Current practice in the seismic design of bridges assumes that their superstructures do not need to be explicitly designed for earthquake loads. They are assumed to remain elastic by virtue of their inherent strength and in-plane stiffness which is required for service loads. As a consequence few codes require detailed design of these members. Whereas this assumption appears valid for concrete box girder superstructures, the performance of steel bridges with concrete decks in recent earthquakes has cast doubt on the validity of this approach for this class of bridge. In particular, damage has occurred within the end cross frames of steel superstructures which are known to be the primary element in the lateral load path from the deck to the substructures. It is also known that designing these end frames with ductile details and allowing braces within the frames to yield, can significantly reduce the lateral loads transferred to the substructures. But little is known about how to maximize this effect while at the same time minimizing the associated damage to the studded connection between the concrete deck and steel girders within these frames.;In this dissertation, finite element analytical studies are described on multi-girder, multi-span, steel superstructures to identify load paths, factors influencing cross frame stiffness, tolerance for drift, and robustness of studded steel-to-concrete connections. Moments and shears transmitted through these connections rotate the girders about their longitudinal axes, and since this rotation is not uniform along the girder, the torsional stiffness of the girder-deck system plays an important role in the behavior of the cross frame. Furthermore, these moments are transmitted through the connections by pairs of tensile and compressive forces which, as the transverse loads increase, can cause yield in the studs and breakout of the concrete. Subsequent damage is difficult to inspect and expensive to repair.;A cross frame detail is developed in this work that minimizes the torsional demand on the superstructure (reduces the so-called system effect) and eliminates the damage observed in the studded steel-to-concrete connections. This frame uses a chevron brace (an inverted V) in the cross frame rather than a conventional X-brace, and directly connects the concrete deck to the lower flanges of the girders. The load path now bypasses the steel-to-concrete connection and it no longer needs to transmit moments. The studs are therefore removed from the upper flanges allowing the girders to rotate below the deck as required.;Experimental studies on a 3-girder cross frame with concrete deck are described. The results of these studies confirm the analytical results and the validity of the decoupled cross frame.;Experimental investigations were conducted on a set of three subassembly specimens to establish their lateral cyclic response including the initial stiffness, ultimate strength and failure modes of subassembly models with various shear connector configurations. The specimens were one-half scale models of a steel girder bridge superstructure prototype. Two of the specimens represented typical end cross frames details without diagonal bracings.;The results of the experiment investigations show the susceptibility of shear connectors near the end cross frames due to combined tension and shear forces. The experimental investigations on the third specimen confirm the analytical results and the validity of the decoupled cross frame.;Simplified analysis and design method are also developed as part of this study to determine the seismic response parameters of single and multi-span steel girder bridges with ductile end cross frames. The proposed methods are based on an iterative solution and show good agreement with results from nonlinear time history analyses.
展开▼