6.6 Analyses on Test Series 1

6.6.1 Deformed ground mesh

In Chapter 4, a series of tests, namely To1, Ts1 and Ts3, was carried out. In these tests, the ground movements develop from small strain to large strain until the collapse of the retaining wall or local failure was captured (Figure 4.30).

In the first series of tests, the soil profile is nearly uniform and for CRISP, the actual condition in the tests is not easy to simulate accurately in the analyses. Thus it is not meaningful for the detail comparisons between the FEM and test results. In order to get a general idea of the capability of the CRISP finite element method to simulate the ground movement mechanism, trial runs of the program were carried out. The input parameters are the same as listed in Table 6.1 which is for the simulation of the second series of tests. On this basis, the effect of different excavation rates used in the two test series on the stress path of retained soil can be evaluated by comparing to the analysis for the second series of tests.

From the analysis for a rate similar to test To1, the result shown on the right hand side of Figure 6.5 shows that in the first four stages, the ground profiles obtained from the test are visually similar with those obtained from the FEM analysis. At the later stages in which failure has occurred, it is beyond the capability of finite element method to analyse. Detail comparison however pointed to a number of differences. First, the test results clearly show the development of slip surface and tension cracks at an area near the wall. The slip surface and tension cracks affect the shape of ground movement, and make the ground surface settlement decrease abruptly when the distance from the wall is beyond the slip surface. This effect becomes more obvious at later stages of excavation near to failure. In contrast, the implicit assumption of a continuum in a FEM analysis means it is unable to simulate the tension cracks and slip surface in the soil. Thus the ground movement extends much longer. As there was no slip elements being used in the analysis at the interface between the diaphragm wall and soil, and between the soil and the right side of the container wall, the connectivity effect is observed. This problem will be studied in some details using slip elements at the interface in a later section. Except for a local range near the boundaries, the magnitudes of ground surface settlements from the tests are generally larger than those from the analysis with un-adjusted input parameters. Accompanying the wall failure is the appearence of a clear slip surface or shear band at an angle of 45^o. Such local shear failure is still not easy to be modelled.

A comparison between the test and numerical analysis results for test Ts1 is shown in Figure 6.12. Due to the presence of the upper strut, the deflection of wall exhibits a similar shape as in the bending action. Similar to the previous case, the magnitude of ground surface settlements from the centrifuge tests are generally larger than those from the finite element analysis. In the case of upper propped wall, the curved slip surface cannot be modelled also.

Generally speaking, most existing programs of FEA cannot perfectly simulate the behaviour of excavation near the collapse. In Figures 6.5 and 6.12 the deformed meshes of the ground are shown to present the general similarities and differences between the experimental observations and FEM. The output of the analysis for test To1 are given from Figure 6.6 to Figure 6.9. Figure 6.6 shows the change of pore water pressure profile with excavation depth. Although there were limited number of pore pressure transducers in the centrifuge model, similar trends can still be found by comparing the predicted values with the observations at points PPT1, PPT2 and PPT3 in Figure 4.3. The change of vertical effective stress profiles and the change of lateral effective stress profiles with excavation depth for test To1 as well as the change of effective shear stress profiles are shown in Figures 6.7, 6.8 and 6.9. Because there were no corresponding observations in the experiment, detailed quantitative comparison and an analysis of mechanisms are not provided in the present stage.

6.6.2 Pore water pressure and stress path at PPT1

Even though the initial condition of To1 is difficult to simulate in the FEM analysis, nevertheless, the FEM analysis would provide some idea of the pore pressure development. For this reason, the predicted pore pressure response at PPT1 is analysed. The total pore pressure at PPT1 and the stress path in p¡¯- q plot are shown in Figures 6.10 and 6.11 respectively.

The initial stress state of point PPT1 was located at Ao on the Ko line (Figure 6.11) which represents the stress state after the one-dimensional consolidation. When the zinc chloride is being drained, the lateral support is removed and the wall is forced to move towards the excavated side. As a result, the stress state was changed by releasing the horizontal stress in the retained soil. Based on the excavation rate in the tests, the analysis shows that the stress change is close to an undrained unloading condition. As shown in Figure 6.11, the deviatoric stress increases towards yield surface with no change in mean effective stress. During pore pressure dissipation, the soil swells and softens, and this reduces the size of the yield locus. The stress state eventually reaches the critical state line.

6.6.3 Ground surface settlements

As shown in Table 3.4, test Ts2 has the same strutting and soil condition as test Ts1. The difference between the two tests is that the excavation depth in test Ts2 is 50 mm while for test Ts1, it is 150 mm. In this case, the effect of zinc chloride solution below the excavated level (50 mm) is significantly lesser than those in test Ts1. The measured and FEM predicted ground surface settlements are presented in Figure 6.13. It is found that by minimising the effect of zinc chloride, the magnitude of measured settlements on the ground surface near LVDT 2 are quite close to the predicted value. However, at locations around LVDT 3 and LVDT 4, the predicted settlements calculated by FEM analysis, are much larger than the measured value even when the right boundary is fixed in the vertical direction. Such difference and its interpretation will be discussed in section 6.8.

v