The objective of the present study is to understand the processes influencing the colloidal silica grout injection into porous media through experimental and numerical investigations. Experimental studies include viscosity measurements, grout injection into columns and grouted sand testing. The grout viscosities are measured during gelation under shear. Colloidal silica grout injection experiments are performed on two different sizes of columns to understand the grouting processes as well as to generate data for numerical simulations. The larger diameter column experiments are performed by varying colloidal silica solution and the type of water. The smaller diameter column (slim tube) experiment was conducted to obtain the data in the absence of fingers. The use of dyes aided in visualizing the grout flow patterns and understanding the processes.

A numerical model is developed by non-iteratively coupling a groundwater flow simulation model (MODFLOW) and a 3-D multi-species reactive transport model (RT3D) and adding modules for the gelling process to simulate the chemical grouting in porous media. A grout gelation model is developed to estimate gel viscosity as a function of reaction time, grout concentration and shear rate. The flow field is periodically updated by taking into account the changing effective hydraulic conductivity in each cell as a result of non-uniform gel viscosity.

The numerical model is verified and validated against (i) an analytical solution developed in this study, and (ii) the available numerical model results from literature, respectively. The grout model is used to analyse the experimental data of sodium silicate grout injection into the sand column as reported in literature. The grout model is also used to analyse the observations made in the colloidal silica grout injection experiment performed as part of this study.

Simulation of grout injection has shown that the shear is an important parameter not incorporated in previous grouting models. The model captured the viscous fingers observed in the column experiments. The model has been able to predict the experimentally observed injection pressures in the mild slope region. The difference between experimental observations and numerical results increased towards the end of the simulation when the pressures increased steeply. It is hypothesized that this is due to the absence of filtration component in the model. The macromolecules filtered out at the pore throats contribute to the need for higher grout injection pressures. The model provides valuable insights in identifying the complex processes such as viscosification, shear, viscous finger formation and filtration, occurring during grouting.