Numerical investigation of upward air-water annular, slug and bubbly

flow regimes

M. Hassani, M. Bagheri Motlagh and R. Kouhikamali *

Faculty of Mechanical Engineering, University of Guilan, Rasht, Iran

Article info: Abstract

Type: Research In this paper, numerical investigation of upward two phase flow of air-water has
Received: 18/07/2018 been studied. Different conditions of flow regimes including annular, wispy
Revised: 08/02/2019 annular, slug, churn and bubbly are simulated based on Hewitt and Roberts map,
and a good agreement between the experimental data of the map and the
Accepted: 15/02/2019
numerical simulation has been observed. Accordingly, a proper CFD model in
Online: 20/02/2019 CFD software of Fluent with the required User Defined Function (UDF) has
Keywords: been obtained to simulate two phase flows of fluids with large density ratio in
Numerical simulation, vertical tubes. The simulation is carried out with the volume of fluid (VOF)
Two phase flow regimes, method and piecewise interface calculation (PLIC) algorithm for tracking the
interface for the annular, wispy annular, churn and slug flow regimes and drift
Volume of fluid method,
flux model for bubbly with proper selection of computational cell and time step
Annular, sizes. Furthermore, water and air momentum fluxes have been changed and the
Slug, changes to the flow patterns are studied.
Bubbly.

1. Introduction field of multiphase flows is based on

experiments. The flow pattern maps of Baker [1]
Many industrial phenomena have been opposed for horizontal tubes and Bennett et al. [2] and
with two phase flow matter. These two phase Hewitt and Roberts [3] for vertical tubes are
flows consisting of different regimes can occur supplied for special geometry and operating
in nuclear power generation, refrigeration and oil conditions. Recently, Waltrich et al. [4], Shaban
and gas industry. and Tavoularis [5] and Ansari and Azadi [6]
Due to the abstruse interaction between phases, have experimentally provided flow patterns of
phases different density and viscosity, extra air-water flow in vertical tubes. Waltrich et al.
turbulence and resulting effective forces, two [4] aimed at long vertical tubes to evaluate the
phase flows have complex dynamics and this axial development of gas-liquid annular, churn
issue can direct the phases to have some different and slug flows. Shaban and Tavoularis [5]
configuration and flow patterns. Therefore, presented an accurate and easy method to
different ranges of operating conditions cause discover the flow regimes. Ansari and Azadi [6]
the multiphase flows to behave in different ways. investigated the differences of the flow patterns
The most and reliable research carried out in the caused by changing the tube diameters. They

*
Corresponding author
Email address: [email protected] 331
JCARME M. Hassani, et al. Vol. 9, No. 2

also considered that different axial location and forepassed method can be resolved. Taha and
different inlet type change the flow patterns. Cui [15] used Volume of Fluid (VOF)
Yang and Shieh [7] found out that beside multiphase model and simulated slug flow
buoyant force and turbulent fluctuations, the regime. The slugs shape, velocity and
surface tension force also affects the flow regime distribution of velocity and wall shear stress
determination in small tubes. Therefore, the were also studied. Schepper et al [16], based on
bubble flow and slug to annular flow transitions Baker chart [1], simulated the gas-liquid and
can happen earlier or later for different fluids vapor-liquid flow regimes with the use of VOF
with different values of surface tension. Kozulin multiphase model and confirmed that a CFD
and Kuznetsov [8] studied the flow regimes and code can correctly compute the variables of
statistical characteristics of two-phase flow of horizontal tube two phase flow. Frank et al. [17]
liquid gas in a vertical micro-channel considered interphase momentum transfer due to
experimentally. Milan et al. [9] investigated the governing drag and non-drag forces and
effects of inlet device, flow history and developed new multiphase flow models for
development length on the downward flow mono- and poly-disperse bubbly flows by
regimes of vertical tubes experimentally. They Ansys-CFX. The effect of inlet conditions on the
resulted in the dependence of mentioned bubble formation of air with three liquids of
conditions on the pattern and boundary water, octane and semi-octane was studied by
transitions. Talley et al. [10] also worked on the CFD [18]. The influences of gas and liquid
flow visualization of air-water in horizontal velocities and inlet gas nozzle size and thickness
pipes and characterized 27 flow conditions of on the bubble size were also studied. It was
bubbly, plug, slug, stratified, wavy and annular concluded that surface tension has a strong effect
flow regimes. Chen et al. [11] visualized rather than viscosity and density. Wei et al. [19]
dispersed bubble, bubbly, confined bubble, slug, used CFD modeling for studying shear stress in
churn, annular and mist flow patterns of R134a a flat sheet membrane bioreactor. For this
in a vertical upward tube. They studied the effect purpose, they used volume of fluid model.
of tube diameter on the transition boundaries of Rahimi et al. [20] investigated air-water slug
flow patterns. Zhang et al. [12] investigated the flow characteristics like its length and pressure
transition mechanism and criterion of bubbly drop in the horizontal tube and studied the
flow to slug flow. They found out that the differences with the flow in inclined pipelines.
transition happens when the velocity ratio The impact of inlet conditions on bubble to slug
reaches its minimum value. Li et al. [13] flow transition of air-water flow was studied
developed new transition criteria between churn- experimentally and numerically by Gregorc and
turbulent and annular flow for downward flow Zune [21]. Their numerical simulation was in a
regimes of air-water in large diameter pipes. As good agreement with their experimental tests.
one of the industrial applications and importance Pouraria et al. [22] investigated the distribution
of two phase flow regimes, Hanifzadeh et al. [14] of phase fraction of water-oil flow in subsea
investigated the effect of air-water flow patterns horizontal pipelines and discovered flow
on the upriser pipe of airlift pump. The regimes patterns for different operating conditions by
they observed were slug, churn and annular, but using the Eulerian-Eulerian approach of CFD
the best performance of airlift pump was simulation.
reported in the slug flow regime. The main goal of this paper is the CFD
Testing conditions and determinate ranges of simulation of different flow regimes of large
important parameters are the biggest density ratio of air-water flow in a vertical tube.
disadvantages of the experimental work. Such flows, due to the larger velocity
Besides, it is sometimes difficult and costly to discontinuity at the interface, accounts with
conduct experiments which probe all relevant extreme complications. These detected flow
phenomena. Therefore, an alternative way is regimes are based on the Hewitt and Roberts
required. With the progressive growth of CFD map [3]. To carry out the simulation, Piecewise
simulation in recent decades, the deficiencies of Linear Interface Calculation (PLIC) algorithm of

332
JCARME Numerical investigation of . . . Vol. 9, No. 2

VOF method has been employed for slug, churn, In these equations, V and P are velocity and
wispy annular and annular flow regimes, but it is pressure, respectively. 𝜌 (density) and 𝜇
needed to consider drag force for detecting (dynamic viscosity) are calculated based on the
bubbles of the bubbly flow regime. Besides, volume fractions of phases in the computational
suitable parameterization of numerical cell. In other words, these governing equations
simulation of these regimes has been attained. are solved for the mixture fluid by using the
Accordingly, the influences of changing the inlet following mixture rule:
conditions on slug, wispy annular and bubbly 𝑛
flow regimes and their pressure drop have been
investigated. 𝜙 = ∑ 𝛼𝑞 𝜙𝑞 , 𝜙 = 𝜌 𝑎𝑛𝑑 𝜇 (4)
𝑞=1

2. Mathematical model F is the volumetric force due to surface tension

at the interface and is calculated via continuum
Different hydrodynamic conditions of phases surface force (CSF) of Brackbill et al. [23]
can cause them to shape differently beside each
other and make various flow patterns. Numerical 𝛼𝑖 𝜌𝑖 𝜅𝑖 ∇𝛼𝑖 + 𝛼𝑗 𝜌𝑗 𝜅𝑗 ∇𝛼𝑗
𝐹 = ∑ 𝜎𝑖𝑗 (5)
modeling of these regimes needs to solve the 0.5(𝜌𝑖 + 𝜌𝑗 )
𝑖,𝑗
governing equations with a proper numerical
approach. In this paper, the volume of fluid In which, 𝜅 is the interface curvature calculated
(VOF) method of Eulerian-Eulerian approach is as the following:
employed to model the two phase flow regimes
∇𝛼𝑞
of annular, wispy annular, churn and slug. A 𝜅𝑞 = (6)
model with proper drag force is used to model |∇𝛼𝑞 |
bubbly flow. These models solve one set of
conservation equations and track the interface of The extra unknown variable of multiphase flow
moving phases to obtain their volume fractions (volume fraction) and its change along the cells
of each computational cell. are computed with volume fraction equation:
For every phase that is considered, one 𝜕𝛼𝑞
additional equation is added to compute its + (𝑉𝑞 ⋅ ∇)𝛼𝑞 = 0 (7)
volume fraction at computational cells. The 𝜕𝑡
quantity of the dispersed phases is solved by the For the bubbly flow regime, the VOF method
volume fraction equation and the continuous which considers the phases with the same
phase volume fraction is obtained from the fact velocity, causes them to gather and form slugs.
that the cell is occupied by phases with volume Therefore, it is needed to consider the relative
fractions sum to unity. velocity and add the relevant terms to the
𝑛 equations. ∇ ⋅ (∑ 𝛼𝑞 𝜌𝑞 𝑉𝑑𝑟,𝑞 𝑉𝑑𝑟,𝑞 ) is added to the
∑ 𝛼𝑞 = 1 (1) right hand side of the momentum equation to
𝑞=1
designate the relative velocities (𝑉𝑝𝑞 ) of phases
[24].
where 𝛼𝑞 is the qth volume fraction and n
𝑝ℎ𝑎𝑠𝑒𝑠𝑛𝑢𝑚𝑏𝑒𝑟
determines the number of phases. 𝛼𝑘 𝜌𝑘
Conservations of mass and momentum are 𝑉𝑑𝑟,𝑞 = 𝑉𝑝𝑞 − ∑ 𝑉 (8)
𝜌 𝑞,𝑘
considered as follows, respectively: 𝑘=1
To obtain the relative velocity, the drag force
𝐷𝜌
⃗ =0
+ 𝜌∇ ⋅ 𝑉 (2) between phases is modeled via Schiller and
𝐷𝑡 Naumann model. Furthermore, ∇ ⋅ (𝛼𝑞 𝜌𝑞 𝑉𝑑𝑟,𝑞 )
𝜕
⃗ ) + ∇ ⋅ (𝜌𝑉
(𝜌𝑉 ⃗𝑉⃗) will be added to the left hand side of volume
𝜕𝑡 fraction equation.
= −∇𝑃 + ∇ ⋅ [𝜇(∇𝑉 ⃗ + ∇𝑉
⃗ 𝑇 )] + 𝜌𝑔
+𝐹 (3)

333
JCARME M. Hassani, et al. Vol. 9, No. 2

Now, it is needed to capture the position of the Eulerian-Eulerian approach. The slug, annular,
interface of the phases. There are different wispy annular, bubbly and churn flow regimes
models for interface tracking [25-28]. Among are captured based on Hewitt and Roberts map
them, the VOF method, due to better mass [3]. This map is created based on low pressure
conservation, simpler use in 3-D geometries and air-water flow and high pressure steam-water in
unstructured grid, is attended more. Piecewise vertical tubes with diameters 1-3 cm with
linear interface calculation (PLIC) algorithm superficial gas and liquid momentum fluxes as
[29] in VOF model is an accurate technique vertical and horizontal axes.
which models the interface by three steps. In the Accordingly, a tube with 25.4 mm diameter has
first step with the help of the amount of phase been considered. Air and water with special
volume fraction in the main and neighboring mass fluxes, from distinct inlets, according to the
cells, the slope of the interface of the phases is desirable flow regime, at atmospheric pressure
determined. The second step consists of and room temperature enter the tube. No slip
modeling interface change caused by the condition is considered for the wall and with the
velocity field and inlet mass flux through the use of the axis boundary, one half of the tube is
face. In the last step, based on the balance of modeled in 2-D. Inlet conditions of air and water
fluxes of the last step, phase volume fraction is for considered flow regimes are presented in
computed. Table 1.
For simulating the turbulences of the problem,
the k-ɛ Realizable turbulence model [30] with the Table 1. Inlet conditions of air and water for
enhanced wall function formulation is used. considered flow regimes at the current study.
Water mass Air mass
In the current numerical simulation, the Flow regime Case flux flux
commercial CFD software of Fluent 6.3 with the name (kg/m2.s) (kg/m2.s)
required User Defined Function (UDF) has been sa 315.9 1.1
used. The pressure implicit with splitting of
sb 706.5 1.1
operators (PISO) is implemented as the pressure- Slug
velocity coupling scheme. The PRESTO! sc 999.1 1.1
scheme interpolates pressure with the sd 706.5 0.14
discretization method used for momentum
equation and turbulent variables is the second- Churn c 315.9 5
order upwind. The problems are solved with ba 29267.2 3.5
adaptive time step based on constant courant
bb 91834.4 3.5
number and average time step size of ∆𝑡 = 5𝑒 − Bubbly
5𝑠 for slug, ∆𝑡 = 𝑙𝑒 − 6𝑠 for bubbly, churn and bc 91834.4 1.9
annular and ∆𝑡 = 𝑙𝑒 − 7𝑠 for wispy annular bd 91834.4 6.1
flow regime.
The under relaxation factors for solving Annular aa 315.9 95.1
pressure, density, body force, momentum, waa 9183.4 95.1
turbulent kinetic energy, turbulent specific
wab 29267.2 95.1
dissipation rate and turbulent viscosity are Wispy annular
considered 0.3, 0.7, 0.7, 0.7, 0.8, 0.8, and 0.8 wac 91834.4 95.1
respectively. wad 91834.4 228.5

3. Geometrical configuration and operating

conditions The grid of the geometry consists of structured
quadrilateral cells with 0.5 mm width and length
A two-dimensional simulation of air-water for slug, churn and annular flow regimes and
upward flow is carried out by the VOF method 0.25 mm for bubbly and wispy annular flow
(for annular, wispy annular, churn and slug) and regimes. Increasing the cell numbers just causes
drift flux model (for bubbly flow regime) of the interface of two phases become more

334
JCARME Numerical investigation of . . . Vol. 9, No. 2

accurate and places every phase (air volume needed to add proper terms of surface tension
fraction of 1 or 0) in every cell. term to the governing equations of drift flux two
phase model.
4. Results and discussion As in Fig. 1b, the experimental vapor slug
inclined a little to the right part of the tube has a
In order to investigate the validity of the circular nose and there are a lot of bubbles in the
numerical procedure for capturing the flow tail of the slug. While the 2-D symmetrically
regimes, the bubbly and slug flow regimes of simulated slug is a little smaller but its form and
experimental work of Ansari and Azadi [6] have nose are coincident. The tailing bubbles are
been numerically modeled and shown in Fig. 1. adhered and next to each other. This issue can be
caused by disability of turbulent model in
simulating the turbulence and vortexes of the tail
of the air slug. The bubbles in the flow are
detected in the numerical visualization but the
numbers are lower. Totally, the numerical slug
and the two phase flow agree with the
experimental work but as explained, there are
some defections.
The pressure drop of the bubbly and slug flow
(a)
regime of the numerical results of Fig. 1 has been
computed and is compared with the separated
flow model pressure drop [31].

𝑑𝑃 𝑑𝑃 𝑑 𝑥 2 𝜈𝑔 (1−𝑥)2 𝜈𝑓
− ( ) = − ( ) + 𝐺2 ( + )+
𝑑𝑧 𝑑𝑧 𝐹 𝑑𝑧 𝛼 1−𝛼
𝑔𝑠𝑖𝑛θ(𝛼𝜌𝑔 + (1 − 𝛼)𝜌𝑓 ) (9)

The frictional pressure gradient is obtained from

(b)
the following relation:
𝑑𝑃 𝑑𝑃
−( ) = −( ) 𝜙2 =
𝑑𝑧 𝐹 𝑑𝑧 𝑓𝑜 𝑓𝑜
Fig. 1. Comparison of volume fraction contours of
2𝑓𝑓𝑜 𝐺 2 𝜈𝑓 2
experimental [6] and numerical bubbly (a) and slug ( ) 𝜙𝑓𝑜 (10)
(b) flow regimes. 𝐷

2
The two phase frictional multiplier 𝜙𝑓𝑜 , is
As shown in Fig. 1a, the bubbly flow regime of
numerical simulation and the experiment [6] has calculated based on the Friedel correlation [31].
been compared. Due to transient behavior of the 3.24𝐴2 𝐴3
2
flow, the bubbles in the numerical simulation do 𝜙𝑓𝑜 = 𝐴1 + 0.045
(11)
𝐹𝑟 𝑊𝑒 0.035
not have the same place as the experimental where
work. But the size of the bubbles seems the 𝜌𝑓 𝑓𝑔𝑜
same; somewhere with fairly large diameter and 𝐴1 = (1 − 𝑥)2 + 𝑥 2 (12)
𝜌𝑔 𝑓𝑓𝑜
somewhere with small one. The volume of the
total air bubbles seems the same too. To simulate 𝐴2 = 𝑥 0.78 (1 − 𝑥)0.224 (13)
0.91 0.19 0.7
bubbly flow regime, it is needed to consider the 𝜌𝑓 𝜇𝑔 𝜇𝑔
relative velocity between phases. Using the VOF 𝐴3 = ( ) ( ) (1 − ) (14)
𝜌𝑔 𝜇𝑓 𝜇𝑓
flow regime causes the bubbles to adhere and 𝐺 2
small slugs to form. As the relative velocity is 𝐹𝑟 = (15)
𝑔𝐷𝜌2
considered, the slugs are separated. But the
commercial CFD codes do not support the two 𝐺 2𝐷
𝑊𝑒 = (16)
phase flow with surface tension. Therefore, it is 𝜌𝜎

335
JCARME M. Hassani, et al. Vol. 9, No. 2

where x is the gas quality, ffo and fgo are friction

factor based on total flow assumed liquid and
gas, respectively, G is the mass velocity and D is
the diameter. The second right term of the Eq. (9)
is the accelerational pressure gradient which is
zero here due to not changing the quality. The
third right term of Eq. (9) is the gravitational
pressure gradient with θ equals to 90 degrees for
vertical tubes.
Table 2 shows the comparison between the
calculated pressure drop of the flows [31] with (a) (b) (c) (d)
the current numerical results.
Fig. 2. Slug flow regime for water and air mass flux
of 315.9 and 1.1 kg/m2.s (a), water and air mass flux
Table 2. Comparison of the pressure drop of the
of 706.5 and 1.1 kg/m2.s (b), water and air mass flux
flows based on Eq.(9) [31] and the current study.
of 999.1 and 1.1 kg/m2.s (c) and water and air mass
Bubbly flow Slug flow (Fig.
Pressure drop (Pa) flux of 706.5 and 0.14 kg/m2.s (d)
(Fig. 1(a)) 1(b))
Based on Eq. (9) 1378 725
Present numerical With increasing water velocity in Fig. 2)b( (case
1720 850
results sb), the volume of the slugs has got smaller in
comparison with Fig. 2)a(, their shapes are more
4.1. Slug and churn flow regimes uniform and their tails are approximately flat. In
addition, the total pressure drop of the flow has
The slug flow regime contains large bubbles increased according to Table 3. In sc case (Fig.
with the diameter nearly equal to the tube 2)c(), water velocity is the highest while air mass
diameter. They have spherical nose and a flat or flow rate is like case sb. It can be observed that
protracted tail. Fig. 2 shows four cases related to not only the sizes of the slugs sizes are smaller
the slug flow regime, introduced in Table 1. To but also their numbers have decreased in a
simulate this regime, air enters from the middle special length of the tube. On the other hand, the
with 7.5 mm length for cases sa, sb and sc and 9 produced slugs, under the influence of higher
mm for the case sd. Smaller inlet length causes water velocity, do not have a continuous shape
the air to have more speed and tends the regime and there can be seen some water bubbles in
to have bubbles between slugs or be bubbly slug. among. As Table 3, the increase of the pressure
It is needed to consider time step size of 5e-5 drop is observed with water mass flux
seconds. increasing. In sd case (Fig. 2)d(), the mass flow
Fig. 2 )a( is the case with the lowest amount of rate of water is equal to the case sb, but the air
water mass flow rate (case sa). As can be seen, mass flow rate is lower. The decrease in the air
the slugs' shape differs slightly, all of them have mass flow rate causes the slugs to become
spherical nose but with different tails. bigger. Formation of bigger slugs with higher
The lengths of liquid slugs (the length between velocity leads to lower numbers in the length and
two air slugs) are approximately equal except for a sharp increase in pressure drop of the flow.
the two last ones. Despite different shapes of In Fig. 1)b(, the slug flow regime of the
slugs' tails, the size of the slugs seems equal. experimental work in comparison with the
This can be explained with the size of the back current simulation reveals the turbulence
liquid slugs which is in the relation with their defection in capturing the bubbles at the bottom
velocity. of the slug tail. But in the slugs of Fig. 2, there is
no vapor flow below each slug and it seems that
in the flow regime with only slugs (not the slugs
with bubbles), the turbulence model does not
have effective influence.

336
JCARME Numerical investigation of . . . Vol. 9, No. 2

Table 3. Pressure drop of flows of cases sa to sd. 4 compares the pressure drop of the bubbly flow
Water mass Air mass regime.
Pressure
Case flux flux
drop (Pa)
(kg/m2.s) (kg/m2.s)
sa 315.9 1.1 310
sb 706.5 1.1 520
sc 999.1 1.1 645
sd 706.5 0.14 1100

In Fig. 3, the resultant changes of increasing air

momentum flux to churn part of the map are
compared with sa case with the same water
momentum flux. The size of the slugs along the
cross section and length is sensibly bigger with
irregular shape and water in among. The
diameters of the slugs are increasing to near the
wall and their lengths are extended as a transition
between slug and annular flow regimes. In spite
of the irregular shape of the churn flow regime,
the increase of the air mass flux has reduced the
total flow density and decreased the pressure Fig. 3. Comparison between slug flow regime with
drop to 100 Pa. water and air mass flux of 315.9 and 1.1 kg/m2.s (case
sa) and churn flow regime with water and air mass
4.2. Bubbly flow regime flux of 315.9 and 5 kg/m2.s

The bubbly flow regime includes small bubbles

dispersed in continuous liquid phase. As shown
in Fig. 4, four cases of bubbly flow have been
considered. This regime must be solved with
small time step near to le-6 seconds. Besides, to
capture the bubbles, the quadrilateral cells with
0.25 mm width and length are generated.
Otherwise, the convergences would not be
obtained.
The bubbly flow regime has to be simulated
based on the drift flux two phase flow by which
a b c d
the relative velocity of phases is considered.
Besides, the force due to surface tension has to Fig. 4. Bubbly flow regime for water and air mass
flux of 29267.2 and 3.5 kg/m2.s (a), water and air
be included in the momentum equation. Without
mass flux of 91834.4 and 3.5 kg/m2.s (b), water and
the relative velocity, the bubbles gather and air mass flux of 91834.4 and 1.9 kg/m2.s (c) and water
slugs are formed while surface tension prevents and air mass flux of 91834.4 and 6.1 kg/m2.s (d)
the elongated air instead of bubble shape. In the
first case (Fig. 4(a)) there seem bubbles of air Table 4. Pressure drop of flows of cases ba to bd.
with different sizes and shapes. Most of the Water mass Air mass
Pressure
bubbles are elongated. By increasing water Case flux flux
drop (Pa)
momentum flux in case bb, the bubbles get more (kg/m2.s) (kg/m2.s)
ba 29267.2 3.5 1210
spherical and their shapes are more uniform. bb 91834.4 3.5 1400
Decrease and increase in air momentum flux in bc 91834.4 1.9 1460
comparison with case bb are shown in Figs. 4c bd 91834.4 6.1 1370
and 4d. Decreasing and increasing the bubble
number and size are the respective results. Table

337
JCARME M. Hassani, et al. Vol. 9, No. 2

According to Table 4, as the gas volume is the same. It can be seen that liquid droplets
increased in the domain, the less the pressure have decreased but gas bubbles in the film
dropped. More water mass flux in comparison increased. The extra amount of air causes the
with the slug flow regime has caused more liquid layer to get wavier too. The pressure drop
pressure drop in the flow. for Fig. 5c, 5d and 5e is obtained 375, 900 and
800 Pa, respectively.
4.3. Annular and wispy-annular flow regimes

In annular flow regime, a liquid film forms on

the wall of the tube. The surface of this film can
be wavy, but the liquid and gas are completely
distinct. In other words, there are no liquid
droplets in the core of the flow. But, the wispy
annular regime consists of a thick liquid film on
the wall which can be full of gas bubbles. On the
other hand, the core of the flow can include
liquid droplets. a b c d e
To obtain the annular and wispy annular Fig. 5. Annular flow regime for water and air mass
regimes, the water inlet is placed along the wall flux of 315.9 and 95.1 kg/m2.s (a), wispy annular flow
and a mean time step value of le-6 and le-7 regime for water and air mass flux of 9183.4 and 95.1
seconds is used for annular and wispy-annular kg/m2.s (b), water and air mass flux of 29267.2 and
95.1 kg/m2.s (c), water and air mass flux of 91834.4
regimes, respectively.
and 95.1 kg/m2.s (d) and water and air mass flux of
As can be seen in Fig. 5a, there is a wavy liquid 91834.4 and 228.5 kg/m2.s (e)
film on the wall. This film is quietly thin and the
boundary between liquid and gas is completely 5. Conclusions
distinguishable. Due to high velocity of gas flow,
all the annular flows simulated in the current In this study, different flow regimes of air-water
study had the form like Fig. 5a and is not (annular, wispy annular, slug, bubbly and churn)
presented. With moving toward the wispy based on Hewitt and Roberts map [3] have been
annular part of the map which is caused by investigated in a vertical tube of 25.4 mm
increasing liquid velocity, the film on the wall diameter. The major purpose of this work is the
got thicker. Therefore, the pressure drops more CFD method which can predict the flow
from nearly 10 to 140 Pa from case aa to waa. In regimes. Therefore, the PLIC algorithm of VOF
Fig. 5b (case waa), a completely wavy liquid two phase flow is used to simulate the regimes
film can be seen. Big droplets of liquid are and tracking the interfaces with proper selection
separating and there are gas bubbles trapped in of computational cell and time step sizes, 2-D
the middle of the liquid film. With increasing symmetrical model blinks some asymmetries. It
liquid velocity in the wispy annular flow part was observed that although the 𝑘 − 𝜀 realizable
(Fig. 5c (case wab)), the waviness of the liquid turbulence model can help predict the total flow,
layer diminishes. The gas bubbles exist in the it is weak to simulate the vortexes caused at the
liquid film consistently and there can be seen tail of the vapor slugs. Besides, for simulating
more gas bubbles in the core of the flow. the bubbly flow regime, it is needed to consider
Like the changes in Fig. 5c, increases in water the relative velocity of two phases.
velocity cause the surface of the liquid layer By comparing the results with the map of flow in
more flat. The amount of gas bubbles in the film the vertical tube, a good agreement is observed.
has got lower but there is a sensible addition in Also, different conditions of each part in the map
the numbers of the water droplets in the middle are considered.
of the tube. In Fig. 5e (case wad), mass flow rate Changing water and air momentum fluxes cause
of air has increased in comparison with the case different results.
5-wac (Fig. 5d), but the mass flow rate of water

338
JCARME Numerical investigation of . . . Vol. 9, No. 2

 In slug flow regime by increasing water [5] H. Shaban and S. Tavoularis,

velocity, air slugs size gets smaller. Their sizes “Identification of flow regime in vertical
are approximately equal and increasing water upward air water pipe flow using
momentum flux decreases their numbers rather differential pressure signals and elastic
than making them smaller to bubble sizes. maps”, Int. J. Multiphase Flow, Vol. 61,
Increasing the water mass flow rate from 315.9 pp.62-72, (2014).
kg/m2.s to 999.1 kg/m2.s with constant air mass [6] M. R. Ansari and R. Azadi, “Effect of
flux of 1.1 kg/m2.s has increased the pressure diameter and axial location on upward gas
drop more than 50% from 310 to 645 Pa in 0.1 liquid two-phase flow patterns in
m length. Decreasing the air mass flux of 1.1 intermediate-scale vertical tubes”, Ann.
kg/m2.s to 0.14 kg/m2.s at water mass flux of Nucl. Energy, Vol. 94, pp. 530-540,
706.5 kg/m2.s has also increased the pressure (2016).
drop nearly 50%. [7] C. Y. Yang and C. C. Shieh, “Flow pattern
 Wispy annular regime has thicker water of air–water and two-phase R-134a in
layer near the wall. With increasing liquid small circular tubes”, Int. J. Multiphase
velocity, the waves of liquid layer decreases and Flow, Vol. 27, No. 7, pp. 1163-1177,
the amounts of liquid droplets in the core flow (2001).
increase. Increasing gas velocity leads to more [8] L. A. Kozulin and V. V. Kuznetsov,
waves of the liquid surface and less amounts of “Statistical Characteristics of two-phase
liquid droplets. gas liquid flow in a vertical
 Bubble number and size increases with microchannel”, Journal of Applied
increasing air momentum flux and decreasing Mechanics and Technical Physics, Vol.
water momentum flux values. The pressure drop 52, No. 6, pp. 956-964, (2011).
for 1.9 kg/m2.s to 6.1 kg/m2.s air mass flux with [9] M. Milan, N. Borhani, J. R., Thome,
91834.4 kg/m2.s water mass flux has changed “Adiabatic vertical downward air-water
few from 1460 to 1370 Pa, respectively. While flow pattem map: Influence of inlet
decreasing water mass flux from 91834.3 to device, flow development length and
29267.2 kg/m2.s decreases the pressure drop hysteresis effects”, Int. J. Multiphase
from 1400 to 1210 Pa. Flow, Vol. 56, pp. 126-137, (2013).
[10] J. D. Talley, T. Worosz, S. Kim, J. R.
References Buchanan, “Characterization of horizontal
air-water two-phase flow in a round pipe
[1] O. Baker, “Simultaneous flow of oil and part I: Flow visualization”, Int. J.
gas”, oil Gas J., Vol. 53, pp.185-195, Multiphase Flow, Vol. 76, pp. 212-222,
(1954). (2015).
[2] A. W. Bennett, A. W. Hewitt, H. A. [11] L. Chen, Y. S. Tian, T. G. Karayiannis,
Kearsey, R. K. F. Keeys, P. M. C. Lacey, “The effect of tube diameter on vertical
“Flow visualization studies of boiling at two-phase flow regimes in small tubes”,
high pressure”, P. I. Mech. Eng., Vol. 180, Int. J. Heat Mass Tran., Vol. 49, No. 21-
No. 3, pp. 260-283, (1965). 22, pp.4220-4230, (2006).
[3] G. F. Hewitt and D. N. Roberts, “Studies [12] M. Zhang, L. m. Pan, P. Ju, X. Yang, M.
of two-phase flow patterns by Ishii, “The mechanism of bubbly to slug
simultaneous X-ray and flash flow regime transition in air-water two
photography”, UKAEA Report AERE phase flow: A new transition criterion”,
M2159, United Kingdom, (1969). Int. J. Heat Mass Tran., Vol. 108, Part B,
[4] P. J. Waltrich, G. Falcone, J. R. Barbosa, pp.1579-1590, (2017).
“Axial development of annular, churn and [13] Z. Li, G. Wang, M. Yousaf, X. Yang, M.
slug flows in a long vertical tube”, Int. J. Ishii, “Flow structure and flow regime
Multiphase Flow, Vol. 57, pp.38-48, transitions of downward two-phase flow
(2013).

339
JCARME M. Hassani, et al. Vol. 9, No. 2

in large diameter pipes”, Int. J. Heat Mass water slow patterns in a subsea pipeline”,
Tran., Vol. 118, pp. 812-822, (2018). Ocean Eng., Vol. 115, pp. 135-148,
[14] P. Hanafizadeh, S. Ghanbarzadeh, M. H. (2016).
Saidi, “Visual technique for detection of [23] J. U. Brackbill, D. B. Kothe, C. Zemach,
gas–liquid two-phase flow regime in the “A Continuum Method for Modeling
airlift pump”, J. Petrol. Sci. Eng., Vol. 75, Surface Tension”, J. Comput. Phys., Vol.
No. 3-4, pp. 327-335, (2011). 100, No. 2, pp.335-354, (1992).
[15] T. Taha, Z. F. Cui, “CFD modelling of [24] M. Manninen, V. Taivassalo, S. Kallio,
slug flow in vertical tubes”, Chemical On the mixture model for multiphase flow,
Engineering Science, Vol. 61, No. 2, pp. VTT Publications 288, Technical
676-687, (2006). Research Centre of Finland, (1996).
[16] C. K. De Schepper, G. J. Heynderickx, G. [25] G. Tryggvason, B. Bunner, A. Esmaeeli,
B. Marin, “CFD modeling of all gas– D. Juric, N. Al-Rawahi, W. Tauber, J.
liquid and vapor–liquid flow regimes Han, S. Nas, Y. J. jan, “A Front Tracking
predicted by the Baker chart”, Chem. Eng. Method for the Computations of
J., Vol. 138, No. 1-3, pp. 349-357, (2008). Multiphase Flow”, J. Comput. Phys., Vol.
[17] Th. Frank, P. J. Zwart, E. Krepper, H. M. 169, No. 2, pp.708-759, (2001).
Prasser, D. Lucas, “Validation of CFD [26] M. Sussman, P. Smereka, S. Osher, “A
models for mono- and polydisperse air- Level Set Approach for Computing
water two-phase flows in pipes”, Nuclear Solutions to Incompressible Two-Phase
Engineering and Design, Vol. 238, No. 3, Flow”, J. Comput. Phys., Vol. 114, No. 1,
pp. 647-659, (2008). pp. 146-159, (1994).
[18] N. Shao, W. Salman, A. Gavriilidis, P. [27] D. Jacqmin, “Calculation of Two-Phase
Angeli, “CFD simulations of the effect of Navier–Stokes Flows Using Phase Field
inlet conditions on Taylor flow Modeling”, J. Comput. Phys., Vol. 155,
formation”, International Journal of Heat No. 1, pp. 96-127, (1999).
and Fluid Flow, Vol. 29, No. 6, pp. 1603- [28] L. D. Youngs, Time-dependent multi-
1611, (2008). material flow with large fluid distortion,
[19] P. Wei, K. Zhang, W. Gao, L. Kong, R. in Numerical Methods for Fluid
Field, “CFD modeling of hydrodynamic Dynamics, Academic Press, New York.,
characteristics of slug bubble flow in a flat pp. 273-285, (1982).
sheet membrane bioreactor”, Journal of [29] J. Li, “Calcul d’interface affin'e par
Membrane Science, Vol. 445, pp. 15-24, morceaux (piecewise linear interface
(2013). calculation)”, C.R. Acad. Sci. Paris S’erie
[20] R. Rahimi, E. Bahrarni far, M. Mazarei IIb: Paris, Vol. 320, pp. 391-396, (1995).
Sotoodeh, “The indication of two phase [30] T. H. Shih, W. W. Liou, A. Shabbir, Z.
flow pattern and slug characteristics in a Yang, J. Zhu, “A new k-epsilon eddy
pipeline using CFD method”, Gas viscosity model for high Reynolds
Processing Journal, Vol. 1, No. 1, pp. 70- number turbulent Flows - model
87, (2013). development and validation”, COMPUT
[21] J. Gregorc and I. Žun, “Inlet conditions FLUIDS, Vol. 24, No. 3, pp. 227-238,
effect on bubble to slug flow transition in (1995).
mini-channels”, Chem. Eng. Sci., Vol. [31] J. G. Collier, J. R. Thome, Convective
102, pp. 106-120, (2013). boiling and condensation, 3rd ed.,
[22] H. Pouraria, J. K. Seo, J. K. Paik, Clarendon Press, Oxford, pp. 48-68,
“Numerical modeling of two–phase oil- (1994).

340
JCARME Numerical investigation of . . . Vol. 9, No. 2

How to cite this paper:

M. Hassani, M. Bagheri Motlagh, R. Kouhikamali, “Numerical
investigation of upward air-water annular, slug and bubbly flow
regimes”, Journal of Computational and Applied Research in
Mechanical Engineering, Vol. 9, No. 2, pp. 331-341, (2019).

DOI: 10.22061/jcarme.2019.3893.1453

URL: http://jcarme.sru.ac.ir/?_action=showPDF&article=1023

341