An evaluation on mechanisms of miscibility development in acid gas injection for volatile oil reservoirs

Produced gas containing the acid gas reinjection is one of the effective enhanced oil recovery methods, not only saving costs of disposing acid gases and zero discharge of greenhouse gases but also supporting reservoir pressure. The subsurface fluid from the Carboniferous carbonate reservoir in the southern margin of the Pre-Caspian basin in Central Asia has low density, low viscosity, high concentrations of H2S (15%) and CO2 (4%), high solution gas/oil ratio. The reservoir is lack of fresh water because of being far away onshore. Pilot test has already been implemented for the acid gas reinjection. Firstly, in our work a scheme of crude oil composition grouping with 15 compositions was presented on the basis of bottomhole sampling from DSTs of four wells. After matching PVT physical experiments including viscosity, density and gas/oil ratio and pressure–temperature (P–T) phase diagram by tuning critical properties of highly uncertain heavy components, the compositional model with phase behavior was built under meeting accuracy of phase fitting, which was used to evaluate mechanism of miscibility development in the acid gas injection process. Then using a cell-to-cell simulation method, vaporizing and/or condensing gas drive mechanisms were investigated for mixtures consisting of various proportions of CH4, CO2 and H2S in the gas injection process. Moreover, effects of gas compositions on miscible mechanisms have also been determined. With the aid of pressure-composition diagrams and pseudoternary diagrams generated from the Equation of State (EoS), pressures of First Contact Miscibility (FCM) and Multiple Contact Miscibility (MCM) for various gases mixing with the reservoir oil sample under reservoir temperature were calculated. Simulation results show that pressures of FCM are higher than those of MCM, and CO2 and H2S are able to reduce the miscible pressure. At the same time, H2S is stronger. As the CH4 content increases, both pressures of FCM and MCM are higher. But incremental values of MCM decrease. In addition, calculated envelopes of pseudoternary diagrams for mixtures of CH4, CO2 and H2S gases of varying composition with acid gas injection have features of bell shape, hourglass shape and triangle shape, which can be used to identify vaporizing and/or condensing gas drives. Finally, comparison of the real produced gas and the one deprived of its C3+ was performed to determine types of miscibility and calculate pressures of FCM and MCM. This study provides a theoretical guideline for selection of injection gas to improve miscibility and oil recovery.


Introduction
Gas injection is an important enhanced oil recovery process, in which interphase mass transfer during multiple contacting of the injected gas and the reservoir oil results in an efficient displacement. Reinjection into the reservoir of a gas produced from an oil field is commonly used. While the reasons are lack of transport pipelines and saving cost of disposing produced gas, much interest is being devoted to gas injection as an enhanced oil recovery process. Obviously, reinjection of the gas reduces the pressure drop associated with production of oil from a field. Moreover, immiscible/miscible gas injection can occur when the injected gas diffuses into the oil, and the oil will swell and oil viscosity reduces. If the interfacial tension is eliminated, miscible gas drive occurs. However, the injected gas also affects the oil/gas equilibrium compositions in the reservoir. Injection of various gases involving CO 2 and acid gas into the oil reservoir has been investigated by some authors. There is the miscible HC gas reinjection project with large volumes of sour gas (3-4% H 2 S and 10-15% CO 2 ) at Harweel field in Oman (Al-Hadhrami et al., 2007). The pilot project is a first contact miscible gas injection with 12% H 2 S content which has recently been running in the Tengiz oilfield in Kazakhstan (Wang et al., 2014;Urazgaliyeva et al., 2014). CO 2 injection has also made a good progress (Luo et al., 2013a, b).
Miscibility development has two different processes: First Contact Miscibility (FCM) and Multiple Contact Miscibility (MCM). If a single phase is formed when the original oil and the injected gas become fully miscible, the oil and gas phases exhibit FCM. If the original oil and the injected gas need multiple contacts to achieve miscibility, MCM will happen. The traditional interpretation of the MCM indicates that the gas may take up components from the oil phase corresponding to the vaporizing gas drive, the oil may take up components from the gas phase corresponding to the condensing gas drive. Later, a combined vaporizing/condensing mechanism was presented to describe MCM (Zick, 1986;Stalkup, 1987;Johns et al., 2002). Multiple contact experiments were conducted to study the effect of phase behavior on CO 2 flood (Gardner et al., 1981). Abrishami and Hatamian (1996) studied the thermodynamic behavior of hydrocarbon fluids in multiple contact processes with non-hydrocarbon (N 2 , CO 2 and their mixtures) through experiments. Ternary diagrams are commonly used to decide whether or not miscibility has been achieved. A cell-to-cell simulation program was presented by Metcalfe et al. (1973). Pederson et al. (1986) used a cell-to-cell model in connection with a ternary diagram to simulate a miscible drive. Nutakki et al. (1991) interpreted the mechanism of oil recovery as a condensing/vaporizing process with significant upper phase extraction on the basis of pseudoternary diagrams and pressure-composition diagrams generated from the Equation of State (EoS). Tang et al. (2005) evaluated the miscible ability and mechanism of gas displacement by simulating the PVT experiment data, pressurecomposition experiment, multiple contact miscible experiment and slim-tube experiment available. Johns et al. (2010) calculated Minimum Miscibility Pressure (MMP) for oils displaced by CO 2 contaminated by mixtures of N 2 , CH 4 , C 2 , C 3 , and H 2 S. Belhaj et al. (2013) evaluated miscible oil recovery utilizing N 2 and/or HC gases in CO 2 injection. It is clear that the major research focuses on CO 2 injection or CO 2 mixtures with other gases, but there are very limited studies on acid gas injection with high H 2 S content (Zhang et al., 2017;Luo et al., 2018).
In this paper, a simulation study was conducted to evaluate the performance of the acid gas miscible injection process. Pseudoternary diagrams are used to represent the mixing processes which take place in a petroleum reservoir into which gas is being injected. Mixtures consisting of various proportions of reservoir oil and injection gas (static mixtures of 0 mol%-90 mol% injection gas) were calculated to generate the pressure-composition diagram for the multicomponent system. Multiple-contact calculations were then performed at a pressure equal to the flooding pressure and the pseudoternary diagram was generated. The principal mechanisms of the oil recovery were then identified using the many-component pressure-composition diagram and pseudoternary diagram.
2 Phase behavior modeling and PVT matching

Phase behavior modeling
In order to access the acid gas flooding, the oil samples utilized are from the Carboniferous carbonate reservoir located in the southern part of the Pri-Caspian basin marginal zone. There are four wells (well X-1, well X-2, well X-3 and well X-6) that have been performed six DSTs to obtain representative bottomhole fluids samples. To make sure the good quality of the sampled fluids, the flowing bottomhole pressure during sampling is larger than the saturation pressure. The routine PVT data includes Constant Mass Expansion (CCE) and Differential Liberation (DL) for the reservoir fluids. The saturation point of the reservoir fluid is found to be 28.4 MPa at 98°C. Column 2 of Table 1 shows the composition of the reservoir oil sample. The acid gas component (H 2 S + CO 2 ) mole fraction in the oil sample is 19.23%.
The solution Gas-Oil Ratio (GOR) is high (640-670 m 3 /m 3 ). The produced fluid was separated by the three-stage separator (Fig. 1). The conditions of high pressure, middle pressure and low pressure separators are 9.6 MPa, 3.0 MPa, 8.5 MPa and 70°C, 60°C, 52°C respectively. The produced gas is exported directly to shore except reinjection gas and fuel gas. Because every separator has different pressure and temperature, GOR and H 2 S concentration of every separator is varied that leads to the compressor with various gas compositions. Column 3 of Table 1 shows the composition of the produced gas injected into the reservoir. To compare the effect of the gas composition, the produced gas is deprived C 3 + , named deprived C 3 + .

Laboratory PVT data matching
A detailed description of sample oils was obtained from experimental analysis. The oil was split into 14 components including 5 pseudocomponents for the heavy fraction. The Peng-Robinson Equation of State (EoS) was selected to predict thermodynamic properties. These component properties (critical pressure P c , critical temperature T c , and ancient factor x) of heavy components were adjusted to match experimental data which include CCE and DL. The regression procedure of Agarwal et al. (1990) was used in the data matching. The oil viscosity, oil density and GOR calculated with 14 components are reported in Figures 2, 3 and 4 respectively showing good agreement with these experimental data. Figure 5 shows the discrepancies between post-match and pre-match of P-T phase diagrams. It is obvious that the P-T phase diagram after matching is enlarged and the saturation pressure at the reservoir temperature of 98°C shows excellent agreement. A 14-component data set of thermodynamic properties was generated (Tab. 2). This is very useful in evaluating mechanisms of miscibility development of the acid gas injection.

Evaluation on mechanisms of vaporizing and condensing gas drive
Gas injection into a petroleum reservoir can be simulated using a cell-to-cell calculation presented by Metcalfe et al. (1973) and Pederson et al. (1986). This method simulates a number of cells of equal volumes in a series as shown in Figure 6. The temperature and the pressure are the same in each cell, and the volume is kept constant. Initially all the cells contain the same fluid compositions. A specified amount of gas is added to cell 1. A flash calculation is performed in each cell when mixing takes place and thermodynamic equilibrium is attained. After mixing of the injected gas and the cell fluid, the excess volume from cell 1 is transferred to cell 2. The excess volume from cell 2 is transferred to cell 3, etc. Pseudoternary diagrams are commonly used to decide whether or not miscibility has been achieved. Based on 14 pseudocomponents of the petroleum mixture, the composition of this mixture is represented using three groups of components, C 1 +N 2 , C 2 -C 5 +CO 2 +H 2 S and C 6 + . In order to use a pseudoternary diagram in connection with a cell-to-cell method, the following steps are used to calculate the MMPs and generate pseudoternary diagrams for MCM processes at a given temperature: 1. Choose a range of pressures. 2. Gas is added to oil at specified gas to oil molar ratio increments and flash calculations are performed until two-phase region is detected. 3. Using the first point in the two-phase region detected in step 2, the flashed liquid is mixed with the original gas at the specified gas to liquid ratio and the flash     (2019) calculation is performed. This process simulates a condensing gas drive process, and generates the portion of the phase envelope.
4. The procedure is repeated until the liquid composition is the same as the vapor composition and MMP is the pressure at which this occurs. 5. If it is not true, then the pressure is increased to a new value and the steps 2-4 are repeated. 6. The procedure is the same for a vaporizing drive process except the third step, where the flashed vapor is mixed with the original oil at the specified gas-oil mixing ratio and then flash calculation is performed.
In our work, the gases that have been hypothesized to be injected into the reservoir are gases which are originated from the separator gas produced from this reservoir itself, as it is available from surface facilities. To further study the effect of gas compositions, removing all C 3 + fraction of separator gas has been considered as a possible candidate. At the same time, to gain an understanding of the mechanisms of vaporizing and condensing gas drive, mixtures of CO 2 and H 2 S of varying composition with CH 4 injection gas are evaluated through simulation.   The pseudoternary diagrams for pure CH 4 , CO 2 and H 2 S injection were calculated using the cell-to-cell approach. These phase envelopes are quite different when MCM was achieved. A bell-shaped phase envelope was formed for pure CH 4 , a distorted hourglass shape for CO 2 and a triangle shape for H 2 S shown in Figure 7. MCM of pure CO 2 injection (19.9 MPa) is lower than that of pure CH 4 injection (32 MPa), and higher than that of pure H 2 S gas (7.8 MPa). The simulation results demonstrated that the sour gas containing CO 2 and H 2 S achieves miscibility easier than CH 4 by the vaporization of light ended hydrocarbons. The pressure-temperature phase diagrams for these three gases were presented in Figure 8. Of course, pure H 2 S injection in real oilfields is impractical because of its hypertoxicity. Simulation calculations of pure H 2 S injection are only for purpose of comparison and evaluation of miscible mechanisms.

Comparison of CH 4 /CO 2 and CH 4 /H 2 S solvent
From Figure 9, the saturation pressures of CH 4 /CO 2 solvent initially increase then decrease, while the saturation pressure curve of pure CO 2 injection keeps decreasing and the curve of pure CH 4 gas increase rapidly. Multiple contact calculations were then performed with 14 components for CH 4 /CO 2 solvents (CO 2 equals 0.75 mole, 0.5 mole and 0.25 mole). The corresponding pseudoternary diagrams are shown in Figure 10. MCM pressures of three mixtures gases are very close. But phase envelopes of pseudoternary diagrams between CO 2 0.5 mole and 0.25 mole are opposite directions. As CO 2 concentrations increase, MCM pressures would decrease slowly. By comparison, the FCM for all injected gas compositions was achieved at higher pressures and it appears that the variation of the injected gas composition had clear effects on FCM pressures.
The trend is very similar for pressure-composition diagrams of CH 4 /H 2 S solvent plotted in Figure 11. Calculated FCMs of CH 4 /H 2 S solvents consisting of various CH4 CO2 H2S Fig. 7. Calculated pseudoternary diagrams with pure CH 4 , CO 2 and H 2 S injection respectively.  proportions of CH 4 being 0, 0.25, 0.5 and 0.75 mole were 29.5 MPa, 29.5 MPa, 31.8 MPa and 53.7 MPa respectively. The shape of phase envelopes of pseudoternary diagrams for CH 4 /H 2 S are different from CH 4 /CO 2 , especially the CH 4 /H 2 S (H 2 S 0.75 mole) solvent appears triangle shape similar to pure H 2 S (Fig. 12).

CH 4 /CO 2 /H 2 S solvent
When the CH 4 concentration equals 0 mole, pressurecomposition diagrams for different mole fractions of CO 2 and H 2 S have a decreasing trend showed in Figure 13.
As the amount of CO 2 /H 2 S solvent injected increases, the saturation pressure is reducing. The increase of H 2 S concentrations in the injected gas decreases the MCM pressures for the reservoir oil by 0.1205 MPa/mol% (Fig. 14). FCM pressures of all mixtures are 29.5 MPa. When the CH 4 mole fraction raised to 0.25 mole, pressure-composition diagrams for different mole fractions of CO 2 and H 2 S change from CO 2 0.75 mole to 0 mole plotted in Figure 15. As CO 2 concentration increases, the decreasing amplitude becomes flattened then decreases rapidly. An increase in the amount of H 2 S in the injected gas would decrease MCM pressures as expected. FCM pressures would level off and attain a constant value of 29.5 MPa when H 2 S concentration is higher than a threshold value of 0.15 mole (Fig. 16). When the CH 4 concentration equals 0.5 mole, pressure-composition diagrams for different mole fractions of CO 2 and H 2 S vary noticeably. These shapes of pressure curves are typically increasing then decreasing and the maximum point appears (Fig. 17). Simulation results demonstrated that MCM pressures decrease relatively small, while FCM pressures decrease with H 2 S increase (Fig. 18). The trend for 0.5 mole of CH 4 is very similar to 0.75 mole of CH 4 . But the maximum points become larger (Fig. 19). MCM pressures of all mixtures equal 31.9 MPa and FCM pressures decrease with H 2 S increase (Fig. 20).
3.2 Comparison of produced gas and deprived of its C 3 + The main difference between produced gas and deprived of its C 3 + is the light component content, which the   E. Luo et al.: Oil & Gas Science and Technology -Rev. IFP Energies nouvelles 74, 59 (2019) concentrations of light components of the latter is higher than the former (Tab. 1). The pressure-composition diagram of a reservoir oil/injection gas system is a valuable tool to help identity the process mechanisms. Simulation results of pressure-composition diagrams for two types of injection gas are plotted in Figure 21. Plots indicate a      vaporizing process rather than a condensing process since the critical point is to the left of the cricondenbar. The process is first contact miscible at 47 MPa and 52 MPa for produced gas and deprived C 3 + gas respectively. It is also clear that the saturation pressure curve of the deprived C 3 + gas is higher than produced gas. The multiple contact pseudoternary diagrams for these two types of injection gas/reservoir oil systems were calculated at several pressures (Figs. 22 and 24). The left figure in Figure 22 for produced gas/reservoir oil system indicates a vaporizing process at 30 MPa because the envelope for this process is not closed and forms an hourglass shape. Therefore miscibility was not achieved. As pressure is increased, the right figure in Figure 22 shows the diagram at 32 MPa. In this case, the phase envelope closes indicating a multiple contact miscible with the oil and the process becomes essentially a vaporizing process. Miscibility is generated by backward contacts of the equilibrium liquid phase with fresh gas (Fig. 23). So the MCM is 32 MPa.
However, from Figure 24, pseudoternary diagrams for reservoir oil/deprived C 3 + gas system is obviously different from reservoir oil/produced gas system. The 20.25 MPa of MCM for deprived C 3 + gas is lower than produced gas. Due to the volatile nature of the sample oil, the full miscibility would achieve at the reservoir pressure of 77.7 MPa.

Conclusion
A 14-component compositional simulation model was built after matching laboratory PVT data. The pressures of FCM and MCM could decrease favorably as the CO 2 and/or H 2 S concentration increased in the injected stream. The envelope shape of pseudoternary diagrams relies on the composition of the injected gas. The following conclusions can be made from the previous results.
The mechanism of recovery can be identified using pressure-composition diagrams and pseudoternary diagrams generated with cubic EoS which have been tuned to match experimental data. Calculated pressures of FCM and MCM are feasible using the cell-to-cell method.
Different injection gases of CH 4 /CO 2 /H 2 S have characteristic shapes of the envelope of pseudoternary diagrams: bell shape, hourglass shape and triangle shape. This is very useful in identifying miscible mechanisms.
The volatile oil showed a nearly linear MCM pressures with mole fraction for each gas component mixed with CH 4 /CO 2 /H 2 S. CO 2 and H 2 S can reduce the pressure of FCM and MCM. As the CH 4 mole fraction increases, MCM pressures of mixtures would increase. FCM pressures are higher than MCM pressures. Moreover, as the CH 4 concentrations increase, the incremental FCM pressures are bigger.