Available online at www.sciencedirect.

com

ScienceDirect
Energy Procedia 75 (2015) 946 955

The 7th International Conference on Applied

Energy ICAE2015

On wellbore heat transfer and fluid flow in the doublet of

Enhanced Geothermal System
Xiaoxue Huanga, Jialing Zhua, Jun Lia*
a
Tianjin University, 92 Weijin Road, Nankai District, Tianjin 300072, China

Abstract

Enhanced geothermal system which utilizes geothermal energy beneath the ground surface with a depth of several
thousand meters has aroused intense interest recently. Wellbores in Enhanced Geothermal System extends several
kilometers from the ground surface, providing large heat transfer areas between the flowing fluid and the surrounding
formation. Various techniques and approximations for predicting flowing process in the wellbore have been presented
in literature. Most studies assume steady one-dimensional flowing without considering heat transfer with the
surroundings. In this paper, unsteady flowing within a vertical injection and a production well, together with wellbore
heat transfer between the fluid and the surrounding formation was modeled. An analytical solution under certain
assumptions is presented using the Laplace transform technique. The time-evolving results show that almost the
entire pipe is under hydrodynamically fully developed condition. Transient effect is significant in the initial phase
after injection and production. Approximate steady condition could be reached within the 30 years of operation under
fixed injection and production conditions.

2015
2015TheThe Authors.
Authors. Published
Published by Elsevier
by Elsevier Ltd.
Ltd. This is an open access article under the CC BY-NC-ND license
Selection and/or peer-review under responsibility of ICAE
(http://creativecommons.org/licenses/by-nc-nd/4.0/).
Peer-review under responsibility of Applied Energy Innovation Institute

* Corresponding author. Tel.: +86-022-27401830; fax: +86-022-27401830.

E-mail address: xiaoxue@tju.edu.cn.

1876-6102 2015 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license
(http://creativecommons.org/licenses/by-nc-nd/4.0/).
Peer-review under responsibility of Applied Energy Innovation Institute
doi:10.1016/j.egypro.2015.07.277
 Xiaoxue Huang et al. / Energy Procedia 75 (2015) 946 955 947

keywords: Enhanced Geothermal System; Wellbore heat transfer; Laplace transform; transient heat transfer; numerical modelling.

1. Introduction

Being exploited for direct use, electricity generation and ground source heat pump, geothermal energy
plays an important role in alleviating the pressure of clean energy supply. To promote development of the
extensive geothermal energy resources, advanced technologies such as exploitation of hot dry rock,
magma bodies and geopressured reservoirs have been attracted. The concept of Enhanced Geothermal
System (EGS) involves mining heat from hot dry rocks via pumping cold fluid to the targeted formation
through the injection well, and bringing hot water from the production well, then utilizing the hot water to
generate electricity or for direct use. Compared to conventional geothermal resources, EGS using hot dry
rock is considered feasible for widespread use with fewer environmental issues [1].

At its most basic level, management of subsurface geothermal resources requires a doublet, the
injection well and the production well, and the target reservoir. In particular, Enhanced Geothermal
System (EGS) for geothermal energy production requires long length wellbores for reaching the deep hot
dry rock. Study of the behavior of fluid through geothermal wells is one of the most fundamental aspects
to understand and predict the subsurface processes, which is required to answer critical questions as to the
design and performance of fluid production, injection, and heat transfer.

Among the wellbore simulations by Hadgu [2], Murray [3], Pruess and Zhang [4], Pan and Oldenburg
[5], generally 1D modeling is carried out for its saving of computational work and ease of being adapted
to geometry of inclined wellbores. The 1D representation, however, implies that the variation of the
temperature is along its vertical axis, and no temperature variation exists in the radial direction. The latter
condition is reasonably valid because of the slenderness of the wellbore, where the temperature variation
in the radial direction is negligible compared to that in the axial direction, whereas heat transfer takes
place along the radial direction. Typically, heat transfer of the wellbore fluid from the inside out consists
of internal convection of flowing fluid, conduction through casing, conduction through grout and
conduction in surrounding rock formation, and thermal resistances of the convection and conduction in
the surrounding dominate the process. Using lumped parameters in 1D assumption is reasonable when the
thermal resistance of convection is negligible. In the studies of Aunzo [6], Fard [7], Jiang [8], heat
exchange between the fluid and the surrounding was often neglected to simplify the mathematical model,
whereas this work estimates the heat transfer process by solving the governing equations of the 3D model
of both the wellbores and the surrounding formations altogether. Sanaz [9] employed correlations for Nu
in horizontal fully developed pipe flow to take account of the convection heat transfer, while the existing
correlations are for horizontal flowing with boundary condition of constant temperature or constant heat
flux. When using computation fluid dynamics (CFD) modelling for 3D model, input of Nu is not
required, and it can be derived from the radial temperature distribution of the results. As experiments are
costly in geothermal power production processes, three-dimensional CFD modelling can be used to
produce reliable results which are able to provide a preliminary understanding.

2. Model description
948 Xiaoxue Huang et al. / Energy Procedia 75 (2015) 946 955

2.1 Physical model

The modelled domain is 4000m4000m6000m. The physical model together with the mesh system is
shown in Figure 2. The fractured reservoir is 500m500m500m located at 4000m from the ground
surface and centered in the x-y plane. Wellbore distance is 400m, and the injection well and the
production well with 0.2m diameters are located symmetrically with respect to the y-z plane.

Rock formations enclosing the reservoir are set to be impermeable, thus only conduction exits within it.
The fractured reservoir is a porous zone with porosity of 0.047. Both injection well and the production
well are fluid regions. Mass-inlet and pressure-outlet boundary conditions are applied to the inlet of the
injection well and outlet of the production well, respectively. Initial temperature increases with depth with
a constant gradient of 3 K/100m . 300K is prescribed at the ground surface, and temperature at the top and
the bottom x-y plane are fixed during the simulation. Lateral boundaries are set to be adiabatic
considering the large scale of the simulated domain.

Meshes were designed to get sufficiently fine resolution in wellbores and the reservoir. Mesh interval
size along z direction) is 1m for wellbores, and in the reservoir, it is 10m firstly and increases downwards
at a growth rate of 1.3. The discretized model totally has around 3.57million numerical elements.

Fig. 1. Schematic of half of the simulated domain with mesh systems

2.2 Mathematical model

Major assumptions made for the governing equations are: 1) the reservoir is fully saturated; 2) the
fractured reservoir is hydraulically equivalent to a uniform porous domain with constant porous and
permeability; 3) under high pressure in the reservoir, fluid is considered to be single-phase. Governing
equations are expressed in the generic form of Equation (1) [10]. Thermophysical properties of water,
including density (1.225 kg/m3), specific heat (1006.43 J/kgK-1), viscosity (1.789410-5 kg/ms-1) and
conductivity (0.0242 w/mK-1), are set to be constant in the simulation.
w ( UI ) w ( U uI ) w ( U vI ) w ( U wI ) w wI w wI w wI
   (* )  *  *  S (1)
wt wx wy wz wx wx wy wy wz wz
 Xiaoxue Huang et al. / Energy Procedia 75 (2015) 946 955 949

Details of each governing equation are shown in Table 1. Standard k  H turbulence model is used for
fluid flow in the wellbore.

Table 1. Terms in the governing equations

I * S

1 0 0

u (K  KT ) wp w wu w wv w ww
  ((K  KT ) )  ((K  KT ) )  ((K  KT ) )
wx wx wx wy wx wz wx

v (K  KT ) wp w wu w wv w ww
  ((K  KT ) )  ((K  KT ) )  ((K  KT ) )
wy wx wy wy wy wz wy

w (K  KT ) wp w wu w wv w ww
  ((K  KT ) )  ((K  KT ) )  ((K  KT ) )  U g
wz wx wz wy wz wz wz

T K / Pr  KT / 0.85 0

k (K  KT / 1.0) wu 2 wv ww wu wv wu ww 2 wv ww 2
KT (2( )  2( ) 2  2( ) 2  (  ) 2  (  ) (  ) )  UH
wx wy wz wy wx wz wx wz wy

H (K  KT / 1.3) H wu 2 wv ww wu wv wu ww 2 wv ww 2
(1.44KT (2( )  2( ) 2  2( ) 2  (  ) 2  (  ) (  ) )  1.92 UH )
k wx wy wz wy wx wz wx wz wy

The fractured reservoir is treated as a uniform porous medium under local thermal equilibrium. Inertial
loss in the porous zone is ignored, and only viscous resistance is considered. Ignoring convective
acceleration and diffusion, the porous medium model then reduces to Darcys Law. In x direction for
example, the governing equation is
3
1
'p x a K v j 'n x
j 1 xj (2)

Assuming equivalent permeability of the fractured reservoir is 0.2510-12 m2 in each direction, the derived
viscous resistance coefficient is 4109 in x direction, and 8109 in y and z direction.

3. Numerical method

The well-known SIMPLE algorithm is used to address the pressure-velocity coupling. First order
upwind differencing scheme is used for discretization of the spatial-derivative terms and a fully implicit
scheme for discretization of the transient terms. The commercial CFD flow solver, Fluent, which is based
on the finite volume approximation, is employed to carry out the simulation.

4. Results and discussion

950 Xiaoxue Huang et al. / Energy Procedia 75 (2015) 946 955

The reported results show that the total pressure gradient along the axial direction (z direction) keeps
constant except interfaces between the wellbores and the reservoir. Considering the total pressure is
mainly dominated by the gravity term, the static pressure gradient (thermodynamic pressure, gravity term
not included) along the axial direction keeps approximately 0.

According to the correlation in Ref. [11], hydrodynamic entry length x can be obtained as 3.3m
derived from ( x / d ) 4.4 Re1/6 . Slices at depths of 0m, 10m, 100m, 1000m, 2000m, 3000m, 3900m,
3990m and 4000m from the ground surface are chosen for simulation results analysis. Z-velocity profiles
keep similar when the distance from inlet of both wells is beyond 10m. Facet average z-velocity keeps
constant, and x- and y- velocity is approximately 0. Z-velocity distribution of selected slices are shown in
Figure 2. Accordingly, it is reasonable assuming fully developed in the whole well if no particular
attention needed for the ends.

(a) (b)
Fig. 2. Z-velocity profiles in the injection well (a) and the production well (b)
 Xiaoxue Huang et al. / Energy Procedia 75 (2015) 946 955 951

The reported Nu is listed in Table 2. The negative sign represents that heat is transferred from fluid to
the surrounding. Thermal-fully developed condition cannot be reached because the boundary conditions
of internal convection are not constant. Due to the temperature gradient in the formations, heat is lost from
the fluid initially and then changes direction as fluid absorbs heat when deeper than 1000m from the inlet.
In the production well, heat is always transferred from the fluid to the surrounding.

Static pressures at different depths are also shown in Table 2. The hydraulic total pressure (sum of the
static pressure and gravity term) changes significantly with depth due to the gravity, thus temperature will
also change a lot due to the significant variation of pressure. Compared to heat transfer, temperature
change resulted from this Joule-Thomson is also obvious. In this simulation, because of the assumption of
fixed properties, enthalpy is the single value function of temperature. In the next step, equation of state of
water will be used to enable calculation of the real properties, and the effect of Joule-Thomson effect can
be fully understood.

Table 2. Static pressures at different depths in wellbores

Depth/m Injection well Production well

Nu Static pressure/ MPa Nu Static pressure/ MPa
0 -5.73 2.25 -7.09 0.4
1000 -0.48 2.03 -5.06 0.62
2000 4.46 1.79 -3.38 0.83
3000 9.31 1.58 -1.68 1.05
4000 10 1.37 -0.01 1.17

Constant themophysical properties which are mainly influenced by temperature are set to be constant
in this simulation. Take the lowest (the injection well inlet of 333 K and 2MPa) and highest temperature
(the production well inlet of 420 K and 40MPa) for comparison, variation is 4.4% for density, 8.2% for
thermal conductivity and 58% for viscosity, thus use of constant properties mainly influences hydraulic
results through viscosity value and tends to underestimate the flowing.

Rock formation temperature distribution of selected depths is shown in Figure 4. Horizontal axis shows
the distance from the wellbore and vertical axis shows the temperature value. It can also be inferred that
heat is transferred from the fluid to the surrounding in the first 1000m in the injection well, then heat
transfer direction changes during the rest 3000m. In the production well, heat transfer takes place from the
fluid to the surrounding, leading to a rise in temperature of the surrounding. The area influenced by heat
transfer is within 10m as shown in figure 3. The time evolving results show that after 720h of operation,
steady state of heat transfer is approximately reached as temperature varies slightly in the horizontal
direction.

(a) (b)
952 Xiaoxue Huang et al. / Energy Procedia 75 (2015) 946 955

Fig.3. Temperature distribution at depths of 0m, 1000m, 2000m, 3000m, 4000m in the rock formations and operation times of
24h, 240h, 720h. (a): surrounding formations of the injection well; (b): surrounding formations of the production well.

According to the numerical results, reasonable assumptions can be employed for an analytical solution.
x 1D fluid flow in the wellbore, with lumped fluid temperature vertically distributed.
x Conduction terms and radial convection term of energy equation of fluid are eliminated.
x Vertical heat conduction of solid formation is neglected.
Thermophysical properties of fluid are assumed constant, and that of surrounding formations are
approximated homogeneous in the vertical direction.
Therefore, the energy equations of fluid in the wellbore and the formation enclosing it can be written as:

wt f wt
U f c pf v f q (3)
wW wz
1 w w t wt
(O r s ) U s c ps ( s ) (4)
r wr wr wr
Heat flux across the wellbore surface is
q h (ts  t f ) (5)
r rw

Initial conditions are

t f ( z, t 0) ts ( r, z, t 0) ti ( z ) (6)
Boundary conditions are
t f ( z 0, t ) tinj (7)
wts ( r, z, t )
O q (8)
wr
lim ts ( r , z , t ) ti ( z ) (9)
r of

Derivation of the solution using Laplace transform is shown in the Appendix. It shows that temperature
variation of fluid and the formation can be obtained analytically with known heat transfer coefficient h .

5. Conclusion

Computational Fluid Dynamics (CFD) modelling of the subsurface heat exchange system of EGS,
including the wellbores, the fractured reservoir and the surrounding rock formation, was performed. Heat
transfer process of fluid flowing in the wellbores with the surrounding was analyzed. The results show
that flowing within almost the whole wellbore is in the hydrodynamic fully developed region. Thermal-
fully developed, which means Nusselt (Nu) number keeps constant cannot be reached because the
boundary conditions of the flowing fluid is not constant, wall of the temperature varies due to the
preexisting geo-temperature gradient. In the injection well, heat loss takes place initially, while at depths
larger than 1000m, the formation temperature is higher than the fluid, thus the fluid gains heat from the
surrounding. In the production well, heat is always transferred from the fluid to the surrounding. Joule-
Thomson effect is significant mainly resulting from the gravity term in total pressure. Vertical conduction
can be neglected in the surrounding formation as horizontal temperature gradient is much more significant.
The transient effects of heat transfer diminishes after 720h of operation, thus in numerical simulation,
steady heat transfer process could be assumed with a longer operation period. Under certain assumptions,
an analytical solution can be arrived using the Laplace transform technique.
 Xiaoxue Huang et al. / Energy Procedia 75 (2015) 946 955 953

Nomenclature
1/ a x j Ts
viscous resistance coefficient dimensionless solid temperature in Laplace space
x x coordinate, m 'p x
y pressure gradient in x direction, Pa/m
y coordinate, m K
z fluid viscosity, Pas
z coordinate, m
'n x thickness of the medium in x direction, m
z* dimensionless z coordinate
r radial coordinate, m KT eddy viscosity, Pas
r* dimensionless radial coordinate k turbulent energy
rw H turbulent dissipation rate
radius of the wellbore, m
u velocity in x direction, m/s O thermal conductivity, W/mK
v velocity in y direction, m/s c pf
specific heat of fluid, J/kg
vj c ps
velocity components in x, y, and z
specific heat of solid, J/kg
directions, m/s
w Nu Nusselt number
velocity in z direction, m/s
K0 zero-order modified Bessel function of second
T temperature,
ti kind
initial temperature, K1
tf first-order modified Bessel function of second
fluid temperature, kind
ts p fluid pressure, Pa
solid temperature,
*
q heat flux between fluid in the wellbore and the
t dimensionless temperature,
surrounding formation, W/m2
tic derivative of initial temperature, q* defined in (A-7)
/m
tinj q* defined in (A-15)
injection temperature,
U s Laplace transform variable
density, m3/kg
S source term
Uf
fluid density, m3/kg T temperature,
Us solid density, m3/kg Z defined in (A-21)
* diffusion coefficient g acceleration of gravity, m/s2
W time, s Re Reynolds number
d diameter of the wellbore, m Nu Nusselt number
C coefficient of the solution Subscripts
D depth of the wellbore f fluid
h heat transfer coefficient, W/m2 s solid
Pr Prandtl number inj
injection
Ts dimensionless solid temperature Superscripts
Tf * dimensionless
dimensionless fluid temperature
T
dimensionless fluid temperature in
f

Laplace space
954 Xiaoxue Huang et al. / Energy Procedia 75 (2015) 946 955

Acknowledgements

The authors would like to express their thanks to the grant of the National Natural Science Foundation
of China (Grant No. 41272263).

References

[1] Tester JW, Anderson BJ, Batchelor AS, Blackwell DD, DiPippo R, Drake EM, et al. The Future of Geothermal Energy-
Impact of Enhanced Geothermal System (EGS) on the United States in the 21st century. Massachussets of Institute of Technology,
2006.
[2] Hadgu T, Zimmerman RW, Bodvarsson GS. Coupled Reservoir-Wellbore Simulation of Geothermal Reservoir Behavior.
Geothermics 1995; 24: 145-66.
[3] Murray L, Gunn C. Toward Integrating Geothermal Reservoir and Wellbore Simulation: TETRAD and WELLSIM. 15th
New Zealand Geothermal Workshop, The University of Auckland, New Zealand; 1993.
[4] Pruess K, Zhang Y. A Hybrid Semi-analytical and Numerical Method for Modeling Wellbore Heat Transmission. Thirtieth
Workshop on Geothermal Reservoir Engineering, Stanford University, Stanford, CA; 2005.
[5] Pan L, Oldenburg CM. T2WELL-An Integrated Wellbore-Reservoir Simulator. TOUGH Symposium, Lawrence Berkeley
National Laboratory, Berkeley, CA; 2012.
[6] Aunzo ZP, Bjornsson G, Bodvarsson GS. Wellbore Models GWELL, GWNACL, and HOLA User's Guide; 2008.
[7] Fard MH, Hooman K, Chua HT. Numerical simulation of a supercritical CO2 geothermosiphon. International
Communications in Heat and Mass Transfer 2010; 37:1447-51.
[8] Jiang F, Luo L, Chen J. A novel three-dimensional transient model for subsurface heat exchange in enhanced geothermal
systems, International Communications in Heat and Mass Transfer 2013; 41: 57-62.
[9] Sanaz S, Rafid AK, Frans B. An Efficient Computational Model for Deep Low-enthalpy Geothermal Systems, Computers &
Geosciences 2013; 51: 400-9.
[10] Shih TM. Numerical heat transfer. CRC Press; 1984.
[11] Munson BR, Okiishi TH, Huebsch WW, Rothmayer AP. Fundamentals of fluid mechanics, seventh edition. John Wiley &
Sons Inc; 2012.

Biography
Li Jun, Geothermal Research and Training Center of Tianjin University, Associate Professor. Research
interests include geothermal power generation, EGS techonology, micro-scale combustion, power MEMS
and underground spontaneous coal fires.

Appendix

The dimensionless forms for radial distance, time, and depth are
r * r / rw (A-1)
*
t tv / D (A-2)
z* z / D (A-3)
The dimensionless temperature functions are
T f ( z * , t * ) [t f  ti ] / [tinj  ti (0)] (A-4)
 Xiaoxue Huang et al. / Energy Procedia 75 (2015) 946 955 955

T s ( r * , z * , t * ) [ts  ti ] / [tinj  ti (0)] (A-5)

The unsteady-state solution in Laplace space can be obtained as (Appendix A)
wT f wT f
 q* (A-6)
wt * wz *
where q* hD[T f ( z * , t * )  T s ( r * , z * , t * ) ] / U f c f v  Dtic( z ) / [tinj  ti (0)]
*
(A-7)
r 1
2
w 2T s 1 wT s U s c ps vrw wT s
and  (A-8)
wr *2 r * wr * DO wt *
with initial conditions
T f ( z*, t* 0) 0 (A-9)
and
T s (r* , z* , t* 0) 0 (A-10)
And boundary conditions
T f ( z * 0, t * ) 1 (A-11)
wT s hrw
(T s  T f ) (A-12)
wr * r* 1
O
lim T s ( r * , z * , t * ) 0 (A-13)
r* of

Applying the Laplace transform to the mathematical model, the equations can be obtained as
d T f
 sT f q* (A-14)
dz *
q* hD[T  T ] / U c v  Dt c( z ) / s[t  t (0)]
f s f pf i inj i (A-15)
r* 1
2
d 2 Ts 1 d Ts U c vr
 ( )  s s ps Ts w
0 (A-16)
dr *2 r * r * DO

dTs hrw
*
(T s *  T f ) (A-17)
dr r* 1 O r 1

T f ( z * 0, s ) 1 / s (A-18)
The derived solutions are
T f ( z * , s ) C ( s ) exp(  pz * )  Dtic( z ) / ps[tinj  ti (0)] (A-19)
hD
p s (Z  1) (A-20)
U f cf v
hrw U s c ps vrw2 hr U c vr 2 U s c ps vrw2 U c vr 2
Z K0 ( s ) / [ w K 0 ( s ps w s )  sK1 ( s ps w s )] (A-21)
O DO O DO DO DO
U s c ps vrw2
s )r*K0 (
* * * hrw D O
Ts (r , z , s) T f ( z , s) (A-22)
O hr U c vr 2 U s c ps vrw2 U c vr 2
w
K 0 ( s ps w s )  sK1 ( s ps w s )
O DO DO DO