Experimental study on variability in mechanical properties of a frozen sand as determined in triaxial compression tests

Large uncertainty exists in soil testing due to the randomness in sampling and system errors, especially in tests on frozen soils. In order to reduce the randomness and improve the test accuracy, the sample preparation method is improved to obtain more homogeneous samples. The standard Chinese sand is used as the soil. An environmental material test apparatus with three-point temperature control was used. Four temperatures and four confining pressures are used for the triaxial compression tests, which makes 16 combinations. For each combination, five repeat tests were carried out to examine the scattering of the mechanical properties of the frozen sand. It is found that under a certain confining pressure, the scattering increased with the increase in temperature and vice versa. Under a certain combination of temperature at ?0.5 °C and confining pressure of 10 MPa, the stress–strain curves are so different that similarity in the curves does not exist. This phenomenon is not scattering because sample randomness or system errors cannot explain it any more and is therefore called variability. It is attributed to pressure melting, pressure crush of soil particles as well as severe phase changes caused by small temperature variations. The difference in the test results is considered as an inherent feature of frozen soils. Strength and strain energy are used, along with temperature and confining pressure to analyze the cause of variability. This work shows that further work must be carried out to develop an optimum testing program in order to make a reasonable analysis for engineering constructions in which frozen soils are involved.