experiment and correlations for co 2 –oil minimum miscibility pressure...

7
Experiment and correlations for CO 2 oil minimum miscibility pressure in pure and impure CO 2 streamsQiaoyan Shang, Shuqian Xia, * Meiqing Shen * and Peisheng Ma Miscible gas ooding is one of the most eective methods for enhanced oil recovery (EOR). A key and important parameter in designing an ecient miscible gas ooding is the minimum miscibility pressure (MMP). It is very essential to determine and predict gasoil MMP for EOR technology. In this work, both the experiment and correlation approaches for obtaining gasoil MMP were studied. Experimentally, a set of apparatus with the vanishing interfacial tension (VIT) technique was set up and utilized to determine the gasoil MMP for three dierent kinds of injection gas. Theoretically, a new model was proposed to predict the gasoil MMP. A total of 156 experimental MMP data were achieved from this work and literature was used to train and test the model. The average absolute relative deviations (AARD%) of the training set for pure CO 2 , CO 2 mole fraction less than 0.5 (x CO 2 < 0.5) and CO 2 mole fraction larger than 0.5 (x CO 2 > 0.5) in the injection gas are 5.85%, 4.06% and 6.49%, respectively. The AARD% of the testing set are 2.08%, 2.97% and 5.26%, respectively. The model is applied to calculate the gasoil MMP in dierent CO 2 purity ooding with CO 2 mole fraction in the injection gas from 0.0352 to 1. 1. Introduction Enhanced oil recovery (EOR) is playing an increasingly important role in the petroleum industry. 1 Miscible gas ooding is one of the most ecient methods for EOR. The general injection gases include ue gases, a mixture of hydrocarbon gases (or enriched gases), nitrogen, methane, and carbon dioxide (CO 2 ), in which CO 2 ooding is the most promising method for EOR technology. CO 2 ooding has the advantages of low cost, a high displacement eciency 2 and reducing the greenhouse eect by sequestrating the emitted CO 2 into the depleted petroleum reservoir. 3 It is not always available for injecting pure CO 2 , however, impure CO 2 can be obtained through many sources, including natural reservoirs, process plant waste streams 4 and the wells undergoing CO 2 ooding reservoir. The gas without purication was reinjected from the wells undergoing CO 2 ooding reservoir can reduce the cost of CO 2 injection. 2 A key and important parameter in designing an ecient miscible gas ooding is the minimum miscibility pressure (MMP), the pressure at which interface tension between injected uid and crude oil in place becomes zero and the local displacement eciency can be theoretically approached 100%. 5 In recent years, there are many researches for the MMP, including experimental methods, theoretical calcula- tion and empirical correlation. Among the existing experi- mental approaches, the slim-tube test, 6 rising bubble apparatus (RBA) 7 and the vanishing interfacial tension (VIT) technique 810 are the most widely used methods for deter- mining the gasoil MMP. Theoretical models based on accu- rate properties of crude oil and reservoir temperature by using an equation of state have been proposed in literatures. 1113 Many correlations of predicting the gasoil MMP have been presented. Yellig and Metcalfe 6 proposed a correlation of CO 2 oil MMP only with temperature as the parameter, and this correlation was used when bubble point pressure exceeds the predicted MMP. Alston et al. 14 correlated MMP with reservoir temperature, characterization of live oil, and weight- averaged critical temperature of the injected mixture gas. Sebastian et al. 4 presented a correlation of gasoil MMP as a function of the pure CO 2 oil MMP and mole-averaged critical temperature of the injected mixture gas. GlasØ 15 correlated pure CO 2 oil MMP as a function of C 7+ molecular weight of the dead oil. Emera and Sarma 16 predicted pure and impure CO 2 oil MMP by re-optimizing the coecients of Alston's 14 and GlasØʹs 15 correlation based on an expanded CO 2 oil MMP database. Shokir 17 used alternating conditional expectation (ACE) algorithm to correlate gasoil MMP. Li et al. 3 proposed a new pure CO 2 oil MMP correlation with Key Laboratory for Green Chemical Technology of the State Education Ministry, Collaborative Innovation Center of Chemical Science and Engineering (Tianjin), School of Chemical Engineering and Technology, Tianjin University, Tianjin 300072, China. E-mail: [email protected]; [email protected] Electronic supplementary information (ESI) available. See DOI: 10.1039/c4ra11471j Cite this: RSC Adv. , 2014, 4, 63824 Received 29th September 2014 Accepted 18th November 2014 DOI: 10.1039/c4ra11471j www.rsc.org/advances 63824 | RSC Adv. , 2014, 4, 6382463830 This journal is © The Royal Society of Chemistry 2014 RSC Advances PAPER Published on 19 November 2014. Downloaded by Northern Illinois University on 25/11/2014 15:32:20. View Article Online View Journal | View Issue

Upload: peisheng

Post on 30-Mar-2017

212 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: Experiment and correlations for CO               2               –oil minimum miscibility pressure in pure and impure CO               2               streams

RSC Advances

PAPER

Publ

ishe

d on

19

Nov

embe

r 20

14. D

ownl

oade

d by

Nor

ther

n Il

linoi

s U

nive

rsity

on

25/1

1/20

14 1

5:32

:20.

View Article OnlineView Journal | View Issue

Experiment and

Key Laboratory for Green Chemical Techn

Collaborative Innovation Center of Chem

School of Chemical Engineering and Techno

China. E-mail: [email protected]; mqs

† Electronic supplementary informa10.1039/c4ra11471j

Cite this: RSC Adv., 2014, 4, 63824

Received 29th September 2014Accepted 18th November 2014

DOI: 10.1039/c4ra11471j

www.rsc.org/advances

63824 | RSC Adv., 2014, 4, 63824–638

correlations for CO2–oil minimummiscibility pressure in pure and impure CO2

streams†

Qiaoyan Shang, Shuqian Xia,* Meiqing Shen* and Peisheng Ma

Miscible gas flooding is one of the most effective methods for enhanced oil recovery (EOR). A key and

important parameter in designing an efficient miscible gas flooding is the minimum miscibility pressure

(MMP). It is very essential to determine and predict gas–oil MMP for EOR technology. In this work,

both the experiment and correlation approaches for obtaining gas–oil MMP were studied.

Experimentally, a set of apparatus with the vanishing interfacial tension (VIT) technique was set up and

utilized to determine the gas–oil MMP for three different kinds of injection gas. Theoretically, a new

model was proposed to predict the gas–oil MMP. A total of 156 experimental MMP data were achieved

from this work and literature was used to train and test the model. The average absolute relative

deviations (AARD%) of the training set for pure CO2, CO2 mole fraction less than 0.5 (xCO2< 0.5) and

CO2 mole fraction larger than 0.5 (xCO2> 0.5) in the injection gas are 5.85%, 4.06% and 6.49%,

respectively. The AARD% of the testing set are 2.08%, 2.97% and 5.26%, respectively. The model is

applied to calculate the gas–oil MMP in different CO2 purity flooding with CO2 mole fraction in the

injection gas from 0.0352 to 1.

1. Introduction

Enhanced oil recovery (EOR) is playing an increasinglyimportant role in the petroleum industry.1 Miscible gasooding is one of the most efficient methods for EOR. Thegeneral injection gases include ue gases, a mixture ofhydrocarbon gases (or enriched gases), nitrogen, methane,and carbon dioxide (CO2), in which CO2 ooding is the mostpromising method for EOR technology. CO2 ooding has theadvantages of low cost, a high displacement efficiency2 andreducing the greenhouse effect by sequestrating the emittedCO2 into the depleted petroleum reservoir.3 It is not alwaysavailable for injecting pure CO2, however, impure CO2 can beobtained through many sources, including natural reservoirs,process plant waste streams4 and the wells undergoing CO2

ooding reservoir. The gas without purication was reinjectedfrom the wells undergoing CO2 ooding reservoir can reducethe cost of CO2 injection.2

A key and important parameter in designing an efficientmiscible gas ooding is the minimum miscibility pressure(MMP), the pressure at which interface tension between

ology of the State Education Ministry,

ical Science and Engineering (Tianjin),

logy, Tianjin University, Tianjin 300072,

[email protected]

tion (ESI) available. See DOI:

30

injected uid and crude oil in place becomes zero and thelocal displacement efficiency can be theoretically approached100%.5 In recent years, there are many researches for theMMP, including experimental methods, theoretical calcula-tion and empirical correlation. Among the existing experi-mental approaches, the slim-tube test,6 rising bubbleapparatus (RBA)7 and the vanishing interfacial tension (VIT)technique8–10 are the most widely used methods for deter-mining the gas–oil MMP. Theoretical models based on accu-rate properties of crude oil and reservoir temperature by usingan equation of state have been proposed in literatures.11–13

Many correlations of predicting the gas–oil MMP have beenpresented. Yellig and Metcalfe6 proposed a correlation ofCO2–oil MMP only with temperature as the parameter, andthis correlation was used when bubble point pressure exceedsthe predicted MMP. Alston et al.14 correlated MMP withreservoir temperature, characterization of live oil, and weight-averaged critical temperature of the injected mixture gas.Sebastian et al.4 presented a correlation of gas–oil MMP as afunction of the pure CO2–oil MMP and mole-averagedcritical temperature of the injected mixture gas. GlasØ15

correlated pure CO2–oil MMP as a function of C7+ molecularweight of the dead oil. Emera and Sarma16 predicted pure andimpure CO2–oil MMP by re-optimizing the coefficients ofAlston's14 and GlasØʹs15 correlation based on an expandedCO2–oil MMP database. Shokir17 used alternating conditionalexpectation (ACE) algorithm to correlate gas–oil MMP. Liet al.3 proposed a new pure CO2–oil MMP correlation with

This journal is © The Royal Society of Chemistry 2014

Page 2: Experiment and correlations for CO               2               –oil minimum miscibility pressure in pure and impure CO               2               streams

Paper RSC Advances

Publ

ishe

d on

19

Nov

embe

r 20

14. D

ownl

oade

d by

Nor

ther

n Il

linoi

s U

nive

rsity

on

25/1

1/20

14 1

5:32

:20.

View Article Online

reservoir temperature and oil characterization. All the abovecorrelations were used only with injected CO2 mole percent aslow as 30%.

The previous empirical correlations have their own limits toeach specic eld and the range of CO2 mole percent in injec-tion gas, although, they are very useful in the prescreeningcandidate reservoirs for CO2 injection. Therefore, it is impor-tant to propose experimental methods and accurate, generalcorrelations for determining the gas–oil MMP.

In this paper, a set of apparatus with the vanishing interfa-cial tension (VIT) technique was designed and manufacturedrstly. It was proved reliable and utilized to determine gas–oilMMP for three different kinds of injection gas, which includespure CO2, CO2 mole fraction less than 0.5(xCO2

< 0.5) and CO2

mole fraction larger than 0.5(xCO2> 0.5) in injection gas. A

general gas–oil MMP model is proposed on the basis of gas–oilMMP database, which includes CO2 mole fraction in the injec-tion gas from 0.0352 to 1. The model relates gas–oil MMP toreservoir temperature, C7+ molecular weight, C7+ mole percent,mole fraction ratio of volatile components (N2 and CH4) tointermediate components (CO2, H2S, and C2–C6) and the char-acterization of injection gas. The new model is able to accu-rately predict MMP for various purity of CO2, and for both liveand dead oils. Generally, dead oils don't have volatile compo-nents, that is the ratio of volatile components to intermediatecomponents is zero. Conversely, live oils have volatilecomponents.

2. Experimental section2.1. Materials

In this study, the simulated dead oil was reconstituted by ninehydrocarbon solvents according to Bohai Oileld in China. Thepurities and suppliers of these nine solvents and two gases usedin this study are listed in Table 1. The characterization of threesimulated oil used in this work and nine crude oil samplesfrom references for testing the new model are shown in Table 2.Table 3 shows the composition of injection gas used in thiswork and references for testing the new model. It can be seenfrom Table 2 that the twelve oil samples include live and deadoils. From Table 3, we can see that the mole fraction of CO2 ininjected mixture gas as low as 0.0352.

Table 1 Purities and suppliers of nine solvents and two gases used in th

Chemical Supplier

n-Pentane Tianjin Guangfu Technolon-Hexane Tianjin Jiangtian Chemican-Heptane Tianjin Jiangtian Chemican-Octane Tianjin Jiangtian Chemican-Nonane Tianjin Guangfu Technolon-Decane Tianjin Guangfu Technolon-Undecane Tianjin Guangfu Technolon-Tridecane Tianjin Guangfu Technolon-Heptadecane Tianjin Heowns BiochemiCarbon dioxide Tianjin Liufang IndustrialNitrogen Tianjin Liufang Industrial

This journal is © The Royal Society of Chemistry 2014

2.2. Interfacial tension measurement

Fig. 1 shows schematic diagram of the experimental apparatusfor measuring the equilibrium interfacial tension (IFT) betweenthe crude oil and injected gas by pendent drop method. Theexperimental setup was designed and manufactured by ourresearch group. A high-pressure view cell (HPVC), syringe pump(SP) and programmable syringe pump (PSP) are the majorcomponents of this apparatus. The cylindrical HPVC is made ofstainless steel with two sapphire windows at both ends of thehole for visual observation. The SP is put in semiconductor coldtrap and used to produce liquid CO2. A stainless steel syringeneedle which is connected to PSP is installed at the top of HPVCand use to produce a pendent drop. Light source (LS) (AFT-BL50W, microvision, Beijing), microscope (AFT-ZL0911,microvision, Beijing) and camera (M&C) (MV-VS078FM, micro-vision, Beijing) are used to capture serial digital images of thedynamic pendant drop oil. The HPVC was placed horizontallybetween the LS and the M&C. The digital image of pendent dropoil at different times is obtained by M&C and automaticallystored in personal computer (PC).

Firstly, liquid CO2 is injected to the HPVC by SP at a pre-specied pressure and room temperature, and then the pres-sure of CO2 increases with the rise of temperature in HPVC. Thecrude oil is injected from PSP to HPVC to form a pendent dropat the tip of the syringe needle aer pressure and temperaturein HPVC were invariant. The dynamic pendant drop areobtained and automatically stored in the computer when acompletely-shaped pendant drop was produced. A drop shapeanalysis program developed by our group was used to determineequilibrium IFT of pendent drop. The method also requires thedensities of the two phases at the equilibrium temperature andpressure. The oil density changes resulting from CO2 dissolu-tion into oil phase have been experimented in ref. 18. Theexperimental results show that the variations of the oil densitybefore and aer CO2 dissolution in atmospheric pressure and12 MPa were almost zero. This behaviour is a result of thebalance between the density increases due to pressure increasesand the reduction in density with increasing CO2 solubility. Thevariations of the oil density with the CO2 dissolution arenegligible in ref. 19–21 when determinating IFT by vanishinginterfacial tension (VIT) technique. In this work, we used the

is study

Purity

gy Development Co., Ltd., China 0.990l Technology Co., Ltd., China 0.990l Technology Co., Ltd., China 0.985l Technology Co., Ltd., China 0.990gy Development Co., Ltd., China 0.990gy Development Co., Ltd., China 0.980gy Development Co., Ltd., China 0.990gy Development Co., Ltd., China 0.985cal Technology Co., Ltd., China 0.990gases Co., Ltd., China 0.999gases Co., Ltd., China 0.999

RSC Adv., 2014, 4, 63824–63830 | 63825

Page 3: Experiment and correlations for CO               2               –oil minimum miscibility pressure in pure and impure CO               2               streams

Table 2 The characterization of three simulated oils used in this work and nine crude oil samples from references for testing the new model

Component/mol/%

Sample

1 2 3 4 5 6 7 8 9 10 11 12

N2 0.00 0.00 0.00 0.60 0.60 0.34 0.01 0.39 0.01 0.01 0.39 0.39CO2 0.00 0.00 0.00 0.58 0.58 0.25 6.66 1.41 6.66 6.66 1.41 1.41C1 0.00 0.00 0.00 9.90 9.90 4.07 32.98 6.35 32.98 32.98 6.35 6.35C2 0.00 0.00 0.00 2.72 2.72 3.11 23.16 7.43 23.16 23.16 7.43 7.43C3 0.00 0.00 0.00 4.98 4.98 4.88 8.39 7.13 8.39 8.39 7.13 7.13C4 0.00 0.00 0.00 4.91 4.91 3.58 2.54 3.73 2.54 2.54 3.73 3.73iC4 0.00 0.00 0.00 1.09 1.09 2.09 1.69 0.89 1.69 1.69 0.89 0.89C5 3.10 1.26 1.52 3.73 3.73 3.07 1.50 3.50 1.50 1.50 3.50 3.50iC5 0.00 0.00 0.00 2.16 2.16 2.45 1.60 0.74 1.60 1.60 0.74 0.74C6 2.71 1.41 2.98 4.85 4.85 5.51 1.88 2.67 1.88 1.88 2.67 2.67C7+ 94.19 97.33 95.50 64.48 64.48 70.67 19.59 65.76 19.59 19.59 65.76 65.76Total 100 100 100 100 100 100 100 100 100 100 100 100MWC5+

183.56 204.14 184.10 187.79 188.65 207.86 187.92 261.66 187.92 187.92 261.66 261.66MWC7+

190.03 207.56 188.94 206.00 207.00 227.94 216.00 281.00 216.00 216.00 281.00 281.00References This work This work This work 30 30 31 32 32 32 32 32 32

Table 3 The composition of injection gas used in this work and references for testing the new model

Component/mol/%

Injected gas

1 2 3 4 5 6 7

CO2 3.52 20 75.00 77.96 75 75 100H2S 25.00 0 0.00 0.00 0 0 0C2 13.65 0 6.00 10.36 25 0 0C3 2.40 0 16.25 6.66 0 0 0C4 0.60 0 2.75 3.33 0 0 0iC4 0.00 0 0.00 0.00 0 0 0C5 0.15 0 0.00 1.33 0 0 0iC5 0.00 0 0.00 0.00 0 0 0C6 0.00 0 0.00 0.34 0 0 0C7+ 0.00 0 0.00 0.03 0 0 0N2 0.30 80 0.00 0.00 0 25 0C1 54.38 0 0.00 0.00 0 0 0References 32 This work 32 32 32 This work 30 and 31, This work

Fig. 1 Schematic diagram of the experiment apparatus. SP,syringe pump; PSP, programmable syringe pump; V, valve;TC, temperature controller; PT, pressure transducer; HPVC,high-pressure view cell; LS, light source; SW, sapphire window;M & C, microscope & camera; PC, personal computer; RD, rupturedisks.

63826 | RSC Adv., 2014, 4, 63824–63830

RSC Advances Paper

Publ

ishe

d on

19

Nov

embe

r 20

14. D

ownl

oade

d by

Nor

ther

n Il

linoi

s U

nive

rsity

on

25/1

1/20

14 1

5:32

:20.

View Article Online

same method as these references. Liquid densities withdifferent temperature are calculated by National Institute ofStandards and Technology22 and the densities of injection gasare obtained by BWRS equation of state.23 The IFT measureprocedure is repeated for at least three times to insure satis-factory repeatability at each equilibrium pressure and temper-ature. Linear least-square lines are tted to each set ofexperimental data and extrapolate to zero of gas–oil IFT todetermine the gas–oil MMP.9

In this study, Firstly, the isotherm IFT data of the (n-decane–CO2) system were measured to check the reliability of theapparatus and the experimental method. Fig. 2 shows acomparison between our experimental IFT data of the (n-decane–CO2) system with the published data.24,25 It can be seenthat the experimental data are consistent well with the pub-lished data. The VIT experimental apparatus was provedreliable.

A total of 9 IFT tests for three simulated oil–pure/impure CO2

systems were conducted under different test conditions

This journal is © The Royal Society of Chemistry 2014

Page 4: Experiment and correlations for CO               2               –oil minimum miscibility pressure in pure and impure CO               2               streams

Fig. 2 Interfacial tension measurements of the (n-decane–CO2)system as a function of pressure for isotherm. ( ) T ¼ 70.0 �C,compared to literature measurements: Georgiadis et al.24 ( ), at 70.40�C; Nagarajan et al.25 ( ), at 71.15 �C.

Table 4 The experimental MMP of three simulated oils with differentinjected gas

Oil sample Injected gas T/�C MMP/MPa

1 Pure CO2 60.0 11.242 Pure CO2 80.0 15.133 Pure CO2 80.0 14.293 CO2 75%, N2 25% 60.0 17.133 CO2 75%, N2 25% 70.0 18.393 CO2 75%, N2 25% 80.0 19.993 CO2 20%, N2 80% 60.0 35.423 CO2 20%, N2 80% 70.0 36.523 CO2 20%, N2 80% 80.0 37.20

Paper RSC Advances

Publ

ishe

d on

19

Nov

embe

r 20

14. D

ownl

oade

d by

Nor

ther

n Il

linoi

s U

nive

rsity

on

25/1

1/20

14 1

5:32

:20.

View Article Online

(temperature and injection gas), which are summarized inTable 4. The constant test temperature were chosen to be t ¼60.0, 70.0, and 80.0 �C according to the Bohai Oileld reservoirtemperature. Table 4 also lists the measured MMPs for thesenine IFT tests.

3. Mathematical model3.1. Factors inuencing the CO2–oil MMP

The gas–oil MMP is proved to be mainly inuenced by reservoirtemperature, oil characterization and injection gas composi-tion.4,14,16,26 Holm and Josendal27 reported that the MMP isstrongly related to reservoir temperature and oil characteriza-tion. In addition, non-CO2 component (e.g. CH4, N2, H2S orintermediate hydrocarbons components) presented in injectiongas has a great inuence on MMP. Rathmell et al.28 proposedthat the MMP was increased as a result of volatile componentsin the crude oil, such as CH4, and decreased due to the presenceof intermediates (C2 to C6). Alston et al.14 demonstrated that theCO2–oil MMP was increased with the rise of the ratio betweenvolatiles to intermediates of crude oil. Generally, the presence of

This journal is © The Royal Society of Chemistry 2014

H2S of injected gas decreases the gas–oil MMP, whereas thepresence of N2 considerably increases the gas–oil MMP.29 Liet al.3 stated that C7+ molecular weight used in correlation wasbetter than C5+ molecular weight.

From the above, we can see that the gas–oil MMP values wereinuenced by many factors, such as reservoir temperature, C7+

molecular weight, C7+ mole percent, volatile (CH4 and N2) tointermediate ratio (C2–C6, H2S and CO2), and injection gascomposition, which should be considered in the gas–oil MMPscorrelations. And some other references were also correlated allthe factors into the models, such as Shokir.17 In this work, wechanged the expressions with a number of parameters andimproved the results.

3.2. Developing pure and impure CO2–oil MMP model

As discussed above, the gas–oil MMPs were inuenced by manyfactors. In this work, it is necessary to take these inuencingfactors into account for gas–oil MMP equations. Eqn (1)–(3) areutilized to calculate MMP for pure CO2, xCO2

< 0.5 and xCO2> 0.5

ooding, respectively.

MMP ¼ exp

��TR

b þ g��ln�MWC7þ

��c

þ h

��exp

�xVOL

xINT

dþ i

�xC7þ

�f� (1)

MMP ¼ expnaTR

bEcf xCO2 þDmixN2 jxCH4

�xC7þ

�d� gxH2S � hxC2�C7

o (2)

MMP ¼ �aTR

b þ EcDf ðexpðxCO2ÞÞg��xC7þ

�dðhxH2S ixC2�C7 þ jÞþ Tm

R (3)

where

E ¼ (MWC7+)xVOL/xINT (4)

D ¼ exp(xN2)exp(xCH4

)/exp(xCO2) (5)

where, TR is the reservoir temperature in �C, MWC7+is C7+

molecular weight, xVOL is the mole percent of volatile compo-nents, xINT is the mole percent of intermediate components, xC7+

is C7+ mole percent, xN2, xCH4

, xH2S, xC2�C7is N2, CH4, H2S, and

C2–C7 mole fraction in injection gas, respectively. a, b, c, d, f, g,h, i, j and m are empirical coefficients.

The objective function OF is expressed as follows:

OF ¼ minX �

yexp � ycal�2

(6)

where yexp and ycal are the experimental and calculated MMPs,respectively. The regressed coefficients of the three equationsare obtained by quasi-Newton methods.

A total of 156 data points includes 9 data points in this studyand 147 data points from literatures were used for the correla-tions. The gas–oil MMP database are presented in the ESI.† Thedatabase included live and dead original oils and differentkinds of injection gases. They are divided into three sets

RSC Adv., 2014, 4, 63824–63830 | 63827

Page 5: Experiment and correlations for CO               2               –oil minimum miscibility pressure in pure and impure CO               2               streams

Table 5 The values of coefficients of three equations

Eqn (1) Eqn (2) Eqn (3)

Coefficient Value Coefficient Value Coefficient Value

a a 1.2739 � 100 a �2.6775 � 103

b 5.3257 � 10�2 b 1.1336 � 10�1 b �1.4801 � 100

c 1.4897 � 100 c 2.3208 � 10�2 c 2.0538 � 10�1

d 2.1651 � 10�3 d 7.7368 � 10�2 d 4.6481 � 10�1

f �2.2589 � 100 f 7.1126 � 10�1 f 5.6514 � 100

g �1.0387 � 100 g 2.0563 � 10�1 g 7.3656 � 100

h 2.2378 � 105 h 1.5048 � 10�2 h 1.1771 � 100

i �1.0006 � 100 i 5.0657 � 100 i 4.8639 � 10�1

j 4.3459 � 100 j �9.6491 � 10�1

m �5.7089 � 10�1 m 6.6667 � 10�1

RSC Advances Paper

Publ

ishe

d on

19

Nov

embe

r 20

14. D

ownl

oade

d by

Nor

ther

n Il

linoi

s U

nive

rsity

on

25/1

1/20

14 1

5:32

:20.

View Article Online

according to CO2 mole fraction in injection gas. The three setsare pure CO2, xCO2

< 0.5 and xCO2> 0.5, respectively. Every set was

split to two parts for training and testing process. The trainingset was used to regress the coefficients in the model to predictgas–oil MMP. The regressed coefficients of three equations arelisted in Table 5.

The square of correlation coefficient (R2) of eqn (1)–(3) are0.9681, 0.9761, 0.9575, respectively.

Fig. 3 Comparison of the calculated MMPs with the experimentalMMP for the pure CO2 (a), xCO2

< 0.5 (b), xCO2> 0.5 (c).

4. Results and discussion

A set of new correlations were developed in order to preciselypredict gas–oil MMP by quasi-Newton methods. The correla-tions present MMP as a function of reservoir temperature, theproperties of crude oil and injection gas. The gas–oil MMPpredicted by this model were compared with the experimentaldata. The accuracy of this model was compared with othercorrelations from literatures. The experimental and the calcu-lated MMP values from this work and literatures are shown inthe ESI.†

Initially, in order to test the proposed gas–oil MMP model,the predicted data were compared with the experimental data.The scatter plot of experimental CO2–oil MMP versus the pre-dicted ones by the eqn (1) is shown in Fig. 3(a). Table 6 showsthe detail calculated results of the newmodel, including averageabsolute relative deviation (AARD%) of 5.85%, and maximumabsolute relative deviation (MARD%) of 23.02%. Fig. 3(a) andTable 6 make clear that eqn (1) can estimate pure CO2–oil MMPdata accurately. In the descending order exactness Li et al.,3

Shokir,17 Emera and Sarma,16 Alston et al.,14 GlasØ15 correlationscame in sequence order, respectively.

Secondly, Table 6 shows the comparison between this modeland other common correlations of mixture gas–oil MMP inliteratures.4,14,16,17 From Table 6, we can see that the AARD% ofeqn (2) and (3) are 4.06%, 6.49%, respectively, and lower thanother correlations in literatures. The scatter plots of experi-mental mixture gas–oil MMP versus the predictions by eqn (2)and (3) are shown in Fig. 3(b) and (c), respectively. The calcu-lated gas–oil MMP data by eqn (2) can match with the experi-mental data well as shown in Fig. 3(b). The predicted gas–oilMMP by eqn (3) are distributed near the diagonal in Fig. 3(c).

63828 | RSC Adv., 2014, 4, 63824–63830

Shokir17 correlation is a closer match to eqn (3) for xCO2> 0.5

ooding from Table 6. In all, it is apparent that this new modelcan precisely predict mixture gas–oil MMP especially for xCO2

<0.5 ooding.

Ultimately, to test and conrm the reliability of the newmodel, 18 experimental MMP data from this work and liter-atures,30–32 which were set as testing database and not used inregressing correlations, were obtained by the three equations.The calculated results were shown in Table 7. Table 7 showsthat the AARD% of the three equations for pure CO2, xCO2

< 0.5and xCO2

> 0.5 in injection gas are 2.08%, 2.97% and 5.26%,respectively, and lower than other gas–oil MMP correlationsfrom literatures3,4,14–17 The results affirmed that the newmodel can predict the experimental gas–oil MMP accurately

This journal is © The Royal Society of Chemistry 2014

Page 6: Experiment and correlations for CO               2               –oil minimum miscibility pressure in pure and impure CO               2               streams

Table 6 Comparison between the performances of this model and common models in pure and impure CO2–oil MMP predictiona

References

Parameter

xCO2¼ 1 xCO2

< 0.5 xCO2> 0.5

AARD/% MARD/% AARD/% MARD/% AARD/% MARD/%

Li et al.3 9.91 29.67 — — — —Shokir17 61.41 1076.23 26.84 139.16 19.11 65.65Emera and Sarma16 25.75 92.59 15.96 24.03 25.03 67.49GlasØ15 26.79 207.56 — — — —Alston et al.14 40.61 181.83 48.53 94.54 28.37 80.83Sebastian et al.4 — — 104.31 370.17 19.2 60.36This work 5.85 23.02 4.06 11.15 6.49 19.16

a ‘—’ the correlations were not suitable for xCO2in the injection gas.

Table 7 Test result comparison between the performances of this model and common models in pure and impure CO2–oil MMP predictiona

References

Parameter

xCO2¼ 1 xCO2

< 0.5 xCO2> 0.5

AARD/% MARD/% AARD/% MARD/% AARD/% MARD/%

Li et al.3 5.87 14.30 — — — —Shokir17 3.75 5.46 51.53 67.29 21.81 35.39Emera and Sarma16 9.31 22.17 12.96 17.31 26.49 81.19GlasØ15 46.29 95.84 — — — —Alston et al.14 9.25 24.20 67.55 79.43 30.82 95.62Sebastian et al.4 — — 189.73 299.16 52.23 86.86This work 2.08 6.31 2.97 4.40 5.26 8.62

a ‘—’ the correlations were not suitable for xCO2in the injection gas.

Paper RSC Advances

Publ

ishe

d on

19

Nov

embe

r 20

14. D

ownl

oade

d by

Nor

ther

n Il

linoi

s U

nive

rsity

on

25/1

1/20

14 1

5:32

:20.

View Article Online

with CO2 mole fraction from 0.0352 to 1 and for both live anddead oils.

5. Conclusion

In this work, a set of VIT apparatus was designed and manu-factured by our research group. It was proved reliable and ninegas–oil MMP data were obtained from the VIT apparatus underdifferent test conditions (temperature and injection gas). Inaddition, three equations correlating MMP as a function ofreservoir temperature, the properties of crude oil and injectiongas were proposed for predicting gas–oil MMP with pure CO2,xCO2

< 0.5 and xCO2> 0.5 in injection gas, respectively. A total of

156 experimental MMP data come from this work and litera-tures were used to train and test models. The new gas–oil MMPcorrelations give the precise prediction with lower AARD% thanother pure and impure CO2–oil MMP correlations. They areapplied to predict gas–oil MMP for pure and impure CO2

ooding with CO2 mole fraction from 0.0352 to 1 and for bothlive and dead oils.

Acknowledgements

This research was supported by National Natural ScienceFoundation of China, no. U1162104.

This journal is © The Royal Society of Chemistry 2014

References

1 M. Cao and Y. Gu, Fuel, 2013, 109, 157.2 A. Shokrollahi, M. Arabloo, F. Gharagheizi andA. H. Mohammadi, Fuel, 2013, 112, 375.

3 H. Li, J. Qin and D. Yang, Ind. Eng. Chem. Res., 2012, 51, 3516.4 H. Sebastian, R. Wenger and T. Renner, J. Pet. Technol., 1985,37, 2076.

5 F. S. Kovarik, SPE Production Technology Symposium,Lubbock, Texas, 1985.

6 W. Yellig and R. Metcalfe, J. Pet. Technol., 1980, 32, 160.7 M. Dong, S. Huang, S. B. Dyer and F. M. Mourits, J. Pet. Sci.Eng., 2001, 31, 13.

8 D. N. Rao, Fluid Phase Equilib., 1997, 139, 311.9 D. N. Rao and J. I. Lee, J. Colloid Interface Sci., 2003, 262, 474.10 M. Nobakht, S. Moghadam and Y. Gu, Ind. Eng. Chem. Res.,

2008, 47, 8918.11 J.-N. Jaubert, L. Avaullee and C. Pierre, Ind. Eng. Chem. Res.,

2001, 41, 303.12 Y. Wang and F. M. Orr Jr, Fluid Phase Equilib., 1997, 139, 101.13 J.-N. Jaubert, L. Wolff, E. Neau and L. Avaullee, Ind. Eng.

Chem. Res., 1998, 37, 4854.14 R. Alston, G. Kokolis and C. James, SPE J., 1985, 25, 268.15 O. Glass, SPE J., 1985, 25, 927.16 M. K. Emera and H. K. Sarma, J. Pet. Sci. Eng., 2005, 46, 37.

RSC Adv., 2014, 4, 63824–63830 | 63829

Page 7: Experiment and correlations for CO               2               –oil minimum miscibility pressure in pure and impure CO               2               streams

RSC Advances Paper

Publ

ishe

d on

19

Nov

embe

r 20

14. D

ownl

oade

d by

Nor

ther

n Il

linoi

s U

nive

rsity

on

25/1

1/20

14 1

5:32

:20.

View Article Online

17 E. M. E.-M. Shokir, J. Pet. Sci. Eng., 2007, 58, 173.18 S. Sayegh, D. Rao, S. Kokal and J. Najman, J. Can. Pet.

Technol., 1990, 29, 30.19 D. Yang, P. Tontiwachwuthikul and Y. Gu, J. Chem. Eng.

Data, 2005, 50, 1242.20 D. Yang and Y. Gu, J. Pet. Sci. Eng., 2005, 23, 1099.21 M. Nobakht, S. Moghadam and Y. Gu, Energy Fuels, 2007, 21,

3469.22 M. Huber, NIST thermophysical properties of hydrocarbon

mixtures database.23 Y. WU and B. CHEN, Oil & Gas Storage and Transportation,

2003, 22, 16.24 A. Georgiadis, F. Llovell, A. Bismarck, F. J. Blas, A. Galindo,

G. C. Maitland, J. P. M. Trusler and G. Jackson, J. Supercrit.Fluids, 2010, 55, 743.

63830 | RSC Adv., 2014, 4, 63824–63830

25 N. Nagarajan and R. L. Robinson, J. Chem. Eng. Data, 1986,31, 168.

26 H. Yuan, R. Johns, A. Egwuenu and B. Dindoruk, SPEReservoir Eval. Eng., 2005, 5, 418.

27 L. Holm and V. Josendal, J. Pet. Technol., 1974, 26, 1427.28 J. Rathmell, F. Stalkup and R. Hassinger, Fall Meeting of the

Society of Petroleum Engineers of AIME, New Orleans, LA, USA,1971.

29 L. W. Lake, Enhanced oil recovery, Prentice-Hall, New Jersey,1989, p. 262.

30 R. S. Metcalfe, SPE J., 1982, 22, 219.31 D. Graue and E. Zana, J. Pet. Technol., 1981, 33, 1312.32 B. Eakin and F. Mitch, SPE Annual Technical Conference and

Exhibition, Houston, Texas, 1988.

This journal is © The Royal Society of Chemistry 2014