Permeability of porous materials is an important characteristic which is extensively used in various engineering disciplines. There are a number of issues that influence the permeability coefficient among which the porosity, size of particles, pore shape, tortuosity, and particle size distribution are of great importance. In this paper a C++ GPU code based on three-dimensional lattice Boltzmann method (LBM) has been developed and used for investigating the effects of the above mentioned factors on the permeability coefficient of granular materials. Multirelaxation time collision scheme of the LBM equations is used in the simulator, which is capable of modeling the exact position of the fluid-solid interface leading to viscosity-independent permeabilities and better computational stability due to separation of the relaxations of various kinetic models. GPU-CPU parallel processing has been employed to reduce the computational time associated with three-dimensional simulations. Soil samples have been prepared using the discrete element method. The obtained results have demonstrated the importance of employing the concept of effective porosity instead of total porosity in permeability relationships. The results also show that a threshold porosity exists below which the connectivity of the pores vanishes and the permeability of the soils reduces drastically.

• Received 27 October 2014
• Revised 15 February 2015

DOI:https://doi.org/10.1103/PhysRevE.91.053301