221530321 Experimental Study

Experimental investigation of miscible and immiscible
Water-Alternating-Gas (WAG) process performance
Madhav M. Kulkarni, Dandina N. Rao
*
The Craft and Hawkins Department of Petroleum Engineering, Louisiana State University, 3516 CEBA Bldg.,
Baton Rouge, LA 70803, United States
Received 18 March 2004; received in revised form 24 April 2005; accepted 2 May 2005
Abstract
Gas injection is the second largest enhanced oil recovery process, next only to thermal processes used in heavy oil fields. To
increase the extent of the reservoir contacted by the injected gas, the gas is generally injected intermittently with water. This
mode of injection, called water-alternating-gas (WAG), is being widely practiced in the oil fields.
This experimental study is aimed at evaluating the performance of the WAG process as a function of gas–oil miscibility and
brine composition. This performance evaluation has been carried out by comparing oil recoveries from WAG injection with
those from continuous gas injection (CGI).
Miscible floods were conducted at 17.24 MPa (2500 psi) and immiscible floods at 3.45 MPa (500 psi) using rock–fluids
systems consisting of Berea cores, n-Decane and two different brines, namely the commonly used 5% NaCl solution and
another being the multi-component reservoir brine from the Yates field in West Texas. The coreflood protocol consisted of a
series of steps including brine saturation, absolute permeability determination, flooding with oil (drainage) to initial oil
saturation, end-point oil permeability determination, flooding with brine (imbibition) to residual oil saturation, end-point
water permeability determination, and finally, tertiary gas injection (in both continuous injection and WAG modes) to recover
the waterflood residual oil.
When oil recovery per unit volume of gas injection was used as a parameter to evaluate the floods, the WAG mode of injection
out-performed the CGI. As expected the miscible floods were found to out-performthe immiscible floods. At increased volumes of
gas injection, the performance of miscible CGI flood, in spite of the high injection pressure, approached that of the low-pressure
immiscible floods. Achange in brine composition from5%NaCl to 0.926%multivalent brine fromYates reservoir showed a slight
adverse effect on tertiary gas flood recovery due to increased solubility of CO
2
in the latter. The results of this study suggest that the
optimum mode of gas injection is a combination process between CGI and WAG modes of gas injection.
D 2005 Elsevier B.V. All rights reserved.
Keywords: Miscibility; CO
2
; Gas injection; WAG; MMP; Brine composition; Waterflood; Tertiary gas injection; Mobility control
0920-4105/$ - see front matter D 2005 Elsevier B.V. All rights reserved.
doi:10.1016/j.petrol.2005.05.001
* Corresponding author. Tel.: +1 225 578 6037; fax: +1 225 578 6039.
E-mail address: [email protected] (D.N. Rao).
Journal of Petroleum Science and Engineering 48 (2005) 1–20
www.elsevier.com/locate/petrol
1. Introduction
1.1. Background
The gas injection EOR processes today help re-
cover a substantial portion of the oil from light oil
reservoirs (Moritis, 2002). The residual oil satura-
tions in gas swept zones have been found to be quite
low, however, the volumetric sweep in gas floods
has always been a cause for concern (Hinderaker et
al., 1996). The mobility ratio, which controls the
volumetric sweep, between the injected gas and dis-
placed oil bank in gas processes, is typically highly
unfavorable due to the relatively low viscosity of the
injected gas phase. This makes mobility, and conse-
quently flood profile control the biggest concerns for
the successful application of the gas injection EOR
process.
1.2. Profile control
Continuous research efforts are being made to im-
prove the flood profile control in gas floods (McKean
et al., 1999, Enick et al., 2000). These include prepa-
ration of direct thickeners with gas-soluble chemicals
such as Telechelic Disulfate, Polyflouroacrylate and
Flouroacrylate–Styrene copolymers, which can in-
crease the injected gas viscosity several fold (e.g. For
CO
2
viscosity increase from 2–100 fold (Enick et al.,
2000)). Alternative methods, such as modifications in
the injected slug by the use of natural gas liquids
(NGL) for highly viscous oils in depleted, poorly
producing and unconsolidated formations are also pro-
posed (McKean et al., 1999; Moritis, 2002). Although
they seem promising on the laboratory/simulator scale,
important issues like feasibility, cost, applicability,
safety and environmental impact still need to be
addressed (Moritis, 1995). Most of these process mod-
ifications are still at inception or experimental stage
and are not accepted as part of the current commercial
technology.
1.3. Water-Alternating-Gas (WAG) process
The development of the Water-Alternating-Gas
(WAG) process was aimed at improving flood profile
control. The higher microscopic displacement effi-
ciency of gas combined with the better macroscopic
sweep efficiency of water significantly increases the
incremental oil production over a plain waterflood.
Nearly all the commercial gas injection projects today
employ the WAG method. However, previous re-
search and field applications have repeatedly shown
the inadequacy of the WAG process, yet it has
remained the default process due to the absence of a
viable alternative. The low recoveries (in the range of
5%–10% OOIP) in WAG field applications have led
to substantial research of the process, resulting in
better understanding of the poor injectivity limitation
and optimization of WAG ratios (Christensen et al.,
1998). In spite of these improvements, the field per-
formance of WAG process (5%–10% OOIP recovery)
is disappointing (Rao, 2001).
In the United States, most of the WAG applications
are onshore, employing a wide variety of injection
gases for a wide range of reservoir characteristics in
the miscible mode. Although many types of injectant
gases have been attempted in the past, CO
2
and
hydrocarbon gases constitute the major share of injec-
tants (~90%). CO
2
is ideally suited for use in gas
injection projects in the U.S. scenario. Abundance of
reserves of almost pure CO
2
and availability of tech-
nical know-how has been the cause for the growth of
CO
2
injection processes in the U.S. Carbon seques-
tration is now an added advantage of the CO
2
injec-
tion projects.
The main design parameters that need to be eval-
uated on a laboratory scale to evaluate the feasibility
of the process are: reservoir heterogeneity, rock type,
fluid characteristics, injection gas, WAG ratio and
gravity considerations. Other parameters that are im-
portant for gas injection, and tertiary recovery in
general, are those of miscibility development and
composition of oil and brine (Kulkarni, 2003).
1.4. Brine composition effects
The migration of small solid materials (dfinesT)
within porous media has long been recognized as a
source of potentially severe permeability impairment
in reservoirs (Eng et al., 1993). One of the primary
factors that determine the migration of clay particles is
the brine composition. Laboratory studies by Eng et
al. (1993) have shown that brine salinity, composition
and pH can have a large effect on the microscopic
displacement efficiency of oil recovery by waterflood-
M.M. Kulkarni, D.N. Rao / Journal of Petroleum Science and Engineering 48 (2005) 1–20 2
ing and imbibition. Although the literature is in agree-
ment about the effects of brine composition on per-
meability reduction and fines migration, there seems
to be little consensus about the effects of brine com-
position on oil recovery (either by waterflooding or
imbibition). Kwan et al. (1989), in their study of
permeability damage via fines migration in extracted
core material, concluded that permeability and oil
recovery were nearly independent of brine composi-
tion. Contrarily, other experimental studies (Jones,
1964; Khilar et al., 1990; Tang and Morrow, 1999;
Filoco and Sharma, 1998), suggested that changes in
brine composition could have a large effect on oil
recovery. To evaluate the effects of brine composition
of clay stability and oil recovery characteristics, two
sets of brines, namely, the commonly used 5% NaCl
solution and Yates reservoir brine were used in this
experimental study.
2. Experimental details
This experimental study is directed towards the
laboratory evaluation of the WAG process perfor-
mance as a function of several variables including:
(i) the effects of brine composition, and (ii) the effect
of miscibility of injected gas with oil and (iii) the
mode of operation, namely Continuous Gas Injection
(CGI) and WAG.
Core-flooding experiments were conducted in 1-ft
long Berea sandstone cores, using n-Decane as the
oleic phase and brines of two different compositions
(5% NaCl brine and Yates reservoir brine) as aqueous
phases along with pure CO
2
as the injectant gas.
2.1. Materials
Analytical grade reagents were used in all the
experiments. n-Decane and the salts that were used
for synthetic Yates brine preparation were from Fisher
Scientific with a purity of 99.9%. To prepare the
brines, deionized water from the Water Quality Lab-
oratory at Louisiana State University was used. The
compositions of the two brines used in the tests are
shown in Table 1. The Berea sandstone (Liver Rock
type) used in the experiments was from Cleveland
Quarries, Ohio. Pure CO
2
was used as the injection
gas in all the floods.
2.2. Experimental design and assembly of coreflood
apparatus
The minimum miscibility pressure (MMP) for
CO
2
/n-Decane was estimated to be 1880 psi (as de-
scribed in Section 3.1). Hence, the immiscible floods
were conducted at 3.45 MPa (500 psi) and the misci-
ble floods at 17.24 MPa (2500 psi). Fig. 1 shows the
flowchart of the experimental design.
The high-pressure coreflood apparatus was set up to
conduct unsteady state coreflood experiments. The
schematic of the apparatus is shown in Fig. 2. It
consists of a high-pressure Ruska pump injecting
fresh (tap) water at desired flow rate and pressure to
the bottom part of the floating piston transfer vessel.
The transfer vessel is filled with the fluid to be injected
into the core. High-pressure steel tubing (0.32 cm
(0.125 inch) ID) carries the fluid injected into the
core with the assistance of a liquid re-distributor
plate. The produced fluids were carried through the
backpressure regulator into a measuring cylinder/elec-
tronic balance to determine fluids production as a
function of run time. A parallel set of tubing was
constructed to facilitate the circulation of core clean-
up fluids using a centrifugal pump. The inlet, outlet,
differential, back and annulus pressures were mea-
sured using electronic pressure transducers (previously
calibrated against a standard dead-weight tester)
Table 1
Composition of the two brines used in coreflood experiments (for
composition of Yates reservoir brine of pH 7.39, see Vijapurapu and
Rao, 2002)
5% NaCl
brine
Pure NaCl salt (200 gm) in 4 L deionized
water
Yates reservoir
brine
Parameter Concentration
(mg/L)
Total dissolved solids 9200
Calcium 425
Magnesium 224
Potassium 50.5
Sodium 1540
Hardness as CaCO
3
1500
Hardness as carbonate 810
Hardness as non-carbonate 730
Bicarbonate 800
Alkalinity 810
Sulfate 660
Chloride 3700
M.M. Kulkarni, D.N. Rao / Journal of Petroleum Science and Engineering 48 (2005) 1–20 3
A. ∆P Transducer
J. G.C.
P
1
B. Core & Holder Assembly P
2
E. BPR
F. Cleanup
Pump G. Separator /
C. Transfer Graduated Cyl
Vessels Solvents
a b c D. Ruska Pump
I. Readout
H. FieldPoint
®
DAQ
Legend for the above schematic:
: Electrical Lines
: 1/8
,,
High Pressure Piping
: Instrumentation Lines
: Cleanup / Accessories Lines
a: n-Decane Transfer Vessel b: Yates Reservoir Transfer Vessel
c: CO
2
Gas Transfer Vessel
Fig. 2. Horizontal core flooding system schematic.
Core Flood Experiments
Immiscible (500 psi) Miscible (2500 psi)
CGI Floods WAG Floods CGI Floods WAG Floods
5% NaCl + n-Decane Yates + n-Decane 5% NaCl + n-Decane Yates + n-Decane
Fig. 1. Design of coreflood experiments.
M.M. Kulkarni, D.N. Rao / Journal of Petroleum Science and Engineering 48 (2005) 1–20 4
mounted on the coreflood apparatus. This experimen-
tal setup used a fresh 1-ft long Berea core as the porous
medium, which had a single coating of epoxy to
prevent damage during handling and processing of
the core such as end facing, polishing and cutting.
2.3. Experimental procedure
There were two types of experiments performed:
Continuous Gas Injection (CGI) and Water Alternating
Gas (WAG). The displacement tests consisted of the
following steps: saturation with brine, determination of
pore volume and absolute permeability, oil flood to
connate water saturation, end point oil-permeability
measurement, waterflood to residual oil saturation,
end point water-permeability measurement and tertiary
gas flood (either CGI or WAG). The core was filled
with brine solution after core cleaning to determine
pore volume and absolute permeability. It is brought
to connate water saturation by flooding with n-Decane
at high flow rates (160 cc/h). The core is then water
flooded (60 cc/h) using the brine of similar composition
as the connate water to bring the core to waterflood
residual oil saturation. This is the imbibition process
(Berea generally being a water-wet rock), which repre-
sents the secondary recovery process. At the end of the
imbibition process, significant residual oil remains in
the core. Then WAG or CGI tests were conducted in
tertiary mode to recover the waterflood residual oil.
Laboratory transient-state displacement processes
are affected by viscous instabilities and discontinuities
at the inlet and, more importantly, the outlet of the
core, which is referred to as the dend effectT. End
effects can be minimized by using large core lengths
and pore volume. The scaling criterion of Rappaport
and Leas (1953) has been used to remove the depen-
dence of oil recovery on injection rate and core length.
The use of this scaling criterion helps the capillary
pressure gradient in the flow direction to be smaller
than the imposed pressure gradient. The scaling crite-
rion is given by,
L: V:lz1 ð1Þ
Where L is the core length (cm), l is viscosity of
displacing phase (cP) and V is fluid velocity (cm/min).
The Rappaport and Leas (1953) scaling criterion
calculations were repeated for each injection fluid to
ensure stable floods. The Rappaport and Leas (1953)
scaling criterion value of 7.5 was used in all the
corefloods conducted in this study. It is interesting
to note that this criterion is generally met in the
reservoir scale floods due to the large distances be-
tween injector and producer. The step-wise detailed
experimental procedure is provided elsewhere (Kulk-
arni, 2003).
3. Results and discussion
The results are subdivided into two major sections:
calculation of the MMP between CO
2
and n-Decane
and tertiary coreflood results. The coreflood results
are grouped according to fluid systems and further
subdivided between immiscible and miscible displa-
cements. The comparison of WAG and continuous
CO
2
floods are included under the respective mode
of displacement.
3.1. Minimum miscibility pressure (MMP) determina-
tion (CO
2
/n-Decane system)
The commercial PVT simulation packages avail-
able often do not yield predictions accurate enough to
be used without any experimental verification. The
volumetric and phase compositional data for various
CO
2
binary mixtures available from visual cell experi-
ments is scattered and not as useful for phase property
calculations (Orr and Silva, 1983). Hence an effort
was made to compare as well as combine the litera-
ture, simulation and empirical correlation predicted
MMP data for confident characterization of the
MMP between CO
2
and n-Decane.
3.1.1. From the literature
Studies of pressure–composition (P–X) diagrams
(Orr and Jensen, 1984; Orr and Silva, 1983)) as well
as pseudo-ternary diagrams (Benmekki and Mansoori,
1986) show the MMP values between CO
2
/n-Decane
system of 13.0 MPa (1880 psia) and 13.5 MPa (1958
psi) respectively.
3.1.2. From empirical correlations
Empirical correlations such as the Extrapolated
Vapor Pressure (EVP) method (Newittelal Equation),
Petroleum Recovery Institute (PRI) equation, Yellig
and Metcalfe Method and Croquist Method were used
M.M. Kulkarni, D.N. Rao / Journal of Petroleum Science and Engineering 48 (2005) 1–20 5
to calculate the MMP of CO
2
/n-Decane system
(Ahmed, 1997). The results for 301 K (82 8F) and
344 K (160 8F) are summarized in Table 2. All the
empirical correlations predicted an MMP lower than
6.9 MPa (1000 psi). This suggests that experiments
with CO
2
/n-Decane system at pressures higher than
1000 psia would positively develop miscibility.
3.1.3. From commercial PVT simulation package
Due to non-availability of experimental facility or
data at the desired temperature, the WINPROPR PVT
package was used to predict the MMP between CO
2
/n-
Decane at 301 K (82 8F). However, the predictions of
the simulator need be evaluated against compositional
systems with known MMP values. To achieve this, two
systems: one simple and other complex, with experi-
mentally known values of MMP were used to calibrate
the simulator. Then, this calibrated simulator was used
to determine the MMP of CO
2
/n-Decane (at 301 K (82
8F)). The simulation results are summarized in Table 3.
Table 3 indicates that the MMP for the CO
2
/n-
Decane system should not be greater than 12.7 MPa
(1840 psia) (at 301 K (82 8F)) for multiple contact
miscibility (MCM) development. Since this predicted
MMP value is higher than those predicted by empir-
ical correlations, P–X diagrams and pseudo-ternary
diagrams, coreflood experiments were conducted at
17.24 MPa (2500 psi) and 301 K (82 8F) to ensure
miscible displacements.
3.2. Coreflood results
The objective of the coreflood experiments was to
determine: (i) the effect of mode of tertiary gas injec-
tion (CGI or WAG), (ii) effect of miscibility and (iii)
brine composition on oil recovery in fresh Berea
cores, without any previous history of exposure to
crude oil. Furthermore, to eliminate the effects of
rock heterogeneity, all the core tests were conducted
on the same core using a non-reactive n-Decane as the
oil phase and adopting a thorough cleaning procedure
in between the various displacements.
Also to facilitate ease of translation of these labo-
ratory coreflood results, dimensional analysis (Kulk-
arni, 2004) on seventeen commercial gas injection
projects was conducted, and the governing dimension-
less groups were identified. The coreflood experimen-
tal design was then dscaledT to duplicate the values of
the dimensionless groups prevalent in the field injec-
tions. This assured the duplication of the controlling
multiphase mechanisms and fluid dynamics into the
laboratory.
3.2.1. Oil flood (drainage)
This step constitutes the process of injection of n-
Decane into the core that is initially saturated with
brine to drive the core to connate water saturation.
The relative permeability of the core to oil at the end
of this step is an important wettability identification
Table 2
Summary of all the MMP values (in psia) obtained from empirical
correlations (Ahmed, 1997) and available experimental values (Orr
and Jensen, 1984; Orr and Silva, 1983; Benmekki and Mansoori,
1986)
Empirical correlation 82 8F 160 8F Experimental
(160 8F)
Newittelal 994.2 2309.6 1880 psia
PRI 998.9 2249.7 –
Yelling and Metcalfe 871.8 2005.3 1880 psia
Croqist 814.9 1479.5 –
Table 3
Comparison between the experimental and calculated values of MMP between CO
2
gas and given oil composition
Oil composition Reference P–R equation of state SRK equation of state Experimental values Temp
MCM FCM MCM FCM MCM FCM
60% C
10
+40% C
4
Orr et al. 1800 psia 2360 psia 1720 psia 2120 psia 1700 psia 1900 psia 160 8F
Complex STO (C
7
to C
28
) Orr et al. 3160 psia 5000+ psia 3000 psia 5000+ psia IFT method: 2400 psig 122 8F
Slim tube: 2300 psig 122 8F
1% C
4
+99% C
10
– 3320 psia 3440 psia 2720 psia 2840 psia – – 160 8F
0.001% C
4
+99.999% C
10
– 1840 psia 3000+ psia 1760 psia 3000+ psia – – 82 8F
0.001% C
4
+99.999% C
10
– 3360 psia 3520 psia 2760 psia 2880 psia – – 160 8F
0.0% C
4
+100.0% C
10
Orr et al. 3360 psia 3520 psia 2760 psia 2880 psia Orr et al: 1800 psig 160 8F
Benmekki: 1958 psig 160 8F
M.M. Kulkarni, D.N. Rao / Journal of Petroleum Science and Engineering 48 (2005) 1–20 6
parameter. The results for this step for all corefloods
conducted are summarized in Part (A) in Tables 4, 5
and 6.
High oil saturations consequently lower connate
water saturations (10.5% to 21.3%) were characteris-
tic of the oil floods for both the brines used. Specif-
ically, low relative oil permeabilities (34.5% to
69.1%) at the end of the oil flood were observed for
mono-valent (5% NaCl solution) brine (Table 4 —
Part (A)).
Berea cores are known to be strongly water-wet,
while displaying wide variations in the displacement
Table 4
Coreflood results for 5% NaCl brine+n-Decane+Berea core system
System: 5% NaCl Brine+n-Decane+Berea core P
TEST
(psi) Abs. perm (D) S
WC
S
OI
End point rel-perms
(A) Drainage (n-Decane) step
Experiment # 1 500 0.2526 12.5 87.5% 34.5%
Experiment # 2 500 0.3435 21.3 78.7% 39.9%
Experiment # 3 2500 0.2895 13.3 86.7% 42.0%
Experiment # 4 2500 0.1825 15.1 84.9% 47.0%
(B) Imbibition (5% NaCl brine) step
Experiment title P
TEST
(psi) S
OR
S
W
Recovery %OOIP End point rel-perms
Experiment # 1 500 35.0 65.0 60.0% 08.01%
Experiment # 2 500 27.7 72.3 64.8% 08.09%
Experiment # 3 2500 32.8 67.2 62.2% 08.05%
Experiment # 4 2500 35.4 64.7 58.1% 08.72%
(C) Tertiary gas (EOR) step
Experiment title P
TEST
(psi) S
L
S
G
Rvry (cc) Recovery %OOIP Utilz. ftr. (MCF/bbl)
Experiment # 1 (CGI – immiscible) 500 47.9 52.1 10.5 8.8% 7.5
Experiment # 2 (WAG – immiscible) 500 – – 9 8.3% 4.5
Experiment # 3 (CGI – miscible) 2500 26.4 73.6 43.5 36.6% 20.2
Experiment # 4 (WAG – miscible) 2500 – – 41 35.0% 9.0
Table 5
Coreflood results for Yates reservoir brine+n-Decane+Berea core system
System: Yates reservoir brine+n-Decane+Berea core P
TEST
(psi) Abs. perm (D) S
WC
S
OI
End point rel-perms
(A) Drainage (n-Decane) step
Experiment # 7 500 0.1311 21.3 78.7 65.5%
Experiment # 8 500 0.1869 19.1 80.9 58.3%
Experiment # 9 2500 0.1443 18.4 81.6 59.1%
Experiment # 10 2500 0.1906 16.9 83.1 66.8%
(B) Imbibition (Yates reservoir brine) step
Experiment title P
TEST
(psi) S
OR
S
W
Recovery %OOIP End point rel-perms
Experiment # 7 500 25.5 74.5 67.6% 11.80%
Experiment # 8 500 27.7 72.3 65.8% 07.51%
Experiment # 9 2500 29.9 70.1 63.4% 11.56%
Experiment # 10 2500 27.0 73.0 64.9% 09.39%
(C) Tertiary Gas (EOR) step
Experiment title P
TEST
(psi) S
L
S
G
Rvry (cc) Recovery %OOIP Utilz. ftr. (MCF/bbl)
Experiment # 7 (CGI – immiscible) 500 27.8 72.2 22 20.4% 4.7
Experiment # 8 (WAG – immiscible) 500 – – 11 9.9% 3.1
Experiment # 9 (CGI – miscible) 2500 19.8 80.2 40 35.7% 19.4
Experiment # 10 (WAG – miscible) 2500 – – 29 25.4% 12.9
M.M. Kulkarni, D.N. Rao / Journal of Petroleum Science and Engineering 48 (2005) 1–20 7
characteristics if the clays are not stabilized using
divalent cations such as [Ca
++
] and [Mg
++
] (Jones,
1964; Khilar et al., 1990; Tang and Morrow, 1999;
Filoco and Sharma, 1998; Hill, 1982). The low end
point oil relative permeabilities and lower connate
water saturations observed in the present study indi-
cate weakly water-wet to intermediate-wet character-
istics as suggested by Craig’s rules of thumb (Ayirala,
2002). The relatively larger variations in the absolute
permeability values with mono-valent brine, suggest
the unstable nature of the system, in spite of the
consistent and rigorous cleaning procedure practiced.
To investigate the effects of brine composition,
another series of experiments using the same core
but with Yates reservoir brine, containing divalent
cations, was designed. The mono-valent brine used
in the previous experiments was replaced with a mul-
tivalent (Yates reservoir) brine to investigate the phe-
nomena of clay stabilization and its effects on
dynamic displacement corefloods. Since the core-
floods were conducted on the same core, the cleanup
was done using the 5% brine initially and then this
brine was miscibly displaced by Yates reservoir mul-
tivalent brine. This was done to ensure that the core
was subjected to the same history as the other tests.
The oil steps conducted with multivalent brine
(Yates brine with n-Decane in a Berea core (Tables
5 and 6 — Part (A)), experiment # 7–12) showed
significant increase in the end point oil permeabilities
(from 40.85% (avg.) to 62.84% (avg.)) as well as the
connate water saturations (10.5% to 21.3%: Tables 4,
5 and 6: Column 4) compared to those with 5% NaCl
brine. Moreover, these values closely agree to similar
experiments conducted by Owens and Archer (1971).
A significant reduction in the standard deviation of the
absolute permeability (Tables 4, 5 and 6: Column 3)
indicates the stabilization of core clays and shift of the
system from weakly water-wet (5% NaCl brine runs)
to more water-wet characteristics.
3.2.2. Brine flood (imbibition)
This step constitutes the process of brine injection
into the core, at connate water saturation, to achieve
waterflood residual oil saturation in the core. Brine is
injected at stable flow rates into the core after primary
drainage step. The results of this step can be an
indicator of the extent of feasible secondary oil recov-
ery. The end-point permeability of the rock to brine
can also be used to infer wettability. The oil recovery,
residual oil saturations, and end-point water relative
permeabilities for the waterfloods conducted are sum-
marized in the Tables 4, 5 and 6 (Part (B)).
The imbibition results are typical of water-wet cases
for both the brine systems (namely 5% NaCl brine and
Yates reservoir brine) used for these tests. Excellent
agreements of recoveries between all the imbibition
Table 6
Coreflood results for Yates reservoir brine+n-Decane+Berea core system using CO
2
saturated Yates reservoir brine for specified steps
System: Yates reservoir brine+n-Decane+
Berea core
P
TEST
(psi) Abs. perm (D) S
WC
S
OI
End point rel-perms
(A) Drainage (n-Decane) step
Experiment # 11 500 0.4503 10.5 89.5 69.07%
Experiment # 12 2500 0.1361 13.3 86.7 58.25%
(B) Imbibition (Yates reservoir brine) step
Experiment title P
TEST
(psi) S
OR
S
W
Recovery %OOIP End point rel-perms
Experiment # 11 (Yates reservoir brine
saturated with CO
2
gas flood)
500 20.6% 79.4% 76.92% 9.64%
Experiment # 12 (unsaturated Yates
reservoir brine flood)
2500 27.0% 73.0% 68.91% 10.26%
(C) Tertiary gas (EOR) step
Experiment title P
TEST
(psi) S
L
S
G
Recovery (%OOIP) Utilz. ftr. (MCF/bbl)
Experiment # 11 (CGI – immiscible) 500 40.7% 59.3% 5 cc (4.80% OOIP) 2.5
Experiment # 12 (WAG – miscible —
Yates reservoir brine saturated with
CO
2
gas alternating with CO
2
flood)
2500 – – 33 cc (27.7% OOIP) 11.2
M.M. Kulkarni, D.N. Rao / Journal of Petroleum Science and Engineering 48 (2005) 1–20 8
displacements were found (Tables 4, 5 and 6 (Part (B):
Column 5)). Higher waterflood oil recoveries, lowend-
point water permeabilities, a sharp breakthrough with
negligible oil production thereafter, all the characteris-
tics of a water-wet rock, were exhibited in the step.
3.2.3. Tertiary gas injection flood (EOR)
The oil recovery, residual oil saturations, and CO
2
utilization factor data for all the corefloods conducted
for both the brines used are provided in Tables 4, 5
and 6 (Part (C)). To facilitate the evaluation of all the
corefloods conducted at various conditions and tertia-
ry recovery modes, require a common parameter for
flood performance evaluation. Two new factors were
defined, as below, to compare the relative merits of all
the corefloods conducted: dCO
2
utilization factor
(UF
CO2
)T and dTertiary Recovery Factor (TRF)T. The
itemized and detailed discussion of the tertiary gas
injection core flood results is provided in Section 4 of
this paper.
3.2.3.1. CO
2
utilization factor. CO
2
utilization factor
(UF
CO2
) is commonly used to evaluate field projects
and is defined as the volume of CO
2
gas injected
under standard conditions, to produce a barrel of oil,
and is calculated as:
UF
CO2
¼
V
CO2
MSCF ð Þ
Q
Oil
Bbl ð Þ
: ð2Þ
3.2.3.2. Tertiary recovery factor (ROIP/PVI-
CO
2
). CGI and WAG mode injections resulted in
an unequal quantity of cumulative gas injections for
each flood. To dnormalizeT the recoveries and avoid
fallacious conclusions from the data, the ROIP/PV-
CO
2
factor was used which is defined below,
Dimensionless TRF
¼ Oil recovered cc ð Þ= S
OR
to WF ð Þ cc ð Þ ½ Š
= Cum: PV CO
2
Injected ½ Š: ð3Þ
The use of these two factors in the analysis was
found to be more appropriate as shown in the follow-
ing comparisons. However, the standard plots, such as
recovery versus Pore Volume Injected (PVI) are also
included for easy cross-reference. All the detailed
results of the tertiary gas injection floods are available
elsewhere (Kulkarni, 2003).
4. Discussion of tertiary gas injection corefloods
The discussions of the coreflood results are item-
ized as: (i) the effect of miscibility development, (ii)
effect of brine composition, (iii) the mode of operation
and (iv) effect of using CO
2
-saturated brine.
4.1. Effect of miscibility
Miscibility affects the microscopic displacement
efficiency in the gas injection EOR process. It influ-
ences the capillary number via the interfacial tension.
A zero interfacial tension value is necessary and
sufficient for attainment of miscibility. This results
in very high capillary numbers and consequently
near perfect microscopic displacement efficiency.
Hence, the development of miscibility is beneficial
from a recovery point of view. Results of the miscible
and immiscible core floods conducted using both
brines are discussed below.
4.1.1. 5% NaCl brine+n-Decane+Berea core
The coreflood results for this system are shown in
two plots: one in the conventional percent residual oil
in place (%ROIP) plot (Fig. 3(a)) and the other the
tertiary recovery factor (Eq. (3)) versus pore volume
injection (PVI) (Fig. 3 (b)).
Fig. 3(a) indicates a significant increase in oil
recovery for miscible floods over immiscible floods.
While the immiscible flood recoveries (both CGI and
WAG) were about 23%, the miscible floods yielded
93.7% recovery for the CGI flood and 84.5% for the
WAG flood. The conventional plot suggests that the
continuous injection of CO
2
yields improved perfor-
mance over the WAG injection.
However, this conclusion is misleading because
the amount of CO
2
injected for the given recoveries
are significantly different in each of these core-
floods. The total recovery obtained from 1: 1
WAG is from one-half the volume of CO
2
gas
injected in the CGI process. Hence, the analysis of
the results on the basis of recovery alone leads to
questionable comparisons. For this reason, the re-
coveries were dnormalizedT on the basis of water-
flood residual oil recovered per pore volume (PV) of
CO
2
gas injected to arrive at the Tertiary Recovery
Factor (TRF) defined earlier. This factor is plotted in
Fig. 3(b).
M.M. Kulkarni, D.N. Rao / Journal of Petroleum Science and Engineering 48 (2005) 1–20 9
The Fig. 3(b) clearly shows the advantage of
the WAG process. Both, miscible and immiscible
processes hasten recovery. It is important to note
the conclusions from Figs. 3(a) and (b) are contra-
dictory. Thereby, the use of Tertiary Recovery
Factor for evaluation of the corefloods is appropri-
ate and is used in analyzing the rest of the exper-
imental results (along with the conventional
recovery plots).
It is interesting to note that, in Fig. 3(b), the TRF
for experiment # 3 for the CGI miscible flood contin-
uously decreases and approaches the data for immis-
cible flood. This has serious implications, in that what
appeared to be the best-case scenario based on the
conventional recovery plot, turns out to be the worst
case due to the process economics (incremental cost
of compressing CO
2
to pressures above the minimum
miscibility pressure for injection).
Effect of Miscibility (1-ft Berea + 5% NaCl Brine)
0%
20%
40%
60%
80%
100%
0.0 0.5 1.0 1.5 2.0 2.5
Total PV (G + W) Injected
R
e
c
o
v
e
r
y

(
%

R
O
I
P
)
EXPT 1: IMM. CGI - 5%NaCl
EXPT 2: IMM. WAG - 5%NaCl
EXPT 3: MIS. CGI - 5%NaCl
EXPT 4: MIS. WAG - 5%NaCl
(a)
Effect of Miscibility (1-ft Berea + 5% NaCl Brine)
0.0
0.2
0.4
0.6
0.8
1.0
1.2
1.4
1.6
0.0 0.5 1.0 1.5 2.0 2.5
Total PV (G + W) Injected
T
R
F

(
%

R
O
I
P

/

P
V
I

C
O
2
)
EXPT 1: IMM. CGI - 5%NaCl
EXPT 2: IMM. WAG - 5%NaCl
EXPT 3: MIS. CGI - 5%NaCl
EXPT 4: MIS. WAG - 5%NaCl
~ 0.6 PVI
(b)
Fig. 3. Effect of miscibility and mode of injection on tertiary recovery in 1-ft Berea core+5% NaCl Brine+n-Decane (non-reactive) system, (a):
recovery as percent of ROIP, (b): fraction of residual oil in place recovered per pore-volume (PV) of CO
2
injected versus total PV (gas+water)
injected.
M.M. Kulkarni, D.N. Rao / Journal of Petroleum Science and Engineering 48 (2005) 1–20 10
Comparison of the recoveries from Figs. 3(a) and
(b) indicates that the maximum utilization (best per-
formance) of the CGI miscible flood occurs up to 0.7
PVinjection. The tertiary recovery factor comparisons
for continued injection (after 0.7 PV injection) show
distinct advantage of the 1: 1 miscible WAG process.
This suggests an optimum mode of gas injection
consisting of the use of CGI mode of injection until
0.6–0.7 PV injection and later switching over to the
WAG process for maximizing the tertiary recovery.
This seems to be the principle behind the patented
processes of dHybrid WAGT (Huang and Holm, 1986)
and dDUWAGT (Tanner et al., 1992) of UNOCAL and
Shell respectively, where a large slug of CO
2
(~0.7
PV) is injected in the reservoir followed by 1: 1 WAG.
4.1.1.1. Important operational differences between the
optimum process identified by this work and dHybrid-
WAGT/DUWAG. In this experimental work, all CGI
experiments showed a TRF peak after about 0.6–0.8
PV injection, and that the TRF values of CGI floods
till this peak are higher than the respective WAG
floods (Figs. 3(b) and 4(b)). However, after this
peak, the CGI flood performance exponentially dete-
riorates. On the other hand, the WAG employment
prevents this exponential TRF decline (after reaching
a peak TRF value), as seen in (Figs. 3(b), 4(b) and, 6),
indicating improved gas utilization factors in both
miscible and immiscible modes. Therefore to optimize
gas utilization (and therefore flood economics), it is
recommended that gas be injected in CGI mode till
0.7 PV injection (or at the TRF peak), followed by
1: 1 WAG injection.
Conceptually the doptimumT process (the combina-
tion of CGI and WAG) recommended by this work is
similar to the patented Hybrid-WAG and DUWAG
processes implemented in the field previously. How-
ever, there are significant differences between the
already patented processes and the optimum process
suggested by this experimental work, which is eluci-
dated below.
The Hybrid-WAG and DUWAG were mainly the
result of field dependant parameters such as market
conditions (Bellavance, 1996) (namely, reduce the
early peak CO
2
demands, maximize utilization of
recycled CO
2
, minimize manpower requirements and
provide flexibility to accelerate or decelerate project
development), and flooding conditions (Bellavance,
1996; Tanner et al., 1992) (namely WAG implemen-
tation only under the circumstances of premature gas
breakthroughs or bgassing outQ of wells).
Another striking feature of the doptimumT process
described in this paper is that the reservoir heteroge-
neity factor has been effectively eliminated in these
experiments by conducting all the CGI, WAG and
Hybrid-WAG corefloods on one Berea core. This is
not the case in the patented processes. For example, in
the Wasson Denver Unit (Tanner et al., 1992) east–
west anisotropy in the continuous CO
2
pilot area
resulted in bnon-radial flood frontsQ. Although the
initial response of the continuous CO
2
pilot was en-
couraging; the bgassing-outQ of production wells sug-
gested subsequent WAG employment to control
premature gas breakthroughs.
The main difference between the patented process-
es and this doptimumT process is the slug-size. Hybrid-
WAG process calls (Bellavance, 1996) for a 9% pore
volume CGI followed by 21% 1: 1 WAG flood;
whereas the DUWAG process (Tanner et al., 1992)
requires 4–6 years of CGI flood (at the pilot rates of
2–7 MMCF/D) followed by 1: 1 WAG till a 40%
HCPV injection is achieved (although simulation
studies (Tanner et al., 1992) suggest a higher HCPV
injection (~60% PV) for higher recoveries). The
doptimumT process suggested by this experimental
work is: approximately 60%–80% pore volume CGI
injection followed by 1: 1 WAG, which conceptually
agrees with the speculation of Tanner et al. (1992) that
b. . .predict that a larger slug size (60% HCPV) could
result in additional EOR recovery. . .without increas-
ing peak gas production ratesQ.
4.1.2. Yates reservoir brine+n-Decane+Berea core
WAG and CGI floods were repeated at similar
flooding conditions, to those of 5% NaCl brine floods,
using Yates reservoir brine. The results are included in
Fig. 4(a) and (b). As noted in the previous case of 5%
NaCl brine, miscible floods showed significantly
higher recoveries compared to immiscible ones (Fig.
4(a)). The CGI recovery increased from 62.9% to
97.6% and WAG recoveries increased from 28.9%
to 72.5% due to miscibility.
The recovery plot of Fig. 4(a) appears to favor the
use of CGI rather than WAG in both miscible and
immiscible cases. However, the use of Tertiary Re-
covery Factor for comparison, as done in Fig. 4(b),
M.M. Kulkarni, D.N. Rao / Journal of Petroleum Science and Engineering 48 (2005) 1–20 11
shows the benefit of using WAG mode floods in both
miscible and immiscible cases. Therefore, even
though the conventional recovery plot shows higher
total recovery for CGI, WAG floods appear to be
better on the basis of recovery per unit volume of
gas injected.
One of the common features of the Figs. 3 and 4
are the immiscible CGI floods (Experiment 1 and 7)
with significant delays in oil production in spite of
continuous gas injection. This delay was further in-
vestigated by plotting volumetric injection/production
plots versus pore volume injection. Mass balance
calculations showed that the water production till oil
breakthrough matched the volume of cumulative CO
2
injection. The difference between injection and pro-
duction observed in Fig. 5 is attributable to the sig-
Effect of Miscibility (1-ft Berea + Yates Reservoir Brine)
0%
20%
40%
60%
80%
100%
0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5
Total PV (G + W) Injected
R
e
c
o
v
e
r
y

(
%

R
O
I
P
)
EXPT 7: IMM. CGI - Yates
EXPT 8: IMM. WAG - Yates
EXPT 9: MIS. CGI - Yates
EXPT 10: MIS. WAG -Yates
(a)
Effect of Miscibility (1-ft Berea + Yates Reservoir Brine)
0.0
0.2
0.4
0.6
0.8
1.0
1.2
0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5
Total PV (G + W) Injected
T
R
F

(
%

R
O
I
P

/

P
V
I

C
O
2
)
EXPT 7: IMM. CGI - Yates
EXPT 8: IMM. WAG - Yates
EXPT 9: MIS. CGI - Yates
EXPT 10: MIS. WAG - Yates
~ 0.75 PVI
(b)
Fig. 4. Effect of miscibility and mode of injection on tertiary recovery in 1-ft Berea core+reservoir brine+n-Decane (non-reactive) system. (a):
recovery as percent of ROIP, (b): fraction of residual oil in place recovered per pore-volume (PV) of CO
2
injected versus total PV (gas+water)
injected.
M.M. Kulkarni, D.N. Rao / Journal of Petroleum Science and Engineering 48 (2005) 1–20 12
nificant density differences between the injected CO
2
(77.85 kg/m
3
(4.86 lbm/ft
3
)) and reservoir brine
(999.23 kg/m
3
(62.38 lbm/ft
3
)).
Longer delays in oil production are observed for
the Yates brine immiscible CGI flood (Fig. 5(a))
compared to that of the 5% NaCl brine (Fig. 5(b)).
Experiment 1: Imsc CGI Flood (5% NaCl Brine)
0
50
100
150
200
250
300
350
0.0 0.5 1.0 1.5 2.0 2.5
PV Injected
P
r
o
d
u
c
t
i
o
n

/

I
n
j
e
c
t
i
o
n

(
c
c
)
Oil
Water
Gas
Injection Rate
The gap between
injection and production
is due to the density difference
SOR = 35.0%
SW = 65.0%
Final
SG = 52.1%
(a)
Experiment 7: Imsc CGI Flood (Yates Brine)
0
50
100
150
200
250
300
350
400
450
0 0.5 1 1.5 2 2.5 3 3.5
PV Injected
P
r
o
d
u
c
t
i
o
n

/

I
n
j
e
c
t
i
o
n

(
c
c
)
Oil
Water
Gas
Cum. CO
2
Injection
SOR = 25.5%
SW = 74.5%
Final
SG = 72.2%
Larger Injection - Production
Gap compared to 5% NaCl Brine
Attributable to higher CO
2
Solubility
and higher free water saturation
(b)
Fig. 5. Investigation of the delayed oil production for immiscible CGI floods using both 5% NaCl brine and Yates reservoir brine, (a): Oil-water-
gas-injection volumetric plot: 5%NaCl brine immiscible CGI flood, (b): Oil-water-gas-injection volumetric plot: Yates brine immiscible CGI flood.
M.M. Kulkarni, D.N. Rao / Journal of Petroleum Science and Engineering 48 (2005) 1–20 13
This is mainly due to the significantly higher solubil-
ity of CO
2
gas in multi-component brines than mono-
valent brines. Also the water-shielding and solubility
requirements are higher in experiment # 7 than exper-
iment # 1 due to higher water saturation (+10%) in the
core (Fig. 5). These results may have serious implica-
tions in the reservoir projects, in that higher costs may
be incurred due to delayed oil productions and in-
creased CO
2
requirements in immiscible mode.
This phenomenon of delayed oil breakthrough is
not observed for miscible floods since CO
2
has
significantly higher density (819.34 kg/m
3
(51.15
lbm/ft
3
)) at 17.24 MPa (2500 psi) injection pressures
resulting in lower density contrasts between core
brine and injected gas. Furthermore the differences
between CGI and WAG oil breakthroughs are sig-
nificantly reduced for the miscible floods compared
to the immiscible floods where this difference could
be as high as 1.8 PVI. Hence for miscible floods the
added benefit of hastened oil breakthroughs by WAG
employment is not available, and the CO
2
-brine
dissolution effect, favoring WAG application in im-
miscible mode, is not as pronounced for miscible
floods.
4.2. Effect of brine composition on tertiary miscible
CO
2
floods
The effects of brine composition on miscible floods
are shown in Fig. 6(a and b). As seen in the previous
sections, the evaluation of the process on the basis of
recoveries alone can lead to misleading conclusions;
hence Tertiary Recovery Factors were used to analyze
the coreflood results (Fig. 6(b)).
While in the case of CGI flood, there appears to be
minimal effect of brine composition (according to
Experiment 3 and 4 in Fig. 6(a)), WAG floods show
a significant dependence on brine composition. The
TRF plot shows that the 5% NaCl brine WAG flood is
the best of the miscible floods followed by Yates brine
WAG. The CGI floods for the two brines were com-
parable, but fared lower than the WAG floods.
4.3. Effect of using CO
2
-saturated brine
The difference in the performance of the two mis-
cible WAG floods can be attributed to the lower TDS
concentration, consequently lower solubility of CO
2
in Yates reservoir brine compared to the 5% NaCl
brine. Solubility experiments with a natural multi-
component brine (Paradox Valley Colorado) and solu-
tions of pure salt such as NaCl, and CaCl
2
, showed
that solubility of CO
2
in natural brines was higher
than in NaCl solution, and that the solubility of CO
2
was higher in the presence of divalent salts in natural
brine (Rosenbauer and Koksalan, 2002). Lower CO
2
solubilities in the brines result in relatively higher
CO
2
volumes available for incremental oil recovery
(by dissolution and swelling) in the 5% NaCl brine
flood than Yates reservoir brine.
To test the validity of the above hypothesis,
Experiments 7 and 10 were repeated with CO
2
-satu-
rated brines. Since there is no water injection in CGI
flood, the secondary waterflood was conducted using
saturated brine, and the drainage (oil flood) and EOR
(immiscible CGI) floods were conducted at conditions
similar to Experiment 7. In the miscible WAG exper-
iment, CO
2
-saturated brine was used in the tertiary
(EOR) mode while conducting the drainage (oil flood)
and imbibition (Yates reservoir brine flood) steps at
conditions similar to Experiment 10. The results of
these two experiments (immiscible CGI: Experiment
11 and miscible WAG: Experiment 12) are summa-
rized in Table 6, and Fig. 7.
In the case of the immiscible CGI flood (Exper-
iment 11), the n-Decane drainage step was similar to
those previously observed whereas the results of the
secondary waterflood with saturated Yates reservoir
brine showed significant pressure fluctuations up to
breakthrough. However the pressure fluctuations
were stabilized immediately after a sharp water
breakthrough. Even after breakthrough, a significant
delay (until 1.59 PVI) in the gas breakthrough (dis-
solved in brine) was observed along with increasing
pressure-drops after breakthrough. These observa-
tions are attributable to the miscible displacement
(consequently replacement) of the connate (unsatu-
rated) core brine by saturated injection brine. The
replacement of the unsaturated core brine with satu-
rated brine, helped significantly in decreasing the oil
and gas breakthrough time for the tertiary (EOR)
CO
2
CGI injection with significantly improved
TRF factors (Fig. 7(a) and (b)).
For the miscible CO
2
WAG flood (Experiment 12),
the drainage and imbibition steps were similar to the
previous WAG corefloods, however significant im-
M.M. Kulkarni, D.N. Rao / Journal of Petroleum Science and Engineering 48 (2005) 1–20 14
provement in the oil production was observed when
the saturated brine was used. Another feature of the
use of CO
2
saturated brine was the corresponding
increase in the observed pressure drops attributable
to 3-phase flow effects and hence decreased injectivity
compared to the previous normal brine WAG flood
(Fig. 8(b)).
The major observations obtained from the com-
parison of the normal (unsaturated) and saturated
brine WAG floods (Experiments 10 and 12 (Fig.
8)) are:
1. Liquid and water productions for both the core-
floods are identical.
Effect of Brine Composition (1-ft Berea + n-Decane + CO
2
)
0%
20%
40%
60%
80%
100%
0.0 0.5 1.0 1.5 2.0 2.5
Total PV (G + W) Injected
R
e
c
o
v
e
r
y

(
%

R
O
I
P
)
EXPT 3: MIS. CGI - 5% NaCl
EXPT 4: MIS. WAG - 5% NaCl
EXPT 9: MIS. CGI - Yates
EXPT 10: MIS. WAG - Yates
(a)
Effect of Brine Composition (1-ft Berea + n-Decane + CO
2
)
0.0
0.2
0.4
0.6
0.8
1.0
1.2
1.4
1.6
0.0 0.5
1.0
1.5 2.0 2.5
Total PV (G + W) Injected
T
R
F

(
%

R
O
I
P

/

P
V
I

C
O
2
)
EXPT 3: MIS. CGI - 5% NaCl
EXPT 4: MIS.WAG - 5% NaCl
EXPT 9: MIS. CGI - Yates Brine
EXPT 10: MIS. WAG -Yates Brine
(b)
Fig. 6. Effect of brine composition (Yates reservoir brine (multi-valent) and 5% NaCl solution (mono-valent)) on miscible tertiary recovery in 1-
ft Berea core+n-Decane (non-reactive) system, (a): recovery as percent of ROIP, (b): fraction of residual oil in place recovered per pore-volume
(PV) of CO
2
injected versus total PV (gas +water) injected.
M.M. Kulkarni, D.N. Rao / Journal of Petroleum Science and Engineering 48 (2005) 1–20 15
2. Experiment 12 with CO
2
-saturated brine recov-
ered significantly higher oil (89.2% ROIP) com-
pared to experiment 10 (72.5% ROIP). This is
attributable to the decreased solubility tendency of
CO
2
in brine (due to previous saturation) and
consequently higher volumes available for oil
recovery.
3. The improved oil recovery can also be partially
attributed to the decreased viscosity contrasts (Fig.
8(c)) between injection fluids leading to improved
sweeps.
4. The TRF maxima (Fig. 8(d)) were achieved at
almost identical pore volume injections (0.84 for
Experiment 10 and 0.82 for Experiment 12).
Expt 7: Immiscible CGI w/unsaturated Secondary WF
0%
20%
40%
60%
80%
100%
0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5
Total PV (G + W) Injected
R
e
c
o
v
e
r
y

(
%

R
O
I
P
)
0.00
0.05
0.10
0.15
0.20
0.25
T
R
F

(
%

R
O
I
P

/

P
V
I

C
O
2
)
Recovery
TRF
(a)
Expt 11: Immiscible CGI w/CO
2
-saturated Secondary WF
0%
5%
10%
15%
20%
25%
0.0 0.1 0.2 0.3 0.4 0.5
Total PV (G + W) Injected
R
e
c
o
v
e
r
y

(
%

R
O
I
P
)
0.00
0.50
1.00
1.50
2.00
2.50
T
R
F

(
%

R
O
I
P

/

P
V
I

C
O
2
)
Recovery
TRF
(b)

Fig. 7. Effect of saturation of brine with CO
2
on immiscible CGI recovery, (a): oil recovery and TRF for CGI flood with unsaturated brine
secondary waterflood, (b): oil recovery and TRF for CGI flood with saturated brine secondary waterflood.
M.M. Kulkarni, D.N. Rao / Journal of Petroleum Science and Engineering 48 (2005) 1–20 16
5. Hastened breakthroughs and higher gas produc-
tions were observed due to the use of CO
2
saturat-
ed brine (Fig. 8(a) and (d)).
The peak TRF values calculated for each of the
corefloods conducted are summarized in Fig. 9. It is
interesting to note that the peak TRF values, as ob-
served from Fig. 9, for the 5% NaCl brine miscible
floods (both CGI and WAG) are higher than the Yates
brine miscible floods. However, this effect has been
reversed for the immiscible floods. This indicates that
although the Yates brine has a higher CO
2
solubility
than 5% NaCl brine at 3.45 MPa (500 psi); this effect
is offset at 17.24 MPa (2500 psi) (miscible) flooding
conditions. The highest TRF factor for CGI flood was
obtained by the use of saturated brine in secondary
mode as expected. This data further fortifies the ear-
lier assumption of relatively higher CO
2
solubility rate
in brine at lower pressures and that this effect is
mitigated at miscible flooding conditions (Experiment
12). Consequently incremental benefits of the brine-
CO
2
solubility reduction (by prior saturation) are more
than offset by miscibility development.
The recoveries, residual oil saturations and gas
utilization factors for the corefloods conducted are
summarized in the Tables 4, 5 and 6 (Part (C)). The
utilization factor, defined earlier, is a good indicator of
the overall efficiency of the process, and is a useful
augmentation, along with the TRF, for the analysis of
the data. The utilization factor is a measure of the
0
10
20
30
40
0 0.5 1.5 1 2
PV Injected
O
i
l

(
c
c
)
Oil (Saturated Brine)
Oil (Normal Brine)
0
5
10
15
20
25
30
35
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2
Total PV (G + W) Injected
P
r
e
s
s
u
r
e

D
r
o
p

(
p
s
i
)
EXPT 12: CO
2
Saturated Yates Brine WAG
EXPT 10: Normal Yates Brine WAG
G
G G
G
G
W
W W W
W
(a) (b)
0.50
0.55
0.60
0.65
0 2000 4000 6000 8000 10000
Pressure (psi)
B
r
i
n
e

V
i
s
c
o
s
i
t
y

(
c
P
)
0.00
0.05
0.10
0.15
0.20
C
O
2

V
i
s
c
o
s
i
t
y

(
c
P
)
Sat. Brine
Normal Brine
CO
2
Gas
0.0
0.4
0.8
1.2
1.6
0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0
PV Injected
T
R
F

(
R
O
I
P

/

P
V
I

C
O
2
)
Saturated Brine
Normal Brine
(c) (d)
Fig. 8. Effect of saturation of Yates reservoir brine with CO
2
on miscible WAG recovery using n-Decane and CO
2
, (a): oil recovery comparison,
(b): DP change during WAG floods, (c): viscosity values for expt. fluids (d) TRF comparisons for WAG floods (using CMG Winprop
R
).
M.M. Kulkarni, D.N. Rao / Journal of Petroleum Science and Engineering 48 (2005) 1–20 17
design CO
2
requirements for the field gas injection
project.
5. Summary and conclusions
1. Based on oil recovery, the CGI flood appeared to
be better in performance than WAG flood. How-
ever, on the basis of the overall Tertiary Recovery
Factor (TRF), where the recoveries were normal-
ized by the volume of CO
2
injected, the WAG
floods clearly out-performed the CGI floods. Fur-
thermore, the TRF performance of the CGI misci-
ble flood approaches the relatively low recoveries
obtained in the immiscible gas floods, indicating
deteriorating returns from the CGI with time.
2. Miscible gas floods were found to recover over
60% to 70% more of the waterflood residual oil
than immiscible gas floods. While the recoveries in
immiscible 5% NaCl brine floods (both CGI and
WAG) were about 23%, the miscible floods yielded
84.5% recovery for the 5% NaCl brine WAG flood
(for 1.02 PVof CO
2
injected) and 96.7% recovery
for the 5% NaCl brine CGI flood (for 2.44 PV of
CO
2
injected). However, about 94% of the oil is
produced in ~1.02 PVof CO2 injected compared to
84.5% for WAG.
3. Experimental results show that for optimization of
tertiary recovery in gas floods, a continuous gas
slug of 0.7 PV (where the CGI flood showed
maximum TRF value) followed by 1: 1 WAG
needs to be injected. This optimized method indi-
cated by our results was found to be similar to the
patented dHybrid WAGT and dDUWAGT processes
employed in the oil industry.
4. Miscible CGI floods showed negligible sensitivity
to brine composition variations. Recoveries of
96.7% and 97.6% where obtained with 5% NaCl
0.112
0.218
2.25
0.0
0.4
0.8
1.2
1.6
2.0
2.4
P
e
a
k

T
R
F

V
a
l
u
e
Immsc. CGI Floods @ 500 psi
1.054
0.844
0.0
0.4
0.8
1.2
1.6
2.0
2.4
Misc. CGI Floods @ 2500 psi
0.229
0.611
0.0
0.4
0.8
1.2
1.6
2.0
2.4
WAG-NaCl (# 2) WAG-Y (# 8)
Experiment
WAG-NaCl (# 4) WAG-Y (# 10) WAG-Sat-Y (# 12)
Experiment
CGI-NaCl (# 3) CGI-Y (# 9)
Experiment
CGI-NaCl (# 1) CGI-Y (# 7) CGI-Sat-Y (# 11)
Experiment
P
e
a
k

T
R
F

V
a
l
u
e


P
e
a
k

T
R
F

V
a
l
u
e
P
e
a
k

T
R
F

V
a
l
u
e


Immsc. WAG Floods @ 500 psi
1.473
1.114
1.523
0.0
0.4
0.8
1.2
1.6
2.0
2.4
Misc. WAG Floods @ 2500 psi
(a)
(b)
Fig. 9. Comparison of peak TRF values for CGI and WAG experiments for 5% NaCl brine and Yates reservoir brine, (a): peak TRF value
comparisons for immiscible and miscible CGI floods, (b): peak TRF value comparisons for immiscible and miscible WAG floods.
M.M. Kulkarni, D.N. Rao / Journal of Petroleum Science and Engineering 48 (2005) 1–20 18
brine and Yates reservoir brine, respectively. In
contrast, the miscible WAG recoveries exhibited
significant dependence on brine composition. The
miscible WAG recoveries showed a significant
decrease (12%) in oil recovery when the connate
brine was changed from 5% NaCl solution to
Yates reservoir brine. While the recovery for the
miscible 5% NaCl brine was 84.5%, it decreased
to 72.5% for Yates reservoir brine. This is attrib-
utable to the higher solubility of CO
2
in natural
multi-component brines than solutions of pure
salts like NaCl, which results in higher volumes
of CO
2
being available for oil recovery in 5%
NaCl brine floods.
5. Solubility of CO
2
in reservoir brine (at lower pres-
sures) may have serious implications in the reser-
voir projects, in that the costs may increase due to
delayed oil productions and increased CO
2
require-
ments for injection in immiscible mode.
6. Unlike immiscible floods, where WAG employ-
ment hastens oil breakthroughs, the miscible
WAG and CGI floods’ oil breakthroughs occur at
near identical pore volume injections (Figs. 3(a)
and 4(a)). The delayed oil breakthroughs in immis-
cible floods are attributable to CO
2
solubility
effects in core-brine. However, miscibility devel-
opment offsets these brine solubility effects and the
need for pre-saturation of injection brine with CO
2
appears to be effectively eliminated.
7. In addition to sweep improvement, if the purpose of
the employment of the WAGprocess to decrease the
quantities of CO
2
injected, then the environmental
benefit of CO
2
sequestration would be minimal.
8. Watered out reservoirs containing high water
saturations serve as good candidates for CO
2
se-
questration through CO
2
dissolution in brine.
Acknowledgments
This paper was prepared with the support of the U.S
Department of Energy under Award No. DE-FC26-
02NT-15323. Any opinions, findings, conclusions or
recommendations expressed herein are those of authors
and do not necessarily reflect the views of the DOE.
The financial support of this project by the U.S. De-
partment of Energy is gratefully acknowledged. The
authors thank Dr. Jerry Casteel of NPTO/DOE for his
support and encouragement. Special thanks to Mr. Dan
Lawrence and Mr. Chandrasekhar Vijapurapu of LSU
for their valuable technical help.
References
Ahmed, T., 1997. A generalized methodology for minimum mis-
cibility pressure. SPE 39034, presented at the Fifth Latin Amer-
ican and Caribbean Petroleum Engineering Conference held in
Rio De Janeiro, Brazil, 30 Aug–3 Sept.
Ayirala, S.C., 2002. Surfactant-induced Relative Permeability Mod-
ifications for Oil Recovery Enhancement, M.S. Thesis, The
Craft and Hawkins Department of Petroleum Engineering,
Louisiana State University and A and M College, Baton
Rouge, LA (Dec).
Bellavance, J.F.R., 1996. Dollarhide Devonian CO
2
flood: project
performance review 10 years later. SPE 35190, Presented at the
SPE Permian Basin Oil and Gas Recovery Conference. Mid-
land, TX, Mar 27–29.
Benmekki, E.H., Mansoori, G.A., 1986. Accurate vaporizing gas-
drive minimum miscibility pressure prediction. SPE 15677,
Presented at the 61st Annual Technical Conference and Exhibi-
tion of the Society of Petroleum Engineers held in New Orleans,
LA October 5–8.
Christensen, J.R., Stenby, E.H., Skauge, A., 1998. Review of the
WAG field experience. SPE 71203, revised paper 39883, Pre-
sented at the 1998 SPE International Petroleum Conference and
Exhibition of Mexico. Villahermosa, March 3–5.
Eng, J.H., Bennion, D.B., Strong, J.B., 1993. Velocity profiles in
perforated completions. Journal of Canadian Petroleum Tech-
nology 32 (8), 49–54 (Oct).
Enick, R.M., Beckman, E.J., Shi, C., Huang, Z., Xu, J., Kilic, S.,
2000. Direct thickeners for CO
2
. SPE 59325, Presented at the
2000 SPE/DOE Improved Oil Recovery Symposium held in
Tulsa, OK, April, 3–5.
Filoco, P.R., Sharma, M.M., 1998. Effect of brine salinity and
crude-oil properties on oil recovery and residual oil saturations.
SPE 65402, Presented at 1998 SPE Annual Technical Confer-
ence and Exhibition, held in New Orleans, LA 27–30 Sept.
Hill, D.G., 1982. Clay stabilization — criteria for best performance.
SPE 10656, Presented at the SPE Formation Damage Control
Symposium, held in Lafayette, LA, Mar 24–25.
Hinderaker, L., Utseth, R.H., Hustad, O.S., Kvanvik, B.A., Paulsen,
J.E., 1996. RUTH — A comprehensive Norwegian R&D pro-
gram on IOR. SPE 36844, Presented at the SPE European
Petroleum Conference held in Milan, Italy, Oct 22–24.
Huang, E.T.S., Holm, L.W., 1986. Effect of WAG injection and
wettability on oil recovery during carbon dioxide flooding. SPE
15491, Presented at 1986 Annual Technical Conference and
Exhibition, New Orleans, LA, Oct 5–8.
Jones, F.O., 1964. Influence of chemical composition of water on
clay blocking of permeability (SPE 631). Journal of Petroleum
Technology, 441–446 (April).
Khilar, K.C., Vaidya, R.N., Fogler, H.S., 1990. Colloidally-induced
fines release in porous media. Journal of Petroleum Technology
4, 213–221 (July).
M.M. Kulkarni, D.N. Rao / Journal of Petroleum Science and Engineering 48 (2005) 1–20 19
Kulkarni, Madhav M., 2003. Immiscible and Miscible Gas–Oil
Displacements In Porous Media, M.S. Thesis, The Craft and
Hawkins Department Of Petroleum Engineering, Louisiana
State University and Agricultural and Mechanical College,
Baton Rouge, LA, Jul 2003. URL: http://etd.lsu.edu:8085/
docs/available/etd-0709103-175151/.
Kulkarni, M.M., 2004. Multiphase mechanisms and fluid dynamics
in gas injection enhanced oil recovery processes, Ph.D. General
Exam Report, The Craft and Hawkins Department of Petroleum
Engineering, Louisiana State University and A and M College,
Baton Rouge, LA (Jul).
Kwan, M.Y., Cullen, M.P., Jamieson, P.R., Fortier, R.A., 1989. A
laboratory study of permeability damage to cold lake tar
sands cores. Journal of Canadian Petroleum Technology 28
(1), 56–62 (Jan–Feb).
McKean, T.A.M., Thomas, A.H., Chesher, J.R., Weggeland, M.C.,
1999. Schrader bluff CO
2
EOR evaluation. SPE 54619, Pre-
sented at the 1999 SPE Western Region Meeting held in An-
chorage, Alaska, 26–26 May.
Moritis, G., 1995. Impact of production and development RD&D
ranked, production editor. Oil and Gas Journal 93 (44) (Oct. 30).
Moritis, G., 2002. OGJ EOR Survey 2002. Oil and Gas Journal
(Apr 15).
Orr, F.M., Jensen, C.M., 1984. Interpretation of pressure–composi-
tion phase diagrams for CO
2
/Crude-oil systems (SPE 11125).
SPE-Journal.
Orr, F.M., Silva, M.K., 1983. Equilibrium phase compositions of
CO
2
/hydrocarbon mixtures — Part 1: measurement by contin-
uous multiple-contact experiment (SPE 10726). SPE-Journal
(April).
Owens, W.W., Archer, D.L., 1971. The effect of rock wettability on
oil–water relative permeability relationships. JPT (July).
Rao, D.N., 2001. Gas injection EOR — a new meaning in the new
millennium (Invited article for the Distinguished Author Series).
Journal of Canadian Petroleum Technology 40 (2), 11–18
(Feb).
Rappaport, L.A., Leas, W.J., 1953. Properties of linear waterfloods.
Transactions of AIME 198, 139.
Rosenbauer, R.J., Koksalan, T., 2002. Experimental determination
of the solubility of CO
2
in electrolytes: application to CO
2
sequestration in deep-saline aquifers. Paper 135-2, Presented
at the 2002 Denver Annual Meeting, The Geological Society
of America, Denver, CO Oct 29.
Tang, G., Morrow, N.R., 1999. Oil recovery by waterflooding and
imbibition — invading brine cation valency and salinity. SCA
9911 Proceedings of the International Symposium of the Society
of Core Analysts, Golden, CO. August.
Tanner, C.S., Baxley, P.T., Crump, J.C. III, Miller, W.C., 1992.
Production performance of the Wasson Denver Unit CO
2
flood. SPE/DOE 24156, Presented at the SPE/DOE Eighth
Symposium on Enhanced Oil Recovery held in Tulsa, OK,
Aug 22–23.
Vijapurapu, C.S., Rao, D.N., 2002. The effect of rock sur-
face characteristics on reservoir wettability. In: Mittal, L.
(Ed.), Contact Angle, Wettability and Adhesion, vol. 3,
pp. 407–426.
M.M. Kulkarni, D.N. Rao / Journal of Petroleum Science and Engineering 48 (2005) 1–20 20