You are on page 1of 12

SPE

SPE 200;3
Appraisal of Analytical Steamflood ModeIs
H-L. Chen, Texas A&M U., and N,D. Sylvester, U. of Akron
SPE Members

CopW9htWSO,SOCletY
of PetroleumErrgirreeca
Inc.
TIIls paperweepreparedfor preaantationat the @Oth
CaliforniaRegionalMeetingheld irrVentura,California,April4-S, 1S90.
Thk paparwasaalectedfor presentation
by errSPE Programcommitteefollowlnareviewof inf-tm
~taiti
in O@abefrecrWbmiff-f by the authort$).~t~fe
of the PWft
aa preaanted,haverw+beenravkwed by the societyof PelrofwmEmirind Me wbl~ to OWTOC~On
by the a~~e). ~ mat~al. M Pfe*ntW, *
~t MI=@
anYiwainonof theSoaietyof PetroleumEngineers,heoffkem,w mamb- pPf-tti
at SPE meetin%e
Wew@M to P@l~t~ rS~SWbyE~tofial@mIIIMOOS
of*V*VW
&~~m.
PmMto~b-dmm~titi*~-.lwftiNy
N@~.~~*M~&n~~
Telex, 7S0SSSSPEDAL.
of whereand by whomthe paperk preeentad.Write PublicationsManager,$PE, P.O. $0x -,
Rkh~*t
~ 7~.

Moving away from the injection well, the steam temperature


drops graduaily as the steam expands in response to the
pressure drop and heat losses to base formations. At a certain
distance, the steam condenses and forms a hot-oil bank. In the
steam zone, oil is displaced by the steam. In the hot oil zone
several changes take place which result in oil recovery. They
include heat losses the formation, thermat expansion of the oil,
and reduction of oil viscosity. In addition, residual saturation
may decrease and changm in relative permeability may occur
due to the variations of temperature and saturation.

$teamflooding in heavy oil reservoirs is one of the


principal thermal oil recovery methods. This paper evaluates
the existing analytical steamflood models with respect to their
mechanisms and predk!ive capabilities and compares them
with field data. The three steamflood models selected were: a
frontal advance model [Jones (1981)], a modified frontal
advance model [Farouq Ali (1982)], and a vertical gravity
override model [Miller and Leung (1985)]. Each model was
somewhat modified to improve its ability for the prediction of
production
rate and/or history match of typical field
production data.

There are three major options available in literature for


predicting the reservoir response to steamflocding. These
include:
empirical
correlations 2) , Simple analytical
models(l 13-7), and muiticomponent, multiphase numeri~al
simulators(8-11 ). Empirical correlations can be useful for
correlating data within a field and for predicting performance
of new wells in that or similar fields, However, use of such
correlations for situations much different from the ones that
led to their development can result in large discrepancies for
Numerical simulators yield rigorous
hist~. ~ matching.
solutions to the material and energy balances, However, their
results are sensitiva to the rock and fluid property input data
and other geological information, some of which may be
unattainable. In addition, large computation time is required
and numerical convergence, and stability problems suggest
that thermai simulators are not appropriate for short-cut
design and/or preliminary evaluation for steamflooding
projects. Thus, the incentive to develop simple analytical
models which account for the important mechanisms invotved
and for routine or approximate engineering prediction is
obvious. The existing analytical steamflood models can be
divided into two categories:

The Jones steamdrive model, with its empirically


determined scaling factors, was found to give a reasonable
history match of oil production for the Kern River field.
Fields with different
characteristics
will require an
adjustment of these scaling factors artdlor field property data
to achieve an acceptable history match. The modified Farouq
Ali steamdrive model gives a good history match without need
for empirical factors or adjustable parameters.
It is thus
recommended for the prediction of steamdrive oil recovery
when fisld production data are unavailable. The Miller-Leung
gravity override steamflood model, which contains two
adjustable parameters, was found to posses the best W3rail
history matching capabilities and is recommended for this
purpose.
AND I ITFRATLW.BUUW
The injection of steam into heavy or pressure depleted oil
reservoirs has been a successful enhanced oii recovery
process for more than three dsoedes. A principat application
of the steam injection is steamflooding which is also termed
steam drive or steam displacement. In this process, steam is
continuously injected into a number of injection wells, and the
dispiaced fluids are produced from the production wells.
Ideally, the injected steam forms a steam saturation zone
around the vkinity of the injection welL The temperature in
the steam zone is nearly equal to that of the injected steam.

1. Frurrta/ advance models: The steam-drive mechanism is


modeled as a horizontal frontal displacement [Figure
1(a)]. The steam zone IS assumed to gmw horizontally
and the tendency of the steam to finger beyond the front
is suppressed by condensation.
2. Verlikal displacemet?t or gravity overz%le models: The
problem of gravity override of the steam due to its low

References and figures at end of paper.


la
. .

APPRAISALOF ANALWICAL STEAMFLOODMODELS

density assumes that the principal direction of steam


zone propagation is vertically downward [figure
2(b)].

SPE 20023

The purpose of this paper is to evaluate existing analytical


steamflood models with respect to their mechanisms and
predictive features. Three typical steam flooding models were
studied and modified by Chen(21 ): Jones(l) frontal advanced
model, Farouq Alis(5) modified frontal advance model, and
Miller and Leungs(6) vertical gravity ovarride model.
History-matching of field data were carried out for each model
to test its applicability.

An early frontal advance model was that of Marx and


Langenheim(l 2) who applied an energy balance of a radially
growing steam zone In which one-dimensional conduction heat
losses, uniform steam zone and reservoir temperature were
assumed. Willman el al.(13) presented a model similar to
that of Marx and Langenheims but included the BuckleyLeverett equation to estimate oil production from a hot water
zone ahead of the steam zone. Mandle and Volek(l $) extended
the concepts of Marx-Langenheim by including convective heat
transfer from the steam zone into the region ahead of the
condensation front at times greater than a critical time. The
model was modified by Myhill and Stegemeier(l 5) to calculate
the thermal efficiency after tha critical time to account for the
disparity observed in physical models versus theory.
Jones(l) noted that the Myhill and Stegemeier model often
overestimates the oil production, especially in the early phase
of a project because of the assumption that the oil displaced by
the steam zone is immediately produced. Thus, there was no
lag in oil production due to fill-up of any gas volume, or due to
the development of an oil bank. Jones(l) thus developed a
modified predictive model including the results of van
Lookeren(l 6) for taking into account the extent of steam
override, and introduced three empirical factors to account for
the dominant mechanisms during the three stages of
production.

Table 1 summarizes the characteristics


and the
parameters for three steamflood models. Complete parameter
sensitivity
analyses for each model are available in
The major modifications for each
Chens(21) dissertation.
model are presented in the Appendix section.

Jones(l)
applied van Lookerens(l 6, method for the
optimal steam injection rate for a given set of steam and
reservoir
parameters,
and utilized
the Myhill and
Stegemeier(l 5) method to predict oil production.
In the
Myhill and Stegemeier model, the average thermal efficiency
of the steam zone was calculated by the Marx and
Langenheim(l 2) solution at early times while the Mandl and
Volek(l 4, method was used to account for heat transfer
through the condensation front after the critical time. Jones
model contains a number of empirical factors (ACD, VODt VPD)
which were obtained through history matching for specific
sets of field production data. Thus, the adjustment of field data
may be necessary (TR, ht,hn ,t.toI) to achieve reasonable
history matching for some projects as shown in Jones Table 1.
In the original Jones model, the steam injection pressure was
calculated assuming a geometric relationship between
pressure and injection rate. The optimum steam injection rate
is taken to be the steam injection rate which gives the
maximum value for the vertical conformance factor (AR D)
Unfortunately, steam Injectivity test data is often not available
in the field. Therefore, the computer program written to
evaluate the Jones model was modified to allow input of steam
injection rate and pressure, This modification was necessary
to permit comparison of model predictions with actual field
data.

Neuman (2S17, and Rhee and Doscher(3) proposed that


the principal direction of steam zone growth is vertically
downward In the horizontal reservoirs. Neumans(17) model
requires the data of relative permeability to oil and water as
functions of temperature.
Also, oil production from the
condensate zone was determined semi-empirically. Aydelotte
and Pope(4) used fractional flow theory and overall energy
and material balances to account for changes in oil cut, gas
production, etc.. Also volumetric sweep efficiency was taken
into account by using van Lookerens( 16) vertical sweep
efficiency
and an empirical correlation given by Farouq
All( 18) for areal sweep efficiency {EA).
This model is
restricted to horizontal,
homogeneous,
isotropic,
and
incompressible
reservoirs
and only five spot sweep
corrections were included.
Doscher and Ghassemi(f 9)
proposed that !he steamflood process consists of the heated oil
displaced by a gas drive mecharrism. Their model showed an
insensitivity of oil recovery to formation thickness, especially
during the early stage of production.
Their experimental
results indicated that the oil/steam ratio increases with a
decrease of oil viscosity.

Farouq Atis(5) model is a modified fontal advar ]d model


which considers the effect of steam gravity override using van
Lookerens( 16) method. At any instant of time during the
production, the model predicts both oil and water productiondisplacement rates, the steam zone volume-thickness, the
heated zone average temperature and the water and oil
saturations. An advantage of the model is that It simulates the
dominant mechanistic features by material and energy
balances and does not employ empirical factors. However, this
produces the modets primary disadvantage in that several
parameters such as Sorst, Sor, Sst and Swir are required
which, unfortunately, are normally unknown and need to be
assumed or defaulted by using acceptable values. Also, it has
been shown by Chen(21 ) that the water saturation during
production affects the relative permeabilities
and the
production rate, and the model predictions are very sensitive
to the accuracy of the Krw and Kro versus SW* which are
difficult to obtain through experiments.
Even though the
experimental difficulties can be overcome, the data may not
represent the actual relative permeability versus saturation

Unlike previous models, Vogel(20) proposed that oil


production was not driven by the growing steam zone, but vice
versa, He pointed out the general weakness of predictive
models based on simple energy balances of a growing steam
zone. With a predominantly overriding steam zone, the heat
balance calculations require that the steam produced In
production be accounted for as well as the steam that migrated
out of pattern, Vogel suggested that the total underground heat
requirement was equal to the heat in the steam chest plus the
heat flow upward and downward from the steam chest. He
concluded ;hat oil production must be determined from some
way other than steam zone growth, Miller and Leung(6)
utilized the concepts of VogeI(20) and Neuman( 17, to
determine tka oil production rate by conductive heating of the
oil below the steam zone.
I
----

H. L. Chen and N. D. Sylvester


relations due to the effect ot temperature and reservoir
heterogeneity. ,. I.} relative permeabllities versus saturations
equations presented by Farouq Ali were based on the curves
presented by Gomma(22), These normalized curves were
obtained through history matching of the Kem River field data
reported by Chu and Trimble (23), The prediction of the
original Farouq Ati model for the Kern River A field production
data indicates it to be totally inadequate at long times (>1.5
years). Several important modifications were able to take into
account heat losses [Figure 2(a)] and displacement mechanism
[Figure 2(b)] to improve its deficiencies. These are discussed
in Appendix (b).

Miller and Leung(6) developed a simple gravity override


model which assumed a complete vertical overlaying steam
zone with a steam-condensate zone between the steam zone and
the oil zone below. They used one-dimensional, unsteady state
heat conduction to calculate the temperature distribution
inside the COIIdWISate and oil zones, and employed tha
Neuman(l 7) method to determine condensate zone thickness as
a function of fract;on of condensed steam that is produced from
the reservoir (fcp: 0.7-O.95). They ciaimed that the modei
overrxedicts the oil twoduction rate for fieid cases with iarge
patterns (> 10 acres) because the steam override may not be
fully developed in those cases. Therefore, another trmPirical
factor, the areal sweep efficiency (EA: 0.4-1,0) presented by
Aydelotte and Pope(4) was introduced for the field cases with
Iar9e pattern area. Chen(21 ) has shown that both values of
fcp and EA have substantial effects on the predicted oil
in addition, the heat baiance which
production rate.
determines the optimum steam injection rate was modified by
Chan(21) to take into account the fact that the steam injection
rate should be based orI cold water fed to a steam generator not
on saturated steam.

The five of fieid projects listed in Table 2 were chosen for


history matching. They represent smail [Kern-A(23), KernCanfield(24),
and Kern-San Joaquin(24)], medium [KernTen Pattern],
and large pattern areas [Tia Juan].
The field production history data for each fieid case was
adapted from the Enhanced 011Recovery Fiei6 Report (27). A
time ir?crement of 1.2 month was used for the prediction of
Kern River A project, and 1.5 month for Kern-Canfield and
Kern-San Joaquin matches. The time increment us d in
medals for Kern-Ten Pattern and Tia Juana was chosen ~~be
one month because the production history data was reported
monthly. it is noted that the Kern River A field data was the
oniy used to test the performance prediction for the modified
Farouq Alimodel because of the availability of reiati~e
permeability versus saturation relations which are required
by this model. The other four field production histories were
used to compare the predictive performance of the Jones and
Miiler-Leung models.

SPE 20023

Figure 4(a) shows that the Jones model predicts a lower


oil production rate at the beginning and a highar production
rate for the longer times for Kern-Canfieid project. Figure
4(b) shows that aithough the Jones modei underpredicts the
cumulative oil production, the prediction improves as time
increasp. As shown in Table 3, at the eyf of the 7.5 years, the
Jones modei overestimates the cumulative production by
it is seen in Figure 4 that the prediction of the
2.1 EYO.
Miiier-Leung model is superior to the Jones model for this
fieN case.
The comparison between the Jones model and Kern-San
Joaqukt field is similar to the Kern-Canfield case. That 1s,the
oil production rate is underestimated at short times and
overestimated at long time as shown in Figure 5(a)t while the
prediction of the Miller-Leung model is just the reverse. The
Miller-Leung model with a iag time (z) of 61 days is capable
of predicting the production up to about 1.75 years. The
computer run was terminated after two years bacause the
thickness of the condensate and steam zones became iarger that
the net thickness of the reservoir. Table 3 shows that the
Jones model overestimates the cumulative production by
8.1770 at the end of the third y~ar, and the Miller-Leung
model overestimates the cumulative production by 3.39 % at
the end of the second year.
Figure 6(a) shows that both the Jones and Miiler-Leung
modeis underestimate the oii production rate for Kern-Teil
Pattern field for the first two years. it also can be observed
from Figure 8(a) that the Milier-Leung
prediction is
superior to Jones modei during this time. Aithough the
predicted production rate of the Miller-Leung model decreases
sharpiy after 5.5 years, the MiIler-Leung model gives a more
accurate cumulative oil production up to about 5 years as
shown in Figure 6(a) and Tabie 3.
Figure 7(a) shows that neither model does weli in
predicting the measured oii production rates for the large
pattern case of the Tia Juana field aithough Figure 7(b) shows
that both models do reasonably well in predicting the
cumulative oil production. It should be noted that the Tia
Juana case is a poor candidate for triatory matching because the
less productive wells were steam stimulated, there were a
large number of unrepaired welis in the pattern, and the two
productive zones had oils of different viscosity.
This may
explain the observed decline of oil production rate.

The following conclusions can be drawn form the


resuits of the steamflood model modification and evaluation:
1, The Jones modei with input of steam injectivity data
can be used to predict oil production for steamflooding projects
with properties similar to the Kern River field. For other
cases, the empirical factors or input data may require
adjustment to achieve better history-matching.
2. The modified Faro~q Ali model is the most realistic
steamflood modei because it simulates both 011and water phase
dominant mechanisms (such as the combination 6f frontal
advanoa and steam override) by matedal and energy baiances.
In addition, this model gives reasonably good prediction and
history-matching results without requiring any empirical
However, retatlve
factors or adjustable parameters.
permeablilty versus water saturation data is needed for fields
otker than Kern River A to obtain reasonable historymatching.

Figure 3(a) shows, the performance prediction for the


Kern River A field using the modiflad Farouq Ali model. Also
shown are the predictions obtained using the Myhill and
Wegemeier (15) model, the numerical simulation results of
Chu and Trimbie(2~), and the actual field data. Figure 3(b)

compares the calculated cumulative production versus time


results to the field data. The agreement Is good with a
difference after 5 years of only 5.5% for cumulative oil
production. It is apparent in Figure 3(a) that the modified
Farouq All model gives superior predictions to those of
Myhill-Stegemeler and Chu-Trimble.
101

APWWSAL OF A?lALvrloAl STEAMFLOODMODELS


3. For history-matchingof field data, the modified Miller
and Leung model is better than the Jones model. Careful
adjustment of the parameters fcp and EA yields accurate
history-matching.
4. Use of the modified Farouq All model is recommended
for predicting steamflood production when field production
history is not available.
The Miller and Leung model is
recommended for trtstory matching of steamflood performance.

Sw/r

= irreducible water saturation, fraction

= time, hr

tc

= critical time, hr

c D

= dimensionless critical time

At

- time increment, hr

tB T

= steam breakthrough time, hr

T1,2

= temperature at conditions 1 and 2, F

Ts

= steam temperature, F
= initial formation temperature, F

TR

%4)

= dimensionless steam zone size

API

= specific gravity of oil at 60 F, dimensionless

c1
q,
&

= specific heat of phase i, Btu/lbm-F

fc p

= tondensed steam produced, fraction

fsdh
hfs

= cownhole steam quality, fraction

hn

= net zone tl ~ickness,ff

hs

= steam zone thickness, ft

= reservoir thermal diffusivity,

ht

- gross zone thickness, ft

@
z

= constant

v~
B

= areal sweep efficiency


x vertical sweep efficiency

SPE 20023

= bulk volume of the pattern, ft3


= VB -s(rr+l),
fti

oD

= dimensionless displaced oil prtiucad

vpD

= Initial pore void filled with steam as water,


dimensionless

Vs(t) = steam zone volume at time t, f@


VsBT = steam zone volume at breakthrough, ft3

= enthalpy of saturated steam at steam temperature,


Btu/lbm

ist

= steam injection rate, cold water equivalent BWp O

Kh

= thermal oonductfvity of cap rock and base rock,

Kro

Btu/ft-hr-F
x relative permeability to oil, fraction

Krw

= relative permeability to water, fraction

ft2/day

= porosity, dimensionless
(=3.14159)

P
s

= density of phase i, lbm/ft3

= viscosity , cp

Voi

= oil viscosity at initkd reservoir condition, cp

(n)

= at time step n, dimensionless

avg

= average temperature condition

sdh

= steam at downhole condition

= oil phase

= steam phase

= rock phase

= water phase

= lag time, days

Lvdh = latent heat of steam, Btu/lb

b
%

= heat capacity of cap rock and base rock, Btu/ft3-F

= oil originally in place, bbl

= cumulated oil displacement, bbl

P
%

9 heat capacity of steam zone, Btu/ft3-F

= cumulative oil production, bbl


= oil productionrate, BOpD

qoi

= pre-steamoil productionrate, BOPD

qw

= water production rate, BWPD

= heat Injection rate, Btu/hr

QI

- heat bsses to cap rock and steam zone, Btu

G>

= oil displacement rate, BOpD

Qw

= water displacementrate, BWPD

so

= oil saturation, fraction

%c

= condensate zone oil saturation, fraction

Soi = initial

oil saturation, fraction

Sor

= residual oil saturation, fraction

Scrst

= steamflood residual oil saturation, fraction

%s

= steam zone oil saturation, fraction

Sq =steam saturation in the steam zone, fraction


SW = water saturation, fraction
s~ - (~-swir)!(i -Swir-Sorw). dimensionless
mm

1.

Jones, J.:
~~~~~;able
.,

2.

Neurnan, C.H, A Mathematical Mo ,el of Steam Drive


Process-Application; paper SPE 47.,7, presented at the
California Regional Meeting of the SPE, Ventura,April 24, 1975.

3.

Rhee, S.W., Doscher, T.M.: A Method for Predicting 011


Recovery hy Steamflooding Including the Effects of
Dlstillatkm and Gravity Overrlde~ Sot. Pet. Eng. J (Aug.
1980) 249-66.

Steam
Drive Model for Hand-Held
Calculators, J. Pet. Tech. (Sept. 1981)

4,

20, Vogel,
J.V.: Simplified Heat Calculations
Steamflood,
J. Pet, Tech (July 1984) 1127-35.

Aydelotte, S.R, and Pope, G.A.: A Simplified Predictive


Model for Steamdrive Performance:
1983) 991-1002.

SPE 20023

H. L. Chen and N. D. Syfvester

J. Pet. Tech. (May

5.

Farouq All, S.M.: Steam Injection Theories - A Unified


Approacht paper SF 2 10746, presented at California
Regional Meeting of the SPE, San Francisco, March 2426, 1982.

6.

Miller, M.A. and Leung, W.K.: A Simple Gravity Override


Model of Steamdrive paper SPE 14241, presented at the
60th Annual Technkal Conference and Exhibition of the
Society of Petroleum Engineers held in Las Vagas, Sept.
22-25, 1985.

for

21. Chen, H.-L.: Analytical Modeling of Thermal Oil Recovety


by Steam Simulation
and Steamflooding,Ph.D.
Dissertation, The University of Tulsa, Tulsa, Oklahoma
(1987),
22. Gomma, E.E,: Correlation for Predicting Oil RecoveV by
Steamflood; J. Pet. Tech. (Feb. 1980) 325-32.
23. Chu, C. and Trimble, A.E.: Numerical Simulation of Steam
Displacement-Field Performance Applications, J. Pet.
Tech. (June 1975) 765-76.

7.

Wingard, J.S. and Orr, F.M. Jr.: An Analytical Solution


for Steam/Oil/Water Displacement; paper SPE 19667,
presented at the 64th Annual Technical Conference in San
Antonio, TX, Oct. 8-11, 1989.

24. Greaser----., G.R.


... and.
Shore. R.A.: Steamffocd Performance in
the Kern River Field, paper SPE 8834, presented at the
1s! Joint SPE/DOE Symposium on Enhanced Oil Recovery,
Tulsa, OK, April 20-23, 1980.

8.

Coats, K.H., George W.D., Chu, C. and Marcum, B.E.:


Three-Dimensional Simulation of Steamflooding: Sot.
Pet. Eng. J. (Dec. 1974), 573-92.

25. Oglesby, K.D., Belvins, T.R., Rogers. E.% and Johnson*


W.M.: Status of the Ten-Pattern Steamflood Kern River
Field, California J. Pet. Tech. (Oct.1982)2251-57.

9.

Crookston, R.B., Culham, W. E., anfi Chen, W. H.: A


Numerical Simulation Model For Thermal Recovery
processes,- Sot. Pet. Eng. J. (1979) 19, 37s58.

26. de Harm, H.J. and van Lookeren: Early Results of the


First Large-Scale Steam Soak Project in the Tia Juana
Field, West Venezuela J. Pet Tech. (Jan. 1969) 10110.

10. Vinsome, P.K.W., and Westeweld, J.: *A Simple Method


for Predicting Cap and Base Rock Heat Losses in Thermal
Reservoir Simulators, J. Can, ~e?. Tech., 19, No. 3
(1980)
87-90.

27. Enhanced 011 Recovery Field Report, 11, 2, Society of


Petroleum Engineers (1986).

lf.

28. Somerton, W. H., Keese, J.A., and Chu, S.L.: Thermal


Behavior of Unconsolidated Oil Sandst Sm. Pet. Eng. J.
(oct. 1974) 513-21.

Barry, R.: A General Thermal Model: paper SPE 11713,


presented at the California Regional Meeting in Ventura,
March 23-25, 1983.

29. Leung, W.K.: A Simple Gravity Override Predictive


Model, M.S. Thesis, The University of Texas, Austin
(1986)

12. Marx, J.W. and Langenheim, R.H.: Resewoir Heating by


Hot Fluid Injections Trans., AlME (1959) 216, 31215.

APpFNW
13, Willman, B.T., Vallerory, V.V, Runberg, G.W. Cornelius.
A.J., and Powers, L~W.: Laboratory Studies of Oil
Recovery by Steam Injection; J. Pet. Tech. (July 1961j
681-90.

The changes made to the Jones model permit direct input


of steam injection rate and pressure, and dimensionless
volume of displaced oil produced as:

14. Mandl, G. and Volek. C.W.: Heat and Mass Transport In


Steam-Drive Processes, Sot. Pet. Eng. J. (March 1969)
46, 59-79; Trans., AIME.

VO!3=

15. Myhlll, N.A. and Stegemeier,


G. A.: Steam-Drive
Correlation
and Prediction,
J, Pet. Tech. (Feb.
1978)173-182.

NPSoi
(1)
[- ~(~oi.sor)l
;

where Np is used insiead of Nd in the original Jones paper(l)


[Eq(A-25)] since VOD Is a function of the amount of displaced
oil which equals the total amount of mobile oil less the
cumulative oil production.

16. van Lookeren, J.: Calculation Methods for Linear and


Radial Steam Flow in Oil Resewolr; paper SPE 6788
presented at the 52th Technical Conference and
Exhibition, Denver, Colo. Oct. 9-12, 1977.
17. Neurmn, C.H,: A Gravity Override Model of Steamdrive,
J,. F . Tech. (Jan. 1985) 163-6%

Several modifications have been made to the Farouq Ali


model to improve its predictive capability.

18. Farouq All, S.M.: Graphical determination of 011Recovery


in a Five-Spot Steamflood paper SPE 2900, presented at
the Rocky Mountain Regional Meeting of SPE, Casper, WY.,
June 8-9, 1970.

I Tim
The critical time calculation recommended by Mand19 and
Volek(14) was used :

19. Doscher, T.M, and Gh&ssemi, F.: The Influence of Oil


Viscosity and Thickness on the Steam Drive; J. Pet. Tech.
(Feb. 19S3) 291-98.

t.=

[ -xqtcD

4 Kh MA
I

(2)

AWRAISAL OF Analytical

STEAMFLOODfvU3DELS

=
,

SPE 20023

Eq (1O) Is an approximation and Is used for Tavg~ Ts. if


Tavg > Ts, then Tavg is set equal to Ts for all future times.

where
et~D effc~

1
f~dhLvdh

(3)

CW(T~-TR)
When Vs > VsBT, there is the option of either producing
steam over a given Interval, or shutting off the production.
Farouq Ali(5) suggested a simplified treatment which gives
Vs(t) after breakthrough at any time t

with
Lvdh

(4)

94 (705 - TJO38

The thermal conductivity of cap rock as given by Somerton et


ai.(28)was used.
Kh = 1.04 + 1.3 @+ 0.2$KR(1 -S0)

Vs(t)=

(s)

(11)

Eq (11) is modified by perfoming an energy balance acoording


to Vogel(20). Assuming that the reservoir and !he adjacent
formation have the same thermal properties, the heat losses
upward and downward from the steam chest are

The average heat capacity of the steam zone provided by


Jones(l )was used.
-t

-2 KhA(43560)EA(Ts-TR)

(~-~) Mm] / Iv& (Ts-TavJ

where K R = 2.75 Btu/ft-hr-F at 120 F.

Ms = (1 -Ip)pRCR

VsBT+[6i(t-tBT)

$fsdh(l Sor)%cs+
QI = 4KhA(43560)(TS-TR)~~

$S*rpoC*

+ $-(1

Sor)PwCw

(12)

(6)

An overall energy balance as shown in Figure 2(a) gives


where pR = 165 ibm/ft3, CR = 0.20 Btu/lbm-F, co = 0,45

Qin - Qout = Qaccmulation or

Btu/lbm-F, and Cs = C w = 1.0 Btu/Ibm-F.


The heat capacity of cap or base rock was taken to be

OtcD erf~

~(13)

= V~MS(T~-TR)

in Eq (3) was obtained from the

Solving Eq (13) for ~s yields

hi~-4Kd4(43560) (Ts-Td

= 0.255K - 0.284K2+1 .421 K3

v.

M, (Ts-TR)

(8)

- 1.453K4 + 1.061 K5

~
(14)

Since

where

1..12
K

t~D =
.[
0.3276
Thus, ~ is obtained as follows:
(i)
(ii)
(ii i)
(iv )

(Ts-TR)

(7)

Mb= p#R
The term etcD erfc~
Jones(l) expression

Qit-4KhA(43560)

Calculate the value of etcD erf~


Solve Eq (8) for K value.
Determine tcD from Eq (9).
Obtain ~ from from Eq (2).

Qi tBT = VSBT &

(9)

(15)

(Ts-TR)

We can write

from Eq (3).
~ Qi
Substituting

(t-tBT)

+ VSBTMS(TS-TR)

(16)

Eq (16) into Eq (14) gives

Vs (t) = vsEw+lhi(t-tf3d-4 KtrA(43560)


The average temperature of the unswept formation was
calculated by subtracting the heat content of the steam zone
from the total heat injected and dividing by two times the buik
heat capacity of unswept formation:

Tavg =

Qit-V~(t)(TS-TR)MS

+ TR

(Ts-TR)

(lo)

H/=]

/ Ms

(Ts-TR)

(17)

From a time step At [t(n) to t(n+l )], the steam volume


from which the oil and water are displaced due to expansion
and displacement of fluids is

2t&[VB-Vs(t)]
where TR was not included In the original Farouq All model.
1
..-

-1

..
7

SPE 20023

H. L. Chen and N. D. Syfvester

In Figure 2(b), the solid line indicates the extent of


displacement by steam, The displaced volume Is the volume
between the dashed and solid lines. The material baiance for
the displacement element is given below.

~
The only modification made for the Miller-Leung model
is m the calculation of optimum steam injection rate whkh
was originally presented by Leung (29) as:

The oil displacement rate, ~o, is given by:

ist=

(19)

Q.= Av~$(S~)-SOr$t)
The water displacement rate, Qw, is given by:
C)w = AVS@[St)-(l

-Sst -S.rst)]
(20)

1+Sst -+Sor$t)

Qi
5.6146 p~[hfs+fsdhLvdh-& (TR-32)]

For oil:

Qj=4K~A(Ts-TR)@+

Assume that VB

[VB-VY)]I$[S$+)-

- @+)=

v;,

S$)]

-s!+)-

for wate~

then

Qw - qwAt = V:@[s$+ )- s!?]


= V:o[(l

(21

Sg)-(1-swsg)]
(22)

= v@@lw+)]
From Eqs (21) and (22) we have
*=
w

W&[sy+wq

(23)

Qw-v@o(n)--s$+l)l

From the fractional fiow eqution, we can write


(24)

fw=~=~

qo+q~

1+Kro~w
Krwpo

Let,
qo
Kro~w . . c
=
K~oVo
qw

(25)

(n+l)

Substituting Eq (25) into (23) and solving for So

Sy

(1+C)+

gives

Qo-CQw
t$v;

Sy+l) =

(28)

where the amount of heat injected, Q i is calculated


Vogel(20) as:

Then, the overail material balance on oil-water zone between


t(n) and t(n+f ) is as follows:

Qo - qoAt =

(27)

To account for the fact that the sleam injection rate should be
based on cold water fed to a steam generator, Eq (27) becomes
is t-

= AVS$(S$)-

Qi
5.6146 PWLvdhAt

(26)

1+C

-..

111

AhsMs(Ts-TR) (29 )

by

Wb

Me

~qp~g~

Summary of Steamflooding Models

Jones (1981)

Farouq Ali (1982)

Miller and Leung

rype of the Modal

FrontalAdvanoe

Modified Frontal Advance

Vertbal Advanoe Gravity Override

Cftaracteristios

1. Predkts ~,~,

1, Prediits ~,~,qw,So,~,

Ehs, and Fos.

and

AcDtVOD, Vpo areused. Data


suchas TR,hn,~i mayneed

1. Pradiote ~ and ist.


2. Adjustment of fW SW, %s. ~, ad

and TaW

2. Empirkal coefficientssuch as

(1985)

2. Requires defaulted values for

and EA values maybe nacaasaryfor

Sorst,Sor, %irl ad %t.

for reasonablehistory-rnafohiftg.
3. Tuning of field data for history=

3. Km, Km vs. &data needed

matching is not necessary.

to be adjusted to obtain good

when a ffefd 0sss other than

history matching for some

Kern Riier-A field is evaluated.


Tuning of hn may needed for

field cases.

reasonable history-matching.
Comparison d
Predictive Ability

Underpradots ~ at short times

Was notevaluated for field cases

and over-shootsthe measured

other than Kern River-A project.

especially for large ~atterrt area fieftf


cases (see Tabte 3).

values at bnger time for large


.,.

Setter pradktbn than Jones model

pattern area oases.


[see Figures 6(a) and 7(a)]

Sensitive

fcP EA, ~i, hi, S~, ~c

ist~ sdh$orst

isto Soit sdh

Parameters

lBbles
Data Used

Field

Kern River A

T~

TR

(:F)

(~)

kaI(TI )
[cp(F)j

for

WOI(T2)
[GP(F)]

~01
(CP)

History

Matching

f~dh
qoi
(BOPD)

API

Soi

ht

hn

(ft)

(ft)

(acres)

(BWPD)

{tt2/D)

(BTU/tt30F)

380

95

1380(100)

47(200)

1380

25

0.7

15

0.5

75

9rJ

2.5

0.345

225

0.96

35.0

300

100

1700(100)

10[230)

f700

15

0.7

13.5

0.51

125

80

2.7

0.31

300

1.097

3s.4

300

90

1000(100)

10(250)

1000

10

0.75

14.5

0.52

33

29

2.7

0.2s

300

1.097

38.4

400

SO

2710(85)

4(350)

2710

230

0.7

14

0.50

97

97

60.7

0.33

6000

0.870

35.7

400

113

27S0(1 13)

S{350)

2780

1S40

0.6

15

0.71

250

200

137

0.33

5s000

iChu.Trlmble(23)]

Kern-Canfield
iQreaaar-Shoro(24)]

.
i

Kern.SanJoaquin
iGraaaar-Shoro(24)]

Kern-1O Pattern
iO@aabyet

4J2S)]

Tla Juana
[da Haan6
m

Lookardq

-112

0.9s

35,0

20023

Comparisonof \hs History Match Roaulta for


Ulllmato Cumulative 011 Production

Fmb$aNp

%%
udU@@Qcs
124053
-6.50

m
132677
(7.5)

Kem.Canfield

(7.5)

Kern.San

28928(2.0)

42912

Joaquln

37507(3.0)

(3.0)

334346a
(6.0)

3131158
(6.0)

Kern-Ten
Pauorn

10414373

TiaJuana

----

(5.5)

1s.21

NP

Np

Lkl?lsl~
136531
(7. s)

2.t5

136198
(7.5)

2.65

40571

8.17

30943
(2.0)

3.39

6.84

3313459

-0.90

(3.0)
-9.35

3572606

(6.0)

(6.0)
----

10073073

6.39

110800s4
(5.5)

.3.28

(5.5)

1. The numberInaldatheparentfwals
irdkatestheuitimateoil pmdwtbnyearby the fielddataor predkfivemdal.
) x 1~
2. % difference- [(Np,model. NpMd)1fJP.f~~

Heat condidlon

to cap rock

1-T*
-----

Stec.rn ZO=

011zone

--

--

EASE
ROCK

(a) Heai Losses

Heat canductlon

ta base rack

(a) Frontal AdVCWICedDisplacement

Heat ccmductkrn to caD rack

V,a

,/

n --

R,H

Steam zone

,+

4
Haofnowto
undeftyfngzone
Condenwte

0:

W*71W

fkal

INITIAL (1)
%.$:

(1+1

so swat

$WI-seidor.t

SW*8W

& 011zone
(b)

Dlsplaoement

Meohanism

Figure 2. Control Volumo for Energy end Material Eralencee


(b) Vertical
Note: ~

or Gravity Overrtde

Dtsptacement

k the dkectlon of heat transfer

k the dkecflonof steam@owth

Ffguro1. lhe Mechonf$rnof St-m Displocemonf

(Modified Farouq Ali Model]

1-

I
Otn

(a) oil Production

S*V
Tk[IYCARM

1
!

.a

I
TIK

IWRIr

(a) Oil Production Rate vs. Time (At -0.125

Ratevs. Time (At = 0.1 year)

I .m

m.m
Tna

3, History Match of Kern River - A

FigUre 4. HietorY

Data

S.n

(b) Cumulative Oil Production vs. Time (At = 0.125

(b) Cumulative Oil Production ve. Time (At.= 0.1 year)


Figure

year)

year)

Matoh ot Kern - Cenfed

at

*E 20023

am

-s

/
4rn

,$

r :
Ira

rmlmalws

:Mr4 .
~om

iia.

f:euklo!mr-w~

F:a.m

W*O

,maam mm.

,4aKerlaa

d,

aw

1.40
TM!

*mm

1.40

.40

s.o

4.*

9.41

l.a

I.*

PA-

mllm+nMs -

4.U

Tin? m

(a) Oil Production

(a) 011Production Rate vs. Time (At -0.125

Rate vs. Time (At -1

Month)

year)

+
i

d .

Ii :

.~*-

Wm4-lo

:Xsl,m

i!
i *4

4QEa

.a

1.48

S.U

n.m
11=

4.n

5. History

Month)

year)
Figure

Figure

- .m

s.n

(b) Cumulative 011 Production vs. Time (At -1


(b) Cumulative 011 Production vs. Time (At -0.125

Mm

Matoh of Karn - San Joaquin

Data

6. History

Match

of Kern

. Ten

Pattern

Data

sPE

3
88

88

F!-

TIA _

Mw.utuwss
, J@ES-

SOW

-L

8-

m87

@
TIME (YSAM)

(a) Oil Production

Rate vs. Time

(At = 1 Month)

FIEIG 71A JWJU


. WSF1-mMs

, J-

MOOSL

TX= -

(b) Cumulative
Figure

h=88-

S417
-t

=~

20029

Oil Production vs. Time (At = 1 Month)


7. History

Match of Tia Juana Data

116

You might also like