Sunteți pe pagina 1din 9

Production From Naturally Fissured

Reservoirs Intercepted by a Vertical


Hydraulic Fracture
K. Ben Naceur, * SPE, and M .... Economldes, * * SPE, Dowell Schlumberger

Summary. A model is developed to simulate the behavior of naturally fissured and anisotropic reservoirs intercepted by a hydraulically
induced vertical fracture of finite or infinite conductivity. The method is based on a numerical discretization of the formation, which
is represented by a double-porosity model, for which transient interporosity solutions for pressure and cumulative production are presented. It is shown that matching the pressure behavior for those formations with homogeneous isotropic models leads invariably to a fracture
length that is smaller than the actual fracture length.

Introduction
Most tight reservoirs are naturally fissured. Two limiting cases are
very small fissures that are irrelevant to flow (thus reducing the
reservoir to a homogeneous and, likely, isotropic medium) and massive, naturally fractured formations whose production may be affected significantly if the wellbore intersects one of these features.
Warren and Root, I using the work of Barenblatt et al., 2 presented an analysis of fissured reservoirs and introduced the concept of
the' 'two-porosity" system. The complete synopsis of all equations
governing the behavior of naturally fissured reservoirs was offered
by Aguilera 3 and Van Golf-Racht. 4 The implied assumption is that
two different interconnected flow behaviors coexist-one in the fissures and one in the matrix. The fissure characteristics, such as
permeability and porosity, are not affected by the reservoir drawdown. The nonstimulated fissured reservoir was described further
by Kazemi et al., 5 who allowed for transient interporosity flow.
This is in contrast to the previous work, I which assumed a
pseudosteady-state interporosity flow between the fissures and the
matrix. Kazemi et al.'s model was based on an analogy between
a two-layered and a fissured reservoir and used a numerical reservoir simulator to represent the transient phenomena accurately.
Later, Mavor and Cinco-Ley 6 incorporated the concepts of
wellbore-storage and skin effect that are necessary elements in the
interpretation of pressure-transient tests. Appendix A briefly outlines the most important elements in the mathematical developments.
Note that the two-parameter description of the reservoir (through
the interporosity-flow coefficient, A, and the storativity ratio, w)
allows the simulation of both extensively fissured and homogeneous formations.
The hydraulic fracturing of fissured reservoirs requires special
attention during treatment design and execution and during
posttreatment-performance analysis. During execution, the presence of fissures would affect fracture geometry for several reasons:
(1) major natural fissures may lead to a fracture-path change,?
causing problems for proppant transport; (2) high permeability of
fissures results in thief zones where leakoff of fracturing fluid is
enhanced, and depending on the magnitude of the fissures, it is possible that leakoff control (without suitable diverting agents) may
not work (Le., filter cakes from polymer residues may be impossible to form); and (3) for limestone formations, and during acid fracturing, fissures ma~ enhance wormhole growth.
Cinco-Ley et al. provided a complete calculation of verticalfracture behavior in the absence of fissures, giving a basis for
production-performance analysis of finite-conductivity fractures.
Earlier, Gringarten et al. 9 provided the basis for analyzing largeconductivity fractures in homogeneous reservoirs.
It is essential that the term "naturally fractured" be analyzed.
In this paper, the term "naturally fractured well" is used only if
the well behaves, for a sustained period of time, in the manner described by Gringarten et al. 9; Le., when pressure difference is
plotted vs. time on log-log coordinates, it should form a straight
'Now at Schlumberger Cambridge Research. U.K.
"Now at Leoben Mining U.
Copyright 1989 Society of Petroleum Engineers

sso

line with a slope equal to 0.5. Instead of the term "naturally fractured reservoir, " in this work, such heterogeneous reservoirs shall
be called "fissured."
Houze et al. IO studied the behavior of a well penetrated by an
infinite-conductivity fracture in a fissured reservoir. Their work
provides the motivation for the case of a finite-conductivity fracture within a fissured formation.
Permeability anisotropy is particularly important in the performance analysis of fractures that intersect a well. Such an anisotropy
may result from depositional anisotropy or from the effect of earth
stresses (salt domes). Branagan et al. II pointed out the impact of
anisotropy on fracture performance by arguing that a fracture, presumed to operate under an average permeability (obtained from a
pretreatment well test), would lead to a PI increase that would be
far lower in an anisotropic situation. The direction of maximum
permeability generally would be perpendicular to the minimum
horizontal stress and likely would coincide with the direction of
a hydraulically induced vertical fracture.
Fig. I is a schematic of the general model used in this study to
simulate the effects of hydraulically induced vertical fractures, with
finite or infinite conductivity, on the performance of fissured and
anisotropic reservoirs. The open- and closed-fissure distribution depends on the stress anisotropy. More fissures are closed normal
to the maximum stress, and result in a permeability anisotropy with
the maximum value in the direction of the maximum stress. The
model is used primarily to predict pressure behavior during drawdown and production increase after stimulation. Fig. 2 denotes the
contributions of this work within a framework of past work.
The works of Gringarten et al. 9 and Cinco-Ley and SamaniegoV.12 were used to interpret infinite- and finite-conductivity fractures, respectively, in isotropic and homogeneous reservoirs. Houze
et al. 10 superimposed a fracture of infinite conductivity in a fissured reservoir. Branagan et al. II presented the implications of a
fractured anisotropic formation. This work allows for a fracture
of finite conductivity in both infinite and closed reservoirs and limits
the possible solutions by transforming the definitions of dimensionless pressure and time. In addition, finite- and infinite-fractureconductivity solutions for fissured formations are presented.

Model
The model used in this study is based on a finite-difference, pointcentered, Cartesian discretization of the formation and thus allows
for the simulation of a heterogeneous anisotropic reservoir. * Doubleporosity behavior is modeled from Kazemi et al. 's5 analogy of a
layered reservoir vs. a fissured medium. (See Boulton and
Streltsova3 for a complete discussion on the analogy.) Only the
higher-permeability blocks are allowed to flow into the well. A detailed presentation of the model parameters that are used to simulate the performance of a reservoir with a vertical fracture and a
description of their relation to fissure and matrix properties are given
in Appendix A. The vertical fracture is accounted for by modifying the transmissibilities of the blocks that include the fracture. Par'Settari, A.: "Farewell: A ThreeDimensional Reservoir Simulator," unpublished
documentation, Simtech Consulting Services Ltd., Calgary, 1987.

SPE Formation Evaluation, December 1989

Oimensionless time,

to

Fig. 3-Pressure response of an Infinite-conductivity fracture


In a fissured medium (from Houze et a/. 10).

5
4
Fig. 1-Conceptual diagram of open- and closed-fissure distribution leading to permeability anisotropy as a result of
stress anisotropy.

3
Q
0.

o~~~~~~~~~~~~
2
3
4
3
2
10 5 16
16
10
16'
10
10
10

to
Fig. 4-Semilogarlthmic plot for a fissured medium intercepted by a vertical, infinite-conductivity fracture: effect of A and
",=0.1.

conductivity derivatives, which were obtained with an inverse


Laplace transform.
Typical values for the layered-reservoir parameters that were used
to generate dimensionless data in the analog dual-porosity model
correspond to two orders of magnitude ratio between the height
of the two zones and to four orders of magnitude difference in permeabilities. Note that the highest-layer permeability (which is different from the fissure permeability) should always be significantly
lower than the fracture permeability to obtain meaningful results.

Fig. 2-Block diagram of solutions to various models in heterogeneous or anisotropic reservoirs included in this work.

ticularly important is a precise discretization of the reservoir, near


the crack tip and perpendicular to the fracture, caused by the high
transmissibility contrast between the fracture and reservoir gridblocks. The implicit formulation used in the model allows for good
stability with respect to the timestep scheme.
The reservoir model allows for the study of infinite and no-flow
outer boundaries. To eliminate the effects of the varying total compressibility, small values of fluid compressibility were entered into
the PVT description and the rock compressibility was held constant. Either a constant flowing pressure or a constant production
rate can be simulated. In the first case, cumulative production rates
are generated. Finally, the evaluation of the pressure derivatives
is performed numerically with a spline library. The obtained pressure derivatives were generally much smoother than the infiniteSPE Formation Evaluation, December 1989

Infinite-Conductivity Vertical Fracture


In a Fissured Reservoir
Houze et al. 10 considered the problem of a uniform-flux fracture
in an infinite-acting, homogeneous reservoir. Their technique uses
a line-source solution combining Green's functions as given by de
Swaan-O.14 for pseudosteady-state flow. The infinite-conductivity
pressure solution is derived from the uniform flux by calculating
the pressure at a specific point within the fracture. Lancaster and
Gatens 15 later used the results to develop a well-testing methodology for these wells.
Houze et al.'slO solution was derived in the Laplace domain, and
Stehfest' s l6 numerical Laplace inverter was used to develop type
curves of dimensionless pressure vs. dimensionless time as shown
in Fig. 3. Constant-flux- and infinite-conductivity-fracture solutions
were generated. In the latter case, a numerical discretization of the
fracture was tried; it was abandoned, however, because it required
a very large amount of numerical inversions to generate the solutions. Instead, Houze et al. used Gtingarten et al.'s9 finding for
a homogeneous medium, which assumes that the pressure drop in
551

'i5
:;0
E

a.

::10

a.

(/)0
(/).-

~o

c::l

0"0

.0
(/)~

<=0.

C])

Ci

Dimensionless time, to
Fig. 5-Semllogarlthmlc plot for a fissured medium Intercepted by a vertical, Infinite-conductivity fracture: effect of wand
). = 1.

Fig. 6-Dlmensionless cumulative production for a fissured


medium Intercepted by an Infinite-conductivity fracture:
w=0.1.

an infinite-conductivity fracture is the same as the pressure drop


measured at 0.732xF in a constant-flux fracture. This assumption
was corroborated through independent means used in this study with
a numerical simulator. The dimensionless variables used in the analysis are as follows.

lowed by a transition period that is dictated by the magnitude of


>-. (in this case, >-.= 102 ), and finally results in a combined matrix
and fissure flow that follows the curve for a fracture in a homogeneous medium. Fig. 4 shows the behavior of fissured reservoirs with
an infinite-conductivity fracture plotted on a semilog plot. The effect of >-. can be seen for a constant w (0.1). For a given reservoir,
small values of >-. correspond to small hydraulic-fracture lengths
(Eq. 4) or small >-'0. Thus, the characteristic two-parallel-straightline response as shown by Warren and Root l is evident. For larger fracture lengths, fracture behavior becomes the dominant feature with the early square-root-of-time relationship. In other words,
the total system response of the system becomes that of the fissures.
Fig. 5 is a semilogarithmic construction similar to Fig. 4 that allows w to vary with a constant >-. (equal to one). As Fig. 5 shows,
large values of w (large fissure porosity) would tend to diminish
the offset of the two straight lines by reducing the impact of a clearcut transition period. On the other hand, for values of w-O, the
early-time pressure behavior indicates a long transition zone. For
the extreme case where the porosity of the fissures becomes very
small compared with the total porosity of the system (i.e., when
w-O), the system should be described only in terms of lumpedmatrix and fissure properties.
For a fissured medium, and when permeability anisotropy is not
considered, the impact of wand >-. quantities can be assessed by
examining the cumulative production curves. Fig. 6 shows the ef-

PD=

tD=

............................. (1)

14 1. 2qBJ.I
0.OOO264kf t

2' . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (2)

(<pJ.lct)f+maxF

QpD =

ltD

qD( r)dr, .............

(3)

o
and >-'=>-'o(x;'lr~) . ..................

(4)

The interporosity flow constant, >-., for the stimulated system


should be distinguished from the constant for the nonstimulated
reservoir, >-'0. Fig. 3 from Houze et al. lO demonstrates the expected behavior of a fracture in a fissured formation. The dashed line
follows a pressure path (assuming no wellbore storage). The left
portion represents fissure flow (for this example w= 10- 2 ) fol-

10
C])

102

>

015
:; 0

E c.
::10
Q c
(/)0
(/).-

10

a.

1-

C])_
Q

c::l

0"0
.- 0
(/)~

co.
C])

Ci

161
102
103 3
10

1()2

1()1

10

102

10 3

Dimensionless time, to

Fig. 8-Dlmensionless pressure for an infinite-conductivity


fracture in an anisotropic homogeneous medium: effect of
permeability ratio.

Fig. 7-Dlmenslonless cumulative production for a fissured


medium intercepted by an Infinite-conductivity fracture: ). = 1.
552

SPE Formation Evaluation, December 1989

10

Ia.

,a.

to
Fig. 9-Normallzed dimensionless pressure for an Infiniteconductivity fracture In an anisotropic medium wHh an equivalent fracture length.

fect of A on the dimensionless cumulative production for a well


intercepted by an infinite-conductivity fracture for a constant w.
With a typical value (0.1) for the storativity parameter, w,17 Fig.
6 shows that A has relatively little effect on the production performance for values of A between 10 - 2 and 104 . Fig. 7 shows the effect of w on the dimensionless cumulative production, giving a
relatively general type curve for the infinite-conductivity case. Increasing values of w, for a given AO and a given fracture length,
result in increasing cumulative production.

Anisotropic Reservoirs
As Fig. 1 shows, a fissure distribution, resulting from a stress
anisotropy, also should induce a permeability anisotropy. Because
earth stresses are c.ompressive in nature, randomly distributed fissures should close preferentially in a direction normal to the maximum horizontal stress. This closure leaves a permeability
anisotropy, where the maximum value is in the direction of the maximum stress, parallel to the direction of a hydraulically induced fracture. As a result, two conclusions are drawn: a hydraulic fracture
in a naturally fissured formation, with distinct stress anisotropy,
would result in certain unique flow characteristics because of the
permeability anisotropy, and such an anisotropy (where the principal axis of the maximum permeability parallels the fracture) is
the least desirable. This will be demonstrated below. First, planar
depositional anisotropy in a nonfissured reservoir is examined.
A planar anisotropic system is characterized by the direction and
the value of the principal permeabilities, kx and ky (see
Earlougher 18 for a discussion). The system can be transformed into
an equ!valent isotropic system by use of an elliptic set of coordinates,
assummg that the x-axis corresponds to the direction oflarger permeabilities. 19 ,20

XI=~,

...................................... (5)

yl=~, ...................................... (6)


and k=.Jkxk y , ..................................... (7)
where k is the permeability likely to be measured from a singlewell test.
~imilar defInitions apply in the case of a double-porosity representatlOn. 1 The problem of determining the influence of anisotropy
on the performance of a reservoir intercepted by a hydraulic vertical fracture can be solved by using the previous definitions. It will
be ,assumed here, for simplicity, that the fracture direction is parallel
to one of the principal permeability directions.

Fig. 10-Effect of reservoir-permeability anisotropy on a flnH8conductivity fracture: F CD = 5.

and tD=0.OOO264ktlq,p.cr

}eq' ......................... (9)

The new equivalent fracture half-length,

(k

XFeq=XF~ ~=XF (ky)


kx

xFeq'

is defined as

'A

......................... (10)

kx

Fig. 8 compares the effects of a 4OO-ft [122-m] infiniteconductivity fracture for three cases: (1) kx = 1 and ky =0.2 md;
(2) kx=0.2 and ky=l md; and (3) kx=ky=l md (isotropic). The
case where kx S ky is not realistic and is used here for comparison
only. Most actual cases will correspond to the case where kx 2!ky .
As Fig. 8 shows, the pressure drop is the largest for the least favorable case of kylkx=0.2. The three curves cannot be superposed on
the basis of permeability modifications alone, even if anyone of
them is shifted 45 0 However, using Eq. 10 to modify the fracture
length will allow such matching (see Fig. 9). The implications in
terms of the effectiveness of the hydraulic fracture were analyzed
qualitatively by Branagan et al. I I With the new definitions introduced above, the solutions shown in Fig. 8 for different anisotropy ratios collapse into one curve, as shown in Fig. 9.
The concept of equivalent fracture length allows a quantitative
prediction with the classic type curves for infinite-conductivity fractures. 9 The graph in Fig. 9 shows a slope equal to 0.5 at early
time, which is indicative of the infinite-conductivity-fracture
response. The consequence of this is important for well testing of
stimulated anisotropic reservoirs by type-curve matching because
the calculated apparent fracture length will not be the actual hydraulic length. Additional testing (interference) is required to estimate
the directional permeabilities and, therefore, the real fracture length.
In most situations, kx (par~llel to the fracture) will be larger
than ky (hence, larger than k), so the matched fracture length will
be shorter than the real fracture dimension.
Suppose that in a postfracture well test, an apparent fracture length
equal to 500 ft [152.4 m] is calculated. If a permeability anisotropy ratio equal to 0.2 is determined, then (with Eq. 10) a real fracture length approximately equal to 750 ft [228.6 m] may be
extracted. This may explain partially the invariably shorter-thanexpected fracture lengths that are obtained from posttreatment tests.
Finite-Conductivity Fractures in a Homogeneous Anisotropic
Reservoir. For finite-conductivity fractures in homogeneous isotropic reservoirs, the dimensionless fracture conductivity, F CD, as introduced by Agarwal et at. ,21 can be_redefined as an equivalent

dimensio~::: frFlre conductivity, F CD:

~finite-C~nduc~ivity .Fractures in a Homogeneous AnisotropIC ReserVOIr. DImenSIOnless pressure, PD, and time, tD, for a
given i~ite-conductivity fracture ofhalf-lengthxF in an isotropic
reservOIr were presented by Cinco-Ley et al. 8 For anisotropic formations, those definitions should be replaced by

PD=khApI141.2qBp. ............................... (8)


SPE Formation Evaluation, December 1989

F CD =

~k;=kFb
F
--

- jfkxF

......................... (11)

kyXF

kx

553

D. Cinco-Ley at al. 8 ; FeD -5, w= 1

4-

x This ~~I:9p~.:.. ~ __ _
U ~~_e~_a~gq~~~"'_~~ __ _
~ This Mod!I~=:.or:,~~~..:.~ _ ~" __

310

FeD
FeD

0.2

2
Q
Q.

c.

I
o

-2

TTTTTIT-T-rTTn~'-lTnTrTl--T-"-'1Tnr---J- r-nTTTrr

0.0001 0.001

0.01

0.1

1016~4"""""'UJ.W101-;:-3:-'--'-...........10.t,-2...........L.U.OJ10w..,-1,.-J--'~uI.-...........uw110--'-...........wJ10

---T-,I-IlflT!

10

100

to

to

Fig. 11-Comparlson between the results of this model and


published solutions In the limiting cases.

Fig. 12-Dimenslonless pressure for a fissured reservoir Intercepted by a finite-conductivity fracture: '" = 0.1 and A = 1.

Fig. 10 compares the three anisotropy cases given in Fig. 8 for


a finite-conductivity fracture. Even with the new dimensionless variables, the curves cannot be superposed for early times, and they
collapse for later times. The isotropic case and the kylkx =0.2 case
in Fig. 10 show the characteristic bilinear-flow response of a finiteconductivity fracture with a slope of 0.25. In most common situations (i.e., when kx is larger), using the average permeability determined from well testing, the matched fracture length will be shorter
than the a~tuallength, while the dimensional conductivity, kFb F,
based on k, will be larger than the actual value.
Previous relations have an effect on the determination of fracture parameters from well testing. For a fracture with a small storage
capacity, Cinco-Ley and Samaniego-V. 12 showed that an early
bilinear regime prevailed. The only parameter that can be determined is the fracture's dimensional conductivity, kF bF (assuming
that there is no wellbore storage). For high-conductivity fractures
or after the bilinear-flow regime ends, a linear-flow period holds
for which the fracture length can be estimated. Again, the value
determined should be related to tJ:!.e anisotropy ratio. Assuming that
a pretreatment test has provided kh values, the problem is to relate
the value of the matches to the true fracture dimensions. The problem is more complex here than it is for the infinite-conductivity

fracture, because, at early times, a match will not be unique. For


later times, when the curves collapse, a unique interpretation is possible with the definitions of Eqs. 10 and 11.

FlnlteConductlvlty Vertical Fracture In a


Fissured Reservoir
The previously considered case of an infinite-conductivity fracture
(i.e., no pressure drop is assumed along the hydraulic fracture) is
amenable to a quasianalytical solution, although the solution is actually derived with an approximation from the uniform-flux expression. Different models have been used in the literature to simulate
the effects of a finite fracture conductivity (typically for
F CD :s; 1(0). Cinco-Ley22 presents an excellent review of solution
techniques (up to 1982) and the expected pressure trends. For simplicity, this study assumes that there is no wellbore storage, no
damage around the fracture, and no lateral or vertical variations
in conductivity.
A finite-difference numerical model is used in this study for a
finite-conductivity fracture in a fissured medium. Only fissures can
flow into the fracture and, for the case without a fracture, the results agree with the transient interporosity solution (Kazemi et at. 5

40

7
6

FeD

0.2

0-

c.

~O~-4;;~1~O;3~~10~-2~::1:0;1~~~~~10~~~1~02'

to

to
Fig. 13-Semllogarithmlc response for a fissured reservoir
Intercepted by a finite-conductivity fracture: '" O. 1 and A= 1.

554

Fig. 14-Dimensionless cumulative production for a fissured


reservoir intercepted by a finite-conductivity fracture: '" = 0.1
and A=1.

SPE Formation Evaluation. December 1989

7
A 100 .........

10

'S.

:0
a.
o
a.

"-""~-FCD

100

FCD 10

21L--~--~~~~10~--~~~~~102

1(j~04

10 3

102 '

""16 1 ' """,I

w!

10

" "~'Cf

to

FeD
Fig. 15-Correlatlon of hydraulic fracture conductivity and
natural fissure Intensity: A (to 10 and w= 0.1).

and de Swaan-O.14). The vertical fracture is discretized into a set


of gridblocks, and special care is given to limit the effects of the
high variations in transmissibility between the fracture (which typically has a width on the order of 0.1 in. [0.25 cm] and a
permeability> 10 5 md) and the adjacent gridblocks in the
reservoir.
In the fissured reservoir with a finite-conductivity-fracture case,
there are three important parameters: w, A, and FCD ' A general
type curve for interpretation is not feasible because of the number
of combinations. However, it is possible to generate type curves
for a given FCD (low, intermediate, or high), and assuming that
wand AO (Eq. 4) are known from a pretreatment test, the fracture
length may be calculated (see Economides 23 for the homogeneous
case). This is demonstrated later in this paper.
Comparisons were made with other published solutions. Fig. II
shows that the results of this model agree closely with those of Houze
et al.1O and Cinco-Ley et al. 8 for the respective cases of infinite
conductivity (w=O.l and A= 1) and F CD =5 (homogeneous reservoir). The difference between this model's solution and Houze et
al. 's results, which essentially occur when the transition fissure--->
fissure + matrix phase, may result from the fracture storage terms
in the numerical model.
Fig. 12 is the dimensionless pressure solution for a fissured reservoir, with typical values for the fissure parameters (w = 0.1 and
A= 1), intersected by a finite-conductivity fracture with F CD values
ranging from 0.2 to 100. A comparison of these results with those
of Cinco-Ley et al. 8 shows an increase in the dimensionless pressure in the fissured medium for a given fracture conductivity. For
example, the curve of F CD= 10 in the fissured medium corresponds to a curve of F CD :$ 5 in the homogeneous reservoir. This
implies that the fracture's FCD is usually underestimated if the fissured nature of the reservoir is not considered during posttreatment
interpretation.
The analysis of early-time behavior is essential for well-testing
applications. Fig. 12 clearly shows that for a finite-conductivity
fracture, the bilinear-flow regime characterized by Cinco-Ley et
al. 8 still exists for the fissured reservoir. For a homogeneous formation the time at which the bilinear-flow regime ends is a function oc'the fracture conductivity and is given by Cinco-Ley22 :

tebID""O.IIF~D .................................. (12)


for F CD "23,
tebID ""0.02(FCD -1.5) -1.5 ......................... (13)
for 1.6:$FcD:$3,
and t eblD",,(4.551.JFCD -2.5)-4 ..................... (14)
for F CD :$1.6.
SPE Fonnation Evaluation, December 1989

Fig. 16-0lmenslonless pressure and pressure derivative


response for fissured reservoirs (w = 0.1 and A= 1) with finiteconductivity fractures.

For fissured reservoirs. a characteristic time, tm,. is given approximately by


AtDAo =::0.2 ...................................... (15)
for the nonstimulated case, and
AtDA=O.2 ....................................... (16)

for the stimulated case. The dimensionless time tDA corresponds


to the time of transition between the fissure-dominated behavior
and the total system response (fissure+matrix). Depending on the
respective values of A and F CD, the two regimes mayor may not
overlap. For example, for the case considered here (w =0.1 and
A= 1), the bilinear-flow regime ends before the transition period.
For shorter fractures, which lead to a small value of A (see Eq.
4) and. hence, a higher tDA' the bilinear region overlaps with the
transition period between the fissure and the total system response.
The semilogarithmic graph (Fig. 13) of pressure vs. time of the
naturally fissured reservoir with a finite-conductivity fracture follows the general trends of Warren and Root's 1 solution. However,
the two straight lines (one for the fissures and the other for the total system) are not parallel, which is similar to the behavior of the
infinite-conductivity fracture for A"2 I (Fig. 4). The early-time regime is masked and distorted by the vertical-fracture response. Latetime behavior, however, is reservoir-dominated, and the second
slope of the semilogarithmic construction leads to the calculation
of the fissure permeability.
Finally, Fig. 14 denotes the dimensionless cumulative production for a range of F CD for a hydraulically fractured, naturally fissured reservoir. Because w cannot generally be changed (taken here
as 0.1), the effect of A and F CD can be assessed by comparing the
cumulative production at a given dimensionless time (tD = 10 in this
case). Fig. IS shows a relationship between cumulative production and A for different values of FCD. A relationship between A
and F CD can be derived by intersecting the graph with a horizontal line that corresponds to the desired value of QpD. It can be seen
that for that particular value of tD, the effect of A becomes significant only for values of A< 1.

Pressure-Transient Interpretation
Fig. 16 is a graph of the dimensionless pressure and pressure derivative of a finite-conductivity fracture in a fissured medium (w =0.1
and A=I) for three FCD values (1,10, and 100). At early time and
for the lower F CD values, there is an evident slope of 0.25; for
the higher-conductivity fracture, the slope is 0.5. This is consistent for the pressure and pressure derivative, as observed by a number of authors. 8- 1O The fissured-reservoir behavior gradually
emerges with a derivative reaching a local minimum corresponding to the interporosity-transition flow. This derivative is less than
the expected value ofO.25 because the fracture influences the transition period. Finally, the total system reaches the infinite-acting
555

10

10

--_. Po
- - pbto

:0

.. 10-1

--_. Po

1. .0.1

- - pbto

10- ~:,-"........""'-::"-u.u""",,-:,-..............
10 -4
10-3
10 -2
10-1
2

10-3

10 -2

10

-1

10

10 2

,-:-,....L.oL"""'' ' ' ' ' ' ' ' ' ' '"' '-:'' ' ' 'u.uu.uJ

to

Fig. 17-Effect of ). on pressure and pressure derivative


response of a finite-conductivity fracture (FCD :: 5) In a fissured medium: w:: 0.1.

behavior with the flattening out of the derivative at a constant value


of 0.5. The impact of the variation in A is shown in Fig. 17. As
expected, the local minimum derivative "hump" is reached at a
later time, which is inversely proportional to A. The value of Atm., .
still remains largely constant and approximately equal to 0.2.
Finally, Fig. 18 shows the effect of storativity, w. These curves
were generated for F CD = 5 and A= unity . The curve for w = 1 corresponds to the Cinco-Ley et at. 8 solution. For smaller values of
A, the pressure drop increases as indicated by Houze et al. 10 for
infinite-conductivity fractures. The derivative response shows that
for increasing values of w, the hump becomes shallower. For w= 1
(homogeneous case), no hump is observed. Type curves were also
generated for different fracture penetration ratios (Fig. 19); the classic unit-slope behavior is obtained, starting approximately at
tD =0.25(xF lx e)2, where xe is the reservoir extent.

Behavior of Anisotropic Fissured Media. Warren and Root 1 also


considered the effects of anisotropy in a fissured medium. Assuming that the matrix is homogeneous, with a permeability kma' and
that the fissure-permeability tensor has its principal vectors parallel
to the x and y axes, with respective permeabilities kfx and kfy' the
average fissure permeability that would be determined from a well
test is again defined as
kf=.Jkfxkfy , ..................................... (17)

while the interporosity flow parameter, A, is calculated by using


the average permeability for the fissures. The same conclusions regarding the use of equivalent fracture lengths apply as in the case
of infinite-conductivity fractures.

10

to

Fig. 18-Effect of w on pressure and pressure derivative


response of a finite-conductivity fracture (F CD = 5) In a fissured medium: ).:: 1.

Application
For the interpretation of real data, certain assumptions and data
obtained from a pretreatment test are necessary. First, a reasonable idea for the expected fracture conductivity F CD is indicated. 23
Then, the values of AO, w, and kf may be obtained from a pretreatment test. From Fig. 15, a value of FCD may be approximated
from the qualitative trends and slopes. A graph combining the pressure derivative for various values of A (as shown in Figs. 16 and
17) may then be generated.
The superimposed pressure data can move only from right to left,
because the knowledge of the fissure permeability sets the dimensionless pressure for any real pressure difference, until the pressure and the derivative of the real data match a type curve. The
value of A may then be obtained. Note that the value of the real
time at which a hump in the pressure derivative is observed cannot, alone, lead to an estimate of xF' Eq. 4 can then be used to
calculate xF; this value can be corroborated from the time match.
From the choice of F CD , the product kFb F may be extracted.

Example Application for Pressure-Transient Interpretation. A


pretreatment test in a fissured formation led to the following estimates for the reservoir parameters (rw=O.406 ft [0.12 m], h= 10
ft [3 m], 1/>=0.055, ct =10- 5 psi- 1 [1.45 kPa- 1], B=l, and /L=1
cp [0.001 Pa's)): kf =12 md; Ao=0.2XIO- 5 ; and w=O.l. A
hydraulic-fracturing treatment designed to achieve a hydraulic halflength of 500 ft [152.4 m] was performed, followed by a posttreatment test at a constant production rate of 100 BID [15.9 m 3 /d].
Pressures were measured every 30 seconds, and the pressure derivatives obtained had a smooth behavior, with a hump approximately
10 hours after the beginning of drawdown.

10
FCD

1. A = 10

FCD = 1. A= 0.1

a.

to

Fig. 19-Closed-boundary effects of the behavior of an


Infinite-conductivity fracture in a fissured medium.

Time (min.!

Fig. 20-lnterpretatlon of a well test In a fissured reservoir


Intercepted by a finite-conductivity fracture.
556

SPE Fonnation Evaluation, December 1989

Early pressure-drawdown data indicated a quarter slope, suggesting that FCD should be less than 10. Bounds for the value of A were
found by limiting the investigation to lengths ranging from 200 to
1,000 ft [61 to 305 m], leading to Amax = 12 and AlDin =0.5 (Eq.
4). Thus, type curves are generated for three values of A (0.1, 1,
and 10); for clarity, we will consider only two values of dimensionless conductivity (1 and 10).
The match of real data with the type curve is done by choosing
any arbitrary value of the pressure and calculating the corresponding dimensionless pressure from Eq. I; e.g., for a pressure drop
of 100 psi [689.5 kPa], the corresponding dimensionless pressure
is 0.85. Then, shifting the data horizontally (keeping the pressure/dimensionless-pressure relationship constant), as shown in Fig.
20, results in a match with FCD = 1 and A= 1. The shape of the
pressure derivative and the position of the hump are essential for
this interpretation. From the time match (tD=0.95 and t=I,OOO
minutes or 16.7 hours) and the definition of the dimensionless time
(Eq. 2), a fracture half-length of 320 ft [97.5 m) is calculated. With
Eq. 4 and the matched value of A (equal to unity), a fracture halflength of 290 ft [88 m] may be extracted (the value of AO is obtained from a pretreatment test).
The fact that the calculated length was significantly lower than
the expected value could have resulted from a combination of the
fracture-height migration; leakotI through fissures, which led to
a treatment that was less efficient than expected; or unaccountedfor anisotropy in the reservoir. The fracture permeability-width
product value can be estimated from a match with the dimensionless conductivity value; in this case, the value is 3.840 md-ft [12.6
mdm].

Conclusions
1. The effects of anisotropy on an infinite-conductivity fracture
can be analyzed with the isotropic case and an appropriate change
of variables. The most likely situation, where a fracture is parallel
to the direction of the largest pemleability, leads to smaller production improvements. Also, in analyzing pressure drawdown after a treatment, ignoring the anisotropy can lead to the calculation
of smaller fracture lengths.
2. The infinite-conductivity fracture in a fissured medium shows
a chara('1eristic shape for pressure derivatives. A technique to iden
tify fracture lengths is provided. Also, the two-slope behavior characterizes the pressure drawdown for large fractures, and the early
square-root behavior dominates the flow characteristics of shorter
fractures.
3. Finite-conductivity fractures in a fissured medium also exhibit
an early-time, bilinear-flow regin1e for low conductivities. Ignoring the fissured reservoir character when performing a type-curve
match leads to fracture conductivities that are lower than the actual values.
4. The effects of w, A, and F CD on the pressure drawdown, pressure derivative, and cumulative production were described. For
posttreatment-testing purposes, type curves representing a range
of values for A and F CD are used.

Nomenclature
bF = fracture width, ft em]

B = FVF, RB/STB [res m 3 /stock-tank: m 3]


= rock compressibility, psi - I [kPa- 1]
c, = formation total compressibility, psi -1 [kPa -1]
F CI) = dimensionless fracture conductivity ratio
FCD = dimensionless anisotropic fracture conductivity ratio
h = height (formation and vertical fracture), ft [m]
k = reservoir average permeability, md
kf = fissure permeability, md
kF = hydraulic fracture permeability, rod
kma = matrix permeability, md
kx,ky = directional permeability in the x and y directions, md
m = slope of semilogarithmic constructioll, psi/cycle
[kPalcycle]
n = number of gridblocks
P = pressure, psi [kPa]
1lp = pressure drop, psi [kPa]

CJ

SPE Formation Evaluation, December 1989

PD =
q =
qD =
QpD =
rw =
t =
ID =
IDA =
IDAo =
leblD =

V=

x,y,z =
Xe =
XF =
XFeq =

=
'Y =

O!

o=

A=
AO =
jL

(TH
cf>
!.oJ

=
=

=
=

dimensionless pressure
production rate, STB/D [stock-tank m 3 /d]
dimensionless production rate
dimensionless cumulative production rate
wellbore radius, ft em]
time, hours
dimensionless time
dimensionless time at transition fissure/total system
dimensionless time after transition fissure/total
system
dimensionless time at the end of the bilinear flow
regime
volume of reservoir model, ft3 [m 3 ]
Cartesian coordinates
reservoir extent, ft em]
hydraulic fracture half-length, ft [m]
equivalent anisotropic hydraulic fracture half-length,
ft em]
interporosity flow shape factor
fluid gravity term
well terms in Eq. A-I
interporosity flow parameter after stimulation
interporosity flow parameter before stimulation
reservoir fluid viscosity, cp [Pa' s]
horizontal stress, psi [kPa]
porosity
storativity ratio

Subscripts

fissured system
fracture
matrix system
maximum
minimum
r = reference conditions
I = low-permeability zone
2 = high-permeability zone

F
ma
max
min

=
=
=
=
=

Superscripts

= average
/ = derivative and corrected values

Reference.
I. Warren, J.E. and Root, PJ.: "The Behavior of Naturally Fractured
Reservoirs," SPEJ (Sept. 1963) 245-55; Trans., AIME, 228.
2. Barenblatt, 0.1., Zheltov, Y.P., and Kochina, I.N.: "Basic Concepts
in the Theory of Seepage of Homogeneous Liquids in Fissured Rocks, "
J. Appl. Math. Mech. (1960) 852-64.
3. Aguilera, R.: Naturally Fractured Reservoirs, PennWell Books, Tulsa,
OK (1980).
4. Van Golf-Racht, T.D.: Fundamentals of Fractured Reservoir Engineering. Elsevier Science Publishers, Amsterdam (1982).
5. Kazemi, H., Seth, M.S., and Thomas, G.W.: "The Interpretation of
Interference Tests in Naturally Fractured Reservoirs With Uniform Fracture Description," SPE! (Dec. 1969) 463-72; Trans., AIME, 246.
6. MavoI', M.J. and Cinco-Ley, H.: "Transient Pressure Behavior of
Naturally Fractured Reservoirs," paper SPE 7977 presented at the 1979
SPE California Regional Meeting, Ventura, April 18-20.
7. Jeffrey, R.O., Vandamme, L., and Roegiers, J.C.: "Mechanical interactions in Branched or Subparallel Hydraulic Fractures," paper SPE
16422 presented at the 1987 SPE Low Permeability Reservoirs Symposium, Denver, May 18-19.
8. Cinco-Ley, H., Samaniego-V., F., and Dominguez-A., N.: "Transient
Pressure Behavior for a Well with a Finite-Conductivity Vertical Fracture," SPEJ (Aug. 1978) 253-64.
9. Oringarten, A.C., Ramey, H.I. Jr., and Raghavan, R.: "UnsteadyState Pressure Distribution Created by a Single Infinite-Conductivity
Vertical Fracture," SPEJ (Aug. 1974) 347-60; Trans, AIME, 257.
10. Houze, O.P., Home. R., and Ramey, H.J. Jf.: "Pressure-Transient
Response of an Infinite-Conductivity Vertical Fracture in a Reservoir
With Double-Porosity Behavior," SPEFE (Sept. 1988) 510-18; Trans.,
AIME,285.
11. Branagan, P.T., Cipolla, C.L., and Lee, S.J.: "Designing and Evaluating Hydraulic Fracture Treatments in Naturally fractured Reservoirs,"
557

paper SPE 16434 presented at the 1987 SPE Low Permeability Reservoirs Symposium, Denver, May 18-19.
12. Cinco-Ley, H. and Samaniego-V., F ..: "Transient Pressure Analysis
for Fractured Wells," SPEJ (Sept. 1981) 1749-66.
13. Boulton, N.S. and Streltsova, T.D.: "Unsteady Flow to a Pumped Well
in a Fi~sured Water-Bearing Formation," 1. Hydrol. (1977) 257-69.
14. de Swaan-O., A.: "Analytical Solutions for Determining Naturally FracturedReservoir Properties by Well Testing," SPEI (June 1976) 117-22;
Trans., AIME, 261.
15. Lancaster, D.E. and Gatens, J.M. ill: "Practical Well Test Analysis
Methods for Hydraulically Fractured Wells in Dual-Porosity ReservOirs," paper SPE 15924 presented at 1986 SPE Eastern Regional Meeting. Columbus, OH, Nov. 12-14.
16. Stehfest, H.: "Algorithm 368: Numerical Inversion of Laplace Transform," Comm. of the ACM (1970) 1, No. 13.
17. Aguilera, R.: "Well Test Analysis of Naturally Fractured Reservoirs,"
SPEI (Sept. 1987) 239-52.
18. Earlougher, R.J.: Advances in Well Test Analysis, Monograph Series,
SPE, Richardson, TX (1977) 5.
19. Papadopulos, I.S.: "Nonsteady Flow to a Well in an Infinite Anisotropic
Aquifer," Proc., Symposium on Hydrology of Fractured Rocks,
Dubrovnik (1965) 21-31.
20. Ramey, H.J. Jr.: "Interference Analysis for Anisotropic Formations-A
Case History," IPT (Oct. 1975) 1290-98; Trans., AIME, 259.
21. Agarwal, R.G., Carter, R.D., and Pollock, C.B.: "Type Curves for
Evaluation and Performance Prediction of Low-Permeability Gas Wells
Stimulated by Massive Hydraulic Fracturing," IPT(May 1979) 651-54;
Trans., AIME, 267.
22. Cinco-Ley, H.: "Evaluation of Hydrdulic Fracturing by Transient Pressure Analysis Methods, " paper SPE 10043 presented at the 1982 SPE
Int!. Petroleum Exhibit and Technical Symposium, Beijing, March
16-25.
23. Econornides, M.J.: "Observations and Recommendations in the Evaluation of Tcsts of Hydraulically Fractured Wells," paper SPE 16396
presented at the 1987 SPE Low Permeability Reservoirs Symposium,
Denver, May 18-19.

reservoir. The model, based on a point-centered, finite-difference


scheme, implicitly solves the fluid-flow equations. Typically, in
the simulations, nx =ny =25 and nz =5 to 10. Twenty pressure data
were generated for each time cycle to ensure smooth pressure derivatives. Finally, only the high-permeability zone communicates with
the well.

Appendix A-Reservoir-Simulator Model

and PD =27rkfh/q~(Pi -Pf)' ......................... (A-8)

The model is based on the analogy between a fissured reservoir


and a multilayered system. The geometry used in the simulator corresponds to a classic Cartesian representation. The reservoir model
takes advantage of the geometry and uses only one-fourth of the
reservoir horizontaliy and one-half of the vertical layers.
The (single-phase) 3D reservoir-model equations are

!-'(!)-qo=v. k(xyz) (Vp--yVz),


at

................ (A-I)

plJ

where 0 represents the well terms. The porosity, c/J, is related to


the pressure through

c/J=c/J, [1+cf(P-Pr)]' .. ,.......................... (A-2)


j

where the subscript r corre~iJOnds to reference conditions, and cf


is the rock compressibility. The vertical hydraulic fracture intersects the reservoir parallel to the x direction, and a refined grid
is used near the fracture tip, as well as across the fracture, because
of the large contrast in permeability between the fracture and the

558

Dimensionless Variables. The dimensionless variables for the nonstimulated fissured reservoir are defined with respect to the total
system (fissures and matrix).
The storativity ratio, w, and the interporosity flow parameter,
}..O, are given by
(Vc/Jct)f

w=-----=--- . .......................... (A-3)


(Vc/Jc t )j'+(Vc/JCt)ma
and }..o=a(kma/kf)r;, .............................. (A-4)
where a is the interporosity-flow shape factor. The governing equations for flow are

kf
apf
apma
-V2pr(Vc/Jct )f-+(Vc/Jct )ma-- .............. (A-5)
~

kma
apma
and a-(Pma-Pf)+(Vc/Jct)ma--=O . ............. (A-6)

at

For a hydraulically fractured reservoir, the dimensionless time


and pressure drop are given in a coherent system of units by
kft

..................... (A-7)

[(Vc/Jc t )f+ (Vc/Jc t )ma]JlXfo

Generating Equivalent Data for the Layered Reservoir. For the


layered system, the geometry and flow properties of each layer are
the relevant parameters. If the subscripts 1 and 2, respectively, relate to the low- and high-permeability zones and subscripts f and
T/UJ represent the fissured and matrix_systems, respectively, the analogy leads to the following relations (Kazemi et al. 5 ):

c/Jrh2c/J2/(hl +h 2), .............................. (A-9)


c/Jma=h1c/JI/(h l +h 2), ........................... (A-lO)
kr(k1h 1 +k2h2)/(hl +h 2), ...................... (A-ll)
and h=hl +h 2. .................................. (A-12)

SPEFE
Original SPE manuscript received for review March 23, 1988. Paper accepted for publica
tion Aug. 8, 1988. Revised manuscript received June 26, 1989. Paper (SPE 17425) first
presented at the 1988 SPE California Regional Meeting held in Long Beach, March 23-25.

SPE Formation Evaluation, December 1989

S-ar putea să vă placă și