| In a federally-funded project on the seismic performance of curved highway bridges at the University of Nevada Reno, a 2/5th scale model of a 3-span, curved steel plate girder bridge was tested on multiple shake tables with an isolation system comprising 12 lead-rubber isolators. The purpose of this experiment was three-fold: (1) confirm that elastic performance of the columns could be achieved during the Design Earthquake using isolation, (2) study the effect of curvature on the seismic response of an isolated bridge, and (3) identify the limit states for an isolated curved bridge and, in particular, determine the nature and consequences of isolator instability during extreme input motions.;Elastic performance was indeed achieved during the Design Earthquake with no concrete spalling in potential plastic hinge zones and minor cracking on face of the columns. In fact, essentially elastic behavior was observed up to three times the Design Earthquake. Even though this bridge was highly curved (subtended angle was 1.8 radians), curvature had little effect on the response of the isolators. It did cause asymmetry in response and the abutment isolators were subject to higher displacements than those over the piers, but at the Design Earthquake these differences were small and the results of the AASHTO Simplified Method of analysis were adequate for design purposes. However, at three times the Design Earthquake, instability occurred in the isolators at one of the abutments due to excessive displacement. Bridge collapse did not however occur because the isolators at other supports remained stable. Full recovery of the unstable isolators was observed. In fact subsequent seismic excitation applied to the bridge after the instability occurred showed the experience of instability had minimal effect on the isolator stiffness properties.;This dissertation also presents the results of analytical investigations into the seismic response of conventional (not isolated) curved steel plate girder bridges. Different modeling techniques—spine beam, plate-and-beam, and 3D finite element—of curved steel plate girder bridges and their effect on the seismic response were investigated and limitations on their application identified. The spine beam model was able to capture the global seismic response of the bridge and give reasonable estimates of the column forces. However, this model was unable to capture the response of local components such as bearings and cross-frames. The response of the plate-and-beam model was comparable to the 3D finite element model. This model may be used to design local components such as bearings and cross-frames.;In addition, 3D finite element models of the curved steel plate girder bridge were developed to determine the influence of the stiffness of the cross-frames, girders, shear connectors, and reinforced concrete deck on the distribution of seismic forces between the girders and the cross-frames. It was found that the cross-frames were most effective in transferring the seismic forces to the bearings when the cross-frame-to-girder stiffness ratio was about 3.0. The shear connector stiffness had negligible effect on the framing action between the deck and girders. The number and spacing of shear connectors required for service loads, according to AASHTO Specifications, is sufficient to achieve this framing action. |
