See

discussions, stats, and author profiles for this publication at: https://www.researchgate.net/publication/259127961

A heat loss analytical model for the thermal

front displacement in naturally fractured
reservoirs

Article in Geothermics · April 2014

Impact Factor: 2.95 · DOI: 10.1016/j.geothermics.2013.09.002

CITATIONS READS

2 40

3 authors, including:

Fernando Ascencio Fernando Samaniego

Universidad Nacional Autónoma de México Petróleos Mexicanos
34 PUBLICATIONS 62 CITATIONS 78 PUBLICATIONS 397 CITATIONS

SEE PROFILE SEE PROFILE

All in-text references underlined in blue are linked to publications on ResearchGate, Available from: Fernando Ascencio
letting you access and read them immediately. Retrieved on: 04 June 2016
 Geothermics 50 (2014) 112–121

Contents lists available at ScienceDirect

Geothermics
journal homepage: www.elsevier.com/locate/geothermics

A heat loss analytical model for the thermal front displacement in

naturally fractured reservoirs
Fernando Ascencio a , Fernando Samaniego b,∗ , Jesús Rivera b
a
Petróleos Mexicanos (PEMEX), Exploración y Producción, Marina Nacional 329, Col. Huasteca, Delegación Miguel Hidalgo, Torre Ejecutiva 41◦ piso,
México, D.F. 11311, Mexico
b
Secretaría de Posgrado e Investigación, Facultad de Ingeniería, Universidad Nacional Autónoma de México, (UNAM), México, D.F. 04250, Mexico

a r t i c l e i n f o a b s t r a c t

Article history: A theoretical study of the injection of separated cold water into naturally fractured hot geothermal reser-
Received 10 August 2013 voir rock is presented. The hot geothermal reservoir is assumed to be initially at a uniform temperature.
Accepted 18 September 2013 The fractured system is modeled as two interconnected homogeneous systems, one for the matrix and the
other for the fractures. Heat and mass balances are established for the interconnected system, when the
Keywords: cold injected fluid travels through the fractures in close contact with a hot matrix. Solutions to this prob-
Geothermal reservoir
lem are presented for two cases: one in which instantaneous thermal equilibrium takes place between the
Thermal front
injected cold fluid and the rock, and the second considers a non equilibrium thermal condition, for which
Underground fluid injection
Naturally fractured systems solutions are derived for the cases when heat transfer occurs under pseudo-steady state and transient
conditions. Heat interchange with underlying and overlying impermeable formations is also considered.
Type-curves are presented for the rate of advance of the thermal front with dimensionless injection time.
A sensitivity analysis was performed to investigate the effect of several parameters on the rate of advance
of the thermal front.
© 2013 Elsevier Ltd. All rights reserved.

1. Introduction • enhanced heat recovery from the resource through a secondary

Commercial exploitation of liquid-dominated geothermal
resources requires the disposal of large volumes of spent brine in
an environmentally safe way. This relatively cool brine is obtained
as a by-product of the separation process used to obtain the
steam for generating electricity. Separated fluids include non-
condensable gases, mainly H2 S and CO2 , as well as substances
such as silicates and toxic compounds like arsenic, boron and mer-
cury, all of which are concentrated in the liquid phase. Due to
environmental regulations, this brine cannot be discarded at the
surface without prohibitively expensive chemical treatments, and
consequently must be injected in the subsurface. Besides com-
plying with environmental regulations, brine injection into the
geothermal reservoir may provide the following benefits (Horne,
1982a,b, 1985; Schroeder et al., 1982; Pruess and Bodvarsson, 1984;
Stefansson, 1997):
• additional pressure support that can reduce the geothermal
reservoir pressure decline due to fluid withdrawal,
∗ Corresponding author. Tel.: +52 55 5550 8712.
E-mail addresses: ance@unam.mx, pexfsamaniegov01@pemex.com,
ance@servidor.unam.mx (F. Samaniego).
“heat mining” process,
• reduce ground subsidence resulting from fluid extraction.
Despite the positive aspects of underground fluid injection
mentioned, extreme care must be taken when such injection
is to be performed into naturally fractured systems. In these
systems injected cool fluids could rapidly travel through open,
communicating fracture networks, which usually connect injec-
tion and production wells, resulting in a “short-circuit”. When
this “short-circuit” occurs, injected fluids may not have suffi-
cient residence time in the reservoir to receive enough heat from
surrounding hotter rock, resulting in undesirable fluid enthalpy
decrease at producing wells. Since most geothermal reservoirs are
located in highly fractured igneous rocks, there have been sev-
eral cases where detrimental effects due to cold fluid injection
have been experienced (Horne, 1982a,b, 1985; Bodvarsson and
Tsang, 1982; Bodvarsson and Stefansson, 1989; Gringarten et al.,
1975; Gringarten and Sauty, 1975; Stefansson, 1997; Goyal, 1999;
Axelsson and Dong, 1998; Axelsson et al., 2001; Bodvarsson, 1974;
Gevrek, 2000).
Lauwerier (1955) published the first and perhaps the most
widely known solution for the temperature distribution due to
injection of a hot fluid in a reservoir, which includes heat losses to
the impermeable strata surrounding the reservoir. His model of the

0375-6505/$ – see front matter © 2013 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.geothermics.2013.09.002
 F. Ascencio et al. / Geothermics 50 (2014) 112–121 113

Nomenclature
0 initial
AHTb effective heat transfer area per unit of bulk forma- 1 unit temperature drop at the rock-fluid interface
tion volume, m−1
b parameter, Eq. (23)
Bi Biot number, Eq. (25)
hot fluid injection assumes constant injection temperature, linear
c specific heat capacity, J/kg ◦ C
one-dimensional, incompressible flow in a homogeneous sand, infi-
ĥ convective heat transfer coefficient, J/m2 s ◦ C
nite vertical thermal conductivity within the permeable strata, zero
h fracture thickness, m
horizontal formation thermal conductivity and zero permeability in
H permeable fractured stratum thickness, m
the horizontal direction in the surrounding strata. Malofeev (1960)
l rock matrix block characteristic length, m
has shown that Lauwerier’s solution is also applicable in the radial
Pe Peclet number, Eq. (26)
flow case. Avdonin (1964) considered a non-zero value for the
qi volumetric fluid injection rate, m3 /s
thermal conductivity within the reservoir in the horizontal direc-
q* matrix–fracture heat flux interchange rate per unit
tion. All other assumptions were identical to those of Lauwerier.
of total volume, J/m3 s
Bodvarsson and Tsang (1982) investigated the advancement of the
q1 heat flux rate per unit temperature drop at the
thermal front during injection into a fractured reservoir system,
matrix–fracture interface, J/m2 s ◦ C
consisting of equally-spaced horizontal fractures. Chen and Reddell
rb rock matrix spherical block radius, m
(1983) developed analytical solutions of temperature distribution
s Laplace transform parameter
for thermal injection into a confined aquifer, with a cap rock of
t time, h
finite thickness. Heat transfer by horizontal conduction and convec-
T temperature, ◦ C
tion within the aquifer and by vertical conduction in the caprock
 macroscopic (Darcy’s) velocity = m , m/s
and bedrock were considered. Shaw-Yang and Hund-Der (2008)
m microscopic velocity, m/s
developed a mathematical model for simulating the thermal energy
V volume
transfer in a confined aquifer, with different thermo-physical prop-
x horizontal coordinate, Fig. 2
erties in the underlying and overlying rocks. The heat transfer
xf refers to a position vector of any point in the fracture
by horizontal convection occurs along the radial direction and
y horizontal coordinate, Fig. 2
by vertical thermal conduction in the overlying and underlying
z vertical coordinate, Fig. 2
rocks. Boyadjiev et al. (2005) presented a paper concerned with the
Greek letters fractional extension of the Lauwerier formulation of the problem
˛ thermal diffusivity, =/c, m2 /s related to the temperature field description in a porous medium
˛
¯ saturated medium thermal diffusivity [rock-fluid (= saturated with oil.
/c)] When a relatively cold separated geothermal brine is injected in
˛ characteristic parameter of the blocks (=AHTb /l), m−2 the hot reservoir, two distinct displacement fronts begin to develop
T temperature difference, ◦ C and grow around the injection well. The first front is known as
 thermal conductivity, J/ms ◦ C the “chemical front” or the “hydrodynamic front”, Fig. 1. The sec-
 rock–fracture interaction coefficient = r = AHTb /l, ond front, called the “thermal front”, whose temperature is lower
J/m2 s ◦ C than that of the reservoir fluids, travels some distance behind the
 thermophysical parameter, Eq. (21) former. The chemical front has a temperature close to that of the

ˆ thermophysical parameter, Eq. (21) reservoir fluid, and can be identified from differences in concentra-
general space variable; = xD and = rD2 /2 for linear tions of chemical species present in the injected and reservoir fluids,
and radial flows, respectively respectively. The mathematical model described in this paper,
 density presents solutions that allow the computation of the distance that

rock matrix spherical block dimensionless radius, separates the chemical and thermal fronts within the reservoir at
Eq. (17) a given time, so that once the presence of the former is detected
ωf ratio of the energy stored in the fluid and of the
total energy stored in the naturally fractured porous
medium
 fracture porosity
ωr ratio of the energy stored in the rock and of the
total energy stored in the naturally fractured porous
medium

Subscripts
b rock matrix block
D dimensionless
f fluid (or fracture)
HF hydrodynamic (chemical) front
HT heat transfer area of a matrix block (i.e. a sphere)
HTb heat transferred per unit of total volume
i injection
r rock
s under and over lying impermeable strata
TF thermal front
Fig. 1. Hydrodynamic and thermal fronts developed during the injection of cold
brine into a geothermal reservoir.
114 F. Ascencio et al. / Geothermics 50 (2014) 112–121

In general some time is required for thermal equilibrium to be

reached between the injected fluid and surrounding hot reservoir
rock and independent energy conservation equations must be con-
sidered for the fluid and rock matrix, coupled by means of a term
representing the heat transfer between fluid and the rock.
In formulating the mathematical model for cold water injection
into a naturally fractured reservoir (NFR), the following assump-
tions are invoked:

• The injected fluid is incompressible and flows through the frac-

tures.
• The thermal properties of the injected and reservoir fluids are
identical.
• For the radial and planar cases, the well is modeled as a line and
plane source, respectively.
• The well fully penetrates the naturally fractured reservoir (NFR)
of thickness H.
• The fractured and matrix systems are homogeneous, isotropic
and of infinite horizontal extent.
• The thermal properties of the fluid and of the matrix are constant.
Fig. 2. Idealized model for fluid injection.
• The upper and lower layers are impermeable, with identical phys-
ical and thermal properties.
at producing wells from chemical analysis of produced fluids, an • There are no phase changes.
estimate of the location of the thermal front can be obtained. • Gravitational effects are neglected.
• The injected cold water flows only through the fractures.

2. Mathematical model
Initially, at time t = 0, the temperature in the system (the NFR,
We study the heat transfer during the flow of a fluid through a the fluid and the upper and lower layers) is considered uniform
naturally fractured reservoir, considering thermal non-equilibrium and equal to T0 . At the well (r = 0), the injection temperature, Ti ,
between the fluid in the fractures and the matrix blocks. The model and rate, qi , remain constant for t > 0.
includes: (1) convective heat transport though the fractures and In a NFR, fluid flows through the fractured network, and due to
infinite vertical heat transfer conductivity (uniform temperature); the finite matrix block dimensions, thermal equilibrium conditions
(2) heat transfer from the underlying and overlying imperme- do not prevail during injection. Thus, it is necessary to consider
able layers, with finite and infinite thermal conductivities in the energy conservation equations for the fracture and for the matrix.
vertical and horizontal directions; and (3) the temperature dif- A coupling term, q ∗ (t, xf ), is introduced to model the heat flow
ference between the matrix blocks and the fluid in the fractures. between the rock (matrix) and the fluid (fractured).
The temperature new solution applies for radial and linear flow The energy equation for the impermeable upper layer is given
(Malofeev, 1960) through the use of the energy equation of the by Eq. (1):
variable = rD2 /2.  
Our model is a combination of those of Lauwerier (1955) and of ∂Ts ∂2 Ts 0 < r, z<∞
s cs = s 2 (1)
Bodvarsson and Tsang (1982), but differs from the former in that ∂t ∂z t>0
non-thermal equilibrium conditions in the reservoir are consid-
ered, and from the latter because it includes thermal energy transfer Energy equations for the fluid (f) and for the rock matrix (r, rock)
from the overlying and underlying layers (caprock and base rock). in the reservoir can be written as follows (in accordance to Mal-
The mathematical model is illustrated in Fig. 2 for a linear ofeev, a spatial variable ␹ that applies for linear and radial flows is
coordinate system, which can easily be extended to a cylindrical defined in Eq. (17):
coordinate system. Physical limits for this model are: ⎧ ⎫
⎪
⎪ 0<r<∞ ⎪
⎪
∂Tf ∂2 Tf f cf qi ∂Tf
⎨ 1 ⎬
• Permeable fractured stratum: − ∞ < x, y < ∞ ; − H < z < 0. f cf = f − + q ∗ (t, xf ) − H<z<0 (2)
• Upper impermeable stratum: − ∞ < x, y < ∞ ; 0 < z < ∞ . ∂t ∂z 2 2 Hr ∂r ⎪
⎪
2 ⎪
⎪
• Lower impermeable stratum: − ∞ < x, y < ∞ ; − ∞ < z < − H.
⎩ ⎭
t>0

When dealing with heat transfer phenomenon for non- ⎧ ⎫

isothermal fluid flow through a permeable medium, consideration ⎪
⎪ 0<r<∞ ⎪
⎪
⎨ 1 ⎬
should be given to whether such flow is taking place through a ∂Tr ∂2 Tr
(1 − )r cr = (1 − )r − q ∗ (t, xf ) − H < z < 0 (3)
granular porous medium or through a fractured system. For flow ∂t ∂z 2 ⎪ 2 ⎪ ⎪ ⎪
through a granular porous medium, there is a greater chance for ⎩ ⎭
t>0
thermal equilibrium to be rapidly reached between the fluid and
surrounding rock, since flow velocities are usually small (except in The boundary conditions at the interface (z = 0) are:
the vicinities of injection wells) and the grain surface area is large.
On the other hand, if fluid is flowing through fractures, fluid veloc-
⎧ ⎫
⎨0 < r < ∞⎬
ities can be very fast. The surface area available for heat transfer is
Tr = Ts ; Tf =
/ Ts z=0 (4)
smaller than for granular media, and it will require a much larger ⎩ ⎭
time to attain thermal equilibrium than for the granular media. t>0
 F. Ascencio et al. / Geothermics 50 (2014) 112–121 115

⎧ ⎫
⎨0 < r < ∞⎬ • Corresponding dimensionless expressions for TDs and TDf , are:
∂Tf
−f = h(Tf − Ts ) z=0 (5) T0 − Ts T0 − Tf
∂z ⎩ ⎭ TDs = , TDf = (16)
t>0 T0 − Ti T0 − Ti
⎧ ⎫ • Spatial dimensionless variables:
⎨0 < r < ∞⎬ ⎧ ⎫
∂Tr ∂Ts r 1 z
(1 − )r = s + h(Tf − Ts ) z=0 (6) ⎪
⎪ rD = , = rD2 , zD = ⎪
∂z ∂z ⎩ ⎭ ⎪
⎨ (H/2) 2 (H/2) ⎪
⎪
⎬
t>0
(17)
Based on the symmetry of the system shown in Fig. 2, there ⎪
⎪ rb
⎪
⎪
⎪
⎩ xD = x l ⎪
⎭
follows: , lD = ,
=
⎧ ⎫ (H/2) (H/2) (H/2)
⎪
⎪ 0 < r < ∞⎪
⎪
∂Tf
⎨ 1
⎬ • Rock–fracture interaction coefficient:
∂Tr
= =0 z=− H (7) 1
∂z ∂z ⎪ 2
⎪ ⎪
⎪ D = Dr (AHTb /l)H 2 = Dr (AHTbD /lD ) (18)
⎩ ⎭ 4
t>0
• Effective heat transfer area per unit of bulk formation volume:
Integrating Eqs. (2) and (3) with respect to the vertical coordi-
H
nate z, within the limit −H/2 and 0, we have: AHTbD = AHTb (19)
2
 0 
∂Tf 
0 0
∂Tf f cf qi ∂Tf • Parameters ωf and ωr relating the fluid energy to total energy
f cf dz = f  − dz
−H/2
∂t ∂z  2 Hr −H/2
∂r content (rock-fluid), and rock energy to total energy content,
z=−H/2
respectively:
 0
f cf
(1 − ) c 
+ q ∗ (t, xf )dz (8) ωf = , ωr =
r r
(20)
−H/2 c c

 0 where ωf + ωr = 1.
∂Tr 
0
∂Tr • Other thermophysical parameters:
(1 − )r cr dz = (1 − )r
−H/2
∂t ∂z  hH hH
z=−H/2
= , ˆ =
 (21)
 0
2¯ 2r
− q ∗ (t, xf )dz (9) ˛s
−H/2 a2 = (22)
˛¯
Considering the boundary conditions given by Eqs. (4)–(7), Eqs. √
Ds  cs s
(8) and (9) can be rewritten as follows: b= = s (23)
a c ¯
1 ∂Tf
z=0: Hf cf = h(Tf − Ts )
2 ∂t r
⎧ ⎫ Dr = (24)
f cf qi ∂Tf
⎨0 < r < ∞⎬ ¯
1
− + Hq ∗ (t, xf ) (10) • Biot numbers:
4 r ∂r 2 ⎩ ⎭
t>0 ĥH ĥH
Bi = , B̂i = (25)
2¯ 2r
1 ∂Ts ∂Ts
z = 0 : H(1 − )r cr = s + h(Tf − Ts ) • Peclet number:
2 ∂t ∂z
⎧ ⎫ f cf qi ωf (qi /)
⎨0 < r < ∞⎬ Pe = = (26)
1 4 H ¯ 4 H ˛
¯
+ Hq ∗ (t, xf ) (11)
2 ⎩ ⎭ f cf VH ωf v H
t>0
Pe = = (27)
4¯ 4˛
¯
We introduce the following dimensionless variables: The Peclet number Pe is the convective to conductive heat
transfer ratio; this definition is taken from Bachu and Dagan
• Dimensionless time, tD :
(1979). The tilde mark (“ ”) in Eq. (27) indicates the Peclet number
4t
¯ 4˛t
¯ refers to linear flow conditions.
tD = = 2 (12) • Heat transfer dimensionless parameter:
cH 2 H
where ˛ ¯ = /c
¯ is the saturated medium thermal diffusivity H 2 q∗
q∗D = (28)
(rock-fluid) and c and ¯ are the heat capacity and thermal con- ¯ 0 − T)
4(T
ductivity of the saturated medium respectively, defined by Eqs.
(13) and (14) (Hadidi et al., 1956):
Assuming that the horizontal conductive heat transfer rate is
c = f cf + (1 − )r cr (13) negligible for the permeable stratum and the confining strata,
a dimensionless mathematical formulation for the heat transfer
¯ = f + (1 − )r (14) problem (Eqs. (1), (10) and (11)) can be written as follows:
 
• Dimensionless temperature: ∂TDs a2 ∂2 TDs 0 < zD < ∞
= (29)
T0 − T ∂tD ∂zD2
tD > 0
TD = (15)
T0 − Ti
116 F. Ascencio et al. / Geothermics 50 (2014) 112–121

∂TDf ∂TDf 2.2. Mathematical solution

zD = 0 : ωf = Bi(TDs − TDf ) − 2Pe
∂tD ∂
 tD   Lauwerier (1955) presented an analytical solution to calculate
∂TDf () 0< <∞ the temperature distribution in a permeable stratum when hot fluid
− qD1 (tD − )d (30)
0
∂ tD > 0 is injected into a horizontal, homogeneous porous stratum satu-
rated with a cold fluid. He considered only convective heat transfer
∂TDs ∂TDs in the permeable stratum, with instantaneous thermal equilibrium
zD = 0 : (1 − ωf ) = Dr − Bi(TDs − TDf ) between the fluid and rock (fractures and matrix), as well as heat
∂tD ∂zD
 tD   transfer into the confining strata. Both the heat transfer problems
∂TDf () 0< <∞ described in the present paper for a double porosity system for-
+ qD1 (tD − )d (31)
0
∂ tD > 0 mulated in accordance to the well testing Warren and Root model
(1963) (Eqs. (29)–(34)) and Lauwerier’s problem, can be solved by
Initial condition: means of a similar methodology.
⎧ ⎫ Dimensionless fluid and rock temperature distributions, TDf and
⎨0 < ⎬ TDs can be obtained by solving the heat transfer problem stated by
TDf = TDr = TDs = 0 zD → ∞ (32)
⎩ ⎭ Eqs. (29)–(34), in the Laplace space (Appendices A and B, Eqs. (77)
tD = 0 and (78)). This solution is given by Eqs. (38) and (39).
Fluid temperature distribution in the fractures is:
Boundary conditions:   √   
  1 ωr + b/ s − q̄D1 (s)
= zD = 0 T̄Df = exp − √ Bi + q̄D1 (s) + ωf s
TDf = 1 (33) s ωr s + b s + Bi 2Pe
tD > 0 (38)

⎧ ⎫ while the temperature distribution in the rock matrix is (Ascencio

⎨0 < < ∞⎬ et al., 2000):
TDs → 0 zD → ∞ (34)  
⎩ ⎭ Bi + sq̄D1 (s) √
tD > 0 T̄Ds = Ds √ TDf exp(− szD /a) (39)
ωr s + b s + Bi
It is important to mention that the heat flow problem, given by For pseudosteady matrix–fracture heat transfer conditions, the
Eqs. (30) and (31) considers that the fractured system is thermally transfer function is expressed by:
anisotropic, i.e., it has infinite vertical thermal conductivity and a
ωr
negligible (zero) horizontal conductivity. This assumption holds for q̄D1 (s) = (40)
high injection rates. (ωr /D )s + 1
Eqs. (29)–(34) include both linear and radial formulations of the where D is the rock–fracture interaction coefficient defined by:
problem under consideration, depending upon the definition used
1
for the space variable ; thus if ␹ = xD the problem formulation D = Dr ˛ H 2 = Dr (AHTbD /lD ) (41)
4
corresponds to the linear case, whereas if = rD2 /2 the problem
formulation is for a radial system. Parameter ˛ (= AHTb /l) has dimensions of L−2 and l is a character-
The coupling term, q∗D (third term of the right side) in Eqs. (30) istic matrix length. This parameter is basically the same as the one
and (31) describes the heat flow from the matrix to the fluid flowing used to analyze the pressure response in double porosity systems
in the fractures. This parameter will be discussed in the next section. by Warren and Root (1963), as given by the following expression:
4n(n + 2)
2.1. Heat flow from the matrix to the fluid in the fractures, q* ˛ =
l2
Due to the variable temperature at the matrix–fracture inter- where n is the number of perpendicular planes limiting the matrix
face, the heat flow from the matrix to the fluid in the fractures is blocks.
described by a convolution integral, given by Eq. (35): For transient matrix–fracture heat transfer conditions, consid-
 t
ering a spherical block geometry, the resulting expression is given
∂Tf () by Ascencio et al. (2000):
q∗ = q1 (t − )d (35)
∂
0 f (s)
q̄D1 (s) = Dr (AHTbD /
) (42)
where Tf = T0 − Tf (t, x̄f ) is the temperature drop, and q1 is the s(1 + (1/B̂i)f (s))
matrix–fracture heat transfer per unit temperature drop (transfer
where
function). By means of Fourier’s law, q1 can be expressed as:  
 f (s) = (
2 /Dr )ωr s) − 1
q1 (t) = AHTb r ∇ Tb1 (t)
(
2 /Dr )ωr s coth( (43)
(36)
interface
 In Eq. (42), the parameter B̂i is related to the temperature drop
where ∇ Tb1 (t) represents the interface matrix–fracture at the interface between the matrix blocks and the fluid in the
interface
temperature space gradient for a unit temperature drop, Tb in fractures.
Tb = T0 − Tb is the temperature of the matrix blocks and AHTb rep- The following expressions are derived in Appendix B for a nat-
resents the effective heat transfer area per unit bulk formation urally fractured medium with spherical block geometry (see Eqs.
volume. (88) and (93)):
Using dimensionless variables, Eq. (35) can be written as fol-
lows: D = 2 Dr /
2 (44)
 tD
H 2 q∗ ∂TDf ()
q∗D = qD1 (TD − )d (37) lD /
=
3
(1 − ) (45)
¯ 0 − T)
4(T 0
∂ 2
 F. Ascencio et al. / Geothermics 50 (2014) 112–121 117

Substituting Eqs. (44) and (45) in Eqs. (42) and (43) yields:
3 f (s)
q̄D1 (s) = (1 − )D , (46)
2 s(1 + (1/B̂i)f (s))
and
 
f (s) = (ωr /D )s coth ( (ωr /D )s) − 1 (47)
In Eqs. (46) and (47), the transient heat transfer has been
expressed in terms of the same parameters used in the pseu-
dosteady state model (D and ωr ). The only additional parameter
is the fracture porosity .

2.3. Special cases

2.3.1. Short times

For short times, t → 0(s → ∞) and q̄D1 (s) → 0, Eq. (38) can be
written as follows:
1


T̄Df = exp −ωf s (48)
s 2Pe
The real time solution to Eq. (48) is given by:


TD ( , tD ) = U tD − ωf (49) Fig. 3. Development and growth of the thermal front for non-isothermal fluid injec-
2Pe
tion in a naturally fractured system.
where the Heaviside unit function is defined as:
 
1 foru≥0 2.3.4. Long times
U(u) = (50)
0 foru < 0 For long times, t → ∞ (s → 0) and q̄D1 (s) ≈ ωr , considering a
value of b =
/ 0, it can be shown that for both the pseudosteady state
and the transient heat transfer functions, Eq. (38) can be simplified
2.3.2. Thermal equilibrium
as:
Assuming that the fractures are thermally isolated (b = 0; Eq.
1
 √

(23)), for long times, t → ∞ (s → 0), the heat transfer function Eq. T̄D = T̄Df = T̄Ds = exp −(b/ s + 1)s (56)
(40) simplifies to q̄D1 (s) → ωr , and Eq. (38) can be written as fol- s 2Pe
lows: The inverse Laplace’s transform of Eq. (56) can be expressed as:
1

T̄Df = exp − s (51)  
s 2Pe
b( /Pe)


The real space solution of Eq. (51) is given by: TDf ( , tD ) = erfc  U tD − (57)
 2 tD − ( /(2Pe)) 2Pe

TD ( , tD ) = U tD − (52)
2Pe where erfc is the complementary error function.
Equation (57) is analogous to the solution previously pub-
2.3.3. Pseudosteady state solution lished by Lauwerier (1955), which assumes instantaneous thermal
Substituting Eq. (40) in Eq. (38) and considering again b = 0, the equilibrium conditions. However when these conditions are not
following equation is obtained: appropriate it is not possible to find an analytical expression for

1

ωr
  the Laplace’s inverse transformation of Eq. (38). Therefore, for these
T̄Df = exp − + ωf s conditions, Stehfest algorithm (Stehfest, 1970) was used to numer-
s (ωr /D )s + 1 2Pe ically invert Eq. (38).
The real space solution is given by Eqs. (53) and (54) (Ascencio
and Rivera, 1994): 3. Expression for the thermal front (TF)





TDf ( , tD ) = J D , (/ωr ) tD − ωf · U tD − ωf The thermal front is defined as the geometric locus described
2Pe 2Pe 2Pe
by fluid particles flowing through the fracture network, whose
(53)
temperature, TTF has a specified value in between that of the resi-
where dent reservoir fluids and the injected fluid (Pruess and Bodvarsson,
⎧ ∞  k  ⎫
⎪  √ ⎪ 1984).
⎪
⎪  
≤1⎪ ⎪
⎪
⎨
exp[−(v, )]
v
Ik (2 v),
v ⎪
⎬ TTF = Ti + f (T0 − Ti ) (58)
k=0
∞  −k
J(v, ) =  (54)
⎪
⎪   √  ⎪
⎪ For a symmetrical thermal front where f = 0.5, the dimensionless
⎪
⎪ ≥1 ⎪
⎪
⎩ exp[−(v, )] v
Ik (2 v),
v ⎭ thermal front temperature can be expressed as:
k=0
TDTF = 0.5 (59)
In the evaluation of the above solution, the following recurrence
relation has been used (Abramowitz and Stegun, 1972; Luke, 1962): A type-curve was constructed considering the assumption of Eq.
(59) for the thermal front and Eq. (38) was inverted numerically
by means of the Stehfest algorithm to find the fluid temperature
2k
Ik+1 (x) = Ik−1 (x) − I (x) (55) distribution flowing through the fractured stratum. Transient and
x k
pseudo-steady heat transfer conditions between fluid and rock
118 F. Ascencio et al. / Geothermics 50 (2014) 112–121
Table 1
Parameter values used for constructing type curves shown in Figs. 3 and 4.
␭D ␻f   
ˆ b
100,000 0.10 0.1 100 ∞ 1
matrix were considered. To construct the type curve shown in Fig. 3,
the typical parameter values presented in Table 1 were used. Fig. 4
presents the thermal front propagation in a naturally fractured
medium for different values of D .
4. Discussion of the type curve
Four periods can be identified in Fig. 3 during the development
and growth of the thermal front in a permeable fractured stratum.
4.1. Early time period
At early times, both the chemical or hydrodynamic front, HF,
and the thermal front, TF, travel together. During this period, heat
transferred from the rock matrix has not yet started to influence
the propagation of the TF. The thermal front propagation at early
times is described by Eq. (49).
4.2. Transition period
At intermediate dimensionless times a transition region devel-
ops. During this period, heat is transferred from the matrix rock
to the fluid traveling through the fractures, producing a delay in
the TF with respect to the HF, which appears as a departure of the
TF curve from the early straight line. The duration of this period
strongly depends on the heterogeneity of the fractured medium,
and on the thermal properties of the fluid-rock system.
4.3. Thermal equilibrium period
At long dimensionless times, thermal equilibrium is finally
reached in the heat transfer process as the TF curve gets fur-
ther away from the HF curve and approaches the instantaneous
(Lauwerier) thermal equilibrium solution (ITES). During this period,
Fig. 4. Thermal front propagation in a naturally fractured medium, for different
values of the rock/fracture interaction parameter D .
any heat transferred to or from the upper and lower confining
strata is negligible. The thermal front displacement in this period
is described by Eq. (52).
4.4. Late time period
At very long dimensionless times heat transferred from the con-
fining strata makes the TF curve start to deviate from the ITES curve,
bending to the right, as shown in the upper right hand side corner
of Fig. 3.
It can be observed from Fig. 3 that both the HF and the TF fronts
follow parallel log-log straight lines during the thermal equilibrium
period; the distance between the two fronts can be expressed for
the radial and linear cases, respectively, as:

ır = rHF − rTF = (1 − ωf )rHF (60)
ıl = xHF − xTF = (1 − ωf )xHF (61)
From Eq. (38), it can be shown that the parameter ωf has a
direct effect on the duration of the transition period. The duration
of the transition period, t, can be calculated from the following
expression:
 
1
t = tTF − tHF = −1 tHF (62)
ωf
where tHF and tTF are the arrival times for the hydrodynamic and
thermal front, respectively.
It is apparent from Fig. 3 that the transition period starts earlier
for transient heat flow conditions from the matrix to the fluid, and
lasts longer than in the case when the heat is transferred under
pseudo-steady state conditions. It can also be observed that the
transition from the early time period to the thermal equilibrium
period is smoother for the transient heat flow condition than for
the pseudo-steady state one.
Fig. 4 shows that the beginning  of the transition period
 is deter-
mined by the parameter D = (1/4)(H 2 Dr (AHTbD /l)) , which is
indicative of the degree of interaction between the fracture sys-
tem and the porous matrix rock. For large D (greater than 106 ),
the period dominated by rock-fluid thermal equilibrium conditions
starts earlier. This is true when either the matrix block size (l) is very
small, or when the permeable strata thickness (H), and/or the heat
transfer area per unit volume (AHTbD ) is very large. Generally speak-
ing, for a given thickness, in the limit it would mean that the porous
medium behavior (see Fig. 4, where the transition ends earlier as
D increases) is close to homogeneous.
5. An example
To illustrate the application of the model, the following syn-
thetic “field” example is presented. A cool separated brine is
injected back into a hot naturally fractured reservoir at a constant
volumetric flow rate of 0.03 m3 /s. A purely radial fluid flow pattern
away from the injection well is assumed. The temperature differ-
ence between the injected and resident reservoir fluids is set as
50 ◦ C. It is desired to compute the locations of the hydrodynamic
and thermal fronts for dimensionless times of 1.0 × 10−4 , 1.0 × 10−2
and 1. We also want to calculate the positions of the thermal and
chemical fronts under rock-fluid thermal equilibrium conditions,
neglecting any heat exchange with adjacent strata. From the type
curves in Figs. 3 or 4, it is apparent that the three dimensionless
times correspond to three different periods: transition, thermal
equilibrium, and to a late time period when heat transfer from
surrounding strata becomes significant.
 F. Ascencio et al. / Geothermics 50 (2014) 112–121 119

Table 2 The model can treat rock-fluid heat transfer under both transient
Geothermal reservoir parameters for the synthetic example.
and pseudo-steady state conditions.
Density, f , kg/m3 1000 • A type-curve was developed to describe the development and
Fluid Specific heat, cf , J/kg ◦ C 4200 growth of the chemical (or hydrodynamic) front and of the ther-
Thermal conductivity, f , W/m ◦ C 1
mal front, when a relatively cool injected fluid displaces a hot
Porosity,  0.1 resident geothermal fluid through a naturally fractured forma-
Density, fr , kg/m3 2700 tion. The rate of advance of the thermal front is characterized by
Rock
Specific heat, cfr , J/kg ◦ C 1000
four distinct flowing periods: (1) early time period, (2) transition
Thermal conductivity, fr , W/m ◦ C 2
period, (3) thermal equilibrium period, and (4) late time period.
Saturated Heat capacity, c, J/m3 ◦ C 2.8 × 106 • The main parameters affecting the temperature distribution
rock Thermal conductivity, , W/m ◦ C 1.9
within the permeable fractured formation during non-isothermal
fluid injection are: the ratio of the energy stored in the fluid and
of the total energy stored in the naturally fractured porous, ωf ,
The thickness of the fractured permeable stratum, and the
the rock–fracture interaction coefficient, D , the Peclet number,
matrix block radius are assumed to be 100 m and 1.5 m, respec-
Pe, and the permeable fractured stratum thickness, H.
tively. Using these values, D is calculated as 10,000.
• The parameter D determines the beginning of the transition
Thermophysical properties for the fluid, the dry hot matrix rock,
period. It is an indication of the degree of heterogeneity of the
and the fluid saturated hot matrix rock are given in Table 2.
fractured stratum. For a homogeneous medium, D → ∞.
For each one of the dimensionless times considered in the exam-
• Heat transfer from the impermeable confining strata to the per-
ple, a value of the dimensionless variable /2Pe is determined from
meable fractured strata slows down the rate of advance of the
the type curve in Fig. 4. Then, the position of the thermal front is
thermal front.
computed using Eq. (63):
 
f cf qi H Appendix A. Solution process
rTF = (63)
4  2Pe curve
After the Laplace transform is applied to Eqs. (29)–(31), and from
The time at which the thermal front separates from the chemical boundary and initial conditions (32)–(34), the following expres-
front is given by Eq. (64), sions are obtained:
√ 
cH 2 T̄Ds = C exp(− szD /a) 0 < zD < ∞ (67)
t= tD (64)
4¯
The position of the hydrodynamic (or chemical) front, rHF , can T̄Df ( = 0, zD = 0) = 1/s (68)
be obtained from Eq. (31).
For a radial geometry, it is given by:
T̄Ds ( , zD → ∞) = 0 (69)

qi t
rHF = (65)
 H dT̄Df
zD = 0 : sωf T̄Df = Bi(T̄Ds − T̄Df ) − 2Pe − sT̄Df (s)q̄D1 (s) (70)
d
∗ , assuming thermal equi-
The position of the thermal front, rTF
librium conditions and neglecting any contribution from adjacent
∂T̄Ds
stratum, is computed from Eq. (66): zD = 0 : sωr T̄Ds = Ds − Bi(T̄Ds − T̄Df ) + sT̄Df (s)q̄D1 (s) (71)
 ∂zD
∗ qi t Evaluating Eq. (67) at zD = 0:
rTF = ωf (66)
 H 
T̄Ds  =C (72)
zD =0
Table 3 shows results for the thermal front development and for
the locations of the chemical and thermal fronts. The derivative of Eq. (67) with respect to zD , evaluated at zD = 0
gives:
 √
∂T̄Ds 
6. Conclusions
s
 =− C (73)
∂zD a
Based upon the theoretical developments and results presented zD =0
in the preceding sections, the following conclusions can be drawn:
Substituting Eqs. (72) and (73) into Eq. (71) yields:
√
• A mathematical model has been developed to solve the heat Ds s
sωr C = − C − Bi(C − T̄Df ) + sT̄Df (s)q̄D1 (s) (74)
transfer problem for non isothermal fluid injection into a nat- a
urally fractured formation. Expressions are presented for the Solving for C, we obtain:
calculation of temperature, profiles for the fluid flowing through
the fractured system, as well as for the rock matrix, under instan- Bi + sq̄D1 (s)
C= √ T̄Df (s) (75)
taneous and non-instantaneous thermal equilibrium conditions. sωr + b s + Bi
Substitution of Eq. (75) into Eq. (72), taking into account Eq. (70)
Table 3 and rearranging terms gives:
Results for the synthetic example.  √  
dT̄Df ωr + b/ s − q̄D1 (s)
tD /(2Pe) t (d) rHF (m) rTF (m) ∗
rTF (m) −2Pe =− √ Bi + q̄D1 (s) + ωf sT̄Df (s)
−4 −4
d ωr s + b s + Bi
1.0 × 10 3.0 × 10 4.3 1.9 × 10 1
9.74 7.3
(76)
1.0 × 10−2 1.0 × 10−2 4.3 × 102 1.9 × 102 7.3 × 101 7.3 × 101
1.0 6.0 × 10−1 4.3 × 104 1.9 × 103 5.6 × 102 7.3 × 102
120 F. Ascencio et al. / Geothermics 50 (2014) 112–121

Substituting Eq. (75) in Eq. (70), we obtain: The energy balance equation for the rock matrix blocks,
  √  considering pseudo-steady state rock-fluid heat transfer (lumped
1 ωr + b/ s − q̄D1 (s) solution) is:
T̄Df (s) = exp − √ Bi
s ωr s + b s + Bi
∂T̃Db1
  ωr = −D (T̃Db1 − 1) (87)
∂tD
+q̄D1 (s) + ωf s (77)
2Pe From a comparison between Eqs. (86) and (87), it follows that:
2
Solving Eq. (71), the temperature distribution within the rock D = (88)
matrix is: (
2 /Dr )
  In terms of real variables, Eq. (88) can be expressed as:
Bi + sq̄D1 (s) √
T̄Ds (s) = √ T̄Df exp(− szD /a) (78)
ωr s + b s + Bi 2
˛ = (AHTb /l) = (89)
r  2b

Appendix B. Sphere characteristic length From the definition of AHTb , and considering a spherical matrix
block, it thus follows:
The temperature distribution within a spherical matrix block is AHT  4 r  2b 
given by (Ascencio and Rivera, 1994): AHTb rb = rb = r (90)
Vtotal Vtotal b


sin h( sωr /Dr rD )T̄Df /s
rD T̄Db =     ! (79)
1 1
sin h( sωr /Dr
) + sωr /Dr cos h( sωr /Dr
) −

sin h( sωr /Dr

)
B̂i

Considering that there is no temperature drop at the

rock–fracture (fluid) interface (both systems are at thermal equi- where Vtotal is the volume of the representative element.
librium, B̂i → ∞), Eq. (79) reduces to:
Vtotal = (2rb + h )
3
(91)


sin h( sωr /Dr rD ) Rearrangement of Eq. (90) gives:
T̄Db1 =  (80)
srD sin h( sωr
2 /Ds ) (4/3) r  3b V
AHTb rb = 3 = 3 blocks = 3(1 − ) (92)
The solution of Eq. (80) in Laplace’s space is (Carslaw and Jaeger, Vtotal Vtotal
1959): Finally, combining Eqs. (89) and (91), and solving for l yields:

2
 (−1)n
∞
  n rD l=
3(1 − ) 
rb , lD =
3(1 − )

(93)
TDb1 = 1 + exp −n2 2 tD /(ωr
2 /Dr ) sin 2 2
rD n

n=1
Eq. (93) is the expression for the characteristic length of a rock
(81)
block.

The average block temperature, T̃b , can be obtained from Eq. References
(82):
 Abramowitz, M., Stegun, I.A., 1972. Handbook of Mathematical Functions. Dover,
1 New York.
T̃b (xf , t) = Tb (xb , t, xf )dV (82) Ascencio, C.F., Rivera, J., 1994. Heat transfer processes during low or high enthalpy
Vb Vb fluid injection into naturally fractured reservoir. In: Proceedings 19th Annual
Workshop on Geothermal Reservoir Engineering, Stanford University, Stanford,
Substituting Eq. (81) into Eq. (82), the dimensionless evaluation CA, USA, pp. 81–87.
Ascencio, C.F., Rivera, R.J., Samaniego, V.F., 2000. Thermal response during cold water
of the integral yields:
reinjection in naturally fractured geothermal reservoirs. In: World Geothermal

6  1
∞ Congress 2000, Kyushu-Tohoku, Japan.
  Avdonin, N. A., 1964. Some formulas for calculating the temperature field os a
T̃Db1 = 1 − exp −n2
2 tD /(ωr
2 /Dr ) (83) stratum subject to thermal injection. Izvestia VUZ. Neft’i Gaz 3, 37-41.

2 n2
n=1 Axelsson, G., Dong, Z.L., 1998. The Tanggu geothermal reservoir (Tian-
jin, China). Geothermics 27 (3), 271–294, http://dx.doi.org/10.1016/
For sufficiently large times, where tD > (ωr
2 /
2 Dr ), all terms S0375-6505(98)00002-9.
Axelsson, G., Flovenz, A.G., Hauksdottir, S., Hjartarson, A., Liu, J.R., 2001. Analysis
but the first one (n = 1) in the series expansion of Eq. (83) become of tracer test data, and injection-induced cooling, in the Laugaland geother-
negligible; thus: mal field, N-Iceland. Geothermics 30 (6), 697–725, http://dx.doi.org/10.1016/
S0375-6505(01)00026-8.
6   Bachu, S., Dagan, G., 1979. Stability of displacement of a cold fluid by a hot fluid in a
T̃Db1 ≈ 1 − exp −
2 tD /(ωr
2 /Dr ) (84) porous media. Physics of Fluids 22 (1), 54–59.

2
Bodvarsson, G.S., 1974. Geothermal resource energetics. Geothermics 3 (3), 83–92,
Differentiating Eq. (84) with respect to tD , and eliminating tD http://dx.doi.org/10.1016/0375-6505(74)90001-7.
Bodvarsson, G.S., Tsang, C.F., 1982. Injection and thermal breakthrough in fractured
from the resulting expression gives:
geothermal reservoirs. Journal of Geophysical Research 87 (B2), 1031–1048,
http://dx.doi.org/10.1029/JB087iB02p01031.
∂T̃Db1
2 Bodvarsson, G.S., Stefansson, V., 1989. Some theoretical and field aspects of rein-
=− (T̃Db1 − 1) (85)
∂tD (ωr
2 /Dr ) jection in geothermal reservoirs. Water Resources Research 15 (6), 1235–1248,
http://dx.doi.org/10.1029/WR025i006p01235.
Multiplying both sides of Eq. (85) by ωr yields: Boyadjiev, L., Kamenov, O., Kalla, S., 2005. On the Lauwerier formula-
tion of the temperature field problem in oil strata. International Jour-
∂T̃Db1
2 nal of Mathematics and Mathematical Sciences 10 (2005), 1577–1588,
ωr =− 2 (T̃Db1 − 1) (86) http://dx.doi.org/10.1155/IJMMS. 2005.1577.
∂tD (
/Dr ) Carslaw, H.S., Jaeger, J.C., 1959. Conduction of Heat in Solids, second ed. Oxford
University Press, London.
 F. Ascencio et al. / Geothermics 50 (2014) 112–121
Chen, C.-S., Reddell, D.L., 1983. Temperature distribution around a well
during thermal injection and a graphical technique for evaluating
aquifer thermal properties. Water Resources Research 19 (2), 351–363,
http://dx.doi.org/10.1029/WR019i002p00351.
Gevrek, A.I., 2000. Water-rock interaction in the Kizilcahamam Geothermal Field,
Galatian Volcanic Province (Turkey): a modelling study of a geothermal system
for reinjection well locations. Journal of Volcanology and Geothermal Research
96 (3–4), 207–213, http://dx.doi.org/10.1016/S0377-0273(99)00149-3.
Goyal, K.P., 1999. Injection related cooling in the Unit 13 area of
the Southeast Geysers, California, USA. Geothermics 28 (1), 3–19,
http://dx.doi.org/10.1016/S0375-6505(98)00042-X.
Gringarten, A.C., Sauty, J.P., 1975. A theoretical study of heat extraction from aquifers
with uniform regional flow. Journal of Geophysical Research 80 (5), 4956–4962,
http://dx.doi.org/10.1029/JB080i035p04956.
Gringarten, A.C., Witherspoon, P.A., Ohnishi, Y., 1975. Theory of heat extraction
from fractured hot dry rock. Journal of Geophysical Research 80 (8), 1120–1124,
http://dx.doi.org/10.1029/JB080i008p01120.
Hadidi, T.R., Nielsen, R.F., Calhoun, I.C., 1956. Studies on heat transfer during a linear
fluid displacement in a porous media. Producer Monthly 20 (10), 38–47.
Horne, R.N., 1982a. Effects of Water Injection into Fractured Geothermal Reservoirs:
A Summary of Experience Worldwide. Special Report 12. Geothermal Resources
Council, Davis, CA.
Horne, R.N., 1982b. Geothermal reinjection in Japan. Journal of Petroleum Technol-
ogy 34 (3), 495–503, http://dx.doi.org/10.2118/9925-PA.
121
Horne, R.N., 1985. Reservoir engineering aspects of reinjection. Geothermics 14
(2–3), 449–457, http://dx.doi.org/10.1016/0375-6505(85)90082-3.
Lauwerier, H.A., 1955. The transport of heat in an oil layer caused by
the injection of hot fluid. Applied Scientific Research 5 (2–3), 145–150,
http://dx.doi.org/10.1007/BF03184614.
Luke, Y.L., 1962. Integrals of Bessel Functions. McGraw-Hill, New York, USA.
Malofeev, G.E., 1960. Calculation of the temperature distribution in a formation
when pumping hot fluid into a well. Neft’i Gaz 3 (7), 59–64 (in Russian).
Pruess, K., Bodvarsson, G.S., 1984. Thermal effects of reinjection in a geothermal
reservoir with major vertical fractures. Journal of Petroleum Technology 36 (9),
1567–1578, http://dx.doi.org/10.2118/12099-PA.
Schroeder, R.C., O’Sullivan, M.J., Pruess, K., Celati, R., Ruffilli, C., 1982. Rein-
jection studies of vapor-dominated systems. Geothermics 11 (2), 93–120,
http://dx.doi.org/10.1016/0375-6505(82)90011-6.
Shaw-Yang, Y., Hund-Der, Y., 2008. An analytical solution for modeling thermal
energy transfer in a confined aquifer system. Hydrogeology Journal 16 (8),
1507–1515, http://dx.doi.org/10.1007/s10040-008-0327-9.
Stefansson, V., 1997. Geothermal reinjection experience. Geothermics 26 (1),
99–139, http://dx.doi.org/10.1016/S0375-6505(96)00035-1.
Stehfest, H., 1970. Numerical inversion of Laplace transform. Communications of the
ACM 13 (1), 47–49.
Warren, J.E., Root, P.J., 1963. The behavior of naturally fractured reservoirs. SPE
Journal (3), 245–255, http://dx.doi.org/10.2118/426-PA.