Variability and accuracy of shear strength measurements in soft soils

Abstract

Spatial variability and uncertainty regarding the subsoil play an important role in assessment and design of water defences, public infrastructure and houses. The objective of this research was to quantify the uncertainties of the shear strength of the subsoil due to spatial variability in horizontal direction and sample disturbance to optimize the characteristic lower bound of the shear strength and to improve the reliability of assessments of slope stability of dikes. Characteristic lower bound values of the soil parameters, in which the uncertainties are taken into account, are usually applied in assessment and design analyses. In this research spatial variability and sample disturbance are studied. These are two aspects which are important in the common Dutch approach for the assessment of slope stability of water defences in The Netherlands (WBI 2017; I&M, 2017), which uses cone penetration tests (CPTs) and empirical correlations between the cone penetration resistance of CPTs and the undrained shear strength obtained from laboratory tests. CPTs with correlations are used as it is thought that the in situ state of the soil (stress history) is important for the operational shear strength. As the in situ state is very variable CPTs are applied to determine the local in situ state and shear strength. It is common practice for infrastructure projects to perform CPTs with an interval of 100 m or more. For the assessment of slope stability of a dike the nearest CPT or the CPT with the lowest shear strength are used to estimate the undrained shear strength for the slope stability analysis of a cross section of a dike. In this research three series of 15 CPTs with a distance interval of 7 m are performed along the Achterwaterschap. The CPT logs give a detailed insight in the variability of the stratigraphy and cone penetration resistance on the scale of a potential slip failure (50-100 m). It is found that the stratigraphy and cone penetration resistance vary considerably. Calculated horizontal correlation lengths vary between 3 and 33 m, dependent on deposits and location. So the correlation length is not a characteristic of a depositional environment, but it is a stochastic variable. Measurement accuracy of the CPT cones and uncertainty about pore water pressure and soil unit weight also play a role in the analyses. Based on these findings it is concluded that one CPT is only an absolute arbitrary measure of the shear strength. The characteristic lower bound values of the in situ state of the soil and the undrained shear strength have to be derived from a series of CPTs. This series of CPTs can belong to a dike section for example. Sample disturbance is important in the WBI 2017 approach as the CPT correlations rely on the results of laboratory tests. The CPT correlations often show substantial uncertainty, and sample disturbance is thought to be a relevant aspect regarding to this uncertainty. Sample disturbance is studied with a comparison of the sensitivity and undrained shear strength from field vane tests with the sensitivity and undrained shear strength derived from intact samples. This is done for three sites: Waaldijk at Waardenburg, Lekdijk near Schalkwijk and Achterwaterschap. It is concluded that sample disturbance occurs for clayey soils with a soil unit weight of 16 to 19 kN/m3. These clayey soils are found below the dikes at the Waaldijk and Lekdijk sites. At these sites the sensitivity and undrained shear strength based on the laboratory tests is low compared to the results from the field vane tests. Sample disturbance is also determined from the recovery ratio of the samples and the volume changes of the samples in the laboratory due to extrusion of the samples from the sample tubes. At the Achterwaterschap site no effect of sample disturbance is found. When the effect of sample disturbance on the undrained shear strength is compared with the spatial variability of the undrained shear strength based on four CPTs close to the boreholes and the field vane tests, it is concluded that the spatial variability of the undrained shear strength is larger than the effect of sample disturbance. Therefore the relative low shear strength of the intact samples can be partly explained by spatial variability. As sample disturbance cannot be indicated as a very important factor for the uncertainty of CPT correlations, it is likely that various other aspects contribute to this uncertainty. Therefore it is recommended for the derivation of CPT correlations to improve the sampling techniques, to apply cross checks and to analyse the involved data carefully. Field vane tests can play an important role in these cross checks, however further study of the interpretation regarding the drainage conditions and the corrections for strain rate is needed.