Effects of C02 Solubility in
Brine on the Compositional
Simulation of C02 Floods
R.M. Enlck, SPE, U. of Pittsburgh, and S.M. Klara, SPE, U.S. DOE
Summary. A compositional simulator was modified to account for the effects of brine on CO 2 solubility and aqueous-phase density
and viscosity. Simulations of oil displacements using continuous injection of CO 2 , CO 2 slugs, and CO 2 water-alternating-gas (WAG)
injection were conducted to determine the effects of brine on the performance of CO 2 EOR processes.
Introduction
The presence of an aqueous phase will influence the phase behavior
of CO 2 /hydrocarbon systems. The goal of this study was to corre-
late these effects accurately and simply, to incorporate them into
a reservoir simulator, and to evaluate their significance on the
modeling of CO2 floods. This was accomplished by comparing
simulations in which (1) the CO 2 solubility in the aqueous phase
was ignored, (2) the CO 2 solubility in the aqueous phase was equal
to that of CO 2 solubility in pure water, and (3) the effect of dis-
solved solids in the aqueous phase on CO 2 solubility was consid-
ered. Henry's law and an empirical correlation for the effect of
dissolved solids on solubility were used to predict CO 2 solubility
in the aqueous phase. 1
Effects of Dissolved Solids on the Equilibrium
and Transport Properties
CO2 Solubility in Pure Water. In a previous study,1 110 solu-
bilities along 12 isotherms between 25 and 250°C [77 and 482°F]
were compiled over the 4.83 to 72.41 MPa [700 to 10,500 psia]
pressure range to develop correlations for the Henry's constant for
the CO 2 /water system.
One of the empirical correlations for the Henry's constant (used
in the Krichevsky-Kasarnovsky2 equation) is given below 1:
H*= -729475+2475.322(1)-2.356042(1)2 +0.65367
x 10- 3(1)3 ...................................... (1)
The Lyckman et at. 3 correlation was used in conjunction with this
expression to determine the partial molar volume of CO 2 in water.
CO 2 fugacities were determined from the Peng-Robinson 4 equa-
tion of state (EOS).
Effect of Dissolved Solids on CO 2 Solubility. In a previous
study, 1 the reduction of CO 2 solubility on a weight basis was cor-
related to the solubility of CO 2 in pure water on a weight basis,
as suggested by Klins. 5 Although the correlation is completely em-
pirical, it is simple to use and is a function of one parameter, total
dissolved solids (TDS), rather than individual ion concentrations.
This correlation, also used in this study, is presented below.
RC02,b=Rc02,W[1.0-(4.893 414XlO-2)CroS+(0.130 2838
XlO- 2)Cfus-(0.187 119 9XlO-4)Cfus]' ............ (2)
where C ros is the TDS concentration in weight percent. The con-
version between CO 2 solubility in brine on a mass and molar ba-
sis was needed because the simulator was designed for molar
component material balances.
R C02 ,b/44
YC02 b= , ................ (3)
, RC02,b/44+(I-Rco2,b)/Mb
where M b =M(NaC1+ H20 ) = 1051.2/(58.4-0.404 Cros) . .... (4)
Eqs. 3 and 4 represent the brine phase as a water/NaCI solution
of equivalent TDS concentration.
Copyright 1992 Society of Petroleum Engineers
SPE Reservoir Engineering, May 1992
Effect of Dissolved Solids on Aqueous-Phase Density. In review-
ing aqueous-phase density data, we noted that the molar densities
of brines 6, 7 are roughly equal to that of water at the same temper-
ature and are relatively unaffected by pressure. 5,8 Therefore, a
reasonable estimate for the mass density of either NaCI or CaCl2
solutions was obtained by multiplying the molar density of water
by the molecular weight Of the brine (assuming all dissolved solids
to be NaCI).
Pmb=PMbMb=PMwM(NaC1+H20)' .................... (5)
Any other correlation, such as that of Rowe and Chou,9 could also
be used.
Effect of Dissolved Solids on Aqueous-Phase Viscosity. Several
data sources for the viscosity of (H 20 + NaCI) solutions, one
tabular 9 and one graphical, 10 were used to develop the following
expression, which relates brine viscosity to that of pure water for
the 25 to 100°C [77 to 21ZOF] temperature range under the assump-
tion that the effects of pressure and dissolved CO 2 may be
ignored 5 ,8:
#-'(H20+NaCI) =#-'w[1 +(1.892 x 1O- 2)Cros+(1.215 x 10- 4 )
XCfus+(1.941 x 1O-5)Cfus]. ...................... (6)
Although brine can contain many dissolved ions, we noted that on
a weight basis,
#-'b=#-'(H2 0 + NaCI)' . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (7)
Although these particular expressions for brine density and vis-
cosity and CO2 solubility in brine were used in this study, any
correlations can be used in the following procedure. These simple
correlations were used because they were functions of only one brine
parameter, C ros. If more accurate correlations based on specific
ion concentrations are used, additional material balances and cor-
relation parameters must be introduced for each dissolved ion.
Governing Equations for
Compositional Simulation
The compositional model used an implicit-pressure, explicit-
saturation, and -composition (IMPESC) formulation. lIThe trans-
missibilities, capillary pressures, and gravity terms are evaluated
at the nth time level. The L superscript in the following expres-
sions refers to the Lth iteration of the pressure-convergence loop
within a time level. Components 1 through nc include CO 2 , gases,
and hydrocarbons. Component nc + 1 refers to water and nc + 2 to
the dissolved solids. Components 1 through 0: exhibit solubility in
the aqueous phase; components 0:+ 1 through nc are not soluble
in the aqueous phase.
Nghiem 12 presented the basic equations used in an IMPESC
compositional simulator, in which CO 2 solubility in the aqueous
phase is accounted for. To account for dissolved solids in the in-
jected and reservoir brines, an additional component material bal-
ance was used in which the solids were assumed to be soluble in
only the aqueous phase. Note that because the effects of dissolved
solids were correlated to a single concentration, Cros, only one
253
additional component balance is required. In essence, the brine is
modeled as an aqueous solution of NaCI of equivalent weight con-
centration. The equation for the dissolved solids material balance is
t1tCY~c+2,b(M'/1} I -M'gob -gC MJ)+q~c+2 =
(Vblt1t)[(Ync+2,bPMb4>Sb)n+1 -(Ync+2,bPMb4>Sb)n] . ..... (8)
The water balance, in which water is restricted to the aqueous
phase, is identical to Eq. 8, except that the nc+2 subscript is
changed to nc + I. The water material balance can be rearranged
to provide an expression for the aqueous-phase saturation:
Sb(L+I)-
- When an aqueous phase is undersaturated with CO 2 , or any
t1tCy~c+ l,btlFtL + I ) +q~c+ I + (Vb 1t1t )(Ync + l,bPMbcf>Sb)n
(Vbl t1t)(yn c + l,bPMbcf»(L+ I)
.................................... (9)
When the presence of CO 2 and other light gases in the brine are
accounted for, the hydrocarbon balance takes the form
t1[tCyJt"b(M'/1/I_M'gob-gCMJ)]
(Vblt1t)4>(L+ I) (PMoSo)(L+ I). . . . . . . . . . . . . . . . . . . . . . . . (14)
This forces the insoluble components in the aqueous phase to re-
main in the gridblock because, under these conditions, Eqs. 13 and
14 can be combined to yield
(ZmPMoSo)L = (ZmPMoSo) (L+ I) , ..................... (15)
for Components a+ 1 through n. Components 1 through a, how-
ever, can be extracted by the aqueous phase as it passes through
the gridblock. These overall mole fractions will be used in the flash
calculation because only the hydrocarbon phases will be described
with the EOS.
other water-soluble gas (which occurs as injected water displaces
all the gas phase from a miscibly swept block), Ym,b can be ob-
tained from Eq. 13 by setting So=S =0, Sb=l, and Z(L+l) =0
and solving for Y~tl). Therefore, under these conditions for m= 1
through a,
(L+ I) _ t1tCyJt"b[tlFtL+l)] +qJt, + (Vbcf>nlt1t)yJt"bPttb
Y b -
m, (Vb lt1t)(PMbcf»(L+ I)

+t1[t;YJt"g(M'/1/1 +M'gog-g;MJ)] ................................... (16)

+t1[tgyJt,.o(M'/1/I_ g gMJ)]+qJt,
= (Vb lt1t){4>n+ I [zn+ I (PMoSo +PMgSg)n+1
+(Ym,bPMbSb)n+l] _cf>n[zn(PMoSo +PMgSg)n
+(Ym,bPMbSb)n]) . ............................... (10)
To obtain a pressure equation, these component balances are
added after weighting the water and solids equations by a factor S:
(S-I)t1tC( Ync+I,b + Yn c+ 2,b)ntlFt+ I)
+ E t1tPtlFt+I)+S(qnc+1 +%c+2)n+qC
j =o,g,b
-(Vblt1t)[4>(n+l)a{n+I)_cf>nar]=O, ................ (11)
where al =(S-I)(Ync+l,b+Ync+2,b)PMbSb+
j =o,g,b
Otherwise, the amount of light gases, such as CO 2 , in the aque-
ous phase is calculated with the fugacity of CO 2 from the hydrocar-
bon phase, Henry's law, and the empirical factor for decreased
solubility caused by the presence of dissolved solids (see Eqs. 1
through 4 and Ref. I).
The amount of solids in the aqueous phase can be computed by
rearranging the component material balance:
(L+I) =
Y nc+ 2,b
t1t::'Y~c+2,btlF~L+ I) +q~c+2 +(Vb 1t1t )(Yn c +2,bPMbcf> S b)n
(Vblt1t)(PMbcf>Sb)(L+ I)
................................... (17)
The constitutive equation for the aqueous phase can be rearranged
to yield
E PMjSj. a

................................... (12)
The overall composition equation, in which Zi is defined as the To convert to TDS, the following expression may be used:
mole fraction of Component i in the hydrocarbon phases only (not
including any of Component i solubilized in the aqueous phase), 58.4XY~W
is obtained by rearranging the component balance and the overall Cms/lOO= ............... (19)
18xy(L+I) +58.4xy(L+1)
material balances: nc+1 nc+ 2

t1[I-ny n tlF(L+ If no variation in TDS concentration initially exists throughout

Z(L+I) )b m,b b I)] +t1[I-ny n. tlF(L+
)g m,g g I)] the formation (in all gridblocks) and if the injected brine has the
m a same TDS concentration as the formation brine, the TDS in the
aqueous phase will remain constant throughout the displacement.
In this case, solids tracking is not necessary. However, the CO 2
solubility in brine and the aqueous-phase density and viscosity should
+t1[tgyJt"otlF~L+ I)] +qJt, + (Vb 1t1t ) be modified to account for changes in their values caused by the
presence of dissolved solids. In the following examples, solids track-
+t1tg tlF~L+I) +qC+(Vb1t1t) ing was used in all the examples when CO 2 solubility in brine was
modeled, even if the in-place and injected brine had equivalent TDS
concentrations.
{cf>n[zJt,(PMoSo +PMgSg)n +(Ym,bPMbSb)n]
Fig. 1 shows the flow chart of the computational sequence that

tcf>n[(PMOSO+PMgSg)n+ ~ (Ym,bPMb Sb)n J- incorporates the equations for solids tracking, CO 2 solubility in
brine, brine density, and viscosity into a compositional simulator.

Example Calculations and Results

_cf>L( Ym,bPMbSb)L} Previous investigators l2- 16 have concluded that the modeling of
---=~-- ............................ (13)
CO 2 solubility is important for the tertiary recovery of oil because
a large volume of water is not only present in the reservoir, but
also is injected alternately or simultaneously with the CO 2 slug.
Some compositional simulators ignore solubility of gases in the
The denominator of Eq. 13 was changed to an equivalent quantity , aqueous phase; others include the solubility of CO 2 (and possibly
given below, when the oil phase became immobile and no gas phase other gases) in pure water. In the examples that follow, results ob-
was present: tained under these assumptions are compared with predictions ob-

254 SPE Reservoir Engineering, May 1992

 I Input Data I
I InHlallzaUon I
I Recurrent Data I
I Compute kr, T, Pc I
I Compute SourcelSlnks I I Time Step
Loop
I
ISolve For Pressure, aqns. 11, 12J

Pressure
1Solve For Overall ComposlUon, aqn. 13 1
Convergence
Loop I Flash Calculation I
I Solve For Brine Saturation, aqn. 9 I
I Compute Dissolved Solids In Brine, aqn. 171
ICalculate C02 Solubility In Pure H2O Using
Henry's Law, aqn.l • rel.l
I
I Calculate C02 Solubility In Brine, aqns. 2, 3
II Undersaturated, aqn. 16
I
lSolve For Water Concentretlon In Brlna, aqns.17,18 I
I Compute Brine Properties, aqns. 4-6 J
I
1 Computa 011 And Gas 5aturaUons I

I
I
I Compute Material Balance Error;
Next Time Step Size
I
I Print Output .1
~
Fig. 1-Computatlonal sequence for compositional simulator.

tained from aID IMPESC compositional model. This model in-
cludes the effects of dissolved solids in the brine on solubility, den-
sity, and viscosity.
The predicted oil recovery from two scenarios will be present-
ed. In the first scenario, the in-place hydrocarbons are recovered
by the continuous injection of CO 2 , Slugs of CO 2 followed by
brine are used to recover oil in the second scenario. These ID prob-
lems are similar to those presented by Coats, 17 except that the in-
jection gas is CO 2 , At reservoir conditions, multiple-contact
miscibility is attained with the oil, which is 20 mol % methane, 20
mol % butane, and 60 mol % decane. Table 1 lists the 25 simula-
tions that were used to evaluate the effect of dissolved CO 2 , ini-
tial water saturation, and dissolved solids in brine on oil recovery.
A complete data set for the simulations is available elsewhere. 18
Continuous Injection of CO 2 , The continuous injection of CO 2
into a formation with a high oil saturation was expected to yield
the smallest effects of CO 2 solubility in brine. The recovery curves
in Fig. 2 (Runs 1 through 3) confirm this. The loss of CO 2 to the
aqueous phase resulted in a larger time requirement to reach a given
recovery (or a smaller recovery at a given time). The maximum
difference, however, was only several percent between the curves
corresponding to models in which CO 2 solubility was ignored and
in which CO 2 solubility in pure water was considered. The differ-
ence was even smaller when dissolved solids were added because
they diminished the aniount of CO 2 that was solubilized into the
aqueous phase. The computational times for the runs involving
CO 2 solubility in brine and dissolved solids tracking were com-
parable with those corresponding to the runs in which CO 2 solu-
bility was ignored (see Table 1).
The results of the second set of simulations of continuous CO 2
injections (see Fig. 2, Runs 4 through 6) show that the shifts in
the recovery curves were greater because the initial water satura-
tion had increased from 0.2 to 0.8. This resulted in a larger amount
SPE Reservoir Engineering, May 1992
of CO 2 dissolving into the aqueous phase. The difference in time
required to attain a specified recovery (or recovery at a specified
time) was still relatively small, however. For these runs, 17% to
92 % more computational time was required when CO 2 solubility
in the aqueous phase was modeled.
Note that in either of the first two sets of simulations (Runs 1
through 3 or Runs 4 through 6), the ultimate recoveries are equal
after about 500 days, regardless of whether CO 2 solubility in brine
is modeled. This indicates that the phase behavior governing the
miscible displacement is the same for the runs, but longer times
are needed to achieve the displacement because of a loss of CO 2
to the aqueous phase. Essentially, all CO 2 lost to the aqueous phase
is replaced by the continued injection of CO 2 , (Note that the ulti-
mate recovery is slightly less than 100%, even though all the oil
was displaced from the porous media, owing to small oil losses
to the separator gas.)
Injection of CO2 Slugs. The third, fourth, fifth, and sixth sets of
simulations (Figs. 3 and 4; Runs 7 through 9, 10 through 13, 14
through 17, and 18 through 25) indicated that recovery can signifi-
cantly decrease as a result of the CO 2 solubility in the aqueous
phase. Unlike the continuous injection scenario, the size of the
CO 2 slug (18% PV total) was small, making losses to the brine
more significant. Further, the amount of brine in the reservoir during
the displacement was greater than that originally present in the reser-
voir because of the injected brine.
For each of the four values of initial oil saturation (Soi =0.8,0.5,
0.3, and 0.2), the modeling of CO 2 solubility showed a decrease
in ultimate recovery. This result was caused by the diminishing of
the slug size by the CO 2 solubilization into the brine.
The most severe reduction in oil recovery occurred in the WAG
simulation (Run 25 on Fig. 4). A simulation of this same WAG
process without considering CO 2 solubility yielded a recovery
curve similar to Run 18 (see Table 1 and Fig. 4). The amount of
255
 TABLE 1-SUMMARY OF SIMULATIONS

TDS(%) CO 2
Initial In·Place Injected Solubility CPU
Simulation Type ~ Brine Brine Modeled (% change)
--- ---
1
2
Continuous
Continuous
0.8
0.8
a
15
a
15
No
Yes
-
+3
3 Continuous 0.8 a a Yes +4
4 Continuous 0.2 a a No -
5 Continuous 0.2 15 15 Yes +17
6 Continuous 0.2 a a Yes +92
7
8
Slug 18% PV
Slug 18% PV
0.8
0.8
a
15
a
15
No
Yes
-
-17
9 Slug 18% PV 0.8 a a Yes +13
10 Slug 18% PV 0.5 a a No -
11 Slug 18% PV 0.5 15 15 Yes +7
12 Slug 18% PV 0.5 10 5 Yes -24
13 Slug 18% PV 0.5 a a Yes -21
14
15
Slug 18% PV
Slug 18% PV
0.3
0.3
0
10
0
5
No
Yes
-
-25
16 Slug 18% PV 0.3 5 5 Yes -27
17 Slug 18% PV 0.3 0 0 Yes -16
18 Slug 18% PV 0.2 0 0 No -
19 Slug 18% PV 0.2 25 25 Yes +12
20 Slug 18% PV 0.2 20 20 Yes +51
21 Slug 18% PV 0.2 15 15 Yes +16
22 Slug 18% PV 0.2 10 10 Yes + 11
23 Slug 18% PV 0.2 5 5 Yes +40
24 Slug 18% PV 0.2 0 0 Yes +33
25 WAG 18% PV 0.2 0 0 Yes +33

CO 2 introduced into the reservoir was exactly the same as the CO 2
slug simulation (Run 24 on Fig. 4). Because water was injected
in 20-day intervals alternating with four 20-day slugs of CO 2 , more
water was present in the reservoir, and therefore in contact with
the CO 2 slugs, during the displacement process.
Effect of IDS Concentration. In all cases that accounted for CO 2
solubility in brine, an increase in TDS concentration led to an in-
crease in oil recovery. These recovery curves approach a limit,
which is the recovery curve corresponding to the simulation in which
CO 2 solubility in brine was ignored.
Benefit of a IDS Material Balance. The inclusion of the dissolved
solids material balance permitted the simulation of processes in
which the TDS of the dissolved brine differed from that of the in-
100
place brine (Runs 12 and 15 in Table 1) or differed from gridblock
to gridblock. Fig. 5 illustrates the phase saturations, the molar con-
centration of TDS in the aqueous phase, and the mole fraction of
CO 2 in the aqueous phase at 200, 300, and 600 days for Run 12
in Table 1. In this example, the initial oil and brine saturations were
both 50%. The TDS concentration of the brine was 10 wt% (3.31
mol%). An 18% PV slug of CO 2 was injected over an 80-day in-
terval. A 5-wt% (1.59-mol%) TDS brine drive was then continu-
ously injected. After 200 days, the CO 2 slug (gas phase) was
almost gone. The oil had been swept out of the first six gridblocks.
The TDS concentration in the aqueous phase had an S-shaped ap-
pearance because the concentrations in the first gridblocks were
at the drive brine levels, while the TDS concentration in the last
gridblocks were at the initial levels and the intermediate gridblocks
contained a mixture of the brines. Between 200 and 300 days, the
gas phase disappeared. At 300 days, the oil in Gridblock 7 had
reached a level below the residual saturation to water because the
drive brine had "extracted" some of the CO 2 out of the hydrocar-
bon phase, thus shrinking the immobile oil phase. The CO 2 con-
centration in the aqueous phase of the first four gridblocks was about

80
100 -r----------------,
18% PV Slug of C02. TOSln Brine

'0 Brine Drive Inj. In-situ

Q)
0 Initial Oil Saturation = .8 No Solubility
::J 80 15% 15%

~
'0 60 Runs 1·3 0% 0%
0
a: .,
"0

-
U No Solubility
(5 ::J
"0 60
Continuous 0
5% 10%
c:
Q) 40 Injection of C02 .t 5% 5%
0 0% 0%
(5
iii - N o C02 Solubilily
0... --C02 In 15% TDS Brine .,u
C 40
-----C02 In H20
iii
D.
20
20

01L-----r~~_r----_r----_.----_.----_1

o 100 200 300 400 500 600 100 200 300 400 500 600
Time (days) Time (days)

Fig. 2-011 recovery resulting from continuous Injection of Fig. 3-011 recovery resulting from Injection of CO 2 slugs,
CO 2 • Initial 011 saturation = 0.8 and 0.3.

256 SPE Reservoir Engineering, May 1992

 100 s YC02 ' B Yn e .. 2. B
TOSlnBrlne
~~IIiiT J""

....
Inl· In-sflu .05

18% PY Slug of CO2. .04

80 Brine Drive

-0
No$olubllity
15%
5%
0""
15%
10*/.
0%
...
.7

.03

.
Q)
0
::>
-0 60 .02
0 No$olubilily .3
Jt 25*;" .2 .01
25*1.
(5 20% 20'"
15". 15*/0 .1
C 40 1~. 10%
o
5%
8~
Q)
0
Q; ~ WAG 0%
!l.
~~mr-
.....
.05
JIll
20 JIll .04

......
.7 JII.
Brine
JIIO .03
JIIO
. 100 .-

--
200 300 400 500 600
.-
Time (days)
..
.3
.01
Fig. 4-011 recovery resulting from Injection of CO 2 slugs, .1

Initial 011 saturation = 0.5 and 0.2. 0 o

.05
zero because the injected brine contained no CO 2 , At 300 days, .... B.lne .04
the CO 2 concentration in the aqueous phase of Gridblock 20 be-
....
.7
came greater than zero. This indicated that the brine had traveled .03
through the reservoir, carrying dissolved CO 2 to the production
well. ..
.3~~~~~~~'- __~__4 -_ _'-----~''-
.02
Between 300 and 600 days, Gridblocks 8 and 9 also reached satu-
.2 .01
rations below residual oil because the CO 2 was extracted from the
.1
oil phase into the brine. The solids concentration throughout all o
20 gridblocks approached that of the injection brine. The CO 2 con- t 2 3 • S • 7 • • 10 11 12 15 U 15 ,. 17 11 ,. 20

centration profile in the aqueous phase remained relatively station- 1njec1lon Grid Blodcs Production

ary, however. The drive continued to extract CO 2 out of the s •

immobile hydrocarbon phases while leaving the insoluble hydrocar- YC02' B
bons behind. The mobile hydrocarbon phases that were ahead of Yn c .. 2. B •
these gridblocks were being displaced by carbonated water. This
car)Jonated water mixed with the carbonated brine that was already Fig. 5-Phase saturation, TOS concentration, and CO 2
present in the gridblocks and displaced the oil toward the produc- aqueous-phase concentration profiles for Simulation 12.
tion well.

Effect of TDS Material Balance and CO 2 Solubility on Com- 5. The changes in aqueous-phase density and viscosity caused
putational Time. Table 1 lists the percentage change in central proc- by dissolved solids were not significant in these ID simulations.
essing unit (CPU) time for each simulation that incorporated CO 2 These changes should be expected to produce more significant ef-
solubility in the aqueous phase relative to the CPU time for the same fects in 3D simulations where mobility becomes important.
simulation without considering CO 2 solubility in brine. For most 6. Both increases and decreases in computational time require-
cases, the CPU time requirement increased when CO 2 solubility ments were observed when the CO 2 solubility in brine was incor-
was modeled. However, the convergence was apparently stabilized porated into the solution algorithm for the simulations. The largest
for some cases, evidenced by decreases in the amount of CPU time increases (up to 92 %) were observed in the simulations in which
when CO 2 solubility was modeled. The reasons for this behavior the initial water saturation was at its highest values.
were not investigated in this study.
Nomenclature
Conclusions eros = total dissolved solids concentration excluding
1. A ID IMPESC compositional simulator was successfully modi- dissolved gases, wt%
fied to incorporate the effects of CO 2 solubility in brine. Solids D = depth to reservoir (measured positive down), ft [m]
tracking was implemented by introducing an additional component F = driving force,=(L1pg+l_M'~oj -gPMJ)
material balance for the dissolved solids. gj = gradient of Phase j
2. Very small changes in oil recovery occur as a result of CO 2 H* = reference Henry's law constant for CO 2 in water,
solubility in brine in the continuous displacement of oil by CO 2 psia [kPa]
because the aqueous phase becomes saturated with CO 2 and the M = molecular weight
CO 2 that is "lost" to the aqueous phase is replenished by the con- nc = number of hydrocarbon components
tinuous injection of CO 2 into the reservoir.
Paj = absolute pressure of Phase j, psia [kPa]
3. Large decreases in oil recovery can result when a slug of CO 2
P cob = oillbrine capillary pressure, psia [kPa]
is used to displace the hydrocarbons. The CO 2 that is "lost" to
P cog = oil/gas capillary pressure, psia [kPa]
the brine can represent a significant fraction of the total amount
of CO 2 injected into the reservoir. Therefore, the ultimate recov- qj = injection/production rate of Phase j, mollD
ery can be substantially less than that calculated when CO 2 solu- Rij = solubility of Component i in Phase j (weight basis)
bility in brine is ignored. Sj = saturation of Phase j
4. These solubility effects become more pronounced when the Ilt = timestep, days
CO 2 slug size decreases, the initial water saturation increases, the T = temperature, OR [K]
amount of water injected alternately with the CO 2 increases, and Vb = gridblock volume, Ilxllyllz, ft3 [m 3]
the brine salinity decreases. y ij = mole fraction of Component i in Phase j

SPE Reservoir Engineering, May 1992 257

 Reference.
Authors
I. Enick, R.M. and Klara, S.M.: "C02 Solubility in Water and Brine
Robert M. Enlck Under Reservoir Conditions," Chern. Eng. Communication (1990) 90,
Is an associate 23-33.
professor In the 2. Krichevsky, I.R. and Kasarnovsky, I.S.: "Thermodynamic Calcula-
Chemical & Petro· tion of Solubilities of Nitrogen and Hydrogen at High Pressures," J.
leum Engineering Am. Chern. Soc. (1935) 57, 2168.
Dept. at the U. of 3. Lyckrnan, E.W., Eckert, C.A., and Prausnitz, I.M.: "Generalized
Pittsburgh. He also Reference Fugacities for Phase Equilibrium Thermodynamics, " Chern.
Is a part·tlme Oak Eng. Sci. (1965) 20, 685-91.
Ridge Fellow at the 4. Peng, D.Y. and Robinson, D.B.: "Two- and Three-Phase Equilibri-
U.S. DOE Morgan· um Calculations for Systems Containing Water," Cdn. J. Chern. Eng.
town Energy Tech· (1~76) 54, 595-99.
Enlck Klara nology Center. His 5. Klins, M.A.: Carbon Dioxide Flooding, Basic Mechanisms, and Proj-
Interests Include ect Design, Inti. Human Resources Development Corp., Boston (1984).
supercrltlcal fluid technology and fluid flow In porous media. 6. International Critical Tables, E.W. Washburn (ed.), McGraw-Hill Book
He holds a PhD degree In chemical engineering from the U. Co. Inc., New York City (1928) 3.
of Pittsburgh. Scott M. Klara Is an engineer with the U.S. 7. Fabuss, B.M.: "Progress Report," Office of Saline Water (1965) 30,
DOE Morgantown Energy Technology Center In Morgantown, No. 136.
WV, where he has worked since 1986. His current research 8. Sayegh, S.G. and Najman, I.: "Phase Behavior Measurements of
Interests Include thermodynamic phase behavior, reservoir C02-S02-Brine Mixtures," Cdn. J. of Chern. Eng. (April 1987) 65,
and process simulation, and modeling horizontal wellbores. 314-20.
He holds BS and MS degrees in chemical engineering and an 9. Rowe, A.M. and Chou, I.C.S.: "Pressure-Volume-Temperature-
MS degree In petroleum engineering from the U. of Pittsburgh. Concentration Relation of Aqueous NaCI Solutions," J. Chern. Eng.
Klara has recently transferred to the U.S. DOE Pittsburgh Data (1970) 15, 61-66.
Energy Technology Center. 10. Burcik, E.I.: Properties of Petroleum Reservoir Fluids, Iohn Wiley &
Sons Inc., New York City (1957).
11. Nghiem, L.X., Fong, D.K., and Aziz, K.: "Compositional Modeling
Zi = feed (global) composition of Component i excluding With an Equation of State," SPEl (Dec. 1981) 687-98.
aqueous phase 12. Nghiem, L.X.: "A New Approach to Quasi-Newton Method With Ap-
(X = heaviest compound exhibiting solubility in aqueous plication to Compositional Modeling," paper SPE 12242 presented at
phase the 1983 SPE Symposium on Reservoir Simulation, San Francisco, Nov.
(X) = term defined in Eq. 12 16-18.
.:l = difference operator 13. Mansoori, I.: "Compositional Modeling of C02 Flooding and the Ef-
fect of C02 Water Solubility," paper SPE 11438 available at SPE,
r= transmissibility Richardson, TX.
e = weighting factor defined in Eq. 11 14. Li, Y.-K. and Nghiem, L.X.: "Phase Equilibria of Oil, Gas, and
p. = viscosity, cp [mPa's] Water/Brine Mixtures From a Cubic Equation of State and Henry's
Pmj = mass density of Phase j Law," ~in. J. of Chern. Eng. (June 1986) 64, 486-96.
P Mj = molar density of Phase j 15. Fishlock, T.P. and Rodwell, W.R.: "Improvements in the Numerical
Simulation of Carbon Dioxide Displacement," paper presented at the
t/> = porosity
1982 European Symposium on Enhanced Oil Recovery, Paris, Nov.
8-10.
Subscripts 16. Pollack, N.R. et al.: "The Effect of an Aqueous Phase on the
b = brine or block in Vb C02/Tetradecane and C02/Maljammar Crude Oil Systems," SPERE
g = gas (May 1988) 533-41.
h = hydrocarbon 17. Coats, K.H.: "An Equation of State Compositional Model," SPEl (Oct.
1980) 363-76.
i = component index
18. Thele, K.I., Lake, L.W., andSepehrnoori, K.: "A Comparison of Three
j = phase index Equation of State Compositional Simulators," paper SPE 12245 present-
nc + 1 = water phase ed at the 1983 SPE Symposium on Reservoir Simulation, San Francis-
nc +2 = dissolved solids phase co, Nov. 16-18.
0= oil
SPERE
w = water
Original SPE manuscript received for review Nov. 16, 1989. Revised manuscript received
June 3, 1991. Paper (SPE 20278) accepted for publication Aug. 12, 1991.
Superscripts
L = previous iteration level within timestep
L +1 = current iteration level within timestep
n = old timestep
n +1 = new timestep

258 SPE Reservoir Engineering, May 1992