Erosion of a river bed has important implications with respect to scour around river structures such as bridges, transport of contaminants attached to the sediment, and disruption or destruction of aquatic habitats. Erosion occurs when the resistive strength of the sediment is overcome by the hydrodynamic forces produced by the flow of water. This resistance to erosion in a sediment originates from gravity or interparticle forces for coarse sediment (sand and gravel) and fine sediment (silt and clay), respectively. Since the erosion of fine sediment depends on the combination of many interparticle forces, and this combination fluctuates widely amongst different fine sediments, past studies have had difficulty finding a consistent method to estimate fine sediment erosion. This study analyzes sediments that fall in the transition size range between fine and coarse sediments and compares the findings with those from fine sediments (Wang 2013) and sandy coarse sediments (Navarro 2004, Hobson 2008), in order to correlate the erosion rates of both sediment types to their physical characteristics. In this study, kaolin-sand mixtures were prepared by mixing various percentages of Georgia kaolin by weight ranging from 30% to 100% with industrial fine sand and tap water. Geotechnical and other tests of sediment properties were performed to measure water content, bulk density, grain size distribution, temperature, pH, and conductivity of these mixtures. Hydraulic flume experiments measured the erosion rates of each sediment and these rates were used to estimate the critical shear stress correlating to that mixture. Relationships between the physical properties of the sediment and critical shear stress were developed by multiple regression analysis. An alternative option of estimating the critical shear stress by a weighted equation, which uses the combination of fine sediment erosion and coarse sediment erosion equations separately, was explored and found to be a viable and accurate option to estimating both coarse and fine sediment erosion from the same parameters and equation. The results from this study can be used to estimate sediment erodibility and thus river bed stability based on simple tests of physical properties of the river bed sediment and will help predict scour around bridges and other flow obstructions.

Les modèles d'érosion et de transport couramment utilisés sont construits à partir de données empiriques obtenues avec des sédiments de granulométries quasi uniformes. Or, il y a beaucoup de situations pour lesquelles la granulométrie des sédiments n’est pas uniforme. Les expériences réalisées en laboratoire et dans les rivières montrent que l’érosion et le transport des sédiments dépendent de la dispersion du diamètre des grains. Cette observation est à l’origine de cette thèse qui a pour objectif l’étude de l’influence de la distribution du diamètre des grains sur le transport de sédiments. Cette influence est envisagée à partir du développement d’un modèle numérique. La prédiction de l’érosion et du transport de sédiments tient de la résolution de deux problèmes : le premier est lié à l’écoulement au dessus du lit, le second à la mise en mouvement du sédiment. Le modèle développé détermine explicitement le mouvement des grains dans le lit de sédiments lorsque sa face supérieure est soumise à un écoulement. Pour cela, il mobilise la Méthode des Eléments Discrets (DEM), développée par Cundall et Strack (1979). Afin de reproduire l’effet de la topographie du lit sur le champ de vitesse du fluide, le modèle DEM est couplé avec le modèle d’écoulement FLOWSTAR. Le modèle FLOWSTAR est proposé par Carruthers et al. (2004) pour déterminer l’écoulement moyen dans une couche limite turbulente atmosphérique au-dessus des collines de faible pente. Le modèle numérique développé est appliqué à différents types d’arrangements de grains. Il permet d’estimer l’évolution du débit de sédiments au cours du temps pour différentes vitesses de frottement. Les seuils d’érosion des lits et l’évolution des débits de sédiments en fonction de la vitesse de frottement sont conformes à l’expérience. L’utilisation de l’approche DEM permet par ailleurs de connaître le comportement des grains dans et à la surface du lit au cours du temps (profil vertical de la vitesse des grains à l’intérieur de l’arrangement par exemple) / The models for the erosion and transport of sediments that are currently used rely on empirical data obtained from experiments with sediments having a uniform or unimodal distribution. But there are many practical situations for which the size distribution is significantly different from this assumed distribution, and laboratory and field experiments have shown that the erosion threshold and the transport rate depend on the size distribution and the range of particle sizes. The aim of this study is therefore to investigate and explain the influence of size distribution on erosion and transport rates, using a numerical model that has been developed specifically to study this problem. The sediment bed is assumed to consist of individual, non-cohesive, spherical particles, and the physical interactions between the particles are modelled explicitly, using the Discrete Element Method developed by Cundall and Strack (1979). The flow above the bed is computed using the FLOWSTAR model (Carruthers et al 2000) which was originally developed to compute the flow in the atmospheric boundary layer above arbitrary topography. These two models are coupled, and the resulting numerical code has been used to investigate the temporal evolution of erosion and transport rates agree well with experimental measurements, and the DEM provides additional information concerning the temporal evolution of the particle size distribution within the bed.