Reservoir slope reliability analysis under water level drawdown considering spatial variability and degradation of soil properties

There are many factors that affect reservoir slope stability, including inherent spatial variability and degradation of soil properties induced by infiltration of water. To incorporate these two factors into reservoir slope reliability analysis, a probabilistic framework is developed combining an uncoupled finite element method (FEM) and limit equilibrium method (LEM) deterministic model, random field theory and Monte Carlo simulation (MCS). The proposed methodology is illustrated through a typical reservoir slope. The influences of spatial variabilities and degradation coefficients of soil properties on reservoir slope stability are fully studied. The simulation results show that the failure probability increases significantly as the scale of fluctuation lambda(v) increases from 1 m to 10 m (approximately the slope height) and varies slightly after lambda(v) is greater than 10 m under a given drawdown speed without considering the degradation of soil properties. P-f increases with the drawdown speed varying from 0.5 m/d to 3.0 m/d under a given lambda(v). The relative enhancement in P-f between 0.5 m/d and 3.0 m/d decreases sharply from a larger value (118.4 % in this study) to 29.5 % as lambda(v) ranges from 2 m to 10 m and varies slightly at 15.4 % when lambda(v) is greater than 10 m. As expected, P-f increases dramatically to 1.0 as the soil property (cohesion or internal friction angle) degraded to a specific value. A critical degraded value of soil property exists when the effect of spatial variability on P-f is studied. When the current degraded value is greater than the critical degraded value, P-f increases as lambda(v) increases, and vice versa. The critical degraded value varies with the drawdown speed. The larger the drawdown speed is, the greater the critical degraded value.