Measurement of membrane penetration in triaxial specimen through digital image correlation

Membrane penetration during triaxial tests can degrade the accuracy of results which affects the volumetric strain in drained tests or the pore pressure in undrained tests, and this problem has attracted much attention in the research community. However, a major portion of the attention has been focused on sand, and studies on gravelly-sandy soils and sandy-gravelly soils have largely been neglected. In previous investigations, in order to reduce or eliminate the impact of membrane penetration on test results, most of existing methods may use unreasonable assumptions or introduce other unpredictable artifacts into triaxial tests. Thus, this paper describes a method that uses image processing based on the digital image correlation (DIC) technique to perform non-contact, global measurement of membrane penetration of the specimen during triaxial tests. The method allows the full-field axial and radial strain of a cylindrical specimen to be accurately measured for estimating the skeletal volumetric strain. The membrane penetration was determined by the difference between the total volumetric strain and the skeletal volumetric strain. In addition, this method verifies the basic assumption that the radial strains in all directions at the same height of the specimen are approximately equal. In this study, the test results indicate that membrane penetration ΔVm exhibits an exponential relation with the normalized effective confining pressure. As the content of coarse particles increases and the relative density decreases, the amount of membrane penetration increases, but the increasing trend is weakened with the increase of coarse particle content and the decrease of relative density. The specimen gradation has a critical effect on membrane penetration and the shape of the particles also affects membrane penetration. A comparison of the test results with the results of Nicholson's empirical formula shows an error on the part of the formula that increases with sample density.