Using nano-XRM and high-contrast imaging to inform micro-porosity permeability during Stokes-Brinkman single and two-phase flow simulations on micro-CT images

Abstract

Carbonate rocks have particularly complex and multiscale pore systems which are weakly understood. In this study we use combined experimental, modelling, and pore space generation methods to tackle the impact of micro-porosity on the bulk flow properties of Estaillades limestone. First, a nano-core from a microporous grain of Estaillades Limestone was scanned using x-ray nano tomography (nano-XRM). The information from the nano-XRM scan was then used as input into an object-based pore network generator, on which permeability fields were simulated for a range of porosities, creating a synthetic Kozeny-Carman porosity-permeability relationship targeted for the specific micro porous system present in Estaillades. We found a good match between experimental and simulated Mercury Intrusion Capillary Pressure (MICP) range in the imaged geometry and a good match between the imaged and object generated permeabilities and MICP. A micro-core of Estaillades was then scanned using x-ray microtomography (μCT), the differential pressure was measured during single phase flow, and the rock was flooded with highly doped brine to differentiate connected from unconnected micro-porosity. The differential contrast between the dry and doped images was used to assign a porosity to each voxel of connected micro-porosity. The flow through the pore space was then solved using a Stokes-Brinkman solver while a second segmented image with no micro-porosity was solved a Stokes solver. The differences between the measured permeability and the two computed permeabilities was evaluated. We found that there was good agreement between both the computed permeability of the Stokes and Stokes-Brinkman simulation with the measured permeability. However, there was considerable differences in the velocity fields with the Stokes-Brinkman simulation capturing stagnant regions of the pore space that were not present in the Stokes simulations. Additionally, we investigated the implications of including micro-porosity in estimations of relative permeability. Nitrogen was experimentally co-injected through the core with doped brine at a 50% fractional flow and imaged to the two-phase effective permeability. This experimental measurement was compared with the numerical permeability simulated using both Stokes and Stokes-Brinkman models for several saturation points along a synthetic MICP injection curve. We found that the Stokes simulation was not able to predict relative permeability with this method due to the major flow paths in the macro-porosity being impeded by the injected non-wetting phase. The Stokes-Brinkman simulations, however, allowed flow in the microporous regions around these blocked flow paths and was able to achieve a relative permeability prediction that was a reasonable match to the experimental measurement. This method could be used to predict relative permeability in water wet pore-structures with high micro-porosity.

Upscaling the porosity–permeability relationship of a microporous carbonate for Darcy-scale flow with machine learning

“…A choice few studies have attempted to bridge the gap between these scales 4 , 36 . Menke et al 32 applied correlative microscopy to incorporate nano-structural information of Estaillades microporosity imaging using nano-CT into pore-scale flow predictions using the Stokes–Brinkman equation. However, as of yet, no studies have integrated these structures into a single model to achieve accurate upscaling parameters that would allow information about the interactions between structures at multiple scales to inform Darcy-scale simulations.…”

Simulating permeability reduction by clay mineral nanopores in a tight sandstone by combining computer X-ray microtomography and focussed ion beam scanning electron microscopy imaging

“…The effective image resolution of the detector equipped with a CCD camera was 1.22 µm per pixel, while the image size was 3056 × 3056 pixels. We used an advanced reconstruction script described in Moosmann et al (2014) with the MATLAB ® software and binned the images by a factor of 2 before reconstruction to increase the signal-to-noise ratio. This decreased the voxel resolution to 2.43 µm and changed the image size to 1528 × 1528 pixels.…”

Permeability Prediction using multivariant structural regression

“…A significant drawback to traditional techniques for permeability estimation (like Kozeny-Carman) is the challenging and arbitrary nature of the measurement of their constituent parameters [42]. A significant advantage of multivariant regression-based techniques is that all inherent errors in measurement are included when performing the regression -the minimization of the prediction error implicitly includes the minimization of the impact of any systematic measurement errors.…”