Sie sind auf Seite 1von 20

Research Journal of Environmental and Earth Sciences 4(11): 962-981, 2012

ISSN: 2041-0492
© Maxwell Scientific Organization, 2012
Submitted: August 15, 2012 Accepted: September 13, 2012 Published: November 20, 2012

Well Test Analysis in Dual-Porosity Aquifers with Stress-Dependent Conductivity


1
H. Jabbari, 1Z. Zeng, 1S.F. Korom and 2M. Khavanin
1
Department of Geology and Geological Engineering,
2
Department of Mathematics, University of North Dakota, Grand Forks, North Dakota 58202, USA

Abstract: A new model for analyzing the hydraulic head in the vicinity of a vertical well in fractured, confined
aquifers is presented. This study shows that flow dynamics within the fractured aquifers are more complex than
previously believed and the fluid flow behavior can be related to rock deformation through hydraulic conductivity
change with fluid withdrawal/injection. This fluid-solid interaction is particularly significant in stress-sensitive,
fissured rocks where the rate of withdrawal/injection is high. The model is derived for cubic geometry under
hydrostatic confining pressure. The solution, however, can be extended to handle other geometries. Considering the
conductivity changes during the life of an aquifer in this study, several findings are drawn: (1) a fully coupled geo
mechanics and fluid-flow model is developed to interpret well test data from confined aquifers with linear elastic
behavior, (2) for characterizing a fissured, confined aquifer knowing three major parameters may be sufficient,
which can be obtained by proper analysis of recovery data, (3) the concept of stress-dependent skin around the wells
is discussed and (4) even though the coupled effect may not significantly influence the drawdown/recovery data
from a normal (stress-insensitive) aquifer, such a coupled model provides an iterative algorithm to estimate the main
fracture parameters, namely fracture porosity, fracture transmissivity, fracture storage capacity ratio and the
elasticity parameter. In general, the fluid-solid coupling effects cannot be ignored when analyzing stress-sensitive
aquifers unless otherwise it is shown to be negligible.

Keywords: Confined aquifer, dual porosity, recovery data, stress-dependent conductivity

INTRODUCTION • The continuum approach that treats the porous


medium and the fractures as two separate but
Groundwater flow in fractured aquifers behaves overlapping continua (Peters and Klavetter, 1988;
differently from that in homogeneous ones. In Rodriguez et al., 2006; Bear and Cheng, 2010)
homogeneous, single porosity aquifers there is one • The Discrete Fracture Network (DFN) approach in
single flow regime, whereas in fractured aquifers there which the unknown heads at the intersections of
are two: fractures and matrix. Characterizing such a fracture network are calculated based on the laws
heterogeneous formation with two distinct media, for flow through individual fractures (Bacca et al.,
namely matrix and fractures, requires more caution and 1984; Karim-Fard and Firoozabadi, 2003)
a more sophisticated model, such as a dual-porosity • The Dual-Porosity (DP) model which assumes that
solution. As a matter of fact, quantifying the fluid flow the medium consists of two continua, one
in fissured rock masses requires the combination of associated with the fracture system and the other
theoretical, laboratory and field studies. In the with the matrix (Barenblatt et al., 1960; Warren
theoretical part for fissured aquifers, which are and Root, 1963; Kazemi, 1969)
heterogeneous and anisotropic formations, cogent
mathematical models are imperative (Lods and Gouze, There have been several studies to implement the
2008; Murdoch and Germanovich, 2006; Neuman, dual-porosity models for characterizing fractured
2005). These solutions can be obtained based on certain systems. The main goal in dual-porosity modeling is to
idealizations (Warren and Root, 1963; Snow, 1969; simulate the effects of a fractured system on drawdown
Kazemi, 1969; Moench, 1984; Hsieh, 1983). or buildup (recovery) data. This concept was first
In fact, three different approaches may be introduced by Barenblatt et al. (1960). It was later
employed for the analysis of groundwater flow in a applied by Warren and Root (1963) for the case of
fractured rock mass: pseudo-steady state interporosity flow model and by

Corresponding Author: H. Jabbari, Department of Geology and Geological Engineering, University of North Dakota, Grand
Forks, North Dakota 58202, USA
962
Res. J. Environ. Earth Sci., 4(11): 962-981, 2012

Kazemi (1969) for a transient flow model in naturally


fractured oil reservoirs. It took the groundwater
industry some time to adopt the proposed methods and
develop new models and primitive research on
fractured aquifers appears in the 1970s and 80s
(Streltsova, 1976; Boulton, 1977; Neuzil, 1981;
Moench, 1984). Of course, there had been some
previous research on homogeneous (single-porosity)
aquifers in the decades prior (Theis, 1935, 1938; Jacob,
1940; Cooper and Jacob, 1946; Domenico and Mifflin,
1965; Bear, 1972).
The dual-porosity concept is widely used to
analyze hydraulic heads at any point around the vertical
wells in the fractured media (Gerke and Genuchten, Fig. 1: Idealization of the heterogeneous porous aquifer
1993). However, the limitation of the dual-porosity (Warren and Root, 1963)
models is that they are mainly applicable when the
fractured media is represented by regularly-shaped • Pore volume compressibility of the fracture
objects, such as sugar cubes, slabs, or spheres. The (Jabbari et al., 2011b)
sugar-cube representation is an efficient configuration
method, but it cannot be applied when the fractures are In this study, we demonstrate the effects of change
not connected (Karim-Frad and Firoozabadi, 2003; in fracture conductivity on the overall behavior of a
Farayola et al., 2011). fissured, confined aquifer. Another goal of this research
Several approaches have been proposed to is to investigate whether or not the geo mechanical
implement the geomechanical effects into both oil behavior of the formation affects the transient test data
reservoir characterization (Min et al., 2004; Dautriat from a fissured aquifer in a dual-porosity framework.
et al., 2007; Tao et al., 2009) and groundwater flow
modeling (Cey, 2006; Cappa et al., 2008). However, MATHEMATICAL MODEL
most proposed models are based on numerical analysis
and exact analytical solutions for nonlinear Problem statement: The general configuration of a
geomechanics, fluid-flow problems are limited. It is the dual-porosity model for a fissured aquifer is depicted in
objective of this study to develop a three-parameter Fig. 1. The concept of dual-porosity model was
dual-porosity model for investigating the unsteady-state originally developed by Warren and Root (1963) in
groundwater flow in naturally fractured aquifers, with order to quantify the fluid flow in fractured rocks.
the inclusion of the changes in effective-stress of According to their model a fractured medium is
formations. This study is founded mainly on the model assumed to consist of two interacting, overlapping
of Warren and Root (1963) for dual-porosity systems media: a medium of low-permeability and high-storage
which was extended by Jabbari and Zeng (2011) to (matrix) and a medium of high-permeability and low-
incorporate the effect of rock deformation into fractured storage (fracture). The origin of the coordinate system
media modeling. It, in fact, couples the geomechanical is at the wellbore and the equations are derived on the
and fluid-flow aspects for characterizing a stress- horizontal plane (x, y). The developed model should
sensitive fractured formation, where the elasticity reflect the response of a real fractured aquifer.
parameter (ε) is relatively significant (ε>0.01). However, certain assumptions and idealizations are
Before presenting the proposed model, some points inevitable (Warren and Root, 1963):
are clarified so that the mathematical model is more
easily understood. First, by “fluid-solid interaction” we • The material containing the primary porosity is
imply that the change in the fracture hydraulic head homogeneous and isotropic and is contained within
(especially in the vicinity of the wellbore) can cause a systematic array of identical, rectangular
change in the fracture aperture and, thus, fracture parallelepipeds (matrix).
hydraulic conductivity. This change in hydraulic • All of the secondary porosity is contained within
conductivity comes from two sources: an orthogonal system of continuous, uniform
fractures which are oriented so that each fracture is
• Volumetric change of matrix blocks (the pore parallel to one of the principal axes of hydraulic
pressure effect in the matrix is ignored) conductivity (fractures).
963
Res. J. Environ. Earth Sci., 4(11): 962-981, 2012

• The complex of primary and secondary porosities  ∂ 2 h ∂ 2 h 1  S  ∂p1 ∂h 


(matrix plus fractures) is homogeneous. Flow to  ∂x 2 + ∂=  γ b − α n2  ∂t + γ n2 ( β 2 + c f ) ∂t 
 2   
2
y K
the wellbore can only occur from the fracture 
  S  ∂p  p 
network; hence, flow through the primary-porosity − α n2  1 =Csh K1  h − 1 
(from matrix to wellbore) cannot occur.   γ b  ∂t  γ 
(3)
Note, also, that the effect of poroelasticity in the
 p1
matrix has been ignored in this study since we have τ 0)
 I .C. : (= = h = hi
assumed that the matrix conductivity is negligible γ

compared to the fracture conductivity. Additional   ∂h 
Q = 2π T  
assumptions will be made at appropriate points in the  B.C. : (τ 〉 0)  ∂ ln r r = rw
mathematical treatment. 
The governing equation of hydraulic head for a  Q ( re , t ) = 0
(4)
single phase flow of a slightly compressible liquid in
the vicinity of a point sink/source within a uniform All the variables are defined in the section of
aquifer that is horizontal, homogeneous and anisotropic, notations. Here, the x-axis and the y-axis coincide with
is partially described by applying Green’s theorem to the principle axes of the conductivity field. In fact,
the Volume (V) to obtain the applicable form of the three different types of flow mechanics can be
continuity equation (Appendix A): distinguished:

1    S  ∂p ∂h •
∇.( ρ K 2 ∇h)=  − α n2  1 + γ n2 ( β 2 + c f ) Transient behavior in a bounded system
ρ  γ b  ∂t ∂t • Steady-state with outer boundary head
(1)
• Pseudosteady-state denoting a no-flow boundary
Because water in the matrix cannot flow directly to condition. Henceforth, the transient behavior of a
wellbore (dual-porosity assumption), the hydraulic head finite aquifer (bounded system) is employed in this
notation does not apply to the matrix, hence, we used study Eq. (4)
pressure notation for the primary porosity.
In addition to Eq. (1), continuity on a local basis is On the other hand, if we also consider the changes
necessary. The matrix-fracture interaction is described in fracture conductivity due to water withdrawal or
by a pseudo-steady state pressure relation fluid injection, the preceding differential system is
(Appendix A): changed such that it reflects the effect of head and, thus,
conductivity changes over the aquifer. Because in
 S  ∂ p1  p1  fissured formations it is the fracture network that
 γ b − α n2  ∂ t =C sh K1  h − γ 
    (2) constitute the main conduits for the flow and the
fracture aperture variations are stress-dependent, the
study of the coupled effective-stress and fracture
where, Csh , shape factor, reflects the geometry of the
conductivity has attracted significant attention in the
matrix elements and controls the flow between matrix recent years (Bai and Elsworth, 1994; Zhu et al., 1995;
and fractures. Takashi et al., 1995; Suri et al., 1997; Chin et al., 2000;
In the existing models for fractured formations, Gutierrez et al., 2001; Dautriat et al., 2007; Meza et al.,
from both petroleum and hydrology points of view, it is 2010). Notice that in Eq. (6) we defined a new term
assumed that the hydraulic conductivity of a fracture with ε that represents this phenomenon. Therefore, the
network is always constant and it does not vary with
principle differential system Eq. (3) turns out to be as
hydraulic head change (Warren and Root, 1963;
follows (Appendix B):
Kazemi, 1969; Streltsova, 1976; Boulton, 1977; Neuzil,
1981; Moench, 1984). For such an aquifer that is
 ∂2h ∂2h   n  3(1 − 2ν )    ∂h   ∂h  
2 2

confined top and bottom and is bounded at the outer  2 + 2 + γ  β 2 + 3  1 + 2  c f +     +    =


 ∂x ∂y   9  En2    ∂x   ∂y   (5)
boundary with a uniform initial head that is to be

produced at a constant discharge rate Eq. (4), the 1  S  ∂p1 ∂h 
 =  − α n2  ∂t + ( 2n2 β 2γ ) ∂t 
K 2  γ b
differential system which represents both the flow from   

fractures to wellbore and the flow from matrix to   S − α n2  ∂p1 =Csh K1  h − p1 
fracture network can be written as follows:   γ b  ∂t


 γ 

964
Res. J. Environ. Earth Sci., 4(11): 962-981, 2012

The transformed equations in polar coordinates (ξ,  −λ x   −λ x 


) ln(η x ) + Ei 
Φ ( x=  − Ei  
θ), the initial condition and boundary conditions can be  ω (1 − ω )  1−ω 
rewritten as follows (Jabbari and Zeng, 2011): (11)
 −λ x    λ η x   λx   −λ x  
2 3

+ exp    ln   − E1  1 − ω  − Ei  ω  
 1 − ω    256(1 − ω )     
 ∂ 2 h  1  ∂h  ∂h 
2
∂hD ∂ψ
 2D +   D − ε  D =  ω + (1 − ω )  −λ x    (1 − ω )η x   λx   λ x 
 ∂ξ ξ  ∂ξ ∂ξ  ∂τ ∂τ + exp   ln   + E1   − E1   
    ω (1 − ω )    ω   ω (1 − ω )   ω 
 ∂ψ
 (1 − ω ) ∂τ =λ ( hD −ψ )
(6) Ei And E1 are “Exponential Integral” functions,
defined as:

 I .C : (=
τ 0) = ψ=

e −u
Ei ( − x ) =−∫
hDi 0
 i du (x〉 0)
 ∂hD u
(12)
 B.Cs : (τ 〉 0)
x
ξ =1 =
−1
 ∂ξ
 ∂hD x
eu − 1
 ξ =ξ D =0 E1( x ) =
γ + ln( x) + ∫ du ( x〉 0) (13)
 ∂ξ (7) 0
u
= Ei ( x ) − π i
where, the dimensionless groups are defined in
Table 1. and the constant η is:
The effect of the change in fracture hydraulic
conductivity around the wellbore can be quantified by
η = 4 exp 
γ 
introducing a new skin factor which is called stress-  (14)
sensitive skin. Therefore, the total skin factor is 2
composed of two elements:
where, 𝛾𝛾� =
�0.57721 is the Euler-Mascheroni constant
• The mechanical skin due to near-wellbore damage (also called Euler’s constant). Among the three major
(S d ) dimensionless groups (ω, λ, ε) two of them, ω and λ,
• The stress-dependent skin which accounts for the are important; they are the groups by which we can
effective-stress change in the near wellbore region describe the deviation of the behavior of an aquifer with
and its impact on the hydraulic conductivity of the dual-porosity from that of a homogenous one (Warren
fracture network (S’) and Root, 1963). The fracture storage capacity (ω) is a
measure of the fluid stored in fractures as compared
The skin effect condition is defined by Da Prat (1981): with the total water present in the aquifer and it has a
value between zero and unity. A value close to 1
indicates that most of the water is stored in the
 ∂hD 
hwD= hD − S  
(8) fractures, whereas a value of zero indicates that no
 ∂ξ ξ =1 water is stored in the fractures. A value of 0.5 indicates
that the water is stored equally in matrix and fractures.
Therefore, the total skin is obtained as (Jabbari and The inter-porosity flow coefficient (λ) is a measure
Zeng, 2011): of the heterogeneity scale of the system and quantifies
the water transfer capacity from matrix to fracture
=S (1 + 0.75γε ) S d + ε S ′ (9) network. A value of unity for λ indicates the absence of
fractures or, ideally, that fractures behave like the
matrix such that there is physically no difference in
where S' (stress-dependent skin factor) is given by:
petro physical properties; in other words, the formation
τ
is homogeneous. Low values of λ, on the other hand,
π2 9γ 2
Φ ( x)
S′ =
24

32
+ ∫ + 4x
dx (10) indicate slow water transfer between matrix and
fractures. However, the actual range of λ in oil
0

reservoirs could be 10-9, which indicates poor fluid


where, transfer, to 10-3, which indicates a very high fluid
τ : The time during drawdown transfer between fractures and matrix (Warren and
Φ(x) : The elasticity function defined as: Root, 1963).
965
Res. J. Environ. Earth Sci., 4(11): 962-981, 2012

Table 1: Dimensionless parameters


2π T 2π T  p1 (ξ , θ , τ ) 
(ξ , θ , τ )
hD= (h − h (ξ , θ , τ ) ) ψ=
(ξ , θ , τ )  hi − 
Q  γ 
i
Q
x +y −1  y 
2 2

ξ= θ = tan  
rw x
2
Tt Csh K1 rw
τ = λ=
rw ( S + ( β 2 + c f − α )γ bn2 )
2
K2

n2 ( β 2 + c f ) γQ   n2   3(1 − 2ν ) 
ω= ε
=  β 2 + 3  1 + 9   c f + En  
S
+ ( β 2 + c f − α ) n2 2π T   2 
γb

Moreover, the third parameter (ε) reflects the impact of effective-stress change (due to head change) on the
conductivity of fractures and on the behavior of the aquifer. This parameter depends on rate of withdrawal/injection
(Q), geo mechanical properties (E, v), water compressibility (β 2 ), fracture porosity (n 2 ) and Transmissivity (T)
(Table 1).

General solutions in the Laplace domain: The differential system in Eq. (6) is nonlinear and is solved by using
regular perturbation theory (He, 1999, 2000) and the Laplace transformation method. The solution to this nonlinear
system with the prescribed boundary conditions in Eq. (7) is obtained in the Laplace space as follows:

hD ( ξ , s ) =
( ) () ( ) (
I o ξ sf ( s ) .K1 ξ D sf ( s ) + I1 ξ D sf ( s ) .K 0 ξ sf ( s ) )
s sf ( s )  I ( ξ sf ( s ) ) .K ( sf ( s ) ) − I ( sf ( s ) ) .K ( ξ sf ( s ) ) 
1 D 1 1 1 D

 c  I ( ξ sf ( s ) ) .K ( sf ( s ) ) + I ( sf ( s ) ) .K ( ξ sf ( s ) )  
 
o 1 1 o

−ε  ξ sf ( s ) ξ sf ( s )

− I o (ξ sf ( s ) ) ∫ (
χ K 0 ( χ ).ℑ( χ ) d χ + K o ξ sf ( s ) ) ∫ χ I 0 ( χ ).ℑ( χ )d χ 
 sf ( s ) sf ( s )  (15)

where,
S : The Laplace transformation variable, ℎ𝐷𝐷 (𝜉𝜉, 𝑠𝑠) �= 𝑙𝑙[ℎ𝐷𝐷 (𝜉𝜉, 𝜏𝜏)]
I n (x) & K n (x) : Modified Bessel functions of the first and second kind of the nth order, respectively and the
constant c along with the function 𝑠𝑠̃ (𝑥𝑥) are defined as:

ξD ξD

( ) ∫ ( ) ∫
sf ( s ) sf ( s )

I1 ξ D sf ( s ) χ K 0 ( χ ).ℑ( χ ) d χ + K1 ξ D sf ( s ) χ I 0 ( χ ).ℑ( χ ) d χ
sf ( s ) sf ( s )
c=
I ( ξ sf ( s ) ) .K (
1 D 1
sf ( s ) ) − I ( sf ( s ) ) .K ( ξ
1 1 D
sf ( s ) ) (16)

 ∂hD 0  2 
   
 ∂ξ  ξ = χ
sf ( s )
ℑ( χ ) =
sf ( s ) (17)

where h D0 is the solution for the case without considering the effect of elasticity (the inversion of the first fraction
term on the RHS of Eq. (15). The general solution for the case of an infinite acting system can also be obtained by
allowing the aquifer size grow to infinity, ξ D →∞. This gives us the following:
966
Res. J. Environ. Earth Sci., 4(11): 962-981, 2012

Table 2: Different types of interporosity flow models


Block geometry Model
Sugar-cube model (1 − ω ) λ
(Warren and Root, 1963) f (s) = ω +
(1 − ω ) s + λ
Slab-shaped matrix
(Deruyck et al., 1982) λ (1 − ω )  3(1 − ω ) s 
) ω+
f ( s= tanh  
3s  λ 
Spherically-shaped matrix
λ  15(1 − ω ) s  15(1 − ω ) s  
(Deruyck et al., 1982) f (s) =ω+  coth   − 1
5s  λ  λ  

hD ( ξ , s ) =
(
K 0 ξ sf ( s ) )
s sf ( s ) K1 ( sf ( s ) )
 ∞ K0 (χ )   (18)
 ∫ (
d χ   I o ξ sf ( s ) .K1 ) (
sf ( s ) + I1 ) ( ) (
sf ( s ) .K o ξ sf ( s ) ) 
ε 
−   sf ( s ) K 1( sf ( s ) χ ) 
 
s

 ξ sf ( s )
K0 (χ )
ξ sf ( s )
I0 (χ ) 

 o I ( ξ sf ( s ) ) ∫ χ d χ + K ( ξ sf ( s ) ) ∫ χ dχ 
 
o
sf ( s ) sf ( s )

In Eq. (18) it is clear that as we move away from the wellbore, the effect of geomechanics will become
vanishingly small since the coefficient of ε (the terms in the brackets) becomes smaller. The function f(s) depends
on the assumed matrix-fracture flow model. The different inter porosity flow models, which can be used for
different characteristic shapes of matrix blocks, are shown in Table 2.
The solution at the inner boundary (wellbore), i.e., ζ = 1, with the consideration of the total skin around the
wellbore Eq. (9) and for different flow periods, namely the early-, intermediate- and late-time periods; are obtained
for both finite and infinite confined, fractured aquifers.

Solutions for closed outer boundary aquifers: The dual-porosity solution for the case of a finite, confined aquifer
with sugar-cube geometry is obtained for the three time periods. At early time t→0, so s→∞, giving f(s)→ω
(Sageev et al., 1985). The inverse transform of Eq. (15) for this case can then be obtained by using the Heaviside
expansion theorem:

  − n 2π 2τ  
 exp  2  
τ 2(ξ D − 1)  ω (ξ D − 1)   + (1 + 0.75γε ) S + ε  π − 9γ 
∞ 2 2

hwD (1, τ=
) +  1 − 62 ∑  24 32 
(19)
ω (ξ D − 1)  π 
d
3 n =1 n2  
 
 

At intermediate time λ controls the flow and f(s) →λ/s (Sageev et al., 1985). These yields:

hwD (1, τ )
Io( λ ) .K (ξ λ ) + I (ξ λ ) .K ( λ ) + (1 + 0.75γε )S
1 D 1 D 0

λ  I (ξ λ ) .K ( λ ) − I ( λ ) .K (ξ λ ) 
d

 1 D 1  1 1 D

(20)
 π 2 9γ 2 1   λ  γ  2 
 − +  ln  +  
 24 32 2   2  4  
+ε  
1  
ξ λ ξ λ

( ) ( )
D D

−  I1 ξ D λ ∫ χ K 0 ( χ ).Θ( χ )d χ + K1 ξ D λ ∫ 0 χ I ( χ ).Θ ( χ ) d χ 
 λ λ  λ λ  
where,
967
Res. J. Environ. Earth Sci., 4(11): 962-981, 2012

( ) ( )
2
 I 1 ξ D λ . K1 ( χ ) − I 1 ( χ ) . K1 ξ D λ 
Θ( χ ) =  
( ) ( λ ) − I ( λ ) .K ( ξ )
3
 I 1 ξ D λ . K1 λ 
 1 1 D  (21)

At late time, t⟶ ∞, so s⟶ 0 (Sageev et al., 1985), then Eq. (15) for a finite naturally fractured, confined
aquifer reduces to:

 (1 − ω ) 2   −λτ  
τ + λ 1 − exp  ω (1 − ω )   
2     
hwD (1, τ )   + (1 + 0.75γε ) S d
ξD − 1  ξD   −λτ   − λτ   
2 2

+ ln τ + 0.80908 + Ei   − Ei  
 4   ω (1 − ω )   1 − ω   
 π 2 9γ 2 τ Φ ( x ) 1 
 24 − 32 + ∫ 4 x dx − × 
 1 
2

 
+
0
 ξ2 
1 −
  D 

 
+ε   ξ
K0 (χ )
ξ
I0 (χ ) 
( ) ( )
sf ( s ) D sf ( s ) D

  I1 ξ D sf ( s ) ∫ χ d χ + K1 ξ D sf ( s ) ∫ χ d χ  
× −1  sf ( s ) sf ( s ) 



 (
s sf ( s )  I1 ξ D sf ( s ) K1 ) (
sf ( s ) − I1 ) (
sf ( s ) K1 ξ D sf ( s )   
 ) ( )
    (22)
2 2
τ 〉100ωξ D , if λ 〈 〈1 , or , τ 〉100ξ D − 1 / λ , if ω 〈 〈1

The Laplace inversion term in Eq. (22) can be executed numerically since an analytical solution is either
unavailable or complex. The algorithms for numerical Laplace inversion are available in the engineering
mathematics literature (Stehfest, 1970; Bellman et al., 1976; Crump, 1976; Talbot, 1979; Ilk et al., 2005; Iseger,
2006; Al-Ajmi et al., 2008).

Solutions for extremely large aquifers: The solutions for the case of an infinite-acting system, namely an
extremely large aquifer in the radial direction, are obtained as follows:
At early time:

τ π2 9γ 2 
(1, τ ) 2
hwD= + (1 + 0.75γε ) S d + ε  −  (23)
πω  24 32 

At intermediate time:

hwD (1, τ ) =
Ko( λ ) + (1 + 0.75γε )S
λK ( λ )
d
1 (24)
π2 9γ 2
1  λ  γ 1
2
∞ 
+ε  − +  ln  +  − ∫ χK ( χ ).K12 ( χ ) d χ 
 24

32 2  2  4  λ λ K1
3
( λ) λ
0

And at late time:

 1 + 0.75γε   ln τ + 0.80908 + Ei  −λτ  − Ei  −λτ  + 2 S 


wD (1, τ )
h=       d 
(25)
 2   ω (1 − ω )  1−ω  

968
Res. J. Environ. Earth Sci., 4(11): 962-981, 2012

For high values of shut-in time Eq. (26) turns into:

Q (1 + 0.75γε )  t p + ∆t 
hws = hi − ln  
(28)
4π T  ∆t 

where τ p , the dimensionless producing time, in practical


field units is:

T tp (29)
τp =
1440r w
2
( S + ρ bn (β
2 2
+ c f − α ) /144 )

where,
heads : In ft
Q : In ft3/day
T : In ft2/day
rw : In ft
ρ : In lb m /ft3
b : In ft
β 2 , c f & α: In psi-1
Fig. 2: Interpretation of buildup data tp : In minutes

RECOVERY DATA INTERPRETATION Also, the fracture Transmissivity (T) can be


determined from combining the slope of the straight
In hydrology, one way to get information about the lines Eq. (26) with the elasticity parameter ( ε in
fracture properties and the behavior of a fissured Table 1). This, in field units, gives us:
aquifer is by interpreting recovery data. Analyzing a
drawdown test is limited by the flow rate fluctuations 0.0915Q   n  3(1 − 2ν )   (30)
inherent to pumping. The zero flow = rate that T 1 + 1 + 0.01ρ m  β + 3  1 +   c f +   
2

m       
2
9 En
corresponds to buildup (recovery) does not have this 2

problem. By superimposing the drawdown equation for


an infinite-acting system Eq. (25) for the case of an where m defined as:
infinite-acting system, the behavior of a shut-in well
following a constant pumping rate is considered. If the 0.183Q (1 + 0.75γε ) (31)
m=
duration of pumping, t p , is fairly long and if certain T
3
conditions hold τp≻ 𝑎𝑎𝑎𝑎𝑎𝑎 ∆𝜏𝜏 ≻ 100𝜔𝜔 𝑖𝑖𝑖𝑖 𝜆𝜆 ≪ 1, 𝑜𝑜𝑜𝑜,
𝜆𝜆 is the slope of the semi-log straight line, in ft/cycle, to
1
∆𝜏𝜏 ≻ 100 − 𝑖𝑖𝑖𝑖 𝜔𝜔 ≪ 1 (Warren and Root, 1963), then be obtained from the buildup data (Fig. 2). Substituting
𝜆𝜆
the buildup head may be obtained by Eq. (26), which the calculated transmissivity Eq. (30) along with the
uses Horner method and is depicted in Fig. 2. given slope (m) into Eq. (31), we can obtain ε .
The fracture storage capacity, ω, can also be
Q  1 + 0.75γε    t + ∆ t 
τ p +∆ τ
 −λ ∆ τ   −λ ∆ τ   Φ( x)  obtained by combining Eq. (27) and (28) to yield:
=hws hi −    ln   + Ei   − Ei   − ε ∫ dx 
p

2π T  2    ∆t 
  ω (1 − ω )   1−ω  ∆τ 4x 
 τp 
(26) − δ + ε Φ( x)
dx 
where, m 1.151 (1+ 0.75γε ) ∫+ 4x 
h ws = Shut-in head ω =10  0  (32)
∆𝑡𝑡 = The time since pumping stopped (shut-in time)
∆𝜏𝜏 = The dimensionless shut-in time where δ is the vertical displacement between the early
and late semi-log straight lines for a buildup curve
For relatively small times the early straight line in (Fig. 2) Also, the mechanical skin factor due to near
the semi-log coordinates acts as though the aquifer is wellbore damage, S d , can be obtained from the
homogeneous. This behavior is described by the superposition concept as follows:
following relationship:
 hws (1min) − hwf (t p )  T  
− log  + 1.024 
) 
τ S d = 1.151 
Q  1 + 0.75γε   t p + ∆t  Φ( x)  (27) (
 rw S + ρ bn2 ( β 2 + c f − α ) / 144
p 2
 
−ε ∫
m
hws hi −   ln  dx 
2π T  2   ω∆t  0 4 x  (33)
969
Res. J. Environ. Earth Sci., 4(11): 962-981, 2012

ESTIMATING THE FRACTURE PROPERTIES An iterative calculation technique has been


introduced in this study for reaching satisfactory results
Since fracture porosity is a scale-dependent from pumping test analysis. The flowchart in Fig. 3
parameter, any method to estimate it is prone to error. It demonstrates the steps of this trial-and-error algorithm.
can be obtained from well logging or well testing. The method is to try out various values of the
However, the porosity computed from well logs and secondary porosity until the resulting error is
well tests may differ from one another and they are sufficiently reduced or eliminated. This process can use
different most of the time. Also, the radius of the value obtained from well logging, if available, as
investigation in well logging is limited to a few feet the first guess. Using well test data through an iterative
around the wellbore. Therefore, the value obtained from procedure, we can achieve much more accurate fracture
well logs does not represent an average value for the porosity. This fracture porosity can be considered as an
reservoir sector under study, in spite of use of average value for the part of the aquifer involved in the
sophisticated computation methods. cone of depression (radius of investigation) (Jabbari
Note, also, that the value of the secondary porosity and Zeng, 2011).
has a significant impact on the properties of a fractured On the other hand, the procedure to answer a
aquifer Eq. (29) to (33). The value of n 2 can be any nonlinear problem, such as that in Eq. (6), by means of
number at any scale over the aquifer, but the average the iterative algorithm shown in Fig. 3 might run into a
value for the whole aquifer is generally less than 1% convergence problem. The iterative solution is to make
(Garcia, 2005). According to Nelson (2001) fracture an initial guess for the desired variable, here fracture
porosity is always less than 2%; in most formations it is porosity. Then, the procedure calculates the fracture
less than 1%, with a general value of less than 0.5%. porosity and cycle begins again. This continues

Fig. 3: Algorithm to determine the average values of the properties of a fractured aquifer

970
Res. J. Environ. Earth Sci., 4(11): 962-981, 2012

until the fracture porosity settles to a value which is Also, we assume that λ is not as sensitive as fracture
within a specific tolerance limit. This limit, however, properties to matrix-fracture elasticity and fracture
can be altered using various option parameters in the porosity and it can be obtained from the proper analysis
algorithm, such as error in the algorithm shown in of conventional type-curve matching developed by
Fig. 3. If the fracture porosity does not converge within Onur et al. (1993).
a certain number of iterations, the loop will not cease.
If the fracture porosity does not converge, the DISCUSSION
values of input data, such as Young’s modulus (E),
Poisson’s ratio (v), the slope of the semi-log straight As discussed earlier, for stress-sensitive fissured
line (m) and interporosity flow coefficient (λ) should be aquifers, effective-stress changes induce changes in
reviewed to ensure their values are appropriate. If the fracture hydraulic conductivity around the wellbore.
procedure fails to converge, the first guess for the
This, in turn, depending upon the value of elasticity
fracture porosity may be changed. Nevertheless,
convergence problem from the fracture porosity would parameter (ε), would affect the buildup/drawdown
result if its first guess is too low or too high. Almost responses from a stress-sensitive aquifer. In this
always, 1% fracture porosity would be a reasonable section, we investigate the effect of on the response of
first guess for the calculations. drawdown curves from both finite and infinite-acting
However, when a convergence problem is aquifers namely ω , λ , and ε.
encountered, it is better to start with relatively lower Head variations vs. time plots (both dimensionless)
fracture porosity, say 0.5% (Nelson, 2001) and proceed shown in Fig. 4 and 5 clearly show the effect of
with the subsequent suggestions until convergence is elasticity parameter on the transient test curves. The
achieved. The sequence of the suggestions is structured results presented in these figures indicate that
so that they can be incrementally added to the program. drawdown tests, through the elasticity parameter (ε),
All in all, the proposed algorithm in Fig. 3 seems not to can identify whether a fissured aquifer is stress-
have convergence problems if the required data are
sensitive or stress-insensitive. The elasticity parameter,
input as described above.
As stated earlier, we can use the above rule of however, can vary from aquifer to aquifer which is
thumb (n 2 = 0.5 and 1%) for the first guess to input to dependent on the solid-fluid parameters (Table 1). The
the iterative algorithm if another estimate is not range of variation of the elasticity parameter will be
available. discussed later in this study.

Fig. 4: Behavior of closed-boundary, naturally fractured aquifers (ω = 0.01, λ = 5×10ᅳ5)

971
Res. J. Environ. Earth Sci., 4(11): 962-981, 2012

Fig. 5: Behavior of infinite-acting, naturally fractured aquifers (ω = 0.01)

Fig. 6: Asymptotic solutions for a finite, naturally fractured aquifer (ε = 0)

Figure 4 and 5 shows that the influence of the interaction between ω and λ, which is the
formation elasticity on the drawdown curves may not communication between primary porosity and
be significant, whereas one advantage of considering secondary porosity. At late times, the curves show the
such a parameter in modeling is that it provides the behavior of both matrix and fractures together.
estimation of the average fracture porosity via the Also, Fig. 6 and 7 present suites of the solutions in
iterative algorithm in Fig. 3. In Fig. 4 and 5, all the real space (time domain) for the cases of both finite
curves are nearly identical at early times, representing (with ξ D = 500) and infinite-acting aquifers Eq. (22)
well discharge from storage in the fissures, followed, at and (25) and for various values of λ and ω. The
intermediate time, by the transitional curves that elasticity of
connect the two linear portions, representing the
972
Res. J. Environ. Earth Sci., 4(11): 962-981, 2012

Fig. 7: Asymptotic solutions for an infinite-acting, naturally fractured aquifer (ε = 0)

the formation is ignored in these schematics. Note, also, Table 3: Recovery data from an oil well (Najurieta, 1980)
that Eq. (22) (used in Fig. 6) is appropriate when certain ∆t, min (t p +∆t)/∆t ∆h s , ft
1 516668 325.1
conditions hold (τ>100ω 𝜉𝜉𝐷𝐷2 , 𝑖𝑖𝑖𝑖 𝜆𝜆 ≪ 1, 𝑜𝑜𝑜𝑜, 𝜏𝜏 > 100𝜉𝜉𝐷𝐷2 − 2 258335 315.6
1
, 𝑖𝑖𝑖𝑖 𝜔𝜔 <). Hence, for the case where, ξ D = 100, with 4 129168 307.3
𝜆𝜆 8 64584 299.5
ω = 0.001, 0.1, or 0.1, the asymptotic solutions are valid 16 32283 283.3
when τ>25×103, τ>25×104, or τ>25×105, respectively. 32 16147 272.1
Likewise, Eq. (25) (used in Fig. 7) is appropriate 64 8073 264.9
for all values of ω and λ when τ>100. However, for 128
256
4037
2019
245.8
230.3
small values of ω and λ, τ>100ω, if λ<<1, or, τ>100- 512 1010 204.4
1/λ, if ω<<1. Hence, for the case of an infinite-acting 1024 505.6 187.5
aquifer with ω = 0.001, 0.01 , or 0.1 , the asymptotic 2048 253.3 172.1

solutions are valid when τ>0.1, τ>1, or τ>10,


a naturally fractured, oil reservoir, discussed in the
respectively.
study by Najurieta (1980). Table 3 shows the observed
The dual-porosity model of Warren and Root
head data and Fig. 8 shows the corresponding Horner
(1963) for a fractured medium is valid if we assume
plot. Basic data include: Q = 14, 318 ft3/day, r w = 0.375
that there is no change in fracture conductivity with
ft, α 6.9×10-5, psi-1, n 1 = 0.2, β 1 = β 2 = 3.2×10-6 psi-1,
respect to change in hydraulic head. However, changes
b = 18 ft, from which we can obtain S = 4.2×10-3. The
in hydraulic head (pressure) due to water withdrawal or
drawdown at the end of the flowing test, h vf (t p ), is 960
fluid injection (e.g., in disposal wells), theoretically, ft and the producing time is t p = 516, 667 min. Also, we
would affect the aquifer effective-stress and this, in assume n 2 = 1% as the first guess.
turn, would influence the fracture hydraulic As is clear in Fig. 8, the test has not distinctly
conductivity. Hence, this change in hydraulic shown the early straight line. Hence, this line is
conductivity may influence the behavior of an aquifer at obtained from an extrapolation.
the near wellbore region, which can cause changes in From the buildup plot the slope of the semi-log
total skin around the wellbore Eq. (9). straight line, m and δ are obtained as 71.2 ft/cycle and
250 ft, respectively. Using the conventional type-curve
CASE STUDY match presented in the literature (Onur et al., 1993), the
following results are obtained: ω = 0.0004, λ = 2×10-6
The mathematical model presented in this study is and S d = 4.8. Since there is no information available
tested by analyzing a recovery (buildup) test taken from regarding the geomechanical parameters of the
973
Res. J. Environ. Earth Sci., 4(11): 962-981, 2012

Fig. 8: Recovery data interpretation

Table 4: Final results from recovery data interpretation using the iterative algorithm in Fig. 3
Geomechanical 2
Iteration No. Until
Runs parameters T, (ft /d) n 2 , (%) ω ε 𝑠𝑠̅ convergence
♠E 1 = 4.5E6
1 ν 1 = 0.26 36.88 0.38 0.00035 0.0068 5.63 12
2 ν 2 = 0.30 36.86 0.39 0.00036 0.0055 5.61 10
3 ν 3 = 0.35 36.84 0.41 0.00038 0.0041 5.58 9
E 2 = 7.5E6
4 ν 1 = 0.26 36.84 0.41 0.00038 0.0039 5.58 10
5 ν 2 = 0.30 36.83 0.42 0.00039 0.0033 5.57 8
6 ν 3 = 0.35 36.82 0.43 0.00040 0.0025 5.56 7
E 3 = 9.5E6
7 ν 1 = 0.26 36.82 0.42 0.00039 0.0031 5.57 6
8 ν 2 = 0.30 36.82 0.43 0.00040 0.0026 5.56 6
9 ν 3 = 0.35 36.81 0.43 0.00040 0.0020 5.55 6
♠ (psi)

Table 5: Sensitivity analyses on the different parameters affecting ε


Base case: n 2 = 0.3%, Q = 33690 (ft3/day), E = 7.5×106 (psi), v = 0.26, c f = 10-5 (psi-1), and T = 36.8 (ft2/day)
n 2 (%) Q (ft3/day)
0.3 0.6 1 5615 33690 56150
ε 0.0142 0.0082 0.0057 0.0024 0.0142 0.0237
E (psi) v (dimensionless)
4.5×106 7.5×106 9.5×106 0.26 0.3 0.35
ε 0.0223 0.0142 0.0117 0.0142 0.0122 0.0097
C f (psi-1) 2
T (ft /day)
5×10-6 1×10-6 5×10-5 30 36.8 60
ε 0.0133 0.0142 0.0218 0.0174 0.0142 0.0087

formation from which the test is taken, namely E and v, simulation conditions for all 9 cases are identical except
we executed the iterative algorithm (Fig. 3) for 9 for the matrix geomechanical parameters, namely
different cases with different combinations of E and v. Young’s modulus (E) and Poisson’s ratio (v). As is
This helps us determine the effect of these parameters clear in Table 4, the matrix-fracture elasticity
on the resulting fracture parameters (Table 4). The interaction (different geo mechanical parameters) may
results obtained from the iterative algorithm, with a affect the behavior of buildup curves from fissured
relative error of 10-8, are shown in Table 4. aquifers. Although the elasticity parameter seems not to
The effect of the change in fracture hydraulic affect the buildup curve in this specific case since the
conductivity on the recovery data is studied using 9 transmissivity shows little change, using the introduced
simulation runs (runs 1 through 9 in Table 4). The iterative algorithm in this study (Fig. 3), provides a
974
Res. J. Environ. Earth Sci., 4(11): 962-981, 2012

practical method to estimate the values of average the fracture transmissivity is high, the elasticity
secondary porosity (n 2 ) and fracture storage capacity parameter becomes vanishingly small since the relative
(ω), which both of the results are in accordance with change in a high fracture hydraulic conductivity would
what was claimed earlier in the context. In terms of the be insignificant when hydraulic head changes.
estimated fracture porosity, generally speaking, the Conversely, it would be significant when the fracture
iterative algorithm works well in practice since the transmissivity is small.
estimated secondary porosities is near the value stated
by Nelson (2001), namely 0.5%. Moreover, the CONCLUSION
approximated values of the fracture storage capacity are
well in line with the result obtained by Najurieta (1980) In this study a mathematical model, founded on the
which is 0.0004. original model of Warren and Root (1963), is
The estimated elasticity parameter (ε) values for developed for coupling the fluid-flow and rock
different combinations of geo mechanical parameters deformation interactions in aquifers with dual-porosity
show that the fractured medium in this example is behavior. The model is to study the effect of stress-
stress-insensitive. This is because the estimated dependent fracture hydraulic conductivity on the
elasticity parameter values are very small (ε<<0.01)). hydraulic head transient data from a fissured, confined
Also, relatively higher values of skin factor in Table 4, aquifer. The following conclusions are drawn from this
compared with that obtained from conventional type- study:
curves, which is 4.8, are due to the inclusion of
elasticity concept in the calculations, namely the stress- • A general mathematical model for hydraulic head
dependent skin defined in Eq. (9). trends in finite and infinite-acting, confined,
To better understand the concept of elasticity fissured aquifers with the consideration of fracture
parameter and determine its contribution to the head hydraulic conductivity changes is developed. The
traces in drawdown and buildup tests, arbitrary ranges model is successfully employed as a tool to analyze
of variation of each parameter were chosen and their the well test problems in heterogeneous aquifers.
corresponding elasticity parameters were calculated. • Effective-stress change in the system usually does
The results are shown in Table 5. The base-case data not affect the shape of the head trends, since the
are identical for all 12 cases except for the parameter elasticity parameter (ε) is in the order of 10-2
for which the sensitivity analysis was run (Table 5). (Table 5). Therefore, this parameter will not
The results in Table 5 indicate that the effect of significantly influence the fracture transmissivity
matrix block elasticity becomes insignificant as the and Storativity (S). However, it provides us a
secondary porosity (n 2 ) increases. This seems logical
promising method, through the iterative algorithm
since in such a case where fracture compressibility is
(Fig. 3), to quantify the fracture parameters, such
the dominating parameter as for the elasticity of the
combined matrix and fracture system. The rate of as fracture porosity, fracture storage capacity,
withdrawal/injection (Q) is the main controlling elasticity parameter and stress-dependent skin
parameter that is linearly related to ε. The larger the Q, around the wellbore.
the larger the elasticity parameter (ε) and the more • All other parameters being constant, flow rate (Q)
significant the effect of elasticity on the aquifer and fracture Compressibility (C f ) variations have
characterization calculations. As shown in Table 5, for the largest impact on the elasticity parameters-due
the case with a higher pumping rate (56150 ft3/day or to the linear relationship among them (Table 5).
equivalently 10000 BBL/day), a higher elasticity The other parameters, affect the elasticity
parameter of 0.0237 is obtained which seems to be parameter but to a less extent.
fairly influential in the calculations (values of ε in • For a weakly fissured aquifer with a stress-
Table 5). As for the effect of geo mechanical dependent, fracture conductivity, the recovery trace
parameters, one would deduce that the elasticity may show differences from that with constant
parameter is lower in cases of higher values of E and v. conductivity. Specific combinations of the main
This, in fact, makes sense because for a stiffer matrix parameters, namely fracture conductivity level,
block with a higher bulk modulus (K) (due to high E flow rate, fracture compressibility and matrix geo
and v), the external head (pressure) change needed to
mechanical parameters, may reduce the effect of
change the unit block volume is higher. Hence, for a
specific state of stress, the impact of elasticity elasticity on the resulting fracture parameters from
parameter becomes weaker with higher module. Also, the well test interpretations (small ε , Table 5).
the higher fracture Compressibility (C f ) leads to a • Long producing times intensify the severity of the
higher elasticity parameter and, therefore, a higher aquifer stress sensitivity which is evident on the
relative volume change when associated with a change resulting total skin factors, through the stress-
in fracture hydraulic head. Last but not the least, when dependent skin Eq. (9).
975
Res. J. Environ. Earth Sci., 4(11): 962-981, 2012

APPENDIX A If we define the Storativity (S) of a confined aquifer as the product of


the Specific Storage (S s ) and the aquifer thickness (b) (Fetter, 2001),
Considering the continuity equation with isotropy in fracture we have:
conductivity (Warren and Root, 1963), namely K 2x = K 2y = K 2z , we
have: S
S = γ b( n1 β 1 + α ) ⇒ n1 β1 + α (1 − n2 ) = − α n2 (A8)
γb
  ∂ ( ρ n )1 ∂ ( ρ n ) 2
∇.( ρ K 2 ∇= h) + Therefore,
∂t ∂t (A1)
∂ ( ρ n )1  S  ∂p (A9)
= ρ − α n2  1
where, the subscripts 1 and 2 denote primary and secondary porosity, ∂t  γb  ∂t
respectively. To determine the rate of change of mass in matrix we
start with (Bear, 1972): To determine the rate of change of mass in fracture network, we have:

 n1  1 ∂ n2 1 ∂ n2 1 ∂n (A10)
Vs = Vb (1 − nm ) = Vb  1 −  = const cf =
− = =2
 1 − n2  n2 ∂ σ e n2 ∂ p 2 γ n2 ∂ h
∂Vs  1 ∂n1 n1 ∂n2   n1  ∂Vb
⇒ = Vb  − . − .  + 1 −  = 0 ∂ ( ρ n) 2 ∂ p2 ∂ ρ ∂n ∂p
∂t  1 − n2 ∂t (1 − n2 ) 2 ∂t   1 − n2  ∂t = n2 . + ρ 2 = ρ n2 ( β 2 + c f ) 2 (A11)
(A2) ∂t ∂ t ∂ p2 ∂t ∂t

Vs ∂n2 ∂n ∂Vb  Vb  ∂n1 Hence, A1 becomes:


n1 + n2 =
1− ⇒ =
− 1 ⇒ =
 
Vt ∂t ∂t ∂t  1 − n2  ∂t
   S  ∂p ∂h 
(A3) ∇.( ρ K 2∇
= h) ρ  − α n2  1 + γ n2 ( β 2 + c f )  (A12)
 γ b  ∂ t ∂t 
∂n1  1 − n2  ∂Vb ∂σ e  ∂σ t ∂p1  ∂p1
=  × =−(1 − n2 )α  −  =(1 − n2 )α Assuming that the fractures are fully filled with water, one would say
∂t  b  e
V ∂σ ∂t  ∂t ∂t  ∂t
β2 =
� C f . This gives us:
(A4)
where, ∂ ( ρ n) 2 ∂h
V s = Volume of solid phase = 2 ργβ 2 n2
V b = Bulk volume of a matrix block ∂t ∂t (A13)
n m = Matrix porosity
σ e = Effective stress If we also consider pseudo-steady state condition, the following
α = The compressibility of matrix skeleton which is defined as: equation holds for the interaction between matrix and fractures
(Warren and Root (1963)):
− 1 ∂ Vb (A5)
α=
Vb ∂ σ e  S  ∂ p1  p1 
 γ b − α n2  ∂ t =C sh K1  h − γ 
(A14)
   
and from the definition of water compressibility, we have:
Finally, the differential system for a naturally fractured aquifer results
∂ρ ∂ p1 ∂h (A6) in the following:
= ρβ= ργβ 2
∂t ∂t ∂t
1

1    S  ∂ p1 ∂h
This yield:  ρ ∇.( ρ K 2 ∇ h)=  γ b − α n2  ∂ t + ( 2 n2 β 2γ ) ∂ t (A15)
  

∂ ( ρ n )1 ∂p ∂ρ ∂n ∂p   S − α n  ∂ p1 
=C sh K1  h −
p1 
= n1 1 . + ρ 1 = ρ [ n1 β1 + α (1 − n2 ) ] 1   2 
γ 
∂t ∂ t ∂ p1 ∂t ∂t  γb  ∂t 
(A7)

APPENDIX B

Equation (A1) gives the following form, upon taking the derivatives:

       ∂ ( ρ n )1 ∂ ( ρ n ) 2 (B1)
∇.( ρ K 2 ∇ h=
) ρ K 2 ∇ 2 h + K 2 ∇ ρ .∇ h + ρ ∇ K 2 .∇ h= +
∂t ∂t

To determine the gradient terms we consider planar variation of functions for which we obtain:

 ∂ρ   ∂h   ∂ρ ∂h   ∂h 
   ∂x   ∂x   ∂h × ∂x   ∂x  ∂ρ   ∂h   ∂h  
2 2 (B2)
K 2 ∇ ρ .∇ h K 2 =
=  .  K 2  =  .  K 2    +   
 ∂ρ   ∂h   ∂ρ ∂h   ∂h  ∂h   ∂x   ∂y  
 ∂y   ∂y   ∂h × ∂y   ∂y 
     

976
Res. J. Environ. Earth Sci., 4(11): 962-981, 2012

And
 ∂K 2   ∂h   ∂K 2 ∂h   ∂h 
   ∂x   ∂x   ∂h × ∂x   ∂x  ∂K  ∂h
2
 ∂h  
2
(B3)
ρ ∇ K 2 .∇ h ρ  =
=  .  ρ  =  .   ρ 2    +   
 ∂K 2   ∂h   ∂K 2 × ∂h   ∂h  ∂ h   ∂ x   ∂ y  
 ∂y   ∂y   ∂h ∂y   ∂y 
     

Substituting the change in density from Eq. (A6) into (B2) and (B3) yields:

    ∂h  2  ∂h  2

K 2∇ ρ .∇ h K 2 β 2γρ    +  
= 
  ∂x   ∂y  (B4)
 
  ∂K   ∂h  2  ∂h  2 
ρ ∇ K 2 .∇ h ρ 2
=    +   
∂h   ∂x   ∂y  

Hence,

1     1 ∂K 2    ∂h   ∂h 
2 2
 (B5)
∇.( ρ K 2∇ h)= K 2 ∇ 2 h +  β 2γ +     +    
ρ   K 2 ∂h    ∂x   ∂y   

Combining Eq. (A12) and (B5), we obtain the following:

    ∂h   ∂h   1   S ∂h 
2
 1 ∂K 2  ∂ p1
2
(B6)
∇ 2 h +  γβ 2 +     +  =     − α n2  ∂ t + ( 2 n2 β 2γ ) ∂ t 
 K 2 ∂h    ∂ x   ∂ y   K 2   γ b  

Hence, the following non-linear system is obtained:

 ∂2h ∂2h  1 ∂K 2   ∂h   ∂h   1  S
2 2
 ∂p1 ∂h 
 2 + 2 +  β 2γ +    +  =    γ b − α n2  ∂t + ( 2n2 β 2γ ) ∂t  (B7)
 ∂x ∂ y  K ∂ h    ∂x  ∂
  
y K 2   

2

  S  ∂p1  p1 
  γ b − α n2  ∂t =Csh K1  h − γ 
    

Plugging Eq. (C2) (Appendix C) into Eq. (B7) gives us:

 ∂2h ∂2h   n  3(1 − 2ν )   ∂h 


2
 ∂h  
2

 2 + 2 + γ β2 + 3 1 + 2   c f +     +    =
 ∂x ∂y   9  En2   ∂x   ∂y   (B8)

 1  S  ∂p1 ∂h 
 =  − α n2  ∂t + ( 2n2 β 2γ ) ∂t 
K 2  γ b
  

  S  ∂p  p 
− α n2  1 =Csh K1  h − 1 
  γ b
  ∂t  γ 

To transform the equations above to act in polar coordinate we use the dimensionless groups presented in Table 1. The following partial
differential terms can be obtained by transformation of variables in polar coordinates (ξ, θ):

∂h Γ ∂h sin θ ∂hD 
−  cos θ . D −
= . 
∂x rw  ∂ξ ξ ∂θ  (B9)
∂h Γ ∂h cos θ ∂hD 
−  sin θ . D +
= . 
∂y rw  ∂ξ ξ ∂θ 

∂2h Γ ∂ 2 hD sin 2 θ ∂ 2 hD sin 2θ ∂ 2 hD sin 2θ ∂hD sin 2 θ ∂hD 


− 2  cos 2 θ .
= + . − . + . + . 
∂x 2
rw  ∂ξ 2 ξ 2 ∂θ 2 ξ ∂ξ∂θ ξ2 ∂θ ξ ∂ξ  (B10)

∂2h Γ ∂ 2 hD cos 2 θ ∂ 2 hD sin 2θ ∂ 2 hD sin 2θ ∂hD cos 2 θ ∂hD 


− 2  sin 2 θ .
= + . + . − . + . 
∂y 2
rw  ∂ξ 2 ξ2 ∂θ 2 ξ ∂ξ∂θ ξ2 ∂θ ξ ∂ξ 

977
Res. J. Environ. Earth Sci., 4(11): 962-981, 2012

∂h Γ ∂hD Q : Pumping rate, L3/T


= − .
∂t Θ ∂τ (B11) r : Radius, L
∂p1 γ Γ ∂ψ rw : Wellbore radius, L
= − . S : Storativity, dimensionless
∂t Θ ∂τ
S : Total skin factor, dimensionless
where, Г and Θ are defined as:
Sd : Mechanical skin due to near wellbore damage,
Q dimensionless
Γ=
2π T S ′ : Stress-dependent skin, dimensionless
rw2  S  t : Time, T
=Θ  + ( β 2 + c f − α )γ n2  T : Transmissivity, L2/T
K2  b 
(B12) α : Compressibility of aquifer skeleton, LT2/M
β 1 : Water compressibility in primary porosity,
Substituting the terms above into Eq. (B8) and considering symmetry LT2/M
in angular direction, we obtain a simpler nonlinear differential system β 2 : Water compressibility in secondary porosity,
in polar coordinates Eq. (6):
LT2/M
 ∂2h 2 γ : ≅ 0.57721, Euler’s constant
 1  ∂h  ∂h  ∂h ∂ψ
 D
+   D − ε  D=  ω D + (1 − ω ) γ : Specific weight, M/L2T2
 ∂ξ 2
 ξ  ∂ξ  ∂ξ  ∂τ ∂τ

 ∂ψ λ : Inter porosity flow coefficient, dimensionless
(1 − ω ) =λ ( hD −ψ )

 ∂τ (B13)
ω : Fracture storage capacity ratio, dimensionless
ε : Elasticity parameter, dimensionless
APPENDIX C Φ(x) : Elasticity function, dimensionless
The change in intrinsic permeability of fractures, in petroleum
ν : Poisson’s ratio, dimensionless
engineering notations, is obtained as (Jabbari et al., 2011): ξ : Radial coordinate, dimensionless
ξ D : Aquifer size in the radial direction,
1 dk f  φf  3 (1 − 2ν )  (C1) dimensionless
=+
3 1   c f + 
k f dp2  9  Eφ f  ρ : Fluid density, M/L3
Ψ : Pressure decline in primary porosity,
where, dimensionless
k f = Fracture intrinsic permeability
P 2 = Fracture pressure τ : Time, dimensionless
Ф f = Secondary porosity
C f = Fracture compressibility which can be considered equal to β 2 ACKNOWLEDGMENT
(Appendix A)
v = Poisson’s ratio
E = Young’s modulus. The authors would like to thank the following
sponsors for their financial support: US DOE through
Using hydrology notations, we derived the following:
contract of DE-FC26-08NT0005643 (Bakken
Geomechanics), North Dakota Industry Commision
1 dK 2  n  3(1 − 2ν )  (C2)
3γ 1 + 2   β 2 +
=  together with five industrial sponsors (Denbury
K 2 dh  9  En2 
Resources Inc., Hess Corporation, Marathon Oil
NOTATIONS Company, St. Mary Land and Exploration Company
and Whiting Petroleum Corporation) under contract
b : Aquifer thickness, L NDIC-G015-031 and North Dakota Department of
cf : Fracture compressibility, LT2/M Commerce through UND’s Petroleum Research,
C sh : Shape factor, 1/L2 Education and Entrepreneurship Center of Excellence
E : Young’s modulus of elasticity, M/LT2 (PREEC).
H : Hydraulic head in the fractures, L
hD : Hydraulic head in fracture, dimensionless REFERENCES
h wD : Hydraulic head in fracture (at the wellbore),
dimensionless Al-Ajmi, N., M. Ahmadi, E. Ozkan and H. Kazemi,
h ws : Hydraulic head in fracture (shut-in), L 2008. Numerical Inversion of Laplace Transforms
K1 : Matrix hydraulic conductivity, L2 in the Solution of Transient Flow Problems With
K2 : Fracture hydraulic conductivity, L2 Discontinuities. SPE 116255, Study Presented at
n1 : Primary porosity, fraction the SPE Annual Technical Conference and
n2 : Secondary porosity, fraction Exhibition, 21-24 September, Denver, Colorado,
p1 : Matrix pressure, M/LT2 USA.
978
Res. J. Environ. Earth Sci., 4(11): 962-981, 2012

Bacca, R.G., 1984. Modeling flow in fractured-porous Dautriat, J., N. Gland, S. Youssef, E. Rosenberg and
rock masses by finite element techniques. Int. S. Bekri, 2007. Stress-Dependent Permeabilities of
J. Num. Math Fluids, 4(4): 337-348, April, DOI: Sandstones and Carbonates: Compression
10.1002/fld.1650040404. Experiments and Pore Network Modelings, SPE
Bai, M. and D. Elsworth, 1994. Modeling of subsidence 110455. Study Presented at the SPE ATCE, 11-14
and stress-dependent hydraulic conductivity for November, Anaheim, California, U.S.A., DOI:
intact and fractured porous media, Rock Mech. 10.2118/110455-MS.
Rock Eng., 27(4): 209-250, DOI: 10. 1007/ Deruyck, B.G., D.P. Bourdet, G.D.P. Part and
BF01020200. H.J. Ramey, 1982. Interpretation of interference
Barenblatt, G.I., I.P. Zheltov and I.N. Kochina, 1960. tests in reservoirs with double porosity behavior:
Basic concepts in the theory of seepage of Theory and field examples, SPE 11025. Study
homogeneous liquids in fissured rocks. PMM, Presented at the 55th Annual Fall Technical
24(5): 852-864. Conference and Exhibition, New Orleans, LA,
Bear, J., 1972. Dynamics of Fluids in Porous Media. Sep., 26-29, DOI: 10.2118/11025-MS.
American Elsevier Publishing Co., New York. Domenico, P.A. and M.D. Mifflin, 1965. Water from
Bear, J. and A.H.D. Cheng, 2010. Modeling low permeability sediments and land subsidence.
Groundwater Flow and Contaminant Transport, American Geophysical Union, Water Resources
Springer Dordrecht Heidelberg. London, New Research, 1(4): 563-576, DOI: 10.1029/WR001i00
York. 4p00563.
Bellman, R., R.E. Kalaba and J.A. Lockett, 1966. Farayola, K.K., M.T. Oluokun, V.O. Adeniyi and
Numerical inversion of Laplace transform: O.A. Fakinlede, 2011. Hydrocarbon flow
applications to biology, economics, engineering simulations in porous and permeable media using
and physics. American Elsevier Publishing Co. finite element analyses and discrete fracture
Inc., New York. network, SPE 150772. Study Presented at the
Boulton, N.S. and T.D. Streltsova 1977. Unsteady flow Nigeria Annual International Conference and
to a pumped well in a fissured water-bearing Exhibition, 30 July-3 August, Abuja, Nigeria, DOI:
formation. J. Hydrol., 35: 257-270, DOI: 10.2118/150772-MS.
10.1016/0022-1694(77)90004-X. Fetter, C.W., 2001. Principles of Ground-Water Flow,
Cappa, F., Y. Guglielmi, J. Rutqvist, C.F. Tsang and Applied Hydrogeology. 4th Edn., Prentice Hall, pp:
A. Thoraval, 2008. Estimation of fracture flow 114-149
parameters through numerical analysis of hydro Garcia, L.E.P., 2005. Integration of well test analysis
mechanical pressure pulses. Water Resour. Res., into a naturally fractured reservoir simulation. MS
44, W11408, DOI: 10.1029/2008WR007015. Thesis, Dep. of Petrol. Eng., Texas A&M
Cey, E., D. Rudolph and R. Therrien, 2006. Simulation University, TX, USA.
of groundwater recharge dynamics in partially Gerke, H.H. and M.T. van Genuchten, 1993. A dual
saturated fractured soils incorporating spatially porosity model for simulating the preferential
variable fracture apertures. Water Resour. Res., 42, movement of water and solutes in structured
W09413, DOI: 10.1029/2005WR004589. porous media. Water Resour. Res., 29(2): 305-319,
Chin, L.Y., R. Raghavan and L.K. Thomas. 2000. Fully DOI: 10.1029/92WR02339.
coupled geomechanics and fluid-flow analysis of Gutierrez, M., R.W. Lewis, I. Masters, 2001. Petroleum
wells with stress-dependent permeability. SPE reservoir simulation coupling fluid flow and
J., 5(1), 32-45, DOI: 10.2118/58968-PA. geomechanics. SPE REE, 4(3), 164-172, DOI:
Cooper, H.H. and C.E. Jacob, 1946. A generalized 10.2118/72095-PA.
graphical method for evaluating formation He, J.H., 1999. Homotopy perturbation technique.
constants and summarizing well-field history. Comp. Math. Appl. Mech. Eng. 178(3-4): 257-262,
Trans. Am. Geophys. Union, 27: 526-534. DOI: 10.1016/S0045-7825(99)00018-3.
Crump, K.S., 1976. Numerical inversion of Laplace He, J.H., 2000. A coupling method of homotopy
transforms using a Fourier series approximation. technique and perturbation technique for nonlinear
J. ACM 233: 89-96, DOI: 10.1145/321921.321931. problems. Int. J. of Non-linear Mech. 35(1): 37-43,
Da Prat, G., H. Cinco-Ley and H.J. Ramey Jr., 1981. PII: S0020-7462(98)00085-7.
Decline curve analysis using type curves for two Hsieh, P.A., 1983. Theoretical and field studies of fluid
porosity systems. SPEJ, 354-362 (June), DOI: flow in fractured rocks. Ph.D. Thesis, Univ. of
10.2118/9292-PA. Ariz., Tucson, pp: 200.
979
Res. J. Environ. Earth Sci., 4(11): 962-981, 2012

Ilk, D., P.P. Valko and T.A. Blasingame, 2005. Moench, A.F., 1984. Double-porosity models for a
Deconvolution of variable-rate reservoir fissured groundwater reservoir with fracture skin,
performance data using B-Splines, SPE 95571. Water Resour Res., 20(7): 831. DOI:
Presented at the SPE Annual Technical Conference 10.1029/WR020i007p00831.
and Exhibition, 9-12 October, Dallas, TX, DOI: Murdoch, L.C. and L.N. Germanovich, 2006. Analysis
10.2118/95571-MS. of a deformable fracture in permeable material. Int.
Iseger, P.D., 2006. Numerical transform inversion using J. Numer. Anal. Meth. Geomech., 30: 529-561,
gaussian quadrature. Probability Eng. Inform. Sci., DOI: 10.1002/nag.492.
20: 1-44, DOI: 10. 1017/ S0269964806060013. Najurieta, H.L., 1980. A theory for pressure transient
Jabbari, H. and Z. Zeng, 2011. A three-parameter dual- analysis in naturally fractured reservoirs. J. Pet.
porosity model for naturally fractured reservoirs, Tech., Trans. AIME, 269: 1241-1250, DOI:
SPE 144560, Study Presented at the SPE Western 10.2118/6017-PA.
North American Regional Meeting held in Nelson, R.A., 2001. Geologic Analysis of Naturally
Anchorage, Alaska, USA, 7-11 May, DOI: Fractured Reservoirs. Gulf Professional Publishing,
10.2118/144560-MS. Woburn, MA.
Jabbari, H., Z. Zeng and M. Ostadhassan, 2011a. Neuman, S.P., 2005. Trends, prospects and challenges
Incorporating Geomechanics into the Decline- in quantifying flow and transport through fractured
Curve Analysis of Naturally Fractured Reservoirs, rocks. Hydrogeol. J., 13(1): 124-147, DOI:
SPE 147008, Study being Presented at the SPE 10.1007/s10040-004-0397-2.
Annual Technical Conference and Exhibition held Neuzil, C.E. and J.V. Tracy, 1981. Flow through
in Denver, Colorado, USA, 30 Oct-2 Nov. fractures. Water Resour. Res., 17(1): 191-199.
Jabbari, H., Z. Zeng and M. Ostadhassan, 2011b. Onur, M., A. Satman and A.C. Reynolds, 1993. New
Impact of In-Situ Stress Change on Fracture type curves for analyzing the transition time data
Conductivity in Naturally Fractured Reservoirs, from naturally fractured reservoirs. SPE 25873,
ARMA 11-239, Study Presented at the US Rock Study Presented at the SPE Rocky Mountain
Mechanics/Geomechanics Symposium held in San Regional/Low Permeability Reservoirs
Francisco, CA, June 26-29. Symposium, Denver, Co, 12-14 April, DOI:
Jacob, C.E., 1940. On the flow of water in an elastic 10.2118/25873-MS.
artesian aquifer. Trans. Am. Geophys. Union, 21: Peters, R.R. and E.A. Klavetter, 1988. A continuum
574-586. model for water movement in an unsaturated
Karim-Fard, M. and A. Firoozabadi 2003: Numerical fractured rock mass, (Yucca Mountain, Nevada).
simulation of water injection in fractured media Water Resour: Res., 24(3): 416-430, DOI:
using the discrete-fracture model and the Galerkin 10.1029/WR024i003p00416.
method, SPE REE, 6(2): 117-126, DOI: Rodriguez, A.A., H. Klie, S. Sun, X. Gai and
10.2118/83633-PA. M.F. Wheeler, 2006. Upscaling of hydraulic
Kazemi, H., 1969. Pressure transient analysis of properties of fractured porous media: full
naturally fractured reservoirs with uniform fracture permeability tensor and continuum scale
distribution. Soc. Pet. Eng. J. 451-62, Trans. simulations. SPE 100057, presented at the
AIME, 246, DOI: 10.2118/2156-A. SPE/DOE Symposium on Improved Oil Recovery
Lods, G. and P. Gouze, 2008. A generalized solution held in Tulsa, Ok, 22-26 April, DOI:
for transient radial flow in hierarchical multifractal 10.2118/100057-MS.
fractured aquifers, Water Resour. Res., 44(12), Sageev, A., G. Da Prat and H.J. Ramey Jr., 1985.
DOI: 10.1029/2008WR007125. Decline curve analysis for double-porosity
Meza, O.E., V. Peralta, E. Nunez and P. Savia, 2010. systems. SPE 13630, Presented at the California
influence of stress field in the productivity of Regional Meeting, held in Bakersfield, California,
naturally fractured reservoirs in metamorphic
March 27-29, DOI: 10.2118/13630-MS.
basement: A case study of the san pedro field,
Snow, D.T., 1969. Anisotropic permeability of fissured
Amotape group, SPE 138946. Presented at the SPE
media. Water. Resour. Res., 5(6): 1271-1289, DOI:
Latin American and Caribbean Petroleum
Engineering Conference, 1-3 December, Lima, 10.1029/WR005i006p01273.
Peru, DOI: 10.2118/138946-MS. Stehfest, H., 1970. Numerical Inversion of Laplace
Min, K.B., J. Rutqvist, C. Tsang and L. Jing, 2004. Transforms Algorithm 368. Communications of
Stress-dependent permeability of fractured rock ACM, D-5, 13: 47-49.
masses: A numerical study. Int. J. Rock. Mech. Streltsova, T.D., 1976. Hydrodynamics of groundwater
Min. Sci., 41: 1191-1210, DOI: 10.1016/j. ijrmms. flow in a fractured formation. Water Resour. Res.,
2004.05.005. 12(3): 405-414.
980
Res. J. Environ. Earth Sci., 4(11): 962-981, 2012

Suri, P., M. Azeemuddin, M. Zaman, A.R. Kukreti and Theis, C.V., 1935. The relation between the lowering of
J.C. Reogiers, 1997. Stress-dependent permeability the piezometric surface and the rate and duration of
measurement using the oscillating pulse technique. discharge of a well using ground water storage.
J. Petrol. Sci., 17(3): 247-264, DOI: Trans. Am. Geophys. Union, 16: 519-524.
10.1016/S0920-4105(96)00073-3.
Theis, C.V., 1938. The significance and nature of the
Takashi, M., H. Koide and Y. Sugita 1995. Three
principle stress effects on permeability of cone of depression in ground water bodies. Econ.
Shirahama sandstone. Proceeding of 8th Geol., 33: 889-902.
International Congress on Rock Mech., Japan, pp: Warren, J.E. and P.J. Root. 1963. The behavior of
729-732. naturally fractured reservoirs. Soc. Pet. Eng.
Talbot, A., 1979. The accurate numerical inversion of J., Trans. AIME, 228: 245-55.
Laplace transforms. IMA J. Appl. Math, 23(1): Zhu, W., C. David and T.F. Wong, 1995. Network
97-120, DOI: 10.1093/imamat/23.1.97. modeling of permeability evolution during
Tao, Q., C.A. Ehlig-Economides and A. Ghassemi, cementation and hot isostatic pressing. J. Geophys.
2009. Modeling variation of stress and Res., 100(B8), DOI: 10.1029/95JB00958.
permeability in naturally fractured reservoirs using
displacement discontinuity method. ARMA 09-
047, study Presented at the 43rd US Rock Mech.
Symp. held in Ashville, NC, June 28th-July 1st.

981

Das könnte Ihnen auch gefallen