Abstract

The flow of water in an open channel is typically turbulent, exerting shear stresses on its boundaries, i.e. the channel-bed and sidewalls. Knowledge of boundary shear stress (BSS) distributed at a channel section has many important applications. For example, the BSS distributions affect channel stability, fish habitats, and the resuspension and subsequent transport of bed sediments. However, knowledge of the BSS is far from being complete, partly because there are technical difficulties in measuring shear stresses distributed across the bed width and sidewalls. This research thesis aims to investigate BSS distributions by means of large eddy simulation (LES). This numerical technique is superior to the commonly used traditional computational fluid dynamics approaches, with respect to detailed predictions of near-boundary flow.

Water channels of rectangular and trapezoidal shapes under open-water as well as ice-covered conditions were included as LES domains in this study. The LES results of BSS from this study were compared with available data from laboratory experiments. The comparisons show that the LES results capture typical features, as reported in the literature, of BSS distributions at rectangular and trapezoidal channel sections, including the occurrence of inflection points caused by bottom vortices. The BSS is shown to vary across the bed width and sidewalls of the channel sections. The bed shear stress is relatively high in the large central portion of the bed width and drops rapidly toward the lower corners of the channel sections. The sidewall shear stress has a similar shape as the bed shear stress. The normalised bed shear stresses in the corner regions have different distributions between the trapezoidal channel and the rectangular channel section. Secondary flow in the channels are shown to cause BSS spatial fluctuations in the central portion of the channel-bed. The maximum BBS does not necessarily occur at the same location as the maximum primary velocity. As expected, the flow structures and turbulent shear stresses in the channel sections are sensitive to the presence of ice cover as well as to inflow disturbances.

The successful predictions are attributed to a proper implementation of no-slip conditions at the solid boundaries, as opposed to the use of uncertain estimates of solid surface roughness, and to an adequate representation of the viscous sublayer by the LES mesh used. It is crucial to allow mesh refinements adjacent to the bed and sidewalls as well as in the corner regions, and to ensure that the wall distance of the first node off a solid surface does not exceed unity, which is not the case in most of the existing LES applications to open-channel hydraulics. For the first time, this research thesis has explored practical LES strategies for accurate and efficient BSS predictions; these include the types of conditions imposed at the free surface and the ice cover underside, the lateral open boundaries at the upstream and downstream ends of a LES domain, and sensitivity tests. It has been demonstrated that the LES technique offers an attractive complement to laboratory measurements of BSS. With an exponential increase in computing power, it will eventually be feasible to perform LES for high Reynolds number flow in water channels.