Graduation of Sanne van Dijk

09 February 2023 15:00 till 02 February 2023 17:00 - Location: CiTG - Lecture Hall F | Add to my calendar

Assessing the applicability of the Sellmeijer piping design rule for hydrogeological systems in Limburg.

  • Professor of graduation: Dr. J.P. Aguilar-López

  • Supervisors: Dr. T. Bogaard (TU Delft), R. Koopmans (Arcadis)

The Maas overflowed its riverbanks and flooded parts of Limburg during the flood events in the Netherlands in 2021. Despite the high-water levels, the primary flood defenses performed admirably and exceeded expectations, with only 13 piping incidents reported along the Maas. This contradiction ignited renewed interest in the applicability of the piping assessment method, the Sellmeijer design rule (Sellmeijer et al., 2011), in Limburg. The Applicability of the Sellmeijer Design Rule in Limburg is questioned because the heterogeneous and multilayered subsoil differs from the Sellmeijer model, and the average representative grain size in Limburg lies outside the parameter application ranges used to derive the limit state equations.

This thesis investigates whether the Sellmeijer design rule can be applied to Limburg dike sections near Well, Hout-Blerick, Buggenum, and Thorn. This is accomplished by using the software COMSOL Multiphysics to create a finite element numerical model to assess the piping process in the hydrogeological systems of the research locations. The proposed model is built by combining several groundwater and piping model principles, schematizing dike cross-sections, and calibrating the model.

The results of an analytical piping assessment using the Sellmeijer design rule and a deterministic piping assessment using the proposed model show that neither the original (Sellmeijer, 1988) with the new geometry factor (Sellmeijer et al., 2011), nor the revised Sellmeijer design rule (Sellmeijer et al., 2011), applies flawlessly to the investigated Limburg research locations. The critical head determined by the design rule is either conservative or exceeds the dike's crest height, whereas the proposed FEM numerical model indicates that piping does not occur for outside water levels equal to the dike's crest level. Making the critical heads above the crest level insignificant.

Finally, a stochastic evaluation of the Sellmeijer design rule is carried out. In the proposed FEM numerical models, the evaluation includes the piping assessment of 1000 randomly generated scenarios. The scenarios are generated by implementing randomly sampled model parameters into FEM numerical models. The model parameters are sampled from carefully selected uniform distributions, using a statistical method known as Latin Hypercube Sampling (Olsson et al., 2003).

52 out of 1000 scenarios are used to evaluate the Sellmeijer design rule because these are the scenarios in which piping occurs for realistic pipe geometries. The evaluation consists of comparing the stochastically determined scale factor with the scale factor corresponding to the equilibrium condition of the failure scenarios. By performing a linear regression through the data points, it is found that the scale factors differ by a factor of 1.56. This implies that the original Sellmeijer design rule (Sellmeijer, 1988) with the new geometry factor (Sellmeijer et al., 2011), can be applied to the dike cross-section in Limburg with an additional factor of 1.56. This factor may also be applied to other dike sections for which the values of the model parameters fall in between the bounds of the proposed parameter distributions.