Experimental data have demonstrated that a strong relationship exists between the state parameter and the peak strength and dilatancy characteristics of sands. This paper proposes a way of reproducing this behaviour using a modified Mohr-Coulomb failure criterion, which retains its simplicity while improving substantially its modelling capabilities. The formulated constitutive model is calibrated for Nevada sand following a well-defined procedure and used in the prediction of four centrifuge tests investigating the behaviour of axially loaded footings. It is shown that the proposed model reproduces well both the element tests and the more
complex footing problems, demonstrating its usefulness for engineering practice. Moreover, a simplified version which does not require the definition of the Critical State Line is proposed for situations when this aspect of soil behaviour cannot be determined with confidence. It is shown that such simplification results in only slightly less accurate predictions than the full version of the model, while simulating aspects of soil response that cannot be reproduced using constant values for strength and dilatancy parameters.