Depth-Averaged Two-Dimensional Numerical Modeling of Unsteady Flow and Nonuniform Sediment Transport in Open Channels

A depth-averaged two-dimensional (2D) numerical model for unsteady flow and nonuniform sediment transport in open channels is established using the finite volume method on a nonstaggered, curvilinear grid. The 2D shallow water equations are solved by the SIMPLE(C) algorithms with the Rhie and Chow’s momentum interpolation technique. The proposed sediment transport model adopts a nonequilibrium approach for nonuniform total-load sediment transport. The bed load and suspended load are calculated separately or jointly according to sediment transport mode. The sediment transport capacity is determined by four formulas which are capable of accounting for the hiding and exposure effects among different size classes. An empirical formula is proposed to consider the effects of the gravity on the sediment transport capacity and the bed-load movement direction in channels with steep slopes. Flow and sediment transport are simulated in a decoupled manner, but the sediment module adopts a coupling procedure for the computations of sediment transport, bed change, and bed material sorting. The model has been tested against several experimental and field cases, showing good agreement between the simulated results and measured data.     

Modeling Noncohesive Suspended Sediment Transport in Stream Channels Using an Ensemble-Averaged Conservation Equation

The governing conservation equation for the transport of noncohesive suspended sediment in erodible channels is recognized as a stochastic partial differential equation due to the uncertainties in the parameters, and a deterministic ensemble-averaged equation is developed. Variables in this one-dimensional equation are represented as averaged quantities, and their covariances are also taken into account. Lateral inflows and deposition and entrainment of sediment are incorporated in the formulation. A hypothetical test problem is constructed to examine the model behavior. Manning’s coefficient, bed slope and bottom width are taken as the primary random parameters. Results from the solution of the ensemble-averaged equation are compared to results from Monte Carlo simulations. For comparison purposes, predicted values are also obtained by solving the deterministic transport equation without the covariance terms. It is found that predictions obtained from this latter approach deviate significantly from Monte Carlo simulation results. On the other hand, the ensemble-averaged predictions compare favorably to the Monte Carlo simulation results indicating that this promising technique needs further exploration.     

Machine Learning Approach to Modeling Sediment Transport

Inaccuracies of sediment transport models largely originate from our limitation to describe the process in precise mathematical terms. Machine learning (ML) is an alternative approach to reduce the inaccuracies of sedimentation models. It utilizes available domain knowledge for selecting the input and output variables for the ML models and uses modern regression techniques to fit the measured data. Two ML methods, artificial neural networks and model trees, are adopted to model bed-load and total-load transport using the measured data. The bed-load transport models are compared with the models due to Bagnold, Einstein, Parker et al., and van Rijn. The total-load transport models are compared with the models due to Ackers and White, Bagnold, Engelund and Hansen, and van Rijn. With the chosen data sets on bed-load and total-load transport the ML models provided better accuracy than the existing ones.     

Implicit Bidiagonal Scheme for Depth-Averaged Free-Surface Flow Equations

A general fast implicit bidiagonal numerical scheme, based on the MacCormack's predictor-corrector technique requiring the inversion of only block bidiagonal matrices, has been developed and subsequently applied for subcritical and supercritical free-surface flow calculations. The model has been applied to depth-averaged steady flows. There are two main advantages of the proposed method: the technique has fast convergence and utilizes a body fitted nonorthogonal local coordinate system to simulate irregular geometry flows. The model is used to analyze a wide variety of hydraulic engineering problems including flows in a converging-diverging subcritical flume, supercritical expansions at various Froude numbers, and supercritical converging chutes. For each of these test cases, the calculated results are compared with experimental data. The comparisons with measurements as well as with other numerical solutions show that the proposed method is comparatively accurate, fast, and reliable.     

Modeling Nonuniform Suspended Sediment Transport in Alluvial Rivers

Problems and difficulties in modeling sediment transport in alluvial rivers arise when one uses the theory of equilibrium transport of uniform sediment to simulate riverbed variation. A two-dimensional mathematical model for nonuniform suspended sediment transport is presented to simulate riverbed deformation. Through dividing sediment mixture into several size groups in which the sediment particles are thought to be uniform, the nonuniformity and the exchange between suspended sediment and bed material are considered. The change of concentration along the flow direction, size redistribution, and cross-sectional bed variation can then be described reasonably well by the model. In simulating the flow field with big dry-wet flats, moving boundary problems are solved very well by introducing a so-called finite-slot technique. Verification with laboratory data shows that the model has a good ability to simulate channel bed variations. Last, the model was applied to a real alluvial river system. Variables such as water level, sediment concentration, suspended sediment size distribution, and riverbed variation were obtained with encouraging results.     

Limitations of Depth-Averaged Modeling for Shallow Wakes

Large-scale horizontal vortical structures are generic features of shallow flows which are often modeled using the two-dimensional (2D) depth-averaged equations. Such modeling is investigated for the well-defined case of shallow wakes of a conical island of small side slope for which a three-dimensional (3D) boundary-layer (3DBL) model has previously been validated through comparison with experiment. The 3DBL model used a 3D, two-mixing-length, eddy-viscosity turbulence model with a vertical mixing length of classical Prandtl form and a horizontal mixing length some multiple of this. A multiple of six gave good predictions. This mixing length approach is reduced to depth-averaged form, giving a horizontal mixing length of about half the water depth. The shallow wakes may be vortex shedding or steady/stable and are conventionally defined by a stability parameter. The critical value above which a stable wake is formed is considerably overestimated by the depth-averaged model (for a range of mixing lengths) and the length of stable wake bubble is considerably underestimated. It seems likely that this is because the amplification of friction coefficient due to horizontal strain rates is not represented. However, when vortex shedding is prominent the 2D and 3DBL wake structures are quite similar. These results show, for example, the limitations of depth-averaged models for the prediction of solute dispersion.     

Depth-Averaged Drag Coefficient for Modeling Flow through Suspended Canopies

Aquatic suspended canopies are porous obstacles that extend down from the free-surface but have a gap between the canopy and bed. Examples of suspended canopies include those formed by aquaculture structures or floating vegetation. The major difference between suspended canopies and the more common submerged canopies, which are located on the bottom boundary, is the influence of the bottom boundary layer beneath the suspended canopy. Data from laboratory experiments are presented which explore aspects of the flow through and beneath suspended canopies constructed from rigid cylinders. The experiments, using both acoustic Doppler and two-dimensional (2D) particle tracking velocimetry, give details of the flow structure that may be divided vertically into a bottom boundary layer, a canopy shear layer, and an internal canopy layer. The experimental data show that the penetration of the shear layer into the canopy is limited by the distance between the canopy and bottom boundary layer. Peaks in velocity spectra indicate an interaction between the bottom boundary and canopy shear layer. An analytical model is also developed that can be used to calculate a drag coefficient that includes the effect of both canopy drag and bed friction. This drag coefficient is suitable for use in 2D (depth-averaged) hydrodynamic modeling. The model also allows the average velocity within and beneath the canopy to be calculated, and is used to investigate the effect of canopy density and thickness on both total drag and bottom friction.     

Three-Dimensional Modeling of Nonuniform Sediment Transport in an S-Shaped Channel

A three-dimensional numerical model was applied to compute uniform and nonuniform sediment transport and bed deformation in an S-shaped laboratory channel located at the University of Innsbruck, where detailed measurements of the velocity field and bed elevation changes were made. The channel had two bends, a trapezoidal cross section, and a slope of S = 0.005. Gravel with a mean diameter of 4.2?mm was used as movable bed material and for sediment feeding. Wu’s formula for multiple grain sizes was compared with van Rijn’s formula using one grain size. Fairly good agreement was found between the computed and measured bed elevations for both approaches, whereas Wu’s formula could further improve the numerical results. Looking at the physics of the erosion pattern, the computed scour areas were located slightly more downstream than what was observed in the physical model. The current study also includes several parameter tests: grid distribution in vertical, lateral, and longitudinal direction; time step; number of inner iterations/time step; active sediment layer thickness; and the Shields coefficient. The variation of those parameters gave some differences in the results, but the overall pattern of bed elevation changes remained the same.     

Bend-Flow Simulation Using 2D Depth-Averaged Model

The purpose of this paper is to present a 2D depth-averaged model for simulating and examining flow patterns in channel bends. In particular, this paper proposes a 2D depth-averaged model that takes into account the influence of the secondary flow phenomenon through the calculation of the dispersion stresses arisen from the integration of the products of the discrepancy between the mean and the true velocity distributions. The proposed model uses an orthogonal curvilinear coordinate system to efficiently and accurately simulate the flow field with irregular boundaries. As for the numerical solution procedure, the two-step split-operator approach consisting of the dispersion step and the propagation step with the staggered grid is used to numerically solve the flow governing equations. Two sets of experimental data from de Vriend and Koch and from Rozovskii were used to demonstrate the model's capabilities. The former data set was from a mildly curved channel, whereas the latter was from a sharply curved channel. The simulations considering the secondary flow effect agree well with the measured data. Furthermore, an examination of the dispersion stress terms shows that the dispersion stresses play a major role in the transverse convection of the momentum shifting from the inner bank to the outer bank for flows in both mild and sharp bends.     

3D Numerical Modeling of Flow and Sediment Transport in Open Channels

A 3D numerical model for calculating flow and sediment transport in open channels is presented. The flow is calculated by solving the full Reynolds-averaged Navier-Stokes equations with the k ? ε turbulence model. Special free-surface and roughness treatments are introduced for open-channel flow; in particular the water level is determined from a 2D Poisson equation derived from 2D depth-averaged momentum equations. Suspended-load transport is simulated through the general convection-diffusion equation with an empirical settling-velocity term. This equation and the flow equations are solved numerically with a finite-volume method on an adaptive, nonstaggered grid. Bed-load transport is simulated with a nonequilibrium method and the bed deformation is obtained from an overall mass-balance equation. The suspended-load model is tested for channel flow situations with net entrainment from a loose bed and with net deposition, and the full 3D total-load model is validated by calculating the flow and sediment transport in a 180° channel bend with movable bed. In all cases, the agreement with measurements is generally good.     

Two-Dimensional Depth-Averaged Flow Modeling with an Unstructured Hybrid Mesh

An unstructured hybrid mesh numerical method is developed to simulate open channel flows. The method is applicable to arbitrarily shaped mesh cells and offers a framework to unify many mesh topologies into a single formulation. A finite-volume discretization is applied to the two-dimensional depth-averaged equations such that mass conservation is satisfied both locally and globally. An automatic wetting-drying procedure is incorporated in conjunction with a segregated solution procedure that chooses the water surface elevation as the main variable. The method is applicable to both steady and unsteady flows and covers the entire flow range: subcritical, transcritical, and supercritical. The proposed numerical method is well suited to natural river flows with a combination of main channels, side channels, bars, floodplains, and in-stream structures. Technical details of the method are presented, verification studies are performed using a number of simple flows, and a practical natural river is modeled to illustrate issues of calibration and validation.     

Soft Bedrock Erosion Modeling with a Two-Dimensional Depth-Averaged Model

Many rivers in Taiwan have steep slopes, are subject to typhoon-induced flood flows, and contain soft bedrock that is exposed at many locations and easily erodible. The occurrence of extensive bedrock erosion has been a major threat to river infrastructure at many locations. Soft bedrock erosion, therefore, is an important process to consider for river projects in Taiwan. In this study, bedrock erosion models are reviewed. A specific model is proposed by combining two existing models incorporating both the hydraulic and abrasive scour mechanisms. The proposed bedrock erosion model is incorporated into a two-dimensional mobile-bed model, and the integrated model is tested by simulating bedrock erosion downstream of the Chi-Chi weir on the Choshui River in Taiwan. A calibration study is performed to determine appropriate values of the model parameters based on two and a half years of measured data. The model is then assessed based on a verification study that compares model predictions of bedrock erosion of the same reach to two additional years of measured data. The bedrock erosion model is found to be suitable for the river reach studied. Further improvement, however, is still necessary, which points to potential future research.     

Validation of Existing Bed Load Transport Formulas Using In-Sewer Sediment

Granular sediment in pipe inverts has been reported in a number of sewer systems in Europe. Given the range of flow conditions and particle characteristics of inorganic sewer sediments the mode of transport may normally be considered as bed load. Current commercial software for modeling the erosion and transport of sediments in sewer pipes still utilizes well-known, or modified versions of transport equations that were derived for transport of noncohesive sediment in alluvial streams. In this paper the performances of the equations of Ackers and White (originally developed for the transport of river sediments) and of May (derived from laboratory pipe experiments) are examined against two separate data sets. One set is from laboratory erosion experiments on sewer sediment obtained in Paris. A second data set has bed load transport rate measurements recorded in a sewer inlet pipe. The formulas were selected because of their widespread use in the prediction of in-sewer sediment transport both in commercial software and in the latest United Kingdom design guidance for new sewers. The results indicated that both the relationships performed poorly, even in such well-controlled conditions. These formulas have significant difficulties in predicting the erosion thresholds and fractional transport rates for non-uniformly sized in-sewer sediments. An empirical formula to adjust the threshold of motion for individual grain size fractions was developed which significantly improved predictions. Although such techniques have been used in gravel bed rivers, the threshold adjustment function for in-sewer deposits was significantly different from these previously published for fluvial gravels, indicating that a direct transfer of fluvial relationships to sewers may be inappropriate without further research.     

Case Study: Modeling of Sediment Transport and Wind-Wave Impact in Lake Okeechobee

There are increasing demands for reliable engineering tools for sediment modeling and water resources management. The Lake Okeechobee environmental model (LOEM), which was calibrated and verified to simulate sediment resuspension and transport in Lake Okeechobee, Florida, is a dependable tool to meet those demands. The LOEM contains 2,126 horizontal grid cells and 5 vertical layers. The primary hydrodynamic and sediment transport driving forces are wind waves, surface wind stresses, and inflows/outflows. The LOEM was calibrated and verified, using two sets of observed data from May 16 to June 13, 1989 and January 17 to March 3, 2000, respectively. The model results indicate that sediment solids are resuspended primarily by wind–wave action and transported by lake circulation. The strong relationship between significant wave height and suspended sediment concentration in the lake indicates that sediment resuspension is primarily driven by wind-induced waves. To simulate this sediment resuspension, the processes of wind–wave- and current-induced bottom shear stresses on the lake bed were added to the LOEM. Once resuspended, the suspended sediment is transported to different areas of the lake by wind-induced currents. The importance of wind-wave, currents, and their interactions to sediment transport is included and discussed. By using the comprehensive data set for model calibration and verification, the LOEM model is proven to be a useful tool to water sources management in the lake.     

3D Numerical Modeling of Flow and Sediment Transport in Laboratory Channel Bends

The development of a fully three-dimensional finite volume morphodynamic model, for simulating fluid and sediment transport in curved open channels with rigid walls, is described. For flow field simulation, the Reynolds-averaged Navier–Stokes equations are solved numerically, without reliance on the assumption of hydrostatic pressure distribution, in a curvilinear nonorthogonal coordinate system. Turbulence closure is provided by either a low-Reynolds number k?ω turbulence model or the standard k?ε turbulence model, both of which apply a Boussinesq eddy viscosity. The sediment concentration distribution is obtained using the convection-diffusion equation and the sediment continuity equation is applied to calculate channel bed evolution, based on consideration of both bed load and suspended sediment load. The governing equations are solved in a collocated grid system. Experimental data obtained from a laboratory study of flow in an S-shaped channel are utilized to check the accuracy of the model’s hydrodynamic computations. Also, data from a different laboratory study, of equilibrium bed morphology associated with flow through 90° and 135° channel bends, are used to validate the model’s simulated bed evolution. The numerically-modeled fluid and sediment transportation show generally good agreement with the measured data. The calculated results with both turbulence models show that the low-Reynolds k?ω model better predicts flow and sediment transport through channel bends than the standard k?ε model.     

Three-Dimensional Modeling of Sediment Transport in a Narrow 90° Channel Bend

A three-dimensional numerical model was used for calculating the velocity and bed level changes over time in a 90° bended channel. The numerical model solved the Reynolds-averaged Navier-Stokes equations in three dimensions to compute the water flow and used the finite-volume method as the discretization scheme. The k-ε model predicted the turbulence, and the SIMPLE method computed the pressure. The suspended sediment transport was calculated by solving the convection diffusion equation and the bed load transport quantity was determined with an empirical formula. The model was enhanced with relations for the movement of sediment particles on steep side slopes in river bends. Located on a transversally sloping bed, a sediment particle has a lower critical shear stress than on a flat bed. Also, the direction of its movement deviates from the direction of the shear stress near the bed. These phenomenona are considered to play an important role in the morphodynamic process in sharp channel bends. The calculated velocities as well as the bed changes over time were compared with data from a physical model study and good agreement was found.     

Method for Assessing Impacts of Parameter Uncertainty in Sediment Transport Modeling Applications

The predictions from a numerical sediment transport model inevitably include uncertainty because of assumptions in the model’s mathematical structure, the values of parameters, and various other sources. In this paper, the writers aim to develop a method that quantifies the degree to which parameter values are constrained by calibration data and the impacts of the remaining parameter uncertainty on model forecasts. The method uses a new multiobjective version of generalized likelihood uncertainty estimation. The likelihoods of parameter values are assessed using a function that weights different output variables on the basis of their first-order global sensitivities, which are obtained from the Fourier amplitude sensitivity test. The method is applied to Sedimentation and River Hydraulics—One Dimension (SRH-1D) models of two flume experiments: an erosional case and a depositional case. Overall, the results suggest that the sensitivities of the model outputs to the parameters can be rather different for erosional and depositional cases and that the outputs in the depositional case can be sensitive to more parameters. The results also suggest that the form of the likelihood function can have a significant impact on the assessment of parameter uncertainty and its implications for the uncertainty of model forecasts.