Preferential Flow Phenomena in Partially-Saturated Porous Media

Preferential flow in unsaturated porous media, in the form of fingers and channels, has been observed to take place in different types of media and at different scales. The preferential flow is caused by fluid instabilities, created by density or viscosity differences, or because of heterogeneities in the geological media. In this study, two very different types of preferential flow problems are examined. In the first part, drainage processes in sand are visualized with the purpose of examining nonuniqueness found in experiments for measuring unsaturated hydraulic properties. In the second part, preferential flow in unsaturated fractured clayey till are examined by applying multiple tracing experiments. Methods for measuring unsaturated hydrologic properties have shown to exhibit random and nonunique behavior. To examine the causes multiple retention and one-step outflow experiments were conducted on a thin sand sample. Light transmission techniques were applied during the experiments for visualizing the drainage processes at micro-scale. The drainage was found to be composed by a mixture of fast air fingering followed by a slower back-filling process. This fingering and back-filling was controlled by a combination of the size and the speed of the applied boundary step, small-scale heterogeneities, and the initial saturation and its structure. The mixture of these micro-scale processes had influence on the macroscale effective behavior and thereby also the measured unsaturated hydraulic properties. The results thus suggest limitations on the current definition and uniqueness of unsaturated hydraulic properties. Fractures and macropores in clayey till may constitute fast preferential flow paths through an otherwise low-permeable porous media. To examine the influence of preferential flow in unsaturated clayey till, multiple tracing experiments were conducted in an isolated block at Avedøre, Denmark. Experiments were conducted at different steady-state water intensities, and multiple tracers with different molecular diffusion coefficients were applied to quantify the diffusive exchange between fractures and matrix. At high intensities, asymmetrical breakthrough curves were observed, where a fast peak was followed by a long tailing period. The multiple tracers only showed small differences in breakthrough indicating that diffusion processes only have small influence at high intensity and high water content. At the lowest intensity, a double peak behavior was observed for two of the applied tracers. This was hypothesized to be caused by severe retardation into the matrix. The tracer with the lowest molecular diffusion coefficient did not show the same double peak behavior, which may be caused by the differences in diffusion, however, sorption of the tracer may have influenced the transport as well. Colloids applied in the same experiment, peaked earlier than the solutes and did not show the double peak, which certifies that the colloids are experiencing less diffusion into stagnant water. The double-porosity model CXTFIT was not fully capable of describing the observed breakthrough. The model fitted the data fairly well, however, unrealistically parameters were obtained and furthermore did different tracers result in different parameters. The simple mobileimmobile model thus partly failed to predict the transport, and different reasons were hypothesized including fingering and film flow in the fractures, heterogeneity in the matrix diffusion, and heterogeneity in the fractures.