Tunnelling is a sustainable solution to traffic problems in congested cities, but often affects existing underground infrastructure such as pipelines due to the inevitable ground movements associated with it. There are not enough detailed case histories of the effects of tunnelling on existing buried pipelines, while current state of the art methods of assessing the complex pipe-soil interaction problem often rely on simple, but unproven, analytical solutions in elasticity. The findings reported in this thesis enable practitioners to assess the effect of tunnelling on pipelines more effectively and with increased confidence. This is achieved by identifying pipe-soil interaction mechanisms, formulating understanding of relative pipe-soil rigidity and quantifying the effect on continuous and jointed pipeline behaviour. The research was conducted in three stages, comprising centrifuge modelling, numerical analysis and monitoring of an existing large diameter, concrete lined cylinder pipeline in North London.
Large diameter prototype pipelines were modelled at 1:75 scale in the Cambridge centrifuge. The experiments were carried out in dense dry sand, during which the model continuous and jointed pipelines were buried at different depths and proximities to a model tunnel. The model tunnel induced volume loss at a controlled rate that enabled studying greenfield and model pipeline behaviour over a wide range of ground movement. Current knowledge of tunnel-induced ground movement in granular soil was improved by relating the lateral extent of the settlement trough to shear strain. Improved upper and lower approximations of the settlement trough width parameter, K, were formulated and shown to relate well with published field data.
Pipeline behaviour was found to be governed by a number of global and local interaction mechanisms, which affected relative pipe-soil rigidity and subsequent pipeline behaviour. Relative pipe-soil rigidity was found to be a function of pipeline sectional properties, the deformation at pipeline level (and hence the problem geometry) and the stiffness of the pipe material and the soil. A method was formulated to estimate the representative stiffness along the affected length of a pipeline based on the expected greenfield ground movement. By means of an analytical investigation of pipe-soil interaction, the importance of defining greenfield data in relation to the requirement for interaction analyses was illustrated. This provided the basis for formulating a rational pipe design process.
Monitoring of an existing large diameter pipeline subjected to tunnelling and pipe jacking provided the opportunity to apply novel fibre optic sensors to monitor near-continuous pipe strain during construction. The field data showed that pipelines in practice may 'change' behaviour, from initially reacting as continuous pipelines, to gradually changing to fully jointed pipelines in accordance with joint and pipe section behaviour. In conjunction with centrifuge observations and analytical modelling, four design load cases were proposed to account for the most critical stages of pipeline behaviour likely to be encountered during tunnelling. A rational design process was formulated, based on initially applying simple design methods before developing a more complex analysis if the pipe-soil system is sufficiently stiff. Guidance is given on setting suitable limits for design and monitoring, applying interaction mechanisms likely to affect pipeline behaviour and identifying critical parameters required for analysis. Throughout the design process assessment of relative pipe-soil rigidity is considered fundamental in understanding pipeline behaviour, a theme which is drawn throughout the thesis.