Soil-structure interaction (SSI) influences the seismic response of buildings supported by pile foundation. Under a significant input motion that causes damage to piles, the SSI is influenced by the nonlinear behaviors of the free ground (i.e. site nonlinearity), soil close to the foundation (i.e. local nonlinearity), and structural members (i.e. structure nonlinearity). The interaction among piles (i.e. the effect of pile groups) complicates local nonlinearity when two or more piles are installed in a narrow space. Experimental and analytical studies have indicated that lateral resistance per pile in a pile group tends to be smaller than that of a single pile, and depends on its location. However, few studies have conducted on lateral resistance of pile groups under load in the oblique direction. Therefore, this study aims to investigate seismic response of pile groups subjected to input motion in the oblique direction.
The study consists of two steps. In the first step, a shaking table test and its simulation analysis are conducted to investigate influences of direction of input motion on dynamic behaviors of a structure supported by a pile group. In the second step, a cyclic loading analysis on full-scale pile groups is conducted to evaluate hysteresis characteristics of pile-soil springs. The analyses are based on the nonlinear three-dimensional finite element method.
The shaking table test was for a rigid body supported by a 5 x 5-pile group. Piles were modeled using acrylic cylinders with a diameter of 12 mm and a length of 421 mm, and installed in the dry Toyoura sand deposit. Six types of input motion, which contains two different waveforms and three different amplitudes, were applied to 0-degree or 45-degree direction. The experiment provided following findings:
1) The amplification characteristic of the soil-structure system did not depend on direction of input motions. This fact indicates lateral resistance of a whole of the pile group does not depend on direction.
2) Subgrade reaction of each pile differed significantly depending on the pile location and the direction of input motion. This difference was considered to cause that of bending moment.
The cyclic loading analysis was for a 2-pile group, 3-, and 5-pile groups in the series arrangement, and 3 x 3- and 5 x 5-pile groups in the square arrangement. Piles had a diameter of 600 mm and a length of 10 m. Soil was assumed to be homogeneous sand deposit, and modeled using perfect elasto-plastic bodies governed by the Mohr-Coulomb yield criterion. The amplitude of pile displacement was gradually increased, and the loading direction was varied at the interval of 15 degrees. The analysis provided following findings:
1) Maximum subgrade reaction and hysteresis dissipated energy of each pile strongly depended on the loading direction. The relationship between these characteristics and loading direction differed significantly depending on pile location.
2) Equivalent damping ratio of the equivalent single pile did not depend on the loading direction as long as nonlinearity in soil close to piles progressed.
