Gas Reservoir Decline-Curve Analysis

Using Type Curves With Real Gas

Pseudopressure and Normalized Time
M.L. Fraim, SPE, Texas A&M U.
R.A. Wattenbarger, SPE, Texas A&M U.

Summary. A normalized time has been developed that linearizes the rate decline vs. normalized time for a gas reservoir
producing against constant well bore pressure during (external) boundary-dominated flow. This allows type-curve matching of the
exponential decline curve for a reservoir with any shape. Values of G, kl/l, and k~/rf> can be determined from this type-curve
match. It was determined that gas reservoir depletion plotted vs. actual time does not match exponential, harmonic, or hyperbolic
decline curves, and future performance can lead to overestimates of reserves and future rate.

Introduciion
Fetkovich 1.2 introduced the idea of applying log-log type-curve The viscosity and compressibility values in their integrated time
analysis to single-well analysis for both transient- and boundary- function are taken at the bottomhole 'flowing pressure (BHFP),
dominated flow periods. Boundary-dominated flow for constant- which is varying with time.
pressure production is similar to pseudosteady-state flow for This paper is concerned with the longer-term boundary-dominated
constant-rate production. His type-curve analysis was intended to flow of real gases in closed reservoirs. The objective is to improve
be a rapid way to estimate performance when a well is producing the use of Fetkovich's type curves for gas well analysis. A nor-
against a constant bottomhole pressure (BHP). From his type curves, malized time is introduced that applies to the viscosity and com-
future performance can be forecast along with estimates of oil in pressibility at the average reservoir pressure rather than the wellbore
place and ultimate recovery (Fig. I). pressure. This normalized time is used for boundary-dominated flow
For boundary-dominated flow, Fetkovich showed the Arps 3 fa- analysis, as well as for transient analysis. Only Darcy flow is con-
mily of curves with b as a parameter. The b=O case is for exponen- sidered.
tial decline of a liquid reservoir. Fetkovich used the curves with
values of b between 0 and 1 (hyperbolic and harmonic decline) for Real Gas Normalized Time
matching solution-gas-drive depletion and gas reservoir depletion. We will now show that the depletion of a closed real gas reservoir
Matching and extrapolating these curves is equivalent to using the can be expressed as an exponential decline (b=O) if a normalized
harmonic and hyperbolic declines with the usual semilog decline time is used. The normalized time is defined as
curves. He showed cases where these curves were useful.
Carter 4 developed another set of decline curves for boundary-
dominated flow that used a parameter, '1\, to represent variations
in the decline curves from real gas properties:
r" = r
o
(W,)j
J.l(p)c,(ji)
dr (4)

We will first develop the expression for exponential decline with

(J.lC~) i( Ppi - PPllj) the assumption that rock compressibility and water compressibil-
'1\ = ----"------'---'---"--- ..................... (I)
2[(P/Z)i -(p/z)lIjJ ity equal 0, for simplicity. We will then generalize to the total-
compressibility, c/' case in the Appendix.
where The basis of our derivation is a material-balance equation for a
real gas expansion in a closed reservoir:

p p =2 r
o
Pdp,
z(p)J.l(pJ
(ji/z)=(ji/z);(I-Gp/G)

This is the equation for the straight line p/z plot. We also use the
(5)

pseudosteady-state equation for a real gas in terms of Pp expressed

which can also be expressed as in the form of a productivity index equation as 7.8

Pp =2
o
r J.l(pJcg(p)
dlp/z(p J]. . (2) qg =Jg(pp -PPllj)' (6)

The expression for J g is

A value of '1\= 1.0 represents the liquid case, and values of '1\ < 1.0
represent the degree of gas property variation as a result of the
severity of the drawdown. . (7)
Agarwal S and Lee and Holditch 6 used a pseudotime to linea-
rize the transient analysis of gas wells with massive hydraulic frac-
tures. Their pseudotime was defined as

r
Eqs. 5 through 7 are the basis for our derivation. First, we
differentiate Eq. 5 to represent the pressure depletion with time:
fa = df. . (3)
o p,(PII:f )c,(P"f) -q (fJ/z)·
dCp/z) = g I dr (8)
Copyright 1987 Society of Petroleum Engineers G

SPE Formation Evaluation, December 1987 671

 10Ic-.,.,......,...,~Tn1--r-"r'"T~'"T-....-.-.-......TTT""-.......~,.,.,..,...---.~.,...,.
. .......--r--.-.....,"TTT1r-...,...,.,......~..."

••
TRANSIENT + DEPLETION

'O'~~~~~
EMPIRICAL TYPE CURVE SOLUTION
qdD=qltl. _ _
' _ :FDR b>O
• 6
t""", • l00.CCO qj
[1+ bOi1)b
J..

•
J qdD=~ • ..L :FOR b • 0
~
q, eDit

qdD EXPONENTIAL tdO:::Djl

0.1 COMMON TO ANAL YTICAL
B AND EMPIRICAL SOLUTIONS

0,01
B
6

Fig. 1-Composite of analytic and empirical type curves (after Fetkovich 1 ).

Then, we differentiate Eq. 6 to show how the rate changes with and substitute our expression for J, then our exponential decline
depletion: can be written as

2JI: _ In(qlqi) = ----------- (15)

dqg=J~dPf'= _ _ d(plz)
jJ.(p )cg(p )
(9)
V2(-A-
7rr\\.2
_I) 1/2In(2.248~A) CAr\\,
Eqs. 8 and 9 can be combined to form the following expression
The significance of Eqs. 13 and 15 is that we can express real
of rate change with time:
gas depletion exactly in terms of exponential decline when the nor-
malized time is used. Thus, all closed gas reservoirs with constant
-2J~ (rJIZ)iqg BHP fit the Fetkovich b =0 type curve for boundary-dominated flow.
dq~=---"-- df (10)
" jJ.(p )cl:(p) G These results can be generalized to include formation and liquid
compressibility by replacing C Ii with c l • Because saturations are in-
Eq. 10 applies to boundary-dominated flow. To follow the de- cluded in C I , Sg must be accounted for. Note that f" should be
velopment of Fetkovich's type curve, it is useful to extend Eq. 6 modified if formation compressibility is important. If the value of
(Pi-P)(cI+S\\,c\\,) is significant compared to 1.0 at any stage of
and thus Eq. 10 to f=O. The rate obtained from Eq. 5 at f=O is
depletion, then the definition of I" should be changed from Eq. 3
defined as qi, an extrapolated '"initial"' rate. With this extrapola-
to Eq. A-16. A development of this correction is shown in the Ap-
tion of Eq. 6 to f=O, we can integrate Eq. 10 from f=O to any
pendix.
later time, f, that has a flow rate. qg' and an average reservoir pres-
The assumption that Eq. 6 applies at f=O leads to an assumption
sure, p. Thus,
that jJ.(Pi) and C,(Pi) can be evaluated at initial conditions, even
though the gas properties corresponding to qi are based on a slight-
l
.-.2J~.
Jr'ldqg=rJ (P/z)i
df
,
, (11)
ly lower initial pressure. We have found that errors occurring from
this assumption occur only during the transient time and are not
'Ii qg 0 jJ.(p )CI:(p) G
significant if transient cumulative production is used in the material-
balance equation. Thus, gas rate vs. normalized time will trace the
which then becomes log-log type curve in both the transient- and boundary-dominated
flow regimes.
-2J~(plz)i l (Wg)i r
In(qlqi)=--"--- \ -c df (12) Type Curve Variables
G(WI:)i jJ.(p )cg(p) a Fetkovich's decline-curve dimensionless variables have been modi-
fied to include normalized time and the area of a general reservoir
The integral expression in Eq. 12 becomes our basis for defin-
shape. These variables use a value of 19.1785 for CA (sec Refs.
ing normalized time, f", in Eq. 4:
9 and 10) rather than 31.62 to conform with Fetkovich's type curve
construction. Thus,
-2J.,(jJlz)i
In(qlqi)= .' I" (13) 2.2485A )
G(Wg)i qgPsc T V2 In ( 2
19. I 785r\\. . (16)
qdD = - - - - - - - - - - : : . . . - -

Eq. 13 is in the form of an exponential decline. Notice that Eq.

13 has a constant coefficient when the f" variable is used.
and
If we express normalized time in terms of the usual dimension-
less time, 0.00633k~f"

IdD =
(</J w, l;r ,7·
----..-.:~--.:..-...:.:.----- ............... (17)
0.00633k~f"

(7r;} -I)
fD= ,'" (14)
(</JWg)ir,~. '/2 '/2 InC:::;A)

672 SPE Formation Evaluation, December 1987

 TABLE 1-RESERVOIR OATA FOR ALL CASES

Shape Case 1 Case 2 Case 3 Case 4

--~--

h, md 200 200 200 200

k, md 10 10 10 10
T, of 200 200 200 200
cP, fraction 0.30 0.30 0.30 0.30
c t , l/psi 3.0 x 10.- 6 3.0x10- 6 3.0x10- 6 3.0x10- 6
c w ' l/psi 3.0x10- 6 3.0x10- 6 3.0x10- 6 3.0x10- 6
Sg;, % 0.99 0.99 0.99 0.99
r w' ft 0.25 0.25 0.25 0.25
s 0 0 0 0
A, ft2 3.848x 10 7 3.848 x 10 7 3.848x 10 7 3.848x10 7
PI' psia 5,000 5,000 3,000 8,000
Pwt, psia 3,000 3,000 500 1,000
G, Sscf 61.51 61.51 40.52 81.63
Gas gravity 0.601 0.601 0.601 0.601
A 0.721 0.721 0.564 0.428

Note that in the transient time for a nonradial shape (C A <20). and
the decline curve will not trace the equivalcnt (r,'/r,l') transient stem

(~-r,~)ln(
on the Fetkovich curves. For shapes with CA <4, the rate will have
to decline to 40% of the "initial" boundary-dominated rate, qj. . ) 2.2485A )\
before a reasonable match can be obtained on a radial type curve. k,,,/A.=(tdD
, 'P
7r19.1785r,~
-------"---. .. . . . . . . (21)
To obtain normalized time, we must assume a value for original til MP 0.00633
gas in place, G. The first guess is found by matching the curve
of actual time vs. rate decline and calculating a value for G. This
value of G is then used to calculate normalized time and to match
this curve with the base fit b curve closest to the exponential curve. Reservoir Simulation Runs
A new value of G is calculated. This process is repeated until con- Numerical results were generated to test the validity of the
vergence is obtained for the exponential decline curve. From our normalized-time transformation. A three-dimensional (3D), three-
experience. it usually takes three tries for a closed-boundary phase, general-purpose simulator was used to simulate the real gas
reservoir. reservoirs. After each run was made, the results wcre processcd
The following properties can be determined from the match point to transform the time from actual time to normalized time. The re-
(MP): sults of both actual- and normalized-time plots were then compared.
The runs reported in this section werc either two-dimensional
(x,v) coordinatcs or one-dimensional radial coordinates. Some 3D
runs were made for analyzing layered reservoirs. They are report-
ed in another section.
Cases I through 4 were chosen to illustrate the use of the nor-
malized time. The data for these four cases are presentcd in Table
(Gp)ult= -
qg) (t,,)
( q,1D -
S~i
-, (19)
I. The reservoir and gas properties wcre held constant for all four
cases. The only properties that were varied for the four cases were
MP tdD MP A the initial reservoir pressurc, the constant BHFP. and the reser-
voir size and shape.
2.2485A ) The simulated results for these four cases arc shown in Figs. 2
PscT 1/2 In ( ? through 5. Each of these plots shows production rate ys. time. Each
_ _ _ _ _1_9_.1_7_8_5r_,-_1'_ , (20) figure shows three curves. The solid line represents thc analytic
solution for the exponential decline case (b=O) after boundary ef-
fects, and a value of r,)r" = 14.000 for the earlier (flat) part of the

10S.-=r- --, 10 5.-=r --,

"
Hatch Point Hatch Point
4.Jl) 0.51 4"1) 0.56
IJ J)7 0.108 ldl' 0.107

- Liquid Case, re/rw - 14,000 (b-O)

- Liquid Case. re!rw • 14,000 (b-O) 6 Actual Time
6. Actual Time o Pseudo-time
o Pseudo-time
). - 0.721
A • 0.721

-3

time, days
time, days

Fig. 3-Case 2-Actual time and pseudotime compared with

Fig. 2-Case 1-Actual time and pseudotime compared with liquid exponential decline case.
liquid exponential decline case.

SPE Formation Evaluation, December 1987 673

 106..,.- --,

Match Point
qdD= 0.72
(dD~ 0.67

Hatch Point
qdO= 1.54

IJD- 0.195
- Liquid Ca.e. re/rw • 14.000 (b-Q) - Liquid Case, re/rw • 14,000 (b-Q)
6. Actual Ti_ 6. Actual 'lime
D Puudo-tiM o Pseudo-time
). • 0.428

time, days time, days

Fig. 4-Case 3-Actual time and pseudotime compared with Fig. 5-Case 4-Actual time and pseudotime compared with
liquid exponential decline case. liquid exponential decline case.

curve, which represents transient behavior. The second curve in where

Figs, 2 through 5 is shown by triangles that represent the simulat-
ed results plotted as actual time in days. The third curve in Figs. C,=s~Cg+S,I'CII'+Cr ,., ' ,., (23)
2 through 5 is shown by squares that represent the time scale trans-
formed to normalized time. Step 2. Find the (pl;')j for each cumulative production point. G ,lj .
Figs. 2 through 5 show that the normalized-time transformation using the Ramagost and Farshad material-balance equation 13:
brings the boundary-dominated flow region exactly on the analytic
solutions. This confirms that the normalized-time cases can be an-
alyzed with liquid solutions given by Fetkovich's type curve for
h=O.
The simulated results plotted as actual time (triangles) do not fol-
low the Fetkovich type curves for any value of b. This means that ............ , (24)
a closed gas reservoir does not follow an exponential, harmonic,
or hyperbolic decline curve. If only early-time data are available
for a field case, the early boundary-dominated flow behavior might Step 3. Evaluate PI' from,the evenly spaced table.
appear to follow one of the h curves. The analysis based on that Step 4. Calculate f!,.t" from
curve will lead to incorrect results.
Note also in Figs. 2 through 5, as expected, that the normalized
time transformation does not affect the transient solutions. This is ........ , . , . , .. , , .... (25)
true because the average reservoir pressure does not decline sig-
nificantly during transient flow.
The spread between the two curves (the extent of the correction and
from actual time to normalized time) varies for the four cases shown.
The degree of deviation between the two curves is related to the (w, )i
pressure drawdown as compared with the initial pressure. The higher f!,.t"j=---f!,.tj, (26)
this fractional drawdown. the greater the deviation between the two (W', )1'
curves. The A parameter (Eq. I) is an index of the fractional draw-
down. For values of A approaching 1.0, there is little drawdown. where (W'I)" is the correct pressure-average value of (MC I ) over
and little deviation from gas properties between the two curves each time interval.
would be expected. As A decreases. the fractional drawdown is
greater and a greater deviation is observed as shown. for example. Step 5. Evaluate t"i from
in Figs. 4 and 5.
Case 2. shown in Fig. 3. shows the results for a rectangular reser-
voir with an aspect ratio of 2 : I with the well located in the middle t"i = ,-'
LJ At,,!"
'-l ....•.•.•....•.••....••••....•.•... (27)
j=1
of one end. The early transient data do not match the analytic solu-
tion for a radial case. but the data will match the analytic solution
Step 6. Plot qg, vs. t"i' find the match point. and calculate G. If
for a rectangular shape. Because average reservoir pressure is in-
it is significantly different from the assumed G. then repeat the
dependent of reservoir shape, the time transformation is valid for
procedure starting with Step 2.
any reservoir shape.
This procedure will assume radial flow into the well over the net
Now let us turn our attention to the type-curve analysis of the
pay thickness if the Fetkovich curve is used. This is a good assump-
curves shown in Figs. 2 through 5 for the actual-time curves and
tion for long-term rate performance of hydraulically fractured wells.
the normalized-time curves.
We assumed that rock and water properties are constant over the
pressure range and that gravity forces and Darcy flow around the
Procedure for Calculation of Normalized wellbore were negligible. For slowly changing rates, we have found
Time for a Gas Reservoir
that plotting q~/(Pi-Plli) vs. normalized time gives excellent re-
Step J. Make an evenly spaced table of (pi;,) vs. PI" The PI' func- sults. For drastic changes in Pllj. we recommend the superposi-
tion should not vary more than 3 % in a pressure interval. The fine tion principle.
spacing ensures that the finite-difference slopes evaluated from this If any of the reservoir properties vary. then the engineer will cal-
table are accurate. Then. culate average reservoir properties for that drainage area. We have
_ il' d( ph) found that we can use average volumetric compressibilities for water
pl'-2 J (22) and gas for volumetric aquifers with Py's less than four times the
o Mgc l gas reservoir PY.

674 SPE Formation Evaluation. December 1987

 TABLE 2-PVT PROPERTIES (abridged)

Gas Real Gas Storage

Pressure Viscosity Compressibility Compressibility Pseudopressure Pseudopressure
(psi) (cp) Factor (10- 3 psi- 1 ) (10 9 psi 2 /cp) (10 9 psi 2 /cp)
500 0.1272 0.9693 2.2870 0.01965 0.01969
1,000 0.1373 0.9388 1.0500 0.07487 0.07673
1,500 0.1479 0.9184 0.6734 0.16380 0.17820
2,000 0.1570 0.9102 .0.5299 0.28220 0.28780
2,500 0.1676 0.9105 0.3890 0.42550 0.43270
3,000 0.1786 0.9198 0.3046 0.58880 0.59730
3,500 0.1899 0.9364 0.2433 0.76770 0.77660
4,000 0.2018 0.9590 0.1978 0.95790 0.96630
4,500 0.2136 0.9861 0.1635 1.1560 1.1630
5,000 0.2258 1.0165 0.1372 1.3590 1.3630

Example 1 Iteration 1 (Normalized Time). For the match point. tdD = 1.24,
Simulation Case I-Circular Reservoir. In this example, we show qdD=0.511. G=64.3 Bscf [1.82X10 9 std m 3 ], t=I,OOO, and
hpw the rate vs. normalized time matches the type curve for a per- q= 10.000.
fect case. An abridged fluid property table is shown in Table 2. Iteration 2 (Normalized Time). For the match point, tdD = 1.29,
The actual (p/z) increment was 12 psi [83 kPaJ. If data storage is qdD=0.5I, G=62.0 Bscf [1.76X10 9 std m 3 ], t=I,OOO. and
a problem, slice the table into partitions such that each partition q= 10.000.
is accurately approximated with a quadratic or cubic polynomial. Iteration 3 (Normalized Time). For the match point. t dD = 1.30,
The rate-vs.-time data are shown in Table 3. The plot of the rate QdD=O.SI, G=61.50 Bscf [1.74X 10 9 std m 3 ), t= 1,000, and
vs. real time is shown in Fig. 2. The real-time match point is shown, Q= 10,0(}().
but the b used is not shown. Notice that the plot of the rates parallels The third iteration decline curve traces the b=O type curve. The
the exponential decline curve late in life. This phenomenon is caused engineer ~an now calculate the reservoir properties with a high level
by the fluid properties' becoming constant as the average reservoir of confidence:
pressure approaches we)lbore pressure.
Now we will discuss the iteration procedure for this gas well. 2.2485A )
Initial Match (Actual Time). For the initial time match. the rate PscT V21n ( - - - -
is plotted vs. time and match on the type curve. When matching kgh =..!!..L 19. 1785r; (20)
the actual time decline curve. use the b>O curves for the best ac- qdD 1.987xlO- 5 T sc (ppi-Ppwj)
curacy.
For the match point (see Fig. 2),
q(1D =0.52, t,1D = 1.08, q= 10,000, t= 1.000,
(2.2485)(3.848 X 10 7 ) ]
qg til 2(p/Z)iS~i 14.7(660)ln [
G=---- (18) = 10,000 (19.1785)(0.25 2 )
qdD tdD (p.("/ )i(Ppi -p/Jllj)
0.55 (2)(1.985 x 10- 5 )(520)
=( 10.000)( 1.000)
0.55 1.08 1
(2)(4.918.5)(0.99) ]
x----------
(1.3593 X 10 9 -5.888 X 10 8 )
x [ (3.099 x 10- 6 )(1.359 x 10 9 -5.888x 10 8 )

=70.6 Bscf [2x 10 9 std m 3 J. =199.7 md-ft [61 md'm],

TABLE 3-CASE 1 OATA TABLE 4-RESULTS OF MANUAL TYPE-CURVE MATCH

Production Rate Actual Time Third Iteration Pseudotime Data from Match Points Case 1 Case 2 Case 3 Case 4
(Mscf/D) (days) (days)
--- --- --- ---
t dO at t n = 1,000 1.30 1.18 0.75 2.22
30,400 0.052 0.050 qdO at g(On = 10,000 0.51 0.55 0.70 1.54
27,300 O.HlO 0.168 t dO at t= 10,000 1.08 1.07 0.67 1.95
24,600 0.729 0.676 qdO at q(t) = 1,000 0.52 0.56 0.72 1.54
22,200 3.38 3.13
20,100 17.7 16.3 Arps factor, b
18,100 ~3.2 75.3 tn 0 0 0 0
16,400 174 154 t 0.3 0.3 0.4 0.7
14,700 276 239 G, Bscf
13;000 389 330 61.49 61.41 40.2 81.3
431
tn
11,300 520 t 73.30 70.00 52.9 91.5
9,650 668 540 Model 61.51 61.51 40.5 81.6
7,990 880 690
6,330 1,150 872 kh, md-ft
4,750 1,490 1,090 tn 200 200 198 201
3,250 1,950 1,370 t 187 197 191 202
1,800 2,690 1,800 Model 200 200 200 200
921 3,550 2,280
472 4,410 2,750 k/¢, md
tn 32.5 32.4 32.9 31.9
t 27.0 28.9 29.4 27.8
Model 32.8 32.8 33.0 32.5

SPE Formation Evaluation, December 1987 675

 10° TABLE 5-FIELD EXAMPLE-WEST VIRGINIA;
WELL A RESERVOIR AND FLIJID PROPERTIES
10 4 Gas specific gravity 0.57 (air = 1.00)
w_
f-
<co.
N
<f>,% 6.0
CC-
",0 10-' Sw' % 35.0
",W
Wt:! Pi' psi 4,175
.J.J
z<c 11, ft 70
0::>
Uigj 10-' t, OF 160
aJZ rw' ft 0.354
::>CC
CiO p, psi 500
10-'

10-·
0 275 550 825 1100 1375 1750 TABLE 7-GAS RATE vs. TIME, WELl- A

NORMALIZED TIME Time Gas Rate

(days) (MscflD)
Fig. 6-CClse 1-normalized (plz) compared to dimensionless 9.36 2,382.8
rate. 31.57 2,339.8
58.12 1,755.3
89.97 1,832.2
119.02 1,542.6
TABLE 6-~ALCULATEO FLUID PROPERTlE~, WELL A 148.51 1,580.6
167.35 1,555.7
Pressure plz 1/l-tc g Pseudopressure 206.37 1,676.5
(psia) (psi) (psi/cp) (psia"/cp) 230.26 1,386.0
260.95 1,364.2
14.65 14.67 1,114 0.22324 x 10 4 287.52 1,297.9
208.75 212.19 15,508 0.33256 x 10 7 315.07 1,207.3
417.50 431.12 3Q,166 0.13372 x 10 8 349.02 1,194.8
626.25 656.41 43,964 0.30111 x 10 8 375.68 1,229.6
835.00 887.47 56,955 0.53457 x 10 8 406.99 1,156.6
1,043.7ti 1,123.47 69,238 0.83255 x 10 8 428.66 1,057.5
1,252.50 1,363.33 80,943 0.11928 x 10 9 466.85 1,040.5
1,461.25 1,605.76 92,219 0.16126x 10 9 502.28 1,132.8
1,!57O.OO 1,849.27 103,221 0.20885 x 10 9 :i54.18 915 .6
1,878.75 2,092.29 114,101 O,26166x 10 9 635.06 927.0
2,087.50 2,333.22 124,992 0.31929 x 10 9 689.14 712.1
2,296.25 2,570.54 136,004 0.~!3130x 10 9 728.47 834.0
2,505.00 2,802.90 147,216 0.44725 x 10? 759.19 692.9
2,713.75 3,029.18 158,682 0.51672 x 10 9 789.24 775.7
2,922.50 3,248.48 170,430 0.58929 x 10 9 865.26 682.1
3,131.25 3,460.18 162,468 0.66457 x 10 9 969.14 702.2
3,340.00 3,663.88 194,794 0.74221 x 10 9 1,049.65 679.3
3,548.75 3,859.39 207,399 0.82188x 10 9 1,056.52 614.0
3,757.50 4,046.66 220,266 0.90329 x 10 9 1,144.50 580.8
3,966.25 4,225.80 233,383 0.98618x 10 9 1,276.13 519.4
4,175.pO 4,396.99 246,733 0.10703 x 10 10 1,367.30 469.7
4,383.75 4,560.49 260,~P1 0.11555 x 10 10 1,453.01 575.4
1,559.88 410.8
1,662.55 353.1
1,799.91 359.4
and 1,892.29 409.2

kglc/>=
( ~ -r2)ln( 19_._1_78:..:5~r.!!:J;_
tdD _:....7f_ _w
2.2845A ) 2,110.90
2,152.70
346.0
254.0
............. (21) 2,331.22 249.9
tn 0.00633 2,465.40 276.6
2,498.23 220.9
2,583.15 239.1
2,692.33 196.4
2,818.95 184.7
7 2,930.14 210.3
3.848x 10 )
3,104.16 189.1
-----0.252
1.20 ( 7f

1,000 0.00633 type-curve match. We have fOlmd errors around 1% when we used
43.099xlO~6
superposition tp match the slowly changing or stair-step BHP cases.

Normalized (plz) VIS. Normalized Time

7
In [(2.2485)(3.848 X 10 ) ] Plotting nonnalized average reservoir preSS!lre vs. normalized time
(19.1785)(0.25 2 ) is a Qetter method to converge on G when the rate data are unsta-
X -----"-..:-:-~:::....-:...:.-.!........-=--
ble. The metpod is base" on the cumulative production vs. time.
First, Eq, 15 is integrated to find

=32.5 md vol/vol. ... (28)

In[l- Gp/(Gp )ultl = -(-A--)--(-2-.2-4-85-A-)·
The results for the pt4er cases are shown in Table 4. The llver- V2 -·-2 -I V2 In 2
age percentage error was less than I % for Case I with a manual 7fr w CAr w

676 SPE Formation Evaluation, December 1987

 TABLE a-NORMALIZED (Plz) Ys. NORMALIZED TIME, WELL A

Time Normalized Time Normalized (plz)

(days) (days) (days) (p,c g) ;!(p,c g) p
0.00000 0.00000 1.000000 1.000000
9.36030 9.31159 0.991142 0.994796
31.5730 31.0261 0.970311 0.977573
58.1210 56.9002 0.948723 0.974616
89.9798 87.1038 0.926027 0.948045
119.029 113.930 0.906559 0.923474
148.520 140.533 0.888269 0.902071
167.359 157.214 0.876536 0.885434
206.378 190.987 0.851493 0.865557
230.263 211.162 0.836967 0.844663
260.951 236.588 0.820209 0.828557
287.527 258.195 0.806160 0.813015
315.075 280.214 0.792455 0.799289
349.027 306.859 0.776260 0.784784
375.688 327.414 0.763425 0.770962
406.994 351.145 0.748591 0.758035
428.662 367.327 0.739064 0.746834
466.853 395.410 0.723153 0.735336
502.280 420.970 0.707863 0.721470
554.185 457.596 0.686750 0.705635
635.069 512.933 0.657155 0.684154
689.146 548.877 0.639553 0.664683
728.472 574.545 0.627479 0.652711
759.190 594.334 0.618164 0.644205
789.248 613.484 0.609398 0.637091
865.269 661.008 0.587391 0.625149
969.145 723.947 0.558837 0.605910
1,049.66 771.214 0.536749 0.587074
1,056.52 775.186 0.534985 0.578340
1,144.51 825.351 0.514110 0.570152
1,276.14 898.066 . 0.485351 0.552436
1,367.31 946.937 0.467443 0.536045
1,453.02 991.824 0.449654 0.523677
1,559.88 1,046.37 0.428724 0.510403
1,662.56 1,097.50 0.413146 0.497996
1,799.92 1,164.28 0.393709 0.486187
1,892.29 1,208.15 0.379610 0.474938
2,110.90 1,308.56 0.346825 0.459298
2,152.71 1,327.24 0.341845 0.446736
2,331.22 1,405.63 0.323982 0.439150
2,465.40 1,463.13 0.309952 0.428560
2,498.23 1,477.02 0.306708 0.422821
2,583.16 1,512.61 0.298949 0.419164
2,692.34 1,557.75 0.289506 0.413440
2,818.95 1,609.30 0.279921 0.407100
2,930.15 1,653.89 0.271197 0.400988
3,104.16 1,722.35 0.257395 0.393448

From the material-balance equation, Eq. 24, we can derive the

relation for [l-Gpl(Gp)ult]:

S
3
a
w
N
:J
<{
::;;
where a:
oz 1=6

10-1L - _...._...J.._ _~ _......._ - . L . . _ - - I _ - - l

o 500 1000 1500 2000 2500 3000 3500
and
TIME. NORMALIZED TIME - DAYS

Fig. 7-Field case-sixth iteration (plz)n for actual and nor-

malized time.

The right side of Eq. 29 is called normalized and will be denoted

(jjlz)/l" If the press/Jre drawdown is not too large, then Cllf and
Cm . can be approximated by setting them equal to I.

SPE Formation Evaluation, December 1987 677

 TABLE 9-NORMALIZED TIME vs. to, qg, qo, Qo. Well A

Normalized Time
(days) to qg qo Qo
9.31159 228,652.0 2,382.857 0.342112 78,790.0
31.0261 761,867.0 2,339.839 0.335936 264,077.0
56.9002 0.139722 x 10 7 1,755.315 0.252015 456,101.0
87.1038 0.213889 x 10 7 1,832.293 0.263066 657,978.0
113.930 0.279764 x 10 7 1,542.693 0.221488 831,145.0
140.533 0.345087 x 10 7 1,580.630 0.226935 993,829.0
157.214 0.386049 x 10 7 1,555.702 0.223356 0.109819x 10 7
190.987 0.468981 x 10 7 1,676.541 0.240705 0.132095 x 10 7
211.162 0.518522 x 10 7 1,386.024 0.198995 0.145015 x 10 7
236.588 0.580958 x 10 7 1,364.227 0.195865 0.159922 x 10 7
258.195 0.634015 x 10 7 1,297.929 0.186347 0.172418x 10 7
280.214 0.688083 x 10 7 1,207.334 0.173340 0.184608 x 10 7
306.859 0.753512 x 10 7 1,194.827 0.171544 0.199013x 10 7
327.414 0.803986 x 10 7 1,229.633 0.176541 0.210430 x 10 7
351.145 0.862259 x 10 7 1,156.633 0.166061 0.223625 x 10 7
367.327 0.901996 x 10 7 1,057.562 0.151837 0.232099 x 10 7
395.410 0.970956 x 10 7 1,040.594 0.149401 0.246252 x 10 7
420.970 0.103372 x 10 8 1,132.847 0.162645 0.259852 x 10 7
457.596 0.112366 x 10 8 915.629 0.131459 0.278632 x 10 7
512.933 0.125954 x 10 8 927.012 0.133093 0.304956 x 10 7
548.877 0.134780 x 10 8 712.191 0.102251 0.320613 x 10 7
574.545 0.141083x 10 8 834.010 0.119741 0.331353x 10 7
594.334 0.145943 x 10 8 692.993 0.994946 x 10 -1 0.339638 x 10 7
613.484 0.150645 x 10 8 775.724 0.111373 0.347435 x 10 7
661.008 0.162315x 10 8 682.126 0.979345 x 10 -1 0.367010 x 10 7
723.947 0.177770 x 10 8 702.224 0.100820 0.392409 x 10 7
771.214 0.189377 x 10 8 679.369 0.975386 x 10- 1 0.412056 x 10 7
775.186 0.190352 x 10 8 614.004 0.881540 x 10- 1 0.413624 x 10 7
825.351 0.202670 x 10 8 580.809 0.833881 x 10- 1 0.432192 x 10 7
898.066 0.220526 x 10 8 519.489 0.745843 x 10- 1 0.457773 x 10 7
946.937 0.232527 x 10 8 469.746 0.674425 x 10- 1 0.473702 x 10 7
991.824 0.243549 x 10 8 575.430 0.826159 x 10- 1 0.489525 x 10 7
1,046.37 0.256942 x 10 8 410.884 0.589916 x 10- 1 0.508142 x 10 7
1,097.50 0.269498 x 10 8 353.190 0.507083 x 10- 1 0.521998 x 10 7
1,164.28 0.285897 x 10 8 359.430 0.516043 x 10- 1 0.539287 x 10 7
1,208.15 0.296670 x 10 8 409.223 0.587531 x 10- 1 0.551828 x 10 7
1,308.5p 0.321326 x 10 8 346.006 0.496769 x 10- 1 0.580990 x 10 7
.1,327.24 0.325912 x 10 8 254.003 0.364678 x 10- 1 0.585420 x 10 7
1,405.63 0.345162 x 10 8 249.916 0.358811 x 10- 1 0.601308 x 10 7
1,463.13 0.359282 x 10 8 276.662 0.397210 x 10- 1 0.613788 x 10 7
1,477.02 0.362691 x 10 8 220.960 0.317237 x 10- 1 0.616674 x 10 7
1,512.61 0.371432 x 10 8 239.122 0.343314 x 10- 1 0.623575 x 10 7
1,557.75 0.382516 x 10 8 196.469 0.282075 x 10 -1 0.631975 x 10 7
1,609.30 0.395174 x 10 8 184.754 0.265256 x 10- 1 0.640500 x 10 7
1,653.89 0.406123 x 10 8 210.328 0.301974 x 10- 1 0.648260 x 10 7
1,722.35 0.422935 x 10 8 189.105 0.271502 x 10- 1 0.660537 x 10 7

Substituting Eq. 29 into Eq. 28, we find BHP. The least-squares best-fit straight line of the normalized (plzl
data had slope error of less than 2 % and an intercept of 1.015.

In [(P/ZliC(/\'-~PlzlwrCII:tJ Calculated Results

(plzli -(p/zlwrCwr . Table 4 shows four basic closed-boundary reservoir cases that vary
in shape and pressure. In each case, the normalized-time correc-
tion improves the accuracy of the type curves. Cases I and 3 are
radial cases where the pressure was varied. Cases I and 2 have

( 7rrA)I (2.2485A)' a very accurate match because the well's decline curve can be
Vl --2 - 112 In 2 matched in the transient region and in the decline region. Case 4
ll · CArli' is a square with a very high drawdown. The square also behaves
which can also be expressed as like the radial case. Therefore. both regions were used to match
the decline curve. Case 2 is a 2 : I rectangle, and its transient region
In[(plZ)Il] =tdD' (30) cannot be used in the decline-curve match because only circular
shapes are given. For shapes with CA <4, the rate has to decline
Thus. we can plot normalized (Plz) vs. normalized time and get to 40% of the "initial" boundary-dominated flow rate, qi' before
a semilog straight line with an intercept approximately equal to I. a good match can be obtained with the radial-flow type curve.
For most of the field cases we matched. the intercept ranged from
1.02 to 0.95. We used Case I data to test the method. except we Example 2
added a random 5% of the current average reservoir pressure drop Field Case-Low-Permeability Fractured Gas Well. The data for
to the BHP to represent fluid loading in the gas well. The plot of this well come from Ref. 2. The reservoir data are shown in Table
the dimensionless rate and normalized (plz) vs. normalized time 5, the calculated fluid properties are shown in Table 6, and the digi-
is shown in Fig. 6. Note that the normalized (Plz) is almost a straight tized raw rate-time data are shown in Table 7. This well was chos-
line. and a unique match can be made on G and lifetime average en because we wanted an independent actual time match of the data.

678 SPE Formation Evaluation, December 1987

 102~ ....L. ..L... --l ..J

10° 10'

TIME, NORMALIZED TIME - DAYS NORMALIZED TIME

Fig. a-Field case-gas rate for actual and normalized time. Fig. 9-Field case-final match of gas rate, r .Ir w = 24.

This is a hydraulically fractured gas well completed in the Dnon- accurate match. The gas in place calculated from the rate-time match
daga chert in West Virginia. The well was initially produced for should agree with the (plz) match: if they do not agree, the rate-
200 days and then shut in for a 106-day buildup in an attempt to time curve is moved up or down. For large differences, the well-
determine average reservoir pressure. With this average reservoir bore pressure should be changed and the process should be repeat-
pressure, they estimated G to be 3.36 Bscf [0.09 X 10 9 std m 3 ] ed. For manual type-curve matching, we found an rJr". stem equal
using the material-balance equation. The well also has a problem to 20. For a computer-generated type curve, we found an relrll'
of water loading up the wellbore. The time, normalized time, nor- stem equal to 25. Finally, we used a linear optimization program
malized (Plz) and pressure-average (W, )/(j.tc,)p data are shown in that chose an r,Jr". stem equal t024, which is shown in Fig. 9.
Table 8. It took six iterations to converge on G using the normal- After a match is obtained, the best-fit r,'/r". stem is recorded and
ized (plz) plot because we had to increase the line pressure from the permeability and porosity are calculated. With the porosity, satu-
500 to 710 psi [3.4 to 4.9 MPa] to represent the lifetime average rations, and original gas in place, the external radius can be calcu-
BHP. With the 71O-psi [4.9-MPa] BHP, we almost had a perfect lated. Finally, with the external radius and the r,'/r". ratio. the
straight line on the normalized (Plz) vs. normalized-time plot, as effective wellbore radius can be calculated. These results are shown
shown in Fig. 7. in Table 10 and are compared with the best-fit results of Refs. I
In the next step, we plot rate vs. the normalized time that was and 2. Note that they used the gas in place from the buildup test
generated during the (Plz) match. Normalized time, dimension- to help estimate the r,Jrl\' stem. The main advantage of this method
less time, gas rate, dimensionless rate, and dimensionless cumula- is that we are very confident in the future performance prediction
tive production are shown in Table 9. In Fig. 8, we see a large of the Fetkovich curve using normalized time. With the actual-time
range of scatter for the rate data that is caused by the liquid load- method, you must pick the best-fit b curve to extrapolate the future
ing, occasional shut-ins, and blows to the atmosphere. Note that production, which Fetkovich and we have shown to be inaccurate.
two to three data points were skipped between each plotted data In an effort to determine the effect of two-phase now in the frac-
point. With this much scatter in the data, it is easy to get a small ture and the effect of liquid loading the wellbore, we compared the
range of reasonable fits to the type curve. Thus, we calculate a small declines of the normalized (ph) and dimensionless rate, as shown
range of gas in place for the reservoir and we see a small range in Fig. 10. From our experience, the (plz) plot is a good approxi-
of relrl\' ratios that could be used in the skin calculation. But if we mation of the rate decline of a well at perfect conditions. From Ta-
use the normalized time from the (ph) match, we fix the time axis ble 10 we see that the dimensionless-rate plot declines about 1.75
and vary the rate axis for the best fit that gives a more unique and times faster than the (plz) plot. Thus. we have an approximate
method of calculating reserves for a well at completely unloaded
conditions and at current conditions. With these incremental reserves
we can determine whether it is profitable to install equipment or
TABLE 10-AESULTS OF NORMALIZED TIME MATCH
AND ACTUAL TIME MATCH, WELL A
to replace tubing to keep the well bore unloaded.

Normalized (Plz)
Normalized Actual 100 0 0 0 0
Time Time 0 0 (p/z)
0 0
0
Slope of In(j)/z)n vs. t n - 0.343743 x 10- 3 o DO
0
0
0
V-intercept 0.989647 w_ o
0
0
I-N 0
Estimated G, Mscf 0.303451 x 10 7 ~a 0
0
Produced G, Mscf 0.251800x 10 7 (1)0 o 0 0

Pi' psi 4,175 4,175 ffi~

..J..J 10~
Q
0
0
z<c qo 0
Pwf' pSi 710 500 0 2 a
-a:
(1)0 0
Residual 0.999833 0
ffiz 0 0
2a:
Gas Rate cO
Effective r .Ir w 24 20
Estimated G, Mscf 0.303451 x10 7 0.360x 10 7
s -5.07873 -' 5.17 10.2
0 275 550 825 1100 1375 1750
k, md 0.07678 0.070
Slope of In(rate) vs. t n - 0.601474 x 10- 3
NORMALIZED TIME
Residual 0.983423
r., ft 1,128.5 1,242
Fig. 10-Field case-decline comparison of (plz) and dimen-
rw' ft 47.2 62.1
sionless rate.

SPE Fonnation Evaluation, December 1987 679

 1.11.-.------------------,., 1. « - r - - - - - - - - - - - - - - - - - - - "
~ GaB Gravity
Temperatura
0. 1. (Top) 300+ 0, 0. 1. (Top)
2.
0.6
0.7
;: (Bott01ll) ~~~; 3.
4. (Bottoa)
0.8
0.9

0. 0.

1 000

Fig. 11-Variation of the t n integrand with reservoir temper- Fig. 12-Variation of the t n integrand with gas gravity; reser-
=
ature; gas gravity 0.6, Pi 10,000 psia. = =
voir temperature 200 Q F, Pi 10,000 psia. =

Other Cases 5. Ifformation compressibility is important, Eq. A-16 should be

Crossflow and layered-reservoir cases were investigated, and we used to evaluate normalized time rather than Eq. 3.
found that gas behaves like a liquid with normalized time. We came
to the same conclusion as Russell" in that a reservoir with cross- Nomenclature
flow can behave like a homogeneous reservoir. We also came to
A = area, ft 2 [m 2 ]
the same conclusion as Fetkovich I that for a layered reservoir, the
b = decline-curve parameter
decline for the reservoir is the summation of the decline curves for
(j = rock compressibility, l/psi [1/kPa]
each layer. Each layer has its own normalized time. Well bore pres-
sure changes were also investigated. Normalized time is reset back cli = gas compressibility, l/psi [l/kPa]
to zero for each pressure change,' and the material-balance equa- cr = total compressibility, l/psi [l/kPa]
tion still uses total cumulative production. C II ' = water compressibility, l/psi [1/kPa]
CA = Deitz shape factor
Discussion G = original gas in place, Mscf [std m 3 ]
A time transformation is introduced to account for the variation Gp = cumulative gas production, Mscf [std m 3 ]
of fluid properties as the average reservoir pressure depletes. This (Gp)ult = ultima!e cumulative gas production at Pwf' Mscf
normalized time is different from the pseudotime used by [std m 3 ]
Agarwal 5 and Lee and Holditch 6 because the fluid properties, /l h = formation thickness, ft [m]
and c r ' are evaluated at average reservoir pressure in the time in- J = oil productivity index, sefID-psi [std m 3 /d' kPa]
tegration rather than at the well bore pressure. We have also chos- J Ii = gas productivity index, MscfID-psi 2 -cp
en to use a /lC r ratio as the integrand so that the normalized time
[std m 3 /d' kPa 2 • Pa' s]
has the dimensions of time (days).
We have shown that the transformation of simulation results seems
k = formation permeability, md
to linearize the pseudosteady-state results exactly without affect- N = oil in place, STB [stock-tank m 3 ]
ing the transient results. Through this linearization, all the simu-· P = pressure, psia [kPa]
lated results fit the liquid-acting exponential decline case (b=O). P= average reservoir pressure, psia [kPa]
None of the simulated results, plotted as actual time or normalized Pi = initial reservoir pressure, psia [kPa]
time, fit any of the harmonic or hyperbolic decline curves (b>O). PI' = real gas pseudopressure, psi 2 /cp [kPa 2 /Pa's]
By use of the normalized-time plots, it is possible to determine PI' = storage pseudopressure, psi 2 /cp [kPa 2 /Pa's]
G, kh, and k/cjJ from type-curve analysis. Psc = pressure at standard conditions, psia [kPa]
The method has the disadvantage of requiring an estimate of G Pllf = wellbore flowing pressure, psia [kPa]
before the normalized time can be calculated. However, we have
found that two or three iterations are all that is required in estimat-
t:..p = Pi-P
ing G and then calculating the same G from the type-curve match. (p/z)n = normalized p/z (Eq. 30)
Figs. 11 and 12 show how (W1)i/(WIi)p varies with pressure for q,lD = decline-curve dimensionless rate (Eq. 16)
various gas gravities and reservoir temperatures. The curves are qD = dimensionless rate =795,746 qli/[kh(ppi-Ppllf)]
almost linear, making the integration of normalized time fairly easy qli = surface flow rate, MscfID [std m 3 /d]
and accurate. qi = initial rate, applying Eq. 5 at t=O
QD = dimensionless cumulative production
Conclusions =5,015. 8Gp /[h( cjJWr )i r ,7·(Ppi -Ppllf)]
1.. None of the cases of a gas reservoir producing at constant flow- rc = equivalent outer-boundary radius, ft [m]
ing pressure matches any of Fetkovitch's exponential, harmonic, r ll , = wellbore radius, ft [m]
or hyperbolic decline curves when actual time is used. r:" = effective well bore radius, rll,e-·I·, ft 1m]
2. When transformed to normalized time, the boundary- s = skin
dominated-flow curves match the type curve for liquid exponential S = saturation
depletion exactly.
t = time, days
. 3. The normalized-time transformation does not have any sig-
tdD = decline-curve dimensionless time (Eq. 17)
nificant effect on transient behavior.
4. Correct values of G, kh, and k/cjJ can be determined accurate- tD = dimensionless time (Eq. 14)
ly with normalized time type-curve analysis. This applies for any tn = normalized time, days
reservoir shape if sufficient boundary-dominated flow data are T = reservoir temperature, oR [K]
available. T,c = temperature at standard conditions, OF [0C]

680 SPE Formation Evaluation, December 1987

 VI' = reservoir PY, Mcf [m 3 ] Differentiating Eq. A-2 gives
A = defined by Eq. I
/1. = gas viscosity, cp [Pa' s I dq" =J,,(B"iIB,,)O +c,,)dp (A-3)
<p = porosity, fraction
If c" is ignored in Eq. A-3, then we have
Subscripts
dqo =J,,(B,,;lB,,)dp (A-4)
dD = dimensionless decline
D = dimensionless Now, with any fluid flowing, we have the general expression for
g = gas dp:
i = initial conditions
.i = counter -q"B o
() = oil dp = --dr (A-5)
sc = standard conditions Vpc 1
r = total
w = water
When Eq. A-5 is substituted into Eq. A-4, then
It! = well bore flowing conditions

Acknowledgments
We acknowledge the helpful suggestions made by W.J. Lee and
For this case, when C1 =C", cr=O, and S"i = 1.0, then we can ex-
T.A. Blasingame during this work.
press the final result in terms of oil in place, N,

References
1. Fetkovich, M.J.: "Decline Curve Analysis Using Type Curves," JPT
(June 1980) 1065-77.
r
'/'I q"
o
dq" = -J fdr. .
Nc,,( pss
(A-7)

2. Fetkovich, MJ.: "Decline-Curve Analysis Using Type Curves-Case

Histories," SPEFE (Dec. 1987) 637-56. This is then integrated to give
3. Arps,J.J.: "Analysis of Decline Curves," Trans., AIME(l945) 160,
228-47.
4. Carter, R.D.: "Type Curves for Finite Radial and Linear Gas-Flow
-J"
In(q,/q"i)=--r2' (A-8)
Systems: Constant-TerminaJ-Pressure Case," SPEJ (Oct. 1985) 719-28. Nc"
5. Agarwal. R.G.: "Real Gas Pseudo-Timc-A New Function f{)r Pres-
sure Buildup Analysis of MHF Gas Wells, " paper SPE 8279 present-
which is in the form of exponential decline because the coefficient
ed at the 1979 SPE Annual Technical Conference and Exhibition. Las
Vegas. Sept. 23-26. of r is constant.
6. Lee. W.J. and Holditch, S.A.: "Application of Pseudotime to Buildup When crand C II ' are considered and c" is not constant, this deri-
Test Analysis of Low-Permeability Gas Wells With L-ong-Duration Well- vation proceeds in the same manner except for the last step when
bore Storage Distortion." JPT (Dec. 1982) 2877-87. VI' is expressed in terms of N. This modification gives the follow-
7. AI-Hussainy. R., Ramey, H.J. Jr" and Crawford. P.B.: "The Flow ing result:
of Real Gases Through Porous Media," JPT (May 1966) 624-36;
Trans .. AIME. 237.
8. Russell. D.G. and Prats. M.: 'The Practical Aspects oflnterlayer Cross- -J"S"i rt
In(qlqi)=--J - - - - d r (A-9)
flow," JPT (June 1962) 589-94.
N 0 c(O - !!.PCj)
9. Ehlig-Economides, C.A. and Ramey. HJ. Jr.: "Transient Rate Decline
Analysis for Wells Produced at a Constant Pressure." SPEJ (Feb. 1981)
98-104. Eq. A-9 shows that an integral time transformation is required
10. Lee: WJ.: Well Testing, Textbook Series. SPE, Richardson. TX (1982)
(even for the liquid case) to get an exponential decline if the oil
1. 9-11. 76-88.
II. Earlougher. R.C. Jr.: Advances in Well Test Analysis. Henry L. Do- compressibility changes significantly.
herty Series, SPE, Richardson, TX (1977.) S. Now, let's go to the gas case. Eq. A-2 is replaced by
12. Wattenbarger, R.A. and Ramey. H.J. Jr.: "Gas Well Testing With Tur-
bulence. Damage and Wellbore Storage," JPT(Aug. 1968) 99-107: qg=Jg(pp-Ppwj)' (A-IO)
Trans. AI ME. 243.
13. Ramagost, B.P. and Farshljd, F.F.: "P/Z Abnormally Pressured Gas
Reservoirs." paper SPE 10125 presented at the 1981 SPE Annual Tech- which is differentiated to give
nical Conference and Exhibition, San Antonio, Oct. 5-7.

2J~p _
Ap.,endix dqg=--qp (A-II)
The exponential decline case assumes that the exponent is constant, Z/1.
or that q vs. r is exactly a straight Iin~ when plotted on semi log
graph paper. The following investigates when this is true and when Because Eq. A-5 applies to gas production as well as liquid pro-
it is an approxi!llation. d4ction, it is substituted into Eq. A-II to give .
Exponential declil]e seems to apply to the liquid case when C is
constant. At initial conditions (the beginning of pseudosteady state),
the initial rate is given qy d q g-2J~p
= - - qg-- - (Z)(PscT)
- - dr, (A-12)
ZJ.!. Vpc 1 P T",

q"i =J,,(Pi -Pill)' (A-I)

which simplifies to
At later times the production rate is given by
dqg -2J g (P'\CT)
-=-- -- dr (A-l3)
q,,=J,,(B,,;lB,,)(p -PI~r) (A-2) qg Vp /1.C 1 T",

SPE Formation Evaluation, December 1987 681

When VI' is expressed in terms of G, then we get the following Note that this relation assumes that gas is the only mobile phase
form, where t1P=Pi-P: and that the relative permeability for gas is constant.
Step 2. Find the (plz) for each cumulative production point, Gp ,
using Ramagost's material-balance equation:

rp ------''-d(plz)
J
c r(l-t1PCj) --
=(plz);( CpIC) (A-19)
................................. (A-14) o SgiCg

Step 3. Evaluate Pp from the evenly spaced table.

which can then be integrated into the final form Step 4. Calculate t1t ll from

- [(PIZ.)j-1 -(plz)j ]
(/J.c r )"1'-2 _ _ ................... (A-20)
PPj-1 -Ppj

................................. (A-15) and

(!J.C r) i
From Eq. A-15, it can be seen that the formation and water com- t1t ll j=---t1tj' (A-21)
pressibility terms enter the integral transformation. Thus, a more (/J.c t ),,1'
.I
general expression for pseudotime is
where (/J.ct)av is the pressure-averaged (/J.gc t )(l-t1pcj) over each

til = r_ (~Cr)idt ,
o /J.(p )c(p )(l-t1PCj) .
(A-16)
time interval.
Step 5. Evaluate In; from

which simplifies to Eq. 4 when formation and water compressibil-

ity or drawdown are small.
The pseudotime expressed by Eq. A-16 seems to be general and
could be applied to any type curve that has been developed for ex- Step 6. Plot qg; vs. In;.
ponential decline.
51 Metric Conversion Factors
Procedure for Calculating General Pseudotime
cp x I.Q* E-03 Pa's
for a Gas Reservoir,
ft x 3.048* E-Ol m
Step I. Make an evenly spaced table of (plz) vs. PI" ft 2 x 9.290304* E-02 m2
where
ft3 x 2.831 685 E-02 m3

Pp=
o
r 2d(plz)
/J.gC r(l-t1pc;)
(A-17)
of (OF-32)/1.8
psi x 6.894757 E+OO
°C
kPa

·Conversion factor is eXilct. SPEFE

and Original SPE manuscript received for review Sept. 22, 1985. Paper accepted for pUblica-
tion March 19, 1987. Revised manuscript received March 11, 1987. Paper (SPE 14238)
first presented at the 1985 SPE Annual Technical Conference and Exhibition held in Las
Cr=SgCg+S",C",+Cr (A-18) Vegas, Sept. 22-25.

682 SPE Formation Evaluation, December 1987