Variability and Bias in Undrained Shear Strength from Different Sampling and Testing Methods

Abstract

Undrained shear strength is one of the most commonly used geotechnical design parameters for fine-grained soils. Disturbance produced during boring, sampling, and testing introduces an often unquantified and variable bias into undrained shear strength measurements, which in turn introduces bias and uncertainty into geotechnical design. Evaluations performed to quantify variability and bias in undrained shear strength determined using five different sampling and testing methods are described. The variability of undrained shear strength measurements differs considerably among sites, but is generally least for isotropically consolidated-undrained triaxial compression (CIUC) tests performed on high-quality specimens, slightly greater for unconsolidated-undrained triaxial compression (UU) tests performed on high-quality specimens, and greatest for unconfined compression (UC) tests performed on lesser-quality specimens and pocket penetrometer (PP) and handheld torvane (TV) tests. Quantitative bias determined for different sampling and testing methods is generally consistent with conventional wisdom regarding sampling and testing quality. The minimum achievable uncertainty for undrained shear strength is controlled by uncertain bias; coefficients of variation representing uncertainty are likely to be greater than 0.25 if sampling and testing methods that are inappropriate for a specific design application are used. Evaluations presented also suggest site-specific testing for shear strength parameters using stress history and normalized soil engineering properties (SHANSEP) techniques may reduce variability and uncertainty by approximately 10%–40% compared with using empirical estimates.