Influence of specimen nonuniformity and end restraint conditions on drained triaxial compression test results in sand

Drained triaxial compression tests on sand are common in geotechnical engineering due to their application in determining strength and critical state properties, and calibration of constitutive models. However, questions around reliability of test results due to strain localization and specimen nonuniformity have sparked debate around the adequacy of various specimen preparation techniques and configuration of top cap and bottom pedestal, among other variables. A rigorously calibrated and validated three-dimensional finite element model using an advanced constitutive model, NorSand, was used to address some of these concerns. The results generally confirmed that common specimen preparation techniques, moist tamping, and (air or water) pluviation, if done properly, can reasonably estimate the properties of a perfectly uniform specimen with the same density. Unsurprisingly, it was also confirmed that the post-peak response of dense specimens is by far the most sensitive part of sand behaviour characterized by these tests. It was suggested that for two-to-one height-to-diameter ratio specimens, end platen lubrication does not necessarily improve test results and can cause concentration of localized deformations near the platens. A fixed cap appeared to enforce a better boundary condition than a free-to-rotate one by preventing cap rotation driven by the formation of a single failure plane.