Response of Lead Rubber Bearings in a Hybrid Isolation System During a Large Scale Shaking Experiment of an Isolated Building

Seismic isolation systems have been proven to provide superior performance and meet continued functionality performance objectives for many facilities around the world, and are thus being considered for the future generation of nuclear power plants in the United States. Experimental simulation of a hybrid lead-rubber isolation system for a 5-story steel moment frame was performed at Hyogo Earthquake Engineering Research Center (E-Defense) of the National Institute for Earth Science and Disaster Prevention in Japan. The isolation system was developed for the seismicity of a potential nuclear site in Central and Eastern United States (CEUS) site. The isolation system was tested to displacements representing beyond design basis ground motions at the CEUS site and design basis ground motions for a Western United States. Forces in the lead-rubber (LR) bearings were measured by an assembly of load cells. The design of the isolation system was constrained by the experimental setup. The light axial loads on the system necessitated the use of a hybrid system of elastomeric bearings and rolling bearings, known as cross linear (CL) bearings. The CL bearings provided support beneath some of the columns without contributing to the system base shear, so that the target displacement at the desired isolation period could be met. Additionally, the CL bearings provided substantial resistance against the tensile demands generated by overturning as a result of the light axial loads. The following behaviors, many of which have been observed before, were observed in the response of LR bearings during this test program: (1) pinching near the center of the measured bearing hysteresis loop, attributed to the small size of the lead plug; (2) loss of characteristic strength over the duration of an excitation, associated with heating of the lead plug; (3) no loss of shear resistance at large displacements due to the stabilizing influence of the CL bearings; and (4) transfer of axial forces from LR bearings to CL bearings at large displacements, referred to as the load transfer effect, causing the LR bearings to sustain tension that was not induced by overturning. The load transfer effect, occurs due to the rigidity of the frame system connecting the bearings, the discrepancy in stiffness between the CL and LR bearings in the vertical direction, and the effective decrease in stiffness of the LR bearings at large horizontal displacements. A numerical simulation model that represents current numerical approaches for design was developed for the isolation system and the structure. The lead-rubber bearings were modeled with a bilinear force-displacement relation with uncoupled behavior in the horizontal and vertical directions, referred as the uncoupled bearing model. Due to the amplitude dependence of the bearing response, the parameters of the uncoupled model were calibrated independently for each simulation to assess the experimental LR bearing response. Although the uncoupled bearing model could produce base shear and bearings displacements that closely matched the experimental response, the peak bearing responses (base shear and horizontal displacements) were not captured by the uncoupled bearing model. The revised bounding analysis methodology was investigated to determine if the peak bearings responses could be reliably bounded with this approach. The bounding analysis was not 100% reliable to bound the observed experimentally peak horizontal displacement and peak base shear of the LR bearings due to spectral variation of the excitations. However, the new bounding analysis procedure that considers the responses of both upper bound and lower bound to bound both peak displacements and peak forces, was found to be an improvement over current design practices. The uncoupled bearing model could not predict the load transfer effect that was observed during the experiment. Thus, a multi-spring LR bearing model with coupled behavior in the horizontal and vertical directions that could predict the load transfer effect was developed and validated. The numerically predicted horizontal responses obtained from the multi-spring bearing model and uncoupled bearing model were nearly identical. Significant portions of this dissertation were taken from a report (Ryan et al. 2013a) prepared for the sponsor one year following the test. The author of this dissertation worked collaboratively on that report with other authors. All data from the experiments is permanently archived and publicly accessible in the NEES Project Warehouse (Ryan et al. 2013b, 2013c, 2013d).