SPE 75524
Time-Dependent Shape Factors for Interporosity Flow in Naturally Fractured
Gas-Condensate Reservoirs
G. Penuela, SPE, F. Civan, SPE, R. G. Hughes, SPE, and M. L. Wiggins, SPE, the University of Oklahoma
Copyright 2002, Society of Petroleum Engineers Inc.
This paper was prepared for presentation at the SPE Gas Technology Symposium held in
Calgary, Alberta, Canada, 30 April–2 May 2002.
This paper was selected for presentation by an SPE Program Committee following review of
information contained in an abstract submitted by the author(s). Contents of the paper, as
presented, have not been reviewed by the Society of Petroleum Engineers and are subject to
correction by the author(s). The material, as presented, does not necessarily reflect any
position of the Society of Petroleum Engineers, its officers, or members. Papers presented at
SPE meetings are subject to publication review by Editorial Committees of the Society of
Petroleum Engineers. Electronic reproduction, distribution, or storage of any part of this paper
for commercial purposes without the written consent of the Society of Petroleum Engineers is
prohibited. Permission to reproduce in print is restricted to an abstract of not more than 300
words; illustrations may not be copied. The abstract must contain conspicuous
acknowledgment of where and by whom the paper was presented. Write Librarian, SPE, P.O.
Box 833836, Richardson, TX 75083-3836, U.S.A., fax 01-972-952-9435.
Abstract
A rigorous interporosity flow equation incorporating a time-
dependent shape factor is derived and validated for improved
dual-porosity modeling of naturally fractured gas-condensate
reservoirs. The equation expresses the interporosity molar rate
in terms of the pseudo-pressure gradient in the matrix, fracture
surface area, matrix permeability, and a variable matrix-block
shape factor. This approach can accommodate the flow
directed from matrix to fractures in a simulator that represents
the permeability of the interconnected fractures by a tensor
inside each grid-block. This feature distinguishes this model
from the popular sugar-cube approach to modeling naturally
fractured reservoirs.
Compositional simulation is performed to verify the flow
equation using the time-dependent shape factor. Numerical
experiments with various matrix-block sizes indicate that the
shape factor varies with time but converges to values derived
by Lim and Aziz (1995). The average matrix-pressure location
in the matrix block shifts from near the fracture face to the
block center as the fluid flows from the matrix into the
fracture. This phenomenon indicates that neglecting the time
dependency of the shape factor can introduce significant errors
in numerical simulation of naturally fractured reservoirs. The
time dependency can contribute significantly to fluid
production once the pressure front moves through the
reservoir along the highly permeable fracture network. This
phenomenon is not considered in present commercial
simulators that use pseudo-steady state factors.
The model equations are expressed in dimensionless form
so they can be readily integrated into current simulators. The
dimensionless time used in the proposed method includes
pseudo-functions that capture multiphase effects of the gas-
condensate systems. Applications to single- and multiphase
black-oil systems are also discussed.
Introduction
Numerical simulation of naturally fractured reservoirs has
received significant attention and its application has increased
in recent years with the advent of highly efficient computers.
Much of the research on naturally fractured reservoir
modeling has focused on accurately representing the matrix-
fracture fluid transfers. Various mechanisms, including gravity
and capillary effects,1-3 reinfiltration and capillary continuity
of the matrix blocks,4-7 and cocurrent/countercurrent
imbibition phenomena,8-10 have been extensively investigated.
However, in spite of the great level of current model
sophistication, the highly anisotropic and heterogeneous
nature of a fractured formation makes fractured reservoir
modeling a challenging task, frequently with uncertain results
in forecasting.
Typically, numerical simulation of naturally fractured
reservoirs assumes there are two continua, matrix and
fractures, within each grid-block. The flow equations are
written for each system with a matrix/fracture transfer
function to relate the loss or gain of matrix fluids to or from
the fracture. For single-phase fluid flowing through an
interconnected fracture system, the following governing
equation applies:
k f  φ  ~
∇
 µ
(
∆p f − ρg∆D ) = − ∂∂t  B f
 + q ...................(1)
 of 
where the fluid transfer rate per unit volume of rock, q~ , is
commonly calculated as a function of the pressure difference
between the matrix and fracture systems, matrix flow
capacity and matrix geometry considered through a constant
shape factor.
This paper investigates time-dependent effects on the
matrix block shape factors by defining a flow correction factor
based on Darcy’s law that can be applied to accurately
represent the matrix/fracture transfer (interporosity flow).
Initially, different approaches for estimating the shape factor
of a matrix block with a single-phase fluid are compared. A
time-dependent flow correction factor is then introduced, and
2 G. PENUELA, F. CIVAN, R. G. HUGHES, AND M. L. WIGGINS
a correlation to estimate the correction factor in terms of
dimensionless variables is derived. The application of the flow
correction factor to gas-condensate systems is presented by
means of appropriate pseudo-functions that account for the
multiphase behavior. Finally, an extension of flow correction
factors to multiphase black-oil systems is discussed.
Shape Factors
In traditional dual porosity/dual permeability simulation of
fractured reservoirs, the shape factor is taken as a constant
value obtained for an assumed matrix-block size and
geometry. A shape factor is necessary to compute the fluid
transfer from matrix to fracture during numerical simulation of
flow in fractured reservoirs by means of a numerical
discretization scheme, such as the finite difference method. In
general, fluid transfer models that use shape factors can be
grouped into two broad categories depending on the fluid
system in the reservoir. Single-phase models have been
applied in well test analysis while multiphase models have
been predominant in modeling secondary and tertiary recovery
in naturally fractured formations.11 For single-phase fluid
systems, two main groups of models can be found. The first
group only considers matrix geometry to compute the shape
factor, while the other uses only fracture geometry. In the
following, a discussion and comparison of models for single-
phase fluid systems is presented.
Matrix-based shape factors
Barenblatt et al.12 proposed a model for naturally fractured
reservoirs that is analogous to a model used for heat transfer in
a heterogeneous medium. They assumed that the outflow of
fluids from matrix blocks into the fractures was steady-state
and that the fluid transfer rate is a function of the viscosity of
the fluid, the pressure drop between the matrix and fracture
systems, and matrix-rock properties related to geometry
and porous interconnectivity in the matrix block. The fluid
transfer rate per unit volume of rock is calculated from the
following expression:
σ km
q~ =
µ
( )
p m − p f .................................................... (2)
where σ is a shape factor related to the specific surface of the
fractures, p m and pf are the average pressures in the matrix
and fracture domains, respectively, and q~ is the fluid transfer
rate between the matrix and fracture.
Several researchers have adopted Eq. 2 for modeling
interporosity fluid transfer in both dual-porosity and dual-
permeability models in single- and multiphase flow. However,
there is little agreement among the reported studies on the
value of the shape factor. Bourbiaux et al.13 presented a
comparison of shape factors found in the literature. Table 1 is
a modified version of the Bourbiaux et al. table,13 reporting
the numerical value of the product σL2 as calculated by
different researchers. L represents the fracture spacing for one
set of parallel fractures, the side length of a square formed
SPE 75524
between two normal sets of fractures, and the side length of an
isotropic cubic matrix block obtained from the intersection of
three normal sets of fractures.
Warren and Root14 presented an application of Eq. 2 in
their dual-porosity model for well test analysis by assuming
the interporosity flow to occur under pseudo-steady state
conditions. They proposed an analytical approximation to
estimate the shape factor assuming uniformly distributed
sets of parallel fractures. A schematic of three normal sets
of fractures is shown in Fig. 1. Warren and Root did not
present a derivation of their equation but expressed the shape
factor as:
4n(n + 2) ...............................................................(3)
σ=
L2
Kazemi et al.,15 Thomas et al.16 and Coats17 presented
various expressions for the shape factor that were verified
through numerical solutions of multiphase flow equations
similar to those proposed by Warren and Root for a single-
phase dual-porosity model. For instance, using a standard
seven-point finite difference formulation of a single-phase
flow problem,11 Kazemi et al.15 obtained the following shape
factor for a three-dimensional homogeneous matrix block:
 1 1 1
σ = 4 2 + 2 + 2  ....................................................(4)
 Lx Ly Lz 
where Lx, Ly, and Lz are the block lengths along the x, y, and z-
direction, respectively. Equation 4 inherently assumes a linear
pressure gradient between the fracture and the center of the
matrix block.18 As observed in Table 1, shape factors
computed from Eq. 4 have the lowest values. Hence,
this approach is likely to underestimate the efficiency of
the energy available to produce single-phase fluids from a
matrix block.
Lim and Aziz18 verified and extended the shape factors for
dual-porosity simulation presented by Kazemi and Gilman11
by combining the analytical solutions of the pressure diffusion
equation for various matrix blocks. By using matrix blocks of
regular shapes, Lim and Aziz18 obtained shape factors that
consider the geometry of the system and the physics of fluid
transfer without using the pseudo-steady state assumption. The
significance of their approach is that the derived shape factors
properly account for non-linear pressure gradients in the
matrix. For the general case of an anisotropic, rectangular
matrix blocks, they reported the following expression for the
shape factor:
π 2  k x k y k z  ................................................(5)
σ=  + + 
km  L2x L2y L2z 
where km is the geometric average matrix permeability.
Fracture-based shape factors
In the solution to the problem of flow of water in fractured
media, Duguid and Lee22 considered a porous medium with
 TIME-DEPENDENT SHAPE FACTORS FOR INTERPOROSITY
SPE 75524 FLOW IN NATURALLY FRACTURED GAS-CONDENSATE RESERVOIRS
several fractures, which were approximated as cylindrical
tubules of elliptical cross-sectional area. These cracks have
average dimensions, randomly distributed throughout the
matrix, and have porous walls that allow fluid transfer. A
representation of their model is shown in Fig. 2.
The formulation of the flow transfer term assumes 1D flow
from the matrix to the fracture with an interaction at the block
boundary where the pressure is assumed to remain constant
and equal to the pressure in the fracture. The following
approximation was given for the fluid transfer rate per unit
volume of rock:22
 ∞
σ km 
∑(− 1) ( p − p ) e
q~ =
µ 
(
pm − p f + 2 ) j
m i
−ξ j t
 j =1
∞ 
+2
∑( p f − pi ) e−ξ t  ........................................... (6)
j =1
j

where
2 k mπ 2 j 2 ............................................................. (7)
ξj =
φm µct L2f
and the shape factor is defined as follows:
16 φ f ............................................................... (8).
σ=
π wf Lf
Even though the fluid transfer rate as defined by Eq. 6
approaches to the steady-state solution given by Eq. 2, a direct
comparison cannot be made between this and previously
approaches discussed for the following reasons. The shape
factors presented in Eqs. 3 through 5 are expressed in terms of
block geometric properties while Eq. 8 is in terms of fracture
geometric properties (wf, Lf). Moreover, the shape factors
calculated from Eqs. 3 through Eq. 5 are independent
of fracture porosity. Finally, the shape factor according to
Eq. 8 is independent of fracture spacing, which is an
important characteristic in determining block size in most
dual-porosity models.
The importance of models for single-phase fluid systems is
that they provide a boundary limit for the flow problem where
no effects from multiphase interactions are present. For an
ideal fluid transfer function, a single-phase flow model should
account for the appropriate effects of both the matrix and the
fracture as a system. Currently, most models only consider the
effect of either the matrix or the fracture geometry on fluid
transfer. There are no models available accounting for the
combined effect of the coupled fracture and matrix media
where each system has different geometrical characteristics.
Conceptual Model
Consider a fractured rock sample in a single-phase fluid flow
experiment where a pressure difference is applied in a certain
direction as shown in Fig. 3. As a consequence of the
3
anisotropic nature of the fractured rock, the fluid velocity
vectors are not necessarily parallel to the pressure gradient
direction. Instead, they tend to align with the main flow
channels formed by the interconnected open fractures.
Because the flow velocity vector is proportional to the
pressure gradient and the proportionality factor involves the
permeability tensor, the flow velocity may have a different
direction than the pressure gradient in a fractured media.23,24
Consequently, if the network of randomly distributed fractures
is represented by a set of parallel and continuous open
fractures such that the same flow velocity direction, flow rate
and pressure gradients are maintained, the same permeability
tensor would be obtained from the flow experiment for the
mathematical description of the system. This representation of
the fractured rock considers the matrix blocks as slabs of finite
lateral extent where 1D interporosity flow occurs, as
illustrated in Fig. 4.
This conceptualization of a fractured media implies that
once the main flow paths have been established through the
flow channels (interconnected macro and micro fractures) in a
rock, fluids in the matrix portion principally flow towards
these flow channels. All matrix “blocks” within a control
volume do not have uniform pressure gradients in all
directions. Instead they tend to develop fluid flow
perpendicular to the primary flow channels rather than running
parallel to them whereas the traditional idealization of
fractured media as a sugar-cube model is based on the
assumption that the pressure gradient is the same in all
directions within each matrix block in a grid-block. The same
assumption has been considered in multiphase models that
rely on the extension of the single-phase models when using
geometric shape factors. The present model assumes that the
pressure gradient is not the same in all directions because the
flow is normal to the flow conduits. However, this is not a
restriction as the model can also handle the dual-permeability
multi-dimensional flow.
Flow Correction Factor
The interporosity flow rate using a variable correction factor
for a single-phase system is determined in this section. Thus,
assume that the main flow is one-dimensional and laminar,
and therefore Darcy’s law can be used to express the
interporosity flow rate as follows:
k m ∂Φ ..............................................................(9)
q = −A
µ ∂x
where km is the average absolute permeability in the matrix
block, µ is the average fluid viscosity and Φ is the flow
potential. The total fracture surface area, A, is calculated
depending on the geometry of the fracture. For fractures with
surfaces that can be approximately represented as parallel
plates, the total fracture area available for interporosity flow is
given by:
4 G. PENUELA, F. CIVAN, R. G. HUGHES, AND M. L. WIGGINS
φf
A=2 V ................................................................. (10)
wf
where φf is the average fracture porosity computed in the
control volume V, and wf is the average fracture width.
Neglecting gravity effects for simplification, Eq. 9 can be
approximated in a finite difference form as:
k m ∆p ............................................................. (11).
q = −A
µ ∆x
Consider that the pressure drop, ∆p, responsible for the
interporosity flow rate, q, can be calculated as the difference
between the matrix and fracture fluid average pressures.
Further, assume that the two average pressure values are
separated by a distance equal to half the fracture spacing.
Therefore, it is necessary to introduce a flow correction factor,
Fc, into Eq. 11 as follows in order to correct the errors caused
by these assumptions:
k
q = Fc A m
( pm − p f ) ................................................ (12).
µ L
2 definition given by Eq. 14 and a single curve was obtained
This flow correction factor is similar to the shape factor used
in the present interporosity flow calculations for dual-porosity
modeling. While Fc is dependent on the flow geometry as are
the shape factors, it is independent of the matrix block size. In
addition, Fc varies with time, and for single-phase fluids its
value converges to the steady-state shape factor value reported
in previous studies.
To illustrate this point, numerical experiments were
performed on a cubic shape matrix block in contact with a
fracture along one side. Because of symmetry, only half of the
matrix-fracture system was simulated as described in Fig. 5.
Reservoir rock and fluid data are similar to those described
by Lim and Aziz.18 Table 2 presents data used to obtain
both numerical and analytical solutions to this 1D flow
problem. Figure 6 shows the 20×1×1 grid and reports x-
direction grid-block sizes (∆y=∆z=10 ft) employed for the
numerical simulation.
Pressure profiles inside the matrix block for different times
are plotted in Fig. 7. The average matrix fluid pressure has
been indicated as a dotted horizontal line that intersects with
the pressure curve from which the average pressure value was
computed. Figure 8 shows the average matrix pressure and
the location of this value as a function of time. It is observed
from Figs. 7 and 8 that the distance between the location of the
average pressure, ∆x, increases from zero at the fracture
surface up to a steady-state value that is less than the assumed
half fracture spacing. Therefore, the purpose of Fc in the
interporosity flow rate (Eq. 12) is to correct for the actual
location of the average pressure in the matrix.
The analytical solution (see Appendix A) of the average
pressure difference and interporosity flow rate in this 1D flow
problem were used to compute Fc, shown in Fig. 9.
Convergence to the shape factor obtained by Lim and Aziz18
SPE 75524
for one set of parallel fractures was observed. Thus, these
numerical studies indicate:
lim 4 Fc (t )
= σ .....................................................(13)
t→∞ L2
where L is the fracture spacing for one set of parallel fractures
and σ is the corresponding shape factor for that matrix
geometry. The input data presented in Table 2 were used along
with several values for matrix permeability and block size to
determine their effect on Fc (Fig. 10). Rapid convergence of
Fc to the steady-state value is observed in small blocks with
high matrix permeability. As block size increases (larger
fracture spacing) and matrix permeability (or fluid mobility)
decreases, time dependency of Fc becomes important. A
generalized curve for Fc that reflects this time dependency
is obtained by defining a convenient dimensionless time
as follows:25
km
tD = t ...........................................................(14).
φ m µct L2
The data obtained for Fig. 10 was replotted using the tD
(Fig. 11). An equation of the form:
C3
 C  .......................................................(15)
Fc = C1  1 + 2 
 tD 
was used to fit this curve. The asymptotic behavior should be
the analytical result of Lim and Aziz, thus, C1 was taken to be
equal to 2.47. The remaining constant values C2 = 0.0133 and
C3 = 0.5 were obtained by regression analysis (Fig. 12) with a
coefficient of regression R2=0.9992. The solid line curve in
Figs. 10 through 12 is the Eq. 15 correlation.
Gas-Condensate Systems
In this section, the applicability of the flow correction factor
correlation given by Eq. 15 for gas-condensate systems is
shown by introducing an appropriate dimensionless time. Gas
production from a naturally fractured gas-condensate reservoir
is impaired by retrograde condensation, which occurs when
the local pressure is reduced below the dew point. Flow of gas
from the matrix to the fracture is affected by the presence of
the liquid phase, which reduces the relative permeability to the
gas phase. Matrix pressure gradients are also affected by the
presence of saturation gradients resulting from the
condensation of liquids from the fracture-matrix interface to
the block center as pressures drop below the dew point
pressure. However, the gas-condensate fluid problem becomes
identical to the single-phase fluid problem when pseudo-
functions are defined to account for the multiphase effects
observed below of the dew point pressure of the gas.
Appendix B shows the theoretical equivalence between the
single–phase and gas-condensate fluid formulations and the
corresponding solutions for interporosity flow.
 TIME-DEPENDENT SHAPE FACTORS FOR INTERPOROSITY
SPE 75524 FLOW IN NATURALLY FRACTURED GAS-CONDENSATE RESERVOIRS 5

Darcy’s law is again applied to estimate the interporosity This approach was verified by using a compositional
flow rate where the reduction of relative permeability to the simulator whose formulation is described by Penuela.30 The
gas phase because of the presence of the liquid phase is simulator is a semi-implicit, non-Newton-Raphson, equation
considered by means of the relative permeability to the gas. of state (EOS)-based compositional, 1D radial reservoir
Neglecting gravity and capillary pressure effects, the model, which was modified to handle 1D linear flow.
interporosity molar flow rate can be calculated from the Numerical experiments were performed on a cubic shape
following expression: matrix block (L = 10 ft) in contact with a fracture along one
side. Because of symmetry, only half of the matrix-fracture
 k rg k  ∂p ................................. (16)
q t = − Ak m  ρ g + ρo ro  system was simulated using the 20×1×1 grid shown in Fig. 6.
 µg µ o  ∂x Matrix permeability and porosity were respectively k = 0.001
md and φ = 0.05%. Initial matrix pressure was set at the dew
The pseudo-pressure function defined by Jones and point pressure of the gas ( pdew = 6750 psia) and fracture
Ravaghan26 may be used to write Eq. 16 as follows: pressure was kept constant at pf = 6250 psia. Mixture 2 for the
∂p p fluid and set 2 for the relative permeability from Penuela30 are
q t = − Ak m ........................................................... (17) used. The fluid system is represented by means of a mixture of
∂x
five pseudo-components with parameters described in Table
where 3. The relative permeabilities were represented using Corey-
p type functions with coefficients interpolated between the
 k ro k rg 
∫
pp =  ρ o + ρg  dp ..................................... (18). immiscible and miscible limits depending on the capillary
 µo µ g  number evaluated at the matrix/fracture interface. The end-
pref points and exponents of the relative permeability curves for
The pressure gradient expressed in terms of the pseudo- the immiscible limit are given in Table 4, and for the miscible
functions in Eq. 17 may be substituted by the difference in the limit, linear relative permeabilities of saturation are assumed.
matrix and fracture average pseudo-pressures, whose locations For additional information regarding rock-fluid description
in the matrix block are separated by a distance L/2, if a flow and flow modeling, the reader is referred to Penuela.30
correction factor is introduced into Eq. 17 as follows: Simulator output data were used to compute the flow
correction factor, Fc, as a function of dimensionless time. To

q t = Fc Ak m
(p p − p pf ) ............................................... (19). further test this approach, several values of absolute
L permeability and block length were used. One single Fc curve
was obtained as observed in Fig. 13. Some deviations were
2
observed during initial simulation because of fluid
The flow correction factor, Fc, is defined by the Eq. 15 compressibility effects that are not properly captured by the
correlation with the following dimensionless time: implementation of Eq. 21 and the assumptions under which
km this equation was derived as indicated in Appendix B.
tD = t p ............................................................... (20)
φm L2 Parametric Study
where tp is an appropriate pseudo-time function. A parametric study was carried out to investigate the
The use of pseudo-time functions effectively simplifies the applicability of the present approach in multi-dimensional
complex flow problems. Lee and Holditch27 used a pseudo- flow in dual-permeability media. This parametric study was
time function for well testing in gas wells using type curves performed for a single-phase fluid system but its conclusions
developed for slightly compressible liquids in a homogenous are readily extended to gas-condensate systems.
reservoir. They theoretically analyzed the conditions under The proposed interporosity flow equation for a single-
which their pseudo-function linearizes the flow equation for phase fluid, Eq. 12, indicates that the main driving force for
single-phase gas. For gas-condensate systems, Penuela and fluid flow is the matrix pressure gradient. This
Civan28,29 applied the following pseudo-time function, which pressure gradient may or may not be uniform depending on
accounts for multiphase flow effects: either the matrix geometry or the permeability ratio (or
−1
heterogeneity degree31):
( )
t
 ∂ ρo So + ρ g S g
∫  k m .....................................................................(22).
tp = dt ..................................... (21). ωK =
 ∂p p  kf
 
tref
Figure 14 depicts fluid velocity vectors in the matrix for
where the pseudo-pressure function was defined by Eq 18. As
different pressure gradients. Figure 14a represents the matrix
indicated in Appendix B, the pseudo-time function given by
flow in a dual-porosity model. The inherent assumption is that
Eq. 21 can be used along with Eqs. 15 and 20 to compute the
the fracture is highly conductive compared with the matrix
total molar rate of interporosity fluid transfer in a matrix block
(the system has a very low permeability ratio,31 ωK << 1) and
that contains a gas-condensate fluid.
6 G. PENUELA, F. CIVAN, R. G. HUGHES, AND M. L. WIGGINS
1D flow perpendicular to the fracture surface is observed.
Figure 14b shows behavior observed in typical cases of dual-
permeability media because either the matrix is also fluid
conductive or the fracture has low fluid conductivity. This
case shows 2D flow where the main velocity vector
components are perpendicular to the fracture surface. Figure
14c is a special case of the dual-permeability media flow
pattern shown in Fig. 14b. Note that the velocity vectors
display a 2D flow behavior with principal components parallel
to the fracture. This regime was observed in few parametric
studies described below.
To investigate the effect of ωK on the flow correction
factor, a parametric study was performed using the fluid and
rock data reported by Thomas et al.5 and summarized in Table
5. Values shown in Table 6 for fracture porosity and fracture
permeability were calculated assuming the parallel plate
model to achieve different permeability ratios. A 2D
numerical simulation was performed using a modified version
of the BOAST-VHS program.32 The grid chosen was 20×20×1
which was equally divided in the y-direction (∆y=0.5 ft) and
had dimensions in the x-direction as indicated in Fig. 6.
Because of symmetry, only half of the cubic fracture-matrix
system was modeled. To simulate the fracture, grid-block
properties from Table 6 were used in blocks, indexed (1,1) to
(1,20). The single-phase flow across the fracture was
simulated by placing a fictitious well at block (1,1) with a
constant fluid injection rate and a well at block (1,20)
producing fluid at constant bottomhole pressure (pwf = pf). The
interporosity rate was calculated as the rate difference between
the injection and production ports.
Numerical solutions were compared with the analytic
solution (Eq. A-8), which represents the ideal behavior of a
dual-porosity medium where the fracture is infinitely
conductive (ωK ≈ 0). The effects of having different values for
ωK are shown in Figs. 15-17. Figure 15 shows pressure
differences between the fracture and average matrix pressure
as a function of dimensionless time (Eq. 13). The average
matrix pressure was calculated as the arithmetic mean of the
block pressures for all blocks except for those indexed by (1,1)
through (1,20). At low fracture permeability, the fracture is
not able to function as a high conductivity channel and
consequently 2D flow occurs in the matrix. This is the case
represented by Fig. 14c and is also shown in Fig. 16. At high
ωK, the initial pressure difference increases while fluids fill the
fracture (fracture pressurization). Fluids are then produced
simultaneously from the matrix and fracture. Figure 17 shows
that small values of permeability ratio lead to large values of
the interporosity flow rate while large vales for ωK lead to
smaller interporosity flow rates. The limiting case is ωK = 1
where matrix and fracture have the same fluid conductive
capacity. In this specific case, the average pressures in both
media would be the same and thus the interporosity flow rate
would drop to zero. Even though the assumption of 1D flow is
violated and interporosity flow rate is reduced at high ωK, the
flow correction factor shows small changes (Fig. 18).
SPE 75524
Therefore, the interporosity flow equation is applicable to both
dual-porosity and dual-permeability situations.
Discussion
The interporosity flow equation developed in this work offers
several advantages. The first advantage is related to the
physical meaning expressed in the proposed interporosity flow
equation. Assume that fracture surfaces can be approximated
by parallel planes. Hence, the fluid transfer rate per unit
volume of rock can be calculated by substituting Eq. 10 into
Eq. 12 to obtain the following equation:
q~ = 4 Fc
φ f km pm − p f ( ) ..........................................(23).
wf µ L
The interporosity flow rate computed from Eq. 23 is not
only a function of matrix geometry and petrophysical
properties expressed in Eqs. 2 through 5, but also considers
fracture characteristics, such as fracture porosity and width. In
previous works, the shape factor given by Eq. 8 considered
only the fracture geometry for the estimation of the
interporosity flow rate. To see the significance of using the
fracture surface area, consider Fig. 19, which shows two grid-
blocks containing two parallel fractures in the same rock
volume. In case a, the interporosity flow rate estimation is the
same regardless whether Eq. 2 or Eq. 23 is used. On the other
hand for case b, the interporosity flow rate computed from Eq.
23 is higher than case a because the fracture surface area is
higher for the same grid-block volume. Equation 23 would
predict twice the rate for a system that has twice the fracture
porosity, while Eq. 2 would predict the same rate for both
systems. The same conclusion is achieved regardless of the
type of fluid present in the matrix block.
Implementation of Eq. 12 to compute single-phase transfer
is straight forward because the same variables are available in
a dual-permeability, dual-porosity simulator. However, the
implementation of Eq. 19 in current compositional simulators
to estimate the interporosity molar rate requires the
computation of the average pseudo-pressure function as
defined by the following double integral:
L/ 2 p
 k ro k rg 
∫∫
1
pp =  ρ o + ρg  dp dx ......................(24).
L  µo µ g 
− L / 2 pref
The pressure distribution inside each matrix block in a grid-
block will not be available. Instead, the average pressure in a
grid-block is computed at every time step. To overcome this
difficulty, the average pressure may be used to compute the
pseudo-pressure function if Eq. 24 is approximated by the
following expression:
p
∫ f ( p)d p ..........................................................(25)
pp ≈
pref
 TIME-DEPENDENT SHAPE FACTORS FOR INTERPOROSITY
SPE 75524 FLOW IN NATURALLY FRACTURED GAS-CONDENSATE RESERVOIRS 7

( )
where f p is approximated as follows: Numerical experiments show that the flow correction factor
curve remains nearly unchanged in spite of pressure gradient
L/ 2 distortions. This fact validates the use of the proposed flow
 k rg 
()
f p =
1
L ∫ 
k
 ρ o ro + ρ g
µo
 dx
µ g 
correction factor to situations where low fracture permeability
(dual-permeability situations) and intersecting fractures are
−L/ 2
present. Time-dependency of the flow correction factor is even
more important in oil recovery by water injection in a
k ro k rg
≈ ρo +ρg .......................................... (26). naturally fractured reservoir. In this multiphase case, there is a
µo µg slow moving saturation gradient that propagates across the
matrix in addition to the rapidly moving pressure gradient.
The average values needed in Eq. 26 should be readily
available from the compositional simulator at the end of every Nomenclature
time step to explicitly compute the interporosity flow rate. A = Total fracture area, ft2
Performance of this approximation is shown in Fig. 20 where Ao = Half fracture area, ft2
the average pseudo-pressure function computed from Eq. 24 is Bo = Oil formation volume factor, rb/STB
also plotted using fluid data for mixture 2 given by Penuela.30 cf = Fracture compressibility, psi-1
Although the approximation introduces some errors, they are cm = Matrix compressibility, psi-1
reasonably insignificant and Eq. 24 may be used to compute ct = Total compressibility, psi-1
the average pseudo-pressure function from the average matrix C1, C2, C3 = Constants in the flow correction factor correlation
pressure and fluid composition available at the end of each Fc = Flow correction factor, dimensionless
time step in the compositional simulator. k = Absolute permeability, md
Application to co-current two-phase flow may be kr = Relative permeability, fraction
simulated by Eq. 19 with an appropriate dimensionless time L = Fracture spacing, ft
and if capillary forces can be neglected. This conclusion can Lf = Fracture length, ft
be drawn from similarities between the governing equations of n = Number of normal sets of fractures
gas-condensate systems (see Appendix B) and two-phase fluid q = Interporosity flow rate, rb/day
systems. However, in co-current flow in water-wet systems q~ = Interporosity flow rate per unit volume of rock,
capillary pressure plays an important role that may not be
day-1
neglected.8 The importance of capillary pressure effects can be
qt = Interporosity molar rate, lb-mole/day
better observed in counter-current flow, which in a water wet
p = Pressure, psia
system may be the most important energy source for
S = Phase saturation, fraction
imbibition. The water front moving slowly towards the matrix
Swc = Connate water saturation, fraction
block center would generate corresponding pressure gradients
t = Time, sec
that also move slowly, and would therefore create high flow
V = Bulk volume, ft3
correction factors that are strongly time dependent.34
wf = Fracture width, ft
φ = Porosity, fraction
Conclusions
This study introduced an interporosity flow equation that µ = Viscosity, cp
considers time-dependency effects due to the shifting of the ρ = Molar density, lb-mole/ft3
average matrix-pressure location in the matrix block from the Φ = Flow potential, psia
fracture face to the block center as the fluid flows from the σ = Shape factor, ft-2
matrix into the fracture when a finite-difference scheme is ωK = Permeability ratio, dimensionless
used for numerical solution. The proposed equation
incorporates fracture and matrix geometric characteristics Subscripts
while retaining fluid mobility in the matrix as the main b = Bubble point
restriction to single-phase flow. D = Dimensionless
For gas-condensate systems, the use of pseudo-functions is f = Fracture
necessary to apply the proposed correlation for interporosity g = Gas phase
flow rate estimation. The pseudo-pressure function includes i = Initial value
the effect of relative permeability reduction due to the m = Matrix
presence of the condensate while the pseudo-time function o = Oil or condensate phase
accounts for the fluid compressibility effects. An p = Pseudo-function
approximation to compute the integral for the average pseudo- x = x-direction
pressure was also discussed. y = y-direction
The importance of having an interporosity rate, that is z = z-direction
dependent on matrix pressure gradient, was also illustrated.
8 G. PENUELA, F. CIVAN, R. G. HUGHES, AND M. L. WIGGINS SPE 75524

Acknowledgments presented at the SPE Reservoir Simulation Symposium,

This research project has been supported in part by U.S. DOE Houston, Feb. 6-8, 1989.
Grant No. DE-AC26-99BC15212. 18. Lim, K.T. and Aziz, K.: “Matrix-Fracture Transfer Shape
Factors for Dual-Porosity Simulators,” J. Pet. Sci. and Eng.
(1995) 13, 169-78.
References 19. Quintard, M. and Whitaker, S.: “Transport in Chemically and
1. Livak, B.L.: “Simulation and Characterization of Naturally Fractured
Mechanically Heterogeneous Porous Media,” Adv. in Water
Reservoirs,” Prod., Reservoir Characterization Technical Conference,
Res., (1996) 19 (1), 29-60.
Dallas, Academic Press, New York City (1985).
20. Noetinger, B. and Estebenet, T.: “Application of Random Walk
2. Sonier, F., Souillard, P., and Blaskovich, F.T.: “Numerical
Methods on Unstructured Grid to Up-Scale Fractured
Simulation of Naturally Fractured Reservoirs,” SPERE (Nov.
Reservoirs,” paper presented the 1998 European Conference on
1988) 1114-22.
Mathematics of Oil Recovery, Peebles, Scotland, Sep 8-11.
3. Gilman, J.R. and Kazemi, H.: “Improved Calculations for
21. Noetinger, B., Estebenet, T., and Landereau, P.: “Up-scaling of
Viscous and Gravity Displacement in Matrix Blocks in Dual-
Double Porosity Fractured Media Using Continuous-Time
Porosity Simulators,” JPT (Jan. 1988) 60-70.
Random Walks Methods,” Transport in Porous Media, (2000)
4. Labastie, A.: “Capillary Continuity between Blocks of a
39 (3), 315-37.
Fractured Reservoir,” paper SPE 20515 presented at the SPE
22. Duguid, J.O. and Lee, P.C.Y.: “Flow in Fractured Porous
Annual Technical Conference and Exhibition, New Orleans,
Media,” Water Resour. Res. (1977) 13 (3), 558-66.
Sept. 23-26, 1990.
23. Greenkorn, R.A., Johnson, C.R., and Shallenberger, L.K.:
5. Firoozabadi, A. and Ishimoto, K: “Theory of Reinfiltration in
“Directional Permeability of Heterogeneous Anisotropic Porous
Fractured Porous Media: Part I – One-Dimensional Model,”
Media,” SPEJ (June 1964) 124-32.
paper SPE 21796 presented at the Western Regional Meeting,
24. Snow, D.T.: “Anisotropic Permeability of Fractured Media,”
Long Beach, California, March 20-22, 1991.
Water Resour. Res. (1969) 5 (6), 1273-89.
6. Barkve, T. and Firoozabadi, A.: “Analysis of Reinfiltration in
25. Penuela, G.: “Modeling Interporosity Flow for Improved
Fractured Porous Media,” paper SPE 24900 presented at the
Simulation of Naturally Fractured Reservoirs,” PhD
Annual Technical Conference and Exhibition, Washington,
dissertation, U. of Oklahoma, OK (in progress).
D.C., Oct. 4-7, 1992.
26. Jones, J.R. and Raghavan, R.: “Interpretation of Flowing Well
7. Tan, J.C.T. and Firoozabadi, A.: “Dual-Porosity Simulation
Response in Gas-Condensate Wells,” SPEFE (Sept.1988) 578-94.
Incorporating Reinfiltration and Capillary Continuity Concepts: Part I
27. Lee, W.J. and Holditch, S.A.: “Application of Pseudotime to
– Single Gridcell,” paper SPE 29113 presented at the SPE Reservoir
Buildup Test Analysis of Low-Permeability Gas Wells With
Simulation Symposium, San Antonio, Feb. 12-15, 1995.
Long-Duration Wellbore Storage Distortion,” JPT (Dec.1982)
8. Bourbiaux, B. and Kalaydjian, F.J.: “Experimental Study of
2877-87.
Cocurrent and Countercurrent Flows in Natural Porous Media,”
28. Penuela, G. and Civan, F.: “Gas-Condensate Well Test Analysis
SPERE (Aug. 1990) 361-368.
with and without Relative Permeability Curves,” paper SPE
9. Pooladi-Darvish, M. and Firoozabadi, A.: “Experiments and
63160 presented at the Annual Technical Conference and
Modeling of Water Injection in Water-Wet Fractured Porous
Exhibition, Dallas, Oct. 1-4, 2000.
Media,” JCPT (March 2000) 31-42.
29. Penuela, G. and Civan, F.: “Prediction of Gas-Condensate Well
10. Pooladi-Darvish, M. and Firoozabadi, A.: “Cocurrent and
Productivity,” J. Pet. Sci. and Eng. (2000) 28, 95-110.
Countercurrent Imbibition in a Water-Wet Matrix Block,” SPEJ
30. Penuela, G.: “Prediction of the Gas-Condensate Well
(March 2000) 3-11.
Productivity and Field Implementation Using a Compositional
11. Kazemi, H. and Gilman, J.R.: Multiphase Flow in Fractured
Model,” MS thesis, U. of Oklahoma, OK (1999).
Petroleum Reservoirs, in Bear, J. Tsang, C-F, and de Marsily,
31. Panfilov, M.: Macroescale Models of Flow Through Highly
G. (Eds.): Flow and Contaminant Transport in Fractured Rock,
Heterogeneous Media, Kluwer Academic Publishers, Dordrecht
Academic Press, Inc., San Diego, California (1993).
(2000) 11-13.
12. Barenblatt, G.I., Zheltov, I.P., and Kochina, I.N.: “Basic
32. Chang, M-M., Sarathi, P., Heemstra, R.J., Cheng, A.M, and
Concepts in the Theory of Seepage of Homogenous Liquids in
Pautz, J.F.: “Users Guide and Documentation Manual for
Fissured Rocks (Strata),” J. Appl. Math. Mech. (1960) 24, 1286-1303.
BOAST-VHS for the PC,” final report, Contract No. DE-FC22-
13. Bourbiaux, B., Granet, S., Landereau, P., Noetinger, B., Sarda,
83FE60149, U.S. DOE, Bartlesville, Oklahoma (Jan. 1992).
S., and Sabathier, J.C.: “Scaling Up Matrix-Fracture Transfer in
33. Crank, J.: The Mathematics of Diffusion, second edition,
Dual-Porosity Models: Theory and Application,” paper SPE
Claredon Press, Oxford (1975) 44-103.
56557 presented at the SPE Annual Technical Conference and
34.Penuela, G., Hughes, R.G., Civan, F., and Wiggins, M.L.: “Time-
Exhibition, Houston, Oct. 3-6, 1999.
Dependent Shape Factors for Secondary Recovery in Naturally
14. Warren, J.E. and Root, P.J.: “The Behavior of Naturally
Fractured Reservoirs,” paper SPE 75234 presented at the
Fractured Reservoirs,” SPEJ (Sep. 1963) 245-55.
SPE/DOE improved Oil Recovery Symposium, Tulsa, April.
15. Kazemi, H., Merrill, L.S., Porterfield, K.L., and Zeman, P.R.:
13-17, 2002.
“Numerical Simulation of Water-Oil Flow in Naturally
35.
Fractured Reservoirs,” SPEJ (Dec. 1976) 317-26.
16. Thomas, L.K., Dixon, T.N., and Pierson, R.G.: “Fractured
SI Metric Conversion Factor
Reservoir Simulation,” SPEJ (Feb.1983) 42-54. bbl × 1.589 873 E−01 = m3
17. Coats, K.H.: “Implicit Compositional Simulation of Single- cp × 1.0* E−03 = Pa.s
Porosity and Dual-Porosity Reservoirs,” paper SPE 18427 ft × 3.048* E−01 =m
ft2 × 9.290 304* E−02 = m2
 TIME-DEPENDENT SHAPE FACTORS FOR INTERPOROSITY
SPE 75524 FLOW IN NATURALLY FRACTURED GAS-CONDENSATE RESERVOIRS 9

ft3 × 2.831 685 E−02 = m3 computed from Eq. A-5 and the interporosity flow rate from
md × 9.869 233 E−04 = µm2 Eq. A-7 into A-8 yields:
psi × 6.894 757 E+00 = kPa
∑ exp[ − (2 j + 1)π 2 t D ]
∞

2
*Conversion factor is exact. π j =0
Fc = ................(A-9)
[ ]
∞
4
∑ (2 j + 1) exp − (2 j + 1)π 2 t D
−2
Appendix A - Analytical Solution of the 1D Single- j =0
Phase Flow Problem
The pressure diffusivity equation for a 1D linear system that where the dimensionless time is defined by:
represents the single-phase fluid flow from the matrix to the km
fracture surface is given by: tD = t ........................................................(A-10)
φ m µct L2
∂pm k m ∂ 2 pm .................................................. (A-1)
= At large dimensionless time, Eq. A-9 converges to Fc = π2/4.
∂t φm µct ∂x 2
The asymptotic behavior of Fc can be also theoretically
and is subject to the following initial and boundary calculated by using the constant shape factor proposed by Lim
conditions18 (see Fig. 4): and Aziz18 for one set of parallel fractures. They eliminated
pm = pi , − L / 2 ≤ x ≤ L / 2 , t = 0 ..................... (A-2) the time parameter in Eq. A-7 by introducing a constant shape
factor into Eq. 2, resulting in a total interporosity flow rate
equation given by:
pm = p f , x = − L / 2 , t > 0 ................................. (A-3)
π 2 km
pm = p f , x = L / 2 , t > 0 ................................... (A-4).
q= V
L2 µ
( )
p m − p f ..........................................(A-11).

The analytic solution to Eqs. A-1 through A-4 is given The total interporosity flow rate can be calculated by
by:18,33 introducing V = LAo into Eq. A-11:
 (2 j + 1)2 π 2 k m t  π2
pm − p f k
( )
∞
8 q= LAo m p m − p f ......................................(A-12).
= ∑ exp −  .. (A-5).
L2 µ
pi − p f j = 0 (2 j + 1) π
2 2
 φ m µct L2 
This equation can be rearranged by approximating the total
In the dual-porosity model, it is assumed that the
fracture surface area as A ≈ 2Ao in order to get the following
interporosity flow rate per unit volume of rock is related to the
expression:
rate of accumulation in the matrix according to the following
relation:14
q=
π 2 km pm − p f
A
( )
.........................................(A-13).
∂ p m ........................................................ (A-6). 4 µ L
q~ = −φ m ct
∂t 2
Taking partial derivative of Eq. A-5 with respect to time Thus, comparing Eq. 12 with Eq. A-13, the steady-state value
and then substituting the resulting expression into Eq. A-6, the of Fc is given by:
total interporosity flow rate can be computed as: π2
Fc = ≈ 2.47 ......................................................(A-14).
) ∑ exp − (
 2 j + 1) π 2 k m t 
2
8V k m ∞ 4
q=
L2 µ
(
pi − p f
φm µct L2
 ....... (A-7)

j=0 
Appendix B - Analytical Solution of the 1D Gas-
where V is the bulk volume of matrix from which fluids are Condensate Flow Problem
produced into the fracture. In this Appendix, the flow correction factor for computing the
The flow correction factor is computed by solving Eq. 12 interporosity molar flow rate in a gas-condensate system is
for Fc, given by: analytically derived.
qµ L The model assumes a matrix block that is homogenous,
Fc = .............................................. (A-8). isotropic and uniform-thickness, bounded by two parallel
(
2 Ak m p m − p f ) fractures that are kept at constant pressure, pf, below the dew
point pressure of the gas, pdew (see Fig. 4). The model neglects
Consider two parallel fractures that limit a matrix block with a gravity, capillary, inertial and rock compressibility effects.
fracture spacing L. Each fracture is in contact with the matrix Consequently, the diffusivity equation for this 1D linear
along a surface Ao so that the total matrix volume is LAo and A system describing the gas-condensate flow from the matrix to
≈ 2Ao. Therefore, substitution of the pressure difference the fracture surface is expressed as follows:
10 G. PENUELA, F. CIVAN, R. G. HUGHES, AND M. L. WIGGINS SPE 75524

(
∂ ρo So + ρ g S g ) = km ∂  k ro
 ρ o + ρg
k rg  ∂pm 
  .. (B-1).
Thus, taking the partial derivative of Eq. B-8 with respect
to pseudo-time and then substituting the resulting expression
∂t φ m ∂x  µo µ g  ∂x  into Eq. B-10, the total interporosity molar rate can be
expressed as:
The initial and boundary conditions are given by Eqs. A-2
through A-4. In order to linearize Eq. B-1, apply the pseudo- ∞
 (2 j + 1) 2 π 2 k m t p 
pressure function defined by:26 qt =
8Vk m
L2
( p pi − p pf )∑ exp  −
 φ m L2
 ...(B-11).

p j =0
 k ro k rg 
pp =
pref
∫  ρ o
 µo
+ ρg  dp .................................. (B-2).
µ g 
Consequently, the flow correction factor is computed by
solving Eq. 18 for Fc, given by:
qt L
Penuela and Civan28,29 proposed the following pseudo-time Fc = ............................................. (B-12).
function to complete the linearization of Eq. B-1: (
2 Ak m p p − p pf )
−1
( ) If the pseudo-pressure difference (Eq. B-8) and the
t
 ∂ ρo So + ρ g S g
tp =
∫
tref


 ∂p p 

dt .................................. (B-3) interporosity flow rate (Eq. B-11) are substituted into Eq. B-
12, Eq. A-9 is obtained but with the following definition for
dimensionless time:
Substitution of Eq. B-2 and B-3 into B-1 yields: km
tD = t p ........................................................... (B-13).
∂p p 2
k ∂ p p ...................................................... (B-4) φm L2
= m
∂t p φ m ∂x 2 The use of the pseudo-time function introduces small
errors into Eq. B-4, which may be neglected. The reason for
Using the pseudo-function definitions given by Eqs. B-2 and
the errors is that in some cases the pseudo-pressure value in
B-3, the initial and boundary conditions expressed in Eqs. A-2
the left hand side in Eq. B-4 obtained from B-9 and B-10 is
through A-4 become:25
not the same as the pseudo-pressure value in the right hand
p p = p pi , − L / 2 ≤ x ≤ L / 2 , t p = 0 .................. (B-5) side, which may be the case at the beginning of the simulation
run. Note that the left hand side of Eq. B-4 requires
p p = p pf , x = − L / 2 , t p > 0 .............................. (B-6) the pseudo-pressure averaged over the matrix block while
it needs pseudo-pressure values as function of space in the
p p = p pf , x = L / 2 , t p > 0 ................................ (B-7). right hand side. For a detailed analysis on linearization by
means of a pseudo-time function, the reader is referred to Lee
Hence, the analytic solution to Eqs. B-4 through B-7 is and Holditch.27
identical to the solution given for the single-phase fluid TABLE 1 − VALUES OF THE GEOMETRIC FACTOR,
problem in Appendix A but in terms of the prescribed pseudo- σL2
functions:
∞ Matrix Geometry
p p − p pf  (2 j + 1) 2 π 2 k m t p 
∑ 8 Mathematical approximations Square-
= exp  −  . (B-8). Slab
Column
Cube
p pi − p pf
j =0
(2 j + 1)2 π 2  φ m L2 
14
Warren and Root (Analytic) 12 32 60
It is also assumed that the interporosity molar flow rate per 15
Kazemi et al. (Numeric) 4 8 12
unit volume of rock is related to the rate of mass accumulation
expressed in moles of hydrocarbons in the matrix according to 16
Thomas et al. (Numeric) -- -- 25
the following relation: 17
Coats (Analytic) 8 16 24

q~t = −φ m
(
∂ ρo So + ρ g S g ) ....................................... (B-9) Kazemi and Gilman (Analytic)
11
-- -- 29.6
∂t
18
Lim and Aziz (Analytic) 9.9 19.7 29.6
where density and saturation values are volume-weighted
average quantities over the total rock volume, V. Quintard and Whitaker (Numeric)
19
12 28.4 49.6
Then, substitution of the pseudo-time function (Eq. B-3) into 20,21
Eq. B-9 yields: Noetinger et al. (Stochastic) 11.5 27.1 --
13
∂ pp Bourbiaux et al. (Numeric) -- 20 --
q~t = −φ m ........................................................(B-10).
∂t p
 TIME-DEPENDENT SHAPE FACTORS FOR INTERPOROSITY
SPE 75524 FLOW IN NATURALLY FRACTURED GAS-CONDENSATE RESERVOIRS 11

TABLE 2 − DATA USED IN NUMERICAL EXPERIMENTS OF A SINGLE-PHASE FLUID

Matrix porosity, φm, fraction 0.0005

Matrix permeability, km, md 0.001

Total compressibility, ct, psi

-1 -6
3.5×10

Fluid viscosity, µ, cp 1.0

Initial pressure, pi, psia 1000

Fracture pressure, pf, psia 500

Fracture spacing, L, ft 10

Half fracture surface area, Ao, ft

2
100

TABLE 3 − PSEUDO-COMPONENT PROPERTIES FOR FLUID USED IN COMPOSITIONAL SIMULATION30

Pseudo-component
Mixture 2
pC1 pC2 pC3 pC4 pC5

Initial mole fraction, % 10.931 74.064 7.870 2.583 4.552

Molecular weight, lb/lb-mole 44.01 16.28 32.97 68.19 131.85

Critical pressure, psia 1071 663 687.7 503.6 375

o
Critical temperature, R 547.91 341.03 573.97 811.14 1450
3
Critical volume, ft /lb-mass 0.026 0.085 0.068 0.055 0.010

Acentric factor, dimensionless 0.2250 0.0110 0.1091 0.2289 0.4000

Shift factor, dimensionless -0.26 -0.1555 -0.0971 -0.0507 0.07104

Parachor, dimensionless 78 76.3 116.8 216.4 381.1

pC1 0 -- -- -- --

Binary interaction pC2 0.1 0 -- -- --

coefficients, pC3 0.12 0.1 0 -- --

dimensionless pC4 0.1 0.1 0 0 --

pC5 -0.02 0 0 0 0
12 G. PENUELA, F. CIVAN, R. G. HUGHES, AND M. L. WIGGINS SPE 75524

TABLE 4 − PARAMETERS FOR RELATIVE

PERMEABILITY30
Gridblock
Oil Gas boundary
Parameter
phase phase

End-point relative permeability, fraction 1 1

Fracture
Residual saturation, fraction 0.048 0.15

Exponent of the Corey function, dimensionless 2 2.12

TABLE 5 − DATA USED IN THE PARAMETRIC

STUDY L Matrix

Matrix porosity, φm, fraction 0.003

Matrix permeability, km, md 1.0 Fig. 1 − Grid-block in the dual-porosity model following Warren
14
and Root.
Matrix compressibility, cm, psi
-1 -6
3.5 × 10

Fracture compressibility, cf, psi

-1 -6
3.5 × 10

Connate water saturation, Swc, % 20

Oil density, ρo, lb/ft

3
51.14
Gridblock
Oil viscosity at pb, µo, cp 0.21 boundary
Slope of µo above pb, dµo/dp, cp/psi
-5
1.72×10

Oil formation volume factor at pb, Bo, RB/STB 1.8540

Fracture
Slope of Bo above pb, dBo/dp, RB/STB/psi
-5
-4.0×10

Initial pressure, pi, psia 5575

Fracture pressure, pf, psia 5565 Matrix

Bubble point pressure, pb, psia 5560

Fracture spacing, L, ft 10
wf Lf
Half fracture surface area, Ao, ft
2
100

Fig. 2 − Grid-block in the dual-permeability model following

22
Duguid and Lee.
TABLE 6 − FRACTURE PROPERTIES USED IN THE
PARAMETRIC STUDY

Fracture Grid-block Perm.

Bulk properties
width properties ratio
wf φf kf φ2 k2 ωK
ft % Darcys % Darcys

0.000108 0.00108 0.01 2.71 2.5 0.1

0.000503 0.00503 1 12.58 250 0.001

0.002336 0.02336 100 58.40 25000 0.00001

 TIME-DEPENDENT SHAPE FACTORS FOR INTERPOROSITY
SPE 75524 FLOW IN NATURALLY FRACTURED GAS-CONDENSATE RESERVOIRS 13

pf

pm
Flow rate
pf
(a)
pm

Pressure
drop

Naturally pm
fractured rock
Flow rate
(a) Natural flow channels

(b) pf

pm
Pressure
pf
drop
Idealized
fractured rock
pm

y
Fig. 3 − Representation of (a) naturally fractured and (b) idealized
fractured rock sample in a flow experiment. Fluid flows in
x pm
preferential paths conformed by interconnected fractures can be
mathematically described by permeability tensors.
L

(b) Idealized flow channels

Fig. 4 − One-dimensional flow towards flow channels in (a) natural

and (b) idealized fracture-matrix system.
14 G. PENUELA, F. CIVAN, R. G. HUGHES, AND M. L. WIGGINS SPE 75524

Symmetry plane 1100

t = 0.009 sec, p m = 998 psia

1000

t = 4.9 sec, p m = 950 psia

Matrix flow 900
Average pressure

800

Pressure, psia
t = 77 sec, p m = 800 psia

700

600 t = 346 sec, p m = 600 psia

Interporosity flow
500 t = 920 sec, p m = 510 psia
x ∆x
Fracture flow 400
L /2
y Symmetry plane 300
0 1 2 3 4 5

Fig. 5 − Numerically-modeled idealized fracture-matrix system. Distance from the fracture surface, ft

Fig. 7 − Pressure profiles and average pressure locations for 1D

numerical simulation with properties given in Table 2.

Matrix-block center line

1100 3
Distance from matrix-fracture interface, x

Distance from fracture surface, ft

Average matrix pressure, psia

900

pm
∆x
700

500

300 0
0 200 400 600 800 1000
Time, sec
Fracture Distance along fracture, y
Fig. 8 − Average matrix pressure and its location on the pressure
Fig. 6 − Schematic grid system discretization of one-half matrix profile curve (Fig. 7) as functions of time. Note the distance
block into grid-blocks for numerical solution. The fracture is the converges to a steady-state value.
first grid-block. Grid-block ∆x vales are 0.002, 0.004, 0.007, 0.012,
0.02, 0.04, 0.06, 0.09, 0.12, 0.15, 0.18, 0.21, 0.25, 0.3, 0.36, 0.44,
0.53, 0.64, 0.77, 0.815 ft.
 TIME-DEPENDENT SHAPE FACTORS FOR INTERPOROSITY
SPE 75524 FLOW IN NATURALLY FRACTURED GAS-CONDENSATE RESERVOIRS 15

40 100

L=10 ft, km=0.001 md

80
30 L=10 ft, km=0.01 md
Flow correction factor, Fc

Flow correction factor, Fc

L=10 ft, km=0.1 md
L=1 ft, km=0.001 md
Present study 60
L=20 ft, km=0.001 md
Lim and Aziz
20 Correlation

40

10
20

2.47
2.47
0 0
0.1 1 10 100 1000 1.0 E-6 1.0 E-4 1.0 E-2 1.0 E+0
Time, sec tD

Fig. 9 − Comparison of flow correction factors computed from the Fig. 11 − Flow correction factor curve as function of
analytic solution and from the constant shape factor reported by dimensionless time.
18
Lim and Aziz.

100
1.0 E+4

L=10 ft, km=0.001 md F c = 2.47(1 + 0.0133/ t D )

1/2
80
8.0 E+3
L=10 ft, km=0.01 md
Flow correction factor, Fc

L=10 ft, km=0.1 md

60 L=1 ft, km=0.001 md
6.0 E+3
L=20 ft, km=0.001 md
Fc 2

40
4.0 E+3

20
2.0 E+3
Simulated data from Fig. 11
Correlation
2.47
0
0.0 E+0
0.01 0.1 1 10 100 1000
0.0 E+0 5.0 E+4 1.0 E+5 1.5 E+5
Time, sec
1 / tD

Fig. 10 − Variation of flow correction factor for different fracture

spacing and average matrix permeability. Fig. 12 − Generalized correlation for the flow correction factor.
16 G. PENUELA, F. CIVAN, R. G. HUGHES, AND M. L. WIGGINS SPE 75524

50

40 L= 10 ft, k = 0.001 md
L= 10 ft, k = 0.01 md
L= 10 ft, k = 0.1 md
30 L= 1 ft, k = 0.001 md
L= 20 ft, k = 0.001 md
Fc

Correlation
20 10

10 8

2.47
0 6

pm - pf , psi
1. E-5 1. E-4 1. E-3 1. E-2 1. E-1 1. E+0
Analytic soln. (Eq. A-5)
tD ω K = 0.00001
4
ω K = 0.001
ω K = 0.1
Fig. 13 − Flow correction factor computed from compositional
simulation. Dimensionless time effectively reduces to one single
Fc curve data generated for different size matrix block and
absolute permeability. 2

(a)
0
1.0 E-4 1.0 E-3 1.0 E-2 1.0 E-1 1.0 E+0

tD

Fig. 15 − Pressure differences at different permeability ratio.

Analytic solution represents an ideal dual-porosity system.

(b)

Matrix
Fracture

(c)

Fig. 14 − Different types of matrix flow. Case a: Dual-porosity

model. Case b: Dual-permeability model. Case c: Fluid velocity
vectors observed at high permeability ratio.
 TIME-DEPENDENT SHAPE FACTORS FOR INTERPOROSITY
SPE 75524 FLOW IN NATURALLY FRACTURED GAS-CONDENSATE RESERVOIRS 17

Fig. 16 Two-dimensional flow experienced at high permeability ratio (ωK=0.1).

18 G. PENUELA, F. CIVAN, R. G. HUGHES, AND M. L. WIGGINS SPE 75524

30

24 (a) (b)
Interporosity flow rate, rb/day

Analytic soln. (Eq. A-7)

ω K = 0.00001
18 ω K = 0.001
ω K = 0.1

12

Fig. 19 − Two simulation grid-blocks containing a pair of parallel

6 fractures in the same rock volume. In Case b, fluids would be
produced faster from the matrix than in Case a because of a
higher total fracture surface area.

0
1.0 E-4 1.0 E-3 1.0 E-2 1.0 E-1 1.0 E+0
tD

Fig. 17 − Effect of permeability ratio on interporosity flow rate. At

high ωK, fluid flows almost equally into matrix and fracture,
therefore the interporosity flow rate is reduced.

10000
20

8000
16
pp, lb-mole.psi/ft3 /cp

Analytic soln. (Eq. A-9)

ω K = 0.00001
Fc , dimensionless

6000
12 ω K = 0.001
ω K = 0.1
4000
8

Exact Integral
2000
Approximation
4
2.47

0
0 1 10 100 1000
1.0 E-4 1.0 E-3 1.0 E-2 1.0 E-1 1.0 E+0
t, sec
tD
Fig. 20 − The average pseudo-pressure may be computed from an
approximate integral that uses the average values of fluid
Fig. 18 − Effect of permeability ratio on flow correction factor. mobility, which are functions of the average matrix pressure and
fluid composition already available in the compositional
simulator.