Uncertainties in modelling undrained shear strength of clays using Critical State Soil Mechanics and SHANSEP

The determination of the undrained shear strength of clays relies upon the goodness of the available in-situ and laboratory tests. Often, limited soil investigation data is available, or the collected data may suffer of low quality associated with poor test execution or sampling operations. The use of reliable correlations can then play an important role in geotechnical design. In that perspective, it is vital to choose the most appropriate correlation models that are suitable with the local soil conditions and that are possibly characterized by low uncertainty. The SHANSEP empirical model describes the undrained shear strength of clays in terms of normalized properties, where the soil strength is defined by means of the overconsolidation ratio (OCR) and two material coefficients that require empirical calibration. The SHANSEP model can be further combined with analytical solutions based on Critical State Soil Mechanics (CSSM) in order to define the undrained shear strength as a function of two fundamental properties, such as the preconsolidation stress (or the OCR) and the friction angle at critical state. The paper deals with the uncertainties associated with modelling the undrained shear strength of clays using a hybrid CSSM-SHANSEP model. The performance of the model is assessed by comparing the predicted undrained shear strength to an existing multivariate database of field vane data points from Finland. For each data point, the friction angle is estimated indirectly from the plasticity index using a correlation in the literature, while the OCR is taken directly from the database. Bias and uncertainties of the CSSM-SHANSEP model associated with the multivariate database are evaluated. Finally, a sensitivity study on the model parameters is presented.