Abstract

With an increasing population and expanding urban areas, there is a need for tunnels that are used for efficient transportation, water supply, sewage disposal and communications. In urban areas, tunnelling is mainly carried out in soft ground and this process inevitably leads to a change in stress state of the ground with associated changes in strains and displacements, which could potentially result in damage to existing surface and subsurface structures. In low permeability soils, a change in stress state induces excess pore pressures that, in the long-term, dissipate into a new equilibrium regime leading to further ground displacement and complex loading on tunnel supports. As part of the planning process, knowledge of the magnitudes of these ground displacements as well as loading on the tunnel lining in both the short and long-term is often required. Due to the complexity of interactions between groundwater flow, soil-tunnel lining interaction during consolidation, finite element analysis is the only practical method of predicting these ground movements and ground loading on the tunnel lining.

In this thesis, the finite element method was employed to evaluate the short and long-term tunnelling induced ground response at St. James's Park, London. To increase the accuracy of the results, an advanced soil model with the critical state and subloading surface concepts has been proposed. The proposed soil model has all the necessary soil model features that, to date, have been shown to improve predictions in tunnelling problems. The features include stiffness anisotropy, plastic deformation within the yield surface, criteria for stress reversal, small strain stiffness and its non-linearity with strain response, and Matsuoka & Nakai failure criterion. Based on the comparisons between the computed ground response and field measurement data, it was found that for accurate predictions of ground displacements and pore pressure response, stiffness anisotropy must be incorporated in the elasto-plastic soil model and with the K0 value of 1.2 assumed for London Clay. Despite the utilisation of such an advanced soil model, the width of the settlement troughs induced by tunnelling in overconsolidated clay computed using the finite element method is still greater than those observed in the field. The detailed comparisons presented in this research have highlighted the sources of discrepancies between the ground response computed from finite element analysis and field measurement data.

Detailed examination of the field measurement data has suggested that, for St. James's Park, the long-term subsurface ground response mechanism consists of a combination of i) swelling of the soil at a confined zone above the tunnel crown, ii) consolidation of the soil on either side of the tunnel extending to a large offset from the tunnel centre-line, and iii) above these zones, a rigid body downward movement. In order to simulate such a mechanism of the long-term subsurface ground response, suitable permeabilities for divisions within London Clay and non-uniform drainage conditions at the tunnel lining had to be modelled.

The research has improved the understanding of long-term ground response and has illustrated how the long-term ground response is affected by uncertainties in the following factors: stiffness anisotropy, permeability anisotropy, volume loss, K0 and tunnel drainage conditions. Based on this improved understanding, a suite of finite element analyses was carried out. The results obtained from these analyses were generalised and presented in the form of normalised charts.