Professional Documents
Culture Documents
Summary
This paper presents a discussion of the issues related to the interaction between rock deformation and multiphase fluid flow behavior
in hydrocarbon reservoirs. Pore-pressure and temperature changes
resulting from production and fluid injection can induce rock deformations, which should be accounted for in reservoir modeling.
Deformation can affect the permeability and pore compressibility of
the reservoir rock. In turn, the pore pressures will vary owing to
changes in the pore volume. This paper presents the formulation of
Biots equations for multiphase fluid flow in deformable porous
media. Based on this formulation, it is argued that rock deformation
and multiphase fluid flow are fully coupled processes that should be
accounted for simultaneously, and can only be decoupled for predefined simple loading conditions. In general, it is shown that reservoir simulators neglect or simplify important geomechanical
aspects that can impact reservoir productivity. This is attributed to
the fact that the only rock mechanical parameter involved in reservoir simulations is pore compressibility. This parameter is shown to
be insufficient in representing aspects of rock behavior such as
stress-path dependency and dilatancy, which require a full tensorial
constitutive relation. Furthermore, the pore-pressure changes
caused by the applied loads from nonpay rock and the influence of
nonpay rock on reservoir deformability cannot be accounted for
simply by adjusting the pore compressibility.
Introduction
In the last two decades, there has been a strong emphasis on the
importance of geomechanics in several petroleum engineering
activities such as drilling, borehole stability, hydraulic fracturing,
and production-induced compaction and subsidence. In these
areas, in-situ stresses and rock deformations, in addition to fluidflow behavior, are key parameters. The interaction between geomechanics and multiphase fluid flow is widely recognized in
hydraulic fracturing. For instance, Advani et al.1 and Settari et al.2
have shown the importance of fracture-induced in-situ stress
changes and deformations on reservoir behavior and how hydraulic
fracturing can be coupled with reservoir simulators. However, in
other applications, geomechanics, if not entirely neglected, is still
treated as a separate aspect from multiphase fluid flow. By treating
the two fields as separate issues, the tendency for each field is to
simplify and make approximate assumptions for the other field. This
is expected because of the complexity of treating geomechanics and
multiphase fluid flow as coupled processes.
Recently, there has been a growing interest in the importance of
geomechanics in reservoir simulation, particularly in the case of
heavy oil or bituminous sand reservoirs,3,4 water injection in fractured and heterogeneous reservoirs,5-7 and compacting and subsiding fields.8,9 Several approaches have been proposed to implement geomechanical effects into reservoir simulation. The
approaches differ on the elements of geomechanics that should be
implemented and the degree to which these elements are coupled
to multiphase fluid flow.
The objective of this paper is to illustrate the importance of geomechanics on multiphase flow behavior in hydrocarbon reservoirs.
An extension of Biots theory10 for 3D consolidation in porous
media to multiphase fluids, which was proposed by Lewis and
Copyright 2001 Society of Petroleum Engineers
This paper (SPE 72095) was revised for publication from paper SPE 50636, first presented
at the 1998 SPE European Petroleum Conference, The Hague, 2022 October. Original
manuscript received for review 15 February 1999. Revised manuscript received 11 April
2001. Paper peer approved 16 April 2001.
164
Terzaghis 1D consolidation theory has been used widely in settlement problems in saturated soils. Biot10 extended the theory into a
more general 3D case, based on a linear stress-strain relation and a
single-phase fluid flow. Here, we present an extension of Biots
equations for two-phase immiscible and isothermal flow.
Equations for three-phase flow can be found in Lewis and
Sukirman.11 In the following, tensorial notation is used and summation is implied for repeated indices.
For two-phase fluid flow, the generalized Darcys law is given as
HH
kij kr
v i =
p + gh .
x j
Fluid Pressure
(Temperature)
In-Situ Stresses
. . . . . . . . . . . . . . . . . . . . . . (1)
Rock Deformation
Permeability
; . . . . . . . . . . . . . . . . . . . . . . . . . . . . (4)
1
1
vi =
S
xi B
t B
HH
ij Dijkl kl
So
+ o kl
B
3K s t
HH
So 1 ij Dijklkl
+ o
2
B K s
( 3K s )
p + q .
. . . . . . . . . . . . . . . . . . . . (6)
1
S ; . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (7)
t B
(1 ) p
; . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (9)
K s t
and (d) the rate of solid-particle-volume change caused by the
change in mean effective stress,
HH
ij ij
3K s t
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (10)
xi
1
1
vi = S + q .
B
t
B
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (12)
k r = k r ( S o , S w ) ,
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (8)
. . . . . . . . . . . . . . . . . . (11)
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (13)
Pc = po pw = pc ( So , S w ) ,
. . . . . . . . . . . . . . . . . . . . . . . . . (14)
165
and So + S w = 1 .
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (15)
S
B
1 1
1 S p
.
+
B
+
p B p S p t
. . . . . . . . . . . . . . . . . . . . . . (16)
The first term on the right side of Eq. 16 is the change in the
volume factor Bp with pressure, giving the fluid compressibility of
phase p as
1
1
.
= cf =
p
p B
then the second term on the right side of Eq. 16 may be rewritten as
. . . . . . . . . . . . . . . . . . . . . . . . (19)
cp =
1
.
p
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (20)
S
1 S p
= c f + S c p +
+ q .
B
S p t
. . . . . . . . . . . . . . (21)
3K s
x j
kl
t
HH
1 ij Dijklkl
+
+
2
Kf
Ks
( 3K s )
p + q ,
t
. . . . . . . . . . . . . . . . . . . . . . (22)
where K f =1/cf is the bulk modulus of the fluid. This equation
should be compared to the single-phase version of the fluid diffusivity equation (Eq. 21) used in reservoir simulation,
166
. . . . . . . . . . . . . (24)
where K and G=the shear and bulk moduli, respectively, and are
related to the Youngs modulus E and Poissons ratio n as
K=
E
3 (1 2 )
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (25a)
E
.
2 (1 + )
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (25b)
Substituting Eq. 24 into Eq. 3 and the resulting equation into Eq. 2
yields the poroelastic
relates the total
stress-strain relation, which
stress increment dsij to the strain increment deij and pore-pressure
increment dp.
HH
HH
2
dij = K G dvij + 2Gdij dpij ,
3
. . . . . . . . . . . . . (26)
=1
kij kr
p + gh
x j
xi
and G =
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (18)
1 1 p
=
= S c p ,
p p p
HH
2
Dijkl = K G ijkl + G ik jl + il jk ,
3
. . . . . . . . . . . . . . . . . . . . . . (17)
. . . . . . . . . . . . . . . . . . . . (23)
where ct=cf +cp is the total compressibility. Again, for the sake of
simplicity, elastic stress-strain behavior
will be considered, in
which case the constitutive tensor Dijkl equals
p
1
1
S =
S
p
B
t B
t
=
HH
kij
p
p + gh = ct
+q,
x j
t
K
.
Ks
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (27)
kij p
1 p
+ gh = v +
+ q , . . . . . . . . . . . . (28)
t
M
xi x j
B t
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . (29)
H
H
2
kij p
1 p
+ gh =
+
+ q . . . . . . . . . . . . . (31)
K M B t
xi x j
1 2
1
+
.
K MB
( ct )hydro =
. . . . . . . . . . . . . . . . . . . . . . . . . (32)
(c )
p
hydro
1 2
+
. . . . . . . . . . . . . . . . . . . . . . . . . (33)
Ks
K
(c )
p hydro
1
.
K
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (34)
d z = K + G dv dp = 0 .
3
. . . . . . . . . . . . . . . . . . . . (35)
1 p
+ gh =
+
+q .
M M B t
xi x j
. . . . . . . . . . . . (36)
( ct )oedo =
1 2
1
+
,
M MB
. . . . . . . . . . . . . . . . . . . . . . . . . (37)
E (1 )
4
. . . . . . . . . . . . . . . . . . . . . (38)
M =K+ G=
3
(1 2 )(1 + )
Again, neglecting fluid-compressibility and poroelastic effects
yields the pore-compressibility parameter for oedometric condition, which is
(c )
p oedo
1
.
M
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (39)
Consequences for Reservoir Simulation. The previous discussion has shown that the equations of poroelasticity can be reduced,
under specific local assumptions, to a hydraulic-diffusion-type
equation. Whereas the compressibility of a fluid can be considered
an intrinsic property under constant temperature, however, the pore
compressibility depends on the local conditions assumed. It was
shown that assuming local hydrostatic or oedometric conditions
gives two different values of pore compressibilities. The differences in the pore compressibilities for these two conditions can be
significant. It can be shown that the ratio of the two compressibility values is equal to
June 2001 SPE Reservoir Evaluation & Engineering
(c )
(c )
p hydro
p oedo
3 (1 )
1+
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (40)
6.5 km
Reservoir centerline
11 km
3 km
0.3 km
0.7 km
Hydrocarbon
reservoir
N
38.5 km
Fig. 2Rock displacements caused by a uniform pressure
drawdown in a disk-shaped reservoir.
. . . . . . . . . . . . (41)
where [Dp]=the matrix of pore-pressure change, [p]=the current pressure matrix, [C]=the compressibility matrix, [F]=the
permeability matrix, [Dq]=the matrix of fluid fluxes, and
Dt=the time increment. From this equation, the pore-pressure
changes can be solved as
p = C + t
) ( t p + q ) .
. . . . . . . . . . (42)
=
t
L t p t p + q
. . . . . . . (43)
p = L K L + t
t p + q + L K F . . . . . . . . . (44)
t
23 km
4 km
the reservoir and nonpay rock system, while the latter is a diagonal
matrix that can only be made dependent on the pore pressure.
Pore-pressure change is also a function of the applied load
[L]t[K]-1[DF], caused by the total stress changes in the fully coupled analysis. Such total stress changes come from the weight of
the overburden, which is transferred nonuniformly to the reservoir
again, according to the pore-pressure distribution in the reservoir.
Application to a Field Case
The extended Biot equations for three-phase fluid flow in
deformable porous media were discretized by Lewis and
Sukirman,11 and the discretized equations were implemented in the
finite-element code CORES (COupled REservoir Simulator).
CORES is a 3D black-oil (three-phase compressible and immiscible fluid flow) simulator. The reservoir rock is modeled by elastic
and/or elastoplastic constitutive models, and the physical properties of the fluids depend on fluid pressures and saturations. In the
finite-element implementation, implicit procedures are used to
solve the fully coupled governing equations where the rock displacements and fluid pressures are the primary unknowns.
To illustrate the importance of analyzing fluid-flow and geomechanical behavior as fully coupled processes, CORES is
applied to the simulation of an idealized North Sea reservoir. The
reservoir has an area of about 6.5 by 11 km and a thickness of 300
m. The complete model, which includes the nonpay rock, has an
area of about 23 by 38.5 km and a thickness of 4 km (of which 3
km is the overburden, 0.3 km the reservoir, and 0.7 km the underburden). A simplified view of the whole model is shown in Fig.
3. A rough mesh with 8 by 20 by 14 eight-noded brick elements
was used in the simulation. However, the results of the 3D model
were also verified by a 2D model with a much finer discretization. No vertical displacements are allowed at the base of the
model, and no lateral displacements are allowed at the four sides
of the model. It is a common practice in geomechanical modeling
to extend the lateral boundaries as far away as computationally
possible from the main loaded region to simulate infinitely horizontal boundaries and minimize local boundary effects. The top
of the model, which corresponds to the seabed, is allowed to
deform freely.
The initial reservoir pressure is assumed to be 48 MPa, which is
uniformly distributed in the reservoir. The initial effective vertical
stress distribution vs. depth is integrated from the self-weights of
June 2001 SPE Reservoir Evaluation & Engineering
2.5
0.45
0
Underburden
Youngs modulus, E (GPa)
Poisson's ratio,
Absolute permeability, k (md)
13.5
0.45
0
Reservoir
Stiff
Soft
0.85
0.25
35
0.0028
150
0.05
0.25
35
0.0476
150
* Used only for the uncoupled flow model and calculated from Eq. 39 with
the same E and used in the fully coupled models. For comparison,
cf =4.4104/MPa at initial reservoir conditions (used for both uncoupled
and fully coupled simulation).
the different rock layers, while the initial effective horizontal stresses
are assumed to be one-half of the total vertical stresses. Only the
water and oil phases are considered in the reservoir to simplify the
analysis. Undrained conditions (i.e., no fluid flow or zero permeability) are used for the surrounding nonpay rocks. Realistic relative
permeability and capillary pressure curves, and standard oil and
water formation volume factor curves, were used for the fluid-flow
part of the flow simulation. Elastic rock properties were used in the
geomechanical simulation. The rock properties used are given in
Table 1. The model is analyzed for a 14-year production scenario
by specifying production rates in the finite-element nodes corresponding to production wells within the reservoir. The production
rates applied in the production wells are based on recorded production data in the field.
The calculated pore-pressure distribution at the end of simulation in the top reservoir layer is shown in Fig. 4. After 14 years of
production, the reservoir pore pressure has been reduced to approximately 25 MPa in much of the reservoir. However, despite the
continuous production, the pore pressures have increased to
Reservoir pressure
50
Initial reservoir
pressure
40
30
Fully coupled modeling
Reservoir simulation
0
1000
2000
3000
Distance from production well, m
Initial reservoir
pressure
50
Reservoir pressure
40
30
1000
2000
3000
controlled by the interaction between the reservoir and the surrounding nonpay rock, and by the constitutive behavior of both the
reservoir and the nonpay rocks.
The increase in pore pressure above initial value during production is an important effect that cannot be predicted by existing
reservoir simulators. This increase in pore pressure is caused by the
load-term load [L]t[K]-1[DF] from the total stress changes and the
interaction of reservoir and overburden in the fully coupled analysis. The pore-pressure reduction resulted in the compaction of the
central part of the reservoir. In turn, the compaction resulted in the
downward movement and bending of the overburden, causing the
reservoir fluids to be squeezed and the pore pressures to increase
toward the reservoir flanks. This coupled response, which comes
from the interaction between the reservoir and the overburden, is a
complicated process. The deformation of the overburden is
dependent on the pore-pressure distribution in the reservoir; on the
other hand, the pore-pressure distribution in the reservoir is also
controlled by deformation of the overburden. This structural indeterminacy in rock deformation is one of the main reasons why it is
not always easy to decouple rock deformation from fluid flow.22
The increase in pore pressure above initial value during fluid
extraction is analogous to the so-called Mandel-Cryer effect
observed in one of the first applications of Biots 3D consolidation
170
theory. Cryer23 showed that on withdrawal of fluid in a consolidating layer of a fluid-saturated medium, the pore pressure instantaneously jumped, then continued to increase for some time, before
pressure dissipation commenced. The pore-pressure increase is
attributed to the downward movement of the layer above the consolidating layer and the increase in total stresses as the weight of
the layer above the consolidating layer is transferred to the fluids,
causing the pore pressure to increase.
The increase in pore pressure owing to rock deformation has
also been referred to as compaction drive in reservoir engineering.
In standard reservoir simulation, the main mechanism accounting
for the compaction drive is the pore-volume reduction of the
reservoir rock. In fully coupled simulation, the downward movement of the overburden also contributes to the compaction drive.
This contribution, particularly when the pore pressures increase
above the initial reservoir pressure, cannot be accounted for simply by adjusting the pore compressibility in reservoir simulations.
The compaction drive will be very pronounced for soft reservoirs,
but it can also be significant for the case of relative stiff reservoirs,
as shown in Fig. 6. Note that this increase is only to be expected
for reservoirs with low-permeability aquifers. Otherwise, the
increase in reservoir pressure from compaction drive will be dissipated into the aquifer. On the other hand, a compressible aquifer
might also contribute to the increase in pore pressure from the
compaction drive.
Several schemes have been proposed in the literature to couple
the stress-strain behavior of rock and multiphase fluid flow.3,4
Settari and Mourits,24 for instance, present an approach where the
porosity is used as a coupling parameter between a finite-element
stress-analysis code and a reservoir simulator. The geomechanical
and reservoir simulators are used in a staggered manner. Porepressure changes are calculated from the reservoir simulation and
converted to nodal loads. From these nodal loads, the in-situ stress
changes and rock displacements are calculated in the geomechanical simulation. An iterative algorithm is used to ensure that the
porosities calculated from the geomechanical simulator are the
same as those calculated from the reservoir simulator.
The iterative approach, however, does not rigorously address
the coupling of geomechanics and reservoir simulations. One possible drawback of such an iterative approach is that there appears
to be no proof that the approach will converge to a unique solution.
For instance, it is not clear whether the approach can be used in the
case where the rock tends to increase in volume with a reduction in
pore pressure (e.g., owing to dilation during shearing). Such a volume increase would require a negative pore compressibility in the
reservoir simulation and may cause numerical instability.
Conclusions
The issues related to the interaction between fluid flow and rock
deformation in reservoir simulation have been discussed in this
paper. A primary type of interaction concerns stress- induced permeability changes, which in turn affect the fluid-pressure distribution. This type of coupling is particularly important in fractured
and faulted reservoirs, where fracture- and fault-permeability
changes can be orders of magnitude greater than those of the bulk
matrix. Moreover, fracture- and fault-permeability changes can
also influence fluid-flow directionality and sweep efficiency.
A comparison of the governing equations used for reservoir
simulations and Biots theory for multiphase fluid flow in
deformable porous media was made. Based on this comparison and
on the results of a simple case study, it was shown that geomechanics and multiphase fluid flow in hydrocarbon reservoirs
should be analyzed as fully coupled processes. As fully coupled
processes, fluid flow and geomechanics form a set of separate
domains that cannot be analyzed separately. The dependent variables in each domain (e.g., fluid pressures and rock displacements)
cannot be eliminated explicitly, except for simple cases corresponding to specific stress paths. However, in general, the deformation
of the reservoir during depletion and recovery is a complicated
process for which a simple stress path cannot be assumed.
It is shown that reservoir simulators, by not taking these coupling
phenomena into consideration, simplify important geomechanical
June 2001 SPE Reservoir Evaluation & Engineering
z = vertical axis
p = fluid phase (o,w)
Acknowledgment
The assistance of Dr. Hamid Ghafouri in carrying out the fully
coupled modeling of the idealized North Sea reservoir is gratefully acknowledged.
References
1. Advani, S.H. et al.: Fluid Flow and Structural Response Modeling
Associated with the Mechanics of Hydraulic Fracturing, SPEFE (June
1986) 309.
2. Settari, A., Ito, Y., and Jha, K.N.: Coupling of a Fracture Mechanics
Model and a Thermal Reservoir Simulator for Tar Sands, J. Cdn. Pet.
Tech. (November 1992) 31, No. 9, 20.
3. Fung, L.S.K., Buchanan, L., and Wan, R.G.: Coupled GeomechanicalThermal Simulation of Deforming Heavy-Oil Reservoirs, paper 361 presented at the 1992 CIM Annual Technical Conference, Calgary, June 1992.
4. Tortike, W.S. and Ali, S.M.: Reservoir Simulation Integrated with
Geomechanics, paper 391 presented at the 1992 CIM Annual
Technical Conference., Calgary, June 1992.
5. Heffer, K.J. et al.: The Influence of Natural Fractures, Faults and Earth
Stresses on Reservoir PerformanceAnalysis by Numerical
Modelling, Proc., 3rd Intl. Conference on North Sea Oil and Gas
Reservoirs, Trondheim, Norway (December 1992) 201.
6. Koutsabeloulis, N.C., Heffer, K.J., and Wong, S.: Numerical
Geomechanics in Reservoir Engineering, Proc., Intl. Conference on
Computer Methods and Advances in Geomechanics (1994) 2097.
7. Gutierrez, M. and Makurat, A.: Coupled HTM Modelling of Cold
Water Injection in Fractured Hydrocarbon Reservoirs, Intl. J. Rock
Mech. Min. Sci. (1997) 34, 429.
8. Gutierrez, M.: Fully Coupled Analysis of Reservoir Compaction and
Subsidence, paper SPE 28900 presented at the 1994 SPE European
Petroleum Conference, London, 2527 October.
9. Lewis, R.W. and Sukirman, Y.: Finite Element Modelling for
Simulating the Surface Subsidence Above a Compacting Hydrocarbon
Reservoir, Intl. J. Num. Analy. Meth. Geomech. (1993) 18, No. 9, 618.
10. Biot, M.A: General Theory of Three-Dimensional Consolidation, J.
Appl. Phys. (1941) 12, 155.
11. Lewis, R.W. and Sukirman, Y.: Finite Element Modelling of ThreePhase Flow in Deforming Saturated Oil Reservoirs, Intl. J. Num.
Analy. Meth. Geomech. (1993) 17, No. 8, 577.
12. Terzaghi, K.: Erdbaumechanik auf Bodenphysikalicher Grundlage,
Franz Deutike, Vienna (1925).
13. Sornette, D., Davy, P., and Sornette, A.: Structuration of the
Lithosphere in Plate Tectonics as a Self-Organized Critical
Phenomenon, J. Geophys. Res. (1990) 95, No. 11, 353.
14. Chen, H.-Y., Teufel, L.W., and Lee, R.L.: Coupled Fluid Flow and
Geomechanics in Reservoir StudyI. Theory and Governing
Equations, paper SPE 30752 presented at the 1995 SPE Annual
Technical Conference and Exhibition, Dallas, 2225 October.
15. Chen, H.-Y. and Teufel, L.W.: Coupling Fluid-Flow and Geomechanics
in Dual-Porosity Modeling of Naturally Fractured Reservoirs, paper
SPE 38884 presented at the 1997 SPE Annual Technical Conference and
Exhibition, San Antonio, Texas, 58 October.
16. Ghafouri, H.R.: Finite element modelling of multi-phase flow through
deformable fractured porous media, PhD thesis, U. of Wales, Swansea,
Wales (1996).
17. Coussy, O.: A General Theory of Thermoporo-elastoplasticity.
Transport in Porous Media (1989) 4, 281293.
18. Charlez, Ph.: Rock Mechanics, Vol. 2 Petroleum Applications, Technip
Ed., Paris (1997).
19. Boutca, M.: Elements of Poro-elasticity for Reservoir Engineering,
Revue de lInstitut Franais du Ptrole, (1992) 47, No. 4, 479.
20. Morita, N. et al.: A Quick Method To Determine Subsidence,
Reservoir Compaction, and In-Situ Stress Induced by Reservoir
Depletion, JPT (January 1989) 71.
21. Zienkiewicz, O.C.: Coupled Problems and Their Numerical Solutions,
Numerical Methods in Coupled Systems, Wiley, New York City (1984) 35.
22. Lewis, R.W., Schrefler, B.A., and Simoni, L.: Coupling vs.
Uncoupling in Soil Consolidation, Intl. J. Num. and Anal. Meth.
Geomech. (1991) 15, No. 8, 533.
171
SI Metric
bar
ft
mile
Conversion Factors
1.0*
E + 05 = Pa
3.048*
E - 01 = m
1.609 344*
E + 00 = km
172
SPEREE