Sei sulla pagina 1di 15

The Effect of Reservoir and Fluid Properties on

Production Decline Curves


R. W. Gentry, SPE-AIME, Edinger, Inc.
A. W. McCray, SPE-AIME, U. of Oklahoma
Introduction
The extrapolation of production decline curves to predict
future oil production has a long history. The word "de-
cline" is a misnomer; "decline curve" is a descriptive
term for a graphical presentation of some aspect of the
performance history of a well or lease. The graph may
show a decline, an incline, or may remain flat. Through
the years, the most commonly used decline curve, and the
curve usually maintained by oil companies, is the produc-
tion rate vs time curve plotted on semilog graph paper.
The analysis of a decline curve provides two important
items of information: (1) the remaining oil and gas re-
serves to be expected, and (2) the remaining productive
life of a well or lease. In addition, an explanation of any
anomalies that appear on the graph is also useful. In
recent years, it has been increasingly important to deter-
mine the combination of these items so that an accurate
yearly estimate of future production can be made. This
information is obtained by an analysis of the past perfor-
mance as shown on the production rate vs time curve. The
curve then is extrapolated into the future, and estimated
future yearly production is taken from the extrapolated
portion of the curve.
Decline Curve Analysis Problems
Three basic problems associated with the extrapolation
process are connected with the historical development of
decline curve analysis and have been considered basic
assumptions since being defined by ArpS.l,2 These as-
01492136/78/0009-6341 $00.25
1978 Society of Petroleum Engineers of AIME
sumptions are as follow.
1. The extrapolation procedure is strictly empirical,
and a mathematical expression of the curve based on
physical considerations can be set up only for a few
simple cases.
2. Whatever causes governed the trend of a curve in
the past will continue to govern its trend in the future in a
uniform manner.
3. The decline exponent b in the equations developed
by Arps (Table 1) must have a value of 0.0 b 1.0.
Because of empirical extrapolation, a decline curve
usually will have a wide range of interpretations. The
range of interpretations depends on the production stage
of the property. If there is limited prior production history
(i. e., a new well), there is a wider range of interpretations
possible than for a well or property in the stripper stage of
production. Also, each specific interpretation is a func-
tion of the experience, integrity, and objective of the
evaluating engineer.
Various controllable and uncontrollable influences or
causes govern the production performance of a well or
lease. Some of these influences are as follow.
Controllable
1. Prorated production.
2. Remedial work on producing wells.
3. Fluid or gas injection into the producing reservoir.
4. Production problems, shutdowns, etc.
5. Problems with scale, paraffin, etc.
6. Limitations of producing equipment.
7. Changes in operating personnel.
This paper investigates the effects that rock andfluid properties impose on the production
history of a well or lease as revealed by a decline curve. In particular, the effects on the
constants of the exponential, hyperbolic, and harmonic decline curve equations are reported.
An improved method is presented for analyzing production histories, with two field examples
illustrating the method.
SEPTEMBER, 1978 1327
Uncontrollable
1. Physical characteristics of the reservoir.
2. Characteristics of the reservoir fluids.
3. Primary reservoir drive mechanisms.
Controllable influences are caused by man when produc-
ing oil from the property. Uncontrollable influences are
those controlled by nature. Any single or combination of
these influences can change at any time during producing
life. If controllable influences are eliminated from the
analysis, the resulting curve should reflect the combined
effect of the three uncontrollable influences.
In prior years, the extrapolation procedure was empiri-
cal, but with modern technology there is no reason for it
to remain so. Also, a critical examination of the causes or
influences that govern decline behavior show that it is not
reasonable to expect those causes that governed in the
past to continue to do so uniformly in the future.
Investigators
3
,4 also have shown that there are many
instances in nature where the decline exponent b is
greater than 1. O.
Although this discussion of problems tends to discour-
age using decline durves, the extrapolation of the rate-
time curve is still the most used and efficient method for
evaluating oil properties.
Historical Development
In 1908, Arnold and Anderson
5
published the first expla-
nation and mathematical development o{the exponential
decline. Note that this is still the most popular method of
decline curve analysis used by engineers. In 1924,
Cutler
6
proposed a method of extrapolating hyperbolic
declines by plotting the data on log-log paper and shifting
the curve to obtain a straight line. The basics of this
method are still being used today. In 1945, Arps pub-
lished a paperl that classified decline curves into four
general types: (1) exponential, (2) hyperbolic, (3) har-
monic, and (4) ratio declines. The first three are familiar
to engineers, but the ratio decline is seldom used. The
second Arps paper2 (1956) did not contain any new in-
formation concerning decline curves, but it did put the
mathematical formulas into a much simpler form. These
equations are shown in Cols. 1 through 4 of Table 1. In
1972, Gentry7 manipulated the decline equations to ob-
tain two dimensionless equations for each type of de-
cline. These equations are shown in Cols. 5 and 6 of
Table 1. These equations were used to prepare two graphs
that can be used to extrapolate hyperbolic and harmonic
decline curves quickly. SliderS proposed an improved
method of hyperbolic decline curve analysis using trans-
parent overlays.
Since 1945, when Ref. 1 was published, investigators
have been trying to associate the decline exponent b with
the physical characteristics' or the active drive mechanism
(uncontrollable influences) of the reservoir. Mead
9
pre-
sented a list of b values for various types of reservoir
drive mechanisms. Matthews and Lefkovits
10
, 11 investi-
gated gravity drainage reservoirs on theoretical and ac-
tual field performance bases. One of their conclusions
stated that the production declines for homogeneous
gravity drainage reservoirs where a free surface exists
will be of the hyperbolic type with b = 0.5. When tested
on field cases, several instances were founc:i where b
values were greater than 1.0. Matthews and Lefkovits
concluded that this happened when two or more layers of
different permeabilities were producing into a common
wellbore. They predicted that as the more permeable
layers were exhausted, the value of b would decline to an
ultimate value of 0.5. They also implied a physical sig-
nificance to the initial decline rate, ai'
The next logical step was to determine why production
curves exhibit certain values for the three decline con-
stants qi' ai' and b. Fetkovich
12
attempted to find possible
connections between decline constants and characteris-
tics of the reservoir and fluids by using the material
balance equation, rate equations, etc., to develop a set of
correlations. The effects of stratified reservoirs and
backpressure changes also were investigated.
Theoretical Development
The equations in Table 1 were derived from the differen-
tial equation,
dq
a = Kqb = - dt . . ...................... (1)
q
The general solution for the rate-time relationship is
q
- qj
- (1 + ba;t)lIb '
(b =to 1, b =,0) ....... (2)
This can be integrated to obtain the equation for the
TABLE 1-DECLINE CURVE EQUATIONS
Decline
Exponent
b=O
b = 1
1328
Type
of
Decline
Exponential
Hyperbolic
Harmonic
Rate-Time
Relationship
Rate-Cumulative
Relationship
N = qt
p (1 -b)a
i
N
P ai q
al
Relationship

, q
(-1i-r - 1
b
at =
, q
Np =
al
!!..Np
qi t
Relationship
1 - (.9i..qi) -I
In !Ii..
q

Np = __ q,,--_
a;t _ 1
JOURNAL OF PETROLEUM TECHNOLOGY
time-cumulative relationship,
b-l
Np = __ q_t _ [(1 + bqit ) b - 1]. ........ (3)
(b - l)a;
The rate-cumulative relationship may be derived by
substituting the rate-time equation into the time-
cumulative relationship to produce
b
N
p
= qi (q/--b - ql-b) . ............. (4)
(l - b)ai
Eq. 2 can be solved for the dimensionless quantity
aitby
(qiY - 1
ait = q b ........................ (5)
By solving Eq. 2 for ai and substituting into Eq. 4, the
. following dimensionless relationship can be obtained:
[1 _/qiq)b-I] . ............ (6)
qit 1 - b (qi q) - 1
Similar equations can be obtained for the special cases
of b = 0.0 and b = 1.0.
Although b values from 0.0 to 1.0 generally are used,
the above equations can be calculated using b values less
than 0.0 and greater than 1.0. In fact, there is a limiting
case for b = 00. In this case, the dimensionless Eq. 6
would become
.lim Ll Np = lim 1 - (qJq)H ] =
qit L (qJq)b - 1 1 - b qi
................................ (7)
The dimensionless values for b = 0.0, b = 0.5, and b
= 1.0 have been plotted as a background on several
figures in this paper, and in some instances, the dimen-
sionless value for b = 00 also has been included. By using
these as a baCkground and plotting computer-generated
or actual field data on the curve, it can readily be seen
how these data compare with the Arps equations.
An interesting result occurs if b is set equal to 0.5. In
this case,

qi t qt
(b = 0.5),
which can be further manipulated as
Np = (qiq)O.5t. . ........................ (8)
One computer simulator run (discussed later) produced a
curve that can be described by the equation
(L)O.4.
qit qi
This can be written for the general form as
= (L)a or Np = (qi I-a qa)t. . ......... (9)
qit qi
This equation may describe certain decline behavior bet-
ter than the Arps equations.
To solve the Arps equations for any particular decline
curve, three constants must be determined: (1) qi' the
initial production rate; (2) ai' the initial decline rate; and
(3) b, the decline exponent.
Current practice is to take the existing production his-
tory and determine the values of these constants. This is
SEPTEMBER, 1978
done by trial and error, using overlays, by mathematical
analysis, or by graphical methods. Once these constants
are determined, they are put into the equations to calcu-
late future production. However, if we could define some
of the effects of reservoir and fluid characteristics on
these constants, we could determine the constants and
calculate future production without prior production
history.
To find some of these relationships, fluid production
should be monitored from reservoirs with known physi-
cal quantities containing fluids of known compositions.
Production then could be plotted as a decline curve, and
the three constants determined for each particular reser-
voir and fluid.
Reservoir Simulation Model
The production of known fluids from a reservoir of
known characteristics was done with the aid of a com-
puter reservoir model. The model was constructed using
assumed reservoir conditions of volume, porosity, tem-
perature, pressure, water saturation, and absolute and
relative permeability relationships. The reservoir was
assumed to be filled with fluids of known characteristics
of formation volume factor, solution gas content, oil and
gas viscosities, and deviation factors for the reservoir
gas. The reservoir model was constructed assuming the
following limitations.
1. The reservoir is closed, horizontal, circular, and of
constant thickness. It is homogeneous with constant per-
meability and relative permeability characteristics.
2. The material balance method for predicting per-
formance histories of solution gas reservoirs permits a
reasonable approximation of reservoir pressure and GOR
as a function of cumulative oil production.
3. Ths solution of Darcy's radial flow equation for a
bounded reservoir adequately describes the flow of fluid
through the reservoir into the wellbore.
4. The reservoir driving mechanism is solution gas at
or below the saturation pressure of the oil.
5. There are no capillary pressure gradients or gravity
drainage effects.
6. Individual zones are separated by impermeable
strata so that pressure communication between zones
exists only at the well bore .
This computer model was used for calculations of all
simulated production histories. Reservoir engineering
equations for the material balance calculations were ob-
tained from Ref. 13. Two sets of relative permeability
relationships were obtained from Ref. 14 with some
slight modifications. One set typifies a consolidated sand
reservoir, and the other represents a dolomite reservoir.
Two sets of fluid data (oil samples) taken from actual
field data were used in the model. One oil sample was
taken from a Bartlesville sandstone reservoir in Ok-
lahoma County, OK, and the other was from a Hunton
dolomitic reservoir in Kingfisher County, OK.
Experimental Procedure
One relative permeability relationship and one fluid
analysis were loaded into the computer model. The pro-
gram then was run and the production histories were
printed for that particular fluid-permeability system for
absolute permeabilities of 1, 3, 5, 7, 9,11,13, and 15
md. The dimensionless values of ait, Np/qit, and qi/q
1329
were calculated and plotted on semilog paper with the
solutions of the Arps equations in the background. This
showed how b behaved in relation to the production ratio,
qi/q, and to the solutions to the Arps equations. Also, the.
results from two or more absolute permeabilities were
summed to simulate the effect of two or more zones
producing into the wellbore. These results were plotted to
observe the effects of stratification.
This procedure was conducted for four fluid-
permeability systems using all possible combinations of
two fluid analyses and two relative permeability relation-
ships. Additional computer runs were made to investigate
the effect of drainage area, zone thickness, and porosity
on the decline constants.
The initial computer run for each fluid-permeability
system used the following standard set of reservoir data.
Porosity, % 18
Zone thickness, ft 20
Radius of well bore, ft 0.333
Radius to external boundary,
ft (acres) 745 (40)
Reservoir water saturation, % 15
Initial reservoir Saturation pressure
pressure at reservoir
temperature
Fluid Sample 1, psia 2,520 (from sandstone)
Fluid Sample 2, psia 2,962 (from dolomite)
Later, a search was made through actual production
histories thought to have some characteristics of the simu-
lated fluid-permeability systems.
Results of Study
The results from the computer simulation model are
summarized in Table 2. The graphical results of these
computer runs are shown in Figs. 1 through 4.
Fluid-Permeability 1-1
The data used in these model runs were from the fluid
analysis (Sample 1) of a fluid sample taken from a con-
solidated sandstone reservoir and relative permeability
curves (Sample 1) for a typical sandstone reservoir. Fig.
1 shows the decline curve histories for three absolute
permeabilities, and Fig. 2 shows the plot of dimension-
less quantities Np/qil vs q/q. Also shown in the backc
ground of Fig. 2 is the solution of the Arps equations.
Note that the trend of the plotted values is not affected by
absolute permeability. The histories from the lower per-
meabilities define the curve in the lower ranges of q;!q ,
and the histories of the higher permeabilities define the
curve for the larger values of q;!q. The equation for the
average curve plotted in Fig. 2 can be represented by
Np = (q) 0.4
qi
l
--q: .
In the lower range of q;!q, the plotted curve closely
approximates the Arps curve for b = 0.0. If this is the
case, then
N p = .!l.i-=-!l.,
aj
which can be manipulated to yield
!L= 1 - !!iNp-
qj qj
1330
This means that a plot of q/qi vs Np will be a straight line
with an intercept of q/qi = 1 when Np = 0; the slope of the
line will be -a;!qi' This plot is shown in Fig. 3. From this
curve we found that
a. = _q_i_ time-I.
I 97,000
Therefore, ai was caiculated for each absolute permeabil-
ity by simply knowing the initial producing rate. Know-
ing ai' the values of ail were calculated and plotted as
shown in Fig. 4. Fig. 4 also shows the solutions of the
Arps equations in the background.
One important observation is that the b value for the
plotted curve does not remain constant. As shown in Fig.
2, the value of b at a q;lq ratio of 5 is less than 0.0, while
at a ratio of 100 the b value is approaching 0.2.
The next run was made using the same load data,
except the external drainage radius was increased to
1,053 ft (80 acres) and the thickness of zone was reduced
to 10 ft,. When compared with the fIrst run, the data
plotted almost identically on the dimensionless curves.
A third run was made to determine the effect reservoir
volume would have on the curves. All load data were the
same as the initial run, except porosity was loaded at 9%.
This gave a reservoir volume of one-half the value of the
initial run. The dimensionless curves again plotted nearly
identically with the other computer runs. However, the
relationship for ai changed to
a. = qi
I 49,000'
Fluid-Permeability System 1-2
This computer simulation used Relative Permeability Re-
lationship 2 to calculate production histories. These his-
tories are plotted in Fig. 1. Using 'these relative permea-
bility curves, we noticed that permeability to gas does not
occur until oil saturation has been reduced from 85 to
55 %. This means that oil recovery at this reservoir should
be more efficient. This is confirmed by the results ob-
tained. Cumulative production from this reservoir
amounted to 250,000 bbl of oil, while recovery from the
previous system amounted to 100,000 bbl.
Fig. 2 shows the dimensionless plot of qJq vs Np/(qil)
and indicates that the early life of this system will be
harmonic (b = 1). Therefore,

and a plot of Np vs q/qj on semilog paper should produce a
straight line with a slope of -q;laj. This plot is shown in
Fig. 3, and the initial decline relationship was
aj =
The dimensionless a;f relationship is plotted in Fig. 4.
The relationships indicate that this reservoir efficiently
produces oil until a q;lq ratio of 6 is reached. At this
point, significant quantities of gas are being produced,
and as production continues, the relative permeability to
gas increases rapidly, resulting in even larger quantities
of produced gas. As this gas is released from the reser-
voir, it is expended without producing oil; oil production
suddenly becomes inefficient. This can be seen in the
abrupt changes on the curves. This simulated behavior
JOURNAL OF PETROLEUM TECHNOLOGY
indicates that relative permeability characteristics do
exert an influence on production histories.
through 4. Production histories followed a hyperbolic
decline with a b value of 0.3. In fact, these histories could
have been predicted accurately using the Arps hyperbolic
formula.
Fluid-Permeability System 2-1
The fluid analysis used in these computer runs was from a
fluid sample (Sample 2) taken from a well in Kingfisher
County, OK, and was obtained from the dolomitic Hun-
ton formation. This sample had a much higher formation
volume factor and a greater solution gas content than
Sample 1. This reservoir contains less oil and more gas
than a reservoir filled with Fluid 1.
Fluid-Permeability System 2-2
Two computer runs were conducted using this fluid-
permeability system. Again, one run was made using a
40-acre drainage area and a 20-ft zone thickness. The
other run was made using an 80-acre drainage area and a
lO-ft zone thickness. Drainage area and zone thickness
had little effect on the dimensionless relationships. Also,
a; was calculated for both runs to be
The data obtained from this system were similar in
character to those obtained for Fluid-Permeability Sys-
tem 1-2. However, this system will recover less oil and
more gas. The Np/q;t vs q;lq curve (Fig. 2) was shifted
slightly to the right as a result of using Fluid Analysis 2.
The value of b during early life was determined to be
larger than 1.0, and the initial decline relationship was
a; = 71
Q
OOO .
,
Stratified Reservoir
Thus far, all data have assumed the reservoir is
homogeneous with constant absolute permeability
throughout its volume. Calculations were made using
Computer Run 1, which combined data from two or more
a = ~
I 60,000
Graphical analysis of this system is shown in Figs.
TABLE 2-RESULTS OF COMPUTER SIMULATION MODEL
Relative
Permeability
Fluid
Analysis
Sample 1
from Bartlesville
Sandstone-
Characteristics
Ps = 2,520 psi
Bo = 1.42
Rs. = 853 cu ft/bbl
/ L ~ = 0.61 cp
Sample 2
from Hunton
Dolomite-
Characteristics
Ps = 2,962 psi
Bo = 1.779
Rs. = 1,395 cu ft/bbl
/ L ~ = 0.233 cp
SEPTEMBER, 1978
Relative Permeability Relationship 2 from
typical dolomite.
14
Critical gas saturation is 30% (gas starts flowing)
krylk = kr/k = 0.11 at 45% gas saturation
First model run was for 40-acre drainage and
20-ft thickness. Analysis of the dimensionless
curves revealed the following equations.
!J:t = (%) 0.4 or Np = (q/.6qO.4)t
a = -q-i-time-
I
, 97,000
qi = 0.215kro h(pe - Pw) bbllmonth
/LoBo[ln(re1rw) - 0.5]
The decline exponent b varied from about 0.0 at
lowq;lq rates to about 0.2 atq;lq = 100.
An additional computer run was made using
80-acre drainage and 10-ft thickness. The di-
mensionless curves plotted identically with the
previous 40-acre case.
A third computer run was made using 40-acre
drainage, 10-ft thickness, and 9% porosity in-
stead of 18% used in the previous two runs. The
dimenSionless curves again plotted identically
with the two previous cases. However, since the
oil volume in the drainage area is one-half the
previous values, a new equation was found for
a
i
= q;l(49,000).
First model run was for 40-acre drainage and
20-ft reservoir thickness. Analysis of the curves
revealed the following:
a = ~
, 60,000
qi = 0.215kroh(pe - p,,J
/LoBo [In (rJrw) - 0.5]
The decline exponent b maintained a value of
0.3 throughout the qJq range of 1 to 100.
An additional model run was made using 80-acre
drainage and 10-ft reservoir thickness. The di-
mensionless curves plotted identically with the
previous 40-acre case.
Relative Permeability Relationship 1 from
typical consolidated sandstone. 14
Critical gas saturation is 5% (gas starts flowing)
krglk = krolk = 0.12 at 27% gas saturation
Model run was for 40-acre drainage and 20-ft
reservoir thickness. Analysis of the curves re-
vealed the following equations:
a. = _q_i_ time-I
, 112,040
qi = 0.215 kroh(Pe - Pw) bbl/month
/LoBo [In(relr w) - 0.5]
The decline exponent b approximates 1.0 with
low q;lq ratios that reduced to about 0.2 when
q;lq = 100.
Model run was for 40-acre drainage and 20-ft
reservoir thickness. Analysis of the curves re-
vealed the following:
a = ~
, 71,000
q, = 0.215kroh(Pe -Pw)
/LoBo [In(relrw) - 0.5]
The decline exponent b varied from a value of
1.0 at 10wqJq ratios to a value greater than 1.0at
q;lq ratios offrom 5 to 10. The value of b reduced
to about 0.4 at aqJq ratio of 100.
1331
-w
W
N
......
0
c:::
;oc

t""
0
'Tl
'"

;oc
0
t""
tIl
c:::
;:::

::z::
z
0
t""
8
-<
Relative
Permeabi 1 ity
System
Fluid
Sample
OJ
c_
o
...,
Vl
"C
c
ttl
V>
E
0
...
I.J...
"**'
OJ
...,
E
0
0
Cl
E
0
...
I.J...
N
"**'
q
Jill.
Mo.
10,000
q
BbJ
Mo.
1000
100
0
#1 Typical Sandstone
20
t - Years
'0
15 20
- Years
q
illlL
Mo.
q
BbJ
Mo.
#2 Typical Dolomite
t - Years
t - Years
Fig. 1-0ecline curve histories for four fluid-permeability systems.
'''"d.
15 20
'''''d.
15 20

00
-w
w
w
Relative
Permeability
System
#1 Typical Sandstone #2 Typical Dolomite
1 0 0 r ~ __ r ~ ~ 100r---------------------------------------r-__ ~ ~
OJ
c:
0
....,
VI
10
-0
c:
AVERAGE CURVE
ttl
V'l qY,
E
q
0
\..
I.J...
"'"
1.0
1.0
100
OJ
....,
'6
0 10
~
0
Cl
ql/q
E
0
\..
I.J...
N
"'"
1.0
1.0
0.5
Np/qll
0.5
Np/qll
0.0
0.0
10
qil.
q
1.0
100
10
1.0 0.5
Np/q;1
0.5
Np/qjl
Fig. 2-Dimensionless Np/qit curve histories for four fluid-permeability systems.
0.0
0.0
Relative
Permeability
System
Fluid
Sample
OJ
s:::
o
+>
<I>
""0
s:::
ro
Vl
E
o
s...
l.J...
1.01' .........
0.5
#1 Typical Sandstone
O
. - qi
,- 97,000
' ....
.........
.'\..,
.......
"-


,"-.!! ______________________ ____________________
0.0 50 100
Np - Cumu lotive Production 1000's Bbls.
1.0
OJ
+>

E
0
q/qj
0
Cl
E
0
0.5
s...
l.J...
N
'*"
____________________ __ ______ ________
0.0
#2 Typical Dolomite
I.e ...-----------------------------------------------.
0.1
OJ =_q_;_
112,040
0.0 1 L-_________ -'--_________ -'-___
1.0 100 200 300
Np-Cumu lotive Production 1000's Bbls .

100
\
\
\
200
O
. - qi
,---
71,000
300
Np-Cumulotive Production 1000's Bbls.
400
400
Fig. 3-Plots of cumulative production vs q/qi for four fluid-permeability systems.
-w
w
UI
Relative
Permeability
System
#1 Typical Sandstone #2 Typical Dolomite
Fluid
Sample

Q)
c::
0
+-'
'" "0
c:: 10 10
10
Vl
E qYq
qY,
0
q

L.L.

'1>0
1.0 1.0
0.0 10 20 30 40 0.0 10 20 30 40
oil oil
100 100
Q)
+-'

E
0
10 10
0
Cl
q'lq
E
0

L.L.
N
'1>0
30
1.0
0.0 10 20 40
1.0
W 30
oil oil
Fig. 4-Dimensionless ait curve histories for four fluid-permeability systems.
o 1,000
::E
"-
..
:;;
00
LLJ

<[
a::
z


c..>
=>
0
0
a::
a.
100

1336
30
20
10
9
1.0
4 9 10 II 12 13 14 ,e
t- YEARS
Fig. 5-Decline curve histories for reservoirs with two layers of different absolute permeabilities.
o Homogeneous Formot ion
GD Imd+3md -3md+9md -5md+15md
6 Imd+ 5md - 3md + 15md
o Imd+7md
x Imd+ IImd
Imd+15md

0.9 0.8 0.7
GD GD
0.6 0.5 0.4 0.3
N./Qit
16
Fig. 6-Dimensionless Np/qit curve histories for reservoirs with two layers of different absolute permeabilities.
17 18

x
0.2
JOURNAL OF PETROLEUM TECHNOLOGY
absolute permeabilities. Fig. 5 shows the production his-
tories of different permeability combinations. Fig. 6
shows the behavior of the N p/q;t vs q;/q curve for various
permeability combinations in comparison with the stan-
dard curve for a homogeneous reservoir.
It can be seen in Fig. 6 that heterogeneity tends to push
the curve to greater corresponding b values as the degree
of heterogeneity is increased. It is also apparent that for
each curve a maximum b will be reached and then as q;lq
becomes larger the b value will again approach the
homogeneous value. We also found that the dimension-
less relationship of the 1- to 3-md curve plotted the same
as the 3- to 9-md and the 5- to 15-md curves. This can be
seen in Fig. 6, and the conclusion can be drawn that any
permeabilities with a 1:3 ratio would follow this same
curve.
To determine more fully the effect of heterogeneity, a
calculation was made on a hypothetical reservoir con-
structed of2 ftofl5-md sand and 18 ft ofl-md sand. The
graphical analysis of this case is shown in Fig. 7. The
production history shows a bend in the rate-time curve
after about 1 year of production. The remainder of the
production history appears to be almost exponential
(straight line). During the early production history (first
year), the 15-md sand is controlling the curve by con-
tributing the largest part of production. After the first
year, the 15-md sand is essentially depleted and the curve
thereafter is controlled by the I-md sand since it is con-
tributing the largest part of the production. The bend in
the curve is the transition between the portion of produc-
tion history controlled by the 15- and I-md sands.
Fig. 7 shows the effect of the heterogeneity on the
Np/q;t vs q;lq dimensionless curve. This curve indicates
that the b value will be a maximum of about 2 when the
q;lq ratio is about 4:4.5. This is one of the reasons that
decline curves can and do calculate b values that are
greater than 1. Fig. 7 shows the a;t vs q;/q dimensionless
relationship for this 1- to 15-md reservoir. Note that the
points on this curve form a straight line for q;/q ratios
greater than 3.0. This indicates that the later portion of
the curve will be exponential, which confirms the obser-
vation of the rate-time curve. In fact, using the correct
scales, the rate-time curve would exactly overlay the plot
on Fig. 7.
In summary, separate zones producing into a common
wellbore can affect the slope and the analysis of produc-
tion decline curves. This effect can cause b values to be
greater than 1. Also, the analysis technique presented
here of using performance plotted on the dimensionless
curves with the solution to the Arps equations as a back-
ground can be helpful when determining actual reservoirs
with severe heterogeneity problems.
Summary
Changing the fluid systems resulted in a greater change in
the initial decline rate (a;). Also, a; is very sensitive to the
volume of the reservoir. In both instances, where fluid
characteristics were interchanged, the change in a; was
200 to 400% as compared with a change of only 15"10
18% when the relative permeability characteristics were
changed.
Comparison of the production histories of the different
fluid-permeability systems shows that changing the fluid
characteristics had a smaller effect on the exponent b than
SEPTEMBER, 1978
changing the relative permeability characteristics. Com-
parison of Systems 1-1 with 2-1 and 1-2 with 2-2 indicate
that changing the fluid characteristics also changed b
from a value of 0.0 to 0.3 in the first instance and from a
value of 1.0 to a value slightly larger than 1.0 in the
second instance. However, the value of b was changed
significantly when the relative permeability characteris-
tics were changed. The value of b was increased from 0.0
to 1.0 when System 1-1 was compared with 1-2, and the
value of b also was increased from 0.3 to a value greater
than 1.0 when System 2-1 was compared to System 2-2.
The initial production rate (q;) depends on the permea-
bility of the formation at the initial water saturation. The
magnitude of q; also depends on the fluid characteristics
as depicted by Darcy's flow equation.
In short, it appears that the relative permeability
characteristics of the reservoir will have a greater effect
on the decline curve constant b while the fluid charac-
teristics will have a greater influence on the constants a;
and q;. Also, permeability variation or reservoir het-
erogeneity will have a predictable effect on the produc-
tion history.
Field Examples
Field data from two wells were refined and plotted on
dimensionless curves to show how the field data com-
pared with the Arps solutions for hyperbolic curves.
Field Example 1
This is the production history of the well from which
Fluid Sample 1 was obtained. The well is producing from
the Bartlesville sandstone and is located in Oklahoma
County, OK. The reservoir should have relative permea-
bility characteristics similar to Relative Permeability Re-
lationship 1. The history of this well is shown in Figs. 8
and 9. Not much can be learned about the reservoir from
these curves, but some important effects of human ma-
nipulation on the production curve can be learned from
this example. In Fig. 8, we can see that production had
just started to establish a significant decline in 1948,
when the formation was shot with nitroglycerin. This
increased production, and the well had again established
a decline when the reinjection of produced gas began.
In 1961, several off-setting wells were shut in because
of excessive GOR production. This, in effect, in-
creased the drainage area of the well and flattened the
decline. In 1966, the well was allowed to produce at
high GOR 's to "blow down" the reservoir. These effects
can be seen in Fig. 9.
The production history of this well shows the effect of
man-induced stimulation. An estimate of production his-
tory in Fig. 8 shows that without the shot of nitroglycerin
and gas reinjection, the ultimate production from this
well would have been about 95,000 bbl of oil. This
estimate was made using the results from Computer
Run 1-1.
q; = 1,540 bbl oil/month
q = 25 bbl oil/month
t = 256 months
!:1Np = (q;0.6
q
.4)t
dN
p
= (1,5406 X 254) 256 = 75,845 bbl
Ultimate cumulative production would equal this
amountplus 20,000 bbl of oil produced by fluid expan-
1337
10,000
Q
Bbl
Mo.
1000
1.0
.....
0
\i

Z
;.:-
r
0
0.5
"r1
~
0
r
tTl
c::
~
--l
~
::I:
Z
0

><
5 10 15 20
t - Years
50 100
Np-Cumulative Production 1000's Bbls.
1 0 0 r ~ ~ ~
10
Qy'
Q
o
100 ,...-----,---------,..---------------,
10
1.0 L-_____ -'-_____ -'-_____ -'-_____ ---J
1.0 10 20
ait
30 40
Fig. 7 -Four performance curves for a reservoir with 18 ft of 1-md zone and 2 ft of 15-md zone.
o 1,000
::e
"-
."

w
...
<t
a:
z

...
0
=>
0
0
a:
Cl. 100
co
0

.;;;
CO
0
Z

w
:;
."
" ;;:
----.
."
0
(!)
-
0
CO
0
"" 0
;;
..
.<=
E
'" ..
a:
"ii

'"
CO

."
0
Q:

.<=
0


Anticipated Preformance Without Stimulation
------.
-----
----.
(0- Monthly Average Production For Year)
1965 1966 1967 1968 1969 1910 1971 1912 1913
10 1946 1941 1948 1949 1950 1951 1952 1953 1954 1955 1956 1951 1958 1959 1960 1961 1962 1963 1964
4 9 10 12 14 15 16 17 18
t - YEARS
Fig_ 8-Production decline curve history for Field Example 1.
60
50
40
30
20
10
1.0
6Np/Qlt
Fig. 9-Dimensionless Np/q;f curve history for Field Example 1.
SEPTEMBER, 1978
1339
o 10,000
:=;;
';;;
:Q
'"
w
f-
<t
'"
z
o
f-
'-'
::>
o
o
'"
I>-
I-YE ARS
sion above the bubble point.
Field Example 2
This example is the type of production history that
prompted this study. This history exhibits a b value
greater than 1.0. Figs. 10 and 11 show the history of this
well.
The well is completed in the Mississippi limestone
formation over a 300-ft interval. The producing forma-
tion is fractured with a tight matrix. Also, production is
obtained from zones scattered throughout the pay sec-
tion. This would indicate that the production history
should resemble the curves in Fig. 7 where 18 ft of I-md
formation was simulated as producing into the same bore
as 2 ft of 15-md formation. This behavior can be seen in
Fig. 11 for values of NP/(qit) from 1.0 to 0.3. At this
point, the choke setting was changed to increase the
producing rate of the well. This same behavior was re-
peated several times. The result of changing the choke
size always increased or maintained the producing rate by
reducing the back pressure on the producing formation.
In each instance, this moved the Np/(qit) vs qJq curve to
the right and maintained a b value greater than 1.0. In the
future it will be impossible to increase production by
resetting the choke. At that time, the curve probably will
become almost vertical and again will move into the
range of b values ofless than 1.0.
Fig. 1 O-Production decline curve history for Field Example 2.
The dashed line in Fig. 11 shows an estimate of the
resulting curve if the choke had not been adjusted on the
well, and the back pressure had remained constant. This
so
50
40
30
20
0.8 0.7
Np/q;1
Fig. 11-Dimensionless Np/q,t curve history for Field Example 2.
1340 JOURNAL OF PETROLEUM TECHNOLOGY
o
curve also shows the effect of permeability variation.
One other observation can be made. The results of
human manipulation of the choke is not apparent in Fig.
10 (the rate-time curve), but these changes are easily seen
and interpreted in Fig. 11. Plotting Np/( qjt) vs q/q can be
a helpful diagnostic tool for evaluating the production
history of a well or lease. Also, production curves exhibit
b values greater than 1.0 due to human manipulation to
maintain the highest production rate possible.
Conclusions
1. The dimensionless curves Np/ajt vs qh and ait vs
q;lq for a particular fluid-permeability system are not
affected by the absolute permeability or size of the reser-
voir. The behavior of these plots is determined by (l) the
characteristics of the contained fluid, (2) the relative
permeability characteristics of the reservoir rock, (3) the
reservoir drive mechanism, (4) reservoir heterogeneity,
and (5) manual manipulation of production.
2. Reservoir heterogeneity tends to increase the mag-
nitude of b as the degree of heterogeneity is increased. It
also is apparent that b for a heterogeneous system will
increase to a maximum value and then as the ratio q;lq
becomes large, b will decrease and approach its homo-
geneous value.
3. Reservoir heterogeneity can and does cause b val-
ues to be greater than 1.0.
4. Manual manipulation of production can and does
cause b values to be greater than 1.0.
5. The dimensionless plots for heterogeneous systems
of 1 and 3 md, 3 and 9 md, and 5 and 15 md all plotted the
same curve. This indicates that heterogeneous systems in
the ratio of 1:3 will plot congruous dimensionless curves.
6. It appears that the relative-permeability characteris-
tics of the reservoir have the greater effect on the decline
exponent b, while the fluid characteristics have a greater
influence on the constants aj and qj.
7. The equation Np/(ajt) = (q/qj)a may better define
certain decline curves than do the Arps equations.
8. The plotting of production data on the Np/(ajt) vs
q/q curve can be a helpful diagnostic tool for evaluating
the production history of a well or lease.
Acknowledgment
We thank Ward M. Edinger for permitting the use of
facilities and office equipment of Edinger, Inc., when
preparing this study.
Nomenclature
a = decline as a fraction of pro<,iuction rate
aj = initial decline
b = decline exponent
Bo = oil formation volume factor
SEPTEMBER, 1978
BOj = initial oil formation volume factor
h = formation thickness
krg = relative permeability to gas
k
ro
= relative permeability to oil
K = constant
Np = cumulative oil production
!1N
p
= oil production during an interval
Pe = reservoir pressure at external drainage
radius
Pw = wellbore terminal pressure
q = production rate
qj = initial oil production rate
r e = radius of drainage
r w = wellbore radius
RSj = initial gas in solution
t = time
a = arbitrary exponent
11-0 = oil viscosity
References
1. Arps, J. J.: "Analysis of Decline Curves," Trans., AIME (1945)
160,228-247.
2. Arps, J. J.: "Estimation of Primary Oil Reserves," Trans., AIME
(1956) 207,182-191.
3. Ramsey, H. J., Jr., and Guerrero, E. T.: "The Ability of Rate-
Time Decline Curves to Predict Production Rates," J. Pet. Tech.
(Feb. 1969) 139-141.
4. Shea, G. B., Higgins, R. V., and Lechtenberg, H. J.: "Decline
and Forecast Studies Based on Performances of Selected California
Oilfields," J. Pet. Tech. (Sept. 1964) 959-965.
5. Arnold, R. and Anderson, R.: "Preliminary Report on Coalinga
Oil District," Bull., USGS (1908) 357, 79.
6. Cutler, W. W., Jr.: "Estimation of Underground Oil Reserves by
Well Production Curves," Bull., USBM (1924) 228.
7. Gentry, R. W.: "Decline Curve Analysis," J. Pet. Tech. (Jan.
1972) 38-41.
8. Slider, H. C.: "A Simplified Method of Hyperbolic Decline Curve
Analysis," J. Pet. Tech. (March 1968) 235-236.
9. Mead, H. N.: "Modifications to Decline Curve Analysis,"
Trans., AIME (1956) 207, 11-16.
10. Lefkovits, H. C. and Matthews, C. S.: "Application of Decline
Curves to Gravity Drainage Reservoirs in the Stripper Stage,"
Trans.,AIME(1958) 213, 275-280.
11. Matthews, C. S. and Lefkovits, H. C.: "Gravity Drainage Perfor-
mance of Depletion-Type Reservoir in the Stripper Stage,"
Trans., AIME (1956) 207, 265-274.
12. Fetkovitch, M. J.: "Decline Curve Analysis Using Type Curves, "
paper SPE 4629 presented at the SPE-AIME 48th Annual Fall
Meeting, Las Vegas, Sept. 30-0ct. 3, 1973.
13. Craft, B. C. and Hawkins, M. F.: Applied Petroleum Reservoir
Engineering. Prentice-Hall, Inc., Englewood Cliffs, NJ (1962).
14. Amyx, J. W., Bass, D. M., Jr., and Whiting, R. L.: Petroleum
Reservoir Engineering, McGraw-Hili Book Co., Inc., New York
(1960) 204. .
15. Brons, Folkert: "On the Use and Misuse of Production Decline
Curves," Prod. Monthly (Sept. 1963) 22. JPT
Original manuscript received in Society of Petroleum Engineers office Jan. 11,1977.
Paper accepted for publication Nov. 16, 1977. Revised manuscript received June 13,
1978. Paper (SPE 6341) first presented at the SPE-AIME Economics and Evaluation
Symposium, held in Dallas, Feb. 21-22,1977.
1341

Potrebbero piacerti anche