It is apparent from Figures 6.5 to 6.10 and Figures 6.16 to 6.17 that discrepancies between the measured and numerical results are present. Some assumptions have been made in the finite element modelling of the excavation and there are factors that cannot be considered in the finite element simulation. These assumptions and factors may influence the response of excavation of a vertical shaft and they are discussed in this section.
6.7.1 Soil Stratification
The simplified soil profile obtained from BH1 and BH2, illustrated in Figure 3.6, is considered to be representative of the soil stratification at the project site. Thickness of the various soil layers is assumed to be uniform in the finite element model. Literature review of earlier research carried out on Old alluvium confirms that it is a highly variable formation, in terms of its composition, weathering and geotechnical properties. Thus, the assumption of uniform thickness and adoption of same geotechnical properties, listed in Tables 6.4 and 6.4, for the entire soil layers might not be applicable for the modelled area of the project site. The thickness and geological properties of the various soil layers are likely to defer at different panels of the diaphragm wall of Influent Pumping Shaft 2. Although the finite element results agree reasonably well with the instrumented results, the variation of soil thickness and properties are likely to affect the development of stresses and deformations of the excavation support system and might contribute to the discrepancies between measured and predicted results.
6.7.2 Interaction Effects Between Shafts
Excavations at Coarse Screen Shaft and Influent Pumping Shaft 1 are carried out, 46 days and 118 days respectively, after the commencement of excavation at Influent Pumping Shaft 2. The excavation sequence of the shafts is presented together with the predicted and measured hoop strains in Figures 6.5 to 6.10. Due to the close proximity of the three shafts, the zones of influence due to excavation of the shafts are expected to overlap. Similar effects, as shown in Figure 6.18, are likely to occur during the excavations at Influent Pumping Shaft 1 and Coarse Screen Shaft. Hence, the radial soil stresses in between the three shafts might be further reduced and the vertical and circumferential soil stresses between the shafts might be further increased when excavations at Influent Pumping Shaft 1 and Coarse Screen Shaft were conducted.
These changes in soil stresses and the corresponding changes in soil stiffness are likely to affect the development of structural forces and movements of Influent Pumping Shaft 2. The true extent and significance of interaction effects on the response of the three shafts can only be studied more accurately using a three-dimensional numerical software such as ABAQUS, CRISP and FLAC.
However, it can be observed from Figures 6.5 to 6.10 that the hoop strains measured by strain gauges at the same elevation of Influent Pumping Shaft 2 have similar trends, despite of their different positions relative to the neighbouring shafts. Panel S20 is located near Influent Pumping Shaft 1 while Panel S20 is closer to the Coarse Screen Shaft. It is apparent that changes in the measured hoop strains in the diaphragm wall of Influent Pumping Shaft 2, when the neighbouring excavations are carried out, seem to be negligible and are not as significant as the substantial development of hoop strains during its excavation and ring wall construction stages. This may be due to the
additional rigidity provided by internal ring walls as excavations at neighbouring shafts, to the depths where the strain gauges are located, are usually carried out after the ring walls at Influent Pumping Shaft 2 are constructed. The changes in soil stresses and stiffness due to the interaction of the shafts are likely to be distributed to both the diaphragm wall and internal ring walls such that the effects on the diaphragm wall is reduced.
Hence, although interaction effects of the shafts cannot be examined in the axisymmetrical finite element analysis, it is apparent that the adoption of such analysis in modelling the excavation and construction phases at Influent Pumping Shaft 2, together with the simulation of thermal effects, is adequate in predicting the major trends in the behaviour of this circular excavation supporting system.
6.7.3 Simulation of thermal effects
Discrepancies between the predicted and measured increase in compressive stresses induced by the temperature effects of ring wall are present. These discrepancies may be due to the usage of the assumed value for coefficient of thermal expansion of concrete. The coefficient of thermal expansion of concrete varies over a range of values and is dependent on the type of aggregate, cement paste and the degree of saturation of concrete. The temperature variation in the third lift of the first ring wall installation stage is adopted for all the lifts of other ring walls. The temperature inside each lift may vary as the thermal properties of concrete and insulating conditions changes. Thus, the actual thermal stresses acting on the diaphragm wall may not be similar to the stresses adopted, as shown in Table 6.8, which are derived based on the assumed coefficient of thermal expansion of concrete and temperature variation. As
the temperature variation of the third lift of the first ring wall installation stage was only monitored for the first 232.5 hours after casting, thermal stresses acting on the diaphragm wall after the first 232.5 hours after casting cannot be determined and included in the finite element analysis.
As temperature differences between the ring walls, diaphragm wall and the surrounding soils exist, there are complex thermodynamics transfers between the ring walls, diaphragm wall and surrounding soil. The complicated interactive effects of thermal transfer between the ring walls and diaphragm wall and between the diaphragm wall and soil cannot be accounted for in the finite element analysis. These phenomenons are likely to complicate the development of volumetric changes and hoop strains in the internal ring walls and diaphragm wall and result in the discrepancies between the numerical and measured results.
6.7.4 Shrinkage and Creep of Concrete
Shrinkage occurs in concrete and it results in a decrease of concrete volume with time.
This reduction in volume occurs due to physico-chemical changes and changes in moisture content of the concrete and generally, it can be classified into drying shrinkage, autogenous shrinkage and carbonation shrinkage. Concrete also exhibit creep behaviour and its strain increases with time due to sustained stress. These effects of shrinkage and creep of concrete cannot be accounted for in the finite element modelling using PLAXIS and may affect the development of hoop strains and contribute to some discrepancies between the numerical and measured hoop strains.
6.7.5 Strain gauges
Vibrating wire strain gauges are spot welded to the surface of steel reinforcement in the diaphragm wall. Sensors are mounted atop of the strain gauges and readings are obtained from a data logger. Drifts and loss of calibration in strain gauges with time and temperature and inaccuracy of readings due to imperfect bonding and non- uniformity between the strain gauge and steel reinforcement are common problems occurring in these instruments and they are plausible causes of discrepancies between the numerical and the measured hoop strains.