MCEER/NCEER Bulletin Articles: Research
Field and Laboratory Studies on the Seismic Performance of Bridge Bearings
by J. Mander, S. Chen, D-K. Kim and D. Wendichansky
This article presents research resulting from NCEER's Highway Project. Comments and questions should be directed to John Mander, University at Buffalo, at (716) 645-2114, ext. 2418; or email: email@example.com.
The principal objective of this research is to investigate, both experimentally and analytically, the seismic performance of slab-on-steel-girder bridges before and after rehabilitating them with elastomeric bearings. The original bearings used in this class of bridge (that is typical of bridges constructed in the central and eastern United States prior to the 1980's) generally consist of a variety of low (sliding) and high (rocker) steel bearings. According to the recently published FHWA Seismic Retrofitting Manual (Buckle and Friedland, 1995) such bearings are prime candidates for replacement due to their historically poor performance in earthquakes. The three main objectives of this research are to determine the level of seismic excitation where the steel bearings perform satisfactorily; investigate whether simple retrofits to the steel bearings themselves markedly improve overall seismic resistance; and assess the performance effectiveness of different types of elastomeric bearings with and without supplemental energy dissipation capabilities.
Field and Laboratory Experiments
Two (sister) three-span slab-on-steel-girder bridges on the Route 400 highway in western New York were recently rehabilitated by the New York State Department of Transportation (NYSDOT). These bridges, shown in Figure 1, were used during large force/large displacement quick release (snap back) tests to experimentally investigate the dynamic response both before and after replacing the steel bearings (Wendichansky, 1996). Figure 2 is a view along the highway between the two bridges showing the snap back tendons which pull the right hand bridge to the left. The insert photograph in Figure 2 shows the mechanical fuse which is electrically triggered to suddenly release the tensile load in the tendons. Total lateral loads of up to 1,800 kN were applied to the bridge deck prior to the snap. Typically, the first four or five mode shapes and frequencies of vibration can be identified by this approach.
During the course of bridge rehabilitation, steel bearings were retrieved from the field and destructively tested in the laboratory under simulated constant axial gravity load and simulated reversed cyclic lateral earthquake loading. The bearings were first tested with strong anchorages to investigate stability and strength issues pertaining to the bearings themselves. The bearings were tested on reinforced concrete pedestals using the original swedged anchor bolts (also retrieved from the field) to investigate the steel bearing-anchorage-concrete pedestal system interaction. NYSDOT replaced the Northbound bridge steel bearings with conventional elastomeric bearing pads, and the Southbound bridge steel bearings with seismic isolation bearings. All of the different types of elastomeric bearing systems were tested in the laboratory to characterize their performance under large displacements.
The analytical studies have focused on conducting nonlinear dynamic time history analyses of the two bridges supported on the different types of steel and elastomeric bearings. To enable such analyses to be conducted, it was first necessary to make accurate force-displacement models of the steel and elastomeric bearings to capture the highly nonlinear behavior of these elements. Various combinations of nonlinear truss and link elements (that include gap effects) have been used for the steel bearings to capture sliding, prying, and keeper plate and/or anchor bolt fracture. Figure 3 compares experimental observations with the analytical model prediction under transverse loading for a low type sliding bearing. A new thermo-visco-elasto-plastic model has also been developed for modeling elastomeric bearing behavior both with and without a lead core. For a laminated rubber seismic isolation bearing, Figure 4 compares experimental observations for a cold temperature test (-48° C) with the analytical model prediction. With these primary sources of nonlinear bridge performance modeled, the next step was to develop overall nonlinear mathematical/structural models of the bridges. For this purpose, the experimental quick release test time histories were used to validate the computational models. Once validated, the computational models were used to predict expected performance for a large range of strong earthquake ground motions - not only motions that may be expected in low to medium seismic zones, but also very strong near-field motions that include fling effects observed in high seismic zones. In addition to investigating seismic behavior using nonlinear time history analysis, simplified analysis techniques were advanced that use: linearized elastic capacity-demand spectra, and nonlinear inelastic capacity-demand spectra.
Companion Analytical Studies
A companion project that is investigating the nonlinear performance and foundation flexibility of the two bridges is being conducted by Professor Bruce Douglas and his co-workers at the University of Nevada, Reno. The results of their analyses are being used by the University at Buffalo researchers to more accurately model soil-foundation-structure interaction. They are also investigating appropriate methods of identifying nonlinear structural behavior parameters.
Bearing behavior can be largely described by rigid body kinematics (i.e., sliding and rocking) with some yielding of critical parts such as anchor bolts, pintles and guide plates. For each type of steel bearing where sliding is possible, the laws of Coulomb friction are obeyed. Results show that by retrofitting the existing high type steel bridge bearings it is possible to provide sufficient strength and displacement capability to withstand substantial ground shaking. The weak link in the chain of force transmission may thus become the anchor bolts and/or the reinforced concrete pedestal. Experimental results demonstrate the importance of considering the flexibility of the concrete pedestal-anchor bolt system (Figure 5). The bearing assembly elastic stiffness may be determined by assessing the flexural and shear flexibility of each of the constituent bearing parts. Ultimate lateral strength may be determined using either rigid body kinematics or upper bound plastic mechanism analysis.
At extremely cold temperatures, rubber glass-hardens with the bearing behaving in a significantly nonlinear fashion. The hysteretic performance of elastomeric bearings may be characterized by a temperature dependent nonlinear Kelvin model - that is a bi-linear spring and velocity dependent nonlinear viscous dashpot. The cold temperature effects of lead-rubber bearings may also be obtained by similarly adding the visco-elasto-plastic effects of lead to the model of the rubber bearing.
Field Experimentation Observations
The replacement of the original steel bearings with elastomeric isolation bearings produced a significant change in the transverse dynamic characteristics of the bridge. The initial transverse frequency of the Southbound bridge was around 5.5 Hz (with 6% damping). This dropped to less than 2 Hz (an upper bound based on the initial elastic stiffness) and frequencies as low as 1.08 Hz (22% equivalent viscous damping) were observed due to nonlinear seismic response. This latter result was obtained with a deformation of 38 mm or around 33% of the maximum bearing displacement. Even lower frequencies (and higher damping) can be expected for higher deformations that would occur under large seismic loading conditions. For the Northbound bridge, the first transverse frequency dropped from a value of 5.8 Hz (6% damping) for the steel bearings to less than 1.8 Hz (9% damping) for the neoprene elastomeric bearings. Also, new rubber expansion joints were installed a few weeks before testing the rehabilitated bridges. Due to frictional restraint, these joints provided a noticeable contribution to the overall transverse stiffness of the rehabilitated bridges.
Analytical Predictions and Comparisons
As summarized in Figure 6, the analytical portion of this study has shown that when seated on the original steel bearings, the bridges were capable of sustaining earthquakes with peak ground accelerations of at least 0.65 g. For reasons other than seismic retrofitting, if steel bearings are rehabilitated using traditional elastomeric/neoprene bearings and also pinning one abutment to provide some anchorage for thermal expansion, then a torsional imbalance may be expected. Predictions show a seismic resistance of PGA = 0.6 g for this class of rehabilitation of the Northbound bridge. This indicates that the bearing rehabilitation of the Northbound bridge did not improve the seismic resistance of that bridge beyond the expectations of the original steel bearings. On the other hand, if the steel bearings are replaced with a well-conceived and designed seismic isolation system, the seismic resistance may be improved; predictive results for the Southbound bridge show an improvement in seismic capacity to PGA = 0.9 g for lead-rubber isolation bearings from 0.75g for the steel bearings. These results show that current seismic evaluation techniques, recommended in the recently published FHWA Seismic Retrofitting Manual (Buckle and Friedland, 1995), underestimate the real resistance capacity of slab-on-girder bridges seated on steel bearings. This is largely because simplistic elastic analysis models, when used, ignore the beneficial effects of friction in steel bearings. Also, current retrofit techniques generally recommend conventional retrofitting or the use of protective systems. This study has demonstrated that existing bridges designed using the provisions of the 1960's and 70's can sustain the design earthquakes forces. Therefore, by performing a detailed engineering analysis of the structure, unnecessary retrofitting may be avoided in some cases.
For bridges located in low to moderate seismic zones such as those in the eastern and central United States, the use of lead rubber bearings may not be necessary. Partial isolation can be achieved by replacing steel bearings with regular laminated elastomeric bearings. If extra damping or a torsionally imbalanced condition exists, then the behavior of these less expensive bearings can be augmented by using shock absorbing devices.
The use of the proposed methodology to evaluate the capacity/demand (C/D) ratio based on an inelastic response spectrum approach provides a rational basis for studying systems where the overall structural behavior is governed by different types of hysteretic response. The results obtained using this methodology show good agreement with the results obtained from rigorous inelastic time history analysis.
Buckle, I.G., and Friedland, I.M., (1995), "Seismic Retrofitting Manual for Highway Bridges," Federal Highway Administration, Report No. FHWA-RD-94-052, 309 pp.
Wendichansky, D.A., (1996), "Experimental Investigation of the Dynamic Response of Two Bridges Before and After Retrofitting with Elastomeric Bearings," Ph.D. Dissertation, State University of New York at Buffalo.
NCEER Bulletin, July 1996, Vol. 10, No. 3