An analysis of the triaxial apparatus using a mixed boundary three-dimensional discrete element model

Abstract

The triaxial test is probably the most important fundamental laboratory test for geotechnical engineers. The objective of the study presented here was to gain insight into the micro-scale interactions experienced by particles during a triaxial test using the distinct element method (DEM). To achieve this objective, a novel DEM modelling environment to simulate triaxial tests on ideal granular materials was implemented, as described here. In this test environment, both the stress conditions and boundary conditions in the physical tests can be accurately reproduced. This approach was quantitatively validated by simulating a number of physical triaxial tests on specimens of steel spheres. A comparison between the circumferential periodic boundaries used here and the circumferential rigid boundaries used by other researchers emphasises the significance of maintaining a continuous particle-particle contact network orthogonal to the major principal stress direction.Results of an analysis of the micro-scale response in a triaxial simulation are analysed in detail, including the stresses, the distribution of contact forces, the evolution of fabric, the distribution of local strains, and the particle rotations. Significant non-uniformities in the stresses, strains and contact fabric were observed.

Micromechanics of granular material response during load reversals: Combined DEM and experimental study

“…This paper initially includes a discussion on the available approaches to validate DEM codes, prior to a description of a coupled experimental-DEM study. This study extends the earlier research of Cui et al (2007) by demonstrating that DEM codes can accurately capture the response of a granular material subject to non-monotonic loading. The experiments considered are strain controlled quasi -static triaxial tests including 2 pre-peak load reversals.…”

“…This imposes a significant restriction for 2D analyses, however in 3D along the central axis void spaces centered on the origin will alternate with particles centered on the origin to maintain try aix-symmetry. While this imposes as small geometrical constraint on the particles close to the specimen centre, the influence on the macro-scale response is not significant, as evidenced by the ability of the model to capture the specimen response in monotonic triaxial tests (Cui et al, 2007) as well as the load reversal tests presented here. A system of indexing has been developed in the code to differentiate between the "real balls", the images of the balls protruding from the o-a boundary, and the images of the balls protruding from the o-b boundary.…”

“…The laboratory test approach used in this study is also described by Cui et al (2007) for monotonic triaxial tests, however a brief description is included here for completeness. An ideal granular material, Grade 25 Chrome steel balls, was used in the physical tests as these spheres are fabricated with tight tolerances (according to the manufacturer, Thomson Precision Ball, the sphere diameter and sphericity is controlled to within 7.5 x 10 -4 mm during fabrication ), and so the particle geometry can be accurately replicated in the numerical model.…”

“…During the specimen generation stage of the analysis, where balls are introduced close to one of the periodic boundaries a check is introduced to ensure that overlap with balls along the other periodic boundary does not take place. As described by Cui et al (2007), contact is detected between particles close to one periodic boundary and particles along the other periodic boundary by multiplying the particle coordinates by an orthogonal rotation tensor. The rotation tensor is also used to rotate the contact force vector for application to the particles, where appropriate.…”

“…The earlier study of Cui et al (2007) considered monotonic tests, however in the current study the specimens were subject to two load reversals prior to the peak.…”