PETROIEUM BRANCH, AIME

Fidelity Union Building PAPER

NUMBER
707-G
Dallas, Texas
THIS IS A PREPRINT - - - SUBJECT TO CORRECTION

A Theoretical And Experimental Study

Of The Steady State Capillary End Effect
By

G. F. Hadley and L. L. Handy, Member AIME,

Cali~ornia Research Corp., La Habra, Cali~.

Publication Rights Reserved

This paper is to be presented at the 31st Annual Fall Meeting o~ the Petroleum Branch o~ the
American Institute o~ Mining, Metallurgi1!al, and Petroleum Engineers in IDs Angeles, October 14-17,
1956, and is considered the property o~ the Petroleum Branch. Permission to publish is hereby re-
stricted to an abstract o~ not more than 300 words, with no illustrations, unless the paper is
speci~ically released to the press by the Branch Publications Committee Chairman or the Executive
Secretary on his behal~. Such abstract should contain appropriate, conspicuous acknowledgments.
Publication elsewhere a~ter publication in Journal o~ Petroleum Technology is granted on request,
providing proper credit is given that publication and the original presentation o~ the paper.

Discussion o~ this paper is invited. Three.copies o~ any discussion should be sent to the
Petroleum Branch o~~ice; it will be presented at the above meeting with the paper and considered
~or publication in Journal o~ Petroleum Technology.

ABSTRACT There was satisfactory agreement between the the-

ory and experimental results.
Formulas are deri ved ~or calculating the
steady state saturation pro~ile and pressure pro- INTRODUCTION
~ile resulting ~rom the end e~~ect. A numerical
integration is required to obtain the results. The so-called capillary end e~fect or the
Two dimensionless numbers containing various holdup o~ the preferentially wetting phase at the
parameters o~ the system are introduced, which, outlet of a porous medium during the simultaneous
along with the viscosity ratio and the ~ractional flow of two or more ~luids has received con~i~er­
~low o~ one component, characterize the system. able attention from time to time in the oil indus~
It is suggested that the magnitude o~ the end e~­ try. In particular, there has been interest in
~ect can be estimated just knowing these numbers. how much influence the end effect has on the pro-
Sample calculations are carried out showing the duction from oil reservoirs and how much labora-
i~luence o~ the various parameters on the end tory measurements of relative permeability are de-
e~~ect • pendent on the end effect. Even though there has
been quite a bit of qualitative discussion of
A discussion is given to show that in the these matters, there has been very little quanti-
measurement o~ relative permeability curves by tative theoretical or laboratory work performed.
the dynamic method, the end e~~ect has a large The present investigation gives a fairly complete
in~luence on the pre~erentially nonwetting phase theoretical treatment of the capillary end effect
relative permeability curve i~ the pressure drop within the bounds of Darcy's law and the capillary
is measured in the wetting phase. I~ the pres- pressure relation. Experimental results on end
sure drop is measured in the nonwetting phase, effects are also presented for water-oil and gas-
the largest error appears in the wetting phase oil systems.
curve. I~ the pressures are measured in both pha
ses, the errors tend to be sel~-correcting ~or The end effect comes about at the efflux face
each relative permeability curve. of a core or around the wellbore in a reservoir
because of the discontinuity in capillary pressure
End e~~ects were measured experimentally on when the flowing fluids leave the porous medium
a Boise core ~or water-oil and gas-oil systems. and abruptly enter a region where there is no cap-
Three di~~erent rates were used in each series. illary pressure. This discontinuity in capillary
pressure tends to make it more difficult for the
Re~erences and illustrations at end o~ paper
2 A THEORETICAL AND EXPERIMENTAL STUDY OF THE STEADY STATE CAPILLARY END EFFECT 707-G

preferentially wetting phase than for the non- it is necessary to standardize the choice of the
wetting phase to leave the porous medium. Hence, characteristic pressure p in NRL. In doing so, it
there tends to be a buildup of the preferentially seems desirable to choose p to be a capillary
wetting phase at the efflux end of the system. pressure which is most characteristic of the sys-
In fact, it appears experimentally that right at tem. This pressure will be taken to be the pres-
the outlet, the nonwetting phase saturation drops sure where the slope of the capillary pressure
to essentially its residual value. The residual curve is a minimum. Such a pressure will be con-
saturation is defined here as the saturation at sidered the most characteristic for the system
which the permeability goes to zero for the non- since it corresponds closely to the capillary
wetting phase. The central problem in a theoret- pressure for the most probably pbre size of the
ical or experimental study of the end effect is pore size distribution. In a simple model like
the determination of the saturation as a fUnction the capillary tube model, the pressure where the
of the distance from the outlet end of the system slope is a minimum corresponds exactly to the cap-
[the fUnction that defines what is here called illary pressure in the most probable pore size.
the saturation profile]. Once this is known, all
other things of interest may be easily found. With the above definition, NRL becomes essen-
tially a measure of the ratio of the capillary
Only the steady state end effect is treated forces to the viscous forces in the system. As
here, i.e., the cases considered here are those NRL grows in size, the ratio of the capillary to
for which the saturations do not change with time viscous forces increases. Thus, a larger end ef-
This is not too restrictive since transient cases fect is to be expected for large values of NRL
can often be well approximated by the steady than for small values. Similarly, NREK is a meas-
state result. The equations derived hold for ure of the ratio of the gravitational forces to
either preferentially oil-wet or water-wet sys- the viscous forces in the system. Since these
tem, but not for those of intermediate vettabili- same two numbers appear in all studies of two
ty. However, when the sample calculations are phase incompressible, immiscible flow, it seems
made to study the influence of the various perti- desirable to give them names. NRL will be ca.l.led
nent parameters on the end effect, the calcula- the Rapoport-Leas number because of its similarity
tions are made for an oil-wet system. The reason to the scaling coefficient, introduced by Rapoport
for this is that all the experimental data are and Leas. 2 NREl( will be called the relative Eu-
for oil-wet systems, so that to facilitate com- gelberts-Klinkenberg number [relative since the
parisons, the sample calculations are also done density difference appears J because of its fq)pear-
for oil-wet systems. ance in the dimensional analysis work ot Engel-
berts and Klinkenberg. 3 It should be noted that
THEORY for given relative permeability and capillar,y
pressure curves, three numbers,_N~ ~HEK' and
For a one dimensional system, having two in- ~, contain: all the parameters in the system.
compressible, immiscible fluids flOWing, it is Uo
possible through the use of Darcy's law and the Hence, these three numbers fix the system.
capillary pressure relation 10derive the famil-
Equation 1 can be rearranged to read.
iar fractional flow formula. This formula can be
rearranged to read NRL
dX-- dB
.-G- ;/I.....,. ••• [7]
0S
~w' : c( + NmJ3 dX .•. [1]
For steady state conditions, ~>.A/ is a constant and
where is independent of X and S. To integrate Eq.
7, an initial condition is needed. Experimentally;
o([S] = Fw[l - NREK k ro ] • •• [2] the wetting phase saturation builds up at the ef-
flux end of the core. In fact, it appears to ap-
Ms] = Fw k ro--dpc proach the residual saturation of the nonvettlng.
ds • •• (3] phase right at the outlet face. Thus, to inte-
-/
Fw -- 1 +[kro liw]
krw Uo
grate the above equation, it will be assumed that
• •• [4 J the nonwettlng phase goes to its residual satura-
tion right at the efflux face. To say that the
Pc = Po - Pw · •• (5 J nonwetting phase saturation goes to its residual
value at the outlet face at first appears discon-
Pc is positive for a water-wet system and nega- certing, for then the permeability of that phase
tive for an oil-wet system. NRL and NREK are di- goes to zero'and it seems impossible to get any of
mensionless numbers defined as follows: the nonwetting phase out of the core. This does
not present a difficulty, however, because the
NRL = ~i NREK = k (pw - po] g sin e ... [6] saturation gradient and, hence, the pressure grad-
uolq Uoq .lent becomes very large. To see how this comes a-
bout mathematically, consider Darcy's law for the
P is taken to be some characteristic capillary nonwetting phase •.
pressure. To be able to compare various systems,
707-G G. F. RADIEY AND L. L. HANDY 3

which is just the usual equation for the static

capillary pressure curve.

This qw must be the flow rate at the end of the It is now of interest to have a means of
core also, where S~Swr. Now, as the water satu- finding the pressure as a function of position in
ration apprqaches its residual value) kw -7 0 and, the core. Darcy's law for the oil phase gives.
hence,

lim (_ ~ ~oj= 0 ••• [9 J

••• [16J
S~ Bwr.
where
since ~ approaches some value given by
ox
... [17J
_~=(Uoqo)
\ko Swr ••. [lOJ is the Engelberts-Klinkenberg number for the oil
phase. On introducing the saturation as the inde-
Then pendent variable
q _ lim (~
Uw dS
dP e ds)
dx .•• [llJ dpo = _lll - 3i w ] + kro N~) 0 X dS
NRL k ro d S ••• [18]

But from Eq. 1

Thus, as the water saturation approaches its res-
idual value and kw becomes smaller and smaller,
the saturation gradient gets larger and larger in
such a way that Eq. 11 is satisfied. .. .[19]

The saturation profile in the case of an oil so

=fS (( 1 - ) ..J+kro N~}t3
wet system is then simply given by
dS
s 13 [sJ dS
1 - X = Y[S] = - NRL
S Swr Jt.... -~[S] ••• [12]
.
S
wr
(.,(, - J) k ro
.....
••• [20]

Knowing X as a function of S, P can be found as a

Y[sJ = - NRL rs [S, Swr' u~, NREKJ function of X.
o ... [13J
Then
where(i1 represents the integral. This function
depends only on the quantities shown for given Pw = Po - Pc ... [2l]
relative permeability and capillary pressure
curves. For a water-wet case, Swr is replaced by When the saturation is not changing with X, Eq. 20
1 - Sore The problem of finding the saturation cannot be used; instead, the following is used:
distribution caused by the capillary end effect
has thus been reduced to a numerical integration. l:::, Po = _(( 1 - ;!t..,tt
k ro NEIdl:::, X ••• [22]
In the absence of gravity, Eq. 7 can be reduced NRL k ro
to that given in Reference 4. For a radial sys- SAMPLE CALCULATIONS
tem, Y in Eq. 12 is replaced by In r/ro.
EFFECT OF NRL
It is interesting to note the following:
Eq. 7 can be rearranged to read Using relative permeability curves A of Fig.
and the nOrmalized capillary pressure curve A of
Fig. 2, the curves shown in Fig. 3 were calculated
by use of Eq. 12. The curves in Fig. 3 show how
dx - - the end effect varies with NRL for a horizontal
L'.p gleo sin e core [NREK = 0] and with the fluids having a vis-
q - -
cosity ratio of 10. The fractional flow to give
.•• [14] an inlet saturation in the absence of end effect
of .5 was Fw = .518. The relative permeability
and for q = 0, it follows that curves were taken only as representative of an oil
wet system and do not represent those measured
dP c = [p w - p 0] g sin e dx •.• [15 J
4 A THEORETICAL AND EX.PERIKENTAL STUDY OF THE STEADY STATE CAPILLARY END EFFECT 707-G

experimentally on any particular system. The clearly shows that the actual outlet saturation
curve obtained after dividing the capillary pres- chosen is not very critical.
sure by the characteristic capillary pressure
[~= .1 psi in this case] is called the normal- EFFEC'l' OF RELATIVE PERMEABILITY FUNCTION CHANGES
ized capillary pressure curve. The capillary
pressure curve is also just a typical one and To investigate the change brought about by
does not "represent measurements on any particular modifying the shape of the relative permeability
system. functions, curves B of Fig. 1 were used with the
capillary pressure curve A of Fig. 2. In Fig. 7,
Examination of Fig. 3 shows that as NRL in- are plotted the saturation profiles obtained from
creases, the end effect increases. Since the to- relative permeability curves A and B, for the same
tal flow rate appears in the denominator of NRLJ value of NRL' It is seen that the change in shape
it can be said that as the flow rate increases, had little influence on the end effect. It seems
the end effect decreases. The curVes also pre- reasonable to say that in general a small percent-
dict that if the capillary pressure curve could age change in the relative permeability curves
be changed in such a way that p is increased will give a small percentage change in the end ef-
while the normalized capillary pressure curve is fect.
unchanged and also without changing the relative
permeability curves, that the end effect would be PRESSURE PROFIlES
increased. In general, however, it is not possi-
ble to change p and k independently; the shapes Equations 20 through 22 were used to obtain
of the relative permeability and normalized cap- the pressure profiles shown in Fig. 8. These were
illary pressure curves will also change in the calculated for NRL = 5, uw/Uo = 10 and for rela-
usual case. tive permeability curves A of Fig. 1 and the cap-
illary pressure curve A of Fig. 2. It will be
EFFECT OF VISCOSITY noted that the capillary end effect causes a small
er pressure drop in the wetting phase and a larger
Figures 4 and 5 illustrate the dependence of pressure drop in the nonwetting phase than would
the end effect on the viscosity of each of the be obtained if there were no end effect. At the
phases. It is important to note that the viscos- outlet face of the core, the pressure difference
ity dependence cannot be Characterized simply by between the two phases is the capillary pressure
the viscosity ratiO, it is necessary to specify at the residual saturation of the nonwetting phase.
the viscosity of both phases. Figure 4 illus- This pressure is called the entry pressure and wiD
trates the result of varying the viscosity ratio not be zero. The slope of the wetting phase pres-
and keeping the oil viscosity constant, while sure curve at the outlet face of the core is sim-
Fig. 5 illustrates the effect of varying the vis- ply obtained from
cosity ratio and holding the water viscosity con- o
stant. It is seen that the end effect decreases [1 - }....,] +- NEK k ro
as the viscosity of either component is increased NRL k ro .•• [23]
A change in the nonwetting phase appears to pro-
duce a IIDlch larger shift than a change in the
The pressure gradient in the water phase [as was
wetting phase. In both Figs. 4 and' 5, the frac-
discussed in the theory] is infinite. This makes
tional flow of the components was changed with
it very difficult to accurately determine the en-
the viscosity ratio so that in the absence of an
try pressure. The increase in over-all pressure
end effect the curves would have been the same.
drop in the nonwetting phase, caused by the capil-
Flow rates are the same for all curves.
lary end effect, [which is necessary to get the
same amount of nonwetting phase out of the system
EFFECT OF CHANGE IN OUTLET SATURATION
as would be obtained if there were no end effect]
can be appreciable in a small core where the over-
It will be recalled from the theoretical
all pressure gradients are quite small, but should
discussion that in the calculation of the end ef-
be negligible in an actual reservoir where the in-
fect, the outlet saturation of the nonwetting
crease in pressure would be a very small fraction
phase is held fixed at its residual value. For
of the total pressure drop.
this reason, it is of interest to know how criti-
cal this saturation is in determining the shape
ESTIMATION OF THE END EFFECT MAGNITUDE
of the end effect curve. Because it is difficult
to determine accurately the residual saturation,
Inasmuch as the end effect does not change
it would be unfortunate if the shape of the end
radically with a small change in the relative per-
effect curve were sensitive to the saturation cho
meability curves, it seems possible to estimate thE
sen. Figure 6 shows that this is not tru:::. When
the outlet saturation is changed from .88 to .80, size of the end effect just by knowing NRL , uW/Uo'
and the fractional flow. This would imply that the
the curves are so close together that they cannot
capillary pressure curve can be characterized by a
be distinguished on the graph. Even when the sat
uration is taken as low as .70, there is no large single number, the characteristic pressure. This
deviation. The true residual is about .88. This is not rigorously true, of course, but may serve as
10 7-G G. F. HADIEY AND L. L. HANDY - 5

a f'air approximation. Table I gives the values saturation than it should"be.

of' NRL f'or dif'ferent values of uw, and different
Uo Consider now the case where it is desired to
saturations that would exist in the absence of determine both the krw and k ro functions". First,
end ef'fect so that for NRL values less than or suppose that experimentally the pressure drop in
equal to the values shown, the end effect will be the preferentially wetting phase is the only pres
restricted to the last one-tenth of' the core. By sure obtained. The individual flow rates and the
confining the end ef'f'ect to the last one-tenth of' average saturation in the core are also recorded.
the core is meant that for X = .9, the saturation k ro ' for example, is then found from
does not dif'fer by more than .01 from the value
it would have in the absence of end ef'f'ect. 'lheBE
values were calculated using relative permeabil- ••• [26]
ity curves A of' Fig. land the capillary pressure
curve A of' Fig. 2. The capillary pressure curve This is plotted against the average saturation.
used has a f'airly large slope and, hence; the end It would be P9ssible to get the correct result by
ef'f'ects computed will be somewhat larger than calculating k ro from Eq. 26 using the pressure
will be sometimes observed in practice. Hence, gradient at the inlet instead of 6P/L and the sat
the NRL values are conservative, and perhaps a uration at the inlet. Now the inlet saturation oj
little smaller than necessary. Since X i9 direct the wetting phase is amaller than the average sat
ly proportional to NRL , the NRL values to conf'ine uration of the wetting phase and the 6P /L for the
the end ef'f'ect to the last one-hundredth of the wetting phase is less than the pressure gradient
core can be _f'ound by multiplying the values in of the wetting phase at the inlet face of the corE
the table by 0.1, etc. [see Fig. 8]. Thus, k ro , computed from Eq. 26, ie
greater than k ro at the inlet and the average
INFLUENCE OF END EFFECT ON RELATIVE PERMEABILITY wetting phase saturation is greater than that at
MEASUREMENTS the inlet. Now, if k ro refers to the preferen-
tially wetting phase, it must increase as the wet-
It is now of interest to discuss just how ting phase saturation increases. It then follows
the capillary end effect influences the results that there is a self-correcting mechanism "for the
of relative permeability measurements by the dy- computation of the wetting phase relative perme-
namic method. Consider f'irst the case where it ability, Le., the changes brought about by the
is only desired to measure a krw/k ro curve and end effect tend to cancel out.
not the individual k) s • Experimentally, the
f'ractional f'low ~~ and the average saturation in Consider now the nonwetting phase curve.
the core are determined. krw/k ro is, then, cal- 6P/L is still used in computing the relative per-
culated f'rom me ab ili ty and is also amaller than the actual in-
let pressure gradient for the nonwetting phase.
Again the computed relative permeability is great-
er than the correct relative permeability at the
•.. [24J
inlet. However, the nonwetting phase relati ve__
permeability curve decreases as the wetting phase
and this is taken to be the krw/kro at the aver-
saturation increases. Thus, on the nonwetting
age saturation in the core. Because of the pres-
phase relative permeability curve, the end effect
ence of capillary terms, the correct equation for
brings about changes in the saturation and pres-
calculating krw/k ro is
sure which work in the same direction and tend to
shift the curve strongly to the right [see Fig. 9J
l_dS
dPc dSJ
dx To sum up the discussion, the end effect intro-

[ dPw
••. [25J duces relatively little error in the measurement
of the wetting phase relative permeability curve,
dx but introduces a relatively large error in the
nonwetting phase curve when the wetting phase
This will reduce to [24J only at a point where pressure drop is the only pressure measured ex-
there is no saturation gradient. If the rate is perimentally.
low enough, there may be no point in the core
where there is no saturation gradient, i.e., the Without going through the details, it is eas-
end effect may extend all the way through the ily seen that if only the nonwetting phase pres-
core. In general, however, rates are high enough sute drop is measured, then the end effect intro-
so that the saturation gradient at the inlet face duces the larger error into the wetting phase rel-
of the core is negligible. Under such conditions, ative permeability and a much smaller error in the
Eq. 25 reduces to Eq. 24 at the inlet face of the nonwetting phase curve. Furthermore, if pressure
core. Thus, when Eq. 24 is used and the results drops are measured in both phases and if they are
are plotted against the average saturation of the used to calculate the corresponding permeabili-
wetting phase rather than the inlet saturation, ties, then there is a self-correction tendency on
each k rw 7k ro is plotted at a higher wetting phase both curves.
6 A THEORm'ICAL AND EXPERIMENTAL' STUDY OF THE STEADY STATE CAPILT.ARY END w.li'~'Ii: 707-G

COMPARISON WITH EXPERIMENT Ji,.= 0.99979, and NRL - 0.473; [2] rate = .189
per sec., ~~= 1.000, and NRL = 0.0328; and [3]
Calculated end e~fects were compared with
those observed for gas displacing oil and water
rate = 4.65 cc per sec., ~~ = 0.999992, and NRL =
.00133.
displacing oil ~rom Boise sandstone cores. Al-
though the equations were derived for steady The observed oil saturations at the differ-
state conditions, comparisons were made for end ent ~lood rates are compared with the computed
effects observed near the completion qf gas and saturation distributions in Fig. 12. For similar
water ~loods when the fractions of displacing fractions of gas in the produced fluids, but at
fluid in the produced ~luids were greater than different flood rates, substantial differences in
0.99 ~or water and oil and 0.999 for gas and oil. oil saturations were both observed and computed.
At that time, saturations were changing slowly. The end ef~ect extended completely through the
Steady state conditions were approximated closely core at the two lower ~lood rates, but not at the
enough for the steady state equations to apply. highest rate. For the highest rate, the oil sat-
uration profile was calculated by the Buckley-
The properties of the Boise cores are given Leverett method and the computed steady state end
in Table II. The static capillary pressure curve ef~ect added, at the downstream end. With this
was measured ~or water displacing oil and for gas modification, the agreement between the observed
displacing oil ~rom Boise cores previously treat- and calculated oil saturation distribution for
ed with a silicone dri-fi1ming material, General this run was satis~actory.
Electric SC-87, to make the rock sur~ace prefer-
entially oil wet. The capillary pressure data A comparison is shown, also, with4the data o~
are shown in Fig. 2. Relative water and oil and Richardson, Kerver, Hafford, and Osoba in Fig. 13
relative gas and oil per.meabilities were measured The saturation pro~ile was calculated using rela-
by a variation o~ the Welge method. The data are ti ve permeability functions A from Fig. 1 and the
plotted in Fig. 10. capillary pressure curve A from Fig. 2. The abso-
lute permeability and characteristic pressure were
Oil saturation pro~iles vere measured by a taken from the data o~ Richardson et al. The re-
radioactive tracer method. Di-p-tolyl selenide, sults show that the end e~fect can be calculated
prePared from selenium 75, was added to the oil with fair accuracy without knowing the relative
to give an activity o~ about two millicuries per permeability functions or-even the normalized cap-
liter of solution. Radioactivity at various posi illary pressure curve ~or the particular core sam-
tions along the core was measured with a scintil- ple.
lation counter, which was shielded to collimate
the gamma rays. The slit width was one centimen- SUMMARY
ter. The relation betveen observed counting rate
and oil saturation was checked against a material Formulas have been derived ~or calculating
balance ~or the entire core. the steady state saturation pro~ile and pressure
pro~ile resulting ~rom the end e~~ect. Experimen.-
End e~fects in the water-oil system were tal results on water-oil and gas-oil systems give
measured in a 5-fi Boise sandstone core which had satis~actory agreement with the theory. Two use-
been made preferentially oil wet With dri-filming ~ dimensionless numbers are introduced. It is
solution. The core was completely saturated with indicated that a good estimate of the magnitude of
a 1.7 cp oil and water flooded. Three flood rates the end ef~ect can be obtained knowing the frac-
were used. The lowest rate was -0.00197 cc/sec ~Ol tional ~low, the Viscosity ratiO, and the two di-
which NRL was 1.45. The ~raction o~ water in the mensionless numbers. For measurement o~ relative
produced fluids at the time saturation profiles permeabilities by the dynamic method, the theory
were measured was 0.9917. The intermediate rate predicts that the greater error will be introduced
was 0.0161 cc/sec. NRL was 0.168 and the ~ractioD into the curve for the phase other than the one in
of water in the produced liquids was 0.998. The which the pressure was measured.
highest rate was 0.082 cc/sec for which NRL was
0.035 and the ~raction of water was 0.990. The NOMENCLATURE
measured oil saturations are compared with the
computed saturation distributions for the down- f Porosity
stream end o~ the core only in Fig. 11. The a- g Gravitational acceleration
greement is satis~actory. k Absolute permeability
k ro Relative oil permeability
In the gas-oil system, observed and computed krw Relative water permeability
saturation distributions were compared ~or three L Length o~ system
gas ~lood rates. The measurements were in a l-fi p Characteristic capillary pressure
Boise sandstone core with air displacing 1.7 cp Pc Capillary pressure
oil. The gas flow rates, fraction o~ gas in pro- Pc Dimensionless cpaillary pressure
duced fluids, and NRL were as follows for the Po Oil pressure
three runs: [1] rate = 1.31 x 10- 2 cc per sec., Po Dimensionless oil pressure
707-G G. F. BADlEY AND L. L. HANDY 7
Pw Water pressure Pw Water density
Pw Dimensionless water pressure e Angle of elevation of core measured
r Radius from horizontal
ro Wellbore radius
q Total flow rate per unit cross section- REFERENCES
al area
% Oil flow rate per unit cross sectional 1. Engelberts, W. F., and Klinkenberg, L. J.:
area "Laboratory Experiments on the Displacement
qw Water flow rate per unit cross section- of Oil by Water from Packs of Granular Ma-
al area terial," Proc third World Pet Congress,
qg Gas flow rate per unit cross sectional [1951] II, 544.
area 2. Pirson, S. J.; Elements of Oil Reservoir
Qo Oil flow rate iliineering, McGraw-Hill Book Co., New York
Qw Water flow rate 1950] 281-284.
Qg Gas flow rate 3. Rapoport, L.. A., and Leas, W. J.: "Properties
S Water saturation of Linear Waterfloods," Trans. AIME, [1953],
Swr Residual water saturation 198, 139.. --
t Time 4. Richardson, J. G., Kerver, J. K., Haf'ford,
x Distance along core J. A., and Osoba, J. S.; "Laboratory Determi-
X Dimensionaless distance = x/L nation of Relative Penneabili ty," Trans. AIME,
Uo Oil viscosity [1952] 195, 187. --
Uw Water· viscosity
Po Oil density

Table I

NRL VALUES FOR END EFFECT TO BE CONTAINED IN LAST ONE-TENTH OF THE CORE
Wetting Phase Saturation in Absence of End Effect

.26 .52 .66

Uw = 1. .01?O .260 .?O5
Uo

Uw = .1 .00167 .0281 .113

Uo

Uw = .01 .000190 .00284 .0132

Uo
TABLE II

PROPERTIES OF BOISE COffES

Core 1 Core 2

PorOSity 0.253 0.253

Air Permeability 888 md 915 md

'!

Length 28.3 em 86.5 em

Cross Sectional Area 23.3 en?- 14.7 cn?-

Pore Volume 167 ce 322 ce
 1.0...,,...--......,---....,...----;----,----.

'"
10

--'+"-f
I i

0::
1.0
I
I
I
~

---V;
I
I8
II.. !I ---- I 8~ .8
0

-
NRL=500
::IE I Q.
U I Z
~
I ,
0
ILl 6 --- 6~
.«
j:
e( .6
II:: r-
::;)
(/)
(/)
ILl
...J
:::!
Q.
a:
::;)
....e(
NRL= 50
NR~ V.J
II:: « (/) NRLa.5
Q. U
4 4 _..
>-
II::
«
0
ILl
N
:::!
0 '"
..J
...J
:::i
«
0:: ::IE
.2 I - - - - - ! - - - ----j--~ «2
U
A = ASSUMED CURVE
20::
0
z
.2 -+-
F",= .518 -'!...e.-
NRL'fLOLq
B = GAS-OIL CURVE
C = WATER -OIL CURVE NREK =0 }if a 10
0 0
0 .2 .4 .2 4 .6 .8 1.0
OIL SATURATION - PER CENT PER SPACE
OIL SATU~ATION

fiGURE I
FIGURE 3
RELATIVE PERMEABILITY fUNCTIONS.
FIGURE 2 VARIATION OF END EFFECT WITH HRl.
CAPILLARY PRESSURE CURVES

1.0

1.0 1.0

---;J-
B
.8 /. .8

)
--
~w= .1

z
o
fLo

~
/; z
0 &.= 10
~/
z
0
j: .6
«
V
~.6 I j:
e( .6
fLo a: ..0
a: a: ....«:::>
::;)
.... ..!.9- ::;)
....e( .1 I
(/)
e( I !
-----;-: - - _ . - -+-.--------;;'---- _ .. -
I/) (/)
I ...J
4
NRL = 5 (So) OUTLET: 0 • . 88
-1--- I (5
...J .4 ...J 1-----
(5 I
i
fLo CONSTANT
0 '" I
NREK' 0
~ = 10
Ii = .84
0 = .80
I
I
}l-w CON S TA NT
2 ,---
}l-o
- -----T·-·---r • = .70

.2
+ . __ .- ---_ .. _- .2
I
- _. ... -

o I I
I I
i
2 4 6 .8 1.0
o .6 1.0
o .2 .4 .6 B 1.0
:x:
.2 8
'" :r

FIGURE ~
VAR I AT ION OF END EFFECT .ITHb; fL 0 CONSTANT. FIGURE 5 FIGURE 6
fLO
VARIATION OF E"DEHECT WITH~ I'w CONSTANT. I NF LUENCE Of OUTlET SATURAT I ON ON END EFFECT
 .6 r-----~------~----~------~------
1.0
J.!'tI .. 10
JLo

.5 r---~~+--------+------~--------+-----~
.8

z
0
~ .6
--.----- .. ~ - -------- ~ w
~ 4 r-------+-------~~----~--------+-----~

~~
CIl
< CIl
a: w
;::)
a:
.... Q..

< CIl
CIl CIl .3 r-------+--------+-------4--~~--+-----~
.4 w
...J PERMEABILITY CURVES A ...J
A- RELATIVE Z
0
B-RELATIVE PERMEABILITY CURVES B o
CIl
z
NRL = 5 ~ .2 r-----~~~-----+------~--------+---~~
.2 - o
I:!:.w = 10
11. 0

o
.1 ,
.2 4 .6 .8 1.0 Po ' ....... ,,
.......
"- .......
o ~--__~------~----~~----~--~~
.2 .4 .6 .8 1.0

fiGURE 7
CHANGE IN END EFFECT WITH CHANGE IN RELATIVE
PERMEABILITY FUNCTIONS.
FIGURE S
PRESSURE DISTRIBUTION CURVES.
 100r-------r------,-------,------~-------

.----EXPERIMENTAL WHEN END

EFFECT IS IMPORTANT

,, 80~T-----+-------~----~----~L+-------~

>-
>-
I- f I-
...J
...J
III
'I III
~60J-~~--t-------r-------~~~--~-----J
~
UJ
'I :E
NONWETTING / / WETTING 0::
:E PHASE '/ PHASE UJ
0::
a..
UJ
a.. 'I
~40~--~-TT-------r---~-,~------~-----J
~
I-
~
...J
UJ
0::
20j-----~~~--~-7L---+-------L------J
- - - WATER- OIL
GAS- OIL

O~O------~----~~--~~~-----L----~
60 80 100
NON WETTING PHASE SATURATION
o 1.0 PER CENT PORE SPACE
WETTING PHASE SATURATION

FIGURE 10
FIGURE 9
RELATIVE PERMEABILITIES FOR BOISE CORES.
SHIFTING OF RELATIVE PERMEABIL1TY CURVES WHEN END EFFECT
IS IMPORTANT.
 100 I I 100
NRL = 1.45 - - - e - - -
NRL = .168 ---b,----
NRL = .035~
I- 80
z
lJ.J
U
-r--!
,

./
I I
a::
lJ.J
a.. 60 I .// /i a::
~ 60
o 0

I 1.--./ , I
I I e_ .... /
Z
o I ..+-- / Z
o
I-
el
a:: -- ~ __ --- 0
sl I-
el

--1---
I ,
! .7 ~ 401----.-lI-+-~~~-:---
--,
~ 40

----- ~---
0
.-7:;-- ~
el
CJ)

...J
-- ...J
I e
o (5
20 - 201-------""""==---,---~~ ,-~,--,
o NRL = .473
I b, NRL = .0328
e NRL =
o
.75 .80
I
.85 .90 .95 1.0
I
OL-____
o
!
~------~----~------~----~
.6
:x:

FIGURE II FIGURE 12
CO~PARISON Of OBSERVED AND CALCULATED SATURATION CONPARISON OF OBSERVED AND CALCULATED SATURATION DISTRIBUTION
DISTRIBUTION FOR WATER-Oil SYSTE~. FOR GAS-OIL SYSTEN.

1.0
I e

.)
.8

z
• ~
--;-- / . ~

o
~ .6
el :
a:: i
::l
I-
el
CJ)

...J.4
o

2 1- Qg=.264 CC/SEC.

I IQo=2.2 X 10~3 CC/SEC.

0 i
.2 .4 .6 .8 1.0
:x:.

FIGURE 13
CO~PARISON WITH EXPERI~ENTAl DATA OF OSOBA.