Você está na página 1de 13

,

8fE 6548

The Prediction of oil Recove:y by Gravity


Drainage
Herman Dykstra, SPE-AIME,consultant

Introduction
The gravity-drainage mechanism is one of the most efficient ways of producing an oil field. Unfortunately, most
bil fields cannot be produced economically under freefall gravity alone because the effective oil permeability is
too low, the oil viscosity is too high, or the dip of the
formation is too small. Nevertheless, many reservoirs
take advantage of the gravity component, especially
under pressure maintenance in areas where gas is available, such as overseas, or where inert gas can be used.
The most ideal system would be one where oil is produced only under free-fall gravity drainage, giving the
highest possible recovery at the most efficient rate, However, because a producing mechanism must be justifkxl
economically as opposed to another producing mechanism, an engineer must calculate gravity-drainage recovery as a function of time.
An approximate theory of strict gravity drainage has
been presented by Cardwell and Parsons,l They ignored
capillary pressure terms and show how gravity-drainage
equations can be solved. They give only one comparison
of recovery calculated from theory with recovery obtained experimentally from a draining column.
In addition to that study, many other papem have been
published describing how to calculate reservoir performance where gravity effects influence the production
mechanism. An early investigation describing the potentially high recovery with gravity drainage is Katz
discussion of Oklahoma City Wilcox sand, Katz also
0149-2136/78KX)05-SS4SW.25
~ 1978 !hcwyof
Pefmleum

Eng.mee!s of AIME

presented data on the low (6 to 8 percent) residual oil


saturations that could be obtained from a completely
drained portion of the Wilcox sand. Lewis3 enlarged
upon the gravity-drainage mechanism for oil recovery
and gave four examples of gravity-drainage fields. Be
ginning in 1949, more studies4-21discussed methods of
calculating the effect of gravity component on oil recovery during depletion or pressure maintenance, and case
histories of fields where gravity was an important factor
in production. All the papers describing a calculating
procedure used the Buckley-Leverett method.= These
smdies also used historical rates of production or a preselected rate of production to get the advance of gas-oil
front and to calculate recovery, More recently, highspeed computers were used to solve mathematical equations of various reservoir simulators. These reservoir
simulators still required that rates or drawdown be preselected as a boundary condition to the solutions of
mathematical equations, Thus, these simulators could
not be used to determine the free-fall, gravity-drainage
rate of production.
One study attempted !O determir.e production rate
based on the location of the gas-oil front below a stated
level. Essley etal. 15determined an approximate produc-
ing rate by arbitrarily assuming a maximum capacity ftii
each well remaining on production below a given subsea
level and determining the total production rate from that.
They did not state how the arbitraxy assumption was
made, nor how this rate was related to a true gravitydrainage rate. Also, they did not use the Buckley1

An approximate theory offree-fall gravity-drainage recovery is expanded to ticcount for


residual oil saturation. The recovery equation is derived and seven comparisons are made of
observed gravity-drainage recovery with calculated recovery. Examples show how the theory
can be applied.
I
818

I
JOURNALOF PETROLEUMTECHNOLOGY

Leverett equation. The authors derived and used a simplified equation to describe the advance of a saturation
front.
A study by Terwilliger et al.zs presented laboratory
data on recovery vs rate of prod~ction and saturation
distributions as a function of both rate and time for an
unconsolidated sand column. Experimental data were
compared with calculated data obtained using the
Buckley-Leverett method. All experiments were conducted at constant rates and none were made under freefall gravity drainage. Their results showed the high recoveries that could be obtained where gravity was an
important part of production. Recoveries of about 70
percent were obtained at low rates fmm a 13-ft column
and calculated recoveries of about 80 percent wew obtained from a 1,000-ft column. The difference resulted
mainly from the difference in percent of capillary rise
between the two columns,
Only Cardwell and Parsonsl presented a theory for
determining oil recoveries from a porous medium under
strict, or free-fall, gravity drainage; that is, the produ~
tion rate was determined solely by gravity effects, Surprisingly, no application of this method has been described in the literature since then. In this study, I show
seven comparisons of recovery calculated from Cardwell
and Parsons theory with recovery determined cxperimentally. The equations were made more general to
account for an immobile gas saturation at the start of
gravity drainage and for relative permeability of oil decreasing to zero at residual oil saturation rather than at
zero oil saturation, The recovery equation not given in
Cardwell and Parsons paper also is derived here.
Three examples show how this method can be used to
calculate gravity-drainage recovery and rates for an oil
field. The experimental data is not satisfactory, so I also
have suggested where additional research is needed to
substantiate the theory and its application further.

Theory
Downward Movement of the Gas-Oil Interface
(Demarcator)
The differential equation derived by Cardwell and Parsons (their Eq. 9) is
G_-pgdk& .
at
I.@ dS aZ

... . . . . . . . . . . . . . ...(1)

In deriving this equation, the authors dropped capillary


pressure terms from their Eq. 8. .heydid so with the
provision that the joint effects of capillary retention and
low permeability will be taken into account by an appropriate treatment of the terms involving the saturationpermeability relationship. In their example calculation
(results are shown in their Fig. 4), they gave a recovery
curve based on a permeability-saturation relation in
which the permeability decreased to zero at zero liquid
saturation. Cardwell and Parsons showed what the limiting recoveries would be with or without a capillary retention of lo-percent liquid. Their figure implied that the
recovery curve would be unaffected by allowing the
relative permeabdity curve to decrease to zero at 10percent liquid saturation, rather than to zero at Opercent liquid saturation. As I show later, this is not true
at late drainage times.
One way to allow the relative liquid permeabdity to
MAY,1978

decrease to zero at a residual liquid saturation, so that Eq,


1can be solved as described by Cardwell and Pamons, is
to define a new saturation variable and to modify the
subsidiary equations to account for the new definition,
This definition for two-phase flow (such as water and gas
or oil and gas) is
s = s, - S,r
. . . . . . . . . . . . . . . . . . . ...0(2)
1.0-s/,

where SI is the dmining liquid saturation, a function of


time and position, and Slr is the residual draining liquid
saturation at which the relative permeability decreases to
zero. If oil is the draining liquid in the presence of
connate water, oil saturation is based on hydrocarbon
pore volume.
Determining :esidual liquid saturation at which the
effective Iiqui! (and hence relathe) permeability goes to
zero from flow tests alone is difficult if not impossible.
One may argue that effective permeability will not go
truly to zero until liquid saturation goes to zero. Thus, all
one has to do is extrapolate the measured portion of the
permeability curve to zero at zero liquid saturation. That
is a moot point, however, as capillary-pressure data always indicate a definite percent-liquid ~tention at high
capillary pressures. This suggests that capillary-pressure
tests be conducted to establish a zero end point for the
relative liquid permeability curve. A long-time drainage
test using a long enough tube filled with sand could be
used to establish the end point, but this procedure is not
practical. If data are required for an oil field of interest,
centrifuge capillary-pressure tests could determine residual oil saturation with air displacing oil in the presence
of connate-water saturation.
In arriving at an equation for the relative liquid permeability, Cardwell and Parsons used Wyckoff and
Botsets datax on unconsolidated sand and applied the
equation to a gravity-drainage experiment by Stahl
et al.2$ A review of the data indicates a minimum
residual-liquidsaturation of about 8 percent in their draining column, Because both the Wyckoff and Botset
relative-permeability tests and the Stahl et al, gravitydrainage tests were conducted on unconsolidated sands,
the Wyckoff and Botset data could be replotted vs the
new saturation variable given by
~= s. -Swr,= Sw -0.03
. . . . . . . . . . . . . (3)
1- Stor
0.92

With the data plotted against S as defined by Eq. 3, the


exponent B in the equation k, = SE becomes B = 3.2,
where kr is the relative permeability. Here, the exponent
is decreazed to 3.2 from the 3.5 obtained by Cardwell
and Parsons, where they ignored the residual liquid
saturation.
Two other factors should be mentioned before proceeding with the derivation of the gravity-drainage and
recovery equations. First, many fields are produced for a
time before a plan of how best to produce the field is
selected. During this time, pressure may be drawn down
to cause gas saturation. As long as gas saturation remains
below the critical value, there is no problem and one can
derive the equations using an initial oil saturation ofSOf.If
initial gas saturation becomes so high that gaa becomes
mobiie and an attempt is made to adjust the production
rate to a gravity-drainage rate, the possibility exi8ts that
two g=-oil fronts will occur as described by Martin.26
819

Field and well performance would have to be watched


closely to determine whether two gas-oil fmnta & being
formed one moving up and one moving down.
Second, a change in oil-formation volume can occur as
the gas-oil fkontmoveedown the structure intoareservoir
with undersaturated oil, The analysis is based on operating at constant pressure; that is, gas is injected at the top
of the structure to offset the void that occurs when oil is
produced, Both the initial oil-formation volume factor,
130j,and oil-formation volume factor, Bo, in the drained
region, are used in the recovery equation.
Several equations in the Cardwell and Parsons
analysis were modifkd to include various factors discussed. Permeability in the Cardwell and Pamons equations was redefined as #
kJS), where k is the actd
permeability, and k, is the relative permeability, a tlmction of S, the saturation variable defiied in Eq, 2. The
velocity of flow equation in the upper region (less than
100-percentliquid saturation) with the capillary-pressure
term omitted becomes
.........................

(5)

For the lower region (Sgi<Sgc), the velocity of flow


beeomes
v* =

k +kr(SJpg ~ - H
L-Z~
P
(

. , . . . . . ...(6)
)

Here,
s,=

If
- 1 . .,, .,..,., ............. (7)
1- S,r

The equation of continuity becomes


&&~ _#
:
, . . . . . . . . . . . . . . . . . . . . . . . (8)
az
where for a two-phase system,

Cadwell and Parsons.


For the genertd case, their Eq. 16 gives Z as a function
ofS, whereZ is the distance traveled byS during drainage
time, t, when the starting point is a uniform satumtion of
St throughout. This equation can be rewritten as
z = +!&jIs)t,

where the relativqxrrneabdity function has been separated out and where

f(s) = *

. . . . . . . . . . . . . . . . . . . . . . . . . . . (14)

If ~-i is the inverse of the function, f, the saturation is


given by
s =f -

()+!!4J_

. . . . . . . . . . . . . . . . . . . . . .. (16)

+=+

Differentiation of Eq. 4 gives


l&=~sB-,,
,-

. . . . . . . . . . . . . . . . . . . . . . . . .

(I3

BSB-=+@#

., . . . . . . . . . . . . . . . . . . . ..(18)

Solving forS,
s

. ()

Z&j

- . . . . . . . . . . . . . . . . . . . (19)

Soi - SOr
\

T
I

as
=.
-Pgk$$-!7, . . . . . . . . . . . . . . .. (10)
at
J@
which is the same as Eq. 1, except that S is now the
saturation variable defined by Eq. 2,
The demarcator equation becomes

I
I
I

(&=
_~!k :_ytuf
. . . . . . . . . . . . . . . . . . . . (1s)
dt
f$(s, -s=

where S, is defined by Eq. 7 and Sd is the saturation


variable defined by the liquid saturation an infinitesimal
distance above the demarcator, Eqs. S and 6 may be
substituted into Eq. 11 to give the equation for the advance of the demarcator or gas-liquid interface.

O:(J

(17)

which on substitution in Eq, 16gives

and where saturations are based on pore volume.


Eqs. 5 and 8 can be combined to give

(12)
where kr~ is the relative liquid permeability at Sti. The
solution to this equation follows along lines describw by

. . . . . . . . . . . . . . . . . . . . (15)

Because of Eqs. 13and 14, we can write

0 =4(1 - s/r)* . . . . . . . . . . . . . . . . . . . . . . . (9a)


and for a three-phase system,
@=$( l-sOr-s Pars
) . . . . . . . . . . . . . . . . . . (9b)

...............................

. . . . . . . . . . . . . . . . . . . ..(13)

L-~d

SOi

c.=-

I .0
SATURATION-SO

F@.lSaturation distribution after tikne,t.


JOURNALOF PETROLEUMTECHNOLOGY

m=-&&.

. . . . . . . . . . . . . . . . . . . . . ..(20)

Eqs. 4,19, and20maybesubstituted intoEq, 12togive

...............................

(21)

Recovery Equation
The recovery equation can be derived as foliows. At a
dminage time, t, the demarcator has reached a ievel Zfj.
given by the integration of Eq. 21. Above this level, the
saturation distribution is given by Eq. 19. A schematic
diagram of the distribution is shown in Fig. 1. In this
diagram, SOcan be considered a iiquid saturation in a
two-phase, iiquid-gas system or an oii saturation based
on hydrocarbon pore volume in a &&phase, wateroil-gas system. The recovery, R, is given by

R=l

= z~
-

Z*
I&

L&

B.

J~

So dZ. . . . . . . . ..(22)

But, from Eqs. 2, 19, and 20, we get

(~

- +Sor. . . . . . . . . . . . (23)
t)
Eq. 23 can be substituted in Eq. 22 end the resulting
equation integrated to give
so=

(1-s.,)

.1
I

01

. mZd -
t

,, . . . . . . . . . . . . . . . . . . (24)

Zd vs 1 is obtained from integration of@. 21. Comesponding vaiues of Zd and t then are used in Eq. 24 to get
recovery es a function of time.
Eq. 21 and of 24canbe solved using anelectronicdesk
calculator, but calculations are tedious. Fortunate!y,
todays computer programs are problem oriented and
require only a modest amount of programming ability.n * Because of the many calculations required to
check the experimental datq described in the next section,
I used IBMs CSMP system (System/360 Continuous
System Modeling Rogram, 360A-CX- 16X, Users
Manual H29-0367-2) to solve Eqs. 21 and 24. This requirA oniy one-half page of programming steps and cost
a few dollars per run.

Comparison With Experimental Data


In addition to the test mentioned earlier, Stahl et al.m
conducted two more tests with crude oii, three with
MAY. 1978

water, and one with heptane using an 8-ft tube. King%


performed drainage tsts on five different sands in 8-ft
tubes using water as the saturating fluid. These sands
were designated as Sands 20,40,60,80, end 100. Kings
date on Sends 20 through 80 is discussed first.
Kingconducted tests for 2!4 years and then disassembled the tubes and determined the weight penxnt of water
in each 3-in. sand section. Thus, data were avaiiabie to
establish height-saturation distribution, equilibrium
height, H, and residuai water s@uration,$&r. A plot of
the data htdk.wed evaporation loss from the top of the
column and Possibly fmm the bottom, though this was
not nearly so obvious. With the possible exception of
Sand 80, minimum @urations reached an aimost verticai
iine before evaporation became a problem. From this
verticai line and the memred saturation distribution, the
amount of evaporation v as calculated. Material-baiance
calculations then were performed to get the amount initiaiiy in place. This vaiue divided into the tabulated
amount recovered gave the percent of recovery. The last
point was based on the amount remaining from the
height-saturation data plus the evaporation loss, not on
the amount recovered because of the evaporation problem. Materiai-baience data indicate that the columns
may not have been iOO-percentliquid saturated. This
makes the data questior,able; however, this problem
was not serious,
Unfortunately, King did not give the permeability of
the send in each tube experiment, but he did give sand
grain size. At fret, i attempted to determine permeability
horn measurements on unconsolidated sand packs. Data
ftom Slichter,~ Wilsey?O Krumbein and Monk? and
Stahl etal.2Swere used to plot permeability vs grain size.
Using permeabilities obtained from this plot did not give
good agreement between calculated recovery and experimental data, Consequently, I abandoned this approach and instead systematically varied both permeability, k, and ExponentB untii a reasonably good match was
obtained be!weencalculated and experimental recovery.
Comparisons (Fig. 2) are good to about SOto 100hours.
Beyond that time, the observed recovery fell below that
calculated, undoubtedly caused in part by evaporation
loss. At the last point, agreement improved and was
almost exact for Sand 20. Values of k and B used for the
calculated curves are shown on the graph. Values of B

70 ,

w
0
SW
>
0
u
:40

SAND
d!L

1-

LA&L

u
g
% 20
*

10

%1

10

[0

100

80

VA S.2

.12

CALCULA7E0
1000

IWOO

40,W0

DRAIIAAQE
llMEHOIJR5
F@.2Comparisonof c~culated recoverywith expwimentet
data of King.
821

were considerably higher than the 3,2 obtained fmm the


Wyckoff and Botset data and increased with finer, less
permeablesrmd. The calculations also showed that calculated recovery at early times was very sensitive to k, but
was insensitive to B and, at late times, was sensitive to B
but only slightly sensitive to k. Thus, early data served to
define a value fork. Because of the evaporation problem, late data were not as satisfactory for determining a
value of B.
This method woukl establish a wdue ofk at the existing
initial saturation when testing began. Thus, if initial
liquid saturation were only 90 percent, the k determined
from gravity-drainage experiments would be the effective permeability at !?O-percentliquid saturation, and
relative permeability would be based on this value as the
startitlg point. Because the relative permeability equation
is a straight line on log-log paper, a shift in the base of the
curve does not change the slope nor the value of B. Any
error caused by a lack of 100-percent liquid saturation
when calculations assume a 100-percent starting saturation probably could be prevented by changing porosity on
the assumption that an immobile initial gas saturation
could be handled as a part of the rock and effectwe
porosity thereby lowered. 1 did not pursue this course
because of the evaporation problem and because no independent measurement of starting liquid saturation was
available for the King data.
Stahl etal. ran experiments on Wilcox crude oil at 100,
115, and 130F, on water at 64, 100, and 130F, and on
heptane, Unfortunately, cumulative recovery data* were
givemfor the Wilcox crude-oil tests only; thus, these will
be the only tests discussed. Rate data were given for
water and height-saturation curves for four ts%ts.
The height-saturation curves help to establish residual
liquid saturation and equilibrium height. The longest test
lasted less than 3 days; therefore, complete capillary
equilibrium was not established based on Kings work.
Minimum saturation from the test on the lowest viscosity
fluid, heptane, would be closest to residual saturation.
This value appeared to be about 8. percent. The 130F
crude-oil test served to establish that H is 15 cm, inasmuch es calculations indicate that Zd is within 0.S cm of
its equilibrium value after 50 hours for all three temperatures. With B = 3.2 and k = 7.5 darcies, I calculated

recoveriesfor Wilcox crude oil at the three temperatures.


Comparisons between calculated and experimental remwezyam shown in Figs. 3,4, and S. The comparison is
good for the 100Fcmde oil. There is greater deviation at
early times for the 115Fcrude oil. The calculated curve
is 3 to 7 percent below the observed data for the 130QF
crude oil. If permeability is increased to 10 darcies, the
agreement is good throughout the entire range as shown
by the dashed curve. A possible explanation might be that
while taking samples for oil content analysis, permeability was altered. In any event, in view of the good agreement obtained for the 100and 115FresuRs, it seems hard
to beileve that resuks for 130Fwould be so much higher
than the calculated value unless something happened that
affected permeability. Thus, assigning a value to k of 10
darcies for the 13(IFtest does not appear unreasonable.
Earlier, I mentioned that Cardwell and P? sons Fig, 4
implied that the recovery curve wouid be the same regardless of whetherSOrwas used in the relative-permeability
equations. The recovery curve shown in their Fig, 5,
which did not include a residutd oil saturation, showed
good agreement with observed data, whereas the calculated curve shown in Fig. 5 was below the observed data,
The difference between the two calculated curves results
from includingSOrin the k, equation and from usingH =
15 cm, rather than the 10 cm used by Cardwcll and
Parsons. Therefore, to check the effect on recovery of
using O rather than 8 percent SOr,I also calculated the
recovery for SO, = Oand B = 3.5 for the 100Ftest. As
shown by the dashed curve in Fig. 3, the recovery after 10
so -

70
,

60

a
w
g
w

50-

~
t:
8
w
a

______

*O

SM6M
CRUOE

20 /

10

o 0.2

Ttw ordinate IS mislabeled m FQ 4 of Stahl etel. and giwn as a fa2a. The Iagend and
curves Mdlcata cumulatwe

_. __,

1.0

data.

5
PRAINAOE

TIME-

OATA
OIL At IISF

OaSt RVEO
CALCULATED

10
*O
HOURS

50

100

Fig. 4CalcuIatad and experimentaldata,


so

-. - L_ .:. . . ..

70
w

J...

50,

I
---

--.

....

-----

u
w

1
:~so

-.{.

-.*

oo
09

10

v
w
=

4
-:

----

0:

40/

,
.,:

s
OR41NAOE

TIME-

--

:50

--

a~en

10
20
I!Ouns

3Calcdatad andexperhentat data.

so

//

QATA

-~--l:~~f~~~:
L---f
1

~.

822

4.
1

-.

g
20

k
w
/

40 .

--

So .

0
s

E
>W

20 -

/
,/;

-..

10

100

%2

10

Fig.

SM6H
OAt A
CRUOE OIL AT ISO* P

5
ORAINAOC

10
TIME-HOURS

0B5ERVE0
CALCULATE9
~ALc LA7co

20

b~?.6*
k,,04

;0

I&

5Cdcdated and axperimentd data.


JOURNAL OF PETROtEUM TECHNOI..00Y

hours drainage time is greater than that for SO, = 8


percent, but below 10 hours it remains the same. Thus,
using a value for Sapdw decrease the rewvery curve,
but only at later times. Hence, its value should be included in any gravity-drainage calculations.
Additional gravity-drainage data were published by
Templeton et al. ?2 who presented a figure showing saturation distributions at eight different time periods for
gravity drainage from a Mt column. Although the authors stated that the tolume of brine drained was meaSIMed as a function of time, they did not present the
results. Thus, no check could be made between calculated and observed reowery. However, these data could
be used to compare measured with calculated saturation
distribution.
Marx33 used a centrifuge to establish a drainagerecovery curve for small core samples. He did not give
pertinent fluid and rock claw, therefore, no calculations
could be made to check with observed recovery. The
Marx data is discussed later,

Drainage Modulus
Examination of Figs, 3, 4, and 5, where recovery is
plotted vs log time, shows that the calculated curves
parallel each other and that a shift along the time scale
will cause the curves to overlay. This suggesfs the possibility of combining rock and fluid properties so that one
recovery curve resulfs.
Consider a rectangular, dipping formation (Fig. 6) in
which oil is dmining down the structure into a well.
According to Elkins et al. ,s the rate at which oil drains
down the structure undf:rfree-fall gravity is given by
~0= 0.001127 kJdVC@Op.) sincx
. . . . . . . . (2s)
/4$%
The constriction coefficient, C, was not in the original
equation but has been included here to represent a constriction inflow because the formation is open only to the
wellbore and r,ot to the face normal to the direction of
flow.
The quantity of oil in place at the start of drainage is
given by

N=+.!!#&.

............,.,,,.,. .. (26)

The rate of recovery in percent per day is given by


d?= looqo , . . . . . . . . . . . . . . . . . . . . .,, ,.. (27)
dt
N
Eqs. 25 and 26 may be substituted into Eq. 27 to give
:=0.274

MAY. 1978

with fbwcbnvergingintotha wellbore,

$~~

,,.,,,.,,...,,,

(28)

where do is now the density of the oil in grams per cubic


centimeter, on the assumption that gas density is small
relative to oil density and can be ignored. In fact,
Muskat~ points out that if the gas phase is immobile
and not continuous, there will be no buoyancy reaction on
the oil due togas and the Ay should be replaced by the oil
density, YO.This suggests that an engineer use oil density for systems of low pressure or where the oil leg
remains in one hydrocarbon phase and that the density
difference be used for systems with a continuous gas
phase throughout.
The term in parentheses in Eq. 28 has been called the
drainage modulus and is characteristic of a system that
includes both rock and fluid properties. Dimensions of
the dminage modulus are millidarcy-grams per cubic
centimeter per centipoisefeet.
If the calculated recovery for Wilcox crude-oil tests is
plotted against the product of the drainage modulus and
time in days, one curve results (Fig. 7). Also plotted on
Fig. 7 are the actual reported recovery data (using k = 10
darcies for the 130Ftest rather than 7.5 darcies). There is
good agreement between calculated and observed nxovery over the entire range.
The curve in Fig. 7 can bethought of as a dimensionless
curve. It can be applied to any system where the equilibrium height, H, is about 6 percent of the length of the
draining column and where relative permeability data can
be approximated closely by the simple equation, k, = S8.
As mentioned earlier, recoveries at early times are insensitive to B. Calculations have shown that the effect of /3
appears after about 25 percent recovery. Also, the effect
of residual oil saturation does not appear until after about
40 percent recovery is obtained (Fig, 3). Thus, Fig, 7 can
be used with confidence as a starting point for gravitydrainage calculations on any reservoir of interest. Then,
if calculations show that the gravity-drainage production
mechanism is feasible, additional laboratory data can be

90;

FIg.6-Dwhgfonnatbn

Zw

400

400

Fig. 7-Recovefy

0R4tM4GE
lbOOULVSI
TIME
Wo
moo
Iwo
Iwo
1s00
t

1900 2000

2200
I

WOO

as a function of drainage modufuaand time.


823

obtained m determine values for B and SOPand another


grwity-tige
cakxdafion made with the new data.
In addition to Fig. 7, I have given two morecurvesto
amist the engineer in cakulating the productionrate.
Because the cuxve in Fig. 7 can be @ought of as a
dimensionless curve, it can be dficmttiated to obtain a
dimensionless rate curve. Two such rate curves are
given. Fig. 8 is a linear plot of pczccnt fCCWCXYw day
per unit drainage modulus vs the product of the drainage
modulus and time for time in days, while Fig. 9 is a
log-log plot. These two curves and the curve in Fig. 7
provide data to calculate both rate and cumulative recovery as a function of time.
Fig. 9 shows the nature of the decliie curve for strict
gravity-drainage production. For the product of drainage
modulus and time (Dhfxt) greater than 200, the dcclinc
curve is a straight line on a log-log plot. This curve can be
compared with a similar curve developed by Matthews
and Lcfkovits3s for a well located in fhe center of a
cylinder and producing oil so!sly by fre+fall gravity
drainage. Their drainage mechanism was for radial flow
into -a well and did not include downdip flow of oil.
I mentioned earlier that Marx33did not give the rock
and fluid data required to calculate gravity-drainage recovery. However, if onc plots the Marx recovery data vs
log time ORthe same scale as used in Fig. 2 and slides the
graph along the time scale, one obtains almost an exact
overlay with the calculated recovery curve for Sand 20,
except for late times where the Marxcurve rises above the
Sand 20 curve, This indicates that the draiiage modulus
for the Marx systcm could bc determined fbm the drain, agc modulus of Sand 20. Thus, the Marx data indirectly
substantiate the theory.

Constriction Coe~dent
Asshown in Fig. 6, the flow out of the dipping, rectangular formation is restricted at the bottom face to flow only
into the wellbore. Thu8, the f-fall,
gravity-drainage
rate would be less thau if the entire face had been open to
flow, as it was in the above cxperimenfs.
Defmc tic constriction Coetilcicnt as
C=qc/go, . . . . . . . . ..0 . . . . . . . . . .. (29)
where go, the flow through the unconstricted face, is
given by
q,=

.*

. ..................

(30)

and qc, the flow through the constricted face, is given by


~c = 2AKk&
. . . . . . . . . . . . . . . . .. (31)
-mYp
AXIAYwas determined by RodSonand Cardwell%to be
Ax = K(klkJ
(32)
~
K(k1k2)
Substituting for q. and qc gives
c. Ax.2L_.
. . . . . . . . . . . . . .. (33)
AY
W
kz is obtained from the relation of
K(k,) .W
. . . . . . . . . . . . . . . . . . . . . (34)
K(kz)
~

and kl from the relation of

*=!I#.

. . . . . . . . . . . . . . . . . . . . . (35)

a
w
>
0
v
w

K
IL

0
w
1-

a
K

F@.6-Rateof
824

reocveryvcdrainage moddueti-t.

DRAINAGE MOOULUSmTIME
Ffg. O-Rate of reeovery w drainage modulustlmest.
JOURNALOF PETROLEUMTECHNOLOGY

K, K, and F are elliptic integrals with argument kl and


modulusk2. For all problems of interest, rJW<< 1.Oand
k1k2<c 1.0. As shown by Dodson and Cardwell, we can
write
F(k2,ki)=ki,
,,.,,,..,,.,,.,
. . . . . . . .. (36)
,.. (37)
K(klkz) = Ir12, . . . . . . . . . . , .0,...,..
and
K(k#J=ln(4/kik2).

. . . . . . . . . . . . . . . ..t (38)

On substitution, Eq, 33 becomes

c=

~L

~ h (4/klk*)

~
. . . . . . . . . . . .. (39)

This equation was solved over a range of values and the


results are shown in Fig. 10. Value rw = 3-in. goes with
the W values shown on the lower portion of the three
curves, whereasrW= d-in, goes with the W values shown
on the upper portion of the curves. The constriction
coefficient has a relatively narrow range from 0,3 to 0.4
for uniform well spacing,

Application to a Reservoir
Threeexamples show how this method may be applied.
The ft example is of a field that was produced under
controlled production, then produced competitively, and
was again under controlled production, The field was
discovered by a well completed in the gas cap. Gas was
produced until 2 years later, when the oil zone was
discovemi, after which gas-cap production essentially
was stopped and energy conserved by producing oil only
from the oil band. Later, during a period of competitive
operation, gas-cap gas was produced, after which gas
production was again controlled. By this time pressure
had decreased more than 1,000 psi below the original
bubble-point pressure, with the result that excess gas was
produced fmm wells completed far below the original
gas-oil contact.

Though somewhat academic, a comparison between


actual and calculated gravity-drainage recovery is interesting. Management and engineem should be aware of
the potential for high recovery tlom fields that can be
produced by the gravity-drainage mechanism.
Table 1shows the data for this field. An explanation is
needed of how to determine value L used to calculate the
drainage moduhw. The field was developed on a reasonably uniform spacing of 16 acres per well. Assume that
each well with its associated 16acres acts as a rectangular, gravity-drainage block and that L can be calculated
fmm the area being drained by a well. For 16 acres, the
spacing is 835 ft, which divided by cosine of the dip angle
gives L = 964 ft, The drainage modulus then can be
calculated as shown in Table 1.
The recovery calculation is shown in Table 2. The
dimensionless rate shown in Col. 4 is obtained from Fig.
8 or9and convemd to barrels per day as shown in Cd. 5.
The percentrecovery isobtainedfrom Fig. 7. Cumulative
recovery can be obtained either fmm the area under the
rate curve or from percent recovery.
The calculated cumulative and actual recovery tlom
the field are shown in Fig. 11. After 24 years, actual
mcpvery was 19.7 percent and the field rate decreased
below 900 B/D, Calculated gravity-drainage recovery
aftei 24 years was 48.7 percent, at which time the calculated rate was 1,910 B/D. However, the extra recovery
by gravity drainage would have required full pressure
tiaintenanos at bubble-point pressure.
The second example is the Lakeview POOLdescribed
by Sims and l%ailing.7This pool is a good example of
strictly he-fall, fyavity-draiiage production because the
gravity head was essentially the only energy left after
extended blowout of the discovery well. Data are shown
in Table 3. Data not given by Sims and Frailing were
determined or estimated from published information.
Fluid properties were based on conditions at the midpoint
of the oil band. Reservoir temperature was estimated
fmm T = 60 + 0.02 (2,7S0) = 115F, Average pressure
was bssed on depth below gas-oil contact and oil density
assuming atmospheric pressure at the gas-oil contact.
Average gas in solution and formation-volume factor
were based on Standings correlations.37Oil viscosity
was based on Bea136and on Chew and ConnalIy?9 RelaTASLE 1-RESERVOIR AND FLUID PROPERTIES,FiELD A

OH
VkOsity,
Cp

Bubble-pointpressure, peia

3,350
2.3

Bubbfe-pokrtfonnatlon volumefactor
Oil density at bubbbpoint pressure,gmkc

O!liff

Polostly
Connate water eaturatkm
Inftiaioif saturation
Average airpermeabHity 18 wails cored), md
EffeothJoon permeabltity! rem PI data, md

0.229
0.23
0.71
227
5s

Avers@ Up angle,degrees
Numberof wells
,
Awageweil spacing, acre/well
A&fagedstenos betwesn Wetls,ft
of drslnhg column, ft
Wd?diameter,In.
constriction Coemdent

DrainaQernodutus=
CON$TRICTIOWco~FPlclWT, c
*.
MAY, 1978

lo4xfwkti0n000md
0m.

30
48
&6
0.:

k~. C sins rnd+mloo

B1.s.fid

epft

=0.6314 ~rt Indays= 11.5fortinyears


82S

TABLE 2-GRAWTY DRAINAQERECOVERY, FIELD A

Time
(years)
DMxt
.
.
(1)
(2)

Recovew
(pwcenfi
(3)

Rate
(percent/D/
unit
dralnaae
modd&)
(4)

0.268
0,257
0.265
0.253
0.251
0.248
0.245
0.240
0.235
0.229
0.221
0.201
0.184
0.127
0.100
0084
0.073
0.054
0.057
0.051
0.046
0.042
0.038
0.036
0.0325
0.030

11.5
23
34.5
46
57.5
69
80.5
92
103.6
115
138
161
184
207
230
253
276
299
322
345
366
391
414
437
460

3.0
:
6.0
3
6.6
4
11.7
14.6
17.4
:
7
20.2
8
23.0
9
25.7
10
26.2
12
33.1
14
37.2
16
40.4
43.1
1!
45,3
22
47.2
24
48.7
26
49,9
28
51.2
30
52.2
32
53.3
34
54.2
36
55.0
38
55.7
40
56.3
Col. 3 is from Fig. 7.
Coi. 4 is from Figs, tr and 9.

Cumulative
Rate Production
(BOPD) (MM bbl)
(6)
(6)
7,700
7,670
7,610
7,650
7,490
7,400
7,310
7,160
7,010
6,630
6,590
6,000
4,890
3,700
2,960
2,510
2,180
1,910
1,700
1,520
1,370
1,250
1,130
1,040
970
890

2.9
5.6
6.4
11.1
13.9
16.5
19.2
21.8
24.4
26.6
31.4
35.3
38.4
40.9
43.0
44.8
46.3
47.4
48.6
49.6
50.6
51.5
52.3
52.9
53.5

Coi. 5 is the rate in B/D = Col. 4 x minagle~~u]us


x

95. lcp.

tive oil permeability at connate water saturation was


obtained from an average curve based on laboratory data.
Relative oil permeability in the presence of gas and connate water saturation was determined from Gates and
Tempelaar-Lietz*study on California cores.a The effective oil permeability was obtained from the product of
these two values and air permeability. Initial oil saturation was based on a material-balance calculation and
assumed a 10-percent residual oil saturation in the
5,000-acre-ft secondary gas-cap sand.
Sims and Frailings Fig, 7 was used to determine value
L. The distance between wells, except for a cluster Lf

TIME
Ftg.

826

IN

YEARS

1lCumulative recoveryw time.

TASLE 3-RESERVOIR AND FLUID DATA (LAKEVIEW


POOLj MIDWAY SUNSET HELD)
Average oil gravity, API
22.5
Reaarvoirtemperature at midpon~F
115
Pressureat midpoint, psi
175
Estimatedgas In solution at this pressure
Oil kxrnation volume factor
1.;:
Average oil viscosity, cp
Average oil density, gin/cc
o.&9
Porosity
0,331
tXnnate water saturatbn
0.236
Catcuiatsd011band saturation
0.6S9
Initial gas saturation
0.106
Estimatedresidualoil saturatbn
0.10
Air permeability,md
3,655
CakUM8d effective 011permeability,md
1,100
Oil in placebelow-1,620 ft subsea, MM bbi
60.5
Total surface area, acres
Oil band surface area, acres
::
Average dip angle, degrees
Middepth, ff
2,7%
Gas.oli contact at start of drainage, ft subsea
-1,620
Average distance between welis, ft
450
Average length of draining column, ft
493
Well diameter, in,
Constrictioncoefficient
0,3;
Drainagemodulus = k.d. C sirra md.Qm/cc
Cpft
PLS14
= 00697 tin days
= 25.4t in years

wells in the southeast portion of the field, ranges from


400 to 500 ft; a good average number would be 450 ft.
Annual production data for the Lakeview Pool for the
yews since Sims and Frailing published their study are
available from annua] repo~ of the Conservation Committee of California Oil producers. These reports were
used to provide the data in Table 4. For convenience, Jan.
1, 1936, was chosen as the starting point of the gravitydrainage calculations because rapid development of the
pool began near the end of 1935.The percent of recovery
and the average rate for each year also is shown in Table
4. A plot of cumulative recovery vs time is shown in Fig.
12. The curve is derived fmm Fig. 7 and is based on a
drainage-modulus value of 0.0697, calculated at the bottom of Table 3. This calculated curve is somewhat above
the observed data. The closeness of the calculated and
observed recovery is good, considering that some required data had to be estimated. However, if a recovery
curve is calculated on the basis of 13-percent hydrocarbon PV residual oil saturation ( 10 percent PV), the correspondence between observed and calculated recovery
becomes remarkably close. This calculated curve is
shown as a dashed curve in Fig. 12.
Production rate is plotted at midyear in Fig. 13. This
graph illustrates the fluctuations in rate caused by drilling
infill wells or extension wells, which occurred in 1940-41
and again in the 1950s. Additional wells increase the
value of the drainage modulus because of a decrease in L
or, if the wells are drilled further downstructure, the
ultimate level of the gas-oil interface will be lowered.
The third example is the Mile Six Pool, described by
Welge6 and Anders.10 Anders gives a fairly complete
history of the pool. Performan& indicates it was produced by gravity drainage. Pool pressure did not vary
much, averaging about 800 psi during the 25-year life
that Anders reported. Thus, reservoir fluid properties
JOURNAL OF PETROLEUMTECHNOLOGY

TABLE4-Cumulative

PRODUCTIONAND RATE
(LAKEVIEW POOL MIDWAY 8UNSET FIELD)

TABLE 6-RESERVOIR AND FLUID DATA, MILE SIX POOL


(FROM REF. 10)

Recocowy
End
of
Year
1035
1938
1037
1938
1039
1940
1941
1942
1943
10
G
;%3
1947
1948
1949
1960
1951
1952
1063
1954
1855
1968
1957
1968
1959
1960
1961
1882
1583
1964
1886
1986
1967
1868
1969
1970
1971
1972
1973
1974
1975

Time
(years)

Dec. 31,1035
(MM bbl)

2.577
8.718
13.008
16.422
17.883
18.308
21.931
24.114
25,496
28.492
27.451
28,280
29.122
29S60
30.419
30.876
31.301
31.788
32.168
32.664
32.898
33,227
33.673
34.262
34.866
35.463
36.938
36.382
38.745
37.085
37.403
37.681
37.923
38.114
36.273
36.397
38.480
38.681
38.631
38.707

;
4
;
:
9
;:
12
13
14
15
16
17
18
;:
21
::
24
25
26
27
28
29
30
31
32
33
34
::
37
::
40

Average Rate
in Year
(B/D)

Recovery
(percent)
o

7;69
16,826
11,763
6,614
6,186
4,463
7,186
5,Q7$I
3,784
2,733
2,628
2,270
2,309
1,s83
1,656
1,260
1,169
1,279
1,145
1,038
914
002
1,223
1,613
1,623
1,636
1,329
1,216

1:::
21.5
2S.6
29,2
31.9
36.3
39.9
42.2
43.6
46.4
48.8
48,2
49.4
60.3
61.1
61,8
52,5
63.2
53.9
54.4
56.0
55.7
58.7
67.6
68.6
69.4
80,2
60.8
61.3
61,9
62.3
62.7
63.0
63.3
63.6
63.7
63.8
63.0
64.0

%
873
761
662
623
438
339
255
195
102
207

Oil gravtty,API
Average pressure, pal
Gas irradutbn at average pressure,cu fthbl
oilviscosity,Cp
Oii density, gmkc
Formatbn volume factor at average pressure

8$
?%!
0,719
1.18

Porosity
Connate water saturatbn, PV
Inftld 011saturation, PV
Residualoil saturatbn, hydrocarbonPV
Oil permeability,md
Oil In place at start of drainage, MM bbl

0.226
0.31
0,69
0.20
5%?

Area of OHband, acres


SpacinQ,am/well
Horizontaldistance between wails, ft
Dip angle, degrees

Constrktiorr coeftident
Oilwatercontact, ft subsea
Original gas-oil contact, ft subsea
Length abng dip, ft

350
55;
17.5
0.35
-2,440
-1,876
1,679

md-gmkc
Drainage modulus = kodoC Sk
pLSo, IJ
Cpft
300 0 19 035 x sin17, ~
= 1.05Xx 6:2 X 0.69X 0.228

coa17
= 0.263 fort in days

at 800 psi would be satisfactory for gravity-drainage


calculations.
Reservoir rock and fluid properties obtained from
Refs. 8 and 10 are shown in Table 5, Exponent B of the
relative-permeability equation was determined from loglog plots of thedatashown in Fig. 4of Ref. 8 and in Fig. 5
of Ref. 10. In the latter figure, the curve seems to be
drawn to zero k, at 80-percent gas saturation, giving a
residual oil saturation of 20 percent, As the sattnz%ns
are based on hydrocarbon pme volume, the saturation
variable is defined as S = (SO- 0.2)/(1 - 0.2), The
log-log plot of the data showed some curvature but a
1

I.m
~

20/300-

glno

10$)00-

~w-m
** :
00

LA KEVIEW

~ 30
a
w

E 20

POOL

OBSERVED RECOVERY

CALCULATED RECOVERY ~fsO.08

2000 -

0
0

500 -

LAKEVIEW
Q

POOL
o

OBSERVEO

CALCULATED

200 5

10

Is

20

25

30

YEARS

Fig. 12Cumulative recoveryvs time.


MAY, 1978

1
:

0.13

10

5000 -

;
A loco 5

-- CALCULATED RECOVERY ~r

35

40

100
0.5

4
1
1

1
2

I
6

I
10

1
20

I
50

YEARS
Fig. 13-Oi productionrate vs time.
827

Sh3pC
Of3.5 fti thCdttt8reasonablyWC~.
la these examplca, cumulativeremwerycalculations
were performedon the basis that the fiekk were bciig
producedattbcirmaxhm~viV-ti@!em@
Forthe
MileSii Pool, Fig. 9 of Ref. 10shows thatthemaximum
productionratewas notreacheduntilalmost3 yearsafter
the field fmt wm placed on continuous production. If a
flcld is produced at less than gravity-drainage ratcfh then
another approachmust bc followed using the recovery
equation.
This quation can be appliixl8s long 32production rate
is equal to or leas than gravity-drainage rate. At leas than
gravity-drainage rates, the gas-oil intcrfitcc will bc retadcd, but fhc saturation distribution (Eq. 23) above the
retarded zd level at any given time will bc the same as
under maximum rates. Hcncc, if Zd is known from field
performance as a function of time, onc can usc& and the
corrcaponding t in Eq. 24 to calculate recovery, rather
than using Z~ calculated fmm integration of Eq, 21.
Calculations of recovery from Eq. 24 as a function of
lCVC1
of the gas-oil intcrfacc arc shown in Table 6. The
recoveryin Col. 2 is o@aincdby subtmcting0.385 fmm
the recoverygiven in Col. 5, Table 5 of Ref. 10. This
recoveryafter the startof gravitydrainageassumes the
reservoirwas essentially at equilibrium on Sept. 30,
1933. Col. 6 rcprescntathe ratioof volume swept in the
intervalfruminitialgas-oil contactto oil-wotercontact.
For example, the fnt 100-ftdrop sweeps ozJy 48 per-

cent, as muchof the rwervoiras if sand vohtmcwere a


linearfunctionof depth.Thus, rccovcrycalculatedfrom
fhc rccovcryequationmustbc multiplkd by this ratioto
obtain the adjuatcdrccovcry shown in Col. 11. The
agrccmcntbetween calculatedand obsemwclrecoveryis
good. Differencesrange from -1.5 to +2.4 pcrecnt.
Results of this approach suggest that the recovery
equation can bc used to calculate Z* for any selected
productionrate less than the gravitydrainsgc rate. In
addition,the -fall
rstc mustbc detmmhwdusing Eq.
21, startingwith corrcaponding values of Zd and t from
the mcovcry quation. When the -fall
rate becomes
lCSSthan the sclcctcd production rate, the L%avcty equation can no longer be used to obtain Zd. Rather, E@.21
must be used to determine i?(j. This vak and the Corresponding titnc must bc hwted in Eq. 24 to obtain recovety. The calculations bccomc considembly more complicated becauac calculating zd fmm the recovery equation
is a trial-and-error proccas. Calculations could bc prorti
in a computer so that at each time step both the
mcovcry equation and Eq. 21 are solved forzd. Then, the
ike-fall rats can bc compared with the rate used to obtain
zd from the rccovcryequation. Time steps can be made as
small as necessary so that when rates match, the cornputcr thcrciafterwill calculate the recovery and decliniqg
rate for free-fall conditions only.
Thcsc examples illustrate how gravity-drainage recoveries and rates can be predicted and how observed pcr-

TABLE 5-OBSERVED AND CALCULATED RECOVERYAS A FUNCTIONOF


OAS-OILINTERFACE LEVEL AND TIME, MILE SIX FOOL
(6)
(2)
(1)
(3)
(4)
(5)

.Date
9/30/33
9130134
3/3f/36
l/31/ti
10131130
413W42
4/30144
9f30f45
0J31148
0130148
1/31/50
5131152
(7)
YearMonth

o
;.$
4-4
61
8-7
10- 7
12 0
12-11
15. 0
16- 4
18- 8

Recoverv
(MM bblj
o
2.302
8.280
17,244
21.810
24.757
26.454
27233
27.731
29.072
29.655
30.031
(8)

Recovery
(percent)

1;:

E
48.2

Ratio,
AAFIAh:
AFfh

(ft)

1.875
1;975
2,110

0A80
;:
375
435
465
485
405
505
515
520
525

.613
,769
,840
.876
.801
.913
.926
.939
.945
.950

51.5
53.0
54.1
56.6
57.7
58.5

%%
2,340
2,360
2,370
2,380
2,380
2,325
2,400

(9)

(lo)

(11)

(12)

Catculeted
Reoovefy
(percent)
10.%
25.8
41,8
49.2
64.0
57.1
59.1
60.2
82.0
62.9

Adjusted
Calculated
Recoverv

Calculated
Obsefved
Recovery

Zd

Days
365
913
1,582
2,220
3,133
3,563
4,380
4,716
5,475
5,862
6,813

Gsaail
Interface
Level

(ft)

333
781
1,247
1,447
1,546
1,613
1,658
1,679
1,713
1,728
1,746

1:.:
32.0
41.4
47,4
51,5

54.0
55,7
58.2
59.5

+0.5
-0.3
-1,5
-1.0
-0.8

0
+1.0
+1.8
+1.6
+1.8

Cols. 1,2, and 4fmm Table 3 of Ref. 10.


Cot. 9 isCol. 5/sin 17.5.
Col. 10 Is calculatedfrom recoveryequatbn.
Cot. 11 h~l. 10 x COL6.
Cal. 121SCOI.11- COI.3.
8ZS

JOURNALOF PETROtSUM TECHNOWOY

formancecan be compared with calculated performance.


In each example, a constant value for drainage modulus
was used, ha reality, drainage modulus does not remain
constant, For exmple, in the Mile Six Pool, pressure
varied somewhat d?wingthe life of the pool. This caused
both density and vtscosity of the oil to change. In addition, if the oil was initially at bubble-point pressure, a
decrease in pressure would force gas out of the solution,
thereby lowering the effective oil Permeability. These
factors can be accounted for by calculating drainage
modulus as a function of time and using an integrated
average value rather than a constant value.
Another factor affecting the performance is infiil drilling, which occumd with the Lakeview Pool. Additional
drilling decreases the spacing causing L, the effective
length of a drainage cell, to become smaller. Again, in
this case, drainage modulus can be calculated for each
change in well spacing and the modulus integrated with
respectto timeor recovery to obtain an integrated average
value up to any time, t. Accounting for a changing drainage modulus should improve the correspondence between calculated and observed recovery and should result
in more reliable predictions.
Obviously, the value of the drainage modulus is the
key to determining how rapidly oil can be produced by
strict gravity-drainage production. Three terms under an
engineers control are effective oil permeability, oil viscosity, and well spacing, which determines L. The
permeability-viscosity ratio can be kept at maximum
value by maintaining the reservoir at the original
bubble-point pressure, Any drop in pressure below
bubble-point decreases this ratio, For a field such as the
Lakeview pool, this decrease can be as much as a factor
of 5, resulting in a gravity-drainage mte of one-fifth the
rate at the bubblepoint pre+wre. Furthermore, for a
given residual oil saturation, the tank oil left behind also
would be greater for a pressure-depleted field because of
the smaller oil formation-volume factor. Therefore, the
pnxsure must be maintained for maximum gravitydrainage rates and for maximum recovery. If natural gas
is not available or is too expensive, Lfienserious consideration should be given to using inert gas. 1o~

Suggested Research Work


lleexpenmental data discussed earlier is not completely
satisfactory. In each study, some required information
was not obtained or was not reprted. This leaves the
field open for a more thorough .msearcheffort devoted to
obtaining all required information to test the theory and
its application mom fully. For example, in all the experiments, only a few data points wem taken in the early
drainage period, This period is controlled almost solely
by permeability of the porous medium (assuming that
viscosity and density of the liquid are known) and serves
as a check on permeability measured with steady-state
flow.
Tests could be performed with different lengths of sand
tubes, using a variety of liquids so that the capillary-rise
term, H, could be varied as a function of permeability,
density, and interracial tension. The effects of residual
liquid saturation and Exponent6f cotdd be determined by
using a variety of sand packs. In fact, the importanceof
the formof the relativepermeabilityquation also could
be tinted by fitting measuredrelativepameabdity data

mom exactly than can befitted by a simple straight line on


a log-log plot.
The assumption that each well forms a separate drainage unit should be checked also. If this assumption is
correct, there should be an optimum economic spacing
for afield with good gravity-drainage potential.

Summary
Thisstudy broadened the application of Cardwell and
Pamonsmethod of predicting oil recovery under free-fall
gravity drainage and presented three examples of how to
apply the method. The theory was tested against experimental data and was shown to be good, even though
drawbacks existed in experimental data. Several research
ideas weresuggested to overcome these drawbacks and to
verify the assumptions.

Nomenclature
B = exponent in relative permeability equilion
BO= oil formation-volume factor
C = constriction coefficient
dO= oil density, grrdcc
g = accelemtion caused by gravity= 980 crnhed
H = equilibrium height of capihry rise
h = formation thickness
k = permeability
k,= relative permeability
L = length of draining column or formation
m = constant
N= oil in place
Ap = pressure drop
qO= oil rate
R = remvery
S = saturation
t = time
v.= flow velocity in saturated region
vu= flow velocity in undersaturated region
W= formation width
Zd = distanceof demarcator or gas-liquid interface
fmm top of draining column
a = angle of dip
@= porosity,fraction
A = viscosity
p = density
subscripts
i = initial
g=gas
1= liquid
O=oil
r = residual
w = wat~
References
1. Canlwctl,W, T.andParson,R.L.: OraVitY
Trans.,
AIME (1949)179,199-215.

Draiaascl%c4xYt

2. Katz, D. L.: possibilitiesof ScsondaryRccoveIYfor the OklahomaCity Wilcox Sand,*Trans., AtME (1942) :46, 3s-53.
3. Lcwk
(N44)

J. O.: WmvityDninagc in Oil Fklds,

Tram.,

AIMB

133-134.
of aHighRclicf Pool,
4. Bunchdl, f?.P.: RC$mOkfblf~
Trans.,AIMB(1949) 186,171-179.
5. Blkins, L. R,, Pmnch,R, W., md f31enn, W. E.: LanceCreek
MS,

Sumlame Reservoir Pcdorrmncc A UnitizedPrawmbf&IcnuxxRojcct,Tmns.,AIME(1949)179,222248.


6, Wickcnhuuer,
L.J.: 0$8-Driw,OmvitySc@gotiofI,
@ -

829

.,

InjcctIoo(hlcuhtiottsA
PPIlcdtoLl Band, TSc@ncitt,tidM
Field, Veneswls,0i/andGusJ. [k.
29, 1949) 52.

7. sims,W. P. ad Frailin, W. 0.: LckcviewPOOLM~waY


Su~t~ld,T,am.,
A&(l9W)
189,7-18.
8, Welge, H, J.: A SimplitlcdMcdmdforCotnput&gtil Rwovcry

byGaa or WstcrDrive, 2hwrs.,AIME (19S2) 19S,91.9S.


PrwlIctioasxnc4mpolBuog
Gravity
9. McCord, D. R.: FWformaoC@

lhalnagc and (Ma Cap Makitctsam, LL370 Asca. Bolivar


Cua.stalFiild,?%ts., AIME (1953) 199,231-248.
10, ArI&Is, E. L., Jr.: *Mile Six Pool An Evaluationof Rccwcry
Ef?ickncv.Tratts,. AME (1953)19S,27%226.

11. Siewciii

M., @ibwako,D.

Ma&tcnanccby Inert k
12.

Field,lmtrs.,
AIME(19S5)204,49.55.
Stueve,D, R. andWelch, L, W., Jr.: Gas IMvc and Gravity
Drainage Analysis for Prosmirc-MsintcnanceOperations,

Trans., AIME (19S6) 297,136-143.


13. Wooddy, L. D., Jr,, andMoscrip,

R. IIL performance calcul8tlona for Combimtion fXive Reacrvoisa, Trans.. AIME (1956)
207,128-135.
14. Kitby, J. E., Jr., Stamm,H. E., andSchttkz,L.B.: Calculation
on the Depletion History and FuturePctforrnanccofs Gas-cap.
DriveReacrvoir,Trans.,AIME (1957)210,21S-226.
15. Easley,P. L., Hancock,
tl L.,andJones,IC.E.:GscvitY
mnitgcconceptsin Steeply Dipping Rcacrvoira, paper 1029-(3presented at the WE-AIME Pctmlcum Ptuduct(onand Rcscwoir
Enghccring Confcmncc, Tulsa, March 20-21, 19S8.
16. Stanley, L, T,: *AppmximatktttosGas-DtivcRecovery and Front
Movement in the Abqalq Field, Saud! Arabia, Trans., AIME
( 1960)219, 273-2S0,
17. Hall, H. N.: AnalysisofGravityilral~e,~.

PeI. Tcch. (Sept.

1961)927.93G Trans., AIME, 2J2.

of(RavityScgmgationPctfotmanccDuringNatutalCkplction, Sot, Pet. Eng. /. (Sept. (962) 261-274,

18. Ccok, R. E.: Analysis

AIME, 22S.
19. Brcitcnbach, E. A.: A Computer Simulationof Gravity Drainage
in Oil Rsacrvoirs~ paper SPE 895 prcacntad at the WE-AIME
39th Annual FalI Meeting, Houston, Oct. 11-14, 1964.
20. Joslin, W. J.: Applyingthe Frontal Advance Equationto Vertical
Trans.,

ScgmgationRcacrvoira,J. Pet,

Tech.

(Jan. 1964)87-%,Trans.,

AIME,231.
21. Hall, H. N.: hficting Gravity Drainage Pctformancc Using a
Thtec-Dimensional Model, J. Pef. Tech. (May 1%8) 567-S24;
Trans., AIME, 243.
22. BuckIcy,S. E. andLcvctctt; M. C.: Machanism of Fluid Displacement in Sands, Treas., AIME(1942)146, 107-116.
23, Tcrwilligcr, P. L., Wilscy, L, E., Hall, H, N., Bridgca, P. M., and
Morxc,R. A.: An Expcnttwntal and Thcactical Invcatigation01
~8&it Drainage Perfotmancc, Trans., AIME (19S1)192,

.
24. Wyckoff, R. D. and Botsct, H, G.: The Flo., of 0ss Liquid
MixturesTluuughUnconsolidated
Sands,Physics ( 1936)7, 325.

830

2S. Stahl, ft. F., Martin, W. A., and Huntington, R. L.: Gravity
sands, nans, , AIME
DrainMe of LhWidatkool Uocaosolhiatcd
(1943)151.
13LM6.
. . . .. . ---, --- .-.
26. MUtbl, J. c.: sOttlChfatbcrilatid Aqmcta of Two.HOW

With Appliatioiu to ~oodlstgendGtEvltyScgrcgatknProbIcma,htf. Mmthfy (A@ 1958)22, No.6,22.

27,

Brmutcn,R. D. and Silbatbcrg,M. y.: continuousSystem


ModelingRugrama,Sy#remsJ., IBM (Dee, 1%7)6, No. 4.

2s. King, F, H,: Rinciplea md Conditionsof the Movctnwm of


Ground Water,Atttua/l?epo
rt,

USGS (1699) 19,

29. Slichtcr, C. S.: fkaicticcl Invcctigations of tbc Motion of

L:,andKrcMJL
F, K.:*PICCSUIC

Injcctlon in tbe Hlgb ReliefElkBasin

GroundWstcts,Atmta/Repon, USGS (1899] 19, Part2.

30. Wilscy, E, F,: An Expesimcntal and Analytical Study on Permeabilityof Sand,*PI@ Utrais, Dept. of Mcchanks and ffydrcuIicc, U. of Iowa, Iowa City (193S).
31. Krumbcin,W, C, andMonk, G. D,: *Pcmteabilitycca Function
of theSiZOFaramatcraof UncontolldctadSand, Trans., AIME
(1943) 1S1, 153.163.

32. Templeton, E, E., Nielsen, R. F,, and Stahl, C, D.: A Study of


Gravity Countcfflow Scgmgation,*Soc. Per. Ettg. 1. (June1%2)
18S.19XTrans., AIME, 22S.
GravityDrainageCharactcrhticcon
33. Mcrx, J. W.: *Detcrmlning
theCa@ugc, Trans., AIME (19S6)207,88-91.
34. Musket,M.: Physical Principles of Oil Production, McGTcw-HN
Book Co., Inc., Ncw York (1949)487.

3s. Matthwva,C. S. andLdkovita, H, D,: GravityDrainagePerformanceof DepletionType Rcacmoirriin the StripperStagct


Trans.,AIME(1956)207, 26S-274.
36, Dodson,C,R. andcatdwcll,W.T.,Jr.: ITowIntoSIottcdLincts
and an App[icmiosrof the Theory io

(194S)NW,56-64.
37. Swndjng,M. B.: Volumetric

core

Analysis,* Trans.,

AIME

HY&ocar&vt

$stems,

and Phase Behavior Oj Oil Fieid

Smicty of IWroleum Enginccraof A~ME*

Dallcs( 1977).
38. Bcal,C.: ViscosityofAir,Water,Natural(3ss,CtudcOilandIts
AssociatedGasm at Oil Field Tcmptaturcs and Pressures.
Trans., AIME (1946) 1%,94-11S.

39. chew, J. N. and Connally, C. A.: A viscosityC@tcla!ionfor


GassaturatedCrude011s, Trans., AIME ( 19S9)216,23-2S.
40. Gates,J. and Tcmpclaar-Lictz,W.: RelativePctmcability of
California CoICsby tk Capillary Pressure Method, Driii. and
Prod. Pmt., API ( 19S0)285.
41. Baistow. W, F.: Entiinc Exhaust Boosts011RccovcrY, Oi/and
Gas J. (tkatch 26, 1973)71,78,
42, Sincovcc,R, F.: NumcticalReservoir Simulation Using @di.
nasy Diffcrcntisl Equations !ntcgtritor,Sot. Pet. Eng. J. (June
1975)25S-264 Trans., AIME. 2S9.
JPT
Of@nslrnanusmlpl
f.wivedInSocISty
o~tWrobum En@sers ofka Feb. 1,1977.
Paper w@td
for Publkxlm
Aug. 22.1677.
Rewed
mmwwlp!
mmlved JSn. 6,
1s7$. Paper (SPE s548) was presenred the SPE.AIME 47rh Annual Callfofms
RefjbnslMes!hg, Md in 8aketafMd, APH 13.15.1977.
Thmpstw wilbs includedmthe 1978 ~mnsac?msvolume.

JOURNALOF PETROLEUMTECHNOLOGY

Você também pode gostar