4.5 Discussions

The purpose of the tests performed in this chapter was mainly to study the behaviour of an excavation upto the failure of the retaining system. Although field data from the monitoring of the excavations are available, these are mostly data under working condition. Very little data is available from actual excavations that have failed. In order to understand the progress of failure from small deformation to total collapse, the centrifuge model is an effective way to produce these phenomena at a model scale to provide insight into the mechanism. The failure situations simulated and the results monitored shown in this chapter would be very difficult and expensive to obtain in the field.

4.5.1 Time dependent behaviour of excavation

From the observations for the three different types of excavation, it is noted that pore water pressures in retained soil all decrease as the excavation starts and the wall moves towards the cut. These phenomena have been reported in a number of site monitoring data and were also measured in previous centrifuge model tests (Lee et al, 1995). The magnitude of such drop is greater at locations nearer to the wall. Figure 4.29 gives a comparison of pore pressure changes for transducer PPT1 obtained from tests To1, To2, Ts1 and Ts3. It can be seen that the drop of pore pressure for an unpropped excavation is greater than that for the propped ones. This is clearly attributed to the larger deformation of the wall in an unpropped excavation.

Pore water pressure changes cause a reduction of lateral stress in the form of negative pore pressure during excavation which can be characterised as nearly undrained during the excavation. After excavation, the pore pressure then dissipates with time as reflected in Figure 4.29. The rate of dissipation is observed to be quite slow for marine clay. During this period of pressure dissipation, it is expected that the wall will continue to move, though at a much slower rate as compared to the rate during excavation.

The phenomenon can be further explained using the stress path plot as shown in Figure 4.30. Before the excavation, the initial stress state is located at Point A on Ko line. When the excavation starts, due to a reduction of lateral pressure, the total stress is reduced and travels along the line A-C. The expansion of soil leads to a rise of negative pressure as observed in the tests. In the short term, with low permeability of soil, the effective stress is nearly unchanged as indicated by the line A-B and negative pore pressure is developed. In the long term, that is after excavation or during a very slow excavation, the dissipation of negative pore pressure can be represented by the line B-C.

Due to the time dependent behaviour of excavation in soft clay, the related mechanism should be investigated carefully. Regarding the short term behaviour of excavation, the deformation of the wall and the ground movement during the excavation are mainly caused by the reduction of lateral stress. The soil expansion leads to the rise of negative pore pressure which is beneficial for the stability of the excavation in the short term. Slow excavation work or in the long term, the dissipation of negative pore pressure would lead to a further deformation of the wall and hence the settlement of soil surface and may eventually lead to the failure of the retaining system.

4.5.2 Soil movements

Obviously, the soil movements around the wall were quite different between the unpropped and propped excavations. A comparison of settlement-time plots for LVDT2 between tests To1, To2, Ts1 and Ts3 is given in Figure 4.31. The comparison reinforces the earlier finding that an unpropped wall (tests To1 and To2) had very low capability to limit the ground movements. As indicated in the curve for test Ts1 shown in Figure 4.31, the upper strutted wall provides improved capability to limit the ground settlement as compared to unstrutted wall. The reduction of ground movement is very much more in the case of a wall supported by two struts, as indicated in curve Ts3 plotted in Figure 4.31. Due to the presence of the lower strut, the surface settlement is less than 3 mm, which is quite different from the others, especially in the later stages.

At the same excavation depth of 50 mm (5 m in prototype scale), the magnitudes and profiles of surface settlement of tests To1, To2, Ts1 and Ts3 are plotted in Figure 4.32. At this excavation stage, the lower strut for test Ts3 has not been pushed in, thus

the magnitude of surface settlement were quit close to that observed in test Ts1. Because there was no upper strut in tests To1 and To2, the settlements are significant larger than those of Ts1 and Ts3.

4.5.3 Failure Mechanism

The photograph of failure pattern and the corresponding sketches of tests To1, Ts1 and Ts3 are shown in Figures 4.33 (a), (b) and (c), respectively. The unpropped excavation exhibits a linear slip surface which is at 45^o to the vertical representing an undrained behaviour. The wall rotates around a pivot near the wall toe. On the ground surface near the wall, some tension cracks in the retained soil was observed. Such tension cracks and the slip surface cause the ground to subside significantly.

For test Ts1 with only an upper strut, the middle portion of the wall was observed to bend towards the cut at the initial excavation level. The strut effectively limits the soil movements as compared to the same stage of an unpropped excavation as shown in Figure 4.33. In the subsequent excavation stages, due to insufficient wall embedment, the soil beneath the excavation cannot provide enough passive earth pressure so that the wall begins to fail by rotating around the support point of the upper strut.

In test Ts3, the second strut was installed to control excess wall deflection. As a result,

the soil movement significantly reduced. The observed surface settlements shown in Figures 4.31 and 4.32 clearly illustrate the effectiveness of the lower strut. In such case, a critical consideration would be to prevent a toe kick-out. For the propped excavation in soft clay, the penetration depth is an important aspect for stability consideration. Figure 4.33(c) shows a failure mode for the toe kicks-out.

The comparison of bending moments for tests To1, Ts1 and Ts3 measured at an excavation depth of 50 mm is given in Figure 4.34 which indicates a significant difference of bending moment on the different strutted walls in the model tests. In test To1, the cantilever effect is shown from the profile of bending moment in the wall. Due to the large deformation of the wall, the strength of retained soil was fully mobilised, so that the earth pressure acting on the wall is small as in the Rankine active state. In contrast, the strutted wall exhibited much higher bending moment as shown by Ts1 and Ts3 in Figure 4.34, especially in the case of Ts3 where two level of struts were used. Although the small model tests cannot provide very accurate measurement of bending moment profile, but such observation is in general agreement with the understanding on the bending moment in the design of retaining wall.

The stage photographs of test Ts2 (model excavation depth is 50 mm as compared to 150 mm for all other tests in series 1) is shown in Figure 4.35. In this test, as indicated in Table 3.4, the wall was supported with upper strut and the total excavation depth is

50 mm. Below the excavation level, it is the real soil instead of zinc chloride solution as used in the other tests in this series. Failure slip surface was observed at an expected critical failure moment based on the previous tests and calculations. As shown in Figure 4.35, the slip surface appeared but was not fully developed. There was no total collapse as compared to the tests with 150 mm total excavation depth. With both effects of residual strength of soil and the drop of pore water pressure, a stable state after a small slip occurred was maintained as shown in Figure 4.35. Partial data of this test will be shown in Chapter 5 as a comparison with the tests conducted in the second test series.