Journal of Hydroinformatics

Published by IWA Publishing
Online ISSN: 1465-1734
Print ISSN: 1464-7141
General setup for adaptive control.
Model Reference Adaptive Control Scheme.
Open-channels networks used for water distribution are subject to algal developments that can induce major disturbances such as clogging issues on hydraulic devices (pipes, weirs, filters,...). We already studied the use of flushes to manage these algae developments. The flush is carried out by increasing the hydraulic shear conditions using the hydraulic structures of the canal network. In response to the shear stress increase, a part of the fixed algae is detached, then re-suspended into the water column, and finally transported into the canal network. This leads to a peak of turbidity. We develop in this paper a distributed linear model that is used for real-time adaptive control of the flushes. Simulations show the effectiveness of the adaptive controller, that can at the same time estimate the gain of the system, linked to the amount of fixed algae, and perform a flush without overtopping the turbidity limit.
Efficient flood management requires accurate real time forecasts to allow early warnings, real time control of hydraulics structures or other actions. Commercially available computing tools typically use, for flow or level forecasting, hydraulic models derived from the numerical approximation of Saint-Venant equations. These tools need powerful computers, accurate knowledge of the riverbed topography and skilled operators with some hydraulic background. This paper presents an alternative approach in which the river network is modeled as a cascade of interconnected input-output systems. Each system is modeled by an adaptive predictive expert model, which provides real-time fast and accurate forecasts over a moving prediction horizon. The main advantages of the approach are: (1) simplicity in the formulation and low computational burden; (2) no need of topographic information on the river waterbeds; (3) operators do not need strong hydraulic knowledge and the forecast may be done autonomously. The approach is evaluated using real data from the Ebro river basin in Spain.
We combine non-hydrostatic flow simulations of the free surface with a discharge model based on elementary gate flow equations for decision support in operation of hydraulic structure gates. A water level-based gate control used in most of today's general practice does not take into account the fact that gate operation scenarios producing similar total discharged volumes and similar water levels may have different local flow characteristics. Accurate and timely prediction of local flow conditions around hydraulic gates is important for several aspects of structure management: ecology, scour, flow-induced gate vibrations and waterway navigation. The modelling approach is described and tested for a multi-gate sluice structure regulating discharge from a river to the sea. The number of opened gates is varied and the discharge is stabilized with automated control by varying gate openings. The free-surface model was validated for discharge showing a correlation coefficient of 0.994 compared to experimental data. Additionally, we show the analysis of CFD results for evaluating bed stability and gate vibrations.
In order to provide safety against high sea water levels, in many low-lying countries on the one hand dunes are maintained at a certain safety level and dikes are built, while on the other hand large control structures that can be controlled dynamically are constructed. Currently, these structures are often operated purely locally, without coordination on actions between different structures. Automatically coordinating the actions is particularly difficult, since open water systems are complex, hybrid systems, in the sense that continuous dynamics (e.g., the evolution of the water levels) are mixed with discrete events (e.g., the opening or closing of barriers). In low-lands, this complexity is increased further due to bi-directional water flows resulting from backwater effects and interconnectivity of flows in different parts of river deltas. In this paper, we propose a model predictive control (MPC) approach that is aimed at automatically coordinating the different actions. Hereby, the hybrid nature is explicitly addressed. In order to reduce the computational effort required to solve the hybrid MPC problem we propose to use TIO-MPC, where TIO stands for time-instant optimization. A simulation study illustrates the potential of the proposed controller in comparison with the current setup in the Rhine-Meuse delta in The Netherlands.
In many process-based models, parameters have to be estimated from data. It is important to obtain not only the optimum value of the parameters, but also to assess the uncertainty in the parameters and, hence, in the models' output. in this paper, the Bayesian Monte Carlo technique known as Generalised Likelihood Uncertainty Estimation (GLUE) is used to evaluate the parameter-induced predictive uncertainty of a three-parameter model that predicts alongshore currents over a nearshore barred profile. GLUE performs a fully random sampling of feasible-parameters space, assigning non-zero likelihoods to those model simulations that outperform a user-defined threshold. Based on data gathered at six cross-shore position across an inner bar at Egmond aan Zee, The Netherlands, non-zero likelihoods were found for a rather wide range of parameter values, largely induced by an interdependence between two parameters that affect the width of current jets across the bar. The width of the 95% uncertainty interval was found empirically to increase linearly with the predicted magnitude of the alongshore current, from about 0.02-0.06 m/s when the current magnitude is near zero to about 0.2 m/s when it is near its maximum of about 1.1 m/s. These widths are approximately equal to a rough estimate of the errors in the data. In many cases the 95% uncertainty interval brackets the observations, although there are also various instances where this is not the case and apparently model structural errors dominate over parameter-induced errors. Model non-linearity and parameter interdependence cause the marginal parameter posterior distributions to differ remarkably from those obtained from traditional first-order approximations.
Time series analysis using nonlinear dynamics systems theory and multilayer neural networks models have been applied to the time sequence of water level data recorded every hour at ‘Punta della Salute’ from Venice Lagoon during the years 1980–1994. The first method is based on the reconstruction of the state space attractor using time delay embedding vectors and on the characterisation of invariant properties which define its dynamics. The results suggest the existence of a low dimensional chaotic attractor with a Lyapunov dimension, DL, of around 6.6 and a predictability between 8 and 13 hours ahead. Furthermore, once the attractor has been reconstructed it is possible to make predictions by mapping local-neighbourhood to local-neighbourhood in the reconstructed phase space. To compare the prediction results with another nonlinear method, two nonlinear autoregressive models (NAR) based on multilayer feedforward neural networks have been developed. From the study, it can be observed that nonlinear forecasting produces adequate results for the ‘normal’ dynamic behaviour of the water level of Venice Lagoon, outperforming linear algorithms, however, both methods fail to forecast the ‘high water’ phenomenon more than 2–3 hours ahead.
In high head dams, the kinetic energy at the spillway toe is very high and the tail-water depth available for energy dissipation is relatively small. Cascade stilling basins are energy dissipation systems for high head dams, the design of which is based on a trial-and-error procedure. Although such an approach yields feasible designs in which hydraulic and topographic considerations are met, there may exist many cost-effective designs. Therefore, optimization tools can help find the least construction cost while keeping hydraulic and topographic considerations satisfied. Particle Swarm Optimization (PSO) and Genetic Algorithms (GA) were used to determine the optimal design of cascade stilling basins in terms of the height of falls and length of stilling basins. The approach was evaluated by application to the design of an energy dissipation system for Tehri Dam on the Bhagirathi River. Comparison of the proposed methods with dynamic programming and an alternative approach not utilizing an optimization tool revealed that GA and PSO lead to significant savings in the construction cost with less computational effort.
In this paper the capability of PSO is employed to deal with the ANFIS inherent shortcomings to extract optimum fuzzy If-Then rules in noisy area arisen from application of nondimentional variables to estimate scouring depth. In the model, a PSO algorithm is employed to optimize the clustering parameters controls fuzzy If-Then rules in subtractive clustering while another PSO algorithm is employed to tune the fuzzy rules parameters associated with the fuzzy If-Then rules. The PSO models objective function is RMSE by which the model attempts to minimize the error of scouring depth estimation with respect to its generalization capability. To evaluate the model performance, the experimental data sets are used as training, checking and testing data sets. In the dimensional model the mean current velocity, mean grain size, water depth, pipe diameter, shear boundary velocity while in the nondimensional model the pipe, boundary Reynolds numbers, Froude number and normalized depth of water are set as the as input variables. The results show that the model provides an alternative approach to the conventional empirical formulas. It is evident that the PSO-FIS-SO is superior to ANFIS model in the noisy area that the input and output variables slightly related to each other.
This paper deals with modelling and identification of a river system using physical insights about the process, experience of operating the system and information about the system dynamics shown by measured data. These components together form a linear model structure in the state space form. The inputs of the prospective model are physical variables, which are not directly measured. However, the model inputs can be found by a nonlinear transformation of measured variables. Unknown parameters of the model are estimated from measured data. The modelling work focuses on the principle of parsimony, which means the best model approach is the simplest one that fit the purpose of the application. The goal of the model is to control the water level of the river, where the water flow is mainly determined by the demand for energy generation produced by the hydropower stations along the river. The energy requirement increases in the morning and decreases in the evening. These flow variations, caused by the energy demand, have to be compensated by controlling the power plants downstream, in such a way that the water level between the power stations is guaranteed. Simulation of the control system by using an adaptive model predictive controller shows that the water levels vary less and can be maintained at a higher level than during manual control. This means that more electric power can be produced with the same amount of water flow.
Simplified formalisation of concepts in 1-D hydraulics – UML class diagram. 
Elements for a modelling terminology, after Refsgaard and Henriksen. 
Formalisation of concepts for operational validation – UML class diagram. 
Procedural knowledge for model calibration – UML activity diagram. 
Model calibration remains a critical step in numerical modelling. After many attempts to automate this task in water-related domains, questions about the actual need for calibrating physics-based models are still open. This paper proposes a framework for good model calibration practice for end-users of 1D hydraulic simulation codes. This framework includes a formalisation of objects used in 1D river hydraulics along with a generic conceptual description of the model calibration process. It was implemented within a knowledge-based system integrating a simulation code and expert knowledge about model calibration. A prototype calibration support system was then built up with a specific simulation code solving subcritical unsteady flow equations for fixed-bed rivers. The framework for model calibration is composed of three independent levels related, respectively, to the generic task, to the application domain and to the simulation code itself. The first two knowledge levels can thus easily be reused to build calibration support systems for other application domains, like 2D hydrodynamics or physics-based rainfall-runoff modelling.
ABSTRACT B. G. Ruessink Department of Physical Geography, Faculty of Geosciences, Institute for Marine,and,Atmospheric,Research Utrecht, Utrecht University, PO Box 80.115, 3508 TC Utrecht, The Netherlands Tel: þ31 30 2532405 Fax: þ31 30 2531145 E-mail: The physically realistic functions,implemented,in nearshore,process,models,are governed,by parameters that usually do not represent measurable attributes of the nearshore and, therefore, need,to be determined,through,calibration. The classical approach,to calibrate nearshore,process models,is via manual,parameter,adjustments,and visual comparisons,of model,predictions,and measurements. In this paper a hybrid genetic algorithm, comprising a global population-evolution- based search strategy and a local Nelder–Mead simplex search, is used to calibrate nearshore process,models,in an objective and automatic,manner.,The effectiveness,of the algorithm,to find the optimum,parameter,setting are examined,in two case,studies with increasing,complexity: a simple,alongshore,current model,and a more,complex,cross-shore bed evolution,model. Whereas the algorithm,is found to be an effective method,to find the optimum,setting of the alongshore current model, it fails to identify the optimum parameter values in the bed evolution model, related to the strong interaction between,two,of the parameters,in the suspended,sediment transport,equation. Setting one of the interdependent,parameters,to a constant,value within its feasible space while retaining the other in the optimization,procedure,is found to be a feasible solution to the ill-posed optimization,problem,of the bed evolution,model. Key words | calibration, genetic algorithm, nearshore, optimization, physically based simulation modelling NOTATION
| Location map of Hirakud dam in Mahanadi Basin. 
| Pareto optimal solutions obtained for the Hirakud reservoir operation problem by EM-MOPSO and NSGA-II, for a typical run ( f1 ¼ annual squared deficit for irrigation ( £ 10 6 m 3 ) 2 and f2 ¼ hydropower production £ 10 6 kWh).
| Model results for reservoir operation using EM-MOPSO, showing the ten daily release policies and power generated, corresponding to the compromised optimal solution. RI ¼ release to irrigation; RP ¼ release to hydropower production; HP ¼ hydropower.
| The ten daily storage operating policy (corresponding to the compromised optimal solution) obtained for reservoir operation using EM-MOPSO.
Optimal allocation of water resources for various stakeholders often involves considerable complexity with several conflicting goals, which often leads to multi-objective optimization. In aid of effective decision-making to the water managers, apart from developing effective multi-objective mathematical models, there is a greater necessity of providing efficient Pareto optimal solutions to the real world problems. This study proposes a swarm-intelligence-based multi-objective technique, namely the elitist-mutated multi-objective particle swarm optimization technique (EM-MOPSO), for arriving at efficient Pareto optimal solutions to the multi-objective water resource management problems. The EM-MOPSO technique is applied to a case study of the multi-objective reservoir operation problem. The model performance is evaluated by comparing with results of a non-dominated sorting genetic algorithm (NSGA-II) model, and it is found that the EM-MOPSO method results in better performance. The developed method can be used as an effective aid for multi-objective decision-making in integrated water resource management.
This article was published in the journal, Journal of Hydroinformatics [© IWA Publishing] and is available at: In this paper the development of a Case-Based reasoning system for Estuarine Modelling (CBEM) is presented. The aim of the constructed CBEM system is to facilitate the utilisation of complex modelling software by users who lack detailed knowledge about modelling techniques and require training and assistance to implement sophisticated schemes effectively. The system is based on modern computing methods and is constructed as a hybrid of three modules which operate conjunctively to guide the user to obtain the best possible simulation for realistic problems. These modules are: a case-based reasoning scheme a genetic algorithm and a library of numerical estuarine models. Based on the features of a given estuary and the physical phenomenon to be modelled, an appropriate solution algorithm from the system’s library is retrieved by the case-based module after a specifically designed reasoning process. The selected model is then analysed and further treated by the genetic algorithm component to find optimum parameters which can appropriately model the conditions and characteristics of any given estuary. Finally, the user is provided with a procedure that gives the best solution for a problem using the available hydrographic data and under the specified conditions. As an illustrative example and to show the applicability of the present CBEM system under realistic conditions a case study based on the simulation of salinity distribution in the Tay estuary (Scotland, UK) is given in this paper.
Relationship between the data, such as direct observations of nature and recorded measurements, and the models is very complicated in the ‘water domain’. It is not at all as clear and explicit as it is often presented by teachers to students, by consultants to clients, or by authors to readers of publications. A number of aspects of this relationship are discussed using examples to illustrate the author's views. Limitations of data-driven tools (correlations, Artificial Neuronal Networks, Genetic Algorithms, etc.) and data-mining, when applied without physical knowledge of the relevant phenomena, are discussed, as are those of deterministic models. The currently used ‘good practice’ paradigm in modelling (the model is to be set up, calibrated, validated and run) is rejected when deterministic models are concerned. They should not be calibrated. A new paradigm, a new ‘code of good practice’, is proposed instead. Strategic and tactical aspects of various available approaches to modelling of physical phenomena and data exploitation have practical engineering and financial consequences, most often immediate and sometimes very important: hence the significance of the subject that concerns the everyday occupations of modellers, their clients and end-users.
Ships docking and traveling in a branched lower approach channel system are at risk from surges caused by multi-lane locks during emptying operations. For this reason, water-level variations in the lower approach channel in response to discharge, interval running time, and outlet location of lock operations were studied using a 2-D hydrodynamic model validated by physical model tests, and the impact of water level variation on navigation safety under extreme operation scenarios of a quadruple-lane lock group was identified. Results indicated that discharge and interval running time of lock emptying had the greatest impact on the water level variation at the lock head. Water level variation at the lower lock head of the ship lift exhibited a trend of first decreasing and then increasing with the increment of the discharge from the lock chamber into the outer river. Specifically, the surge height at the lock head of the ship lift reached the minimum when approximately 40% of the discharge generated by dual-lane locks during emptying operations was released into the outer river. Overall, the simultaneous operation of quadruple-lane lock group and unit load rejection should be avoided in engineering applications. HIGHLIGHTS A 2-D hydrodynamic model was utilized to analyze the water-level fluctuations in a branched approach channel system.; Water-level fluctuations in response to discharge, interval running time, and emptying outlet location of lock operations were revealed in details.; Extreme multi-lane lock operating scenarios were determined considering the requirement of safe docking of ships at the lower lock head of the ship lift.;
Turbulent flow in meandering open channels is one of the most complicated and unpredictable turbulent flows as the interaction of various forces, such as pressure gradient, centrifugal force, and wall shear stresses severely affect the flow pattern. In order to improve significance in engineering application, understanding the overall flow characteristic is the focus. This paper presents the results of numerical and experimental investigations of flow in a 180° mild bend, which is close to criticality with curvature ratio R/B = 3. Considering the characteristic of various models, three-dimensional (3D) re-normalization group (RNG) k-∈ model was adopted to simulate the flow efficiently. Governing equations of the flow were solved with a finite-volume method. The pressure-based coupled algorithm was used to compute the pressure. The flow velocities were measured experimentally with Micro acoustic Doppler velocimeter. Good agreement between the numerical results and measurements indicated that RNG k-∈ model can successfully predict this flow phenomenon. The flow pattern in this bend is influenced widely by the secondary flow. The variations of velocity components, streamlines, secondary flow, and wall shear stresses are analysed in the study. Some newly discovered phenomenon in this special state are worth noting.
Evaluation of pluvial flood risk is often based on computations using 1D/2D urban flood models. However, guidelines on choice of model complexity are missing, especially for 1D network models. This study presents a new automatic approach for simplification of 1D hydraulic networks (SAHM) using trimming and merging techniques, with performance evaluated in a 1D/2D case study. Decreasing the number of elements in the 1D model by 66% yielded a 35% decrease in computation time of the coupled 1D/2D simulation. The simplifications increased flow in some downstream branches and removing nodes eliminated connection to some areas. This promoted errors in 2D flood results with changes in spatial location of flooding in the reduced 1D/2D models. Applying delayed rain inputs to compensate for changes in travel time and preserving network volume by expanding node diameters did not improve overall results. Investigations on the expected annual damages ( EAD ) showed that differences in EAD are smaller than deviations in the simulated flooded areas, suggesting that spatial changes are limited to local displacements. Probably, minor improvements of the simplification procedure will further improve results of the reduced models.
This paper presents the results of a calibrated 1D/2D coupled model simulating surface and sewer flows in Barcelona. The model covers 44 km2 of the city land involving 241 km of sewers. It was developed in order to assess the flood hazard in the Raval district, historically affected by flooding during heavy rainfalls. Special attention was paid to the hydraulic characterization of the inlet systems (representing the interface between surface and underground flows), through experimental expressions used to estimate the effective runoff flows into the sewers in case of storms. A 2D unstructured mesh with more than 400,000 cells was created on the basis of a detailed digital terrain model. The model was calibrated and validated using four sets of well-recorded flooding events that occurred in 2011. The aim of this paper is to show how a detailed 1D/2D coupled model can be adequately calibrated and validated using a wide set of sewer sensors and post-event collected data (videos, photos, emergency reports, etc.). Moreover, the created model presents significant computational time savings via parallel processing and hardware configuration. Considering the computational performances achieved, the model can be used for real-time strategies and as the core of early warning systems.
A water supply system including radial wells (RWs) is usually large, extending several tens of kilometers horizontally and is dozens of meters deep. An RW is a vertical shaft with lateral screens, which radially penetrate the soil. A large and complex 3D finite element (FE) mesh also needs to include laterals with cross-sectional dimensions measured in centimeters. An adequate representation of lateral screens by line elements, with nodes coinciding with the 3D FE mesh nodes, is desirable in order to simplify modeling, render the computation efficient, and present the results in an easily readable form. Line elements are introduced for the lateral screens and accuracy of the results is analyzed. It was found that the domain size of an RW (or laterals) has a more pronounced effect on accuracy than mesh density. The authors' conclusion is that the concept of 1D RW lateral screens representation is adequate for practical purposes.
Test case I: Evolution of J with respect to K g for each method. Coloured dots represent the different MC evaluations  
The calibration of parameters in complex systems usually requires a large computational effort. Moreover, it becomes harder to perform the calibration when non-linear systems underlie the physical process, and the direction to follow in order to optimize an objective function changes depending on the situation. In the context of shallow water equations (SWE), the calibration of parameters, such as the roughness coefficient or the gauge curve for the outlet boundary condition, is often required. In this work, the SWE are used to simulate an open channel flow with lateral gates. Due to the uncertainty in the mathematical modeling that these lateral discharges may introduce into the simulation, the work is focused on the calibration of discharge coefficients. Thus, the calibration is performed by two different approaches. On the one hand, a classical Monte Carlo method is used. On the other hand, the development and application of an adjoint formulation to evaluate the gradient is presented. This is then used in a gradient-based optimizer and is compared with the stochastic approach. The advantages and disadvantages are illustrated and discussed through different test cases.
The paper concerns the numerical solution of one-dimensional (1D) and two-dimensional (2D) advection–diffusion equations. For the numerical solution of the 1D advection–diffusion equation, a method, originally proposed for the solution of the 1D pure advection equation, has been developed. A modified equation analysis carried out for the proposed method allowed increasing of the resulting solution accuracy and, consequently, to reduce the numerical dissipation and dispersion. This is achieved by proper choice of the involved weighting parameter being a function of the Courant number and the diffusive number. The method is adaptive because for uniform grid point and for uniform flow velocity, the weighting parameter takes a constant value, whereas for non-uniform grid and for varying flow velocity, its value varies in the region of solution. For the solution of the 2D transport equation, the dimensional decomposition in the form of Strang splitting technique is used. Consequently, this equation is reduced to a series of the 1D equations with regard to x- and y-directions which next are solved using the aforementioned method. The results of computational experiments compared with the exact solutions confirmed that the proposed approaches ensure high solution accuracy of the transport equations. HIGHLIGHTS Solution of 2D advection–diffusion equation using the Strang splitting method.; Solution of 1D advection–diffusion equation using the modified finite element method.; Modified equation analysis yields the coefficients of numerical dissipation and dispersion.; For variable flow velocities, the weighting parameter varies in space and time.; The adaptive algorithm ensuring the constant order of approximation up to the 4rd order.;
In this work, an one-dimensional (1D) finite volume numerical model for the unsteady simulation of the flow hydrodynamics and water quality is developed. The water dynamics is formulated with the 1D shallow water equations, and the water quality evolution is described by the Water Quality Analysis Simulation Program (WASP) model, allowing us to interpret and predict the transport and fate of various biochemical substances along any river reach. This combined system is solved with an explicit finite volume scheme based on Roe's linearization for the advection component of both the flow and the solute transport equations. The proposed model is able to consider temporal variations in tributaries and abstractions occurring in the river basin. This feature is transcendent in order to predict the chemical composition of natural water bodies during winter and summer periods, leading to an improvement in the agreement between computed and observed water quality evolutions. The combined model has been evaluated using literature tests in a steady state and a real-field case of the Ebro river (Spain), characterized by a marked unsteady regime. In the real case, we found that the water temperature was very sensitive to both the solar radiation and the average air temperature, requiring a careful calibration of these parameters. The numerical results also demonstrate to be reasonably accurate, conservative and robust in real-scale field cases, showing that the model is able to predict the evolution of quality parameters as well as hydrodynamic variables in complex scenarios.
In the pumping station, the main feedwater system and the reactor system of nuclear power plant, power-supply failure causes strong hydraulic transients. 1D-MOC is used to calculate the transient process in the pipeline system while the 3D computational fluid dynamics is employed to analyze the turbulent flows inside the pump and to obtain the performance parameters of the pump, and the data exchanges on the boundary conditions of the shared interface between 1D and 3D domains are updated based on the MpCCI platform. Based on the equation of motion of the pump motion parts, the relationship between the external characteristics and the internal flow field in the pump is further investigated because the dynamic behavior of the pump and the detailed fluid field evolutions inside the pump are captured during the transition process, and the transient flow rate, rotating speed, and pressure inside impeller are comprehensively investigated. Meanwhile, compared with the data gained by experiment and traditional 1D-MOC, the relative errors of rotating speed and the flow rate obtained by 1D-3D coupling method are smaller than those by 1D-MOC. Furthermore, the influences of the main coupling parameters and coupling modes on the calculation results are analyzed, and the cause of the deviation is further explained.
| Examples of the random-field realizations. (a) and (b) represent the 35th and 50th realizations of the 2D random field of the floodplain Manning's roughness coefficient; (c) and (d) represent the 35th and 50th realizations of the 1D/2D random field of the channel/floodplain Manning's roughness coefficients over the modelling domain. Note: the modelling domain is divided into 76 (in x axis) × 48 (in y axis) grid elements; the 2D covariance function for the random field n(x) corresponds to the normalized Cartesian coordinates x 1 , x 2 ∈ [0, 1] and η x ¼ η y ¼ η nc ¼ 0.03.
| Comparison of statistics of h(x) simulated by optimal SRSM with Set 1 and MCS along the cross-section profile x N ¼ 21/76. (a), (c) and (e) represent the means of h(x); (b), (d) and (f) represent the Stds of h(x) under Scenarios 3, 4 and 5, respectively. Note: Scenarios 3, 4 and 5 are assumed with inflow levels at 36.5, 146 and 219 m 3 /s, respectively; and R 2 and RMSE are shown in sequence in the legend for each combination of parameters.
shows a comparison of statistics of results forecasted by SRSM-253 with Set 1 and the corresponding MCS (with 10,000 realizations) along the cross-section profile x N ¼ 21/76. It appears that more grid elements would get inundated when inflow level increases. This leads to a
| Boxplot comparison of SRSM-253 vs. MC in predicting h(x) at three different grid locations. (a) Inflow ¼ 36.5 m 3 /s; (b) inflow ¼ 73 m 3 /s; (c) inflow ¼ 146 m 3 /s; (d) inflow ¼ 219 m 3 /s. Note: the middle line within the box is mean of the results; the top and bottom lines of the box are the 75 and 25 percentiles of the predicted results, respectively; the bars at the top and bottom represent upper and lower whiskers, respectively. Besides, at the top of the figure, the first line of indexes are relative error of the predicted means between the MCS and SRSM-253 and the indexes in the second are the relative error of the predicted confidence interval for each grid location, respectively.
An uncertainty assessment framework based on Karhunen-Loevexpansion (KLE) and probabilistic collocation method (PCM) was introduced to deal with flood inundation modelling under uncertainty. The Manning's roughness for channel and floodplain were treated as 1D and 2D, respectively, and decomposed by KLE. The maximum flow depths were decomposed by the 2nd-order PCM. Through a flood modelling case with steady inflow hydrographs based on five designed testing scenarios, the applicability of KLE-PCM was demonstrated. The study results showed that the Manning's roughness assumed as a 1D/2D random field could efficiently alleviate the burden of random dimensionality within the analysis framework, and the introduced method could significantly reduce repetitive runs of the physical model as required in the traditional Monte Carlo simulation (MCS). The study sheds some light on reducing the computational burden associated with flood modelling under uncertainty which is useful for the related damage quantification and risk management.
Transient flow characteristics and dissipation mechanism in pressurized pipeline were investigated based on 1D friction models and 3D turbulence models, where the pressure–density model was combined into the 3D continuity equation allowing for the elasticity of the fluid and the pipes. The applicability of 3D realizable k–ε and 3D SST (shear stress transport) k–ω turbulence models was verified with comparison to 1D traditional water hammer models and the experimental data for fast closing of the valve in the reservoir–pipe–valve system. The valve closure rule was instantaneously carried out using the grid slip CFD (computational fluid dynamics) technique. The SST k–ω turbulence model has the highest accuracy in predicting the pressure attenuation of transient flows. The 3D detailed flow field confirms that the asymmetric flows induced by the change of valve opening within approximately three-fourths of the pipe inner diameter before the valve are captured. In the pressure wave cycles, the unsteady inertia, axial pressure gradient, viscous shear stress and turbulent shear stress mainly influence the velocity variations. During the pressure wave propagation, the viscous and turbulent dissipation are critical in the pressure attenuation in the wall region; the viscous dissipation is mainly concentrated in the viscous sublayer, while the turbulent dissipation increases to the maximum values at y+ = 13–23. HIGHLIGHTS The 3D weakly compressible and dissipation mechanism models are introduced to investigate transient flow characteristics.; The asymmetric flows induced by the change of valve opening are captured well within approximately three-fourths of the pipe inner diameter before the valve.; The viscous dissipation is mainly concentrated in the viscous sublayer, while the turbulent dissipation has maximum influence at y+ = 13–23.;
| Diagram sketching model comparisons.
Simulation times for different hydrographs (H1, H2, H3 and H4) and for different number of 1D cells on river bed running in CPU/GPU.
Normalized simulation times for different hydrographs (H1, H2, H3 and H4) and for different number of 1D cells on river bed running in CPU/GPU.
Coupled 1D2D models emerged as an efficient solution for a two-dimensional (2D) representation of the floodplain combined with a fast one-dimensional (1D) schematization of the main channel. At the same time, high-performance computing (HPC) has appeared as an efficient tool for model acceleration. In this work, a previously validated 1D2D CPU model is combined with an HPC technique for fast and accurate flood simulation. Due to the speed of 1D schemes, a hybrid CPU/GPU model that runs the 1D main channel on CPU and accelerates the 2D floodplain with a Graphics Processing Unit (GPU) is presented. Since the data transfer between sub-domains and devices (CPU/GPU) may be the main potential drawback of this architecture, the test cases are selected to carry out a careful time analysis. The results reveal the speed-up dependency on the 2D mesh, the event to be solved and the 1D discretization of the main channel. Additionally, special attention must be paid to the time step size computation shared between sub-models. In spite of the use of a hybrid CPU/GPU implementation, high speed-ups are accomplished in some cases.
Following the release of the OpenMI 2.0 standard for model coupling with reference object classes (interfaces) in C# and Java, a set of tools including a Software Development Kit (SDK) and Graphical User Interface (GUI) is expected to accompany it. These are necessary to enable numerical model developers to easily adapt their models to become OpenMI compliant and to allow modellers to easily assemble and run compositions of them. FluidEarth 2 is an HR Wallingford initiative providing these open source tools for the .net 4.0 Framework together with training, community support and sample models. They are the only such open source tools available so in this sense they act as the reference SDK and GUI for OpenMI 2.0 with .net. The purpose of this paper is to outline these and demonstrate a set of examples. A series of components were successfully constructed and compositions built. These include training models designed to demonstrate different aspects of model coupling, moving to industry strength model codes simulating dam-break bathymetry updates. The FluidEarth 2 tools have been designed to be cross-platform and have been tested under Windows and Linux (using Mono). Usage is successfully demonstrated, providing an environment for integrated modelling with OpenMI 2.0.
Satellite-based precipitation products and reanalysis precipitation products have the potential to overcome the lack of information in regions where there are no or insufficient rain gauges to achieve any hydrological study. The Google Earth Engine (GEE) data analysis platform has products in its repository with global coverage that offers different geospatial information capable of measuring the amount of precipitation. However, it is necessary to evaluate the reliability of the products. There are precipitation information biases in Mexico due to the scarce presence of gauging stations, failed operations, access difficulty, and data capture errors. This study evaluates the reliability of satellite and reanalysis precipitation products hosted in the GEE repository against rain gauge observation from 2001 to 2017 using data from 4,658 stations over Mexico. The evaluation was carried out using statistical indicators comparing the behavior across different topographic, climatic, and temporal conditions. The results exhibit that the performance of the products hosted in GEE seems to depend on elevation conditions for other climatic regions in Mexico. The results show that all products can capture the general precipitation patterns at annual, seasonal, and monthly scales; however, the accuracy of the product is clearly lower at a daily scale. All products are highly biased on low precipitation events. HIGHLIGHTS The intensity and distribution of precipitation in Mexico depends on topography and climatic regions.; ERA5 product in Mexico showed the lowest correlations of the entire region, while CHIRPS is the product with the best score.; There is a more significant correlation of products in the regions and seasons with the highest presence of rainfall.;
The River Drava is one of the major, inexhaustible water sources not only for Croatia, but also for the other European countries it flows through. This study is based on the observations of 15 water variables at three sampling stations in the lower River Drava over a 24 year period. Although the obtained results revealed an improvement of most of the parameters, the values of some of them (i.e. NH4-N, NO3-N, BOD5, total coliform and heterotrophic bacteria) are still above the approved limits for water Class II. The results of principal component analysis (PCA) confirmed an existence of. three clearly separated zones. The first zone corresponds to a rural upstream part of River Drava, which is characterised with low level pollution. The influences of untreated domestic waters become more noticeable in the second more densely populated suburban zone (II) located upstream of the city of Osijek. According to the results of the PCA, untreated wastewaters from Osijek are becoming contributing factors to the high pollution level of the river in the third (Ill) suburban zone. This study shows the usefulness of the PCA method for analysis and interpretation of complex data sets as well as for determination of pollution sources.
An important feature of the two-layer shallow flow model is that the resulting system of equations cannot be expressed in conservation-law form. Here, the HLLS and ARoe solvers, derived initially for systems of conservation laws, are reformulated and applied to the two-layer shallow flows in a great variety of problems. Their resulting extension and combination allows to overcome the loss of the hyperbolic character, ensuring energy or exactly balanced property, guarantees positivity of the solution, and provides a correct drying/wetting advance front without requiring tuning parameters. As a result, in those cases where the rich description of internal and external waves cannot be provided by the ARoe solver, HLLS is applied. Variable density is considered in each layer as a result of a bulk density driven by the mixture of different constituents. A wide variety of test cases is presented confirming the properties of this combination, including exactly balanced scenarios in subcritical and subcritical-transcritical scenarios, dam-break problems over bed variations and wet/dry fronts, non-hyperbolic conditions, transcritical exchange flow with loss of hyperbolicity. Despite the complexity of the test cases presented here, accurate and stable simulations are guaranteed, ensuring positivity of the solution without decreasing the time step.
The two-dimensional (2D) axial-symmetric model is applied to investigate the transient cavitating flows in the reservoir-pipeline-valve (RPV) system. Firstly, the MacCormack scheme is used to solve the governing equations, and compared to the numerical results of the one-dimensional (1D) model, the 2D head peaks and time-dependent evolutions predicted by five-region turbulence model in the frozen form are in better agreement with the experimental results, and the comparisons show that the maximum head relative errors of the 2D model are generally smaller than those of the 1D model. Then, further numerical simulations are carried out to investigate the performances of different turbulence models incorporated in the 2D model. The comparisons between the numerical results and the experimental ones show the head curves of the two-region turbulence model are similar to those of the five-region turbulence model, which indicates that the transient cavitating flows are insensitive to the magnitude and distribution of the eddy viscosity in the core region. In addition, the sensitivity to the quasi-steady form and the frozen form of the five-region turbulence model is implemented in the 2D model; the numerical results predicted by the two forms both agree better with the experimental results, as the non-dimensional parameter P increases.
a) Plane view of the flume. The shaded area stands for the zone covered by sand. b) Position of the water level gauges.
Configuration 1 measured topography after 20 s in metres.
Final topography as measured by Soares-Frazão et al. (2019) with the focus put only on the second part of the channel that was rotated by 270° for convenience reasons.
Unstructured mesh of the 90° bend test case.
We propose a finite-volume model that aims at improving the ability of 2D numerical models to accurately predict the morphological evolution of sandy beds when subjected to transient flows like dam-breaks. This model solves shallow water and Exner equations with a weakly coupled approach while the fluxes at the interfaces of the cells are calculated thanks to a lateralized HLLC flux scheme. Besides describing the model, we ran it for four different test cases: a steady flow on an inclined bed leading to aggradation or degradation, a dam-break leading to high interaction between the flow and the bed, a dam-break with a symmetrical enlargement close to the gate and a dam-break in a channel with a 90° bend. The gathered results are discussed and compared to an existing fully coupled approach based on HLLC fluxes. Although both models equally perform regarding water levels, the weakly coupled model looks to better predict the bed evolution for the four test cases. In particular, its results are not affected by an excessive numerical diffusion encountered by the coupled model. Moreover, it usually better estimates the amplitudes of the maximum deposits and scours. It is also more stable when subject to high bed–flow interaction.
A numerical modelling of flow dynamics in a tidal river mouth of comprehensive morphology is assumed to be one of the most effective methods of both scientific research and civil engineering projects. Realistic results of simulations can be obtained only on the basis of field observations. This approach is realized for a 2D hydrodynamic model of the Northern Dvina River mouth area. The Northern Dvina delta has a very complicated distributary network and suffers from both spring snow-melt floods and autumn storm surges. The STREAM_2D software package based on the 2D shallow water equations was used for the model development. The model was calibrated and validated on the background of water level data at state gauges and special water discharges measurements in the essential delta branches during the semi-diurnal tidal cycles. Sensitivity tests were provided to evaluate the most significant reasons for model errors. It was discovered that the distribution of roughness coefficients amidst delta channels and floodplain does not affect the flow dynamics in the delta significantly. However, the tidal range variations over a neap-spring cycle and mean sea level changes along the delta marine edge are of major importance.
A numerical model is presented for simulation of hydrodynamics of a 2D vertical free surface domain consisting of an arbitrary partitioned porous and non-porous area. To this end, modified Navier–Stokes equations are considered which could be applied in surface water and in subsurface flows, simultaneously. A wide range of Reynolds number has been considered, from which non-Darcy effects have also been taken into account. A fractional step method has been used in the time discretization procedure, where the convection and diffusion terms are separated from the pressure term while solving the momentum equations. To include the variation of surface elevation in computation, the domain has been divided into two parts, namely, ‘interior subdomain’, which never gets dry during the simulation period, covered by fixed unstructured triangular grids and ‘top layer’ with only a one layer structured grid, the position of which varies with the water surface. The validation of the proposed model has been achieved by comparison of its results with both theoretical and experimental data reported in the literature.
2D non-uniform polygonal meshes allow representation of the impact of landscape elements and small infrastructures on water flows. The initial vectorial mesh, derived from the intersection of several geographical information systems f layers, can have highly non-convex or sliver polygons. These bad-shaped elements compromise accurate numerical flow computation. We propose a flexible divide-and-conquer strategy to decompose polygons into physiographical meaningful parts using shape descriptors to better represent the surface terrain and hydrologic connectivity. We use the convexity index (CI) and the form factor (FF) to consider convex and square like optimum shapes. The strategy was applied to two peri-urban areas whose hydrologic response was simulated using distributed modeling. Good-quality meshes were generated with threshold values of CI.0.8 and FF.0.2, and CI.0.95 and FF.0.4 for undeveloped and highly urbanized zones, respectively. We concluded that the mesh segmentation facilitates the representation of the spatially distributed processes controlling not only the lumped response of the catchment, but also the spatial variability of water quantity and fluxes within it at medium and small scales.
The numerical modelling of circulations in shallow lakes is a relevant tool for all environmental applications in which flow advection processes are of interest, e.g. for studies on nutrients, microorganisms, pollutants and sediment dynamics. While 3D models are needed to properly describe the flow fields of basins with the main circulations in the vertical plane, 2D models are commonly deemed to yield adequate results for lakes with prevailing horizontal circulations. However, the depth-averaged approximation is more limiting for wind-driven flows than for gravity-driven ones, such as rivers, as the driving force is a surface rather than a volume one, distributed along the depth through turbulence. In this work, the effects of such inaccuracy on the reproduction of circulation layouts are evaluated through compared simulations between a 2D Shallow Water solver and a 3D Reynolds-Averaged Navier-Stokes one. The models are first applied to a simple enclosed elliptical test basin and then to the real case of the Superior Lake of Mantua, a shallow fluvial lake in Northern Italy, thereby also investigating the influences of the interaction of wind with a riverine current and of a complex bathymetry on the compared results.
Top-cited authors
Shie-Yui Liong
  • National University of Singapore
Vladan Babovic
  • National University of Singapore
Fayzul Pasha
  • California State University, Fresno
Sivapragasam Chandrasekaran
  • Kalasalingam Academy of Research and Education
Maarten Keijzer