InterPore2026

Cellular automata–based modelling of pore microstructure and water retention in fine-grained soils

20 May 2026, 15:35

1h 30m

Poster Presentation (MS09) Pore-Scale Physics and Modeling Poster

Speaker

Arghya Das (Indian Institute of Technology Kanpur)

Description

Accurate estimation of the pore structure of fine-grained (clay mineral-rich) soils is a challenging task, as these soils exhibit complex pore morphologies, strong heterogeneity, and a non-granular fabric, distinguishing them from coarse-grained soils such as sand. Therefore, to study its hydraulic characteristics, such as permeability and water retention behaviour, we need an accurate and representative model of fine-grained soil microstructure. Few experimental methods exist (i.e., gas adsorption, MIP, or imaging with SEM and X-ray tomography), but each has its own significant limitations. As a result, no single technique provides a complete three-dimensional description of fine-grained soil pore structure.
In this work, a cellular automata-based framework is employed to generate three-dimensional, voxel-scale microstructures representative of fine-grained soils. The pore structure is represented on a regular three-dimensional grid, where each voxel evolves according to local neighbourhood rules and can exist in either a solid or pore state (binary values 1 and 0). The persistence or transformation of solid voxels depends on the number of neighbouring solid voxels, enabling controlled growth of a connected solid matrix and an interconnected pore space. To avoid artificial regularity during skeleton formation, two distinct neighbourhood definitions are applied alternately during the evolution process, promoting irregular macro-scale pore morphology. Additional micro-scale heterogeneity is introduced through localised stochastic pore generation within selected solid regions, resulting in embedded microporosity without disrupting the global connectivity of the pore network (typical figure).
The generated pore structure was quantitatively analysed using pore network extraction in terms of porosity, pore size distribution, and pore–throat connectivity. The extracted pore networks exhibit a broad connectivity distribution and non-spherical pore geometries, indicating a heterogeneous and non-granular pore structure characteristic of fine-grained soils. Finally, a numerical framework is presented for predicting the hysteretic soil-water retention curves (SWRCs) for fine-grained soils utilising the cellular automata-based pore network. The framework integrates two fundamental aspects of water retention behaviour, namely capillary forces, which control water retention in soil pores, and adsorptive forces, which govern water retention on soil mineral surfaces. The capillary contribution to the SWRC is modelled using the simulated network. Concurrently, the adsorptive contribution is assessed independently using the soil’s specific surface area to quantify adsorbed water on soil mineral surfaces. The simulated capillary and adsorbed water content at a particular matric suction are combined to derive the SWRC. The proposed model is tested on various soils with varying compositions of sand, silt, and clay, and the predicted SWRCs are in good agreement with the experimental data.