| dc.description.abstract | Numerical analyses are widely used in practical geotechnical engineering to assess the deformation
behaviour of deep excavations, in particular when the influence on existing infrastructure,
such as buildings or adjacent tunnels, has to be evaluated. The key issue of numerical analysis
is selecting a applicable constitutive model and determining the input parameters.
Some commonly used constitutive models are studied, and the adaptabilities and limitations
of these models are also been analysed. It is found that the hardening soil model which
can define the shear hardening, compression hardening and unloading-reloading characteristics
gives a very reasonable prediction in deformation analysis. But the models which can only
simulate the shear hardening behaviour, such as Mohr- Coulomb model, are not suitable for
the deep excavation analysis. When the heave of the foundation after excavation of great interest,
the hardening soil with small-strain gives a acceptable estimation, while the prediction
from hardening soil has big errors. That may because the unloading-reloading modulus is not
associated with stress and strain level in hardening soil model, and can not define the nonlinear
unloading-reloading characteristics.
Then, the determination for parameters of hardening soil model, especially for stiffness parameters
is discussed in detail. A selection scheme is proposed on the basis of empirical correlations
and back calculation for geotechnical engineering design with available soil data. Two
case study are performed to help determining the soil parameters. It is found that the initial estimation
of soil parameters need to be calibrated for undrained simulation, while it can be used
directly for the drained case in deep excavation.
Finally, the foundation excavation of Shanghai Bank building has been analysed. Finite element
method software PLAXIS is employed to perform the numerical simulation. Three soil
models are used in this study, and the soil parameters are determined by the selection scheme
proposed before. The model results are compared with the on site measurements to prove the
reasonableness of the parameters selecting scheme. | en |
