Journal of Hydraulic Engineering

Many investigators have observed that in mobile sand bed streams, there is a range of flows over which two subcritical depths of flow are possible. This phenomenon is typically attributed to a hysteresis effect in which the discharge is changing more rapidly than the bed forms can adjust to the new discharge. This theory is supported by the fact that higher depths of flow are typically observed on the ascending limb of the hydrograph, while shallower depths tend to occur on the descending limb of the hydrograph. The magnitude of the discontinuity would then primarily be a function of the time rate of change in the discharge, and would not have to be a unique value. An alternative hypothesis, presented by the writers, demonstrates that in mobile sand bed channels where bed forms occur, two depths of flow are numerically and physically possible. Data collected by other researchers were used to develop a bed form friction factor predictor for narrow flumes and wide sand bed channels with appropriate corrections for side roughness. The friction factor predictor is used in conjunction with Einstein's method for the separation of wall roughness from bed roughness to predict the depth of flow for a given discharge. It is demonstrated that, under some hydraulic conditions, there are two numerical solutions to the system of equations used to predict the depth of flow. Both of the solutions appear to be physically possible. The two depths of flow are referred to as sequential depths. Application of the method to field data collected by Colby shows that the magnitudes of the sequential depths are approximately equal to the two depths observed by Colby.

Turbulent flow in compound open channels is studied numerically with nonlinear k-ε turbulence models of low-Reynolds type. Emphasis is given to the flow characteristics and the performance of the models for conditions of low relative depths (up to 0.2547) for which the interaction between main channel and flood plain flow is significant. Experiments and computations indicate that the velocities follow the law of the wall in the interaction region even for such low relative depths. Secondary currents are predicted by a modified version of a previous model; however, the predicted currents are not as strong as the experimental ones. Turbulent shear stresses and especially $-\overline{\mathit{uw}}\text{,}$ which is significant for low relative depths, are predicted reasonably by the model, except in the interaction region in the main channel. The turbulent intensities v′ and w′ are captured by the model while u′ is underestimated. A depth-averaged analysis identifies two basic mechanisms that influence the variation of the bed shear stress and suggests a means of assessing the contribution of each mechanism to the deviation of the bed shear stress from its respective two-dimensional value.

In relatively shallow coastal waters, a submarine outfall pipeline is often laid in a dredged trench and the wastewater jet discharges inside a concentric enclosure (a protective riser tube) before release at the seabed. The effect of such lateral confinement over a finite vertical length on the mixing of a buoyant jet is studied experimentally for a wide range of geometries. A confinement index β=H_r/(D_r−D) is proposed to categorize the confined buoyant jet, where H_r and D_r are the height and diameter of the confinement, respectively, and D is the jet diameter. Compared to the dilution of the corresponding free buoyant jet, it is found that mixing is impaired for β≥ 3. For β>> 1, the jet is highly confined and there is virtually no dilution within the riser. For β≤ 1.5, the confinement has no effect. For 1.5 ≤β≤ 3, a wall-attached swirling plume with enhanced dilution is observed, and the mixing behavior is complex. Simple relations for predicting the initial dilution of a laterally confined plume are presented.

Numerous inland Australian streams contain density-stratified or saline pools. Saline pools consist of a layer of saline water underlying a layer of fresh water. They are associated with seasonal low flows and influxes of saline groundwater and have important water quality and environmental implications. A theoretical model of the dominant mixing process (flushing) associated with flows of fresh water across the top of the saline layer is presented. Laboratory results show that progressive removal of saline water from saline pools occurs because of a thin saline layer flowing up the downstream depression slope when the surface layer flows over the saline layer at a sufficiently high velocity. The behavior of this thin layer is controlled by a balance between shear and buoyancy forces that act on it. Experimental results describe the initiation of this outflow, the rate of the outflow, the composition of the outflow, and the range of scour hole geometries over which flushing is the dominant mixing process.

This paper presents the results of a laboratory study on flow past cylindrical piers placed on smooth, rough, and mobile beds. Experimental results are presented on the flow in the plane of symmetry, including the frontal downflow and the effects of bed roughness and the scour hole on it. The Clauser-type defect scheme describes the velocity profiles better than the log-law and defect law. Frontal downflows as large as 95% of the approach flow were observed. Experimental results are also presented on the deflected flow and bed shear stress field. Bed roughness increased the magnitude of bed shear stress and the area over which the shear amplification was felt and also resisted the skewing of the flow near the bed, thus leading to smaller yaw angles of the bed shear vectors.

An explicit numerical approximation of the inertial equation governing pump speed changes is shown to have many computational advantages over the conventional implicit approach when modeling complex pumping stations. The pump boundary condition is numerically decoupled from the transient time step by solving the first-order differential torque equation explicitly. A step-by-step finite difference method is used to integrate the torque equation, but the energy equation is solved separately by Newton's method. The explicit approach is demonstrated on a simple forcemain and shows good agreement with the conventional implicit approach. Applications show the ease with which complex and variable speed pumping arrangements are efficiently modeled using the explicit approach. In particular, the explicit approach does not require: (1) contraction of the system of equations as a decelerating pump is eliminated upon pump check valve closure; (2) expansion of the system of equations as an accelerating pump comes up to speed; or (3) dedicated code for combinations of operating, failing, or speed-changing pumps.

A general hydrodynamic numerical model for sewer/channel networks is presented. The Priessmann Slot assumption is adopted to extend open channel flow equations to closed conduits under surcharged conditions. The “superlink” algorithm with a staggered grid, implicit scheme is developed to improve the stability and speed of the computation. The SUPERLINK computer model was established without omission of any terms in the St. Venant equations under both subcritical and supercritical flow conditions. The model reduces the order of the sparse matrix equation so that a considerable savings in computational effort is achieved. The numerical experiments show that the model is robust and reliable, and produces consistent results under different grid setups. The application to a large looped sewer network shows a reasonable agreement with field measurement. A comparison between the results from the SUPERLINK and the SWMM EXTRAN model demonstrates that while maintaining the computational stability and accuracy, the computation can be sped up using the SUPERLINK algorithm for the sewer network system, which experienced a wide variety of hydrodynamic conditions.

Inclusion of longitudinal dispersion enhances realism in the assessment of transport in streams by accounting for contaminant mixing. Here we revisit the problem of estimating the longitudinal dispersion coefficient D from readily measurable bulk hydraulic variables, which is useful in applications. We develop and calibrate, through field testing, a predictor for the longitudinal dispersion coefficient for streams that is structured after the fundamental formula of G. I. Taylor. Comparison of predicted with observed values of D shows the new optimized formula providing closer estimates than the established formula of Fischer, with respect to both mean accuracy and deviations from the mean, achieving in 14 of the 24 cases studied predictive accuracy of 0.5 ≤D_observed/D_predicted≤ 2.

In the United States, water-level datums are computed by applying National Ocean Service algorithms to records obtained at fixed gauges. These physical gauges are expensive to install and maintain, and they suffer certain practical limitations such as requiring a storm-resistant platform and convenient access. As a supplement to physical gauges, in this paper a modeling approach is explored whereby physical gauges at boundaries drive a calibrated numerical model from which the calculated time series of water level at any internal grid point may be considered as a “numerical” gauge. A proof-of-concept study for the upper Laguna Madre, Texas, is described that compares monthly and annual tidal datums and range determined from a co-located numerical gauge and physical gauge. The favorable comparison indicates the viability of the numerical gauge concept.

The pattern of primary isovels, pathways of sediment transport (both bed load and suspended load), and the evolution of channel planform are all strongly affected by secondary currents. The way in which secondary currents affect channel morphology in single-thread meandering rivers is now well understood; however, to date there has been no field-based study of secondary currents in a large, sand-bedded braided river. The River Survey Project of FAP-24, Bangladesh, offered the opportunity to rectify this gap in knowledge through a joint study between the University of Nottingham, Delft Hydraulics, and the Danish Hydraulics Institute. A comprehensive set of field data, including measurements of primary and secondary velocities and channel bed topography, were taken around a braid bar in the Brahmaputra River near Bahadurabad, Bangladesh. An acoustic doppler current profiler was used to measure three-dimensional velocity profiles. The results presented here were taken from two cross sections: at the flow bifurcation upstream of the study bar and at the bend apex of the left anabranch around the study bar. The results demonstrate the presence of large-scale secondary current cells at the flow bifurcation and bend apex.

Two simplified dynamic reliability models for hydraulic design are developed, which consider both hydraulic and hydrologic uncertainties. To incorporate the safety factor into the reliability model properly, the actual design flood is introduced and defined as the product of the safety factor and the design flood. The reliability models represent unique relationships between actual design flood and reliability. If actual design floods are the same, regardless of changes between the safety factor and the design flood, the same reliability results are obtained by the simplified models, given that the service periods are identical. The simplified models can be reduced to the conventional reliability models when only hydrologic uncertainty is considered. The risk due to hydraulic uncertainty can be estimated from the difference of the reliability results by the conventional reliability models and those by the simplified models, i.e., the difference between single and total reliabilities. The simplified models can be used to design new hydraulic structures and to evaluate the reliability level for existing hydraulic structures.