You are on page 1of 22

Res Chem Intermed (2014) 40:1021–1042

DOI 10.1007/s11164-013-1018-2

A combined CFD modeling with population balance


equation to predict pressure drop in venturi scrubbers

Azam Sharifi • Ali Mohebbi

Received: 21 August 2012 / Accepted: 2 January 2013 / Published online: 16 January 2013
Ó Springer Science+Business Media Dordrecht 2013

Abstract A venturi scrubber is one of the most important devices for air pollution
control. Although there are different models for predicting the pressure drop in
venturi scrubbers, most of them have some defects and cannot predict the pressure
drop correctly. In this study, for the first time, an Eulerian–Eulerian computational
fluid dynamics (CFD) model is combined with a population balance equation to
predict the pressure drop in venturi scrubbers. This simulation takes into account a
multiple size group model for droplet dispersion and droplet size distribution, which
is based on a population balance equation. Flow field has been calculated by solving
the time averaged continuity and Navier–Stokes equations along with the standard
k–e turbulence model. The equations included drag, turbulent dispersion, and
buoyancy forces. The calculated pressure drop with and without considering the
population balance equation was compared with the experimental data to evaluate
the accuracy of the CFD modeling. The size distribution of droplets in the venturi
scrubber was studied at different points for different liquid to gas ratios and throat
gas velocities. The results show that the maximum break-up of droplets happens at
the liquid injection point. Finally, the effects of nozzle diameter and nozzle
arrangement on pressure drop in venturi scrubbers were investigated.

Keywords Pressure drop  Venturi scrubber  CFD  Multiple size group (MUSIG)
model  Eulerian–Eulerian method  Droplet size distribution

A. Sharifi  A. Mohebbi (&)


Department of Chemical Engineering, College of Engineering,
Shahid Bahonar University of Kerman, Kerman, Iran
e-mail: amohebbi2002@yahoo.com; amohebbi@mail.uk.ac.ir

A. Mohebbi
School of Chemical Engineering, The University of Adelaide, Adelaide, SA, Australia

123
1022 A. Sharifi, A. Mohebbi

List of symbols
Variables
Ap Area of the droplet projected in the flow direction (m2)
BB Birth rate of group i due to break-up of larger droplets (1/m3 s)
Bc Birth rate of group i due to coalescence of group j and group k droplets (1/m3 s)
L/G Liquid to gas ratio (m3/m3)
CD Drag coefficient
CTD Turbulent dispersion coefficient
Cl k–e turbulent model coefficient
C1e k–e turbulent model coefficient
C2e k–e turbulent model coefficient
DC Death rate of group i due to coalescence with other droplets (1/m3 s)
DB Death rate of group i droplets due to break-up into smaller droplets (1/m3 s)
Eij Mean rate of strain tensor
F Inter-phase momentum transfer (N/m3)
fi Volume fraction of droplets of group i
g Gravity acceleration (m/s2)
g Break-up frequency (s-1)
k Turbulent kinetic energy (m2/s2)
P Pressure (Pa)
Q Specific droplet coalescence rate (m3/s)
Si Source term of the rate of mass transfer in group i (kg/m2 s)
t Time (s)
Vgth Gas velocity in the throat (m/s)
di Droplet diameter of group i (m)
ds The Sature mean diameter (m)
Re Reynolds number
ni Number density of droplets of group i (m-3)
Urel Relative velocity between the phases (m/s)
U Velocity vector (m/s)
mi Volume of droplets of group i (m3)
Greek letters
a Volume fraction
e Turbulent energy dissipation rate (m2/s3)
lt Turbulent viscosity (Pa s)
q Density (kg/m3)
q0 Reference density (kg/m3)
mr Relative axial velocity between gas and liquid (m/s)
dk k–e turbulent model coefficient
de k–e turbulent model coefficient
Subscripts
d Dispersed phase
c Continuous phase
th Throat

123
CFD modeling with population balance equation 1023

Introduction

Venturi scrubbers are widely used for the simultaneous separation of fine particulate
matter and toxic gases from industrial gaseous effluents. They are usually compact
equipment of simple construction, and can handle both hot gases and sticky or
inflammable particles [1]. The venturi scrubber utilizes a moving gas stream to
atomize and accelerate the liquid droplets. The gas velocity is controlled by
adjusting the area of the venturi throat. This can be used to control performance
under varying gas flow rates by maintaining a constant pressure drop across the
venturi throat. Due to the absence of moving parts, scrubbers of this type may be
especially suitable for the collection of sticky particles. Liquid is most often
introduced as jets though orifices at the beginning of the throat. The high energy of
the gas promotes the atomization of the jets into a large number of small droplets
[2].
The collection of particulate matter occurs due to collisions between particles and
droplets, and is a complex function of scrubber geometry, gas and liquid flow rates,
jet penetration, dust size, droplet size distribution, droplet dispersion in the throat,
and the fraction of liquid flowing as a film on the walls of the equipment [3, 4].
Collection efficiency in a gas atomized venturi scrubber increases with pressure
drop. Pressure drops of 2500 water gauge or higher are utilized to collect sub-micron
particles.
Viswanathan et al. [5] developed an annular flow model for accurate prediction
of pressure drop in Pease-Anthony venturi scrubbers. This model, which considered
the primary design parameters of liquid to gas ratio, throat gas velocity, venturi
geometry, and liquid film flow rate, accurately predicted the measured pressure
gradients and overall energy losses. Their model did not provide a correlation to
estimate the fraction of water flowing on the wall, and they used their empirical data
to introduce this parameter in their model. Furthermore, Viswanathan [6]
investigated liquid film characteristics in the prediction of pressure drop in a
venturi scrubber by measuring the film flow rate, film thickness, and pressure
drop by varying the injection orifice diameter, the throat gas velocity, and the liquid
loading in a pilot-scale venturi.
Gonçalves et al. [7] evaluated the most important models for the prediction of
pressure drop in venturi scrubbers. The prediction of different models was compared
to experimental data from venturi scrubbers of different sizes, geometries, operating
variables, and liquid injection arrangements. They concluded that most of the
models must be used with caution and much attention must be paid to the validity of
the assumptions employed in the mathematical models.
Pak and Chang [8] developed a computational model for the interactive three-
phase flow in a circular venturi scrubber to estimate pressure drop and collection
efficiency. The Eulerian–Lagrangian method was used to solve the model
numerically. Their model under-predicted the pressure drop, while the prediction
of droplet size was inaccurate.
Nasseh et al. [9] applied artificial neural networks to predict pressure drop in
venturi scrubbers. They used three sets of experimental data to design three
independent ANNs. Nasseh et al. [10] also used some correlations for film flow rate,

123
1024 A. Sharifi, A. Mohebbi

core and film friction factors in the model of Viswanathan to get better results. In
their models the genetic algorithm was used to optimize the parameters of the ANNs
to improve the results.
Lu and Wang [11] established a 3D model of heat and mass transfer with phase
change in order to predict the pressure drop, collection efficiency, velocity, temperature,
and mole fraction of vapor in an industrial venturi scrubber with water spraying for
converter gas cooling. The coupled problem of heat, force, and mass transfers between
gas flow and liquid droplets was solved by a commercial computational fluid dynamics
(CFD) package, FLUENT. The numerical results showed that the water injections had
an important influence on the distributions of pressure, velocity, temperature, and mole
fraction of vapor, especially for the spraying region in the throat.
Ahmadvand and Talaie [12] developed a modified 2D mathematical model to
simulate droplet dispersion through a cylindrical venturi scrubber based on an
Eulerian approach. A Rosin–Rammler distribution function was used to take into
account the distribution of droplet size. Their results showed that the prediction
of droplet concentration distribution according to their model was in a better
agreement with experimental data than that predicted using the previous models,
which were based on a constant Peclet number.
Over recent years, as computing costs have decreased, CFD has led a shift of
scientific study and engineering design through experiments to numerical simula-
tions [13, 14]. In environmental engineering, CFD has been applied to evaluate the
indoor air quality and to simulate the pollution dispersion, and the diffusion of
automobile exhaust into tunnels [15–17].
Flow fields in chemical process unit operations are complex and often involve
multiphase flows. The analysis of multifluid or multiphase handling devices is not
easy. CFD has been identified as an available tool for the analysis of such devices
and associated flow fields. Multiphase flows refer to situations in which more than
one fluid is present and all fluids are immiscible. The term ‘phase’ is used in a much
wider sense and does not necessarily refer to thermodynamic phases but instead to
multiple fluids mixed at a macroscopic level and having a mixing scale larger than
the molecular scale. Multiphase flows can be classified broadly into continuous–
dispersed flows and continuous–continuous flows [18].
CFD models are based on the first principles of mass, momentum, and energy
conservation, as described by the Navier–Stokes equations [19]. CFD models for
multiphase flows can be divided into four categories: the Algebraic slip model, the
Drift flux model, Eulerian–Lagrangian models, and Eulerian–Eulerian models. The
conservation equations of mass, momentum, and energy are solved in each of these
methods. The presence of each phase is quantified by volume fraction, which
represents the fraction of volume occupied by that particular phase. Volume fraction
is obtained by solving individual continuity equations for each phase [20].
An Eulerian–Eulerian model is the most general model for multiphase flows. It is
based on the principle of interpenetrating continua. Each phase is governed by the
Navier–Stokes equations. It is also described by its distinctive physical properties
and has its own velocity, pressure, concentration, and temperature field. The
interphase transfer between phases is computed using empirical closure relation-
ships. The Eulerian–Eulerian model is applicable for continuous–dispersed and

123
CFD modeling with population balance equation 1025

continuous–continuous systems. For continuous–dispersed systems, the velocity of


each phase is computed using the Navier–Stokes equations. The dispersed phase can
be in the form of particles, drops, or bubbles. The forces acting on the dispersed
phase are modeled using empirical correlations and are included as part of the
interphase terms. Drag, lift, buoyancy, and virtual mass effects are some of the
forces that might be acting on the dispersed phase [18]. Although there are some
CFD models for venturi scrubbers,more physics of venturi scrubber needs to be
included in the modeling for predicting pressure drop.
In this study, for the first tim,e a 3D CFD modeling based on an Eulerian–
Eulerian two-fluid model was combined with a population balance model to predict
pressure drop in a rectangular, Pease-Anthony-type venturi scrubber. In the venturi
scrubber, the multiple size group (MUSIG) model was used for predicting the
droplet size distribution of the dispersed phase. The performance of the model was
compared with the experimental data. Moreover, the capability of the combined
model of CFD simulation with the population balance model for predicting the
droplet size distribution in venturi scrubbers was investigated.

Flow equations

The numerical simulations presented are based on the two-fluid model Eulerian–
Eulerian approach because it gives better opportunities to study phase interactions and
phase separation than the Lagrangian method. In a venturi scrubber, gas and liquid water
are continuous (c) and dispersed (d) phases, respectively. In this study, the flow has been
considered turbulent, incompressible, isothermal, and steady-state. The time averaged
continuity and momentum equations for each phase are written as follows [21]:
Continuity equation of the continuous phase:
r  ð qc a c U c Þ ¼ 0 ð1Þ
Continuity equation of the dispersed phase:
r  ðqd ad Ud Þ ¼ 0 ð2Þ
where q, a and U represent the density, volume fraction, and velocity vector,
respectively.
Momentum equation:
½r  ðqk ak Uk U k Þ ¼ ak rP þ ak qk g þ ðlk þ lt Þr2 ðak U k Þ þ Fkm ðk; m ¼ c; dÞ
ð3Þ
where l, lt, and P are dynamic viscosity, turbulent viscosity, and pressure,
respectively. The terms on the right hand side of Eq. 3 consider all the forces acting
on phase k in each control volume: the pressure gradient, the gravity, the viscous
stress term, and the interphase transfer of momentum between phases.
The vector term Fkm accounts for all forces considering the interphase
momentum transfer. This term includes the drag, the buoyancy, the lift, the virtual
mass, and the dispersion forces. Only the effects of drag, buoyancy, and dispersion
forces have been considered in the present simulation.

123
1026 A. Sharifi, A. Mohebbi

The interphase momentum transfer between gas and liquid due to drag force is
given by [21]:
D 3 1
Flg ¼ CD ql ag jU rel jU rel ð4Þ
4 dS
where Urel is the relative velocity between the phases, dS is the Sauter mean
diameter, and CD is the drag coefficient. In this study, the Grace drag coefficient was
used.
The local Sauter mean diameter (dS), based on the calculated values of the scalar
fraction fi and the droplet diameter of the MUSIG-size group-i, di, can be calculated
as follows [21]:
1
dS ¼ P fi ð5Þ
i di

The droplets are dispersed due to turbulent flow. The turbulent dispersion force is
a function of the magnitude of turbulence in the continuous phase and the gradient
of the volume fraction [22]:
FcT ¼ CTD qc kc rac ð6Þ
where CTD is the turbulent dispersion coefficient, with the value CTD = 0.1 as
recommended in the ANSYS CFX code.
For calculating the turbulent viscosity, lt in Eq. 3, a k–e model with the standard
values of the constants according to the following equation is applied [23]:
k2
lt ¼ qCl ð7Þ
e
where k and e are the turbulent kinetic energy and the turbulent dissipation rate,
respectively. The transport equations for k and e are as follows [21]:
     
oðakÞ oðakÞ oðakÞ lt
q u þv þw ¼r a lþ rk þ 2alt Eij Eij  aqe ð8Þ
ox oy oz dk
     
oðaeÞ oðaeÞ oðaeÞ l e
q u þv þw ¼ r  a l þ t re þ C1e 2alt Eij Eij
ox oy oz de k
e2
 C2e aq ð9Þ
k
the mean rate of strain tensor Eij is:
 
1 oui ouj
Eij ¼ þ ð10Þ
2 oxj oxi
Cl, C1e, C2e, dk and de are the standard k–e turbulent model coefficients. Their
respective values are: 0.09, 1.44, 1.92, 1.0, and 1.3.
The buoyancy force due to the density difference between the continuous and
dispersed phases can be defined for the dispersed phase as follows [21]:

123
CFD modeling with population balance equation 1027

Fdbuo ¼ ad ðqd  q0 Þg ð11Þ


where q0 is a reference density and g is the acceleration of gravity vector.

Multiple size group model (MUSIG)

Droplets in the dispersed phase can be distributed amongst several size groups. Each
size group may be treated as a separated phase in the multiphase flow calculations.
One of the attributes of polydispersed multiphase flow is that the different sizes of
the dispersed phase interact with each other through the mechanisms of break-up
and coalescence. To deal with this type of flow, a population balance equation must
be formulated. In this work, the implementation of population balance equations
into ANSYS CFX 11 developed by Lo [24], namely MUSIG, is used to take into
account the non-uniform droplet size distribution. The MUSIG model is carried out
with an Eulerian–Eulerian two-phase model and assumes all droplets of the same
group travelling with the same velocity and having the same properties; therefore, it
is only necessary to solve one set of momentum equations for the dispersed phase.
The following continuity equation of the droplet size groups is solved to represent
the i-th size distribution [25]:
r  ðqd ad fi U d Þ ¼ Si ð12Þ
where Si is the source term of the rate of mass transfer in the MUSIG-size group
i due to the breakup and coalescence process, and fi is the fraction of dispersed
phase volume fraction in the MUSIG-size group i.
The droplet number density of the i-th (ni) and the droplet volume of i-th (vi) are
related to the dispersed phase volume fraction by [25]:
ni v i ¼ ad f i ð13Þ
The general form of the population balance equation is:
r  ðU d  ni Þ ¼ r  ðad fi U d Þ ¼ BB  DB þ Bc  DC ð14Þ
The break-up model is taken from Lou and Svendsen [26] for the prediction of droplet
break-up in turbulent dispersions and assumes isotropic turbulence. The birth rate of the
MUSIG-size group i due to break-up of larger droplets can be given by [21]:
X
N  
BB ¼ g v j : v i nj ð15Þ
j¼iþ1

the function of g represents the droplet break-up kernel. The death rate of the
MUSIG-size group i droplet due to break-up into smaller droplets can be given by
[21]:
D B ¼ gi ni ð16Þ
According to the Prince and Blanch [27] model, the coalescence of two droplets
occurs in three steps: at first, the droplets collide trapping a small amount of liquid
between them; then, this liquid film drains until it reaches a critical thickness; and as
a last step, this film disappears and the droplets join together.

123
1028 A. Sharifi, A. Mohebbi

The birth rate of the MUSIG-size group-i droplet due to the coalescence of group
j and group-k droplets can be given by [21]:
1X i X i
Bc ¼ Qjk nj nk ð17Þ
2 j¼1 k¼1

where the function of Q represents the specific droplet coalescence rate. The death
rate of the MUSIG-size group-i droplet due to coalescence with other droplets can
be given by [21]:
X
N
D C ¼ ni Qij nj ð18Þ
j¼1

In this study, both phases travel at the same averaged velocity, thereby the
aerodynamic fragmentation does not happen. Hence, the only process considered
is the turbulent break-up. In the same way, coalescence processes are also not
considered.

Venturi scrubber geometry

Figure 1 shows the venturi scrubber geometry [28] which has been used in this
study. The water was injected into the scrubber through 34 orifices (17 of them on
each side) perpendicularly oriented with respect to the air stream. The 17 orifices
were arranged along two rows with center-to-center spacing of 13 mm. The first row
of openings, which was 2.54 cm below the entrance to the throat section, had 9
orifices, while the bottom row had 8. The holes of the lower row were staggered
midway between those of the top row to improve the distribution of water. The
diameter of the orifices was 2.108 mm. In the venturi scrubber, the gas phase is
continuous and the liquid phase is dispersed. In Fig. 1, the y axis is the flow
direction, z is vertical to the flow direction in a horizontal plane, and x is the vertical
axis.

Numerical solution

The venturi scrubber process was studied in steady-state. The mass and momentum
equations were solved by using the commercial CFD package ANSYS CFX-11. The
CFX code uses the finite volume method (FVM) to solve generalized unstructured
meshes in Cartesian coordinates. The discrete system of linearized equations is
solved using the algebraic multigrid (AMG) method accelerated by the incomplete
lower upper (ILU) factorization technique. A Rhie and Chow [29] formulation is
used for the pressure–velocity coupling in a single cell of the collocated grid. This
solution approach uses a fully implicit discretization of the equations based on the
false time step technique. The UPWIND differencing and the High Resolution
schemes were used for the volume fraction and the other terms, respectively.

123
CFD modeling with population balance equation 1029

Fig. 1 The venturi scrubber


geometry [28]

(dimensions are in cm)

The mass flow, static pressure, and no-slip boundary conditions at the inlet, the
outlet, and walls were applied, respectively.
For testing the grid independency of the solutions, three different 3D meshes,
which contained 200,000, 300,000, and 400,000 cells were used, respectively.
Figure 2 shows a comparison of axial pressure drop for liquid-to-gas flow rate ratio
of 0.4 m3 liquid/1,000 m3 gas and throat gas velocity of 45.7 m/s with different
cells. As one can see in this figure, increasing the number of cells did not show
important changes in the results. Therefore, the 300,000 cells were used as a grid-
independent case for the computations. Moreover, these selected cells were
confirmed by the comparison between simulation results and experimental data.
Some parameters must be chosen before calculation in the MUSIG model. After
running the code several times, it was found that the number of the droplet size
groups, the minimum droplet diameter, the maximum droplet diameter, and break-
up calibration for the MUSIG model are 10, 50 lm, 500 lm and 3, respectively.
The percentage of breakdown distribution of droplet size at the inlet for two
different liquid-to-gas flow rate ratios and three throat gas velocities is presented in
Tables 1 and 2 [30]. Figure 3 shows the droplet size distribution in the inlet of the
venturi scrubber for Vgth = 45.7 m/s, L/G = 0.0015 m3/m3.

123
1030 A. Sharifi, A. Mohebbi

400000 300000 200000


2
1.8

Pressure drop (kPa)


1.6
1.4
1.2
1
0.8
0.6
0.4
0.2
0
0 0.5 1 1.5 2 2.5
Distance along venturi scrubber (m)

Fig. 2 The comparison of modeling results with different cells for L/G = 0.0004 m3/m3 and
Vgth = 61 m/s

Table 1 Breakdown percentage for size distribution of droplets at inlet for L/G = 1.5 m3/1,000 m3 and
Vgth = 61, 45.7, and 76.2 m/s [30]
Number of size Size group Percentage Number of size Size group Percentage
group (lm) breakdown (%) groups (lm) breakdown (%)

Vgth = 61 m/s
1 50 0.092 6 300 0
2 100 0.22 7 350 0
3 150 0.29 8 400 0
4 200 0.195 9 450 0
5 250 0.11 10 500 0
Vgth = 45.7 m/s
1 50 0.118 6 300 0.074
2 100 0.218 7 350 0
3 150 0.25 8 400 0
4 200 0.2 9 450 0
5 250 0.143 10 500 0
Vgth = 76.2 m/s
1 50 0.09 6 300 0.048
2 100 0.254 7 350 0
3 150 0.3 8 400 0
4 200 0.21 9 450 0
5 250 0.098 10 500 0

Results and discussion

Single-phase air

First, the model was validated for dry air flow through the venturi scrubber with
axial pressure drop data for three different gas velocities. The results are shown in
Fig. 4. It can be seen that the model agreement with the experimental data [31]

123
CFD modeling with population balance equation 1031

Table 2 Breakdown percentage for size distribution of droplets at inlet for L/G = 0.4 m3/1,000 m3 and
Vgth = 61, 45.7, and 76.2 m/s [30]
Number of size Size group Percentage Number of size Size group Percentage
group (lm) breakdown (%) groups (lm) breakdown (%)

Vgth = 61 m/s
1 50 0.201 6 300 0
2 100 0.3 7 350 0
3 150 0.35 8 400 0
4 200 0.23 9 450 0
5 250 0.13 10 500 0
Vgth = 45.7 m/s
1 50 0.16 6 300 0.123
2 100 0.23 7 350 0
3 150 0.25 8 400 0
4 200 0.21 9 450 0
5 250 0.17 10 500 0
Vgth = 76.2 m/s
1 50 0.09 6 300 0
2 100 0.3 7 350 0
3 150 0.35 8 400 0
4 200 0.195 9 450 0
5 250 0.065 10 500 0

30

25 24
20.8 20.4
20
Frequency

15 13.3
10.8
10
7

5 2.4
1 0.3
0
50 100 150 200 250 300 350 400 450 500
Droplet size (µm)

Fig. 3 Droplet size distribution histogram in the inlet of the venturi scrubber (Vgth = 45.7 m/s,
L/G = 0.0015 m3/m3)

increases when the gas velocity at the throat increases and also the pressure drop
increases due to the friction gas-walls.

Air–water flows

In Figs. 5, 6, 7, 8, 9, and 10, the pressure drop along the venturi scrubber obtained
with the proposed model is compared with the experimental data of Viswanathan

123
1032 A. Sharifi, A. Mohebbi

et al. [5] for a pilot Pease-Anthony venturi scrubber with a rectangular cross-section
for two different liquid-to-gas flow rate ratios (i.e. L/G = 0.4, and 1.5 m3 liquid/
1,000 m3 gas) and three throat gas velocities (i.e. Vgth = 45.7, 61, and 76.2 m/s).
As one can see from these figures, there is a good agreement between the results of
combined CFD model with the population balance equation and the experimental
data. In these figures, the mean droplet diameter for CFD work without population
balance was calculated by Boll’s equation [32]. The comparison between the dry

This work Experimental data [31] (Vgth= 61 m/s)


Experimental data [31] (Vgth= 45.7 m/s) Experimental data [31] (Vgth= 76.2 m/s)
6

5 L/G=0.0004 m3/m3
Pressure drop (kPa)

0
0 0.5 1 1.5 2 2.5
Distance along venturi scrubber (m)

Fig. 4 The comparison of predicted and experimental axial pressure drop [31] for only gas flow

Experimental data [5]


CFD work without population balance
CFD work with population balance
4
3.5
Pressure drop (KPa)

3 Vgth = 61 m/s

2.5
2
1.5
1
0.5
0
0 0.5 1 1.5 2 2.5
Distance along venturi scrubber (m)

Fig. 5 The comparison of modeling results of this work with the experimental data reported by
Viswanathan et al. [5] for L/G = 0.0004 m3/m3 and Vgth = 61 m/s

123
CFD modeling with population balance equation 1033

Experimental data [5]


CFD work without population balance
CFD work with population balance
6

5 Vgth= 76.2 m/s


Pressure drop (kPa)

0
0 0.5 1 1.5 2 2.5
Distance along venturi scrubber (m)

Fig. 6 The comparison of modeling results of this work with the experimental data reported by
Viswanathan et al. [5] for L/G = 0.0004 m3/m3 and Vgth = 76.2 m/s

Experimental data [5]


CFD work without population balance
CFD work with population balance
3

2.5
Vgth= 45.7 m/s
Pressure drop (kPa)

1.5

0.5

0
0 0.5 1 1.5 2 2.5
Distance along venturi scrubber (m)

Fig. 7 The comparison of modeling results of this work with the experimental data reported by
Viswanathan et al. [5] for L/G = 0.0004 m3/m3 and Vgth = 45.7 m/s

pressure drop and the wet pressure drop shows that the pressure drop increases when
the liquid droplets enter the flow. At each fixed throat gas velocity, the pressure drop
rises exponentially up to the throat where the gradient becomes essentially linear.

123
1034 A. Sharifi, A. Mohebbi

Experimental data [5]


CFD work without population balance
CFD work with population balance
8
7
Pressure drop (kPa)

6 Vgth= 61 m/s

5
4
3
2
1
0
0 0.5 1 1.5 2 2.5
Distance along venturi scrubber (m)

Fig. 8 The comparison of modeling results of this work with the experimental data reported by
Viswanathan et al. [5] for L/G = 0.0015 m3/m3 and Vgth = 61 m/s

Experimental data [5]


CFD without population balance
CFD work with population balance
12

10 Vgth= 76.2 m/s


Pressure drop (KPa)

0
0 0.5 1 1.5 2 2.5
Distance along venturi scrubber (m)

Fig. 9 The comparison of modeling results of this work with the experimental data reported by
Viswanathan et al. [5] for L/G = 0.0015 m3/m3 and Vgth = 76.2 m/s

Figure 11 shows the velocity vectors inside the venturi scrubber. This figure
indicates that the phases flow towards the central section after entering the venturi
scrubber, which causes the velocity to increase in this region.

123
CFD modeling with population balance equation 1035

Experimental data [5]


CFD work without population balance
CFD work with population balance
4.5
4
Vgth= 45.7 m/s
Pressure drop (KPa)

3.5
3
2.5
2
1.5
1
0.5
0
0 0.5 1 1.5 2 2.5
Distance along venturi scrubber (m)

Fig. 10 The comparison of modeling results of this work with the experimental data reported by
Viswanathan et al. [5] for L/G = 0.0015 m3/m3 and Vgth = 45.7 m/s

Fig. 11 The velocity vectors in the venturi scrubber for L/G = 0.0004 m3/m3 and Vgth = 61 m/s

The contours of droplet mean diameter for two different liquid-to-gas ratios and
Vgth = 61 m/s are shown in Figs. 12 and 13. As one can see from these figures, the
maximum break-up of droplets occurs at the throat in which the liquid is injected.

123
1036 A. Sharifi, A. Mohebbi

Fig. 12 The contours of droplet mean diameter for L/G = 0.0004 m3/m3 and Vgth = 61 m/s

Fig. 13 The contours of droplet mean diameter for L/G = 0.0015 m3/m3 and Vgth = 61 m/s

123
CFD modeling with population balance equation 1037

The distribution of droplet size in the venturi scrubber was evaluated at different
points. The coordinates of the mentioned points are shown in Table 3. The locations
of the selected three points are distributed at the throat after the liquid injection
point (i.e. point 1), at the entrance of diffuser (i.e. point 2), and the end of the
scrubber (i.e. point 3). The changes in droplet size distribution along the venturi
scrubber are shown in Figs. 14 and 15 for two different liquid-to-gas ratios and
throat gas velocities. In this figure, the model results at two different positions (i.e.
points 1 and 3) were superimposed. As the liquid droplet moves from the points 1–3,
a movement to the left in the droplet size distribution can be observed, indicating a
decrease in the size of the droplets. This behavior is in accordance with our model,
because in the model only the droplet break-up was considered. However, the break-
up percentage of size groups for L/G = 0.0015 m3/m3 is not as high as the
percentage for L/G = 0.0004 m3/m3. These results confirm that the droplet break-
up is the controlling parameter at the scrubber inlet.
Moreover, the effect of throat gas velocity on mean droplet size can be observed
in Figs. 16 and 17. As one can see from these figures, the mean droplet size
decreases with increasing the throat gas velocity and this size decreases along
the venturi scrubber for low L/G. On the other hand, at high L/G value (i.e.
L/G = 0.0015 m3/m3), there are no significant changes in droplet size along the

Table 3 Coordinate of points


Points Coordinate (x, y, z), (m)
for analyzing droplet size
distribution in different locations
1 -0.00253, -0.83, 0.122
in the venturi scrubber
2 -0.00253, -1, 0.122
3 -0.01, -1.95, 0.1

30
Vgth= 45.7 m/s
25 L/G = 0.0004 m3/m3

20
Frequency

15 Point 1
Point 3
10

0
50 100 150 200 250 300 350 400 450 500
Diameter (µm)

Fig. 14 Variation in droplet size distribution along the venturi scrubber (L/G = 0.0004 m3/m3 and
Vgth = 45.7 m/s)

123
1038 A. Sharifi, A. Mohebbi

venturi. In most cases, an increase in the throat gas velocity caused a decrease in
droplet size, because increasing the gas velocity causes an increase in the total
surface area required for a finer spray. This is in agreement with both the Nukiyama
and Tanasawa [33] and Boll et al. [34] experiments and correlations.
Pease-Anthony-type venturi scrubbers that employ a system of nozzles to inject
liquid into the gas stream are common in industry. The cleaning efficiency in such
scrubbers is sensitive to many parameters, such as throat gas velocity, liquid-to-gas
ratio, nozzle diameter, nozzle type, and the nozzle arrangement. In this work,
the effects of nozzle diameter and nozzle arrangement on pressure drop are
investigated.

35
Vgth= 61 m/s
30
L/G = 0.0015 m3/m3
25
Frequency

20
Point 1
15
Point 3
10

0
50 100 150 200 250 300 350 400 450 500
Diameter (µm)

Fig. 15 Variation in droplet size distribution along the venturi scrubber (L/G = 0.0015 m3/m3 and
Vgth = 61 m/s)

190
Mean droplet diameter (µm)

170

150

130

110 point 1
point 2
90 L/G = 0.0004 m3/m3
Point 3
70

50
20 30 40 50 60 70 80
Throat gas velocity (m/s)

Fig. 16 Variation in mean droplet diameter with throat gas velocity for points 1, 2, and 3 for
L/G = 0.0004 m3/m3

123
CFD modeling with population balance equation 1039

Effect of nozzle diameter and nozzle arrangement

Typically, nozzles are located on the longest side of the throat (i.e. z direction).
Three configurations that may be encountered in industrial applications were chosen
for this study. These were: (1) a single line in which the nozzles are arranged in one
straight line; (2) a triangular configuration in which the nozzles in the second row
are at the midpoint of two nozzles in the first row and vice versa that the nozzles are
of different diameters; the top row of nozzles is 1 mm and the bottom row of
nozzles is 1.5 mm; and (3) a triple line in which the nozzles are arranged in three

150
Mean droplet diameter (µm)

130

110

90 Point 1
L/G = 0.0015 m3/m3
Point 2
70
Point 3

50
20 30 40 50 60 70 80
Throat gas velocity (m/s)

Fig. 17 Variation in mean droplet diameter with throat gas velocity for points 1, 2, and 3 for
L/G = 0.0015 m3/m3

1 row of nozzle 2 rows of nozzle


3.5

3
Vgth=61 m/s
Pressure drop (kPa)

2.5

1.5

0.5

0
0 0.5 1 1.5 2 2.5
Distance along venturi scrubber (m)

Fig. 18 The effect of one straight line nozzle configuration for L/G = 0.0004 m3/m3 and Vgth = 61 m/s

123
1040 A. Sharifi, A. Mohebbi

3 rows of nozzle 2 rows of nozzle


4
3.5 Vgth= 61 m/s
Pressure drop (kPa)

3
2.5
2
1.5
1
0.5
0
0 0.5 1 1.5 2 2.5
Distance along venturi scrubber (m)

Fig. 19 The effect of three straight lines nozzle configuration for L/G = 0.0004 m3/m3 and
Vgth = 61 m/s

different diameters for nozzle (first row, d= 1 mm; second row, d= 1.5 mm)
same diameters for nozzle (d= 2.108 mm)
3.5

3 Vgth=61 m/s
Pressure drop (kPa)

2.5

1.5

0.5

0
0 0.5 1 1.5 2 2.5
Distance along venturi scrubber (m)

Fig. 20 The effect of different nozzle diameters and configuration for L/G = 0.0004 m3/m3 and
Vgth = 61 m/s

straight lines. The results for L/G = 0.0004 m3/m3 and Vgth = 61 m/s are shown in
Figs. 18, 19, and 20. As can be seen from these figures, no significant changes in the
pressure drop are observed between the single, triangular, and triple arrangements.
Figure 20 also shows that there is no difference in pressure drop between the nozzle
diameters of 1, 1.5, and 2.108 mm. This finding shows that the amount of injected
liquid has more influence on the pressure drop in venturi scrubbers than nozzle
diameter and nozzle arrangement.

123
CFD modeling with population balance equation 1041

Conclusions

In this study, a 3D CFDs simulation based on a population balance model has been
applied to predict the pressure drop in venturi scrubbers. The simulation results
were validated by the experimental data. A CFD modeling was combined with a
population balance model to study complex flow fields in a venturi scrubber.
Successful simulation of a complex multiphase flow requires full understanding of
the underlying models. The results of the simulation showed that droplet size
distribution and pressure drop can be well predicted along the venturi scrubber for
the air–water system. The droplet size distribution changed along the scrubber and
there were smaller sizes in the exit. Furthermore, the effects of nozzle diameter and
nozzle arrangement on pressure drop were investigated. The results showed that, for
different configurations and diameters of nozzles, the pressure drop in the venturi
scrubber is identical. Finally, from the results of this work, it can be concluded that
CFDs is a promising tool for monitoring the droplet size alteration along venturi
scrubbers and that these findings can help industries in the simulation and
optimization of scrubbers. This simulation can also be applied for predicting
collection efficiency in venturi scrubbers.

References

1. S. Calvert, How to choose a particulate scrubber. Chem. Eng. 19, 54–68 (1977)
2. Technical Manual TM 5-815-1/AFR 19-6, Air pollution control systems for boilers and incinerators
(Headquarters Department of the Army and the Air Force, Washington, D.C., 9 May 1988)
3. M. Taheri, C.M. Sheih, Mathematical modeling of atomizing scrubbers. AIChE J. 21, 153–157
(1975)
4. S. Viswanathan, Modeling of venturi scrubber performance. Ind. Eng. Chem. Res. 36, 4308–4317
(1977)
5. S. Viswanathan, A.W. Gynp, C.C. St Pierre, Annular flow pressure drop model for Pease-Anthony-
type venturi scrubbers. AIChE J. 31, 1947–1958 (1985)
6. S. Viswanathan, Development of a pressure drop model for a variable throat venturi scrubber.
J. Chem. Eng. 71, 153–160 (1998)
7. J.A.S. Gonçalves, D. Fernández Alonso, M.A. Martins Costa, B.J. Azzopardi, J.R. Coury, Evaluation
of the models available for the prediction of pressure drop in venturi scrubbers. J. Hazard. Mater. 81,
123–140 (2001)
8. S.I. Pak, K.S. Chang, Performance estimation of a venturi scrubber using a computational model for
capturing dust particles with liquid spray. J. Hazard. Mater. 138, 560–573 (2006)
9. S. Nasseh, A. Mohebbi, Z. Jeirani, A. Sarrafi, Predicting pressure drop in venturi scrubbers with
artificial neural networks. J. Hazard. Mater. 143, 144–149 (2007)
10. S. Nasseh, A. Mohebbi, A. Sarrafi, M. Taheri, Estimation of pressure drop in venturi scrubbers based
on annular two-phase flow model, artificial neural networks and genetic algorithm. J. Chem. Eng.
150, 131–138 (2009)
11. T. Lu, K. Wang, Numerical simulation of three-dimensional heat and mass transfer in spray cooling
of converter gas in a venturi scrubber. Chin. J. Mech. Eng. 22, 745–754 (2009)
12. F. Ahmadvand, M.R. Talaie, CFD modeling of droplet dispersion in a venturi scrubber. J. Chem.
Eng. 160, 423–431 (2010)
13. S. Bernardo, M. Mori, A.P. Peres, R.P. Dionı́sio, 3D computational fluid dynamics for gas and gas-
particle flows in a cyclone with different inlet section angles. Powder Technol. 162, 190–200 (2006)
14. K. Pant, C.T. Crow, P. Irving, On the design of miniature cyclones for the collection of bioaerosols.
Powder Technol. 125, 260–265 (2002)

123
1042 A. Sharifi, A. Mohebbi

15. J.M. Desantes, X. Margot, A. Gil, E. Fuentes, Computational study on the deposition of ultra-fine
particles from diesel exhaust aerosol. Aerosol Sci. 37, 1750–1769 (2006)
16. X. Wang, K.F. Mcnamara, Evaluation of CFD simulation using RANS turbulence models for
building effects on pollutant dispersion. Environ. Fluid Mech. 6, 181–202 (2006)
17. C.H. Chang, R.N. Meroney, Numerical and physical modeling of bluff body flow and dispersion in
urban street canyons. J. Wind Eng. Ind. Aerodyn. 89, 1325–1334 (2001)
18. H. S. Pordal, Practicing the science of computational fluid dynamics, Stress Engineering Services,
Inc. (2006)
19. S.V. Patankar, Numerical Heat Transfer and Fluid Flow (McGraw-Hill, New York, 1980)
20. R.H. Perry, D. Green, Chemical Engineers’ Handbook (McGraw-Hill, New York, 1984)
21. ANSYS CFX 11 User’ Guide, ANSYS, Ltd., Copyright 2009
22. L.S. Soo, Multiphase Fluid Dynamics (Science Press, Hong Kong, 1999)
23. B.E. Lunder, D.B. Spalding, The numerical computation of turbulent flows. Comput. Methods Appl.
Mech. Eng. 3, 269 (1974)
24. S. Lo, Application of Population Balance to CFD Modeling of Bubbly Flow via the MUSIG Model
(AEA Technology, Lockwood, 1996)
25. R. Sadeghi, A. Mohebbi, M. Baniasadi, CFD modeling of the launder of settler of an industrial
copper solvent extraction plant: a case study on Sarcheshmeh copper complex, Iran. Int. J. Miner.
Process. 98, 55–65 (2011)
26. H. Luo, H. Svendson, Theoretical model for drop and bubble break-up in turbulent dispersions.
AIChE J. 42, 1225–1233 (1996)
27. M.J. Prince, H.W. Blanch, Bubble coalescence and break-up in air-sparged bubble columns. AIChE
J. 36, 1485–1499 (1999)
28. S. Viswanathan, Examination of liquid film characteristics in the prediction of pressure drop in a
venturi scrubber. Chem. Eng. Sci. 53, 3161–3175 (1998)
29. C.M. Rhie, W.L. Chow, Numerical study of the turbulent flow on airfoil with trailing eddy sepa-
ration. AIAA J. 21, 1525–1532 (1983)
30. S. Viswanathan, D.S. Lim, M.B. Ray, Measurement of drop size and distribution in an annular
two-phase, two-component flow occurring in a venturi scrubber. Ind. Eng. Chem. Res. 44, 7458
(2005)
31. N.V. Ananthanarayanan, S. Viswanathan, Effect of nozzle arrangement on venturi scrubber perfor-
mance. Ind. Eng. Chem. Res. 38, 4889–4900 (1999)
32. R.H. Boll, Particle collection and pressure drop in venturi scrubbers. Ind. Eng. Chem. Fundam. 12,
40–50 (1973)
33. S. Nukiyama, Y. Tanasawa, Experiment on atomization of liquid by means of air stream. Trans. Soc.
Mech. Eng. Jpn. 4, S13–S14 (1938)
34. R.H. Boll, L.R. Flais, P.W. Maurer, W.L. Thompson, Mean drop size in a full scale venturi scrubber
via transmissometer. J. Air Pollut. Control Assoc. 24, 934–938 (1974)

123

You might also like