You are on page 1of 10

Inflow Performance Relationships

for SolutionGasDrive Reservoirs


R.G. Camacho-Y., * SPE, and Rajagopa. Raghavan, * * SPE, U. of Tulsa

Summary. In this theoretical study, a numerical model was used to examine the influence of pressure level and skin factor on the
inflow performance relationships (lPR's) of wells producing under solution-gas-drive systems. Examination of the synthetic deliverability
curves suggests that the exponent of the deliverability curve is a function of time and that the exponent is usually greater than unity.
The implication of this observation to field data is discussed. The accuracy of procedures given in the literature to predict oilwell delivera-
bilities is also examined. It is shown that these methods can be used to predict future performance provided that the exponent of the
deliverability curve is known and that extrapolations over large time ranges are avoided. If single-point tests are used to predict future
performance (such tests assume that the exponent of the deliverability curve is constant), then errors in predictions will be minimized
if test conditions correspond closely to actual producing conditions. Under these circumstances, errors in rate predictions will be directly
influenced by the magnitude of the error that results from assuming that the exponent of the deliverability curve is known. Although
relative permeability and fluid property data are required, the Muskat material-balance equation and the assumption that GOR is indepen-
dent of distance can be used to predict future production rates. This method avoids problems associated with other methods in the litera-
ture and always yields reliable results.
New methods to modify the IPR curve to incorporate changes in skin factor are presented. A new flow-efficiency definition based
on the structure of the deliverability equations for solution-gas-drive reservoirs is proposed. This definition avoids problems that result
when the currently available methods are applied to heavily stimulated wells.

Introduction
This paper addresses three concerns pertinent to IPR's under can be eliminated. Unfortunately, the curves developed by Stand-
solution-gas drive. First, by means of numerical simulations, we ing need to be corrected for all values of s. Incidentally, the defini-
examine the IPR when a skin region exists around the wellbore. tion we present is consistent with the definition of Ef used for gas
This phase of our work is intended to address the question of whether wells. A new chart to modify the IPR curve based on the appropri-
the generalized form of the IPR curve is governed by the existence ate deliverability equation (Vogel or Fetkovich) is presented.
of a skin region. Although the skin region has a dominant effect
on the well response, the influence of a skin region has been ig- Numerical Model
nored for all practical purposes. (The seminal work of Vogel, 1
We consider a homogeneous, closed, circular reservoir with the
however, briefly addresses the influence of a skin region.)
well located at the center. The well is capable of producing either
Second, we address the prediction of future IPR's from test data.
at constant oil rate or at constant wellbore pressure. Gravity ef-
Many methods have been proposed in the literature 2- 6 to predict
fects are considered to be negligible. The reservoir initially is
future performance. The methods of Standing 2 and Fetkovich 3 are
assumed to be at the bubblepoint. An annular region concentric with
based on empirical observations of solution-gas-drive reservoir per-
the wellbore, with a permeability different from formation perme-
formance. All methods presented in the literature assume that the
ability, is used to incorporate the influence of a skin region.
mobility function, kro/(JJ.oBo), is a linear function of pressure, p,
Several sets of relative permeability and fluid properties were
where kro is the relative permeability to oil, JJ.o is the viscosity of
used in this study. Figs. I and 2 show the fluid properties for two
oil, and Bo is the FVF of oil. Although the methods of Standing
of the data sets we considered, and Fig. 3 presents the relative per-
and Fetkovich can be justified on the basis of this assumption, this
meability data used in this study. Table 1 presents other informa-
requirement (mobility function is a linear function of pressure) is
tion regarding reservoir properties used in this work. The results
not a necessary condition for the methods given in Refs. 2 and 3
given in this work, however, can be applied over a wider range
(or for that matter, in the other references cited above) to be valid.
of conditions.
We examine the validity of the procedures suggested by Standing
A finite-difference model was used to obtain the results present-
and Fetkovich by means of simulations for a wide range of producing
ed here. The procedures we followed to ensure that the results are
conditions. We show that the success of these methods rests primar-
accurate are similar to those presented in Refs. 10 and 11 and will
ily on the fact that the deliverability equations suggested by Vogel
not be considered here.
and Fetkovich incorporate features that reflect the nonlinear aspects
Non-Darcy flow effects are not considered. The influence of non-
of flow in the reservoir and, more importantly, are calibrated by
Darcy flow for wells producing by solution-gas drive represents
measurements. Their procedures attempt to predict only the
a formidable analytical problem. Many fundamental considerations
endpoint-the final answer. We demonstrate this point by compar-
are involved, which are beyond the scope of this work.
ing mobility-function profiles predicted by methods given in the
literature.
Background
The third part of this paper examines procedures to modify the
IPR curve to incorporate the influence of the skin factor, s. Brown 7 From numerical simulations, Vogel showed that the deliverability
shows that Standing's8 method to modify the IPR curve can result curve for a well producing under solution-gas drive is given by
in inconsistent predictions if s is negative. Whitson 9 says that qolqo.max = 1-0.2(Pw/p)-0.8(P'iv/p2). . ............. (1)
Standing'S method leads to nonphysical results if s<O. Whitson's
observations suggest that Standing'S procedure should be incorrect Here qo.max is the maximum possible flow rate (rate correspond-
if s>O. Here, we show that inconsistent or nonphysical predictions ing to Pwf=O), and qo is the flow rate corresponding to average
are a result of the definition of flow efficiency used by Standing. pressure, p, and wellbore flowing pressure, pwf. In 1973, on the
Standing defined flow efficiency on the basis of liquid flow and basis of field experiments, Fetkovich suggested that the delivera-
not on the quadratic form suggested by Vogel's equation. In this bility curve for solution-gas drive systems is given by the follow-
work, we redefine flow efficiency, Ef , on the basis of the quad- ing relation.
ratic forms of the deliverability equations suggested by Vogell and qo =J~(t)(ji2 _p'i:r)n . ............................... (2)
Fetkovich 3 and show that the inconsistencies mentioned by Brown
Here J~(t) is the productivity index for a well producing under
Now at Inst. Mexicano del Petr61eo
"Now at Texas A&M U.
solution-gas drive. Field tests presented in Ref. 3 indicate that the
Copyright 1989 Society of Petroleum Engineers exponent n should be in the range 0.5:$ n:$ l. On the basis of cur-
Journal of Petroleum Technology, May 1989 541
18~--~--~--~--~--~--~

1.0 - Sor=0.234, Sgc=0.07


SET I
16 - - Sor=Sgc =O,Swi =0.3
.-..J>= 0.8
14 ffi

wo
12 0.025 ~~0.6
we.
Il. ....
w~
> 0.4
~
..J
8 0.02 W
a:: 0.2

0.015

Fig. 3-Relatlve permeability data.

TABLE 1-RESERVOIR PROPERTIES USED IN SIMULATION

Set 1 Set 2
Drainage radius, r., ft 656.17 1,000
Fig. 1-Fluld-property data, Set 1. Porosity, cp, fraction 0.3 0.119
PV, Vp ' MMbbl 1.124 1.665
Permeability, k, md 10 10
Well radius, rw' ft 0.32808 0.5
Initial pressure, Pi, psi 5,704.78 1,500
Initial water saturation, S wi 0.3 0.0
SET 2 Initial system compressibility, 1.085x 10- 5 1.334x10- 4
fLg X400(cp) c~, psi- 1
Initial oil viscosity, J.toi' cp 0.298 1.7645
6 Thickness, h, ft 15.55 25

rent knowledge of gas well behavior, Fetkovich attributes expo-


5 1.25 nent values less than unity to non-Darcy flow, although he
acknowledges that n can be less than unity strictly as a result of
variation in fluid properties. Fetkovich also reports that Vogel's
equation yields n=1.24. As we discuss later, this observation is
4 important. In terms of qo,max, Eq. 2 may be written as
qo/qo,max=[1-(p~/p2W . ......................... (3)
Fetkovich (see also Handy12) further suggested that if the varia-
tion of kro/(ll-oBo) with pressure were known, then well delivera-
bility for a well located at the center of a circle could be calculated
by the following equation.
p(l)
1.1 qo=J] A(p)dp, .................................. (4)
Pwj(t)
where J =kh/{l41.2[ln(re/rw)- * +s]} .................. (5)
1.05 and A(p)=kro/(ll-oBo). Theoretical confirmation ofEq. 4 is given
~-Bg X 102 (RB/SCF) in Ref. 13.
To predict future performance with Eq. 1, Standing suggested
oL-~~~~~~~dl that qo,max in the future can be obtained by the relation
o 400 800 1200 1600 qo,max,ho,max,p = [PjA(PJ)]/[ppA(fii,)] . ................ (6)
PRESSURE, p, psi Here the subscripts f and p refer to future and present conditions,
respectively.
Fig. 2-Fluid-property data, Set 2. Fetkovich proposes that future performance can be predicted by
the relation J~j(t)/J~p(t)=plpp, and in this case, we obtain (for
n=l)
qo,maxiqo,max,p =p]!p). . .......................... (7)

542 Journal of Petroleum Technology, May 1989


10 5 10 5

"'~
SET I Np/N
5 = II 51 I 0.001 @
,.,
~"O.9'
,., 4 Q 10
4
2
3
0.026
0.066
Q 10 4 0.093

N
X

N~
~
-
I
Np/N
N~

N
X

-
~
I
5 0.108

n=0.984
I~ 10 3 0=1.05 I 0.001
e 10
3

2 0.026
3 0.066 n=1.279 SET I
4 0.093 5 = -I
5 0.108
10 2
10- 1 10 10 2 103
Qo,STB/O

Fig. 4-Log-log deliverability plots, s = 11.51. Fig. 5-Log-log deliverability plots, s = - 1.

If we wish to establish a basis for the equations presented by Fet- Results


kovich or Standing, a convenient method is to assume, as Handy In this section, we first consider the influence of sand p on the
proposed, that A(p) varies linearly with pressure; i.e., shape of the deliverability curve. Methods to predict future per-
A(p) =A(P)- ([A(P) - A(O)]lp} (p-p) . ................. (8) formance are then examined. Then, we consider modifications to
the IPR curve to incorporate changes in s.
Substituting the right side of Eq. 8 for A(p) in Eq. 4, we obtain
The IPR Curve. Fig. 4 is a log-log plot of the deliverability curve
qo,max,/ qo,max,p = {Pf[ Af(p) + Af(O)]} I {pp [Ap (p) + Ap(O)]}. for a well producing under solution-gas drive. The variable of in-
...................................... (9) terest is the fraction of initial oil produced, NplN (or average pres-
sure,p). The data shown here are for s>O. Fig. 5 presents a similar
As long as A(O) is proportional to A(p) [in Vogel's case, this con- plot for s<O for identical values of NpIN. Set 1 data are used in
stant of proportionality is 119; in Fetkovich's case, A(O) is assumed both plots. For each specific value of NplN, well-defined straight
to be zero], we obtain Eq. 6. Fetkovich then proceeds to make the lines (within engineering accuracy) are obtained. Careful exami-
assumption A(p)=exp2; i.e., A(p)=expp, where ex is a constant nation of these responses suggests that some curvature exists at high
that is independent of time, and under these circumstances, Eq. 7 rates and that the curvature is concave downward. Considerable
results ~for n"*
1, in the general case, this may imply that Af(p)/Ap care was taken to estimate qo max for each case. The arrows that
(p) =Pfntl Pjnp]. point downward in these figu'res denote the value of qo,max' All
It is easy to show that Eq. 2 for n= I can be derived from Eq. data shown here are for boundary-dominated flow; the alignment
4 if Eq. 8 is valid. As Fetkovich mentioned, we can assume that of the deliverability curve is insensitive to production mode. Un-
A(O)=O. Fetkovich's equation can also be derived if we assume like single-phase flow, however, a family of curves is obtained dur-
that A(O)={3(P+P1)' where {3 is a constant of proportionality. In- ing boundary-dominated flow. Results similar to those shown here
spection of Eq. 1 suggests that it can be derived if we assume that were obtained for Set 2 data. In this case, however, values of n
A(O)={3 p. are greater than unity even for small values of NplN, and this re-
Note that the assumption given in Eq. 8 is merely a convenient sult can be attributed to the existence of a critical gas saturation.
way to establish a basis for Eqs. 1 and 2 (n = 1) or Eqs. 6 and 7 From a practical viewpoint, the most important points to note
(n= 1) to be valid. Eq. 8 represents only a sufficient, not a neces- are that (1) the value of n is a function of time or depletion, (2)
sary, condition to establish these results. For example, we can read- the change in n is not monotonic with time, and (3) for most cases,
ily show that the relation the value of n is greater than unity. The value of n is approximate-
ly unity only for exceedingly low values of gas saturation.
qo =J{A(iJ)(p-Pwf) - [A(p)- A(0)][(p-Pwf)2/2p]} . .... (10)
The above results indicate that n is a function of system com-
generally will be incorrect, 13 although Eqs. 1 and 3 may be valid. pressibility or gas saturation. The change in n with time reflects
This point may become more readily apparent if we note that Eq. the change in the mobility function as a function of pressure. The
4 will predict flow rates accurately if the area under the A(p) curve fact that n> 1 in many cases suggests that the variation in A(p) with
is accurate. Many A(p) functions can yield identical values of the pressure (distance) is much shallower than the profile suggested
integral. Our point is that the success of the methods of prediction by Fetkovich [A(p)=expp]. Note that Vogel also obtained expo-
of Standing, Fetkovich, and others does not imply that the A(p) nent values greater than unity (n = 1.24). The results in Fig. 6 dem-
function for the reservoir is a straight line or that we can predict onstrate this point where the variation of A(p) with pressure for
the A(p) function from measured data (qo' Pwf' and p). a specific case is shown for both transient and boundary-dominated
At this stage, it should be noted that the expressions for stabi- flow. Here, the symbol tiAD represents dimensionless time based
lized productivity index, J~(t), and productivity index at zero draw- on the drainage area, A, and is given by
down, J*, that are derived in the literature on the basis of the
expression for productivity index, J, given by Eq. 5 and the as- tiAD = (0.OO6328kk roi t)/(c/>C t iJl.oiA ) . ................... (11)
sumption that A(p) is a linear function of pressure generally will
be incorrect. We make this observation on the basis of several sets These profiles are typical of cases where Sgc=O and the well is
of computations we have made. 10 In fact, on the basis of a num- produced at a constant rate. If Sgc >0 and/or the well is produced
ber of computations, it can be concluded that the methods given at a constant wellbore pressure, then the profiles are shallower than
in Refs. 2 and 3 are successful because the expression for J given those shown here. The dashed lines represent profiles predicted by
by the right side of Eq. 5 is never calculated explicitly, although the Fetkovich hypothesis. The Fetkovich hypothesis clearly pre-
the expression for J given by Eq. 5 is used in deriving the final dicts mobility-function profiles that are much steeper than those
result. obtained by simulations.

Journal of Petroleum Technology, May 1989 543


z
o
SET I
i= s'O"eD'500,qo'80 ST8/0
NplN SET 1,5' 11.51
~ 0.8
10.001 -n'l 0
=>

l!~i~ ~y$
LL
>-
j~0.6
m~
0_

Y' '{/
::!:~04
0-< .
10.
UJ
N
:J
~ 0.2

~
o 0.25 0.5 0.75
NORMALIZED PRESSURE, p(r,tl/pi

Fig. 6-Mobility-function profiles as a function of time. Fig. 7-Log-log deliverability plot-.1p mode of plotting.

It is difficult to make further generalizations regarding the form qo =C(t)(p-pwf)n(t), then we can show that n(t) will be less
mobility-function profile without additional assumptions. Ifwe as- than unity. In this case, as expected, n = 1 only if the slope of the
sume for convenience that the mobility function is given by Eq. mobility function is zero. Note that even if n = I, estimates of
8 and note that A(O) must be zero for n to be unity. then n can be q o.max obtained by extrapolating these straight lines will be in-
greater than unity under three circumstances: (1) A(O) > 0; (2) correct.
A(O)<O and IA(O)I<m(p+pwf); and (3) A(O)<O and The results shown above substantiate Fetkovich's claim that a
IA(O)I > m(p+pwf)' where m is the slope of the straight line (see near-wellbore effect is the probable cause for field curves exhibit-
the Appendix). ing slopes much less than unity. The fact that all field experiments
Although we can conclude on the basis of our results and those presented by Fetkovich yield values of n:S 1 and that simulations
of Vogel that the exponent of the deliverability curve usually will yield estimates of n ~ 0.9 leads to the following observations: (1)
be greater than unity, this observation does not imply that the for wells under solution-gas drive, the non-Darcy flow component
deliverability equation for single-phase flow will be valid; i.e., data should not be ignored; (2) the value of n generally will change with
should not be plotted in terms of the pressure difference (p -Pwf)' depletion, unless the change is completely overshadowed by the
[In the gas case, if n> 1, then this result would suggest that the non-Darcy flow effect; and (3) if field experiments suggest that
deliv~rability equation is of the form qo=C(t)(P-pwf)'] The re- n"" 1, then this result does not necessarily imply that non-Darcy
sults in Fig. 7 are a replot of the data shown in Fig. 4 in terms flow is insignificant. (Of course, one could argue that the field re-
of the pressure difference, dp. These plots exhibit an upward bend sults of Fetkovich may not be representative of other cases, but
at higher flow rates; i.e., the decline in oil rate will be much faster the evidence in Ref. 3 is convincing. Similar conclusions can be
than that predicted by the equation for single-phase flow. It is pos- derived if the results in Figs. 9 and 10 of Ref. 14 are examined.)
sible to fit straight lines through portions of the data. The dashed Ifwe compare the results in Figs. 4 and 5, nothing in these figures
lines shown in Fig. 7 represent a fit through data at lower rates. suggests that the shape of the deliverability curve is influenced by
The exponent corresponding to these straight lines is approximate- s. Fig. 8 presents results we obtained for both positive and nega-
ly equal to 0.9. Straight lines with slopes equal to unity are also tive s, and solutions are presented along the lines presented by
shown for reference (unbroken lines). If we perform an analysis Vogel. I The unbroken line represents the solution of Vogel and
similar to that shown in the Appendix under the assumption that the dashed line represents the deliverability equation given by
A(p) is a straight line and that the deliverability equation is of the Fetkovich 3 (n = 1). Fig. 8a through c presents information for

(a) (b) a,b,e


0.8 0.8
S
0.6 ~ 0.6

-
IQ. IQ.
..... ..... ~ 0 0
t1 0.4 A~o 'i
Q. 0.4
~ 011.51
(;, -I
0.2 Np/N = 0.001 0.2 Np IN=0.066 -VOGEL EQN.
--FETKOVICH EQN.
(n. I)
00 0
0.2 0.4 0.6 0.6 0 0.2 0.4 0.6 0.6
qo Iqo,max qo Iqo,max

0.8
(e)
0.8 ... ~
(d)
d

Np/N
I~
IQ.
.....
'i
0.6 IQ.
.....
'i
0.6
0-
... 0- .. 0.0037
0.038
Q. 0.4 c. 0.4 reo =2000 I~:\ 0.078
s =20 I ~ 0.105
0.2 N p /N=0.108 0.2
-VOGEL EQN.
rso=2.8
- - FETKOVICH EQN.
0 0 (n al)
0 0.2 0.4 0.6 0.6 0 0.2 0.4 0.6 0.8
qo Iqo,max qo Iqo,max

Fig. 8-Deliverability plots-Cartesian coordinates.

544 Journal of Petroleum Technology, May 1989


s
o 0 ?
o 11.51 3.7,7.3 1SET I o 11.51 3.7,7.3 5SET'
"-I 3.7 "-I 3.7
20 2.8 SET 2 20 2.8 SET 2
- - STANDING'S "'-
IQ. - FETKOVICH'S
EQUATION
EQUATION

"
'i~ 10

I,OL_-:1-'--''-'-..J...L.......,I''--.......-I......./-U..LU.L- . . I - . . I -........u..u,..L-=2-.....
I 10 10 0
qo,max,p / qo,mox,f qo,mox,p /qo,mox,f

Fig. 9-Predictlon of future performance, Standing method. Fig. 10-Prediction of future performance, Fetkovich method.

several values of s for fixed values of NplN (Set 1 data), and Fig. bo12 ...... 4 implies that data at Pressure Level 2 were used to predict
8d presents data for several values of NplN for a fixed value of the value of qo,max at Pressure Level 4. The results given here sug-
s (Set 2 data). The results shown here and in Figs. 4 and 5 are in- gest that if data are used to predict responses over a short time range,
tended to demonstrate that the shape of the deliverability curve is then Standing's method should provide an adequate engineering ap-
essentially unchanged whether results are presented along the lines proximation to the actual value of qo.max,J. For example, consider
presented by Fetkovich (log-log plots) or by Vogel (Cartesian plots). Points 4 ...... 5,3 ...... 5,2 ...... 5, and 1 ...... 5 for Set 1 data. Points 4 ...... 5 and
The results shown in Figs. 4, 5, and 8 clearly show that there is 3 ...... 5 are closer to the unbroken line than Point 2 ...... 5 is, and all
no basis for the contention in Refs. , and 8 that the deliverability three of these points are closer to the unbroken line than Point ' ...... 5
curve changes from a quadratic form to a linear form as s increases is. Thus, predictions over a longer time span are less reliable.
(see Ref. 15). In fact, these results show that Eqs. 1 and 3 can be If we consider the alignment of the data points with the unbroken
used to analyze data if s*O, provided that we assume that the ap- line, we note that all data points fall on or to the left of this line.
propriate relations are valid. This observation also suggests that This result implies that if A(jJ) values are estimated accurately, then
the correction suggested by Couto and Golan 5 to incorporate the estimates of qo,max,J will be lower than the correct values of
change in s is unnecessary. We return to this point later in the paper. qo,max,f- Thus, values of qo predicted by this method will be low-
er than actual values of qo.
Prediction of Future IPR Curves. The data we have generated As mentioned by Fetkovich, Standing's method requires that the
enable us to test the validity of the equations suggested by k ro curve be well-defined. Fetkovich suggested, on the basis of em-
Standing 2 and Fetkovich 3 to predict the shift in the deliverability pirical observations, that Eq. 7 be used to avoid the need for rela-
curve with time. As mentioned in the Introduction, to our knowl- tive permeability data. Fig. 10 presents information that can be used
edge, verification of these suggestions is yet to be published. Fig. to verify this hypothesis. The unbroken line in Fig. 10 represents
9 presents data that test the validity of Eq. 6, which is denoted by Eq. 7. The data points reflect information extracted from the deliver-
the unbroken line. Data for several values of NplN and s are ability curves for several values of jJ. Except for the fact that re-
presented. Results for both sets of data that we considered are shown sults are presented in terms of average pressure, the information
here. The data points reflect results obtained from log-log deliver- given here is identical to that given in Fig. 9. The results shown
ability plots similar to Figs. 4 and 5 and estimates of A(jJ). The here suggest that if predictions are made over a short time span,
alignment of the data points with the unbroken line is good. For then the recommendation of Fetkovich should yield an accurate es-
purposes of discussion, most of the data points are labeled. The timate of qo,max,J' For example, Points 1 ...... 2, ' ...... 3, and 3 ...... 4 are
nomenclature used here is as follows. The first number refers to much closer to the unbroken line than Points 1 ...... 5 or 2 ...... 5 are (the
present conditions; the second, to future conditions. Thus, ' ...... 2 nomenclature used here is identical to that used in Fig. 9). On the
represents values at Pressure Levels 1 and 2, respectively, and 2 ...... 4 basis of the results shown here, we can conclude that Fetkovich's
represellts values at Pressure Levels 2 and 4, respectively. The sym- suggestion will yield estimates of qo,max,J that are good approxi-

1.1 .---~--...,---T"----r---r---r----,

SET 2, reo' 2000, s=20, 'so = 2.8, Pwf Ipj=0.5333 SET 2, reo = 2000, s = 20, rSO = 2.8, Pwf IPi =0.5333
TEST I, 'iAO= 0.058, Pwf Ijj = 0.5354 o TEST I' 'iAO=0.2903, Pwf/ji = 0.5429
TEST 2' 'IAO= 0.122, Pwf/jj = 0.5376 ~
TEST2' 'iAO =0.5230, Pwf lji = 0.5498
~o 0.9

w w
~ 0.8
!;i 0.8
0::
o o
~ 0.7
~ 0.7 ::::i STANDING
:J <l
<l ~
0.6 FETKOVICH
~ PIVOT-POINT
~
0::
0.6
~ 0.5
Qo,2 IqO,1 =0.9838
o I 2 3
0.5 l:---J.---!1!-----'----:2l:----'----:!3!---...J DIMENSIONLESS TIME, 'iAD
0
DIMENSIONLESS TIME, 'iAD
Fig. l2-Prediction of rate responses-influence of test con
Fig. ll-Predlctlon of rate responses-comparison of vari- ditions.
ous approaches.

Journal of Petroleum Technology, May 1989 545


1.1.---,--""--.-----,----,--,-----,
SET I, S:: 0, reo:: 2000 FETKOVICH PROCEDURE
"j' 0.48 s = 20. reo = 2000. SET 2

~" 0.46
tiAo=10.27, Pwf/Pj::0.5941

CD ui
~ !;[ 0.8
Q. 0:
20.44 faN 0.7

-z
~

Q. ::::i
, .( ~
0.6 p
Pwf/p J .STB/D/Psi 2
~0.42 0: I 0.0679 1.054X10-5
oZ 0.5 2 0.5429 8.43XI0-6
o 3 0.9162 7.21 X10- 6

~ 0.40 0.4 !:----'---+1----'----:!2:----'-----,3:------'


0
z DIMENSIONLESS TIME. tiAD
:::>
~
~0.38 Fig. 14-lnfluence of test conditions on rate predictions.
~
:J
a5 0.36 conclusion was also reached by Levine and Prats for the s=O case.
o Note that this approach does not require rate or pressure measure-
~ ments; relative permeability and fluid property data, however. are
0.34~------~--------~--------
needed. (The method we have used is slightly different from that
0.85 0.9 0.95 used in Ref. 16. The authors in Ref. 16 assumed that the pressure
and the GOR obtained from the differential material-balance equa-
NORMALIZED PRESSURE, Pwt'i> tion correspond to the values at the outer boundary. In Ref. 13,
we show that results from the differential material-balance equa-
Fig. 13-Mobility-function profile predictions. tion correspond to the average reservoir pressure; hence, in this
work we assume that the GOR given by the Muskat material-balance
corresponds to p. In some cases, these differences can be sig-
mations of the actual values of qo,max,f, provided that extrapola- nificant.)
tions are conducted over a short time range. From a practical An unfortunate aspect of the prediction methods given in the liter-
viewpoint, this method will provide estimates as accurate as the ature is that it is not possible to derive definite conclusions regard-
method suggested by Eq. 6, although the procedure recommended ing the accuracy of these methods. If test conditions are different,
by Standing (Fig, 9) will result in more consistent predictions. From then the differences in rate predictions can be significant. This point
the information given above, we can conclude that the suggestions is demonstrated in Fig. 12, where all conditions except the test con-
of Standing or Fetkovich can be used to predict future performance. ditions are different. In this particular case, the two-point methods
Care should be taken in applying the information given in Figs. are superior to the single-point methods. Nevertheless, in this case
9 and 10 to field data, particularly if single-point tests are used to also, the single-point methods are reasonably accurate if short time
predict future performance. First, the information given in Figs. spans are considered. This example emphasizes the need to obtain
9 and 10 assumes that the deliverability curve is well-defined- data at regular intervals and to not rely on extrapolations over long
i.e., that n is known. Second, the procedures suggested by Stand- time ranges. The prediction methods cited above can be used only
ing and Fetkovich provide no information on the future value of for constant-pressure production. We have not considered produc-
n. One would have to assume n to be unity or that n is unchanged tion at a constant rate because this mode of production is not perti-
(implicitly assumed in Refs. 2 and 3). In the following, we con- nent to rate predictions. For the constant-rate case, only the
sider the consequences of assuming that n is constant (unity or un- constant-GOR method yields consistent results.
changed) in performance prediction. In summary, on the basis of the results in Figs. 11, 12, and others,
Fig. 11 presents results that compare rates predicted by various we conclude that (1) all methods predict rates that are reasonably
methods. For convenience, we also present results obtained by the accurate for times immediately following test times-in this regard,
two-rate methods suggested in Refs. 4 and 6, which require data two-test methods and single-point methods are equally good for en-
at two pressure levels. (In this work, we make no distinction be- gineering purposes; (2) predictions based on two-test methods gener-
tween the two Kelkar-Cox 6 methods because the differences in ally are better than single-point methods, and predictions become
rates predicted by the two methods are negligible.) All results given more accurate if the gas saturation in the reservoir is greater than
here are normalized on the test rate at the first time level, qo,l (the the critical gas saturation; (3) Standing's method generally predicts
arrows denote times corresponding to the two tests). Predictions rates that are lower than all other predictions; (4) in many instances,
for the single-point methods were made with Test 1 data. Because Fetkovich's method predicts rates as accurately as the two-test
*
these methods have never been tested when s 0, results shown methods; and (5) predictions based on the Muskat material-balance
here are for s = 20 and are typical of the results obtained in this equation and the assumption that the GOR is independent of dis-
study. For constant-pressure production, we have found that the tance are reliable.
Vogel-Standing (n= 1.24) method generally predicts rates that are Although the results presented in Figs. 11 and 12 and in other
lower than the Fetkovich method (n= 1). Although not of direct computations not shown here suggest that future production rates
interest to this work, the Fetkovich method predicts rates essen- can be predicted with reasonable accuracy, one cannot conclude
tially identical to the two-test methods of Uhri and Blount 4 or that A(p) can be predicted by any of these methods. In Fig. 13,
Kelkar and Cox 6 in most cases, although the former requires only we plot A(p) predicted from some of the methods discussed in the
one data point. Unfortunately for the specific case considered here, literature at a given instant in time. The results shown here dem-
errors with the Couto-Golan 5 procedure are exceedingly high. Rate onstrate the point we made earlier: the success of these methods
predictions with Eq. 4 and the constant-GOR assumption suggest- in predicting future production rates does not require that A(p) be
ed by Levine and Prats l6 are also shown. The GOR for each spe- known accurately. The principal reason for the inability of these
cific average pressure was obtained from the Muskat l7 material- methods to predict the mobility function is that the expression for
balance equation. Unlike the other methods,2-6 this procedure pre- J given by the right side of Eq. 5 generally is not valid if it is
dicts actual rates very well after the onset of boundary-dominated assumed that the mobility function is a straight line. Note, how-
flow, and in most of the cases, it yields the best results. A similar ever, that the constant-GOR assumption predicts the mobility func-

546 Journal of Petroleum Technology, May [989


- - TWO-PHASE FLOW
- - SINGLE-PHASE FLOW
.0. .e.
~ 0.75 ~ 0.75
~ ~
0. Ef
e. Ef
0 0.60 0 0.60
fi
a: fia:
~ 0.45
a: w 0.45
::;) a:
::;)
C/)
C/)
C/) 0.30 C/) 0.30
w w
a: a:
Q. Q.
0.15 0.15

00 00

Fig. 15-Modifications to IPR curve-effect of the definition Fig. 16-IPR curves using quadratic definition of flow efficien-
of flow efficiency. cy, influence of V.

tion fairly well. In summary, neither the value of A(p) as a function cy; the dashed lines are solutions when Standing's definition of flow
of pressure nor the value of A(p) at any instant in time can be pre- efficiency is used. The curves are distinct for all values s or Ef ,
dicted on the basis of test data and the assumption that A(p) is a except for Ef = 1. For the Ef = 1.2 and 1.6 cases, as indicated by
linear function of pressure. Brown, the dashed line bends over and indicates that flow rate
The results presented in Figs. II and 12 and in the literature as- decreases as pwf is lowered. This incorrect behavior is avoided if
sume that the well will be produced under conditions identical to Eq. 12 is used. It is also interesting that all unbroken curves predict
test conditions. This assumption may not always be true. Because that qolq;r;;.~ =Ef when Pwf=O. This is the expected result. How-'
n can be different from unity and because it also changes with time, ever, none of the dashed lines reflect this result, further confirma-
additional errors can be introduced if test conditions are widely tion that the liquid-flow definition of flow efficiency should not be
different from actual producing conditions. Fig. 14 presents results used. Similar results are obtained when the Fetkovich deliverabili-
that address this situation. Here we compare predictions based on ty equation is used. The above observations are also applicable to
three values of J~(t) obtained from three points on a deliverability the results presented in Ref. 18.
curve under the assumption that n = 1. The actual value of n was At this stage, two points regarding the use of Standing'S method
1.362. Curves 1 through 3 represent predictions made with the deserve clarification. For purposes of discussion, let us assume that
highest, intermediate, and lowest flow rates, respectively. Differ- Standing's definition of flow efficiency is correct. In this case, it
ences in the predictions are significant, and these differences are can be shown that the point at which qo begins to decrease is given
a result of the use of different values for qo,max,p. If the operator's by P wf lp=(Ef -1.l25)IEf . If we examine this relationship, it is
intent is to produce the well at conditions similar to those of Test clear that problems with this approach will arise if Ef > 1.125. It
2, then the value of J~p(t) corresponding to Test 2 should be used can also be shown that the ideal well response, P:Vf' will be less
in predicting future production rates. Fig. 14 also presents simula- than zero when Pwf1p < (Ef-l)IEf . Again, this relationship sug-
tor results corresponding to well conditions identical to those in gests that no problems will be evident if Ef < I (damaged wells).
Test 2. From these results, the testing of wells at conditions that Note that the point at which the flow rate is a maximum is not iden-
reflect actual producing conditions is clearly preferable. tical to the point at which Pwf=O. Note also that computations that
suggest that Pwf$O are consistent with the van Everdingen 19 and
Modifications to IPR Curve To Incorporate Changes in Skin Hurst 20 concept of skin factor.
Factor. Because Vogel' s 1 deliverability equation was developed Fig. 16 compares responses for both the Vogel and Fetkovich
for s=O, StandingS suggested a simple method to incorporate the deliverability equations under the assumption that the quadratic form
effect of a change in s. His method is based on the assumption that of the definition of flow efficiency is valid. The unbroken lines are
the definition of flow efficiency for single-phase flow will also be the Vogel solutions; the dashed lines are solutions corresponding
valid for solution-gas drive. As mentioned earlier, Brown 7 and to the Fetkovich deliverability curve. In this case, all curves indi-
Whitson 9 showed that inconsistent results are obtained when this cate that qolq;!;;;a~ =Ef when Pwf=O and, more importantly, the
method is used if s<O. The deliverability plots shown in Figs. 4, incorrect behavior or nonphysical result is not evident. [Actually,
5, and 8 conclusively suggest that quadratic forms of the delivera- the suggestion of Brown 7 that a log-log plot be used to determine
bility curves suggested by Vogel and Fetkovich also apply if s *0. qo.max when incorrect behavior is evident tacitly assumes the quad-
These results imply that the definition of flow efficiency must also ratic definition of flow efficiency; the procedure recommended by
reflect the quadratic form of the deliverability equation. Thus, in him must be modified, however, because he assumes that the liq-
the following discussion, we define flow efficiency by uid definition of flow efficiency is valid before the incorrect be-
havior (see Figs. 2.22 and 2.23 of Ref. 7). Also, the solutions in
Er [(1 + VP:Vip)(1-P:Vf 1p)]![(1 + VPwf1p)(1-Pwf1p)] . .. (12)
Fig. 2.25 of Ref. 7 do not show that qolq;!;;;a~ =Ef at Pwf=O.]
Here V takes on values of 0.8 (Vogel) or 1 (Fetkovich), and P:Vf Results summarized in Table 2 demonstrate the advantages of
is the well pressure when s=O. Note that this definition of flow our recommendations. The log-log deliverability curves generated
efficiency is consistent with definitions commonly used in gas wells, in the course of this study were used to obtain the results given
where the quadratic form of the deliverability equation usually is here. Three data sets are considered in Table 2. Cols. 2 and 3 list
valid. For this definition of flow efficiency, the flow rate when values of s and the pressure ratio, Pwip, for each value of s. Col.
EF* 1 can be obtained by 4 presents qo.max obtained from the deliverability plots (values in
parentheses in the column are n.) Cols. 5 and 6 present appropri-
qolq;!;;;a~ =Ef [1 + V(Pwflp)][I-(Pwflp)] . ............. (13)
ate values of flow efficiency computed with results in Col. 3. Cols.
Fig. 15 presents solutions of Pwflp vs. qolq;!;;;a~. The unbroken 7 and 8 present values of q~(max obtained from the Vogel and Fet-
lines are solutions for the V = 0.8 case (Vogel's relationship of the kovich relations with the information in Col. 3. Differences be-
deliverability curve) when Eq. 12 is used to define flow efficien- tween the results obtained from the deliverability curve and the
Journal of Petroleum Technology, May 1989 547
TABLE 2-EFFECT OF DEFINITION OF FLOW EFFICIENCY ON FLOW-RATE PREDICTION

Skin Maximum Production


Data Factor, Pressure Maximum Flow Efficiency, E f Rate, qa,max
Set s Ratio, P wf1p Rate, q a,max' log-log Vogel Fetkovich Vogel Fetkovich
0 0,8993 440 (0.93) 1.0 1.0 508.37 460.23
-1 0.9174 530 (0.93) 1.2082 1.2068 614.21 555.44
11.51 0.6619 151 (0.964) 0.3348 0.3404 170.20 156.64
0 0.7942 238 (0.986) 1.0 1.0 261.52 238.36
2 -1 0.8273 275 (0.984) 1.1726 1.1700 306.66 278.87
11.51 0.2822 102.6 (1.05) 0.3825 0.4011 100.02 95.61
0 0.3130 100 (1.04) 1.0 1.0 102.44 97.56
3 -1 0.4444 131 (1.204) 1.1405 1.1240 116.83 109.65
Set 1, reO = 2,000.
Flow rate for all computations: q a = 88 STB/D.

equations of Vogel and Fetkovich reflect the influence of n. Vogel's flow. Although these extensions demonstrate considerable insight
method predicts larger values of qo max than Fetkovich's method, regarding the mechanics of fluid flow under solution-gas drive, in-
because Vogel's equation assumes 'n> I, and Fetkovich's proce- consistencies exist, particularly regarding the influence of s, the
dure is used with the assumption that n = I. If we now compute value of n, and the shape of A(p). One of the objectives of this work
qff!max Iq~t,;a~ for a specific data set, then this ratio will be equal is to attempt to resolve some of these inconsistencies by examining
to the appropriate value of Ef (Cols. 5 and 6). These results con- well performance in a rigorous manner and to lay a foundation for
firm that the definition of flow efficiency given by Eq. 13 is correct. further work. For example, we have shown that s does not influ-
On the basis of the results presented in Figs. 15 and 16 and Ta- ence the shape of the deliverability curve and that problems en-
ble 2, we can again conclude that the general shape of the deliver- countered in mOdifYing the IPR curve to study the effect of a change
'*
ability curve is unchanged if s 0 and that there is no basis for the in s are a result of ignoring the shape of the deliverability curve.
conclusion that the curve will be linear if s > 0. 1. 8 . 18 On the basis of this work, the following conclusions are warranted.
Results in Fig. 17 demonstrate the effect of using the quadratic \. The exponent n is a function of time or pressure level, and
definitions of flow efficiency. The test conditions are noted on the changes in n are not monotonic with time. In most of the simula-
figure; also noted are the values of q~{;.~. As indicated by Brown, tions we conducted, n> \. Even for early stages of depletion, n
the results obtained by Standing's method suggest that the flow rate can be different from unity. Because many of the synthetic deliver-
will decrease as the wellbore pressure approaches zero. If the quad- ability curves predict n> I (this conclusion can also be reached on
ratic definition of flow efficiency is used, then the maximum flow the basis of Vogel's work) and field results suggest that n < I, it
rate is obtained, as expected, when Pwf=O. Also shown in this appears that non-Darcy flow is the norm rather than the exception
figure is the response predicted by Harrison's equation (see Eq. in solution-gas-drive reservoirs.
2.39 of Ref. 7). At lower pressures, both the Standing and the Har- 2. The shape of the deliverability curve does not change from
rison methods, which are based on the definition of flow efficien- a quadratic to a linear form if the well is damaged. Methods pro-
cy for single-phase flow, predict rates much lower than the responses posed in the literature that assume that the shape of the curve is
predicted by the quadratic definition of flow efficiency, and these a function of s are incorrect.
differences are significant. 3. The suggestions by Standing and Fetkovich can be used to
predict future IPR curves, provided that extrapolations over long
Conclusions time spans are avoided. Predictions with single-point tests will be
A finite-difference model was used to evaluate IPR's in solution- as accurate as predictions based on two-point tests. If single-point
gas-drive reservoirs for a wide range of producing conditions. The tests are used to predict future production rates, then test condi-
effects of pressure level, well condition, and production mode were tions should reflect actual producing conditions. Under these cir-
studied. The consequences of using the Vogel l and Fetkovich 3 cumstances, errors in future production rates will be primarily a
deliverability equations (under identical conditions) are examined. function of the error in the initial estimate of n.
Many of the methods proposed in the literature for analyzing IPR 4. The method of prediction that uses the Muskat material-balance
curves have been derived by extending ideas based on single-phase equation to compute A(ji) and the constant-GOR assumption always
yields reliable results for boundary-dominated flow for all values
of s. This conclusion is an extension of the results in Ref. 16 to
2000~~~-.~~~'-~~~r-~~-r~~--'
skin factors different from zero. This procedure requires relative
permeability and PVT data, but it does not require any data point
.c;; TEST DATA' qo =200 STB/D as the other methods do; thus, concerns pertaining to extrapola-
Cl. Pwf =1500 psi tions over long time spans are not pertinent.
'i 1500 P =2000 psi 5. Problems that arise in applying Standing's modification to the
Cl. Ef =2 IPR proposed by Vogel to incorporate changes in s can be avoided
w qo,max if Ef is defined after consideration is given to the factors that
a: govern the pressure distribution in solution-gas-drive reservoirs;
=> STBID
~ 1000 VOGEL 250.0 i.e., the quadratic form of the deliverability equation must be used
W FETKOVICH 228.6
a: to define EI , and this definition should be used for both positive
a.. HARRISON 281.7
and negative s. Differences in predicted rates by the Vogel and Fet-
w STANDING 285.7
a: kovich (n= I) relations do not appear to be significant.
g 500
...J
...J Nomenclature
W
:;: a = function of time (Eq. A-2)
A = drainage area, ft2
h = function of time
h' = function A(p) at p=O (Eq. A-I)
Bg = FVF, RB/scf [res m 3 /std m 3 ]
Fig. 17-Comparison of methods on rate predictions.
Bo = oil FVF, RB/STB [res m 3 /stock-tank m 3]

548 Journal of Petroleum Technology, May 1989


CI = system compressibility, psi -1 [kPa -1] 2. Standing, M.B.: "Concerning the Calculation of Inflow Performance
C(t) = function of time (Eq. A-3) of Wells Producing From Solution Gas Reservoirs," JPT(Sept. 1971)
Ef = flow efficiency 1141-42.
3. Fetkovich, M.J.: "The Isochronal Testing of Oil Wells," paper SPE
h = formation thickness, ft [m]
4529 presented at the 1973 SPE Annual Technical Conference and Ex-
J = productivity index, defined by Eq. 5 hibition, Las Vegas, Sept. 30-0ct. 3.
J~(t) = productivity index for well producing under solution- 4. Uhri, D.C. and Blount, E.M.: "Pivot Point Method Quickly Predicts
gas drive, STB/[D-(psi2)n] Well Performance," World Oil (May 1982) 153-64.
[stock-tank m 3 /d'(kPa 2 )n] 5. Dias-Couto, L.E. and Golan, M.: "General Inflow Performance Rela-
J* = productivity index at zero drawdown, STBID-psi [stock- tionship for Solution-Gas Reservoir Wells," JPT (Feb. 1982) 285-88.
6. Kelkar, B.G. and Cox, R.: "Unified Relationship to Predict Future
tank m 3 /d' kPa]
IPR Curves for Solution Gas-Drive Reservoirs," paper SPE 14239
k = absolute permeability, md presented at the 1985 SPE Annual Technical Conference and Exhibi-
krg = relative permeability to gas tion, Las Vegas (Sept. 22-25).
k ro = relative permeability to oil 7. Brown, K.E.: The Technology ofArtificial Lift Methods, PennWell Pub-
m = slope of straight line (Eq. A-I) lishing Co., Tulsa, OK (1984) 4, 13-21.
n = exponent of inflow performance curve 8. Standing, M.B.: "Inflow Performance Relationships for Damaged Wells
Producing by Solution Gas Drive," JPT(Nov. 1970) 1399-1400.
N = initial oil in place, STB [stock-tank m3 ]
9. Whitson, C.S.: "Reservoir Well Performance and Predicting Deliver-
Np = cumulative oil produced, STB [stock-tank m3 ] ability," paper SPE 12518 available at SPE, Richardson, TX.
P = pressure, psi [kPa] 10. Camacho-V., R.G.: "WeIl Performance Under Solution Gas Drive,"
t:.p = pressure difference, psi [kPa] PhD dissertation, U. of Tulsa, Tulsa, OK (1987).
Pi = initial pressure, psi [kPa] 11. Camacho-V., R.G. and Raghavan, R.: "Performance of Wells in So-
Pwf = wellbore flowing pressure, psi [kPa] lution Gas Drive Reservoirs," paper 16745 presented at the 1987 SPE
Annual Technical Conference and Exhibition, Dallas, Sept. 27-30.
P'wJ = wellbore flowing pressure when s=O (Eq. 12), psi [kPa] 12. Handy, L.L.: "Effect of Local High Gas Saturations on Productivity
P = average reservoir pressure, psi [kPa] Indices," Drill. & Prod. Prac., API (1957) II 1-22.
q0 = oil flow rate, STB/D [stock -tank m 31 d] 13. Camacho-V., R.G. and Raghavan, R.: "Some Theoretical Results Use-
qo,1 = oil flow test rate at the first level ful in Analyzing WeIl Performance Under Solution Gas Drive," paper
qo,max = maximum oil rate, STBID [stock-tank m 3 /d] 16580 available at SPE, Richardson, TX.
r = radius, ft [m] 14. Millikan, C.V. and Sidwell, C.V.: "Bottomhole Pressures in Oil
Wells," Petroleum Development and Technology, Petroleum Div.,
re = external drainage radius, ft [m]
AIME (1931) 194-205.
rsD = dimensionless skin-zone radius 15. Raghavan, R.: "Well Test Analysis for Multiphase Flow," paper SPE
r w = wellbore radius, ft [m] 14098 presented at the 1986 IntI. Meeting on Petroleum Engineering,
R = producing GOR, scf/STB [std m 3 /stock-tank m3 ] Beijing, March 17-20.
Rs = solution GOR, scf/STB [std m 3 /stock-tank m 3 ] 16. Levine, J.S. and Prats, M.: "The Calculated Performance of Solution-
s = skin factor Gas-Drive Reservoirs," SPEJ (Sept. 1961) 142-52; Trans., AIME,
222 .
Sgc = critical gas saturation
17. Muskat, M.: "The Production Histories of Oil Producing Gas-Drive
So = oil saturation Reservoirs," J. App. Phys. (March 1945) 147-59.
Sor = residual oil saturation 18. Nind, T.E.W.: "Skin Factor and Flow Efficiency in Wells Producing
Swi = initial water saturation Under Solution Gas Drive," paper CIM 85-36-9 presented at the 1985
t = time, days CIM Annual Technical Meetirig, Edmonton, Alta., June 2-5.
tiAD = dimensionless time based on A and initial conditions 19. van Everdingen, A.F.: "The Skin Effect and Its Influence on the Produc-
tive Capacity ofa Well," JPT(June 1953) 171-76; Trans., AIME (1953)
V = constant (0.8 or I), see Eq. 12
198.
Vp = PV, MMbbl [m 3 ] 20. Hurst, W.: "Establishment of the Skin Effect and Its Impediment to
z = gas compressibility factor Fluid Flow Into a Wellbore," Pet. Eng. (Oct. 1953) 25, B-6.
ex = constant independent of time 21. Carter, R.D., Miller, S.C., and Riley, H.G.: "Determination ofStabi-
(3 = constant of proportionality Iized Gas Well Performance From Short Flow Tests," JPT(June 1963)
'Y = Euler's constant, 0.57721, .. 651-58.
"A(p) = mobility function (Eq. 4)
Appendix-Discussion on the Value of the
/A-g = gas viscosity, cp [Pa' s]
Exponent of the Dellverabllity Curve
/A-o = oil viscosity, cp [Pa' s]
<P = porosity In this section, we examine the value of n when "A(p) is a straight
line. Carter et at. 21 present a similar analysis for gas wells.
Subscripts Let the mobility function be given by
D = dimensionless "A(p)=mp+b' . .................................. (A-I)
1 = future conditions
i = initial conditions Then Eq. 4 can be written as
max = maximum
qo(t)=a(p L p2,q)+b(p-pwf)' .................... (A-2)
o = oil
P = present conditions where a and b can be functions of time but independent of
pLp~f'
Acknowledgments If we follow Fetkovich, 3 the deliverability curve is given by
We thank M.J. Fetkovich, M.D. Bradley, D. Epps, S. Joshi, and qo(t) = C(t)(p~rp~f )n(t). . ....................... (A-3)
T. Thrasher for their advice, suggestions, and comments. We ac-
knowledge the assistance of our colleague, Kermit E. Brown, who Combining Eqs. A-2 and A-3 and assuming C(t) to be independent
has patiently explained the significance of this topic to Raghavan of (p2_p~), we obtain
for the past 10 years. Camacho-V. acknowledges the support of n(t) = {a+b[d(p-pwf)]/[d(p2 -p2,q)j}[a+bl(p+Pwf)]-I.
the Inst. Mexicano del Petr6leo. Computing time was provided by
the U. of Tulsa. .................................... (A-4)
Eq. A-4 suggests that n will not remain constant for isochronal
References curves if "A(p) is given by Eq. A-I.
I. Vogel, J.V.: "Inflow Performance Relationships for Solution-Gas Drive In the following discussion, we will first assume that a > O. Equa-
Wells," JPT (Jan. 1968) 83-92; Trans., AIME, 243. tion A-4 suggests that if b=O [i.e., 1(0)=0], then n= 1.
Journal of Petroleum Technology, May 1989 549
We also have Case B (Eq. A-7 Is Valid). Combining Eq. A-4 and Inequality
A-7, we have (1) if b> 0, then n> 1, and (2) if b<O, five possibil-
[d(p-Pwt))/d(p 2 -p2,.t) = (p+Pwt+(P-Pwf)
x {1+ [2/(dpldpwr l)J} t 1................ (A-5)
ities exist. We will restrict our attention to the cases that yield values
of n observed in practice: if Ib[d(p-Pwt)/d(p2 -p4)] I<a, then
O<n< I, and if IbII(p+pwta, then n> 1.
From Eq. A-4 we note that the value of n will be dictated by one
If a =0, then for fixed value of p, Eqs. A-4 and A-5 suggest that
of the following inequalities.
n ~ 1. This result is well known, and this behavior also results when
d(p- Pwt)/d(p L p2,.t) < II(P+Pwt) ................. (A-6) data that should be approximated by the liquid-flow equations are
plotted in terms of (p2 -p2,.t).
or d(p-Pwt)/d(p 2 -p4 II(P+Pwt) . ................ (A-7)
If neither of these inequalities is valid, then n = I. The inequality 51 Metric Conversion Factors
given in Eq. A-6 will apply if the well is produced at a constant bbl x 1.589873 E-Ol m3
pressure or if dpldPwt= 1. (Note that dpldpwf= 1 for single-phase cp x 1.0* E-03 Pa's
liquid flow, constant-rate production, and that dpldPwt:51 for the ft x 3.048* E...,OI m
production of real gas under the assumption that p.z is proportional ft3 x 2.831 685 E-02 m3
to p). Inequality A-7 will apply if dpldPwt< 1 or if we study the md x 9.869233 E-04 p.m 2
shape of the deliverability curve for fixed values of p. psi x 6.894757 E+OO kPa
Case A (Eq. A-6 Is Valid). Combining Inequality A-6 and Eq. Conversion factor is exact. JPT
A-4, we have (1) if b>O, then O<n< 1, and (2) if b<O, then there
are five possibilities. Only two that yield values of n that have been Original SPE manuscript received for review March 8, 1987. Paper accepted for publica-
tion Feb. 14, 1989. Revised manuscript received Jan. 23. 1989. Paper (SPE 16204) first
observed in practice are of interest to this work: if Ibl/(p+pwf) <a, presented at the 1987 SPE Production Operations Symposium held in Oklahoma City, March
then n> 1, and if Ibd(p-Pwt)/d(p2_p 2,.t)1 >a, then O<n< 1. 8-10.

550 Journal of Petroleum Technology, May 1989

You might also like