Study On Variability Of Soil Parameters And Sample Size In Reliability Analysis Of Slope

Two limitations are associated with the deterministic method used in stability analysis of slopes in practice. Firstly, too much focus is put on computing the factor of safety. The uncertainty and the variability in geotechnical engineering are ignored due to simply averaging relevant index and parameters. Secondly, sufficient understanding and effective treatment are lacking for the variability of parameters, especially the influence of sample size (n) on variation. In this dissertation, the analysis of uncertainty and reliability is investigated, the influence of variation and sample size on failure probability is studied systematically.GEO-SLOPE, a software of slope stability analysis, uses the Monte Carlo method with Limit Equilibrium Method, and considers the safety factor and failure probability of slopes. The safety factor combined with the failure probability (P_f) and the reliability index (β) increase the precision of the results.The study on uncertainty of soil parameter, the theory of random field and the recursion spatial method are introduced to determine the variability and the distance. The variability of parameters re-organized and analyzed. Then the influences of inherent variability, correlation distance are studied carefully.The paper uses a standard slope to carry out the analysis, the impact analysis in the safety factor, the index of reliability and the failure probability about the homogeneous and layered slope. It is shown that:1. As the coefficient of variability increases, the index of reliability decreases and the failure probability increases. When the coefficient of variability range from 0.05 to 0.15, the curve of 3 - 8 relation is steeper; which means the index of reliability is much more sensitive. And as 8 from 0.2 to 0.4, the index of reliability is less sensitive. It provides theoretical basis for a reasonable sample size and the application of reliability analysis in landfill which variability of the parameters is much larger.2. As the mean value and the coefficient of variability are identical of every layers, the more layers, which means the number of variables to consider more, thesmaller the probability of failure.3. Produce five different samples, using Matlab software, which are based on specifications, and study the relationship between sample size and the index of reliability and the failure probability, in order to find out the reasonable number of n, which is accuracy while cost smaller.4. Calculate the coefficient of variability, considering the slope with layered, it is bigger than the coefficient of heterogeneous slope. And the probability of failure is much smaller. The formation of slope n is a natural process, with such a close relationship to the layer composition and deposition era, therefore, will obtain the reasonable probability of slope failure.The compositions of MSW are complex and diverse, whose uncertainties are far beyond soil. The variability and the reliability index were calculated based on the example of Qizishan Landfill, Suzhou, China.. The stability of landfill, which introduces uncertainty analysis and reliability analysis, is an important area of development, and it can be used as a domestic landfill design reference.