A predictive view on Durbans flood safety

Abstract

Durban is the third largest city of South-Africa, located in the province of KwaZulu-Natal. The city suffers from severe floods from time to time, finding its cause in both the Indian ocean as well as the Umgeni river. The eThekwini municipality wishes a better insight in the occurrence of these floods and searches for a structural solution to protect the coastline. The eThekwini municipality has models in operation to predict the hydraulic characteristics in the ocean and the river. However, the existing models don’t represent the reality sufficiently, since the interaction between the Indian ocean and the Umgeni river is not modelled properly yet. An analysis on the area of interest has been executed. The conclusion was drawn that the Umgeni river delta was (partly) tide-dominant, meaning that the Indian ocean imposes the downstream water level. Furthermore, the wave climate was observed, as well as a look into present coastal protections. The link between the Indian ocean and the Umgeni river has been modelled using Delft3D. Since the Indian ocean imposes a downstream boundary condition (in terms of a water level) for the Umgeni river, a backwater curve might occur. First, the link is made by extending the Delft3D-model which was present for the Indian ocean only. The model has been extended all the way up to the Inanda dam. The part of the river included in the new model is approximately 32 푘푚 long. When comparing the models output at the river mouth, at the same location as a measurement point, similar behaviour can be observed. The same phase (lag) is observed, contrary to the tidal range. The tidal range in the model differs from reality, but this is due to a lack of calibration in the amplitudes of the different tidal constituents taken into account. Hence, the renewed model seems to work, but more validation still has to be done. This was not possible yet, as there is a lack of measurement stations along the river. Next to an extension of the Delft3D model, a script has been written in Python. This script is based on the empirical fit of Bresse and shows an elegant function. The results from the function in Python and the model in Delft3D are similar in a qualitative and a quantitative way. Both the models show an influence of the Indian Ocean, reaching easily to about 12 푘푚 upstream of the river mouth. This can be explained by the mild bed slope in this part. A structural solution for the flooding on the promenade at the height of North Beach was found in the form of a seawall. The most important design demand is to protect against a high water level of a 200 year return period combined with a 50 year return period wave height. These storm conditions are input for the ocean-river model, which delivers wave characteristics at the beach front, linking the structural design to the ocean-river model. After a pre-selection on design options, a Multi Criteria Analysis is carried out for the remaining eight design options. Grading is done based on criteria, representing the viewpoints of the many stakeholders involved and leading to a highest grading of a seawall in combination with an emergency barrier. Following, the water-retaining height for a vertical wall is determined. Given the the ground level height of the promenade to be 푀푆퐿 + 2.2 푚 and a total water-retaining height of 푀푆퐿 + 2.944 푚 this leads a practical construction height of 0.80 푚. Due to the limited height a reinforced concrete seawall is designed with emergency barriers for the beach entrances. The emergency barriers are designed of pinewood. Additionally, in order the fit properly in the surroundings, an integrated design is added with features like benches, thatch umbrellas and plants to disguise the construction and protect the Golden Mile in style.