TO THE STATISTICAL ANALYSIS OF SOIL SHEAR TESTS RESULTS

Abstract

The result of any shear tests, which are the most popular laboratory methods for determination the strength characteristics of soils, is a set of values of the ultimate shear resistance and the corresponding normal pressure along the shear section. The theoretical background for the analysis of the soil shear tests results is the theory of probability and mathematical statistics. At the same time, we have questions about the determination of the calculation values of soil strength parameters. So, during the processing of shear tests results according to a consolidated-drained scheme for sandy soils the calculation values of soil specific cohesion are often negative, which is incorrect. The article considers an example of processing of shear tests data of sandy loam for various normal pressures, and analyzes the results of statistical processing of soil tests. The normative methods for probabilistic assessment of soil strength characteristics are based on using of normal distribution law of a random variable. It is assumed that the valuesof a random variable can vary from -¥ to +¥, while the soil strength parameters can vary in a limited range, i.e. accept only positive valuesfrom 0 to +¥. This circumstance does not matter for the majority of clayed soils, which strength parameters are far enough from zero. However, it has a significant effect on the result for small sought-for values, particularly, in assessing of specific cohesion for sandy soils. Therefore, for such cases, the article presents a variant of the statistical processing of the shear tests results with using the reduced normal distribution law of a random variable.