Evaluation of different soil shear strength testing procedures taking inherent soil variability into account

Abstract

For several decades, geotechnical engineers relied on the Working Load Design approach, where the uncertainty and variability of soil were addressed using subjective global factors of safety. The Limit States Design concept was first introduced in the late 1990s and later adopted in South Africa for geotechnical engineering. This approach involves selecting characteristic values based on experience, cautious estimation, standard tables, and/or statistical methods. However, relying solely on past experience and standard tables may not be sufficient to ensure safety against failure. The most logical approach to estimating the value of a soil parameter seems to be the use of statistical methods. Ensuring a high degree of reliability and confidence in the selected characteristic value requires extensive testing. Most testing standards only require a small number of tests, typically three, to obtain an average or “representative” value. Due to cost and time constraints, it is often unfeasible to conduct a large number of tests on a particular soil. This reduces our understanding and knowledge of the inherent soil variability at hand. To address these challenges three shear strength testing procedures, the vane shear, (undisturbed) fall-cone, and pocket penetrometer, were evaluated to enable more accurate testing and decision-making for measuring inherently variable soils. This leads to more reliable data analysis and establishing a less conservative characteristic value for design inputs. Testing was conducted on cohesive soils across multiple sites in the Mangaung Metro Municipality. The vane shear test proved to be inadequate due to a low shear strength capability. The fall-cone yielded very low penetration readings and underestimated the shear strength compared to the pocket penetrometer. Using Hansbo’s equation to convert penetration data to shear strength values scatters the data and introduces higher variability with an increase coefficient of variation. In contrast, the pocket penetrometer produced more consistent results with lower variation in testing data. None of the test methods appeared to be an ideal fit for a lognormal distribution. To improve shear strength measurements on undisturbed soil samples, the study recommends using a larger cone factor and a modified fall-cone apparatus