Faculty of Chemical and Natural Resources Engineering, University College of Engineering and
Technology Malaysia, KUKTEM, MEC Town, 25200 Kuantan, Pahang D. M., Malaysia.
Department of Chemical and Environmental Engineering, Faculty of Engineering, Universiti
Putra Malaysia, Selangor D. E., Malaysia.
10
15
20
This work presents a Computational Fluid Dynamics calculation to predict and to evaluate the effects of temperature, operating pressure and inlet velocity on the collection efficiency of gas
cyclones. The numerical solutions were carried out using spreadsheet and commercial CFD code FLUENT 6.0. This paper also
reviews four empirical models for the prediction of cyclone collection efficiency, namely Lapple [1], Koch and Licht [2], Li and
Wang [3], and Iozia and Leith [4]. All the predictions proved to be
satisfactory when compared with the presented experimental data.
The CFD simulations predict the cyclone cut-off size for all operating conditions with a deviation of 3.7% from the experimental
data. Specifically, results obtained from the computer modelling
exercise have demonstrated that CFD model is the best method of
modelling the cyclones collection efficiency.
Keywords
25
30
35
1. INTRODUCTION
Cyclones are devices that employ a centrifugal force generated by a spinning gas stream to separate particles from the
carrier gas. Their simple design, low capital cost and nearly
maintenance-free operation make them ideal for use as precleaners for more expensive final control devices such as baghouses or electrostatic precipitators. Cyclones are particularly
well suited for high temperature and pressure conditions because of their rugged design and flexible component materials.
Cyclone collection efficiencies can reach 99% for particles bigger than 5 m [5], and can be operated at very high dust loading.
Cyclones are used for the removal of large particles for both air
pollution control and process use. Application in extreme con-
40
45
50
55
60
65
70
162
J. GIMBUN ET AL.
75
80
85
90
95
calculations (with reference to the conservation of mass, momentum and energy equations). It is ideally suited for incompressible to mildly compressible flows. The conservation of
mass, momentum and energy in a fluid flow are expressed in
terms of non-linear partial differential equations that generally
defy solution by analytical means. The solution of these equations has been made possible by the advent of powerful workstations, opening avenues towards the calculation of complicated
flow fields with relative ease.
For the turbulent flow in a cyclone the key to the success of
CFD lies with the accurate description of the turbulent behavior
of the flow [6]. To model the swirling turbulent flow in a cyclone
separator, there are a number of turbulence models available in
FLUENT. These range from the standard k- model to the more
complicated Reynolds stress model (RSM). The comparison of
the different RANS-based turbulence models available in FLUENT 6.0 is presented in Table 2. The k- model involves the solution of transport equations for the kinetic energy of turbulence
and its dissipation rate and the calculation of a turbulent contribution to the viscosity at each computational cell. The standard
k-, RNG k- and Realizable k- models were not optimized for
the strongly swirling flows typically found in cyclones [8, 9].
Turbulence may be stabilized or destabilized in the parts of flow
domain where strong streamline curvature is present. However,
to reduce the computational effort, the RNG k- model can be
used with about 12% deviation on experimental data [6]. The
numerical studies carried out by Fredriksson [10] reveal that the
TABLE 1
Cyclone geometry used in this simulation
Geometry
Stairmand High
0.5 0.2
0.5
0.5 1.5
4
0.375
Efficiency
Kim and Lee
0.33 0.225 0.257 1.157 1.447 3.05 0.482
(1990) cyclone I
Bohnet (1995)
0.533 0.133 0.333 0.733 0.693 2.58 0.333
RNG k- model underestimates the variation of the axial velocity profile across the radial direction and also overestimates the
magnitude of the tangential velocity and the cyclone pressure
drop.
The Reynolds stress model requires the solution of transport
equations for each of the Reynolds stress components as well
as for dissipation transport without the necessity to calculate an
isotropic turbulent viscosity field. The Reynolds Stress turbulence model yields an accurate prediction of swirl flow pattern,
axial velocity, tangential velocity and pressure drop on cyclone
simulations [810].
The finite volume method has been used to discretize the
partial differential equations of the model using the SIMPLE
method for pressure-velocity coupling and the Second Order
Upwind scheme to interpolate the variables on the surface of
the control volume. The segregated solution algorithm was selected. The Reynolds stress (RSM) turbulence model was used
in this model due to the anisotropic nature of the turbulence in
cyclones. Standard Fluent wall functions were applied and high
order discretization schemes were also used.
Under the RSM second order upwind for discretization there
is a difficulty to reach the convergence in simulation [11].
The residuals may exhibit cyclic tendencies, which means that
the transient pattern occurs. In this instance, the solver must
be changed to a transient solver and this makes the time step
something in the region of 0.025 seconds or a tiny fraction of
the residence time of the cyclone. The simulation is then solved
with a coupling of unsteady and steady state solvers in FLUENT.
For the simulation using RNG k- model the steady state solver
is sufficient to reach convergence.
To calculate the trajectories of particles in the flow, the discrete phase model (DPM) was used to track individual particles
through the continuum fluid. The particle loading in a cyclone
separator is typically small (35%), and therefore it can be safely
assumed that the presence of the particles does not affect the
flow field (one-way coupling). The equation of motion for an
individual particle can be written as Crowe et al., [12]
dv
f
= (u v) + g
dt
v
100
105
110
115
120
125
130
[1]
where the other contributions to the force on the particle (buoy- 135
ancy, virtual mass and Basset term) are negligible because of
the small fluid-to-particle density ratio. The response time of
the particle, v is defined in terms of the particle density, particle
diameter and the viscosity of the air as:
v =
p d 2p
18
[2]
C D Rer
24
140
[3]
163
TABLE 2
Comparison of the different turbulence models in FLUENT 6.0
Model
Standard k-
RNG k-
Realizable k-
RSM
Strength
Weaknesses
with
Pentium IV 2.8 GHz HP workstation XW8000 with 512 cachememory, 1 GB RAM-memory, and 110 GB hard-disc memory. 160
Rer =
g d p |u v|
g
[4]
where Rer is the relative Reynolds number and C D is the drag coefficient. In FLUENT, the drag coefficient for spherical particles
is calculated by using the correlations developed by Morsi and
145 Alexander [13]. For non-spherical particles, the correlation was
developed by Haider and Levenspiel [14]. The ordinary differential equation (Eq. (1)) was integrated along the trajectory of an
individual particle. Collection efficiency statistics were obtained
by releasing a specified number of monodispersed particles at
150 the inlet of the cyclone and by monitoring the number escaping through the underflow. Collisions between particles and the
walls of the cyclone were assumed to be perfectly elastic (coefficient of restitution is equal to 1).
The numerical calculation was made with a fine numerical
155 grid as shown in Fig. 2. The numerical grid of cyclone A, B
and C contains 28871, 33056, and 18045 nodes respectively, to
yield a reasonable prediction. The details of the CFD setting are
presented in Table 3. The CFD simulation was performed with a
FIG. 2.
164
J. GIMBUN ET AL.
TABLE 3
Detail on CFD setting
Boundary condition
Inlet
Outlet
Cyclone wall
Viscous
Turbulence
Velocity inlet
Outflow
Standard wall function
Reynolds stress model (RSM)
RNG k-
Discretization
Pressure
Pressure-velocity coupling
Momentum
Turbulence kinetic energy
Turbulence dissipation rate
Reynolds stresses
Discrete phase modelling
Assumption
Maximum number of step
(phase integration)
d pc =
Presto!
SIMPLE
2nd order upwind
2nd order upwind
2nd order upwind
2nd order upwind
20000
i = 1 exp{1 }
0.25
De
D
1 = 2(S + L)/a
0.5
185
[9]
[15]
[16] 195
1.4
[14]
[7]
175
[13]
where
ab
dc = 0.47D
D2
(1 )K ww
Dr rwn
190
HS
[(dc /B) 1] for dc > B [8a]
(D/B) 1
for dc < B
[8b]
zc = H S
(1 n)( p g )d 2 Q
18 b rw1n rn1n
Spherical particle
9 Q
p z c vt2max
[12]
where
and
1
c0 (rw rn ) exp K (1+n)
r 1+n
c(r, ) =
rw
1
1+n dr
rn exp K (1+n) r
K =
[10]
[11]
The tangential velocity of a particle is equal to the tangential velocity of the gas flow, i.e. there is no slip in the
tangential direction between the particle and the gas.
The tangential velocity is related to the radius of cyclone by: u R n = constant.
205
[17]
where
8K c
K a2 K b2
(12D)0.14
T + 460 0.3
n = 1 1
2.5
530
2
p d pi
i =
18
G=
[18]
[19]
[20]
165
FIG. 3. Particle trajectories from CFD simulation of different particle size in the Bohnet cyclone at T = 1073 K.
210
d pc =
220
9 b
2 Ne vi ( p g )
12
[21]
1
H h
h+
a
2
[22]
1
1 + (d pc /d pi )2
[23]
FIG. 4. CFD flow field simulation on Bohnet cyclone (vi = 8 m/s, T = 293 K).
166
235
240
245
250
255
J. GIMBUN ET AL.
FIG. 5. Calculated and measured collection efficiencies for Kim and Lee [16]
cyclone (P = 1 Bar, T = 293 K, vi = 4.25 m/s, D = 0.311 m). Data point
from Kim and Lee (1990).
260
265
270
275
167
TABLE 4
Comparison of measured and predicted cut-off size of different cyclones
Models
Cyclone type and experiment value
Kim and Lee [16]
Dirgo and Leith [17] 5 m/s
Dirgo and Leith [17] 15 m/s
Ray et al. [19]
Bohnet [20] 873 K
Bohnet [20] 1073 K
Average deviation (%)
2.86
6.24
3.06
2.61
2.52
3.12
0
CFD
Li and Wang
Lapple
2.91
6.14
3.27
2.54
2.75
3.12
3.67
3.05
5.91
3.06
2.67
3.38
3.83
11.85
1.7
6.73
3.34
2.84
1.85
1.96
21.69
0.82
4.72
2.43
2.46
1.54
1.91
33.28
2.52
8.22
4.19
3.57
2.48
2.48
23.24
3.7% to the measured value. The Li and Wang, Lapple, Iozia and
Leith, and Koch and Licht models were found to be inconsistent
in the cut-off size prediction with the deviation ranging from 310
11.9 to 33.3% from the measured value.
6. CONCLUSIONS
The Li and Wang model and CFD code both predict very
well the cyclone efficiency and cut-off size for any operational
conditions. The Li and Wang model and FLUENT CFD code
produce a better fit to the Ray, Dirgo and Leith, and Kim and
Lee experimental data respectively. In all operating conditions
and cyclone types the FLUENT CFD and Li and Wang model
were found to be much closer to the experimental measurement.
However, only the FLUENT CFD code is consistently predicts
the cyclone cut-off size. Therefore, both the Li and Wang model
and FLUENT CFD code can be used to evaluate the collection
efficiency in the cyclone design except for the extreme operating temperatures, which is Li and Wang model is less accurate.
The Lapple and Koch and Lich models considerably underestimate the efficiency for large particles and overestimate efficiency
for small particles. Iozia and Leith logistic model show a good
agreement with an experimental data for the cyclone size range
of D = 0.250.4 m, but it is unable to predict correctly the efficiency for small cyclone (D < 0.1 m). Iozia and Leith model
is only suitable for efficiency prediction of cyclone diameter
around 0.25 m.
FIG. 9. Separation efficiency of Bohnet (1995) cyclone at high temperature (P = 1 Bar, vi = 8.61 m/s, D = 0.15 m). Data point from Bohnet [20].
315
320
325
330
168
ACKNOWLEDGEMENT
The authors would like to thank Dr. Tom Fraser, Fluent India
and Fluent Europe UK for their guidance and support.
u, v
Rer
CD
RANS
NOMENCLATURE
Greek Letters
375
v
particle response time (s)
g
gas viscosity (m2 /s)
slope parameter
angular coordinate
characteristic value
grade efficiency of particle size at mid-point of ini
ternal i (%)
g
gas density (kg/m3 )
385
p
particle mass density (kg/m3 )
L
a
b
D
De
340 H
h
S
B
c0 , c1
345 d p
Dr
d pc
335
n
Q
r
R
T
w
355 wn , ww
d pi
g
G
i
360 K a
Kb
Kc
Ne
350
365
J. GIMBUN ET AL.
vi
K
zc
dc
370 vt max
REFERENCES
1.
2.
3.
4.
5.
6.
7.
8.
9.
10.
11.
12.
13.
14.
15.
16.
17.
18.
19.
20.
390
395
400
405
410
415