Fuel 247 (2019) 77–86

Contents lists available at ScienceDirect

Fuel
journal homepage: www.elsevier.com/locate/fuel

Full Length Article

Incorporation of multi-phase solubility and molecular diffusion in a T

geochemical evaluation of the CO2 huff-n-puff process in liquid-rich shale
reservoirs
Ji Ho Lee, Moon Sik Jeong, Kun Sang Lee
⁎

Department of Earth Resources and Environmental Engineering, Hanyang University, Seoul 04763, Republic of Korea

ARTICLE INFO ABSTRACT

Keywords: Despite the development of horizontal well and hydraulic fracturing technologies, natural depletion is limited in
Geochemical reaction low productivity, and thereby considerable hydrocarbon remains in low permeable shale reservoirs. Recent
Aqueous solubility researches proved that the deployment of CO2 huff-puff process could extract the remained hydrocarbon. They
Molecular diffusion quantified various mechanisms behind the process and proposed the potential for CO2 sequestration in the shale
CO2 huff-n-puff
formation. However, thorough understanding of fluid transport in the CO2 huff-n-puff process has not been
Shale reservoir
completed due to complexity in tight shale formations. They had a lack to analyze potential factors affecting
fluid flow in the system of oil/CO2/brine/shale formation: 1) aqueous solubility; 2) molecular diffusion in
aqueous and oleic phases; and 3) geochemistry. Therefore, this study diagnoses the potential factors affecting
fluid flow during CO2 huff-n-puff process in the shale oil reservoirs. Then, their effects are quantified by ana-
lyzing shale oil recovery and CO2 storage capacity.
Since a fraction of injected CO2 dissolves in water and oil during the CO2 huff-n-puff process, transportability
is acquired by molecular diffusion in aqueous and oleic phases. In addition, dissolution of CO2 changes the pH of
in-situ brine, which is a sensitive factor in geochemistry, and thereby would change physical properties of
formation. Implementing these mechanisms, this study simulates the CO2 huff-n-puff process in shale oil re-
servoirs. It is observed that effects of aqueous solubility and geochemistry decrease oil production by 3.2% and
6%, respectively. Approximately, up to 9.5% of CO2 to be injected is sequestrated in depleted shale formation via
solubility mechanism. Molecular diffusion in oleic phase increases the oil production by 2%, but that in aqueous
phase is negligible. For the accurate estimation of oil production and CO2 storage capacity in the simulation of
CO2 huff-n-puff process for shale oil reservoir, the comprehensive model should consider potential factors such
as aqueous solubility, molecular diffusion, and geochemical reactions. Because the potential factors have been
neglected in the works of other types of shale reservoirs, the proposed approaches expect to expand our un-
derstanding in the CO2 injection to tight and shale gas/oil reservoirs.

1. Introduction the oil production in the U.S. [8].

While multi-stage hydraulic fracturing enhances oil productivity
Unconventional, liquid-rich shale reservoirs have very small pore from shale reservoirs, the rapid decline of the production rate is a
throat and ultralow permeability on the order of micro- to nano-scale challenging problem. Therefore, enhanced oil recovery (EOR) has been
[1]. Because of the extremely low permeability, natural depletion limits proposed to shale oil reservoirs. Extensive studies of laboratory ex-
to recovery of hydrocarbon from shale reservoirs [2,3]. Recent devel- periment, numerical simulation, and field-scaled investigation have
opment of technologies including horizontal well and multi-stage explored EOR scheme injecting gas in shale oil reservoirs. Laboratory
fracturing increases the productivity and economics of shale reservoirs studies tried to quantify major factors to extract oil from shale re-
[4–7]. These technologies enlarge contact area between well and for- servoirs via CO2 injection [9–11]. Gamadi et al. [9] experimentally
mation of reservoir by bridging natural fractures and hydraulic frac- conducted a sensitivity study controlling parameters of injection de-
tures. The success of these technologies has accelerated production signs of CO2 huff-n-puff process such as soaking period, soaking pres-
from low-permeability shale reservoirs, accounting for almost half of sure, and number of cycles. The study observed enhancing oil

⁎
Corresponding author.
E-mail address: [email protected] (K.S. Lee).

https://doi.org/10.1016/j.fuel.2019.03.033
Received 7 January 2019; Received in revised form 2 March 2019; Accepted 6 March 2019
Available online 11 March 2019
0016-2361/ © 2019 Elsevier Ltd. All rights reserved.
J.H. Lee, et al. Fuel 247 (2019) 77–86

Nomenclature His Henry’s law constant of species i at the saturation pressure

of H2O, temperature, and zero salinity, MPa
ai ion acitivity of componenti I ionic strength
ai ion size parameter of componenti K eq, equilibrium constant of aqueous reaction
fi, j fugacity of species i in phase j , Mpa K eq, solubility product constant of mineral reaction
k andk 0 permeabilities after and before mineral reaction, md Mik molecular weight of solventk
ksalt , i salting-out coefficient of componenti N andN 0 total moles of minerals per bulk volume after and before
k rate constant of mineral reaction, mol/(m2⋅s) mineral reaction, mol
mi molality of componenti Q ion acitivity product of aqueous reaction
na number of aqueous components that exist only in the Q ion activity product of mineral reaction
aqueous phase Raq total number of aqueous reactions
nc number of gaseous components that are soluble in the Rmn total number of mineral reactions
aqueous phase T temperature, K
naq total number of components in the aqueous phase Tr,H2O reduced temperature of H2O
n total number of mineral components T temperature, °C
pHs 2O saturation pressure of H2O, Mpa j association factor of solventk
vbi partial molar volume at the boiling point, cm3/mol i ionic activity coefficient of componenti
vi stoichiometric coefficient of aqueous reaction k consumption or production rate of species k in brine due
vk stoichiometric coefficient of mineral reaction to mineral reaction, mol/(m3⋅s)
v̄i partial molar volume of species i , cm3/mol rate of mineral reaction , mol/(m3⋅s)
zi charge of componenti µk viscosity of solvent k , cp
A reactive surface area of a mineral, m2/m3 k density of solvent k , kg/m3
Dij binary molecular diffusion coefficient between two species kr pseudo-reduced molar density
i and j , cm2/s mineral molar density, m3/mol
Dik* molecular diffusion coefficient of species i in solvent k , 0 0
k Dij low-pressure gas-density-diffusion product
cm2/s and 0 porosities after and before mineral reaction
Hi Henry’s law constant of species i at zero salinity, MPa
Hsalt , i Henry’s law constant of species i at a specific salinity, MPa

extraction from Eagle Ford and Mancos shale cores by varying the in tight shale reservoirs, they neglected potential factors affecting the
parameters. Jin et al. [10] focused on quantifying the mechanism be- fluid flow in shale formation: 1) interactions between injected CO2 and
hind CO2 EOR in Bakken shale formation. The study concluded that CO2 formation water; 2) molecular diffusion of multi-components in aqu-
greatly enhanced a diffusion process, and then improved oil trans- eous and oleic phases; and 3) chemical reactions occurred in the system
portability in tight shale formation. Another study [11] observed effects of oil/CO2/brine/shale formation. In the high-pressure conditions of
of interfacial tension reduction and oil swelling on increasing oil re- CO2 huff-n-puff process, a fraction of CO2 can dissolve into the in-situ
covery in CO2 EOR. In addition to the EOR potential, the study of Kang brine and they might have an additional transportability by diffusion in
et al. [12] experimentally clarified that shale oil reservoir could be a water. The amount of CO2 dissolved in water also affect an effective
good formation to sequestrate CO2. The numerical studies [13–17] have concentration of CO2, which would interact with oil in the shale matrix.
successfully simulated the deployment of CO2 huff-n-puff process in In addition, CO2 dissolution in water decreases pH of the in-situ brine
shale oil reservoirs considering multiple physics (molecular diffusion in due to geochemistry. It would cause geochemical reactions including
hydrocarbon phase, phase behavior of CO2-oil mixture, oil viscosity mineral dissolution in shale formation with carbonate minerals, and
reduction, and oil swelling). They considered multi-component trans- then physical properties of shale formation might change. As a result,
port in the hydrocarbon phase and observed increasing oil recovery the change in the fluid flow by the potential factors would influence the
from shale reservoirs. Alfarge et al. [13] modeled a field-scaled simu- oil production and CO2 storage during CO2 huff-n-puff process in shale
lation using laboratory measurements and compared its result to a pilot formations.
test result. The study cautioned a careful assessment in molecular dif- Therefore, this study develops an improved model of the CO2 huff-n-
fusion to be required. The study also summarized the potential me- puff process by incorporating potential factors, i.e., solubility of CO2 in
chanisms of CO2 EOR in shale reservoirs: 1) diffusion; 2) reduction in water, comprehensive molecular diffusion in oleic and aqueous phases,
capillary forces; 3) re-pressurization; 4) extraction; 5) oil swelling and and geochemical reactions (Fig. 1). The developed model is examined
pressure maintenance; and 6) oil viscosity reduction. Torres et al. [16] in the Eagle Ford shale oil reservoirs which have a high fraction of
incorporated laboratory- and field-base data to construct drill spacing carbonate minerals. Spanning a region from north of the Maverick
unit-scaled model and demonstrated a potential of CO2 storage as well Basin to the Gulf of Mexico, Eagle Ford shale is a major source rock
as EOR in shale reservoirs. There have been a number of field trials [20]. Eagle Ford formation mainly consists of organic-rich calcareous
injecting gas, particularly, in Bakken and Eagle Ford reservoirs. Thomas mudstones and chalks that were deposited during two transgressive
et al. [18] reported improvements in production up to 30%70% over sequences, the upper and lower Eagle Ford. Because formation is
primary recovery due to gas injection. Hoffman [19] summarized data characterized by a variety of thermal maturities related to structural
sets of seven pilot projects and concluded that huff-n-puff process in- and depositional configuration, various fluid types and petrophysical
jecting gas is effective to enhance oil production and the process is properties have been developed [21,22]. For the carbonate-rich Eagle
proven to be economically viable in Eagle Ford shale formations. Ford shale oil reservoirs, this study investigates an overall effect of the
However, gas to be injected in huff-n-puff process is limited in natural potential factors on the oil recovery and CO2 storage capacity of CO2
gas, not CO2. huff-n-puff process.
Previous studies have been encouraged to assess governing factors This article is organized as follows: In 2. BASIC THEORY, potential
of the CO2 EOR scheme in shale oil reservoirs. Thoroughly quantifying factors including aqueous solubility, molecular diffusion, and geo-
the effects of these factors on the transport of hydrocarbon components chemistry are explained in detail. Mathematical formulation of each

78
J.H. Lee, et al. Fuel 247 (2019) 77–86

where k indicates the solvent; i and j indicate species; Dik* is the mo-
lecular diffusion coefficient of species i in solvent k ; j is the association
factor of solvent k ; Mik is the molecular weight of solvent k ; µk is the
viscosity of solvent k ; vbi is the partial molar volume at the boiling
point; and yi, k and yj, k are the mole fractions of species i and j in solvent
k.
For the oleic phase, the Sigmund correlation, which is a function of
reduced density, successfully predicts the molecular diffusion coeffi-
cient, Dij (Eq. (5)). This correlation is applicable to oleic and gas phases,
but not the aqueous phase. For the multi-component system, the ef-
fective diffusion coefficient of each component is calculated via Eq. (6).
0 0
k Dij 2 3
Dij = (0.99589 + 0.096016 kr 0.22035 kr + 0.032874 kr )
k (5)

Fig. 1. Schematic description of the system of oil/CO2/brine/shale formation 1 yi, k

incorporating aqueous solubility, geochemistry, and molecular diffusion of CO2.
Dik* = 1
y D
j i i, k ij (6)

mechanism is explained in Appendix. Based on shale oil reservoir model where Dij is the binary molecular diffusion coefficient between com-
and design of CO2 huff-n-puff process presented in 3. NUMERICAL ponents i and j in the mixture; k0 Dij0 is the low-pressure gas-density-
SIMULATIONS, extensive simulations have been performed. Results of diffusion product; and kr is the pseudo-reduced molar density.
simulations are used to systematically examine the effects of various
factors on the oil production and CO2 sequestration in 4. RESULTS. In 5. 2.3. Geochemical reactions
CONCLUSIONS, the findings of this study are summarized.
This study considers major geochemical reactions including aqueous
2. Basic theory and mineral reactions. In the aqueous phase, ions can attach to other
ions and form complexes. The reactions of aqueous complex formation
2.1. Aqueous solubility are relatively fast to achieve equilibria. The equilibrium of an aqueous
reaction between components generally follows the law of mass action
The gaseous component of CO2 is soluble in aqueous and oleic (Eq. (7)). Details calculation of the equilibrium of aqueous reaction are
phases. The solubility in each phase is determined via the equality of described in Appendix B.
fugacity of all phases (Eq. (1)). While the fugacities of CO2 in the oleic Q K eq, = 0, with = 1, …, Raq (7)
and gas phases are calculated using the Peng-Robinson equation of state
(EoS), calculation of its fugacity in the aqueous phase follows Henry’s where indicates the aqueous reaction; K eq, is the equilibrium con-
law (Eq. (2)). Henry’s law determines how much amount of CO2 dis- stant of the aqueous reaction, Q is the ion activity product of the
solves in water at specific reservoir condition (temperature, pressure, aqueous reaction; and Raq is the number of reactions between compo-
and salinity). In detail, the constant of Henry’s law is a function of nents in aqueous phase.
pressure, temperature, and salinity [23,24]. Further calculations of The mineral reaction of dissolution and precipitation involves
Henry’s law constant considering the pressure, temperature, and sali- multiple phases, i.e., solid and aqueous phases. Because the mineral
nity are illustrated in Appendix A. reaction is a slow kinetic reaction, equilibrium of mineral reactions
requires a relatively longer time than for the aqueous reaction.
fi, o = fi, g = fi, aq (1) Therefore, the mineral reaction follows the rate law (Eq. (8)).
fi, aq = Hi x i (2) Q
=A k 1 , With = 1, …, Rmn
where i indicates species, i.e., CO2; j indicates phase, i.e., oleic, gas, and K eq, (8)
aqueous phases; fi, j is the fugacity of species i in phase j ; Hi is the
where defines the mineral reaction; r is the mineral reaction rate; k
Henry’s law constant for species i at zero salinity; and x i is the mole
is the reaction rate constant of mineral reaction; A is the reactive
fraction of species i in the aqueous phase.
surface area of a mineral; K eq, is the solubility product constant of the
mineral reaction; Q is the ion activity product of the mineral reaction;
2.2. Molecular diffusion
and Rmn is the number of reactions between minerals and aqueous
components.
Various correlations are developed to predict the molecular diffu-
This mineral reaction affects the aqueous reaction controlling con-
sion coefficients of components in oleic and aqueous phases. This study
centrations of aqueous species. Mineral dissolution generates aqueous spe-
adopts the Wilke and Chang [25] and Sigmund [26] correlations. The
cies in the water, while mineral precipitation consumes aqueous species.
correlations of Wilke and Chang and Sigmund are widely used to cal-
Based on this relationship, the rate of mineral dissolution/precipitation is
culate the molecular diffusion coefficient of CO2 in the aqueous phase
closely related to the formation/consumption rate of aqueous species (Eq.
and its diffusion coefficient in the oleic phase, respectively. The Wilke
(9)). In addition, the mineral dissolution or precipitation changes the pore
and Change correlation is an empirical modification of the Stokes-
volume of the reservoir, i.e., porosity. The change in porosity might affect
Einstein relation. In the aqueous phase, the correlation estimates the
the permeability of the system. This study uses the formulation of Kozeny-
molecular diffusion coefficient of CO2 (Eqs. (3) and (4)).
Carman describing relationship between porosity and permeability. Details
in these calculations are described in Appendix B.
0.5
7.4 × 10 8Mik T
Dik* = = vk
µk vbi0.6 (3) k (9)

y M where vk is the stoichiometric coefficient of the mineral reaction; and k is

i j j, k j j
Mik = the consumption or production rate of ionic species in brine due to the
1 yi, k (4) mineral reaction.

79
J.H. Lee, et al. Fuel 247 (2019) 77–86

3. Numerical simulations

The numerical simulations were performed with GEMTM software,

which simulates the transport of multi-component, multi-phase in
porous media, the EoS of the fluid, and the comprehensive geochemical
reactions. In this study, the two-dimensional shale oil reservoir system
is characterized with a dual permeability model describing the natural
fracture network and matrix-matrix, fracture-fracture, and matrix-
fracture communications. The reservoir system has an initial pressure of
6425 psi. This study simulates only one single hydraulic-fracture-sti-
mulated reservoir volume on the basis of symmetry to reduce the
computational work. The size of the system is 495 × 1515 × 200 ft3.
Each grid block has a size of 15 × 15 × 200 ft3. In the reservoir model,
the porosities of natural fracture and matrix are 1 × 10−5 and 0.06,
respectively, and their permeabilities are 0.08 md and 1 × 10−4 md. In
the hydraulic fracture system, fracture permeability is 83,300 md. The
fracture width and half-length are 0.001 ft and 500 ft, respectively.
Fluid modeling of shale oil follows previous work on Eagle Ford shale
[27] and is simulated with WINPROPTM software. The pressure-tem-
perature diagram of the fluid is described in Fig. 2.
Production is operated with two years. The CO2 huff-n-puff process
uses a single well for production and injection. Deployment of natural
depletion is designed for one year ahead of the initiation of the CO2 Fig. 3. Cumulative oil production of primary recovery and CO2 huff-n-puff
processes in a shale oil reservoir.
huff-n-puff process. Then, the CO2 huff-n-puff process is deployed for
the next year. During the period of production, the bottom-hole pres-
sure of the producer is set to 2000 psi to avoid in-situ gas generation.
geochemical reactions on performance of the CO2 huff-n-puff process.
Subsequently, the CO2 huff-n-puff process is designed with four cycles
The reservoir rock is assumed to be composed of 30% calcite mineral.
and each cycle consists of one month of CO2 injection and two months
of production.
This study illustrates two sets of simulations of the CO2 huff-n-puff 4. Results
process in tight shale oil reservoirs. The first set describes the results of
simulations neglecting geochemical reactions. In the first set, the phe- 4.1. CO2 huff-n-puff
nomena of CO2 solubility in water and molecular diffusions in aqueous
and oleic phases are implemented in the simulations of the CO2 huff-n- 4.1.1. Aqueous solubility
puff process. The second set of simulations considers the geochemical In the hydraulically-fractured reservoir system, primary natural
reactions. Because the Eagle Ford shale reservoir has a high fraction of depletion and secondary CO2 huff-n-puff processes are deployed. To
calcite minerals, the simulations of second set explores the role of confirm the performance of the CO2 huff-n-puff process, production by

Fig. 2. Pressure-temperature diagram of oil phase behavior.

80
J.H. Lee, et al. Fuel 247 (2019) 77–86

only natural depletion for two years is also simulated and investigated. 4.2. CO2 huff-n-puff coupling with geochemistry
First, the simulation neglects the aqueous solubility of CO2, molecular
diffusion, and geochemical reactions. Fig. 3 describes the hydrocarbon The performance of the CO2 huff-n-puff process is evaluated by in-
production from the shale oil reservoir. Over two years, the primary corporating geochemical reactions. Table 1 describes the geochemical
recovery of natural depletion process is approximately 2.10 × 104 bbl reactions, including aqueous and mineral reactions, implemented in the
of oil. To enhance shale oil production, the secondary CO2 huff-n-puff simulations of CO2 huff-n-puff process. The database of geochemical
process is deployed. It is the base case of the CO2 huff-n-puff process reactions is referred from the work of Wolery and Daveler [29]. Because
throughout this article. The four cycles of CO2 huff-n-puff process are the Eagle Ford shale reservoir has a large fraction of calcite, the si-
initiated after one year of natural depletion. The CO2 huff-n-puff pres- mulation considers the calcite mineral reaction and aqueous reactions.
surizes the reservoir system and results in some dissolution of CO2 into
oil, contributing to viscosity reduction and oil swelling. During the CO2
huff-n-puff process, a total 1.8 × 108 mol of CO2 is injected, and ap- 4.2.1. Geochemical reactions
proximately 57% of the injected CO2 is produced (Fig. 4). As a result, When the CO2 huff-n-puff process is applied, some fraction of CO2
the CO2 huff-n-puff process increases the total oil production by up to dissolves in water following Henry’s law. The dissolved CO2 reacts with
2.31 × 104 bbl, which is about 10% higher than the primary recovery water and provides H+, thereby decreasing the pH of the in-situ brine.
case (Fig. 3). The production history clearly shows that the CO2 huff-n- In a low-pH condition, calcite mineral potentially would dissolve in-
puff process has EOR potential in tight shale oil reservoirs. creasing pore volume, i.e., porosity. Fig. 8 compares the pore volumes
Taking into consideration the aqueous solubility of CO2, simulation of the base case and the CO2 huff-n-puff process with geochemical re-
of the CO2 huff-n-puff process in the shale reservoir is carried out. Some actions. During natural depletion, pore volume decreases due to the
fraction of the CO2 to be injected would be soluble into water following pressure depletion. When CO2 huff-n-puff process is initiated, CO2 in-
Henry’s law. Because the solubility of CO2 in the aqueous phase is jection increases pore volume and production decreases it. In the base
sensitive to pressure, CO2 would dissolve in water during CO2 injection. case, the pore volume returns to its initial state as pressure decreases.
Some of the dissolved CO2 degasifies out of the water during produc- However, simulation with geochemical reactions shows the tendency of
tion. Simulation of the CO2 huff-n-puff process with aqueous solubility increasing pore volume despite of pressure depletion. The difference in
is compared to the base case. Fig. 5A shows the amount of CO2 dis- pore volumes is attributed to the dissolution of calcite minerals. Fig. 9
solved in the aqueous phase during four cycles of CO2 huff-n-puff due to shows the total amounts of H+ and HCO3–, which are the results of
aqueous solubility. Up to 9.5% of the injecting CO2 dissolves in the geochemical reactions. During CO2 injection, the amount of H+ in-
aqueous phase. This aqueous solubility affects the dissolution of CO2 in creases. Increase in the amount of H+ indicates decreasing pH, thereby
oil, resulting in relatively less contact between CO2 and oil. Fig. 5B calcite mineral dissolves in the low-pH condition generating HCO3–. In
shows the amount of CO2 dissolved in oil. Dissolution of CO2 in the addition, the geochemical reactions influence the solubility of CO2 in
aqueous phase leads to 10% less dissolution of CO2 in oil and also af- water and oil. Fig. 10 describes the results of dissolved CO2 in water and
fects the contact between oil and CO2 (Fig. 6A and B). As a result, CO2 oil for the CO2 huff-n-puff process considering geochemical reactions.
huff-n-puff process with aqueous solubility produces total Results of the simulation with geochemical reactions are compared to
2.26 × 104 bbl of oil (Fig. 7). Comparing to the base case of the CO2 those of simulations considering only aqueous solubility of CO2 (Figs. 5
huff-n-puff process, the aqueous solubility of CO2 reduces total oil and 10). Simulation with geochemical reactions shows less total
production by 3.2% (Figs. 3 and 7). This result clearly shows that the amount of CO2 dissolved in water by 29%. This reduction indicates a
aqueous solubility of CO2 has a negative impact on oil extraction from decrease in CO2 sequestration via solubility mechanism. In addition, the
shale reservoirs. However, aqueous dissolution of CO2 implies the po- geochemical reactions reduce the total amount of CO2 dissolved in oil
tential of CO2 storage in the depleted reservoir. It is known that CO2 can by 24%. The reduction in solubility of CO2 influences the performance
be sequestrated in reservoirs by four mechanisms: 1) structural trap- of the CO2 huff-n-puff process. As a result, the simulation of CO2 huff-n-
ping, 2) residual trapping, 3) solubility trapping, and 4) mineral trap-
ping [28]. After the CO2 huff-n-puff process, up to 9.5% of CO2 is stored
by the solubility trapping mechanism (Fig. 5A).

4.1.2. Molecular diffusion in the aqueous and oleic phases

Once CO2 dissolves in water, transportability by aqueous molecular
diffusion would be predicted following the modified Wilke and Chang
correlation. Additional analysis is performed to evaluate the effect of
aqueous molecular diffusion on production of the CO2 huff-n-puff
process. Aqueous molecular diffusion slightly increases the amount of
CO2 dissolved in water, though it is not significant (Fig. 5A). Ad-
ditionally, it hardly changes the amount of CO2 dissolved in oil
(Fig. 5B). As a result, aqueous molecular diffusion of CO2 is a negligible
factor to affect the oil production in the CO2 huff-n-puff process (Fig. 7).
The influence of aqueous solubility of CO2 is highly significant whether
aqueous molecular diffusion exists or not (Fig. 7). Further investigation
considering only molecular diffusion in the oleic phase, not aqueous
solubility of CO2 and molecular diffusion in the aqueous phase, is
performed and observes that about 10% more CO2 dissolves in the oleic
phase during the CO2 huff-n-puff process (Fig. 5B). It is also verified
that more CO2 dissolves in oil due to molecular diffusion in oil and
slightly better sweep efficiency is achieved comparing to other cases
(Fig. 6). Finally, incorporation of the effect of molecular diffusion in the
oleic phase leads to a slight increase in total production, up to Fig. 4. Cumulative volume of injecting and producing CO2 during CO2 huff-n-
2.37 × 104 bbl, about 2% higher than the base case (Figs. 3 and 7). puff process.

81
J.H. Lee, et al. Fuel 247 (2019) 77–86

Fig. 6. Effects of aqueous solubility of CO2 and molecular diffusion on the

distribution of the CO2 dissolution in oil (mole fraction) after the CO2 huff-n-
puff process: (A) base case; (B) aqueous solubility; and (C) aqueous solubility
and molecular diffusion.

Fig. 7. Cumulative oil production during the CO2 huff-n-puff process.

Fig. 5. Amount of CO2 dissolved in (A) water and (B) oil during the CO2 huff-n-
puff process. Table 1
Geochemical reactions.
Reactions
puff process with geochemical reactions results in 6% less oil produc-
tion compared to the CO2 huff-n-puff process simulation, which con- Aqueous reactions H++HCO32–↔CO2 + H2O
siders only aqueous solubility of CO2 (Figs. 7 and 11). H2O + Ca2+↔H++CaOH+
CaHCO3+↔Ca2++HCO3–
H++OH–↔H2O
4.2.2. Mineral reaction CaSO4 ↔ Ca2++SO42-
The simulations in the previous sections clarify the effect of geo- MgSO4 ↔ Mg2++SO42-
chemical reactions on the performance of CO2 huff-n-puff process. HSO4-↔H++SO42-
Another assessment explores why geochemical reactions affect the oil Mineral reaction CaCO3(s) + H+↔Ca2++HCO3–
recovery of CO2 huff-n-puff process. An additional simulation of the
CO2 huff-n-puff process ignores the reaction of mineral dissolution/
precipitation and considers only the aqueous reactions. When CO2 Ignoring the effect of the mineral reaction results in a relatively lower
dissolves in water, the amount of H+ in the water increases (Fig. 9A). pH condition (Fig. 12). In addition, the amount of HCO3–, the main
When the mineral reaction exists, the mineral potentially dissolves in product of calcite mineral dissolution, hardly changes (Fig. 9B). In
the low-pH condition and consumes H+, increasing the pH of the brine. terms of CO2 solubility, the CO2 huff-n-puff process without mineral
When the mineral reaction is ignored, H+ is not used in the mineral reaction shows a larger amount of dissolved CO2 in both phases
reaction, and a higher amount of H+ remains in the reservoir (Fig. 9A). (Fig. 10). As a result, 6% higher oil production is observed when the

82
J.H. Lee, et al. Fuel 247 (2019) 77–86

Fig. 8. Pore volume change of reservoir in the base case and the CO2 huff-n-puff
process with geochemistry.

simulation of CO2 huff-n-puff process ignores the mineral reaction

(Fig. 11), showing comparable results with the simulation that con-
siders only aqueous solubility of CO2. The solubility of CO2 in both
phases (Figs. 5 and 10) and oil production (Figs. 7 and 11) are com-
parable in the two cases. The existence of aqueous reactions produces
negligible change in performance of the CO2 huff-n-puff process. The
role of the mineral reaction and calcite dissolution is of importance
compared to that of aqueous reactions.

4.2.3. Molecular diffusion in the oleic phase

The simulations in Section 4.1.2 reveal that molecular diffusion in
the oleic phase has a positive effect on oil production of the CO2 huff-n-
puff process. The impact of molecular diffusion of CO2 in the oleic
phase is also investigated in the simulation considering geochemical
reactions. This simulation includes modeling of both aqueous and mi-
neral reactions. While molecular diffusion in the oleic phase hardly
changes the solubility of CO2 in either phase (Fig. 10), cumulative oil
production slightly increases by 3% (Fig. 11). Regardless of geochem-
istry, molecular diffusion in the oleic phase encourages oil to flow more
easily into porous media.
The results of this work clearly confirm the combined effects due to
aqueous solubility, geochemistry, and overall molecular diffusion on Fig. 9. Amounts of (A) hydrogen and (B) bicarbonate ions during the CO2 huff-
the performance of CO2 huff-n-puff process in carbonate-rich shale re- n-puff process with geochemistry.
servoirs. However, there is a limitation in the assessment of geochem-
ical reactions that affect oil flow and CO2 storage in shale oil reservoirs.
5. Conclusion
Because geochemical and mineralogical conditions vary in other shale
formations, a variety of the conditions might cause mineral dissolution
This study simulates the CO2 huff-n-puff process in a shale oil re-
or mineral precipitation or both [30]. In addition, geochemical condi-
servoir and investigates the effects of CO2 solubility in water and
tion of the hydraulic fracturing fluid could change the conditions near
comprehensive molecular diffusion on the process. Because decreasing
well bore in shale formation [31]. The mineral reaction changing the
pH, caused by CO2 dissolution in water, is an important factor in geo-
porosity and permeability enhances or hinders fluid flow in porous
chemistry, the effect of geochemical reactions is investigated in the
media. In addition, the precipitation of carbonate minerals is a long-
simulations of CO2 huff-n-puff process. The following conclusions are
term mechanism of CO2 storage. Because the target Eagle Ford shale
drawn from the numerical simulations.
reservoir of this work has a high fraction of carbonate minerals, reac-
tion of calcite mineral is of importance and calcite mineral dissolution
(1) In the simulation of CO2 huff-n-puff process, the aqueous solubility
affects the fluid flow. An assessment in other shale reservoir formation
of CO2 mitigates the dissolution of CO2 in oil. Decreasing the cu-
could draw different role of mineral reaction on the oil flow and CO2
mulative oil production by 3.2% compared to the base case, simu-
storage during CO2 huff-n-puff process. Therefore, a thorough geo-
lation with aqueous solubility stores up to 9.5% of injecting CO2 in
chemical analysis is required for an accurate assessment of CO2 huff-n-
the depleted shale formation. Consideration of aqueous solubility is
puff process in shale oil reservoirs.

83
J.H. Lee, et al. Fuel 247 (2019) 77–86

Fig. 11. Cumulative oil production during the CO2 huff-n-puff process with
geochemistry.

Fig. 10. Amounts of CO2 dissolved in (A) water and (B) oil during the CO2 huff-
n-puff process with geochemistry. Fig. 12. Distribution of pH of in the CO2 huff-n-puff process with geochemistry:
(A) with mineral reaction and (B) without mineral reaction.

required to avoid an overestimation in oil production. In addition, it

clarifies the potential of CO2 storage via solubility mechanism.
CO2 in water decreases by 29%. Consideration of geochemical re-
(2) While molecular diffusion in the aqueous phase hardly affects the
actions is necessary to accurately predict the performance of the
process, CO2 huff-n-puff process considering molecular diffusion in
CO2 huff-n-puff process in shale oil reservoirs.
the oleic phase shows 2% higher production compared to the base
(4) When the simulation of the process ignores the mineral reaction, its
case. Regardless of geochemistry, molecular diffusion in the oleic
performance is comparable with that of the process considering
phase has a positive effect on oil production, whereas the diffusion
only aqueous solubility. Existence of the mineral reaction affects
in the aqueous phase does not.
the fluid transport in the tight shale formation.
(3) Dissolution of CO2 into water decreases the pH of the in-situ brine,
(5) Because effects of aqueous solubility and geochemical reactions on
thereby calcite in the shale formation dissolves in the low-pH
oil recovery are contrary to the effect of oleic molecular diffusion, it
conditions. The geochemical reactions change the amounts of CO2
is hard to interpret and differentiate respective contribution of the
dissolved in oil and water, affecting both hydrocarbon production
potential factors to the shale oil recovery. In addition, the degree of
and CO2 storage capacity. Compared to the simulation of CO2 huff-
effect of each factor on the production would depend on the con-
n-puff process considering only aqueous solubility, the simulation
dition of the system. Therefore, an analysis in the CO2 huff-n-puff
with geochemical reactions produces 6% less oil. In terms of CO2
process for shale oil reservoirs would be sensitive to the reservoir
storage capacity via solubility mechanism, the amount of dissolved
characteristics and designs of CO2 huff-n-puff process.

84
J.H. Lee, et al. Fuel 247 (2019) 77–86

(6) This simulation work of CO2 huff-n-puff guarantees that in- Acknowledgments
corporation of aqueous solubility, geochemical reactions, and
overall molecular diffusion prevents the underestimation and This research was supported by Basic Science Research Program
overestimation in oil recovery and CO2 storage for carbonate-rich through the National Research Foundation of Korea (NRF) of South Korea
shale reservoirs. funded by the Ministry of Education (No. 2016R1D1A1B03933738).

Appendix A

Aqueous solubility
The Henry’s law determines the fugacity of dissolved CO2 in aquoue phase at reservoir condition by introducing pressure-, temperature-, and
salinity-dependent Henry’s constant (Eq. (A.1)). When temperature and pressure change, Harvey [23] published a correlation to calculate Henry’s
law constant of CO2 (Eq. (A.2)). In addition, solubility of CO2 in aqueous phase is affected by the salinity of aqueous phase. Generally, the aqueous
solubility of light gaseous component decreases with increasing salinity, which is referred as a salting-out phenomenon. Bakker [30] published a
correlation of salting-out coefficient to predict the salting-out phenomenon of CO2. The Henry’s law constant of CO2 at specific salinity is determined
by incorporating its Henry’s law constant in pure water and salting-out coefficient (Eqs. (A.3) and (A.4)).
1 p
ln Hi = ln His + v¯i dp
RT pHs O
2 (A.1)

ln His = ln pHs 2 O + 9.4234(Tr ,H2O) 1 + 4.0087(1 Tr ,H2O )0.355 (Tr ,H2 O) 1

0.41
+ 10.3199[exp(1 Tr ,H2O )](Tr ,H2 O) (A.2)

Hsalt , i
ln = ksalt , i msalt
Hi (A.3)
2 3
ksalt , i = 0.11572 6.0293 × 10 4T + 3.5817 × 10 6T 3.772 × 10 9T (A.4)
where i indicates CO2 dissolved in aqueous phase; is the Henry’s law constant at the saturation pressure of H2O, temperature, and zero salinity;
His
pHs 2O is the saturation pressure of H2O; v̄i is the partial molar volume of species i ; Tr,H2O is the reduced temperature of H2O; Hsalt , i is the Henry’s law
constant at the specific salinity condition; ksalt , i is the salting-out coefficient; msalt is the molaltiy of the dissolved salt; and T is the temperature.

Appendix B

Aqueous reaction
Chemical reaction occurs between components in the intra-aqueous phase, i.e., aqueous reaction. In the reaction, the ion activity product is
constructed by introducing the acitivty of the ion (Eq. (B.1)). The acitivity of ion indicates an effective concentration of ion and is defined with
acitivity coefficient and molality of ion (Eq. (B.2)). The activity coefficient is a function of ionic strength and charge of the ion. This study uses the B-
dot model (Eq. (B.3)) for calculating the activity coefficient [32].
n aq
Q = aivi , with naq = nc + na
i=1 (B.1)

ai = i mi , with i = 1, ···,naq (B.2)

A z i2 I
log i = + BI
1 + ai B I (B.3)
where i indicates the component of aqueous reaction; naq is the total number of components in the aqueous phase; nc is the number of gaseous
components that are soluble in the aqueous phase; na is the aqueous components that exist only in the aqueous phase; ai is the activity; i is the ionic
activity coefficient; mi is the molality; I is the ionic strength; z i is the charge of the ion; ai is the ion size parameter; and A , B , and B are the
temperature-dependent parameters.
Mineral reaction
Another chemical reaction occurs between minerals and aqueous components, i.e., mineral reaction. In the reaction, calculation of ionic acitvity
product is analogous to that of acitivity product for the aqueous reaction (Eq. (B.4)).
n
v
Q = ak k,
k=1 (B.4)
where k indicates the component in mineral reaction; n is the number of mineral components; ak is the activity of component k ; and vk is the
stoichiometric coefficient of the mineral reaction.
The mineral dissolution or precipitation controls the pore volume of the system. The change in pore volume is calculated based on the change in
total moles of the mineral (Eq. (B.5)). The change in porosity also affects the absolute permeability of the system. The formulation of Kozeny-Carman
describes the relationship between porosity and permeability (Eq. (B.6)).
n
0
N N0
=
=1 (B.5)

85
J.H. Lee, et al. Fuel 247 (2019) 77–86

3 0 2
k 1
=
k0 0 1 (B.6)
where and are the porosities after and before mineral reaction, respectively; N and N are the total moles of mineral per bulk volume after and
0 0

before mineral reaction; is the mineral molar density; and k and k 0 are the permeabilities after and before mineral reaction.

References [16] Torres JA, Jin L, Bosshart NW, Pekot LJ, Sorensen JA, Peterson K, Anderson PW,
Hawthorne SB. Multiscale modeling to evaluate the mechanisms controlling CO2-
based enhanced oil recovery and CO2 storage in the Bakken formation. In:
[1] Qiu X, Tan SP, Dejam M, Adidharma H. Simple and accurate isochoric differential Proceedings of SPE/AAPG/SEC Unconventional Resources Technology, Austin,
scanning calorimeter measurements: phase transitions for pure fluids and mixtures Texas, USA; 2325 July 2018.
in nanopores. Phys Chem Chem Phys 2019;21. 224231. [17] Yu W, Zhang Y, Varavei A, Sepehrnoori K, Zhang T, Wu K, Miao J. Compositional
[2] Dejam M, Hassanzadeh H, Chen Z. Pre-darcy flow in porous media. Water Resour simulation of CO2 huff-n-puff in Eagle Ford tight oil reservoirs with CO2 molecular
Res 2017;53(10). 81878210. diffusion, nanopore confinement and complex natural fractures. In: Proceedings of
[3] Dejam M, Hassanzadeh H, Chen Z. Pre-Darcy flow in tight and shale formations. SPE Improved Oil Recovery Conference, Tulsa, Oklahoma, USA, 1418 April 2018.
Proceedings of the 70th Annual Meeting of the APS Division of Fluid Dynamics, [18] Thomas WR, Helms LW, Driggers TK, Trice DW, Thomas GL. Energy resources
Denver, Colorado, USA; 1921 November 2017. (EOG) earnings call; May 2016.
[4] Arukhe JOI, Aguilera R, Harding TG. Solutions for better production in tight gas [19] Hoffman BT. Huff-n-puff gas injection pilot projects in the Eagle Ford. Proceedings
reservoirs through hydraulic fracturing. Proceedings of SPE Western Regional of SPE Canada Unconventional Resources Conference. Calgary, Alberta, Canada,
Meeting, San Jose, California, USA, 2426 March; 2009. 1314; March 2018.
[5] Curtis JB. Fractured shale-gas wystems. AAPG Bull 2002;86(11). 192138. [20] Fan L, Martin RB, Thompson JW, Atwood K, Robinson JR, Lindsay GJ. An in-
[6] Dejam M. Advective-diffusive-reactive solute transport due to non-Newtonian fluid tegrated approach for understanding oil and gas reserves potential in Eagle Ford
flows in a fracture surrounded by a tight porous medium. Int J Heat Mass Transf shale formation. Proceedings of Canadian Unconventional Resources Conference,
2009;128. 13071321. Calgary, Alberta, Canada; 1517 November 2011.
[7] Zhang L, Kou Z, Wang H, Zhao Y, Dejam M, Guo J, et al. Performance analysis for a [21] Gherabati SA, Hammes U, Male F, Browning J. Assessment of hydrocarbon in place
model of a multi-wing hydraulically fractured vertical well in a coalbed methane and recovery factors in the eagle ford shale play. SPE Res Eval Eng 2018;21(02).
gas reservoir. J Pet Sci Eng 2018;166:104–20. 291306.
[8] U.S. Energy Information Administration (EIA). U.S. field production of crude oil, [22] U.S. Energy Information Administration (EIA). U.S. crude oil and natural gas proved
https://www.eia.gov/dnav/pet/hist/LeafHandler.ashx?n=PET&s=MCRFPUS1& reserves, Year-end 2017; 2018.
f=M (accessed 15.02.19.). [23] Harvey AH. Semiempirical correlation for Henry's constants over large temperature
[9] Gamadi TD, Sheng JJ, Soliman MY, Menouar H, Watson MC, Emadibaladehi H. An ranges. AIChE J 1996;42(5):14914.
experimental study of cyclic CO2 injection to improve shale oil recovery. [24] Bakker RJ. Package FLUIDS 1. computer programs for analysis of fluid inclusion
Proceedings of SPE Improved Oil Recovery Symposium, Tulsa, USA; 1216 April data and for modelling bulk fluid properties. Chem Geol 2003;194:323.
2014. [25] Wilke CR, Chang P. Correlation of diffusion coefficients in dilute solutions. AIChE J
[10] Jin L, Sorensen JA; Hawthorne SB, Smith SA, Bosshart NW, Burton-Kelly ME, Miller 1955;1(2):26470.
DJ, Grabanski CB, Harju JA. Improving oil transportability using CO2 in the Bakken [26] Sigmund PM. Prediction of molecular diffusion at reservoir conditions. Part II -
system – A laboratory investigation. Proceedings of SPE International Conferenence estimating the effects of molecular diffusion and convective mixing in multi-
and Exhibition on Formation Damage Control, Lafayette, Louisiana, USA; 2426 component systems. J Can Pet Tech 1976;15(03):5362.
February 2016. [27] Orangi A, Nagarajan NR, Honarpour MM, Rosenzweig JJ. Unconventional shale oil
[11] Jin L, Hawthorne S, Sorensen J, Pekot L, Kurz B, Smith S, Heebink L, Bosshart N, and gas-condensate reservoir production, impact of rock, fluid, and hydraulic
Torres J, Dalkhaa C, Gorecki C, Steadman E, Harju J. Extraction of oil from the fractures. Proceedings of SPE Hydraulic Fracturing Technology Conference, The
Bakken shales with supercritical CO2. Proceedings of SPE/AAPG/SEC Woodlands, Texas, USA, 2426 January; 2011.
Unconventional Resources Technology, Austin, Texas, USA; 2426 July 2017. [28] Metz B, Davidson O, de Coninck H, Loos M, Meyer L. IPCC Special Report on Carbon
[12] Kang SM, Fathi E, Ambrose RJ, Akkutlu IY, Sigal RF. CO2 storage capacity of or- Dioxide Capture and Storage. UK and New York, NY, USA: Cambridge University
ganic-rich shales. Proceedings of SPE Annual Technical Conference and Exhibition, Press; 2005.
Florence, Italy; 1922 September 2010. [29] Wolery TJ, Daveler SA. EQ6, a computer program for reaction path modeling of
[13] Alfarge D, Wei M, Bai B, Almansour A. Effect of molecular-diffusion mechanisim on aqueous geochemical systems: theoretical manual, user’s guide, and related doc-
CO2 huff-n-puff process in shale-oil reservoirs. Proceedings of SPE Kingdom of umentation (version 7.0). Lawrence Livermore National Laboratory; 1992.
Saudi Arabia Annual Technical Symposium and Exhibition. Dammam, Saudi [30] Pearce JK, Turner L, Pandey D. Experimental and predicted geochemical shale-
Arabia; 2427 April 2017. water reactions: Roseneath and Murteree shales of the Cooper Basin. Int J Coal Geol
[14] Jia B, Tsau J-S, Barati R. Role of molecular diffusion in heterogeneous shale re- 2018;187:3044.
servoirs during CO2 huff-n-puff. Proceedings of SPE Europec featured at 79th EAGE [31] Paukert Vankeuren AN, Hakala JA, Jarvis K, Moore JE. Mineral reactions in shale
Conference and Exhibition, Paris, France; 1215 June 2017. gas reservoirs: barite scale formation from reusing produced water as hydraulic
[15] Pankaj P, Mukisa H, Solovyeva I, Xue H. Boosting oil recovery in naturally fractured fracturing fluid. Environ. Sci. Technol 2017;51(16). 93919402.
shale using CO2 huff-n-puff. Proceedings of SPE Argentina Exploration and [32] Bethke CM. Geochemical reaction modelling. NY, USA: Oxford University Press;
Production of Unconventional Resources Symposium, Neuquen, Argentina; 1416 1996.
August 2018.

86