Discrete geometric modeling of granular soils based on statistical percolative principles: A 2D implementation

This study proposes a new geometric modeling approach for a discrete representation of a realistic pore-scale morphology by using an amorphous, non-Euclidean, geometry of the shape of soil grains. The soil-skeleton composition process follows a consistent, statistically based, algorithm by implementing ideas derived from both fractal and percolation theories. Although the model yields random amorphous soil morphology, the global (geotechnical) scale of the obtained soil skeleton is consistent and well defined. The paper drives the fundamental principles of the model and suggests specific functions for practical use. An inverse analysis is also shown, in which the model parameters can be determined based on the specific grading curve characteristics that are required. The model is implemented in 2D but can be used as a general basis for 3D modeling.