Natural alluvial foundations are inherently heterogeneous. To enhance seepage safety, a cut-off wall is commonly embedded in a dam foundation. However, walls can also have stochastic defects. The dual uncertainties arising from soil heterogeneity and wall defects pose significant challenges for seepage safety evaluation. In this study, systematic numerical simulations were conducted on an internally unstable dam foundation based on a four-constituent mixture framework. Soil heterogeneity was characterized by stochastic initial hydraulic conductivity and initial fines content. An erosion model, specifically incorporating the influence of overburden pressure, was employed to quantify suffusion. A probabilistic assessment utilizing Monte Carlo simulations revealed that suffusion in heterogeneous fields could be more severe than that in homogeneous fields. Various combinations of stochastic soil properties and defect locations can result in substantial disparities in seepage and erosion fields. The mean values of the total flux and the fines eroded ratio are insensitive to the spatial variation length, while their deviations increase with increasing spatial variation length, leading to larger uncertainties in the leakage channel morphology. For highly heterogeneous alluvial foundations with large spatial variations, conventional seepage and suffusion analyses that rely on homogeneous assumptions may considerably underestimate the internal erosion risk.
Read full abstract