Cost-Effective Strategies for Assessing CO₂ Water-Alternating-Gas (WAG) Injection for Enhanced Oil Recovery (EOR) in a Heterogeneous Reservoir

Abstract

This study evaluates the feasibility of CO₂ Water-Alternating-Gas (WAG) injection for enhanced oil recovery (EOR) in a highly heterogeneous reservoir using cost-effective and efficient tools. The Rule of Thumb method was initially used to screen the reservoir, confirming its suitability for CO₂-WAG injection. A fluid model was constructed by comparing several component lumping methods, selecting the approach with the least deviation from experimental data to ensure accuracy. The minimum miscibility pressure (MMP), a critical parameter for CO₂-EOR, was estimated using three methodologies: 1D simulation based on the slim tube test, semi-empirical analytical correlations, and fluid modeling. These techniques provided complementary insights into the reservoir’s miscibility conditions. The CO₂ Prophet software version 1 was employed to history-match production data and evaluate different development strategies. The Kinder Morgan CO₂ Scoping Model was used to perform production forecasting and assess the economic viability of implementing CO₂-WAG. Quantitative comparisons showed that the CO₂ Prophet version 1 model revealed minimal deviations from the history match results: oil production estimates differed by only 3.5%, and water production estimates differed by −4.11%. Cumulative oil recovery was projected to reach approximately 20.26 MMSTB over a 25-year production period. The results indicate that CO₂-WAG injection could enhance oil recovery significantly compared to water flooding while maintaining economic feasibility. This study demonstrates the practical integration of analytical tools and inexpensive models to evaluate and optimize CO₂-EOR strategies in complex reservoirs. The findings provide a systematic workflow for deploying CO₂-WAG in heterogeneous reservoirs, balancing technical and economic considerations.

1. Introduction

Global warming, primarily driven by increasing atmospheric greenhouse gases, with carbon dioxide (CO₂) being the most significant contributor, has emerged as one of the most pressing environmental challenges of our time. The rise in CO₂ emissions is largely attributed to the combustion of fossil fuels, a consequence of growing global energy demands fueled by rapid population growth and industrialization [1,2]. If left unchecked, the rising concentration of atmospheric CO₂ threatens to exacerbate climate change, leading to severe ecological and socio-economic consequences. As such, the development of effective mitigation strategies is essential to address this global crisis. Among the various approaches to reducing CO₂ emissions, Carbon Capture and Storage (CCS) has emerged as a promising solution. CCS technologies aim to capture CO₂ from industrial processes and store it securely in geological formations, thereby preventing its release into the atmosphere [3]. Geological storage, in particular, has gained significant attention due to its potential for large-scale CO₂ containment over extended periods [4].

The direct injection of flue gas into gas hydrate reservoirs has been identified as an effective method to enhance CO₂ capture efficiency while simultaneously recovering methane from existing gas hydrate deposits. Studies by Hassanpouryouzband and his team [5] highlight the high efficiency of flue gas hydrate formation in both frozen and unfrozen sediments, with CO₂ capture rates exceeding 92% under certain conditions. This method offers a dual benefit: reducing the costs associated with CO₂ separation and storing CO₂ in a stable hydrate phase. The kinetics of hydrate formation, particularly in mesoporous media, have been thoroughly investigated, revealing the important role of pressure and temperature in optimizing CO₂ capture [6].

One of the most notable applications of geological CO₂ storage is its integration with enhanced oil recovery (EOR) techniques, commonly referred to as CO₂-EOR. In this process, CO₂ is injected into depleted hydrocarbon reservoirs to mobilize residual oil while simultaneously sequestering CO₂ within the subsurface [7,8]. This dual-purpose approach not only enhances hydrocarbon recovery but also aligns with climate change mitigation goals by reducing atmospheric CO₂ concentrations [9]. However, the success of CO₂-EOR projects depends on numerous technical and economic factors, including reservoir pressure, minimum miscibility pressure (MMP), fluid properties, and the cost-effectiveness of injection operations [10,11]. Recent research highlights the dual potential of CO₂-EOR to advance hydrocarbon production and contribute to global decarbonization objectives. A notable case study by Blecich et al. [12] demonstrated Croatia’s pioneering CO₂-EOR initiative, which successfully repurposed CO₂ captured from a natural gas processing facility for injection into mature oil reservoirs, simultaneously boosting production and sequestering emissions. Expanding on this synergy, Núñez-López and Moskal [13] emphasized CO₂-EOR’s capacity to transition toward negative emissions when integrated with permanent geological CO₂ storage.

The concept of minimum miscibility pressure (MMP) is particularly critical in the design and optimization of CO₂-EOR projects. MMP refers to the minimum pressure required for the injected CO₂ and reservoir oil to achieve complete miscibility, resulting in a single-phase mixture that maximizes oil displacement efficiency. Accurate estimation of MMP is essential for determining the feasibility of gas flooding operations and optimizing both injection strategies and surface facility designs. Traditional laboratory methods, such as the slim tube test and rising bubble apparatus, have long been regarded as the gold standard for MMP determination [14]. While these methods provide high accuracy, they are often time-consuming, resource-intensive, and impractical for rapid assessments or small-scale projects.

To overcome the limitations of traditional laboratory techniques, researchers have developed computational methods for MMP estimation. These methods are typically categorized into semi-empirical correlations and advanced analytical models, which leverage reservoir fluid characteristics to predict MMP with varying degrees of precision [15]. Semi-empirical correlations offer a cost-effective and expedient alternative to laboratory tests but are often constrained by the specific assumptions and conditions under which they were developed [16]. Consequently, their application to diverse reservoir settings can result in significant uncertainties [17].

One of the widely recognized tools for assessing the feasibility of CO₂-EOR is the Kinder Morgan CO₂ Scoping Model [18]. This model provides a comprehensive framework for preliminary feasibility analysis by estimating recoverable oil volumes and evaluating the economic viability of CO₂ injection projects. Through extensive industry experience, Kinder Morgan developed the Kinder Morgan Scoping Model [18], which incorporates reservoir-specific input parameters, such as oil and gas properties, reservoir pressure, and injection costs, to deliver a robust screening tool for potential CO₂-EOR projects [19]. The model’s ability to balance simplicity with predictive accuracy makes it particularly valuable during the initial stages of project planning, allowing operators to identify viable candidates for detailed simulation studies. While the Kinder Morgan CO₂ Scoping Model [18] is less computationally intensive than advanced compositional models, it remains an effective tool for bridging the gap between basic analog methods and complex reservoir simulations, providing critical insights that align technical feasibility with economic considerations [15].

Similarly, the CO₂ Prophet version 1 model [20], developed with the support of the U.S. Department of Energy, offers a mid-level simulation capability that bridges the gap between basic screening tools and full-scale compositional models. This model is designed to predict well performance and fluid flow dynamics, incorporating various reservoir-specific parameters such as injection rates, fluid saturation, and well pattern descriptions. The CO₂ Prophet version 1 model [20] is particularly useful for evaluating the outcomes of waterflood and CO₂-EOR projects, providing reliable estimates of recoverable oil volumes and CO₂ storage potential. Its relatively moderate computational demands and robust predictive capability make it a valuable tool for operators seeking to perform detailed feasibility assessments without the resource intensity of compositional models. Additionally, the CO₂ Prophet version 1 model [20] has been instrumental in enhancing the understanding of subsurface dynamics and optimizing CO₂ injection strategies, thereby improving the overall efficiency and economic viability of CO₂-EOR projects [21].

Even though CO₂-EOR has numerous benefits, small-scale processes such as viscous fingering and molecular diffusion can have a big impact on how well the oil is swept and recovered. Viscous fingering happens when the injected CO₂, which is much thinner than the oil in the reservoir, moves through unevenly as it creates unstable flow paths. This can lead to inefficient displacement and early gas breakthrough of the injected CO₂ at the production well, resulting in reduced overall recovery. On the other hand, molecular diffusion helps CO₂ mix with the oil more effectively over time, helping to improve miscibility and the displacement of the oil.

Despite the growing body of research on CO₂-based enhanced oil recovery (EOR), a critical gap remains in developing simple, cost-effective, and less time-consuming tools for evaluating CO₂-EOR feasibility. Most existing studies rely on commercial reservoir simulation software and thorough experimental work, which, although robust, can be prohibitively expensive and require significant computational time and resources. This creates a barrier for operators with limited budgets or smaller-scale projects. Addressing this gap by offering a more accessible approach to minimum miscibility pressure (MMP) estimation and feasibility analysis can greatly facilitate the wider adoption of CO₂-EOR and support both economic and environmental objectives.

This work presents an approach for assessing the feasibility of CO₂-EOR using innovative and cost-efficient methodologies. By systematically comparing various component lumping techniques, semi-empirical correlations, and conventional simulation approaches, we identify a reliable and economical method for determining minimum miscibility pressure (MMP) while evaluating the overall feasibility of CO₂-EOR from technical, economic, and environmental perspectives. This integrated multidisciplinary framework provides a practical foundation for optimizing CO₂-EOR strategies, ultimately advancing enhanced hydrocarbon recovery and promoting sustainable practices.

2. Materials and Methods

2.1. Rule of Thumb

Based on the Rule of Thumb, this reservoir is a good candidate for CO₂ flooding. The Rule of Thumb states that a successful waterflooding project will likely result in a successful CO₂ flood. This analogy is mostly valid because the waterflooding process usually builds up the reservoir pressure, which contributes significantly to attaining the MMP. In [22], the field successfully underwent a waterflooding process, yielding 29.4 MMSTB by converting three pressure monitoring wells and three of the highest water production wells into water injection wells. This confirms the effectiveness of the waterflooding and hence makes a successful CO₂ flood process highly probable. The current initial oil saturation of the field is about 0.52. This fraction indicates that a significant amount of oil remains, making a CO₂ flood a good option to consider, assuming the economic factors influencing the CO₂ flood are favorable.

The field under study is at a depth of about 8000 ft, which is higher than the acceptable depth (2500 ft) for CO₂ flooding. At this depth, higher reservoir temperature and pressure are more likely as these properties increase with increasing depth. Higher reservoir temperature ensures higher oil API gravity due to the thermal cracking of heavier oil components into intermediate and light components. CO₂ miscibility is favored by the presence of intermediate-molecular-weight hydrocarbons, i.e., C1-C12, whilst heavier-molecular-weight hydrocarbons increase MMP, thus making it difficult to achieve miscibility. The reservoir oil has an API gravity of 30 and hence can be deduced to have a good number of lighter components, as seen in its relatively lower oil viscosity of 0.7 cp. This oil viscosity is within the acceptable oil viscosity range deemed for a successful CO₂ flood.

The temperature and pressure of the field favor miscibility for a CO₂ flood process. The estimation of the minimum miscibility pressure (MMP) based on a slim tube test and correlations detailed in later sections shows that this field meets the MMP requirements in order to implement a CO₂ flood.

Table 1. Summary of the requirements of successful CO₂ flooding [23].

Oil and Reservoir Characteristic	Requirement	Value
Oil gravity, API	>27 to 30	30
Oil viscosity, cP	≤10 to 12	0.70
Reservoir depth, ft	<9800 and >2000	8000
Reservoir temperature, F	<250	210
Oil saturation after waterflooding, frac	>0.25 to 0.30	0.52
MMP achievable	Yes	2750 (using slim tube test)

shows the summary of the requirements of successful CO₂ flooding.

Compared to other reservoirs, the field demonstrates very good compatibility with CO₂-WAG due to its moderate API gravity, favorable MMP conditions, and successful waterflood history, which increases sweep efficiency and CO₂ injectivity. The selected reservoir, with a depth of approximately 8000 ft, falls well within the optimal range for CO₂ flooding as deeper formations tend to have the pressure required to attain miscibility with the injected CO₂. Additionally, its temperature of 210 °F is within the acceptable range to maintain CO₂ in a supercritical state, ensuring better interaction with crude oil. When compared to relatively shallower reservoirs like those <2000 ft in depth, additional compression may be required to help the reservoir attain miscibility. Conversely, reservoirs with a depth greater than 10,000 ft often require significantly higher injection pressures and operational costs, making them less economically viable. The selected reservoir provides a balance between these extremes. The API gravity of 30 positions this reservoir in a favorable range for CO₂-EOR. Reservoirs with lighter crude oils (API > 35) tend to have higher gas solubility, making it easier to achieve miscibility, but may suffer from early gas breakthroughs. Conversely, heavier oils (API < 25) require significantly higher pressures for miscibility, making CO₂ flooding less effective. Furthermore, its prior success with waterflooding, which recovered 29.4 MMSTB, suggests high residual oil saturation, making it a good candidate for CO₂-EOR.

According to the National Energy Technology Laboratory [23], reservoirs that have had a successful waterflooding process are ideal candidates for CO₂ flooding projects. The screening criteria include minimum miscibility pressure, residual crude oil volume, and the ability of CO₂ to contact the crude oil. Reservoirs in the Permian Basin are known to exhibit favorable conditions, such as extensive waterflooding history, low geothermal gradients, which reduce miscibility pressure, and high permeability. Operators rely on screening tools like CO₂ Prophet version 1 for technical and economic assessments before full-scale numerical simulations. Table 1 outlines the essential criteria for effective CO₂ flooding.

Analyzing the constant composition expansion and differential vaporization data, the field supports the idea that it will be successful to embark on a CO₂-EOR project. The following expands on our reasoning. The ultimate oil recovery for a CO₂-EOR project is only realized when miscibility is attained. In general, miscibility is influenced by the pressure and temperature, oil composition, and contaminants of the displacing fluid. Ideally, a lower thermodynamic miscibility pressure, higher reservoir temperature, lighter oil composition, and little to no contaminants of the displacing fluid are required, which the field under study meets. Analyzing the PVT data, the following can be inferred:

Pressure: Generally, higher pressure favors a CO₂-EOR project because higher pressure causes the CO₂ to be in the supercritical state. The density of the supercritical CO₂ (ScCO₂) possesses a density similar to that of the oil and hence results in the effective displacement of the oil by the ScCO₂. The initial pressure of the reservoir is estimated to be about 4027 psi, which is high and above most thermodynamic minimum miscibility pressures (MMP) recorded using a variety of approaches that are discussed in detail in the subsequent sections.

Temperature: Based on the literature, temperatures below 120 °F are more likely to result in three phases, excluding the water phase, which mostly results in poor sweep efficiency. The field shows a temperature of about 210 °F, making it more likely to prevent a three-phase scenario. Hence, there is a higher probability of better sweep efficiency. In addition, relatively higher temperatures increase the chances of the breakdown of heavier oil components into intermediate ones and the intermediate oil components into lighter ones, making the oil relatively lighter and more favorable to being displaced by a displacing fluid.

Oil composition: Having more lighter components, fewer heavier components, and little to no contaminants are ideal cases for attaining miscibility for CO₂-EOR projects. More than 50% of lighter components are present with relatively fewer heavier components. This signifies a relatively lighter oil. This mixture of oil composition is a good candidate for miscibility projects as the extraction of the relatively heavier components by the displacing fluid (CO₂) becomes faster and enhances miscibility.

Table 2. Statistics of the dataset.

Core Data
Total samples	208
Depth	7792 ft to 8736 ft
Statistics	Min	Max	Average	Standard deviation
Porosity, (v/v)	0.09	0.30	0.21	0.05
Permeability, md	0.14	383.96	73.64	85.70
Well Log Data
Total samples	10,550
Depth	7514 ft to 8899.5 ft
Statistics	Min	Max	Average	Standard deviation
Gamma ray, API	24.81	124.04	79.63	27.76
Bulk density, g/cm3	2.04	2.64	2.46	0.11
Resistivity, ohm-m	0.33	287.94	17.26	17.13

gives a summary of the statistics of the log and core data utilized in this study.

2.2. Fluid Modeling

Using available differential liberation and constant composition expansion (CCE) test data available after performing laboratory tests on the reservoir fluid sample, a fluid model is constructed and tuned to an equation of state by minimizing the error between the laboratory data on the fluid sample and the equation of state (EOS), which was utilized through a regression process using the CMG Winprop version 2022.1 software [24]. The workflow utilized in tuning the fluid model to the EOS is detailed below.

In constructing the fluid model, the Peng Robinson 1976 EOS [25] was used. This EOS is a modification of a cubic EOS that was also developed by Peng Robinson in 1976. The changes made to the equation developed initially were to enhance the phase behavior predictions. This EOS has been used extensively in chemical and petroleum engineering applications.

Reservoir fluids are usually composed of different hydrocarbon mixtures, which can make the fluid modeling process very tedious. To simplify the computational processes required, some components that make up the fluid can be lumped together to simplify the process while ensuring the original characteristics or properties of the fluid are not altered. Lumping looks to group together hydrocarbons that have a similar molecular weight to reduce the number of pseudo-components for fluid modeling processes. This approach often results in a significant reduction in the simulation run time. However, while lumping facilitates more efficient computational workflows, it also introduces some trade-offs between model simplification and tuning accuracy. It is important that the lumping process does not impact significant phase behavior characteristics.

The choice of lumping strategy can directly impact the accuracy of predictions. Hence, tuning the fluid model with an appropriate lumping strategy is necessary to ensure the behavior of the reservoir fluid is accurately represented. It is, however, important to make sure that the behavior of the reservoir fluid is accurately represented by not compromising on the model accuracy in the process to prevent potential incorrect predictions of the phase behavior, which will negatively impact reservoir recovery forecasts and flow rates. Inaccurate phase behavior predictions resulting from tuning that is not accurate will most likely result in incorrect estimates of various key parameters, such as fluid production rates. Therefore, tuning the fluid model for accurate phase behavior predictions is paramount to obtaining reliable reservoir forecasts. This study implements several lumping techniques, which include Lee et al.’s [26], Danesh et al.’s [27], and Pedersen [28] and Whitson’s [29] lumping schemes on the reservoir fluid prior to conducting further fluid modeling processes. The implemented lumping techniques are briefly described in the following section:

Lee’s mixing rules, referenced by Rouighi [30], are widely used for characterizing lumped fractions in fluid systems. These rules are strongly dependent on Kay’s mixing rules and provide a systematic workflow for calculating the thermodynamic properties of multi-component mixtures in lumping schemes.

The process begins by defining the normalized mole fraction (${\varphi }_i$) of component i within the lumped fraction L, calculated as:

$${\varphi }_i={\displaystyle \frac{z_i}{{Ʃ}_jϵL{z}_j}}$$

(1)

where z_i is the mole fraction of component i in the mixture.

Using these normalized mole fractions, Lee’s rules calculate the thermodynamic properties of the lumped fraction (L) as weighted averages of the individual components’ properties:

Molecular Weight (M_wi):

$$M_{wL}=\sum _{iϵL}{\varphi }_{i}M{w}_1$$

(2)

Specific Gravity ${(\gamma }_L)$:

$${\gamma }_L=\sum _{iϵL}{\varphi }_i{\gamma }_L$$

(3)

Critical Volume ($V_{cL}$):

$$V_{cL}=\sum _{iϵL}{\varphi }_i{\displaystyle \frac{M_{wi}}{M_{wL}}}{V}_{ci}$$

(4)

Critical Pressure ${(P}_{cL})$:

$$P_{cL}=\sum _{iϵL}{\varphi }_i{P}_{ci}$$

(5)

Critical Temperature ${(T}_{cL}):$

$$T_{cL}=\sum _{iϵL}{\varphi }_i{T}_{ci}$$

(6)

Acentric Factor ${(\omega }_L):$

$${\omega }_L=\sum _{iϵL}{\varphi }_i{\omega }_i$$

(7)

2.3. Lee et al.’s Lumping Scheme

Lee et al. [26] developed a straightforward procedure to guide the grouping of oil fractions into pseudo-components. The method was based on the premise that crude oil fractions with similar physicochemical properties can be accurately represented by a single pseudo-component. This similarity is assessed by examining the slopes of property curves, such as molecular weight or boiling point, when plotted against characteristic independent variables. The approach ensured that components within a group share comparable properties, simplifying the fluid characterization while preserving the accuracy of thermodynamic predictions [31].

2.4. Danesh et al.’s Lumping Scheme

Danesh et al. [27] introduced a lumping method that classifies components according to their concentration and molecular weight in a mixture. This approach looks to arrange the original components in ascending order of their normal boiling point temperatures and then groups them into pseudo-components. The goal is to ensure that the summation of the weighted logarithms of molecular weights is approximately equal across all groups. This criterion is referred to as the quasi-equal-weight criterion.

Mathematically, the grouping satisfies the following condition:

$$0 \le \sum _{i=1}^n{z}_i\ln \left(M_{w, i}\right)-{\displaystyle \frac{1}{N_p}}\sum _{i=1}^p{z}_i\ln \left(M_{w,i}\right)$$

(8)

where

$N_p$: Number of pseudo-component groups,

$n$: Total number of components in the mixture,

$z_i$: Mole fraction of component i,

$M_{w,i}$: Molecular weight of component i.

The grouping process ensures the following:

$${\displaystyle \frac{1}{N_p}} \sum _{i=1}^n{z}_i\ln \left(M_{w, i}\right)\approx {\displaystyle \frac{1}{N_p}}\sum _{i=1}^n{z}_i\ln \left(M_{w,i}\right)$$

(9)

Slight deviations in the equation above are allowed due to practical constraints. Each group was constructed by combining components with similar boiling point ranges and molecular weights until the quasi-equal-weight criterion was met.

This method provides a systematic approach to balance the representation of heavier and lighter components in lumped fluid models. It is particularly effective for maintaining accuracy in thermodynamic calculations for compositional simulations, where boiling point and molecular weight distributions significantly impact phase behavior predictions [30].

2.5. Pedersen et al.’s Lumping Scheme

Pedersen et al. [28,32,33,34,35] proposed a lumping method that groups components based on their weight fractions. This approach ensured that each of the groups within the C7+ fraction had approximately the same total weight, giving equal importance to all hydrocarbon segments of the C7+ fraction.

The C7+ fraction was divided into three or more groups, depending on the desired level of detail and specific reservoir conditions. The weight of each pseudo-component group was calculated as:

$$W_j=\sum _{iϵj}{z}_i{M}_{w,i}$$

(10)

where

W_j: Total weight of pseudo-component group j,

z_i: Mole fraction of component i,

M_w,i = Molecular weight of component i.

This method ensured that the weight fraction of each pseudso-component group was balanced and made it effective for compositional modeling, where a simplified, yet representative, fluid characterization is needed. Once the group was defined, their thermodynamic properties could be determined using the mixing rules, such as Lee’s method [26].

2.6. Whitson’s Lumping Scheme

Whitson [29] proposed a lumping methodology that simplified the compositional distribution of the C7+ fraction by grouping components into multiple-carbon-number (MCN) groups. This approach ensured that the complexity of the fluid description was reduced while keeping its thermodynamic accuracy. The number of MCN groups required to characterize the C7+ fraction was determined using the following empirical formula:

$$N_G=Int \left[1+3.3\log \left(N-n\right)\right]$$

(11)

where

N: Carbon number of the heaviest component in the fraction,

n: Carbon number of the lightest component in the fraction (typically, n = 7 for C7+).

For black oil systems, N_G may be reduced by one without significantly impacting the accuracy.

The molecular weight (Mw) boundaries defining each MCN group are calculated as follows:

$$M_{wi}=M_{w7}\ast ({{\displaystyle \frac{M_{wN}}{M_{w7}}})}^{{\displaystyle \frac{1}{N_G}}}$$

(12)

where

M_w7: Molecular weight of the lightest component in the fraction,

M_wN: Molecular weight of the heaviest component in the fraction,

i: Group index, ranging from 1 to N_G.

Whitson’s [29] scheme is particularly effective in the compositional modeling of reservoir fluids, where the representation of heavier hydrocarbon fractions is critical [31]. Figure 1 is a flowchart illustrating the steps followed to achieve the objectives.

2.7. MMP Determination

Numerous analytical correlations have been proposed for MMP estimation, many of which are derived from experimental data. These are some of the notable correlations [7,8,36,41,42,43]. These models have contributed significantly to the industry’s understanding of the relationship between oil composition, temperature, and MMP. One of the earliest correlations for MMP estimation was developed by Benham et al. [41]. Their work involved miscibility estimation for various reservoir oils displaced by rich gases, spanning a pressure range from 1500 to 3000 psia. By testing five distinct reservoir fluids with varying C5+ fractions, Benham et al. [41] laid the groundwork for several subsequent MMP correlations. This correlation uses reservoir temperature and oil composition as input parameters and remains influential in guiding further research on MMP prediction.

Holm and Josendal [7] extended Benham et al.’s [41] work by incorporating a broader range of oil compositions and reservoir temperatures. They introduced the concept of the volatile oil fraction (primarily C1 and N2), the intermediate fraction (C2–C4, CO₂, and H₂S), and the molecular weight of the heavier C5+ components. This correlation proved especially useful for estimating MMP with CO₂ as the injection gas, although it has limitations when applied to fluids with high C5+ molecular weights, especially those exceeding 200 g/mol. Yellig and Metcalfe [42] conducted an extensive experimental program to explore the impact of temperature and oil composition on MMP. Their findings suggested that temperature has a significant effect on CO₂ MMP, with the correlation indicating an approximate increase of 15 psi/°F over a range of 95–192 °F. Although they noted that oil composition did not substantially affect MMP, their correlation’s applicability is limited to specific reservoir oils, and deviations can occur when applied to more diverse compositions.

Glaso [36] proposed a generalized correlation based on the graphical data from [43]. Glaso’s [36] correlation accounts for reservoir temperature, the molecular weight of C7+ fractions, methane content in the injection gas, and intermediate molecular weights (C2–C6) in the oil. This correlation is widely used in CO₂ injection scenarios, although it may overestimate or underestimate MMP in cases where compositional effects deviate from the typical reservoir oils it was calibrated against. We estimated the MMP using analytical correlations proposed in five studies.

2.8. Simulation Model

After an accurate fluid model was developed, we proceeded to calculate the minimum miscibility pressure (MMP) of the reservoir. The process was modeled in the CMG (Computer Modeling Group) GEM version 2022.1 (Generalized Equation-of-State Model) [44] environment by designing a 1D slim tube test to estimate the MMP. The test was performed under different ranges of pressures, and the oil recovery after injecting 1.2 PV of CO₂ gas was recorded. With MMP determination using the slim tube method, a percentage recovery of about 90% is targeted since this value is a representation of the pressure at which enough miscibility has occurred to result in the efficient displacement of the oil. Figure 2 shows the setup of the MMP determination process.

The properties of the slim tube simulation design are specified in

Table 3. Simulation setup for 1D slim tube test.

Properties	Value
Number of grids (I-, J-, and K-directions), ft	20 × 1 × 1
Grid size, ft	1 × 1 × 1
Porosity, fraction	0.22
Permeability, mD	10,000
Connate water saturation, fraction	0.25
Initial oil saturation, fraction	0.52
Saturation pressure, psi	1856.2

. It is important to design the slim tube experiment using relatively high permeability values to ensure that the flow within the experimental design is controlled primarily by the effects of the miscibility and not due to heterogeneity. Using high values of permeability helps to reduce the pressure drops that may potentially occur as the injected CO₂ gas moves through the slim tube, where the flow should be controlled by the miscibility between the injected CO₂ and the oil rather than due to any physical restrictions that may be present. The high permeability also helps in creating a better interaction between the CO₂ and the oil, which results in a better displacement of the oil and makes the process more efficient. This results in a more distinct profile across different pressures and contributes to the accurate determination of the MMP. Also, this study briefly looks at the impact of nitrogen as an impurity in the determination of the MMP.

2.9. CO₂ Prophet

The CO₂ Prophet version 1 software [20] generates a stream tube model for a reservoir. Streamlines are built depending on the injection and production rates specified by the user. The streamlines help to define the flow paths for all the fluid in the reservoir. Stream tubes are then created from the streamlines, which are separated into various sections for finite difference calculations [45]. The software is capable of executing a number of patterns for the injection and production wells. The patterns include configurations like the 5-spot pattern and some line drives and also allow for a custom pattern to be specified.

The CO₂ Prophet version 1 software [20] uses a mixing parameter approach to simulate flow behavior. This method captures key displacement mechanisms, including mixing and viscous fingering, by adjusting the viscosities of the injected fluid and reservoir oil rather than using a full equation of state (EOS). This makes computations simpler while still approximating the essential physics of miscible flow. The software typically assumes homogeneous rock properties and maintains isothermal conditions. The reservoir is considered an isolated system with closed boundaries, ensuring that no external pressure influences the modeling process, with injection and production wells operating under a fixed rate.

In this study, a custom pattern was designed to mimic the location of the injector and production wells of the field. Figure 3 shows the porosity distribution and the location of all 15 wells in the field.

In the pattern, the location of each of the injection and production wells is specified by defining the X and Y coordinates. The reservoir area in this study is about 22 acres, and hence, a distance of 1000 ft by 1000 ft was specified within the software. Figure 4 and Figure 5 show the injection and production well locations. A total of 6 injection wells and 9 production wells were utilized in the waterflood scenario within the CMG GEM version 2022.1 reservoir simulator [44], and hence, a similar development strategy was implemented within the CO₂ Prophet version 1 software [20].

A history match was then performed on the field, and the results were compared to that obtained using the CMG GEM version 2022.1 simulator [44] for the 15-year waterflooding stage in the study by [22]. The initial oil saturation after the primary recovery phase prior to water flooding was 0.75. The aim of this history match process was to ensure the results from the CO₂ Prophet version 1 model [20] were closely aligned with observed field data and CMG GEM version 2022.1 simulation [44] results. We performed a waterflood of the field utilizing 6 water injection wells and 9 oil-producing wells. The CMG GEM version 2022.1 simulator [44] specified a bottomhole pressure of 4207 psi for all the water injection wells. For the CO₂ Prophet version 1 model [20], the average field injection rate specified for each of the six injectors was 1500 bbls/day of water since it did not have the option for a bottomhole pressure specification. Also, a total of about 0.8 PV of water was injected. This rate was chosen to achieve a similar injection volume and recovery to that observed in the CMG GEM version 2022.1 model [44]. The production rate within the CMG GEM version 2022.1 simulator [44] averaged a total field production rate of about 11,000 bbls/day. With each of the 9 production wells for the CO₂ Prophet version 1 model [20], a production rate of about 1200 bbls/day was specified for each of the production wells. Figure 6 and Figure 7 show the injection rate specification for one of the injection wells and the production rate specification for the producer wells.

Table 4. CO₂ Prophet version 1 simulation model inputs.

Property	Value
Dykstra–Parsons coefficient	0.8
Reservoir temperature, oF	210
Oil viscosity, cp	0.7
Oil formation volume factor, RB/STB	1.4
Average reservoir pressure, psia	2600
Oil specific gravity, oAPI	30
Gas specific gravity	0.75
Water viscosity, cp	1.3
Water salinity, ppm	70,000
Area, acres	22
Porosity, fraction	0.22
Initial oil saturation, fraction	0.52
Water injection rate for 1 well, bbls/day	1500
Oil production rate for 1 well, bbls/day	1200

below specifies the input parameters for the simulation using the CO₂ Prophet version 1 software [20].

3. Results

3.1. Fluid Modeling

After implementing the various fluid modeling methods, Danesh et al.’s [27] lumping scheme was selected as the best-performing approach on this dataset when compared to the other lumping methods when the mean squared error and the root mean square error were calculated after comparing the results to the experimental data. Appendix A provides more detail regarding the delineation of the C7+ components into various groups using the lumping schemes that have been discussed earlier. With pseudo-components ranging from C₇H₁₈ to C30+, Danesh et al.’s [27] lumping scheme was applied to create four distinct groups. The performance of each lumping method was assigned a rank based on its performance, with a lower rank indicating better performance.

Table 5. Comparison of various fluid properties for the different component lumping methods.

Lumping Method	Danesh [27]	Lee [26]	Pedersen [28]	Whitson [29]
Experiment type	CCE (Constant Composition Experiment)
Fluid property	ROV (Relative Oil Volume)
MSE	0.132381	0.143046	0.151196	0.143479
RMSE	0.363842	0.378214	0.388839	0.378786
Rank	1	2	4	3
Fluid property	CL (Liquid Compressibility)
MSE	1.72 × 10−12	2.17 × 10−11	1.5 × 10−11	1.73 × 10−12
RMSE	1.31 × 10−6	4.66 × 10−6	3.87 × 10−6	1.31 × 10−6
Rank	1	4	3	2
Experiment type	Differential Liberation
Fluid property	ROV (Relative Oil Volume)
MSE	0.000543	0.000615	0.000624	0.000557
RMSE	0.0233	0.024808	0.024977	0.023598
Rank	1	3	4	2
Fluid property	DL (Liquid Density)
MSE	0.000114	0.000178	0.000183	0.000161
RMSE	0.010665	0.013349	0.013531	0.012683
Rank	1	3	4	2
Fluid property	FVF (Formation Volume Factor, Oil)
MSE	0.000958	0.001033	0.001023	0.000946
RMSE	0.030949	0.032142	0.031987	0.030758
Rank	2	4	3	1
Fluid property	SGV (Gas Relative Density)
MSE	0.006947	0.001565	0.001636	0.008806
RMSE	0.083349	0.039564	0.040443	0.09384
Rank	3	1	2	4
Fluid property	MUL (Liquid Viscosity)
MSE	0.00461	0.004905	0.005495	0.007789
RMSE	0.067894	0.070036	0.074127	0.088253
Rank	1	2	3	4
Fluid property	MUV (Gas Viscosity)
MSE	5.16 × 10−6	4.34 × 10−6	3.82 × 10−6	6.39 × 10−6
RMSE	0.002272	0.002083	0.001955	0.002528
Rank	3	2	1	4

shows the results of the comparison between the various lumping methods.

Figure 8, Figure 9, Figure 10, Figure 11 and Figure 12 show the results for Danesh et al.’s [27] lumping method for various fluid properties.

3.2. MMP Determination

3.2.1. Analytical Methods

Table 6. Summary of the estimation of the MMP using analytical correlations proposed by different authors.

MMP from Analytical Correlations
Correlation	MMP (Psia)	MMP (MPa)
Yuan et al., 2004 [39]	3268.34	22.53
Delforouz et al., 2019 [40]	3459.13	23.85
Li et al., 2012 [38]	3350.84	23.10
Lai et al., 2017 [37]	2952.64	20.36
Glaso, 1985 [36]	2596.07	17.90

and Figure 13 illustrate the comparison of MMP values across these studies [36,37,38,39,40], showing significant differences that emphasize the importance of selecting an appropriate correlation based on specific reservoir conditions.

3.2.2. Simulation Method

• Pure CO₂ injection

The plot below shows the results for a pure (100%) CO₂ injection by simulating a slim tube experiment to help determine the MMP. Determining the MMP is important since it helps us know the pressure above which the injected CO₂ will effectively mix with the reservoir oil to potentially help increase the recovery factor.

The MMP of 2475 psi defines the pressure above which the injected CO₂ will mix with the oil to result in a more efficient displacement for this reservoir. Figure 14 shows a chart of test injection pressure and the % recovery factor for 100% CO₂ as the solvent.

• Impact of impurities on MMP

The presence of impurities in a CO₂ stream can have a very significant impact on the MMP determination process. This is important because the efficiency of the injected CO₂ may be impacted and could negatively impact the budget of projects. Some common impurities in the CO₂ stream may include nitrogen (N₂), methane (CH₄), and hydrogen sulfide (H₂S). These impurities tend to affect the MMP differently since they all have different properties and interact differently with the reservoir fluids [35]. It is important that the different interactions that may result from the presence of impurities are understood since their impacts on the MMP can greatly affect the feasibility of injecting CO₂ to improve oil recovery.

The concentration or amounts of impurities also play a very important role with regard to the impact on the MMP. Generally, gases with a lower molecular weight than CO₂ tend to increase the pressure needed to attain miscibility, while gases having a higher molecular weight than CO₂ tend to have the opposite effect. This results in the need to increase the reservoir pressure to attain miscibility, causing operating costs to increase and potentially making the CO₂ flood less viable. To meet the increased pressure requirements, there are significant modifications that may need to be made to the existing facility to be able to handle the increased pressure for injection.

On the other hand, other heavier hydrocarbons such as ethane and propane tend to have a positive impact on the MMP by promoting the miscibility of the CO₂ at considerably lower pressures. In these cases, the pressure needed to attain miscibility is greatly lowered and results in a reduction in operational expenses. Although H₂S has a higher molecular weight than CO₂, it presents corrosion and safety concerns, making it undesirable. The presence of H₂S in the CO₂ stream can cause corrosion to the production equipment, and thus, some investment needs to go into purchasing corrosion-resistant equipment to sustain production. Figure 15 highlights the impact of N₂ on the MMP.

3.3. CO₂ Prophet

Following earlier investigations by [22], where various field development strategies were considered using waterflood, this study proceeds further to evaluate the impact of implementing a CO₂ flood on the recovery of oil. The Dykstra–Parsons coefficient [46] is an important parameter that helps to quantify the degree of heterogeneity of the reservoir and is an important requirement for the CO₂ Prophet version 1 model [20].

3.3.1. Dykstra–Parsons Coefficient

The Dykstra–Parsons coefficient [46] is used to characterize the heterogeneity within the CO₂ Prophet version 1 software. Figure 16 shows the permeability versus the % of samples with greater permeability and their corresponding permeability. Ensuring the variation in reservoir permeability is calculated accurately is important since it enables the simulator to make better predictions of the CO₂ performance.

Heterogeneous reservoirs generally have a coefficient closer to 1, with the more homogeneous reservoirs having a coefficient closer to 0. The Dykstra–Parsons coefficient [46] is calculated from Equation (13).

$$V={\displaystyle \frac{k_{50}-k_{84.1}}{k_{50}}}$$

(13)

where k₅₀ is the median permeability value of the reservoir, k_84.1 is the permeability value at the 84.1th percentile of the data, and V is the coefficient of variation. The coefficient of variation was obtained by dividing the standard deviation of the permeability by the mean permeability. The coefficient of variation was calculated as 0.796. This value indicates that the reservoir is very heterogeneous in nature, as shown in Figure 17.

3.3.2. History Match Results

• Cumulative oil and water production using the CO₂ Prophet version 1 software

Figure 18 is a chart that shows the field cumulative oil and water production for the 15-year secondary recovery time period using the CO₂ Prophet version 1 software [20]. In the figure, the green curve represents the cumulative oil production and shows a steady increase until it reaches about 30.47 MMSTB. The blue curve, which represents the cumulative water production, is seen to also rise steadily to a value of 6.05 MMSTB. Comparing the results to those obtained by [22] using the CMG GEM version 2022.1 reservoir simulator [44], the cumulative production for oil and water was obtained as 29.4 MMSTB and 6.3 MMSTB, respectively.

3.4. Reservoir Performance Forecasting Using the Kinder Morgan Scoping Model

3.4.1. Oil Rates

Using the Kinder Morgan Scoping Model [18] to evaluate the feasibility of the WAG process after the field has undergone a secondary recovery phase using waterflooding, field cumulative oil production was forecasted for a 25-year period. The cumulative oil recovery at the end of the 25-year production period was estimated at about 20.26 MMSTB, as shown in Figure 19. The graph also highlights the impact of a CO₂ flood on oil recovery. From the plot, the black curve shows that of the oil rate without the injection of CO₂ and typically represents only a waterflood situation. The steady decline observed in this curve highlights the natural depletion of the reservoir pressure and a reduction in oil recovery a reservoir undergoes with an increase in production time. In contrast, the red curve represents the cumulative oil recovered due to the utilization of CO₂ over the simulation time.

3.4.2. Production Rates

In Figure 20, the oil production rate begins at ~10 mbbl/d and initially increases slightly as the CO₂ flood mobilizes residual oil. With increasing production, oil production is observed to decrease gradually as the field depletes. In the plot, oil production declines to about 1.41 mbbl/d by the end of the 25-year period. Water production, which is indicated by the blue curve, is observed to increase significantly throughout the 25-year period, attaining a stable value at approximately 54.51 mbbl/d.

3.4.3. Injection Rates

Figure 21 illustrates the CO₂ and water injection rates over 25 years of WAG injection in the reservoir. The alternating injection of CO₂ is indicated by the red line and that of water is shown by the blue line. CO₂ injection begins at high rates, going above 60 mmcf/d, but alternates significantly due to the alternating cycles with water.

3.4.4. Undiscounted Cumulative Net Cash Flow (NCF) Above the Working Forecast (WF)

The undiscounted NCF refers to the total net cash flow generated over a period of time without taking into account the time value of money. Figure 22 illustrates the undiscounted cumulative net cash flow (NCF) above the working forecast (WF) over a 25-year period. The red curve shows the undiscounted cumulative NCF above the WF before taxes. This value generally lies above the after-tax (AT) curve, indicating higher cash flow values prior to tax deductions, as seen in the green curve. In the plot, there is a slight initial dip, which is likely due to early-stage investment and operational costs. In both curves, the difference in the trend can be attributed to taxation.

3.4.5. Present Value (PV) of Profit After Tax

The plot below, Figure 23, highlights the relationship between the present value (PV) of profit after tax (PVPAT) and the nominal discount rate, which ranges from 0% to 50%, while the y-axis represents the PV of profit after tax in MM$. A downward trend can be observed in this curve, from which we can infer that as the nominal discount rate increases, the present value of profit after tax decreases. Hence, the higher the discount rate, the lower the present value of future cash flows. This reflects the time value of money.

4. Discussion

With CCE, Danesh et al.’s [27] lumping scheme showed the best performance, which can be observed for the ROV and CL fluid properties, which shows the least MSE and RMSE difference compared to the observed data. Also, with the DL measurements, Danesh et al.’s [27] lumping scheme showed the best performance with the ROV, DL, and MUL fluid properties, when compared to the experimental data. The consistency in the results for Danesh et al.’s [27] lumping method across different fluid properties proves its reliability and accuracy in modeling fluid behavior. Whitson’s [29] lumping method showed the best results with the oil FVF, with Lee et al.’s [26] and Pedersen et al.’s [28] lumping schemes showing the best performance when the SGV and the MUV were compared to the experimental data, respectively. The 1856.8 psig result for the fluid saturation pressure for Pederson’s [28] and Whitson’s [29] lumping methods showed the least deviation from the saturation pressure of the experimental data, which was 1857 psig, whereas Danesh et al.’s [27] and Lee et al.’s [26] lumping methods recorded saturation pressures of 1858 psig and 1857.6 psig, respectively.

From the analytical methods, Glaso’s [36] correlation yielded the lowest MMP, which aligns with its design to incorporate the effects of lighter oil components and intermediate molecular weights on miscibility. This makes it particularly suited for reservoirs with a higher concentration of intermediate hydrocarbons. Conversely, Delforouz et al.’s [40] correlation produced the highest MMP value, which could be attributed to its incorporation of broader reservoir conditions and its emphasis on heavier hydrocarbon fractions, increasing the required pressure for miscibility. The MMP values of 2952 psi from Lai et al. [37] and 3350 psi from Li et al. [38] fall within the mid-range values of the results, suggesting that these correlations balance the influence of light and heavy hydrocarbon components. Yuan et al. [39] obtained an MMP of 3268 psi, indicating a strong sensitivity to reservoir-specific properties, such as temperature and pressure gradients. According to Table 6 and Figure 13, Glaso’s [36] correlation is effective for lighter oil compositions, while Delforouz et al.’s [40] approach may be more relevant for reservoirs with complex fluid properties. These findings underline the necessity of validating analytical MMP estimates with experimental or simulation-based methods, such as the slim tube test, to ensure accuracy. While analytical methods provide a quick and computationally efficient means of estimating the MMP based on empirical correlations, they may not always capture the total complexity of reservoir conditions. Simulation-based methods offer a more detailed approach by considering phase behavior modeling and reservoir-specific parameters, leading to potentially more accurate MMP estimations. However, simulations require more data inputs and computational resources, making them more time-intensive compared to analytical approaches. This comparative analysis demonstrates that no single correlation universally applies to all reservoirs. The selection of the appropriate method should consider the specific characteristics of the reservoir and the goals of the CO₂-EOR project, balancing computational efficiency with accuracy.

When determining the MMP using the simulation method, an increasing trend is generally observed for oil recovery as the pressure is increased. As the injected CO₂ begins to dissolve into the oil phase and causes the viscosity to reduce, this process is known as oil swelling and enhances the flow characteristics of the oil, making it easier for the CO₂ to displace it. The combination of reduced viscosity and an increase in volume helps improve both the mobility of the oil and the overall efficiency of the displacement process. From the trend, higher pressure enhances the interaction between the CO₂ molecules and the reservoir oil, which leads to an enhanced miscibility situation, resulting in an increased oil recovery. At a pressure of around 2475 psi, the recovery is observed to begin to stabilize, and further increases in the pressure result in relatively reduced amounts of the oil recovered. At this point, the miscibility threshold has been attained and further increases in the pressure may not yield any significant oil recoveries, making further injection inefficient. The impact of 10% N₂ added to the CO₂ is evidenced. In Figure 15, the % recovery factor increases with an increase in pressure as well. However, the MMP for the CO₂ and N₂ mixture increases to about 4900 psi compared to the pure CO₂ case, which is about 2475 psi. The addition of nitrogen to the CO₂ stream results in a significant increase in the MMP and hence results in a higher pressure requirement to be able to attain recovery results similar to those with only the injection of pure CO₂. The changes observed in the MMP values between pure CO₂ injection and the 90% CO₂ and 10% N₂ mixture can be explained by the differences in the phase behavior of the two gas mixtures. The addition of N₂, which is less soluble in oil compared to CO₂, causes a reduction in the overall miscibility of the injected gas with the reservoir oil. This results in an increase in the pressure required to reach miscibility and results in the MMP value being relatively higher compared to the pure CO₂ injection case. The MMP value for pure CO₂ is lower because CO₂ has a higher affinity for dissolving into the oil phase, which leads to better miscibility at lower pressures. The addition of nitrogen to the CO₂ stream introduces a higher injection pressure requirement to achieve similar levels of oil recovery as pure CO₂. Increasing the pressure also raises operational costs as more energy and infrastructure are needed to support the injection at relatively higher pressures. In addition, the presence of contaminants like nitrogen may require gas purification to meet injection standards, which further adds to the overall project expense. The costs associated with purification and the need for specialized equipment to manage higher pressures can have a significant impact on the economic viability of implementing a CO₂-EOR project. This, in effect, affects the economic outlook of the project where impurities are present.

The impact of varying the Dykstra–Parsons coefficient [46] on the cumulative oil recovery using the CO₂ Prophet version 1 model [20] was evaluated by varying the value of the coefficient from 0.55 to 0.8. The results show that as the reservoir became more heterogeneous, the cumulative oil recovery decreased, as shown in Figure 17. Khataniar and Peters [47] found the reservoir heterogeneity to be detrimental to the field recovery performance. The severity of this effect was observed to increase with an increase in the Dykstra–Parsons coefficient [46]. This effect often leads to the development or formation of different fingering patterns for the various permeability distributions.

The deviation in the production estimates in Figure 18 between the two software programs was minimal, with differences of 3.5% observed for oil and −4.11% for water cumulative production after a 15-year period of water injection. The slight deviation in the production results between the different software could be attributed to the underlying assumptions made within each simulator regarding relative permeability, the modeling of water influx, or the heterogeneity of the reservoir. The CO₂ Prophet version 1 software [20] generally uses a relatively simplified fluid flow model compared to the CMG GEM version 2022.1 simulator. While the CO₂ Prophet version 1 software [20] provides a fast and efficient means of evaluating CO₂-EOR performance, it has certain limitations compared to the CMG GEM version 2022.1 simulator software [44]. The reliance of the CO₂ Prophet version 1 software on a mixing parameter approach rather than an EOS model can limit the accuracy of phase behavior predictions, especially when dealing with complex fluid systems. Additionally, the CO₂ Prophet version 1 software assumes more simplified reservoir conditions and does not offer the same level of detail in capturing multiphase flow interactions, heterogeneity effects, and compositional changes over time. In contrast, CMG GEM version 2022.1 software [44] incorporates a more advanced numerical framework, allowing for a more detailed representation of the reservoir dynamics.

In Figure 19, which highlights the oil rates, there is a clear increase in oil production compared to that of the base case as CO₂ improves recovery by reducing the oil viscosity and swelling the oil, thereby resulting in a much better mobilization of the oil. Given ample time, the production begins to decline as the injected CO₂ breaks through at the producer wells. The green curve represents the cumulative oil recovered due to a CO₂ flood over time. The curve rises steadily with time, showing the significant contribution of the CO₂ injection to overall recovery.

The overlap observed in Figure 20 water production is indicative of WAG injection cycles. Gas production, which is represented by the red curve, shows an increasing trend in the early years, which is likely an indication of CO₂ breakthrough at the producer wells. After the breakthrough, gas production alternates due to a corresponding water injection until it attains a stable value close to 3.46 mmcf/d.

In Figure 21, CO₂ injection rates decline after about 11 years. Water injection starts later in the WAG process, gradually increases, and becomes the dominant injection phase after year 11. By the end of the simulation, water injection rates reach approximately 42.84 mbw/d. A notable decline in the water injection rate occurs briefly after year 20. The reductions in water injection rates after 20 years are most likely due to reservoir pressure limitations, water breakthroughs, and operational adjustments as the reservoir enters its later stages during the EOR operation. The projected water injection is capped at 36,000 bw/day. Hence, around year 20, the injection rate was adjusted to help optimize production costs, mitigate risks, and control any potential water-handling limitations. Injecting water continuously after an extended period of WAG injection helps maximize oil recovery by maintaining pressure and sweep efficiency.

At discount rates close to 0 in Figure 23, the PV of profit after tax is positive, indicating that the project or investment has a high value. As the nominal discount rate increases, the present value decreases and likely indicates that the discounted future cash flows may no longer cover the costs or investments with higher values of the discount rate.

At the point where the curve crosses zero, the present value of cash inflows equals the cash outflows, resulting in no net profit or loss, which helps determine the economic viability of a project.

5. Conclusions

This study highlights the importance of fluid characterization and reservoir performance analysis in optimizing CO₂-EOR processes. Among the fluid lumping methods that were utilized, Danesh et al.’s [27] method emerged as the best-performing method, delivering accurate predictions across different fluid properties.

This study also examined the impact of impurities like nitrogen in the CO₂ stream, revealing that even small amounts can significantly alter phase behavior. Notably, nitrogen lowers the MMP, which could impact injection strategies and overall recovery efficiency. This finding highlights the need for careful control over CO₂ purity in EOR operations.

MMP was determined using both analytical correlations and slim tube simulations. While analytical methods, particularly Glaso’s [36] correlation, produce results that closely match simulation-based estimates, analytical methods provide quick estimates but often do not factor in specific reservoir condition data, which are necessary to improve the accuracy of the constructed models.

Reservoir heterogeneity, as measured by the Dykstra–Parsons coefficient [46], played an important role in the flow dynamics and sweep efficiency. This study found that higher heterogeneity reduces CO₂ displacement efficiency, making it a key factor in designing effective WAG operations. The CO₂ Prophet version 1 model [20] performed well in history matching and production trend analysis, offering a quick and practical approach to injection optimization. However, its predictive accuracy depends heavily on the quality of input data, and it may not be as robust as more advanced simulators when dealing with complex multiphase flow and compositional effects.

The CO₂-WAG process, designed using the Kinder Morgan Scoping Model [18], is recommended for its operational simplicity. The alternating injection strategy proved effective in mobilizing residual oil, mitigating gas breakthroughs, and sustaining reservoir pressure. The model’s adaptability to various field conditions and its economic feasibility study provide a relatively quick means to determine the feasibility of implementing the WAG strategy using CO₂ gas on a given field. However, operational challenges such as water production, injectivity issues, and CO₂-induced corrosion must be carefully managed. Water production, if excessive, can increase costs and reduce injectivity, while CO₂-related corrosion requires appropriate material selection and chemical inhibition strategies to prevent the degradation of equipment.

CO₂ recycling and separation add further complexity, impacting project economics. Ensuring efficient gas handling and reinjection is crucial to maintaining long-term feasibility. Additionally, uncertainties in reservoir properties, pressure variations, and heterogeneity can affect model predictions, emphasizing the need for conducting sensitivity analysis to better understand the project dynamics.

Future works could focus on evaluating the impact of other impurities like methane, ethane, and H₂S on the CO₂ stream. The studies could also investigate other analytical methods or machine learning techniques to improve the accuracy of MMP determined using correlations and could perform experimental measurements to validate the accuracy of the results. Also, the long-term impact of different WAG cycles on the integrity of CO₂ sequestration efficiency could also be investigated. Conducting a sensitivity analysis on factors like temperature, pressure, and the composition of CO₂ and reservoir oil could help reduce uncertainties in the MMP estimate. Additionally, looking into IFT measurements and modeling would provide a better understanding of the CO₂–oil interactions.