Two 1/3 scale three-story, two-bay steel frames, a special moment resisting frame (WMRF) and a post-tensioned energy dissipating frame (PTED) designed and previously tested for seismic performance on earthquake simulators, were adapted for quasi-static collapse testing by simulating the structural response after the sudden failure of a column. A sliding mechanism was designed, which permitted the frames to be tested up to large displacements, while providing realistic boundary conditions. High capacity hydraulic actuators implemented the gravity loading, by 'pushing down' the frame to the point of inability to sustain the load. The experiments demonstrated the capacity of the frame specimens to prevent the triggering of a progressive collapse sequence for expected design loads, after the removal of a base vertical load-bearing element. The WMRF specimen provided further resistance through wide-spread yielding at members' ends and connection panel zones, until a series of fractures compromised its strength. On the other hand, the collapse resistance of the PTED specimen was primarily dependent on the ultimate ductility and failure mode of the wire strands that provided the post-tensioning force. The sequence of damage in both frames is identified and traced back to the measured resistance. Supplementary numerical analyses using both commercial and research-oriented software attempt to replicate the experimentally observed global behavior deep into the inelastic, large-displacement range. The tracking of the damage sequence during the push-down tests provides new insight into the simplified macroscopic idealization of the post-peak force vs. displacement behavior (damage) of steel framed structures, particularly with respect to the modeling approach. The degradation of both steel frames can be captured using parallel bi-linear spring models with rapid strength loss after achieving a given strain. The numerical models are able to capture the salient features of the global collapse behavior measured in the experiments.