عنوان مقاله [English]
In present paper, a biaxial test on a granular soil sample is simulated by the discrete element method. An ambiguous subject in the modeling of this test is the effect of boundary conditions on the micro and macro behavior of the sample. In this context, four different types of boundary condition have been considered for the lateral boundaries across which the confining pressure maintains a constant. The first type of lateral boundary condition is rigid boundary that uses rigid plates for the imposition of stress. The second and third types of boundary condition are flexible boundary, in which a string of particles acts as a plastic membrane and external forces imposed onto the string's particles. The last type of boundary condition used in present paper is that the outer particles of the samples are considered as the boundary particles and the external load imposed on them. The obtained results from the analyses show that the rigid boundary condition provides higher shear strength, with respect to the flexible ones, and there is also no considerable softening in the stress-strain behavior. However, as the constraints are imposed onto the sample's particles by rigid plates, no shear band is detected. On the other hand, under all flexible boundary conditions, shear band formation is obvious, but they suffer from an instability problem. It has been observed that under the boundary condition wherein particles of the sample themselves act as the boundary, instability is inevitable. However, for boundary conditions in which strings of particles are used, orientations of external forces have a significant influence on the stability of the sample. When the external force acts normal to the deformed shape of the boundary, no considerable instability has been observed, but when the external force acts in a constant direction from the beginning to the end of the test, instability occurs. It has been also observed that at micro scale, the sparseness of force chain distribution results in the global instability of the sample.