Introduction
The resources in low-permeability, tight oil reservoirs are abundant in China, which is significant to ensure stable and increased crude oil production. The CO2 flooding is a key technique for enhanced oil recovery (EOR) in low-permeability, tight oil reservoirs. It is also the key to successfully implementing carbon capture, utilization, and storage (CCUS) in the context of the “carbon peaking and carbon neutrality” strategy [1-2].
The efficiency and economic benefits of CO2 flooding, however, are negatively affected by gas channeling, due to the strong heterogeneity of the reservoir. More specifically, improving the sweep efficiency of CO2 flooding is one of the key issues of this technology [3-4]. Under the conditions of strong heterogeneity in reservoir, the sweep- controlling methods for CO2 flooding, such as parameter adjustment for injection/production and water-alternating-gas, perform poorly due to the complicated application scenarios [3,5]. As for the chemical anti-channeling technologies, the challenges mainly lie with injection difficulty, shear degradation, formation damage for polymer- based agents, low structural strength for foam-based agents, and sediments-related formation damage for inorganic- based agents [4,6⇓⇓ -9]. Recently, surfactant-based sweep-controlling systems which would be thickened and improve the flow resistance of CO2 after interacting with CO2 have been developed to tackle these issues [10⇓⇓⇓⇓⇓⇓-17]. The system has the characteristics of low viscosity, easy injection, underground self-thickening, and no damage to reservoirs. It has provided a new solution to solve the technical issues in chemical anti-channeling. So far, relevant research mostly focuses on primary amine-based single-chain small molecule surfactants, such as tetramethyl propylene diamine-sodium dodecyl sulfate (TMPDA-SDS), etc., which have problems of high response concentration, low interfacial activity, and poor water solubility [14⇓⇓-17]. These problems hinder these surfactants from becoming an effective industrialized solution for sweep-controlling during the CO2 flooding.
Compared with primary amine groups, tertiary amine groups are more sensitive to CO2 [18-19]. Also, compared with small single-chain molecules, long-chain Gemini surfactants have the advantages of lower self-assembly concentration, higher interfacial activity, and better heat and salt-resistant [20⇓-22]. Therefore, in this study, we used tertiary amine groups as the response component to design and synthesize smart response fluids for sweep-controlling during the CO2 flooding (hereinafter referred to as the smart fluids) based on Gemini surfactant with self-assembly capabilities. The structure of the smart fluid was confirmed by 1H-NMR spectroscopy, cryo-transmission electron microscopy and rheometer. The smart response characteristics and mechanism of self-thickening and interfacial tension reduction of the smart fluid during interaction with CO2 and oil were sequentially studied by rheological test, core displacement system and interfacial tension test. In addition, the self-healing capability of thickened aggregates, and the temperature and salt resistance of the smart fluid and thickened aggregates were also studied. The feasibility and effectiveness of the smart fluid for enhancing the sweep efficiency during CO2 flooding were examined and validated.
1. Research approach and technical principles
1.1. Design of the molecular structure and preparation of the smart fluid
The Gemini surfactants were designed to possess asymmetric structures to improve their self-assembly capability and interfacial activity, etc. [22]. Firstly, two types of single-chain tertiary amine-based surfactants, as shown in Eq. (1), were prepared from unsaturated and saturated fatty acids, respectively. R, R' in Eq. (2) are C11-C21 saturated and unsaturated alkyl chains, respectively. The asymmetric quaternary ammonium-based Gemini surfactant (as shown in Eq. (2)) was obtained by the two surfactants linking through a bridging agent. By controlling the molar ratio of tertiary amine-based single-chain surfactants to the quaternary ammonium-based Gemini surfactants in the product, the CO2-responsive smart fluid can be produced. The smart fluid is based on Gemini surfactants, with tertiary amine groups as the response component [23].
1.2. Mechanism of the smart fluid interacting with CO2 and oil exposure
1.2.1. Mechanism of self-thickening with CO2 exposure
Before being exposed with CO2, uncharged tertiary amine-based single-chain surfactants are interspersed in Gemini surfactants, with relatively high stacking parameters and low viscosity, being mutually dispersed vesicular micelles [20,24]. While after CO2 exposure, the carbonate group protonates the tertiary amine group with positive charges, and the charge repulsion decreases the packing parameters. As a result, the assembly mode of the micelles changes. From a micro perspective, the micelle transforms from a vesicle-like to a worm-like structure [20,24], and entangles each other, forming three-dimensional structures. Macroscopically, the fluid is thickened after CO2 exposure (Fig. 1 ).
Fig. 1. Schematic diagram of the self-thickening mechanism of the smart fluid during the interaction with CO2. |
1.2.2. Mechanism of interfacial tension reduction with oil exposure
The amphiphilic properties of Gemini surfactants can greatly reduce oil-water interfacial tension. The smart fluid thus improves the oil displacement efficiency with oil exposure. In addition, the lipophilic segment of Gemini surfactants has structurally similar miscibility to oil, so the dispersed vesicular micelles formed during this process can spontaneously solubilize the oil [25]. When the vesicular micelles contact with the oil, the oil molecules can enter into the vesicle core and transform the vesicular micelles into spherical micelles [20,24,26] (Fig. 2 ).
Fig. 2. Schematic diagram of the interfacial tension reduction mechanism of the smart fluid during the interaction with oil. |
1.3. Principle of sweep-controlling by the smart fluid
The heterogeneity of the reservoir can cause a non- uniform sweep of CO2 flooding, resulting in high CO2 content and low oil saturation in the high-permeability area. The injected smart fluid will mainly enter the high-permeability area and thicken when interacts with CO2 to form thickened aggregates. Therefore, the subsequent CO2 will be forced to infiltrate into the low-to-medium-permeability area to expand the swept volume. Another small amount of the smart fluid that enters the highly oil-saturated, low-to-medium-permeability area can maintain its low-viscosity state due to its oil solubility. During this process, the smart fluid only performs as an active substance to improve the oil displacement efficiency, without damaging low-to-medium-permeability areas. By applying multiple cycles of the smart fluid-altering-CO2 injection, the favorable sweep efficiency during CO2 flooding in heterogeneous reservoirs can be obtained.
2. Experiment
2.1. Experimental materials and devices
Experimental materials: (1) Synthetic reagents. Oleic acid (unsaturated fatty acid), stearic acid (saturated fatty acid), N, N-dimethyl propylene diamine (DMAPA), 1,3- dibromo propane (bridging agent), analytical reagent, were produced by the Sinopharm Chemical Reagent Co., Ltd. Company. (2) Water for experiments. Deionized water was self-made in the laboratory. The formation water was collected from a low-permeability reservoir flooded by CO2 in Jilin Oilfield, with a total salinity of 21 738 mg/L. Unless otherwise specified, the water used in the experiments was deionized. (3) Oil for experiments. The simulated normal dodecane C12 oil (the purity is 99.9%) was produced by the Aladdin Reagent Co., Ltd. The crude oil was collected from a low-permeability reservoir flooded by CO2 in Jilin Oilfield. (4) CO2 for experiments. CO2 (the purity is 99.9%) was provided by Beijing Zhaoge Gas Technology Co., Ltd. (5) The models of core. According to the standard (SY/T 5336-2006 core analysis method) [27], core models made of quartz sand (as shown in Table 1 ), were used to simulate relatively high-permeability bands in the low-permeability, tight oil reservoirs where gas channeling occurs. The size of the core models was 2.54 cm (diameter) × 100 cm (length), with a permeability of (38-216)×10−3 μm2. Before the experiment, the core models were saturated with deionized water.
Table 1. Parameters of the core models |
| Core number | Permeability/ 10−3 μm2 | Porosity/ % | Pore volume/ mL |
|---|---|---|---|
| 1# | 206.0 | 29.9 | 151.5 |
| 2# | 205.6 | 31.7 | 160.6 |
| 3# | 38.8 | 23.1 | 117.0 |
| 4# | 198.5 | 28.5 | 144.4 |
| 5# | 216.1 | 32.6 | 165.1 |
Experimental devices: (1) Devices for synthesis. The devices include dropping funnel, four-neck flask and condensing reflux device, produced by Beijing Xinweier Glass Instrument Co., Ltd. (2) Analytical and testing devices. The devices include Mars high-temperature and high-pressure rheometer and Glacios cryo-transmission electron microscope, produced by Thermo Fisher Corporation of the United States; AvanceIII nuclear magnetic resonance spectrometer, produced by Bruker Company of the United States; TX500C full-scale spinning drop interfacial tensiometer, produced by Kono Company of the United States; Displacement simulation device with a displacement flow rate ranging from 0.001 to 400.000 mL/min, working pressure ranging from 0.1 to 20.0 MPa, pressure measurement ranging from 0.01 to 5.00 MPa (accuracy 0.25% F.S.), produced by ShanDong ShiYi Science and Technology Co., Ltd. of U.P.C.
2.2. Experimental method
2.2.1. Synthesis, preparation of the smart fluid and its structural characterization
(1) 1.5 mol of DMAPA was added dropwise to 1 mol of oleic acid and then stirred at 170 °C for 6 h. Real-time reaction monitoring was implemented, and a small amount of reactant was taken and titrated by potassium carbonate standard solution (pH=11.0) until the acid value of the system was below 5 mg/g before stopping heating. Finally, excessive DMAPA was distilled at reduced pressure to obtain single-chain oleic acid amide propyl dimethyl tertiary amine. Single-chain stearamidopropyl dimethyl tertiary amine with stearic acid was prepared by the same method. (2) Above two single-chain surfactants were mixed at a molar ratio of 1:1, and dispersed with ethanol. Equal molar amounts of the bridging agent (1,3-dibromopropane) were added to the system, and heated up to 80 °C for 12 h to complete the reaction. Gemini surfactants with asymmetric structures can be obtained as the product. (3) Single-chain oleic acid amide propyl dimethyl tertiary amine (the molar ratio 7:6) was introduced into the Gemini surfactant to obtain the water-soluble single-phase smart fluid.
Two kinds of single-chain surfactants, Gemini surfactants, and smart fluids should go through centrifugal and vacuum drying, respectively, as preparations for structural testing by 1H-NMR spectrometer. The micelle assembly morphology and modulus changes of the smart fluids before and after CO2 exposure was also examined by cryo-TEM and rheometer.
2.2.2. Simulation and evaluation of the characteristics and action mechanisms of the smart fluid
The characteristics and action mechanism of the smart fluid relate to various factors, such as fluid mass fraction, CO2 content, oil content, shear rate, temperature and salinity. To evaluate these factors, a series of experiments, including rheological tests (unless otherwise specified, the test pressure was 0.1 MPa, and the shear rate was 7.34 s−1), core displacement (unless otherwise specified, the simulation temperature was 25 °C, the displacement rate was 0.4 mL/min, system back pressure is 2 MPa), interfacial tension test (temperature of 25 °C), were conducted sequentially. With the help of these tests, the performance of the smart fluid with CO2/simulated oil exposure, the shear behavior of thickened aggregates, and the temperature and salt resistances of the smart fluid and thickened aggregates were evaluated and analyzed.
2.2.2.1. Test on the responsive characteristics of the smart fluid with CO2 exposure
(1) Mass concentration limit of CO2 for triggering the response of the smart fluid: CO2 was continuously injected into the smart fluid with mass fractions of 1.25% and 2.50% to obtain thickened aggregates after complete response, then placed the aggregates into a rheometer. The CO2 content was controlled by heating the aggregates to specific temperatures. The required critical mass concentration of CO2 corresponded to a specific CO2 content is determined according to the viscosity change.
(2) Thickening ratios of the smart fluid at different mass fractions in response to CO2: CO2 was injected into the smart fluids with mass fractions of 0.05% to 2.50%, to obtain thickened aggregates after complete responses. The viscosities and thickening ratios of the samples were measured before and after thickening.
(3) The influence of CO2 phase state and pressure on the thickening performance of the smart fluid: the smart fluid with mass fractions of 0.05%, 0.75% and 1.00% were placed in the rheometer respectively before sealing, and pressurizing with CO2 to obtain thickened aggregates from complete responses. The temperature was set at 45 °C and the pressure at 0.1, 8.0, and 10.0 MPa, respectively. The viscosity of the sample is measured after injection of gaseous CO2 and supercritical CO2.
(4) Simulation of self-thickening performance of the smart fluid in the core model interacting with CO2: 0.75% smart fluid (S) and CO2 were injected into the 1# core model via S-CO2 injection cycles. The responsive characteristics of the smart fluid in the core interacting with CO2 were analyzed through the changes in injection pressure and apparent viscosity. The apparent viscosity can be calculated by the Eq. (3).
$\mu \text{app}=\frac{K\Delta p}{vL}$
To determine the apparent viscosity of the initial smart fluid in the core model without CO2 contact, S was first injected before CO2. CO2 injection lasted for 30-40 min, and the gas-liquid volume ratio in the core model was maintained at 1:2. After each round of injection, the system was sealed for 2.5 h before the next round to ensure sufficient diffusion and mass transfer of S-CO2 in the core model.
(5) Simulation of increased swept volume by the smart fluid in heterogeneous core models: 2# and 3# core models were connected in parallel to make a heterogeneous model with a relative permeability ratio of 5.3. Firstly, the heterogeneity model was displaced by CO2 until the pressure became stable, and then 0.5 PV (pore volume) of 0.75% smart fluid was injected into the model. After 2.5 h of sealing, the model was displaced by CO2 again until the pressure was stable. The pressure during injection process and swept volume change of high- and low-permeability cores were analyzed to evaluate improvement of the swept volume. During the flooding process, the swept volume in the high- and low-permeability cores was estimated by the volume of the displaced liquid.
2.2.2.2. Evaluation of the responsive characteristics of the smart fluid with simulated oil exposure
(1) The capability of the smart fluid to reduce the oil-water interfacial tension after oil exposure: the interfacial tension between the smart fluid and the C12 simulated oil was tested by an interfacial tensiometer. The mass fractions of the smart fluid at 0.10%, 0.50%, 0.75% and 1.00% were chosen. The interfacial tension was measured according to the Surface and Interfacial Tension Measurement Method (SY/T 5370-2018) [28].
(2) Responsive behavior of the smart fluid first interacted with oil and then with CO2: The C12 simulated oil at varied mass fractions (1%, 3%, 5%, 7% and 10%) was added to 0.75% smart fluid, separately. CO2 was injected into the smart fluid with simulated oil, and the viscosity of the sample was tested after a complete response.
(3) Simulation of the behavior of the smart fluid first interacted with oil and then with CO2 in the core model: S-C12-CO2 was injected into 4# core for multiple cycles to simulate the state of CO2 and oil co-existence. The responsive characteristics of the smart fluid in the core model were analyzed by the injection pressure change. The mass fraction of the smart fluid was 0.75%. The mass ratio of C12 to S was 1:9. The volume ratio of CO2 to S was 1:2, and the soaking time after each round of injection was 2.5 h.
2.2.2.3. Test on the self-healing behavior of thickened aggregates
Viscosity change was monitored during shear-static- shear-static cycles, to identify the shear characteristics of the thickened aggregates. The temperature was set at 25 °C and the system back pressure at 0.1 MPa. When the equivalent shear rate in the 5# core model was adjusted to 100 s−1, the pressure change was measured during migration-static-migration-static cycles. The thickened aggregates were produced after a complete response of the 0.75% smart fluid with CO2.
2.2.2.4. Evaluation of temperature, salt resistances of the smart fluids and the thickened aggregates
The crude oil and formation water were from a low-permeability reservoir of CO2 flooding in Jilin Oilfield. The long-term stability evaluation for the temperature and salt resistances of the smart fluids and the thickened aggregates was performed at 90 °C of reservoir temperature. The initial aggregates were prepared by a complete response of the smart fluid (0.75%) and formation water (salinity of 21 738 mg/L). Next, the initial aggregates were placed in an ageing tank for long-term aging at 90 °C. At different stages of ageing, the interfacial tension between the smart fluid and crude oil, and the viscosity of the initial aggregates were recorded. At the same time, injecting CO2 to the aged smart fluids to form thickened aggregates from a complete response, and measuring the viscosity of the aged aggregates.
3. Results and analysis
3.1. Structural characteristics of the smart fluid
Fig. 3. 1H-NMR spectra of the smart fluids and intermediate products (data represents intergral value of 1H-NMR spectra, CDCl3 is chloroform-d). |
Fig. 4. Morphology of micellar assembly in the smart fluids before and after CO2 exposure via cryo-TEM. |
Fig. 5. Modulus changes of the smart fluid before and after CO2 exposure. |
3.2. Responsive characteristics of the smart fluid with CO2 exposure
3.2.1. Response limit and thickening ratios of the smart fluid with CO2 exposure
Fig. 6. (a) Response limit and (b) thickening ratios of the smart fluid with CO2 exposure. |
The thickening ratios of the smart fluids at different mass fractions interacting with CO2 are shown in Fig. 6b . The fluid viscosity was relatively low without CO2 and increased significantly after CO2 exposure. Correspondingly, the thickening ratio was 9 to 246, and the viscosity increased to 13 mPa·s to 3100 mPa·s. Based on the results above, the window of mass fraction for the smart fluid to respond to CO2 is wide, which is favorable for overcoming unfavorable conditions, such as dilution by formation water and adsorption by rocks in underground porous media. Therefore, the needs of sweep-controlling for different flow channels in the reservoir could be satisfied by merely adjusting the mass fraction of the smart fluid on-site.
The influence of phase state and pressure of CO2 on the thickening performance of the smart fluid is shown in Fig. 7 . Comparing the thickened aggregates obtained from supercritical CO2 (45 °C, 8 MPa) and the gaseous CO2 (45 °C, 0.1 MPa), the viscosity of the former increased slightly after a complete response. The phenomenon is either due to the viscosity increase of the supercritical CO2, or the interaction between CO2 and the smart fluid. When the pressure was increased to 10 MPa, the viscosity of the thickened aggregate continued to increase gradually. Therefore, the supercritical CO2 and increased pressure are favorable for the thickening performance of the smart fluids.
Fig. 7. Influence of CO2 phase state and pressure on the thickening performance of the smart fluid. |
3.2.2. Self-thickening characteristics of the smart fluid in core models with CO2 exposure
Fig. 8. Pressure change during two cycles of S-CO2 injection in the core model. |
3.2.3. Improvement of swept volume in the heterogeneous core models
Fig. 9. The sweep efficiency and injection pressure changes in high- and low-permeability core models. |
3.3. Responsive characteristics of the smart fluid with simulated oil exposure
At 25 °C, when the mass fractions of the smart fluid were 0.10%, 0.50%, 0.75% and 1.00%, the corresponding oil-liquid interfacial tensions were 0.950, 0.087, 0.076 and 0.069 mN/m, respectively. Compared with the interfacial tension of 52.8 mN/m for C12-water [29], it is evident that the smart fluid at a low mass fraction can greatly reduce the oil-liquid interfacial tension. When the smart fluid with mass fraction was 0.5%-1.0%, the oil-liquid interfacial tension can be reduced to the magnitude order of 1×10−2 mN/m, reflecting the high interfacial activity of Gemini surfactants. Therefore, a small amount of the smart fluid can be used as an active substance to improve the oil displacement efficiency for the reservoirs with high oil content and low permeability.
Without simulated oil, the viscosity of 0.75% smart fluid after CO2 exposure was 77.8 mPa·s. By increasing the mass fractions of C12 simulated oil from 0 to 1%, 3%, 5%, 7% and 10%, the viscosities of 0.75% smart fluid after CO2 exposure changed to 77.8, 71.1, 52.6, 23.9, 11.3, 2.0 mPa·s, respectively. It can be seen that with more simulated oil in the system, the sample viscosity gradually decreased, and changed to its initial fluid viscosity once the content of simulated oil exceeded 10%. The results show that when the content of simulated oil is below 10%, some vesicular micelles in the oil transform into spheres, and some vesicular micelles assemble into worm-like micelles with CO2 exposure, improving the sweep efficiency and oil displacement. Once the simulated oil content exceeds 10%, the vesicular micelles can completely transform into spherical micelles by solubilizing oil, facilitating oil displacement.
The responsive characteristics of the smart fluid when CO2 and oil coexist in the rock are presented in Fig. 10 . The equilibrium pressures p0, p1, and p2 in different cycles of fluid injection were similar. Also, no significant change in transmittance was observed between the fluids collected at the outlet and the inlet of the core model, indicating that when CO2 and oil coexisted and oil content reached a certain level, the thickening response did not occur.
Fig. 10. Responsive characteristics of the smart fluid when CO2 and oil coexist in core models. |
During CO2 flooding on-site, the smart fluid does not respond to the crude oil in the potential reservoir with high oil content, so it does not thicken and maintains high interfacial activity. The smart fluid can not only avoid damage to reservoirs in low-to-medium-permeability areas, but also play the role of oil displacement.
3.4. Self-healing characteristics of the thickened aggregates
Fig. 11. Experimental results of the thickened aggregates after shearing cycles. |
The thickened aggregate is a three-dimensional structure formed by the intertwinement of worm-like micelles. The winding structure can reorient during shearing and recover during resting, which may explain its structural stability after multiple cycles of shearing, manifesting as the self-healing nature after shear. The self-healing capability of the thickened aggregates is significantly different from that of the chemical cross-linked gel system [30], as the aggregates can maintain structural strength, while the chemical cross-linked gel system degrades during deep migration. The thickened aggregates showed self- healing behavior as shear thinning and static thickening, and thus they can be safely placed in the deep reservoir to effectively maintain the sweep efficiency of CO2 flooding in heterogeneous reservoirs.
3.5. Temperature and salt resistance characteristics of the smart fluid and thickened aggregates
Table 2. Temperature and salt resistances of the smart fluid and thickened aggregates |
| Time/d | Fluid performance | Viscosity of the initial thickened aggregates/ (mPa·s) | |
|---|---|---|---|
| Oil-water interfacial tension/ (mN·m−1) | Viscosity of fluid after complete response with CO2/(mPa·s) | ||
| 0 | 0.073 | 77.8 | 77.8 |
| 10 | 0.075 | 77.3 | 75.4 |
| 30 | 0.080 | 76.5 | 74.7 |
| 60 | 0.083 | 75.4 | 72.7 |
| 120 | 0.087 | 73.4 | 73.2 |
| 300 | 0.087 | 72.8 | 71.1 |
4. Conclusions
The smart fluid for sweep-controlling during CO2 flooding is a water-soluble single-phase system. This system is based on Gemini surfactants with self-assembly properties and tertiary amine groups as the response component. The system has four major characteristics: when interacting with CO2, the smart fluid will thicken to form thickened aggregates. This process will help alleviate gas channeling in high-permeability areas and expand the swept volume. When exposed to crude oil, the smart fluid will not thicken with reduced interfacial tension, and will improve the displacement efficiency in low-permeability areas. The thickened aggregates can self-heal after shear thinning and static thickening in porous media. This feature contributes to the safe placement of the smart fluid in the deep reservoir. The long-term stability of temperature and salt resistance of the smart fluid and thickened aggregates have also been proven. With the above characteristics, the smart fluid can meet the technical requirements for sweep-controlling of enhanced oil recovery (EOR) during CO2 flooding in low-permeability, tight oil reservoirs. The development of the smart fluid is promising to implement CO2 capture, utilization and storage-enhanced oil recovery (CCUS-EOR) in the context of “carbon peaking and carbon neutrality”. The smart fluid can be used in high-efficiency development of low-permeability reservoirs, effective recovery of tight oil, and conversion of development mode for complex reservoirs. All these facts help to continuously improve the swept volume and increase oil recovery in low-to-medium-permeability areas.
Nomenclature
K—permeability, μm2;
L—length of sand filling pipe, cm;
p0—stable injection pressure when the fluid migrates in the core model before contacting CO2, MPa;
p1—stable injection pressure when the fluid migrates in the core model after contacting CO2 for the first time, MPa;
p2—stable injection pressure when the fluid migrates in the core model after contacting CO2 for the second time, MPa;
Δp—injection pressure difference, 105 Pa;
v—injection speed, cm/s;
μapp—apparent viscosity, mPa·s.