w 9

... -

-
z Soeietv of Petroleum Engineers

SPE 49222

Analyzing Well Production Data Using Combined Type Curve and Decline Curve
Analysis Concepts
Ram G. Agatwal, SPE, David C. Gardner, SPE, Stanley W. Kleinsteiber, SPE, and Del D. Fussell”, SPE, Amoco
Exploration and Production Co.

Cwyright1996 Smety of Petroleum Engineers, Inc

result is the development of these new production decline type
This papr was prepered for presentation at the 19% SPE Annual Ttimcal Conference and curves.
Exhibfiion held in New Orleans, Louls!ana, 27-30 September 199S
These new production decline type curves represent an
This pa~r was selected for presentabcm by an SPE Prcgram Committee following rev!aw of
Imformat!on W“tai”ed M a“ abstract $&mifted by tha authm(s) Co”te”ts of the pap,gr,
aa advancement over previous work because a clearer distinction
presentad, have not baen reviewed by the Swety of Patroleum Engmaers and ara subject to
mrrecbon by the author(s) The material, as presented, does not necessarily reflect any
can be made between transient and boundary dominated flow
poslbon of the Society of Petroleum Engineers, Its offIcars, or membrs Papers presented at periods. The new curves also contain derivative finctions,
SPE meetmgs are sublect to publl=bon rev!aw by Edltorlal Committees of the Smtety of
Petroleum Engineers Electromc reproduction, d!stribubon, or storage of any part of this papar similar to those used in the pressure transient literature to aid
for wmmem!al pu~ses w!thl the wrttten wnsant of the Society of Petrolaum Engineers IS
prohrbltad Perm!sslon to raprcdum Im print is rastrlcled to an abstract of not more than 300
in the matching process. These production decline curves are,
words. !llustrabons may not be cop!ed The abstract must contain Wnsplcuous to our knowledge, the first to be published in this format
~nowledgment of where and by whom the pa~r was presented Wr!te L!brarlan, SPE, P O
Box 8338w, Rtiardson. ~ 75083-38%, U. S A, fax 01-972.952-9435 specifically for hydraulically fractured wells of both infinite
and finite conductivity. Finally, these new curves have been
Abstract extended to utilize cumulative production data in addition to
This paper presents new production decline curves for commonly used rate decline data.
analyzing well production data from radial and vertically
fractured oil and gas wells, These curves have been developed Introduction
by combining Decline curve and Type curve analysis concepts Estimation of hydrocarbon-in-place and reserves for oil and
to result in a practical tool which we feel can more easily gas reservoirs is needed from the time when such reservoirs
estimate the gas (or oil ) in place as well as to estimate are first discovered to future times when they are being
reservoir permeability, skin effect, fracture length and developed by drilling step out wells or infill wells. These
conductivity, etc. Accuracy of this new method has been estimates are needed to determine the economic viability of
verified with numerical simulations and the methods have the project development as well as to book reserves required
been used to perform analyses using production data from by regulatory agencies.
several different kinds of gas wells. Field and simulated During the last fifi years, various methods have been
examples are included to demonstrate the applicability and developed and published in the literature for estimating
versatility of this technology. reserves from high permeability oil reservoirs to low
Decline curve analysis methods, in a variety of forms, permeability gas reservoirs. These methods range from the
have been used in the petroleum industry for more than fifiy basic material balance methods to decline/type curve analysis
years to analyze production data and forecast reserves. Type techniques. They have varying limitations and are based on
curve analysis methods have become popular, during the last analytical solutions, graphical solutions (known as type curves
thirty years, to analyze pressure transient test (e.g. buildup, and decline curves), and combinations of the two. Examples
draw-down) data. of these range from Arp’s i decline equations for liquids to
Pressure transient data can be costly to obtain and may not Fetkovich’s2 liquid decline curves, Carter’s3 gas type curves
be available for many wells, while well production data is and Blasingame4 et. al.’s gas equivalent to liquid decline
routinely collected and is even available from industry data curves. Several other papers on the related subject, too many
bases. In the absence of pressure transient data, a method that to quote here, have appeared in the SPE literature,
can use readily available well production data to perform
pressure transient analysis would be very beneficial. The Purpose
The purpose of this paper is to document ne~v production
decline type curves for estimating reserves and determining
* Retired from Amoco
other reservoir parameters for oil and gas wells using
References and illustrations at the end of paper

585
2 RAM G. AGARWAL, DAVID C. GARDNER, STANLEY W. KLEINSTEIBER AND DEL D. FUSSELL SPE 49222

performance data. Depending on the amount of performance mechanism of a reservoir.

data available, these methods can provide lower bound and/or Reservoir limit test is based on the constant rate
upper bound estimates of a well’s hydrocarbon-in-place. The drawdown solution. A well is produced at a constant rate and
accuracy of such estimates for reserves and other reservoir wellbore pressure response is plotted as a finction of time on
parameters will depend on the quality and kind of the Cartesian graph paper (see Figure 1). During the pss
performance data available. condition, the slope of this line is inversely proportional to the
Using the methods of Palacio and Blasingame4, it will be fluid in place. This kind of test is not limited to, but is usually
demonstrated and confirmed, that solutions for constant rate run on new exploration type wells for estimating reserves.
or constant bottomhole pressure production for oil and gas can Arp’s Decline Equation is based on empirical
be converted in most cases to equivalent constant rate liquid relationships of rate vs. time for oil wells and is shown below:
solutions.
Initially, we will review background material. Next, we 9i
q(t) = -.--(1)
will briefly discuss the various methods which are commonly
(1+ bD,, )+
used for estimating gas reserves. Finally, we will present the
new production decline type curves and demonshate their In this equation, b=O and b= 1 represent exponential and
utility and application by means of both synthetic and field harmonic decline, respectively. During pss for liquid systems,
examples. the exponential decline is a characteristic of constant pressure
Although the technology discussed in this paper is production whereas the harmonic decline is due to constant
applicable to both oil and gas wells, we will limit our rate production. Any other value of b represents a hyperbolic
discussion mainly to gas wells. decline (see Figure 2). Although Arp’s equation is strictly
applicable for pss conditions, it has been ofien misused for oil
Background Material and gas wells whose flow regimes are in a transient state.
Fetkovich Liquid System Decline Curves were published
Transient & Pseudo Steady-State Flow Conditions. in 1980 for analyzing oil wells producing at a constant
When a well is first opened to flow, it is under a transient pressure. He combined early time, analytical transient
condition. It remains under this condition until the production solutions with Arp’s equations for the later time, pseudo
from the well affects the total reservoir system. Then the well steady-state solutions. A reproduction of his type curves are
is said to be flowing under a pseudo steady-state (pss) shown in Figure 3.
condition or a boundary dominated flow condition. Transient Although the value of the b stem ranges from O to 1,
rate and pressure data are used to determine reservoir curves for b> 1 are often added to this Figure and are misused
permeability and near wellbore condition (damage or to match transient data. These liquid system curves are not
improvement), fracture length and/or fracture conductivity recommended for gas wells when the amount of pressure
whereas the pss data are required to estimate the fluid-in-place drawdown is moderate to large. Alternately, these curves (b=O
and reserves. and b= 1) may be used for gas wells if gas well data are
Transient and pss flow conditions are schematically shown converted to an equivalent liquid system data. This concept
on a Cartesian paper in Figure 1 and on a log-log paper on has been used in this study and will be discussed later.
Figure 2. Carter Gas System Type Curves were developed in 1985
to fill the gap which existed with the Fetkovich decline curves.
Review of Various Reserve Estimation Methods. They are shown in Figure 4. Carter used a variable L
Volumetric method is used to make an initial estimate of reflecting the magnitude of pressure drawdown in gas wells.
gas-in-place using petrophysicai data such as hydrocarbon His k = 1,0 curves corresponds to b = O on the Fetkovich
porosity, pay thickness, initial reservoir pressure, reservoir liquid decline curves and represents a liquid system curve with
temperature, PVT data, and reservoir size (or well spacing). an exponential decline. Curves with X = 0.75 and 0.55 are
Such estimates are useful and should always be made. used for gas wells with an increasing magnitude of pressure
Material balance method for gas wells makes use of the drawdown. Obviously, Carter type curves are better suited to
estimate reserves for gas wells.
material balance equation, Palacio & Blasingame Type Curves, first presented in
:=(:)(’-:) and ‘he 1993, provide a major advancement in the area of analyzing
=F is plotted as a function of
oil and/or gas well performance data using type curves. Their
material balance graph where
z type curves are shown in Figure 5. This paper is an excellent
cumulative gas production. Although the material balance culmination of their work and the work of other investigators
graph may have a limited application in certain cases (for whose goals were to convert gas well production data into
example tight gas wells) it provides useful guidelines for equivalent constant rate liquid data. They also established a
reserves and can also give insight regarding the drive clear relationship among the previously discussed decline

586
SPE 49222 ANALYZING WELL PRODUCTION DATA USING COMBINED TYPE CURVE AND DECLINE CURVE ANALYSIS CONCEPTS 3

curves. the independent variable. For example, with t~ as the

Palacio-Blasingame type curves provide a useful tool to *
estimate gas in place (GIP), reservoir permeability and skin. Similarly, l/dlnPwD’ is
‘dependent ‘ariab’e’ ‘D’ = dt~ “
However, the transient stems are strictly valid only for radial
flow and thus may not be suitable for analyzing gas wells with defined as the reciprocal of the derivative of M with respect
relatively long vertical hydraulic fractures of infinite or finite to the log of the independent variable. For example,
conductivity. It is also diticult to pick up a clear transition @}@
between the transient and the pss flow periods from these and l/dinPwD’ ~~}
= 1/{— for the independent variable, t~.
the other previously discussed decline curves. Palacio-
Blasingame utilize derivative methods to help with the type Equivalence Between Constant Rate and Constant BHP
curve matching process but this results in multiple curves even Liquid Solutions.
for the radial flow system. Details about their type curves as
To demonstrate this, two radial liquid system cases were
well as a comprehensive list of pertinent references on this
considered. The two systems were identical, except in one
subject can be found in their paper (Reference 4).
case (LR) the well was produced at a constant rate and in the
uttier case (LP) the well was produced at a constant
Discussion bottomhole pressure.
Our first objective with this study was to verify, using a single
Figure 6 shows graphs for both cases in terms of l/b,
phase finite difference reservoir simulator, a major finding of
PwD’ and l/dlnPwD’. A comparison of these two cases show
Palacio and Blasingame, that constant rate and constant
that during the transient period, the two sets of results are very
bottomhole pressure solutions for liquid and gas systems, can
similar. However, they are quite different during the pseudo
be converted to an equivalent constant rate liquid solution.
steady-state (or boundary dominated flow period). This is to
Constant rate liquid solutions are well understood for both
transient and pss conditions and are widely used for pressure be expected since constant rate solutions during the pss period
show a harmonic decline (straight line with negative unit slope
transient analysis (PTA) purposes. With constant rate liquid
solutions, one can take advantage of the many well known on log-log paper), whereas constant bottomhole pressure
PTA techniques for plotting decline curve data on different solutions result in an exponential decline (concave line on log-
log paper). If constant BHP results are replotted using a
types of graph papers and for utilizing appropriate plotting
variables such as pressure, rate, cumulative production, and modified time, t. = (Cumulative Production)/(instantaneous
rate) and compared with the constant rate solutions, they
time functions, and also the appropriate derivative functions.
become equivalent. This equivalence behveen the two
Plotting Dimensionless Variables. solutions is shown in Figure 7. Notice during transient flow,
Constant rate liquid solutions are commonly used for pressure PwD’ has a negative unit slope while l/dlnPwD’ is at a
constant value of 2.0 (zero slope). During pss flow their roles
transient analysis. Dimensionless variables which are
reverse, that is 1/dlnPwD’ has a negative unit slope, while
frequently used in type curves for pressure transient analysis
PwD’ has a constant value which is inversely proportional to
are dimensionless pressure, M and its derivatives with
GIP, Therefore, it is possible to estimate GIP and to determine
@@ whether this value is a lower or upper bound estimate.
respect to dimensionless time, — and with respect to log
dt~

of dimensionless time,
@w
— To make a type curve graph
Equivalence Between Constant Rate and Constant BHP,
Gas Solutions.
dhlt~ “
To illustrate this equivalence, two radial gas system cases,
appear like a decline curve, one can use the reciprocals of ~ constant rate case (GR) and constant BHP case (GP) were
@,@ considered. Except for the mode of production, the two
to produce graphs of 1/~, ‘, and l/{~} plotted systems were identical. These cases should provide a difficult
dt~
litmus test for verifying the desired equivalence. Figure 8
against dimensionless time. The reciprocal log time
shows graphs for both gas cases and is similar to Figure 6 in
*W terms of plotting variables. During the transient period, both
derivative, n{-~ti~ }does for the rate decline plot, what ~
constant rate and constant BHP cases appear identical.
does for the pressure build-up plot, namely, helping to identi& However, the difference between the two cases during the pss
flow regimes and to estimate permeability. period becomes significant. This is to be expected because
References to these 2 types of derivatives are used quite additional complications are caused not only due to
extensively in this paper, so for convenience in the text and in differences in modes of production but also due to varying gas
the figures, a short hand form representing the derivatives is properties. We go through the conversion to an equivalent
used. PwD’ is defined as the derivative of M with respect to constant rate liquid solution in a stepwise manner. First we use
the dimensionless time based on the modified time, t= instead

587
4 RAM G AGARWAL, DAVID C. GARDNER, STANLEY W, KLEINSTEIBER AND DEL D. FUSSELL SPE 49222

of real time as was done for the liquid cases. The second step During the transient period, 1/dlnPwD’ has a constant value of
is to redefine time in terms of pseudo time where gas 2.0 (zero slope), while PwD’ has a negative unit slope. These
properties product (pc~) is calculated as a function of average characteristics are reversed during the pss period where
reservoir pressure ~ . Pseudo equivalent time, t, was l/dlnPwD’ has the negative unit slope while PwD’ has a
developed by Palacio and BlasingameJ by extending earlier constant value of 27t and the zero slope. This negative unit
work of Fraim and Wattenbargers and is shown below: slope behavior is also referred to as harmonic decline. Both
time and log time derivatives on log-log graphs are also very
useful because of their distinguishing features during pss and
transient flow periods, This kind of graph is also useful for
estimating gas-in-place.
In Figure 10, where data has been plotted using
Results, after this conversion is made, are found to be dimensionless time based on r~ti, a single curve is obtained
identical to those shown in Figure 7. That is, the two gas cases
during the transient flow period for l/pW~ for all rJrW, values
become identical during both the transient and pss periods.
and also for each of the two types of derivatives. However,
This verifies that it is possible to convert constant BHP liquid
these curves start peeling off for each rJrW, value during the
as well as constant rate and constant BHP gas cases into an
pss period. The use this kind of type curve graph is
equivalent constant rate liquid case. This is significant because
recommended for estimating reservoir parameters such as
it will permit us to focus our attention to mainly constant rate
permeability and skin effect.
liquid systems. We have also found that for gas wells, when
2. Infinite Conductivity Fracture.
both the rate and pressure vary smoothly such that 1/pW~ is
Figures 11 and 12 show similar graphs but for vertically
monotonically decreasing, that the data can be converted into
fractured wells with infinite conductivity fractures. Here, data
the constant rate liquid analogue.
are plotted in a manner similar to Figures 9 and 10. The main
However, there is a computational problem. The
calculation of pseudo time in Equation. 2 assumes that we a,
differences here in plotting the results are that, x$x~ (or —
know GIP (a parameter value we normally don’t know and
would like to determine). The implication of this assumption
:,
suggests an iterative procedure for GIP. This can be easily has been used as a parameter compared to rd’rW,(or —
rw
accomplished using a spreadsheet program.
which was used for the radial system cases and dimensionless
Production Decline Type Curves. time is based on the fracture half length, x~ instead of rW,. In
These new decline type curves will be presented under three this case xJxf values of 1, 2, 5, 25 have been used.
categories: I ) Rate - Time, 2) Rate - Cumulative Production, Comparison of Figures I I and 9 shows that for the fractured
and 3) Cumulative Production - Time. Under each category, well multiple derivative curves are obtained as opposed to
we will present decline type curves for radial systems and for obtaining a single curve for each kind of derivative for the
vertically fractured wells with infinite and finite fracture radial case.
conductivity, as appropriate.
3. Finite Conductivity Fracture.
Rate-Time Production Decline Type Curves. Figures 13 and 14 show graphs of l/pW~ and I/dlnPwD’ for
I. Radial System vertically fractured wells with finite conductivity fractures.
For the radial system log-log type curves, we generated three These graphs are similar to those shown for infinite
cases corresponding to rd’rW. = 10,000, 1,000, and 100, and conductivity vertically fractured wells with one exception.
utilized the previously discussed approach for plotting the Here we have also included the effect of varying
results. Dimensionless time variable for the x-axis was krw
dimensionless fracture conductivity, FCD= — . Three
calculated in two different ways: ( 1) based on the drainage [) kx ~
area, A. and (2) the apparent wellbore radius squared, ~~. . values of Fcd have been used which range from 500 to .05.
Results are shown in Figures 9 and 10, respectively. The value of 500 corresponds to a fracture of infinite
In Figure 9, where data has been plotted as a function of conductivity whereas the value of 0.05 represents _very low
dimensionless time based on the area (A), 1‘pW~is a function fracture conductivity. In this case x,/x~ values of 1, 2, 5, and
of r&rW~during the transient flow period. These different 1~pWD 25 have been used for each value of fracture conductivity.
curves merge together during the pss flow period and take a
unit slope line. This graph looks similar to decline curves 11. Rate-Cumulative Production Decline Type Curves.
published by Fetkovich, Carter, Palacio-Blasingame and Another kind of graph which is commonly made by
others but has certain distinct advantages. A single curve is operations and field engineers is to plot rate, q(t), or
obtained for each of the two derivatives for all rJrW. values. normalized rate, q(t)/Am(p) as a function of cumulative

588
 SPE a222 ANALYZING WELL PRODUCTION DATA USING COMBINED TYPE CURVE AND DECLINE CURVE ANALYSIS CONCEPTS 5

production. A recent paper on this topic is due to Callard6 et. has a unit slope line except at dimensionless times less than
al. To investigate the character of these graphs, the about 100. During pss flow, the curves peel off and become
dimensionless groups I!pW~ and the derivative of pW~ with flat when going from smaller to larger values of rJrW,.
respect to dimensionless cumulative production. Q~A (Q~~ = In Figure 19, dimensionless cumulative production, Q.~
tDA/pWD), were plotted as functions of Q~~ on log-log for a fractured well has been plotted on a log-log graph paper
coordinates. Results in the form of type curves for radial flow as a function of dimensionless time based on fracture half-
systems for r&’rW,= 10,000, 1,000, and 100 are shown in Iength, Xfi In this case a separate curve is obtained for each
Figures 15, and 16. value of dimensionless fracture conductivity. During transient
Figure 15 shows that during the transient flow period, flow, for each fracture conductivity value, a single curve is
separate 1/pWD curves are obtained for each r,/rW~ value. obtained for all x~x~ values. Their slope ranges from a unit
However, during pss flow they asymptotically merge into a slope for low conductivity fractures to half-slope for infinite
single value of Q~A = l/(27t) = O.159. We call it an anchor conductivity fractures. During pss flow, these curves peel off
point value and find it useful in estimating GIP. and become flat for each value of xJx~ .
Figure 16 is a Cartesian (linear) graph of the same 1/pWD
data as is used in Figure 15. Notice that during the pss flow Comments About These & Other Published Type Curves.
period, the different 1,’pW~curves become linear and converge The type curves presented in Figures 9-19, in total
at Q~A = 1/(2n). The significance of this attribute is that for an represent a new contribution to technology and contribute to
optimistic estimate of GIP the trajectory of the field data will recent work within the industry in advanced type curve
under shoot the anchor point and it will over shoot the anchor methods. Our contributions in this area are mainly in the
point for a pessimistic estimate of GIP. We find this graph integration of recent developments from several published
very useful in converging to a correct value of GIP. sources into more complete sets of type curves for both radial
Figure 17 shows l’pWD and the derivative of pW~ with and fracture flow, in both transient as well as boundary
respect to dimensionless cumulative production based on r~a , dominated flow conditions, These new contributions include
derivative type curves for both radial and fracture flow in
Qat) (Q,D = t,w’pw~) plotted against Q,~ A notable feature of
this graph is that 1/pW~ along with this derivative, form an terms of dimensionless rate and rate derivatives vs.
envelope, A vertical tangent to this envelope corresponds to dimensionless time which are based on normalized rate and
the GIP. The derivative plot shows a negative unit slope line equivalent fluid properties. We have also made contributions
during the transient period but it assumes a rapidly increasing through the creation of new dimensionless rate and rate-
positive slope during the pss period. These characteristics can derivative type curves in terms of dimensionless cumulative
be utilized to identify the transient and pss flow periods and production, Q~~ (which is also based on normalized rate and
the transition between them. Moving to the Iefi of this graph equivalent fluid properties), For lack of a better name, this
where the rJrW, value decreases, I‘pWD and its derivative new suite of type curves are referred to as Agarwal-Gardner
become closer to one another. Although not shown, they (A-G) &pe czirves, not because we discovered the previously
intersect one another for smaller values of rJrW, We utilize discussed mathematical relationships, but because we have
similar graphs for infinite and finite conductivity vertically compiled the requisite information from various recently
fractured wells but they are not included in this paper.. published sources and have constructed original type curves
The above discussed characteristics of cumulative type based on those relationships.
curves have been found to be advantageous for diagnostic These A-G type curves consider both the transient and the
purposes as well as type curve matching purposes. pss flow conditions, as well as the transition between the two,
in a rigorous manner. They can be easily generated using a
111. Cumulative-Time Production Decline Type Curves. simulator for any desired flow system such as radial and that
We generated cumulative production vs. time decline type resulting from vertically fractured wells. Analyses of tight gas
curves using the same data as we used and discussed for radial wells require that a much longer transient period be included
systems and vertically fractured wells. We find such curves with such type curves. This can be easily accomplished with
useful because field cumulative production data is ofien these type curves. Alternately, published infinite and finite
smoother than the corresponding rate data. Moreover, such conductivity fracture type curves (such as published by
curves have their own characteristics which can be used to our Agarwal, Carter, and Pollock7 and Cinco et al.8) can be used
benefit in estimating reservoir parameters and reserves. for the analysis of transient data. Infinite conductivity fracture
Figures 18 and 19 show such curves respectively for radial type curves published by Gingarten et al.9 can be used for
wells and finite conductivity fractured wells. transient as well as pss periods. Similar comments apply to
In Figure 18, dimensionless cumulative production, Q,D other published type curves.
has been plotted on a log-log graph paper as a function of
dimensionless time based on wellbore radius. During the
transient fiow a fi~e curve is obtained for ti r@W, values. It

589
6 RAM G. AGARWAL, DAVID C. GARDNER, STANLEY W. KLEINSTEIBER AND DEL D. FUSSELL SPE 49222

Application of A-G Type Curves Field Application of A-G Type Curves

These type curves such as shown in Figures 9-19 can be Figures 22 and 23 show history matched parameter
easi Iy programmed into a spreadsheet program. Modem estimates for GIP, permeability, etc. for an infill gas well in
spreadsheet programs provide a convenient medium to history the low permeability Red Oak sand of southeastern
match tield data to determine parameter values with these new Oklahoma, The well was drilled in December, 1991 and has
type curves. slightly over six years of production history with a cumulative
A set of basic data is required to do the type curve production of 1,9 Bscf. This production data is typical of that
matching for gas wells regardless of which kind of type curve obtained from industry or field data bases and the noise or
is being used, They are listed below: inaccuracies in measured or reported data is reflected in the
1. Reservoir Data: Initial reservoir pressure (pi), reservoir dimensionless data.
temperature (T), formation thickness (h). reservoir Figure 22 shows that the plot of l/pWD vs Qo~ is
permeability (k), hydrocarbon porosity (+s~). converging nicely to the GIP anchor point and that the
2. Gas Properties Data: Tables of viscosity (p), z-factor, estimate of GIP can be used with some confidence. The
and gas compressibility (c~) vs. pressure including values of character of the derivative data on the plot of l/pWD vs tDA
vi, zi, and Cti. Tables of real gas pseudopressure [m(p)] vs. (Figure 23) clearly shows that the well has transitioned to
pressure. boundary dominated flow. This plot also illustrates a match
3. Performance Data: Well rate (q), bottomhole pressure for estimating permeability, fracture half-length, and
(PBHP), and cumulative gas production (Q) as a function of dimensionless fracture conductivity. Only minor
producing time (t). modifications to the parameter values obtained from this type
Normally the parameters to be determined that are of curve match were needed in order to match the well’s history
primary interest are GIP, formation flow capacity (kh), and using a finite difference simulator.
wellbore skin or for fractured wells, fracture length and Figure 24 is a plot of the same production data on a
conductivity. Palacio-Blasingame type curve. The estimates of GIP
Estimating GIP between the two type curves are in excellent agreement,
With these new decline/type curves, we recommend that differing by less than three percent. There is, however, some
GIP be determined first. This is because the independent ambiguity about which transient stem to match the data with
variable, Q~~, used to estimate GIP, is independent of in order to calculate permeability and wellbore skin. This
permeability, whereas estimates of permeability, etc., are illustrates a benefit of using these new type curves for low
dependent upon GIP. permeability fractured wells which typically have long linear
A good initial estimate of GIP is not necessary, or bilinear transient flow periods.
convergence to the proper GIP is usually very rapid. The
cumulative gas production can provide a lower bound estimate Summary and Conclusions
for GIP, whereas a volumetric estimate obtained from 1. A new set of rate-time, rate-cumulative, and
petrophysical data could provide an upper bound estimate. A cumulative-time production decline type curves and their
value somewhere between the two would suffice for an initial associated derivatives have been developed using pressure
estimate, transient analysis concepts and are presented in this paper.
Figure 20 shows the rate-cumulative production decline 2. They have been developed for radial systems as well as
type curves and illustrates the concept of graphically vertically fractured wells with infinite and finite conductivity
estimating GIP based on responses to changes in GIP values fracture, These production decline curves are, to our
relative to the type curves. In this Figure, we use values of knowledge, the first to be published in this format especially
plus and minus 20 percent of the true value of GIP to for fractured wells.
showcase this effect. Notice that with the correct GIP, the 3. These new production decline type curves represent a
data follows the trajectory of one of the 1.’pw~rays. It doesn’t advancement over previous work in that a clearer distinction
matter which of these rays are used, all focus to the same point can be made between transient and boundary dominated flow
on the Q~~ axis of l/(27t). periods.
Estimating formation f70w capacity, &h 4. They provide a practical tool to field engineers for
Figure 21 shows rate-time production decline type curves estimating gas (or oil) -in-place as well as to estimate reservoir
and illustrates the concept of graphically estimating kh based permeability, skin effect, fracture length, and fracture
on responses to changes in kh values relative to the type conductivity. Also because this technology can be easily
curves. programmed into an electronic spreadsheet, it is more readily
In this Figure, we use kh values of 2.0 and 0.5 md, which used.
are twice and 1/2 the true value of 0.10 md, in order to 5. These type curves enable us to utilize routinely
showcase the sensitivity to permeabilhy. colleeted production data and b avaihibie from industry

590
SPE 49222 ANALYZING WELL PRODUCTION DATA USING COMBINED TYPE CURVE AND DECLINE CURVE ANALYSIS CONCEPTS 7

data bases in the absence of costly pressure transient data. v = Viscosity, cp

6. These concepts can be extended to other well and/or n = 3.1416
reservoir models such as horizontal wells or naturally
fractured reservoirs, to name a few. Acknowledgments
The authors wish to thank Amoco Exploration and Production
Nomenclature Company for permission to publish this paper. We would like
A= Drainage Area, sq fi to also express appreciation for the company’s support for the
b= Arp’s decline curve exponent creation, the dissemination and the application of technology.
Cg= Gas compressibility, l/psi
(Cg)i = Gas compressibility at initial reservoir pressure, References
1/psi 1, Arps, J. J.: ‘bAnalysis of Decline Curves,” Trans. AIME ( 1945)
(.t =
Total system compressibility, I/psi ~, 228-247,
2. Fetkovich, M. J.: “Decline Curve Using Type Curves,” JPT
Dit = Arp’s dimensionless decline rate
e= 2.7183, Napierian constant (June 1980), 1065-77.
3. Carter, R. D.: “Type Curves for Finite Radial and Linear Gas
Fc~ = dimensionless fracture conductivity
Flow Systems: Constant Terminal Pressure Case,” SPEJ
Gi = Initial gas in place, MMSCF or BSCF (Oct. 1985), 719-728.
GP = Cumulative gas produced, MMSCF or BSCF
4. Palacio. J. C. and Blasingame. T, A.: ‘bDecline-Curve Analysis
k= Effective permeability to gas, md Using Type Curves - Analysis of Gas Well Production Data,”
k~ = Fracture permeability, md paper SPE 25909 presented at the 1993 Joint Rocky Mountain
m(p) = Real gas pseudo pressure, psiazlcp Regional and Low Permeability Reservoirs Symposium,
A[m @)]= m(pi)-m (~) ,psizlcp Denver, CO, April 26-28, 1993.
5. Fraim, M. L. and Wattenbarger, R. A.: “Gas Reservoir Decline-
A[m (p)]= m(Pi)-m (PBHP ,psi2/cp
)
Curve Analysis Using Type Curves with Real Gas
PBHP = Bottomhole producing pressure, psia Pseudopressure and Normalized Time,” SPEFE (Dec. 1987)
~=
Average reservoir pressure, psia 671-82.
pi= Initial reservoir pressure 6. Canard, J. G.: “Reservoir Performance History Matching Using
Dimensionless wellbore pressure Rate/Cumulative Type Curves,” paper SPE 30793 presented at
PwD =
the 1995 SPE Annual Technical Conference and Exhibition,
q,= Initial flow rate, MSCF/D
Dallas, Oct. 22-25, 1995.
q(t) = Flow rate, MSCF/D
7. Agarwal, R. G., Carter, R. D., and Pollock, C. B.: “’Evaluation
QDA = Dimensionless cumulative production based on and Performance Prediction of Low -Permeability Gas Wells
Area (A) Stimulated by Massive Hydraulic Fracturing,”’ JPT
QaD = Dimensionless cumulative production based on (March 1979), 362-372.
r2 or x; 8. Cinco, L. H., Samaniego, V. F., and Dominguez, A. N.:
“Transient Pressure Behavior for a Well with a Finite-
rW~= Ap~a;ent wellbore radius, ft
r-e= Reservoir radius, ft Conductivity Vertical Fracture,’” SPEJ (Aug. 1978), 253-264.
9. Gringartcn, A. C., Ramey. H. J., Jr., and Raghavan, R.:
s = skin factor “Unsteady-State Pressure Distributions Created by a Well with
t = time, days a Single Infinite-Conductivity Vertical Fracture.” SPEJ
T = reservoir temperature, degrees Rankine (Aug. 1974), 347-360.
t. = Pseudo equivalent time
t, = Equivalent time (cumulative(t)/q(t)), days Appendix A
tDA= Dimensionless time based on area, A
tc= Q(t)/q(t) (A- I)
t,. = Dimensionless time, based on ~~a

t~x~ = Dimensionless time, based on x;

w = Fracture width, H
xc = Distance from well to reservoir boundary
(Cartesian coordinate system) where, A[6)l=tii)--@)
x~= Fracture half length, R
q = Initial gas compressibility factor [)

y= Gas compressibility factor at average pressure p’dp’

and, ~P) = 2 [P(P’)Z(P’)
o
Greek letters:
4 = Hydrocarbon porosity, fraction t~=a. t (A-3)
k = Carter’s draw-down variable

591
 SPE 49222
8 RAM G. AGARWAL, DAVID C. GARDNER, STANLEY W. KLEINSTEIBER AND DEL D. FUSSELL

t,~=a. t. (A-5)

(2.637 X 10-’ )(24)k

where, a =
$(Pc, ),r;.

1422T~t)
—.
1
(A-6)
P,D M)

Pu/) = (P/) +~) (A-7)

9.OT ‘ q(t’)dt’
Qu,, . % . (A-8)
pw,, $hAm(p)r~. ,,J v(~)cx(~)

(A-9)

where: Am(p) = [m(pl) - m(pBHp)]

Dimensionless times in Eqs. (A-3) through (A-9) are based

on r~ti . These are multiplied by r~ti /A to convert them so that
r2
they are based on the area, A. For example, ‘)),4= taj>”~ ,

r2 ‘I)A
and Q,,,d = Q.{) . ~ = —
P“/)

S1 Metric Conversion Factora

acre x 4.046 873 E+03 = m2
bbl X 1.589 873 E-O I = m3
Cp x 1.0” E-03 = Pas

ft X 3.048” E-o I =m
md-ft x 3.008 142 E+02 = pmz
psi x 3.048” Etoo = kPa
psi-i x I .450 377 E-04 = Pa-l
CU ft X 2.83 I 685 E-02 = m3
md x 9.869 233 E-04 = ~mz
=. K
“R x5/9

‘ Conversion factor is exact

592
 T

II
.
m

.-M
L

Cc ccc

.. ..
— -J m
.-oil
L &9-
, I

: u
Un
—

.-
‘*
0

1’
O.:*
;
l“:
—

,
I
- ●
●

— ,.

7
—

—— —

-,,- WI
 I E+ll

T
--t
/ -~
—
●

–t-– —
——

*“
,’ .——
——.
.“
. + .

I_
L,
—.
I
I
~

— .- .j_
—:.—
+“ ;
— I
0
I
d
$ ?
, w,
. -----

1.EW E+02

1 I

-- ‘-T —i— E+O1

.,.:
I
_——, .—
I —. Etcm

lj*
—— l!
1.

.’
I

i
I
..—

..
I
I - L—--+ —

..‘“;
j. ——

.,
#‘1 L’
—IL
1-- I

-j
i
—
,,

I #
#’
i, —. #
1 Em
‘: I
“
ii ‘s
t“.

I I 1 EQS
I .-w
L
 I

l!
-.. ~. ...+ ------ –-+

,
/

------+

_.__: -...~...~ +.: /

‘--;{
.r+. .

1’ ‘/”/“
—. _.,—-.’ —
I /:
1;
2
I ,“ {
g -- ,:——
.1
+ /~’ .,’

_, –. ——
—-—.———.—
/L ~i.::_
.?
–..
1/’

~ /’ ‘ -$--------
//
-“T-—-”

y ~z I
.’I
..”
--- .. .
— I
0
:
. . .
.

I I I
l!
I

I I
1}

!
I
I
,
.,
\
I

I
I
L

I m
.

.-M
L

---- ----
,W ‘w!
 ,. , .; . ..
1

,,
I

,,

..
~
.-M
l.L

E*

E*O7

E+02

r-
.-M
E+O1
.-aiJ LL
L

597
,- ~.. --.._.
r 1-
k. —I
I

i_...—. ——. —
I
t
.;
——— .—.
.,—-.
I i; ,. ,“

—1 ‘,amdlt

598