Professional Documents
Culture Documents
Petroleum Economics
Analysis of Decline Curves
By J. J. ARPs, * MEMBER A.I.M.E.
(Houston Meeting. May 1944)
show the possible ultimate production as certain minerals by states or nations. Such
a function of the initial production rate. a curve will rise from zero value at the
W. W. Cutler,S in 1924, pointed out, after time of first production to a maximum
an intensive investigation of a large and then slowly decline, presumably to
number of oil-field decline curves, that zero value. The possibilities of various
the assumption of constant percentage mathematical relationships and different
decline and a straight-line relationship methods of curve fitting are described in
on semilog paper generally gave results this paper. The best results were obtained
that were too conservative in the final with a curve of the type:
stage. In his opinion, a better and more
reliable straight-line relationship could
be obtained on log-log paper, although some
horizontal shifting usually was necessary. R. H. Johnson and A. L. Bollens,12
This implied that the decline curves show- in 1927, introduced a novel statistical
ing such characteristics were of the hyper- method for extrapolation of oil-well decline
bolic rather than the exponential or curves. With their so-called "loss-ratio
geometric type. He also recommended the method," the production rates are tabu-
use of the family decline curve, either lated for equal time intervals, then the
graphically constructed or statistically drop in production is listed in a second
determined, which is a representative column and the ratio of the two, or "loss
average decline curve for a given area ratio," is listed in a third. A curve to be
based on a combination of the actual rate- investigated with this method usually
time data from a number of wells in the shows, after proper smoothing out, either
area. a constant loss ratio or a constancy in the
C. S. Larkey,9 in 1925, showed how the differences of successive loss ratios. Some-
method of least squares could be applied times it may be necessary to take these
successfully to decline curves belonging to differences two or three times before
both the exponential and the hyperbolic constancy is reached, and often additional
types. He also demonstrated that the smoothing out of the data is required. This
application of this well-known statistical procedure furnishes an easy and convenient
method makes a strict mathematical method for extrapolation. It is only neces-
extrapolation of a given decline trend sary to continue the coh1mn with the
possible. constant figures in the same manner and
H. M. Roeser,lO in 1925, showed that then work backward to the production-
equally reliable results can be obtained rate column.
when, instead of the rigorous method of H. N. Marsh,13 in 1928, introduced
least squares, a somewhat simpler method the rate-cumulative curve plotted on
of trial and error to determine the neces- coordinate paper and pointed out that
sary constants is followed. He illustrated this relationship generally appears to be
his method with examples of both the or approaches a straight line. Although
exponential and the hyperbolic types of this is only mathematically exact for
decline curves. In his paper was also the decline curves of the exponential type, as
first reference to the matliematical relation- will be shown later, it was pointed out
ship between cumulative and time for in his paper that the errors in estimating
hyperbolic type of decline. ultimate recovery with this method in
C. E. Van Orstrand,ll in 1925, investi- most other cases were generally small or
gated the empirical relationship of produc- negligible. A distinct advantage of this
tion curves representing the <1utput of type of curve is its simplicity in appraising
J. J. ARPS 23 1
t he effect of differell t methods of production method, originated by A. F _ van Ever-
control on the same well. dingen in Houston, proved particularly
R. E. Allen,14 in 1931, mentioned four valuable for prorated Gulf Coast water-
types of decline and classified them accord- drive production.
ing to a simple mathematical relationship. For depletion-type or gas-drive-type
The decline types were: pools without water encroachment, how-
1. Arithmetic, or constant decrement ever, a parameter other than oil or water
decline. percentage had to be found to replace
2. Geometric, constant rate or expo- the production rate.
nential decline W. W. Cutler and H. R. Johnson,17 in
3. Harmonic, or isothermal decline. 1940, showed how potential tests, taken
4. Basic, or fractional power decline. periodically on prorated wells (or calcu-
Type I is of little practical value for lated from bottom-hole pressure and
production-decline curves. Type 2 is productivity-index data) can be used to
the well-known straight-line relationship reconstruct or calculate the production-
on semilog paper, and type 3 is the special decline curve, which the well would have
case of hyperbolic decline where the decline followed if it had been permitted to produce
is proportional to the production rate. at capacity.
It was not possible to reconcile the equa- H. C. Miller,18 of the Bureau of Mines,
tion given for the type 4 decline, as the introduced in 1942 the pressure-drop
nominator and the denominator were of cumulative relationship on log-log paper
the same order, indicating a possible and showed how changes in reservoir
misprint. performance may be detected by abrupt
S. J. Pirson,15 in 1935, investigated the changes in the slope of such a curve.
mathematical basis of the loss-ratio method C. H. Rankin,19 in 1943, showed how
and arrived at the rate-time relationships the bottom-hole pressure can sometimes
for production-decline curves having a be used to advantage as a substitute for
constant loss ratio, constant first differences the rate of production of the rate-cumula-
and constant second differences. Those of tive curve on prorated leases. Apparently,
the first type appeared to be identical this method applies only in pools where
with the simple exponential or constant water drive is absent or negligible and
percentage decline curves, which straighten where productivity indexes are constant.
out on semilog paper; those of the second In the Oklahoma City field, which is
type were the hyperbolic type of decline well known as a typical example of gravity
curves, which can be straightened on drainage, a plot of fluid level against the
log-log paper and those of the third type cumulative production has been used
appeared to have such complicated mathe- successfully to estimate the reserves of
matical equations as to be unsuitable for wells with constant production rates.
practical purposes. P. J. Jones,21 in 1942, suggested for
During the period of production curtail- wells declining at variable rates an approxi-
ment, interest centered upon suitable mation whereby the decline-time relation-
curves for reserve estimates that did not ship follows a straight line on log-log paper.
require the usually constant or almost This corresponds to an equation:
constant actual rate of production.
H. E. Gross,16 in 1938, showed the log D = log Do - m log t
advantages of substituting oil percentage
in gross fluid for the production rate in in which Do designates the initial decline
the Marsh rate-cumulative curve. This and m is a positive constant. Integration
23 2 ANALYSIS OF DECLINE CURVES
of this relationship will lead to a rate-time In most actual pools, however, the
equation of the general form: aforementioned idealized conditions do
Dot 1- - not occur. Pressures usually are not
P = P.eIOo(m I) proportional to the remaining oil, but seem
to decline at a gradually slower rate as
It may be noted that this relationship the amount of remaining oil diminishes.
will not straighten out on semilog or At the same time the productivity indexes
log-log paper, but shows the interesting are generally not constant, but show a
characteristic of straightening out when tendency to decline as the reservoir is
the log-log of the production rate is plotted being depleted and the gas-oil ratios
against the log of the time. increase. The combined result of these
F. K. Beach,20 in 1943, showed, with two tendencies is a rate-cumulative re-
examples from the Turner Valley field, lationship, which, instead of being a
Canada, how cumulative-time curves some- straight line on coordinate paper, shows
times can be extrapolated as straight lines up as a gentle curve, convex toward the
in their last stage by plotting the antilog origin.
of the cumulative production against time. If the curvature is very pronounced,
Such a straight-line relationship is mathe- the curve can sometimes be represented
matically correct only for the case of by an exponential equation and the rate-
harmonic decline, where the decline itself cumulative relationship straightened out
is proportional to the production rate, as on semilog paper. This type is called
will be discussed later. harmonic decline, and its equation is
identical with Eq. 14, derived on page 12.
RESERVOIR CHARACTERISTICS
By differentiation, it can be shown that
AND DECLINE CURVES
in this case the decline percen tage is
In order to analyze what influence directly proportional to the production
certain reservoir characteristics may have rate.
on the type of decline curves, it was first When the curvature of the rate-cumula-
assumed that we are dealing with the tive relationship is not pronounced enough
idealized case of a reservoir, where water to straighten out on semilog paper, it can
drive is absent and where the pressure is usually be represented as a straight line
proportional to the amo'unt of remammg on log-log paper after some shifting. This
oil. It was further assumed that the identifies it as a hyperbola and it can be
productivity indexes of the wells are shown that it will fit Eq. 13 (p. 12) for
constant throughout their life, so that the the general case of hyperbolic or log-log
production rates are always proportional decline.
to the reservoir pressure. From this general discussion, it is evident
In such a hypothetical case, the relation- that the hyperbolic type of decline curve
ship between cumulative oil produced should be the most common and that
and pressure would have to be linear and, harmonic decline ~s a special case, which
consequently, also the relationship between occurs less frequeptly.
production rate and cumulative production. The exponential or semilog decline,
This linear relationship between rate however, although less accurate, is so
and cumulative is typical of exponential much-simpler to handle than the other two
or semilog decline, as will be shown later that it is still quite popular for quick
(Eq. 4), and simple differentiation will appraisals and approximate estimates;
lead to the basic equation for this type of particularly since a large number of decline
decline in Eq. I. curves actually show an apparent constant
J. J. ARPS 233
decline over limited intervals. The decline The average value over the period from
percentage in such calculations is then July 1940 to January 1944 is 86.8 and this
usually taken somewhat lower than the value was used to extrapolate the produc-
actually observed value in order to evaluate tion rate to January 1947 .in the lower
the possibility of a smaller decline in the half of the tabulation. The procedure
final stage.
TABLE I.-Loss Ratio on a Lease in the
Cutbank Field, Montana
EXPONENTIAL DECLINE
(TYPICAL CASE OF EXPONENTIAL DECliNE)
Exponential decline, which is also called
Loss in Loss Ratio
" geometric," "semilog" or "constant Monthly Production (on
Rate dur-
percentage" decline, is characterized by Month Year Produc-
tion ing6
Monthly
Basis).
Months
the fact that the drop in production rate Rate. P Interval. a = 6.!'...
M' M'
per unit of time is proportional to the
production rate.
July ...... 1940 460
January .. 1941 431 -29 -89.2
Statistical A nalysis and Extrapolation July ...... 1941 403 -28 -86.4
January .. 1942 377 -26 -87.0
July ...... 1942 352 -25 -84.5
The simplest method' to recognize January .. 1943 330 -22
-21
-90.0
July ...... 1943 309 -88.3
exponential decline by statistical means is ~nuary .. 1944 288 -21 -82.3
the loss-ratio procedure. 12 With this July ...... 1944 269.4 -I8.6 -86.8
January .. 1945 252.0 -174 -86.8
method the production rates P at equal July ...... 1945 235.7 -16.3 -86.8
January .. 1946 220.4 -I5.3 -86.8
time intervals are tabulated in one column, July ...... 1946 206.1 -I43 -86.8
the production drop per unit of time, tlP in January .. 1947 192.7 -134 -86.8
a second column and the ratio of the two Average loss ratio July 1940 to January 1944. 86.8.
(a = loss ratio) in a third. If this loss
n ccI me IOO
percentage 86.8 = LIS per cent.
ratio is constant or nearly constant, the Extrapolation until January 1947 by means of
curve can be assumed to be of the expo- average loss ratio. 86.8.
nential type. The mathematical basis for followed in this extrapolation is self-
this will be discussed hereafter. explanatory; the same method that was
It will often be found, if time intervals used to arrive at the loss ratio from the
of one month are used and when the decline known production rates in the upper half
percentage is small, that' the general trend of the tabulation is used in reverse to find
is disturbed considerably by irregularities the unknown future production rates from
in the monthly figures, and in such cases the constant loss-ratio values.
it is better to take the production rates
further apart. As an example, Table I Mathematical A nalysis 16
shows the data from a lease in the Cutbank Rate-time Relationship.-The rate-time
field, Montana, where the monthly produc- curve for the case of exponential decline
tion rates are taken at six-month intervals. has a constant loss ratio, as shown in the
Since the loss ratio is defined as the preceding section, which leads to the follow-
production rate per unit of time divided ing differential equation (see list of symbols
by. the first derivative of the rate-time on page 20):
curve, it is necessary in this case to intro-
P
duce a 'factor 6 in the last column to correct dPjdt = -a [I}
the drop in production rate during the six
months interval back to a monthly basis. IIIwhich a is a positive constant. After
The loss ratios in the fifth column of the integration of this equation, and after
table appear to be approximately constant. elimination of the integration-constant by
234 ANALYSIS OF DECLINE CURVES
Graphical Extrapolation and Practical this strip for various values of the decline
Shortcuts percentage D, it is possible to carry the
As pointed out before, the rate-time multiplication out on the same graph paper
curve for exponential decline will show used for the curves, and read the answer
a straight-line relationship on semilog on its vertical log scale. Figs. I and 2 show
paper and can, therefore, be extrapolated how such a calculator, designed for deter-
by continuing the straight line. mination of both decline percentage and
J. J. ARPS 235
CAlCUL-ATOf~
nll- Exponential
Decline
mJ
f~~ure
ProductiQ~-
tOO,OI!,g - r'" ______y_.,..,......
,..~,..,.',...,~"":r..'"""""'.L_' __ ...........
~'= ~~<>1I .:;
SO,ooo - f---"-~-"- - . - - - - , - - - ~
70,00" _ _ _ _ _ _ _ _ _ _ _ .. _ _ _ _ _ _ __
&0,000 - t- - - - - - - - - - --,---------- ~
5 0 ,000 - : : 1 1 - - - - - - - - - - -------i~
40,000-
$ t,ooo --11""---
~ - ~~"'I"<
-'1"''''< ~~ : /
.-g -vj ~IQ ~
9 <> ~~
.., ',000___ ---1+'10--- :; / ----,---",- - , -
~ :~: - ;- V "" ~~ __ ~~ j / ---------+----t
II 700 -j--------Fr~==::i!!~~;;;;::-g 0 -,---,----+----i
D. Iloo--lf--------\ MonthllJ ~ ,-~Icl: / ' 0,5-t---------+---..-
500-::11---------\ Dec:line ~ ~ P"..... t----------- - - -
400_:t_ _ _ _ _ _ -L!:i=::4-='1o=O==::J-I---_~?~ ............
"'0,30
r-------..- - -
300-_,______________~I--___________~ar__---_------4A~----
100,000
--I--- . 90,;000
f-. 80,000
------~.------ '---.<~-----I-~-----+--..-,'F, 70.000
---~-.'''~" -.~~.~ ,~--,,---I-, ___ ~~ __ -+-_,_____ .~-+ __ '.- -::
-
;o~OOq
';):~f~Q.
II
- 40,00':'
~
<I) - lO}OCeJ
..
S
; " t~,Ooo
.;:
g.-
..J
10,aGO
9,0'00
s,uoo
'1,OOQ
f.-
r-----------+----------~ -- I- 6,000
~--------4--,-------- SIO Gb
E 4,~OO 0
~,coo '"
-;
.....
~
CO
- 1.,000
::..
1,000
~uo
S6t,
-1,00
'tOO
0.'-
J. J. ARPS 237
future production, is used. The monthly m.atched with this production rate of
decline percentage was read off from scale 190 bbl. per month and the future recovt:ry
Be in :Fig. 1 as 4 per cent and the constant is read off opposite arrow E as 4750 ban:els.
.
~"Q.OO
<)0,00
.:-:
..:
1
tU,OIll
"'0.00
40,#0\)
~O.OO 0
.-
0-:
40,011 0-:
" ...0-: 'J
tCpO 0-:
t~
.,..0-1
.. 't ~:;
tO,OO
oe
.~
~
rJ1'. t
- - --- - -- .... +:.,, , ,
I
,
I
, ,
,
, , ,
c: ',00 0-: .:. OrnU",
... I
! .5 ~,..~
---- - -- ~~
1\1-\
~.00i'J 0_ - -t--
I
1_ ",00 o~ I '-
I
I
l.
~
0-'
",000
,
I
,
I '-
'-
'-
~~
: :'1,000 - I
r' '.
.... "\.~
,,
j
'- ,
:. 't,OO 0-:
,
,," ~i ,
(i '- I '-
~ --
li
~
I '- ,,
I '- ,,
I :c~ ~.
.. .......,
"'0 t.O~ 0
, '"J
I T, ,
! :: , , I"
"
, '- , I
r-_
, ,, ,,
II '"
,,- I
4. ...~-
.. .. _, - - - ,,\
I
-- -- .. --~
I"
--~;T
<>-
m~
to
,0 ~
0
IUS 195& IU7 .9311 I'~~ 19,,0 1541 194't "4~ 1944 1945 1S4i iS4? I!i<'>l)
t:; Tim.e
FIG. 3.-GRAPHICAL EXTRAPOLATION OF HYPERBOLIC RATE-T~!lE CURVE ON SEWLOG PAPER.
ratios ue constant or n~arly um::;tant I2 ,1i> to find the proper'ld.lues. The loss ratios
As an e;;:a;~lple, Table 2 shows the loss ratio thus ol.,tained indicated a fairly uniform
for production data frum a lease producing arithmetic series and consequently the
from the Arbuckle lime in Kansas. This differenCes between successive loss ratio
lease had been producing under conditions values b are reasonably constant. The
of capacity production since the completion average is 0.508.
of dtilling and shows a ratetime curve on These differences represent the d.:riva-
sel.llilog f!::tper, curvIng steadily to the tives of the loss ratios with respect to
right (Fig. 3). To eliminate ineg,.llarities, time; and since sixmoIith intervals are
it waJ nece;sary to smooth out the origh,al used, a cOirectioll factor of 76 was intro-
data (see curve J B on Fig, 3). The p,.'oduc- duced to find the proper values of b. The
tion rates listed in Table 2 are identical average value for b was used to extrapolate
with the circles on the cur,'e in Fig. 3. the curve to July 1948 by reversing the
TARLE 2.-Loss Ratio for Lease Producing process used in the upper part of the
frvm Arbuckle Lime in Kt!nsas tabulation. From these data, it is evident
(TYPICAL CASE OF HYPEkBOLIC DECLINE) that the lease can be expected to reach
- - - - - - - - - - ' - - - - - / - - - - ; : s t ~: its economic limit of 400 bbl. per month
Monthly .LOSS In Loss rivative
during the second half of 1947.
Pr~duc- ~rod:lc. "Ratio on of Loss As will be shown later, the mathematical
bon tlOn.Rakl Monthly Ratio,
Monti. Y car Rate, P dunng 6 Basis b = equations of the rate-time and rate-
(Cune Months
J B, InterVal.\a = 6 Mi
P {6P}
!J. ::iP cumu1ative curves for hyperbolic decline
hg. 3 ) ! J . P -6- are essentidly of the same type and it is
therefore also possible to use the loss-ratio
Jan ... 1937 28,200 method for extrapolation of rate-cumula-
July .. 1937 15.680 -12,520 - '7. 5~
Jan ... 1938 9,,/00 5.980 - 9i2 '-037 tive dat",. The only difference from the
July .. 1938 6.635 3.065 --1297 --054
Jan ... 1939 4,775 1,860 - IS. 39 -040 procedure in Table 2 is that the time
July .. 1939 3,628 1,147 - 18.96 -0.59
Jan ... 1940 2.850 778 -21.96 -0.50 column is replaced by cumulative produc-
July .. 1940 2,300 550 -25.08 -0.52 tion figure", and that the intervals therefore
Jd.n ... 1941 1,905 395 -2895 -0.64
July .. 1941 1,6ro 295 -32.76 -0.63 may not be constant. The loss ratio in that
Jan ... 1942 1.365 245 --3443 -0.28
July. 1942 I 177 188 - 3697 -0.42 case is the production rate at a given point
Jan ... 1943 1,027 ISO -41. IS -070
divided by the ratio of the drop in produc-
July .. 1943 904 123 -4420 --0508
Jan ... 1944 802 102 -4725 -0508 tion rate to the total production during
July .. 1944 717 85 - 5030 -050S
Jan .. 1945 644 73 -5335 -050S the preceding interval. In a similar way,
July .. 1945 582 62 -5640 -050S
Jan ... 1946 529 53 -- 59. 45 -0.50S
the first derivative should be determined
July .. 1946 4 83 46 -62 50 -0 50S as the increase in lo:>s ratio over the given
Jan ... 1947 442 41 -65.55 -0508
July .. 1947 406 36 -68.60 - o. 508 interval divided by the total production
Jan ... 19i18 375 31 -?I.65 -0 50S
July .. 94S 347 28 -7470 -0.508 during the same interval. In hyperbolic
decline, the first derivative should be
First derivative of loss ratios approxiInately con-
stant; a.verage b = - 0.508. approximately constant. To extrapolate
Extrapolation until July 1948 by means of this
average b value of -- 0.50S. the data and find the ultimate recovery
As in the case of exponential decline, the for a given economic limit, the average
production rates were posted at six-month first derivative can be used to extrapolate
interval~ to eliminate monthly fluctuations the tabulation in a manner similar to
and to embrace the general trend of the that of Table 2.
curve without too much work. Sir,ce the
loss ratio a is defined as the production Mathematical Analysis
rate divided by the first derivative of the I. Rate-time Relationship.-When the
rate-time curve, a factor 6 was introduced first differences of the loss ratios are
J. J. ARPS
P
dP/at = -bt -- ao [8) In the special case, where b = I, the
integration results in the expression for
in which ao is a positive constal).t, represent harmonic decline as can be easily verified:
ing the loss ratio for t ,= o. Eq. 8 can he
simplified to:
c= aoP o (log Po - log P) [14]
the production rate has dropped to lOO bbl. tion construction for the hyperbolic-type
per month. decline curve on semilog paper as illustrated
Three-point Rule.-The hyperbolic de- hy Fig. 3 and discussed hereafter.
dine curve shows another interesting
Graphical Extrapolation .~reth(ds
feature, which can sometimes be used to
advrmtage. It can be expressed as: "For Log-log Paper.--As pointed out before,
;my tviO points on a hyperbolic rate-time both the rate-time and rate-cumulative
\;urve, of which the production rates are in curves for hyperbolic decline can be repre~,
'l given ratio, the point midway between sen ted and extrapolated as straight lines
.viii have a production rate which is a on log-log paper after some shifting. The
fL'I'ed number of times the rate of either rate-cumulative curve for the special case
the first or last point, regardless of where of harmonic decline where b = I, however,
~he first two points are chosen." can be straightened only on semilog paper.
In other words, if on a curve with an Log-log paper extrapolation has the
exponent b = 0.$, the first point has a disadvantage of giving the least accuracy
production rate of ?A bbl. and the last at the point where the answer is required;
point a rate of A bbl., the pOint midway it is also somewhRt laborious on account
between will have a value of I.374A bbl., of the e:N:tra work involved in shifting until
egarJlcss <If where the first "et of points the best straight-line relationship is found"
's selected on the curve and regardless of Semilog Paper.--Although log-log paper
the time interv~l. The validity of this is used to a large extent for production
::tatement can be shown as follows: curves of the hyperbolic type, there are
According to Eq. 10, the production still some companies that continue to plot
+
rates at time t -- v, t Zl'.nd t 'II Will be: their production curves on semilog paper,
9, = Po ( ! + !!... e)-1I6
or P,-b = Po-b (I + ~ e) [181
" aD
lI.nd
P.+. = Po { I + ~ (t + v) } -lib or
By adding together the right sides of even though the decline may be of the
~qs. 11 and 19. the time interval 'II is hyperbolic type. The reason seems to be
.Jiminated and %1I.n cl!pression is obtained that this procedure allows a wide range
tb~t i!' twice t.he value of the right side of in small space on the vertical log scale
F~q. xlt Therefore: and at the same time has a simple linear
2P,-t = P,_.-b + p,{.-'J; [-zo] horizontal time scale. The curvature in the
rate-timl': relationship for this case, how-
If the rate at the first point is n times ever. makes extrapolation difficult and
the rate at the last point, the value of the uncertain.
rate at the middle point (P,) can be With the help of the" three-p";':C!; rule"
c:<tpre!!sed as: for hyperbolic decline. it is now possible
!
to extrapolate such a curved hyperbolic
PI = (t'-I> ~t- 1) - b PH> [nJ rate-time curve on semilog paper with a
fair degree of accuracy by simple graphica.l
This relationship was used advan- construction. This procedure is shown on
tageously for a simple graphical extrapola- Fig, 3. Three points, A, E and B, are
tlJO 400 S!vG f>C)Q GQO SOC IpOO
Q
:>
a:
::;)
U
U ,
a.
cr:
~
eOO I
I
'/w.
,! II +
...0
V
...Z:z:
~
'00-
600 I ".
.. .. .. 15 40,000
~ s.. I I 0
tl-t
:::;
\0,000 II:
':-'
r-
<Ii
I&j
a.
':-'
.J
. i If)
:.-
1
..J
lSI 1
ell ~
til III 'tI
'00
! ~
1&1
iii
!( !(
1(
cr: .0,000
a:
9,000
z0 8.~OQ
Z
'1/;)00
i= 0
j:
,.
'.000
u &,000
U
::l
Q "',ClOO. ::l
0 0
cr: ',000 0
cr:
a.
I.
'&.,000 Il.
"
CI.
Q.
0
19~5 19~6 19~7 mtl 1'~9 1940 1941 \942. 1943 1944 1945 1946 1945 190;,0 1951
Tll\AE"I t:
FIG. 4.-STRAIGHT-LINE DECLINE CHART 1'0R lIYPE.Il:"OLIC DECLINE. b = 0.5.
To be l:sed f('\"( production curves wh~:re rle<:line is proportiona- t<> 1(,,. "O'lare root of the pO~l1ctiOJl late.
:n~~<.~;<;,:\,l,;t~1 ~~<~b iiac., ir!,\ttl ~~~ ;;,~,::,s ".:,:. i:r.\;~~~, !}j<!dllC'n~ ::.:""6/U ~}u}' '\.':~ j~ Ur"~,
ANALYSIS OF D}:CUNE CURVES
The chart for b ,,- 0.5 is shown in Fig. 4 right should be used in conjunction with
and the data from Table 2 are plotted on the linear cumulative scale on the top of
this chart to show the straight-line extra- the chart,'while the scale on the left should
polation procedure. The scale on the right be used in combination with the linear
o
e
Q. 40,OOO~ 30,000Ci.
Q.
"
Q. Cl.,
1000..,
qOO-
800 - . 30,000
700 - 20,000
600-
500 -:
400- 20,000
200
300- 10,000
~
100 ~.J...J....'--'-L_~':-'-:::'
1000 1000
t=Time
a 100
FIG. 5.--STRAIGHT-LINE DECLINE CHART FOR
EXPONENTIAL DECLINE. FIG. 6.--STRAIGHTLINE DECLINE CHART FOR
a = constant; b = o. HYPERBOLIC DECLINE.
(For curves with constant decline,) b = 0.25.
(To be used.if decline is proportional to the
is designed to match the b value of the one ~ power of the production rate.)
on the left, so that it will fit the rate- time scale on the bottom. Both curves
cumulative relationship. The scale on the can then be plotted and extrapolated as
244 ANALYSIS OF DECLINE CURVES
-;traight lines, simultaneously. Vertical points are selected on the available curve
scales for similar charts, designed for in such a manner that the production rate
b values of 0, '0.25 and 1.0 are shown of the first point is twice the rate at the last
In Figs .. 5,6 and 7, respectively. point. The production rate at the midway
point is then read off and its ratio to the
c= Cumulotive 0"11 production 100000
iOO,OOO last point determined. If this ratio has a
5000- I I qO;OOO
r- 80,000 value between 1.414 and 1.396, the chart
2000 -: t-. 70,000
r- 60.000
for b = should be used; if it is between
1.396 and 1.383, the chart for b = 0.25
100 8-: ., I:- 50000
90 - "is will be better; if it is betw6Cn 1.383 and
800- ~ ::- 40,000
'"
100-, "is u
1.352, the chart for b = 0.50 should be
600--' VI
'" E f-- 30,000 preferred, and if the ratio is between 1.352
<l> ::E
500.., ~ c and 1.333 the chart for harmonic decline
<S
~
400-'
--'
H
c Sf-- 20,000 (b = I) will give the best results. If these
ratios are too close together, other values
f-- can be calculated with the help of Eq. 21.
~ A simpler method is to plot the rate-time
...
>
-300- f--IO,OOO a curve on semilog paper (b = 0) and if it
~
... C:- 9000 ~
:l f-- 8000 I shows a persistent curvature three repre-
<.)
t-- 1000 ~ sentative points should be replotted on the
'4>0
I
I;),.
r 6000 ~ chart for b = 0.5. If the three points do
... 5000 'i not lie on a straight line, but show curva-
~ t--4000 ~
~ !:
0
ture to the right, the chart for b = I
E 200- should be selected; if the curvature is
t--3000 ~
. jqO-
-+-
:J
~ downward, the chart for b = c.25 should
0
~ 180- ....
a. give better results.
t--2000 h
no- Q,. Another method is to set up a loss-ratio
" 160-
~ tabulation and actually determine the
average value of the first differential b.
150- The chart with the closest b value should
flOOO
qOO
800 then be chosen. This method was followed
140- - 100 in Table 2, and since the b value obtained
- 600
(0.508) was very close to 0.50, the chart
130- =- 500
for this latter value was used (Fig. 4).
:c.. 400
OTHER EMPIRICAL DECLINE CURVES
12<1
- 300 In additi.on to the exponential type of
decline, which is the simplest empirical
110- - 200 relationship and has found widespread
100
m I
t=Time
I
-
100
application for approximate estimates
because of its simplicity, and the hyperbolic
type of decline, which is more complicated,
but-also generally more accurate, there are
FIG. 7.-STRAIGHT-LINE DECLINE CHART FOR several empirical equations that can
HARMONIC DECLINE.
b = I. sometimes be used to represent production-
(To be used if decline is proportional to the decline curves if the simpler types are
production rate.) inadequate. Three of the more important
To determine which chart should! be used, types are discussed in the following
the three-point rule c~.'l b~ used: Two pages.
J. J. ARPS 245
Loss Ratios Form a Geometric Series TABI.E 4.-Loss Ratio for the Family
(Ratio Decline) Decline Cur~e of a Field Producing from
the Wilcox Sand in Oklahoma
A curve of this type has the charac- (TYPICAL CASE OF RATIO DECLINE)
teristic that the decline percentage-time
relationship is similar to the rate-time ] Loos In Loos Ratio of
Produc-
Monthly tion Rate Ratio. on Succes c
relationship for exponential decline and M onth Year Pr'!duc- during 6 Monthly sive L05~
Basis, Ratios,
can be plotted as a straight line on semilog tI~n Months p
RatIO. P Interval. a=6t.P r =a~ -
- - - - - - - - ._--- _.__
paper. In other words, the decline fraction t.P an_I
..
.-
itself is declining at a constant percentage
an ... 1 20.360
per month. The differential equation for uly .. 1 13.260 -7.100 -11.206
an ... 2 8.990 -4. 270 -12.63' 1.127
the rate-time curve is: July ..
Jan ...
3 6.390
4.650
-2.600
-1.740
-14.746
-16.034
1.167
1.087
July .. 3 3.490 - 1.160 -18.052 1. 126
P
- ... _-- = aorl Jan ...
uly ..
4 2.700
2,140
-- 790
560
- 20. 506
-22929
1. 136
1.118
4
dPldt -
.
I Jan ... S 1.740 400 - 26. 100 1. 138
July .. 5 1.440 - 300 -28.800 1.103
Jan ... 6 1,220 - 220
- 170
-33. 2 73 1. 15~
in which r is the constant ratio of two July .. 6 1,050 -37059 1.114
successive values of the loss ratio a. After Jan ... 7 918 - 13' -41. 769 1. >27
July .. 7 81 4 - 10 4 -47. 0 78 1.127
'ntegration this leads to: Jan ...
July ..
s
8
731
664 -
- 83
67
-5306.
-59.806
1. 127
1. 127
Jan ... 9 610 - 54 -67407 1.127
(I-r-I)
- 1.127
--
July .. 9 565 45 -75974
P = Poe a.logr an .. . 10 528 37 -85.631 I. 127
July .. 10 497 31 -96.5 14 1. 127
The simplest way to recognize this type The ratio of successive loss ratios is approximately
constant; average value, 1.127.
d decline, and to extrapolate it, is by Extrapolation until the tenth year. in the lower
half of the tabulation. by means of this average value
means of the loss-ratio tabulation. The
equation for the rate-cumulative curve, First Derivatives of Loss Ratios Form
which can be found by integration of Eq. an Arithmetic Series
25, is too complicated for practical use. The first derivatives of the loss ratios
As an example of the statistical treatment form an arithmetic series and the second
(,f production curves of this type, Table 4 derivatives are constant. S. T. Pirson 16
shows a loss-ratio tabulation of the family worked out the three possible mathematical
decline curve from a Wilcox sand pool in solutions for the rate-time equations, and
Oklahoma. As before, the per well produc- complete details may be found in his paper.
tion rates at equal time intervals are It has been found that these equatiom
t.abulated in column 3, the drop in produc- are generally too complicated for practical
tion rate in column 4 and the loss ratio usc. The simplest way to extrapolate a
in column 5. In this case the loss ratios curve, showing these characteristics, is by
form approximately a geometric series. means of the loss-ratio method.
This is evidenced by the fact that the
figures in column 6, which represent the Straight-line Relationship between
ratios of successive loss-ratio values, are Decline Percentage a.nd Time on Log-log
approximatel)C constant. Their average Paper
value is 1.127 and this figure was used This type of decline was discussed in a
for the extrapolation of column 6 in thc general way on pages 4 ~,) 5, and for
lower half of the table. The extrapolated more details we refer to the orieinal artick
values for the production rate were then by P. J. Jones. 21
found by reversing the process used in the Aside from the fact that there is a
upper part of the tabulation. straight-line relationship between decline
ANALYSIS OF DECUNE CURVES
and time on log-log paper, this type of If the loss ratio is COllstan t, the decline
curve can l':lso be extrapolated as a straight curve must be of the exponential type.
line by plotting. the log-log of the produc- If the loss-rf!.tio figures are not constant,
tion rate against the log of the time. but form an arithmetic series, the decline
Statistical extrapolation by means of the will be of the hyperbolic or' harmonic type,
loss-ratio method is possible but too com- depending on the value of the increment b.
plicated for practical use. If the loss-ratio figures indicate a geometric
series, the curve must be of the ra.tio-
TENTATIVE CLASSIFICATION OF
decline type.
DECLINE CURVES, BASED
On this table is shown also a summary
ON Loss RATIO
of the graphical and other methods that
To summarize the discussions in this can be used to extrapolate the different
paper, a tabulation was prepared (Table 5) types of curves.
showing the mathematical interrelationship
SUMMARY
between the commoner types of decline
curves. At the same time, it is shown how Most production-decline curves can be
these decline curves can be classified classified into a few simple types, which
according to the loss-ratio method. can be recognized by graphical, statistical
C = alP, - P)
:t:i~~_~:::;~~:g_logi;a~e:;:e;::::ng)_ - To=plica~ed ._-
Cumulative-rate re- - PH) -logP)
lationship (C, Pl (Straight line on special decline charts)
(Straight line on co- (Straight line on log- (Straight line on
ordinate paper log paper after shift- semilog paper)
ing)
.---.---.--~-.--------