Consistent prediction of absolute permeability in carbonates without upscaling

We describe a study focused on the absolute permeability of reservoir carbonate rocks from the Middle East and involving comparison of experimental data and numerical estimates obtained by combining digital-rock and Lattice-Boltzmann Methods (LBM). The question of the "representativeness"of the site at which the simulation is performed is addressed as follows. First, a low-resolution, CT X-ray scan of the core plug is performed to identify regions of large porosity (millimeter-sized vugs, etc.). These regions are then avoided to postselect smaller sites (site volume ~ 1 mm) which are to be scanned at higher resolutions (voxel size < dominant pore-throat size of the core plug). A "representativeness"criterion based on an empirically-inspired "representativeness"measure (R-measure) is used to eliminate those sites for which R > b, where b is an upper bound (typically, b = 1). Essentially, the measure estimates how well the postselected sites capture the experimental porosity and the dominant pore-throat size of the core plug. This leads to a small set of sites for which the simulations are both computationally manageable and yield a reasonable estimate of the permeability: the experimental and predicted values differ by a factor of about 3 on average, which is a particularly significant result given the challenging heterogeneous pore space of carbonate samples. We believe the suggested methodology to be an adequate and practical way to circumvent upscaling.