SPE 69

jpt forum ,~g

A Method for Extrapolation of Cut vs Recovery Curves

Mroductim
For a fully developed waterflood with no major opera- semilog paper is quite linear and maybe expressedas
tional changes planned, a plot of fmctional water cut vs ~k =~ebsw,
total recovery is used often to obtain a quick estimate of , .,.,,..,. . . . . . . . . . . . . . . . . . (3)
kw
ultimate recovery at a given economic water cut. Ex-
trapolationof past performanceon the cut-cum*plot is where a and b are constants. On substitution. Eq. 2 may
a complicatedtask. The difficulty arises mainly because be written as
curve fitting by simple polynomial approximation does fw= 1 (4)
not result in satisfactoryanswers in most cases. l+ AebsW
This paperpresentsan equationthat is a fair representa-
where
tion of the waterflood process, based on the concepts of
fractional flow and the Buckley-Leverett recovery A=a~.
formula,* which may be written as
ER=m Xi-n, . . . . . . . . . . . . . . . . . . . . . . . . . . (1) As shown by Welge,* the water saturation at the pro-
ducing end (SW)may be expressed as
ER = over-all reservoir recovery, volume of
hydrocarbonrecovereddivided by oil in
place,
=Sav- *9
Sm ,.,,,.,,,.,..O . . . . ...(5)

()
X= In J--l
f.
fw= fractional water cut,
-~, where
sav = average watersaturation throughout the
system.
where m and n are constants. Since
Derivation of the Equation sav =ER(l -Swl)+swi,

The fractional flow equation (after neglecting the capil- then

lary pressureand gravity terms) may be written as
Sw=&(l-SWJ+SWl- 1~WW
-f , . . . . . . ...(6)
f.= 1 . . . . . . . . . . . . . . . . . . ...(2)
1+#.& In these equations, SW,is the initial water saturation.
& Pa From Eu. 4, the first derivative can be obtained and
The main portion of the ko/kw curve vs Sw plotted on writtena;
Welge, H. J.: Simp[ifled Memod for C%mput!ng Oil Rarnverbe by Gee or Water 014S-21 26/78/0002-S977S00 .25
Drhre; Trans., AlME (1952) 19s, M.s@, @ Sooletyof Petroleum Engineereof AlME

JPT Forum articles are limited to 1,500 words including 250 words for each table and figure, or a maximumof two
pagas in JPT. A Forum article may present preliminary results or conclusions of an investigation that tha author
wishes to publish bafore completing a full stud~ R may impart general technical information that does nnt war-
rent publication as a full.length paper. All Forum articles are subject to approval by an aditorial committea.
Letters to the editor are published under Dialogue, and may cover technical or nontechnical topics. SPE.AIME
reserves the right to edit letters for style and conterk

FEBRUARY, 1978 203

 Rmavmv. IUCW*

Fig. IPast performanceof the example waterflood, The

smooth line is the projection usingm and n from Fig, 2.

f. =
-A bebsw
(1 +Ads W)2
=- b~W(l-jJ.,, . . ...(7)
2Plot of recoveryvs X = -
Ftg.
W-HI
On substitution, Eq. 5 maybe expressed as data point on either side of the inflectionpoint separately,
Because the objective is always to project a waterflood
SW=E.7(1-SW, )+ SW,+L. . . . . . . . . . . (8) performance into the future, only water cuts higher than
b fw 0.5 should be used in the linear regression model.
From Eq. 4, the following relationship becomes tree. Fig. 1 shows the production history of a pattern flood
in a fault block, The application of this method to the
fu.=l./l +Ae* [
~R(*-swf)~swf + b.fwl . . . . (9) flood using a linear regression computer program re-
sulted in parameters m and n. These values then were
Solving for recoverygives used in Eq. 1to obtain the extrapolationsshown in Figs, 1
and 2.
ER= 1 ~n~~-1)-~]
b(lSw,) Advantages and Limitations
The proposedtechnique is based purely on actual perfor-
1 Swl +~lnA . .. ...... (lo) mance of a waterflood project. It implicitly considers
1Sw,( )
reservoir configurations, heterogeneity, and displace-
l%isequation now maybe written m ment effkiencyoOne major assumption is that the operat-
ing method will remain relatively unchanged. Variations
ER=mx+n,
in operational procedure, resulting in a shift of actual
where performance,can be integrated easily in updating runs.
1 As an akmative, it is common to plot WOR vs fmc-
= b(l SwJ tional recovery on semilog paper and extrapolate tie
results, This assumption, based on the derivations shown
X=ln
()~-l
fw

1
-1,
fw
in this paper, is equivalent to neglecting the term l/b O.fw
in Eq. 10. Thus, the method presented here is more
precise,
n = ( Swi+ ~ lnA .
Iswi\ ) One interesting application of the proposed method is
that from the linear plot of ER vs X, the two constants m
and n can be determined. These constants, in turn, may
Method of Application be used to derive a field kw/kO plot from Eo 3.
Given the past performance data, a table of cut vs frac- An sstimate of SW,is required to compute the constant a.
tional recovery can be constructed. From this table, a
~ = & e -b[w(l - Swt) + sw/l,
graph of fictional recovery vs X = In (l&w- 1) l~w
Au
would rtwlt in a straight line. The straight line may be
extrapolatedto any desired water cut to obtain the corres- b=
ponding recovery, m(l Swt)
In applying this equation, the recovery maybe used as
a fm.ctionof hydrocarbon in pla~e (E~), fraction of total Acknowledgment
pore volume (E~), or actual volume of oil produced We wish to thank L. L. Handy, who reviewedthe manu-
(NP), the last two values being proportional to ER, script and made helpful suggestions.
The fmctional flow curve has a point of inflection at
f.= 0.5. The proposedequationshould be appliedto tie Iraj Ershaghi, SPE.AIME
Osazuwa Omoregie
OrPhal rnmwript rewivad h $4@0~O! PetroleurI Snglmsr$ofliwAug. 12,1977,
Paper(SPE S&7) a~tedforpublimti &t. 7,1$77. Rewiti marwwpt raeelved U. of Southern California
Oec.e, 1s77. Los Angeles

204 JOURNAL OF PETROLEUMTECHNOLOGY