Transitions on dikes have a large impact on the dike stability during wave overtopping events. It is quite common to find dikes with integrated hard structures such as roads, which also represent bottom irregularities and surface roughness changes which directly influence the erosion rates of dike covers from the overtopped waves. All these effects may be approximated and simulated with complex fluid dynamics numerical solutions, but their computational burden makes them sometimes unfeasible for their use in flood risk analysis.
For tackling this challenge, Dr. Juan Pablo Aguilar-López et al. from the WEM department developed a surrogate modeling methodology which allowed to estimate the spatially distributed failure probability of the grass cover of a dike with a road on top. This was done by training response surfaces at each location of interest along the dike based on the output from few runs of a two-dimensional Reynolds average Navier-Stokes (RANS) with k epsilon turbulence closure approximation in two phases. These already trained response surfaces allowed to estimate the bottom shear stress which was later used as input for an erosion model. These coupled models where implemented in a crude Monte Carlo reliability method for estimating the failure probabilities along the dike.
The results from this study showed that dikes with roads on top result in higher failure probabilities (5 × 10−5 > Pf >1 × 10−4) with respect to dikes without a road (Pf <1 × 10−6); all tested for extreme type boundary conditions for representing climate change scenarios with different grass qualities. Grass quality also proven to have significant influence in the failure probability. Nevertheless, the resulting probabilities where low enough to assume that the dikes are still quite safe when roads are present on top.
You can download the full paper (open access) from: doi.org/10.3390/jmse6030074