Vous êtes sur la page 1sur 27

Journal of Petroleum Science and Engineering, 1 (1988) 195-221 195

Elsevier Science Publishers B.V., Amsterdam - - Printed in The Netherlands

WELL TEST INTERPRETATION FOR RESERVOIRS WITH A


SINGLE LINEAR NO-FLOW BARRIER

F I K R I J. K U C H U K a n d S H A H K A B I R

Schlumberger-Doll Research, Old Quarry Road, Ridgefield, CT 06877-4108 (U.S.A.)


Schlumberger Well Services, P.O. Box 2175, Houston, TX 77252-2175 (U.S.A.)

(Accepted for publication November 20, 1987)

Abstract

Kuchuk, F.J. and Kabir, S., 1988. Well test interpretation for reservoirs with a single linear no-flow barrier. J. Pet.
Sci. Eng., 1: 195-221.

Sealing faults, which cut formations with or without contrasting rock properties, have a profound impact on tran-
sient pressure behavior of nearby wells. New analytical solutions with wellbore storage and skin effects are presented
for the interpretation of the wellbore pressure with and without wellbore flow rate (measured simultaneously with
the pressure) for a well located at a distance from a single fault. In the absence of measured downhole flow rate, both
constant wellbore storage and exponential decline flow rate models are considered in this work. These new solutions
offer a potential for the estimation of permeability, skin, and wellbore storage coefficient. Additionally, the well-to-
fault distance can also be estimated.
Application of the new solutions, using the logarithmic convolution, deconvolution, and nonlinear least-squares
estimation methods, is demonstrated through examples.

Introduction t e s t can be generally c h a r a c t e r i z e d into t h r e e


periods. T h e s e periods are:
In t r a n s i e n t testing, m a n y factors often cause (1) E a r l y - t i m e period: t h e c o m b i n e d effects
the wellbore p r e s s u r e b e h a v i o r to differ f r o m of wellbore storage, d a m a g e skin, p e r f o r a t i o n ,
t h e i n f i n i t e - a c t i n g c o n s t a n t - r a t e behavior. F o r p a r t i a l p e n e t r a t i o n , acidized zone, fractures,
example, t h e storage c a p a c i t y of t h e wellbore, n o n - D a r c y flow, etc., d o m i n a t e t h e t r a n s i e n t
wellbore g e o m e t r y , a n d n e a r - w e l l b o r e com- p r e s s u r e b e h a v i o r o f t h e well.
plexities affect the e a r l y - t i m e b e h a v i o r of the (2) M i d d l e - t i m e period: d u r i n g this period,
well, while e x t e r n a l b o u n d a r i e s such as gas cap radial flow is established. C o n v e n t i o n a l l y ,
a n d faults affect t h e l a t e - t i m e t r a n s i e n t b e h a v - semilog t e c h n i q u e s are u s e d to d e t e r m i n e for-
ior. D u r i n g the analysis of p r e s s u r e - t i m e data, m a t i o n p a r a m e t e r s including average r e s e r v o i r
each p h e n o m e n o n a n d its d u r a t i o n m u s t be p r e s s u r e f r o m t h e m e a s u r e d wellbore pressure.
recognized for t h e i d e n t i f i c a t i o n o f t h e well- (3) L a t e - t i m e period: d u r i n g this period,
reservoir model so t h a t its p a r a m e t e r s can be o u t e r b o u n d a r y effects s t a r t to d i s t o r t t h e well-
e s t i m a t e d . T h e influence o f t h e above p h e n o m - bore p r e s s u r e b e h a v i o r f r o m t h e i n f i n i t e - a c t i n g
e n a on the t r a n s i e n t b e h a v i o r of a well pro- c o n s t a n t rate behavior.
gresses with time. T h e d u r a t i o n o f a t r a n s i e n t S o m e t i m e s t h e s e p a r a t i o n of t h e s e periods

0920-4105/88/$03.50 © 1988 Elsevier Science Publishers B.V.


196

from each other is impossible. For example, the lougher and Kazemi (1980) gave criteria for
effect of a sealing fault, which is the subject of producing time necessary for the development
this study, may start during the early-time pe- of the doubling of slope on the Horner graph.
riod if it is in close proximity of the well. However, Gray (1965) and Martinez and
Well test analysis for a well adjacent to a no- Cinco-Ley (1985) suggested that the distance
flow barrier or a fault presents a significant to a fault can be estimated even when the sec-
interpretation challenge. The interpretation ond semilog line does not develop on the Hor-
difficulty stems from the fact that the wellbore ner graph due to short producing time.
storage and skin effects may dominate wellbore Reservoir anisotropy is also an important
pressure behavior during which the effects of a factor for the detection of faults as indicated by
no-flow barrier or fault are felt at the wellbore. Overpeck and Holden (1970). They showed
Consequently, the conventional semilog anal- that the distance to the fault could be in error
ysis is difficult to perform. Because of the dom- by as much as 20% if anisotropy is not ac-
inating influence of the inner- and outer- counted for.
boundary conditions, an appropriate interpre- Solutions have also been developed fbr par-
tation model is necessary for estimating reser- tially communicating fault systems (Stewart et
voir parameters. al., 1984; Yaxley, 1985) in which some degree
Since the pioneering work of Horner (1951), of lateral continuity for fluid flow is retained.
wellbore pressure behavior in the presence of a The pressure derivative technique has been
single fault has been the subject of numerous used recently by Clark and van Golf-Racht
studies; most of them, reported between 1951 (1985), and by Proano and Lilley (1986) to
and 1980, are given in the reference list. Some solve field examples involving single and mul-
authors examined pressure behavior in systems tiple faults. The interpretation of simultane-
where faults intersect at an angle (Jones, 1962; ously measured wellbore pressure and flow rate
van Poollen, 1965; Prasad; 1975) and others, in was presented for reservoirs with a single no-
the presence of parallel faults (Matthews and flow barrier by Kabir et al. (1986).
Russell, 1967; Ramey et al., 1973; Tiab and Ku- In most of the works cited above, it is com-
mar, 1980). Recently, models have also been monly assumed that the infinite acting period
proposed for locating reservoir pinchout is achieved before the effect of the no-flow bar-
boundaries (Home and Temeng, 1982; Gerard rier is felt at the wellbore and after the effect of
and Horne, 1985 ). the wellbore storage diminishes. In many cases,
Horner (1951) first reported the doubling of the early-time data that are distorted by well-
slope on a pressure buildup semilog graph in bore storage and skin effects, may also contain
the presence of an adjacent no-flow barrier. information about the no-flow barrier. In other
Russell (1963) and Pinson (1972) have shown words, the storage-dominated early period and
the existence of similar characteristic features the outer-boundary-dominated late-time pe-
during two-rate flow testing. A number of other riod may cause either the middle-time period
studies (Dolan et al., 1957; Bixel et al., 1963; to disappear or cause too severe a distortion to
Davis and Hawkins, 1963; Standing, 1964; Gray, allow a conventional semilog analysis. In such
1965; Earlougher and Kazemi, 1980) have pre- a situation, conventional methods are not suit-
sented methods for the detection of no-flow able for transient analysis. The type curve
boundaries from pressure buildup tests. analysis is only applicable if the wellbore stor-
The time of intersection of the two semilog age remains constant during the test, and it
straight lines has also been used to obtain in- cannot be used as an identification (diagnos-
formation about the distance from the well to tic) tool if the wellbore storage dominates the
the fault. In particular, Gray {1965) and Ear- effect of the no-flow barrier.
197

/ x pe ( rD ,tD ) = -- ~ Ei \ 4to]
/ \ LINEAR BARRIER ( FAULT}

WELL (i)
Jl _. I I' ~x
RESERVOIR
', , | /
BOUNOARY
\x ,,,"r ~ 1 ii and its Laplace transform is:

fiB (S)= -1 [Ko(rD~s)"t-Ko(dex/~)] (2)


S
Fig. 1. Schematic representation of a well with a no-flow
barrier in a n infinite reservoir. where rD= r/rw, dD=2d/rw, r=radial distance
from the wellbore axis, and d=distance from
The interpretation methods presented in this the wellbore axis to the no-flow linear barrier.
paper are an extension of the work done by Ka- The basic solution given by eqs. 1 or 2 will be
bir et al. (1986). New formulas are presented a fundamental building block for developing
for the determination of the no-flow barrier interpretation models for the analysis of well-
distance from the semilog, derivative, and con- bore pressure, with or without downhole flow
volution analyses. Analytical models are pre- rate, for a well adjacent to a no-flow barrier.
sented incorporating the constant wellbore The Laplace transform of PD given by eq. 2 is
storage, exponential flow rate decline, and used for the computation of wellbore pressure
measured downhole flow rate cases. A few well with the wellbore storage and skin effects.
test examples are given to explore the applica- For a constant producing rate in the absence
tion of the interpretation methods presented of wellbore storage effect, the interpretation of
here. transient data using eq. 1 is straightforward, as
shown by H o m e r {1951) and Earlougher
Mathematical development {1977). However, the wellbore storage may
partially or entirely mask the effect of the no-
In this section, the basic solution (influence flow barrier on the pressure response of a well.
or unit response function) for the constant rate Conceivably, the characteristic behavior of a
case without wellbore storage and skin effects reservoir with a no-flow linear barrier could be
is presented first. This solution will then be in- distorted if the distance, d, is short and/or the
corporated with different wellbore flow (stor- storage duration is long.
age) models or with measured downhole flow The effect of wellbore storage could be easily
rate. eliminated or reduced to a large degree if the
A fully penetrated well in a horizontal and downhole flow rate is measured and analyzed
infinite homogeneous reservoir containing a together with the bottomhole pressure. Prac-
slightly compressible fluid with a constant vis- ticality may, however, limit downhole flow rate
cosity is considered here. The fault cuts the in- measurement for every well. Thus, we consider
finite reservoir vertically at a distance (d) from both cases here: transient analysis of only pres-
the wellbore as shown in Fig. 1. sure or of pressure and flow rate data. As stated
The dimensionless pressure (influence or above, the wellbore flow rate behavior will be
unit response function) for the well described approximated by either the exponential or con-
above, producing at a constant rate without the stant wellbore storage flow model if it is not
effects of wellbore storage and skin, is given by measured. The constant weUbore storage model
Horner (1951) as: has been commonly used to develop type-curves
198

for different well reservoir systems for the last Fig. 2 presents a log-log plot of the dimen-
two decades. sionless wellbore pressure, PwD, and its deriva-
Given the constant rate behavior (influence tive with respect to In tD for d= 100 ft, CD ----100,
function) of the well-reservoir system, the re- and S= 10. Solutions given in Fig. 2 are ob-
lationship between downhole flow rate (input) tained by combining eqs. 5 and 2 and using the
and pressure (output) signals at the wellbore Stehfest (1970) numerical Laplace inversion
can be described as a convolution operation: method. Also presented in Fig. 2 are the solu-
tl)
tions without the effect of a no-flow barrier, but
/I
with the same wellbore storage and skin effects.
PwD (tl)) = ~ qI~ (r)PsD (tD - r ) d r As can be seen from Fig. 2, the difference in
0
the pressures with and without the effect of the
tl)
f. no-flow barrier is minimal. The derivative does
= | qD (r) [p~)(tD --r) + not show any distinguishing features of a res-
0 ervoir with a no-flow barrier. Thus, identifica-
tion of the no-flow barrier appears impossible
d(tD-r)S]dr (3) from pressure or its derivative when the well-
provided that qD ( 0+ ) is zero. bore storage and skin effects are dominant. In
In eq. 3, qD is the normalized downhole flow this situation, the presence of a no-flow barrier
rate, and PwD is the dimensionless downhole and its distance to the wellbore have to be
pressure, both of which are measured at the tool known independently for the use of semilog and
location. The dimensionless pressure, PD, for type curve matching methods.
the constant rate case is given by eq. 1 and The pressure or pressure derivative type
PsD=PD+S, where S is the van Everdingen- curves (from eq. 5 combined with eq. 2 ) can be
Hurst steady-state skin factor. developed for different distances to the fault.
The convolution integral given by eq. 3, which However, it is better to use nonlinear least-
is a subset of Volterra integral equations, is well squares estimation methods (automated type
known for providing techniques for solving curve matching) to analyze well test data be-
time-dependent boundary value problems. It is cause a set of type curves is needed for each
also known as the superposition theorem, and distance to the fault. A brief introduction to the
has played an important part in transient well nonlinear least-squares estimation methods will
test analysis. be given later.
The dimensionless wellbore pressure for the For the exponential wellbore flow rate case,
constant wellbore storage case, PwD, is given by the dimensionless wellbore pressure, PwD, is
van Everdingen and Hurst (1949) and Agarwal given by van Everdingen (1953) and Hurst
et al. (1970) as an integro-differential equation (1953) in terms of the convolution integral as:
of a convolution type, namely: tD

tD PwD (tD) = ~ (1 --e -/j~)


t)
PwD (tD) = 1 -- CD dr
0 [p~(tD--r)+d(tD-r)S]dde (6)
and its Laplace transform is:
[dpD(tD-r)~-d(tD-r)Sldr(4)dtD
fl/SsD (s)
i6wD (S) = - - (7)
and its Laplace transform is: fl+s
psD (s) Fig. 3 presents a log-log plot of the dimen-
15wD(s) - l +C.s2Ps D (s) (5) sionless wellbore pressure, PwD, and its deriva-
199

l0 t

X
N

A
A
A
A
&,

10 °
e-
C~ A

r~
-- P,~l) with no-flow houndac,

C D = lO00 ---- dP~D/dlnt D v*ith no-Ilow boundar?

S = 10 0 P~D without n,)-llo~, houndac.

d = lOOft /X P~D/dlntl) v, lthout no-Ilow houndar,.

1 0 -t
1 0 -t 10 ° l0 t 10 2 10 ~ 10 4 10 ~

ID/CD

Fig. 2. Dimensionless wellbore pressure, PwD, and its derivative for the constant wellbore storage case.

10 t

10 °
A

-- P~D with no-llow boundary


13 = 0.00007 ---- dP~D/dlnt D with no-flow boundary

S =10 O P~D without no-flow boundary

d = lOOft ~. P~o/dlnt D without no-flow boundary

10-'
1 0 -2 10-' 10 ° l0 t 10 2 l0 s 10 4

PtD
Fig. 3. D i m e n s i o n l e s s w e l l b o r e p r e s s u r e , PwD, a n d i t s d e r i v a t i v e for t h e e x p o n e n t i a l flow r a t e case.
200

rive with respect to In tD for d-- 100, fl= 0.00007, TABLE I


and S = 10. Like Fig. 2, solutions given in Fig.
Rock a n d fluid data
3 are obtained by combining eqs. 7 and 2, and
using the Stehfest ( 1970 ) algorithm. The same Porosity, 0 = 0.20
Permeability, k = 40.0 m D
figure also presents a log-log plot o f p w D and its V i s c o s i t y , # = 0.8 cp
derivative without the effect of a no-flow bar- Compressibility, ct= 10 -:' psi t
rier, but with the same wellbore storage and skin Well radius, rw=0.35 ft
effects. As can be seen from Fig. 3, the behavior F o r m a t i o n thickness, h = 100 ft
Initial pressure, p~ = 4400.0 psi
of pressure and its derivative for the exponen- P r o d u c t i o n rate, q = 3000 R B / D
tial wellbore flow rate case is similar to that of P r o d u c t i o n time, tp = 5000 h
the constant wellbore storage case. Skin, S = 10

Interpretation methods sure and flow rate data. This model is given by:

Interpretation of well test data from a well in dpwD ( tD )


qD (tD) = 1 -- CD (8)
a reservoir with a no-flow barrier or a fault pre- dtD
sents two problems that must be solved simul- However, the interpretation methods pre-
taneously. The first is the identification sented in-this paper are aimed at both the mea-
(diagnosis) of the system; that is, identifica- sured pressure and simultaneously measured
tion of the presence of a no-flow barrier or bar- flow rate and pressure.
riers without independent information.
Geological and geophysical methods may pro- Conventional
vide clues about the presence and location of
the tectonic boundaries. However, these sources In the conventional analysis, log-log pres-
do not usually reveal the nature of the no-flow sure and log-log pressure derivative plots are
barrier - - sealing, partially sealing. The second used as identification (diagnostic) tools. Log-
problem is the estimation of the distance of the log pressure and log-log pressure derivative type
no-flow barrier to the wellbore and other un- curve methods, either graphical (Clark and van
known reservoirs parameters, some of which are Golf-Racht, 1985; Proanov and Lilley, 1986) or
necessary for the estimation of the distance. automated (nonlinear least-squares estima-
Solutions of both problems are attempted tion, described later), can be used if the well-
using the conventional, logarithmic convolu- bore storage is constant. However, for various
tion, deconvolution, and the nonlinear least- reasons, the wellbore storage a n d / o r flow rate
squares estimation methods. Examples are (for drawdown tests) may not be constant for
given to address the identification and estima- the duration of a test. In those situations, the
tion problems. use of this model at early times must be done
Four synthetic well test examples are used with caution.
for the application of the interpretation meth- In general, the conventional methods depend
ods presented below. All examples have the on the presence of one or two semilog straight
same rock and formation properties and are lines. The first conventional method is that of
given in Table I with different distances to the Horner ( 1951 ), which depends on the presence
no-flow barrier: 5 ft for Example 1, 20 ft for of two semilog straight lines. The distance to
Example 2, 100 ft for Example 3, and 500 ft for the no-flow barrier from the Horner method is
Example 4. For convenience, the constant well- given by Earlougher and Kazemi (1980) as:
bore storage model with a C of 0.0274 bbl/psi is
d = 0.01217~/ktin/O/~ct (9)
used to generate the synthetic wellbore pres-
201

where tin is the time at which the two semilog radial flow with the effect of the no-flow barrier
straight lines intersect each other. and is not the end of the wellbore storage effect.
The skin from the second semilog straight can The wellbore storage effect should diminish be-
be obtained as: fore the start of the second semilog straight line,
however.
S = 2.3026 [_Aplh~--log ~ The estimation of the distance from eqs. 11
[_ m2 ~JllCtrw or 12 depends on a single time point. Thus,
pressure derivative plots are better suited for
+3.2275+log(2d/rw) 1 (10) the determination of tdv and tsl2 and their de-
termination from semilog plots is not recom-
where 3Plhr=Pws(att=O)-pw~(3t=l hr), and mended. Nevertheless, eqs. 11 and 12 provide
m2--the slope of the second Hornet semilog an alternative to eq. 9 for the estimation of the
straight line. distance to the no-flow barrier. If the distances
The second conventional method is the de- obtained from eqs. 9, 11, or 12 are different from
superposition given by Gray (1965) and Mar- each other, the reservoir model and the first and
tinez and Cinco-Ley (1985), which depends on second semilog straight lines should be ques-
the existence of a first semilog straight line and tioned. If applicable, the distance obtained from
a transition period. The onset of the deviation either eq. 9, 11, or 12 can be used as an initial
of the pressure or pressure derivative from the guess for the nonlinear parameter estimation.
first straight line can also be used to estimate Example 1:d=5 ft. Fig. 4 presents a syn-
the distance as: thetic buildup pressure and afterflow rate data
after a constant-rate flow period of 5000 h for
d=/0.0002637 ktav In (0.86859ml) (111
Example 1. As can be seen from this figure, the
measurable afterflow period is about one hour.
Fig. 5 presents a log-log plot of the wellbore
where Q,-- the time at which the pressure or its pressure and derivative of the wellbore pres-
derivative deviates from the first semilog sure with respect to In t (for convenience, t in-
straight line, and ml--the slope of the first stead of At is used throughout this paper).
semilog straight line. Usually, the derivative of the wellbore pressure
In the desuperposition method, type-curve with respect to In ( tp + t ) should be used to min-
matching is not needed to obtain the distance imize the effect of the producing time, tp. How-
ifeq. 11 is used. Eq. 11 is also useful if the sec- ever, the producing time for the buildup
ond semilog straight line does not develop. examples is long enough so that its effect can
The method given by Earlougher and Ka- be neglected.
zemi (1980) for estimating the no-flow barrier As expected, Fig. 5 does not show the effect
distance to the wellbore can also be modified of a no-flow barrier; rather, it exhibits a typical
as: radial system behavior since its effect on the
d = 0.00431 x/kt~e/~)].~ct ( 12 ) wellbore pressure and its derivative is masked
by the wellbore storage and skin effects. Thus,
where t~2 is the start of the second semilog the presence of the no-flow barrier and its dis-
straight line which can be obtained accurately tance to the wellbore have to be known for the
from the pressure derivative plot. use of semilog and type curve matching meth-
To apply eq. 12, the wellbore storage effect ods. Otherwise, the test would have been ana-
should be negligible at the start of the second lyzed as a radial system without a no-flow
semilog straight line. In other words, t~2 is the barrier as shown in Fig. 6. The Horner plot, Fig.
start of the second semilog straight line due to 6, exhibits a semilog straight line which yields
202

4400 . . . . . . . . 3750

i
4000 - ............... ~.~ 3000
~++++~+
++%

++÷
3600 2250
m

4 *

÷
3200-
+
- 1500
n-
,2 {
¢. O
+
l,
2800 - 750

2400 . . . . . . .A~~.... -o

•~ Pressure 1

2000 17 10"=
. . . . . . . .

10'
. . . . . . . .

10"
. . . . . . . . {

10=
+ Spinner {

10'
Time, hrs
Fig. 4. Wellbore pressure and afterflow rate for Example 1 (d=5 ft).

10 ~

f _ % % ~

10 ~
<1
¢-
c$

i0'
.fl$ O Measured pressure
, ~ Computed pressure
+ Measured pressure derivative
°° Computed pressure derivative

0 -3 10 2 10-, 10 o 10 ~
Shut-in time, hr

Fig. 5. Comparison of measured and computed pressures and their derivatives for Example 1 ( d - - 5 ft).
203

4200-
16~67 10.'99 7.259 4.795 3.167 2.()94 1.3'84 0.9152 0.6()51
Real time, hr

4000-

O.

d 3800 . . . . . . . . . . . . . . . . . . aAA~Aaa
(n
fD

Q.

c AA
3600 . . . . .... ~ _ 4 ...........
A
c-
u) AA

Slope = -206.50 p s i / c y c l e ",


3400 kh = 1889.80 md-ft t . . . . . . . . . . . . AAA
k = 18.90 md ",
S = 3.09
p. = 4422.37 psi

3500
L
10' 10'
Horner time, (At+tp)/At
Fig. 6. Homer plot for Example 1 (d=5 ft).

k=18.9 mD, S=3.1 and p*=4422.4 psi. The Example 3: d= 100 ft. Figs. 9 and 10 present
permeability and skin calculated from the Hor- a log-log pressure, log-log pressure derivative,
ner semilog straight line are incorrect. These and the Horner plots for Example 3. What is
results can be modified to obtain the correct said about Examples 1 and 2 is also true for this
values of the parameters if the of the fault is example.
known. The desuperposition method and Eqs. Example 4:d=500 ft. Fig. 11 presents log-
9, 11, and 12 are not applicable to this test be- log pressure and pressure derivative plots for
cause the first semilog straight line is not pres- Example 4. Ascan be seen from this figure, the
ent due to the wellbore storage effect. log-log plot clearly shows the effect of a no-flow
Example 2: d= 20 ft. Fig. 7 presents pressure barrier. The log-log pressure derivative indi-
and its derivative with respect to In t on a log- cates an upward trend after a dip. This upward
log graph for Example 2. As in Example 1, Fig. trend of the pressure derivative curve can be
7 does not show the effect of a no-flow barrier. seen in Fig. 12 (the enlarged plot of Fig. 11 ). It
Here, the fault is located 20 ft away from the is clear from Fig. 11 that the effect of the no-
wellbore. The Horner plot for Example 2 is not flow barrier becomes a dominant factor follow-
much different than the Horner plot for Ex- ing the dissipation of the wellbore storage ef-
ample 1. As shown in Fig. 8, the semilog straight fect, lasting around 10 h. The remaining part
line yields k = 18.7 mD, S-- 2.3, and p* -- 4428.7 of the test is the transition period between the
psi. Theoretically, both semilog straight lines first and second straight line. Thus, in a real
in Figs. 6 and 8 should yield the same results; test, identification may become a problem if we
however, they are not drawn spanning the same do not know the presence of a fault. In other
time period. words, it will be difficult to obtain a smooth
204

10 3 -

\
102 -

<1

¢..

e-,

/'

0 Measured pressure
--Computed pressure
+ Measured p r e s s u r e derivative
=-Computed p r e s s u r e derivative
I0 ° ........ I ........ I ........ I ........ I
I 0 -~ I 0 -~ i 0 -l I0 ° 101
Shut-in time, hr
Fig. 7. C o m p a r i s o n of m e a s u r e d a n d c o m p u t e d pressures a n d t h e i r derivatives for E x a m p l e 2 (d = 20 ft).

4200
10.99 7.259 4.795 3.167 2.094 1.384 0.9152 0.6051
Real time, hr

4000 -

a~ 38oo

8Q .
.~.
3600
i-
>,

Slope = -209.24 psi/cycle


3400 kh = 1865.07 m d - f t t - " ~'
k ffi 18.65 m d [
S = 2.33
P" = 4 4 2 8 , 6 7 Dsl
J
3200
102 10s

Horner time, (At+tp)/At


Fig. 8. H o r n e r plot for E x a m p l e 2 ( d = 2 0 ft).
205

10 3 -

a.
,,=,,)
/ ,.
=
•.~ 102 - / ~%x,+
.....................
~ ......................

e~

~ 10 I -

r a Computed pressure
+ Measured p r e s s u r e derivative
- ° C o m p u t e d p r e s s u r e derivative

10° ...... I ........ I ........ I ........ I ........ I


10 3 I0 ~ 10 ' 10 ° 10'
Shut-in time, hr
Fig. 9. C o m p a r i s o n of m e a s u r e d a n d c o m p u t e d p r e s s u r e s a n d t h e i r derivatives for E x a m p l e 3 ( d = 100 ft).

4200-
16:67 - ~ . . . . . 7. . 2T5 9
10.99 4 . 7r9 5 ,
3.167 2 . 0,9 4 r . . . . . . . 0.9152
1.384 .~ - 0.6()51
R e a l time, hr

4050-

rl 3900 -

d
W

3750-
Q.
A~kA ~
¢
&
A
JC A
o) 3600 -
A
&
Slope = -193.86 psi/cycle A
kh = 2012.98 md-ft &
3450- k = 20.13 m d
S = 2.23
p"
= 4 4 0 0 . 7 3 asl

3300-

10' 10'

Horner time, (At÷tp)/At

Fig. 10. H o r n e r p l o t for E x a m p l e 3 (d= 100 ft).


206

• i
1 0 3 --:

,.; 10 2 ~

.~ lO t
e-,

10 ° ~ Computed pressure
+ Measured pressure derivative
-° Computed pressure derivative

........ I ........ I ........ I ........ I ........ I


0 -( I0 ~ I 0 -2 10-' 10 ° lO t
Shut-in t i m e , hr
Fig. 11. C o m p a r i s o n o f m e a s u r e d a n d c o m p u t e d p r e s s u r e s a n d t h e i r d e r i v a t i v e s for E x a m p l e 4 ( d = 5 0 0 ft ).

100 0 The Horner plot, as shown in Fig. 13, ex-


90 hibits a semilog straight line which yields
80 )
k-- 35.9 mD, S=8.3 andp* =4244.5 psi. As can
be seen from Fig. 12, the Horner semilog
~. 7O straight line is drawn through the transition
period. Thus, permeability and skin are lower
than the true values. The extrapolated pressure
# Onset of the effect of the
' 'co' ' ',7 no-flow boundary is incorrect because it is obtained from the first
Ooo
50
c<o
,, ° ooo U-~
Horner semilog straight line. On the other hand,
°°Oocx-~. . . . . . ~ . ~ o e O ')
the extrapolated pressures from the second
40 . . . . . . . ]
semilog straight line for Examples 1, 2, and 3
I 00 are correct. However, for Example 4, the well-
~hul-in time, hr
bore storage effect slightly overlaps the no-flow
Fig. 12. S e m i l o g p l o t o f t h e d e r i v a t i v e o f t h e w e l l b o r e p r e s - barrier effect. As can be seen from Fig. 12, the
s u r e for E x a m p l e 4 ( d = 5 0 0 f t ) . slope has not doubled at 24 h; meaning the cor-
rect second semilog straight line did not de-
pressure derivative of real data to diagnose a velop at that time.
subtle effect of the no-flow barrier from either
Fig. 11 or 12. Nevertheless, permeability and Logarithmic convolution
the distance from Fig. 12 are calculated to be
k = 36.7 mD and d = 538 ft, using eq. 11. We note The logarithmic or sandface rate convolu-
that permeability from the pressure derivative tion method, which is similar to the conven-
plot is slightly more accurate than that of the tional multirate method, is well documented in
Horner plot which is discussed below. the literature (Fetkovich and Vienot, 1984;
207

4100
Real time, hr

4000-

CL 3900-
d

3800. "%
Q.
A
A

"A i
r-
u) 3700 -
A
b ' '
A
A

t Slope = -108.64 psi/cycle A

kh -- 3592.18 md-ft
3600- k = 35.92 md
S
p-
= 8.29
= 4244.52 psi

3500 I

10' 10'

H o m e r time, (~t+tp)/At
Fig. 13. Horner plot for Example 4 (d--500 ft).

Meunier et al., 1985, 1987; Kuchuk and Ayes- d2


taran, 1985). 0.000264~(t_ ~:)]dT} (14)
Substituting eq. 1 with skin in eq. 3 yields:
where:
tD

PwD(tD)= f q~(r) {1[


0
---~Ei 4(o 1,) J~i =/lPwf/qD and is called "reciprocal produc-
tivity index" or "rate normalized pressure"
(13) t
f, ct(t,qD) = [1/qD(t)] _t qD(~) l o g ( t - r ) d ~
Substituting the logarithmic approximation for 0
E i ( - 1/4tD) in eq. 13 when tD> 100, which is
attained for most systems after a few seconds if = logarithmic convolution time
the permeability is greater than 1 mD, and di-
viding by qD, yields (in oil-field units): rn = 162.6q/t / kh
O.4343 b = log (r//r~) - 3.2275 + 0.86859S
J~i (t) = m { [f~ct(t,qD) + b ]-
qD (t) If the well-to-fault distance, d, is small (a few
t wellbore radii from the wellbore), the argument
of the exponential integral in eq. 14 becomes
× I q~(~)Ei sufficiently small after a few seconds for rea-
0
sonably permeable systems. Thus, using the
208

logarithmic approximation for the exponential fault distance of 20 ft). The logarithmic con-
integral in eq. 14 yields: volution plot exhibits two straight lines (the
slope of the second straight line is slightly less
d~i (t) = 2m [/it t (t,qD) + b ' ] (15)
than the double of the first one ) from which the
where: permeability, skin, and the distance are ob-
tained as k = 4 0 . 6 mD, S = 1 0 . 2 and d = 1 6 . 3 ft.
b' = log (tl/r~) - 3.2275 + 0.4343S These estimated values are close to the actual
-log(2d/rw ) values. We point out that the conventional
methods did not work for this case because the
A linear plot of J ~ i vs. [lct(t,qD) should also effect of the no-flow barrier is masked by the
yield a straight line with a slope of 2m and an wellbore storage. The identification of the pres-
intercept of 2rob'. The permeability and skin ence of a no-flow barrier and estimation of its
can be obtained from the slope and intercept. distance to the wellbore, permeability, and skin
Although interpretation is straightforward for are possible from the logarithmic convolution
the case of a small but known d, the identifica- plot.
tion of a no-flow barrier from transient well- For large d (distance) and sufficiently small
bore pressure a n d / o r flow rate data may not be t (time), the argument of the exponential in-
possible from logarithmic convolution as in tegral in eq. 14 becomes large; the exponential
conventional methods. integral thus approaches zero. In this case, the
Fig. 14 presents a logarithmic convolution plot logarithmic convolution method can be used
of the normalized pressure as a function of the during the early time at which the no-flow bar-
logarithmic convolution time for Example 1 (a rier has a negligible effect. Thus, for a radial
fault distance of 5 ft). After a short time, the system, a linear plot of Jmi vs./let (t,qD) should
plot exhibits a straight line. The permeability yield a straight line with a slope m and an in-
and skin obtained from the slope and intercept tercept rob. The permeability and skin can be
of the straight line are k = 4 1 . 2 mD and S = 10.5, obtained from the slope and intercept.
which are almost the same as actual values. As The logarithmic convolution method can also
in the conventional methods, without knowl- be used during the late time if the flow rate is
edge of the fault distance, the permeability and measured. For instance, flow rate can be mea-
skin would be calculated as k = 2 0 . 6 mD and sured during the late time of a drawdown test.
S=3.9. However, the flow rate approaches zero as pres-
If d (distance) is not very small and the test sure builds up; thus, the flow rate measurement
duration is sufficiently long, the logarithmic during the late time of a buildup test is often
convolution plot should yield two straight lines. impractical.
The first straight line is described by eq. 14 and Fig. 16 presents a logarithmic convolution plot
the second one by eq. 15. The distance to the of the normalized pressure as a function of the
fault can be obtained from the intersection of logarithmic convolution time for Example 3 (a
the two straight lines as: fault distance of 100 ft). The plot exhibits only
d = 0.01217 e 1'513 ¢r,ct,, x / ~ l x c ' (16) one straight line period after which the after-
flow becomes too small to be measured accu-
where (/let) i --- logarithmic convolution time at rately in practice. The straight line in Fig. 16
the intersection of the first and second straight yields k = 4 0 . 0 mD and S = 10.0, which are the
lines on the logarithmic convolution plot. same as actual values. Note that the normalized
Fig. 15 presents a logarithmic convolution plot pressure deviates from the straight line at about
of the normalized pressure as a function of the one hour. One can also use the deviation point,
logarithmic convolution time for Example 2 (a as given in eq. 12, to compute the distance.
209

1600
0.0008 o o~2e ooze3 o o~ss 0.0~70 02~5 05~19
Real time, hr

1500
/
o.

:3 1400 /
/
CO
¢0

1300

/
m
o
"O
"o
a)
N 1200

E
,= 1100
L A A ~
n- AA A
& / Slope -- 189.46 psi/cycle
1000 . " " kh = 2059.73 m d - f t _
/ /
k = 41.19 md
S = 3.57

900 I I I
-4.0 -3.5 -3.0 -2.5 -2.0 -t5 -1.0 -0.5 0.0

logarithmic convolution time, ftct


Fig. 14. Logarithmic convolution plot for Example 1 (d=5 ft).

Note that for these three examples, permea- fault distance of 500 ft). As in Example 3, the
bility and skin calculated from the Horner sem- logarithmic convolution plot also exhibits only
ilog straight line is much lower than those one straight line period after which the after-
obtained from the logarithmic convolution. For flow becomes too small to be measured. The
most flow systems, the late time Horner perme- straight line in Fig. 17 yields k=40.0 mD and
ability and skin must be equal to or greater than S = 10.0. However, the distance to the no-flow
the early time logarithmic convolution perme- barrier cannot be estimated from eq. 10, be-
ability and skin. Thus, the contrast can be taken cause as we stated earlier, the Horner semilog
as an indication of a no-flow barrier effect if straight line given in Fig. 13 is the first one. The
other possibilities are thoroughly examined. estimates from the logarithmic convolution
Further, the Horner permeability is one-half of method also support this argument because they
the logarithmic convolution permeability. This are close to those from the Horner method.
condition most likely will not occur in many A reasonably long-duration drawdown test
reservoirs, except for single no-flow barrier sys- instead of the buildup tests, as given in Exam-
tems. After estimation of permeability and skin ples 3 and 4, would be a better choice for elimi-
from logarithmic convolution, Fig. 16, the dis- nating any ambiguity arising from the low
tance is calculated from eq. 10 as d= 74 ft, which flowmeter response after approximately one
is 26% lower than the actual value since eq. 10 hour. The logarithmic convolution plot of the
is very sensitive to errors in S, APlhr, and ml. drawdown test would have exhibited two
Fig. 17 presents a logarithmic convolution plot straight lines from which the permeability, skin,
of the normalized pressure as a function of the and distance could have been calculated.
logarithmic convolution time for Example 4 (a Thus, logarithmic convolution of rate and
210

18oo~_ o.~5 ----T


0.0028 o.o'.7 0.0777 o.3~s 1.3'34 " ~
5.982 . . . . . . . .

1650 - --
,r,=

d
U)
1500

3.
¢1

"0 1350
"0
0

as
E
1200. --

=1
¢1::
1050 - - - - - ---

90O ...... T- T
-4.50 -3.75 -3.00 -2.25 -tS0 -0.75 0,00 0.75 t50

logarithmic convolution time, flct


Fig. 15. Logarithmic convolution plot for Example 2 (d=20 ft).

pressure data from a buildup test works for the wellbore. In other words, the near wellbore ef-
identification of no-flow barriers and the esti- fects must be resolved as a constant skin region,
mation of reservoir parameters, if the no-flow and the wellbore storage due to the volume be-
barrier is not too close to the wellbore. Since the low the tool must become negligible before the
wellbore storage effects are minimized by meas- onset of the radial flow on the logarithmic con-
uring the downhole flow rate, the very early time volution plot.
data, during which the effect of the no-flow bar-
rier is small, can be used for the estimation of
permeability and skin from the first straight line Deconvolution
on the logarithmic convolution graph. If the
distance is not too large, the plot may also ex-
hibit a second straight line from which the dis- Deconvolution is defined as determining di-
tance is obtained. If the distance is large, then mensionless pressure (or pressure for actual
the logarithmic convolution and the semilog well tests ), PSD, (given in eq. 3 ) at the wellbore
methods can be combined to obtain it. How- for the constant flow rate from simultaneously
ever, one should remember that the effects of measured downhole pressure and flow rate. In
the wellbore volume below the tool and of other other words, deconvolution involves comput-
near-wellbore complexities such as perfora- ing the pressure behavior of a well-reservoir
tions, acidized zone, etc., should be small so that system as though the well produces at a con-
very early time transient pressure-rate data can stant rate, from any arbitrary pressure and flow
be used to detect the no-flow barrier near the rate history. Unlike the conventional and con-
wellbore and to estimate its distance from the volution methods, deconvolution of wellbore
212

1650 - i - T l i • i 1
0.0005 0.0026 0.0147 0.0780 0.3530 1.323 5.949
Real time, hr

I
1500 -
0. w 1
o3
¢t) 1350 -
a)
P,
__=
¢,
10 12oo - r
I
"10 rt I
i
!

¢I i
I
J
E I
4
1050 - t 1
,=
.= J /
n"

t
J
Slope = 97.39 p s i / c y c l e /
900 - -
kh = 4007.13 m d - f t

75O
-4.50 -3.75 -3,00 -2.25 -1.50 -0.75
k
S
= 40.07 m d
= 10.03

I
0.00
i
0.75
f
1.50

logarithmic convolution time, fict


Fig. 17, Logarithmic convolution plot for Example 4 (d--500 ft).

contains measurement errors. For wellbore geometry and fluid type. In any event,
PSD [(tD)l ], the problem becomes worse be- the deconvolution methods mentioned above
cause incorrect PSD [ ( ~D ) 0 ] is used for the com- give unstable solutions if measurements have a
putation of the numerator in eq. 18, and the small a m o u n t of noise (errors) associated with
numerator has to be again divided by qD [ (tD) 1 ]. them.
The error in PSD (tD) will grow as ~D increases To circumvent the stability problem, Ku-
since eq. 17 is recursive. Thus, the procedure chuk et al. (1985) introduced the constrained
given by eq. 17 for deconvolution is an unstable deconvolution method. This method is used to
operation if the downhole flow rate contains compute deconvolved pressure for the exam-
some measurement noise (errors). ples given above. A brief discussion on con-
Kuchuk and Ayestaran (1985) also pre- strained deconvolution is given for the sake of
sented a curve-fit deconvolution for eq. 3. As thoroughness.
discussed by them, the success of the curve-fit Let us assume t h a t the downhole flow rate
deconvolution method depends on how well the contains measurement noise or error which can
approximation function represents the behav- be expressed as:
ior Of PsD.
qI) (tD) = q D (tD)-4-~(tD ) (19)
All measurements, no matter how carefully
obtained, are subject to some errors. At the where ql) is the measured flow rate and e (tD) is
wellbore, usually errors in pressure measure- the noise in the flow rate measurements. In the
ments are smaller t h a n errors in flowmeter constrained deconvolution, PSD is computed in
measurements. Sometimes errors can be large such a way t h a t it minimizes I]e(tD)]] subject
in flowmeter measurements depending upon to:
213

tD These estimated values are close to the ac-


tual values. Unlike the conventional and loga-
P w D ( t D ) = f OD(3) X rithmic convolution methods, identification of
0
the presence of a no-flow barrier and estima-
[P~D(tD --~) + 5(tD-r)S]dz (20) tion of its distance from the wellbore, permea-
and: bility, and skin are possible from the
deconvolved pressure. However, accurate
PSD (tD) >-- for tD > 0 (21) downhole flow rate measurements are essential
PSD (tD) >--- for tD > 0 (22) to have successful deconvolution during the
very early time.
PSD(D)
,r / t -~ for tD > 0 (23) As in Example 1, Fig. 19 for Example 2 also
P SD (tD) is monotone increasing for tD >--0 exhibits two semilog straight lines: the first one
yields k=38.0 mD and S=9.4. The distance
(24) from the intercept of the two straight lines us-
where the prime denotes time derivative of ing eq. 9 is calculated as d= 18.5 ft. The perme-
PSD ( tD ). ability and skin from the second straight line
The general solutions of the diffusivity equa- are k=40.0 mD and S=10.6. Note that both
tion for a second kind of internal boundary straight lines in Fig. 19 are well developed.
condition in homogeneous porous media with In contrast, the semilog plot for Example 3
nonperiodic initial and outer boundary condi- exhibits essentially one straight line as Fig. 20
tions will satisfy the conditions given in eqs. demonstrates. In this case the measurable af-
21-24 when the real time is greater than 1 s and terflow period is not long enough to allow the
permeability is not less than 1 mD. development of the second straight line. The
The constraints given by eqs. 21-24 were first permeability, skin, and the distance are ob-
introduced by Coates et al. (1964). Thus, de- tained as k--40.0 mD, S-- 10.0, and d--121 ft
convolution becomes a linear least-squares from the first semilog straight line.
problem subject to simple bounds on the vari- The point of deviation occurs at 0.5 h. The
ables. The active set method is used to solve the estimated distance is 21% higher than the ac-
above minimization problem for the well test tual value because eq. 11 is used with the semi-
examples. log pressure plot instead of the derivative plot.
Figs. 18, 19, 20, and 21 present a semilog plot The presence of a no-flow barrier is assumed to
of the deconvolved pressures as a function of occur at the deviation point to allow computa-
time for Examples 1, 2, 3, and 4, respectively. tion of the distance. The deviation from the first
As can be seen from Fig. 18, the semilog plot straight line, without the development of the
exhibits two straight lines. The first one with a second semilog straight line, is necessary but
very short duration yields k--32.1 mD and not sufficient for the identification of a no-flow
S = 7.6. The distance from the intercept of the barrier. This is also true if the desuperposition
two straight lines using eq. 9 is obtained as method is used.
d = 5.1 ft. The second straight line with a much As in the previous example, Example 4 also
longer duration (1 h) than that of the first one exhibits only one straight line (Fig. 21) and
yields k=40.6 mD and S--10.3 (from eq. 10). does not show the effect of a no-flow barrier.
Although the permeability values computed The permeability and skin obtained from the
from both straight lines are not very different, semilog line are k-- 40.0 mD and S = 10.0.
permeability computed from the second straight In general, the above results from deconvo-
line with a longer duration should be consid- lution show the possibility of identifying a no-
ered more reliable. flow barrier system if the distance is not too
214

1650

. . . . . . . . . . . . . . . r. . . . . . . . .
.m J /
Q.
1500 . . . . . . . . . . . . . . . . .

I]J
Q.

"0

_> 12oo . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12~

f't 1050 . . . . . . . . . . . . . . . . . . . . . . . .

900 . . . . . . . . . . ---.-
104 10''~ 10" 10e
Shut-In time, hr

Fig. 18. Semilo~ plot of deconvolved pressure for Example l (d = 5 ft ).

. . . . . . . . . . . . . . . . . .
1600

•~ 15oo . . . . . . . . .
] I
1,-

'~ 1400
¢/)
t/)
G)
t._

~' 1300
i
G)
"0 1200 r
> j . . . .
i
0
>
f.. 1100 -
Slope = 102.58 p s i / c y c l e t
0 kh 3804.10 m d - f t
(J
(9 k 38.04 md
f-t 1000 - = 9.38
- ,i/~Z- .... 1 s
Slope~ = 187.14 psi/cycle
/ J

900 ..... ,j

10"' 10"~ 104 10o


Shut-In t i m e , hr

Fig. 19. Semilog plot of deconvolved pressure for Example 2 (d = 20 ft ).


215

1425

•~ 1350
¢L
d
U) 1275
(n
l_

Q.

.~, 1200

10
10
.~> 1125
0
>
t.-
0 Slope = 97.58 psi/cycle
o kh ffi 3999.20 md-ft
G) 1050 k = 39.99 md
a S = 10.00

975
10"= 10-, 10" 10°
Shut-in time, hr
Fig. 20. Semilog plot of deconvolved pressure for Example 3 (d = 100 ft).

1440 -

"~ 1360-
0,.

l,l,,

(/) 1280- _ m
(/)

Q,.
1200- .............. J
10

j
10
:> 1120
0
>
r"
0 J Slope = 97.82 psi/cycle
0 [kh = 3989.21 md-lt
1040 = 39.89 md
= 9.96

960 I
g

10.4 10"= 10 "t 10" 10'


Shut-in Ume, hr

Fig. 21. Semilog plot of deconvolved pressure for Example 4 (d = 500 ft).
216

small or too large (for drawdown test, there is


no upper limit for the distance ) and to estimate
its parameters, provided that the downhole flow
rate is measured simultaneously with pressure.
c:

Nonlinear least-squares estimation

The purpose of the nonlinear least-squares


estimation is to construct an objective func-
tion, J, that can be minimized to obtain esti-
mates by ensuring the best fit between the
measured pressure and the model behavior. The . . . . . I
Ill;
nonlinear least-squares estimation, also called '~hlll iI~ liIl~,', ht

"automated type-curve or history matching", Fig. 22. Comparison of measured and computed pressure
is similar to type-curve matching. It can be per- derivatives for Example 1 (d = 5 ft).
formed using wellbore pressure with (Guillot
and Horne, 1986; Kuchuk et al., 1986) and nonlinear estimation method does not have any
without (a great deal of information and ref- identification power before the estimation be-
erences on this subject can be found in Rosa gins. However, a systematic use of different
and Horne, 1983; Barua et al., 1985a, b) down- models may eventually lead to the correct
hole flow rate data. model. For a complete interpretation of a well
Thus, the nonlinear estimation refers to test, the nonlinear estimation should be carried
mlmmlzlng: out to verify the model behavior with the ob-
served behavior of the system.
1 Nm
J(~)=~ ~ [Ap~f(x~,ti)-(Zlpwmf(ti)] 2 (25) Using the measured pressure and flow rate
i=l with the mathematical model ( eq. 3 with eq. 1 ),
where the model behavior {computed pres- Fig. 5 presents the match between the mea-
sure), Apwbf (x,ti), is given by either eq. 3, 5, or sured and computed pressures and their deriv-
7, ~ = unknown parameter vector (k, S, d, etc. ) atives for Example 1. The estimates obtained
to be estimated, Apwmbf(ti)=measured pres- from these matches are:
sure, and Nm = number of measured data points. k-- 39.6 mD, S=9.8 and d=4.7 ft
c -* m
The derivative of Apwbf(x,ti) and Apwbf(ti) from the pressure match, and
with respect to in t are also used in eq. 25 for
the estimation of reservoir parameters. The de- k = 39.7 mD, S = 10.4 and d--8.7 ft
rivative of ztp~bf with respect to In t is com- from the pressure derivative match.
puted numerically from ApCbf. Although both matches given in Fig. 5 look
The nonlinear estimation method (type- exact, the enlarged plot of the pressure deriva-
curve matching with rate), which is described tive curve, Fig. 22, shows a mismatch between
above, is used to estimate parameters of the the measured and computed pressure deriva-
system. During estimation, the no-flow barrier tive. The estimated distance from the pressure
distance is also assumed to be unknown. How- derivative is 74% higher than the actual value.
ever, in the nonlinear estimation method the It took 11 iterations to obtain the pressure
model (well-reservoir system) is assumed to be match and 146 iterations to obtain the pressure
known. Unlike conventional, logarithmic con- derivative match starting from the same initial
volution, and deconvolution methods, which guesses.
can also be used as an identification tool, the Fig. 7 presents the match between the mea-
217

140 1600
o Measured ,ressure derivative
[ ~ C o m p u t e ( pressure derivative
1400

"~. 1200

m
1ooo
~ °Oo
~. °o
[l()O

600

80 . . . . . . . I 400 . . . . . . . . I . . . . . . . . I
lO' (1 ' 1() ~ lO ~
Shut-in time, hr SI;uL-in Lime, h r

Fig. 23. Comparison of measured and computed pressure Fig. 24. Comparison of measured and computed pressures
derivatives for Example 2 (d=20 ft). for Example 4 (d=500 ft).

sured and computed pressures and their deriv- Fig. 11 presents the match between the mea-
atives for Example 2. The estimates are: sured and the computed pressures and their de-
rivatives for Example 4. The estimates are:
k=39.9 mD, S=9.9 and d = 19.2 ft
k--28.58 mD, S=5.8 and d--941.6 ft
from the pressure match, and
from the pressure match, and
k=40.9 mD, S = 7 . 6 and d=l.O ft
k=36.5 mD, S=8.6 and d=919 ft
from the pressure derivative match.
As in Example 1, Fig. 7 does not show the from the pressure derivative match.
mismatch between the measured and com- The mismatches between the measured and
puted pressure derivative. However, the mis- computed pressures and their derivatives are
match is clearly shown in Fig. 23. The number shown in detail in Figs. 24 and 25. The esti-
of iterations is 87 for the pressure and 23 for mated distance is almost 100% higher than the
the pressure derivative. The estimated distance actual value from both matches. However, the
from the pressure derivative did not converge
I00
to the correct value. o o Measured pressure der va ,e
/\ --Cnmp, pres. derivative with unknown k and
Fig. 9 presents the match between the mea- 90-
sured and the computed pressures and their de-
rivatives for Example 3. The estimates are: B0-

k=39.8 mD, S--9.9 and d=99.1 ft _=


70-

from the pressure match, and


k--40.2 mD, S-- 10.0 and d--97.1 ft
60-

50
°Oo oo

from the pressure derivative match.


The numbers of iterations are 148 for the 4O . . . . . . . ]

pressure and 69 for the pressure derivative. For lO~ 10'


Shut-in time, hr
this case, the estimated distance from the pres-
sure and pressure derivative did converge to the Fig. 25. Comparison of measured and computed pressure
correct value. derivatives for Example 4 (d--500 ft).
218

permeability and skin values from the deriva- amples, this threshold value is assumed to be
tive match are close to the actual values. As can 44 B / D .
be seen in eq. 12, the no-flow barrier has neg- The buildup examples given here cover a
ligible effects on the pressure behavior of the range of well-to-fault distances, 5 to 500 ft, to
system before 10 h. Thus, permeability and skin illustrate the strengths and weaknesses of each
is estimated as k = 39.9 mD and S - - 9.8 from the interpretation method. In general, the fault
pressure match using a radial model without a identification may become a problem if the dis-
no-flow barrier, with 10 h of pressure and flow tance is too small or too large from the well-
rate data. When these permeability and skin bore. Independent information about the no-
estimates are held fixed (known), the esti- flow barrier is essential for these cases. How-
mated distance using the pressure derivative ever, for intermediate distances, such indepen-
data is d-- 507.6 ft. As can be seen from Fig. 25, dent information, although useful, is not a
the pressure derivative match is improved. prerequisite.
In this paper a fully penetrated well in a sin-
gle-layer infinite reservoir in which only single
Discussion and conclusions
phase fluid flows, is considered. Multiple fault
and layered systems are common occurrences
In this section the interpretation results from in many well known reservoirs. The methods
different methods will be discussed. This dis- as presented here can be extended to multifault
cussion, although limited to buildup tests, may a n d / o r layered reservoirs. For the multilayer
easily be extended to drawdown tests. Three case, a single buildup test is not enough to ob-
factors largely influence the transient response tain reservoir parameters; multilayer testing is
and its interpretation of a well in a faulted res- therefore necessary.
ervoir. First, the afterflow period during which The pressure match gives better results for
the flow rate is measurable, is an important the small and intermediate distances, while the
factor on interpretation methods. This rate pressure derivative approach works well for
measurement problem does not arise for draw- large distances. Of course, one has to remember
down tests, except for a few seconds at the be- that these observations could be a consequence
ginning of the test. Second, the well-to-fault of the nonlinear methods that are used here.
distance, coupled with storage, dictates whether The results presented here show that when
identification of the no-flow barrier is possible the effect of a no-flow barrier is not accounted
from a test. Third, the producing time, which for, the estimated parameters (k, S, and p*) can
is not discussed here, also impacts the devel- be subject to very large errors. The magnitude
opment of the characteristic features of a of the errors is dependent upon the well-to-fault
nearby fault. Earlougher and Kazemi (1980) distance. For example, if the distance is large it
provided a detailed discussion about the effect is possible that only the first straight line may
of the producing time on the estimation of the be observed on the Horner plot. The permea-
distance. This factor becomes important if the bility and skin determined from the first
distance is large and when the conventional straight line will be correct. However, the ex-
methods are used. trapolated pressure will be wrong. On the other
There are also other factors which affect hand, if the distance is small it is possible that
interpretation in practice, such as accuracy and only the second straight line may be observed
smoothness of downhole pressure and flow rate on the Horner plot. The permeability and skin
measurements and the threshold value of the will be incorrect but the extrapolated pressure
flowmeter, beyond which flow rate measure- is determined accurately. Using conventional
ments become unreliable. For the synthetic ex- methods along with logarithmic convolution,
219

deconvolution, and nonlinear estimation PD ----dimensionless sand/ace pressure for the


constant-rate case
methods will lead to the identification of the
PSD = PD q- S
no-flow barrier, a unique estimation of the PD = kh [Pi-Psf(t) ]/141.2qBl~
permeability, skin, reservoir pressure, and PwD ---- kh[pi-pwf(t) ]/141.2qB#
distance. PwbD = kh[pl-pwbf(t) ]/141.2qB/1
Pd = deconvolved pressure, psi
We did not attempt to compute the reservoir
Pwbf = downhole pressure, psi
pressure (average or extrapolated) for all cases. q = flow rate, S T B / D
It could be done easily either from the second qD = n o r m a l i z e d d o w n h o l e rate, qsf/q~B = qwbf/q~B
semilog straight line, if it is present, from the qr = reference flow rate, R B / D
qsf ---- sand/ace flow rate ( f l o w m e t e r readings, rps ),
reservoir drainage volume and geometry, or
RB/D
from the infinite-acting model with a no-flow qwbf = d o w n h o l e flow rate ( f l o w m e t e r readings,
barrier if the drainage volume is unknown. rps), RB/D
Finally, interpretation of pressure and flow r -- radial d i s t a n c e to t h e reservoir boundary, ft
rate data from a reservoir with no-flow bound- rw = wellbore radius, ft
rD = d i m e n s i o n l e s s radius
aries (this may not be available prior to the S = d a m a g e skin
interpretation) does not require the use of all s = Laplace image space variable
the methods as described above. However, in t -- time, h r
an inverse problem (identification of a system tdv = t i m e o f deviation from first semilog line, h r
tD ---- d i m e n s i o n l e s s time, 0.0002637kt/(~/ACt r 2 )
and estimation of its parameters), the tran-
tin = t i m e o f i n t e r s e c t i o n o f two semilog lines, h r
sient pressure and/or flow rate should be inter- t~12 = t i m e to t h e s t a r t o f s e c o n d semilog line, h r
preted using different methods with all ~j = p o i n t in t h e subinterval [ tj,ti + 1]
available prior information, such as past pro- = positive c o n s t a n t
duction, geological, geophysical, well logging, ]~ ~- OL~Ct r~ / (0.0002637k)

etc., to reduce the non-uniqueness of the /~ = oil viscosity, cp


¢ = system porosity
problem. z = d u m m y i n t e g r a t i o n variable
~/ = hydraulic diffusivity, k~ (~pCt)
Acknowledgment 0 = weighting factor for a u n i f o r m w e i g h t i n g for
all tj
= error in flow rate m e a s u r e m e n t , R B / D
We are thankful to Schlumberger manage- J(t) = Dirac delta f u n c t i o n
ment for permission to publish this work.
Subscripts
Appendix -- Nomenclature D = dimensionless
exp -- exponential
i = initial
B = f o r m a t i o n volume factor, R B / S T B
sf = sand/ace
C = wellbore storage c o n s t a n t , b b l / p s i
w = b o t t o m hole, well
c~ = d i m e n s i o n l e s s wellbore storage c o n s t a n t
wbf = wellbore flowing
CD -- 5.6146C/(2gOcthr~ )
wf = wellbore flowing
Ct = s y s t e m total compressibility, p s i - 1
d = well-to-fault distance, ft Superscripts
dD = d i m e n s i o n l e s s well-to-fault distance, 2d/r
= logarithmic c o n v o l u t i o n t i m e c = model or c o m p u t e d
flct
h -- f o r m a t i o n t h i c k n e s s , ft m = measured
k -- permeability, m D - = Laplace t r a n s f o r m o f
Ko -- m o d i f i e d Bessel f u n c t i o n o f s e c o n d kind,
order zero References
ml = first H o r n e r semilog slope, p s i / c y c l e
rft 2 = s e c o n d H o m e r semilog slope, p s i / c y c l e Agarwal, R.G., 1980. A n e w m e t h o d to a c c o u n t for produc-
p -- pressure, psi ing t i m e effects w h e n d r a w d o w n type curves are used to
220

analyze buildup and other test data. SPE 9289, 55th Gerard, M. and Horne, R.N., 1985. Effects of external
Annual Technical Conference and Exhibition, Dallas, boundaries on the recognition of reservoir pinchout
Texas, September 21-24. boundaries by pressure transient analysis. Soc. Pet. Eng.
Agarwal, R.G., A1-Hussainy, R. and Ramey, H.J., Jr., 1970. J., June, pp. 427-436.
An investigation of wellbore storage and skin effect in Gladfelter, R.E., Tracy, G.W. and Wilsey, L.E., 1955. Se-
unsteady liquid flow, I. Analytical treatment. Soc. Pet. lecting wells which will respond to production-stimu-
Eng. J., Sept., pp. 279-290; Trans., AIME, 249. lation treatment. Drill. Prod. Pract., API, pp. 117-129.
A1-Hussainy, R., Ramey, H.J., Jr. and Crawford, P.B., 1966. Gringarten, A.C. et al., 1979. A comparison between differ-
The flow of real gases through porous media. J. Pet. ent skin and wellbore storage type-curves for early-time
Tech., May, pp. 624-636. transient analysis. SPE 8205, 54th Annual Technical
Barua, J., Kucuk, F. and Gomez-Angulo, J., 1985a. Appli- Conference and Exhibition, Las Vegas, Nevada, Sept.
cation of computers in the analysis of well tests from 23-26.
fractured reservoirs. SPE 13662, California Regional Guillot, A.Y. and Horne, R.N., 1986. Using simultaneous
Meeting, Bakersfield, CA, March 27-29. flow rate and pressure measurements to improve anal-
Barua, J., Horne, R., Greenstadt, J.L. and Lopez, L., 1985b. ysis of well tests. SPE Formation Evaluation, 1 (3): 217-
Improved estimation algorithms for automated type 226.
curve analysis of well tests. SPE 14255, 60th Annual Horner, D.R., 1951. Pressure build-up in wells. Proc., 3rd
Technical Conference and Exhibition, Las Vegas, NV, World Petroleum Congress, The Hague, Sec. II, pp. 503-
Sept. 22-25. 523.
Bixel, H.C., Larkin, B.K. and van Poollen, H.K., 1963. Ef- Hutchinson, T.S. and Sikora, V.J., 1959. A generalized
fect of linear discontinuities on pressure build-up and water-drive analysis. Trans. AIME, 216, 169.
drawdown behavior. J. Pet. Tech., August, pp. 885-895; Hurst, W., 1953. Establishment of the skin effect and its
Trans., AIME, 228. impediment to fluid flow into a well bore. Pet. Eng., Oct.,
Bostic, J.N., Agarwal, R.G. and Carter, R.D., 1980. Com- B-6 through B-16.
bined analysis of postfracturing performance and pres- Jargon, J.R. and van Poollen, H.K., 1965. Unit response
sure buildup data for evaluating an MHF gas well. J. function from varying rate data. J. Pet. Tech., Aug., pp.
Pet. Tech., Oct., pp. 1711-1719. 965-969; Trans., AIME, 234.
Cinco, H., Ramey, H.J., Jr. and Miller, F.G., 1975. Un- Jones, P., 1962. Reservoir limit test on gas wells. J. Pet.
steady-state pressure distribution created by a direc- Tech., June, pp. 613-619.
tionally drilled well. J. Pet. Tech., Nov., pp. 1392-1400; Kabir, S., Kuchuk, F.J. and Gomez-Angulo, J., 1986. Well
Trans., AIME, 251. test interpretation in faulted reservoirs. Proc., 6th Off-
Clark, D.G. and van Golf-Racht, T.D., 1985. Pressure-de- shore South East Asia Conference, Singapore, Jan. 28
rivative approach to transient test analysis: a high- 31, pp. 152-161.
permeability North Sea example. J. Pet. Tech., Nov., Katz, D.L., Tek, M.R. and Jones, S.C., 1962. A generalized
pp. 2023-2039. model for predicting the performance of gas reservoirs
Coats, K.H., Rapoport, L.A., McCord, J.R. and Drews, subject to water drive. SPE 328, 37th Annual Fall Meet-
W.P., 1964. Determination of aquifer influence func- ing, Los Angeles, CA, Oct. 7-10.
tions from field data. J. Pet. Tech., Dec., pp. 1417-1424; Kuchuk, F.J., 1987. Gladfelter deconvolution. SPE 16377,
Trans., AIME, 231. California Regional Meeting, Ventura, Ca., April 8-10.
Davis, E.G., Jr. and Hawkins, M.F., Jr., 1963. Linear fluid Kuchuk, F. and Ayestaran, L., 1985. Analysis of simulta-
barrier detection by well pressure measurements - - dis- neously measured pressure and sandface flow rate in
cussion. J. Pet. Tech., Oct., p. 1077. transient well testing. J. Pet. Tech., Febr., pp. 323-334.
Dolan, John P., Einarsen, Charles A. and Hill, Gilman A., Kuchuk, F., Carter, R.G. and Ayestaran, L., 1985. Numer-
1957. Special application of drill stem test pressure data. ical deconvolution of wellbore pressure and flow rate.
Trans., AIME, 210: 318-324. SPE 13960, SPE, Richardson, TX.
Earlougher, R.C., Jr., 1977. Advances in well test analysis. Kuchuk, F.J., Karakas, M. and Ayestaran, L., 1986. Well
SPE of AIME, Dallas, 5: 124-126. testing and analysis techniques for layered reservoirs.
Earlougher, R.C., Jr. and Kazemi, H., 1980. Practicalities SPE Formation Evaluation, Aug., pp. 342-354.
of detecting faults from buildup testing. J. Pet. Tech., Martinez, S. and Cinco-Ley, H., 1985. Detection of linear
Jan., pp. 18-20. impermeable barriers by transient pressure analysis.
Evrenos, A.I. and Regda, E.A., 1965. A digital computer Paper submitted to SPE, Richardson, Texas.
application to the investigation of aquifer properties. J. Matthews, C.S. and Russell, D.G., 1967. Pressure build-up
Pet. Tech., July, pp. 827-838. and flow tests in wells. SPE of AIME, Dallas, Texas, 1.
Fetkovich, M.J. and Vienot, M.E., 1984. Rate-normaliza- Meunier, D., Wittmann, M.J. and Stewart, G., 1985. Inter-
tion of buildup pressure using afterflow data. J. Pet. pretation of pressure buildup test using in-situ mea-
Tech., Dec., pp. 2211-2224. surement of afterflow. J. Pet. Tech., Jan., pp. 143-152.
221

Meunier, D.F., Kabir, C.S. and Wittmann, M.J., 1987. Gas Annual Technical Conference and Exhibition, San
well test analysis: use of normalized pseudovariables. Francisco, CA, Oct. 5-8.
SPE Formation Evaluation, Dec., pp. 629-636. Russell, D.G., 1963. Determination of formation charac-
Overpeck, A.C. and Holden, W.R., 1970. Well imaging and teristicsfrom two-rate flow tests.J. Pet. Tech., Dec.,
fault detection in anisotropic reservoirs. J. Pet. Tech., pp. 1347-1355; Trans., AIME, 228.
Oct., pp. 1317-1325; Trans., AIME, 249. Standing, M.B., 1964. Linear fluidbarrierdetection by well
Pascal, H., 1981. Advances in evaluating gas well deliver- pressure measurements -- discussion. J. Pet. Tech.,
ability using variable rate tests under non-Darcy flow. March, pp. 259-260.
SPE/DOE 9841, SPE/DOE Low Permeability Sym- Stehfest, H., 1970. Algorithm 368, numerical inversion of
posium, Denver, Colo., May 27-29. Laplace transforms. D-5, Communications of the A C M ,
Pinson, A.E., Jr., 1972. Convenience in analyzing two-rate 13(I): 47.
flow tests. J. Pet. Tech., Sept., pp. 1139-1141. Tiab, D. and Kumar, A., 1980. Detection and location of
Prasad, Raj K., 1975. Pressure transient analysis in the two parallelsealing faultsaround a well. J. Pet. Tech.,
presence of two intersecting boundaries. J. Pet. Tech., Oct., pp. 1701-1708.
Jan., 89-96; Trans., AIME, 259. Thompson, L. and Reynolds, A., 1986. Analysis of vari-
Proano, E.A. and Lilley, I.J., 1986. Derivative of pressure: able-rate well-test pressure data using Duhamel's prin-
application to bounded reservoir interpretation. SPE ciple. SPE Formation Evaluation, June, pp. 217-228.
15861, European Petroleum Conference, London, Oct. Thompson, L., Jones, J.R., Reynolds, A. and Raghavan,
20-22. R., 1986. Analysis of pressure buildup data influenced
Ramey, H.J., 1965. Non-Darcy flow and wellbore storage by wellbore phase redistribution. SPE Formation Eval-
effects in pressure build-up and drawdown of gas wells. uation, June, pp. 217-228.
J. Pet. Tech., Febr., pp. 223-233; Trans. AIME, 234. Van Everdingen, A.F., 1953. The skin effect and its influ-
Ramey, H.J., Jr., 1976. Verification of the Gladfelter-Tra- ence on the productive capacity of a well. Trans., AIME,
cey-Wilsey concept for wellbore storage-dominated 198: 171-176.
transient pressures during production. J. Cdn. Pet. Van Everdingen, A.F. and Hurst, W., 1949. Application of
Tech., April-June, pp. 84-85. the Laplace transformation to flow problems in reser-
Ramey, H.J., Jr. and Earlougher, R.C., Jr., 1968. A note on voirs. Trans., AIME, 186: 305-324.
pressure build-up tests. J. Pet. Tech., Febr., pp. 119- Van Poollen, H.K., 1965. Drawdown curves give angle be-
120.
tween intersecting faults. Oil Gas J., Dec., 20: 71-75.
Ramey, H.J., Jr., Kumar, A. and Gulati, M.S., 1973. Gas Witherspoon, P.A., Javendal, I., Neuman, S.P. and Freeze,
well test analysis under water-drive conditions. AGA R.A., 1967. Interpretation of aquifer gas storage condi-
Monograph 61-51, Arlington, VA. tions from water pumping tests. AGA Monograph NS-
Rodgers, J.S. and McArthur, B.W., 1972. Implicit solu- 38, New York City, NY.
tions and precision pressure measurements provide re- Yaxley, L.M., 1985. The effect of a partially communicat-
liable transient analysis. SPE 4056, 47th Annual ing fault on transient pressure behavior. SPE 14311, 60th
Technical Meeting, San Antonio, TX, Oct. 8-11. Annual Technical Conference and Exhibition, Las Ve-
Rosa, A.J. and Home, R.N., 1983. Automated type-curve gas, Nev., Sept. 22-25.
matching in well test analysis using Laplace space de-
termination of parameter gradients. SPE 12131, 58th

Vous aimerez peut-être aussi