You are on page 1of 13

Journal of Natural Gas Science and Engineering 101 (2022) 104519

Contents lists available at ScienceDirect

Journal of Natural Gas Science and Engineering


journal homepage: www.elsevier.com/locate/jngse

Thermodynamic modeling for confined fluids in nanopores using an


external potential-advanced equation of state
Yufeng Chen, Hao Zhan, Zhiyong Zeng *
School of Energy Science and Engineering, Central South University, Changsha, Hunan, 410083, China

A R T I C L E I N F O A B S T R A C T

Keywords: Quantitatively understanding thermodynamic properties of confined fluids in nanoporous media is of great
Nanopores significance to the development of shale gas. Due to the complex intermolecular forces in the nanopore, it is
Confined effect difficult to accurately predict the thermodynamic properties of fluid molecules. Focusing on the fluid-fluid and
Fluid-wall molecules interaction
fluid-wall molecules interactions in shale reservoir system, the thermodynamic model of fluid molecules in pores
Shale gas
is constructed, and a modified equation of state is proposed by systematically coupling the original Soave-
Redlich-Kwong equation of state with Tjatjopoulos-Feke-Mann potential model in this work. The advanced
EoS could facilitate a good prediction on thermodynamic properties of confined fluids without any introduction
of new empirical parameters. For verification, fluid density as the important thermodynamic property was tar­
geted and pure methane at a wide pressure range was employed to represent the fluid. The results indicated that
the calculated densities accord well with the reported ones in the free gas zone. The deviation of discrete density
ranges from 0.239% to 1.7329%. The fluid density distribution in the nanopores is found to be nonuniform,
exhibiting a greater value near the wall than that in the pore center, which would be ascribed to the more
dominant fluid wall molecule interaction. For example, the local density is 16.90 kg/m3 in the pore center, while
it increases to 26.67 kg/m3 in the region which is 0.76 nm to the wall at 350 K, 3 MPa, and 5 nm (radius).
Moreover, effects of other critical factors on fluid density distribution were also conducted, and it was indicated
that higher pressure, lower temperature, and smaller pore size could be favorable for the occurrence of confined
fluid. In general, the novel EoS could provide a quantitative and simple method in predicting the thermodynamic
properties of confined fluids relating to applications of shale gas storage and exploitation.

confined state in the nano-scale pores, and it is categorized into three


types: free gas in pores and fractures, adsorbed gas on the shale particle
1. Introduction
surface and in kerogen, and dissolved gas in kerogen and asphalt (Liu
and Zhang, 2019). It is evident that the nanopores is a critical factor
1.1. Motivation and issue statement
influencing the occurrence characteristics of shale gas, which deserves a
deep evaluation from qualitative and quantitative perspectives. Under­
Based on the BP Statistical Review of World Energy, the nature gas
standing of occurrence state, resource assessment, and production
imports and the foreign-trade dependence (FTD) of China continue to
mechanism at a molecular level is of great significance for the efficient
rise (Fig. 1) (BP, 2020), which poses a great threat to the security of
exploitation of shale gas. In such case, it is indispensable to investigate
energy supply. Recently, shale gas is now deemed as a promising nature
the thermodynamic properties of shale gas because of their impacts on
energy resource for ensuring the cleaner energy supply (Energy Infor­
the component and the mobility of so-called confined fluid (Liu and
mation Administration (EIA), 2018). In China, the current shale gas
Zhang, 2019). At present, previous studies have mostly focused on the
resource is facing a huge unbalance that its reserve has ranked the first in
qualitative exploration of confinement systems by considering factors
the world (31.58 × 1012 m3) while the current production is still staying
like pore size (Tao et al., 2021; Xiong et al., 2017a), particle size (Ji
at a very middle level (Fig. 2). Thus, in-depth studies are urgent to
et al., 2012), pore shape (Liu et al., 2020), surface roughness (Kim et al.,
improve the exploitation of shale gas. Different from the occurrence of
2011), and surface chemical composition (Li et al., 2020). As is known,
conventional nature gas, most shale gas is usually residing in the

* Corresponding author. Permanent address: School of Energy Science and Engineering, Central South University, No. 932 Lushan South Road, Changsha, Hunan,
410083, China.
E-mail address: zengzhiyong@csu.edu.cn (Z. Zeng).

https://doi.org/10.1016/j.jngse.2022.104519
Received 30 November 2021; Received in revised form 18 February 2022; Accepted 8 March 2022
Available online 11 March 2022
1875-5100/© 2022 Elsevier B.V. All rights reserved.
Y. Chen et al. Journal of Natural Gas Science and Engineering 101 (2022) 104519

Nomenclature ext External


w Pore wall
Symbols ff Fluid molecule parameter
T Temperature, K ww Wall molecule parameter
P Pressure, MPa surf Surface
V Molar volume, m3⋅mol-1
R Universal gas constant, m3⋅Pa⋅K-1⋅mol-1 Abbreviations
rc Distance between the molecule and wall, nm FTD foreign-trade dependence
b van der Waals volume, m3⋅mol-1 PC-SAFT Perturbed-chain statistical associating fluid theory
a(T) Attraction parameter, Pa⋅m6⋅mol-2 EoS Equation of state
E Molecule interaction potential, J⋅mol-1 DFT Density functional theory
σ Molecule diameter, nm PR Peng-Robinson
ε Energy parameter, J SRK Soave-Redlich-Kwong
ρ Density, kg⋅m-3 VLE Vapor-liquid equilibrium
Z Compressibility factor CPA Cubic-plus-association
k Boltzmann constant, J⋅K-1 vdW van der Waals
LJ Lennard-Jones
Subscripts TFM Tjatjopoulos-Feke-Mann
attra Attraction GCMC Grand canonical Monte Carlo
repul repulsion MD molecular dynamic
fw Fluid-wall molecules interaction SWCNT Single-walled carbon nanotube
cc Critical properties of confined fluid LCA Local Density Approximation theory

state of fluid molecules, making them bound to or nearby the surface of


the pore wall (Barsotti et al., 2016; Palmer et al., 2011). For another, it
would also weaken the existing fluid-fluid molecules interaction. As a
consequence, there would be an in-homogenous fluid distribution in
nanopores (Jin and Firoozabadi, 2016), namely, confined fluids exhibit
nonuniformity in some thermodynamic properties, especially in the
fluid density (Liu et al., 2018; Wang et al., 2019). Hence, the fluid-wall
molecules interaction is what we concern in this work. The specific goal
of the work is to establish an integrated thermodynamic model by
considering fluid-wall molecules interaction, and further to validate the
proposed model by investigating the thermodynamic properties (refer­
ring particularly to fluid density) of confined fluids.

1.2. Literature background

Based on the objectives of this paper, previous studies referring to


keywords of ‘confinement effect’, ‘thermodynamic properties’ and ‘shale
gas’ have been summarized and reviewed in this section, including
empirical, numerical simulation, and theoretical methods (Liu and
Zhang, 2019).On one hand, Langmuir-related models have been regar­
Fig. 1. The nature gas imports and the foreign-trade dependence of China.
ded as one of the most common empirical methods that can directly
determine adsorption characteristics of shale gas based on the fitting of
shale gas as a type of confined fluid is different from bulk fluids due to massive experimental data (Tang et al., 2017). In the study of Sandoval
the confinement effect. Moreover, the confinement effect would be et al. (2018), six empirical models were adopted to describe the
enhanced with the decreasing pore size, making the forementioned adsorption capacity of shale gas, and reliable results were achieved with
difference more significant. From these perspectives, it is of great an acceptable deviation. However, the requirement of existing experi­
challenge to fully understand the thermodynamic properties of shale gas mental data and the inaccuracy of excess adsorption are two obstacles
and the quantitative evaluation on this is also lacking. Besides this, restricting the application of this empirical method. On the other hand,
thermodynamic properties of fluid molecules in nanopores are also the density functional theory (DFT) and molecule simulation are two nu­
research priority relating to hydrogen storage (Cracknell, 2002; Wil­ merical simulation methods that are able to describe the thermodynamic
liams and Eklund, 2000), catalytic reaction, and gas separation. As properties of confined fluids. The phase behaviors of fluids in nanopores
summarized above, it is fully expected to establish an integrated model have been accurately predicted by Sermoud et al. (2020) and Li et al.
with suitable theoretical principles that can satisfy a quantitative eval­ (2014) using DFT models combined with different equations of state
uation on the thermodynamic properties of confined fluids. (EoS), respectively. Monte Carlo simulation, being a typical molecule
As reflected above, it is evident that the confinement effect is the simulation method, has been adopted by some scholars (Li et al., 2020;
essence when evaluating the thermodynamic properties of confined Lin et al., 2020; Xiong et al., 2017b) to investigate phase behaviors and
fluids. In general, the confinement effect involves the fluid-wall mole­ critical properties of confined fluids, and the effects of relevant factors
cules interaction and the capillary condensation (Barsotti et al., 2016). like temperature, pressure, fluid composition, pore size, and moisture
The former is generally manifested as an attraction that brings two ef­ content could be clearly ascertained by the target simulation method. In
fects as follows. For one thing, it would alter the original equilibrium spite of the attainment of reasonable and accurate results for two

2
Y. Chen et al. Journal of Natural Gas Science and Engineering 101 (2022) 104519

numerical simulation methods, it’s worth noting that there is an obvious there was a strong correlation between the LJ potential and the
disadvantage that they are facing a variety of complex and intensive fluid-fluid molecules interaction. Furthermore, as a developed model for
computing calculations. LJ potential, a Tjatjopoulos-Feke-Mann (TFM) potential model was
Possessing simple formulas and offering flexible solutions, EoS demonstrated to offer a better description about the interaction between
modeling has been proved to be an ideal theoretical method for a sys­ a fluid molecule and the pore wall (Tjatjopoulos et al., 1988). As
tematical description on the overall thermodynamic properties of fluids revealed, this target LJ potential model (TFM) would be a feasible so­
(Liu and Zhang, 2019). In general, there are two categories involved: lution to the description of the expected correction term.
theoretical EoS (e.g., SAFT type) and empirical EoS (e.g., cubic type).
SAFT EoS as a theoretical type is to reveal the molecule interactions, 1.3. Main contributions
which was demonstrated to be reliable with a wider application scope
(Zeng et al., 2008, 2009). However, the corresponding computing pro­ In comparison, the EoS modeling (especially the cubic EoS) is sug­
cedures of this modeling are considered to be more complex. In com­ gested to be a simpler and more convenient method that provides a
parison, cubic EoS as an empirical type is a much simpler modeling that quantitative evaluation on the thermodynamic property of confined
can give an accurate description on fluids in engineering fields, like fluids, exhibiting a more potential application in the oil and gas in­
Peng-Robinson (PR) and Soave-Redlich-Kwong (SRK) EoS (Soave, 1972; dustries. Nevertheless, for the existing cubic EoS modelings, an impor­
Wang and Gmehling, 1999). As a result, cubic EoS seems to be a more tant factor, namely the interaction of fluid-wall molecules, has been
suitable theoretical method. usually ignored, which inevitably leads to the inaccuracy and inappli­
Regarding the fluid molecules in the nanopores, it is inappropriate to cability of the existing models. Thus, the keystone of this work lies in the
directly describe their thermodynamic properties by using original cubic modification of the cubic EoS model by introducing the external po­
EoS owing to the presence of the fluid-wall molecules interaction. In this tential term, aiming at a precise prediction on the thermodynamic
situation, some cubic models have been developed by scholars with the properties of confined fluids. The main contributions of this paper can be
consideration of fluid-wall molecules interaction (Yang and Li, 2020). In summarized as follows:
particular, a correction term was introduced to modify the original cubic
EoS, which was used to reflect this interaction. At present, some previ­ (1) A thermodynamic model of confined fluids has been proposed
ous studies have confirmed the correction term by the regressions of under the consideration of the dual interactions of fluid-fluid and
experimental and molecular simulation data. For instance, for the fluid-wall molecules. Specifically, a cylindrical pore is chosen as
modification of EoS involving van der Waals (vdW) (Zarragoicoechea the research target in this work. A modified EoS model has been
and Kuz, 2002, 2004), SRK (Zuo et al., 2018) and PR (Yang et al., 2019) constructed by introducing an external potential term into the
types. Restricting by the insufficient existing data for regression, the classical SRK EoS, and the corresponding external potential term
accuracy of the resultant correction term is actually to some extent is expressed and calculated by the TFM potential model.
limited for the shale gas system. Moreover, this regression method fails (2) Centering on an important thermodynamic property (i.e., fluid
to provide a good description on the overall thermodynamic properties density), its distribution characteristics of confined fluids are in-
of continuous confined fluids across the nanopores. Therefore, devel­ depth studied using the modified EoS model, and the effects of
oping a more advanced correction term for the modification of the cubic relevant factors including pressure, temperature, and pore size
EoS is urgent for the evaluation of the shale gas system. are also ascertained.
For the determination of advanced correction term, a suitable (3) The results of this work can provide insights on the occurrence
expression on fluid-wall molecules interaction is essential. Based on this, state of confined fluids in nanopores, and also offer guidance on
idealized potential models have been introduced. Specifically, the the resource assessment and the exploitation mechanism of shale
Lennard-Jones (LJ) potential model was adopted by some previous gas.
studies (Cracknell, 2002; Gu, 2001), and it was widely accepted that

Fig. 2. Location of the world’s shale play, the volume of technically recoverable shale gas, and the shale gas production in the major resource countries (U. S. Energy
Information Administration (EIA), n.d.; World Resources Institute, n.d.).

3
Y. Chen et al. Journal of Natural Gas Science and Engineering 101 (2022) 104519

2. Modified cubic EoS with an external potential model a(T)


Pattra = − (4)
V(V + b)
2.1. Physical models and assumptions
Where P is the pressure, MPa; V is the molar volume, m3⋅mol− 1; T is the
Owing to high organics content in shale rocks, most shale gas re­ temperature, K; R represents the universal gas constant, which is 8.314
sources are residing in the adsorbed state and free state. For a better m3⋅Pa⋅K− 1⋅mol− 1; a(T) represents the attraction parameter in SRK EoS,
understanding of the occurrence system, a physical model is proposed to Pa⋅m6⋅mol− 2; b is the van der Waals volume, m3⋅mol− 1;
represent the shale pore, and the detailed sketch map is shown in Fig. 3. In the nanopores, with pore sizes ranging from a few nanometers to
In this study, the physical model has been appropriately simplified to tens of nanometers (Ross and Marc Bustin, 2009), the molecule
facilitate theorization. The model representing the actual pore structure dimension is comparable to the pore size, and the distance between the
will be considered in future studies. Thus, some reasonable assumptions fluid and wall molecules is reduced to the nanometer scale. Thus, the
are made as follows: (1) The actual pores are simplified to smooth cy­ fluid-wall molecules interaction becomes so strong that it can’t be
lindrical pores; (2) Considering that organic matter is the main ignored in nanopores in comparison with the intermolecular force.
component of shale, single-walled carbon nanotube (SWCNT) is selected Considering the joint effect of the fluid-wall molecules interaction,
as a representative object for characterizing shale wall (Zhang et al., fluid-fluid molecules attraction interaction, and fluid-fluid molecules
2003b); (3) The configuration of the fluid molecule is simplified to a repulsion interaction, the expression of pressure is presented as follows,
sphere. In Fig. 3, r and rc represent the pore radius and distance between
the molecule and the wall. P = Prepul + Pattra + Pfw (5)

Where Pfw is originated from the fluid-wall molecules interaction.


2.2. Fluid-wall molecules interaction Based on the above concepts, some scholars have proposed some
modified EoSs to calculate confined fluid thermodynamic properties by
For the Cubic EoS, the pressure is manifested as a combination of a introducing a new term on the basis of the traditional EoS. However,
repulsion pressure (Prepul) and an intermolecular attraction pressure some empirical parameters are added and regressed from the reported
(Pattra). experimental data (Xiong et al., 2021; Yang et al., 2019; Zarragoi­
P = Prepul + Pattra (1) coechea and Kuz, 2002). The development and application of these
modified EoSs may be hampered by the limited experimental data and
In the previous study, some EoSs for real gases are developed, in weak theoretical foundations. This paper is devoted to constructing a
which the semi-empirical equation is widely used for its high conve­ new EoS with a solid theoretical basis and without introducing any
nience and relatively high computational accuracy. The vdW, SRK, and empirical parameters. Then, the key issue is whether another action
PR equations are relatively simple and basic equations, and some should be considered besides the above repulsion and attraction term for
extended equations, such as CPA, and PC-SAFT, are proposed for higher the confined fluids. As mentioned above, a molecule-wall term, abbre­
precision and wider application. The SRK EoS is modified in this study. viated as Pfw, is introduced and studied in this work. Besides, the other
The original SRK EoS is the sum of a repulsion term and an attraction key issue is how to reflect the influence of the fluid-wall molecules
term, it could be expressed as: interaction in the EoS. According to a dimensional analysis of the classic
RT a(T) cubic EoS, both the repulsion and attraction term can be expressed as the
P = Prepul + Pattra = − (2) ratio of energy to volume [E/(V-b)]. In fact, the energy reflects the effect
V − b V(V + b)
of the intermolecular action, and the volume can be regarded as the
RT molecular motion space. In the Pfw, the energy originated from the
Prepul = (3)
V− b fluid-wall molecules interaction is a kind of potential energy due to the
relatively static condition for the wall molecules, and the volume is
determined as the same as the repulsion term, which considers the
volume of the fluid molecules. Moreover, the fluid-wall intermolecular
force is a combination of attraction and repulsion terms. Based on the
Lenard-Jones potential model, the molecule-wall intermolecular force is
attractive generally, except that when the intermolecular distance is less
than molecule diameter. So, the Pfw is identified as a negative force as
the same as the attraction term Pattra. As described above, the Pfw can be
expressed as following and the force analysis of a single molecule is
shown in Fig. 3.
E
Pfw = − (6)
V− b

where Pfw is the fluid-wall molecules interaction force term; and E is the
fluid-wall molecules interaction potential. Taking the fluid-wall mole­
cules interaction into consideration, the original SRK EoS can be
modified as,
RT a(T) E
P = Prepul + Pattra + Pfw = − −
V − b V(V + b) V − b
(7)
RT − E a(T)
= −
V− b V(V + b)
It could be assumed as c = 1 − E/RT, and Eq (7) can be rewritten as,
Fig. 3. The sketch map of the physical model. The blue, yellow, and red cycles
represent the adsorbed gas, free gas, and pore wall molecules, respectively. (For cRT a(T)
interpretation of the references to colour in this figure legend, the reader is P= − (8)
V − b V(V + b)
referred to the Web version of this article.)

4
Y. Chen et al. Journal of Natural Gas Science and Engineering 101 (2022) 104519

Eq. (8) is the modified SRK EoS for the confined fluids in the pore,
Eext = ϕfw (r, σ, ε, rc ) × R (15)
which is called Pore-SRK EoS, or abbreviated as P-SRK EoS in this work.
For the P-SRK EoS, there is an inflection point at the critical point.
Where F[α, β, γ] is a hyper-geometric function, ρw is the number density
Mathematically, the first and second derivatives of pressure with respect
of the wall of a cylindrical pore; the subscript w represents the pore wall;
to volume at a constant temperature are equal to zero at the critical
and εfw and σfw are the cross-interaction energy and molecular param­
point,
eters, which are also obtained from the Lorentz–Berthelot combining
( ) ( 2 )
∂P ∂P rules.
= =0 (9)
∂V T=Tcc , ​ P=Pcc , ​ V=Vcc ∂V 2 T=Tcc , ​ P=Pcc , ​ V=Vcc ( )0.5
εfw = εff × εww (16)
Based on Eqs. (8) and (9), the parameters a(T) and b can be derived ( )/
as: σ fw = σff + σ ww 2 (17)
2 2
cR Tcc2 where the subscript ff and ww represent the parameters of fluid molec­
a(T) = 0.42748 α (10)
Pcc ular and wall molecular.
cRTcc
b = 0.08664 (11)
Pcc 2.4. Modified SRK EoS
where Tcc and Pcc represent the critical temperature (K) and pressure
According to the previous section, the P-SRK EoS is expressed as
(MPa) for the confined fluid, respectively. The detailed derivations of a
follows,
(T) and b are listed in Appendix A.
RT a(T) Eext
P = Prepul + Pattra + Pfw = − −
V − b V(V + b) V − b
2.3. Theorization of fluid-wall molecules interaction (18)
cext RT a(T)
= −
V − b V(V + b)
In this part, the energy, E in Eq (6), is theorized in detail. It is re­
ported that the pore radius (r), molecular diameter (σ), and energy As the pore size approaches infinity in the bulk, Eext infinitely ap­
parameter (ε) are the key factors on the fluid-wall molecules interaction proaches zero, which further causes Pfw to be ignored. In this case, the P-
potential (E) (Singh et al., 2009; Xiong et al., 2021; Yang et al., 2019), SRK EoS is mathematically simplified to the original SRK EoS. As a key
thermodynamic property, the compressibility factor (Z) should be
E = F(r, σ, ε) (12)
figured out. Referring to the compressibility factor expression for the
In the literature (Yang et al., 2019) and (Xiong et al., 2021), E is fitted original SRK EoS, the expression for the P-SRK EoS (Zext) is demonstrated
by the r/σ directly, and it is acceptable for phase behavior prediction. as follows. The detailed derivations are listed in Appendix B.
However, this method lacks the support of a theoretical basis, and the ( )
Zext 3 − cext Z 2 − B2 + cext B − A Zext − AB = 0 (19)
theoretical mechanism is not clear enough. As described above, the fluid
molecules are attracted by the pore wall, and the interaction between The EoS constants are calculated by
molecules or atoms is generally expressed by Lennard-Jones potential
model for the nonpolar particles. Then, it is well known that the inter­ A=
a(T)P
(20)
molecular action is determined by the intermolecular distance and en­ (RT)2
ergy parameter in the Lennard-Jones potential model. For the fluid
molecules in the pore, the detailed position of a specific molecule should B=
bP
(21)
be determined by the r and rc. For the specified molecules of fluid and RT
wall, the energy parameter and molecular diameter are certain. Thus,
cext = 1 − Eext /RT (22)
the potential energy is the function of r and rc, which is expressed below:
The density is calculated as follows,
E = F(r, σ, ε, rc ) (13)
P
Due to the three-dimensional structure of the cylinder pore, the ρ= (23)
RTZext
systematic action on a specified molecule is complex to calculate. In this
study, an external potential model (Eext), derived from the LJ potential Thus, the local molecule density in the nanopore can be calculated by
model of the cylindrical pore, is adopted for describing the fluid-wall equation (23). The solving process of calculating the pure component
molecules interaction. It is obtained by the integration of Lennard- density distribution by P-SRK EoS is shown in Fig. 4.
Jones potential over the whole surface of the cylindrical pore. Because
of the different mathematical techniques, there are various potential 3. Results and discussion
models of the cylindrical pore. This work selects TFM potential (Tjat­
jopoulos et al., 1988) for reliability and applicability, and its detailed In this section, the application scope of the proposed P-SRK EoS was
information can be found in the literature (Tjatjopoulos et al., 1988) and firstly discussed, and then the accuracy of P-SRK EoS for predicting the
(Zhang et al., 2010). confined fluid density was validated within this application scope.

⎧ [ ]− 10 [ / / ] ⎫
⎪ 63 rc ⎪

⎨ (2 − rc /r) F − 9 2, − 9 2; 1; (1 − rc /r)2 ⎪

32 σ
ϕfw (r, σ, ε, rc ) = π2 ρw εfw σ2fw (14)
⎪ [x

]− 4 [ / / ( )] ⎪

⎩ − 3 (2 − rc /r) F − 3 2, − 3 2; 1; (1 − rc /r)2 ⎭
σ

5
Y. Chen et al. Journal of Natural Gas Science and Engineering 101 (2022) 104519

confinement effect on fluid thermodynamic properties is studied


through investigating the adsorption and capillary condensation. For the
adsorber layer molecules, owing to the pressure difference between the
adsorption phase and the free gas phase, the capillary pressure needs to
be considered by the force system with the fluid-fluid molecules inter­
action force and fluid-wall molecules interaction force. However, the
molecules in the free gas zone are subjected to a combination of fluid-
fluid molecules interaction force and fluid-wall molecules interaction
force. Previous studies suggest that there are two obvious adsorbed
layers in nanopores (Ambrose et al., 2012; Ma and Jamili, 2016; Pang
and Jin, 2019), and the thickness of each adsorbed layer is equal to the
molecule diameter (A-B and B-C in Fig. 5) (Mosher et al., 2013; Tian
et al., 2017). Moreover, Pang et al. (Pang and Jin, 2019) proposed that
there is an extra weak adsorbed layer beyond the second adsorbed layer
under high pressure (It is not specifically labeled in Fig. 5). The density
of the extra adsorbed layer is much higher than the bulk density. This
proves that the fluid in the free gas zone is affected by the fluid-fluid
molecules interaction and fluid-wall molecules interaction. P-SRK EoS
reflects the fluid-wall molecules interaction based on the original SRK
EoS. Thus, it is applicable to the prediction of thermodynamic properties
of free zone fluid.

3.2. Model validation

In this section, first of all, we compared the bulk phase densities


predicted by the P-SRK EoS, molecular dynamic (MD), and SUPER­
TRAPP (i.e., National Institute of Standards and Technology’s (NIST)
thermophysical properties of hydrocarbon mixtures database), and the
results are listed in Table 1. The small deviations between the three
prediction results confirm the accuracy of the P-SRK EoS for free fluid
thermodynamic property prediction from the side. Besides, the local
Fig. 4. Schematic representation of solution process of the P-SRK EoS.
fluid densities are further calculated by the P-SRK EoS and compared
with the reported data. Pang et al. (Pang and Jin, 2019) characterized
the continuous density profiles of methane by Grand canonical Monte
Carlo (GCMC) simulations. Ambrose et al. (2012) used the MD simula­
tion to calculate methane discrete density profiles (statistical interval:
every 0.38 nm wide, i.e., one methane molecular diameter). For com­
parison with discrete density, Local Density Approximation theory
(LCA) is adopted in this work, and the discrete density is calculated by
averaging the density within each 0.38 nm wide integral across the pore.
Fig. 6 presents the comparison between the continuous fluid density
profiles calculated from the P-SRK EoS and collected from Pang et al.
(Pang and Jin, 2019), and Fig. 7 compares the discrete density from the
P-SRK EoS and Ambrose et al. (2012). Considering that the density
distributions across the pores are axially symmetric, only the density
distributions of radius width are shown in Figs. 6 and 7. In Fig. 6, it can
be clearly found that the fluid density distribution calculated from the
P-SRK EoS is comparable to that obtained from GCMC simulations in
both 4-nm and 8-nm pore, and the difference between the above two
densities is not obvious. Moreover, from the discrete fluid density, the
results calculated from P-SRK EoS could accord well with that obtained
from MD simulations in Fig. 7. Three discrete densities are presented in
the pore with a radius of 1.965 nm. As the rc increases, the differences
between the density from P-SRK model and density from MD simulation
are 2.3, 0.3, and 0.3 kg/m3, respectively, and the differences account for
Fig. 5. Schematic representation of molecular occurrence state in nanopores.
1.729%, 0.239%, and 0.243%. An area with a relatively high density
appears near the wall, which is owing to the stronger fluid-wall mole­
Subsequently, the characteristics of confined fluid density distribution
cules interaction. The above phenomenon verifies the point of the third
were studied in depth. Finally, the impacts of various factors, including
adsorbed layer from Ambrose et al. (2012) and Pang et al. (Pang and Jin,
temperature, pressure, and pore radius, on the density distribution were
2019).
investigated.
3.3. Characterization of fluid density distribution in free gas zone
3.1. Model application scope
3.3.1. Methane density distribution in single pore system
Considering multiple factors, the nanopores zone is divided into the In order to better understand the occurrence state of molecules in the
adsorbed layer and free gas zone, as shown in Fig. 5. Generally, the free gas zone, the methane local density distribution is predicted by the

6
Y. Chen et al. Journal of Natural Gas Science and Engineering 101 (2022) 104519

Table 1
Prediction of bulk density from P-SRK EoS, MD simulation, and NIST-SUPERTRAPP.
Fluid r[nm] T [K] P[bar] ρbulk(P-SRK) [kg/m3] ρbulk(MD) [kg/m3] ρbulk(NIST) [kg/m3] Relative deviationa [%] Relative deviationb [%]

CH4 1.95 353.15 152.10 91.32 90.00 89.00 1.47 2.61


1.95 353.15 209.81 123.84 124.00 122.00 0.13 1.51
1.95 353.15 253.45 146.07 147.00 144.00 0.63 1.44
1.95 353.15 285.51 161.02 160.00 159.00 0.63 1.27
1.95 353.15 303.64 168.96 168.00 167.00 0.57 1.17
1.95 353.15 316.19 174.25 176.00 172.00 0.99 1.31
1.95 353.15 330.94 180.27 178.00 175.00 1.27 3.01
1.95 353.15 432.43 216.24 214.00 211.00 1.05 2.48
1.95 353.15 503.31 236.79 231.00 229.00 2.51 3.40
1.95 353.15 520.55 241.33 235.00 233.00 2.69 3.58
1.95 353.15 600.32 260.40 253.00 250.00 2.93 4.16

NOTES: (1) The bulk density of MD, i.e., ρ bulk (MD), is collected from Ambrose et al. (2012); (2) The bulk density of NIST, i.e., ρ bulk (NIST), is calculated by
NIST-SUPERTRAPP®; (3) “Relative deviationa” indicate the deviation between the bulk density calculated by P-SRK and MD, and “Relative deviationb” indicate the
deviation between the bulk density calculated by P-SRK and NIST.

Fig. 6. Comparisons of the continuous fluid density distributions of methane obtained from P-SRK EoS and GCMC simulations (Pang and Jin, 2019) in (a) 4-nm and
(b) 8-nm radius cylindrical pores. Operating conditions: temperature of 333.15 K; pressure of 10–50 MPa.

P-SRK EoS under the condition of 350 K, 3 MPa, and 5 nm (radius) in prediction of the above four types of hydrocarbons in the 5-nm radius
Fig. 8. Moreover, Fig. 8 also presents the bulk density calculated by the pore is investigated in this section (Fig. 9). Compared to the lighter
original SRK EoS. In the pore center, the gap between the local density components, the heavier components present higher densities both at
(16.90 kg/m3) and the bulk density (16.88 kg/m3) is very small, which the pore center and the area near the wall, suggesting that the heavier
means that the fluid-wall molecules interaction is weak here. However, components are more affinity to shale. This is because the carbon chains
as rc decreases, the fluid-wall molecules interaction suddenly becomes of the heavier components are longer and the structures are more
stronger, leading to a dramatic increase of local fluid density. Specif­ complex. Moreover, it can be found that the densities of heavier com­
ically, the fluid density in the region where rc = 0.76 nm (26.67 kg/m3) ponents increase sharply as rc decreases. Thus, at the location near the
is 1.58 times that at the pore center. It is worth noting that the methane wall, the density differences between heavier components and lighter
local density is always greater than the bulk density across the 5-nm components are much larger than those at the pore center. For example,
radius nanopore (The illustration in Fig. 8 shows that the local density the difference between the n-butane density (121.02 kg/m3) and
in the pore center is still greater than the bulk density). Thus, the mol­ methane density (16.90 kg/m3) is 104.12 kg/m3 in the pore center. As
ecules are affected by the fluid-wall molecules interaction in the entire the rc decreases to 1.5 nm and 1.0 nm, such density differences increase
5-nm radius nanopore under the condition of 350 K and 3 MPa. to 125.49 kg/m3 and 440.27 kg/m3. This suggests that the heavier
component molecules form tighter confined layers on the pore wall as a
3.3.2. Comparison of density distribution of common alkanes result of stronger fluid-fluid molecules interaction.
Shale gas is heterogeneous with regard to components, containing Fig. 10 shows the variation of the CH4, C3H8, and CO2 densities with
methane (dominating components), ethane, propane, n-butane, and the distance to the pore center under the same working condition. It can
other hydrocarbons. To fully understand the distribution of different be seen that the density of CO2 in the pores is between that of CH4 and
components in the free gas zone, we extended density distribution pre­ C3H8 under the same working conditions, indicating that the affinity of
diction from CH4 to C2H6, C3H8, and n-C4H10. Table 2 lists the potential CO2 to shale is higher than that of CH4, but less than that of C3H8 to
parameters of four kinds of hydrocarbons and SWCNT, including the shale. In the process of shale gas exploitation, the injection of CO2 breaks
energy parameter (ε), molecule diameter (σ ), and number density of the the adsorption-desorption dynamic equilibrium of methane molecules,
pore wall molecules (ρsurf ). The calculation conditions are set as the and competitive adsorption between CH4 and CO2 occurs. CO2 mole­
temperature of 350 K and the pressure of 3 MPa, and the density profile cules with stronger affinity have a clear advantage in competing for

7
Y. Chen et al. Journal of Natural Gas Science and Engineering 101 (2022) 104519

Table 2
Parameters of potential EoS for CH4, C2H6, C3H8, n-C4H10, CO2 (Poling et al.,
2001), and SWCNT (Zhang et al., 2003a).
CH4 C2H6 C3H8

σCH4 (nm) εCH4 /k (K) σC2 H6 (nm) εC2 H6 /k σ C3 H8 εC3 H8 /k


(K) (nm) (K)
0.3758 148.6 0.4443 215.7 0.5118 237.1

n-C4H10 CO2

σC4 H10 εC4 H10 /k σCO2 (nm) εCO2 /k


(nm) (K) (K)
0.4687 531.4 0.3750 236.10

SWCNT

σSWCNT εSWCNT /k ρsurf (atoms/


(nm) (K) nm2)
0.3400 28.0 38.2

Note: k represents Boltzmann constant.

Fig. 7. Comparisons of the discrete fluid density distributions of methane ob­


tained from P-SRK EoS and MD simulations (Ambrose et al., 2012). Operating
conditions: temperature of 353.15 K; pressure of 20.98 MPa; pore radius of
1.965 nm.

Fig. 9. Extension of the density distribution from CH4 to C2H6, C3H8, and n-
C4H10 of radius = 5 nm at 350 K and 3 MPa.

Fig. 8. Methane density distribution of radius = 5 nm at 350 K and 3 MPa.

adsorption sites, while methane with weaker affinity deviated from the
sites. Therefore, the injection of CO2 is beneficial to improve the re­
covery rate of CH4, and it is theoretically feasible to replace CH4 with
CO2. At the same time, CO2 is effectively sequestered in the process. But
for heavy components, such as propane, the effect of injecting carbon
dioxide is questionable.

3.4. Parameter analysis

In sections 3.3, only the effect of the component on fluid density


distribution is analyzed. For a sufficient understanding of the contri­ Fig. 10. Comparison of the density distribution of CH4, C3H8, and CO2 of
butions of other factors, a parameter analysis (contributing factors: radius = 5 nm at 350 K and 3 MPa.
temperature, pressure, and pore size) is performed in this section.

8
Y. Chen et al. Journal of Natural Gas Science and Engineering 101 (2022) 104519

temperature effect is presented in Fig. 12, and the pressure and pore
radius are fixed at 3 MPa and 5 nm, while the temperature increases
from 293.15 to 373.15 K. Simply, the temperature shows an inverse
correlation with the density, Moreover, a continuous decrease in the
temperature is accompanied by a gradual increase in the density
increment (Fig. 13(b)). Beyond that, the density increments near the
wall are much larger than that in the pore center in Fig. 13(b). Under the
high-temperature condition, the density near the wall drops sharply,
suggesting that the confinement effect is strongly suppressed owing to
the weakening of fluid-wall molecules interaction in terms of high
temperature. To take one example, when the temperature decreases
from 373.15 to 353.15 K, the density increment of the pore center is
1.59 kg/m3, while such value is 3.51 kg/m3 for the region where rc =
0.80 nm. Remarkably, the above results are consistent with the finding
of the previous study (Heller and Zoback, 2014).

3.4.2. Effect of pore size


Considering the anisotropy of shale gas reservoirs, the pore size
distribution is very complex, ranging from less than 2 nm–100 nm in
radius (Chen et al., 2019). As one of the most critical factors affecting the
density distribution, the pore size is studied in this section at 350 K and
Fig. 11. The variations of methane density distribution versus pressure in a 5-
3 MPa. Five different pore radii are selected: 2 nm, 4 nm, 6 nm, 8 nm,
nm radius pore at 350 K.
and 10 nm. As revealed by Fig. 14, the density distribution varies greatly
with the pore radii. At the pore center, the density difference of each
3.4.1. Effect of pressure and temperature pore with different radii is quite small, the density values are close to the
It is well known that shale gas is stored in reservoirs vary in tem­ value of bulk density. This phenomenon is especially obvious when the
peratures and pressures, which could result in very different densities. pore radii is larger than 2 nm. Nonetheless, it can still be observed that
Therefore, it is necessary to investigate the effects of temperature and there is a gap between the density of small pore and the density of large
pressure on density contribution. Fig. 11 presents the variation of pore (see Fig. 14(b) and orange box in Fig. 14(a)). Owing to the com­
methane density distribution with pressure, ranging from 5 to 45 MPa, pacting effect of gas molecules at the pore center, density at the pore
in terms of temperature at 350 K and pore radius at 5 nm. As revealed by center is higher in the small pore compared to that at the large pore. This
Fig. 11, a pressure increase is accompanied by a density increase. The gap is more obvious in the region close to the wall. As proof, in the pore
greater pressure causes the molecules to be more severely compressed, of 2-nm radius, the density of the region where rc = 0.76 nm is 29.90
resulting in the molecular getting closer and the density increasing kg/m3, while this value is only 26.02 kg/m3 in the 10-nm radius pore,
accordingly. In addition, there is a gradual dip in the density growth rate which is 12.97% lower than the former. It is found that the density of the
with the continuous rise in the pressure (see Fig. 13(a)). For instance, in region close to the wall in the 2-nm pore is significantly higher
the pore center, the density difference between 15 and 5 MPa is 59.30 compared to that of the 10-nm pore owing to much stronger attraction
kg/m3, while that between 45 and 35 MPa is 32.70 kg/m3. In the region from the pore surface.
where rc = 0.80 nm, the above two values are 84.72 kg/m3 and 56.72
kg/m3, respectively. It implies that this state is gradually approaching 4. Conclusion
the compressible limit state.
A visual depiction of the methane density distribution and In this study, a thermodynamic model of confined fluids in nano­
pores has been established under a comprehensive consideration of the
dual interactions of fluid-fluid and fluid-wall molecules. In the ther­
modynamic model, a novel EoS (P-SRK) was proposed by introducing an
external potential term in the classical SRK EoS. The external potential
term is expressed as E/(V-b), and E is calculated by the TFM model.
Some reported data were collected and compared to validate the reli­
ability of this model. From our point of view, further studies could
concentrate on the considerations of pore geometry, surface roughness,
and components of the shale wall.
From this paper, the main conclusions are summarized as follows.

(1) The P-SRK EoS was developed for pure hydrocarbons (including
methane, ethane, propane, and n-butane), and this model did a
good job for predicting the fluid density distributions of shale gas.
The results from P-SRK EoS have been validated with reported
data in the free gas zone. As the rc increases, the differences be­
tween the density from P-SRK model and density from MD
simulation are 2.3, 0.3, and 0.3 kg/m3, respectively, and the
differences account for 1.729%, 0.239%, and 0.243% at 353.15
K, 20.98 MPa, and 1.965 nm (radius).
(2) According to the P-SRK EoS, the fluid density distribution was
found to be inhomogeneous, exhibiting a significantly higher
Fig. 12. The variations of methane density distribution versus temperature in a value near the wall than that in the pore center. For example, the
5-nm radius pore at 3 MPa. fluid density in the region where rc = 0.76 nm (26.67 kg/m3) is

9
Y. Chen et al. Journal of Natural Gas Science and Engineering 101 (2022) 104519

Fig. 13. The effect of the (a) pressure and (b) temperature on the fluid density difference. Note: in the legend, “3 nm”, “2 nm”, and “0.80 nm” represent the region
where rc =3 nm, 2 nm, and 0.80 nm, respectively.

Fig. 14. (a) The effect of the pore size on the fluid density distribution; (b) partical enlarged view.

1.58 times that (16.90 kg/m3) at the pore center at 350 K, 3 MPa, Credit author statement
and 5 nm (radius). From the comparison of different components,
it could be intuitively found that the density of heavier compo­ Yufeng Chen: Conceptualization, Visualization, Methodology, Soft­
nents is significantly higher than that of lighter components, ware, Formal analysis, Data curation, Writing – original draft. Hao Zhan:
because of the stronger fluid-wall molecules interaction on the Software, Writing – review & editing, Funding acquisition. Zhiyong
former. Zeng: Supervision, Project administration, Writing – review & editing,
(3) The fluid density of shale gas is dependent on pressure, temper­ Funding acquisition.
ature, and pore size. In other words, higher pressure, lower
temperature, and smaller pore diameter all contribute positively
to the occurrence of confined fluids. As the pressure continues to Declaration of competing interest
increase and the temperature continues to decrease, the change of
fluid density will gradually slow down, and the density will The authors declare that they have no known competing financial
gradually approach the limit. For instance, in the pore center, the interests or personal relationships that could have appeared to influence
density deviation between 15 and 5 MPa is 59.30 kg/m3 at 350 K the work reported in this paper.
in a 5 nm pore, while that between 45 and 35 MPa is 32.70 kg/
m3. In the region where rc = 0.80 nm, the above two values are Acknowledgments
84.72 kg/m3 and 56.72 kg/m3, respectively.
This work was supported by the National Natural Science Foundation
In the near future, it is foreseeable that the thermodynamic proper­ of China (Grant No. 52074348 and No. 51906247) and the Fundamental
ties of confined fluid will become a continuing hot topic, and the work Research Funds for the Central Universities of Central South University
provides an insight for the quantitative evaluation of that, accordingly. (No. 1053320192040).

10
Y. Chen et al. Journal of Natural Gas Science and Engineering 101 (2022) 104519

Appendix A

The calculation processes of a and b in the P-SRK EoS are shown as flows:
The P-SRK EoS is given,
RT a E RT − E a
P= − − = − (A.1)
V − b V(V + b) V − b V − b V(V + b)
We assume c = 1 − E/RT
It can be rewritten as,
cRT a
P= − (A.2)
V − b V(V + b)
The first and second derivatives of pressure with respect to volume at the critical point are yielded,
( )
∂P − cRTec a(2Vec + b)
= 2
+ 2
=0 (A.3)
∂V T=Tec (Vec − b) V 2
ec (Vec + b)

( ) [ ]
∂2 P 2cRTec 2a Vec (Vec + b) − (2Vec + b)2
= + =0 (A.4)
∂V 2 T=Tec (Vec − b)3 Vec3 (Vec + b)3

From Eq. (A.3), it is given as,


cRTec a(2Vec + b)
= (A.5)
(Vec − b)3 Vec2 (Vcc + b)2 (Vec − b)

Imposing Eq. (A.5) on Eq. (A.4), it is yielded as,


[ ]
2a(2Vec + b) − 2a Vec (Vec + b) − (2Vec + b)2
2
= 3
(A.6)
Vec2 (Vec + b) (Vec − b) 3 (V + b)
Vec ec

2
2Vec + b 3Vec + 3Vec2 b + b2
= (A.7)
Vec − b Vec (Vec + b)
It can be rewritten as,
Vec3 − 3Vcc2 b − 3Vcc b2 − b3 = 0 (A.8)

We assume b = k⋅Vec, and Eq. (A.8) can be rewritten as,


k3 + 3k2 + 3k − 1 = 0 (A.9)
Solving for k,
k = 0.2599 (A.10)
Thus,
b = 0.2599Vec (A.11)
Imposing Eq. (A.11) on Eq. (A.3), it is written as,
− cRTec a(2 + k)
+ =0 (A.12)
(1 − k)2 Vec (1 + k)2
Thus,
[ ]
cRTec Vec (1 + k)2
a= = 1.282337cRTec Vec (A.13)
(2 + k)(1 − k)2
Applying Eq. (A.2) to the critical point and substituting Eqs. (A.11) and (A.13) into Eq. (A.2), it is obtained as,
Pec Vec
= 0.333 (A.14)
cRTec
From Eqs. (A.2), (A.11), and (A.14), the parameters a and b are given (the effect of temperature on attraction coefficient has been considered),

c2 R2 Tec2 (1 − E/RTec )2 R2 Tec2


a = 0.42748 = 0.42748 α (A.15)
Pec Pec

cRTec (1 − E/RTec )RTec


b = 0.08664 = 0.08664 (A.16)
Pec Pec

11
Y. Chen et al. Journal of Natural Gas Science and Engineering 101 (2022) 104519

Appendix B

The derivation processes of compressibility factor for the P-SRK EoS are shown as flows:
The P-SRK EoS is given,
cvext RT a(T)
P= − (B.1)
V − b V(V + b)
It can be rewritten as,
PV(V − b)(V + b) = cvext RTV(V + b) − a(T)(V − b) (B.2)

[ 3 ]
PV 3 − cvext RTV 2 − Pb + cvext RTb − a(T) V − a(T)b = 0 (B.3)

Imposing V = ZRT
P
on Eq. (B.3), it is yielded as,
[ ]
3 3
(RT) 3 cvext (RT) 2 cvext (RT)2 b RTa(T)
Z − Z − RTb 2
+ − Z − a(T)b = 0 (B.4)
P2 P2 P P

[ ]
(Pb)2 cvext Pb Pa(T) a(T)bP2
3
Z − cvext Z − 2
2
+ − 2
Z− =0 (B.5)
(RT) RT (RT) (RT)2
a(T)P
We assume A = (RT)2
and B = RT
bP
,
It can be rewritten as,
( 2 )
Z 3 − cvext Z 2 −B + cvext B − A Z − AB = 0 (B.6)

References Liu, J., Xie, H., Wang, Q., Chen, S., Hu, Z., 2020. Influence of pore structure on shale gas
recovery with CO2 sequestration: insight into molecular mechanisms. Energy Fuel.
https://doi.org/10.1021/acs.energyfuels.9b02643.
Ambrose, R.J., Hartman, R.C., Diaz-Campos, M., Akkutlu, I.Y., Sondergeld, C.H., 2012.
Ma, Y., Jamili, A., 2016. Modeling the density profiles and adsorption of pure and
Shale gas-in-place calculations Part I: new pore-scale considerations. SPE J. 17,
mixture hydrocarbons in shales. J. Unconv. Oil Gas Resour. 14, 128–138. https://
219–229. https://doi.org/10.2118/131772-PA.
doi.org/10.1016/j.juogr.2016.03.003.
Barsotti, E., Tan, S.P., Saraji, S., Piri, M., Chen, J.H., 2016. A review on capillary
Mosher, K., He, J., Liu, Y., Rupp, E., Wilcox, J., 2013. Molecular simulation of methane
condensation in nanoporous media: implications for hydrocarbon recovery from
adsorption in micro- and mesoporous carbons with applications to coal and gas shale
tight reservoirs. Fuel 184, 344–361. https://doi.org/10.1016/j.fuel.2016.06.123.
systems. Int. J. Coal Geol. 109–110, 36–44. https://doi.org/10.1016/j.
BP. Statistical, 2020. Review of World, Energy|69th Edition.
coal.2013.01.001.
Chen, F., Lu, S., Ding, X., Ju, Y., 2019. Evaluation of the density and thickness of
Palmer, J.C., Moore, J.D., Brennan, J.K., Gubbins, K.E., 2011. Simulating local
adsorbed methane in differently sized pores contributed by various components in a
adsorption isotherms in structurally complex porous materials: a direct assessment of
shale gas reservoir: a case study of the Longmaxi Shale in Southeast Chongqing,
the slit pore model. J. Phys. Chem. Lett. 2, 165–169. https://doi.org/10.1021/
China. Chem. Eng. J. 367, 123–138. https://doi.org/10.1016/j.cej.2019.02.105.
jz1015668.
Cracknell, R.F., 2002. Simulation of hydrogen adsorption in carbon nanotubes. Mol.
Pang, W., Jin, Z., 2019. Revisiting methane absolute adsorption in organic nanopores
Phys. 100, 2079–2086. https://doi.org/10.1080/00268970210130236.
from molecular simulation and Ono-Kondo lattice model. Fuel 235, 339–349.
Energy Information Administration (EIA), 2018. Annual Energy Outlook 2018 with
https://doi.org/10.1016/j.fuel.2018.07.098.
Projections to 2050. https://www.eia.gov/analysis/studies/worldshalegas/.
Poling, B.E., Prausnitz, J.M., O’Conneel, J.P., 2001. The properties of gases and liquids.
(Accessed 17 October 2021). accessed.
Phys. Today. https://doi.org/10.1063/1.3060771.
Gu, C., 2001. Simulation study of hydrogen storage in single walled carbon nanotubes.
Ross, D.J.K., Marc Bustin, R., 2009. The importance of shale composition and pore
Int. J. Hydrogen Energy 26, 691–696. https://doi.org/10.1016/S0360-3199(01)
structure upon gas storage potential of shale gas reservoirs. Mar. Petrol. Geol. 26,
00005-2.
916–927. https://doi.org/10.1016/j.marpetgeo.2008.06.004.
Heller, R., Zoback, M., 2014. Journal of Unconventional Oil and Gas Resources
Sandoval, D.R., Yan, W., Michelsen, M.L., Stenby, E.H., 2018. Modeling of shale gas
Adsorption of methane and carbon dioxide on gas shale and pure mineral samples.
adsorption and its influence on phase equilibrium. Ind. Eng. Chem. Res. https://doi.
J. Unconv. Oil Gas Resour. 8, 14–24. https://doi.org/10.1016/j.juogr.2014.06.001.
org/10.1021/acs.iecr.7b04144.
Ji, L., Zhang, T., Milliken, K.L., Qu, J., Zhang, X., 2012. Experimental investigation of
Sermoud, V.M., Barbosa, G.D., Barreto, A.G., Tavares, F.W., 2020. Quenched solid
main controls to methane adsorption in clay-rich rocks. Appl. Geochem. 27,
density functional theory coupled with PC-SAFT for the adsorption modeling on
2533–2545. https://doi.org/10.1016/j.apgeochem.2012.08.027.
nanopores. Fluid Phase Equil. 521, 112700. https://doi.org/10.1016/j.
Jin, Z., Firoozabadi, A., 2016. Phase behavior and flow in shale nanopores from
fluid.2020.112700.
molecular simulations. Fluid Phase Equil. 430, 156–168. https://doi.org/10.1016/j.
Singh, S.K., Sinha, A., Deo, G., Singh, J.K., 2009. Vapor-liquid phase coexistence, critical
fluid.2016.09.011.
properties, and surface tension of confined alkanes. J. Phys. Chem. C 113,
Kim, H.Y., Cole, M.W., Mbaye, M., Gatica, S.M., 2011. Phase behavior of Ar and Kr films
7170–7180. https://doi.org/10.1021/jp8073915.
on carbon nanotubes. J. Phys. Chem. A 115, 7249–7257. https://doi.org/10.1021/
Soave, G., 1972. Equilibrium constants from a modified Redlich-Kwong equation of state.
jp200410y.
Chem. Eng. Sci. 27, 1197–1203. https://doi.org/10.1016/0009-2509(72)80096-4.
Li, Z., Jin, Z., Firoozabadi, A., 2014. Phase behavior and adsorption of pure substances
Tang, X., Ripepi, N., Luxbacher, K., Pitcher, E., 2017. Adsorption models for methane in
and mixtures and characterization in nanopore structures by density functional
shales: review, comparison, and application. Energy Fuel. 31, 10787–10801. https://
theory. SPE J. 19, 1096–1109. https://doi.org/10.2118/169819-PA.
doi.org/10.1021/acs.energyfuels.7b01948.
Li, J., Rao, Q., Xia, Y., Hoepfner, M., Deo, M.D., 2020. Confinement-mediated phase
Tao, T., Wang, S., Qu, Y., Cao, D., 2021. Displacement of shale gas confined in illite shale
behavior of hydrocarbon fluids: insights from Monte Carlo simulations. Langmuir 36,
by flue gas: a molecular simulation study. Chin. J. Chem. Eng. 29, 295–303. https://
7277–7288. https://doi.org/10.1021/acs.langmuir.0c00652.
doi.org/10.1016/j.cjche.2020.09.015.
Lin, K., Huang, X., Zhao, Y.P., 2020. Combining image recognition and simulation to
Tian, Y., Yan, C., Jin, Z., 2017. Characterization of methane excess and absolute
reproduce the adsorption/desorption behaviors of shale gas. Energy Fuel. https://
adsorption in various clay nanopores from molecular simulation. Sci. Rep. 7, 1–21.
doi.org/10.1021/acs.energyfuels.9b03669.
https://doi.org/10.1038/s41598-017-12123-x.
Liu, X., Zhang, D., 2019. A review of phase behavior simulation of hydrocarbons in
Tjatjopoulos, G.J., Feke, D.L., Mann, J.A., 1988. Molecule-micropore interaction
confined space: implications for shale oil and shale gas. J. Nat. Gas Sci. Eng. 68,
potentials. accessed 17 October 2021 J. Phys. Chem. 92, 4006–4007. https://doi.
102901. https://doi.org/10.1016/j.jngse.2019.102901.
org/10.1021/j100324a063, 2018, URL. https://www.eia.gov/analysis/stud
Liu, B., Qi, C., Mai, T., Zhang, J., Zhan, K., Zhang, Z., He, J., 2018. Competitive
ies/worldshalegas/ (accessed 7.23.21).
adsorption and diffusion of CH4/CO2 binary mixture within shale organic
nanochannels. J. Nat. Gas Sci. Eng. 53, 329–336. https://doi.org/10.1016/j.
jngse.2018.02.034.

12
Y. Chen et al. Journal of Natural Gas Science and Engineering 101 (2022) 104519

Wang, L.S., Gmehling, J., 1999. Improvement of the SRK equation of state for Yang, G., Fan, Z., Li, X., 2019. Determination of confined fluid phase behavior using
representing volumetric properties of petroleum fluids using Dortmund Data Bank. extended Peng-Robinson equation of state. Chem. Eng. J. 378, 122032. https://doi.
Chem. Eng. Sci. 54, 3885–3892. https://doi.org/10.1016/S0009-2509(99)00025-1. org/10.1016/j.cej.2019.122032.
Wang, S., Feng, Q., Javadpour, F., Hu, Q., Wu, K., 2019. Competitive adsorption of Zarragoicoechea, G.J., Kuz, V.A., 2002. Van der Waals equation of state for a fluid in a
methane and ethane in montmorillonite nanopores of shale at supercritical nanopore. Phys. Rev. E - Stat. Physics, Plasmas, Fluids, Relat. Interdiscip. Top. 65,
conditions: a grand canonical Monte Carlo simulation study. Chem. Eng. J. 355, 1–4. https://doi.org/10.1103/PhysRevE.65.021110.
76–90. https://doi.org/10.1016/j.cej.2018.08.067. Zarragoicoechea, G.J., Kuz, V.A., 2004. Critical shift of a confined fluid in a nanopore.
Williams, K.A., Eklund, P.C., 2000. Monte Carlo simulations of H 2 physisorption in Fluid Phase Equil. 220, 7–9. https://doi.org/10.1016/j.fluid.2004.02.014.
finite-diameter carbon nanotube ropes. Chem. Phys. Lett. 320, 352–358. https://doi. Zeng, Z.Y., Xu, Y.Y., Li, Y.W., 2008. Calculation of solubility parameter using perturbed-
org/10.1016/S0009-2614(00)00225-6. chain SAFT and cubic-plus-association equations of state. Ind. Eng. Chem. Res. 47,
WORLD RESOURCES INSTITUTE. Location of World’s Shale Plays, Volume of 9663–9669. https://doi.org/10.1021/ie800811f.
Technically Recoverable Shale Gas in the 20 Countries with the Largest Resources, Zeng, Z.Y., Xu, Y.Y., Hao, X., Li, Y.W., 2009. Application of the simplified perturbed-
and the Level of Baseline Water Stress. n.d. [WWW Document]. URL. https://www. chain saft to hydrocarbon systems with new group-contribution parameters. Ind.
wri.org/water-for-shale (accessed 7.23.21). Eng. Chem. Res. 48, 5867–5873. https://doi.org/10.1021/ie8019246.
Xiong, J., Liu, X., Liang, L., Zeng, Q., 2017a. Adsorption of methane in organic-rich shale Zhang, X., Cao, D., Chen, J., 2003a. Hydrogen adsorption storage on single-walled
nanopores: an experimental and molecular simulation study. Fuel 200, 299–315. carbon nanotube arrays by a combination of classical potential and density
https://doi.org/10.1016/j.fuel.2017.03.083. functional theory. J. Phys. Chem. B 107, 4942–4950. https://doi.org/10.1021/
Xiong, J., Liu, X., Liang, L., Zeng, Q., 2017b. Adsorption behavior of methane on jp034110e.
kaolinite. Ind. Eng. Chem. Res. 56, 6229–6238. https://doi.org/10.1021/acs. Zhang, X., Wang, W., Chen, J., Shen, Z., 2003b. Characterization of a sample of single-
iecr.7b00838. walled carbon nanotube array by nitrogen adsorption isotherm and density
Xiong, W., Zhao, Y.L., Qin, J.H., Huang, S.L., Zhang, L.H., 2021. Phase equilibrium functional theory. Langmuir 19, 6088–6096. https://doi.org/10.1021/la026924c.
modeling for confined fluids in nanopores using an association equation of state. Zhang, X., Shao, X., Wang, W., Cao, D., 2010. Molecular modeling of selectivity of single-
J. Supercrit. Fluids 169, 105118. https://doi.org/10.1016/j.supflu.2020.105118. walled carbon nanotube and MCM-41 for separation of methane and carbon dioxide.
Yang, G., Li, X., 2020. Journal of natural gas science and engineering modified Peng- Separ. Purif. Technol. 74, 280–287. https://doi.org/10.1016/j.seppur.2010.06.016.
Robinson equation of state for CO 2/hydrocarbon systems within nanopores. J. Nat. Zuo, J.Y., Guo, X., Liu, Y., Pan, S., Canas, J., Mullins, O.C., 2018. Impact of capillary
Gas Sci. Eng. 84, 103700. https://doi.org/10.1016/j.jngse.2020.103700. pressure and nanopore confinement on phase behaviors of shale gas and oil. Energy
Fuel. 32, 4705–4714. https://doi.org/10.1021/acs.energyfuels.7b03975.

13

You might also like