Development and experimental validation of a multiscale DEM-FEM model for core overtopping in embankment dams

当水位上升超过防渗心墙，进入分区筑堤坝的反滤层时，心墙漫顶发生。它可能导致在核心和过滤器之间的界面处的内部侵蚀，并最终导致溃坝。通过保持上游水库的低水位，可以避免堆芯漫顶。然而，水库水位低可能对大坝所有者产生经济后果。对于复杂坝形、瞬变流和宽级配心墙材料，心墙漫顶过程中接触侵蚀进展的试验数据很少。因此，对于大坝业主来说，更好地了解心墙漫顶的实际影响是很重要的。该项目的主要目标是收集真实几何形状、材料和流动条件下的岩心漫顶实验数据，并为这一现象建立一个数值模型。在项目的第一阶段，将用实验室模型对不同的岩心和滤层粒径分布进行实验复制。试验材料的选择应确保某些组合不符合公认的抗侵蚀稳定性设计标准。将对具有广泛级配心墙的加拿大土石坝代表性组合进行试验。在项目的第二阶段，一个数值模型将被编程以复制实验结果。在模拟依赖于微观行为的现象时，如内部侵蚀，在实践中广泛使用的连续模型表现不佳。数值模型将基于一种新的多尺度方法。它将把联合收割机颗粒尺度建模与离散单元法结合起来，在离散单元法中对单个颗粒的位移进行建模（YADE软件），并将大尺度建模与有限元法结合起来，在有限元法中将土壤建模为连续体（COMSOL软件）。目前加拿大有超过10，000座水坝。一个实用的多尺度模型和实验数据的侵蚀阈值和侵蚀率为现实的几何形状和材料将允许采取适当的行动，在核心越流的情况下。它也将有助于设计新的结构。

Core overtopping occurs when the water level rises over the impermeable core, into the filter of zonedembankment dams. It can lead to internal erosion at the interface between core and filter, and eventually to damfailure. Core overtopping can be avoided by maintaining a low water level in the upstream reservoir. However,low reservoir levels can have economic consequences for dam owners. Experimental data on the progression ofcontact erosion during core overtopping for complex dam geometries, transient flow and widely graded corematerials are scarce. As a consequence, it is important for dam owners to get a better understanding of the realimpact of core overtopping. The main objectives of this project are to gather experimental data on coreovertopping for realistic geometries, materials and flow conditions, and to develop a numerical model for thisphenomenon. In the first phase of the project, core overtopping will be replicated experimentally with alaboratory model for different core and filter grain size distributions. The tested materials will be chosen so thatsome combinations fail to respect the accepted design criteria for stability against erosion. Combinationsrepresentative of Canadian embankment dams with a widely graded core will be tested. In the second phase ofthe project, a numerical model will be programmed to replicate the experimental results. The continuummodels that are widely used in practice perform poorly when modelling phenomena that are dependent onmicroscale behaviour, such as internal erosion. The numerical model will be based on a novel multiscaleapproach. It will combine particle scale modelling with the discrete element method, where the displacement ofindividual particles is modelled (software YADE), and large scale modelling with the finite element method,where the soil is modelled as a continuum (software COMSOL). There are currently more than 10,000 dams inCanada. A practical multiscale model and experimental data on erosion threshold and rate of erosion forrealistic geometries and materials will allow appropriate actions to be taken in case of core overtopping. It willalso help design new structures.