Seismic Analysis and Response of Highway Bridges with Hybrid Isolation

AbstractIn this study hybrid isolation is proposed as an alternative to (full) seismic isolation for minimizing bridge damage in strong earthquakes. In this technique isolators are only used at the abutments, and monolithic or pinned connections remain at the piers. Both full and hybrid isolation can be shown to reduce the demand on critical members such as columns and foundations, but hybrid isolation also reduces superstructure displacements and the corresponding size of abutment movement joints. A key factor, however, to the success of hybrid isolation (where isolators are placed only at the abutments) is the ability of the abutments to take high lateral loads without damaging the underlying piles.This technique is applicable to a wide range of bridges including those not suitable for full isolation, such as a bridge with a continuous superstructure that is monolithic with its piers. Hybrid isolation may be used for both new construction and retrofit work.This research has focused on validating hybrid isolation as a viable alternative to (full) seismic isolation, gaining insight into the response of bridges with hybrid isolation, and quantifying the advantages and disadvantages of this approach. For this purpose, the AASHTO Simplified Method for analyzing fully isolated bridges (AASHTO, 2010) was modified to include yielding substructures (as necessary), and soil-abutment-pile-structure interaction. Of note is a parallel study conducted on the stiffness and capacity of piles in sloping ground, since many abutment piles are located in sloping embankments under the end spans of bridges. This numerical investigation used the Deep Foundation System Analysis Program (DFSAP) and the results indicated the effect of slope on lateral stiffness and capacity of long, laterally loaded piles can be significant. Reduction factors for pile stiffness were obtained, and utilized in the improved Simplified Method for studying the effect of embankment slope under the end spans on the response of hybrid isolated bridges.The methodology of hybrid isolation was numerically validated by conducting finite element analyses of a prototype bridge using SAP2000 for several different configurations of the bridge: (1) bridge with conventional details and unrestrained girders at the abutments, (2) fully isolated bridge, and (3) hybrid isolated bridge with/without soil-abutment-pile-structure interaction. The numerical analyses not only validated the advantages of hybrid isolation, but also its limitations by giving insight into the load sharing mechanism at the abutments. In addition, the numerical analyses illustrated the accuracy and reliability of the improved Simplified Method.To further validate the hybrid isolation methodology, a series of experiments on a large-scale model of a hybrid isolated bridge were conducted using the NEES shake table array in the Large-Scale Structures laboratory at the University of Nevada Reno. With a scale factor of 2/5th, a three-span curved bridge model had a 12-ft wide superstructure consisting of three steel girders and a concrete deck slab, and was 145 feet long with a radius of 80 feet at the centerline. The model spanned four shake tables and was subjected to scaled ground motions with increasing amplitudes from the 1994 Northridge earthquake. The results showed that hybrid isolation (1) was effective at keeping the piers elastic under the Design Earthquake, and essentially elastic at the Maximum Considered Earthquake, (2) reduced the superstructure displacements by a factor between 3 and 4, and (3) increased the shear force (lateral load) demand on the abutments by a factor about 3, compared to full isolation. In addition, the experimental results showed the isolators were stiffer than assumed in design, with properties that were very dependent on shear strain.A 3D finite element model of the above bridge was developed, and analyses were performed with actual material properties. Overall, good agreement was obtained with the experimental results, but it was found that the conventional properties used for the design of lead rubber isolators (target properties) needed to be modified for the relatively small strain range that is typical for hybrid isolators (0-50%)