Pore-Scale Gas–Water Two-Phase Flow Mechanisms for Underground Hydrogen Storage: A Mini Review of Theory, Experiment, and Simulation

Abstract

In recent years, underground hydrogen storage (UHS) has become a hot topic in the field of deep energy storage. Green hydrogen, produced using surplus electricity during peak production, can be injected and stored in underground reservoirs and extracted during periods of high demand. A profound understanding of the mechanisms of the gas–water two-phase flow at the pore scale is of great significance for evaluating the sealing integrity of UHS reservoirs and optimizing injection, as well as the storage space. The pore structure of rocks, as the storage space and flow channels for fluids, has a significant impact on fluid injection, production, and storage processes. This paper systematically summarizes the methods for characterizing the micro-pore structure of reservoir rocks. The applicability of different techniques was evaluated and compared. A detailed comparative analysis was made of the advantages and disadvantages of various numerical simulation methods in tracking two-phase flow interfaces, along with an assessment of their suitability. Subsequently, the microscopic visualization seepage experimental techniques, including microfluidics, NMR-based, and CT scanning-based methods, were reviewed and discussed in terms of the microscopic dynamic mechanisms of complex fluid transport behaviors. Due to the high resolution, non-contact, and non-destructive, as well as the scalable in situ high-temperature and high-pressure experimental conditions, CT scanning-based visualization technology has received increasing attention. The research presented in this paper can provide theoretical guidance for further understanding the characterization of the micro-pore structure of reservoir rocks and the mechanisms of two-phase flow at the pore scale.

1. Introduction

In response to the significant challenges posed by global climate change, China has taken a leading role in proposing the “3060 dual carbon” targets. To meet these objectives, a transition from a fossil fuel-based energy system to one primarily reliant on renewable energy sources is essential [1]. However, renewable energy sources such as wind, solar, and tidal power suffer from inherent intermittency, leading to periods of both surplus and shortfall in energy production. This cyclic variability poses challenges to the stable operation of the power grid. Energy storage, especially underground storage, can improve the grid management and guarantee energy security by balancing supply and demand [2]. A promising solution proposed by researchers involves using surplus renewable energy to electrolyze water, producing hydrogen, which can then be stored in depleted oil and gas reservoirs, aquifers, or other suitable underground geological formations, as illustrated in Figure 1. From a geological standpoint, underground storage is particularly suitable for hydrogen due to the capacity of such formations to securely contain gases under pressure and their natural sealing properties. Moreover, Underground hydrogen storage (UHS) is a technology that utilizes underground geological structures (such as salt caverns, depleted gas fields and gas reservoirs, and aquifers) for large-scale hydrogen storage, providing an effective solution for large-scale, long-term energy storage [3].

Currently, UHS facilities in China are not yet widely deployed, and most related research remains theoretical. There are several technical challenges that still need to be addressed, particularly concerning the unstable displacement and two-phase seepage processes during hydrogen injection and extraction. To enable efficient UHS, it is essential to ensure both adequate storage capacity and sealing integrity. This requires reservoirs to possess high porosity to accommodate the necessary gas volumes and sufficient permeability to facilitate effective gas injection and extraction at adequate rates [4]. It has been proven that geological formations suitable for UHS include salt caverns, aquifers, and depleted oil and gas reservoirs. Salt caverns are currently the most developed storage option due to their low cover permeability and stable mechanical properties, though they are prone to significant creep and volumetric contraction over long-term operation [5]. Aquifers, which are abundant in sedimentary basins, are porous and permeable layers of underground rock filled with either fresh or saline water. However, aquifers present challenges in terms of exploration and site selection [6]. In contrast, depleted oil and gas reservoirs offer large storage volumes, widespread distribution across China, and natural trap structures with well-developed reservoir-cap rock combinations [7]. Compared to other geological formations, depleted oil and gas reservoirs offer distinct advantages and considerable development potential. There has been extensive research and practical experience both domestically and internationally in the storage of these reservoirs, with several global CO₂ storage projects underway to reduce greenhouse gas emissions and foster the development of related technologies. Australia’s Gorgon project, one of the world’s largest CO₂ sequestration initiatives, mitigates emissions from liquefied natural gas (LNG) production by injecting CO₂ into underground salt caverns [8], and similarly, in Ordos, where CO₂ is injected into underground formations for long-term storage. This initiative is aimed at providing emission reduction solutions to support the clean utilization of coal in the region [9]. The key to successful UHS lies in effective “injection, extraction, and sealing” [10,11]. To better understand the UHS process, extensive research has been conducted that mainly focuses on three primary types of storage reservoirs: depleted oil and gas reservoirs, salt caverns, and aquifers [12]. As shown in

Table 1. Articles about UHS and its main concerns.

Citation	Main Concerns	Research Directions
Rana et al., 2024 [13]; Leng et al., 2025 [11]	Suitable reservoir types	Screening and evaluation of geological structures for hydrogen storage
Zeng et al., 2023 [14]; Bensing et al., 2022 [15]	Geological requirements
Thiyagarajan et al., 2022 [16]; Bhadariya et al., 2024 [17]	Microbial activity	Interactions between hydrogen and subsurface environments
Kalam et al., 2023 [18]; Gbadamosi et al., 2023 [19]	Rock–fluid reactions
Juez-Larre et al., 2023 [20]; Lubon et al., 2023 [21]	Injection and withdrawal processes	Engineering technologies and operational optimization
Song et al., 2024 [22]; Alfarge et al., 2025 [23]	Safety and risk management

, this table illustrates the research conducted over the years on key issues related to underground hydrogen storage.

In this paper, four topics closely related to UHS, including pore-structure characterization, mathematical models, pore-scale numerical simulation approaches, and microscopic visualization seepage experiments for gas–water two-phase flow, were reviewed and analyzed in an integrated way. The work presented in this paper wishes to further enhance the understanding of the two-phase flow mechanisms at the pore-scale, as well as to offer valuable information and crucial guidance for understanding and evaluating site selection and operation optimization for UHS.

2. Acquisition and Characterization of Pore Structure

Characterizing the pore structure of rocks is essential for understanding the complex network of pores and throats, including their geometry, size, spatial distribution, and interconnectivity. This detailed characterization is critical for predicting and analyzing the transport behaviors, such as gas–water two-phase flow within the rock’s pores.

The pore size distribution (PSD) obtained through rock pore structure characterization plays a critical role in determining the potential for hydrogen storage or sequestration. Different types of rocks, such as sandstone and carbonate rocks, exhibit distinct pore structures that influence the behavior of hydrogen within these reservoirs. Sandstone typically has a more uniform pore structure with higher porosity, which allows hydrogen molecules to be stored and transported efficiently. Hydrogen primarily occupies medium and larger pores, minimizing hydrogen trapping. Its well-connected pore network facilitates the flow and distribution of hydrogen, making sandstone an ideal medium for underground hydrogen storage [24]. On the other hand, carbonate rocks, such as limestone, often have more heterogeneous pore structures, consisting of a mixture of macropores and micropores, with complex and irregular pore geometries. While these rocks may have lower porosity, they possess high permeability, which helps trap hydrogen gas within their pores. Additionally, the presence of natural fractures and vugs in carbonate rocks further enhances their ability to retain hydrogen, making them an ideal choice for hydrogen sequestration. The PSD of these rocks influences the distribution and retention of hydrogen, which in turn affects storage efficiency and the long-term stability of hydrogen reservoirs. Therefore, understanding the PSD of different rock types is crucial for optimizing hydrogen storage and ensuring safe and effective sequestration strategies [25].

In recent years, with the development of experimental testing equipment and analytical techniques, laboratory methods for obtaining and characterizing pore structures have become increasingly advanced and diverse, forming an integrated analysis and testing system that includes qualitative description, semi-quantitative analysis, and quantitative characterization of pore structure features. Based on differences in testing methods and theories, existing testing technologies can be divided into two main categories: fluid invasion methods and imaging methods. Fluid invasion methods primarily include mercury intrusion porosimetry (MIP), nitrogen adsorption, and nuclear magnetic resonance (NMR). While NMR is often associated with fluid invasion techniques, it is important to note that NMR can also be used for solid-state measurements, making it versatile in characterizing both fluid-filled and solid pore spaces. On the other hand, imaging methods mainly include techniques such as thin section casting, scanning electron microscopy (SEM), and CT imaging, which use optical radiation technologies to capture images of the pore structure. These images, combined with digital image analysis, enable the quantitative characterization of pore structures, as shown in Figure 2.

2.1. Mercury Intrusion Porosimetry

Mercury intrusion porosimetry is a technique for characterizing pore structure by injecting a non-wetting phase fluid (mercury) into a porous medium under applied pressure, overcoming surface tension. The mercury intrusion method characterizes pore structure features primarily by recording the relationship between the intrusion pressure and the volume of mercury injected, which in turn reflects the characteristics of the pore capillary pressure curve [28]. This method can be used to calculate a series of parameters related to the pore throat distribution and size, such as displacement pressure, median saturation pressure, and mercury injection/withdrawal efficiency, as well as indicators that measure the spatial heterogeneity of the pore structure, such as skewness, mean coefficient, and sorting coefficient. Based on different techniques, the mercury intrusion method can be further divided into high-pressure mercury intrusion [29] and constant-speed mercury intrusion [30].

This method can calculate parameters such as the size of the core’s pore throats and specific surface area. However, it does not allow for direct observation of the sample, nor can it detect isolated pores. Furthermore, once used, the cores cannot be reused.

2.2. Nitrogen Adsorption

The nitrogen adsorption method is a pore structure characterization technique for porous media based on the Langmuir monolayer adsorption theory [31]. Scholars such as Brunauer, Emmett, and Teller extended the monolayer adsorption theory to develop the multilayer adsorption theory, also known as the BET theory [32]. This theory has become the most commonly used theoretical foundation in gas adsorption characterization methods for porous media. It has been widely applied in the study and testing of particle surface adsorption properties, such as specific surface area, pore volume, and pore size distribution.

The nitrogen adsorption method is primarily used to obtain data on specific surface area, pore volume, and pore size distribution, particularly for characterizing micropores. This method provides detailed information about the pore structure; however, it requires strict sample preparation and relatively long experimental times, as it may take a longer period to reach adsorption equilibrium, leading to poorer reproducibility of the experiments. Additionally, nitrogen adsorption at low temperatures can be influenced by the chemical properties of the sample surface, necessitating thorough preliminary analysis. Moreover, compared to nitrogen, hydrogen molecules are smaller, allowing hydrogen to enter smaller micropores. Therefore, during pore characterization, hydrogen adsorption can provide more detailed information about micropore and ultra-micropore distribution than nitrogen, making it especially suitable for studying small pores.

2.3. Nuclear Magnetic Resonance

Nuclear magnetic resonance (NMR) is a physical process in which atomic nuclei with non-zero magnetic moments undergo spin energy level Zeeman splitting under the influence of an external magnetic field and resonate by absorbing radiofrequency radiation at a specific frequency [33]. In 1956, Brown and Fatt discovered that the NMR relaxation times of fluids in a free state and those confined in rock pores differ significantly, with the former being much longer than the latter. This finding indicated that the pore size in which the fluid is present has a significant impact on its relaxation time. For rock samples, since the solid components of the sample generally do not contain hydrogen atoms, and the magnetic sensitivity of other common elements in the pores is negligible compared to that of hydrogen nuclei, the hydrogen nuclei in the pore fluids of rock cores are the main subjects in NMR testing [34].

Currently, the nuclear magnetic resonance (NMR) technique used for pore structure characterization generally refers to low-field NMR analysis conducted under magnetic field strengths below 1T. It provides parameters that reflect the structural characteristics of a sample by measuring different relaxation times. For rock materials, a large amount of experimental and theoretical research has shown that when the surface characteristics of the solid and the properties of the fluid are similar, the differences in relaxation time T2 reflect the differences in the pore structure of the rock sample (pore size) [35]. In general, the larger the pore size, the greater the number of hydrogen nuclei present, resulting in slower decay of the detected NMR signal, which manifests as longer relaxation times on the T2 spectrum.

This method allows for the precise characterization of the full-size pore structure in a non-destructive manner. However, NMR signals are easily interfered with, and the accuracy of the measurement results is greatly influenced by the analysis of the NMR signals and relaxation times.

2.4. Thin Section

The thin section analysis method is one of the most commonly used techniques in geological exploration and development. It involves a series of steps, including cleaning, drying, impregnation, cutting, grinding, mounting, trimming, polishing, and sealing, to prepare rock samples into transparent two-dimensional thin sections with micrometer-level thickness [36]. These thin sections are then observed under an optical microscope to examine the pore and mineral characteristics of the rock. By analyzing the optical properties of different minerals, the mineral composition can be determined, and the cementation types and pore characteristics can be studied [37].

This method has the advantages of being simple, efficient, and cost-effective, and is the most widely used method. However, its limitation lies in the ability to only observe the 2D structure of rocks. Additionally, mechanical damage to the sample may occur during the sample preparation process, which could affect the accuracy of the test results.

2.5. SEM

The SEM method uses an electron beam to induce secondary excitation on the surface of solid samples, obtaining scanning images that reflect the surface structure of the specimen. It can be used for studies related to the surface morphology of rocks, mineral composition, and other aspects. When using scanning electron microscopy, the most commonly used techniques are acquiring secondary electron images (SEI) [38] and backscattered electron images (BSE) [39]. The former provides details of the sample’s microscopic morphology, while the latter, combined with energy-dispersive X-ray spectroscopy (EDS), can provide compositional information of the sample.

This method offers ultra-high imaging resolution and measurement accuracy. Compared to conventional microscopes, it has a greater depth of field, a wider field of view, and produces more three-dimensional images, providing more comprehensive surface information of the sample. However, SEM measurements require sample preparation suitable for observation under the scanning electron microscope, ensuring the surface is flat and free of contaminants, which increases the complexity and cost of the experiment. Additionally, samples need to be observed in a vacuum environment, which limits the types of samples and their applicability.

2.6. CT Scanning

Both thin section and SEM scanning require pre-treatment of the rock samples, such as slicing, grinding, and polishing. This series of processes can cause some degree of damage to the sample surface, leading to changes in the pore structural features of the sample. In contrast to these two methods, CT scanning technology, which reconstructs the sample’s internal structure using projection data obtained through X-ray imaging, can be considered a non-contact, non-destructive 3D imaging technique [40].

An X-ray CT scanning system consists of three main components: the X-ray source, the sample stage, and the detector [7]. In the X-ray source, an electron beam is emitted from the cathode and strikes the anode target, exciting the atoms in the target material and releasing energy in the form of X-ray photons. These X-ray photons pass through the sample placed on the sample stage and are used to image the internal structure of the sample. The detector collects the varying attenuation signals, presenting a two-dimensional X-ray image (spectrum) of the current scanning position based on the X-ray absorption intensity. By rotating the sample stage, 2D projection data from different positions of the sample are acquired. Combining these projection data ultimately allows for the reconstruction of the sample’s 3D structure, as shown in Figure 3.

3. Theoretical Model of Fluid Flow in Porous Media

The prediction of fluid transport properties in the rock at the pore scale is the basis for analyzing macroscopic flow characteristics. Due to the complexity of the pore structure inside the rock, fluid flow within the rock often exhibits highly complex multi-scale features. At the nanoscale or mesoscopic scale, fluid flow can be modeled using fictive methods such as the lattice Boltzmann method, where the behavior of the fluid is described by the interactions of fluid molecules or pseudo-particles in random motion. This approach captures the random and discrete nature of fluid flow at small scales. However, at the macroscopic scale, the fluid flow process can be considered as continuous, with the flow characteristics described by classical fluid dynamics equations. The study of hydrogen flow in reservoirs is crucial for understanding its permeability in porous media. Due to the small size of hydrogen molecules, it exhibits higher diffusivity and greater permeability, which causes its flow behavior in low-permeability rocks to differ significantly from that of water or oil. Understanding the flow mechanisms of hydrogen in reservoirs is essential for the development of hydrogen storage and recovery technologies, particularly in the context of underground hydrogen storage and two-phase flow dynamics.

The flow of fluids in porous media is influenced not only by the complexity of the pore structure but also by various displacement mechanisms. In the case of two-phase flow, fluid movement within the pores typically exhibits different displacement patterns. Common displacement mechanisms include piston-like displacement, breakthrough phenomena, and hysteretic displacement. Piston-like displacement generally occurs in porous media with larger pore sizes or uniform pore structures, where the fluid moves in a relatively stable manner, resulting in a well-defined interface [41]. In contrast, breakthrough phenomena arise in more complex pore structures, where the fluid interface is irregular and accompanied by a significant pressure gradient. These displacement mechanisms have a significant impact on the macroscopic behavior of fluid flow, including permeability, seepage velocity, and other related properties. Therefore, understanding these mechanisms and incorporating them into flow models is essential for describing two-phase flow behavior in porous media, particularly in the context of underground hydrogen storage [42].

Darcy’s law is the most fundamental and widely used theory for describing the macroscopic flow behavior of fluids in porous media, and it is derived from steady-state flow experiments, as shown in Equation (1). The expression for two-phase flow in porous media can be derived from Equation (1), as shown in Equation (2) [43]:

$$k={\displaystyle \frac{q\mu L}{\nabla pA}}$$

(1)

$$q_{\alpha }=-{\displaystyle \frac{k{k}_{r\alpha }}{\mu }}(\nabla p)$$

(2)

where q is the flow rate, m³/s; k and kr represent the absolute permeability and relative permeability respectively, m²; μ is the fluid’s viscosity, Pa·s; ∇p is the pressure gradient, Pa; α equals 1 or 2, representing different fluid phases; A and L represent sample dimension, i.e., cross-sectional area and length, respectively.

Although Darcy’s law is the fundamental theory for describing fluid flow in porous media, in certain special cases, particularly when the pore structure is complex, flow channels are narrow, or two-phase flow occurs, the fluid flow behavior often deviates from the predictions of Darcy’s law. In these situations, the flow no longer follows the linear behavior described by Darcy’s law, but instead exhibits typical non-Darcy flow characteristics. As a result, researchers have started using non-Darcy flow models to describe two-phase flow [44]. To accurately simulate these complex flow mechanisms, several modified theoretical models have been proposed, taking into account the multi-scale nature of the flow and the impact of displacement mechanisms.

The MD equation characterizes the overall transport properties of fluids by studying the motion laws of all fluid particles in the statistical model, which can precisely describe the distribution of fluid particles in the model, thereby obtaining properties such as adsorption and diffusion of the fluid in the model. Figure 4 illustrates the molecular dynamics model for fluid migration in nanopores [45]. The MD equation, which solves the motion equations for each particle to obtain the changes in particle positions, differs significantly from the disordered pore structure of real porous media, and the basic mathematical governing equation is shown in Equation (3):

$$m_i{\displaystyle \frac{d^2{r}_i(t)}{d{t}^2}}=-{\displaystyle \sum _{i<j}\frac{\partial V(r_{ij})}{\partial r_{ij}}}$$

(3)

where m_i represents the mass of particle i, g/mol; r_i is the position vector of particle i; V(r_ij) is the interaction potential function between particles i and j; r_ij is the relative position between particle i and j; t is time, in picoseconds (ps).

The NS equation is based on the continuum hypothesis, which treats the fluid as a continuous medium filling the entire flow domain. This allows the definition of properties such as fluid density, velocity, temperature, and pressure at each point in the flow domain. The continuum hypothesis is a fundamental assumption in fluid mechanics and can be seen as an approximation of fluid behavior. The motion of the fluid satisfies the laws of conservation of mass, momentum, and energy, and can be mathematically described by a set of partial differential equations [46], as shown in Equation (4):

$$\left\{\begin{array}{c}{\displaystyle \frac{\partial \rho }{\partial t}}+\nabla \cdot \left(\rho u\right)=0\\ {\displaystyle \frac{\partial (\rho u)}{\partial t}}+\nabla \cdot \left(\rho uu\right)=\nabla \cdot \sigma \\ {\displaystyle \frac{\partial (\rho e)}{\partial t}}+\nabla \cdot \left(\rho ue\right)=\sigma :\nabla u-\nabla \cdot q\end{array}\right.$$

(4)

where u and e represent the velocity vector and the specific internal energy, respectively; σ is the stress tensor, Pa; q is the heat flux caused by thermal conduction and radiation, J/s.

To better reveal the inherent mechanism of the two-phase flow process under various scenarios, researchers have developed robust theoretical models to describe the flow process, which not only improve our understanding of the underlying fluid dynamics at the pore scale but also aid in optimizing the displacement efficiency of two-phase systems [47].

Flow in porous media can be described using different methods at various scales. At the pore scale, the flow behavior is typically analyzed by directly solving flow equations based on physical mechanisms. For single-phase flow, the Navier–Stokes (NS) equations are commonly used to describe the balance between fluid inertia and viscous forces. In the case of two-phase flow, methods such as the phase field or level-set methods are employed to solve the interface evolution equations, while considering microscopic forces like capillary pressure and surface tension. Molecular dynamics (MD) simulations are primarily used at the molecular scale, focusing on phenomena like wettability, surface chemistry, and nanoscale confinement effects. MD describes fluid flow through molecular interactions, such as collisions and transfer, while NS models flow as a continuous medium, applicable to the entire fluid domain. At the macroscopic scale, flow behavior is typically described using the continuum assumption and macroscopic equivalent equations. Common models include Darcy’s law, which is used for single-phase saturated flow, establishing the relationship between permeability and pressure gradient to determine flow velocity. The Richards equation, an extension of Darcy’s law, is used for unsaturated porous media and couples the dynamics of saturation and pressure fields. At the macroscopic scale, parameters such as permeability and relative permeability can be derived from the analysis of local flow fields.

Song et al. [48] developed a cyclic hydrogen injection and production system and used Darcy’s Law to create a mathematical model for investigating hydrogen production and recovery, as shown in Equations (5) and (6):

$$v=-{\displaystyle \frac{k}{\mu }}\nabla p$$

(5)

$${\rho }_g={\displaystyle \frac{T_{sc}{Z}_{sc}{\rho }_{gsc}}{p_{sc}}}\cdot {\displaystyle \frac{p}{TZ}}$$

(6)

where Tsc, Zsc, p_gsc, and p_sc represent the temperature, compressibility, gas-phase density, and pressure under standard conditions, respectively. p denotes the reservoir pressure, T represents the reservoir temperature, and Z is the actual compressibility factor.

To investigate the migration and mass transfer behavior of saturated and unsaturated gas–liquid two-phase systems, researchers [49] developed a mathematical model to describe the flow process on the basis of Darcy’s Law as shown in Equations (7) and (8), in which Equation (7) describes the mass balance of the gas and liquid phases:

$$\phi {\displaystyle \frac{\partial S_t}{\partial t}}+\nabla \cdot \overrightarrow{v_i}=0$$

(7)

$$\overrightarrow{v_i}=-{\displaystyle \frac{k_{r,i}}{{\mu }_i}}K(\nabla p_i-{\rho }_i\overrightarrow{g})$$

(8)

where the flux of the fluid phase ($\overrightarrow{v_i}$) can be described by Darcy’s law extending to two-phase flow; φ is the porosity; g is the acceleration due to gravity, m/s².

The mathematical models mentioned above all assume that the gas–water two-phase flow follows Darcy’s law. However, in certain cases, such as when the medium has low permeability, extremely small flow channel sizes, and under critical flow velocities, Reynolds numbers, and nonlinear flow conditions, the flow behavior deviates from Darcy’s law. In such cases, the high flow velocity and the interaction between the rock matrix and fluids within the pores lead to nonlinear seepage characteristics. As a result, scholars have turned to non-Darcy flow models to describe two-phase flow. Simultaneously, they have extended or modified Darcy’s law to account for multiphase flow and both miscible and immiscible conditions [50,51,52]. The theoretical model is a mathematical description of the complex pore-scale flow process and serves as the foundation for experimental and numerical simulation research. The continuous improvement of theoretical models is essential for better understanding complex flow processes.

4. Pore-Scale Visualized Experiments for Two-Phase Flow

The study of two-phase flow in porous rock, particularly the microscopic mechanical mechanisms, is crucial for UHS. Due to the complexity and irregularity of the pore structure, understanding the microscopic mechanisms involved in the flow process has long been a challenging issue and remains insufficient. This has resulted in a gap between theoretical research and experimental observations. Pore-scale visualized experiments, a technique capable of capturing the flow process and fluid distribution within the flow channel (i.e., pore structure), have undergone decades of development. In the past decade, with the development of advanced imaging technologies such as CT and NMR, micro-scale visualized seepage experiments have made significant progress. These advancements have been used to reveal more complex flow mechanisms, greatly accelerating the development of microscopic seepage mechanics.

4.1. Microfluidic-Based Visualized Experiment

The development and application of microfluidic chips provide a new approach to overcoming the limitation of not being able to visually observe the internal fluid flow process due to the opacity of rocks. These models use transparent materials to depict interconnected pore networks, enabling optical visualization of complex fluid flow processes occurring at the pore scale. Experiments, utilizing high-precision microscopes and recording equipment, implementing image analysis and experimental measurement techniques to achieve visualized studies of microscopic flow processes and dynamic mechanisms in porous media [53]. With the help of visualized microscopic seepage experiments, micro-models based on real rock core pore structures can achieve qualitative and quantitative research on the microscopic seepage process of reservoir cores, revealing the microscopic seepage characteristics of fluids within reservoirs. Those microfluids manufactured using different methods are shown in Figure 5.

By using microfluidic-based visualization experiments, there is fruitful research on gas–water two-phase flow mechanisms, which are essential for understanding the migration and trapping mechanisms of fluids within complex porous media, as shown in Figure 6. In previous studies, researchers explored various aspects of hydrogen-water two-phase flow in microfluidic systems. Researchers have investigated the flow behaviour of hydrogen in water-saturated porous media [60]. The bubble dynamics and two-phase flow in proton exchange membrane electrolytes have also been captured and studied by using microfluidics combined with high-speed in situ imaging technique [7]. The storage capacity and residual trapping characteristics of hydrogen during cyclic injections are also investigated in microfluidic channels, which offer insights into underground storage applications [61]. Thanks to the advantage of easy cyclic operation in microfluidic experiments, researchers have conducted multi-parameter cyclic injections and production simulations to explore the key factors influencing hydrogen underground storage capacity and recovery efficiency [62]. Collectively, these studies contribute to a deeper understanding of hydrogen–water flow behavior and its potential applications in energy production and storage systems.

In microfluidic experiments, the study of hydrogen–water two-phase flow has revealed the flow characteristics of hydrogen in water-saturated porous media. The research shows that hydrogen flow in microfluidic channels is significantly influenced by the pore structure. Larger pore channels facilitate the flow of the gas phase, while smaller pores may lead to blockages or local retention of hydrogen, which is crucial for understanding gas migration and trapping mechanisms in underground hydrogen storage. For instance, cyclic injection experiments demonstrate that the residual storage characteristics of hydrogen in porous media, such as residual gas saturation and gas sequestration capacity, are closely related to the pore structure. These findings are essential for understanding hydrogen’s underground storage, recovery efficiency, and storage capacity.

4.2. NMR-Based Visualized Experiment

NMR technology, as a rapid, non-destructive, and safe measurement method, it can measure the pore characteristics of reservoir porous media and conduct in-depth studies on the seepage characteristics of fluids within rocks. Furthermore, NMR technology can effectively separate fluids from the porous medium skeleton, obtaining intuitive information about the distribution of fluids within the core that traditional measurement methods cannot achieve, as well as the distribution state of fluids during the displacement process. When NMR is applied, relaxation phenomena occur, which are closely linked to the pore structure of the material. By employing mathematical analysis techniques, the distribution of fluids with different T2 relaxation times can be determined, producing what is known as the NMR T2 spectrum. According to the NMR relaxation mechanism, signals with longer T2 relaxation times in the spectrum correspond to fluids in larger pores, while those with shorter T2 times are associated with fluids in smaller pores [63]. Nuclear magnetic resonance (NMR) technology allows for the real-time observation of fluid morphology and the dynamic seepage process within porous structures, making it particularly effective for studying the seepage laws of fluid flow as well as the fluid-solid interactions in porous media, which are crucial for applications such as underground hydrogen storage.

NMR technology is utilized to study the two-phase fluid flow state in reservoir porous media, which is primarily divided into two approaches. One is to evaluate the permeability of the core by establishing a regression relationship between the saturation of mobile fluids, porosity, and permeability [64,65]; the other is to directly observe the distribution of fluids within the core using NMR imaging [66,67]. Cheng et al. 2021 [68] used nuclear magnetic imaging to analyze the gas–water two-phase flow characteristics of the core. By analyzing NMR images under different gas saturations, it was found that both gas and water preferentially flow through larger pores, but there is also flow through smaller pores during the process, leading to plugging of these smaller pores. Although this increases the bound water saturation, it further expands the gas phase flow channels, thus increasing the gas phase permeability and reducing the relative permeability of the water phase. NMR imaging reveals the dominant flow channels during the displacement process in the core and the distribution characteristics of gas and water within pores of different sizes. Goodarzi et al. 2025 [69] analyzed the visualization method of fluid flow and transport in porous media based on NMR imaging technology, as shown in Figure 7. As the volume fraction of the invading phase in the pores increases, the displacement front gradually stabilizes, and the saturation of the defending phase gradually decreases, as shown in Figure 7b. By comparing the saturation curves calculated by NMR and experimental testing, it is concluded that the liquid phase saturation curves obtained by both methods are essentially consistent, as shown in Figure 7c. Research on the dynamic flow experiments of gas–water two-phase fluids in underground reservoirs using NMR technology has shown that NMR imaging technology accurately and intuitively reflects the fluid distribution information in the core during the two-phase flow process. Compared with conventional methods of studying flow characteristics, it has the advantages of high experimental efficiency and the ability to obtain fluid distribution information across different sections without destroying the rock.

Nuclear magnetic resonance (NMR) experiments have elucidated the distribution characteristics of hydrogen in pore structures during gas–water two-phase flow. In these experiments, NMR is used to analyze hydrogen flow across different pore scales, revealing that hydrogen preferentially migrates through larger pores, whereas it may be partially trapped in smaller ones. NMR imaging illustrates the dominant flow pathways of the gas phase and the spatial distribution of the water phase during drainage/imbibition processes. These results provide critical insights into hydrogen flow pathways, changes in saturation, and hydrogen-water interfacial interactions during the flow process. Such findings are crucial for understanding fluid distribution and migration mechanisms in underground hydrogen storage systems.

4.3. CT-Based Visualized Experiment

Computed tomography (CT) imaging technology, which is fast, non-destructive, and safe, provides a three-dimensional visualization tool for characterizing rock pore structures and gas–water two-phase flow dynamics. It also delivers real-time, visualized results. These technologies have been widely applied in various fields, overcoming the resolution and dimensionality constraints of traditional planar micro-scale models to visualize fluid distribution at the pore scale. It allows can investigate microscale two-phase flow behaviors in reservoir cores while preserving their original geometry and internal structure. For example, by studying the displacement process between hydrogen and brine, researchers found that hydrogen experiences a certain degree of retention effect in the pore media [70]. By imaging different parts of the core and various displacement stages within the experimental setup, the dynamic evolution characteristics of the core’s internal microstructure, fluid flow, and distribution during the displacement experiment can be obtained [71]. Combined with monitoring information such as pressure and flow rate in the experiment, as well as CT cross-sectional imaging technology, the distribution characteristics of fluid saturation and heterogeneity within the core can be acquired. This enables a qualitative description and quantitative characterization of micro-displacement efficiency and the distribution of two-phase fluids.

To investigate the seepage pathways of gas–water two-phase flow in underground rock reservoirs, Researchers utilize insights from CT technology, including wet and dry images, to better inform pore-scale models. By visualizing the throats and pores within the rock matrix, we can gain insights into the factors influencing the efficiency of gas–water displacement within the rock’s pore structure. Guo et al. 2022 [72] used CT imaging to study the displacement between hydrogen gas and brine. The study revealed that after a 12-h period without flow, trapped hydrogen ganglia underwent significant rearrangement. Although the total mass of hydrogen remained unchanged, smaller gas ganglia tended to disappear while larger ones grew. Additionally, researchers can utilize CT imaging technology to assess the impact of rock physical properties on gas–water two-phase seepage dynamics. The distribution of two-phase fluids within porous media is affected by several factors, including rock type, surface wettability, and the capillary number et al. [73]. Furthermore, researchers conducted a quantitative and comprehensive study on the influence of rock properties on two-phase seepage characteristics, employing CT scanning and pore-scale multiphase flow modeling. They analyzed how rock surface wettability and heterogeneity influenced hydrogen transport in porous media [74]. Song et al. 2025 [75] conducted core displacement experiments involving hydrogen injection and extraction in sandstone samples, two sets of micro-CT images were captured during the primary drainage and imbibition stages. Image analysis revealed that after the hydrogen injection stage, hydrogen was distributed across both large and small pores and throats, but after water injection, hydrogen predominantly remained in larger pores, where capillary pressure barriers were lower. Thanks to the in situ CT scanning imaging technology, which allows for three-dimensional imaging of the heterogeneity of rock pore structures and reveals the differences in pore structure at different slice locations, the volume fraction of fluid at each slice position can be obtained. This results in an uneven distribution of fluid volume fractions across different strata of underground hydrogen storage pores, as shown in Figure 8a. The hydrogen volume fraction remained virtually unchanged between the H₂ imbibition and the post-12-h storage period. This behavior highlights hydrogen’s high mobility in porous media under reservoir conditions, where its strong diffusivity facilitates rapid migration, resulting in marked redistribution patterns [76]. Moreover, observing fluid flow behavior is essential when studying two-phase or multiphase systems, Huang et al. 2021 [77] conducted an in situ study on immiscible-phase fluid displacement in oil-wet reservoirs and found that under immiscible conditions, the gas phase exists as separated ganglia in medium-sized pore spaces, as shown in Figure 8b.

CT imaging technology provides a three-dimensional visualization tool for studying gas–water two-phase flow, enabling clear observation of hydrogen distribution and dynamic evolution within reservoir pore media. Using CT imaging, researchers can examine the gas flow paths, fluid saturation, and the dynamic distribution of the gas–water phases within the pores. For example, by studying the displacement process between hydrogen and brine, researchers found that hydrogen experiences a certain degree of retention effect in the pore media. The rearrangement of hydrogen clusters and changes in gas distribution are closely linked to the pore structure. The application of CT technology helps to deepen our understanding of hydrogen flow behavior in complex pore media, particularly in optimizing fluid distribution and flow paths when considering hydrogen storage and extraction processes.

5. Comparison of Interface Tracking Methods in Pore-Scale Two-Phase Flow Simulation

Recent advancements in digital core technology have enabled the quantitative characterization of the internal microstructure of reservoirs, offering deeper insights into the variation patterns of reservoir rock properties and the micro-scale mechanisms of seepage. On this basis, the pore-scale numerical simulation technology, compared to experiments, not only saves experimental costs, shorten experimental cycles, and overcomes the limitations of experimental conditions, but also provides intuitive and visually clear simulation results. Moreover, it allows for lateral comparative simulation experiments under different conditions using the reconstructed models. With the development of computer technology and computational fluid dynamics theory, pore-scale numerical simulation technology based on micro-CT images of rocks has become increasingly favored by researchers.

In two-phase or multiphase flow systems within porous media, a distinct phase interface exists, and this interface is continually evolving as the system progresses. Accurately capturing the dynamics of the gas–liquid phase interface is a critical aspect of understanding two-phase seepage mechanisms. However, due to the complex flow patterns inherent in two-phase flow and the intricate coupling between the flow field and the fluid exchange mechanisms at the interface, verifying the accuracy of interface conditions remains challenging. Unlike single-phase flow, the defining characteristic of two-phase and multiphase flows is the presence of a well-defined phase interface within the mixed system. As the mixture moves through the porous media, the shape and state of these phase interfaces are continuously changing and evolving, further complicating the study of their behavior [78].

To develop an accurate pore-scale two-phase flow model, it is crucial to precisely describe the fluid interfaces. In practical applications, the evolution of fluid interfaces is closely related to the distribution of residual fluids, especially during the interaction between gas and liquid, as shown in Figure 9 [79]. Researchers worldwide commonly employ two primary methods to track the motion of these interfaces: the Eulerian and Lagrangian approaches. The Lagrangian method tracks the motion of material points that reside on the interface itself, which is numerically implemented by following grid points and either the local fluid velocity or the velocity of grid movement. However, this method faces challenges, particularly in dealing with the deformation of interfaces. These challenges include the need for grid remeshing and interpolation, which can lead to increased computational costs and errors. In contrast, the Eulerian method tracks the flow of the two fluid components on a fixed grid. The interface is captured by an isosurface of a globally defined function, and at each time step, the interface is reconstructed or determined using a scalar indicator function. The Eulerian method generally outperforms the Lagrangian method, particularly in handling complex interface features, such as irregular contours, sharp corners, merging, and destructive interfaces. The most widely used Eulerian methods for two-phase flow modeling include the level set method [80], the volume of fluid (VOF) method [60], and the phase field method [81]. In addition to these methods, the pore network (PN) model is also a critical technique for pore-scale modeling in hydrogen storage, as well as in applications like enhanced oil recovery (EOR) and carbon sequestration. The PN model simulates fluid flow by discretizing the pore structure into pores and throats, which is well-suited for capturing phenomena such as capillary pressure, relative permeability, and fluid distribution at the pore scale. PN has proven valuable for accurately simulating the behavior of multiphase fluid systems in porous media, making it an essential tool for hydrogen storage applications. The integration of PN with other numerical methods, such as the Eulerian or Lagrangian approaches, can provide a more comprehensive understanding of fluid dynamics in hydrogen storage systems.

Table 2. Application of mathematical model of two-phase seepage in porous media.

Numerical Methods for Multiphase Flow	Formula	Scope/Characteristics
VOF Method Zhu et al., 2023 [82]; Wang et al., 2022 [83]	$\begin{array}{c}{\displaystyle \frac{\partial (\rho \overrightarrow{V})}{\partial t}}+\nabla \cdot (\rho \overrightarrow{V}\overrightarrow{V})=-\nabla p+\nabla \cdot [\mu (\nabla \overrightarrow{V}+\nabla \overrightarrow{V^T})]+\rho \overrightarrow{g}+\overrightarrow{F_s}\\ {\displaystyle \frac{\partial a}{\partial t}}+\nabla \cdot (au)+\nabla \cdot \left[a(1-a)u_c\right]=0\end{array}$ where p represents the pressure; g represents the acceleration due to gravity; ρ represents the density; $\overrightarrow{V}$ denotes the velocity vector of the fluid; μ denotes the dynamic viscosity of the fluid; $\overrightarrow{F_s}$ describes the Laplace pressure acting at the interface; The volume fraction of the water phase is aw. α is represents flow volume; uc is represents compress velocity.	It is applicable to immiscible multiphase flow interfaces with very distinct phase boundaries. However, the VOF (volume of fluid) function is a discontinuous function at the phase interface, leading to poor continuity.
Phase Field Method (Using the Cahn-Hilliard equation) Safari et al., 2024 [80]	$\begin{array}{c}{\displaystyle \frac{\partial \phi }{\partial \mathrm{t}}}+u\nabla \phi =\nabla \cdot {\displaystyle \frac{\gamma \lambda }{{\epsilon }^2}}\nabla \psi \\ \psi =-\nabla \cdot {\epsilon }^2\nabla \phi +({\phi }^2-1)\phi \end{array}$ where φ represents the phase field variable in different regions, which varies between −1 and 1, ψ represents the chemical potential, also known as the phase field auxiliary variable, ε is the control parameter for the thickness of the two-phase interface, and γ denotes the migration regulation parameter; u is the velocity.	It is suitable for studying the wettability and two-phase seepage mechanisms in porous structures with complex pore throat geometry, as it can accurately capture the interface; however, it is unable to maintain mass conservation.
Level Set Liu et al., 2025 [79]	${\displaystyle \frac{{\partial }_{\phi }}{{\partial }_t}}+u\cdot \nabla \phi =\gamma \nabla \cdot (\epsilon \nabla \phi -\phi (1-\phi ){\displaystyle \frac{\nabla \phi }{\left|{\nabla }_{\phi }\right|}})$ where ∂φ/∂t represents the accumulation term with respect to time; u∙∇φ represents the advection term, where u denotes the velocity; when the fluid is compressible, the velocity divergence is not zero, in which case the advection term is ∇(uφ); γ is the reinitialization parameter; ε is the interface thickness control parameter.	It is suitable for tracking free interfaces in complex fluids, with good continuity at the phase interface, but it cannot quantitatively satisfy mass conservation.
VOSET Equation Ling et al., 2019 [84]	$\begin{array}{c}\nabla \cdot \overrightarrow{u}=0\\ {\displaystyle \frac{\partial \overrightarrow{u}}{\partial t}}+\overrightarrow{u}\cdot \nabla \overrightarrow{u}={\displaystyle \frac{1}{p(\varphi )}}\left\{-\nabla p+\nabla \cdot u(\varphi )\left[\nabla \overrightarrow{u}+{(\nabla \overrightarrow{u})}^T\right]+p(\varphi )\overrightarrow{g}+\overrightarrow{F_s(\varphi )}\right\}\end{array}$ where ф is the symbol for the distance function; p is pressure, in Pa; p(ф) and u(ф) refer to the gas–liquid mixture density and the mixture viscosity; u is the velocity; g represents the acceleration due to gravity.	Coupling the VOF and LS methods, and using a geometric interface front construction approach to build the phase interface.
PNM Method Cai et al., 2022 [85]	$\begin{array}{c}{\displaystyle \sum _{i\to j}{q}_{ij}}=0\\ q_{ij}=g_{ij}(P_i-P_j)\\ k={\displaystyle \frac{Q\mu L}{\Delta PA}}\end{array}$ where j refers to all pores connected to pore i; qij is the flow rate between pore i and pore j; gij is the conductivity of the throat connecting pore i and pore j; Pi and Pj are the pressures at pore i and pore j; k is the permeability of the pore network model; p is the applied pressure gradient between the inlet and outlet of the model; L is the length of the model in the flow direction.	Two-phase flow processes can be simulated by constructing models that incorporate parameters such as the distribution, size, and aspect ratio of pores and throats. Analyzing how these parameters evolve during the flow process can predict flow patterns in porous media.

provides a comparative analysis of the applicability of various interface tracking methods for two-phase fluids.

6. Conclusions and Prospects

Investigation of the gas–water two-phase flow mechanism in reservoir rocks is crucial for UHS. This paper reviews pore structure characterization approaches, two-phase flow models, imaging-based pore-scale flow experiments, and interface tracking methods, offering insights into the migration and transport characteristics of two-phase fluids within porous media.

(1)

Accurate characterization of the pore structure in porous media is essential for predicting rock properties and fluid transport behavior. Various testing methods offer qualitative, semi-quantitative, or quantitative insights into the micro-pore structure, but each has limitations. A multi-technique integrated approach is crucial for a more comprehensive and precise analysis of rock properties and fluid flow within porous media.

(2)

Microscopic seepage experiments using visualization methods provide insights into fluid flow and pore-scale mechanics. Micro-seepage models quantify gas–water two-phase flow, optimizing conditions for improved production. These models aid in studying complex flow phenomena and, combined with 3D simulations, enhance understanding of reservoir seepage patterns and recovery factor research.

(3)

Recent advances in digital core technology and computational fluid dynamics have improved pore-scale two-phase flow simulations. Eulerian methods like VOF, level set, and phase field track phase interfaces but face challenges in mass conservation and accuracy. Integrating Eulerian and Lagrangian methods and enhancing computational efficiency will benefit applications in oil recovery, carbon sequestration, and hydrogen storage.