Useful Concepts for

DeclineĆCurve Forecasting, Reserve

Estimation, and Analysis
M.J. Fetkovich, SPE, E.J. Fetkovich, SPE, and M.D. Fetkovich, SPE, Phillips Petroleum Co.

Summary
q(t) + q ińe
NJDińƪ1*ǒpwfńpRǓƫNjt , . . . . . . . . . . . . . . . . . . . . . . . (1)
The purpose of this paper is to give engineers responsible for making
forecasts and determining reserves for numerous operated or nonoper-
ated wells some guidelines and fundamental concepts to allow them to ƪ ƫ
where D i( p wf) + q i ńN puoń 1 * ǒp wfńp RǓ . . . . . . . . . . . . . . (2)
make these forecasts and determinations more quickly and accurately.
Both qi and Npuo are functions of the constant pressure, pwf . For pwf +0,
Introduction
Di + qi maxńN puo,o, . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (3)
Decline-curve analysis, based on the Arps1 equations, has always been
considered to be purely empirical with no basis on physical laws gov- where Npuo,o is the ultimate recoverable to q(t)+0 and pwf +0.
erning the flow of oil and gas through the formation. The works of Fet- Note that different levels of constant wellbore pressures, pwf , al-
kovich2–11 and others have attempted to place decline-curve analysis ways result in an exponential decline and Di is the same for all levels
on a sound, fundamental basis using the constant wellbore pressure of backpressure. (This is not the case for derived forms of the hyper-
analytical solution and simple combinations of material-balance equa- bolic equation.) The goal of any prudent operator to maximize pro-
tions and pseudosteady-state rate equations to derive rate/time decline duction and recoverable reserves and to minimize drainage of re-
equations for oil and gas wells. (See additional references listed in Ref. serves to offset wells is to maintain the flowing pressure, pwf , as
14.) The derivations illustrate under what circumstances specific values close to 0 as is economically possible. Therefore, pwf [0 was the ba-
of the hyperbolic decline exponent, b, should result. It is from these der- sis for deriving hyperbolic forms of decline equations which turn
ivations that the variables in Arps equations can be expressed in terms out to be very useful for understanding some important concepts of
of reservoir variables and reservoir engineering concepts. Decline- decline-curve analysis.
curve analysis, or more specifically rate/time analysis, is not simply an
art based on applying a purely empirical equation to be analyzed with Solution-Gas Decline Equations (Derived) (Below Bubble-
statistical approaches. Reliance on a statistical analysis, void of reser- point). As presented and discussed in Ref. 3, the material-balance
voir engineering concepts, often leads to unrealistic and unreliable and rate equations used to derive rate/time equations for solution-
forecasts and reserve estimates. gas-drive oil wells are

Oil and Gas Decline Equations p R2 + * ǒp Ri2 ńN puoǓN p ) p Ri2 , . . . . . . . . . . . . . . . . . . . . . . . (4)

Arps Equations. Conventional decline-curve analysis is based on the
(p Rńp Ri) ǒp R2 * p 2wfǓ
n
empirical equations of Arps.1 In his work, Arps proposed a set of and q o + J Ȁoi . . . . . . . . . . . . . . . . . . . . (5)
equations described by an exponent, b, having a range between 0 and
1. The resulting rate time and cumulative production equations are The ratio ( pR / pRi ) is used in the rate equation to approximate the
shown in Table 1 (see also Ref. 14). He used his equations analyzing decrease in relative permeability, kro , with pressure depletion. The
some 149 oil fields’ production data, assembled by Cutler–1924, in reader is referred to Ref. 6, Table 9 and Fig. 25 where the works of
1945, for determining the distribution of b. He found that the range Levine and Pratts and Vogel were used to justify the pressure ratio
of exponent, b, was between 0 and 0.7 with 90% having b less than approximation to the kro change in Eq. 5.
0.5. No harmonic declines were found and less than 15% had a b val- The derived rate/time equation for an oil well producing against
ue less than 0.1. The rapid development and virtual wide-open pro- a constant pressure, pwf , is for all backpressure curve exponents n
duction of many of these oil fields resulted in the ideal situation for u0.5, hyperbolic.
constant wellbore pressure production over their entire life. qi
Although the original Arps exponential and hyperbolic equations q(t) + (2n)1)ń(2n*1)
. . . . . . . . (6)
were developed empirically on the basis of oil production data, Arps NJ1 ) ƪ(2n * 1)ń2ƫǒq ińN puoǓ tNj
forms of equations can be rigorously derived for volumetric dry gas
systems. This permits a physical basis to be placed on each of the and exponential for n+0.5
variables, qi , q(t), b, and Di .
q(t) + q ińe *ǒqińNpuoǓ t . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (7)
Exponential Decline (Oil) (Derived) (Above Bubblepoint). The High-capacity, flowing oil wells that are “tubing limited,” friction
single-phase liquid, exponential decline equation, valid for undersat- pressure drop dominated, will tend to exhibit exponential decline,
urated oil or water flow, has been derived in Refs. 2 and 3. The deriva- as is the case for tubing limited gas wells.
tion is based on a simple material-balance equation and pseudostea- With the derivations based on pwf +0, qi +qi max. At backpressur-
dy-state or stabilized backpressure curve equation. Both equations es other than pwf +0, qi is a rate from the stabilized backpressure
are depicted graphically in Fig. 1. The use of this approach has also curve at the specified flowing pressure, pwf . Npuo is equal to the orig-
been used in Ref. 3 to derive the Arps form of equations for single- inal oil-in-place times a recovery factor (RF)
phase gas and solution-gas-drive volumetric systems.
The complete form of the derived exponential equation for N puo + N RF, . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (8)
constant wellbore pressure production is
where the recovery factor may be obtained from correlations, histor-
Copyright 1996 Society of Petroleum Engineers
ical performance, or Tarner type material-balance forecasts at some
selected gas/oil relative permeability curve.
Original SPE manuscript received for review Oct. 17, 1994. Revised manuscript received
Dec. 11, 1995. Paper peer approved Dec. 11, 1995. Paper (SPE 28628) first presented at the
Di , in terms of reservoir engineering terms, is for both the hyper-
1994 SPE Annual Technical Conference & Exhibition, New Orleans, Sept. 25–28. bolic and exponential forms

SPE Reservoir Engineering, February 1996 13

 TABLE 1—SUMMARY OF PRODUCTION DECLINE EQUATIONS
Decline Type Hyperbolic Exponential Harmonic

Rate–Time qǒ t Ǔ + q i ńǒ1 ) bDi tǓ

1ńb
qǒ t Ǔ + q ińe
D t
i qǒ t Ǔ + q i ń ǒ1 ) D i t Ǔ

Time to q(t) t+ NJƪq ńqǒtǓƫ

i
b
*1 NjńbD i t + lnƪq i * qǒ t ǓƫńD i t + NJƪq ińqǒ t Ǔƫ * 1NjńD i

Cumulative–Time ƪ
Q p + ƪq ińǒ1 * bǓD iƫ 1 * ǒ1 ) bD itǓ
ǒ b*1ńb Ǔ
ƫ ǒ
Q p + ǒq ińD iǓ 1 * e Ǔ
*D t
i ƪǒ
Q p + ǒq ińD iǓ ln 1 ) D i t Ǔƫ
Rate–Cumulative ƪ
Q p + ƪq ibńǒ1 * bǓD iƫ q (1–b)–qǒ t Ǔ
i
(1–b)
ƫ Q p + ƪq i * qǒ t ǓƫńQ D
i
Q p + ǒq ińD iǓ lnƪq ińqǒ t Ǔƫ
From Rate–Cum. D i + ƪ1ńǒ1–bǓƫǒq ińQ puoǓ D i + q ińQ puo Di is not definable; (Qpuo is infinite).
Di at q(t)+0
Di (oil) D i + ƪǒ2n ) 1Ǔń2ƫǒq ińN puoǓ n + 0.5; D i + ǒq ińN puoǓ Not derivable
Npuo +N (RF) where RF+f(kg /ko )
Di (gas) D i + 2nǒ q ińG Ǔ n + 0.5; D i + ǒ q ińG Ǔ Not derivable

G + Gi ǒRFǓ where RF + ƪ1 * ǒp ńp Ǔƫ
wf R
b(oil) where pwf [0 b + ǒ2n * 1Ǔńǒ2n ) 1Ǔ where n is between 0.5 and 1
b(gas) where pwf [0 b + ǒ2n * 1Ǔń2n where n is between 0.5 and 1

D i + ƪ(2n ) 1)ń2ƫǒq ińN puoǓ . . . . . . . . . . . . . . . . . . . . . . . . (9) Arps equations, qi , q(t), b, and Di , can be made from the derivation
of the single-phase gas equations.
The Arps decline exponent, b, can be expressed by the backpressure Derivation of the gas well decline equations is completely analo-
curve exponent, n, for wells producing at very low flowing pressures gous to that previously used to derive the single-phase liquid expo-
nential decline. The rate decline q(t) with time for a single-phase,
b + (2n * 1)ń(2n ) 1). . . . . . . . . . . . . . . . . . . . . . . . (10) single-layer, gas flow system, as in the single-phase liquid solution,
For n+1, the value that one would assume in the absence of an actu- is based entirely on reservoir pressure depletion of a closed finite
al multipoint test having been run on any well in the field, b would system. For single-phase gas, we use the pseudosteady state or stabi-
be equal to 0.33. A slope n+1 should be considered as being typical lized backpressure equation and material-balance equation, as de-
for most solution-gas-drive reservoirs and also would be equivalent picted graphically in Fig. 1. The derived rate/time equation for a gas
to assuming the Vogel IPR relationship. With n+1, well producing against a constant pressure, pwf , is for all backpres-
sure curve exponents nu0.5 a hyperbolic
D i + 3ń2 (q i ńN puo). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (11)
q(t) + q ińƪ1 ) (2n * 1) ǒ q ińG Ǔ tƫ
2nń(2n*1)
In the absence of a clearly defined decline exponent from field data, , . . . . . . . . . . . (12)
a value of b+0.3 should be assumed for a solution-gas-drive reser- and an exponential for the backpressure curve exponent n+0.5.
voir; i.e., n+1 in Eq. 10. However, indication of an unfavorable rel-
ative permeability, kg /ko , relationship would dictate a decline expo- q(t) + q ińe ǒq ińG Ǔt. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (13)
nent approaching exponential decline, b+0, because of the
anticipated lower recovery factor (see Fig. 2). With the derivations based on pwf +0, qi +qimax, and G+Gi , where
qimax is the stabilized absolute open-flow potential and Gi is the origi-
Gas Well Decline Equations (Derived). Most of the fundamental nal gas-in-place. At backpressures other than pwf +0, qi is a rate from
reservoir engineering concepts and definitions of the terms in the the stabilized backpressure curve at the specified flowing pressure,

Fig. 1—Rate and material-balance equations used to derive rate/time decline equations for gas and oil wells.

14 SPE Reservoir Engineering, February 1996

 Fig. 2—Decline exponent b as a reflection of recovery efficiency or drive mechanism.

pwf . G is the recoverable gas to an abandonment pressure equal to pwf , ponent b for 149 oil fields also tends to support the 0.5 value as a
where G is equal to the original gas-in-place times a recovery factor. physical upper limit of b. He found that 90% of the oil fields studied
had b values less than 0.5.
G + G i (RF), . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (14) As will be discussed later, a layered no-crossflow reservoir sys-
tem, or its equivalent, can result in decline exponents that cover the
where RF [ ƪ1 * ǒp wfńp RǓƫ . . . . . . . . . . . . . . . . . . . . . . . . (15) range of b between 0.5 and 1—values of b greater than 0.5 can be
used to identify layered no-crossflow reservoirs. In Arps distribu-
These same type definitions involving in-place volume and recov- tion study he found no value of b greater than 0.7. It should also be
ery factor were also used for single-phase and multi-phase oil sys- pointed out that attempting to fit all or some of the “transient” pro-
tems—i.e., qi and Npuo . duction rate data of a well with the Arps pseudosteady-state equa-
Note that the basic form of the gas well decline equations, Eqs. tion will result in an “apparent” b value higher than it really is. In
12 and 13, are identical to that of the Arps form (see Table 1). They some cases, it will even be greater than 1.
exactly reproduce or overlay the Arps type curve over its entire
length. Di expressed in more familiar and readily available reservoir Cumulative Production Equations. The cumulative production/
engineering terms is for both the hyperbolic and exponential forms time equation is for nu0.5, the hyperbolic form
D i + 2n ǒ q ińG Ǔ . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (16) 1ń(1*2n)
G pńG + 1 * ƪ 1 ) (2n * 1) ǒ q ińG Ǔ tƫ , . . . . . . . . (18)
All the terms in Eq. 16 can be calculated or estimated from initial
and for n+0.5, the exponential form
multipoint, pressure transient analysis results and geological data
G pńG + 1 * e *ǒ qińG Ǔ t, . . . . . . . . . . . . . . . . . . . . . . . . . . (19)
before any rate decline data is even available. Or in the case of a re-
placement well or offset location to be drilled, values from offset
wells can be used. both of which can be readily reduced to the Arps cumulative produc-
With regard to qi , qi is a rate from the stabilized wellhead or the bot- tion-time equations and rate/cumulative production equations
tomhole backpressure curve. If tubular friction is not significant, the shown in Table 1 (see also Ref. 14, Eqs. 6 through 11).
bottomhole and wellhead curves will be essentially the same except for
a hydrostatic head term. qi is not simply a producing rate at early time, Rate/Cumulative Production Equations (Derived). The rate/cu-
it is very specifically a pseudosteady-state rate at the surface. It can be mulative production equations for single-phase liquid and gas can
substantially less than actual early time transient flow rates as would be be directly derived from the backpressure curve and material-bal-
produced from low-permeability wells with large negative skins. ance equations (see Fig. 1). From the backpressure curve equation
The Arps decline exponent, b, can be expressed in terms of the at pRi and pR(t) where
backpressure curve exponent, n, for wells producing at very low
C gi + C g(t) + qńǒp R2Ǔ , . . . . . . . . . . . . . . . . . . . . . . . . . . (20)
n
flowing pressure
b + (2n * 1)ń2n . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (17)
then q(t)ńq i + ƪp R (t)ńp Riƫ . . . . . . . . . . . . . . . . . . . . . . . . . . (21)
2n

The exponent n from a gas well backpressure performance curve

can therefore be used to calculate or estimate b and Di . Using Eq. 17, From the material-balance equation slope
we can determine the physical limits of b, which is between 0 and
0.5, over the accepted theoretical range of the backpressure curve p R (t)ńp Ri + ƪG i * G p(t)ƫńG i. . . . . . . . . . . . . . . . . . . . . . . . (22)
exponent, n, which is between 0.5 and 1 for a single-layer homoge-
neous system (see Table 2). Substituting Eq. 22 into 21 and simplifying, we get
The harmonic decline exponent, b+1, cannot be obtained. In
q(t)ńq i + ƪ1 * G p(t)ńG iƫ . . . . . . . . . . . . . . . . . . . . . . . . (23)
2n
fact, no other investigators have been able to derive an exponent b
greater than 0.5 for any reasonable single-layer, homogeneous res-
ervoir system or drive mechanism. Arps study of the range of the ex- For n+0.5 or exponential decline, b+0,
q(t) + q i * ǒq ińG iǓG p(t). . . . . . . . . . . . . . . . . . . . . . . . . . . (24)
TABLE 2—DECLINE EXPONENT b AS A FUNCTION OF
BACKPRESSURE EXPONENT n The Arps hyperbolic rate/cumulative production equation con-
verted to our nomenclature with q(t)+0 is
Decline Exponent, b Backpressure Exponent, n
0 .50 (High k) G p + NJq bińƪ(1 * b) D iƫNj ƪq (1*b)
i
ƫ . . . . . . . . . . . . . . . . . . . . . (25)
.1 .56
.2 .62 Using the gas relationship b+(2n*1)/2n and Di +2n(qi /G), the hy-
.3 .71 perbolic rate/cumulative equation given in Table 1 reduces to
.4 .83
q(t)ńq i + ƪ1 * G p(t)ńGƫ , . . . . . . . . . . . . . . . . . . . . . . . . (26)
2n
.5 1.0 (Low k)

SPE Reservoir Engineering, February 1996 15

 Fig. 3—Decline exponents b for various reservoir drive mechanisms.

which is identical to Eq. 23. A similar approach was used for the nisms from the least efficient, totally undersaturated reservoir with
single-phase liquid and solution-gas-drive decline equations with a b+0, to a much more efficient gravity drainage or waterdrive re-
similar results. covery mechanism with a b+0.5. Different drive mechanisms and
Because both the rate/time and rate/cumulative equations are derived typical recoveries are also depicted in between.
using the same backpressure and material-balance equations, both Also shown on this figure is a corresponding rate/time decline
methods of analyzing production data should get identical results. type curve consisting of a transient period followed by depletion
stems ranging from b+0 to 0.5. Note that as b increases, the percent
Recovery or Drive Mechanisms and b. In many instances, rate/ recovery increases. (These same recovery values are also depicted
time data existing in the depletion period is of such poor quality or on the Cole pressure-% recovery plot.) With the transient or infi-
limited extent that a unique value of b cannot be determined. Re- nite-acting production period fixed, it requires a larger value of b,
liance on a statistical analysis, void of reservoir engineering con- or better recovery efficiency mechanism, to get a larger area under
cepts, to determine the decline exponent often leads to unrealistic or the rate/time curve to achieve higher fraction recoveries. Before the
illogical values of b and unreliable or indefensible forecasts and re- outermost boundary is encountered, the transient period would be
serves estimates. the same regardless of what later drive mechanism is established
From basic reservoir engineering principles, several of the ex- once a depletion process begins.
pected values of b have been derived for different reservoir drive or
recovery mechanisms. These values should be used when produc- Relationship Between Backpressure Curve Exponent n and De-
tion data is poor, insufficient, or totally lacking. They can also be cline-Curve Exponent b. Because gas and oil are sold at the well-
used to support or confirm clearly defined values of b determined head, we must consider the affect that the tubing string has on pro-
from good quality production data. duction for flowing gas and oil wells. It is the exponent, n, of the
Following is a tabulation of the values of b that should be ex- wellhead backpressure curve that affects the decline exponent, b.
pected for homogeneous single-layer or layered crossflow systems Examination of field performance backpressure curves7 indicates
(see Fig. 3). The range of the expected b values for these systems is that low-permeability gas wells yield bottomhole backpressure
from 0 to 0.5. curves with n values more nearly approaching 1.0 while high-
b+undeterminable. Constant-rate or increasing-rate production permeability gas wells yield n values approaching 0.5.
period. Flow rates are all in the transient or infinite-acting period The Forchheimer form of the backpressure equation is
with no supplemental engineering or geological information.
b+0, exponential. Single-phase liquid (highly undersaturated p 2R * p 2wf + Aq ) Bq 2, . . . . . . . . . . . . . . . . . . . . . . . . . . (27)
oil wells). High-pressure gas. Low-pressure gas with wellhead
backpressure curve exponent n[0.5 (tubing limited wells, both gas where A is a laminar flow pressure drop term and B is a turbulent
and oil flowing wells). Depletion or solution-gas-drive with unfa- flow pressure drop term. When kh is large, the Aq term becomes
vorable kg /ko . Poor waterflood performance.Wells with a high back- small and we have
pressure, pwf /pR³1. Gas wells undergoing liquid loading. Gravity
drainage with no free surface (derived value). q [ ǒ1ń ǸBǓ (p 2R * p 2wf ) 0.5. . . . . . . . . . . . . . . . . . . . . . . . (28)
b+0.3.Typical for solution-gas-drive (a derived value).
b+0.4 to 0.5.Typical for gas wells (derived values). b+0.5 for Similarly, when kh is small Aq becomes large with the Bq2 term
pwf [0; b+0.4 for pwf +0.1 pRi . becoming negligible when compared with the laminar pressure drop
b = 0.5. Gravity drainage with a free surface (derived value). Full term. We would then have
waterdrive in oil reservoirs (field data observations).
An unpublished study conducted several years ago on West Texas q [ ǒ1ńAǓ(p 2R * p 2wf ) 1.0 . . . . . . . . . . . . . . . . . . . . . . . . . . (29)
fields being waterflooded found values of b ranging from exponen-
Expressing the flow equation at a surface or wellhead datum and in-
tial, b+0, to b+0.9 or nearly harmonic. The decline exponent, b,
cluding a pressure drop term for the tubing or flow string, we have
was essentially the same for different leases within a given “field”
but different fields had different values of b. p 2c * p 2t + ǒA WHǓq ) (B WH ) T WH) q 2, . . . . . . . . . . . . . (30)
Fig. 2 is an attempt to illustrate the concept that b is a reflection
of recovery efficiency or drive mechanism. The pressure-percent re- where TWH represents a friction pressure drop term for flow through
covery figure, after Cole,12 depicts typical values of percentage re- the tubing and is indistinguishable from the reservoir turbulent pres-
coveries (recovery efficiency) for various reservoir drive mecha- sure drop component (BWH ) at the surface datum. Eq. 30 is more

16 SPE Reservoir Engineering, February 1996

 Fig. 4—Well performance plots of material-balance, backpressure curve, and rate/time semilog and log-log graphs.

useful for performance monitoring when expressed as the wellhead CTWH that reproduces the actual results obtained (i.e., a 50% in-
backpressure curve crease in production).
When a well is tubing limited, n³0.5, it should be recognized that
q [ C WH (p 2c * p 2t ) n , . . . . . . . . . . . . . . . . . . . . . . . . . . . (31) a significant deterioration in the bottomhole performance curve will
not be reflected in any change in the wellhead backpressure perfor-
where the range of n is between 0.5 and 1.
mance curve. Conversely, any improvement in the bottomhole per-
It should be recognized that a storm choke, flowline, and other sur-
formance curve, such as a stimulation or restimulation, will not re-
face equipment can be included in Eqs. 30 and 31, even further driv- sult in any increase in surface production; i.e., there will be no
ing the value of n to 0.5 in high-capacity flowing gas and oil wells. change in the wellhead deliverability curve and no increase in pro-
Note that as a limiting condition in Eq. 30 if TWH is large compared duction. Clearly, one will not be able to calculate reservoir variables
with AWH and BWH , a very large bottomhole deliverability well with from rate/time analysis on tubing limited gas or oil wells since most
a relatively small diameter tubing string, Eq. 30 reduces to of the pressure drop in the well is tubing friction pressure drop.
q [ C TWH (p 2c * p 2t ) 0.5 . . . . . . . . . . . . . . . . . . . . . . . . . . . (32)
Backpressure Effects on b. The level of backpressure, pwf , does not
When the exponent (n) of the wellhead backpressure curve ap- affect b for single-phase liquid flow, it’s always exponential—i.e.,
proaches 0.5, the well is considered to be “tubing limited” and will b+0. The effect of backpressure on a gas well is demonstrated for
always exhibit exponential rate/time decline. Once tubing limited a backpressure curve exponent n+1 in Fig. 9 given in Ref. 3. Back-
“always tubing limited”—the backpressure performance curve pressure is expressed as a ratio of pwf / pRi where pwf is the constant
does not shift to the right as reservoir pressure, pR declines. The pressure against which the rate declines with time and can also be
wellhead backpressure curve based on 25)years of performance considered as an abandonment pressure, pwfa . The rate/time type
data presented in Fig. 4 has an exponent n+0.55. The effect of a tub- curve for a backpressure ratio, pwf / pRi +0 has a decline exponent b
ing changeout, 27/8 to 3½ in., is shown both on the backpressure value of 0.5 and would yield a recovery factor [1*pwf / pRi ] of
curve and the semilog production plot. A 50% increase in produc- 100%. At a backpressure ratio of pwf / pRi +0.9, where pwf³ pRi the
rate/time type curve exhibits exponential decline, b+0, and would
tion was achieved by simply increasing the tubing diameter. Yet the
yield a recovery factor [1*pwf / pRi ] of only 10%. Backpressure ra-
well is still “tubing limited”. (Well depth and casing programs often
tios shown on the figure yielding 20% and 50% recoveries will not
dictate a maximum tubing size.) Note that the log-log production
trace any of the Arps type curve stems over its entirety—they cut
plot initialized after the tubing changeout to a larger diameter fits a across several of the b stems. From a practical standpoint, no opera-
decline exponent, b = 0.1, stem of the type curve. The decline expo- tor would produce a well that is on decline at such high backpressure
nent is exactly what would be predicted by using Eq. 17 and simply to abandonment. A more realistic ultimate backpressure ratio limit,
knowing that the backpressure curve exponent is 0.55. pwf / pRi , for volumetric gas wells in particular, would be 0.1. This
This example illustrates that the wellhead deliverability curve would yield a maximum recovery factor of 90% and result in a de-
and exponent n could, in some instances, be totally described by the cline exponent b+0.4. The b value of 0.4 should be considered as
pressure drop through the tubing string. We could use Eq. 32 to es- a good limiting value for gas wells when not clearly defined by actu-
tablish for a given tubing size a maximum position for a wellhead al production data.
curve, its wellhead potential qimax or qi , exponent n[0.5 and it will
exhibit exponential decline, b+0. For any other diameter flow Some Basic Concepts. Fig. 1 depicts in a graphical form the
string, one need only ratio (D2.612new/D2.612present) CTWH to (pR *pwf ), pseudosteady-state rate equation, and the pR *Np , mate-
draw in its new wellhead deliverability curve. Using 3.476 and rial-balance equation, from which the exponential decline equation
2.992 in. ID for the tubing sizes in Fig. 4, we get a curve shift 1.5 was derived. The drawdown backpressure curve is also a depletion

SPE Reservoir Engineering, February 1996 17

 TABLE 3—DECLINE EXPONENT b AS A FUNCTION OF ǒ q ińG Ǔ or ǒq ińN puoǓ AND VOLUME RATIO, V R
R R

Single-Phase Gas Single-Phase Liquid

VR or GR VR or (Npuo )R
(qi /G)R 2/1 1/1 1/2 1/5 (qi /Npuo )R 2/1 1/1 1/2 1/5
1 .4 .4 .4 .4 1 0 0 0 0
3 .45 .5 .5 .5 3 .1 .2 .25 .2
5 .5 .55 .65 .65 5 .2 .3 .4 .45
10 .6 .65 .7 .8 10 .3 .4 .5 .65
15.5 .6 .7 .75 .8 15.5 .35 .45 .6 .7
qtotal+q1)q2; individual layer b+0.4 for gas and b+0 for single-phase liquid
q iT qi 1 qi 2
Gas : q total + + )
NJ1 ) bTƪ1ńǒ1 * bTǓƫǒqiTńGTǓtNj
1ń0.4 1ń0.4
NJ1 ) 0.4ƪ1ńǒ1 * 0.4Ǔƫǒqi1ńG 1ǓtNj NJ1 ) 0.4ƪ1ńǒ1 * 0.4Ǔƫǒqi2ńG2ǓtNj
1ńb T

qi T qi 1 qi 2
Liquid : q total + + )
NJ1 ) b ƪ1ńǒ1 * b Ǔƫǒq Ǔ tNj
1ńb T
ǒqi 1ńNpuo1Ǔt ǒqi 2ńNpuo2Ǔt
i TńN puoT
T T
e e

curve in that all flow rates q(t), including qi , will trace down the is of considerable significance in long-term forecasting and reserve
curve as the reservoir pressure, pR , declines as a result of production. estimates. If crossflow exists in a layered reservoir, adjacent layers
The backpressure curve is fixed, Jo is constant, for all stages of pres- can simply be combined into a single equivalent layer using the av-
sure depletion providing there are no skin changes or the drainage erage reservoir properties of the crossflowing layers. It will then
radius, re , does not change. In equation form, the pseudosteady-state perform as an equivalent homogeneous single-layer system. For a
backpressure equation is homogeneous single-layer system, the maximum value of b is 0.5.
kh ( p R(t) * p wf) Decline-curve exponents, b, ranging between 0.5 and 1 are a pre-
q(t) + , . . . . . . . . . . . . . . . . (33) dicted response for layered, no-crossflow reservoirs10 and can
141.2 mB ƪln ǒr eńr wǓ * 34 ) sƫ therefore be used to identify them. We have found that layered,
no-crossflow reservoirs have the greatest potential for increasing
or in its simplest form current production and recoverable reserves.
q(t) + J o ƪp R(t) * p wfƫ 1.0. . . . . . . . . . . . . . . . . . . . . . . . . . . (34)
Low-permeability, stimulated wells’ production performance
can appear similar to layered, no-crossflow reservoir responses on
For a total field or lease backpressure curve, we would have a semilog production curve. However, a log-log type curve plot can
be used to distinguish between the two. Further confirmation of
q T (t) + WJ o avg( p R * p wf), . . . . . . . . . . . . . . . . . . . . . . . . (35) no-crossflow can be made by measuring layer pressures and having
some idea of the well’s permeability level.
where W is the total number of wells and Joavg represents our aver-
For two or more layers producing against a common flowing
age well PI. Or we could sum the individual wells to obtain a total
pressure, pwf , the commingled production rate, qT, is simply the sum
field or lease composite backpressure curve.
of separate producing rates or forecasts from each of the individual

ȍJ layers. Producing at a constant wellbore pressure, production from

W
q T (t) + oj( p R * p wf). . . . . . . . . . . . . . . . . . . . . . . . . (36) each layer is independent of all the other layers present.
j+1
q T + q 1 ) q 2 ), . . . , ) q n . . . . . . . . . . . . . . . . . . . . . . . (38)
The material-balance equation, graphically depicted in Fig. 1, re-
lates the cumulative production, Np , to the reservoir pressure, pR , is For a two-layered system, or its equivalent (most multilayered
systems can be reduced to an equivalent of two layers by combining
p R + * ǒ p RińN puoǓN p ) p Ri . . . . . . . . . . . . . . . . . . . . . . . (37) layers with similar (q/G) or (q/Npuo ) values, we can write
A basic assumption again is that re does not change; i.e., the well (or qi 1 qi 2
qT + ) , . . . . . . . . . . (39)
lease) is neither draining nor being drained by offset wells. For a to-
tal field, the reservoir volume, or re , always remains fixed.
ǒ1 ) b1Di 1tǓ 1ńb1 ǒ1 ) b2D i 2 tǓ1ńb2
An important reservoir engineering concept useful for interpret-
ing decline performance and understanding and applying decline- or q T + ǒq i 1ńe D i 1tǓ ) ǒq i 2ńe Di 2tǓ, . . . . . . . . . . . . . . . . . . . . (40)
curve analysis is that each well in a common reservoir undergoing
pseudosteady–state depletion drains a volume in proportion to its qi 1 qi 2
or even q T + ) , . . . . . . . . . . . . . . . (41)
producing rate. From this we can point out a couple of important ǒ1 ) b1D i 1tǓ1ńb1 e Di 2 t
concepts that result.
1. The decline rate, Di or qi /Npuo , theoretically should be the same where Di is defined in Table 1 and Layer 1 normally represents the
for all wells in the common reservoir. Wells that have a noticeably higher-deliverability layer.
different Di are not in communication. The sum of two or more separate forecasts usually results in a
2. When a new well(s) is added (drilled) or an old well(s) is resti- forecast with the decline exponent, b, greater than that of each indi-
mulated in a common reservoir, the decline rate, Di , will increase for vidual layer. We have, for a commingled system using single-phase
each well. The depletion of the total field will be essentially acceler-
gas variables, for example
ated. All other wells will lose some of their remaining reserves.
q iT
qT + , . . . . . . . . . . . . . (42)
Layered, NoĆCrossflow Decline Behavior NJ1 ) ƪ b Tń(1 * b T) ƫǒq iTńG TǓ t Nj
1ńb T

Most reservoirs consist of several layers with reservoir properties

varying between layers. The fact that no-crossflow reservoirs are where qiT is the sum of each layer’s flow rates, qi , and GT is the total
perhaps the most prevalent and important, reservoir heterogeneity ultimate recoverable gas in each layer. The exponent bT is the single

18 SPE Reservoir Engineering, February 1996

 TABLE 4—LAYER SKIN VALUES diversion, the stimulation seems to preferentially occur in the higher
permeability layer(s). This preferential stimulation of the high-perme-
s1 s2 ability layer(s) will most likely occur in openhole completions that
*7 0 have been acidized commingled. By restimulating wells with the ob-
0 )72 jective of targeting the low-permeability layers, substantial increases in
*7.6 *4 production and recoverable reserves can be achieved. This potential
*2.2 )50 can normally be identified through high decline exponents, b. Note
from Table 3 that the higher the b value, the more volume in the low-
permeability layer and therefore the more potential for increased pro-
value of b that results in the best match, or fit, of the commingled duction and recoverable reserves. With a single-layered system exhibit-
production rates, excluding late-time commingled production rates ing a low b value, restimulation generally results in accelerated
when necessary. Very-late-time production rates will ultimately re- production. With a layered, no-crossflow system exhibiting a high b
sult in the b value of Layer 2 producing by itself. Use of the early value, restimulation can result in substantial increased recoverable re-
to middle time production data permits reasonable calculated values serves, depending on the volume of the low-permeability layer(s). The
of qiT and GT from actual production performance data. (hydraulically) fractured well example given in Ref. 3 had a b of be-
Table 3 contains predicted values of the decline exponent, b, for a tween 0.6 and 1 to fit both the before and after stimulation data. The
two-layer system as a function of (qi /G) ratio vs. volume ratio, VR or rather successful treatment reflected in both the increase in production
GR for single-phase gas and (qi /Npuo ) ratio vs. volume ratio, VR or and additional recoverable reserves is an example of what one could
[Npuo ]R , for single-phase liquid. The high-deliverability layer is nor- expect from a proper restimulation of a layered, no-crossflow well.
mally the numerator in both ratios. The qi /G or qi /Npuo ratio can be de- Following is a list of some of the more important characteristics
scribed as a rate-of-take or rate-of-depletion indicator. The layer that of layered, no-crossflow behavior:
has the higher qi /G or qi /Npuo value is depleting its reserves faster. It is 1. High value of the decline exponent, b; bu0.5. This is reflected
important to recognize that the qi for the high-deliverability layer is not as an early rapid decline in rate followed by an extended period of
just a reflection of layer permeability, k, but also of layer skin, s, and a low percentage decline.
layer shut-in pressure, pR . With regard to layer volume, V or (G or 2. Reservoir has an indicated unusually long producing life.
Npuo), the PV ratio, VR , will essentially remain constant for all produc- 3. Unusually rapid decline in reservoir shut-in pressure. Later, a
ing times, whereas the gas-in-place ratio, GR or [Npuo ]R , changes with nearly constant measured shut-in pressure with increasing cumula-
time due to differential pressure depletion. This becomes significant tive production. [This is because the pressures usually represent the
when we consider a constant rate production period followed by a de- most permeable layer, but production is coming mainly from the
clining production rate period; i.e., the volume ratio, GR , will be differ- low-permeability layer(s).]
ent at the beginning and end of the constant-rate period. 4. Large discrepancy between pressure performance oil or gas-
From Table 3, we see that the highest b values listed are for (qi/G in-place and volumetrics. (Indicated low recovery factor.)
or qi /Npuo ) ratios between 10 and 15.5 (see Ref. 10 for the results of 5. Thick reservoirs have a very high likelihood of exhibiting lay-
larger ratios). Further, for any given (qi /G or qi /Npuo ) ratio, b in- ered, no-crossflow behavior. (There appears to be a strong correla-
creases as the volume ratio decreases. In either case, this occurs be- tion of increasing b with thickness.)
cause the reserves in the high-deliverability layer are being depleted 6. Thin reservoirs don’t automatically mean they will behave as
faster with respect to the low-deliverability layer. As a result, deple- a single layer. (We have measured differential depletion in a 9-ft
tion of the high-deliverability layer occurs over a shorter time inter- sand.)
val with respect to the low-deliverability layer. The composite or 7. Naturally fractured reservoirs permit a contrast in (qi /G)R re-
combined surface flow rate will have b values that are driven further quired for layered, no-crossflow behavior (i.e., high b values as seen
towards a value of b equal to 1 as either the qi /G or qi /Npuo ratio in- in the Monterey formation).
creases, the VR decreases, or both. The (qi /G)R ratio is written as 8. Layer pressures indicating differential depletion.
9. Production logs indicating wellbore backflow during shut-ins
(q i ńG) R + ǒq i1 ńG 1Ǔńǒq i2 ńG 2Ǔ, . . . . . . . . . . . . . . . . . . . . . . (43) or rate reductions. Lack of backflow does not mean lack of layered,
where Layer 1 is generally the higher-deliverability layer. Assum- no-crossflow behavior.
ing, for simplicity, that pi1+pi2 and (mct ) are equal for both layers The reader is referred to Ref. 10 for additional material on lay-
and substituting reservoir variables, we have ered, no-crossflow reservoir behavior.
We recognize the similarity of rate/time and pressure cumulative
k 1 ńf 1 ƪ lnǒ0.472 r e2ńr wǓ ) s 2ƫ r 2e2 production and a few of the other listed characteristics are similar
(q i ńG) R + . . . . . . . . . . (44) to that of a composite reservoir model; however, we find it to be an
k 2 ńf 2 ƪ lnǒ0.472 r e1ńr wǓ ) s 1ƫ r 2e1 unrealistic model on an individual well basis. It is a possible model
With a (qi /G) ratio of 10, one normally assumes that this would on a fault-block or a field-wide basis. The question of which it may
represent a layered, no-crossflow system with a 10 to 1 contrast in be is readily resolved by the taking of layer pressures and comparing
permeability. From Eq. 44, this would be true if f1+f2, re1+re2, pressure data on an areal basis.
and s1+s2. Note that thickness, h, has cancelled from the equation.
However, if layer k, f, and re are all equal, we would still have a Production Data Treatments
(qi /G) ratio of 10 for the following skin conditions shown in Table Production data from a commercial database is usually available on
4 on each layer. a monthly production basis, with the production sometimes being
Some small contrast in permeability would lead to even more rea- erratic and difficult to analyze. We have found that 6-month or
sonable values of skin on each layer to still maintain the 10 ratio and 12-month averaging, with the average rate qavg (t) plotted at the mid-
the resulting high decline exponents, b, reflected in the table. For all point of the time period, works very satisfactorily (see Ref. 4, Figs.
reservoir variables of each layer being equal, including skins, s, an 20 and 21 for examples). Material is neither created nor destroyed
re1+630 ft and re2+2,000 ft would also result in a ratio [10. One this way, as can be done with purely mathematical smoothing ap-
need not have a high contrast in permeability to have significant lay- proaches. Further, this averaging is nearly equivalent to cumulative
ered, no-crossflow reservoir production performance behavior that production/time plots that are often made to make the same erratic
exhibits high values of the decline exponent, b. rate/time data “look” better. A derivative at the midpoint of the year
In a layered, no-crossflow system, the maximum recoverable re- would result then in rate/time data for a rate/time plot. It is much eas-
serves from the low-deliverability layer will occur as (qi /G)R ap- ier to relate rate/time data to one’s experience, particularly transient
proaches 1. We, therefore, must have as large a negative skin, s2, on the behavior, than it is cumulative production time plots.
low-permeability layer(s) as possible and as early in the production life Transient production data, however, result in the opposite recom-
as possible. On commingled wellbore systems, even with attempts at mendation. Average monthly production on low-permeability,

SPE Reservoir Engineering, February 1996 19

highly stimulated wells or fields will not reveal some useful data ponent, b, and Npuo , would change and the value of b will be greater
(see Ref. 4, Figs. 32 and 33). We have had extremely good success than exponential decline depending on the kg /ko relationship. Pres-
using daily production from meter charts on individual wells in the sure-cumulative production and gas/oil-ratio performance often
Carthage Cotton Valley field. help indicate when the reinitialization should be done. Some other
conditions where the production data might be reinitialized are:
Normalization. Another important technique we use in rate/time 1. Primary, secondary, and tertiary recovery periods would all be
analysis for low-permeability stimulated wells is rate normaliza- reinitialized. The decline exponent, b, however should be similar for
tion—i.e., q(t)/[pi *pwf (t)]. Often, during transient production, both each period.
the rate q(t) and flowing pressure, pwf (t) are both declining smoothly 2.ĂAn abrupt change in the number of wells on a lease or in a field.
and monotonically. Normalizing the rate by dividing by [pi *pwf (t)] An infill well on a one-well section, for example, will result in a
is identical to rigorous superposition and will permit the normalized change in qiT, (qi1)qi2)/G compared with qi1/G before the infill well.
data to overlay the qD *tD type curves. Extrapolation of the fit into 3. Restimulation of an original well would be similar, qi new/G, as
the future by tracing along the type curve results in a rate/time fore- compared to qiold/G.
cast as if the well were being produced at the last value of pwf . Pro- 4. Also similar would be a tubing changeout to a larger diameter
duction rates at a different backpressure, pwf , would then require a where the well was tubing limited. Not only would qi change after-
superposition calculation. wards, but the decline exponent, b, could also change due to the in-
Abrupt changes in flowing pressure, pwf , normally a reduction in crease in wellhead backpressure exponent, n (see Fig. 4 and Eq. 17).
pwf to increase production and recoverable reserves, is adequately Finally, anytime during the depletion period, one can reinitialize
discussed in Ref. 3 (Table 2 and Fig. 14), Ref. 5 (Tables 9 and 9A the problem and get the same resulting forecast. However, there be-
and Fig. 9), and Ref. 13 (Pages 438–41). The principle of superposi- comes a point where there is not sufficient rate/time data to uniquely
tion also applies to an increase in pwf2 as would be applicable in at- match or define the decline exponent, b. Using a different value of
tempting to handle the gas bubble curtailment period of the b would result in a different forecast and reserve estimate. One of
mid-1980’s. See Figs. 34 and 35 of Ref. 4 for the negligible effect the problems with reinitializing is obtaining a new pR to use in at-
of numerous field shut-ins on the overall total field production. tempting to calculate reservoir variables.
One additional normalization procedure that we have used is well
count normalization when the well count is increasing smoothly and Stimulation and Restimulation (Depletion). The effects of a stim-
monotonically as a result of continuous drilling in a field or lease. One ulation or a restimulation, providing a well is not tubing or equip-
of the more noticeable effects we have seen was in the Sooner Trend ment limited, result in a change in deliverability, qi , and possibly the
field between 1970 and 1982. The total field production data matched remaining recoverable gas or oil, G or Npuo . The decline exponent,
a decline exponent b+1 during this period of continuous drilling. b, normally can be assumed to remain the same.
Normalized by well count to an average well basis, the decline expo- For an individual layer, a convenient “rule-of-thumb” equation13 to
nent was found to be b+0.23 instead of b+1 unnormalized. approximate an increase in rate due to a stimulation or restimulation is
Reinitialization of Data. There are several obvious situations q i new + q(t) old ƪǒ7 ) s oldǓń(7 ) s new)ƫ, . . . . . . . . . . . . . . . (45)
where rate/time data must be reinitialized because the drive or pro-
duction mechanism has changed. The most obvious situation is pro- where q(t)old is the producing rate just prior to stimulation, and qinew
duction above and below the bubblepoint pressure. Above the bub- is the new qi for the reinitialized forecast. We would then have
blepoint production will follow exponential decline, b+0. Further,
the calculation of reservoir variables, kh, s, and original oil-in-place, q(t) new + q i newńǒ1 ) b D i newtǓ 1ńb, . . . . . . . . . . . . . . . . . (46)
N, will be most accurate when calculated during the undersaturated
period. Below the bubblepoint pressure, the values of the decline ex- where b is assumed to be the same before and after stimulation and

Fig. 5—Composite of analytical and empirical type curves with comments.

20 SPE Reservoir Engineering, February 1996

 D i new + 1 ǒq i new
(1 * b) G or N puo remaining
Ǔ
. . . . . . . . . . . . . . (47)
where kh
ƪlnǒ0.473r eńr waǓƫ
+
141.2mB
ǒp R * p wfǓ
q i . . . . . . . . . . . . . . . . (54)

By assuming some value of skin, or rwa , we can calculate a value of kh.

G or Npuo , if assumed to be the same as that obtained prior to stimula-
Calculate the pseudosteady-state PI for the rate equation
tion, would be simply acceleration, which would be the case if all off-
J + q ińƪp R * p wfƫ. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (55)
sets were stimulated or restimulated at the same time. If offsets were
not restimulated, then the remaining reserves would increase in direct
proportion to the increase in producing rate as a result of a radius of Using the hyperbolic expression for Di from Table 1, we can then
drainage readjustment; i.e., drainage of offset wells would occur. calculate the recoverable reserves
For layered, no-crossflow systems, we would assume that the
low-deliverability layer was not stimulated originally and calculate G or N puo + ƪ1ń(1 * b)ƫ ǒq ińD iǓ . . . . . . . . . . . . . . . . . . (56)
its increase due to restimulation to qi2 with the high-deliverability
layer, qi (t), remaining the same. In this case, the increase in reserves Then, gas or oil-in-place at the start of decline is
would be a result of reducing the amount of differential depletion G i or N + ǒG or N puoǓńRF . . . . . . . . . . . . . . . . . . . . . . . (57)
between layers.

Calculating Reservoir Variables Conclusions

There are two periods of rate decline, transient decline followed by Rate/time analysis (decline-curve analysis) has been placed on a sound,
depletion decline. These two periods, and their production response fundamental basis using the constant wellbore pressure analytical solu-
are illustrated on the log-log type curve of Fig. 5. tions. Derivations combining the material-balance and pseudosteady-
state rate equations result in rate/time decline equations for oil and gas
Transient Decline. Production data existing during the transient de- wells. It is from these derivations that the variables in Arps equations
cline period can be matched to a constant wellbore pressure type can be expressed in terms of reservoir variables. Using these solutions
curve to calculate both kh and skin, s, from the following equations with sound reservoir engineering concepts, it becomes possible to pro-
vide more accurate production forecasts and better oil and gas reserves
kh + 141.2mBNJ[qń(p i * p wf )]ńq DNj . . . . . . . . . . . . . . (48) for both wells and/or individual fields.
match

and r wa + Ǹƪ0.00634 khńǒmc tǓfhƫǒtńt DǓ match , . . . . . . . . . . . (49) Nomenclature

b +decline exponent
where r wa + r w e *s , . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (50) Cg +gas-well PI (backpressure curve coefficient)
or in terms of skin Mscf/D/(psi)2n
Di +initial decline rate, t*1
s + * lnǒr wańr wǓ . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (51) G +gas-in-place, surface-measured
Gp +cumulative gas produced, surface-measured
Eq. 48 is written to emphasize the need to normalize the rates q J o +PI, STB/D/psi
by using flowing pressure, pwf , when necessary. To calculate kh and J Ȁo +PI, (backpressure curve coefficient) STB/D/(psi)2n
s from transient data, we also prefer to use the more general radial n +exponent of backpressure curve
flow constant wellbore pressure solution type curve in terms of Np +cumulative production, STB
qD *tD , Fig. 2-A of Ref. 3, with the effective wellbore radius, rwa .
Npuo +ultimate recoverable oil at q(t)+0, STB
This type curve covers a much broader range of qD *tD (rwa ) that ap-
Npuo,o +ultimate recoverable oil at q(t)+0 and pwf +0, STB
plies to most of the very low-permeability, successfully stimulated
N +oil-in-place, STB
wells. Wells with large, positive skins of any permeability level lose
pc +wellhead shut-in pressure, psia
uniqueness in that the rate/time data would be in the region where
pR +average reservoir pressure, psia
q(t) appears to be nearly constant because of large tD (rwa ) values.
pt +flowing wellhead tubing pressure, psia
Eq. 49 is written in terms of kh and fh to emphasize the point that
pwf +bottomhole flowing pressure, psia
when we pick h we are assigning an oil or gas-in-place—not simply
qi +initial surface rate of flow from stabilized curve
determining k. The significance is now apparent in that we are mak-
qi(max) +initial surface rate of flow from stabilized curve at
ing production forecasts. It’s one thing to reduce the k by a half by
pwf or pt +0
assigning a range of thickness, h, and quite different to cut the oil or
q(t) +surface rate of flow at time t
gas-in-place by half. Generally, people who do pressure transient
qD +dimensionless rate
analysis seldom apply the results to production forecasting. See
Refs. 3 through 5 for some good field examples. qDd +decline-curve dimensionless rate
Qp +oil or gas cumulative production, Np or Gp , surface
Depletion Decline. Rapidly declining transient production data in measure
tabular form, or even on a semilog plot, often is misinterpreted to be rwa +effective wellbore radius, ft.
depletion. The primary use of the composite type curve (Fig. 5) is tD +dimensionless time
to be able to distinguish between transient and depletion production tDd +decline-curve dimensionless time
decline by making a log-log plot of the production data, matching VR +layer volume ratio, PVs
it and then forecasting future production by extending the match Vp +reservoir PV, ft3 or BBl
down the appropriate depletion stem. If production data exist only
in depletion, we can calculate oil or gas-in-place and a combined kh Subscripts
and rwa or skin and PI, J. a +at abandonment
In its simplest form, using the Arps equation and match points g +gas
from a type curve match on the depletion stem, b, calculate i +initial
o +oil
q i + ƪq(t)ńq Ddƫ match . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (52) R +ratio
T +total of all layers’ productions
and D i + ǒ t Ddńt Ǔ match, . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (53) u +ultimate

SPE Reservoir Engineering, February 1996 21

Acknowledgments 14. Fetkovich, M.J., Fetkovich, E.J., and Fetkovich, M.D.: “Useful Con-
cepts for Decline Curve Forecasting, Reserve Estimation, and Analy-
We wish to thank Phillips Petroleum Co. for permission to publish
sis,” paper SPE 28628 presented at the 69th Annual Technical Confer-
this paper. Special thanks to Kay Patton for the excellent typing of ence and Exhibition, New Orleans, LA (Sept. 25–28, 1994).
this paper and the several drafts that preceded it.
SI Metric Conversion Factors
References
ft 3.048* E*01 +m
1. Arps, J.J.: “Analysis of Decline Curves,” Trans., AIME (1945) 160, 228–47.
2. Fetkovich, M.J.: “A Simplified Approach to Water Influx Calcula- in. 2.54* E)00 +cm
tions—Finite Aquifer Systems,” JPT (July 1971) 814–828. *Conversion factor is exact. SPERE
3. Fetkovich, M.J.: “Decline Curve Analysis Using Type Curves,” JPT
(June 1980) 1065–77.
4. Fetkovich, M.J. et al.: “Decline Curve Analysis Using Type Cur- Michael J. Fetkovich is a Phillips Fellow Emeritus in the Reservoir
ves*Case Histories,” SPEFE (Dec. 1987) 637–656. Engineering Branch, Research & Services Div. of R&D, Phillips PeĆ
5. Fetkovich, M.J. et al.: “Case Study of a Low-Permeability Volatile Oil troleum Co. in Bartlesville, OK. He holds a BS degree in petroleum
Field Using Individual-Well Advanced Decline Curve Analysis,” paper and natural gas engineering from the U. of Pittsburgh and a
SPE 14237 presented at the SPE 60th Annual Technical Conference and D.Ing. degree from the Norwegian Inst. of Technology. Fetkovich
Exhibition, Las Vegas, NV (Sept. 22–25, 1985). received the 1993 Lester C. Uren Award and the 1989 Reservoir
6. Fetkovich, M.J.: “The Isochronal Testing of Oil Wells,” paper SPE 4529 Engineering Award. He is a Distinguished Member and was a
presented at the 48th Annual Fall Meeting, Las Vegas, NV (Sept. 1977ć78 Distinguished Lecturer. Eddie J. Fetkovich recently transĆ
30–Oct. 3, 1973). ferred from Phillips' Research & Services Div. in Bartlesville to
7. Fetkovich, M.J.: “Multipoint Testing of Gas Wells,” SPE Mid-Conti- Odessa, TX, to be the asset team reservoir engineer for PhillipsĆopĆ
nent Section, Continuing Education Course, Tulsa, (March 17, 1975). erated New Mexico properties. He holds a BS degree in mechaniĆ
8. Fetkovich, M.D. et al.: “Oil and Gas Relative Permeabilities Determined cal engineering with a petroleum option from Oklahoma State U.
From rate/time Performance Data,” paper SPE 15431 presented at the 61st in Stillwater. He was the 1994ć95 publicity chairman for the BartlesĆ
Annual Technical Conference and Exhibition of SPE, New Orleans, LA ville Section. Michael D. Fetkovich currently is a reservoir engineer
(Oct. 5–8, 1986). for Phillips Petroleum working in the western Gulf of Mexico. He
9. Fetkovich, M.J.: “Advances in Well Deliverability and Production Fore- holds a BS degree in chemical engineering from Oklahoma State
casting,” PhD dissertation, U. of Trondheim (1988). U., Stillwater, and an MS degree in petroleum engineering from
10. Fetkovich, M.J. et al.: “Depletion Performance of Layered Reservoirs the U. of Tulsa.
Without Crossflow,” SPEFE (Sept. 1990).
11. Fetkovich, M.J., Needham, R.B., and McCoy, T.F.: “Analysis of Kansas Hu-
goton Infill Drilling, Part II: Twelve Year Performance History of Five Re-
placement Wells,” paper SPE 20779 presented at the 65th SPE Annual Tech-
nical Conference and Exhibition, New Orleans, LA (Sept. 23–26, 1990).
12. Cole, F.W.: Reservoir Engineering Manual, Gulf Publishing Co., Hous-
ton, TX (1969) 48.
13. Golan, M. and Whitson, C.H.: Well Performance, Prentice Hall, Engle-
wood Cliffs, NJ, Second Edition (1991), 669. M.J. Fetkovich E.J. Fetkovich M.D. Fetkovich

22 SPE Reservoir Engineering, February 1996