Seepage features and reliability analysis of deep overburden dam foundations considering the spatial variability in permeability

Abstract

Seepage stability is an important issue for dams constructed on overburden layers. However, less interest has been focused on the influence of the spatial variability of the overburden layers of dam foundations on seepage features and stability. These layers are too deep such that the cut-off walls cannot reach the rock base because of the limitations of the technology. We propose an efficient way to evaluate the effect of spatial variability on seepage features for different lengths of the cut-off walls, and we perform reliability analysis. By combining the Cholesky decomposition method with a single exponential autocorrelation function (SNX), we generate many random fields for probability analysis. We study the influence of spatial variability and the autocorrelation coefficient of hydraulic conductivity, together with different cut-off wall lengths and drainage prism configurations, on the typical maximum hydraulic gradient (i_t at the base of the cut-off wall and i_s at the downstream overburden) and downstream seepage flow (Q_e). The results show that the deterministic seepage calculation underestimate the critical hydraulic gradient, because soil material heterogeneity often makes the seepage channel more focused. The existence of a drainage prism enlarges the critical maximum hydraulic gradient at the bottom of the cut-off wall. A low risk of seepage failure is shown only at the deep cut-off wall configuration but is also greatly affected by spatial variability. This work improves the understanding of how spatial variability influences the seepage reliability of overburden dams and is critical for controlling the seepage safety of dams with cut-off walls built upon deep overburden foundations.