You are on page 1of 93

Production Testing

2009-Oct-27
AStech Inc.
A Seminar by
Dave Pridie

Objectives
1.
Mechanics of a production test
- know mechanical components of test
- know basic definitions
- understand how to conduct a test
- understand simple pressure chart
- know mechanical and electronic pressure charts
- know what the test accomplishes
2.
Visual analysis of pressure charts
- know basic quality checks
- understand mechanical and electronic recorders
- see pressure difference plot, know effect of moving liquid level
- see Horner plots and recognize different types of reservoirs
3.
Rates and Sampling
- know flow rate definitions and calculations
- know sampling techniques
4.
Pressure Transient Analysis
- know definitions of parameters
- know assumptions
- understand basic equations
- see old style mathematics (previous to Stehfast)
- understand semi-log and log-log methods
- understand superposition and Horner analysis
i. see examples
- understand log-log analysis
i. dimensionless parameters
ii. EI type curve
iii. equivalent time
- understand well bore storage
- understand skin
i. skin factor
ii. skin effect and rate calculations
- understand derivatives of flow equations
i. derivative plot and Gringarten
ii. see log-log plots and recognize different types of reservoirs
- understand similarity of models
- understand equivalency of models in production forecasting
- what to look for in a complete analysis
- current examples
5.
Common Reservoir Models
- understand hydraulically fraced reservoirs and linear flow
- review whats new in multiply fraced horizontal wells
- boundaries, single, L-shaped, parallel
- understand radial composite reservoirs
- horizontal wells (to be done)
6.
Designing PTs
- design AOFs, Reservoir Parameter tests, Reserve tests
- know requirements for testing after completion
- know pretest plan
- know costs
- know how PT fits into larger picture

Part 1 Mechanics of a Production Test


What Happens
Fluid is removed from the formation, i.e. the well is flowed, causing the pressure in the well bore
and near formation area to drop. The well is then shut in and the pressure rises.
What It Does
A production test can either be a snapshot of a wells ability to produce or an indicator of its future ability
to produce. A mechanically successful PT can accomplish several objectives. It can recover high quality
samples of the formation fluids; establish flow rates, fluids ratios and absolute open flow potential, AOF
(the AOF being an ERCB-legislated gas-well requirement to be done within the first three months of
production). It can measure formation parameters including pressure, permeability, skin factor, hydraulic
fracture length and conductivity, plus boundaries. Recently, advances have been made that allow the
accurate measurement of reservoir size, which in conjunction with the reservoir parameters can yield a
production forecast upon which economic decisions can be made. With good PT data, different scenarios
of stimulation can be forecast as well.
The four main Components of a Production Test are:
1.
Perforations and Stimulation History
2.
Conduit (tubing and other components of production string)
3.
Pressure Gauges
4.
Production Test equipment (vs. Inline Testing)
1. Perforations: These allow the formation fluids to enter through the casing into the well bore.
Depth, width, linear density plus whether shot under-or-over balanced, all reflect on the efficiency
of the perforations. Of equal importance are the stimulation procedures following perforating,
including acid or solvent washes or injections plus plasma, propped or acid fracs. In an open
hole or barefoot completion, the formation is completely open to the well bore so there is no need
of perforations but fracing and acidizing operations are much more difficult.
2. Conduit: Obviously, the smaller the tubing size, the more restricted the gas rate. Consideration
must also be given to lifting water where smaller tubing sizes are better (see attachments for
nomographic calculation for gas rates required to lift water). Of particular importance is the depth
to which the tubing is run, since, for testing, it may be important to keep the tubing end to the midpoint of perforations distance a minimum.
3. Pressure Gauges: Most testing is done with strain-gauge type, down-hole electronic recorders
run in tandem but consideration must be given to running surface gauges with or without fluid
level sensing, or down hole quartz-type gauges (more accurate, more sensitive?). As we will see,
running the gauges to mid-point perforations is a concern when testing for boundaries and
reserves.
4. Production Testing Equipment: Size, H2S rating, sand handling capability plus heating (in case
of hydrates or winter conditions) are prime considerations for choosing production test equipment.
Cost is secondary but still a very important consideration because production testing of new wells
has to compete with the philosophy of tying in and testing after. Equipment should be sized to
the job.

Some Quick Definitions

1. Deliverability: Assumed to be defined by the simplified equation Q = C (PR2 Pwf2) n at sand face
(bottom hole) conditions. This seminar does not deal with the LIT (Laminar-Inertial-Turbulent)
equation for deliverability due to the declining significance of deliverability (which is a snapshot) in
favor of production forecasting (productivity over time).
2. AOF: Absolute Open Flow. Deliverability when flow pressure (pwf) is zero.
3. AOF Test: The name given to the test the ERCB requires of gas wells. Can be as simple as
calculating Qmax (letting pwf=0 and assuming n=1.0 in the deliverability equation).
4. Production Test: Generic name given to deliverability testing.
5. Stabilized Deliverability: Actually does not exist. Is defined as the rate at the point in time when the
transient from a well test touches the (assumed radial) boundary and converts from infinite acting
radial (IARF) flow to pseudo steady state (PSS) flow. Often assumed to occur at the time when the
radius of investigation reaches 804.7 meters, which represents the edge of one section of lad. (Some
use 908 meters radius to represent the area of 1-section.).
6. Modified Isochronal Test: A multiple rate test defined by the ERCB that calculates n. Usually run
with four increasing flow rates interspersed with three shut ins, all of the equal duration (usually two to
four hours) with the final rates being choked back to about the third rate (the rate at line pressure) and
then held for an extended period of time (1-5 days).
7. Single Point Test: Assumes n = 1.0 (deliverability eqn). Usually run at expected line pressure.
8. Reservoir Parameters: Permeability (k), skin (s) and pressure (p). pi is the initial pressure (when
drilled). p* is any extrapolated pressure and pR is average reservoir pressure notably one that is
calculated from a linear p* or equal to a curvilinear p*. Frac parameters also fall under this category
and include fracture half length (Xf), fracture conductivity (kfx) where x is the fracture width and sf
which is the frac face skin.
9. Boundaries: Barriers to flow. Can be single or multiple, intersecting or parallel, linear or radial, noflow, partial flow, changing transmissibility or constant pressure in nature.
10. Reserves: Usually GIP (gas in place) or OIP (oil in place). Is OGIP (original GIP) and OOIP if done
before any hydrocarbons removed from the zone. Is RGIP (remaining GIP) and ROIP if done after
production has occurred?

How to Conduct a Reserves-Calculating Production Test on a New Well


1. Complete the well. If reserves calculation is a consideration, allow sufficient time to elapse for well to
return to equilibrium.
2. Run gauges. If reserves calculation is a consideration, run to MPP (mid-pt perforations). If sand face
AOF is known to be below 21 103m3/d, surface gauges will suffice but conversion to BHP must be
done.
3. Attach surface equipment.
4. If reserves calculation is a consideration, flow well for a short period of time and shut in to obtain
initial pressure.
5. Flow well at approximate line pressure to estimate actual deliverability. Do modified isochronal test, if
necessary (Expected AOF is greater than 300 103m3/d, in case where line pressure cannot be
achieved (or is not known) or if testing in Saskatchewan). Measure rates and pressures. Take
samples at end of flow. If reserves calculation is a consideration, flow well for sufficient time to
remove gas yielding an acceptable pressure decline 50 kPa is good rule of thumb.
6. Shut in and allow pressure to build up to establish final pressure plus transient analysis.
7. Pull gauges and do analysis.
Some Things to Remember
1. If the test is being run only to satisfy ERCB requirements, an initial bottom hole pressure and surface
rate and pressure data may be all that is required. Often, the cleanup (done with the service rig)
yields enough data to estimate deliverability accurately enough for ERCB purposes. If it does not,
production during the first three months can be used.
2. As can be seen, the preparation and running of a test that yields a reserve-estimate is much more
involved than one that simply yields deliverability and reservoir parameters.
3. Running a reserves test immediately upon completion requires special programming and special
consideration for the completions department. Prior knowledge helps immensely to keep the overall
test time and costs down.
4. Strive for simplicity in design. Nature will provide the complications.

What Each Step Accomplishes


1. Stabilization time may be required to allow transients from previous operations (cleanups etc.) to
dissipate when testing for reserves. With good planning, this step can be eliminated by incorporating
the cleanup into pretest flow. If not, a shut in equal to 10 times the cleanup flow time will suffice.
2. The pretest flow (2 hours) simply removes enough fluid to ensure that there is no supercharging and
gives an indication of the rates and fluids expected. The short duration helps prevent the influence of
heterogeneities/boundaries being seen in the following buildup.
3. The first shut in (22 hours) defines pi (the initial reservoir pressure). The long duration of the shut in
combined with the shortness of the preceding flow makes the extrapolation distance quite small and
thus pi can usually be determined with a linear extrapolation. This requires that the pressure be
stable prior to the 2 hour flow.
4. The modified isochronal (MI) flow and buildup period establishes Q vs. P and allows calculation of the
exponent n in the deliverability equation Q = C (delta P2)n. (Rarely done any more. Single points
are enough except when AOF is greater than 300 103m3/d)
5. The extended flow (2-5 days) investigates further into the reservoir and helps establish stabilized
rates.
6. The main shut in (2-4 weeks) provides pressure data that is analyzable for permeability and skin,
boundaries & heterogeneities plus reservoir size.
Pressure and Pressure Charts
Modern pressure gauges are very sensitive and very accurate and can retain a very large number of data
points. Choosing gauges is not the concern that it used to be. Nonetheless, differences in pressure
measurements between gauges can still be as much as 30 kPa at reservoir pressures. For use in long
term P/Z plots, this accuracy is more than sufficient and as a matter of fact, even the accuracy of the old
mechanical recorders is acceptable. However, for calculating reserves from a short-term production test,
one should use the same gauge on the same trip in the hope at MMP (or lower) for both initial and final
pressures. (What is more important in this case, accuracy or sensitivity? How about repeatability?). This
eliminates the errors in accuracy that occur between gauges. This is important since the tests are often
designed to measure 50 kPa difference in reservoir pressures.
One should not use DST pressures, pressures from a previous test or static gradient and assume them to
be equal to the pressure prior to the production test. Even with the greatest of care in correlating initial
pressures from DSTs or static gradients to a datum, errors can creep in. In an unpublished study at
Chevron, I looked at 30 production tests where the wells later went on production and found that reserves
were between 0-50% pessimistic except one case that was 100% pessimistic. Strangely enough, ALL
cases were pessimistic indicatingwhat??? DSTs are affected by supercharging? What else?

Part 2 Flow Rates & Sampling


Rates
1.
2.
3.
4.

Instantaneous
Average
Final
Stabilized

Instantaneous Rates - Gas


Qgi = rate measured at surface
Note: Each type of orifice meter will have its own equation and chokes have different equations than
orifice plats.
The general equation for a choke is
Qi = Ki C P
(GT)1/2
Ki = various correction factors (most are very near 1.0)
C = choke coefficient
P = absolute upstream flow pressure (Average atmospheric pressure in Alta is 93 kPa)
T = absolute temperature = Celsius + 273 or Fahrenheit + 459
Note: Upstream pressure must be at least 2x downstream pressure to maintain critical flow through the
choke.
Choke Coefficients
Inches
1/8
3/16

5/16
3/8
7/16

5/8

mm
3.175
4.7625
6.35
7.9375
9.525
11.1125
12.7
15.875
19.05

metric coeff
19.164
44.277
81.287
133.812
187.686
261.032
345.63
551.132
800.268

imperial coeff
6.25
14.44
26.51
43.64
61.21
85.13
112.72
179.74
260.99

Assuming a six-inch bean choke with 0.7 gravity gas and 32 degrees Celsius, the equation becomes:
Qg = C P
13 409

in m3/d for P in kPaa


C from above table

Average Rates
Liquid: qav =

total recovery during time period


total time of flow period
(most production tests average liquid rates over hour periods)

Gas:

qav =

(qtf x tf)
total time

volume (running volume is listed on most reports)


time

Final Rates
The final rate is measured just prior to shutting in. This is an average/instantaneous rate.

Stabilized Rates
A stabilized rate on a PT simply means that one does not change over the last hours of a test. It may
fluctuate but should not show an increasing or decreasing trend. (Do not confuse this with the stabilized
rate that is calculated from an assumed area of drainage and is assumed to apply over long periods of
time.)
Effective Flow Time sometimes called Horner flow time.
teff

Volume Produced
Final Flow Rate

From the definitions of stabilized rate, final rate and effective flow time, what would be the effect on teff (as
compared to tactual) of changing the rate just prior to shutting in?
Effective flow time was mostly used before the advent of computers when superposition time was timeconsuming to use. In the case where a flew had large fluctuations or a rate change in the first third of the
period, it was found that the effect was small and could easily be corrected for using effective flow time.
Note: A common production test report contains gas rates that are _______________ rates and liquid
rates that are _______________ rates. A well test analysis of a gas well uses the __________________
Instantaneous gas rate but corrects the flow time based on the total ______________.

Accounting for Rate Changes


It is possible to mathematically account for rate changes made during the main flow period but the
correction is not perfect. The closer the rate changes are to the beginning of the flow period, the smaller
the effect on the buildup analysis. Therefore, it is recommended that any choke changes made to yield
an acceptable flow rate for the test be made during the first 1/3 of the flow period.
Note that changes made in the last hours of the main flow are the most difficult to account for PLUS they
can also suffer from well bore storage and therefore be quite inaccurate. Often, these changes are
ignored and the previous rate is extrapolated to where it intersects the buildup for the best estimate of
skin factor.

Sampling
DST sampling is done by catching a flow sample between two valves at the end of the final flow. This is
one of the best means of catching a flowing sample but DST flows are often contaminated with mud and
filtrate. Offshore techniques have evolved to the point where they flow at an extremely low rate for a
number of hours and then catch the sample at a pressure just lightly below reservoir for the most
representative conditions of all. The main problem with DST sampling is retrieving the fluid since even
mercury displacement does not retrieve all the liquid.
Most PT gas sampling is done close to, but downstream of, the separator so that the gas is still at
separator conditions. This is the most representative place since the gas can then be recombined with
separator liquids to recreate actual in-situ fluids. Other sampling locations are secondary to this one.
Liquid PT sampling is mostly done on the separator as well. This too is the most representative spot.
Note that samples that are taken into evacuated cylinders cause liquid portions to flash and may be
unrepresentative of actual flow conditions.
Note that (except in very special circumstances) liquid samples from separators flash some gas to the
atmosphere as they are being taken into non-pressurized containers. Separator shrinkage is not
commonly measured in Canada.
H2S should be measured with Draeger/GasTec tubes or Tutweiller with lab rep on site for best results
especially if concentrations are low. H2S can plate out on the inside of steel sample cylinders and
obscure 0-100 ppm readings.

Part 3 Pressure Charts


How to Preview and Evaluate Data-Quality on a PT Chart
Are the draw downs and buildups smooth?
Was the drawdown pressure smooth for at least the final 2/3 of the flow time or were there
choke changes just prior to shutting in?
Any other obvious anomalies?
Can the anomalies be explained? Do the anomalies correspond to actions recorded on the
production report? Are there special considerations like multiphase flow, injection test etc?
What does temperature do? Does temp fall when gas is flowing? Does it rise during the shut ins?
Flow characteristics.
Does the pressure stabilize or does it continue to decline? Does it decline linearly signifying
depletion?
Does flow pressure increase signifying cleaning up loading up?
Build Ups:
How quickly does pressure build during shut-ins?
Does it build quickly at first then more slowly?
Does the second buildup occur more/less quickly than the first?
Does the second buildup appear to aspire to the same pressure as the first?
Does the second buildup have a lower shoulder than the first?
Does either buildup appear to go into decline?

Rising and Falling Liquid Levels Between RRD and MPP


A pressure-difference plot simply measures how well two recorders track. It helps to ascertain whether
certain pressure fluctuations are related to actual conditions in the well bore or to gauge
peculiarities/malfunction. One of the most common fluctuations seen is a sharp change in the pressure
difference caused by a change in the fluid that separates the gauges.

Pressure Difference = Bottom Top

The above chart indicates that the fluid between the gauges at the
beginning of the shut in was water, then oil and finally gas.
Because of the top-gauge pressure was subtracted from the lower-level
gauge pressure, we know that the liquid level was rising/dropping. A
dropping liquid level only affects the gauge pressure when the level is
between the gauges and the point of entry into the formation. Why?
By finding the point in time where the level drops past the gauges (from
the above plot) one can review the data after that point to see where
the new gradient may stabilize and the data may be legitimate for
analysis again. (In above case, the gauges were within 0.1 meter of
the mid-point perforations and within 2 meters of the bottom of the
perforations).
What liquid was between the gauges prior to the main flow? Is the
pressure prior to the test comparable to the final buildup pressures? If
the buildup had only been held 300 hours, would the pressures be
comparable? What is the most likely reason the flow pressure
increased? What evidence of this would be seen after the buildup
say on a static gradient?
Would a static gradient run to MPP be an aid in confirming the final
gradient?
Why might such a gradient show oil or water near the
bottom. (Hint, how does one get a pressure gauge into the tubing at
surface. Does this involve flowing a small amount of gas?)

The bold data is


the lower gauge.
The top gauge
sees the liquid
level drop by first
(it rises first).
This is how we
know the liquid
level is dropping.
Why does the
pressure rise at
a different rate
after the liquid
level has fallen
past the gauge?

Above is a blow up of the place where the second liquid level drops past the gauges.
Hint: What would the pressure be doing at the liquid entry point? Building_________.
How would the gauge pressure differ with a liquid gradient between the two?
How would it differ with a gas gradient between the two? How would it differ during transition?

What to recognize on PT Pressure Chart

Vertical scale does not start at zero.

10

Vertical scale starts at zero.


High Perm zones have 1) high flow rates, 2) small, flat draw downs and 3) quick buildups. Make
sure to view the overall scale so that zoomed portions like the top chart do not make the drawdown
appear relatively larger than it is.
What to recognize on PT Pressure Chart continued

11

High skin zones have 1) low flow rates and 2) large but flat draw downs, and 3) quick buildups.

Low perm zones have 1) low flow rates, 2) increasing draw downs, and 3) more rounded corners plus
slower buildups. (Check the vertical scale.)

12

What to recognize on PT Pressure Chart continued

Frac in Higher Perm

Frac in Low Perm


Hydraulically fractured zones build quickly initially then slow down. Some (lower perm) may appear to
deplete. Only further analysis can determine depletion.

13

Parallel boundaries. Builds up quickly at first then slowly. Why? Pseudo-depletion? = Yes.

U-shaped boundaries.
What is the general characteristic that fraced zones, radial composite and parallel/U-shaped
boundaries share?

14

What to recognize on PT Pressure Chart continued

Radial composite zone with low contrast.

Radial composite with high contrast. Flow shows depletion of high perm zone (only). Builds quickly at
first (due to the high perm) then slowly. Can be difficult to distinguish from hydraulically fraced zones or
parallel boundaries. Buildup can also mimic depletion (called pseudo depletion).

15

What to recognize on PT Pressure Chart continued

Depletion can be seen by the drop in final buildup pressure and sometimes by a linearly declining flow
pressure. What can one do to distinguish simple depletion from interference? Run shut in long enough
to see pressure decline.

Constant pressure boundaries. Unlike depletion, buildup returns to initial pressure. What if you do not
know initial pressure? What will flow pressure do? What will rates do? What is the down side?

16

Part 4 Pressure Transient Analysis


Pressure buildup analyses (from a shut in) are derived from pressure drawdown (flow) equations.
Pressure drawdown equations are based on the fact that when fluid is removed from a reservoir, the
pressure will drop. The initial correlation between pressure drop and time for a constant rate, in the most
simple reservoir, is the exponential integral (EI) function which is very very closely approximated by the
natural logarithmic function (ln) for times large enough to be meaningful in a well test. The magnitude of
the pressure drop depends upon the initial pressure, the flow rate, the formations ability to flow
(permeability, thickness, damage, fractures, etc.) and the time at which the measurement is made. By
measuring flow rate and pressure drop over time and by independent measurement of other parameters
like thickness, porosity and water saturation (e.g. from logs), the remaining parameters (permeability,
damage, etc.) can be calculated. Two factors that detract from this simple theory are well bore storage
and skin factor. Well bore storage occurs (for example in a gas well) because the flow out of the well
bore initially is independent of the formation (it is due to expansion of gas in the well bore). Skin factor
occurs due to damage or stimulation of the near well bore area that causes an increase or decrease in
the expected pressure drop over that zone compared to virgin formation. These concepts will be dealt
with in detail later. The final correlation between pressure drop and time for a constant rate is called
pseudosteady (PSS) and is linear in nature. Imagine that the pressure wave (transient) has reached the
outer boundary of the reservoir and has become frozen in shape so that the entire wave drops at the
same linear rate throughout the reservoir.
Pressure buildup analysis simply states that when a well is shut in after a flow, the pressure at the sand
face will increase. The existing pressure wave (transient) can be imagined to be anchored at the final
flowing pressure but works its way back into the reservoir and the ln function describes its shape. The
magnitude of the pressure increase in time is dependent on exactly the same parameters as for the
drawdown that just preceded it.

Buildup analysis is preferred to drawdown analysis because the

equations were written for constant flow rates and at no time is the rate more constant than during a shut
in (what is it?). Also, pressures can be measured to tens of kPa accuracy and to a decimal of kPa
sensitivity but rates cannot be measured anywhere near so precisely - so a zero rate (as occurs in a
shut in) is the most desirable number to work with. In addition, it is very difficult to maintain constant flow
rates during a PT, but it has been shown that a buildup analysis will yield very good results by using the
final rate of the preceding drawdown. Analysis results are better the longer the rate is kept steady prior to
shutting in. Computers can correct for changing flow rates throughout a well test but the best analysis is
still done on a well that has been produced at a constant rate like the theory requires.
Pressure draw downs and build ups all go though distinct phases and flow regimes. From the above, one
would already expect to see wellbore-storage dominated flow followed by altered pressures due to skin
factor, then, with luck, we would see the transient radial-flow (logarithmic decline) regime (hours to
months) from where we can determine reservoir parameters.

Later there could be various

heterogeneities and boundaries affecting the data until the pressure wave encounters last boundary and
goes into pseudo-steady state decline (hours to years). An aquifer or gas cap can provide pressure
support and dominate the final PSS regime. Overlap between phases and regimes can occur so that
some phases and regimes are not seen at all.

17

A drawdown transitions from logarithmic decline to linear decline.

The buildup is dependent on the same parameters as the preceding drawdown (its shape is dependent
on the shape of the drawdown but is not the same).

18

Definitions
Permeability Capability for transmitting fluid. Units of millidarcies. Commonly abbreviated k.
In Canada, perm is based on 100 kPa* rather than atmospheric pressure and is
abbreviated K. It differs from the commonly used oilfield permeability, k, by a factor of
1.01325.
Conductivity kh where h is the thickness in meters. Also called productive capacity.
Transmissibility kh/ where is viscosity of fluid in mPaS, (older reports sometimes in PaS).
Skin A zone of altered permeability considered to occur immediately at the sand face.
Skin Factor a term in the flow equation that relates how much pressure change at the sand face is due
to skin.
Pressure Buildup Parameters in Canadian Metric Units
p [=] kPa

pressure

r [=] m

radius

[=] decimal fraction

porosity

[=] mPaS

viscosity

-1

compressibility. Units actually m3/m3kPa

ct [=] kPa
t [=] hr

time (note that most PTs report in hours)


3

[= ] rm /m

formation volume factor (converts surface volume to reservoir


volume)

h [=] m

formation thickness

Z [=] unitless

gas

supercompressibility

factor

(deviation

compressibility from an ideal gas)


By definition ln =

1/t dt

and

EI =

e-t/t dt

Assumptions
1. Single phase fluid
2. No turbulence effects (DArcys law is applicable)
3. Gravitational effects are negligible
4. Reservoir is homogeneous and isotropic
5. Permeability, porosity and viscosity are constant (not dependent on pressure)
6. Compressibility is small and constant
7. Pressure gradients are small
8. Flow is isothermal
9. Flow rate is constant
Which assumptions are likely violated? Why do we use it?

* Canadian STP is 100 kPa and 15C while oil field STP is 14.696 psi and 60F.

19

in

actual

Infinite Acting Radial Flow (IARF) also calledTransient Response


The general flow equation IARF conditions is
pD = 0.5 lntD +0.40454 +s

(This formulation uses total skin factor)

Canadian Metric Equations for Infinite Acting Radial Flow (IARF) reservoir (transient response)
Liquid
p = pi pwf = 2121 q [log(

Kt

) - 5.09 + 0.87 s]
2

(ctrw )

Kh
Gas

p2 = pi2 pwf2= 1491 quZT log(

Kt

) - 5.09 + 0.87 s]
2

Kh

(ctrw )

What is important here? The pressure difference increases logarithmically in time for an infiniteacting radial flow reservoir.
Why would liquid have -5.09 and gas sometimes have -2.09? Hint: Think viscosity. What other number
would change? Why is the log used instead of ln? Note that 1842 = 2121 x 2 / ln(10)
Canadian Metric Equations for a Bounded Reservoir (pseudo steady state response)
(Circular reservoir, modified to account for average reservoir pressure pR rather initial pressure pi and
since time was not involved, not converted from In to log function.
Liquid

p = pR pwf = 1842 q [In re


Kh

- 0.75 + s]

rw

Gas

p2 = pR2 - pwf2= 1295 quZT [In re


Kh
Note:

- 0.75 + s]

rw

Most modern gas analyses use pseudo pressure () rather than pressure squared.

This is

because it is very simple to calculate and convert to and from pseudo pressure with a computer. One
would rarely calculate such by hand. As a rule of thumb in the past, pressure-squared equations were
used below 21 mPa and simple pressure was used above 21 mPa for gas analysis (except in Canada
pressure squared was mostly used throughout). Pseudo pressure does give the best results because it
is a measure of the integral of (p/Z)dp and therefore includes viscosity and compressibility effects.
Remember most old Canadian PTs are analyzed in terms of p2. What concern is it? Skin factors were
slightly lower (10%) than they should have been.
What is important here? The pressure difference is constant in time during pseudo steady state.

20

Methods of Transient Analyses


Two general methods have evolved for analyzing pressure transient data. The first method, the semi-log
method, is to plot pressure against the log of time and when the plot is linear, permeability can be
determined from the slope of the line. From the equation for infinite acting reservoir (transient response),
the technique is obvious but for buildups, one has to use the theory of superposition (Horner analysis).
DTS analyses mostly still use the Horner method.
The other method is called type curve (or log-log) analysis and it uses the fact that all (perfect) reservoirs
will build up according to the exponential integral curve so that by non-dimensionalizing the variables, the
pressure build up can be plotted and compared to a curve to obtain permeability. To exaggerate the
characteristics of the curve (for easier matching) the plot is done on log-log scales. To account for
wellbore storage and skin, dual variables can be plotted on the x axis and multiple curves are plotted
respectively. To further define characteristics of the curve, the derivative of the semi-log plot is plotted as
well (Gringartens Curves). In addition, many reservoir heterogeneities can be easily identified on the loglog w/derivative plot.

Semi-Log Methods
From the drawdown equation
p = pi pwf = 2121 q [log kt - 5.09 + 0.87 s]
Kh

ctrw2

On a semi-log plot of p vs log t, the slope of the line will be


m = 2121 q
Kh
Superposition and Horners Equation
The pressure in a reservoir during a buildup is equal to the drop caused by the previous flow plus an
increase equal to that caused by adding an equal but negative flow rate from the point of shutting in.
(Remember the assumption of constant rate.)
Let

tp = time producing
t = time shut in

21

Note how the rate of (-q) is superimposed onto the original solution after tp to produce the solution for q =
0 which is the condition for the shut-in.
So

pws

= pi pws = pi (pwf - p(-q)) = pi - (ptp+t p)


=2121 q [log k(tp + t) - 5.09 + 0.87 s]
Kh
-

ctrw2

2121 q [log kt) - 5.09 + 0.87 s]


Kh

ctrw2

pws = 2121 q log tp + t


Kh

Horners Equation

As time increases, what does (tp + t)/ t do?


At t = infinity, what is the value of (tp + t)/ t? What is the Log of that number?
What is the definition of pws at t = infinity?
Horner Analysis
Plot pws vs. log (tp + t) and obtain a straight line with
t
m = slope = 2121 q
kh
So by measuring the slope, one can calculate k.
Note that the intercept of the ordinate is Pi. What buildup time does the intercept signify?

22

Equations Associated with Horner Analysis


Transmissibility

kh = 2121q

Conductivity

kh = Transmissibility x

Permeability

k = 2121q

= conductivity/___?

Transmissibility x ___?

hm
for gas with p2

k = 1491qZT
hm
Skin

s = 1.151 [pi-pwf - log ( kt ) + 5.09]


ctrw2

m
Radius of Investigation

rinv = 0.00389 [

kt ]1/2
ct

What time/time function should be used in rinv?


Time to Stabilization

tpss = 6.94 x 104 [ctre2]

re is 402 m for oil, 804 m for gas

k
(a.k.a time to pseudo steady state for a circular reservoir)
What is important here? Horner analysis is simple and therefore not overly precise when it comes
to heterogeneities (i.e. comparisons of plots by eye are difficult). It is used mostly on drill stem
tests which are also simple and imprecise.

23

Interpreting the Semi-Log Plot

Early Time

Middle Time

24

Late Time

B,D,I First parts of E,J K L = similar


C,G,H, Last parts of J,L = similar
K, stratified layers = NOT.
How would measuring initial pressure help distinguish between D & I ? What is main difference in
character between D & E?

25

Dimensionless Parameters and Type Curve Matching


If all reservoirs behaved according to basic theory we would only need the exponential integral function
(which is very closely approximated by a logarithmic function at any other than extremely early times). If
the EI function was non-dimensionalized, it would be the type curve. Analysis of pressure transient data
would be as simple as plotting the actual pressures against time on the proper scale, matching the plot to
the type curve then backing out the parameters from any matched point using the definitions of the
dimensionless parameters below.

See Appendix 1 for an elucidation of these dimensionless

parameters. Note that type curves use log-log axes to give the buildup data a more easily matched
(exaggerated) shape.

Dimensionless Pressure

Practical Can

Darcy

Metric Units

Units

pD

Khp

2 khp

1842 q
Dimensionless Time

= 3.6 x 10-6Kt =

tD

kt

ctrw2
Dimensionless Wellbore Storage =

CD

Cs
cthrw2

ctrw2
=

Cs
2cthrw2

EI Function

The practical Canadian units can be used to convert a match point back to Canadian permeability (using
Canadian metric oilfield units). DArcy units are the ones that were used in constructing the curves.

26

Lets check. The equation for flow in IARF reservoir was


p = pi pwf = 2121 q [log Kt - 5.09 + 0.87 s]
ctrw2

Kh

And we know that 2121 = 1842 x ln(10) / 2 so


p = pi pwf = 1842 ln(10)
2
Khp

q [log Kt - 5.09 + 0.87 s]

= 0.5 ln(10) x [ log ( 3.6 x 10-6 Kt ) log (3.6 x 10-6) -5.09 +0.87 s ]
ctrw2

1842 q
Khp
1842 q
so that

pD

ctrw2

Kh

= [0.5 ln (3.6 x 10-6 Kt

) + 0.4045 + s]

ctrw2
= 0.5 lntD + 0.4045 + s just like the known general flow equation for IARF

Which makes one wonder which came first, the chicken or the egg.
What is important here?

The Ei type curve is the basis for all radial flow type curves.

27

Equivalent or Agarwal Time


In order to analyze the buildups against curves that were generated for a flow period, the derivative can
be calculated with an equivalent time or Agarwal time defined by
tequiv = tf x t

Equivalent or Agarwal time

tf +t

This is based on the assumption that the radius of investigation during the flow period is set by tf and can
only be duplicated at infinite shut in time. This method therefore back plots the shut-in pressure to a
corresponding inverted flow pressure. (This exemplifies why not to expect good results from short flow
tests even if they are followed by long shut-ins.) The correction has debatable tenets but is sufficient to
make derivative plots analyzable using drawdown curves. However, when the method first came out,
many analysts used it correct the buildup time but then they also changed the axis on the log-log plot to
Agarwal time thus making a double correction. This caused the data to curve upward and was very
noticeable in later time. Overall accuracy of the analysis was not affected since computers generate type
curves based on the preceding flow period and match the raw data using the same time function as for
the generated curve. Thus, no matter what time function is used, the match was still the same. This
occurred between about 1998 2001. Current practice is to use the Agarwal derivative plotted against
elapsed time. The benefit is that the final plot looks more like the textbook (drawdown) plot for the
assumed model (e.g. a channel or linear flow remains at a slope of 0.5 rather than continuing to curve
upward).
What is important here? We have a correction for shut in times that allows us to analyze buildups using type curves generated for draw downs. And we even know how to use it properly now.

28

Parallel boundaries calculated with Agarwal time and plotted against Agarwal time.

Parallel boundaries calculated with Agarwal time and plotted against elapsed time.

29

Deviations from the Theoretical


(or Welcome to Reality)
Buildup data is affected in early time by wellbore storage and skin. It can also be affected by well bore
dynamics (mostly liquid fallback and other phase segregation). It is affected in late time by boundaries,
depletion, interference from other wells, pressure support (aquifer) and even earth tides. If the recorders
are not at the same depth as a liquid level in the formation, the data can be affected at any time by rising
and falling liquid levels between the recorders and the formation. To add insult to injury there are cases
where wellbore storage runs straight into boundary effects so the pressure response never actually
follows the expected curve.
Whats important here? Life aint easy.
Well Bore Storage
When a well is opened to flow at surface, the initial flow comes entirely from the wellbore. As the wellbore
begins to deplete, the pressure progressively drops and fluids influx from the formation at progressively
higher rates. When, the influx rate catches up to the surface rate, well bore storage is said to have
ceased. Note that there will be a well bore storage effect every time the surface rate changes. Why?
Conversely, when the well is shut in at surface, flow into the wellbore does not stop immediately, it
continues until the compressibility of the wellbore fluids is near that of the formation. Well bore storage
either depends on the volume of the well bore and the compressibility of the single fluid in it or on the
area of the well bore and the gradient of the liquid that is at the sand face.
Cs [=] m3/kPa [=] cm3/atm

Single Phase WB

Moving Liquid Level

Cs = cf x Vwb

Cs = A/f

cf is compressibility of fluid in well bore in kPa-1


V is volume in wellbore in m3
A is area of well bore in m2
f is gradient of fluid in kPa/m
At the beginning of the flow, i.e. for pure well bore storage, the pressure change in Darcy or non-practical
units is:
p = pi pwf = qt/Cs
Note that this equation in not dependent on K.
What is important here? No analysis can be done on the pressure data when flow or buildup is
dominated by well bore storage.
We can see from the above equation that in DArcy units
pD = tD/CD so that log(pD) = log(tD/CD)
If one were to plot pD vs. tD/CD on log-log scales, the above indicates that pure well bore storage would
have a slope of 1.0.
Note: If a buildup was not run long enough to get past well bore storage and produce a Horner straight
line, the test is said to have insufficient curve development for extrapolation or analysis. (You see this
comment a lot on DST analyses.)
What is important here? The slope on a log-log plot of pD vs. tD/CD is 1.0 during pure wellbore
storage.

30

Skin
Skin is an altered permeability zone around the wellbore. Positive skin causes an increased pressure
drop near the wellbore and indicates damage. Negative skin causes a decreased pressure drop around
the wellbore and indicates stimulation.
Positive skin, or damage, can be caused by the motion of a bit crushing the formation, by mud (actually it
is most likely the mud filtrate) or working fluid (e.g. during completion) invading the formation and either
changing the wettability or reacting with the formation or formation liquids to form precipitates or by
swelling clays. This damage may indeed be a skin (concentrated at the wellbore) or may extend out as
far as a meter or so into the reservoir whence it becomes known as deep damage. Skin can also be
caused by fines migration that conglomerate near the wellbore but this phenomenon occurs much less in
nature than is suspected (fines migration can only occur when the wetting phase of the rock is flowing!).
Negative skin or stimulation can be due to natural or induced fractures or any other process (acidizing,
hydraulic fracturing) that serves to improve flow through the formation near the wellbore.
Skin factor has a mathematical definition that relates it logarithmically to the flow area into the wellbore
but that turns out to be less important than understanding the concept of skin, its components and how
skin number can be used.
Definition of Skin
Mathematically
(pD)skin =

pskin

Practical Can

Darcy

Metric Units

Units

1842 s q

s q

kh

2kh

Above is the Van Everdingen 1953 formulation modified to include .


Measurement of Skin
Semi-Log Method for a drawdown
m = slope =

and

2121 q

ps = m s

Kh

1.151

Consider that
pactual =
=

pi - (pwf)actual = pi - (pwf)theoretical + pskin


(pi - pwf ) - ps = m(log

kt

-5.09) + m s

ctrw2
So

= 1.151 [pi-(pwf)actual - log

Kt

(Using the IARF eqn)

1.151

+ 5.09]

ctrw2

In order to keep the calculation simple, assume t = 1 hour so that


s

= 1.151 [pi- (p1 hour)actual - log

+ 5.09]

ctrw2

m
For a buildup at the last moment of flow
(pi - pwf ) - ps = m(log

Ktp
ctrw2

-5.09) + m s
1.151

31

Skin Equation

For a buildup from the Horner equation


(pi - pws) =

m log tp + t

Note that this is not affected by s.

t
so that
(pws - pwf) = m (log tp t + Log
= 1.151 [p1 hour-pwf - log

ctrw

+ 5.09 + s )

but for t =1, tp t /( tp + t ) ~ 1

ctrw2

tp + t
s

+ 5.09]
2

Whats important here? The semi-log method uses the difference between actual and theoretical
drawdown pressures to calculate skin factor.

Log-Log Method
The log-log method for determining skin factor is to draw type curves with various values of skin in the
flow equation and match the actual buildup data to the curve.
Components of Skin Factor
Be sure to read the superscripts and subscripts when noting the skin factor. Superscript prime (s)
indicates total skin at the wellbore while s by itself indicates mechanical skin (i.e. removable damage at
the wellbore). Subscripts and methods for estimating the various factors are as follows:
Sit = skin due to inertia/turbulence. Is non-removable. What happens when fracd?
= 2.99 10-4 G/(k1/3rwh)
Sa = skin due to deviated formation or slant drilling. (positive or negative value?)
= (a/41)2.06 [(a/56)1.865 log(0.01h/rw)] a = degrees well to vertical
Sp = skin due to partial penetration. A frac can certainly remove it!
= (h/hp-1)[In(h/rw) (kh/kv)0.5 -2]
Note that s = s + sit + sa + sp
Spr = pseudoradial skin. Is a number calculated by comparing the logarithm of the area of a frac against
the original well bore area. It assumes the frac has infinite conductivity. It is the maximum value the post
frac skin can have for a given frac length.
Spr = In (2 rw)

Maximum Expected Skin for Hydraulic Frac

( Xf)
Sf = frac face skin. Only use I have seen for this is to determine probability of frac face damage verses
choked flow in the frac. For sf > 0.2, suspect choked flow.

32

Skin Effect
For wells that have changing transmissibility away from the wellbore, it has become prudent to consider
the near wellbore transmissibility in combination with the true wellbore skin factor and judge the effect of
both on the outer or controlling transmissibility. The combination thus construed is termed the skin effect,
se. One way to estimated se is from a Horner plot much in exactly the same fashion as a wellbore skin
factor. By placing the straight line on the latter (outer) transmissibility portion and calculating skin factor
as usual, the skin effect is estimated. Note that many wells with very low outer transmissibility are not
(cannot practically be) tested long enough to actually define how low that transmissibility is.

When

analyzed, the log-log plot simply shows a transition to tighter conditions (an upward slanting derivative)
and the analyst can only estimate the outer parameters for his model. In this case, a maximum outer
permeability with an attending maximum (but negative) skin is often used to estimate the skin effect
usually read from the Horner plot.

If prevailing transmissibility is known with any certainty, a more

accurate method to estimate the skin effect is to vary the skin of an IARF model until the forecast matches
the radial composite (or other) model with the same outer or prevailing permeability. Accuracy still fails if
prevailing perm is not known, of course, but both methods are useful to indicate the futility of stimulation
since the measured skin effect estimates often range from -4 to -6 (and the actual skin effect has to be a
more negative number why?). Note that any data that shows an increasing derivative on the loglog plot is already stimulated with respect to the prevailing permeability. The chosen model should
be production forecasted to see the effects of stimulation before such is recommended to operations.
How Skin Affects Production
Skin ranges from about -7 to 200. Until recently, they rarely got below -5 and still rarely get above 20. A
skin of -7 indicates a good set of multiple hydraulic fracs on a horizontal well. A skin of -6 to -5 indicates
a good hydraulic frac in tight reservoirs, a skin of -4 indicates a good hydraulic frac in permeable
reservoirs and a skin of -2 indicates a good acid job. A skin of down to -1 can be associated with natural
fracturing. Most non-hydraulically fractured zones will have a skin between -2 and 10. Drilling and
completions damage rarely go above 20. Skins of 200 and more can occur on extremely permeable gas
wells although the cause is not known, stimulation does not appear to remove them. Up until recently,
this was about all the skin factor told us. Now, however, the rule of eights can be applied to a PT analysis
to estimate the post stimulation production rate (calculated at the same flowing pressure) in the following
manner.
Rule of Eights:
qstim

qpt x 8 + spt

where sstim = -2 for an acid job


= -4 for a frac above 1 md

8 + sstim

= -5 for a frac below 1 md


= -7 for multiple fracs on Hz, below 1 md
Remember: In the cases where the permeability changes away from the wellbore (radial composite
reservoirs), the skin factor at the wellbore cannot be used with the above equation. Rather, the skin
effect (one based on only the outermost permeability) should be used. Why?

33

The Derivative of the Flow Equations


(or how to make a geologist go to sleep.)
The Derivative of the General IARF equation
Formation - IARF
pD = 0.5 lntD +0.40454 +s
dpD/dlntD = 0.5

Yes, it is that simple. But notice that is the derivative is w.r.t. lnt. Also,

dpD/dtD = 0.5/ tD since dlntD /dtD = 1/ tD

(by definition ln =

1/t dt

and

EI =

e-t/t dt )

The Derivative of Well Bore Storage Flow


Wellbore Storage Flow
pD = tD/CD

so

dpD/dlntD = dpD/dtD dtD/dlntD = 1/CD

tD = tD/CD = pD which is a line of slope = 1.0.

What is important here? Both well bore storage and the derivative of well bore storage have a
slope of 1.0.
Old Style Type Curves
The EI type curve was expanded upon to account for varying well bore storage and skin factors. This
lead to a series of curves that moved to the left as well bore storage increased, and got taller as skin
factor increased (Ramey curves). Even with the log-log scale exaggerating the characteristics, these
plots were awkward to use because matches had to be made in two dimensions and then compared (in
goodness of fit) with other matches (from memory).
Whats important here? The old style type curves are difficult to use but easy to convert to
permeability and skin once the match is made.
Gringarten Style Type Curves
Gringartens series of curves uses tD/CD on the x-axis so that a single line of WBS (unit slope) suffices for
all curves.

To incorporate effects of skin, the functions are plotted are for various values of CDe2s.

Obviously, the arithmetical conversions from match point back to permeability, skin and well bore storage
from a Gringarten curve are a little more complicated than the simple dimensionless parameters, but are
still easy enough to accomplish easier yet with a computer. To top it all off, Gringarten added the
derivative curves that belonged to each of the CDe2s curves. Remember that the derivative curve is
actually the derivative of the Horner Plot (dpD / dlnt). The derivative curves all have a slope of 1.0 at the
beginning of wellbore storage and all flatten out to 0.5. During transition from WBS to radial flow the
derivative plot is much more sensitive that the pressure plot and can be used to refine the match. Note
that with the unit slope of well bore storage locked and with the final level of the derivative at 0.5 locked,
matching is a two step process.
What is important here? The most widely used type curves are Gringarten style and they sacrifice
simplicity-in-calculation for simplicity-in-matching. This is a good trade.

34

35

Log-Log Derivative Plots


In the examples of derivative plots that follow, where the derivative plot stays horizontal, the Horner plot
will have a straight line (which can indicate infinite acting radial flow) and thus can be analyzed for
reservoir parameters.

From there, the derivative can only rise, or fall or level off again or some

combination of the three so many reservoir models have the same characteristic derivative curve.
What will the Horner plot be doing if the derivative plot has a slope of ? What will the Horner plot be
doing if the derivative plot is curving upward?

General Observations of the Log-Log with Derivative Plot


Note: Many production tests are designed to see boundaries or other late time effects. These are most
accurately modeled from the log-log w/derivative plot.
After wellbore storage effects have dissipated and IARF is seen, a further upward trend in the derivative
plot usually indicates something bad. It means that the well is failing to maintain the flow pressure
expected for an infinite acting radial reservoir yet may not be suffering depletion. (How does flow
pressure fit into the argument? Hint: Can the Horner plot reach a pressure higher than pi?). If the
derivative curves upward then stabilizes again (at one-half or one-quarter the original permeability), the
usual model is either one or two (right-angled) boundaries. If it stabilizes at something other than or
original transmissibility, the model is usually radial composite. (An equivalent model to radial composite
would be intersecting boundaries at varying degrees). If the derivative shows a or slope, the model
may be a narrow channel, a U-shaped reservoir, or a hydraulic frac (all of which incur linear flow) or it
could be a radial-composite reservoir (with very low outer kh). Of these preceding models, only those that
indicate a secondary stabilization of the derivative plot (radial composite, intersecting boundaries,
hydraulic fracture) will go on to deliver at a stabilized rate. The others (parallel or U-shaped boundary)
will continue to decline more quickly. (Why? Do any reservoirs actually stabilize? Hint: What does PSS
stand for?). A properly designed production test should see if a well will stabilize at commercial rates.
A curve downward in the derivative plot can be good or bad. It means that the flow pressure was held
artificially high (compared to expected) for a given permeability, skin and IARF or that the reservoir
pressure dropped in relation to the flow pressure.
Thus, it can mean increasing kh away from the wellbore (radial composite again) or it can indicate
pressure support (top or side gas, bottom or side water). By a quirk of analyzing buildup pressures, it can
also indicate depletion. Depletion can often be confirmed by a loss in reservoir pressure over the test
and/or by a linear decline in the flow pressure during the test so some distinction can be made between
possible models.
Two specialty cases are where the derivative falls off then returns. If the fall off and return are continuous
and smooth and return to the same level, the reservoir may have dual porosity (secondary porosity i.e.
natural fractures or a perm streak between). If the fall off is complete and the return is a separate
occurrence, the reservoir is compartmentalized with a partial barrier between compartments, (which is a
form of secondary porosity but with the added partial barrier and possible different permeability). One
can start to see the complexity of well test analysis if one were to imagine only seeing parts of these

36

models i.e. only the drop or only the return. They could very easily be mistaken for entirely different
models.
Radial composite reservoirs (where the permeability can go up or down at increasing radii) can be made
to match almost any log-log derivative plot and can sometimes serve a catch-all (used simply to model
the curve for a good pressure extrapolation without truly describing the reservoir).

It has be my

experience that, even when such a model is not correct according to geology, it will give an accurate
pressure extrapolation and an accurate production forecast if a) the test sees far into the reservoir and
b) the model fits the test data well (assuming good data to begin with). Even after ten years, I would
appreciate any feedback that tends to confirm or deny this assertion.
Ghost Boundaries
When a dropping fluid level passes the gauges on the way to the formation entry point but does not get
there by the end of the test, it is possible to see a change in the derivative plot (an upturn or an upturn
followed by a downturn) that could be interpreted as a boundary. This boundary can appear in one place
on one test but in another place (or not at all) on another test. If the level started above the gauges, it
helps to check the difference plot but if the level started between the bottom gauge and the entry point,
there would be no way to discern the falling liquid level from a true boundary.

We are presently

researching this phenomenon.


Below is an example of a rising liquid level (the gauges were below the mid-point of the perforations)
where the speed of the rise was too quick to be interpreted as a boundary but did indeed show an upward
turn in the derivative.

37

Examples of Different Models on Log-Log with Derivative Plot

Infinite acting radial flow IARF.

Infinite acting radial flow with well bore storage.

38

Infinite acting radial flow with well bore storage and positive skin.

Infinite acting radial flow with well bore storage and negative skin.

39

Single no-flow boundary. Note: The higher the derivative rises, the lower the effective permeability.

L-shaped no flow boundary.

40

Radial Composite. M = 0.5.

Radial Composite. M = 0.25

41

No-flow boundaries intersecting at 30 degrees.

Radial Composite with R=L for 30degree faults above plus M = 30/360 = 0.833.

42

Hydraulic frac. Higher perm is measurable.

Hydraulic frac. Lower perm is estimable.

43

Hydraulic frac. Perm is too low to measure (buildup is too short even at 1000 hours).

Parallel Boundaries
What is main different between the two models above?
How would one analyze a frac in parallel boundaries?

44

Bounded reservoir, irregular boundaries.

Bounded reservoir circular or square (all boundaries seen at once).

45

Single constant-pressure boundary.

Parallel constant-pressure boundaries.

46

Radial composite increasing kh/. Made to look like depletion or constant pressure boundaries.

Radial composite model that levels out at a higher outer permeability level.

47

Similarity of Models
From the preceding it can be seen that many models have the same derivative characteristics and indeed
have similar pressure responses. One of the more obvious equivalencies is intersecting boundaries at
various degrees with radial composite of various outer-permeability. Both models start out as infinite
acting with a measurable near wellbore permeability. After transition in the case of the boundaries, the
second plateau (or permeability equivalent) is determined by the number of degrees between the
boundaries as compared to a circle.
In a less obvious case, a hydraulic fracture can look like radial composite because it sees frac perm first
then transition (linear flow) to virgin permeability. If the radial composite model can be made to fit the
linear flow, as it often can, then the models are very similar.
Some similarities in models are listed below:
Upward acting then flattening out of derivative (early IARF may not be seen)
-

boundaries intersecting at varying degrees

radial composite

hydraulic fracture

large negative skin factor

Upward acting derivative


-

parallel boundaries, U-shaped reservoir

possible effect of offset injection (reservoir pressure rising)

early part of intersecting boundary

early part of radial composite with very low outer permeability

early part of hydraulic fracture/ negative skin factor

late part of dual porosity

Downward acting derivative with a stabilization/return


-

dual porosity (the derivative will return to same level)

leaky boundary (the derivative may return to a different level)

radial composite with higher out perm (drops and levels out)

Downward acting derivative


-

bounded reservoir

pressure support (gas cap, water drive or constant pressure boundary)

compartmentalized reservoir

interference from offset production

early part of dual porosity

early part of leaky boundary

early part of radial composite

48

Equivalency of Production Forecasts


It has been noted over hundreds of analyses at AStech, that if a model fits the test data most
importantly in the later time the production forecast based on the model will be equivalent to the
forecast of any other model that fits the data with the same general goodness of fit. If the model also
correctly predicts where the derivative will go after the test data has stopped, then the forecast will be
accurate. In dealing with the case of the general upward turning derivative, there will be an optimistic
case (the derivative flattens out) and a pessimistic case (the derivative curve continues upward). Based
on test times recommended in this seminar, the optimistic and pessimistic cases, as above, rarely differ
enough to cause economic concern. This may change, however, as prices allow lower and lower rate
wells to be tied in.

Would the case of depletion verses pressure support give similar production

forecasts? What would be the prime difference between the two models? Would this be distinguishable
from a single flow and buildup test?
The later data is the best indicator of the overall ability of the formation to produce long term (the early
time effects dissipate when put on production as fast as they do in the test buildup). When analyzing the
data, the effects of superimposing wellbore storage and other early time effects sometimes can distort the
later match if the buildup is short. If this is the case, the match should be made without considering the
early time effects at all (set them to zero). One can always compensate somewhat for them later for the
sake of esthetics.
One notable exception to this rule in PanSys is the Compartmentalized Reservoir model. In this case,
the size of the additional compartment can be orders of magnitude different without significantly affecting
the model. The size does, however, have a huge impact on the production forecast.
In the hands of a qualified well test analyst, an equivalent model for forecasting purposes is a powerful
tool. In the hands of a neophyte, it can be dangerous.
What is important here? If a model fits a data set (more importantly in late time) then the forecast
is correct for the time limited by the data set. If the model correctly estimates what the derivative
will do beyond the data set, then the forecast is correct. Inherent in this is that if two models fit
the data equally, the forecasts will be equivalent. This underlies the use of radically composite
models to yield correct forecasts even if the model is not geologically correct.

49

Comparison of Frac and Radial Composite Models

Hydraulically fractured reservoir. (Perm = 1 md)

Radial composite model fitted to hydraulically fractured data. (Prevailing Perm = 2.5277x0.3955 = 1 md)

50

Comparison of Forecasts for frac and radial composite models.

Production forecast for fraced reservoir model.

Production forecast for equivalent radial composite model.

51

What to Look for in a Complete Analysis


Quality of surface flow data:
Was the original flow data corrected for the proper gas gravity? If a flow-prover was used, was there any
water/liquid-hydrocarbon production that would lead to overly optimistic readings?
Was there any evidence of hydrates (down hole vs. surface pressures)? H2S mentioned?
Were the final results calculated or computer generated? (Mistakes possible in hand calculations?)
Was water production highlighted? Mentioned?
Any condensate/oil production?
How many stages were in the separator? i.e. were water and hydrocarbon rates measured independently
or just as fluid with a water cut (less accurate)?
Were choke changes made late in the flow period?
Was the zone flowed prior to the test and has that flow time been reported?
Were other flow measurement devices used turbine meters etc?

Did they agree with traditional

methods?
Was there a change in flow rate just prior to shutting in? (Especially on AWS tests. Why so?)
Quality of pressure data:
Any obvious anomalies? Data not smooth? Anomalies explained?
Was a down hole gauge used? Is the data from modern gauges?
Were real time gauges used? (Real time gauge data can STILL be of poor quality compared to down
hole gauges.)
If surface data, how does the calculated BHP compare in general? Is the character totally different?
What is the Pressure/Depth gradient for the formation? (Anything above 10 kPa/m should be reviewed.)
Was data corrected to MPP (gas) or pool datum (oil)?
If two gauges were run, were they compared? If they differ, why was one chosen for the analysis?
Was there the possibility of liquid level motion between RRD and MPP?
Are the log-log plots from two gauges compared? On the same plot?
What does the temperature plot look like? Are there anomalies on it?
What flow rate and pressure were used? Average rate? Was a short term rate change ignored just prior
to the final shut in? Should it have been? Was an appropriate flow pressure used if a short change was
indeed ignored?
Quality of Analysis:
Model:
Was the buildup held long enough to establish permeability?
Was the choice of model in the analysis explained?
Do the log-log plots show the later trend of the model pressure and derivative as compared to the
data or does it stop at the same time as the data stops? If the data has an upward or downward
trending derivative, does the chosen model continue the trend or does it flatten out (or rise or fall
differently) beyond the recorded data?
Are there alternative models to the one chosen? (Usually the answer is yes.)

Were the

alternatives addressed? Would alternatives lead to more optimistic/pessimistic forecasts?


Was the use of radial composite models incurred? Justified?

52

Radius of Investigation:
Was it calculated using shut in, flow or equivalent time? (Best is flow time.)
Is it deep enough for the test to be considered representative of the formation?
Depletion:
If depletion was mentioned, how many of the three indicators of depletion were seen.
1) Most indicative: A drop in pressure between pi and p* of a well run test.
2) Quite indicative: A linear drop in the flow pressure (PSS depletion)
3) Speculatively Indicative: A drop of the derivative on the log-log plot
How and when was the initial pressure measured?
Skin Factor:
Does the author distinguish between skin factor and skin effect?
Are the skin factors extremely high? Skin factors on extremely permeable wells may not be
removable. These wells can carry a high skin factor (as much as 100) and no amount of
stimulation will remove it. It is NOT inertia or turbulence induced (it is often orders of magnitude
higher than IT skin factors). It MAY have to do with the size/number of the perforations through
which the fluid actually flows, but this is speculation.
Does the rule of eights approximately confirm the expected production forecast for a simulated
stimulation?
Pressure:
Is the extrapolation linear? Is it over a long distance? (The longer the flow time the longer the
extrapolation distance.)
Is the flow longer time possibly longer than the tpss? Horner plots should not use a time greater
than tpss.
Rising derivatives (upward curving Horner plots) are difficult to extrapolate accurately.
Falling derivatives are more likely to yield accurate extrapolations than rising derivatives.
General:
Does the author give an estimate of the confidence that should be places on the extrapolations,
model, reserves and forecast?
Was tpss surpassed by the production time (Area calculated?

Area assumed?)?

Note that

modern analyses that solve the flow equations in LaPlace space actually use Pi and therefore
also use the entire flow period when a closed system model is used. One must remember to use
tpss if an infinite model is used so as to not overestimate Pi.
Were estimates of stable rates made and compared to the production forecast to confirm the
validity of each?

53

Does the input appear reasonable?


Gas

Oil

Input Data
= porosity

Less than 5% - Is there flow from the reservoir?


5-10 % = poor
10-20 % = average
20-33 % = good to excellent
>33 % = questionable

rw = wellbore radius

check caliper or hole size


0.10 meter = average

h = net pay

check logs
it is usually difficult to make a commercial well with less than
1 meter of net pay because of the small reserves number.

cfr = formation compressibility

1.87 10-6 -0.415 kPa-1


7 x 10-7 kPa-1 = average

c = compressibility

5 10-7 for water

cg ~ 1/(1.2P)

6 10-7 oil, low GOR


10 10-7 oil, high GOR
T = formation pressure

C = 5 + 0.0273 x meters depth


Rs = G(P(100.0125API/100.00091T)/18)1.205

GOR = gas/oil ratio

20 = low
50 = average (GTS)
300 = high
3

GWR

2 scm/m = average

= relative density

1.0 air

1.0 is water

0.6 for dry gas

0.9 is heavy (26 API)

0.65 average

0.85 is average (35 API)

0.7 solution gas

0.80 is light (45 API)

Note: = 141.5 (API + 131.5) API = 141.5/ -131.5


= formation volume factor

~100/P

1.03 is low (low GOR oil)


1.2 is average
1.5 is high

= viscosity

0.7 mPaS is water


0.008 mPaS is low

0.3 mPaS is low

0.012 mPaS is average

2.0 mPaS is average

0.018 mPaS is high

8.0 mPaS is high

Gas viscosity goes up with formation pressure.

54

Gas

Oil

Does the output appear reasonable?


Output data
k = permeability

rinv = radius of investigation

0.1 md requires stimulation to bed productive


0.5 md is low

2md is low

1 md is acceptable

5 md is acceptable

10 md is good

20 md is good

1 m may not have seen beyond damage


10 m is low
100 m is ok
300 m is excellent

s = skin

above 10 is extreme
10 is high
0 is average
-1 to -2 for naturally fraced reservoirs
-2 to -3 = for an acid job, small to large
-4 = frac in good perm
-5 to -6 = frac in tight reservoir, small to large frac.
-6 to -7 = multiple fracs on Hz well in tight reservoir

55

Current Examples
Following are some excerpts from new and interesting analyses performed by AStech. These highlight
the benefits of good data preparation and close observation plus making use of as much information as is
available.
Why data preparation is important.

The thinner derivative plot was from the upper gauge. It was reading a slightly more erratically than the
bottom gauge as can be seen in early time. However, ONE wild point (the final data point), in conjunction
with smoothing effects, made the derivative appear to go horizontal then drop.
Why knowing the software is important. This is an excerpt from an AStech report.

For the chosen 2-cell model in the original report, the second cell volume had to be
assumed and therefore could not be used in later calculations to reverse out reserves
numbers. The usual case with PanSystem plots is to use Pi to represent extrapolated
pressures (which can be used for calculating reserves). However, when closed-system
models are analyzed, Pi is used to represent initial pressure and a Pav to represent
extrapolated pressure. The 2-cell model is a closed-system model but PanSystem did
not list Pav, which led to the confusion of using Pi in the reserves calculations.

56

Why understanding well bore storage is important.

This is an injection test with a bend in the fall off data at about the 120 hour mark.

This is the log-log plot showing the change in well bore storage when the liquid fell away from the
wellhead at the 120 hour mark causing a change in well bore storage from the single phase type to the
moving liquid level type.

57

A perfect half slope that was dismissed as well bore dynamics by other than an AStech analyst.

A review of the pressure difference plot showed liquid between the gauges before point 1. The half slope
was legitimate. Contrary to other analysis, this well does not have stimulation potential.

58

AWS data. Upper is corrected to MPP. Lower is pure casing pressure.


This AWS test was used to back up the claim, from the previous example, that the well had stimulation
potential. Notice that the derivative of the MPP data continued to drop well into the 500 hour range.
HOWEVER, the pure casing pressure derivative rose throughout the later portion of the test in agreement
with the first test. Astech has never seen MPP corrections to CP that completely reversed the derivative
trend.

59

Multiple fracs in Hz well. Overall skin effect was -6.8. Perm was estimated/measured ?

60

Double hump on multiple fracs in Hz well. Obviously due to liquid level motion.

Liquid fall back before 1. Liquid level (L/L) falls below gauges at 1. Rising L/L flattens derive before 2.
L/L rises above gauges at 2. Quarter slope gives way to half slope. 1989 hours shut in not enough.

61

Optimistic frac model, derivative flattens. K = 0.1108 md. Overall skin effect was se = -7.1.

Pessimistic model, derivative continues upward. K = 0.03 md.

62

Multiple (4) radial changes in transmissibility. Prevailing perm = 0.1108 md.

Getting a little liberal with the derivative curve. Equivalent model of a frac with deep damage. Prevailing
perm = 0.335 md. Just look at the fit of the data

63

Forecast for optimistic case of multiple fracs in Hz well.

Forecast for equivalent IARF model with same prevailing permeability. Skin had to be decreased to -7.1
to match rates and volumes. Which way would skin go if perm was lower?

64

The derivative was rising too quickly to match with any boundaries. Permeability was an estimate.

Note the length and steepness of the extrapolation on the Horner plot. Accuracy of P*?

65

Part 5 Common Reservoir Models in Detail


Although IARF reservoirs are the simplest, they are not the most common found today. Most production
test analyses deal with a rising derivative. This is reflected in the fact that many zones are hydraulically
fractured prior to testing and that most of the reservoirs found today are small or else have close/closeparallel boundaries. Also with small reservoirs comes the possibility of changing reservoir thickness
within the radius of the test. Since the test measures kh/, this is one type of radial composite reservoir.
Radial composite models (with multiple changes in kh, as needed) can be used as fit alls for difficult
analyses but should be used with caution.

Infinite-Conductivity Hydraulically Fractures


The infinite acting analytical solution for an infinite conductivity frac (IAASicf) is
pD = ( tDXf)1/2 [erf(0.134/ (tDXf)1/2 +erf(0.866/(tDXf)1/2 ] - 0.067Ei(-0.018/tDXf) - 0.433Ei(-0.750/tDXf)
assuming constant rate draw down and no well bore storage or frac face skin.
=

Dimensionless Time based on Xf

tDXf

= 3.6 x 10-6Kt =
ctrXf2

kt
ct Xf2

Where Xf = frac length (m)


Linear Flow
For tDXf <0.016, the erf tends to 1.0 and the Ei tends to 0. So
pD = ( tDXf)1/2 + sf

with the addition of sf which is frac face skin.

The linear flow equation has a slope of on a log-log plot.


Root Time Analysis
p = pi pwf = 1842 q [(3.6x10-6 Kt )1/2 +1/2 + sf ]
Kh

( ctXf2 )

Shows that a plot of p vs. t1/2 will yield a straight line for linear flow but is not unique in K and Xf.
(Remember that Xf is now variable where rw was constant so K cannot be determined from the slope nor
from a tDXf match.)
The Derivative of Linear Flow
dpD/dlntD = dpD/dtD dtD/dlntD = ()1/2 tDXf-1/2 x tD = ( tDXf)1/2 = pD

ignoring skin

The derivative also plots as a line with slope on a Gringarten type log-log plot but is cycle
offset to the right of the pressure plot.
Pseudo Radial Flow
For tDXf >3, the IAAS solution becomes
pD = 0.5 lntDXF + 1.1 +sf

or

pD = 0.5 lntD + 0.40454 + spr +sf


where spr = ln (2rw/Xf)

actually 2.005rw/Xf

What is important here? The skin factor improvement for an infinite conductivity frac (i.e. the best
improvement that one can hope for ) is spr = ln (2rw/Xf) .

66

67

Well Bore Storage


There may be some unit slope on the log-log plot but linear flow usually dominates even the early time.
Log-Log Analysis
For type curve analysis, both the pressure and derivative plot of linear flow have a slope of . The
derivative plot is offset cycle to the right of the pressure plot. In later time, the plot devolves to radial
flow where permeability can be measured.

Once permeability is known, frac half length can be

determined.

Finite Conductivity Hydraulic Fractures


The infinite acting analytical solution for a finite conductivity frac (IAASfcf) is just too complicated to merit
study. Brockenbrough developed a trilinear flow equation that was easily solvable in LaPlace space with
the Stehfast algorithm and it agreed with previous numerical and Greens & Source functions solutions so
special cases of that equation will be presented. Note that although flow starts with linear compressible
flow in the frac it is of such short duration that it is of no consequence. The three flow regimes that
actually manifest themselves chronologically are bilinear flow, pseudo linear flow and pseudo radial flow.
Bilinear Flow for an infinitely long frac
Frac Bilinear Flow (linear flow down the frac and linear flow into the frac from the formation).
The IAASfcf reduces to
pD = 2.45 (FCD)-1/2 (tDXf)1/4

Note is a function of both Kf and K so K cannot be determined.

Dimensionless Fracture Conductivity=

FCD

= Kfw
KXf

where w = frac width (m) and Kf = fracture permeability

so that kfw = frac conductivity

The bilinear flow equation has a slope of on a log-log plot.


Derivative of Bilinear Flow
dpD/dlntD = dpD/dtD dtD/dlntD = (2.45)(FCD)-1/2 tDXf-3/4 x tD = (2.45)(FCD)-1/2 tDXf1/4 = pD
The bilinear derivative has a slope of on a log-log plot. The derivative is offset cycle to the
right of pressure.
Pseudo Linear Flow
For FCD > 15, the bilinear flow regime, after a transition, is followed by linear flow described by
pD = ( tDXf)1/2 + / 3FCD

Note is a function of both Kf and K so K cannot be determined.

Thus, during this period, the response is like an infinite conductivity case with skin (which makes sense if
you think about it). So why does it make sense?
The pseudo linear flow equation has a slope of on a log-log plot.
Derivative of Pseudo Linear Flow
dpD/dlntD = dpD/dtD dtD/dlntD = ( tDXf )-1/2 x tD = ( tDXf )-1/2 = pD

ignoring the skin

The derivative pseudo linear flow has a slope of on a log-log plot. The derivative is offset
cycle to the right of pressure.
Pseudo Radial Flow
For tDXf > 2.5

the IAASfcf devolves to pD = 0.5 ln (tD) +0.40454 +spr

Function of K only.

The pseudo radial flow derivative devolves to 0.5.


Well Bore Storage
There may be some unit slope on the log-log plot but linear flow usually dominates even the early time.

68

Log-Log w/Derivative Analysis


The pressure response for a finite conductivity frac may have a unit slope but quickly turns to a quarter
slope then possibly back to half then on through a quarter slope (formation bilinear blow) and on to radial
flow. The attending derivative plot will show unit slope, quarter slope offset cycle, possible half slope
offset by cycle then thru a quarter slope and on to 0.5.
What is important here? The type curve match must measure or assume permeability to calculate
the associated frac length.
Comments
For FCD > 300, the well behaves like an infinite conductivity well. FCD as low as 0.5 is quite possible.
For production dominated by pseudo steady state, maximized production occurs for FCD = 1.26.
For production with a large transient component, FCD between 10 and 30 is accepted as optimum.

These curves are for finite-conductivity hydraulically-fractured formations.


In computer use, the curves are fitted to the data which is backwards from the initial intent but which
works very well with 1000 data points that would have to be continually updated on a screen.

69

Horizontal Wells
The complicated approach (to Analysis)
For an infinite, homogenous but anisotropic reservoir assuming uniform flux and using a line source
TDL/2
1/2

PDh = () /4 (k/ky)

1/2

[erf ((k/kx + xD )/ (2()1/2) + erf ((k/kx - xD )/ (2()1/2)] x exp(-yD2/4)

x [ 1+2 exp[(-n22LD2) cos (nzD)cos(nzwD)] d/1/2


n=1

Where
k = (kxky)1/2

LD = L/2h (kz/k)1/2

yD = 2 y/L (k/ky)1/2

tDL/2 = kt/((ct(L/2)2)

xD = 2 x/L (k/kx)1/2

zD = z/h

What is important here? Even if you assume kx = ky, kz is still important (notably in earlier time).
This can all be solved by the computer but still needs an estimate of kz. kz only concerns itself with the fit
of the data in earlier time.
The less complicated approach (to understanding Horizontal Well flow Regimes)
A horizontal well can go through as many as six distinct flow regimes. They are technically called
spherical flow, radial vertical flow, hemi-radial flow, linear flow, pseudo-radial flow and reservoir linear
flow.
1. Spherical flow: If L<h, spherical (or ellipsoidal) flow can occur in very early time.
2. Radial vertical flow: The dominating regime is radial flow into the well bore along its length. It is
dependent on both horizontal and vertical perm. One usually assumes horizontal perm is isotropic.
3. Hemi-Radial vertical flow: If the well bore is close to the top or bottom of the reservoir. Is as per radial
vertical flow but with a boundary.
4. Linear flow: The dominating regime is linear flow into the well bore along its length. This is mostly
influenced by horizontal permeability.
5. Pseudo-radial flow: When the EFFECTIVE length of the horizontal portion becomes insignificant to the
distance that the transient has travelled into the reservoir. The flow equation is as per simple radial flow.
6. Formation linear flow: When boundaries in the formation cause linear flow.
The simple approach (to understanding Horizontal Well flow Regimes)
With the added volume of the horizontal portion, Hz well bore storage is often tripled and dominates the
early time flow. Assuming L/h large, the only flow regimes that are likely to be manifest in an actual test
are radial vertical and/or hemi-radial vertical, linear and pseudo-radial. The derivative plot follows.

70

Vertical Radial Flow


The general flow equation for this phase of flow is
PDL = 0.5 ln (tD) +0.40454 +spr +sell
Where kav = (kzky)1/2

TD = (kzky)1/2 t / (ctrw2)

PDL = p2(kzky)1/2L / (q)

And sell is a small correction factor which compensates for the distortion of the well to an ellipse under the
coordinate transform and can be included in an s skin factor.
Analysis is the same as for simple radial flow but one has to estimate a value of kz, usually taken
as 1/10 of ky.
Hemi-Radial Vertical Flow
This is exactly the same as introducing a single no-flow boundary in the simple radial flow case. If both
radial and hemi-radial flow manifest, then one will see a doubling of the derivative. If not, and one
mistakenly assumes that hemi-radial flow is actually radial flow, then the permeability will be in error by a
factor of a half (i.e. it will be double what was measured).
Linear Flow
During the linear flow regime, the horizontal well can be considered equal to a fraced well with a frac
height of 2rw and a single wing so that for kx = ky and sufficient time for the Error functions to go to 1.0
PDh = 2(tDL)1/2 + sf

where

PDh = p2kyh/q

and

TDL = kyt/(ctL2)

and sf is a complicated equation that in essence accounts for the effect of flow convergence near
the well bore. It depends on h, L, z, rw and kz/ky.

71

Horizontal Pseudo-Radial Flow


The flow equation in later time reduces to
PDh = 0.5 ln (tD) +0.40454 +se
Where se now includes flow convergence plus factors to account for the prior linear flow period.
You will not see the use of se in text books; it is used here to relate to my own definition of se which is the
concept of equivalent skin.
The Dave Pridie approach (to understanding horizontal well test analysis)
All three regimes that are likely to be seen reduce to formulations that we are already familiar with. The
only quirk is that early time radial/hemi-radial flow depends on kz. In general, these regimes yield the
sequence of plateau (Vert Rad), rise (transition slope can =1/2), plateau (Vert H-Rad), rise (linear flow
slope of ) and finally plateau (Pseudo Radial). If one sees the entire sequence, analysis is simple.
What one usually sees is well bore storage followed by one (usually a rise), two or maybe three
components of the sequence and there is nothing to indicate where they fit into it. What this means, for
example, is that with only a rise in the derivative after well bore storage, there are three more unknowns
than definitive parts of the plot (ky, kz and Leff), so there are three families of solutions that will fit the data.
These can be ranged by pushing each of the unknowns to its limit and finding what the other
parameters need to be to get a fit.

High Perm. (Was in fact the best fit.)

72

High Lw. (Not as likely but still possible.)

High kz. (Possible but not likely.)

73

Balanced but not unique solution. This solution assumed that the model derivative would go horizontal as
soon as possible after the test data stopped, i.e. was the optimistic case for permeability.

74

75

The best method of categorizing horizontal wells is to run them long enough to determine reservoir
permeability and then calculating the overall skin effect. Horizontal wells usually produce an skin effect
of -3 to -4 depending on permeability.
What is important here? A Hz well skin between -3 to -4 indicates a successful well.

76

Multiply-Fraced Horizontal Wells


There is no simple equation for the complex flow pattern that can set up for multiple fracs in a horizontal
well but the later time response for the entire system is still a half slope like a single frac and can
therefore be compared to a single frac. The flow eventually will go radial but some work has been done
on elliptical flow that shows promise. I passed a John Lee paper on to the authors of PanSys who are
presently evaluating it for inclusion to their software. This pattern may yield estimates of permeability
earlier than waiting for radial flow to develop.

77

If the formation perm is, say, not much less than 0.1 md and the test run long enough to see the reservoir
perm, then the above suggests that the results can be compared to a single frac. These multiple fracs
have been seen to induce skins of -7 and even -7.2. This is getting very near to the most negative skin
possible and may require a reevaluation of the old Rule of Eights. The 8 is actually more like 7.8 and
is based on one section drainage. Perhaps the rule needs a more precise number than 8 which will most
likely be found by considering the transient flow equation. I will look into this soon.
For the moment consider this.
spr = ln (2rw/Xf) so that
Xf = 2rw/es = 2 x 0.1 / e-7.2 = 268 meters = 879 feet. Quite an EFFECTIVE frac length.
What is important here? Multiple fracs on horizontal wells work!
Consider also from the unmodified rule of eights.
Single fracs rarely achieved a skin below -5.
8/(8-5) = 2.7 x the flow rate.
8(8-7) = 8 x the flow rate. This is about 3 x as effective as extremely successful single fracs.
Until multiple fracs, the largest skin factor I had seen on single (and let me say they were huge) fracs was
-6 to =-6.2 and I had only seen it about a half dozen times.
8/(8-6) = 4 so multiple fracs are routinely twice as good as the best single fracs ever were.
What is important here?

You wont find these numbers in any one elses seminar

78

Reservoirs with Boundaries


Single Boundary
Method of Images (or superposition in space)
To model the buildup for a well a distance L from a linear no-flow boundary one can use the method of
images. By placing an imaginary well at a distance L on the other side of the boundary and subtracting
an equivalent p (from an equivalent q) caused by the image well from the actual well, the buildup for an
actual well with a boundary can be synthesized. Note that skin at the image well only affects the zone of
altered permeability and has no effect of the drawdown at the actual well (p from an equivalent q!).
Semi-Log Analysis
On a semi-log plot, a no-flow boundary manifests itself as a doubling of the slope. The plot sees true
perm until the transient hits the boundary.
Log-Log w/Derivative Analysis
The IARF derivative will rise from 0.5 to 1.0.
Comments
Doubling the slope or doubling the derivative is equivalent to halving the permeability. Indeed, one is
taking away flow from half the reservoir so that semi-radial flow (from half a reservoir) at K is equivalent to
radial flow (from the entire reservoir) at K. If the boundary is so close that well bore storage or other
early time phenomenon masks the true perm, one would never know that a boundary had been
encountered.
L-Shaped Boundary
Log-Log w/Derivative Analysis
The IARF derivative will rise from 0.5 to 2.0. Similar arguments to those above.
Wedge Shaped Boundaries
Log-Log w/Derivative Analysis
Using the formula D = 0.5 x 360/wedge-angle, one can make the derivative rise as high as one likes
before it goes horizontal. Since the wedge is part of an infinite circle, the derivative will always go
horizontal.
Parallel Boundaries
Think of parallel boundaries as a fixed-width channel that cuts nearly the exact same length of arc out of
an ever expanding circle as the transient radiates outward. This means that the amount of flow down the
channel is constant compared to the ever increasing circumference of a transient if it were in an infinite
reservoir. Thus, the amount of flow coming down the channel compared to the flow from an IARF well is
as a constant divided by an ever increasing number. At infinity, the ratio is 0. So too can you consider
the ratio of apparent permeability of a channel to that of an IARF well. The derivative for a reservoir of
infinite-length parallel-boundaries rises continually.
U-Shaped Boundaries
Same as parallel boundaries but with one half of the channel blocked off. This is the fastest continuallyrising derivative that can be modeled using boundaries.
Other Types of Boundaries
Reservoir boundaries can also be constantpressure. For an expanding gas cap on an oil reservoir or
water drive gas reservoir pressure would be fairly constant. For a side drive aquifer, the boundary would

79

more likely be a change in kh/ than constant pressure. Yet another type of boundary is the partially
sealing fault. In a more complicated version, a reservoir can be broken up into cells that communicate
across leaky boundaries with various combinations of no-flow boundaries thrown in.

When simple

changes in kh/ take place in a radial direction, the reservoir is called radial composite. The simplest
case is for a water injection well where viscosity changes at the water front.
Radial Composite (RC) Reservoirs
Where there are single or multiple changes in transmissibility, kh/, in a radial direction, the derivative
curve will drop for increases in transmissibility and rise for decreases in transmissibility. Thus, radial
composite models can be fit to almost any derivative action. In the hands of a neophyte, this could be
dangerous because of the inherent ease of matching any given data set. Indeed, if a radial composite
model is fit to a set of data and it correctly predicts where the derivative will continue past the
measured data, then the production forecast from that radial composite model will be correct. It
should be left in the hands of experts to decide where the derivative will go beyond the measured data.
Radial composite models can be fit to hydraulically fractured data or to any of the boundary conditions
described above. For parallel and U-shaped boundaries the RC outer-perm simply has to be small
enough not to cause the derivative to bend appreciably during the phase when the derivative is rising in
the actual data.
For tests that end with a rising derivative, the parallel boundaries or U-shaped boundaries model can be
used to model the pessimistic forecast (prevailing permeability appears to go to zero) and the radial
composite model can be used to model the optimistic forecast (permeability stabilizes).
What is important here? For tests that end with a rising derivative, the parallel boundaries or Ushaped boundaries model can be used to model the pessimistic forecast and the radial composite
model can be used to model the optimistic forecast.
Exotic and Combination Reservoirs
Compartmentalized, dual porosity (naturally fractured), dual permeability, pressure supported (gas caps
or aquifer, bottom or edge drive), partially penetrated, partially-sealing faulted and other exotic reservoirs
or reservoirs with combinations of models and boundaries should in general be avoided unless there is
compelling evidence not to. For example, a hydraulically fractured with parallel boundaries would be very
difficult to analyze since fractured well are dominated by linear flow in the early time and get prevailing
permeability in later time while parallel boundaries determines permeability in earlier time and sees linear
flow in later times.
Boundary Effects
The late time can be affected by various configurations/combinations of no-flow boundaries, changes in k,
h or away from the wellbore, leaky boundaries, pressure support (gas caps or aquifer, bottom or edge
drive), dual porosity effects, interference with other producing wells, perhaps even changes in k with time
(actually with pressure), tides and earth tides plus some things that havent even been thought of yet.
The shape of the any buildup plot is not unique to any model or boundary so input is required from
geology, geophysics, previous pressures/tests, the logs etc.

80

Part 6 PT Design
Well tests can be categorized by their objectives.

The following have been arranged so that the

objectives of any chosen test will contain the objectives of all the previous tests.
Shoot and Test
Shoot and Test usually requires use of Convolution or Deconvolution technique to analyze data and is
only correct in about 33% of cases (by DEI estimates) and ballparkish in about 33% of cases. Problem is
determining if any particular analysis is one of the good 33% or not. They work best where the perm is
high enough to measure by regular test methods. They also are short term and may not see thru
damage. Are also often done with surface gauges so cannot gauge liquid influx. Not recommended.
Simple Cleanup/Flow Test
Done immediately upon completion. Primary objectives are to clear well bore of work-over fluids and see
what formation fluids flow and at what rates.
It is a fallacy that wells have to be cleaned up prior to testing. The only time a well needs to be cleaned
up prior to a test is if a multi-point isochronal test is being done. For single point tests the well can clean
up during the main flow period without affecting the results. This is a special consideration when flow
volumes are restricted (flow volumes include cleanup volumes). Also, in the case of reserves testing, the
well should be shut in 10 times the cleanup time prior to measuring initial pressure so short cleanup times
make sense.
What is important here? A well only needs to be cleaned up to the point where it will flow without
swabbing so that the rig can be moved off and the test begun.
Quick AOF (Absolute Open Flow) Tests (Valid at ERCB for AOFsf<21 103m3/d)
Quick AOF tests are done mainly to satisfy ERCB initial deliverability requirements (for gas zones
only). The primary prerequisite for an AOF test is reservoir pressure (pi which in this case is also pR) and
that is usually obtained from a static gradient done at the time of completion (which satisfies the ERCB
initial pressure requirements for all zones). If a year or more has elapsed between completions and
the test, another static gradient should be run to confirm the pressure. The flow data (rates and surface
pressures) can be taken from the completion cleanup but is most often taken during the first three months
of production. (The ERCB allows 90 days on production time before the AOF is required). The AOF is
calculated by converting surface flowing pressures (ptf) to down hole flow pressures (pwf) and
extrapolating to the maximum rate using the simplified flow equation and assuming an exponent of n =
1.0 as per

Q = C (p2)n = C (pR2 - pwf2)n


AOF = C (pR2)

Where Q = flow rate in 103m3/d


C = flow co-efficient
p2 = pressure squared in 106 kPa2 (pR2 pwf2)
n = flow exponent
Note that is test can be run very inexpensively. What is the procedure?

81

The flow period should be at least long enough to overcome well bore storage which, for a gas well, is
given by:
tws = 2.65 106 VwCw

In Canadian metric units.

Note: Vws = depth x rw2

Kh
For depth = 1500m, rw = 0.15m, h = 12m, cws = 7.8x10-5, k = 20md and = 0.015 mPaS, tws = 1.37 hours.
Reservoir Parameter Tests
Reservoir parameter tests require recorders to be down hole for the final part of the flow and for the entire
buildup. This is the type of test most people are familiar with and can be programmed and run to
investigate a certain distance into the reservoir using
tflow = 6.94 x 104 [ctrtest2]

rtest is radius of investigation required

K
But since permeability is most times unknown, the usual case is to simply flow for one to three days and
shut in for 4 to 7 times the flow period up to 30 days. (7 times is preferred for best pressure and model
definition). Only tight (k<0.1 md) reservoirs benefit from buildups longer than 30 days. This test can yield
quality samples plus determine reservoir pressure, permeability, skin, and, should such be the case, frac
parameters. Boundaries can sometimes be determined but ghost boundaries can still be a problem.
(How does recorder depth factor in here?)
A reservoir parameters test can be run immediately after a well has been flowed or cleaned up so long as
the recorders are down hole to capture the final flow pressure. In such a case, the recorded flow prior to
the buildup should be held for ten times the time the well was shut in to run gauges so that the effect of
the short shut in is minimized to the point it can be neglected if so desired. An accounting for the
preceding flow and buildup times prior to the test are sufficient to correct for the stray transients induced.
Reserve Tests
Reserve tests are the most difficult to plan. The linchpin of the overall test is the pretest flow and buildup
(2 and 22 hours) that allows that calculation of the initial reservoir pressure. The combination of 2 and 22
was decided upon through 5 years of trial and error and found to be the best to a) flow long enough to
relieve any supercharging but not long enough to encounter boundaries, b) buildup long enough to yield pi
with a straight line extrapolation and c) work well with operational considerations. The 22-hour buildup
can be shortened to slightly less for wells that flow 30 103m3/d or more.
The main flow period should designed to remove enough reservoir fluid to drop the pressure by 50 kPa
for the minimum economic limit of reserves. For example, if calculations dictate that 30 106m6 (1bcf) are
required to make the tie in economic, and the reservoir pressure is expected to be 10, 000 kPa, then the
amount of reserves to be removed is

Voltest = 50/pi x voleconomic


= 50/10000 x 30 106m3
= 150 103m3n the example
And the flow time for a well that flows 30 103m3/d is

82

t = voltest/Qexcepted
= 150/30 days
= 5 days in the example
The amount 50 kPa was also found by trial and error but was mostly dictated by the required flow
volumes and times for most practical reserves tests. 50 kPa usually keeps the flow periods below 5 days
which most production engineers in conjunction with the ERCB are apt to sanction. It also is just large
enough to prevent extrapolation errors from dominating the measurement. (It is not based on gauge
accuracy or sensitivity).
Rates
Single point tests are most often run at expected line pressure. High productivity wells are often run at
higher rates especially when the possibility of forming hydrates is present. Choice of rates for a modified
isochronal test should be multiples of the initial rate to space them evenly on a log-log scale.
Testing After Completions Operations
Completions operations often include cleaning up the zone of interest bringing back kill fluid, spent acid
or frac fluid. As previously noted, the effect on a reservoir-parameters test is minimal so long as the
pretest flow and buildup (cleanup) times are accounted for. However, this is not the case for a reservestest however and obtaining an initial pressure can be quite difficult. It is therefore an absolute necessity
that the nature of the reserves test should be known prior to completions operations. The cleanup
operations can be held to a minimum and in some cases, the cleanup can even be accomplished during
the test, i.e. with gauges in the hole. Because most of AStechs tests are single point (the test does not
calculate the n exponent) the well can clean up during the test without any adverse effects. This applies
to the pretest flow and buildup as well. A well need only be cleaned up to the point where the testers do
not need to swab to bring it on. If cleanup operations must go ahead without gauges, then the well
should be left standing for ten times the cleanup flow period prior to running the pretest flow and buildup.
This will minimize the effect of the cleanup of the following test. Because completions operations cannot
be forecasted, the ensuing well tests remain flexible in design until they are run. Even the main flow
periods cannot be determined with any precision until rates have been proved in the cleanup or pretest
phase.

83

Pretest plan
Possible Test Priorities

Possibility of Achieving Design Criteria

Fluid Sample

excellent

Initial Pressure

excellent

Satisfy ERCB requirements

excellent

Flow Rate/AOF

very very good

Permeability/Skin

very good

Frac Parameters

good

Boundaries

good

Depletion

fair to good

Forecast

fair to good

Note that priorities down the list will include all the ones above it.
Test at all?
Yes

if good chance of achieving objectives

No

if existing, cleanup or initial production data can fulfill requirements

Test on Completion?
Yes

if rates/reserves are needed immediately for economics of tie in


if equipment is already on site for cleanup

No

if pipeline is to be built regardless

Allow well to stabilize after Completion?


Yes

if well was cleaned up for a long period of time and reserves are required

Test in Line?
Yes

if pipeline is to be built regardless


if only to satisfy ERCB requirements

Sampling
None

rare case large benefit to cost ratio

Routine

usual. Routine duplicate samples of gas, oil and water taken by testers give a
good indication of the expected production

Special recombination for condensate wells. Best to have lab company on site
H2S on slightly sour wells (small amounts of H2S can scrub out on the walls of
metal cylinders). Draeger/GasTec tubes or Tutweiller with/without lab rep on site

84

Recorder Type
Mechanical

rarely used anymore, electronic more cost effective and reliable

Elec. Strain

good enough for most applications - $100/day

Crystal

good enough for most applications - $100/day

Quartz

high quality, expensive, slower to stabilize but more accurate - $150/day

Piezo

Piezo-resistive have not made a big impact yet

Real Time Gauges


No

Usual. Cost is factor. ($2500/day for real time, $750/day for drop off)

Yes

Minimize on bottom time


Ensure test goes according to plan
Make changes on the fly
Ensure buildups are developed enough to analyze

Real time testing simply means that the bottom hole pressure is available at surface as the test
occurs. This is accomplished either with gauges that are tethered to surface with a conducting
line or with gauges that send electromagnetic signals through the ground (wireless telemetry).
The wireless gauges do not usually work below 3500 meters, rarely in cased hole unless special
insulator portions are run in the casing or sometimes not in certain geographical areas of
Alberta. (If asked, the operators will happily yield such shortcomings of their gauges but if not,
will happily run them to any depth anywhere you want). Wireless gauges typically have less
accuracy and sensitivity than tethered ones.
Placement of Recorders
Above Perforations

OK for finding reservoir parameters BUT may not yield accurate


pressure plus may indicate ghost boundaries in late time due to
liquid movement in wellbore.

Mid-Point Perforations

Necessary for reserves tests.

Very good idea if looking for

boundaries.
Below Perforations

Probably the best location but length of tubing extension is often


prohibitive. (Liquid is usually assumed between RRD and MPP.
We are presently checking out some cases of rising liquid levels
which would make this location less desirable than the best).

Cushion (gas) for DST


No

Usual

Yes

Reduce initial rate (rare)


Zone suffers from fines migration
Zone has unconsolidated sands

Closed Chamber for DST


No

Usual

Yes

Rare.

Test times
Initial Flow

2hr

Standard for initial pressure

85

Initial Shut In

22hr

Standard for initial pressure

Main Flow

1-5 days

As required to prove reserve numbers

Main Flow Shut In

5-30 days

Prefer 7x flow time to max of 30 days

Overall time ~35 days


Test Rates
Pretest use highest rate to cleanup (higher than for main flow period). Thus no sand etc.
should be produced during main flow. This is especially important if cleanup/pretest flow is done
to a P-tank and a test separator is moved in for the main flow.
Main Flow
Run at expected line pressure if possible
Do not change the choke after first third of flow period. It is better to have a smooth
pressure curve than a constant rate. Mathematics can account for the smooth decline in
rate more easily than the jumps imposed by choke changes.
Note that flow periods are subject to flaring volume restrictions, H2S release rates and public
acceptance.
Minimum rates for lifting water can be estimated from the nomograph in Appendix 2. Evenly
spaced rates can be also determined roughly by taking the minimum rate, quarter difference, half
difference and maximum rate. More accurate would be minimum rate, 0.215 difference, 0.464
difference and maximum rate.

86

PT Costs
The majority of cost of a PT is composed of the price paid to the slickline company for running, pulling
and rent of the gauges plus that paid to the production testing company for their separator or P-Tank plus
crew.

Incidentals include damage to equipment (washed out chokes etc.), methanol required to

overcome hydrating and hauling of fluids. These costs are constantly rising as per summer/fall of 2001.
Recorders

Testers

Total

Southern Alberta

$ 8 000

$10 000

$18 000

Northeastern B.C.

$12 000

$15 000

$27 000

Costs for routine fluid analyses are about $300 and pressure transient analysis can cost $800 to $2500
from industry or about $300 in house.
How PTs Fit into the Larger Picture
A drill stem test is often run to help make the decision whether or not to case a well. DSTs ($20 000)
measure the initial capability of the well to produce hydrocarbons so that positive results lead to casing
($50 000 - $70 000) and completing ($50 000 - $300 000) of the well. For gas, a production test is then
run ($15 000 - $30 000) to confirm the rates and to measure reserves so that a long term production
forecast can be made. Positive results from a production test lead to the well being tied in ($100 000 - $1
000 000). For oil, the well is usually swabbed to establish a rate then, if economic, it is set up with a
pump (if necessary) and tied to a battery. (Note that it is very expensive to test oil wells for reserves
unless they flow to surface why?).
Whenever PTs are run for a purpose other than the above, one should question the motives. Some
considerations are:
1.

Will the test results really be used? (Are negative results going to be ignored?). What happens if
the results are gray?

2.

If the well is cased for one zone, and another presents an exploration possibility, is it cheaper to
test the second one (shoot ($20 000) and possibly frac ($40 000 - $60 000) and production test
($20 000) plus lost production during the test time ($?)) before putting on production verses
drilling another well later? (How much later is later?). Is the second zone up hole from the first?
Can the well be dually completed?

3.

As in 3, if the logs are good enough to book reserves, why PT it? The PT could actually be
damaging to the reserves booking!

87

Comments on Data from Databases


The data in prepared formats should be checked against the original since it has been
manipulated and is known to contain errors.
Do not use the final buildup pressure as equal to the reservoir pressure unless you have seen the
PT/DST chart and can confirm that the buildup curve is well enough developed to for such an
assumption. See below.
Buildup pressures sometimes get classified according to how accurately they depict reservoir pressure.
Rather than go over anyone elses system, here is a scale of your own.
Stabilized flow or shut-in pressure:
(Stabilized flow meaning well killed itself during flow period.)
Very good:

OK to use as reservoir pressure, will probably be a few kPa low.

Extrapolated:
Best:

If Horner is straight line (straight line extrapolation is best overall)


BUT may suffer supercharge if low permeability (on pretest buildup).

Good:

If Horner does not show large deviations

Good:

If Horner plot flattens horizontally (pressure stabilizes due to support).

Poor:

If Horner plot is curing upward. May overestimate pressure!

Poor:

If Horner plot is curving downward. Could be interference. Difficult to


distinguish interference from small reservoir or support unless Cartesian
begins to decline.

Not Extrapolated:
Good:

If buildup curve is well developed and getting near the horizontal. Final
buildup pressure will probably be less than 5 % lower than the
extrapolation.

Poor:

If buildup is slow and still climbing.


storage or curving upward.).

88

(Horner will still be in wellbore

Abbreviations
Some of the abbreviations in the databases can be difficult to understand, here are a few explained.
(They apply to DSTs too.).
VO

valve open

SI

shut in

SIP

shut in pressure

FP

flow pressure

HP

hydrostatic pressure

XP

extrapolated pressure

GTS

gas to surface

SFC

surface

TSTM too small to measure


SAB

strong air blow

WAB

weak air blow

SGB

strong gas blow

SPR

spray

REC

recovery

cut (as in OCM is oil cut mud)

SWTR salt water


O

oil

DEN

density

OH

open hole

mud

thousand

thousand

million ( Halliburton DST reports) These have all been researched.

mm

million

MM

million

89

Appendix 1
Dimensionless Parameters
(This is not a derivation so much as an explanation of the form)
From the section on equations, for a liquid:
pi-pwf = 2121 q [log

p =

kh

kt
ctrw

-5.09 + 0.87 s]
2

but this equation was modified from the original In to make plotting easy, so going backward
2121 = 1842 x x In(10)
1842 q x x In(10) x [log

p =

1842 q x [In [
kh

-5.09 + 0.87 s]

ctrw2

kh
p =

kt

kt ] -5.86 + s]

ctrw2

now homogeneous units kt/ ctrw2 is already unitless (as you would expect for an In term) but in
practical metric units the conversion comes out to be 3.6 x 10-6
so
1842 q x [In [ 3.6 x 10-6 kt ] - x In 3.6 x 10-6 + 5.86 + s]

p =

kh

[ ctrw2

1842 q x [In [ 3.6 x 10-6 kt ] -0.405 + s]

p =

kh

[ ctrw2

(This now has the same constant as the equation for pseudo steady state production!)
Since p has units of kPa and the In term has no units, the factor 1842 q/kh must have units of kPa
so to make the equation non-dimensional we introduce
pD =

Kh p

= Dimensionless Pressure

1842 q
and

tD = 3.6 x 10-6 K t

= Dimensionless Time

ctrw2
and independently,
rD = r

= Dimensionless Distance

rw
in addition,
CsD =

Cs

= Dimensionless Wellbore Storage

cthrw2
{=} m3/kPa
m3/kPa

90

91

Appendix 3 - Some old-time concepts that still pervade Well Testing - Explained
The concepts of stabilized AOF and stabilized deliverability are nice but do not exist in reality. All flow
potentials decline with time. Rates can be held constant, in come cases, by choking the well back below
its potential but that is not the stabilized rate that the textbooks speak of. When the potential falls below
the maintained rate, production will again decline according to the regime of the time (transient or pseudo
steady state).
Production forecasting done at stable down hole pressures are much more accurate and useful at
predicting the future production of a well. However, the concept of stabilized rate or AOF is still with us so
here are some useful concepts to accompany it.
Rate for a gas well in transient flow
pi2 pwf2

q =

(1)

1632 ZT (Log ( kt )

-3.23 + 2s )

ctrw2

kh

2.303

Rate for a gas well in pseudo-steady state flow, circular reservoir of radius re
(an equation we never usually see)
pi2 pwf2

q =
1422 ZT

(2)

-4

(2 2.637 x 10 kt) + 2 x 1632 ZT (Log (0.472 re ) + 2s


ctre2

kh

kh

rw

2.303

which, if we define
pR2 = pi2 - 1422 ZT (2 x 2.637 x 10-4

kt

) q

Note no ln terms

(3)

ctre2

kh
becomes the (almost) familiar
pR2 pwf2

q =

(4)

2 x 1632 ZT (Log (0.472 re ) + 2s )


kh

rw

2.303

So, with the above equations, we can make the following observations:
1. During the transient phase, the drop in rate is proportional to the reciprocal of the logarithm of time.
2. During the pseudo-steady state phase, the drop in rate, although proportional to time and size (eqn.
2), can be made proportional to the average reservoir pressure and size (eqn. 4).
Why would anyone develop equation 4? (Hint think Pi.) Does pR change hourly? Daily? Weekly?
Hint:

When does transient flow end for a circular reservoir of re?


When does PSS flow begin for a circular reservoir of re?
What is the reservoir pressure doing during PSS?

So a plot of d(p2)/dt vs. dV/dt should yield what? Same as p2 vs. V. Same as p vs. V?
Note: pR is defined as pi f(q,t,area). pR can be calculated by using a straight line (infinite acting model)
extrapolation to p* on a semi-log plot then subtracting an (assumed) area-based correction (MDH, MBH
methods) or it can be extrapolated graphically using curvilinear extrapolation (bounded model). The
tenets of PSS flow, assuming constant rate, are 1) that the pressure gradient is static throughout the
reservoir from the well bore to the farthest boundary and 2) the pressure is declining everywhere at the
same (linear) rate in the reservoir.

92

What is important here? The volume of the reservoir can be measured by measuring the pressure
decline for a constant rate.

From a plot of equations, we can see that

1. The transient decline is the best (most optimistic) case


2. The true PSS case is always below the transient case i.e. the lines never cross, i.e. there is not time
at which the solutions are equal. In fact, time to PSS is actually the time to the point where the
derivatives are equal. It is defined by
tpss = 945 ctre2
k
3. The form of the PSS equation with average pressure does cross the transient line at the point where
tps = 376 ctre2
k
but this point has not been recognized as having as much significance in the literature, (though that is
more by preference than mathematical design).
4. Because of the shape of the PSS curve, the rate at tpss is higher for smaller reservoirs and yet as
can be seen, the rate may never hold stable for those small sizes.
Note that modern computerized analyses solve the flow equations in LaPlace space using the
Stehfast algorithm so that the transition from transient to pseudosteady state behavior is
continuous and smooth.

93

You might also like