You are on page 1of 13

SPE 63077

Frequently Asked Questions in Well Test Analysis


S. Daungkaew, F. Hollaender and A.C. Gringarten, Imperial College, London

Copyright 2000, Society of Petroleum Engineers Inc.


This paper was prepared for presentation at the 2000 SPE Annual Technical Conference and
Exhibition held in Dallas, Texas, 14 October 2000.
This paper was selected for presentation by an SPE Program Committee following review of
information contained in an abstract submitted by the author(s). Contents of the paper, as
presented, have not been reviewed by the Society of Petroleum Engineers and are subject to
correction by the author(s). The material, as presented, does not necessarily reflect any
position of the Society of Petroleum Engineers, its officers, or members. Papers presented at
SPE meetings are subject to publication review by Editorial Committees of the Society of
Petroleum Engineers. Electronic reproduction, distribution, or storage of any part of this paper
for commercial purposes without the written consent of the Society of Petroleum Engineers is
prohibited. Permission to reproduce in print is restricted to an abstract of not more than 300
words; illustrations may not be copied. The abstract must contain conspicuous
acknowledgment of where and by whom the paper was presented. Write Librarian, SPE, P.O.
Box 833836, Richardson, TX 75083-3836, U.S.A., fax 01-972-952-9435.

Abstract
Despite the tremendous progress achieved in well test analysis
in the last twenty years, many important and practical
questions that are frequently asked by practicing engineers
have received little attention in the literature. Three of these
are addressed in this paper. The first one is whether it is
possible to obtain more information from build-up data than
from drawdown data in a testing sequence. By introducing a
new definition of the radius of investigation as opposed to
the radius of drainage based on the ability to interpret data
using pressure derivatives, we show that the information
obtainable from a long build-up following a short drawdown is
actually limited by the gauge accuracy and noises as the shutin time increases. The second frequently asked question
(FAQ) relates to the minimum amount of detailed rate history
which is required in order to obtain correct pressure
derivatives, as a truncated or averaged rate histories may
modify the derivative shapes at late times and therefore induce
erroneous interpretations. A new rule of thumb is proposed,
which combines the most recent flow-rate history with an
effective time based on a fraction of the cumulative well
production. The last FAQ concerns the possibility to
distinguish a non-uniform mechanical skin effect from a
uniform one from well test data. Current interpretations with
fully penetrating wells yield a single value for the mechanical
skin effect, which implicitly assumes that damage is uniform
over the well surface. Various skin distributions are
investigated with a multilayered model to determine how a
non-uniform skin distribution around the wellbore affects the
interpretation. It is shown that non-uniform skin distribution
can be identified during specific flow-regimes and therefore
must be taken into account in the interpretation.

Introduction
Very significant advances have been made in well test analysis
over the last twenty years with the introduction of a systematic
interpretation methodology, pressure derivative analysis and
new interpretation models. As a results, well test analysis has
not only become more powerful, but also easier and faster to
perform. There remains, however, a number of issues which
are faced by practicing engineers in their day-to-day work, but
have not received much attention in the literature. This paper
aims at giving some answers to these frequently asked
questions (FAQ) in well test analysis.
FAQ # 1: Can we see more in a build-up than in a
drawdown?
This has been a subject a disagreement for many years.
Some defend the concept that once you start producing a well,
the pressure disturbance is felt everywhere instantaneously,
and therefore, by shuting the well in indefinitely, you should
be able to obtain all possible information on the reservoir,
even that which has not been obtained in the drawdown.
Others believe that if you dont see it in the drawdown, you
dont see it in the build-up. The answer to the question can be
obtained from the concept of the radius of investigation of a
well test, which is the distance at which a given feature of the
reservoir can be interpreted, taking into account the influence
of noises in the data and the rate history of the test.
Radius of drainage
The radius of investigation, as defined in this paper, is
different from the radius of drainage which is routinely used
in well test analysis. The radius of drainage has been the
subject of many publications, mainly in the 1950s and early
1960s (summarized by Van Poolen1) with a few more recent
ones2-4. They all propose similar expressions, of the form
reD=1+D.tD0.5 in dimensionless parameters4. In most cases, the
use of the line-source solution with an infinitely small
wellbore radius simplifies this equation to reD=D.tD0.5. A
summary of the various definitions is given in Appendix A.
Some drainage radius equations have been developed for
pressure drawdowns1,3-8, others for pressure build-ups2,9-11.
Most definitions are based on the start of the deviation from
radial flow in an infinite acting, homogeneous reservoir
behavior model, which in most practical cases does not
correspond to the actual reservoir conditions. Ehlig-

S. DAUNGKAEW, F. HOLLAENDER, A.C. GRINGARTEN

Economides12 has extended the concept of radius of


investigation to that of area of investigation. Another
interesting approach uses interference tests in order to
determine pressure disturbance travel time in a reservoir13.
Theoretical radius of investigation
One of the most-commonly used definition of the radius of
investigation as a function of the test duration has been
developed by considering the drainage area1,7:
riD = 2 t D ........................................................................................( 1)
This definition does not take into account the ability to
interpret features located at that distance. In this paper, we
focus on the requirement that a feature at a distance equal to
the radius of investigation must be interpretable using pressure
derivatives.
We consider a single drawdown at a constant rate q and a
homogenous reservoir bounded by a sealing fault at a distance
d from the well. To derive the radius of investigation, we first
assume that the fault is identifiable if the second derivative
stabilization due to the fault has reached 90% of its final level.
We will also determine the radius of investigation for a 10%
development of the second transition period.
Mathematically, the dimensionless pressure response for a
line source well with a sealing fault is obtained by using the
principle of superposition in space:
PD =

1
1

1
Ei
(2d D )2 ..................................( 2)
Ei
2
4tD
4tD

with the exponential-integral function defined as:


Ei ( x) =

e
du .........................................................................( 3)
u

For tD large enough, the derivative of the dimensionless


pressure drop tends to 1, each exponential-integral term
derivative tending to 1/2. Assuming that the first stabilization
has been reached, the condition for the fault to be interpretable
on a log-log pressure derivative plot is obtained in Appendix
B as:

1
2
d E i (d D )
0.9
tD
1


1 ...............................................( 4)
d ln t
2

The corresponding value of tD may be obtained


numerically or analytically. Both methods yield:
A numerical solution yields:
riD = 0.379 t D ................................................................................( 5)
The corresponding dimensionless radius of investigation
for 10% of the fault stabilization level instead of 90% is:
riD = 1.623 t D ................................................................................( 6)
Eqs. 5 and 6 yield radii of investigation which are smaller
than the ones defined by most previous authors. The radius
from Eq. 5 is compared with that from Eq. 1 in Figure 1.
Figure 1 represents the pressure and derivative behavior in a
homogeneous reservoir with a sealing fault at a distance of
400ft away from the well. The other model parameters are
based on the North Sea field of Balmoral14. Assuming that the

SPE 63077

radius of investigation targeted is 400ft, Eqs. 1, 5 and 6 yield


test durations of 1.02 hour, 28.3 hours and 1.55 hours,
respectively. As can be seen in Figure 1, a test of duration of
1.02 hour does not allow identification of the fault located at
400ft. A test duration of 28.3 hours does, whereas a test
duration of 1.55 hour only suggests that a change of flow
regime is taking place.
Practical radius of investigation
The radius of investigation defined above does not take into
account the influence of gauge resolution or noises in the
pressure data. Noises cause errors on the pressure, which are
amplified on the derivative. This is particularly important
when the change in pressure is small compared with the gauge
resolution, for instance in the case of small flow rates or highmobility reservoirs. Figures 2 to 4 show the effect of noises on
the derivative. In the following, we first study the influence of
noises on the derivative and then examine the case of a buildup following a single drawdown.
Using the algorithm developed by Bourdet15, it is possible
to determine a condition for the pressure change over a given
time to yield a satisfying derivative. This algorithm calculates
an average value of the derivative over a time window of halflength L (generally expressed as a percentage of the total timerange of the test), using the mid-point at which the derivative
must be estimated, (Pi, ti), and the first and the last point in
the window, (P1, t1) and (P2, t2) respectively:
Pi P1
P Pi
(ln t2 ln ti ) + 2
(ln ti ln t1 )

ln
ln
ln
t
t
t2 ln ti
dP

..............( 7)
1
i

=
ln t2 ln t1
d ln t i

If we use points at the limits of the window (actual or


interpolated), Eq. 7 simplifies to:
P2 P1 ..................................................................( 8)
dP

=
ln
2L
d
t

i
A measured pressure value can be expressed as a
theoretical, exact value with added noise:
P (t ) meas = P (t ) real + noise ...........................................................( 9)

If we want to know the derivative within 10% (an


acceptable error on the mobility), Eq. 8 yields:
noise 2 noise1
0.1 ....................................................................( 10)
(P2 P1 )theo

With noises belonging to [-Anoise, Anoise], Eq. 10 finally


yields:
Anoise
P2 P1

0.05 .....................................................................( 11)


theo

This is the condition for obtaining an interpretable


derivative. It is a function of the noise amplitude, of the size of
the time window used to calculate the derivative and of the
type of reservoir studied. Given the characteristics of a given
reservoir and the quality of the derivative needed (a larger
window gives a better global aspect but tends to hide smallscale features), it is possible to determine a minimum flowrate allowing the derivative to be interpreted with confidence.

SPE 63077

FREQUENTLY ASKED QUESTIONS IN WELL TEST ANALYSIS

This condition considers a worst-case scenario where the noise


is such that the derivative is affected most. Actually, this
condition should be much less restrictive. On a probabilistic
point of view, the likeliness of having pressure points with
noise of maximum amplitude (thus affecting the derivative
most), is infinitely small. If the structure of noises is known, it
is possible to modify equation 11 by adding a probabilistic
confidence interval.
Interpretation of build-up data
Well test analysis is usually performed on build-up data,
because drawdown data tend to be affected by rate fluctuations
and wellbore dynamics. A major problem with long buildups, however, is that the pressure tends to stabilize as the
reservoir pressure goes back to equilibrium and pressure
variations are getting smaller and smaller. Consequently, as
the test goes on, the derivative becomes harder to interpret
because the signal-to-noise ratio decreases.
Two methods can be used to interpret build-up data, based
on the principle of superposition in time. The first one is based
on build-up or multirate type-curve analysis and considers the
pressure changes from the beginning of the build-up period.
The derivative is then calculated using Horners superposition
time. The second one, which has been the object of much
research but still does not work satisfactorily, consists of
reconstructing a drawdown type-curve by correcting for the
build-up effects by inverting Duhamels principle of
superposition (deconvolution). In the following, we examine
the influence of noises on the first method and illustrate on an
example how the usable duration of build-up data can be
estimated.
With the pressure drop in a build-up defined as the
difference between the build-up pressure at an elapsed buildup time, p(t), and the pressure at the start of the build-up,
p(t=0), the dimensionless build-up type curve is given by16:
pBU D (t D ) = pD (t P + t ) D + pD (t ) D + pD (t P ) D ..................( 12)

As t increases, the pressure tends to the pressure at


equilibrium, the variations pBU become smaller, and the
influence of noises becomes greater and tends to make the
interpretation harder. If we consider a reservoir of infinite
extent with homogenous reservoir, Eq. 12 becomes17:
PBU =

70.6 .Bq 0.00105k t P + t ..........................( 13)

ln
2

kh
.c t (2 RW ) t.t P

Using Eq. 10 with Horner time leads to the condition:

70.6 .Bq t P + t.e L


A
2 L noises ................................( 14)
ln
L
0.05
kh
t P + t.e

which yields the maximum interpretable duration for the


build-up:
2L

t = t P

Anoise kh
3 . 53 . Bq

e L e L .e

A
kh
2 L noise
3 . 53 . Bq

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

Equation 15 is applicable to any model showing


stabilization on the derivative: infinite homogeneous reservoir,
sealing fault, composite behavior, etc. We can notice that the

most influential parameter is the drawdown duration in this


equation. The usable build-up duration is directly proportional
to the drawdown duration.
To illustrate the use of Eq. 15, we consider a build-up
following a first drawdown of 10 hours at a rate of 5000
bbl/day in a reservoir with a permeability-thickness of
5000mD.ft, a fluid of viscosity of 1cP and a formation volume
factor of 1. Assuming a maximum noise amplitude of 0.2 psi,
a 0.15 log cycle derivative window, Eq.15 yields a maximum
build-up duration of 43 hours (after 43 hours, the error in the
derivative may be greater than 10%). With a rate of
2500bbl/day, this duration would only be 16.5 hours. On the
other hand, changing the size of the calculation window from
0.3 to 0.8 log cycle would increase this duration up to 134
hours but this operation may well mislead the interpretation.
An application of this is shown in Figures 5 to 7. Figure 5
shows a case where the use of build-up data can clearly
provide more information than the preceding drawdown. A
fault that didnt appear during the drawdown is visible and
interpretable in the build-up. Inversely in Figure 6 noises are
larger and it is impossible, or at least harder, to make a correct
interpretation. Finally, Figure 7 shows the impact of the
duration of the drawdown on the amount of data collectable.
With the same level of noise as the one used in Figure 5 but a
shorter drawdown it is impossible to interpret the fault.
It therefore appears that we can see more in a build-up than
in a drawdown but the extra amount of information collectable
is limited and largely dependent upon the drawdown duration.
Conclusions
1. A new definition for the radius of investigation has
been developed, which takes into account the ability to
interpret a geological feature located within this radius
from a pressure derivative plot
2. The influence of gauge resolution and other noises has
been considered and the use of equation 15 provides a
safety guideline concerning the amount of build-up
data confidently interpretable. The critical point is to
estimate the amplitude of noises on a statistical basis.
3. It has been shown that if, theoretically build-up
analysis can provide more information than drawdown
data, the additional amount of information is limited by
the existence of noises on the pressure.
FAQ # 2: How much of the rate history is required for
a correct analysis?
Along with pressure measurement errors, an imperfect
knowledge of the flow-rate history can lead to erroneous
model identifications or/and wrong parameters estimations.
This is particularly true with the pressure derivative method
which is very powerful, but at the same time is more sensitive
to the accuracy and completeness of rate and pressure data that
previous, less powerful techniques. The reason is that
derivatives in multirate tests are taken with respect to the
superposition function which depends on the rate history:

S. DAUNGKAEW, F. HOLLAENDER, A.C. GRINGARTEN


i 1

With qi being the flow-rate between t j and


j =1

t
j =1

the superposition time is:


n 1
n 1

tsup = (qi qi 1 ) log t j + t (qn 1 qn )log(t ) ..(16)


i =1
j =i

Yet, well test data are still reported as a single build-up, with
an average production rate and a duration calculated from the
Horner equivalent time. On the other hand, it is unrealistic to
expect a complete rate history for wells with many years of
production. The question is thus how much rate history detail
should be considered to obtain a correct analysis?
The need to take into account the entire rate history has
been acknowledged in the past18 and several methods have
been proposed to account for production history effects such
as the use of the Horner equivalent time19 or the introduction
of a modified rate and time20 when the production has not
been stabilised prior to shut-in. Several studies have been
published on build-up data analysis but most of them focused
on pressure analysis by mean of the Horner plot. Among the
existing methods, the use of the Horner equivalent producing
time is the most common. Its advantages and disadvantages
have been summarized by Swift21. More recently, Spivey et
al.22 have discussed the effects of rate history on well test
analysis but without quantifying the errors made. All these
studies have usually been based on radial flow-regimes, i.e.
the logarithmic approximation and have not been updated for
derivative analysis. Cinco-Ley et al.23 considered other flowregimes and introduced the second derivative as a way to
estimate the initial pressure. Later, Cinco-Ley et al.24 extended
the use of pressure derivative type-curves and pointed out the
limitations of the use of the superposition time.
In this section, we examine several typical rate histories
and the impact that truncating or averaging them has on the
calculation of pressure derivatives. The model used is an
infinite, homogenous behavior reservoir with wellbore storage
and a zero skin effect. We then develop rules of thumb to
define the minimum rate history information required for a
confident analysis.
Impact of typical rate history approximations
Truncation of the rate history
We consider here the influence of truncation by using only
part of the rate history defined as a percentage of the
cumulative production, i.e. we account for the last 20%, 40%,
etc., and ignore the first 80%, 60%, etc., respectively. The
impact on the derivative is shown in Figure 8. Figure 8 is a
dimensionless graph, with the model dimensional pressure on
the y axis:
pD =

kh
p ..(17)
141.2qB

and the model dimensionless time on the x axis:


tD
kh p ..(18)
= 0.000295
C
CD

SPE 63077

When only part of the rate history is used, the cumulative


production is less than what it should be and the derivative
exhibits an upward trend at late times. Even with the last 80%
of the cumulative history accounted for, the error in the
derivative is greater than +/- 10%, which is the acceptable
error band for the estimate of the permeability-thickness
(represented by the two horizontal lines above and below the
derivative 0.5 stabilization line in Figure 8). In terms of model
recognition, the derivative plots may wrongly be diagnosed as
indicating a boundary effect. The rate history used for Figure 8
is shown at the top left corner. It corresponds to the most
unfavorable but practical case of a rate declining with time,
with most of the cumulative production occurring at early
times. The deviation from the correct derivative would be
much less severe with rates increasing with time.
Use of Horner equivalent time
In this case, the rate history is simplified and reduced to
one single drawdown with a rate equal to the last rate value
prior to shut-in, qlast, and a duration, tpe, equal to the correct
cumulative production, Q, divided by that last rate19:
t pe =

24 Q ..(19)
qlast

Contrary to the preceding case, where the rate is truncated,


here, the correct cumulative production is maintained and the
derivative calculated with tpe should match the correct one at
large values of t (see Effect of production time below).
Before that, however, the shape of the derivatives is different
and therefore a wrong interpretation is likely to be made.
A comparison between the dimensionless derivative based
on the Horner equivalent time and the correct one is shown in
Figure 9.
New rate history approximation
As pointed out by Cinco-Ley and al.24, multi-rate build-up
data are mainly influenced by two parameters: the recent rate
history, which determines the pressure response of early-time
data, and the cumulative production, which has a dominant
influence on late-time data (t>3tp, where tp is the total
duration of the rate history). In-between, data are influenced
by both. This suggests a method for approximating the rate
history which can provide correct early and late time
derivative: the actual production history is decomposed into
two terms, the first one being the most recent rate history,
expressed as a fraction of the cumulative production, and the
second being the Horner equivalent time for the earlier,
remaining rate history. This rate history approximation is
referred to as %+tpe in the remaining of this paper. It is well
suited to the kind of information available in pratice for
production wells, namely monthly production data and the rate
history measured prior to a test.
The quality of the derivative based on the %+tpe rate
approximation depends on the choice of the length of the two
rates components. Figure 10 shows the results with detailed
description of the rates during the most recent 0, 20, 40, 60,

SPE 63077

FREQUENTLY ASKED QUESTIONS IN WELL TEST ANALYSIS

80 and 100 percent of the cumulative productions. The poorest


derivative is obtained with 0%, which amount to using only
the Horner equivalent time only. The time of maximum error
is mainly a function of the total production time. Three
simulations have been run to verify this. In all three cases, the
rate history has the same shape but the dimensionless rate
history duration is different in each case, 4.100, 4.102 and
4.104, respectively. Results in Figure 11 show that the
maximum error is located around t=tP. This may help
estimate which part of the data obtained from multi-rate buildup analysis may be wrong.
We have tested the %+tpe rate approximation with various
rate histories in order to estimate the minimum amount of rate
information that must be taken into account for a correct well
test interpretation. In each case, we have compare derivatives
calculated with 100, 80, 60, 40, 20 and 0% of the cumulative
production. The 100% case corresponds to the correct
derivative, with 100% of the rate history. The 0% case
corresponds to the usual Horner equivalent time
approximation. The condition for determining the minimum
amount of rate history needed is that the corresponding
derivative must stay inside this 10% window around the 0.5
stabilization line.
The various rate histories investigated include alternating
drawdowns and build-ups (Figure 12); linearly decreasing
rates (Figures 13 and 14); linearly increasing rates (Figure 15);
and exponentially decreasing rates (Figure 10). In all of these
cases except the first one, the minimum detailed rate history
required for a correct derivative corresponds to the last 20% of
the cumulative production, with the previous 80% being
described with a Horner equivalent time. The first case
requires the last 40% of the cumulative production.
As rate measurement uncertainties are usually important (a
10% accuracy is usually accepted) it is interesting to assess the
importance of rate uncertainties compared to the effect of
simplification of the rate history. As shown in Figure 16,
differences between the actual rate history and the one used
for analysis (i.e. the measured one) does not cause much errors
as long as a right cumulative production is used. Only a
slightly higher amount of rate history is needed in order to
obtain a level of stabilization with an accuracy of at least 10%
(40%+tPe gives a satisfying result).
Real field data example
In order to check the validity of this theoretical study,
several real field examples were considered and the impact of
using a simplified rate history was observed. These studies
confirmed the results obtained analytically: if using 20% of
the rate history along with Horner equivalent time can
generally allow us to perform a satisfying interpretation, using
30% of the rate history gives very accurate results. It is
therefore confirmed that the use of 30%+tPe can provide a very
satisfying interpretation of pressure derivative curves.
Figures 17 to 19 show an example of real field data
analysis using the new method. As the test duration is much
smaller than the producing period, late-time data are not

affected by the use of a wrong rate history. However, using a


part of the recent rate history improves significantly the
quality of early-time data.
Conclusions
1. A new approximation of the rate history is proposed,
which combines a Horner equivalent time with a
detailed history for some part of the production history
before the test.
2. From the study of various rate histories, it appears that
rates should be described accurately for the last 40% of
the cumulative production, with the first 60% being
approximated with a Horner equivalent time.
3. Rate measurement uncertainties, usually about 10%, do
not really take a major part on the errors but so does the
knowledge of the cumulative production.
4. The same type of study has been performed using real
field data. The same order of magnitude has been
deduced: the use of 30%+tPe is advisable to perform an
accurate analysis. However, this seems to provide a
very satisfying accuracy.
FAQ #3: Is it possible to identify a non-uniform skin
effect from well test data in a fully penetrating well?
Numerous investigations have been carried out in the area
of skin effects since 1953. However, most studies have
concentrated on uniform skin distribution. In reality, it is
likely that the skin along the wellbore is not uniform. Only
two previous publications have dealt with this.
Ozkan and Raghavan25 studied the performance of
horizontal wells under the combined influences of wellbore
friction and wellbore damage. The role of non-uniform
damage on well performance was discussed including the
effect of the skin distribution on the productivity of horizontal
wells. In their study, three types of non-uniform skin
distributions were investigated: conical skin distribution,
arbitrary skin distribution and partially open horizontal wells.
They showed that if there is a conical skin region around
the wellbore, then the late time performance of the horizontal
well is affected by the arithmetic mean of the skin factor
distribution. Both arithmetic mean and the distribution of skin
factors affect the early and determinate time responses.
If only some segments of well are damaged, then early and
intermediate time flow characteristics are influenced by the
skin factor distribution. If the damaged segments have
identical skin factors, then the early time responses are
governed by the arithmetic mean of the skin factors. For the
case of a limited-entry well, they showed that the distribution
of open intervals would yield significant changes in the flow
characteristics at intermediate time.
They also indicated that the correct vertical permeability
cannot be determined from the analysis of the well test data
when the skin distribution is unknown. Similarly, it is not
possible to infer the distribution of skin factors from the
characteristics of the pressure and derivative responses of
horizontal wells and this, therefore, makes production logging

S. DAUNGKAEW, F. HOLLAENDER, A.C. GRINGARTEN

data an essential input in the evaluation of horizontal well


performance.
Yildiz and Cinar26 studied the influences of the formation
anisotropy, total perforated ratio and mechanical skin
distribution on well productivity. They concluded that pseudo
skin due to selective completion is interrelated to the
mechanical skin distribution and its magnitude. They showed
the influence of skin distribution on pressure and pressure
derivative responses when the degree of damage is kept
constant. They concluded that when the skin distribution is
strongly non-uniform, the pressure and derivative pressure
responses are substantially different during early time radial
flow and the transition flow period. When there are few
variations in the skin values, the pressure derivative responses
in a damaged well are identical to those of an undamaged
well. Thus, they concluded that it would be rather difficult, if
not impossible, to estimate the mechanical skin across each
open segment by examining only the pressure data: the rate
distribution should be monitored during well testing
operations to estimate the skin distribution.
This study investigates the effect of non-uniform skin
distribution in a vertical well when the degree of damage is
kept constant. A muti-layered reservoir model27 is used to
simulate different skin factors in an otherwise homogeneous
reservoir. Two types of non-uniform, continuous skin
distributions (conical skin distributiona and alternating change
skin distributionb) and one type of non-uniform, discontinuous
skin distribution (limited-entry wells), are examined for both
infinite and channel reservoirs. Finally, results are compared
with those presented by Ozkan and Raghavan25 and Yildiz and
Cinar26.
Non-uniform, continuous skin distributions, infinite
reservoir
The skin distributions used are presented in Appendix D.
Distribution G9 is uniform; G10, G11 and G12 correspond to
conical skin distributions. Distribution G22 is uniform; G18 to
G21 are alternating change skin distributions. For both
distribution types, pressure and derivative responses have been
calculated for various total skin factors. Results for a negative
total skin factor, St=-3 and St=4.76 are shown in Figures 20
and 21, respectively. There are only some differences between
uniform and non-uniform skin distributions in Figure 20, but
these are not significant enough to allow differentiation in
actual tests. G12, which shows the largest difference is more
likely to be analyzed as a uniform skin of higher value than as
a non-uniform skin. Results in this case exhibit more
distinguishable curves between uniform and non-uniform
distribution compared to the case with a positive total skin
factors, which are not shown in this paper. There is more
a

In a conical skin distribution, the value of the mechanical


skin factor continuously decreases from segment to segment
along the vertical well25.
b
In an alternating change skin distribution, the mechanical
skin factor changes arbitrarily.

SPE 63077

difference of practical significance in Figure 21. Systems G18,


G19 and G20 show more differences than system G21. This
can be explained by the fact that the heterogeneity across the
wellbore is more important in this case. Inversely, in the case
of an alternating skin distribution, the case of a high positive
skin factor gives more distinguishable curves than when using
a low positive or negative total skin factor.
Limited entry wells, infinite reservoir
Pressure drops and derivatives for a high total skin factor,
St=20, and five skin distributions (G23 to G27 in Appendix D)
are shown in Figure 22. G27 has a uniform skin distribution,
whereas G23 corresponds to a limited entry. Except for G23
(partial penetration), the limited entry curves corresponding to
G22 to G26 would be difficult to distinguish in practice from
the total penetration curves (G27). These results do not agree
with those presented by Yildiz and Cinar26. They suggested
that the pressure and derivative pressure responses for each
case are substantially different during early and intermediatetime regimes. The reason for this is the fact that their case is
composed of only two different skin zones. Their skin
distribution can be identified as a conical skin distribution
instead of a limited-entry well. Their results agree then with
the case of a conical skin distribution presented above.
Non-uniform, continuous skin distributions, with a
channel boundary
In this part, a channel boundary was included into the
models, with the distance between this boundary and the well
varying from 2000 down to 25 ft. Since Ozkan and
Raghavan25 have investigated the effect of skin distribution for
a horizontal well and ttheir case is similar to the one
investigated here, their results will be compared with those
obtained in this study.
Figures 23 and 24 show the pressure drop and derivative of
systems G9 to G12 for several distances to the boundary.
Again, the impact of non-uniform skin distribution cannot be
seen clearly except when the channel boundary is at 2000 ft,
there is more discrepancy between the pressure curves
produced from non-uniform skin distributions at the early time
regime. Ozkan and Raghavan25 concluded that the nonuniform skin effect can be seen in both pressure and derivative
pressure responses in a horizontal well at the early and
intermediate time regimes. Their conclusion does not
correspond to the case of positive skin factors, but it only
agrees with the case of negative total skin factors where the
boundary is close to the well.
In the case of non-uniform, alternating change skin
distribution, Figures 25 and 26 show results of systems G18 to
G22, having a total skin factor of 4.76. The results show that
for the pressure response, the difference between uniform and
non-uniform skin can be seen clearly when the boundary is
closer, as indicated in figure 25. The results from this study
agree with Ozkan and Raghavan who concluded that the effect
of different skin distributions can be noticed at intermediate

SPE 63077

FREQUENTLY ASKED QUESTIONS IN WELL TEST ANALYSIS

and late-time regimes on the pressure response. However, the


results also indicate the difference between uniform and nonuniform skin distributions at the early time regime with both
positive and negative total skin factors, but there is no
different between non-uniform skin distributions compared
with each other.
Limited entry wells, with a channel boundary
Finally, Figures 27 and 28 show the pressure and pressure
derivative responses of systems G23 to G27 with a total skin
factor of 20. From figure 28, it appears that the non-uniform
skin distribution affects the derivative pressure responses at
the intermediate-time regime. On the other hand, this effect
was found on the pressure responses at middle and late-times.
The results from this study agree with the results presented by
Ozkan and Raghavan. It is important to note that in figure 28 it
is difficult to diagnose systems G24, G25 and G26 as a limited
entry well because the negative half slop at middle-time
regime is not significant for these systems. In contrast, this
characterization is clear for system G23.
The non-uniform skin can cause errors in the estimation of
kh up to 38% when the boundary is at 500 ft. Similarly, when
the boundary distance is of 50 ft, its distance can be wrongly
interpreted up to 23% of error when the effects of non-uniform
skin distribution are not taken into account for the analysis.
Figure 27 indicates that when the channel is close from the
well, the middle-time pressure responses display more
distinguishable curves between different non-uniform skin
distributions because of the boundary effects.
Conclusions
1. Practically, it is not possible to identify a non-uniform
skin distribution using log-log plots in the case of a
conical skin distribution.
2. Different types of non-uniform skin distribution affect
pressure and derivative pressure responses at different
flow regimes. This difference can be seen more clearly
in the limited-entry well and in the alternating change
than in the conical skin distributions. However, it also
depends on the value of total skin factor.
3. When the boundary is added into a system, the distance
to the boundary influences the regimes affected by a
non-uniform skin distribution.
Nomenclature
PD = dimensionless pressure drop
= dimensionless time
tD
dD
= dimensionless distance
riD
= dimensionless radius of investigation
k
= permeability (in mD)
h
= thickness (in feet)

= viscosity (in cP)


B
= form volume factor
q
= oil flow-rate

= porosity
= total compressibility (in 1/psi)
ct

rW
= wellbore radius (in feet)
L
= window half- length
Anoise = amplitude of noises (in psi)
tP
= production time (in hours)
t
= build-up duration (in hours)
= Horner equivalent time
tPe
m
= slope of the Horner curve

= block shape parameter (in feet-2)


C
= wellbore storage coefficient (bbl/psi)
Definition of dimensionless parameters
kh
pressure drop
P =
P
141.2 .Bq
0.000264.k
tD =
t
.ct rW2
D

dD =

d
rW

time
distance

r
rw
(k h / ) i
i =
(kh / ) total
rD =

rw
(k h h / ) total
(ct h) i
i =
(ct h) total
0.8936C
CD =
[(ct h) total ]rw 2

radius

mobility ratio

i =

interporosity flow parameter


storativity ratio
wellbore storage

References
1. Van Poolen, H.K.: Radius of Drainage and Stabilisation-Sime
Equations, The Oil and Gas Journal, Sept. 1964,p138-146
2. Hurst, W.: The Radius of Drainage Formula, The Oil And Gas
Journal, July 1969, pp66-69
3. Johnson, P.W.: The Relationship Between Radius of Drainage
and Cumulative Production, SPE 16035, 1986
4. Kutasov, I.M., Hejri, S.: Drainage Radius of a Well Produced
at Constant Bottomhole Pressure in an Infinite Acting
Reservoir, SPE 13382, 1984
5. Jones, P.: Reservoir Limit Test on Gas Wells, JPT, June 1962,
pp613-618
6. Tek, M.R., Grove, M.L., Poettmann, F.H.: Method for
Predicting the Back-Pressure Behaviour of Low-Permeability
Natural Gas-Wells, Trans. AIME, pp210-302,1957
7. Muskat, M.: The Flow of Compressible Fluids Through Porous
Media and some Problem in Heat Conduction, Physics 5, 71,
March 1934
8. Chatas, A.T.: A Practical Treatment of Nonsteady-State Flow
Problems in Reservoir Systems, part 3, Petroleum engineer,
B-44, August 1953
9. Gray, K.E.: Approximating Well to Fault Distance from
Pressure Build-Up Tests, JPT, July 1965, pp761-767
10. Hurst, W., Haynie, O.K., Walker, R.N.: Some Problems in
Pressure Build-Up, SPE 145, presented at the 36th Annual Fall
Meeting of the Society of Petroleum Engineers in Dallas,
October 8-11, 1961

S. DAUNGKAEW, F. HOLLAENDER, A.C. GRINGARTEN

11. Ishteiwy, A.A., Van Poolen, H.K.: Radius of Drainage


Equation for Pressure Build-Up, SPE 2468, presented at the
Libyan Association of Petroleum Technologists Annual
Meeting, January 25-26, 1967, Tripoli
12. Ehlig-Economides, C.A.: Computation of Test Area of
Investigation in Nonradial Geometries, SPE 25020, presented
at the European Petroleum Conference in Cannes, 16-18
November 1992
13. Narujieta, H.L., Robles, O.O., Edwards, D.P.: Interference
Well Testing to Predict Early Water Breakthrough in Naturally
Fractured Reservoirs, SPE 15636, presented at the 61st Annual
Technical Conference and Exhibition of the Society of
Petroleum Engineers in New Orleans, October 5-8, 1986
14. Robertson ed.: Oil and Gas Fields of the North Sea: Volume 2:
Fields of the UK Sector of the Central and Northern North Sea
and the Morecombe and Cuytch Form Fields, Robertson
group, 19
15. Bourdet, D., Ayoub, J.A., Pirard, Y.M.: Use of Pressure
Derivative in Well-Test Interpretation, SPEFE, June 1989,
pp293-302
16. Ramey, H.J., Jr. and Cobb, W.M.:"A General Pressure Build-up
Theory for a Well in a Closed Drainage Area," J. Pet. Tech.(
Dec., 1971) 1493-1505; Trans., AIME ( 1971), 252.
17. Agarwal, R.G.: A New Method to Account for Producing Time
Effect when Drawdown Type Curves are Used to Analyse
Pressure Buildup and Other Test Data, SPE 9289, presented at
the 1980 Dallas Annual Technical Conference and Exhibition
18. Odeh, A.S., Nabor, G.W.: The Effect of Production History on
Determination of Formation Characteristics from Flow Tests,
JPT, October 1966, pp1343-1350
19. Horner, D.R.: Pressure Build-Up in Wells, Proc., Third World
Pet. Cong., The Hague, Sec. II, 1951, p503-523
20. Odeh, A.S., Selig, F.: Pressure Build-Up Analysis, VariableRate Case, JPT, July 1963, pp790-794
21. Swift, S.C.: Application of Equivalent Drawdown Time in
Well Testing, SPE 17547, 1988
22. Spivey, J.P., Lee, W.J.: Effects of Flow Rate Variations and
Producing Time on Well Test Analysis, Harts Petroleum
Engineer International, January 1998, pp93-104
23. Cinco-Ley, H, Kuchuk, F., Ayoub, J.A., Samaniego-V., F.,
Ayerstaran, L.: Analysis of Pressure Tests Through the Use of
Instantaneous Source Response Concepts, SPE 15476, 1986
24. Cinco-Ley, H., Samaniego, F.: Use and Misuse of the
Superposition Time Function in Well Test Analysis, SPE
19817, presented at the 64th Annual Technical Conference and
Exhibition of the Society of Petroleum Engineers in San
Antonio, October 8-11, 1989
25. Ozkan, E., Raghavan, R., Estimation of Formation Damage in
Horizontal Wells, SPE 37511. Paper presented at the SPE
Production Operation Symposium, Oklahoma, March 9-11,
1997.
26. Yildiz, T. and Cinar, Y., Inflow Performance and transient
Pressure Behaviour of Selectively Completed Vertical wells,
SPE Reservoir Evaluation & Engineering, October 1998, pp.
467-468.
27. Bidaux, P., Whittle, T. M., Coveney, P. J. and Gringarten, A.
C., Analysis of Pressure and Rate Transient Data From Wells
in Multi-layered Reservoirs: Theory and Application,
SPE24679. Paper presented at 67th Annual Technical
Conference and Exhibition, District Columbia, October 4-7,
1992.
28. Brownscombe, R., Kern, L.R.: Graphical Solution of SinglePhase Flow Problems, The petroleum engineer, B-70, 1951

SPE 63077

29. Miller, C.C., Dyes, A.B., Hutchinson, G.A.: The Estimation of


Permeability and Reservoir Pressure from Bottomhole Pressure
Buildup Characteristics, JPT 2, pp91-104, April 1950

Appendices
Appendix A: existing definitions of the radius of
investigation
This part is the extension of Van Poolens article in 19641,
in order to summarize the different functions obtained and the
condition used for each of them. In order to obtain a
comparable form for all of these equations, they will all be
expressed in their dimensionless form reD=D.tD0.5.
Drawdown-based equations
The first definitions have been developed taking into
account the propagation of the signal in the reservoir and the
establishment of a steady-state flow-regime. The equation
obtained, i.e. the value of D, is largely depending upon the
condition chosen.
Jones5 considered that the radius of drainage is the
distance at which the pressure disturbance generated
during a drawdown reaches 1% of the pressure drop
at the wellbore1, obtaining a value of 4 for D. This,
however, is an approximation.
Tek, Grove and Poettmann6 had a similar approach to
Jones but considered the radius at which the radial
flow-rate equals 1% of the flow-rate at the wellbore.
They obtained a slightly larger value for D at 4.29.
Muskat7 considered the time for the flow-regime to
reach a steady-state in a bounded reservoir, relatively
to the portion of fluid removed from the reservoir. D
is there equal to 2. Chatas8 used the same approach
and extended it to linear flow, obtaining a value of
1.41 for D.
Van Poolen1 used the Y function defined by Jones5 to
deduce the time at which connected pore space has
been proven. He obtained D=2.
Johnson3 considered the radius of drainage as linked
to a fraction of the net production during a test. He
obtained a D coefficient of 2.81.
Kutasov and Hejri4 generalised the value of D as a
function of the dimensionless time at which the
radius of drainage is calculated. They expressed that:
D=D0.tDn and obtained values for D ranging between
2.03 and 2.14.
Build-up-based equations
Brownscombe and Kern28 focused on stabilisation
time and considered the time needed for the transients
to become negligible. They assumed that the
stabilization time is the time when the reservoir is
within 2% of equilibrium. They obtained D=1.783.
This approach has been extended to two-phase flow
by Miller, Dyes and Hutchinson29.
Hutchinson and Kern used a differential network to
derive the same problem and ended with D=1.5
Hurst, Haynie and Walker10 considered the graphical
end of the match between build-up curves and the

Author
Jones
Tek, Grove And
Poettmann
Muskat
Chatas (linear)
Van Poolen
Johnson
Kutasov and Hejri
Brownscombe and Kern
Hutchinson and Kern
Hurst, Haynie and Walker
Hurst
Suggestion

D
4
4.29
2
1.41
2
2.81
2.03-2.14
1.783
1.5
2.6408
2.8284
0.379-1.623

Appendix B: condition of equation 4


Equation 2 gives the expression of the dimensionless
pressure drop using a homogenous reservoir bounded by a
sealing fault. Two stabilizations are observed in this case on
the derivative: a first one for a value of corresponding to the
radial flow before the fault appears and a second one for a
value of 1 corresponding to the second radial flow accounting
for the fault. The condition chosen is that the second
stabilization must reach 90% of its final level on the
logarithmic plot. Assuming that the first stabilization is fully
developed Equation 2 becomes after derivation:

which finally becomes Equation 4

1
2
d E i (d D )
0.9
tD
1


1
d ln t
2

Appendix C: total skin factor calculation


CASE I: s1 = s 2
sT = s1 = s 2
CASE II: s1 s 2
=0
sT = (1 )s1 + s 2
0
s12 = s1 s 2

If

sT =

s1 = no flow
A + (B + Ds 2 )

1
If s 2 = no flow
A + ([1 ]B + Ds1 )
sT =

If s1 = s 2 = open to flow
A + ([1 ]Ds1 B + Ds1 B + Ds1 Ds2 )
sT =
B[1 ]Ds1 + Ds2
Parameters A, B, C, Ds1 and Ds2 are as defined in table 2

Appendix D: Skin distribution models used


G9

-3

-2.5

dPD
log
d ln t D
log (1 / 2)

Layer
-1.5

-1

-4.5

-4

-3.5

-3

-2.5

-2

-1.5

-1

-0.5

3
4
5

7
8
G9
9

7
8

G10

10
10

11

11

12

12

13
14

13

15

14
15

16

System G9

System G10

G11

G12

Layer
-6

-5

-4

-3

-2

Layer

-1

-8

-6

-4

-2

6
Skin Factor

7
8
G11
9

7
8
G12
9

10

10

11

11

12

12

13

13

14

14

15

15

16

16

System G11

System G12

G13
G14

Layer
0

10

20

30

40

50

60

Layer

70
0

10

20

30

40

50

60

70

1
1

2
2

3
3

4
4

Skin Factor

7
8
G13
9
10

7
G14

8
9

11

10

12

11

13

12

14

13

15

14
15

16

System G18

0.1

-0.5

The condition can be expressed as:

Replacing the stabilized derivatives by their values yields:

-2

1 2
dEi rD
1
dPD
tD
= 1

d ln t D 2
d ln t D

dPD

dPD
log
log
d ln t

ln
d
t
nd

D
D
stab

2
1 0.9
dPD

dPD

log

log

d ln t st
D 1 stab
d ln t D 2 nd stab

G10

Layer
-3.5

Skin Factor

Summary

thus, replacing the derivative by its expression

1 2
dEi rD
0.1
1
tD 1
1
2
2
d ln t D

Skin Factor

steady-state regime. A numerical solution gave


D=2.6408. Later, Hurst2 developed an analytical
solution for this problem, obtaining an approximate
value of 2.8284 for D.
Ishteiwy and Van Poolen11 developed a method to
obtain a graphical estimate of the radius of drainage
based on the deviation time from the infinite model,
considering a boundary.
Gray9 published a method to approximate well-tofault distance from pressure build-up tests by a
graphical mean. This method uses a trial-and-error
approach to obtain the distance and didnt generalize
it to obtain a radius of drainage equation as usually
known.

Skin Factor

FREQUENTLY ASKED QUESTIONS IN WELL TEST ANALYSIS

Skin Factor

SPE 63077

System G19

10

S. DAUNGKAEW, F. HOLLAENDER, A.C. GRINGARTEN

G15

G16

Layer
30

50

60

70

15
14

13

12

11

5
Skin Factor

10

6
7
Skin Factor

16
40

9
G16
8
7

G15
9

10

11

12

13

14

15

10

20

30

16

60

70

G23

Layer

Layer
0

0.5

1.5

2.5

3.5

4.5

6
Skin Factor

7
8
G17
9
10

7
8
G23
9
10

11

11

12

12

13

13

14

14

15

15

System G22
1

G25
Layer
5

0
1

7
8

G24

Skin Factor

7
8
G25
9
10

10

11

11

12

12

13

13

14

14

15

15

16

System G24

System G25

G26

G27
Layer

16
0

14

13

12

11

10

15

20

25

10

9
G26
8

Skin Factor

Skin Factor

15

7
6

7
8
G27
9
10

11

4
3

12

13
14

1
0

15

16

Layer

System G26

System G27

Tables and figures


Table 1- Reservoir Parameters

(pav)i
C

5000 psi
0.01 bbl/psi

ki(xy)
ki(z)

200 mD
20mD

Table 2- Equations used to calculate the total skin factor in a


multi-layered model.
s1 < 0 and s1 no flow and s 2 < 0 and s 2 no flow and
Else
(s12 < 0 or s2 = no flow)
(s12 > 0 or s 2 = no flow)

Ds1 = 0
Ds 2 = s12

Ds 2 = 0
Ds1 = s12

Ds1 = s1
Ds 2 = s 2

A = s1

A = s2

A=0
C =

C = e

2 s1

K0
B=

C = e
C
(1 )

C
C
K1
(1 )
(1 )

2 s2

10

Eq. 5

101

Eq. 1

10

-3

-2

-1

10

10
10
10
10
Elapsed time (hours)
Fig. 1- Derivative responses for a sealing fault for test durations
given by Eqs. 1 and 5
10 3

System G23

G24
Layer

10

10

16

16

Skin Factor

50

System G21

G17

Skin Factor

40
Layer

System G20

Pressure change and derivative

20

Pressure change and derivative (p Pressure change and derivative (p

10

10

10 1

10 0
10

-1

10

-3

10

-2

-1

10
10
10
10
Elapsed time (hours)

10

Fig. 2- Influence of noise on the derivative, amplitude 0.01 psi


3
10
10

10

10

10

-1

10

-3

10

10

10

10

10 -3

10

-2

-1

10
10
10
10
10
Elapsed tim e (hours)
Fig. 3- Influence of noise on the derivative, amplitude 0.05 psi
Pressure change and derivative (ps

SPE 63077

10 -2

10 -1
10 0
10 1
Elapsed time (hours)

10 2

Fig. 4- Influence of coherent noise (earth-tide) on the derivative

10

Build-up
Drawdown
10

10

10 -3

10-2

10 -1
100
10 1
Elapsed time (hours)

10 2

10 3

Oil flow -rate

30

20

Build-up
Drawdown
10

10

1
-3

-2

10

-1

10

1x104

2x104
3x104
Dimensionless time tD/CD

4x104

0.5
Correct derivative

0.3
0.2
0.1

10-1

100

101

102

Build-up
Drawdow
10

10

10

-3

10

-2

-1

Oil flow -rate

30

20

tPe only
10

1x104

2x104
3x104
Dimensionless time tD/CD

4x104

20%+tPe

40%+tPe

0.5
60%+tPe

0.3
0.2

-1

10

10

10

10
10
10
Elapsed time (hours)

10

10

10

tPD=4x10
50

5
3

tPD=4x102

30

tPD=4x10

20

10

1x104

2x104
3x104
Dimensionless time tD/CD

4x104

0.5
0.3
0.2

10-1

100

101

102

103

104

105

0.1
Dimensionless time tD/CD

Fig. 11- Effect of producing time on the localization of errors


1

30

0.8

20

20% of QCum

10

1x10

40% of QCum

2x10
3x10
Dimensionless time tD/CD

4x10

60% of QCum

0.5
0.3

80% of QCum
-1

10

10

10

10

0.1

100% of QCum
3

10

10

10

3
2

Oil flow -rate

Oil flow -rate

10

Dimensionless time tD/CD

0.2

10

Fig. 10- Use of part of the rate history along with tPe

Dimensionless pressure drop and derivaitve

10

0.1

50

105

50

40

104

40

Dimensionless pressure drop and derivaitve

10

103

Fig. 9- Effect of the use of Horners equivalent time

40
40

Fig. 7-Influence of drawdown duration (Anoise=0.5psi)

Dimensionless pressure drop and derivaitve

Horner equivalent time

10

Oil flow -rate

Pressure change and derivative

10
10
10
10
10
Elapsed time (hours)
Fig. 6- Effect of noises on the extra amount of interpretable data
(Anoise=1psi)
4
10

Pressure change and derivative

40

Dimensionless time tD/CD

Fig. 5- Interpretable build-up showing more than the preceding


drawdown (Anoise=0.5psi)
4
10

10

11

50

Dimensionless pressure drop and derivaitve

10

FREQUENTLY ASKED QUESTIONS IN WELL TEST ANALYSIS

Dimensionless pressure drop and derivaitve

Pressure change and derivative (psi)

SPE 63077

0.6

0.4

100%

0.2

1x10

2x10
3x10
Dimensionless time tD/CD

4x10

60%+tPe

40%+tPe

0.5
20%+tPe
0.3
0.2
0.1

tPe only
10-1

100

101

102

103

104

105

Dimensionless time tD/CD

Dimensionless time tD/CD

Fig. 8- Effect of rate truncation

Fig. 12- Effect of imperfect rate knowledge, interrupted rate

12

S. DAUNGKAEW, F. HOLLAENDER, A.C. GRINGARTEN

SPE 63077

20

Oil flow -rate

1400
10

1200

3
5x103

1x104
1.5x104
Dimensionless time tD/CD

tPe only
20%+tPe

2x104

1000
Flowrate (Mscf/D)

Dimensionless pressure drop and derivaitve

15

40%+tPe

1
0.5
0.3

800

600

400

100%

0.2

10-1

100

101

102

103

104

200

105

0.1

Dimensionless time tD/CD

Fig. 13- Effect of imperfect rate knowledge, linearly decreasing


rate history

40000

80000

120000

120000

Fig. 17-Real field example: rate history


80

60

40

30

70

Complete rate history

20

Normalised pseudo pressure (psia)

Oil flow -rate

Dimensionless pressure drop and derivaitve

50

10

1x104

2x104
3x104
Dimensionless time tD/CD

4x104

20%+tPe

tPe only

40%+tPe

1
0.5

40%+tPe

60

30%+tPe

50

10% difference with the


correct interpretation

40

0.3
100%
0.2

10-1

100

101

102

103

104

105

30
260

0.1

280

300

320

Dimensionless time tD/CD

340

360

380

Superposition function (Mscf/D)

Fig. 14- Effect of imperfect rate knowledge, interrupted linearly


decreasing rate

Fig. 18-Real field example: Horner plot


2

20

100%

10

2
4

1x10

2x10
3x10
Dimensionless time tD/CD

4x10

60%+tPe
1

0.5
20%+tPe

tPe

0.3
-1

10

10

0.1

10

10

10

10

10

Dimensionless time tD/CD

Fig. 15- Effect of imperfect rate knowledge, linearly increasing


rate history

40
Oil flow -rate

Dimensionless pressure drop and derivaitve

30

tPe only

20

10

1x10

2x10
3x10
Dimensionless time tD/CD

4x10

20%+tPe

40%+tPe

0.5
60%+tPe

0.3
0.2

40%+tPe

10
-1
10

10

10
Elapsed time (hours)

10

Fig. 19-Real field example: log-log plots


102

50

Complete rate history

10

40%+tPe

0.2

30%+tPe

Pressure change and derivative (psi)

Oil flow -rate

Dimensionless pressure drop and derivaitve

30

Pressure change and derivative (psi)

10

40

-1

10

10

10

10

100%
3

10

10

0.1
Dimensionless time tD/CD

Fig. 16- Effect of a simplification of the rate history

10

G12

10

10

G10

G11

G9

10-1
100
101
Elapsed time (hours)
Fig. 20- Log-log Plots of Systems G9 to G12
10-4

10-3

10-2

102

103

SPE 63077

FREQUENTLY ASKED QUESTIONS IN WELL TEST ANALYSIS

102
G20
G22

1
G21

100 -2
10

-1

10

10

102

10

10

Elapsed time (hours)


Fig. 21-Log-log plots of System G18 to G22
Pressure change and derivative (psi)

10

102
G23 G24

G25
G26

G27

10

100 -3
10

-2

10

-1

10
10
10
Elapsed time (hours)

10

10

Pressure change and derivative (psi)

10

Channel at 50ft
Channel at 25ft
Channel at 500ft

Channel at 2000ft

10

100 -3
10

10-2

10-1
100
101
Elapsed time (hours)

102

103

Pressure change and derivative (psi)

Fig. 23- Influence of a boundary: pressure plots in the case of a


conical skin distribution
103

Channel at 25ft

10

Channel at 50ft

Channel at 500ft

10

10 -3
10

Channel at 2000ft

10-2

10-1
100
101
Elapsed time (hours)

Channel at 25ft

Channel at 50ft
Channel at 500ft

102

101

100 -2
10

-1

10

10

10

10

10

Channel at 25ft

Channel at 50ft

10

Channel at 500ft
Channel at 2000ft

10

10 -2
10

10-1

100

102

101

103

Elapsed time (hours)


Fig. 26- Influence of a boundary: derivative plots with an
alternating change skin distribution
4
10

Fig. 22-Log-log Plots of System G23 to G27

102

10

Elapsed time (hours)


Fig.
25-Influence of a boundary: pressure plots with an
alternating change skin distribution
103
Pressure change and derivative (psi)

10

102

103

Fig. 24- Influence of a boundary: derivative plots in the case of a


conical skin distribution

Pressure change and derivative (psi)

G18 G19

Pressure change and derivative (psi)

10

103

Channel at 25ft
Channel at 50ft
Channel at 500ft

102

10 -2
10

-1

10

10

102

10

10

Elapsed time (hours)


Fig. 27- Influence of a boundary: pressure plots with a limitedentry well
4
10
Pressure change and derivative (psi)

Pressure change and derivative (psi)

13

Channel at 25ft
Channel at 50ft

10

Channel at 500ft

10

1
-2
1010

Channel at 2000ft

-1

10

10

10

102

10

Elapsed time (hours)


Fig. 28- Influence of a boundary: derivative plots with a limitedentry well

You might also like