[Back]

W. Morales Peñuela:
"River dyke failure modeling under transient water conditions";
Supervisor, Reviewer: D. Adam, S. Springman; Institut für Geotechnik, 2013; oral examination: 2013-12-10.

Knowledge of the performance of river dykes during flooding is necessary when designing governmental assistance plans aimed to reduce both casualties and material damage. This is especially relevant when floods have increased in their frequency during the last decades, together with the resulting material damage and life costs.
Most of previous attempts for analyzing dyke breaching during flooding have neglected to consider the soil mechanics component and the influence of infiltration and saturation changes on the failure mechanisms developed in the river dyke. This research project aimed to fill that gap in knowledge by analyzing, in a comprehensive manner, the effect of transient water conditions, represented by successive flood cycling, on the seepage conditions and subsequent breaching of dykes. Therefore, three key sub-projects were carried out: the anal-ysis of the results from an overflow field test, the physical modeling of small-scaled models under an enhanced gravity field, and the numerical modeling of the flow response and the resulting stability of both the air- and water-side slopes.
The overflowing field test was carried out in a section of a dyke enclosed within a rectangular sheet-pile box along the Rhone River in Baltschieder, south west Switzerland. This formed the concluding part of another research project (Mayor, 2013), which had been devoted to the analysis of the response of a dyke to successive flood cycles.
The grass cover and a low erodability gravel on the crown of the dyke prevented the it from being eroded superficially, as was expected. An instability event on the air-side slope, fol-lowed by internal erosion, was observed instead. Laboratory tests were carried out to deter-mine the unsaturated flow parameters of the silty sand composing the main body of the dyke. These, together with parameters estimated from empirical relationships, such as the Kozeny-Carman (Carrier, 2003) and modified Kovacs (Aubertin et al., 2003) equations, allowed nu-merical modeling of the experiment to be completed with success. Both the flow response and stability of the dyke were simulated to represent the response that was observed during the test.
The physical modeling was performed by testing 12 small-scale models at an increased gravity of 33-g. These represented dykes of 5 m height at prototype scale, with three different slope gradients (1:2.0, 1:2.5, 1:3.0), and included one or two protective measures (a toe fil-ter, a cut-off wall) plus a homogeneous dyke. The goal was to analyze the effect of each pro-tective measure on the groundwater flow during flood cycles and on the breaching mecha-nism that developed during an overflow event. Therefore, two cycles of floods, with a subse-quent overflow were applied to all of the models. The flood cycles had a sinusoidal shape, each one with a duration of 20 days. The overflow was intended to replicate a hydrograph measured in a Swiss river during a flood in 2005.The sand used to build the models was characterized by compiling information from previous research projects, which had used the same soil. Specific tests were performed in order to determine the mechanical (water content controlled triaxial test and suction controlled oe-dometer tests) and hydraulic parameters (Soil Water Retention Curve) under unsaturated conditions.
Two types of dyke breaching mechanisms were identified. If a cut-off wall was not included, water started eroding the soil surface, creating a breach throat, through which water flowed rapidly, which, in turn, increase the size of the throat. A second type of breaching mechanism was observed when a cut-off wall was placed. A breach throat started to develop in the crest of the dyke, closer to the wall, in a similar manner to that observed for the dykes without the wall. When the throat reached the cut-off wall, it could not continue increasing towards the water-side. Instead, the soil in front of the wall, i.e. on the air-side, started to be eroded, cre-ating a narrow and shallow breach zone in the vicinity of the wall.
A numerical simulation of the unsaturated groundwater flow for all twelve dykes was carried out with commercial software based on the Finite Element Method (FEM), which allows the governing equation for flow through unsaturated porous media to be solved. Additionally, the variation in time of the stability of both air- and water-side slopes was investigated using a limit equilibrium approach.
The results from the numerical simulations matched accurately with the results obtained with the centrifuge modeling, including the prediction of local instabilities during the flood cycles for those dykes that did not include a toe filter. This was a consequence of an appropriate definition of the boundary conditions of the problem, together with an accurate estimation of the soil parameters through specific laboratory tests.

RIVER DYKE FAILURE MODELING

http://publik.tuwien.ac.at/files/PubDat_223739.pdf