Study On Slope Stability Reliability Based On The Finite Element Method

Because of the variability of soil parameters, it is reasonable to utilize reliability method based on probabilistic theory when conducting slope stability analysis. For the present, at home and abroad, the principle methods of slope stability reliability analysis are the first-order second-moment method, second-order second-moment method, response surface method, monte-carlo method, etc. However, when using the first-order two-moment method and the two-order two-moment method, the difficulty lies in obtaining the partial deviration of implicit performance function precisely. When using the response surface method, the difficulty is that the sample points, especially the test points, are needed to be fitted well to form the response surface function. And the monte-carlo method requires burdensome compulation. In response to the above problems, a slope stability reliability analysis method based on the finite element method is investigated in the paper. Moreover, based on the truncated distribution, the sensitivity analysis of the effect of soil parameters’ left-right truncation point on slope stability is conducted. The principle researches are as follows:(1) This study proposes a method to determine the slip surface and safety factor of slope according to the equivalent plastic strain distribution of slope and the limit equilibrium method. This method exhibits the advantages of the finite element and limit equilibrium methods for slope stability analysis. The result of this method is compared with that obtained using the limit equilibrium method to verify its precision. In the meanwhile, the reasonability of determining the critical slip surface of slope according to the just right penetrated plastic area is studied in the paper. Also, the effect of gravitational acceleration selection on slip surface position and safety factor of slope, and the effect of segment number on safety factor of slope are studied.(2) A new method of four order moments for reliability analysis of slope stability is proposed. The process is conducted by integrating uniform design method, the RBF neural network technique, and maximum entropy principle. Finally, the approximate expression of probability density function of slope safety factor, the slope failure probability, and corresponding reliability index are available. The result of this new method is compared with that obtained using Monte-Carlo method to verify its precision.(3) The sensitivity law of the effect of soil parameters’ left-right truncation point on slope stability is obtained in the paper. The slope reliability index corresponding to different truncation points of soil parameters is obtained using the four order moments method for reliability analysis of slope stability. Based on grey incidence analysis, the sensitivity analysis of soil cohesion and internal friction angle left-right truncation point is conducted. The result will serve as references for slope stability investigation based on truncated distribution.