- Email: [email protected]
Displacement mechanisms of enhanced heavy oil recovery by alkaline flooding in a micromodel
Particuology 10 (2012) 298–305
Contents lists available at SciVerse ScienceDirect
Particuology journal homepage: www.elsevier.com/locate/partic
Displacement mechanisms of enhanced heavy oil recovery by alkaline flooding in a micromodel Mingzhe Dong a,b,∗ , Qiang Liu c , Aifen Li a a
College of Petroleum Engineering, China University of Petroleum (East), Qingdao, Shandong 255666, China Department of Chemical and Petroleum Engineering, University of Calgary, Calgary AB T2N 1N4, Canada c Brenntag Canada Inc., Calgary AB T2C 0A8, Canada b
a r t i c l e
i n f o
Article history: Received 17 June 2011 Accepted 23 September 2011 Keywords: Micromodel Immiscible displacement Enhanced oil recovery Emulsion flow Alkaline flooding
a b s t r a c t Enhanced oil recovery (EOR) by alkaline flooding for conventional oils has been extensively studied. For heavy oils, investigations are very limited due to the unfavorable mobility ratio between the water and oil phases. In this study, the displacement mechanisms of alkaline flooding for heavy oil EOR are investigated by conducting flood tests in a micromodel. Two different displacement mechanisms are observed for enhancing heavy oil recovery. One is in situ water-in-oil (W/O) emulsion formation and partial wettability alteration. The W/O emulsion formed during the injection of alkaline solution plugs high permeability water channels, and pore walls are altered to become partially oil-wetted, leading to an improvement in sweep efficiency and high tertiary oil recovery. The other mechanism is the formation of an oil-in-water (O/W) emulsion. Heavy oil is dispersed into the water phase by injecting an alkaline solution containing a very dilute surfactant. The oil is then entrained in the water phase and flows out of the model with the water phase. © 2011 Chinese Society of Particuology and Institute of Process Engineering, Chinese Academy of Sciences. Published by Elsevier B.V. All rights reserved.
1. Introduction Alkaline flooding improves oil recovery for conventional oils by using an in situ surfactant generated from the reaction of alkalis and the natural organic acids in oil. The in situ surfactant adsorbs at the oil/water interface, reducing the interfacial tension (IFT). Recovery of the trapped oil is maximized at the minimum IFT (Chatterjee & Wasan, 1998). When the oil/water IFT is reduced to an ultralow level (less than 10−2 dyn/cm), residual oil can be emulsified in the reservoir (Cooke, William, & Kolodzie, 1974; Dong, Liu, Zhang, & Zhu, 1986; Jennings, Johnson, & McAuliffe, 1974; Wasan, Shah, Chan, Sampath, & Shah, 1978). Johnson (1976) summarized three possible mechanisms of alkaline flooding for improving conventional oil recovery. These include (1) dispersion and entrainment of oil, (2) wettability reversal, and (3) emulsification and entrapment of oil. Johnson (1976) also pointed out that each mechanism worked under different injection conditions with respect to the oil, formation rock, and injection water properties; therefore, each process should be designed to improve oil recovery in a different manner.
∗ Corresponding author at: Department of Chemical and Petroleum Engineering, University of Calgary, Calgary AB T2N 1N4, Canada. Tel.: +1 403 210 7642. E-mail address: [email protected] (M. Dong).
Wagner and Leach (1959) presented laboratory tests showing improved oil recovery through the injection of an alkaline solution that reversed rock wettability from oil-wetted to water-wetted. Their results demonstrated that the injected chemical solution could change the wettability of the core samples. The chemicals included acids, bases, and some salts. They believed that the application of alkaline solution should be limited to oil-wetted reservoirs, where wettability could be reversed from oil-wetted to water-wetted. Cooke et al. (1974) reported a mechanism by which sodium hydroxide could improve waterflood oil recovery. They observed that under the proper conditions of pH, salinity, and temperature, some oil reservoirs were converted from water-wetted to oil-wetted. As a result, discontinuous, non-wetting residual oil was converted to a continuous wetted phase, which provided a flow path for what would otherwise be trapped oil. They also found that low interfacial tension could induce the formation of a W/O emulsion. The emulsion droplets tended to block the flow of water and induced a high pressure gradient in the region where they formed. The high pressure gradient, in turn, could overcome the capillary forces already reduced by the low IFT, thereby further reducing residual oil saturation. Laboratory experiments by Jennings et al. (1974) showed that if the IFT were sufficiently low, residual oil in a preferentially waterwetted core could be emulsified in situ, moved downstream within the flowing alkaline solution, and entrapped again by pore throats
1674-2001/$ – see front matter © 2011 Chinese Society of Particuology and Institute of Process Engineering, Chinese Academy of Sciences. Published by Elsevier B.V. All rights reserved.
doi:10.1016/j.partic.2011.09.008
M. Dong et al. / Particuology 10 (2012) 298–305
that are too small for the oil emulsion droplets to penetrate. This would reduce water mobility and improve the sweep efficiency. McAuliffe (1973a, 1973b) also provided experimental evidence that injection of O/W emulsions can improve waterflood oil recovery via the emulsion entrapment mechanism. Jennings et al. (1974) conducted coreflood tests on preserved core samples using a 187 mPa s oil. Their results showed that caustic flooding with 0.1 wt% NaOH solution could significantly improve the waterflood oil recovery. It was believed that in situ O/W emulsions tended to plug growing water fingers and channels and diverted the flow to improve sweep efficiency. Alkaline flooding EOR has been extensively studied for conventional oils, including numerous laboratory experiments and some field tests. For heavy oils, investigations on alkaline flooding EOR are very limited due to the unfavorable mobility ratio between the water and oil phases. Compared with conventional oils, Western Canadian heavy oils are more viscous, and the performance of waterflooding for these oil reservoirs is very poor. The viscosities of these oils range from 1000 to more than 10,000 mPa s. Recent research (Liu, 2006; Liu, Dong, Ma, & Tu, 2007; Liu, Dong, Yue, & Hou, 2006; Ma, 2005; Ma, Dong, Li, & Shirif, 2007) has shown that waterflood recovery of these heavy oils could be greatly improved by alkaline flooding. Laboratory sandpack flood tests in these studies showed that a dilute chemical solution injection can cause either (1) heavy oil to be broken up into small droplets, entrained in the water phase, and carried out of the oil sands or (2) water flow into channels to be blocked because of dispersion of the water-in-oil phase, resulting in improved sweep efficiency. If the displacement mechanisms are well understood, waterflooding for these heavy oil reservoirs can be improved by applying an optimal chemical formula to realize a displacement process most suited to the situation. The objective of this study is to analyze the heavy oil EOR mechanisms in alkaline flooding by observing two distinct displacement processes in micromodel tests. The heavy oil tested by Liu (2006) was used in this study. High tertiary recoveries were obtained for this oil from alkaline or alkaline/surfactant injection in sandpack flood tests. Alkaline flooding tests for other heavy oils from Western Canada showed similar tertiary recovery phenomena by applying the above two displacement mechanisms (Liu, 2006; Ma, 2005). Therefore, the displacement mechanisms discussed in this paper are believed to be applicable to other heavy oils.
2. Experimental Oil and formation brine samples were collected from a heavy oil reservoir in Alberta, Canada. The heavy oil was centrifuged at 10,000 rpm at 35 ◦ C for 2 h to remove water and solids in the oil. The oil had a viscosity of 1800 mPa s and a density of 0.964 g/cm3 at 22 ◦ C. The formation brine had a salinity of 2.7 wt%. The chemical agents used in this study included NaOH, Na2 CO3 , and the surfactant Stepanol Me Dry (sodium lauryl sulfate from Stepan, Canada). The micromodel used in this study was made of two glass plates, and a two-dimensional network of pores and throats was etched into the top plate (Chatzis, Morrow, & Lim, 1983). The micromodel consists of a porous area of 8.0 cm × 4.5 cm and a pore volume (PV) of 0.15 cm3 . The transparent nature of the micromodel allowed the pore-scale multi-phase displacement and wettability of the pore surfaces to be visually observed. The glass micromodel was cleaned first with Varsol (commercial paint thinner) and then with ethanol to remove any residual oil in the model after a displacement test. After cleaning, the model was blown dry using air to remove the solvent. To keep the glass model strongly water-wetted, the model was then heated in a muffle furnace at 400 ◦ C for 1 h to remove trace
299
organic compounds adsorbed on the pore walls. The displacement procedure of the micromodel test is as follows: (1) (2) (3) (4)
Saturate the micromodel with brine; Inject the heavy oil at 60 ◦ C; Conduct waterflooding for two pore volumes at 60 ◦ C; Conduct an alkaline flood at ambient temperature.
The injection rate in all the stages was 0.075 cm3 /h (0.5 PV/h or a velocity of 0.5 m/day). In Step 2, a higher temperature was applied to obtain a relatively high water saturation in the model. In Step 3, the mobility of heavy oil was higher at 60 ◦ C, which allowed a better sweep efficiency for water flooding or a relatively low residual oil saturation. The purpose of achieving low residual oil saturation was to create disconnected heavy oil ganglia in the two-dimensional pore network model for testing the two displacement mechanisms. The oil recovery was recorded as the percentage of the initial oil in place (IOIP) during different displacement stages. Polyethylene tubing with an ID of 2.0 mm was connected to the outlet of the micromodel. The produced oil volume was calculated from the length of oil slugs in the tube and the ID of the tube. The water content in the emulsion was assessed under a microscope, and the oil production volume was corrected accordingly. 3. Results and discussion 3.1. Two displacement mechanisms The distribution of the heavy oil and connate water in the micromodel is shown in Fig. 1. Fig. 1(a) is a picture of the micromodel containing heavy oil (in black) and irreducible water at the end of the oil injection. The transparent parts of the picture are sintered glass without pores. Fig. 1(b) is a pore-level image of the oil and water distribution, which is an enlarged image of a region in Fig. 1(a). This image shows water films surrounding the solid boundaries and the continuous oil phase occupying the central portion of pores and throats of the pore network. The oil color in Fig. 1(b) is brown due to the high brightness of the light. The image also illustrates that the micromodel is water-wetted. Fig. 2 illustrates the distribution of oil in the micromodel during and after waterflooding. Fig. 2(a)–(c) shows oil saturation changes at two locations, L1 and L2. Fig. 2(d) shows a larger portion of the micromodel with disconnected oil ganglia that were formed and trapped in the pore network after waterflooding. As an example, an isolated oil ganglion is labeled in Fig. 2(d). With the injection of brine, the heavy oil at L1 and L2 became separated (Fig. 2(b) and (c)). After waterflooding, the glass pore network remained waterwetted. As shown in Fig. 2(d), the oil became discontinuous after waterflooding, when the water cut in the effluent samples was nearly 100%. The disconnected oil ganglia could not be recovered by further injection of water. The trapped residual oil was in the form of either small oil drops or ganglia and occupied many pore bodies of the pore network. Two displacement mechanisms of the residual oil were observed in the micromodel tests. One was the formation of a W/O emulsion and partial wettability alteration, and the other was emulsification and entrainment of heavy oil in the water phase. It has been shown by Ma (2005) and Ma et al. (2007) that injection of an alkaline solution (NaOH and Na2 CO3 ) could markedly improve heavy oil recovery in channeled sandpack flood tests, which simulate water channeling phenomena in heavy oil reservoirs. Fig. 3 shows pictures of the micromodel network with injection of a chemical slug containing 0.40 wt% NaOH and 0.20 wt% Na2 CO3 . During the alkaline injection, the injected water phase penetrated
300
M. Dong et al. / Particuology 10 (2012) 298–305
Fig. 1. Distribution of heavy oil and water in the micromodel. (a) Heavy oil in the model, (b) enlarged image of a region of (a). Pore-level distribution showing water films.
into the residual oil phase to form some discontinuous water ganglia inside the oil phase, or a water-in-oil (W/O) emulsions. The formation of a W/O emulsion was confirmed by examining the produced oil under a microscope. The produced oil could contain up to 40% water. Sandpack flow tests by Arhuoma, Dong, Yang, and Idem (2009) showed that the viscosity of a W/O emulsion is much higher than the viscosity of the water phase and even higher than the viscosity of the oil phase. Based on their experimental results, Wang,
Dong, and Arhuoma (2010) derived the following correlation for the effective viscosity of a W/O emulsion in sandpacks: log(e ) = k + log(o ),
(1)
where e is the effective viscosity of the W/O emulsion, o is the oil phase viscosity, and is the emulsion quality. The value of the coefficient k is larger than unity and depends on the porosity and permeability of the porous medium through which the emulsion
Fig. 2. Trapping of heavy oil during waterflood in the micromodel. (a)–(c) Pore-level images of the trapping process, (d) ganglia of residual oil in the micromodel (an example is illustrated).
M. Dong et al. / Particuology 10 (2012) 298–305
301
Fig. 3. W/O emulsion formation and mobilization of trapped oil ganglia during alkaline injection in a micromodel test. Alkaline slug: 0.40 wt% NaOH + 0.20 wt% Na2 CO3 . (a) Formation of a W/O emulsion, (b) oil front approaching Oil 1, (c) oil front approaching Oil 2, (d) oil front approaching Oils 3, 4 and 5.
is flowing. Eq. (1) shows that in the same porous medium (i.e., k is constant), the effective viscosity of the W/O emulsion increases with increasing emulsion quality. With increasing emulsion quality (water content), the effective viscosity of the W/O emulsion becomes much higher than the oil viscosity. This phenomenon can be exploited in chemical flooding for heavy oil recovery to reduce the water mobility in water channels, which results in an improved sweep efficiency. Therefore, the oil can be displaced in the form of a W/O emulsion with minimal water fingering effects. Some oil drops also touched the pore walls, indicating that the pore wall became partially oil-wetted (Liu, 2006; Liu, Dong, Asghari, & Tu, 2007). The wettability alteration also aided in reducing water flow in the high-water-saturation regions. 7 The images in Fig. 3 also show that the water phase takes the central portion of the pore spaces occupied by residual oil during the formation of the W/O emulsion, instead of bypassing the oil ganglia by flowing through the existing water paths. This is how the process makes the trapped oil continuous and recoverable. Fig. 3(a)–(d) shows the process in which trapped oil ganglia 1–5 are merged with the displacement front through the W/O emulsification mechanism in a micromodel test. In Fig. 3(a), the oil bank is driven by the W/O emulsion and moves (from left to right) forward to approach the oil ganglia. In Fig. 3(b), the oil front is approaching Oil 1. After being merged with the W/O emulsion, Oil 1 is mobilized and driven forward to touch Oil 2 (Fig. 3(c)) and then Oils 3–5 (Fig. 3(d)). With this oil mobilization mechanism, the trapped oil ganglia become continuous and are displaced out of the micromodel. Fig. 4 shows the pressure drop curve for the above micromodel test of alkaline flooding. The pressure drop along the micromodel
grew from 0 to 0.8 pore volume (PV) during the oil saturation process. During waterflooding, the pressure drop continued to increase for the 0.2 PV injection and then decreased with the reset of the waterflood. When the pressure drop declined to a certain low value (at 2.4 PV), a chemical slug injection was initiated. As shown in Fig. 4, the pressure drop increased continuously up to 12.4 kPa due to the formation of a W/O emulsion, which had a higher viscosity than the oil. At 3.9 PV, a rapid decrease in the pressure drop was
Fig. 4. Pressure drop as a function of time in a micromodel test. Chemical slug: 0.40 wt% NaOH + 0.20 wt% Na2 CO3 .
302
M. Dong et al. / Particuology 10 (2012) 298–305
Fig. 5. Cumulative oil recovery as a function of fluid injected in a micromodel test. Chemical slug: 0.40 wt% NaOH + 0.20 wt% Na2 CO3 .
observed because the in situ formed W/O emulsion broke through the model outlet. After breakthrough, no further oil was recovered. Fig. 5 shows the oil recovery in the micromodel test plotted as a function of PV injection during the waterflooding and chemical injection. Waterflooding recovered approximately 28% of the IOIP, and the alkaline flooding recovered an additional 33% of the IOIP. These results show the improvement in sweep efficiency and oil recovery provided by the W/O emulsion flow.
Fig. 6 shows pore-level microphotographs taken during the above test and shows the oil and water distribution in different displacement stages. A water film between the oil and pore wall existed before the injection of the alkaline slug. After the alkaline flooding, an oil film existed between the water and the surface of the pores, indicating that the pore walls became preferentially oilwetted. The oil/water meniscus in Fig. 6(b) was convex to the oil phase, suggesting that the pore wall was oil-wetted. To investigate why the alkaline solution prefers to penetrate into the oil phase rather than flow through the water paths in the above test, additional micromodel tests were carried out by injecting a 0.40 wt% NaOH plus 0.20 wt% Na2 CO3 solution. This solution had a pH of 12.8. The experimental results showed that a W/O emulsion was formed by injecting NaOH rather than Na2 CO3 . NaOH is a strong base, and the interaction between NaOH in the water phase and organic acids in heavy oil is dramatic. It is hypothesized that mass transfer occurs more rapidly at the oil/water interface with a NaOH solution than with a Na2 CO3 solution. To initiate the penetration of the alkaline solution into the residual oil phase, a low interfacial tension between the oil and water is necessary. In the presence of a 0.40 wt% NaOH solution, the IFT of oil/water is lower than 0.1 dyn/cm. The lowered IFT is believed to be caused by the ionization of organic acids in the oil to form in situ surfactants as a result of the added alkalis. In the water-wetted glass network model, oil is in the central portion of the pores with thin water films between the oil and pore walls. A rapid reaction results in a very low instantaneous IFT between the injected alkaline solution and the residual oil, and the alkaline solution contacts the oil
Fig. 6. Pictures of one location in the micromodel at four stages in the oil displacement process. Chemical slug: 0.40 wt% NaOH + 0.20 wt% Na2 CO3 . (a) After water flooding, (b) after alkaline flooding, (c) 50 h after alkaline flooding, (d) 150 h after alkaline flooding.
M. Dong et al. / Particuology 10 (2012) 298–305
303
Fig. 7. Emulsification and entrainment of heavy oil in the water phase in a micromodel test. (a) Pore-level image, (b) image of the micromodel during oil displacement.
drops first in the central portion of the pores to form a W/O emulsion. During the process of emulsion movement, the pore walls also become partially oil-wetted because of the interaction of the heavy oil/alkaline solution interfaces with the glass walls. In sandpack flood tests (Liu, 2006; Ma, 2005), Na2 CO3 was also used together with NaOH. The calcium ion in brine can make the ionized in situ surfactants saponified and thus lose surface activity. The use of Na2 CO3 in the water phase can precipitate Ca2+ in brine, lowering the oil/brine IFT more efficiently. Therefore, the combination of NaOH and Na2 CO3 is favorable for lowering the IFT and enhancing oil recovery. In sandpack flood tests by Ma (2005) and Liu (2006), it was found that the O/W emulsion mechanism occurred for enhanced oil recovery when injecting alkaline/surfactant formulas. In micromodel tests, it was confirmed that the O/W emulsion mechanism dominated the displacement in EOR by injecting an alkaline/surfactant (A/S) formula (0.20 wt% NaOH + 0.40 wt% Na2 CO3 + 450 mg/L surfactant). Fig. 7 shows the emulsification and
entrainment of the heavy oil in the water phase with the injection of an A/S slug. Because of the synergy of the surfactant and alkali (Liu, 2006), the interfacial tension between the oil and water was lowered to 10−3 dyn/cm. The oil was emulsified and entrained in the water phase. The emulsion flow in the porous media also led to an increase in the pressure drop and an improvement in the oil recovery in sandpack flood tests. However, the increase in the pressure drop in this process was not as large as in the W/O emulsion process. Using water as the external phase of the O/W emulsion results in this type of emulsion having a similar viscosity as the water phase. 3.2. Comparison of the sweep efficiency of two displacement processes The two displacement processes resulted in different recovery efficiencies, which could be visualized in the micromodel tests. Fig. 8 shows pictures of the entire micromodel at different stages of
Fig. 8. Photos of the micromodel during an alkaline flood test showing the improvement of sweep efficiency. (a) After waterflood, (b) early stage of alkaline injection, (c) alkaline slug reached production end, (d) at 1.0 PV alkaline slug injection.
304
M. Dong et al. / Particuology 10 (2012) 298–305
Fig. 9. Photos of the micromodel during alkaline/surfactant injection, showing the formation of water channels during chemical injection. (a) After waterflood, (b) early stage of chemical injection, (c) chemical slug reached production end, (d) at 1.0 PV chemical slug injection.
alkaline flooding. Fig. 8(a) was taken after the initial waterflood and shows that some water channels (light colored) were created and some regions were not approached by the waterflood. The sweep efficiency by waterflooding was approximately 70% based on the area swept by the waterflood. When the alkaline solution was injected, as shown in Fig. 8(b), the water channels were eliminated by re-mobilized residual oil, and the area behind the displacement front became relatively uniform. It is the blocking of water channels that diverted the injected alkaline solution phase to the unswept region of the model to improve the sweep efficiency. Fig. 8(c) shows the micromodel when the alkaline slug front reached the outlet and shows a relatively uniform oil saturation distribution over the entire model. The alkaline solution reached almost every location inside the micromodel and improved the sweep efficiency to about 90%. Fig. 8(d) shows the model after approximately one pore volume of alkaline solution injection. From the distribution of the water phase (the color of the model), it is seen that the oil saturation was still uniform and greatly reduced at the end of the displacement when compared with the model in Fig. 8(c). Fig. 9 shows pictures of the entire micromodel at different stages of the A/S injection (0.20 wt% NaOH + 0.40 wt% Na2 CO3 + 450 mg/L surfactant). Fig. 9(a) shows the oil saturation distribution after waterflooding, and some water channels can be seen. In the early stage of the chemical injection, as shown in Fig. 9(b) and (c), the sweep efficiency was improved by the injection of the A/S solution. The water channels disappeared in the region behind the displacement front. By injecting an A/S slug, an O/W emulsion was formed in the pores, and the residual oil was mobilized by the O/W emulsion flow. After the injection of one PV of the chemical solution, as shown in Fig. 9(d), new water channels were created. In this situation, the oil production declined rapidly, and the injection of additional A/S solution had a poor sweep efficiency. Water channeling is a common problem during waterflooding in heavy oil reservoirs because of the high viscosity of heavy oils and the heterogeneity of oil reservoirs. To improve sweep efficiency
and enhance heavy oil recovery, water mobility in water channels and high permeability zones needs to be reduced significantly to divert the flow to oil-rich channels. The above experimental results showed that water channels could be more effectively blocked by injecting an alkaline slug rather than an A/S slug. 3.3. Application of the two displacement mechanisms The mechanisms of in situ W/O flows and of O/W emulsion flows in porous media are different and should be applied for different reservoir conditions for enhanced heavy oil recovery by chemical flooding. Numerous experimental results by Ma (2005) and Liu (2006) indicate that the viscosity of a W/O emulsion varies from 2000 to 20,000 mPa s, depending on the ratio of water/oil and on the alkaline concentration in the system. An O/W emulsion has a much lower viscosity than a W/O emulsion because it tends to have the viscosity of the external phase, i.e., the water phase. An O/W emulsion is produced using alkalis and hydrophilic surfactants. The synergy of alkalis and the hydrophilic surfactant can reduce the water and heavy oil IFT to an ultralow value. An O/W emulsion can be created with a minor disturbance at the oil and water interface. Numerous emulsification tests have shown that very gentle turning of a test tube that contains the above oil and water system can make O/W emulsions. However, the W/O emulsion observed in the micromodel displacements could not be easily formed in bottle tests, even when the oil and water phase were mixed by vigorous hand shaking. A possible reason is that the oil was so viscous that the oil and alkaline solution did not have good contact, whereas the alkaline solution and the oil in a porous medium can have pore-level contact. It was also observed that a W/O emulsion can be easily generated in sandpack flood tests by injecting an alkaline slug (Ma, 2005). Many sandpack flood tests have shown that the in situ-formed W/O emulsion can effectively block water channels, which leads to highly enhanced oil recovery. For some heavy oil recovery processes, the O/W emulsion mechanism can be applied
M. Dong et al. / Particuology 10 (2012) 298–305
to improve heavy oil recovery by dispersing oil in the water phase. In this case, residual oil is entrained in the water phase and flows to the production well. 4. Conclusions In alkaline flooding for enhanced heavy oil recovery, the alkaline solution can penetrate into heavy oil in porous media and form W/O emulsions. Because of the very high viscosity of the W/O emulsion, the resistance to water flow in high-water-saturation zones can be increased significantly to improve sweep efficiency and oil recovery. When an alkaline/surfactant formula is used to create an ultralow IFT and an O/W emulsion for a heavy oil, the oil can be produced by the O/W emulsion flow mechanism. Each of the two displacement processes can be designed to improve heavy oil recovery in accordance with the different displacement mechanisms. Acknowledgements The authors acknowledge the Petroleum Technology Research Center (PTRC) in Regina, Saskatchewan, Canada and the Natural Sciences and Engineering Research Council of Canada (NSERC) for their financial support of this work. The authors thank Dr. I. Chatzis of the University of Waterloo for providing the glass micromodel used in this study. References Arhuoma, M., Dong, M., Yang, D., & Idem, R. (2009). Determination of water-in-oil emulsion viscosity in porous media. Industrial & Engineering Chemistry Research, 48, 7092–7120. Chatterjee, J., & Wasan, D. T. (1998). A kinetic model for dynamic interfacial tension variation in an acidic oil/alkali/surfactant system. Chemical Engineering Science, 53(15), 2711–2725.
305
Chatzis, I., Morrow, N. R., & Lim, H. L. (1983). Magnitude and detailed structure of residual oil saturation. Society of Petroleum Engineers Journal, 23, 311–326. Cooke, C. E., Jr., William, R. E., & Kolodzie, P. A. (1974). Oil recovery by alkaline waterflooding. Journal of Petroleum Technology, 26, 1365–1374. Dong, M., Liu, P., Zhang, L., & Zhu, Y. (1986). An improved model for interfacial activity of acidic oil/caustic system for alkaline flooding. In SPE 14855 presented at the international meeting on petroleum engineering March 17–20, Beijing, China. Jennings, H. Y., Jr., Johnson, C. E., Jr., & McAuliffe, C. D. (1974). A caustic waterflooding process for heavy oils. Journal of Petroleum Technology, 26, 1344–1352. Johnson, C. E., Jr. (1976). Status of caustic and emulsion methods. Journal of Petroleum Technology, 28, 85–92. Liu, Q. (2006). Interfacial phenomena in enhanced heavy oil recovery by alkaline flood. Unpublished doctoral dissertation, University of Regina, Regina, Saskatchewan, Canada. Liu, Q., Dong, M., Asghari, K., & Tu, Y. (2007). Wettability alteration by magnesium ion binding in heavy oil/brine/chemical/sand systems—Analysis of electrostatic forces. Journal of Petroleum Science and Engineering, 59, 147–156. Liu, Q., Dong, M., Ma, S., & Tu, Y. (2007). Surfactant enhanced alkaline flooding for Western Canadian heavy oil recovery. Colloids and Surfaces A: Physicochemical and Engineering Aspects, 293, 63–71. Liu, Q., Dong, M., Yue, X., & Hou, J. (2006). Synergy of alkali and surfactant in emulsification of heavy oil in brine. Colloids and Surfaces A: Physicochemical and Engineering Aspects, 273, 219–228. Ma, S. (2005). Enhanced heavy oil recovery by dilute alkaline flooding. Master thesis, University of Regina, Regina, Saskatchewan, Canada. Ma, S., Dong, M., Li, Z., & Shirif, E. (2007). Evaluation of the effectiveness of chemical flooding using heterogeneous sandpack flood test. Journal of Petroleum Science and Engineering, 55, 294–300. McAuliffe, C. D. (1973a). Oil-in-water emulsions and their flow properties in porous media. Journal of Petroleum Technology, 25, 727–733. McAuliffe, C. D. (1973b). Crude-oil-in-water emulsions to improve fluid flow in an oil reservoir. Journal of Petroleum Technology, 25, 721–726. Wagner, O. R., & Leach, R. O. (1959). Improving oil displacement efficiency by wettability adjustment. Petroleum Transactions, AIME, 216, 65–72. Wang, J., Dong, M., & Arhuoma, M. (2010). Experimental and numerical study of improving heavy oil recovery by alkaline flooding in sandpacks. Journal of Canadian Petroleum Technology, 49(3), 51–57. Wasan, D. T., Shah, S. M., Chan, M., Sampath, K., & Shah, R. (1978). Observations on emulsion stability and interfacial properties in improved oil recovery by chemical flooding. American Chemical Society, 23(2), 705–727.
Contents lists available at SciVerse ScienceDirect
Particuology journal homepage: www.elsevier.com/locate/partic
Displacement mechanisms of enhanced heavy oil recovery by alkaline flooding in a micromodel Mingzhe Dong a,b,∗ , Qiang Liu c , Aifen Li a a
College of Petroleum Engineering, China University of Petroleum (East), Qingdao, Shandong 255666, China Department of Chemical and Petroleum Engineering, University of Calgary, Calgary AB T2N 1N4, Canada c Brenntag Canada Inc., Calgary AB T2C 0A8, Canada b
a r t i c l e
i n f o
Article history: Received 17 June 2011 Accepted 23 September 2011 Keywords: Micromodel Immiscible displacement Enhanced oil recovery Emulsion flow Alkaline flooding
a b s t r a c t Enhanced oil recovery (EOR) by alkaline flooding for conventional oils has been extensively studied. For heavy oils, investigations are very limited due to the unfavorable mobility ratio between the water and oil phases. In this study, the displacement mechanisms of alkaline flooding for heavy oil EOR are investigated by conducting flood tests in a micromodel. Two different displacement mechanisms are observed for enhancing heavy oil recovery. One is in situ water-in-oil (W/O) emulsion formation and partial wettability alteration. The W/O emulsion formed during the injection of alkaline solution plugs high permeability water channels, and pore walls are altered to become partially oil-wetted, leading to an improvement in sweep efficiency and high tertiary oil recovery. The other mechanism is the formation of an oil-in-water (O/W) emulsion. Heavy oil is dispersed into the water phase by injecting an alkaline solution containing a very dilute surfactant. The oil is then entrained in the water phase and flows out of the model with the water phase. © 2011 Chinese Society of Particuology and Institute of Process Engineering, Chinese Academy of Sciences. Published by Elsevier B.V. All rights reserved.
1. Introduction Alkaline flooding improves oil recovery for conventional oils by using an in situ surfactant generated from the reaction of alkalis and the natural organic acids in oil. The in situ surfactant adsorbs at the oil/water interface, reducing the interfacial tension (IFT). Recovery of the trapped oil is maximized at the minimum IFT (Chatterjee & Wasan, 1998). When the oil/water IFT is reduced to an ultralow level (less than 10−2 dyn/cm), residual oil can be emulsified in the reservoir (Cooke, William, & Kolodzie, 1974; Dong, Liu, Zhang, & Zhu, 1986; Jennings, Johnson, & McAuliffe, 1974; Wasan, Shah, Chan, Sampath, & Shah, 1978). Johnson (1976) summarized three possible mechanisms of alkaline flooding for improving conventional oil recovery. These include (1) dispersion and entrainment of oil, (2) wettability reversal, and (3) emulsification and entrapment of oil. Johnson (1976) also pointed out that each mechanism worked under different injection conditions with respect to the oil, formation rock, and injection water properties; therefore, each process should be designed to improve oil recovery in a different manner.
∗ Corresponding author at: Department of Chemical and Petroleum Engineering, University of Calgary, Calgary AB T2N 1N4, Canada. Tel.: +1 403 210 7642. E-mail address: [email protected] (M. Dong).
Wagner and Leach (1959) presented laboratory tests showing improved oil recovery through the injection of an alkaline solution that reversed rock wettability from oil-wetted to water-wetted. Their results demonstrated that the injected chemical solution could change the wettability of the core samples. The chemicals included acids, bases, and some salts. They believed that the application of alkaline solution should be limited to oil-wetted reservoirs, where wettability could be reversed from oil-wetted to water-wetted. Cooke et al. (1974) reported a mechanism by which sodium hydroxide could improve waterflood oil recovery. They observed that under the proper conditions of pH, salinity, and temperature, some oil reservoirs were converted from water-wetted to oil-wetted. As a result, discontinuous, non-wetting residual oil was converted to a continuous wetted phase, which provided a flow path for what would otherwise be trapped oil. They also found that low interfacial tension could induce the formation of a W/O emulsion. The emulsion droplets tended to block the flow of water and induced a high pressure gradient in the region where they formed. The high pressure gradient, in turn, could overcome the capillary forces already reduced by the low IFT, thereby further reducing residual oil saturation. Laboratory experiments by Jennings et al. (1974) showed that if the IFT were sufficiently low, residual oil in a preferentially waterwetted core could be emulsified in situ, moved downstream within the flowing alkaline solution, and entrapped again by pore throats
1674-2001/$ – see front matter © 2011 Chinese Society of Particuology and Institute of Process Engineering, Chinese Academy of Sciences. Published by Elsevier B.V. All rights reserved.
doi:10.1016/j.partic.2011.09.008
M. Dong et al. / Particuology 10 (2012) 298–305
that are too small for the oil emulsion droplets to penetrate. This would reduce water mobility and improve the sweep efficiency. McAuliffe (1973a, 1973b) also provided experimental evidence that injection of O/W emulsions can improve waterflood oil recovery via the emulsion entrapment mechanism. Jennings et al. (1974) conducted coreflood tests on preserved core samples using a 187 mPa s oil. Their results showed that caustic flooding with 0.1 wt% NaOH solution could significantly improve the waterflood oil recovery. It was believed that in situ O/W emulsions tended to plug growing water fingers and channels and diverted the flow to improve sweep efficiency. Alkaline flooding EOR has been extensively studied for conventional oils, including numerous laboratory experiments and some field tests. For heavy oils, investigations on alkaline flooding EOR are very limited due to the unfavorable mobility ratio between the water and oil phases. Compared with conventional oils, Western Canadian heavy oils are more viscous, and the performance of waterflooding for these oil reservoirs is very poor. The viscosities of these oils range from 1000 to more than 10,000 mPa s. Recent research (Liu, 2006; Liu, Dong, Ma, & Tu, 2007; Liu, Dong, Yue, & Hou, 2006; Ma, 2005; Ma, Dong, Li, & Shirif, 2007) has shown that waterflood recovery of these heavy oils could be greatly improved by alkaline flooding. Laboratory sandpack flood tests in these studies showed that a dilute chemical solution injection can cause either (1) heavy oil to be broken up into small droplets, entrained in the water phase, and carried out of the oil sands or (2) water flow into channels to be blocked because of dispersion of the water-in-oil phase, resulting in improved sweep efficiency. If the displacement mechanisms are well understood, waterflooding for these heavy oil reservoirs can be improved by applying an optimal chemical formula to realize a displacement process most suited to the situation. The objective of this study is to analyze the heavy oil EOR mechanisms in alkaline flooding by observing two distinct displacement processes in micromodel tests. The heavy oil tested by Liu (2006) was used in this study. High tertiary recoveries were obtained for this oil from alkaline or alkaline/surfactant injection in sandpack flood tests. Alkaline flooding tests for other heavy oils from Western Canada showed similar tertiary recovery phenomena by applying the above two displacement mechanisms (Liu, 2006; Ma, 2005). Therefore, the displacement mechanisms discussed in this paper are believed to be applicable to other heavy oils.
2. Experimental Oil and formation brine samples were collected from a heavy oil reservoir in Alberta, Canada. The heavy oil was centrifuged at 10,000 rpm at 35 ◦ C for 2 h to remove water and solids in the oil. The oil had a viscosity of 1800 mPa s and a density of 0.964 g/cm3 at 22 ◦ C. The formation brine had a salinity of 2.7 wt%. The chemical agents used in this study included NaOH, Na2 CO3 , and the surfactant Stepanol Me Dry (sodium lauryl sulfate from Stepan, Canada). The micromodel used in this study was made of two glass plates, and a two-dimensional network of pores and throats was etched into the top plate (Chatzis, Morrow, & Lim, 1983). The micromodel consists of a porous area of 8.0 cm × 4.5 cm and a pore volume (PV) of 0.15 cm3 . The transparent nature of the micromodel allowed the pore-scale multi-phase displacement and wettability of the pore surfaces to be visually observed. The glass micromodel was cleaned first with Varsol (commercial paint thinner) and then with ethanol to remove any residual oil in the model after a displacement test. After cleaning, the model was blown dry using air to remove the solvent. To keep the glass model strongly water-wetted, the model was then heated in a muffle furnace at 400 ◦ C for 1 h to remove trace
299
organic compounds adsorbed on the pore walls. The displacement procedure of the micromodel test is as follows: (1) (2) (3) (4)
Saturate the micromodel with brine; Inject the heavy oil at 60 ◦ C; Conduct waterflooding for two pore volumes at 60 ◦ C; Conduct an alkaline flood at ambient temperature.
The injection rate in all the stages was 0.075 cm3 /h (0.5 PV/h or a velocity of 0.5 m/day). In Step 2, a higher temperature was applied to obtain a relatively high water saturation in the model. In Step 3, the mobility of heavy oil was higher at 60 ◦ C, which allowed a better sweep efficiency for water flooding or a relatively low residual oil saturation. The purpose of achieving low residual oil saturation was to create disconnected heavy oil ganglia in the two-dimensional pore network model for testing the two displacement mechanisms. The oil recovery was recorded as the percentage of the initial oil in place (IOIP) during different displacement stages. Polyethylene tubing with an ID of 2.0 mm was connected to the outlet of the micromodel. The produced oil volume was calculated from the length of oil slugs in the tube and the ID of the tube. The water content in the emulsion was assessed under a microscope, and the oil production volume was corrected accordingly. 3. Results and discussion 3.1. Two displacement mechanisms The distribution of the heavy oil and connate water in the micromodel is shown in Fig. 1. Fig. 1(a) is a picture of the micromodel containing heavy oil (in black) and irreducible water at the end of the oil injection. The transparent parts of the picture are sintered glass without pores. Fig. 1(b) is a pore-level image of the oil and water distribution, which is an enlarged image of a region in Fig. 1(a). This image shows water films surrounding the solid boundaries and the continuous oil phase occupying the central portion of pores and throats of the pore network. The oil color in Fig. 1(b) is brown due to the high brightness of the light. The image also illustrates that the micromodel is water-wetted. Fig. 2 illustrates the distribution of oil in the micromodel during and after waterflooding. Fig. 2(a)–(c) shows oil saturation changes at two locations, L1 and L2. Fig. 2(d) shows a larger portion of the micromodel with disconnected oil ganglia that were formed and trapped in the pore network after waterflooding. As an example, an isolated oil ganglion is labeled in Fig. 2(d). With the injection of brine, the heavy oil at L1 and L2 became separated (Fig. 2(b) and (c)). After waterflooding, the glass pore network remained waterwetted. As shown in Fig. 2(d), the oil became discontinuous after waterflooding, when the water cut in the effluent samples was nearly 100%. The disconnected oil ganglia could not be recovered by further injection of water. The trapped residual oil was in the form of either small oil drops or ganglia and occupied many pore bodies of the pore network. Two displacement mechanisms of the residual oil were observed in the micromodel tests. One was the formation of a W/O emulsion and partial wettability alteration, and the other was emulsification and entrainment of heavy oil in the water phase. It has been shown by Ma (2005) and Ma et al. (2007) that injection of an alkaline solution (NaOH and Na2 CO3 ) could markedly improve heavy oil recovery in channeled sandpack flood tests, which simulate water channeling phenomena in heavy oil reservoirs. Fig. 3 shows pictures of the micromodel network with injection of a chemical slug containing 0.40 wt% NaOH and 0.20 wt% Na2 CO3 . During the alkaline injection, the injected water phase penetrated
300
M. Dong et al. / Particuology 10 (2012) 298–305
Fig. 1. Distribution of heavy oil and water in the micromodel. (a) Heavy oil in the model, (b) enlarged image of a region of (a). Pore-level distribution showing water films.
into the residual oil phase to form some discontinuous water ganglia inside the oil phase, or a water-in-oil (W/O) emulsions. The formation of a W/O emulsion was confirmed by examining the produced oil under a microscope. The produced oil could contain up to 40% water. Sandpack flow tests by Arhuoma, Dong, Yang, and Idem (2009) showed that the viscosity of a W/O emulsion is much higher than the viscosity of the water phase and even higher than the viscosity of the oil phase. Based on their experimental results, Wang,
Dong, and Arhuoma (2010) derived the following correlation for the effective viscosity of a W/O emulsion in sandpacks: log(e ) = k + log(o ),
(1)
where e is the effective viscosity of the W/O emulsion, o is the oil phase viscosity, and is the emulsion quality. The value of the coefficient k is larger than unity and depends on the porosity and permeability of the porous medium through which the emulsion
Fig. 2. Trapping of heavy oil during waterflood in the micromodel. (a)–(c) Pore-level images of the trapping process, (d) ganglia of residual oil in the micromodel (an example is illustrated).
M. Dong et al. / Particuology 10 (2012) 298–305
301
Fig. 3. W/O emulsion formation and mobilization of trapped oil ganglia during alkaline injection in a micromodel test. Alkaline slug: 0.40 wt% NaOH + 0.20 wt% Na2 CO3 . (a) Formation of a W/O emulsion, (b) oil front approaching Oil 1, (c) oil front approaching Oil 2, (d) oil front approaching Oils 3, 4 and 5.
is flowing. Eq. (1) shows that in the same porous medium (i.e., k is constant), the effective viscosity of the W/O emulsion increases with increasing emulsion quality. With increasing emulsion quality (water content), the effective viscosity of the W/O emulsion becomes much higher than the oil viscosity. This phenomenon can be exploited in chemical flooding for heavy oil recovery to reduce the water mobility in water channels, which results in an improved sweep efficiency. Therefore, the oil can be displaced in the form of a W/O emulsion with minimal water fingering effects. Some oil drops also touched the pore walls, indicating that the pore wall became partially oil-wetted (Liu, 2006; Liu, Dong, Asghari, & Tu, 2007). The wettability alteration also aided in reducing water flow in the high-water-saturation regions. 7 The images in Fig. 3 also show that the water phase takes the central portion of the pore spaces occupied by residual oil during the formation of the W/O emulsion, instead of bypassing the oil ganglia by flowing through the existing water paths. This is how the process makes the trapped oil continuous and recoverable. Fig. 3(a)–(d) shows the process in which trapped oil ganglia 1–5 are merged with the displacement front through the W/O emulsification mechanism in a micromodel test. In Fig. 3(a), the oil bank is driven by the W/O emulsion and moves (from left to right) forward to approach the oil ganglia. In Fig. 3(b), the oil front is approaching Oil 1. After being merged with the W/O emulsion, Oil 1 is mobilized and driven forward to touch Oil 2 (Fig. 3(c)) and then Oils 3–5 (Fig. 3(d)). With this oil mobilization mechanism, the trapped oil ganglia become continuous and are displaced out of the micromodel. Fig. 4 shows the pressure drop curve for the above micromodel test of alkaline flooding. The pressure drop along the micromodel
grew from 0 to 0.8 pore volume (PV) during the oil saturation process. During waterflooding, the pressure drop continued to increase for the 0.2 PV injection and then decreased with the reset of the waterflood. When the pressure drop declined to a certain low value (at 2.4 PV), a chemical slug injection was initiated. As shown in Fig. 4, the pressure drop increased continuously up to 12.4 kPa due to the formation of a W/O emulsion, which had a higher viscosity than the oil. At 3.9 PV, a rapid decrease in the pressure drop was
Fig. 4. Pressure drop as a function of time in a micromodel test. Chemical slug: 0.40 wt% NaOH + 0.20 wt% Na2 CO3 .
302
M. Dong et al. / Particuology 10 (2012) 298–305
Fig. 5. Cumulative oil recovery as a function of fluid injected in a micromodel test. Chemical slug: 0.40 wt% NaOH + 0.20 wt% Na2 CO3 .
observed because the in situ formed W/O emulsion broke through the model outlet. After breakthrough, no further oil was recovered. Fig. 5 shows the oil recovery in the micromodel test plotted as a function of PV injection during the waterflooding and chemical injection. Waterflooding recovered approximately 28% of the IOIP, and the alkaline flooding recovered an additional 33% of the IOIP. These results show the improvement in sweep efficiency and oil recovery provided by the W/O emulsion flow.
Fig. 6 shows pore-level microphotographs taken during the above test and shows the oil and water distribution in different displacement stages. A water film between the oil and pore wall existed before the injection of the alkaline slug. After the alkaline flooding, an oil film existed between the water and the surface of the pores, indicating that the pore walls became preferentially oilwetted. The oil/water meniscus in Fig. 6(b) was convex to the oil phase, suggesting that the pore wall was oil-wetted. To investigate why the alkaline solution prefers to penetrate into the oil phase rather than flow through the water paths in the above test, additional micromodel tests were carried out by injecting a 0.40 wt% NaOH plus 0.20 wt% Na2 CO3 solution. This solution had a pH of 12.8. The experimental results showed that a W/O emulsion was formed by injecting NaOH rather than Na2 CO3 . NaOH is a strong base, and the interaction between NaOH in the water phase and organic acids in heavy oil is dramatic. It is hypothesized that mass transfer occurs more rapidly at the oil/water interface with a NaOH solution than with a Na2 CO3 solution. To initiate the penetration of the alkaline solution into the residual oil phase, a low interfacial tension between the oil and water is necessary. In the presence of a 0.40 wt% NaOH solution, the IFT of oil/water is lower than 0.1 dyn/cm. The lowered IFT is believed to be caused by the ionization of organic acids in the oil to form in situ surfactants as a result of the added alkalis. In the water-wetted glass network model, oil is in the central portion of the pores with thin water films between the oil and pore walls. A rapid reaction results in a very low instantaneous IFT between the injected alkaline solution and the residual oil, and the alkaline solution contacts the oil
Fig. 6. Pictures of one location in the micromodel at four stages in the oil displacement process. Chemical slug: 0.40 wt% NaOH + 0.20 wt% Na2 CO3 . (a) After water flooding, (b) after alkaline flooding, (c) 50 h after alkaline flooding, (d) 150 h after alkaline flooding.
M. Dong et al. / Particuology 10 (2012) 298–305
303
Fig. 7. Emulsification and entrainment of heavy oil in the water phase in a micromodel test. (a) Pore-level image, (b) image of the micromodel during oil displacement.
drops first in the central portion of the pores to form a W/O emulsion. During the process of emulsion movement, the pore walls also become partially oil-wetted because of the interaction of the heavy oil/alkaline solution interfaces with the glass walls. In sandpack flood tests (Liu, 2006; Ma, 2005), Na2 CO3 was also used together with NaOH. The calcium ion in brine can make the ionized in situ surfactants saponified and thus lose surface activity. The use of Na2 CO3 in the water phase can precipitate Ca2+ in brine, lowering the oil/brine IFT more efficiently. Therefore, the combination of NaOH and Na2 CO3 is favorable for lowering the IFT and enhancing oil recovery. In sandpack flood tests by Ma (2005) and Liu (2006), it was found that the O/W emulsion mechanism occurred for enhanced oil recovery when injecting alkaline/surfactant formulas. In micromodel tests, it was confirmed that the O/W emulsion mechanism dominated the displacement in EOR by injecting an alkaline/surfactant (A/S) formula (0.20 wt% NaOH + 0.40 wt% Na2 CO3 + 450 mg/L surfactant). Fig. 7 shows the emulsification and
entrainment of the heavy oil in the water phase with the injection of an A/S slug. Because of the synergy of the surfactant and alkali (Liu, 2006), the interfacial tension between the oil and water was lowered to 10−3 dyn/cm. The oil was emulsified and entrained in the water phase. The emulsion flow in the porous media also led to an increase in the pressure drop and an improvement in the oil recovery in sandpack flood tests. However, the increase in the pressure drop in this process was not as large as in the W/O emulsion process. Using water as the external phase of the O/W emulsion results in this type of emulsion having a similar viscosity as the water phase. 3.2. Comparison of the sweep efficiency of two displacement processes The two displacement processes resulted in different recovery efficiencies, which could be visualized in the micromodel tests. Fig. 8 shows pictures of the entire micromodel at different stages of
Fig. 8. Photos of the micromodel during an alkaline flood test showing the improvement of sweep efficiency. (a) After waterflood, (b) early stage of alkaline injection, (c) alkaline slug reached production end, (d) at 1.0 PV alkaline slug injection.
304
M. Dong et al. / Particuology 10 (2012) 298–305
Fig. 9. Photos of the micromodel during alkaline/surfactant injection, showing the formation of water channels during chemical injection. (a) After waterflood, (b) early stage of chemical injection, (c) chemical slug reached production end, (d) at 1.0 PV chemical slug injection.
alkaline flooding. Fig. 8(a) was taken after the initial waterflood and shows that some water channels (light colored) were created and some regions were not approached by the waterflood. The sweep efficiency by waterflooding was approximately 70% based on the area swept by the waterflood. When the alkaline solution was injected, as shown in Fig. 8(b), the water channels were eliminated by re-mobilized residual oil, and the area behind the displacement front became relatively uniform. It is the blocking of water channels that diverted the injected alkaline solution phase to the unswept region of the model to improve the sweep efficiency. Fig. 8(c) shows the micromodel when the alkaline slug front reached the outlet and shows a relatively uniform oil saturation distribution over the entire model. The alkaline solution reached almost every location inside the micromodel and improved the sweep efficiency to about 90%. Fig. 8(d) shows the model after approximately one pore volume of alkaline solution injection. From the distribution of the water phase (the color of the model), it is seen that the oil saturation was still uniform and greatly reduced at the end of the displacement when compared with the model in Fig. 8(c). Fig. 9 shows pictures of the entire micromodel at different stages of the A/S injection (0.20 wt% NaOH + 0.40 wt% Na2 CO3 + 450 mg/L surfactant). Fig. 9(a) shows the oil saturation distribution after waterflooding, and some water channels can be seen. In the early stage of the chemical injection, as shown in Fig. 9(b) and (c), the sweep efficiency was improved by the injection of the A/S solution. The water channels disappeared in the region behind the displacement front. By injecting an A/S slug, an O/W emulsion was formed in the pores, and the residual oil was mobilized by the O/W emulsion flow. After the injection of one PV of the chemical solution, as shown in Fig. 9(d), new water channels were created. In this situation, the oil production declined rapidly, and the injection of additional A/S solution had a poor sweep efficiency. Water channeling is a common problem during waterflooding in heavy oil reservoirs because of the high viscosity of heavy oils and the heterogeneity of oil reservoirs. To improve sweep efficiency
and enhance heavy oil recovery, water mobility in water channels and high permeability zones needs to be reduced significantly to divert the flow to oil-rich channels. The above experimental results showed that water channels could be more effectively blocked by injecting an alkaline slug rather than an A/S slug. 3.3. Application of the two displacement mechanisms The mechanisms of in situ W/O flows and of O/W emulsion flows in porous media are different and should be applied for different reservoir conditions for enhanced heavy oil recovery by chemical flooding. Numerous experimental results by Ma (2005) and Liu (2006) indicate that the viscosity of a W/O emulsion varies from 2000 to 20,000 mPa s, depending on the ratio of water/oil and on the alkaline concentration in the system. An O/W emulsion has a much lower viscosity than a W/O emulsion because it tends to have the viscosity of the external phase, i.e., the water phase. An O/W emulsion is produced using alkalis and hydrophilic surfactants. The synergy of alkalis and the hydrophilic surfactant can reduce the water and heavy oil IFT to an ultralow value. An O/W emulsion can be created with a minor disturbance at the oil and water interface. Numerous emulsification tests have shown that very gentle turning of a test tube that contains the above oil and water system can make O/W emulsions. However, the W/O emulsion observed in the micromodel displacements could not be easily formed in bottle tests, even when the oil and water phase were mixed by vigorous hand shaking. A possible reason is that the oil was so viscous that the oil and alkaline solution did not have good contact, whereas the alkaline solution and the oil in a porous medium can have pore-level contact. It was also observed that a W/O emulsion can be easily generated in sandpack flood tests by injecting an alkaline slug (Ma, 2005). Many sandpack flood tests have shown that the in situ-formed W/O emulsion can effectively block water channels, which leads to highly enhanced oil recovery. For some heavy oil recovery processes, the O/W emulsion mechanism can be applied
M. Dong et al. / Particuology 10 (2012) 298–305
to improve heavy oil recovery by dispersing oil in the water phase. In this case, residual oil is entrained in the water phase and flows to the production well. 4. Conclusions In alkaline flooding for enhanced heavy oil recovery, the alkaline solution can penetrate into heavy oil in porous media and form W/O emulsions. Because of the very high viscosity of the W/O emulsion, the resistance to water flow in high-water-saturation zones can be increased significantly to improve sweep efficiency and oil recovery. When an alkaline/surfactant formula is used to create an ultralow IFT and an O/W emulsion for a heavy oil, the oil can be produced by the O/W emulsion flow mechanism. Each of the two displacement processes can be designed to improve heavy oil recovery in accordance with the different displacement mechanisms. Acknowledgements The authors acknowledge the Petroleum Technology Research Center (PTRC) in Regina, Saskatchewan, Canada and the Natural Sciences and Engineering Research Council of Canada (NSERC) for their financial support of this work. The authors thank Dr. I. Chatzis of the University of Waterloo for providing the glass micromodel used in this study. References Arhuoma, M., Dong, M., Yang, D., & Idem, R. (2009). Determination of water-in-oil emulsion viscosity in porous media. Industrial & Engineering Chemistry Research, 48, 7092–7120. Chatterjee, J., & Wasan, D. T. (1998). A kinetic model for dynamic interfacial tension variation in an acidic oil/alkali/surfactant system. Chemical Engineering Science, 53(15), 2711–2725.
305
Chatzis, I., Morrow, N. R., & Lim, H. L. (1983). Magnitude and detailed structure of residual oil saturation. Society of Petroleum Engineers Journal, 23, 311–326. Cooke, C. E., Jr., William, R. E., & Kolodzie, P. A. (1974). Oil recovery by alkaline waterflooding. Journal of Petroleum Technology, 26, 1365–1374. Dong, M., Liu, P., Zhang, L., & Zhu, Y. (1986). An improved model for interfacial activity of acidic oil/caustic system for alkaline flooding. In SPE 14855 presented at the international meeting on petroleum engineering March 17–20, Beijing, China. Jennings, H. Y., Jr., Johnson, C. E., Jr., & McAuliffe, C. D. (1974). A caustic waterflooding process for heavy oils. Journal of Petroleum Technology, 26, 1344–1352. Johnson, C. E., Jr. (1976). Status of caustic and emulsion methods. Journal of Petroleum Technology, 28, 85–92. Liu, Q. (2006). Interfacial phenomena in enhanced heavy oil recovery by alkaline flood. Unpublished doctoral dissertation, University of Regina, Regina, Saskatchewan, Canada. Liu, Q., Dong, M., Asghari, K., & Tu, Y. (2007). Wettability alteration by magnesium ion binding in heavy oil/brine/chemical/sand systems—Analysis of electrostatic forces. Journal of Petroleum Science and Engineering, 59, 147–156. Liu, Q., Dong, M., Ma, S., & Tu, Y. (2007). Surfactant enhanced alkaline flooding for Western Canadian heavy oil recovery. Colloids and Surfaces A: Physicochemical and Engineering Aspects, 293, 63–71. Liu, Q., Dong, M., Yue, X., & Hou, J. (2006). Synergy of alkali and surfactant in emulsification of heavy oil in brine. Colloids and Surfaces A: Physicochemical and Engineering Aspects, 273, 219–228. Ma, S. (2005). Enhanced heavy oil recovery by dilute alkaline flooding. Master thesis, University of Regina, Regina, Saskatchewan, Canada. Ma, S., Dong, M., Li, Z., & Shirif, E. (2007). Evaluation of the effectiveness of chemical flooding using heterogeneous sandpack flood test. Journal of Petroleum Science and Engineering, 55, 294–300. McAuliffe, C. D. (1973a). Oil-in-water emulsions and their flow properties in porous media. Journal of Petroleum Technology, 25, 727–733. McAuliffe, C. D. (1973b). Crude-oil-in-water emulsions to improve fluid flow in an oil reservoir. Journal of Petroleum Technology, 25, 721–726. Wagner, O. R., & Leach, R. O. (1959). Improving oil displacement efficiency by wettability adjustment. Petroleum Transactions, AIME, 216, 65–72. Wang, J., Dong, M., & Arhuoma, M. (2010). Experimental and numerical study of improving heavy oil recovery by alkaline flooding in sandpacks. Journal of Canadian Petroleum Technology, 49(3), 51–57. Wasan, D. T., Shah, S. M., Chan, M., Sampath, K., & Shah, R. (1978). Observations on emulsion stability and interfacial properties in improved oil recovery by chemical flooding. American Chemical Society, 23(2), 705–727.