Anda di halaman 1dari 5

A Comparison of Two-Phase Inflow

Performance Relationships
Frederic Gallice,* SPE, and Michael L. Wiggins, SPE, U. of Oklahoma

Summary Deliverability Methods


Petroleum engineers are routinely required to predict the pressure/ Vogel developed one of the earliest IPRs based on simulation data
production behavior of individual oil wells. These well-perfor- for 21 reservoir data sets representing a wide range of reservoir
mance estimates assist the engineer in evaluating various operating rock and fluid properties. Vogel noticed that the shapes of the
conditions, determining the optimum production scheme, and de- pressure/production curve for these cases were very similar. He
signing production equipment and artificial-lift systems. made the curves dimensionless by dividing the pressure at each
In this paper, commonly used empirical, inflow performance point by the reservoir pressure and by dividing the flow rate by the
relationships for estimating the pressure/production behavior dur- maximum flow rate to obtain the dimensionless inflow perfor-
ing two-phase flow are investigated. Relationships studied include mance curve. He observed that all the points fell within a narrow
those proposed by Vogel; Fetkovich; Jones, Blount, and Glaze; range and developed the following relationship to describe the
Klins and Majcher; and Sukarno and Wisnogroho. Each method is dimensionless behavior.


described briefly, and the methods used to develop the relationship
qo pwf pwf 2
are discussed.
= 1 0.2 0.8 . . . . . . . . . . . . . . . . . . . . . . . (1)
On the basis of actual vertical-well data, the relationships are qo,max pR pR
used to predict performance for 26 cases. The predicted perfor-
mance is then compared to the actual measured rate and pressure Fetkovich3 proposed the isochronal testing of oil wells to esti-
data. The variation between the predicted and measured data is mate their productivity. This relationship is based on the empirical
analyzed, and from this analysis, an assessment is made on the use gas-well-deliverability equation proposed by Rawlins and Schell-
of inflow performance relationships and of the quality of the per- hardt.8 Using data from multirate tests on 40 different oil wells in
formance estimates. six fields, Fetkovich showed that the following approach was suit-
able for predicting performance:
Introduction
When considering the performance of oil wells, it is often assumed qo = CpR2 pwf
2 n
, . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (2)
that production rates are proportional to pressure drawdown. This
which can be expressed in a form similar to Vogels IPR, as follows:
straight-line relationship can be derived from Darcys law for
steady-state flow of a single, incompressible fluid and is called the
productivity index (PI).
Evinger and Muskat1 were some of the earliest investigators to
qo
qo,max
= 1
pwf
pR
2 n
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (3)

look at oilwell performance. They pointed out that a straight-line This method requires that a multirate test be conducted to obtain
relationship should not be expected when two fluid phases are the values of C and n. A log-log plot of the pressure-squared
flowing in the reservoir. They presented evidence, based on mul- difference vs. flow rate is expected to plot as a straight line, where
tiphase flow equations, that a curved relationship existed between the inverse of the slope of the curve yields the deliverability ex-
flow rate and pressure. ponent n required in Eq. 3.
This work led to the development of several empirical inflow Using Forchheimers9 model to describe non-Darcy flow, Jones
performance relationships (IPRs) to predict the pressure/ et al.4 proposed the following relationship between pressure and rate.
production behavior of oil wells producing under two-phase flow
conditions. These estimates assist the engineer in evaluating various pR pwf
= A + Bqo. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (4)
operating conditions, determining the optimum production scheme, qo
and designing production equipment and artificial-lift systems.
This paper reviews and compares five IPRs proposed in the This method requires that a multirate test be conducted to deter-
literature for predicting individual-vertical-well performance in so- mine the coefficients, A and B, in which A is the laminar-flow
lution-gas-drive reservoirs. The IPRs studied are Vogel2; Fetkov- coefficient and B is the turbulence coefficient. From Eq. 4, it is
ich3; Jones, Blount, and Glaze4; Klins and Majcher5; and Sukarno evident that a Cartesian plot of the ratio of the pressure difference
and Wisnogroho.6 Each IPR was developed for various conditions to the flow rate vs. the flow rate yields a straight line, with the
but essentially represents vertical wells producing from a single y-intercept being A and the slope, B. Once the coefficients are
solution-gas-drive reservoir under boundary-dominated flow con- estimated, the flow rate at any flowing pressure can be determined
ditions. A homogeneous reservoir is assumed in all the methods with Eq. 5.
except for Fetkovichs; however, Wiggins et al.7 have shown that
this assumption does not restrict the applicability of an IPR A + A2 + 4BpR pwf
qo = . . . . . . . . . . . . . . . . . . . . . . . . (5)
method. Using data from 26 field cases, the five IPR methods are 2B
used to predict the pressure/production behavior for the individual
cases, and the predictions are compared to the actual well perfor- On the basis of Vogels work, Klins and Majcher5 developed an
mance and to the other methods predictions to develop an under- IPR that incorporated the bubblepoint pressure. The authors simulated
standing of their reliability. 21 wells using Vogels data and developed 1,344 IPR curves. Using
nonlinear regression analysis, they presented the following IPR.

* Now with Kerr-McGee, Houston.


qo
qo,max
= 1 0.295 pwf
pR
0.705
pwf
pR d
, . . . . . . . . . . . . . . . . . . (6)
Copyright 2004 Society of Petroleum Engineers
in which


This paper (SPE 88445) was revised for publication from paper SPE 52171, first presented
at the 1999 SPE Mid-Continent Operations Symposium, Oklahoma City, Oklahoma, 2831 pR
March. Original manuscript received for review 1 July 1999. Revised manuscript received 5 d = 0.28 + 0.72 1.235 + 0.001pb . . . . . . . . . . . . . . . . . . (7)
March 2004. Paper peer approved 6 March 2004. pb

100 May 2004 SPE Production & Facilities


Sukarno and Wisnogroho6 developed an IPR based on simula-
tion results that attempts to account for the flow-efficiency varia-
tion caused by rate-dependent skin as the flowing bottomhole pres-
sure changes. The authors developed the following relationship
using nonlinear regression analysis.
qo,actual
qo,max,r=0
= FE 1 0.1489
pwf
pR
0.4416
pwf
pR 2
0.4093
pwf
pR
3
,

. . . . . . . . . . . . . . . . . . . . . . . . . . . (8)
in which

FE = a0 + a1
pwf
pR
+ a2
pwf
pR
2
+ a3
pwf
pR
3
, . . . . . . . . . . . . . . . (9)

and

aR = bo + b1s + b2s2 + b3s3 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (10)


In Eq. 10, s is the skin factor, and a and b are the fitting coeffi-
cients shown in Table 1.

IPR Comparison multirate methods have differences of less than 10%. The average
To compare the various IPRs, data from 26 cases presented in the absolute difference for Fetkovichs method is 4%, while Jones
literature are analyzed. Each case uses actual field data represent- et al.s is 7%. The single-point methods have an absolute average
ing different producing conditions. Data from each case are used to difference ranging from 18 to 31% for Klins and Majcher and
select rate and pressure information for test points, and these points Vogel, respectively. In general, the difference tends to increase
are used to predict well performance with each IPR method. The with increasing pressure drawdown. This increased difference in
predictions are then compared to the actual measured production predicted vs. actual performance is expected. Because each IPR is
data at drawdowns greater than the test data. Several cases are used actually used to extrapolate performance behavior at drawdowns
to demonstrate the analysis and to provide insight into the behavior greater than the test point, one would expect these estimates to
of the various predictive models. Complete details of the analysis increase in error as one moves further from the known data point.
are presented by Gallice,10 while the cases analyzed are summa- Because the test data cover a wide range of pressure draw-
rized in Table 2. downs, they allow an investigation of the effect of drawdown on
performance estimates. Table 6 presents a summary of the average
Case 1. Millikan and Sidewell11 presented multirate-test data for a absolute differences for each method based on drawdown percent-
well producing from the Hunton Lime in the Carry City Field, ages (8, 21, 38, 51, and 78%) of the test point. As shown, the average
Oklahoma. The test was conducted in approximately 2 weeks, with
the well producing at random rates rather than in an increasing or
decreasing rate sequence. The average reservoir pressure was
1,600 psi, with an estimated bubblepoint pressure of 2,530 psi
and an assumed skin value of zero. The field data are summarized
in Table 3.
Table 4 presents the performance predictions for test informa-
tion at a flowing bottomhole pressure of 1,267 psi, representing a
21% pressure drawdown. As can be seen, the maximum well de-
liverability varies from 2,562 to 3,706 STB/D. The largest flow
rate was calculated with Vogels IPR, while the smallest rate was
obtained using Fetkovichs method.
Fig. 1 shows the various IPR curves generated from the test
data. Visual inspection indicates that the methods of Fetkovich and
Jones, Blount, and Glaze estimate the actual well performance
more accurately. The other methods capture the general shape of
the data but overestimate actual performance. If the straight-line
PI is used in this case, a maximum flow rate of 6,054 STB/D
would have been predicted from the test point. This estimate is
more than 60% greater than the highest rate predicted by the IPR
methods and shows the importance of using a multiphase flow
relationship to evaluate well performance when multiphase flow
occurs in the reservoir.
Table 5 shows the percent difference between the recorded
flow-rate data and the computed rate for the five IPR methods. The

May 2004 SPE Production & Facilities 101


absolute differences in the performance predictions for all the reservoir pressure decreased from 1,734 to 1,609 psi, or 7%, be-
methods decrease as the test-point drawdown percentage increases. tween tests. These cases were selected to demonstrate the effect of
For example, Vogels method predicted a maximum flow rate depletion on the IPR methods.
of 5,108 STB/D at an 8% pressure drawdown, compared with Owing to limited test data, performance predictions were made
2,564 STB/D at a drawdown of 78%. This is almost a 100% from test information at pressure drawdowns of 13 and 12% for
reduction in the maximum well deliverability. In addition, the reservoir pressures of 1,734 and 1,609 psi, respectively. As antic-
average differences in the performance estimates decrease from ipated, the various methods provide a range of performance esti-
72% at an 8% drawdown to 1.7% at a 78% drawdown. mates for both reservoir pressures. Table 7 summarizes the abso-
All the methods show that the average absolute differences lute differences in the IPR estimates. When the data were plotted,
decrease similarly in the predicted performance. By increasing there was little to distinguish the multipoint methods from single-
the pressure drawdown of the test point from 8 to 21%, the av- point ones for the first case. However, the second case clearly
erage absolute differences were decreased by more than 100% showed a definite difference between the multipoint and single-
for each method. For this particular case, a 20% pressure draw- point methods. The average absolute difference in the performance
down appears sufficient to predict the well performance. This estimates also changed between the two cases, indicating that the
is consistent with the observations of Wiggins12 who recom- reliability of the various performance methods may change during
mended, on the basis of simulation results, that a minimum pres- the life of a well.
sure drawdown of 20% be used for all well testing used to predict
oilwell performance.
In summary, Fetkovichs relation provided the best estimates of Summary. The additional cases and their analysis are presented in
well performance for this cases entire range of interest. In general, detail in Ref. 10. Table 8 presents a summary of the average
the difference in performance predictions increased as the pressure absolute difference for each method for all cases examined. As
drawdown increased from the test pressure. Also, the average ab- indicated, no one method always provided the most reliable esti-
solute difference in the predictions decreased as the test pressure mates of the actual well data analyzed. However, some general
drawdown increased. comments can be made on the basis of this table and all the cases
analyzed in this study.
Cases 2 and 3. The next cases represent one well located in the The multipoint methods of Fetkovich and Jones et al. tend to do
Keokuk pool, Seminole County, Oklahoma, where test data were a better job of predicting well performance than the single-point
collected 8 months apart at two different reservoir pressures.13 The methods. As a matter of fact, the total average absolute difference
is almost twice as great for the single-point methods as compared
to Fetkovichs multipoint method15 compared to 8%. The
method of Jones et al. had an average difference of 12%. Overall,
the single-point methods of Vogel, Klins and Majcher, and
Sukarno provided similar average differences in the cases exam-
ined14 to 15%.
Case 5 demonstrates the variation in the predicted performance.
In this case, Fetkovichs method performed the poorest in estimat-
ing actual performance, while Vogels IPR did the best. This case
clearly shows that one should not depend on a single IPR method
to make reliable performance predictions in all reservoirs.
Case 9 provides another anomaly in this analysis. Each method
provides very similar estimates, except for Jones et al. In this
case, Vogels method provides a somewhat better estimate than
Fetkovich. However, the multipoint method of Jones et al. pre-
dicted rates that are significantly different from the actual perfor-
mance. For this case, this method estimated performance with an
average absolute difference of 58%, compared with 16 to 18% for
the other methods.
As a final note, the available data or costs of obtaining data will
influence selecting an IPR method to predict performance. Overall,
multipoint methods will provide more information and are recom-
Fig. 1Predicted inflow performance curves compared to ac- mended to estimate well performance; however, it costs more to
tual field data for Case 1. obtain the data compared to single-point methods. In the end, the

102 May 2004 SPE Production & Facilities


benefit of multipoint methods must be carefully considered in 4. Because of depletion effects, one IPR method may be reliable at
relation to the expense of obtaining the information. one reservoir pressure but unreliable at another. This may be
caused by changes in reservoir parameters with time that can
Conclusions lead to changes in reservoir flow properties. Once again, this
suggests the use of multiple IPR methods to estimate well per-
In this study, five different methods to predict the pressure/ formance.
production performance of oil wells producing from solution-gas-
drive reservoirs have been presented. These are the methods of
Nomenclature
Vogel; Fetkovich; Jones, Blount, and Glaze; Klins and Majcher;
and Sukarno and Wisngroho. Each method requires parameters a fitting parameter defined in Eq. 10, dimensionless
that are normally available from a production test. The methods A laminar-flow coefficient, mL4/t, psia/STB/D
can be separated into multipoint and single-point methods. The b constant in Eq. 10, dimensionless
primary concern of this study was to evaluate the reliability of the B turbulence coefficient, mL7, psia/(STB/D)2
IPR methods on the basis of actual production-test data. Detailed C flow coefficient, L3+2nt4n1/m2n, STB/D/psia2n
analysis and comparisons for 26 different cases were performed. d exponent defined in Eq. 7, dimensionless
From this study, the following conclusions were drawn.
FE Sukarno and Wisnogroho flow efficiency, defined in
1. There is no one method that is the most suitable for every test.
It has been observed that in one case, one method will provide Eq. 9, dimensionless
the most reliable estimates, while providing the worst estimates n deliverability exponent, dimensionless
in the next case. From this observation, consideration should be pb bubblepoint pressure, m/Lt2, psia
given to using more than one method in predicting performance pR average reservoir pressure, m/Lt2, psia
to provide a range of possible outcomes. pwf flowing bottomhole pressure, m/Lt2, psia
2. Of the well-performance methods evaluated in this study and the qo oil flow rate, L3/t, STB/D
field data analyzed, Fetkovichs multipoint method tended to be qo,max maximum oil flow rate, L3/t, STB/D
the most reliable. It has been shown, on the basis of the test data, s skin factor, dimensionless
that the overall absolute difference for Fetkovichs method was
less than for the others. Also, Fetkovichs method provided
consistent performance predictions throughout the pressure- References
drawdown range, while the single-point methods appeared to be 1. Evinger, H.H. and Muskat, M.: Calculation of Theoretical Productiv-
more sensitive to the drawdown pressure of the test point. ity Factors, Trans., AIME (1942) 146, 126.
3. The selection of a drawdown pressure for testing purposes is an 2. Vogel, J.V.: Inflow Performance Relationships for Solution-Gas Drive
important parameter related to the reliability of the IPR meth- Wells, JPT (January 1968) 83; Trans., AIME, 243.
ods. It appears that a minimum drawdown pressure of 20% of 3. Fetkovich, M.J.: The Isochronal Testing of Oil Wells, paper SPE
the average reservoir pressure is required to obtain reliable es- 4529 presented at the 1973 SPE Annual Fall Meeting, Las Vegas,
timates of well performance for any IPR method. In general, it Nevada, 30 September3 October.
is recommended that test information be obtained as near to 4. Jones, L.G., Blount, E.M., and Glaze, O.H.: Use of Short Term Mul-
operating conditions as possible. tiple Rate Flow Tests To Predict Performance of Wells Having Tur-
bulence, paper SPE 6133 presented at the 1976 SPE Annual Technical
Conference and Exhibition, New Orleans, 36 October.

May 2004 SPE Production & Facilities 103


9. Forchheimer, P.: Wasserbewegung durch Boden, Ziets V. Deutsch
Ing. (1901) 45, 1782.
10. Gallice, F.: A Comparison of Two-Phase Inflow Performance Rela-
tionships, MS thesis, U. of Oklahoma, Norman, Oklahoma (1997).
11. Millikan, C.V. and Sidewell, C.V.: Bottom-hole Pressures in Oil
Wells, Trans, AIME (1931) 194205.
12. Wiggins, M.L.: Inflow Performance of Oil Wells Producing Water,
PhD dissertation, Texas A&M U., College Station, Texas (1991).
13. Haider, M.L.: Productivity Index, API Drilling and Production Prac-
tice, API, Dallas (1936) 181190.
14. Sukarno, P.: Application of the New IPR Curve Equations in Sangatta
and Tanjung Miring Timur Fields, Proc., Indonesian Petroleum Assn.
Sixteenth Annual Convention (1987).
15. Walls, W.S.: Practical Methods of Determining Productivity in Res-
ervoirs on Leases by Bottomhole Pressure and Core Analysis, API
Drilling and Production Practice, API, Dallas (1938) 146161.
16. Kemler, E. and Poole, G.A.: A Preliminary Investigation of Flowing
Wells, API Drilling and Production Practice, API, Dallas (1936)
140157.

SI Metric Conversion Factors


bbl 1.589 873 E-01 m3
psi 6.894 757 E+00 kPa
psi2 4.753 8 E+01 kPa2

Frederic Gallice is a geoscientist at Kerr-McGee Corp., Hous-


ton. e-mail: fgallice@kmg.com. Gallice holds a BS degree in
physics from the U. of Blaise Pascal and an MS degree in pe-
troleum engineering from the U. of Oklahoma (OU). Michael L.
Wiggins is a professor of petroleum and geological engineer-
ing at OU. e-mail: mwiggins@ou.edu. He has industry experi-
ence with major and independent exploration and produc-
5. Klins, M.A. and Majcher, M.W.: Inflow Performance Relationships tion companies. His teaching and research interests include
for Damaged or Improved Wells Producing Under Solution-Gas production operations, well performance, stimulation, artificial
Drive, JPT (December 1992) 1357. lift, and production optimization. Wiggins holds BS, MEng, and
6. Sukarno, P. and Wisnogroho, A.: Generalized Two-Phase IPR Curve PhD degrees in petroleum engineering from Texas A&M U. He
Equation Under Influence of Non-linear Flow Efficiency, Proc., Soc. is a Distinguished Member of SPE and currently serves on the
of Indonesian Petroleum Engineers Production Optimization Intl. Sym- Editorial Review Committee as Executive Editor of SPE Produc-
posium, Bandung, Indonesia (1995) 3143. tion & Facilities. He has been a member of the Production and
Operations Symposium (POS) technical program committee
7. Wiggins, M.L., Russell, J.E., and Jennings, J.W.: Analytical Devel-
since 1992 and served as the General Chairman for the 2003
opment of Vogel-Type Inflow Performance Relationships, SPEJ (De- POS. He was a Director of the Oklahoma City Section from
cember 1996) 355. 1999 to 2001. He has served as a member of the Engineering
8. Rawlins, E.L. and Schellhardt, M.A.: Backpressure Data on Natural Registration Committee, as the Faculty Adviser for the OU SPE
Gas Wells and Their Application to Production Practices, U.S. Bureau Student Chapter, and as a committee member for the Petro-
of Mines (1935) 7. leum Computer Conference.

104 May 2004 SPE Production & Facilities

Anda mungkin juga menyukai