Silo design relies in the wall friction parameter for the correct choice of hopper, for a particular material. Usually, the wall friction angle is obtained from a real test on the actual material to be stored. This value is a conservative one, and it would not hint any relation to the particle characteristics from the sample. In design, this friction value is considered constant in time and in position all along the silo and hopper walls.

It is now known that the friction is not constant all over the wall surface. Before this friction distribution is obtained, the stress distribution has to be measured, which is not an easy task. At real silo scales, many attempts have been made to measure stresses over the wall. From using probe sensors to photoelasticity, no real experimental method has proven to give results that would, undoubtedly, allow the prediction of the material behavior under given circumstances. Full-scale experiments are expensive and difficult to conduct, and the outcome is sometimes not up to the effort expended. Recently, more use of numerical analysis tools, instead of this full-scale experimentation, such as FEA (Finite Element Analisys) and DEM (Discrete Element Modeling) has become a common trend. Using mathematical models that try to predict the material behavior under many different boundary and initial conditions, FEA analysis relies more on the continuum assumption, which sometimes is not adequate for particulate systems. A more adequate model for bulk materials is the so-called DEM, in which particles are created and their interaction under certain conditions is monitored using force laws between particles and Newton's Second Law to track particle movement.

In this project, a 2D DEM model of a system inside a Jenike cell, which sometimes is used to measure the friction angle, was created to observe the behavior of a particulate system and its interaction to the wall. The wall has grooves simulating the asperity of a surface. The study compares the behavior of the bulk material under small normal forces (in the friction angle vs. normal load curve), which produces bigger friction angles than at higher loads. The groove size was altered to observe the movement of the sample during a shear test, with all other parameters constant. Cohesion and particle-structure interaction was also studied in this model (bonding parameters changed as others were kept constant). The particle characteristics are incorporated into the model and the resultant bulk movement is compared to actual bulk tests. It is expected that in the near future, particle characteristics will be incorporated into a DEM model to study different systems for design purposes, instead of carrying out the equivalent bulk tests needed to obtain the same parameters.