One of the most important characteristics of shale gas formation is that the majority of the natural gas in both organic and inorganic nanopores exists as adsorbed gas and, due to the pressure draw down, the desorbed gas flows into the adjacent fracture network from the shale matrix. Therefore, numerical investigations on the transport behavior of natural gas in the nanopores become increasingly crucial for better understanding the shale gas production performance due to the limitations of current experimental measurements.
In this paper, we applied boundary-driven non-equilibrium molecular dynamics (BD-NEMD) simulation to estimate natural gas transport diffusivity coefficients in both organic and inorganic nanopores. Type II kerogen molecules and montmorillonite clay molecules are used to model the nano-channel with various sizes and methane is used to model natural gas. Driven by an external force, high- and low-density gas regions have been formed and constant flow rate has been established as the system reaches steady-state. Pressure-dependent transport diffusivity coefficients of methane can be then determined based on 1st-order Fick's law. Due to the work done by the external force, the temperature of fluid will inevitably increase and introduce certain artifacts to the diffusivity coefficients. Therefore, fluid temperature must be carefully controlled during the simulation to mimic real isothermal flow experiments. Both convection and diffusion contribute to the gas transport in nano-channel. Results indicate that the molecular mean free path is smaller than the free gas region due to the nano-scale confinement effect. Transport diffusivity coefficients depend on Knudsen number as well as pore geometry. In the continuum regime (Kn < 0.1), transport diffusivity coefficients are mainly dominated by convection and independent on pore size. In the transition regime (0.1 < Kn < 10), gas transport can be estimated by Knudsen diffusion that transport diffusivity coefficients are size-dependent. Meanwhile, in the organic nanopore, similar correlation is observed with smaller characteristic Knudsen number. Deviations may be caused by the surface roughness.