Integrated Simulation of CO₂ Injection in Heavy Oil Reservoirs with Asphaltene Precipitation Effects

Abstract

The results of core flooding experiments can guide the formulation of development plans for similar oil reservoirs. However, for cores from heavy oil reservoirs, asphaltene deposition often occurs during flooding due to changes in pressure, temperature, and petroleum composition, affecting the determination of injection parameters. Taking core samples from the Xia 018 well block as the research object, this study determined that the crude oil sample exhibits normal CO₂ sensitivity based on PVT experiments and core flooding results. A corresponding asphaltene precipitation model was established and coupled with core-scale numerical simulation, forming an integrated core-scale numerical simulation method considering asphaltene precipitation. Through orthogonal experimental design, the optimized fracturing production parameters for Well Y were determined as follows: fracturing stage length of 1000 m, CO₂ injection volume of 100 m³ per stage, fluid volume per stage of 1000 m³, proppant volume of 1000 m³, and injection rate of 14 m³/min. Finally, the optimized parameters were applied to simulate a case well, where the asphaltene deposition model combined with pressure nephograms during production provided effective guidance on unplugging timing. Compared with results without using the asphaltene deposition model, cumulative production decreased by 1300 m³ when the model was applied. These findings can provide a reference for the development of similar reservoirs.

1. Introduction

China possesses abundant unconventional oil and gas reserves, among which heavy oil reservoirs, as a crucial type of unconventional reservoirs, will become important alternative energy sources in the future [1,2,3,4]. CO₂ has long been widely applied in enhanced oil recovery (EOR) research. On one hand, CO₂ demonstrates favorable miscibility with in-situ crude oil. Its dissolution effectively reduces oil interfacial tension and viscosity, thereby improving crude oil mobility in reservoirs, making it extensively utilized in heavy oil reservoir development [5,6,7,8,9]. On the other hand, the application of CO₂ resources in oilfield development aligns with the Carbon Capture, Utilization, and Storage (CCUS) policy, serving as a vital measure to mitigate the global climate crisis.

During CO₂ injection for enhanced heavy oil recovery, asphaltene precipitation tends to occur [10,11,12,13,14,15]. Numerous domestic and international scholars have investigated the mechanisms of asphaltene generation through CO₂-heavy oil interactions, establishing various asphaltene deposition models [16,17,18,19,20,21,22,23]. Mehdi Mahdavifar and colleagues employed real-time microfluidic experiments and a lattice Boltzmann immersed boundary numerical model to investigate pore-scale asphaltene deposition. The experimental results directly observed porosity reduction in different pore geometries, providing a detailed explanation of asphaltene deposition phenomena [24,25,26]. Hassan Sadeghi Yamchi and others conducted numerical simulations of asphaltene deposition induced by solvent injection by implementing liquid-liquid equilibrium (LLE) data and a reaction-based non-equilibrium mass transfer model, exploring the interaction between viscous fingering and asphaltene deposition during dimethyl ether (DME) injection. This study was applied in solvent-based recovery processes and paved the way for integrating LLE data into numerical simulations of asphaltene deposition [27,28]. Nasser Sabet and co-authors studied the mixing kinetics of paraffin solvents with oil during miscible displacement for enhanced oil recovery in porous media, proposing a rigorous formulation of the problem physics and performing high-resolution nonlinear numerical simulations to investigate the impacts of asphaltene precipitation/deposition and the resulting formation damage [29,30]. Moreover, heavy oil extraction requires collaborative optimization of drilling techniques and reservoir modification. For instance, horizontal well segmental fracturing can expand the oil drainage area [31], while acid fracturing before CO₂ injection improves reservoir permeability [32]. As Li et al. demonstrated in hydrate reservoir studies of the northern South China Sea, the coupling effect of wellbore stability and fluid production significantly influences development efficiency [31]. Similarly, CO₂ flooding in heavy oil reservoirs requires considering the impact of fracturing stage length on fluid distribution, aligning with Li et al.’s concept of ‘wellbore-reservoir collaborative modification’ [32].

However, the traditional integrated numerical simulation workflow cannot directly account for the impact of asphaltene changes on production, and since asphaltene plays a pivotal role in production processes, this study takes this as a starting point. Using the Xia 018 well block as an example, core flooding experiments and core-scale numerical simulations were conducted. By coupling the heavy oil component model and the asphaltene deposition model developed based on PVT experimental results for the block, an integrated core-scale numerical simulation method considering asphaltene precipitation was formed. Furthermore, the calibrated asphaltene deposition model was applied to optimize the parameters for CO₂ fracturing injection production in Well Y of the block, providing reasonable production recommendations.

2. Interaction Mechanism Between CO₂ and Heavy Oil

The dissolution of CO₂ in crude oil and its subsequent alteration of oil properties constitute critical factors in enhancing oil recovery through CO₂ flooding. Upon injection into crude oil, CO₂ reduces interfacial tension between oil molecules, induces volumetric expansion, and decreases oil viscosity, thereby significantly improving fluid mobility. However, this process also disrupts the equilibrium between resin molecules and asphaltenes in crude oil, leading to asphaltene deposition. For different types of heavy oil, the interaction patterns between CO₂ and heavy oil vary considerably, thus necessitating experimental investigations to elucidate these mechanisms and provide data support for subsequent numerical simulations.

2.1. Oil Swelling Tests

The oil sample was selected from the Xia 018 well block, located in the eastern part of the Junggar Basin. The reservoir structure generally presents a nose-shaped structure with a north-high and south-low trend, and the formation dip angle is 1.7–9.1°. The oil-bearing series is the Jurassic BD Formation, with a maximum thickness of 190 m. The lithology is mainly gray-green medium sandstone and fine sandstone, which is divided into sublayers a, b, and c from bottom to top. The viscosity of the degassed oil on the ground at 50 °C is 85–18,334 mPa·s, which is greatly affected by faults, making it a typical heavy oil reservoir.

Following SY/T 5542-2009 “Test Method for Reservoir Fluid Physical Properties” [33] and referencing the phase behavior test report of formation fluids from Reservoir Xia 018, formation crude oil was reconstituted under conditions of 51.8 °C and the current average reservoir pressure of 31.75 MPa. The reconstitution results of the obtained samples are presented in

Table 1. Reconstituted crude oil property.

Gas-Oil Ratio, m3/m3	Formation Oil Volume Coefficient	Formation Oil Density, g/cm3	Formation Oil Viscosity, mPa·s	Saturation Pressure, MPa
26.80	1.06	0.89	234.40	10.62

.

To investigate phase compatibility characteristics of different gas injection schemes and elucidate the mechanisms of enhanced oil recovery (EOR) through gas injection, swelling experiments were conducted on formation crude oil under gas injection. These experiments measured key high-pressure physical properties including saturation pressure, formation volume factor, swelling factor, gas-oil ratio, density, and viscosity after gas injection.

The gas injection swelling tests were performed using a DBR visual PVT system (DBR Systems Inc., Calgary, AB, Canada), and the schematic diagram is illustrated in Figure 1 and Figure 2.

Experimental Workflow:

• Referencing the phase behavior test report of formation fluids from Well B, formation crude oil was reconstituted under conditions of 51.8 °C and reservoir pressure of 31.75 MPa;
• Prior to testing, the PVT instrument was thoroughly cleansed, purged with high-pressure nitrogen, oven-dried at experimental temperatures, and evacuated;
• The equilibrated crude oil was transferred into the PVT instrument, stabilized to the current reservoir temperature and pressure conditions, and its volume was measured;
• Injection gases (CO₂ and dry gas) were injected incrementally from low to high percentages into the PVT cell. The mixture was pressurized and agitated until single-phase stability was achieved. Phase behavior characteristics (saturation pressure, P_b) and high-pressure physical properties (formation volume factor, density, viscosity, swelling factor, gas-oil ratio) under varying gas injection percentages were measured. Gas injection volume versus high-pressure property curves were subsequently plotted.

During testing, gas injection volumes were incrementally increased in 10% steps (typically 5–6 data points). Results were analyzed alongside phase behavior simulations to clarify post-injection phase evolution.

Dry gas injection (97.12% methane + 2.88% ethane) revealed that the formation crude oil exhibited heavy components with high density and viscosity. Moderate solubilization and swelling effects were observed post-injection. Rapid saturation pressure increase occurred during gas injection. Limited improvement in crude oil physical properties and moderate gas-oil compatibility were noted. The experimental results of CO₂ injection into in-situ crude oil demonstrate that after CO₂ is introduced, the crude oil exhibits significant solubilization and swelling, with a moderate rate of increase in saturation pressure—conditions favorable for solution gas drive. The injected gas improves crude oil’s physical properties to a certain extent, and gas injection shows good compatibility. Detailed experimental results are summarized in

Table 2. Test Results of Gas Injection Swelling Experiments (dry gas/CO₂).

Injected Gas Volume	Gas-Oil Ratio	Swelling Factor	Bubble Point Pressure	Surface Crude Oil Density	Saturated Crude Oil Density	Saturated Crude Oil Viscosity	Formation Volume Factor
mol%	m3/m3	/	MPa	g/m3	g/m3	cP	/
0	26.8/26.8	1.0000/ 1.0000	10.62/10.62	0.9234/ 0.9234	0.8854/ 0.8854	216.8/216.8	1.0704/ 1.0704
10	37.9/37.9	1.0163/ 1.0170	16.52/12.87	0.9236/ 0.9237	0.8744/ 0.8835	183.4/171.4	1.0878/ 1.0886
20	54.4/54.4	1.0349/ 1.0443	24.51/15.51	0.9238/ 0.9241	0.8645/ 0.8836	138.5/120.8	1.1078/ 1.1178
30	68.9/68.9	1.0629/ 1.0718	33.26/19.41	0.9241/ 0.9243	0.8538/ 0.8841	110.5/79.4	1.1377/ 1.1473
40	92.6/92.6	1.0909/ 1.1188	48.92/26.55	0.9243/ 0.9245	0.8433/ 0.8863	82.3/55.1	1.1677/ 1.1976
50 (CO2 only)	124.5	1.1561	38.64	0.9248	0.8934	40.3	1.2375

.

CO₂ injection into formation crude oil demonstrates significant viscosity reduction effects. However, due to the relatively high subsurface density of CO₂, the density alteration of crude oil post-CO₂ injection remains marginal.

Comparative analysis of swelling test results between dry gas and CO₂ injection reveals:

• Enhanced Phase Compatibility: CO₂ exhibits superior phase compatibility with formation crude oil compared to dry gas;
• Lower Saturation Pressure: Equivalent gas injection volumes yield reduced saturation pressure under CO₂ flooding conditions;
• EOR Mechanism Superiority: This pressure reduction facilitates solubility-driven swelling effects, substantially improving oil recovery efficiency.

2.2. Asphaltene Precipitation Experiments

To elucidate CO₂ injection characteristics and EOR mechanisms, asphaltene precipitation experiments were conducted on formation crude oil under CO₂ injection, measuring the onset pressure and precipitated asphaltene content during pressure depletion. This study focused on defining asphaltene precipitation thresholds and quantifying precipitation volumes for crude oil from the Xia 018 Well Block.

Materials and apparatus used in the experiment are as follows: For materials, a 500 mL heavy oil sample, along with high-purity gases including CO₂ (with a purity of ≥99.99%) and CH₄ (with a purity of ≥99.95%), were employed. As for the apparatus (shown in Figure 3), a crude oil mixer with an operating range of 0–70 MPa and 0–200 °C, a Thermogravimetric Analyzer with a temperature range of 0–1200 °C, a High-Pressure Reactor capable of operating within the range of 0–70 MPa and 0–200 °C, and a RUSKA PVT System equipped with visual FC/FPC cells (operating range: 0–70 MPa, 0–200 °C) were utilized.

The experimental steps are as follows:

• Prepare formation crude oil at 51.8 °C under saturation pressure conditions, referencing the formation fluid phase behavior test report of Well Y;
• Inject CO₂ into the live oil until the current formation pressure is reached. Fully stir the mixture and allow it to stand for 12 h to ensure equilibrium;
• After standing, displace approximately 30 mL of the oil sample under constant pressure for a four-component analysis to characterize the initial fluid composition;
• Reduce the pressure in steps according to a preset pressure gradient of 3.2 MPa: decrease the pressure to 28.8 MPa in the first step, wait for flash vaporization to complete, and let the sample stand for another 12 h;
• Repeat Step 4 iteratively, reducing the pressure by 3.2 MPa each time, until the pressure drops to atmospheric pressure;
• Conduct a final four-component analysis of the sample to measure its asphaltene content. Test results are presented in Figure 4a,b.

It can be seen from the figures that during the pressure reduction process with production proceeding, the asphaltene content in the extracted heavy oil samples first decreases and then increases. When asphaltene content decreases, it is because, after CO₂ injection, CO₂ molecules more easily diffuse into crude oil, disrupting the stable dispersed distribution of asphaltenes in crude oil. Similarly, as pressure increases, more CO₂ dissolves in crude oil-high-pressure physical property experiments show that crude oil density decreases with CO₂ dissolution. As a large number of CO₂ molecules disperse in crude oil, due to polarity effects and molecular motion, CO₂ molecules gradually occupy the surfaces of asphaltene particles, replacing resins originally covering asphaltene surfaces. Exposed asphaltene molecules then attract and agglomerate with each other, forming large particle precipitates that settle at the bottom of the PVT cell, reducing asphaltene content in the sample. As pressure further increases, the asphaltene deposition tendency strengthens. After exceeding a certain pressure (approximately 16 MPa), the asphaltene deposition amount starts to decrease with weakened tendency, possibly because when the CO₂/crude oil system reaches miscibility, a further pressure increase leads to increased density of the miscible system, enhancing asphaltene dissolution capacity and reducing precipitated asphaltenes. Experiments show that the initiation pressure for asphaltene precipitation in the target block crude oil is approximately 25.6 MPa. When pressure drops to 16 MPa, asphaltene precipitation reaches a maximum of 2%, accounting for 44.44% of the original asphaltene mass, which is consistent with the asphaltene precipitation range (30–50%) of ordinary heavy oil. Therefore, the sensitivity of asphaltenes in the study area oil samples to CO₂ injection falls within the normal range. Meanwhile, the relationship between asphaltene precipitation saturation and mobility multiplier was obtained, as shown in Figure 5.

3. Simulation Setup and Model Validation

In order to further investigate the mechanisms of asphaltene precipitation and deposition during CO₂ injection and reveal the impact of asphaltene precipitation on development performance in reservoir exploitation, this study utilized the Petrel integrated simulation platform to construct an asphaltene precipitation model compatible with experimental results. An integrated numerical simulation study was then conducted for CO₂ flooding development in heavy oil reservoirs considering asphaltene deposition.

3.1. PVT Fitting

This study used the PVTi module in the phase behavior simulation software ECLIPSE (Version 2021.4) to fit and calculate the PVT experimental data and gas injection swelling experimental data of the crude oil in the Well B area. The fitting work for phase behavior experimental results mainly involves splitting heavy components in crude oil, component aggregation, and fitting the reconstructed component model to saturation pressure, differential liberation experiment data, single flash experiment data, and CO₂ injection swelling experiment data. The significance of this approach lies in both improving the speed and accuracy of simulation calculations by reducing the number of components (thereby avoiding non-convergence in simulations) and ensuring that the properties and attributes of the component model remain nearly unchanged before and after recomposition. Through fitting calculations, equation of state (EOS) characteristic parameters capable of characterizing the phase behavior of formation fluids under actual reservoir conditions can be obtained, providing basic data for subsequent component model construction.

• Pseudocomponent Division

Based on parameters such as molecular weight, critical pressure, critical temperature, and critical volume of well stream components, these were grouped into 7 pseudocomponents (original components are shown in Appendix A 

Table A1. Well Stream Composition Data.

Composition	Flash Oil Composition		Flash Gas Composition	Composition
	(mol%)	(wt%)	(mol%)	(mol%)	(wt%)
H2S	0.00	0.00	0.00	0.00	0.00
N2	0.00	0.00	0.00	0.00	0.00
CO2	0.00	0.00	0.00	0.00	0.00
C1	0.00	0.00	97.96	29.16	1.91
C2	0.00	0.00	0.89	0.26	0.03
C3	0.00	0.00	0.61	0.18	0.03
iC4	0.00	0.00	0.13	0.04	0.01
nC4	0.00	0.00	0.27	0.08	0.02
iC5	0.00	0.00	0.08	0.02	0.01
nC5	0.00	0.00	0.06	0.02	0.01
C6	5.93	1.46	0.00	4.16	1.43
C7	1.64	0.46	0.00	1.15	0.45
C8	1.21	0.38	0.00	0.85	0.37
C9	1.61	0.57	0.00	1.13	0.56
C10	2.18	0.86	0.00	1.53	0.84
C11	2.80	1.21	0.00	1.97	1.18
C12	3.73	1.76	0.00	2.62	1.73
C13	4.89	2.51	0.00	3.43	2.46
C14	5.10	2.84	0.00	3.58	2.78
C15	5.26	3.17	0.00	3.69	3.11
C16	4.88	3.17	0.00	3.43	3.11
C17	5.44	3.78	0.00	3.82	3.70
C18	3.47	2.55	0.00	2.44	2.51
C19	2.86	2.21	0.00	2.01	2.16
C20	2.83	2.28	0.00	1.99	2.24
C21	2.44	2.08	0.00	1.71	2.04
C22	2.18	1.95	0.00	1.53	1.91
C23	2.09	1.95	0.00	1.47	1.91
C24	2.03	1.97	0.00	1.42	1.92
C25	2.21	2.24	0.00	1.55	2.19
C26	2.40	2.52	0.00	1.68	2.47
C27	2.87	3.14	0.00	2.01	3.07
C28	2.75	3.13	0.00	1.93	3.06
C29	2.73	3.22	0.00	1.92	3.16
C30	2.12	2.59	0.00	1.49	2.54
C31	1.76	2.22	0.00	1.24	2.18
C32	1.34	1.75	0.00	0.94	1.71
C33	1.25	1.68	0.00	0.88	1.65
C34	1.22	1.68	0.00	0.85	1.64
C35	1.26	1.79	0.00	0.88	1.75
C36+	15.52	36.89	0.00	10.90	36.16
Total	100.00	100.00	100.00	100.00	100.00
molecular weight of C36+, 811
density of C36+, 0.9885 g/cm3

, and the recomposed components are shown in

Table 3. Composition of Pseudocomponents in Formation Crude Oil after Splitting and Recomposition.

Composition	CO2	C1	C2–C15	C16–C26	C27–C35	C36+	C40+
Mol fraction%	0	29.17	24.72	23.06	12.14	9.40	1.51

). According to the asphaltene precipitation experimental results above, the C40+ components with appropriate molar mass were classified as asphaltene components.

• Constant Composition Expansion Experiment Fitting

The constant composition expansion experiment refers to measuring the relationship between the volume of a fluid sample and pressure under reservoir pressure and temperature conditions with unchanged component composition, thereby obtaining basic phase behavior parameters such as the fluid’s bubble point pressure, relative volume at different pressures, fluid viscosity, density, etc., also simply referred to as PVT relationship testing. Based on the constant composition expansion experiment data, experimental fitting adjustments were performed, with fitting results shown in Figure 6. It can be seen from the figure that the variation law of the relative oil volume of the recomposed components with pressure is almost identical to that of the original components. The fitting error value is 0.5%, indicating a good fitting effect.

• Differential Liberation (DL) Experiment Fitting

The differential liberation (DL) experiment primarily measures parameters such as crude oil formation volume factor, relative density, crude oil viscosity, gas production volume at different pressures, gas compressibility factor, gas relative density, and natural gas viscosity. Fitting results are shown in Figure 7 and Figure 8,

Table 4. Crude Oil Solution Gas-Oil Ratio Fitting.

Pressure	Experimental Value of Solution Gas-Oil Ratio	Fitted Value of Solution Gas-Oil Ratio	Deviation
MPa	m3/m3	m3/m3	%
12.91	27.8	26.9	3.08
10	23.1	21.2	8.27
7	16.8	15.1	10.25
4	9.9	8.8	10.89

and

Table 5. Crude Oil Deviation Factor Fitting.

Pressure	Experimental Value of Deviation Factor	Fitted Value of Deviation Factor	Deviation
MPa	m3/m3	m3/m3	%
10	0.8778	0.8897	1.36
7	0.9061	0.9119	0.64
4	0.9427	0.9434	0.07

.

As shown in the figures, the solution gas-oil ratio of recomposed components is slightly lower than the value obtained from original component experiments, but the error remains within an acceptable range. The deviation factor shows excellent fitting at low pressures, but as pressure increases, the deviation between fitted and experimental values gradually widens. This issue may arise from inherent deviations in phase equilibrium prediction by the selected Peng-Robinson 3 Parm (PR3) equation of state at high pressures or from the loss of high-pressure phase behavior details due to insufficient pseudocomponent division. Since the error values are within acceptable limits and further adjustment without affecting other fitting results is difficult, the current optimized results are compromisely adopted.

In PVTi data fitting, since the parameters for adjusting viscosity differ from those of other experiments, to avoid mutual interference between fitting results, crude oil viscosity is generally fitted separately after other experimental parameters are completed. In this fitting, the Lohrenz-Bray-Clark (Modified) Formula (1) was used for viscosity calculation. Fitting results are shown in Figure 9a,b. It can be seen from the figures that both crude oil viscosity and gas viscosity exhibit good fitting effects. The fitted values of gas viscosity are slightly lower than the experimental values overall, possibly due to insufficient applicability of the LBC model for specific components or pressure ranges. It may also be caused by the difficulty of measuring high-pressure gas viscosity, potential systematic errors in experimental equipment (such as vibrating string viscometers), or inadequate control of experimental conditions (temperature, pressure). Therefore, further parameter optimization is unnecessary.

The calculation expression for viscosity is:

$$\eta =A\cdot \exp (B/(T-C))$$

(1)

• CO₂ Injection Swelling Experiment Fitting

The fitting of the CO₂ injection swelling experiment primarily involves adjusting the fluid component model based on the experimental test data of crude oil viscosity under different CO₂ injection amounts. Finally, the corresponding experimental fitting comparison is completed as shown in Figure 10, with good fitting results.

3.2. Asphaltene Deposition Model

To characterize the impact of asphaltene precipitation on the reservoir during gas injection, this paper uses a three-component model to simulate the process, with all content in this section referenced from the Eclipse Software (Version 2021.4) Manual [34]. The principles of the three-component model are detailed in Appendix B.

Using the experimental results from a series of experiments in Section 2.1, Section 2.2, Section 3 and Section 3.1, the code parameters required for the three-component model are defined. First, the ASPHALTE code is activated, which primarily consists of strings specifying characterization conditions, strings specifying the permeability impairment model, and strings specifying the viscosity model. The characterization condition string defaults to the WIGHT string, with data specified using single-variable codes (ASPP1P, ASPREG) or two-variable codes (ASPPP2P, ASPPW2D). This paper uses single-variable codes, which specify the relationship between pressure and the molar mass percentage of asphaltene dissolved in crude oil: the ASPP1P code designates the corresponding property “pressure,” and the ASPREG code specifies the specific relationship (displayed in a list).

The codes for specifying the permeability impairment model use the three-component model corresponding to SOLIDMMS or SOLIDMMC, which describes solid adsorption caused by changes in fluid mobility. The difference lies in that the two codes correspond to the mobility coefficient and the concentration/saturation of solids adsorbed by the formation, respectively. As the concentration/saturation of solids adsorbed by the formation increases, the mobility coefficient gradually decreases. Since a saturation experiment was conducted in the previous section, the SOLIDMMS code is selected to define the asphaltene model.

The ASPVISO code provides data for modeling oil viscosity changes during asphaltene precipitation. Three models are used to describe viscosity changes:

• Generalized Einstein Model (Single-Parameter): Specifies the slope of relative viscosity as a function of concentration to describe viscosity changes;
• K & D Model (Two-Parameter): Specifies the relationship between mass concentration at maximum packing and intrinsic viscosity to describe viscosity changes;
• Direct Relationship Table: Provides a table of the relationship between the mass fraction of asphaltene precipitate and the oil viscosity multiplier (i.e., the oil viscosity multiplier as a function of the mass fraction of asphaltene precipitate).

After defining all the above codes in combination with the experimental results and PVT fitting results, a unique asphaltene deposition model for the target oil sample in Well Block Xia 018 is formed. However, before using it for numerical simulation, the asphaltene deposition model must first be verified.

Shown as

Table 6. Mapping relationship between asphaltene code settings and experimental data.

Asphaltene Code	Corresponding Experimental Data	Charts of Experimental Results	Determination Method
ASPP1P	Results of Asphaltene Precipitation Experiments in Section 2.2	Figure 4	Directly specify the selection of Pressure as the research variable
ASPREG	Results of Asphaltene Precipitation Experiments in Section 2.2	Figure 4	According to the results of Figure 4, directly present the relationship between pressure and the percentage of asphaltene molar mass
SOLIDMMS	Results of Asphaltene Precipitation Experiments in Section 2.2	Figure 5	According to the results of Figure 5, directly present the relationship between Asphaltene Precipitation Saturation and the Mobility Multiplier
ASPVISO	Results of Asphaltene Precipitation Experiments in Section 2.1 and Section 3.1	Figure 9 and Figure 10	Select the third viscosity damage model, and directly present the relationship table between the mass fraction of asphaltene precipitates and the oil viscosity multiplier based on the experimental results in Figure 9 and Figure 10

is the mapping relationships between model parameters and experimental data.

3.3. Model Validation

Core flooding experiments simulate reservoir conditions using field cores to evaluate EOR performance under realistic formation conditions through short- and long-core tests. Optimal injection parameters are determined by varying huff-n-puff parameters, providing guidance for developing similar reservoir exploitation strategies.

During CO₂ flooding of crude oil-saturated cores, CO₂ reacts with the crude oil, causing asphaltene precipitation that clogs pore throats and increases injection pressure. Based on this phenomenon, this study validated the asphaltene deposition model by conducting CO₂ flooding experiments on oil-saturated cores and monitoring changes in injection pressure.

3.3.1. Core Flooding Experiment

Based on a high-temperature and high-pressure core flooding experimental setup, high-purity CO₂ was used as the injection medium to conduct core flooding experiments. The experimental process of saturating short cores first and then performing flooding better simulates actual reservoir production processes. This study analyzed pressure evolution at the core scale, gas-injection agent development characteristics, and the distribution pattern of remaining oil.

• Experimental Equipment and Materials

Materials: 200 mL heavy oil sample from Well Block Xia 018; CO₂ with a purity of 99.99%; Field cores, etc.

Equipment: Constant-speed and constant-pressure pump: pressure range 0–70 MPa, temperature range ambient—200 °C, flooding rate 0.001–20 mL/min; Core holder: diameter 2.5 cm, length 10 cm; Piston container: volume 1 L, pressure range 0–70 MPa, temperature range ambient—200 °C; High-pressure pipelines: pressure range 0–70 MPa, temperature range ambient—200 °C; Pressure gauge: accuracy 0.01 MPa (as shown in Figure 11).

• Experimental Procedures
  • Cores were cleaned, dried, and placed into the core holder, followed by a tightness test. The imbibition method was used to saturate formation water, and the pore volume and porosity of the core assembly were measured;
  • The backpressure valve was set to formation pressure, and formation water was used to raise the core pressure to reservoir pressure. Formation water was injected into the short core at a rate of 0.01 mL/min for water saturation. When the pressure stabilized and continuous water production was observed at the outlet, water-phase saturation was considered complete;
  • Water was injected into the core through pre-fixed pipelines with a small diameter to generate high pressure. The high-pressure water flow-induced fracturing to achieve the expected effect;
  • The core setup was placed in a thermostatic chamber, and dead oil was injected into the core at a rate of 0.01 mL/min for oil saturation. When the pressure stabilized and continuous oil production was observed at the outlet, the core was considered fully oil-saturated. Approximately 2.0 pore volumes (PV) of dead oil were required, and the initial oil saturation of the core was calculated simultaneously;
  • CO₂ gas was injected into the core at a rate of 0.5 mL/min for CO₂ flooding. Each core was flooded for 24 h, with data such as pressure, liquid production, oil production, and gas production recorded every 2 h.
• Experiment Result

As shown in Section 3.3.2 Figure 12, taking Samples 1, 2, and 3 as examples, their injection pressure changes over 24 h were recorded. The figure indicates that the injection pressure remained initially stable; as injection time increased, it began to rise around 10 h, peaked at approximately 15 h, and finally dropped rapidly to a constant value. This behavior is attributed to the following: when CO₂ is injected, it first destabilizes asphaltene components, causing asphaltene precipitation that partially clogs pore throats, leading to a rapid increase in injection pressure. As the injection pressure further increases, CO₂ becomes miscible with the crude oil, increasing asphaltene solubility and unclogging the blocked pore throats, causing the injection pressure to quickly return to a normal level. However, due to the irreversibility of the damage, the injection pressure after unclogging remains slightly higher than before the clogging occurred.

3.3.2. Core Flooding Numerical Simulation

After successfully establishing the asphaltene deposition model, the calibrated fluid composition model was imported into Petrel, forming an integrated numerical simulation technology that accounts for the coupling of multiple physicochemical processes such as asphaltene precipitation and migration. Before applying this technology, the accuracy of the model was first validated.

A core-scale mechanistic model matching the cores used in the flooding experiments was built using the Petrel platform. The model was parameterized with physical property data from the core samples, including average permeability, average porosity, and dimensions (length and diameter). The final properties of the mechanistic model were determined as follows: total length 6 cm, diameter 2.5 cm, porosity 0.2, permeability 100 mD, and grid size 25 mm × 25 mm × 1 mm.

The established mechanistic model is shown in Figure 13. To avoid convergence issues, reduce computation time, and simplify the calculation process, a 1D (stratified only in the k-direction) rectangular mechanistic model was used for simplification. The core-scale model was vertically divided into 60 layers, with three horizontal wells evenly spaced at the bottom—each equipped with 3 uniform perforation clusters—serving as CO₂ injection wells to simulate the gas injection end of the core flooding experiment. At the top, nine production wells with uniform spacing were used for oil production, simulating the fluid outlet end of the experiment. The pre-imported and calibrated composition model was applied to the defined simulation scheme, and asphaltene deposition codes were added to activate the asphaltene deposition model. To match the actual gas injection rate and flooding time in the experiments, the numerical simulation was set with a constant injection rate of 0.5 mL/min for 24 h, recording results every 1 h. After configuration, the simulation scheme was run. The flooding results are shown in Figure 14.

As shown in Figure 12, as the flooding simulation progresses, CO₂ injected from the bottom displaces crude oil upward throughout the core, corresponding to a decrease in oil saturation starting from the lower part of the core. After 16 h of flooding, oil saturation stabilizes, indicating most of the crude oil in the core model has been displaced. During the simulation, uneven oil saturation distribution occurs in the middle section of the core model, likely due to asphaltene deposition—initially treated as a solid phase (unrecognized by oil saturation properties)—reverting to a free asphaltene state (recognized as a liquid component by oil saturation properties) due to the unclogging effect. Additionally, a comparison of injection pressure between the flooding experiment and core flooding simulation is shown in Figure 12. The figure demonstrates that simulation results without considering the asphaltene deposition model exhibit significant discrepancies with experimental data, whereas results incorporating the asphaltene deposition model show good agreement with experimental results. This preliminarily validates the effectiveness of the asphaltene deposition model.

4. Field Applications

Based on the previous descriptions of the block’s formation and oil sample characteristics, the crude oil in Well Block Xia 018 exhibits high viscosity and poor reservoir physical properties. The field plans to enhance production in Well Y through reservoir fracturing stimulation. Determining multiple construction and production parameters for reservoir fracturing requires integrating a reservoir mechanistic model with actual fracturing pumping schedules. Therefore, this chapter establishes a numerical simulation model that incorporates both the compositional model and asphaltene deposition model, conducting a systematic study on parameter optimization and screening. Based on this, optimal construction and production parameters are selected, and preliminary production recommendations are provided.

4.1. Mechanistic Model Development

To simplify calculations, demonstrate the workflow, and eliminate the influence of formation property heterogeneity, this chapter uses a computationally efficient mechanistic model for the study.

Based on the location of the target horizontal well Y, suitable regional size and thickness were defined to form a rectangular block measuring 1600 m × 900 m × 50 m, with a grid size of 20 m × 20 m × 5 m and a total of 36,520 grids. Property modeling for the mechanistic model was referenced from the geological data provided by the field, using average values. Key properties required for fracturing simulation and subsequent production simulation include porosity, permeability, water saturation, oil saturation, and geomechanical parameters (Poisson’s ratio, Young’s modulus, maximum and minimum horizontal principal stresses, overburden pressure, pore pressure, etc.), as listed in

Table 7. Key Parameters of Reservoir Mechanistic Model for Well Y in Xia 018 Well Block.

Porosity	0.178	Poisson’s Ratio	0.34
Permeability (mD)	180	Young’s Modulus (GPa)	71
Water Saturation	0.4	Maximum Horizontal Principal Stress (bar)	308.7
Oil Saturation	0.6	Minimum Horizontal Principal Stress (bar)	238.6
Overburden Stress (bar)	249.5	Pore Pressure (bar)	149.9

.

The established mechanistic model is shown in Figure 15, where Well Y is depicted as a horizontal well, preparing for the creation of zonesets in subsequent fracturing simulations.

4.2. Fracturing and Production Optimization

Both fracturing operations and production prediction simulations require determining numerous parameters, whose selection ranges are primarily based on field data combined with operational experience. Under normal conditions, the fracturing parameters to be optimized include cluster spacing, fracturing stage length, fluid volume, proppant volume, CO₂ injection volume, injection rate, well spacing, etc.; production regime parameters include daily oil (gas) production and production time. Since Well Y already has existing perforation data, cluster spacing will not be adjusted in this optimization.

Given the large number of fracturing parameters to optimize, an orthogonal array design was chosen to ensure uniform dispersion of parameters and minimize the number of experiments while enabling main effect analysis. The fracturing parameters consist of 5 factors with 4 levels each and 1 factor with 5 levels (well spacing), which cannot be directly accommodated by standard orthogonal arrays. Instead, the pseudo-level method was applied: using the standard orthogonal table L₁₆(4⁵) (16 experiments for 5 factors with 4 levels), the 6th factor (well spacing, 5 levels) was integrated into the table through pseudo-level adjustment, forming L₂₀(4⁵ × 5¹) to maintain orthogonality of main effects. For details, the designed experimental schemes are listed in Appendix A2. The final optimized fracturing parameters are fracturing stage length of 1000 m, fluid volume per stage of 100 m³, proppant volume per stage of 100 m³, CO₂ injection volume per stage of 110 m³, injection rate of 14 m³/min, and well spacing of 260 m.

After optimizing fracturing parameters with SRV (Stimulated Reservoir Volume) as the target, the selected parameter set was further used to optimize production regimes with cumulative oil production as the objective.

Production simulation results (Figure 16, Figure 17, Figure 18 and Figure 19) show that as daily oil production increases, cumulative oil production rises gradually, while stable production time decreases and the pressure depletion radius in the well block expands. Since stable production cannot be sustained, increasing daily production beyond 15 m³/d leads to negligible growth in total production; thus, 15 m³/d is selected as the optimal daily oil production rate. As production years increase, pressure contours at the end of each period indicate an expanding depletion area. While cumulative oil production continues to grow, the average annual oil production declines progressively. Comprehensively considering input costs and economic benefits, 15 years is determined as the optimal production time.

4.3. Case Well Production Simulation

Using the production regime optimized in Section 4.2, a constant-production-rate prediction simulation was conducted for Well Y with a set daily production rate of 15 m³/day and a production duration of 15 years. The code for the asphaltene deposition model was activated during the simulation, and the results are shown in Figure 20. As indicated in the figure, with the progression of the production simulation, CO₂ in the fracturing fluid reacts with heavy oil in the formation, causing asphaltene deposition in the wellbore vicinity. This leads to a decrease in reservoir permeability and the formation of local abnormal high pressure in the near-wellbore region, as shown in Figure 20b,c. However, as the simulation proceeds, the local abnormal high pressure gradually decreases. It is inferred that the increased pressure promotes miscibility between CO₂ and asphaltene, enhancing asphaltene solubility and causing partial debottlenecking. Nevertheless, the reduced reservoir permeability cannot fully recover—by the end of 2039 (the final production stage), the pressure in the near-wellbore region remains slightly higher than in other areas, as shown in Figure 20d.

A comparative simulation without activating the asphaltene deposition model was also performed, with results compared in Figure 21 and Figure 22. Without the asphaltene model (which does not account for asphaltene-induced damage), reservoir permeability remains unchanged, and the simulated curves for daily oil production, bottomhole flowing pressure, and cumulative oil production follow normal constant-production-rate trends. When the asphaltene deposition model is activated, bottomhole flowing pressure decreases more slowly, and daily production declines due to simulated permeability damage from asphaltene. Asphaltene plugging reaches its maximum severity by mid-2026, after which debottlenecking occurs and bottomhole pressure decreases. However, due to the irreversibility of permeability reduction (modeled by setting the flocculation rate higher than the dissolution rate in the asphaltene deposition model), bottomhole pressure remains slightly higher than in the non-asphaltene model scenario. Daily production is also lower than the non-asphaltene case, resulting in lower cumulative production: the standard scenario yields 13,000 m³, while the asphaltene-inclusive scenario yields 11,700 m³. For field production guidance, periodic debottlenecking operations may be required between 2026 and 2029 to maintain stable well productivity and improve ultimate recovery. From 2029 to 2039, the dosage and frequency of debottlenecking agents can be reduced.

5. Conclusions

This study addresses the challenges of production enhancement in heavy oil reservoirs and damage caused by asphaltene deposition. Adopting a technical approach comprising “mechanism experiments—model development—parameter optimization—field validation,” it investigates the reaction mechanisms of CO₂ injection into oil samples from the target block. Component and asphaltene deposition models specific to the oil samples were developed, with the accuracy of the asphaltene deposition model validated through core flooding experiments and numerical simulations. Fracturing and production parameters for the target area were optimized and successfully applied in field operations, leading to the following key findings:

• This study reveals the duality of the interaction mechanism between CO₂ and heavy oil. Experiments confirm that CO₂ reduces heavy oil viscosity through molecular diffusion and expands crude oil volume, significantly enhancing fluidity. However, CO₂ dissolution disrupts the resin-asphaltene equilibrium: asphaltene precipitation is triggered when pressure drops below 25.6 MPa, reaching a peak of 2% at 16 MPa. Controlling the pre-injected CO₂ volume in parameter optimization is essential to avoid reducing fracture conductivity.
• A research approach integrating the asphaltene deposition model into a unified simulation was proposed. PVT parameters were fitted using experimental data and the ECLIPSE-PVTi module. An asphaltene code was defined and applied to construct a three-component asphaltene deposition model, with its reliability validated through dual verification of core flooding experiments and numerical simulation. Enabling the asphaltene model in subsequent unified simulation cases improved prediction accuracy, demonstrating significant practical value;
• Collaborative optimization of fracturing operation parameters is completed. Based on the established orthogonal array, the highest unit-length Stimulated Reservoir Volume (SRV) was achieved at a fracturing stage length of 1000 m. Combining economic evaluation, the optimal parameter combination was determined as: fluid volume per stage of 1000 m³, proppant volume of 1000 m³, and injection rate of 14 m³/min. A design framework for similar reservoirs is provided;
• Conducted production simulation and prediction for case wells using optimized fracturing parameters and production regimes, integrated with the validated component and asphaltene models. A 3D geomechanical model was built for Well Y in the Xia 018 block, and simulations were performed using the optimized parameters. Compared with the scenario without the asphaltene deposition model, the 15-year predicted cumulative oil production was 1.17 × 10⁴ m³, a 10% decrease. Pressure contour plots of the well area revealed that the combined effect of declining pore pressure and CO₂ during production would induce asphaltene precipitation in the near-well region. These results guide planning debottlenecking timing in field operations, demonstrating significant practical significance.