Journal of Hydraulic Engineering

A temperature-control curtain (TCC) is an effective facility to regulate the outflow temperature in reservoirs, and its safety and stability are of great concern. The pressure difference between the front and back of a TCC induces a thrust force on the TCC. Previous studies examined the effect of TCCs on the discharged temperature of reservoirs, but the force analysis of the TCC in a stratified reservoir has not been studied. Taking Sanbanxi Reservoir as a case study, this work explains the thrust force exerted on a TCC. A three-dimensional hydrodynamic and temperature model is built and validated using the physical model test and prototype observed data. Using a numerical simulation under typical scenarios, the pressure difference on the TCC is identified, and the regularity of the thrust force is estimated. The research results are as follows: (1) For a bottom TCC in a reservoir with strong thermal stratification, because of the temperature difference between the two sides of the TCC, the thrust force on the TCC consists of hydrostatic and hydrodynamic pressure differences, and both are in the downstream direction. Between the two parts, the proportion of the hydrodynamic pressure difference is much larger; (2) when the flow height decreases, the thrust force on the TCC rapidly increases. Both hydrostatic and hydrodynamic pressure differences increase, and the latter changes more rapidly; (3) when the thermal stratification becomes stronger, the hydrostatic pressure difference increases, the hydrodynamic pressure difference decreases, and the thrust force on the TCC decreases; (4) for a top TCC, the hydrostatic pressure difference points in the upstream direction. However, because of the giant hydrodynamic pressure difference, the thrust force on the top TCC is much greater than that on the bottom TCC. In addition, three TCC implementation schemes are proposed with a focus on engineering. This study quantitatively estimates the mechanism of thrust force on a TCC and provides a theoretical foundation for practical application.

The aim of this paper is to investigate the complex nature of the transient energy dissipation by using a computational fluid dynamics (CFD) model with a high spatial and temporal resolution together with one-dimensional (1D) models incorporating different unsteady friction (UF) formulations. The analysis focuses on a transient event in a single pipe system with a smooth-wall turbulent flow (with an initial Reynolds number equal to 7,638) generated by an instantaneous valve closure. For the considered flow condition, the numerical experiments point out the importance of the used UF model with regard to the wall shear stress simulation. In such a context, it is shown that the convolution-based UF models better describe the pressure signal than the instantaneous acceleration–based ones because they take into account a set of previous time steps. This is due to the fact that, similarly to the laminar regime, to simulate the characteristics of low turbulent transients the flow time history plays a crucial role.

A two-dimensional (2D) finite-volume Eulerian-Lagrangian method (FVELM) on unstructured grid is proposed for solving the advection equation in free-surface scalar transport models. A backtracking band is defined along the backward trajectory of a side center as the dependence domain of a cell face, over which scalar concentration distribution is integrated to evaluate the advective flux through the cell face. Using the cell-face advective fluxes, a finite-volume cell update is finally carried out to obtain new cell concentrations, when mass is conserved both locally and globally by the unique flux at a cell face. The FVELM is then tested by a solid-body rotation experiment, a laboratory bend-flume experiment, and a real-river test (in a 365-km reach of the Yangtze River). In solid-body rotation tests, the FVELM is revealed to at least achieve a performance of existing second-order accuracy advection schemes. Relative to explicit Eulerian methods, the FVELM extends the dependence domain of a cell face from the upwind cell to the backtracking band, and therefore allows large time steps for which the Courant–Friedrichs–Lewy number (CFL) can be greater than 1. Accurate and stable FVELM simulations can be achieved at a CFL as large as 2–5 in these three tests. Efficiency issues of the FVELM are clarified by using the bend-flume test with refined grids (193,536 cells) on a computer with 16 cores; a parallel run using the FVELM is 14.2 times faster than a sequential run. In solving a transport problem (using 16 kinds of scalars and 16 cores), a parallel run using the FVELM is 2.3 times faster than a parallel run using an existing subcycling finite-volume method.

Two-phase flows occurring at flow constrictions such as bridges or open sediment check dams are complex, especially for steep rivers with bedload transport. Dangerous bedload deposition and backwater effects may occur in steep mountain rivers at bridges. In contrast, sediment deposition is desirable at open check dams combined with sediment traps. For design purposes, the discharge and bedload capacity across these flow constrictions must be known. The energy losses, discharge capacity, and bedload transport capacity of vertical and lateral flow constrictions are experimentally studied in a rough, 2% inclined, trapezoidal channel. Both free surface and pressurized flow conditions, as caused by lateral and vertical flow constrictions, respectively, were analyzed because both may occur at bridges and check dams. The experiments demonstrate that the vertical flow constrictions cause a faster increase in the backwater depth, with increasing discharge, than lateral constrictions. The resulting upstream flow conditions can be described by the upstream Froude number, defined as a function of the constriction dimensions (height and width). The bedload transport capacity through the flow constriction decreases with the upstream Froude number. The practical relevance of the findings is illustrated by a design example of flow constrictions at open sediment check dams.

This paper deals with two-phase flow models of transient cavitating pipe flow, showing how bubble-flow models and shallow-water models can be considered in a unitary framework by only changing the geometry of the cross section. A bubble-flow cavitation model and a shallow-water cavitation model are proposed, taking into account the convective terms. Computational results are compared with those of two previous simpler bubble-flow models and with experimental measurements reported in the literature. Both quasi-steady one-dimensional (1D) and quasi-two-dimensional (2D) forms of all models are considered. For the examined cases, numerical analysis shows that 1D models converge for similar grid resolution; 2D models reach convergence with coarser longitudinal grids than, or at most with same grids as, 1D models. Comparison with measurements reveals that in order to properly investigate the capabilities of models, the computation must proceed way in time. Indeed, the first cavitation event (or, in some cases, the first few cavitation events) is very well reproduced with all models, both 1D and 2D, but over time remarkable differences in numerical results can be observed. The 1D models overestimate head values, generally do not reproduce well the shape of the function, and are delayed in time with respect to measurements, whereas 2D models reproduce very well the peaks and shapes of the head as function of time, but their results are anticipated in time with respect to measurements, suggesting that energy dissipation is overestimated. The comparison of results of the different 2D models shows, for the examined cases, very small differences, justifying the adoption of simpler models.

There is limited knowledge on the erosion rates of large grains. The available erosion models of granular material at large flow velocities are only validated up to a median particle diameter ($d_{50}$) of 1.5 mm. To validate these models with larger grains, erosion tests have been performed on split (crushed rock) with a $d_{50}$ of 8.5 and 21.5 mm. The test results are compared with the erosion models of Fernandez Luque and van Rijn with a combined data set. Results showed that for both theories the measured erosion velocities can be well predicted. It can be concluded that the empirical parameter in the pickup function of Fernandez Luque (${\alpha }_{FL}$) increases with particle diameter. For particle diameters larger than approximately 6 mm, ${\alpha }_{FL}$ becomes more or less constant. The pickup function of van Rijn is applicable over the total range of tested particle diameters. Unfortunately for particle diameters larger than 21.5 mm, it is still not clear which theory is most reliable to estimate the erosion velocities. For the time being, it is recommended to use both functions, with the function of Fernandez Luque as a conservative lower limit.

Automated, real-time suspended sediment concentration (SSC) measurement techniques are important for the design and operation of hydraulic structures as well as the sediment management of river and reservoir systems. Traditional methods provide high-accuracy data but are inadequate to provide high temporal and spatial resolution. Surrogate methods, such as turbidity, laser diffraction, or acoustic backscatter, allow higher temporal and spatial resolution, but accuracy is negatively affected by natural variations in sediment grain size, color, or shape and the maximum measurable concentration is limited. To overcome these limitations and to close the gap between traditional and surrogate technologies, a new automatic density measurement device called a densimeter for semicontinuous SSC measurements (DENSE), for high SSC values, is presented. Laboratory tests have been performed for the evaluation of its performance and precision. The results show that the new DENSE technology has an accuracy of $0.3\mathrm{g}/\mathrm{L}$.

Multiport diffusers are used in many environmental applications to distribute a discharge flow uniformly through a number of jet nozzles into the same or another fluid. For example, a sewage diffuser can be designed to achieve approximately equal port discharges along the diffuser to maximize environmental benefit, i.e., approximately the same port discharge to achieve similar dilutions for all ports. When the effluent density is smaller than that of the ambient receiving water (i.e., positively buoyant), the energy head increases with elevation (in the inshore direction); as the total diffuser flow varies, the relative port discharge distribution remains similar. However, for dense (negatively buoyant) discharges, this may not be the case. This paper presents an experimental and theoretical study of the internal hydraulics of a diffuser used for dosing sodium hypochlorite solution (specific gravity 1.2) for sewage disinfection. Because of the significant density difference, the jet discharge undergoes a flow switching (reversal of relative discharge distribution) as the total discharge decreases. The predictions were validated with full scale experiments carried out with water jets into air; the design implications for plant operation are discussed. The findings are of general interest to the design of diffuser manifolds for dense effluents.

Continuous monitoring of gravel transport in rivers is necessary for understanding the effect of dam removal on river systems. Impact plates, such as those deployed on the Elwha River in the state of Washington, are a promising technique; however, relating the data generated by the plates to sediment load is difficult. In order to refine methods described in a recent study, coarse bed load transport was measured in a laboratory channel and compared to the output from two accelerometer-equipped impact plates that were full-scale replicas of 72 plates deployed in the Elwha River. Relations were developed for converting the signal amplitude generated by particles impacting the plates to particle mass. Mean discrepancy ratios indicate that estimates of coarse bed load transport were within 25% of the measured values if one outlier from the experiments was removed. Methods to transfer these relations to field locations are also discussed.