You are on page 1of 11

Journal of Petroleum Science and Engineering 42 (2004) 245 255

www.elsevier.com/locate/petrol

Intermittent gas lift in Aghajari oil field, a mathematical study


Shahaboddin Ayatollahi a,*, Mostafa Narimani a, Mahmood Moshfeghian a,b,1
a

School of Chemical and Petroleum Engineering, Shiraz University, P.O. Box, 71345-1719, Shiraz, Iran
b
Department of Chemical Engineering, University of Qatar, P.O. Box 2713, Doha, Qatar

Abstract
The gas lift is a widely used method of artificial lift, for which there are different options. Previous work in this regard to
optimize the continuous gas lift (CGL) of Aghajari oil field located in the south of Iran was promising. A new intermittent gas
lift (IGL) simulator is developed considering all aspects of this artificial lift method including the temperature differences
between the injected and produced fluids. In order to check the validity of the mathematical model, the intermittent gas lift
experimental data, i.e. pressure gradient as a function of time, of Brown [Gas Lift Theory and Practice, Prentice-Hall,
Englewood Cliffs, NJ (1967)] were compared with the results obtained by this simulator. The calculated pressure gradient
matched well with the experimental data. For the same system, the pressure under the liquid slug as a function of opening
pressure of gas lift valve was also calculated and compared with the experimental data. It was found that ignoring the heat
transfer between the injected gas and the liquid slug shifts away the calculated results from the experimental points and
therefore, decreasing the accuracy of the model. The model then was used to study the Aghajari oil field. The results showed
that the production of this oil field will increase significantly using intermittent gas lift method.
D 2003 Elsevier B.V. All rights reserved.
Keywords: Gas lift; Artificial lift; Intermittent; Computer simulation; Heat transfer; Two phase flow

1. Introduction
Artificial lift is used in petroleum production when
the energy of the reservoir is not enough to sustain the
flow of oil in the well up to the surface with
satisfactory economic return. High pressure gas is
injected into the well to lighten the column of fluid
and allow the reservoir pressure to force the fluid to
the surface. The gas that is injected is produced with
the reservoir fluid into the low pressure system.

* Corresponding author. Fax: +98-711-628-7294.


E-mail addresses: shahab@shirazu.ac.ir (S. Ayatollahi),
moshfeg@qu.edu.qa (M. Moshfeghian).
1
On sabbatical leave from Shiraz University.
0920-4105/$ - see front matter D 2003 Elsevier B.V. All rights reserved.
doi:10.1016/j.petrol.2003.12.015

Oil production operations in Aghajari oil field in


south of Iran are mature. Initially discovered in the
1938, cumulative recovery from this giant oil field
with more than 15 billion barrels of recoverable oil
has reached 10 billion barrels so far. Irans Aghajari
Field, which experienced peak production of 1
million barrels a day in 1974, and then produced a
steady 850,000 barrels per day for 17 consecutive
years, is now going to consume 3 billion cubic feet
of South Pars gas to prop up production of only
187,000 barrels per day. This field operates 165
wells, 60 of them under artificial gas lift process.
Typical wellbores include 3 1/2 and 4 1/2 in. tubing
set in 9 5/8 in. production casing. Due to the
importance and demand of gas lift process, developing a mathematical model to simulate this process is

246

S. Ayatollahi et al. / Journal of Petroleum Science and Engineering 42 (2004) 245255

of high interest. The model may be used to find the


optimum operational condition.

2. Intermittent gas lift


Selection of the proper artificial lift method is
critical to the long-term profitability of the oil well;
a poor choice will lead to low production and high
operating costs. The gas lift is a widely used
method of artificial lift, where gas is injected in
the production well providing energy to the flow.
Previous work in this regard to optimize the continuous gas lift of Aghajari oil field located in the
south of Iran was promising (Ayatollahi et al.,
2001). In some instances, the continuous injection
of gas, named continuous gas lift (CGL), is not
efficient, and the intermittent (periodical) injection
of gas, named intermittent gas lift (IGL), becomes
the more economical alternative. Different design
options are available to implement the IGL: conventional IGL (CIGL), IGL with plunger (IGLP),
IGL with chamber (IGLC) and GL with pig, also
known as pig-lift (PL). In the IGLP system, a solid
plunger separates the oil and gas flowing in the
well, to prevent fallback of oil. In the IGLC
system, a chamber is used to accumulate the oil
at the bottom of the well, reducing the backpressure against the reservoir formation. In the PL
system, a foam-pig separates the oil and gas flowing in the well to prevent fallback of oil, and a
double column is used inside the well. There are
some empirical, though questionable, rules of
thumb to choose between the CGL and IGL. Table
1 shows a brief description of these rules (Odair
et al., 2001).
The vertical flow of liquid slugs is more complex than either multiphase continuous vertical or
horizontal flow. The unsteady state nature of liquid
Table 1
Practical criteria for selection of continuous (CGL) and intermittent
(IGL) gas lift (Odair et al., 2001)
Static head (hs/Hw)

Productivity index (PI, m3/day MPa)


High ( z 20)

Medium (5 20)

Low ( V 5)

High ( z 0.7)
Medium (0.4 0.7)
Low ( V 0.4)

CGL
CGL/IGL
IGL

CGL/IGL
CGL/IGL
IGL

IGL
IGL
IGL

slug flow has discouraged attempts to describe the


phenomenon analytically. Computer simulation of
the IGL method which would be a powerful tool
for production engineers is not in public domain.
On the other hand, most of these softwares are
developed for special cases which are difficult to
be adapted to the operational conditions in the
field.

3. Literature review
An extensive overview of artificial lift design
considerations was presented by Clegg et al.
(1993). Chacin (1994) discussed the state of the
art of the design of IGL methods, presented a
simplified algorithm for the calculation of the
production rate, and a procedure to select the best
IGL method. Brown and Jessen (1962), Brill et al.
(1967), and Neely et al. (1974) presented some
experimental results on specific field installations
of conventional IGL. They concluded their own
empirical rules for the setting of the operational
parameters. White (1963) developed the first simple
mathematical relationships for the conventional IGL
and did experiments on laboratory installations.
Liao (1991) obtained theoretical results that showed
good agreement with Brown, Brill and Neely.
Odair et al. (2001) studied different types of IGL
in the oil industry. This work presents a numerical
model to study the behavior of the Conventional
IGL, the IGL with chamber (IGLC), the IGL with
plunger (IGLP) and the IGL with pig. Simulations
were done under various reservoir conditions, for
different settings of the operational parameters.
Abdel-Wally et al. (1996) optimized the gas lift
process in Gulf of Suez Field, and resulted production increase from 17,000 bbl/day of oil to
19,000 bbl/day. Ayatollahi et al. (2001) used PVT
data combined with fluid and multiphase flow
correlations to optimize the continuous gas lift
process in Aghajari oil field. From actual pressure
and temperature surveys and determining the point
of injection, a gas lift performance curve was
constructed. In order to determine the optimal gas
lift condition, nodal method was used to determine
optimum injection depth, optimum well-head pressure, optimum production rate and minimum injec-

S. Ayatollahi et al. / Journal of Petroleum Science and Engineering 42 (2004) 245255

247

Fig. 1. Intermittent gas lift process.

tion gas volume as well as the appropriate valve


spacing.

4. Theory
The use of artificial lift methods is generally
required at some stage during the life of a reservoir.
The general models for simulating the performance of
artificial lift system for designing and optimizing
operations usually assume steady-state flow conditions for calculating the influx from the reservoir.
However, the bottomhole flowing pressure in cyclic
artificial lift methods, such as intermittent gas lift, is
anything but constant. Fluctuation of this pressure,
which controls the pressure of the well, takes place
depending on the characteristics of the well and the
operating conditions of the particular lift method.
Accurate prediction of the reservoir performance in
the simulation of any artificial lift system is necessary
since the behavior of the system itself as well as
reservoir production and buildup are affected by
artificial lift parameters directly.
Intermittent gas lift is a cyclic production method
in which a liquid slug is first allowed to build up in
tubing string. Fig. 1 shows a schematic diagram of
this artificial lift method. When the combination of
surface back pressure, weight of gas column, and

hydrostatic pressure of the slug reaches a specified


value, gas is injected into the casing annulus through
some type of control at the surface for a definite
injection time. Under ideal conditions the liquid, in
the form of a slug or piston, is propelled upwards by
the energy of expanding and flowing gas beneath it.
Production of oil slug and the gas results in pressure
decreases in the tube causes the gas lift valve to be
closed and stopping gas injection. The cycle starts
again by liquid buildup in the bottomhole where gas
injected from the control valve pushes the next oil
slug upward.
4.1. Pressure gradient for liquid slug
Widely varying pressure at the top and bottom of
the moving liquid slug are characteristic in IGL. In

Table 2
Physical properties correlations (Vasquez and Beggs, 1980; Beggs
and Robinson, 1975; Brill and Beggs, 1984; Soave, 1972)
Properties

Correlation

Solution gas oil ratio


Oil formation volume factor
Oil viscosity
Gas viscosity
Gas compressibility factor

Vasquez
Vasquez
Beggs and Robinson
Lee et al.
SRK EOS

248

S. Ayatollahi et al. / Journal of Petroleum Science and Engineering 42 (2004) 245255

Fig. 2. Schematic diagram of slug size and pressure gradient in IGL method.

order to simulate and then optimize the IGL process,


the information of pressure gradient beneath the slug
is very important as it finds its way from the wellbore to the surface. Also because of the number of
variables involved, the analytical study of IGL is
extremely complex. The different possibilities of
flow regimes in two phase flow of oil and gas in
vertical pipe, the source and nature of different
forces acting on the system, the interfacial instabilities involved, operational principles of existing gas
lift valve and unsteady state nature of the IGL
process are some of these difficulties that must be
overcome in the simulation of this process. Since
IGL is an extremely unsteady state phenomenon and
the pressure gradient changes with time, an incremental study is performed to simulate the IGL
process. Two types of average are used. The first
was an arithmetic average of variable at a given time
and over a finite distance. The averaged pressure in a
static gas column found using this technique. The
second type, referred to as time average, represents
an arithmetic average of a given variable over a
finite time increment. The governing equations and a
summary of the solution technique are shown in
Appendix A. The physical properties used in the
modeling are found from correlations and SRK EOS
(Soave, 1972) as indicated in Table 2. A schematic
diagram of the IGL process at two different time
increments (t and t + Dt) is shown in Fig. 2. It should
be noted that the gas lift valves are located at the
perforations.

4.2. Solution methodology


The governing equations were solved using a
complete implicit technique. The time for liquid slug
to reach the surface was found based on a timed
average of velocity and acceleration. The pressure at
the gas liquid interface at the top of the liquid slug
was composed of the separator pressure and the gas
phase pressure above the slug. On the other hand, the
pressure of the gas liquid interface beneath the liquid
slug is composed of the pressure at the gas liquid
interface at the top of the liquid slug and the pressure
due to liquid slug column. In order to calculate the
pressure gradient due to the friction in the two phase
flow regime of vertical tubing, Poetman and Carpenters (1952) equation was used. To calculate the

Table 3
Experimental input data (Brown, 1967)
Gas injection
temperature (jF)
Productivity index
(bbl/(psi day))
Diameter of control
valve (in.)
Reservoir
temperature (jF)
Separator pressure
(psia)
Gas injection
pressure (psi)

80
1
1
150
65

600

Oil API
Length of
tubing (ft)
Fallback (ft/ft)
Gas specific
gravity in separator
Control valve
opening pressure
(psi)
Reservoir pressure
(psi)

34
5940
0.00007
0.65
350

1000

S. Ayatollahi et al. / Journal of Petroleum Science and Engineering 42 (2004) 245255

249

Fig. 3. Comparison of Brown experimental data with the simulation results (pressure gradient).

pressure at the operating valve, the pressure values of


different regions consist of gas liquid interface and
gas column under liquid slug, pressure of falling-back
liquid and finally the pressure of liquid feed-in must
be added together. It is noted that the amount of
liquid feed-in is subjected to the injecting valve
pressure. In order to complete the pressure gradient
calculation, a trial and error scheme was performed

based on the known pressure values at two successive


time increments.
Heat transfer between the injected gas and oil
slugs has been ignored in the most of the investigation. Incremental technique used in this modeling
allowed for detail calculation of heat transfer and
temperature gradient along the tubing which affect all
the properties of gas and oil. An empirical correlation

Fig. 4. Comparison of Brown experimental data with the simulation output (maximum pressure under slug vs. valve opening pressure).

250

S. Ayatollahi et al. / Journal of Petroleum Science and Engineering 42 (2004) 245255

Table 4
Aghajari oil field data (Taghae, 1984)
Gas injection
temperature (jF)
Productivity index
(bbl/(psi day))
Outside diameter
of Tubing (in.)
Inside diameter of
casing (in.)

110

Formation gas
liquid ratio
(scf/bbl)
Gas injection
pressure (psi)
Valve opening
pressure (psi)
Reservoir
temperature (jF)
Control valve
opening (in2)

550

3
3
9.625

1300

Oil API

5. Results
34

Length of
tubing (ft)
Water cut
percent
Gas specific
gravity in
separator
Inside diameter
of tubing (in.)

8500

Reservoir
pressure (psi)

2150

Oil production
rate (bbl/day)
Overall heat
transfer coefficient
(Btu/h ft2 jF)

1050

2
0.76

2.625

500
180
0.8

19

along with trial and error schemes is used to accurately calculate the temperature gradient (Dongwoo
and Ghajar, 2002).

5.1. Accuracy of the model


Few experimental artificial lift data are available.
One of these data is by Brown (1967), shown in
Table 3. In order to check the validity of the
mathematical model, the intermittent gas lift experimental data, i.e. pressure gradient as a function of
time, of Brown (1967) were compared with the
results obtained by this simulator. As it is shown
in Fig. 3, the calculated pressure gradient matched
well with the experimental data. A key factor in
designing of an optimized intermittent gas lift process is the pressure under the liquid slug as a
function of opening pressure of gas lift valve.
Therefore, for the same system, the pressure under
the liquid slug as a function of opening pressure of
gas lift valve was also calculated and compared with
the experimental data in Fig. 4. It was found, as
shown in Fig. 4, that ignoring the heat transfer
between the injected gas and the liquid slug shifts
away the calculated results from the experimental
points and therefore, decreasing the accuracy of the
model.

Fig. 5. Calculated pressure gradient of gas lift process in Aghajari oil field.

S. Ayatollahi et al. / Journal of Petroleum Science and Engineering 42 (2004) 245255

251

Fig. 6. Calculated GLR vs. the casing pressure of gas lift process in Aghajari oil field.

5.2. Modeling of Aghajari oil field


The information of a typical well in Aghajari field
is shown in Table 4. The pressure gradient at depth of
6000 ft with respect to time for the mentioned well is

shown in Fig. 5. The gas liquid ratio (GLR) determines the success of an artificial lift process. Fig. 6
shows the GLR with respect to the pressure of the
casing for an opening pressure of 500 psi for a typical
well of Aghajari oil field. Increasing the pressure of

Fig. 7. Calculated GLR vs. valve opening pressure for the IGL process of Aghajari oil field.

252

S. Ayatollahi et al. / Journal of Petroleum Science and Engineering 42 (2004) 245255

Fig. 8. Calculated production rate with respect to the percent of fallback for intermittent gas lift process in Aghajari oil field.

the casing results in ever-increasing of GLR up 8000


(Scf/bbl) at pressure of 1400 psi.
The opening pressure affects the size of each slug as
well as the number of the cycles and changes the oil
production rate. Fig. 7 shows the GLR as a key
parameter of gas lift process with respect to the opening
pressure, indicating that the opening pressure should
not be less than 500 psi to achieve sufficient oil
production. Liquid fallback due to the friction between
the liquid slugs is another symptom of conventional
IGL process. As the gas elevates the liquid slug, leaving
behind a film of liquidthe fallbackdecreasing the
oil production rate. The amount of liquid fallback was
defined as the percent lost in slug length for every 1000
ft of slug movement upward. Using this definition,
losing 0.00001 ft of slug length for every ft of the slug
traveling upward denotes as 1% fallback. This parameter is set manually in the simulator using the results
from Brown (1967) experimental investigations. Fig.
8 shows the effects of liquid fallback on the total oil
production per day for the indicated Aghajari oil well.
Decreasing the amount of liquid fallback, using techniques introduced previously such as IGLC, IGLP and
PL, from 8% to 4% can double the rate of oil production
to 200 bbl/day. This mathematical simulation shows
that the oil production from the Aghajari oil field would

increase to over 30,000 bbl/day from almost nil using


intermittent gas lift when the continuous gas lift process cannot produce more oil.

6. Conclusions
In this paper, a model of intermittent gas lift for
pressure depleted reservoirs was presented. On the
basis of this model, the main problems concerning the
calculation of transient pressure gradient could be
formulated and mathematically solved. The proposed
methodology allowed the hydrodynamic investigations of the intermittent gas lift in order to obtain
the distributions of the pressure, oil and gas production parameters at different locations of the artificially
lifted oil reservoirs. The physical analysis of the
energy changes of the hydrodynamic parameters due
to the different temperature of the injected gas and
producing oil has indicated that the most effective
model can be obtained by introducing the energy
balance into the reservoir simulator. Finally, it is
shown that the oil production from a giant oil field
in the south of Iran, Aghajari oil field, can be
increased using this type of artificial lift if more
pressure depletion is experienced in the field.

S. Ayatollahi et al. / Journal of Petroleum Science and Engineering 42 (2004) 245255

Nomenclature
AT
aT
d
Dv
DZ
ftp
G
gc
H
h
hs
Hw
j
k
LA
LS
LB
LF
LP
M
m
nT
P
Pr
qf
R
T
t
U
VS
VT
W
x
Z
D
q
l
a

Cross-section area, ft2


Acceleration, ft/s2
Diameter, in.
Valve space, ft
Length of increments, ft
Two phase flow friction factor
Gravity acceleration, ft/s2
Dimensional constant, lbm ft/lbf s2
Height, ft
Heat transfer coefficient, Btu/ft2 jF
Reservoir static head, ft
Well depth, ft
Productivity index, bbl/psi day
Ratio of heat capacities
Gas column above the slug, ft
Slug length, ft
Gas column below the slug, ft
Length of liquid fallback, ft
Length of produced slug, ft
Gas molecular weight, lb/lb mol
Mass flow rate, lbm/s
Number of gas moles, lb mol
Pressure, psi
Prandtl number
Oil flow rate into the well, ft3
Universal constant gas
Temperature, jF
Time, sec
Overall heat transfer confident, Btu/ft2 jF
Slug velocity, ft/s
Volume, ft3
Mass flux, lbm/ft2 s
Flow quality (mg/(mg + ml)
Compressibility factor
Deviation
Density, lbm/ft3
Dynamic viscosity, lbm/ft h
Void fraction (Vg/(Vg + Vl)

el
f
g
i
l
n
s
sp
tp
TV

Refer to two successive time increment


Acceleration
Above the slug
Below the slug
Casing

Elevation
Friction
Gas
Initial
Liquid
No slip
Reservoir
Separator
Two phase
Control valve

SI metric conversion factors


cp  1e  03 = Pa. s
ft  3.048e  01 = m
ft2  9.290304e  02 = in.2
ft3  2.831685e  02 = m3
lbf  4.448222 = N
Psi  6.89475 = kpa

Appendix A . Governing equations and solution


technique
Pressure drop is due to friction loss, acceleration
loss and elevation change through the piping system:

dp

dz

Subscripts
1, 2
acc
A
B
CV

253

dp
dz

dp
dz

dp
dz

dp
dz

el

dp
dz

 
dp

dz acc
f



ftp W 2
1 g

q
144 gc n 2:9652  1011 qn d 5
f

el

1 g
q
144 gc

acc

1 g
aT
144 gc

254

S. Ayatollahi et al. / Journal of Petroleum Science and Engineering 42 (2004) 245255

To calculate the key parameters of Fig. 2, first the


operating valve pressure ( PTV2) at time increment
t + Dt is calculated:

PTV1 PTV2
2

PTV

14

jps  p TV
5:614
86400

VS

VS1 VS2
2

a T

VS2  VS1
Dt

DLF

q f Dt
AT

16

LF2 LF1 DLF

17

VT2 LB2  AT

18

DLT Dt V

LA2 DV  LS1 

tDt
X

q f

15

DLT

to

PA2 Psp expM LA2 g=144ZRTg


c

PB2


n 
X
dp
PA2
DZ
dz
j1

nT 2

PB2 VT2
Z2 RT2

19

nT 1

PB1 VT1
Z1 RT1

20

10

DnT nT2  nT1


LB1

t
X
DLT  DLF

11

to

DnT is compared with the net moles of the entering


gas to the casing at time increment Dt (Brown, 1967).
2

LB1 LB1 DLT  DLF


Dn
PTV2 PB2 PLF2 PLP2

PTV2 PB2 exp

M LB2 g
T gc
144ZR

qtp LP2 g
144gc

12


13

To find the operating valve pressure, LF2 is needed


which can be calculated from the following equations
using trial and error technique.

12PTV Ae 4
gc
MRTCV
ZTV



qtp LF2 g

144gc

21

PCV
PTV



0
 
 2k1
k
2k @ PCV
k1
PTV

!# 12
 k1
k
Dt

22

And finally when DnT is converged to Dn the final


solution is achieved.
Heat transfer between the injected gas and the
liquid slug was considered in the simulator using the
following correlation of heat transfer coefficient for
two phase flow regime (Kim et al., 2000). Incremental temperature difference was used for the

S. Ayatollahi et al. / Journal of Petroleum Science and Engineering 42 (2004) 245255

energy balance to calculate the temperature of the


liquid slug.
x 0:04 a 1:21
htp
1 0:27
1x
1a
1  ah1
 0:66  0:72
lg
Prg

23
Prl
ll
References
Abdel-Wally, A.A., Darwish, T.A., El-Naggar, M., 1996. Study
optimizes gas lift in Gulf of Suez Field. Oil Gas J., 38 44.
(June 24).
Ayatollahi, Sh., Bahadori, A., Moshfeghian, M., 2001. Method optimizes Aghajari oil field gas lift. Oil Gas J., 47 49. (May).
Beggs, H.D., Robinson, J.R., 1975. Estimation the viscosity of
crude oil systems. JPT, 1140 1141. (September).
Brill, J.P., Beggs, H.D., 1984. Two Phase Flow in Pipes University
of Tulsa, Tulsa.
Brill, J.P., Doerr, T.C., Brown, K.E., 1967. An analytical description
of liquid slug flow in small diameter vertical conduits. Trans.
AIME 240, 419 432. (JPT, March).
Brown, K.E., 1967. Gas Lift Theory and Practice. Prentice-Hall,
Englewood Cliffs, NJ.
Brown, K.E., Jessen, F.W., 1962. Evaluation of valve port size,
surface chokes and fluid fallback in intermittent gas-lift installation. Trans. AIME 255, 315 322. (JPT, March).
Chacin, J.E., 1994. Selection of optimum intermittent gas-lift

255

scheme for gas-lift wells. University of Tulsa Centennial Petroleum Engineering Symposium held in Tulsa, Oklahoma, U.S.A.,
29 31 August, SPE #27986.
Clegg, J.D., Bucaram, S.M., Hein Jr., N.W 1993. Recommendations and comparison for selecting artificial lift methods. JPT,
1128 1167. (December, SPE #24834).
Dongwoo, K., Ghajar, A.J., 2002. Heat transfer measurements and
correlations for air water flow of different flow patterns in a
horizontal pipe. Exp. Therm. Fluid Sci. 25, 659 676.
Kim, D., Ghajar, A.J., Dougherty, D.L., 2000. Robust heat transfer correlation for turbulent gas liquid flow in vertical pipes.
J. Thermophys. Heat Transf. 14 (4), 574 578.
Liao, T., 1991. Mechanistic Modeling of Intermittent Gas-Lift, PhD
thesis, University of Tulsa, OK, USA.
Neely, A.B., Montgomery, J.W., Vogel, J.W., 1974. A field test
and analytical study of intermittent gas-lift. Trans. AIME 257,
502 512. (SPE Journal October).
Odair, G.S., Sergio, N.B., Francisco, J.S.A., 2001. Study of the
dynamics, optimization and selection of intermittent gas-lift
methodsa comprehensive model. JPSE 32, 231 248.
Poettman, F.H., Carpenter, P.G., 1952. The multiphase flow of gas,
oil and water through vertical flow string with application to the
design of gas-lift installations. Drill Prod. Prac. API, 257 317.
Soave, G., 1972. Equilibrium constants from a modified Redlich
Kwong equation of state. Chem. Eng. Sci. 27, 1197 1203.
Taghae, G., 1984. Aghajari Gas Lift Manual, Ahvaz, Iran. National
Iranian Onshore Oil Company, Iran.
Vasquez, M., Beggs, H.D., 1980. Correlations for fluid physical
prediction. JPT, 968 970. (June 1980).
White, G.W., 1963. An analytical concept of the static and dynamic
parameters of intermittent gas-lift. Trans. AIME 228, 301 308.
(JPT, March).