Experimentally Constrained Kinetic Modeling of Long-Term Permeability and Injectivity Bounds in CO₂–Brine–Sandstone Systems

Abstract

Maintaining injectivity is essential for the success of CO₂-enhanced oil recovery (CO₂-EOR) and CO₂ storage. This study develops a simple, experimentally based framework to evaluate how sandstone reservoir properties evolve over time during CO₂ injection. Static aging experiments were conducted on Berea sandstone for exposure periods of up to three months using CO₂-saturated brine at representative reservoir conditions (150 °F and 1500 psi). Changes in rock and fluid properties were quantified using X-ray diffraction (XRD), aqueous chemistry (ICP-OES), and dry mass-loss measurements to quantify reaction progress. Dissolution kinetics were derived from these data and combined with a mass-balance porosity model and the Kozeny–Carman porosity–permeability relationship to project time-dependent porosity and permeability evolution. Short-term experimental results show limited alteration of the quartz-dominated sandstone framework over the duration of the laboratory tests, while the measured permeability decreased by 5.7% and 11.7% after 1 and 3 months of CO₂ exposure, respectively. However, kinetic projections indicate that slow dissolution of reactive carbonate cements can gradually enhance pore connectivity over longer timescales. Scenario-based projections over a 30-year injection period predict a potential permeability increase exceeding 140%, accompanied by a model-projected absolute porosity increase of approximately 5.5 percentage points; the permeability projection is contingent on the assumptions of the Kozeny–Carman relationship. The framework links laboratory geochemical measurements with long-term permeability assessment for CO₂-EOR and CO₂ storage.

1. Introduction

Geological CO₂ sequestration is increasingly viewed as a critical pathway for large-scale decarbonization, integrating subsurface engineering, reservoir management, and CO₂–brine–rock interaction science [1,2]. CO₂ migration and trapping depend on interacting flow, capillary, gravitational, and mass-transfer processes, which require appropriate representation in storage models [1,2,3]. Injectivity is a central design constraint for both CO₂-enhanced oil recovery (CO₂-EOR) and geological CO₂ storage. The ability to place large volumes of CO₂ into the subsurface over decades depends on how porosity and permeability evolve in response to coupled hydraulic, geochemical, and geomechanical processes. When CO₂ dissolves into formation brine, it acidifies the aqueous phase and promotes mineral reactions that can modify pore geometry and rock strength. The equilibrium solubility of CO₂ depends on temperature, pressure, and the concentrations and identities of dissolved salts, making brine composition an important consideration in evaluating CO₂–brine–rock interactions [4,5]. Where advection and convective mixing are weak, molecular diffusion becomes an important mechanism governing CO₂ transport through the aqueous phase [6]. These solubility and transport effects provide context for interpreting static-aging experiments and extrapolating their results to flowing reservoir systems. In sandstone and saline aquifer settings, mineral reactions are often subtle because the framework is quartz-dominated, yet even modest changes in pore connectivity can significantly influence injectivity over field time scales.

A growing body of experimental work has examined how CO₂-rich fluids alter the hydraulic properties of sandstone at the core scale, but the reported trends are not fully consistent. Al-Yaseri et al. [7] performed a series of coreflood experiments on Berea and Bandera Gray sandstones under reservoir conditions, cycling dead brine, live brine, and supercritical CO₂. They observed that permeability could either increase or decrease modestly depending on lithology and flow history, with Berea samples generally showing permeability increases after prolonged exposure to live brine, while the overall sensitivity to fluid type and injection rate remained limited. Chen et al. [8] injected CO₂-saturated synthetic brine through a quartz-rich Berea core at 60 °C and 20 MPa and reported a 1.8% increase in porosity and a 5.1% reduction in permeability. Tarokh et al. [9] investigated the poromechanical response of pristine and thermally damaged Berea sandstone subjected to liquid CO₂ injection, combining drained/undrained mechanical tests, ultrasonic velocities, acoustic emission monitoring, X-ray diffraction (XRD), and petrographic analysis. They found that CO₂ exposure decreased strength by 10–15% and increased porosity by approximately 10% and permeability by up to 100%, while poroelastic moduli changed only slightly. Collectively, these studies demonstrate that both permeability increases and decreases have been reported in sandstone during CO₂ coreflooding and that the sign and magnitude of change vary across rock types, mineralogies, and test conditions.

Ahmed et al. [10,11,12] contributed a series of studies examining how CO2-brine-rock interaction affects the geochemical, petrophysical, and multiphase-flow properties of sandstone. Ahmed et al. [10] performed static aging experiments on Berea sandstone at 150 °F and 1500 psi to evaluate how one month of CO₂ exposure impacts mineralogy, fluid chemistry, porosity, permeability, and mechanical properties. XRD and brine chemistry showed carbonate dissolution accompanied by halite precipitation, with minor reductions in porosity and permeability and minor stiffening of the rock framework. Ahmed et al. [11] then developed a data-driven framework for CO₂–brine relative permeability in sandstone reservoirs, combining laboratory measurements with machine-learning models trained on reservoir and fluid properties, and concluded that one-month CO₂ storage in quartz-dominated Berea under typical saline aquifer conditions did not measurably alter CO₂–brine relative permeability. Building on these petrophysical insights, Ahmed et al. [12] quantified the impact of CO₂ exposure on interfacial tension and rock properties relevant to saline aquifer storage, reporting interfacial tension, porosity, permeability, XRD mineralogy, and ICP fluid chemistry data after one month of CO₂ exposure at 150 °F and 1500 psi. These studies collectively indicate that over laboratory time scales (~30 days) and in quartz-rich, carbonate-cemented Berea sandstone, geochemical reactions are active but produce only modest changes in measured rock and fluid properties, and they did not explicitly translate those property changes into long-term injectivity forecasts.

More recently, Ahmed et al. [13] and Tale & Dindoruk [14,15] extended this work to longer storage times and coupled reactive-transport modeling. Ahmed et al. [13] evaluated the impact of CO₂ storage duration on Berea sandstone over 30- and 90-day static exposure periods at 150 °F and 1500 psi, incorporating a brine-only baseline core to isolate natural brine reactions from CO₂-driven reactions. By jointly analyzing porosity, permeability, mineralogy, fluid chemistry, and mechanical stability, they showed that carbonate dissolution and halite precipitation occurred over time and that only small increases in porosity and small decreases in permeability were observed over three months. In parallel, Tale and Dindoruk [14,15] developed and validated a coupled numerical–experimental framework for permeability evolution during CO₂-saturated brine injection into sandstone, integrating reactive transport, mineral kinetic laws, fines migration, and the Kozeny–Carman relationship. Their work showed that CO₂-saturated brine reduced permeability by up to 64% under certain mineralogical and injection-rate conditions. Together, these studies highlight that static CO₂ exposure can yield small changes in porosity and permeability over short laboratory durations, whereas dynamic CO₂–brine injection can yield substantial near-wellbore permeability changes under certain conditions. Similarly, Xu et al. [16] investigated coupled flow, mechanical, and geochemical processes during CO₂ injection, reporting that competing dissolution and precipitation reactions produce spatially and temporally variable porosity and permeability responses. At larger scales, Boison et al. [17] extended a geochemical model evaluated against laboratory flow-through experiments to a field-scale CO₂-storage model, predicting limited net porosity change because mineral dissolution and precipitation largely offset one another.

Despite this progress, two important gaps remain. First, most experimental studies—both from the literature and from our group—are limited to weeks to a few months of exposure, which is much shorter than the operational lifetimes of CO₂-EOR and storage projects. Second, while dynamic coreflood models such as those of Tale and Dindoruk [14,15] explicitly couple geochemistry and flow, they primarily resolve near-wellbore behavior over coreflood time scales (capturing the impact of faster reactions) rather than providing simple tools for long-term injectivity forecasting in sandstone reservoirs where geochemical reactions are slow but persistent. In this work, we address these gaps by (i) deriving reaction kinetics for carbonate-cement dissolution in Berea sandstone from carefully controlled static aging experiments at representative reservoir conditions, (ii) combining these kinetics with a mass-balance porosity model and the Kozeny–Carman porosity–permeability relationship project porosity and permeability evolution over longer durations, and (iii) embedding these property trends into injectivity calculations relevant to CO₂-EOR and storage. The resulting framework provides an experimentally constrained method to bracket long-term injectivity evolution in carbonate-cemented sandstones, complementing both short-term coreflood observations and fully coupled reactive-transport simulations.

2. Materials and Methods

2.1. Rock Material and Aging Conditions

Berea sandstone cores were used due to their well-characterized mineralogy and relevance as a sandstone reservoir analog. Static aging experiments were conducted using CO₂-saturated brine at representative reservoir conditions of 150 °F and 1500 psi. Exposure periods of 1 and 3 months were investigated, along with a brine-only control test.

Five sister cores of gray Berea sandstone were obtained from the same block, and three homogeneous cores were selected by X-ray CT, one for each aging condition: brine only for 1 month (A), CO₂–brine for 1 month (B), and CO₂–brine for 3 months (C). Each selected 1.5-in-diameter core was sectioned into pieces for different post-aging measurements, including a 0.25-in disc for wettability characterization, a 2.5-in plug for routine core analysis RCA and NMR, and an approximately 8-in section reserved for relative-permeability testing. The sections exposed under each condition had a combined bulk volume of approximately 311 cm³ (Table 1). Pre-aging XRD showed quartz-dominated mineralogy with minor feldspars, clays, and carbonate cements; total carbonate content ranged from 4.3 to 6.8 wt.% among the three tested cores. The samples were vacuum-saturated for 38 h with synthetic brine (100,000 ppm; 80 wt.% NaCl and 20 wt.% KCl), pressurized to 2000 psi for 24 h, and aged under static conditions for 30 or 90 days. For the CO₂ tests, CO₂ was introduced into a 1-in headspace above the brine to 1500 psi, and the system was maintained at 150 °F and 1500 psi. Diffusive CO₂ saturation of the brine was estimated to require ~10 h [18], which is negligible relative to the 30- and 90-day aging periods, and the CO₂ gas cap maintained saturation throughout aging. Before and after aging, the rock samples were oven-dried at 110 °C for 24 h and weighed in triplicate (±0.1 g); mass loss was determined from the difference between the pre- and post-aging dry masses. Porosity was measured on the dedicated RCA plug using the Boyle’s-law method, and Klinkenberg-corrected permeability was obtained using the unsteady-state method at net confining stress. Details of the experimental setup and procedures for the CO₂ static exposure test were reported in our previous publications [10,11,13]. Static conditions were intentionally selected to isolate intrinsic geochemical reactions from flow-induced effects such as fines migration, capillary end effects, and salt precipitation associated with CO₂ dry-out.

Table 1. Induced property changes in Berea sandstone during static aging experiments (150 °F, 1500 psi).

Exposure Condition	Duration (Months)	Mass Loss (%)	Initial Porosity (%)	Initial Permeability (mD)	Total Aged-Rock Bulk Volume (cm3)	Porosity Change (Absolute, %)	Permeability Change
Brine only (control)	1	0.13	19.3	174	311.02	+0.77	(−6.9%)
CO2–brine	1	0.24	19.6	211	311.28	+0.17	(−5.7%)
CO2–brine	3	0.62	19.6	214	311.24	+0.59	(−11.7%)

2.2. Analytical Measurements

Reaction progress was quantified using:

•

Dry mass loss to track cumulative mineral dissolution;

•

X-ray diffraction (XRD) to identify mineralogical changes;

•

Aqueous geochemistry (ICP-OES and titration) to quantify dissolved species;

•

Routine porosity and permeability measurements before and after exposure.

These measurements provide the experimental basis for kinetic modeling. Details of the static CO₂ exposure experiments, analytical techniques XRD, ICP-OES, and porosity–permeability characterization are provided in Ahmed et al. [13].

2.3. Kinetic Model for Mineral Dissolution

Mineral dissolution was modeled as a first-order depletion process acting on a finite pool of reactive minerals as in Equation (1):

$${\displaystyle \frac{d{M}_{\mathit{rem}}}{\mathit{dt}}}=-k_d{M}_{\mathit{rem}}$$

(1)

where M_rem is the remaining reactive mineral mass, and k_d is the apparent rate constant. The cumulative dissolved mass is W(t):

$$W(t)=M_0\left(1−e^{-k_{d}t}\right)$$

(2)

Model parameters were calibrated using experimentally measured mass loss at 1 and 3 months.

Model parameters were estimated from the experimentally measured mass losses at 1 and 3 months. For the short-term model, M₀ and k_d were both fitted to these measurements using the finite-pool first-order model. For the long-term bounding scenario, the maximum reactive inventory was constrained using the XRD-derived total carbonate fraction and the initial core mass, and k_d,LT was fitted to the same mass-loss data with this inventory fixed.

One independent core was tested for each aging condition (n = 1). Pre- and post-aging dry masses were each measured in triplicate, with identical readings at the balance resolution; the balance accuracy was ±0.1 g. Because M₀ and k_d were determined from two experimental observations, a regression-based confidence interval could not be estimated. Propagation of the balance accuracy gives k_d = 0.167 ± 0.107 month⁻¹. For the long-term bounding scenario, the 6.8 wt.% pre-aging carbonate content of the 3-month core corresponds to M₀,_LT = 45.31 g. Using the observed sister-core variability in total carbonate content (SD = 1.27 wt.%) as the inventory uncertainty gives k_d,LT = 0.0321 ± 0.0063 month⁻¹.

2.4. Porosity Evolution

Porosity evolution was computed by converting dissolved mineral mass into added pore volume:

$$\varphi (t)={\varphi }_0+{\displaystyle \frac{W\left(t\right)}{{\rho }_s{V}_{\mathit{bulk}}}}$$

(3)

where ρ_s is the grain density, ∅₀ is the initial porosity pre-aging, V_bulk is the bulk rock volume, and W(t) is the total weight dissolved at time t.

Equation (3) assumes that the experimentally measured net solid mass loss is converted to additional pore volume using grain density. Secondary precipitation and spatial redistribution of reaction products are not represented explicitly and are reflected only through their net contribution to the measured pre- and post-aging rock mass.

2.5. Permeability Evolution via Kozeny–Carman

Permeability evolution was obtained by mapping the reaction-driven porosity changes into permeability space using the Carman–Kozeny (KC) relationship. The KC framework assumes a homogeneous pore structure with uniform enlargement of pore bodies and throats, such that permeability is governed by pore volume, pore surface area, and tortuosity. The general KC form for consolidated sandstones may be expressed as:

$$k\left(t\right)={\displaystyle \frac{\varphi (t{)}^3}{C\hspace{0.17em}{S}_v^2\hspace{0.17em}[1-\varphi (t){]}^2}}$$

(4)

where ϕ(t) is porosity, S_v is the specific internal surface area per unit pore volume, and C is a geometric/tortuosity constant. Under the assumption that the pore-shape constant and specific surface area scale proportionally with the initial state, the KC expression can be recast in normalized form as:

$$k(t)=k_0{\left({\displaystyle \frac{\varphi \left(t\right)}{{\varphi }_0}}\right)}^3{\left({\displaystyle \frac{1-{\varphi }_0}{1-\varphi \left(t\right)}}\right)}^2$$

(5)

where k₀ and ϕ₀ are the initial permeability and porosity, respectively. This normalized formulation preserves the geometric coefficients without requiring explicit calibration of C or S_v It is widely used for consolidated sandstones under dissolution-driven pore enlargement.

The KC model provides a direct functional link between evolving porosity and permeability:

k(t) = f(ϕ(t))

(6)

and therefore enables long-term permeability prediction once ϕ(t) is supplied from the reactive dissolution model described in Section 2.4. This approach represents an upper-bound (optimistic) scenario in which dissolution uniformly enlarges flow pathways without fines mobilization or secondary precipitation.

2.6. Time-Dependent Injectivity Formulation

Injectivity was evaluated using the injectivity index (II), defined as the ratio of injection rate to the pressure differential between the wellbore and the reservoir. Under radial flow assumptions, injectivity can be expressed as:

$$\mathit{II}={\displaystyle \frac{q}{\Delta p}}={\displaystyle \frac{2\mathit{\pi }\mathit{kh}}{\mathit{\mu }\mathit{B}\hspace{0.17em}\left[\ln \right(r_e/r_w)+s]}}$$

(7)

where q is the injection rate, Δp is the pressure difference between bottom-hole and reservoir pressure, k is permeability, h is formation thickness, μ is fluid viscosity, B is the formation volume factor, r_e and r_w are the drainage and wellbore radii, respectively, and s is the skin factor. For a given well geometry and fluid system, injectivity scales linearly with permeability.

Substituting the time-dependent permeability function yields:

$$\mathit{II}(t)={\displaystyle \frac{2\mathit{\pi }\mathit{h}}{\mathit{\mu }\mathit{B}\left[\ln \left(\frac{r_e}{r_w}\right)+s\right]}}\cdot k(t)$$

(8)

This formulation transforms injectivity from a static reservoir assumption into a time-dependent, reaction-driven prediction, directly linking laboratory-measured geochemical kinetics to field-scale injectivity evolution.

Because the present study employed static aging experiments, injectivity was not measured directly through flow tests. Instead, changes in injectivity were inferred from experimentally derived permeability evolution. This approach isolates the intrinsic geochemical contribution to injectivity by excluding flow-induced effects such as fines migration, salt precipitation, and capillary end effects. Under these conditions, the injectivity ratio (IR) can be approximated as:

$$\mathit{IR}(t)={\displaystyle \frac{\mathit{II}\left(t\right)}{I{I}_0}}\approx {\displaystyle \frac{k\left(t\right)}{k_0}}$$

(9)

where k₀ and k(t) are the initial and time-dependent permeabilities, respectively. Using this relationship, reaction-driven permeability changes predicted by the kinetic model were directly translated into time-dependent injectivity trends.

3. Results and Discussion

3.1. Short-Term Rock Property Changes During Static Aging

Table 1 details the petrophysical evolution of Berea sandstone after exposure to reservoir conditions (150 °F, 1500 psi). The data reveals a clear decoupling between porosity and permeability evolution. Across all conditions, the rock lost mass and gained porosity, indicating mineral–fluid interaction. Notably, the brine-only control showed the largest porosity increase (+0.77 percentage points), demonstrating that porosity change alone cannot be attributed specifically to CO₂. The control also showed a small mass loss (0.13 wt.%), whereas the CO₂-exposed samples showed larger mass losses of 0.24 and 0.62 wt.%. The mineralogical and fluid-chemistry results reported for these experiments in our companion study [13] further support enhanced mineral dissolution under CO₂ exposure.

Crucially, bulk dissolution did not translate to improved hydraulic conductivity. Despite the increase in pore volume, Klinkenberg permeability declined across all samples. For the 3-month CO₂ exposure, permeability dropped from 214 mD to 189 mD (−11.7%). This counter-intuitive behavior—where porosity increases while permeability decreases— may reflect localized pore-throat obstruction or grain redistribution. A similar response was reported during CO₂-saturated brine injection into Berea sandstone, where fines movement following cement dissolution and/or compaction were proposed as possible explanations [8]. However, those flow-through conditions differ from the present static-aging experiments, and the permeability decrease in the brine-only control precludes attributing the entire change specifically to CO₂.

Our mineralogical and fluid chemistry analyses provide evidence consistent with net mineral dissolution. As detailed in our companion study [13], XRD analysis confirmed the complete disappearance of siderite and a reduction in chlorite after just one month. Concurrently, ICP-OES data showed a sharp spike in Ca²⁺ and Fe²⁺ concentrations. These observations support the dissolution of reactive mineral phases. Localized secondary precipitation or fines redistribution may nevertheless occur within the pore network and could contribute to pore-throat obstruction; however, these mechanisms were not directly quantified in the present study.

3.2. Implication

Pre- and post-aging MICP measurements from the same sister-core experimental program showed condition-dependent changes in threshold capillary pressure, with no systematic trend toward pore-throat constriction with exposure time; notably, the 3-month core showed the largest permeability decrease despite a lower threshold pressure [19]. The specific pore-scale mechanism responsible for the measured permeability decrease remains unresolved.

Figure 1 tracks the cumulative dissolved mass during CO₂–brine aging in grams. The reaction trajectory follows a decelerating first-order trend, which physically corresponds to the rapid depletion of the finite reactive mineral inventory. As noted in the XRD analysis, the highly reactive siderite and calcite phases were largely consumed within the first months [13], leaving the slower-reacting silicate framework (quartz and feldspar) to dominate the later-stage kinetics. Fitting the measured mass losses of 1.60 g at 1 month and 4.10 g at 3 months gives M₀ = 10.4 g (~1.56 wt.%) and k_d = 0.167 month⁻¹, corresponding to a half-life of 4.15 months. The fitted reactive inventory is smaller than the total carbonate content estimated form XRD wt.% and total core mass, indicating that only an effective fraction of the total carbonate inventory reacted under the closed-system laboratory conditions. This decelerating trend aligns with standard rate laws [20] but requires careful extrapolation. Because the reactive carbonate content in Berea is low (~5–7 wt.%), the reaction is self-limiting. Once these cements are exhausted, the dissolution rate will drop by orders of magnitude as it shifts to the sluggish kinetics of aluminosilicates.

Figure 1. Cumulative mineral dissolution during static CO₂–brine aging of Berea sandstone. Symbols represent experimentally measured mass loss of 1.60 g (0.24 wt.%) at 1 month and 4.10 g (0.62 wt.%) at 3 months. The solid curve shows the short-term first-order fit with M₀ = 10.4 g and k_d = 0.167 month⁻¹. The dotted horizontal line indicates the 10.4 g asymptotic dissolution capacity, and the dashed vertical line marks the 4.15-month half-life. For the long-term bounding scenario, the XRD-derived carbonate fraction (~6.7 wt.%) and the initial core mass (~665–670 g) give a maximum reactive inventory of approximately 45 g. This finite carbonate pool is assumed to be consumed progressively during prolonged CO₂–brine exposure. With M_0,LT fixed at 45 g, fitting the same mass-loss data gives an effective k_d,LT ≈ 0.032 month⁻¹.

Figure 1. Cumulative mineral dissolution during static CO₂–brine aging of Berea sandstone. Symbols represent experimentally measured mass loss of 1.60 g (0.24 wt.%) at 1 month and 4.10 g (0.62 wt.%) at 3 months. The solid curve shows the short-term first-order fit with M₀ = 10.4 g and k_d = 0.167 month⁻¹. The dotted horizontal line indicates the 10.4 g asymptotic dissolution capacity, and the dashed vertical line marks the 4.15-month half-life. For the long-term bounding scenario, the XRD-derived carbonate fraction (~6.7 wt.%) and the initial core mass (~665–670 g) give a maximum reactive inventory of approximately 45 g. This finite carbonate pool is assumed to be consumed progressively during prolonged CO₂–brine exposure. With M_0,LT fixed at 45 g, fitting the same mass-loss data gives an effective k_d,LT ≈ 0.032 month⁻¹.

Figure 2a shows long-term porosity evolution based on the kinetic model. The predicted rapid early porosity increase followed by a plateau reflects progressive depletion of carbonate cements and the very slow dissolution rates of silicates at 150 °F. To verify that the analytical porosity evolution function used in Figure 2a is consistent with laboratory observations, the model was evaluated at the CO₂–brine exposure durations (1 and 3 months) and compared against the measured porosity increases obtained from Routine Core Analysis (RCA) on the aged Berea samples. The model predicts porosities of 19.79% and 20.10% after 1 and 3 months, compared with measured RCA values of 19.77% and 20.19%, respectively (Table 1). The absolute differences (0.02 and 0.09 percentage points) are within the estimated RCA measurement uncertainty (±0.2–0.3 percentage points). Because the kinetic parameters were calibrated using mass losses at the same two times, this agreement indicates consistency with the available measurements rather than statistical validation. The porosity function is therefore used as a mass-balance-consistent input for the subsequent permeability and injectivity projections, which extend beyond the experimental window.

Figure 2. (a) Analytical porosity evolution function plotted over 0–30 years and (b) comparison of the model with measured porosity increases from CO₂–brine aging experiments at 1 and 3 months. The analytical curve reproduces the observed porosity gains within experimental uncertainty, validating its use for forward permeability and injectivity projections that extend beyond the laboratory exposure window.

Figure 2. (a) Analytical porosity evolution function plotted over 0–30 years and (b) comparison of the model with measured porosity increases from CO₂–brine aging experiments at 1 and 3 months. The analytical curve reproduces the observed porosity gains within experimental uncertainty, validating its use for forward permeability and injectivity projections that extend beyond the laboratory exposure window.

Figure 3 shows estimated permeability based on the Kozeny–Carman relationship, which relates permeability to porosity and internal surface area for granular media. The Kozeny–Carman formulation is widely used in reactive flow simulations to approximate permeability evolution under homogeneous pore-structure assumptions [21]. Under these assumptions, dissolution-induced porosity increases yield permeability enhancement.

3.3. Kozeny–Carman Projection and Measured Permeability Response

The Kozeny–Carman scenario disagrees with the measured permeability decreases in Table 1. At 1 and 3 months, the Kozeny–Carman projection gives k/k₀ = 1.03 and 1.09, respectively, whereas the measured CO₂–brine values are 0.94 and 0.88 (Table 1; Figure 3). This contrast confirms that the Kozeny–Carman curve represents a mechanistic upper-bound scenario rather than an extrapolation of the measured permeability response. This discrepancy is consistent with CCUS laboratory observations showing that permeability may either:

•

Increase due to connectivity enhancement [22,23];

•

Decrease due to pore-throat blocking, fines migration, or secondary precipitation [24,25], depending on prevailing pore-scale mechanisms.

Therefore, Figure 3 should be interpreted as a mechanistic upper-bound scenario rather than an extrapolation of measured permeability.

Figure 3. Projected normalized permeability (k/k₀) of Berea sandstone under long-term CO₂–brine exposure, derived from reaction-driven porosity changes using the Kozeny–Carman relationship (Equations (3) and (5)) with the long-term scenario parameters (M₀,LT = 45.3 g; k_d,LT = 0.0321 month⁻¹), plotted on a logarithmic time axis. The shaded region marks the experimental window. Symbols show the measured RCA permeability ratios from Table 1 (filled: CO₂–brine, Cores B and C; open: brine-only control, Core A). The measured decreases contrast with the predicted increase, so the curve represents a mechanistic upper-bound scenario rather than an extrapolation of the measured permeability.

Figure 3. Projected normalized permeability (k/k₀) of Berea sandstone under long-term CO₂–brine exposure, derived from reaction-driven porosity changes using the Kozeny–Carman relationship (Equations (3) and (5)) with the long-term scenario parameters (M₀,LT = 45.3 g; k_d,LT = 0.0321 month⁻¹), plotted on a logarithmic time axis. The shaded region marks the experimental window. Symbols show the measured RCA permeability ratios from Table 1 (filled: CO₂–brine, Cores B and C; open: brine-only control, Core A). The measured decreases contrast with the predicted increase, so the curve represents a mechanistic upper-bound scenario rather than an extrapolation of the measured permeability.

Figure 4 shows the projected evolution of the normalized injectivity index, IR (II/II₀), over a 30-year operating horizon. The trajectory is obtained by coupling experimentally derived first-order dissolution kinetics for carbonate cements with a Kozeny–Carman porosity–permeability relationship and embedding the resulting k(t) function into a radial well model following Peaceman [26]. This workflow is consistent with standard geochemical and reactive-transport practice in which Arrhenius-type rate laws are used for carbonate dissolution, reactive surface area controls are applied, and porosity changes are mapped to permeability via Kozeny–Carman [20,27,28]. The model response exhibits an early-time enhancement regime followed by late-time stabilization. The initial enhancement is associated with the rapid dissolution of the small but reactive carbonate-cement fraction measured in the Berea sandstone specimens, which increases pore connectivity and reduces flow resistance near the wellbore. Similar observations have been documented in laboratory studies, where dissolution of carbonate cements in sandstone under CO₂-acidified brines resulted in measurable increases in porosity and permeability [14,15]. This aligns with laboratory characterizations showing that carbonate dissolution rates are initially high when reactive surface area and acid availability are both large [20,27]. As dissolution progresses, the system transitions to a reactant-limited regime in which the carbonate inventory dictates the remaining reaction extent. Under these conditions, permeability increments decrease and the injectivity index II approaches an asymptotic value over long times. This type of asymptotic behavior is consistent with reactive-transport simulations of CO₂–rock systems, which show that porosity and permeability evolve rapidly at early times and then approach steady values as dissolution fronts migrate and geochemical equilibrium is approached [27,28].

Figure 4. Normalized injectivity evolution under long-term CO₂–brine exposure, derived by embedding reaction-driven permeability evolution into the radial flow injectivity formulation. Injectivity increases rapidly during the early years of CO₂ exposure and stabilizes as reactive carbonate minerals are depleted.

Figure 4. Normalized injectivity evolution under long-term CO₂–brine exposure, derived by embedding reaction-driven permeability evolution into the radial flow injectivity formulation. Injectivity increases rapidly during the early years of CO₂ exposure and stabilizes as reactive carbonate minerals are depleted.

The injectivity projections presented in Figure 5 in this study are derived from static batch aging experiments conducted over limited durations (up to three months) and therefore do not capture flow-induced mechanisms such as advective transport, dynamic fines migration, or near-wellbore salt precipitation that may occur under field injection conditions. The reaction-driven scenario assumes that dissolution-induced porosity gains translate effectively into permeability enhancement at the Darcy scale, which represents an upper-bound response in well-swept, brine-saturated zones. In contrast, the measured-trend scenario incorporates the permeability decline observed during short-term static CO₂–brine exposure experiments and extrapolates this behavior over a 30-year horizon to define a conservative lower bound on long-term injectivity. The measured-trend lower bound assumes a first-order exponential decline, II(t)/II₀ = k(t)/k₀ = exp(−λt), calibrated to the 3-month measurement (k/k₀ = 0.883; Table 1), giving λ = 0.50 yr⁻¹ (0.0414 month⁻¹). Because this form asymptotically approaches zero, it represents a hypothetical unmitigated lower bound rather than an expected field response. This extension assumes a closed system with no brine renewal, no advective flushing of reaction products, and no operational intervention, and therefore should not be interpreted as a literal field forecast. As such, the results should be interpreted as bounding scenarios rather than deterministic forecasts. Extension of this framework to specific field applications requires calibration against dynamic coreflood experiments, well-test data, or coupled reactive-transport simulations. This lower-bound trajectory represents an unmitigated decline scenario. In practical field operations, such a decline would typically trigger well stimulation treatments (e.g., acidizing or hydraulic fracturing) to restore injectivity long before the asymptotic low is reached.

Figure 5. Injectivity bounding scenarios derived from static CO₂–brine aging experiments: the reaction-driven scenario (upper bound) assumes that permeability enhancement driven by dissolution-induced porosity increases, whereas the measured-trend scenario (lower bound) incorporates experimentally observed early-time permeability reductions. Model inputs used for Figure 2, Figure 3, Figure 4 and Figure 5: φ₀ = 0.193, k₀ = 200 mD, ρ_s = 2.65 g/cm³, and Vbulk = 311 cm³. The short-term kinetic model uses M₀ = 10.4 g; the long-term bounding scenario uses M_0,LT = 45.31 g.

Figure 5. Injectivity bounding scenarios derived from static CO₂–brine aging experiments: the reaction-driven scenario (upper bound) assumes that permeability enhancement driven by dissolution-induced porosity increases, whereas the measured-trend scenario (lower bound) incorporates experimentally observed early-time permeability reductions. Model inputs used for Figure 2, Figure 3, Figure 4 and Figure 5: φ₀ = 0.193, k₀ = 200 mD, ρ_s = 2.65 g/cm³, and Vbulk = 311 cm³. The short-term kinetic model uses M₀ = 10.4 g; the long-term bounding scenario uses M_0,LT = 45.31 g.

3.4. Injectivity Bounds and Limitations

The reaction-driven and measured-trend scenarios represent bounding cases rather than deterministic field forecasts. The reaction-driven upper bound assumes that dissolution-driven porosity gains translate into permeability enhancement through the Kozeny–Carman relationship, with no fines mobilization or secondary precipitation; it represents the potential for matrix enhancement in well-swept zones. The measured-trend lower bound extrapolates the experimentally observed permeability decline under closed-system conditions, with no brine renewal, no advective flushing of reaction products, and no operational intervention. Because its exponential form asymptotically approaches zero, it represents a hypothetical unmitigated damage end member rather than an expected field response. In practice, such a decline would typically trigger stimulation treatments (e.g., acidizing or hydraulic fracturing) long before the asymptotic low is reached. The injectivity calculation also assumes single-phase flow with constant viscosity, formation volume factor, thickness, drainage and wellbore radii, and skin factor, so that injectivity changes reflect permeability changes only; these simplifications may not hold in the near-wellbore region during CO₂ injection.

Static aging may favor retention of fines or reaction products because hydrodynamic drag forces are absent; under flowing conditions, mobilization and redistribution may shift the response toward the upper-bound scenario. The static aging experiments also do not capture flow-induced processes such as advective transport, dynamic fines migration, CO₂ dry-out and salt precipitation, relative-permeability effects, or near-wellbore skin evolution. Field-scale injectivity may therefore evolve between these end members depending on mineralogy, flow regime, stress state, operational conditions, and completion design. Application to specific reservoirs requires calibration against dynamic coreflood experiments, well-test data, or coupled reactive-transport simulations.

4. Conclusions

• Measurable Mineral Dissolution: Static CO₂–brine exposure produced cumulative mass loss increasing with time (0.24% at 1 month and 0.62% at 3 months). These trends are consistent with the dissolution of minor carbonate accessory minerals identified by XRD, confirming active geochemical alteration even in quartz-dominated lithologies.
• Porosity Enhancement: Modest absolute porosity gains (up to +0.59% after three months) were observed under CO₂–brine conditions. This confirms that dissolution successfully generates additional pore volume, consistent with carbonic-acid-driven dissolution of reactive carbonate cement.
• Transport-Limited Permeability Decline: Despite porosity gains, Klinkenberg permeability decreased across all static cases (e.g., −11.7% after 3 months). Complementary MICP measurements showed condition-dependent pore-throat changes but no systematic constriction with exposure time [14]. The specific pore-scale mechanism responsible for the permeability decrease remains unresolved.
• Kinetic Characterization: The short-term mass-loss data were represented by a first-order depletion model with M₀ = 10.4 g (~1.56 wt.%) and k_d = 0.167 month⁻¹ (t½ = 4.15 months). For the separate long-term bounding scenario, the reactive inventory was set to approximately 45.3 g (~6.7 wt.%) based on the XRD-derived carbonate content, giving an effective k_d,LT ≈ 0.032 month⁻¹. This model captures the decelerating reaction trajectory associated with the progressive depletion of finite reactive carbonate phases.
• Long-Term Projections: Upscaling these kinetics using Kozeny–Carman and radial-flow formulations indicates that, under homogeneous flow assumptions, permeability and injectivity will increase rapidly at early times before stabilizing as carbonates are depleted.
• Performance Envelope: The study defines a bounding performance envelope. The projected increasing trends represent a geochemical upper bound (intrinsic dissolution-driven behavior), whereas the measured static decline represents a conservative lower bound (transport-limited damage). Field-scale injectivity will likely evolve between these limits, depending on the combined effects of mineral reactions, transport processes, flow conditions, and operational factors.