Abstract
The cap model proposed by DiMaggio and Sandler and subsequently modified by Desai et al. involves only two stress invariants. An extension of this model is proposed herein by incorporating the third stress invariant into the formulation. The model requires only one additional constant which can be obtained from simple laboratory tests. An artificial soil is tested in a truly triaxial device and the stress-strain data obtained from this test program is used to verify the model. Failure surface as well as stress-strain predictions are compared with that of experimental observations. The agreement between predictions and observations are found to be excellent.
