In seismically active, densely populated areas, buildings within a city block interact with one another during an earthquake. This phenomenon, whereby two adjacent buildings interact with each other through the surrounding soil during an earthquake, is often called structure-soil-structure interaction (SSSI). SSSI effects are less understood than soil-foundation-structure interaction (SFSI) effects. There are a lack of high-quality case histories that clearly show SSSI, which is a key reason that SSSI is less understood than SFSI. SSSI effects can potentially be detrimental and lead to more damage within the soil-foundation-structure system. Accordingly, it is important to understand when SSSI effects are important, and include them in engineering analysis and design when necessary.
This dissertation describes three centrifuge tests designed to simulate SSSI and SFSI case histories. All centrifuge test described within this dissertation were performed at the University of California at Davis Center for Geotechnical Modeling (UCD-CGM). The first test, Centrifuge Test-1, examined two inelastic moment-resisting frame structures atop a bed of dry, dense sand. One frame structure represented a prototypical three-story moment-resisting frame structure founded on spread footings. The other frame structure represented a prototypical nine-story moment-resisting frame structure founded on a three-story basement. The two structures were located a significant distance apart, and thus, SSSI effects were masked. Accordingly, the purpose of Test-1 was to examine SFSI effects of inelastic frame structures and to serve as a baseline test (i.e., a control test). The second test, Centrifuge Test-2, examined the same two structures atop a bed of dry, dense sand. In Test-2, however, the two structures were located adjacent to each other. Therefore, the purpose of Test-2 was to examine SSSI effects. By comparing results from Test-1 with results from Test-2, insights into SSSI effects were made.
The third test, Centrifuge Test-3, examined three structures atop a bed of dry, dense sand. Two of the structures were identical, and represented prototypical three-store moment-resisting frame structures founded on spread footings. These structures were nearly identical to the three-story structures used during Test-1 and Test-2. The third structure was a rigid rocking wall founded on a large mat foundation, which was identified as the transmitter structure. One frame structure, which was identified as the receiver structure, was located adjacent to the transmitter structure. The other frame structure, which was identified as the control structure, was located a significant distance away from the transmitter-receiver pair of structures. The design goal of the transmitter-receiver pair was to maximize interaction between the two structures. By comparing the seismic response of the control structure with the seismic response of the receiver structure, insights into SSSI were made.
The earthquake motions employed during the three centrifuge tests described within this dissertation are critically important. A preliminary centrifuge test (Test-0) was performed after an earthquake motion selection process. The purpose of Test-0 was to calibrate a suite of earthquake motions that could be used at the UCD-CGM. This dissertation describes an earthquake motion selection and calibration process that future researchers can use to create test-specific earthquake motions for their research projects.
Kinematic SFSI and SSSI effects were examined during Test-1 and Test-2. Specifically, the earthquake motions recorded in the free-field at the surface, which is the earthquake motion most often used by practicing engineers for dynamic analyses, was compared to the earthquake motion recorded under the basement, in the soil. Because of kinematic interaction effects, which include base slab averaging and embedment effects, the earthquake motion recorded under the basement has smaller amplitude and smaller high-frequency content than the earthquake motion recorded in the free-field at the surface. This is an established observation, and Test-1 and Test-2 data corroborate with current kinematic interaction estimation procedures. When comparing the results from Test-2 with Test-1, however, it was seen that basement-level earthquake motion differed less from the free-field surface motion during Test-2. This result indicates that kinematic interaction effects may be masked in urban environments.
The seismic responses of the shallowly embedded frame structure footings were also examined during Test-1, Test-2, and Test-3. More specifically, the vertical displacement (settlement and uplift), horizontal displacement (sliding), and rocking were examined. By comparing results from Test-2 with results from Test-1, it was seen that the deeply embedded basement "restrains" the adjacent footings. In other words, the adjacent footings displace and rotate less than the footings that are not adjacent to the basement (i.e., the free footings). This asymmetrical footing response leads to additional demands on the superstructure, which may be unacceptable. In addition, the seismically-induced column moments measured above the restrained footings are larger than those measured above the free footings. Therefore, SSSI effects were seen to be potentially detrimental (i.e., lead to more superstructure damage) during Test-2.
During Test-3, the same footing restraining effect observed in Test-2 was found to be not as large. However, there is evidence that the transmitter structure affected the seismic response of the adjacent receiver structure. More specifically, as the transmitter structure rocked and settled during the higher-intensity earthquake motions, the adjacent footings of the receiver structure did uplift, and this caused asymmetry in the superstructure. A general observation from Test-3 is that the seismic footing response of frame structures founded on shallowly-embedded footings is erratic. Future work in this area will examine possible explanations for the observed erratic response.
