You are on page 1of 25

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/334436126

Multifracture response to supercritical CO2 ‐EGS and water‐EGS based on


thermo‐hydro‐mechanical coupling method

Article  in  International Journal of Energy Research · July 2019


DOI: 10.1002/er.4743

CITATIONS READS

0 107

7 authors, including:

Kelvin Bongole Sun Zhixue


China University of Petroleum China University of Petroleum - Beijing
8 PUBLICATIONS   10 CITATIONS    14 PUBLICATIONS   218 CITATIONS   

SEE PROFILE SEE PROFILE

Yueying Wang
China University of Petroleum
15 PUBLICATIONS   176 CITATIONS   

SEE PROFILE

All content following this page was uploaded by Kelvin Bongole on 29 October 2019.

The user has requested enhancement of the downloaded file.


Received: 26 April 2019 Revised: 26 June 2019 Accepted: 1 July 2019
DOI: 10.1002/er.4743

RESEARCH ARTICLE

Multifracture response to supercritical CO2‐EGS and water‐


EGS based on thermo‐hydro‐mechanical coupling method

Kelvin Bongole | Zhixue Sun | Jun Yao | Asif Mehmood | Wang Yueying |
James Mboje | Ying Xin

School of Petroleum Engineering, China


University of Petroleum (East China),
Summary
Qingdao 266580, China Hydraulic‐fracturing treatments have become an essential technology for the
development of deep hot dry rocks (HDRs). The deep rock formation often con-
Correspondence
Zhixue Sun, School of Petroleum tains natural fractures (NFs) at micro and macroscales. In the presence of the
Engineering, China University of NF, the hydraulic‐fracturing process may form a complex fracture network
Petroleum (East China), Qingdao 266580,
caused by the interaction between hydraulic fractures and NF. In this study,
China.
Email: upcszx@upc.edu.cn analysis of carbon dioxide (CO2)‐based enhanced geothermal system (EGS)
and water‐based EGS in complex fracture network was performed based on
Funding information
the thermo‐hydro‐mechanical (THM) coupling method, with various rock con-
National Natural Science Foundation of
China, Grant/Award Number: 51774317; stitutive models. The complexity of the fracture geometry influences the fluid
National Science and Technology Major flow path and heat transfer efficiency of the thermal reservoir. Compared with
Project, Grant/Award Number:
CO2‐based EGS, water‐based EGS had an earlier thermal breakthrough with a
2016ZX05011004‐004; Fundamental
Research Funds for the Central Universi- rapid decline in production temperature. CO2 can easily gain heat rising its
ties, Grant/Award Number: 18CX02100A temperature thus reducing the effect of a premature thermal breakthrough.
Both CO2‐based EGS and water‐based EGS are affected by in‐situ stress; the
increase in stress ratio improved the fracture permeability but resulted in an
early cold thermal breakthrough. When the same injection rate is applied to
water‐based EGS and CO2‐based EGS, water‐based EGS displayed higher injec-
tion pressure buildup. Water‐based EGS had higher reservoir deformation area
than CO2‐based EGS, and thermoelastic constitutive model for water‐based
EGS showed larger deformed area ratio than thermo‐poroelastic rock model.
Furthermore, higher values of rock modulus accelerated the reservoir deforma-
tion for water‐based EGS. This study established a novel discussion investigat-
ing the performance of CO2‐based EGS and water‐based EGS in a complex
fractured reservoir. The findings from this study will help in deepening the
understanding of the mechanisms involved when using CO2 or water as a
working fluid in EGS.

KEYWORDS
CO2‐EGS, geothermal reservoir, thermo‐poroelastic, THM, water‐EGS

Int J Energy Res. 2019;1–24. wileyonlinelibrary.com/journal/er © 2019 John Wiley & Sons, Ltd. 1
2 BONGOLE ET AL.

1 | INTRODUCTION study by injecting CO2 into fractures to study their geo-


physical characteristics and observed a significant seismic
Geothermal energy is one among the clean, renewable, velocity upon CO2 injection. The author's result focused
and sustainable source of heat and baseload power. Effi- on a single fracture in various topology. However, the
cient utilization of geothermal resources will reduce seismic velocity could be elevated for a complex fracture
dependence on fossil fuels, leading to a cleaner global distribution system, and the use of a single fracture over-
environment. The techniques and designs of geothermal looks the fracture and matrix pore variation.18 Shi
energy extraction from the subsurface reservoirs have et al19,20 suggested the use of multilateral wells for heat
evolved in multiple ways such as geothermal closed‐loop extraction in EGS with CO2 as the heat transmission
systems including wellbore heat exchanger1 and the use fluid. The authors molded planar fractures connecting
of abandon oil and gas wells.2,3 High‐enthalpy geother- injection and production wells into the reservoir geome-
mal resources are stored in deep formations of hot dry try. They pointed out the significance of multilateral wells
rocks (HDRs), having less formation permeability and to improve heat extraction performance. Wang et al11
porosity. Enhanced geothermal system (EGS) was performed numerical simulations using a coupled
designed to exploit these HDRs by hydraulic‐fracturing thermo‐hydro continuum model. The authors showed
(HF) treatments to reactivate and induce fractures in that higher heat extraction could be achieved in CO2‐
the subsurface rock. The successful implementation of based EGS than water‐based EGS for more substantial
EGS project in Fenton Hill4 motivated other countries formation permeability. The continuum model used by
to exploit and develop their deep HDRs resources.5 Wang et al11,21 assumed fracture properties embedded in
Water and carbon dioxide (CO2) are popular working the matrix; this fails to mimic a fractured reservoir accu-
fluid mostly discussed in the context of EGS. Water has rately. Guo et al22 conducted a performance evaluation of
been used as heat transmission fluid in multiple EGS pro- EGS using CO2 as a working fluid under THM coupling
jects compared with the idea of using CO2. Brown6 ini- technique. The authors showed that the complexity of
tially proposed CO2‐based enhanced geothermal system the fractured reservoir to mimic an EGS should include
(CO2‐EGS) in which supercritical CO2 was used as a stochastic and discrete fracture networks embedded into
working fluid for heat extraction. The author pointed porous media. Also, high‐fracture aperture and perme-
out multiple advantages of using CO2 as a reservoir circu- ability reduce the heat mining rate for the CO2‐based
lating fluid for effective heat mining. Numerous studies EGS, while higher connectivity of fracture networks
have reported the significance of using supercritical CO2 improves the heat extraction rate. The constitutive
in thermal reservoirs7,8 because of its high heat extraction models employed in numerical analysis, particularly with
rate,9 large expansivity, and compressibility.10 Tempera- fluid properties that vary with reservoir conditions, influ-
ture variation in CO2 fluid can produce a high‐density ence the accuracy of the simulated results. Various consti-
ratio that induces significant buoyancy force, reducing tutive models of thermo‐poroelastic,23,24 poroelastic,25,26
the pumping power required to drive the working and thermoelastic27 on single fracture have widely been
fluid.6,10 Furthermore, with the recent rise in global tem- discussed.
perature, efforts to reduce CO2 emissions are highly However, in the above studies, the complexity of the
encouraged, and the idea of using CO2 as working fluid fracture geometry and propagation was not considered.
coupled with CO2 sequestration and heat mining in geo- Also, the previous studies used a single constitutive rock
thermal reservoirs has attracted a great deal of attention model in the analysis of CO2‐based EGS and water‐based
among researchers and industry communities.11,12 EGS. Water and CO2 flow behavior may vary depending
EGS at reservoir scale involves complex geometry and on the connectivity and density of the fracture network.
propagation of fractures resulting from reactivation of the In this paper, a comparison between CO2‐based EGS
existing natural fractures, fracture roughness,13 and inter- and water‐based EGS is performed in a hydraulic frac-
action between hydraulic stimulated fractures with natu- tured geothermal reservoir (Figure 1) with the thermo‐
ral fractures. The structure and assembly of the fracture hydro‐mechanical (THM) coupling method.28 The study
void14 and matrix pores15 influence the fluid flow behav- considers the application of the thermo‐poroelastic con-
ior and are the source of complexity in determining the stitutive model in comparison with the thermoelastic
reservoir porosity to be used in numerical analysis. Luo rock model with various parameters of the investigation.
et al16 performed simulations using water and CO2 as The term carbon dioxide (CO2) used in this study infers
working fluid and compared the performances of CO2‐ to the supercritical CO2. Matrix deformation is analyzed
based EGS and water‐based EGS. The authors used a sim- based on the Coulomb failure criteria for stress difference
plified EGS model, a single fracture with various fracture between a fracture and the nearby matrix. Furthermore,
permeability values. Borgia et al17 performed a simulation the influence of varying in‐situ stress on water‐based
BONGOLE ET AL. 3

FIGURE 1 Distribution of stimulated


and natural formation fractures (modified
from Zhang et al34) [Colour figure can be
viewed at wileyonlinelibrary.com]

EGS and CO2‐based EGS is studied based on the interaction30 and changing geometries of hydraulic frac-
thermoelastic model. ture growth.33 As shown in Figure 1, the primary hydrau-
lic fractures are generated by the initial fluid intrusion
into the formation from the wellbore perforation.29
2 | D E S C R I P T I O N O F TH E Microfractures consist of small natural fractures and the
FRACTURE INTERACTIONS A ND reactivated cracks after the fracturing process. The stimu-
NETWORK SYSTEM lated reservoir volume (SRV) is an extensive distributed
fracture network system with high connectivity close to
HF treatments have become an indispensable part of the fractured wells. This region plays a significant role
wells completion in deep geothermal reservoirs. The mul- in the effective fluid flow and heat exchange process in
tistage fracturing of horizontal wells is commonly used to fractured geothermal reservoirs.
improve the permeability in unconventional petroleum
reservoirs and in the deep formations of the HDRs.29
The deep HDRs' formation may contain natural fractures 3 | F L U I D PR O P E R T I E S WI T H
that may interact with the hydraulic fractures during the TE M P E R A T U R E AN D PRE S S U R E
stimulation process. The interaction between fractures
may alter the way hydraulic fracture propagates, resulting The fluids used in this study are sensitive to reservoir
in a compound fracture network. There are several conditions. Unlike water properties, CO2 physical proper-
reported types of fracture interaction: crossing, slippage, ties rapidly vary with slight changes in temperature and
dilation, activation, crossing with activation, and offset.30 pressure. To accurately represent CO2‐based EGS, a
The formation of complex fracture network and their precise model of CO2 variation with temperature and
interaction has a strong dependence on in‐situ stresses, pressure is required. Span and Wagner35 proposed an
rock mechanical properties, natural fracture strike,31 equation of state that is proven to model CO2 at a wide
and the hydraulic‐fracture treatment methods, including range of temperature and pressure (216.95 K < T <
injection scheme and fracturing‐fluid properties.32 1100 K, 0.52 MPa < p < 800 MPa) with high accuracy.
At high confining temperature, Frash et al33 The proposed series of equations by Span and Wagner,35
experimented HF of granite rock, a similar rock found Heidaryan et al,36 and Jarrahian and Heidaryan37 are
in deep geothermal reservoirs. The authors showed multi- used to model the different properties of CO2. Readers
ple hydraulic fractures of different geometries growing are urged to visit the National Institute of Standards
outwards from the fracturing well. Also, using the same and Technology (NIST) chemistry webbook for detailed
well, multiple hydraulic fracture stimulation treatments clarification on fluid properties variation.38
may be performed to achieve significantly increased res- Under high temperature and pressure, water proper-
ervoir conductivity and well‐fluid communication. In this ties are not constant; their physical properties depend
study, the 2D model (Figure 1) is constructed to mimic on reservoir conditions. Various authors discussed
the complex fracture network34 with random fracture models and equation of states that relates changes in
4 BONGOLE ET AL.

water properties with temperature and pressure.39,40 The Kd


αβ ¼ 1 − (8)
density of water varies with temperature (Ts) and pres- Ks
sure (p), can be written as28
The matrix and fracture storage coefficient “S” is
1=ρf ¼ 3:086 − 0:899017ð4014:15−T s Þ 0:147166
accounted by the porosity (θ), fluid compressibility “x f ”,
(1) Pa−1, solid drained (Kd), and solid bulk modulus (Ks),
− 0:39ð658:15−T s Þ−1:6 ðp − 225:5Þ þ δ;
Pa. Furthermore, there is a separate input file for the var-
where the value “δ” does not exceed 6% of 1/ρ f . Temper- iation of fluid properties with temperature and pressure,
ature variation affects the dynamic viscosity of water and and their updated values at each time step are supplied
is expressed as41 to the model equations.

η ¼ υρf ; (2)
4.2 | Equilibrium model
0:01775
υ¼ ; (3) The equilibrium equation for stress tensor in a porous
1 þ 0:033T f þ 0:000221T 2f medium saturated with fluid can be written as28
where, υ is the kinematic viscosity of the flowing water, σ ij; j þ F i ¼ 0; (9)
Pa·s. Temperature affects the density and viscosity of
the flowing water, which, in turn, influence the water
E E
transport and heat transfer behavior in geothermal ui; jj þ uj; ji − αB p;i
2 ð1 þ ν Þ 2ð1 þ νÞð1 − 2νÞ
reservoirs.
E
− αT T s;i þ F i ¼ 0;
ð1 − 2νÞ
4 | GOVERNING EQUATIONS
where σij,j is the stress tensor; u is the displacement, m; F i
4.1 | Fluid flow model is the body force per unit area in the i‐coordinate (i = x, y,
in 2D); E is the elastic modulus, Pa; ν is Poisson's ratio; p is
Fluid flow equations in the matrix and fracture medium the hydraulic pressure, Pa; αT is the thermal expansion
are governed based on the modified Darcy law in a coefficient. The parameters “αBp,i” and “αTTs,iE/(1 − 2ν)”
poroelastic rock model42 correspond to the hydraulic pressure and thermal‐
  mechanical interactions, respectively.
∂p km ∂ε A pair of surfaces represents the fracture by which
ρf S þ ∇⋅ρf − ∇p ¼ −Qfr − ρf αβ ; (4)
∂t η ∂t shear and normal displacements are permissible. The
two fracture surfaces (top and bottom surfaces) can move
∂p ∂ε perpendicular (up and down movement) or parallel (shear
ρf dfr S þ ∇τ ⋅dfr ρf vf ¼ −dfr ρf αβ þ dfr Qfr ; (5) displacement) to each other. The normal displacement of
∂t ∂t
the fracture surfaces results in opening or closing of the
kfr fracture void, while the shear movement of the fracture
vf ¼ − ∇τ p; (6) surfaces results to shearing of the fracture. Both the shear
η
and normal displacements of the fracture surfaces lead to
where S is the specific storage term, Pa−1; p is the hydrau- changes aperture, thus affecting the fracture permeability.
lic pressure, Pa; km is the matrix permeability, m2; η is the The fracture deformation equation can be written as28
dynamic fluid viscosity, Pa·s; ρ f is the fluid density,
kg/m3; kfr is the fracture permeability, m2; ∇τ is the gradi- un ¼ σ ′n =τ n ; us ¼ σ ′s =τ s ; (10)
ent operator constrained to the fracture's tangential
plane, dfr is the fracture aperture, m; αβ is the Biot's
σ ′n ¼ σ n − αB p; σ ′s ¼ σ s ; (11)
constant, and ε is the volumetric strain. The term “
∂ε
ρf αβ ” describes the mechanical influence of the solid where; u is the displacement, m; σ is the total stress, Pa; σ′
∂t is the effective stress, Pa; τ is the fracture stiffness, N/m.
on the seepage process, and Qfr is the term that describes
Fracture stiffness refers to the resistance in the movement
the fluid flow exchange between the fracture and the
of the fracture surfaces under the application of external
matrix.
force. The subscripts “n” and “s” represent the normal
 1 − αβ and tangential directions, respectively, on the fracture
S ¼ ϕx f þ αβ − ϕ (7)
Kd plane.
BONGOLE ET AL. 5

4.3 | Energy balance model 4.4 | Description of the constitutive rock


models
The nonequilibrium method of heat transfer considers
two different equations for matrix and fracture systems. Thermo‐poroelastic constitutive rock model is established
Heat transfer in porous media involves the matrix and by combining the influence of thermal stresses and the
fluid inside the rock pores. In the fracture network, heat difference between matrix expansion to rock stresses
transfer occurs within the flowing fluid and heat together with fluid diffusion when rock structure changes
exchange between the fracture and the surrounding (variation in pore volume). In this study, the thermo‐
matrix. Therefore, the fracture equation has fluid compo- poroelastic rock model follows the equations defined in
nents except for the fracture aperture (Equation 13). The the fluid flow model (Section 4.1), the equilibrium model
following energy balance equations describe heat transfer (Section 4.2), and the energy balance model (Section 4.3).
in porous media and fracture network43,44 In summary, the thermo‐poroelastic constitutive rock
model uses the governing equations as described from
∂T s Equations 4 to Equation 14.
ðρCÞeff ¼ λeff ∇2 T s þ W s ; (12)
∂t The thermoelastic constitutive rock model combines
the effect of thermal stress and the difference between
  matrix expansion to the rock stresses without considering
∂T f
dfr ρf Cf þ dfr ρf Cf ⋅vf ∇τ T f ¼ dfr ∇τ ⋅ λf ∇τ T f þ W fr ; the fluid dispersal because of the changes in rock struc-
∂t
ture (poroelastic rock behavior). The Biot's constant is
(13)
the main leading parameter determining the degree of
change in pore structure during fluid withdraw. Biot's
where (ρC)eff = (1 − ϕ)ρsCs+ϕρ f C f and λeff = (1 − ϕ)λs
constant can be defined as the ratio of the volume change
+ϕλ f are effective heat capacity and effective thermal con-
of the fluid‐filled porosity to the volume change of the
ductivity, respectively, obtained by volume average
rock when the fluid is free to move out of the rock (ie,
method. Ts is the average matrix temperature, K; T f is
at constant hydraulic pressure). Therefore, for
the average temperature of the fluid in fractures, K; ρs is
thermoelastic constitutive rock model, the Biot's constant
the density of the matrix, kg/m3; ρ f is the density of the
has a unit value (αβ = 1). The fluid flow equations in the
fluid, kg/m3; v f is the fluid flow velocity in fractures as
matrix and fracture medium for the thermoelastic model
expressed in Equation 6, m/s. Cs is the specific heat capac-
can be defined as
ity of the solid, J/kg/K; C f is the specific heat capacity of
the fluid, J/kg/K; λs is the heat conductivity of the solid,  
∂p km ∂ε
and λ f is the heat conductivity of the fluid, W/m/K. Ws ρf S þ ∇⋅ρf − ∇p ¼ −Qfr − ρf ; (15)
∂t η ∂t
is the external heat source supplied to the matrix, W/m3;
and Wfr is the heat exchange between the matrix and its  
∂p k fr ∂ε
surrounding fractures, W/m. For deep geothermal reser- ρf dfr S þ ∇τ ⋅dfr ρf − ∇τ p ¼ −dfr ρf þ dfr Qfr : (16)
voirs, constant heat flux from the surrounding HDRs sup- ∂t η ∂t
plies thermal flux to the targeted EGS. The heat flux
supplied to EGS from the surrounding HDRs is expressed S ¼ ϕx f (17)
as “Ws” as seen in Equation 12. While, within the EGS,
the heat exchange between the matrix and its nearby frac- The equilibrium equation for stress tensor in a porous
ture is expressed as “Wfr” as seen in Equation 13. medium saturated with fluid considering thermoelastic
The velocity of fluid inside the porous rock is smaller model can be rewritten as
than in fractures. Therefore, it can be assumed that the
σ ij; j þ F i ¼ 0; (18)
temperature of the fluid in the matrix pores is identical
to that of the rock. Heat exchange between the fluid in
the fractures to the surrounding matrix follows Newton's E E
ui; jj þ uj; ji − p;i
heat transfer equation.43 2ð1 þ νÞ 2ð1 þ νÞð1 − 2νÞ
E
  − αT T s;i þ F i ¼ 0:
W fr ¼ h T s − T f ; (14) ð1 − 2νÞ

For reservoirs with a high content of rock‐forming


where h is the convection coefficient, W/m2/K. When “h” minerals, the thermo‐poroelastic constitutive model is
is sufficiently large, the temperature of rock and that of preferable since the mineral constituent of rock affects
water are equal at the fracture surface. its structure under the influence of stress or pore
6 BONGOLE ET AL.

 
pressure. The thermo‐elastic constitutive model can be kfr ¼ k o exp −ασ ′n ; (19)
applied in reservoirs with high rock strength, ie, the rock
structure cannot deform easily under influence pore pres- where kfr is the fracture permeability, m2; κo is the base-
sure or external stresses. line fracture permeability, m2; σ ′n is the effective normal
stress on the fracture plane, Pa; and α is a normalizing
coefficient, Pa−1.
5 | COUPLED PARAMETERS A ND Yu et al46 conducted a series of laboratory experiments
THEIR C HARACTERISTICS to study the thermal physical and mechanical properties
of the granite. At temperatures of 0°C to 200°C, the
This section explains the underlying coupling phenom-
authors developed mechanical‐temperature models for
ena that are associated with THM processes. The cou-
granite rock. The Young's modulus and Poisson's ratio
pling effect in EGS refers to the mutual influence of
vary with temperature as follows:46
stress field, temperature field, and flow field in thermal
reservoirs. As shown in Figure 2, the geomechanical rock v ¼ 0:0004T s þ 0:1185; (20)
property has a two‐way link with the heat variations and
fluid flow. Changes in temperature, either by injection of
cold fluid or external heat flux, would result in thermal E ¼ −0:014T s þ 29:997; (21)
stresses, and the produced stresses can cause rock failure
where Ts is the matrix temperature; °C; E is elastic mod-
or alter the fracture permeability. Fluid flow is also
ulus, GPa; v is the Poisson's ratio.
affected by temperature changes by altering fluid density
and viscosity. The flow in fractures and fluid inside the
matrix pores can impact the rock effective stress through 6 | M OD E L VAL I D A T IO N
the fluid pressure variations.
Permeability is a vital parameter of fluid flow, espe- Before the numerical investigation of the water‐based
cially in the densely fractured reservoir, and depends on EGS and CO2‐based EGS with multifractures, the accu-
the rock stress variation. Because of poroelastic behavior racy of the proposed model should be verified. Our math-
considered in this study, the initial fracture permeability ematical models and numerical solving procedures follow
(ko) depends on the initial pore pressure, where the Biot's the earlier work by Sun et al.47 The detailed analysis for
constant (αβ) is the determinant factor in the level of 2D THM coupling validation was performed in the previ-
influence. Under tensile stress (positive normal stress), ous work.47 Sun et al47 validated the THM coupling pro-
the fracture will open and increase the permeability; oth- cess using a 1D thermal consolidated model with a
erwise, the permeability will decrease under compressive series of analytical solutions proposed by Bai.48
stress. Stress to fracture permeability relation can be The accuracy of the thermo‐hydraulic (TH) analysis is
expressed as45 validated by the analytical model proposed by Cheng

FIGURE 2 A thermo‐hydro‐mechanical flow diagram [Colour figure can be viewed at wileyonlinelibrary.com]


BONGOLE ET AL. 7

et al.49 In this study, the presented solution is based on an (2) The rock matrix extends in the y‐direction to ±∞
idealized geometry, as depicted in Figure 3. The geometric value, while the fracture plane extends to infinity
assumptions for the numerical model consider a single‐ x‐direction.
rectangular fracture of thickness “dff” penetrating the (3) The fluid is assumed to flow only in fracture at a
entire height of the reservoir and dividing a 3D block into constant velocity, and the fracture thickness is
two homogenous, isotropic, and impermeable individual unchangeable. Furthermore, the rock is considered
blocks. Hence, the solution geometry is a two‐dimensional impermeable.
geometry by considering analysis through the fracture (4) Heat convection occurs along the fracture plane in
plane, as shown in Figure 4. Also, the fracture aperture is the x‐direction, and heat conduction occurs in
constant, and the fracture walls are impermeable. Further- matrix block in y‐directions (Figure 4). At the frac-
more, water is assumed to be incompressible, and all the ture plane, the temperature of the fluid and the rock
thermal properties, including conductivity and heat capac- are assumed to be equal.
ity, are constant. The numerical solution is solved by (5) The physical properties of water are constant, ie,
COMSOL50 using finite element method (FEM). independent to pressure and temperature changes.
Water does not change its phase throughout the
flow process.
6.1 | The basic assumptions and
mathematical expression for the TH
The temperature distribution on the block matrix is
analytical model
given by49
The derivation of TH analytical models49 adopts simplified
T o − T f ðx; t Þ
1D heat conduction with a single‐fracture surface. One‐ T sD ðx; y; tÞ ¼
T o"− T wo
dimension geometry (see the lower part of Figure 4) with  sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi#
λm x uf ρm cm
a single‐fracture line is considered in the derivation of ¼ erfc jyj þ   :
dff uf ρw cw 4λm uf t − x
the analytical solution. The assumptions considered in
obtaining the analytical solution for the mathematical (22)
model are as follows:
As the fluid is assumed to flow only along the fracture,
(1) No external heat is applied to the system bound- the temperature distribution along the fracture is identi-
aries; thus, the viscous diffusion and radiative heat cal to the fluid distribution temperature. The analytical
transfer are ignored. model for fluid temperature distribution is given by49

FIGURE 3 An idealized model for heat


extraction with a single fracture [Colour
figure can be viewed at wileyonlinelibrary.
com]
8 BONGOLE ET AL.

FIGURE 4 Solution domain for the mathematical problem

T o − T f ðx; t Þ TABLE 1 Computational parameter for the validation model


T fD ðx; t Þ ¼
T o"− T wo
 sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi# λm, W/m/K ρm, kg/m3 Cw, J/kg/K
λm x uf ρm cm
¼ erfc   ; (23) 3 2700 4200
dff uf ρw cw 4λm uf t − x
2dff (m) To (°C) Two (°C)
0.0004 150 60
where ρw and ρm are the density of water and matrix,
Cm (J/kg/K) ρw (kg/m )3
uf (m/s)
respectively, kg/m3; cw and cm are the specific heat capac-
1000 1000 0.01
ity of water and matrix, respectively, J/kg/K; λm is the
matrix thermal conductivity W/m/K; dff is the fracture
aperture, m; u f is the water flow velocity inside the frac-
ture, m/s; x and y are material coordinates, m. 7.1 | Boundary and initial conditions
A numerical solution is presented to model 3D heat
The investigation of the water‐based EGS and CO2‐based
transfer and extraction problem with a finite domain of
EGS is simulated based on the geometric model displayed
100 × 100 × 100 m. The numerical model is used in com-
in Figure 6. The THM analysis for the two EGS is per-
parison with the analytical model, and the computational
formed for 7000 days with a simulation time step of 10
parameters used in this model are listed in Table 1.
days. Absolute tolerance of 1 × 10−6 is set and regarded
Figure 5A shows temperature variation along the fracture
as the convergence criterion of the numerical solutions.
plane at different times, ie, 20, 90, and 500 days. The iso-
If the iterative error is smaller than the absolute toler-
thermal temperature distribution across the matrix block
ance, the numerical solutions will achieve convergence
at t = 1000 days is displayed in Figure 5B. We can observe
with the desired minimum error.
a good agreement between the analytical solution and the
Furthermore, a segregated solution approach is used to
numerical solution.
solve different mathematical models in COMSOL.50 The
initial and boundary conditions are as follows:
7 | A C ONCEPTUAL M ODEL OF
THE HYDRAULIC FRACTURED (1) Displacement field: The outside boundaries of the
RE SERV OIR model are all constrained in the normal direction.
Fracture movement is permissible, and their move-
Advances in hydraulic fracturing make it possible to ment defines fracture closing or opening as a result
stimulate large reservoir areas over 2 km2. However, frac- of stress changes. Injection and production wells
ture connectivity is an essential factor for effective heat are fixed.
extraction performance. Fractures in our 2D model were (2) Thermal field: The initial reservoir temperature is at
constructed to have a higher level of connectivity mim- 473 K, and the injection fluid is at 333 K for both
icking a real‐hydraulic fractured reservoir. The geometric water‐based EGS and CO2‐based EGS. The outside
model has dimensions of 915 × 585 m with two produc- boundaries of the reservoir are thermally insulated;
tion wells placed at sides of the model approximately more focus is put on heat exchange within the stim-
220 m from the center injection well (see Figure 6). An ulated reservoir area.
imaginary line AB is drawn on the geometric model for (3) Seepage field: The injection well is located at the
various parametric analysis. center of the model at an injection pressure of 17
BONGOLE ET AL. 9

simulation accuracy because this area is an important


part where the heat exchange and fluid flow processes
occur. Furthermore, areas around of fracture intersection
and the outer boundaries of the model are more refined
to improve simulation precision (see circled regions in
Figure 7). To guarantee the simulation results are mesh‐
independent, multiple simulations at different mesh qual-
ity (see Table 2) were performed. The wells production
temperatures after 7000 days under various mesh quality
are shown in Figure 8. It can be seen that the production
temperature evolution for different mesh qualities has a
negligible difference. A further increase in mesh quality
has no significant impact on the accuracy of the results,
instead increases the simulation time (Table 3).

8 | FAILURE PARAMETERS

Failure expressions are mostly incorporated in reservoir


research to understand the integrity of matrix/fractures53
or wellbore54 under the influence of stress. In this study,
we consider the fluid pressure together with the stress
changes from heat flux between the fracture and its sur-
rounding matrix. The formulated mathematical equation
of Coulomb failure relates fracture failure, three principal
stresses (resultant stresses from heat and fluid fluxes),
and the pressure difference between a fracture and its
surrounding matrix. Our 2D geometric model considers
a plane strain with a thickness far less than the height
and width dimensions. The parameters related to granite
FIGURE 5 Analytical temperature distribution results. (A) failure in Equation 24 are derived from Lajtai and Gadi
Temperature variation along the fracture at a different time and (1989),55 and the failure equation can be expressed as53
(B) isothermal matrix temperature distribution [Colour figure can
be viewed at wileyonlinelibrary.com]
   
ψ¼ σ 3 þ C1*ΔpðtÞ − Q* σ 1 þ C1*ΔpðtÞ
 (24)
MPa and production wells placed at sides of the geo- þ N*ð1 þ ðσ 2 − σ 1 Þ=ðσ 3 − σ 1 ÞÞ=C2 ;
metric model at a pressure of 12 MPa. The outer
boundaries are restricted to cross flow. Furthermore,
for water‐based EGS, the initial fluid in the reservoir Q ¼ ðð1 þ sinθÞ=1 þ sinθÞ; (25)
is water and for CO2‐based EGS; the reservoir was
initially saturated with supercritical CO2. The initial N ¼ ðð2 cosθÞ=ð1 − sinθÞÞSo ; (26)
CO2 saturation for the CO2‐based EGS comes from
the assumption that the supercritical CO2 was used where, σ1, σ2, and σ3 are the principal stresses, Pa. It
in the fracturing process in creating the artificial should be noted that the principal stresses are calculated
reservoir.51,52 within the model and updated continuously during the
simulation. ψ is the failure coefficient; P(t) is the pressure
difference between the fracture and its surrounding
matrix, Pa; θ is the friction angle, degrees; So is the Cou-
7.2 | Simulation mesh lomb Cohesion force, Pa; C1 and C2 are calibration
parameters. The parameters related to the “fail” expres-
An example of the meshing scheme for the 2D model is sion are listed in Table 4. The important turning parame-
illustrated in Figure 7. The mesh configuration consists ters of failure expression are; ψ = 0 indicates the onset of
of triangular, vertex, quads, and edge elements. The rock failure; ψ < 0 designates failure, and ψ > 0 predicts
meshes in the region of SRV are refined to improve stability.
10 BONGOLE ET AL.

FIGURE 6 2D geometric model of the


stimulated reservoir [Colour figure can be
viewed at wileyonlinelibrary.com]

FIGURE 7 Finite element mesh for 2D


domain [Colour figure can be viewed at
wileyonlinelibrary.com]

TABLE 2 Model meshing parameters between the injector and producer wells. Figure 9
Total Number of Mesh Quality
displays the evolution of the spatial distribution of reser-
Mesh ID Elements (%) voir pressure for water‐based EGS and CO2‐based EGS
at t = 360 days. It can be seen that CO2‐based EGS
Coarser mesh 78 984 77.07
(Figure 9A,C) has a broader region of high pressure than
Fine mesh 145 340 82.221 water‐based EGS. The broader pressure region for CO2‐
Extra fine mesh 523 016 88.02 based EGS results from larger flow potential of CO2.
The higher flow potential of CO2 is a result of higher den-
9 | R E S U L T S AN D D I S C U S S I O N sity to viscosity ratio (ρf/ηf) of CO2 than that of water.
To further explain the fluid pressure distribution
9.1 | Fluid pressure distribution within within the model, a 3D plot with pressure values
the fractured model displayed as height (z‐axis) is shown in Figure 9C,D. It
can be seen that fractures alter the fluid pressure distribu-
A better understanding of pressure distribution is vital for tion as displayed by the pressure ripples in the circular
monitoring the performance of EGS. The flow behavior in region of Figure 9D. Furthermore, pressure around the
fractured reservoir depends on the pressure difference injection well is high for both cases (water and CO2‐based
BONGOLE ET AL. 11

FIGURE 8 Variation in well


production temperature with different
mesh qualities [Colour figure can be
viewed at wileyonlinelibrary.com]

TABLE 3 Computational parameters TABLE 4 List of failure parameters53,55

Parameters Value Units Friction angle (θ) 49.8 degrees


7
Thermal conductivity of matrix (λs) 3 W/m/K Coulomb Cohesion force (So) 1.6 × 10 Pa
Thermal conductivity of water (λf) 0.6 W/m/K Calibration parameter 1 (C1) 14.7 Unitless
3
Density of carbon dioxide f (T,P) kg/m Calibration parameter 2 (C2) 40 Unitless
Density of solid (ρs) 2700 kg/m3
Density of water (ρw) f (T) kg/m3
Matrix permeability (κ) 1.0 × 10−16 m2 D). The production well is maintained at 12 MPa, slightly
Fracture initial permeability (κo) f (initial stress) m2 lower than the initial reservoir pressure. The reduced
Heat capacity of solid (cs) 1000 J/kg/K pressure at the production well permits a continuous flow
Heat capacity of fluid (cf) f (T) J/kg/K
of the reservoir fluids through the wellbore to the surface
equipment. In this study, a constant injection‐production
Fluid compressibility (xf) f (P) 1/Pa
pressure difference of 5 MPa was adopted for both water‐
Viscosity of fluids () f (T,P) Pa·s based EGS and CO2‐based EGS. A similar injection‐
Matric porosity (ϕ) 0.015 Unitless production pressure difference was used by Cui et al56
Fracture porosity (ϕ f ) 0.75 Unitless and Wang et al11,57 in numerical simulations for geother-
Fracture aperture (dfr) 0.005 m mal recovery during CO2 injection.
To better understand the fluid pressure distribution
Poisson's ratio (v) 0.25 Unitless
within the reservoir, imaginary line AB (see Figure 6) is
Coefficient of heat transfer (h) 300 W/m2/K drawn along the x‐axis in the model. Figure 10 shows
Injection temperature (T inj) 333 K pressure distribution along the line AB for water‐based
Initial reservoir temperature (To) 473 K EGS and CO2‐based EGS at two different simulation
Normalizing coefficient (α) 0.2 × 10 −6
Pa−1 times. It can be observed that CO2 fluid exerts high pres-
sure along the line AB with protruding pressure spikes
Biot's constant for the poroelastic 0.7 Unitless
model (αB)
because of fractures. The low viscosity value of CO2
results in high mobility and velocity of the fluid, making
Note. f (T) = function of temperature; f (P) = function of the pressure.
the fluid flow through the fractures to have high‐pressure
values. Therefore, CO2 circulates through the fractured
EGS), which diffuses gradually to the production well reservoir with higher fluid pressure than water. However,
with time. However, a significant broader pressure the pressure difference between the simulation time t =
decline is observed for CO2‐based EGS (Figure 9C) and 360 days and t = 5000 days is more evident for water‐
a sharp pressure decline for water‐based EGS (Figure 9 based EGS than CO2‐based EGS.
12 BONGOLE ET AL.

FIGURE 9 Pressure distribution for the thermo‐poroelastic model. (A) plane view for CO2‐EGS at t = 360 days; (B) plane view for water‐
EGS at t = 360 days; (C) 3D view with pressure distribution displayed as height for CO2‐EGS at t = 360 days; and (D) 3D view with pressure
distribution displayed as height for water‐EGS at t = 360 days, the circle shows pressure waves within the locality because of the presence of
fractures [Colour figure can be viewed at wileyonlinelibrary.com]

9.2 | Influence of injection rate on from high flow potential and low reservoir flow
pressure buildup impendence for CO2 than water. In this study, we consid-
ered a 2D model; therefore, the injection rate is low in
In this section, we investigate the effect of injection flow value (M = 0.5 or 1 kg/s). However, in a 3D model with
rate (M) on pressure buildup for water‐based EGS and large reservoir geometry, the injection rate might be
CO2‐based EGS. It is advised within the context of the higher in value.
EGS project to reduce the injection pressure to avoid pos- Figure 12 shows the evolution of injection pressure for
sible seismicity that may be felt at earth's surface. There- different fracture permeability at M = 0.5 kg/s. Fracture
fore, the preferred injection rate should be employed to permeability and connectivity influence the injection
overcome seismic issues in EGS projects.58 pressure; the extent of injection pressure is significant in
Figure 11 shows the injection pressure buildup for var- water‐based EGS than CO2‐based EGS for a progressive
ious injection rate for water‐EGS and CO2‐based EGS. increase in fracture permeability. As shown in Figure 12
For the same fluid injection rate (M = 1 kg/s), water‐ A, the injection pressure for water‐EGS increased from
based EGS displays higher pressure buildup (55 MPa) 26 to 79 MPa as fracture permeability (kf) decreased from
than CO2‐based EGS (26 MPa). The injection pressure 5 × 10−11 to 5 × 10−13 m2. Whereas the injection pressure
then declines rapidly with time for CO2‐based EGS to a for CO2‐based EGS (Figure 12B) increased from 19 to 28
constant value of about 16 MPa while that of water‐based MPa for the same range of fracture permeability (kf)
EGS has a slight decline to about 53 MPa. Water injection decrease.
rate at M = 0.5 kg/s has a maximum pressure buildup From the results of Figure 12, it is clear that the pres-
almost equal to CO2‐based EGS at M = 1 kg/s. The trend sure drop within the reservoir is a direct function of
in the injection pressure for the CO2‐based EGS results equivalent reservoir permeability, taking fracture/matrix
BONGOLE ET AL. 13

FIGURE 10 Pressure distribution along


the imaginary line AB [Colour figure can
be viewed at wileyonlinelibrary.com]

FIGURE 11 Injection pressure buildup


at various injection rates. [Colour figure
can be viewed at wileyonlinelibrary.com]

average permeability. The increase in fracture permeabil- extraction efficiency. However, the cost of obtaining
ity lowers the reservoir flow impendence, which is favor- higher permeable fractures is high during hydraulic frac-
able for efficient fluid flow and heat extraction process. turing as it consumes much power.
The reservoir flow impendence refers to the power
consumption of unit production rate for penetrating
the fractured reservoir and depends on the inlet 9.3 | Mechanical stress and reservoir
impendence (flow near the injection well), main reservoir damage
impendence, and outlet impendence (flow around the
production well).59 High‐fracture permeability lowers The induced matrix deformation is caused by variation in
the reservoir flow impendence resulting in reduced pore pressure and thermal stresses. Fluid withdraws in
pumping power and smooth fluid flow through the reser- regions close to the production wells and pressure
voir to the production wellbore, thus improving the heat increase near the injection wells accounts for the change
14 BONGOLE ET AL.

FIGURE 12 Injection pressure buildup for various fracture permeability at M = 0.5 kg/s. (A) Water‐based EGS and (B) CO2‐based EGS
[Colour figure can be viewed at wileyonlinelibrary.com]

in pore pressure. Thermal stresses originate from the than distortion caused by rock thermal contraction. For
temperature difference between the reservoir and the thermo‐poroelastic constitutive model, water‐based EGS
injected fluid resulting in matrix contraction. Figure 13 displayed higher deformation area ratio than CO2‐based
illustrates the deformation area for water‐based EGS EGS throughout the heat extraction period. The choice
and CO2‐based EGS at different times of the simulation of the constitutive rock model to be used in simulation
obtained by solving Coulomb failure criteria in Equa- analysis affects the extent of deformation results. The
tion 24. It can be seen that water‐based EGS has large thermo‐poroelastic constitutive model should be used in
deformation area than CO2‐based EGS at t = 360 days the reservoir formation with a high amount of the rock‐
and t = 2100 days. The injected cold water absorbed a forming minerals. The type of mineral constituting the
significant amount of heat energy from the reservoir, rock affects its strength leading to variation in rock struc-
resulting in a rapid decrease in matrix temperature. ture under the influence of stress. Thermoelastic rock
The rapid change in temperature accelerates rock ther- model can be applied to reservoir formation with high‐
mal contraction, leading to a thicker deformed region rock stiffness. The rock stiffness refers to the extent to
of the matrix. Higher heat capacity and mass of water which the rock resists deformation or changes in struc-
contribute to significant deformation extent of the reser- ture in response to an applied force.
voir in water‐based EGS. Figure 15 shows the effect of different rock modulus
To better understand the extent of the deformed on the deformation area ratio using thermo‐poroelastic
matrix‐fracture region, we define a variable “R,” which constitutive model. It can be seen that the deformation
relates the area of the deformed region to the total area area ratio increases considerably when the rock elastic
of the reservoir. The term “R” can be expressed as modulus is altered. Thus, the rock deformation is more
significant with higher rock modulus. Furthermore, the
∬A ðψ < 0ÞdA deformation extent for water‐based EGS is more than that
R¼ ; (27) of CO2‐based EGS for the same value of rock modulus.
∬A dA
The difference in heat capacity and densities between
the two fluids results in different deformation extent for
where, ψ is the Coulomb failure expression as expressed the two EGSs under the same rock modulus. Therefore,
in Equation 24; ∬AdA is the surface area integral. the role of mechanical interaction is significant in heat
Figure 14 shows the ratio “R” for water‐based EGS and recovery efficiency and is linked to the hydraulic process
CO2‐based EGS with various constitutive rock models. It in EGS.
can be seen that water‐based EGS under thermo‐elastic The deformation level in thermo‐poroelastic model is
constitutive model has a higher area deformation ratio. further explained by the mechanical stress variation
Meanwhile, CO2‐based EGS under thermo‐poroelastic along the imaginary line AB. Figure 16 shows the
model shows the least deformation area ratio because of mechanical stress variation along the line AB for water‐
the poroelastic behavior of the rock that can absorb stress based EGS and CO2‐based EGS at t = 360 days. It can
changes by pore volume variation. The extent of the be seen that the stress difference for water‐based EGS is
deformation caused by changes in pore volume is less higher during the earlier period of the simulation,
BONGOLE ET AL. 15

FIGURE 13 Critical deformed region based on Coulomb failure criteria for thermo‐poroelastic model; (A) CO2‐based EGS at t = 360 days;
(B) Water‐based EGS at t = 360 days; (C) CO2‐based EGS at t = 2100 days and (D) Water‐based EGS at t = 2100 days [Colour figure can be
viewed at wileyonlinelibrary.com]

FIGURE 14 Deformation area ratio for


water‐based EGS and CO2‐based EGS for
various constitutive models [Colour figure
can be viewed at wileyonlinelibrary.com]

strongly affected by the differential flow between the deformation area of water‐based EGS based on the
matrix and fracture systems. At t = 360 days, the mechan- mechanical difference.
ical stress difference for water‐based EGS is about 5.25 Initial development of geothermal reservoirs requires
MPa; while that of CO2‐based EGS is about 3.125 MPa. HF treatments to induce and reactivate fractures within
Therefore, we can support the results of higher the reservoirs. The induced fractures should have
16 BONGOLE ET AL.

FIGURE 15 Deformation area ratio for


water‐based EGS and CO2‐based EGS for
different rock elastic modulus for the
thermo‐poroelastic constitutive model
[Colour figure can be viewed at
wileyonlinelibrary.com]

FIGURE 16 Mechanical stress


distribution for the thermo‐poroelastic
constitutive model along line AB at t =
360 days [Colour figure can be viewed at
wileyonlinelibrary.com]

sufficient permeability to allow the smooth flow of the fracture permeability. The fracture will open with an
working fluid. Fracture permeability reflects the ability increment of permeability when the normal displacement
of reservoir conductivity and is a major influencing factor is positive (tensile stress). Otherwise, the permeability of
for heat extraction performance.60 Despite the fracturing fracture will decrease (compressive stress), and the frac-
process aiming to increase fracture permeability, the ture will close. Similar cases of fracture permeability
coupled effects of THM process during heat extraction increase in geothermal reservoirs were reported by Sun
affect the effective reservoir permeability. et al47 and Yao et al.61 The relation between effective res-
The matrix surrounding the injector tends to contract ervoir stress and fracture permeability is linked by Equa-
as the cold fluid flows out of the injection well to the res- tion 19. Figure 17 shows fracture permeability evolution
ervoir. The rock contraction results from temperature for water‐based EGS and CO2‐based EGS for various con-
variation between the reservoir and the injected fluid, stitutive rock models. It can be seen that water‐based EGS
leading to matrix displacement. The rock displacement for both constitutive rock models display higher‐
may alter the fracture aperture leading to variation in permeability evolution because of the extent of the
BONGOLE ET AL. 17

FIGURE 17 Fracture permeability


evolution for the CO2‐based EGS and
water‐based EGS [Colour figure can be
viewed at wileyonlinelibrary.com]

mechanical stresses experienced in these reservoirs. How- within the SRV; this signifies that fracture network and
ever, significant growth in fracture permeability is connectivity is vital for effective fluid flow and matrix‐
observed for the thermo‐elastic rock model because of fracture temperature exchange.
its rigidity in the rock structure, ie, no pore volume Water‐based EGS shows earlier thermal breakthrough
changes. than CO2‐based EGS as displayed from the wells produc-
Reduction in fracture permeability is observed during tion temperature in Figure 19. The breakthrough of the
the initial period of simulation for thermo‐poroelastic cold thermal water occurred as the low‐temperature
rock model caused by pore changes because of fluid with- region extends to the production well. The temperature
draw. Because of the poroelastic property of the rock, the of the produced CO2 stayed stable for about 3000 days
fracture void and matrix pores may contract during fluid then gradually begins to drop during the later period of
withdraw, reducing the effective reservoir permeability. the simulation. CO2 is quickly heated up to higher tem-
However, the poroelastic permeability decrease (fracture perature (gains heat easily) because of its less mass and
closure) is not as distinct as the fracture opening because heat capacity, thus, reducing the effect of an earlier ther-
of thermal stress. The trend in fracture permeability evo- mal breakthrough at the production well. A similar trend
lution for the constitutive rock models in this study corre- in production temperature between water‐based EGS and
lates to a single fracture aperture evolution discussed by CO2‐based EGS was observed by Wang et al11 in which
Ghassemi and Zhou.62 water had earlier thermal breakthrough than CO2. How-
ever, no rock mechanical effect or coupling was consid-
ered in his study. The similarities suggest that the
9.4 | Flow and temperature fields simulation results in this study are reliable.
CO2 has a significant advantage because of its higher
Figure 18 displays the evolution of the spatial distribution flow mobility expressed in terms of density to viscosity
of temperature for water‐based EGS and CO2‐based EGS ratio (ρf/ηf). At pressure ranges of 12 to 17 MPa and tem-
at t = 360 days and t = 2100 days. At the initial stage of perature ranging from 333 to 473 K, the (ρf/ηf) of CO2
heat mining, the low‐temperature region is seen around varies from 6 × 106 to 13.5 × 106 s/m2, and (ρf/ηf) of
the injection well, and heat exchange occurs between water varies from 2.1 × 106 to 6.2 × 106 s/m2. (ρf/ηf) of
the matrix and fluid flowing through the fractures. Heat CO2 is about 2 to 4 times that of water. Figure 19 shows
convection occurs in fractures as fluid flows through, the production temperature for water‐based EGS and
while heat conduction takes place in matrix rock. Heat CO2‐based EGS under different rock constitutive models.
conduction accounts for a gradual decrease in matrix As shown in Figure 19, the thermo‐elastic constitutive
temperature extending to the production well with time. model displayed a steep decline in production tempera-
From the temperature distribution plot (see Figure 18), ture than thermo‐poroelastic rock model. The difference
fluid flow and the low‐temperature region are confined in production temperature between the two constitutive
18 BONGOLE ET AL.

FIGURE 18 Temperature distribution plot for the thermo‐poroelastic model. (A) CO2‐based EGS at t = 360 days; (B) water‐based EGS at t
= 360 days; (C) CO2‐based EGS at t = 2100 days and (D) water‐based EGS as reservoir fluid at t = 2100 days [Colour figure can be viewed at
wileyonlinelibrary.com]

FIGURE 19 Well production


temperature for thermo‐poroelastic and
thermoelastic constitutive models [Colour
figure can be viewed at wileyonlinelibrary.
com]

models resulted from a distinct fracture permeability The effect of different rock elastic modulus for the
evolution during heat extraction, by which thermo‐ thermo‐poroelastic constitutive model was investigated.
elastic rock model has significant fracture permeability As shown in Figure 20, the temperature decline varies
evolution than thermo‐poroelastic rock model (see considerably with changes in rock elastic modulus. The
Figure 17). decline in production temperature has a direct relation
BONGOLE ET AL. 19

FIGURE 20 Well production


temperature for water‐based EGS and
CO2‐based EGS for different rock elastic
modulus using thermo‐poroelastic
constitutive model. [Colour figure can be
viewed at wileyonlinelibrary.com]

to the increase in rock modulus. The declining trend is where, Wwell is heat mining rate from the individual pro-
attributed to the extent of matrix‐fracture deformation duction well, W; Qpro is the production flow rate, kg/s;
resulting in preferential flow within the fracture system, Tpro is the fluid temperature at the production well, K;
causing an earlier thermal breakthrough. C f is the specific heat capacity at constant pressure of
the extracted fluid, J/kg/K; and T inj is the fluid injection
temperature, K.
9.5 | Heat mining efficiency for CO2‐based
Figure 21 shows the production flow rate and heat
EGS and water‐based EGS
mining rate for CO2‐based EGS and water‐based EGS. It
can be seen that the heat mining rate and flow rate for
The heat mining rate describes the heat extraction effi-
CO2‐based EGS are higher than that of water‐based
ciency for CO2‐based EGS and water‐based EGS and
EGS. For CO2‐based EGS, the higher flow rate at the pro-
can be described as11,63
duction well results from higher mobility of the CO2
  fluid. The delayed thermal breakthrough for the CO2‐
W well ¼ Qpro T pro × Cf − T inj × C f ; (28) based EGS compared with water‐based EGS (see

FIGURE 21 Production flow rate and


heat extraction rate for CO2‐based EGS
and water‐based EGS based on the
thermo‐poroelastic model [Colour figure
can be viewed at wileyonlinelibrary.com]
20 BONGOLE ET AL.

Figure 19) improves the heat mining rate for the CO2‐ decline in heat mining rate for CO2‐based EGS at t =
based EGS. Therefore, the heat mining rate for the EGS 3500 days is because of the drop in temperature of the
depends on the flow rate and the temperature of the extracted CO2 (the beginning of thermal breakthrough).
extracted fluid at the production wells. As shown in The higher heat extraction efficiency of the CO2‐based
Figure 21, the heat mining rate for CO2‐based EGS first EGS have also been reported in various studies.7-9,11
increases and later fall gradually at t = 3500 days. The

9.6 | Effect of in stress


TABLE 5 External stress variation parametric and fracture per-
meability evolution values
Major higher thermal energy reservoirs are located deep
S/N σx, MPa σy, MPa Stress Ratio, σy/σx in the subsurface, where in‐situ stresses are significant.
CASE_I 5 10 2 In‐situ stresses influence fracture propagation and con-
nectivity leading to variation in fluid flow and reservoir
CASE_II 5 15 3
productivity. In this section, we investigate the influence
CASE_III 5 16.5 3.3
of varying in‐situ stresses for CO2‐based EGS and water‐

FIGURE 22 Fracture permeability


evolution for different in‐situ stress
direction [Colour figure can be viewed at
wileyonlinelibrary.com]

FIGURE 23 Wells production


temperature for different in‐situ stress
direction [Colour figure can be viewed at
wileyonlinelibrary.com]
BONGOLE ET AL. 21

based EGS systems. Three cases are considered with their temperature than thermo‐poroelastic rock model for
stress values displayed in Table 5. Case_I, Case_II, and both CO2‐based EGS and water‐based EGS.
Case_III have increasing stress ratio, respectively. Stress 3) The mechanical stresses in water‐based EGS are rela-
value of “y” to “x” direction gives the stress ratio (σy/σx). tively high than CO2‐based EGS; the stress difference
The “x” and “y” directions within the model are displayed is attributed to the fluid flow behavior and fluid prop-
in Figure 6. erties. Furthermore, water‐EGS has higher injection
As shown in Figure 22, the equivalent fracture perme- pressure buildup than CO2‐based EGS for the same
ability increased with stress ratio for individual water‐ injection rate.
based EGS and CO2‐based EGS, ie, Case_III is more 4) In‐situ stresses have a significant influence in both
significant than Case_II and Case_II higher than Case_I. CO2‐based EGS and water‐based EGS. The increase
However, the extent of the growth is more pronounced in in stress ratio improves the fracture permeability
water‐based EGS than CO2‐based EGS. Further increase but leads to a steep decline in the production temper-
in stress ratio enhances the fracture permeability. There- ature. The stress ratio enhances the fracture perme-
fore, the equivalent reservoir permeability is sensitive ability leading to channelized flow resulting in the
external stress and has a positive correlation to stress earlier thermal breakthrough at the production well.
ratio.
Figure 23 shows the wells production temperature for
CO2‐based EGS and water‐based EGS for different stress
ACKNOWLEDGEMENTS
ratio. It can be seen that the increase in stress ratio has
a steep decline trend in production temperature leading This study was jointly supported by the National Natural
to an earlier thermal breakthrough. The production Science Foundation of China (Grant NO.51774317), the
temperature decline is higher for large stress ratio. How- Fundamental Research Funds for the Central Universities
ever, the production temperature decline rate is signifi- (Grant No.18CX02100A), and National Science and Tech-
cant for CO2‐based EGS (Case_1 to Case_II) than water‐ nology Major Project (Grant NO.2016ZX05011004‐004).
based EGS.

CONFLICT OF INTEREST
1 0 | CONCLUSIONS
The authors declare that they have no conflict of interest.
This study investigates the multifracture response to
water or supercritical carbon dioxide injection in a NOMENCLATURE
hydraulically fractured geothermal reservoir. The proper-
ties of water are quite different from those of carbon diox- ϕ Matrix porosity
ide. Therefore, the fluid flow, heat flux, and mechanistic χf Fluid compressibility (Pa‐1)
response of the fractures differ significantly. The specific
p Hydraulic pressure (Pa)
key points from this study are summarized below: dfr Fracture aperture
km Matrix permeability (m2)
1) The reservoir deformation area for water‐based EGS αB Biot's constant
is higher than CO2‐based EGS; the higher heat capac- R Area deformed ratio
ity and mass of water contributes to wider deforma- ∇τ Gradient operator to fracture plane
tion area in water‐based EGS. A thermoelastic u Displacement (m)
constitutive model displayed higher reservoir defor- E Elastic modulus (Pa)
mation than the thermo‐poroelastic model. Further- Ts Matrix temperature (K)
more, the extent of reservoir deformation is σ Total stress (Pa)
accelerated by the strength of the rock (rock elastic τ Fracture stiffness (N/m)
modulus). Cs Heat capacity of solid matrix (J/kg/K)
2) Compared with CO2‐based EGS, water‐based EGS Cf Heat capacity of fluid (J/kg/K)
displayed earlier thermal breakthrough at the pro- Wfr Fracture to matrix heat exchange (W/m)
duction wells. The injected CO2 gains heat quickly Ws Matrix external heat source (W/m3)
because of its low heat capacity; its temperature rises ko Initial permeability at zero stress (m2)
fast thus reducing the effect of an earlier thermal σ ′n Effective normal stress on the fracture plane (Pa)
breakthrough. Furthermore, the thermoelastic con- σ1 First principle stress (Pa)
stitutive rock model has a rapid decline in production C1 Calibration parameter 1
22 BONGOLE ET AL.

on Geothermal Resevoir Engineering. Stanford (California):


ψ Failure coefficient Stanford University January 24‐26; 2000.
C2 Calibration parameter 2
7. Atrens AD, Gurgenci H, Rudolph V. Electricity generation using
T inj Temperature of the injected fluid (K) a carbon‐dioxide thermosiphon. Geothermics. 2010;39(2):
ϕ Fracture porosity 161‐169.
η Dynamic fluid viscosity (Pa·s) 8. Atrens AD, Gurgenci H, Rudolph V. CO2 Thermosiphon for
ρf Density of fluid (kg/m3) competitive geothermal power generation. Energy Fuel.
S Specific storage term (Pa‐1) 2009;23(1):553‐557.
vf Fluid velocity in fractures (m/s) 9. Pruess K. On production behavior of enhanced geothermal sys-
Qm The source‐sink term of the seepage process tems with CO2 as working fluid. Energ Conver Manage.
ε The volumetric strain of matrix rock 2008;49(6):1446‐1454.
kfr Fracture permeability (m2) 10. Pruess K. Enhanced geothermal systems (EGS) using CO2 as
σij Stress tensor (Pa) working fluid—a novel approach for generating renewable
energy with simultaneous sequestration of carbon. Geothermics.
Fi Body force per unit volume in the (Pa)
2006;35(4):351‐367.
v Poisson's ratio
11. Wang CL, Cheng WL, Nian YL, Yang L, Han BB, Liu MH. Sim-
αT Thermal expansion coefficient (K‐1)
ulation of heat extraction from CO2‐based enhanced geothermal
σ' Effective stress (Pa) systems considering CO2sequestration. Energy. 2018;142:
Tf Fluid temperature in fracture network (K) 157‐167.
ρs Density of matrix (kg/m3) 12. Sun Z, Yao J, Huang X, et al. CO2 injection for heat extraction
λs Heat conductivity of matrix (W/m/K) and carbon sequestration in a geothermal site: Huizhou Sag,
λf Heat conductivity of the fluid (W/m/K) the Pearl River Mouth Basin. Geothermics. 2016;64:331‐343.
σ3 Third principle stress (Pa) 13. Xin Y, Sun Z, Zhuang L, et al. Numerical simulation of fluid
h Convection coefficient (W/m2/K) flow and heat transfer in EGS with thermal‐hydraulic‐
α Normalizing coefficient mechanical coupling method based on a rough fracture model.
θ Friction angle (rad) Energy Procedia. 2019;158:6038‐6045.
So Coulomb Cohesion force (Pa) 14. Golsanami N, Sun J, Liu Y, et al. Distinguishing fractures from
matrix pores based on the practical application of rock physics
Qpro Production flow rate (kg/s)
inversion and NMR data: a case study from an unconventional
σ2 Second principle stress (Pa)
coal reservoir in China. J Nat Gas Sci Eng. 2019;65:145‐167.
Tpro Temperature of the extracted fluid (K)
15. Yan W, Sun J, Sun Y, Golsanami N. A robust NMR method to
measure porosity of low porosity rocks. Microporous Mesoporous
Mater. 2018;269:113‐117.
ORCID 16. Luo F, Xu RN, Jiang PX. Numerical investigation of fluid flow
and heat transfer in a doublet enhanced geothermal system with
Kelvin Bongole https://orcid.org/0000-0001-5399-553X CO2as the working fluid (CO2‐EGS). Energy. 2014;64:307‐322.
17. Borgia A, Oldenburg CM, Zhang R, et al. Simulations of CO2
injection into fractures and faults for improving their geophysi-
R EF E RE N C E S
cal characterization at EGS sites. Geothermics. 2017;69:189‐201.
1. Jiang P, Li X, Xu R, Zhang F. Heat extraction of novel under- 18. Yan W, Sun J, Golsanami N, et al. Evaluation of wettabilities
ground well pattern systems for geothermal energy and pores in tight oil reservoirs by a new experimental design.
exploitation. Renew Energy. 2016;90:83‐94. Fuel. 2019;252:272‐280.
2. Mehmood A, Yao J, Yan Fun D, Zafar A. Geothermal energy 19. Shi Y, Song X, Shen Z, et al. Numerical investigation on heat
potential of Pakistan on the basis of abandoned oil and gas extraction performance of a CO2 enhanced geothermal system
wells. J Pet Environ Biotechnol. 2017;08(03):1‐6. with multilateral wells. Energy. 2018;163:38‐51.
3. Mehmood A, Yao J, Fan DY, Bongole K. Utilization of 20. Shi Y, Song X, Li J, Wang G, Zheng R, YuLong F. Numerical
abandoned oil and gas wells for geothermal energy production investigation on heat extraction performance of a multilateral‐
in Pakistan. Tunisia: Springer International Publishing; 2019: well enhanced geothermal system with a discrete fracture net-
181‐183. work. Fuel. 2019;244:207‐226.
4. Kelkar S, Woldegabriel G, Rehfeldt K. Lessons learned from the 21. Wang T, Sun Z, Zhang K, Jiang C, Xin Y, Mao Q. Investigation
pioneering hot dry rock project at Fenton. Geothermics. on heat extraction performance of fractured geothermal reser-
2016;63:5‐14. voir using coupled thermal‐hydraulic‐mechanical model based
5. Breede K, Dzebisashvili K, Liu X, Falcone G. A systematic on equivalent continuum method. Energies. 2019;12(1):127.
review of enhanced (or engineered) geothermal systems: past, 22. Guo T, Gong F, Wang X, Lin Q, Qu Z, Zhang W. Performance of
present and future. Geotherm Energy. 2013;1(1):4. enhanced geothermal system (EGS) in fractured geothermal res-
6. Brown D. A hot dry rock geothermal energy concept utilizing ervoirs with CO2 as working fluid. Appl Therm Eng.
supercritical CO2 instead of water. In: Twenty‐Fifth Workshop 2019;152:215‐230.
BONGOLE ET AL. 23

23. Ghassemi A, Zhang Q. Porothermoelastic analysis of the 39. Saul A, Wagner W. A fundamental equation for water covering
response of a stationary crack using the displacement disconti- the range from the melting line to 1273 K at pressures up to 25
nuity method. J Eng Mech. 2006;132(1):26‐33. 000 MPa. J Phys Chem Ref Data Monogr. 1989;18(4):1537‐1564.
24. Ghassemi A, Nygren A, Cheng A. Effects of heat extraction on 40. Wagner W, Pruss A. Revised Release on the {IAPWS} Formula-
fracture aperture: a poro‐thermoelastic analysis. Geothermics. tion 1995 for the Thermodynamic Properties of Ordinary
2008;37(5):525‐539. Water Substance for General and Scientific Use. J Phys Chem
25. Zhou X, Ghassemi A. Three‐dimensional poroelastic analysis of Ref Data Monogr. 2002;31(2):387‐535.
a pressurized natural fracture. Int J Rock Mech Min Sci. 41. Wu C. Hydraulics. Beijing (in Chinese): Higher Education Press;
2011;48(4):527‐534. 1982.
26. Bahrami D, Danko G, Fu P, et al. Poroelastic and self‐propped 42. Biot MA. Mechanics of deformation and acoustic propagation in
single fracture THM models for EGS studies. Proc 40th Stanford porous media. J Appl Phys. 1962;33(4):1482‐1498.
Geotherm Work. 2015:1‐20. 43. Xu C, Dowd PA, Tian ZF. A simplified coupled hydro‐thermal
27. Slatlem Vik H, Salimzadeh S, Nick HM. Heat recovery from model for enhanced geothermal systems. Appl Energy.
multiple‐fracture enhanced geothermal systems: the effect of 2015;140:135‐145.
thermoelastic fracture interactions. Renew Energy. 2018; 44. Chen B, Song EX, Cheng X. A numerical method for discrete
121:606‐622. fracture network model for flow and heat transfer in two‐
28. Zhao Y, Feng Z, Feng Z, Yang D, Liang W. THM (Thermo‐ dimensional fractured rocks. Chinese J Rock Mech Eng.
hydro‐mechanical) coupled mathematical model of fractured 2014;33(1):43‐51.
media and numerical simulation of a 3D enhanced geothermal 45. Rice JR. Fault stress states, pore pressure distributions, and the
system at 573 K and buried depth 6000–7000 M. Energy. weakness of the San Andreas fault. Int Geophys.
2015;82:193‐205. 1992;51(C):475‐503.
29. Kumar D, Ghassemi A. Hydraulic stimulation of multiple hori- 46. Yu W, Bao‐lin L, Hai‐yan Z, Chuan‐liang Y, Zhi‐jun L, Zhi‐qiao
zontal wells for egs reservoir creation. Trans ‐ Geothermal Res W. Thermophysical and mechanical properties of granite and its
Counc. 2016;40:373‐382. effects on borehole stability in high temperature and three‐
30. Cheng W, Jin Y, Chen M. Experimental study of step‐ dimensional stress. Sci World J. 2014;2014:1‐11.
displacement hydraulic fracturing on naturally fractured shale 47. Sun Z, Zhang X, Xu Y, et al. Numerical simulation of the heat
outcrops. J Geophys Eng. 2015;12(4):714‐723. extraction in EGS with thermal‐hydraulic‐mechanical coupling
31. Gu H, Weng X, Lund JB, Mack MG, Ganguly U, Suarez‐Rivera method based on discrete fractures model. Energy. 2017;
R. Hydraulic fracture crossing natural fracture at nonorthogonal 120:20‐33.
angles: a criterion and its validation. SPE Prod Oper. 48. Bai B. One‐dimensional thermal consolidation characteristics of
2012;27(01):20‐26. geotechnical media under non‐isothermal condition. Eng Mech.
32. Chen P, Rahman MM, Sarma HK. Interaction between hydrau- 2005;22(5):186‐191.
lic fracture and natural fracture—a new prediction model by 49. Cheng AHD, Ghassemi A, Detournay E. Integral equation solu-
adaptive neuro‐fuzzy inference system (ANFIS). In: Abu Dhabi tion of heat extraction from a fracture in hot dry rock. Int J
International Petroleum Exhibition and Conference. Society of Numer Anal Methods Geomech. 2001;25(13):1327‐1338.
Petroleum Engineers; 2014:1‐16. 50. COMSOL Co. Ltd., Stockholm S. COMSOL multiphysics. User's
33. Frash L, Gutierrez M, Hampto J. Scale model simulation of Guide Version 5.3a. 2017.
hydraulic fracturing for EGS reservoir creation using a heated 51. Campbell S, Fairchild NR Jr, Arnold DL. Liquid CO2 and sand
true‐triaxial apparatus. In: Effective and Sustainable Hydraulic stimulations in the lewis shale, San Juan Basin, New Mexico: a
Fracturing. InTech; 2013:959‐977. case study. In: Proceedings of SPE Rocky Mountain
34. Zhang D, Dai Y, Ma X, et al. An analysis for the influences of Regional/Low‐Permeability Reservoirs Symposium and Exhibi-
fracture network system on multi‐stage fractured horizontal tion. Denver, Colorado,USA: Society of Petroleum Engineers;
well productivity in shale gas reservoirs. Energies. 2018; 2000.
11(2):414. 52. Li X, Li G, Yu W, et al. Thermodynamic Behavior of Liquid‐
35. Span R, Wagner W. A new equation of state for carbon dioxide Supercritical CO2 Fracturing in Shale. In: Proceedings of
covering the fluid region from the triple‐point temperature to the 5th Unconventional Resources Technology Conference.
1100 K at pressures up to 800 MPa. J Phys Chem Ref Data Tulsa, OK, USA: American Association of Petroleum Geologists;
Monogr. 1996;25(6):1509‐1596. 2017.
36. Heidaryan E, Hatami T, Rahimi M, Moghadasi J. Viscosity of 53. Vishal V, Jain N, Singh TN. Three dimensional modelling of
pure carbon dioxide at supercritical region: measurement and propagation of hydraulic fractures in shale at different injection
correlation approach. J Supercrit Fluids. 2011;56(2):144‐151. pressures. Sustain Environ Res. 2015;25(4):217‐225.
37. Jarrahian A, Heidaryan E. A novel correlation approach to esti- 54. Suarez‐rivera R, Begnaud BJ, Martin JW. Numerical analysis of
mate thermal conductivity of pure carbon dioxide in the open‐hole multilateral completions minimizes the risk of costly
supercritical region. J Supercrit Fluids. 2012;64:39‐45. junction. In: Rio Oil & Gas Expo and Conference. Rio de Janeiro;
38. NIST. National Institute of Standards and Technology (NIST) 2004.
Chemistry WebBook. National Institute of Standards and 55. Lajtai EZ, Gadi AM. Friction on a granite to granite interface.
Technology. Rock Mech Rock Eng. 1989;22(1):25‐49.
24 BONGOLE ET AL.

56. Cui G, Zhang L, Ren B, Enechukwu C, Liu Y, Ren S. Geother- coupling method based on discrete fractures model.
mal exploitation from depleted high temperature gas reservoirs Geothermics. 2018;74(2017):19‐34.
via recycling supercritical CO2: heat mining rate and salt precip- 62. Ghassemi A, Zhou X. A three‐dimensional thermo‐poroelastic
itation effects. Appl Energy. 2016;183:837‐852. model for fracture response to injection/extraction in enhanced
57. Wang Y, Li T, Chen Y, Ma G. Numerical analysis of heat mining geothermal systems. Geothermics. 2011;40(1):39‐49.
and geological carbon sequestration in supercritical CO2 circu- 63. Cheng W‐L, Wang C‐L, Nian Y‐L, Han B‐B, Liu J. Analysis of
lating enhanced geothermal systems inlayed with complex influencing factors of heat extraction from enhanced geothermal
discrete fracture networks. Energy. 2019;173:92‐108. systems considering water losses. Energy. 2016;115:274‐288.
58. Majer EL, Peterson JE. The impact of injection on seismicity at
The Geysers, California Geothermal Field. Int J Rock Mech
Min Sci. 2007;44(8):1079‐1090.
59. Murphy H, Brown D, Jung R, Matsunaga I, Parker R. Hydraulics How to cite this article: Bongole K, Sun Z, Yao J,
and well testing of engineered geothermal reservoirs. et al. Multifracture response to supercritical CO2‐
Geothermics. 1999;28(4‐5):491‐506. EGS and water‐EGS based on thermo‐hydro‐
60. Watanabe K, Takahashi H. Parametric study of the energy mechanical coupling method. Int J Energy Res.
extraction from hot dry rock based on fractal fracture network 2019;1–24. https://doi.org/10.1002/er.4743
model. Geothermics. 1995;24(2):223‐236.
61. Yao J, Zhang X, Sun Z, et al. Numerical simulation of the heat
extraction in 3D‐EGS with thermal‐hydraulic‐mechanical

View publication stats

You might also like