Journal of Petroleum Science and Engineering, 2 (1989) 281-303 281

Elsevier Science Publishers B.V., Amsterdam - - Printed in The Netherlands

APPLYING FRACTIONAL FLOW THEORY TO SOLVENT

FLOODING AND CHASE FLUIDS

MARK P. WALSH and LARRY W. LAKE

Department o[ Petroleum Engineering, Texas A & M University, College Station, TX 77843 (U.S.A.)
Department of Petroleum Engineering, The University of Texas at Austin, Austin, TX 78712 (U.SA.)

(Received March 30, 1988; revised and accepted August 23, 1988)

Abstract

Walsh, M.P. and Lake, L.W., 1989. Applying fractional flow theory to solvent flooding and chase fluids. J. Pet. ScL
Eng., 2: 281-303.

The classification of and much of the understanding behind miscible flooding is based on fractional flow theory.
The purpose of this paper is to extend such understanding to the displacement of oil by a miscible solvent in the
presence of an immiscible aqueous phase. We pay special attention to the effects of simultaneous water-solvent
injection and of the behavior of chase fluids following a solvent-water slug.
Besides the general enhancement of understanding, practical results of this work are: (1) graphical procedures
which are valid for arbitrary water-solvent ratios and arbitrary initial conditions; (2) development of the notion of
an optimum water-solvent ratio (Caudle and Dyes' definition is valid only for secondary floods ); (3) a procedure for
sizing solvent slugs; (4) procedures for selecting the best chase fluid; and (5) the concept of and methods to compute
a water-solvent ratio in the chase fluid which minimizes solvent usage.
Our results must be interpreted within the limits of the fractional flow assumptions. Nevertheless, the insight
provided by the fractional flow solutions yields a strong base for interpreting the behavior of complex solvent floods.

Introduction gogical importance of fractional flow theory: (a)

the mathematics are relatively uncomplicated;
The easiest way to understand complicated (b) many times solutions can be obtained
displacements of fluids in permeable media is graphically; and (c) the solution usually forces
through fractional flow theory. This theory is an intuitive viewpoint which is useful even in
an application of a subset of the method of interpreting displacements which violate frac-
characteristics (MOC), known as coherent tional flow conditions. Fractional flow theory
wave theory (Courant and Hilbert, 1962; Helf- has also a role in process design that is fre-
ferich and Klein, 1970), to solve species con- quently overlooked. Demonstrating this role is
servation and constitutive equations in one-di- another objective of this paper.
mensional flow. Such solutions applied to the The history of fractional flow theory in oil
special case of miscible solvent flooding is the recovery analysis begins with the works of
subject of this paper. Buckley and Leverett (1942) and Welge ( 1952 )
There are three basic reasons for the peda- for waterflooding and continues through an

0920-4105/89/$03.50 © 1989 Elsevier Science Publishers B.V.

282

NOMENCLATURE impressive variety of processes and conditions

(Fayers and Perrine, 1959; Taber et al., 1961;
A cross-sectional area, L 2
Ci overall concentration of c o m p o n e n t i
--
Welge et al., 1961; Koval, 1963; De Nevers, 1964;
C,j = concentration (volume fraction) of c o m p o n e n t Wachmann, 1964; Davis and Jones, 1968; Pat-
i in phase j ton et al., 1971; Shutler and Boberg, 1972; Fos-
/j = fractional flow of phase j ter, 1973; Claridge and Bonder, 1974; Hales and
Fi = overall fractional flow of c o m p o n e n t i
Odeh, 1976; Larson and Hirasaki, 1978; Pope
g = gravitational c o n s t a n t
I = identity m a t r i x (appendix) et al., 1978; Larson, 1979; Helfferich, 1980;
k = absolute permeability Pope, 1980; Hirasaki, 1981; Stalkup, 1983; Du-
krj = relative permeability of phase j more et al., 1984; Paul et al., 1984; Cere and
L = system length Zanotti, 1985; Dai and Orr, 1987). Neverthe-
r~ s o p t i m u m solvent amount, PV
M = mobility ratio (Eq. 18)
less, only two works, those of Paul et al. (1984)
MOC = M e t h o d of characteristics and Stalkup ( 1983 ), deal specifically with mis-
N~ = n u m b e r of c o m p o n e n t s cible solvent flooding in the presence of mobile
Np ~- n u m b e r of phases water saturations. The work of Paul et al., the
NI,I) dimensionless cumulative oil recovered
most detailed application to date, gives a dis-
Nj = saturation of phase j
t = time cussion of the MOC and several example cal-
tD dimensionless time, PV culations including examples with Koval's
tDs solvent + water slug size, P V model (1963) of representing viscous finger-
U -- vector of d e p e n d e n t variables (appendix)
ing. This work has little variety with respect to
U volumetric flux or Darcy velocity
uj = d e p e n d e n t variable (appendix) initial and injected conditions, does not ad-
Ucl c o n c e n t r a t i o n velocity of i dress the issue of design, and, most impor-
w~ = simultaneous water-solvent ratio tantly, has no graphical analysis.
X position Our analysis is restricted to one-dimen-
XD fractional length
sional, dissipation-free displacements of oil by
Greek a miscible solvent in the presence of an immis-
= dip angle cible aqueous phase. Because in the limit of no
= t r a n s f o r m variable (appendix) dissipation, first-contact miscible displace-
= relative mobility of phase j
ments are indistinguishable from multiple-
PJ = density o f p h a s e j
= viscosity o f p h a s e j contact miscible displacements (Helfferich,
d = (operator) discrete change 1980; Hirasaki, 1981 ), our treatment is equally
= porosity applicable to all forms of developed miscibility
displacements. We include arbitrary initial and
Subscripts
cw = chase water
injected conditions, limited solvent solubility in
D = dimensionless the aqueous phase, trapped oil saturation and
i = components arbitrary chase fluid identity. These effects will
I -- initial condition lead to insights concerning solvent slug sizing,
j = phases
J = injected condition
optimal injected water-solvent ratios, and se-
o = oil lection of the best chase fluid.
OB = oil b a n k
OM = trapped oil phase
r = residual saturation
s = solvent
Assumptions
T = trapped oil phase
All theoretical analyses, fractional flow the-
w = water
ory more than most, rest on a set of simplifying
 283

assumptions. In this section we itemize these boundary or injected conditions consist of si-
assumptions and briefly discuss their multaneous injection of water and solvent (or
implications. water and chase fluid) which can only change
( 1 ) The flow is one-dimensional through an in a step-wise manner.
isothermal permeable medium. Unidimension- (6) At most there are three components
ality rules out viscous instabilities which are present: water, oil and solvent. Similarly, there
inherently multidimensional. The condition of are no more t h a n two phases: an aqueous phase
isothermal flow means we need to treat only which contains only water (we relax this in one
mass conservation equations. case below ), and an oleic phase which contains
(2) The rock making up the permeable me- solvent and oil but no water.
dium and all the fluids therein have properties Assumptions 1-5 imply that the species con-
that are independent of pressure. This is prob- servation equations will be reducible in the
ably the most useful assumption required by method of characteristics sense (Courant and
fractional flow theory because it obviates the Hilbert, 1962 ) so that coherent wave theory will
need to calculate pressures, but it is also the apply (see Appendix). Even so, the boundary
most controversial, particularly when applied conditions given in assumption 5 could gener-
to gas-like solvents. Some solvents such as su- ate regions of noncoherence within the me-
percritical CO2 are fairly incompressible and the dium. Such regions can be asymptotically tran-
approximation is reasonable. In other cases of sitory (Liu, 1977) but difficult to treat; hence,
high permeability a n d / o r low flow rates, pres- we avoid these cases.
sure gradients are small so that expansion of
the fluids becomes unimportant. W h e n these Theory
conditions are not met the fractional flow cal-
culations are in error; however, even here the In this section we discuss the basic equa-
insights gained from the solution are useful. tions, physical properties and give a sketch of
(3) The fluids flowing through the medium the solution technique.
are in local thermodynamic equilibrium, do not
react with the rock or solid phase, and mix ide- Basic equations
ally. The first part of this assumption means
that interphase mass transport reaches local Subject to the above assumptions the species
equilibrium at a rate substantially faster t h a n conservation equations for component i are:
the fluid flow rate. This assumption is often ex-
cellently approximated in displacements con- OCi OFi
F~x =o i=1 ..... ,No (1)
ducted at reservoir rates. Ideal mixing means atD
that we can effectively transform mass concen- where Ci and Fi are the overall concentration
trations into volume fractions, a concentration and fractional flow of component i, respec-
basis that is more convenient for flow in perme- tively, and are defined as:
able media. Np
(4) Dissipation is negligible. By dissipation Ci= 2S~C~j i = 1 ..... ,No (2)
we mean transport caused by both flowing- j=l
phase capillary-pressure gradients, as would Np
exist between immiscible phases, and disper- Fi= ~,fiC,~ i = 1,....,Nc (3)
j=l
sion/diffusion between components in a single
phase. where: Cij denotes the volume fraction of spe-
(5) Initial conditions consist of only water cies i in phase j; Sj is the saturation of phase j;
and oil at arbitrary but uniform saturation. The fj. is the fractional flow of phase j; N~ is the num-
284

ber of components; and Np is the number of is defined as the quotient of the relative perme-
phases. In our treatment, No is three and the ability krj and the phase viscosity #j. The re-
components are water, oil and solvent; Np is two maining variables are defined in the
and the phases are the aqueous and oleic phases. Nomenclature.
See Lake (1989) for a discussion of how Eq. 1 The oleic phase viscosity ~o and density p,,
is derived as a reduction of more general equa- are functions of Cso and the relative permeabil-
tions. The independent variables in Eq. 1 are ities kr~ are functions of both Sw and Cso. How-
dimensionless time and position: ever, our theoretical t r e a t m e n t does not depend
t x on the specific form of fw(Sw,Cso) - although

0
S dx
0
the specific solutions certainly do - and we be-
lieve that the cases discussed below are general
t D - OL.4' X D - A-L enough to encompass all possible forms of the
fractional flow relation. In addition, for our ap-
Dimensionless time defined above is the cu- plication of the MOC, coherent wave theory fw
mulative volume of fluid injected divided by the cannot be a function of position (except im-
medium pore volume, designated as PV in the plicitly through Sw and Cso). This means that,
following. The bulk fluid flow rate uA can even though Eq. 4 is rigorously correct for spa-
change with time but not with position owing tially-varying u, we must hold u constant or take
to the assumption of incompressibility. The di- c~= 0 in our examples. We take ~ = 0 in the ex-
mensionless position definition can accommo- amples that follow. Finally, we construct all
date a medium with a spatially-varying cross- water-solvent fractional flow curves from the
sectional area A (x) having an average value A;. water-oil curve by simply replacing the oil vis-
The variable cross-section feature of XD is the cosity by the solvent (or chase fluid) viscosity
basis for the combination of fractional flow so- in Eq. 4. This means that we assume that the
lutions with streamline models (Lake et al., k~j are functions of Sw only, according to Fig. 1.
1978).
Initial and boundary conditions
Fractional flow relations
In accordance with assumption 5, the initial
The essential nonlinearity in fractional flow condition (t D = 0 ) , denoted with a subscript I,
problems lies in the relationship between the is a uniform water saturation Swi. There is no
fractional flow of a phase and the saturation solvent present initially. The injected condi-
and properties of that phase. In this case the tion (XD= 0) is in general a two-phase mixture
relevant fractional flow is that of the aqueous of solvent and water specified by the water
phase which we take to be a function of the fractional flow at the injection end fwJ. The in-
water saturation Sw and volume fraction of sol- let water fractional flow is related to the water-
vent in the oleic phase C~o. The aqueous-phase solvent ratio by:
fractional flow is given by the following
WR
equation: fwJ - - - (5)
1 + WR
fw -
). rw -..l_- ~ ro ~ 1
)
r ? pg sinOL (4) where WR is the ratio of water to solvent in-
jected simultaneously, both expressed in res-
for a dipping reservoir with constant dip angle ervoir volumes. Water and solvent are nor-
~, where ~ ~1is the relative mobility of phase j, mally injected alternately in field practice, not
and Ap is the difference of the aqueous and oleic injected simultaneously as suggested by Eq. 5.
phase densities (Pw-Po). The relative mobility In laboratory experiments, where the frac-
 285
,o ~ /
OSw ~ Ofw OS~= o
OtD c)Sw (~XD
O(C~ofo) O(Cso[o) O(C~oSo)
=0 (6a)
OtD c)(CsoSo) (~XD
OI -

where we have expanded the spatial terms ac-

>-
cording to the chain rule for partial derivatives
.J
(Helfferich, 1980). Independently of this, we
W set the total differential for dSw and d (Cso fo)
I=
W OOl . . . . .
to zero and form the concentration velocities:
a.
tU
WATER - - OIL
I'-

-J
(OSw/OXD)
(dxD) (O(C,oSo)/OtD)
0 001 . . . . . . . . . . . C~oSo= - (O(C~oSo)/OxD) (6b)

Eliminating derivatives between Eqs. 6a and 6b

gives the final form for concentration velocities:

0
WATER SATURATION
I0
Vow= \OSwJx.
-- , vc~= [0O(C~oSo)
so o>1x,~ (7)

Fig.1.Relativepermeabilitycurvesusedinthiswork. Because of the nature of the dimensionless

variables XD and tD, the velocity in Eq. 7 is nor-
tional flow assumptions are most accurate, si-
malized by the bulk fluid velocity. Strictly
multaneous injection is common. Intuition
speaking, we should call this the specific con-
suggests that alternating and simultaneous in-
centration velocity.
jection become the same as the size of the water/
Equations 7 are the velocity for a continuous
solvent slugs decrease (at the same ratio);
variation in concentration. W h e n the concen-
however, we are not able to demonstrate this
tration change is abrupt or a shock, it travels
analytically. Dai and Orr (1987) have shown
with velocity:
small though persistent deviations between si-
multaneous and alternating solvent-water in- ~fw A( Csofo) (8)
jection in numerical simulations. In some cases Vjc,,-AS w ' v~c,-A(CsoSo)
we will follow the w a t e r / s o l v e n t slug with a
where the A's represent changes between the
water/chase fluid mixture whose inlet condi-
upstream and downstream value of the indi-
tion is defined analogously to Eq. 5. The sub-
cated quantity.
script K will denote chase fluid injection. In
The method of solution is based on the rec-
neither case oil is injected.
ognition that problems, as formulated here,
must satisfy the coherence condition (see Ap-
Method of solution pendix) which says that the concentration ve-
locity for all species must be equal. From Eq. 7
Coherent wave theory rests on the idea of we then have:
concentration velocities which are derived as
follows. First combine Eqs. 1-3 and apply as- dfw d(C~ofo)
sumption 6 to give: dSw - d (C~oSo) =vc (9)
286
The relationship of Eq. 9 is satisfied by two into the trapped oil saturation. Including these
fundamental relations between the dependent p h e n o m e n a changes Eq. 11 to:
variables (see Dumore et al., 1984 for a similar
fw-[1/(1-C~w)]
treatment of immiscible solvent flooding):
VC-Sw_{[l_SoM(l_C~T)]/(l_C~w)}
(1) For C~,, is constant the relation 9 is ob-
viously satisfied, since f,,= 1 -fw, and the con- (12a)
centration velocity is:
where Csw is the volume fraction of solvent in
dfw (10a) the aqueous phase, SOM is the miscible-flood
vc - dSw residual hydrocarbon saturation, CsT is the vol-
ume fraction of solvent in the residual phase
This velocity is simply a dimensionless form of and Vc is again constant. Equation 12a follows
the Buckley-Leverett equation and corre-
from making a solvent material balance around
sponds to an immiscible displacement. The
the miscible displacement front which is pre-
concentration velocity is the slope of the frac-
sumed to be a shock. We say "presumed" be-
tional flow curve at constant Cso. The concen-
cause we are unable to show that the miscible
tration velocity, being a function of saturation,
wave is indifferent if Caw is greater than zero
means that the wave corresponding to this
(Cere and Zanotti, 1985 ). The small water sol-
change can be a shock since Sw can change so
ubility of most commercial solvents probably
that vc increases in the downstream direction.
makes this assumption quite good. Equation
If so, Eq. 10a is replaced with a discrete version:
12a is of the form:
J/w fw - b (13a)
v jc - ~ (10b)
JSw v c - Sw - a
The graphical interpretation of the velocity in where:
Eq. 10b is a secant on the fractional flow curve
(C~,, constant) between the upstream and a=
1--SoM(1-- CsT) (13b)
downstream water saturations. For fractional 1 - C~w
flow curves with inflection points we are some- 1
times required to set Eqs. 10a and 10b equal to b=-- (13c)
1 - C~
determine a shock saturation.
As for Eq. 11, the interpretation of Eq. 12a is a
(2) If vc is a constant the equation:
straight line emanating from point
f, =L'(,S,, (11) (Sw,fw) = (a,b) and intersecting the solvent-
also satisfies Eq. 9. The changes represented by water fractional flow curve; the slope of this line
Eq. 11 are indifferent waves, waves where all is vc.
concentrations have the same velocity (Pope et A similar material balance on the oil across
al., 1978). Miscible displacement waves are in- the same wave gives:
variably indifferent. There is no analogous [w-1
shock equation for Eq. 11 because the wave is (12b)
VC--Sw_I_SoM(I_C~T)
already indifferent. The concentration velocity
Equation 12b is of the form:
in Eq. 11 is the slope of a line constructed from
the upper right corner of a fw vs. Sw plot through [w-1 (14a)
a point on the curve. VC - - S w - c

Several other interesting effects may be in- where:

serted at this point: solvent water solubility, a
trapped oil saturation, and solvent partitioning C= I - - S o M ( 1 - - C s T ) (14b)
 287

The interpretation of Eq. 12b is a straight line these diagrams will be discussed later.
emanating from point (Sw,fw)= (c,1) and in- The graphical solution procedure is greatly
tersecting the oil-water fractional flow curve; simplified for the special case where there is no
the slope of this line is vc. solvent solubility in the aqueous phase
Figure 2a illustrates the application of Eqs. (Csw=0). In this case, points (a,b) and (c,1)
12a and 12b for the case of a tertiary WAG dis- coincide, and Eqs. 12a and 12b are clearly iden-
placement with W R = 1; a trapped oil satura- tical. Hence, for the special case of Csw--0, a
tion given by SOM=0.15; solvent solubility in single line drawn from the point
the aqueous phase given by Csw=0.10; and sol- (Sw,fw) = [ 1 - S o M ( 1 - C s , r ) , l ] and intersect-
vent partitioning into the trapped oil satura- ing both the solvent-water and oil-water frac-
tion given by CsT = 0.20. The designation "ter- tional flow curves yields the solvent velocity vs
tiary" denotes that the displacement starts at as well as the oil bank oil saturation OB.
residual oil saturation Sw~= 1-Sor. Applica- In the remainder of the cases considered we
tion of Eqs. 13a, 13b and 14a yields values for neglect solvent solubility in the aqueous phase
a, b and c of 0.98, 1.11 and 0.88, respectively (C~w=0). As a result, the simplified solution
(see Fig. 2a). The injected fractional flow is lo- procedure just outlined applies. Given the ex-
cated at point J on the water-solvent curve. ample shown in Fig. 2, any examples neglecting
Equation 12a says that the solvent propagates solvent solubility in the aqueous and trapped
through the medium with a miscible wave of oil phases can be easily modified to include
velocity (Vc = vs in the figure ) which is the slope these effects.
of a straight line through the point J and the Equations 10 and 12 are the culmination of
point (Sw,/w)= (a,b). On the other hand, co- the mathematical development; the remainder
herence requires that this velocity is the same of this paper deals with their application.
as the velocity of the rear of the oil bank so that Though the equations are quite simple, the ap-
Eq. 12b says that a line through the point plication can be obscure because we have not
(Sw,/w) = (c,1) with slope vs must pass through specified the saturations or concentrations
the water saturation in the oil bank designated which go into the evaluation of the specific ve-
by OB. locities. These saturations and concentrations
The line from (c,1) of course passes through are case-dependent, but they all follow the gen-
several fractional flow curves (not shown on the eral rule that the correct solution avoids mul-
figure) between the water-oil and water-sol- tiple values in both the saturation and concen-
vent limits, but these all collapse to a single tration. To eliminate multiple values we apply
water saturation since the miscible displace- one or more of the following rules.
ment wave is indifferent. For waves emanating from a common origin:
Once on the water-oil fractional flow curve, (1) For the immiscible waves replace Eq. 10a
there is an immiscible displacement of condi- by Eq. 10b. This replaces continuous satu-
tion I by OB which follows the composition path ration changes with shocks in the same
given by either Eq. 10a or 10b depending on m a n n e r as in the Buckley-Leverett theory;
whether the velocity (the slope of the [w-Sw or:
curve) monotonically increases from I to OB. (2) For miscible waves select the saturations in
For the case shown in Fig. 2a the velocity de- Eq. 12a so that the line drawn from
creases and there exists a shock in saturation { [1--SoM(1--C~T) ],1} has a slope (veloc-
between I and OB with velocity Vc = VOB.Given ity) that is consistent with the other veloc-
the graphical solution shown in Fig. 2a, satu- ities in the displacement. A velocity is con-
ration profiles and effluent histories (Figs. 2b sistent for a given saturation if its position
and 2c) can easily be deduced. Application of is upstream of waves with greater velocities
288

and downstream of waves with lesser lows from the saturation velocity by allowing
velocities. tD to vary with fw (recall that the fractional flow
For sets of waves with different origins: curve insures a relationship between Sw and fw)
(3) Select time delays (slug sizes) such that while holding XD constant at unity. The ef-
these waves do not intersect wave begun fluent history also provides a means for calcu-
earlier. lating cumulative oil recovery from the area
For all cases studied thus far these rules are under the curve:
sufficient; however, a uniqueness proof does not tD
exist for the general R i e m a n n problem.
N,D = JFo ]xD=ldtD (15)
o
Results
where N,D is the cumulative oil recovered ex-
We present several cases of increasing com- pressed as a fraction of the medium pore
plexity. However to begin the development let volume.
us discuss the representations about to be Finally, Fig. 2d is a time-distance diagram
employed. which shows lines of constant saturation and
saturation changes on a plot of XD v s . t D. This
Graphical representations diagram is arranged so that it matches with Fig.
2b on the left and Fig. 2c above. Shaded areas
A major challenge in this work is to present on this diagram represent spreading waves and
the diversity of the solutions without undue dotted lines are miscible waves. On such a dia-
complexity. We also would like to retain the gram the slopes of the lines representing the
graphical nature of the solutions as much as waves are the specific velocities.
possible.
Figure 2, a composite of four smaller figures, Continuous solvent injection
illustrates how the results will be presented. The
upper left plot, Fig. 2a, illustrates the fractional Figure 3 shows the results for the case of a
flow curves upon which the graphical solution tertiary displacement using continuous solvent
will be constructed. The fractional flow curves injection, and WR=fwj = O. The initial and in-
shown in Fig. 2a and all subsequent figures are jected conditions are denoted by points I and J,
constructed assuming the oil, water, and sol- respectively, on the fractional flow curves. A
vent viscosities are 2, 1, and 0.1 m P a s, respec- miscible wave construction as in Fig. 2 cannot
tively. Immediately below this on Fig. 2b is a pass directly to J as there would t h e n be a dou-
profile, a plot of saturation and concentration ble intersection with the water-oil fractional
versus XD at a fixed tD. The profile is rotated flow curve which would violate the rule about
90 ° so that it will match up with the fractional single-valued saturations. Instead, there is a
flow curve above. On this diagram immiscible spreading immiscible wave from J which fol-
waves are represented by heavy solid lines and lows the water-solvent fractional flow curve in
miscible waves by wavy light lines. Saturation accordance with Eq. 10a. To find the termina-
profiles are constructed from the specific veloc- tion of this immiscible wave, point S, construct
ity definition v = XD/tD by holding tD constant the tangent to the water-oil fractional flow
and allowing XD to vary. curve as suggested by Eq. 12 and extend this
To the right of the fractional flow curve, Fig. line to the water-solvent curve. This construc-
2c, is an effluent history, a plot of fractional tion is permitted because we have not previ-
flux and concentration at the effluent end of ously specified the saturations in Eq. 12. The
the medium versus tD. The effluent history fol- oil bank saturation OB is now the saturation at
 289

(o) F R A C T I O N A L FLOW
CURVES (c) ELUTION HISTORY
Z~(c,I)2? ~(°'b)

OIL
WATER BANK SOLVENT

~__ t WATER- /'1 / I I

/ I
o WATER SATURATION 0 0 DIMENSIONLESS TIME i.o
I, I.Oi

7, -

LEGEND (a,b,c,d)
SOM : 0.15
wa> z o~ CSW = O. IO
l-WJ w C ST = 0 20
WR =1

0 , , , ,-0 ol I i i I 11.0
OIL SATURATION 0 DIMENSIONLESS TIME

(b) SATURATION PROFILE AT (,d) T I M E - D I S T A N C E DIAGRAM

0 , 2 5 PV I N J E C T E D

Fig. 2, Tertiary miscible displacement using a water-solvent ratio (WR) of unity and including a trapped oil saturation and
solvent solubility in water and trapped oil phases.

the tangent and we again follow the water-oil velocities in the slower wave indicate that it is
fractional flow curve to the initial condition I. very difficult to remove completely mobile water
In this case, however, the immiscible wave is from the medium by injecting immiscible
spreading from OB to OB', a saturation deter- solvent.
mined as the tangent from I and a shock there- Figure 3e shows the oil production curve, a
after, both of which are applications of Eq. 10. plot of recovery efficiency versus dimension-
The diagrams now exhibit two immiscible less time, where recovery efficiency here is de-
spreading waves: a spreading immiscible por- fined as pore volumes of oil recovered N p D n o r -
tion ahead of the miscible wave and a very small, malized by the initial oil saturation Soi. The
slow immiscible displacement of a solvent- curve in Fig. 3e follows from Eq. 15 and is given
water mixture by pure solvent. These are the in greater detail by application of the Welge
shaded fan-like regions in Fig. 3d. The small (1952) integration, or:
29O

N,D =Soi-So Ix.=1 +Fo I=D=ltD (16) and So I xD = 1 is the oil saturation corresponding
to Fo [ xD=1- Equation 16 applies for secondary
where Fo [ xD= 1 represents the effluent oil frac- floods as well as tertiary floods.
tional flux (oil cut) at tD pore volumes injected, Secondary displacements using continuous

1.0- (e) OIL PRODUCTION CURVE

>-
0
Z
w

b_
I

/,
w
>-
n~
w
>
0
OIL BT SOLVENT BT

V '/,
w
II
I I ' I I I
I DIMENSIONLESS TIME 1.O

(a) FRACTIONAL FLOW | l

CURVES (C) ELUTION HISTORY

][~1 ~I~SOM 1.0

0
.J
h
If;/', I x
Z
0
,,o.._../F/~
~= I V I / L
J/f;' 'I, ]I ,.-' WATER OIL SOLVENT
I'"
0 I/i/i
|/A I
; I
I I
rr
b- I//1! ,I
n,'
OB -- -----I
I--
O B ~

r ]HI
--~ . . . . . .

= I I I I
o1 i i i i
,%
0 WATER SATURATION i.0 0 DIMENSIONLESS TIME

_..,.o I LEGEND (e,b,c,d & e )

SOM -- O. I0
CSW = 0
CST = O
3( WR=O
Z
,,~
,"n Vs M = 30
.J
O
SOLVENT

i i i i i v , k

OI SATURATION O 0 DIMENSIONLESS TIME

I'.0
(b} SATURATION PROFILE AT (d) TIME-DISTANCE DIAGRAM
0.25 PV INJECTED

Fig. 3. Tertiary miscible displacement by water-free solvent.

 291

solvent injection and having Sw] = Sw, as the the displacement front which, in turn, governs
initial condition result in a piston-like dis- the severity of viscous fingering. Our proce-
placement of oil by solvent. Because this result dures also allow estimation of local displace-
is straightforward, we do not include a sche- ment efficiency and definition of an optimal
matic illustrating it. WR.
An important aspect of solvent floods is their Injection schemes range from injecting con-
inherently unfavorable or adverse ( M > 1) mo- tinuous solvent at one limit to injecting contin-
bility ratio. M > 1 results in viscously unstable uous water at the other. We have just examined
displacements in actual floods (Koval, 1963). the former limiting case and have observed that
To quantify the mobility ratio which exists be- very poor mobility ratios ensue. Clearly as WR
tween the oil bank and driving solvent, we de- approaches infinity, the displacement effi-
fine the effective displacement mobility ratio: ciency must begin to decrease at some point and
eventually approach that of a waterflood. Fig-
M- ( k r o / f l s ) + (krw/]Aw)solvent (17) ure 4 summarizes the effect of WR on the dis-
(kro/~to) nt- (krw/~tw)oil bank placement efficiency and attending mobility
where the term in the numerator represents the ratio. This figure shows WR across the top hor-
total relative mobility of fluids immediately up- izontal axis and the injected solvent fraction
stream of the indifferent solvent wave and the (1-/wj) across the bottom. The left vertical
term in the denominator represents the total axis plots the mobility ratio as defined by Eq.
mobility of fluids immediately downstream of 17; the right vertical axis plots the displace-
the solvent wave. We will use this definition of ment efficiency in terms of pore volumes of to-
mobility ratio to infer the likelihood and extent
WR
of viscous fingering. For the tertiary flood ex-
IO 4 2 I O5 0
ample shown in Fig. 3 the mobility ratio is 30.
LEGEND
For the case of the corresponding secondary
SOM : 0 I0
flood (not shown) the mobility ratio is simply CSW : 0
JO00 I0
a ratio of the oil and solvent viscosities, or 20 CSO : 0
in this case. These values reflect the highly un- 5OO o~
~-O
stable nature of floods using 100% solvent in- 3OO
jection. They also show that tertiary solvent 2OO
r_,,~
~ - SECONDARY FLO0[

floods usually have more adverse mobility ra- 0

I.-I00
tios than secondary floods, an argument for be- w w
ginning solvent floods before tertiary condi- 05~
~ 50 ~"~:E:T: AIY : L : C : "
.J
tions are reached. al 3O 0
0 F
~ 20
Water-solvent injection I LsEco.oAR FL00
I0 !
I
Injecting water with solvent came into being I
as a means of lowering the usually adverse mo-
I
bility ratio between the solvent and the oil. Such
lowering impedes viscous instability and in- ~ 1 ~ /TERTIARY
i / OpMUM
OPTIMUM
creases volumetric sweep efficiency in multi- 0.2 0 4 0 6 0.8 0
dimensional displacements. Our one-dimen- INJECTED SOLVENT FRACTION

sional fractional flow theory cannot estimate Fig. 4. The behavior of displacement-front mobility ratio
sweep efficiency for an unstable displacement. and time to complete recovery as functions of water-sol-
But we can estimate the mobility ratio across vent ratio ( W~ ).
292

tal fluids injected required to reach complete creases the solvent usage decreases, this defi-
recovery of the movable oil. The movable oil is nition of optimal WR implies minimum solvent
defined as the initially in-place oil less the trap- usage while yet realizing maximum displace-
ped oil (Sol -- SOM ). ment efficiency. This definition of optimality
Figure 4 shows that the displacement effi- includes no explicit statement as to minimizing
ciency and mobility ratio remain unchanged as the mobility ratio, although clearly the effect of
WR increases from zero until a critical WR is using the optimal WR is to improve the mobil-
reached, for both secondary and tertiary floods. ity ratio greatly over those cases using contin-
At this critical WR, the mobility ratio decreases uous solvent injection or values of WR less than
sharply. Increasing the WR above this critical the optimal. The results in Fig. 4 show that the
WR results in only a small change in mobility mobility ratio is not necessarily minimized at
ratio while gradually diminishing the displace- the optimal Wm Because the mobility ratio as
ment efficiency. We define this critical WR to defined by Eq. 17 approaches one as the WR
be the optimal WR. Because as the WR in- approaches infinity, it is unlikely that the mo-

(0) FRACTIONAL FLOW

CURVES (c) ELUTION HISTORY
E , J I0

z
o OIL
F-
SOLVENT

n~
U. . . . .

l
n~
,,, vwf/j,/
WATER

0 WATER SATURATIOI ~
i
I
I
I
I
I
I
I
oo
0

~ O I L
'

BANK
'

/ '
DIMENSIONLESS

SOL
'
TIME
/ i'o I
I

0
I i

. . . . . . . . . / ~ / ~ SOLVENT- WATER
<

,w~
.-J
0 --,7' / ~"/. " W-A-T-ER
/ I"
LEGEND ( o , b , c , d )

/~./-~ SOM : O IO

w~ CST : O
CSW : O
w<
z
W R : 05
M = 20
OF ~ i i E ~ !.
OIL SATURATION o 0 DIMENSIONLESS T I M E ~1.0

(b) SATURATION PROFILE AT (d) TIME-DISTANCE DIAGRAM

0.50 PV INJECTED

Fig. 5. Secondary miscible displacement below optimal water-solvent ratio (W~).

 293

(a) FRACTIONAL FLOW

CURVES (c) ELUTION HISTORY
I 0
o
J
x OIL SOLVENT
._
J
h
Z ----
_o
b--
.---~j
<~

U.
JLL INJECTED
h CONNATE W ER WATER
Vw / ~/ / ', / hi
I--
'

~ ¢
o.d/j • ,
P
0 WATER SATURATION 1.0 DIMENSIONLESS TIME I'0

t I
1.0
OIL BANK /

n,- I I l /-'~--------7~ SOLVENT WATER

/.~CONNATE
----V ------/ WATEy //J INJECTED WATER
/ / / ////VW
~ /l- I

o 1 o3
-- --h(~l . .... /
J /
---Z)- ---- LEGEND (a,b,c,d)

// SOM: 0 I0
F-- Z / CST = 0
> 7~
E3 0J Csw = 0
bJ
I-- 03
W R =2
0 M =2.2
L,J
Z
i i [ I 0 0 i I i I , Im
OIL SATURATION 0 0 DIMENSIONLESS TIME t'O
(b) SATURATION PROFILE AT (d) T I M E - D I S T A N C E DIAGRAM
0.40 PV INJECTED

Fig. 6. Secondarymiscible displacementabove optimal water-solvent ratio (WR).

bility ratio at the optimal WR will be a and to show why secondary and tertiary flood
minimum. optimal values of WR differ.
Figure 4 shows that the optimal WR for sec- Because the fractional flow analysis does not
ondary and tertiary floods are different. Among include all factors affecting recovery, the opti-
other things, the optimal WR is a function of mal WR defined here may not in practice be the
the initial condition. For the case in Fig. 4, the WR which yields the greatest displacement ef-
secondary flood optimal WR is 1.33 and the ter- ficiency. For instance, if oil trapping because of
tiary flood optimal Wa is 0.82. The optimal WR water shielding is important, lower recoveries
for other initial conditions lies between these may result by using the optimal WR as defined
two limits. The purpose of the examples which here rather than using a WR much less than
follow is to clearly establish the idea of an op- optimal (Brigham et al., 1963; Raimondi and
timal WR, to show how the WR selection affects Torcaso, 1964; Stalkup, 1970; Tiffin and Yellig,
the displacement efficiency and mobility ratio, 1983 ). In this case, as well as other cases which
go beyond the scope of this work, it may be nec-
294

(a) FRACTIONAL FLOW

CURVES (c) ELUTION HISTORY

Lt.
I I sO WN

Vw/V I, ! = -
r-/,/' I; / ,.-' CO,NATE

I ,~ I ~ i I - - - - -

'

o
~

I
- . . . . .

I-
-I! ~ 'OIL BANK

/ // (
t I I~ | /-- / 'SOLVENTW A T E R -
_ (e,b,c,d)
I0

II ° 1 ~ Ir 1
/ / / // / Csw:O
Cs,-O
// M = 2.6

~ - 0 0 I' , , , , , I~
OIL SATURATION 0 o DIMENSIONLESS TIME I'.o

(b) SATURATION PROFILE AT (d) TIME-DISTANCE DIAGRAM

0,50 PV INJECTED

Fig. 7. Definition of optimal water-solvent ratio (W~) for secondary miscible displacements.

essary to alter the W R from the optimal as de- true on all our representations, we indicate
fined here. miscible waves on the profile and history dia-
grams with wavy lines.
Secondary displacements For the under optimal WR case the displace-
To see what causes this optimal WR for sec- ment forms a miscible solvent wave whose ma-
ondary cases, consider Figs. 5-7 which show terial balance construction (Eq. 12a) passes
cases below (Fig. 5), above (Fig. 6), and at (Fig. through the residual water saturation. As stated
7 ) the optimal WR. Figure 6 is the same case as earlier, application of Eqs. 12a and 12b for the
is shown in Fig. 2 except with solvent-water and simplified solution procedure requires that the
solvent-trapped oil solubility neglected and line emanating from point (Sw,fw) = (0.9,1.0)
with a construction of the injected water front must pass through both the solvent-water and
added. The injected water miscibly displaces the oil-water fractional flow curves. The position
resident water as an indifferent wave with ve- of this line must pass through point I inasmuch
locity Uwgiven by Eq. 11 applied to water. As is as this is the only position which yields a phys-
 295

(a) FRACTIONAL FLOW

CURVES (c) ELUTION HISTORY
I I.O

it I/j, x
t I
°tv° L
WATER 0 IL SOLVENT

" OB' O ~ - - - -4- ---- U_ . . . . . . . . . .

~: o L ~ s~,/1~-'-~---I- I .... _4 ...... -4 . . . .

o WATEF S A T U R A T I O N Lo o DIMENSIONLES ~ T I M E I0
I.O I.O

J I I
: "
--SOLVENT-WATER

I I
/
Zl
I-- : "~1 1
/
F "JI 11
t
~z ol I f
/I L.EGEND (a,b,c,d}
SOM = 011 0
IX:
W
r,.."
........
z~
O .:!iii?¸
... .. CST= O
. ~i.~~ ¸ ; ¸ IDENT~ / f f
Csw = 0
.J /,:// WATER/ /
WR = 0.5
bJ I
I-- F__7O,," . I M = 30
hi
z ~ o_

/~/ INJECTED WATER

f <n~ f
i J i O o-~'
OIL SATURATION 0 O DIMENSIONLESS TIME 1.0

(b) SATURATION PROFILE AT (d) TIME-DISTANCE DIAGRAM

0 . 2 5 PV INJECTED

Fig. 8. Tertiary miscible displacement below optimal water-solvent ratio (WR).

ically admissable or consistent solution. Be- with the additional result that solvent, unac-
cause the injected condition point J is uniquely companied by water, is actually displacing the
different from point I, an additional wave ap- oil. Such a circumstance is precisely what the
pears and is a shock denoted by a line between injected water was intended to ameliorate.
points I and J. Figure 6 shows the opposite case where WR
The miscible water front propagates with a is tOO large. N o w the injected water is of such a
velocity given by the slope of the line from the large amount that it propagates faster than the
origin through the point J. The displacement solvent - that is, there is a waterflood preceding
consists of an immiscible shock from J to I the solvent injection. This circumstance is again
where the miscibly displaced resident water in undesirable because of a delayed oil produc-
turn displaces the solvent. The solvent propa- tion. More importantly, the rapid water ad-
gates faster than both the injected water and vance can leave a disconnected oil saturation
the immiscible front. The slow water waves are for the solvent to displace. Such saturations
the consequence of so little water being injected have been observed in strongly water-wet rock
296

(o) F R A C T I O N A L FLOW
CURVES (c) ELUTION HISTORY
?= I I.O
0
J
u_ × OIL SOLVENT
d
:::)
d
z ____b.. _ .
0
!,--
O B ~
c) W
n.- d INJECTED
b- w i u. WATER
r~ U-
hi RESIDENT WATER
i,i I
I

/_ /
I , i
1 i i ] L i O ~ i
ii D
~ I M E N S I O N L~ E S S ~ I
o WATER SATURATION I.O TIME 1Oi

I o 0
I OIL BANK /
SOLVENT WATE R ~
I

rr
I
I
I c~uJ" / J
WATV//////"
RESIDENT ~ / /
INJECTED WATER

Vow-/ /v~ ,'vw

Z i
/ / ,, ~GEND~°,b,c,"~
~, / / ,, SoM: o.~o
O i . g - / /,- Csw:O
I
~. / J ," cs,:°
--~. .... 7--7_," w~ : 2
£3
///

i
I-w
uJ<~
z

i O O,
y
OIL SATURATION DIMENSIONLESS TIME 1.0

(b) SATURATION PROFILE AT (d) T I M E - D I S T A N C E DIAGRAM

0 . 2 5 PV INJECTED

Fig. 9. Tertiary miscible displacement above optimal water-solvent ratio (W~).

to be partially shielded from the solvent by a Tertiary displacement

continuous water phase with a loss of ultimate The optimum for the tertiary case (dotted
recovery (Brigham et al., 1963; Raimondi and line in Fig. 4) is a little less clear-cut. Now M
Torcaso, 1964; Stalkup, 1970; Tiffin and Yellig, increases and then decreases slightly with de-
1983). creasing WR before again making a discontin-
In the optimal case, Fig. 7, WR is chosen so uous jump to a large constant value. The opti-
that the miscible solvent wave and the immis- mal WR again exists at or slightly larger than
cible wave from J to I have equal velocity. Now where the jump occurs ( WR = 0.82) in Fig. 4.
the mobility ratio across the displacement front The tertiary optimum requires a larger injected
experiences the full effect of the injected water solvent fraction than the corresponding sec-
but without unduly delayed production. This ondary case, another argument for beginning
optimal WR was originally proposed and exper- solvent floods before tertiary conditions are
imentally verified by Caudle and Dyes (1958). reached.
 297

(0) FRACTIONAL FLOW

CURVES (c) ELUTION HISTORY

1.0-

-
,,-, WATER 01L SOLVENT
g -

5 yV>v, ,
,_, -

RESIDENT INJECTED
Vw OB~ !-- -- !_~ WATER WATER

~: o ' I ,I , 1 o I i i i
0 WATER iSATURATION I.O o I DIMENSIONLESS TIME i'.o
i
II
I . 1.0 >
I'01 SOLVENT - WATER /'
II /
II /
/
/
/ / LEGEND (o,b,c,d)
,,=, / / SOM = 0.10
l--" Z vs / Vw CST = 0
/
g
/ / CSW = 0
J / W R = 0.82
5 / M = 1.9
J
/
hl /
nr"
WATER / /
/
/
INJECTED WATER

J tlJ
~- 0
Z /
w "~ /
14.1 /
.J Q.-
,.-, 0 0.. - -
IJJ CO" CO ~:[
I--W rr
LLI /

--i i i i i 0 o--p
i OIL SATURATION o o DIMENSIONLESS TIME ho

(b) SATURATION PROFILE AT (d) TIME-DISTANCE DIAGRAM

0.25 PV INJECTED

Fig. 10. Definition of optimal water-solvent ratio ( WE ) for tertiary miscible displacements.

The optimal WR for tertiary floods is based preceding the solvent. The oil fractional flow
on slightly different conditions from the sec- in the oil bank is rather small and the time for
ondary case. The below optimal case in Fig. 8 complete recovery is large. The value of
shows the solvent front moving faster than the WR=0.82 in Fig. 10 combines the advantages
injected water which in turn banks up the sol- of large oil fractional flow, and quick recovery
vent. The mobility ratio here is between the without a substantial sacrifice in M. This fig-
solvent at condition S and the oil bank satu- ure shows that this optimal WR is given by the
ration at OR. The mobility ratio is greater than intersection of the miscible wave line with the
the secondary case in Fig. 5 because the addi- water-solvent fractional flow curve con-
tional flowing water lowers the mobility of the structed so that the line is tangent to the water-
oil bank. The above optimal case in Fig. 9, like oil curve. If there is no point of tangency, the
the corresponding secondary case, shows the optimal WR is the same for both tertiary and
immiscible water displacement of the oil bank secondary displacements.
298

(o) FRACTIONAL FLOW

CURVES (c) ELUTION HISTORY
I 1.0-
0/
b.
I ] ~ -
Z
O
o
I..-
o Vos hZl
/ II•J---I-I-- -~ - - CHASE
I1:
~w / V/I; ~ I I ~ - I WATER
llff
w
I--

/
r--/ !/111
4-i ! I

r--..-, . . . . .
"
-
I

I
, ,, ,11 , I ~ O-
WATER SATURATION i.o 0 DIMENSIONLESS TIME Io
' II I.b

UJ
~ Z

ID(a,b,c,d_._)
010
W 0
..J
Z
_o 0
0.82
--Z

uJN
~ w ~
7,
(.3 ~')
U.h~

-0
OIL SATURATION o Y---~ SLUG DIMENSIONLESS TIME 1,0
1
(b) SATURATION PROFILE AT (d) TIME-DISTANCE DIAGRAM
0 . 5 0 PV INJECTED

Fig. 11. The behavior of chase water injection for a tertiary miscible flood.

Chase water injection resident water as in Fig. 7, and water displacing

the solvent-water mixture. The wave between
A largely overlooked use of fractional flow the chase water and the solvent-water fluid is
theory is in predicting the behavior of chase an immiscible spreading wave whose velocity is
fluids. Such predictions are well-suited to esti- given by Eq. 10a, which is the slope of the frac-
mations of optimal solvent-water slug size. tional flow curve at J. This is an extremely fast
Figure 11 shows a tertiary displacement, at moving wave so that, even though it is spread-
the optimal WR, of a solvent-water slug being ing and even though its initiation is delayed, it
displaced by water after tDs = 0.45 PV have been nearly catches up with the miscible solvent wave
injected. The point corresponding to the chase at the effluent end of the medium. The speed of
fluid is identical to the initial condition; both this wave is caused by the chase water trapping
are shown as I on Fig. 11. For this case we have the solvent in the residual hydrocarbon phase
essentially two separate problems: the solvent- so that only a small fraction of the injected sol-
water mixture displacing the residual oil and vent is actually produced. The chase water it-
 299
(a) FRACTIONAL FLOW
CURVES (c) ELUTION HISTORY
O-
0
.J WATER/OIL .-f r
h
WATER/SOLV ENT~, .~ I x
J
OIL SOLVENT CHASE GAS

i
Z
0 k-
l--
~J -----hZi
n,,*
WATER
nr
LU tATER/_ - - - -
As-1- "
[LI 2_-----:~ . . . .

~ o , , ,11 , I
0 WATER SATURATION 0
i L i
I DIMENSIONLESS TIME ,.o
II I •
~ l O - - J f£1.~ Z :

;I
I
I
I F-- ~ CHASE GAS =/"
I z I
w
I > /
I j /
O /
/
Ms / / Vc w
/
/
/
,/
/ ~ C H A S E WATER
/
/ LEGEND (a,b, c , d )
bO /
<~ /
(.9 / S O M = O I0
bJ / CST = O
03
bjnr <t CSW = 0
(DbJ I
WR = 0.82

i i , 0
OIL SATURATION 0 SLUG ..~ Ol M E N S I O I ~ L E S S TIME ,'.0 '
I"
(b) SATURATION PROFILE AT (d) TIME-DISTANCE DIAGRAM
0 . 5 0 PV INJECTED

Fig. 12. Definition of optimal water-solvent ratio ( WR) in the chase fluid for tertiary miscible displacements.

self miscibly displaces the water injected with

the solvent at an intermediate velocity Vcw as 1 1
tDs - (18a)
shown in Fig. l l d . VC s (U Cw) max

The case shown in Fig. 11 has about 0.03 PV where ( V C w ) m a x i s the velocity of the leading
of solvent and water produced at its injected wave induced by chase water injection. From
fractional flow. We could effect complete oil re- Eq. 18a it follows that the optimal solvent usage
covery if we reduced the slug size by this amount is:
to 0.42 which is an optimal slug size for this
case. This optimum occurs when the slug size ms = (1--fwj)tDs (18b)
is just large enough so that the fastest chase The solvent amount ms suggested by these
water wave intersects the slow solvent wave at equations has been doubly optimized, first with
XD= 1. Elementary geometry then suggests the respect to the Wa ratio and then to minimize
optimal slug size (solvent plus water) ex- solvent slug size. In the case just considered,
pressed in pore volume units is: and in many cases in general, since (VOw)maxis
300

quite large, the optimal slug size tDs c a n be ap- has been selected to minimize solvent usage.
proximated by: The example of Fig. 12, however, shows a case
1 where the actual solvent-water slug size is
tDs ~ - - (18c) tDs = 0.30 which is greater than the optimum to
UC~
minimize solvent usage. The WR in the sol-
Since the reciprocal of vc~ represents the sol- vent-water slug in Fig. 12 is at the optimum
vent breakthrough time, this equation states suggested in Fig. 10. If we select the chase fluid
that the optimal slug size can be approximated WR ( WR is now the water-chase fluid ratio ) to
by the solvent breakthrough time. be the point K determined by the intersection
The word "optimal" sometimes suggests an of the miscible wave line with the water-chase
economic objective function whereby rate of re- fluid fractional flow curve, the velocities at the
turn is maximized with respect to several vari- front and rear of the solvent slug will be equal.
ables. We clearly are not doing this here be- From either Eq. 18a or Fig. 12d this circum-
cause fractional flow theory lacks the generality stance suggests that complete oil recovery can
of numerical simulation even in elementary be effected with a zero solvent slug size. The
cases. (We cannot treat slug sizes less than the optimal chase fluid WR is slightly greater than
optimal because this would entail wave inter- the optimal solvent WR in agreement with the
ference within the medium, a circumstance idea of keeping the mobilities of the solvent-
whose analysis eludes graphical treatment.) water and chase fluid-water roughly equal.
Nevertheless, the optimums discussed here are In reality, oil recovery will begin to decline
predicated on avoiding effects, like dispersion before the solvent slug size actually reaches zero
and viscous fingering, which numerical simu- because effects we have neglected in the frac-
lators treat crudely if at all. Our analysis there- tional flow analysis (dispersion, fingering, het-
fore should be viewed as an alternative to or a erogeneity) will become important. Even con-
comparison with simulation. ceding the validity of the fractional flow
assumptions, the tDs = 0 case will be in error be-
Chase fluid design cause the chase fluid is generally not miscible
with the oil. Nevertheless, Fig. 12 suggests that
The final example involves chasing the sol- solvent usage can be affected by the chase fluid-
vent-water mixture with a mixture of water and water ratio in the chase fluid and that this vari-
a less expensive chase gas which is miscible with able should be included in design considera-
the solvent. tions. To our knowledge, this analysis is the first
As shown in Fig. 12a there is now an addi- to theoretically analyze simultaneous chase
tional curve for the water-chase gas fractional water-chase fluid injection.
flow which will be displaced to the right of the
water-solvent curve (presuming, as is usually Conclusions
the case, that the chase gas is less viscous than
the solvent). The chase-fluid injected condi- This paper is primarily an exposition of
tion K can be anywhere on this curve. As long methods to graphically analyze solvent dis-
as the solvent-oil bank and solvent-chase fluid placements based on fractional flow theory.
waves do not intersect, the problem can be ana- Specific conclusions, therefore, depend on the
lyzed as individual cases of the types given specific fractional flow curves used. However,
above with the origin of the chase fluid waves application of the methods given here to the
offset by the slug size tDs. large variety of such curves attainable in prac-
We show a particular case in Fig. 12 where tice will certainly yield at least qualitative
the ratio of water to chase gas in the chase fluid trends between performance and reservoir
 301

properties. A good example here would be the

effect of reservoir wettability on oil bank satu- OAi ~_c~Fi=
rations, optimal water-solvent ratios and chase Ot 3x 0 i=l,....,N (A1)
fluid design.
The analyses, limited to cases where the oil where:
and solvent are mutually miscible or develop Ai =Ai (ul ,...UN ) is a known overall concentra-
miscibility in the absence of dissipation, yield tion function;
the following three important and novel gen- Fi =Fi(ul .... UN) is a known flux function; and
eral contributions: ui (x,t) is the dependent variable, a measure of
( 1 ) A general fractional flow analysis which local concentration.
can include solvent-water solubility, solvent- Equation (A1) describes a very large number
trapped oil saturation, solvent partitioning into of dissipation-free conservation laws with non-
the trapped oil, and arbitrary fractional flow linear constitutive and concentration func-
curves, water-solvent ratios and initial tions. The boundary conditions for (A1) are:
conditions. Ai(ul (x,O ),..UN(X,O ) ) =Aii (initial or right),
(2) An extension of the general analysis to
include the performance of chase fluids. Fi(ul (O,t),..UN(O,t) ) =Fi.j (injection or left),
(3) Simple estimations of optimal water where Aii and Fij are specified.
usage in both the solvent slug and chase fluid, It is easier to proceed with Eq. A1 in the fol-
and optimal solvent slug size. lowing form:

Acknowledgements Out Out

Ai~~-~-+ Fij-~--x= 0, i=l,...,N (A2)

We thank Amoco Production Company for

where the second subscript indicates differen-
permission to publish this work, a portion of
tiation with respect to ui (holding all the other
which was supported by the Center for En-
ui constant) and a repeated index indicates
hanced Oil and Gas Recovery Research at The
summation. The boundary conditions corre-
University of Texas. We also thank Thormod
sponding to this form are ui(x,O)=ua and
Johansen for suggestions leading to the
Ui(0,t) =UiJ.
Appendix.
The change from the boundary conditions
given above to these is possible only if the ma-
Appendix - A demonstration of trix A (with elements Air) and F (with ele-
coherence ments Fit) are nonsingular. If we let a nonsub-
scripted symbol indicate the matrix or vector
The objective of this appendix is to demon- containing the subscripted elements, we can
strate that coherence is a property of reducible, write Eq. A2 as:
hyperbolic equations having Riemann bound-
ary conditions. Ut+J U~ = 0 (A3)

where J = A - 1F and the subscripts represent

Formulation partial differentiation with respect to t and x.
Note that A and F (and hence J ) remain
We restrict ourselves to rarefied (spreading known functions of the ui and the matrix mul-
or smooth) waves initiated by the following tiplication with the inverse of A is possible since
mathematical description: A is non singular.
302

Transformation
j 0ul ~ Ouk _ OUN OUh
+ +... + =,t
We transform the partial differential equa-
tion (A3) into an ordinary differential equa- since it is homogeneous in t/. W h e n we subtract
tion with the transform variable tl= x / t . Equa- the above two equations we have:
tion A3 now becomes:
OUk OUk
+ = 0 (A7)
(J- ~/I)~=0 (A4)
Thus t/, defined by Eq. A6, is the concentration
where I is the identity matrix. The boundary "velocity" or the slope of a curve in (x,t) space
conditions are also transformed to: along which uk is constant. Since k is arbitrary,
U(x,O) -~ U ( o o ) : U I and U(O,t) = U(O) = Uj the velocity is the same for all species. This is
the coherence condition used in Eq. 9.
Equation A4 is an eigenvalue problem with ~ as
the eigenvalue and d U / d ~ as the right
References
eigenvector.
Let us consider the kth scalar equation in this
Brigham, W.E., Dew, I.N. and Reed, P.W., 1963. Recovery
set for any one eigenvalue ~: process for producing petroleum. U.S. Patent No.
3249157.
dul _ duk Buckley, S.E. and Leverett, M.C., 1942. Mechanism of fluid
Jh1 + .... (15)
displacement in sands. Trans. AIME, 146: 107-116.
Caudle, B.H. and Dyes, A.B., 1958. Improving miscible dis-
+~ dUN duk placement by gas-water injection. Trans. AIME, 213:
281-284.
Cere, A. and Zanotti, F., 1985. Sharpening behavior and
The left side of this equation is just the total dispersion in chemical flooding. Presented at European
derivative of J with respect to ~/or: Enhanced Oil Recovery Symp., Rome, Italy.
Claridge, E.L. and Bonder, P.L., 1974, A graphical method
dJk dub for calculating linear displacements with mass transfer
and continously changing mobilities. Soc. Pet. Eng. J.,
d~ = ~ d r (Dec.): 609-618.
Courant, R. and Hilbert, D., 1962. Methods of Mathemat-
an equation containing only total derivatives. ical Physics. Interscience, Vol. 1.
The eigenvalue ~/does not have a subscript and Dai, K.K. and Orr, F.M., Jr., 1987. Prediction of CO2 flood
since k is arbitrary we must have: performance: Interaction of phase behavior with micro-
scopic pore structure heterogeneity. S.P.E. Res. Eng.,
d J1 d J2 dJk dJN (Nov.): 531-542.
dul - du2 ..... duk - - d U N - - YI (A6) Davis, J.A. and Jones, S.C., 1968. Displacement mecha-
nisms of micellar solutions. J. Pet. Technol., (Dec.):
1415-1428; Trans. AIME, p. 243.
De Nevers, N., 1964. A calculation method for carbonated
Coherence waterflooding. Soc. Pet. Eng. J., (Mar.): 9-20; Trans.
AIME, p.2 31.
Returning to Eq. A3 let us look at its kth Dumore, J.M., Hagoort, J. and Risseeuw, A.S., 1984. An
analytical model for one-dimensional, three-compo-
equation: nent condensing and vaporizing gas drives, Soc. Pet.
Eng. J., (Apr.): 169-179.
Ouk _ 0ul _ Ouk _ OUN
Fayers, F.J. and Perrine, R.L., 1959. Mathematical de-
at + "'" + + '" + =° scription of detergent flooding in oil reservoirs. Trans.
AIME, 216: 277-283.
But Eq. A5 can be written in terms of x-deriv- Foster, W.R., 1973. A low tension waterflooding process
atives as: employing a petroleum sulfonate, inorganic salts, and a
 303

biopolymer. J. Pet. Technol., (Feb.) 205-210; Trans. predictive model for C02 miscible flooding. S.P.E. 13238,
AIME, 255. presented at 59th Annu. Tech. Conf. Soc. Pet. Eng.,
Hales, H.B. and Odeh, A.S., 1976. An improved method for Houston, Tex., (Sep.): 16-19,
simulating ideal low tension flooding processes. Soc. Pet. Pope, G.A., 1980. The application of fractional flow theory
Eng. J. (Apr.): 53-56. to enhanced oil recovery. Soc. Pet. Eng. J., (June): 191-
Helfferich, F.G., 1980. General theory of multicomponent, 205.
multiphase displacement in porous media. Soc. Pet, Eng. Pope, G.A., Lake, L.W. and Helfferich, F.G., 1978. Cation
J. (Feb.): 51-62. exchange in chemical flooding: Part 1 - basic theory
Helfferich, F.G. and Klein, G., 1970. Multicomponent without dispersion. Soc. Pet. Eng. J., (Dec.): 418-434.
Chromatography. Marcel Dekker, New York, N.Y. Raimondi, P. and Torcaso, M.A., 1964. Distribution of the
Hirasaki, G.J., 1981. Application of the theory of multi- oil phase obtained upon imbibition of water. Soc. Pet.
component, multiphase displacement to three-compo- Eng. J., (Mar.): 49-55.
nent, two-phase surfactant flooding. Soc. Pet. Eng. J., Shutler, N.D. and Boberg, T.C., 1972. A one-dimensional
(Apr.): 191-204. analytical technique for predicting oil recovery by
Koval, E.J., 1963. A method for predicting the performace steamflooding. Soc. Pet. Eng. J., (Dec.): 489-498.
of unstable miscible displacements in heterogeneous Stalkup, F., 1983. Miscible Displacement. S.P.E.
media. Soc. Pet. Eng. J., (June): 145-154. Monograph.
Lake, L.W., 1989. Enhanced Oil Recovery, Prentice-Hall. Stalkup, F.I., 1970. Displacement of oil by solvent at high
Lake, L.W., Johnston, J.R. and Stegemeier, G.L., 1978. water saturation. Soc. Pet. Eng. J., (Dec.): 337-348.
Simulation and performance prediction of a large-scale Taber, J.J., Kamath, I.S.K. and Reed, R.L., 1961. Mecha-
surfactant/polymer process. Soc. Pet. Eng. J., 18, p. 12. nism of alcohol displacement of oil from porous media.
Larson, R.G., 1979. The influence of phase behavior on Soc. Pet. Eng. J., (Sep.): 195-212.
surfactant flooding. Soc. Pet. Eng. J., (Dec.): 411-422; Tiffin, D.L. and Yellig, W.F., 1983. Effects of mobile water
Trans. AIME, 267. on multiple-contact miscible gas displacements. Soc.
Larson, R.G. and Hirasaki, G.J., 1978. Analysis of the Pet. Eng. J., (June): 447-455.
physical mechanisms in surfactant flooding. Soc. Pet. Wachmann, C., 1964. A mathematical model for the dis-
Eng. J., (Feb.): 42-58. placement of oil and water by alcohol. Soc. Pet. Eng. J.,
Liu, Tai-Ping, 1977. Large-time behavior of solutions of (Sep.): 250-266; Trans. AIME, 231.
initial and initial-boundary value problems of a general Welge, H.J., 1952. A simplified method for computing oil
system of hyperbolic conservation laws. Commun. Math. recovery by gas or water drive. Trans. AIME, 195: 91-
Phys., 55: 166-177. 98.
Patton, J.T., Coats, K.H. and Colegrove, G.T., 1971. Pre- Welge, H.J., Johnson, E.F., Erving, S.P. and Brinkman,
diction of polymer flood performance. Soc. Pet. Eng. J., F.H., 1961. The linear displacement of oil from porous
(Mar.): 72-84; Trans. AIME, 251. media by enriched gas. J. Pet. Technol., (Aug.): 787
Paul, G.W., Lake, L.W. and Gould, T.L., 1984. A simplified 796; Trans. AIME, 222.