Slope Stability Prediction Based On Mechanical Parameter Variability Characterization

The prediction of slope deformation and the evaluation of slope stability are the primary means and basic work to preventing catastrophic slope failure from occurring.Nonetheless, the variability, existing in parameters of rock and soil within slope in practice, inevitably leads to the uncertainty in the prediction of slope deformation and the evaluation of slope stability. The variability in parameters of rock and soil is mainly manifested as two aspects, which are the variation in space and the uncertainty in parameters. On one hand, the heterogeneity existing objectively in rock and soil results in the variation of parameters in space. On the other hand, our inability to characterize it makes it uncertain. With such a circumstance, the prediction is essentially probabilistic rather than determinate, and quantitative characterization of the variability in parameters of slope rock and soil can efficiently reduce the uncertainty in the prediction and assess the reliability of the prediction.To this end, this thesis focuses on the variability of mechanical parameters (i.e.,effective Young's modulus E', effective cohesion c' and effective internal friction angle ?'). Firstly, the cross-correlation between each mechanical parameter and horizontal displacement u_x in time and space is analyzed. Based on this and the principle of Successive Linear Estimator, the method for characterizing the variability of mechanical parameters of slopes is established. The elegant way for assimilating geological reconnaissance data into the characterization and the sampling strategy for slope horizontal displacements are proposed afterwards. Subsequently, an actual complex scenario with variably saturated conditions is introduced. Under this scenario,the effect of the variability of slope mechanical parameters on slope stability is discussed, the variability of slope mechanical parameters is characterized, and the prediction on slope deformation and stability is demonstrated. Thereafter, a feasible method for the assessment of prediction uncertainty and reliability is established based on probability theory, and the uncertainty and reliability existing in the prediction of slope deformation and stability is quantified. Finally, the probabilistic framework is proposed for slope stability prediction. The main conclusions and achievements are as follows:(1) The cross-correlation analysis between mechanical parameters and horizontal displacements u_x within a slope: u_x is negatively correlated with E' 's around the x-shaped conjugate plane on the side close to the slope surface, and is positively correlated with E' 's around the x-shaped conjugate plane on the side away from the slope surface, and is positively correlated with E' 'sat the portion horizontally to the side close to the slope surface, and is negatively correlated with E' 's at the portion horizontally to the side away from the slope surface. u_x is positively correlated with c' or ?' at the portion on the side close to the slope surface, and is negatively correlated with c' or ?' at the portion on the side away from the slope surface.(2) The characterization of the variability in slope mechanical parameters: the variability of E' generally can be characterized clearer than those of c' and .The characterization of the variability in c' and ?' is limited by the portion around the significant plastic zone, which makes the results as the indicator for the weak zone in the slope. the characterization resolution of ?' increases with the magnitude of the effective stress. With a reasonable design for the location and the amount of monitoring points and the sampling time, c' and ?' around the non-significant plastic zone and E' can be characterized clearly, and the corresponding uncertainty gets small.(3) Engineering geological reconnaissance data of slope contains large amounts of information about the material distribution and structural characteristics within the slope. These data can be elegantly assimilated into the characterization of the variability in slope mechanical parameters, with the proper way depends on the availability of the magnitude and the location of data.(4) The sampling strategy aiming at characterization for slope horizontal displacements: firstly, the location of monitoring points determines the characterization scope. Drillings and monitoring points should be placed at equal interval and be distributed over an area covering the region where our interest is rather than clustered together. Secondly, increasing the amount of monitoring points is useful but inefficient,with is costly. Therefore, the sampling scheme should be designed according to project level and requirements. In addition, exploring multi-type data fusion technology to improve the resolution for slope parameter variability characterization is highly required. Thirdly, chose horizontal displacement monitoring data coming from moments, when external stimuli (such as reservoir water level changes, rainfall, etc.)which are as different as possible, are taking effects from different directions, with different magnitudes and effects. This can improve the resolution of the characterization based on the same sampling scheme. Fourth, indicate the key areas which need further investigation in the light of the characterization of the parameter variability and the evaluated reliability of our prediction on slope stability. In addition,the patterns how the variability of slope mechanical parameters influences slope stability can be used to identify potential weak zone. These four strategies provide a new theoretical basis for the existing geological survey of slope engineering.(5) The effect of the variability of slope mechanical parameters on slope deformation and stability: Firstly, by directly changing the magnitude of each mechanical parameter at different locations within the slope. Secondly, via adjusting the stress distribution of the slope, which extends the influenced scope. Specifically,E' mainly controls the deformation at different portions within the slope, which makes the spatial deformation diverse. c' and ?' controls the strength or yield limit at different portions within the slope, which makes it possible that the yield damage occurs at any parts within the slope. The cross-correlation analysis between each mechanical parameter and horizontal displacement in time and space can be used to reveal the law how the variability of slope mechanical parameters influences slope horizontal displacement.(6) This thesis exploratively constructs a "variability characterization-prediction"framework. This framework can characterize the spatial variability of parameters within slopes in a high-resolution way based on available geological reconnaissance data, and reduce the uncertainty around the temporal and spatial prediction on slope deformation and slope failure. In addition, it can quantify the uncertainty existing in the prediction, identify the reliability of our prediction at different parts within the slope,and locate the key area where need further characterization in a feedback manner.Furthermore, it employs the laws how the variability of slope mechanical parameters influences slope stability to identify potential weak zone and grasp the evolution trend of slope stability, eventually achieve the reliable prediction of slope deformation and stability. Results indicate that the prediction based on the characterization of parameter variability on slope horizontal displacement and stability is accurate, which can delineate the differences and overall trends of slope spatial deformation. These results demonstrate the framework that prediction based on characterization of parameter variability is effective.