Analysis and interpretation of ground and building movements due to EPB tunnelling

As the number of tunnels increases to meet the demands of a world with a rapidly expanding population, the importance of tunnel design and construction efficiencies also grows; such new efficiencies must ensure minimal damage to nearby existing structures and facilities. When tunnels are constructed in an urban area, it is important, at the design stage, that the response of nearby structures can be predicted with sufficient accuracy. These predictions will affect the choice of tunnelling method and dictate the form and scope of potential ground improvement measures. While the 'Gaussian empirical method' has been shown to provide good estimates of the shape of the settlement profile in 'greenfield' conditions (i.e. where no structures are present), this method cannot be used to assess structural movements associated with the tunnelling, for which the profession now generally employs the finite element method. This thesis investigates the surface ('greenfield') and building movements induced by two bored tunnels in the central business district (CBD) of Perth, Western Australia. The tunnelling was carried out by an Earth Pressure Balanced (EPB) Tunnel Boring Machine and the stratigraphy along the tunnel route comprised normally consolidated dune sand overlying the interbedded layers of alluvial silts, clays and sands. Volume losses during EPB tunnelling were generally less than 0.29% and, as a consequence, settlements associated with tunnelling were small and no building damage was observed. The greenfield measurements (observed using electro-level beams and settlement pins in a railway yard) indicated wider tunnelling-induced settlement troughs in soil profiles comprising a higher proportion of clay layers. Time dependent or consolidation settlements observed after tunnelling was completed were not significant. Contrary to expectations, the volume losses associated with the second tunnel boring were less than those induced during the initial boring. All 'greenfield' settlement troughs were of a Gaussian nature and this form could not be predicted with an acceptable level of accuracy using both 2D (Plaxis version 9.02) and 3D (Plaxis 3D Tunnel, Version 2.0) ii finite element (FE) analyses combined with relatively advanced soil models. Reasons for the mismatch are discussed in this thesis and ways in which the Gaussian form may be predicted are investigated. Observed movement data as well as finite element back analyses for a multi-storey building in Perth CBD indicated that the stiffness of the building altered the free field (Gaussian) form of the settlement trough in its vicinity. 2D FE analyses are used to illustrate how building stiffness and soil type and/stiffness influence the shape of the predicted settlement trough. A comparison of measurements with predictions suggests that the FE approach adopted is a reasonable method for assessing soil structure interaction effects. To assist the numerical predictions, the thesis also present results from a targeted laboratory and in-situ test investigation on the upper horizons of the alluvial deposit found beneath Perth’s dune sand; little information was available on the mechanical properties of these horizons. The effectiveness of using compensation grouting in sand to reduce the volume loss due to tunnel boring and the applicability of numerical methods to model the grouting is also investigated.