Development of a 2D hydraulic model for the Rhine Valley using open data
Kriebel, M. (2016) Development of a 2D hydraulic model for the Rhine Valley using open data.

PDF
4MB 
Abstract:  For current Dutch river management, flood waves are predicted using the Generator of Rainfall and Discharge Extremes for the Rhine and Meuse basins (GRADE). One of the components of GRADE is a onedimensional hydraulic model that is made with the SOBEK modelling software. However, this could be replaced by a twodimensional hydraulic model, which has a more continuous representation of the area’s topography. Furthermore, a 2D model can calculate both longitudinal and transversal movements, making it more accurate in simulating floodplains. This could yield different results than the current calculations for the maximum flood wave. This report describes the development of such a 2D model for the Rhine valley in Delft3D Flexible Mesh, and discusses its performance compared to the 1D hydraulic model that is currently implemented in GRADE. The development of the twodimensional hydraulic model started with finding the minimum required model extent. This is done by calculating the probable extent of flooding of the river’s floodplains in combination with the backwater adaptation length in each of the tributaries that could occur in the case of an extreme discharge event. Then, an estimation of the basic bathymetry of the Rhine is made, which is then incorporated in the digital elevation model (DEM) that is created by the Shuttle Radar Topography Mission (SRTM). After this, a computational grid is created and assessed on characteristics such as orthogonality and smoothness. Lastly, initial and boundary conditions have been extracted from the 1D model where possible, or made as similar to the 1D model as possible to be able to make a good comparison. Different possibilities in these components have resulted in the development of several slightly different models. These models have been compared to each other first so that the effects of different model properties could be evaluated, after which the twodimensional model has been compared to the onedimensional hydraulic model of GRADE. Comparison of the twodimensional models with varying properties resulted in three main findings:  The minimum required model extent for an extreme discharge event in the Rhine valley can be accurately determined by calculating the probable extent of flooding of the river’s floodplains in combination with calculations on the probable tributary backwater adaptation length.  A twodimensional hydraulic model that uses a relative fine computational grid simulates less flooding in terms of volume compared to a coarser model, resulting in more discharge at Lobith. Despite this difference, the coarse grid model in this study can be appointed as the most optimal solution since its computational time is almost four times lower.  Changes in the initial water depth only influence the simulated discharges at the start of the simulation, and cause relatively little difference. The use of a socalled restart file as a starting point for the twodimensional model appears to have a significant effect on the simulated discharges. Comparison of the onedimensional model with the twodimensional model resulted in three main findings:  Discharge peaks that are simulated by the twodimensional model arrive later compared to the peaks in the onedimensional model.  The discharges that are simulated by the twodimensional model are, generally speaking, lower than those that are simulated by the onedimensional model.  In general, the hydrographs of the one and twodimensional models at Maxau and Bonn share the most similarities, while their hydrographs at Mainz and Lobith differ the most. An important reason for these differences is the fact that the 1D model starts at Maxau, while the 2D model starts at Basel. Because comparison of the two models was needed, the input of the 1D model is also used in the 2D model. Since this data did not contain a discharge time series for Basel, the time series for Maxau has been used here. As a consequence, it takes the water an estimated 1.4 days longer to reach Lobith compared to the one model. This delay can also cause discharge peaks in tributaries to flow into the Rhine before the flood wave in the Rhine arrives at the tributary, causing the peak discharge to be more spread out over time. Furthermore, the differences can be explained by the fact that the twodimensional model simulates lower flow velocities than the onedimensional model. The reason for this is twofold. Firstly, the crosssectional area of the river is much larger during a flood, causing the flow velocity to decrease. This increase of crosssectional area is better simulated by the twodimensional model. Secondly, transversal movements that are simulated by the twodimensional model cause the longitudinal velocity to decrease. These transversal movements are not simulated by the onedimensional model, explaining the differences in discharge and discharge peak timing. This is also the reason why the differences between the one and twodimensional model are the largest at Mainz and Lobith, since these are the areas that experience the most flooding according to the twodimensional model. 
Item Type:  Essay (Bachelor) 
Faculty:  ET: Engineering Technology 
Subject:  56 civil engineering 
Programme:  Civil Engineering BSc (56952) 
Link to this item:  http://purl.utwente.nl/essays/70421 
Export this item as:  BibTeX EndNote HTML Citation Reference Manager 
Repository Staff Only: item control page