Professional Documents
Culture Documents
Earlougher and Kersch Wel Test Analysist
Earlougher and Kersch Wel Test Analysist
Earlougher and Kersch Wel Test Analysist
By Type-Curve Matching
Robert C. Earlougher, Jr., SPE-AIME, Marathon Oil Co.
Keith M. Kersch, SPE-AIME, Marathon Oil Co.
Introduction
Occasionally, insufficient transient test data are avail- culated, or both. These approaches are tedious and
able for analysis using semiiogarithmic i plotting not always successful.
methods. This usually happens when data collection In spite of its disadvantages, curve matching seems
stops before wellbore storage (afterflow) has become to be the most promising of the methods, particularly
negligible. Under those circumstances, the semi- for the engineer who does not have a computer avail-
logarithmic straight line does not develop, and com- able. Cooper et aI.7 present type curves and an analysis
mon semilogarithmic analysis methods cannot be technique for specific flow and injection tests with the
used. When such methods cannot be used, the engi- well shut in before testing. At the start of the test,
neer either obtains no information from the test or the pressure instantaneously changes to some new
must use the available, short-time data to estimate value. Then both pressure and flow rate vary during
reservoir characteristics. This paper presents a tech- the test. The Cooper-Bredehoeft-Papadopulos type
nique for the approximate analysis of such short-time curves are useful for analyzing data taken during the
transient test data. The method applies to buildup, flow period of a drills tern test. 11 ,12 Agarwal et aI.9
falloff, drawdown, and injectivity tests when wellbore point out that neglect of the skin effect makes the
storage effects are important. It should not be used if Cooper-Bredehoeft-Papadopulos type curves of dubi-
data can be analyzed by more conventional, semi- ous value. In any case, those curves do not apply to
logarithmic plotting methods. the more common transient testing situations: buildup,
It has long been recognized that wellbore storage falloff, injectivity, and drawdown. Ramey 8 and Agar-
(afterfiow) can impede pressure transient test anal- wal et aI.9 suggest type-curve matching for these kinds
ysis. 2,3 Several ways have been suggested for deter- of transient tests. They present applicable type curves
mining when well known semilogarithmic techniques that form several families of curves with skin
can be used for transient test analysis. I, 3, 4 Gladfelter factor and wellbore storage coefficient as parameters.
et al. 3, I and Russell" present calculational methods for Ramey's curve-matching method requires that the
analyzing the portion of transient test data influenced data plot be slid both horizontally and vertically to
by wellbore storage. Curve matching 7 - 12 and regres- obtain a match. This feature and the fact that the
sion techniques '3 have also been proposed for accom- curves have very similar shapes make the matching
plishing such analyses. technique difficult to use unless there are data at least
All these methods have disadvantages. The tech- onto the start of the semilog straight line.
niques presented by Gladfelter et aI.5 Russell,6 and McKinleylO uses a similar approach, but with a
Earlougher and Kersch 13 utilize either trial-and-error different kind of type-curve plot. He plots his type
analysis or require that the afterflow schedule be cal- curves so there is only one family of curves and
?
.2.5
I.
"
0.1 ~ °1
val
(\J , 0"
8
0.1_
<J <J
a
+"==-+
D112S
II
1.!
SPE 4488 ANALYSIS OF SHORT TIME TRANSIENT TEST DATA 8Y ~;;;p:;:;-.
P:~'.5
TYPE CURVE MATCHING
't8. R.C. EARlOUGHER. Jr. and K.M. KERSCH
0.01
m Ts
© 1973, MARATHON Oil COMPANY
25 9 1
kh LH
p:' C
2.5
t
md ft hr
cp' bbl/psi
10 3 10 4 10 5 106
-..j
'C>
VI Fig, l-Master type curve (reprinted by perm1ssion of Marathon Oil Co.).
on the type curve of Fig. 1. curves to the same data. The calculated value of the
4. Place the type curve data plot over Fig. 1 so skin factor will probably be inexact, also. However,
the asymptote calculated in Eq. 5 overlies the value the skin factor calculated from Eq. 8 should indi-
of 1.0 on the ordinate of Fig. 1; that is, so cate the relative amount of damage or improvement.
This curve-matching approach will give a much
Il P 24C) = 10 more accurate value of formation transmissibility if
( Ilt qB Fig. 1 •.
the wellbore storage coefficient and skin factor are
5. Slide the data plot horizontally until the best known independently. Then in the matching process,
match is obtained with one of the curves on Fig. 1. one has the horizontal alignment from the wellbore
To get a good match it may be necessary to add a storage coefficient and can pick the curve in Fig. 1
slight amount of vertical movement to the data plot. that should be matched.
In any case, it is important that the grids of the two
type curves be kept parallel to each other.
Examples
6. Sketch the matched curve onto the data plot. Example 1
From Fig. 1 read the value of A pressure drawdown test on a new oil well appears
to be strongly influenced by wellbore storage. Never-
( CD e2.) Fig.1,M
theless, enough data exist to determine formation
properties from the semilog plot. We analyze this test
Pick any convenient match point and read the values with the techniques presented in this paper, and
of compare the results with results from the semilog
analysis. Table 1 gives pressure time data. Other
( ~) M ,(llt)M known data are
qo = 179 STB/D, Ct = 8.2 X 10- 6 psi-I,
from the data plot and the values lying directly under Bo = 1.2 RB/STB, rw = 0.276 ft,
this point from Fig. 1: h = 35 ft, 1> = 18 percent.
kh
C (kh
p.
flt)
C Fig.l,M
TABLE 1-PRESSURE DATA FOR EXAMPLE 1
Time, Ilt Pressure Change, IIp IIp/Ilt
- ---'-'--(-:-Il-,t),...:;M----'''--'-- (7) (hours) (psi) (psi/hr)
p.
0.2 19.7 98.50
where C is from Eq. 6. 0.3 28.1 93.67
9. Estimate the skin factor from 0.5 43.1 86.20
0.7 58.3 83.29
s
= ~I
2 n
(1) C t hrW 2 (CD e28hig.l,M)
0.89359 C .
(8) 1.0 75.1 75.10
2.0 114.5 57.25
Although we feel this method is an improvement 3.0 135.5 45.17
5.0 152.2 30.44
over other type-curve matching methods presented in 23.31
7.0 163.2
the literature, it is still not exact. Values of kh/ p. may 10.0 166.7 16.67
be uncertain within about a factor of three, as illus- 20.0 171.2 8.56
trated in the Examples Section. This uncertainty 30.0 173.9 5.80
occurs because of the similarity in shape of the type 50.0 175.2 3.50
curves and the possibility of matching two or three 70.0 177.1 2.53
'00r------r------,----------.,
'"'"
w
~
z
'2
'"~
"
is 0.1
~l·m
OJ ~
.a!<i
0.01'-:--------'--;--------''-:-------'
103 104 10 5 '~L,- - - - - - ' - - - - - - - - - , L- - - - - - - " 0 0
kh at MO FT HR O
P:" C' ----cP. BBl/PSI 6t,hr
Fig. 2-Example I-drawdown test on a new oil well. Fig. 3-Example 2-buildup test in a pumping well.
and B = 1.0 RB/STB. Although McKinley does not using Fig. 1. The results given in Table 3 compare
specifically give a value of <PfLCtfw2/k, we assume it to quite well with the parameters used by McKinley to
be 1.028 X 10- 7 cp sq ft/(md psi), the value used for calculate the data.
his type curves.
This example indicates that errors of a factor of Example 5
2 to 3 can occur in the permeability calculation. N ev- Russe1l 6 provides the field data shown in the type
ertheless, it is encouraging that even though CD e 28 curve of Fig. 6. These data are also used by McKin-
varied by a factor of 10 3 in this example, kh/ fL varied ley in his Example F -2."0 Since there are no data points
by a factor of only 2. When all other methods fail, it at very early times, these data are difficult to analyze.
is better to know permeability within a factor of 2 or Equally good data matches can be obtained using the
3 than not at all. Unfortunately, the uncertainty in CD e 28 = 10"0 or the CD e 28 = 10 20 curves, shown as
skin factor can be worse. If the uncertainty in CD e 28 the dashed and solid lines, respectively, in Fig. 6.
is lOX, then 'the skin factor will be uncertain by about Using Russell's data of q = 157 STB/D; B = 1.6
the additive quantity, 1.15 X. RB/STB; fL = 0.3 cp; h = 4.0 ft; <P = 0.10; Ct =
2 X 10- 5 , and assuming fw = 0.25 ft, we obtain the
Example 4 results shown in Table 4. Both Russell's and McKin-
Fig. 5 is a type-curve plot of McKinley's10 calculated ley's results are shown in the table for comparison.
data (McKinley's Fig. 6a and 6b). The triangles rep- In spite of the marginal quality of the data, the
resent calculated test response for a well with s = - 2. most useful results are obtained using the matching
The circles represent response calculated for a well procedure presented in this paper. If several CD e 28
with s = + 2. McKinley's other data are kh/ fL = 1,000 curves seem to match the data equally well, then we
md ft/cp; q = 100 STB/D; B = 1.0 RB/STB; C = can estimate a fange of transmissibility. This gives an
0.178 bbl/psi; (<PfJ_Ctfw2)/k = 1.028 X 10- 7 cp sq ft/ idea of the accuracy of the results. In this case, we
(md psi). conclude that the transmissibility is between 370 and
Fig. 5 also shows the type-curve matches obtained 640 and that the well is badly damaged.
IO'r------,-------,-------, - .6.t,hr
I 0 0 r - - - - - - , - - - - - - - , - - - - - - - ,~
il.1,hr 61,hr
Fig. 4-Example 3-calculated data from Fig. 5-Example 4-calculated data for damaged
Fig. 4 of McKinley,'· and improved wells, from McKinley,'·
100.-------,---:-------,1:---------,
{
[~l: 13.4 po;/h,
Coe2SSJOlO-
CoelS: 10 10 ------
[~.2,cl· 1.01
6t qB
FiV.I,M
[JI'l •• 16h, I
(
[~'~l
p. C
"IO.. mdft _ _
Fi9.1, M cp
h'_
bbl/psi
~ X
~ 100h~=='----+-------t-~------1
ilOl~==~~~.~.~~.~C.~~~M~~~~h~p~o;n~I------~------~
10"
!
COe'~
:I~
[~l"
t:.t
100 p,i
If hr
2! •
Coe =100 ••••.
l!.P
Tt"qa
[
z4cl
'" 0.1
Fit· I ,"
[.61] .. t 0.165 hr
[~'~l
fl. C
'"
to·~·_h'_
Fit·I,M
cp bbl/psi
10L------~------L-----~ IL-------ILO-----~~~----~IOOO
O.! 10 JOO
61, hr
at. hr
Fig. 6--Example 5-pressure buildup test, Fig. 7-Example 6--buiJdup test on a gas· lift well,
data from RusselL" data from Example F·3 of McKinley.'o