Use of Simple Mathematical Models for

Predicting Reservoir Behavior

J. G. Richardson, SPE-AIME,Esso Production ResearchCo.
R. J. Blackwell, SPE-AIME,Esso Production ResearchCo.

Introduction
Forecasting the behavior of a reservoir is one of the
more important but complicated tasks of engineers
in the oil industry. Knowledge of reserves remaining
in a reservoir is vital to planning optimum depletion
of a field. Unfortunately, the engineer assigned the
task of predicting reserves often faces a CM%CLA
choice. For the most accurate answer, he can use a
computer program that takes into account all of the
pertinent factors, but this approach is usually expensive and time consuming, and requires a detailed
knowledge of the reservoir. On the other hand, he
can use conventional one-dimensional displacement
calculations that are easily applied but that in some
cases do not adequately describe the reservoir flow
system. The purpose of this paper is to describe a
middle-ground approach that in special situations has
many of the advantages of the above methods with. .
~rnwbacks, This approach
out their more seriom ---SJS:S~.ath~matical models that describe the principal
flow mechanisms and can be quickly appiied by hand
calculations.
..~ A~~V d-. predicting reservoir behavior
The Icv,,nv.wb,
has grown steadily since the pioneering work of
Muskat and of Buckley and Leverett., 3 Muskats
tank-type or zero-dimensional method of predicting
behavior in dissolved gas drive reservoirs has been
invaluable to the industry. Another milestone was
reached with the Buckley -Leverett method of predicting linear displacement of oil by water or gas when
flow was principally along the bedding plane. The
classic work of Hurst,4 Muskat.5 and van Everdingen
and Hurst6 laid a firm foundation for problems in-

volving unsteady-state flow of fluids. Later, progress

was made by Welge7 in solving onedirnensional displacement equations more easily. The advent of digital
computers led to the development of methods of
solving problems of greater and greater complexity.
~ndi~ating the progress being made with computers,
Douglas er al. in one papefl inciuded tiie ei?ect of
capillary pressure in one-dimensional flow, and in
another papere dealt with the flow of two phases in
two dimensions. The utility of computers in predicting reservoir behavior has continued to grow as programs have become more user-oriented and as computers have become faster and more economical to
employ. But even today, the time, effort, and money
required to use computers to solve reservoir problems cannot always be justified. Thus, other tools are
needed.
An excellent example of another approach is given
in a paper by Jos!in. 10 ~~ ~nalvzin~ a gas injection
project in a large Venezuelan re&vo;r, Joslin realized
that gas had overridden the entire oil sand became
production was above the critical rate. However, the
pre=nce ~f pan~~ke-like shale members penetrated
by the wells prevented coning of gas into perforations
located below the shale near the base of the sand. Oil
recovery was predicted by assuming that gas displaced
oil vertically downward throughout the producing
area. Another paper demonstrating the practical use
of simple mathematical models is that of Matthews
and Lefkovitsll for predicting producing rates for
wells in depletion-type reservoirs.
We shall describe here technology that can be used

~.HeTe is an approach
jor solving reservoir flow problems
.. ..
.*
{nr
i by a rate-lzmztlng step, Simple mo~en are d zvelope~J I-,
f70w occurs, jor water unfierrunning cj viscous oils, jor
, banks in gas caps, and jor control of coning by injection
(
----l#QC?

7Q71

d7r-

where behavior is dominated

uravitY drainage where vertical
~.
gravity segregation of water
oj oil.

1145

to solve several other types of reservoir problems.

Our purpose is not to write an exhaustive treatise on
each of the subjects covered, but to show an approach
to solving reservoir flow problems.

Use of Simple Mathematical Models for

pre~,c~lfig p=e~er~oir &havior
Understanding the basic mechanisms that govern flow
of oil, water, and gas in any given reservoir situation
is necessary in developing reliable methods of predicting behavior in that reservoir. Four kinds of
forces viscous, gravitational, and capillary or diffusional can be involved in the dkplacement of oil
by water, gas, or solvent from a three-dimensional
Thus, analysis of a displacement
~OiCW
ItRd!U~..
process in a particular reservoir can become quite
complex if each of these forces and flow ht each
..
-+ 12firt11ma~e!y,~tudies of some
IIIIIJUI L~lL. s1 .
mmension are- :---+
situations indicate that one force is dominant and
that only one dimension is involved in the ratelimiting step. In these circumstances, solutions can
be direct and simple. In the following sections we
shall describe several types of such solutions.
A necessary step before applying these models to
a particular reservoir is to study its past behavior.
If reservoir conditions agree with the simplifying
assumptions used in developing the model, then the
model can be used to try to match past behavior. If
past reservoir behavior can be matched, the model
can then be used in predicting future behavior.

Second, oil drains vertically downward through the

gas-invaded region, joining the main oil column by
flowing along the bottom of the interval.
The Mathematical Model
A ve~ simple mathematical model accurately deof oil by gas drive and
~cfibes ~he ~isplacement
gravity drainage when the rate is less than one-half
the critical rate. The first calculation determines the
gas saturation just above the gas-oil contact, using
the Welge procedure to solve the fractional flow
equation.

Gravity Segregation
Physical Description of the Problem
In steeply dipping reservoirs containing sands with
high vertical permeabilities, gravity drainage of the
oil can be much more effective than one would calculate if he assumed that all flow must be parallel
to the bedding planes. When sutlicient vertical permeability exists, gravity drainage by flow of oil vertically
downward through the sand can be quite rapid. If
the dip is great enough, oil that drains through the
thickness will flow in a thin layer along the bottom
-.--..1 +~ +h. h~ce nf the ~eservok as shown
d he ~lllG1
in Fig. 1. Thus, the displacement process occurs in
two steps. First, gas invades the sand and the main
gas-oil contact moves downdip. The tilt angle of the
front can be predicted using Dietzs method, if the
rate is less than the critical rate given by
Val

LU

.1!-

--w

-.

. .

0.044 k Ap
(9/~)criti.al

or

PO

@ = total flow rate, cu ft/D.

This saturation is found by plotting F, vs S, and finding the tangent to the curve passing through the origin.
For ease of calculation, the gas-oil contact is assumed
to move at a constant rate.
The next calculation determines the quantity of
oil that drains from the region invaded by gas in a
given time. For ease of calculation, this region is
divided into arbitrary lengths and the amount of oil
produced by vertical gravity drainage is calculated
for the average time since passage of the gas front.
For vertical drainage of oil, the rate is given by
Darcys law, with the driving force arising from the
density difference between gas and oil. We have
assumed, as did Cardwell and Parsons,19 that resistance to flow of gas and capillary effects is negligible.
u~ =

0.044 k. ~p
/&7

9 ..,-..

(3)

in which
U. = oil fiow rate, cu ft/D
k. = effective permeability to oil, darcies;

(1)

=
_. PO

in

Sill

Ffl = fraction of flowing stream that is gas

k gr

which
4 = total flow rate through area A, cu ft/D
A = area of cross-section normal to bedding
plane, sq ft
k = perrneabllity, darcies
Ap = density difference, lb/cu ft
viscosity of oil, cp
=
l% viscosity of gas, cp
k = relative permeability to oil, fraction
k; = relative permeability to gas, fraction
a = angle of dip, degrees
p.

,,AJC

Fig. lMechanics of gravity drainage.

JOURNAL OF PETROLEUM TECHNOLOGY

and from continuity

o
dz
X6.

1 du. ,..

+ dS.

(4)

in which

h.
k, Ap
P@
by the time required for flow along the bedding plane,
r,=

z = vertical distance, ft
t = time, days
= porosity, fraction
oil saturation, fraction.
:=
The rate of movement of a particular oil saturation,
dz
~ , , can be determined by plotting UOcalculated

()

fr~rn Eq. 3 vs saturation, taking slopes to determine

duJdSo and dividing by porosity, as indicated in Eq.
4. The amount of oil drained from each region during
the time since passage of the front can be calculated by graphical integration of a plot of height vs
saturation.
As a first approximation, the time for oil to flow
downdlp in the thin layer along the base of the
formation can be neglected, as can the volume of
..-,
.
~mt p-?e ~~ceeds
od m ttus layer. H the displacer%...
one-half the critical rate, oil tends to accumulate
rather than flow away along the bottom of the interval. More accurate calculations also require consideration of the thickness of the transition zone arising
fr~m capillary effects above this layer of oil, especially
if the sands are less than 10 ft thick.
:. +
- m;fi1- nre
Recoveries calculated by. .thl~
te-hl.+--- quite
sensitive to the values of relative permeabdity to oil
at very low oil saturations. Conventional laboratory
data can be extended to low oil saturations by plotting measured values of effective permeability to oil
on log-log paper vs *

so _ &*
in which SO.*
S,ni S..*

is the irreducible oil satu~ation in the presence of

connate water.
If a measured value of S., is not available, a value
is chosen to yield a straight line through the data so
ko, =

SO,*
n
0
1 S.,* Swi )

The slope
of the line, n, should be approximately 4
..
..
-+ =..,
A 14 ~lu~ may be
according to the meory of Ckiie .,y =,
as large as 6. The value of S..* should be nearly zero.
Eqs. 3 and 4 can be solved analytically. Recovery
TABLE

data can be correlated by the dimensionless parameter,

/& hP
which is derived by dividing the time rek,, Ap ~
quired for vertical drainage,

UT

lAVERAGE RESERVOIR PROPERTIES

FOR THE HAWKINS FIELD

Average permeability, darcies

Average vertical permeability, darcies
Average porosity
Average initial water saturation, percent
Angle of dip, degrees
Average thickness (in vertical direction), ft
Average length (along bedding planes), ft
Average reservoir pressure, psi
Average vkmsity of Oi!r Cp
Average viscosity of gas, Cp
Average density of oil, lb/cu ft
Average density of gas, lb/cu ft
Average total flow rate per unit area,
cu ft/sq ft.D

3.40
2.38
0.279
8
6
49
3,500
1,500
4.45
0.0185
51.7
5.35
0.0365

t+

k, is the permeability in the vertical direction and h,,

is the thickness in the vertical direction.
Example Problem
The utility of this simple model can be illustrated by
predicting recovery by gas drive and gravity drainage for an actual resewoir.
Given: Average reservoir properties as presented in
Tables 1 and 2.
Solution: Using Eq. 1 and the average reservoir prop
erties given in Tables 1 and 2, the critical ra= foi
Hawkins is calculated to be 0.173 cu ft/D-sq ft. Since
the average actual rate of 0.036 cu ft/D-sq ft is only
21 percent of the critical rate, the simplified model
should apply. Using Eq. 2, the average gas saturation
just above L5e ga~.ei! c~ntac~ is found to be 45 percent by the Welge procedure, and the actual rate of
frontal movement is found to be
Uac

~~

+=

(0.0365)(365) = ~06ft,yr
(0.279)(0.45)

Time to gas breakthrough = 3,500/106 = 33 ye=

Recovery at breakthrough may now be estimated by
dividing the reservoir into seven blocks 500 ft long
and 49 ft thick. The average vertical distance a given
saturation drains in each block can be calculated from
Eqs. 3 and 4 knowing the time interval since passage
of the gas front. The residual oil left in each block
is determined by graphical integration of a piot of
So vs height. Results of these calculations are shown
in Table 3.
.-.,.-7 tm
is 0,92 0.076
w rewr.,
.----uac hretifirough
0.92
(loo) = 91.5 percent of the oil originally in place.

TABLE 2-REIATIVE
PERMEABILITY-SATURATION
REUTION FOR THE HAWKINS FIELD
Gas
--Saturation
(percent)

c).10
0.20
0.30
0.40
O.m
0.60
0.70
0s0
0.90

Relative
Permeability
to Ga.1s
(dareies)

Relatwe
Pe~oe;ility
(darcies)

0.018

0.49

0.054

0.28
0.15

0.105
0.184
n
e-901 .

0.420
0.60
0.78
0.97

0.07
0.027
0.0076
0.0015
0.00019
0.000017
1147

TABLE 3-AVERAGE
Block
~um&~
1

2
3
4
5
6
7

SATURATIONS BEHIND GAS FRONT

Orainage

Dis/f;Ee

tD

500

500 to lrooo
1,000 to 1,500
1,500 to 2,000
2,000 to 2.500
2,500 to 3,000
3,000 to 3,500

Dsys

11,180
9,460
7,740
6,020
4,300
2,580
860

Time
Yesrs

Residual Oil at
Gas Breakthrough
(psrcent)

30.6
25.9
21.2
16.5
11.8
7.07
2.36
Average

m
5.5
6.0
6.7
7.5
9.0
13.5
7.6

Comparisons with Field Data

Additional calculations were made using the model
for ranges of rates, vertical permeabilities, and sand
thicknesses. The results are correlated in Fig. 2, which
is a plot of recovery at gas breakthrough vs the dimensionless parameter uT P, h,/kr Ap L. The recovery of
87 percent observed in the Hawkins field is shown
for comparison in Fig. 2. This recovery was obtained
by dividing the oil produced from the gas-invaded
area of the field by the volume of oil originally in
place in that area calculated from oil isopachous
=-A.
Airted
~r.. .. by the
maps. 1he 4.s percent recowly m
simple model is in excellent agreement with the 87
percent value observed in the field.
Another check was made of the validity of the
simple model by comparing results with calculations
made with a two-dimensional, two-phase reservoir
simulator using alternating direction implicit procedure (ADIP). Although slightly different values of
relative permeability to oil at low saturation were
used in the simulator calculations, the value of 87.2
percent recovery at breakthrough for conditions used
in the computer solution coincides with the curve
P!Qtted through results of the simple model. Independent calculations show that the slightiy different
relative permeability values used should make less
than 1 percent difference in recovery, so the agreement of the computer results with the curve suggests
that both are essentially correct solutions.
By contrast, the recoveries observed in the field
and predicted by models that permit flOW in WO
dimensions are more than 15 percent greater than
those calculated by conventional one-dimensional
techniques. These differences arise because vertical
drainage through the sand thickness is not accounted
for in the conventional one-dimensional calculation
methods. which assume that flow occurs only along
the bedding plane.
Summarizing, this study has shown that a simple,
gravity drainage model can be readily app!ied tO PRdlct recoveries by gas drive and gravity drainage when
there exist certain conditions: flow rates are less than
one-half the critical rate and permeabilities in the
..
---with the
I+ h qecllltq
~it? ..x&.
. -.-. + obtiined
.
verucai ciire~uuzi
~-&_ei Me in good agreement with recoveries observed
in the Hawkins field and also agree with solutions
obtained using the two-dimensional, two-phase reservoir simulator. By contrast, attempts to use twodimensional models to match past reservoir behavior
in gas-invaded areas of Friendswood field were un.
successful. Although ~~.~~.wfind
~ , ,W.lu=.--has
_ thick, highly

permeable sands, low-viscosity oil, and displacement

rates much less than the critical rate, recovery predicted by the two-dimensional models is much greater
than that observed in the field. G Permeabilities in
the vertical direction at Friendswooci are apparently
very limited. So, as a practical matter, the simple
model should be used in predicting reservoir behavior only when it can be shown to match past
behavior. When more complex models are needed,
those proposed in the literature should be studied
for applicability.-
Although the simple mathematical model can also
apply to some water drive reservoirs, the utility of
this approach is restricted for several reasons. (1) The
mobility ratio is favorable in many water drive reservoirs, and viscous forces can be quite efficient in the
initial displacement phase. Thus, little mobile oil is
left to drain by gravity. (2) Segregation rates are less
than in gas-invaded areas since the density difference
between water and oil is less. (3) In reservoirs containing viscous oils, practical dkplacement
rates
usually exceed the critical rate and this first model
does not apply. Another type of model, described
below, may apply in the latter case.

W-ater Lhxierrunning
Physical Description of the Problem
When water displaces viscous oils. conditions are
seldom favorable for uniform contacting of the sand
by water. Production rates, because of the demands
of economics, must usually exceed critical rates,
making it impossible for gravitational forces to maintain a stable displacement front. The mobility ratio
is slightly unfavorable in these reservoirs and water
tends to channel and bypass oil. Since water is more
dense than oil, it seeks the bottom of the interval and
channels or tongues under the oil. The phenomenon
and
has been discussed @ a number of authors9
ElkinsZ published a case history describing the
process in the Fosterton field.
When water arrives in the region under a producing
well, it tends to cone up into the well, if vexlical
permeability exists, and be produced with the oil as
shown in Fig. 3. Production histories in such reservoirs are characterized @ early breakthrough of
water into all of the producing wells, followed by an
extended penod of gradually increasing water-oil
ratios. Study of the fluid mechanics involved in these

>5

{H 20 I
:
a

o
~-b

oGRAVITYOSAINAGEMOOEL
AU~WK&JSFl~ RECOVERY
2-O,2.PNASE
COMFUISRCALCULATION
,
,..5

Fig, 2Recoveries

predicted

,.-4

using gravity drainage

I
1
,:.3

model

compared with field and computer results.

JOURNAL OF PETROLEUK1TECHNOLOGY

reveals that the dominant displacement force

arises from the viscous pressure gradients exerted on
the oil by water flowing in the layer under the oil.
For practical purposes, the time required for mobile
oil to rise by gravity from the water layer can be considered instantaneous compared with the time required for it to flow through the oil layer along the
bedding planes to the producing wells.
The Mathematical Model
Consider a cross-section of sand containing viscous
oil as shown in Fig. 4. Assume that water underruns
the oil because it is more dense than oil, and assume
that the water layer is of constant thickness. At a
patiicular time, the fraction of the total thickness
occupied by water is x and the fraction occupied by
oil is 1 x. 0]1 and water flow horizontally with
the viscous pressure generated by the invading water.
It is assumed that as oil is produced, the water layer
moves vertically upward with no resistance. The
average oil flow rate along the length is one-half the
oil production rate since no oil enters the inflow face
and the average water flow is the rate of water production plus one-half the rate of oil production. If we
assume that both water and oil fiow horizontally
through unit width, we obtain

df

q,.

cases

L+(l

Stri

= k
2ko It,,

xdx

1 -l

dx
.2

(8)

(9)

lC.

.$Or)

ro

P(I

Integrating Eq. 8, we obtain

= k Po[- xln(l-x)
2k. P,.
.

]--$-,
.

in which W,, is the total water production in dkplaceable pore volumes. Water influx is found by
w,. = w,, + x. Cumulative oil recovery, Nj,, in displaceable pore volumes is equal to x. After breakthrough, the instantaneous produced water-oil ratio is
given by
k....

qtr
=

2ko

q.

ILO

}1,,

()

(10)

mz

Example Solution
The ~!i!ity of this model may be shown by matching
behavior observed in an actual field where the displacement process is the underrunning of oil by water.
Given:

and

SW, = 0.11
0.20
s
50 Cp
w = 0.4 Cp
k. = 1.0 darcy
0.25 darcy
k
or

()%

p.

~.L

APO=

k~(l

(6)

x)

w,

in which k,,.,. is the effective permea~llitY to Water

at residual oil saturation.
Produced oil is replaced by water moving vertically
upward, so
q. =+

L$(l

SW, s.,).

(7)

By assuming that the pressure drop is the same

through the oil and water layers, equating (5) to (6).
solving for qo, eliminating qo with Eq. 7 and grouping variables, we obtain

To

Substituting the above data into Eq. 9 provides the

following
we =

(5o) (0.25) ~_ ~ ln(l

(2) (0.4) (1)

x)]+

= 15.7 in (1 x) 15.2X.

A table may now be prepared relating x, the fraction

of the thickness invaded (or recovery in percent of
displaceable oil), to water influx and produced wateroii ratio (Table 4).

ZONE
wmn
... . mmn
.. . . . . . m!
. . ..

WATERZW

q02
~.
v
Fig. 3-Mechanics
.Fn-rr

R#nrl?

1071

of water underrunning

and coning.

Fig. 4-Model

WATERLAYSU

q W2

for water underrunning.

1149

TABLE

&PREDICTED
BEHAVIOR
UNDERRUNNING

Oil Recovery
(displaceable
volume or
fraction swept)

Weter Influx
(displaceable
volumes)

WITH

Water-Oil

Ratio

0.1

0.13

~.~g

0.45
1.04
1.92
3.26
5.28

3.42
6.22

0.6

10.00
15.00
23.00

Water Cut
(percent)
~~e~

Comparisons with Field Data

Results obtained by the above calculations are compared in Fig. 5 with recoveries and water cuts observed in the field as functions of water influx. Note
that the field data are closely matched by the curves
derived by the simple mathematical model. Calculations were also made of conditions existing in the
Hawkins field, Recovery at water breakthrough was
predicted to be 40 percent, which compares well with
a value of 42 percent calculated by dividing the oil
production by the volume originally in place in the
water-invaded regions. Thus, it is concluded that this
model can be used to predict future production char. . . ..
.
-lc,.nmpl
IA d that
1A. 1.
:. a,av
actenstlcs m mese wvo cmm>.
IL IS
~ul.w,~-e...-.
.,~ht=r~
., .W.accllmntio~S
. ..... .. r.._ Usecl in deriving the model apply,
it may have utility in predicting behavior in other
reservoirs if it can be demonstrated that it will match
past history. While this model predicts that water will
channel and bypass oil, it assumes complete areal
coverage. In reservoirs containing areal heterogenei. .,
.- _.-:
..r ha
- enptacted
-w ..-... ties, only part or me rescrvulr iii~ii may
by water, and adjustments or conformance factors
will be needed to match reservoir behavior with the
model.

Gravity Segregation of Water

Injected into a Gas Cap
Physical Description of the Problem
One method of accelerating recovery of gas from
reservoirs where the gas cap and oil zone are associated is to inject wqter near the crest of the structure.
Although reservoir pressure can be maintained by
replacing oil and gas-cap gas with water, the danger
exists that water will flow down into the oil zone and
cause counterblow of oil into the gas cap, The problem is then twofold. First, how can the water be concentrated near the crest of the structure? Second, how
much water can be injected without endaiigeririg the
oil coiwnn1
The first problem can be dealt with directly. Water
injection wells should obviously be located high on
the structure parallel to the crest. Optimum spacing
would be approximately that where the bubbles of
water growing around each injection well join just as
injection ceases. The rate of water injection should
be greater than twice the critical rate given by

u.
++

Ap sin a
9..
- u.

~lcr

(11)

~gr

where A is the total cross-sectional area of the gas

~~t~,
Water
will fill the entire
---- A+ ,.,,k the +~ca]
~a~.
_i

77.4
86.1
90.9
93.7
95.8

The model predicts that water breakthrough wiii

have occurred in all wells when 5 percent of the displaceable oil is produced and that water cuts will
exceed 90 percent before 40 percent of the displaceable oil is produced.

., =,-l

().()44
(9/~)crit*c.l

0.2
0.3
0.4

Q.5

WATER

n--

. ..-

---

sand thickness. Also, its areal distribution around

each injection well will approach a circular shape.
(At very low rates, the water will underrun the gas
and sag dowdiP into a tear-drop shape because of
gravitational forces.) Solution of the second problem
requires knowledge of how much time would be required for .wa~er to flow by grav@ to the oil zone
after water injection ceases. The problem is one of
predicting the countercurrent flow of water downdip
and gas updip under the influence of gravity.
The Mathematical Model
After injection ceases, water tends to underrun the
gas as shown in Fig. 6. A simpie mathematical model
can be derived by assuming that the thicknesses
through which the gas and water flow are constant.
In the upper part of the thickness, which contains a
reduced water saturation, gas is displaced updip by
water flowing downward at the same rate by gravity
through a region of immobile residual gas saturation.
By assuming that capillary forces are negligibly small
and by applying Darcys law for each phase and
equating the flow rates, the following equation can
be derived.
a,./A

0.044 k LP sin ff
p.,,,
k<:; , + k,,,(l A J

(12)

If the area occupied by gas is such that q,t/A is a

maximum, then
A, =

~~w
l+~w

(13)

...

where M is the mobility ratio defined by

kJwc
~ =
.
kw/w,

..

(14)

Eq. 22 can be used to predict the rate of movement of tie water bank downdip once injection ceases.
This information. along with knowledge of the position of the water bank when injection ceased, permits
us to determine whether or not water will reach the
oil zone before it is depleted.
Example Problem
To illustrate the use of this model, consider the following problem.
Given:
L=
H=
w=
k=
+=

5,000 ft (distance from crest to gas-oii

contact)
20 ft
5,000 ft
0.1 darcy

0.20
s,.; = 25 percent
S,u = 20 percent

JOURN.Al_OF PETROLEUKf TFCHNOLOG}

k,.,.,, = 0.50
k,, = 1.0
0.35 Cp
p.lo

/Jg = 0.015 Cp
a = 6
6Q.4 ]b/cu ft
~, = 12.0 lb/cu ft
=

pt.

And water is injected at the crest until the water

front has moved one-half the distance toward the oil
column.
To find:
The time required for water to flow by gravity through
the remaining 2,500 ft to the gas-oil contact.
Solution:
~ -_ ~0.5/0.35

-tk AC have been

treatment in 1935, numerotis rnu.l.o-.
proposed for reducing production of unwanted gas
(or water) that cones into wells. For example, wells
are usually completed as low in the oil zone as possible, or even in the water zone, in efforts to reduce
gas coning. In other cases, wells are shut in temporarily to allow cones to heal. Ka~ et al. discussed a different approach in which impermeable,
horizontal barriers are placed (preferably just above
the production perforations) to reduce coning. In this
paper. we shall discuss injection of oil into the gas
cones to form a liquid barrier. This method, although
of limited practica! va!ue in most reservoirs, may have
application in a few reservoirs containing thick, permeable sands.
Physically, a liquid barrier can be generated as
shown in Fig. 8. The well is dually completed so

= (3.0214

1.0/0.013
A~ =

\/O.0214
_
1 + VO.0214

= 0.128

(0.044) (0.1) (o. 1045) (50.4)

= o 015/1 X 0.128 -r 0.35/0.5 X 0.872

= 0.02519 ft/D.

:y=ikd!ii

The time for water to advance the remaining 2,500

f~is

= (0.20) (2,500)(1

0.2

0.25) (0872)

o!

1.0

0.02519

w, -

= 9,519 days, or 26.1 years.

F@

ODSERVEDWATER CUT

2.0

3.0

4.0

4 20

5.:

DISPIACEAELEPORE VOLUMES

of predicted and actual

Comparison

bebatibr.

reservoir

Comparisons with Experimental Data

The validity of this mathematical model was checked
by comparisons with data from a sarid-jmcked model
scaled to the condition shown in Fig. 7. Water was
injected updip at about three times the critical rate
until the front had advanced one-half the length of
the model. Injection was stopped and the position of
the water bank was recorded at various times as it
segregated downdip. The scaled time required for
water to reach the outflow end was 79 years, which
compares well with a value of 60 years calculated by
the mathematical model. The clifferences between
calculated and observed results are in the direction
one would predict because unscaled capillary forces
,n tie ~xper;m,enta] mode! would maintain too high
a water saturation in the gas flow path and restrict
movement of gas updip. We conclude that the time
for segregation in a reservoir would be more closely
approximated
by results from the mathematical
model.

TOTAL i
AREA
=A

for gravity segregation of water.

Fig. &Model

CEASES

AS lNJ~N

1
%!#l?ER

b
o

(
K = o.1 DAR~

GAS AND
CONNATEWATER

16.5

Control of Coning
Physical Description of the Problem
When gas exists in sands above the oil producing
interval of a well, the pressure drawdown associated
with production of oil causes the gas-oil contact to
cone down into the producing interval. Since
Muskat and Wyckoff presented their analytical

16.5
!jOoo
o~
Fig. 7Shapes of water-gas fronts.
1151

that a liquid such as stock-tank oil can be injected

down the annulus thr:u~h perforations in the casing
into the gas cone. The inJected oil flows into the reservoir as shown and drains down. filling the cone to
foml a liquid barrier. This barrier impedes flow of
gas into the productive interval and permits production of more oil from the reservoir for a given volume
~f ~a~. This !iquid barrier increases the effective wellbore radius and controls coning. much like a fixed
horizontal barrier discussed by Karp et ol.
Muskat and Wyckoff gave the following equation
for approximating the maximum rate at which a well
can produce oil without coning. *
q. = 0.0246 ko Ap (h, h,) ,..
.

injection rates, q,, many times the net oil production

rate. From a practical viewpoint. the size of the barrier will usually be limited either by the distance
between perforations or by the rate at which liquid
is injected to form the barrier. Within the above constraints, it is frequently possible to approximate the
performance of a fluid barrier by assuming that it is
a fixed horizontal barrier having a radius equal to Ah
.,
h
rv=rfnratd
The
-~-...= .lle
ana located just duwvV
~wl..,----- ~n~rv~l.
corresponding oil production reference rate is then
calculated by substituting Ah of Fig. 8 into Eq. 15
in place of r,c:

(15)

in which q,, is the oil production rate in barrels per

day, When a circular horizontal barrier is present, the
radius of that barrier is substituted into Eq. 15 for r,.
and the distance from the barrier to the bottom of
the oil zone is substituted for Al,. (For anisotropic
sands, h,. can be replaced by h. ylkr.~kh.)
Conditions favorable for control of coning are (1)
thick. permeable sands containing low-viscosity oils
and (2) a barrier with as large a radius as possible.
The Mathematical Model
Laboratory model studies and computer calculations
suggest that the benefits from a fluid barrier can be
approximated by a fixed horizontal barrier placed a
short distance above the perforations in the oil zone
and having a radius of the same order but slightly
smaller than the fluid barrier. The radius of the fluid
barrier out to a radius greater than one or two times
the distance between the injection and production
h..) wi]l usually require fluid
P3rfOfZikHiS (A.4 ~ h, ..,,,
More accurate estimates of critical coning rates can be obtained
t al.= In
usxng the curves and procedures
developed
by Chancy
the current
application,
however,
Muskats
oldar equation
ia more
is included
cmvenient
to uae since the effect of wallbore diameter
explicitly.

Production rates with and without fluid injection are

then calculated using Eqs. 15 and 16.** The difference between these two rates can then be used as a
measure of the incremental production rate that
would be possible with oil injection. This increase
can rarely be achieved without coning some gas, but
the increased gas production should not be serious.
Example Problem
Given:
ko = 1 darcy
p. = 2.31 CP
54.3 lb/cu ft
#l. =
p,, = 5.73 lb/cu ft
h, = 32 ft
hp=8ft
Ah = 32 ft
r. = 1,024 ft
rw=lft
Oil is injected into the upper perforations as shown
in Fig. 8.
To find:
How much oil production rates can be increased
above the critical coning rate by injection of liquid.
Solution:
From Eq. 15, the limiting gas-free OHprcxhittiort rate
.L1 ..Wu,-c
..*hfis.* q...
Ii IIi A inipcti~ff
pcrssiule
..--------

&

~0 = (0.0246)(1 .0) (54.3 5.7) (32 82)

2.31 in ~
()
= 72.0 B/D.
By injecting oil, a liquid barrier having an effective
radius of about 32 ft can be formed. The corresponding production rate is now calculated to be 144 B/D,
or twice the previous rate. Thus, the model predicts
that a production increase of 72 B/D should be possible with only a moderate increase in the net GOR,
although it is not possible, using this simple model
alone, to predict how much the GOR will increase.
*.lf

kr~sA~
Fig. 8-Modei

1152

for coning control

by injection.

desired,

the more

accurate

method

of Chanaj

at sf.~

can be

used for the case with no iniectjon

and this critical
rate can be
multiplied
by the ratio On r./r_)/(ln
r./Ah)
tO obtain a closer estimate for the raferanca
rate with liquid injaction.

JOURNAL OF PETROLEUM TECHNOLOGY

Comparisons with Computer Results

The above example shows how the simple model may
whether !iquid injection is attracbe used to prdki
tive. This model can be used as a tool to screen
prospective projects. An actual project could best be
.-.
4-11.,find
optlmlzed by experlmetkwy
1,,. m=u tbe
. injection rate
produc~hat~ieid~ a ~inim-~m GOR for the desired
?~~~rate.
The simple model was checked by comparisons
with resuhs of computer calculations.:~ Computed
net GORS for the example problem are shown in
F& 9. With no injection. the GOR for a production
rate of 150 SlT3/D is 780 cu f{/bbl (so!utlon gas was
458 cu ft/bbl). This is in reasonable agreement ~th
the wa!ue of about 120 B/D obtained using the correlations of Chancy et al.,:nbut rougil~~ twice the rates
predicted in this case by using Eq. 15. Above 150
STB/D, the GOR increases rapidly. At a production
rate of 300 B/D, the ratio has increased to 13,630
cu ft/bbl. By injecting oil at a rate of 150 B/D, a
liquid barrier is formed and the ratio is reduced to
9,800 cu ft/bbl. When the injection rate is increased
to 600 B/D, the ratio drops to 6,200; but if the injection rate is increased to 800 B/D, the net GOR
also increases substantially because the higher injection rates tend to block influx of oil as well as gas.
Thus. an optimum injection rate exists for any given
net Oii pro~uc~on ~~tie. This optimum depends upon
the stage of depletion and the ratio of vertical permeability to horizontal permeability, as well as upon
local reservoir heterogeneities.
Obvious questions then are: Can a si.mific~t improvement be made in the performance of an actual
well that has clifferent sand- and oil-zone thickness,
viscosity, a different horizontal -to-vefiical permeability ratio? What is the optimum injection rate?
Although the capability exists for computing the benefits of liquid barriers on well performance and the
optimum rates for different reservoir situations, these
predictions are only as good as the input data, and
frequently the reservoir description is woefully inadequate. Although much of the needed data could
be obtained from the well by measuring GORS for
various oil production rates, it appears that a somewhat better approach would be to first screen candidates using the simple model and then determine the
optimum injection rate by field experiments.
Comparisons with Field Data
A test of fluid injection techniques to control gas
coning was conducted in a field in southwest Texas.
In the test well, the GOR had risen steadily over a
period of 6 years to more than 13,000 cu ft/bbl and
the daily net oil production rates had declined from
more than 60 to less than 7 B/D. For this case, the
..
..:..1 m~ei
predicted a higher lfiitsimple marnerndilual
,lS
ing rate of 16 B/D and a potential increase of about
10 B/D by injection of oil to create a liquid barrier.
In the test itseif, it was fomtd that an initial net oil
production rate of about 25 B/D with a net GOR
of about 3,000 could be realized by injetting between
70 and 80 bbl of stock-tank oil per day. Subsequently,
the net GOR increased slowly to above 4,000. The
-- .--=

,n-1,

benefits of fluid injection were demonstrated very

strikingly. when the test was stopped about a Year
later. In less than a month after injection was stopped,
the producing GOR ciimbed d20Ve 30,000.

Conclusions
It is concluded that reservoir flow problems can often
be analyzed readiiy by applying simple mathematical
models if only one force and one dimension are iri,.
km,, h fl@w
volved m the rate-hmltmg step, eveii tllv-g..
..
occurs in three dimensions. It is further concluded
that the mathematical models described in this paper
can be used to predict reservoir behavior if reservoir
conditions agree with the simplifying assumptions
. .
1--; ~ tbe
and if past behavior
used m develvPm6
.. - model
... .
can be matched. For these special situations, the following conclusions can be reached.
1. More than 80 percent of ti!e oil in place can
be recovered from reservoirs produced by gas drive
when conditions are ideal for gravity drainage of oil
vertically downward.
2. Early water production and long histories of
rising water production from reservoirs containing
viscous oils can be matched and predicted using
mathematical models that assume underrunning of
water.
3. Gravity segregation rates of water injected into
gas sands can be estimated adequately using simple
calculation procedures.
4. Caicu!ations can be made using a simple procedu re to determine if liquid injected into a we!! can
reduce coning.

Nomenclature
A
h
k
L
N
P
9

=
=
=
=
=
=
=

cross-sectional area normal to flow

thickness
permeability
length
volume of oil
pressure
flow rate

251

<
%
E

20 -

15 -

G
~
4

10

1 1

INJECTION
RATES
n 150 bbl/day
A 600

bbl/day

o 800

bbl/day

o~
o

L
100

200

NET OIL PRODLKTiON

Fig. 9Application

300

400

A
500

RATE - bb!/day

of liquid injection to control gas coning.

1153

S = saturation
t = time
= flow rate per unit of area
volume of water
&
x = fraction of the thickness
z = vertical distance moved by a saturation
a = angle of dip
P = viscosity
= density
$ = porosity
Subscripts
e =
s =
H =
1 =
0

P
r
T
v

=
=
=
=

influx (or effective in coning)

gas
horizontal
initial or injection
oil
production
relative or residual
total
vertical

References
1. Muskat. M.: Physical Principlesof oil Production McGraw-Hill Book Co., Inc., New York (1949).
2. Leverett, M. C.: Capillary Behavior in Porous Solids,
AINf~
( 1941) 142, 152-169.
?~~n~., ----3. Buckley, S. E. and Leverett, M. C.: MechanismOi Fhsi6
( 1942) 146, 107Displacemertt
in Sands, Trans., AIME
116.
---4. Hurst, W.: Unsteady Flow of Fluids in 011 Reservoirs,
Physics (Jan., 1934).
5. Muskat, M.: The Flow of Compressible Fluids through
Porous Media and Some Problems in Heat Conduction,
Physics (March, 1934).
6. van Everdingen, A. F. and Hurst, W.: The Application
of the Laplace Transformation to Flow Problems in Reservoirs, Trans., AIME ( 1949) 186, 305-324.
7. Welge, H. J.: A Simplified Method for Computing Oil
R..~r;8by
Gas or Water Drive, Trans., AIME (1952)
-.
8. Do;glas, J., Jr., Blair, P. M. and Wagner, R. J.: Calculation of Linear Walerilood Behavior Including the Effects
of Capillary Pressure, Trans., AIME ( 1958) 213,96-102.
9. Douglas, J., Jr.. Peaceman. D. W. and Rachford, H. H.,
Jr.: A Method for Calculating Multi-Dimensional Immiscible Displacement, Truns., AIME ( 1959) 21Q 297308.
10. Joslin, W. J.: Applying the Frontal Advance Equation
to Vertical Segregation Reservoirs, J. Pet. Tech. (Jan.,

1964) 87-94.

..C4

~a~~~e%, c S. and Lefkovits, H. C.: Gravity Drainage Performance .f Depletion-Type Reserweirs in the
Stripper Stage, Trans., AJME (1956) 207, 265-274.
12. Dictz. D. N.: A Theoretical Approach to the Problem

II.

of Encroaching and By-PassingEdge Water, Proc. Ken.

Neder. A dad. Wetenshaffen ( 1953) series B-56, 83.

13. Cardwell, W. T., Jr., and Parsons, R. L.: Gravity Drainage Theory, Tram., AIME ( 1949) 179, 199-215.

14. Corey, A. T., Rathjens, C. H., Henderson, J. H. and

Wyllie, M. R. J.: Three-Phase Relative Permeability,
Trans.. AIME (1956) 207, 349-351.
15. King, R. L., Stiles, J. H., Jr., and Waggoner, J. M.: A
Reservoir Study of the Hawkins Woodbute Field, paper
SPE 2972 presented at SPE 45th Annual Fall Meeting,
Houston, Oct. 4-7, 1970.
16. BlirEE&jrEnquist,
B. R., Jr., Good, L. W., and Wal A Reservoir Study of the Friendswood
Fie~d; J. #et.>ech. (June, 1971) 685-694.
17. Shreve, D. R, and Welch, L. W., Jr.: Gas Drive and
Gravity Drainage Analysis for pressure Maintenance
Operations, Trans., AIME (1956) 207, 136-143.
18. Martin, John C.: Reservoir Analysis for Rcasurc
Maintenance Operations Based on Complete Segregation
of Mobile Fluids, Trans., AIME ( 1958) 213, 220-227.
19. Cook, R. E.: Analysis of Gravity Segregation Performance During Natural Depletion, &c. Pet. Eng. J. (Sept.,
1962) 261-274.
20. Hall, H. N.: Predicting Gravity Drainage Performance
Using a Three-Dimensional Model, J. Per. Tech. (May.
1968) 517-524.
z 1, Terwil]igcr, p. L., Wilsey, L. E., Hall, H. N., Bridges,
P. M. and Morse, R. A.: Experimental and Theoreticrd
Investigation of Gravity Drainage Performance, Trans.,
AIME (. 1951). 19Z 285-296.
22. Elkins, L. F.: F~sieit~fi Fie!d -AmUnusual probl?m
of Bottom Water Coning and Volumetric Water Invasion
Etljciency, Trans., AIME ( 1959) 216, 130-137.
23. Muskat, M. and Wyckoff, R. D.: An Approximate
Theory of Water Coning in Oil Production, Trans.,
AIME (1935) 114, 144-163.
24. Karp, J. C., Lowe, D. K. and Maruaov, N.: Horizontal
Barriers for Controlling Water Coning, Trans., A3ME
( 1962) 22S, 783-790.
25. Chancy, P. E., Noble, M. D., Henson, W. L. and Ri%
J. D.: How to Perforate Your Well to Prevent Water
and Gas Coning, Oil and Gas J. (May 7, 1956) 55, 108.
26. Nolen, J. S. and Berry, D. W.: A Study of the Reliability of a Semi-Implicit Reservoir Simulator, paper
SPE 2981 presented at SPE 45th Annual Fall Meeting,
XPT
Houston, Oct. 4-7, 1970.
Original
manuscript
received in Society of Petroleum
Engineers
office Aug. 7, 1970. Revised
manuscri~t
recaived June 29, 1971.
at SPE 45th Annual
Fall Meating,
Paper (SPE 2928) was presanted
held in Houston,
Oct. 4-7, 1970. ~ Copyright
1971 American
lnstind petroleum Enginaare, Inc.
tue of Mining,
Metallu~ical,
This paper will
will cover 1971.

be

printed

JOURNAL

in

Traneectione

OF PETROLEUM

volume

251,

which

TECHNOLOGY