Sie sind auf Seite 1von 7

. .

q-m?%
.)
-(
.
[ -
.-
,+

..

) RESERVOIR
ENGINEERING

.
Approximating Well=to-Fault Distance from
Pressure Build-Up Tests
K. E. WJAY
MEMBER AIME
I THE Lt. OF TEXAS
AUSTIN, TEX.

Abstract During shut-in time At,the pressure behavior at the


sand face of a single oil well which has been produced at
The method used by Horner to calculate distance from a constant rate q, for a time f, in a horizontal formation
a well to a linear barrier fault in an otherwise infinite re- of constant thickness h, uniform permeability k, and por-
servoir is an approximation, the accuy-acy of which- de- osity +, at a distance d, from a linear barrier fault in an
pends upon transient duration relative to fault distance otherwise infinite reservoir, is given in dimensionless form
and reservoir rock and fluid properties. Using calculated (in cgs units) by
build-up curves jor an undainaged well, Horneis method
and several alternative methods not requiring a computer
are illustrated and discussed. p.=
HEi(-H-Ei(-&-)
Introduction
The unstepdy-state pressure behavior of a given ,well
+&&i(+)] ) - ~ )
indicates the net effect of all transients operating to pro- w
e
duce Pressure changes at that well. OnlY oqe or two tran-
sients would be i_nv-olved,for example, during a constant-
rate drawdowo or build-up test, respectively, of a single
. J
where Ei(.v) = ~ du and the dimensionless quan-

[ities pressure, rate &rd time are defined in cgs units by


well in a new reservoir of large extent. On the other hand,
amultitude of individual transients might be contributing Pn = p,. p,.: kAt
,..
to the pressure behavior at the well: For examPle> rate, p,.. = - jbpct,,
changes at the well in question, interferetice from an offset.
well or wells, a permeability or phase change withh the Pi
reservoir, faults, piqchouts, etc., influence a wells pressure- l) = %hkp. .
time hktory. It is not always possible to isolate each pres- ._ k(t+At) kAt
sure transient from the composite pressure-thne history at t* = 7- .
- m- 4+pcd
a test well, since both pressure-increasing ?nd -decreasing
transients ,might be involved, and since some interfer- where p,. is the wellbore pressure and P,,. is tie stabilind
ence transients might arrive at the well simultaneously, formation shut-in pressure. The remaining quantities are
Depending upon the. situation, however, prior pressure defined in the nomenclature. Using the constant terminal
changes may have stabilized at the well before the arrival rate solution to the radial diffusivity equation rind the
of another transient, in which ca..e we can usefully analyze Lord Kelvin point source, given by Hurst and van Ever-
that, transient. djrtgen, as well as the principle of superposition, Eq. 1
The presence of a fault in a reservoir is obviously of can be derived by the rnet!md of images. The effect of the
great importance, and this paper is concerned with the fault is duplicated by an image well identical to the real
pressure behavior of a well near a sealing fault, artd how, well and located at a mirror image of the real well from
under idealized conditions, distance to the fault might be the fault plane. The physical significance of the irtiage well
calculated from measured well-pressure data. No implica- N that there is no fluid flow across the fault, ,ie., it is a
tion that othk! interference phenomena are unimportant ii sealing one. While such an ~ssumption is necessa~ to the
intended. An ideal well is used to provide the best situa- mathematical development df Eq. 1, it afso limits its strict
tion we ,could hope for. If problems in analyses arise here, application, because obviously not all faults tire sealing.
. then we might suspect that unambiguous analyses of actual The real and image wells flow at constant rate for a
well -data--might be at- tiftx%impossible; this implication is dime t arid are then shut hi fuf tinte At.There are; theg, -j . . --+
intentional. Ifoui individual constant-rate transients, the superposed
effects of which are to be measured at the real well. The
Original manuscript rseelvsd in Society of Petroleum Engineers oilke first two Ei functions in Eq , 1, representing the &awdown -
Od,, 22. 1964. Revi*d mamsecrirst received June S, 1065. Paper (sPE -,.
91K0. presented at SPE Annual I+Ili M-ting hel~l In H~@On. ~et.
11-14, 1964, . References given nt end Of lmwr. .- =., .
.
76 I
-. ,,- -
flJLY, 1965
,. J
I .-.
.
,. ,
... . . .. . ,, .

,7
and build-up transients of the real well, simplify to the shut-in, except in this case the well flowed only 12 iiours
logarithmic approximation in [At/( t+At)] after dimen- prior to shut-in. Note that the first straight-line per.
sionless time tn reaches 25. For normal values of wellbore tiori has too low a slope (k derived therefrom would be
radius, t,, becomes greater than 25 after teal times on the too high) and that the second straight-line portion, wbiie
order of seconds, about double the first slope, is also too low (extrapolation
As Horner has previously shown, Eq. 1 can be written to infinit,c shut-in time would give a low value of p,,,).
in field units as For Fig, 2 (t= 12 hours) the siope of the first appavett~
straight-line portion is 74.5 psi/cycle when the image
9d@ ,n - AI : problem is used, whereas for no fault the slope is 81,3
P. = Pw, +70,62 _ kh
[ t+At psi/cycle, The reason for this difference is that the draw-
down transient from the image well is measurably dropphzg
~i _ 3793 .6z/@ _ ~i _ 3793 .6@d
the pressure at the reai we!! during shut-in hours; that is,
( k(t+At) ),( k~t. )1
the drawdown transient of the image ,weil at the real well
., .,, ., .. . . . . . . (2) has not stabilized because flowing time t is rather short.
While the slope at any value of Acj(t+Az) is slowly
where the Ei terms represent, respective] y, the drawdown changing, the pressure intercept of a tangent at that point
and mtild-up transients of the image well as measured at is decreasing with ti~e.
the real well, The field units in Eq. 2 are: p. and pm.= . In Fig. 1 (t=1,2oO hours) on the other hand, the first
psia; q., =STB/D; P=CP; /3=reservoir bbl/STB; k= md; apparent straight-line portion hag a slope of S 1 psi/cycle
h = ft; t and At= hours; # = porosity, fraction; c = psi; for the image problem compared to t31.3 psi/cycie for no
d= ft. - fault, Much closer agreement resuits here because iong
Using appropriate dimensionless quantities in Eq. 1 for flowing time allows the drawdown transient of the image
constant ,fiow rate, an equation for pressure behavior of a weli at the real well to stabilize prior to shut-in,
gas well near a sealing fault analogous to Eq. 2 can be We emphasize the apparent straight-iine portions in Figs.
written ks 1 and 2 because pm plotted bs in [At/ (t+At)] from Eq.2
for exampie, is not mathematically linear, Rather the
2

( . . . . . . . . . . . )1
3793.6&f2 _Ei _ 3793.6#@2 3000
+Ei -
k~t + At] )( kjW 1
(3) 2930

In Eq. 3, ~ and Z are average values for viscosity and _ 2900

gas deviation factor, respectively: over the pressure drop a

involved: T = formation temperature, ORankine; Q = : Z-b60


~
const~nt production rate, Mcf/i2 at 14,7- psia and 60 F; :
and p = volumetric average reservoir pressure, assumed ,0
: 2800
m
constant. .~.

Horner points out that fault distance d is usually large ~ 2750

enough so that during early hours of shut in; the first Ei -


term in Eqs. 2 or 3 is practically constant, while the sec- 2709
.,
ond Ei term is essentially zero. Then the first part of the
build-up, plot is a straight line having a slope of ~ ~ 265o

7062 26.00 ,.
Oc! .CC2 004 cm iJ! 0? 04 G6. 08 02 04 ;C ,9
+(+) ~.
t*At
when pm(pm,) is plotted vs in [At/( as usual. In-
t+-At)]
terwell permeability k may be obtained from tids first Fig; l-Buildup plots for tile first 1,000 hours shut-in
time for no fault and for a fmdt Iocatesi
straight-line portion, but extrapolation to iniinite shut-in 100 ft from the well.
time would yield art incorrect value for stabilized forma-
tion pressure, p,., (p,O,:). 3010
As A? becomes larger, however, finally the arguments of 3000
both Ei terms in Eqs. 2 or 3 becorr~ less than 0.01 and 2990
the logarithmic approximation can be used. The latter part 2900
of the build-up plot has a slope twice that of the first 2970
pttrt.,l%is second straight line can be extrapolated to in- 2960
finite shut-in time to, obtain pm. (p.,). 29$0

Because thd form of Eqs. 2 and 3 is the same, onIy ,the 2940

oilwell case is illustrated below. .


Fig. 1 shows ~uild-up plots for the first 1,000 hours
shut-in time for no fault and for a fault located at 100 ft
2900-
from the well. The well was assumed to flow at 100 reser- 2$s0 ;.. .--.: ---
voir BUIWYfor 1,20(f Iiotirs prior to shut in; Resemoif,
01 08 .09.10 1522556 $,6 38910
well, rind fluidproperties are shown in F]g. 1, For a case
of a fault, note tiie transition range between the two ap-
parently straight-fine portions.
Fig. .&Build.up plots for fatal! and for no fttuh, weII
Fig. 2 shows the same situation for the first J 00 hours flowed only 12 hours prior to shut-in.
7
/
,

slope of ii build-up plot for a fault changes continuously Calculation of fault distance for the two cases shown
even though the curve has portions that appear to be in Figs. 1 and 2 using Horners method will illustrate
linear. \ the problem. From Fig. 1, the two straight-line portions
intersect at
Methoda for Approxfr&tfhg Fault Distance
At.
From Build-Up Plot = 0.017.
t + At,
Having the build-up plots shown in Figs. 1 and 2 at
hand, we now desire to calculate the distance from the well Thus,
,-
to the fault, assuming of course all the ideali~d condi- 3793,6 (0.15) (1) (2CI X 10) d
tions upon which Eq. 2 is based.
Horners Method
Ei
[

10 (1200) 1
Horners metho& for calculating fault distance d con- = 2.303 log& = 4,07.
sists of first extending the two apparent straight-line por-
tions of Eq. 2 until they intersect at some value of in From Ei function table% Ei (- 0.00!V) = 4.07.
At, Therefore,
corresponding to p.: as shown in Fig. 1.
/+ At. 0.0097(10) (1200) _= ~ 025 ~ ~o,
Horner then equates (r=
3793.6 (0.15) (1) (20 X 10-)
3793.6 @cd- = ,n t+4t,
Ei
(
~t
) 70
(4) d = lol,ft. ,
Recall that the fault is 100 ft from the well; the result
to solve for fault distance. for t = 1,200 hours is perfectly acceptable.
Because problems arise in using this method, it seems From Fig. 2, the two straight-line portions intersect at
worthwhile to; develop Eq, 4 to illustrate the situation. As
was stated earlier, Homer points out that d is usually At.
= 0.63
large enough that during the, early hours of shut-in, the t + At.
first. Ei term in 13q. 2 is practically cm.stant while the sec-
3793.6 (0.15) (1) (20 X 10-)d
ond Ei term is essentially zero. We Can then write from Ei
Eq, 2 an expression for the first straight-line portion [ 10 (12) 1
, 1
Pm = p., + 70.627
fl.@ In , +t = 2.303 log~ = 0.4606..
[ rtAf
3793.6 &J.cti Ei( 0.5933) = 0.4606.
+. Ei . . . . , (5)
( ~~ kt )1 0.5933(10) (12)
#=- --= 62.5 x 10.
Much later, when At has become large enough to make 3793.6 (0.15) (1) (20 X 10-;)
t,, >25 for the last Hi ierm in Eq. 2, logarithmic approx-
imations can be made for both Ei terms, and t second d=79 ft.; .
straight line results, the cqwition of which is This value. is too low by 21 per cent. In general, other
factors held constant, Horners approximation becomes . .
70.62 q,,@ z In At
p. = P,. e +
kh I ,
~
.,,
1 (:) less accurate as flowing time t is reduced.
It can. be noted here that k = 10 md, corresponding
to a slope of 81,3 psi/cycle, was used in both calculations.
At the ,intersection of the straig~t lines represerited by
In Fig. 1 the first -$traight-line portion has an apparent
Eqs. 5 and 6, p. becomes p.. anti since P.,, satisfies
slope of 81 psi/cycle while in Fig. 2 the value is only
both equations at In ~ ~r~t ~the ~esulting identity yields 74.S psi/cycle, which corresponds to k = 10.8 md. If
s: 10,8 md had been used above, a vaiue of d = 82 ft, or
3793.6#pc$ = ;n
Ate an error of 18 per cent, would result.
,(
Ei -, ~t
) t + M,
,.,
,.
(7) This slope difference of 81 psilcycle for r = 1;200
hours and 74. Lpsi/cycle for t = 12 hours is worthy of
repeated emphas~s. In the first instance, the drawdown of
or
. 144 psi at the real well caused by the image well remains
_ ~i : 3793.6@ccf = ,n .
t + At,
( ~ kt At,)
The probiem which aris& in the Horner method can
, (4) essentially constant during the early shut-in hours, In the
second case, however, the image-well drawdowri at the
real well is measurably increasing during the first few
be stated in the following way. The second straight line. hours of shut-in. This gives a qualitative rule that Homers .
Eq. 6, passes through p,,. and has a slope double that method will yield acceptable accuracy if the image-well
for no fault; thus the second straight line is fixed. The drawdown at the real well has stabilized.
first straight line, Eq, 5, results in a -parallel displacement DoIan et aL have used the Horner method to develop
from the no-fault jine at toe instant of bhut-in, At = O. an expression to approximate the distance from a wellbore
The amount ~f vertical displacement is that a barrier can be detected duking a ds@-stem test.
Since the flowing time in a DST is often Short, Horners
70.62q.,p13 .Ei _ 3793.6 #pcti ii
1 method could lead to rather large- e,mors,, In deriving the
, kh ( kt ) appro$rnation, these quthors also .jssurne.$la! the & f~fif~ ... . ,
,. ... ,-, -
and thus depends upon flowing .~m-e i for a given fault .tion ar@s&nt i~lek+stliin 0,01: For nortial VWW of the
distande, d. As the flowing time decreases, the time ratio diffusivity constant, thii approximation would be true onlY
at the intersection point betweert the two straight lines if fault distance were very small. .Notwithstanding the
leads to increasingly erroneous values for d calculated small volume of resetioir tested during a norrnat DST,
withEq. 4. the variable inflow rate usually involved and other corn-
.,
..
. 763
.. . . . . . . .. i. . .. . . . . . .. ,. . . . . ,
. . .
,.
---- .,

placations to the theory, the accuracy of fmdbdisttince and 2 will show, Eq. 10 is noi eqttiwtlent to Horners
determination from a DST record could be improved with method. Eq. 4, because the argument of the Ei term in
less assuming, if somewhat longer calculations. Eq. 4 is not necessarily less than 0,01, a condition neces-
Two-rate flow tests m described by Russell ciur be sary to use of the logarithmic approximation.
used to determine the same reservoir characteristics as are Standing uses Eq. 9 and a correction factor to develop
obtained from build-up tests, often with less operational the following equation for build-up:
problems. The two,rate test involves a constant-rate draw-
k~t. c .
down followed by a second, different constant-rate draw-
,down. The Horner method for fatslt distauce can be used
as Russell has shown, but the duration of the first image-
d = 0.0 I22
(1 -,--*
, )( ~ )
Using Eq. 11 for the extimple in Fig. 1 yields d = 108 ft
(11)

.
well drawdown is important, since it influences the inter- rmd for the example in Fig, 2 yields d = 126 ft. These
section point of the two straight-line sections, and hence values rigree very well with Standings interpretation of
,the calculated fault distance. Horners method for. if in Eq, 4 ~ .+AI, were used in-
Davis and HawkinsS have derived a for.mulu for ex- stead of r, the calculated values of d = 103 ft for F&. ~
plicitly determining fault distance using the time at inter- ilnd d = 130, ft for Fig. 2 would result,
section of the two straight4ine portions of consthnt-rate
drawdown pressure curves. When the test well has been [iuugc-WN BuilsMJp %kthod
tlowingan appreciable time prior to shot-in, the indicateti A second melhod for calculating fuult distance. which
intersection time M, of the two straight-line portions of was suggested to the author in general form by Elkins.
the build-up plot is used in place of the flowing time to has been used with some success. This method involves
yield. in the nomenclature of this paper, graphically measuring the pressure difference between the
first straight-line portion and the actuaI build-up curve
k At,
(
d == l.4ti?. lo-
+fl~ ) (8) during the transition part of the curve where the image-
well build-up becomes significant.
[n Fig. 1. for example, at .M = 60 hours, the apparent
\Vhetl~+~~- is less than shout 30. Davis and Hnwliins image-well build-up is about 43 psi. ,This pressure build-up
,
is equated to the second Ei term in Eq. 2 as follows. -
recommend using Fforncrstnethod, Eq. 4.
For a given fault distance. Horners method gives the ._70,620/.8
kh -

1:=
wune results as Eq. S. only for rather long values of
Ilowing time. /. For example. from Fig. l,A~, = 20.75
hours. Thus. 1
.: ~. _ ~79$6 (0.15) (1) (20
. ..
10 (60)
;; ...
. . . .. 10-) d-
-.. 43.

d == fif~-~;.~fi2~.75)
[ (O.15)(1)[2OX1O-) 1
15
= 101.3 ft,

and from Fig. 2. M, = 20.43 fiours and d =: 100.5 ft.


fj~ z?!:$..(.~ ~5 ) ( 1) ( 2~_.: .JO)fr

== [
1.218.
If) (60)
.-
--1 I
Note that Eq. 8 gives excellent results despite the fact .Ej(-o.201) = 1.218.
t + M, 0,201 (10) (60)
that ~quals 59 ~or Fig. 1 and 1~59 for Fig. 2.
M, r = 3793.6 (0.15) (1) (20 x 10-;) -
In a discussion of Davis agd Hawkins papeti, Standingn = 1;06 X 10.
suggests a method of calculating linear barrier distances
d = 103 ft.
t + N.
when is as low as two, without recoursq to Ei In this case (t = 1,200 hours), the calctilated fault.
At. :,
distance agrees very well with the actual value. The as-
tables or charts. While the method k intended tobe a sumption is, of coursd, that there has been little change
close approximation to that of Horner, standings inter- in the image-well drawdown transient during the shut-in . .
pre~~tion of-the Horner nlethod is somewhat different time involved. At At, = 1 hour, the ityage-welI drawdown
from that of the present author and,others., Specifically. is 144 psi and at At = 60 hours the drawdown is 146
Standing writes the Horner method, not as shown in Eq. 4, psi, so that only an additional 2-psi -drawdown resulted
but as ciuring the 59 hours following At == 1 hour.
For the case shown in Fig. 2 (t = 12 hours), both the
3793.6 .+Fcd >, ,nZ i- At,
Ei
( .
k[t + At,]
in Eq. 9 the duration of the image-well drawdown tran--
) At,
(9) , drawdown and build-up transients from the image well
are producing significant pressure changes at the real -well
throughout the range of shut-in time. The result is that
sient up to the timeof intersection of the two straight the measured pressure difference between an extension of
lines is used, whereas in Eq. 4 only tlowingtime is used. the initial straight line and later portions of the build-up
In, a- check-of Horners paper, however, At. itself is not cti,we does not represent the true, net pres%re of the . ,
used. Using Eq, 9. and the Iogaritbrnic approximation image-well transients, because the points making up the .,
.kl(-x) = in x + 0.5772, standing oblains the same early portion of the curve actually have different intercept
Eq.,8 as developed by Ddvii and. -Hawkirts. If tbis:ap- value?. To illustrate the magnitude of error, from Fig. 2
proach is used employing Homers method,, Eq. 4. fhe tit At = 60 hours, the measured pressure difference from .,~,,
~~s~,~ti$~ :.- ,. . . - ----- . . .. ,~- . : the. kytenrfeci .tlrs~straight part to the L@ral cttrye is 10.? ,. . .. j, -<
p~i. -, -,-....- . . ....<..

I
,, (i =[l.wo-i~+ (,, ~,,, )lO:!
As calctshstions of fault dista,nce using Eq. 10 for Figs. 1
_Ei

= [
_ .

0,286.
3793,6 (0.15) (1) (20 X 106)dJ


10 (60) .1
1,
,.
.
761. . JO IJSSN,?I. OF PETJSOLtWM r ECHNOLIifwy
. ..> .. . .. . .
. . -..= ...,. ...-..... .. .,,., ....-.... :-.
,.

-- Ei ( ().846) = 0.286. ,, = ___ 101 (lo) (20)


from which 70.62(66.67)(1)(1.5)

d= 211 ft. By trial-and-error, d = 100 ft satisfies the identity.


The same approach can be used for Fig., 2, where. at
a result which is too high by 111 per cent. In this case. M= 60 hours the net pressure effect of the image well
at At = 1 hour the image-well drawdown is 9.6, psi, and is 7-5.4 psi. Again, d = 100 ft satisfies the correct identity.
at At = 60 hours it is 50 psi. At M = 60 hours the image-
v.ellbuild-up is 44.7 psi. To this point it has been assumed that the straight-line
build-up plot for ,no fault has been correctly established.
Here again the correct Wdlli? of k = 10 rod was us;d In practice this may not be clear cut, but tie can use the
in the calculations. A valueof k = 10.S md obtained from actual buiid-up ~as a guide (the initial and final slopes as
the slope of 74.5 psi/cycle would have yielded c1= 219 ft.
well as indicated p,,..) to draw in a trial straight line for
hli~l@-\tFeliDrawdown Method no fault. Then trial-and-error calchlations for fault dis-
tance would be made at two values of shut-in time, one
The two methods previously used only involve die
at a small value of At and one, @ some hours later. When
actual build-up plot with fault, Suppose now we are able
the straight line is drawn in .it the correct slope ( assum-
to draw in the straight-line buifd-up plot for no fault. ing p,, . is correct), then ceitxtlated fault distance will he
using p,,, = 3,000 psi as otte point and the known ,slope constant for any value of .kt used. I
of 81.3 psi/cycle. Now fault distance can be calculated
.by equating the graphically measured image-well draw- Additioiinl Methods .
down pressore to the first Ei term in Eq. 2. In Pig. 1 both straight-line portion~ of the actual curye
From Fig. 1 (t = 1,200 hours) at At = 1 hour, the are well established. The first ,section has virtually the cor-
pressure drawdown i+ 144 psi. rect slope, the second straight line has a slope double that
of the first, and pw. is easily obtained. In Fig. 2, however.
the slopes and indicated p,,, are somewhat low. In the

.:~i ..2-.-.-.2
[ 1() .....1=1~~
37936 (() 15) ( 1) (20 ~. It)-)d
(i201 )
general case k, p,.. and d are unknown. These three vari-
ables can be determined by a triaI-and-error procedure
king a computer.
3793.6 (0.15) (1) (20 :< 10-)
d- Another procedure for czzlcukting fault distance involves

=
Ei
;[
--
1
4,075,
10 (1201)
a ,curve.fitting technique using only the actual build-up
plot. This method also is best handled on z computer, and
since the present paper is concerned with ,hand-calculation
from which, as before, methods, these two methods are deferred to a future paper.

d =- 101 ft. Limitations to Pra&ical Application


From Fig: 2 (t = 12 hours) at Ar = 1 hour. drawdown The idealized conditions upon which the foregoing
is -9.6 psi. theory is based have been enumerated in the beginning of
,_Ei _ ~~7~.6 (0.15) (1) (20 X. 10)([- this paper. Thus, the build-up curves shown in Figs. 1 and

:: [
0.272,
10 (13)
.
-1 z may serve as Iimitirig cases for conditions which obtain
in a real reservoir. In applying the foregoing methods to
actual situations several Iimitat ions should be remembered.
whiqh gives the result One Iimita;ion to bear in mind is that in a large nurYl-
d = 99.9 ft. .ber of cases, the familiar swans neck caused by skin
effect ,and/or afterflow, masks a substantial portion of the
Inn~ge-Well Drawdown and hild.Up Medtod early part of the build-up cutve. It is possible that such
The three methods shown above involve direct calcu- distortion, particularly in the case of afterflow, might last
lation for fault distance. Accuracy of the; Horner and long enough to, obscure completely the first straight-line
image-well build-up methods depends, other things, fixed. portion, Thus, not only would there be no clear indica-
upon the duration of the flowing time. The image-well tion of a faiilt, but k derived from the remaining second
drawdown method is not sensitive to this problem, provid- straight-line portion would be too low by a factor of two.
ing the straight line for no fault is correctly established. Also, for a, well extremely close to afault, or if the dif-
A ,fourth method involves a trial-and-error approach in fusivity, constant were extreme!y high, the double slope,
which the graphically measured pressure difference b,e- could be obtained almost immediately and the fault would
tween the acftral build-up plot with fault and a straight not be seen. If geophysical or geologic data are available.
line for no fault is equated to the net effect of both image- the calcldated fatit distance could be compared with that
well drawdown. and build-up transients from Eq. 2. As- evidence.
suming the straight, line for no fault is correctly estab- A second limktion isimposed by the necessity of keej-
lished, the calculation can be made at any value of shut- ing shut-in tim~ as small a? possible. The two Ei functions
in time. ., , in- Eq. 2 do not degenerate to the log approximation until
From Fig. 1, for example, at At = 60 Iiours, the net Jt becomes large enough to reduce the Ei function argu:
pressure effect of the image well is 101 psi: Thus, from ments to less than 0.01. In practice, for usuai values of
Eq. 2. , the diffusivity constarit, unless the fault dist;@ce is verY
small, At must become rather large before the second
-+:E/ L 3793.6
--- 10;15)
,o (1) (20 .~:
(1260), to:)ir - stiafght:lhie portion is firmly eitabliiihcd. If ~~g. f ~,for-
-- [ 1 example, at At =.; 1,000 hours, the argpment of t!e image-.
.,well build-up transient is 0.01138, so that even for that
3793.6 (0.15) (1) (20 X 10-)cf long a shut-in time the second line of double slope is not
-- Ei .. .-. .
[ -- 10 (60) .1 quite established. Lkmg build-up times am seldom desir-
,- c
-, . . .- -.
,.
able and avaihtble data points may lie on the transition At f hwr$)
46 8 16 20 40 60 80 100
part of tie curve. Here, again, a fault might not be evi- r
I
1
I I t
dent, and ?-xtrapolation of the data to infinite shut-in would , ] .]~ + % 3WOWL-4
yield low values for pi. ?9s0

For naturalfy flowing wells, the two limitations dis- 2980

cussed above can be circumvented by using two-rate flow 2970

tests as described by Russell. 2960

A third problem which commonly arises in transient 2950

analyses has to do with the use of drawdown transients 2940

alone or build-up transients aione, and the duration of 2930


these transients relative to each Other, When a well is first 2920
placed on production at constant fate there is only a draw- a= 2910
down transient operating. At shut-in, a build-up transient 2900
is introduced. There can be no build-up transient without -. ..- , - . +--- \ I I
?890 -
there also being a drawdown transient which may, or may 1 I I f 1 f I
2860
not, be causing measurable pressure changes at the well. 01 .02 ,04 .06 08 .10 20 40 60 80 10

The at plot, in which p,. is plottad vs hsAt, is so com- _&


t+At
mon in the literature that many engineers have assumed it .,

At Fig. 4--Re-plotted example of calculations for fnult


is equivalent to piotting p,, vs i in . The problems 6hown in Figs. 1 and 2.
,t i- At r
arising, from this errdneous assumption have been dis-
cussed elsewhere.= The calculated, example of a fault veloped for an infinite reservoir within which a sealing
,shown in Figs. 1 ~d 2 is replotted in Figs. 3 and 4 to fault exists., This limitation is not too serious because tl-
show the two common methods of plotting build-up d~ta, nite reservoirs will behave as though they were infin%
In Fig. 3 (t = 1,200 hours) the two plots are quite simi- until some b~undary effect appears. The methods shown
lar, except that at extended shut-in times the At plot begins above can be applied, using the portions of the build-up
to curve. In Ftg. 4 (t = 12 hours) the two plots: are in curve which reflect infinite behavior.
noway sirnllar. Not only wordd lhere be no indication of
a fault, but the slope of the Ar plot is misleading. As has Summary and Concessions I
been shown, the At plot is-not mathematically linear, and
the twcf plots have the same slope only when At = 0. The Horner method and several alternative approaches
i.e., at shut-in. The reason for the strange shape of the for calculating the approximate distance from a well to .
At pIot [n Fig. 4 is, of course, that the drawdown transient a linear barrier fault using a buildap curve which reflects
is also producing measurable pressure changes at the well such pressure behavior have been shown by example cal-
during shut-in hours. This is not to say that a At plot is culat ion,
not useful; it simply must be used very carefully. DeDendinz
...-. uuon the fault distance. Homers m&hod
Even when jhe flowing time has been fairly long, a A/ might yield substantial error in calculated fault distance
plot can be misleading. In Fig. 3, for example, the two for initird tesfirsg of a, new well or for DST analyses. .
owing to the relatively short duratitin of the flowing time.
straight portions of the AZ plot intersect at AL = 11,2
hours. Recall that the value of At, obtained from the time-
ratio plot was 20.7S hours. Use of M. =11,2 hours in
On the ~other hand, Homers method gives quite accept-
~able results when flowing time has been rather long. The
image-well build-up method likewise becomes inaccurate
~ 1
:$
..

Eq. 8 would yield a value of d = 73.8 ft as compared to


the vaIue of d = 101.3 ft when At= from, the time-ratio at rather small values of t.
plot is used. Although the calculations are longer, the image-weIl
A fjnal limitation to the, methods shown arises. because drawdown, or image-well drawdown-build-up procedures
alI reservoirs are physically finite, whereas @. 2 was de- should be reasonably reliable, even though both methods
require the determination of a build-up plot for no fiudt,
The drawdown method proposed by Davis and Hawkins
At { houls]
8 2 4 $eio 20 40 60 !00 200 .006C0 100D 2000 9000 !0,000
and modiffed- for build-tip data might give acceptable
5000
accuracy; a vahte for d determined in that manner should .
be used as a check against ,. the longer image-weII caictr-
2950
Iations.
Use of a At plot for build-up analysis-is certainly not
~ 2900 precluded, but great care should he exercised in its use.
L
: 2650
8 . Nomestclature
K
u
=, 2800
F:. =, wellbore pressure, psia . -
s
Plc = stabilized formation pre&sre, psia
: 2750
E flu, = pressur% psia, c&responding to intersection of
,, apparent sfraight-line portions of build-up curve
270G /
for fault
2650 q. $ = constant rate of ,prodtt~on. ~B/D
p = viscosity, cp
. .
26110
.0001 0002 .Mry .00[
- /3 ~ f&mation-tiolume.factor,-rcsWvoir bbl;STB. ~ ,. . ~
002 .00+ ,006.01 .02 .04 .06 ,10 20 40 .60 10
II*
-..=2-
,/ - k = effective permeability, rnd
I* AI h = net paythicknes?, ft
Fig, 2...-plottedted exam le of eahxdations fck fault ,. = natural logarithm, base e
- shown in # ]gs. 1 and 2. 1: = comrngri [ogarithm,,...@ise 10,

JO UkNAL OF PETtl OLELKM TECHNOLOGY


.,
., j
.- , :
, --. + .-, . . -----
t-
t = flowing time; hours 7. Russell, D. G,: Determinntion of Formation Characteristics
At = shut-in time, hours ~~m~ Two.Rate Flow Tests, Jour. Pet. Tech, (Dec., 1963)
lCH1.
At. = shut-in time, hours, corresponding to intersec-
8, Davis, E; Grady, Jr. and Hawkins, M. F., Jr.: Ulnear Fluid
tion of apparent straight-line portions of build- Barrfer Detection by Well Premure Measurements, Jour. Pet.
up curve for fault Tech. (Oct., 1963) 1077.
+ = effective porosity, a fraction 9, Standing, M, B.: Discussion of Linear Fbrid.Barrier Detec.
c = system compressibility, psi- tion by Well Pressure Measurement, Jour. Pet. Tech. ( hlareh,
1964) 259.
d = distance from well to fault, ft
10. Elkirrs, Lincoln F,: personal communication.
rW = wellbore radius, ft
11. Gra , K. E.: Pressure Buildup Analysis, Sobio Petroleum
Co. k errort (June. 1960).
Reference 12. Gray, K, E.:- *How to Plot Pressure Build-Up Cur~ed, World
*.
Oi.f (Feb. 1. 1962) 82.
L varr Everdingen, A. F. and Hu~at, W.: Applications of the
Lrrplace Transformation to Flow Problems in Reservoirs, K. E. GRAY rs an assistant professor
Tran.rt,AIME (1949) 186, 305.

L
2. Horner, D. R.: pressure. BuildJJp in wells, ProC., Third of ~etroleum engineering at The U. of
World Pet. Congress, Sec II (1951) 5Q3. Texas. After receiving BS and MS de-
, 3, Gray, K, E.: Lectures in Fluid Flow Through Poro& Media, grees [n petroleum engineering frotn the
The U. of Texas (1962), U. of Tulsa in 1956 and 1957, he
4, Nide, Robert G.: How to Use the Exponential Integral, Pet. worked for two years as a drilling en-
Eng. (Aug., 1956) B-171. gineer for the California Co. at Venice,
5. Tables o}, Sine, Cosine and Exponential Integ~als, Vol. 1 and La., and for one year as a reservoi ett-
II,. Federal works Agency, WPA. %rpe~ntendent of Docu- gineer for Sohio Petroleum at Okla 1 oma ~.
ments, Washington, D.C. J&&&w
City. He then returned to schcol to earn
6. ljolap, J, P., Eharsen, C, A. and Hill, G.-A.: %recial A Pla-
A!ME
his PhD degree ii petroleum enghreering from The U. of
cations of Drill-Stem Test Pressure Data, Trans., ,..
( 1957) 2109318. Texas in 1962.

. .

L: .
.. ., .-..,
,- .. .
.,

., .,

, ).

.,
/
t . ,
!,

/ i,,,

,:

., . . .. . ...=_ .

. .
,- -a

-. 767
.

.- .. i --- ,.. .. .. . . ... . . .. . . . ... .. . . ... . . . . . . -,

Das könnte Ihnen auch gefallen