3.5 General Design of Centrifuge Model

In the present study, three types of tests were proposed to be carried out. The first type was to model the excavation retained by a cantilever diaphragm wall. The second was an excavation in which the wall was propped by an upper support. The third employed support at two levels to prop the diaphragm wall. The lateral supports were provided by an already formed slab. So the tests generally represented an excavation supported by slabs in a top-down construction. In order to simulate the field situation shown in Figure 2.3, the lower slab was installed in the middle of the excavation progress during centrifuge flight.

All the tests were conducted using the National University of Singapore (NUS) Geotechnical Centrifuge as shown in Figures 3.1 and 3.2. The centrifuge has a payload capacity of 40 g-tonnes and a maximum working g-level of 200g. Based on the working space of the swing platform of the NUS centrifuge and the usual working acceleration of 50-200g, the internal dimensions of the model container for the experiments was selected to be 570 mm long by 380 mm wide by 500 mm high and the working acceleration was selected to be 100g.

For relatively small size box shaped excavation, the corner effect plays an important role for stability and deformation of excavations. To simulate this, three-dimensional model excavation should be carried out. For large size excavation, the mid-section may be simplified as a plane strain problem. Model tests involving two-dimensional plane strain situation can be conducted and controlled with relative ease. In such cases, the soil movement and the failure pattern can be directly observed through the perspex window in the non-strain plane. In view of the above advantages, two-dimensional model tests were conducted in the present work. The development of an inflight excavator for three-dimensional excavation is currently in progress at the university.

A photograph and side elevation of the model setup are shown in Figures 3.3 and 3.4, respectively. The model container consists of a stainless steel alloy back plate with a number of stiffening beams and a perspex front window 80 mm thick that is able to withstand high-g stresses. Both the back plate and front perspex can be opened to allow for the pre-cutting of the excavated soil and the installation of the retaining wall. This strong box provides a plane strain section of a long excavation and retaining wall. The internal faces of the model container are coated with silicon grease to reduce the soil-wall friction.

Based on the prediction using the classical theory on retaining wall in an uniform soil, failure zones behind the diaphragm wall would be projected at an angle of (45^o + f /2) from the base of the wall. A distance of 200-300 mm behind the wall must be adopted to accommodate this failure zone. As shown in Figure 2.5, the distribution of soil settlement as a function of distance from the cut as determined by O’Rourke is approximately 1.5 to 2.5 times the maximum depth of excavation for soft clay. Suppose the maximum depths of excavation in this study is 5 cm and 15 cm respectively, its approximate extent of influence would be 125 to 375 mm. Thus in this model, the distance from the back of the retaining wall to the container boundary is designed to be 420 mm to accommodate the anticipated failure and settlement zones.

This study was designed to investigate the behaviour of excavation in Singapore marine clay. For the first series of tests, the multi-layered Kallang formation shown in Figure 2.6 was simplified by a uniform marine clay of 200 mm thick. In practice, the diaphragm wall is rested on a hard stratum beneath the Kallang formation. The hard stratum was simulated by a stiff residual soil. The thickness of this hard stratum was 170 mm which was deemed to be sufficient to eliminate the boundary effect of the container base. For the second series, the upper soil consisted of marine clay with strength increasing linearly with depth and the lower soil was again stiff residual soil.