Numerical Modelling of Cracking in Embankment Dams

Concentrated leak erosion is a form of internal erosion and piping which occurs in embankment dams, levees, and their foundations. Internal erosion and piping is the cause of about half of large embankment dam failures. Most internal erosion and piping failures occur due to concentrated leak erosion which is initiated by a crack in an embankment or its foundation. This research is focused on the development of suitable numerical simulation procedures for predicting the potential location of cracks, crack width and depth in embankment dams. The sub-modelling technique is employed in the methodology in order to reduce the otherwise prohibitive computational time required to model crack growth. For this technique, two procedures, global modelling and sub-modelling are required. The global modelling is to analyse the development of potential cracking zones during and after construction. The global model is based on the use of commercial finite element software Abaqus. The parametric studies of the global model investigated the relative importance of various factors on the potential cracking zones. The sub-modelling is to analyse crack propagation in the potential cracking zones predicted by the global model. The sub-model is based on the use of the scaled boundary finite element method combined with the quadtree mesh. The framework of the linear elastic fracture mechanics is used to investigate the fracture behaviour of core soils. The maximum circumferential stress theory including the T-stress is suggested for the crack propagation criterion. Three case studies of Dartmouth Dam, Lake Buffalo Dam and Red Willow Dam have been performed to test the applicability of the developed simulation procedures in practice. The predicted location of cracks, crack depth and width show reasonable agreements with the field observations.A novel, simple and efficient one-point return mapping SBFEM formulation for the elastoplastic analysis has been developed. The accuracy and efficiency of the formulation are demonstrated using numerical benchmark examples. The developed formulation has been also applied to the image-based elastoplastic analysis. It demonstrates that the developed SBFEM formulation combined with the quadtree image decomposition provides an attractive way to perform virtual testing of elastoplastic responses of materials and structures