Sie sind auf Seite 1von 12

SPE 106440

Decline Analysis for Horizontal Wells of Intercampo Field, Venezuela


Hong’en Dou, SPE, Changchun Chen, and Yuwen Chang, SPE, RIPED, PetroChina; Yanjun Fang, Daqing Oilfield,
PetroChina; Xinbin Chen, RIPED, PetroChina; Wei Liu, Liaohe Oilfield, PetroChina; and Wenxin Cai, CNODC, CNPC

Copyright 2007, Society of Petroleum Engineers


heavy crude oil, edge water and bottom water [1]. CNPC
This paper was prepared for a presentation at the 2007 SPE Production and Operations American Ltd drilled 131 new wells, including 54 horizontal
Symposium held in Oklahoma City, Oklahoma, U.S.A., 31 March–3 April 2007.
wells by the end of 2004, of which the first horizontal well
This paper was selected for presentation by an SPE Program Committee following review of
information contained in an abstract submitted by the author(s). Contents of the paper, as
was drilled in 1999. In the new wells, except for twenty seven
presented, have not been reviewed by the Society of Petroleum Engineers and are subject to ESP wells, 77 wells are produced by continuous gas lift.
correction by the author(s). The material, as presented, does not necessarily reflect any
position of the Society of Petroleum Engineers, its officers, or members. Papers presented at The existing decline curve analysis techniques, which
SPE meetings are subject to publication review by Editorial Committees of the Society of
Petroleum Engineers. Electronic reproduction, distribution, or storage of any part of this paper
included three Arps models (exponential, hyperbolic, and
for commercial purposes without the written consent of the Society of Petroleum Engineers is harmonic, 1945), and Fetkovich model (1980), are derived
prohibited. Permission to reproduce in print is restricted to an abstract of not more than
300 words; illustrations may not be copied. The abstract must contain conspicuous empirically. But the Arps models are still a preferred method
acknowledgment of where and by whom the paper was presented. Write Librarian, SPE, P.O.
Box 833836, Richardson, Texas 75083-3836 U.S.A., fax 01-972-952-9435.
for forecasting the oil production and proven reserve. The
methods have played a very important role in exploration and
Abstract development of oilfields in the world [2-6].
Intercampo oilfield is characterized by heavy oil with mid- Li and Horne (2001) developed their decline analytical
high permeability and high oil saturation, which is an model, called the Li-Horne model [7-8], and based on fluid flow
unconsolidated reservoir driven by edge water and bottom mechanisms study. In fact, Li and Horne made a conceptual
water. This paper presents the oilfield’s application of three mistake in their reasoning of the Arps models. Our research
decline analysis models to the 13 horizontal and 13 vertical found that the Arps’ models does not have any association
wells of the oilfield respectively, as well as the comparison of with fluid flow mechanism, reservoir types, and the fluid’s
the decline features between vertical and horizontal wells. The characteristics, such as steady or unsteady flow and single or
authors point out that the decline analysis models are not only multiphase flow.
applicable for vertical wells but also for horizontal wells. The two new decline analysis models are introduced in this
Although the water cut rising of horizontal wells are different paper; they are named the Orstrand-Weng model [2, 9] and T
from vertical wells, the authors emphasize the well type and model [10]. By using the two models, the production decline
reservoir properties are not associated with the decline analysis and reserve prediction both will be done.
analysis models. In regards to screening and comparing of The main purpose of the paper is to conduct a comparison
decline analysis models; how to select and use the models; among the four models of the Arps model, Li-Horne model,
their application condition; and the models features are clearly Orstrand-Weng model and T model for the application
presented in this paper. condition and results in the oilfields.
The paper introduces two new decline analysis models for Reservoir Description
horizontal wells. They have been applied to predict production
rate of the oilfields, reservoirs, and wells, as well as their In the last decade, horizontal well technology has been
future trends. The results accord well with the actual data from employed widely in the production of heavy oil and extra
the oilfield. The classical models screened out and the new heavy oil in the reservoirs of the Intercampo oilfield. The
models were recommended analyzing the production reservoirs are currently classified into three types. The first
performance of the oilfields and reservoirs for the types of type includes two areas of BASUP and BAMED. These
wells. reservoirs are situated in the upper Miocene, characterized by
heavy-oil, mid-to-high permeability, and high-saturation,
Introduction unconsolidated sand reservoirs. In the second type, we have
The Intercampo Oilfield, located in Zulia in the northeast part BAINF, LAGNA and LGINF, in the middle Miocene. They
of Lake Maracaibo in Venezuela, was taken over by CNPC are middle-heavy crude oil, mid-to-high permeability, high-
American Ltd. in 1998. The water depth of the lake ranges saturation, unconsolidated sand reservoirs. Finally, the third
from 7 to 26 meters and the entire area is 39.56 square meters type B-2-X features, a middle-heavy crude oil, low porosity-
in extent. Most of the reservoirs are buried from 3560 to 7500 permeability and high-oil-saturation sand reservoir, situated in
ft, with heavy crude-oil gravity ranging from 12.5 to 23.3 API. the lower Eocene.
The reservoir is characterized by strong heterogeneity, The mid-high permeability, heavy-oil reservoirs include
substantial differences in physical properties, such as middle– BASUP53 and BAMED78, which contain multi oil-water
systems, have an approximate aggregate of 500 MMB of
2 SPE 106440

initial petroleum in place and cover 27.9% of the entire qi


contract area. q (t ) = (4)
(1 + Di t )
The high-permeability, middle-heavy reservoirs include
BAMED58, BAINF60, LAGUNA10 and LGINF11. They In terms of cumulative production, the harmonic decline can
contain multi oil-water systems, the OOIP estimated at 558 be expressed as:
MMB and account for 31.2% of the entire contract area. At q q
present, being in a mid-later stage of oil recovery, the reserve N p = i ln (5)
Di q i
recovery percentages from LA10 and LG11 (in the Miocene),
are as high as 34% and 39.6%, respectively. A linear trend between the cumulative oil production versus
In the high oil-saturation reservoirs occurring in the the natural logarithm of production rate will be expected.
Miocene, porosity ranges from 26.7 - 37.8%, giving a mean Its judgment equation of the decline type is
porosity of 30.38%. Air permeability varies from 409 - 2962 1/q(t) = a + bt
md, the mean being 1456.93 md. Oil saturation is about 57 -
86%, the mean being 70.17%. The Hyperbolic Model. For 0<n<1, Equation 1 is defined as
The average API gravity is 14.3 degree and viscosity is the hyperbolic model
q (t ) = q i (1 + nD i t )
− 1n
about 40~763cp under the heavy oil reservoir conditions in the (6)
main BASUP-53 reservoir. The API gravity usually ranges
from 11 to 17, with the lowest API to 10 and the highest API In terms of cumulative production, the hyperbolic decline
to 23. The well liquid property is poor. The reservoir is a can be expressed as:
heavy oil reservoir and very difficult to produce. Presently, a
pilot test is going on. The BASUP-53 reservoir data is shown
q n ⎛ 1 ⎞ 1- n
Np = i ⎜ ⎟ qi − q
Di ⎝ 1 − n ⎠
(1- n
) (7)
in Table 1 [11].
Analysis of Past Decline Models In this case of hyperbolic-type decline curves the rate-time
relationship as well as the rate-cumulative relationship can be
The Arps decline model [2-3] represents the relationship straightened out after shifting to become straight lines on log-
between production rate and time for oil wells, shown as log coordinate.
follows: Its judgment equation of the decline type is
−1 (1/q(t))n = a + bt
q(t ) = q (1 + nD t ) n
i i
(1)
The Li-Horne Model. The Li-Horne production decline curve
where q(t) is the oil production rate at production time t and qi analysis model [7, 8] is expressed as follows:
is the initial oil production rate, n, denoting decline index, and 1
Di, denoting initial decline rate, are two constants. In equation q( t ) = ao − bo (8)
R( t )
(1), when production time t → ∞ , oil production q(t) → 0.
Generally, the models are only suitable for production rate where R(t) is the recovery at time t, The two constants ao and
decrease phase. When the Arps models were used, the key is bo are constant of the regression.
to judge the decline type. Equation 8 is a quadratic curve, according to discriminant
When n=0, Equation 1 represents an exponential decline, of the quadratic curve ∆ ≠ 0 and the invariant of the
and n=1 represents a harmonic decline in oil production. For quadratic curve δ > 0, so, the Li-Horne model is a
0<n<1, Equation 1 is defined as the hyperbolic model. The hyperbola.
three cases will be discussed as follows.
According to the function of the Arps models, all of the
The Exponential Model. For n=0, an exponential decline in asymptote equations for exponential decline, hyperbolic
oil production can be obtained and expressed as follows: decline and harmonic decline, respectively are R(t) =0, while
q( t ) = qi e − Dt (2) the asymptotic line equation of the Li-Horne model is q(t) =-
bo. This implies that the value of R(t) increases with the
The exponential decline model can also be expressed in
production time gradually. And that the production rate will be
terms of cumulative production:
close to -bo, a negative value. Therefore, we think that the
N = ( q − q) D
p i
(3)
applicability of the Li-Horne model has limitation.
Unquestionably, their model will produce negative production
where Np is the cumulative oil production. In the case of
prediction values, which should not meet actual oilfield
exponential decline, one should get a linear trend by plotting
status[12].
the cumulative oil production versus the production rate.
Its judgment equation of the decline type is On one hand, we cannot find an exact function of recovery
log q(t) = a + bt. factor R(t), with time. In other words, the relationship between
the oil production rate q(t) versus time cannot be obtained
The Harmonic Model. In the Arps model (Equation 1), n=1 directly. On the other hand, the recoverable reserve NR cannot
represents a harmonic decline in oil production, which can be be obtained from the Li-Horne model. Therefore, the Li-Horne
expressed as follows: model is difficult to implement for predicting oil production
and analyzing production decline.
SPE 106440 3

New Decline Analysis Models dN P


= at b (14)
The Orstrand-Weng Model. In fact, in 1945, J.J Arps have [2] dtNp
reported C. E. Van Orstrand’s model (1925). Integrating equation 14 on both sides with respect to Np and
q( t ) = At m e − Bt (9) t respectively, we obtain a new model as follows.
where A ,B and m are the model constants, and t is production (
a b +1
t ⋅ )
time. In equation 9, when production time t → ∞ , oil N P = N Re b +1
(15)
production q(t) → 0.
In 1984, a Chinese geophysicist, called Wenbo Weng, put We differentiate both sides of equation 15 with respect to t,
forward the same model as equation 9. He thought that the all thus obtaining
developmental processes experience a cyclical growth and a b +1
( t ⋅ )
depletion, phase. The change of things is directly proportional q(t) = AN R t b e b +1 (16)
to the function of the m-th power of time t and inversely
proportional to a negative exponential function of time t, just where NR is the upper limit that the oil production approaches
as the above model. Professor Weng referred to it as the Life asymptotically as t → ∞ , or the maximum oil production
Cycle Model [9]. Here, the model is named, the Orstrand-Weng that can be attained; but the oil production q(t) → 0 at
model. production time t → ∞ ; a and b are the model’s constants
Equation 9 can be used to predict and analyze the respectively.
production decline of the oil well, reservoir and oilfield during When we take the natural logarithm of both sides of
their life cycle. equation 11, a linear regression correlation will be obtained
In order to get the constants, both sides of equation 9 are according to the ln [q(t)/Np] ~lnt using actual data from the
divided by tm, and then taken the common logarithm. Equation oil fields or wells. The constant b is the slope of the straight
9 is changed into: line of the fitting curve. The constant a is calculated from the
q(t ) B intercept, and the parameter A in equation 16 is ea. Then the
log m = log A − t (10) constant a and b can be used to determine NR in equation 15
t 2. 303 and 16.
Assume a value of m, by setting t as x axis and log[ q(t)/tm] Taking natural logarithm in two sides of equation 15, we get
as y axis, log[q(t)/tm]is a function of time t, a curve of the following equation:
log[q(t)/tm]~t was obtain. Given a proper value of m, the linear
correlation coefficient of the regression line of log[q(t)/tm] and ln N p = A 1 − cx (17)
t can be the maximum value. In this way, m is determined. where
Meanwhile, A can be derived from the intercept of the straight d a
line, and B can be obtained from the slope of the regression A = ln N , x = t ,
1 R c=− ,d = b + 1
b+1
line. Once the parameters m, A and B are got, the Orstrand- A regression straight line of the actual data was plotted by
Weng model is determined. the relationship curve of lnNp~td, the parameter NR can be
T Model. By studying a large number of field data, we found derived from the intercept. Hence the model of equation 16 is
that a differential equation was followed between oil obtained.
accumulative productions with production time in the oil Equation 16 can analyze production decline only when b is
production process, and it can be written as in the equation less than negative 1. However, when b is a positive value, the
below [10] function is a cumulative curve.

dN p Application Results and Analysis


= f (t ) (11)
N p dt By three decline analysis models, a large number of
where Np is the cumulative oil production at the production Intercampo oilfield data of horizontal and vertical wells of
time, t; f (t) is an information function of the oil production several different reservoir types were studied. Total oil
rate change with oil production time, a power function production rate is 23474 BOPD at the end of 2004,
relationship between a ratio of the oil production rate to comprehensive water cut is 38.8%, and average GOR is 596
accumulative oil production with production time t, the scf/bbl, accumulation oil production rate is 2.0 ×108 bbl. oil
function is mathematically given by: production recovery factor is 12.33%, oil recovery percent of
q(t) residual recoverable reserve is 3.73%, recovery factor of
= at b (12)
recoverable reserve is 52.17%, the oil production is 134
Np
BOPD per well. Production decline analysis of 13 horizontal
And and vertical wells respectively was done by the Arps model,
dNp Orstrand-Weng Model and T model. Four wells as examples
q(t) = (13) are described as follows.
dt
Well BA 2376. This is a horizontal well located in Busup-53
Combining equation 11, 12 and 13, we obtain
reservoir, produced by ESP. the well was put into production
4 SPE 106440

in July, 2003, with the initial oil production rate of 333 bbl per was put into production in April, 2000, with initial oil
day. At present, the production rate is 233 BOPD. Daily, production rate of 690 BOPD. At present, the production rate
monthly and quarterly production rate were analyzed by the is 60 BOPD. Daily, monthly and quarterly production rates
three models. Daily and monthly decline type is the hyperbolic were analyzed by the three models. Decline types of daily,
of decline index n=0.53, 0.1 respectively, and quarterly monthly and quarterly are exponential declines of decline
decline type is the harmonic decline of decline index n = 1.0. index n = 0, in the three cases, decline rates are 5.7%, 5.5%
In three cases, decline rate is 1.9 %, 1.2% and 5.7 % and 16.6% respectively. (Table 2)
respectively (Table 2). Figs. 10 through 12 show that prediction results of the three
Figs. 1 through 3 show that the prediction results of the models are very close, after daily, monthly and quarterly
three models are very close. Comparing the Arps model with production declines were analyzed by the three decline
the Orstrand-Weng model and the T model all match the models. In the earlier stage of decline, the prediction results of
actual data relatively accurately in the early decline phase. The the Orstrand-Weng model and the T model is in accordance
prediction result of the T model is slightly bigger than the with actual production data. Prediction results of the T model
other two models for daily, monthly and quarterly production. are slightly higher than the other two models in the earlier and
later decline phases, while the prediction results of the T
Well BA 2384. This well is located in BAMED-78 reservoir,
model are less than the other two models in the middle decline
and it is a horizontal well, produced by ESP. the well was put
phase.
into production in Oct. 2002, with initial oil production rate
116 BOPD was produced. At present, the production rate is Comparison and Discussion. By analyzing the production
195 BOPD. Daily production rate, monthly production rate decline of vertical and horizontal wells, the decline rate of
and quarterly production rate were analyzed by the Arps vertical wells is larger than horizontal wells, because pressure
model. Decline types of daily, monthly and quarterly are drawdown in vertical wells is faster than that in the horizontal
exponential decline of decline index n = 0, in the three cases, wells. The average decline rate is compared between vertical
decline rate is 1.0%, 0.09% and 2.8% respectively (Table 2). and horizontal wells, a vertical well is 7.1%, 8.2% and 19.7%
Figs. 4 through 6 show that prediction results of the three for daily, monthly and quarterly production and a horizontal
models are very close. Prediction result of the production well is 3.7%, 3.5% and 11.4%. In addition, from 1999 to 2000,
decline using the Arps model and the Orstrand-Weng model the water cut rising of 13 horizontal wells are faster than that
of 13 vertical well in the same phase in the oilfield (Fig. 13).
are consistent for daily, monthly and quarterly decline. But the
A large number of actual production data was analyzed in
T model result is slightly higher than the other two models in
terms of the three models. We found that the decline analysis
the earlier and later decline phase, while at the middle phase
models are available not only for vertical wells but also for
prediction result of the T model is lower than the other two horizontal wells. The models should be used to estimate
models. reserves of the wells, reservoirs and oilfields.
Well BA 2313. This is a vertical well, located in BAMED-78 As is well known, there is a significant difference between
reservoir and is produced by gas lift. The well was put into vertical and horizontal wells regarding the fluid flow through a
production in Sept. 1999, with an initial oil production rate of porous medium. However, a large amount of statistical data
100 BOPD. At present, the production rate is about 50 BOPD. comes from the oilfield, the decline analysis of wells and
Daily, monthly, and quarterly production rates were analyzed oilfields are both in accordance with the Arps model and the
by the three models. Daily decline type is exponential decline other two new models. These models do not have any
association with the well types, fluid flow mechanism,
of decline index n =0, monthly decline type is also an
reservoir types, and fluids characteristics.
exponential decline of decline index n = 0, and quarterly
Figs. 1 thought 12 show the comparison among the Arps
decline type is a harmonic decline of decline index n = 1. In
model, Orstrand-Weng model and T model in terms of the
the three cases, decline rate is 2.9%, 3.3% and 9.5%
analysis of the production rate versus time. We studied a large
respectively (Table 2).
amount of data of the Intercampo oilfield and from other
Figs. 7 through 9 show that the prediction results of the
oilfields [11] using the Arps model. We found that the
three models are very close. Daily, monthly and quarterly
prediction results of the Arps model coincided with actual
production declines were analyzed by the three decline
oilfield data accurately. The Arps model represents the
models. In the earlier stage of decline, the prediction result of
relationship among the production rate, cumulative
the Orstrand-Weng model and T model is in accordance with
production, and reserve with time. Therefore production
actual production data. Prediction results of the Arps model
decline analysis and reserve can be predicted easily using this
also coincide with the Orstrand-Weng model closely for daily,
model. And the Arps model’s analysis results show that it is
monthly and quarterly production decline, but the T model is
quite simple and highly accurate. In addition, it has another
less than the two models for the three cases in the middle stage
advantage, being the decline type and decline rate can be
of decline. However, the prediction results of the T model
clearly described for each decline phase in the oilfield life
stand in the top of the curves of the two models in the earlier
cycle.
and later decline phases.
As for production decline analysis, of daily, monthly,
Well BA 2343. This well is a vertical well, located in quarterly and yearly, all three models (Arps, Orstrand-Weng
LAGNA-10 reservoir and is produced by gas lift. The well or T model) produce matching results, however those results
SPE 106440 5

lie in the middle of actual results derived from analyzing the Substituting equation 17 into 9, we obtain
production data of vertical and horizontal wells. Therefore, ⎛ 1 ⎞
q(t) = bo ⎜⎜ - 1⎟⎟ (18)
their prediction results met the decline analysis of the oilfield ⎝ [R(t)] R ⎠
in any oil production decline phases. For the decline analysis
of horizontal well, Figs. 1 through 12 show that the Orstrand- In equation 18, when production time t → ∞ , R(t)]R → 1 oil
Weng and T model matching results are highly accurate in the production q(t) → 0. When the Li-Horne model was used to
early decline phase. However, prediction results of the Arps analyze actual oilfield data, there are three regression forms,
model coincide with the Orstrand-Weng model apparently in they are q(t)= a/[R(t)]R -b, q(t)= -a/[R(t)]R -b and q(t)=
phase of decline, while prediction results of the T model is a a/[R(t)]R+ b. We cannot obtain a regression form, as equation
slightly larger than the other two models in the earlier and 18 according to the actual data of oilfield. Therefore, the Li-
later decline phase. For the decline analysis of vertical wells, Horne model is inaccurate. Therefore this model is not useful
the prediction results of the Arps model seem to coincide with for production decline analysis. The analysis results were
the Orstrand-Weng model throughout the decline phase. When shown in Figs. 14 and 15.
actual data shows a faster decline trend in the early decline The T model can be used to analyze the production decline
phase, the T model closely coincides with the actual data. But and estimate the reserve. The model does not need to judge the
generally, the model has a slightly larger prediction results decline type, and the model constant may easily be obtained
than the other two models in the later decline phase. by using Microsoft Excel.
The Orstrand-Weng model is an entire prediction process The Orstrand-Weng model was first used by these authors
model, and it can be used to predict the oil production from to analyze the production decline and predict production. The
start to finish. Also, the model can be used to deal with the model is unnecessary to judge decline type. Its prediction
production decline analysis when setting the power item m<0. results are in accordance with actual production results of the
As for the Li-Horne model, a negative value occurred in the oilfield. In addition, the model can be used to predict the entire
oilfield life cycle. On this point, we consider the Li-Horne development process in the oil and gas field. Furthermore, it
model lacking in accurate results in regards to actual oilfield has higher forecast accuracy.
production and its prediction results [12]. We can see that the magnitude predicted using the three
We see that by assuming the q(t) is to close zero, R(t) = different models have prediction results which are closely
ao/bo from the Li-Horne model. Some data from the matched. The comparison presented in Figs. 1 through 12
SPEREE June 2005(Line 27 of page 200) were analyzed as demonstrates that the Arps, Orstrand-Weng and T model may
following: have a better accuracy for the cases studied. But the Orstrand-
Weng and T model are more flexible than the Arps model.
ao =0.0055 OOIP/year, bo =0.04386 OOIP/year Particularly, the Orstrand-Weng model can be used to predict
4
If OOIP= 4925.9×10 m3 (310 MMBO) [13] in the field, if not only production trends but also reserve trends during the
the oilfield life t =30 years, R(t)t=30years= 12.6% .The oilfield life cycle.
4
calculation value of ao and bo are, ao= 1625.54×10 m3, bo=
4 3 Conclusions
12963 ×10 m .
Substituting above data into equation 9, we obtain 1. The Arps, Orstrand-Weng, and T model are not only
4
suitable for vertical wells but also for horizontal wells
q(t)t=30year=[ ao/R(t)] -bo= -61.89 ×10 m3 although the water cut rises differently in vertical wells
Obviously, the oil production rate q(t)t=30 year encounters a and horizontal wells. The authors emphasize that the well
negative value during oilfield life cycle. Here, we need to type and reservoir properties are not associated with the
explain the data, which is oil recovery of OOIP, it is 1695× decline analysis model.
4
10 tons [14] (106.73 MMBO), However, the recovery percent 2. The Arps model is a widely applicable, very useful and
of OOIP has reached 16.9%, according to the statistics data of highly accurate for production decline prediction of
2004 Petrochina Reservoir Engineering Annual Report. Li and vertical and horizontal well. In addition, the Decline rates
Horne used an incorrect OOIP data of the Reference 13, an of horizontal wells are lower than for vertical wells on a
example from Yianling field in northeastern China. We cannot daily, monthly and quarterly basis.
help but to say Li and Horne’s prediction R(t) = 12.6% is a 3. The Orstrand-Weng model was not only applied to
incorrect. At the same time, when time t is close to infinite, analyze production decline in the decline phase, but also
to predict the oilfield production and its trends for
q(t)=0, the Li-Horne model q(t) is unable to satisfy this
horizontal and vertical wells throughout the oilfield life
mathematical characteristic.
cycle.
Because the Orstrand-Weng model and T model used the
4. The T model has been applied to production decline
same parameter, recoverable reserve NR, we need to shift the analysis and production prediction of horizontal and
oil recovery of OOIP into the recoverable reserve in the Li- vertical wells. Prediction results met actual data obtained
Horne model in order to make sure the data is uniform. from the oilfield. Therefore, the T model will be
Therefore, a relationship between the recovery percent of recommended for the reservoir engineering.
recoverable reserve, [R (t)]R , with the oil recovery factor, 5. The Li-Horne model has a significant limitation in that it
R(t), was given as follows, encounters a negative value in the oilfield life cycle
R (t) = ao [R (t)]R / bo (17)
6 SPE 106440

according to the model’s function performance. In prepared for presentation at the 2005 SPE International
addition, the Li-Horne model is difficult to use in Thermal Operations and Heavy Oil Symposium held in
predicting production decline analysis due to the fact that Calgary, Alberta, Canada, 1–3 November 2005
the model cannot give NR or R (t) expression equation 12. Hongen Dou, Changchun Chen, Discussion on Decline
directly. Analysis Models for Production Prediction: SPE 62552,
6. The production prediction results of horizontal and 77544, 83470 and 93878, SPE-107143, 2006-09-28,
vertical wells using the three models are very similar. The http://manuscripts.spe.org/spej/cgi-
Orstrand-Weng model and T model coincide in their bin/main.plex?el=A3C6LLd1A3lsD3F1A9KGj2VQJuVS
accuracy during the early decline phase of horizontal hHh35hjSRZwZ, 2006-10-6
wells. However, for the decline analysis of vertical wells, 13. Allan, J and Sun, S. Q. , Control on Recovery Factor in
the prediction results of the Arps model coincide well Fractured Reservoirs: Lessons Learned from 100
with the Orstrand-Weng model throughout the entire Fractured Fields, paper SPE 84590 presented at 2003 SPE
decline phase. In spite of the type of well, the prediction Annual Technical Conference and Exhibition, Denver, 5-8
results of the T model are slightly larger than that of the October, 2003
other two models in the earlier and later decline phase. 14. Fenghan Bai, Yongqing Shen et al, N2 Injection Pilot for
Improving Oil Recovery in Yanlin Oilfield, ACTA
Acknowledgement
Petrolei SINICA, 19 (4), 61-68, Oct., 1998
The authors wish to thank CNPC America Ltd., Venezuela for
granting permission to publish this paper. Also, thank RIPED, SI Metric Conversion Factors
Petrochina Company Ltd. for support. bbl×1.589 873* E-01 = m3
References bbl/D×1.589 873* E-01 = m3/D
ft3 ×2.831 685 E-02 = m3
1. Hong'en Dou, Dandan Hu et al , Sand production acre×4.046 873 E+03 = m2
prediction and the selection of completion methods for *
psi×6.806 690 E-03 = MPa
horizontal well in Intercampo oilfield,Venezuela, SPE
ft×0.305* E-02 = cm
93821, this paper was prepared for presentation at the
2005 Asia Pacific Oil & Gas Conference and Exhibition cp×1.0* E-03 = Pa.s
held in Jakarta, Indonesia, 5 – 7 April 2005 (℉-32)/1.8 =℃
2. Arps, J.J.: “Analysis of Decline Curves,” Trans. AIME 141.5/(131.5+ o API) = g/cm3
*
(1945) 160, 228-247. Conversion factor is exact.
3. Arps, J.J.: “Estimation of Primary Oil Reserves,” Trans.,
AIME (1956) 207, 182-191.
4. Fetkovich, M.J.: “Decline Curve Analysis Using Type
Curves,” JPT (June 1980) 1065-1077.
5. Fetkovich, M.J., Vienot, M.E., Bradley, M.D., and Kiesow,
U.G.: “Decline-Curve Analysis Using Type Curves Case
Histories,” SPE Formation Evaluation (December 1987)
637- 656.
6. Fetkovich, M.J., Fetkovich, E.J., and Fetkovich, M.D.:
“Useful Concepts for Decline-Curve Forecasting, Reserve
Estimation, and Analysis,” SPERE (February 1996) 13-22.
7. Kewen Li., et al. An Analytical Model for Production
Decline-Curve Analysis in Naturally Fractured Reservoirs
(SPE 83470). SPE REE, 8(3), 197-203. June, 2005
8. Li, K., Chow, K., and Horne, R.N.: “Effects of Initial
Water Saturation on Spontaneous Water Imbibition,” SPE
76727, presented at the 2002 SPE Western Region
Meeting/AAPG Pacific Section Joint Meeting held in
Anchorage, Alaska, May 20-22, 2002
9. Wenbo Weng, Theory of Forecasting, International
Academic Publisher, A Pergamon-CNPIEC Joint Venture,
Beijing, 1991, 98-99
10. Fusheng Huang, Yongsheng Zhao et al, A New Model for
Oilfield Performance Prediction, Petroleum Geology and
Oil Development Daqing (P.G.O.D.D) 6(4), 55-62, Dec.,
1987
11. Hong'en Dou, Chang Yuwen et al, Application of Gas Lift
Technology to High Water-Cut Heavy Oil Reservoir of
Intercampo Oilfield, Venezuela, SPE 9730, the paper was
SPE 106440 7

TABLES

TABLE 1-BASIC DATA OF BASUP-53 RESERVIOR


Sedimentary Lake-river delta Original pressure (MPa) 13.3
2
Oil-bearing area (km ) 9.243 Buble point pressure (MPa) 12.16
Net thickness (m) 32.92 Reservoir temperature (℃) 76
4 3
OOIP (10 m ) 5829.25 Coefficient of pressure 0.95
Porosity (%) 30 Temperature gradient(℃/100m) 3.42
Oil saturation (%) 72 Datum depth (m) 1402
API 14.3 Put into production (year) 1955
3
Oil density (g/cm ) 0.971 Well spacing (m) 300
Oil-viscosity @ underground(cp) 40~700 Well pattern T hree-spot
Natural water drive and
Reservoir depth (ft) 4330 -5103 Drive type
solution gas drive
-3 783 Recovery percent (%) 2.99
Permeability (10 µm2 )
Shale content (%) 2.38 Water cut (%) 48

TAB LE 2 CO M PARISO N VERTICAL AND H O RIZO NTAL DECLINE


M ean Daily Production M on thly Production Quarterly Production
W ell T yape W ell No. Arps Decline Arps Declin e Arps Declin e Arps Decline Arps Decline Arps Decline
Rate Index Rate Index Rate Index
BA 2386 0.375 0.3 0.470 0.4 0.645 0
BA 2313 0.029 0 0.033 0 0.095 0
BA 2354 0.023 0 0.026 0 0.090 0
BA 2314 0.034 0 0.038 0 0.108 0
BA 2381 0.039 0.5 0.030 0.2 0.109 0.4
BA 2342 0.061 0.1 0.057 0 0.201 0
Vertical BA 2343 0.057 0 0.055 0 0.178 0
W ells BA 2345 0.038 0 0.041 0 0.115 0
BA 2350 0.088 0.4 0.133 0.55 0.286 0.5
BA 2351 0.045 0 0.045 0 0.161 0
BA 2355 0.024 0 0.023 0 0.069 0
BA 2360 0.041 0 0.044 0 0.119 0
BA 2364 0.062 0.38 0.077 0.36 0.387 0.7
Average 0.071 0.082 0.197
BA 2332 0.088 0.75 0.048 0.2 0.124 0
BA 2348 0.044 0.3 0.043 0.22 0.168 0.54
BA 2367 0.029 1 0.024 0.48 0.064 0.35
BA 2376 0.019 0.53 0.012 0.1 0.057 1
BA 2384 0.010 0 0.009 0 0.028 0
BA 2392 0.088 0 0.102 0 0.320 0
Horizontal BA 2403 0.029 1 0.025 0.7 0.057 0
W ells BA 2408 0.051 0 0.067 0 0.286 0.5
BA 2414 0.027 0.2 0.025 0 0.089 0
BA 2299 0.006 0 0.007 0 0.029 0
BA 2330 0.038 0 0.035 0 0.104 0
BA 2333 0.041 0 0.043 0 0.128 0
BA 2339 0.009 0 0.009 0 0.033 0
Average 0.037 0.035 0.114
8 SPE 106440

FIGURES

500
Actual Production
450
Arps Model
Mean Daily Production(bbl) 400 Orstrand-Weng Model
350 T Model
300
250
200
150
100
50
0
0 5 10 15 20 25 30 35 40 45 50
Month

Fig. 1-Mean Daily Production Decline Analysis of Well BA2376

12000
Actual Production
Arps Model
10000
Orstrand-Weng Model
Monthly Production(bbl)

T Model
8000

6000

4000

2000

0
0 5 10 15 20 25 30 35 40 45 50
Month

Fig 2-Monthly Production Decline Analysis of Well BA2376

35000
Actual Production
30000 Arps Model
Orstrand-Weng Model
Quarterly Production(bbl)

25000 T Model

20000

15000

10000

5000

0
0 2 4 6 8 10 12 14 16 18 20
Quarter

Fig. 3-Quarterly Production Decline Analysis of Well BA2376


SPE 106440 9

180
Actual Production
160 Arps Model

Mean Daily Production(bbl)


Orstrand-Weng Model
T Model
140

120

100

80

60
0 10 20 30 40 50
Month

Fig. 4-Mean Daily Production Decline Analysis of Well BA2384

6000
Actual Production
Arps Model
5000
Orstrand-Weng Model
Monthly Production(bbl)

T Model
4000

3000

2000

1000

0
0 10 20 30 40 50
Month

Fig. 5-Monthly Production Decline Analysis of Well BA2384


16000
Actual Production
14000 Arps Model
Quarterly Production(bbl)

Orstrand-Weng Model
12000
T Model
10000

8000

6000

4000

2000

0
0 5 10 15 20 25
Quarter

Fig. 6-Quarterly Production Decline Analysis of Well BA2384


10 SPE 106440

160
Actual Production
140
Arps Model

Mean Daily Production(bbl)


120 Orstrand-Weng Model
T Model
100

80

60

40

20

0
0 10 20 30 40 50 60
Month

Fig. 7-Mean Daily Production Decline Analysis of Well BA2313

4500
Actual Production
4000
Arps Model
3500 Orstrand-Weng Model
Monthly Production(bbl)

3000 T Model

2500

2000

1500

1000

500

0
0 10 20 30 40 50 60
Month

Fig. 8-Monthly Production Decline Analysis of Well BA2313

14000
Actual Production
12000 Arps Model
Quarterly Production(bbl)

Orstrand-Weng Model
10000 T Model

8000

6000

4000

2000

0
0 5 10 15 20 25
Quarter

Fig. 9-Quarterly Production Decline Analysis of Well BA2313


SPE 106440 11

1000
Actual Production
900
Arps Model
800

Mean Daily Production(bbl)


Orstrand-Weng Model
700 T Model
600
500
400
300
200
100
0
0 10 20 30 40 50 60
Month

Fig. 10-Mean Daily Production Decline Analysis of Well BA2343

30000
Actual Production
25000 Arps Model
Orstrand-Weng Model
Monthly Production(bbl)

20000 T Model

15000

10000

5000

0
0 10 20 30 40 50 60
Month

Fig. 11-Monthly Production Decline Analysis of Well BA2343

80000
Actual Production
70000 Arps Model
Quarterly Production(bbl)

60000 Orstrand-Weng Model


T Model
50000

40000

30000

20000

10000

0
0 5 10 15 20 25
Quarter

Fig. 12-Quarterly Production Decline Analysis of Well BA2343


12 SPE 106440

60
Avg. Water-Cut of 13 Horizontal Wells
50 Avg. Water-Cut of 13 Vertical Wells

40

Water-Cut (%)
30

20

10

0
Feb-99 Aug-99 Mar-00 Oct-00 Apr-01 Nov-01 May-02 Dec-02 Jun-03 Jan-04 Aug-04
Time

Fig 13-Average Water-cut of Horizontal Wells and Vertical Wells

52000
47000 Monthly Prodction vs. 1/R

42000 Quarterly Production vs. 1/R


y = 0.0001x + 12677
2
37000 R = 0.8918
Production(bbl)

32000
27000
22000
17000 y= 0.0013x + 5212.1
2
R = 0.5188
12000
7000
2000
0 2E+07 4E+07 6E+07 8E+07 1E+08 1E+08 1E+08 2E+08 2E+08 2E+08
1/R

Fig. 14-Li-Horne Production Decline Analysis of Well BA2376

14000
Monthly Prodction vs. 1/R
12000
y = 471.34x + 1112.9 Quarterly Production vs. 1/R
2
R = 0.6706
10000
Production(bbl)

8000 y= 0.0013x + 5212.1


2
R = 0.5188
6000

4000

2000

0
0 10 20 30 40 50 60 70 80 90 100
1/R

Fig 15-Li-Horne Production Decline Analysis of Well BA 2313

Das könnte Ihnen auch gefallen