| Traditionally,deterministic method was used to evaluate slope stability.However,as an evaluation index of deterministic analysis,stability coefficient can not be used to obtain reliable results of slope stability.To solve this problem,in the study of reliability,a twodimensional stationary random field model of slope is established considering the spatial variation characteristics of soil strength parameters.Random field theory is combined with elastic-plastic finite element method to analyze the random reliability of slope,and the influence of spatial variation characteristics of soil shear strength parameters on slope reliability is revealed.The research results have a certain reference value for the design and control of natural or loess excavation slope.Specific research contents and results are as follows:1.Collected a large amount of loess stratum data in the study area,and completed the mathematical statistics of soil physical and mechanical indexes by stratum age and working point.The results show that the coefficient of variation of cohesion is mostly above 0.35,and the maximum is 0.85.The coefficient of variation of the internal friction Angle will not exceed 0.3.The fitting test of c and φ values of large samples is carried out by K-S method using MATLAB programming.The results show that c values obey lognormal distribution and φ values obey normal distribution or Weibull distribution.The variation range of vertical autocorrelation distance of Malan loess is between 0.2m and 1m by using recursive space method.2.Firstly,commercial software GEO-Studio was used to conduct simplified probability analysis under the stochastic variable model,and then the program developed by Griffiths was used to conduct stochastic finite element analysis.Finally,based on the finite element analysis results,the difference of influence on slope reliability when considering spatial variability was studied.The numerical results show that:(1)Bounded by a certain value of the autocorrelation distance parameter Θ=0.25～0.5,when Θ≤0.25,the failure probability increases with the increase of the correlation distance under the same coefficient of variation,but the change law is different when Θ≥0.5.(2)There is a critical value of the influence of the variation coefficient of the strength parameter on the slope failure probability.For the loess strength parameter in the normal variation range,when the variation coefficient of the shear strength parameter is below the critical value,the simplified probability analysis method ignoring the spatial variability will overestimate the failure probability of the slope.Under the condition of high variation coefficient,when the variation coefficient of shear strength parameter exceeds the critical value,ignoring the spatial variation characteristics will lead to the underestimation of the failure probability.(3)The influence of stability coefficient on slope failure probability also has a critical value,and the increase of parameter variation coefficient will increase the critical value of stability coefficient.When the stability coefficient is less than the critical value,the failure probability obtained by the simplified probability analysis method ignoring the spatial variability is less,which leads to unreasonable evaluation of the safety of the slope.3.Due to the unique failure mechanism of stochastic finite element,the area with the weakest soil strength can be naturally obtained by local average of strength parameters,so as to determine the location of the sliding surface.Under different correlation distance and parameter variation coefficient,the deformation and failure forms of slope obtained by random finite element method are quite different.The deepest sliding surface of the slope may pass through the bottom of the slope,which is the foundation failure,or it may only be the surface failure. |
PDF Full Text Request