Modelling issues in simulation of deep excavations

This paper investigates three different modelling approaches available in ABAQUS/Standard finite element program for the simulation of short-term behaviour of deep excavations in saturated clayey soils below the water table. The first approach is based on the total stress principle with the undrained soil properties, which is the conventional way of analysing deep excavations using the finite element method with the undrained assumption. The other two approaches are based on the effective stress principle but the analysis time is very short preventing any excess pore pressure dissipation, replicating undrained behaviour. In the first approach based on the effective stress principle, a partially coupled analysis is carried out considering the excess pore pressures and effective stresses. ABAQUS does not have the ability to incorporate initial hydrostatic pore pressure distribution in the partially coupled excess pore pressure analysis. To simulate the hydrostatic pressures applied on the wall during dewatering of the excavation, body loads are applied along the wall. In the second approach, which resembles the real scenario, a fully coupled analysis was carried out considering the total pore pressures and effective stresses. In this analysis, initial hydrostatic pressures can be established before the excavation begins. Undrained and drained shear strength parameters relevant for the same clayey soil derived from triaxial tests were used to perform the total and effective stress analyses, respectively. Results obtained from a hypothetical case show that the use of the conventional way of analysis based on the total stress principle assuming undrained behaviour is not suitable to predict wall and ground deformations during excavations if the excavation is deep (more than 10 m). Partially-coupled analysis over predicted wall deformations and ground surface settlements compared to the fully coupled analysis, confirming that a fully coupled analysis is necessary to obtain the wall deformation and ground surface settlement corresponding to the short-term behaviour. Finally a case study reported in the literature for a staged deep excavation carried out in Taiwan is simulated using the two methods based on the effective stress principle. These results show that both partially and fully coupled approaches are suitable for the simulation of deep excavations, when excess pore pressure dissipation is allowed during the excavation.