Professional Documents
Culture Documents
Introduction
With the advent of sensitive pressure gauges and on interference data was examined first in the
because of the need to obtain precise information on groundwater hydrology literature. 2,3 In petroleum
reservoir heterogeneity for implementation of ter- engineering literature, Prats and Scott 4 were the first
tiary recovery programs, interference tests have to consider the effect of well bore storage on the
become popular. It is well-established that the results resporise of a pulse test. They concluded that the
of a muItiwell test are influenced more by reservoir porosity-compressibility product, cj>c I' of the for-
heterogeneity than those of a single-well test. The line mation would be overestimated and the hydraulic
source solution presented first by Theis 1 serves as a diffusivity, k!cJ>clp.., would be underestimated if the
starting point for the analysis of interference test well bore storage effect was significant and not
data. This solution assumes that the storage capacity considered in the data analysis. Also, they noted that
of the flowing well and the skin region around the care should be taken in measuring pressure data if the
sand-face have a negligible effect on the observation- well bore storage effect was significant, because the
well response. This assumption is valid if the distance magnitude of the actual pressure response would be
between the flowing well and the observation well is lower than expected.
large. However, if .the distance is small, then the They did not quantify the effect of a skin region
effect of storage and skin on the observation-well around the well but did discuss qualitatively the
response can be significant. The pressure response at influence of the skin factor. However, Prats and
any point in the reservoir will be damped and delayed Scott assumed that the storage effect existed at the
as a result of the storage effect at the active well (with observation well and not the flowing well. In a
or without the skin effect), because the main effect of subsequent paper, we show that their results can be
the storage capacity of the flowing well is to cause a used to analyze pulse test data influenced by the
lag in the time required for the surface rate to equal storage effect at the active well.
the sand-face rate. For given reservoir properties, the For the problem at hand, Jargon 5 was the first to
magnitude of the delay depends on the well bore quantitatively examine the influence of well bore
volume, the nature of the storage phenomenon, and storage and skin effects. Jargon showed that if
the skin effect. well bore storage and skin at the active well influence
To our knowledge, the effect of wellbore storage
01492136/80/00018029$00.25
'Now with the Ins!. Mexicano del Petroleo. 1980 Society of Petroleum Engineers of AIME
This paper presents new solutions for analyzing interference test data affected by
wellbore storage and skin effects at the flowing well. Three parameters of interest-
wellbore storage phenomenon, skin effects, and distance between two wells - are
considered simultaneously. The new type curves should improve significantly the
ability to analyze such data.
JANUARY 1980 171
the pressure behavior at the observation well, and if been evaluated to obtain the pressure drop as a
the line source solution is used to analyze data, the function of time. The main difficulty appears to be
permeability-thickness product will be un- the numerical evaluation of the Laplace inversion
derestimated and the porosity-compressibility integral. An algorithm to invert the Laplace trans-
product will be overestimated. Jargon was the first to form numerically came to our attention after this
present type curves to estimate the permeability- work was completed. We consider this algorithm in
thickness and porosity-compressibility products. the Discussion section of this paper.
However, his attempts to include skin effect at the We have computed the wellbore pressure drop by
active well as a correlating parameter were un- the principle of superposition, because the sand-face
successful. He was able to obtain correlations only as flow rate has been tabulated in the literature by
a function of the wellbore storage constant, CD' for Ramey and Agarwal 8 and Ramey et 01.9 Admittedly
specific values of the skin factor, s, where CD and s this is an approximate procedure, because the
are defined 6 as: solutions will not apply at small distances (less than a
few feet) from the well bore. However, this limitation
is not important, because we principally are in-
5.615C C terested in analyzing interference test data and in
C D =2 h 2= h 2' . . . . . . . . . . (1)
7rc/>C trw 27rc/>Ct r w such cases the distance between the active and ob-
servation wells is larger. We discuss the limitations of
this procedure in the Discussion section.
khilpskin 27r X 10 - 9 khilpskin
s= =------..::..:.:.:=. ..... (2) Applying the principle of superposition, the
141.2qBp. qBp. pressure drop in the reservoir may be computed by
The first equality in Eqs. 1 and 2 should be used qBp. f. (qB) sfi
when quantities are measured in conventional oilfield Pi-p(r,/)=-2 10- 9 kh ~ B
units, and the second should be used when 7rX 1=1 q
measurements are made in SI metric units. 2
In this paper, we present a generalized method ( o[
. 1f2El - 1 9 c/>c t w ]J ' . . . . . . (3)
wherein all three parameters of interest - the 3.6xlO- 4k(t-t i )
well bore storage phenomenon, the skin effect, and
the distance between wells - are considered where (qB) sf is the sand-face rate, q is the surface
simultaneously. The basis for simultaneously con- rate, and - Ei ( - u) is the exponential integral. 10 All
sidering the three parameters is demonstrated quantities in Eq. 3 are expressed in SI metric units. If
rigorously. (Jargon obtained correlating parameters conventional oilfield units are used, the constants
in an empirical fashion.) New type curves are O.57rx 10- 9 and 3.6x 10- 9 should be replaced by
presented for analyzing data when the wellbore 141.2 and 2.64 x 10 -4, respectively. By introducing
storage and skin effects at the active well influence dimensionless variables, II Eq 3 becomes 0
Q.
o
Q.
o
a:
o
r 2 w
.Ei[- 4(tD~tDi)J . ................. (8)
a:
~
(/)
w
a:
Q.
I
Noting that (qB) sfjlqB is a function of IDICD and (/)
(/)
W
17-----20
2
C D exp(2s), Eq. 8 indicates that the dimensionless ..J
2
5X 10
o ' - -_ _ _ 10 4
pressure drop should correlate in terms of three u; - - -_ _ 10 10
parameters: IDICD , I D lrD 2 and C D exp(2s).
2
w , - - -_ _ _ 1020
::;
Because I D I r D 2 is used as a parameter in all in- o COexp(2s)
for all combinations of CD' s, and rD' In this in- r:L LINE SOURCE
0
a: SOLUTlON~
vestigation, the results provided by Ramey et 01. and 0 .....-:~
w
Ramey and Agarwal were used to test the above a:
=> ./
hypothesis for a number of combinations of CD's, en
en
/
w
a:
/
and rD' In all instances it was found that the above Q. /
hypothesis was justified. en
(I) 1 /
w 10
Z
..J I
Results ~ I
z
Figs. 1 through 3 are sample type curves obtained in w
2
/
this study. The dimensionless pressure drop was 0
computed from Eq. 8 using the tabulated values of I~~~~~~~~~~~~~~~~~~~
2 3
10' I (0 10 10
qBsflqB given by Ramey et 01.... Type curves were DIMENSIONLESS TIME TO DIMENSIONLESS DISTANCE SQUARED"D/r~
prepared for 18 values of CDlrD2 in the range 2.5 X
10- 3 '5.CD lrD 2 '5.10 for all values of CD exp(2s) Fig. 2-povs toir02 due to well bore storage and skin at a
presented by Ramey et 01. Appendix A * * presents flowing well in an infinite reservoir, C o ir02 1. =
tabulated values of PD vs tDlrD2 for the range of
10
parameters of interest. The line source solution 1
co/r~ .. 10
(CD =0, s=O) is shown as a dashed line on Figs. 1 ., Q
through 3. The shape of the CD exp(2s) >0 curves is ..-0 LINE SOURCE ~ ___
a: SOLUTION~.",.,........-- __
seen to be similar to the line source solution. This 0
w ,/ .,
implies that it would be difficult to determine if a:
,/
interference data are affected by wellbore storage '"'"w" ,/ I
0:
Il.
/ . 20
and skin unless data also are obtained at the flowing U) / 5XI0
2
"Large-scale type curves of parameters shown here and Appendix A may be Fig. 3-povs to/(02 due to well bore storage and skin at a
obtained from R. Raghavan, U. of Tulsa, Dept. of Petroleum Engineering, 600 S.
College Ave., Tulsa, OK 74104. flowing well in an infinite reservOir, Co 1(02 10. =
JANUARY 1980 173
the well bore storage-skin cases is large at early times.
The magnitude of the difference depends on C D lrD 2 Discussion
and C Dexp(2s). Ultimately, all cases merge with the In this section we discuss the conditions under which
line source solution, indicating that the wellbore the solutions presented are valid. Also presented is an
storage effect at the flowing well has become overview of the parameters that control the pressure
negligible. A comparison of the various curves for distribution in the reservoir and the correlating
constant values of C Dexp(2s) indicates that if groups that can be used to describe the pressure
CD I r D 2 is small, the results are close to the line distribution. As is well known, the pressure response
source solution. On the other hand, if CDlrD2 is at an observation well due to production (or in-
large, the difference in computed pressure drops jection) through a well of finite radius can be
between the line source solution and the well bore represented by the line source solution I for all values
storage-skin curves is large. This result can be ex- of tDlrD 2 provided rf. ~20 and at all values of rD
plained readily. A large value of C D lrD 2 implies provided t D l r D 2 ~70. 4 .
that either the storage constant is large or the Similarly, we should not expect the solution
distance between the two wells is small. Con- presented here to be representative of the pressure
sequently, one would expect a considerable dif- drop at all values of rD for all times of interest. The
ference in pressure behavior at early times. A small range of conditions for which our solutions are
value of CD I r D 2 implies that either the storage applicable can be established only by comparing the
constant is small or distance between the two wells is results of the rigorous solution with the approximate
large. Thus, for this case the difference in pressure solution.
drops would be small. In addition to the above As shown by Agarwal et al., the complete solution
observations, the results shown in Figs. 1 through 3 to the problem under consideration may be obtained
indicate that as the skin effect increases for a given by the application of the Laplace transformation
value of C D lrD 2 , the difference in computed technique. In Laplace space, the pressure distribution
pressure drops between the line source and well bore in the reservoir is given by
storage-skin cases increases.
Fig. 4 presents a comparison of the results ob-
tained here and those obtained by Jargon. The results
of our study compare favorably with Jargon's work.
Although the approach used in this study is more
general, many of Jargon's conclusions pertaining to
the effects of well bore storage and skin on in- ................................... (9)
terference data are still valid. Thus, it does not
appear worthwhile to repeat all of the useful in-
formation presented in Conclusions 1 through 5 and where p is the variable in Laplace space (s, which is
9 of Ref. 5. commonly' used to denote the Laplace space, is not
Finally, one difference between the type curves used here because s represents skin factor); PD (rD)
obtained here and those by Jargon should be is the Laplace transformation of p D (r D' t D); and
mentioned. The Ramey et al. and Ramey and Ko (x) and KI (x) are modified Bessel functions of
Agarwal studies represented the skin region by the the second kind of order zero and one, respectively. 7
van Everdingen-Hurst I2 ,13 skin effect. Jargon, The pressure distribution may be obtained by
however, represented the negative skin factor by a applying the inversion theorem for Laplace trans-
region of high permeability and finite thickness. formation 7 to Eq. 9. However, the inversion integral
Thus, for small values of C Dexp(2s), the results would have to be evaluated numerically. This can be
obtained here and those in Ref. 5 are different. a formidable problem from a computational
viewpoint; it may be simpler to solve the problem by
the method of finite differences, as was done by
Jargon. However, it was shown recently that the
algorithm presented by Stehfest 15 may be used to
...
Q
'"
a: We found that Eq. 8 predicted the pressure
CL
U)
Coup (2.) response accurately for all times of interest for
10 4
rD ~20. We also found that if C D lrD 2 ~O.25, Eq. 8
U)
'"
..J
10
5
~
U) 10
6 was accurate if rD ~ 10. Therefore, we concluded
;o
that the correlating parameters (t D I r D 2 , C Dexp(2s) ,
and C D lrD 2 ) are sufficient for analyzing all in-
IclLL.,...I..LJ.....,.,.I4.u..IJ-.....L...J...J....L..I.JwJ........JL......L.J..J..&.I.LI.L;;-..I-..L...J..l..U.W
2 3
terference test data dominated by wellbore storage
10"' I 10 10 10 skin, because the dimensionless distance between
OtMENSIONLESS TIME TO DIMENSIONLESS DISTANCE SQUARED, I 0/' ~
wells for all interference tests is at least one order of
Fig. 4 - Comparison of results of this study with the magnitude greater than 20.
Jargon study. In the process of testing the solutions, we also
174 JOURNAL OF PETROLEUM TECHNOLOGY
TABLE 1 - APPLICABILITY OF THE STEHFEST C D exp(2s) and tDICD if C D exp(2s) is large. If
ALGORITHM C D exp(2s) is small, it is a function of t D , CD' ands.
The Stehfest algorithm does not work for the following Cases 2 and 3 are of academic interest only as far
range of values of Coexp (2s) and C o Jro 2 . as interference testing is concerned. Thus, we can
Co exp(2s) C o lf02 conclude that the three dimensionless groups
1 ~7.5 X 10- 2 [tDlrD2, CDlrD2, and C D exp(2s)] are sufficient. to
10 ~7.5 X 10- 1 analyze interference test data. If these correlatmg
10 2 ~ 7.5 groups are tested to determine if they describe the
pressure response at distances close to the well bore,
found that the Stehfest algorithm is unsuitable for
they will break down. These groups do not (and were
the problem under consideration if s:::;; - 2. That is,
never intended to) apply when rD is small.
the algorithm may not apply for small values of
C Dexp(2s). This results because there is a pole on the Example
positive real axis of the complex p-plane If s < 0 and
A computer-generated example intended to
the inversion theorem does not take this into ac- demonstrate the use of the type curves is presented
count. This observation also applies to the results here. Table 2 presents pressure data obtained during
obtained by Agarwal et al. even though Eq. 9 is the interference test. The two observation wells are
correct for all values of s when s<O. Our in- located approximately 30 m (100 ft) from the in-
vestigation indicated that no solution can be obtained jection well (Fig. 5). From Fig. 5 it is evident that
if the numerical inverter is used for the conditions well bore storage controls pressure behavior at the
given in Table 1. injection well at early times. However, we note that
The validity of our solution cannot be established, the test has been run long enough to analyze pressure
then, unless the inversion integral is evaluated
data at the flowing well by the,semilog approach. The
properly when s:::;; - 2. To our knowledge, no simple
data at the flowing well were analyzed first to
method exists whereby a pole on the positive side of
determine the effective permeability and skin factor.
the teal axis can be taken into account while applying
We estimated the permeability to be 20 md and the
the inversion theorem. Agarwal et al. avoided this
skin factor to be 20.2. The dimensionless storaye
problem by using the effective well bore radius
constant, CD' computed from the unit slope line 8
concept-i.e., by replacing tD by tDexp(2s) and CD
on the log-log graph was found to be 9.91 x 10 3 .
by C D exp(2s) and assuming the skin factor term to
These calculations are not presented here as the
be zero. Therefore, the results presented here can not
procedure has been explained thoroughly in the
be validated independently for s:::;; - 2 unless the
literature.
problem is solved by a different approach. However,
Jargon has shown that the time at which the
considering that all the other solutions are accurate
well bore storage effect will become negligible at an
for rD ~20, further verification is n~t neede~. T~e observation well is given by
Stehfest algorithm should be used with care If s IS
negative. More details are given in Ref. 16. tD (CD )0.86
Although it is established in the literature, one -2 =(230+ ISs) -2 . . ............ (10)
final point regarding the correlation presented here rD rD
should be noted. Both Earlougher and Kersch 17 and
Ramey et al. have pointed out that C D exp(2s) is not Using Eq. 10 we find that the well bore storage effects
a good correlating function to represent the well bore at Wells A and B would become negligible only after
pressure drop and sand-face flow rate, respectively, 173 and 167 hours, respectively. Consequently, the
if C D exp(2s) is small. This limitation also applies to type curves presented here, which consider well bore
the results presented here. We used C Dexp(2s) over storage and skin at the active well, should be used to
the r~n~e 1:::;; C Dexp(2s) :::;; 1040 for several values of
CDlrD to test its validity. We found that if rD ~20,
then C Dexp(2s) is an adequate correlating parameter
SEMI-LOG ANALYSIS -2 2
even for small values of CDexp(2s). Only if rD <20 PERMEABILITY,k a l.97XI0 fLm (20md) 0 ~ 0 0
SKIN FACTOR,' a 20.2 0 START SEMI-LOG
does the group C D exp(2s) break down as a STRAIGHT LINE
correlating parameter. o
In summary, an examination of the pressure
distribution surrounding a well su,bject to well bore o
,;
storage and skin effects indicates the following. ,;
APPROXIMATE END
~
1. If rD ~20 (in a number of cases rD ~ 10), the . OF STORAGE PERIOD
dimensionless pressure drop is a function of the three UNIT SLOPE LINE
groups (tDlrD2, C D exp(2s) , and C D lrD 2 ) for all
eB
(CO a 9.91 X 10 3 ) T
27m
N
I
values of the parameters of interest.
~ A
2. If r D < 20 and if C Dexp(2s) is large (~I000),
the pressure drop is a function of rD' t D , and
~3Im--i
C D exp(2s) .
3. If rD <20 and C D exp(2s) < 1000, the pressure
FLOW TIME,' hours
drop is a function of rD' t D , CD' ands.
4. The well bore pressure drop is a function of Fig. 5 - Pressure change vs flowing time at the injector.
Pressure Data
Flowing Well Observation Well A Observation Well B
Time Pressure Change Time Pressure Change Time Pressure Change
(hours) (psi) (kPa) (hours) (psi) (kPa) (hours) (psi) (kPa)
0 0 0 0 0 0 0 0 0
-~
Po = 6.2X I02,LlP= 10kPa
102 __ _-1 'O/r~ a 1.17,' a 2.78 hours
lo2l__ 'O/r~ a 8.7XI6~' a2.78hour.
16
1
'o/r6 I 10
1
'o/r~
I 2
I 2 1 10 10
16
10-1 10 10
FLOW TIME,', hour.
Fl OW TIME,', hour.
Fig. 6 - Type-curve match of the pressure data at Well A. Fig. 7 - Typecurve match of the pressure data at Well B.
(~ 1000), the well bore storage constant and skin Jl = viscosity of flowing fluid, Pa s (cp)
factor can be combined in terms of the group <I> = porosity, fraction of bulk volume
C D exp(2s),
Subscripts
Nomenclature D = dimensionless
B=formation volume factor, res i= initial
m 3 /stock-tank m 3 (RB/STB) r = radial dimension
c t = total system effective isothermal sf = sand-face
compressibility, kPa - I (psi - I) t= total
C= unit storage factor, m 3 IkPa (RB/psi) w=well
CD = dimensionless storage constant wf = well flowing
-Ei( -u) = exponential integral,