Академический Документы
Профессиональный Документы
Культура Документы
PII: S0009-2509(18)30110-6
DOI: https://doi.org/10.1016/j.ces.2018.02.048
Reference: CES 14072
Please cite this article as: M. Romero-Flores, A.J. García-Cuéllar, A. Montesinos-Castellanos, J. Luis Lopez-Salinas,
Modeling surfactant adsorption/retention and transport through porous media, Chemical Engineering Science
(2018), doi: https://doi.org/10.1016/j.ces.2018.02.048
This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers
we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and
review of the resulting proof before it is published in its final form. Please note that during the production process
errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
Modeling surfactant adsorption/retention and transport through porous
media
Salinas a *
a
Tecnologico de Monterrey, Escuela de Ingeniería y Ciencias, Ave. Eugenio Garza Sada 2501, Monterrey,
*Corresponding author.
Abstract
The mathematical modeling of adsorption/retention behavior of surface active materials in a porous medium
composed of a complex network of macropores, mesopores, and micropores was studied herein. In this paper,
we propose a model for these processes and discuss selecting boundary conditions for parameter fitting,
analyze tracer and surfactant signal sizes, contrast calculated results of reversible and irreversible adsorption,
and address the difference between local equilibrium and the rate-limited process. We used experimental data
from the literature to adjust the parameters of the proposed model, taking macroporosity and mesoporosity
into account. Our results show that at least two types of porosity should be used for modeling porous media.
Moreover, the boundary condition at the outlet was found to significantly affect the output response. This
effect is greater in systems with low Péclet numbers (high dispersion and/or diffusion, i.e., NPe < 5).
Therefore, an appropriate boundary condition should be used if an analytical solution is employed to fit
experimental parameters for the tracer. In addition, we observed that two input signal characteristics, namely
slug size and rectangular pulse, proved to be of great importance in determining the output response when
they are smaller than the corresponding values that would cause the system to reach adsorption saturation. A
local equilibrium assumption is only valid when the flow conditions result in a Stanton number greater than
1
10. Our model should be helpful in guiding the design of dynamic adsorption experiments and understanding
the ways in which heterogeneities in the rock influence the interpretation of experimental results.
1 Introduction
Crude oil recovery is of the utmost economic importance for producers around the world. Owing to the
depletion of oil resources in some countries, enhanced oil recovery (EOR) techniques are being considered to
increase production rates and overall recovery. Surfactant injection, which is an important EOR technique, is
being applied in reservoirs to alter the interfacial tension (IFT) between oil and reservoir fluids. The residual
oil recovery can be enhanced by modifying mobility; however, the optimal recovery of oil via surfactant
injection requires a comprehensive understanding of the transport of components and phases in porous media.
A common IFT-based EOR process involves injecting chemical components that can modify either the IFT
between oil and the aqueous phase, or the wettability of oil in the rock formation, or both. Injected
hydrotopes, and ionic species are used to achieve this goal. Therefore, the success of the EOR injection
process is strongly related to transport of species through the reservoir. To guarantee a successful extraction,
adsorption/retention or selective adsorption of any of these species must be predicted with certainty. This goal
is challenging due to formulation complexities and rock structure intricacies. Thus, a mathematical model that
is capable of uncoupling all these elements will aid in a better understanding of adsorption mechanisms.
Solute transport in porous media can be analyzed using various approaches, all of which divide pore space or
porosity into different size-based classes. Bai et al. (1997) indicated that different transport processes take
Porosity distributions and structure types inside porous media have been identified in several studies that
explain and model these structures. In different fields, several approaches with different degrees of
simplifications have been used to model flow and transport of components in porous media. These include
studies of flow in naturally-fractured reservoirs (Warren and Root, 1963); studies on the transport of inert
species in porous media (Coats and Smith, 1964); an analysis of the transport of solutes that undergo sorption
2
(Brusseau et al., 1989); studies of fluid transport during miscible displacement (Brigham et al., 1961, Bai and
Elsworth 1995); measurement and modeling of the transport of different species during immiscible
displacement (Mohanty, 1983; Valiollahi et al., 2012), and research on the effects of mineralogy (Liu et al.,
2017). However, computational capabilities in the past have limited some of these studies and models, forcing
researchers to simplify them to reproduce and explain experimental results. For instance, Bai et al. (1997)
coupled the boundary condition with the mass balance in the mesoporous region and assumed the same
concentrations for both the mesoporous region and the solid surface. As a result, their model neglected
Solute transport in porous media is also related to solute adsorption on a solid surface. Surface adsorption of
compounds has been widely studied, and this mechanism can be simplified using an isotherm equation.
Although the adsorption process is time-dependent, an instantaneous equilibrium condition can be used to
simplify the model. Schwartz et al. (2000) used the concept of instantaneous equilibrium to study the
distribution of species within porous media, assuming a linear relation between the adsorbed species and bulk
concentrations. However, this study neglected adsorption and diffusion within the microporous region, which
may become important at such small scales. Van Genuchten and Wierenga (1976) adopted the same
instantaneous adsorption assumption and used the Freundlich isotherm in a porous medium. They also
considered boundary conditions (BCs) that were applicable to semi-infinite media. Lapidus and Amudson
(1952) analyzed the assumption of instantaneous adsorption and compared it with another approach in which
adsorption rate was incorporated in a porosity model. Balhoff et al. (2007) also found that boundary condition
effects play a fundamental role on adsorption in porous media. However, their model was simplified in order
to achieve an analytical solution, and differences were observed at low flow rates. Problems related to semi-
infinite BCs were discussed assuming a single porosity type. Sardin et al. (1991) presented a broader
spectrum of mathematical models and simulated the effects of representing micropores using different
geometries. The ways in which instantaneous equilibrium and adsorption/desorption kinetics affected
adsorbed species were compared; however, their study did not involve any experiments.
In recent years, increased computational power has enabled the use of more complex models, such as those
including non-equilibrium approaches or combining nanopores, micropores, and macropores (Le et al., 2015;
Maraqa and Khashan, 2014). However, complex models are required to gain a better understanding of solute
3
transport in porous media. Such models should include pore distributions from real rock formations and
different adsorption mechanisms. Crude oil recovery operations involve diverse rock and geological
formation types. Different surfactant blends, each presenting its own adsorption mechanism, may be used to
reduce IFT. Therefore, a robust model is required to cover the full spectrum of potential surfactant
formulations.
The aforementioned ideas on pore distribution were used herein to build a model involving macropores,
mesopores, and micropores. The modeling results were compared with experimental data before performing
additional calculations, and the effects of diverse parameters were analyzed using additional results.
2 Model development
Solid porosity can be broadly classified as consisting of connected or unconnected pores and by pore size as
macroporous, mesoporous, or microporous. These different porosity types coexist and interact with each
other. There are two common ways to model a medium with both macroporosity and mesoporosity. One
involves specifying dual porosity with a single permeability, and the other specifying dual porosity with dual
permeability. Another less common approach involves considering the two sections with differing porosities
as a single heterogeneous section, allowing porosity and permeability to vary rapidly and discontinuously
over the entire domain. However, this approach has a high computational cost (Chen, 2006) and is not
considered herein.
Using concepts developed by Salter and Mohanty (1983), a mathematical model was proposed herein to
examine component transport in porous media. This model can be used to understand and predict surfactant
adsorption in consolidated porous media. In addition, the model can be tuned or tested by analyzing
experimental results, and it can be generalized for the liquid phase in the macroporous, mesoporous, and
4
where vi, and fi represent the velocity field, dispersion/diffusion coefficients, and normalized porosity,
respectively. The values of i represent different types of porous regions: i = 1 (macroporous), 2 (mesoporous),
and 3 (microporous). is component A concentration in a region of type m, and and are the
mass transfer coefficient and specific area for mass transfer per unit volume, respectively, between regions l
and m. Regions l and m may be 1, 2, or 3 (porosity types as for i above), 4 for a solid in contact with
macropores, 5 for a solid in contact with mesopores, and 6 for a solid in contact with micropores. Lastly, is
The two LHS terms of equations 1 and 2 represent, from left to right, advection and diffusion. Similarly, the
RHS of equations 1 and 2 contains terms accounting for accumulation, mass exchange between porous
regions, and mass exchange between phases, respectively. Advection is considered for macroporosity and
mesoporosity but not for microporosity (only diffusion is accounted for in equation 3). Each equation includes
mass transfer between the fluid and its respective solid interface (solid–liquid interface). These mass transfer
terms are required in the equations if the solute can be adsorbed. The equilibrium phase term is discarded in
When using the double-permeability model, two different velocities, v1 and v2, arise because of the two
permeability values that are considered. Darcy’s law can be used to determine v1 and v2 for the two
corresponding regions. This approach is similar to the one used for transport in fractured reservoirs, and Jia
and Niu (2012) used a similar approach to model horizontal wells. When considering homogeneous media, v2
In equations 1 and 2, represents the difference between liquid- and solid-phase solute
Some simplifications can be made depending on the system selected. For instance, the solute concentration
may be assumed to equilibrate instantaneously during adsorption with the solid phase (instant local
equilibrium). Hence, assuming that mass exchange sites (areas wherein solute adsorption may occur) are
5
randomly distributed throughout porous media, the equilibrium phase term can be replaced with the solute
These assumptions reduce the number of differential equations for each solute in porous media to three. After
simplification, the model is easier to handle and the number of parameters to be estimated is reduced. In these
equations , and represent the derivative of the adsorption isotherm with respect to aqueous phase
solute concentration for the macroporous, mesoporous, and microporous regions, respectively.
The simplification of equations 4–6 when complete terms are neglected (e.g., eliminating mesoporosity and
microporosity) can lead to various models used by researchers in the past that have been shown to have
An advantage of the model presented herein is that it includes the previously discussed pore networks.
Therefore, it can be used to adequately represent surfactant flow through intricate porous media.
The solute adsorption process is complex; thus, dealing with surface active materials or polymers cannot be
adequately described using a single isotherm. In order to capture this complex behavior, most models use the
Langmuir isotherm and its derivative. Lee et al. (2005) have found that this isotherm is adequate for systems
6
where a is the adsorption in mg/g of rock, A is the adsorption coefficient, k is the desorption coefficient, and
For comparison, the Redlich–Peterson isotherm is used herein (Maurya and Mittal, 2006):
where and are the first- and second-term adsorption coefficients, and are the first- and second-
term desorption coefficients, and and are the first- and second-term exponential coefficients.
In addition, a hybrid Langmuir isotherm may be used. For this isotherm in macroporosity and mesoporosity,
where is adsorption for region i, Ai,I and Ai,II are the adsorption coefficients, and ki,I and ki,II are the
Finally, a network with only macroporosity and mesoporosity was selected to evaluate the appropriateness of
our assumption of non-local equilibrium (NE). The process is described via equations 1, 2, 15, and 16:
7
In the resulting model, concentration is the solute concentration in a solid contacting the liquid phase in
the macroporous region. Concentration is the solute concentration in a solid contacting the liquid phase in
The equations describing one-dimensional solute transport in a porous medium are recast in a dimensionless
where and are dimensionless velocity terms for the macroporous and mesoporous regions, respectively
(for the cases studied here and ), and τ are the dimensionless distance and time, respectively
(this dimensionless time is also known as pore volume or PV), is the dimensionless radius of spheres
forming the microporous solid aggregate, is the Stanton number, and is the Péclet number, where
8
Here, x is the system coordinate, t represents time, is the system velocity, L is the actual core length (L is the
characteristic length used to perform dimensionless analysis), R is the radius of the spheres forming the
For the non-equilibrium system, equations 1, 2, 15, and 16 take the following forms:
When modeling the transport of a conservative tracer or inert solute, the values of F’4 and F’5 in equations 27
When solving the model, Robin or third-type BCs are considered for the injection, and Neumann or second-
type BCs are considered for production (outlet). Upstream perturbations include rectangular steps or
rectangular impulses with normalized concentrations in the range from zero to one directly injected into the
flowing fraction (i.e., into the macroporous region). Walls are considered to be impermeable.
The Robin and Neumann BCs are known as Danckwerts BCs (Danckwerts, 1953) in a mass balance context.
9
Several studies concerning appropriate BCs are available, including Hulburt (1944) and Danckwerts (1953);
more recent discussions have been reported by Mott and Green (2015) on this aspect.
Danckwerts BCs are used in this study because they are appropriate for finite systems, such as those typically
found in laboratories when working with core materials for reservoir applications, and they satisfy the entire
For initial solute concentrations, the following initial conditions are used:
, (31)
(32)
(33)
Additional BCs for the microporosity-mesoporosity interface exist. Convective mass transfer between
mesoporous and microporous regions equals the diffusion at the aggregate surface wherein microporosity
Microporous structure of the solid aggregate can be represented as a rectangular slab-type structure when n =
1, as a cylinder when n = 2, and as a sphere when n = 3. A reversible Langmuir isotherm was used herein.
10
3 Methodology for model application
Various experimental datasets from the literature (Lopez-Salinas, 2013; Mannhardt et al., 1992; Mannhardt
and Novosad, 1991) were studied to test if model simplifications would reproduce observed behavior. In this
section, macroporosity and mesoporosity, along with one-dimensional (1-D) transport and local equilibrium
(equations 17 and 18 with initial conditions and BCs as established in equations 31‒33) were assumed. For
The model was adjusted by comparing the experimental datasets from three references (Table 1) using model
calculations resulting from the best fit of the adsorption and desorption coefficients. The remaining
parameters were obtained directly from the corresponding references. The parameters used for each
comparison (Nst, NPe, f1, f2, and φ) are shown in the results and discussion section.
11
3.2 Adsorption model analysis
After adjusting the model, six key elements affecting adsorption were analyzed to determine their importance
in predicting transport in porous media. Tables 2 and 3 present the parameter values used in each case. In the
following subsections, the importance of each selected element are presented: isotherm model, isotherm
parameters, downstream boundary condition, slug size, non-local equilibrium, and porosity.
Studies in the literature have reported the effect of isotherm characteristics, such as reversibility,
irreversibility, and pseudo-hysteresis in heterogeneous systems, and have applied these characteristics to
porous media for both batch and flow-through methods (e.g., Limousin et al., 2007). However, the effect of
the porosity scale as the isotherm model changes has not been as thoroughly studied. Here, we analyze the
effect of the isotherm model by comparing two irreversible isotherms, the Langmuir (equation 8) and
Redlich–Peterson (equation 10) isotherms, for the cases wherein macroporosity and mesoporosity (equations
Studies reporting the surfactant performance under different sacrificial agent or polymer conditions, wherein
only the isotherm’s plateau region was considered, are available (Budhathoki et al., 2016; ShamsiJazeyi et al.,
2014). Considering this practice, we performed a test to establish the importance of considering the entire
isotherm in predicting adsorption. Using our model, we varied the IP on the adsorption process; the results
were obtained and examined with two different irreversible Langmuir isotherms showing the same surfactant
concentration in the solid phase at equilibrium, with an aqueous normalized concentration equal to one.
We considered the length of porous media as both finite and infinite to examine the effect of downstream
BCs. The infinite length approach was previously considered by Balhoff et al. (2007) and van Genuchten and
Wierenga (1976). The adsorption of a solute at x = 35 cm in a system of length L = 35 cm was compared with
solute adsorption at x = 35 cm in a longer system (5 L ≈ 178 cm), which can be considered to represent an
12
infinite BC. These tests were performed using an irreversible Langmuir isotherm with high (NPe = 5) and low
The effect of slug size (rectangular signal) on adsorption model calculations in the unsaturated case was tested
Both local and NE assumptions will be applied to the adsorption model using irreversible Langmuir
isotherms. Results were compared to determine when a non-equilibrium assumption was important. Only
The effect of microporosity in adsorption processes has often been neglected under the assumption that it has
a weak impact. However, this may not be true and could depend on the proportion of microporosity in the
porous network and rock properties. To test this effect, equations 17–19 was solved using irreversible
Langmuir isotherms.
We numerically solved a system of one-dimensional partial differential equations using the commercial
The basis for COMSOL Multiphysics is the finite element method, which is used for spatial discretization. To
discretize time, the software uses a backward differentiation formulation with adaptive time steps.
We found that this approach was appropriate to satisfactorily solve the proposed model in all cases.
13
Table 2 Adsorption key elements for analysis and model parameters used in calculations.
14
Adsorption model element Tag Nst 1,2 Nst 2,3 Nst 1,4 Nst 2,5 NPe,1 NPe 2,3 A1 k1 n1 A2 k2 n2 PV
Isotherm
Redlich–Peterson IT-RP 1 - - - 10 - 6 5.17 2 0.21 0.28 1 3
Type
Langmuir IT-T 1 - - - 10 - 0 0 1 - - - 3
DB-L1-Pe5 1 - - - 5 - 1.9 4 1 - - - 3
Downstream
DB-L1-Pe20 1 - - - 20 - 1.9 4 1 - - - 3
Boundary
DB-L5-Pe5 1 - - - 5 - 1.9 4 1 - - - 3
Condition
DB-L5-Pe20 1 - - - 20 - 1.9 4 1 - - - 3
NE1-T 1 - - - 20 - 0 0 1 - - - 1.46
15
a
The values for Pore Volume (PV) were 0.5, 1, 1.5, 2, 2.5, and 3.
b
The values for Nst14 = Nst25 were 0.1, 0.5, 1, and 10.
c
The values for Nst25 were 0.1 and 10.
- Denotes a parameter that does not apply for that particular case
16
Table 3 Porosity (φ), macroporosity fraction (f1), and mesoporosity fraction (f2) used for analyzing key element.
Isotherm type
Isotherm parameters
0.45 0.9 0.1
Downstream boundary
Slug size
P-M,m-T
0.5 0.5 0.5
P-M,m-S
Porosity
P-M,m,
0.5 0.5 0.25
P-M,m,S
The values of coefficient of determination (R2) in Tables 4 and 5 result from comparing experimental and
model predicted values of concentration of component A ( and , respectively) using the model
parameters shown for each set of N experimental observations (belonging to the same Tag) according to the
following equation.
Experimental results (Lopez-Salinas, 2013) were used to adjust model parameters (A and k), which are shown
in Table 4.
17
Table 4 Adjusted model parameters using irreversible isotherm for comparison with experimental results
Coefficient of
Tag NSt NPe φ f1 A k PV
determination (R2)
Figure 1 shows the comparison of model results to experimental data of a normalized (C0 = 1% wt)
concentration as a function of PV. As expected, the model predicts that the tracer exits first and is followed by
the surfactant. A delay in the response exists for C1-T model results with respect to experimental results. The
high macroporosity in C2 (90%) resulted in a rapid and total recovery of both the tracer and the surfactant.
The C1-S model results show good agreement with experimental results. The difference between model and
experimental results for C1-T may be caused by an erroneous determination of rock properties, such as
1.0 1.0
0.8 0.8
0.6 0.6
C/C0
C2-T
C/C0
0 1 2 3 4 5 6 7 8 9 10 0 1 2 3 4 5 6 7 8 9 10
PV PV
(a) (b)
Fig. 1 Exit concentration (C/C0) as a function of pore volume (PV). Comparison between model results and experimental
results obtained for (a) C1-T (Tracer) and C1-S (Surfactant) and (b) C2-T (Tracer) and C2-S (Surfactant).
We used experimental results (Mannhardt et al., 1992; Mannhardt and Novosad, 1991) to adjust hybrid IP,
Table 5 Adjusted model parameters using hybrid isotherm for comparison with experimental results.
18
Macroporous Mesoporous region
Coefficient of
Tag Nst Npe φ f1 region (M) (m)
PV determination
A1 k1 A2 k2 A1 k1 A2 k2 (R2)
Figure 2 shows how the model results compare with the experimental data of a normalized concentration as a
function of PV. For C3-S, differences appear after 5 PV. It is also important to note that after this PV value, a
constant experimental outlet surfactant concentration is observed. The results from C3-T seem to have a better
agreement with experimental results than C3-S. For C4-S, after adjusting our model parameters to the
experimental data, we obtained a better agreement than with the model Mannhardt proposed for PV > 2.5.
Considering mesoporosity in our model may have resulted in this improved agreement.
C/C0
0 1 2 3 4 5 6 7 8 9 10 0 1 2 3 4 5
PV PV
(a) (b)
Fig. 2 Exit concentration (C/C0) as a function of pore volume (PV). Comparison between computed (our model and
Mannhardt’s model) and experimental results obtained for (a) C3-T (Tracer) and C3-S (Surfactant) and (b) C4-T (Tracer)
19
4.1 Model element analysis
In Fig. 3, the area under each component’s curve is the fraction of mass produced, and we used the difference
between tracer and surfactant to quantify adsorption. When the Redlich–Peterson isotherm is used, a delay in
the surfactant signal reaching the outlet is observed. This isotherm also results in a steeper slope compared to
that of Langmuir isotherm, resulting in a different adsorption prediction with the two isotherms. Thus, the
isotherm model effect must be considered to track surfactant transport in porous media.
Table 6 shows adsorption for both isotherms. It can be seen that, based on the parameters used, the Redlich–
1.0
IT-T
0.8 IT-L
IT-RP
0.6
C/C0
0.4
0.2
0.0
0 1 2 3 4 5 6
PV
Fig. 3 Model prediction of effluent concentration for tracer (IT-T), Langmuir isotherm (IT-L), and Redlich–Peterson (IT-
Table 6 Surfactant adsorption (in mg of surfactant per g of rock) for model prediction when varying isotherm type.
IT-L 0.41
IT-RP 0.52
IT-T 0
20
4.1.2 Isotherm Parameters
Figure 4 (a) shows the model results for two isotherms with identical analytical forms but with parameter
values, as presented for tags IT-A and IT-B in Table 2, plotted together. Both isotherms reach the same solid-
phase concentration at an aqueous phase concentration equal to 1 but present different shapes. Isotherm A has
a steeper slope than isotherm B, resulting in a larger area under the curve before reaching a concentration
equal to 1 in the aqueous phase. Usually, both isotherms would be considered to cause the same adsorption,
but Table 7 shows the difference. In Fig. 4 (b), we see that the response corresponding to isotherm A, with a
larger area under the curve, has a larger adsorption than the response corresponding to isotherm B. It is
therefore possible to infer that the shape of the curve significantly influences the adsorption analysis, even
though the isotherms have the same analytical form. In addition, effluent concentration resulting from using
isotherm A shows a steeper slope than that from using isotherm B. These patterns may change the
interpretation of results and the formulation of conclusions while analyzing experimental results.
1.0
0.4 0.8
0.6
C(s) (mg/g)
C/C0
0.0 0.0
0 1 2 0 1 2 3
C(aq) (% wt) PV
(a) (b)
Fig. 4 Model prediction of (a) solid phase ( ) and aqueous ( ) concentrations for two Langmuir isotherms: isotherm
A (IP-A: A = 10, k = 25.32) and isotherm B (IP-B: A = 1.9, k = 4) and (b) effluent concentration for two types of Langmuir
Table 7 Surfactant adsorption (in mg of surfactant per g of rock) for model prediction when varying isotherm parameters.
IP-A 0.568
21
IP-B 0.488
IP-T 0
Figure 5 shows the effect of considering different BCs at the outlet on surfactant concentration as a function
of PV. Two different core lengths were considered, L = 35 cm (DB-L1) and 5 L ≈ 178 cm (DB-L5). For the
same core position, DB-L5 results show an output signal displaced toward the right with respect to DB-L1
results. This is more evident for a Péclet number of 5. Signal displacement may be caused by the diffusion
affecting surfactant transport in porous media. In the DB-L1 calculations, the χ = 1 position represents the end
of the core, whereas for DB-L5, the same position represents a continuity in the core.
Diffusion and dispersion in the outlet and inlet may modify the results, which in this case resembles a zeroth-
type BC at the outlet. As a result, increasing output signal (surfactant concentration) variation is related to
For Péclet numbers greater than 20, the outlet BC effect seems negligible. In contrast, for systems with low
Péclet numbers (high dispersion and/or diffusion, e.g., NPe < 5), the BC at the outlet greatly affects the
downstream response. Choosing the appropriate BC (i.e., dC/dχ=0 at the outlet of a relatively small core
instead of C=0 at the outlet of a very long core) is a decisive factor when analyzing relatively short cores used
in experimental tests.
1.0
DB-L1-Pe20
0.8 DB-L1-Pe5
DB-L5-Pe20
0.6 DB-L5-Pe5
C/C0
0.4
0.2
0.0
0 1 2 3 4 5 6
PV
22
Fig. 5 Model results using different downstream boundary conditions (continuity and discontinuity) at the location x = 35
cm with two different Péclet numbers (5 and 20) (DB: downstream boundary condition, PV: pore volume, core lengths:
Figures 6 (a) and (b) show the effect of varying slug size (SS) on effluent concentration and surfactant
adsorption respectively. The legend of Fig. 6 (a) shows SS in terms of PV following the SS identifying tag. In
Fig. 6 (a), we see that when SS < 2, an outlet concentration less than 1 is observed. Thus, a complete
saturation of the rock is guaranteed when SS ≥ 2. Figure 6 (b) shows the impact of SS on adsorption.
Saturation is observed for SS > 1. The use of SS values higher than those needed to reach adsorption
saturation is strongly recommended to obtain reliable results and avoid confusing non-saturation with porosity
effects.
1.0 0.5
SS-0.5
0.8 SS-1 0.4
Adsorption (mg/g)
SS-1.5
0.6 SS-2 0.3
C/C0
SS-2.5
0.4 SS-3 0.2
0.2 0.1
0.0 0.0
0 1 2 3 4 5 6 0 1 2 3
PV Slug size (PV)
(a) (b)
Fig. 6 Model prediction of (a) outlet concentration using different slug sizes (SS: slug size, PV: pore volume) and (b)
Figure 7 shows surfactant concentration as a function of PV for varying Nst. Figure 7 (a) shows results for Nst
1,4 = Nst 2,5, whereas in Fig. 7 (b), the value Nst 1,4 = 1 is constant and Nst 2,5 varies. The Stanton number values
can be found in legends used in Fig. 7, following the Nst tag. Figure 7 (a) shows that when both Stanton
numbers increase, the response approaches that of instantaneous equilibrium. Figure 7 (b) shows that
changing Nst 2,5 by a factor of 100 has a small effect on outlet concentration. Hence, Nst 1,4 seems to be the
23
dominant factor in reaching instantaneous equilibration; this may be owing to mass transfer in the
macroporosity section which apparently regulates overall adsorption because it is in direct contact with the
fluid phase. This causes a bottleneck when the Stanton number (Nst14) is less than that of the mesoporosity
section (Nst25).
1.0 1.0
NE1-T NE2-T
0.8 NE1-S-Nst0.1 0.8 NE2-S-Nst0.1
NE1-S-Nst0.5 NE2-S-Nst10
0.6 0.6
NE1-S-Nst1 Instant equilibrium
C/C0
C/C0
0.4 NE1-S-Nst10 0.4
Instant equilibrium
0.2 0.2
0.0 0.0
0 1 2 3 4 5 6 0 1 2 3 4 5
PV PV
(a) (b)
Fig. 7 Model prediction of effluent concentration for tracer (NE1-T and NE2-T) and surfactant (NE2-S and NE2-S) as a
function of pore volume (PV). (a) Results using equal values for Nst14 and Nst25 in non-equilibrium and (b) results using a
constant value of Nst14 = 1 while varying Nst25 in non-equilibrium. In both legends, the value for the varying Stanton
4.1.6 Porosity
The effect of microporosity on model outlet tracer and surfactant concentration prediction for a rectangular
pulse injection is shown in Fig. 8. The figure depicts four results that are identified in its legend as follows.
For tracer (T) and surfactant (S), the response obtained from a medium comprising only macroporosity and
mesoporosity (P-M-m) is compared to the response obtained considering macroporosity, mesoporosity, and
microporosity (P-M-m-μ).
The flowing fraction (f1) presented in Fig. 8 remains the same (i.e., same macroporosity) in all cases, and
when microporosity is considered (P-M,m,µ), a lesser degree of both tracer and surfactant penetration is
observed. In addition, the output signal requires more time to reach a concentration of zero, compared with a
system without microporosity. This may be caused by the adsorption/desorption mechanism in the
24
mesoporosity–microporosity interface, where only diffusion occurs, causing the surfactant to slowly penetrate
the microporosity while leaving it at the same rate; thus, more time is required for the surfactant to completely
leave the system. When only macroporosity and mesoporosity were considered (P-M,m), the concentration
response had a lower maximum compared with when microporosity was included in the model. This may be
caused by the diminishing mesoporosity fraction (from 50% to 25%) when microporosity was included. As a
result, the macroporosity to mesoporosity fraction ratio doubles and the flowing phase enters and leaves the
1.0
P-M,m-S
0.8 P-M,m,-S
P-M,m-T
0.6 P-M,m,-T
C/C0
0.4
0.2
0.0
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
PV
Fig. 8 Model prediction of outlet surfactant and tracer concentrations as a function of pore volume (PV) for various
porous network scenarios. Continuous lines represent surfactant (S) concentration and dashed lines represent tracer (T)
concentration. Porous networks are described in the legend as macroporosity (M), mesoporosity (m), and microporosity
(μ).
5 Conclusions
The model used in this study makes it possible to obtain good fits between experimental and calculated
results. This model addresses, from a theoretical point of view, the ways in which key effects impact the
adsorption phenomena.
Using an isotherm that reaches a maximum before its plateau region contributes to the final adsorption results.
This can be achieved using the Redlich–Peterson isotherm but not the Langmuir isotherm. In general, for
adsorption analysis, when the isotherm curve plateaus, maximum adsorption is considered to have been
25
reached, leaving aside the analysis of an isotherm peak. A correct selection of the appropriate isotherm type is
The outlet BC appears to be insignificant for systems with a large Péclet number; however, the outlet BC has
a significant impact on downstream response for small Péclet systems. Therefore, when using an analytical
solution to fit experimental parameters to the tracer, the appropriate BC should be used.
We observed that input signal characteristics, such as SS or rectangular pulse, proved to be significant for
signals smaller than that required to meet the adsorption saturation. Therefore, it is highly recommended to
A local equilibrium assumption is only valid where flowing conditions result in a Stanton number greater than
10. To correctly apply the instant local equilibrium assumption, high velocity flows (v1 for macro porosity)
must be considered, ensuring that the specific area (av) of the solid surface is higher than that of the
On adjusting its parameters and considering macroporosity and mesoporosity, the proposed model can explain
experimental results that were presented. It is strongly recommended to use at least two types of porosity to
model surfactant transport through porous media. When microporosity was included, a behavior resembling
irreversible adsorption appeared. This may help in differentiating irreversible adsorption from reversible
adsorption present in microporosity. Thus, if microporosity is overlooked, the model results could be
misinterpreted as adsorption.
Acknowledgements
The authors gratefully acknowledge support from Tecnologico de Monterrey through the Energy and Climate
Change Research Group. M. R.-F. gratefully acknowledges the support from Tecnologico de Monterrey
References
26
Bai, M., Elsworth, D., 1995. On the modeling of miscible flow in multi-component porous media. Transp.
Bai, M., Elsworth, D., Inyang, H.I., Roegiers, J.C., 1997. Modeling contaminant migration with linear
Balhoff, M.T., Thomson, K.E., Hjortsø, M., 2007. Coupling pore-scale networks to continuum-scale models
Brigham, W.E., Reed, P.W., Dew, J.N., 1961. Experiments on mixing during miscible displacement in porous
Brusseau, M.L., Jessup, R.E., Rao, P.S.C., 1989. Modeling the transport of solutes influenced by multiprocess
Budhathoki, M., Barnee, S.H.R., Shiau, B.J., Harwell, J.H., 2016. Improved oil recovery by reducing
surfactant adsorption with polyelectrolyte in high saline brine. Colloids Surfaces A Physicochem. Eng.
Chen, Z., Huan, G., Ma Y., 2006. Computational Methods for Multiphase Flows in Porous Media. Society for
Coats, K.H., Smith, B.D., 1964. Dead-end pore volume and dispersion in porous media. Soc. Pet. Eng. J. 4,
73–84. doi:10.2118/647-PA
Danckwerts, P.V., 1953. Continuous flow systems: Distribution of residence times. Chem. Eng. Sci. 2, 1–13.
doi:10.1016/0009-2509(53)80001-1
Hulburt, H.M., 1944. Chemical processes in continuous-flow systems. Ind. Eng. Chem. 36, 1012–1017.
doi:10.1021/ie50419a010
Jia, R.N.Y.M.Y., Niu, F.Z.X.Y.X., 2012. Dual porosity and dual permeability modeling of horizontal well in
Lapidus, L., Amundson, N.R., 1952. Mathematics of adsorption in beds. VI. The effect of longitudinal
diffusion in ion exchange and chromatographic columns. J. Phys. Chem. 56, 984–988.
27
Le, T.D., Moyne, C., Murad, M.A., 2015. A three-scale model for ionic solute transport in swelling clays
doi:10.1016/j.advwatres.2014.10.005
Lee, V.K.C., Porter, J.F., McKay, G., Mathews, A.P., 2005. Application of solid-phase concentration-
dependent HSDM to the acid dye adsorption system. AIChE J. 51, 323–332. doi:10.1002/aic.10290
Limousin, G., Gaudet, J.-P., Charlet, L., Szenknect, S., Barthès, V., Krimissa, M., 2007. Sorption isotherms:
A review on physical bases, modeling and measurement. Appl. Geochemistry 22, 249–275.
doi:10.1016/j.apgeochem.2006.09.010
Liu, M., Shabaninejad, M., Mostaghimi, P., 2017. Impact of mineralogical heterogeneity on reactive transport
Lopez-Salinas, J.L., 2013. Transport of Components and Phases in a Surfactant /Foam EOR Process for a
Mannhardt, K., Novosad, J.J., 1991. Chromatographic effects in flow of a surfactant mixture in a porous
Mannhardt, K., Schramm, L.L., Novosad, J.J., 1992. Adsorption of anionic and amphoteric foam-forming
surfactants on different rock types. Colloids and Surfaces 68, 37–53. doi:10.1016/0166-6622(92)80146-
Maraqa, M.A., Khashan, S.A., 2014. Modeling solute transport affected by heterogeneous sorption kinetics
doi:10.1016/j.jconhyd.2013.11.005
Maurya, N.S., Mittal, A.K., 2006. Applicability of equilibrium isotherm models for the biosorptive uptakes
doi:10.1061/(ASCE)0733-9372(2006)132:12(1589)
Mohanty, K.K., 1983. Multiphase flow in porous media: III. Oil mobilization, transverse dispersion, and
wettability, in: SPE Annual Technical Conference and Exhibition. Society of Petroleum Engineers, San
28
Mott, H. V, Green, Z.A., 2015. On Danckwerts’ boundary conditions for the plug-flow with
Sardin, M., Schweich, D., Leij, F.J., van Genuchten, M.T., 1991. Modeling the nonequilibrium transport of
linearly interacting solutes in porous media: A review. Water Resour. Res. 27, 2287–2307.
doi:10.1029/91WR01034
Schwartz, R.C., Juo, A.S.R., McInnes, K.J., 2000. Estimating parameters for a dual-porosity model to
doi:10.1016/S0022-1694(00)00164-5
ShamsiJazeyi, H., Verduzco, R., Hirasaki, G.J., 2014. Reducing adsorption of anionic surfactant for enhanced
oil recovery: Part I. Competitive adsorption mechanism. Colloids Surfaces A Physicochem. Eng. Asp.
Valiollahi, H., Ziabakhsh, Z., Zitha, P.L.J., 2012. Mathematical modeling of chemical oil-soluble transport for
van Genuchten, M.T., Wierenga, P.J., 1976. Mass transfer studies in sorbing porous media I. Analytical
Warren, J.E., Root, P.J., 1963. The behavior of naturally fractured reservoirs. Soc. Pet. Eng. J. 3, 245–255.
doi:10.2118/426-PA
29
A model of adsorption/retention of surfactants in a complex porous media of macro, meso
and micro pores is proposed.
The model shows agreement with experimental results when macro and meso porosity are
considered.
The proposed model is a tool for guiding the design of dynamic adsorption experiments.
The model will help understanding how rock heterogeneities influence interpretation of
experimental results.
30