SPE-169888-MS Experimentally Based Empirical Foam Modeling+++
SPE-169888-MS Experimentally Based Empirical Foam Modeling+++
SPE-169888-MS Experimentally Based Empirical Foam Modeling+++
This paper was prepared for presentation at the SPE Improved Oil Recovery Symposium held in Tulsa, Oklahoma, USA, 12–16 April 2014.
This paper was selected for presentation by an SPE program committee following review of information contained in an abstract submitted by the author(s). Contents of the paper have not been
reviewed by the Society of Petroleum Engineers and are subject to correction by the author(s). The material does not necessarily reflect any position of the Society of Petroleum Engineers, its
officers, or members. Electronic reproduction, distribution, or storage of any part of this paper without the written consent of the Society of Petroleum Engineers is prohibited. Permission to
reproduce in print is restricted to an abstract of not more than 300 words; illustrations may not be copied. The abstract must contain conspicuous acknowledgment of SPE copyright.
Abstract
This paper presents fine-scale numerical simulations and mathematical analysis of the empirical foam model for representing
foam-surfactant flow in a vertical column of laboratory sand-pack based on two sets of experimental data conducted at
variable total velocities and variable foam qualities. The empirical foam model of CMG-STASRS is used for parametric
matching of laboratory data, and relevant foam parameters are calibrated. The paper discusses experimental setup, procedure
and measurements to provide apparent foam viscosity data needed for foam modeling. In the first set of lab tests, foam
quality is constant and the total fluid superficial velocity varies for foam shear thinning effect; while in the second tests, foam
quality is varied at a fixed total superficial velocity to capture different flow regimes and foam dry-out characteristics.
Employing an analytical method and 1-D numerical simulations of the foam flow in the sand-pack, the empirical foam model
is tuned to the first data set of variable velocity and used to predict the second data set of variable quality as a consistency
check. The model predictions for the second data set as well as the associated sensitivity analysis prove that the foam
modeling procedure of this paper is unique and applicable for large-scale predictions.
Introduction
Practical application of foams for chemical and solvent EOR processes1,5, mainly for gas mobility control, requires a
representative and predictive model for foam flow behavior and its mobility control characteristics at different scales. Foam
applications are specially targeted for mobility control in heterogeneous reservoirs to divert flow from high permeability
zones to low permeability regions for improved recovery6-8. Fractured reservoirs with their inherent high contrast between
fracture and matrix flow mobility are candidates for foam applications to divert injected chemical EOR fluids or solvents into
matrix9,10. Many parameters such as surfactant concentration, shear rate, capillary number, oil saturation, and salinity affect
foam flow rheology and behavior11. Accordingly, different mechanistic and empirical foam models have been investigated in
the literature and also implemented in commercial simulators. Usually parameters in these models are tuned to match lab
data for a particular foam-surfactant and oil system characteristics in porous media.
Foam in porous media is defined as a dispersion of gas in liquid such that the liquid phase is continuous and at least some
part of the gas phase is made discontinuous by thin liquid films called lamellae12-14. Lamellae are produced in-situ with
various mechanisms and conditions14-17, thus resulting in foams with wide range of differing characteristics. Foam is trapped
within a pore structure and the effect of this trapping is reduction in foam mobility18-22, either by reduced gas relative
permeability or increased gas viscosity23. There exists a minimum pressure gradient to generate strong foam by mobilizing
the trapped bubbles and by dividing liquid lamellae to refined foam texture11,24. At high foam quality, the amount of liquid is
insufficient and the lamellae cannot sustain the suction capillary pressure. So, foam coalesces and coarsens at a specific
capillary pressure, called limiting capillary pressure, ܲ כ, making further desaturation difficult.
In a model of capillary tubes25, it has been shown that foam can only exist in qualities less than a certain break point
quality and that foam texture stays constant above this break point quality, corresponding to the limiting capillary pressure.
The limiting capillary pressure imposes a condition for foam stability where capillary pressure increases to a level such that
the rate of foam coalescence and generation equate and foam starts to dry-out at lower water saturations15. For this reason,
two regimes exist in foam flow through porous media: the low-quality regime and the high-quality regime26. In the low-
quality regime, bubble trapping and mobilization dominate foam flow27; and in the high-quality regime, gas bubble
separation by thin liquid lamellae is prevalent, with bubble coalescence behavior occurring at the limiting capillary
pressure15,28. As the gas velocity increases and the liquid velocity is held constant, local water saturation decreases and
2 SPE-169888-MS
capillary pressure increases. When the limiting capillary pressure is maintained, the water saturation at this state remains
relatively constant.
There are two main groups of foam models in the literature: mechanistic12,29 and empirical28. The mechanistic methods are
based on bubble population balance theory, which describe physically the generation and decay of foam lamellae, and
dynamically track the mobility of foam through foam texture and foam bubble size variations. Foam texture (defined as the
number of liquid lamellae per unit volume present in porous media) is modeled explicitly with a bubble population-balance
equation and gas mobility is expressed as a function of foam texture, water saturation and other factors30. The main challenge
in use of the mechanistic population balance foam models is their complexity and difficulty in determining model parameters.
The empirical methods, on the other hand, are based on the fractional flow theory, and the effect of foam on gas mobility
reduction is expressed with a mobility reduction factor, applied to reduce the relative permeability of the gas phase depending
on surfactant concentration, oil saturation and capillary number31-34. Local steady-state condition is assumed and lamellae
creation and coalescence are balanced in the empirical model. Foam texture depends on water saturation and pressure
gradient at saturations above the critical water saturation. The rheological shear thinning and the limiting capillary pressure
effects for foam coalescence are included in terms of critical water saturation.
The modified expressions of underlying foam flow concepts have been included in the CMG-STARS commercial
reservoir simulator35,36 through empirically-based expressions, and are widely used in the oil and gas industry to describe
foam flow through porous media37-44. One of the main features of the CMG-STARS empirical foam model is that it accounts
for different characteristics of low-quality and high-quality regimes. The model allows for increase in apparent foam
viscosity with increase in foam quality in the low quality region and for decrease in apparent viscosity with increase in foam
quality (or gas fractional flow) in the high quality regime. Thus, apparent foam viscosity is maximum at a transition foam
quality between the two flow regimes for a fixed total flow rate. The empirical model also has several parameters that are
typically calibrated to laboratory experimental data through mathematical relationships or numerical simulations of
laboratory experiments28,45.
This paper presents results of foam-surfactant flow experiments in a high permeability sand-pack to provide data for
calibrating the empirical foam model implemented in CMG-STARS commercial reservoir simulator. The lab data include
the important effects of shear thinning and limiting-capillary foam dry-out. Both mathematical methods and numerical
simulations are presented to illustrate the intricacies of calibrating the empirical foam model to the sand-pack foam flooding
experiments. The developed empirical foam model is targeted for use in flow simulation studies for field pilot design of a
foam-assisted chemical EOR process in a naturally fractured reservoir.
and qualities. Foam apparent viscosity, which is the pressure gradient normalized with respect to the permeability and the
total flux of surfactant solution and gas, is calculated from measured pressure drop and phase velocities through 1-D Darcy’s
law under steady-state conditions:
−k∇p
μ fa = ………………………………………………………………………………………………..... (1)
uw + ug
The calculations in Tables 1 and 2 consider uniform foam quality along the sand-pack and do not include the impact of
any existing pressure profile on foam quality and other parameters. The resulting calculated apparent foam viscosities are
shown in Fig. 3, where the red data points show the apparent foam viscosity computed at constant quality of 0.7. It is seen
that apparent foam viscosity decreases as total velocity increases, which makes the foam a non-Newtonian shear thinning
fluid. It is an advantage for foam to be a shear thinning fluid for improved sweep efficiency purposes in reservoir
applications. This is because near the injection wellbore where the velocity is maximum, apparent foam viscosity will be low
and the injectivity will be high. However, deep in the reservoir, apparent viscosity will be high so foam will act as an
efficient mobility control agent in transferring injected surfactant solution from high permeability to low permeability media,
such as fractures to matrix in naturally fractured reservoirs.
Foam quality is a basic characteristics and an inseparable term used in foam flooding scenarios. Two distinct foam
regimes of high-quality (high-gas-fraction, dry foam) and low-quality (low-gas-fraction, wet foam) are observed in the
measured data of Fig. 4. Up to a point within the range of fg =0.9-0.95, apparent foam viscosity is increased as the foam
quality increases (when injected gas fraction is increased in the low quality regime); while for fg > 0.9-0.95, foam strength is
decreased by increasing the foam quality (when injected gas fraction is increased in the high-quality regime). At the
boundary of the two regimes, for foam quality of about 0.9-0.95 and for a given surfactant concentration and total superficial
velocity, foam attains its maximum apparent viscosity. The injected gas fraction at the boundary of these two regimes is
known as transition foam quality28, fg*. Thus in principle foam characteristics are different for low quality and high quality
foams. The reduction in apparent foam viscosity for fg > fg* is due to foam coarsening and drying out or losing its surfactant
solution by the limiting capillary pressure mechanism. The foam dry-out is an important feature in reservoir applications in
fractured reservoirs. As fluid velocities in the reservoir decrease and gravity effects become dominant the liquid surfactant
solution will flow down relative to the gas, the quality of foam will increase and the foam will lose its strength; and hence the
importance of this functionality.
As it has been shown in the literature (N2 foam46, methane foam47 and CO2 foam48) that foam is stronger if the surfactant
concentration is above the CMC (critical micelle concentration). For the surfactant solution of the experiments in this study,
the CMC is about 0.1 wt%, while the injected surfactant solution is of 1 wt% concentration, which is well above the CMC.
Therefore, the critical surfactant concentration above which foam strength is independent of surfactant concentration in the
foam model is maintained in these sand-pack experiments.
Foam Model
Various foam models have been proposed in literature to simulate foam flow through porous media, among which the most
promising ones are the mechanistic methods based on bubble population balance and the semi-empirical methods based on
fractional flow theory28,36. The population-balance foam model12,29 has received wide attention because it describes the
generation and decay of foam lamellae and tracks the mobility of foam. The model is capable of simulating foam behavior in
both the high-quality and low-quality regimes upon the modification of net foam generation scheme49. However, difficulties
in obtaining the model parameters, especially at the field scale, and determining the minimum pressure gradient for foam
generation are some drawbacks of the mechanistic foam models.
In the semi-empirical methods, the effect of foam on gas mobility reduction is expressed by a mobility reduction factor,
which reduces the relative permeability of the gas phase as a function of surfactant concentration, oil saturation and capillary
number31. According to the fixed-ܲ כmodel15,32, the effect of limiting capillary pressure above which foam collapses is
included in terms of a critical water saturation, or the foam dry-out expression. As such, the semi-empirical method
incorporates the mechanistic physics of foam transport in the mobility reduction term, using the fractional flow theory and
some basic capillary-governing relationships. The semi-empirical foam model, based on either the fractional flow theory35,
or the CMG-STARS foam model36, has been widely used to describe foam flow through porous media in the oil and gas
industry for a variety of foam-surfactant injection systems.
1
FM = …………………………………………………..…………….... (3)
1 + fmmob ⋅ F1 ⋅ F2 ⋅ F3 ⋅ F4 ⋅ F5 ⋅ F6 ⋅ F7
Various F factors in Eq. 3 account empirically for surfactant concentration, oil saturation, gas flow velocity (both
generation and shear thinning effects), oil composition, salinity and foam coalescence or dry-out that affect foam strength.
The foam dry-out function, F7, is also supported in the simulator even though there is no explicit reference to it in CMG-
STARS instructions. In the following, these parameters are reviewed briefly.
Base Foam Mobility Factor fmmob: The term fmmob is a reference foam mobility reduction factor considered to
introduce a base foam mobility reduction factor without the impact of other factors.
Surfactant Concentration Effect, F1: This is a power-law relationship of surfactant concentration controlled by parameters
fmsurf and epsurf, where fmsurf is the critical surfactant concentration above which gas mobility is independent of surfactant
concentration, Cs. In general, fmsurf depends on the surfactant type and it has a value larger than the critical micelle
concentration, CMC. The exponent term -4<epsurf<4 is a parameter to fit measured foam data.
⎧⎛ C ⎞
epsurf
⎪⎜ s
⎟⎟ for C s < fmsurf
F1 = ⎨ ⎜ fmsurf …………………………………………………………………………. (4)
⎝ ⎠
⎪1 for C s ≥ fmsurf
⎩
Oil Saturation Effect, F2: This function models reduction in foam strength in the presence of oil. fmoil is critical oil
saturation; where for So>fmoil, foam will not be stable and will colescence. floil is lower oil saturation value; where for
So<floil, the presence of oil is considered not to impact the foam rehology. The exponenet is 0<epoil<5.
epoil
⎛ fmoil − S o ⎞
F2 = ⎜⎜ ⎟⎟ .………………………………………………………………………………………… (5)
⎝ fmoil − floil ⎠
Shear Thinning Velocity Effect, F3: This function represents the influence of non-Newtonian power-law shear thinning
foam rheology on the foam strength. Nc is capillary number, 0<fmcap<1 is reference rheology capillary number value and
-10<epcap<10 is the exponent.
epcap
⎛ fmcap ⎞
F3 = ⎜⎜ ⎟
⎟ ……………….……………………………………………………………..……………...... (6)
⎝ Nc ⎠
Foam generation effect, F4: The function models condition for generation or coalescence of foam. 0<fmgcp<1 is critical
generation capillary number, and -10<epgcp<10 is the exponent for the generation capillary number contribution.
epgcp
⎛ fmgcp − N c ⎞
F4 = ⎜⎜ ⎟⎟ ……………….……………………………………………………………..…….….... (7)
⎝ fmgcp ⎠
Foam dry-out effect, F7: This function captures the impact of changes in foam strength as foam quality increases and
foam coalesces or dries out by the limiting capillary pressure phenomenon. fmdry is critical water saturation above which the
maximum foam strength is reached; and below which foam coalesces and weakens. epdry is the exponent for the dry-out
contribution. The modeling of F7 is important in the calibration of the CMG-STARS empirical foam model.
tan −1[epdry(S w − fmdry)]
F7 = 0.5 + ……….……………………………………………………..……………. (8)
π
The behavior of F7 for a hypothetical example of fmdry=0.3 and epdry=10, 100, 1000 is shown in Fig. 5. The figure
indicates that foam dry-out starts at Sw*=0.33 and foam coalesces rapidly below Sw*=0.33. The influence of fmdry of 0.1, 0.3,
0.5 at epdry=1000 is also shown in Fig. 5, with the intention to emphasize the impact of foam dry-out critical saturation.
F5 and F6: These functions relate to specific oil component and salt concentration effects, respectively. They could be
included in the foam model if the corresponding laboratory or field data are available to calibrate these fucntion parameters.
Gas (or foam) fractional flow, fg, is also function of Sw when other parameters remain unchanged:
ug 1
f g (S w ) = = ……………………………………………………..…………. (13)
uw + u g k rw ( S w ) μg
1+ ⋅
μw k rgf ( S w , u g )
Corey-type expressions are used for water and gas relative permeabilities. Foam relative permeability is given by Eq. 2.
Without considering the shear thinning function F3:
nw
⎛ S w − S wc ⎞
k rw = 0 ⎜ ⎟
k rw ……….……………………………………………………..…………………….. (14)
⎜1 − S − S ⎟
⎝ gr wc ⎠
ng
⎛ S w − S wc ⎞ 1
k rgf = 0 ⎜1 −
k rg ⎟ ⋅ …………………..………….. (15)
⎜ 1 − S gr − S wc ⎟ ⎧ tan −1[epdry( S − fmdry)] ⎫
⎝ ⎠ 1 + fmmob ⋅ ⎨0.5 + w
⎬
⎩ π ⎭
Combining Eqs. 12-14 results in a relationship for water saturation, which is the common link between all the equations
listed above:
1
⎡ μ w (1 − f g ) ⎤ nw
S w = S wc + (1 − S gr − S wc ) ⎢ 0 ⎥ ……….…………………………………………..………………. (16)
⎢⎣ k rw μ foam,app ⎥⎦
The solution for fmdry is obtained from a cross plot of fmmob vs. fmdry in Fig. 6 with contour plots of foam quality, fg,
and apparent foam viscosity, μaf. The rest of the parameters are listed in Table 3, which are based on published results30,40.
The selected value of epdry=1000 in Table 3 has uncertainty and imparts a degree of non-uniqueness to the solution. The
determination of fmdry is done by using the experimental data with a total superficial velocity of 31.0 ft/day. This results in
fg=0.88 and μaf=625 cp. In these calculations, it is assumed that the measurement with the highest apparent foam viscosity in
כ
Fig. 4 to be at the transition foam quality (݂= כ0.88 and ߤ =625 cp) because the foam dry-out behavior is not observed yet.
There are intricacies in the nature of these calculations, and some non-uniqueness and numerical artifacts could be present50.
Fig. 7 compares the analytical foam modeling results with experimental data, using the derived values of fmmob=38,800
and fmdry=0.044, as well as other parameters listed in Table 3. The computed model results match the trend of the
experimental data with some underestimation at low gas fraction.
epv
⎛ ug ⎞
F3 = ⎜ ⎟ ……….……………………………………………………..…………………………………. (17)
⎜ u g ,ref ⎟
⎝ ⎠
Incorporating the F3 term in the presence of dry-out function F7, alters Eq. 15 to:
ng
⎛ S w − S wc ⎞⎟ 1
0 ⎜
k rgf = k rg 1− ⋅ ………………...… (18)
⎜ 1 − S gr − S wc ⎟⎠ ug ⎧ tan −1[epdry(S w − fmdry)] ⎫
⎝ 1 + fmmob ⋅ ( ) epv ⋅ ⎨0.5 + ⎬
u g ,ref ⎩ π ⎭
Rearranging and substituting for various terms:
ng
⎧ 1 ⎫
0 μ
k rg af ⎪ ⎡ μ w (1 − f g ) ⎤ nw ⎪
⎨1 − ⎢ 0 ⎥ ⎬ −1
μg fg ⎪ ⎢⎣ k rw μ af ⎥⎦ ⎪ epv
⎩ ⎭ ⎛ ug ⎞
= fmmob ⋅ ⎜ ⎟ ………………….. (19)
⎧ 1 ⎫ ⎜ u g ,ref ⎟
⎡ μ w (1 − f g ) ⎤ nw ⎝ ⎠
⎪ ⎪
tan −1 ⎨epdry{S wc + (1 − S gr − S wc ) ⎢ 0 ⎥ − fmdry}⎬
⎪ ⎣⎢ k rw μ af ⎦⎥ ⎪
0.5 + ⎩ ⎭
π
According to Eq. 19, a linear regression in the form of Eq. 20 can be used to estimate parameters fmmob and the shear
thinning exponent epv, where
⎛ fmmob ⎞
y = epv ⋅ x + log ⎜ ⎟ …………………………………………………………………………………..… (20)
⎜u epv ⎟
⎝ g , ref ⎠
x = log(u g ) ……….……………………………………………………..…………………………………....... (21)
⎧ ⎫
⎪ ⎧ 1 ⎫
ng ⎪
⎪ 0 μ
k rg ⎪ ⎡ μ w (1 − f g ) ⎤ nw ⎪ ⎪
⎪ af
⎨1 − ⎢ 0 ⎥ ⎬ −1 ⎪
⎪ μg fg ⎪ ⎣⎢ k rw μ af ⎦⎥ ⎪ ⎪
⎪ ⎩ ⎭ ⎪
y = log ⎨ ⎬ ……….…………………..…. (22)
⎪ ⎧ 1 ⎫⎪
⎪ ⎪ ⎡ μ w (1 − f g ) ⎤ nw ⎪
−1
tan ⎨epdry{S wc + (1 − S gr − S wc ) ⎢ 0 ⎥ − fmdry}⎬ ⎪
⎪ ⎪
⎪ ⎪ ⎢⎣ k rw μ af ⎥⎦ ⎪⎪
⎪ 0.5 + ⎩ ⎭⎪
⎩ π ⎭
For these calculations, fg=0.7 and other needed parameters are from Table 3 for the nitrogen-surfactant foam experimental
data of this paper. According to the regression curve indicated in Fig. 8, the parameters are:
epv = −0.8693 ……….……………………………………………………..………………………………….... (23)
fmmob
log = 5.6883 ……….……………………………………………………..……………………........ (24)
(u g ,ref ) epv
As an example, if ug,ref is equal to 21.7 ft/day for total superficial velocity of 31.0 ft/day, then fmmob=33,614 for the
estimated value of fmdry=0.044 from the previous analytical section. Using these computed parameters, a plot of apparent
foam viscosity versus foam quality is shown in Fig. 9 along with measured experimental data45. Figure 9 exhibits a less sharp
transition between the high quality and low quality regimes. This is due to the fact that at a fixed total velocity, a decrease in
gas fraction increases mobility reduction as a result of the shear-thinning behavior. To improve the estimation of the shear
thinning parameters, several sets of experiments at fixed gas flow rates could be conducted, where the F3 term in each set of
experiments is kept constant so that fmmob and fmdry can be regressed upon more rigorously.
SPE-169888-MS 7
Model Setup.
A vertical Cartesian 1-D model is considered to capture the physics of the foam flow problem (Fig. 10 shows an example
with 18x1x1 gridblocks; however, the actual model used in simulations is 100x1x1). The simulation model has the same
cross sectional area and height as the experimental sand-pack column. Because injection and production in the laboratory
setup occur uniformly through the entire inlet and outlet faces of the column, these faces are represented by production and
injection wells in this 1-D numerical simulation model. Fluid is injected into the bottom most gridblock through a well
located there, which then flows across the whole cross section into the top most girdblock of the simulation model and is
produced from a production well in that gridblock. To make the simulation model as close as possible to laboratory
experiment conditions where injection and production occur nearly uniformly across the inlet and outlet faces of the sand-
pack column, the first and the last gridblocks are considered to be thin and of high permeability for minimal pressure drops
within these well gridblocks.
Foam is coinjected through nitrogen gas and aqueous surfactant solution streams. It is assumed that surfactant
concentration does not affect the gas-water interfacial tension; thus surfactant is considered to be a water-like component and
act as a foam bubble stabilizer agent. For modeling purposes, surfactant is defined either as a component in gas or in liquid
with proper definitions of the associated k-values.
For each gridblock and at a certain time step, all the F functions in the empirical foam model of Eq. 2 along with the
computed fmmob value from the analytical section are used in the calculation of the interpolation parameter, FM. Depending
on the case, some of these F terms are absent. For example in the case with no oil, F2 and F5 are not considered in the
calculations. A set of foam-gas relative permeabilities are assigned based on the value of the interpolation foam mobility
reduction parameter, FM, and each set is characterized by a corresponding DTRAPN value (a parameter specific to CMG-
STARS). The first set, which is the no foam case, has a DTRAPN value of one and is the main unaltered gas-water relative
permeability curves. DTRAPN values decrease accordingly for the later sets, as apparent foam viscosity increases and gas-
water relative permeability decreases. For each gridblock, in general, an interpolation scheme between pairs of DTRAPN
values is required. Furthermore, because two values of water saturation give the same apparent foam viscosity (one in the
low quality wet foam region and another in the high quality dry foam region), as shown in Fig. 11, the F7 function in the
empirical foam model is conditioned to represent this phenomenon in the simulation model.
For each measured data point, all values of F’s are calculated and used along with the calculated value of
fmmob=219,974 to find the related FM or DTRAPN values. Functions F1, F3, and F7 are the only functions considered in this
modeling because the experiments are under steady-state flow conditions at high surfactant concentration and without any oil
present. Modeling the F3 function requires knowledge of capillary number to determine fmcap. The simulator calculates
8 SPE-169888-MS
capillary numbers differently than the ones stated in the analytical foam modeling section. Local capillary number for
multiphase flow in the simulator is computed using the pressure gradient-based definition:
k∇p
NC = ……………………………………………………………………………………………………..… (26)
σ
This contrasts to the corresponding velocity-based definition of the analytical foam model:
vg μ g
NC = …………………………………………………………………………………………...…………. (27)
σ
However, the pressure gradient-based definition can be converted to the velocity-based definition using the Darcy law for
pressure drop and substituting it in the pressure gradient-based capillary number definition. Thus:
vg μ g
NC = …………………………………………………………………………………………………......... (28)
k rgf σ
Gas viscosity is constant in this modeling, and the last point on the gas (foam) relative permeability for each experimental
data point is estimated using
μg f g
krgf (Sw , ug ) = ……………………………………………………………………………………………...…. (29)
μaf
The Euclidean averaging of the capillary number calculation in the CMG-STARS simulator makes the above approach
not exactly the same as the local capillary number, where the local capillary number for a gridblock is the average of the local
capillary numbers of all faces connecting to that gridblock. Despite this inherent approximation, the above procedure for
capillary computation modeling gives the closest approximation for the CMG-STARS local capillary number. This same
definition is used for the calculation of the reference velocity, fmcap. When fmcap and the corresponding capillary number
of each lab velocity data point are known, F3 can be calculated as follows:
u ref μ
fmcap = ………………………………………………………………………………………...….. (30)
σ gw k rg ,ref
and, ut=ug/fg = 2.5/0.7 = 3.57 ft/day. This value of total velocity is close to the first data point of laboratory measurement
with a local velocity of 4.18 ft/day. Therefore, the apparent foam viscosity of 2,391 cp of the measured data point at the
lowest total superficial velocity of 4.2 ft/day in Fig. 3 can be used to approximate gas relative permeability at the reference
velocity. The calculated foam relative permeability from Eq. 29 is equal to 5.83x10-6, which results in fmcap=3.025x10-3 for
Eq. 30.
because in actual field injections, the flow rate in the immediate vicinity of injection wells could be very high, imparting the
highest shear thinning effect for injectivity considerations.
Conclusions
1. Experimental data are obtained for steady-state foam-surfactant solution with a proprietary formulation of combined
zwitterionic, anionic and cationic surfactants in a high permeability sand-pack column.
2. Measured apparent foam viscosity data include experiments at fixed foam quality with variable injection rate, and
experiments at constant injection rate with varying foam quality. Non-Newtonian foam rheology with shear thinning
effect is measured, and foam dry-out behavior is captured at high foam quality limits.
3. An analytical method for fitting the parameters of the empirical foam model is presented to determine certain foam
model parameters as a pre-set for use in the CMG-STARS reservoir simulator. A unique group of parameters is
illustrated to form the basis of analytical modeling.
4. The critical surfactant concentration, fmsurf, in the foam model is at least one order of magnitude above the CMC. Thus,
the concentration dependency of the calibrated foam model is not considered explicitly in this study.
5. The shear thinning effect is modeled adequately, using the empirical foam model of CMG-STARS commercial reservoir
simulator. The simulations necessitate capturing the manner that numerical algorithms of the simulator represent local
capillary numbers.
6. The foam dry-out characteristics imparted by the mechanism of limiting capillary pressure is augmented in the empirical
foam model, and the related fmdry parameter is calibrated against experimental data.
7. The constructed empirical foam model is shown to function within the framework of CMG-STARS reservoir simulator
and predict adequately apparent foam viscosities measured in laboratory experiments.
Nomenclature
C = mass concentration, wt%
f = fractional flow
FM = a dimensionless interpolation factor in CMG-STARSTM foam model
k = permeability, md
kr = relative permeability
krgcw = end-point gas (foam) relative permeability in CMG-STARS TM
0
k rw = end-point relative permeability of aqueous phase
0
k rg = end-point relative permeability of gaseous phase
Nc = capillary number
p = pressure, psi
Pc = capillary pressure, psi
pc* = limiting capillary pressure, psi
u = superficial (Darcy) velocity, ft/d
S = saturation
μ = viscosity, cp
μaf = apparent foam viscosity, cp
10 SPE-169888-MS
Superscripts
f = with foam
ng = exponent in krg curve
nw = exponent in krw curve
Subscripts
g = gaseous phase
rg = residual gas
s = surfactant
t = total (water + gas)
w = aqueous phase
wc = connate water
References
1. French, T., Broz, J., Lorenz, P., and Bertus, K.: "Use of Emulsions for Mobility Control during Steamooding," SPE California
Regional Meeting, 2-4 April 1986, Oakland, California.
2. Chambers, D.: "Foams for Well Stimulation. Foams: Fundamentals and Applications in the Petroleum Industry," L. Schramm (ed.),
American Chemical Society, 1994, Washington, DC, ACS Symposium No. 242.
3. Millard, M., Getzlaf, D., and Griffth, J.: "Selection of Foam Flushes and Foamed Cement Proves Effective for Remedial Operations
of Low-Pressured Formations Case Histories," SPE Production and Operations Symposium, 22-25 March 2003, Oklahoma City,
Oklahoma.
4. Heller, J. P.: "CO2 Foams in Enhanced Oil-Recovery," Foams: Fundamentals and Applications in the Petroleum Industry, 1994,
242, 201-234.
5. Li, R.F., Yan, W., Liu, S.H., Hirasaki, G.J., Miller, C.A.: "Foam Mobility Control for Surfactant Enhanced Oil Recovery," SPEJ,
2010, 15, 4, 934-948.
6. Lawson, J. and Reisberg, J.: "Alternative Slugs of Gas and Dilute Surfactant for Mobility Control During Chemical Flooding,"
SPE/DOE Enhanced Oil Recovery Symposium, 20-23 April 1980, Tulsa, Oklahoma.
7. Djabbarah, N., Weber, S., Freeman, D., Muscatello, J., Ashbaugh, J., and Covington, T.: "Laboratory Design and Field
Demonstration of Steam Diversion With Foam," SPE California Regional Meeting, 4-6 April 1990, Ventura, California.
8. Hirasaki, G., Miller, C., Szafranski, R., Tanzil, D., Lawson, J., Meinardus, H., Jin, M., Londergan, J., Jackson, R., Pope, G., and
Wade, W.: "Field Demonstration of the Surfactant/Foam Process for Aquifer Remediation," SPE Annual Technical Conference and
Exhibition, 5-8 October 1997, San Antonio, Texas.
9. Yan, W., Miller, C.A. and Hirasaki, G.J.: “Foam Sweep in Fractures for Enhanced Oil Recovery,” Colloids & Surfaces A. Physico.
Eng. Aspects, 282-283 (2006), 348-359.
10. Abbaszadeh, M., Rodriguez-de la Garza, F., Yuan, C., Pope, G.: “Single-Well Simulation Study of Foam EOR in Gas-Cap Oil of
the Naturally-Fractured Cantarell Field,” SPE paper 129867, presented at the SPE Improved Oil recovery Symposium held in Tulsa,
Oklahoma, USA, 24-28 April 2010.
11. Gauglitz, P.A., Friedmann, F., Kam, S.I., Rossen, W.R.: "Foam Generation in Homogeneous Porous Media," Chem. Eng. Sci.,
2002, 57, 19, 4037-4052.
12. Falls, A.H., Hirasaki, G.J., Patzek, T.W., Gauglitz, D.A., Miller, D.D., Ratulowski, T.: "Development of a Mechanistic Foam
Simulator: the Population Balance and Generation by Snap-Off," SPE Reservoir Eng. 1988, 3, 3, 884-892.
13. Hirasaki, G.J.: "The Steam-Foam Process," J. Pet. Tech., 1989, 41, 5, 449-456.
14. Kovscek, A.R., Radke, C.J.: "Fundamentals of Foam Transport in Porous-Media.," Foams: Fundamentals and Applications in the
Petroleum Industry, 1994, 242, 115-163.
15. Khatib, Z., Hirasaki, G., and Falls, A.: "Effects of Capillary Pressure on Coalescence and Phase Mobilities in Foams Flowing
Through Porous Media," SPE Reservoir Engineering, 1988, 3:919-926.
16. Chou, S.I.: “Conditions for Generating Foam in Porous Media,” paper SPE 22628, ATCE, Dallas, Texas, October 6-9, 1991
17. Tanzil, D., Hirasaki, G.J., Miller, C.A.: "Conditions for Foam Generation in Homogeneous Porous Media," SPE/DOE Improved Oil
Recovery Symposium, Tulsa, Oklahoma, 2002.
18. Kovscek, A.R., Bertin, H.J.: "Foam Mobility in Heterogeneous Porous Media - (I: Scaling concepts)," Transport Porous Media,
2003, 52, 1, 17-35.
19. Bernard, G.G., Holm, L.W.: "Effect of Foam on Permeability of Porous Media to Gas," SPEJ, 1964, 4, 3, 267-274.
20. Zhang, C.Y., Oostrom, M., Grate, J.W., Wietsma, T.W., Warner, M.G.: "Liquid CO2 Displacement of Water in a Dual-Permeability
Pore Network Micromodel," Env. Sci. Tech., 2011, 45, 17, 7581-7588.
21. Hirasaki, G.J., Lawson, J.B.: "Mechanisms of Foam Flow in Porous Media - Apparent Viscosity in Smooth Capillaries," SPEJ,
1985, 25, 2, 176-190.
22. Tang, G.Q. and Kovscek, AR.: “Trapped Gas Fraction during Steady-State Foam Flow”, Transport in Porous Media, 2006, 65(2),
287-307.
23. Li, R.F., Hirasaki, G.J., Miller, C.A., Masalmeh, S.K.: "Wettability Alteration and Foam Mobility Control in a Layered 2-D
Heterogeneous System," SPE International Symposium on Oilfield Chemistry, the Woodlands, Texas, USA, 2011.
SPE-169888-MS 11
24. Rossen, W.R.: "Theory of Mobilization Pressure-Gradient of Flowing Foams in Porous Media.1. Incompressible foam," J. Colloid
Interface Sci., 1990, 136, 1, 1-16.
25. De Vries, A. and Wit, K.: "Rheology of Gas/Water Foam in the Quality Range Relevant to Steam Foam," SPE Reservoir
Engineering, 1990, 5.
26. Osterloh, W.T., Jante Jr, M.J.: "Effects of Gas and Liquid Velocity on Steady-State Foam Flow at High Temperature," SPE/DOE
Enhanced Oil Recovery Symposium, Tulsa, Oklahoma, 1992.
27. Rossen, W.R., Wang, M.W.: "Modeling Foams for Acid Diversion," SPEJ, 1999, 06, 92-100.
28. Alvarez, J.M., Rivas, H.J., Rossen, W.R.: "Unified Model for Steady-State Foam Behavior at High and Low Foam Qualities,"
SPEJ, 2001, 6, 3, 325-333.
29. Patzek, T.W.: "Description of Foam Flow in Porous Media by the Population Balance Method," ACS Symp. Ser., 1988, 373, 326-
341.
30. Kam, S.I., Nguyen, Q.P., Li, Q., Rossen, W.R.: "Dynamics Simulations with an Improved Model for Foam Generation," SPEJ,
2007, 12, 1, 35-48.
31. Mohammadi, S.S., Coombe, D.A., Stevenson, V.M.: "Test of Steam-Foam Process for Mobility Control in South-Casper Creek
Reservoir," J. Can. Pet. Tec., 1993, 32, 10, 49-54.
32. Zhou, Z., Rossen, W.R.: "Applying Fractional Flow Theory to Foam Processes at the Limiting Capillary Pressure". SPE Advanced
Technology Series 1995, 3, 1, 154-162.
33. Cheng, L., Reme, A.B., Shan, D., Coombe, D.A., Rossen, W.R.: "Simulating Foam Processes at High and Low Foam Qualities,"
SPE/DOE Improved Oil Recovery Symposium, Tulsa, Oklahoma, 2000.
34. Dholkawala, Z.F., Sarma, H.K., Kam, S.I.: "Application of Fractional Flow Theory to Foams in Porous Media," J. Pet. Sci. Eng.,
2007, 57, 1-2, 152-165.
35. Cheng, L., Reme, A.B., Shan, D., Coombe, D. A., Rossen, W.R.: "Simulating Foam Processes at High and Low Foam Qualities,"
SPE/DOE Improved Oil Recovery Symposium, Tulsa, Oklahoma, 2000.
36. Computer Modeling Group, STARSTM User's Guide. Calgary, Alberta, Canada, 2007.
37. Renkema, W.J., Rossen, W.R.: "Success of Foam SAG Processes in Heterogeneous Reservoirs," SPE Annual Technical Conference
and Exhibition, Society of Petroleum Engineers, Anaheim, California, U.S.A., 2007.
38. Masalmeh, S.K., Wei, L., Blom, C.P.A.: "Mobility Control for Gas Injection in Heterogeneous Carbonate reservoirs: Comparison of
Foams versus Polymers.," SPE Middle East Oil and Gas Show and Conference, Manama, Bahrain, 2011.
39. Zanganeh, M.N., Kam, S.I., LaForce, T.C., Rossen, W.R.: "The Method of Characteristics Applied to Oil Displacement by Foam,"
SPEJ, 2011, 16, 1, 8-23.
40. Ashoori, E.; van der Heijden, T.L.M., Rossen, W.R.: "Fractional Flow Theory of Foam Displacements with Oil,” SPEJ, 2010, 15, 2,
260-273.
41. Liu, M., Andrianov, A., Rossen, W.R.: "Sweep Efficiency in CO2 Foam Simulations with Oil," SPE EUROPEC/EAGE Annual
Conference and Exhibition, Society of Petroleum Engineers: Vienna, Austria, 2011.
42. Vassendan, F., Holt, T., Ghaderi, A., Solheim, A.: "Foam Propagation on Semi-Reservoir Scale," SPE Reservoir Eval. Eng., 1999,
2, 5, 436-441.
43. Chalbaud, C.A., Moreno, R.A., Alvarez, J.M.: "Simulating Foam Process for a Venezuelan Pilot Test,". SPE Annual Technical
Conference and Exhibition, San Antonio, Texas, 2002.
44. Skauge, A., Aarra, M.G., Surguchev, L., Martinsen, H. A., Rasmussen, L.: " Foam-Assisted WAG: Experience from the Snorre
Field," SPE/DOE Improved Oil Recovery Symposium, Tulsa, Oklahoma, 2002.
45. Ma, K., Lopez-Salinas, Jose L, Puerto, Maura C. Miller, Clarence A. Biswal, S.L. and Hirasaki, G.J.: "Estimation of Parameters for
Simulation of Steady State Foam Flow in Porous Media. Part 1: The Dry-Out Effect," Energy Fuels of ACS publication, 2013, 27,
2363-2375.
46. Kovscek, A.R., Apaydin, O.G.: "Surfactant Concentration and End Effects on Foam Flow in Porous Media," Transport Porous
Media, 2001, 43, 3, 511-536.
47. Mannhardt, K., Svorstol, I.: "Surfactant Concentration for Foam Formation and Propagation in Snorre Reservoir Core," J. Pet. Sci.
Eng., 2001, 30, 2, 105-119.
48. Lee, H.O., Heller, J.P.: "Laboratory Measurements of CO2-Foam Mobility," SPE Reservoir Eng. 1990, 5, 2, 193-197.
49. Chen, Q., Gerritsen, M.G., Kovscek, A.R.: "Modeling Foam Displacement with the Local-Equilibrium Approximation: Theory and
Experimental Verification," SPEJ, 2010, 15, 1, 171-183.
50. Ma, K., Farajzadeh, R., Lopez-Salinas, J.L., Miller, C.A., Biswal, L.S., Hirassaki, G.I.: “Estimation of Parameters for the
Simulation of Foam Flow Through Porous Media: Part 3; Non-uniqness, Numerical Artifact and Sensitivity,” SPE paper 165263,
presented at the SPE Enhanced Oil Recovery Conference held in Kula Lumpur, Malaysia, 2-4 July 2013
Acknowledgement
We thank the management of PEMEX E&P for permission to publish this paper.
12 SPE-169888-MS
Experimental set up
N2
Relief
valve
Surfactant
pump
Second P
E
Porous media
section Pressure
holder
transducer
First
section P P
Gas flow E E
controller
N2
P
E
Heat in
Oven
Thermocouple T Heat out
Figure 1 - Schematic of the apparatus for foam flooding experiments in sand pack system
SPE-169888-MS 13
Rubber stopper
Sieve
Rubber stopper
Mesh 200
5 1/2
Second Section
(Downstream)
Silica sand Mesh 20-40
Internal
Taps
First Section
(Upstream)
Sieve
Mesh 200 Rubber stopper
5 1/2
Rubber stopper
Figure 3 - Computed foam apparent viscosity as a function of gas quality and shear rate
Figure 4 - Measured apparent foam viscosity as a function of foam quality at total injection rate of 31 ft/d
14 SPE-169888-MS
F7
0.4 0.4
0.3 0.3
fmdry=0.3, epdry=10 fmdry=0.1, epdry=1000
0.2 fmdry=0.3, epdry=100
0.2 fmdry=0.3, epdry=1000
0.1 fmdry=0.3, epdry=1000
0.1 fmdry=0.5, epdry=1000
0 0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Sw Sw
Figure 5 – Effect of foam dry-out parameters on F7. Left = “epdry”, Right = “fmdry”
fmmob = 38,800
fmdry = 0.044
Figure 6 - Estimation of parameters fmmob=38,800 and fmdry=0.044 at total superficial velocity 31.0 ft/day.
Figure 7 – Comparison of experimental result and model fit for total superficial velocity of 31.0 ft/day.
SPE-169888-MS 15
5.4
5.2
4.8
y
4.6
y = -0.8693x + 5.6883
4.4 R² = 0.9988
4.2
4
0 0.5 1 1.5 2
x
Figure 8 - Fit to experimental data at different total velocities and for fixed foam quality of 0.7
3000
u= 31.0 ft/d u data for fg=0.7 model u=52.0 ft/d
model u=32.6 ft/d model u=31.0 ft/d model u=16.4 ft/d
model u=6.7 ft/d model u=4.2 ft/d
2500 4.2 ft/day
Apparent Viscosity ,cp
2000
6.7 ft/day
1500
1000
16.4 ft/day
52.0 ft/day
0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Injected Gas Fraction
45
Figure 9 - Comparison of modeling results and experimental data at various injected velocities and foam qaulities
500
450
400
350
Appareny viscpsity
300
250
200
150
100
50
0
0 0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08 0.09 0.1 0.11 0.12 0.13
Sw
Figure 10 – The sandpack simulation model and Figure 11 - The nature of double-valued apparent foam viscosity
inlet-outlet gridding scheme
16 SPE-169888-MS
Figure 12 - Simulated against the experimental data for the apparent viscosity as a function of total velocity
Figure 13 - Match of experimental data through simulations by CMG-STARS empirical foam model