Resolved DNS of bedload transport with realistic grain morphology

The prediction of bedload sediment transport remains challenging due to the multi-scale interactions that link grain-scale dynamics, bed morphology, and turbulence. Grain-scale processes are influenced by particle shape and contribute to the spread in existing bedload models. Experimental access to these processes is limited, making numerical simulations a valuable complementary tool. Most numerical studies to date represent sediment grains as spheres to reduce computational cost or intentionally exclude shape effects. More recent work has demonstrated the importance of shape using ellipsoidal approximations which capture the overall grain form but loose finer surface irregularities. Only a few simulations have employed more realistic clumped-sphere grain approximations and have shown that grain shape introduces significant uncertainty in entrainment and transport predictions.

This contribution advances the quantification of grain-shape effects by using realistic representations of grain geometry obtained from measurements in the literature. It presents direct numerical simulations of turbulent bedload transport with low particle loading in a highly mobile regime using fully resolved, realistic sand grains. Three simulations with monodisperse but polymorph particles are considered, such that only grain shape is varied. One configuration represents smooth, well-rounded sand grains, the second consists of more angular and irregular grains. A third simulation with uniform spheres serves as a reference. The realistic grain samples are generated statistically following an established methodology that yields two distinct sand populations. The grains in these populations are characterized using sphericity and roundness and are classified using the Zingg diagram. Although the Zingg-class of all grains is spheroid, the grain populations can be subdivided based on the distributions of the other two shape descriptors, with sphericity capturing larger-scale morphology and roundness reflecting smaller-scale surface irregularities.

Across the three simulations, the Shields parameter increases with increasing grain irregularity. Furthermore, the particle ensembles in all simulations show oscillatory dynamics during statistically steady bedload transport, attributed to the recurring formation of particle clusters, with the characteristic period increasing with grain irregularity. Shape-conditioned statistics obtained through double averaging show that in the case of the angular grain population more rounded grains accumulate near the channel bottom, while more angular grains are transported to higher elevations. This sorting is not observed for the well-rounded grain population. Additionally, for both grain populations, the rotational energy of the grains increases with irregularity, although rotation remains overall weak compared to translational motion in the present highly mobile regime.