A comparison of probabilistic distributions of undrained shear strength of soils in Nipigon River, Canada

Abstract

In probabilistic reliability analysis and design, critical geotechnical variables such as soil shear strength are usually regarded as random variables with a probability distribution rather than deterministic values or constants. In this paper, the vane shear test is briefly introduced and used to obtain undrained shear strength of soil in the area of Nipigon river landslide, Ontario, Canada. Then the maximum entropy method is presented to generate an unbiased probabilistic distribution for soil properties based on optimal-order moments from observed soil samples. A comparative study between maximum entropy distributions and traditional lognormal and normal distributions is conducted to evaluate the performance of fitted probabilistic distributions. Kolmogorov-Smirnov goodness of fit test shows that the maximum entropy distribution with four order moments fit the undrained shear strength best. The analytical entropy distribution obtained can be used in probabilistic reliability analysis.