Dual-scale pore network reconstruction of vugular carbonates using multi-scale imaging techniques

Vuggy carbonate rocks have a complex geometry with pore size ranges from nano to several millimeters (e.g., vugs). Compared to other rocks (e.g., sandstones), this multiscale nature makes pore-scale studies very challenging. Vugs have significant effects on rock properties (e.g., porosity, permeability, capillary pressure, etc.) by establishing high conductive pathways. Multiscale pore network models can capture pores across various length scales and can be used to predict the flow behavior of these porous systems. In this study, a dual-scale PNM is implemented to reconstruct the behavior of a vuggy carbonate sample. The rock sample is CT scanned at two different scales. At the macro-scale (i.e., vugular-network), a medical-CT scanner is used to image the rock sample at the resolution of 100 ism. For better visualization of macropores (i.e., vugs) of the rock sample, imaging is conducted at both dry and saturated states with an X-ray attenuating brine (i.e., NaI solution). The rock is also imaged by a micro-CT scanner at the resolution of 0.75 ism to extract the micro-scale properties (i.e., micro network). The images at both scales are preprocessed first. Then, the networks of both scales are extracted by two network extraction algorithms (i.e., the sub-network of the over-segmented watershed (SNOW) and the Maximal Ball (MB) method), and the results compared together. Then, a stochastically equivalent network based on the extracted micro-network properties is generated with a larger field of view (FOV). Then, vugs are randomly added to the reconstructed micro-network based on the properties of the macro-scale CT images. The result is a dual-scale unstructured irregular PNM. This modeling approach can efficiently preserve the vug-to-vug and vug-to-pore connectivity of overlapping vugs. The results show that the reconstructed dual-scale PNM has very close properties to the laboratory measurement data of the real rock sample.