Evaluation of the Salinity Gradient Concept
in Surfactant Flooding
G.J. Hirasaki, SPE. Shell Development CO.
H.R. van Domselaar, Koninklijke/Shell E&P Laboratorium
R.C. Nelson, SPE. Shell Development Co.
Abstract
Salinity design goals are to keep as much surfactant as
possible in the active region and to minimize surfactant
retention. Achieving these is complicated because (1)
compositions change as a result of dispersion,
chromatographic separation of components distributed
among two or more phases, and retention by adsorption
onto rock and/or absorption in a trapped phase; (2) in the
presence of divalent ions, optimal salinity is not constant
but a function of surfactant concentration and
calcium/sodium ratio; and (3) the changing composition
of a system strongly influences transport of the
components.
A one-dimensional (ID) six-component finite-
difference simulator was used to compare a salinity gra-
dient design with a constant salinity design. Numerical
dispersion was used to evaluate the effects of dispersive
mixing. These simulations show that, with a salinity gra-
dient, change of phase behavior with salinity can be used
to advantage both to keep surfactant in the active region
and to minimize retention. By contrast, under some con-
ditions with a constant salinity design, it is possible to
have early surfactant breakthrough and/or large surfac-
tant retention.
Other experiments conducted showed that high salinity
does retard surfactant, and, if the drive has high salinity,
a great amount of surfactant retention can result. The
design that produced the best recovery had the
waterflood brine overoptimum and the drive underop-
0197-752018310006-8825$00.25
Copyright 1983 Society of Petroleum Engineers of AIME
486
timum; the peak surfactant concentration occurred in the
active region and oil production ceased at the same
point.
Introduction
The phase behavior of surfactant/oil/brine systems for
different salinities is shown in Fig. I. I
Low salinities, called "underoptimum" or "Type
lI( -)" phase behavior, are shown at the top of Fig. 1.
In this kind of system, surfactant is partitioned
predominantly into the aqueous phase.
High salinities, called "overoptimum" or "Type
1I( +)" phase behavior, are shown at the bottom of Fig.
1. In this kind of system, surfactant is partitioned
predominantly into the oleic phase. When the oleic phase
has a low oil concentration, the oil is said to be
"swollen" by the surfactant and brine.
At moderate salinities, the system can have up to three
phases and is called' 'Type III. " This is illustrated in the
middle of Fig. 1. The salinity at which the middle phase
has a WOR of unity is called "optimal salinity" because
the lowest interfacial tensions (1FT's) usually occur near
this salinity. 2
As salinity increases, there is a steady progression
from Type 1I( -) to Type III to Type lI( +) phase
behavior. The middle-phase composition moves from
the brine side of the diagram to the oil side. The two-
phase regions that correspond to the Type lI( -) and
Type lI( +) systems can be seen above the three-phase
region in Fig. 1.
SOCIETY OF PETROLEUM ENGINEERS JOURNAL
 PHASE TERNARY DESCRIPTIVE
VOLUMES DIAGRAM NOMENC LA TURE

SURFACTANT

OIL )a/a+b TYPE II (-)

II UNDER OPTIMUM"
MICRO-I\ b/ 0+ b LOW SALINITY
EMLSN'if

BRINE OIL

SURFACTANT

OIL
MICRO-
EMLSN.
BRINE
! a/a+b
e/e +1

d/c+d
TYPE m
1I0PTIMUM"
INTERMEDIATE SALINITY

BRINE OIL

SURFACTANT

TYPE II(+)
MICRO-} c/c+d
EMLSN. "OVER OPTIMUM"
HIGH SALINITY
BRINE} d/c+d

BRINE OIL

Fig. 1-Types of phase behavior at different salinities.

With sodium chloride brine, optimal salinity is
relatively constant, but complications are caused by field
brines that contain divalent ions, such as calcium and
magnesium. Then optimal salinity is a function of sur-
factant concentration and ratio of divalent to sodium
ions 3,4 and decreases to low values at low surfactant
concentrations. Exchange of divalent and sodium ions
with surfactant and clays further complicates the
dependence of optimal salinity on the composition of the
system.
Discussion of Simulation Methods
The following sections describe the simulator used, in-
cluding the equations for numerical dispersion, residual
saturations, and relative permeabilities.
The Finite-Difference Simulator
The basis for the ID six-component finite-difference
compositional simulator used in this work was described
JUNE 1983
by Pope and Nelson.5 Several important modifications
we made are described here.
In Pope and Nelson's work, optimal salinity was con-
stant. However, we used a "salinity requirement
diagram" 3 that describes optimal salinity as a function
of surfactant concentration. This change in optimal
salinity is a consequence of divalent ions interacting with
surfactant or of surfactant "pseudocomponents" parti-
tioning in different proportions. However, since these in-
teractions have not been described well enough yet by
any researcher, ion exchange effects were neglected in
our simulations.
The curves for optimal salinity, maximum salinity for
Type III phase behavior, and minimum salinity for Type
III phase behavior as functions of surfactant concentra-
tion were included in the simulator as empirical func-
tions. Phase behavior was determined as in Ref. 5.
1FT's between microemulsion and excess brine, aWM,
and between microemulsion and excess oil, a OM, were
487
z
o WETTING PHASE"
~
""
~ !-N-O:-N--W-E--T--T-IN-:G:--:PH~A-::S~E-:,;--~
<t.
o=>
;;;
UJ
""
CAPILLARY NUMBER, +
cr "r
(log scale)
Fig. 2-Trapping functions.
computed from empirical correlations of Reed and Hea-
ly.2 Separate functions, g(a) and f(a), were used to
describe residual wetting and non wetting phases, respec-
tively, as a function of the capillary number. 6.7 These
are shown in Fig. 2. Appropriate 1FT's for these func-
tions are described in the section on residual saturations.
Although the capillary number is a function of the total
mobility and velocity as well as the 1FT, the first two
variables are not mentioned in this discussion because
they are of only secondary interest.
Numerical Dispersion Calculations
The truncation error of the finite-difference approxima-
tion (numerical dispersion) was used to approximate the
effects of dispersion on a process. 8 For calculating a
single-phase miscible displacement with a forward-
difference approximation in time and single-point
upstream mobility, the length of the mixing zone, D.L M
(the distance over which the concentration changes from
10 to 90% of the injected concentration), was expressed
as
V t (AxIL-.1V I ) ] ,/,
D.L MIL=3.625 [ 2 ........... (1)
and
=3.625
V axiL] If,
2
[ ................ (2)
_1_-
In Eq. 2 it is assumed that the time step size, .1 VI , in Eq.
1 is so small that it can be omitted.
Let N denote the number of grid blocks. Then the mix-
ing zone at I PV of throughput is
D.LMIL= 3.625(2N) -'/,. . .................. (3)
488
SI
S2
R' Sarw B" B'
@
R'
DETRIL
Fig. 3-Ternary saturation diagram showing regions of mobile
phases. The inset at the upper right shows the ternary
composition; the numbers indicate the number of
mobile phases.
The effect of dispersion on the process was evaluated by
changing the number of grid blocks to approximate the
mixing zone length described by Eq. 3.
A solution with no dispersion, calculated by the
method of characteristics, 9 provided the limiting case of
the finite-difference solution as the dispersion ap-
proaches zero. Composition routes and velocities were
plotted on a ternary phase diagram to provide a priori
understanding of the effects of dispersion on a process.
All calculations were made without incorporating ad-
sorption and ion exchange. The emphasis of this work is
on better understanding of mechanisms involved in
maintaining good displacement efficiency with a finite
slug in the presence of dispersive mixing.
Residual Saturations
in a Three-Phase Region
Residual saturations and relative permeabilities should
change continuously as the composition of a system
changes from three phases to two. For example, as the
excess oil saturation vanishes, the remaining phases are
microemulsion and excess brine; and, as the excess brine
saturation vanishes, the remaInIng phases are
microemulsion and excess oil.
If we assume that an excess brine phase wets rock
preferentially to a microemulsion phase and that a
microemulsion wets preferentially to an oleic phase, then
(1) in the absence of an excess brine phase, the
microemulsion is the wetting phase; (2) in the absence of
an excess oil phase, the microemulsion is the non wetting
phase; and (3) when all three phases are present, the
microemulsion is a spreading phase between the excess
oil and the excess brine.
SOCIETY OF PETROLEUM ENGINEERS JOURNAL
 WATER

SAND
/,...,.i*--GRAIN

(a) OIL MICROEMULSION

(a)

WATER

MICROEMULSION
(b)
(b)

Fig. 4-Trapping of nonwetting phase(s): (a) shows trapping of Fig. 5-Trapping of wetting phase(s): (a) shows trapping of
microemulsion and oil by brine, and (b) trapping of oil by microemulsion and water by oil, and (b) shows trapping
microemulsion. of water by microemulsion.

Recent experimental observations by Reed 10 bring in-

to question the preceding assumptions about the order of
wetting. His data show that (I) at low salinities, the
microemulsion can approach being the wetting phase in
preference to the excess brine and (2) at high salinities,
the microemulsion can approach being the non wetting
phase in preference to the excess oil phase. However, the
approach to the change in state of wetting occurs as the

-/;".,
LOG a

composition of the microemulsion approaches that of the

excess phase. Thus, it may be just a case of two phases
changing the state of wetting as the phase compositions
approach one another. Also, because IFT's approach
tlawMi
zero as phase compositions approach one another, the
-S3
capillary number becomes large. This causes the relative
permeabilities of the two phases to near those of miscible
fluids (i.e., zero residuals and straight-line relative
permeability curves), in which case the state of wetting PRESENT
i MODEL
is immaterial.
The regions of mobile phases are shown in the ternary
saturation diagram of Fig. 3. The equations describing
the straight lines that are the boundaries of these regions
L _______ ~ ::__ SALINITY. CSE
____________J
also are shown. The inset at the upper right shows the
ternary composition, a part of which includes a three- Fig. 6-Relationships between 1FT, residual oil saturation, and
salinity.
phase region; the numbers indicate the number of mobile
phases. The inset at the bottom shows detail for small
values of microemulsion saturation. As the microemul-
sion saturation approaches zero, the residual nonwetting-
phase saturation approaches the value of the waterflood
residual oil saturation, SOTW'

Trapping of Nonwetting Phases by Excess Brine

Phase. Trapping of non wetting phases (i.e., microemul-
sion and oil) by excess brine is illustrated in Fig. 4a.
This trapping is described by the nonwetting-phase trap-
JUNE 1983 489

SALINITY
Fig. 7-Residual microemulsion saturation as a function of
salinity.
ping function, f(a), and the appropriate tension is the
water/microemulsion 1FT, aWM. Thus the trapping is
described by
S3 +S2 =f(awM) .......................... (4)
and is shown as Line AN on Fig. 3.
Trapping of Wetting Phases by Excess Oil Phase.
Trapping of wetting phases (i.e., water and microemul-
sion) by excess oil is illustrated in Fig. 5a. This trapping
is described by the wetting-phase trapping function,
g(a), and the appropriate tension is the oil/microemul-
sion 1FT, a OM. Thus the trapping is described by
S3+SI=g(aOM) .......................... (5)
and is shown as Line BB' on Fig. 3.
Trapping of the Excess Oil Phase by the Microemul-
sion. Illustrated in Fig. 4b, this trapping is described by
the nonwetting-phase trapping function, f(a), and the ap-
propriate tension is the oil! microemulsion 1FT, a OM.
This trapping then is
S2 =f(aoM) .............................. (6)
and is Line A" A'" in Fig. 3. This line intersects Line
AA' if aOM < aWM'
Trapping of the Excess Brine Phase by the
Microemulsion. Shown in Fig. 5b, this trapping is
described by the wetting-phase trapping function, g(a),
and the appropriate tension is the water/microemulsion
1FT, aWM. Then
S I = g (a WM) .......... .................... (7)
490
and is Line B"B'" in Fig. 3. Thi\ linc mtcr,cch Line
BB' ifawM < am..l.
Residual Saturation Functions. Thc\c arc dc,crihcd hy
three expressions:
(a WMJ
SIr =max ... (Xa)
[
(a OM) -S,
If(a OM)
SJr=max~
If(a OM) -S,
and
:(a WM )-S~
S3r=max
t 0
(aOM )-S I
.. (XC)
These functions are shown on the appropriate line., in
Fig. 3.
Dependence of Residual Saturation on Salinit~. All
previous discussion has been about three pha\e\ of in-
variant composition at a given sal inity. As sal inity
changes, microemulsion composition as well as the ten-
sion between it and the excess brine and excess oil
phases also change. These tension changes mean that
residual saturations calculated according to the preceding
discussion also must change. The residual oil saturation
as a function of salinity is illustrated in Fig. 6. where the
microemulsion saturation, S,. is assumed constant.
At low salinities, the oil/microemulsion 1FT is eon-
trolling, and the residual oil saturation is given by
S2r =f(a OM), which is Line A" A'" in Fig. 3. At slightly
above optimal salinity (where a OM = a WAf). the
water/microemulsion 1FT. aWM' becomes controlling.
Above this salinity, the residual oil saturation is given by
S2r=f(awM)-S" which is Line AA' in Fig. 3. These
curves show that the residual oil saturation is reduced at
higher salinities by a higher microemulsion saturation.
If one were measuring the 1FT of an oil drop contain-
ing a thin film of microemulsion, the shape of the drop
would be determined by the sum of the 1FT's,
aOM+aWM, denoted by the curve in the upper part of
Fig. 6. If this sum of tensions were controlling the
residual oil saturation. the residual oil saturation would
appear as the curve labeled "sum residual oil." The
curve labeled "present model" (the model we used)
closely approximates this curve, and both curves have a
minimum residual oil saturation close to optimal salinity.
The residual microemulsion phase saturation. S,,.. is
shown as a function of salinity in Fig. 7. Here the oil
saturation, S 2, is assumed constant. At high water
saturations, where the microemulsion is being trapped by
water, the residual microemulsion saturation is either
zero or S3r=f(awM)-S2' At low salinities. where the
water/microemulsion tension, a WM. is low. the residual
microemulsion saturation is zero. As salinity and the ten-
SOCIETY OF PETROLEU\l E;-";GI;-";EERS JOCR;-";AL
 ~----~----~----~r-----------------~ 1. ~-----------------;--~~----~------~t.

I I
I I
I I
I I kC
I I r2
I I
I I
I I
I I
c I I
kn I I
S3", S3
,/
,/
""
""
"" A
,/
"" ,/
,/

" S1 A'
- S2r------

Fig. 8-Brine relative permeability. Fig. 9-0il relative permeability.

sion increase, the residual microemulsion saturation
increases. This clearly shows the merit of having
low salinity in the drive-to avoid trapping the
microemulsion.
Relative Permeabilities in a Three-Phase Region
Relative permeabilities for three-phase surfactant
flooding differ from those of three-phase systems of
water, oil, and gas in several respects.
1. Endpoint saturations can change. 7
2. Saturations can become less than residual because
of phase behavior and mass transfer.
3. Curvatures of the relative permeability curves can
change. 7
4. Microemulsion relative permeability should
become the wetting-phase relative permeability at low
salinities and the nonwetting-phase relative permeability
at high salinities.
Excess-Brine-Phase Relative Permeability. The brine
relative permeability, k rl' is expressed as a function of
the normalized brine saturation, S 1 :
The brine relative-permeability exponent, EW, is an
empirical function of the capillary number, is equal to
the exponent value of the water relative permeability in
the absence of surfactant at low capillary numbers, and
approaches unity at large capillary numbers.
The surface of the brine relative permeability is shown
in Fig. 8 for when a WM > a OM, which means that Line
B"B'" does not intersect Line BB' (the same situation as
in Fig. 3). If the lines do intersect, then the isorelative
permeability contours are not straight lines but appear as
in the case for the oil phase shown in Fig. 9.
Excess-Oil-Phase Relative Permeability. The oil
relative permeability, k r2, is expressed as a function of
the normalized oil saturation, S2:
kr2 =k rF(S2)EO , ....................... (lOa)
- S2 -S2r
S2 = , ................. (lOb)
1-g(aOM )-S2r
and

krl =krf (Sr)EW , ........................ (9a)

- SI-Slr The oil relative-permeability exponent, EO, is an em-

SI = , ................... (9b) pirical function of a capillary number similar to EW. The
1 - f (a WM ) - SIr surface of the oil relative permeability is shown in Fig. 9
and for when a WM > a OM, which means that Line A" A'" in-
tersects Line AA'. If the lines do not intersect, then the
krf = l.O-(l.O-krt)f(awM )/S orw ' ........ (9c) isorelative permeability contours appear as straight lines.
JUNE 1983 491
~----------~~--~~~--------------~ 1.

100
TYPE I I (+)

80
:3
en
o
N 60
TYPE III
>-
f-
Z 40 2~}1~e~_~~~1~J]~__ --

---7~~::===-----------------~
-l
cr:
en
~/ SLUG
~~
20 ~
~
v
TYPE I I H

51 A' 00 . 01 . 02 . 03 . 04 . 05
-52r----- SURFACTANT CONC., VOLUME FRACTION
Fig. 10-Microemulsion relative permeability. Fig. ll-Constant salinity design with field brine, slug, and
drive all at 30% "D" sand water.

Microemulsion Relative Permeability. The micro-
emulsion relative permeability is also a function of a nor-
malized saturation, S3, but the endpoint relative
permeability should approach that of brine or oil in
limiting cases of w = 1.0 or 0:
The coefficient and exponent of the microemulsion nor-
malized saturation approach those of brine or oil as the
weighting factor, w, approaches unity or zero. The
microemulsion relative-permeability surface is shown in
Fig. 10.

[ C
kr3 = (w)k r1 +(l-W)kr2 C] (S3P
- , ........... (lIa) Comparison of Various Salinity Systems
and Simulation Results
when Discussion of System With
Underoptimum Salinity 5,7,9
y=(w)EW+(I-w)EO, ................. (lIb)
The process that exists throughout in an underoptimum
salinity environment is the simplest. The surfactant is in
- S3 -S3r the aqueous phase, so it is transported with the water.
S3= , ................. (lIc)
1 - Sir - S 2r - S 3r The drawback of this design is that 1FT's are usually not
low enough for good oil displacement.
S2T
w= , ....................... (lId)
SIT+S2T Discussion of System With Overoptimum Salinity
In overoptimum, Type II( +) systems, 1FT's usually are
not low enough to reduce the residual oil saturation, but
oil can still be displaced by oil "swelling" and
, .................... (lIe) "solubilization" mechanisms. 1 In fact, if the injected
surfactant concentration is great enough, piston-like
displacement can result. 9 These mechanisms have ap-
and peal because they do not require ultralow 1FT's.
Another paper by Hirasaki 9 showed that without
dispersion an overoptimum system with either a con-
tinuous or a finite slug can produce good oil recovery.
. . ................... (lIf) However, his calculations showed that with dispersion a
continuous slug gives good oil recovery, but a finite slug
becomes ineffective. Dispersion causes the peak surfac-
492 SOCIETY OF PETROLEUM ENGINEERS JOURNAL
 1.00
POl YMER
BREAKTHROUGH ~
0.8
>-
(.)
z <t 0.06
~ CONTINUOUS
-
(.)
LL
0.6 SLUG
LL
w
SURFACTANT
>- BREAKTHROUGH" L'lL/L
IT
w O. 4
> .29
0 0
(.)
W
IT
10% PV SLUGS
0.2 .51
.81
O. 0 L -_ _----''--~~"____~~"____~~"____~_~~
O. 0 O. 2 O. 4 O. 6 O. 8 1. 0 1. 2
CUM. I NJECT I ON
Fig. 12-Recovery performance of constant salinity system.
tant concentration to decrease, which retards the surfac-
tant because the partition coefficient increases and the
surfactant loses its ability to cause swelling of the oil.
Nelson and Pope I reported a laboratory experiment in
which a continuous slug of an overoptimum system ef-
fectively displaced oil. Later in this paper, we describe
our laboratory experiment with a finite slug and an
overoptimum salinity environment where only half the
oil and very little surfactant were produced.
Constant Salinity System
Discussion
Because ultralow 1FT's occur at such a narrow range of
salinities, it would seem desirable to formulate a surfac-
tant system that is optimal at the field brine salinity and
to flood with the slug and drive at the same salinity.
Then changes in salinity would be of no concern.
Such a system, called a "constant salinity" design, is
illustrated in Fig. 11. The field brine, slug, and drive are
all at 30 % SDSW. The system is close to optimal salinity
until the surfactant concentration drops below about 0.01
(volume fraction), when it becomes overoptimum, or
Type II( +).
Simulation Results
Recovery as a function of cumulative injection is shown
in Fig. 12 for a continuous slug and finite slugs with dif-
ferent levels of dispersion. As with the overoptimum
system, recovery is good with continuous injection but is
a strong function of dispersion with a finite slug. It is in-
teresting that surfactant breakthrough occurred at about
0.3 to 0.4 PV throughput for the continuous slug, but
surfactant was not produced with the finite slugs.
The concentration profiles in Fig. 13 illustrate the dif-
ference between the continuous and finite slugs at 0.5
PV injection.
For the continuous slug, the profile shows a low sur-
factant concentration all the way to the outflow end,
even though only 0.5 PV has been injected. This is
because the microemulsion phase in the three-phase
region is assumed to be a spreading phase between the
brine and oil. If the oil saturation is sufficiently large, as
JUNE 1983
z
0
i=
u
0::
LL
~
Q = 0.5 PY
0
> 0.04
0 CONTINUOUS SLUG
z
~
10%
u PY SLUG
f- 0.02
z
<t
f-
U
~ 0.00
0::
:::J 0.0 0.2 0.4 0.6 0.8 1.0
(j)
DISTANCE, FRACTION
Fig. 13-Concentration profiles of constant salinity system.
in the oil bank, then the residual saturation of the
microemulsion phase is assumed to be zero. Also, since
the polymer is assumed to partition entirely into the most
water-rich phase,5 the microemulsion phase no longer
contains polymer for viscosity control. Thus, the
microemulsion phase with zero residual saturation and
low viscosity quickly travels through the oil bank. In this
case, the microemulsion phase without polymer has a
viscosity of 3 cpo
For the finite slug, the profile in Fig. 13 shows that
most of the surfactant is at a concentration of about 0.01.
As discussed and shown in Fig. 11, this is the approx-
imate concentration at which the system becomes Type
II( +). Thus, most of this surfactant was trapped by be-
ing absorbed into the trapped oleic phase, the same thing
that occurs in an overoptimum design. This result is con-
sistent with the observations of Glover et al. 4
The performance we simulated with a continuous slug
helps explain why the surfactant ran ahead of the
polymer in Hill's series of experiments in which the slug
depended on polymer for viscosity. * Fig. 14 shows the
phase volumes and some viscosities when the slug was
diluted with preflood water and equilibrated with crude
oil. The preflood had 50% more field brine than the slug
and contained some added sodium chloride. The slug
had a viscosity of about 50 cpo At the onset of three
phases, the microemulsion had a viscosity of 3 cp, but
the excess brine phase had more than 2,000 cpo
The effluent history for a similar system is shown in
Fig. 15.* The experiment was designed with a salinity
contrast in which the preflood, slug, and drive salinity
were approximately 7.5, 5, and 1 % formation water,
respectively. The surfactant front (50% of injected con-
centration) arrived at about 0.25 PV throughput ahead of
the polymer front. The surfactant breakthrough was
earlier than that which would be expected with stable
displacement. The drive caused the effluent concentra-
tion to overshoot the injected concentration. Apparently,
a minor, polymer-rich phase becomes a major phase
when contacted by the drive and then is readily displaced
by the drive.
·HiII, H.J.: personal communication, Shell Development Co., Houston (1975).
493
 1.0 .--------------.-------,--------------~
I
I
100 ORIGINAL RESERVOIR
OIL u = 6 cp
? TYPE I I (t)
I
0.8
80
3:
\.I = 50 cp [f)
z
o 0.6 0 TYPE I I I
;-
() \ a.e 60
([
a: \
u.
\.I = 3 cp
-;
w f-
I: 1'1 .....
::J
<5 Z 40 5RLl N~-----·'-.-----
0.4 .,. \\,\~\..._----- ... .
>
MICRO-
-l
IT
O~:.--- ..... :®
....... · .... · .. · .... · .. · .. ·~·LUG)
14 cp [f)
EMULSION
20
0.2
BRINE
TYPE I I (-)

\.I 2000
o - __ I
0~G-G-~~~~~L-----~------~------~
> cp
o .01 .02 .03 .04 .05
o 0.2 0.4 0.6 0.8 1.0 SURFACTANT CONC., VOLUME FRACTION
Fig. 14-Dilution diagram of polymer-thickened slug with Fig. 16-Salinity gradient design.
preflood brine. (Courtesy H.J. Hill, Shell Development
Co.)

1.6 . - - -_ _, - -_ _---,____---,____---,____- ,_____ Toconfinn that early surfactant breakthrough resulted

1. 4
CHEM I CAL
SLUG

.-----. CI -,
---I
l
MEQ/GM
I
.. j
,0-

--- .
'
from low microemulsion viscosity caused by lack of
polymer, we conducted a simulation in which the
microemulsion phase had enough viscosity without
......---. M++ ,,
•. _ ... SURFACTANT
polymer (i.e., an inherently viscous system). In this
• - - . POLYMER ~OLYMER-=-r
1.2
~ fo,OILCUT

j
. -.-. DRIVE case, the surfactant front was stable and surfactant did
not break through early.
,f

Therefore, to be effective a constant salinity design re-

, " .......
1. a ~/'\ quires that the microemulsion have sufficient viscosity
1 \ without polymer and that the optimal salinity re-
, "
main above the design salinity at low surfactant
i3
~ 0.8 /: \ concentrations.
,u
o SURFRCTANT i :~MER \
Salinity Gradient System
u
.} 0.6 L-f' i 4.
3.
a
a
2. a
Discussion
\ 1. a
The salinity gradient design has a surfactant fonnulation
0.4 a
\ O.
with an optimal salinity range that is less than the salinity
\ of the field brine and a drive salinity that, at most, is
0.2 equal to the optimal salinity extrapolated to zero surfac-
\
tant concentration (see Fig. 16).
The idea is to have overoptimum salinity ahead of the
0.5 1. a 1.5 2. a 2.5 3. a
surfactant bank and a drive salinity that is less than or
CUMULATIVE INJECTION. PV FROM STAAT OF SLUG equal to the lowest value of the optimal salinity; then,
with dispersive mixing, salinity will become optimal
Fig. 15-Effluent history of system with polymer-thickened slug. somewhere in the surfactant bank. Overoptimum salinity
(Courtesy H.J. Hill, Shell Development Co.)
ahead of the surfactant bank produces a Type 1I( + )
system, which retards transport of surfactant, as already
discussed in the section on an overoptimum design. Low
salinity in the drive produces Type 1I( -) behavior,
which causes surfactant to be transported in the aqueous
phase.
Thus, the salinity gradient design has the merits of all
three designs previously discussed-(l) the salinity pro-
file passes through the optimal salinity; (2) the design
slows the movement of surfactant into the high-salinity
region ahead of the surfactant bank; and (3) the design
does not allow surfactant to be trapped behind the surfac-
tant bank, because the surfactant is in the aqueous
phase-and none of the disadvantages.
494 SOCIETY OF PETROLEUM ENGINEERS JOURNAL
 ~L I L TYPE TYPE TYPE
1.00 .29 1.0 III~_ . III +1_11+)

10% PV SLUG .51 0.8

.81
0.8 >-
0>- >- 0.6r
u
Z Z
w ......
u 0.6
« 0.41-
V)

u.
u.
W
0.2
0>-
rr
w O. 4 O.O! L L I ..J L . I 1 ---.J
> 0.0 0.2 0.4 0.6 0.8 1.0
0
u
w
rr >-
0.2 L

~
O. 2 z
~
U i
~
0.0 L -_ _ _ _L -_ _ _ _L -_ _ _ _ _ _ _ _ _ _ _ _ _ _
0.0 0·2 O. 4
~

O. 6
~

O. 8
CUM. INJECTION, PV
~

1. 0
~

1.2
""
:::>
V)
oJ0.0 02 04 06 08 1.0
I

Fig. 17-Recovery performance of salinity gradient system.

0.4'-

~
0

Simulation Results «
>-
0.2~
0
The recovery perfonnance of a lO%-PV slug with a >-
salinity gradient design is shown in Fig. 17. Dispersive
mixing has only a slight effect on recovery compared 00 ~-'-=_ ~_._~ __ ~_~_ I I
0.0 0.2 0.4 0.6 0.8 1.0
with other designs. As expected, breakthrough of surfac-
tant and polymer occurs earlier when dispersion is DISTANCE
increased. Fig. 18-0il, surfactant, and salinity concentration profiles for
Concentration profiles for salinity, surfactant, and oil salinity gradient system.
are shown in Fig. 18. The salinity wave has traveled
about half the length after 0.5 PV injection. The peak
surfactant concentration is in the region of changing
salinity at a salinity that is close to optimum. The rapid T Y P E II (+1 0 E S I G N
reduction in oil concentration shows that displacement
occurred near the region of optimal salinity and peak sur-
factant concentration. SLUG
NO DISPERSION
The bottom of Fig. 19 shows surfactant concentration Q=i.QPV

.29
profiles for a salinity gradient design at 0.5 PV injection It.LIL
with different levels of dispersive mixing. The top of the . 81 =~--,''-----=~

figure shows profiles for an overoptimum, Type II( +) DISTRNCE

design at 1.0 PV injection. 9 The peak surfactant concen-
SALINITY GRRDIENT DESIGN
tration is affected by the dispersion level in both cases;
but, in the salinity gradient design, the location of the
SLUG
peak is not affected. AL/L
a = 0.5 PV .29
The relationship between the salinity profiles and sur-
factant concentration is shown by a crossplot of salinity
vs. surfactant concentration superimposed on a salinity
requirement diagram. In Fig. 20, three different times 00 O. I 0.2 O. 3 O. 4 0.8 0.9 1.0

during the flood have been crossplotted. As expected, DISTRNCE

the peak surfactant concentration decreases as the flood

progresses. The significant observation is that peak sur- Fig. 19-5urfactant concentration profiles for Type II( +) and
factant concentration occurs near optimal salinity. salinity gradient systems.
Factors affecting the front mixing zone of the surfac-
tant bank were investigated further. Simulation of a con-
tinuous slug injection showed that the surfactant front
was not nearly as advanced as in the case of constant
salinity. This shows that high salinity and the resulting
Type II( + ) phase behavior ahead of the surfactant bank
retard the surfactant front. Another simulation was con-
ducted in which a finite slug and the microemulsion had
sufficient viscosity without polymer (15 cp). In this case,
the surfactant front was much sharper than in the cases
JUNE 1983 495
 100 RESERVOIR
TYPE I I (t)
PV
80 100
...... ......
3:
...... ......
(JJ ......
0 80
60 TYPE I I I
~ ~
TYPE 11 1+1
.; 0>
I- 60 ...... ...... CHEMICRL
OPTIM~~_~~~!~~~Y
w
Z ...... SL~
z 40 TYPE III
~
-.J
IT
(JJ
SLUG~
'"
'"
~ 40 --
~
V">
TYPE III
20 ~
TYPE 11(-) 20

TYPE II I-I
I I oL-__ ____ __ ____ ____L -_ _ __
00 .01 .02 .04 .05
~ ~ ~ ~ ~

o 1.0 2.0 3.0 4.0 5.0 6.0

SURFACTANT CONC. SURFRCTRNT BLEND, VOL %

Fig. 20-Crossplot of salinity and surfactant concentrations Fig. 21-Salinity requirement diagram for the system in Run 1
superimposed on a salinity requirement diagram. (unmatched Bolt ratios).

shown in the bottom of Fig. 19.
Even more benefits of a salinity gradient design
become apparent when we examine phase behavior and
phase-trapping relationships as a function of salinity. In
the concentration profiles of Fig. 18, much of the oil
displacement occurred at a salinity higher than optimal.
We saw from Fig. 6 that the residual oil saturation in the
presence of a microemulsion phase reached a minimum
value at a salinity somewhat greater than optimal salini-
ty. We saw from Fig. 7 that the microemulsion residual
saturation was less at low salinities. In addition, at low
salinities the microemulsion has more water and less sur-
factant and oil; the microemulsion can even become
miscible with the brine. Thus it is desirable for
microemulsion trapping to occur at a low salinity.
Further Discussion of
Salinity Gradient Benefits
A salinity gradient design also has an advantage over a
constant salinity design when optimal salinity is a func-
tion of surfactant concentration, because some surfactant
components do not partition together. For example, in
the slugs containing various alcohols described by
Hedges and Glinsmann, 11 some of the alcohols parti-
tioned preferentially into the excess brine or the excess
oil phases rather than remaining in the microemulsion
with the surfactant. 12
As a result of chromatographic separation and dilution
from dispersive mixing, alcohol concentration in a
microemulsion will become less during displacement. If
loss of alcohol causes the system to become under-
optimum-Type II( -) phase behavior-then surfactant
will travel with the velocity of the aqueous phase. A
higher salinity ahead of the surfactant bank will help
restore the system to optimal salinity by dispersive mix-
ing. If loss of alcohol causes the system to become over-
optimum-Type II( +) phase behavior-then the surfac-
tant will travel with the velocity of the oleic phase,
which is the slower phase in the case of an aqueous slug.
If the drive has a lower salinity, then as the drive water
overtakes the surfactant-containing oleic phase, it will
496
help restore the system to optimal salinity.
Gupta and Trushenski 7 showed that, at low salinities
relative to optimal salinity, surfactant retention was
small but oil recovery was poor when 1FT's were too
high. At high salinities, recovery was poor because of
high surfactant retention. The best recovery occurred at a
salinity where 1FT was low enough but surfactant reten-
tion was not too high. With a salinity contrast they
showed that the best recovery took place when the
waterflood had high salinity and the drive had low salini-
ty. Nelson 3 has discussed their results with respect to a
salinity requirement diagram.
Pope et al. 14 conducted a number of simulations to
determine the sensitivity to a number of parameters.
They showed that the most important factor is the salini-
ty gradient. Overoptimum initial salinity and under-
optimum drive salinity produced the best recovery.
Limitations of Simulations
The purpose of this investigation was to compare alter-
nate salinity design strategies in the presence of disper-
sive mixing rather than to give an absolute prediction for
a given design. We have only a qualitative understanding
of some of the most significant parameters. Very little is
known about the appropriate level of dispersive mixing
for a field-scale design. A correlation from the literature
was used for the dependence of 1FT on phase behavior,
but we do not know how well this correlation predicts
tensions for our systems. The model for viscosity of
microemulsions is unverified. The effects of adsorption
and cation exchange were neglected. Data for the salinity
requirement diagrams were prepared with a constant
ratio of divalent to sodium ions in the brine, but the ratio
does not remain constant in typical experiments. Thus
the data had to be adjusted to obtain results similar to
typical experiments.
Subsequent work has investigated ion exchange in the
presence of surfactant 15 and the effect of divalent ions
and chromatographic separation on optimal salinity. 16
The gradients of surfactant components and/or cosur-
factant are a significant effect and should be considered.
SOCIETY OF PETROLEUM ENGINEERS JOURNAL
 40 , . . - - - - - - - - - - - - - - - - - - - - - - - - ,

[,OO~~~~~~-------------------.
if!.
Z 30 MATCHED BOLT RATIO EXPERIMENT
o
UNMATCHED ~
a:
0.80 a:
w
o o
BOLT RAnOS ::J
~
u 0:
a: "- (f) FIRST SECOND
o U
..J ..J 20
:r:
u z 0.60 DRIVE DRIVE
o o
o
z ~ ..J
a:
0: U
:::> ::J
~ 0 o
z 0 0.40
w a: (f)
10
..J 0. w
0: a:
> Z
;::; 0
UNMATCHED BOLT RATIO EXPERIMENT
0.20

OL-_L-~L-~ __ ~ __ ~ __ ~ _ _L_~

o [,0 2.0 3.0 4.0

O.OB ,--------------------------------,
UNMATCHED BOLT MATCHED BOLT
[. 00 r---\,'I~--~'k--------------------,
RATIO EXPERIMENT RATIO EXPERIMENT
o
u
MATCHED "-
U 0.06
0.80
w BOLT RAT IDS
o o z
u o
a: "-
o U
~
..J
:r: 0.60 U
u z ::J
o o 0.04
o o
z ~ a:
0: U n.
:::>
~ 0 ~
z
w a:
0 0.40 z
..J 0.
a:
~
0:
>
;::;
Z
0
~ 0.02
'"-
a:
O. 20 ::J
(f)

OL-~~~_~ __~===±___ L_ _L_~

o o [,0 2.0 3.0 4.0
o 0.5 1.0 1.5 2.0

CHEMICAL SLUG I DRIVE INJECTED, PV CHEMICAL SLUG/ DRIVES INJECTED, PV

Fig. 22-Effect on ion exchange of matching the Bolt ratio of the Fig. 23-Comparison of unmatched Bolt ratio experiments,
polymer drive to that of the formation brine. The top showing an extreme example of salinity-dependent
shows Run 1 (unmatched Bolt ratios): the bottom surfactant retention and its effect on oil production by
shows Run 2 (matched Bolt ratios). chemical flooding.

Experiments and Results isobutyl alcohol. It was made up in 50 vol % SDSW. A

Surfactant Retardation
Coreflood experiments described in the following sec-
tions show how surfactant transport rate through a core
depends on phase type. We found evidence that, during a
first drive, a surfactant-rich oleic phase that was
generated by Type 1I( +) conditions was trapped in the
core until a second, less saline drive changed the phase
type.
Two chemical floods were conducted at 168°F in
Berea cores that were 22 in. long and 2 in. square. The
oil in both was 27 vol % iso-octane and 73 vol % stock-
tank oil from Shell's South Pass Block 27, N 4 RB reser-
voir. The formation brine in both was synthetic "D"
sand water (SDSW) containing 120,000 ppm total dis-
solved salts, of which 3,000 ppm were multivalent cat-
ions. The chemical slug in both contained 5.43 wt%
Petro nate TRS IOB® (petroleum sulfonate), 1.10 wt%
NEODOL® 25-3S (alkyl-ethoxysulfate), and 1.00 wt%
JUNE 1983
chemical slug of 10% PV was used in both floods.
Run 1 (Unmatched Bolt Ratios). The drive contained
1,800 ppm of Kelzan MF® biopolymer in a 13 vol %
SDSW brine. As can be seen from a salinity requirement
diagram 3 for the system (Fig. 21), this salinity gradient
put much of the surfactant in the Type III region for most
of the flood. Furthermore, since the 2.3 Bolt ratio 13 of
the drive was considerably below the 6.2 ratio of the for-
mation brine, ion exchange in this flood was
favorable-that is, the reservoir clays removed
multivalent cations (M ++) from the flowing liquids as
the overall salinity decreased because of the salinity
gradient.
The magnitude of this effect can be seen in Fig. 22.
The concentration of M ++ in the effluent liquids
dropped to what it would have been in 6 % SDSW, even
though the only liquids put into the core contained
100%, 50%, and 13% SDSW. This favorable ion ex-
497
 0.4
BR[NE OIL BANK TRANS[T[ON ZONE
04 tant partitioned into the oleic phase). Essentially no oil
(TWO PHASES) and little surfactant were produced during injection of
+
i'l
. v,
v
0
the last 1.0 PV of drive. In fact, almost half the oil and
more than 93 % of the surfactant still remained in the
>
~
~ Z
0.. 0.3 ~
w
~
~
0.3 Q core.
z' ~ ~ ~
:;: Then we switched to the second drive, which was
'" it'"
I
Q it 0..
eo
::;: 0
w
0
Z
w
made up in 8 % SDSW. Because of its low salinity and
~

'"=> ~ I ~ V favorable ion exchange characteristics, the second drive

~
Z
::;: promoted Type lI( -) phase behavior. After 0.6 PV of
V) 0.2 028

0
~

z this new drive had been injected (1.6 PV of total drive

4.
=>
0 C9
=>
V)
w 0.[ ~
o[
'" :;<
v POLYMER DR[VE
""
~
v
0 0
0 0.4 0.8 12 1.6
CHEM[CAL SLUG I DR[VE [NJECTED, PV
Fig. 24-0il and surfactant production during experiment with
finite, overoptimum chemical slug and varying salinity.
change accentuated the salinity gradient.
Fig. 23 shows that oil recovery by this unmatched-
ratio flood was good. About 76% of waterflood residual
oil was produced, leaving a residual oil saturation, S orlt"
of about 0.08. Fig. 23 also shows that surfactant produc-
tion was normal. Surfactant concentration in the effluent
liquids peaked at about 6% of injected concentration,
just after 1 PV of chemical slug and drive had been in-
jected. Very little additional oil was produced after the
surfactant concentration peaked.
Run 2 (Matched Bolt Ratios). This was the same as
Run 1 except that two drives were used and enough
sodium nitrate (40 giL) was added to the first drive to
raise its Bolt ratio to 6.2. Fig. 22 shows that this match-
ing of the Bolt ratios for the first drive and the formation
brine eliminated the difference between CI- and M ++
production in the effluent liquids. That is, equalizing the
Bolt ratios suppressed ion exchange. 13
However, increasing the Bolt ratio of the first drive by
adding sodium ion increased the total salinity of the
drive; the sodium ion concentration was equal to that in a
46% SDSW brine. This increase in salinity plus the sup-
pression of favorable ion exchange meant that the first
drive promoted Type lI( +) phase behavior and, as a
result, oil production was not very good. Fig. 23 shows
that the system produced about 50 % waterflood residual
oil, leaving a residual oil saturation of about O. 17.
Very little surfactant appeared in the effluent liquids.
The peak in surfactant concentration, which occurred
just after 1 PV of chemical slug and drive had been in-
jected, was less than I % of injected concentration.
Essentially no additional oil and little more than trace
amounts of surfactant were produced after that small sur-
factant peak.
At this point in Run 2, 0.1 PV of chemical slug and 2
PV of the first drive had been pumped through the core.
That drive, with high salinity and suppressed ion ex-
change, had created a Type lI( +) phase environment
when it contacted the surfactant and oil (i.e., the surfac-
498
~ since any significant amount of oil had been produced), a
V
new oil bank emerged. As shown in Fig. 23, residual oil
:::
~ saturation in the core dropped during the next PV of sec-
V)
ond drive to about 0.08, the same residual oil saturation
after chemical flood achieved in Run 1.
Fig. 23 shows that a new surfactant bank also formed
in response to the less saline second drive. Surfactant
concentration in the effluent peaked at about 7 % of in-
jected concentration just after 1.0 PV of the second
drive. In other words, except for the later oil
breakthrough, which resulted from the oil saturation's
being lower at the beginning of second drive injection
than at the beginning of first drive injection, the second
drive behaved as though it were pushing a fresh chemical
slug because surfactants were trapped in the core while
the 2 PV first drive passed them by.
These data indicate, but do not prove, that surfactants
remobilized by the second drive had been trapped in the
0.17 PV of oil remaining in the core after the first drive.
Another possibility is that the high salinity of the first
drive caused high adsorption of surfactant on the rock,
most of which then desorbed into the less saline second
drive. Ref. 3 provides direct evidence that oil, not rock,
plays the major role in trapping surfactant under high-
salinity, Type lI( +) conditions.
Experiments by Glover et al. 4 for a similar system
show that much of the surfactant retention can be caused
by phase trapping. They also show that much of this re-
tained surfactant can be remobilized with a low-salinity
drive.
Detailed Observation of Salinity Gradient
In a laboratory flow experiment described in Ref. 3,
detailed observations were made to determine whether
surfactant is retarded by high salinity in and ahead of the
slug and then mobilized by a low-salinity drive, thus ac-
cumulating in a Type III environment. We also wished to
determine whether oil displacement takes place in a Type
III environment.
Results are depicted in Fig. 24. Near the bottom of the
figure we indicate that we injected a O.l-PV chemical
slug before switching to polymer drive. Across the top
we show that brine alone was produced during the first
0.27 PV of slug and drive injected. Then a clean oil bank
was produced at 42 % oil cut until surfactant
breakthrough at 0.82 PV of slug/drive injected. Surfac-
tant breakthrough came later in this experiment, as has
been our experience with less saline chemical slugs and
reservoir brines. This is consistent with the concept of
surfactant retardation caused by Type lI( + ) conditions in
the front mixing zone and in the chemical slug.
Phase relationships in the front mixing zone, as
observed in the effluent liquids, were as follows.
1. Initially, two phases were produced. As predicted
SOCIETY OF PETROLEUM ENGINEERS JOURNAL
from phase diagrams, the system was Type lI( +), an oil-
rich phase containing surfactant and some brine in
equilibrium with essentially pure brine. (The upper
phase was dull black, characteristic of oil-rich
microemulsions. The bottom phase was transparent and
pale yellow, characteristic of brine containing a small
amount of the more water-soluble surfactant species.)
The oil-rich-phase cut decreased from 30% to 25% to
18% in the three sample collection tubes showing Type
II( +) phase behavior.
2. Next, in agreement with the phase diagrams, three
phases (Type III) were observed in two collection tubes.
The upper phase was glossy black, identical to the oil of
the clean oil bank. The middle phase was a dull gray-
black microemulsion. The bottom phase matched the
bottom phase of the preceding Type lI( +) system. The
top, middle, and bottom phases occupied 5, 11, and
84 %, respectively, of the collection tube in which the
middle phase was most prevalent.
3. Then the collected samples again became two-
phase. However, the top phase was glossy black and the
bottom phase was translucent and orange-red. That is,
the effluent liquids were in a Type lI( - ) environment of
essentially pure oil on top of a brine-rich microemulsion.
The oil cut was 2 % in the first collection tube of this
Type II( -) region and became less in subsequent tubes.
These oil-cut figures and the residual oil saturation
plotted in Fig. 24 show again that oil production by sur-
factant after a three-phase region was practically nil,
even though most of the surfactant was traveling there.
However, although surfactant lag still appeared to exist,
the overoptimum (as injected) slug did deliver its peak
surfactant concentrations, CIC o, in a salinity range
necessary for Type III production.
Conclusions
1. The salinity design that always keeps surfactant in
an underoptimum or Type lI( - ) environment is
straightforward because surfactant is transported in the
aqueous phase. However, 1FT's are generally not low
enough for good oil displacement.
2. An overoptimum or Type lI( +) salinity design ob-
tains oil displacement by an oil-swelling mechanism
without ultralow 1FT. However, in the presence of
dispersion, a finite slug of this design quickly becomes
ineffective after the surfactant concentration decreases
from its peak.
3. A constant salinity design has the same salinity in
the formation brine, slug, and drive. This design is ap-
pealing because salinity is independent of all mixing
processes. However, with a continuous slug, the
microemulsion phase can travel rapidly through the oil
bank if polymer is required for mobility control of that
phase. With a finite slug, much surfactant can become
trapped in the residual oleic phase if, at low surfactant
concentration, optimal salinity decreases to below the
constant-salinity value.
For a constant salinity design to be effective, the
microemulsion must have sufficient viscosity without
polymer and the optimal salinity must remain above the
design salinity at low surfactant concentrations.
4. A salinity gradient design has the salinity decreas-
ing from the field brine to the drive. Optimal salinity oc-
curs near the peak surfactant concentration. Therefore,
JUNE 1983
the surfactant bank passes through regions of low 1FT's.
Unlike in a constant salinity design, surfactant is retard-
ed ahead of the surfactant bank because of the over-
optimum salinity there. Behind the surfactant bank,
salinity is low, so the surfactant travels either in the
aqueous phase or as a microemulsion with high water
content. Thus, the surfactant is not trapped by the drive.
Nomenclature
CIC o = concentration divided by injected
concentration, meq/g
EO = oil relative-permeability exponent
EW = brine relative-permeability exponent
fa = oil cut
f(a) = nonwetting-phase trapping function
g(a) = wetting-phase trapping function
k rj = relative permeability to phase j
k l = endpoint relative permeability
k rjw = endpoint relative permeability for
waterflood
L = system length, m (ft)
ALM = mixing zone length, m (ft)
N = number of grid blocks
Sj = saturation of phase j, %
Sjr = residual saturation of phase j, %
Sj = normalized saturation of phase j, %
SjT = trapped saturation of phase j
Sorw = waterflood water saturation, %
S Irw = residual water saturation, %
u = superficial velocity, m/s (ft/sec); in ula'A"
capillary number
V t = cumulative injection, PV
AVt = time-step size, fraction of grid-block PV
AX = grid block size, m (ft)
'A r = relative total mobility, l/poise (Pa' S - I )
1.1. = viscosity, cp (Pa' s)
a = interfacial tension, dyne/cm (mN/cm)
a OM = interfacial tension between excess oil and
microemulsion, dyne/cm (mN/cm)
a WM = interfacial tension between excess brine and
microemulsion, dyne/cm (mN/cm)
w = weighting function
Subscript
j = 1, excess brine; 2, excess oil; 3,
microemulsion
References
I. Nelson, R.C. and Pope, G.A.: "Phase Relationships in Chemical
Flooding," Soc. Pet. Eng. 1. (Oct. 1978) 325-38.
2. Reed, R.L. and Healy, R.N.: "Some Physico-chemical Aspects
of Microemulsion Flooding: A Review," Improved Oil Recovery
by SUlfactant and Polymer Flooding, D.O. Shah and R.S.
Schechter (eds.), Academic Press Inc., New York City (1977)
383-437.
3. Nelson, R.C.: "The Salinity Requirement Diagram-A Useful
Tool in Chemical Flooding Research and Development," Soc.
Pet. Eng. 1. (April 1982) 259-70.
4. Glover, C.J. et al.: "Surfactant Phase Behavior and Retention in
Porous Media," Soc. Pet. Eng. 1. (June 1979) 183-93.
499
 5. Pope, G.A. and Nelson, R.C.: "A Chemical Flooding Composi- 13. Hill, HJ. and Lake, L.W.: "Cation Exchange in Chemical
tional Simulator," Soc. Pet. Eng. J. (Oct. 1978) 339~54. Flooding: Part 3~Experimental," Soc. Pet. Eng. J. (Dec. 1978)
6. Stegemeier, G.L.: "Mechanisms of Entrapment and Mobilization 445~56.
of Oil in Porous Media," Improved Oil Recovery by Surfactant 14. Pope, G.A., Wang, B .. and Tsaur, K.: "A Sensitivity Study of
and Polymer Flooding, D.O. Shah and R.S. Schechter (eds.), Micellar/Polymer Flooding." Soc. Pet. Eng. J. (Dec. 1979)
Academic Press Inc., New York City (1977) 55~91. 357~68.
7. Gupta, S.P. and Trushenski, S.P.: "Micellar Flooding~Com­ 15. Hirasaki, G.J.: "Ion Exchange With Clays in the Presence of Sur-
positional Effects on Oil Displacement," Soc. Pet. Eng. 1. (April factant," Soc. Pet. Eng. J. (April 1982) 181~92.
1979) 116~28. 16. Hirasaki, GJ.: "Interpretation of the Change in Optimal Salinity
8. Lantz, R.B.: "Qualitative Evaluation of Numerical Diffusion With Overall Surfactant Concentration," Soc. Pet. Eng. J. (Dec.
(Truncation Error)," Soc. Pet. Eng. J. (Sept. 1971) 315~20. 1982) 971~82.
9. Hirasaki, GJ.: "Application of the Theory of Multicomponent,
Multiphase Displacement to Three-Component, Two-Phase Sur-
factant Flooding," Soc. Pet. Eng. J. (April 1981) 191~204. SI Metric Conversion Factors
10. Reed, R.L. and Healy, R.N.: "Contact Angles for Equilibrated
Microemulsion Systems," paper SPE 8262 presented at the 1979 ep X 1.0* E~03 Pa 's
SPE Annual Technical Conference and Exhibition, Las Vegas, OF (OF-32)11.8 °C
Sept. 23~26. in. X 2.54* em
E+OO
II. Hedges, J.H. and Glinsmann, G.K.: "Compositional Effects on
Surfactant Flood Optimization," paper SPE 8324 presented at the
1979 SPE Annual Technical Conference and Exhibition, Las "'Conversion factor is exact. SPEJ
Vegas, Sept. 23~26.
12. Salter, S.J.: "The Influence of the Type and Amount of Alcohol
Original manuscript received in Society of Petroleum Engineers office Feb. 18, 1980.
on Surfactant-Oil-Brine Phase Behavior and Properties," paper
Paper accepted for publication Oct. 22, 1981. Revised manuscript received March
SPE 6843 presented at the 1977 SPE Annual Technical Con- 30. 1983. Paper (SPE 8825) first presented at the 1980 SPEIDOE Enhanced Oil
ference and Exhibition, Denver, Oct. 9~12. Recovery Symposium held in Tulsa. April 20-23.

500 SOCIETY OF PETROLEUM ENGINEERS JOURNAL