Transient Flow in Unsaturated Soils: Numerical Analysis and Case Study

Abstract

Numerous authors have described issues of numerical oscillation and slow or inaccurate convergence in finite element analyses of transient non-linear phenomena. Karthikeyan and Tan have discussed numerical issues such as the effect of element size and time step in analyzing transient flow through unsaturated soils. Unsaturated flow problems are highly non-linear because of the variations of soil hydraulic conductivity and water content (soil-water characteristic curves) with matric suction. Edgers and Nadim (2003) analyzed the pore pressure response of soils during a precipitation event that triggered a debris flow in Norway. They used the computer program SEEP/W for a two-dimensional finite element analysis with one directional flow. Their numerical analyses showed reasonable agreement between the time computed for the development of positive pore water pressures, approaching full hydrostatic, and the observed elapsed time to failure for the debris flow. This paper follows up on these earlier studies by analyzing the effects of time step, element size, and element type (4- vs. 8- noded elements) on the numerical solutions to the Edgers and Nadim (2003) case study. This paper shows that for this case study, uncertainty in the assumed hydraulic conductivity is more important than inaccuracies caused by numerical oscillations and slow or inaccurate convergence.