Modelling of deep tunnel behaviour in clay

Several aspects of the ground response to tunnel excavation in deep clay formations have been studied. Such strata may be only lightly overconsolidated and an unsupported tunnel could have a high stability ratio. Consequently, at large depths there may be construcz ion difficulties similar to those encountered in soft ground at shallow depths. Problems which may be encountered are squeezing ground, i.e. a continually increasing volume loss with time, and possible collapse after a 'stand-up' time. A range of parameters likely to influence the time dependent deformation behaviour at an unsupported tunnel face was investigated in a number of small scale model tunnel tests. The tests were performed in large cylindrical samples of kaolin clay under axisymmetric conditions, with measurement of tunnel face displacement and pore pressure changes in the clay. After removal of the tunnel face support, soil intruded into the tunnel at a rate which gradually increased with time until a constant rate was reached. The steadily increasing deformation was frequently associated with increasing pore pressures close to the tunnel face as water flow occurred towards the tunnel face due to the changed boundary conditions. Two parameters, the initial pore pressure in the soil and the initial load factor, were shown to have a major influence on the time dependent behaviour, and were incorporated into a new deformation time factor. Initial undrained pore pressure changes caused by the removal of tunnel face support have been compared with simplified closed form solutions for the thick cylinder and thick sphere analogue. As a result two zones were identified at the tunnel face, in which either approximately cylindrical or spherical behaviour was observed. The small scale tests were modelled numerically using the finite element program CRISP. Elasto-plastic soil behaviour and consolidation were included in the analyses. Although the predictions were affected by the complex geometry and boundary conditions of the model tests, the mechanics of the time dependent deformations were demonstrated. Deformations at the tunnel face were poorly predicted and pore pressure changes were confined to smaller zones than in the model tests. However, comparisons of the finite element predictions and the closed form solutions for the plastic zone were more favourable. Longe term pore pressure predictions showed only limited agreement with the experimental behaviour. For both the model tunnel and thick cylinder, CRISP predicted localized zones of softened soil close to the unsupported boundary which developed over a very short time period. The rapid local drainage implies that time dependent movements will be observed which are governed by changes in pore pressure and water flow.