A Physically-Based Model of Bedload Transport

A physically-based bedload transport model is presented that aims to relax the constraints of classical empirical models, yet which is simple enough to be used within a variety of modelling frameworks. The model is derived from analysis of static discrete element model (DEM) beds of grains. A probabilistic erosion algorithm is used to "erode" the beds, where the probability of each grain being entrained is inverse to the force required to entrain it. The data produced by this process is then used to derive two relationships. The first relates the shear stress to the proportion of the bed surface which is active. The second uses this proportion, in conjunction with the surface grain size distribution (GSD), to predict the bedload GSD. These relationships predict bedload coarsening as shear stress increases. The probabilistic erosion model is based on the parameters of grain exposure and grain pivoting angle, as measured from the DEM beds. Field measurements were used to compare the values of these parameters in the DEM beds with those measured in fluvial gravels in the field. This enables investigation of the effect of grain shape (DEM grains are spherical, whereas gravels generally are not) and grain packing (as affected by water working) on the values of grain exposure and pivoting angles. A terrestrial laser scanner was used in the field to capture the gravel surface, from which, along with associated field measurements, grain exposure and pivoting angle were measured. An implementation of the bedload model is demonstrated within a 1D flume model, run in both feed and recirculation modes. The equilibrium GSDs of the bedload and surface are consistent with expected flume behaviour, suggesting that the model provides an reasonable reproduction of sediment entrainment, transport and sorting.