Direct Modeling of Granular Materials

Abstract

Granular packing is one of the long-standing problems that have been studied for decades. Most of the current methods are trying to infer some statistical properties to describe and then regenerate a granular medium. These techniques, however, remain at the lower orders and cannot extract very complex information and, thus, they produce unrealistic models that are physically unlikely. The texture of the generated particles using these techniques are mostly different from the real sharpness of granular materials. In this paper, a new technique for direct modeling of granular materials (DGM) based on using high-order statistics is introduced that can extract more information and, finally generate very accurate models. This method, indeed, does not employ any particle descriptor and uses the existing 2D/3D images directly. The results indicate a very good agreement between the simulated and original data. The computational cost of this method is very low.

Additional details