A New Approach to the

Hyperbolic Curve
D.R. Long, * SPE, Williamson Petroleum Consultants Inc.
M .... Davis, * * SPE, Williamson Petroleum Consultants Inc.

Summary. An efficient and timesaving approach to hyperbolic-decline-curve analysis has been developed that has made it
possible to determine the hyperbolic b exponents characterized by thousands of wells. After extensive use of the technique, we
concluded that the range of the b exponent previously prescribed by Arps is too narrow.

Downloaded from http://onepetro.org/JPT/article-pdf/40/07/909/2221036/spe-16237-pa.pdf by Universidad Nacional Autonoma De Mexico user on 21 April 2021
Introduction
The extrapolation of decline curves is a valuable yet often misused line. This extrapolation, or constant-percentage-decline projection,
tool for predicting the future production rate and ultimate recov- portrays a decline exponent b=O and accurately characterizes the
ery of a producing well. Despite all the evidence of the common production performance of many producing wells. The b exponent
applicability of hyperbolic curves to the extrapolation of historical is the rate of change of the slope function, at, of the curve with
production performance, many engineers continue to use only the respect to time. Hyperbolic declines frequently and with better relia-
constant-percentage-type decline analysis, mainly because of the bility describe the anticipated trend of future production for select
difficulty historically associated with use of the hyperbolic curve. wells. These declines exhibit a concave upward appearance on
While some engineers may use a French curve or a series of graphic production plots of semilog rate vs. time.
constant-percentage declines with sequentially decreasing decline The curvature or bending feature of the hyperbolic decline curve
rates to compensate for the conservative results of the constant- has not been conducive to widespread use in extrapolating produc-
percentage-decline assumption, these procedures have neither the tion rates because the procedures have been complicated, not well
experience nor the theoretical background documentation availa- understood, and/or time consuming. The prevailing technique used
ble for hyperbolic declines. in the past to fit historical data to a hyperbolic curve by graphic
This paper emphasizes several circumstances, among many ob- methods involved repetitive plotting of data points using trial-and-
served throughout the U.S., where the production curve exhibits error techniques to obtain a straight line. One overlay technique
a decline exponent b> 1.0 and demonstrates the ease with which commonly requires many different overlays. Computer software
engineers may use hyperbolic decline curves by using overlays in has generated a variety of curve-fitting solutions that are difficult
estimating future production rate and ultimate recovery. Relative- to carry around in a briefcase.
ly high b exponent values make the correct use of the hyperbolic In addition to the excessive amount of time or awkwardness gener-
curve more essential than ever before. When this overlay technique ally inherent in hyperbolic-decline-curve-analysis techniques, there
is used to match the history of a producing well, the engineer ac- are other restraining factors. Some of the petroleum economic cash
quires a visual perception of the uniqueness of the solution that is flow programs do not support b exponents> 1.0. Another restraint
not readily apparent in a computer's numerical solution. stems from the continued flattening of the hyperbolic curve with
When Arps 1 presented his analysis of decline curves in 1944, time, which may lead to unrealistically high projected lifetimes and
he concluded that all hyperbolic decline exponents ranged from reserve estimates. To compensate for these prevailing criticisms,
o:5 b:5 1. O. About 25 years passed before authors began to ques- many engineers use a series of sequentially decreasing exponential
tion this conclusion, even then restricting themselves to special cases decline rates in the construction of type curves; however, this does
in specific areas. Gentry and McCray 2 demonstrated the theoreti- not depict the true character of a well's performance or provide
cal consistency of b exponents> 1.0 with a reservoir model. Brown for determination of the b exponent that may be characteristic and
et al. 3 attributed high b exponents to transient flow behavior in valuable in the analysis of other nearby wells.
the Hay Reservoir of Wyoming. McNulty and Knapp4 also saw An overlay technique was developed to permit the engineer to
this in a selected well in Oklahoma. Bailey 5 noted a wide range extrapolate the hyperbolic decline curve of a production plot of semi-
(0.2 :5 b :5 3 .4) of the b exponent in his analysis of producing gas log rate vs. time with almost as much ease as a straight line constant-
wells in the Wattenberg field in Colorado. From experience with percentage-decline projection. This technique uses a suite of type
the technique described here, widespread evidence exists that the curves plotted on a clear plastic overlay that has been custom de-
b exponent exceeds 1.0 in many areas. signed for the scale of graph paper being used. The type curves
We hope that the observations presented in this paper and the are marked to exhibit instantaneous equivalent annual decline rates.
hyperbolic-curve-analysis technique described will lead to a better In addition to its simplicity, this technique provides the engineer
understanding of the hyperbolic decline equations and will encourage with a straightforward visual understanding of historical perform-
practicing engineers, where applicable, to make greater use of the ance and a clear grasp of potentially expected future production
hyperbolic decline curve. trends.

Decline-Curve Analysis Type-Curve Overlay

The extrapolation of production decline performance with rate/time The type-curve-overlay technique is similar in theory to the methods
plots of oil and gas volumes historically produced is one of the presented by Slider 6 and Fetkovich. 7 Each technique recognizes
primary tools used by the petroleum engineer to evaluate produc- the importance of a unique b exponent in describing the historical
ing properties. The most common decline-curve technique involves production trends of a well. The new overlay, however, offers sever-
plotting monthly volume~ on a logarithmic scale vs. time on a linear al advantages. Slider's approach requires a different overlay for
scale and extrapolating the observed trend into the future as a straight each b exponent, which causes the engineer to sift through any num-
• Now with Long Consultants Inc.
ber of overlays in his attempt to match historical data to the in-
"Now with Needlepoint of Midland. digenous b exponent. Fetkovich used an overlay technique with
Copyright 1988 Society of Petroleum Engineers multiple b exponent curves on the same overlay, but his method

Journal of Petroleum Technology, July 1988 909

 10,000 a constant-percentage decline. This second projection will thus ef-
.
eo
HYPERBOLIC DECLINE TYPE CURVES
ao~ 1 month
3-CYCLE SEMILOG x 20 YEARS
fectively be a tangent to the hyperbolic curve. We refer to this sec-
ond segment as the CPD (constant-percentage-decline) tail. Certain
selected tangent points to the hyperbolic decline curve are indicat-
ed by tick marks on the type-curve overlay. These tangent points
l000~~~~,,~------------ ____________________ ~
are noted to indicate the instantaneous constant-percent annual
decline rate.
By positioning a properly sized type-curve overlay on a semilog
rate-vs.-time production plot and moving the overlay on the graph,
the engineer obtains a best fit with a particular b exponent in much
the same way he would with a straightedge in working with a
constant-percentage decline. The single rule is that the overlay must
be kept square with the graph paper. The overlay may be moved
up and down and left and right to obtain what is judged to be a
best fit of all the data or a selected portion of them. When a best
fit is judged to have been found, the engineer is ready to extrapo-

Downloaded from http://onepetro.org/JPT/article-pdf/40/07/909/2221036/spe-16237-pa.pdf by Universidad Nacional Autonoma De Mexico user on 21 April 2021
late and document an estimate of the future expected or potential
Fig. 1-The hyperbolic-curve overlay. performance trend.
Obtaining forecast parameters of initial rate, qi' initial slope
applied only to rate/time production plots on log-log paper. The function, ai' and the b exponent takes no more than a few minutes.
type-curve approach suggested here uses a single overlay with mul- By using the equations in Table 1, an engineer may, with a scien-
tiple curves representing a suite of b exponent values ranging from tific calculator, quickly estimate reserves or code directions for entry
0.3 to I .7 and enables the engineer to choose the best fit of a curve into a petroleum economics cash-flow program. It is even easier
readily. This is particularly important because a much broader range with a programmable calculator. The following section shows an
of b exponents above the 0=5b=51.0 range prescribed by Slider 6 example problem that uses these equations and demonstrates the
and Arps I is supported by today's data. The overlay, illustrated technique.
in Fig. 1, has been developed for rate/time production plots on semi-
log graph paper, which is commonly used by practicing engineers. Example Problem: A Spraberry Trend Well
All the curves on the overlay have a common initial starting point
Hyperbolic Segment. Fig. 2 displays production data of a Spraberry
and slope and fan out as a result of having different b exponents.
trend well in west Texas. Fig. 3 shows the overlay fit to the histor-
A criticism of the hyperbolic curve expressed by many engineers
ical data. The hyperbolic curve that best represents past produc-
is that it frequently yields an unrealistically high reserve estimate
tion history has a b exponent of 1. 7.
and lifetime because the curve continually flattens with time. To
The fundamental equation defining the relationship between the
compensate for this just criticism, a solution has been implement-
hyperbolic slope function, at, and the b exponent is
ed that limits the minimum decline rate value. Once the projected
hyperbolic curve flattens to a predetermined instantaneous decline
value, the forecast should be changed from a hyperbolic curve to at =ai +bt . ....................................... (1)

TABLE 1-DECLINE-CURVE EQUATIONS

Cumulative
RatelTime Volume Time/Rate Other Useful
Decline-Curve Type Relationship Relationship Relationship Relationships
Constant percent, b = 0.0 -(qj-q) -In q/q -1
q=qj(l-d)1 N=--- t=----- ad=---
In(l-d) In(1 -d) In(1-d)

or or or d= 1-e -.lIa

1 qj t(qj-q)
q=qj;U; N=a(qj -q) t=a In-- N=--
q In q/q
--
qj
a q =8j--
q

Harmonic, b = 1.0
q=qj( 1 +:/aj)
qj
N=qja j In-
q
t=a j ( ~ -1) at=aj+t

-1
aj=
In(1-d j)
--

at =aj +bt
(
btr'/b a·q·b a·
Hyperbolic, b";1.0, b>O.O q=qj 1+a;- N= 1'-'b (qj l-b _ql-b) t= -'-(e b In qjlq -1) -1
b . aj=
In(1-d j)

1. The percent decline function d is normally expressed in annual terms but the decline function a (or the initial decline rate term a J) is common~y expressed in monthly units.
The consistency of units is imperative to the successful use of all these equations.
2. Many forms of similar decline-curve equations use an a that is the reciprocal of the a used in these equations. The reciprocal a value is sometimes referred to as D. For example,
if the decline rate is 12% per year, the annual value for a is a ~ -1/1In(1 - 0.12)] ~ 7.823 years; the monthly value for a is a ~ 7.823 x 12 ~ 93.872 months; the daily value for
a is a ~ 7.823 x 365 ~ 2,855 days; and the monthly value for 0 is 0 ~ 1193.872 ~ 0.01 065 months -1 .
3. t refers to the duration of the interval between an initial rate (q; or q 0 if working from the origin of an overlay) at the beginning of a decline-curve segment and any other rate q.

910 Journal of Petroleum Technology, July 1988

 ,O.ooor+----------------.:..:H.:..:Yp:...:E::.R::.:B::.:O::.:L::.:IC;:D:.;E~CrL;;;IN;;:E;.TYrP-=E..::C:.::U:::R:.:,V::;ES
95 .0= mon
.. 3-CYCLE SEMI LOG • 20 YEARS

,oor---------------------------------------~

72 84 96 108 120 132 144 156 168 110 192 20C 216 2211
1OL...-'f.;2~2t.""-:!.,.""-fo'.""-fo
..~72c"-;!;
..~..~'..~'20;;'-;-!";;-'2-;-!'44~1..~,..:;;'-;:!1..".w.,92!:-'-;;!204~2,~8'-::22!::-.'-'
'.'-'-'-!;:2-'-!:2.~36~.8~..~72!:-'-::..1:-'-:..
!:-'-:'±..:::'-:'::!::20:'-:'::!::'2'""~..~'..:-w.'..".w.'..~19L2~...L~2L,.~2L2•...-1

Downloaded from http://onepetro.org/JPT/article-pdf/40/07/909/2221036/spe-16237-pa.pdf by Universidad Nacional Autonoma De Mexico user on 21 April 2021
TIME - MONTHS
TIME - MONTHS
Fig. 3-Production history superimposed with type-curve
Fig. 2-Production history of a Spraberry trend well. overlay.

In preparing the type-curve overlay presented here, the value for Rearrange and substitute into Eq. 1 to solve for the time interval
the slope function at for each hyperbolic curve on the overlay at between qi and q2:
time zero, ao, was set to 1 month. Time zero refers to the time
at the origin of the overlay in the upper left comer. In rare circum- ad-a;
stances, it is convenient to use type curves with an alternative ori- t = - - ....................................... (4)
gin slope value, but an origin slope of 1 (month) suits almost all b
circumstances and is the only one illustrated. The at slope func-
tion changes with time as the curve becomes flatter. Conversely, (193.9 months-145.5 months)
the b exponent is unitless and remains constant with time and rep- -------------------- = 28.5 months.
resents the rate of change with time of the slope function, at. 1.7
The next step is to determine the quantitative value of the hyper- The general hyperbolic equation for production rate as a func-
bolic slope function at the beginning of the hyperbolic forecast seg- tion of time is:
ment. From Fig. 3, the time interval between the origin of the curve
and the beginning of the projection is 85 months. Considering the
need to know the values for the parameters of the hyperbolic fore- q=qi(l+ :tfllb . ................................ (5)
cast segment at the end of that time, Eq. 1 becomes a;=ao+bt I
= 1 month +(1.7)(85 months) = 145.5 months because the value of
the slope at the end of the overlay segment (the segment of the over- Substituting values into Eq. 5 to calculate the rate at the end of
lay b= 1.7 curve between zero time and 85 months) is equal to the the hyperbolic segment gives
slope at the beginning of the forecast segment. The equation for
reserves to be produced during the hyperbolic segment is (1.7)(28.5 months)l-III.7
q2 = 850 bbllmonth [ I +
a;qi b 145.5 months .
N = ---(q; I-b -q I-b) . . . . . . . . . . . . . . . . . . . . . . . . . . . . (2)
I-b
=718 bbllmonth [114 m 3 /month].
At the beginning of the projection and from the fit of the curve
to the data, qi is 850 bbl/month [135 m 3 /month] for the hyper- The calculation procedure demonstrated above is a more precise
bolic forecast segment. If the engineer knows that a 6% CPD tail method of arriving at q2 than a visual inspection of the extrapo-
is representative of the formation in this particular area, he reads lated curve, but it is a little more time consuming.
off the producing rate at the tick mark labeled 6, which is 720
bbl/month [114 m 3 /month] for the rate at the end of the hyper- CPD Segment. The general equation for reserves to be produced
bolic forecast segment. Substituting values into Eq. (2) gives during the constant-percentage-decline segment of the forecast is

(145.5 months)(850 bbllmonth) 1.7 -(q; -q)

N N= ..................................... (6)
1-1.7 In(l-d)
x [(850 bbllmonth) 1-1.7 -(720 bbllmonth) 1- 1.7] N is obtained if we use an assumed economic limit, q3' of 100
bbllmonth [16 m 3 Imonth] and other parameters previously deter-
=21,768 bbl [3461 m 3 ]. mined, then substitute q2 for qi and q3 for q in Eq. 6 for the
constant-percentage-decline segment:
If the engineer is not content with reading q2 (the rate at the end
of the hyperbolic segment) off the overlay, it may be obtained by (720 bblimonth -100 bbl/month)(l2 months/yr)
making the following calculations. The first thing that must be solved N=-------------~-
for is the hyperbolic slope function at the end of the hyperbolic In(l-0.06/yr)
segment that is numerically equivalent to the slope at the begin-
=120,242 bbl [19 117 m 3 ].
ning of the CPD tail. This was previously assumed for example
purposes to be a 6% annual decline rate:
Total reserves are therefore 21,768+ 120,242= 142,010 bbl [22 578
-I m 3 ].
...............' ..................... (3) This same type of analysis may also be applied to the harmonic-
In(1-d) decline curve as well as those circumstances where the exhibited
( - 1)(12 months/yr) hyperbolic performance is better described by a b exponent < 1.
193.9 months. The equations for the unique harmonic case where b = 1 are as shown
In(l-0.06/yr) in Table 1. The same technique is applicable.
Journal of Petroleum Technology, July 1988 911
 10,_ 10,000

'000 .000
\.
.00 ~ '00
'mu
•• 12 2" 38 41 10 72 . 98 108 120 132 144 156 168 180 192 204 216 228
TIME - MONTHS
•• n~~u~n"M~mm~m~~=_m~
TIME - MONTHS

Downloaded from http://onepetro.org/JPT/article-pdf/40/07/909/2221036/spe-16237-pa.pdf by Universidad Nacional Autonoma De Mexico user on 21 April 2021
Fig. 4-Production history of a Clinton sand well. Fig. 6-Production history of a Sooner trend well.

10,000 Nomenclature
a = negative reciprocal of slope of a constant-

II\:~
percentage-decline curve
ad = value of negative reciprocal of slope of a decline
.000
I"
- curve with a slope equivalent to d
ai = negative reciprocal of slope of hyperbolic decline
curve at beginning of a forecast segment
at instantaneous decline hyperbolic slope function or
.00
negative reciprocal of slope of hyperbolic curve at
time t subsequent to beginning of a curve segment
ao = negative reciprocal of slope of hyperbolic decline-
curve overlay at upper left origin of a suite of
curves
•• n~~u~n"M~mm~m~~=_m~ b = hyperbolic exponent
d = constant-percentage-decline rate
TIME - MONTHS

Fig. 5-Production history of an Austin Chalk well. N = reserves scheduled to be produced during a
production-decline-curve segment
q = producing rate at time t
Conclusions
qi = producing rate at beginning of a forecast period or
Several papers have indicated observations of b exponents <:! 1.0 segment
in isolated circumstances. Through the use of the hyperbolic-overlay
qt = producing rate at specified point of time t
technique described, many areas throughout the U.S. where the
t = length of time period, particularly, from beginning
decline exponent is consistently > 1.0 have been identified. The
characteristic of the producing intervals in these areas includes rocks to end of a decline-curve segment
that are fractured and tight or relatively tight. A b exponent of 1.7
is believed to be the highest that commonly occurs based on our Acknowledgments
experience and observations. Higher exponents are the exception We thank Williamson Petroleum Consultants Inc. for permission
but are noted in the referenced literature and in our own experi- to publish this paper and its staff members who contributed to our
ences and observations. experience and helped make this paper more complete and of greater
Four outstanding areas where the decline curves of wells charac- value.
teristically exhibit b exponents > 1.0 are the Spraberry trend of
west Texas (Figs. 2 and 3), the Clinton sands of the Appalachian References
basin (Fig. 4), the Austin Chalk formation of south Texas (Fig. I. Arps, J.J.: "Analysis of Decline Curves," Trans., AIME (1944) 160,
5), and the Sooner trend of Oklahoma (Fig. 6). These areas, where 228-47.
examples are available by the thousands, make up only a fraction 2. Gentry, R.W. and McCray, A.W.: "The Effect of Reservoir and Fluid
of the number of sedimentary basins in which the hyperbolic b ex- Properties on Production Decline Curves," JPT(Sept. 1978) 1327-41.
ponent > 1.0 has been noted. 3. Brown, c.A., Erbe, C.B., and Crafton, J.W.: "A Comprehensive Reser-
The curve-fitting technique with the type-curve overlay described voir Model of the Low Permeability Lewis Sands in the Hay Reservoir
here is efficient and practical in analyzing decline curves of wells Area, Sweetwater County, Wyoming," paper SPE 10193 presented at
the 1981 SPE Annual Technical Conference and Exhibition, San Anto-
that exhibit the characteristics of hyperbolic production perform- nio, Oct. 5-7.
ance. This technique uses commonly generated historical produc- 4. McNulty, R.R. and Knapp, R.M.: "Statistical Decline Curve Analy-
tion plots and requires neither special graph paper nor repetitive sis, " paper SPE 10279 presented at the 1981 SPE Annual Technical Con-
trial-and-error data plotting. The overlay permits the experienced ference and Exhibition, San Antonio, Oct. 5-7.
evaluation engineer to extrapolate rate/time production data directly 5. Bailey, W.: "Optimized Hyperbolic Decline Curve Analysis of Gas
while preferential weight and judgment are given to the data be- Wells," Oil & Gas J. (Feb. 15, 1982) 118.
lieved to be representative. The reserve implications of alternative 6. Slider, H.C.: "A Simplified Method of Hyperbolic Decline Curve Anal-
fits of a curve may be calculated quickly when a simple scientific ysis," JPT (March 1968) 235-36.
calculator is used. As a result of both the usefulness and speed of 7. Fetkovich, M.L: "Decline Curve Analysis Using Type Curves," JPT
(June 1980) 1065-77. •
this overlay technique, it has been possible to determine the hyper-
bolic b exponents characterized by literally thousands of produc- JPT
ing wells and to conclude that b exponents exceeding 1.0 are typical Original SPE manuscript received for review March 8, 1987. Paper accepted for publica-
tion Nov. 17, 1987. Revised manuscript received March 21,1988. Paper (SPE 16237)
of the production performance character of wells in many areas first presented at the 1987 SPE Production Operations Symposium held in Oklahoma City,
of the U.S. March 8-10.

912 Journal of Petroleum Technology, July 1988

Discussion of A New Approach
to the Hyperbolic Curve
B.G. Kelkar, SPE, U. of Tulsa
G. Perez, SPE, U. of Tulsa

In "A New Approach to the Hyperbolic Curve" (July 1988 JPT, unit month is plotted as a function of time, along with the extrapo-
Pages 909-12), Long and Davis conclude that the value of bin Arps' lation. Although not shown, the data could be matched by using
decline curves can exceed 1 in many cases. They state that the type the type curve presented by Fetkovich where
of well that exhibits such behavior is relatively tight (low-
permeability) fractured wells. They suggest an empirical method re/rw = 100,
to analyze such wells that allows the use of hyperbolic decline curves b = 1.0,

Downloaded from http://onepetro.org/JPT/article-pdf/40/07/909/2221036/spe-16237-pa.pdf by Universidad Nacional Autonoma De Mexico user on 21 April 2021
in the beginning and a constant-percentage decline curve later. This qi = 2,494 units/month, and
discussion is intended to show that this method is not necessary Di = 0.02286.
if the data analysis is carried out in a manner suggested by
Fetkovich I and Fetkovich et ai. 2
With a conventional technique, the remaining reserves can easily
When Arps' curves are used for decline-curve analysis, two as-
be calculated. For example, if the currenfrate is 720 bbllmonth
sumptions are implicitly made: the well is producing in a boundary-
[114 m 3 /month] and the abandonment rate is 100 bbllmonth [16
dominated region, and the production data are collected at constant
m 3 /month], the remaining reserves can be calculated by
wellbore pressure. If data are not collected at constant wellbore
pressure, the data should be normalized to reflect constant-wellbore-
pressure production. If the well is not producing in a boundary-
q q
Np=-ln-, .................................. (D-l)
dominated region, the data should be analyzed with transient-state D qa
type curves, as suggested by Fetkovich. 1 For most wells, the
boundary-dominated state is reached in a matter of days. For tight where D, current decline rate, is calculated as
gas reservoirs and/or fractured gas reservoirs, however, the tran-
sient ~tate may continue for years. If data in transient state are mis- 1 1
interpreted with Arps' decline curves, the analysis results in -=-+t= +90, ....................... (D-2)
artificially high values of b, sometimes even exceeding 2. D Di 0.02286
Long and Davis present several examples of reservoirs where
high values of b were noted (Figs. 2 and 4 through 6 in their paper). therefore D=0.00748 month-I. Substituting Eq. D-2 in Eq. D-l,
To check the necessity of using a value of b> 1, we reanalyzed the remaining reserves are 190,018 bbl [30 210 m 3 ]. Figs. D-2
the data with the method suggested by Fetkovich. I Fig. D-l shows and D-3 show matches for the Clinton sand well and the Sooner
the match for the Spraberry trend well where production rate per trend well, respectively. For both wells, b is observed to be 0.8.

Start Date: 01/01/80

loooo.oor---~--~--'---'---'I'---r---.---~----~----'

qj 24114
Dj 0.0228511
b 1.00

~~ I
1000.00 ~

I
flloo.oor---r---+---+---~---~---r----r---+---+---1
i
I
I

10.00 '::-::-----:-::'=---:-::'=---::-::'-::::----::~:____:_;;f_::::____:_:::'-::::----:--:-;'-;:::___:~::___:=""::::___:=_'
0.00 20.00 40.00 60.00 80.00 100.00 120.00 140.00 160.00 180.00 200.00
Time. Ifonth.

Fig. D-1-Extrapolatlon of future production for Spraberry trend well.

Journal of Petroleum Technology, December 1988 1617

 Slarl Dale: 01/01/80

qj 451
Dj D.042793
b 0.80

.- \
--
0'
......
'-"'
0'
1/
"0

Downloaded from http://onepetro.org/JPT/article-pdf/40/07/909/2221036/spe-16237-pa.pdf by Universidad Nacional Autonoma De Mexico user on 21 April 2021
Q
0'

tDd =Dit
1\
Fig. D-2-Arps decline curve match for Clinton sand well.

Starl Dale: 12/30/80

qj 4803
Dj 0280984
b 0.110

Fig. D-3-Arps decline curve match for Sooner trend well.

Neither well exhibited any transient-state data. Note that the data proposing a new technique, the authors should have checked their
are read directly from the paper and may involve small reading data against a conventional technique.
errors. However, overall analysis is fairly accurate. Although we
did not analyze the Austin Chalk well, it also could be analyzed
with a conventional technique. Acknowledgment
From the above analysis, it seems that the conventional technique, The data in this discussion were analyzed with software developed
as explained by Fetkovich, 1 is an adequate tool to analyze produc- by Mannon Assocs. Inc. We thank Mannon Assocs. for donating
tion decline data and is based on a sound theoretical basis. Before this software to the U. of Tulsa.

1618 Journal of Petroleum Technology, December 1988

Nomenclature Subscripts
b = decline exponent a = abandonment
D = decline rate i = initial
Np = cumulative production References
q = production rate 1. Fetkovich, M.J.: "Decline Curve Analysis Using Type Curves," JPT
qDd = decline curve dimensionless rate (June 1980) 1065-77.
r e = reservoir radius 2. Fetkovich, M.J. et al.: "Case Study of a Low-Penneability Volatile
r w = wellbore radius Oil Field Using Individual-Well Advanced Decline Curve Analysis,"
paper SPE 14237 presented at the 1985 SPE Annual Technical Con-
t = time ference and Exhibition, Las Vegas, Sept. 22-25.
tDd = decline curve dimensionless time
(SPE 18728) JPT

Downloaded from http://onepetro.org/JPT/article-pdf/40/07/909/2221036/spe-16237-pa.pdf by Universidad Nacional Autonoma De Mexico user on 21 April 2021
Author's Reply to Discussion of
A New Approach to the Hyperbolic Curve
D.R. Long, SPE, Long Consultants Inc.

Kelkar and Perez indicate that the empirical method discussed in Perez write about "where the value of r e/ r w = 100 .... " Neither
our paper is not necessary if the data analysis is carried out in a the drainage radius nor the wellbore radius is a parameter in the
manner suggested by Fetkovich 1 and Fetkovich et al. 2 I have no Arps equation. They are writing about an equation set forth by
argument with their conclusion. Our paper was intended to provide Fetkovich, not the empirical Arps equation. They do not make this
a perspective of the range of the Arps b exponent and to explain clear to the reader.
our experience with a technique that is both fast and accurate when In their Spraberry trend well example, Kelkar and Perez did not
used under appropriate circumstances. The empirical method was manage to match the earliest-time data. At least the line on their
not represented as the only method of solution for wells that exhibit graph does not pass through the data for the initial period that was
characteristics of high Arps b exponents. Our technique requires about 12 months long. After 12 months of capacity rate production,
the plotting ofthe rate/time data once on sernilog rate-vs.-time graph I usually feel that I have a pretty good idea of what a Spraberry
paper. While the technique set forth by Fetkovich may have a well is going to produce ultimately, but Kelkar and Perez don't
stronger technical basis, it normally requires multiple plotting of match that part of the history. This appears to be a problem with
the data on log-log paper. For most of the work that I do, I need their use of the Fetkovich technique, not a problem with the
a technique that requires less time. I fmd that the technique suggested technique itself.
in our paper is frequently all the analysis that is warranted, The major advantage to the technique illustrated in our paper is
considering the availability and accuracy of the data. the ease and efficiency of use. Where the accuracy of the data and
That early production performance in low-permeability reservoirs the value of the answer warrant the additional time, the Fetkovich
frequently may be characterized by high values of Arps equation technique is possibly the preferred procedure. A study of his work
b exponent (b> 1.0) is not contested by Fetkovich et a[.3 They is certainly important to a better understanding of what decline
acknowledge that the curve characteristics of high Arps b exponents curves indicate about the reservoir depletion process. His most
fit their transient constant-wellbore-pressure rate/time data. The recently published paper3 presents not only the most compre-
question is whether or not it is appropriate to extrapolate the decline hensive but also the most comprehensible discussion of the technique
curve characterized by such high Arps b exponents into the future. he first set forth in 1973.
Fetkovich et at. attribute occurrences of high Arps b exponents to
transient flow conditions and present some convincing technical References
support. They take the position, however, that data fit to the Arps 1. Fetkovich, M.J.: "Decline Curve Analysis Using Type Curves," JPT
equation will extrapolate to infinity with no rational basis of (June 1980) 1065-77.
terminating the forecast. My point of disagreement is that I believe 2. Fetkovich, M.J. et al.: "Case Study of a Low-Permeability Volatile
that through the use of experience in a particular area and given Oil Field Using Individual-Well Advanced Decline Curve Analysis,"
other consistent conditions, a rational basis of terminating the paper SPE 14237 presented at the 1985 SPE Annual Technical Con-
forecast does exist through what I call the constant-percent-decline ference and Exhibition, Las Vegas, Sept. 22-25.
(CPD) tail. 3. Fetkovich, M.J. et al.: "Decline-Curve Analysis Using Type Curves-
Case Histories," SPEFE (Dec. 1987) 637-56.
Kelkar and Perez present their match for the Spraberry trend well
with a b exponent of 1.0. We obtained a match for this well with
the Arps equation and a b exponent of 1.7. Note that Kelkar and (SPE 18942) JPT

Journal of Petroleum Technology, December 1988 1619