Pore-Scale Characterization and PNM Simulations of Multiphase Flow in Carbonate Rocks

Abstract

This paper deals with pore-scale two-phase flow simulations in carbonate rock using the pore network method (PNM). This method was used to determine the rock and flow properties of three different rock samples, such as porosity, capillary pressure, absolute permeabilities, and oil–water relative permeabilities. The pore network method was further used to determine the properties of rock matrices, such as pore size distribution, topological structure, aspect ratio, pore throat shape factor, connected porosity, total porosity, and absolute permeability. The predicted simulation for the network-connected porosity, total porosity, and absolute permeability agree well with those measured experimentally when the image resolution is appropriate to resolve the relevant pore and throat sizes. This paper also explores the effect of the wettability and fraction of oil-wet pores on relative permeabilities, both in uniform and mixed wet systems.

Stacked ensemble machine learning for porosity and absolute permeability prediction of carbonate rock plugs

“…Moreover, the manual segmentation algorithms cannot be generalized to all samples 33 . On the other hand, automatic segmentation algorithms are less subjective, more efficient, and generalizable 47 , 48 . As a result, automated segmentation algorithms are more implemented in the DRP workflow 49 .…”

Influence of Heterogeneity on Carbonate Permeability Upscaling: A Renormalization Approach Coupled with the Pore Network Model

“…In the DNS method, the transport governing equations are solved for a segmented version of the 3D image using the lattice Boltzmann method (LBM), whereas a PNM involves idealized representation of the pore-scale rock geometry considering pores as spheres and throat as cylinders . Notably however, LBM is time-consuming and requires high computational power in particular when the simulations are conducted with large-size digital rock images, , whereas the PNM is computationally cost-effective and thus widely used to characterize the rock structure and predict the fluid flow behavior in porous medium. − Based on variable pore space configurations and sizes, different pore and throats generation algorithms have been proposed such as medial axis algorithm, , maximal ball algorithm, , and subnetwork of the oversegmented watershed (SNOW) algorithm to reflect the topological structure of the rock samples. The absolute permeability of the rock sample is thus estimated using flow simulation through the pore network model extracted from the segmented 3D images, in which the balls represent pore spaces and the sticks represent throats …”

“…The absolute permeability of the rock sample is thus estimated using flow simulation through the pore network model extracted from the segmented 3D images, in which the balls represent pore spaces and the sticks represent throats. 30 A score of publications applied DRP modeling to analyze the carbonate rock properties. 36−40 While the PNM approach has demonstrated good accuracy in predicting absolute rock permeability, 28,41−44 there are two key challenges in terms of the PNM approach.…”