University of Twente Student Theses
An analysis to test the sensitivity of output values of flood modelling software D-HYDRO to characteristics of the river, dike and hinterland
Lambers, M. (2020) An analysis to test the sensitivity of output values of flood modelling software D-HYDRO to characteristics of the river, dike and hinterland.
|
PDF
3MB |
| Abstract: | Flood models are used to sketch an image of the possible effects of a flood from sea, rivers, lakes, regional and local water systems. Within flood models, 1D, 2D and 1D2D calculations are used to determine the course of a flood. Companies and institutes are working on developing new software packages which are able to simulate flood patterns even faster and more accurate than the already existing flood models can do. HydroLogic is a research and consultancy office which strives to come up with innovative water solutions with the help of ICT. To live up to these standards, HydroLogic is collaborating with Deltares, which is an independent institute for applied research in the field of water and subsurface, to develop new software which can be used for the modelling of floods in coastal areas, estuaries, rivers and rural and urban areas. The new software, D-HYDRO, is in contrary to most other flood modelling software able to make a flexible grid of the study area with both squares and triangles with different sizes. This makes it possible to adapt the level of detail within a study area. HydroLogic has already done two pilot studies to test the possibilities of the new software. Within this study, preliminary sensitivity analysis were performed to gather information about the way the output values of the flood model in D-HYDRO were influenced by different values of the input parameters. A more systematic and comprehensive sensitivity analysis is needed to evaluate the effects of changes in input parameters on output of D-HYDRO. Therefore, the aim of the research is to “determine the sensitivity of the output of D-HYDRO software to changes in input parameters which describe characteristics of the river, the dike and the hinterland”. The research will focus on designing a simplified flood model with only a few standard characteristics adapted the study area like a height map and an uniform friction. Furthermore, this study area will be a small area of 4 by 6 kilometer. Within the study, 4 research questions were established. The first question was made to gather information about how to systematically conduct the sensitivity analysis of the research. The second research question was established to investigate which characteristics of the river, the dike and the hinterland should be tested within the sensitivity analysis of the research and find out how these characteristics are translated to input parameters and what the ranges of these tested parameters are. The third and fourth research question consisted of the sensitivity analysis of the characteristics of the river, the dike and the hinterland. In the research, the determination of the method which should be used for the sensitivity analysis, is done by means of literature research. Hereby, literature regarding flood modelling with 1D, 2D and 1D2D calculations was searched for. It could be concluded that an independent and local sensitivity analysis needs to be done. This should be done by changing the input parameters One-At-a-Time (AOT). Furthermore, the chosen parameters should be varied over their whole parameter range. Before the sensitivity analysis can be conducted, a reference situation needs to be determined. When the different simulations runs of the sensitivity analysis has been completed, the output values of these runs should be compared with the output values of the reference situation. Hereby, a ranking can be made of the relative contribution to the output variability of the different input parameters which has been tested. For the determination of the characteristics which are going to be tested within the sensitivity analysis, literature has been consulted again. Next to the guideline prepared by Deltares (2018), the reports of the pilot studies from HydroLogic were used to provide information about which input parameters, with which ranges, should be used when building a flood model. In the end, for the tested characteristics of the river and dike, the location of the breach would be varied over two locations and the material of the dike would be varied over 5 values. For the characteristics of the hinterland, the roughness of the surface would be varied over 5 uniform friction values. Furthermore, also the roughness map of the pilot study of the Grebbedijk would be applied to test the influence if a non-uniform friction of the surface is used. Next to that, line element sand waterways will be added to different simulation runs on 8 different locations with different directions to test the sensitivity of the output values of the flood model to these characteristics. Based on the guideline of D-HDYRO and with support from HydroLogic, the simplified flood model has been designed. Next to the height map of the pilot study of the Grebbedijk and an uniform friction equal to the friction of grassland, boundary conditions for the water level of the river and the discharge through the breach were added. These boundary conditions were derived from the pilot study of the Grebbedijk. The results of the sensitivity analysis of the characteristics of the river and dike showed that the material of the dike has a far greater influence on the output values of the flood model in D-HDYRO than the location of the breach. A change in the material from the dike from clay to sand caused a doubling in the width of the breach which also resulted in a doubling of the maximum discharge through the breach. For the sensitivity analysis of the roughness of the surface of the hinterland and the presence of line elements and waterways, less clear results were shown. Variation of the uniform friction always has an effect of the output values whereas the line elements and waterways only have an influence if they disturb the flow direction of the flood. Furthermore, the presence of waterways were often the reason for a smaller inundated area whilst the presence of line elements often lead to a bigger inundated area. This could be because a waterway makes it more easy for the flood to travel to the lower places in the area, whereas a line element makes it harder for the flood to travel to these places since they can block (some of) the easy routes of the flood. Whilst it was difficult to draw a clear conclusion, it could still be concluded that, based on the ranking that was done, the roughness of the surface caused the biggest ranges for most of the output values compared to the other characteristics of the hinterland that has been tested. A big limitation of this study is the use of the small study area and the use of the simplified model. This meant that some interesting characterises of the river, the dike and the hinterland could not be tested with the sensitivity analysis. On top of that, simplistic representations have been used for, for example, the line elements and waterways. Therefore, the results of the sensitivity analysis should be taken with a grain of salt. Furthermore, the limited time that was set for the research leaded to a univariate sensitivity analysis where the mutual relationships of characteristics could not be tested. Moreover, this limited time ensured that a link between the sensitivity analysis of the characteristics of the river and dike with those of the characteristics of the hinterland could not have been made. For further research it is therefore advised to make use of a real existing study area and a fast super computer which will make it possible to carry out a lot of simulation runs where the sensitivity analysis can be based on. For the final conclusion of the research, it can be said that the aim of the research is party fulfilled. The sensitivity of the output of D-HYDRO software is determined for the characteristics which were chosen to test. Still, due to the scope and other limitations of the research, not all of the important characteristics could have been tested and no link could be made between the sensitivity analysis of the characteristics of the river and dike and those of the characteristics of the hinterland. |
| Item Type: | Essay (Bachelor) |
| Faculty: | ET: Engineering Technology |
| Programme: | Civil Engineering BSc (56952) |
| Link to this item: | https://purl.utwente.nl/essays/85020 |
| Export this item as: | BibTeX EndNote HTML Citation Reference Manager |
Show download statistics for this publication
Show download statistics for this publicationRepository Staff Only: item control page