Kinetic Energy Method for Predicting Initiation of Backward Erosion in Earthen Dams and Levees

Current methods to evaluate the potential risk of earthen dam and levee failures by internal erosion fail to consider appropriate failure criteria for the initiation, continuation, and progression phases and are often based on subjective assessment. Six different internal erosion failure modes may occur within a dam or levee, its abutments, or the foundation that are triggered or facilitated by different mechanisms and therefore have different failure criteria. In non-cohesive soils, suffusion, backward erosion, heave, or concentrated leak erosion (scour) are possible, although the physical mechanisms driving the failure criterion for each of these are very different. Currently, no credible failure criteria exist for evaluation of the initiation of backward erosion. This article presents derivation of a specific failure criterion for initiation of backward erosion in non-cohesive soils using the critical kinetic energy (E-kcrit) of initiation. Laboratory experiments conducted with different soils showed that E-kcrit is affected by the physical characteristics of the soil, effective stress conditions, and angle of the seepage path. In internally stable, non-cohesive soils (e.g., uniform medium sands) with up to 6 percent non-plastic fines, backward erosion initiated when the E-kcrit of seepage exceeded 0.075 Joules. In non-cohesive soils with 10 percent non-plastic fines, which are also prone to suffusion, the E-kcrit for backward erosion is reduced by a factor of as much as 100. Moreover, in soil with 20 percent plastic fines, initiation of backward erosion did not occur. Concentrated leak erosion along hydraulic-induced fractures was the dominant process of internal erosion in soils with plastic fines. A new backward erosion assessment methodology based on the factor of safety is proposed for initiation of backward erosion that relates the laboratory-derived E-kcrit and the anticipated maximum actual kinetic energy, as measured in the field or estimated during design. The advantage of this method is that laboratory-derived E-kcrit can take into account the natural factors that affect initiation of backward erosion. Several examples are presented to demonstrate the application of the factor of safety methodology under typical field conditions.