Scientia Iranica C (2012) 19 (6), 1653–1660

Sharif University of Technology

Scientia Iranica
Transactions C: Chemistry and Chemical Engineering
www.sciencedirect.com

A simple mathematical model for slug liquid holdup in

horizontal pipes
R. Nobakht Hassanlouei, H. Firouzfar, N. Kasiri ∗ , M.H. Khanof
Laboratory of Computer Aided Process Engineering, CAPE, School of Chemical Engineering, Iran University of Science & Technology, IUST, Tehran,
P.O. Box 16846-13114, Iran

Received 6 January 2012; revised 2 March 2012; accepted 1 May 2012

KEYWORDS Abstract This paper presents a methodology for calculation of a slug, two-phase flow hold-up in a
Two phase flow; horizontal pipe. The advantage of this method is that the slug unit hold up can be calculated directly
Slug flow; from the solutions of flow field equations with no need to use correlations. An experimental apparatus to
Liquid holdup. measure air–water hold up was setup. The flow pattern and liquid holdup in horizontal and inclined pipes,
from angles 5° to 40°, for air–water two-phase flow, are experimentally observed. The test section, with
an inside diameter of 30 mm and 3 m in length, was made of plexy-glass to permit visual observations
of the flow patterns. The proposed model was tested extensively against experimentally collected data.
Furthermore, other data sources for slug flow in horizontal pipes, for air–water and air–oil systems, were
also used for comparison. The presented methodology was compared against four recently developed
models of a two phase, slug flow holdup in horizontal pipes. Not only does the presented model
demonstrate good agreement, with less than 6.8% error, compared to the experimental data, but also has
less error compared to other models. These results substantiate the general validity of the model.
© 2012 Sharif University of Technology. Production and hosting by Elsevier B.V.
Open access under CC BY-NC-ND license.

1. Introduction a unit cell. Based on this approach, Taitel and Barnea [2] con-
cluded that across the boundary between the liquid slug and
Slug flow can be observed in many two-phase flow engineer- the long gas bubble region, flow appears steady, with mass and
momentum conserved. The unit cell concept bred a number of
ing applications, such as flow in oil and gas pipelines and other
models for calculating slug hydrodynamic parameters [3].
process industries. The formation of the slug regime is transient
The first comprehensive model was developed by Dukler
in nature, passing from stratified to wavy flow and then onto
and Hubbard [4]. Further papers employing the unit cell concept
slug flow. The existence of an interface between the two phases
improved the steady state slug flow model: Nicholson et al. [5];
that will change easily is a challenge for modeling [1]. Initial at-
Kokal and Stanislav [6] and, later, De Henau and Raithby [7] de-
tempts to model slug flow neglect its non-deterministic nature veloped a steady-state two-fluid model for the pressure gradi-
by considering the alternating liquid pistons and gas bubbles in ent and liquid holdup.
an orderly periodic way. They tested their model against a limited number of exper-
The flow is reduced to periodic cells moving downstream imental data. Good agreement was reported for air–water and
composed of a liquid piston trailed by a gas bubble, also named air–oil slug flow in horizontal pipes.
One of the most significant studies in this field was carried
∗ out by Taitel and Barnea published in their second paper [8].
Corresponding author.
E-mail addresses: ferzal.reza@gmail.com (R. Nobakht Hassanlouei),
They employed a consistent method for calculating the pressure
hastifirooz@yahoo.com.au (H. Firouzfar), capepub@cape.iust.ac.ir (N. Kasiri), gradient in steady state slug flow in inclined pipes. For
khanof@iust.ac.ir (M.H. Khanof). calculation of slug holdup, they considered variations of film
Peer review under responsibility of Sharif University of Technology. thickness and, based on this assumption, momentum balance
was developed. Then, the length of liquid film holdup, and the
velocity at the end of the liquid film were the model outputs.
Taitel and Barnea [8] used the holdup to calculate pressure,
while assuming that the film thickness was uniform, and
1026-3098 © 2012 Sharif University of Technology. Production and hosting by Elsevier B.V. Open access under CC BY-NC-ND license.
doi:10.1016/j.scient.2012.05.010
1654 R. Nobakht Hassanlouei et al. / Scientia Iranica, Transactions C: Chemistry and Chemical Engineering 19 (2012) 1653–1660

simplified the momentum balance. The reformulated version

of the simple sub model of Taitel and Barnea [8] for calculating
pressure drop was presented by Orell [9]. Orell [9] reformulated
the Reynolds number for calculating the friction factor, and
improved the accuracy of the Taitel and Barnea [8] model for
slug pressure drop in horizontal pipes.
However, Orell [9] had to use empirical correlation for
slug holdup calculation, and the correlation of Andreussi and
Bendiksen [10], as reformulated by Andreussi et al. [11], was
used to calculate the slug holdup for both air–water and
air–viscous oil systems.
In the present work, the Orell [9] model has been developed
for calculation of average liquid holdup in a slug unit. In this
study, unlike Orell [9], the liquid film length in the present
Figure 1: Schematic diagram of the test facility.
model is evaluated when deriving the slug liquid holdup
through solving the equations. So, the present model enables
direct calculation of the slug holdup from the solution of flow the size of the holes diminishes through the internal tube. After
field equations. It is independent of any specific experimental mixing gas and liquid flow, the mixed flow passes through the
correlation and can be used for different fluids (high viscose or test section.
low viscose with high or low velocity). The air inlet flow rate was measured by a digital flow
On the other hand, compared to Taitel and Barnea [8], the meter within the range of 0–12 000 l/h. Water flow rate was
liquid film thickness in the present model is assumed uniform measured by two rotameters, within ranges of 0–600 and
for slug liquid holdup calculation, so the momentum equation 1000–6000 l/h. During the experiments, the superficial gas
is simplified. Fanning friction factors of the liquid film and velocity was maintained at a set value.
gas bubble are calculated by the correlation used by Orell [9] Corresponding to literature, no previous study provides
instead of the correlation employed by Taitel and Barnea [8]. conclusive experimental evidence or estimates of the minimum
Also, liquid holdup in the film zone is defined by one required length for the flow development section [20]. Also,
half of the angle that subtends the liquid–gas interface. The some of the reported experimental observations are for
present work has a much simpler structure and equations are temporary or developing flow patterns [21]. Based on these
solved analytically for horizontal pipes. Nowadays, attempts lengths, the setup is set at 3 m. This provides around 2 m for the
to estimate liquid holdup by slug tracking can be seen [12]. flow to develop. Observations show that flow almost reaches
Generally, numerous correlations and models have been development at this stage and hold up measurements could
developed to predict the hold up of slug flow, some of the proceed.
newest being: Zhang et al. [13]; Gomez et al. [14]; Issa and The liquid holdup was measured once the system had
Kempf [15]; Abdul-Majeed [16]; Soleimani and Hanratty [17]. reached steady state conditions, having changed the superficial
liquid velocity. The liquid holdup was measured by isolating the
test section with two rapidly closing valves. Having isolated the
2. Experiments
test section, the first valve is opened, measuring the amount
of water isolated between the two valves; the holdup being
The experimental facility is shown schematically in Figure 1.
calculated through the division of this amount of water by the
The main components of the system consist of a test section,
total volume.
air supply, water supply and an instrumentation system. The
test section has a fixed and a movable part. Each part, with Four differential pressure transducers of range 0–100 psig
1.5 m length and 30 mm diameter, is made of plexy-glass were installed to measure the two phase pressure drop across
to permit visual observations of the flow pattern. A digital the test section. The system was allowed to approach steady
camera is used to record the two-phase flow regimes under state conditions before the air and water flow rates, flow
different conditions. The experiments are run for different inlet pattern and holdup were recorded. Each data was repeated
gas flow rates (0–200 l/min), liquid flow rates (150–5500 l/h) three times to ensure repeatability and the average was
and different slopes of the tube, ranging from 0° (horizontal) to reported. A total number of 720 experiments were carried out to
40° (inclined upward). measure the two-phase flow pressure drops, liquid holdup and
The water was at room temperature, and the atmospheric flow regimes in different gas/liquid flow rates and tube slopes.
pressure was pumped from the storage tank through two ro-
tameters, and then to the air–water mixer and the test section. 2.1. Experimental result
Surplus water was sent back to the storage tank. Air was
supplied to the test section by a compressor. In a mixer, both The liquid holdup was measured by two quick closing valves
the air and water streams were brought together. This part in horizontal and inclined pipes. The experimental results were
is designed according to previous work [18,19]. The mixer shown in Figures 2–6 for inclination angles: 0°, 10°, 20°, 30°, 40°
consists of a PVC pipe (6 mm inside diameter) inserted into (the experimental results for 5°, 15°, 25°, 35° are not displayed
the liquid stream by means of a compression fitting. The end of as they are very similar to these). In order to ensure result
the tube was soldered. Four holes (three rows of 1.5 mm, four repeatability, each measurement was repeated three times; the
rows of 3 mm and eight rows of 4 mm) were positioned at 90° average being reported here. In doing so, the resulting outliers
intervals around the tube perimeter. have been eliminated. Figures 2–6 show that in horizontal and
The hole centers are distanced at 1.3 cm. Passing air through inclined pipes, liquid holdup tends to go in the same direction
the PVC pipe mixes with the water that exists in the plexy-glass as gas and liquid flow rates. In other words, naturally, by
pipe. In order to prevent water from entering into the air flow, increasing the liquid flow rate (at constant gas flow rate), the
 R. Nobakht Hassanlouei et al. / Scientia Iranica, Transactions C: Chemistry and Chemical Engineering 19 (2012) 1653–1660 1655

Figure 2: Variation of liquid holdup by gas flow rate in a horizontal pipe with Figure 5: Variation of liquid holdup by gas flow rate in a pipe with Θ = 30°
D = 0.03 m. and D = 0.03 m.

Figure 3: Variation of liquid holdup by gas flow rate in a pipe with Θ = 10° Figure 6: Variation of liquid holdup by gas flow rate in a pipe with Θ = 40°
and D = 0.03 m. and D = 0.03 m.

Figure 4: Variation of liquid holdup by gas flow rate in a pipe with Θ = 20° Figure 7: Variation of liquid holdup with inclination angle at GF = 20 l/m and
and D = 0.03 m. D = 0.03 m.

liquid slippage increases, and as the gas inertia forces are not increases, because the liquid is naturally pulled to the bottom
enough to overcome it, the gas phase cannot carry over the by gravity. Figure 7 shows this change at 20 l/min gas flow rate.
liquid, resulting in an increase in liquid holdup. The value of In other gas flow rates, the changes of liquid holdup by the
liquid holdup seems to increase when the inclination angle inclination angle are the same.
1656 R. Nobakht Hassanlouei et al. / Scientia Iranica, Transactions C: Chemistry and Chemical Engineering 19 (2012) 1653–1660

Figure 8: The physical model for slug flow.

The experimental results show that the inclination angle has

a significant effect on liquid holdup. In horizontal and inclined
pipes, liquid holdup goes in the same direction as gas and liquid
flow rates. The value of liquid holdup seems to increase when
the inclination angle increases, because the liquid was naturally
pulled to the bottom by gravity when the inclination angle
increased.
Figure 9: Film zone flow geometry.
3. Mathematical model
present data bank relies on this condition), the slug length is
3.1. The model governing equations between 10 and 20 D, the average of which (15 D) is considered
the fixed slug length in the present model. The value of slug
The first step in evaluating the average liquid holdup of slug length for other different conditions can be estimated from
flow is to predict the major variables affecting it. So, a sketch of previous studies, such as Xin et al. [23].
an idealized slug has been presented in Figure 8. In this figure, Liquid mass balance, relative to a coordinate system that
l is the length of the slug unit that consists of two sections: a travels at the translational velocity of the slug unit, yields [3]:
liquid slug zone of length lS and a film zone of length lf [3]. (Ut − Uf )Hf = (Ut − US )HS . (8)
Assume a liquid film of a uniform thickness overlaid by an
Unlike the Taitel and Barnea [8] model, the film region thickness
elongated gas bubble. The liquid and gas mass balances over the
is assumed uniform and end effects have been neglected, so
slug unit yield:
the film zone can be considered as stratified flow. Figure 9
lS lf displays the film zone flow geometry. A momentum balance on
USL = US HS + Uf Hf (1) the liquid and gas yields:
l l
lS lf dP
USG = US (1 − HS ) + UG (1 − Hf ) , (2) −Af = τf Sf − τi Si , (9)
l l dx
where Hs and Hf are the liquid holdups in the slug and film dP
−AG = τG SG + τi Si . (10)
zones, and USL and USG are the superficial liquid and gas velocity, dx
respectively. Following Taitel and Barnea [8], the aerated liquid In the above equations, Af & AG refer to the film and gas cross-
slug velocity, US , was evaluated by: section areas, respectively, with τf , τG & τi referring to the film,
gas and interfacial zone shear stresses, respectively, with P
US = Um = USL + USG , (3) standing for pressure.
where Um is the superficial velocity of the gas–liquid mixture. The pressure gradient at the pipe cross-section within the
The value of l was evaluated by the Dukler and Hubbard [4] cor- film zone is uniform. Simplification of Eqs. (9) and (10) results
relation. They presented this correlation based on experimental in:
pressure drop data for a 0.0381 m diameter smooth pipe. (τG SG + τi Si ) (τf Sf −τi Si )
= . (11)
Ut Af AG
l= . (4) The shear stresses, τf , τG and τi , are given by:
vS
Gregory and Scott [22] used slug data in 0.75 and 1.5 inch 1
τf = ff ρL Uf2 , (12)
pipes and correlated the slug frequency, νS , by the slug Froude 2
number, Frs , where, in the derived equation, g is the gravity 1
acceleration constant and D is the internal pipe diameter: τG = fG ρG UG2 , (13)
2
υS = 0.0226 Fr1s .2 , (5) 1
τi = fi ρG (UG − Uf )2 (14)
2
  
USL
Frs = /(19.75/Um + Um ) , (6) where fi , fG and ff are the Fanning friction factors of the liquid
gD
film, gas bubble and gas–liquid interface, respectively, ρL and
where Um , the mixture velocity, is in ft/s and the slug frequency ρG being liquid and gas density. For slug flow in a smooth pipe,
is in s−1 . Following Taitel and Barnea [8], Ut in Eq. (4), which ff and fG are expressed in terms of the Reynolds number of each
is the translational velocity, is calculated by the following phase and are given by Orell [9]:
equation for slug flow in horizontal pipes: Cf
ff = , (15)
Ut = 1.2 Um + 0.54 gD. (ρL Uf Dhf /µL )n

(7)
Xin et al. [23] concluded that when the gas velocity is between CG
fG = . (16)
3 and 6 m/s and the liquid velocity is lower than 2 m/s (the (ρG UG DhG /µG )m
 R. Nobakht Hassanlouei et al. / Scientia Iranica, Transactions C: Chemistry and Chemical Engineering 19 (2012) 1653–1660 1657

The hydraulic diameters (Dhf , DhG ) of the liquid film and gas other air–water and air–oil data sources in horizontal pipes. The
bubbles are [9]: specification of these data sources are recorded in Table 1.
4Af The developed model was compared with Gomez et al. [14],
Dhf = , (17) Leung [25], Abdul-Majeed [16] and Zhang et al. [13] models.
Sf Parameter ranges of these empirical correlations are shown in
4AG Table 2.
DhG = . (18) The Leung [25] model was evaluated based on a wide exper-
(SG + Si )
imental database by Ghajar and Woldesemayat [26]; they sug-
For laminar flow of both phases: gested this correlation is an accurate empirical correlation for
Cf = CG = 16 n = 1. (19) holdup calculations in horizontal and inclined pipes.
For turbulent flow of both phases: The parameter range for which this assessment was carried
out is shown in Table 2. The Zhang model [13] is a mechanistic
Cf = CG = 0.046 m = 0.2. (20)
model for inclined and horizontal pipes, which is valid for gas
The interfacial friction factor suggested by Hanratty and Cohen velocities over 0.1 m/s.
[24] correlation: The accuracy of the models was evaluated by calculating the
fi = 0.0142. (21) percentage error of individual data points (IPE), the Average
The above value can be used for slug flow in inclined pipes too. Percentage Error (APE) and Absolute Average Percentage Error
The film zone flow geometry is shown in Figure 9 and the liquid (AAPE) of each data source, which is defined below [9]. The
holdup in the film zone is defined by one half of the angle that regression coefficient (R2 ) of all data sources is also reported.
subtends the liquid–gas interface, Θ [9]: n
 n

2
IPEi |IPEi |
D i=1 i=1
Af = (θ − 0.5 sin 2θ ), (22) APE = , AAPE = , (28)
4 n n
D2 Predicted − Measured
AG = (π − θ + 0.5 sin 2θ ). (23) IPE = × 100. (29)
4 Measured
The wet perimeter of film, gas and interfacial zones (Sf , SG , Si ) The result of this comparison has been shown in Table 3. This
are expressed by: table shows that the theoretical model matches the holdup of
Sf = Dθ , SG = D(π − θ ), Si = D sin θ , (24) 125 data points within an average percent error of 6.72 and
absolute average percent error of 11.14. As demonstrated in
θ − 0.5 sin 2θ Table 3, the error of presented model was the lowest when
Hf = . (25)
π compared to other methods studied.
By contrast, Abdul-Majeed’s [16] model, developed to calcu-
3.2. Computational procedure late slug liquid holdup in horizontal and slightly inclined two-
phase slug flow, has shown scattered results. This empirical
The simple sub model of Taitel and Barnea [8] for horizon- model is not valid for superficial liquid velocities higher than
tal gas–liquid slug flow has been reformulated by Orell [9]. This 2.316 m/s, so outlier data are eliminated for comparison.
model could not predict the slug unit liquid holdup theoreti- Abdul-Majeed [16] and his colleagues did not provide veri-
cally, so the experimental correlation of Andreussi and Bendik- fication results of their model for horizontal pipes. They mixed
sen [10] was used for slug holdup calculations. horizontal data with upward and downward inclined flow data
Also, the slug length is not obtainable from this model. In the (−5 < Θ < 10), then compared and calculated the APE of their
present study, no experimental correlations for film and slug model with horizontal-upward inclined flow and horizontal-
holdups are used, the slug unit liquid holdup being calculated by downward inclined flow data sources.
the equations referred to in the following solution procedure: The error was lower than ±10%, but comparison of Abdul-
1. Following parameters e.g. ρL , ρG , µL , µG , D, USL , USG should Majeed’s [16] model to the experimental data of horizontal
be defined as inputs. pipes shows that this empirical model is not accurate enough
2. Eqs. (1), (2) and (11) constitute the basic equations of the for horizontal pipes.
model. Substituting the required parameters from Eqs. (8) One reason for this problem could be that the empirical cor-
and (12)–(25), result in a set of equations. relation presented by Abdul-Majeed [16] only contained liquid
3. The equations are solved analytically providing the slug flow to gas velocity ratios. Corresponding to literature, the empirical
characteristic variables, Uf , UG , Hf , HS . A computer program correlations which used other significant flow characteristics
was used to solve the equations; (e.g. gas density, liquid density, surface tension, etc.) are more
4. Using the equations below, the average liquid holdup in slug accurate.
unit, HSU , and the average void fraction in the slug unit, ESU , Gomez et al. [14] developed a dimensionless correlation for
were defined: calculating slug liquid holdup in horizontal and upward vertical
lS lf pipes. The dimensionless groups of this correlation are inclina-
HSU = HS + Hf , (26) tion angle and Reynolds number.
l l
The Reynolds number was defined by a mixture velocity
lS lf
ESU = (1 − HS ) + (1 − Hf ) . (27) without considering the slippage of phases, and the effect of the
l l gas phase was ignored for calculation of density and viscosity.
Thus, it is not surprising that the recent model, which considers
4. Results and discussion the slippage of phases and effects of both phases in flow char-
acteristics, provides more accurate results.
The proposed model was tested against the experimental Armand and Massima’s model [25] was considered in the
liquid holdup data of air–water from an available setup and category of K εH correlations by Ghajar and Woldesemayat [26].
1658 R. Nobakht Hassanlouei et al. / Scientia Iranica, Transactions C: Chemistry and Chemical Engineering 19 (2012) 1653–1660

Table 1: Details of data bank used for model validation.

Data source Diameter (m) Gas velocity (m/s) Liquid velocity (m/s) Fluids Data points

Available set up 0.03 1.7–4.2 0.39–1.8 Air–water 25

Heywood and Richardson [27] 0.042 0.3–4.5 0.48–1.47 Air–water 43
Abdul-Majeed [16] 0.051 1.98–3.07 0.22–1.52 Air–oil 9
Nicholson et al. [5] 0.0512 0.02–4.0 0.12–1.86 Air–oil 31
Kokal and Stanislav [6] 0.0763 0.1–5 0.5–1 Air–oil 17

Table 2: The range of flow parameters of empirical correlations.

Empirical correlation Angle (°) Diameter (cm) Fluids

Gomez et al. 0–90 7.6, 17.8, 20.3, 5.1 Air–kerosene, freon–water, air–water, nitrogen–diesel, air–oil
Model [14]
Leung [25] 0–90 5.25, 10.22, 2.54, 3.81, 4.55, 7.80, 1.9, 5.08, 1.27 Air–water, air–kerosene, natural gas–water
Abdul-Majeed [16] −10–+9 7.62, 5.10, 17.145, 20.32, 2.58 Air–kerosene, air–light oil, freon–water, air–water, nitrogen–diesel

Table 3: Comparison of slug unit liquid holdup predictions with data sources.

Data source Present model Gomez et al. model [14] Leung [25] Abdul-Majeed [16] Zhang et al. [13]
APE AAPE APE AAPE APE AAPE APE AAPE APE AAPE

Available set up 0.394 1.44 33.69 33.69 26.42 26.42 56.01 56.01 26.53 26.53
Heywood and Richardson [27] 11.62 15.93 29.57 29.57 1.75 3.89 56.27 56.27 38.48 38.48
Abdul-Majeed [16] 14.61 14.61 8.90 8.90 41.67 41.67 12.61 12.61 43.08 43.08
Nicholson et al. [5] 11.49 12.80 34.69 36.56 1.03 15.44 36.14 36.14 38.45 38.45
Kokal and Stanislav [6] 6.24 8.41 24.19 32.30 27.76 27.76 34.67 34.67 36.43 36.43
Total error 6.72 11.14 29.12 30.72 12.46 16.87 44.26 44.26 36.13 36.13

Table 4: Comparison of slug unit liquid holdup predictions with all data sources.

Presented model Gomez et al. model [14] Leung [25] Abdul-Majeed [16] Zhang et al. [13]

R2 0.94 0.83 0.89 0.78 0.82

This correlation consists of constant (K ) and some functional

multiple of the no-slip void fraction, εH . This model was not
developed specifically for slug flow, but it was based on a wide
range of data bank for horizontal two phase flow [26]. This has
better results compared to other empirical models.
It can be concluded that the empirical models which are
based on a wide range of data are more reliable than other
empirical correlations. However, they depend on their data
banks. The present model, which is a combination of some
correlations and theoretical approaches, is still more accurate.
In the Zhang model [13], slug liquid holdup can be calculated
by a balance between the turbulent kinetic energy of the liquid
phase and the surface free energy of dispersed spherical gas
bubbles. To evaluate slug liquid holdup in various inclination
angles, these two terms are entered in the Zhang model [13].
As Table 3 shows, the Zhang et al. [13] model does not produce
better accuracy compared to the present model.
One of the important reasons can be found in the slug length
Figure 10: Comparison of developed model to all experimental data.
term. In the Zhang model [13], 32D is used as the slug length
value (ls ), but as mentioned above, Xin et al. [23] concluded that
when the gas velocity is between 3 and 6 m/s and the liquid The presented model matches the data sources perfectly
velocity is lower than 2 m/s (the conditions of the present data with a regression coefficient of 0.94. Meantime, the results
bank), the slug length is between 10 and 20 D. So, it seems that of the developed model versus gas superficial velocity for
the Zhang model [13] is not appropriate for the presented data air–water and air–oil data sources are shown in Figures 11 and
and the high error of the model might be the result of this fact. 12, confirming the model validity.
Hence, for horizontal pipes, the recent model is suggested.
For a better comparison, the proposed model results are 5. Conclusions
compared to all experimental data sources, the results being
shown in Figure 10. The regression coefficient (R2 ) is also In the present model, the value of slug unit length (l) was
calculated for all models in Table 4. calculated by the Dukler and Hubbard [4] correlation. On the
 R. Nobakht Hassanlouei et al. / Scientia Iranica, Transactions C: Chemistry and Chemical Engineering 19 (2012) 1653–1660 1659

References

[1] Ansari, M.R. and Shokri, V. ‘‘Numerical modeling of slug flow initiation in
horizontal channels using a two-fluid model’’, Int. J. Heat Fluid Flow, 32,
pp. 145–155 (2011).
[2] Taitel, Y. and Barnea, D. ‘‘Two-phase slug flow’’, Adv. Heat Transfer, 20,
pp. 43–132 (1990).
[3] Mazza, R.A., Rosa, E.S. and Yoshizawa, C.J. ‘‘Analyses of liquid film models
applied to horizontal and near horizontal gas–liquid slug flows’’, Chem.
Eng. Sci., 65, pp. 3876–3892 (2010).
[4] Dukler, A.E. and Hubbard, M.G. ‘‘A model for gas–liquid slug flow in
horizontal and near horizontal tubes’’, Ind. Eng. Chem. Fundam., 14,
pp. 337–347 (1975).
[5] Nicholson, M.K., Aziz, K. and Gregory, G.A. ‘‘Intermittent two phase flow
in horizontal pipes: predictive models’’, Can. J. Chem. Eng., 56, pp. 653–663
(1978).
[6] Kokal, S.L. and Stanislav, J.F. ‘‘An experimental study of two-phase flow in
slightly inclined pipes-II: liquid holdup and pressure drop’’, Chem. Eng. Sci.,
44, pp. 681–693 (1989).
[7] De Henau, V. and Raithby, G.B. ‘‘A transient two-fluid model for the
simulation of slug flow in pipelines-I, theory’’, Int. J. Multiphase Flow, 21,
pp. 335–349 (1995).
[8] Taitel, Y. and Barnea, D. ‘‘A consistent approach for calculating pressure
Figure 11: Comparison of predicted slug unit liquid holdup and air–water drop in inclined slug flow’’, Chem. Eng. Sci., 45, pp. 1199–1206 (1990).
[9] Orell, A. ‘‘Experimental validation of a simple model for gas–liquid slug
experimental data.
flow in horizontal pipes’’, Chem. Eng. Sci., 60, pp. 1371–1381 (2005).
[10] Andreussi, P. and Bendiksen, K. ‘‘An investigation of void fraction in liquid
slugs for horizontal and inclined gas–liquid pipe flow’’, Int. J. Multiphase
Flow, 15, pp. 937–946 (1989).
[11] Andreussi, P., Minervini, A. and Paglianti, A. ‘‘Mechanistic model of slug
flow in near-horizontal pipes’’, AIChE. J., 39, pp. 1281–1291 (1993).
[12] Al-Safran, E.M., Taitel, Y. and Brill, J.P. ‘‘Prediction of slug length
distribution along a hilly terrain pipeline using slug tracking model’’,
J. Energy Res. Technol., 126, pp. 54–62 (2004).
[13] Zhang, H.Q., Wang, Q., Sarica, C. and Brill, J.P. ‘‘A unified mechanistic model
for slug liquid holdup and transition between slug and dispersed bubble
flows’’, Int. J. Multiphase Flow, 29, pp. 97–107 (2003).
[14] Gomez, L.E., Shohama, O. and Taitel, Y. ‘‘Prediction of slug liquid holdup:
horizontal to upward vertical flow’’, Int. J. Multiphase Flow, 26, pp. 517–521
(2000).
[15] Issa, R.I. and Kempf, M.H.W. ‘‘Simulation of slug flow in horizontal and
nearly horizontal pipes with the two-fluid model’’, Int. J. Multiphase Flow,
29, pp. 69–95 (2003).
[16] Abdul-Majeed, G.H. ‘‘Liquid slug holdup in horizontal and slightly inclined
two-phase slug flow’’, J. Pet. Sci. Eng., 27, pp. 27–32 (2000).
[17] Soleimani, A. and Hanratty, T.J. ‘‘Critical liquid flows for the transition from
Figure 12: Comparison of predicted slug unit liquid holdup and air–oil the pseudo-slug and stratified patterns to slug flow’’, Int. J. Multiphase Flow,
experimental data. 29, pp. 51–67 (2003).
[18] Azzi, A., Friedel, L., Kibboua, R. and Shannak, B. ‘‘Reproductive accuracy of
two-phase flow pressure loss correlations for vertical 90° bends’’, Forsch.
other hand, liquid slug length (ls ) was estimated by the Gregory Ingenieurwesen, 67, pp. 109–116 (2002).
[19] Azzi, A., Alger, U.S.T.H.B. and Friedel, L. ‘‘Two-phase upward flow 90° bend
and Scott [22] correlation. Two critical unknowns for calculat- pressure loss model’’, Forsch. Ingenieurwesen, 69, pp. 120–130 (2005).
ing slug unit liquid holdup (HS , Hf ) are found by solving mo- [20] Balakotaiah, V., Ungar, E., Motil, B.J., Lastochkin, D. and Dao, E.K. ‘‘Modeling
mentum equations. and experimental studies on gas–liquid two-phase flows at normal and
microgravity conditions’’, The University of Houston, Isso Annual Report
So, unlike the Orell model [9], the calculation of slug liquid (2000).
holdup is not dependent on a specific range and is not as [21] Brennen, C.E., Fundamentals of Multiphase Flow, Cambridge University
sophisticated as the Taitel and Barnea [8] model. Press, New York (2005).
[22] Gregory, G.A. and Scott, D.S. ‘‘Correlation of liquid slug velocity and
The experimental work was carried out to gather liquid frequency in horizontal co-current gas liquid slug flow’’, AIChE J., 15,
holdup air–water data in horizontal and inclined pipes. The pro- pp. 933–935 (1969).
posed model was tested extensively against 125 experimental [23] Xin, W., Liejin, G. and Ximin, Z. ‘‘Development of liquid slug length
in gas–liquid slug flow along horizontal pipeline: experiment and
data sets. The average percentage error and absolute average simulation’’, Chinese. J. Chem. Eng., 14, pp. 626–633 (2006).
percentage error were calculated for model comparison. [24] Hanratty, J.T. and Cohen, S.L. ‘‘Effects of waves at a gas–liquid interface on
The results show that the total average percentage error a turbulent air flow’’, J. Fluid Mech., 31, pp. 467–469 (1968).
[25] Leung, L.K.H. ‘‘Two phase flow, thermal hydraulics branch, Chalk river
of the present model was 6.72 and the average absolute laboratories’’, AECL, UNENE Thermal Hydraulic Scours, (2005).
percentage error was 11.14, both of which are lower than other [26] Ghajar, A.J. and Woldesemayat, M.A. ‘‘Comparison of void fraction
empirical and mathematical models. These results substantiate correlations for different flow patterns in horizontal and upward inclined
pipes’’, Int. J. Multiphase Flow, 33, pp. 347–370 (2007).
that the new modified model is independent of various physical [27] Heywood, N.I. and Richardson, J.F. ‘‘Slug flow of air–water mixtures in
properties of flowing fluid, so it has an adequate prediction a horizontal pipe: determination of liquid holdup by X-ray absorption’’,
potential for a broad range of situations for slug liquid holdup Chem. Eng. Sci., 34, pp. 17–30 (1979).
calculations in horizontal pipes.

Reza Nobakht Hassanlouei obtained B.S. and M.S. degrees in Chemical

Acknowledgments Engineering from Iran University of Science and Technology (IUST), in 2008
and 2011, respectively. He is currently working as a researcher in the catalyst
research and development department at IUST. His research interests include:
The authors will like to thank the support received for this fluid flow, neural network, CFD and catalyst, and he has published papers in
project from the National Iranian Gas Company. these fields.
1660 R. Nobakht Hassanlouei et al. / Scientia Iranica, Transactions C: Chemistry and Chemical Engineering 19 (2012) 1653–1660

Hasti Firouzfar began her studies at Arak University, Iran, in 2002, and obtained he established the Computer Aided Process Engineering Research Laboratory
an M.S. degree in Chemical Engineering from Iran University of Science and (CAPE).
Technology (IUST), in 2009. She has worked as a lecturer in Azad University, His research interests include software consultation, training, design,
Iran, and has been teaching assistant for a Modeling and Simulation course at writing and development, process feasibility evaluation, equipment & layout
IUST since 2009. Her research interests include: fluid flow and multiphase flow. design, simulation, optimization & analysis.
She is currently working in the Computer Aided Process Engineering Laboratory
at IUST as a research assistant.
Mohammad Hassan Khanof began his studies at Sharif University of
Technology, Tehran, Iran, in 1975, after which he obtained an M.S. degree
in Chemical and Mechanical Engineering from the University of Southern
Norollah Kasiri began his studies at Glamorgan University, UK, and obtained California, USA, in 1977. He has been lecturer in the Chemical Engineering
M.S. and Ph.D. degrees in Chemical Engineering from Swansea University, UK, Department at Iran University of Science and Technology, since 1989, where
in 1988 and 1993, respectively. Since then, he has been lecturer at the Chemical he also manages the Fluid Flow Laboratory. His research interests include fluid
Engineering Department in Iran University of Science and Technology, where flow and multiphase flow, in which he has published papers.