True triaxial tests were conducted on artificially frozen sand. The effects of the intermediate principal stress coefficient, temperature and confining pressure on the strength of frozen sand were studied. The stress-strain curves under different initial conditions indicated a strain hardening. In response to increases of either the intermediate principal stress coefficient or the confining pressure or to a decrease of temperature, the strength typically increased. Furthermore, a new strength criterion was proposed to describe the strength of artificially frozen sand under a constant b-value stress path, combining the strength function in the p-q and pi planes. Considering the low confining pressure, the strength criterion in the p-q plane fitted the linear relationship in the parabolic strength criterion well. The strength criterion in the pi plane was combined with stress invariants, and a new strength criterion was established. This criterion considers unequal tension and compression strength, and integrates temperature. Test results indicated its validity. All parameters of the strength criterion could be easily determined from the triaxial compression and triaxial tensile tests.

A comprehensive three-dimensional elastoplastic constitutive model is presented to characterize the stress-strain behavior of cement stone under the true triaxial stress state. This constitutive model incorporates a threedimensional yield function and a three-dimensional potential function. The three-dimensional yield function is designed to accurately represent the true triaxial stress state during hardening. The three-dimensional potential function is devised to depict the plastic flow direction under true triaxial stress state. The yield and potential functions include parameters that control the shape of the deviatoric and meridian planes, and these parameters vary with the plastic internal variable. Consequently, the yield function can accurately describe the stress state, and the potential function can precisely capture the variations in plastic flow direction. Additionally, a detailed procedure for determining the parameters of the yield function and potential function is proposed based on the full deformation process. The constitutive model is presented in the form of analytical solution. The comparison of experimental data with the constitutive model confirms its accuracy and validity. A sensitivity analysis of the deviatoric and meridian parameters in the potential function is performed, shedding light on their impact on the model behavior. Furthermore, the significance of incorporating Lode angle dependence into the potential function is discussed, emphasizing its essential role in accurately capturing strain in the direction of the intermediate principal stress.