Accurate estimation of soil shear strength parameters

Abstract

The cost and safety of geotechnical engineering are highly depending on the accuracy of soil shear strength parameters. There are three methods often used to estimate soil shear strength parameters, i.e., moment method, 3-sigma rule and linear regression method. In this study, the accuracy of these three methods is compared. Traditional linear regression method (LRM) can only offer the mean of shear strength parameters. Some engineers misuse the standard error of shear strength indexes as the standard deviations. Such misuse may highly underestimate the uncertainty and induce high risk to the geotechnical design. A modified LRM is proposed to determine both the mean and variance of shear strength parameters. The moment method, three-sigma rule and LRM are used to analyze the tri-axial test data in Xiaolangdi Hydraulic Project and three numerical shear strength tests. The results demonstrate that: 1) The modified LRM can offer the most accurate estimation to shear strength parameters; 2) A dimensionless formula is much preferred in LRM rather than a dimensional formula. The stress ratio formula is much better than stress relation in the shear strength parameter analysis. The proposed method is applicable to shear strength parameter analysis for tri-axial test data, direct shear test and the un-drained shear strength test of stratified clay.