See

discussions, stats, and author profiles for this publication at: https://www.researchgate.net/publication/276078316

A new approach for modeling of gas-

condensate flow through pipelines under
industrial operating conditions

Article in Journal of Natural Gas Science and Engineering November 2014

DOI: 10.1016/j.jngse.2014.09.016

CITATIONS READS

3 28

2 authors, including:

Mohammad Reza Talaie khozani

Shiraz University
61 PUBLICATIONS 488 CITATIONS

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Removal of H2S and COD Using UV, Ferrate and UV/Ferrate from Municipal Wastewater View project

All content following this page was uploaded by Mohammad Reza Talaie khozani on 07 April 2017.

The user has requested enhancement of the downloaded file. All in-text references underlined in blue are added to the original document
and are linked to publications on ResearchGate, letting you access and read them immediately.
 Journal of Natural Gas Science and Engineering 21 (2014) 540e555

Contents lists available at ScienceDirect

Journal of Natural Gas Science and Engineering

journal homepage: www.elsevier.com/locate/jngse

A new approach for modeling of gas-condensate ow through

pipelines under industrial operating conditions
M.R. Talaie a, b, *, K. Soleimani Deilamani b
a
Department of Chemical and Environmental Engineering, Faculty of Engineering, University of Nottingham, Malaysia Campus, Malaysia
b
Department of Chemical Engineering, Faculty of Engineering, University of Isfahan, Iran

a r t i c l e i n f o a b s t r a c t

Article history: The main objective of this study is to achieve a comprehensive integrated two-phase/single-phase hy-
Received 1 February 2014 drodynamic model for gas-condensate ows through transmission pipelines under industrial operating
Received in revised form conditions, i.e. large pipe size and elevated pressure. The model developed in the present work covers
9 September 2014
single-phase gas, mist ows, and also all ow patterns occurring in transition from stratied to annular
Accepted 12 September 2014
Available online
having liquid volume fraction from 0.005 to 0.3. Flanigan's correlation was corrected in the way that it
was applicable for mist two-phase ow under industrial operating conditions. For liquid lm holdup,
Grolman and Fortuin model and Taitel and Duckler approaches were employed for uphill and downhill
Keywords:
Mathematical modeling
two-phase ows respectively. Grolman and Fortuin and BJA methods were applied to nd pressure drop
Multiphase ow through upward and downward pipelines respectively. Two new relations for predicting liquidegas and
Natural gas liquidewall friction factors, which were obtained based on tting to the eld data, were used in Grolman
Pipeline and Fortuin model. 200 eld pressure data were collected by conducting a eld experiment on an in-
Pressure drop dustrial gas-condensate pipeline. 160 of these eld data were applied to make the required corrections to
Liquid holdup the model including the development of two new correlations for both liquidegas and liquidewall
friction factors. The rest of the data points were utilized to verify the model. The comparison of the
present model results with the eld data revealed that the proposed model is capable of predicting
pressure drop accurately in gas-condensate pipelines.
2014 Elsevier B.V. All rights reserved.

1. Introduction
Two-phase ow phenomena can be observed in different parts
of petroleum industries. The transmission of extracted natural gas
from wells to reneries using pipelines is one of the important
cases in which two-phase ow can be formed. Although natural
gas is normally dried by passing through wellhead equipment, the
variations of pressure and temperature through transmission
pipelines may eventually result in the formation of liquid phase
due to retrograde condensation behavior. Gas-condensate ows
* Corresponding author. Department of Chemical and Environmental Engineer-
ing, Faculty of Engineering, University of Nottingham, Malaysia Campus, Malaysia.
E-mail address: mreza.talaei@nottingham.edu.my (M.R. Talaie).
usually fall into the category of low liquid loading in two-phase
ow science. Many attempts have been made to propose a
model for predicting pressure drop and liquid holdup in two-
phase ow pipe with low liquid loading. However, apparent dis-
crepancies emerge among the predictions of these different
models while being applied for undulating gas-condensate pipe-
lines. The rst reason is that these models are usually adjusted
based on the experimental data which is obtained under labora-
tory conditions, i.e. small pipes and low pressures, being too far
from industrial operating conditions. The second reason can be
explained by the fact that nearly all two-phase ow models were
tuned for a particular limited range of liquid holdups. Nonethe-
less, in gas-condensate pipelines the liquid holdup varies greatly
from low values (less than 0.005) to high values (0.3) depending
on the pipeline inclination. In other words, gas transmission

http://dx.doi.org/10.1016/j.jngse.2014.09.016
1875-5100/ 2014 Elsevier B.V. All rights reserved.
 M.R. Talaie, K.S. Deilamani / Journal of Natural Gas Science and Engineering 21 (2014) 540e555 541

pipelines are usually characterized by undulating topography.

Consequently, various liquid holdups are formed in different
sections of the pipeline regarding their inclinations. That is why
the available models of gas-condensate ow suffer from poor
prediction of pressure drop while applying for an undulating
pipeline. According to the aforementioned characteristic of gas-
condensate ows with low liquid loading, it is necessary to nd
a reliable comprehensive model which is able to predict pressure
drop and liquid holdup in a wide range of liquid holdup under
industrial operating conditions accurately. Having large pipe
diameter, being at high pressure and considering mass transfer
between gas and liquid phases are meant by industrial operating
conditions. Such a model should satisfy the following criteria:

1. The model should predict pressure drop for both two-phase and
single-phase gas ow cohesively.
2. The model should be so comprehensive that it covers ow
patterns in a wide range of liquid holdups including single-
phase gas, mist, stratied and annular ows up to slug ow.
3. Gasewall, liquidewall and gaseliquid friction factors must be
calculated accurately for high pressure and large diameter
pipelines.
All of the models having been proposed so far for gas-
condensate ow with low liquid loading are designed for a
limited range, and they have not considered as a comprehensive
model. Also, in the recent decades, several efforts have been made
to developed phenomenological models which have been based
on physical facts of gaseliquid ow phenomenon. Johannessen
(1972) has indicated that Lockhart and Martinelli method had a
physical base for the case of stratied two-phase ow. As a result,
Lockhart and Martinelli model could be recognized as one of the
phenomenological ones. Although several phenomenological
models have been developed within the two recent decades, only
few of them are appropriate for gas-condensate system. Taitel and
Duckler (1976, 1977) conducted pioneering work which has been
the base of many further phenomenological models. Oliemans
(1976) put forward a theory to predict pressure drop for gas-
condensate pipelines. He suggested the hypothesis that liquid is
approximately stationary around the inner side of the pipe wall.
This fact has two impacts on two-phase ow, i.e. reducing effec-
tive pipe diameter and modifying pipe roughness. These effects
both cause an increase in friction factor. Oliemans (1987) has also
obtained new relations for predicting pressure drop and effective
pipe diameter. He has not proposed any method to evaluate liquid
holdup. Based on the experimental data from pipe diameters 17
Fig. 2. The thin-layer liquid lm conguration.
and 20 inches, he has come to this conclusion that Lockhart and
Martinelli method predict liquid holdup well. It should be noticed
that Oliemans model was used only for horizontal pipes. Baker
et al. (1988) have developed a new model based on Oliemans
theory to predict both pressure drop and liquid holdup. Similar to
Oliemans model, liquid leads to a decrease in effective pipe
diameter but an increase in friction factor. Baker et al. have
employed a homogenous model to predict pressure drop, whereas
they utilized Taitel and Duckler phenomenological model to
evaluate liquid holdup. Taitel and Duckler have made the
assumption that gaseliquid interface was at, as shown in Fig. 1.
Baker et al. model which sometimes is called BJA model could also
be used for inclined pipes. Hamersma and Hart (1986) have
further developed Oliemans theory. They suggested that liquid
wetted a fraction of pipe wall perimeter as a thin liquid lm with
constant thickness, as indicated in Fig. 2. It causes a decrease in
effective cross sectional area but a signicant increase in apparent
roughness. In the end, the friction factor increases as a result of
the extra roughness created due to the wave on the surface of the
liquid lm. Hamersma and Hart have introduced the relations to
nd wetted wall fraction, q, and apparent roughness. Hart et al.
(1989) have created a new model called Apparent Rough Surface
(ARS) by further developing their own previous model. The
application of this model was restricted to horizontal pipelines.

Fig. 1. The at liquid surface conguration. Fig. 3. The curved-interface conguration.

542 M.R. Talaie, K.S. Deilamani / Journal of Natural Gas Science and Engineering 21 (2014) 540e555

Following to Hart et al. theories of two-phase ow, Chen et al.
(1997) have offered Double Circle Model in which new
gaseliquid interface shape was employed according to Fig. 3.
Comparing with experimental data, Grolman and Fortuin (1997a,
1997b) have revealed that the amount of pressure drop calculated
from ARS model underpredicted the experimental data. Based on
this fact, they modied ARS model and reached to Modied
Apparent Rough Surface (MARS) model. MARS model could pre-
dict pressure drop and liquid holdup in pipes with inclination
slopes between
3 and 6. (
3 < b < 6 ). In other words, they
have modied the previous models by taking the effect of pipe
inclination into account. The inuence of liquid droplets entrained
in gas stream on pressure drop has been taken into consideration
by Meng et al. (1999). Asante (2002) has established his model
based on the values of liquid holdup. According to this model, if
the volume fraction of liquid is less than 0.005, all liquid is
accompanied with gas as droplets, and hence a homogenous
model could be used to predict pressure drop. However, if the
volume fraction of liquid is between 0.005 and 0.05, a Double-
Circle model was suggested to be utilized. Otherwise, if the vol-
ume fraction of liquid is greater than 0.05, Taitel and Duckler's
model was applied to predict pressure drop. Also in order to nd
liquid holdup, Asante recommended empirical relations of
Butterworth (1975) which were only used for horizontal pipes.
Unfortunately, all above models are not perfect to predict pressure
drop in gas-condensate pipelines, and they suffer from the lack of
reasonable accuracy, especially for industrial operating condi-
tions. Although these models need further improvement, no other
relevant work can be found in the literature. It seems that more
urgent topics such as two-phase ow in micro channels suppress
further developing gas-condensate ow models in pipelines in the
recent decade.
The objective of this research is to achieve a model which can
be employed in accurately predicting pressure drop and liquid
holdup of gas-condensate pipelines under industrial operating
conditions in a vast range of liquid holdup. For this purpose, two
new correlations for calculating liquidewall and liquidegas fric-
tion factors were obtained by tting the results of the model to the
eld data.
and mist ow pattern is established. In the above equation, (dp/
dx)sl and (dp/dx)sg are supercial liquid and gas frictional pres-
sure gradients respectively.
2. There is no slip between gas ow and entrained-droplet
movement, so the gas and tiny droplets are considered as
a single homogeneous phase. Also, the properties of the
gas-droplet mixture are determined using no-slip liquid
holdup.
3. When LockharteMartinelli parameter
q
(c dp=dxsl =dp=dxsg ) is more than 0.01, a part of liquid
ows as a segregated lm. In this case, the model of Grolman and
Fortuin (1997a, 1997b) is used for horizontal and upward ows
and the one of Baker et al. (1988) is employed for downhill ows.
4. The thermodynamic equilibrium will be established as soon as
gas and liquid come into contact. Thus, both temperature and
pressure of liquid and gas phases are the same at each point
along the pipeline.
5. There is no possibility for mass transfer between phases unless
through liquid evaporation or gas condensation as a result of
temperature or pressure variations.
2.2. Governing equations for a single pipeline
The objectives of modeling gaseliquid ow through a pipeline
are to obtain pressure, liquid holdup, temperature, mass ow rate
and composition variations for each phase with known values of
pipe characteristic parameters (diameter, length, pipe inclinations
and roughness) and also with known inlet condition of the two-
phase ow, i.e. inlet temperature, pressure, gas and liquid ow
rates and total composition.
2.2.1. Liquid holdup and pressure drop
In order to calculate pressure drop and liquid holdup, the model
employs following three different approaches depending on the
value of LockharteMartinelli parameter:
1 LockharteMartinelli parameter is more than 0.01 and the ow is
upward:

In this case, pressure and liquid lm volume fraction excluding

2. Mathematical modeling
liquid droplets are obtained utilizing momentum conservation
equations for both lm and core phases. The steady-state mo-
Liquid can ow both as a lm segregated from the gas phase and
mentum equations are as follows:
as tiny droplets entrained in the gas phase. The gas phase and the
entrained tiny droplets are considered as a homogenous phase vp     swf s
named as core phase, and the segregated liquid phase is called lm af MAX 0;
Mfc uc
MAX 0; Mfc uf t it
vx A wf A i
phase. The shape of interface between core and lm phases can  
change from a at surface in stratied ows up to as much curved
rf af g sin b (1)
as the one in annular ows depending on the conditions. The
proportions of liquid lm and droplets depend mainly on the gas vp     swc s
velocity and liquid holdup. The lower is liquid holdup and the ac MAX 0; Mfc uf
MAX 0;
Mfc uc twc
i ti
vx A A
higher is gas velocity, the higher is amount of liquid entrainment in  
gas phase.
rc ac g sin b (2)

2.1. Assumptions where u, r and a are velocity, density and volume fraction of each
phase respectively. The subscript f and c stand for lm and core
In the proposed model the following assumptions were made: phases. g is gravity acceleration. b and A are inclination angle of
pipeline and pipeline cross sectional area. b is positive for upward
1. When LockharteMartinelli parameter and negative for downward pipes. p is the pressure at each point
q along the pipeline respectively. si, swf, swc are the interfacial surface
(c dp=dxsl =dp=dxsg ) is less than 0.01, all of the liquid area between core and liquid lm, the wall surface wetted by liquid
phase is in the form of liquid droplet entrained in the gas phase, lm and the wall surface wetted by core per unit of pipe length
 M.R. Talaie, K.S. Deilamani / Journal of Natural Gas Science and Engineering 21 (2014) 540e555 543

respectively. These three parameters (si, swf, swc) strongly depend liquid droplets entrained in gas stream. This correction was made
on the ow pattern, and they were found by geometrical principles. using the following relation:
twf, twc and ti are the shear stresses exerted from wall to the lm
phase, wall to the core phase and core phase to lm phase fg
fwc (10)
respectively. Mfc is the total mass transfer rate from core phase to E2
liquid lm. If Mfc is negative it means mass transfer occurs from the The correction factor E can be calculated as follows:
liquid lm to the core phase.
By combining the equations (1) and (2) to eliminate dp/dx, one r
0:1
can obtain the following equation: E 0:1084 (11)
1:1215r 1:3439
h     i swf
MAX 0;
Mfc uf
MAX 0; Mfc uf t ac where
A wf
(3)
swc s     ugs

twc af i ti
rc
rf af ac g sin b 0 r (12)
A A R0:32
All terms in the above equation are a function of af. Thus, the In the above equation R is the ratio of liquid droplet to gas ow
above equation can be solved for lm volume hold up, af. It should rates (in sbbL/mmscf). If gas stream contains no entrained liquid,
be noted that the summation of af and ac wil1 be equal to unity. i.e. r approaches to innitive, the correction factor E will be equal to
The shear stresses in equation (3) were predicted by the one.
following relations: In order to nd R, one need to know the amount of liquid
entrained in gas as droplets. The correlation developed by Petalas
  and Aziz (1998) for fraction of liquid entrained in gas in the form
fwf rf uf uf  of tiny liquid droplets is used to take liquid entrainment into ac-
twf
(4)
2 count. They have suggested the following relation:

fwc rc uc juc j NA
twc
(5) FE (13)
2 1 NA

where FE is the mass fraction of liquid entrained in gas, and NA can

f r uc
ui juc
ui j
ti i c (6)
2 
where fwf, fwc and f i are the lmewall, coreewall and
lmecore moody friction factors respectively, and ui is inter-
phase velocity.
For upward ows, Grolman and Fortune model was employed to
nd the shape of gaseliquid interface and the parameters of si, swf,
swc and ui which are required to solve equation (3) for lm liquid
holdup. Also, for calculating fwc, fwf and fi new correlations were
developed by tting the results of the model to the eld data. fi
were obtained using the following equation:
1:66
fi    2 (7)
15 kn
Ln Re 3:715d
c
where Rec is the real core Reynolds number, and kn is the roughness
due to interfacial waves. kn could be found as bellow:
af
kn 0:6325D (8)
q mlc ml FE
where af, q and D are the volume fraction of liquid lm, the wetted
wall fraction and pipe diameter respectively. Also, the equation for
friction factor between liquid lm and wall is given as follows:
0:731
fwf 0:0633fwfs (9)
where fwfs is the wall-lm supercial friction factor, and it was
predicted by ColebrookeWhite correlation using lm supercial
Reynolds number, Refs. Additionally, droplet-laden gas ow was
treated by a homogenous model. In order to nd gas-pipe wall
friction factor, fwc, rst gas friction factor, fg, was predicted by
American Gas Association (AGA) relation and using gas real Rey-
nolds number, Reg. Then, it was corrected to consider the effect of
be calculated using the following relation:
0:2
uls
NA 0:735NB0:074 (14)
ugs
where ugs and uls are gas and liquid supercial velocities. The
parameter NB is determined using the following correlation:
m2l u2gs rg
NB (15)
s2 rl
where ml, s, rg and rl are liquid viscosity, liquid surface tension and
gas and liquid densities. The mass ow rate of segregated liquid
lm which ows as a layer wetting a part or whole of the pipe wall
can be found as follows:
mf ml 1
FE (16)
The mass ow rate of liquid fraction entrained in gas stream is:
(17)
In the above equations ml is the total mass ow rate of liquid, mf
is the mass ow rate of the liquid layer and mlc is the mass ow rate
of liquid entrained in gas phase.
In order to nd total liquid ow rate including lm and liquid
droplets and gas ow rate the following procedure was
employed.
Total and component mass balances in steady-state mood sim-
ply express that total mass ow rate and composition, {zi}, are
constant along the pipeline. As a result of thermodynamic equi-
librium assumption, performing ash calculation using T, P and {zi}
at each point, one can determine the compositions of liquid and gas
phase, {xi} and {yi}, and the mass ratio of gas to liquid, v. Being
known total mass ow rate, the mass ow rates of both liquid and
544 M.R. Talaie, K.S. Deilamani / Journal of Natural Gas Science and Engineering 21 (2014) 540e555

gas phases can be calculated using the ratio v based on the

following relations: Mfc mc jxDx
mc jx (31)

Finally, the pressure distribution can be determined using either

ml mt 1
v (18)
equations (1) or (2).

mg mt v (19) 2 LockharteMartinelli parameter is more than 0.01 and the ow is

downward:
mc mg mlc (20)
For this case the same procedure as the one for upward ow was
ml includes both mass ow rates of liquid lm and droplet used except for the shape of gaseliquid interface and the param-
entrained in gas phase. eters of si, swf, swc, fwc, fwf, fi and ui which were calculated based on
The supercial velocities are given in the following: the correlations introduced by Taitel and Duckler (1976). In this
model the gaseliquid interface shape was assumed to be at. For
 the pressure drop the homogeneous model of Baker et al. (1988)
mg Mw
g
ugs (21) which is called BJA was used.
A
. 3 LockharteMartinelli parameter is less than 0.01:
mf Mw
l
ufs (22) For this case, it was assumed that the liquid phase is distributed
A
throughout the gas phase as tiny droplets, and there is not any
  liquid lm (af 0). The droplets moves with the same speed as the
 gas ow, and the slip between liquid and gas phases will be equal to
mg Mw mlc=Mw
g l
ucs (23) zero (ug uc). Therefore, the total liquid holdup matches the no-slip
A liquid hold up which can be calculated using equation (28).
Phase velocities can be found using the following equations: Although the total mass ow rate is constant though the pipe based
on mass conservation law, the liquid and gas ow rates can vary as a
ucs result of evaporation/condensation. Liquid and gas ow rates were
uc (24)
ac determined using equations (18) and (19) and applying ash
calculation to nd gas to liquid mass ratio, v. The pressure drop can
ufc be calculated using the following relation expressing homogeneous
uf (25) momentum conservation:
af

By assuming no slip between gas and droplet ows, the gas vp    

velocity, ulc, and gas velocity, ug, both will be equal to uc.
Being known af, The holdup of core phase can be evaluated by
the following relation:
af ac 1
Because the core phase consists of gas and droplet phases the
following relation will be held:
pD twc
rc g sin b (32)
vx
Where D is pipe diameter, twc is the shear stress between core
phase and pipe wall calculated using equation (5) and rc is core
phase density determined by equation (29). The friction factor
(26)
between core phase and pipe wall was evaluated using equations
(10)e(12). The velocity of core phase was determined using equa-
tions (23) and (24) with knowing that ac lc.

ag alc 1 (27) 2.2.2. Temperature variation

Temperature variation was predicted by applying the following
Where alc and ag are droplet and gas holdups based on free liquid equation expressing the steady-state energy balance:
lm. By assuming no slip between gas and droplet ows, alc will be      
equal to no-slip llc. No-slip droplet holdup relation can be written: d 1 d   1
arL uL HL u2L 1
a rG uG HG u2G
dx 2 dx 2 (33)
mlc=Mw
llc  l  (28) arL uL 1
arG uG g sin b qt
mlc m
Mwl Mwg g
where HG and HL are the specic enthalpy of gas and liquid. The
enthalpy and density of gas and liquid phases are determined by
The core phase comprises both gas phase and entrained liquid using the PRSV equation of state Stryjek and Vera (1986). qt is total
droplets. Because it was assumed that there is no slip between heat transferred from two-phase ow to environment. It can be
gas and liquid droplets, the physical properties of the core phase calculated using following relation:
are determined using no-slip holdup of liquid droplets, as
follows: qt Uave pDT
Ts (34)
 
rc llc rl 1
lc rg (29) where Uave, Ts, T and D are average heat transfer coefcient, ambient
temperature, bulk temperature of phases and pipe diameter
  respectively.
mc llc ml 1
lc mg (30)

The average of total mass transfer rate from core phase to liquid 2.2.3. Physical properties
lm, Mfc, in a denite length of pipe, Dx, can be found using the The methods developed by Chung for gas viscosity, Grun-
following relation: bergeNissan for liquid viscosity and MacLeodeSugden for surface
 M.R. Talaie, K.S. Deilamani / Journal of Natural Gas Science and Engineering 21 (2014) 540e555 545

Fig. 4. The conguration of gas-condensate transmission network comprising two different parts.

tension evaluations were employed in the present work (Poling

et al., 2001).

3. Field data

Field data were obtained on gas-condensate ow network

operating in one of the Iranian central oil elds. The oil and gas
produced from about 100 wells are sent to the seven primary
separation units named KA, KB, KC, and KD, 1, 2 and 3. After elim-
inating water and condensate, gas is transmitted to renery
through a simple undulating network pipelines. The gas is
condensed due to retrograde condensation behavior as a result of

Fig. 5. The variables involved in the rst part of the network.

Table 1
The characteristics of gas streams entering gas network A.

Component Entering line to r KA input KB input KC input KD input

enery from
upstream pipelines

Mole percent
N2 5.77 5.40 5.65 5.54 5.77
CO2 1.67 1.64 1.72 1.63 1.69
C1 85.33 82.44 84.68 85.87 85.41
C2 3.96 3.88 3.99 3.98 3.86
C3 1.19 1.28 1.23 1.20 1.13
iC4 0.29 0.43 0.32 0.3 0.27
nC4 0.41 0.77 0.46 0.42 0.39
iC5 0.20 0.54 0.26 0.20 0.21
nC5 0.16 0.55 0.20 0.16 0.18
C6 0.30 0.88 0.39 0.23 0.34
C7 0.69 2.02 1.07 0.45 0.45
Other properties
SG (C7) 0.7519 0.7511 0.7542 0.7547 0.7480
Mw (C7) 107.82 104.93 110.96 108.89 113.63
Temperature,  C 44.7 49 52 51 51
Pressure, Bar 83 89 95 98 101
Density, g/cc e e e 0.0838 0.0874
Fig. 6. The variables involved in the second part of the network.
546 M.R. Talaie, K.S. Deilamani / Journal of Natural Gas Science and Engineering 21 (2014) 540e555

Table 2
The characteristics of gas streams entering gas network B.

Component Liquid removed from slug catcher 2 Liquid removed from slug catcher 1 Input of unit 1 Input of unit 2 Input of unit 3

Mole percent
N2 0.62 0.62 4.09 4.40 4.35
CO2 0.93 0.92 1.42 1.37 1.36
C1 24.10 25.17 88.51 87.88 88.06
C2 4.11 4.08 3.41 3.60 3.60
C3 4.19 3.14 0.99 1.08 1.06
iC4 1.89 1.41 0.25 0.26 0.25
nC4 3.89 2.75 0.35 0.37 0.35
iC5 3.80 2.66 0.18 0.18 0.17
nC5 3.62 2.39 0.14 0.18 0.13
C6 8.67 6.45 0.21 0.17 0.16
C7 44.13 50.35 0.45 0.54 0.51
Other properties
SG (C7) 0.7619 0.7673 0.7470 0.7474 0.7488
Mw (C7) 114.7 121.31 107.08 109.23 111.91
Temperature,  C 44 45 46 46 45
Pressure, Bar 83 89 84.5 85 84.5
Density, g/cc e e e 0.6604 0.6492

Table 3
Gas and liquid densities e eld data compared with the results from PRSV and SRK.
connection points of different pipelines. The following measure-
Density, g/cc ments were performed during the eld experiment on the gas
Field data Calculated Calculated network:
using PRSV using SRK

Gas from unit KC 0.0838 0.0820 0.0787

1. The pressures, temperatures, ow rates of gas stream entering
Gas from unit KB 0.0874 0.08508 0.0812 pipe network and the stream to the renery were measured in
Liquid removed 0.6604 0.6643 0.5903 different time of a year. 200 data points were attained during
from slug catcher 1 this experiment. The gas compositions were obtained by using
Liquid removed 0.6492 0.6548 0.5818
gas chromatography method.
from slug catcher 2
2. Liquid and gas densities of four samples were measured in
operating conditions.
3. Phase envelope of a gas sample was attained experimentally.

pressure and temperature drop through transmission pipeline. The
network conguration, the positions of separation units and pipe-
line sizes are shown in Fig. 4. The pipeline network comprises
several pipelines which are connected to each other at points called
nodes. Figs. 5 and 6 represent the variables and the nodes existed in
two parts of this network. m, T and a are mass ow rate, temper-
ature and liquid holdup respectively. The subscripts s, o and i stand
for source point, pipeline input and pipeline output respectively.
The numbers 1, 2, 3 and 4 denote the node points which are the
The compositions of seven inlet gas streams are given in Tables 1
and 2. The liquid and gas densities measured for four samples are
shown in Table 3. Also, the four points of phase envelop regarding
to a gas sample taken from the inlet of 48 inches pipeline are shown
in Fig. 7.
4. Results and discussion
As mentioned previously, one of the primary goals of the pre-
sent work is to develop accurate correlations for calculating

Fig. 7. Comparison between eld data and phase equilibrium calculation results for
the condensate sample taken from 48-inch-pipeline slug catcher. Fig. 8. Comparison between the results of proposed model and eld pressure data.
 M.R. Talaie, K.S. Deilamani / Journal of Natural Gas Science and Engineering 21 (2014) 540e555 547

Fig. 9. Comparison between the results of Lockhart and Martinelli model and eld Fig. 12. Comparison between the results of BJA model and eld pressure dat.
pressure data.

Fig. 10. Comparison between the results of Beggs and Brill model and eld pressure Fig. 13. The effect of decreasing gas ow rate on liquid holdup and pressure drop for
data. the presented model.

Fig. 14. The effect of decreasing gas ow rate on liquid holdup and pressure drop for
Fig. 11. Comparison between the results of AGA model and eld pressure data. BJA model.
548 M.R. Talaie, K.S. Deilamani / Journal of Natural Gas Science and Engineering 21 (2014) 540e555
liquidewall and liquidegas friction factors for gas-condensate ow
through an undulating pipeline under industrial operating condi-
tions (high pressure and large pipe size). These two new correla-
tions were obtained through adjusting the comprehensive model to
simulate the eld data accurately. Talaie and Nasiri (2004) have
shown that thermodynamic model has an important inuence on
gas-condensate pressure drop prediction. As a result, efforts were
made to improve the thermodynamic model applied in the
comprehensive model. Therefore, these thermodynamic calcula-
tions were veried rst.
Table 3 shows the comparison between the eld liquid and gas
densities and the calculated ones using both PRSV and SRK equa-
tions of state. It is obvious that applying PRSV revealed much more
satisfactory results than using SRK. Thus PRSV equation of state was
used in the present model. Fig. 7 compares the phase envelop ob-
tained through using PRSV with the eld data. Apparently, some
discrepancies can be observed between eld data and calculated
results. The binary coefcient between the components C1 and C7
was determined to t the eld data. The best-t value for that co-
efcient was equal to 0.055.
As it was mentioned, 160 of 200 data points were applied to
develop equations (7)e(12). Then, the results of the model were
compared with the rest of the eld data points. Fig. 8 indicates that
the maximum errors are quite less than two percent.
Moreover, the predictions of pressure drop of previous
prominent models which are used widely in pipeline simulation
software such as PIPESIM, PIPEPHASE and PIPESYS are compared
with the proposed model results. These models include Lockhart
and Martinelli, Beggs and Brill, AGA and BJA. The prediction re-
sults of the mentioned models are illustrated in Figs. 9e12. As it
is clear in the gures, only BJA model predicts pressure values in
an acceptable range (10%), while other models are poor in
evaluating pressure specically at higher values. Among these
models Beggs and Brill predicts pressure with much more than
10% error.
The liquid holdup and inlet pressure in an undulating pipeline
connecting KD unit to KC unit were calculated for different gas
ow rate. The input pressure of the pipeline is considered equal to
101 bar. These results are presented in Fig. 13. This gure reveals
that decreasing gas ow rate leads to a reduction of pressure drop
and liquid holdup to some extent. However, further decreasing of
gas ow rate causes a rise in both liquid holdup and pressure drop.
It can be attributed to the fact that in uphill segments of this
pipeline, liquid is accumulated due to reduction of gas velocity.
This results in augmenting liquid holdup and hence pressure drop.
This experience was observed during warm season when the gas
consumption is at its lowest level. The present model is capable to
demonstrate this fact. Fig. 14 shows the variations of liquid holdup
and inlet pressure of the same pipeline versus gas ow rate
applying BJA model for all segments of the pipeline (uphill and
downhill). As it is apparently displayed, BJA model is not able to
predict the rise of liquid holdup for low values of gas ow rate.
This attribute is one of the most prominent features of the new
model.
5. Conclusion
A comprehensive integrated single/two phase model to calcu-
late pressure drop and liquid holdup in a gas-condensate pipeline
was developed under industrial conditions. The two correlations
were introduced to calculate gaseliquid and liquidewall friction
factors. These correlations can be used in a comprehensive model
under industrial conditions, i.e. large pipe sizes and high pressure.
The present model predicts a considerable amount of pressure
drop and liquid holdup in uphill pipelines compared to the hori-
zontal ones, especially where the slug pattern begins to form. The
results are in good agreement with the eld data. Thus, the model
can be used as a powerful tool to simulate gas-condensate pipe-
lines and networks. The relations derived in this study were ob-
tained based on pressure data. Using the experimental data of the
liquid holdup, ow pattern and gaseliquid interface shape for
high-pressure gas-condensate ows through large pipelines can
improve the model.
Acknowledgment
The authors wish to acknowledge the nancial support provided
by Iranian Central Oil Field Company for this research.
Appendix A. Basic part of computer code:
The Computer codes include different modules each specied
for certain reason. These modules are module data, module ash,
module input, module main physic and module main calcula-
tions. Visual Basic (VBA) was chosen as programming language/
software. Here, some part of the computer codes are presented
briey so that readers might apply it more generally to other
cases.
M.R. Talaie, K.S. Deilamani / Journal of Natural Gas Science and Engineering 21 (2014) 540e555 549
550 M.R. Talaie, K.S. Deilamani / Journal of Natural Gas Science and Engineering 21 (2014) 540e555
M.R. Talaie, K.S. Deilamani / Journal of Natural Gas Science and Engineering 21 (2014) 540e555 551
552 M.R. Talaie, K.S. Deilamani / Journal of Natural Gas Science and Engineering 21 (2014) 540e555
M.R. Talaie, K.S. Deilamani / Journal of Natural Gas Science and Engineering 21 (2014) 540e555 553
554 M.R. Talaie, K.S. Deilamani / Journal of Natural Gas Science and Engineering 21 (2014) 540e555
 M.R. Talaie, K.S. Deilamani / Journal of Natural Gas Science and Engineering 21 (2014) 540e555 555

References Meng, W., Chen, X.T., Kouba, G., Sarica, C., Brill, J.P., 1999. Experimental study of
low liquid loading gas-liquid ow in near horizontal pipes. SPE 56466. In:
Presented at the 1999 SPE Annual Technical Conference and Exhibition,
Asante, B., 2002. Two-phase ow: accounting for the presence of liquids in gas
Houston, Texas.
pipeline simulation. In: Pipeline Simulation Interest Group Conference 34th
Oliemans, R.V.A., 1976. Two-phase ow in gas transmission pipeline. In: ASME
Annual Meeting, Portland, Oregon.
Meeting, Mexico City, pp. 76e125.
Baker, A.K., Nielsen, K., Gabb, A., 1988. Field data test new holdup, pressure loss
Oliemans, R.V.A., 1987. Modeling of gas-condensate ow in horizontal and inclined
calculations for gas-condensate pipelines. Oil Gas J. 86 (12), 78e86.
pipes. In: ASME Pipeline Engineering Symposium-ETCE, Dallas.
Butterworth, D., 1975. A comparison of some void-fraction relationships for co-
Petalas, N., Aziz, Kh, 1998. A mechanistic model for multiphase ow in pipes. In:
current gas-liquid ow. Int. J. Multiph. Flow 1, 845e850.
49th Annual Technical Meeting of the Petroleum Society of the Canadian
Chen, X.T., Cai, X.D., Brill, J.P., 1997. Gas-liquid stratied-wavy ow in horizontal
Institute of Mining, Metallurgy and Petroleum, Calgary, Alberta, June 8e10.
pipeline. J. Energy Resour. Technol. 119, 209e216.
Poling, B.E., Parausnitz, J.M., Connell, J.P.O., 2001. The Properties of Gases and Liq-
Grolman, E., Fortuin, J.M.H., 1997a. Gas-liquid ow in slightly inclined pipes. Chem.
uids, fth ed. McGraw Hill.
Eng. Sci. 52, 4461e4471.
Stryjek, R., Vera, J.H., 1986. An improved Peng-Robinson equation of state for pure
Grolman, E., Fortuin, J.M.H., 1997b. Liquid holdup, pressure gradient and ow pat-
components and mixtures. Can. J. Chem. Eng. 64, 323e333.
terns in inclined gas-liquid pipe ow. Exp. Therm. Fluid Sci. 15, 174e182.
Taitel, Y., Duckler, A.E., 1976. A model for predicting ow transition regime transi-
Hamersma, P.J., Hart, J., 1986. A pressure drop correlation for gas/liquid pipe ow
tions in horizontal and near horizontal gas liquid ow. AICHE J. 22 (1), 47e55.
with small liquid holdup. Chem. Eng. Sci. 42 (5), 1187e1196.
Taitel, Y., 1977. Flow pattern transitions in rough pipes. Int. J. Multiph. Flow 3,
Hart, N.P., Hamersma, P.J., Fortuin, J.M.H., 1989. Correlations predicting frictional
597e601.
pressure drop and liquid holdup during horizontal gas-liquid pipe ow with
Talaie, M.R., Nasiri, M., 2004. The Effects of C7 Characterization on Pressure Drop
small liquid holdup. Int. J. Multiph. Flow 15 (6), 947e964.
of Wet Gas Pipelines. 8th Iranian Congress of Chemical Engineering. Tarbiat
Johannessen, T., 1972. A theoretical solution of the Lockhart and Martinelli ow
Modarres University, Tehran, Iran.
model for calculating two-phase ow pressure drop and hold-up. Int. J. Heat
Mass Transf. 15 (8), 1443e1449.

View publication stats