Pore-Scale Numerical Simulation of CO₂ Miscible Displacement Behavior in Low-Permeability Oil Reservoirs

Abstract

CO₂ miscible flooding provides dual advantages in enhancing oil recovery and facilitating geological sequestration, and has become a key technical approach for developing low-permeability oil reservoirs and carbon emission reduction. The pore-scale flow mechanisms governing CO₂ behavior during miscible flooding are crucial for achieving efficient oil recovery and secure geological storage of CO₂. In this study, pore-scale two-phase flow simulations of CO₂ miscible flooding in porous media are performed using a coupled laminar-flow and diluted-species-transport framework. The model captures the effects of diffusion, concentration distribution, and pore structure on the behavior of CO₂ miscible displacement. The results indicate that: (1) during miscible flooding, CO₂ preferentially displaces oil in larger pore throats and subsequently invades smaller throats, significantly improving the mobilization of oil trapped in small pores; (2) increasing the injection velocity accelerates the displacement front and improves oil utilization in dead-end and trailing regions, but a “velocity saturation effect” is observed—when the inject velocity exceeds 0.02 m/s, the displacement pattern stabilizes and further gains in ultimate recovery become limited; (3) higher injected CO₂ concentration accelerates CO₂ accumulation within the pores, enlarges the miscible sweep area, promotes a more uniform concentration field, leads to a smoother displacement front, and reduces high-gradient regions, thereby suppressing local instabilities, and improves displacement efficiency, although its effect on overall recovery remains modest; (4) CO₂ dynamic viscosity strongly influences flow stability: low-viscosity conditions promote viscous fingering and severe local bypassing, whereas higher viscosity stabilizes flow but increases injection pressure drop and energy consumption, indicating a necessary trade-off between flow stability and operational efficiency.

1. Introduction

With the continued acceleration of global industrialization, fossil energy consumption has risen sharply, leading to a substantial increase in CO₂ emissions. This trend has intensified the greenhouse effect and contributed to more frequent extreme climate events, posing serious challenges to human security and sustainable development [1,2,3,4,5,6]. Among the available emission-reduction pathways, CO₂ capture, utilization, and storage (CCUS) is widely regarded as one of the most promising and feasible strategies for mitigating climate change [7,8]. In oilfield operations, CO₂ flooding not only enables geological sequestration but also significantly enhances recovery in low-permeability oil reservoirs, thereby offering dual strategic benefits [9,10,11,12,13,14].

Depending on reservoir temperature, pressure, and crude oil composition, CO₂ flooding can be classified into miscible, immiscible, and near-miscible displacement modes [15,16]. When the reservoir pressure exceeds the minimum miscibility pressure, CO₂ and crude oil form a single-phase system, and the displacement process transitions into the miscible stage. Under miscible conditions, transport of the CO₂–oil system in porous media is mainly governed by convection and diffusion [17]. A portion of the injected CO₂ dissolves into the oil phase, forming a miscible zone that reduces oil viscosity and density, lowers interfacial tension, and thereby improves oil mobility and displacement efficiency [18,19,20,21,22]. Figure 1 illustrates a simplified schematic of the CO₂ flooding process: CO₂ passes through the pore structure, undergoes mass transfer, and interacts with the oil phase, forming a miscible region that propagates along pore channels and progressively displaces the oil.

However, during CO₂ miscible flooding, a series of pore-scale displacement phenomena such as viscous fingering, preferential flow paths, limited sweep area, oil trapping, and pore blockage are frequently observed [8,23,24]. These non-uniform displacement behaviors significantly restrict macroscopic sweep efficiency. While most existing studies focus on the diffusion and dissolution of CO₂ into crude oil, comparatively less attention has been devoted to the detailed transport behavior of CO₂ within porous media [25,26,27,28]. Therefore, elucidating the multiphase flow characteristics of the CO₂–oil system at the pore scale is of great significance for deepening the understanding of the displacement mechanisms and optimizing CO₂ injection strategies.

Numerical simulation provides an effective tool for investigating multiphase fluid flow in porous media at the pore scale [29,30,31,32,33,34,35]. Commonly used pore-scale modeling approaches include artificially constructed pore network models [31] and pore structures reconstructed from 2D or 3D images of real cores obtained through various scanning techniques [33,34,36]. Among these, generating artificial pore structures and numerically solving the governing equations has emerged as a promising approach for simulating CO₂ flooding in porous media [37,38,39]. Typical numerical methods include the finite difference method (FDM), finite volume method (FVM), finite element method (FEM), lattice Boltzmann method (LBM), phase-field method, and the “laminar flow + diluted species transport” approach [40,41,42]. Zhu et al. [31] employed a phase-field method to numerically simulate immiscible CO₂ displacement in a homogeneous porous medium. They found that once CO₂ breaks through the outlet, the pressure along the primary CO₂ channel drops significantly, leading to oil flowing back into pores previously occupied by CO₂. Song et al. [43] applied the “laminar flow + diluted species transport” method to investigate the key factors influencing CO₂ miscible flooding and demonstrated that pore structure markedly affects displacement efficiency, leading to the formation of multiple types of residual oil; furthermore, variations in CO₂ density and injection velocity can reshape effective flow channels and regulate the spatial distribution of residual oil. Zhang et al. [23] performed numerical simulations of liquid CO₂ flooding using two heterogeneous pore-network models with contrasting permeability zones and found that, at low injection velocities, the displacement mechanism transitions from capillary fingering to viscous fingering as the injection rate increases. Shi et al. [30] employed LBM to systematically study the effects of viscosity, wettability, and gravity on the flow behavior of non-Newtonian fluids in porous media. Wang Yuxia et al. [44] developed a transient numerical model for immiscible CO₂ displacement in low-permeability sandstone reservoirs, incorporating CO₂ dissolution and gas slippage mechanisms. Ma Qingsong [42] applied the “laminar flow + diluted species transport” method to simulate CO₂–Oil displacement under miscible, near-miscible, and immiscible conditions, showing that under miscible conditions, CO₂ preferentially displaces oil from larger pore throats and subsequently invades smaller throats, thereby significantly improving the mobilization of oil trapped in narrow pores. Overall, comparative studies indicate that the “laminar flow + diluted species transport” method offers advantages in computational accuracy, numerical stability, and efficiency, and can effectively capture fluid transport and mass-transfer processes. It is therefore suitable for simulating multiphysics coupling processes, such as fluid flow and mass transfer, in low- and ultra-low-permeability reservoirs with complex pore structures [45,46,47,48].

Based on the above understanding, this study develops a pore-scale CO₂ miscible displacement model to quantitatively analyze the migration and displacement behavior of CO₂ in crude oil within a two-dimensional randomly packed porous medium. The model adopts a coupled “laminar flow + diluted species transport” framework to solve the incompressible flow field and CO₂ concentration field within the pore network, thereby capturing the convection–diffusion coupling mechanism that governs miscible flooding. CO₂ injection velocity, injection concentration, and dynamic viscosity are selected as the primary control variables. Their effects on CO₂ migration rate, concentration distribution, and displacement-front propagation are systematically investigated, and their contributions to ultimate oil recovery are evaluated. The results provide a theoretical basis for clarifying pore-scale displacement mechanisms and optimizing CO₂ injection parameters for field development.

2. Materials and Methods

2.1. Physical Model

The geometric model shown in Figure 2a consists of a randomly generated by COMSOL Multiphysics 6.0 (Stockholm, Sweden), two-dimensional pore-channel structure with an outer boundary size of 6.2 mm × 3.2 mm. Figure 2b illustrates the initial state of the porous medium, in which all pore space is fully saturated with crude oil. The computational domain is discretized using an unstructured triangular mesh comprising 31,182 cells; the final mesh layout is shown in Figure 2c. A prescribed inlet pressure is applied at the left boundary, and zero-pressure outlet conditions are imposed on the right boundaries of the model. In contrast, no-slip boundary conditions are applied along the pore walls. Due to the strong geometric randomness of the pore structure, the channel widths and connectivity exhibit pronounced non-uniform spatial distribution, and the relevant parameters are summarized in

Table 1. Boundary conditions of the model.

Boundary Type	Boundary Condition	Value
Inlet	Pressure	p = p0
Outlet	Pressure	p = 0
Grain wall	Wall	No-slip
Symmetry plane	Symmetry	——

. During the simulation, CO₂ is injected from the left side of the model and migrates toward the outlet on the right, displacing the resident oil along the pore channels and ultimately producing a corresponding volume of oil at the outlet.

2.2. Mathematical Model

In CO₂ flooding of low- and ultra-low-permeability reservoirs, the migration and interaction of CO₂ and crude oil within porous media are complex processes governed by coupled convection and diffusion. At the pore scale, this process cannot be accurately described using simple algebraic equations or empirical correlations. Nevertheless, the continuity equation and the governing equations of mass and momentum conservation remain valid. Accordingly, a pore-scale flow model is established in this study, and a numerical framework combining laminar flow and diluted species transport is employed to solve the coupled flow and mass-transfer processes. This approach provides favorable numerical stability, high computational accuracy, and good efficiency, enabling reliable characterization of fluid displacement dynamics. Compared with alternative methods, the present framework directly couples pore-scale flow with two-phase transport without requiring explicit interface tracking [41,42,44,49]. The analysis focuses on three key control parameters: CO₂ injection velocity, CO₂ concentration, and dynamic viscosity. These parameters directly influence the driving force of fluid motion, the diffusive capability, and the flow resistance, respectively, thereby significantly affecting displacement efficiency and ultimate oil recovery, providing a basis for optimal design of CO₂ flooding parameters.

The pore-scale miscible CO₂ flooding process involves highly complex flow and mass-transfer behavior, making numerical modeling computationally intensive and intricate. To simplify the simulations and improve computational efficiency while maintaining physical plausibility, the following assumptions are adopted:

(1)

During the oil displacement process, molecular diffusion of CO₂ in the oil phase is relatively slow; moreover, within the pressure–temperature range considered in this study, the influence of compositional variations on the diffusion coefficient is secondary. Therefore, the variation in the CO₂ diffusion coefficient in the oil phase is neglected [15].

(2)

Both the CO₂ phase and the crude oil phase are treated as incompressible Newtonian fluids; that is, their density and viscosity do not vary appreciably with pressure and/or concentration, and the shear stress is linearly proportional to the shear rate [42].

(3)

CO₂ flow in the porous medium is assumed to follow Darcy’s law, and inertial effects as well as potential non-Darcy terms that may arise under high flow rates or in the near-wellbore region are neglected [50].

(4)

Two-phase CO₂–oil flow within the porous medium is assumed to be isothermal; thus, the Joule–Thomson effect, heat release/absorption associated with dissolution and/or miscible mixing, and viscosity–density variations induced by temperature gradients are ignored [51].

(5)

When the pressure reaches or exceeds the minimum miscibility pressure (MMP) and miscibility is achieved, the interfacial tension is assumed to approach zero, so capillary forces are not explicitly considered. In addition, oil swelling, extraction/exchange of light components, phase behavior/compositional evolution, and chemical reactions induced by miscible conditions are treated as secondary effects and therefore neglected; the displacement behavior is represented using an “effective miscibility” approximation [15,52].

2.2.1. Fluid Flow Equations

In the numerical simulations of this study, the Reynolds number of the flow system is Re < 1 ($\mathrm{Re}={\displaystyle \frac{\mathsf{\rho }\mathrm{uL}}{\mathsf{\mu }}}={\displaystyle \frac{1.977 \times 0.04 \times 0.002}{0.015}}=1.0544\times {10}^{-2}$), indicating that the model operates in a typical viscosity-dominated laminar flow regime throughout. Under these conditions, inertial effects are neglected, and fluid motion is primarily controlled by viscous resistance. Therefore, the fluid dynamics during the displacement process can be described using the principles of mass and momentum conservation. By resolving the spatial and temporal distributions of fluid velocity, density, pressure, and dynamic viscosity, the flow behavior of fluids within complex porous media can be elucidated [53]:

$$\left\{\begin{array}{c}\rho {\displaystyle \frac{\partial u}{\partial t}}+\rho \left(u\cdot \nabla \right)u=\nabla \cdot \left[-pI+K\right]+F\\ \rho \nabla \cdot u=0\\ K=\mu \left[\nabla u+{\left(\nabla u\right)}^T\right]\end{array},\right.$$

(1)

In the governing equations, the variables and parameters are defined as follows: $\rho$ is the fluid density (kg/m³); u is the fluid velocity (m/s); t is time (s); p is the reservoir pressure (MPa); I is the unit vector; $\mu$ is the dynamic viscosity of the fluid (mPa·s); F is the interfacial tension (N/m³); K is the viscous stress tensor.

2.2.2. Convection–Diffusion Equation

The convection–diffusion equation is widely used to describe the coupled processes of fluid flow and chemical species transport, and is well suited for simulating mass transport in porous media. This equation describes solute transport in a fluid when the solid skeleton remains stationary, and can also represent convection and diffusion within a gaseous medium under the assumption of a stationary gas phase. Based on the principles of mass conservation and dispersion (diffusion), the convection–diffusion process can be expressed as [54]:

$$\left\{\begin{array}{c}{\displaystyle \frac{\partial c_i}{\partial t}}+\nabla \cdot J_i+u\cdot \nabla c_i=R_i\\ J_i=-D_i\nabla c_i\end{array},\right.$$

(2)

In the governing equations, $c_i$ is the CO₂ concentration (mol/m³); $J_i$ is the diffusive flux (m/s); $R_i$ is the source term; $D_i$ is the diffusion coefficient.

During the displacement process under miscible conditions, the continuity equation, flow (momentum) equation, and convection–diffusion equation can be written as follows [50]:

$${\displaystyle \frac{\partial {\rho }_{C{O}_2}}{\partial t}}+\nabla \cdot \left({\rho }_{C{O}_2}\cdot u_{C{O}_2}\right)=0$$

(3)

In the governing equations, $u_{C{O}_2}$ denotes the velocity (m/s); ${\rho }_{C{O}_2}$ is the density (kg/m³).

$${\displaystyle \frac{\partial ({\rho }_{C{O}_2}{u}_{C{O}_2})}{\partial t}}+\nabla \cdot \left({\rho }_{C{O}_2}{u}_{C{O}_2}{u}_{C{O}_2}\right)=-\nabla p+\nabla \left[{\mu }_{C{O}_2}\left(\nabla u_{C{O}_2}+{\left(\nabla u_{C{O}_2}\right)}^T\right)-{\displaystyle \frac{2}{3}}\mu \left(\nabla \cdot u_{C{O}_2}\right)\right],$$

(4)

In the governing equations, ${\mathsf{\mu }}_{{\mathrm{CO}}_2}$ is the viscosity of CO₂ (mPa·s); ${\mathrm{c}}_{{\mathrm{CO}}_2}$ is the CO₂ concentration in the pore space of the rock matrix (mol/m³).

$${\displaystyle \frac{\partial c_{C{O}_2}}{\partial t}}+\nabla \cdot \left(-D\nabla c_{C{O}_2}\right)+u_{C{O}_2}\cdot \nabla c_{C{O}_2}=0$$

(5)

In the governing equations, D is the diffusion coefficient of CO₂ in the sandstone pore channels (m²/s).

2.3. Numerical Verification

To ensure the reliability and applicability of the constructed pore-scale numerical model, a combined numerical analysis and physical-experiment approach was employed. The experimental component consisted of pore-scale microfluidic CO₂ flooding tests, with the experimental setup shown in Figure 3. The microfluidic chip was fabricated based on the real pore structure obtained from CT scanning of reservoir cores from a field demonstration area in the Yanchang Oilfield, China. The pore geometry was subsequently processed via glass-etching techniques to produce a two-dimensional pore network whose geometric morphology and scale closely match those of the numerical model (Changyuan Oilfield, Shaanxi, China); the detailed fabrication procedure is illustrated in Figure 4. To validate the pore-scale model, a glass-etched chip with the same pore structure as in the numerical simulations was selected for microfluidic experiments. The key parameters used in both experiments and simulations are summarized in

Table 2. Input parameters used in the physical experiments and numerical simulations.

Physical Quantity	Value	Description
u0	0.1/1 µL/min	CO2 injection velocity
T0	323.15 K	CO2 injection temperature
p0	8 MPa	CO2 injection pressure
mu0	14.9 mPa·s	CO2 dynamic viscosity
c0	0.001 mol/m3	CO2 concentration

.

During the experiments, the glass-etched chip was mounted in a micro-visualization model holder and saturated with crude oil under a confining pressure of 8 MPa and a temperature of 50 °C. CO₂ was injected at flow rates of 0.1 μL/min and 1 μL/min, respectively, until no additional oil was produced at the outlet. Based on this, the oil recovery under different injection rates was calculated, as shown in Figure 5.

Figure 6 presents the numerical simulation results under the same operating conditions. A comparison between the residual oil spatial distributions obtained from microfluidic experiments and simulations after flooding indicates that, with increasing injection rate, the mobilization of residual oil in dead-end pores and corner regions is significantly enhanced, and the simulated trends align well with the experimental observations. The oil recovery at different injection rates is determined by using Adobe Photoshop to extract and quantify the oil-colored pixels in the final images and combining this with the oil-colored area extracted from the initial fully oil-saturated image. Furthermore, as shown in Figure 7, the relative error between simulated and experimental oil recovery remains within 10%. These results demonstrate that the developed pore-scale numerical model effectively captures flow behavior and residual oil distribution during miscible CO₂ flooding, thereby confirming the model’s accuracy and feasibility.

3. Results

The numerical investigation is divided into three groups of simulation cases.

(a)

In the first group, the effects of injection rate on the displacement mechanism and oil recovery are examined at a fixed CO₂ concentration and dynamic viscosities of 0.001 mol·m⁻³ and 15 mPa·s, respectively (A1–A9,

Table 3. Numerical simulation schemes.

Number	$u_{C{O}_2}$ (m/s)	${\mathbf{c}}_{C{O}_2}$ (mol/m3)	${\mu }_{C{O}_2}$ (mPa·s)	Number	$c_{C{O}_2}$ (mol/m3)	$u_{C{O}_2}$ (m/s)	${\mu }_{C{O}_2}$ (mPa·s)	Number	${\mu }_{C{O}_2}$ (mPa·s)	$u_{C{O}_2}$ (m/s)	${\mathbf{c}}_{C{O}_2}$ (mol/m3)
A1	0.001	0.001	15	B1	0.001	0.01	15	C1	10	0.01	0.001
A2	0.005			B2	0.0015			C2	15
A3	0.01			B3	0.002			C3	20
A4	0.015			B4	0.0025			C4	25
A5	0.02			B5	0.003			C5	30
A6	0.025			B6	0.0035			C6	35
A7	0.03			B7	0.004			C7	40
A8	0.035			B8	0.0045			C8	45
A9	0.04			B9	0.005			C9	50

).

(b)

The second group investigates the influence of CO₂ injection concentration on the displacement process at a constant injection velocity of 0.01 m·s⁻¹ and a continuous dynamic viscosity of 15 mPa·s (B1–B9, Table 3).

(c)

The third group analyzes the effect of CO₂ dynamic viscosity on oil recovery under fixed injection velocity and CO₂ concentration (C1–C9, Table 3).

3.1. Effect of CO₂ Injection Velocity on Displacement Performance

The injection velocity reflects the driving force exerted by CO₂ during flooding. The distribution of CO₂ concentration was analyzed at different injection velocities ranging from 0.001 to 0.04 m/s. As shown in Figure 8, within the miscible zone, CO₂ concentration increases more rapidly with higher injection velocities. However, when the CO₂ injection velocity is ≥0.02 m/s, the final displacement pattern shows only minor differences. As shown in Figure 9, at higher injection velocities, the residence time of CO₂ in the flow region is relatively short, leading to earlier breakthrough. The displacement front position exhibits an approximately linear relationship with the CO₂ injection velocity. As illustrated in Figure 10, increasing injection velocity enhances the displacement driving force: more CO₂ enters the porous medium, the spatial extent of CO₂ distribution expands, and the mobilization of residual oil in dead-end pores, cluster-like regions, isolated pockets, and trailing zones within the miscible region is significantly enhanced. This phenomenon is particularly pronounced in larger pore channels, whereas in porous media with smaller pores, displacement predominantly follows the dominant flow paths.

3.2. Effect of CO₂ Injection Concentration on Displacement Performance

A parametric sweep of CO₂ injection concentration was conducted to simulate miscible CO₂ flooding in porous media, with different injection concentrations ranging from 0.001 mol/m³ to 0.005 mol/m³. As shown in Figure 11, higher CO₂ injection concentrations lead to faster CO₂ accumulation within the domain and produce a significantly larger swept area over the same time period.

According to Equation (7), increasing the CO₂ injection concentration decreases the spatial variance of the concentration field, indicating more uniform mixing and a higher overall CO₂ concentration within the porous medium. Although the contrast concentration fields differ markedly across cases, the resulting CO₂ flooding efficiency and the final displacement pattern show only minor differences. At high CO₂ injection concentrations, the displacement front advances more smoothly, with reduced fluctuations in front velocity and a smaller proportion of high-gradient regions. These trends suggest that higher CO₂ concentration is beneficial for suppressing local instabilities in the displacement front and enhancing miscible flooding efficiency. It should be emphasized, however, that these findings are inferred indirectly from the evolution of the concentration field. Quantitative conclusions regarding ultimate oil recovery require further evaluation based on saturation distributions or material-balance calculations.

$$\overline{C}={\displaystyle \frac{1}{\left|\Omega \right|}}{{\displaystyle \int }}_{\Omega }C\left(x,y\right)d\Omega ,$$

(6)

In the governing equations, $\Omega$ denotes the spatial region of the reservoir domain; $\left|\Omega \right|$ is the area of the domain; $C\left(x,y\right)$ is the local CO₂ concentration at the spatial position $\left(x,y\right)$; $\overline{C}$ is the spatial mean CO₂ concentration within the domain; $d\Omega$ is the infinitesimal area element.

$${\sigma }_{\mathrm{C}}^2={\displaystyle \frac{1}{\left|\Omega \right|}}{{\displaystyle \int }}_{\Omega }{\left[C\left(x,y\right)-\overline{C}\right]}^{2}d\Omega ,$$

(7)

In the governing equations, ${\sigma }_{\mathrm{C}}^2$ is the spatial variance of the concentration field.

3.3. Effect of CO₂ Dynamic Viscosity on Displacement Performance

During miscible displacement of crude oil by CO₂, interfacial effects become negligible, and fluid dynamics govern the flow behavior. To assess the influence of CO₂ viscosity, simulations were performed over a viscosity range of 0.001–0.005 mol/m³, while simultaneously monitoring the corresponding reservoir pressure response. As shown in Figure 12, CO₂ dynamic viscosity strongly affects the smoothness of the pressure field and the distribution of pressure gradients. At lower CO₂ viscosity, the mobility ratio M increases ($M={\displaystyle \frac{{\mu }_{oil}}{{\mu }_{C{O}_2}}}$), indicating stronger pressure heterogeneity and more pronounced local high-gradient zones. As the displacement front advances through the porous medium, viscous fingering is more likely to occur, leading to an unstable front, aggravated channeling, enlarged unswept regions, and consequently reduced macroscopic oil recovery. With increasing CO₂ dynamic viscosity, the mobility ratio of the displacement system improves, the spatial distribution of the pressure field becomes more uniform, pressure fluctuations are reduced, flow stability is enhanced, viscous fingering is significantly suppressed, and the local sweep area increases. As shown in Figure 13, reservoir pressure in the porous medium exhibits an approximately linear relationship with CO₂ dynamic viscosity. However, excessively high CO₂ viscosity increases formation pressure, resulting in a larger injection pressure drop and higher energy consumption, which can be detrimental to oil recovery at a fixed injection velocity. Therefore, in engineering applications, an optimal viscosity should be selected by balancing displacement stability against injection energy consumption.

4. Discussion

In this study, a pore-scale miscible CO₂ flooding model was developed based on a coupled “laminar flow + diluted species transport” framework, and the effects of CO₂ injection velocity, CO₂ injection concentration, and CO₂ dynamic viscosity on miscible displacement behavior were systematically analyzed. By combining numerical simulations with microfluidic experiments, the mechanisms of miscible CO₂ flooding and implications for engineering optimization are discussed as follows:

1.

Increasing CO₂ injection velocity accelerates the propagation of the displacement front and enhances the mobilization of residual oil in dead-end pores and trailing zones. However, a velocity saturation effect is observed. When the injection velocity exceeds 0.02 m/s, the displacement pattern tends to stabilize, indicating that further increases in injection velocity yield only marginal improvements in ultimate oil recovery.

2.

Higher CO₂ injection concentration leads to faster CO₂ accumulation within the pore space, a substantial enlargement of the miscible swept area, greater spatial uniformity of the concentration field, and smoother advancement of the displacement front. At higher concentrations, front-velocity fluctuations are reduced, and the proportion of high-gradient regions decreases, suggesting that higher injection concentration helps suppress local instability of the displacement front and enhances miscible flooding efficiency. However, the final displacement configuration and overall displacement efficiency showed only minor variation across the concentration range considered.

3.

CO₂ dynamic viscosity strongly influences flow stability. Low viscosity promotes viscous fingering, resulting in channeling and bypassing of oil. Conversely, excessively high viscosity improves flow stability and suppresses fingering, but increases injection pressure drops and higher energy consumption. Therefore, an optimal viscosity must balance improved displacement stability against the operational cost associated with higher injection pressures.

In summary, injection velocity, dynamic viscosity, and injection concentration collectively control the flow and displacement characteristics of CO₂ in porous media. Injection velocity primarily governs the driving force for front advancement, viscosity regulates flow resistance, and concentration controls solute transport behavior. This study integrates pore-scale numerical simulations and microfluidic experiments to elucidate, from a microscopic perspective, the flow mechanisms and key controlling factors of CO₂ miscible flooding, thereby providing a basis for understanding macroscopic displacement behavior and for model development. Meanwhile, we also recognize that extending pore-scale insights to reservoir-scale applications typically requires a multiscale framework and parameter upscaling. To this end, future work will focus on multiscale modeling and parameter transfer, using the key mechanisms identified at the pore scale to assist in determining critical parameters in reservoir-scale models. This will strengthen the linkage between mechanistic understanding and engineering application, and provide more targeted theoretical support for optimizing gas-injection development parameters.