Digital Rock Research Articles

A computationally efficient modeling of flow in complex porous media by coupling multiscale digital rock physics and deep learning: Improving the tradeoff between resolution and field-of-view

Digital rock physics is at the forefront of characterizing porous media, leveraging advanced tomographic imaging and numerical simulations to extract key rock properties like permeability. However, fully capturing the heterogeneity of natural rocks necessitates imaging increasingly larger sample volumes, presenting a significant challenge. Direct numerical simulations at these scales become either prohibitively expensive or computationally unfeasible due to limitations in resolution and field of view (FOV). This issue is particularly pronounced in carbonate rocks, known for their complex, multiscale pore structures, which exacerbate the resolution-FOV tradeoff. To address this, we introduce a machine learning strategy that merges multiscale imaging data from various resolutions with a 3D convolutional neural network (CNN) model. This approach is innovative in its ability to identify cross-scale correlations, thereby enabling the estimation of transport properties in larger volumes—properties that are difficult to simulate directly—using trainable proxies. The integration of multiscale imaging with deep learning allows for accurate permeability predictions at scales beyond those feasible with traditional direct simulation methods. By employing transfer learning across different scales during the training phase, our multiscale machine learning model achieves robust performance, with an R² exceeding 0.96 when evaluated on diverse lower-resolution domains with larger FOVs. Notably, this method significantly enhances computational efficiency, reducing the computational time by orders of magnitude. Originally developed for the intricate pore structures of carbonate rocks, our approach shows promise for application to a wide range of multiscale porous media, offering a viable solution to the longstanding tradeoff between imaging resolution and FOV in digital rock physics.

Stable diffusion for high-quality image reconstruction in digital rock analysis

Stable diffusion for high-quality image reconstruction in digital rock analysis

The effect of topographic density variations on the geoid and orthometric heights in Hong Kong

Utilizing the adopted average topographic density of 2670 kg/m3 in the reduction of gravity anomalies introduces errors attributed to topographic density variations, which consequently affect geoid modeling accuracy. Furthermore, the mean gravity along the plumbline within the topography in the definition of Helmert orthometric heights is computed approximately by applying the Poincaré-Prey gravity reduction where the topographic density variations are disregarded. The Helmert orthometric heights of benchmarks are then affected by errors. These errors could be random or systematic depending on the specific geological setting of the region where the leveling network is physically established and/or the geoid model is determined. An example of systematic errors in orthometric heights can be given for large regions characterized by sediment or volcanic deposits, the density of which is substantially lower than the adopted topographic density used in Helmert's definition of heights. The same applies to geoid modeling errors. In this study, we investigate these errors in the Hong Kong territory, where topographic density is about 20% lower than the density of 2670 kg/m3. We use the digital rock density model to estimate the effect of topographic density variations on the geoid and orthometric heights. Our results show that this effect on the geoid and Helmert orthometric heights reach maxima of about 2.1 and 0.5 cm, respectively. Both results provide clear evidence that rock density models are essential in physical geodesy applications involving gravimetric geoid modeling and orthometric height determination despite some criticism that could be raised regarding the reliability of these density models. However, in regions dominated by sedimentary and igneous rocks, the geological information is essential in these applications because topographic densities are substantially lower than the average density of 2670 kg/m3, thus introducing large systematic errors in geoid and orthometric heights.

Three-Water Differential Parallel Conductivity Saturation Model of Low-Permeability Tight Oil and Gas Reservoirs

Addressing the poor performance of existing logging saturation models in low-permeability tight sandstone reservoirs and the challenges in determining model parameters, this study investigates the pore structure and fluid occurrence state of such reservoirs through petrophysical experiments and digital rock visualization simulations. The aim is to uncover new insights into fluid occurrence state and electrical conduction properties and subsequently develop a low-permeability tight sandstone reservoir saturation model with easily determinable parameters. This model is suitable for practical oilfield exploration and development applications with high evaluation accuracy. The research findings reveal that such reservoirs comprise three types of formation water: strongly bound water, weakly bound water, and free water. These types are found in non-connected micropores, poorly connected mesopores where fluid flow occurs when the pressure differential exceeds the critical value, and well-connected macropores. Furthermore, the three types of formation water demonstrate variations in their electrical conduction contributions. By inversely solving rock electrical experiment data, it was determined that for a single sample, the overall cementation index is the highest, followed by the cementation index of pore throats containing strongly bound water, and the lowest for the pore throats with free water. Building on the aforementioned insights, this study develops a parallel electrical pore cementation index term, ϕm′, to account for the differences among the three types of water and introduces a parallel electrical saturation model suitable for logging evaluation of low-permeability tight oil and gas reservoirs. This model demonstrated positive application effects in the logging evaluation of low-permeability tight gas reservoirs in a specific basin in the Chinese offshore area, thereby confirming the advantages of its application.

Integrated Physical and Digital Chalk Relative Permeability Evaluation: A Case Study

The Valhall chalk field has produced more than 1 billion barrels of oil equivalents over the last 40 years, primarily from the homogeneous Tor Formation. The underlying Hod Formation is more heterogeneous and is less maturely developed. The extent of heterogeneity poses a challenge in the evaluation of multiphase fluid flow properties. The objective of the work was to use digital core analysis to generate early relative permeability data to leverage and compare with conventional physical steady-state relative permeability data. Accurate digital and physical description of capillary pressure and relative permeability in the high-porosity chalk is complicated by both low permeabilities and heterogeneity. The main challenge with chalk is that flow occurs in a nano-environment. Physically, the nano-environment translates to low permeability, difficult rock preparation, and extensive experimental time, especiallyfor steady-state flow experiments. Representative three-dimensional (3D) digital rocks were generated using a combination of X-ray computed tomography (CT) and focused ion beam-scanning electron microscopy (FIB-SEM) methods. The digital rocks were used to simulate two-phase flow and generate relative permeabilities and sensitivity to wettability. Physical steady-state and digital relative permeabilities on several core plugs and subsets are compared in this study, which discusses the advantages of performing both as part of the formation evaluation process. The physical and digital results compare reasonably well. The physical results provide a relative permeability anchor, and the digital results provide the leverage of early results, parametric sensitivities, and quality assurance. Hence, integration between digital and physical core analysis yields a robust understanding and input for uncertainty modeling.

Super-resolution reconstruction of 3D digital rocks by deep neural networks

Digital rock technology provides valuable insights into the pore structure and fluid flow properties of geoenergy resources. Artificial intelligence technology in vision and image processing, especially the image super-resolution, has great potential for digital rock reconstruction and resolution enhancement. However, the analyzed core samples are typically sandstones/carbonates in micro-scale resolutions and in two-dimensional (2D) space, whereas the shale rocks in nano-scale resolutions for unconventional resources or three-dimensional (3D) digital cores are rarely investigated. Additionally, previous studies primarily emphasized image quality from a computer vision perspective, with little consideration given to estimating physical properties of digital rocks using super-resolution techniques.This study presents a very deep super-resolution (VDSR) algorithm, specifically designed to generate high-resolution 3D digital rock images, for nano-scale shale matrix and micro-scale hydraulic fractures. We compare both image quality and permeability accuracy between the original high-resolution images and the super-resolution images reconstructed by the proposed method. The results reveal that the reconstructed images using the proposed method closely resemble the actual images, and effectively reduce errors in permeability computations. This study highlights the applicability of the proposed VDSR algorithm in establishing the detailed structures of 3D nano-scale shale matrix and hydraulic fractured rocks, thus advancing super-resolution techniques in digital core analysis for geoenergy resources development.

A parallel programming application of the A* algorithm in digital rock physics

Pore-scale analysis and characterization of reservoir rocks provide valuable information for the definition and management of underground hydrogen storage and CO2 storage or sequestration.This article presents an optimized implementation of the A* algorithm, the most popular pathfinding method in the presence of obstacles. In this context, the algorithm is applied to recognize the minimum length connected paths through each flow direction of 3D images of a porous medium representative of a reservoir rock. The identification of the main paths allows the characterization of the pore space and the calculation of fundamental petrophysical properties such as tortuosity and effective porosity, which can be used for permeability estimation. Compared to other algorithms available for pore-scale characterization, such as the pore centroid, A* provides a better approximation of the pore space available for the flow and, therefore, a reliable characterization of the petrophysical properties. On the other hand, path identification is significantly consuming in terms of time and memory. In this paper, an efficient and optimized implementation based on C++/OpenMP programming language is presented.The proposed implementation aims to the analysis of large-scale models profiting from parallelization, memory optimization, and enhanced managing of dead paths. Three test cases of increasing sizes are presented, to analyze the advantages and the disadvantages of the algorithm as the number of explored points increases. The 3D binary images analyzed are related to a synthetic domain (1 million voxels) and two actual sandstone samples (about 4 and 64 million voxels respectively). The code is validated against a Matlab serial implementation, showing a significant improvement in efficiency. Remarkable test cases of several millions of voxels were afforded, overcoming the memory and execution slowness issues. Moreover, the proposed implementation is suitable for large pore-scale models run in HPC environments.

Quantitative characterization of the carbonate rock microstructure considering topological features: a case study from the Gaoshiti–Moxi block of the Sichuan Basin

The precise characterization of the rock microstructure is crucial for predicting the physical characteristics, flow behavior, and mechanical properties of rocks. This is particularly important for carbonate rocks, which depict a complex microstructure with multimodal pore radius distribution and natural fractures. Here, topological features that are typically ignored are taken into account to quantify the carbonate microstructure. Carbonate samples used are obtained from the Gaoshiti–Moxi block of the Sichuan Basin, which showed remarkable potential for oil and gas. Specifically, nuclear magnetic resonance (NMR), X-ray micro-computed tomography (micro-CT), and mercury injection capillary pressure (MICP) techniques are performed to describe the topological and geometric characteristics. The results indicate that NMR and MICP techniques can describe more rock pores than micro-CT. However, due to the presence of pore shielding in MICP tests, the pore radius obtained by MICP is smaller than that obtained by micro-CT and NMR. Furthermore, the effective method used for characterizing the pore structure is NMR technology. The hardest part is that the coefficient between the pore radius and T2 relaxation time is difficult to calculate. Therefore, a better calculation method must be found. In addition, micro-CT is an irreplaceable technique for obtaining a large number of topological and geometric features, and multi-phase or single-phase flow simulations can be conducted via digital rock models. However, for carbonates, micro-CT is not sufficient to describe the complete pore systems because macropores cannot be fully represented and sub-resolution micropores cannot be described. Those macropores and micropores have a very important effect on their seepage properties. Therefore, multi-scale digital rock modeling involving small and large pores is essential for complex rocks, which is of great significance for the analysis of pore systems and the simulation of rock physical properties.

Microscopic seepage simulation of gas and water in shale pores and slits based on VOF

Abstract The microscopic pore-fracture structure and wettability have a significant influence on the two-phase seepage of shale gas and water. Due to the limitation of experimental conditions, the seepage patterns of gas and water in shale pores and slits under different wetting conditions have not been clarified yet. In this study, the three-dimensional digital rock models of shale inorganic pores, organic pores, and microfractures are established by focused ion beam-scanning electron microscopy scanning, and gas-driven water seepage simulation in shale microscopic pore-fracture structure under different wetting conditions is carried out based on volume of fluid method. The simulation results show that the gas–water relative permeability curves of microfractures are up-concave, and the gas–water relative permeability curves of inorganic and organic pores are up-convex; the gas–water two-phase percolation in microfractures is least affected by the change of wettability, the gas–water two-phase percolation in inorganic pores is most affected by the change of wettability, and the organic pores are in between; the gas–water two-phase percolation zone of microfractures is the largest, and the isotonic saturation is the highest; under the water-wet conditions, the critical gas saturation of microfractures, inorganic pores, and organic pores are 0.13, 0.315, and 0.34, respectively, and the critical gas saturation of organic pores under non-water-wet conditions is 0.525, indicating that under water-bearing conditions, the shale gas flow capacity in water-wet microfractures is the strongest, followed by water-wet inorganic pores, water-wet organic pores, and hydrophobic organic pores, respectively.

Modeling of flow and transport in multiscale digital rocks aided by grid coarsening of microporous domains

Many subsurface porous media such as soils, carbonate rocks, and mudstones possess multiscale porous structures that play an important role in regulating fluid flow and transport therein. A pore-network-continuum hybrid model is promising for numerical studies of a multiscale digital rock. It is, however, still prohibitive to the REV-size modeling because tens of millions of microporosity voxels may exist. In this work, we develop a novel and robust algorithm for coarsening microporosity voxels of a multiscale digital rock. Then, we combine coarsened microporosity grids with the pore network of resolved macropores to form efficient computational meshes. Furthermore, a pore-network-continuum simulator is developed to simulate flow and transport in both a synthesized multiscale digital rock and a realistic Estaillades carbonate rock. We show that the coarsening algorithm can reduce computational grids by about 90%, which substantially reduces computational costs. Meanwhile, coarsening microporosity has a minor impact on the predictions of absolute permeability, gas production curves, and breakthrough curves of solute transport. We illustrate the mechanisms of flow and transport in multiscale porous media induced by microporosity. Finally, the efficient hybrid model is used to predict the absolute permeability of an Estaillades digital rock. The numerical prediction matches well with the reported experimental data. We highlight the importance of characterizing mean pore-size distributions in microporosity for the prediction of rock permeability and local flow fields. The developed pore-network-continuum hybrid model aided by grid coarsening of microporosity serves as a useful numerical tool to study flow and transport in multiscale porous media.

Controllable image expansion of rock castings based on deep learning

Abstract Digital rock physics (DRP) offers an effective method of deriving elastic parameters from digital rock images, but its practical application is always limited to limited datasets. Recently, deep learning techniques have presented a promising avenue for generating more extensive and cost-effective samples. However, generating controllable samples according to user definition remains very difficult due to high dependence on sufficient datasets. To resolve this problem, a new network was proposed based on the UNet framework through image translation (UNet-IT) to expand rock castings by given porosity in relatively fewer datasets. Practical tests on carbonate rock images demonstrate that the proposed method can generate samples tailored to specific porosity requirements, which achieved a minimum porosity relative error of less than 1%. Compared with the unextended samples, the generated ones have completely different pore structures in terms of two-point probability, two-point cluster, and lineal path functions. Furthermore, the elastic parameters of the generated images obtained through the finite element method (FEM) and practical logging data matched well, with an average relative error of ∼9%. This indicates that the generated samples can be used as effective data to estimate fine rock physics templates and then improve inversion accuracy.

Stability Analysis of Soil and Rock Mixed Slope Based on Random Heterogeneous Structure

Due to the complexity in the heterogeneous internal structure and interactions between rocks and soil, the slide of soil–rock mixed slope is usually more complex than that of a homogeneous soil slope. This paper investigated the stability of soil–rock mixed slopes with finite element method (FEM) based on random heterogeneous structure. An image-aided approach was used to generate the 2-D and 3-D digital rocks to ensure the morphology of digital rocks was similar with the real rocks. The 2-D and 3-D soil–rock mixed slopes were then generated by placing the digital rocks into the soil matrix. The generated heterogeneous structures of soil–rock mixed slope were imported into ABAQUS for numerical analysis. The effect of rock content, spatial distributions, material properties, and rock–soil interface on the stability of soil–rock mixed slopes were analyzed. Results show that the stability factor of the soil–rock mixed slope increases with the increase of rock content. The rocks can play a certain degree of antislide effect in the slope. The uneven spatial distribution of rocks has effect on the overall stability of soil–rock mixed slope. This effect is more significant when the rock content is moderate. Rocks distributed in the middle layer of the slope may improve the overall antisliding performance of the slope. The stability factor decreases with the increase of rock density. While the effect of rock elastic modulus on stability of soil–rock mixed slope is relatively limited. The contact condition at the soil–rock interface has effect on the overall stability of soil–rock mixed slope. It is recommended to properly determine the interface properties for stability analysis of soil–rock mixed slope.

Micro‐Continuum Modeling: An Hybrid‐Scale Approach for Solving Coupled Processes in Porous Media

AbstractMicro‐continuum models are versatile and powerful approaches for simulating coupled processes in two‐scale porous systems. Initially oriented for modeling static single‐phase flow in microtomography images with sub‐voxel porosity, the concept has been extended over the years to multi‐phase flow, reactive transport, and poromechanics. This paper introduces an integrated micro‐continuum framework to model coupled processes in porous media. It reviews state‐of‐the‐art models and discusses applications in geosciences including Digital Rock Physics with sub‐voxel porosity, moving fluid‐solid interface at the pore‐scale due to geochemical reactions, fracture‐matrix interactions, and solid deformation. Finally, the paper discusses future developments in micro‐continuum models.

Acoustical-electrical models of tight rocks based on digital rock physics and double-porosity theory

The acoustical-electrical (AE) properties of reservoir rocks are affected by their microstructure (pores, microfractures, cracks and their geometry) and saturating fluids. To aid the interpretation, digital rock physics (DRP) is a useful technique to characterize this microstructure, as well as equivalent petrophysical models (EPM) to obtain joint AE properties. We perform thin-section analysis on rocks from tight-oil reservoirs, along with X-ray diffraction (XRD) and computed tomography (CT) to study rock lithology and minerals. We also perform porosity, permeability, ultrasound, and electrical conductivity experiments as a function of confining pressure to analyze pore structure. First, the 3D digital multicomponent cores are created based on a geology-driven multiphase segmentation workflow and images of the samples, and verified based on the porosity and minerals. Then, the AE properties and permeability are calculated by using numerical simulations (finite difference and finite volume methods). Next, the rigid and microcracked (soft) porosities are determined by using the Shapiro model to create the rock skeleton. We develop a joint EPM based on the effective medium AE and the Biot-Rayleigh equations with double porosity to describe the rock properties. The ultrasonic, sonic and seismic multiple data are used to compare and analyze the two approaches. The results show that the DRP techniques based on real cores are effective in characterizing the microstructure and the proposed EPM can describe the AE properties of real rocks, indicating a potential for quantitative characterization of reservoirs.

Homogenisation method based on energy conservation and independent of boundary conditions

The foundation of homogenisation methods rests on the postulate of Hill–Mandel, describing energy consistency throughout the transition of scales. The consideration of this principle is therefore crucial in the discipline of Digital Rock Physics which focuses on the upscaling of rock properties. For this reason, numerous studies have developed numerical schemes for porous media to enforce the Hill–Mandel condition to be respected. The most common method is to impose specific boundary conditions, such as periodic ones. However, these boundary conditions influence both the effective property and the size of the REV. The recent study of Thovert and Mourzenko (2020) has shown that most boundary conditions still result in the same intrinsic effective physical property if the averaging is applied outside the range of the boundary layer. From this discovery, it becomes logical to question the status of Hill–Mandel postulate in porous media when homogenising away from the boundary. In this contribution, we simulate Stokes flow through random packings of spheres and a range of rock microstructures. For each, we plot the evolution of the ratio micro- vs macro-scale of the energy of the fluid transport outside the boundary layer, for a growing subsample size of porous media. Here, we prove that we naturally find energy consistency across scales when reaching the size of the Representative Elementary Volume (REV), which is a known condition for rigorous upscaling. Furthermore, we show that this index for the energy consistency is a more accurate indicator of REV convergence since the mean value is already known to be unitary.

Exploring CO2 storage potential in Lithuanian deep saline aquifers using digital rock volumes: a machine learning guided approach

The increasing significance of carbon capture, utilization and storage (CCUS) as a climate mitigation strategy has underscored the importance of accurately evaluating subsurface reservoirs for CO2 sequestration. In this context, digital rock volumes, obtained through advanced imaging techniques such as micro-Xray computed tomography (MXCT), offer intricate insights into the porous and permeable structures of geological formations. This study presents a comprehensive methodology for assessing CO2 storage viability within Lithuanian deep saline aquifers, namely Syderiai and Vaskai, by utilizing petrophysical properties estimated from digital rock volumes of samples from analogous formations. It also demonstrates the potential of integrating advanced imaging techniques, machine learning, and numerical modeling for accurate assessment and effective management of subsurface CO2 storage.

Spatial Characterization of Wetting in Porous Media Using Local Lattice-Boltzmann Simulations

Wettability is one of the critical parameters affecting multiphase flow in porous media. The wettability is determined by the affinity of fluids to the rock surface, which varies due to factors such as mineral heterogeneity, roughness, ageing, and pore-space geometry. It is well known that wettability varies spatially in natural rocks, and it is still generally considered a constant parameter in pore-scale simulation studies. The accuracy of pore-scale simulation of multiphase flow in porous media is undermined by such inadequate wettability models. The advent of in situ visualization techniques, e.g. X-ray imaging and microtomography, enables us to characterize the spatial distribution of wetting more accurately. There are several approaches for such characterization. Most include the construction of a meshed surface of the interface surfaces in a segmented X-ray image and are known to have significant errors arising from insufficient resolution and surface-smoothing algorithms. This work presents a novel approach for spatial determination of wetting properties using local lattice-Boltzmann simulations. The scheme is computationally efficient as the segmented X-ray image is divided into subdomains before conducting the lattice-Boltzmann simulations, enabling fast simulations. To test the proposed method, it was applied to two synthetic cases with known wettability and three datasets of imaged fluid distributions. The wettability map was obtained for all samples using local lattice-Boltzmann calculations on trapped ganglia and optimization on surface affinity parameters. The results were quantitatively compared with a previously developed geometrical contact angle determination method. The two synthetic cases were used to validate the results of the developed workflow, as well as to compare the wettability results with the geometrical analysis method. It is shown that the developed workflow accurately characterizes the wetting state in the synthetic porous media with an acceptable uncertainty and is better to capture extreme wetting conditions. For the three datasets of imaged fluid distributions, our results show that the obtained contact angle distributions are consistent with the geometrical method. However, the obtained contact angle distributions tend to have a narrower span and are considered more realistic compared to the geometrical method. Finally, our results show the potential of the proposed scheme to efficiently obtain wettability maps of porous media using X-ray images of multiphase fluid distributions. The developed workflow can help for more accurate characterization of the wettability map in the porous media using limited experimental data, and hence more accurate digital rock analysis of multiphase flow in porous media.

Ultrahigh-Resolution Reconstruction of Shale Digital Rocks from FIB-SEM Images Using Deep Learning

Summary Accurate characterization of shale pore structures is of paramount importance in elucidating the distribution and migration mechanisms of fluids within shale rocks. However, the acquisition of high-resolution (HR) images of shale rocks is limited by the precision of the scanning equipment. Even with higher-precision devices, compromising the image field of view becomes inevitable, making it challenging to faithfully represent the actual conditions of shale. We propose a stepwise 3D super-resolution (SR) reconstruction method for shale digital rocks based on the widely used focused-ion-beam scanning electron microscope (FIB-SEM) technique. This method effectively addresses the issues of inconsistent horizontal and vertical resolutions as well as low 3D image resolution in FIB-SEM images. By adopting this approach, we significantly enhance image details and clarity, enabling successful observations of pores smaller than 10 nm within shale and laying a foundation for further pore-scale flow simulations. Furthermore, we extract the pore network model (PNM) from the SR reconstructed digital rock to analyze the pore size distribution, coordination number, and pore-throat ratio of shale samples from the Jiyang Depression. The results demonstrate a pore radius distribution in the range of 0 nm to 40 nm, which aligns with the results from nitrogen adsorption experiments. Notably, pores with radii smaller than 10 nm account for 50% of the total connected pores. The proportion of isolated pores in the SR reconstructed shale PNM is significantly reduced, with the coordination number mainly distributed between 1 and 4. The pore-throat ratio of shale ranges from 1 to 3, indicating a relatively uniform development of pores and throats. This study introduces a novel method for accurately characterizing the shale pore structure, which aids researchers in evaluating the pore size distribution and connectivity of shales.

Pore to Core Scale Characterization of Hydraulic Flow Units Using Petrophysical and Digital Rock Analyses

This article illustrates an integrated method for characterizing hydraulic flow units in subsurface reservoirs using multiscale images and petrophysical data analysis from well cores. This method is applied to 35 meters of cores drilled from a Late Jurassic Upper Jubayla Formation outcrop in Saudi Arabia, which is analogous to the lower section of the Arab‐D reservoir. Scanning electron microscope (SEM) imagery, thin‐section petrography, and X‐ray computed tomography (CT) images of whole cores and core plugs are used to visualize and characterize these carbonate rocks across multiple length scales. Core plug porosity and permeability data are used for petrophysical characterization and deriving hydraulic flow units (HFU). The multimodal pore types associated with the HFUs and their connectivity using thin section and SEM image analysis, combined with mercury injection porosimetry are used. Whole‐core CT textures and heterogeneity logs are correlated with HFU and digital image analysis of core plugs. Whole‐core CT data prove to be a highly valuable tool for bridging the scale gap between well data and reservoir models. This methodology can be applied to datasets from a large number of wells, especially when combined with machine learning tools, allowing improved characterization of complex carbonate reservoirs.

Multiscale fusion of tight sandstone digital rocks using attention-guided generative adversarial network

Tight sandstone exhibits heterogeneity due to its multiple mineral compositions and multiscale pore structures, presenting significant challenges for digital rock modeling. To capture the multiscale information inherent in heterogeneous rocks, we utilized an attention-guided generative adversarial network (AttentionGAN) to combine X-ray micro-computed tomography (micro-CT) and scanning electron microscope (SEM) images of tight sandstone, thereby producing large-scale, high-precision rock images. The reconstructed digital rock images comprehensively showcase the multiscale pore structures, including intergranular and clay micropores, as well as the distribution of minerals. Mineral content, two-point correlation functions, representative volume elements, and pore radius distributions were calculated for both the real and reconstructed images to confirm the efficacy of the proposed AttentionGAN. The results demonstrate that AttentionGAN is capable of generating high-precision rock images with a wide field of view. This offers a beneficial methodology for characterizing heterogeneous rocks and numerically simulating rock physics processes.