Groundwater seepage can lead to the erosion and failure of streambanks and hillslopes. Two groundwater instability mechanisms include (i) tension failure due to the seepage force exceeding the soil shear strength or (ii) undercutting by seepage erosion and eventual mass failure. Previous research on these mechanisms has been limited to non-cohesive and low cohesion soils. This study utilized a constant-head, seepage soil box packed with more cohesive (6% and 15% clay) sandy loam soils at prescribed bulk densities (1.30 to 1.70 Mg m À3 ) and with a bank angle of 90°to investigate the controls on failure mechanisms due to seepage forces. A dimensionless seepage mechanism (SM) number was derived and evaluated based on the ratio of resistive cohesion forces to the driving forces leading to instability including seepage gradients with an assumed steady-state seepage angle. Tension failures and undercutting were both observed dependent primarily on the saturated hydraulic conductivity, effective cohesion, and seepage gradient. Also, shapes of seepage undercuts for these more cohesive soils were wider and less deep compared to undercuts in sand and loamy sand soils. Direct shear tests were used to quantify the geotechnical properties of the soils packed at the various bulk densities. The SM number reasonably predicted the seepage failure mechanism (tension failure versus undercutting) based on the geotechnical properties and assumed steady-state seepage gradients of the physical-scale laboratory experiments, with some uncertainty due to measurement of geotechnical parameters, assumed seepage gradient direction, and the expected width of the failure block. It is hypothesized that the SM number can be used to evaluate seepage failure mechanisms when a streambank or hillslope experiences steady-state seepage forces. When prevalent, seepage gradient forces should be considered when analyzing bank stability, and therefore should be incorporated into commonly used stability models.

Mechanistic models have been proposed for soil piping and internal erosion on well‐compacted levees and dams, but limited research has evaluated these models in less compacted (more erodible) soils typical of hillslopes and streambanks. This study utilized a soil box (50 cm long, 50 cm wide and 20 cm tall) to conduct constant‐head, soil pipe and internal erosion experiments for two soils (clay loam from Dry Creek and sandy loam from Cow Creek streambanks) packed at uniform bulk densities. Initial gravimetric moisture contents prior to packing were 10, 12 and 14% for Dry Creek soil and 8, 12, and 14% for Cow Creek soil. A 1‐cm diameter rod was placed horizontally along the length of the soil bed during packing and carefully removed after packing to create a continuous soil pipe. A constant head was maintained at the inflow end. Flow rates and sediment concentrations were measured from the pipe outlet. Replicate submerged jet erosion tests (JETs) were conducted to derive erodibility parameters for repacked samples at the same moisture contents. Flow rates from the box experiments were used to calibrate the mechanistic model. The influence of the initial moisture content was apparent, with some pipes (8% moisture content) expanding so fast that limited data was collected. The mechanistic model was able to estimate equivalent flow rates to those observed in the experiments, but had difficulty matching observed sediment concentrations when the pipes rapidly expanded. The JETs predicted similar erodibility coefficients compared to the mechanistic model for the more erodible cases but not for the less erodible cases (14% moisture content). Improved models are needed that better define the changing soil pipe cross‐section during supply‐ and transport‐limited internal erosion, especially for piping through lower compacted (more erodible) soils as opposed to more well‐compacted soils resulting from constructing levees and dams. Copyright © 2013 John Wiley & Sons, Ltd.

No abstract