You are on page 1of 20

SPE

Society of PetroIeun Engineers of AIME

SPE 10733

A Simplified Predictive Model for Micellar-Polymer Flooding


by George W. Paul, Intercomp Resource Development & Engineering, Inc.; Larry W. Lake and Gary A.
Pope, University of Texas at Austin; and GenevieveB. Young, Intercomp Resource Development &
Engineering, Inc.

Members SPE-AIME

This paper was presented at the 1982 California Regional Meeting of the Society of Petroleum Engineers held in San Francisco, CA,
March 24-26, 1982. The material is subject to correction by the author. Permission to copy is restricted to an abstract of not more than
300 words. Write: 6200 N. Central Expwy., Dallas, TX 75206.

ABSTRACT process. 3 Oil rate and recovery algorithms were then


developed from theory and numerical simulation, and are
The purpose ,of this paper is to present a simple, reliable the subject of this paper. The full CFPM includes
and inexpensive predictive model which allows the eco- procedures for the estimation of field development costs
nomic assessment of micellar-polymer flooding. An oil and chemical requirements, a DCFROR calculation and
recovery algorithm is developed from theory and the other economic criteria. A study of the effects of
results of numerical simulation. The model contains reservoir and process variables on the economics of
correlations which relate factors impacting oil recovery chemical flooding will be the subject of a future paper.
to reservoir and process data: capillary number (perme-
ability, depth, spacing), heterogeneity (Dykstra-Parsons Numerical simulation was used to validate and to
coefficient), crossflow (k/k ), surfactant sorption (clay construct correlations for the CFPM. Micellar-polymer
h
fraction) and wettability (relative permeability). Oil flooding was simulated with a multicomponent, three-
breakthrough, peak oil rate and project life are estimated phase, three-dimensional, finite difference simulator."
from oil-wat'er fractional flow theory, augmented with an The simulations incorporated, among other things;
effective mobility ratio to represent heterogeneity. The oil-water-surfactant-salinity dependent equilibrium,
model is validated against the results of numerical simu- three-phase relative permeability, capillary pressure, and
lation and is compared with the results of several field compositional dependent fluid properties and chemical
projects. sorption.

INTRODUCl1ION In the sections that follow, we estimate the chemical


flood target oil (TO), or that portion of the remaining
The chemical flood predictive model (CFPM) is one of oil-in-place previously swept by water (Figure 1). The
a series of models developed for the U. S. Department of fraction of TO produced during a chemical flood is
Energy which are designed to identify reservoirs for defined as the overall recovery efficiency, ER . The
enhanced oil recovery processes. In the traditional recoverable oil is simply ER x TO. E is determmed as a
"binary screening" approach to EOR,1,2 reservoirs are function of field capillary number an~ sweep efficiencies
selected on the basis of oil saturation, viscosity, perme- for vertical heterogeneity and the mobility buffer. An oil
ability, etc. above or below an assigned value. With the rate vs. time curve, or production function, is estimated
predictive m:odels, however, the idea is to select reser- with a fractional flow treatment. The area under the
voirs based upon economic criteria with minimal technical production curve, which is by definition the recoverable
constraints, :;;upported by a physical model in which the oil, is used to estimate the peak oil rate.
technical parameters are in proper relationship to each
other. TARGET OIL CALCULATION

The CFPM was developed for sandstone reservoirs, and If we define TO as the oil remaining in the waters wept
only two te1chnical constraints were used - formation portion of the reservoir, and further reduce this by the
temperature and salinity (total dissolved solids). The fraction of the reservoir above bottom water (f ) and
below a gas cap (f )' then S bw
limits above which reservoirs were rejected, 230°F and gc
80,000 ppm TDS, respectively, will likely be relaxed with
new teChnology. For the CFPM, a screen on divalent ions
is probably more meaningful, but the data are less readily TO = [Sorw ][N p - OOIP (1 - Bo/BOf)]x
S .- S
available than TDS. 01 orw (1 - f ) (1 - f ) (1)
bw gc'
In the first phase of the work, available field and where we have assumed complete resaturation of the
laboratory data were assessed in order to identify the unswept portion of the reservoir, and that the areal sweep
critical quk!tities impacting the micellar-polymer of the chemical flood is the same as the preceding
water flood. The floodable pore volume, VP' follows from
References and illustrations at end of paper.

137
A SIMPLIFIED PREDICTIVE MODEL FOR MICELLAR-POL YMER FLOODING SPE 10733

VP = (TO) (Bof)/Sorw . (2) SURFACTANT RETENTION ON SANDSTONE

If data are unavailable for OOIP, NP and the B 's, TO may It is convenient to express surfactant retention in
be estimated from equation (2) wIth VP det~rmined by units of pore volumes of injected surfactant slug required
A¢h. to satisfy all adsorption. This quantity, denoted by D ,
may be obtained by equating the total surfactknt injecteSd
RATE AND CAPILLARY NUMBER to the total adsorbed:
.!...::...1- pa 1
With the steady-state rate equation given by Muskat 6 ( (jl- ) (~ PsC s 1000 '
(8)
for a 5-spot, and defining an injectivity coefficient, C~, by

t.P _ C' D/l-l (3)


where a is the surfactant retention in mg surfactant/g
11 = p 0 rock, an~ C is the volume fraction surfac~ant in the
injected slug~ The quantity V /D, where V I is the pore
ps s ps
it can be shown thae
volumes slug injected, is the dimensionless slug size. This
ratio is subject to optimization; for a rough! screen V
0.003541C'pkhD ~~~u:~ bbee Vma/~in:~:2 g~e1~~~r than Ds· Usually this turR~
q = j..l 0(5.58 + t In (A»
bbl/day, (4) ps s
The quantity a may be estimated from BI correlation
where A = area/injection well (spacing) in acres. between surfactan~ adsorption on sandstones with total
clay fraction, as shown on Figure 2. Only aI few of the
Recalling the definition of capillary number many data points used for the correlation are 'shown. The
original work 3 should be consulted for the details of the
N = Uj..l/cr (5)
procedures used. The lower line with equation a = 3.3 x
cap
wt. fro clay is recommended if no laboratory ~ata are
and using Parsons 8 results for median velocity in a 5-spot
available. If clay fraction is also not known, a = 0.4
mg/g may be used as an order of magnitude es~imate.
0.000055C'pkD The upper line on Figure 2 is an estimate ofl the actual,
u = ft/day, (6)
historical field retention. The field datal are higher
j..l 0 IX (5.58 + ! In (A» because they probably reflect other retention mechanisms
in addition to clay adsorption.
taking j..l '" j..l 0' and cr = 10 -3 clyne/cm, we obtain
7 OIL RECOVERY ALGORITHM
= (1.9 x 10- ) C'pkD (7)
N The tertiary oil recovery efficiency in the absence of
cap ..x (5.58 + ! In A) crossflow, E~, may be decomposed into displacement,
vertical sweep, and mobility buffer (polymer) sweep
ThUS, the estimation of both rate and capillary number efficiencies
depends on C'. If j..l = j..l, C' can be as high as 0.7
psia/ft if the iRjectors £re injectiRg at their fracture limit (9)
and the producers are completely pumped off, or as low as
0.27 psia/ft, if the producers have a water head. In the following paragraphs, methods are presented for
estimating the various efficiencies.
Using injection/Rroduction data for 1~ micellar-
polymer field tests, ,7 equation (4) was used to calculate Displacement Efficiency. The micrdscopic dis-
values for C'. The values ranged from 0.13 to 3.63 placement efficiency is defined as
psia/ft, with ~ average of 0.3 psi a/ft. Correspondingly,
C' = 0.3 is used as a default in the CFPM when other (10)
d!Ra are not available. Note that this agrees with the ED = (S orw - Sorc)/S orw '
prod\lcer water head limit mentioned above. However, where Soc is the oil saturation in zones swept by the
because the oil production function is sensitive to C' , micellar s~ug. Because ED is a microscopic efficiency, it
historical water flood data and practical experience gain~d is independent of sweep and is a function lof capillary
from familiarity with given reservoirs (fractures, con-
number via the capillary desaturation (CD) curve. ED is
dition of wells) should be used to estimate C' whenever
possible. p the "breakover" coreflood recovery achie~ed at suf-
ficiently large V /D; in this sense ED is similar to oil
Oil recovery, being a loga.rithmic function of N c ,is . h CpsO 2 s.mJec
. ti on a t and a boye
not as sensitive to C' as is rate. More importaH~ to
I •.
recovery WIt mmimum
recovery is interfaciJil tension, which may change by miscibility pressure.
orders of magnitude in situ. The one millidyne value for
interfacial tension was chosen for the N estimation on There are three methods by whiCh ED may be obtained
the grounds that it 1) represents what cft\~y be obtained for the CFPM: 1) Estimate N from equation (7) and use
with current technology, 2) provides a relative comparison a CD curve developed for tlfl'l-eservoir rock! in question.
between projects and allows one to address questions of However, CD curves are not generally available. 2)
rate and pattern size sensitivity, and 3) provides a signifi- Obtain ED directly from large V /D slug tests in field
cant correlation between oil recovery and N for the ps s I
aforementioned 16 field tests. 3 cap cores. 3) Compute N caD from equation (7)1 and use CD
curves for Berea roc\( such as the ones developed by
The procedures discussed above for 5-spot patterns Gupta-Trushenski 9 (Figure 3). In the CFPM, the influence
have also been developed for line-drives. 3 of wettability on ED is approximated by a li~ear interpo-

138
SPE 10733 G. W. PAUL, L. W. LAKE, G. A. POPE, and G. B. YOUNG

lation on, R = kO /ko where the kO,s are relative theory, a well-designed micellar-polymer flood will
rw ro' generate a clean oil bank of constant saturation Shand
permeability endpoints, with R = 0.1 (water-wet curve) fractional flow f h = f (S b). This bank will be driSen by
and R = 10.0 (oil-wet curve). a surfactant fron~naviRg g specific velocity
Vertical Sweep Efficiency. A procedure is given in the
Appendix I for the estimation of E as a function of 1
(15)
heterogeneity, through the Dykstra - ¥arsons coefficient 1 ° 1+D - S
s orc
(V DP ), and VpiDs. From equation (A-IS),
The specific· velocity is the actual front velocity divided
by the interstitial velocity. Equation (15) presumes no
(11)
great curvature in the low interfacial tension fractional
flow curve, an assumption that is probably accurate
enough for the CFPM. The velocity v can also be
where C and F are obtained from equations (A -11) and expressed in terms of the oil saturation Sand fractional
(A-19), r~spectiv%y. Plots of EV vs. V /Ds for various flow change at the rear of the oil bank
VDP are ~hown in Figure 4. ps
1 - f
Mobiliity Buffer Sweep Efficiency. EMB , defined' as wb
= I-S (16)
the volufli1e produced oil/volume mobilized- oil, would be S - S -S
ob orc orc wb
expected I to be a function of VMB' the pore volumes of
polymer injected, as well as VpiDs and VDp. Conse- Equations (15) and (16) give a relationship for the oil bank
saturation which must be solved simultaneously with the
quently, EMB = EMB (V MB' EV) since EV = EV (V pi D5' water-oil fractional flow curve f = f (S ). The solution
VDP). A function of the form follows graphicallyll as the inter~ecti~n d't a straight line
passi~g through the points fw = 0, Sw = - Ds and fw = 1,
S = 1 - S with the fractional flow curve (Figure 6).
w orc
Assuming S constant allows an expression for the
specific velo<m<ty of the oil bank front
where

E MBo = EMB I V (13) v


ob
= (17)
MB = 0

is proposed for EMB since it satisifies the limits where S . is the initial oil saturation. For tertiary
applicati8hs Soi = Sor and foi = O.
lim E!\I(B = 1.0, lim EMB .... 1.0.
The specific velocities in equations (15) and (17) lead
V .... 00 EV.... 0 to expressions for the dimensionless breakthrough time of
MB
the oil bank, t ' and surfactant, t Ds
Dob
The secohd limit, which follows from the observation of
numerical simulation results, reflects that for smaller
VpiDs' ~ence poorer EV' EMB tends towards unity (for a t Ds = l/vs . (18)

given VMs? because lesser amounts of oil are mobilized


which! reqUIre production. For homogeneous media the dimensionless production
I function is oil cut = foi unt.il tD = t Dob' oil cut = fob for
Evaluation of the constants a, Sand E B of equation
(12) was 'Performed using the results of ~wo<?.dimensional t Dob < tD < t Ds ' then oil cut = 0 for tD > t Dob • This
cross-section simulations without crossflow at various function is shown in Figure 7 a.
VMB ' VpiDs and VDP • With ED as determined from the
one-dimensional, large V /D analogs, and EV calculated Heterogeneous Media. The oil production function
I ps s
by a pro1cedure for finite layers, EMB was determined proposed here will be triangular, and based on the veloci-
from ties in equations (15) and (17) now modified to account for
layering. The triangular shape is selected because it can
(14) be described with only four variables: oil breakthrough
time, time of peak oil rate, peak oil rate, and time of
zero rate.
The E from equation (14) were found to correlate
reason&b~y well with equation (12) with a = 0.4, S = 1.2 From the Appendix, VDP is related to an effective
and E : = 0.71 - 0.6V DP ' as shown in Figure 5. The mobility ratio Me
MB
points wrth the largest deviation from the 45 line °
(perfect correlation) on Figure 5 are from the simulations VDP
run at very high VDp. 10g(M ) = 02 (19)
e (1 - V ).
DP
PRODUQTION FUNCTION
and the flow capacity F of the layered assembly is related
Homogeneous Media. According to fractional flow to the storage capacity C by

139
A SIMPLIFIED PREDICTIVE MODEL FOR MICELLAR-POL YMER FLOODING SPE 10733

where the total fluid injection rate q is from equation (4)


(20)
and VP is from equation (2). I
I
The procedure discussed above gives production
The specific bulk fluid velocity of the fastest layer is, functions that are reasonably consistent with experience.
therefore, The procedure could be easily modified to calculate a
production function for fields undergoing ~tagewise
v = F"C = 0 M (21)
e development.

Correction for Crossflow. Two-dimensional cross-


The velocity of the oil bank and surfactant fronts in section and three-dimensional numerical simUlations were
layered media relative to the overall average interstitial performed with variable VDP' V ID, and dimensionless
velocity are crossflow number ps s
v'
ob
= vobMe and v's=sv e
M (22)
RL
1

= (k/kh)' (L/H). (29)

The results are shown "in Figlfre 8. In these sibulations


from which follow the dimensionless breakthrough times
k/k varied from about 10 (R = 0.025) to about 0.6
of the oil bank and surfactant in layered media h L
(Rr = 40). Examination of the simulation results showed
thli't oil breakthrough
.
(tat» and sweepout (tsw ) tiines
' were
(23) generaIIy mdependent 0 R . Thus, as an appr0ximation
to the crossflow production runction with oil rec6very E~,
the low cross flow recovery is increased by tJ. ER , Where
(24) I

tJ.E R = E~ - E~ = 0.04 log R:L + 0.08 , (30)


It is reasonable to assume that the peak oil rate will
occur at t at which point the overall oil cut will be
Ds
for RL > O. Equation (26) with E~ replaced by E~ yields
f =f . F , (25) the peak oil rate. The crossflow production, function,
opk ob F' = M I then, is identical in timing to the no crossflow case,
e Vs vob except with a higher peak rate and area.
The last term on the right side of equation (25) expresses
the flow capacity of the reservoir at a flow capacity VALIDATION OF CFPM BY NUMERICAL SIMULATION
I
where the surfactant velocity in the highest permeability
The CFPM is compared with the results of three
layer (vs~. ) equals the oil bank velocity (vobF') in a lower simulations in Figures 9 - 11. The simulator inp).lt data is
permeaDlllty layer. For t Dob < tD < t Ds the oil cut is summarized in Table 1. With the exception of vertical
increasing linearly because successively more layers begin sweep efficiency, the CFPM performance cur~es were
producing oil. For tD > t Ds the oil cut falls linearly since determined by the procedures outlined above. ,For com-
successively more layers are producing surfactant or parison with the two four-layer simulations (Figures 9 and
10), Ev for the CFPM was dete~mined by a prodedure for
polymer. In equations (15) - (25) it is assumed that there
finite layers with M = k .jk. This finite lkyer pro-
will be no change in surfactant velocity when the polymer cedure is not the oneeprese'Rfea in Appendix A fbr infinite
front overtakes the surfactant fronts. This will happen layers. The latter procedure is more useful fori reservoir
predominantly in the lower permeability layers which will screening, since the detailed geologic data r~quired to
not contribute much to the initial portion of the pro-
define layer properties is usually not available. Even with
duction function. only four layers, the difference between the I two pro-
cedures is small and diminishes with the numberl of layers
The final point on the triangular production function,
the dimensionless time at zero oil rate, tD ,is selected simulated (Figure 12).
so that the total chemical flood oil recoJfry--the area Figure 10 shows the result of the cross flow correction
under the production function--agrees with that given by on CFPM performance. Note the symmetrical inature of
equation (9): the simulator derived curve in Figure 10, illustrating the
reduction in effective heterogeneity due to crosSflow.
(26)
For ~he two-dimensional areal case (V DP 1= 0), the
CFPM WIth M = 1.3 was found to reproduce Ithe simu-
The dimensionless production function is shown in Figure lation shown {fi Figure 11 as well as a number of simu-
7b. lations in which permeability, depth and spacing were
varied. The value of M greater than 1.0 in the areal case
The dimensionless production function may be reflects the reduced erecovery caused by a~eal con-
converted to a dimensional function (Figure 7c) of oil rate formance. In a cross-section, a value of M greater than
(stb/D) vs. time (days) by the following definitions: 1.0 reflects the reduced recovery due to everhcal con-
formance. I
I
qo = qf 0 (27)
The CFPM oil recovery and timing agrees I favorably
with the simulator. A more rigorous test of tM CFPM is
t = tD(V p/q) (28)
a comparison with field results.

140
SPE 10733 G. W. PAUL, L. W. LAKE, G. A. POPE, and G. B. YOUNG

COMPARISON OF CFPM WITH FIELD RESULTS NOMENCLATURE

The CfPM is compared with the Sloss Field test l2 - 11t active surfactant retention, mg/g rock
(Figure 1~), the Big MUddy;.ilot I5 ,16 (Figure 14), the = spacing, acres/injection well
North B~rbank Unit f,ilot 1 23 (Figure 15) and the = formation volume factor
Robinson ~19R project 2 ,25 (Figure 16). = storage capacity
= surfactant concentration, vol.fr.
For Sloss, the CFPM overestimates oil recovery, per- depth
haps due to productivity problems in the field. When = retardation factor for surfactant, surfac-
comparedl with Big Muddy the CFPM is low on recovery, tant sorption in pore volumes/pore volumes
probably because crossflow was not considered. For both injected
these tests, oil timing is predicted well within acceptable = micellar-polymer oil recovery efficiency
limits for I economic calculations. = microscopic displacement efficiency
vertical sweep efficiency
Figure 15 shows that CFPM performance is reasonably mobility buffer sweep efficiency
close to lthe performance of one of the four confined = fractional flow, phase flux/total flux
producers, at North Burbank (97-06) and another confined = fraction of reservoir above bottom water
producer, not shown. However, poor response for two of fraction of reservoir below gas cap
the centra.l producers resulted in the combined four-well = flow capacity
curve shoiwn. In the CFPM calculations, D was based on layer thickness
a laboratory measured surfactant retentionsof about 2000
total internal thickness
Ib/acre ft. Post pilot analysis indicated that retention permeability
may be as high as 9500 Ib/acre ft. at North Burbank. thickness averaged permeability
Were this higher retention used, the CFPM performance geometric mean permeability
would be hlUch closer to the lower curve on Figure 15. permeability above which 8.41% of
reservoir permeabilities fall
For 2li9R, the predicted ER (0.31) agrees well with the maximum layer permeability
field esti~ate of 0.27 - 0.33. Figure 16 shows that the
CFPM approximates the magnitude of peak oil rate and = oil relative permeability at residual water
project l~fe, but misses on peak rate location and oil saturation
breakthro~gh. There may be several reasons for this: 1) = water relative permeability at residual oil
Great un~ertainty in the retention and relative perme- saturation
ability data. 2) The simplified fractional flow treatment L = injector-producer distance
Lorenz coefficient
~e
in the CRPM may be a poor approximation for viscous,
relativelyl high oil content micellar slugs, for which a = effective mobility ratio
more spe1cific procedure is available. 11 3) The sym- n upper limit on partial sums in equations
metrical :character of the field curve, with a hetero- (A-I) and (A-2)
geneity factor of 0.62, may reflect the effects of high N total number of layers
vertical c~ossflow. N = capillary number
Ncap = ultimate water flood recovery
p
Consiqering the assumptions made in the development OOIP original oil-in-place
of the. CIj'PM, and the uncertainty of much of the data q steady-sta te production/injection rate
required for its application, the comparative results are QD dimensionless cumulative injection
good. In addition, the above comparisons indicate that = dimensionles crossflow number, equation
the CFPM might be used as a history matching or design RL
(29)
tool to p~ecede more costly, fully compositional simu- S = sa tura tion
lations. ~part from constructing a water-oil fractional t time
flow curVje, the oil rate and recovery algorithms in the = dimensionless time, or injected pore
CFPM m~y be executed in a few minutes with a hand tD
volumes of fluid
calcula tOIj • TO target oil
u = darcy velocity
CONCL USIONS dimensionless velocity for homogeneous
v
media
1. A slimplified screening model for chemical flooding -
has been ~eveloped which, in general, yields oil production v average interstitial velocity
v' dimensionless velocity for heterogeneous
functions !which overlay field test results.
media
2. Ba:sed on comparisons with the field floods V Dykstra-Parsons coefficient
DP
presentedl here and with other field results, the CFPM
VMB pore volumes graded mobility buffer
Vp floodable pore volume
tends to be optimistic for oil breakthrough and peak rate. pore volumes surfactant slug
However,' this is not altogether inappropriate for a pre- VPS
dictive mbdel. XD dimensionless longitudinal position

3. A ~ore rigorous approach is needed to model the a,S = constants in equation (12)
liP injector-producer pressure drop
effects oJi vertical crossflow. This should be done through = cumulative frequency, equation (A-5)
A
a reductibn in the effective mobility ratio, which would viscosity
directly impact vertical conformance and polymer sweep II
'J = variance of lognormal permeability distri-
efficiency. bution, equation (A-5)

141
I
A SIMPLIFIED PREDICTIVE MODEL FOR MICELLAR-POL ¥MER FLOODING SPE 10733

p density Flooding -Compositional Effects on Oil Dis-


a = interfacial tension placement," Soc. Pet. Eng. J. (April 1979) 116-128.
4> = porosity I
Subscril2ts 10. Dykstra, H. and Parsons, R. L.: ''The Preoiction of
f final (at initiation of chemical flood) Oil Recovery by Waterflood," Secondary IRecovery
i = initial (at discovery) of Oil in the United States, 2nd ed., AIPI (1950)
m = layer where XD = 1 at production well 160-174.
0 = oil
ob = oil bank 11. Pope, G. A.: ''The Application of Fractional Flow
orc = oil residual to chemical Theory to Enhanced Oil Recovery," Soc. Pet. Eng.
orw = oil residual to water J. (June 1980) 191-205. i
pk peak I
r = rock 12. Froning H R. and Treiber, L. E.: "Devel08ment and
s surfactant Selection of Chemical Systems for Miscible Water-
sw = sweepout flooding," paper SPE 5816 presented at Sl?E Fourth
w water Symposium on Improved Methods for Oil Recovery,
wb = water bank Tulsa, March 22-24, 1976.
SUl2erscril2ts
c cross flow 13. Basan, P. B.: McCaleb, J. A. and Buxtqn, T. S.:
0 = without crossflow "Important Geological Factors Affecting Ithe Sloss
= thickness weighted average Field Micellar Pilot Project," paper &PE 7047
presented at SPE Fifth Symposium on IImproved
ACKNOWLEDGEMENTS Methods for Oil Recovery, Tulsa, April 16-~9, 1978.
The authors wish to thank A. Goldburg, consultant to 14. Yanosik, J. L., Treiber, L. E., Myal, Fj. R. and
Gary Energy Co., and the following persons from Calvin, J. W.: "Sloss Mi~ellar Pilot: Project Design
Intercomp: J. E. McElhiney, H. S. Price, S. R. Aydelotte, and Performance," paper 7092 presented at SPE
T. L. Gould, G. C. Harris, M. B. Moranville and R. M. Fifth Symposium on Improved Method~ for Oil
Stearns; and, J. W. McGhee of DOE for data on the North Recovery, Tulsa, April 16-19, 1978. I
Burbank project.
15. Big Muddy Field Low Tension Demonstration
Project, First Annual Rel2ort, Al2ril 19178-March
REFERENCES 1979, DOE/SF!01424-13 (August 1979).
I

1. Lewin and Assocs. Inc.: The Potentials and Ec- 16. Enhanced Oil-Recovery Field Reports, Spciety of
onomics of Enhanced Oil Recovery, U. S. Federal Petroleum Engineers of AI ME (J anua~y 1975-
Energy Admin. Report B-75/221 (April 1976). December 1977).

2. Geffen, T. M.: "Oil Production to Expect From 17. North Burbank Unit Tertiary Recovery plilot Test,
Known Technology," Oil and Gas J. (May 7, 1973) Annual Rel2ort, BERC/TPR-76!2 (July 1976).
66-76. I
18. North Burbank Unit Tertiary Recovery Plilot Test,
3. Selection of Reservoirs Amenable to Micellar Second Annual Rel2ort, May 1976-May 1977,
Flooding, U. S. Department of Energy Report BERC/TPR-77!5 (August 1977).
DOE/BC/00048-20 (December 1980). I
19. North Burbank Unit Tertiary Recovery Piilot Test,
4. Intercomp Resource Development and Engineering, Third Annual Rel2ort, May 1977-May 1978,
Inc.: EI Dorado (High Water Content Process) BETC/TPR-78/8 (August 1978).
Micellar-Polymer Pilot, Report submitted to Gulf
Universities Research Consortium (May 1980). 20. North Burbank Unit Tertiary Recovery plilot Test,
Final Rel2ort, DOE!ET/13067-60 (June 1980).
5. Lake, L. W., Stock, L. G. and Lawson, J. B.: I
"Screening Estimation of Recovery Efficiency and 21. Trantham, J. C. and Clampitt, R. L.: I'Determi-
Chemical Requirements for Chemical Flooding," nation of Oil Saturation After Waterflooding in an
paper SPE 7069 presented at SPE Fifth Symposium Oil-Wet Reservoir-The North Burbank Unit Tract 97
on Improved Methods for Oil Recovery, Tulsa, April Project," paper SPE 5802 presented at SPE Fourth
16-19, 1978. Symposium on Improved Methods for Oil Recovery,
Tulsa, March 22-24, 1976. I

6. Muskat, M.: The Flow of Homogeneous Fluids


Through Porous Media, J. W. Edwards, Ind., Ann 22. Boneau, D. F. and Clampitt, R. L.: "A ~urfactant
Arbor, Mich. (1966). System for the Oil-Wet Sandstone of the North
Burbank Unit," J. Pet. Tech. (May 1977) 49~-500.
7. Lake, L. W. and Pope, G. A.: "Status of Micellar-
Polymer Field Tests," Petroleum Engineer 23. Trantham, J. C., Patterson, H. L. and B~neau, D.
(N ovember 1979) 38-60. F.: "The North Burbank Unit, Tract 97
Surfactant/Polymer Pilot - Operation anq Control,
8. Parsons, R. W.: "Velocities in Developed Five-Spot J. Pet. Tech. (July 1978) 1068-1074.
Patterns," J. Pet. Tech. (May 1974) 550.
24. Gogarty, W. B., Meabon, H. P. and Milton, H. W.,
9. Gupta, S. P. and Trushenski, S. P.: "Micellar Jr.: "Mobility Control Design for Miscible Type

142
SPE 10733 G. W. PAUL, L. W. LAKE, G. A. POPE, and G. B. YOUNG

Water Floods Using Micellar Solutions," J. Pet. heterogeneity to simplify the following mathematics and
Tech. (February 1970) 141-147. conform to prior definitions. This restriction is not
I severe as permeability variations are usually several
25. Howell, J. C., McAtee, R. W., Synder, W. D. and factors of ten larger than porosity variations. The result
Tonso, K. L.: ''Large-Scale Field Application of of this is that the porosities divide out from equations
Micellar-Polymer Flooding, J. Pet. Tech. (June (A-I) - (A-3).
1979) Q90-696.
If, as would be true if N + 00, the layer
26. Craig, IF. F., Jr.: The Reservoir Engineering Aspects thicknesses are all effectively equal, the layer assembly
of Wa1ierflooding, SPE Monograph Vol. 3, Society of can be characterized by the Dykstra-Parsons coef-
Petroleum Engineers of AIME, Dallas (1971). ficient 1 0 defined as

27. Aithison, J. and Brown, J. A. C.: The Lognormal


k-ko.Sltl
Distri~ution, Cambridge University Press, New York (A-4)
(1957).1 k
I
2S. Koval,' E. J.: "A Method for Predicting the Per- where k O. Sitl is the permeability above which S4.1 % of
formaljlce of Unstable Miscible Displacements in the distribution falls. We choose VDP as a measure of
Heterogeneous Media," Soc. Pet. Eng. J. (June 1963) heterogeneity because it is more common in reservoir
145-154. engineering.

29. Nelson, R. C. and Pope, G. A.: "Phase Relationships To relate F to C we assume that the permeability
in ChE;mical Flooding," Soc. Pet. Eng. J. (October assembly is log-normally distributed; hence, the relation-
1975) ~25-33S. ship between cumulative frequency A and permeability
2
APPENDIX is "

Representing Heterogeneity With An Effective Mobility A= i r1 - erf {In(klk) ~, (A-5)


Ratio. An elaborated definition of one of the traditional
measures ofi reservoir heterogeneity26 - the Lorenz coef-
L .tzV J
ficient - is Ibased upon the deviation of a plot of flow where k is the geometric or log-mean of the distribution,
capacity F W's. storage capacity C from a straight line, and v is the variance of the distribution. The relationship
where
between k and k is given by27
n N n
F = I
i=1
k.h·1 Ik.h. = I
k.h./Hk
1 l i=1 1 1 i=1 1 1
(A-I)
-k =. ""k e vl2 (A-6)

n N n
C =I <p! .h·1 I <p .h. =I <p .h./H~ (A-2) Hence if we identify A with the storage capacity C, using
i=1 ,1 l i =1 11 i=1 11 equations (A-3), (A-5) and (A-6) we obtain
I
In these equations k., h. and <p. are the permeability, In(ev 12 F') ]
porosity, and thickn~ss 10f layJr i; k and ~ are the C = i [ 1 - erf { 2v }, (A-7)
thickness-w~ighted average permeability and porosity; N
the total number of layers; H the total interval thickness
(layers are in parallel); and n the upper index of the where F' = dF IdC. Equation (A -7) may be solved for F'
partial sums, in the numerators. Equations (A-I) and (A-2) and then integrated subject to the boundary condition F =
define F and C as parametric functions of n. Letting n C = 0,
take on valJes from 1 to N, and plotting F vs. C at each n
gives the fl6w-storage capacity curve. The curve passes c
through F = IC = 0 and F = C = 1, consisting of straight line F = J exp{ - ~ + I2V erf-1 (1 - 2x) } dx. (A-S)
segments, aljld will have a discontinuously decreasing slope o
if the layers are arranged in order of decreasing k./<p .• Of
course, layers do not naturally occur this way, but 1f there Equation (A-S) must be integrated numerically to give the
·is no transtrerse flow of fluid, the naturally occurring F-C curve for fixed v. The results of such an integration
order isunifllPortant. It follow~irom the .shape of the are shown in figure A-I where the filled points are the
F-C curve that the slope of the n segment IS results of the integration and VDP is used instead of v. It
follows from equations (A-4) an<HA-5) that

dF (A-3) V = 1 - e-IV (A-9)


DP
de =
and, furthermore, that the relationshi~ among Lorenz and
Equation (A-3) states that the slope of the F-C curve at Dykstra-Parsons coefficients, and v is 7
any point n :is equal to the interstitial velocity in layer n
of a non-cross flowing, unit mobility ratio displacement. IV -In (1 - V DP )
In the limit 'of large N, equations (A-I) and (A -2) become L = erf(T) = erf ( 2 ) (A-10)
integrals arid dF/dC in equation (A-3) becomes a con- c
tinuous funcltion of C.
Note from equation (A-10) that L and VDP are bounded
I. while v is not. c
From thIS point on we consider only permeability

143
A SIMPLIFIED PREDICTIVE MODEL FOR MICELLAR-POL YMER FLOODING SPE 10733

Given the three heterogeneity measures in equation (2) there is no dispersion of the micellar'slug;
(A-10) it must seem odd to propose a fourth, but none of
the measures discussed so far directly relate to porous (3) the mobility ratio of all fronts is unity and there
media flow. To ameliorate this we propose an effective is no crossflow; and, I
mobility ratio M as a fourth measure of heterogeneity.
Me is defined by e (4) the oil saturation is reduced to S J everywhere
there has been surfactant present. or9
F = (1 + ~ (1 ~ C ) f 1 , (A-ll) Corresponding to these assumptions, the mobilized oil
e fraction is liS IS in any layer receivin~ a slug size
greater than D~, Sid" somewhat smaller if its slug size is
where equation (A-ll) follows from observing the pointed
similarity between a homogeneous media fractional flow less than D. If V is the overall slug size injected into
S
curve having straight line relative permeabilties and zero the layered reserv6fl., the slug size into any ldyer is F'V •
I ps
residual phase saturations, and the points generated on Now the surfactant front travels with relative velocity
figure A -1. In fact, the solid lines on figure A-I are F'/(1 + D ) and the mobility buffer with velocity F'. At
calculated from equation (A-ll) with M adjusted to give some dimsensionless position X the mobility Ibuffer front
the best least square fit to the calculat~d points. Hence, D
will overtake the surfactant front
there is a unique correspondence between V and Me
DP
which is shown in figure A-2 as the filled points. It
follows from equations (A-B) and (A-ll) that M + co as xD -- F'
I+D QD = F'Q D - F'Vps· (A-IS)
e S I
VDP +1 (infinitely heterogeneous) and Me +1 as V +0. Eliminating QD' the cumulative injection in pore volumes,
DP
In between these limits the relation between VDP and M ~~ I

is given by the following empirical fit to the points 011


figure A-2
X = I+D -F'V . (A-16)
D sps
VDP
10g(M ) = ----==__,,..~ (A-12)
e (1 - VDP)0.2 Now there will always be a layer m where the mobility
buffer front will overtake the surfactant fro~t exactly at
which is also shown on figure A-2. the production well (X = 1)
D

The Me - VDP relationship could also follow directly


from equations (A-3) and (A-4) as ~_ 1 (A-:17)
D - F'
s m
dF
VDP =1 - (dC)C = 0.B41 (A-13) Equation (A-17) says that the VpsID s ratio will be equal
to the reciprocal slope of the F-C curve at a point m
Equation (A-13) leads to a quadratic polynomial in M which separates the reservoir into layers .whose mobilized
whose solution relates only two points on the F-C curv~ oil fraction is liS IS ,F' > P' , and l~yers whose
whereas equation (A-12) is a fit involving several points. o orw - m I

It should be noted that the figure A-2 curve is an mobilized oil fraction is less than this, F' ~ F' m. . This
expanded version of the curve presented by Koval 28 which means that the overall mobilized oil fraction i~
is also shown.

Using M and equation (A-ll) is'an improvement over II So V I

equation (Ae-B) and VDP because the interstitial fluid + ~ (1 - F ) }(A-IB)


= Sorw { Cm D 1m
s
velocity in any layer is gIVen by
completely partially
swept swep~
- kX
v = -V F' = v ~ (A-14)

where C is the x-coordinate of the point! defined by


equation (A-17). Using equations (A-ll) an~ (A-14) the
from equations (A-3) and (A-ll) (we have reintroduced y-coordinate of the layer m may be solved fori directly
the porosity heterogeneity). The velo<;!.ty of the highest
permeability layer is, therefor~, v = v M ; that of the D
lowest permeability layer is v = v/Me. e
(Me (_s_ ))t M
~
V e
Qs
Estimating EV For Infinite Layer Case In estimating EV
we present here a brief summary of earlier workS using
1- M
e
'D
s
-< Me
the heterogeneity model of the previous section. Fm (A-19)
V
It is assumed that 1, ~ > M
D el
s
(1) the movable oil saturation lIS o = Sorw - Sorc
and the frontal advance lag D (surfactant adsorption in where the second part of equation (A-19) hoHls only if all
pore volumes) are uniform; s the layers are completely swept by surfactantL

144
TABLE 1

SIMULATOR INPUT DATA

Pattern spacing/type 40 Acre/5-Spot


Depth 5000 ft
Porosity 0.215
Thickness 18 ft
Ave. Permeability 250 md
kma/k 2.16, 1.0
Dykstra-Parsons Coefficient 0.6, 0
k/kh 0, 0.6
Surfactant Sorption (D s) 0.082
Cation Exchange Capacity 0.004 meq/g
Phase Behavior Reference 29
Surfactant Concentration, vol. fro 0.05
Surfactant Slug Viscosity 9.5 cp
Oil Viscosity 3.8 cp
Brine Viscosity 0.6 cp
Residual Oil Saturaton 0.3
Connate Water Saturation 0.3
k O
/k o 0.15 (water wet)
rw ro
Mobility Ratios
Surfacant Slug/Oil-Water Bank 1.0
1st Polymer Slug/Surfactant Slug 1.0
REMAINING OIL IN PLACE

TARGET OIL

MMBBLS
ROIP
TO
RECOVERABLE OIL

RO=
ERX TO

Fig. 1 - Chemical flood oil recovery

1.5~----------~------------~-----------r------------.---------~

• LAB DATA
o FIELD DATA
z
o
~ 1.0 o
IJ.I
t- FIELD RETENTION
IJ.I
.0::
IJ.I
t-
<{
Z
o
I.L
3 0 .5 o
CJ)

IJ.I
>
t-
U
<{
8 o

O~~ ______- L__________L -________- L_ _ _ _ _ _ _ _ _ _ ~ ______ ~

o 0.05 0.10 0.15 0.20


WEIGHT FRACTION CLAY

Fig. 2 - Surfactant retention on sandstone


1.0.-------------------r-------------------r-------~--------~

0.8

WATER WET

u 0.6
....
0
en ....~
I

....~ en
0
0
en
II
0 0.4
ILl

0.2

O~------------------~------------------~------------------~
4
10- 10- 1
CAPILLARY NUMBER

Fig. 3 - Capillary desaturation for berea rock

1.0 I----------:-:--=--T-:::::::==::::=---~--,--__=====:::::J=====--~--.,--I

0.8

0.6

Ev
0.4

0.2

1.0 2.0 3.0 4.0


Vps/Ds

Fig. 4 - Effect of slug size/retention ratio on vertical sweep


1.0r-----..,.----"T"""----,-----,----..,.

0.8

EMS
(SIMULATOR)

o 0.2 0.4 0.6 0.8 1.0


EMS (ESTIMATED)

Fig. 5 - Mobility buffer sweep efficiency cross plot

1.0 . - - -__r-------,-----r--,.----.------.._......

0.8

0.6

fw
0.4

0.2

o 0.2 0.4 0.6 0.8 1.0

Fig. 6 - Fractional flow diagram


A. Ho.Mo.GENEo.US
0..50.

to o.25t-

t-

I I I I I I I I I
o.
0. 0..2 0..4 0..6 0..8 1.0
to

B. HETERo.GENEo.US
0..50.

to 0..25

a~~~--~--~--~~--~--~--~--~~~~~
a 0..2 0..4 0..6 0..8 La

C. ACTUAL
16

12

Qo
8
MiSTB/O

a
0. 2 3 4 5 6
TIME,YEARS

Fig. 7 - Production functions


4 LAYERS
o 2DXS SIMULATION
Vps/Ds = 1.21
VOp=0.6 • 30 SIMULATION
0.8

0.6

o
e .......
0.4 o o

0.2

0.1 1.0 10.0 100.0


CROSSFLOW NUMBER, R L

Fig. 8 -"Effect of crossflow on simulated oil recovery

300r---------------~--------------~----------------~--------------~
30-4 LAYERS
Vop=0.6 SIMULATION
Vps/Os= 1.21 CFPM
RL = 0.025
200

qo,
bbl/d /
/
100 /

O~--------~L---L- ______ ~ ______ ~ ______________ ~ ________ ~~ __ ~

o 1.0 2.0 3.0 4.0


TIME,yr.

Fig. 9 - Production functiOns for low crossflow simulation


400~------------~--------------~---------------r--------------,

3D -4 LAYERS
VOp=0.6 - - - - SIMULATION
Vps/Os = 1.21 - - - - CFPM, RL =0.025
300 - - - CFPM, RL = 40
R L =40

qo .
bbl/d 200
I

10Q

Q~--------~~~--------------~--------------~--------~~--~
o 1.0 2.0 3.0 4.0
TIME,yr

Fig. 10 - Production function for high crossflow simulation

400r--------,---------.---------.---------r--------~

20 AREAL
Vop=O
Vps/Os= 1.21
300

- SIMULATION

~~~id 200 --- CFPM


\
\
100
\
\
\
0
0 1.0 2.0 3.0 4.0 5.0
TIME, y r

Fig. 11 - Production function for areal simulation


1.0

0.8

0.6

CFPM (INFINITE LAYERS) - - -


SIMULATION: 4 LAYERS ~ ~
0.4 12 LAYERS • •

0.2

OL---------~--------~--------~--------~~--------~--------~----~
o 0.5 1.0 1.5 2.0 2.5 3.0
Vps/Ds

Fig. 12 - Vertical sweep efficiency for finite and infinite layers


70
1\
I \
" PREDICTED AND OBSERVED Oil
I \ PRODUCTION FUNCTIONS FOR
\ SLOSS PilOT
60 I \
I
I
50 I \
I \
\
qo' OBSERVED--
BBl/D 40 I \ CFPM---
I
30 I
I
I
20
I
I
10 I
I
I
0
0 100 200 300 400 500 600 700
TIME, DAYS

Fig. 13 - Production functions for Sloss Field Test


5or------.-------r------.------.-------r------~----~

PREDICTED AND OBSERVED OIL


40 PRODUCTION FUNCTIONS FOR
BIG MUDDY PILOT

OBSERVED - -
30
CFPM - - -
f'
I
20 I
I
I
I
10 I
I
I
f
°0~~~~~-~-7~---6~0~0~--8~0-0~--~10~0-0---,2~0-0----J,400
TIME, DAYS

Fig. 14 - Production functions for Big Muddy Pilot

50~----~-----~-----~----~-----_r----~---~---~----__,

PREDICTED AND OBSERVED OIL


PRODUCTION FUNCTIONS FOR
40 NORTH BURBANK PILOT

.....- OBSERVED, WELL 97-06


- - - OBSERVED, 4 CONFINED
30 PRODUCERS
---CFPM

20

" "- "-


"-
"-
O~--~~------~------~-----~--~~------~------~~----~----~
o 200 400 600 800 1000 1200
"
1400
;-...
1600 1800
TIME, DAYS

Fig. 15 - Production functions for No. Burbank Unit


1000~----~~----~------~------~------~-------.-------r-------r------,

PREDICTED AND OBSERVED OIL


800 PRODUCTION FUNCTIONS FOR
219 R PROJECT

600 OBSERVED - - - -
CFPM---

4qO

200

0L-.L::._--'-_ _ _...1....-_ _--L._ _ _...L.-_ _---L_ _ _ ..l..._~___1 _ _ _......L..._ _____I

o 200 400 600 800 1000 1200 1400 1600 1800


TIME,DAYS

Fig. 16 - Production function for 219R project


1.0 r-----,r-----,-----y--~~~!!II"I

0.8

0.6

0.4 VOP Me

• 0.25 1.71
• 0.50 3.41
• 0.75 10.0

OL-----~------L-----~------~----_J
o 0.2 0.4 0.6 0.8 1.0
c
Fig. A-1 - Flow capacity-storage capacity curves

18.0

16.0

14.0
• CALCULATED

12.0

10.0
Me
8.0 VOP
LOG(M e ) = --..::..:.....-
(1- V op )O.2

6.0

4.0

KOVAL'S FIG.20
2.0

0
0 0.2 0.4 0.6 0.8
VOP
Fig. A-2 - Relation between effective mobility ratio and heterogeneity

You might also like