Abstract

The prevention and control of bridge scour is a challenging element in bridge-pier foundation design. Sediment scour from around the piers has been a main cause of all bridge failures. For the cost-effective and reliable design of pier foundations, one needs accurate prediction of flow-inducing scour. Such prediction can be obtained from numerical modelling, as a good extension of experimental results. General CFD software packages are incapable to simulate sediment transport (bedload) and bed-level change (scour/deposition). Some numerical models have been developed for bedload and scour simulations, but there are two major limitations: (1) the bed sediments are assumed to have uniform grain size which is not true in natural river channels; (2) the modelling techniques are not computationally efficient for applications at the field scale. Thus, new modelling techniques for scour prediction are needed.
The objectives of this research are: (1) to modify an existing shallow-water hydrodynamics model to allow for non-hydrostatic pressure corrections; (2) to improve the prediction of bed shear stress, a key parameter for bedload prediction; (3) to incorporate a new module for calculating bedload of mixed sediment-grain sizes; (4) to verify the model’s prediction with existing experimental data.
For pressure corrections, a seven-diagonal linear system is added to the model, which is symmetric and positively defined. This system is numerically solved for non-hydrostatic pressure through preconditioned conjugate gradient iterations. Then, corrections to velocity and water surface elevation due to non-hydrostatic pressure are obtained. Fractional bedload calculations are based on a surface-based transport function, which depends on a particle-hiding factor, bed shear stress, and grain size distribution. Bed level change caused by bedload is calculated using the Exner equation added to the model.
The new model successfully predicts 3-D velocities around a circular pier in a fixed scour-hole and scour development on a mobile bed with uniform and non-uniform sediments. We improve bed shear stress prediction by using near-bottom velocities, as opposed to the widely used bottom-layer velocity, in the wall function method. Velocity and scour depth predictions agree well with experimental data. We show that scour emerges from the lateral sides of a pier, and scour patterns move toward its upstream nose. On the upstream side, relative to the pier’s centre, scour depth increases until the bed slope reaches the angle of repose of sediments. On the downstream side, scour continues until equilibrium is reached. Scour is deeper on the upstream than downstream side. Non-uniformity in grain sizes tends to reduce the magnitude of scour. The presence of the pier causes a strong vortex at its foot on the upstream side, which effectively moves sediments toward downstream. On the upstream side, the scour-hole’s outer shape is almost half a cone, true for both uniform and non-uniform sediments. These findings have implications to foundation design. The modelling techniques presented in this study are computationally efficient and are practical for applications at the field scale, which have been difficult so far.