Three-dimensional numerical and physical modelling of soft soil improvement using concrete injected columns

Publication Type:
Issue Date:
Full metadata record
Concrete injected columns (CICs) are a popular method for improving soft soil properties to support road and bridge approach embankments due to quick construction, absence of spoil, and limited post-construction settlement. While the limited settlement of CICs makes them attractive for cases where there are stringent settlement criteria, low-cost methods of improving soils are used where there are no such limitations. The lack of comprehensive experimental studies on CICs in the available literature showed the necessity of further laboratory modelling. Moreover, the equivalent comparison between frictional and socketed CICs has not been thoroughly studied. In this study, a well-instrumented physical modelling of soft clay improved with CICs was performed. A granular layer was used to model the load transfer platform (LTP), and a geotextile layer was utilised to model the geosynthetic reinforcement (GR) layer. The load was applied and controlled in stages using a large loading frame on top of the granular layer. Pore pressure dissipation, stresses transferred to the soft soil and CICs, and the strains in the geotextile were monitored with time. A three-dimensional numerical model was also developed using finite difference software FLAC³ᴰ, and the results were validated against the experimental data. The numerical model considered coupled flow-deformation allowing prediction of the excess pore water pressure (EPWP) dissipation with time, while the permeability of the soft soil varied with time. Modified Cam-Clay (MCC) soft soil model was used as the constitutive model for the soft clay deposit, while elastic-perfectly plastic Mohr- Coulomb failure criterion was used to simulate the LTP layer. Hoek-Brown constitutive model was used to model the unreinforced concrete used for CIC construction. A good agreement was perceived between the numerical results and the measurements from the experiment. Referring to both measurements and predictions, despite the low permeability of the soft clay, a rather quick dissipation in the EPWP occurred due to the load transfer mechanism between the soft soil and CICs. The stress concentration ratio decreased at the beginning of the loading stages and then later increased with time, and was higher for higher applied loads. This thesis also sets out to investigate the options available for the transition zone from CICs to other ground improvement methods away from the abutment. Two possible alternatives were numerically simulated using FLAC3D software considering the dissipation of pore water pressure and variation of soil permeability with time. A geosynthetic layer was introduced into the load transfer platform (LTP) located above the CICs, and interface elements were incorporated to simulate CIC-soil interaction. The first option for the transition zone was widely spaced CICs socketed into stiff material and the second was using shorter, closely spaced, frictional CICs. A comparison was then made between the predicted ground settlement, the force mobilised in the geosynthetic, the excess pore water pressure, and stresses in the CICs for the two scenarios. The total length of the CICs and thus the total volume of the concrete used for their construction were kept the same for both alternatives. Indeed, the embankment on frictional CICs experienced less settlement, the forces mobilised in the geosynthetic were reduced, and the bending moments and shear forces generated in the columns were less than the corresponding values for the case of socketed CICs. This study showed that for a given volume of concrete, shorter, frictional CICs perform better than longer CICs socketed into stiff strata. Furthermore, a comparison was made between drained and coupled flowdeformation numerical analyses. This study revealed that while performing drained analysis by simply assigning drained parameters to the material was less computationally demanding, it lead to inaccuracies in the predictions. The perceived discrepancies were attributed to the difference in the stress-path of drained and coupled analyses. The results from this study can be beneficial for the practicing engineers for designing structures on CIC-improved grounds, particularly for predicting the timedependent performance of the system.
Please use this identifier to cite or link to this item: