Finite Element Modeling of Installation Effects of Soil-Cement Columns

Abstract: Since the 1970's deep mixing columns have been widely used all over the world to improve the performance of soft soil in regard to bearing capacity or deformation behaviour. They are installed by mixing a binding agent, e.g. cement, in situ with the soil. The choice of installation method affects the properties of the column and the surrounding as the soil is disturbed by the installation process. However, the effects of the installation are often neglected during design even though they are plentiful. Besides the lateral displacement that could destabilize neighbouring constructions, the soil in the direct vicinity of the installed column is affected. Laboratory and field tests revealed the formation of three distinct zones outside the nominal diameter of the column which have different strength properties than the initial clay. They are formed due to cylindrical expansion, clay fracturing, and the migration of ions from the binding agent and their strength changes with time due to consolidation, cementation, heating, and thixotropy. Within this thesis, the installation effects that occur in the direct vicinity of the column have been studied in the context of a construction project in Sweden, where deep mixing columns are considered for the reduction of settlements of road and parks areas located on a thick clay layer. Based on analytical calculation methods and field measurements described in the literature, the occurrence and the magnitude of the installation effects have been assessed. The influence of considering these effects was then studied numerically using the finite element program PLAXIS. The simulation included one column within a column group and was performed in 2D assuming axisymmetry. The presence of the neighbouring columns was considered over the boundary conditions. The installation effects in the vicinity of the column comprised three zones which were implemented in the numerical model. For comparison, the simulations were also performed using the "wish-in-place" approach for the column that ignores the occurrence of any installation effects. The stabilized soil is loaded with two layers of new filling material which results in excess pore pressures and settlements that have been studied. The results for the model in which the installation effects were considered could be compared to the results for the model in which the column was wished-in-place. The comparison showed that the consideration of the installation effects leads to a faster consolidation and a significant reduction in settlements. This was observed for different installation patterns, i.e. triangular and square, and varying column spacings of 1.2 and 2.4 m. The positive installation effects were greatest for a smaller spacing and a triangular installation pattern. For a square installation pattern with a spacing of 2.4 m, the consolidation time and the final settlements were both reduced by more than 40%. Even though the assumptions and simplifications require verification, a clear positive influence can be seen for the project in Sweden. If these numerical results are confirmed by field observations, more efficient construction designs could be obtained which ultimately result in reduced costs and carbon dioxide emissions.