You are on page 1of 6

Journal of Earth Science, Vol. 26, No. 1, p.

047052, February 2015


Printed in China
DOI: 10.1007/s12583-015-0518-y

ISSN 1674-487X

Coupling of Thermal-Hydraulic-Mechanical
Processes for Geothermal Reservoir Modelling
Ali Karrech*l, Oussama Beltaief2, Ruyan Vincec1, Thomas Poulet3, Klaus Regenauer-Lieb3, 4, 5
1. School of Civil and Resource Engineering, The University of Western Australia, Crawley WA 6009, Australia
2. Tunisia Polytechnic School, Rue Alkhawarizmi, La Marsa, Tunisia
3. Earth Science and Resource Engineering, CSIRO, Kensington WA 6151, Australia
4. School of Petroleum Engineering, University of New South Wales, Sydney NSW 2052, Australia
5. School of Earth and Environment, The University of Western Australia, Perth WA 6000, Australia
ABSTRACT: This paper uses a fully coupled framework of thermal-hydraulic-mechanical processes to
investigate how the injection and extraction of fluid within a geothermal reservoir impacts on the distributions of temperature, pore pressure, and deformation within the rock formations. Based on this
formulation, a numerical model is developed in light of the thermodynamics of porous materials. The
proposed procedure relies on the derivation of dissipative flow rules by postulating proper storage and
dissipation functions. This approach opens new horizons for several resource engineering applications.
Since it allows for full coupling, this formulation can play a key role in predicting risks when used for
reservoir simulation. The results indicate that the injection-extraction process and temperature change
have a definite impact on altering the in-situ properties of the reservoir.
KEY WORDS: poro-mechanics, resource engineering, fluid injection and extraction, temperature
change, pore pressure, stress, deformation, uplift, subsidence.
0

INTRODUCTION
As the population of the world is set to double by the end
of this century, sustaining our current lifestyle requires the
production of clean energy at affordable prices. While fossil
fuels are still expected to be important energy resources for the
next decades, the share of renewable energies is expected to
increase significantly. The need to increase the supply of renewable energy sources has led to an increasing amount of
research on harvesting geothermal energy. The geothermal
energy process involves the utilisation of the Earths natural
geothermal gradient to extract heat and transform it into a directly useful energy such as electricity. To maintain recoverability and limit the footprint of geothermal energy, the process
involves injecting cold fluids into an underground reservoir of
hot permeable rock, allowing the fluid to flow through the rock
formation and then extracting the heated fluid. The injection
and extraction of fluids and subsequent temperature change
throughout the reservoir impacts the physical properties of the
permeable formations. The basic mechanism underlying the
geo-mechanical response of a reservoir is related to the magnitude of the pore pressure and temperature variations. Deep
underground reservoirs that have low values of porosity, permeability and compressibility result in greater pressure
*Corresponding author: ali.karrech@uwa.edu.au
China University of Geosciences and Springer-Verlag Berlin
Heidelberg 2015
Manuscript received August 13, 2014.
Manuscript accepted October 21, 2014.

changes, generating large surface deformation responses along


the top of a reservoir. As fluid is injected, the pore pressure
increases above its initial state, decreasing the effective stress
in the ground. The expansion that occurs due to fluid injection
is primarily caused by the vertical effective stress release due to
the pore pressure increase. There is also the need to consider
the thermal effect on the deformation in a reservoir. If the temperature of the injected fluid is lower than the in situ temperature, the poro-elastic and thermal effects on stresses and deformations tend to counteract and cancel each other out.
From the sites investigated by Teatini et al. (2011), two
major conclusions where determined: (i) the observed land
uplift may vary from a few millimetres to tens of centimetres
over a time interval of some months to several years according
to the specific location and (ii) the largest uplift depends on a
number of factors, including the fluid pore pressure increase,
the depth, thickness and areal extent of the pressurized and
heated geological formation, and the hydro-geo-thermomechanical properties of the porous medium involved in the
process. Teatini et al. show that if the fluid injection in a reservoir occurs at a deep enough level, the uplift and subsidence is
expected to be uniform between the injection and extraction
wells. If the injection occurs at a shallow level, the surface
uplift and subsidence may be irregular due to the influence of
the reservoir boundaries. Similarly, Chen (2012) investigated
the induced stresses and deformations in hydrocarbon and geothermal reservoirs. The purpose of his study was to investigate
deformation-based monitoring techniques. This involves using
surface deformation, such as tiltmeter monitoring, to control
hydraulic fractures, subsurface fluid migration and heat trans-

Karrech, A., Beltaief, O., Vincec, R., et al., 2015. Coupling of Thermal-Hydraulic-Mechanical Processes for Geothermal Reservoir
Modelling. Journal of Earth Science, 26(1): 4752. doi:10.1007/s12583-015-0518-y

Ali Karrech, Oussama Beltaief, Ruyan Vincec, Thomas Poulet and Klaus Regenauer-Lieb

48

port as fluid is injected or extracted from a reservoir. Chen used


a numerical model of an idealised ellipsoidal poro-elastic reservoir embedded in a surrounding rock body at a specified
depth to study the effects of fluid injection. The numerical
model predicted the expansion and/or con-traction of a reservoir due to pressure change. The pore pressure, fluid mass content and temperature within the reservoir were allowed to
change over time, while the same values in the surrounding
rock were fixed. He also included analytical solutions for the
induced stresses and deformations obtained using the theory of
inclusion and inhomogeneity. Chen concluded that both the
shape and mechanical properties of a reservoir can have a significant effect on any resulting deformation that occurs due to
fluid injection given that the reservoir and surrounding media
are homogenous. Due to the elliptical reservoir geometry, the
surface tilt vectors that occurred due to fluid injection were in a
radial pattern, pointing outwards from the centre of the reservoir and corresponding fluid injection point.
The usefulness of coupled thermal-hydraulic-mechanical
processes for poro-materials goes beyond the modelling of
geothermal reservoirs as it is relevant for several resource engineering applications such as mineral exploration and mining,
nuclear waste storage, and carbon dioxide storage. Most of
these applications involve large deformation of porous materials and/or large temperature variations. Considerable progress
has been made in the last decades to develop such a framework
based on the pioneering work of Biot (1972) and the comprehensive formulations of Coussy (2003). The main purpose of
this paper is to apply a simplified version of the developments
of Karrech et al. (2012) to geothermal reservoir modelling and
predict their responses under fluid injection and extraction.
1

GOVERNING EQUATIONS
The idea of describing processes in a coupled manner assumes that energy is smooth in the space of state variables
(Karrech et al., 2012). The smoothness of energy potentials is a
classifying but powerful assumption. Along with Schwartz
theorem, this assumption replaces many particular hypotheses
such as the Onsager and Maxwell principles as will be shown at
a later stage. State variables are selected to cover the different
processes that govern the behaviour of a mechanical system.
These variables should be, a priori, independent and enough to
control the system and determine its evolution with respect to
time. In the current framework, we express Helmholtz free
energy in terms of the common measure of deformation, ,
porosity, , and temperature, T, as follows

s s( , , )

(1)

Using the second law of thermodynamics and the independence of processes, it can be shown that
s
s
s

,
, and pf

(2)

where is the Cauchy stress, is the entropy, pf is the pore


pressure, and is the density of the solid-fluid mixture. The
subscripts f and s denote the fluid and solid phases. The
above relations can be expressed as follows: ds=:
ddT+pfd. Instead of using the free energy s, it is cus-

tomary in poromechanics to consider its dual which can be


obtained by Legendre transformation with regard to as follows: =s(pfp0f). This transformation shows that

,
, and

( pf p0 f )

(3)

The free energy function is assumed regular with respect to its variables. In particular, its mixed derivatives exist
and they are continuous. Therefore, differentiation of the different terms in equation (3) and application the Schwartz
theorem results in the following incremental relationships
2
d ( G )d1 2Gd 3KdT 1 bdpf 1
3

d 3d C

d bd 3 dT

3 dpf

(4)

dp f
N

where =tr, K is the bulk modulus of the porous material and


G is its shear modulus, is the thermal expansion coefficient of
the porous material, b is the Biot coefficient, C is the heat capacity of the porous material at constant volume, is the latent
heat coefficient associated with the change of porosity, and N is
the Biot modulus. Note that the isochoric and isobaric heat
capacities are duals. The relationship between them involves
temperature as follows: Cp=C+9T2K. In addition, the following thermo-elastic relationships can be deduced

1
1
1 b
,

,
s (1 ) s 1
C p (1 ) C ps , (b ), s
b 1

(5)

where =/J is the porosity in the current configuration, J is


the determinant of the gradient of deformation, which is assumed to be close to unity. As explained in details by Karrech
et al. (2011) the assumption of local equilibrium as well as the
second law of thermodynamics deliver the following expression for dissipation

D Vf ( pf f g ) J

(6)

where Vf=(fs) is the infiltration vector, f is the fluid velocity, s is the solid velocity and J is a heat flux, g is a body
force, and f is the density of the pore fluid. The right hand
terms represent the hydraulic and thermal dissipations, respectively. In this study, we maximise the dissipation associated
with the hydrothermal behaviour and use the theorem of
Schwartz (if h: n R has continuous second partial
derivatives at a point (p 1 ,,p n ), then i , j 1n,
2h
2h
( p1, , pn )
( p1 , , pn ) ) to show that
i j
j i
J Lqf ( pf f g ) Lqq

and Vf L ff ( pf f g ) L fq

(7)

Coupling of Thermal-Hydraulic-Mechanical Processes for Geothermal Reservoir Modelling


In the above equation Lij are symmetry terms which are
similar to the Onsager coefficients (Onsager, 1931a, b). For
simplification we consider Lij=0 if i=j, in this study. This assumption allows us to use the classical Darcy and Fourier laws
without unnecessary complexity. When included in the local
equations of conservation of mass, momenta, and energy the
above fluxes deliver the following partial differential equations
that govern the behaviour of the fully coupled system. We describe first the equation of motion
g 0, n d at , u u d at u

(8)

where Td is a specified stress vector and ud is a specified displace geo-infiltration problems involve small fluid velocities,
pf
the terms Vf
3 f Vf can be neglected. In the above

equations denotes the surface boundaries corresponding to


the respective Dirichlet and Neumann conditions. Therefore,
the partial differential equation of hydraulic flow in porous
media can be obtained

ence temperature and pressure, respectively. In addition the


permeability varies with respect to porosity in accordance with
2

Carman-Kozeny expression: ff ff 0

1 0

. The per0 1

meability is related to the hydraulic conductivity via the rela f g ff


tionship
. In this study the permeability is as-

(9)

where , n pf qf at qf , pf pd at p

Table l

where 1/M=1/N+/Kf, N is the Biot modulus, Kf is the fluid


bulk modulus, m=+f, f is the fluid thermal expansion,

ff
L ff
, Kff is the intrinsic permeability, f is the fluid
f

dynamic viscosity, qf is a specified pore pressure gradient, and


pd is a specified pore pressure. Similarly, the equation of heat
transfer reads
d
Cf Vf
dt
where , n qq at qq , T Td at T

The model has dimensions of 20 ml 000 ml 000 m. The


width and height of the model were selected large enough as
they are the primary directions, which allow for fluid flow and
temperature variation to occur throughout the model without
the edges of the model affecting the results. The thickness of
the model is small in comparison to the other dimension to
allow for shorter simulation times. Table l summarises the parameters of simulation: The density of water considered in this
study varies with respect to temperature and pressure as:

0
p p0
1 f
, where T0 and 0 are referf
f
1 f ( 0 )

sumed to be orthotropic and the ratio of vertical permeability


over horizontal permeability is of 1/3.

ff
1 dp f
d
d
( pf f g )
b
3 m

dt
dt
ff
dt

(C p Cf )

49

(10)

Material properties used for simulations

Property

Solid

Fluid

Density (kg/m )

s=2 600

f(T,pf)|(0,0)=1 000

Bulk modulus (GPa)

Ks=6.89

Kf=2.2

Cps=2 500

Cf=4 193

s=2.5

f=0.6

s=1.510-6

f=610-4

=0.25

Specific heat (J/(kgK))


Thermal conductivity
(W/(mK))
Thermal expansion
(K-1)
Poisson coefficient

Reservoir surface, z = T d

where Lqq /0 is the thermal conductivity, qq is a specified temperature gradient, and Td is a specified temperature.
2

NUMERICAL MODEL
From the equations of conservation of mass, energy and
momenta described previously, we formulate the quasi-static
coupled thermo-hydro-mechanical problem. As the governing
equations are difficult to solve analytically, we develop a numerical model to determine the impact of fluid flow and temperature change on the initial conditions of a geothermal reservoir. The numerical model calculates the changes in temperature, pore pressure, stress and deformations that occur over
time. The model idealises a geothermal reservoir containing
porous sedimentary rock that is fully saturated. Fluid is injected
into one side of the rock and extracted at the same rate on the
opposite side of the model. The general purpose commercial
Finite Element tool Abaqus (Anonynous, 2011) is used for
simulation. The coupling of temperature, fluid flow, and temperature is described by dimensional element C3D8PT, which
contains 8 degrees of freedom, including pore pressure and
temperature, allowing for variations in these areas to be simulated, and the resulting stress and deformation to be calculated.

, y T = qq
, y pf = qf

z
Injection

y Extraction

, y T = qq
, y pf = qf

u y =0

u y =0

u z =0

Figure l. Fluid flow path and boundary conditions.

The 3D block used for simulation is subjected to sliding


boundary conditions restricting its normal motion at all sides
except the top. This allows the reservoir to expand and contract
vertically as its motion is restricted due to adjacent rock masses
delimiting the rest of its periphery. Out of the injection zones,
Dirichlet boundary conditions are applied in terms of temperature and pore pressure in accordance with the initial gradients.
In other words, temperature and pore pressure at these bounda-

50

Ali Karrech, Oussama Beltaief, Ruyan Vincec, Thomas Poulet and Klaus Regenauer-Lieb

ries remain unchanged. The magnitudes of the initial gradients


of temperatures and pore pressure are respectively 0.03 Cm-1
and f g. The centre of the model is initially positioned
at 2.17 km, below the Earths surface. The initial vertical
and horizontal stresses are respectively expressed by
1 667 s g ( z 1 667)((1 ) s f ) and h=0.55. The
loading process comprises the injection and extraction of fluids
at a rate of 50 L/s across 40 m span at the centre of the left and
right hand sides, respectively. The injection area is characterised by an imposed temperature of 50 C.
Figure 2a shows the temperature distribution along the
centreline from the injection area to the extraction area, for
different time instants. Figure 2b shows the evolution of temperature in the vicinity of the extraction point. The initial temperature of the injection-extraction path is at 90 C. It starts
decreasing progressively due to diffusion and advection effects.
The profile of temperature represents a front like pattern which
progresses towards the extraction point. The front disperses
with respect to time due to diffusion. It is worthwhile noticing
that temperature decreases slightly at the extraction point due to
the heat flux induced by the extraction process. This phenomenon is affected by the gravity, which enhances the cooling
process at the extraction point.

Figure 3a shows the spatial distribution of pore pressure


along the injection-extraction path at different instants of time.
The results show that the problem is not symmetric at least due
to the effect of coupling and exchanged feedbacks. In the absence of thermal feedback, pore pressure stabilises within one
day. In the vicinity of the injection area the size of the cold
zone increases progressively and affects the distribution of pore
pressure. This behaviour is explained by the following coupled
incremental equation: dpf/M=dmf/fbd+3mdT It can be seen
that at constant water content and deformation, the pore pressure increases with temperature. Figure 3b shows the evolution
of pore pressure with respect to time in the vicinity of the injection and extraction points. There are three distinguishable phenomena that can be reported. The first is the opposite overall
increase of pore pressure at the injection area and decrease at
the extraction area by around 4 MPa. This behaviour is mainly
attributed to the variation of the fluid content dmf/f. The second component is the small decrease of pore pressure observed
at the injection point at around t=5 000 s. This variation is
mainly explained by variation of the local volume of the solid
skeleton bd. The third phenomenon acts progressively on the
injection and extraction area. It reflects the variation of temperature with respect to time 3mdT.

Figure 2. (a) Spatial distribution of temperature along the centreline between the injection and extraction areas, (b) temporal
evolution of temperature at the vicinity of the extraction area.

Figures 4a and 4b show respectively the spatial distribution of the porosity along the injection-extraction path in different instants and the evolution of porosity with respect to time
at the vicinities of the injection and extraction points. Initially,
the porosity is selected equal to 0.3, but it reduces to 0.292 3
due to geostatic loading. As for pore pressure, porosity varies in
dp f
accordance with dt
bd 3 s d . It increases when
N
pore pressure increases, the skeleton expands and/or temperature decreases. As shown in Figs. 3 and 4, the behaviours of
pore pressure and porosity with respect to time and space are
similar since they are thermodynamic duals.
Figure 5a shows the distribution of uplift at the surface of
the model with respect to time. After 1 day of injection, there is
a positive vertical displacement (uplift) occurring on the injection side of the reservoir and a negative vertical displacement
occurring on the extraction side of the reservoir (subsidence).
This initial displacement is due to the increase and decrease in
pore pressure that occurs after 1 day due to the injection-

extraction process. We can see that the deformation reaches a


maximum after at this particular stage and initial deformation is
equal and opposite on either side of the reservoir surface, which
is the same as the initial change in pore pressure. After 1 year,
the effect of temperature change in the reservoir begins to impact the vertical displacement. We see that the uplift near the
extraction side decreases after 1 year. This is due to the temperature decreasing near the injection point, causing the reservoir to contract. After 10 years, there is a further decrease in the
uplift on the injection side of the reservoir, as there is only a
significant change in temperature on the left side of the model.
There is subsequently no change in subsidence between the 1
and 10 years mark because the temperature does not change
significantly within this period. Figure 5b shows the effect of
the permeability on the expansion of the rock mass. Changing
the permeability of the reservoir model affects the ease to
which a fluid can move through the pore spaces. The values of
permeability used for this analysis are in the typical range of
sedimentary rocks, being 0.25, 0.5 (as seen in the reference set of

Coupling of Thermal-Hydraulic-Mechanical Processes for Geothermal Reservoir Modelling

51

Figure 3. (a) Spatial distribution of pore pressure along the centreline between the injection and extraction areas, (b) temporal evolution of pore pressure at the vicinity of the injection and extraction areas.

Figure 4. (a) Spatial distribution of porosity along the centreline between the injection and extraction areas, (b) temporal
evolution of porosity at the vicinity of the injection and extraction areas.

Figure 5. (a) Uplift along the surface of the model at different time instants, (b) impact of permeability on uplift.

results), 0.75, 1 and 1.25 Darcy. It can be seen that the lower the
permeability the higher the induced deformation of the reservoir.
3

CONCLUSION
This paper introduces the effects of coupled thermalhydraulic-mechanical processes on the response of an idealised
geothermal reservoir. The model shows that the temperature
change throughout the reservoir is extremely important as its
affects the in-situ properties of the reservoir. Through the sensitivity analysis we determine that the permeability of the reservoir
does not have a significant impact on temperature change, but it
affects significantly the pore pressure distribution. The results
show that the pore pressure changes to a new equilibrium level
immediately after the injection and extraction of fluid begins. As
time progresses and the reservoir changes in temperature so does
the pore pressure. Decreasing the permeability or increasing the

injection rate in the model results in the initial increase of pore


pressure. These changes in temperatures and pressures affect the
stress state directly through coupling terms such as the expansion
and Biots coefficients, respectively.
The deformation in the reservoir is affected by the pore
pressure and temperature. On the injection side, the pore pressure initially increases and induces a change in stress, which
corresponds to uplift. As the temperature decreases over time,
the subsequent increase in pore pressure and stress causes expansion in the reservoir. The temperature decrease also results
in the reservoir contracting. On the extraction side of the reservoir, subsidence occurs due to a negative change in pore pressure and stress. As temperature decreases, these values increase
to a larger negative amount, and its contraction is added to the
contraction caused by a temperature decrease. Changing the
permeability and injection rate results in the deformation vary-

52

Ali Karrech, Oussama Beltaief, Ruyan Vincec, Thomas Poulet and Klaus Regenauer-Lieb

ing drastically over time.


This investigation focuses on specific parameters and
outputs to identify how pore pressure changes results in deformations occurring. In order to gain a complete understanding of
the behaviour of geothermal reservoirs, it would be highly
recommended to further research the effect of changing other
parameters. These parameters would include those that affect
the rate at which the temperature decreases throughout the reservoir over time, other parameters affecting the pore pressure
and if it is possible to change the in-situ stress without changing the pore pressure. A more comprehensive analysis of the
expansion and contraction of the reservoir due to temperature
change would also be beneficial. Combining all of these different investigations could potentially allow for the surface
deformation in reservoirs to be reduced while still enabling the
reservoir to produce a fluid that can economically produce
energy.
REFERENCES CITED
Anonynous, 2011. Abaqus User Manual Version 6.11. Hibbitt,
Karlson and Sorensen Inc., USA

Biot, M., 1972. Theory of Finite Deformations of Porous Solids.


Indiana University Mathematics Journal, 21(7): 597620
Chen, Z. R., 2012. Poroelastic Model for Induced Stresses and
Deformations in Hydrocarbon and Geothermal Reservoirs.
Journal of Petroleum, Science and Engineering, 80: 4152
Coussy, O., 2003. Poromechanics. Wiley, Chichester. 312
Karrech, A., Poulet, T., Regenauer-Lieb, K., 2012. Poromechanics of Saturated Media Based on the Logarithmic Finite Strain. Mechanics of Materials, 51: 118136
Karrech, A., Regenauer-Lieb, K., Poulet, T., 2011. Frame Indifferent Elastoplasticity of Frictional Materials at Finite
Strain. International Journal of Solids and Structures,
48(34): 397407
Onsager, L., 1931a. Reciprocal Relations in Irreversible Processes i. Physical Review, 37: 405426
Onsager, L., 1931b. Reciprocal Relations in Irreversible Processes ii. Physical Review, 38: 22652279
Teatini, P., Gambolati, G., Ferronato, M., et al., 2011. Land
Uplift due to Subsurface Fluid Injection. Journal of Geodynamics, 51: 116

You might also like