SPE 102483

The Use of Well Testing for Evaluation of Connected Reservoir Volume

M.M. Levitan, SPE, and M.J. Ward, SPE, BP plc.; J.-L. Boutaud de la Combe, SPE, Total S.A.; and M.R. Wilson,
Well-Test Solutions Ltd.

Copyright 2006, Society of Petroleum Engineers

studying the dynamic pressure behavior in response to
This paper was prepared for presentation at the 2006 SPE Annual Technical Conference and appropriately designed sequence of well rate changes provides
Exhibition held in San Antonio, Texas, U.S.A., 24–27 September 2006.
a way to evaluate some of these properties. This technique
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
has been historically used for evaluation of formation
presented, have not been reviewed by the Society of Petroleum Engineers and are subject to permeability, large-scale reservoir heterogeneities and
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 boundaries, reservoir connectivity, well productivity, and for
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
diagnosing possible well productivity problems.
for commercial purposes without the written consent of the Society of Petroleum Engineers is Depending on test objectives, a well test may last from
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 several days to several weeks and even months. The longer
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.
the test, the larger is the reservoir volume investigated during
the test. When used for exploration and reservoir appraisal,
Abstract these tests normally involve flowing hydrocarbons to surface
In its search for new oil and gas reserves the oil industry and disposal of some of these hydrocarbons through flaring.
moves to more and more remote areas of the world and to Completion of a reservoir appraisal well and subsequent well
technically challenging areas of deep water. Development of testing is a long and very expensive operation which carries
hydrocarbon resources in these environments is extremely significant operational risks. A decision to perform a well test
expensive. To be economically viable, the newly discovered and accept these costs and risks should only be supported if
fields must be developed and effectively exploited with very the test will provide the information that is critical to shaping
few wells. This forces the oil companies to concentrate on the appropriate reservoir development plan.
high quality reservoirs that yield highly productive wells with There are alternative techniques for evaluation of
large reserves per well. High costs prohibit extensive formation rock properties through well logging and the use of
appraisal activity and drive development decisions based on modern formation evaluation tools. These methods allow
very few wells. Whilst these limited penetrations are often estimation of the short-term productivity of the well.
logged extensively using modern formation evaluation tools, However, these techniques cannot confirm that the well is
the acquired static data cannot confirm that the wells will connected to a sufficiently large volume and will be able to
drain sufficient reserves. Evaluation of reservoir connectivity drain sufficient reserves. Well testing still remains the only
over large distances from the well requires relatively long and method for direct evaluation of reservoir connectivity over
expensive well tests. large distances from the well.
In this paper we review a number of critical points Development of hydrocarbon resources in deep water
associated with design and execution of such tests. The key environment is extremely expensive. To be economically
here is to ensure that the data acquired during the test will viable, the newly discovered fields must be developed and
contain sufficient information to draw conclusions about the effectively exploited with very few wells. This forces the oil
reservoir connectivity and to estimate the reservoir volume companies to concentrate on high quality reservoirs that yield
that is in communication with the well. We discuss the well highly productive wells with large reserves per well.
test sequences used for this purpose, the tools and operational Assessment of reservoir connectivity becomes one of the main
aspects of well test execution, the data acquisition, the well objectives in appraisal of such fields. An appropriately
test analysis techniques, the accuracy and the degree of designed and executed well test that confirms good reservoir
uncertainty of test results. We illustrate this application of connectivity may potentially decrease the total number of
well testing for reservoir connectivity with several real test wells required for appraising the field and reduce the overall
examples. costs of the reservoir appraisal program.

Introduction Reservoir Connectivity

Well testing is one of the techniques used for reservoir and There are several ways to characterize reservoir connectivity.
well evaluation. Well testing studies dynamic reservoir Probably, the most useful and clear description is in terms of
behavior in response to changing flow conditions at the well. the reservoir volume connected to a well. In some cases, well
The dynamic reaction of well bottomhole pressure to rate test results may indicate that the well is connected to a closed
changes depends on the reservoir and well properties. Hence, reservoir compartment. In this case the volume of the
2 SPE 102483

compartment is determined from the well test results. If a volume. It reflects combined effects of fluid and rock
relatively short duration well test yields this type of result, compressibility:
then it is likely that the reservoir compartment volume is
relatively small and the reservoir may have a connectivity
c t = c r + c o s o + c w s w ................................................(3)
problem. This would be an unfortunate and undesirable result
of the test. However, such a result is critically important for
assessment of commerciality and quantifying potential The main uncertainty in this characteristic is associated with
downside risks associated with the development proposals. the rock compressibility c r . Rock compressibility may
A more encouraging and promising conclusion would be
become the main component of total compressibility in the
when the test results indicate that the pressure behavior is
case of oil reservoirs and soft or unconsolidated rock
characteristic of a homogeneous open reservoir. The well test
formations. This is less of a problem for gas reservoirs
in this case can be characterized in terms of its radius of
because the total compressibility is mostly controlled by the
investigation. The radius of test investigation in this case also
gas compressibility component. The uncertainty range for c t
implies reservoir connectivity within at least this distance
from the well. in this case is much smaller.

Radius of well test investigation. The fluid production and Connected reservoir volume. Well tests often produce
the changes in well operating conditions during a well test pressure behavior that may be interpreted as that caused by
form a pressure field in the region around the well that causes reservoir heterogeneities and no-flow boundaries while at the
the fluid to flow towards the well. The size of the region same time still indicating open reservoir behavior. For
affected by induced pressure changes increases with time. example, a well may be completed in a reservoir (or a part of
Since we can monitor only the bottomhole pressure in the well the reservoir) constrained by two parallel no-flow boundaries.
and not the pressure field in the formation, we estimate the If these boundaries are close enough to the well to affect the
size of the affected region from the pressure measurements in pressure during the test they will produce the pressure
the well. There is a simple formula that relates time to the transient behavior characteristic of parallel boundary
radius of this region. This expression is valid only for the case configuration. While affected by these boundaries, the
of a homogeneous reservoir with no boundaries. It defines the pressure during the test will still reflect the open character of
so-called radius of well test investigation: this boundary configuration. The concept of well test radius
of investigation cannot be applied in this situation. However,
it is still possible to define a volumetric characteristic that
k quantifies reservoir connectivity. This characteristic is the
Rinv = 4η Δt = 4 α Δt . ................................ (1)
φ μ ct "minimum connected reservoir volume" supported by the test
results. The minimum connected reservoir volume is
determined based on the minimum sized closed reservoir
The Δt here is the duration of a flow period during the test model that reproduces all the relevant pressure behavior
when the pressure data indicate there is homogeneous during the whole well test sequence.
reservoir behavior. According to this expression, the radius of Let us consider the above example of a reservoir in the
well test investigation depends on the reservoir formation and form of a channel constrained by two parallel boundaries. We
fluid properties. The mobility k μ is normally defined by the can reproduce this test pressure behavior with an open
test data. However, the remaining parameters in Eq. 1 (the reservoir model that includes only these two boundaries. It is
porosity φ , and total compressibility c t ) are obtained from also possible to reproduce the same pressure behavior with a
closed-reservoir model. This model would include two
other measurements and uncertainty in these characteristics additional boundaries that truncate the channel on both sides
affects the value of the radius of investigation estimate. of the well. As long as the two truncating boundaries are
The radius of investigation can also be presented as the sufficiently far away from the well, this closed reservoir
volume of fluid within the region investigated by the test: model will still be consistent with the test data. We can
gradually reduce the size of this model by moving the
kh s o truncating boundaries closer to the well until the first signs of
Vinv = π Rinv
2
h φ s o = 4π α Δt . ......................... (2) a discrepancy between the test data and the model prediction
μ ct
appear. This then defines the minimum-size closed reservoir
The fluid volume defined by Eq. 2 could be viewed as the model that is consistent with the test data. The fluid volume in
connected reservoir volume supported by the test. The this model defines the minimum connected reservoir volume
parameter combination kh μ is defined by the transient that is supported by the test results. We should note that this
pressure data during the test. Hence, the parameters that affect is a conservative estimate of the true reservoir volume
the connected volume estimate which are not defined by the connected to the well. The actual connected reservoir volume
test data are the oil (or gas) saturation and total may be much larger.
compressibility. Total compressibility is probably the main
source of uncertainty in quantifying connected reservoir
SPE 102483 3

Considerations for a Well Test to Prove Connected The PBU pressure behavior could indeed be produced by a
Reservoir Volume homogeneous reservoir model. However, it could also be
Here we discuss the main considerations that go into the produced by a variety of heterogeneous reservoir models.
design of a well test that provides the information necessary For example, the initial pressure derived from a well-test
for assessing reservoir connectivity. The purpose is to analysis is normally compared to an independent static
optimize the well test sequence, i.e. to minimize the test measurement of this value (e.g. via a wireline formation
duration and the volume of hydrocarbons produced and still be logging tool). The initial-pressure from the well-test derived
able to prove the connected reservoir volume targeted for the value is considered to be acceptable when it is within the
test. accuracy of the independent measurement. However, while
If a well in a limited-size reservoir compartment is this comparison appears to be a straight-forward, there is a
produced at constant rate then the well bottomhole pressure degree of uncertainty in the initial-pressure obtained from a
will eventually develop a decreasing linear trend on the well-test analysis which comes from the extrapolation of the
Cartesian plot of bottomhole pressure vs. time. This flow type-curve match back to the initial conditions of the system.
regime is called the pseudo steady-state regime. The rate of The range of these extrapolations depends on the magnitude of
pressure decline during pseudo steady-state flow is inversely the pressure changes induced into the reservoir by the test.
proportional to the size of reservoir compartment drained by When the range in the extrapolation has a similar magnitude to
the well. This serves as the basis for a so-called reservoir the accuracy of the independent initial-pressure measurement,
limits test. The time to the onset of this regime depends on the then the well-test can fail to provide any useful information
volume of reservoir compartment and its areal shape. In most about reservoir connectivity.
cases the time to the onset of this regime is prohibitively long. This can be demonstrated by calculating the PBU response
Also, it is difficult to confirm establishment of the linear in a simple unbounded homogeneous reservoir. The following
pressure trend and distinguish it from the preceding pressure case is based on a test sequence comprised of a 100 hour
transient regime. All these factors make reservoir limits constant rate flow followed by a 24 hour shut-in. The initial
testing impractical, expensive, and an environmentally pressure of the system is 5000 psi which is known to an
unfriendly testing option if produced hydrocarbons have to be accuracy of ±5 psi. For simplicity, the PBU is shown as a de-
flared. This is why reservoir limits or drawdown/production convolved "response curve" i.e. the shape of the type-curve
tests alone are not often used for reservoir appraisal. model required to match the pressure data and then extrapolate
Likewise, pressure data from a single PBU test does not to a specified initial-pressure. Shown in Fig.1 are two de-
have enough information to identify reservoir boundaries and convolved response curves for a test run at a low production
heterogeneities1 and cannot be used for evaluation of reservoir rate that imposes small pressure changes on the reservoir. The
volume. Even in the trivial case when the pressure buildup initial-pressure specified for each de-convolution corresponds
data exhibit so-called infinite homogeneous reservoir behavior to the ±5 psi accuracy of the known 5000 psi value. Each de-
manifested by a horizontal pressure derivative on the convolved curve initially shows the expected homogeneous
derivative plot, the type-curve model responsible for this behavior, but beyond that time the response reflects an
pressure behavior is not uniquely defined by the single PBU. extrapolation back to the specified initial pressure.
RESPONSE FUNCTION (PSI/STB/D)

.02
.02
10-2

Pi = 5005 psi
.02
10-3

.02
Pi = 4995 psi
10-4

10-3 10-2 10-1 100 101 102

Delta-T (hr)

Fig.1 - Derivative plot for a low rate test showing that heterogeneous models can match the PBU response for a homogeneous system given a
small uncertainty in the initial-pressure.
4 SPE 102483

The different extrapolations have the appearance of either a balance considerations. The objective of a test is to extract a
reservoir boundary (for the higher initial-pressure) or a certain volume of fluid from the reservoir and to measure the
permeability-thickness that increases away from the well (for change in reservoir pressure resulting from this production.
the lower initial-pressure). While this shows the initial Given the compressibility of reservoir rock-fluid system and
pressure from the PBU analysis agrees with the independent using the material balance equation, one can translate the
value within its measured accuracy, the analysis does not cumulative volume produced and the resulting reservoir
provide any useful information about reservoir connectivity pressure decline into the reservoir volume connected to the
because the range of possible matches covers both bounded well. One can see that this approach is based on the
and open reservoir models. assumption that the fluid production during the test will result
The above case is not entirely hypothetical. The rates used in some reservoir pressure decline and that this reservoir
in a well-test can be limited for a variety of reasons; sand- pressure decline can be unambiguously identified from the
control may create a high skin-factor limiting well bottomhole pressure data measured during the test.
productivity, flaring limits may constrain the rates, avoiding Normally, we design a test to prove a specific reservoir
two phase flow by keeping the reservoir flowing pressure volume. This would require an appropriate volume of fluid to
above the bubble/dew point pressure of reservoir fluid, etc. be produced and as a result, we plan for some specific
While high resolution gauges capture sufficiently detailed reservoir pressure decline caused by this production. If after
pressure data at low rates for the analysis, the test design for performing a well test we determine that the reservoir pressure
this situation must ensure all other data is also captured at a declined more than was expected, then this is an indication
high resolution. For example, in the above case the accuracy than the connected reservoir volume is smaller than the
of the independent initial pressure measurement would need to volume the test was designed to prove. If on the other hand
be reduced significantly below ±5 psi by doing a detailed the reservoir pressure declined less than was expected, then
gauge calibration between the wireline formation tool and the reservoir volume proven by the test is actually larger than
well-test pressure gauges, or by modifying the test sequence to it was designed for. Finally, if the test data show no reservoir
have the well-test gauge also record the initial pressure pressure decline, this is an indication of infinite reservoir
immediately after perforating the well. behavior during the test, and that the reservoir volume proven
This example of initial-pressure sensitivity is but one issue by the test can be quantified in terms of the minimum
related to designing a test to obtain an unambiguous reservoir connected reservoir volume discussed earlier.
"signal" over the "noise" in the measurements. Carrying out a
well-test to determine reservoir connectivity requires a The Well Test Sequence. We start with the premise that the
complete look at all of the data that will feed into the analysis, test objectives clearly define the reservoir volume that has to
and ensuring the test is configured to provide a response that is be confirmed by the test. In order to prove this reservoir
greater than all of the measurement errors. volume, an appropriate volume of fluid has to be extracted
The well test sequences that are used in exploration and from the reservoir during the test in order to cause a reservoir
appraisal well testing normally include several production and pressure decrease that could be reliably determined from the
pressure buildup periods. The reasoning behind a test test data. We normally evaluate the reservoir pressure by
sequence tailored for reservoir appraisal is based on material performing a pressure buildup test.

Clean-up flow
Final PBU
5000.

Initial PBU

Main Flow
pressure PSI
4800.
4600.
4400.

0. 50. 100. 150.

2000. 4000.
rates STB/D
-1000.

0. 50. 100. 150.

Time (hours)

Fig. 2 - Typical well test sequence used for evaluating reservoir volume.
SPE 102483 5

Hence, in order to determine the reservoir pressure change Test Design. Well test objectives determine the design of the
caused by the production targeted for the test we need to test. Normally, the primary objective of a test aimed at
include in the test sequence two pressure buildup tests; one evaluation of reservoir connectivity is formulated in terms of
before the main flow period and another immediately after the the reservoir volume Vt that has to be proven or confirmed by
main production. As a result, the well test sequence required the test. Well test design also relies on some assumptions and
for this test is shown in Fig. 2. Here we assume that the well expectations of the reservoir and fluid properties. Normally
test is performed immediately after the well is completed. The these properties are estimated from reservoir well logs, cores,
test starts with a Clean-up or Initial flow period followed by MDT tests, the well tests performed on other wells on the
Initial PBU. After the Initial PBU we have the Main flow same structure, or from geological analogs. Interpretation of
period followed by the final PBU. This is the most common seismic information may point to possible no-flow boundaries
test sequence used in exploration and appraisal well testing. in the vicinity of the well. All this information is taken into
The purpose of the Initial flow period is to remove account in setting up a model that is used in well test design.
completion fluid from the wellbore and to prepare the The main questions that have to be addressed from the
wellbore and the reservoir for the subsequent test. The start are the total volume of fluid to be produced during the
volume of fluid produced during this period should be small Main flow period, the duration of this flow period, and the
so it does not have noticeable effect on reservoir pressure. At duration of the Final PBU. We decide on the cumulative
the same time, the clean up volume and flow rate should be volume to be produced based on the target for reservoir
sufficient to remove the completion fluid and have only pressure decline that must result from this production. Let us
reservoir fluid in the wellbore by the end of this period. The assume that based on previous experience with the well tests
Initial PBU is intended to provide the information on the performed in reservoirs with similar geology it is possible to
initial reservoir pressure as well as the reference transient identify the reservoir pressure decline of Δp from the
pressure data that are used in well test analysis.
bottomhole pressure data measured during the two pressure
The key point is to have a valid reference transient
buildups before and after the Main flow period. Then to
pressure response. Thus, it is better to be certain the well has
generate this reservoir pressure decline in a reservoir of the
been sufficiently cleaned-up, even though this requires a
higher volume of fluid to be produced, and even if the data volume Vt we have to extract the volume of fluid
from the initial PBU does not yield the exact initial pressure.
Today, it is more likely that the initial pressure will come from ΔV = c t Vt Δp . ............................................................(4)
a formation evaluation tool which can be used to guide the
analysis of the Initial PBU.
The purpose of the Main flow period is to produce the Note that the Vt here is the reservoir volume that has to be
fluid volume targeted for the test. The Final PBU provides the proved by the test according to the test objectives. The fluid
main pressure transient data used in well test analysis. It also volume to be produced during the Main flow period is
provides the information on the final reservoir pressure after estimated here based on the volume to be proved by the test
the main production. and on the reservoir pressure decrease targeted for the test.
Together, the Initial and the Final PBUs contain the Next, based on the expected reservoir and fluid properties, on
information on the reservoir pressure decline induced by the the type of well completion, and on the constraint on the
production during the Main flow period. The reservoir pressure drawdown allowed during the test we estimate the
pressure decline targeted for the test is generally small because safe production rate q w for the test. The duration of the Main
we try to minimize the fluid produced during the test since
some of this fluid will have to be flared. Hence, the success in flow period is then estimated as Δt = ΔV q w .
quantifying the reservoir volume proven by the test hinges on The duration of the Final PBU is based on the radius of
our ability to identify a small decline in reservoir pressure and investigation targeted for this flow period. Again, we estimate
unambiguously associate it with the fluid production during the radius of investigation based on the volume Vt that has to
the Main flow period. The problem is that some other be proven by the test. For example, in the case when there are
phenomena may affect the bottomhole pressure during the test no boundaries close to the well, the target for the radius of
and it is critically important to eliminate, minimize, or account investigation can be estimated from Eq. 1. This radius is
for these effects during the analysis. We will discuss these
achieved at the time Δt = Rinv2
4 η , where η is the diffusivity
effects that may compromise the test data later. At this point,
however, we would like to emphasize importance of paying coefficient. As a result, the duration of Final PBU period can
close attention to the characteristics of the pressure gauges be estimated according to
employed during the test and to the appropriate placement of
the gauges in the wellbore. For example, it is critically μ ct
important that the pressure behavior during the whole test Δt = Vt . ......................................................(5)
4π h s o k
sequence be recorded by the same pressure gauges which
remain downhole in fixed locations during the entire test
sequence. In a high-permeability formation, the rate of pressure build up
decreases rapidly with time during pressure buildup period.
6 SPE 102483

The pressure build up may become so slow during the later There are several important points in conjunction with this
part of PBU that it may not be possible to resolve the buildup well test analysis approach that require discussion.
trend by using modern gauges with their finite resolution of
about 0.01 psi. The problem may be compounded by gauge Material balance. Development of reservoir-well model that
drift which could overwhelm the pressure trend due to PBU. honors the pressure data acquired during a test is normally
It is therefore important to simulate the pressure behavior done with the help of modern well test analysis software. The
during the test sequence using a model that reflects the catalog of pressure transient models available in a software
reservoir properties and well completion and to evaluate the package should be extensive enough to cover a variety of
rate of pressure build up during the Final PBU period for a pressure transient behaviors associated with different types of
range of expected reservoir and fluid properties. In this reservoirs, reservoir heterogeneities, boundaries, and well
evaluation it would also be useful to have a model that completions that may be reflected by the test data. The critical
accounts not only for the transient pressure behavior prerequisite from the pressure transient models used for this
associated with fluid flow in the reservoir but also for the type of analysis is that they have to be sufficiently accurate
noise and the drift characteristics of the gauge and tidal and honor material balance. Software that does not honor
effects. If simulation does indicate that the gauges may not be material balance can not be used for the type of analysis
able to resolve the PBU pressure trend then the duration of the described in this paper.
Final PBU period estimated from Eq. 5 will have to be The pressure transient models supplied by most
reduced appropriately. commercial well test analysis tools normally honor material
balance in the liquid case when rock and fluid properties are
Well Test Analysis constant. Therefore, for oil reservoirs this is not generally a
The idea behind the use of well testing to prove connected problem. However, in the case of gas reservoirs where gas
reservoir volume is based on material balance considerations: properties depend on pressure, the fluid flow problem
(1) produce a volume of fluid that for the reservoir volume becomes non-linear. The standard approach in this case is to
targeted to be proved by the test would create the approximately linearize the problem by using the so-called
corresponding decrease of reservoir pressure that would be pseudo-pressure transform. Unfortunately, this method does
possible to identify and quantify from the well test results, (2) not ensure material balance. The way to enforce material
quantify the reservoir pressure change caused by the fluid balance for gas problems is by using the so-called material
production during the test, and (3) translate this reservoir balance correction.4 Hence, only the well test analysis
pressure change into the connected reservoir volume. While software that incorporates this correction is suitable for the
well test design evolves around these considerations, analysis application of this type of analysis to gas well tests.
of well test data does not normally follow this sequence
directly. The reason is that the pressure acquired during a well The mechanics of model calibration. In well test analysis,
test is the bottomhole pressure measured at the well and there the goal is to develop a simple reservoir model that honors all
is no simple and easy way to translate the bottomhole pressure the relevant static and dynamic data acquired during reservoir
into the average reservoir pressure needed to compute a appraisal. A simple model allows us to identify and
reservoir volume. understand the main reservoir characteristics that are
There is an alternative approach to achieve the same responsible for the reservoir behavior observed during the test.
objective of quantifying the reservoir volume proven by the Development of a model generally starts with evaluation of
test. With this approach there is no attempt to directly the character of the pressure transient behavior during pressure
translate the test bottomhole pressure data into the reservoir buildup periods on diagnostic derivative plots. By comparing
pressure change. Instead, we develop a reservoir-well model derivative plots it is possible to identify the features that are
that honors all the relevant static and dynamic data acquired common for the two PBU's of the test sequence and identify
before and during the well test. The model is used to simulate the features of the transient behavior that are related to the
the fluid flow in the reservoir and predict the bottomhole reservoir properties. This helps to understand the reservoir
pressure. As such, the pressure predicted by this model can be characteristics that are “seen” by the PBU pressure data.
directly compared with the pressure data acquired during the These characteristics will have to be incorporated in the
test. The model is calibrated to accurately reproduce the reservoir-well model in order to honor the test pressure data.
pressure transient behavior during individual pressure buildup In setting up the reservoir-well model we also take into
periods. It is also calibrated to simultaneously reproduce the account the information on the type of the well completion
pressure behavior during the two pressure buildups of the well (vertical, horizontal, deviated, partial penetration), and on the
test sequence. As we remarked earlier, pressure buildup data possible locations of reservoir boundaries and faults indicated
somehow reflect the average reservoir pressure at the time of by seismic data. After an appropriate reservoir model is
PBU. By calibrating the model to simultaneously reproduce selected, the next step is to adjust the model parameters in
the pressure behavior during the two pressure buildups of the order to reproduce all the relevant pressure behavior during
test sequence, we ensure that the model correctly accounts for the test. This includes the transient behavior during the two
the reservoir pressure change caused by the production during pressure buildups of the test sequence as displayed on
the Main flow period between these two pressure buildups. superposition (semi-log) and derivative plots. In addition, the
Having calibrated the model against the relevant test pressure model also has to honor the pressure behavior during the
data, we can use this model to quantify the connected reservoir whole test sequence on the pressure vs. time Cartesian plot.
volume measured by the test. Most importantly, the model should simultaneously and
SPE 102483 7

accurately reproduce on this plot the pressure behavior during test results. There may be several and in some situations
the later part of the two PBUs. This will ensure that the model many models that are consistent with the reservoir geology
is consistent with the possible reservoir pressure depletion and the static reservoir data while at the same time honor all
caused by the production during the main flow period between the relevant dynamic pressure behavior during the test. Non-
these PBUs. uniqueness of well test interpretation is a fact of life and
should not be ignored. There is a natural tendency when
Pressure-rate deconvolution. The transient pressure behavior analyzing well tests to present just one answer, one
during the Final PBU provides the information on the interpretation and ignore other possibilities. A variety of
reservoir characteristics within the radius of investigation of possible well test interpretations reflects the degree of
this PBU. The pressure data during the whole test sequence, uncertainty in the estimates of reservoir properties inherent in
which is significantly longer, contains more information about the data. Therefore, we should not expect to have one value
reservoir. However, this information is hidden by rate for the connected reservoir volume that is considered as
variations during the test. Pressure-rate deconvolution2,3,5 “proven” by the test. This value may be defined within a
allows one to remove the effects of this rate variation and certain range by the test data and determination of this range is
reconstruct the characteristic drawdown pressure transient also a task of well test interpretation. Hence, we should
behavior on the time scale of the whole test sequence. This consider a variety of models that may be consistent with the
reconstructed drawdown response helps in identifying and test data and evaluate the corresponding range for the
calibrating the reservoir-well model on a larger time interval connected reservoir volume resulting from these test
that covers the whole test sequence. interpretations.
While recognizing in principle the non-unique nature of
Evaluation of connected reservoir volume. The connected well test interpretation, the goal is to design and implement a
reservoir volume supported by the test is estimated using the well test in such a way that it produces the data that contain
model calibrated to honor the well test data. We always the volumetric reservoir information we are looking for. A
attempt to reproduce the relevant test pressure behavior using test design based on material balance considerations discussed
the least complicated model. If as a result of this effort we in this paper is a way to achieve this objective.
end up with a closed-reservoir model, then the test has The following simulation experiment illustrates that it is
determined a result and the volume of fluid in the model is the possible to recover the connected reservoir volume even if for
connected reservoir volume supported by the test. some reason we end up with a wrong model while interpreting
If on the other hand we end up with an open-reservoir the test. This experiment illustrates that we can obtain a
model, then the test data do not have enough information to reasonably accurate estimate of connected reservoir volume as
define the true connected reservoir volume. In this case we long as the well test data contain this information and the
estimate the minimum connected reservoir volume that is model is calibrated to honor the relevant parts of the pressure
supported by the data. To estimate this volume we introduce record. Fig. 3 presents the simulated bottomhole pressure data
additional boundaries to close the model. If these additional for a well in a homogeneous reservoir constrained by two
boundaries are placed far enough from the well they will not parallel boundaries. The reservoir is open at the both ends of
affect the pressure behavior during the test and therefore this this channel.
closed-reservoir model will still honor the relevant pressure We use two different models to reproduce the simulated
behavior during the test. Next, we bring these additional test data and estimate the minimum connected reservoir
boundaries closer and closer to the well until the first signs of volume supported by the data. First, we use a homogeneous
discrepancy between the model prediction and the test data reservoir model that is consistent with the model used to
can be identified. This then defines the minimum size closed generate the data. The pressure buildup data show the
reservoir model that is consistent with the test results, and characteristic behavior associated with parallel boundary
hence the minimum connected volume supported by the test configuration. Therefore, the width of the channel is defined
results. by the data. In order to estimate the minimum volume
When adjusting the model to define the minimum supported by the data we truncate the channel from both ends.
connected volume, it is not possible to identify the exact As a result, we end up with a closed compartment 800 ft wide
boundary locations where the signs of deviation from the test by 5400 ft long. This closed-model is the minimum size
data begin to appear. There is subjective judgment involved in model consistent with the data. The original oil in place for
this determination, and this causes some uncertainty in the this model is 10.8 million barrels. For the second estimate of
resulting minimum connected volume. We usually make this the minimum volume supported by the data, we intentionally
determination by evaluating the match on several plots. We start with a wrong model. This time it is a linear composite
verify the accuracy of simultaneous match of the test pressure model that has two regions with different rock properties. It is
during the later part of the two PBUs on pressure vs. time possible to approximately reproduce the test pressure using
Cartesian plot. We also evaluate the match of the this model. Fig. 3 and Fig. 4 demonstrate the quality of the
deconvolved drawdown response on a derivative plot. And linear-composite match on a Cartesian plot of pressure data vs.
finally, we check the match of the final (and the longest) PBU time. Fig. 5 presents the match of the second PBU data on
on a derivative plot. derivative, and Fig. 6 presents the match of the deconvolved
drawdown response also displayed on derivative plot.
Non-uniqueness of interpretation. It is a well recognized fact
that well test analysis does not yield unique interpretation of
8 SPE 102483

4200. 4400. 4600. 4800. 5000.

pressure PSI
-1000. 2000. 5000. 0. 50. 100. 150.
rates STB/D

0. 50. 100. 150.

Time (hours)

Fig. 3 - Simulated test data (markers) for a well in a channel. Continuous line is the simulation using linear composite model.
4750. 4800. 4850. 4900. 4950.
pressure PSI

0. 50. 100. 150.

-1000. 2000. 5000.
rates STB/D

0. 50. 100. 150.

Time (hours)

Fig. 4 - The close-up view of the late-time PBU data in Fig. 3.

DP & DERIVATIVE (PSI/STB/D)
10-2 10-3 10-1

10-2 10-1 100 101 102

Delta-T (hr)

Fig. 5 - Model match of second PBU data using linear composite model
SPE 102483 9

RESPONSE FUNCTION (PSI/STB/D)

. 04

10-1
. 02

. 04

. 02
10-2 10-3

10-3 10-2 10-1 100 101 102

Delta-T (hr)

Fig. 6 - Model match of the drawdown response derived from the second PBU data

The resulting model is a rectangle 730 ft wide by 3470 ft long. affect the pressure data measured by the gauges. When the
Along the length of this rectangle, there are two regions: the pressure change caused by fluid production is small it is
first is 1215 ft long and the second is 2255 ft long. The well is difficult to distinguish it from the pressure changes caused by
located in the first region. The storativity characteristic these other effects.
( φ c t h ) ratio of the two regions is 2.15. Again, this match A bottomhole pressure gauge does not directly measure the
can be adjusted to define a minimum-size model that is reservoir pressure. It does not even measure the bottomhole
consistent with the test data. Note that we identified the pressure next to the perforations. It measures the pressure in
minimum-size model by gradually moving the most distant the wellbore at the location of the gauge. The pressure
model boundary closer to the well until the model solution measured by the gauge differs from the sand-face pressure (the
started to show signs of deviation from the test data at late pressure at the mid-point of perforated interval) by the
time as shown in Fig. 4 through Fig. 6. Assuming that the hydrostatic head of the column of fluid in the portion of the
porosity φ and the total compressibility c t in the two model wellbore between these two locations. If the fluid between the
gauge location and the mid-point of the perforations has the
regions is the same, the change in storativity is then a change same density during the Initial and the Final PBU's, then the
of reservoir thickness and we can estimate the initial volume hydrostatic head of this fluid column does not change and we
of oil for this model as 11 million barrels. This is about can make judgements about the reservoir pressure changes
2.5%larger than the minimum connected volume estimate based on the measurements obtained during these two pressure
resulting from the correct homogeneous model. Note that the buildups. However, if the well is not cleaned up sufficiently
purpose of this experiment was to assess uncertainty of the fluid within this column during the two PBUs may be
connected reservoir volume due to non-uniqueness of the test different and the measured pressure will be affected by the
interpretation. We used the same value of total corresponding change in hydrostatic head. The only way to
compressibility when we generated the synthetic test data and avoid this problem is to place the pressure gauges at the mid-
later when we analyzed the data to determine the minimum point of the perforations. Unfortunately, it is almost always
connected volume. The uncertainty of connected reservoir impossible to position gauges exactly at this depth and an
volume due to uncertainty of total compressibility is probably operational compromise is required. Frequently gauge carriers
much larger than the uncertainty of connected volume due to are positioned in the test string bottom-hole assembly, but
non-uniqueness of test interpretation. these can still be some distance from the perforated interval.
To minimize the possible interpretation errors that may creep
Effects that Could Compromise Well Test Data in from operational artifacts it is important to make every
Quality. attempt to position gauges as close to the perforated zone as
We need to identify and quantify reservoir pressure changes possible. The use of multiple gauges positioned with vertical
caused by production during the Main flow period of a well offsets provides additional back up information which
test sequence in order to translate this information into the provides an ability to understand fluid gradient variations
connected reservoir volume. The incentives to reduce well between these gauges. Optimizing the placement of pressure
test costs and pollution caused by flaring of produced gauges thus becomes critically important and should never be
hydrocarbons push us to minimize the fluid volume produced overlooked. Another potential source of gauge measurement
during the test. As a result, we have to work with small error can be caused by string dimensional changes. Various
reservoir pressure changes induced during a test. We estimate well test string designs exist and some of these designs will
the reservoir pressure changes based on the measurements inevitably require gauges to be positioned in a part of the
recorded by the pressure gauges located at the bottom of the string that may move due to thermal and pressure induced
well. The problem is that there are several physical length changes. In some cases, these movements could
phenomena in addition to fluid production that could also
10 SPE 102483

introduce additional artifacts on the gauge data that will be The above indications of potential interpretation error
very difficult to estimate or remove. Whilst the errors sources should help optmize the operational design of a test.
associated with hydrostatic column variation below the However, there are other operational elements associated with
lowermost gauge or string movement can not be exactly well testing that also should not be overlooked when acquiring
quantified, optimized gauge placement and string design will test data. Managing operations to minimize the introduction
minimize the uncertainty in the connected reservoir volume of unwanted or spurious signals is essential as is the need to
estimate obtained from the test analysis. carefully document all activities and data sets that may
When it is not possible to place a gauge carrier next to the provide relevant data for the detailed well test interpretation.
perforated interval, it is recommended that one gauge carrier Conducting a relatively "noise free" test and fully
be placed as close as possible to the perforations and a second documenting all activities and measurements may well help
gauge carrier be placed high enough above the first one. With limit the number of interpretation models that need to be
two gauge carriers, one can check if the fluid gradients at the considered during the analysis.
end of both PBUs are similar. When the gradients are
different, it means that fluids densities are different and Field Examples
comparison of pressure data from the two PBUs is Following are three examples of the well tests performed to
meaningless. While having similar fluid gradients from this evaluate long term production potential of the wells. The first
gauge configuration does not guarantee there was no fluid two examples demonstrate the tests performed on gas wells
movement below the gauges, it is a positive indication that and the third example presents an oil test.
comparison of the PBU pressures is probably valid.
Tidal oscillations are another physical phenomenon that Test Example 1. This is an example of an extended well test
may compromise the pressure data acquired during a test and (EWT) performed on an appraisal gas well. Fig. 7 presents
complicate a well test analysis. The bottomhole pressure data the pressure and rate data. The total test duration is 780 hours
acquired during well tests often contain an oscillating pressure (32.5 days). The test starts with a clean-up flow followed by
component caused by tides. This happens most often in the first pressure buildup, PBU 1. Next, a modified isochronal
offshore fields and is related to the rise and fall of the sea test is performed. It is followed by the second pressure
tides. The amplitude of this oscillating pressure signal in the buildup, PBU 2. The purpose of the modified isochronal test
test data varies with geographical location and can be sequence is to obtain the pressure data that provide
correlated with the amplitude of sea tides. There are regions information necessary for estimating the non-Darcy flow
on the earth where tidal effect is significant and there are coefficient. The second pressure buildup is followed by the
places where it is small. extended flow period. The extended flow lasted for 413
Before the pressure data measured during a test can be hours. During this period, there are two accidental shut-ins,
used for well test analysis, the tidal signal has to be identified PBU 3 and PBU 4, triggered by the emergency shutdown
and removed from the data. There are special algorithms6 that system. The test sequence ends with the final pressure
do this reasonably well. Tidal signal identification and buildup, PBU 5, which follows the extended flow period. The
extraction requires a reference tidal measurement. The best duration of this PBU is 137 hours. The test pressure data
way to obtain this reference measurement is by using a show remarkably consistent pressure transient behavior during
pressure gauge placed on the sea floor in the vicinity of the all pressure buildups of the test sequence. Fig. 8 demonstrates
well. The gauge should be set to record the sea floor pressure this for the PBU 2 and PBU 5 periods.
during the entire time period of the test.
( )

PBU 1 MIT PBU 2 PBU 3 PBU 4 PBU 5

5000. 5400.
pressure PSI

Extended f low
4600.
4200.

0. 200. 400. 600.

20000.
rates MSCF/D
-10000.

0. 200. 400. 600.

Time (hours)

Fig. 7 – Test Example 1. Pressure (markers) and rate data. Continuous line is the test sequence simulation using partial-penetration
homogeneous reservoir model.
SPE 102483 11

( )

PBU 2

DP & DERIVATIVE (MPSI2/CP/MSCF/D)

10-2

PBU 5
10-3 10-4

10-2 10-1 100 101 102

Delta-T (hr)

Fig. 8 – Test Example 1. Comparison of derivative plots of PBU 2 and PBU 5. Continuous line is the simulation using partial-penetration
homogeneous reservoir model
( )
5340. 5360.
pressure PSI
5320.
5300.

0. 200. 400. 600.

20000.
rates MSCF/D
-10000.

0. 200. 400. 600.

Time (hours)

Fig. 9 – Test Example 1. Close-up view of the Fig. 7 homogeneous model match during later part of pressure buildup periods.

A slightly higher Δp during PBU 2 in Fig. 8 indicates that the for the decreasing skin factor during the extended flow and as
skin factor decreased during the extended flow period between a result does not honor the pressure during this period. Fig. 9
the PBU 2 and PBU 5 and that the well continued to clean up. presents a close-up view of the late time pressure buildup data
The decreasing derivative trend between 0.2 hours and 4 hours from Fig. 7. Fig. 9 clearly demonstrates that while
from the start of the pressure buildups in Fig. 8 is most likely reproducing the pressure behavior during PBU 2, the
caused by spherically converging flow towards perforations in homogeneous reservoir model solution falls below the test
the vicinity of the well. This is known as partial penetration pressure during the PBU 5 period. This is an indication that
effect. Later, the derivatives stabilize and develop a horizontal the pressure support during the test is better than the pressure
trend reflecting radial flow geometry further away from the support in the homogeneous infinite reservoir represented by
well. The solid lines in Fig. 8 represent the simulation using a the model. This can only happen if the formation
partial-penetration homogeneous reservoir model. The model permeability-thickness increases at some distance away from
is calibrated to reproduce the partial penetration and the radial the well. The same conclusion is supported by the
flow behaviors and to honor the skin at the time of PBU 2. deconvolved drawdown pressure responses2, 3 derived from the
Fig. 7 presents the match of the pressure data during the whole PBU 2 and PBU 5 data. The drawdown responses are shown
test sequence using this model. The model does not account in Fig. 10. The longest of these responses reconstructed from
the PBU 5 data is defined on the time interval of 780 hours
12 SPE 102483

and shows the derivative decreasing at late time. This is also this reason, we use a linear composite model to match the late-
the sign of increasing formation permeability-thickness further time transient pressure behavior of the drawdown responses in
away from the well. Fig. 10. This model has a region with increased permeability-
Hence, the test data indicate the pressure behavior of an thickness on one side of the well. We calibrate this model to
open reservoir and that the production during the extended reproduce the relevant test pressure behavior and then
flow period does not produce any identifiable decline of the constrain it to identify the minimum size closed reservoir
average reservoir pressure. Geological evidence indicates model that is consistent with the test data.
presence in the reservoir of linear-type heterogeneities. For
RESPONSE FUNCTION (MPSI2/CP/M)
10-2

.15
.21
10-3

.15
10-4

.21

10-2 10-1 100 101 102 103

Delta-T (hr)

Fig. 10-Test Example 1. Deconvolved drawdown responses derived from PBU 2 and PBU 5 data
5340. 5360.
pressure PSI
5320.
5300.

0. 200. 400. 600.

20000.
rates MSCF/D
-10000.

0. 200. 400. 600.

Time (hours)

Fig. 11 – Test Example 1. Closed-reservoir linear composite model match during the later part of pressure buildup periods.

Fig. 11 demonstrates the match of the late time pressure from the PBU 5 data. The model solution derivative deviates
buildup data using this minimum size closed reservoir model. from the drawdown response derivative at late time. This
The model simultaneously reproduces the pressure buildups discrepancy is what identifies this closed model as the
performed before and after the extended flow period and, minimum size model supported by the test data. Note, that the
therefore, correctly represents the reservoir volume linear composite model does not represent partial penetration
investigated by the test. Fig. 12 demonstrates the match using and, as a result, does not honor the early-time transient
this model of the deconvolved drawdown response derived pressure behavior in Fig. 12.
SPE 102483 13

RESPONSE FUNCTION (MPSI2/CP/M)

10-2
. 21

10-3 10-4

. 21

10-2 10-1 100 101 102

Delta-T (hr)

Fig. 12-Test Example 1. Closed-reservoir linear composite model match of deconvolved drawdown response derived from PBU 5 data
( )
DP & DERIVATIVE (MPSI2/CP/MSCF/D)
10-3 10-2

PD=1/2
10-4

10-2 10-1 100 101 102

Delta-T (hr)

Fig. 13-Test Example 1. Closed reservoir linear composite model match of PBU 5 data

This, however, is not detrimental for correct and accurate of two PBUs in Fig. 15 indicates larger skin factor during
representation of the test long term pressure behavior that PBU 1. Increasing derivative trend at late time indicates that
reflects the reservoir volumetric properties. Fig. 13 the well pressure is influenced by no-flow boundaries.
demonstrates the model match of the last PBU data on However, the derivatives do not decrease rapidly at late time.
derivative plot. This closed reservoir model provides an Such behavior would indicate a closed reservoir compartment.
estimate of the minimum connected gas volume proven by this This seems to contradict the pressure behavior during the
extended well test. whole test sequence in Fig. 14. The pressure in Fig. 14 builds
up to a lower pressure during PBU 2 compared to PBU 1.
Test Example 2. This is an example of a well test performed Normally, this pressure “depletion” is a sign of closed
on an appraisal gas well.3 Fig. 14 presents the pressure and reservoir compartment. The apparent contradiction in this test
rate data. The total test duration is 210 hours. The test data between the transient pressure behavior during individual
sequence includes two pressure buildup periods. The longest pressure buildups and the pressure behavior during the entire
buildup, PBU 2, lasted 121 hours. test sequence has been explained using pressure-rate
Fig. 15 compares the derivative plots of the two pressure deconvolution.3
buildups. The plots in Fig. 15 show reasonably consistent
pressure transient behavior. The separation of the Δp curves
14 SPE 102483

4800. 5000. 5200. 5400. 5600.

pressure PSI
PBU 1 PBU 2

0. 50. 100. 150. 200.

rates MSCF/D
10000. -5000.

0. 50. 100. 150. 200.

Time (hours)

Fig. 14-Test Example 2. Pressure (marker) and rate data. Continuous line is the test sequence simulation using the model developed for
matching the system response in Fig. 16.
DP & DERIVATIVE (MPSI2/CP/MSCF/D)

PBU 1
10-2

PBU 2
10-3 10-4

10-3 10-2 10-1 100 101 102

Delta-T (hr)

Fig. 15-Test Example 2. Comparison of derivative plots of two PBUs

RESPONSE FUNCTION (MPSI2/CP/M)

f rom PBU 1
. 20

. 20
. 12
10-2

. 12 f rom PBU 2
10-3 10-4

10-3 10-2 10-1 100 101 102

Delta-T (hr)

Fig. 16-Test Example 2. Closed reservoir model match of the constant-rate drawdown response derived from PBU 2 pressure data.
SPE 102483 15

( )

5550. 5600. 5650. 5700. 5750.

pressure PSI

0. 50. 100. 150. 200.

rates MSCF/D
10000. -5000.

0. 50. 100. 150. 200.

Time (hours)

Fig. 17-Test Example 2. Closed reservoir model match during the later part of pressure buildup periods
DP & DERIVATIVE (MPSI2/CP/MSCF/D)
10-2
10-3

PD= 1/2
10-4

10-3 10-2 10-1 100 101 102

Delta-T (hr)

Fig. 18-Test Example 2. Closed reservoir model match of PBU 2 data

This well is located in a very narrow reservoir channel simultaneously reproduces the pressure behavior during the
truncated from one end. The pressure support from the two PBUs. One can notice that the model pressure is slightly
channel open end is very weak. As a result, the pressure below the test pressure at the very end of PBU 2 period. This
behavior during the entire test sequence is very close to the deviation is intentionally introduced to define the minimum-
pressure behavior in a closed compartment. size model that is supported by the test data. Fig. 18 presents
We use the channel reservoir model developed to match the minimum size closed model match of the PBU 2 pressure
the test data3 to evaluate the reservoir volume proven by this data on a derivative plot. Again, the discrepancy between the
test. We close the model by truncating the channel from the model and the test derivative at the very end of the pressure
second end. The location of this closing boundary is chosen to buildup defines this model as the minimum size model that is
introduce small discrepancy between the model solution and consistent with the test data.
the test data. Fig. 16 demonstrates the match of the
deconvolved drawdown response derived from PBU 2 data by Test Example 3. This is an example of a well test performed
using this closed reservoir model. The derivative curve of this on an appraisal oil well. Fig. 19 presents the test pressure and
drawdown response develops an increasing trend at late time rate data. The total test duration is 299 hours. The test
and the slope of this trend is close to one. The minimum-size sequence includes two pressure buildup periods. The duration
closed reservoir model matches this drawdown response of the longest buildup, PBU 2, is 99 hours. Fig. 19 seems to
derivative behavior at late time in Fig. 16. Fig. 17 presents a indicate that the pressure during PBU 2 builds to a
close-up view of the pressure data during the later part of the significantly lower pressure than during PBU 1 period.
two pressure buildups of the test sequence. The model
16 SPE 102483

2250. 2300. 2350. 2400. 2450. 2500. 2550.

pressure PSI
PBU 1

PBU 2

50. 100. 150. 200. 250. 300. 350.

3000.
rates STB/D
-1000. 1000.

50. 100. 150. 200. 250. 300. 350.

Time (hours)

Fig. 19-Test Example 3. Pressure (marker) and rate data during the well test sequence.
DP & DERIVATIVE (PSI/STB/D)
10-2

PBU 2

PBU 1
10-3
10-4

10-4 10-3 10-2 10-1 100 101

D elta-T (hr)

Fig. 20-Test Example 3. Comparison of derivative plots of two PBUs

As in the Example 2 discussed above, this could be a sign of pressure data recorded by the bottomhole pressure gauges.
reservoir depletion. Fig. 20 compares the derivative plots of The tidal pressure oscillations in the test pressure data are
the pressure data during two pressure build-up periods. Fig. 20 identified and removed using the methodology described in
clearly shows that the test pressure is affected by tidal effect. Ref. 6. Fig. 21 presents the tidal pressure oscillations
The amplitude of the tidal oscillations during the PBU 1 identified in the test pressure data. Indeed, the amplitude of
period is noticeably higher than during the PBU 2 period. The oscillation in the bottomhole pressure data in Fig. 21 during
tidal oscillations have to be identified and removed from the the PBU 1 period is larger than during the PBU 2 period. This
test pressure data before proceeding with well test analysis. is consistent with the observations from the derivative plots in
This is done by using a reference pressure signal recorded by Fig. 20 discussed earlier. Fig. 22 compares the derivative
one of the two pressure gauges placed on the sea floor next to plots of the same two pressure buildups after the tidal signal is
the well. These gauges recorded the sea floor pressure during removed from the test pressure data. Fig. 22 shows
entire test sequence. The sea floor pressure record shows a reasonably consistent pressure transient behavior during the
pattern of oscillations caused by tides. A similar pattern of two pressure buildups.
oscillations, albeit with smaller amplitude is present in the test
SPE 102483 17

.100
-0.75E-08
PSI
-.050
-.100
-.150

50. 100. 150. 200. 250. 300. 350.

3000.
STB/D
-1000. 1000.

PBU1 PBU2

50. 100. 150. 200. 250. 300. 350.

Time (hours)

Fig. 21-Test Example 3. The tidal signal identified in the bottomhole pressure data
DP & DERIVATIVE (PSI/STB/D)
10-2

PBU 2
PBU1
10-3
10-4

10-4 10-3 10-2 10-1 100 101

Delta-T (hr)

Fig. 22-Test Example 3. Comparison of derivative plots of two PBUs after the tidal signal has been removed from test data

The PBU 2 is significantly longer than PBU 1 and the pressure behavior in Fig. 23. However, this model significantly over
data during PBU 2 has been used for pressure transient estimates the pressure during the PBU 2 period on the
analysis. The pressure derivative stabilizes at middle time in Cartesian plot of the whole test pressure history in Fig. 24.
Fig. 23 reflecting radial flow geometry near the well. This Normally, this problem can be corrected by closing the
provides an estimate of the formation kh . After ten hours model on the fourth side. However, in this specific test this
from the start of PBU 2, the derivative curve develops an approach does not work. It is not possible to simultaneously
increasing trend with the slope close to ½. After twenty hours reproduce the test pressure behavior in Fig. 23 and Fig. 24
into the pressure buildup, the slope steepens which is a using such closed reservoir model. Indeed, the minimum size
signature of a channel-type reservoir truncated at one end. closed model that still honors the pressure buildup behavior on
Except for the final derivative trend at the very end of pressure derivative plot in Fig, 25 over predicts the PBU 2 pressure in
buildup, a model with such three-sided boundary Fig. 26.
configuration honors the middle and the late-time pressure
18 SPE 102483

DP & DERIVATIVE (PSI/STB/D)

10-2
10-3

PD=1/2
10-4

10-4 10-3 10-2 10-1 100 101 102

Delta-T (hr)

Fig. 23-Test Example 3. Match of PBU 2 test data using truncated channel model
2250. 2300. 2350. 2400. 2450. 2500. 2550.
pressure PSI

50. 100. 150. 200. 250. 300. 350.

3000.
rates STB/D
-1000. 1000.

50. 100. 150. 200. 250. 300. 350.

Time (hours)

Fig. 24-Test Example 3. Match of well test data using truncated channel model
DP & DERIVATIVE (PSI/STB/D)
10-2
10-3

PD=1/2
10-4

10-4 10-3 10-2 10-1 100 101 102

Delta-T (hr)

Fig. 25-Test Example 3. Match of PBU 2 data using closed reservoir model
SPE 102483 19

2250. 2300. 2350. 2400. 2450. 2500. 2550.

pressure PSI

50. 100. 150. 200. 250. 300. 350.

3000.
rates STB/D
-1000. 1000.

50. 100. 150. 200. 250. 300. 350.

Time (hours)

Fig. 26-Test Example 3. Match of test data using closed reservoir model

In order to reproduce the test pressure behavior including the pressure decrease between PBU 1 and PBU 2 is not due to
pressure decrease from PBU 1 to PBU 2, a region with closed reservoir compartment, but due to deterioration of
significantly lower mobility is introduced in the above model reservoir quality further away from the well. Having
with three-sided boundary. This low-mobility region begins at developed the open reservoir model that reproduces all the
some distance from the well and extends in the open reservoir relevant features of test pressure behavior, we can now
direction. The low mobility region reduces the pressure evaluate the minimum connected reservoir volume proven by
support and allows reproducing the pressure decrease between the test. We do this by introducing a closing boundary at the
the two PBUs. It also improves the quality of PBU 2 pressure open end of the above model. We adjust the position of this
match on a derivative plot. This is illustrated in Fig. 27 and closing boundary to identify the minimum size model that still
Fig. 28 by the red solid curves representing the simulation honors all the relevant well test data. The simulation results
results using this heterogeneous open reservoir model. The using this minimum size model are shown in Fig. 27 and Fig.
red curve in Fig. 28 is somewhat obscured by the blue curve 28 by the solid blue lines.
representing another simulation using the closed reservoir
model discussed below. This simulation reveals that the
DP & DERIVATIVE (PSI/STB/D)
10-2
10-3

PD=1/2
10-4

10-4 10-3 10-2 10-1 100 101 102

Delta-T (hr)

Fig. 27-Test Example 3. Match of PBU 2 data using the linear composite truncated channel model (red) and the closed-reservoir composite
model (blue)
20 SPE 102483

2250. 2300. 2350. 2400. 2450. 2500. 2550.

pressure PSI

50. 100. 150. 200. 250. 300. 350.

3000.
rates STB/D
-1000. 1000.

50. 100. 150. 200. 250. 300. 350.

Time (hours)

Fig. 28-Test Example 3. Match of test data using the linear-composite truncated channel model (red) and the closed-reservoir composite
model (blue)
10-1
RESPONSE FUNCTION (PSI/STB/D)

.15
.09

.15

.09
10-2
10-3
10-4

10-4 10-3 10-2 10-1 100 101 102

D elta-T (hr)

Fig. 29-Test Example 3. Match of the drawdown responses deconvolved from PBU 1 and PBU 2 data using the closed reservoir composite
model

Fig. 29 presents the two constant rate drawdown pressure

responses deconvolved from the PBU 1 and PBU 2 data. The Conclusions
drawdown pressure responses are very consistent allowing to 1. Evaluation of connected reservoir volume through the use
accurately determine the initial reservoir pressure.3 The solid of well testing requires appropriately designed well test
blue curve in Fig. 29 represents the simulation using the same sequence. The volume of fluid produced during the test
minimum size bounded reservoir model. The model must be correlated with the reservoir volume targeted to
accurately reproduces the deconvolved responses on the time be proven by the test.
interval of the whole test duration. Fig. 27 and Fig. 29 2. As a minimum, the test sequence must include two
illustrate that moving the closing boundary even closer to the pressure buildup periods immediately before and after the
well would cause the solution to deviate from the test data at main production period. Together, these two pressure
the very end. Hence, this is the minimum size model still buildups contain the information that reflects the volume
consistent with the test data. of the part of the reservoir investigated by the test.
Two other field test examples using this methodology for 3. The pressure buildup immediately following the main
evaluation of minimum connected reservoir volume proven by production period should be longest of the two PBUs. It
the test have been documented in Ref. 7. provides the main data for pressure transient analysis
SPE 102483 21

4. The pressure gauges used during the test should be Acknowledgements

located in the well as close as possible to the tested The authors would like to thank the managements of BP and
reservoir interval. The gauges should be kept in the same Total for the support and permission to publish this paper
location during the entire well test sequence. They should
have suitable gauge resolution and gauge drift References
characteristics. 1. Larsen, L. Uncertainties in Standard Analyses of Boundary
5. The reservoir volume proven by the test is estimated Effects in Buildup Data, SPE Reservoir Evaluation &
based on material balance considerations. This is not a Engineering, October 2005, 437-444.
uniquely defined characteristic. It has some degree of 2. Levitan, M. M. Practical Application of Pressure/rate
deconvolution to Analysis of Real Well Tests, SPE
uncertainty. The main uncertainty is associated with total reservoir Evaluation & Engineering, April 2005, 113-121.
compressibility. Non-uniqueness of test interpretation is 3. Levitan, M. M., Crawford G.E., Hardwick, A. Practical
also a contributing factor to this uncertainty. Considerations for Pressure-Rate Deconvolution of Well-
Test Data, SPE Journal, March 2006, 35-47.
Nomenclature 4. Bourgeois, M.J., Wilson, M.R. Additional Use of Well Test
c = compressibility, psi-1 Analytical Solutions for Production Prediction, SPE 36820,
h = thickness, ft Presented at the 1996 SPE European Petroleum
k = permeability, md Conference, Milan, Italy, Oct. 22-24.
p = pressure, psi 5. Von Schroeter, T., Hollaender, F., Gringarten, A.C.,
Deconvolution of Well-Test Data as a Nonlinear Total
Rinv = radius of investigation, ft Least-Squares Problem, SPE Journal, December 2004, 375-
s = saturation 390.
t = time, hrs 6. Levitan, M. M., Vinh, P. Identification of Tidal Signal in
V = volume, ft3 Well Test Pressure Data, SPE 84376, Presented at the 2003
φ = porosity SPE Annual Technical Conference and Exhibition, Denver,
η = diffusivity coefficient, ft2/hr Colorado, Oct. 5-8.
7. Boutaud de la Combe, J.L., Akinwumni, O., Dumay, Ch.,
μ = viscosity, cp Tachon, M., Use of DST for Effective Dynamic Appraisal:
α = units conversion factor in Eq. 1, α = 0.0002637 Case Studies From Deep Offshore West Africa and
(ft2 cp)/(hr psi md) Associated Methodology, SPE 97113, Presented at the
2005 SPE Annual Technical Conference and Exhibition,
Dallas, Texas, Oct. 9-12.
Subscripts
r = rock
o = oil
w = water