Flume Experiments to Determine the Erodibility of Gravel Streambank Soils

Abstract

Abstract. Fluvial erosion in streambank stability models is typically modeled using an excess shear stress equation based on two soil parameters: the erodibility coefficient (k d ) and critical shear stress (I„ c ). For cohesive soils, methods exist such as jet erosion tests (JETs) for measuring k d and I„ c . For noncohesive bank materials such as sands and gravels, I„ c is commonly estimated using the Shields diagram based on the median particle diameter. However, no universally accepted relationship exists for estimating k d for noncohesive soils. Also, recent research has proposed the use of more fundamentally-based mechanistic detachment models in place of the excess shear stress equation, but limited research has been performed in deriving the parameters of these alternative detachment models. One such model investigated in this research is the Wilson Model, which is based on two parameters: b 0 and b 1 . The objectives of this research included the following: (i) conduct flume experiments on noncohesive gravels to quantify detachment rates relative to the imposed shear stress, (ii) derive k d and I„ c and b 0 and b 1 for these gravels, (iii) determine if relationships can be developed relating k d versus I„ c and b 0 versus b 1 , (iv) compare the k d -I„ c relationship to previously proposed relationships, and (v) compare the relative performance of the models in being able to fit the scour data. Flume experiments were conducted at the USDA-ARS Hydraulics Laboratory. Samples of gravels were extracted from streambanks on the Barren Fork Creek in eastern Oklahoma. The samples were sieved into particle size classes and then at least triplicate flume experiments were performed with average gravel sizes of 0.45, 0.60, 1.30, and 1.90 cm (0.18, 0.25, 0.50, and 0.75 inches, respectively). Scour depth was measured using a point gage during the flume experiment. The k d -I„ c relationships derived from the laboratory experiments followed a power law relationship with the form: k d =2.23I„ c -0.50 (R 2 = 0.65). The Wilson Model parameters b 0 and b 1 derived from this research appeared to have a similar relationship but different magnitude than k d and I„ c . The k d -I„ c or b 0 -b 1 relationships can be used to model fluvial erosion of the noncohesive gravel and therefore the resulting streambank failure of composite banks.