EOS Modeling and
Experimental Observations of
Three-Hydrocarbon-Phase Equilibria
S.P. Godbole* and K.J. Thele, SPE, Arco E&P Technology, and E.W. Reinbold, SPE, Arco Alaska Inc.
Summary
An equation-of-state (EOS) fluid characterization has been devel-
oped to model three-hydrocarbon-phase equilibria as part of the en-
riched-gas EOR research program for the Kuparuk hydrocarbon
miscible flood on the Alaskan North Slope. Laboratory experiments
with Kuparuk reservoir oils and miscible injectants have identified
an apparent second equilibrium liquid that is heavier than the origi-
nal oil and macroscopically reversible. A three-phase EOS fluid
characterization was developed to match both two- and three-phase
experimental data by use of a modified Peng-Robinson EOS. Exper-
imental observations identified a relationship between the third
phase and the oil's pentane-insoluble fraction. The characterization
uses a total of 16 components, with the heaviest two approximating
the pentane-soluble and pentane-insoluble portions of the C36+
fraction. The properties of these heavy pseudocomponents control
the development and behavior of the heavy phase. The EOS fluid
characterization was used to study three-phase, enriched-gas-drive
displacement mechanisms. For the Kuparuk fluids, a condensing/
vaporizing displacement mechanism is predicted in a three-phase
environment. Reservoir-condition micromodel displacements were
conducted to observe the phase-behavior and displacement mecha-
nisms during Kuparuk enriched-gas displacements. Micromodel
observations are consistent with a near-miscible condensing/vapor-
izing mechanism and qualitatively agree with EOS predictions.
Introduction
A hydrocarbon miscible flood is being conducted in the Kuparuk
River field on the North Slope of Alaska. 1 The EOR project covers
5,120 acres at Drillsites I Y and 2Z. The flood is in Sands A and C
of the Kuparuk River formation, a stratified and highly faulted sand-
stone reservoir. The enriched gas injected at the project is a blend of
lean produced gas with scrubber and low-temperature separator liq-
uids (enriching fluids) produced on site. Project design calls for in-
jection of a 30%-hydrocarbon-PV (HPV) slug of enriched gas alter-
nately with water at a I : I volume ratio for mobility control.
Fluid properties at Kuparuk vary somewhat across the areal ex-
tent of the reservoir. Oil gravities range from 18 to 25° API, and pen-
tane insolubles range from 3 to 17 wt% of the C7 + fraction. An ex-
tensive laboratory program was undertaken to provide phase-behavior
data for designing and operating the EOR project. The laboratory
program identified a third hydrocarbon phase when miscible injec-
tant was mixed with reservoir oil.
The phase equilibria observed with the Kuparuk oils and enriched
gases differ from those reported in the literature for other enriched-
gas/crude-oil mixtures 2-5 and C02/hydrocarbon mixtures. 5-8 In
those systems, three phases exist in a relatively narrow temperature/
pressure/composition window and consist of a liquid, a vapor, and
a second liquid lighter than the first and rich in components that
make up the injectant. In some systems, asphaltenic flocculation
and deposition may also occur. The third phase seen with Kuparuk
fluids occurs over a large region of pressure/composition space, ap-
pears liquid-like, and is heavier than the original oil. This heavy
phase is > 30 wt% pentane insolubles and also contains significant
'Now with BP Chemicals.
Copyright 1995 Society of Petroleum Engineers
Original SPE manuscript received for review Oct. 4, 1992. Revised manuscript received Jan.
12,1995. Paper accepted for publication Feb. 8, 1995. Paper (SPE 24936) first presented at
the 1992 SPE Annual Technical Conference and Exhibition held in Washington, DC, Oct. 4-7.
SPE Reservoir Engineering. May 1995
quantities of lighter components (e.g., 3 wt% methane and 7 wt%
C2 through C6). Small quantities of what are assumed to be asphal-
tenes also deposited on the cell walls in some Kuparuk experiments.
Several models are discussed in the literature for modeling multi-
phase behavior. The appearance and disappearance of the upper liq-
uid phase with mixtures of oil and C02 or enriched gas have been
modeled with an EOS,?·9 The literature describes modeling of floc-
culation and precipitation of asphaltenes by irreversible flocculation
of asphaltic particles 10 and reversible asphaltene precipitation. I 1.12
For simplicity, we considered only three-phase models for
approximating the vapor, liquid, apparent second liquid, and asphal-
tene phases of the Kuparuk system. The apparent second liquid
could be approximated as a mixture of aggregated asphaltenes (sol-
id phase) entrained in a portion of the other liquid phase. This ap-
proach offers the possibility of modeling the asphaltenic deposition
as well as the second liquid phase; however, it assumes that asphal-
tenes have aggregated in the liquid phase and that the upper liquid
and the lighter portion of the lower liquid have similar composi-
tions. Although we did not test the lower phase for aggregated as-
phaltenes, compositional data indicate that the lighter portion of the
lower liquid phase is different from the upper liquid phase.
Another three-phase approach is to ignore the asphaltenic deposi-
tion and model the apparent second liquid as a true equilibrium
phase by use of the same EOS that is used to model the other liquid
and vapor phases. The formation of this heavy phase needs to be re-
versible for this model and experimentally was found to be at least
macroscopically reversible. Because we do not believe that asphal-
tenic deposition plays a significant role in the phase behavior at Ku-
paruk and because this model is derived from the data more directly,
we chose this approach for our modeling.
Baker et al. l3 and Michelsen l4 •15 have presented algorithms
based on cubic EOS's to model reversible multi phase behavior of
hydrocarbon mixtures. These models determine the correct number
of phases by minimizing the overall Gibbs free energy. An in-house
phase-behavior package based on Michelsen's flash algorithm and
the three-parameter Peng-Robinson EOSI6 was used in this study.
A micromodel apparatus was constructed to enable visual observa-
tion of the complex phase-behavior and associated displacement
mechanisms. Micromodels are transparent pore networks that allow
pore-level observation of fluid flow. Commonly, micromodels are 2D
pore networks etched on a glass surface, although many construction
techniques and materials have been used. 17 Micromodels have been
used to study a wide range of displacement phenomena. 18
This paper presents laboratory data, discusses the development of
a three-phase EOS fluid characterization, describes a three-phase
condensing/vaporizing displacement mechanism, and compares EOS
predictions with observations from micromodel displacements.
Experimental Data
The laboratory program included conventional PVT studies; slim-
tube, coreflood, and micromodel displacements; and single-con-
tact, multicontact, and heavy-phase onset experiments. The data
discussed in this paper are for recombined oils from Wells I Y-9,
IE-8, and 2D-12. The Well IY-9 oil is representative of the EOR
project area and is low in pentane insolubles relative to the field av-
erage. WelllE-8 and 2D-12 oils have more pentane insolubles and
yield larger amounts of the heavy third phase.
PVT Studies. Conventional PVT studies included constant-com-
position-expansion, differential-liberation, and separator-flash ex-
101
 TABLE 1-COMPOSITIONS (MOLE FRACTION) OF FLUIDS IN TERMS OF EOS PSEUDOCOMPONENTS
Well Well Well
Component MW 1E-8 Oil 20-12 Oil 1Y-9 Oil Lean Gas EF-1 EF-2 60/40 MI 70/30 MI 77123 MI
- -- ---
CO 2 44 0.0061 0.0068 0.0028
C1 16 0.3954 0.3867 0.4016 0.8510 0.1355 0.1409 0.5660 0.6387 0.6753
C2 30 0.0507 0.0722 0.0460 0.0850 0.2028 0.0874 0.1311 0.0861 0.0853
C3 44 0.0349 0.0556 0.0386 0.0450 0.2980 0.1529 0.1466 0.0780 0.0768
C4 58 0.0237 0.0304 0.0307 0.0140 0.2141 0.1772 0.0936 0.0626 0.0534
Cs 72 0.0158 0.0166 0.0206 0.0050 0.1041 0.1083 0.0450 0.0340 0.0281
C6 84 0.0243 0.0155 0.0163 0.0455 0.0962 0.0177 0.0270 0.0214
C7 ~6 0.0210 0.0300 0.0316 0.2101 0.0662 0.0541
Cs 107 0.0299 0.0395 0.0401 0.0270 0.0074 0.0056
C9 121 0.0303 0.0295 0.0298
C 10 through C 13 155 0.1084 0.0866 0.0984
C14 through C 19 227 0.1043 0.0953 0.1059
C20 through C26 317 0.0637 0.0563 0.0620
C27 through C3S 430 0.0399 0.0328 0.0329
C 36 S 722 0.0428 0.0414 0.0414
C361 2,717 0.0088 0.0048 0.0013
Bubblepoint pressure 2,800 2,800 2,720
at 160°F, psig

periments. These experiments provided saturation pressures, densi-
ties, relative volumes, GOR's, and viscosity data.
Displacement Experiments. Slim-tube displacements were con-
ducted to define minimum miscibility enrichments (MME's) for
various enriched-gaslKuparuk-oil mixtures. The slim-tube data
were not used in developing the fluid characterization so that they
could be used to validate the characterization. To date, the coreflood
results have not been used quantitatively with the EOS model and
are not discussed in this paper. The micromodel experiments, like
the slim-tube displacements, were used to test the characterization
and are discussed later.
A series of constant-pressure, variable-enrichment, slim-tube
displacements at 3,200 psi and 160°F was performed with Well
IE-8 oil. Table 1 lists the compositions of all fluids used in the ex-
periments discussed in this paper. The first enriching fluid, EF-I in
Table I, was mixed with lean gas in various ratios for this slim-tube
series. Recoveries at 1.2 injected PV were plotted against enrich-
ment, giving an MME of 33 mol% enriching fluid, with ultimate re-
coveries of 96% for enrichments greater than the MME.
Single- and Multicontact Experiments. A multicontact experi-
ment was performed with Well lE-8 oil and miscible injectant (MI)
consisting of 60% lean gas and 40% EF-I (labeled 60/40 MI) to pro-
vide phase properties and compositions along a portion of the same
compositional path followed in one of the slim-tube displacements.
In the first stage, injectant was added to the reservoir oil in a 1 : 2 vol-
ume ratio and allowed to equilibrate at reservoir conditions. The en-
tire equilibrium vapor phase was removed, and its volume, density,
and composition were determined. Approximately 10 vol% of the
equilibrium liquid was removed from the top of the cell for analysis.
Fresh injectant was added to the cell, and the process was repeated
for a total of seven stages.
On completing the multicontact experiment, we found that an un-
expected heavy phase remained in the cell. Because the experiment
was performed in a blind cell, when the third phase formed was un-
clear; however, material balance suggested that the heavy phase
formed during the first stage. For confirmation, the first stage was
repeated, and a heavy-liquid phase was identified, occupying 12%
of the original cell volume. The heavy phase was titrated with pen-
tane at room conditions and found to contain a significant quantity
of pentane insolubles.
Heavy-Phase-Onset Experiments. Onset experiments were per-
formed to determine the minimum ratio of injectant to oil required
to form the heavy phase. Injectant was added to reservoir oil at a
1 : 10 volume ratio in a blind cell at reservoir conditions. The pres-
sure was raised to the bubblepoint pressure of the mixture, and a liq-
uid sample was removed from the top of the cell. The density and
composition of the sample were measured at experimental condi-
tions. The sample was then flashed to atmospheric conditions, and
the pentane-insoluble content of the liquid was determined. More
injectant was added to the cell, and the process was repeated for a
total of eight stages. A decrease in the pentane-insoluble content of

TABLE 2-COMPARISON OF PREDICTED AND EXPERIMENTAL PHASE PROPERTIES AT

POINTS A THROUGH 0 IN p-xDIAGRAM FOR WELL 20-12 OIL AND 70/30 MI
Pressure Relative Volume (%) Density (g/cm3) C s Insoluble (wt%)
Point (psig) Phase Experimental EOS Experimental EOS Experimental EOS
A 2,800 Vapor 31.8 33.3 0.204 0.209 N/A N/A
Liquid 62.2 61.1 0.695 0.665 3.6 2.2
Heavy 6.0 5.6 0.941 0.938 51.2 49.6
B 1,900 Vapor 47.8 50.2 0.119 0.128 N/A N/A
Liquid 50.0 47.4 0.722 0.703 7.2 5.0
Heavy 2.2 2.4 0.935 41.9 50.3
C 5,000 Vapor 0 23.4 N/A 0.475 N/A N/A
Liquid 87.6 62.6 0.600 0.622 0.8 0.3
Heavy 12.4 14.0 0.886 0.888 41.4 32.2
0 2,800 Vapor 50.2 43.9 0.277 0.226 N/A N/A
Liquid 44.9 49.7 0.638 0.592 1.1 0.2
Heavy 4.9 6.4 0.901 64.7 36.7
'Not measured owing to experimental difficulties.

102 SPE Reservoir Engineering, May 1995

 6000
H-L
5000 "'C
L •
(i 4000
'iii H-L-V
.eo
Q)
3000
:; "'A "'0
til
til
~ L-V
11. 2000 "'8
_ Predicted Phase Boundaries
• C7 through C35 range. The C36S and C361 pseudocomponents were
1000
• Experimental Phase Boundaries
Experimental Point
'"
0
0 0.2 0.4 0.6 0.8
MI Mole Fraction
Fig. 1-Predicted pox diagram and experimental points; Well
02-12 oil and 70/30 MI at 160°F.
the liquid sample (upper phase) signifies transfer of heavy compo-
nents to a second, heavy phase. For mixtures of Well I E-8 oil and
60/40 MI, the heavy-phase onset point is ... 30 mol% MI.
On a p-x diagram, these heavy-phase-onset experiments follow
the bubblepoint-pressure curve from the original oil composition to
the composition where the heavy phase starts forming. All these
data were used in developing the EOS fluid characterization.
Miscellaneous Experiments. Additional slim-tube displacements
and another multicontact experiment were performed with lighter
oils and other injectants. Five more onset experiments and several
single-contact experiments were also performed. Table 2 and Fig.
1 give results from some of the onset and single-contact experi-
ments. Owing to the nature of the heavy phase, some concern ex-
isted about reaching equilibrium within 1 or 2 days. Phase densities,
compositional analyses, and pentane-insoluble measurements from
three single-contact experiments identical in procedure except for
time allowed for equilibration (1 day, 1 week, and 1 month) sup-
ported the hypothesis that three-phase equilibria were being at-
tained in all experiments; this was a key requirement for the EOS
modeling described in this paper. Constant-composition experi-
ments (near Point B in Fig. 1) showed that the heavy liquid will ap-
pear or disappear by altering the pressure, indicating that the three-
phase behavior is macroscopically reversible, another key
requirement for the EOS model.
A few oils were titrated with pentane, and the pentane-soluble
portions were analyzed by gas chromatography (GC) and their mo-
lecular weights determined by benzene-freezing-point depression.
The data from these titrations were used to define some of the
pseudocomponents in the fluid characterization.
EOS Fluid Characterization
Pseudocomponent Selection. The EOS fluid characterization uses
16 components, including six pseudocomponents to describe the
heavier fractions. The first 10 components, C02 and methane
through nonane, were kept separate to provide better definition of
the injectant composition. The remaining six pseudocomponents
were selected on the basis of C7 through C36 + GC analyses of 24
oil samples. Components in the C 10 through C 35 range were lumped
into four pseudocomponents, representing approximately 1/3 , 1/3,
1/6, and 1/6 of this fraction on a molar basis. The carbon-number
splits are CIO through CI3, CI4 through C19, C20 through C26, and
C27 through C35.
The characterization was developed around the idea that the pen-
tane insolubles playa significant role in development of the second
liquid phase. Experimental data confirmed that the second liquid
phase contains large quantities of pentane insolubles. Composition-
al analyses and molecular-weight determinations indicate that the
SPE Reservoir Engineering, May 1995
pentane insolubles reside in the C36 + fraction and have a consider-
ably higher molecular weight than the rest of the C36 + fraction.
The C36 + fraction is split into two pseudocomponents represent-
ing the pentane-soluble (C36S) and pentane-insoluble (C361) por-
tions of the C36 + fraction. The C36 + split is based on room-condi-
tion pentane-insoluble measurements of the C7 + cuts. This
approach allows the molecular weight of the C36 + fraction to vary
with the pentane-insoluble content of the oil. However, we did not
try to model the amount of precipitate in the pentane titrations per-
formed at room conditions.
Molecular weights and normal-alkane boiling points from Katz
and Firoozabadi 19 were assigned to all ~seudocomponents from the
assigned molecular weights of 722 and 2,717, respectively; these
were backcalculated on the basis of experimental molecular weights
OfC7 + cuts, GC analyses, and Katz-and-Firoozabadi-assigned mo-
lecular weights for C7 through C35. Twu's correlations 20 were then
used to compute critical temperatures and pressures and acentric
factors; volumetric shift parameters 16 were based on Katz and Fi-
roozabadi's specific gravities.
Regression Procedure. An in-house phase-behavior regression
package was used for regression on the component properties of the
six heaviest pseudocomponents and all the binary interaction coeffi-
cients. The regression was performed in a sequence of steps.
1. Regression was performed on the component properties and
binary interaction coefficients until a satisfactory match was ob-
tained for all two-phase data, excluding viscosities (similar to devel-
oping a conventional two-phase characterization).
2. The characterization was tested by checking for three phases
at the experimental three-phase conditions. If three phases were not
predicted, three were induced by decreasing the volatility of the two
heaviest pseudocomponents (by decreasing critical pressure, in-
creasing critical temperature and acentric factor, andlor increasing
the binary interaction coefficients between them and the very light
components). However, decreasing the volatility of C 36S or C 361 too
much caused three phases to be predicted where only two phases
physically exist, such as in standard black-oil separator tests.
3. All three-phase data except for the multicontact three-phase data
were added to the database, and the regression was continued. (Multi-
contact data were difficult to match without a good initial three-phase
characterization because EOS calculations for the second and subse-
quent contacts depend on calculations from all earlier contacts.) The
critical properties of C361 and the binary interaction coefficient be-
tween C361 and methane tend to have a much greater effect on three-
phase than on two-phase calculations. Decreasing the C361 volatility,
however, did make the bubblepoint pressures overly sensitive to the
pentane-insoluble content of the oil.
4. The multicontact three-phase data were added to the regression
database, and the regression was continued. Matching the multicon-
tact data for both light and heavy oils without sacrificing the two-
phase matches was very challenging.
5. After a satisfactory match was achieved between the data and
the model for phase volumes, densities, and compositions, the vis-
cosity data were added and regression was conducted on the critical
compressibility factors for the last six pseudocomponents to reduce
the error between experimental and computed viscosities by use of
the Lohrenz et at. 21 correlation.
Characterization Validation. The quality of the final characteriza-
tion is tested in three ways. The first is to determine whether the crit-
ical properties and binary interaction coefficients follow the correct
trends. In the final characterization, all hydrocarbon-component
critical temperatures, acentric factors, specific gravities, and visco-
sities monotonically increase and the critical pressures monotoni-
cally decrease with increasing molecular weight. The binary inter-
action coefficients also follow desired trends.
The second is to determine how well the results match the data
that were used to develop the characterization. Most of the two-
phase data matched quite well. The average discrepancy between
measured and computed bubblepoint pressures is 75 psi, with the
computed bubblepoint pressures generally being too high for the
103
 0 0 0
0 Vapor
[] Liquid
• Heavy Phase
0 0 o

0"
•• Heavy
non-Vapor
M
u
......
-9
o.g

~~ • • ~
u
~ 1
0 0 ~
en en
Q)
0.6 Original Oil)
Q) :;:; Liquid Phase
E '00
::l c (Injectant)
"0 Q)
> Cl
Q)
en
ttl
.c
Q)
en
ttl
.c
0.4
/ 6
6

Vapor Phase
{), ~

a... a... ~
0.2
Solid symbols - Expt.
Open Symbols - EOS
o °0·L-----~--~2----~3~----4~--~5----~6----~7
Stages 1 through 7
Stage

Fig. 2-Phase volumes for multicontact experiment; Wen 1E-8 Fig. 3-Phase densities for multicontact experiment; Wen 1E·8
oil and 60/40 MI at 3,200 psig and 160°F. oil and 60/40 MI at 3,200 psig and 160°F.

heavy oils and too low for the light oils. The largest relative error
between the model and the data for the phase densities and volumes
from the differential-liberation and constant-composition-expan-
sion experiments is 2%. The errors in the viscosities are consider-
=
ably higher, usually 10%, but still satisfactory.
The multicontact experiments played a key role in tuning the
characterization. Figs. 2 and 3 show the experimental and com-
puted phase volumes and densities for each of the seven contacts
with Well IE-8 oil, and Figs. 4 and 5 show the results with Well
I Y-9 oil. Owing to the experimental procedure, the volumes of the
individual heavy and liquid phases and the densities of the heavy
phase could not be determined except for the first stage with Well
lE-8 oil and the last stage with Well I Y-9 oil. For all other stages,
Figs. 2 and 4 show experimental volumes as non vapor (may be liq-
uid or heavy plus liquid phases) and vapor phases. (The densities in
Figs. 3 and 5 for contact zero are those of the original oil and injec-
tant at cell conditions before mixing.) The pentane-insoluble con-
tent of the liquid phase in the Well I Y-9 oil experiment decreased
some in the first three stages (possibly because of asphaltene coating
on the cell walls), but a substantial decline did not occur until the
fourth contact. The EOS predicts heavy-phase formation starting in
the third contact. Overall, the EOS matched phase volumes, densi-
ties, and compositions (not shown) of both experiments very well.
The third is to determine how well the characterization predicts
data that are not used in its development. Fig. I shows the predicted
Pox diagram and experimental data for We1l2D-12 oil and 70/30 MI.
The only data in Fig. I used in the regression are at Point A. Table
2 shows phase volumes, densities, and pentane-insoluble weight
fractions for experimental conditions at Points A through D. Al-
though limited phase-boundary data are available, the characteriza-
tion appears to capture the broad extent of the three-phase region on
the Pox diagram. Within the three-phase region, the characterization
predicts quite reasonable phase properties.
Slim-tube displacements were simulated in a lD compositional
simulator equipped with three-phase flash capabilities. For Weill E-8
oil displaced by lean gas enriched with EF-I at 3,200 psi, the pre-
dicted MME is 38% vs. an experimental MME of33%. For Well1E-8
oil displaced by an MI blend with EF-2 (a heavier enriching fluid) at
2,825 psi, the predicted MME is 26% vs. an experimental MME of
28%. Additional data indicate that lighter Kuparuk oils have lower
MME's for the same enriching fluid; however, the characterization
does not capture this behavior. Though the dynamic model has some
shortcomings, the predicted three-phase displacement mechanism is
shown to agree qualitatively with micromodel observations.
Three-Phase Displacement Mechanism
The displacement mechanism for Kuparuk enriched-gas displace-
ments in a three-phase environment was investigated by simulating
ID slim-tube displacements. Fig. 6 presents simulation results as
gray-scale density/saturation plots for displacements of We1l2D-12
oil by 77123 MI at 2,700,3,000, and 3,500 psig. In this figure, phase
density is proportional to the gray-scale intensity and phase satura-

(Original Oil) o Heavy Phase

o

0 Vapor o.~ / o

D Liquid '"'E ~
~
u

en
Q)
•• Heavy
non-Vapor
~
.s
en
Q)
0.6
Liquid Phase
•o
E :;:;
::l '00 (Injectant)
"0 c
>
/
Q)
0 0.4
Q)
en Q)
ttl en Vapor Phase
.c ttl
.c
a... a... ~ ! !
0.2
Solid symbols - Expt.
Open Symbols - EOS
%~----~----~2~----~3------4~----~5----~6
Stages 1 through 6
Stage
Fig. 4-Phase volumes for multlcontact experiment; Wen1V·9 Fig. 5-Phase densities for multicontact experiment; 1V·9 oil
011 and 70/30 MI at 2,800 psig and 160°F. and 70/30 MI at 2,800 psig and 160°F.

104 SPE Reservoir Engineering, May 1995

Pressure. 2700 psi tion is proportional to the bar height in each cell. Simulations are
~ Lean Gas Front
shown for 200 cells in the displacement direction. The displacement
~t

~ .... ' ReservOir' OJI

moves from injection on the left to production on the right. The den-
___ HPO)IYl!qul:!-_ ~ _~-~ _ _ _ _ ~~~_ _ _ _ _ __ sity/saturation profiles are shown at 0.25 HPV injected.
Simulation results for the three-phase system show features char-
Pressure 3000 psi
!!!

. ~- ~ ~
acteristic of a two-phase condensing/vaporizing (mixed) mecha-
nism. 22 In a two-phase system, the mixed mechanism is character-
ized by a lean-gas bank ahead of the displacement front resulting
'" ~----------- - from intermediate components condensing into the oil; a sharp, two-
PntSSUTO = 3500 psi
phase, near-miscible transition zone; and a small residual saturation

'.
.---------------- resulting from incomplete vaporization at the trailing edge of the
transition zone. The mixed mechanism is considered effectively
-"',%.~;,,'>_::.~0.h'<-- ~ _ __ miscible because it achieves a high unit-displacement efficiency.
In the three-phase system, a heavy-second-liquid phase is devel-
oped in the transition zone as MI contacts the reservoir oil. Inter-
Fig. 6-Oensity/saturation plots at 0.25 HPV injected for n/23 MI mediate hydrocarbon components (C3 through C8) condense from
displacing Well 20-12 oil.
the MI into the reservoir oil and heavier components (C8 +) transfer
from the oil to the heavy phase. Both these mechanisms shift the va-
por/oil equilibrium ratio (K value) closer to unity and reduce the
density and viscosity of the oil phase, achieving a high unit-dis-
1~ placement efficiency. The density of the heavy phase (Fig. 7) is also
0.8 Heavy Liquid reduced in the condensing region owing to the mass transfer be-
Reservoir Oil
tween the original reservoir oil and the heavy phase; however, the
heavy phase ultimately becomes stripped, leaving a residual similar
'B
" 0.6 to that seen with the two-phase mixed mechanism.
~
As Fig. 7 shows, the densities of the oil and MI phases approach

f :: mM~;~":":~:::::==..=.::..::..=..=..=.=..::..::..=..::.:::...
each other in the middle of the transition zone. For the Well 2D-12
fluid system at 2,700 psia, no vaporizing signature is evident with
the oil phase; only the heavy phase shows evidence of a vaporizing
mechanism (phase density increasing from right to left). At 3,000
psig, Fig. 7 shows that the density of the reservoir oil increases at the
oL-----~---~----~-----~---~ trailing edge of the transition zone, characteristic of the condensing/
o 0.1 0.2 0.3 0.4 0.5
vaporizing mechanism. At 3,500 psig, the vaporizing mechanism
Dimensionless Distance
becomes significant and the two liquid phases merge in the vapor-
ization region to form a single, heavy residual phase.
Fig. 7-Phase-density profiles at 0.25 HPV injected for n/23 MI
displacing Well 20-12 oil.
Micromodel-Displacement Experiments
Micromodel experiments were conducted to observe the phase-be-
havior and displacement mechanisms of Kuparuk enriched-gas
floods. The displacement experiments were conducted in 2D
etched-glass micromodels. The pore-level observations provide
qualitative measurements ofthe phase densities, interfacial tensions
(lFT's), and saturations during the displacement. The results pre-
sented in this paper are for displacements ofWe1l2D-12 oil by 77/23
MI at reservoir conditions of 3,000 psi and 160°F.
The pore network for micromodels used in this work is a graphics
pattern resembling the pore structure of porous media. Fig. 8 shows
a diagram of the micromodel. The simple network pattern was se-
lected to expedite the experimental program, recognizing that pore
geometry could influence pore-level displacement behavior. Pore-
throat dimensions for the micromodel range from 60 to 190 ,urn, and
network depth is .., 120 ,urn. Network-pore dimensions are approxi-
mately one order of magnitude larger than actual reservoir rock.
The simple etched-glass micromodels do not capture the detailed
pore geometry, topology, and mineralogy of reservoir rock. These
factors generally control pore-level fluid dynamics. The primary ob-
jective of this work was to investigate and to observe phase-behavior
mechanisms. Many of the micromodel-displacement experiments
used a slim tube upstream of the micro model to develop a transition
2.600" DIameter zone. Because fluid dynamics and phase behavior are coupled, the
micro model observations are viewed with some caution.

Micromodel-Displacement Observations. Observations from the

Kuparuk micromodel-displacement program are consistent with a
II , III near-miscible condensing/vaporizing displacement mechanism. At
the beginning of the displacement, a clearly immiscible vapor phase
1/4" Total Thlclmesa fingered through the pore network and no evidence of mass transfer
was immediately apparent. The initial vapor phase is a lean-gas
front that was developed in a slim tube placed upstream of the mi-
Fig. 8-Etched-glass micromodel design. cromodel and travels ahead of the transition zone as a result of high-

SPE Reservoir Engineering, May 1995 105

 Fig. 9-Lean gas fingering through pore network. Fig. 1O-Oillightens in color from condensation of intermediate
components from MI.

Fig. 11-Minimum density and vaporlliquid 1FT are achieved. Fig. 12-Low saturation of altered oil surrounding individual
grains and grain clusters.

er mobility. Fig. 9 shows lean gas fingering through the pore net-
work. Flow is from right to left in all micromodel pictures.
As the displacement progresses, mass transfer of intermediate
components from the enriched gas alters the composition and color
of the oil phase. The oil initially is dark and opaque, gradually light-
ens in color, and eventually becomes nearly transparent. Figs. 10
and 11 illustrate the compositional change associated with the con-
densing mechanism. The condensing mechanism was also evi-
denced (not shown) by swelling of oil that saturates isolated grain
clusters in the micromodel.
A dark hydrocarbon phase develops from the altered-oil phase
during the condensing stage of the displacement. The mass transfer
of intermediate components from the vapor to the liquid phase ap-
pears to shift the system phase equilibrium, causing the third hydro-
carbon phase to form. The third phase initially is entrained as small
(= 1 .urn) particles or droplets in the altered-oil phase but quickly
adheres to the surface and coalesces into small liquid droplets on the
pore network. Fig. 11 shows the third hydrocarbon phase "speck- Fig. 13-Saturation of altered oil decreasing as a result of reva-
ling" the surface of the pore network within the altered-oil phase. porization.
Evidence of a low vaporniquid IFT is observed as a result of the
compositional changes occurring in the system. The vaporniquid The low altered-oil saturation is subsequently observed to vapor-
interface becomes difficult to identify and deforms easily with fluid ize, leaving a dark residual-oil saturation similar in appearance to
movement. The reduction in IFT reduces the capillary pressure, al- the heavy third phase coating the grain boundaries throughout the
lowing the altered oil to drain easily from the network,leaving a low model. Vaporization of the remaining altered-oil phase is easily seen
saturation oftransparent liquid surrounding and wetting the individ- in the time-lapsed video recordings and is illustrated by contrasting
ual grains and isolated grain clusters. the saturation around isolated grain clusters in Figs. 12 and 13.

106 SPE Reservoir Engineering. May 1995

Discussion
The micromodel-displacement experiments were conducted at 3,000
psig where EOS simulation results predict a condensing/vaporizing
mechanism that is dominated by condensing behavior. The exper-
imental observations are consistent with these simulation results.
At higher pressures, the simulations indicate a more balanced
condensing/vaporizing mechanism. Micromodel-displacement ex-
periments were not conducted at higher pressures. However, the
multicontact experiment with Well1E-8 oil at 3,200 psi indicates a
condensing/vaporizing mechanism as shown in Fig. 3 by the densi-
ties of the upper liquid and vapor phases converging in Contacts I
through 5 (condensing mechanism) and then diverging in Contacts
6 and 7 (vaporizing mechanism). The density differences are small
because the multicontact experiments represent conditions only at
the inlet and near the beginning of the displacement process. Only
the condensing mechanism appears in Fig. 5 for the Weill Y-9 multi-
contact experiment at 2,800 psi (note, however, that the oil, injec-
tant, and number of contacts also differ).
Micromodel displacements were also performed with Well I Y-9
oil, and the displacement behavior was similar to that with Well
2D-12 oil. However, enriched-gas displacements of Well I Y-9 oil,
which is relatively low in pentane insolubles, resulted in significant-
ly smaller volumes of the third hydrocarbon phase, consistent with
EOS predictions. The heavy-phase saturation is largely controlled
by the quantity of pentane insolubles in the oil.
No significant plugging was observed in any of the displacement
experiments conducted in this study. The heavy residual phase is
limited in saturation and mobility and shows no indication of prefer-
ential deposition in pore-throat locations. This observation is con-
sistent with laboratory-displacement experiments and field experi-
ence because no injectivity problems have been observed.
Development of a transition zone limited the volume of the third
hydrocarbon phase that formed during enriched-gas displacements.
Experiments conducted with the slim tube upstream of the micromo-
del developed a transition zone and produced a minimal volume
( < 1% PY) of the heavy residual phase. Without a slim tube, substan-
tial volumes (5% to 10% PY) of the heavy phase were observed. The
transition zone provides a compositional gradient or "buffer" that iso-
lates the oil from intimate contact with the MI, thus avoiding the com-
positional environment conducive to the formation of the third phase.
These results are consistent with static PYT experiments where high
levels of mixing generate large volumes of the heavy phase and with
slim-tube and coreflood results where a transition zone is developed
and limited evidence of a third phase is found.
The heavy phase developed in the Kuparuk fluid system appears
as a liquid at reservoir conditions; however, the liquid phase may
contain aggregated asphaltene micelles. Experimental observations
indicate that the heavy phase forms in the compositionally altered-
oil phase and remains entrained for a short time before coalescing
and adhering to the surface of the pore network. After sufficient ma-
terial coalesces, liquid droplets with smooth, curved interfaces can
be identified clearly.
Conclusions
1. A three-hydrocarbon-phase system is observed in reservoir-
condition laboratory experiments for Kuparuk oil and enriched-gas
mixtures over a wide range of pressure and composition. The third
hydrocarbon phase is rich in pentane insolubles and heavier than the
original oil. Experimental evidence indicates that the third-phase vol-
ume is proportional to the amount of pentane insolubles in the oil.
2. For the Kuparuk fluid system, the formation of a third hydro-
carbon phase is a macroscopically reversible process. Constant-
composition experiments traversing the three-phase region show
the appearance and disappearance of the third hydrocarbon phase.
3. A three-phase EOS fluid characterization was developed for
Kuparuk-reservoir-oillenriched-gas mixtures. The characterization
matches quite well phase volumes, compositions, and properties
measured in multicontact experiments with both heavy (WelllE-8)
and light (Well 1Y-9) oils. This is perhaps the first instance where
a cubic EOS is successful in matching and predicting the phase be-
SPE Reservoir Engineering, May 1995
havior in static experiments containing a third equilibrium phase
rich in pentane insolubles and heavier than the original oil.
4. The fluid characterization has some difficulty capturing the dis-
placement behavior in slim-tube experiments. With the EOS charac-
terization in ID simulations, the MME predictions have a higher error
(± 5% enrichment) than the experimental uncertainty ( ± 2%).
5. With the EOS characterization, ID simulations show a three-
phase, near-miscible condensing/vaporizing mechanism for en-
riched gas displacing Kuparuk oil.
6. Observations from micromodel-displacement experiments
provide qualitative validation of the phase-behavior and displace-
ment mechanisms predicted by the three-phase EOS fluid character-
ization.
Acknowledgments
We thank the managements of Arco E&P Technology, Arco Alaska
Inc., and BP Exploration (Alaska) Inc. for their support ofthis work
and permission to publish this paper. We also thank our coworkers
for their contributions, in particular, Andy Spence, Russ Bone, Jeff
Freeman, Jim Bradford, Wayne Kriel, and Jerry Haley for their roles
in performing the PYT experiments and Tak Sing Lo, Kishore Mo-
hanty, and Gary Jerauld for their contributions to the micromodel
work.
References
1. Bokhari, S.W. et al.: "Early Perfonnance and Evaluation of the Kuparuk
Hydrocarbon Miscible Flood," paper SPE 24930 presented at the 1992
SPE Annual Technical Conference and Exhibition, Washington, DC,
Oct. 4-7.
2. Mohanty, K.K., Gupta, A., and DeRuiter, R.A.: "Pore-Level Mecha-
nisms of Residual Oil Fonnation During Miscible Displacement," J.
Colloid & Interface Sci. (1994) 163, 199.
3. DeRuiter, R.A., Nash, L.J., and Wyrick, M.S.: "Solubility and Displace-
ment Behavior of C02 and Hydrocarbon Gases With a Viscous Crude
Oil," SPERE (May 1994) 101.
4. Mohanty, K.K. et al.: "Role of Three-Hydrocarbon-Phase Flow in a Gas
Displacement Process," paper SPE 24115 presented at the 1992 SPEI
DOE Symposium on Enhanced Oil Recovery, Tulsa, April 22-24.
5. Shelton, J.L. and Yarborough, L.: "Multiple Phase Behavior in Porous
Media During C02 or Rich-Gas Flooding," JPT (Sept. 1977) 1171.
6. Turkek, E.A., Metcalfe, R.S., and Fishback, R.E.: "Phase Behavior of Sev-
eral C02/West Texas Reservoir Oil Systems," SPERE (May 1988) 505.
7. Negabiban, S. and Kremesec, V.J.: "Development and Validation of
Equation-of-State Fluid Descriptions for C02-Reservoir Oil Systems,"
SPERE (Aug. 1992) 363.
8. Stalkup, F.1. Jr.: Miscible Displacement, Monograph Series, SPE, Rich-
ardson, TX (1983) 8.
9. Fussell, L.T.: "A Technique for Calculating Multiphase Equilibria,"
SPEJ (Aug. 1979) 203.
10. Leontaritis, K.J. and Mansoori, G.A.: "Asphaltene Flocculation During
Oil Production and Processing: A Thennodynamic-Colloidal Model,"
paper SPE 16258 presented at the 1987 SPE IntI. Symposium on Oilfield
Chemistry, San Antonio, Feb. 4-6.
11. Hirschberg, A. et al.: "Influence of Temperature and Pressure on As-
phaltene Flocculation," SPEJ (June 1984) 283.
12. Burke, N.E., Hobbs, R.D., and Kashou, S.F.: "Measurement and Model-
ing of Asphaltene Precipitation From Live Reservoir Fluid Systems,"
JPT(Nov. 1990) 1440; Trans., AIME, 289.
13. Baker, L.E., Pierce, A.C., and Luks, K.D.: "Gibbs Energy Analysis of
Phase Equilibria," SPEJ (Oct. 1982) 731.
14. Michelsen, M.L.: "The Isothennal Flash Problem. Part I. Stability,"
Fluid Phase Equilibria (1982) 9, 1.
15. Michelsen, M.L.: ''The Isothermal Flash Problem. Part II. Phase-Split
Calculation," Fluid Phase Equilibria (1982) 9, 21.
16. Jhaveri, B.S. and Youngren, G.K.: "Three-Parameter Modification of
the Peng-Robinson Equation-of-State To Improve Volumetric Predic-
tions," SPERE (Aug. 1988) 1033.
17. Buckley, J.S.: "Multiphase Displacements in Micromodels," Interfacial
Phenomena in Petroleum Recovery, Surfactant Science Series, Norman
R. Morrow (ed.), Marcel Dekker Inc., New York (1990) 36,157-189.
18. Mahmood, S.M. and Maerefat, N.L.: "New Techniques of Pore-Scale
Visualization of Fluids in Porous Media: The Effect of Pore Structure on
Fluid Distribution," topical report, Fossil Energy, U.S. DOE, Contract
No. NIPER-372 (DE89000731), NIPER, Bartlesville, OK (1989).
107
19. Katz, D.L. and Firoozabadi, A.: "Predicting Phase Behavior of Conden- Sanjay P. Godbole is a technical specialist at BP Chemicals in
sate/Crude-Oil Systems Using Methane Interaction Coefficients," 1PT Cleveland, OH, with research interests in phase-behavior mod-
(Nov. 1978) 1649; Trans., AIME,265. eling and simulation. Previously, he was a senior research engi-
20. Twu, C.H.: "An Internally Consistent Correlation for Predicting the neer at Arco Oil & Gas Co. in Plano, TX, and an associate profes-
Critical Properties and Molecular Weights of Petroleum and Coal-Tar sor of petroleum engineering at the U. of Alaska, Fairbanks.
Liquids," Fluid Phase Equilibria (April 1984) 137. Godbole holds a BS degree in chemical engineering from U. of
21. Lohrenz, J., Bray, B.O., and Clark, C.R.: "Calculating Viscosities of Bombay, MS degrees in chemical and petroleum engineering
Reservoir Fluids From Their Compositions," 1PT (Oct. 1964) 1171; and a PhD degree in chemical engineering from U. of Pitts-
Trans., AIME, 231. burgh, and an MS degree in engineering management from U.
22. Zick, A.A.: "A Combined CondensinglVaporizing Mechanism in the of Alaska. Kent J. Thele is currently pursuing a career in second-
Displacement of Oil by Enriched Oases," paper SPE 15493 presented ary education. He previously was a senior research engineer for
at the 1986 SPE Annual Technical Conference and Exhibition, New Or- Arco E&P Technology in Plano and an engineer for Cities Ser-
leans, Oct. 5-8. vice. Thele holds a BS degree from the U. of Tulsa and an MS de-
gree from the U. ofTexas, both in petroleum engineering. Eric W.
51 Metric Conversion Factors Reinbold is a senior petroleum engineer for Arco Alaska Inc. in
Anchorage. He has experience in production and reservoir en-
acre x 4.046 873 E - 01 = ha gineering, with an emphasis on development drilling and mis-
o API 141.5/(131.5 + API)
0 = g/cm3 cible-gas injection projects and has also worked in the research
OF CF - 32)/1.8 = °c of miscible-gas injection processes. Reinbold holds a BS degree
psi x 6.894 757 E + 00 = kPa in petroleum engineering from the U. of Alaska, Fairbanks.
SPERE

Godbole Thele Reinbold

108 SPE Reservoir Engineering, May 1995