Statistical evaluation of cyclic strength of sand

A statistical model is proposed to estimate the cyclic shear strength of a saturated sand deposit. Large-scale shaking table test results are used to develop the basic model since they reproduce the in situ test conditions more closely than other laboratory tests. Due to the unavailability of large-scale shaking table test results for different soils and test conditions, cyclic triaxial test results are used to modify this basic model. The model is first developed to represent laboratory conditions. This laboratory relationship is then modified for in situ conditions. Factors affecting laboratory relationships that are considered in this study are methods of sample preparation, soil fabric, system compliance, mean grain size, multidirectionality of earthquake motions, and secondary factors. Factors affecting the in situ relationship are previous strain history, age, cementation, etc. It is observed that considerable uncertainty is expected in the estimation of the in situ cyclic shear strength. The uncertainty associated with the in situ relative density contributes significantly to the overall uncertainty. This type of uncertainty needs to be considered in any analysis where the cyclic shear strength of sand is an important parameter, e.g., liquefaction.