Consequences of sample disturbance when predicting long-term settlements in soft clay

An approach for assessing the effects of sample quality is presented. Soil samples were taken using a 50 mm Swedish STII piston sampler and the Norwegian University of Science and Technology (NTNU) mini-block sampler from a soft clay test site. Differences in laboratory test results are identified for several stress paths, assisted by simulations made using an advanced constitutive model. Hitherto such comparisons have focused on differences in basic engineering properties such as strength and stiffness. The effect of choosing alternative model parameters from piston and block samples is demonstrated through the analysis of the long-term settlement of an embankment. The simulations show that substantially larger settlements and lateral displacements are predicted using parameters obtained from the piston samples. Furthermore, the magnitude of the differences is larger than expected. This demonstrates that for this application, relatively small differences in the assessed sample quality, using traditional laboratory data interpretation methods, are amplified when applied to a prototype boundary value problem. It is suggested that a little more care in sampling and testing can result in large cost savings as a result of the more reliable model parameters that can be extracted, particularly when the improved sampling is combined with the use of an advanced constitutive model.