Numerical modeling of the effect of preferential flow on hillslope hydrology and slope stability

The topic of this thesis is the quantification of the influence of preferential flow on landslide-triggering in potentially unstable slopes. Preferential flow paths (e.g., cracks, macropores, fissures, pipes, etc.) commonly exists in slopes. Flow velocities in preferential flow paths can be significantly larger than in the matrix. Under large rainfall or snow-melt events, preferential flow can bypass the adjacent soil matrix and directly reach the groundwater table. The fast pressure build-up caused by preferential flow can reduce the effective stress and shear strength, which is an important triggering factor for landslides. Single-permeability models can not appropriately simulate preferential flow. Hence, hydro-mechanical models of landslide need the inclusion of preferential flow.

Preferential flow also affects tracer transport in subsurface flow systems. The celerity in unsaturated flow represents the maximum water velocity in a soil, and it may be used to predict the first arrival time of a conservative tracer. The celerity function is derived from the soil hydraulic conductivity function for unsaturated flow, and is used to derive the breakthrough curve of a conservative tracer under advective transport. Analysis of the bimodal hydraulic function for a dual-permeability model shows that different parameter sets may result in similar soil hydraulic conductivity behavior, but distinctly different celerity behavior.

In Chapter 4, a 2D hydro-mechanical model is developed using COMSOL multi-physics modeling software to couple a dual-permeability model with a linear-elastic model. Numerical experiments are conducted for two different rainfall events on a synthetic slope. The influence of preferential flow on slope stability is quantified by comparing the simulated slope failure area for single-permeability model and dual-permeability models. The single-permeability model only simulate regular wetting fronts propagating downward without representing the preferential flow. In contrast, the dual-permeability model can simulate the influence of preferential flow including the enhanced drainage that facilities pressure dissipation under low-intensity rainfall, as well as the fast pressure build-up that may trigger landslides under high-intensity rainfall. The dual-permeability model resulted in a smaller failure area than the corresponding single-permeability model under low-intensity rainfall, while the dual-permeability model resulted in a larger failure area and earlier timing than the corresponding single-permeability model for high-intensity rainfall.

In Chapter 5, a parsimonious 1D hydro-mechanical model is developed for field application by coupling a 1D dual-permeability model with an infinite slope stability analysis approach. The numerical model is benchmarked against the HYDRUS-1D for the simulation of non-equilibrium flow. In Chapter 6, the model is applied to simulate the pressure response in a clay-shales slope located in northern Italy. In the study area, preferential flow paths such as tension cracks and macropores are widespread. Intense rain-pulses in the summer can cause nearly-instant pressure responses which may reactivate landslide movement. The water exchange coefficient of the dual-permeability model is calibrated for two single-pulse rainfall-events in the summer, while all other parameters are obtained from field investigations. Results from the dual-permeability model are compared to previously published outcomes using a linear-diffusion equation, where the diffusion coefficient was calibrated for each rainfall event separately. The dual-permeability model explicitly accounts for the influence of both matrix flow and preferential flow on water flow and pressure propagation in variably saturated soils, and is able to simulate the measured pressure response to multi-pulse rainfall-events quite well even in the winter time. Results indicate that the dual-permeability model may be more appropriate for the prediction of landslide-triggering when the pore water pressure response is influenced by preferential flow under high-intensity rainfall.