Dynamic soil-foundation interaction is an important consideration in the design of structures subjected to dynamic loads such as earthquakes, wind, machine vibration, vehicle loading, and impacts. One reason for this importance is that soil-structure interaction (SSI) can have detrimental effects on the dynamic response of soil-foundation-structure systems. However, present theories and solutions for dynamic soil-pile interaction can be quite complex and contain several parameters that are not known with a high degree of certainty. Additionally, several aspects of dynamic soil-pile interaction problems are difficult to characterize accurately, such as the highly-nonhomogeneous spatial distribution of soil properties, the nonlinear and stress-dependent mechanical response of soil, variable soil-pile contact conditions, and complexities of 3D wave propagation in nonhomogeneous media. Despite significant advancements in theoretical and experimental research on dynamic soil-pile interaction, many of the available simplified approaches as well as sophisticated numerical models fail to accurately capture the observed responses from realistic multi-modal experiments.

To help bridge the knowledge-gap between existing theories and experimental observations for dynamic soil-pile interaction problems, a program of full-scale dynamic field tests were performed in this study using two identical H-piles at the same site containing soft clay. One pile was installed in the natural soil profile, and the other was partially embedded in an improved soil-cement zone. A new servo-hydraulic inertial shaker testing system and modular pile-cap were developed, then used to perform forced-vibration tests on the piles using random vibration techniques. Three different types and intensities of broadband excitation were applied to the system using the shaker installed on the pile cap in three different testing configurations. For the first time, the multi-modal vertical-eccentric (VE) dynamic test, first developed for scaled-model geotechnical centrifuge experiments, was performed on full-scale pile foundations in natural soil conditions, and verified to simultaneously capture the important aspects of the vertical, horizontal, and coupled horizontal-rocking modes of vibration. A new set of theoretical centroidal accelerance solutions was developed for the dynamic response of soil-pile systems for which an inertial shaker moving in rigid body motion with the pile-cap provides the excitation.

On the theoretical side, an existing approximate method was evaluated against the experiments as well a rigorous 3D boundary element program. It was shown that the approximate method lacks accuracy in the low frequency region and also in characterizing the vertical mode of vibration for the actual soil and pile conditions encountered. Two sets of shear modulus and damping profiles were introduced based on cone penetration tests and widely-used empirical design equations, and the effect of the dynamic shear strain level in the soil surrounding the pile was investigated. The numerical models were then calibrated to minimize the mismatch between theory and experiment by introducing a multi-modal error function which accounts for the three centroidal modes of vibration, while employing relative weighing factors developed through parametric studies and engineering judgment. The method of impedance modification factors (IMF) and the advanced three-domain computational disturbed-zone model developed from centrifuge experimentation were extended for the first time to the full-scale pile tests and natural soil conditions. The results verified that the IMF method can capture the experimental response for the pile in unimproved soil very well. The three-domain disturbed-zone computational continuum model showed promise for simultaneously capturing the experimental centroidal horizontal and rotational peak frequencies using modulus and damping profiles in the disturbed zone that were generated through reasonable modifications of the far-field profiles.