Graduation of Rokus de Bie

Reducing hydraulic loads on river dikes by construction of longitudinal mounds of local soil

• Professor of graduation: Prof. dr. P.M.J. Herman

• Supervisors: Dr. ir E. Mosselman (TU Delft) and Ir. M. Peters (Sweco)

For the past 2000 years river training works have been performed on the Dutch rivers. These training works have shaped the river system of today. Another result of the river training works, especially over the last two centuries, is a decrease in biodiversity. To protect the biodiversity some floodplains in the Netherlands have been classified as Natura2000 areas. Construction works on these floodplains are not allowed, unless these measures are a last resort. Also, if this is the case, compensating measures have to be taken elsewhere.

Before 2050 about 1500 kilometres of dikes and 500 sluices and pumping stations need reinforcements. Dike reinforcements could be executed by only adding soil to the dike. Another option is to add structural elements to the dike. A soil-based approach is preferred because there is more experience and a higher level of security of the reliability for a soil-based structure. 

One such soil-based approach is a longitudinal mound. A longitudinal mound is a body of soil which is parallel with the dike, with the goal to reduce the wave height at the dike itself. As a result of the wave height reduction the necessary dike crest level will be reduced as well. Therefore, a reinforcement of the dike itself is not needed. The crest of this longitudinal mound is lower than the crest height of the dike. The longitudinal mound will be submerged during (flood) design conditions and will act like a submerged wave breaker.

Costs, emissions and construction time could potentially be reduced by using local soil. This local soil can be obtained in two different ways. Firstly, it is possible to use the surplus of soil of another local project for the longitudinal mound. Secondly, the soil for the longitudinal mound could be taken from the floodplain itself. 

However, only little is known about the hydrodynamic effects of a longitudinal mound on the floodplain. This thesis research is done to find possible locations for a longitudinal mound, the hydrodynamic effects and the differences between a simple and more complex model of the longitudinal mound. This is done with a multicriteria analysis for the location study and with a conceptual model and a 2D D-Flow FM model for the hydrodynamic effects.

In the multicriteria analysis the studied criteria are the size of the floodplain, structures on the floodplain and inside the dike, the availability of clay on the floodplain, the habitats on the floodplain and the wave height at the dike.

The multicriteria analysis has been performed from the point of view from multiple stakeholders. For all locations a compromise is necessary. Different locations for a longitudinal mound are preferred depending of the point of view of the stakeholders.

In the conceptual model three design parameters for the longitudinal mound are taken into account, the crest height, the crest width and the slope. For each combination of these three parameters the conceptual model calculates the new equilibrium water level and the transmitted wave height from the longitudinal mound towards the dike. The transmitted wave height is calculated with the best empirical fit on multiple datasets by Friebel and Harris in 2003. 

With the Van der Meer overtopping formula the freeboard of the dike above the water level can be determined. This is done for the original situation without longitudinal mound and subsequently for the situation with all combinations of the longitudinal mound. From these calculations it can be concluded that the necessary dike crest height decreases when a longitudinal mound is present. However, more soil is needed than for a traditional dike reinforcement.

Also the conceptual model does not include a backwater effect. The water level does not immediately jump to the new equilibrium water level, so the water level increase should be smaller than calculated in the conceptual model. On the other hand, in the conceptual model all waves are assumed to be perpendicular to the dike. If waves are not perpendicular the necessary freeboard is smaller. The absolute dike crest height reduction with a longitudinal mound is therefore smaller for non-perpendicular waves than for perpendicular waves.

The 2D D-Flow FM model has been supplied by Deltares. The grid consists of cells of 20 by 10 square metres on the main river channel and 20 by 20 square metres on the floodplain. To model the longitudinal mound with a higher accuracy the grid on the floodplain has been refined to 5 by 5 square metres. On this refined grid three different variants have been modelled. All variants have a crest height of about half a metre below design water level and their alignment is identical. For Variant 2 a connection of half the longitudinal mound height has been made with the dike. For Variant 3 the same volume of soil needed for the longitudinal mound has been removed from the floodplain by lowering it by 0.3 metres. 

There are only small differences between the three variants. Compared to the original situation there was only a difference in the order of millimetres of water level at the main river channel. The main differences are found between the dike and the longitudinal mound. In this area the Bernoulli effect is found, at locations of increased flow velocity lower water levels are found and vice versa. The subsequent difference in water level is about 5 to 10 centimetres. 

The flow velocity depends on the difference of flow area in longitudinal direction between the longitudinal mound and the dike. So, the main contributor to the water level change on the floodplain is the alignment of the longitudinal mound. Therefore, the alignment of the longitudinal mound is an important design parameter and can be used to find a trade-off between increased water levels and increased flow velocity.

As this process is not incorporated in the current version of the conceptual model the results between the conceptual model and the 2D D-Flow FM model are different. Therefore, it is recommended that the water levels between the longitudinal mound and dike are calculated separately in the conceptual model. To do this the area between the dike and longitudinal mound can be split into multiple segments. With energy and momentum balances the water levels in these segments can be calculated. 

It is also recommended that the 2D D-Flow FM model is used at a smaller floodplain as well to see if the effect on the main river channel is similarly small. Next, it could be helpful to try different alignments for the longitudinal mound to see how these influence the water levels and flow velocities.

Finally, in this research only the flow has been modelled in 2D. However, the wave reduction is also of importance. The next step is to add a wave model to the 2D model to as well. With this addition it would be possible to make the comparison between the wave height reduction in the conceptual model relative to a 2D model as well as for the water level.