Sunteți pe pagina 1din 6

Effect of Solvent Viscosity

on Miscible Flooding
B.L. O'Steen, Westinghouse Savannah River Co., and E.T.S. Huang, SPE, Unocal Corp.

Summary. This paper presents experimental results for tertiary miscible displacements in strongly water-wet Berea cores. Floods
with both continuous solvent and simultaneous solvent/water injection were studied at ambient conditions. Three solvent/oil systems
are compared: heptane (0.4 cp) displacing mineral oil (6 cp), tagged heptane displacing heptane, and tagged mineral oil displacing
an equal-viscosity mineral oil (6 cp). Saturation profiles from microwave absorption measurements and production histories are present-
ed. The experimental results show that the manner in which residual oil is mobilized and propagated depends on solvent/water injection
ratio, oleic/water-phase viscosity ratio, and fractional-flow hysteresis.

Introduction
The miscible displacement of residual oil is often carried out by 1 % p-xylene (Table 1), and Solvent 2 was heptane tagged with 1 %
the injection of an alternating sequence of solvent and water slugs. nonane. The first type of displacement test was with an equal-
The motivation behind this practice is to reduce solvent mobility viscosity fluid system-i.e., tagged mineral oil displacing mineral
and thus to stabilize the miscible displacement. Numerical simula- oil or tagged heptane displacing heptane. The second type of dis-
tion of simultaneous injection of solvent and water indicates that placement test was with heptane displacing the 6-cp mineral oil.
miscible instability can indeed be controlled in this manner. 1 Un- For the equal-viscosity. displacement tests, the fraction of solvent
fortunately, this practice has some drawbacks. The injection of water in the effluent cuts was measured by gas chromatography (GC).
can also block direct oil/solvent contact, resulting in low oil recov- For the unequal-viscosity tests, the heptane fraction was determined
ery.2-5 In addition, some ofthe oil may exist as a dead-end or den- by refractive index measurement and GC.
dritic fraction; i.e., it is connected to the flowing oleic phase but
does not flow. This oil should be recoverable through a diffusive Experimental Procedure
mechanism, with the recovery rate dependent on an appropriate The Berea core was first saturated with 2 % KCl brine. Then the
mass-transfer coefficient. 5-7 core was flooded with the 6-cp mineral oil to a connate water satu-
A previous paper 7 presented displacement results for an octane ration, Siw, of about 35 %. Subsequent waterflooding resulted in
(O.4-cp solvent) 1 /mineral-oil (12-cp) system that demonstrated the a residual mineral oil saturation of about 36 %. A simultaneous
effect of dendritic oil on production history. A model of miscible water/solvent injection was then initiated at a typical injection rate
displacement, incorporating both trapped and dendritic fractions, of 50 cm 3 /h. Produced cuts were measured volumetrically in 5-
also was developed, and a mechanistic study of the experimental to 1O-cm 3 increments with an accuracy of about 2 %. Injection con-
results was performed. A major assumption of this model was that tinued until the fraction of mineral oil in the effluent cuts was < 1%.
both the trapped and dendritic fractions were functions of only the The core was flushed with the mineral oil until all the solvent in
oleic-phase saturation. Agreement between this model and the ex- the core was removed. Then the core was waterflooded to residual
perimental production histories appeared to be good. However, only oil saturation (ROS). At this point, the core was ready for another
two floods at 1: 1 and 2: 1 water/solvent injection ratios were displacement experiment. The scanning microwave apparatus was
modeled for the water-wet system. Miscible instability seemed to used to measure the ROS before solvent injection began, and satu-
be present in these two floods and was· accounted for with the ration profiles were measured at least three or four times during
Koval 8 method. This result is in contrast to 2D simulation results, 1 the displacement experiments.
which suggest that miscible displacements at these injection ratios For the runs with the tagged mineral oil as a solvent, the water/sol-
should be stable. vent ratio was varied from 0: 1 to 9: 1. For the runs with heptane
In this paper, we explore a broader range of water/solvent injec- as a solvent (displacing 6-cpoil), the ratio was varied from 0: 1
tion ratios and also examine the effect of solvent viscosity on ter- to 2 : 1. For the heptane-displacing-heptane tests, the coreflooding
tiary miscible displacements in a water-wet system. Results are procedure was slightly modified. When a miscible displacement
presented for miscible displacements with a high-viscosity (6-cp) experiment was completed, the core was always flooded with the
solvent displacing mineral oil (6 cp) and heptane displacing hep- mineral oil to establish Siw and to remove the residual solvent left
tane. Additional displacements also were run with heptane (0.4 cp) in the core from the previous run. The core (at Siw) was then
displacing the 6-cp oil. These sets of results are compared directly flooded with heptane to displace mineral oil and waterflooded be-
through their production histories and saturation profiles. fore a water/tagged-heptane injection test began. ROS achieved in
this manner were about 32 %.
Experimental Apparatus and Fluids
The coreflood equipment used in this study and the scanning micro" Experimental Results
wave adsorption apparatus for saturation distribution measurements Solvent/Oil Viscosity Ratio=l, (6 cp/6 cp and 0.4 cp/O.4 cp).
are described in Ref. 9. The microwave beam measures water satu- Four displacements were performed with the oil and solvent both
ration averaged over a I-in. -diameter spot. Because of the narrow having viscosities of about 6 cp (Runs EV 0 : 1, EV 1 : 1, EV 2 : 1,
core (1.8 in. wide), only a 10 saturation distribution could be meas- and EV 9: 1). Three displacements with oil and solvent having vis-
ured. All the displacement experiments were performed at ambient cosities of 0.4 cp also were run (Runs EV 1 : 9, C7, EV 1 : 3, C7,
conditions. The Berea core was a strongly water-wet sandstone slab and EV 1: 1, C7). The solvent and oil were distinguished in the
with a permeability of 1,100 md. The core was 42.5 X 1.8 xO. 7 in. produced oleic phase by the addition of tracers. Table 2 gives the
(Table 1). The oil used in this study was a mixture of an ll-cp injected solvent fraction, injection rate, oil recovery, and the aver-
mineral oil (67%) and carbon tetrachloride (33%). The viscosity age initial saturation, SOT' and final saturation, Soss' These satura-
and density of the oil were 6 cp and 0.98 g/cm 3 respectively (Ta- tions were measured with the microwave apparatus. Fig. 1 shows
ble 1). the corresponding production histories and other histories discussed
Two types of displacement tests were performed and two types later. Runs with similar water-alternating-gas <:WAG) ratios are ar-
of solvents were used. Solvent 1 was the mineral oil tagged with ranged on the same column, and those with the same fluid system
are on the same row to facilitate cross-referencing and comparison
Copyright 1992 Society of Petroleum Engineers of different runs.

SPE Reservoir Engineering, May 1992 213


The 1 :9,1: 3, and 1: 1 heptane-displacing-heptane results (Figs.
TABLE 1-BEREA CORE AND FLUIDS USED Ie through Ig) are different from the mineral oil floods. While these
displacements are quite efficient, with little delayed oil production,
Berea Core
only 19%,16%, and 12% of the ROS were recovered at the sol-
Length. in. 42.5 vent breakthrough (Table 2). Overall recovery was 61 %,52%, and
Width. in. 1.8
34 % of the ROS. These results appear to be in agreement with other
Thickness. in. 0.7
low-viscosity solvent floods-Le., significant oil trapping occurred.
PV. cm 3 205
Porosity 0.25 While the 1: 1 steady-state oleic-phase saturation was only 0.378,
k. md 1.100 the ROS was also low at 0.317.
krw. at 80' 0.082 Oleic-phase saturation profiles were determined by microwave
8 or . 0.35 absorption at several different times during each flood. All exhibited
kro. at 81w 0.71 the·"oil humps" and/or oscillations reported in Refs. 9 and 10.
8;w 0.35 An example of these saturation distributions is shown in Fig. 2a
for the 2: 1 mineral oil flood (Run EV 2: 1).
Fluids Used
Mineral oil Solvent/Oil Viscosity Ratio < 1, (0.4 cp/6 cp). Many displace-
Composition 67% 11-cp mineral oil. ments were performed with heptane (O.4-cp solvent) displacing the
33% CCI 4 6-cp mineral oil. However, the results of only five of these floods
Density. g/cm 3 0.98
are presented here. Table 3 gives the injected solvent fraction, in-
Viscosity, cp 6
Solvent 1-tagged mineral oil jection rate, oil recovery, and average initial and final saturations.
Composition 66% 11 cp mineral oil, Production histories for the heptane floods are shown in Figs. Ih
33% CCI 4 , 1% p-xylene through lk.
Solvent 2-heptane tagged with 1% C g The 0: 1 and 1: 3 displacements (Figs. Ih and Ii) were quite ef-
Solvent detection methods ficient, with virtually all the recovered oil produced by 1 PV. The
p-xylene by GC 0: 1 displacement recovered all the ROS, while the I: 3 flood left
C 7 by refractive index 6% of the ROS trapped. This efficient oil recovery is similar to
C 7• <::5 C g by GC that seen in the 0: I and 1 : I mineral oil (solvent) floods (Figs. la
and Ib). The I: 1 and 2: I heptane floods behave differently (Figs.
The results for the mineral oil floods (Figs. la through Id) dem- Ij and Ik). Here, the oil recovery is still far from complete at 2
onstrate that the ability of a viscous solvent to bank and to displace PV. In both instances, a significant fraction of the residual oil is
residual oil efficiently in a miscible flood is excellent for the floods efficiently displaced (52 % and 32 %, Runs LV 1: 1 and LV 2: 1
with 0: I and 1 : 1 water/solvent injection ratios, but that this abili- in Table 3), but following this bank, oil production rapidly
ty degrades significantly as this ratio increases. Note that the I : 1 decreases. A small and slowly decreasing flow of oil continues un-
flood recovered 48 % of the ROS at solvent breakthrough compared til the end of the flood. Complete recovery would certainly take
with only 23 % for the 9: 1 flood (Table 2). As the size of the solvent- many PV's of additional injection. Clearly, as far as production
free oil bank decreases (from Run EV 1: 1 to EV 9: 1), however, results are concerned, the combination of efficient short-term and
the amount of slow, long-term oil production increases. Thus, given small, relatively constant long-term oil production is the distinguish-
ample time, the majority of the residual oil still appears to be ing feature of the 1: 1 and 2: I heptane floods.
recoverable. The 9: 1 flood recovered 64 % of ROS at 4 PV. and Oleic-phase saturation profiles during these heptane floods are
the fraction of oil in the produced cut of the oleic phase at this point shown in Figs. 2b through 2d. The profiles for the 1 : 1 and 2: I
was still 30%. This suggests that there might be little trapped oil displacements (Figs. 2c and 2d) show considerably more struc-
in a water-wet system when a moderately viscous solvent is used ture than seen in the mineral oil floods. The initial oil banks and
to displace miscibly an oil of equal or lower viscosity. This is a trailing plateau or shoulder regions are unusual, and we believe
new and somewhat surprising result for water-wet cores. this to be the first reported observation of su~h behavior. Note that

TABLE 2-EQUAL-VISCOSITY FLOODS

Oil Solvent Recovery


Oil Viscosity Viscosity Rate At PV
Run Fraction (cp) (cp) (cm 3 /h) 8 *
--E!-
% 8 0r Injected Remarks

Mineral Oil Displacing Mineral Oil


EV 0:1 5.8 5.8 0.347 68 0.47'* Fig.1a
98 1
EV 1:1 0.5 6.1 5.8 34 0.37 0.468 48 0.55* • Fig.1b
100 2
EV 2:1 0.32 5.8 5.8 48 0.353 0.42 34 0.62" Fig.1c
82 2
89 2.5
EV 9:1 0.1 5.8 5.8 40 0.359 0.395 23 1.3" Fig.1d
37 2
64 4

Heptane Displacing Heptane


EV 1:9, C 7 0.9 0.4 0.4 50 0.32t NA 19 0.27" Fig.1e
61 1
EV 1 :3, C 7 0.75 0.4 0.4 48 NA 16 0.24" Fig.1f
52 1
EV1:1,C 7 0.475 0.4 0.4 57 0.317 0.378 12 0.25" Fig.1g
34 1

*By microwave device .


• • Solvent breakthrough.
tS or assumed equal to Sor of Run EV 1:1, C 7 .

214 SPE Reservoir Engineering, May 1992


.J

C
o
... a b ., Run No.
EV2:1 ... Run No.
EVI:1 d
.J
«
a:
w
... 1.1 1.1
z

•.. ,.. •.. ~


u
o
c:;;U u ... .J
«
a:
C>
a:
::!; • +-,
Run No.
..-liP_."'.• ..,.....,..,..,.. . . . .-;:,~.;..V0;::;:1:..-1•., ... ,,
... , . 1.1 II
w
z
i
,,-- TIME. rm VOLUMII

f ... IoU
e
'.1
I!"\
~
~ .... o
0 0
,. oCum.OIl
Z
«
t-
Il.
W
J:
I.'
o Solvent Cut iii
z
II o Oleic Cut ~
Q.
Run No. Co Solvent
Conc. IoU
EV 1:3, C7 J:
,~~~~~~~~
1.5 15 a.' 0 0.5 1.5 U

.. k
..J
o
..J
«
a:
'.1 1.1 w
z
... I.'
i
iii
...
Run No. Run No.
I.'
Run No.
.. Run No.
z
«
t-
Il.
W
I+-.~""I~.I..,.....~'..,........-T,\V;..:°M:l-.-lU O+-,~<\.,~
.• ..,.....~,..,........,..;:.t~IV:,..,1;:;:3~a.0 '., II I~V 1:1 u G+,....."'~..,.....~.............
I.~'i-Vn2:1~.. J:
lIMt rOlE VOLUMIS liME. rOlE VOLUMES liME. POA£ VOLUMES TIME. 'Olf VOLUMES

WAG Ratio 0:1 1:3 1:1 2:1 9:1

Fig. 1-0il recovery data, experimental results.

the oil bank in the 1: 1 displacement (Fig. 2c) expands much more mixing mechanism. The 9: 1 flood (Run EV 9 : 1) is strong confir-
rapidly than the trailing shoulder while the reverse is true for the mation of this mixing trend. A clean oil bank exists but contains
2 : 1 flood (Fig, 2d). Comparison of the saturation profiles and the only 23% of the ROS. Mixing between oil and solvent appears to
production histories (Figs. Ij and Ik) shows that variations in one be capacitive at all times. Oil recovery is 64 % at 4 PV with the
are generally related to variations in the other. The concentration oleic-phase cuts still containing 30% oil and slowly decreasing.
of the tagging agent (solvent) in the produced oleic phase is partic- While the time required for this recovery is large for the labora-
ularly interesting in this regard. It remains zero across some fronts, tory floods, this will not be the case (because of scaling considera-
rises rapidly at others, and tends to remain constant (nonzero) tions) in an actual field displacement. 5 The above findings suggest
throughout the shoulder regions. that mass transfer from a dendritic oil saturation is the primary mode
of solvent/oil transfer at high water/solvent injection ratios and that
Discussion of Experimental Results most of the oil is in a dendritic state and hence recoverable.
Solvent/Oil Viscosity Ratio = 1, (6 cp/6 cp and 0.4 cp/O.4 cp). The results for the heptane-displacing-heptane floods (Figs. Ie
There are two major observations related to the displacements for through Ig) were, in many respects, contradictory to the mineral
the tagged mineral-oil/mineral-oil/water system. First, the mixing oil results. The majority of the ROS remained trapped and there
region between solvent and oil in the oleic-phase production in- was little trailing oil production. Examining the I : 1 heptane flood
creases dramatically with increasing water/solvent injection ratio, (EV 1: 1, C7 in Table 2), we find that the steady-state oleic-phase
and this mixing exhibits a strong capacitive character at high saturation for this flood (0.378) is slightly lower than that found
water/solvent ratio. Second, oil recovery is very high at all for the 9: 1 mineral oil flood (0.395); thus, on the basis of the min-
water/solvent injection ratios. eral oil results, a large dendritic saturation and high recovery
The trend in mixing is clear from the production histories (Figs. ( > 70 %) should be expected. Experimentally, we find little indi-
Ia and Id). The 0: 1 flood (Run EV 0: 1) produced a nearly piston- cation of a dendritic saturation (Fig. Ig), and oil recovery is only
like displacement of the residual oil, with 68 % of the ROS pro- 34 %. If one considers the difference between the steady-state satu-
duced at 0.47 PV (solvent breakthrough) and 98% recovery at 1.0 ration and ROS (AS=O.06 in Table 2) as a more appropriate meas-
PV (Table 2). The mixing zone is small and appears to be primari- ure of the trapped and dendritic saturations, then the 1 : 1 heptane
ly dispersive in character, although there is some asymmetry at high flood should be compared with the 2: I mineral oil flood
solvent cuts. For the 1: 1 flood (Run EV I: I), approximately 48% (AS=0.067). However, a large degree of trailing oil production
of the ROS is recovered at solvent breakthrough (0.55 PV) and was still evident in the 2: 1 mineral oil flood and long-term oil recov-
recovery is 100% of ROS by 2 PV. The mixing zone is signifi- ery was nearly complete. The 1: 9 and 1: 3 floods (Figs. Ie and
cantly larger, and while still quite dispersive in overall appearance, If) produced similar results. While oil recoveries for these floods
there is clearly more trailing oil production than observed in the were somewhat higher at 60% and 52% ROS, they were still far
o: 1 flood. At an injection ratio of 2: 1 (Run EV 2 : I), a much differ- from complete. However one compares the high- and low-viscosity
ent production behavior is seen. Only 34 % of the ROS is recov- results, the conclusion that the trapped and dendritic saturation be-
ered at solvent breakthrough (0.62 PV), and oil production continues havior is different seems inescapable. Apparently, viscous forces
at 2 PV with 82 % recovery. The mixing region between solvent are important in determining the trapped and dendritic saturations.
and oil is much larger than in the 1: 1 flood, and a break or "knee"
in the solvent cut occurs at about 1.1 PV. This apears to be a tran- Solvent/Oil Viscosity Ratio < 1, (0.4 cp/6 cp). In analyzing the
sition from a primarily dispersive type of mixing to a capacitive displacement results for the heptane (O.4-cp solvent) and 6-cp min-

SPE Reservoir Engineering, May 1992 215


TABLE 3-LOW-VISCOSITY FLOOD (HEPTANE DISPLACING MINERAL OIL)

Oil Solvent Recovery


Oil Viscosity Viscosity Rate At PV
Run Fraction (cp) (cp) (cm 3 /h) Sor * Soss * % Sor Injected Remarks
LV 0:1 5.8 0.4 51 0.348 0.576 34 0.32** Fig.1h
98 1.3
LV 1:3 0.78 5.8 0.4 54 0.353 0.41 33 0.33** Fig.1i
94 1.3
LV 1:1 0.48 5.8 0.4 47 0.351 0.371 33 0.34** Fig.1j
52 In bank
58 2
LV 2:1 0.33 5.8 0.4 50 0.352 0.374 20 0.37* * Fig.1k
32 In bank
43 2
'By microwave device.
• 'SoIvent breakthrough.

~ oil system, one must consider the saturation distributions dur- We begin by examining the 0: 1 heptane flood (Run LV 0: I in
iDa tIW floods and the production histories. This enables one to Fig. Ih). About 34% of the ROS is produced at solvent breakthrough
understand better what might otherwise be mistaken for miscible (0.32 PV), and overall recovery is 98% at 1.3 PV (Table 3). While
instability or simply "noise" in the production behavior. The ab- the efficiency of this flood is very similar to that of the 0: 1 flood
"sence of miscible instability is indicated by the production data, for the mineral oil (solvent) system, there are some significant differ-
as diicllssed below. ences in the solvent/oil mixing behavior. In the mineral oil flood,
a solvent-free oil bank (68% ROS) was produced for 0.3 PV and
then the solvent cut rose very rapidly to 80% of the produced oleic
phase. The heptane flood produced clean oil for only 0.12 PV. Af-
a Run No. EV 2:1
ter solvent breakthrough, the heptane cut rises rapidly (Fig. Ih),
8.7 but at about a 45 % cut, it breaks to a more gradual increase. Note
Final that the solvent breakthrough and rapid increase in solvent cut occurs
l I.e. as the oleic-phase fractional flow approaches unity, while the more
f
!_
0.5

:::=:;4=!~
O.OB PV
\.., .!'," 1)1" gradual increase in solvent cut occurs at constant oleic-phase frac-
tional flow (unity) and saturation. The solvent/oil mixing in the hep-
!E
g
0.4
0.3 Initial
. r:;=::'t-~ tane flood is not dominantly dispersive in character (as was seen
in the mineral oil flood), and miscible instability is not indicated.
0.3 PV The lack of instability is probably a result of the small core width
(1.8 in.). Thus, we are left with capacitive mixing at constant satu-
b
0.7
...... ....... ....... ....... ration as the best candidate for explaining the latter part of the sol-
8 ."
"" vent breakthrough behavior, though it is certainly not suggested
c> C> ;; ;;
~ 0.6
\. ....:" ,,"- .. ....
by the 0: 1 mineral oil flood.

! 0.5 ..:';\
"r,'
.
""- ------ \ The 1 : 3 heptane flood (Run LVI: 3 in Fig. Ii) exhibits satura-
tion profiles and a production history that are almost identical to
I... 0.4 :,
- .....---.~_,. __ l.._....,\ .• - .....- ._.......,
,Final those of the LV 0: 1 heptane flood, except for the trailing water
Ii shock. This shock virtually ends the oil flow, leaving only a small
m 0.3 Initial amount of trailing production. The shock in oleic-phase fractional
Run No. LV 1:3 flow occurs simultaneously with a shock in solvent concentration
C (to pure solvent), leaving the solvent cut almost constant. Once
0.7 ....... ....... ...
c..
....... again, 33 % of the ROS is produced before solvent breakthrough
(0.33 PV), but the ultimate recovery is only 94 % at 1.3 PV (Table
~
i;::
0.6
r~' 3).
0.5 The efficient oil banking (and high recovery) observed in the 0: 1

ItI 0.4
and 1: 3 heptane (displacing mineral oil) floods is produced by the
high viscosity (6 cp) and consequent high oleic saturation in the
0.3 mobilized oil bank. The dramatic decline in oil production in the
is 1 : 3 flood is, of course, produced by the shock in both oleic-phase
d flow and solvent concentration. This results in a low oleic-phase
saturation because only water and heptane are flowing (Le., high
,..
0.7 ...
:>
....
:>
trapping).
....
\~\
u We now consider the 1: 1 heptane (displacing mineral oil) flood.
fiI 0.5
The results (Run LV 1: 1) are shown in Fig. Ij. The saturation pro-
files clearly show that a single oleic-phase bank initially forms (Fig.
....::0e 0.4 2c). However, a smail oscillating disturbance develops behind this
U>
::! bank and evolves into a slowly growing shoulder behind the rapid-
c:>
0.3 ly expanding bank. The produced cuts confirm this behavior. At
Run No. LV 2:1
0.2 breakthrough, the oleic-phase cuts rise to about 85 % of the pro-
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 O.B 0.9 1.0 duced fluids, and this level is sustained for about 0.1 PV. No sol-
NORMALIZED DISTANCE. X/L vent is produced during this period. Subsequently, the oleic-phase
fraction decreases to about 55 % and remains constant for a brief
interval of about 0.05 PV before decreasing to the steady-state level
Fig. 2-Experlmental 011 saturation dlstrlb·utlons. of 48 %. This period of oleic-phase production at a cut of 55 % cor-
responds to the saturation shoulder at an oleic-phase saturation of

216 SPE Reservoir Engineering. May 1992


about 42 %. The solvent fraction rises rapidly to about 25 % of the Conclusions
oleic-phase production as the fraction decreases from 85 % to 55 %. I. For displacements in a water-wet Berea core with high-
At this point, the solvent fraction remains constant, mirroring the viscosity solvent (6 cp) and oil (6 cp), little oil is trapped during
behavior of the oleic-phase cuts Then the solvent fraction once again simultaneous water/solvent injection and a large dendritic oleic-
rises rapidly to about 93 % of the oleic-phase production as the oleic- phase saturation exists over most of the saturation range. This is
phase cuts decrease to steady-state. However, the region of con- true even at high water/solvent ratio (e.g., 9: 1).
stant oleic-phase fractional flow and solvent cut creates the appear- 2. When conventional miscible flood experiments-i.e., low-
ance of a slightly larger mixing zone. viscosity solvent (0.4 cp) and water (1 cp) displacing 6-cp oil-
Comparing the I : I heptane-displacing-heptane flood (Run EV were performed in the same core, a more typical oil-trapping be-
1 : 1, C7) and the I: 1 heptane-displacing-mineral oil flood (Run havior for a water-wet core was confirmed.
LV I: I), we see that while the ROS for the heptane displacement 3. A low-viscosity solvent (0.4 cp) and water (1 cp) displacing
(32 %) is lower than that for mineral oil (35 %), the steady-state satu- an equal-viscosity oil (0.4 cp) produced a large degree of oil trap-
rations are approximately equal. The oil recovery for the heptane- ping and little recovery of dendritic oil. This implies that the den-
displacing-mineral oil flood (61 %) is much greater than for the I : 1 dritic behavior depends on the viscosity ratios of oleic and water
heptane-displacing-heptane flood (31 %). This is true even if the phases.
slow tailing oil production is excluded in the heptane-displacing- 4. The oleic phase trapped and dendritic fractions depend on the
mineral oil flood. The dominant factor in mobilization of residual oleic-phase saturation and viscosity. This suggests that viscous forces
oil is clearly the viscosity of the residual oil itself. The trailing oil are important in determining the distribution of the nonwetting phase
production is apparently a vestige of the residual mineral oil be- during a drainage process.
cause it is not observed in the heptane-displacing-heptane floods. 5. The macroscopic saturation distribution during the displace-
We note in passing that reproducibility was sometimes difficult to ment of a viscous oil by a less viscous solvent is strongly influ-
achieve in the heptane-displacing-mineral oil floods. In a repeat I : 1 enced by fractional-flow hysteresis. This leads to saturation/solvent
flood (not presented here), the oscillating disturbance after the ini- fronts and saturation shoulders trailing the main oil bank that are
tial oil bank developed into a completely separate oil bank instead not predicted by the current models of miscible displacement.
of a shoulder. Much of the displacement behavior observed in these experiments
Finally, we consider the 2: 1 flood (Run LV 2: I in Fig. Ik). can be qualitatively explained with a model containing trapped and
The results for this flood exhibit even more structure than the I : 1 dendritic oleic-phase saturations.
floods. The saturation profiles in Fig. 2d clearly show a complex,
evolving saturation distribution. The first profile reveals an oil bank Nomenclature
and trailing shoulder. At breakthrough, there is an initial oil bank k = absolute permeability, md
followed by a prominent shoulder at an oleic-phase saturation of k ro = oil relative permeability
42 %. Following this shoulder is another region of constant satura- krw = water relative permeability
tion (39%) just slightly higher than the eventual steady-state satu- Siw = connate water saturation, fraction
ration of 37.4 %. A small secondary oleic-phase bank finally ends Sor = waterflood ROS, fraction
the displacement and steady state is achieved. In contrast with the Soss = steady-state oleic-phase saturation, fraction
1: I flood, the initial oil bank expands very slowly, while the trail- /1S = difference between steady-state saturation and ROS,
ing shoulder(s) exhibit a rapid growth. Thus, the initial increase fraction
in the produced oleic-phase cut, although large, is short lived, while
the subsequent production representing the shoulder(s) is sustained Acknowledgment
for about 0.25 PV. The saturation profiles after breakthrough and
We thank the Unocal Corp. management for permission to publish
the produced cuts suggest that the second saturation shoulder at an
this paper.
oleic-phase saturation of 39% is never actually produced as such.
The saturation at the exit end of the core appears to be at or near References
the saturation of the first shoulder in both of the postbreakthrough
saturation scans. In addition, the produced oleic-phase cuts are near- 1. Gorell, S.B.: "Modeling the Effect of Trapping and Water Alternate
Gas (WAG) Injection on Tertiary Miscible Displacements," paper SPE
ly constant at about 37% from 0.35 PV until they fall abruptly to
17340 presented at the 1988 SPE/DOE Enhanced Oil Recovery Sym-
steady state (0.55 PV). Solvent breaks through at about 0.4 PV, posium, Tulsa, April 17-20.
jumping immediately to 28 % of the oleic-phase production and re- 2. Raimondi, P. and Torcaso, M.A.: "Distribution of the Oil Phase Ob-
maining nearly constant until 0.5 PV. At this point, the solvent cut tained Upon Imbibition of Water, " SPEJ (March 1964) 49-55; Trans.,
rises rapidly to 85 % of the oleic-phase production and thereafter AIME,231.
increases gradually to 95% at 2 PV. Solvent-free oil production 3. Lin, E.C. and Huang, E.T.S.: "The Effect of Rock Wettability on Water
is only 20% of the ROS. Overall oil recovery at 2 PV is 43% of Blocking During Miscible Displacement," SPERE (May 1990) 205-12.
4. Wang, F.H.L.: "Effect ofWettability Alteration on Water/Oil Rela-
the ROS (Table 3).
tive Permeability, Dispersion, and Flowable Saturation in Porous Me-
A comparison of the 2 : 1 and I : 1 results (Runs LV 2 : 1 and LV dia," SPERE (May 1988) 617-28.
1: 1) shows that less oil is banked in the 2: 1 flood (20% vs. 30% 5. Stalkup, F.I.: "Displacement of Oil by Solvent at High Water Satura-
at solvent breakthrough) and that the overall recovery at 2 PV is tion," SPEJ (Dec. 1970) 337-48.
substantially lower (43% vs. 60%). During the period of slow oil 6. Salter, S.l. and Mohanty, K.K.: "Multiphase Flow in Porous Media:
production following the oil bank, however, the fraction of oil in I. Macroscopic Observations and Modeling," paper SPE 110 17 present-
the oleic-phase cuts is greater in the 2: 1 flood than in the I : 1 flood. ed at the 1982 SPE Annual Technical Conference and Exhibition, New
If one assumes, as suggested, that the trailing oil production is a Orleans, Sept. 26-29.
7. O'Steen, B.L. and Huang, E.T.S.: "Modeling of Trapped and Den-
consequence of the mineral oil viscosity, then the above behavior dritic Oil Mobilization During Miscible Displacement," In Situ (1990)
is reasonable. 14, No.3, 285.
The unusual behavior of the saturation profIles and produced frac- 8. Koval, E.l.: "A Method for Predicting the Performance of Unstable
tions for the 2: 1 and I: I heptane floods (Figs. 2d and 2c) can at Miscible Displacement in Heterogeneous Porous Media," SPEJ (June
least be partially understood by appealing to fractional-flow hyste- 1963) 145-54; Trans., AIME, 228.
resis. 9- 11 Models of miscible displacement that do not allow for 9. O'Steen, B.L. and Holm, L.W.: "Saturation Distribution During Dis-
fractional-flow hysteresis can produce only a single oil bank (two placements Generated by the Simultaneous Injection of Oil and Water, "
SPERE (Feb. 1988) 130-38.
fronts) in simulating the experiments performed here. This behavior
10. Gladfelter, R.E. and Gupta, S.P.: "Effect of Fractional Flow Hyste-
is discussed in great detail by Gladfelter and Gupta. 10 resis on Recovery of Tertiary Oil," SPEJ (Dec. 1980) 508-20.

SPE Reservoir Engineering, May 1992 217


11. O'Steen, B.L.: "Time-Dependent ~ow Behavior During Immiscible
Authors Displacement in Porous Media," paper SPE 14369 presented at the
1985 SPE Annual Technical Conference and Exhibition, Las Vegas,
Lance O'Steen Is Sept. 22-25.
a research engI-
neer In the EnvIron- 51 Metric Conversion Factors
mental Transport
Section of the U.S. cp x 1.0* E-03 Pa's
DOE Savannah in. x 2.54* E+OO in.
River Laboratory In md x 9.869233 E-04 /lm2
Aiken, SC. He Is
currently Involved 'Converslon factor is exact. SPERE
In modeling studies Original SPE manuscript received for review April 22, 1990. Revised manuscript received
of the atmospheric July 1. 1991. Paper accepted for publication Aug. 23. 1991. Paper (SPE 20181) first present-
Huang o 'Steen boundary layer. He ed at the 1990 SPEIDOE Symposium on Enhanced Oil Recovery held in Tulsa. April 22-25.
previously con-
ducted studies In miscible displacement as a senior research
engineer at the Unocal Science & Technology Center. O'steen
holds a BS degree from Clemson U. and a PhD degree from
the california Inst. of Technology, both In chemical engineer-
Ing. Edward T.S. Huang, a research associate at Unocal
Science & Technology Dlv. In Brea, CA, researches fluid
phase behavior, miscible flow mechanisms, foam mobility
control, and rock wettablllty effects on 011 recovery. He has
been Involved In the design of several miscible flood projects.
He previously worked for Sohlo Petroleum and Gulf R&D Co.
He holds MS and PhD degrees from the U. of Kansas. Huang
was a member of the Editorial Review Committee from 1986
to 1991.

218 SPE Reservoir Engineering, May 1992

S-ar putea să vă placă și