WE LL TES TiNG
Bottomhole pressure
Testing Design and Analysis
Bruno Deruyck Christine Ehlig-Economides
Montrouge, France Clamart, France
Well testing is performed in so many different guises that it is easy to lose
These are to confirm the economic potential
In its simplest form, testing provides short-
term production of reservoir fluids to the
surface permitting the operator to confirm
the show—indicated by cuttings, cores and
logs—and estimate reservoir deliverability.
In its subtlest form, measured pressure tran-
sients caused by abrupt changes in produc-
tion can characterize completion damage,
reservoir permeability and distant reservoir
heterogeneities.
The logistics of well testing are simple in
concept, but complex in practice. Flowing
an exploration well requires a temporary
completion. Flowing any well not con-
nected to downstream facilities requires
heavy surface equipment including separa-
tors and flares. Obtaining pressure transients
requires alternately shutting and opening
the well, preferably downhole, and making
accurate downhole measurements of pres-
sure. Increasingly, testing is performed in
combination with perforating and produc-
28
Jeffrey Joseph
London, England
tion logging to measure downhole flow.
They are routinely run in horizontal as well
as vertical wells.
Developing the multifarious and intricate
hardware to accomplish all these tasks is a
design engineer’s dream. And juggling the
many options for conducting a well test pro-
vides endless challenges in the field (see
“The Nuts and Bolts of Well Testing,” page
14). This article concentrates not on hard-
ware but on the information well tests give
and how tests are designed and interpreted.
Primary concerns in testing exploration
wells are obtaining representative samples
and estimating reservoir producibility.1 Fluid
samples are needed to determine various
physical parameters required for well test
analysis, such as compressibility and viscos-
ity, and for pressure-volume-temperature
(PVT) analysis that unlocks how the hydro-
carbon phases coexist at different pressures
and temperatures.2 For oil, a critical PVT
parameter is bubblepoint pressure, the pres-
sure above which oil is undersaturated in
gas and below which gas within oil starts
being released. Maintaining reservoir pres-
sure above bubblepoint is key to successful
testing since the principle of transient analy-
sis, described below, holds only if flow in
the reservoir remains monophasic. Estimat-
ing reservoir producibility requires achiev-
Bubblepoint
∆pA–B
∆pA–C
Wellbore Potential
B
A C
Crossflow rates
Production rate
nEstimating producibility by altering the
production rate and noting changes in
bottomhole pressure (top). A well’s pro-
ductivity index, or inflow performance, is
the slope of the straight line, measured in
barrels of oil per day per psi. The straight-
line response curves downward once
pressure falls below bubblepoint and gas
starts coming out of solution.
In a layered reservoir, individual pro-
duction rates measured using a produc-
tion logging tool—layers A, B and C in this
example—are plotted versus each layer’s
wellbore potential, the wellbore pressure
normalized to a datum. This so-called
selective inflow performance technique
reveals individual layer inflow perfor-
mances and also pressure imbalances
between layers that can promote crossflow.
In this article, COMPUTEST (wellsite computer sys-
tem), FPE (Fluid Properties Estimation), IMPULSE
(measurement while perforating), MDT (Modular For-
mation Dynamics Tester), PLT (Production Logging
Tool), RFT (Repeat Formation Tester), SPG (Sapphire
Pressure Gauge), STAR (Schlumberger Transient Anal-
ysis and Report) and ZODIAC (Zoned Dynamic Inter-
pretation Analysis and Computation) are marks of
Schlumberger.
1. Barnum RS and Vela S: “Testing Exploration Wells by
Objectives,” paper SPE 13184, presented at the 59th
SPE Annual Technical Conference and Exhibition,
Houston, Texas, USA, September 16-19, 1984.
2. Freyss H, Guieze P, Varotsis N, Khakoo A, Lestelle K
and Simper D: “PVT Analysis for Oil Reservoirs,”
The Technical Review 37, no. 1 (January 1989): 4-15.
Oilfield Review
 ing stable flow rates at several choke sizes
and then determining the productivity index
from the slope of the flow versus drawdown
pressure data (previous page ).
The type of oil as determined by a sample
and the ability of the well to produce are
the first steps toward commercial exploita-
tion. If well productivity is less than
expected, then wellbore damage may be
the cause. This is the next concern in testing
exploration wells. Estimating the near-well-
bore condition to perform necessary reme-
dial action and ultimately to plan a well
completion strategy for the field is accom-
plished from the transient analysis part of a
well test.
Transient analysis, however, reaches
deeper than just the near-wellbore region.
Today, it contributes so much to characteriz-
ing the reservoir that engineers increasingly
refer to well testing as reservoir testing.
Analysis can indicate the likely producing
mechanism of the formation—for example,
how much production comes from frac-
tures, how much from intergranular poros-
ity—and it can determine the producing
zone’s permeability-thickness product, kh. It
can see to the limits of the reservoir indicat-
ing the probable shape (but not orientation)
of the reservoir boundaries and can show
whether the primary recovery mechanism is
from water or gas-cap support. This infor-
mation becomes crucial in the appraisal and
production stages of field development
when engineers combine testing interpreta-
tion results with seismic and geologic data
to refine their understanding of the reservoir.
How does transient pressure testing work?
Imagine first an oil well in stable production
with a certain pressure drawdown between
the far limits of the reservoir and the well.
Now shut in the well. In the formation, a
sort of concertina effect takes place (below).
Oil near the wellbore is the first to sense the
shut-in and gets stopped in its tracks as it
tries to push more oil ahead of it, getting
compressed in the process. Then, the shock
is felt farther away as news of the shut-in, so
to speak, travels to the outer regions of the
reservoir. Gradually, the pressure builds up
everywhere, eventually reaching the reser-
voir pressure that drives production.
The reservoir engineer follows this chain
of events by measuring the pressure
buildup, or transient, and through analysis
determines information about the reservoir
from near the wellbore to its limits. An anal-
ogous chain of events occurs if instead of
shutting in the well, the well is opened and
allowed to flow. Again, it is oil near the
wellbore that first senses the disturbance,
but it is only a matter of time for oil deeper
in the reservoir to respond and begin flow-
ing too. Drawdown pressure measurements
to track these events practically mirror the
buildup response. In fact, transients can be
obtained simply by increasing or decreasing
the flow rate.
Transient testing depends on accurate
pressure measurements taken long enough
after the flow rate change to observe what
the test was designed to detect. Impulse test-
ing, for example, measures the transient that
occurs as a well is perforated, allowed to
produce for a short time and then shut in.

sight of its two real purposes.

of a discovery well and to maximize the cost efficiency of production in a developing or mature field.

Integrated with other measurements, well tests help provide the basis of reservoir characterization.

Fracture

Sealing fault

nPlan view showing the movement of wave fronts of a pressure transient

progressing away from a well. This shows a hydraulically fractured well
drilled near a sealing fault. Once wellbore storage disappears, the tran-
sient moves into the formation along linear paths perpendicular to the
fracture. Farther from the well, the transient moves radially from the
borehole. On hitting the sealing fault, the transient begins reflecting back
toward the borehole.

April 1992 29
The primary target is the near-wellbore Impulse Conventional Interference
region (right ).3 The goal is to assess forma-
tion damage and, if necessary, perform stim-
ulation. Tests last just an hour or two. In a
conventional test conducted to investigate
reservoir boundaries, often called a limit
test, the transient must be long enough for
the pressure disturbance to reach the
boundaries and then create a measurable
response in the well. How long this takes
depends on formation and fluid characteris-
tics. In particular, the lower the formation
permeability, the more time is needed—tests
can continue for days. Longest lasting are
interference tests, in which the effect of a
transient created in one well is observed in
another, yielding information about reser-
voir transmissivity and storativity.
The analysis and interpretation of well
tests have evolved remarkably since the
technique became established in the 1930s.
Today, a unified methodology has devel-
Flow

oped to obtain the maximum information

from any transient.4 The conventional test
on a new well comprises two flow periods
and two shut-ins (next page). The first flow
period, perhaps an hour long, is designed to
Pressure

clean up the near-wellbore region and give

the field crew time to manipulate chokes to
establish a practical, stable flow rate. The
well is then shut in and pressure builds up
to reservoir pressure, an important parame-
ter for the reservoir engineer. Then begins a
long flow period, followed by a shut-in last-
ing at least 1.2 to 1.5 times as long. This last
step generates the transient designed to
yield the reservoir’s secrets. Of course, there
are many variants on this theme (see “Text-
book Well Test from the Congo,” page 33).
Time
nThree types of well testing: Impulse, conventional and interference. Impulse testing
measures the transient caused by a very brief flow, typically just as the well is perfo-
rated. Results yield skin and permeability and may indicate if remedial stimulation is
required. Conventional well testing measures the shut-in transient after a lengthy flow
period and is often used to detect reservoir limits. Interference testing measures the
transient in a well caused by one or more flow pulses in a nearby well. Results yield
details about interwell transmissivity and storativity.

3. Ayestaran L, Ayoub J, Campbell J, Fairhurst D, Herrera

IC, Munsell S and Sneed BJ: “IMPULSE Testing,” The
Technical Review 36, no. 4 (October 1988): 37-45.
4. For a review:
Horne RN: Modern Well Test Analysis: A Computer-
Aided Approach. Palo Alto, California, USA:
Petroway, Inc, 1990.
For the development of well test analysis:
Ramey HJ Jr: “Advances in Practical Well Test Analy-
sis,” paper SPE 20592, presented at the 65th SPE
Annual Technical Conference and Exhibition, New
Orleans, Louisiana, USA, September 23-26, 1990.
5. Bourdet D, Whittle TM, Douglas AA and Pirard YM:
“A New Set of Type Curves Simplifies Well Test
Analysis,” World Oil 196 (May 1983): 95-106.
Bourdet D, Ayoub JA and Pirard YM: “Use of Pressure
Derivative in Well Test Interpretation, paper SPE
12777, presented at the 1984 California Regional
Meeting, Long Beach, California, USA, April 11-13,
1984.
The basic data obtained are change in
pressure, ∆p, versus elapsed time since the
transient was initiated, ∆t. In traditional
analysis, ∆p is plotted against the logarithm
of (tp + ∆t )/∆t , a dimensionless variable in
which tp is the duration of the flow period.
This is the Horner plot—( t p + ∆ t )/∆ t is
called Horner time (next page)—and the
transient is analyzed by tracing the
progress of the data from right to left.
First comes wellbore storage, which refers
to the obfuscating role of the wellbore fluid
when a transient is initiated. The moment a
well is shut in or allowed to flow, fluids in
the wellbore must first compress or expand
before formation fluids can react. If flow is
controlled from the surface, the entire well’s
fluids contribute to wellbore storage and the
effect can dominate the pressure transient
for hours afterward. The effect is exacer-
bated if well pressure toward the top of the
well drops below bubblepoint and part of
the well is filled with compressible gas.
Wellbore storage is substantially reduced by
shutting in the well downhole, minimizing
the volume of fluids that contribute.
As wellbore storage dissipates, the tran-
sient begins to move into the formation.
Pressure continues building up, but at a
slower rate as the transient moves far
enough to achieve radial flow toward the
wellbore. This is the so-called radial-flow
regime that appears as a straight line trend
on the Horner plot. The radial-flow regime
is crucial to quantitative interpretation, since
it provides values for kh and skin, S, a mea-

30 Oilfield Review
sure of the extra pressure drop caused by
wellbore damage. Skin takes positive values
in a damaged well when pressure drop near
the wellbore is greater than expected and
negative values when stimulation creates
less pressure drop. Next, the transient
encounters the limits of the reservoir and
pressure departs from its straight-line radial-
flow response.
The definition of Horner time is based on
a step change in flow rate, with one flow
period followed by a buildup. In actual
tests, there are always at least two prior flow
periods, often many more, and each affects
the pressure response after it occurs. Never-
theless, their cumulative effect can be deter-
mined using the superposition principle,
which states that transients occurring
sequentially simply add up. This results in
generalized Horner time that takes into
account the flow rates and flow times for all
previous flow periods. Using generalized
Horner time, the Horner plot retains its
validity in determining kh and skin for the
most complex series of drawdowns and
buildups, providing that the radial-flow
regime is present in the response.
Although the Horner plot is acceptable for
interpreting the radial-flow regime of easy-
to-interpret tests, a straight-line trend is often
difficult to pick out. Alternatively, there may
be several straight-line trends, of which only
one represents radial flow. Also, the plot
fails to provide ready insight into the nature
of reservoir limits. As pressure measure-
ments improved in accuracy, it was this
aspect that increasingly engaged the atten-
tion of reservoir engineers. The solution, dis-
covered in in the early 1980s, was a double
logarithmic, or log-log, plot of two sets of
data versus ∆t (below).5 One set is simply
∆p, the other is the gradient, or derivative,
of the response on the Horner plot. The
virtue of the log-log plot is that reservoirs
similar in construction but perhaps differing
in thickness, porosity and permeability give
rise to similar looking responses and can be
recognized as belonging to a class.
A pressure transient breaks into several
regimes on the log-log plot, each seeing
deeper than the last. The first regime typi-
cally reflects wellbore storage, during which
both the pressure and derivative curves
overlay and increase along a straight line of
unit slope. As wellbore fluids stabilize, pres-
sure continues building up, but at a slower
rate. The derivative curve swings down,
eventually flattening out as the transient
moves far enough from the wellbore to
achieve radial flow. Since the radial-flow
regime is a straight-line trend on the Horner
plot, the derivative curve on the log-log plot
is constant and traces a horizontal line. The
interpreter’s first task always is to identify
this derivative plateau, but this may require
waiting a long time in tests dominated by
wellbore storage (page 34, top).
Lengthy wellbore storage can totally mask
earlier flow regimes that occur for certain
borehole-formation configurations and for-
mation types, causing distinct perturbations
(continued on page 34)

Data

1st Flow 1st 2nd Flow 2nd Buildup

Buildup
Pressure

Horner Plot
Limits Radial
flow

Time
Wellbore
∆p

nElements of a conventional two-stage buildup tran- storage

sient test. Testing engineers use the first flow period to
clean up formation damage and adjust the choke to
gauge the producing capacity of the well. The first
buildup provides a first estimate of reservoir pressure.
Then begins a long flow period, followed by a longer
buildup. Analysis of the transient measured during this
(tp + ∆t )/∆t
second buildup reveals details of the near-wellbore
Log-Log Plot
region, formation characteristics such as permeability,
and distant limits of the reservoir.
Traditional analysis centered on the Horner plot
(middle), in particular the straight-line trend that sig-
∆p and Derivative

nals radial flow. Today, the log-log plot (bottom) of ∆p Wellbore

and the derivative, the slope of the Horner plot, is used storage
to first diagnose the various flow regimes of the tran-
sient. Then, specialized plots such as the Horner plot
are used to estimate specific parameters such as per- Limits
Limits
meability, skin and reservoir pressure. Radial flow

∆t

April 1992 31

Downhole shut-in
100
∆p and Derivative, psi
10-1
10-2
10-3
10-2 10-1 100
Time, hr
on the derivative response. The signs are
varied (next page ). A partially penetrated
formation produces a linear trend on the
derivative curve with a slope of –1/2. In wells
where the formation is strongly layered or
naturally fractured, the derivative tends to
dip before it rises to the radial-flow plateau.
If wellbore storage is not too dominating, the
transient can be analyzed to pinpoint the
most likely explanation.
The last regime on the log-log plot occurs
when the pressure transient has travelled far
from the well and encounters the reservoir
or drainage-area limits. Testing theorists
have worked out the transient response to a
catalog of boundary geometries ( right ). In
most cases, the transient responses alone do
not offer enough differentiation to enable
the interpreter to definitively establish the
boundary type. The choice of the type as
well as the orientation of the boundary
geometry must be guided by geologic, seis-
mic and log data.
There are three categories: no-flow
boundary, constant pressure boundary and
34
nReduction of Impermeable Boundary
wellbore storage
with downhole Well
shut-in. The log-log
plot compares two
well tests, one shut
in at the surface,
the other shut in
downhole. In the
surface shut-in test,
wellbore storage
Surface shut-in Partially Sealing Fault
masks the radial-
flow plateau for Well
over 100 hours
(4 days) (square
data points). The
plateau emerges
clearly in the down-
hole shut-in data
after just one hour
(triangular data
points). Intersecting Impermeable Boundaries
101 102 (From Joseph et al,
reference 6.)
Well
the special case in which the test is long
enough to reach all the no-flow boundaries,
thus forming a closed system. Examples of
no-flow boundaries include sealing
faults—perhaps several of them—pinchouts, Parallel Impermeable Boundaries
and channels. Because no-flow boundaries
reflect the transient back toward the well,
they cause ∆p to rise at higher than its nor-
mal radial-flow rate, so the derivative curve Well
jumps to a higher level. A sealing fault
causes the plateau value to double. With
two intersecting sealing faults, the jump is
correspondingly higher. If a fault is partially
Truncated Channel
sealing, the derivative curve starts to jump
but then falls back to its radial-flow value.
Constant-pressure boundaries, like a gas
cap or aquifer, allow the pressure transient Well
to flatten out at the boundary pressure, so
the derivative takes a nosedive, which is
instantly recognizable. In a closed system,
pressure is completely contained within the
reservoir. How this affects the ∆ p and Pinchout
derivative curves depends on whether the Well
transient is a drawdown or buildup. In
drawdown, both curves track a line of unit
slope, again an easily recognizable effect. In
buildup, the derivative curve starts moving
toward the line of unit slope but takes a
nosedive before reaching it, somewhat simi-
lar to the constant-pressure boundary case. nA representative selection of no-flow
These reservoir models are simpler than boundaries studied by well-test theorists.
nature generally allows—in reality, a mix- Any of these may appear in the late-time
ture of responses should be expected. portion of a transient’s diagnostic log-log
Thanks to the superposition principle, how- plot. The associated parameters defining
the boundary may then be estimated by
ever, responses may be combined to pro- regression analysis.
duce a realistic transient response for even
the most complex situation. Simulating data,
though, is the easier forward task. More dif-
Oilfield Review
 Log-log Horner Specialized Flow Regimes

Wellbore

∆p, Derivative
storage
Radial

∆p

∆p
flow
Other

Time Horner time Function of time

Homogeneous
Reservoir
Wellbore storage
coefficient

∆t
Partially
Penetrating Well

Infinite Conductivity
Vertical Fracture
(Fracture half-length)2
× matrix permeability

∆t
Finite Conductivity
Vertical Fracture
Fracture permeability
× fracture width

4
Sealing Fault ∆t

nResponse of log-log plot (left column) to

several common reservoir systems, show-
Constant Pressure ing different flow regimes (see legend).
Boundary
The log-log plot is used by analysts to
diagnose the flow regimes present in the
transient. Once regimes are identified, the
Horner plot (semi-logarithmic) and other
specialized plots (linear) are used to eval-
uate parameters characterizing the system.

Linear Channel

Matrix permeability
× (channel width)2

∆t
Dual-Porosity

35
ficult for the analyst is the inverse procedure
of finding the best model to match actual
test data (below ).6
For the traditional test comprising two
flow periods and two buildups, transient
analysis focuses on the second buildup. The
first step is to identify the various regimes on
the log-log ∆p and derivative-curve plots
and then choose the most likely model for
each. Estimation of model parameters is
then made using specialized plots that allow
a focused analysis of each flow regime (pre-
vious page ). For example, wellbore storage
in the early data is determined from the
slope of the straight-line portion of a linear
∆p versus ∆t plot. Confirmation and charac-
terization of a vertical, high-conductivity
fracture, recognized by a half-slope deriva-
tive trend on the log-log plot, come from a
plot of ∆ p versus ∆ t . The radial-flow
plateau is best analyzed using the general-
ized Horner plot. And so on.
Openhole logs
PVT data
Production logs
36
Using a workstation, the reservoir engi-
neer interacts with a computer program,
such as STAR Schlumberger Transient Anal-
ysis and Report and ZODIAC Zoned
Dynamic Interpretation Analysis and Com-
putation programs,7 to build a comprehen-
sive model using all the parameters found
for the various flow regimes, predict what
the entire transient should look like, and
compare the results with the data. In this
forward modeling process, the interpreter
tweaks parameters, either manually or auto-
matically using a nonlinear regression
scheme, and perhaps alters the choice of
model for one of the regimes to obtain the
best possible fit. There may be several com-
binations of models that match the data
equally well. In this case, other data must
be sought to decide which model is the
most appropriate (next page ).
The final interpretation step, called history
matching or verification, uses the model
Raw data
Preprocessing
Model diagnosis
Specialized plots
Parameter estimation
History matching
Results
nThe order of business
in pressure transient
analysis.
established in the second buildup to predict
pressure response throughout all four peri-
ods of the test and confirms that the model
satisfactorily accounts for all data. This may
result in more parameter adjustment
because every period must now be matched
simultaneously, even though the second
flow period is planned intentionally long to
minimize the influence of previous periods.
In some cases, interference from earlier
well manipulations may obscure key
regimes of the transient being analyzed.
Interpreters then resort to a process called
desuperposition that attempts to isolate the
transient from earlier ones and in particular
reform the given transient’s data to mimic
how the reservoir would have reacted if the
flow rate change had been an isolated, per-
fect step.8
Designing well tests involves many of the
same steps the interpreter uses. This is
because once a test has been proposed,
both the pressure data and the data’s inter-
pretation can be simulated to show that the
test as designed meets its goals—design sim-
ulation requires estimates of formation and
fluid parameters from nearby wells or the
well in question. By predicting the likely
shape of the log-log ∆ p and derivative
curves, the engineer can demonstrate the
feasibility of detecting and characterizing
the anticipated reservoir features. For exam-
ple, design simulation ensures that wellbore
storage does not smother the feature being
sought and guarantees a test that is long
enough to view suspected reservoir bound-
aries. Another important feature of simula-
tion is determining the accuracy and preci-
sion required of the pressure gauges.
The design phase not only maps out the
mechanics of a test, but also ensures that,
once underway objectives are met. For
example, the progress of the planned tran-
sient can be followed at the wellsite and
compared with that forecast during the
design. To avoid the costly mistake of rig-
ging down before the transient indicates a
desired feature, wellsite validation of data
during the test remains a must. This is best
accomplished with surface readout of
downhole gauges and enough computing
power at the surface to produce appropriate
plots, notably the log-log diagnostic plot. If
the reservoir response is quite different from
that assumed in the design, wellsite diagno-
sis permits an instant correction of the job,
perhaps a lengthening of the transient, to
Oilfield Review
 Sealing Fault ensure optimum use for the data. In certain
101
cases, real-time readout is not feasible and
downhole recording must be used. Data
validation can still be performed onsite right
∆p and Derivative

after retrieving the gauges.

Integral to well test design is selection of
100 hardware, which involves many options. To
minimize wellbore storage, should the well
be shut in downhole rather than at surface?
In a low producer, will the act of shutting in
actually kill the well? How sensitive must the
pressure gauges be? To some extent, these
10-1
questions are decided by the operator’s stan-
Two Intersecting Perpendicular Faults
101 dard practices, the current status of the hole,
the configuration of the downhole hardware
and, not least, safety considerations.
The options have expanded in recent
∆p and Derivative

years. While drillstem test (DST) equipment

has always guaranteed downhole shut-in in
100
new wells, downhole shut-in devices for
completed wells did not become commer-
cial until the early 1980s. Pressure gauges
have evolved from crude mechanical
devices to quickly reacting, highly accurate
10-1 quartz gauges. Perhaps the most unexpected
Dual-porosity Model innovation is a downhole flow measurement.
101 Traditional well testing theory dispensed
with a flow measurement because it
assumed constant wellbore storage,
∆p and Derivative

enabling flow to be estimated from early

pressure data. But reality is less predictable.
100 Wellbore storage often varies as the fluids in
the wellbore change during the test, and a
downhole flow measurement in fact offers a
valuable complement to conventional pres-
sure data.
Downhole flow measurements are cur-
10-1
rently performed using production logging
Dual-permeability Model
101 6. Joseph J, Ehlig-Economides CA and Kuchuk F: “The
Role of Downhole Flow and Pressure Measurements
in Reservoir Testing,” paper SPE 18379, presented at
the SPE European Petroleum Conference, London,
∆p and Derivative

England, October 16-19, 1988.

Ehlig-Economides CA, Joseph JA, Ambrose RW Jr and
Norwood C: “A Modern Approach to Reservoir Test-
100 ing,” Journal of Petroleum Technology 42 (December
1990): 1554-1563.
7. The newer ZODIAC program includes all the features
of the older STAR program.
8. Ehlig-Economides CA, Ambrose RW and Joseph JA:
“Pressure Desuperposition Technique for Improved
Late-Time Transient Diagnosis,” paper SPE 20550,
10-1 presented at the 65th SPE Annual Technical Confer-
10-1 100 101 102 103 ence and Exhibition, New Orleans, Louisiana, USA,
Time September 23-26, 1990.

nFinding the best model to fit the data. In this case, four scenar-
ios fit quite well, but the dual-permeability model fits best. Dual
permeability means a two-layered formation with a different
permeability in each layer.

April 1992 37
 tools and therefore not deployable when
downhole shut-in is planned. The most
∆p and Derivative

common uses are to measure transients

caused by either shutting in or altering the
flow rate at the surface—the tool, entered
into the well through a conventional riser
and lubricator, is suspended just above the
producing zone. A drawback of the current
technology is the threshold and resolution
of the spinner in the logging tool that mea-
Time Model sures flow. Well testing experts dream about
a robust flow measurement incorporated
T into the DST tool, enabling continuous flow
P(t) = 0
q(τ) p(t-τ) dτ measurements in newly drilled wells.
If flow measurements are at hand, how
does that change testing analysis? The nec-
essary theoretical ground was broken in the
Pressure 1980s. The underlying principle is, once
again, superposition.9 Suppose the pressure
response to a unit step change in flow is
p(t – t 0 ), where t 0 is when the step change
occurs. Then the pressure response, P(t ), to
a gradual change in downhole flow rate,
Flow rate q (t ), may be computed by approximating
the gradual change with a series of stair
steps, then considering each step as provid-
Time ing a minitransient, and finally through the
superposition principle summing all the
nThe convolution integral that converts pressure response to a unit step change in flow, minitransients (left ). As the steps are made
p(t), and actual measured flow rate, q(t), into measured pressure response, P(t). Convo-
lution revolutionizes transient analysis when downhole flow measurements are avail- smaller, this sum becomes the convolution
able, for example as measured by production logging in a flowing test. The mathemat- integral:
ical manipulation virtually wipes out wellbore storage, leaving later portions of the
T
transient clearly visible.
P ( t ) = q (τ) p (t– τ ) d τ .
nAn example
0

showing wellbore
103 storage virtually During a test, downhole pressure gauges
eliminated using measure P (t ) and a flowmeter measures q(t ).
Convolution derivative Pressure change the convolution But p(t ) is what the interpreter wants. Get-
derivative. The
∆p and Derivative, psi

transient was intro- ting at it requires the reverse process of

102
Pressure derivative duced by chang- deconvolution, which unfortunately is a
ing the well’s pro- rather unstable numerical procedure. More
duction rate and commonly, interpreters favor a procedure
101 downhole flow was
measured using called logarithmic convolution that converts
production logging. the two measurements more easily into
something that fits existing analytical tech-
100 niques (left ).
Logarithmic convolution is a mathemati-
cal trick in which a form for p ( t ) is
assumed—usually the response for infinite-
10-1
acting radial flow—that simplifies the above
10-5 10-4 10-3 10-2 10-1 100 101
convolution integral to a simpler expression
Time, hr involving a rate-normalized pressure
P(t )/q (t ), written J (t ), and a new time-scale
called sandface rate convolution time, tsfrc .
J (t ) and its derivative with respect to tsfrc
offer the same diagnostic power as the con-
ventional well testing analysis described
earlier with the advantage that most of the
wellbore storage is removed.

38 Oilfield Review
 There are several advantages to testing a
well with downhole pressure and flow mea-
surements under drawdown—and one dis-
advantage. The disadvantage is that reservoir
shut-in pressure is not measured. The advan-
tages are:
•in producing wells, little production is lost
since the well is never shut-in.
•in poor producers, production is not killed
as may occur during a shut in.
•in layered reservoirs, testing under draw-
down reduces the possibility of crossflow
between producing layers, while this can
easily occur in a buildup test complicating
the interpretation.
The technique’s most popular application in
layered reservoirs, though, is in analyzing
individual layer kh and skin values.10
This involves measuring a series of tran-
sients created by changing the production
rate, one for each layer with the production
logging tool situated at the top of the layer
(right ). The amount of data acquired is huge
and can be analyzed in several ways with
varying degrees of sophistication. The key,
however, is to first analyze the transient
measured with the tool situated just above
the bottom layer, yielding that layer’s reser-
voir properties. Then, a second transient is
measured with the tool situated above the
next layer, revealing reservoir properties of
the new layer and bottom layer combined.
Since reservoir properties for the bottom
layer are already estimated, the transient
can be analyzed to reveal just the new
layer’s properties. The process continues up
the well.
Layered reservoir testing (LRT) was origi-
nally conceived to investigate production
wells. Recently in offshore Congo, AGIP
used the technique to evaluate a layered
reservoir encountered by an exploration
well. Conventional testing of individual pay
zones in an exploration well would nor-
mally call for a separate DST-perforation run
for each zone. But using layered reservoir
testing, AGIP obtained reliable kh, skin and
productivity index values for individual
zones with only one trip in the hole, at a
considerable cost savings (see “Exploration
Layered Reservoir Testing in the Congo,“
next page).
The drawback of using an LRT in the
exploration setting is that production from
different zones commingles, ruling out rep-
resentative sampling from different pay
zones. Fortunately, a recent technological
April 1992
A B C
Surface flow rate
A
Downhole
C
Pressure
B Flow
A B C
Time
nSequence of downhole pressure and flow transients measured using a production log-
ging tool in a layered reservoir test. A separate transient is measured with the tool posi-
tioned at the top of each zone. Analyzing the transients yields individual zone perme-
ability and skin values.
innovation provides a solution. Samples of 9. Meunier D, Wittmann MJ and Stewart G: “Interpreta-
extraordinary reliability may now be tion of Pressure Buildup Test Using In-Situ Measure-
obtained from any number of zones using ment of Afterflow,” Journal of Petroleum Technology
37 (January 1985): 143-152.
the new wireline-conveyed MDT Modular 10. Kucuk F, Karakas M and Ayestaran L: “Well Test
Formation Dynamics Tester, but this has to Analysis of Commingled Zones Without Crossflow,”
be planned in advance because the sam- paper SPE 13081, presented at the 59th SPE Annual
Technical Conference and Exhibition, Houston
pling takes place in open hole (see “The Texas, USA, September 16-19, 1984.
MDT tool: A Wireline Testing Break- Ayestaran L, Ehlig-Economides C, Shah P, Kuchuk F,
through,” page 58). Nicolson B and Wittmann M: “Layered Reservoir
Testing,” The Technical Review 35, no. 4 (October
In addition to convolution and layered 1987): 4-11.
reservoir testing, there are other advantages
to supplementing conventional pressure
data with production logging measure-
ments. A flow profile run during stabilized
production or shut-in can pinpoint where
production is coming from and provide
invaluable data on crossflow between
zones. The information may directly influ-
ence testing interpretation. For example, if a
zone is producing only from its upper part,
a portion of the transient will react as if the
well were only partially completed. The
diagnosis must be adjusted accordingly. The
fluid density measurement in production
logging also plays a role by indicating
whether gas is coming out of solution, giv-
ing a warning that a test may be occurring
at below bubblepoint conditions.
Perhaps the most valuable contribution of
downhole flow measurements is in testing
(continued on page 45)
39
horizontal wells.11 Horizontal wells pose
two special problems for the reservoir engi-
neer. The first is the unavoidably large well-
bore storage effect. Horizontal sections may
extend for thousands of feet and cannot be
isolated from the transient. The second is
the more complex nature of the transient.
Once wellbore storage is stabilized, three
regimes possibly replace the radial-flow
regime of a conventional test (right ).
First is radial flow in a vertical plane
toward the well, indicated by a plateau on
the derivative curve on the log-log
plot—this regime is termed early-time,
pseudo-radial because permeability
anisotropy (vertical to horizontal) actually
causes an elliptical flow pattern. The second
regime begins when the transient reaches
y
the upper and lower boundaries of the pro- z
ducing zone and flow becomes linear
toward the well within a horizontal plane.
The derivative curve traces a line of slope
1/ 2. The third regime occurs as the transient

moves so far from the well that flow

becomes radial again, but this time in the
horizontal plane. The derivative curve
enters a second plateau.
Although this makes diagnosis more diffi-
cult, it also offers benefits. As in conven-
tional testing, the first plateau gives kh and
skin, but k is now the geometric average of
permeability in the vertical plane perpen-
dicular to the horizontal wellbore trajec-
tory, k y k z , the wellbore trajectory being
considered parallel to the x-axis. The inter-
mediate linear flow period gives horizontal
permeability along the y axis, ky , and the
second plateau gives the average perme-
ability in the horizontal plane, kx k y . In
theory, the three regimes together can pro-
vide a breakdown of permeability into its
three components.
The key to a successful interpretation is
recognizing the first plateau, not only
because this alone gives k z , but also
because it is the only regime that can
directly provide skin. However, it is the
regime most likely to get swamped by the
large wellbore storage occurring in a hori-
zontal well. The key to this dilemma is
either downhole shut-in, or downhole flow
measurements and logarithmic convolution.
Because of the length of a horizontal
well’s producing zone, supplementing test
x
nPhases in a horizontal well transient test. After wellbore storage has disappeared,
flow is first radial toward the well in the vertical y-z plane, then linear in the y-z plane,
finally radial in the x-y plane. The first and third regimes produce plateaus on the log-
log diagnostic plot and can be analyzed to provide vertical and horizontal permeability.
data with flow profiles measured during The future of testing is assured, of course.
production logging is even more crucial for What will accelerate its use and impact is
pinpointing production and recognizing better integration with other reservoir data,
crossflow (see “Horizontal Well Testing in improved downhole pressure and flow sen-
the Gulf of Guinea” page 42).12 Crossflow is sors, further development of transient theory
common in horizontal wells as in vertical and a continued evolution of the interactive
wells, particularly during a buildup test, and computer software that now aids interpreters.
may seriously jeopardize interpretation. —HE
Drawdown tests are therefore recommended
as an insurance policy, particularly for new 11. Clark G, Shah P, Deruyck B, Gupta DK and Sharma
wells in developed fields where differential SK: “Horizontal Well Testing in India,” Oilfield
Review 2, no. 3 (July 1990): 64-67.
depletion may exacerbate crossflow. Shah PC, Gupta DK, Singh L and Deruyck BG: “A
The underpinnings of horizontal well test- Field Application of the Methodology for Interpreta-
ing theory are developing rapidly. Interfer- tion of Horizontal Well Transient Tests,” paper SPE
20611, presented at the 65th SPE Annual Technical
ence testing of horizontal wells is being Conference and Exhibition, New Orleans, Louisiana,
worked out, as is the influence on the hori- USA, September 23-26, 1990.
zontal well-test response of the same range 12. Ahmed U and Badry R: “Production Logging as an
Integral Part of Horizontal Well Transient Pressure
of reservoir heterogeneities and boundaries Test,” paper SPE 20980, presented at Europec 90,
that are now well understood for conven- The Hague, The Netherlands, October 22-24, 1990.
tional testing.

April 1992 45