Samenvatting

This research held two main intentions, the first being to analyze what subsurface layouts have what kinds of effects on pipe length for the dike failure mechanism, and the second being to create a model using this information and knowledge. The first intention was completed using two different programs: D-Geo Flow and Timflow. D-Geo Flow was a valuable program for this research since it is a visual program that shows the inputs and output of the model. Many different iterations were completed using the program, but it is a timely process to complete the iterations. It acted as a good foundation for understanding the outcome of different changes, but completing every iteration with the program was challenging.

Had this been known previously, more time should have been spent with Timflow instead of D-Geo Flow. A few iterations regarding a clay layer spanning entirety, the left half, and right half of the aquifer width were completed. Due to constraints with time, more iterations/iteration types, and more in-depth improvement to the Timflow approach could not be completed.

In the future, it could be useful to optimize these models and to complete more iterations so that there could be more data surrounding many different variations in the subsurface. Additionally, a recommendation for even further in the future would be to implement transient flow modules. The author only explored Static flow due to time and knowledge constraints, but applying realistic water conditions to the model may produce more accurate results.

For the second intention to create the model itself, many different variations in modeling method and application were applied. GemPy proved to be simple to use, but its outputs were overly simplified and took too long to calculate. The IPyWidget model had a friendly interface when complete but also took extremely long to load. The final model was then created using custom Python functions.

This final model allows the user to apply custom parameters to the function which then creates a custom subsurface visualization model. It is applicable elsewhere in Zeeland and the Netherlands as long as the inputted dataset is in the correct format. The model also is capable of visualizing different types of sections such as a cross section which passes through a dike, or a section that follows the shape of the dike, and could include only data from the inner toe and or/crest of the dike.

This method is also capable of applying a conclusion from D-Geo Flow, that an intermittent clay layer in the subsurface acts as a barrier to the flow of water, and therefore is a major factor in whether or not piping can occur at a location. This finding gets applied and then the visualization is simplified underneath these clay layers.

To continue this research, it would be extremely valuable to find a method to include the other D-Geo Flow findings outlined in this report. Currently only one major finding is included, but information on anisotropy for example, would be an additional valuable piece of data that could be included in the piping specific visualization of the subsurface.