Thesis: Suresh Dash

Doctor of Philosophy, Somerville College, University of Oxford, Michaelmas Term 2010

Lateral Pile-Soil interaction in Liquifable Soils

The p-y curves are widely used to define lateral pile-soil interaction (LPSI) in a Beam on Non-linear Winkler Foundation (BNWF) model, where ‘p’ refers to the lateral soil pressure per unit length of the pile and ‘y’ refers to the lateral relative pile-soil displacement. The parameters required for a p-y curve definition for normal soil condition (i.e., non-liquefied soil in this context) are well understood and have been used in practice with confidence for the last 30 years. However, these parameters for liquefied soil are still not well defined. For liquefied soil, current practice suggests a reduced strength p-y curve from its non-liquefied state p-y curve, which still keeps a very high initial stiffness and low stiffness at large deflection. However, in contrast, many experimental observations have shown that the initial stiffness of p-y curve in liquefied soil is very small and increases significantly at large deflection.

This thesis is aimed at investigating the lateral soil-pile response in liquefiable soils and formulating a reasonable p-y curve to represent this response. A detailed analysis of a set of centrifuge test results, 1-g shaking table tests and numerical studies form the basis of this investigation.

The p-y curve has been back calculated from the centrifuge test results of 13 pile groups. The study showed that the p-y curve has very small initial stiffness which increases with increase in pile displacement. The cyclic p-y curves have further been analysed to estimate equivalent viscous damping of the soil, which has shown quite high values of damping from liquefied soil, with the damping ratio going up to 50% in some cases.

To study the p-y curve response directly, a set of 1-g tests were carried out with plane strain idealisation of LPSI at a particular depth. The tests have shown that as the degree of liquefaction increases in the soil, the shape of the p-y curve changes from strain softening to strain hardening type and the ultimate strength reduces.

Based on the above experimental observations, collated element test data on liquefied soil and numerical study, a monotonic p-y curve for fully liquefied soil has been proposed. This p-y curve was derived directly by scaling the stress-strain curve, and hence, retains the characteristics of the stress-strain behaviour of liquefied soil and also shows qualitative agreement with the p-y curves observed in some other experiments. However, the associated limitations of this simplified p-y curve model, which represents a very complex and nonlinear liquefied soil, indicates the requirement of better experiments to confirm its form and to provide more robust numerical values for its wide applicability.

To demonstrate the application of the proposed p-y curve model, a well known case history of Showa Bridge pile failure in liquefiable soils was considered, which failed during the 1964 Niigata earthquake. The characteristics of the proposed p-y curve were compared with the p-y curve in conventional practice and its implication in capturing the failure mechanisms for pile design has been demonstrated. Although, the aim of this study was not to find the actual cause of Showa bridge pile failure, the analysis captured some major field observations of the failure by using the proposed p-y curve model.