Evaluation of regular-based pore networks for simulation of Newtonian two phase flow

Abstract

Pore network modelling is a well-established and powerful tool for investigating the pore-scale mechanisms in porous media. It can be used for simulation of vast areas involving fluid flow through porous media. The main constraint in this approach is to develop a network structure which is close enough to the real structure of the porous medium. Although recently proposed approaches can capture the real structure of a medium, acquiring a three-dimensional CT image or two-dimensional cross-section of the rock sample is quite expensive for many individual researchers. However, a pore network structure which adequately represents the porous medium and behaves similarly is sufficient for performing the simulations if the aim is to understand flow behaviour in porous media or to perform sensitivity analyses on fluid and rock properties, where. Early results of using regular-based networks in simulations were not promising. However, their inefficiency might be due to the over-simplified equations used in the simulations and ignorance of the presence of wetting layers. This study attempts to understand whether these networks can be a good representation of the porous medium when advanced equations are used in the calculations.To achieve this aim, a simple regular cubic network was distorted by removing some of the throats randomly. The descriptive parameters of the network were tuned to match the macroscopic properties of the real porous medium. It was then used for simulation of primary drainage and the imbibition process using a quasi-static method. The tuning and validation processes were repeated for water-wet Bentheimer and Berea sandstones and a sample of carbonate rock. This study showed that if advanced equations and concepts (e.g. the shape factor and wetting layers) are used in calculations, the results of using a simple network can be adequately representative of the complexities of a real rock structure. Although the network obtained by tuning is not unique, however, these regular-based networks can be used for sensitivity analysis of pore-scale mechanism, with acceptable results, when a CT image or cross-section image of the rock is not available.