Seepage caused tension failures and erosion undercutting of hillslopes

Abstract

Summary Seepage has been suggested as an important factor in gully and river bank erosion. This study investigated the underlying mechanisms of instability by seepage in laboratory studies. A 25-cm tall, 50-cm wide, and 20-cm long soil block with a focused inflow reservoir was constructed to investigate seepage gradient forces and the three-dimensional nature of seepage particle mobilization (i.e., seepage erosion) and undercutting. Experiments included sand and loamy sand soil blocks packed at prescribed bulk densities (1.30–1.70 Mg m−3) and with an outflow face at various angles (90°, 75°, and 60°). Constant heads of 15, 25, and 35 cm were imposed on the soil to induce flow. A laser scanner was utilized to obtain the three-dimensional coordinates of the bank and undercut surfaces at approximately 15–30 s intervals. The bulk density of the two different soil types controlled which seepage failure mechanism occurred: (1) tension or “pop-out” failures due to the seepage force exceeding the soil shear strength which was being concurrently reduced by increased soil pore-water pressure, or (2) particle entrainment in the seepage flow, particle mobilization, bank undercutting, and bank collapse when the initial seepage force gradient was less than the resistance of the soil block. For cases experiencing particle mobilization and undercutting, seepage erosion initiated as unimodal (i.e., concentrated at one point) or as multimodal (i.e., initiating at several locations across the bank face), and this result was largely controlled by the bank angle. A five parameter Gaussian function was fitted to the measured three-dimensional undercut shapes to derive parameters for the maximum depth of undercutting, position of the center of the peak, and the vertical and lateral spreads of the undercut. The parameters of this distribution can be useful in the development of improved sediment transport functions and the incorporation of this failure mechanism into hillslope stability models.