Elasto-viscoplastic modelling of ground-improvements via vacuum-assisted prefabricated vertical drains with time-dependent boundary conditions

Abstract

In this thesis, time-dependent boundary conditions are introduced to a creep based elasto-viscoplastic (EVP) model, which can be used to predict soft soil deformations with vacuum consolidation. Use of vacuum-assisted prefabricated vertical drains (PVDs) is a relatively new method and is getting popularity in ground-improvement projects due to its ability to consolidate deep-buried soft clay layers in a comparatively short period. Simple approximations have been reported to idealise vacuum consolidation equivalent to surcharge preloading in previous research. Conversely, in this thesis, it has been shown that considering vacuum consolidation as a time-dependent boundary condition unfolds a large number of possibilities to accurately represent ground improvements with vacuum-assisted PVDs. Combined with a creep based EVP model, it is illustrated that the accuracy of both short and long-term settlement and excess pore pressure (EPP) predictions can be improved. Finite element analyses (FEA) of several case histories and laboratory experimental data have been used in this thesis to illustrate the improvements made in the predictability of the soft soil behaviour. Both axisymmetric and plane strain (PS) FEA are carried out for a period over three years for the Ballina test embankment vacuum applied section. Improvements made in the proposed method are highlighted against the field data and previous FEA attempts. In PS conditions, the implications of unit cell width on the FEA results such as settlements, EPP and lateral deformations are demonstrated which would serve as a guide for predicting the performance in similar scenarios. Also, time-dependent boundary conditions are successfully used in this thesis to capture the removal and re-application of vacuum to model practical scenarios such as vacuum pump breakdowns and recoveries. Validations have been carried out against published laboratory results and with a case history from Singapore. Simple yet effective methods to avoid numerical instabilities in simulating vacuum removal and re-application are proposed and improvements achieved with these solutions are illustrated. Later in the thesis, implications of vacuum distributions with depth of PVDs are discussed. Subsequently, a convenient method to model complex yet practically realistic vacuum distributions encountered in PVDs is presented and validated.