Journal of Theoretical and Applied Mechanics, Sofia, Vol. 47 No. 1 (2017) pp.

49-68
DOI: 10.1515/jtam-2017-0004

A COMPUTATION FLUID DYNAMIC MODEL FOR

GAS LIFT PROCESS SIMULATION IN A VERTICAL OIL WELL

A R A S H K A D I VA R , E B R A H I M N E M AT I L AY
Department of Chemical Engineering, Faculty of Engineering,
University of Kashan, Kashan, Iran
[Received 30 November 2015. Accepted 20 March 2017]

A BSTRACT: Continuous gas-lift in a typical oil well was simulated using com-
putational fluid dynamic (CFD) technique. A multi fluid model based on the
momentum transfer between liquid and gas bubbles was employed to simu-
late two-phase flow in a vertical pipe. The accuracy of the model was inves-
tigated through comparison of numerical predictions with experimental data.
The model then was used to study the dynamic behaviour of the two-phase
flow around injection point in details. The predictions by the model were com-
pared with other empirical correlations, as well. To obtain an optimum condi-
tion of gas-lift, the influence of the effective parameters including the quantity
of injected gas, tubing diameter and bubble size distribution were investigated.
The results revealed that increasing tubing diameter, the injected gas rate and
decreasing bubble diameter improve gas-lift performance.

K EY WORDS : Continuous gas lift, two-phase flow, computational fluid

dynamic (CFD), optimization.

1. I N T RO D U C T I O N
To meet the ever-increasing global demand of non renewable resources, the oil and
gas industry is forced to rationalize and optimize its production and consumption.
Majorities of the oil wells flow naturally in the early years of their lifetime. The well
is not able to flow to the surface by passing the time and decreasing the driving force.
Continuous gas lift, as one of the most economic artificial lift methods, has been
always a point of interest to researchers led to more effective use of oil resources.
In this process, the injection of high pressure natural gas into the wellbore leads to
lighten the column of fluid and allows the reservoir pressure to flow the fluid to the
surface [1]. The process was sketched in Fig. 1.
Determining the optimal operational conditions in gas lift process has been inten-
sively studied in many papers. The initial step of gas lifting optimization was taken by
Mayhill, developing the first correlation to formulate gas lift performance [2]. Gomez
continued Mayhill study and introduced a two order one [3]. Kanu et al. optimized

Corresponding author e-mail: enemati@kashanu.ac.ir

Unauthenticated
Download Date | 12/12/17 12:14 PM
continued Mayhill study and introduced a two order one [3]. Kanu et al. optimized the
gas lift process economically through a new graphical method [4]. Mahmudi and
Sadeghi [5] used an integrated mathematical model to design and operate a gas lift
system. Moreover several studies have been conducted on design and optimization of
gas lift, using different techniques, involving linear programming (LP), non-linear
programming (NLP), dynamic programming (DP)...etc. [6,7].
50 Arash Kadivar, Ebrahim Nemati Lay

Fig. 1. Schematic of the continuous gas lift process.

Fig. 1. Schematic of the continuous gas lift process

In this work, a multi fluid model developed by CFD technique was used to simulate the
behaviour tohfe gtw as lift process economically through a new graphical method [4]. Mahmudi and
o-phase fluid. The CFD model has be en powerful an efficient tool to
Sadeghi [5] used an integrated mathematical model to design and operate a gas lift
understandsythstemc.oM mopreleovxerhsyevderroaldsytundaiem s ihcasveabned enm coencdh ucatnedisom n sdeo sifgngandloiq ptuimidizatwtioon -phase flows
and it has obfegeans lisfut,cucsiensgsdfiufflelryenttrtieecdhniaqnueds,tiensvotelvdinignlinm eaarnpyrogrreasmem airncgh(eLsP,),innocnl-ulidneianrg the oil and
the gas indust . The accuracy of the multi fluid model w as investigated by means of
p ro g ram m i n g (N L P ) , d y na m ic p ro g ra m m i n g ( D P )..., et c . [ 6,7 ] .
comparison oIfncthi alscwuloarkte, ad mpurletislfsuidemdordoepl dsew veiloth peedxbp y eCrFim
D etencthanliqaun e dwam s uesaesdutroesdimf-ield data.
ulate the behaviour of two-phase lf uid. The CFD model has been powerful and ef-
The multi fficient
luid m odel was used to investigate the dynamic behaviour of two-phase flow
tool to understand the complex hydrodynamics and mechanisms of gas liquid
around thetwinoj-pehcatsieonflop wo s ianntd iint has tbyepen icsaulccoeislsfw ulleylltriu ednd anedr tgesatesd linftmiannythresesao rcuhteh
s, of Iran. The
calculated ipnrcelusdsiungrethde rooilpanads tthheegadsoim ndiunstaryn.t Tphaertacocfurg acaysolfiftht esm im ulutillfautidonmow dealswcao s mpared with
i nv e st ig ate d b y m ea ns of c o m pa ris o n o f c a l cu
other empirical or sem i- mpirical models (i.e. Kabir and Hasan, Orkiszewski, Aziz andl at ed p re ss ur e dr op s w ith e xp e rim e ntal
Govier, Chainedrm iceiasaurned fiSelcdldoactac.hi and Duns and Ros) [8, 9, 10, 11, 12]. To reach the
The multi fluid model was used to investigate the dynamic behaviour of two-phase
flow around the injection point in a typical oil well under gas lift in the south of Iran.
The calculated pressure drop as the dominant part of gas lift simulation was compared
with other empirical or semi-empirical models (i.e. Kabir and Hasan, Orkiszewski,
Aziz and Govier, Chierici and Sclocchi and Duns and Ros) [8-12]. To reach the opti-
mum condition of gas lift process, the influence of operational parameters, including
tubing diameter, injected gas rate and bubble size distribution was studied.

Unauthenticated
Download Date | 12/12/17 12:14 PM
 A Computation Fluid Dynamic Model for Gas Lift Process Simulation in ... 51

2. T WO - P H A S E FL OW M O D E L L I N G
In this work, the CFD model was used to introduce a procedure for the analysis of
two-phase flow in a continuous gas lift system. A multiphase flow has been ex-
pressed, using its governing equations. The multiphase flow considered in this study
consists of liquid as continuous phase (q = l) and gas as dispersed phases (q = g). By
neglecting mass transfer between phases as well as assuming incompressible isother-
mal multiphase flow, the governing equations can be written as [13]:
Conservation of mass:
(q q )
(1) + (q q Uq ) = 0 .
t
Conservation of Momentum:

(q q Uq )
(2) + ( U U )
q q q q
t
= q pq + q q g + (q (Sq + S Re
q )) + (pi Pq ) q + Fq ,

where g, pi , q and q are the gravitational acceleration, the interfacial pressure, the
volume fraction and the density of phase q, respectively. Uq , pq and Sq a re mean
velocity, pressure and viscous stress of phase q. The unknown terms including inter-
facial force density, (Fq ), Reynolds stresses of phase q, (Sq ) and interfacial pressure
difference, (pi Pq ), which arise from averaging of the instantaneous momentum
equation and require modelling.
2.1. I N T E R FAC I A L P R E S S U R E D I FF E R E N C E

Lamb considered the potential flow around single phase and expressed interfacial
pressure difference, as follows [14]:

(3) pi Pq = Cp c |Ur |2 c ,

where Cp = 0.25, |Ur | is the relative velocity and c and c are density and void
fraction of continuous phase, respectively. Drew pointed out, that by assumption
of incompressible phases and without expansion or contraction there is microscopic
instantaneous pressure equilibrium, i.e., Cp = 0 [15].
2.2. I N T E R FAC I A L FORCES

For bubbly flow in pipes, the interfacial force acting on a dispersed phase is decom-
posed into several terms, expressed as

(4) Fd = F drag + F lif t wall td vm

d d + Fd + Fd +F d .

Unauthenticated
Download Date | 12/12/17 12:14 PM
52 Arash Kadivar, Ebrahim Nemati Lay

The components of interfacial force density are drag, lift, wall, turbulence dis-
persion and virtual mass, respectively. It was assumed that in bubbly flow regime
the interfacial force between dispersed phases were neglected [16]. Therefore, by
Newtons law the interfacial force acting on continuous phase is

(5) Fd = Fc

The drag force acting on the dispersed phase travelling steadily through the fluid
is illustrated as [17]

(6) d = Klg (Ul Ug ) ,

3 CD
F drag
(7) Klg = g l ,
4 d

where d is the bubble diameter of a dispersed phase and CD is the drag coefficient
and can be written by the Schiller and Naumann correlation [18].

(
24(1 + 0.15Re0.687
d )/Red Red 1000
(8) CD =
0.44 Red > 1000
d|Ur |
(9) Red = ,
l

where Red is the particle Reynolds number and l is the liquid (as continuous phase)
kinematics viscosity.
A bubble moving through a fluid that is in a shearing motion will be subjected to
a lift force transverse to the direction of motion. In general, the lift force is expressed
as [15]

(10) F dlif t = CL g l Ur ( Ug ) ,

where CL is the lift coefficient. Different constant values are reported for this coeffi-
cient. In this study, the modified lift coefficient was used, based on the Tomiyamas
correlation and achieved by Behbahani et al. [16,19].

(11) CL = CLmod CL,Tomiyama ,

Unauthenticated
Download Date | 12/12/17 12:14 PM
 A Computation Fluid Dynamic Model for Gas Lift Process Simulation in ... 53

0.2 Eo d < 1.17

0.1 1.17 < Eod < 2.65
0.05 2.65 < Eod < Eod,crit = 6.06
(12) C mod
L = 0.13 6.06 < Eod < 9.5
0.04(14.9 Eod)
+ 0.09 9.5 < Eod < 14.9
5.4

0.04(23.5 Eo d )
+ 0.05 14.9 < Eod < 23.5
8.6
Eod 0.5
(13) dH = ,
g(l g )

where Eod is the Eotvos number, is the surface tension coefficient and dH is the
long axis of a deformable bubble.
Liquid flow rate between bubble and wall is lower than that between bubble and
outer flow. This hydrodynamic pressure difference is the origin of wall force. There
are several models for wall force, which have been reported in the literature. Here,
Tomiyamas model, which has been proposed for flow in pipe geometry with tuned
wall force coefficient [16,19] was selected which has the following general form:

l dg |Ur |2 1
(14) F dwall = CW nW ,
2 y2
1

W (D yW )2
where D is the tubing diameter, ym is the distance from wall and nW is the wall nor-
mal vector. The modified wall force coefficient, based on the Tomiyamas correlation,
concluded by Behbahani et al. [16,19], was used in this study.
mod C
(15) CW = CW W,Tomiyama ,
(
0.1 Eod < 1.17
(16) CLmod =
0.05 Eod > 1.17

Different models are introduced to estimate the interfacial turbulent dispersion

force. As the Favre averaged- drag (FAD) model leads to accurate predictions for
bubbly flow in vertical pipes, the FAD model was used in this study, which can be
written as [20]
t
Dlg l g
(17) F td = Klg Udr = Klg ,

d
pr lg l g
where Udr is a drift velocity, pr dc is a dispersion Prandtl number equal to 0.75 [16],
t is the turbulent diffusivity.
and Dlg

Unauthenticated
Download Date | 12/12/17 12:14 PM
54 Arash Kadivar, Ebrahim Nemati Lay

Another component, the virtual-mass force, is considered when bubbles acceler-

ate. The virtual mass force is calculated by Drew [15]
Dl D U
(18) F dvm = Cvm g l g g ,
Ul Dt
Dt

where Dp /Dt is the material time derivation of each phase, both continuous and
dispersed phase, and Cvm is the virtual mass constant and is equal to C m = 0.5.

2.3. T U R BU L E N C E I N M U LT I - P H A S E M O D E L L I N G

For gas lift optimization, it is usually desired to operate in the bubbly flow regime
[21]. Two-equation turbulence models have been commonly used for bubbly flow in
the vertical pipes. Here, the k model, introduced by Harlow and Nakayama [22]
was used, where k is the turbulence kinetic energy and is the rate of dissipation of
the turbulent energy.
(l l kl )
(19) + ( k U ) = T Re : (U )
l l l l l l l
t
l l
t
kl l l l + Rk
pr K E

(l l l )
(20) + ( U ) = C T Re : (U ) C
l l l l 1 l l l 2 l l l
t k

l lt
+ R
pr DR l

Using the definition of the turbulent length scale the eddy viscosity, ct is written as

k 2l
(21) lt = C ,
l

where TcRe is the Reynold stress. The source terms, Rk and R , were used based on
BelFdhila and Simonins correlation [23]. The constants in Eqs (19), (20) and (21)
take the values, given in Table 1 [13].

Table 1. The values of the constants in the k model

C C1 C2 pr K E pr DR
0.09 1.44 1.92 1.0 1.272

Unauthenticated
Download Date | 12/12/17 12:14 PM
 A Computation Fluid Dynamic Model for Gas Lift Process Simulation in ... 55

2.4. PRESSURE V EL OCITY CO UPLIN G

Here, the adopted PISO algorithm for two-phase flow developed by Issa and Oliveira
was used for the pressure-velocity coupling [24,25]. It involves one predictor step
and two corrector step. The steps of the algorithm are defined as:
1: The continuous phase momentum equation is solved
X
(22) Al + l l u = H n (u ) P B P [P n ]P
V

0 il l il l ji j
t
c c V
+ Fq un un V + S c + un .
ig il ui ic
t
2: The dispersed phase momentum equation is solved

X
g g V
(23) g
A + +F V u = H n u P B P [P n ]P
0 q g ig g ji
t ig
j

d V
+ Fq u V + S l + un .
il ui
t il
Where , p and n denote intermediate value, considered phase and time
(or iteration) level, respectively. The superscripts P for the cell-center and f
for the face along direction l = j were considered for the locations of variables
were
P computed. The P contribution of surrounding cells will be denoted as H () =
Af f , with A0 = Af . Also, all quantities were assumed to be located at cell-
f f
center. Bji is the i-component of the area-vector along j-direction, Bf is the face
area and V is the cell volume and S ku contains all terms not explicitly written.
3: The corrected pressure p0 is assembled using the overall continuity equation. The
pressure and velocities are updated based on Issa& Oliveira equations [26]
X
(24) P pP = Af p + Su ,
0 p 0
Ap f
p

l l V p
X 0 P
un+1 P

(25) uil = l Bi P ,
t il j

g g V X
un+1 u B P P 0
P
(26) + Fq V = p ,
ig ig g i
t j

(27) P n+1 = P n + P 0 .

In the same way, the fluxes, F , are corrected and defined, as [27]
0 f
F fn+1 P
g = Ff g Af g P ,
f
(28)
Unauthenticated
Download Date | 12/12/17 12:14 PM
 = Ffl AfPl P 0
f
F fn+1
l f

Unauthenticated
Download Date | 12/12/17 12:14 PM
56 Arash Kadivar, Ebrahim Nemati Lay

4: The turbulence quantities should be solved for k and . With new values of k and
, liquid and gas effective viscosities are updated.
5: The void fraction is updated through dispersed phase continuity equation, which
is solved implicitly.
d V
(29) Aa + + M ax [div (u ) , 0] = H n ( )

o g a
t
g V n

+ M ax [div (ug ) , 0] +
t
The solution will be advanced in time until the residuals of all the equations are
smaller than a specified value.

3. PRESSURE D RO P C A L C U L AT I O N

The prediction of pressure drop for bubbly flow in the tubing is very important in the
gas-lift. The overall pressure drop for gas-liquid flow can be written as the sum of
three individual components, i. e., acceleration, gravitational and frictional pressure
losses.
p p p p
(30) = + +
x x A x H x F

In the bubble flow regime, the gravitational component usually forms more than
90% of the overall two-phase pressure drop [28]. As the drift flux model gives very
good predictions of the void fraction for bubbly flow regimes, the gravitational com-
ponent of the overall two-phase pressure gradient is estimated using the model, based
on drift flux model, presented by Zuber and Findlay [29].
p g
(31) = m ,
x H
gc

(32) m = g g + (1 g ) l ,

where m is the in-situ mixture density, calculated through gas void fraction, which
estimated by multi fluid modelin each step of calculation. There are many empirical
and semi-empirical correlations for predicting this parameter. Here, the gas void
fraction was calculated via drift-flux model [29]. In general, the acceleration term in
right-hand side of Eq. (22) could be neglected in the bubbly flow [8]. The frictional
component can be roughly predicted by the correlation, suggested by Wallis [28].
p 2
2fm vm m
(33) = ,
x F gc D

(34) vm = vsl + vsg ,

Unauthenticated
Download Date | 12/12/17 12:14 PM
 A Computation Fluid Dynamic Model for Gas Lift Process Simulation in ... 57

where vm , vsl and vsg are mixture, liquid and gas superficial velocity, respectively.
p is the pressure, g is yhe acceleration due to gravity and gc is the conservation
factor, (32.2 lbm.ft/ft/s2 ). The friction factor, fm , was estimated from an empirical
correlation reported by Chen [30].

4. B O U N DA RY CONDITIONS

It is impossible to simulate the whole tubing with actual length of 2531 m. On the
other hand, as fully developed flow was assumed at the outlet, the length of the tube
should be considered long enough to have a fully developed profile at the outlet
boundary. Therefore, only the lowest 15 meters of the well, at least 80D, is regarded
using Behbahani et al. criterion [16]. Since the computational domain is symmetric
around the center axis, the symmetry axis boundary condition was employed at the
tubing string center line [16]. At the tubing walls no slip boundary condition coupled
with the standard wall function for the turbulence model [31] was imposed. Uniform
velocity inlets were employed as boundary conditions at the gas and oil inlets. The
influence of the gravitational force on the flow was taken into account. The coarse
mesh size was considered in a way, that the computational domain was divided into
320,000 cells. As can be seen in Fig. 2, a finer grid treatment was employed near the
injection point. A typical grid, boundary conditions, and coordinates system (xr)
are shown in Fig. 2.

Fig. 2F.igT. 2h.eTgherigdridtotoppoolloggyyanadnbdoubnoduarnydcaonryditcioonns.ditions

5. Materials and methods

The two-dimensional (vertical) model was used to study the unsteady
behaviour of two-phase fluid in an oil well under gas lift. The calculations were performed
with a time step of 0.01 s. The governing equations, which were derived assuming
isothermal incompressible multiphase flow, along with the boundary conditions,
have been integrated over a control volume. The subsequent equations have been
discretized over the control volume, using a finite volume method to yield algebraic
equations, which can be solved in an iterative manner for each time step. The conservation
equations are solved by the segregate solver, using implicit scheme. The
discretization form for all the convective variables were taken to be second ordUenrauuthpenticated
winding. Here, the adopted PISOalgorithm introduced by Issa and OlDivowenilroaadwDaatse |u1s2/e1d2/1f7o1r2:14 PM
the pressure-velocity coupling [24, 25]. The k model with the modified coefficients
58 Arash Kadivar, Ebrahim Nemati Lay

5. M AT E R I A L S AND METHODS
The two-dimensional (vertical) model was used to study the unsteady behaviour of
two-phase fluid in an oil well under gas lift. The calculations were performed with a
time step of 0.01 s. The governing equations, which were derived assuming isother-
mal incompressible multiphase flow, along with the boundary conditions, have been
integrated over a control volume. The subsequent equations have been discretized
over the control volume, using a finite volume method to yield algebraic equations,
which can be solved in an iterative manner for each time step. The conservation
equations are solved by the segregate solver, using implicit scheme. The discretiza-
tion form for all the convective variables were taken to be second order up winding.
Here, the adopted PISOalgorithm introduced by Issa and Oliveira was used for the
pressure-velocity coupling [24, 25]. The k model with the modified coefficients by
Behbahani et al. [17] was used to treat turbulence phenomena in both phases. In this
study, the residual value of the mass, velocity components and volume fraction was
considered as a convergence criterion. The numerical computation was considered
converged, when the scaled residuals of these variables in the control volume lowered
by four orders of magnitude.

6. MODEL VA L I DAT I O N A N D C O M PA R I S O N
In order to investigate the validity of the multi fluid model, the calculated total pres-
sure drops, as the most important characteristic of a two-phase flow, were compared
with the experimental data. Of the 44 test data, 43 were reported by Dhotre and Joshi
[32] and one was taken from Mahmudi and Sadeghi [5]. According to Dhotre and
Joshi, data were taken in a way to cover a wide range of column diameter and gas
velocity inlet for the bubbly flow regime. The statistical results were obtained using
arithmetic average of percent deviation, AP D, and standard deviation, SD.
P
n
P Di
i=1
(35) AP D = ,
v n
uP
u n
u (P D i AP D)2
t i=1
(36) SD = ,
n 1

PBH,m PBH,c
(37) PD = 100 .
PBH, m

The results of statistical analysis, applied to all 44 test data are shown in Ta-
bles 2 and 3. The small values of AP D and SD imply the good agreement between
measured pressure drops and calculated ones. For example, the overall percent and

Unauthenticated
Download Date | 12/12/17 12:14 PM
A Computation Fluid Dynamic Model for Gas Lift Process Simulation in ... 59

Table 2. Summary of data and comparison results

Code vl (m/s) vg (m/s) p/l (Kpa/m) PD

Measured Calculated
D 0.05 0.13 9.44 8.90 5.71
D 0.05 0.21 10.35 10.13 2.05
D 0.05 0.29 11.63 11.44 1.64
D 0.05 0.37 12.21 11.97 1.95
D 0.05 0.45 12.48 12.20 2.28
D 0.09 0.13 8.92 8.09 9.24
D 0.09 0.21 10.67 10.24 3.96
D 0.09 0.29 11.15 11.12 0.3
D 0.09 0.37 11.12 11.20 -0.73
D 0.09 0.45 11.49 12.52 -9.00
Airwater

D 0.14 0.13 9.2 8.80 4.39

D 0.14 0.21 9.02 8.98 0.37
D 0.14 0.29 9.6 9.14 4.82
D 0.14 0.37 9.74 9.74 0.01
D 0.14 0.45 9.44 9.54 -1.06
D 0.19 0.13 9.51 8.87 6.71
D 0.19 0.21 9.55 8.70 8.86
D 0.19 0.29 9.68 9.50 1.84
D 0.19 0.37 10.34 9.99 3.29
D 0.19 0.45 12.59 12.87 -2.16
D 0.05 0.13 9.29 8.77 5.67
D 0.05 0.21 7.29 6.89 5.47
D 0.05 0.29 6.88 6.89 -0.2
D 0.05 0.37 8.14 8.11 0.38
D 0.05 0.45 8.45 8.53 -1.02
D 0.09 0.13 7.48 7.27 2.85
D 0.09 0.21 7.98 7.73 3.16
D 0.09 0.29 8.49 8.37 1.36
D 0.09 0.37 8.17 8.16 0.06
D 0.09 0.45 9.08 9.11 -0.41
Airethanol

D 0.14 0.13 7.88 6.79 13.81

D 0.14 0.21 7.52 7.19 4.42
D 0.14 0.29 7.77 7.65 1.54
D 0.14 0.37 10.07 10.39 -3.15
D 0.19 0.13 7.63 7.08 7.26
D 0.19 0.21 7.88 7.24 8.10
D 0.19 0.29 8.17 8.05 1.39
D 0.19 0.37 8.47 8.43 0.46
D 0.19 0.45 8.95 8.80 1.62
Gasliquid M 2.48 3.41 9.53 10.01 -5.30
Code: D Dhotre and Joshi [32]; M Mahmudi and Sadeghi [5]

Unauthenticated
Download Date | 12/12/17 12:14 PM
on is normally distributed about their mean value, the measured pressure drops
be predicted within 6.13 percent (l standard deviation) of the average percent
nce. This statistical analysis is also exhibited graphically in Fig. 3. The results
e the ability of the multi fluid model to predict pressure drop in the bubbly flow
. 60 Arash Kadivar, Ebrahim Nemati Lay

14
Calculated pressure drop (Kpa/m)

13
12
11
10
9
8
7
6
6 8 10 12 14
Measured pressure drop (Kpa/m)

Fig. 3. CompaFirgi. s3.oCnomopafrism

on e
ofamseu
asrue
red anadncadlcuclaa
teldcpu
relssautreegd
radp
ierne
t. ssure gradient

standard deviation values are 2.16 and 6.13 percent, respectively. It indicates, that
if percent deviation is normally distributed about their mean value, the measured
pressure drops would be predicted within 6.13 percent (l standard deviation) of the
average percent difference. This statistical analysis is also exhibited graphically in
Fig. 3. The results indicate the ability of the multi fluid model to predict pressure
drop in the bubbly flow regime.

Table 3. Overall comparison results

Method APD SD
Multi fluid method +2.16 6.13

7. GAS LIFT S I M U L AT I O N
The multi fluid model was used to investigate the performance of gas lift in an oil
well. The typical well, considered in this study, is located in one of the oil fields in
the south of Iran. It started to produce light to medium crude oil in 1968. As time
progressed, the reservoir pressure dropped as several wells did not flow naturally and
were put on to gas lift. The study consists of a typical well tubing string with a
diameter of 0.127 m and a length of 2531 m. The tubing contains crude oil, which
hydrocarbon gas enters at the specific injection point. The computational domain is
shown in Fig. 4. In addition, the properties of injected hydrocarbon gas are given
in Table 4, while the relevant information and properties of the well and crude oil is
illustrated in Table 5.

Unauthenticated
Download Date | 12/12/17 12:14 PM
the south of Iran. It started to produce light to medium crude oil in 1968. As time
progressed, the reservoir pressure dropped as several wells did not flow naturally and
were put on to gas lift. The study consists of a typical well tubing string with a diameter
of 0.127 m and a length of 2531 m. The tubing contains crude oil, which hydrocarbon
gas enters at the specific injection point. The computational domain is shown in Fig. 4.
In addition, the properties of injected hydrocarbon gas are given in Table 4, while the
A Computation Fluid Dynamic Model for Gas Lift Process Simulation in ... 61
relevant info rmation and properties of the well and crude oil is illustrated in T able 5.

Fig. 4F.igT
. 4h. e
Thceocm
ompputtaa
titoinoalndaolmd
aio
n.main

TableTa5bl.eI4n. jIenjcecteted ggaas sproppreortpiees rties

Symbol
Symbol
Definition
Definition
value
value
Density (kg/m3)
Density (kg/m3 )
145.60
145.60
g

vg Visicsocsoitysi(tKyg(/m
V K/gs)/m/s) 00.0.00020
02 2

QQgg GaGsasininjjeectiioonnrartae t(ekg(/ks)g/s) 22

767.760.70

dd BBuubbbblle ddiaim
ameter (mm)
ter (mm) 3
3
Table 5. Typical well information

Symbol Definition Value

BHP Bottom hole pressure (Psi) 2920.15
TBH Reservoir temperature ( C) 92.59
TT op Surface temperature ( C) 27.86
GOR Gas oil ratio (scf/bbl) 506
L Length of tubing (m) 2531
D Tubing inside diameter (m) 0.127
API Oil API 30
l Viscosity of crude oil (cP) 0.207
WC% Water cut 0.00
Surface tension (N/m) 0.0086

Unauthenticated
Download Date | 12/12/17 12:14 PM
 TTop Surface temperature ( C) 27.86
GOR Gas oil ratio (scf/bbl) 506
L Length of tubing (m) 2531
D Tubing inside diameter (m) 0.127
API Oil API 30
Viscosity of crude oil (cP) 0.207
WC% Water cut 0.00
Surface tension (N/m)
Arash Kadivar, Ebrahim Nemati Lay
0.0086
62

8. Results and discussion

8.
TRheE SpUeLrTfoS rAmNaDncDeI SoCfUgSaSsI OlNift on the oil production is shown in Figs 5 and 6. In
theseTfhiegupreersfo, rm thaencm e ioxftugraes d lieftnsointyt,hesuopilerpfriocdiaulctiooinl ivs eslhoocw itny iannFdiggs.as5 vanodid6.fraInction at
different sections of the tubing are in dicated. A s it is obvioaunsdingatshevsoeidfifgruacretiso,naat tthe gas
th es e f
i g u re s , th e m ix tu r e de n s ity , s u p e r f
i c ia l oi l v el o ci ty
injectdioifnferpenotinstectthioenrse oifs thae ttuurbnininggarpeoiinndti.caW ted.henAsgaits isisobinvjieocutseidn, tgheas vfiogluurm ese, aftraction
increases significantly, so the flu id density decreases considerably at this pouim
th ga s i j e c t io n p o i n t t h e re i s a tu rn i n g po i n t. W h en ga s i s in j ec ted , g as v ol nte. In the
bubblfyracftliow n inrcergeiam sees, sgigrnaivfiictatnitolyn,asl optrheessfluuried dro nspityisdeth creeadseosmcionnasnidt ep raabrltys aot fthtihse total
pressuporeint.dIrnopth,e tbhuebrbelfyolfreo,w trhegeimper,egsrsauvrietatigornaadl ipernetssudrecdrreoapseiss thseigdnom ifiicnantlpyartasboofve the
injection valve. It enables the BHP to accelerate the velocity ofifitchaentflyloawbionvge liquid
t h e to ta l pr e ss u r d ro p , the re fo re , t he p re s su r e g r a d ie n t de c re as e s sig n

vertictahlelyinje
upcwtioanrdvaalnved. iInt cerneaabsles tohiel BpH roPdtuocaticocnel,earasteththeeoviel lvoecitloycoitfythwe aflsow
inicnrgelaisqeudidto 0.29
m/s avterthtiecaolluytuleptw
. aArdlsaon, ditinicsresahsoew s onilinproFdiguc. tion
6 b, aths atthebyoilpvueslhoicnitgy twhaes oinilc,resauspeedrftoicial oil
veloc0it.29
y inmc/rseatstehseaobuotlveet. tAhelsoin, jiet cistesdhopwoninitn. Fig. 6b that by pushing the oil, superficial
oil velocity increases above the injected point.
800
(a)
760
Mixture density

720

680
640
0 5 10 15
X (m)

Fig. 5.Fig.
The effect of injected gas on the mixture density and static pressure
5. The effect of injected gas on the mixture density and static pressure.
0.6 0.5 X=2m
X=2m
(a)0.6 X=2m
0.5
(b) X=2m
(a) X=2.05m (b) X=2.05m
X=2.05m X=2.05m
X=3m 0.4 X=3m
X=3m 0.4 X=3m
Oil velocity (m/s)

0.4 on X=10m X=10m

0.4
Gas void fraction

ti
ca X=10m
X=15m X=10m
fr X=15m X=15m 0.3 (m/s) X=15m
y
vo it 0.3
id
s 0.2 vel
0.2 Ga oc
0.2
Oil 0.2

0 0 0.1
0 0 20 20 40 60 0 0.1 20 60
r (mm) r40(mm) 60 r (mm) 40 60
0 20 40
r (mm)

Fig. 6.FTigh.e6v.aTrhiaetivoanrioaftigoansovfogidasfrvaocitdiofnra(cat)ioannd(ao) ialnsdupoeirlfsiucipaelrvfieclioaclivtyel(obc)itiyn (tbh)e itnubthinegtubing

Fig. 6. The variation of gas void fraction (a) and oil superficial velocity (b) in the tubing.
The caTlhcuelcaatelcduglartaevditagtriaovniatal tpiorensasluprreedssruorpe bdyrom publtyi m fluuildti m floudidelmwoadselcowmaspacroem d pared
with thosthe proe d icted b y sd ome em peirica l an d s emi e pirica lp m h dseot[dh8eo,ld9w
w iT het h c a s e l c u
p l
rae t
d ei dc tge r a v
b i
y ta stoi o
m n a l e p
m r e
p s
i s
ri uc rael da r
n od p b
sm
e y
m m
i e um lt i i l
f
r u
ie i
ct ad lo m s,
a[18s0,,c9o1,m p a r
11,01,21]1.e d , 12]. The bwTeisthhte tahbgoerseetem
pargeerndetiecm
bteedlnobtnygbesedolomtnoeg ethmeptoicroitcrhareel
laacntoidrornseelm
aintitreonmdiupncitreriodcadlbum
yceeOdthrbkoyidzseO[v8rsk-k1iiz2e]bv.ysTk0hi.e4bb1ye%s0t .41%
averaagagevreeerram rgoeern.etTrbhroeelrop.nT rgehdeedicptioroentdhsiecbticyoonK rsreablbayitrioKannabdinirHtraoansddauncH eadansdbaynAO aznirzdkieA
ztezvailsz.kewi
tbeayrel.0tw .h4e1r% seatmahveersaangm dee and relativreellyatsivateilsyfascattoirsyfabctyor3y.3b0ya3n.d303.a6n3d%3e.6r3ro%r, ererrsopre,crteivspeelyc.tiTvheleyc.
oTrhrelactoirorneldaetivoenlodpeevdeloped
by ChbieyriC cihaienrdicG i aiuncdcG i biuyc6ci.2b5y%6.e2r5ro%r w erarsorinwtahse inetxhtelenveexlt. lTevhe .wTohrset w proerdsitcptiroendiw ctaiosn was
c o rnela
by cobrryelatiro in rodiucted y Du t ito n n ro db u c ed bn ys D
and osd, w ic o er p eedricpte the grta triaovniatal tional
u n sRa n Rh o s ,hw hv ic h orv rd
e di c t ed hv eita
g
pressuprreedssruorpebdyrolpargbey elarrogre ueprrotor 1 u1p% to. 1It1c%an. Ibt ecacnonbcelucdoendc,luthdaetdt,htehamt uthltei m fluuildti m floudidelmodel
is ableistoabplreedtoicptrperdeiscstuprreedssroupreindrooipl w ineollisl wietlhlsaw n iathccaenptaacbcle atcacbuleraaccyc.uracy.

8.1. O8p.t1i.mOupmtim opuem raotipoenraaltcioonnadlitcioonndition

An opA n aolpgtiamsallifgt aospleirfat toiopnerm
tim atuiosnt jm usutsiftyjuthstriefey tyhprees toyfpaeismosf (ai)immsa(xii)mmizaaxtiim onization
of porofducperodduocile,d(ioi)il,m(aixi)immizaaxtiim onizaotfionprooffit parnodfit (iaini)d o(piitimiozpattiim onizadteiosingndeosfigngasof gas
injectiionnjescytisotenms.ysGtem
ne.rG alelyn,etrhaellye,fftehcetievfefepcatirvaempeaterarsmoentetrhseoonptthim e aolpgtiamsallifgtapsrolicfetspsroccaenss can
be catbegeocraizteegdoorinzetdwo nimtw poritamnptocrrtiatnerticar. iT tehriea.fiTrshteofniersitsoonpetiim s aolpotipmearaltoinpgerpaatirnagmpeaterarsmeters
like ploiskietiopnosfiotironinjfeocrtiionnjecptoioinnt, ptouibnitn,gtu(boirngstr(ionrg)stsrizeg)asnidzethaendwathye ow f aiynjeocfteidnjegcatsed gasUnauthenticated
distribduitsitornib.uTtihoen.oTthheer onheerisoneestiim s aetsintigmathtiengoptthiemaolptgim asalingjeacstiionnjecratitoen. A rasDteoow. nnA
loysadtohDnealtye |t1h2e/12/17 12:14 PM
lower lpoawrterofpathrte otuf bthinegtuisbisnim
g uislastiemdu,ltahteede,fftehcet eofffethcte odfetphteh doefpitnhjeocftiionnjepcotiionnt cpoouinldt cnooutld not
be discbuesdsiesdc.uH ssoewd.eH
veorw, tehveeor,ththeer poathraem
r peaterarsmaerteerisnvaerestigvateesdtiginattehde innetxhtesencetxiotnse.ction.
 A Computation Fluid Dynamic Model for Gas Lift Process Simulation in ... 63

error. The predictions by Kabir and Hasan and Aziz et al. were the same and rela-
tively satisfactory by 3.30 and 3.63% error, respectively. The correlation developed
by Chierici and Giucci by 6.25% error was in the next level. The worst prediction
was by correlation introduced by Duns and Ros, which over predicted the gravita-
tional pressure drop by large error up to 11%. It can be concluded, that the multi
fluid model is able to predict pressure drop in oil wells with an acceptable accuracy.

8.1. OPTIM UM O P E R AT I O NA L C O N D I T I O N

An optimal gas lift operation must justify three types of aims (i) maximization of
produced oil, (ii) maximization of profit and (iii) optimization design of gas injection
system. Generally, the effective parameters on the optimal gas lift process can be
categorized on two important criteria. The first one is optimal operating parameters
like position for injection point, tubing (or string) size and the way of injected gas
distribution. The other one is estimating the optimal gas injection rate. As only the
lower part of the tubing is simulated, the effect of the depth of injection point could
not be discussed. However, the other parameters are investigated in the next section.

8.1.1. E FFE C T O F G A S I N J E C T I O N R AT E

The quantity of injected gas is one of the most important parameter, which directly
affect the performance of gas lift. Figure 7 shows the variation of gas void fraction,
oil superficial velocity and frictional pressure gradient in different axial distance from
the injection point. The results concluded using bubbles with 3 mm diameter, as
observed in Figs. 7a and 7b, the higher amount of injected gas, the larger gas void
fraction near the wall and center of tubing. The density of the oil reduces by aerating
the fluid resulting in lightening the two-phase flow fluid. According to Eq. (23),
a lightened two-phase fluid developed less gravitational pressure drop, hence more
amount of oil produced. It is clear in Fig. 7c, that the 24.5% increment in oil velocity
is observed, when the gas injection rate is increased by 149.3%.
However, by increasing the amount of gas injected higher amount of fluid passes
through tubing. As it can be notified in Eq. (25), higher mixture superficial velocity
causes increasing the frictional pressure drop, accordingly (see Fig. 7c). The upward
trend of frictional component results in decreasing the positive effect on oil produc-
tion until reaching the economical optimum point (E.O.P.), where well fluid density
reduction is equal to friction force increasing. At this point, the maximum quantity
of well production is obtained. When the rate of injected gas keeps increasing, the
reverse effect on production will be observed. Moreover, increasing the quantity of
injected gas leads to increasing operational costs, which should be taken into consid-
eration.

Unauthenticated
Download Date | 12/12/17 12:14 PM
 paspsaesseths rothurgohugthubtiunbgi.ngA.sAits citancabne bneotnifoiteidfieidn iEnq.Eq(2. 5()2,5h),ighhiegrhemr ixmtuixreturseupseurpfiecrifaicl ial
vevleolvcoeictliyotycciatcyuauscesaseusisneincscrei irneaccasrirseneinagasgsitinhntghgeetfthrhfireecictffitroriioicncntatiaiololnpnpararlelespspsrsureuersressesudurdreroeodppdr,,oraoapcpc,cc,aooacrrcddoiinorndgrgdillniyng(gl(ysley(es(eF sFeigeF.F i77
gi.gc.7)..7cT)c.h
)T .ee T
hehe
upuw pw uappradwrdatrrtedrnendtrdeonofdffo rofirffcictffitroriiionccntatiialoolnncacoalolmmccpopoom omnnpepeononntnteernrenetstsururleltestsssuuliitlnnstsdidnieenccdrrdeeeacacsrseirinenagasgisnittnhghgeethtpheoespipotiosvisetiitvieevffeecfeftfefcoetncto
o i l
n
onil oil o il pr propdruocdtuinocntiuou
o d u c t i o nnntitluilnrtteriiaell acrrcheehiaaniccnghhgiitnnhtghgeettehheceecooneencocom om nniociocm amalilicocaoaplplttioim ompptuuitm
immuupm pm ooiipnnpottoin(i(nE t
t.(OE (E..PO ..O
).P ,
,.P.w).,)h,wew rheehrew erew llw i
eflelullfidlfuliudid d
dednesndisteiyntysrietrydeu drucetcditouioncntiiosinnseieqssquuaeealqqluuto ataoll ftrftooircictffitroriioicncnttiiofofononrrcfcefoeorrciicneneccrirenienacacsrsierinenagasgsi..ninA
gAg.tt.
AttA hhtiitssththpipsiosipnpot,oinittnh,te,thtm eheam xm iamxauiuxm im
m um um
quqaunaqtnuitaiytnyotifotyfwo weflelw llpeprlorlodpdurrucoctddito uiuonccnttii osoisnnoob iisbstatoaoibnibntetaeadidin.n.eW
eW dd.h.hW eWennhhtethehnenetrhtrahaeteteerarooatfetfeioinnofjjfeeicncintjteejedctegtedadsgk gakseaeesekpkesepienspcsirneicnarscseiirnanesga
,i,nign,g,
s
thtehertehrveevrreesrveseerfsefefefcetcfftoeoncntpoprnnorodppdurruococtddituoiuocncnttiw ioowninlilw lw bibiellellobobbebessoeoerbrbvsvseeedrdrv.v.eM eM dd.oo.MrrM eeooorvvereeorrov,,veiinern,rcc,irrneiencarcser inaesgagisntihngegthqtehueaqnuqtauitnay
tniotoyifftyofof inijnejceitncejtdecdtgedagsasglaelsaedaldlsesadtotsso ittnoionccrierinneacacsrsirenienagasgsiinnogogppeeororapaptetieiororanantaitaoilolnnaccaloolsscttscos,o,sstsw tw s,
,hhiw c
w h i s
c s lh o u ld
i chic hshosuhloduldbe bteakteankenintointo t
b e t a k i
e n t i n to
cocnosncisd o6ine4drseairdtaietoironan.ti.on.. Arash Kadivar, Ebrahim Nemati Lay

0.4 (a) Qg =3.64 (m3/s) Qg =3.64(m3/s)

0 (b)
Qg =2.45(m3/s) 0.8 Qg =2.45(m3/s)
0.3 (a) Qg =3.64 (m3/s)
Qg=3.64
=3.64(m3/s)
(m3/s ) Qg =1.46(m3/s)
.4 (a)
(a) Qg

Gas void fraction

0.4 Qg =3.64(m3/s)
QgQg=3.64(m3/s)
=3.64(m3/s)
Qg =2.45(m3/s)
Qg=2.45(m3/s)
Qg =2.45(m3/s)
0.8 (b)
0.8
0.8 (b)
(b)
0.6 QgQg=2.45(m3/s)
Qg =2.45(m3/s)
=2.45(m3/s)
Gas void fraction
to

Qg
Qg =1.46 =1.46
=1.46(m3/s)
Qg(m3/s) (m3/s)
in .3ction 0.3 o ti
n onon QgQg=1.46(m3/s)
Qg =1.46(m3/s)
=1.46(m3/s)
ac a ti
0.3 c .6 ticcti0.6
0.6
r frcfr a raf ra
f

0.2 fra
f
0.2 s 0 0.4
oid .2 void void 0.2 v oid.4 voidvoid0.4
v

s ss
a Ga G 0.4s
G

.1
Ga s
0.1 0.1 0.1
a0.20.2
GaGa
0.2
0.2
0
0
0 2
00 4 6 8 10 12 14
00 22 44 X (m)
66 8 10 12 14 00 00
0 2 4 6 8X (m)8 10 1012 1214 14
00 00 55 55 10
10 1010 15 15 15
X (m)X (m) X (m)X X
X (m) (m)
(m)
0.45 1
1.8 (d)
(c)
0.40.35 0.45
5 0.45 .8 1.8
1 1.8
Oil Velocity (m/s)

(c) (c)
(c) 1.6 (d) (d) (d)
(

0.35 .6 1.6 1.6

0.3 5 0.35
) ))
dp/dx (Kpa/m)

m/s) (m/s)
(m/s) /m 1.4 /m/m1.4
pa 1 .4papa 1.4
5 locity
0.2
0.25 (K (K
ocity locity (K
O V l

0.250.25 1.2 1.2

1 .2 1.2
Qg=3.64 dp/dx dp/dx
e 0.15 Ve
5Ve Qg=3.64(m3/s)
(m3/s) s) dp/dx
Qg= 1.46 (m3/s)
Qg=3.64
(m3/s)(m3/s)s) 1
Oil 0.15
il 0.1 Oil 0.15
Qg= 1.46
Qg=3.64 (m3/ 11 1
Qg=
Qg=2.45 (m3/s)
5
Qg=
2.45 1.46 (m3/s)
1.46(m3/s)
Qg=(m3/s)
0.05 0.8
Qg= 2.45 (m3/s) 0.8
.8
0 0.05
0 2 2 44 66 88 1010 1212Qg= 142.45
14 (m3/ 0.8
0.0 0.05 1 1 1.5 1.5 2 2 2.5 2.5 3 3 3.5 3.5 4 4
0 2 4 X (m)
6(m) 8 10 12 14 1 1 1.5 1.5 2 2 2.5Qg (m3/s)
2.5 3 3 3.5 3.5 4 4
0 2 4 X 6 8 10 12 14 Qg (m3/s)
Qg (m3/s)
Qg (m3/s)
X (m)
X (m)

Fithe
g. 7gas
. Thvoid
e fraction in center of tubing (a), near FigF. i7g.. T 7h.
Fig. 7. The effect of the injected gas rate on
eTehfefefefcecftffftoeofccftthtooheffeittnhihnjeeejeicincntejtjededctgtegeadadssgrgraastteerraoaotntenetothohnenetghtghaeaessgvgvaoasosiivddvofofirrdiaadccfrtftiairoacnctitoiinonncineinctecenertnoetfrertoufobftiuntbugibnigng
the(a(wall)a,),(nane(b)
)ae,ranrton
ehtahethe
rew tthwoil
aeeallvelocity
w
lw l(aba(lbl)ll)o((obnb(c)
n))tohtoand
hnenetothohifrictional
elielvovoeieillllovovceceilitlpressure
otyoycc(i(ticcyt)y)(adrop
a(cncn)d)daafnf(d).
rnrdiidccfttrfiiiorocinnctiataoillonpnarralelpsprseurserseseusdurerroedprdo(rdpdo))p
(d()d)

8.1.2. E FFE C T OF T UBIN G SIZE

Another parameter, which can significantly influence the performance of continues

gas lift, is the size of tubing. The effect of tubing size is not simple and there are not
apparent conclusions in literature. It considerably depends on flow rate and should
be determined in each well particularly. Figure 8 involves the effect of tubing size on
the pressure gradient, oil superficial velocity and gas void fraction. In this figure, the
bubble diameter was assumed 3 mm and the quantity of injected gas was 2.45 m3 /s.
Three standard tubing diameters with the size of 0.0706, 0.102 and 0.127 meters,
regarding the operational limitations, are investigated. As observed in the following
figure, by increasing the tubing size, gas void fraction, gravitational pressure, gradient
and oil velocity in the center of tubing decrease. In fact, the frictional pressure drop
increases, when the size of tubing diameter decreases. This increment of the frictional
pressure drop overcomes the slight reduction of gravitational pressure drop.

8.1.3. E FFE C T O F I N J E C T E D G A S D I S T R I BU T I O N

Unauthenticated
Download Date | 12/12/17 12:14 PM
The bubble diameter is considered as a critical parameter to discuss the effect of the
injected gas distribution on gas lift performance. All simulation was performed in a

Unauthenticated
Download Date | 12/12/17 12:14 PM
 should be determined in each well particularly. Figure 8 involves the effect of tubing
sizesiozne tohnP e souilpesrufpiceiraflicviaellovceitlyocaitnyd agnads gvaosidvofirdacti
froanct.ioIn. tIhnis this
atpghreas4ps2ruarsesugrreadgiraandt,iaonitl,

figufrieg,utrhee, tbhuebbluebbdlieamdeiatemrew tearswaassuam sseudm3edm3mmam nd atnhde tqhueanqtuitayntoitfy inojfecintejedctgeads gwaasswas

3
2.452.m 45/s.m"T 3 h2 rThree
8./s. .eE e FsFtaEnstandard
CdTardOFtuIbNin JgECdT
tubing i aE mDetG erA
diameters s SwD ith IStT hthe
with Re IsBiz UeToIO
size fofN0.0.07
"07s0h6 o,ul, 00.102
d.1b0e2reapnand ldac0e.0.127 d12b7y "8.1.3. EFFECT
OF mete rs , re g
mete sI,NrJeEg C a r d in g
arTdE t
inD hgGe o p er
thAeSoD a ti o n
pIeSraTtR a l li
ioInBaU m it
l TliIm a ti
OiNota"tions, are inves iga ted. As observed tihne the
n s , a r e i n ve s tig a te d . A s ob se r v e d in
follofowlilnogwifniggufrieg,ubrye,in bcr
y einacsrin eagsith neg tuhbeitnugbisnizge,s igzaes, gvaosidvofriadcftiroanct,iogra n ,vgirtatviiotnatailopnraelspsruerses,ure,
gradgireandtiaenPdtaao gned4o3 i velocity in the cente r of tubing decrease. In fath
i l v e lo c ity i n th e c e nt e r o f tu b i n g d ecr e a s e . I n fa ct, cte, fthriectfiroinctailopnraelspsruersesure
dropdrionpcrienacsreesa, sewsh, ew n htehne tshiezesiozfe toufbitnugbidnigamdeiatemred teercrdeeacsreesa.seTsh.isThiniscrienmcreenm t eonft tohfe the
A C o m p u a t i o
frictfiroinctailopnTraehlse pstlruergsesed F
unrdo l u i d D y n a
e podfrooFvpiegro .cv 8oebmm is
c
rciesom stM o
cheresam d l f o
sthleibeglhesd r Ghdthuricsetdfiioiugncutoiroefngshoraofvugilrtdatvbiioetnartaeiloppnlaracelisnepsd.ru.e.brseys:du6roe5pd.
lti.rgeT a s L f t P r o c e s s S i m u l a t i o n
rop.
1.4 0.22
(a)
1.4 D= 0.076 m (b)
1.2 0.22 0.22
D= 0.102
D=m0.076 m
1 1.2 (a) 0.21 (b) (b)

Gas void fraction

D=m0.102 m
Oil velocity (m/s)

D= 0.127

Gas void fraction

0.8 1 D= 0.127 0.21 0.21
on
(m/s) 0.2ti
ca
0.6 m fr
0.8 0.2
yit void
0.2
0.4 D=m
veloc D= 0.0706
0.190.19s
0.2 0.6 0.07
Ga 0.19 D= 0.102 m D=
06 m 0.07
0 Oil 0.4 D= 0.127 m
0.180.18 06 m
0 0.2 2 4 6 8 10 12 14 0 00.18 5 5 10 10 15 15
X (m) X (m)
X (m) 15
1.36 0 5 10
0
1.36 X (m)
0 2 4 6 8 10 12 14 (c)
X (m) 1.34 (c)
)
/m 1.32
dp/dx (Kpa/m)

pa
1.32
(K 1.3
1.3
dp/dx
1.28
1.28
1.26 1.26
0.07 0.07
0.08 0.08
0.09 0.090.1 0.10.11 0.11
0.12 0.12
0.13
0.13
D (m) D (m)

FigF.iF
g8.i.g
8T8h.heeTvvhaaerriivataiorinnatoo
.T ifofgnagsoavsfogvidaosifdrvafcortiiadocnft:iroa(anc)t:io(oialn)s:uo(piael)rsifoucipilaeslruvfpeicleoiracfli itcyvi,ea(llbo)vcgeirtlayov,cii(tatbyt)i,o(ngbar)lapvgriertsastvuiiortenatailonal
drop, (c) with tubing diapmreetsepsr.u rersesd uro e pd,r(ocp), w (ci)thwtiutbhitnugbdiniagmdeiatemreter

constant gas injected rate equal to 24.50 m3 /s. Figure 9 illustrates the gravitational
pressure gradient, gas void fraction and oil superficial velocity in different size of
bubble diameter. It can be seen in Fig. 9a, decreasing the bubble size causes increas-
ing gas void fraction in the center of the pipe and successively decreasing it near the
wall. So, the smaller bubbles improve gas distribution in the tubing. Also, it is shown
in Figs. 9b and 9c, when the smaller bubbles were injected, gravitational pressure
drop decreased and higher oil velocity were observed in the tubing.

9. CONCLUSION
In this study, the gas lift process in a typical well in one of the oil fields in the south of
Iran was simulated. A multi fluid model, based on CFD technique was developed to
study the behaviour of two-phase flow in the vertical pipe. A statistical analysis was
conducted and the accuracy of the multi fluid model was investigated. The results
showed that the proposed method, performed reasonably well to predict the complex
behaviour of two-phase flow. The model was used to investigate the performance of
gas lift around the injection point in an oil well under gas lift. The calculated pres-
sure drop was compared with those predicted by other empirical and semi empirical
models. The low values of errors showed the accuracy of the model to simulate gas

Unauthenticated
Download Date | 12/12/17 12:14 PM
 The bTuhbebbleubdbialemdeitaem r iestecroinssicdoenrseiddearseda acsritaiccarlitpicaaral m paertaem r teotedristocudsisscthuessetfhfecetffect
ooff tthheeoifinntjh jeeecctiie
tn n
eddjeggcataessdddg igissattrsriibdbuiustttiirooinbnuootninogngaaossnlliigfftat pspeelrirfftoop rrm erafna oncrcm ee..aA nclllees
A l .siA
A
mlululslsaaitm
i tiioounnlaw tiao
w asnsnppw eerarffsooprrmerefd eodrim inned in
aa ccoonnassttcaaonnntt sggtaansstiingnjajeescctitienenddjercraatetteed eerqaqutueaalleq ttooua22l44t..o550024m.533/0/ss..m FF3i3i/ggsu.urrFeeig99uirilellluu9ssttriralalttueesstrttahhteesggtrrhaaevviitgtaartataiiovonintaaltlional
pprreessssupurreessggurraraeddiiegenrnatt,d, igg eanasst, vvgooaiisdd vffroraaicdcttiio fornnacatainnodd n ooain illdssuu oppileerrsffiuiccpiiaealrlfivvceiealloloccviitteyyloicinnityddiififnfnfeerrdeeinnftftesrsieizzneet osoiiffze of
bbuubbbblbleeudbdibiaalm e edettieaearm .. IeIttteccraa.nnItbbceeasn
rm seeebenen isineneeFFniiggi.n. 99F--iaag,,.dd9ee-ccarr,eedaaessicinnrgegattshhienegbbutuhbbebblbleeusb siibzzleleeccsaaiuzuseseecssaiu innsccerrseeaianssicinnrgegasing
ggaass vvgooaiidsd vfvfrro oaaiccdttiifoornn
aciitnniottnhheeincceteh nnetteecrreoonfftetth rheeoofppitiphpe apaninpddessauunccdcceesssussicivvceelslyysidvdeeclcyrreedaaessicinnrgegaisitti nngeeaairtr tnthheeearw wtahallell..wall.
SSoo,, tthhSeeos,sm thaaellllseemrr bablluulbebrrbblbleeussbiibmleppsrrooim vveepgrgoaavsseddg iissattsrriibdbuiustttiirooinbnuiitnniottnhheeinttuutbhbieinngtgu..bA Als.o, it is shown in Fniiggin
in l s
g o. ,A itlsi s
o ,s ih
to iw s n
s hi o nw F .. Fig.
99--bb aa9nn-dbdb 9a- n c d
, 9w - ch , e n wt h ee n
9a-c, w,hen the s ealler bubbles were inj cit n s t
m h e
al l se mr ab lul ebrb l
b e us bb w le r
s e w i n ej
r e c tie ed, gravitaation l pressure drop drop
n dje , cgt e
r a
dv, i t
g a rta
i o
v ni taaltio p nr e
a s
l s u
p r e ss d ur r
o ep
ddeeccrreedaa6essec6edrdeaansnedddhhaiignghdheehrriogoihillevvreeolloiolccviA itteyyrlaow
scheietrK
w ryeeaw
oodbeebivsrseaerro,vv bE
eesdbd erirainvnhetithm
dheeiN nttuuetbhbmieinangtguiu..bLian
yg.
0.3 0.5
0.3
0.3 (a)
0.3 0.5 d=1mm
(a) 0.5 (b)0.5 d=1mm
0.25 (a) (a) (b) (b) d=1mm
d=2mm d=1mm
0.2 5
0.25 (b) d=2mm

Gas void fraction

d=2mm
on 0.25 0.4 on d=3mm d=2mm
Gas void fraction

on d=3mm
cc 0.2 ation ti
0.2
on
ti 0.4on ti 0.4 d=3mm d=3mm
a 0 .2frti ca 0.4
fr
c ca
fr ctifra
0.2 fr
fr
voi
ad
void 0.15 void
0.155 dvoi
s vo
s 0.1 s 0.3 0.3
vo d=1mm
d=1mm Ga 0.3 ids
Ga d=1mm Ga 0.3
Ga 0.1.1 id
s 0.15 d=2mm Ga
s
0.1 d=2mm
0 Ga d=2mm
d=1mm
d=3mm
d=3mm
d=3mm 0.2
0.055 0.1 d=2mm
0.0 00 0.0522 44 66 88 10
10 12
12 14d=3mm14
14 00 0.2 2
2 44 66 88 10
10 12
12 14
14 14
4
0 2 4 X (m)6 8 10 12 0.2 0 2 4 6 8 10 12 1
0.05 X (m) 0.2 X (m)
X (m) 0 2 4 X (m)6 X (m) 8 10 12
0.5 0 2 4 6 8 10 12 1.52
X (m) X (m)
0.5 1.52
0.5 (c)
0.5 1.5 2 1.52
0.4 (d) (d)
Oil velocity (m/s)

(c) 1.44 (d) (d)

0.4
(c) (c)
0.4 0.4 1.44
1.4 4 1.44
(m/s) )
(m/s)
dp/dx (Kpa/m)

0.3 (m/s) ) )/m

y 0.3 ityity 0.3
0.3 /m /m
pa 1.36
1.3 pa
pa 1.36
it 6 (K 1.36
0.2 veloc0.2 (K (K
veloc 0.2 d=1mm d=1mm
veloc0.2 Oil d=1mm d=1mm dp/dx
/dx dp/dx
dp 1.28 1.28
Oil d=2mm d=2m m
Oil 0.1 0.1 d=2mm d=2mm 1.21.28
8
0.1 d=3mm m
0.1 d=3m
d=3mm d=3mm
0 00 1.2.2 1.2
1.2
0 0 20 42 64 86 108 1210 1412 1
14 0 00 1 11 2 22 3 33 4 4
0 20 42 64 86 108 10
12 12
14 14 0 1 2 3 4 4
X (m) X (m) r (mm) rr (mm)
X (m) X (m) r (mm) (mm)

FFiiggFig.
..99F .F. TiTgh.9
9.hThe
ee.iinTT
nfh llueueeinncfcleeuo
finfluence eeonn
fof
fcbbeuub
o
obbffblbleeud
bubble
b
dib
iaalm
diameter
e edetitiear rmooegas
aeon
m nnteggraavoid
ossnvvg
o
gofraction
iaaidds fvfrroaaiccdttin
iifofornthe
aac
n iintncenter
iotthnheeincceof
etnhnetetubing
teecrreoonfftet(a),
turb
uboiifnnear
ngtgubing
((aa)),, nnt((eheaaea)rr,wtnthahelelea(w
rw t a h l l
e
b)a, loln(b (wb )
a ,l o
l
oai)l, soupe n
( b o
) ,ilo sn u op e
i lr f
s i
u p a
el
r f v
ei c l
ia
lo c i
v tey l (
oy c )
i t a
n)o,riflcsi aulpveerl foicciitayl(vc)ealoncdiegt ylroa(vcit)aati nocdn)agl rpar n (dc
v ) gt
g aran o
v p (d)
dveisistaaurtatariieonidintraaoltlpipop(rnrdee)as.slssupurreesddsruroorpep d(d(drdro)o)

99.. C Coo9nn.ccC lluuossniioocnlnusion

lifItInnfottrhhitiIshs nesstothuutidhdsyeysr,,tw tuthhdeeeyllg,sgataihsnseltlihigffetaat spoprilrloioficcteeplsdrsrsoo.
iciInnnesaasdttiy p i c at
idyniptaioc nayl,ptwl w l
hicealrlew i n
is nuelo o n e o
ltlsniisenhoow f t h
ntehdeotfhe o it l
o aihltefiniofi e l d s n t h
ejeliidlcstffi iioieennldtoshf ein e the ssouthsogoua fsthrIa r oioasnfnedIw
o u th o f I ra wrthaaness
ow sw
siilim
am eulssloaaictm
vsu teietduduy.ll.au
aA ttpA edtom .m0uA u.l2ltt9iimm ffullu/lustiidadtfm lomuuioiotdd ldeeetm
l.l,,obdbaeasls,eeddbaoosnendC CoFFnD DCtteFecD chhnntiieqqcuuheeniw wuaaess
q was
ddeevveelldooeppveeeddTlohttopoeem sdsttuutlodtyiysflttuhhidedeybm bteteohhhhdeaaevvbliietoohhuuearnrvoiw ooffautstrw uosofe--pdpthw
w htaoaoss-s-eepurhffvlaloeoswe tffhiilnenow
yw etthfhfeieincvvtteteohrrfetetitichcvaaeellroptpiiceppareael.t.piA
oA ipnpeaessl.t.taapA ttaiisrsatstimitccaaat-
illstical
aannaallyyasseniistsaew rysasaisiissnsccw
lw olounanddsdiunccc
ugocttneted u dbuaciacntntegdddstthiahzeneeda,acitctnhchcjueuerrcaatceccydyuorgoafafcsttyhhdeeoism trtuihubleluttiitm
fm iffolulu nulititdaidinfm
dluootidhdeem llrw aowtdaeaessolifiw
nw nivvanesesjsesitctniitggveaadettseegtdidag..saTTtohehnd
ee. The
the performance of gas lift. The results revealed the sensitivity of the continuous gas
lift to its design and operational parameters. The oil production rate was increased
by decreasing bubble size as well as increasing the injected gas rate. As oil velocity
rose by 24.5%, the gas injected rate was increased by 149.3%. However, the effect of
tubing size is more complicated. For the considered well in this study, increasing the
tubing size declined oil superficial velocity, which caused to decrease oil production.

REFERENCES
[1] TAKACS , G. Gas Lift Manual, Tulsa, Penn Well, 2005.
[2] M AYHILL , T. D. Simplified Method for Gas Lift Well Problem Identification and Di-
agnosis, SPE 5151, SPE 49th Annual Fall Meeting Houston, USA, Texas, October 6-9,
1974.
[3] G OMEZ , V. Optimization of Continuous Flow Gas Lift Systems, M. S. Thesis, USA,
Oklahoma, Tulsa, 1974,

Unauthenticated
Download Date | 12/12/17 12:14 PM
 A Computation Fluid Dynamic Model for Gas Lift Process Simulation in ... 67

[4] K ANU , E. P., J. M ACH , K. E. B ROWN . Economic Approach to Oil Production and
Gas Allocation in Continuous Gas Lift. J. Pet. Technol., 10 (1981), No. 3, 1887-1892.
[5] M AHM UDI , M., M. T. S ADEGHI. The Optimization of Continuous Gas Lift Process
using an Integrated Compositional Model. J. Pet. Sci. Eng., 108 (2013), 321327.
[6] FANG , W. Y., K. K. L O . A Generalized Well Management Scheme for Reservoir
Simulation. SPE Reserv. Eng., 11 (1996), 116-120.
[7] A LARCON , G. A., C. F. T ORRES , L. E. G OMEZ . Global Optimization of Gas Alloca-
tion to a Group of Wells in Artificial Lifts Using Nonlinear Constrained Programing. J.
Energy Resour. Technol., 124 (2002), 262-268.
[8] K ABIR , C. S., A. R. H ASAN . Performance of a Two-phase Gas/Liquid Flow Model in
Vertical Wells. J. Pet. Sci. Eng., 4 (1990), 273289.
[9] O RKISZEWSKI , J. Predicting Two-phase Pressure Drops in Vertical Pipes. J. Pet. Tech-
nol., 19 (1967), No. 6, 829838.
[10] A ZIZ , K., G. W. G OVIER , M. F OGARASI . Pressure Drop in Wells Producing Oil and
Gas. J. Can. Petrol. Technol., 11 (1972), 3847.
[11] C HIERICI , G. L., G. S CLOCCHI. Two-Phase Vertical Flow in Oil Wells - Prediction of
Pressure Drop. J. Pet. Technol., 26 (1974), No. 8, 927938.
[12] D UNS , J R . H., N. C. J. ROS . Vertical Flow of Gas and Liquid Mixtures in Wells, In:
Proc. 6th World Petroleum Congress, Germany, Frankfurt am Main, 19-26 June, 1963,
451465.
[13] T ROSHKO , A. A., Y. A. H ASSAN . A Two-equation Turbulence Model of Turbulent
Bubbly Flows. Int. J. Multiph. Flow., 27 (2001), 19652000.
[14] L AMB , H. Hydrodynamics, New York, Cambridge University Press, 1932.
[15] D REW, D. A., R. T. L AHEY J R . Application of General Constitutive Principles to the
Derivation of Multidimensional Two-phase Flow Equation. Int. J. Multiph. Flow., 5
(1979), 243264.
[16] B EHBAHANI , M. A., M. E DRISI , F. R ASHIDI, E. A MANI. Tuning a Multi-Fluid
Model for Gas Lift Simulations in Wells. Chem. Eng. Res. Des., 90 (2011), No. 4,
471486.
[17] I SHII , M., N. Z UBER . Drag Coefficient and Relative Velocity in Bubbly, Droplet or
Particulate Flows. AIChE J., 25 (1979), 843855.
[18] S CHILLER , L., A. N AUM ANN . A Drag Coefficient Correlation. V. D . I. Zeitung, 77
(1935), 318320.
[19] T OMIYAMA , A. Struggle with Computational Bubble Dynamics. Multiph. Sci. Tech-
nol., 10 (1998), No. 4, 369-405.
[20] F RANK , T H ., P. J. Z WART, E. K REPPER , H. M. P RASSER , D. L UCAS . Validation of
CFD Models for Mono- and Polydisperse AirWater Two-phase Flows in Pipes. Nucl.
Eng. Des., 238 (2008), 647659.
[21] G UET, S., G. O OMS . Fluid Mechanical Aspects of the Gas-lift Technique. Annu. Rev.
Fluid Mech., 38 (2006), 225249.

Unauthenticated
Download Date | 12/12/17 12:14 PM
68 Arash Kadivar, Ebrahim Nemati Lay

[22] H ARLOW, F. H., P. I. NAKAYAM A . Transport of Turbulence Energy Decay Rate, Los
Alamos Scientific Laboratory report LA-3854, 1968.
[23] B ELF DHILA , R., O. S IMONIN . Eulerian Predictions of a Turbulent Bubbly Flow
downstream a Sudden Pipe Expansion, In: Sommerfeld, M. (Ed.), Sixth Workshop on
Two-Phase Flow Predictions, Germany, Erlangen, 1992.
[24] I SSA , R. L. Solution of the Implicitly Discretized Fluid Flow Equations by Operator
Splitting. J. Comp. Phy., 62 (1986), 40-65.
[25] I SSA , R. L, P. J. O LIVEIRA . On The Numerical Treatment of Interphase Forces in
Two-phase Flow. Numer. Methods Multiphase Flows, 185 (1994), 131-140.
[26] B EHZADI , A., R. L. I SSA , H. RUSCHE . Modeling of Dispersed Bubble and Droplet
Flow at High Phase Fractions. Chem. Eng. Sci., 59 (2004), 759-770.
[27] I SSA , R. L, P. J. O LIVEIRA . Numerical Prediction of Phase Separation in Two-phase
Flow through T-junctions. Comput. Fluids, 23 (1993), No. 2, 347-372.
[28] WALLIS , G. B. One Dimensional Two-Phase Flow, New York, McGraw-Hill, 1969.
[29] Z UBER , N., J. A. F INDLAY. Average Volumetric Concentration in Two-phase Flow
Systems. J. Heat Transfer. Trans., ASME, 83 (1965), 453-468.
[30] C HEN , N. H. An Explicit Equation for Friction Factor in Pipe. Ind. Eng. Chem. Fun-
dam., 18 (1997), No. 3, 296-297.
[31] L AUNDER , B. E., D. B. S PALDING. The Numerical Computation of Turbulent Flows.
Comput. Method Appl. Mech. Eng., 3 (1974), 269289.
[32] D HOTRE , M. T., J. B. J OSHI . Two-dimensional CFD Model for the Prediction of
Flow Pattern, Pressure Drop and Heat Transfer Coefficient in Bubble Column Reac-
tors. Chem. Eng. Res. Des., 82 (2004), 689307.

Unauthenticated
Download Date | 12/12/17 12:14 PM