Journal of Hydraulic Engineering

The flow, sediment transport, and bathymetry at equilibrium conditions in loose-bed open channels are predicted using a fully three-dimensional (3D) Reynolds-Averaged Navier-Stokes model that was validated for cases in which the bed-load transport is dominant. The paper focuses on two test cases corresponding to laboratory experiments performed in an S-shaped channel in which the suspended sediment load is comparable to the bed load. A modified version of the original model that accounts for gravitational force effects due to the total sediment load is proposed. It is shown that the way bed-slope effects are accounted for plays a major role in improving the predictive capabilities of the model for a case in which the equilibrium bathymetry is characterized by relatively large values and sharp variations of the transverse slope. The model is then used to qualitatively and quantitatively analyze the distributions of several flow variables that are difficult to measure in the laboratory. They include the distributions of the bed-shear stress and of a Chezy-type friction coefficient assumed to be constant by most lower-order (e.g., two-dimensional depth-averaged) models.

In dredging practice sand is eroded at very high flow velocities using water jets. Breaching of dikes or dams is another process where sediments are eroded under the influence of high flow velocities. The existing pick-up functions were developed for relative low values of bed shear stress and hence overestimate erosion. Conventional pick-up functions can be made suitable for dealing with high-velocity erosion by taking the permeability of the sediment into account. This can be achieved by modifying the critical Shields parameter. The procedure is demonstrated using the van Rijn pick-up function. The theory is compared with experiment, and agreement is found to be good.

The effectiveness of a novel approach of using vanes, installed at a low angle and attached to the bank, for bank protection and for the restoration of river meanders has been investigated in a laboratory study. Experiments were carried out in a large-scale meandering mobile-bed channel with graded sediment. The bed topography, three-dimensional flow pattern, and turbulence characteristics in the meandering channel with or without structures are analyzed. When a single or an array of such vanes is installed, the scour hole at the base of the outer bank is infilled and the thalweg is relocated toward the center of the river. The structures induce a secondary flow cell near the outer bank which counteracts the main spiral flow in the bend. In contrast to common spurs and bendway weirs, large-scale horizontal vortices are not generated behind the structures. Vanes which grade to the bed from bankfull level at the bank show better performance than low level ones, whereas multiple structures show positive effects as far downstream as the crossover section.

Local scour at circular bridge piers embedded in a clay-sand-mixed bed was investigated in laboratory flume experiments. The effects of clay content, water content, and sand size on maximum equilibrium scour depth, equilibrium scour hole geometry, scouring process, and time variation of scour were studied at velocities close to the threshold velocities for the sand in the clay-sand mixture. It was observed that clay content and water content were the key parameters that effect the scouring process, scour hole geometry, and maximum equilibrium scour depth. The bridge pier scouring process in clay-sand mixtures involved different dominating modes for removal of sediment from scour hole: chunks-of-aggregates, aggregate-by-aggregate, and particle-by-particle. Regression-based equations for estimation of nondimensional maximum scour depth and scour hole diameter for piers embedded in clay-sand mixtures having clay content of $<40\mathrm{\%}$ and water content of $<40\mathrm{\%}$ were proposed as functions of pier Froude number, clay content, water content, and bed shear strength.

Experiments have been carried out in a mobile-bed laboratory flume in order to study the sediment exchange process between the main channel and the groyne fields. The flume represented half the width of a schematized river reach with a series of groynes. The experiment was designed to represent typical dimensions of the Dutch River Waal at a geometrical scale of 1:100. The conditions were set to guarantee bed load as well as suspended load sediment transport. Conditions with submerged and emerged groynes were investigated. In addition to traditional measurements, viz., bed-level changes, suspended sediment concentrations, and flow velocities, bed-form propagation was measured in two dimensions using a the particle image velocimetry technique. The results were analyzed with focus on sediment exchange mechanisms and sediment transport patterns. The results demonstrate that under all flow conditions there is a net import of sediment into the groyne fields. The prevailing transport mechanisms vary with the flow stage: if the groynes are emerged it is mainly advection by the primary circulation cell, whereas if the groynes are submerged it is rather residual advection by large-scale coherent flow structures (in a straight reach). Additional entrainment of sediment by enhanced turbulence complicates the erosion/deposition patterns.

A large group of reactions that affect water quality in distribution networks occur on the pipe wall surface. Existing simulation models are usually based on cross-sectionally averaged variables that use mass-transfer coefficients derived for constant-concentration (Dirichlet) boundary conditions to account for cross-sectional variations. In the case of a first-order wall-demand problem, the boundary condition is however of Robin type. We derive a simple one-dimensional (1D) model for the radial concentration profile of a solute of arbitrary Schmidt number $\left(\mathsf{Sc}\right)$ reacting with pipe walls in a fully developed turbulent flow. A modified van Driest mixing length model was used to approximate the Reynolds-averaged velocity and eddy diffusivity. Numerical solutions of the 1D model agree well with a two-dimensional mass transport model and experimental data. An asymptotic solution for high $\mathsf{Sc}$ is derived, which is in excellent agreement with the 1D model for $\mathsf{Sc}>100$ . A comparison with the mass-transfer coefficients for constant-concentration boundary conditions shows that the differences between the two boundary conditions are small.

In this paper, attention has been focused on the near-exit region of a round turbulent free jet to study the large-scale coherent structures and to document the signatures of the vortices over a range of Reynolds numbers. Particle image velocimeter measurements were conducted at three jet exit Reynolds numbers of 10,000, 30,000, and 55,000. The large-scale structures in the near field $\left(X/D<12\right)$ were investigated by performing a proper orthogonal decomposition analysis of the velocity fields. A vortex identification algorithm was complemented by swirling strength maps to identify the vortex centers, rotational sense, size, and circulation of the vortices. The influence of the Reynolds number on the distribution of the number, size, and circulation of the identified vortices was studied. Proper orthogonal decomposition of the velocity fields revealed that Reynolds number has a strong influence on the mean circulation of vortices. The present results show that the axial location where vortices first appear and the number of vortices close to the nozzle exit $\left(X/D<5\right)$ are dependent on the Reynolds number.

Full-scale field acoustic Doppler velocimeter measurements of the radial, axial, and tangential flow components were undertaken in a large-diameter turbulent swirling jet generated by a raft-mounted axial flow-mixing propeller discharging through a draft tube during isothermal and stratified conditions. The results were compared with classical jet theory and showed significant differences in the zone of flow establishment though similar behavior in the zone where established flow was found. It was concluded that the efflux from the axial flow-mixing propeller could not be described adequately by conventional jet theory due to the jet size and swirling characteristics.

Submerged vanes are submerged foils of low height and larger length, constructed in a river at an angle of attack $\alpha$ to the flow to modify the near-bed flow pattern and redistribute flow and sediment transport within the channel cross section. At a Froude number $\left(\mathsf{F}\right)$ of 0.13, the local scour development around the submerged vane without a collar was not enough to dislodge the vane whereas at $\mathsf{F}=0.25$ , there was a significant local scour hole around the vane and the vane was dislodged. With the introduction of a collar at the leading edge of a submerged vane, the scour depth at the leading edge of the vane was reduced to zero. A collar of circular shape was found more suitable for a rectangular vane. Recommendations for sizing collars at two values of $\mathsf{F}$ are given. The optimal $\alpha$ for a rectangular vane with a collar was found close to 40°. The study clearly indicates the advantages of using collars in case of submerged vanes and provides insight into selection of appropriate collar shapes.

The amplification of supercritical waves in steep channels is examined analytically using a one-dimensional dynamic solution of the Saint-Venant equations. Existing methods were modified to describe the amplification of surface waves over a normalized channel length rather than over a single wavelength. The results are strikingly different, and a generalized graph shows that short waves amplify the most over a fixed channel length. The maximum amplification parameter over a normalized channel length is 0.53 when $\mathsf{F}=3.44$ . Applications to the flood drainage channel F1 in Las Vegas indicate that the amplitude of waves shorter than 100 m would increase by 65% over a channel length of 543 m. These theoretical results await field verification. Supercritical waves could be dampened by increasing channel roughness to reduce the Froude number below 1.5.

An unsteady mathematical model for predicting flow divisions at a right-angled open-channel junction is presented. Existing dividing models depend on a prior knowledge of a constant flow regime. In addition, their strong nonlinearity does not guarantee compatibility with the St. Venant solutions in the context of an internal boundary condition treatment. Assuming zero crest height at the junction region, a side weir model explicitly introduced within the one-dimensional St. Venant equations is used to cope with the two-dimensional pattern of the flow. An upwind implicit numerical solver is employed to compute the new governing equations. The performance of the proposed technique in predicting super-, trans-, and subcritical flow bifurcations is illustrated by comparing with experimental data and/or theoretical predictions. In all the tests, lateral-to-upstream discharge ratios $\left(R_q\right)$ are successfully reproduced by the present technique with a maximum error magnitude of less than 9%.

This work extends and improves a one-dimensional shallow flow model to two-dimensional (2D) for real-world flood simulations. The model solves a prebalanced formulation of the fully 2D shallow water equations, including friction source terms using a finite volume Godunov-type numerical scheme. A reconstruction method ensuring nonnegative depth is used along with a Harten, Lax, and van Leer approximate Riemann solver with the contact wave restored for calculation of interface fluxes. A local bed modification method is proposed to maintain the well-balanced property of the algorithm for simulations involving wetting and drying. Second-order accurate scheme is achieved by using the slope limited linear reconstruction together with a Runge-Kutta time integration method. The model is applicable to calculate different types of flood wave ranging from slow-varying inundations to extreme and violent floods, propagating over complex domains including natural terrains and dense urban areas. After validating against an analytical case of flow sloshing in a domain with a parabolic bed profile, the model is applied to simulate an inundation event in a $36{\text{km}}^2$ floodplain in Thamesmead near London. The numerical predictions are compared with analytical solutions and alternative numerical results.

Models to predict mudflow and debris flow velocities are tested with 350 field and laboratory measurements. Overall, the turbulent model performs best while the dispersive stress approach only compares well with the measurements when the flow depth $h$ is less than 50 times the median particle diameter $d_{50}$ . The analysis of field measurements shows that the ratio of mean flow velocity $V$ to shear velocity $u^{\ast }$ is approximately 10, rarely exceeds 30, and increases slightly with relative submergence $h/d_{50}$ . The best overall agreement with laboratory and field velocity measurements is obtained with $V=5.75u^{\ast }\text{log}h/d_{50}$ .