In Situ Soil Pipeflow Experiments on Contrasting Streambank Soils

Soil piping has been attributed as a potential mechanism of instability of embankments and streambanks. Limited field work has been conducted on quantifying and modeling pipeflow and internal erosion processes in the field with either natural or artificially created soil pipes. This research utilized an innovative constant-head trench system to conduct constant-head soil pipe experiments in two contrasting streambanks: Dry Creek in northern Mississippi and Cow Creek in northern Oklahoma. Experiments included open pipes, in which the soil pipe was directly connected to the constant-head trench and open at the streambank face, and clogged pipes, which involved plugging the outlet of the soil pipe using soil excavated adjacent to the pipe. A tensiometer network was used to measure soil water pressures surrounding open and clogged pipe outlets on the streambank face. When pipeflow occurred, flow and sediment samples were collected using flow collection pans to quantify sediment concentrations. Flow and sediment data were used with an existing turbulent pipeflow and internal erosion model to estimate erodibility and critical shear stress properties of the soils, which were subsequently compared to similar properties derived from jet erosion tests. Clogged soil pipes resulted in pore water pressure increases in the soil adjacent to the pipe, which generally remained below saturation during these experimental periods, except at locations close to the plug. Depending on the density of the plugged soil material, the clogged soil pipes either burst, resulting in turbulent pipeflow, or were manually punctured to establish pipeflow. Calibrated critical shear stress from the turbulent pipeflow and internal erosion model matched that observed from jet erosion tests for the less erodible soils on the Dry Creek streambank, where sediment concentrations were consistently below 2 g L even with fairly large hydraulic gradients on the pipe (0.3 m m). Calibrated erodibility coefficients were much smaller than those measured with jet erosion tests. For the more erodible streambank soils of Cow Creek, sediment concentrations approached 40 g L. There is a need for improved pipeflow modeling that accounts for rapidly changing pipe geometries, partially filled soil pipes, and pipeflow/soil matrix interactions.