Experimental and numerical study of the rod shear test for determining steel-sand interface behavior

Abstract

One laboratory technique to determine the unit side resistance of piles is the rod shear test. In a rod shear test a model pile is placed inside a cylindrical soil specimen. A confining pressure is then placed around the specimen/pile assembly and the model pile axially loaded. In some cases rod shear tests in sands have produced anomalously high interface friction angles (δ) which may be, at least in part, due to differences in how the normal stress on the model pile has been interpreted. Therefore, the usefulness of the rod shear test as a means to determine interface behavior for pile design is questionable. In this study, a new rod shear device was constructed to determine the behavior of steel-sand interfaces, and a numerical model was used to further evaluate the results. The interfaces tested were smooth and rough model piles in dense and loose sand. Results for the smooth pile tests are consistent with published values. However, tests with the dense and loose sands using rough piles produced interface friction values that were, in some instances, higher than the dilatant friction angle (φ(d)) for the sand at comparable stress levels. Elastic and elasto-plastic (modified Cam-clay) models failed to predict the high interface friction angles. Therefore it is suggested that the rod shear test imposes heretofore unrecognized kinematic restraints to volume change that, coupled with the discrete nature of the sand, lead to the development of a complex arching system The result is that effective normal stress on the pile is increased beyond the applied confining pressure. Evidence for such behavior included the observation of secondary features in the dense sand following pile displacement, the existence of force chains and work with discrete element systems showing that stress distribution in dense arrays is nonuniform.