Uncertainty in calibration of hypoplastic model for clay attributed to limited number of laboratory tests.

The paper quantifies the uncertainty of material parameters of the hypoplastic model for clay resulting from the experimenter's choice about what laboratory data are utilised in the calibration process. The results of several 1D compression tests and undrained triaxial shear tests performed on twenty different fine-grained soils are considered in various combinations to yield a scattered set of material parameters for each soil. A hierarchical probabilistic model separating the variations and correlations of parameters within one soil from variations and correlations among different soils is formulated. The posterior distribution of the probabilistic model parameters is obtained by means of the Bayesian inference. The results allow for stochastic simulations on a structural level even when only point estimates of material parameters are available. • Material parameters of hypoplastic model for clay are determined for twenty soils. • Calibration uncertainty of the parameters is quantified for general fine-grained soil. • Covariance within one soil and between different soils is formulated separated. • Results are directly applicable in stochastic analysis with different clayey soils. [ABSTRACT FROM AUTHOR]