In this work, a series of dynamic centrifuge tests were performed on a rigid foundation supporting a containment structure. This foundation was supported on different soil stratifications. The seismic response of the soil bed was compared for these soil stratifications. It was observed that seismic response of layered, inhomogeneous and remediated soil affects the overall behaviour of the rigid foundation. It was concluded that inherent layering could change the acceleration experienced at the base of the structure. This effect varied from negligible to profound. Tilt and ultimate settlement of the foundation was strongly dependent on the residual resistance provided by the sheared soil, which was a function of the degree of softening, intensity of shaking, local permeability variations and transient flow conditions from the surrounding soil. The likely zone of influence affected by the interaction between the soil and the structure was identified.

It was shown that the ‘vibration isolator’ capabilities of the liquefiable layer can be exploited while designing remediation schemes for liquefiable sites. This result can have important implications to the practising engineer. It was concluded that the degree of heterogeneity of the soil profile has a greater role to play in site-specific response studies than presently accounted for. As part of the experimental work a ‘miniature air hammer’ was designed which was used ‘in flight’ to characterize the soil softening during consecutive earthquakes providing valuable information.

The role of Soil-Structure interaction was also investigated and it was clearly seen that the foundation input motion was different from the free field motion and it was often higher in magnitude. No visible input loss was seen for low intensity earthquakes. In high intensity earthquakes such assumptions of neglecting kinematic interaction may be justified. Site response analysis for the layered soil also showed that there was visible advantage of using a 2D non -linear program over 1D program, as this would enable an economic design.

Numerical simulation was successfully carried out for the reported centrifuge test results by using the finite element code SWANDYNE. The success was due to incorporating the criteria developed in the numerical work presented in this thesis for proper mesh selection and time discretization. It was shown that most of the dynamic simulations reported in the literature use a far coarser time step than required for non-linear liquefaction- type problems. Numerical analysis has also shown the effects of frequency content on the rate of pore pressure generations in liquefiable soil. Careful modelling and selection of sensible parameters have shown that the present ‘u-p’ formulations can be used for predicting the deformations, accelerations and excess pore pressure for different strength earthquakes and bearing pressures in layered soils. 

         The seismic soil structure interaction of a rigid foundation embedded in non-homogeneous, layered and liquefiable soil is a complex problem. This boundary value problem has been investigated using dynamic centrifuge modelling and numerical modelling based on finite element. In this thesis, it is shown that these methods can complement each other in providing an insight into the SSI interaction during earthquake shaking. The model test data obtained from the present centrifuge tests can be used to improve the design of the foundation of a high risk structure like a nuclear containment in an earthquake zone founded in inhomogeneous soil. These results are equally relevant for other heavy foundation structures like power plants, industrial structures and buildings resting on raft foundations.