Journal of Hydraulic Engineering

Hydraulic two-phase transport applied in the dredging, mining, and deep-sea mining industries involves the transportation of sand, gravel, polymetallic nodules, or other particulate tailings as a solids phase and water as a liquid phase. Regardless of the type or size of the granular material, the slurry flow is always subject to transient behavior. Most transient behavior can be attributed to the centrifugal pump as variations in pump pressure and mixture velocity over time, but transients can also be caused by microscopic slurry mechanisms, specifically the amplification of density waves in a pipeline. Density wave amplification in horizontal pipelines at mixture velocities just above the deposition limit velocity was reported and researched in the 1990s. New experiments showing a density wave amplification in a system with combined vertical and horizontal pipelines and at mixture velocities far above the deposition limit suggest that another type of density wave amplification mechanism exists. The newly proposed density wave amplification mechanism is hypothesized to be caused by a change in average particle velocity as the slurry flows from a vertical pipe into a horizontal pipe. Density waves that grow too large cause system blockages or possibly a failure of the pump drive. This article considers centrifugal pump–induced transients and density wave amplification effects separately and how these effects influence each other. Three case studies showing density wave amplification are analyzed, one from the literature and two from new data sets. Furthermore, the causes of these transients are discussed, and where possible, solutions are proposed to avoid these undesirable instabilities.

This paper examines the relationship between the flow rate and outflow depth for flow in a rectangular pipe with the outflow running partially full of a free nappe through a series of computational fluid dynamics (CFD) simulations. The results are presented for two different upstream inlet conditions, namely, flooded, in which the pipe inlet is fully submerged, and partially full, wherein there is a free surface at the pipe inlet. The results are presented for the outflow normalized brink depth as a function of the nondimensional flow rate $q^{*}=(q/\sqrt{g{D}^3})$. For $q^{*}$ less than 0.575, the outflow conditions are the same regardless of the inflow boundary conditions. However, for larger $q^{*}$, there is a bifurcation in the outflow behavior. When the pipe inlet is fully submerged and $q^{*}>0.575$, the flow transitions to the so-called bubble-washout regime, and the brink depth is larger than for the case in which the pipe inlet is only partially full. These results indicate that if the brink depth from a pipe is to be used to quantify the pipe discharge, it is important to know the upstream flow conditions in order to know which flow regime is controlling the outflow.

This paper aims at improving the weighting function based-method (WFB) for modeling the transient behavior of a laminar flow in cylindrical pipes in a one-dimensional approach. Two improvements for the numerical computation of the unsteady friction term are presented. First, a rational approximation of the weighting function in the Laplace domain is preferred instead of an exponential series fit in the time domain. It allows the WFB method to be improved in terms of validity for small time steps, accuracy, and computational efficiency. Second, the use of auxiliary differential equations to compute convolution makes the high order time-integration of the frequency-dependent friction term straightforward, without the assumption of a constant acceleration during the time step. The simulation results for a well-known experimental test case show a good agreement of the derived methods with the experiment. Finally, the time stability of the discretized problem is fully analyzed, and a stability condition for the WFB method is brought out.

Floodborne debris impact loads depend on several factors, including debris orientation and velocity. Due to the random nature of turbulent flood flows, debris motions are inherently random. This study presents a probabilistic characterization of debris motions in steady-state currents with relatively high Reynolds numbers. An item of debris was released, multiple times, in the open channel flows, at two angles and three current velocities. The debris motions were recorded via a camera. The characteristics of the debris motions including the trajectory, orientation, and velocity were extracted using image processing and were analyzed statistically. The results indicate that the debris orientation and lateral displacement followed a Gaussian distribution. The current velocity and release angle had a smaller effect on the debris lateral displacement than did the orientation. The analysis also highlights the combined effect of the current velocity and release angle on the lateral displacement and orientation of the debris. Empirical relationships were developed for estimating the mean trajectory and orientation of debris in steady currents, as a function of the release angle, and Froude number.

Temporal concentration profiles resulting from an injected pulse of fluorescent tracer were recorded at multiple locations along a pipe during controlled unsteady flow conditions. A linear temporal change in discharge over durations of 5, 10, or 60 s for both accelerating and decelerating flow conditions was studied. Tests were performed for flows that changed within the turbulent range, between Reynolds numbers of 6,500 and 47,000, and for laminar to turbulent flows, between Reynolds numbers of 2,700 and 47,000. Analysis of the data shows the limitations of employing steady-state routing of temporal concentration profiles in unsteady flow. Employing a flow weighted time routing approach, using tracer mean velocity and dispersion coefficients, provides accurate predictions of mixing in unsteady flow. For decelerating flows, longitudinal dispersion coefficients were lower than for the equivalent mean steady discharge. Previously unreported disaggregation of the tracer cloud was observed during all experiments accelerating from laminar to turbulent conditions.

Vegetation can have an appreciable impact on the hydrodynamics and scour potential in natural rivers, but this effect is generally unaccounted for in high-fidelity computational fluid dynamic models. In this study, we have incorporated trees into the flow domain using two different approaches to study the hydrodynamics of the American River in Northern California under flood conditions. In the first approach, we resolved numerous trees as discrete objects. The second method incorporated a vegetation model into our in-house numerical model to treat the vegetation as a momentum sink along the banks. The flood flow of both cases was modeled using the large-eddy simulation. The computed hydrodynamics results of the cases were compared with a baseline case, which did not include any trees. Although both the tree-resolving and vegetation model approaches compared well with one another with respect to the flow field, they significantly altered the computed river flow dynamics and bed shear stress near the banks and the midwidth of the river compared with that of the no-tree case. Both methods that accounted for the resistance of the trees obtained lower and higher bed shear stresses and velocities along the banks and the midwidth of the river, respectively, than that of the baseline case. This research identified the important role that vegetation plays in natural rivers and provided researchers and engineers with the conceptual tools needed to incorporate vegetation into numerical models to improve the accuracy of the model results.

This paper presents a case study of a prototype B-type hydraulic jump produced downstream an irrigation-regulating gate in a natural river. The inflow Froude number is ${\mathrm{Fr}}_1=5.1$ and the Reynolds number is $\mathrm{Re}=9\times {10}^5$. Detailed air–water flow properties and free-surface fluctuating features are investigated on site in the central flow region using intrusive phase-detection probes and nonintrusive acoustic displacement meters. Challenges in employing the phase-detection probe in high-momentum natural water are addressed, including the necessary postprocessing for uncertainty mitigation. While good agreement between the present data and previous laboratory measurements is achieved for the air concentration and interfacial velocity distributions, scale effects are observed for bubble frequency, challenging the validity of previous empirical prediction deriving from laboratory experiments. A greater percentage of large bubbles are detected in the full-scale jump. The free-surface fluctuating characteristics are investigated together with the surface wave propagation in the tailwater. The jump length and roller length in the B-jump are discussed with comparison to existing model equations. Although the investigation is limited to one case, it provides the first physical data obtained in full-scale natural water and an important benchmark data set for future study of scale effects in self-aerated open channel flows.

A three-dimensional (3D) numerical study is undertaken to investigate dam-break flows over 3D structures. A two-phase flow model has been developed within the large-eddy simulation (LES) framework. The governing equations have been discretized using the finite-volume method, with the air-water interface being captured using a volume-of-fluid method while the Cartesian cut-cell method deals with complex geometries. The robustness and versatility of the proposed numerical approach are demonstrated first by applying it to a 3D dam-break flow over a cuboid. Good agreement is obtained between the simulation results and the corresponding experimental data and other numerical solutions. Then, a horizontal cylinder and a sphere are subjected to the same dam-break flow. Snapshots of water surface profiles are presented and discussed, and turbulent vortical structures are identified in the flow. In addition, the internal kinematics, hydrodynamic loading on the structure, and energy dissipation during dam-break flow impact are analyzed and discussed, providing more insight into such flows.

Natural river velocity distributions are often studied by laboratory flume flows because of similarities of the bed shear stress effects between the two flows. Nevertheless, the velocity distributions in natural rivers are often affected by winds over water surfaces, which are often negligible in laboratory flows. The objective of this research is then to study how water surface shear stress and wind-induced turbulent mixing affect open channel velocity distributions in flume flows and natural river flows. To this end, we first introduce the recent second log-wake law from symmetric and antisymmetric channel flows to open channel flows. We then demonstrate that: (1) a laboratory flume velocity distribution is a superposition of a half symmetric channel flow solution and a half antisymmetric channel flow solution, and (2) a natural river velocity distribution is a superposition of a half symmetric channel flow solution and a complete antisymmetric channel flow solution. These solutions are confirmed by both laboratory and field data including velocity dip and inflection phenomena.

Pooled stepped chutes with openings as a design requirement are associated with high flow transmission efficiency and smaller downstream dissipation structures; however, these design openings may negatively affect flow characteristics. In this study, 3D numerical simulations were conducted on large pooled stepped chutes equipped with openings and different pool heights, which benefited from the volume of fluid (VOF) method, to track free-surface configuration, and the use of the $k-\epsilon (RNG)$ turbulence model. Results indicated that, for a pooled stepped chute with $\theta =8.9°$, increasing pool heights increased flow resistance, and for a chute with a steeper slope, $\theta =26.6°$, flow resistance remained quasi-constant. Four configurations for openings were further tested: centered, offset, inline, and staggered configurations. Results demonstrated that the staggered configuration reduces residual head at the end of the chute. Additionally, the data revealed that, as the dimensions of the openings increases, flow resistance decreases. The findings of this research could be used as a general guideline for practical cases where openings should be embedded in pooled stepped chutes.