You are on page 1of 9

A NEW DECONVOLUTION METHOD TO ANALYZE

WELLBORE STORAGE DISTORTED DATA FROM PRESSURE


BUILDUP TEST
by
Taufan Marhaendrajana*

Sari
Wellbore storage adalah phenomena yang umum yang teramati selama well test (pressure drawdown
tes dan pressure buildup tes). Selama pengaruh wellbore storage, data yang diperoleh dari tes tidak
dapat dianalisa untuk mendapatkan parameter-parameter reservoir (seperti permeabilitas dan skin)
dengan menggunakan metoda yang konvensional (analisa menggunakan semilog, yaitu metoda Horner
dan MDH). Jika hanya data periode ini dianalisa menggunakan metode type-curve, hasilnya tidaklah
unik. Karena itu, tes yang dilakukan dilapangan dijalankan dengan waktu yang cukup sampai periode
pengaruh wellbore storage ini terlewati. Kadang-kadang waktu yang diperlukan untuk melewati
periode wellbore storage ini cukup panjang dan hal ini tidak diinginkan. Hal lain untuk mengatasi
fenomena wellbore storage ini adalah dengan melakukan penutupan di dasar sumur, namun biaya yang
diperlukan juga lebih mahal.
Makalah ini menyajikan metoda baru untuk menganalisa data yang dipengaruhi wellbore storage, yang
memungkinkan pemanfaatan data ini untuk mendapatkan permeabilitas dan skin. Respon dari reservoir
yang bebas dari pengaruh wellbore storage dipisahkan dengan menggunakan teknik dekonvolusi.
Kemudian, data yang sudah terpisahkan dari pengaruh wellbore storage dapat dianalisa menggunakan
teknik metode Horner atau MDH. Teknik dekonvolusi ini diuji menggunakan data yang diperoleh dari
program simulasi reservoir numerik. Teknik ini juga diuji dan digunakan terhadap data lapangan.
Kata kunci : kapasitas sumur, dekonvolusi, tekanan transien, tes sumur.

Abstract
Wellbore storage is a common phenomenon observed in well testing (pressure drawdown or pressure
buildup test). It obscures radial flow response from reservoir, during which it is difficult to obtain
reservoir flow properties (i.e. permeability and skin factor) using conventional analysis methods
(semilog analyses, i.e., Horner and MDH). In this period, it also posses non-uniqueness problem if the
wellbore storage distorted data alone is analyzed using type-curve method. To alleviate this problem, in
todays practice, the test is conducted for sufficient time to allow the wellbore storage diminishes.
Sometimes it requires a long test duration (i.e. it requires shut-in the well for a long period for pressure
buildup test), which is not desirable. Another technique to overcome the wellbore storage effect is to
shut-in the well at downhole, but this operation is costly.
This paper proposes a new method to analyze wellbore storage distorted data, hence it enables the
analysis of short-time pressure transient test data to obtain accurate estimates of reservoir permeability
and skin factor. The true reservoir response is extracted from the wellbore storage distorted data using
a deconvolution technique. The extracted data then can be analyzed using Horner or MDH methods.
The deconvolution method as a result of this study is validated using synthetic cases generated from
numerical reservoir simulator. The applications of this method to field data are also presented.
Keywords : wellbore storage, deconvolution, pressure transient, well test.
* Department of Petroleum Engineering, Institut Teknologi Bandung

I. INTRODUCTION
The use of deconvolution technique to analyze
well test data was first introduced by Gladfelter
et al.1 It was utilized to determine the constant
rate behavior of the well-reservoir system from
simultaneously measured downhole flow rate
(afterglow) and pressure data. They stated that
the reciprocal productivity index (pws/qsf)is a
linear function of the logarithm of time for
buildup tests where pws is the change in shut-in
pressure and qsf is the rate during afterglow.
The validity of the Gladfelter deconvolution was
investigated by Kuchuk2 for different wellbore
geometries such as line, cylindrical, and
spherical source wells, as well as fractured well.
He concluded that the Gladfelter deconvolution
is valid only if the downhole flow rate varies
linearly with time, and it does not work if the
downhole rate varies arbitrarily. One important
conclusion was that the commonly assumed first
semilog straight line due to the Gladfelter
deconvolution (if it works) is a tangential line
that is parallel to the final semilog straight line.
There have been many authors studied this
subject. Among others are Winestock and
Colpitts,3 Ramey,4,5 Fetkovich and Vienot,6
Kuchuk and Ayestaran,7,8 Thompson et al.,9
Thompson and Reynold,10 Meunier and
Wittmann,11 Stewart et al.,12 Mendes et al.,13
Johnston and Lee,14 Fair and Simmons,15 Baygun
et al.,16 and von Schroeter et al.17
Stewart et al.12 modeled the measured sandface
rate using a discrete linear function. Analytical
expression of the deconvolved equations was
then obtained for infinite acting homogeneous
reservoir, and for a reservoir with linear
discontinuity. Mendes et al., 13 and Fair and
Simmons15 developed a technique for
deconvolution computation in Laplace domain,
and others16,17 developed deconvolution methods
to improve the accuracy dealing with the noisy
downhole flowrate data. All these methods are
used if both downhole pressure and flowrate are
measured.
Johnston and Lee,14 developed a method which
calculates sandface rates after shut-in from the
pressure data measured during a buildup test for
gas wells; therefore, rates need not be measured
during a test to be able to remove the effects of
afterglow from test data. However, we need to
know the cross-sectional area of the production
string as a function of depth.
In this study, we attempt to develop a
deconvolution method by only using information

of measured downhole pressure data. Specifically, the objectives of this paper are:
z To investigate a method to analyze wellbore
storage distorted data, hence it enables the
analysis of short-time pressure transient test
data to obtain accurate estimates of reservoir
permeability and
skin factor. The true
reservoir response is extracted from the
wellbore storage distorted data using a
deconvolution technique. The extracted data
then can be analyzed using Horner or MDH
methods.
z To validate the deconvolution method as a
result of this study using synthetic cases.
z To provide an analysis procedure for using
the deconvolution method.
z To
demonstrate the application of the
deconvolution method to field data.

II. MODEL AND ASSUMPTIONS


In this work we assume a vertical well producing
from a homogeneous reservoir. A single-phase
flow is considered in this paper. It is also
assumed that during testing only bottom-hole
pressure is recorded. The test is conducted at
constant surface rate, and the relationship
between surface rate/pressure and sandface
rate/pressure can be represented by this
equation.18
dp
dp
q D = 1 C D wD tD ...................... (1)
dt D
dt D
where qD is dimensionless sandface rate; CD is
dimensionless wellbore storage coefficient; pwD
is dimensionless bottom hole flowing pressure;
ptD is dimensionless tubing pressure; and tD is
dimensionless time.
The tubing pressure variation during the test is
assumed very small and hence it is considered
constant. Therefore, Eq. 1 becomes:
dp wD
..................................... (2)
qD = 1 CD
dt D
The pressure response of a variable-rate system
is given by the well-known convolution integral
(Duhamel theorem) which takes the form:
tD

p wD = q D ( )
0

dp sD
(t D )d ................ (3)
d

where psD is the dimensionless pressure solution


of a vertical well for constant-rate system. This
includes near-wellbore skin effect. Writing Eqs.
2 and 3 in Laplace domain, we have:
1
q D = sC D p sD ...................................... (4)
s

and
p wD = sq D p sD ..........................................(5)
Combining Eqs. 4 and 5 we obtain
1

p wD = s sC D p wD p sD ......................(6)
s

To solve Eq. 6, we assume psD to behave as


combination of linear functions. This means that
psD is divided into small sections. Each section is
modeled by a linear function, which is:
p sD = at D + b ............................................(7)
where:
dp
a = sD
dt D

dp sD
dt D
Transforming into Laplace domain, Eq (7)
becomes
a b
............................................(8)
p sD = 2 +
s
s
Substituting Eq. 8 into Eq. 6 and rearranging we
obtain
p wD = p sD aC D p wD bsC D p wD ..........(9)
Inverting Eq. 9 into real domain, we get
dp wD
p wD = p sD aC D p wD bsC D
....(10)
dt D
We now substitute a and b into Eq. 10 to obtain
C dp sD
p wD = p sD D t D
p wD
tD
dt D
b = p sD t D

dp sD dp wD
CD
t D
p sD t D
tD
dt D
dt D
...................................................(11)
Defining well test derivative as y=[ x(dy/dx) ],
Eq. 11 can be written as
C
'
p wD
p wD = p sD D p sD
tD

CD
'
'
p sD p sD
p wD
tD
...................................................(12)
Since our goal is to extract reservoir response out
of wellbore storage distorted data and to analyze
it using semilog plot (straight linefor radial
flow) and log-log plot (horizontal linefor
radial flow), that is psD=1/2. Hence, Eq. 12
reduces to
C
p wD = p sD D p wD
2t D

C
CD
'
'
p sD p wD
+ D p wD
2t D
tD
...................................................(13)
Rewriting Eq. 13 in term of psD, we obtain

CD
'
p wD p wD
2t D
p sD =
+ Err (t D )
C
'
1 D p wD
tD
...................................................(14)
In Eq. 14, the unknown variable, psD, in the lefthand-side is isolated from the known variable
(recorded pressure data) in the right-hand-side.
Eq. 14 is our deconvolution equation and it is a
basis for analyzing wellbore storage distorted
data. Notice that, we add error term in the righthand-side to represent the inaccuracy due to
assumption during mathematical process. For a
vertical well, the comparison of deconvolution
equation to the analytical solution of the
constant-rate system is shown in Fig. 1. It shows
that the error term can be correlated using the
following simple formula:
p wD +

p p'
wD
Err (t D ) = 1.021 wD
tD

C p wD
D

p p'
wD
- 0.00645 wD
+ 4.1512 10 7
tD

C p wD
D

...................................................(15)

We choose to simplify and shorten Eq. 15 into a


more compact formulation, which is
2

p p'
wD
wD
................................(16)
Err (t D ) =
tD p
wD
C
D

And therefore, substituting Eq. 16 into Eq. 14 we


obtain our deconvolution equation.

p sD =

CD
'
p wD p wD
2t D
C
'
1 D p wD
tD

p wD +

p p'
wD
+ wD
tD

C p wD
D

...................................................(17)

Eq. 17 is validated for various value of CDe2s


ranging from 103 to 10100. The comparison

between analytical and deconvolution equations


shows good agreement (Fig. 2). To use Eq. 17,
we must convert it into a dimensional form. The
definition of the dimensionless pressure and time
variables are:
0.894c s
.........................................(18)
CD =
hc t rw2
Pressure Drawdown
kh
p sD =
p i p wf , dc ..............(19)
141.2qBi i

p wD =

tD =

kh
p i p wf
141.2qBi i

0.0002637kt

c t rw2

] ................(20)

...................................(21)

Pressure Buildup
p sD =
p wD =

tD =

kh p ws , dc p wf ( t = 0)

141.2qBi i

..............(22)

kh
p ws p wf ( t = 0) .(23)
141.2qBi i

0.0002637kt

c t rw2

.................................(24)

Substituting Eqs. 18, 19, 20 and 21 for pressure


drawdown case and Eqs. 18, 22, 23 and 24 for
pressure buildup case into Eq. 17 we obtain
Pressure Drawdown
'

p
p wf , dc = p i

'
1 p /[m wbs t ]

70.6qBi i
kh

p p '

'
m wbs t p

p p '

'
m wbs t p
...................................................(25)

141.2qBi i
kh

Pressure Buildup
'

p
p ws , dc = p wf ( t = 0) +

'
1 p /[m wbs t ]
70.6qBi i p p '
+

'
kh
m wbs t p

141.2qBi i p p '
+

'
kh
m wbs t p
...................................................(26)

III. ANALYSIS PROCEDURE


The deconvolution equation in Eq. 25 or Eq. 26
contains permeability, k, in the right-hand-side of
the equation and this parameter is unknown.
Therefore, procedure to use the deconvolution
equation requires an iteration process. The
procedure is as follows.
1. Plot pwf(t=0)+ p/[1-p/[mwbst]] versus
log(t) or (log of Horner time) and calculate
slope of straight line, mj, and permeability, kj.
162.6qBi i
kj =
.....................................(27)
mi h
2. Calculate pws,dc using Eq. 26.
3. Plot pws,dc versus log(t) (or log of Horner
time) and calculate slope of straight line, mj+1,
and permeability, kj+1.
4. If |kj+1-kj|< then calculate skin factor:
p ws ,dc p wf ( t = 0)
s = 1.151
m j +1

+
3
.
23
log

c t rw
...................................................(28)
5. Otherwise, repeat step 2.

IV. EXAMPLES
In this section we provide two examples of using
the deconvolution equation for analyzing
wellbore storage distorted data. We choose
examples from pressure buildup tests, since this
is where use of deconvolution equation is the
most relevant. The first example is generated
from a numerical reservoir simulator, and the
second example is a field case taken from
Bourdet paper.19
Example 1:
Table 1Data for Example 1
Initial reservoir pressure, pi
Reservoir thickness, h
Oil FVF, Bo
Oil viscosity, o
Oil rate, qo
pwf(t=0)
Producing time, tp
Permeability, k
Well radius, rw
Skin factor, s
Wellbore storage coeff., cs
Porosity,
Total compressibility, ct

: 5000 psia
: 50 ft
: 1.2 bbl/STB
: 0.8 cp
: 50 STB/D
: 4899.3 psia
: 1000 hrs
: 20 mD
: 0.3 ft
: +5
: 0.1 bbl/psi
: 0.15
: 3x10-6 psi-1

: 489

Fig. 3 shows semilog plot (MDH plot) of shut-in


pressure versus shut-in time. The pressure
responses distorted by wellbore storage is
denoted by solid circles (results from numerical
simulator). The open circles denote the pressure
deconvolution computed using procedure
outlined in section III.

It can be seen from this figure that after


deconvolution the radial flow response (semilog
straight line) can be extracted from the wellbore
storage distorted data for the most part of the
early time (t<30 hrs). Exception from this
occurs at very early time (t<0.2 hr). This may
be caused by the assumption used in the
development of the method.
Horner plot in Fig. 4 shows gives a better results.
This is true because Horner plot include
producing time (production history), while MDH
plot does not.
From these exercises, we say that use of the
deconvolution method may extend the straightline up to about 2 log-cycle longer. The results
from analysis of the deconvolved data agree with
the input data of the numerical simulator (same
permeability of 20 mD and skin factor of 4.97
compared to 5).
Example 2:
In this example, we use field case taken from
Bourdets paper. The results of using
deconvolution are presented in the Agarwal plot
shown in Fig. 5. The deconvolved data are
denoted by open circles.

V. DISCUSSION AND CONCLUSIONS


The deconvolution method is advantageous in
the pressure buildup test analysis if the test does
not show radial flow signature. However, the
availability of deconvolution method such as
presented in this study is not to propose a new
well test design replacing current method. As
much as possible the test should be run long
enough until the radial flow is attained. However,
short test may be performed and the data can be
analyzed using the deconvolution method for
economic or practical reasons.
From this study we conclude:
1. A new deconvolution method has been
developed which require only pressure data.
Hence this method is compatible with the
conventional well test which require only
measurement of the bottom hole pressure.
2. The method has been validated using
analytical and numerical solutions. Its
application to field data has also been
demonstrated.
3. The use of deconvolution method can extend
the semi-log straight line longer, which provide
an aid to the analysis.
4. We also observed that this method may not
work for entire early-time (wellbore storage
distorted) data. At worst, the quality of the
deconvolved data may only be good for a half to
one log-cycle.

NOMENCLATURE

Table 2Data for Example 2


Reservoir thickness, h
Oil FVF, Bo
Oil viscosity, o
Oil rate, qo
Producing time, tp
Well radius, rw
Porosity,
Total compressibility, ct

: 107 ft
: 1.06 bbl/STB
: 2.5 cp
: 174 STB/D
: 15.33 hrs
: 0.29 ft
: 0.25
: 4.2x10-6 psi-1

Quality of the deconvolved data for this case is


not as good as for the simulated case. At very
early time, data are very scattered. This may be
caused by this method is possibly very sensitive
to the data noise or to any violation to the model
assumptions (such as, constant wellbore storage
assumption). Nevertheless, we still able extract
the radial flow response and extend the semi-log
straight line for about 1 log-cycle longer. This
certainly provides an aid to the analysis. From
this analysis we obtain permeability of 10.78 mD
and skin factor of 7.62.

psD
psD
qD
q
t
tp
t
k
h
B

ct
pwf
pwf,dc
pws
pws,dc
cs
s

= dimesionless pressure for constant-rate


system
= dimensionless pressure
= dimensionless flowrate
= flowrate, STB/D
: 4899.3 psia
= time
= producing time
= shut-in time
= reservoir permeability, mD
= net pay thickness, ft
= formation volume factor, bbl/STB
= viscosity, cp
= porosity
= total compressibility, 1/psia
= well flowing pressure, psia
= deconvolved well flowing pressure, psia
= well shut-in pressure, psia
= deconvolved well shut-in pressure, psia
= wellbore storage coefficient, bbl/psi
= skin factor

CD
rw
pi

= dimensionless wellbore storage


coefficient
= well radius, ft
= initial pressure, psia

REFERENCES
1. Gladfelter, R.E., Tracy, G.W. and Wilsey,
L.E.:Selecting Wells Which Will
Respond
to
Production-Stimulation
Treatment, Drill. And Prod. Prac., API
(1955) 117-129.
2. Kuchuk, F.J.:Gladfelter Deconvolution, SPE
16377; Proceeding of the 1987 SPE
California Regional Meeting, Ventura,
California, April 8-10, 1987.
3. Winestock, A.G. and Colpitts, G.P.:Advances
in Estimating Gas Well deliverability, J.
Cadn. Pet. Tech. (July-Sept. 1965) 111119.
4. Ramey, H.J., Jr.:Non-Darcy Flow and
Wellbore Storage Effects in Pressure
Buildup and Drawdown of Gas Wells, J.
Pet. Tech. (Feb. 1965) 223-233; Trans.
AIME, 234.
5. Ramey, H.J., Jr.:Verification of the
Gladfelter-Tracy-Wilsey Concept for
Wellbore Storage-Dominated Transient
Pressures During Production, J. Cadn.
Pet. Tech. (April-June 1976) 84-85.
6. Fetkovich, M.J. and Vienot, M.E.:Rate
Normalization of Buildup Pressure Using
Afterflow Data, J. Pet. Tech. (Dec. 1984)
2211-24.
7. Kuchuk, F. and Ayestaran, L.:Analysis of
Simultaneously Measured Pressure and
Sandface Flow Rate in Transient Well
Testing, J. Pet. Tech. (Feb. 1985) 323-34.
8. Kuchuk, F. and Ayestaran, L.:Authors Reply
to
Discussion
of
Analysis
of
Simultaneously Measured Pressure and
Sandface Flow Rate in Transient Well
Testing, J. Pet. Tech. (Oct. 1985) 186769.
9. Thompson, L., Jones, J.R., Reynolds, A., and
Raghavan, R.:Analysis of Pressure
Buildup Data Influenced by Wellbore
Phase Redistribution, SPE 12782;
Proceeding of the 1984 SPE California
Regional Meeting, Long Beach, CA, April
11-13.
10. Thompson, L. and Reynolds, A.:Analysis of
Variable-Rate Well-Test Pressure Data
Using Duhamels Principle, SPE 13080;
Proceeding of the 1984 SPE California
Regional Meeting, Long Beach, CA, April
11-13.

11. Meunier, D. and Wittmann, M.J.: A Simple


Approach to Sandface Rate-Convolution
and Its Application to Early-Time Well
Test
Analysis,
SWS
Reservoir
Engineering Coordination and Technical
Meeting, Houston, Texas, March, 1985.
12. Stewart, G. Wittmann, M.J. and Meunier,
D.:Afterflow
Measurement
and
Deconvolution in Well Test Analysis,
SPE 12174; Proceeding of the 58th
Annual Technical Conference and
Exhibition, San Francisco, CA, October 58, 1983.
13. Mendez, L.C.C., Tygel, M. and Correa,
A.C.F.:A Deconvolution Algorithm for
Analysis of Variable-Rate Well Test
Pressure Data, SPE 19815; Proceeding of
the 64th Annual Technical Conference
and Exhibition of the Society of Petroleum
Engineers, San Antonio, TX, October 811, 1989.
14. Johnston, J.L. and Lee, W.J.:Interpreting
Short-Term Buildup Tests From LowProductivity
Gas
Wells
Using
Deconvolution, Paper SPE 21503,
Proceeding of the SPE Gas Technology
Symposium, Houston, Texas, January 2325, 1991.
15. Fair, P.S. and Simmons, J.F.:Novel Well
Testing
Applications
of
Laplace
Transform Deconvolution, Paper SPE
24716, Proceeding of the 67th Annual
Technical Conference and Exhibition of
the Society of Petroleum Engineers,
Washington, DC, October 4-7, 1992.
16. Baygun, B., Kuchuk, F.J. and Arikan,
O.:Deconvolution Under Normalized
Autocorrelation
Constraints,
SPE
Journal, V.2 (September 1997).
17. von Schroeter, T., Hollaender, F. and
Gringarten, A.C.:Deconvolution of Well
Test Data as a Nonlinear Total Least
Squares Problem, Paper SPE 71574,
Proceeding of the 2001 SPE Annual
Technical Conference and Exhibition,
New orleans, LA, 30 September 3
October, 2001.
18. Fair, W.B.:Pressure Buildup Analysis with
Wellbore Phase Redistribution, SPEJ
(April 1981), p. 259-270.
19. Bourdet, D., Ayoub, J.A. and Pirard,
Y.M.:Use of Pressure Derivative in WellTest Interpretation, SPEFE (June 1989),
p. 293-302.

Figure 1. Error analysis.

Figure 2. Semilog plot of dimensionless pressure versus dimensionless time functioncomparison


between Eq.15 and exact solution for various CDe2s.

Figure 3. simulated data-MDH plot.

Figure 4. simulated data-Horner plot.

Figure 5. Bourdet data exampleAgarwal plot.

You might also like