Risk Assessment of Slope Accounting for the Cross-Correlation of Soil Properties

Abstract

The spatial variability of soil properties is inherent and apparent, and the cross-correlation of them is also exhibited. Research on risk of slope failure considering the cross-correlation of soil properties is not sufficient now. Accounting for this disadvantage, the cross-correlation random fields describing the spatial variability of soil properties are generated by using the covariance matrix decomposition method, and the procedure of the method is briefly introduced. Failure probability and slip mass are employed to quantify the risk of failure. A simple method programming method with the let-in Fish code in FLAC3D is proposed to determine slip mass. Taking a frictional/cohesive slope for example, a series of sensitivity analyses are performed through Monte Carlo simulation, and effects of the cross-correlation of soil shear parameters on probability, mechanism and risk of failure are investigated. The results indicate that the cross-correlation has a significant influence on the stability of slope. Ignoring the negative cross-correlation of shear strength parameters, both failure probability and failure risk will be overestimated. With the cross-correlation coefficient increasing from negative to positive, both failure probability and failure risk increase, and the critical slip surface becomes more scattered. Slip mass is more sensitive to the variation of cross-correlation coefficient than safety of factor. The paper also suggests that the influence of failure mechanism should be considered in risk analysis of slope. These conclusions can improve our understanding about the effect of the cross-correlation of soil properties on the slope stability, and provide reference for risk assessment of slope failure.