Evaluation of dynamic pile behavior by centrifuge tests considering kinematic load effect

Abstract

In order to predict the dynamic behavior of a pile foundation, a pseudo-static analysis is widely used to convert dynamic loads to additional equivalent inertial loads for seismic design. Among the diverse methods of modeling a soil-pile system for pseudo-static analysis, an equivalent soil-spring model using p-y curves which can consider non-linear soil behavior is most frequently used in practice. Due to the complexity of the dynamic analytical p-y models, however, the p-y curves which were determined for static or cyclic loading condition are still applied for pseudo-static analysis in the seismic design of a pile. In addition, no studies have been carried out on the pile installation effect on dynamic p-y curves during earthquakes, and same p-y curves were applied in practical seismic design regardless of installation method. Field investigation and subsequent analyses after recent earthquakes confirmed that not only the inertial effects of superstructures but also the kinematic effect induced by ground movement had significant impact on the pile foundations. Therefore, it is required to take both effects into account in designing pile foundation. However, pseudo static analysis cannot adequately consider the kinematic effect. In addition, little is known concerning the degree of contribution of the two effects, although much research on pile behavior under seismic loading has been performed. The objective of this dissertation is to suggest the dynamic pile behavior evaluating method quantitatively during earthquake by dynamic centrifuge model tests. This study is divided into three themes. In the first theme, dynamic p-y curve for pre-installed pile, which could simulate in-situ pile, was suggested by a series of centrifuge tests. The centrifuge tests were carried out for a single pile in dry sand, changing the conditions such as pile diameters, relative densities, input acceleration amplitudes and frequencies. Based on the results, the dynamic p-y curve was formed as a hyperbolic function by connecting the peak points of the resulting experimental dynamic p-y curves, which corresponded to the maximum soil resistances. In order to represent the p-y backbone curve numerically, the initial stiffness and the ultimate subgrade reaction of soil, which are required to formulate the p-y backbone curve, were proposed as a function of the properties of a pile and soil. The dynamic p-y backbone curve was compared with the p-y curves that are currently used in practice. Also, the applicability of the dynamic p-y backbone curve was verified using psuedo-static analysis, and the analysis result with the backbone curve more successfully simulated the other centrifuge tests results than that with the existing p-y curve. In the second theme, the installation effect on dynamic pile behavior was evaluated, and a dynamic p-y curve for jacked piles that could simulate a driven pile was suggested. Dynamic centrifuge tests with different types of piles in dry sand -- pre-installed, 1g-jacked and Ng-jacked piles -- were carried out to analyze the effect of pile installation on the dynamic p-y curve. According to the results of these tests, the subgrade reaction of a jacked pile in 40g was found to be greater than that of a jacked pile in 1g and a preinstalled pile in 1g. It was also found that differences in the subgrade resistance decrease with the depth of the pile. The applicability of the suggested dynamic p-y curve for the pre-installed pile was evaluated by comparison with the results of the centrifuge tests, and in addition, a dynamic p-y backbone curve for jacked piles was developed by modifying the results for pre-installed piles. In the third theme a series of dynamic centrifuge tests was carried out under a seismic loading condition, to analyze the effect of inertial and kinematic forces on a pile foundation, and to evaluate the influence of various input parameters on each component. The tests were conducted in dry sand and liquefiable saturated sand deposit. Based on test results, evaluation method of kinematic force was suggested by using inertial force of trapezoidal soil wedge. In addition, the dynamic pile behavior predicting method considering kinematic effect was suggested by combining suggested dynamic p-y curves and kinematic force evaluation method.