Está en la página 1de 9

Effect of a Partially Communicating

Fault on Transient Pressure Beh-vior


L.M. Yaxley,* SPE, Shell U.K. E&P Ltd.

Summary. A mathematical model is presented that describes the effect of a partially communicating fault on transient pressure
behavior. The well is treated as an infinite line source and the partially communicating fault as an infinitely long, vertical
semipermeable barrier. Analytic solutions are giv~n for the interference response at an observation well ahd the draw down
behavior of the active well. These solutions can improve th~ design and analysis of interference tests, which are affected by
partially corrurmnicating faults. They are initially developed for constant formation thickness but are later extended to the case of
unequal formation thickness on opposite sides of the fault.
lntroductio"
A question that arises frequently in the developmental planning of
oil and gas fields is to what extent the faults that have been identified by seismic and geologic studies will act as barriers to fluid
flow. The question is important because it may have a major jmpact on the number of wells required to exploit the field's reserves.
Faults in a hydrocarbon-bearing structure may be either sealing
or nonsealing. A sealing fault will completely impede lateral fluid
flow and maY actually form part of the trapping mechanism for the
hydrocarbon accumulation. The throw of a sealing fault is such that
permeable strata on one side of the fault plane are juxtaposed against
impermeable strata on the other side, as illustrated in fig. 1.
A nonsealing fault, which will usually have insufficient throw
to cause complete separation of the penneable strata on opposite
sides of the fault plane, will always allow some degree of lateral
fluid flow. Because of various mechanical processes, such as grain
crushing, bed deformation, and clay smearing, however, the transmissibility of the fault zone may be much lower than the transmissibility of the undisturbed permeable strata. Such a situation, as
illustrated in Fig. 2, is referred to as a partially communicating fault.
Interference testing would seem to be an obvious method for measuring the transmissibility of a partially communicating fault. The
design and analysis of conventional interference and pulse tests are
based on the homogeneous reservoir model, however, which may
be inadequate. At best, it may give an average interwell transmissibility, but it will not quantify separately the transmissibility of
the fault and the transmissibility of the continuous reservoir.
In practice, though, many interference tests across faults have
had the simple, qualitative objective of demonstrating whethe.r communication exists. Such an application was suggested by Johnson
et al. 1 in respect of pulse testing. On the other hand, there have
probably been many atte~pts to measure the transmissibility across
a fault by interference or pulse tests, which have yielded no result.
Qualitatively, a negative test result would imply that the fault was
sealing, but the field pressure performance may later show that it
was nonsealing. In such a case, the interference or pulse test may
have failed because it was designed with an inappropriate flow model
(i.e., the homogeneous reservoir model and exponential integral
solution).
The influence of a partially communicating fault on interference
testing was first considered by Stewart et al., 2 who introduced the
idea of modeling the fault zone as a linear, vertical semipermeable
barrier of negligible capacity. Of course, the model is a great simplification of the complex physical nature of the fault zone, but as
Stewart et al. 2 noted, it h,as the essential property of imposing a
linear flow pattern at the fault plane. It is also the way in which
partially communicating faults are modeled in reservoir simulation
studies. In fact, the set of drawdown type curves Stewart et al. 2
developed was obtained by numerical simulation.
*Now at She!ll International Petroleum Mij.
Copyrig~t

590

1987 Society of Petroleum Engineers

In essence, the partially communicating fault is a type .of linear


discontinuity in reservoir properties. Although the subject of linear
discontinuities has often been addressed in the transient well testing Hterature, 3-7 it has usually been concerned with s~aling faults.
Bixel et al. , 6 however, considered the case of an abrupt, linear
change of reservoir or fluid properties, as shown in Fig; 3a. For
comparison, the vertical semipermeable barrier is shown in Fig.
3b. The two problems look similar, but the mathematical models
needed to describe them are ~uite different. Nevertheless, the solution technique Bixel et al. used can also be used to solve the
problem of the vertical semipermeable barrier.

Problem Description and Definitions


The problem being considered is the drawdown distribution resulting
from constant..:rate production from a well in an infinite reservoir
that contains a linear, vertical semip~rmeable barrier.
The mathematical model relies on the assumptions that (1) the
reservoir fluid is single-phase and slightly compressible, having constant compressibility and viscosity; (2) the reservoir is homogeneous in all rock properties and isotropic with respect to permeability
on each side of the semipermeable barrier; (3) the formation thickness is constant; (4) the well can be approximated by an infinite
line source; (5) the semipermeable barrier is infinitely long and has
negligible capacity; and (6) the fluid leakage rat~ through the semipermeable barrier is always proportional to the instantaneous pressure difference across the barrier.
The last two assumptions allow the partially communicating fault
to be approximated by a vertical plane. This should be valid, provided that the width of the fault zone is small compared with the distance between the fault plane and the well. According to the last
assumption, the leakage through the fault per unit time per unit
length of the fault can be expressed by

Vx=Tt(P2 -pi), (1)


where Tt=kthllpt is defined as the specific transmissibility of the
partially communicating fault, and the parameters kf and lt represent the effective permeability and effective width of the fault zone,
respectively (see Figs. 3b and 4a).
Stewart et al. 2 characterized the partially communicating fault
by the parameter ratio ktf It, which was defined as the fault conductivity. However, the specific transmissibility is preferred here
because it simplifies the terminology required when the model is
extended to the case of unequal formation thickness on opposite
sides of the fault;
On both sides of the fault, the pressure. behavior obeys the diffusivity equation, which is usually expressed in radial coordinates
for problems of fluid flow to a well in a porous medium. The problem considered here, however, is more eas~ly solved by expressiori of the diffusivity equation in Cartesian coordinates with a
separate formulation for each side of the semipermeable barrier.
SPE Formation Evaluation, December 1987

SEALING FAULt

------

P2 Pt
.........

.._...Vx

--

(a)

Pt
Fig. 1-Sealing fault prevents lateral fluid flow.

h1

P2 - ......... Vx

hz

--------

(b)
PARTIALLY COMMUNICATING FAULT

Fig. 4-Partially communicating fiu.ilt modeled as a vertical


plane: (a) uniform thickness; (b) unequal thickness .

.. ~ y

Observation Well
Region
(x, y)

""'

L ______.

(b, 0)
Observation
Well

...

'

Fig. 2-Partially communicating fault inhibits lateral fluid flow.

Active Well
Region

'Y/1

Fault

(a)

Fig. 5-Coordinate system for .the partially communicating


fault model.

k1
f.l1
'Y/1

(b)

k1

is located at the point (b,O) in the active-well region and is treated


as a line source of strength q/~cth,
The diffusivity equationfor x>O is

fl-1
'Y/1

a2 ilpi

a2 ilpi

qp,

1 ailpi

&2

~2

-.-+--+-o(x-b)o(y)=---., ........... (2)

If

Fig. 3-Types of linear discontinuities: (a) abrupt change of


rock or fluid properties; (b) vertical semipermeable barrier.

in which o(x-b) is the Dirac delta function defined by the relations

o(x-b)=O, for x=t=b,


As shown in Fig. 5, the semipermeable barrier lies along the y
axis. In the half-plane x>O, which is defined as the active-well
region, the pressure-drawdown distribution is ilp 1(x,y,t). In the
half-plane x < 0, which is defined as the observation-well region,
the pressure-drawdown distribution is ilp 2 (x,y,t). A producing well
SPE Formation Evaluation, December 1987

and

591

10 ~--------~----------~--------~--------~----------,

10- 2 ~~~~~~~~~~~~--~~~~~~~~~~~~~~~~

w-!

10

Dimensionless Time tvL


Fig. 6-Logarithmic plot of pressure drawdown for an observation well near a partially communicating fiiUit.

For x < 0, the diffusivity equation is

a2 ilp 2 a2 llp 2 1 oilp2


,
- - + - - = - - - ......................... (3)
2
2
ox
oy
'ri ot
Eq. 2 is the same special form used by Bixel et al. , 6 which incorporates the constant-flow-rate boundary condition at the wellbore by means of the source density term (qlh)o(x-b)o(y). The
initial condition and remaining boundary conditions are
ilpn(x,y,O)=O, n=1,2, ............................. (4)
ilpn----.o as lxl-oo, ................................ (5)
ilpn----.o as

IYI-oo, ................................ (6)

-~= I x~o ~

0
:

x~o'

....... ... (?)

and
- kh oilpt
p,
OX

=
x=O

kjt (ilP2 -ilpt) ................... (8)


lfP,

Eqs. 5 and 6 ensure transient-flow conditions by stating that the


drawdown is negligible at the outer infinite boundaries of both
regions.
Eq. 7 ensures conservation of mass for flow through the semipermeable barrier, while Eq. 8 is a restatement of Eq. 1, which
introduces the flow resistance effect of the semipermeable barrier.
For convenien~, the following dimensionless variables are in~ _
troduced.

kh
Pvn=27r-dpn, n=1,2, ........................... (9)

xv =x/L, ........................................ (11)


YD =y/L, ......................... ; .............. (12)
bv =b/L, ................. -....................... (13)

and
"L

~c:

)/(:). . . . . . . . . . . . . . . .

(14)

where L is the displacement of the observatiort well from the active well perpendicular to the f~ult (see Fig. 5).
The dimensionless parameter a L is defined as the specific transmissibility ratio with respect to the interwell distance, L For the
. case of constant reservoir thickness on both sides of the fault, a L
is equivalent to the dimensionless fault conductivity defined by
Stewart et al. 2

These definitions allow Eqs. 2 through 8 to be expressed in dimensi9nless form. The resultant equations are giveri in Appendix A,
which also outlines the solution procedure.

Observation-Well-Region Solution
The drawdown distribution in the observatiort-well region as a solution to Eqs. 2 through 8 is

rL [exp(

4aiu-

y:Jerfc [2aLviU + (IX~j;D)] ~].

................................... (15)
In the special case when the observation Well is directly opposite
the active well (i.e., Yv=O and lxvl+bv=1), Eq. 15 reduces to

qp,

where ilpn =pi -pn and Pi is the initial pressure. Also,

2
.
(
[fDL
1 ) du
Pv=-liraLJ
exp(4aLu+2aL)erfc 2acJu+-- - .

tvL ='f/t/L 2 , .............. ; ...................... (10)


592

2~ ~

........................... ' ....... (16)

SPE Formation EvalUation, December 1987

~
~

3.0 +-----~~-----+------~~~r-r---~~

c.
~

12.0 +-------t------:E

i:S

10
Dimensionless Time loL

Dimensionless Time loA

Fig. 7-Semilogarithmic plot of pressure drawdown for an_observation well near a partially communicating fault.

Fig. 8-Pressure drawdown for an active well near a partially -;;ommunicating fault.

An interesting feature of this solution is that the draw down response


at the observation well does not depend on the position of the fault,
but only on its distance from the active well. The integral in Eq.
16 can be evaluated numerically, provided that care is taken to compute the integrand with sufficient accuracy. Some solutions for various values of the specific transmissibility ratio are plotted in Figs.
6 and 7. They show that the partially communicating fault attenuates the drawdown response relative to the exponential integral solution (i.e., the no-fault case). The semilog plot in Fig. 7 shows
that at late time all the response curves have the same straight-line
slope and are parallel to. the exponential integral solution.
Hence as time increases, the difference between the exponential
integral solution and Eq. 16 must tend toward a constant value that
is a function of the specific transmissibility ratio. If is shown in
Appendix B that the late-time behavior of ~q. 16 is given by

A more convenient form of the solution is obtained by setting L=b,


because the location of an observation well on the opposite side
of the fault is not relevant to the active well region. Hence the solution becomes

[(xv -1)2 +yzS]]


4tvA

YE) erfc [ 2a.AJU+-----(xv+1)] -du] ,


JrtvA[ exp ( 4a1u:._4u

2JU

JU

Pv=-VzEi(---)+e 2aLEi(-2aL) . ......... ; .... (17)


4tvL
.
In principle, the type curves shown in Fig. 6 could be used for
matching interference test data, provided that the duration of the
test corresponds to at least two to three log cycles on the dimensionless time axis. However, if the direction of the fault is not perpendicular to the line of intersection between the active well and
the observation well, then Eq. 15 should be used to generate a different set of type curves. Fig. 6 is still useful for rough design calculations because the specific transmissibility ratio will not be known.
Even if the observation well is not directly opposite the active well,
a sufficiently accurate test design can be made from Fig. 6 if L
is taken as the distance between the wells rather than the displacement indicated in Fig. 5.

ActiveWeiiRegion Solution
The drawdown distribution in the active well region as a solution
to Eqs. 2 through 8 is

. .................................. (19)

where

tvA ='rftlb 2

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

(20)

and

aA~c;: )!(:).

............................

(21)

The new dimensionless parameter, a A, is defined as the specific


transmissibility ratio with respect to the fault distance b.
Drawdown at the active well is obtained by setting x = b- r w and
y=O. Ifthe wellbore radius (rw) is very small compared with the
fault distance, b, then xv + 1 =:::: 2 and the wellbore pressure
response is given by

[(xv -bv) +yE]]


Pv(xv.Yv)=-VzEi [ - - - - - - 4tvL

( - -r~w)
1 ) -.,J;ra.A
.
Pvw=-lhEi
. - -VzEi ( - 4tvA
tvA

xfVA exp(4a]u+4a.A)erfc(2a.AJU+ - 1-) du,


0

tvL

J
0

Yv

2) erfc [2a.LJU+
[exp (4a.Lu-4u
2

(xv +bv)
2JU

J-JUdu ] .

................................... (18)
SPE Formation Evaluation, December 1987

...... (22)

JUJU

where rvw=r wlb.


Eq. 22 contains, as its first two terms, the well-known method-ofimages solution for a sealing fault 4 7 8 and degenerates to this form
completely when aA =0. Fig. 8 compares the active-well draw593

10

Co=L
s

= 1.5 (skin factor)

-~0.9

aA

..._---+----l----+-,l.'+--.....::....,----+---+----1

= 0.1

blrw = 200

"-

Q.
::1

~
~

0.8

-~

-.:
~
~ 0.7

...~
"

Homogeneous Reservoir

i5

10_,

10
Dimensionless Time

102

IO'+-~~~~~~~~~~~~~~~~~~~~
10 '
10'
10'
10'
10
10

10'

loA

Dimensionless Time t 0 /C 0

Fig. 9-Pressure-derivative drawdown for an active well in


the absence of wellbore storage.

Fig. 10-Pressure derivative drawdown example for an active well with wellbore storage and skin.

down behavior for different values of the specific transmissibility


ratio. It shows that for all finite, nonzero values of specific transmissibility ratio, the drawdown response curves are bounded by
(1) the no-fault case for which the response is a single semilog
straight line of slope m and (2) the sealing-fault case for which the
response consists of two semilog straight lines joined by a smooth
transition. The first has a slope of m and the second a slope of 2m.
Initially, the drawdown curves are indistinguishable from the nofault case, but eventually the influence of the semipermeable barrier appears as a gradually increasing slope. They never reach the
second straight line of the sealing-fault response, however, because
of the pressure support provided by leakage .through the semipermeable barrier. Therefore, the slope of each curve increases to some
maximum that is always less than 2m. Thereafter, the slope
decreases until each curve becomes parallel to the no-fault response.
By this time, the leakage through the semipermeable barrier has
stabilized, whereas it is time-variant during the transition period.
After a sufficiently long flow period, the partially communicating fault drawdown response, and the'no-fault drawdown response

differ by a constant pressure difference, op, which is a function


of the specific transmissibility ratio. The late-time behavior derived
in Appendix B shows that

op=-e 401 AEi(-4aA)

...-.......................... (23)

In theory, therefore, the drawdown response of an active well


near a partially communicating fautt could also be used to measure
the specific transmissibility of the fault. This is an important result
because it would not be possible, for instance, to investigate a partially communicating fault by interference testing if an appropria~e
observation well had not been drilled. In practice, though, the first
semilog straight line may be obscured by wellbore storage and the
second straight line may not fully develop if the flow period is too
short.
Interpretation of the active well response might be improved by
utilizing the type of pressure derivative approach advocated by Clark
and Van Golf-Racht 9 for various other well and reservoir models.
F~g. 9 illustrates the pressure derivative response in the absence

lXA = 0.25
fJ:?Ap= 1000

2.04-------+~~,-,-~~-~-----4-------+-----~

'
/',,,
slope= m

'

Schematic Horner Plot


''
'

'

' ',

'

Naturally Fractured
Reservoir

10

102
{tp

103

+ D..t)l D..t

Fig. 11-Pressure buildup example f.or an active well near a partially communicating fault.
594

SPE Formation Evaluation, December 1987

TABLE 1-VALUES OF THE FUNCTION


f(x) = - e 2 x Ei(- 2x)
X

0.001
0.002
0.005
0.010
0.020
0.050
0.100
0.200
0.500
1.000
2.000

__!N_
5.6507
4.9681
4.0785
3.4225
2.7907
2.0146
1.4934
1.0478
0.5963
0.3613
0.2063

Also, Eqs. 7 and 8 become

h1 a!lp1
ax

Active-Well Pressure Buildup


The pressure-buildup response at the active well after a single-rate
drawdown test is given by the superposition equation

x=O

=h2 a!lp2
ax

- kh2 a!lp1
JL
ax

kh
op=

(Pi-p*), ............................ (25)

141.2qJ.L
where p* is the extrapolated pressure of the first semilog straight
line. The specific transmissibility ratio of the fault can then be obtained from Eq. 23 with the help of Table 1. In practice, the usual
problems of wellbore storage and insufficient drawdown or buildup time would make such an ideal response difficult to measure.
Nevertheless, the technique might still be practical if the test could
be conducted with a bottomhole shut-off tool. A pressure-derivative
approach would probably assist in the interpretation of actual data.

Effect of Unequal Formation Thickness


A useful generalization of the partially communicating fault model
is the case of unequal formation thickness on opposite sides of the
fault, as illustrated in Fig. 2. If the diffusivity remains constant,
then the solution already obtained for the observation well still applies but with modified definitions for the specific transmissibility
ratio and for the dimensionless pressure. If h1 and h2 are the formation thicknesses in the active-well and observation-well region,
respectively (see Fig. 4b), then the specific transmissibility used
in Eq. 1 must be redefined as
Tf=kfhlltll ..................................... (26)

where

ltJL

Hence the new definition of the specific transmissibility ratio is

~(~:)/(::

) .............................. (29)

After some algebraic manipulation of the transformed equations


given in Appendix A, it can be shown that Eq. 16 is still valid,
provided that the dimensionless pressure is redefined as
PD =27rkh!lp21qJL. . ............................... (30)

In a similar manner, one can obtain the following solution for the
active well:

Pvw=

-1f2Ei(-

Ei.(--1-)-c

rvw ) - 112C 1
4tvA

2I(ct.A,tvA);

4tvA

. ..................... ..... ; ....... (31)

where l(aA,tDA) is the integral term in Eq. 22 and


Pvw=27rkh 1flp 11qJL, .............................. (32)

aA

~ ~:
(

) / ( : ) . .... , . . . . . . . . . . . . . . . . . . . . . . . . . (33)

c 1 =2(h 1th 2 )-1,

................................. (34)

and

c2 =2h[ lh 2(h 1 +h2).

. ............................ (35)

Conclusions
Analytic solutions have been obtained for the transient pressure behavior of a constant-rate well in an infinite reservoir that contains
a linear, vertical semipermeable barrier. The solutions may be useq
to improve the design and analysis of interference tests between
wells separated by partially communicating faults. The type curves
generated by these solutions will yield separate estimates of the formation transmissibility arid the transmissibility of the fault itself.
An explicit solution for drawdown at the active well offers the
possibility of deriving fault transmissibility from the drawdown and
buildup behavior of the active well alone. One could use this method
if a convenient observation well is not available for interference
testing.
It was found that the draw down and buildup behaviors of the active well resemble inverted forms of the characteristic behavior of
a well in a naturally fractured reservoir.
Interference type curves were generated for the special case of
a fault that is perpendicular to a line of intersection joining the active and observation wells. The general analytic solution for the
pressure-drawdown distribution, however, can be used to generate type curves for other fault orientations.

Nomenclature
b = x coordinate of active-well location, ft [m]
ct = total compressibility, psi - 1 [Pa - 1]
CD

SPE Formation Evaluation, December 1987

x=O

= kfh (flP2 -flp1) ................. (28)


x=O

Pvs=PvwCtvAp+fltvA)-Pvw(lltvA), (24)

where tDAp is the producing time before shut-in.


Eq. 24 was used to generate the theoretical Horner buildup plot
in Fig. 11, which shows two parallel semilog straight lines separated by an amount op on the dimensionless-pressure axis. This behavior resembles the characteristic buildup response of a naturally
fractured reservoir, 7 8 but is inverted (see insert in Fig. 11). In
theory, the slope of the first straight line would yield the reservoir
transmissibility the dimensionless-pressure increment, op, would
yield the speciflc transmissibility ratio of the fault, and the second
straight line would extrapolate to initial pressure. The dimensionless pressure difference, op, can be calculated for actual field data
from

I ..... : ................. (27)

and

aL

of wellbore storage for different values of specific transmissibility


ratio. Wellbore storage can always be incorporated into the par. tially communic'ating fault model by the desuperposition
method. 10 11 An example is shown in Fig. 10. Note that the second "hump" resembles an inverted form of the characteristic
pressure-derivative response of a naturally fractured reservoir, as
described by Bourdet et al. 12
.

= dimensionless

storage constant
595

F = Fourier transformation operator


h = formation thickness, ft [m]

h = average formation thickness (see Eq. 26), ft [m]


I= function defined by Eq. A-21
k-=
k1 =
11 =
L =
=
m =
p

op
D.p
q

rw
s
t

D.t
Tf

u
Vx
w
x

z
a
{3

formation permeability' md
effective permeability of fault zone, md
effective hoi-izontal thickness of fault zone, ft [m]
interwell distance perpendicular to fault, ft [m]
Laplace transformation operator
semilog slope of pressure change for infinite-acting,
homogeneous reservoir
= pressure, psi [Pa]
= incremental pressure difference (see Eq. 25 and
Fig. 11)
:;= pressure drawdown, psi [Pa]
= production rate, RIVD [res m 3 /d]
= wellbore radius, ft [m]
= Laplace transform parameter
= time, hours
= shut-in time, hours
= specific transmissibility= k1hllp.t,
m 2 /Pas=0.00l}27k1hlltp,, md/cp or RB/D-psi-ft
with kt in md and p, in cp (see .Eq. 1)
= variable of integration
= .volumetric leakage rate per unit length of fault,
RB/D-ft [res m 3 /dm]
= transformed pressure drawdown (see Eq. A-10)
= distance perpendicular to fault, ft [m]
= distance from active well p~rallel to fault, ft [m]
= component of transformed pressure drawdown (see
Eqs. A-18 and A-19)
= specific transmissibility ratio (see Eqs. 14 and 21)

3.
4.
5.
6.
7.
8.
9.
10.

11.
12.
13.
14.

Appendix A-Solution of Eqs. 2 through 8


Basic Equations. In dimensionless form, the differential equations,
initial conditions, and boundary, conditions for the partially communicating fault model are

= ..Jw 2 +s

11 = hyd~aulic diffusivity, khpctp,, m 2 /s=

0.0002637k/c,p,, ft 2 /hr with k in md, c1 in psi -l,


and p, in cp
p, = viscosity, cp [Pa s]
= porosity, fraction
w = Fourier transform parameter

Subscripts
A = active well
D = .dimensionless
f =fault
i = initial
L = with respect to interwell distance
p = producing
S = shut-in'
W = wellbore
1 = active-well region
2 = observation-well region

Acknowledgments
I thank Shell U.K. E&P and Shell Inti. Petroleum Mij. for permission to publish this paper. I also gratefully acknowledge Sandra
M. Robertson's assistance with various computer applications.

References
1. Johnson, C.R., Greenkorn, R.A., and Woods, E.G.: "Pulse Testing:
A New Method for Describing Reservoir Flow Properties Between
Wells," JPT (Dec. 1966) 1599-1604; Trans:, AIME, 237.
2. Stewart, G., Gupta, A., and Westaway, P.: "The Interpretation oflnterferen~e Tests in a Reservoir With Sealing and Partially Communicat-

596

ing Faults," paper SPE 12967 presented at the 1984 SPE European
Petroleum Conference, London, Oct. 22-24 ..
Matthews, C.S. and Russell, D.G.: Pressure Buildup and Flow Tests
in Wells, Monograph Series, SPE, Richardson, TX (1967) 10, 1.
Horner, D.R.: "Pressure Build-Up in Wells," Proc., Third World Pet.
Cong., The Hague (1951) 503-,23.
Gray, K.E.: "Approximating Well-to-Fault Distance From Pressure
Build-Up Tests," JPT (July 1965) 761-67.
Bixel, H.C., Larkin, B.K., and van Poollen, H.K.: "Effect of Linear
Discontinuities on Pressure Build-Up and Draw down Behavior,'' JPT
(Aug. 1963) 885-95; Trans., AIME, 228.
Streltsova, T.D. and McKinley, R.M.: "Effects of Flow Time Duration on Buildup Pattern for Reservoirs With Heterogeneous Properties,"
SPEJ (June 1984) 294-306.
Earlougl1er, R.C. Jr.: Advances in Well Test Analysis, Monograph Series, SPE, Richardson, TX (1977) 10, 5.
Clark, G. and Van Golf-Racht, T.D.: "Pressure Derivative Approach
to Transient Test Analysis: A High-Permeability North Sea Example,"
JP-T (Nov. 1985) 2023-39.

Gringarten, A.C., Ramey, H.J. Jr., and Raghavan, R.: "UnsteadyState Pressure Distributions Created by a Well With a Single InfiniteConductivity Vertical Fracture," SPEJ (Aug. 1974) 347-60; Trans.,
AIME, 257.
Chen, H.K. and Brigham, W.E.: "Pressure Buildup for a Well with
Storage and Skin in a Closed Square,'' paper SPE 4890 presented at
the 1974 SPE California Regional Meeting, San Francisco, April4-5.
Bourdet, D. et al.: "Interpreting Well Tests in Fractured Reservoirs,"
World Oil (Oct. 1983) 77-87.
Gradshteyn, I.S. and Ryzhik,-I.M.: Table of Integrals, Series, and Products, corrected and enlarged edition, Academic Press Inc., New York
City (1980) 307-10, 931.
Abramowitz, M. and Stegun, I.A.: Handbook of Mathematical Functions, Natl. Bureau of Standards, 1968.

azpz azpz apz


- - + - - = - , x<O, ....................... (A-2)
ax 2
ay 2
at
apl
ax

apz
=--;;;'
x=O,

............................... (A-3)

apz

- = a ( pi -pz), x=O, .......................... (A-4)

ax
Pn=O at t=O, n=l,2 ............................ (A-5)
Pn~o as lxl~oo,

............................... (A-6)

and
Pn~o as IYI~oo

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

(Note that the subscripts of the dimensionless variables-i.e., D


and L-have been omitted throughout the appendices for convenience of notation.)

Solution Procedure. The problem, as stated by Eqs. A-1 through


A-7, was solved by successive application of the Laplace transform
with respect to time, and the complex Fourier transform with respect
toy; i.e.,
[pn(x,y,t)l = rx> e -stPn(x,y,t)dt=jin(x,y,s) .......... (A-8)
0

and

F [ji n(x,y,s)] =

rX> e -iwyjin(x,y,s)dy=p n(x,w,s). . ......

(A-9)

-00

SPE For~ation Evaluation, December 1987

To simpiify the notation, let


Wfi

=p n(X,W;S) .

Appendix &"-Limiting Cases


of the Function I (a, t)

........ ......... ; ........... . (A-10)

The differential equations, Eqs. A-1 through A-4; become

d 2 w1
27t
- --(w 2 +s)w 1 =--o(x-b), x>O, ........... (A-11)
2
dx
s
d2w2

- - - ( w +s)w2=0, x<O, .................... . (A-12)

dx2

Limiting Cases. There are three limiting cases of the general so-

lut~oh W exarnin~ that depend ori the behavior of the function /(a, t).

For the sake of brevity, the discussion is restricted to the


obserVation-well tesponse for which b+x= 1 and y=O.
Case 1: a--+0 (i.e., a sealing fault). Eq. A-21 clearly shows that
l(a,t)--+0 as a--+0, which means that the drawdown. in the
observation-well region becomes negligible.
Case 2: a...:..oo (i.e., homogeneous reServoir). With the asymptotic representation of the probability integral function as given by
Gradshteyn arid Ryzhik, 13 it can be shown that as a~ oo for finite

x=O, .............................. (A-13)


dx

u, exp(4a 2 u+2a)erfc(2a.JU+ - -_ )- e- '>4u (

dx

.fir

2../U

-./U ).

4au+ 1

........ ' ......................... (B-1)

arid
dw 2
=a(Wt -w2), x=O . ........................ (A-14)
dx
Eqs. A-ll through A-14 areordinary differential equations for
which the solutions are readily obtained:
7r

w 1 =-e-13lb-xl+

s(3

1re -{3(b+x)
. , x>O .............. (A-15)
s((3+2a)

and
.

7r

w2=-e
s(3

_r.llb-xl
fJ

Hence it can be deduced that


-1;.\u
(
1)
l(a 1t)-+ 1h r-e--dit=- 1f2Ei - - ............... (B-2)
0 u
4t

Case 3: Late-time behavior (i.e., t-+ oo). To study this case fully
we hav~ to find an aiternative form for the ihtegnil term in Eq.
A-21. Gradshteyrt and Ryzhik 13 give the following integral relationship:

1re-{3(b+lxi)

----'----, x<O, ............ (A-16)


s((3+2a)

where
.................................. (B-3)

(3=-Jw 2 +s . ................................... (A-17)

Now let w 1 =z1 +z2. This means that W:2 =z 1 -z 2 and the inversions of w 1 ahd w2 are obtained simultaneously after die corrimon
components z 1 and z2 are inverted independently:

Therefore l(a,t) (Eq. A-21) cim be rewritten as


.

te-l 14u

7r

z 1 =-e-13lb-xl

(
v2
)
exp ---2av dvdu.
b+ lxl
4u
-

/(a,t)=a~~xp[2a(b+lxbJ)---I

............................... (A-18)

s(3

oo

.................................. (B-4)

and
t2 =

1re - {3(b + lxb


.
. .............................. {A-19)
s((3+2a)

Eqs. A-18 and A-19 were inverted by taking the inverse Laplace
transformation and applying the integral inversion formula for the
compiex Fourier transfotmation. The results are
.
[ [(b-x)2 +y2]]
F- 1- 1(z1)=-lhEi
.......... (A-20)
.
4t
and

After chatiging the order of integration and integtating by parts,


Eq. B-4 simplifies to
'
l(a,t)=- 1hEi (

. [(b+ lxl)2 +yi]]


.

4t
00

ue -2aue -(i +u 2 )14tdu

-exp[2a(b+lxi)]I - - - - - - b+lxl
y2+u2

...... (B-5)

For the special case of b+ ixl = 1 and y=O, Eq. B-5 reduces to

p- 1 - 1 (z2 )=- 'hEi [

[(b+ lxl)2 +y2]

-l(a,t),

4t

......... : ....................... (A-21)


where

tf [
y2
l(a,t)=..hra exp[2a(b+lxi)]J lexp 4a 2u- u
4

l(a,t)= -\>E;(-

X erfc

[ 2a.JU +

.JU

JU .

SPE Formation Evaluation, December 1987

-j\a,t), ..................... (B-6)

where

oo

(b+lxi)J du]

:J

f(aJt)=e 2a

e-2aue-u2/4tdu

.....................

(B-7)

597

But as

t--+oo,

f(cx)=e 2a

j
1

Eq. B-7 becomes


00

e -2au
--du= -e 2aEi( -2cx) . ............. (B-8)
u

Therefore, for large enough t, the observation-well response is

l(a,t)

~-

hEi(-

:J

-f(a), ...................... (B-9)

sionless time and specific transmissibility ratio given by Eqs. 20


and 21 , we obtain ,

Pvw= -'hEi(- rbw) +j(2cxA),_ .................. (B-10)


4tvA
where f(2cx A)= - e 4aA Ei( -4cx A).

Sl Metric Conversion Factor


psi X 6.894 757

E+OO

kPa

SPEFE
where j(cx) is independent oft.
Some useful values of the functionf(cx) are listed in Table 1. The
late-time behavior for the active well can be deduced by combining Eqs. 18 and B-5. After incorporating the definitions of dimen-

598

Original SPE manuscript received for review Sept. 22, 1985. Paper accepted for publicatioh Jan. 16, 1986. Revised manuscript received Jan. 2, 1987. Paper (SPE 14311) first
presented at the 1985 SPE Annual Technical Conference and Exhibition held in Las Vegas, Sept. 22-25.

SPE Formation Evaluation; December 1987

También podría gustarte