Graduation of Maurits Kampen

27 juni 2017 | 9:30 - 11:00
plaats: Room G, faculty of Civil Engineering and Geosciences
door Webredactie

“Economic optimisation of multi-layer flood safety: Applicability of the risk based optimisation process under sensitivity to uncertainties and budget constraints. Case study: Nyaungdon, Myanmar”| Professor of graduation: Matthijs Kok, supervisors: Sander van Nederveen (TU Delft), Martine Rutten (TU Delft), Saskia van Vuren (HKV, TU Delft)


Living in the vicinity of rivers has many advantages, but also has one major drawback: fluvial flooding. People have been trying to reduce the risk of flooding for a long time. Risk is hereby divided into the product of probability of flooding and its consequences. This means that flood risk can not only be reduced by lowering the probability of floods, but also by reducing the consequences of a flood.

This notion is captured in the multi layered safety approach, which classifies flood safety measures into three layers: prevention, spatial planning and crisis management. With the current computation power, optimal flood management systems can be designed so that the total costs of investments in measures and the remaining flood risk are as low as possible. This so called risk based economic optimisation is presently being implemented in the more developed countries, but offers potential for developing countries too. With a limited budget, every penny needs to be well spent. It is however questionable if this method fits the circumstances in developing countries. Sufficient data is needed to get reliable results, while developing countries are often data scarce. The sensitivity to uncertainties in the optimisation process due to working with limited data is explored by means of a case study concerning the area of Nyaungdon, Myanmar, which inundates regularly due to the Ayeyarwady River.

The risks of the case study area are mostly estimated on the basis of open source data. Only prerequisite is that a time series of the river is available to derive the probabilities of flooding. Inundation maps can be constructed based on NASA’s Digital Elevation Model, after which they can be verified  with Landsat images of historic flood events. Landsat images are furthermore used to construct land use maps. When these maps are combined in a geographic information system together with land use values and depth-damage curves, a spatial and quantitative estimation of the flood risk over the area is found. It is however essential that information about the current embankments is available, otherwise the probabilities of flooding cannot be determined with high confidence. The quantification of the risk is sensitive to the inclusion of data uncertainties. Within the case study the risk has a mean present value of $18.7 million with a standard deviation of $6 million. Especially the estimation of loss of life is a source of high uncertainty.

The risk reducing measures that are chosen for this case study are dike heightening of the current dike, placing poles under the dwellings located in the floodplain and construct shelters to reduce the loss of life. While the uncertainties in risk and investment costs are large, the choice for optimal design remained the same. Lowest total costs are found when dike heightening is used. This layer also corresponds to the lowest needed investments. Dike heightening can be combined with shelters, which become especially attractive when a high statistical valuation of life is used. Within the case study, the spatial planning layer is not only more receptive to the influence of uncertainties, from economic perspective it is also least attractive due to its high investment costs. The amount of dike heightening also remained stable under the inclusion of uncertainties. The shape of the curve that relates dike heightening to Total Costs was as such that its minimum value will always be located around the same location, irrespective the uncertainties: one meter dike heightening, corresponding to a new danger level of 8.5m +MSL. It must however be noted that there is worked with symmetrical uncertainty distributions, other shapes will have more effect on the optimal dike heightening. The optimisation results can still serve as a good indicator of economically most appropriate designs.

The robustness of the optimisation process within the case study does not necessarily offer conclusions for other cases as well, as this is situation and measure specific. It is expected that the outcomes will be robust under uncertainties as long as the Total Costs curve has the same shape as   in the case study: fast declining at first, before slowly rising again. With respect to the allocation of resources over different safety layers under budget constraints, it is advisable to invest first in those measures that have the steepest declining Total Costs curve. These measures have the highest ratio between risk reduction and needed investments. The derived risk assessment method is nevertheless executable for other areas that are comparable in size, due to its use of information that is openly available.


© 2017 TU Delft