Summary
Miscible gas injection is the most widely applied enhanced oil recovery (EOR) method in light oil carbonate reservoirs as a tertiary and secondary method. Miscible gas has high displacement efficiency and usually results in a low residual oil saturation (Sorm) in the parts of the reservoirs that are in contact with the gas. Accurate determination of Sorm and understanding the parameters that affect displacement efficiency are crucial for successful miscible gas EOR projects.
In this paper, we present a comprehensive experimental program designed to investigate the effect of a number of parameters on oil recovery, displacement efficiency, and Sorm of miscible and near-miscible carbon dioxide (CO2) injection. The parameters investigated in this study are the experimental pressure, pore volume (PV) injected, injection rate, rock type, and initial water saturation (Swi). The coreflood experiments were performed using live crude oil at pressures starting below the minimum miscibility pressure (MMP) to pressure well above the MMP, using reservoir core samples of up to 1 ft long and 2 in. diameter. All CO2 injection experiments were performed using vertically oriented cores, with gas injection from the top to ensure stable displacement.
The experimental results show that (1) Oil recovery decreases as pressure decreases with Sorm increasing by more than 20 saturation units as the pressure decreases from 4,250 psi to 2,700 psi; (2) CO2 breakthrough was much earlier at lower pressure, which leads to more CO2 recycling and potentially lower CO2 sequestration volume; (3) the recovery factor (RF) is strongly affected by the PV injected, and this effect is much more significant for the experiments performed at lower pressure; (4) the injection rate has an insignificant impact on oil recovery and Sorm for miscible or near-miscible CO2, due to the low interfacial tension (IFT) between oil and CO2; (5) rock heterogeneity has a strong effect on oil recovery and CO2 breakthrough and hence on CO2 recycling and economy of the projects; and (6) the presence of mobile water at the beginning of CO2 injection resulted in lower displacement efficiency and increased Sorm. However, this water blocking effect should be determined experimentally for a given reservoir rock/fluid system. The results of this study cannot be generalized for other reservoirs.
The results of this study have important implications for the design and performance predictions of CO2 injection in the reservoirs under study. Starting CO2 injection at reservoir pressure, which, in some cases, is more than 1,500 psi above MMP, is recommended due to its superior displacement efficiency and less CO2 recycling due to later breakthrough. However, a higher pressure may negatively impact the required CO2 volume, the compression cost, and project economics.
Introduction
Miscible gas injection, particularly using CO2, is a proven and economically viable EOR method. It is the most widely applied EOR technique, especially in light oil reservoirs, and has been implemented as a secondary and tertiary recovery method in carbonate and sandstone reservoirs.
The primary objective for miscible gas injection is to improve the displacement efficiency, or microscopic sweep efficiency, and reduce residual oil saturation (Sor) to levels below the values typically obtained in waterflooding. However, gas injection suffers from poor sweep efficiency due to low gas viscosity (high mobility), gravity override, and viscous fingering, leading to unfavorable mobility ratios and premature gas breakthrough. Furthermore, the macroscopic sweep of gas injection is further diminished in highly heterogeneous reservoirs, where the uneven flow paths exacerbate the challenges of gravity segregation and viscous fingering. These sweep efficiency challenges are critical barriers that must be addressed to maximize the effectiveness of miscible gas injection as an EOR technique. However, this study mainly focuses on displacement efficiency, as proper evaluation of sweep efficiency requires combining laboratory experiments with reservoir-scale simulation.
Miscible gas injection, such as CO2 or rich hydrocarbon gas, is usually operated so the reservoir remains above the MMP. Miscible gas or water-alternating-gas (WAG) injection (CO2 or rich hydrocarbon gases) can theoretically achieve 100% microscopic displacement efficiencies. However, several practical constraints can limit the design and implementation of truly miscible gas injection:
Gas Availability and Cost: The availability and cost of the required gas volumes can be limiting factors.
Reservoir Pressure: Maintaining the reservoir pressure above the MMP may not always be feasible, especially in mature fields.
Gas Processing Capacity: Lack of sufficient gas processing capabilities, particularly in offshore environments, can restrict the design of miscible gas injection projects.
Environmental Constraints: Environmental regulations and concerns, especially for offshore operations, can impose additional constraints on using miscible gases.
These practical limitations often result in the need to operate in a near-miscible or even immiscible regime, leading to lower displacement efficiency and residual oil saturation (Sor) higher than usually measured for miscible gas injection (Sorm). Moreover, even in 1D laboratory experiments conducted under miscible conditions, residual oil saturations of up to 10% or higher have been reported in the literature [see Lange (1998) and Masalmeh et. al. (2023) and the references therein]. Given these practical limitations and the observed residual oil saturations, understanding the key parameters affecting the displacement efficiency of gas injection is paramount for the successful application of miscible gas EOR. This is the primary focus of the current study.
The study by Shyeh-Yung (1991) showed that for CO2 flood, Sorm increases linearly (oil recovery decreases) as pressure decreases, and no dramatic loss of recovery is observed below the MMP, contrary to slimtube test results, which usually show a significant change in the slope of oil recovery curve. The presence of water in the core appears to reduce the effect of pressure on oil recovery. The work also highlighted that two oil recovery mechanisms, low IFT displacement and extraction of oil components, were necessary. This was deduced from the analyses of produced oil compositions, and it was concluded that lower IFT (higher pressure) and lower initial water saturation enhance displacement efficiency and reduce pore level bypassing. Extraction becomes a more dominant mechanism after solvent breakthrough, and extraction increases with pressure and decreases with water saturation. The study suggests that reservoir-condition coreflood experiments should be used instead of slimtube tests whenever possible, as they provide more realistic data on pressure sensitivity, miscible flood residual oil saturation, and solvent mobility.
Contrary to the above findings, experimental observation in the study by Kamali et al. (2015) showed that almost similar oil recovery is achieved during miscible and near-miscible displacements. In contrast, a significant reduction in oil recovery is recorded in the immiscible displacement. In this study, the experiments at a pressure just below MMP (near-miscible) and above MMP showed no measurable difference in oil recovery.
Kazemi et al. (2015) investigated the impact of initial water saturation on bypassed oil recovery during CO2 injection at different miscibility conditions. They found that the maximum oil recovery always occurs at near-miscible conditions independent of initial water saturation. They also found that the presence of water at near-miscible and first-contact-miscible regions decreased the oil recovery factor. The study was performed under water-wet conditions, decane was used as the oil phase, and gas injection was performed horizontally. The gravity effect is expected to be significant in their study due to the large density difference between the oil phase and CO2.
Stern (1991) carried out a series of tertiary multiple-contact miscible CO2 injection experiments to study the effect of flow rate, core length, oil viscosity, wettability, WAG ratio, and initial water saturation on displacement mechanisms. The study concluded that wettability significantly affects oil recovery by WAG or tertiary CO2 injection after waterflooding. A significant water blocking effect is observed in water-wet rocks compared to mixed-wet or oil-wet rocks, though water blocking is still observed in mixed-wet rocks. Water shielding was also investigated in several other studies, such as Spence and Ostrander (1983), Huang and Holm (1988), Lin and Huang (1990), Shyeh-Yung (1991), Rao et al. (1992), and Yeh et al. (1992). They all concluded that a significant water blocking effect is observed for water-wet Berea cores, and the effect decreases for mixed-wet and oil-wet cores.
The length of the core sample to be used in miscible gas injection experiments was also investigated in the literature. Shelton and Schneider (1975) observed only a small effect on core length when using cores between 1 ft and 8 ft, while Yellig (1982) and Negahban et al. (1990) found that recovery is higher in longer cores. Negahban et al. (1990) concluded that cores longer than 10 ft are required to develop miscibility, which is similar to conclusions reached earlier by Rathmell et al. (1971). On the other hand, both Stern (1991) and Shyeh-Yung (1991) concluded that for experiments performed at a pressure well above MMP, miscibility develops quickly over a distance of less than 1 ft. Therefore, they found a negligible effect on Sorm in core lengths ranging from 1 ft to 32 ft.
The effect of flow rate in both tertiary and secondary CO2 injection has been investigated in the literature. In secondary floods, Shelton and Schneider (1975), Tiffin and Kremesec (1988), and Watkins (1978) observed an increase in Sorm with flow rate. They suggested that this results from increased viscous fingering at higher flow rates, which leads to the bypassing of oil. Here, viscous fingering refers to macroscopic bypassing caused by core-scale heterogeneity. On the other hand, Jones (1985) found that, in tertiary floods, Sorm decreased with increasing flow rate, suggesting that this resulted from capillarity-induced bypassing. The capillary entry pressure is higher in small pores, so large pores are entered first, resulting in pore-level bypassing. This effect is absent in secondary floods, where water is not mobile, as there are no capillary forces during miscible gas displacing oil. Sehbi et al. (2001) analyzed factors affecting microscopic displacement efficiency in CO2 floods. They concluded that lower injection rates and higher residence time increase mass transfer between the oil and CO2, reducing Sorm. Ajoma and Sungkachart (2021) concluded that gravity effects are more significant at low-rate CO2 injection than at high-rate CO2 injection, leaving more pores unswept. When examining their experiments, we believe that their conclusions are affected by the fact that high-permeability core samples were used in the study, the injection was performed in the horizontal direction, and the density difference between CO2 and oil was significant.
This study investigated the impact of several parameters on displacement efficiency, Sorm, and oil recovery of gas injection, such as the effect of pressure, PV injected, initial water saturation, flow rate, core length, and rock type. The coreflood experiments were performed using live crude oil at pressures starting below MMP to pressure well above MMP, using reservoir core samples of up to 1 ft long and 2 in. diameter. All gas injection experiments were performed using vertically oriented cores, with gas injection from the top to ensure stable displacement.
Our aim in this study is to understand better the parameters that influence displacement efficiency, residual oil saturation, and overall oil recovery during miscible and near-miscible CO2 injection. This knowledge can help operators to optimize the design and implementation of gas-based EOR techniques. Maximizing displacement efficiency and minimizing residual oil saturation are critical for improving the economic viability of CO2 flooding projects, which can be capital-intensive. The insights gained can help operators make more informed project feasibility and design decisions.
While limited in scale, laboratory experiments can provide valuable data on the fundamental mechanisms and sensitivities involved in miscible CO2 displacement. Integrating these findings with reservoir-scale simulations helps bridge the gap between laboratory and field performance. The data generated in this study provide valuable input to improve the accuracy and predictive capabilities of reservoir models and simulation tools used for CO2 flooding evaluation and optimization.
In summary, a comprehensive investigation of the factors affecting displacement efficiency and residual oil during miscible CO2 injection is essential for advancing the technical and economic success of this critical EOR method. The insights gained can improve project design, operations, and overall CO2 flooding performance.
Core Materials and Fluid Properties
Reservoir Cores
Five reservoir cores used in the coreflood experiments were obtained from carbonate reservoirs RA, DY (two samples), BB, and SB (see properties in Table 1 ). The cores were cut vertically to obtain long samples that are suitable for miscible gas injection. The cores were relatively high-purity limestone, containing less than 2% clay and 4% silica.
Properties of cores RA1, DY1, DY2, BB1, and SB1 used in this work.
Properties . | RA1 . | DY1 . | DY2 . | BB1 . | SB1 . |
---|---|---|---|---|---|
Diameter (in.) | 2.0 | 2.0 | 2.0 | 1.98 | 2.01 |
Length (ft) | 0.8 | 0.57 | 0.98 | 0.57 | 0.67 |
Porosity, ϕ (%) | 22.7 | 18.0 | 19.4 | 28.0 | 30.5 |
Permeability to brine, k (md) | 1.1 | 0.8 | 1.9 | 9.8 | 3.1 |
Properties . | RA1 . | DY1 . | DY2 . | BB1 . | SB1 . |
---|---|---|---|---|---|
Diameter (in.) | 2.0 | 2.0 | 2.0 | 1.98 | 2.01 |
Length (ft) | 0.8 | 0.57 | 0.98 | 0.57 | 0.67 |
Porosity, ϕ (%) | 22.7 | 18.0 | 19.4 | 28.0 | 30.5 |
Permeability to brine, k (md) | 1.1 | 0.8 | 1.9 | 9.8 | 3.1 |
To prepare the cores for the experiments, they were thoroughly cleaned with toluene and methanol injection. Each core was loaded into a high-pressure core holder, and its PV was measured. Later, the cores were fully saturated with formation water (FW), and permeability measurements were performed.
Fluids
The crude oils from the four different reservoirs (labeled RA, DY, BB, and SB) were centrifuged to separate possible water content before being injected into the core. The formation brines were prepared volumetrically at atmospheric conditions, stirred, and degassed before testing. The compositions and properties of all brines are given in Table 2 . The saturates, aromatics, resins, and asphaltenes (SARA) analysis of the crude oils used in the study is presented in Table 3 .
Formation brine compositions used in coreflood experiments on RA, DY, BB, and SB reservoirs.
Salt Component (ppm) . | RA . | DY . | BB . | SB . |
---|---|---|---|---|
Na | 58,454 | 49,898 | 49,933 | 51,935 |
Ca | 15,369 | 14,501 | 14,501 | 16,522 |
Mg | 1,409 | 3,248 | 3,248 | 3,055 |
K | 0 | 1,990 | 0 | 0 |
SO4 | 557 | 234 | 234 | 335 |
HCO3 | 0 | 162 | 162 | 174 |
Cl | 121,071 | 109,830 | 111,885 | 111,810 |
Total dissolved solids | 196,818 | 179,853 | 179,963 | 183,831 |
Salt Component (ppm) . | RA . | DY . | BB . | SB . |
---|---|---|---|---|
Na | 58,454 | 49,898 | 49,933 | 51,935 |
Ca | 15,369 | 14,501 | 14,501 | 16,522 |
Mg | 1,409 | 3,248 | 3,248 | 3,055 |
K | 0 | 1,990 | 0 | 0 |
SO4 | 557 | 234 | 234 | 335 |
HCO3 | 0 | 162 | 162 | 174 |
Cl | 121,071 | 109,830 | 111,885 | 111,810 |
Total dissolved solids | 196,818 | 179,853 | 179,963 | 183,831 |
SARA analysis of the crude oils used in this study.
Crude Oil . | Saturates . | Aromatics . | Resins . | Asphaltenes . |
---|---|---|---|---|
RA | 73.10 | 20.00 | 6.50 | 0.40 |
DY | 63.54 | 26.88 | 8.79 | 0.79 |
BB | 68.90 | 25.90 | 4.80 | 0.40 |
SB | 73.08 | 19.12 | 7.63 | 0.17 |
Crude Oil . | Saturates . | Aromatics . | Resins . | Asphaltenes . |
---|---|---|---|---|
RA | 73.10 | 20.00 | 6.50 | 0.40 |
DY | 63.54 | 26.88 | 8.79 | 0.79 |
BB | 68.90 | 25.90 | 4.80 | 0.40 |
SB | 73.08 | 19.12 | 7.63 | 0.17 |
The composition (mole fraction) of the produced gases of the reservoirs is shown in Table 4 . The live oil was prepared by recombining the dead crude oil with the associated gas. The mixing process was performed at the corresponding reservoir pressure and temperature of each reservoir. It was conducted to match the gas oil ratio (GOR), oil formation volume factor, bubble point, and viscosity of the live fluid for RA, DY, BB, and SB reservoirs. As shown in Table 4 , the produced gas contains light and intermediate components (C2–C10). Thus, the recombined oils were representative of the real reservoir fluids.
Composition of the produced gases of RA, DY, BB, and SB reservoirs used in this work.
. | RA . | DY . | BB . | SB . |
---|---|---|---|---|
Component | Concentration (mol%) | |||
CO2 | 5.10 | 6.13 | 3.50 | 3.32 |
C1 | 58.91 | 66.5 | 70.54 | 53.63 |
C2 | 12.37 | 12.19 | 10.02 | 14.31 |
C3 | 9.81 | 7.59 | 7.45 | 13.23 |
i-C4 | 2.41 | 1.49 | 1.46 | 2.99 |
n-C4 | 5.11 | 2.64 | 3.58 | 6.54 |
i-C5 | 1.10 | 0.87 | 1.06 | 2.10 |
n-C5 | 2.60 | 0.85 | 1.60 | 2.04 |
C6 | 1.44 | 0.69 | 0.61 | 1.39 |
C7 | 0.73 | 0.98 | 0.11 | 0.42 |
C8 | 0.29 | 0.05 | 0.07 | 0.03 |
C9 | 0.13 | 0.02 | 0 | 0 |
. | RA . | DY . | BB . | SB . |
---|---|---|---|---|
Component | Concentration (mol%) | |||
CO2 | 5.10 | 6.13 | 3.50 | 3.32 |
C1 | 58.91 | 66.5 | 70.54 | 53.63 |
C2 | 12.37 | 12.19 | 10.02 | 14.31 |
C3 | 9.81 | 7.59 | 7.45 | 13.23 |
i-C4 | 2.41 | 1.49 | 1.46 | 2.99 |
n-C4 | 5.11 | 2.64 | 3.58 | 6.54 |
i-C5 | 1.10 | 0.87 | 1.06 | 2.10 |
n-C5 | 2.60 | 0.85 | 1.60 | 2.04 |
C6 | 1.44 | 0.69 | 0.61 | 1.39 |
C7 | 0.73 | 0.98 | 0.11 | 0.42 |
C8 | 0.29 | 0.05 | 0.07 | 0.03 |
C9 | 0.13 | 0.02 | 0 | 0 |
Table 5 lists the results of PVT properties experimentally measured for the recombined live oils at reservoir conditions. The MMP of CO2 with the reservoir fluids is also shown in Table 5 , and the data were measured using slimtube experiments (see Alshuaibi et al. 2019). In our research, the MMP was defined as the pressure at the linear intersection point of the measured recovery factor plotted against injection pressure data within both immiscible and miscible pressure ranges, a criterion commonly referenced in the existing literature. There are varying opinions in the literature regarding the appropriate criteria for determining MMP from slimtube experiments. A comprehensive discussion on this subject can be found in the work of Zhang and Gu (2015).
PVT properties of the recombined fluids (RA, DY, BB, and SB) used in these coreflood experiments.
Parameter . | Reservoir RA . | Reservoir DY . | Reservoir BB . | Reservoir SB . |
---|---|---|---|---|
GOR (std cm3 gas/std cm3 oil) | 121 | 160.22 | 213.73 | 156.73 |
Oil viscosity (cp) | 0.411 | 0.361 | 0.343 | 0.322 |
Water viscosity (cp) | 0.410 | 0.404 | 0.415 | 0.414 |
CO2 viscosity (cp) | 0.046 | 0.048 | 0.042 | 0.047 |
Formation volume factor (res cm3/std cm3) | 1.46 | 1.50 | 1.58 | 1.63 |
Bubblepoint (psig) | 2,430 | 2,570 | 3,500 | 2,220 |
Pressure (psig) | 4,250 | 4,485 | 3,900 | 3,950 |
Temperature (°C) | 129 | 121 | 121 | 121 |
CO2 MMP | 2,850 | 2,570 | 3,500 | 3,500 |
Parameter . | Reservoir RA . | Reservoir DY . | Reservoir BB . | Reservoir SB . |
---|---|---|---|---|
GOR (std cm3 gas/std cm3 oil) | 121 | 160.22 | 213.73 | 156.73 |
Oil viscosity (cp) | 0.411 | 0.361 | 0.343 | 0.322 |
Water viscosity (cp) | 0.410 | 0.404 | 0.415 | 0.414 |
CO2 viscosity (cp) | 0.046 | 0.048 | 0.042 | 0.047 |
Formation volume factor (res cm3/std cm3) | 1.46 | 1.50 | 1.58 | 1.63 |
Bubblepoint (psig) | 2,430 | 2,570 | 3,500 | 2,220 |
Pressure (psig) | 4,250 | 4,485 | 3,900 | 3,950 |
Temperature (°C) | 129 | 121 | 121 | 121 |
CO2 MMP | 2,850 | 2,570 | 3,500 | 3,500 |
Experimental Setup and Procedures
Experimental Setup
A sketch of the experimental setup is shown in Fig. 1 . In the high-pressure coreflood rig used in this study, a temperature-controlled air oven is used to house the core holder, injection fluids, the backpressure regulator (BPR) lines and connections at a constant temperature. The apparatus has four storage cells containing the injection fluids (i.e., crude oil, gas, and water). Two pairs of pumps are used to pump fluids around the flow system (core and bypass lines) and to apply overburden pressure on the core and supply pressure to the BPR.
In the experiments reported here, the orientation of the core was vertical, and the overburden pressure was kept at 500 psi above the pore pressure (pressure at BPR). Two pressure transducers were connected to the inlet and outlet of the core to measure and record the differential pressure across the core. A BPR was used to maintain the pressure of the core outlet and deliver the core effluent at atmospheric pressure. While running an experiment, the effluent from the BPR flowed into a dual outlet separator, where the liquid was collected in a graduated cylinder. A CO2 analyzer was connected to the production line to detect CO2 and to recognize CO2 breakthrough in the produced gas. The material balance performed on oil and brine was used to determine the system’s final average oil and water saturations. Overall pressure drops were measured to determine fluid mobilities.
Experimental Procedure
The core samples were first cleaned under reservoir temperature and then dried in an oven at 40°C to evaporate any remaining solvent. The core was then loaded into the core holder and vacuumed for approximately 24 hours to remove any air from the core pores. Helium and brine (FW) PV tests were carried out with an overburden pressure of 500 psi higher than the core pressure.
Using FW, brine permeability measurement was performed at room temperature and reservoir pressure. The net pressure was kept at 500 psi. The initial water saturation (Swi) was established by injecting a series of viscous mineral oils provided by Sigma Aldrich. These mineral oils are inert and do not lead to the formation of emulsions in the core during Swi establishment. After establishing Swi, the viscous mineral oil was displaced sequentially with lighter mineral oils to reach a viscosity close to crude oil viscosity. The viscosity of these mineral oils’ ranges from 87 cp to 2 cp.
The core holder was moved into the oven, and crude oil was injected through the core. During the injection, the pressure in the core was increased to reservoir pressure as the net stress was kept at 500 psi. The temperature was increased gradually to corresponding reservoir temperatures. The differential pressure across the core and the density of the produced fluids were used as the quality control parameters to ensure that the mineral oils at each step were removed from the core completely. At a rate of one PV per week, the aging process was carried out for 4 weeks under reservoir conditions using dead crude oil.
The recombined fluid was transferred into the oven after preparing the live crude oil. Live crude oil was first injected through the bypass, BPR, and gas meter to measure GOR and formation volume factor (swelling factor). Subsequently, the live oil was pumped through the core for 4 PV until the GOR calculated from the core effluent had reached the initial value measured from the recombined oil through the bypass and PVT tests.
A series of secondary and tertiary multiple-contact miscible CO2 injection experiments were carried out to study the effects of pressure, flow rate, rock type, core length/slimtube, and initial water saturation on displacement efficiency and residual oil saturation to miscible gas (Sorm). The coreflood experiments were performed above and below MMP using the core samples shown in Table 1 . The effect of pressure on displacement efficiency and oil recovery was investigated using two different procedures:
First procedure:
Initialize the core sample at connate water of 10%.
After aging to restore wettability, start CO2 injection at 2,700 psi until no further oil is recovered.
Repeat the same procedure using the same core sample, but every time, start CO2 injection at different pressures of 2,800 psi, 2,900 psi, 3,000 psi, 3,100 psi, 3,150 psi, 3,200 psi, 3,400 psi, 3,800 psi, and 4,250 psi, respectively.
Second procedure:
Initialize the core sample at connate water of 10%.
After aging, start CO2 injection at a low pressure of 2,700 psi until no further oil is recovered.
Increase the injection rate to investigate the impact of the rate on oil recovery.
Increase the pressure in steps until reaching 5,000 psi; at each pressure step, inject CO2 until no further oil is recovered.
The Swi of 10% was used in the coreflood experiments as it represents the Swi in the carbonate oil reservoirs under study.
Experimental Results and Discussions
CO2 miscible/near-miscible injection experiments were performed under reservoir conditions using limestone reservoir cores and live crude oil under different injection strategies. All gas injection experiments were performed using vertically oriented cores, and the gas was injected from the top. This is done to ensure a stable displacement. The density difference between the crude oil and CO2 under reservoir conditions is 0.05 g/cm3 on average, which translates into a gravity gradient of 0.022 psi/ft. This shows that gravity forces are negligible. The details of the experiments and main results are discussed in the subsequent sections.
Effect of Pressure on Oil Recovery and Sorm
The effect of pressure on miscible gas/WAG reservoir performance has been an active area of interest and research; see Shyeh-Yung (1991) and the references therein. While operating gas injection at higher pressure (higher than MMP) leads to better microscopic displacement, a lower-pressure process can be attractive for economic and operational reasons. Both gas volume requirements in standard cubic feet and costs for compression are reduced at lower pressures. Moreover, near-miscible displacement (low IFT) can potentially lead to high displacement efficiency similar to miscible displacement. In addition, injecting CO2 at lower pressure (lower than MMP) but still at near-miscible conditions can lead to improved sweep efficiency due to the formation of two phases, which lowers the mobility of the injected gas; see Shyeh-Yung (1991) and Burger and Mohanty (1997).
Several experiments were performed in this study to investigate the impact of pressure on displacement efficiency and Sorm. The pressure investigated using the first procedure discussed above ranged from 2,700 psi to 4,250 psi. The MMP of CO2 with the crude oil used in these studies is 2,850 psi; hence, the experiments covered the range below and above MMP. Fig. 2 shows the results of the secondary CO2 injection experiments performed at different pressures; each experiment started at immobile connate water saturation of ~10%. The figure shows the recovery factor as a function of PV injected for the pressure range from 2,700 psi to 4,250 psi. Fig. 3 shows Sorm and recovery factor as a function of pressure. The data show a clear correlation between Sorm and pressure, where the Sorm decreases as the pressure increases. The recovery factor of the experiment performed at a reservoir pressure of 4,250 psi is ~97%. This shows the high displacement efficiency of CO2, which is expected for such light oil as the experiment is performed at a pressure well above MMP.
The data shown in Figs. 2 and 3 demonstrate a strong impact of pressure on the recovery factor and Sorm. Fig. 3 shows that Sorm decreased (recovery factor increased) almost linearly with increasing pressure in the 2,700–3,200-psi range, where Sorm decreased from 26% to 5%. The data also showed no dramatic change in the slope at pressures below the MMP (2,850 psi), as slimtube results suggest. At higher pressures, the Sorm only slightly decreased to 3% when the pressure increased to 4,250 psi. The data show that the maximum displacement efficiency is obtained when performing experiments at a few 100 psi above MMP. However, such a pressure can be significantly lower than reservoir pressure in cases where MMP is relatively low, like the case investigated in this study.
Fig. 4 illustrates how pressure and injected CO2 PV affect the oil recovery factor. The curve at 1 PV injected shows that the recovery factor increases by more than 30% of oil in place as the pressure increases from 2,700 psi to 4,250 psi. This trend is consistent across the other curves in the figure, indicating that CO2 injection displacement efficiency improves as PV injected and/or pressure increases. For example, a recovery factor of ~90% can be obtained for this specific reservoir at 3,000 psi with 6 PV injected or at 4,000 psi with 1 PV injected. A recovery factor of 80% is obtained by injecting 6 PV at 2,900 psi, 2 PV at 3,000 psi, or 1 PV at 3,200 psi. This demonstrates that a modest increase in pressure (300 psi) can achieve the same recovery factor with significantly less injected volume (1 PV instead of 6 PV). Additionally, the curves in Fig. 4 show that the effect of the injected PV is reduced as the pressure increases. For example, at 4,250 psi, the recovery factor increases from 91% to 96% at 4,250 psi when the injected PV increases from 1 PV to 4 PV, compared with a rise from 57% to 70% at 2,700 psi for the same increase in injected PV.
Oil recovery factor as a function of pressure at different PVs injected.
The data also show that the ultimate recovery factor is lower at low pressure, with a higher Sorm even at high injected PV. For example, the recovery factor is only 71% after injecting 6 PV at 2,700 psi, while increasing the pressure to 2,900 psi (just above MMP) increases the recovery factor to ~80% at 6 PV injected. At higher pressures, for example, higher than 3,200 psi, the ultimate recovery factor or Sorm does not change much. Still, the same recovery factor is obtained at lower PV injected as the pressure increases. Note that the recovery factor discussed here applies only to displacement efficiency and does not necessarily reflect reservoir scale results due to the lower sweep efficiency expected due to heterogeneity and gas override.
Selecting the operating pressure is based on the project’s economy. When MMP is much lower than reservoir pressure, miscible gas injection can be operated at a lower pressure that combines maximum displacement efficiency and lower compression cost. For the case under study, the reservoir pressure is 4,250 psi, but maximum displacement efficiency (minimum Sorm) is achieved at around 3,200–3,400 psi. On the other hand, operating the reservoir at lower pressure will require fewer compression facilities and lower injection volume at surface conditions, but it requires injecting higher reservoir PV for the same recovery factor. Therefore, there could be an optimal operating pressure that maximizes displacement efficiency and the project economics, which can be determined through reservoir scale simulation. From the discussion above, operating the reservoir at a higher pressure seems to have a clear advantage by significantly reducing the required reservoir PV injection for the same recovery factor. This advantage may overcome the relatively small increase in the surface volume and compression cost.
Fig. 5 shows CO2 production during the same experiments plotted in Fig. 2 . The data show that CO2 breakthrough is delayed as the pressure increases. The breakthrough at 2,700 psi is at 0.42 PV injected and increases to 0.67 PV injected at high pressure. This delay in breakthrough means less CO2 recycling and better sweep efficiency. At lower pressures, CO2 viscosity decreases, mobility increases, and displacement efficiency decreases. All this explains the earlier breakthrough at lower pressure. Therefore, operating miscible gas injection at high pressure means higher recovery at lower PV injected, late CO2 breakthrough, and less CO2 recycling.
CO2 production as a function of PV injected for the experiments performed at different pressures.
CO2 production as a function of PV injected for the experiments performed at different pressures.
Increasing Pressure during CO2 Injection
The experiments performed at different pressures demonstrated the significant impact of pressure on miscible gas performance. This section examines whether increasing the pressure during CO2 injection can achieve the same displacement efficiency as operating at high pressure from the beginning. The recovery factor was measured as the pressure increased from 2,700 psi to 5,000 psi in steps, covering a range from below MMP to greater than 2,000 psi above MMP. The last pressure step of 5,000 psi is well above the initial reservoir pressure and was chosen to analyze the effect of higher pressure near the CO2 injector and to extend the pressure range in the study.
Table 6 summarizes the pressure steps, injection rate, cumulative PV injected, oil saturation, and recovery factor. In total, significantly higher cumulative PV was injected compared with the individual experiments reported in Figs. 2 and 3 . Note that the injection rate was increased from 5 cm3/h to 900 cm3/h at the first pressure of 2,700 psi to investigate the effect of the rate on the oil recovery factor. The injection rate was 5 cm3/h during all other pressures.
Experimental pressure, injection rate, PV injected, Sorm, and recovery factor (RF).
Pressure (psi) . | Rate (cm3/h) . | PV Injected . | Sorm (%) . | RF (%) . |
---|---|---|---|---|
2,700 | 5 | 4.5 | 26.5 | 70.5 |
2,700 | 50 | 4.7 | 26.2 | 70.8 |
2,700 | 200 | 4.9 | 26.2 | 70.8 |
2,700 | 600 | 5.4 | 26.0 | 71.1 |
2,700 | 900 | 6.0 | 26.0 | 71.1 |
2,750 | 5 | 7.7 | 24.5 | 72.7 |
2,800 | 5 | 10.5 | 22.5 | 75.0 |
2,850 | 5 | 13.3 | 21.1 | 76.6 |
2,900 | 5 | 15.3 | 20.2 | 77.5 |
2,950 | 5 | 18.0 | 19.5 | 78.3 |
3,100 | 5 | 21.2 | 16.0 | 82.2 |
3,300 | 5 | 23.6 | 14.6 | 83.8 |
3,600 | 5 | 26.3 | 13.4 | 85.0 |
3,900 | 5 | 27.4 | 13.0 | 85.5 |
4,250 | 5 | 34.6 | 11.6 | 87.1 |
5,000 | 5 | 43.7 | 10.7 | 88.1 |
Pressure (psi) . | Rate (cm3/h) . | PV Injected . | Sorm (%) . | RF (%) . |
---|---|---|---|---|
2,700 | 5 | 4.5 | 26.5 | 70.5 |
2,700 | 50 | 4.7 | 26.2 | 70.8 |
2,700 | 200 | 4.9 | 26.2 | 70.8 |
2,700 | 600 | 5.4 | 26.0 | 71.1 |
2,700 | 900 | 6.0 | 26.0 | 71.1 |
2,750 | 5 | 7.7 | 24.5 | 72.7 |
2,800 | 5 | 10.5 | 22.5 | 75.0 |
2,850 | 5 | 13.3 | 21.1 | 76.6 |
2,900 | 5 | 15.3 | 20.2 | 77.5 |
2,950 | 5 | 18.0 | 19.5 | 78.3 |
3,100 | 5 | 21.2 | 16.0 | 82.2 |
3,300 | 5 | 23.6 | 14.6 | 83.8 |
3,600 | 5 | 26.3 | 13.4 | 85.0 |
3,900 | 5 | 27.4 | 13.0 | 85.5 |
4,250 | 5 | 34.6 | 11.6 | 87.1 |
5,000 | 5 | 43.7 | 10.7 | 88.1 |
Fig. 6 shows the recovery factor and oil saturation as a function of pressure at the end of each pressure step. Similar to Fig. 3 , the residual oil saturation to miscible gas (Sorm) decreased, and the recovery factor increased with increasing pressure. Sorm decreased almost linearly from 26% to 15% of oil in place as the pressure increased from 2,700 psi to 3,100 psi; Sorm only reduced to 11% as the pressure increased to 5,000 psi.
Recovery factor and Sorm as a function of pressure for the experiments shown in Table 6 .
Recovery factor and Sorm as a function of pressure for the experiments shown in Table 6 .
Comparing Figs. 3 and 6 shows that starting the experiment at higher pressure results in lower Sorm and higher recovery factor than increasing the pressure during the experiment. The minimum Sorm in Fig. 6 is 11% measured at 5,000 psi and 46 PV injected. This is the same as the Sorm measured at 3,000 psi and 6 PV injected, as shown in Fig. 3 . Therefore, starting CO2 injection at a higher pressure yields better displacement efficiency than starting at a lower pressure and then increasing the pressure while injecting CO2. When comparing the data at the same pressure of 3,200 psi, Sorm is ~10% lower when starting CO2 injection directly at that pressure compared with beginning at 2,700 psi and then increasing the pressure in steps to ~3,200 psi. Note that, when starting at low pressures (i.e., 2,700 psi), the Sorm at the end of this pressure step consists of heavier components, making the recovery less efficient even after increasing the pressure. As noted by Inoue et al. (1985), oil produced after CO2 breakthrough had lower viscosity due to the core’s remaining heavy components (C7+), leading to a higher MMP of the remaining oil with CO2 than with the original oil.
The MMP was assessed for crude oils with varying molecular weights of the C5 + fraction by Holm and Josendal (1974 and 1980). Their findings confirmed a notable increase in MMP with the rising molecular weight of the C5 + fraction. Mungan (1981) further expanded these data to include higher molecular weights of the C5 + fraction, demonstrating that as crude oil becomes heavier, an asymptotic increase in the MMP is observed. This aligns with our findings that oil remaining at lower pressures has significantly higher MMP with CO2 than with the original oil, explaining the lower recovery factor measured when increasing the pressure while injecting CO2 (shown in Fig. 6 ) compared with the data in Fig. 3 , where CO2 injection was performed at higher pressures. Analysis of the produced oil from experiments done at different pressures was performed and confirmed the change in the oil composition, as discussed in subsequent sections.
Slimtube vs. Coreflood Experiments
Slimtube test is a well-established and accepted procedure in the oil industry to measure MMP, which is the lowest pressure at which crude oil and solvent develop miscibility dynamically, and it is usually the target reservoir pressure for designing a miscible process.
Oil recoveries from slimtube tests typically are high at pressures above the MMP and decline steeply as pressures are reduced below the MMP. This bend-over point in the recovery vs. pressure laboratory data is the MMP. The main drawbacks of slimtube tests are that the “porous medium” is a bead pack bearing little resemblance to reservoir rocks, and the water is not present during the test.
The slimtube test done using the same crude oil and CO2 is shown in Fig. 7 , and the coreflood data are measured at 1 PV, 3 PV, and 6 PV injections. The main differences between the slimtube curve and the core injection experimental curves are as follows:
The slimtube oil recovery is higher at the same PV injected. The slimtube recovery factor is usually measured at 1.2 PV injected.
The oil recovery decreases sharply at a pressure below MMP in the slimtube test. In contrast, no dramatic change in the slope of the recovery curve is observed at a pressure below MMP during core injection experiments.
The change in slope of the recovery factor during the core injection experiments occurs at a higher pressure, about 3,200 psi. However, the recovery curve’s slope change is still smoother than the slimtube experiment.
Oil recovery factor as a function of pressure at different PVs injected compared with that from slimtube test.
Oil recovery factor as a function of pressure at different PVs injected compared with that from slimtube test.
The above observations demonstrate that the effect of pressure on secondary CO2 core injection experiments is different compared with the impact of pressure on slimtube experiments. This aligns with the results reported by Shyeh-Yung (1991) and Lange (1998). It was also concluded by Moortgat et al. (2013) that there is a vast difference between CO2 injection in a 1D slimtube and in a core where there may be a 2D flow. Hence, corefloods should be used whenever possible to determine displacement efficiency, Sorm, and recovery factor, while slimtube can still measure MMP. However, the meaning of MMP measured using a slimtube may lose its importance because the recovery factor increases gently with pressure. The coreflooding data still show that performing the experiments at higher pressure significantly impacts displacement efficiency. Also, the recovery factor approaches the values obtained in the slimtube experiments at high pressure and high PV injected.
Effect of Injection Rate on Recovery Factor and Sorm
The impact of injection rate was evaluated during the first experiments performed at 2,700 psi, 2,800 psi, 2,900 psi, and 3,000 psi, with rates increasing from 5 cm3/h to 900 cm3/h. The recovery factor and injection rate of the experiments performed at 2,700 psi and 3,000 psi are shown in Fig. 8 , while for the 2,700 psi experiment, they are already reported in Table 6 . The results show that the injection rate has minimal effect on the recovery factor, which is expected given that the experiments were performed at near-miscible and miscible conditions with very low IFT (close to zero) between CO2 and oil. Due to the low IFT, the capillary forces are negligible and there is hardly any capillary end effect. Therefore, the injection rate does not affect the recovery factor of the secondary miscible or near-miscible gas injection experiments, consistent with the findings of Stern (1991) on secondary CO2 injection at immobile water saturation.
Recovery factor (RF) and injection rate of two CO2 injection experiments performed at 2,700 psi and 3,000 psi.
Recovery factor (RF) and injection rate of two CO2 injection experiments performed at 2,700 psi and 3,000 psi.
Fluid flow during these experiments was predominantly controlled by viscous forces, which were significantly higher than capillary forces due to the low IFT between oil and CO2. Regarding gravity forces, the density of CO2 at 2,700 psi and 3,000 psi was 340 kg/m3 and 390 kg/m3, respectively, while the crude oil density was 644 kg/m3 and 650 kg/m3, respectively. This results in a gravity gradient of 0.13 psi/ft and 0.12 psi/ft at those pressures, leading to gravity pressure drop across the core of about 0.1 psi, which is much lower than the viscous pressure drop measured at the lowest injection rate (5 cm3/h) used in the experiment. Furthermore, as the injection rate was increased to 900 cm3/h, as shown in Fig. 8 , the viscous pressure drop increased to 120 psi, which is about three orders of magnitude higher than the gravity pressure, clearly demonstrating that viscous forces dominated the fluid flow during these experiments.
It is important to note that the discussion focuses solely on displacement efficiency; sweep efficiency—when comparing experiments starting at different injection rates—may be negatively affected by increasing the rate due to early breakthrough. This was not investigated in this study. Moreover, gravity may have more effects at the reservoir scale, which depends on the thickness of the reservoir, the well spacing, and injection rates.
Effect of Rock Type and Rock Heterogeneity on Recovery Factor and Sorm
The same crude oil (RA) was used to perform CO2 injection experiments using Core Samples RA1 and DY2 at pressures of 2,900 psi and 4,250 psi. Fig. 9 shows the dimensionless tracer concentration profiles vs. tracer pore volumes injected for both cores. The tracer profiles show that Sample RA1 is more heterogeneous, while DY2 is relatively homogeneous. As reported in Table 1 , both core samples have low permeability—1.1 md for RA1 and 1.9 md for DY2.
Profile of the dimensionless concentration of tracer vs. tracer pore volumes injected for Cores RA1 and DY2.
Profile of the dimensionless concentration of tracer vs. tracer pore volumes injected for Cores RA1 and DY2.
The recovery factor as a function of the PV injected is shown in Fig. 10 for the four experiments. The figure shows that a similar ultimate recovery is obtained at the same injection pressure using both core samples. However, the oil is produced much faster for Core Sample DY2. A higher injected PV is required to obtain the same recovery factor for Core Sample RA1. Fig. 11 shows the CO2 production data, and it is evident that the CO2 breakthrough occurred much earlier in the experiments run on Core Sample RA1. The results demonstrate the strong effect of heterogeneity on oil recovery and CO2 breakthrough and consequently on CO2 recycling and project economics.
Recovery factor vs. PV injected for two experiments performed on Cores RA1 and DY2 at 2,900 psi and 4,250 psi, respectively.
Recovery factor vs. PV injected for two experiments performed on Cores RA1 and DY2 at 2,900 psi and 4,250 psi, respectively.
CO2 production as a function of PV injected during the same experiment is shown in Fig. 10 .
CO2 production as a function of PV injected during the same experiment is shown in Fig. 10 .
Effect of Flowing the CO2 through Slimtube before CO2 Injection in Core Sample
The significant effect of pressure on Sorm and recovery factor raised the question of whether the results are due to experimental artifact. As CO2 is multicontact miscible with the oil, a minimum core length is required to develop miscibility and maximize oil recovery. As demonstrated by Stern (1991), when experiments are performed at pressures well above MMP, core length has a negligible effect, with no significant difference in recovery factor observed on core samples from 1 ft to 32 ft. This suggests that miscibility develops over a distance smaller than 1 ft. However, at pressure closer to MMP, a longer core may be required to develop miscibility, which may explain the strong dependence of oil recovery on experimental pressure. Because it is challenging to use a reservoir core longer than 1 ft, a new experiment was designed where CO2 flowed through a 60-ft slimtube full of oil before entering the 1-ft-long reservoir core. This allowed for mass exchange between CO2 and oil (through vaporizing and condensing mechanisms) and achieved a miscible front before entering the core. This is similar to using a longer core where a miscibility front is developed in the first part of the core, and hence, such an experiment may replace using a longer core in such experiments.
Experiments were conducted with and without the slimtube at 2,900 psi to test this experimental procedure. The results of the two experiments are shown in Fig. 12 . The data show that using the slimtube did not significantly reduce Sorm, and the reduction was less than 1%. The results confirm that, for the conditions of our experiments, the 1-ft-long core was sufficient for achieving miscibility, as adding the slimtube had no material change on oil recovery.
Recovery factor vs. PV injected for two experiments performed on Core DY2 at 2,900 psi with and without the slimtube.
Recovery factor vs. PV injected for two experiments performed on Core DY2 at 2,900 psi with and without the slimtube.
This finding aligns with the experiments of Shyeh-Yung (1991), which showed that Sorm was not strongly influenced by the presence of the slimtube at pressures above MMP. Overall, the results validate that the observed pressure effects on CO2 core injection are genuine and not due to core length. Further investigations into pressure effects on recovery factor using cores of different lengths are ongoing and will be discussed in future publications.
Produced Oil Compositions
Oil produced from selected CO2 injection experiments was analyzed by gas chromatography, focusing on crude oil components ranging from C6 to C35. Six or seven samples were collected during each experiment at different PVs injected, with the first sample collected before the CO2 breakthrough to represent the original oil composition. The weight percent of a single carbon number from C6 to C35 was reported for each sample.
The normalization method proposed by Shyeh-Yung (1991) was adopted, where the weight percent of a component in the produced oil sample was divided by its weight percent in the original crude oil to highlight changes in composition. When a component is preferentially extracted, its normalized composition will increase above 1, while a value around 1 suggests no extraction occurs, or the oil is miscibly displaced.
Fig. 13 plots normalized produced oil compositions for the experiments performed at 2,800 psi, 3,100 psi, and 3,200 psi, with each curve representing different PVs injected. It is important to note that the high carbon numbers (i.e., above C30) have low absolute values, leading to higher uncertainty, as demonstrated by the peak at C34 at 5.5 PV injected in Fig. 13a . All samples were collected post-CO2 breakthrough. The different figures demonstrate that lower molecular weight components (lower carbon numbers) are produced first at lower PV injected, while heavier components are produced as more CO2 is injected. Comparing Figs. 13a (at 2,800 psi) and 13c (at 3,200 psi) demonstrates that heavier components are produced at higher PV injected in lower-pressure experiment, supporting the earlier discussion on the effect of pressure on recovery factor and Sorm. The data also explain why CO2 injection at higher pressure is more efficient, as the heavy components are produced much faster.
Normalized produced oil composition at different PVs injected (a) at 2,800 psi, (b) at 3,100 psi, and (c) at 3,200 psi for crude oil RA.
Normalized produced oil composition at different PVs injected (a) at 2,800 psi, (b) at 3,100 psi, and (c) at 3,200 psi for crude oil RA.
The same data are plotted in Fig. 14 ; however, the different curves in the figure are taken from different experiments performed at different pressures but at similar PV injected. Fig. 14a shows fluid samples collected at ~2.5 PV injected from experiments at 2,700 psi, 2,800 psi, 3,000 psi, and 3100 psi, while Fig. 14b shows samples collected at ~3.5 PV injected from five experiments, including one at 3,200 psi. The data confirm that the fraction of heavier components produced increases as the pressure increases. Consequently, the remaining oil saturation at lower pressure primarily consists of heavy components that are no longer miscible with fresh CO2. This aligns with the earlier analysis that the lower recovery measured at lower pressure is due to leaving behind a significant portion of heavy components that can only be extracted at high pressure.
Normalized produced oil composition taken at (a) 2.5 PV and (b) 3.5 PV injected for the experiments performed at different pressures.
Normalized produced oil composition taken at (a) 2.5 PV and (b) 3.5 PV injected for the experiments performed at different pressures.
Effect of Initial Water Saturation on Sorm and Recovery Factor of CO2 Injection
All experiments discussed were performed as secondary CO2 injections at immobile connate water. The presence of mobile water was recognized as an essential factor affecting oil recovery during miscible CO2 experiments, which may explain the difference between slimtube and coreflood experiments. As reported by several authors (Stalkup 1970; Shelton and Schneider 1975; Tiffin and Yellig 1983; Spence and Ostrander 1983), part of the oil saturation can be trapped by mobile water even when it is miscible with CO2. The degree of oil trapping is strongly dependent on the rock wettability. To study the effect of the presence of water on the recovery factor and Sorm during miscible CO2 injection, three types of experiments were performed:
Secondary CO2 injection in core samples containing irreducible water saturation of 10%.
Tertiary CO2 injection at high mobile water saturation and Sorw. The water saturation increased to 68–74% after the waterflood.
Secondary CO2 injection using core initialized at 40% and 60% mobile water saturation.
The experiments were performed using five core samples and four crude oils (see Tables 1 and 2 ). The results of the secondary and tertiary experiments are tabulated in Table 7 and plotted in Fig. 15 . Table 7 shows the experimental type (secondary or tertiary), core sample, crude oil, pressure, MMP, initial water saturation (Swi), water saturation at the start of CO2 injection (SwiCO2), Sorm, and water saturation at the end of CO2 injection (Swf). Fig. 15 shows Sorm vs. Swi in both secondary and tertiary experiments. The results show that secondary CO2 injection at immobile connate water of ~10% recovers more oil than tertiary floods at similar pressures. In secondary CO2 injection at immobile water saturation, the higher initial oil saturation is better connected, while the target oil in tertiary CO2 injection is mainly trapped by water, hence poorly connected. The results of the experiments demonstrate that the presence of a mobile water phase in the core hinders CO2 flood oil recovery performance.
Experimental (Exp.) type, core sample, crude oil pressure, MMP, Swi, SwiCO2, Sorm, and Swf. Sec. = secondary; Ter. = tertiary.
Exp. Type . | Core . | Crude Oil . | Pressure (psi) . | MMP (psi) . | Swi . | SwiCO2 . | Sorm . | Swf . |
---|---|---|---|---|---|---|---|---|
Sec. | DY1 | DY | 4,485 | 2,570 | 0.4 | 0.4 | 0.14 | 0.34 |
Sec. | DY1 | DY | 4,485 | 2,570 | 0.6 | 0.6 | 0.13 | 0.35 |
Sec. | DY1 | DY | 4,485 | 2,570 | 0.09 | 0.09 | 0.06 | 0.09 |
Sec. | DY2 | DY | 4,250 | 2,570 | 0.09 | 0.09 | 0.03 | 0.09 |
Sec. | DY2 | RA | 4,250 | 2,850 | 0.09 | 0.09 | 0.03 | 0.09 |
Sec. | RA1 | RA | 4,250 | 2,850 | 0.04 | 0.04 | 0.02 | 0.04 |
Sec. | BB1 | BB | 3,900 | 3,500 | 0.09 | 0.09 | 0.07 | 0.09 |
Sec. | SB1 | SB | 3,950 | 3,520 | 0.06 | 0.06 | 0.07 | 0.06 |
Ter. | DY1 | DY | 4,485 | 2,570 | 0.4 | 0.68 | 0.12 | 0.32 |
Ter. | DY1 | DY | 4,485 | 2,570 | 0.6 | 0.73 | 0.13 | 0.32 |
Ter. | DY2 | DY | 4,485 | 2,570 | 0.11 | 0.74 | 0.09 | 0.38 |
Ter. | BB1 | BB | 3,900 | 3,500 | 0.09 | 0.69 | 0.1 | 0.37 |
Ter. | SB1 | SB | 3,950 | 3,520 | 0.06 | 0.72 | 0.11 | 0.39 |
Exp. Type . | Core . | Crude Oil . | Pressure (psi) . | MMP (psi) . | Swi . | SwiCO2 . | Sorm . | Swf . |
---|---|---|---|---|---|---|---|---|
Sec. | DY1 | DY | 4,485 | 2,570 | 0.4 | 0.4 | 0.14 | 0.34 |
Sec. | DY1 | DY | 4,485 | 2,570 | 0.6 | 0.6 | 0.13 | 0.35 |
Sec. | DY1 | DY | 4,485 | 2,570 | 0.09 | 0.09 | 0.06 | 0.09 |
Sec. | DY2 | DY | 4,250 | 2,570 | 0.09 | 0.09 | 0.03 | 0.09 |
Sec. | DY2 | RA | 4,250 | 2,850 | 0.09 | 0.09 | 0.03 | 0.09 |
Sec. | RA1 | RA | 4,250 | 2,850 | 0.04 | 0.04 | 0.02 | 0.04 |
Sec. | BB1 | BB | 3,900 | 3,500 | 0.09 | 0.09 | 0.07 | 0.09 |
Sec. | SB1 | SB | 3,950 | 3,520 | 0.06 | 0.06 | 0.07 | 0.06 |
Ter. | DY1 | DY | 4,485 | 2,570 | 0.4 | 0.68 | 0.12 | 0.32 |
Ter. | DY1 | DY | 4,485 | 2,570 | 0.6 | 0.73 | 0.13 | 0.32 |
Ter. | DY2 | DY | 4,485 | 2,570 | 0.11 | 0.74 | 0.09 | 0.38 |
Ter. | BB1 | BB | 3,900 | 3,500 | 0.09 | 0.69 | 0.1 | 0.37 |
Ter. | SB1 | SB | 3,950 | 3,520 | 0.06 | 0.72 | 0.11 | 0.39 |
These findings are inline with the literature data. As already documented in the literature (Inoue et al. 1985; Holm 1986; Shyeh-Yung 1991; Stern 1991; Rao et al. 1992; Yeh et al. 1992), water blocking can significantly hinder oil recovery of CO2 injection in water-wet rock. For mixed-wet to oil-wet rocks, the impact of water blocking on oil recovery is less pronounced than for water-wet rock due to the presence of wall-coating oil films that enhance contact between oil and solvent (see Shyeh-Yung 1991; Stern 1991). In this study, only mixed-wet rock was investigated. While the presence of mobile water reduced recovery factor by about 10%, this effect is minor compared with the reduction reported in literature for water-wet rock.
So far, the impact of water saturation on the CO2 injection recovery factor and Sorm was investigated by comparing secondary and tertiary CO2 injection experiments, where the sample in both experiments is initialized at connate water. As discussed above, the oil is better connected in the secondary experiments, which could explain the higher recovery.
To investigate whether the oil connectivity is the only reason for the lower recovery in tertiary experiments, secondary CO2 injection experiments were performed using a core initialized at mobile water saturation. The initialization was performed using a steady-state experiment starting from a 100% water-saturated core to reach the target oil saturation and ensure a more homogeneous saturation profile. Core Sample DY1 was initialized at mobile water saturation of Swi ~ 40% (first experiment) and Swi ~ 60% (second experiment). In this case, oil and water are mobile, and the oil is well connected and located in the big pores. Both tertiary and secondary CO2 injection experiments were performed. The results are shown in Table 7 and Fig. 15 and confirm that the presence of mobile water still has a similar effect on Sorm (i.e., the oil recovery decreased and Sorm increased). During secondary CO2 injection at high mobile water saturation, oil distribution is favorable for CO2 displacement compared to tertiary CO2 injection at similar saturation, as oil occupies the big pores and is better connected than tertiary CO2 injection. However, recovery factor and Sorm values are similar irrespective of the oil distribution or connectivity. This demonstrates that the presence of mobile water, either after waterflooding or at initial conditions, negatively affects the displacement efficiency of CO2 injection.
Fig. 16 illustrates the effect of mobile water on Sorm in three experiments: (a) secondary CO2 injection at immobile connate water Swi = 9%, (b) secondary CO2 injection at mobile initial water saturation Swi = 40%, and (c) tertiary CO2 injection where waterflood started at Swi = 40% and CO2 injection at Sw = 68%. The results show that Sorm is ~6% for the secondary CO2 injection starting at connate water, while Sorm values for secondary and tertiary injections starting at mobile water saturation are 14% and 12%, respectively. In the tertiary CO2 experiment, more than 60% of Sorw was recovered. In the secondary CO2 injection starting at Swi = 40%, it seems CO2 displaced both water and oil, and the water trapped some of the oil, leading to similar Sorm as measured in tertiary experiments.
Oil, water, and CO2 saturation in three different experiments: (a) secondary CO2 injection at Swi ~ 10%, (b) secondary CO2 injection at Swi ~40%, and (c) tertiary CO2 injection.
Oil, water, and CO2 saturation in three different experiments: (a) secondary CO2 injection at Swi ~ 10%, (b) secondary CO2 injection at Swi ~40%, and (c) tertiary CO2 injection.
Better oil recovery at lower initial water saturation may not be universally applicable; it may depend on the crude oil composition and/or wetting conditions of the rock [see Masalmeh et. al. (2023) and the references therein]. Therefore, the effect of water blocking on oil recovery during miscible gas/WAG injection should be experimentally evaluated for a given reservoir rock/fluid system.
Additional Discussion on Experimental Results and Practical Implications for Field Application
The results of this laboratory CO2 core injection study have important implications for the design and performance prediction of CO2 gas injection processes. Although no dramatic changes in the recovery curve slope were observed at a pressure below MMP during core injection experiments, a strong correlation was found between experimental pressure and the oil recovery. CO2 injection at lower pressure has significantly reduced the oil recovery factor in 1D coreflood experiments. Unlike the conclusion by Shyeh-Yung (1991), this work demonstrates that the reduction in oil recovery and increase in Sorm as the pressure decreases is significant. This is unlikely to be offset economically by reducing the mass of CO2 required to inject or the compression cost. Higher pressure not only improved recovery but also delayed CO2 breakthrough, thereby reducing CO2 recycling costs and CO2 footprint. Field-scale simulations should be conducted to further assess the recovery and economic viability of the CO2 projects under miscible or near-miscible conditions.
This work also demonstrated that the presence of mobile water during CO2 injection is crucial, even for mixed-wet rock. Even when the core was initialized at mobile water saturation (where oil is better connected and located in the big pores), the recovery factor decreased compared with secondary CO2 injection at immobile water. In this case, CO2 displaces both oil and water, and some of the oil is trapped by water. This phenomenon needs to be further investigated by running corefloods, incorporating the proper monitoring tools or experimenting with micromodels.
The study also examined whether increasing the pressure during injection will lead to a similar recovery factor as starting injection at reservoir pressure. The results show that starting a gas/CO2 injection project at low pressure is inefficient, as lighter oil components are extracted first, leaving behind heavier components, forming a more viscous oil with higher MMP with CO2. Once the pressure is increased, the fresh CO2 will not be as efficient in extracting this oil. Therefore, starting CO2 injection at the target pressure is more efficient.
In a case where the reservoir pressure is low at the start of CO2 injection, is it more efficient to restore pressure to initial conditions by waterflooding or by CO2 injection? Restoring the pressure with waterflooding and then starting the CO2 injection may lead to recovery loss due to water blocking effect, as shown in the experiments discussed above. On the other hand, restoring the pressure while injecting CO2 leads to lower displacement efficiency and higher Sorm. The choice between those two development strategies should be investigated experimentally using field-scale simulation. For example, for the case under study, running the experiment at low pressure (2,800 psi or 2,900 psi), the measured Sorm is 23% or 17%, respectively, after 7 PV injected. The recovery factor of the two experiments at 1 PV injected was 60% and 67%, respectively, as shown in Fig. 3 . However, in the tertiary CO2 injection experiments where the waterflood started at connate water, the measured Sorm after injecting 1 PV of CO2 was ~10%. This suggests that increasing the pressure by water injection before CO2 injection leads to higher displacement efficiency as water blocking is insignificant. However, the conclusion may be different for water-wet reservoirs as the water blocking is much more significant.
The study demonstrates that rock heterogeneity strongly affects CO2 flooding performance. It significantly impacts oil recovery and leads to premature CO2 breakthrough, which increases CO2 recycling needs and project economics. While the core samples used in this particular study were relatively homogeneous, a moderate level of heterogeneity was sufficient to demonstrate the high sensitivity of gas injection performance to rock heterogeneity. This highlights the critical importance of thorough characterization and accounting for reservoir heterogeneity when designing and evaluating the performance of CO2 flooding projects. The impact of heterogeneity can have far-reaching consequences on the technical and economic viability of the EOR operation.
The results obtained in this study suggest combining slimtube tests with reservoir-condition coreflood tests for more representative data for reservoir application and field process design. While slimtubes can determine MMP, the reservoir condition core injection experiments provide more representative data with respect to displacement efficiency and Sorm.
This study also explored how various parameters, such as injected PV, injection rate, and rock types, affect recovery factor and Sorm of CO2 injection. A strong impact of PV injected, especially at low pressure, was evident from the data presented in Fig. 3 , and a strong impact of rock heterogeneity/rock type was also evident in Figs. 10 and 11 . However, the injection rate had a negligible impact due to the low IFT system (see Fig. 8 and Table 6 ).
To assess core length effects on Sorm and recovery factor, an experiment was conducted where CO2 flowed through a slimtube so that CO2 developed miscibility with the oil before flowing through the core sample. The results of this experiment were compared with the experiment where CO2 was injected directly using the same core sample and the same pressure. The experiments showed that adding the slimtube did not affect the outcome, suggesting that the core length did not affect the results within the tested pressure range. Further research is ongoing to examine the impact of the core length using reservoir core samples.
In summary, one of the primary outcomes of this work is that Sorm is a function of pressure drop, PV injection, and mobile water saturation. This needs to be captured in reservoir simulation models so that a proper evaluation of CO2 injection projects at the reservoir scale can be evaluated. In addition, this conclusion may depend on the fluid-rock system; therefore, the results cannot be applied to other reservoirs with different fluid and rock properties. At least some of the experiments discussed in this work must be performed to help design CO2 injection projects for other reservoirs.
Limitation of the Study and Future Work
Gas-based EOR methods generally exhibit high displacement efficiency. However, they suffer from poor sweep efficiency due to gravity segregation, channeling, and viscous fingering. The study presented in this paper focused on evaluating the displacement efficiency of CO2 injection under reservoir conditions using core samples and fluids from four carbonate reservoirs in Abu Dhabi. While this study focused solely on displacement efficiency, laboratory experiments can provide insights into sweep efficiency and gravity override. However, laboratory experiments should be combined with reservoir-scale simulation using properly calibrated models to gain a more comprehensive understanding of gas injection sweep efficiency, gravity override, and oil recovery. The impact of gravity override and heterogeneity on miscible gas injection sweep efficiency is reported in Masalmeh et al. (2014).
In a previous publication (Masalmeh et. al., 2023), the authors conducted two types of experiments to investigate the effects of gravity segregation on immiscible and miscible gas injection and WAG processes:
Vertical and horizontal immiscible gas injection and WAG experiments using 2-in.-diameter core samples.
Miscible gas injection experiments using 2-in.- and 4-in.-diameter core samples, with the gas injection performed horizontally.
The results of these experiments demonstrated the significant impact of gravity segregation and override, even at the core scale:
The ultimate oil recovery from the vertical experiment was about 10% higher for immiscible gas injection than the horizontal injection.
For the WAG process, the ultimate recovery factor from the vertical coreflood was around 8% higher than the horizontal coreflood, indicating that the WAG cycles could not mitigate the impact of gravity during the horizontal gas injection.
The miscible gas injection experiments further highlighted the upscaling challenges. Comparing the results between the 2-in.- and 4-in.-diameter core samples, the ultimate oil recovery from miscible CO2 injection was reduced by 14% when the core diameter was doubled. These findings suggest that the sweep efficiency of miscible gas injection is significantly reduced as the scale increases, and this problem will be even more pronounced at the reservoir scale, where the reservoir section can be up to 100 ft or 150 ft thick. Additionally, the issue is expected to worsen with increasing well spacing as it gives more time for gas override. Note that these experiments are performed using low permeability (1–2 md) samples; gravity override will increase for high-permeability reservoirs.
The main limitations of this core-scale study are as follows:
The experiments were conducted on low-permeability and relatively homogeneous core samples. The conclusions may not apply to samples of high permeability and/or samples that exhibit stronger heterogeneity.
The study used light oils with low viscosity (0.32–0.41 cp). The findings cannot be generalized to heavy oil reservoirs.
The injection rates investigated in this study only apply to laboratory-scale experiments and displacement efficiency and may not reflect the effect of rate on sweep efficiency or reservoir scale oil recovery. This should be investigated using reservoir-scale simulation, and the results are likely to be case-dependent and strongly affected by geology and heterogeneity.
The study was performed at near-miscible/miscible conditions with low IFT and fluids (oil and CO2) of similar density where the density difference is 0.05 g/cm3. This shows that the effect of gravity and capillary forces is insignificant. These conditions are representative of the reservoirs under study. The conclusions cannot be extended to reservoirs at lower pressure where the density difference between the fluids is much larger (see Ajoma and Sungkachart 2021) or where the CO2 is denser than the crude oil (see Moortgat et al. 2013).
The laboratory data cannot be used to draw quantitative conclusions on utilization factors or CO2 sequestration volumes; the data can only provide a qualitative assessment. For example, the utilization factor is more efficient at higher pressure as more oil is produced at the same CO2 volume injected. Moreover, more CO2 volume can be sequestered at higher pressures as higher CO2 saturation is achieved at the end of CO2 injections, and less CO2 recycling is observed due to later breakthroughs. A more accurate quantitative assessment would require reservoir-scale reservoir simulation in addition to the data presented in this paper.
Future research is required to address these limitations and expand the understanding of CO2 injection in carbonate reservoirs. While the study acknowledges certain limitations, this does not diminish the overall value of the research. Instead, it provides a realistic and responsible assessment of the application scope of the data. By clearly defining the study’s boundaries, the authors caution readers against making unsupported generalizations beyond the specific context examined. This transparency helps readers understand the appropriate uses and limitations of the findings, ensuring that they draw well-grounded conclusions from the evidence presented.
Conclusions
The results of several CO2 core injection experiments performed using mixed- to oil-wet carbonate rocks are presented in this paper. The CO2 core injection experiments were conducted to investigate and understand the parameters that affect the displacement efficiency of CO2 injection under miscible and near-miscible conditions in the presence and absence of mobile water saturation. The results reported in this study are derived from core-scale experiments, and appropriate upscaling is required before applying them in field development. Based on the results, the following conclusions can be drawn:
The data demonstrate that the displacement efficiency of CO2 injection increases as PV injected and/or pressure increases.
Oil recovery decreases as pressure decreases, with Sorm increasing by more than 20 saturation units as the pressure decreases from 4,250 psi to 2,700 psi.
A higher recovery factor was achieved at a lower PV injected for the experiments performed at reservoir pressure ~1,500 psi higher than MMP.
In addition to higher displacement efficiency, operating CO2 injection projects at a pressure higher than MMP leads to late CO2 breakthroughs and less CO2 recycling.
The recovery factor is strongly affected by the PV injected; however, the effect of the CO2 injection volume is much more significant at lower pressure.
The injection rate has an insignificant effect on oil recovery and Sorm for miscible or near-miscible CO2 due to the low IFT between oil and CO2. This also shows that the residual oil saturation is already achieved at a low rate; increasing the rate does not mobilize more oil.
Rock heterogeneity has a strong effect on oil recovery and CO2 breakthrough and, hence, on CO2 recycling and the economy of the projects. Note that the core samples used in this study are homogeneous; however, even a moderate level of heterogeneity observed in one of the samples was sufficient to demonstrate the high sensitivity of gas injection performance to rock heterogeneity.
The presence of mobile water at the beginning of CO2 injection resulted in lower displacement efficiency and increased Sorm. However, this water blocking effect should be determined experimentally for a given reservoir rock/fluid system. The results of this study cannot be generalized to other reservoirs or fluid-rock systems.
Reservoir condition coreflood should be combined with slimtube tests where the slimtube is still used for determining MMP. In contrast, core injection experiments provide more realistic Sorm values for different injection scenarios, which can be used in miscible or near-miscible process design and simulation.
Nomenclature
Acknowledgments
The coreflood experiments were performed in the Centre for Enhanced Oil Recovery and CO2 Solutions of Heriot-Watt University. We would like to thank ADNOC for their financial support and for granting permission to publish this paper.
Article History
This paper (SPE 218525) was accepted for presentation at the SPE Conference at Oman Petroleum & Energy Show, Muscat, Oman, 22–24 April 2024, and revised for publication. Original manuscript received for review 26 April 2024. Revised manuscript received for review 18 November 2024. Paper peer approved 10 December 2024.