A Compositional Model for CO2 Floods

Including CO2 Solubility in Water

Yih-Bor Chang,* SPE, B.K. Coats,** SPE, and J.S. Nolen,*** SPE, Western Atlas Software

Summary

This paper presents a three-dimensional, three-phase compositional

model for simulating CO2 flooding, including CO2 solubility in
water. In this model, CO2 is allowed to dissolve in the aqueous
phase, while all other components, except water, exist in the oil and
gas phases. Oil- and gas-phase densities and fugacities are modeled
by a cubic equation of state. The aqueous phase properties are
functions of the amount of dissolved CO2. CO2 solubility is
computed using a CO2 fugacity coefficient table that is converted
internally from input CO2 solubility data as a function of pressure
at reservoir temperature. Correlations for computing the solubility
of CO2 in water and other properties of CO2-saturated water are
presented. The results for simulation runs with and without CO2
solubility in water are shown for comparison.

(EOS) is used to model oil- and gas-phase densities and fugacities.

An input table of CO2 solubility in water, water formation volume
factor, water compressibility, and water viscosity is entered as
functions of pressure at reservoir temperature. These data then are
used to compute the amount of CO2 in water and other aqueous
phase properties during simulation.
A new empirical correlation is presented for the solubility of CO2
in distilled water as a function of pressure and temperature. The
calculated solubility in distilled water can be adjusted further for the
effects of salinity to obtain the solubility of CO2 in NaCl brine.
Furthermore, correlations for computing the formation volume
factor, compressibility, and viscosity of CO2-saturated water are
given. Finally, we present simulation results that compare reservoir
performances with and without CO2 solubility in water.

Introduction

Compositional models using a cubic equation of state usually are

used to simulate the enhanced recovery process of gas injection. In
most of the published models, for example, Coats1 and Young and
Stephenson,2 all hydrocarbon components exist in the oil and gas
phases but are not allowed to dissolve in the aqueous phase.
Usually, this assumption is adequate because the hydrocarbon
solubility in water is low over the range of temperature and pressure
for gas injection. Carbon dioxide, however, is an exception to this
assumption. The solubility of CO2 in water is much higher than that
of hydrocarbon components and is a factor that cannot be neglected
in the simulation process.3 This is especially true when CO2 is
injected into a previously waterflooded reservoir or when CO2 is
injected with water for mobility control.4 The objective of this
paper is to model oil recovery processes involving CO2 injection
while taking into account the effects of CO2 solubility in water.
The effects of the presence of an aqueous phase on the phase
behavior of CO2/hydrocarbon systems have been experimentally
studied by Pollack et al.5 It was found that the presence of water
reduces the amount of CO2 available for mixing with the hydrocarbons and shifts the pressure-composition diagram of the CO2/
crude oil system. The solubility of CO2 in water is a function of
temperature, pressure, and water salinity.3 A thorough study of CO2
solubility data in distilled water was presented by Dodds et al.6 In
general, CO2 solubility increases with pressure and decreases with
temperature. An increase in salinity of the reservoir water decreases
CO2 solubility significantly. Li and Nghiem7 used Henrys Law to
estimate CO2 solubility in distilled water and used the scaledparticle theory to take into account the presence of salt in the
aqueous phase. Enick and Klara8 also used Henrys Law to predict
CO2 solubility in distilled water. The decreased solubility of CO2
in brine was accounted for empirically by a single factor correlated
to the weight percent of dissolved solid; however, a wide scatter of
data characterizes their correlation.
A compositional model for simulating CO2 floods, including
CO2 solubility in water, is presented. In this model, hydrocarbons
and CO2 are allowed to exist in the oil and gas phases, whereas only
CO2 and water exist in the aqueous phase. A cubic equation of state
* Now with Texaco EPTD
** Now with Landmark Graphics Corp.
*** Now with Geoquest
Copyright 1998 Society of Petroleum Engineers
Original SPE manuscript received for review 6 August 1996. Revised manuscript
received 22 January 1998. Paper peer approved 27 January 1998. Paper (SPE
35164) first presented at the 1996 SPE Permian Basin Oil & Gas Recovery Conference, Midland, Texas, 2729 March.

SPE Reservoir Evaluation & Engineering, April 1998

General Description of the Model

The simulator described here is an isothermal, three-dimensional,

compositional model. Both fully implicit and implicit-pressure,
explicit-saturation (IMPES) formulations are included. This model
can simulate water, oil, and gas flow in reservoirs of heterogeneous
permeability and porosity. This simulator does not model the
three-hydrocarbon-phase phenomenon that has been observed for
some CO2/hydrocarbon systems.
Oil- and gas-phase densities and fugacities are calculated from
a cubic equation of state.9 Lohrenz et al.10 correlations are used for
oil- and gas-phase viscosities. The aqueous phase properties are
computed using a method discussed later. The fluid flow is simulated with Darcys law, incorporating viscous, gravitational, and
capillary forces. The model consists of mass balances for water and
n c hydrocarbon components and associated constraint equations.
It is assumed that instantaneous equilibrium exists between oil,
gas, and water, and that water and hydrocarbon components are
mutually insoluble. Only CO2 is allowed to exist in all three phases,
because very large quantities of CO2 are injected for CO2 floods,
and CO2 is the dominant component dissolved in water. Thus, we
assumed that the concentration of other components in water,
including H2S and light hydrocarbons, is negligible compared to
CO2. Under normal CO2 flood conditions, the CO2 solubility errors
introduced by this assumption should be small. Using this assumption, we were able to simplify the phase behavior calculation in the
aqueous phase and to reduce computation time greatly. There may
be cases where more than one dominant component dissolves in the
aqueous phase, and the presence of other components may substantially change the solubility of CO2 in water. The effects of
multiple dominant components dissolved in water are not considered in this model.

Correlations for CO2-Saturated Water Properties

Correlations are developed for estimating properties of CO2-saturated water for binary CO2/water systems as functions of temperature, pressure, and salinity.
CO2 Solubility in Water. The solubility of CO2 in distilled water
may be estimated as

Rsw 5 a z p z 1 2 b z sin

Rsw 5 Rosw 1 m z ~ p 2 po !

czp
p
z
2 czp11

DG

if p , po . . . . . . . (1)

if p $ po . . . . . . . . . . . . . . . . . . . (2)
155

ai z 1023i z T i . . . . . . . . . . . . . . . . . . . . . . . . (3)

bi z 1023i z T i

where a 5

i50
4

b5

0 , b , 1 . . . . . . . . . . . . . . . (4)

i50

O
4

c 5 1023 z

ci z 1023i z T i . . . . . . . . . . . . . . . . . . . . (5)

i50

po 5

2
z
p

sin21 ~b2 !
2
c z 1 2 z sin21 ~b2 !
p

. . . . . . . . . . . . . . . . . (6)

Rosw 5 a z po z ~1 2 b3 ! . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (7)

H F S

m 5 a 1 2 b sin
1

p
czp
z
2 c z po 1 1
o

c z po
c z po
p
p
z
o
2 cos
2 ~c z p 1 1!
2 c z po 1 1

DGJ

. . . . . . . . . . . . (8)

R sw is CO2 solubility in scf of CO2 per STB of water, T is

temperature (F), p is pressure (psia), and the coefficients are
shown in Table 1.
This correlation matches the solubility data of Wiebe11 for liquid
and supercritical CO2 in water within 610 scf/STB for temperatures between 54 F and 212 F and pressures up to 10,000 psia.
Fig. 1 is a comparison of the computed data and the measured data
by Wiebe11 at 104 F, 122 F, 167 F, and 212 F. This correlation
also matches the data by Gillespie and Wilson12 within the stated
accuracy between 85 F and 200 F. The two data sets, however,
are not consistent at lower temperatures, and the data by Wiebe11
are used in developing our correlation.
The calculated solubility in distilled water can be adjusted further
for the effects of salinity to obtain the solubility of CO2 in brine:

log

S D

Rsb
5 20.028 z C z T20.12 ,
Rsw

Fig. 1Comparison of measured11 and calculated CO2 solubilities in distilled water.

Fig. 2Comparison of measured13,14 and calculated CO2 solubilities in NaCl brine at 695.5 psia.

. . . . . . . . . . . . . . . . . . . . . (9)

where R sb is CO2 solubility in scf of CO2 per STB of brine, S is

the salinity of brine in weight percent of solid, and T is temperature
(F). The measured data in NaCl solutions from Malinin and
Savelyeva,13 Malinin and Kurovskaya14 (Fig. 2), and McRee15
(Fig. 3) were used to obtain this correlation. Eq. 9 also matches
these CO2 solubility data in NaCl solution within 610 scf/STB.
Water Formation Volume Factor. The formation volume factor
of CO2-saturated water (or brine) is calculated with

rw, sc 1 0.02066 z Rsb

Bw 5
, . . . . . . . . . . . . . . . . . . . . . . . . (10)
rw, atm 1 0.0058 z Rsb
where B w is water formation volume factor in reservoir barrel per
STB of water (rb/STB), r w,sc is water density at standard temperature and pressure in lb/ft3, and r w,atm is water density at reservoir
temperature and 14.7 psia in lb/ft3. Both r w,sc and r w,atm can be
calculated from, for example, Rowe and Chou16 correlations. Eq.

10 is developed using the data of Parkinson and De Nevers17 and

Sayegh and Najman.18
Water Compressibility. For pressures below 5,000 psia, the compressibility of CO2-saturated water is calculated using the correlation from Rowe and Chou.16 For pressures greater than 5,000
psia,

1
1
5
1 7.033 z ~ p 2 5000!,
cw cw, 5000

. . . . . . . . . . . . . . . . . . (11)

where c w,5000 is water compressibility at reservoir temperature and

5,000 psia from the Rowe and Chou16 correlation in psia21. The
above correlation is derived from the water-compressibility correlation of Osif.19 The effects of dissolved CO2 in water on water
compressibility is assumed negligible when using the Rowe and
Chou16 and Osif19 correlations.

TABLE 1VALUES OF COEFFICIENTS IN EQS. 3 TO 5

ai
bi
ci
156

i50

i51

1.163
0.965
1.280

216.630
20.272
210.757

i52

i53

i54

111.073
0.0923
52.696

2376.859
20.1008
2222.395

524.889
0.0998
462.672

SPE Reservoir Evaluation & Engineering, April 1998

O
nc

i51

O
nc

xi 5 1

yi 5 1

and

So 1 Sg 1 Sw 5 1.

. . . . . . . (16)

i51

IMPES Formulation. For the IMPES formulation, the variable n i

in Eq. 12 is expressed as

ni 5 F z zi 1

Sw z RCO2w
z di, CO2 i 5 1, 2, . . . , nc . . . . . . . . . (17)
Bw

where F 5 jo So 1 jg Sg
and zi 5 L z xi 1 V z yi

Fig. 3Comparison of measured15 and calculated CO2 solubilities in NaCl brine at 100F.

Water Viscosity. The effects of dissolved CO2 in water on water

viscosity were found to be very small by Sayegh and Najman.18
Water viscosity can be calculated from the Kestin et al.20 correlation as a function of temperature, pressure, and NaCl concentration, but this neglects the effects of dissolved CO2.
Reservoir Simulation Model

This reservoir simulation model consists of 2n c 1 5 equations for

each gridblock, where n c is the number of hydrocarbon components, including CO2. The n c hydrocarbon components and water
mass balance equations are

Rmbi 5

Vm
kro
@~f z ni!n11 2 ~f z ni!n# 2 D Tm z joxi z ~Dp 2 goDD!
Dt
mo

F
F

krg
2 D Tm z jg yi z ~Dp 1 Dpcgo 2 gg DD!
mg

krw wco2
2 D Tm z
z
z ~Dp 2 Dpcwo 2 gw DD! z di, co 2
m w Bw
1 qi 5 0
Rmb w 5

i 5 1, 2, . . . , nc . . . . . . . . . . . . . . . . . (12)

FS D S D G
F

Vm
Dt

f z Sw
Bw

2 D Tm z

n11

f z Sw
Bw

krw 1
z
z ~Dp 2 Dpcwo 2 gw DD! 1 qw 5 0,
m w Bw

. . . . . . . . . . . . . . . . . . . . . . . . . . (18)
i 5 1, 2 , . . . , nc .

. . . . . . . . . . (19)

The unknown variables are V, y i , i 5 1, . . . , n c ; z i , i 5 1, . . . ,

n c ; R CO2w, F, S w , and p for a predominantly liquid cell, and L, x i ,
i 5 1, . . . , n c ; z i , i 5 1, . . . , n c ; R CO2w , F, S w , and p for a
predominantly vapor cell. For a three-phase gridblock, the 2n c 1
3 equations of fugacity constraints, mass balances, and saturation
constraints are first linearized in terms of these 2n c 1 5 unknowns.
Then the variables y n c (or x n c) and z n c are eliminated using the mole
fraction constraints. Fig. 4 shows the structure of the Jacobian for
the diagonal term with three hydrocarbon components. We have
assigned CO2 to be component number one in this figure. In this
figure, the symbols x and *, respectively, denote new and
modified elements resulting from CO2 solubility in water. The
symbol o denotes elements that are not affected by considering
CO2 solubility in water. Only five more elements from the material
balance equations are added to the Jacobian compared with the
Jacobian without including CO2 solubility in water. These extra
elements are the derivatives of CO2, total hydrocarbon, and water
material balance equations with respective to R CO2w and S w . They
can be eliminated easily, and then the resulting Jacobian assumes
the same form as the one without including CO2 solubility in water.
Thus, we can apply the same solution procedure as suggested by
Young and Stephenson.2
Implicit Formulation. For the implicit formulation, the variable
n i in Eq. 12 is expressed as

ni 5 jo So xi 1 jg Sg yi 1

Sw z RCO2w
z di, CO 2
Bw

i 5 1, 2, . . . , nc .

. . . . . . . . . . . . . . . . . . . . . . . . . . (20)
The unknown variables are y i , i 5 1, . . . , n c ; x i , i 5 1, . . . , n c ;
R CO2w , S o , S w , S g , and p. In the case of a three-phase gridblock,
the 2n c 1 2 equations of fugacity constraints and mass balances
are first linearized in terms of these 2n c 1 5 unknowns. Then the
variables y n c, and x n c are eliminated using the mole fraction
constraints, and the variable S o is eliminated using the saturation

. . . . . . . . . . . . . . . . . . . . . . . . . . (13)
where d i, CO2 5 1, if Component i is CO2. Otherwise d i, CO2 5 0.
In Eq. 12, only CO2 is allowed to dissolve in the aqueous phase,
whereas all components except water exist in the oil and gas phases.
Summing Eq. 12 over all hydrocarbon components yields the total
hydrocarbon equation, which our model uses instead of the mass
balance equation for component n c .
The n c 1 1 fugacity constraints (for a three-phase block),

Rf i 5 fi, o 2 fi, g 5 0

i 5 1, 2, . . . , nc . . . . . . . . . . . . . . . . . (14)

and RCO2 5 fCO2 , x 2 fCO2, w 5 0,

. . . . . . . . . . . . . . . . . . . . (15)

where x is one of the hydrocarbon phases (o or g), expressing the

requirement that oil- and gas-phase fugacities must be equal for
each hydrocarbon component, and that CO2 fugacities in the
hydrocarbon phase and the aqueous phase must be equal. There are
two mole fraction constraints and one water saturation constraint
SPE Reservoir Evaluation & Engineering, April 1998

Fig. 4 Structure of diagonal Jacobian of the IMPES formulation

for three components. The symbol o represents elements
without CO2 solubility in water, x represents elements due to
CO2 solubility in water only and * represents elements for
which the CO2 solubility term has been added to the hydrocarbon flow terms.
157

constraint. In the solution procedure, Gaussian elimination is used

to solve for the n c 1 1 unknowns of y i , i 5 1, . . . , n c 2 1; x 1 ,
and R CO2w in terms of the remaining n c 1 1 unknowns of x i , i 5
2, . . . , n c 2 1; S w , S g , and p. The remaining n c 1 1 unknowns
are called primary unknowns, whereas the n c 1 1-eliminated
unknowns are called secondary unknowns. The linearized mass
balance equations, after eliminating the secondary unknowns using
the constraint equations, form a set of n c 1 1 equations in terms
of n c 1 1 primary unknowns. The linear system is solved iteratively, and the solution vector then is used to compute the secondary
unknowns through back-substitution, as described by Coats.1
Phase Equilibrium Calculations. During simulation, we need to
determine the numbers and types of phases in a grid cell. A
sequential approach is used in the model. For a cell with hydrocarbons, CO2 and water, a two-phase hydrocarbon/water flash is
performed first to compute the amount of CO2 in water and the
composition of the hydrocarbon phase. CO2 hydrocarbon/water
fugacity constraint is the primary equation, and R CO2w is the
primary unknown for this flash calculation.
Next, the computed hydrocarbon composition is tested by a
two-phase oil/gas flash or a phase-stability calculation for the
possibility of splitting the single hydrocarbon phase into two
phases. If only one hydrocarbon phase can be found at the end of
the calculation, then the phase-equilibrium calculation is complete.
If the calculation yields two hydrocarbon phases, then we know that
oil/gas/water three-phase equilibria exist in this grid cell, and we
perform a three-phase flash to determine the compositions of the
oil, gas, and aqueous phases.
Aqueous-Phase Properties. An input table of CO2-saturated water properties of a binary CO2/water system is needed in our model.
These data include CO2 solubility in water, water formation volume
factor, water compressibility, and water viscosity. They are entered
as a function of pressure at the reservoir temperature. The format
and use of this table are similar to the conventional black oil table,
except that now the solution gas is CO2, the liquid phase is water
instead of oil, and the formation volume factor is for CO2-saturated
water instead of gas-saturated oil.
A table of CO2 fugacity coefficients in the aqueous phase, fCO2w ,
is computed internally from the input CO2 solubility for a saturated
binary CO2/water system. The phase-equilibrium constraint for the
CO2 component in a binary CO2/water system is

fCO2, w 5 fCO2, v

or

vCO2 z fCO2, w 5 yCO2 z fCO2, v ,

. . . . . . . (21)

where we use the subscript w to denote the aqueous phase and v to

denote the CO2-rich phase. y CO2 is the mole fraction of CO2 in the
CO2-rich phase, and vCO2 is the mole fraction of CO2 in the
aqueous phase, which can be computed from the input CO2 solubility in water. At the conditions of typical CO2 floods, the water
vapor pressure is small relative to the total pressure and CO2 vapor
pressure. If we assume that the water mole fraction in the CO2-rich
phase is negligible (i.e., y CO2 ' 1), the fugacity coefficient of CO2
in the aqueous phase can be approximated as

fCO2, w <

EOS
fCO
2, v

vCO2

, . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (22)

EOS
where fCO
is the fugacity coefficient of pure CO2 in the vapor
2, v
phase computed using an EOS.

For a binary CO2/water system, the computed fugacity coefficient of CO2 in the aqueous phase, fCO2,w , is a function of
EOS
temperature and pressure but not composition, because fCO
is for
2, v
pure CO2 and vCO2 is for the binary CO2/water system. Both are
functions of temperature and pressure only at a constant water
salinity. For isothermal reservoir simulations, fCO2, w is assumed to
be a function of pressure only and is used in Eq. 15 for computing
CO2 content in the aqueous phase with an arbitrary hydrocarbon
system. The assumption that fCO2, w is not a function of composition may introduce error for the amount of CO2 in the aqueous
phase; however, because no more than a few percent of CO2 is
dissolved in the aqueous phase, the error should not be significant.
After computing CO2 content in the aqueous phase using the
phase-equilibrium algorithm mentioned earlier, water viscosity,
saturated water formation volume factor (B sat
w ), water compressibility (c w ), and water saturation pressure ( p sat
w ) can be calculated
from the input table of CO2-saturated water properties using linear
interpolation. These aqueous-phase properties are assumed to be
functions only of aqueous-phase CO2 content. The water formation
volume factor (B w ) at block pressure p is calculated as

1
1
5 sat z @1 1 cw z ~ p 2 psat
w !#.
Bw Bw

. . . . . . . . . . . . . . . . . . . . (23)

Simulation Results

The objective of these simulation runs is to investigate the effects

of CO2 solubility in water on oil recovery. For each case, we made
a pair of simulations in which CO2 solubility in water was included
and CO2 solubility was ignored. The reservoir and fluid properties
are from Killough and Kossack.21 The same three-dimensional 7 3
7 3 3 grid is also used in our study. The injection well is located
in the gridblock (i, j) 5 (1, 1) with all three layers perforated. The
production well is in the gridblock (i, j) 5 (7, 7) with only the third
layer perforated. We changed the injected gas from a mixture of C1,
C2, and C6 to CO2. The solubility of CO2 in water is calculated
using Eq. 1 to 9, and the water-formation volume factor is from Eq.
10 at 160F and 10 wt% salinity. For the purpose of comparison,
constant values of water compressibility and viscosity, 3.3 3 1025
psi21 and 0.7 cp, respectively, are used for all runs. Sayegh and
Najman18 also have observed that the effects of dissolved CO2 in
water on water viscosity were very small, and nearly constant
values of viscosity were reported in their paper.
Two water alternating gas (WAG) injection cases were designed
in our study. Both cases start with 2 years of oil production at a
maximum rate of 2,000 STB/D with no injection. For Case A, a
3-month WAG cycle begins at the end of Year 2. The WAG cycles
last for 12 years, and they are followed by a 6-year chase water
period. For Case B, waterflood begins at the end of the second year
and continues for 4 years. The waterflood is followed by 8 years
of 3-month WAG cycles, which are followed by 6 years of chase
water. Case A mimics secondary CO2 flooding, whereas Case B
mimics tertiary CO2 flooding. Two runs were made for each case,
the first one simulated an immiscible CO2 flood, and the second
simulated a multiple contact-miscible CO2 flood. Tables 2 and 3
summarize run conditions.
First, we examined the ratio of total CO2 dissolved in water to
the total amount of CO2 in the reservoir. About 8% to 13% of the
CO2 in the reservoir dissolved in water during WAG cycles. The
dissolved CO2, which is considered lost, is not available for
mixing with oil. We also compared the ratio of total CO2 dissolved

TABLE 2SUMMARY OF SIMULATION CASES

Case A
Case B
158

Initial
Production
(yr)

Waterflood
(yr)

WAG
(yr)

Chase
Water
(yr)

1 to 2
1 to 2

3 to 6

3 to 14
6 to 14

15 to 20
15 to 20

SPE Reservoir Evaluation & Engineering, April 1998

TABLE 3SUMMARY OF SIMULATION RUNS

Run
Run
Run
Run

A-1
A-2
B-1
B-2

Max. Inj. qw
(STB/D)

Max. Inj. qg
(scf/D)

Max. Pbh (psia)

Injector

Min Pbh (psia)

Producer

7,500
10,000
7,500
10,000

15,000
20,000
15,000
20,000

4,500
4,500
4,500
4,500

500
1,500
500
1,500

in water to cumulative CO2 injected. For all runs, this ratio reaches
its peak value near the end of the first year of WAG cycles, and then
it gradually declines until the end of the WAG cycles. The maximum ratios are about 7% and 12% for Cases A and B, respectively.
Oil recoveries were compared for simulation runs with and
without considering CO2 solubility in water. For Case B, including
CO2 solubility not only delays the oil recovery but also reduces the
final oil recovery. The effect is more pronounced for Run B-2
(multiple contactmiscible conditions) than for Run B-1 (immiscible conditions). At the end of Year 9, the runs with CO2 solubility
produced about 200 MSTB less oil than the runs with no CO2
solubility. This is about 5% of post-waterflood oil recovery at year
nine. For Case A runs, the major impact of CO2 solubility in water
is delaying oil recovery, and at the end of year eight, the runs with
CO2 solubility produced about 180 MSTB oil less than the runs
without CO2 solubility.
Although a simple reservoir model was used in the above
simulations, similar effects of CO2 solubility in water on oil
production were also found in one of our field simulations. The
reservoir had about 50 years of primary production and waterflooding, and it is planned for CO2 flooding for almost 30 years of
WAG cycles. The CO2 flood is followed by another 15 years of
chase-water injection. Our simulation results show that about 10%
to 15% of total injected CO2 dissolved in water, or about 11% to
15% of the CO2 in the reservoir dissolved in water during the WAG
cycles. A maximum of 15% of post-waterflood oil production, or
8% of total oil production, are delayed when CO2 solubility is
simulated. These results are in general agreement with Enick and
Clara,4 who also reported that the ultimate recovery can be substantially less than that calculated when CO2 solubility is ignored.
The simulation results presented here, however, should not be
generalized to all reservoirs without considering other factors. The
effect of CO2 solubility in water on CO2 flood oil recovery will also
depend on reservoir heterogeneity, rock and fluid properties, and
the injection/production history of the reservoir.
Conclusions

A three-dimensional, three-phase compositional model is presented

for simulating CO2 floods, including CO2 solubility in water. Oiland gas-phase densities and fugacities are modeled by a cubic
equation of state. CO2 solubility in water is computed using a CO2
fugacity coefficient table, which is internally converted from input
CO2 in water solubility data as a function of pressure at the reservoir
temperature.
Correlations for estimating CO2 solubility in water and NaCl
brine, the water formation volume factor of CO2-saturated water,
water compressibility, and water viscosity are presented. The CO2
solubility in water can be estimated with the correlations within
610 scf/STB for most CO2 flood conditions.
Using a simple three-dimensional model, simulation results show
that about 10% of the CO2 injected is dissolved in water and is
unavailable for mixing with oil. This may be considered lost to
the aqueous phase. The solubility effects are more pronounced for
tertiary CO2 floods than secondary CO2 floods. As much as 5% of
post-waterflood oil production can be delayed from our simulation
results.
Nomenclature

B w 5 water formation volume factor, rb/STB

c w 5 water compressibility, 1/psi
SPE Reservoir Evaluation & Engineering, April 1998

D 5 depth, ft
F 5 hydrocarbon phase molar density, lbmole/ft3
f i,o , f i,g 5 fugacity of component i in oil and gas phases
k ro , k rg , k rw 5 oil, gas, and water relative permeabilities
L 5 oil mole fraction in hydrocarbon phase
n i 5 total molar density of component i
p 5 pressure, psia
p cwo , p cgo 5 oil/water and oil/gas capillary pressures, psia
qg 5 gas well rate, scf/D
q i 5 hydrocarbon component i well rate, lbmole/D
q w 5 water well rate, STB/D
R CO2w 5 solubility of CO2 in water, lbmole/STB
R sw 5 solubility of CO2 in distilled water, scf/STB
R sb 5 solubility of CO2 in NaCl brine, scf/STB
C 5 salinity of NaCl brine, wt%
S o , S g , S w 5 oil, gas, and water saturations, fraction
T 5 temperature, F
T m 5 transmissibility
t 5 time, day
V m 5 gridblock bulk volume, bbl
V 5 gas mole fraction in hydrocarbon phase
x i 5 mole fraction of component i in the oil phase
y i 5 mole fraction of component i in the gas phase
z i 5 overall mole fraction of component i in the
hydrocarbon phases
Greek symbols
D 5
f 5
fCO2,v 5
fCO2, w 5
go, gg, gw 5
mo, mg, mw 5
rw 5
vCO2 5
jo, jg 5

difference operator
porosity, fraction
fugacity coefficient of CO2 in CO2-rich phase
fugacity coefficient of CO2 in water
oil-, gas-, and water-specific gravities
oil, gas, and water viscosities, cp
water density, lb/ft3
mole fraction of CO2 in water
oil and gas molar densities, lbmole/ft3

Subscripts and Superscripts

o, g, w 5 oil, gas, and aqueous phases, respectively
i 5 component index
n 5 time-step index
Acknowledgments

The authors wish to thank Western Atlas Software for permission

to publish this paper.
References
1. Coats, K.H.: An Equation of State Compositional Model, SPEJ
(October 1980) 363; Trans., AIME, 269.
2. Young, L.C. and Stephenson, R.E.: A Generalized Compositional
Approach for Reservoir Simulation, SPEJ (October 1983) 727.
3. Klins, M.A.: Carbon Dioxide Flooding Basic Mechanisms and Project
Design, International Human Resources Development Corp., Boston
(1984).
4. Enick, R.M. and Klara, S.M.: Effects of CO2 Solubility in Brine on the
Compositional Simulation of CO2 Floods, SPERE (May 1992) 253.
5. Pollack, N.R., et al.: Effect of an Aqueous Phase on CO2/Tetradecane
and CO2/Maljamar-Crude-Oil Systems, SPERE (May 1988) 533.
159

6. Dodds, W.S., Stutzman, L.F., and Sollami, B.J.: Carbon Dioxide

Solubility in Water, Industrial and Engineering Chemistry (1956) 1,
No. 1, 92.
7. Li, Y.-K. and Nghiem, L.X.: Phase Equilibria of Oil, Gas and Water/
Brine Mixtures from a Cubic Equation of State and Henrys Law,
Canadian J. of Chemical Engineering (1986) 64, No. 3, 486.
8. Enick, R.M. and Klara, S.M.: CO2 Solubility in Water and Brine Under
Reservoir Conditions, Chemical Engineering Communications (1990)
90, 23.
9. Coats, K.H.: Simulation of Gas Condensate Reservoir Performance,
JPT (October 1985) 1870; Trans., AIME, 279.
10. Lohrenz, J., Bray, B.G., and Clark, C.R.: Calculating Viscosities of
Reservoir Fluids from their Compositions, SPEJ (October 1964) 1171;
Trans., AIME, 231.
11. Wiebe, R.: The Binary System Carbon Dioxide-Water Under Pressure, Chemical Reviews (1941) 29, 475.
12. Gillespie, P.C. and Wilson, G.M.: Vapor-Liquid and Liquid-Liquid
Equilibria: Water-Methane, Water-Carbon Dioxide, Water-Hydrogen
Sulfide, Water-nPentane, Water-Methane-nPentane, Research Report
RR-48, Gas Processors Assn. Tulsa, Oklahoma (April 1982) 11.
13. Malinin, S.D. and Savelyeva, N.I.: The Solubility of CO2 in NaCl and
CaCl2 Solutions at 25, 50 and 75 Under Elevated CO2 Pressures,
Geochemistry International (1972) 9, No. 1, 410.
14. Malinin, S.D. and Kurovskaya, N.A.: Solubility of CO2 in Chloride
Solutions at Elevated Temperatures and CO2 Pressures, Geochemistry
International (1975) 2, No. 2, 199.
15. McRee, B.C.: CO2: How It Works, Where It Works, Petroleum
Engineer International (November 1977) 49, No. 12, 52.
16. Rowe, A.M. and Chou, J.C.: Pressure-Volume-Temperature-Concentration Relation of Aqueous NaCl Solutions, J. of Chemical and
Engineering Data (1970) 15, No. 1, 61.
17. Parkinson, W.J. and De Nevers, N.: Partial Molal Volume of Carbon
Dioxide in Water Solutions, Industrial and Engineering Chemistry
Fundamentals (1969) 8, No. 4, 709.
18. Sayegh, S.G. and Najman, J.: Phase Behavior Measurements of CO2SO2-Brine Mixtures, The Canadian J. of Chemical Engineering (1987)
65, No. 2, 314.
19. Osif, T.L.: The Effects of Salt, Gas, Temperature, and Pressure on the
Compressibility of Water, SPERE (February 1988) 175.
20. Kestin, J., et al.: Effect of Pressure on the Viscosity of Aqueous NaCl
Solutions in the Temperature Range 20150C, J. of Chemical and
Engineering Data (1978) 23, No. 4, 328.
21. Killough, J.E. and Kossack, C.A.: Fifth Comparative Solution Project:
Evaluation of Miscible Flood Simulators, SPE 16000 presented at the

160

1987 SPE Symposium on Reservoir Simulation, San Antonio, Texas,

14 February.

Metric Conversion Factors

cp 3 1.0*
ft3 3 2.83 685
F (F232)/1.8
psi 3 6.894 757

E203 5Pa z s
E202 5 m3
5C
E100 5 kPa
SPEREE

*Conversion factor is exact.

Yih-Bor Chang works with Texaco EPTD in Houston. Previously,

he worked in reservoir simulation and phase-behavior modeling at Western Atlas Software and Landmark Graphics Corporation. His research interests include reservoir engineering, modeling, simulation, and EOR. Chang holds a BS degree in
chemistry from Tunghai U., Republic of China, and MS and PhD
degrees in petroleum engineering from the U. of Texas at Austin.
Brian K. Coats is a senior software engineer with Landmark
Graphics Corp.s Integrated Products Group in VIP simulator
development. He has developed or codeveloped, supported,
and maintained thermal, compositional, and black oil models
for reservoir simulation and regression models for use in volatile
and heavy oil equation-of-state characterization. His current
interests include simulation on (not necessarily k-orthogonal)
unstructured grid systems and continued development and
support of the VIP-THERM reservoir simulator. Coats holds BS and
MS degrees in chemical engineering from the U. of Texas at
Austin. James S. Nolen is manager of reservoir simulation research for Schlumberger GeoQuest in Houston. Previously, he
worked for Exxon Production Research and Intercomp. In 1979
he founded J.S. Nolen & Associates, Inc., which was merged
into Western Atlas Software in 1988. He joined GeoQuest upon
the sale of Western Atlas Software in 1996. He holds a BS degree
in chemical engineering from Louisiana State U. and a PhD
degree in chemical engineering from Tulane U.

Chang

Coats

Nolen

SPE Reservoir Evaluation & Engineering, April 1998