Abstract
The many large earthquakes of recent years, including those at Northridge, Kobe, Turkey, Taiwan and India have highlighted the risks that these events can pose to life and infrastructure. Whilst these events cannot be prevented and can only be predicted in the broadest sense, in that risk levels can be assigned to particular areas, a knowledge of the behaviour of structures under earthquake loading can allow them to be designed to survive these catastrophic events
This thesis details the results of an investigation into the behaviour of slopes of liquefiable sand under earthquake loading and the influence of these laterally spreading slopes on inclusions such as pile foundations passing through them.
A study of the behaviour of these slopes has been carried out using the techniques of dynamic centrifuge modelling. Eight tests were carried out on laterally spreading slopes and a further five on slopes containing instrumented pile foundations. Each model was subjected to a sinusoidal input motion using a Stored Angular Momentum earthquake actuator, causing liquefaction of the sand and lateral spreading of the slope. Data from instruments measuring acceleration, fluid pressure, total stress and bending moment were logged during the earthquake and analysed to reveal information relating to the performance of these slopes during earthquakes.
The experiments highlighted the importance of the dilation of liquefied soil to the behaviour of liquefiable slopes. Slope movements were limited by dilation during each cycle of the earthquake which prevented significant soil flow velocities from building up and large pressures were applied to piles from the liquefied soil owing to dilation of soil close to the pile foundation. The lateral pressures measured were observed to exceed the recommendations of the Japan Road Association code of 1996. Further research is needed to develop better design guidance for this situation.
The data from these experiments was compared with the results of a number of numerical models constructed during this work in order to simplify the prediction of the behaviour of these slopes. The displacements of the slopes were compared with those predicted using a Newmarkian sliding block approach modified to include the effects of excess pore pressures. This was shown to give reasonable agreement with centrifuge data, though it requires input of a measured or predicted time-history of excess pore pressure to calculate threshold accelerations. Using linearised pore pressures overestimated the measured slope displacements as dilation limits the movements measured in the centrifuge tests. The bending moments induced in pile foundations and the contact stresses applied to them during lateral flow were modelled using a pseudo-dynamic finite element pile model, attached to the spreading soil with hyperbolic p-y curves whose stiffness varied with excess pore pressure. It was seen that this analysis method predicted many of the phenomena observed from the centrifuge models, though again this was with the aid of measured pore pressure time histories. Using linearised synthetic pore pressure time-histories gave good prediction of the measured residual bending moments but underestimated the peak values owing to the influence of dilation on the centrifuge model results.
In summary, the behaviour of liquefiable slopes and their interaction with pile foundations is significantly influenced by the dilative behaviour of the soil, both in terms of limiting slope deformations and in exerting large transient pressures on piles.