Professional Documents
Culture Documents
Summary ments that used fired Berea sandstone with the help of chemical
A thorough understanding of foam fundamentals is crucial to the tracer. Tang and Kovscek (2004) also reported 56 to 80%
optimal design of foams for improved oil recovery (IOR) or trapped-gas saturation in their study that used a gas-tracer tech-
enhanced oil recovery (EOR). This study, for the first time, presents nique in combination with CT scans in Berea sandstone. Nguyen-
anomalous foam-fractional-flow solutions that deviate significantly Quoc et al. (2002) used an X-ray CT-scan apparatus to measure
from the conventional solutions at high-injection foam qualities by the trapped-gas saturation in a Berea core, and they measured 50
comparing method-of-characteristics and mechanistic bubble-popu- to 70% gas trapped in porous media from two series of flow
lation-balance simulations. experiments in which foam with 90% quality is injected at the
The results from modeling and simulations derived from core- total rates of 0.3 and 0.4 cm3/min. Kil et al. (2009), a study of CT
flood experiments revealed the following: images based on a conservative estimate of diffusion coefficient,
There are three regions—Region A with relatively wet (or however, estimated flowing-gas fraction one or two orders of
high fw) injection conditions in which the solutions are consistent magnitude lower than other studies.
with the conventional fractional-flow theory; Region C with very Similar to other miscible and immiscible processes such as
dry (or low fw) injection conditions in which the solutions deviate polymer flooding (Pope 1980) and solvent flooding (Walsh and
significantly; and Region B in between, which has a negative dfw/ Lake 1989), fractional-flow analysis has been used as an efficient
dSw slope showing physically unstable solutions. tool to characterize the nature of foam displacement in porous
For dry-injection conditions in Region C, the solutions media. This mathematical tool was first introduced by Buckley
require a constant state (IJ) between initial (I) and injection (J) and Leverett (1942) to interpret secondary-oil-recovery processes
conditions, forcing a shock from I to IJ by intersecting fractional- such as waterflooding and gas injection. The fractional-flow solu-
flow curves, followed by spreading waves or another shock to tions typically consist of a shock and a series of spreading waves
reach from IJ to J. that describe the changes in water saturation in porous media,
The location of IJ in fw vs. Sw domain moves to the left (or which can, in turn, be translated into saturation profiles, effluent
toward lower Sw) as the total injection velocity increases for both history, time/distance diagrams, and pressure profiles. Previous
weak and strong foams until it reaches limiting water saturation. studies investigated many different aspects of foam displacement
Even though foams at high-injection quality are popular for in porous media implemented into the framework of fractional-
mobility control associating a minimum amount of surfactant solu- flow analysis, including foam flow with no oil present (Zhou and
tions, foam behaviors at dry conditions have not been thoroughly Rossen 1995; Alvarez et al. 2001; Kam and Rossen 2003; Ashoori
investigated and understood. The outcome of this study is believed and Rossen 2012; Roostapour and Kam 2012), foam flow with oil
to be helpful to the successful planning of foam IOR/EOR field present (Mayberry and Kam 2008; Zanganeh et al. 2011), and
applications. foam-diversion modeling in two isolated layers (Rosman and
Kam 2009).
The experimental study of Gauglitz et al. (2002) shows that
Introduction foams can exhibit three different foam states (strong-foam, inter-
Foams have been used in the petroleum industry as a versatile mediate-foam, and weak-foam), and the surface that represents
means in IOR and EOR processes to control gas-phase mobility the steady-state pressure gradient can be expressed by a 3D sur-
and to overcome permeability variations (Kovscek and Radke face, as illustrated by Fig. 1. This concept, later called foam catas-
1994; Schramm 1994; Rossen 1996). Numerous field-scale treat- trophe (Kam 2008), indicates that the steady-state pressure
ments and pilot tests show a significant production enhancement gradient during foam flow is multivalued when injection rates are
after surfactant/foam applications, including Snorre field in the fixed, but is single-valued if injection pressure is fixed. The con-
North Sea (Blaker et al. 2002), Midway-Sunset field (Duerksen ventional fractional-flow methods have been attempted to merge
and Ploeg 1985), Prudhoe Bay (Krause et al. 1992), and Guada- with mechanistic foam modeling to reproduce such foam proper-
lupe field (Mohammadi and McCollum 1989). A recent increase ties (Dholkawala et al. 2007; Afsharpoor et al. 2010).
in oil price allows surfactant foams to be a more viable option The main objective of this study is to investigate the mecha-
with other chemical additives such as polymers and alkaline (Sri- nisms of foam displacement in porous media by use of fractional-
vastava et al. 2009). flow analysis and mechanistic foam simulation. The focus is espe-
Previous studies show that successful foam applications for cially on the deviation from conventional fractional-flow analysis
mobility control are typically accompanied by a large fraction of at dry (or low fw) foam-injection conditions, which is a common
gas trapped in the media that does not contribute to the flow practice in field foam applications. Such an anomalous behavior
(Kovscek and Radke 1994; Falls et al. 1989; Bertin and Kovscek at very dry injection conditions has been reported experimentally
2003), and this high trapped-gas saturation indirectly causes a (Kibodeaux and Rossen 1997; Wassmuth et al. 2001; Xu and Ros-
large pressure buildup for foam propagation in linear-flow experi- sen 2004), but the mechanisms behind it have not been clearly
ments, as evidenced by various experimental techniques such as understood.
magnetic-resonance imaging, computed tomographic (CT) scan-
ning, and chemical tracers. For example, Radke and Gillis (1990)
reported a trapped-gas saturation higher than 70% in their experi- Methods
Earlier studies of Falls et al. (1988), Patzek (1988), and Ettinger
and Radke (1992) show how mechanistic foam modeling and sim-
Copyright V
C 2013 Society of Petroleum Engineers
ulation can be achieved by combining material balance and bub-
This paper (SPE 152907) was accepted for presentation at the SPE Improved Oil Recovery ble-population balance.
Symposium, Tulsa, 14–18 April 2012, and revised for publication. Original manuscript
received for review 19 December 2011. Revised manuscript received for review 16
The mass-balance equation for the immiscible gas/liquid two-
September 2012. Paper peer approved 10 December 2012. phase system can be simplified into a fractional-flow equation:
e
coars
t
e
foam rock and fluid properties used in the modeling and simulation in
wR ate this study.
Liquid Flo
Fig. 2 shows mechanistic foam-fractional-flow curves at dif-
ferent total injection velocities (ut) ranging from 2.8 105 to
Fig. 1—Multivalued foam-catastrophe surface represented by
7.0 105 m/s to demonstrate how the multivalued foam-catas-
three different foam states (Gauglitz et al. 2002).
trophe surface (Fig. 1) affects the curves. This characteristic of
the foam-catastrophe surface results from the bubble-generation
mechanism derived from the “minimum mobilization pressure”
@ ut @
ðSj Þ þ ð fj Þ ¼ 0 . . . . . . . . . . . . . . . . . . . . . . ð1Þ and the “lamella mobilization and division mechanism” in porous
@t / @x media. [Please see Kam and Rossen (2003) for more details.] The
where Sj and fj are the saturation and fractional flow of phase j, nature of velocity-dependent fractional-flow curves results in the
respectively; / is the porosity of the media; ut is total injection following observations: at given injection condition (J) specified
velocity; and t and x are time and space, respectively. Lake (1989) by fw on the y-axis, the number of intersections is either one or
presents in-depth discussions about derivations and assumptions three; and, if three, the intersection at the lowest and highest water
to reach Eq. 1 that include 1D flow, isothermal conditions, rock saturations (Sw) corresponds to the strong-foam and weak-foam
and fluid properties with no sorption, reaction or interphase mass states, respectively, whereas the saturation value in the middle is
transfer, and immediate attainment of a local steady state. for the intermediate state, which cannot be obtained in fixed-rate
Bubble-population balance keeps track of the change in foam foam-injection experiments (Kam et al. 2007; Kam 2008). [There
texture (i.e., the number of foam films or lamellae) in time and still might be two intersections on rare occasions when the injec-
space. The following equation applies when the medium is rela- tion condition (i.e., a horizontal line in the fw-vs.-Sw domain)
tively wet (Falls et al. 1988; Kovscek et al. 1995): touches the inflection point.] It should be noted that any part of
the curve with a negative slope (or negative dfw/dSw value) is
@ @ physically unstable and therefore cannot be observed in actual
/ ðSg nf Þ þ ðnf ug Þ ¼ /Sg ðRg Rc Þ . . . . . . . . . . ð2Þ foam-displacement experiments (Dholkawala et al. 2007). This
@t @x
study investigates the nature of foam displacements in a broad
where Rg and Rc are the rates of lamella creation and coalescence, range of injection velocities, but it presents two representative
nf is foam texture, and Sg is gas saturation. Note that this bubble cases: ut ¼ 3.7 105 m/s for weak-foam propagation and ut ¼
population balance equation is valid only when the capillary pres- 7.0 105 m/s for strong-foam propagation. Mechanistic foam
sure (Pc) at the time and location of interest are lower than the simulation shows that an abrupt transition from weak foam to
limiting capillary pressure (P*c ), or Pc < P*c . The solution of this strong foam, or so-called foam generation, occurs at approxi-
equation becomes trivial with nf ¼ zero (or no foam films present), mately ut ¼ 5.0 105 m/s. In all cases investigated in this study,
if Pc > P*c or the medium is relatively dry. The use of the limiting the initial condition (I) of the media is 100% saturated with sur-
capillary pressure (P*c ) can be translated into the use of the limit- factant solutions. The results at other injection ut values attempted
ing-water saturation (S*w) such that Pc < P*c is equivalent to in this study, although not shown because of space limitation, are
Sw > S*w, and Pc > P*c is equivalent to Sw < S*w. consistent with the results with these two ut values.
Among many studies available in the literature, this study fol-
lows the mechanistic modeling and simulation technique pre-
sented by Afsharpoor et al. (2010), which is an extension of Kam Weak-Foam Propagation. Fig. 3 shows the fractional-flow
and Rossen (2003), Kam et al. (2007), and Kam (2008) with the curve at ut ¼ 3.7 105 m/s, which consists of three foam states.
concept of trapped-gas saturation (Kovscek and Radke 1994) and If the focus is on the weak-foam state only, one can find that there
effective foam viscosity (Hirasaki and Lawson 1985). The model are, in general, three different types of injection conditions (J):
is derived from the minimum pressure gradient for mobilization Region A with a relatively wet-injection condition in which dfw/
(and subsequent lamella mobilization and division) for lamella dSw > zero; Region C with a relatively dry-injection condition in
creation (Rossen and Gauglitz 1990; Kam and Rossen 2003) (cf. which dfw/dSw > zero; and Region B in between with dfw/
Rg term in Eq. 2) and the limiting capillary pressure for lamella dSw < zero. To examine the effect of foam propagation in differ-
coalescence (Khatib et al. 1988) (cf. Rc term in Eq. 2). Details of ent regions, this study selects four different injection conditions:
this foam model are summarized in Appendix A. J1 at fw ¼ 0.30 in Region A; J2 at fw ¼ 0.11 in Region B; and both
J3 at fw ¼ 0.045 and J4 at fw ¼ 0.002 in Region C, at the same ini-
tial condition (I) of Sw ¼ 1. The selection of these four injection
Results and Discussion conditions (J1 through J4) is somewhat arbitrary. But, on the basis
As shown by Eq. A-1 the Appendix A, the foam model borrowed of numerous injection conditions investigated in this study (not all
in this study has both a lamella-creation mechanism driven by the shown), these four injection conditions are believed to represent
mobilization pressure gradient and a lamella-coalescence mecha- all different types of displacement solutions that cover all of
nism driven by the limiting capillary pressure. This, in turn, Regions A, B, and C.
means that the resulting mechanistic foam-fractional-flow curve is Figs. 4a and 4b show the saturation profiles at J ¼ J1 and
a function not only of saturation but also of total injection veloc- J ¼ J2. In the case of injection conditions in Region A with J1
ity, as pointed out by Dholkawala et al. (2007) and Kam (2008). (Fig. 4a), the nature of displacement in mechanistic foam simula-
Table 1 lists input parameters describing foam characteristics and tion (solid lines) is consistent with the conventional understanding
Velocity Increasing
0.6 0.6 0.6
0 0.0 0
0 0.2 0.4 0.6 0.8 1 0 0.2 0.4 0.6 0.8 1 0 0.2 0.4 0.6 0.8 1
Water Saturation, Sw Water Saturation, Sw Water Saturation, Sw
Velocity Increasing
1.0 1.0 1
Water Fractional Flow, fw
0.2
ut = 7.0×105 m/s 0.2
ut = 5.3×105 m/s 0.2
ut = 4.2×105 m/s
0.0 0.0 0
0 0.2 0.4 0.6 0.8 1 0.0 0.2 0.4 0.6 0.8 1.0 0 0.2 0.4 0.6 0.8 1
Water Saturation, Sw Water Saturation, Sw Water Saturation, Sw
Fig. 2—Velocity-dependent mechanistic foam-fractional-flow curves in a wide range of total injection velocities (ut) resulting from
foam-catastrophe surface.
of fractional-flow solutions (dotted lines), which is the entire dis- oscillations, the average Sw behind the shock is very close to the
placement process governed by a single saturation shock from J1 Sw value corresponding to J2 in Fig. 3. Similar observations are
to I. In the case of injection conditions in Region B with J2 (Fig. found at other injection conditions in separate trials (not shown)
4b), although the propagation of the shock is well-captured, the within Regions A and B.
simulation shows oscillations in terms of Sw behind the shock. Figs. 5a and 5b show the saturation profiles at J ¼ J3 and
This is similar to what Kam et al. (2007) reported—mechanistic J ¼ J4, and Fig. 6 shows more-detailed information about the
simulation does not provide a stable Sw value behind the shock locations of J3 and J4, plotted together with the boundaries
because of a negative dfw/dSw value, or physically unachievable between Regions A and B [i.e., dfw/dSw ¼ 1 at (Sw, fw) ¼ (0.4005,
solution. It is interesting, however, to point out that even with 0.13)], and Regions B and C [i.e., dfw/dSw ¼ 1 at (Sw,
fw) ¼ (0.407, 0.046)], the tangent point “T” from I [i.e., (Sw,
fw) ¼ (0.403, 0.031)], and the constant state “IJ” [i.e., (Sw,
1.0 fw) ¼ (0.343, 0.007)]. In both injection conditions, simulation
results disagree with fractional-flow solutions. More specifically,
for injection condition J3 (Fig. 5a), fractional-flow analysis con-
strong foam state
te
te
intermediate sta
0.6
we
0.4
followed by spreading waves to J4.
tl
Two important aspects are worth noting: in both J3 and J4, the
en
ng
saturation shock
saturation shock
PVI=0.081
0.8 PVI=0.121 0.8
Water Saturation, Sw
Water Saturation, Sw
PVI=0.162
PVI=0.243
0.6
0.6 PVI=0.243
"J" PVI=0.364
0.4 "J" PVI=0.324
PVI=0.485 0.4
PVI=0.405
PVI=0.607
0.2
0.2
0
0 0.2 0.4 0.6 0.8 1 0
Dimensionless Distance, XD 0 0.2 0.4 0.6 0.8 1
Dimensionless Distance, XD
(a) (b)
Fig. 4—Saturation profiles at two different injection conditions at total injection velocity (ut) 5 3.7 3 1025 m/s: (a) J1 with fw 5 0.30
(Region A in Fig. 3) and (b) J2 with fw 5 0.11 (Region B in Fig. 3).
1 "I" 1 "I"
PVI=0.121
PVI=0.121
Water Saturation, sw
Water Saturation, sw
0.8 0.8
PVI=0.243
PVI=0.243
PVI=0.364
0.6 PVI=0.364 0.6
PVI=0.485
"J" PVI=0.485
0.4 0.4 PVI=0.607
PVI=0.607
0 0
0 0.2 0.4 0.6 0.8 1 0 0.2 0.4 0.6 0.8 1
Dimensionless Distance, XD Dimensionless Distance, XD
(a) (b)
Fig. 5—Saturation profiles at two different injection conditions at total injection velocity (ut) 5 3.7 3 1025 m/s: (a) J3 with fw 5 0.045
and (b) J4 with fw 5 0.002 (both in Region C in Fig. 3).
J3
0.8 (0.4005,0.13) 0.04
Tangent Point
(0.407, 0.045)
(0.403, 0.031)
0.6 0.03
∞
dfw/dSw=∞ A
(0.407,0.046)
0.4 0.02
J4
0.2 0.01 (0.252,0.002)
Swc "IJ"
B (0.04, 0) (0.343, 0.007)
0.0
C 0.00
0.0 0.2 "IJ"0.4 0.6 0.8 1.0 0.0 0.2 0.4
Fig. 6—Location of constant state (IJ ) in comparison with injection conditions (J3 and J4) in Region C (ut 53.7 3 1025 m/s): Simula-
tion forces the shock to be constructed by intersecting the fractional-flow curve.
Water Saturation, Sw
Water Saturation, Sw
0.8 0.8
PVI=0.243 PVI=0.243
PVI=0.485 PVI=0.485
"J"
0.4 PVI=0.607 0.4
PVI=0.607
0 0
0 0.2 0.4 0.6 0.8 1 0 0.2 0.4 0.6 0.8 1
Dimensionless Distance, XD Dimensionless Distance, XD
(a) (b)
Fig. 7—Saturation profiles with injection condition (J ) at (a) J3 and (b) J4 in comparison with Fig. 5: After constant state (IJ ) is
forced, fractional-flow analysis can reproduce simulation results.
This behavior is not typical but anomalous because it violates the at fw ¼ 0.20 in Region A; J2 at fw ¼ 0.017 in Region B; and J3 at
entropy condition (Luckhaus and Plotnikov 2000) that does not fw ¼ 0.008 and J4 at fw ¼ 2.86 106, both in Region C.
allow the fractional-flow curves to be intersected by the shock Figs. 9a and 9b and Figs. 10a and 10b show the saturation
construction. This strange behavior is discussed in later sections. profiles at J1, J2, J3, and J4 with conventional fractional-flow solu-
Fig. 7 shows saturation profiles similar to those in Fig. 5 with tions (dotted lines) and mechanistic simulation results (solid
J3 and J4 (i.e., dry foam injection in Region C); this time the con- lines). These four cases leading to strong foams are very similar
stant state (IJ) borrowed from the simulation is incorporated into to those responses in weak foams (see Figs. 4 and 5): The injec-
the factional-flow analysis. The results in Fig. 7 show that the tion condition in Region A (i.e., J1) shows good agreement
new fractional-flow solution agrees well with simulation results between the two solutions (Fig. 9a); the injection condition in
when the predetermined constant state IJ from the simulation is Region B (i.e., J2) shows oscillations behind the shock (Fig. 9b);
implemented in the modeling. and the injection conditions in Region C (i.e., J3 and J4) lead to an
identical constant state (IJ), which requires the shock waves to
intersect fractional-flow curves, thus deviating from the conven-
Strong-Foam Propagation. Fig. 8 shows the fractional-flow tional fractional-flow solution of constructing a tangent line.
curve at ut ¼ 7.0 105 m/s, which ends up with strong-foam Fig. 11 shows that the location of the constant state IJ (0.0585,
propagation in simulation. Similar to the weak-foam case (see 7.3 106) observed in simulations (see Figs. 10a and 10b) at the
Fig. 3), the fractional-flow curve at higher total injection velocity injection condition of J3 and J4, plotted with the boundaries
also consists of three different regions: Region A with a relatively between Regions A and B [i.e., dfw/dSw ¼ 1 at (Sw, fw) ¼ (0.0585,
wet-injection condition (dfw/dSw > 0); Region C with a relatively 0.0186)] and Regions B and C [i.e., dfw/dSw ¼ 1 at (Sw,
dry-injection condition (dfw/dSw > zero); and Region B in be- fw) ¼ (0.248, 0.008)] and the tangent point “T” [i.e., (Sw,
tween with dfw/dSw < 0. Four different injection conditions are J1 fw) ¼ (0.247, 0.007)]. Figs. 12a and 12b show the updated
0.8 0.04
Region "B" and "C"
Region "A"
A
dfw/dSw > 0
0.6 0.03
weak foam state
0.4 0.02
"J2"
B
"J1" dfw/dSw < 0
0.2 0.01
"J3"
C
dfw/dSw > 0
0.0 0.00
0 0.2 0.4 0.6 0.8 1 0 0.1 0.2 0.3
Water Saturation, Sw "J4" Water Saturation, Sw
Fig. 8—Strong-foam fractional-flow curve at ut 5 7.0 3 1025 m/s: four different injection conditions (J1 through J4) are examined to
see the effect of different Regions (A, B, and C).
Water Saturation, Sw
Water Saturation, Sw
PVI=0.306
saturation shock
saturation shock
PVI=0.306
0.6 0.6 PVI=0.459
PVI=0.459
PVI=0.612
0.4 PVI=0.612 0.4
PVI=0.765
PVI=0.765
"J"
0.2 0.2
"J"
0 0
0 0.2 0.4 0.6 0.8 1 0 0.2 0.4 0.6 0.8 1
Dimensionless Distance, XD Dimensionless Distance, XD
(a) (b)
Fig. 9—Saturation profiles at two different injection conditions at total injection velocity (ut) 5 7.0 3 1025 m/s: (a) J1 with fw 5 0.20
(Region A in Fig. 8) and (b) J2 with fw 5 0.017 (Region B in Fig. 8).
1 1
"I"
PVI=0.153
PVI=0.153
0.8 0.8
Water Saturation, Sw
PVI=0.306
PVI=0.306
saturation shock
Water Saturation, Sw
PVI=0.459
saturation shock
PVI=0.612
PVI=0.612
0.4 PVI=0.765 0.4
PVI=0.765
"J"
0.2 0.2
(a) (b)
Fig. 10—Saturation profiles at two different injection conditions at total injection velocity (ut) 5 7.0 3 1025 m/s: (a) J3 with fw 5 0.008
and (b) J4 with fw 5 2.86 3 1026 (both in Region C in Fig. 3).
(0.0585,0.0186)
A
Tangent Point "T"
Region "B" and "C"
Region "A"
(0.247,0.007)
0.6 0.03
0.4 0.02
"J2" B
"J1"
0.2 0.01
"J3"
C "IJ"
(0.0585, 7.3×106)
0.0 0.00
0 0.2 0.4 0.6 0.8 1 "J4" 0 0.1 0.2 0.3
Swc
Water Saturation, Sw (0.04,0) Water Saturation, Sw
Fig. 11—Location of constant state (IJ ) in comparison with foam injection at dry conditions (J3 and J4) in Region C (ut 5 7.0 3 1025
m/s): Simulation forces the shock to be constructed by intersecting the fractional-flow curve.
saturation shock
Water Saturation, Sw
saturation shock
PVI=0.306
Water Saturation, Sw
PVI=0.459
0.6 0.6 PVI=0.459
PVI=0.612
PVI=0.612
0.4 PVI=0.765 0.4
PVI=0.765
"J"
0.2 0.2
(a) (b)
Fig. 12—Saturation profiles with (a) J 5 J3 and (b) J 5 J4 in comparison with Fig. 10: After constant state (IJ ) is forced, fractional-
flow analysis can reproduce simulation results.
fractional-flow solutions when the constant state (IJ) is borrowed the fraction of flowing-gas saturation (Xf) play a role in foam
from simulations, which shows good agreement between the two. modeling are provided in the Appendix, and the equations to deal
Once again, without implementing IJ, the conventional fractional- with these two fractions (Eq. A-3) are from earlier studies of
flow analysis shows a deviation from simulation results. Kovscek and Radke (1994). For clarification, Xtmax sets the upper
limit of Xt, but Xt value itself, which varies with time and space,
Effect of Trapped-Gas Saturation. Although only two total interacts with other foam properties (e.g., lamella creation and co-
injection velocities (ut ¼ 3.7 105 and 7.0 105 m/s) are pre- alescence mechanisms, foam viscosity, and foam texture) in
sented here, the same trials are repeated in a wide range of injec- mechanistic foam modeling.
tion velocities. The overall responses are the same, which is Fig. 14 presents a series of mechanistic foam-fractional-flow
summarized in Fig. 13: When the injection condition is very dry curves with ut ¼ 3.7 105 m/s in a wide range of Xtmax from 0 to
and in Region C (e.g., J3 or J4), the conventional fractional-flow 0.82 (meaning that from no gas to up to 82% of gas, respectively,
solutions are not consistent with mechanistic simulation results. can be trapped during foam flow). The curve at Xtmax ¼ 0.80 is
The existence of such a constant state (IJ) that forces the satura- identical to that in Fig. 3. Note that a reduction in Xtmax means
tion shock to intersect the fractional-flow curve is caused by less foam trapped, which indirectly implies a lower effective
trapped-gas saturation, which is often used as an input parameter foam viscosity and a coarser foam texture. Therefore, the entire
in foam modeling and simulations. This is further explained in fractional-flow curves tend to shift to the right, approaching a typ-
Figs. 14 through 16. It should be noted that in all modeling and ical gas/water two-phase fractional-flow curve, when Xtmax comes
simulations in Figs. 2 through 12, the maximum fraction of nearer to zero. On the other hand, an increase in Xtmax makes the
trapped-gas saturation (Xtmax) is assumed to be 0.8 (Table 1). fractional-flow curve shift to the left and become curvy. The solid
Details about how the fraction of trapped-gas saturation (Xt) and circles in Fig. 14a show the position behind the shock from
1 “I” “I”
1
“Region A”
“Region A”
Water Fractional Flow, fw
“J1” “J1”
∞
dfw/dSw=∞
dfw/dSw=∞
“B”
“J2”
“B”
“J2”
dfw/dSw=∞
“C”
“J3”
“C”
“J3” dfw/dSw=∞
“J4”
“IJ” “J4” “IJ”
0 1 0 1
Water Saturation, Sw Water Saturation, Sw
(a) (b)
Fig. 13—Schematics summarizing mechanistic foam-fractional-flow solutions: (a) weak foam and (b) strong foam. In both cases,
there exists a constant state (IJ ) forcing the saturation shock front to intersect the curves.
X tmax X tmax
0.8 0.8
0.7 0.7
0.4 0.4 0.6
0.6
0.4
0.4
0.2
0.2 0.2 0.2
0
0
0.0 0.0
0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0
Water Saturation, Sw Water Saturation, Sw
(a) (b)
Fig. 14—Effect of trapped-gas saturation on mechanistic foam fractional flows in a wide range of maximum trapped-gas saturation
(Xtmax): (a) saturation behind shock from conventional fractional-flow analysis (tangent line) and (b) saturation behind shock from
simulation (weak foams with ut 5 3.7 3 1025 m/s).
1.0 1.0
X tmax X tmax
0.85 0.85
0.8 0.8
Water Fractional Flow, fw
0.75 0.75
0.7 0.7
0.4 0.4
0.65 0.65
0.6 0.6
0.4 0.2 0.2
0.4
0.2 0.2
0 0
0.0 0.0
0 0.2 0.4 0.6 0.8 1 0 0.2 0.4 0.6 0.8 1
Water Saturation, Sw Water Saturation, Sw
(a) (b)
Fig. 15—Effect of trapped-gas saturation on mechanistic-foam fractional flows in a wide range of maximum trapped-gas saturation
(Xtmax): (a) saturation behind shock from conventional fractional-flow analysis (tangent line) and (b) saturation behind shock from
simulation (strong foams with ut 5 7.0 3 1025 m/s).
0.35 0.30
0.30
0.25
Water Fractional Flow, fw
0.25
0.20
0.20
0.15
0.15
0.10
0.10
Simulation
0.05 Simulation
0.05
Fractional Flow Fractional Flow
0.00 0.00
0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0
Water Saturation, Sw Water Saturation, Sw
(a) (b)
Fig. 16—Direct comparison of saturations behind shock between two methods: (a) ut 5 3.7 3 1025 m/s (weak foams) and (b)
ut 5 7.0 3 1025 m/s (strong foams).