J. F. Wilson Member Aime Union Oil Co. of California Bi?Ea, Calif
J. F. Wilson Member Aime Union Oil Co. of California Bi?Ea, Calif
J. F. Wilson Member Aime Union Oil Co. of California Bi?Ea, Calif
dp=
PETROLEUM
TRANSACTIONS
+wr o@
4iif34
em -@
ABSTRACT
Steady-state flow theory, previously applied to disphrcenrents with two mobile phases, is extended fo cover downstructure flow involving three mobile phases: oil, gas and
water. When used with norinal reservoir fhiid properties,
the theory predicts the existence of four distinct flow
regimes and the conditions under which each regime
exists. The predicted behavioris verified by a parallel-plate
model study,
(e E)
<g(pap,
INTRODUCTION
Special problems exist when a water flood is contemplated in adipping reservoir having a substantial gascap.
Normally, the gas-cap region is one of very low oil saturation. If oil is moved into tfregas cap during the flood, a
loss in recovery results. Efficient recovery of the oil under-
lying thegasis therefore difficult.
In two reported field water floods, injection into the gas
cap has been used to solve these problems, At tire West
Norfolk Ciarr Sand Unit, injection into the up-structure
region of the reservoir was made for the stated purpose of
filling up the gas cap with water. The purpose in the
Sholem Alechem Fault Block A Sims Sand Unit was
to form a continuous water barrier between the gas cap
and oil band. A laboratory model study preceded. this
flood and was described in the same publication,
Both of these gas-cap water injections gave satisfactory
results, but the published data are insufficient to allow
generalizing to other field conditions. The present investigation was also undertaken to obtain results for a specific
field problem. However, these results will be shown to be
in agreement with an extension of the Dletz36mathematical model to three-phase flowing systems. The agreement
between the theory and the model studies indicates that
the mathematical model can be used to describe completely the flowing system under some conditions, and to
indicate the generaI type of flow to be expected under all
condhions.
FLOW THEORY
The Dletz equations for two.phase flow are derive&
for the case of constant-velocity linear flow in a homog-
(1)
-~=(,Y
(H)
+tana
,.
&J(P: - p,) Cosa
where subscripts 1 and 2 refer to the displaced
displacing phases, respectively.
tany=
(,,
and
--%
Or[ginal manuscript
received in Society of Petroleum Engineers
otlke Aug. 6, 1962. Revised manuscript received Nov. 2, 1962. Paner
presented at 37th Annual Fall kketing of sPE, Oct. 7.10, 1962, in
Las Axrgeles.
References given rt end of paper.
Disct&slon of this and all followlng techni&J DaDem Is invited. Dis.
cuss16n [n writ[ rig- (three copies) may be sent to the otlica of the
Jav477Lal of Petrpkwn
!lkchnologrf. Any diecusskm offered after Dec. s 1,
1062 ehordd he in the form of a new paper. No discussion should ezceed 10 Der cent of the manuscript heimx discussed,
DEcEMBER,
)sinrr,
FIG.
1962
2FLew MODEL,
THaIiE.PHASE
Fr.ow.
1sss
...
cotiguration
is assumed to be stable. That is, all flow
vectors are parallel to the top and bottom edges of the
section and are equal in magnitude. Along each interface
the pressure in one phase differs from that in the other
by the capillary pressure, a constant for each two-phase
system. Therefore, the pressure gradients taken along an
interface are equal in each of the two fluids. As in a
two-phase system:-a the pressure gradients are obtained
from Darcys law and then equated, giving an equation
identical with Eq. 2 for each interface.
When tan y becomes zero, the interface is parallel to
the top or bottom of the section and the displacement is
unstable, Setting tan y = O in Eq. 2, we can solve for a
critical velocity u=.
u==
(m-p)
/41 k
k,
sina
(3)
MODEL STUDY
k,
Property
,<g(p,.
p,)
sin a .
(lb)
(?+-)
For normal reservoir fluid properties, the critical velocity
will be somewhat higher than for the water-oil system.
The situation is different for the displacement of oil by
gas, Eq. 1 becomes
~
P.
() kO
P@ <g(p,
k.
pg) sina.
(It)
Prototype
Descrfpttan
Typical
section
Lmath
[ftl
1~*,, . .....
.
Medal
Flow channel betwmen
two Vertically.
oriented parallel
class plates
1990
Thlcknas$ift]
14
0.204
1
29
1
DIP
Permeablllty (dorcias)
Single Nuld
011 In Oil Zone
0.300
2.102
10!
0.1s1
2,102
10!
0.0251
2.102 x 104
0.181
2,102
10*
1 ,Wo
0.226
TABLE 2-FLUID
unfavorable so that ~
reservoir
PROPERTIES
Labaratc.rv 28t
Propartf
Re$erv61r
Density [sin/cc]
%$
.0591
water
011
. No. 2
.No. 3
No. 4
0.0012
:~4;
o.oot2
y;:
0.000s
,7525
0.ss41
.98S9
m.1817
1:1s17
0.01S5
4.73
te.59
0.035
4.73
18.59
0,255
O.OQ1O
0.255
0.0019
0.22s
0.0037
0.238
0.0069
0.254
9.253
0.249
0,249
Vi$co::g [Cp]
water
. No. 1
.0134
,a4
.50
0.8S7S
1.1s20
0,0155
.:;;4
Meb[~fyyR~tie
kwbg
.233
.0037
DmdW Dlff.
Ratio
(Pw.Po]/(Pw.Pg],254
JOURNAL
OF PETROLEUM
TECRNOLOCY
..
..-;
(p.
--
f%)
g km
~~~
(pm
Pa)
refined oil was used in the flow control system. The oil
displaced the inj?ction fluid into the model from a Plexiglass vessel. The flow control set-up is shown schematically in Fig. 3.
~ k.
1
u pm
u p.
or
(P.
p.)
gkto
,*~
U.
,._
u p.m
p, pg
A Buckley-Leverett calculation using the measured relative-permeability curve for the reservoir indicated, as
would be expected from the favorable mobility ratio, that
the displacement of oil by water proceeds with little nlovcment of oil behind the flood front. Thus the Dietz model,
which posttdrdes no oil flow behind the front, should be a
good approximation. The same can be assumed for the
displacement of gas by water. Although the displacement
of oil by gns would not be expected to. flt the Dietz
assumptions very closely, Hawthornes studies in a porousmedia model showed that the theory adequately describes
such systems,
WE \fol)EI,
A Hele-Shnw (or spaced-pl@e) type of model was
chosen for the study. In this model, flow takes place
between two closely spaced plates. To simulate a vertical
section, the plane through the two major dimensions of
the crack is oriented vertically.
Poiseuilles law for fluid flow between two parallel
plates may
.;==
be written
VI
12\L
Darcys law for flow in a porous medium is
;.
k
~
v 4J
(4)
(5)
Comparison of Eqs. 4 and 5 shows that they are completely analogous, and that the parallel plate model has a
permeability of (lt2/12) sq cm, or (1.013 X 10hz/12)
darcies.
The analogy has been appreciated for a long time and
has been utilized extensively by hydrologists. It has been
used for petroleum reservoir modeling only recently in
studies of water coning: gravity drainage: viscous ftngerin~ and displacements in vertical sections.~
The. flow channel of the model, 2.45-in, high x 14-ft
long, was formed by two pieces of %-in. plate glass separated at the top and bottom edges by 0.020-in. cellulose
acetate strips, C-clamps were spaced 3-in. apart along
the top and bottom edges to hold the assembly together.
Strips of cold-finish steel bar of rectangular cross-section
(% X ?4 in,) were placed between the clamps and the
glass to distribute the load. The clamps were adjusted to
give a spacing of 0.019 to 0.020 in., as measured in the
%-in. section Ieft between the outer edges of the spacer
strips and the outer edges of the glass. This outer groove
was then tilled with epoxy resin. Connections for flow into
and out of the model were made through Plexiglass end
fittings cemented to the model with epoxy resin. The completed model was set at a dip angle of 1.
From single-fluid flow tests made at both the minimum
and maximum pressure drops used in the waterflood
studies, the plate spacing was calculated by Eq. 4 as
0,0499 cm (0,0196 in.), Model properties are compared
with those of the prototype reservoir section in Table 1.
F1OWcontrol was achieved by using a controlleric to hold
a constant pressure drop across various sections of porous
media selected to give the desired flow rate. A 28-cp
tVSJIDS
Four different sets of fluids were used in the laboratory.
Their properties, together with the reservoir fluid properties, are listed in Table 2. The two waters were made
from water, glycerine, potassium iodide and Schiilings
blue food coloring. The three oils consisted of blends of
kerosene, light mineral oil, Kel F No. 1 oil and Schillings
red food coloring. The two gases were air and neon.
The sets were designed to give different gas nobilities
relative to the other fluids. Water-gas mobility ratios of
0.0010, 0.0019, 0.0037 and 0.0069 were used to find if is
reasonable degree of uncertainty in the gas mobility would
have a marked effect on results.
Exploratory runs showed that cleaning the model to is
completely uniform surface wettabllity was difficult. As a
result, irregularly shaped and nonreproducible interface
configurations were obtained, caused by differences in the
capillary pressure across the interface at various points in
the model. To minimize this probIem, a cleaning procedure was adopted designed to leave a low surfactant
concentration on the glass, and 500 ppm of the same
surfactant was used in both the oil and water phases. This
reduced the oil-water interracial tension to about 1 dyne/
cm and eliminated the irregular-interfiice problem:
RUN PKOCEINJRIi
The procedure for a run was as follows. The previously
cleaned and dried model was filled with oil, A portion of
the oil was replaced by gas to form a gas cap. The flow
control system was started and tluid allowed to flow
briefly into the burette shown in Fig, 3, Then both the
burette and pump were opened to the bottom of the model
inlet
causing
the contents
to be rapidly
purged
out
C.OTriton X-1OO
--
Ftc.
fitting,
the top fitting. The run was started with the simultaneous
closing of the line to the burette and the top inlet fitting,
and the opening of the line from the model outlet fitting to
the production graduate.
The run was recorded on 16-mm motion picture film
by a camera controlled by a device that r+owed variation
in the picture-takkg rate. From 40 to 100 ft of film was
taken for each run, Production-vs-time data were taken
to check the rate, and a clock was kept in camera focus
at all times to relate the observed behavior to time. Numbered markers l-ft apart along the length of model also
aided following the progress of the displacement.
3FLOW DIAGRAM.
1ss5
Individual frames from the movie of each run were projected and traced, Interface tilt angles were measured and
compared with the values calculated by Eq. 2. Run conditions are listed in Table 3, and the measured tilt angles
at various stages of the displacements are compared with
the calculated equilibrium values in Table 4, Three of the
four predicted flow regimes were studied experimentally.
Figs. 4, 5 and 6 are copies of the tracings included to
illustrate the flow behavior.
U <0.00074
TYPE 2 I)ISPLACEMENTS
One run was made at a flow rate greater than the critical
for the displacement of oil by water, but less than the
criticai for the displacement of gas by water, The behavior
is illustrated in Fig. 4, taken from the picture record of
Run 5. As predicted, the initially-horizontal gas-oil interface tiltg down-structure, butthe gascap is not displaced.
On the other hand, the oil is, moved from under the gas
with 100 percent vertical sweep efficiency.
DISCUSSION
For convenience in the subsequent discussion, the four
flow regimes defined by the critical velocities for the three
displacements will be referred to as Types 1, 2, 3 and 4,
in order of increasing rate,
TYPE 1 DISPLACEMENTS
Rates below the critical for the displacement of oil by
water were not studied for the three-phase system, The
general description of Type 1 flow given in the Flow
Theory secti~n. has been deduced from the nature of
Type 2 flow and from the behavior of slow down-structure
displacements with the gas cap absent.
Model
[cm/se.)
size
of
GypvVI
Pree:pe
0:34s
0.209
1,044
0.34s
0.209
0.34s
0,209
3
3
4
0.039
0.039
0.039
0.039
0.039
0.039
0.039
0.039
0.142
0.142
0.142
0.142
0.142
0<0062
0.0062
0.0334
0.0965
0.0015
0.02s7
0,0095
0.005s
0.02s7
0,0095
0.003s
0.0095
0,0058
TYPE 3 DISPLACENENTS
Five runs were made under conditions at which a[l three
displacements are stable, Scaled prototype rate, density
differences and thewater-oil mobility ratio were held substantially constant. The mobility of the gas re[ative to the
other two fluids and the size of the gas cap were varied.
Run data are shown in Tables 3 and 4, and typical flow
behavior is shown in Fig. 5 (Run 8).
Type
ef Flow
0.05s3
0.0577
0.312
0.900
yX:9
;
4
4
3
4
A
4
3
TASIE
Run No,
1
pv.
Y..
0.009
0.210
0.273
0.571
:.,5:
&
.*
**
..
**
7.3
6.4
6.7
6.3
6,0
6.6
1;.8
;::
6,6
6.3
179.6
179.7
!79.8
~;:j
:::
6,5
I79:7
179.s
4-RUN
0.044
0.110
0.2S4
0.434
0.573
:;9;
0.066
0.105
0.142
0.325
0.575
0,857
0.;2;
Pv*
0.182
0.226
0.512
:;s4;
0.130
0.2
0,8
Eq. 2
0.113
0.41s
:.3;
0.115
32.0
G. S. T.***
31.7
33.3
32.0
29.S
29.S
34,9
10
179.0
179,0
2.8
1.4
i::
0.143
0.4S6
:7s;
33.6
33,2
31,7
34,1
s.
T.
13.s
13.6
13.3
13.0
0.261
7.2
2:
4.9
Eq. 2
34,2
33.7
33.8
34.1
-0.07
0.141
0.381
Eq. 2
14.1
13,7
12.0
12
0.079
0.169
y;
I ,4
,1,5
1,7
I ,5
G. S. T.
-0.04
0.375
0.541
0.791
0,23s
0.506
::5:
13
~
G. B. 1.
0!500
0.s07
Eq. 2
11
G. B. T,
-63?
-63}
63.S
DATA
Run No.
s
2
2 flow occurs at
CONDITIONS
.%pdlclal Velocity
SeO. Table
Type
179.6
179,s
;;;,;
G. 8. T.
179.a
1So.o
180.3
179.s
sol
7,6
7.4
6.9
12,9
12.s
12.0
0.165
0,305
0.500
gG713;
G. S. T.
2.2
7.9
:::
I .5
1.5
,.
179.5
179.5
179.6
179.7
179.s
**OaS breakthrough.
\?Appreclable
JOURNAL
OF PETRGLEUM
TRC23NOL0CY
4 DISPLACEMENTS
A:
PORE
B:
0,21
VOLUMES
INJECTED
13
Ik
PORE
VOLUMES
Ii
0,57
PORE
.. . .
VOLUMES
INJECTEO
FLOW _
NOMENCLATURE
g = gravitational constant
k = plate spacing in FIele-Shaw model
k = permeability
u = superficial velocity (flow rate per unit area)
u. = critical velocity defined by Eq. 3
x = Cartesian co-ordinate in direction of flow
WATER ~
OIL=
as=
I
_.._.m___=..
Ik
C:
CONCLUSIONS
INJECTED
-==--+
A:
FIG.
~.
poRE
4-TYPE
VOLUMES
FLOW-RUN
INJEcTED
,tJ
0.29
VOLUMES
PORE
4
OIL
Fm.
c:
0,0s
PoRE
VOLUME
Ii
PORE
VOLUMES
DECEMBER,
19G2
Ii
INJECTEO
II
FLOW -
WATER m
,.
lNdECTEO
i?
D: 0,23
5-TYPE
INJECTEO
[2
FLOW
GAS m
-.. ,:,.;/.,:% ,,
.INJEOTEO
VOLUMES
13
.VOLUMES
PORE
0.04
INJECTED
II
0.73
INJECTEO
[
fi
C:
VOLUMES
13
12
PORE
PORE
I
S\
S:
5.
FLOWRUN
8.
OIL
OAs n
FIG.
&TvPE
WATER ~
~
4
FLOW-RUN
7.
1s87
!.
6,
7,
80
9,
A.:
Flooding
of Gas Cap in the Norfolk Garr
Jour, Pet. Tech. (Oct., 1958) 10, No. 10, 14,
H. G,, Craig,
F. F., Jr.
and
Wagner,
of
B56, 3&
Ertghedzg,
Second Ed., Mc.
Graw-Hill Book Co,, Inc., N, Y. (1958) 429-432,
Geertsma, J,, Croes, G. A. and Schwarz, N,: Theory of
Dimensionally
Scaled
Models of PetroIcum
ReservoirB,
Trans., AIME (1956) 207, 118.
Todd, D. K,: Ground IFater Hydrology, John Wiley & Sons,
Inc,, N. Y. (1959) 314-316.
Meyer, H. L and Searcy, D. F,: Analog Study of Water
Coning*, Jour Pet. Tech, (April, 1956) 8, No. 4, 61.
Hollingworth,
B. J. and Meyer, H. L: Two.Dimensional
G#t{9
Drainage Profilcs, Trans., Am, Geophys. Un. (1958)
.
Sa;man
and Taylor: The Penetration
of a Fluid into a
Porons Medium or Hele-Shaw Cell Containing a More Viscous Liquid, Pruc,, Royal Sot, ( 1958) A245, 312.
11, Chuoke,
Instability
tiacement
12.
REFERENCES
2.
Proldem
81.
H,
Unit,
~ater,
10.
Sand
Edge
ACKNOWLEDGMENT
1. Nelson,
and Bypassing
(1953) Proc,
1 = displaced fluid
2 = displacing fluid
~ = gas
o = oil
w = water
1}, N,:
Encroaching
Eretenschap
sUBSCRIPTS
l)ietz,
188,
Pirson, S, J, and Salvi, G, V,: %me Reservoir Applications
of the Hele.Shaw Fluid Model, Paper 801.G presented at
~N~5~;gimlal
Meeting, The U. of Texas, Austin, ( Feh, 14-
Vertical
Model,
***
---,
I
I
I
,
18s6
JOURNAL
..
.. .
. .. .
OF PETROLEUM
. .
..
TECHNOLOGY
..
.=
... .,