Estimation of Fracture Permeability from Aperture Distributions for Rough and Partially Cemented Fractures

Natural fractures in subsurface reservoirs are frequently partially cemented with mineral precipitates, and it is unclear if fracture permeability models developed for rough barren fractures are applicable for fractures where roughness originates from cement linings. Here, we use a digital rock physics workflow to quantify the error in fracture permeability predicted by these models for five digitally synthesized rough fractures and four fractures imaged using three-dimensional X-ray computed microtomography. Samples include a rough, artificially-induced barren fracture in sandstone, a cement-lined natural fracture in limestone sampled from outcrop, and two cement-bridged natural fractures in tight-gas sandstones sampled from reservoir core. The images are then processed, segmented, characterized to determine statistical moments of the aperture distribution, and used in lattice Boltzmann model flow simulations. We address complications in measuring aperture distributions from images when the fracture pore space morphology deviates from the typical theoretical description of rough fractures and evaluate three different methods of measuring local aperture. The alternative cubic law using the nominal mean aperture is found to overestimate fracture permeability by upwards of one to two orders of magnitude, while the fracture permeability models using statistical moments of the aperture distribution are far more accurate for both rough barren and partially cemented fractures. We also define an empirical description of the upper and lower bounds of fracture permeability estimates as a function of relative roughness that is applicable to both rough barren and partially cemented fractures.