Journal of Hydraulic Engineering

This paper presents analytical solutions based on Boussinesq models with moving boundaries and their applicability to the phenomena of water waves: sloshing in a rectangular tank on an oscillation table generated by lateral moving boundaries and tsunami generation by a moving bottom. In the application to sloshing, the linear solutions in Boussinesq and hydrostatic models were derived, and their characteristics were examined by comparing them with experimental results. It was shown that the result for the Boussinesq models was in close agreement with measurements for high wave numbers. In the application to tsunami generation, linear solutions were compared with numerical simulation results. The linear solutions with the effect of vertical acceleration on the bottom and across the depth improved the reproducibility of the wave profiles. The reproducibility of the linear solutions was confirmed to be good, except for the case of faster bottom motion with linear temporal change. These analytical solutions will be useful to engineers in understanding the phenomena considered.

Reservoir sedimentation is of worldwide concern, involving not only reservoir benefits, but environmental and ecological issues. Sediment flushing is one of the most common and effective methods for alleviating sedimentation and its negative impacts. To date, however, it has remained unknown whether sediment flushing can be accomplished through a bottom tunnel once its intake has been covered by sediment deposits. Here, a total of 65 runs of flume experiments were conducted to investigate the threshold conditions for occurrence of sediment flushing. The results demonstrate that higher water level, smaller cover layer thickness, shorter tunnel length, and steeper tunnel slope favor the occurrence of sediment flushing. Physically, as the gate located at the tunnel intake is opened, the sediment deposits immediately upstream of the intake collapse and enter the tunnel, entraining water. A mass of hyperconcentrated water-sediment mixture is generated, in which the amount of sediment depends on the cover layer thickness. If the water level is sufficiently high, the cover layer thickness is low, the tunnel slope is steep, or the tunnel is short, the pressure difference between the inlet and outlet of the tunnel suffices to drive the water-sediment mixture to move down to the outlet, overcoming the resistance due to the tunnel boundary. This way, sediment flushing is accomplished. Based on dimensional analysis of the observed data, a threshold condition is proposed for the occurrence of sediment flushing. It is also shown that an equilibrium scour hole is formed immediately upstream of the tunnel intake, and an empirical model for the scour hole geometry is proposed. The present findings can be exploited to improve sediment-rich reservoir planning, design, and operation for benefit maximization and eco-environmental wellbeing.

By comparing model results from tests of steady and oscillatory flows over a submerged bar (of interest for both riverine and marine environments), we highlight similarities and differences between a state-of-the-art Boussinesq model and a recently developed non-hydrostatic, quasi-three-dimensional (3D) model. To make the comparison as clean as possible, the complexities due to breaking-induced turbulence are avoided by suitably tuning the flow strength. Although the main flow features are similarly described by the two models, the non-hydrostatic model predicts the occurrence of extra, small-scale stationary oscillations on top of a submerged obstacle in the case of a steady current. This is attributed to a pattern of alternating upward and downward vertical velocity over the sill top, presumably due to rapidly converging flows as they climb over the submerged obstacle. The non-hydrostatic model overall allows for a more complete representation of flow dynamics in the vertical direction, with respect to the Boussinesq-type solver. This becomes necessary when tackling fundamental and applicative problems characterized by intense vertical flows (e.g., interactions of fluids with structures).

A novel numerical model for describing highly transient mixed flows is presented based on the Preissmann slot approach. To overcome the spurious oscillation problems caused by this approach, a numerical scheme named “random choice method” (RCM), in which flow variables of the next time level are obtained by picking a local Riemann solution state at random, is applied herein. Three numerical tests are performed to verify the ability of the proposed model in simulating from single flows to mixed flows. The results show that the RCM gives sharper shock resolutions as compared with the Godunov-type scheme, which causes the smearing of discontinuities. The proposed model can eliminate the numerical oscillations under flow conditions of switching from free-surface flow to pressurized flow, because of its unconditional stability. On further analysis of the experimental verifications, a hybrid method is presented to improve the performance of RCM in the smooth (nonuniform) parts of the flow.

A new empirical equation for free alternate bar length at the fully developed state is introduced. The equation is developed on the basis of dimensional analysis and all readily available data. A special effort is also made to align the formulation with findings from existing theoretical and numerical analyses of the development of alternate bars, and more specifically those resulting from a stability analysis of the phenomenon. When applied to the existing laboratory data, the equation results in significantly less dispersion around the best agreement line than that associated with previous empirical equations, yielding 70% of the data within the 20% error range, against 47% for previous equations. It is shown that previous equations perform notoriously poorly in the case of fully rough subcritical flows and all transitionally rough flows, with the present equation addressing this matter. When applied to field cases, the present equation yielded realistic values of bar length, including for large sand streams with high or very high values of relative depth (e.g., the Tagus and Mississippi Rivers).