Study On The Variation Characteristics Of Soil Strength Parameters And Reasonable Sample Number In The Slope Reliability Analysis

In the slope stability analysis,the reliability design method based on probability theory and considering various uncertainties has entered the practical stage.This method mainly needs to make clear the uncertainty of geotechnical design parameters.According to the sources of uncertainty,it can be classified as the inherent variability of parameters from the rock and soil and the statistical uncertainty.Traditionally,based on the assumption that the soil space is homogeneous,on the basis of the stochastic variable model to describe the variability of soil parameters.But actually,due to the deposition process of soil and the stress history of different soil parameters,cause the variability in space.Due to the ignoring of the physical and mechanical properties differences in the soil spatial by the traditional probability statistics method,it is unable to meet the need to make objective analysis and evaluation the parameters of soil spatial variability.In the practical engineering,finite samples are obtained by reconnaissance sampling test fistly,then use statistical values under small samples to the parameters estimation of the true value effectively.But there are no specific standards for the selection of the sample numbers in order to reflect the overall value,thus the statistical uncertainty is affected by different sample numbers.Based on the survey data of railway engineering,5 clay dominated engineering geological units are selected,the point and spatial variability characteristics of the soil shear strength index c??are analysised by using probability statistics method and random field theory respectively.Differences between the two theoretical analysis results are discussed,and the necessary of considering the spatial autocorrelation of parameter c??is analysised,the associated features between the spatial variability of c?? and other physical and mechanical indexes are studied based on the grey correlation degree.The sampling test of practical engineering exploration is simulated by computer random sampling method,sample statistical value space of geotechnical parameters and the corresponding reliability index result space are constructed respectively,the influence of different sample sizes on reliability index dispersion is compared and analyzed,and the reasonable number of samples under different variation levels are discussed.The main achievements of this dissertation are as follows:(1)Based on the traditional probability statistics method and random field theory,the point variation and spatial variability coefficient of cohesive soil parametersc?? are calculated respectively.The results show that there is an obvious difference of the coefficient of point and spatial variability of the two parameters.Compared with the coefficient of point variation,the spatial variability coefficients of c?? can be reduced by 25%and 15%respectively.The spatial variance of soil is obtained by the point variance reduction method in the random field theory.The coefficient of point variation is transformed into the coefficient of spatial variability,therefore,the parameters variation characteristics are described effectively.(2)The grey correlation theory is used to discuss the effect of physical and mechanical parameters on the variation of shear strength such as IL a1-2 w e IP.It shows that the correlation between IL and spatial variation coefficient of c?? is the highest and structural index a1-2 and Water bearing index w has the next influence on variation coefficient.(3)For the discrete degree of the results of reliability analysis affected by the sample number and the soil parameter variability,a method is proposed by inversing the number of samples in order to control the spatial variability of the reliability analysis results and reasonable sampling numbers corresponding to different soil parameters variation level are discussed.It shows that when the coefficient of variation are 0.30,0.25,0.20,the sample numbers corresponding to a number of small sample parameter variation level are 6,9,15,and the sample numbers should be increased to 11,15,20 and 18,20,30 corresponding to medium and large variability levels respectively.