SOCIETY OF PETROLEUMENGINEERS OF AIME

6200 North Central Expressway RR SPE 3600

Dallas, Texas 75206

THIS IS A PREPRINT’--- SUBJECT TO CORRECTION

A Three-Phase, Experimental and Numerical

Simulation Study of the Steam Flood
Process
BY
A. Abdalla andJf. H. Coats, Members AIME, INTERCOMP

O Copyright 1971
American Institute of Mining, Metallurgical, and Petroleum Engineers, Inc.

This paper was prepared for the 46th Annual Fall Meeting of the Society of Petroleum Engineers
of AIME, to be held in New Orleans, La., Oct. 3-6, 1971. Permissionto copy is restrictedto an
abstractof not more than 300 words. Illustrationsmay not be copied. The abstract should contain
conspicuousacknowledgmentof where and by whom the paper is presented. Publicationelsewhereafter
publicationin the JOURNAL OF PETROLEUMTECHNOLOGYor the SOCIETY OF PETROLEUM ENGINEERSJOURNAL is
usually granted upon request to the Editor of the appropriatejournalprovided agreementto give
proper credit is made. “i

Discussionof this paper is invited. Three copies of any discussionshould be sent to the
Society of Petroleum Engineersoffice. Such discussionmay be presented at the above meeting and,
with the paper, may be consideredfor publicationin one of the two SPE magazines.

ABSTRACT show earlier steam breakthroughthan those with

higher viscosity.
A numerical model of steam-driveoil
recovery was developed and tested. The implicit INTRODUCTION
pressure-explicitsaturation (IMPES) technique
was used to solve the three-phase fluid flow The first part of the work presented here
equations for compressiblefluids. A method is a physical laboratory model of steam injec-
was developed and applied to determine the tion in a linear system. A constant pressure
temperature and the rate of steam condensation boundary condition was used. Two runs were
implicitly from the heat-balance equation. Both performed on the same model using two different
techniques were used in computer simulators for sets of injection and production pressures. Oil
linear and two4iimensionalsystems. recovery and temperature distributiondata were
obtained. Each run was repeated to check
A steam-injectionexperimental study was reproducibilityof results.
perfoxmed in a linear model. The results of
this experimentalwork showed good agreement The second part of this work describes the
with the results obtained from the linear developmentand applicationof numerical
numerical computer simulators. The results from simulationtechniques to solve equations
the two-dimensionalnumerical computer simulator describing the steam-injectionprocess. This
was also found to be in good agreement with simulationmodel was the impli&:9p:essure-
-..L12 _L_A .L..- 22—.--:---1 -.—--2---A-7 ---..lL- .5-14 n<+ ..+,,-.+4 *- +-nhrt; “,-c, ~G ~ofiv.e
puuALtmeu lJwu-JJIltxls.LulleLL eiiperuilexlu.iu I-eauLIJb. GA~L&bLU D=UIAL-ULVL4 UGL,lUU~UG .? -

difference equations describing the multiphase

The numerical simulatorswere also used to
study the effect of some parameters on the
steam-drivenrocess.
-.-—. r------- lt was —-—.—
found that n~eri-
cal.model results were very sensitive to
capillary pressure values. It was also found
that the relative permeabilitydata has a minor
effect on the results obtained. The oil
viscosity was found to affect the process to a
large etient. Oil recovery from steam-drive
process decreases as the oil tiscosity ficreases~
Also, the recovery curves for low viscosity oils
References and illustrationsat end of paper.
flow system. The solution of the heat-balance
equation yields the temperature ahead of the
steam front and the rate of steam condensation
behind the front.
THE LINEAR PHYSICAL MODEL
Few linear physical models have been
reported in the literature. The most recently
published one is that of Will.manet al.23
However, they did not publish enough data so the
numerical simulator developed here could be
I
 A THREE-PHASE.EWEUUWWAL AND NUMERICAL
2. CTMTIT Am-rfm cmmiv m
OJ.L-lUU LAW.,“A“u. “.
TH VIT17AMli’llMTl
----- ----- PR(Y!T?.S.S
.-.--—-- SPE 3603

properly tested. Therefore, a physical model of each phase in the reservoir will be obtained.
was designed that not only helped in the under-
L—..
s2.
-— -m *L- —------- L..
* -1 -- -.-A.- ..4-A
scanang WI me prucesa, Uub CrAau pL-uv J.uGu
sufficient data to check the simulator.
a(pouxo) a(pou ~)
A schematic diagram of the apparatus is ax - ay + qvo =
shown in Rig. 1. It consisted of a condensing
steam trap, filter, adjustable coil heater, a(oposo)
Wet pressure gauge, core holder, thermo- (l-A)
at ;******** ““”
couples, outlet pressure gauge, condenser and
backpressure regulator. for the water phase
The steam used in the experimentswas a a (PWUW)
saturated steam from The U. of Texas utility + ~ + qvc =
lines. The injection pressure was adjusted by ax - ay
a pressure regulator mounted on the steam lines.
The steam coming from the pressure regulator
passed through the condensing steam trap. This
knocked out the steam condensate. The steam
then passed through a filter which removed and for the steam phase
impurities that could cause clogging of the sane
a(p~ux~) a(p~u ~)
pack. A coil heater was wrapped around the
injection line. The temperature of the heater ax - ay ‘%- %.=
was adjusted by a variable autotransformerto a
temperature slightly higher thsn the saturation
temperature of the injected steam. TiiiseMmi-
nated any possibility of having condensatein
the injected steam. where U& and uy_iare given by Darcyts law as
follows:
The oil used was primol 185 with a viscos-
ity of ,43cp at 80°F and at 2600F. Curves of kxkri api
viscosity and specific gravity vs temperature u= - 6.33 - (2-A)
-—- mIuwIk
tire-1,----. .
L-I Fig. 2.
~i r- ~“”””
1
The sand used was an unconsolidated ssnd kk. api
of 2.54 darcies permeability snd 35.4 percent - 6.33 -& (2-B)
porosity. ‘yi ‘ ~“”””
i
Two steam injection runs were performed and i = O,w,s.
,,.+=IT
L4.4.=
rli
=+..”.
Pfnwnn+
“...
~~7&ma~&m=r nnndi+.i
w“..--.--..”.
nne. The
. ..”
fi?~+.
--- - .

run was performed with an injection pressure Substitutionof Eqs. 2 into Eqs. 1 gives

n
of 40.0 psia and a production pressure of 28.2
psia. The results are plotted in Figs. 3 and 4 kxkrOpO apO
The second run was performed with injection 6.33 &
llo T )
pressure of 39.6 psia and production pressure o: L

14.7 psia. The results are plotted in Figs. 5

1
and 6. Both runs were repeated and the results k kryo ape,

(y
+a
were in good agreement. ~ ~; + Clvo
lJo
—.
THE DIFFERENTIALFORM OF THE PROBLEM
a(oposo)
= . .(3-A;
Differential equations describing the fluif at .“ “ “ ““” “ “ ● “
and heat flow for the stesm~rive process are
presented here.
..,-a-–-- .-L.---
Fhua !Low Jlql,lazlons

The mathematical relationshipsdescribing

multiphase fluid flow a pear in the a[ k krwpw PPW
literatWe05,12717,20t2f The development of + qvc
+ x\ Ilw T )] + qw
such relationshipsis based upon mass balance
and Darcyts law for each phase. When both
relationshipsare combined, the partial- a(~pWsw)
differential equation describing the fluid flow = .(3-B:
at********** “
SPE 3600 A. AHIALLA end K. H. COATS

Mp s
Ps = R(T + 460) ‘ i.e., an ideal gas.

2. Viscosities of the water, oil and

steam depend upon temperature only.

3. Water and stesm relative permeabilities

are functions of their relative saturations.
a(~psss) The oil relative permeability is a function of
= . (3-c) both oil and water saturations.
at””-”””=”””
The saturations are related as follows: 4. The capillary pressure between oil and
water is a function of the water saturation only
SO+ SW+ S5 =1..... .00 (4) Capillary pressure between oil and steam is a
function of both water and oil saturations.
The pressures of the different phases are re-
lated by the capillary pressures as follows: 5* The heat loss term is explained in
detail in Appendix A. The difference form of
P =PO-PWO. O.. .0.. (5-A) the partialdifferential equation described here
c o-w is presented in Appendix B. The application of
the IMPES technique to solve the difference equa
P = Ps -P~* ●. . . .,.. (5-B) tion is given in Appendix C. The equations
co-s given in both appendices are for the linear mode
All symbols used are described in the Nomen- for simplicity.
clature.
‘ DESCRIPTION OF COMPUTER ~~
Heat Flow Equation
The Line~Simulator
I
The development of the mathematical
relationshipdescribing the heat flow in porous The techniques discussed here were incor-
media is based upon the heat balance, Fourier, porated into a Fortran IV computer program. The
and Darcy’s equations. When those equations arc grid system used is shown in Fig. 7. This
combined, the following differential equation i: program computes at each time step the satura-
obtained. tions, pressures and temperaturedistributions.
Also, it computes the steam condensationrates

- ‘,+ ~ %’*(DY
(DX %) in each block and the injection and production
rates. The check for the convergenceis based
upon the change in pressure, temperature,and
a steam condensationrates between two successive
- ~ (Uxphn) - &uyPhn) + qvshinj iterations. Between the three checks, the rate
of steam condensationis found to be the ccm-
trolling one.
= & [$(PSh) + (1 - $) PrCrZ] , (6)
The program has a msximum grid-size system
where of 100. The execution times are dependent on
the weight factor used in the calculationof the
uiphn = u. POhn + uiwpwhn rate of steam condensationdescribed in Appendix
10 0 w
Do A value of 0.85 is found to be most suit-
+U ispshn able. An average execution time is 0.08 seconds
s “- ● .* (7) per
●
//--time step for 10 blocks system on the CDC
●

bbuu computer.
I
Sph = Sopoho + Swpwhw + S~p5,S, (8)
A generalized flow chart of the program is
given in Fig. 8. All the necessary data other
snd i = x,y. than the steam viscosity, specific heat, and
rock properties are read into the program prior
Q and ~ are given byEqs 2. to the main computationloop. At the start of
this loop, the relative permeabilities,the
In this study, the functionaldependencies capillary pressures, the densities, the
of the parameters are assumed to be as follows. viscosities, and the transmissibilitiesare
determined. A table look-up is used for this
1. Densities of water and oil are func- procedure. In the calculationof the trans-
tions of temperature only. The density of stean missibilities,all the parameters are evaluated
is expressed by the equation 100 percent upstream. Calculation of the pres-
sure distributionthen follows. The steam
 A THREE-PHASE.EXPERIMENT
-.——
SIMULATION STtiY OF !l ; STEAMFLOOD PROCESS
saturation temperaturesare determined from the
stesm pressures using a table look-up procedure.
Calculation of the saturationsthen follows.
Computation of the rate of steam condensationor
the temperature is done using the heat bslance pressure drop of about 11.8 psi and an injec-
equation. This is followed by the convergence
check.
In the program, steam viscosity, rock
density, and specific heats of oil, water and
rock are constants. However, stesm viscosity
and specific heats of oil and water can be used (Fig. 4). About 84 percent of the oil in place
in the program as temperaturedependent. Fixing was produced by the end of the experiment.
the former quantities is merely due to the
relatively small pressure drops used in testing
the model.
The Two-DimensionslSimulator
A computer program was written based on
the techniques discussed here. The grid system
used is shown in Fig. 9. As in the linear simu-
lator, the program computes pressures, satura-
tions and temperature distributions. The
program slso computes the rate of steam conden~
sation and injection and production rates.
Although the controllingparameter in the con-
vergence is the rate of steam condensation,
the program computes the change in the three
variables, namely, pressure, temperature and
rate of steam condensation.
The progrsm has a maximum grid-size system
of 20 x 20. Execution times are dependent upon
the weight factor used in the calculationof
the rate of steam.condensationas stated in
AppendixD. A value of 0.$35was found to be
suitable. An average execution time is 0.25
second per time step on the CDC 6600 computer
f~~ .aj ~ ~ g~~~ ~y~~~~,~~udyo
A generalized flow chart of the program is
given h Fig. 8. The program follows the same
outltie as the linear simulator. However, the
values of the parameters in the transmissi-
bilities calculationare taken at the block
under considerationexcept for the relative
permeabilities,which are 100 percent up-
streamed.
COMPARISONWITH EXPERIMENTALRESULTS
The Linear Model
As mentioned earlier, two experimentalruns
with different boundary conditionshave been
performed. The difference between the runs was
the pressure drop. This gave different injec-
tion rates, which in turn affected the cumula-
tive heat loss. The pressure level has great
significancein the stesm-injectionprocess.
Saturation temperature and steam enthalpies are
functions of pressure level. The higher the
pressure level, the higher the temperature
level, which, in turn, gives larger rate of heat
AND NUMERICAL
SPE 3600
Loss. Data used in the computer program for
both experimentalruns are given in Appendix F.
~
The first expertient was performed with a
tion pressure of about 40.0 psia. The experi-
mental and calculated results are plotted in
Figs. 3 and 4. The experiment was terminated
approximately40 hours from the start. Although
3 PV had been produced, only about one-half of
the model had saturated steam temperature level
To test the linear numerical simulator,
a computer run was made using the same boundary
conditions. Data used in the program are given
in Appendix E. The value of the surface ove~all
.,.–-——-l
Gnerma -.-ems-,--L .._-J:-
cuezzzc~eub usw
Al.-....... .....
ML IAe pL-U~L-eJII WU=
aboout double the value determined in the labora-
tory. However, it was found that the vslue of
the”over-allthermal coefficientused behind the
steam front is the one that is important in
getting a good agreement between the calculated
and the experimentalresults. Accordingly, the
difference in values can be due to two factors:
(1) the over-all thermal coefficientis tempera-
ture dependent to some degree. The value of
this coefficientfor liquid phases was deter-
mined expefientally at 1400F using hot water
injection, while the temperature in the steam
injection runs reachec”vsluesup to 270°F and
(2) the over-all thermal coefficientfor steam
is small compared with that for liquids. Steam
condensatemight have developed a thin layer
around the inside wall of the core holder in the
region behind the steam front. This will in-
crease the coefficientfor this region to some
degree.
Results plotted in Fig. 3 show that experi-
mental and calculated results agree closely
when the proper value of the over-all thermal
coefficientis used.
The second experimentwas performed with
a pressure drop of 24.9 psi and an injection
pressure of 39.6 psia. Both experimental. md
calculatd results are plotted in Figs. 5 and 6.
The experimentwas terminated after appro~atel
m -,,------
——-_l..
_._A
ii ‘hours. n~cnougn ouy ~
““” ‘- ‘-‘--1-- rv were prouuv=u,
three-fourthsof the model had reached steam
temperature (Fig. 6). Comparing this result
with the one in the former experiment shows the
effect of the pressure drop on the heat loss.
About 80 percent of oil in place was produced
by the end of the experiment.
The linear simulatorwas run for the bound-
ary conditions of the second experiment. All
the parameters used were the same as those used
for the first tun, including the value for the
surface over-all thermal coefficient.
Results plotted in Fig. 5 show good agree-
ment between experimentaland calculated
results when the proper value of the over-all
thermal coefficientis used.
The Two-DimensionalModel
The only published results on two-
A:.....”.<
UL111C11O-LV11U
-“..1 --A-1
Uluuc.1-a
_ ---
d~-c
-4..-,..
&vcll
L..
Uy
ck..41
CJIJUULCI-.
his publication, he listed the parameters and
the recovery curve for one-eighth of a five-spot
model. No temperature distributionwas reported
The data are given in Appendix E.
The two-dimensionalsimulator was used with
the data reported. Fig. 10 shows the experi-
mental and the calculated results. A good match
between both results is evident.
DISCUSSION
Capillary Pressure
To determine the importance of the
.m.
capillary pressure in tinesteam-drivemodei, two
computer runs have been performed using the
two-dimensionalexperimentdata given in
Appendix E. One run uses the capillarypres-
sures as tabulated in the above mentioned appen-
dix, and the second run uses scaled values, such
that
1? — P
Cscaled Ctabulated.
10
The recovery curves are shown in Fig. 10.
The recovery curve of the run which uses scaled
capillary pressure values shows a delay in the
water breakthrough,and an early steam break-
through when compared with the recovery curve of
the run which uses the tabulated values. This
might be due to the fact that low capillary
pressure values give low steam pressure, which,
in turn, give low steam saturationtemperature.
This will decrease the heat loss that is a
function of the temperaturelevels and accounts
for an early steam breakthrough. In this case,
more heat will be used to heat the producing
zone, giving low oil-to-waterviscosity ratio
which will result in a delay in the water break-
throllgh:
The above discussion shows the importance
of the capillary pressure values in the steam-
drive model. The recovery curve, resulting from
the use of scaled capillary pressure values, is
closer to the experimentalresults than the one
determined through the use of the tabulated
values. To explain such a trend in the results,
a comparison was made between the values tabu-
lated and values calculated from Leverett’s9
imbibition J-curve using values of the inter-
racial tensions at atmospheric conditions. It
-L-,.,-A
SLIUWGU
+ho+
ULia U
Lfi+h
UU IAL
.WIA
alc
nP
UL
~~,e ~&T,e
magnitude. However, Hough et al.7 shows that
the value of the interracial tension at the
temperature and pressure used in the experiment
drops to as low as one-third of its value at
atmospheric conditions. This tends to give
lower capillary pressure values than the one
tabulated.
Relative Permeability
-- 20 ~fi Dal .++,,a
LI.GLC.”LVGno-e.h+l
p’G.u,GcLIJL*A”J
.-a
++.,.r.l,,a.
v-u=a
a G
-L
~
+h.+
“..-”
20 percent off the tabulated values ti Appendix
E have been used in the two-dimensional
simulator. The other parameters are the sane as
those used in the experiment. The recovery
curves obtained from both runs show less than
1.2 percent difference. Breakthroughvalues
did not show any change. This indicates that
the steam-drivemodel is not very sensitive
to variations on the order of L 20 percent in
relative permeabilityvalues.
Oil Viscosity
The-steam-driveprocess has been intro-
duced to the industry as a solution to the
problem of producing highly viscous oils; thus,
the importance of investigatingthe effect of
viscosity on the process.
Three different oils (I, II, III) with wide
ranges of viscosity (Fig. 11) have been used in
the two<imensional simulator. The other
parameters are the same as those of the two-
dimensional experiment given in Appendix F. The
three resulting recovery curves are shown in
Fig. 12. The curves show the following.
1. Recovery curves for low viscosity oils
show earlier steam breakthroughthan those with
higher viscosity. This is due to the fact that
the driving front moves more slowly in case of
high viscosity oils than it moves in case of
low viscosity ones. This will increase the
heat loss which, in turn, delays the steam
breakthrough.
2. Recovery curves for high viscosity oils
showed earlier water breakthroughthan those
with low viscosity. This is due to the fact
that for high viscosity oils, the mobility of
water is much greater than the mobility of oil!
which will accelerate the water production.
3. Although oil recovery from steam-drive
process decreases as the oil viscosity
increases, it still gives much higher values
than those obtained from the waterflood process.
The recovery curve for a waterflood in a five-
spot pattern and for oil-to-waterviscosity
ratio of ’754is shown in Fig. 10. Such oil is
comparable to the one used in the two-
dimensional experiment. Comparing the two re-
covery curves snows the superiorityof the
~~uer of 5tewf100d prOcess over the wat,erf~~~dnrncess.
=..--__.
However, as mentioned before, the recovery
curves of the steamflood process differ con-
siderably with the magnitude of the heat loss.
 A THREE-PHASE,EXPERIMENTALAND NUMERICAL
6 SIMULATION STUDY OF THE STEAMFLOOD PROCESS SPE 3600

Dtiensional.,. Two-Phase Flow in Oil and Gas

NOMENCLATURE
Reservoirs”, Sot. Pet. lhvz.J. (Dec., 1967)
377-388.
A= cross-sectionalarea, sq ft
6. Higgins, R. V. and Leighton, A. J.: “A
c= specific heat, Btu/lb day OF
fractional flow, dimensionless Computer Method to Calculate Two-Phase
f=
H= reservoir sand thickness, ft Flow in Any Irregularly Bounded Porous
h= M&um”, J. Pet. Tech. (June, 1962) 679-
enthalpy, Btu/lb
h~j = .
enthslpy of injected steam, Btu/lb
7. Hough, E. W., Rzasa, M. J. and Wood, B. B.:
k= absolute permeability,darcies l!Interfa~i~Tensions at Reservoir pres-.
kr = relative permeability,dimensionless
heat loss, Btu/D sures and Temperatures,Apparatus and the
&h =
molecular weight of steam the Water-MethaneSystem”, Trans., AI14E
M=
pressure, psia (1951) ~, 57-60. —
8. Lauwerier, H. A.: !~~e Trmsport of Heat
P: : capillary pressure, psi
in an Oil Layer Caused by the Injection
q. mass injection rate, lb/D of Hot Fluid;’,Applied Sci. Res.-(1955)
qv . volumetric injection term, lb/unit bulk Sec. A, ~, 145.
reservoir volume er day Leverett, M. C.: ~?capilla~Behavior fi
9*
R= gas constant, psiaTcu ft lb mol ‘R Porous Solids”, Trans;, AIME (194.1)l&,
r. radius 152-169.
R= heat residual, Btu/D Little: T. w*: f~Recove~.Of ViscOu5 Crude
10,
s= saturation,dimensionless Oil by In Situ Combustion;”, MS thesis,
T= temperature,‘F if not subscripted,and Petroleum Engineering Dept., The U. of
transmissibilityif subscripted,lb/D Texas (July, 1958).
psi 11. Marx, J. W. and Langenheim, R. H.: “Reser-
t= time, days voir Heating by Hot Fluid Injection”,
u. Darcy’s velocity, ft/D Trans., AIMi (i959) ~, 312-315.
u= over-all thermal coefficient,Btu/D Sq 12. MacDonald, R. C. and Coats, K. H.:
ft ‘F l!Methodsfor Numerical Stitiation of Water
VP . block pore volume, cu ft and Gas Coning”, Sot. Pet. En.g.J. (Dec.,
X,y,z = Cartesian coordinates,ft 1970)L25-436.
13. Rame~. H. J.: Discussion on “Reservoir
Greek Heat”&g by Hot Fluid Injection”, Trans.,
~. difference operator AIM (1Q5Q) 216. 36L+-365=
k-t..{ _r
p= viscosity, cp 14. Ramsey, P. E.: !!Recoveryof Viscous Crude
q= dimensionlessheight Oil by Steam Injection”,MS thesis, Petro-
~= porosity, dimensionless leum EngineeringDept., The U. of Texas
p= density, lb/cu ft (MaY, 1958).
r= dimensionlesstime 15. Reid, S.: lt~e Flow of Three Immiscible
Fluids in Porous Media”, PhD dissertation,
Subscripts Chemical En ineering Dept., U. of
Birmingham f1956).
a. ambient condition 16. Richtmyer, R. D.: Difference Metinodsfor
c= condensate Initial-ValueProblems, Interscience
i. grid index in the x-direction Publishers,New York (1957) 101.
j= grid index in the y+irection 17. Sheldon, J. W., Harris, C. D. and Baviy,
A?= liquid D .: 11AMethod for General Reservoir
~. old-time step Behavior Simulation on Digital Computers”,
n+l = new-time step Paper SPE 1521-G presented at the 35th
Annual SPE Fall Meeting, Denver, Oct. 2-5,
REFERENCES 1960.
18. Snell, R. W.: t!~ree-phaseRelative
1. Buckley, S. E. and Leverett, M. C.: ‘T4ech. Permeabilityin an UnconsolidatedSand”,
anism of Fluid Displacement in Sands”, J. Inst. Pet. (March, 1962),!@,80.
Trans., AIME (1942) ~, 107-116. 19. Shutler, N. D.: ~~N~eric~, Three-Phase
2. Coats, K. H.: ncomput,er simulation ‘f Simulation of the Linear Steamflood
Three-DimensionalThree-PhaseFlow in Process”, Sot. Pet. EnR. J. (June, 1969)
Reservoirs”,UnpublishedReport, The U. 232-246.
of Texas at Austin (Nov., 1968). 20. Shutler, N. D.: t~N~ericd Three-Phase
3. Coats, K. H.: P. En. 383.21 class notes, Model of the Two-DimensionalSteam Flood
Petroleum Engineertig Dept., The U. of Process”, Sot. Pet. EnR. J. (Dec., 1970)
Texas at Austin (1967). 405-417.
4. Coats, K. H.: Private Cormnunication(19681 21 Spillette, A. G. and Nielsen, R. L.: “Two-
●

5. Coats, K. H., Nielsen, R. L., Terhune, M. Dimensional Method for Predicting Hot
H. and Weber, A. G.: “Simulationof Three- Waterflood Recovery Behavior”, J. Pet. Teck
‘E 36c0 A. ABDALIA and K. H. COATS 7

22 (June, 1968) 627-638. a2T aT

Stone, H. L. and Garder, A. D., Jr.:
●
Kz —= Prcr=. . . . . . . (A-3) ●

!?An~ysis of Gas-Cap or Dissolved-Gas az2

Drive-Reservoirs”,Sot. Pet. E%q. J. (June, The boundary conditions are
1961) 92-104. H
23. Willmsn,
- B. T., et al.: “Laboratory T=Ti atz=7 and t = ti . . (A-4)
Studies of Oil Recovery by Steam Injectiont’
J. Pet. Tech. (July, 1961) 681-690.
T+T
APPENDIX A aas z+rn) . . ● . . . . . (A-5)

Heat Loss Calculation The initial condition is

In the computer simulator developed in T=T a at t = O and all z . . . (A-6)

this study two procedures are used to calculate
the heat loss. One is used in the testing of Let
the physical models, and the other is used in 4 Kzt
the field case studies. ‘c= . . ...* . . . . . (A-7)
H20rcr
Heat Loss Caicuiation for the Physicai Modeis
. (A-8)
Physical models are made with limited in- ~= e”””’””””” ““”
sulation thickness. A representationof the
heat loss in terms of an average over-all (A-9)
thermal coefficientthat can be determined in Z=T-T ● ● ● ● ● ● ● ● ● ● * ●

a
the laboratory will best suit such cases. The
Then Eqs. A-3 through A-6 will be
following is the equation used for cylindrical
insulations around a cylindricalcore holder: a2z az (A-1O
—=E”
2 “ “ “ “ “ “ ● “ “ “ “ “
Heat loss = n dU Z x, . . . . (A-1) an

where d = outside diameter of the insulation,fi

2=2 iatrl= 1 and -r = Ti . . (A-n
U= over-all thermal coefficient,Btu/D
sq ft ‘F (A-12
Z+l) asq +m* ● “ ● “ ● “ “ ● ●

Z = difference in temperature across the

insulation, ‘F
x = block length, ft
z = O at T = O and allne w ● 4 (A-13
Since the over-all thermal coefficientof Performing Laplace transform on Eqs. A-10
-.“-..
=+.eam -- --..“.v.....
i Q di ffaven+. +.hnm
.. . +.hn+.
..... of’~iq~d~, a and A-ii and using Eq. A-13 we get
weighted average value was used in-this”study
and is given by the following equation: az- S~=O. . . . . . . . . . . (A-14
u = u~s~ + Ul(l - Ss) . . . . (A-2)
av
Heat Loss Calculation for Field Cases =land~=~i. .(A-15
In field cases, the overburden and the The solution to Eq. A-14 is
underburden can be considered as infinite insu-
lations. The equations under considerationare Q/’-is -r)v%-
Z=cle +c2e (A-16
those describing the heat flow in a semi- .0..

infinite slab. Their solution is made using

T.-1
udp~acc
. . . 4 . . . . . . .
u~-ala~uillla
..#--- ~~: A-lz uives ~j, =
m
be 0-”--
Consider a reservoir ~
sand of thiclmess h. The ‘ ~.ce2 -T- ds”
. ..*. . . . . . (A-17
z-axis runs parallel to
the heat flowing to the Using Eq. A-15 in Eq. A-17, we get
overbtien as in the fig- ! overburden ,
ure. The partial- Zi
-/5
differential equation de- H ..... 3--=c2e
scribing the heat flow is 7, ‘“ ‘“””‘-”
+r,l 1 n,.,=.
as A.uA4.uwa.
reservoir then z
i&_
I C2
=+
s
o ~_._ ._._..
 A THREE-~SE , EXPERIMENTALAND NUMERICAL
SIMULATION STUDY OF THE STEAMFLOOD PROCESS SPE 3600

kxkro P.
a. ‘— . . . . . . .*.** (B-2)
!-l.
-(l-l-l)&
~=z%
iS
. ..*.*. (A-18) The p values in these spatial differences
are unders?ood to apply at the new time level
Performing the inverted Laplace transform
Lo Eq. A-18, we get ‘n+l”
n-l The backward time difference approximation
z(~, ~)=zierfc —’.” *O (A-19;
●
is used for the time derivatives as follows:
26“

Using Leipnitz rule to differentiateEq. A-19, % (+h)%))n+l

- (Qp($())
n ,.
&@Poso) = At
we get
-(T1-1)2 . . . . . . .*.*** ● ☛☛☛☛☛
(B-3)
Zi (n-l) e 4~
az
—=— Using these finite+ifference approximationsin
a~ 2G Eq. 3 and multiplying both sides of the equation
rT3 “ ““ ““ ‘A-20’ byAAx, we get
lrn ai
Rate of heat loss per unit volume = Prcr
1 z
L AXTOAXPO = ‘~At(poSo) . (B-4a)
_(n-i)2
2Kz Zi 4T 4Kz Zi
= fm(n-l) e .— AXTWAXPW + qc = ‘&t (pwsw) . (W4b)
h2 Td= 1 h2 d=
. . ...* .. *.** ● ☛☛☛☛✝
(A-21
AxT~AxP~ - qc = ‘$At(P&) . (B-4c)
Eq. A-21 gives the rate of heat loss per unit
volume. where
APPENDIX B AxTOAxpo = To (Po -J?o)
i+l\2 i+l i
Finite-DifferenceExpansion
- To (Po -Po)
Before starting any finite-difference i-l\2 i i-1
expansion, the grid system must be specified. . . . . . . . ..*** ● ***** (B-5)
The selection of such a system depends upon the
boundary conditions to be used. kxkropo
= 6.33 ‘=
‘o. i+l\2 ‘M)
For the linear model to be developed here, 1+1/2 1-’o
predeterminedinjection and production pressure
will be used as boundary conditions. The grid and A is the cross-sectionslarea perpendicular
system shown in Fig. 7 is the most suitable for to the flow.
+.h;~r.a~e.
“A.-s----- Au finite-differenceexpansions
that follow in this chapter pertain to such a Applying the same finite-differenceapproxi
grid system. mations to the heat-bslanceEq. 6 and multi–
plying both sides by A&, we get
Eqs. 3 contain second-orderspatial deriv-
atives and first-ordertime derivatives. The DXA
standard central-differenceapproximationfor ~A;zn - in - AAx(uPhn) + qshinj
the spatial derivatives is as follows.

v
. fiAt(~ph + $+rZ), ● (B-i’)

.
IPO -pOm, -a. P. - P. where
30.
1+1/2\ i+l il i-l\2 ( i i-1 Ax(uopoho) = A(uoPo)l+l/2hnoi
.
AxL -A(uoPo)i_l/2hno
. . . . . . . ..*** ● ***** (B-1) . . . . . . . ...0. ● .*** ‘:l(B-8)
where = TO (Po - PO.)
‘(uOpO)i+l/2 i+l
i+l/2
. ..*.* ● ...*. .***** (B-;)
m 3600 A. ABDAILA anc

At (pOSOhOl = (Pos&)n+l - (~osoho)

n }, Atso = () ,
- a3ps
. . . . . . . . . . . . . .*.. (B-lo) n+lj

A:zn = Zn -2zn+ Zn ● (B-II)

then
i+l i i-1
‘o
n+l
APPENDIX C a3=~ “ “ ● “ “ “ “ “ “ “ “ ● (c-3a:
n+l
IMPES Application
P
IMPES is a technique in which the pressure ‘n+l
in the flow term, ATAp, is handled implicitly, al=a3P “ “ “ “ “ “ “ “ “ ● “ (C-sb:
while the saturation and saturation-dependent ““n+1
parameters are handled explicitly. This tech-
nique is described in the literature.17,22 In SubstitutingEq. 5 in Eq. 13, we get
this anaiysis tlnistechniq~ is applied in a
manner described by Coats. AXTAXPO -a A TA P
1 x w x co_,J

Eq. B-4 can be rewritten as follows. + a3AxTsAxPC + (al - a3)qC

s-Cl
AxToAxpo =
‘P
= ~(alsw AtPw + So At~o
‘P & n n
(C-la)
~ ~on+l *ts(3+ At On ‘tpO
Atps), . ● . . c . ● . ● (C-4)
+ a3%=
AxTwAxpCg + qc =
where
v
‘P AtSw +%Ap . (C-lb)
~Pw At Wn tw”
n+ 1 ‘= alTw+To+a3Ts” “*”””(C-5)
AxTsAxp~ - qc = In forming AxT~PO, care must be taken ti
leaving al and a3 outside the spatial differenc(
‘P ‘P ~
EQsn+l ‘ts~ + m Sn ‘tps “ “ (C-lc) Since oil and water densities have been
..-,.. considered in this study ljQbe flm.ctiQnsof
rm.uupqimg Eq. C-lb by al and Eq. C-it by a3
and adding the three equations,we get temperature only and steam density is a func-
tion of both temperature and pressure, the term
AXTOAXPO + alAxTwAxp,d + a3AxTsAxps ~Ps can furtherbe expanded as follows:

Atp5 = Atp; + p:Atp , . . . . ● (C-6)

+ (a, - a3)qC
L = ‘~
At
p. AtSO where
n+l
Atp; ‘ps(zn+l,Ps ) - Ps(znfP~ )(C-7)
1 n n

+ alpwn+l ‘tsw + a3psn+1 ‘tss..’ Ps(zn+~lPn+l) - Ps(zn+lfP5 )

n
P: = -p
P5 s
n+l Ii
.&s on ‘tpo + Wwn ‘tpw
. . . . . . . . . . . ...* . . . (c-8)

From Eqs. C-4 and C-6 we get the following:

. (C-2) k+l + ~k
AXTAXPO = GAtpo , . . . (C-9)
+ a3%n%ps “ ‘ “ “ “ “ “ “ ●

.-
where k
Using Eq. 4 and choosing al and a3 such that B= (al -a3)q~ - alAxT,dAxPc
o-w
+aA?AD
3 x s xc
s-q
v
 A ~~.~~~ ; -IMENTAL AND NUMERICAL
) SIMULATION STUDY OF THE STEWFLOOD PROCESS SPE 3600

D-1 --1.- ..G+L V--* C+a.m

A P*] . D-LuGfi3 W.1.L/LL XLGG UVV-U
. (c-lo)
+ son%~o + a3%n t s
After solving the fluid flow equations for
and the pressure distribution,the steam saturation
temperatures for blocks with free steam are
G=v$a3ssp~ . . . . . . . . (C-II) determined. The use of these temperaturesin
n the heat-balanceequation will result in resid-
uals. These residusls are due to the use of the
The superscript (k) shows that the value at rate of steam condensationat the old iteration
the old iteration is to be used. The super- in solvin~ the fluid flow equations. Correction
script (k+l) shows that the value at the new of such v~ues will reduce ~he residuals to
iteration is to be used. within limits of tolerance.

APPENDIX D Denoting the residual of the heat-balance

equation at any grid point by R, we then have
Rate of Steam Condensation

The calculation of the rate of steam con-

densation is made by the use of two sets of
equations. The first set is for blocks that
where (wf) is a weight factor to be chosen in
have no free steam, i.e., their temperatures are
a way that will accelerate the convergence.
below the saturation temperaturesof steam. The
second set of equations is for blocks that have
APPENDIX E
free stesm, i.e., their temperatures are equal
to the saturation temperaturesof steam.
Data Used for Calculations
Blocks with No Free Steam
This appendix contains data used in the
operationalmodels. The relative perme-
In these blocks, all the steam coming in
abilities, capillary pressures and dispersion
from adjacent blocks is condensing,i.e.,
coefficientsfor the linear model study are
obtained from Shutler.20
q= = Ts (Ps -PSI
x. “ i-l,j i,j
1-1/2 ,j E.1 Linear Experiment I

E.2 Linear Experiment II

+T (PS -PSI E.3 Two-dimensionalexperiment
‘Y.l,j-l\2 i,j-1 i,j
. . . . . . . ..*.. . ..*.. (D-1)

E.1 E.2 E.3

k (darcys) 2.54 ‘2.54 132

.354 .372

3=42 3.9

H (ft.) .83

UL(Btu/day.ft. F) 6.2 6.2

Us(Btu\day.ft. F) .204 .204

80 80 24

1<7
A“# 167 167

Cr(Btu\lb. F) .1855 .1855 .2156

Ta ( F) 80 80 80

Sw .229 .229 .1
i
m 3600 A. AEDALIA and K. H. COATS 11

E.1 E.2 .E.3

T inj ( F) 267.25 266.63 400

P inj ( Psi) ‘25.3 24.9 260

( Psi) 13.5 0 190

‘prod

E.1 and E.2

Pc P
Sw so c
— o -w — .—
s -o

.2287 2.2 .3 .38

.30 1.0 .4 .29

.40 .7 .5 .21

.50 .52 .6 . 16

.60 .37 ,7 ___

12

.70 .23 .8 .11

.90 .1

k
SW k ro
— rw o -w

.2287 o 1.0

.3 .002 .922

.4 .009 .8

.5 .012 .58

.6 .019 .26

.7 .022 .06

.9 .042 0.0

k
so ro k
— s-o rs

.2 .0008 .175

.4 .01 .105

.5 .04 .05

.6 .125 .01

.7 .38 .001

.8 .7 .0
 A THREE-FHASEEXPERIMENTALAND NUMERICAL
SPE 3600
12 SIMULATION STUDY OF THE STEAMFLOOD PROCESS

E.3
Pc
Sw 50 s-o
— —
.1 4.1 .1 4.517

.2 .095 .2 .067

.3 .072 .3 .042

.4 .061 .4 .02

A,-, .5
.5 .U3L

.6 .041 .6 -.022

.7 .031 .7 -.043

. 8 .021 .8 -.064

. 86 .011 . 89 -.085

k
k ro
rw o -w

.1 o 1.0

.2 .0016 .875

.3 .0081 .735

.4 .0259 .590

.5 . 0672 .42

.6 .1 .21

.7 . 14 .07

.8 .20 .016

. 86” .25 G

k k
ro
s -o rs
—

.1 0 .52

.2 .009 .41

. 3 . 031 .31
.22
.4 .062

.5 .11 .14

.6 .19 .08

.7 . 335 .03

. 8 .570 .005

. 89 1.0 0
SPE 3600 A. ABDALLA and K. H. COATS 1:

Temperature Viscosity (Cp)

80 800

100 330
140 110
1$?n
. .. ~~

240 18
280 11
360 5.26
‘ 450 2.9

TABLE 1 - RESULTS OF EXPERIMENT 1, LINEAR MODEL, PORE VOLUME = 494.14cc

Pressure TemperatureF
Fluids
Time (Psi) Produced(cc)
(min.) DistancefromInlet

In- out- 011 Total 1.2” 8.9” 15.6” 23.3” 31!1 38.7”
let let
25.2 13.4 0 0 80 80 80 80
3: 25.1 13.3 6.6 6.6 84 80 80 80 % %
60 25 13.2 13.4 13.4 96 80 80 80 00 00
90 25.1 13,2 20.3 20.3 107 80 00 80 80 80
120 25.2 12.9 27.9 27.9 116 80 80 80 80 80
150 25.2 13,2 35.8 35.8 124 80 80 80 80 80
180 25.1 13.3 42.9 42.9 132 84 @ 80 80 80
210 25.2 13.1 50 50 137 85 80 80 80
240 25.3 13.2 57.8 57.8 138 88 80 f% 80 80
..,s ~~ ~ ~~ ~ ~~~ w On
Clu 12.S 68.2 w
25:3 12.9 77.5 77;5 147 92 80 %
~% 25.2 12.8 87.6 87.6 154 93 80 80
360 25.2 13.2 98.5 98.5 161 96 80 80
390 25.2 13 109.2 109.2 168 98 80 80 00
420 25.2 13.2 121 121 174 100 81 80 ::
450 25.2 13.1 134.8 134.8 183 101 82 80 t% 80
4s0 25.2 13.1 149.8 149.8 192 105 83 80 80 80
510 24,8 13.1 171 171 200 108 05 80
24.9 13.3 191.8 191.8 205 112 06 :; 80 ::
570 24.9 13.1 219.5 258 117 87 80 80
600 24.9 13.1 212 260.5 279 125 88 % 80, 80
640 24.9 13.2 221.3 309.3 279 144 90 82 80
680 24.9 13.2 226.5 358.5 279 180 98 83 z 80
720 24.8 13.2 232.5 408 279 223 106 84 80 80
760 24.8 13.2 237 454 279 247 115 80 80
000 24.9 13.2 241.5 503 279 275 128 % 80
850 24.9 13.1 247.6 564 279 275 143 95 82 ::
24.9 13.2 250.9 599.2 279 275 152 99 82 80
n 24.9 13.2 254.9 636.7 279 275 162 102 83 80
940 24.4 13.0 259.1 676.6 279 275 172 106 84 80
970 24.9 13.2 263.6 715.5 279 275 182 109 86 80
1000 24.9 12.9 268.4 753.9 279 275 192 117 89 80
1030 24.8 12.8 272.5 790.6 279 275 199 119 90 80
1060 24.4 13.3 277.3 830.6 279 275 211 124 92 80
1090 24.8 13 282.3 874.8 279 275 228 129 94 83
1120 24.9 13 287.4 917.4 279 275 244 135 84
11s0 24.8 12.8 295.8 1008,1 279 275 266 149 1:: 85
~2~Q 24.8 ~z.g 298,2 1050,6 279 275 266 157 104 85
1240 24.8 12.9 301.8 1095.4 279 275 266 164 106 86
1270 24.0 12.8 305.3 1140.1 279 275 266 170 108 86
1305 24.8 12.8 308.9 1186.7 279 275 266 175 110
1345 24.8 12.3 312.9 1251.7 279 275 266 184 114 :;
1385 24.0 12.2 317.4 1310.2 279 275 266 198 123 92
1430 24.8 13.0 321 1360.5 279 275 266 205 127 93
1460 24.9 13.0 324 1406.4 279 275 266 210 130 94
1500 24.8 13.0 326.2 1451.6 279 275 266 212 134 98
 TABLE 2 - RESULTS OF EXPERIMENT 2, LINEAR MODEL, POKE VOLUMi = 494.14CC TMERMOCWPLES

INLET
TemPeP.ture F CONOENSEO PRESSURE
?mfi FluId* sTEAM
Time P,oiueea i,.)
Distance frca Inlet TRAP
(min.)
.....
In- out- 011 Total 1.2” 8, 9“ 16.6” 23.?” 31” 30.7”
let let !
o 24,8 0 0 0 W608060W N3

40 25,3 0 20.4 20.4 llo 00 W 00 00 W MC K- PRESSUFIE

TO COIL HEATII13 REWLATOR
so 24.0 0 41,2 41,2 133 23 00 00 60 eo

120 24.7 0 65 65 lmzzboeom No

u

154 24,7 0 91.5 91.5 176 96 W 00 80 24

Fig. 1 - Schematic diagram of the epparatus.
2U0 24.5 0 122.7 122.7 210104 s4m2a 00

240 24.3 0 160.3 160.3 279 11629WW 84

220 24,6 0 lm.2 222 279 170 92 E4 20 60

320 24, e 0 207.5 305.5 279 234 102 06 00 6.3

360 24.0 0 222.1 381,0 279 275 172 90 20 80

4s0 24,9 0 237.3 448 279 275 264 106 6~ en

440 24.9 0 247.2 514,6 279 275 265 144 90 w
490 24,6 0 262,6 394.e 279 275 26S 170 98 20

320 24.9 0 275.5 677.7 279 2?5 265 254 120 80

560 24,0 0 226.4 76E.2 279 275 265 2S 186 96

6SCI 24,9 0 2’J7 .3 849.6 279 275 265 234 214 102

640 24,9 0 3043,1 929.1 279 275 265 254 236 124

E/L77

f!

[: //
.1
s
o .4 .a hp. 1.6 do M h
CUMPiJLATIV”-TOTALFLuIOS PRODUCED-PORE VOLUMES

Fig. 2 - Viscosity emd specific gravity vs temperature for Primol 185. Fig. 3 - Experimental and calculated results of linear Experiment 1,
AP =11.8 Psi.

SW
A A TEMPERATUREAT 8,9’
EI.
● $!50 FROMlNJECTIONPOINT

&’J AA
m 100
#/

“~ A =
1.
r
a
-.
m .?T

&
●
*SO -

‘R.. b
o
>
.6
b

u
.5-
= too - :
~ g .4-
\
: A tSipERlUEt4TAL ●

p ,s0 - -o- CALCULATED

Z .~-
\
~ .e-
<._
100 - a
J -
-1

L———————
0 to s 1s 20 !?6 so 66 L. a Inctms
Zio

o .4 .6 l.z 1,6 Zo 6.4 M

OISTANCE FROM tN.3ECT10N POINTS - INCHES CUMULATIVE TOTAL FLUIDS PRODUCED- PORE VOLUMES

Fig. h - Temperature distribution at the end of Experiment 1. Fig. 5 - Experimental md calculated results of linear !kPeriMent 2,
AP =24.9 Psi.
 Read Data Compute enthlalples and
Initialize the Model saturation temperatures
A

t I
and
Prtnt
Initialized
Data
Conditions
I Compute saturations
I

I
+ t

I
A EXPERIMENTAL Compute temperature and

;1
Compute: ralte of steam.condensation
-o- CALCULATE from the bleat balance
- Pore volume
;- Inltfal oil and -———————l——————
water
3 - Flow rate
Iln place
L—
coefficient

+
c Check Convergence 1
50 I I I I 1 1 I ~—
0 5 10 Is 20 25 30 35 L~141 inchc$ ~-
OIST’ANCE FROM INJECTION POINTS - INCHES

Fig.6 - Temperature
distribution
at the end of Experiment
2.
Compute

1 - Relative
from tabl(e

p(ermeabillty
look-up:
I Compute material
heat ba[lances
and
I
2 - Capillary [pressure
- Viscosities
: - Densities t
Compute transmissibil ities

I
I Incrememt
and print
time
results

t
r—
Update
l’———

~1 Establish F1OW Rate L#

densities

t
1 2 3 N-2 N-1 N L —~

Fig. 7 - Grid system used in the linear numerical simulator. Fig. 8 - Nlurnerica.1 simulators
flowchart.
 .6,
r I I I I !
● **.*

$ ‘5
~ 2–s’’’”’””
● **OS /
i,.i+l STEAM S.T. A
! ,4 /
●
● *
i-; ,j i~j i+l ,j
[

w
B .x\
>1 .- Y
a
:
8
a .2
d en- Celculoted using tabulated
5
-c- Calculated using Pc t.ab.lo?edltO
Fig. ; - Grid w.t,n used in the
tho-airqensicmal numerical : /;-’r
H,@gins, .!01 water fled (11)

s ir.uiator.

.2 .4 .6 .8 1.0 1.2 1.4 1.6

PoRE VOLUMES PROOUCEO

Fig. 10 - Experimental and calculated results of

the two-dimensional experiment.

I 0,000, ! 1 I 1 d

I JMo
N i
0.6

z ~ 0.5
I S
~
: 100
o
z >
o ~ 0.4
.RtMwL gil Waco sitv fat Room
v a Tomverotwe, c
~ g K II
>
I
● 0.3
-

g
10
: 0.2
J
G
0.1

I I I I I I 0
106
I 0 0,2 0.4 0.6 0.s 1.0 1,2 1.4
200 250 300 350 400
50 100 150
PORE VOLUMES PROOUCEO
~EMpzR~~,~:~ .~

m, “
. .=. JQ .
.. rmn->,+.d Gil
.. ...=....- recQvery curves for different oil viscosities.
Fig. 11 - Viscosity vs temperature