Effect of 2-D spatial variability of soil properties on cofferdam stability considering degradation of soft soil foundation

The inherent spatial variability and degradation of soil properties in soft soil foundation may have a significant effect on cofferdam stability. To incorporate these factors into cofferdam stability and reliability analysis, this paper presents a framework combining uncoupled cofferdam stability analysis with a calibrated random forest surrogate model and Monte Carlo Simulation (MCS). A typical cofferdam model is used to illustrate the proposed methodology. The results show that the probabilistic critical slip surface of cofferdam considering the spatial variability of soil properties significantly shrinks as compared to that from deterministic cofferdam stability analysis. When the vertical (λ_v) and horizontal (λ_h) fluctuation scales are 2m and 20m, respectively, the percentage range of critical cumulative density functions is 62.83%-71.41% as speed varying from 0.2m/d to 2.0m/d, suggesting that the deterministic cofferdam stability analysis may overestimate the minimum safety factor ( \({FS}_{\text{min}}^{v}\) ${\mathit{FS}}_{\text{min}}^v$ ) value. The percentage range of critical cumulative density functions decreases considerably as spatial variability of soil properties grows insignificant. The failure probability P_f of cofferdam downsteam increases as spaial variability of soil properties vanishes. The vertical spatial variability poses more significant effect on P_f than horizontal spatial variability. It is worthy to note that the effect of spatial variability of soil properties on P_f of cofferdam downstream would be complicated by the existence of degradation on soil properties, especially when the degradation coefficient (D_i) decreases to a critical value.