Effects of salinity and slug size in miscible CO2 water

Journal of Industrial and Engineering Chemistry 52 (2017) 99–107

Contents lists available at ScienceDirect

Journal of Industrial and Engineering Chemistry journal homepage: www.elsevier.com/locate/jiec

Effects of salinity and slug size in miscible CO2 water-alternating-gas core flooding experiments Si Le Van, Bo Hyun Chon* Department of Energy Resources Engineering, Inha University, 100 Inharo, Nam-gu, Incheon 22212, Republic of Korea

A R T I C L E I N F O

Article history: Received 2 December 2016 Received in revised form 15 March 2017 Accepted 19 March 2017 Available online 27 March 2017 Keywords: CO2 flooding Salinity Slug size Water alternating gas Enhanced oil recovery

A B S T R A C T

This experimental study focuses on the variation of the salinity and the injected volume of individual WAG miscible flooding scheme for the improvement of oil recovery. In total, three slug sizes and five salinities were investigated in WAG flooding cycles and the recovered oil was measured accordingly. A response surface function was introduced for the purposes of estimating the available design and finally optimizing the variables. The experimental results indicated the unstable performances of oil recovery to the variations in the design parameters, confirmed the dependence of the oil production on a nonlinear function of the salinity and injected volume. © 2017 The Korean Society of Industrial and Engineering Chemistry. Published by Elsevier B.V. All rights reserved.

Introduction The division of an oil production project into three stages has been approved and is commonly carried out in most oil fields; these processes leave a large amount of oil in the reservoirs after the primary and secondary stages, especially when the crude oil is not the light-oil type. Thereby, the methods for recovering the remaining oil in the tertiary stage need to be prudently considered since they depend on a large number of factors such as the reservoir conditions, well operating conditions, or economic feasibility. Among highly technical methods, thermal conduction, chemical flooding, and gas injection are the most widely applied with the purpose of effectively mobilizing the trapped oil from the pores to the production well. Thermal methods are utilized most appropriately in heavy-oil reservoirs, whereas chemical flooding can be deployed in either light or heavy oil reservoirs [1,2]. Nevertheless, chemical adsorption, a slow response in the producer when injecting a viscous polymer solution, environmental concerns, and the uncertainty in the economic feasibility are restrictions for employing chemical flooding in enhanced oil recovery (EOR) projects [3]. The employment of CO2 in the oil fields has been demonstrated as an efficient measure to partly compensate for the decrease in the pressure in the reservoir during oil production and enhances the oil recovery as a consequence of reducing the residual oil saturation [4]. In terms

* Corresponding author. E-mail address: [email protected] (B.H. Chon).

of the environment, the use of CO2 from anthropogenic sources for injection underground, where CO2 can be stored permanently, will significantly reduce global greenhouse-gas effects; therefore, combined CO2-EOR promises a highly attractive profit from investment [5,6,7]. Indeed, the most recent practical report of the Farnsworth field for history performance and anticipation has demonstrated the substantial benefits of applying CO2-EOR in both enhancing oil production and CO2 storage, with the possibility of recovering over 30% original oil in place and sequestering more than 75% anthropogenic carbon underground [8,9]. Among the major uses of CO2 for enhancing oil recovery such as CO2 huff-npuff, continuous CO2 injection, and WAG, the use of multicycle injection for water and supercritical CO2 seems to be employed more popularly and efficiently owing to the ability to control the fluid mobility to the proper value [10,11]. During the WAG process in suitable reservoirs where the crude oil contains a sufficient amount of light hydrocarbon components, the occurrence of compositional exchange between the gas and the oil will result in swelling of the oil and a reduction in the oil viscosity; thus, the oil becomes moveable [12,13,14]. Further, the use of water after the injection of each designed volume of CO2 aims to mitigate the mobility of the displacing fluid, even though viscous fingering is unavoidable during the process. Depending on the reservoir pressure and crude-oil properties, the gas EOR method can be subcategorized as immiscible and miscible flooding; generally, the miscible process is more efficient because it results in greater improvements in both microscopic and macroscopic displacement than immiscible flooding [15,16,17]. Theoretically, miscibility is only achieved when the domain pressure is equal to or higher than

http://dx.doi.org/10.1016/j.jiec.2017.03.030 1226-086X/© 2017 The Korean Society of Industrial and Engineering Chemistry. Published by Elsevier B.V. All rights reserved.

100

S. Le Van, B.H. Chon / Journal of Industrial and Engineering Chemistry 52 (2017) 99–107

minimum miscibility pressure (MMP), or it can be defined as the point at which the practical maximum recovery efficiency is observed [18]. As in He et al., the MMP is the most important criterion in screening procedures for CO2-EOR for determining whether the reservoir pressure is adequate for developing miscible flooding [19]. Another concern of the WAG process is the loss of CO2 in the aqueous phase, or more precisely, the injected gas is partly dissolved in water [20]. Chang et al. have concluded that the solubility of CO2 in water is much higher than that of the hydrocarbon components and should not be neglected during investigation [21]. They also confirmed the dependence of the CO2 solubility in water on temperature, pressure, and salinity. Following their results, increases in these variables decrease the solubility of CO2 in the injected brine [22]. In the work of Tadesse et al., the injection of low-salinity water after seawater flooding significantly produced additional oil, and the following CO2 injection possibly further improves the recovery [23]. Furthermore, the mathematical studies of Jalal et al. argued that when carbonated water contacted the crude oil during the EOR processes, CO2 would migrate from the water into the oil phase, as it has higher solubility in hydrocarbons, thereby improving the oil mobility in the reservoir [24]. Regarding prediction methods, the gravity-enhanced process suggested by Liwei et al. has successfully estimated oil recovery and CO2 storage capacity by response surface models (RSMs) from scaling group development [25]. They also pointed out the predominant use of Latin hypercube sampling (LHS) for obtaining reliable estimates rather than the Box–Behnken method. In addition, the work of Feng et al. demonstrated the application of an RSM integrated with Monte–Carlo simulations for optimization and its utilization for an uncertainty analysis within an acceptable range [26]. Since this mathematical tool has been successfully derived for assessment and estimation, presumably it can be developed in a more complicated framework to assist the economic assessment in clearly understanding the impact of reservoir uncertainty in both oil recovery and effective carbon sequestration [27]. Even though the gas solubility has been investigated by numerical simulation in previous studies, there is still a lack of knowledge about the effects of the salinity on the other design parameters and oil-production performance. This work focuses on the impacts of the water salinity in WAG processes for the alternation of an injected slug size in experimental core-flood processes. The WAG ratio of 1:1 is fixed; thus, the water slug size is equal to that of the CO2 for all injection cycles initialized by water injection. A viscous oil sample is used for the experiment, and the pressure in the tests will be maintained at a higher value than the MMP after it has been determined by the slim-tube method, thereby maintaining the miscible condition during the tests. For the purposes of compensating for the lack of experimental data and optimization by computation, this work proposes the application of the response surface methodology to each of the specific available recorded data sets. Presumably, the combination of experimental testing and a mathematical tool will significantly decrease the time, costs, and materials while obtaining highly reliable results before application to real fields.

composition information, this study employs a slim-tube apparatus for simply measuring the MMP, which is also practically used in the oil industry. Fig. 1 shows a schematic of the Inha slim-tube apparatus (INST) used for the MMP measurement, which was constructed by winding a long steel tubing with a diameter of 0.457 cm and a length of 18.5 m in the form of a spring. Uniformsize glass beads or sand can be used as a filler inside the steel tubing. In the MMP measurement experiment, a 60–70 mesh filler was used, which is within the range obtained from literature research [28,29]. After filling the inside of the steel tubing with glass beads, the porosity and permeability were measured. The measured values are listed in Table 1. The oil sample used in the experiments is Van Gogh (VGH) crude oil with a viscosity of 355.5 cP at 71 C, as detailed in Table 2. Geographically, the Van Gogh oil field is located 53 km north of Exmouth, Western Australia. The reservoirs of this field virtually belong to Upper Triassic, Jurassic and Lower Cretaceous sandstones beneath the Lower Cretaceous seal. It is possible that the oil sample used in this work might fall out of range of the screening criteria of oil properties, where crude oils have gravity greater than 18 API and viscosity less than 10 cp (reservoir condition), to achieve miscible gas flooding, whereas the oil can be feasibly recovered under the immiscible scheme when its gravity is higher than 12 API and viscosity is lower than 600 cp [15]. However, since some practical heavy oil field testing has demonstrated the feasibility of achieving the partially miscible gas injection process [30], in particular there is always a big gap between laboratory tests and field applications in terms of controlling the flooding scheme. The utilized samples and procedures of carrying out experiments in this work can also result in well validated conclusions and reliable references for other CO2 injection projects. Before measuring the MMP, the inside of the steel tubing was cleaned and dried before the injection of oil to obtain accurate experimental results. The properties of the oil used in the MMP measurement experiment are listed in Table 2. After saturating the steel tubing with oil, CO2 was injected at the designated pressures. The injection pressures were 3.45, 6.89, 10.34, 13.79, 15.86, and 17.93 MPa, and the MMP was measured at 70 C, which is the same temperature as that used in the core experiment. The CO2 injection rate was maintained at a constant value of 0.2 ml/min while measuring the MMP. Although the measurement experiments using a slim tube have the disadvantage of requiring a relatively longer time compared to other methods, it is the simplest experiment for determining the phase behavior effects of CO2 and the oil. MMP measurement experiments were carried out to measure the pressure at the point where the oil recovery factor is 80% or

Experimental procedure Minimum miscibility pressure determination Currently, there are many possible methods for determining the MMP, including the use of a rising bubble apparatus, interfacial tension, or the slim tube and mathematical methods using empirical calculations or simulation programs based on the available components of oil samples. Because of the lack of oil

Fig. 1. Schematic of the Inha slim-tube apparatus (INST).

S. Le Van, B.H. Chon / Journal of Industrial and Engineering Chemistry 52 (2017) 99–107 Table 1 Range of specifications for the slim-tube apparatus from the literature compared with INST [28]. Criteria

Literature

INST

Inner diameter, cm Length, m Material Mesh Permeability, Darcy Porosity, % Displacement velocity, m/day

0.3048–1.6 1.52–36.6 Glass bead, sand 50–270 2.5–250 32–45 9.15–198.17

0.457 18.5 Sand 60–70 10.5 36 18.29

Table 2 Crude oil type used in the experiments. Crude oil

Gravity ( API)

Viscosity at 71 C Total sulfur (%, (cP) W/W)

Nickel (mg/ kg)

VGH (Van Gogh)

17.1

355.5

1.7

0.38

reaches 90–95% when CO2 is injected at 1.2 PV. In the present MMP measurement experiment, CO2 was injected to 1.2 PV, and the oil recovery factor at that point was measured to derive the MMP. When the experiment was completed, the slim-tube apparatus was cleaned using toluene and placed into an oven to dry it. The experimental processes including oil saturation and aging, CO2 injection, cleaning, and drying lasted for a total of three days to measure the MMP under each condition. Core-flood experiment To measure the oil recovery factor, an experimental system is configured as shown in Fig. 2. The core sample used in the forthcoming experiments is water-wet Berea sandstone whose general components are listed in Table 3. As the rock wettability

101

Table 3 Compositions of the Berea sandstone core sample used in the experiments. Component

Chemical formula

Content (%)

Silica Alumina Ferrous oxide Magnesium oxide Ferric oxide Calcium oxide

SiO2 Al2O3 FeO MgO Fe2O3 CaO

93.13 3.86 0.54 0.25 0.11 0.1

alteration has been determined in several studies [23,31], this issue is not taken into consideration or verified in this work. The core flooding experiment was conducted in four separate stages: water injection, oil injection, water injection, and CO2 injection. Before injecting oil, the core was 100% saturated with water (water with soluble salt) for 24 h under vacuum conditions using a saturation apparatus. After that, oil injection was carried out until the water cut reached 1% or less to simulate the initial saturation conditions of the reservoir. When oil injection completed, the sample was aged for approximately 30 h at a constant pressure and temperature to restore the wettability of the core sample to the same conditions as in the reservoir. Thereafter, 5 wt% water was injected at a rate of 0.5 ml/min up to a pore volume (PV) of 2.0 to determine the residual oil saturation. CO2 injection was conducted at a WAG ratio of 1:1 and salinities of 1, 2, 3, 4, and 5 wt% while changing the CO2 slug size to 0.2, 0.4, and 0.6 PV. The CO2 injection rate into the core was maintained at 0.2 ml/min. In order to obtain the miscible flooding condition, the confining pressure was maintained at a higher value than the MMP, which was measured from the slim-tube test. Results and discussion The oil recovery factors according to the CO2 injection pressures in the MMP measurement are shown in Fig. 3. From the figure, the

Fig. 2. Schematic of the experimental system for WAG core flooding.

102


partial molar volume is an important parameter for correcting Henry’s constant when the gas pressure is high and can be used to estimate the density of the solution [36,37]. The empirical correlation between this parameter for CO2 and the temperature was established as follows [36]: vi ¼ 37:51 90585 102 T þ 8:740 104 T 2 5:044 107 T 3 ðT : CÞ

ð3Þ

The above correlations are only applied when the concentration of the salt dissolved in water is zero. When salts are dissolved in an aqueous solution, Henry’s constant for the gas is corrected with the salting-out coefficient ks: ! salt kH ð4Þ ln ¼ ks bsalt kH Fig. 3. MMP determination using the slim-tube method.

where bsalt (mol/kg H2O) is the molality of the dissolved salt, and ks is computed using the correlation by Bakker [38]:

oil recovery factor is observed to have different dependencies on the pressure in two different sections. The section with large increases in the oil recovery factor according to the pressure is the non-mixing section, and the section with small increases in the oil recovery factor according to the pressure for an oil recovery factor greater than 88% is the mixing section. In this case, the MMP can be regarded to be 12.62 MPa, which is the point where the linear trend lines drawn to fit the data in the two sections using the least-squares method intersect each other. According to the result of a slim-tube test, miscibility could obviously be achieved for an operating confining pressure of at least 12.62 MPa. In fact, core-flood experiments have been carried out at 15.17 MPa; therefore, the observed data absolutely reflect the results of WAG miscible flooding processes in which the injected CO2 satisfies the supercritical state (the critical point at 10 MPa and 40 C) [32]. Furthermore, since viscous dead oil was used in this study, it is argued that the oil has been displaced via multi-contact miscibility (MCM) processes in the core-flood tests instead of firstcontact miscibility (FCM). Theoretically, the increase in the salinity of the injected fluid or aquifer reduces the solubility of CO2 in the aqueous phase, as presented in many previous studies. However, since there have been insufficient quantitative investigations of the salinity with regard to the EOR performance, the full effects of the salt concentration are still in question, especially when the design of artificial brine for injection might contribute to the success of enhancing oil recovery [33]. The correlation for Henry’s constants in terms of the gas solubility in water was developed over a large temperature range by Harvey as follows [34]: i

lnkH ¼ lnPs1 þ

A Bð1 T Þ þ T T

0:355

þ

Ce1T

0:41

ðT Þ

ð1Þ

where kHi is Henry’s constant at the considered temperature and water saturation pressure, T* = T/Tc (Tc is the critical temperature of water and equal to 647.14 K) is the reduced temperature, and P1s (MPa) is the saturation pressure of water at the temperature T (K). The empirical parameters A,B, and C for CO2 are 9.4234, 4.0087, and 10.3199, respectively. The dependence of Henry’s constant on the pressure is expressed as [35]: i

lnkH ¼ lnkH þ

1 RT

ZP vi dP

ð2Þ

Ps1

where kH is Henry’s constant at the pressure P (MPa), and vi (cm3/ mol) is the partial molar volume of CO2 in the aqueous phase. The

ks ðCO2 Þ ¼ 0:11572 0:00060293T þ 3:5817 106 T 2 3:7772 109 T 3 ðT : CÞ

ð5Þ

According to the correlations, Henry’s constant will obviously increase as the salinity increases; as a consequence, the gas solubility in the solution decreases. The schematic applicability of Henry’s concept is briefly presented in Fig. 4 in which the investigated salinities and pressure used for experiments in this work are involved. Practically, low salinity water flooding has been proven as an effective EOR method in either technical or economic aspects since the change from high to low salinity brine causes an alteration of rock wettability from mixed-to-oil wet or oil wet to water wet system that helps to extract more oil out of the porous media; hence, the flooding efficiency strongly depends on brine composition, crude oil composition, and mineralogy of the rock system [39,40]. In CO2 flooding, using low salinity water basically improves the oil recovery by elevating the solubility of CO2 in water that significantly reduces interfacial tension of oil and brine; however, as oil swelling and oil viscosity reduction are also a function in EOR efficiency by CO2 injection, the suitable levels of water salinity in a WAG process are not really stable among different contexts. In total, 15 core-flood experiments were conducted with three slug sizes and five salinities after finishing 2.0 PV water flooding as aforementioned. In other words, the experiments in terms of changing injection slug sizes and salinities were commenced at residual oil saturation; hence, the recovery factor of each experiment will be computed as the percentage of oil produced in that remaining oil volume in order to make the comparisons more understandable. Besides residual oil saturation (Sor), porosity and absolute permeability of the core were also recorded contemporarily, whose results are clearly listed in Table 4. As presented, Sor varies slightly among experiments, with the maximal deviation of 3.24%, whereas the margin errors in porosity and permeability are 1.48% and 12.9 mD, respectively. Technically, these experimental error thresholds are absolutely acceptable to proceed further with the analysis. For a fixed WAG ratio of 1:1, the slug sizes of water and CO2 are equal and constant for all cycles for each slug-size case. The results for core flooding are presented in Fig. 5. In the present experiments, to obtain results following sufficient CO2 injection, the experimental results were compared at the assumed final injection volume of 2.5 PV in terms of oil recovery factor. Oil started to be recovered after approximately 0.2 PV of injection. The starting point of recovery is affected by the length of the core and the injection rates. The time for the additional recovery of oil is longer


103

Table 4 Porosity, permeability of the testing core and residual oil saturation measurement results of 15 water flooding experiments. Slug size (PV)

Salinity (%)

Porosity (%)

Permeability (mD)

Residual oil saturation after water flooding (Sor, %)

0.2

1 2 3 4 5 1 2 3 4 5 1 2 3 4 5

18.01 17.80 17.32 17.96 18.45 17.78 17.89 17.81 17.37 18.36 16.97 18.22 17.86 17.79 17.87

151.9 146.7 152.8 156.0 143.1 146.7 143.1 154.5 149.4 143.2 143.5 151.1 143.8 152.5 150.4

42.38 42.21 43.86 43.13 43.00 41.87 41.45 40.69 42.89 41.55 42.57 41.32 42.78 44.24 43.93

0.4

0.6

at actual sites because the scales of the oil recovery systems are larger than the system in the present experiments. In addition, the oil recovery is closely related to the injection rates. The artificial water which was injected should push the mixed fluid at a uniform speed, and if the injection rate increased, the sweep efficiency might decrease, leading to a decrease in the oil recovery rate. Therefore, when the method is applied to actual sites, the injection rate should be considered so that it can be optimized for the reservoir conditions through reservoir simulations.

Fig. 4. Henry’s constant versus temperature: (a) at the water saturation pressure and testing pressure (0% salinity) and (b) at the testing pressure for different salinities.

As observed in Fig. 5, the ultimate oil recovery factors in case of 0.2 PV slug size are proportional with the salinity of the injected fluid within its verified constraint, whereas for higher slug sizes, the highest produced oil volume is achieved at a salinity of 4% and decreases as salt concentration is beyond this point. In detail, a maximal oil recovery of 47.07% is obtained corresponding with 5% salinity and slug size of 0.2 PV, while the peaks of oil volume recovered for 0.4 PV and 0.6 PV slug sizes are 52.33% and 49.91%, respectively. From Henry’s theory, it is implied that a reduction of CO2 solubility in water caused by increasing salinity helps the crude oil to be contacted and displaced with more supercritical CO2, as a result more oil can be recovered till the limit of salt concentration. These results evidently demonstrate the significant influences of conjugate salinity and injected slug size on the WAG core-flood experiments, in other words they should be prudently involved when considering other design parameters. It should also be realized that the limitation of enhancing oil recovery following the rises of injection slug size and salinity, as the design of 0.4 PV and 4% is most dominant among 14 other experiments. In addition, the results absolutely affirm the existence of an optimal water salinity in a WAG process of which the lower or higher points just leave more oil in the pore volumes. Since the experiments have been conducted under reservoir conditions, the results can be presumably extended to verify at higher scales with consideration to other parameters, that might be involved in future works. First, it is argued that the optimal design of the slug size and salinity for achieving the highest oil recovery factor might fall within the ranges of both parameters. Second, there is not a sufficient amount of available measured data to understand the behavior of the flooding processes after a specific injected volume corresponding with the changes in the design parameters. Finally, since the compositional information of the crude oil is lacking, the use of a simulator for core-flood history matching simulation is not applicable. Therefore, the association between a set of recorded data with a mathematical tool needs to be considered for the purpose of clarifying the questions from the tests. The response surface methodology is proposed in this study to determine the relationship between the design variables and the objective function in order to predict the responses of the unmeasured data and optimize the variables following a target. Since the oil recovery factor is the main parameter for evaluating the WAG processes, it is assumed to be the response function, whereas the CO2 slug size of each cycle and the salinity are considered as the design variables. The response surface can be described as follows: RF ¼ A0 þ A1 X þ A2 Y þ A11 X 2 þ A22 Y 2 þ A12 XY

ð6Þ

104


Fig. 5. Experimental results of the core-flood test for slug sizes of (a) 0.2 PV, (b) 0.4 PV, and (c) 0.6 PV.

where RF is the oil recovery factor, and X and Y are the designed CO2 slug size (PV) and designed salinity (wt%), respectively. The coefficients An (n = 1, 2, 11, 22, 12) are determined from the measured data of RF, X, and Y, and the validation of these values depends on the quality of the measured data and the factorial design. In the basic experiment results, an LHS design is proposed for computation by interchanging the pair of variables from 0.2, 0.4, and 0.6 PV with 1, 4, and 5 wt% (4 wt% is selected for design as the oil recovery is peaked at this point) to form nine design samples. The coefficients are computed by matrix transformation method with a step size of 0.2 PV of the total injected volume. The results are presented in Table 5.

The calculated coefficients are validated by comparing the recomputed response surface with the initial design response function values. As shown in Fig. 6, the matching curves— representing the recomputed values—fit the design data well. In addition, the estimated results and available data, which have not been used in the design samples, also exhibit good agreement, as evidenced by the high value of R-square (R2 > 0.95), indicating that the mathematical models are fully validated and acceptable. By using the response coefficients, it is completely possible to investigate the changes in the core-flood process performance after the injection of any specific fluid volume according to changes in the CO2 slug size and the salinity of the aqueous phase.

Table 5 Computed coefficients by the RSM. Injected volume (PV)

A0

A1

A2

A11

A22

A12

0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2 2.2 2.4 2.5

– 0.981 0.030 12.566 25.270 24.588 22.468 21.609 20.390 18.308 15.353 18.382 20.383 21.001

–

– 0.150 1.019 0.559 0.007 0.419 2.683 2.340 5.101 6.417 7.430 6.324 6.690 6.766

–

–

3.072 19.495 1.527 15.379 11.575 24.093 45.434 49.131 63.317 99.201 91.936 80.903 77.849

0.000 0.073 0.003 0.027 0.010 0.236 0.171 0.655 0.820 1.014 0.763 0.802 0.818

– 0.256 0.497 0.033 1.352 0.968 0.055 0.116 0.221 0.783 0.179 0.067 0.430 0.588

8.018 23.584 2.805 17.918 9.165 21.029 39.903 45.375 61.763 88.974 81.898 71.073 68.101


105

Fig. 6. Estimated results for the measured data: (a) slug size = 0.2 PV, (b) slug size = 0.4 PV, (c) slug size = 0.6 PV, (d) salinity = 2 wt%, and (e) salinity = 3 wt%.

As presented in Fig. 7, the behaviors of the oil recovery factor are nearly linear with the changes in the variables within 1 PV of the injected volume, e.g., increases in both the salinity and slug size enhance the recovery factor with a maximum of approximately 31% of the OIP. However, when the cycles are continuously carried out, the enhancement in RF reaches the peak for certain values that fall within the constraints of either the slug size or salinity. In detail, for 1.5 PV, a peak appears around a salinity of 3–4 wt% and a slug size of 0.4–0.5 PV. Compared to the core-flood experimental conclusion in which a salt concentration of 4 wt% and a slug size of 0.4 PV result in the highest oil recovery, these results from the RSM method are extremely reliable. The next target is to determine the optimal values of the salt concentration and slug size that result in

the highest oil recovery. Presumably, the most predominant design might not give the highest recovery factor throughout the flooding process in comparison with others. Indeed, the optimal result in Fig. 8 evidently shows that the most outstanding curve only achieves the highest oil recovery after an injected volume of 2 PV, affirming the accuracy level of the optimization estimation. Furthermore, the optimal values of the variables achieved from both the experiment and estimation are very close (Table 6), indicating the good quality of the measured data, which physically and quantitatively reflect the WAG flooding processes. Undoubtedly, the application of an RSM for matching and estimation for the purpose of supporting the experimental data can be convincingly considered to be a relevant tool since it might shorten the time to

106


Fig. 7. Oil recovery performance for changes in the salinity and CO2 slug size after the injection of a specific fluid volume: (a) 1 PV, (b) 1.5 PV, and (c) 2.5 PV.

Table 6 Optimal values from the experiment and RSM methods.

Experiment Estimated by RSM Optimal by RSM

RF (%)

X (PV)

Y (wt%)

52.33 50.71 51.0

0.4

4

0.455

4.313

Fig. 8. Optimization results obtained by the RSM.


carry out more experiments, e.g., the requirement of only nine samples in this work, and provide greater understanding of the physical aspects of flooding processes. Conclusions A total of 15 WAG core-flood experiments have been conducted after multiple tests for determining the MMP using the slim-tube method. The variation in the salinity of the aqueous phase from 1 wt% to 5 wt% and the CO2 slug size for each cycle at a fixed WAG ratio of 1:1 were investigated in order to understand their associated impacts on oil recovery. The experimental results clearly indicate that the salinity of the injected water still has a positive influence within a threshold on recovery enhancement for a suitably designed slug size of the solvent. Since the results might differ among experimental conditions, or between experiments and practical field scale, a larger scale of investigating these conclusions is strongly proposed in future works. The response surface methodology has been applied as a mathematical method on re-computing the core-flood processes using the available measured data by constructing response surfaces. The method has demonstrated good agreement between the response models and the recorded data and was evaluated via the quality of estimation, which properly reflects the impacts of an individual variable after any specific injected volume and optimizes the physical variables with the aim of obtaining the highest recovery factor. Even though low-salinity water is most commonly used in real WAG processes, the results of this work propose that the appropriate salt concentration for the flooding schemes be prudently designed to achieve the highest EOR performance instead of neglect it. Further, the RSM method utilized in this study evidently cannot substitute for a core-flood history matching simulation; however, the use of this tool for estimating unmeasurable data with a small number of design samples from available measurements becomes much more attractive in terms of lowering the time and expenses while still providing highly reliable consequences. Acknowledgement This work was supported by Inha University Research Grant. References [1] S. LeVan, B.H. Chon, J. Ind. Eng. Chem. 38 (2016) 200. [2] J. Binshan, W. Yu-Shu, Q. Jishun, Oil Gas Sci. Technol. Rev. IFP 70 (2015) 951.

107

[3] A.J. Hadi, D. Morteza, C. Zhangxin, J. Pet. Sci. Eng. 142 (2016) 85. [4] E. Amin, W.L. Larry, L.B. Steven, Int. J. Greenh. Gas Control 25 (2014) 79. [5] R.D. James, C.M. Charles, F. Scott, L. Yaghoob, D.W. Nathan, G. Nathan, Y. Fang, B. Peter, Energy Procedia 63 (2014) 7694. [6] Z. Xiaoliang, R. Zhenhua, L. Xinwei, J. Nat. Gas Sci. Eng. 29 (2016) 275. [7] C. Chalbaud, M. Robin, J.M. Lombard, H. Bertin, P. Egermann, Oil Gas Sci. Technol. Rev. IFP 65 (2010) 541. [8] W. Ampomah, R.S. Balch, R.B. Grigg, The SPE Improved Oil Recovery Conference, Tulsa, Oklahoma USA 11–13 April, 2016. [9] W. Ampomah, R. Balch, M. Cather, D. Rose-Coss, Z. Dai, J. Heath, T. Dewers, P. Mozley, Energy Fuels 30 (2016) 8545. [10] H. Longyu, G. Yongan, Energy Fuel 28 (2014) 6811. [11] E.J. Koval, SPE J. 3 (1963) 145. [12] A. Yaser, E.E. Seyed, B. Peyman, H. Mahdi, K. Yousef, V. Atena, J. Pet. Sci. Eng. 133 (2015) 123. [13] F. Huang, H. Huang, Y. Wang, J. Ren, L. Zhang, B. Hassan, S. Ren, G. Chen, J. Pet. Sci. Eng. 145 (2016) 328. [14] P. Shen, X. Chen, J. Qin, Pet. Explor. Dev. 37 (2010) 211. [15] A.A. Ahmad, B. Baojun, J. Pet. Sci. Eng. 79 (2011) 10. [16] M. Yahya, Al-Wahaibi, Energy Fuel 24 (2010) 1813. [17] S. Ji-Quan, X. Ziqiu, D. Sevket, Energy Procedia 1 (2009) 3205. [18] O. Alomair, A. Malallah, A. Elsharkawy, M. Iqbal, Oil Gas Sci. Technol. Rev. IFP 70 (2015) 967. [19] L.P. He, P.P. Shen, X.W. Liao, Q.C. Gao, C.S. Wang, F. Li, J. Pet. Sci. Eng. 134 (2015) 199. [20] A.C.J. Curtis, R.T. Michael, SPE J. 24 (1984) 197. [21] Y.B. Chang, B.K. Coasts, J.S. Nolen, SPE Reserv. Eval. Eng. 1 (1998) 155. [22] W.T. Tadesse, A. Najeeb, K. Hossein, Y. Xiaolong, M.G. Ramona, The SPE Western North American and Rocky Mountain Joint Regional Meeting, Denver, Colorado USA 16–18 April, 2014. [23] W.T. Tadesse, A. Waleed, M.G. Ramana, K. Hossein, M.A. Ali, J. Pet. Sci. Eng. 142 (2016) 101. [24] F. Jalal, J. Mahmoud, S. Mehran, Fuel 174 (2016) 325. [25] L. Liwei, K. Saeid, T.J. Russell, M.D. Robert, Int. J. Greenh. Gas Control 42 (2015) 502. [26] P. Feng, J.M. Brian, D. Zhenxue, J. Wei, L. Si-Yong, A. William, V. Hari, E. Rich, Int. J. Greenh. Gas Control 51 (2016) 18. [27] Z. Dai, H. Viswanathan, J.F. Rahn, R. Middleton, F. Pan, W. Jia, S.Y. Lee, B. McPherson, W. Ampomah, R. Grigg, Energy Procedia 63 (2014) 7685. [28] J.Y. Oh, B.H. Chon, H.S. Yu, J. Korean Soc. Geosyst. Eng. 49 (2012) 521. [29] H.S. Yu, M.S. Thesis, Inha University, Korea, 2013. [30] C. Bardon, L. Denoyelle, in: E. Okandan (Ed.), Co2 Injection to Enhance Heavy Oil Recovery, Martinus Nijhoff, The Hague, 1984, pp. 177–209. [31] G. Thyne, The SPE Western Regional Meeting, Anchorage, Alaska USA 23– 26 May, 2016. [32] S. Ji-Quan, X. Ziqiu, D. Seveket, Int. J. Greenh. Gas Control 5 (2011) 75. [33] W. Ning, L. Xiaochun, T.D. Robert, L.D. Casie, Int. J. Greenh. Gas Control 37 (2015) 170. [34] A.H. Harvey, AIChE J. 42 (1996) 1491. [35] Computer Modelling Group Ltd, GEM User Guide, Computer Modelling Group Ltd (CMG), 2014 Calgary, Alberta Canada. [36] J.E. Garcia, Density of Aqueous Solutions of CO2, Lawrence Berkeley National Laboratory, 2011, 2017. [37] D.Y. Peng, D.B. Robinson, Ind. Eng. Chem. Fundam. 15 (1976) 59. [38] R.J. Bakker, Chem. Geol. 194 (2003) 3. [39] P. Gamage, G. Thyne, The International Symposium of the Society of Core Analysts, Austin, Texas, USA 18–21 September, 2011. [40] P. Vledder, J.C. Fonseca, T. Wells, I. Gonzalez, D. Ligthelm, The 2010 SPE Improved Oil Recovery Symposium, Tulsa, Oklahoma, USA 24–28 April, 2010.