Вы находитесь на странице: 1из 12

SPE

6014

TransientPressureBehaviorfora WellWitha Finite-Conductivity


VerticalFracture
HEBER ClN(X) L.
F. SAMANIE(30,V.
N. DWINQUEZ A.

INSTITUTO MEX\CANO DEL P&ROLEO


MEXICO CITY, MEXICO

MEMBERS WE-AIME

ABSTRACT
A mathematical
model was developed
to study tbe
transient behavior o/a well with a finite-conductivity
vertical
/racture in an iniinite
slab reservoir.
For
values 0/ dimensionless
time 0/ interest,
tD 210-3,
the dimensionless
wellbore
pressure,
PwlD~ can be
correlated
by the dimensionless
group, wkl/xf k ,
and
where w, kf, and x are tbe width, permeability,
half length o/ t i e /racture,
respectively,
and k
represents
the {ormation perrneabi!ity.
Results
when
plotted
as a [unction of pwfD us
10g
tD
give,
for !urge t~, a 1. 151+iope
straight
semilogaritbmic
pressure
analysis
line;
hence,
methods
can be app!ied.
When p[ottcd in terms of
log pwjD vs log tD, a family o/ Curves 0/ cbaraCter.
istic shape result, A type-curve
matching procedure
can be used to analyze early time transient presszdre
and
jracture
data
to
obtain
the
formation
characteristics.

INTRODUCTION
Hydraulic fracturing is an effective technique for
increasing
the productivity
of damaged wells or
wells producing from low permeability fortnations.
Much research has been conducted to determine the
effect of hydraulic fractures on well performance
and transient pressure behavior. The results have
been used to improve the design of hydraulic
fractures. Many methods 1-14 have been proposed to
determine formation properties and fracture characteristics from transient pressure and flow rate data.
These methods have been based on either analytical
or numerical solutions of the transient flow of fluids
toward fractured wells. Recently, Gringarten et al.14
made an important contribution to the analysis of
transient
pressure data of fractured wells, They
presented
a type-curve analysis and tttree basic
solutions: the infinite-fracture
conductivity solution
(zero pressure drop along a verticai fracture), the
uniform flux solutiori for vertical fractures, and the
Original manu;!cript received AnSociety of Petro:eum Ertglneem
Feb. 17, 197?.
office Aug. 15, 1976. Paper mcepted for publication
Revised
nranuscrlpt
received
April 28, 1978. paper (SpE 6014)
first preaentcd
at the SPE-AIME
5 Itt Annual Fmll Technical
Conference
and Exhibition,
held in New Orleans,
Oct. 3-6, 1976.

0037-9999/78/0008-60
14$00.2S
@ 197s Society of Petroleum Englneerc of AIME
AUGUST,1~

uniform flux solution for horizontal fractures.


Although the assumption of an infinite fracture
conductivity is adequate for some cases, we must
consider a finite conductivity for large or very low
flow capacity fractures. Sawyer and Locke15 studied
the transient pressure behavior of finite-conductivity
vertical fractures in gas wells. Their solutions
cannot be used to analyze transient pressure data
because only specific cases were presented.
In this study, we wanted to prepare general
solutions for the transient pressure behavior of a
vertical
well intersected
by a /inite.conductivity
fracture. The solutions sought should be useful for
short-time or type-curve analysis. We also wanted
to show whether conventional
methods could be
applied to analyze transient pressure data for these
conditions.
A combination
of both methods, as
pointed out by Gringarten et aft, 14 should permit an
level
concerning
the
extraordinary y confidence
analysis of field data.
STATEMENT OF THE PROBLEM AND
DEVELOPMENT OF FLOW MODELS
The transient pressure behavior for a fractured
well can be studied by nnalyzing the solution of the
di~ferential equations that describe this phenomenon
To
with proper initial and boundary conditions.
simplify the derivation of flow models, the following
assumptions are made.
1. An isotropic, homogeneous, horizontal, infinite,
slab reservoir is bounded by an upper and a lower
impermeable
strata.
The reservoir
has uniform
k, and porosity, +,
h, permeability,
thickness,
which are independent of pressure.
2. The reservoir contttins a slightly compressible
fluid of compressibility,
c, and viscosity,
p, and
both properties are constant.
fractured
3. Fluid i~ produced through a vertically
/initer
well intersected
by a /ully penetrating,
of half length, Xp width, W?
conductivity
/racture
permeability,
kf, and porosity, @ . These fracture
characteristics
are constant.
F {uid entering the
wellbore comes only through the fracmre.
A system with these assumptions
is shown in
Fig. 1. In addition, we assume that gravity effects
are negligible and also that laminar flow occurs in
the system,
Under these conditions,
the flow
2s3

;>

,,;:

+4

phenomenon may be described by the diffusivity


the
equation in two dimensions. 16 TO facilitate
solution of this equation, two flow regions will be
considered (1) the reservoir and (2) the fracture.
FRACTURE

FLOW MODEL

Boundary conditions,
aPf

q#

zE- ~o-ziqi
and

fracture is considered as a homogeneous,


finite, slab, prous medium of height, h, half length,
xl, and width, w. Fluid enters the fracture at a rate
q (x, t) per unit of fracture length, and flow across
the edge of this porous medium is ne~ligible
because the fracture width is very mall compared
with the fracture length. lhe Iaat assumption allows
us to consider a linear flow in the fracture and
permits simulation of well production by a uniform
flux plane source of b and w, located at the wellbore
axis (Fig. 2).
Unsteady-state
flow in the fracture may be
described by the equation,
lhe

a2Pf

qffx.t)

+kf

~f~cft

kf

~=

ax2

mf
E- . X=X -O*

(2)

In Eq. 1, q x, t) is a source term that represents the


fluid flow / from the reservoir to the fracture.
The solution of Eq. 1 with initial and boundary
conditions Siven by Eq. 2 is expressed in dimen=
sionless form by this equation:

apf
at

O<%<%=........(l)
&

subject to the, following conditions.


Initial conditia,
pf(x,

t=o)

2n+l

= Ppo~x~xf

qfD(x8T)

-/

2n-1

(xD-x )2

WELLBORE

- [ 4(kf@ct/k$fc

f ~)

dx

d~ *

z~ ~

lAtPERM\ABLE
BOUND~RIES

[
I
[
t
I

I
i
I
I
I

i
I
[
I FRACTURI

where

(3)

Wpf(w)]

I
I
I
I
I

pf#~9

D)

141.2

qwB~

&

D =

--

0.000264

t.

kt

$W pf2

and
FIG.
1 FINITE-CONDUCTIVITY
VERTICAL
FRACTURE IN AN INFINITE SLAB RESERVOIR.

2qf(x
qfD(x90

,T)

FIG. 2W

FRACTURE

FLOW MODEL,

*
(4)

=
f

Eq. 3 gives the dimensionless


pressure &op in
the fracture at location XD and dimensionless
time
tD. his equation was obtained by applying Green
and source functions and the Newman product method
extensively discussed by Grin8arten and Ramey.22
SOCIETY OF PSTaOLWM

CNG1NES8S JOURNAL

RESERVOIR FLOW MODEL

and

The transient pressure behavior in the reservoir


may be studied by considering
the fracture as a
plane source of height, b, length, 2X1, and flux
density q@ t) (Fi8. 3). The dimensionless pressure
&op at any point in the reservoir may be obtained
from the following equation:

tD)

- qD(~,

qfD(x#~)

o . .

. .(8)

for

A combination of Eqs. 3, 5, 7, and 8 and use of


Poissons summation formula yields
o
(Iyx

qD(%,T)

cfDf

)*+yD2

4 (tD-T)

-[

-1

dx dc . w (5)

(tD-T)

*1
1 ~
7r2rlfD n=l n

tD+

-q ~ Dn2w2tD
I

Wpi-p(x,y,t)]
=

qD(x ,T)

141.2

D =

~BIJ

29.@L.Q~f

qw

and

L.

n.

-1

-nf

. . . . . . .,

-x)

. (6)

.,

...,

dx t d~

tJ)) 0 . 0

fDf

(7)

. . . . . . .,,

,,.

,.

(9)

where ~
= pD(xD}YD=O,

Dn2T2 (tD-~)

X=

Eq. 5 also was derived using Green and source


functions.
To solve Eqs. 3 and 5 simultaneously,
continuity
between the two flow regions must be established,
The dimensionless
pressure drop pff) (x~, t~) and
flux density qlD (XD, tD) in the fracture model must
eqUd
the dimensionless
pressure drop pD (xf)J yD$
tD) and flux density qD (xD, tD) on the plane
source
of the reservoir model, respectively.
That
is,
pfDtxD$tD)

(n7rxJ

l-e

where
P#~9Y~9tJ

cos

ft
= Txf @ct

and
k#c ~
..
%D = Wfc f t

PLANE SOURCE(FRACTURE)

%(x,t)

. . . . . . . .

(lo)

Eq. 10 is a Fredholm integral equation where the


Mknown is qfj (%D ~f$. CfDf is the dimensionless
fractw: storage capacity, and qfD is the dimensionless hy&aulic diffusivity of the fracture.
METHOD OF SOLUTION

FIG.
A1GUW. 1978

3 -

RESERVOIR FLOW MODEL.

Eq. 9 can be solved by discretization


in time and
space so thar the fracture is divided into 2N equal
segments (Fig. 4) and time is divided into K different
intervals.
It is assumed that fr !cture flux has a
stepwise distribution
in time r.:~d space. In other
words, the flux density qD ~,e of a fracture interval
w

is constant for a given segment i and time inteival


For a fracture segment j, Eq. 9 becomes

1.

1-1/2
N

Di

%,2-I = DL

D&l

+ erf

and

112Ti2
~f DAtK,

k-l

-e

~i,j%
n

-$

6i,jEi
[

The arguments
defined a;

II

of the erf and Ei functions

are

= j-i+l/2
2N

~i,j

~ j -j-i-l/2

B
where

= j+;y2
id

and

6~j=H#!L
9

12 H SEGMENTS

/.

# ,

)
1

123

FIG. 4-

z%

2N

1 ,.

FRACTURE DIVIDED INTO N EQUAL


SEGMENTS.

By writing Eq. 11 for all fracture segments, a


system of equations is obtained where the unknowns
are the g~i,t s. Solution of such a system for each
time interval produces values for the fracture flux
The dimensionless
pressure drop at
distribution.
any point of the system can be calculated by using
the discretized forms of Eqs. 3 and 5. Although the
theory presented here does not consider formation
damage near the fracture caused by fracturing fluid
SOCIETY OF PETROLEUM

ENGINEERS

JOURNAL

loss, the equations may be modified to include


variable iikin damage along the fracture.

A computer program was writtsn to determine the


flux distribution
and dimensionless
pressure drop
A sensitivity
analyais
was
along the fracture.
conducted to obtain accurate results.
We found
that soluticma do not change appreciably when more
than 20 segments are taken per fracture half length,
x . Results also indicated that the solutions were
/
accurate enough for practical purposes if at least
10 intervals were considered in each log cycle of
logarithm of dimensionless
time. Therefore, in all
cases studied, the fracture half len@h was divided
into 20 equal segments and 10 time intervals wete
taken in each log cycle of dimensionless time.
Cases were simulated for values of CfDf ranging
from 2 x 104 to 10-3 and values of q p from 10
to 108. Iheae ranges were based on published
fracture
characteristics
data.
Analysis
of the
results showed that as soon as most of the fluid
produced at the wellbore comes from the formntion
(i.e.,
the expansion
of the fracture system is
negligible),
solutions
can be correlated
by one
parameter that depends on CfDi and ql~ constants.
Fortunately, this holds for times of interest. This
correlating parameter was found to be

cfDf

fD

~kxf

. . ...*

(12)

An important feature of this variabie is that it does


not depend on the porosity and total compressibility
of the formation and fracture. It is essentially
the
dimensionless
fracture flow conductivity,
With regard to the symbol for this correlating
parameter, Ramey 23 suggested using a product of
two dimensionless variables, such as

fD

as

a T&

DISCUSSION OF RESULTS

kf

which may be expressed

capacity,

The use of the term capacity


is a misnomer. The
correct term is conductivity.
In the following,
the dimensionless
fracture conductivity
will be
considered as (k ~ . tu(D). Although there was a
constant n in t { e original correlating parameter
(see Eq, 12), and (tr/2) in Prats expression,
wc
&oppcd the constants for the sake of simplicity.
The solutions
obtained
in this study were
compared, where applicable, with solutions published
in the literature.
Results for a highly conductive
fracture (CfDf = 10 3, q ID = 107, and kfp wjD =
I@/m) ~;how excellent a8rcemcnt with the infiniteconductivity
solution
of Gringarten
et
al.lg
Differcnccs between rhc two solutions arc less than
1% for small values of dimensionless
time, and
less than 0.025% for other times of interest.
Fig. 5 shows the dimensionless
flux along the
fracture at different values of !D. For small values
of [D, the fhtx density is uniform along the fracture.
Also, for small times, the flow rate from the
formaticm into the fracture, is less than the total
WCII f14Dw rate. This results
from the storage
capacity
of the fracture.
For intermediate
and
large vtdues of t~, the well flow rate is generated
by the expansion of rhc system outside the fracture;
under these conditions,
the total area under each
flux density curve in Fig. 5 is equal to unity. And
finally, for large values of tD, the flux density
becomes stabilized
as discussed
by Gringarten
fracture. Fig. 5
et al. for an infinite-conductivity

7fD+.=

f
= ~

.
3iJ

lo
@kf
rack

It

.,04

and

t~>s
tB=2
The first is the relative fracture permeability and
the second represents
the dimensionless
fracture
width. Large values for the product (kfD tu~D) represent
highly conductive
fractures;
conversely,
small values represent fracturea of low conductivity.
Small values of the product may be caused either by
low fracture permeability or large /racture length.
For much of the following discussion,
we refer to

the condition of low fracture conductivity


conductivity.
remember that we mean a dimensionless
EitAer low fracture permeability
or long fracture
lcn@h, or both, may be the physical phe mena
involved.
Solutions for the atcady-stato
flow case were
co~related by Pratsls
using the relative fracture
AllOUST, 1970

t*rl~lo-4

-1

Ialo+

Ttlls STUDY
6RlM6ARTEN,~.

(Iatlnll. contucllsltp solution)


1

.2

,4
XD?*

.s

.8

FIIG. 5 - FLUX DISTRIBUTION AT VARIOUS


TIMES ALQNG A HIGHLY CONDUCTIVE
VERTICAL FRACTURE.
m

also shows the stabilized flux distribution presented


by Gringartcn et al. Good a8reement was found
between both solutions.
It is of interest to know the effect of fracture
flux density along
conductivity on rite stabilized
the fracture, Fig. 6 shows that for a highly
conductive ftacture (i ., kfD w fD 2 300), the flux
density is high at the portions of the fracture away
horn the wellbore.
As fracture conductivity
decreases,
the flux density changes so that flow
entering the. portion of the fracture close to the
wellbore becomes steadily
more important.
For
instance, in a low conductivity fracture (kp I#fl) =
0.63), about 70% of the flow comes from the nearest
half of the fracture. However, approximately twothirds of the total flow comes from the farthest half
in a highly conductive fracture (k@ w\D 2 300).
Ibis emphasizes
the importance of creating high
to
overcome
the flow
conductivity
fractures
restrictions
created by the wellbore damage zone.
These findings agree with results presented by van
Poollen.zO
Fig. 7 shows a graph of stabilized dimensionless
pressure &op vs dimensionless
distance along the
values of k/D wffj. Thk pres sure
fracture fOr W/cd
&op is the difference between the pressure at any
point on the fracture and the pressure at the tip of
the fracture. The curves on this figure show that,
for highly conductive
fractures, the pressure drop
the fracture ,is
small
and sometimes
along
negligible. As the fracture conductivity decreases,
the pressure
&op becomes increasingly
greater,
and as the fracture permeability
approaches
the
formation permeability, the pressure drop distribution
(not shown here) corresponds to that for radial flow.
Fig. 7 also presents the results published by Prats18
for steady-state
flow. Excellent
agreement was

found,
The wellbore pressure drop reduction caused by
a fracture is usually handled as a pseudo-skin
factor,
St, which is defined
as the difference
between the dimensionless
pressure
drop for a
fractured well and that for an unfractured well.
Although s is a function of tD, it becomes a
function o { the geometry of the system only for
]ar8e values of tD.
l%e pseudo-skin factor for a fractured well in an
infinite reservoir may be applied to fractured wells
in a finite, circular reservoir whenever the radius
of influence, ~f, of the fracture is smaller tnan the
external radius of the reservoir.
The radius of
influence of a fracture is defined as the radius
beyond which the pressure distribution created by
the fracture is similar, for practical purposes, to
that for radial flow. The radius of influence for an
infinite-conductivity
fracture is about 4 x!. This
means that values of s obtained for an infinite
/
system may be used for a finite reservoir when
re/x ~ 4, These results also are valid for a finitecon d uctivity fracture because rfi for this case is
less than r~i for an infinite-conductivity
fracture.
Fig. 8 presents SI as a ~nction of dimensionless
frac~re conductivity,
~/D k D, for a system where
re/rw = 2,000 and ~e/x, = f O. A particular case is
presented in this figure to compare solutions from
this study with those published in the literature.
Fig. 8 shows, that s is negative, indicating an
increaae in well pro i uctivity. It also shows that
there is a strong variation of s for small values of
w@ klD, and as the value of / racture conductivity
increases,
Sf approaches a stabilized value. Data
published by McGuire and Sikora21 and Pratsls
appear tc agree well with the results of this study.
A general correlation for the fracture skin factor,
St, may be obtained if S1 is expressed as a function

THIS STUDY
PRATS (Stoo4y-Stoto)

- s

z -

=(

o
t
.2

.4

.*

,0

Xos + ,
FIG. 6 STABILIZED FLUX DISTRIBUTION FOR
DIFFERENT FRACTURE CONJXICTIVITIES.
zm

FIG. 7 DIMENSIONLESS PRESSURE DROP


DISTRIBUTION ALONG A FINITE-CONDUCTIVITY
FRACTURE (tD > 5).
JOURNAL
SOCIEIWOF ETROLEUM ENGINEERS

l%is
can be shown by
WjD kiD
n$ x//w
cmnbining the line source solution and the definition
for fracture skin factor. Fig. 9 shows that a graph
of (s/ + k %//rw) V$ (w/L#@) ttiil give a $ingle
curve chat may be used to estimate Sf if values for
Of

straight line for different values of w DkfD. This


time varies between tD equal to 2.5 {or very low
fracture conductivities,
and tD equal to S for high
fracture conductivities.
This is in agreement with
et al,
for infinitethe findings of Gringarten

.dlmcnsion
wellbore
and ~Dk{Dess ace
govidedpressure drop vs the
time for ,various values
logarithm .of dimensionless
of k D wff) is shown in Fi8. 9 and presented in
Tab {e 1. Analysis of these results shows that for
times of itjterest$ tD ~ l@3, solutions
can be
correlated using only one parameter. If practical
values of q D and CID1 are considered,
a unique
solution wi { be obtained for a constant value of
Fig. 9 indicates
that, as the fracture
fD /D
conductivity increases, the dimensionless
wellbore
pressure drop for a fixed time decreases, and for
wfDkFd greater than 300, the solution is essentially
equal to the infinite-conductivity
solution
of
Gringarten et al. All the curves on this figure do
follow, for large times, a straight line of slope
1.151, characteristic
of the semilo8arithmic methods
of pressure analysis.
Also indicated by a dashed
line is the approximate start of the semilogarithmic
The
w

I
-3

TABLE 1 DIMENSIDNLES8 PRESSURE FOR A WELL WITH


A FULLY PENETRATING, FINITE=C43NDUCTIVITY
VERTICAL FRACTURE
Pwm

THIS SIUOY

A MCGUIRC@ndSIKORA

t~ = 0.W0254 M

1412 Qt3/A

+/Actx,z
Pwfr)

kf wl kxf = kfDwfD

.
1x

~~(/+ - Pwf)

tf)
103

2
3
4
6
6
7
8
9
1 x 10-2
2
3
4
6
6
7
8
Q
1 x 10-1
2
3
4
5
6
7
8
Q
1
2

o.211

Iolr

loon

. lr

. 2f7

0.5449

0.2443

0.1733

6m6

Em

E65ii6

0.6330
0,7024
0.7520
0,7926
0,6273
0.6576
0A846
0.2090
0,W13
1.0s37
1.1661
1.2838
1.326e
1,2602
1.4266
1.4676
1.504Q

0,2681

0.2056
0.2280
0.2475
0.2632
0.2770
0,2ss3
o.3m
0.3110
0.3207
0.3630
0.4460
0.4=0
0.5297
0.8030
0,5229
o.62tB
0.6456
0.6s91
0.s453
0.!M65
1.ms
1.1442
1.2121
1.2715
1.3243
1.3719
1.4153
1.7156

O.llfxl
0.1277
0.1424
0.1553
0.1666
0.1773
0,1871
0. 1W2
0,2047
0.2720
0.3221
0,3833
0.3W8
0.4304
0,456s
0.4640
o.5mo
0,5316
0,7015
0.82CB
0.9143
0.SS18
1.0562
1.1163
1.1682
1.21m
1.2577
1.5546

0.0346
0.1120
0.1265
0.1392
O,lom
0.1610
0.1708
0.1796
0.1s81
0.2540
0.3047
0,34%
0.3502
0.4122
0.4408
0.4665
0,4905
0.5129
0.6620
o.8cm8
0;6S40
0.0712
1.0374
1.0355
1.1472
1.1s30
1.236s
1.5320

0.W14
0.0SS6
0.1130
0.1258
0,1360
0.1472
Oclm
0.1658
0.1742
0.2407
o.2scr3
0.3310
0.3662
0,3074
0.4255
0.4514
0.4753
0.4s75
0.5651
0.7545
0.6774
0.9544
1.=05
1.0754
1.1300
1.1768
1,21=
1.5152

2.0414
2.1340
2.211s
2.2164
2.3372
2.3607
2.4371
2.757$

0.31s0
0.3432
0.3633
0.28W
0,3s!3
o.4cm
0.4224
0.4341
0.6181
0.5788
0,6272
0.6682
0,7040
0.7361
0.7653
0,7s21
0,8170
1.W1O
1.1269
12282
1.3CW2
1.37s4
1.4iot
1.41m
1,5424
1.5865
1.6m

O,a?

17

GO
3.0914
3.2W?
3.28SS
3.3684
3.4312
3.48s3
3.5414
3.6883
4.0670
4.2304
4.3417
4.4327
4.5m7
4.5763
4.6351
4.8578
5sm41
6.2387
5.3505
S.4220
5.5532
mm
5.7270
5.768S
5.s353

=3
2.2144
2.3212
2.40s2
2.463!3
2.54s0
2.W5
2.6561
2.s205
3.2~
3.3430
3.4540
3.8458
3.6228
3.68s2
3.74s0
3m05
4.1486
4,34s2
4.4s2s
4.6045
4.8056
4.7736
4.68S4
4.6SS3
4.Q5m

1.5357
1 .772s
1.9263

2olr

kfwl kxf

tfj
-
FIG. 8 PSEUDO=SKIN FACTOR FOR A FINITECONDUCTIVITY VERTICAL FRACTURE.

FIG. 9- PSEUDCkSKIN FACTOR FOR A WELL WITX-1A


FINITE-CONDUCTIVITY
VERTICAL FRACTURE,
Alxdm.

1970

3
4
5
6
7
8
9
1 x 101
2
3
4
6
6
7
8
0
1 x 102
2
3
4
5
6
7
8

Ix!@

2fl

lolr

207r

1oo17
,

1.S016
2.0371
2.1435
2.2311
2.3056
2.3705
2.4270
2.47S4
2$620s
3.0214
3.1643
32753
3@3s81
3.4430
3.50s5
3.8683
3.6208
3#668
4.16%
4.3131
4,4247
4.5156
4.5s28
4.65W
4.7185
4.7711

1.73s4
1,8730
1.07ss
2.0671
2.1414
2.2ml
2.2684
2.3147
2.6553
2.8561
2.swB
3.lfns
3.2007
3.2775
3.3440
3.4a27
3.4553
3.6013
4.0038
4.1475
42ss0
4.3502
4.4272
4.4s39
4.5528
4.6055

1.7175
1.85$~
1.9577
2.0450
2.1193
2.1830
2.2412
2.232q
2.6331
2.6338
2.0768
3.0576
3.17s4
3.2551
3.3217
3.ss04
3.4330
3.776!3
3.s815
4.1252
4.2367
4.3278
4.4049
4.4716
4.83m
4.5631

1.8W8
1.6~0
1.9308
2.0270
2.1013
2.1659
22231
2.2745
2.6150
2.6157
2.9=5
3.0805
3.1602
3.2370
3.3035
3.3623
3.4148
3.76m
3.s633
4.1070
4.2166
4.30s7
4.3657
4,4535
4.5123
4.5650
2s

81

1/

feature of the solutions can be used


This
to analyze field data by a type-curve matching
technique.
From this kind of analysis
we can
determine the formation permeability, k, the halffracturc length, x,, and the fracture cr?nductivity,
k,w.
that estimates
for formation
We assume
porosity,
+,
fluid
viscosity,
p, and total
compressibility,
Ct, are available.
Log-1og type-curve
matching is a technique
commonly used in well test analysis.
As pointed
of this
out by Gringarten et al,, a combination
technique
with conventional
semi log analytical
methods permits a hi8hly confident analysis of
field data.
Solutions presented here do not include wellbore
storage
effects.
However, the /racture
capacity in a highly conductive fracture creates an
effect on the transient wellbore pressure behavior
similar to that caused by wellbore storage capacity.
Recently, Ramey and Gringarten12 presented finitedifference solutions for the transient behavior of a
well crossed by a high-volume, infinite conductivity
vertical
fracture.
They defined a dimensionless
storage coefficient that appears equal to C D~ Fig.
12 presents the dimensionless pressure so [ution for
~,D/ = 0.1, and also shows data presented by, Ramey
and Gringarten for the same case. A good agreement
exists between the two solutions; differences are
less than 2.5%.
for the transient
pressure
Although
results
behavior of a fractured well in a finite drainage
system are not presented in this study, they may
be generated
by means of a desuperposition
technique.19 AIso, wellbore storage effects can be
fD ID

p=-unlfonll nux WLullon


IllFllllTC
WilOSCTIVITV
$OLUTION
CnlnoABTcn,u

r
o

I
108

1
,0-$
,, ,+y

IOs

storage

FIG, 10 km
Vs tD FOR A WELL WITH A FINITECONDUCTIVITY VERTICAL FRACTURE.
fractures. Thent provided that sufficient
portion
of the pressure
data on the straight-line
curve are available before boundary effects influence
a test, the formation flow capacity may be obtained
in the usual way by conventional semilog analytical
methods. Also plotted in Fig. 9 is the uniform flux
solution
for vertical
fractures
presented
by
Gringarten et al. This solution follows the infiniteconductivity solution at small values of times. For
intermediate
times, this solution behaves as a
variable conductivity fracture solution, eventually
following a finite-conductivity
fracture solution of
about equal to 4.4.
jD~fD
Fig. 11 shows the results of Fig. 10 plotted as a
function of the logarithm of pWD vs the logarithm of
ID. At small values of dimensionless
time, the
curves have a distinct form for different values of
conductivity

lo,~!
..

,.,

,,,

,,

II***,

9919.

II

,8

.-,.
m
.

*..,*

. . . .

;.:1:;

.,

..

--T.
.
.,,

...i.

,
,

I,i,:
---!

--+--

:..

;_,
.,

--:-

+;

.-

4 ..
...; :.:,

{ .

tmFR

..:,

1.-L.LL!L

, ,

I+tlll

1..1ill!

,.. I i 11 11 ill
1

[ [1

t I ~ --l
..1..:.1---- L.-.1
F SEMILO~~ ~ ,;
.--....-.,
.-,
,1 ,I
J---.-. . .
.. r-lr
d.
I
1
.
,,
:-I .

..:

.
I
..

I I t!

io-s

* *i&i

FIG. 11
m

f%fD

*7

Vs ~D FOR

,,,

1 , , ,

40

.,,

,,,

* .. IO*

lo~

264kt
t,8 :.00;
+PQ*

FINITE-CONDUCTIVITY

VERTICAL

FRACTURE.

SOCIETY

OF PETROLEUM

ENGINEERS JOlkNAL

incorporated into the solution by using the principle


We expect that in certain field
of superposition.
cases, the pressure behavior for a fractured well
may not follow the solutions presented here, These
deviations may be related to several causes, such
as a high dependence of fracture conductivity on
pressure, wellbore storage, and partial penetration,
to mention some. Sometimes, the nature of the
deviation can be inferred from a careful examination
of field data, For instance, a weH crossed by a
fracture with a conductivity highly dependent on
pressure will exhibit a different behavior in both
bui Idup and drawdown tesrs, in a drawdown test,
the fracture conductivity will decrease continuously,
while in a buildup test, fracture conductivity will
exhibit an increase. In Mb cases, pressure data
will cut across the curves presented in Fig, 11.

TAf3LE 2PRES8URE DRAWDOWNDATA FOR A WELL


CRC%8EDBY A FINITE=CONDLJCTIVITY FRACTURE
(EXAMPLE 1)

c,of~

*:1

% = 260 STf3/f)

Pi - Plvf
Jh&/

JE!9!L
57
68
79

0,25
0.00
1.
2.s
s.
10
20
30
40
00
80
70
80
90
lW
160

108
134
188
210
238
201
200
228
311
321
334
343
384

dimensionless
pressure
drop, pu D, and the
dimension less time, tD, respectively. h e definitions
of pw~ and @ are given by Eq. 4. Once these two
parameters are known, the fracture conductivity,
klw, can be estimated from kfD wtD data match. The
definition of kf~w@ is given by Eq. 13.
EXAMPLE 1

This represents the results of a pressure drawdown


test on an oil well. Pertinent
drawdown and
reservoir properry data are given in Table 2. Fig.
13 shows the application of the type-curve matching
technique for this case. A good and unique match
is obtained for kfD wfD = 2rr and it appsars that the
test was not run long enough to reach the semilog
straight line. Thus, type-curve matching is the
best method to analyze the data of this test.
The formation permeability may be estimated from
the pressure match, Ap = 100 psi, pw~D = 0.47, and
data from Table 2, as follows:
= 0.47

pwfD)M

10

p = 0.85 Cp

DrawdownDsta

EXAMPLES OF APPLICATION
The examples
presented
here are synthetic
pressure drawdown tests for a well crossed by a
finite-conductivity
fracture. The pressure data were
analyzed using a type-curve matching procerdun.,
This technique consists of plotting the presw re
&op, Ap, on the ordinate vs flowing time, t, on the
abscissa
of a log-log paper of the sane size as
Fig. 11. Normally, a tracing paper is placed over
the type. curve, and the major grid lines are traced
for reference. The grid of Fig. 11 is used to plot
actual data on the tracing paper. Next, the data
plot is moved vertically and horizontally over Fig.
11, keeping the grids of the type-curve and those
of the data plot parallel to each other until the
best match is obtained with a curve of Fig. 11.
From this figure, the value of klD !UfD corresponding
to the curve that fits the pressure data fs read. A
convenient match point is pieced and the values of
(Ap)M and (At)M are read from the data plot. The
corresponding values lying directly under this point
on Fig. 11 are (pW@M ~d (~D)M.
h, and the half-fracture
The formation permeability,
length, xi, may be obtained by substitution of the
match
point
data
into
the expressions
for

h=30ft

+ = 0.3
Ct = 20 x 10+ pal-*
B = 1.6s bbl/STB
rw = 0.26 ft

(k md) (30 ft) (100 psi)


STB/D) (0.85 ,Cp) (1.65 ,bbl/STB]

0.1

141. 2(250,

I0

I
10L

MATCH

POIHT

Niloo Dll, t#o.4?


Allloohr$ ,lC$I.C

I0$
I
t :s1J

1
101

,.~

10+

Id

101

10-1

t,

10

FIG. 12 #WtD VS tD FOR A HIGH-VGLi.JME, INE INITECONDUCTIVITY FRACTURE.


AUGISr. 197s

10

,0.1

Io?
I

10

,08

IOJ

,~$

&tlhr81

FIG, 13 APPLICATION OF THE TYPE-CURVE


MATCHING TECHNIQUE (EXAMPLE 1).
261

This yields
k-

7.76m~.

The half-fracture
length then is calculated from
the time match, t= 100hours and tD = 1.6; that is,
(tD)M

= 1.6

=
md)(100

0.000264(7.76
0. 3(0.85

Cp) (20x10-6

P&

hrs)
(xf2

tion of the type-curve and the semilogarithmic


techniques, rcapertively.
Fig. 14 indicates that a
good data match is obtained with the curve for
k ~9@~D = 10 rr. Furthermore, some data points of
t 1 is test fall in the semilog straight line.
The pressure match may be taken as I$p = 100
PSi 2UK! /)w/D
= 0.4.
Similarly, the tirric match may
be chosen as At = 100 hours and tn = 3.3. These
match points,
in conjunction with- the additional
data given in Table 3 yield

ft2)

k _ 5.05

md

Hence,
= 158.4

ft.

Xf = 83,23

estimation of the fracture


conductivity,
&jw,
from the kfD #jD match and the
may be obtained
valuesofk
andxf already computed; thus:
An

(kfw
fDwfD

= 2

= (158.4

= 7.72

x 103

= 13.2

%
10= md-ft

and

red-f t)
ft)(7.76

= -5.06

rnd)

From
Fig.
15, the slope of the semi logarithmic
and P 1 h~~r =
straight line is m = -285 psi/cyck
2,732 psia; then,

and
kfw

wkf

md-f~.

Now, the pseudo-skin factor is calculated from


Fig. 9 by using kfDwjD and %f/ru, values. Since
kpwp
= 2rrandxflrw
=633.60,

162.6~

k=.

qB1.i ~ -

k=

162. 6x195x1.
-285x55

5.1

4x1.8

md

= -5.65

Although the dimensionless


fracture conductivity
has an intermediate
value, the fracture is large
enough to yield a good improvement in well
productivity.
EXAMPLE

*
g

2
MATCH

Table 3 presents the data fora drawdown test in


a fractured well. Figs. 14 and 15 show the applica-

1d

TABLE 3-

PRESSURE DRAWDCWNTEST DATA FORA


FRACTURED WELL (EXAMPLE 2)

4 = 0.18

h=55ft

C, = 18x10-13w1

p = l,8cp

B = 1.4 bbl/STB

II

II
,
;02

10

lt

10

At (h~)ld

103

FIG. 14 APPLICATION OF THE TYPE-CURVE


MATCHING TECHNIQUE (EXAMPLE 2).

.
Drawlown Dsta
.
/+ - Pwf
(houm)
-i!@!J__

I&)

10-

lG

%. = 195sTB/D

t ~ =o,25ft

POIMT

AP~loo?Di,
~,o*o!4?
~t,look~, ..1.. .9.3

I1

10-2

262

ft

81

1
2
3

103
128

4
.5
6
7
8
9
10
12
14
16
20

144
157
17C
182
192
201
207
223
2s
247
267

Pi - Pwf

24
30

_@Q2s3
307

40
50
60
70
80
90
100
120
150
200
250
3W

333
358
378
3m
411
424
439
45s
4s4
522
54s
571

00

2400 -

00%%

Pwf
(

psia)

m=-aC~C#&
\

2000 L

,m~
10-

10

102

103

r ( hm)

FIG. 15 SEMILOG GRAPH FOR EXAMPLE 2.


SWIETY

OF PETROLEUM

ENGINEERS

JOURNAL

= 1.1513

2. For practical values of time, solutions can be


correlated
as a function
of one parameter,
$L) W/D = kfw/kx/. This parameter.is the dimensionin the
less fracture
conductivity.
A decrease
dimensionless
fracture conductivity may be caused
by a decrease in fracture permeability, an increase
in fracture length, or botb. This appears to be an
important reason why type-curve matching with the
original fracture type curves sometimes resulted in
small apparent fracture lengths for large fracture
jobs.
3,. For k~D wfD values equal to or greater than

hour-pi
rn

[oM;r:]
+3.22751
~

-W3

2732-2600
-285

= 1*1513

.18X1.8X18X10-1X

(.25)2

Results
from both procedures are in excellent
agreement; however, additional data concerning the
fracture geometry may be ~ound from the rype-cutve
analysis.
Comparison of Fig. 14 with Fig. 15 is
interesting. Only four data points lie to the right of
the dashed line in Fig. 14, while 10 or 11 points
seem ro lie on the semilog straight line in Fig, 15.
This would include all points to the right of the
arrow in Fig. 14. This results because the analytic
solutions
approach
the semilog
straight
line
asymptotically.
This practical
application
of the
criteria for the start of the semilo~ straight line
indicates the rules may be stretched with acceptable
results.
If pressure data are not available for early times,
a uniqueness
problem arises
when type-curve
matching is applied. This means that data will
match any of the curves in Fig. 11 because they
of time. As a
are similar
for large values
consequence,
fracture geometry parameters cannot
be estimated,
and the only applicable technique
will be the semilogarithmic method.

CONCLUSIONS

all

NOMENCLATURE
6=

f3=
c=

relative
capacity
parameter
defined
Prats,ls dimensionless
formation volume factor, bbl/STB
compressibility,
psi-l
fracture storage capacity, ft3/psi
formation thickness, ft
permeability
pressure
fracture flux density, STB/D - ft
well flow rate, STB/D

%+cjt

x,y =

197s

for

? .~e=u;~~ when plotted


as a function
of
wellbore
pressure
dimensionless
drop vs the
logarithm of dimensionless
time do follow (for
large values of time) straight lines of slope equal
to 1.151. Thus, commonly used semilogarithmic
methods of analysis can be used.
6. The approximate start of the .w:milogarithrnic
straight line is a function of k~D w~D. The dimensionless time for this point ranges from 2.5 to 5 for
practical values of fracture conductivity,
but this
to lower times when
range may be stretched
necessary.
7. Results
when plotted as a fimction of log
~w~D vs log tf)show, at small values of t~, a set
of curves of distinct shapes for the different values
Of kjD wjD . Hence, a type-curve matching procedure
can be used to obtain both the formation and
fracture characteristics.
8. Pressure
data for a well with a low or an
intermediate conductivity fracture (k~D WjD < 300)
does not exhibit a one-half slope, straight line in a
log-log graph.

9W =
t= time, hours
w. fracture. width,

Almlsr,

are

40 The uniform-flux, vertical fracture solution of


Gringarten
et al. behaves
like the infiniteconductivity
solution at small values of time; at
intermediate
times, it follows a variable fracture
conductivity solution, For large values of time, it
follows a finite fracture conductivity
solution of

The main purpose of this study was to provide a


solution that could be applied to analyze transient
pressure data for wells with a finite-conductivity
vertical
fracture.
From the results
of this
investigation,
the following conclusions
can be
reached.
1. Solutions for the transient pressure behavior
for a well with a finite-conductivity
vertical
fracture can be correlated by two dirnensiotdess
parameters, CIDf= w~fc~l /nxf et, and ~/D = kf+ct/

solutions

identical
to the infinitefracture solution of Gringarten

et al.

= -5.04

finite-ccmductivity

purposes
conductivity vertical
5.1

Log

+ 3.2275

300, the
practical

c,

k =

k=
p=
9/ =

space

by

ft

coordinates,

ft
w

T = hydraulic

diffusivity,
=
viscosity,
cp
P
@ = porosity, fraction

Fractured Oil Wells, paper SPE 334S presented et


SPE.AIME Rocky Mountain Regional
Meeting,
Billings,
MT, June 2-4, 1971.

md-psi/cp

the

10. Rsgtmvsn, R., Cady, G. V,, and Ramey,


~~Well Test
Analysis
for Vertically
Wells, t? ,J, Pet,
Tech.
(Aug.
1972)
Trams,, AIME, Vol. 253,

SUBSCRIPTS
D = dimensionless

11.

/ = fracture
i = initial
wellbore

d%

12.,. Ramey, H. J., Jr., nd Gringsrten, A. C.: iEffect of


High Volume
Vertical
Fractures
on Geothermal
Steam Well Behavior,)$ Proc., Second United Nationa
Symposium on the Use nd Development of (3eothermai
Enc?gy, San Francisco, May 20=29, 1975.
13. Locke, C. D. and Sawyer, W. K.: gConstant Preasuro
in II Fractured Reservoir-History
Injection
Test
Match Using Numerical Simulation nd Type Curve
Analyais,C~ paper SPE S594 presented t the SPE=
AIME 50th Annual Fall Technical
Conference
and
Exhibition, Dallas, Sept. 2S=Oct. 1, 1975.

r-

references
P.,
1. Prats, M., Hazebroek,
Jl%ffectof Vertical Fracturea

and Strickler,
W. R.:
on Reservoir Behavior
- CompreaaiMe.Fluid Case, Sot. Pet. 8mg. J.
(June 1962) 87-94; Trams., ASME, Vol. 225.
2. Scott, J. o.: ~c~e Effect of Vertical Fracturea on
Tranai~nt Pressure Behavior of Wells, J. Pet. Tech.
(Dec. 1963) 1365-1369; Trans., AIME, Vol. 22S.
D. G. nd Truitt, N. E.: !Transient Pres3. Ruascll,
sure Behavior in Vertically
Fractured Reservoirs,
J. Pet. Tech. (Oct. 1%4) 11s9.1 170; Truss., AIME,
Vol. 231.
Lee, W. J., Jr.: Analysis
of Hydraulically Fractured
paper SPE
Wells With Pressure Buildup Teats,
1820 presented at the SPE-AIME 42nd Annual Fall
Meeting, Houston, Oct. 1-4, 1967.

5. Millheim,

Analyzing
J.

Pet.

L.: Testing
and
K. Kt and Cichowicz,
Low-Permeability
Fractured Gaa Wells,
(Feb. 1968) 193-19S; Trans., AIME,
Tech

Vol. 243.
Pressure
Testing of Frac6. Clark, K. K.: ~ Transient
tured Water Snjection Wells,
J. Pet. Tech
(June
1%S) 639-643; Trans., AIME, Vol. 243.
R. A. and Remey, H. J., Jr.: well
Interpretation
of Vertically
Fractured
Gas
Wella, t$ ), Pet, ?ecb. (May 1969) 625-632; Trans..
AIME, Vol. 246.

7. Wsttenbarger,

Test

8. van Everdingen,

.% F. and Meyer, L. J.: ~Analyais


of Buildup Curves Obtained After Well Treatment,
/. Pet. Tech
(April 1971) S13-S24; Tram.,
AIME,
vol. 2s1.
Buildup Infor9. Evans, J. G.: ~~~e Use of pressure

mation

au

of
Fracturing Trcatmenta in ow

Permeability Reservoirs, pt!per SPE 453u presented


t the SPE-AIME 4Sth Annual Fail Meeting, Las
Vegas, Sept. 30-Ott. 3, 1973.

SPECIAL FUNCTIONS

4.

jr.:
Fractured
1014-1020;

Schrider, L. A. nd Locke, C. D.: $Effectiveness


Different Hydraulic

t = total
w=

EL J.,

to

Analyze

Non-Respondent

Vertically

14. Gringmten, A. C., Ramey, H. J., Jr., and f?aghavan,


R.: ~6Applied Pressure Analysis for Fractured Wells,$
J. Pet. Tech. (July 197S) 887-S92.
15. Sswyer, W. K., Locke, C. D., nd Overbey, W. K., Jr.:
Vertical Fracture
tcsimulation
of Finite@apacity
in a Gas Reservoir,$ paper SPE 4593 Presented at
the SPE-AIME 4Sth Annual Fall Meeting, Las Vegas,
Sept. 3&Oct. 3, 19730
16. Matthews, C. S. and Russell, D, G.: Ptessurs BMift&
and Flow Teats im Wells, ?donwrrwahseries, society
of Petroleum Engineers,
Dallaa (1967).
H. S. and Jaeger, J. C.: Conduction of
17. Csrslfiw,
2nd cd., Oxford at the Clarendon
Heat in Solids,
Press (1959) 275.
1s. Prats, M,: ##Effect of ve~ica~ Fractures on Rese~oir
Behavior - Incompressible
Case, Sot. Pet. Eng. ~.
(June 1961) 105-1 1S; Tnrtisd, AIME, VO1. 222.
19. Gringarten, A. C., Ramey, H. J., Jr., and Raghavan,
R.: t$Unateady-statePreaswe Distributions Created
Vertical
by a Well With a Single infinite.Conductivity
Fra~ture,~~ Sot. Pef, Eng. J* (Aug. 1974) 347-360~
Tram,, AIME, Vol. 257.
H. K,: c~ProductivWy vs Permeability
20. van Poollen,
Damage
in Hydraulically
Produced
Fracture,
Drill. and Prod. Prac., API (1957) 103.
21. McGuire, W. J. and Sikora, V. J.: The Effect of
Vertical
Fracturea on Well Productivity,$
Tram,,
AIME (1960) Vol. 219, 401-403.
22. Gringarten, A. C. and Ramey, H. J., Jr.: The I.@
of Source and Greens Functions in Solving Unsteudy
Flow Problems in Reservoirs, ~ Sot. Pet. Eng. J.
@ct. 1973) 285.296; Ttams., AIME, Vol. 2S5.
23. Ramey, H. J., Jr.: Personal
U., Stanford, CA (1977).

communication,

Stanford
***

SOCIETY

OF P8TROLEUM

CNGINZ8M8

JOU8NAL

Вам также может понравиться