Fig. 1. Real digital rock sample with 100-cubed voxels (voxel size of 2.82 nm).
Fig. 2. Convergence processes of volumetric velocities, p_0=0.2 MPa.
Fig. 3. Computed permeabilities in a production pressure range.
Note that the expensive computational costs at high pressure conditions due to cell refinement can be avoided by using a single LBM simulation to calculate the intrinsic permeability, which can be used together with a single DSBGK data at low pressure (Kn>10) to calibrate the Klinkenberg correlation model as a data fitting formula. Then, the calibrated Klinkenberg correlation model is valid to predict the apparent permeability inside the whole production pressure range. With the help of this hybrid DSBGK-LBM simulations, our software can handle digital rock geometry with 1000-cubed (a billion) voxels by using about 1000 CPU cores for few days.