Influence of Residual Dyke Strength on Dyke Reliability Using the Random Material Point Method

Abstract

The material point method (MPM) is used to model both rotational and horizontal sliding failure mechanisms of dykes under external (water) loading. To model the different failure mechanisms, an external hydrostatic water pressure has been applied on the canal side of the dyke by applying a newly developed boundary condition. The boundary condition detects the material boundary and distributes the applied load to the nodes of the background mesh. The definition of dyke failure has also been investigated using MPM. In conventional dyke assessment, using (for example) the finite element method (FEM), the dyke is considered to have failed as soon as an initial failure occurs. However, the dyke may still be able to resist the flow of water, and this continuing ability to resist water flow is known as residual dyke strength. By taking account of residual dyke strength, for example with MPM as shown in this paper, the computed reliability can increase compared to conventional assessment, as in some cases total dyke failure does not occur after an initial slope failure. Finally, spatial variability is considered using the random material point method (RMPM), which combines random fields with MPM in a Monte Carlo framework. When considering spatial variability, a significant gain in reliability due to residual dyke strength has been observed, but further investigation is required to fully understand the effect of spatial variability on residual dyke strength. In order to simplify this preliminary investigation, the adopted soil properties in this paper have not been based on actual soils used in dyke construction; the results are only intended to indicate the capabilities of RMPM and develop hypotheses on the effect of residual dyke strength.