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

net/publication/308188952

The flow pattern and entropy generation in an axial inlet cyclone with reflux
cone and gaps in the vortex finder

Article  in  Powder Technology · September 2016

DOI: 10.1016/j.powtec.2016.09.019

CITATIONS READS

18 2,583

5 authors, including:

Lu Duan Ji Zhongli
Tsinghua University China University of Petroleum - Beijing
4 PUBLICATIONS   38 CITATIONS    38 PUBLICATIONS   333 CITATIONS   

SEE PROFILE SEE PROFILE

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

Evaluation of gas-liquid separation performance of natural gas filters View project

All content following this page was uploaded by Lu Duan on 05 June 2018.

The user has requested enhancement of the downloaded file.

 Powder Technology 303 (2016) 192–202

Contents lists available at ScienceDirect

Powder Technology

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

The flow pattern and entropy generation in an axial inlet cyclone with
reflux cone and gaps in the vortex finder
Lu Duan a,c, Xiaolin Wu a,c,⁎, Zhongli Ji b,c, Zhiyi Xiong a,c, Jingxian Zhuang a,c
a
Department of Chemical Engineering, China University of Petroleum, Beijing 102200, PR China
b
Department of Mechanical and Transportation Engineering, China University of Petroleum, Beijing 102200, PR China
c
Beijing Key Laboratory of Process Fluid Filtration and Separation, China University of Petroleum, Beijing 102249, PR China

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

Article history: This paper aimed at studying the contribution of the reflux cone and the gaps in the vortex finder to the reduction
Received 1 February 2016 of energy consumption in the cyclones. The flow field was calculated using Reynolds Stress Model (RSM). Based
Received in revised form 14 August 2016 on the numerical investigations, the entropy generation analysis method was used to explain the mechanism of
Accepted 9 September 2016
energy consumption inside cyclone separators. The regional entropy generation in four parts viz. the outlet pipe,
Available online 14 September 2016
the inlet part, the region around vortex finder and the region below vortex finder was calculated and analyzed to
Keywords:
identify the zones where the energy is largely consumed. The results show that the reflux cone and the gaps help
Cyclone separator reduce the entropy generation in the gas-outlet pipe, in the regions around and under the vortex finder, whereas
Vortex finder increase the entropy generation in the inlet part. In addition, the reflux cone restrains the reflux flow and mainly
RSM reduces the entropy generation in the region around the vortex finder. The gaps reduce the flow rate under the
Entropy generation analysis vortex finder and mainly reduce the entropy generation in that region.
© 2016 Elsevier B.V. All rights reserved.

1. Introduction
Cyclone separators are widely used in the industrial process to re-
move particles larger than 10 μm from fluid flows. They are simply con-
structed, contain no moving parts, and have low manufacturing and
maintenance costs. As a result, the cyclone separators, manufactured
from materials including alumina, alumina-silica, zirconia, magnesia,
beryllia, and silicon carbide [1], and the unique constructions such as
the convex downward top or the vault top [1,2], can be used under ex-
treme conditions. For example, the cyclones can protect downstream
equipment from contaminants in a high-pressure gas network [3].
Since high operating pressure provides a dense gas atmosphere and
then leads to a high pressure drop, the structure of the cyclone separator
must be optimized to reduce its pressure drop.
Many theoretical and experimental studies have been carried out to
determine the effect of the eight geometrical dimensions of the cyclones
on performance [4,5]. The results indicated that the shape and size of
the vortex finder play a vital role in the pressure drop and separation ef-
ficiency. Hoekstra [6] studied the tangential and axial mean velocities in
the cyclones by Laser-Doppler Anemometry (LDA) and Computational
Fluid Dynamics (CFD) and tested the effect of the non-dimensional
⁎ Corresponding author at: Department of Chemical Engineering, China University of
Petroleum, Beijing, 18th Fuxue Rd., Beijing 102200, PR China.
E-mail address: wuxl_du@126.com (X. Wu).
diameter of the vortex finder in a range of 0.3 to 0.5. As the diameter
of the vortex finder decreased, the tangential and axial mean velocities
increased significantly, causing an increase in pressure drop, which was
also observed using Large Eddy Simulation (LES) by Elsayed and Lacor
[7]. El-Batsh et al. [8] found that the length of the vortex finder is less im-
portant than its diameter in affecting pressure drop. A performance map
was developed to allow selection of the diameter and length of the vor-
tex finder. Lim et al. [9] compared the performance of ten self-designed
cyclones and found that cyclones with a cone-shaped vortex finder ex-
hibit lower pressure drop and moderate separation efficiency compared
with the cylinder-shaped vortex finders. Raoufi et al. [10] also evaluated
the effect of the shape and diameter of vortex finder on cyclone perfor-
mance using CFD method. The results showed that the low-pressure
zone in the middle of the cyclones expands with the small vortex finder
divergence angle, decreasing separation efficiency and pressure drop.
Usually, separation efficiency improves simultaneously with an in-
crease in the pressure drop [7]. Jin et al. [11,12] developed a high-
efficiency guide-vaned cyclone, called the PSC type cyclone, and found
that gaps in the vortex finder improve flow field and performance.
Xiong et al. [13] ameliorated the vortex finder configuration by adding
a reflux cone and spiral gaps to reduce the pressure drop. They designed
cyclones with six types of cone-shaped vortex finders and tested perfor-
mance at different flow rates and particle concentrations. They found
that the cyclone separator with a vortex finder that has a reflux cone
and spiral gaps with dextrorotation of 15° was optimal, and reduced

http://dx.doi.org/10.1016/j.powtec.2016.09.019
0032-5910/© 2016 Elsevier B.V. All rights reserved.
 L. Duan et al. / Powder Technology 303 (2016) 192–202
the pressure drop by 73% compared with the one without the reflux
cone and spiral gaps. However, this study did not describe the flow
field patterns or the mechanism of energy consumption reduction.
The CFD method is a high effective, reliable, labor-saving, and time-
saving approach to predict the flow behavior and performance of cy-
clones [14,15,16,17,18]. The entropy generation analysis method,
based on the numerical results, is a favorable approach to evaluate ener-
gy consumption because the exergy loss is proportional to the entropy
generation [19]. This approach has been widely used to study and im-
prove the flow and heat transfer equipment including the centrifugal
fan [20] and heat exchangers [21,22,23,24,25]. Therefore, the entropy
generation analysis method can analyze the mechanism of energy con-
sumption and help identify the high energy consumption regions inside
the cyclone separator.
In this study, the flow field of cyclone separators with four different
vortex finders was simulated using Reynolds Stress Model (RSM). The
numerical results were compared with the experimental data to assess
the reliability of the simulation results. Entropy generation was calculat-
ed based on the simulated results. The effects of the configuration of
vortex finder on energy consumption were studied using the entropy
generation analysis method.
2. Material and methods
2.1. Construction of the cyclones
Fig. 1 presents the basic geometrical dimensions of the cyclone sep-
arator and its flow pattern. In the inlet part, eight vanes were arranged
symmetrically, as shown in the cross-section A-A, to allow swirling flow
in the clockwise direction.
Fig. 1. Geometry and inside flow of the cyclone separator.
193
Four vortex finders are considered in this study, as presented in
Fig. 2. Type A was a basic cone-shaped vortex finder. This was modified
by adding a reflux cone to give Type B. Type B was modified by slotting
18 straight gaps to give Type C, and 18 spiral gaps with 15°
dextrorotation for Type D. The gaps are slanted in the opposite direction
of gas flow, as shown in cross-section B-B in Fig. 2. Two different vortex
finders with spiral gaps with either a dextrorotation of 30° or a
levorotation of 15° were also studied in Xiong's work [13]. The perfor-
mance of these two vortex finders was not as good as the Type D cy-
clone, so were not included in this study.
2.2. CFD method
2.2.1. Turbulent model
The flow field in the cyclone separator is three-dimensional, high
swirling and intense anisotropic turbulence. RSM and LES were verified
to be the favorable turbulent models to simulate the flow field in cy-
clones [26,27,28]. However, the RSM model has lower computer capac-
ity requirements and shorter running time compared with LES [27].
RSM also provides a better prediction in the boundary layer with the rel-
ative coarse grids [29]. In addition, LES does not provide the turbulent
dissipation rate directly, which required to determine the entropy gen-
eration [30]. Therefore, RSM was selected to predict the flow field in the
cyclones. The details of RSM model are provided in reference [31].
2.2.2. Gas-solid two-phase flow model
The inlet particle concentration ranged from 100 to 500 mg/m3, for a
volume fraction far less than 10%. Thus, the particle interaction is negli-
gible and the Eulerian-Lagrangian approach, called discrete phase
model (DPM), can be used to track individual particles [31]. Only the
drag force and the gravity are taken into consideration here due to the
194 L. Duan et al. / Powder Technology 303 (2016) 192–202

Fig. 2. Geometries of the vortex finders.

small fluid-to-particle density ratio [15]. According to Newton's laws of [37,38] who focused on wall friction. The entropy generation caused
motion, the equation of motion for a particle can be written as [31]: by direct dissipation can be neglected because its accounts for less
than 1.5% of the total entropy generation in a cyclone separator [37,


! ! 38]. Thus, in this study, entropy generation due to turbulent dissipation
d up 18μ C D Re
! !  g ρp −ρ and wall friction is taken into consideration, which are listed as follows:
¼ u−up þ ð1Þ
dt 2
ρp dp 24 ρp
• In turbulent flow, the turbulence kinetic energy, defined as k, is dissi-
pated into heat by viscous force in the Kolmogorov microscales [39],
! !
where the u is the fluid phase velocity, u p is the particle velocity, μ is resulting in energy dissipation. The volumetric entropy generation
the molecular viscosity of fluid, ρp is the particle density, dp is the parti- rate due to turbulent dissipation is associated with the turbulent ki-
cle diameter, Re ¼
ρdp jup −uj
is the relative Reynolds number, and CD is the netic energy dissipation, which can be written as [33–35]
μ
drag coefficient, which can be calculated by the correlations developed
by Morsi and Alexander [32].
ρε
Discrete random walk (DRW) is used to evaluate the effect of turbu- S‴gen;t ¼ ð2Þ
T
lent dispersion on particle motion. The separation efficiency was deter-
mined by counting the particle numbers at the inlet and outlet of the
where ρ is the gas density, ε is the turbulent energy dissipation rate,
cyclone separator.
and T is the gas temperature.
In the RSM turbulent model, two transport equations for k and ε are
2.2.3. Computational methodology
solved to calculate the turbulence kinetic energy and turbulent dissipa-
The following methods were used in the simulation: the SIMPLEC
tion rate [31].
(semi-implicit method pressure-linked equations consistent) algorithm
The entropy generation due to turbulent dissipation can be deter-
was used for pressure-velocity coupling, the PRESTO! (Pressure Stag-
mined by integrating Eq. (2) over the flow in the cyclone:
gering Option) pressure interpolation scheme was applied for pressure
discretization because of the high rotating flows, the QUICK ΔSgen;t ¼ ∭ Ω S‴gen;t dV ð3Þ
discretization scheme was used to calculate momentum, and the
second-order upwind discretization scheme was used to obtain the tur-
in which Ω denotes the entire volume of the cyclone.
bulent kinetic energy and turbulent dissipation rate. The simulation was
performed via the commercial software Fluent 6.3.26 on a workstation • The isotropic assumption for ε in RSM deviates considerably because ε
with an Intel Xeon 2.9 GHz CPU and 64 GB RAM. is strongly anisotropic in the low-Reynolds-number near-wall regions
[40–42], resulting in an under estimate of entropy generation [33–35].
2.2.4. Boundary condition Therefore, a model to calculate wall entropy generation was devel-
The velocity inlet boundary condition was set at the inlet. The flow oped by Duan et al. [37,38], and is written as
rate ranged from 214 to 482 m3/h. The parameters of the inlet boundary
condition were set according to reference [31]. The outflow boundary
Z
condition was utilized at the outlet. Air enters the inlet with different τ w vp
velocities at 300 K. The flow is assumed to be fully developed flow at ΔSgen;w ¼ dA ð4Þ
T
the inlet and outlet. The no-slip condition and the adiabatic condition
are assumed at the walls. The standard wall function is applied to where vp is the fluid velocity at each node in a grid cell and τw is the
solve turbulent flow problems in the wall regions. wall-shear stress.

2.3. Entropy generation analysis method 2.3.2. Exergy loss

Since turbulent dissipation and wall friction are the main
2.3.1. Entropy generation factors determining energy consumption [37,38], the resulting entropy
In a non-reacting flow system, the energy consumption in cyclones generation determines the exergy loss, which can be expressed as
results from direct dissipation, turbulent dissipation, and wall friction  
[33–35]. There are several models to predict entropy generation includ- I ¼ T 0 ΔSgen;t þ ΔSgen;w ð5Þ
ing the model of Bejan [36] that relied on direct dissipation, Kork and
Herwig [33–35] who utilized turbulent dissipation, and Duan et al. where T0 is the inlet gas temperature.
 L. Duan et al. / Powder Technology 303 (2016) 192–202 195

Table 1
Exergy loss in the cyclone meshed with different wall-adjacent grid thicknesses.

Type A Type B Type C Type D

Grid levels n y+ δ n y+ δ n y+ δ n y+ δ

1 90,360 875 27.6 132,920 564 7.7 159,589 452 9.2 175,221 657 6.3
2 228,412 529 14.2 275,356 430 5.3 331,874 324 10.3 313,089 420 4.9
3 454,848 423 1.1 471,088 379 3.1 589,589 306 1.3 629,408 370 2.9
4 1,090,294 385 4.5 924,716 319 0.6 968,076 287 6.3 1,071,277 274 11.0
5 1,846,419 320 6.0 1,728,576 264 8.8 1,460,842 197 10.1 1,786,738 230 7.1

The exergy loss can also be determined by the exergy analysis meth-
od, written as
1

Iex ¼ H1 −H2 −T 0 ðs1 −s2 Þq_ m þ q_ m V 21 −V 22
2 field and the entropy generation for Type A, Type B, Type C, and Type
where q_ m is the mass flow rate. s1 and s2 represent the thermody-
namic entropy at the inlet and outlet, respectively. V1 and V2 represent
the gas velocity at the inlet and outlet, respectively.
3.6%. With the further increase of grid number, there was no significant
improvement in the relative deviation of pressure drop, suggesting that
the grids of level 3 produce grid independent results.
Combining the grid independency and exergy loss results, the flow
ð6Þ
D were reasonably predicted with grid numbers of 454,848, 471,088,
589,589, and 629,408, respectively.
3. Results and discussion

2.4. Grid division and grid test 3.1. Reduction rates in energy consumption

The three complex vortex finders with reflux cone and gaps were
created using ProE.4.0 and were imported into the Gambit code. The
four cyclone separators were established using Gambit 2.4.6. They
were meshed mainly using the structured grid and partially applying
the unstructured grid. Since the grid has a crucial impact on calculating
the flow field and the entropy generation [38], five levels of grids for
each cyclone were tested to evaluate the entropy generation analysis
method and the grid independency. As the grid level increased, the
thickness of the wall-adjacent grids decreased and the grid number in-
creased. The turbulent wall coordinates (y+) can be obtained in the
post-process, and generally decrease with the decrease of the thickness
of the wall-adjacent grids. The grid number and the turbulent wall coor-
dinate for each level of grids are shown in Table 1.
According to Section 2.3.2, the exergy loss can be determined by two
methods viz., entropy generation analysis and exergy analysis, with lit-
tle deviation at a reasonable y+ [39]. This deviation, can be written as
Eq. (7), and is proposed to evaluate the effect of the thickness of the
wall adjacent grids on the entropy generation.
Table 3 compares the numerical results and experimental data for
the relative reductions of pressure drop of the modified cyclone separa-
tors at different inlet flow rates. The relative reduction of pressure drop
of Type A versus Type B cyclones is used to estimate the effect of reflux
cone on saving energy. The pressure drops of the Type C and Type D cy-
clones are compared with that of the Type B cyclone to illustrate the ef-
fect of different types of gaps on saving energy. As is shown in Table 3,
the numerical results are generally in good agreement with the experi-
mental data, except for the testing results for the Type D cyclone. The
CFD method underestimates the reduction of pressure drop of the
Type D cyclone separator. Overall, the simulation results indicate the
trend of the pressure drop reduction for the three modified types of cy-
clone separators.
Table 4 shows the reduction of exergy loss of the modified cyclones.
The exergy loss for each cyclone separator is in direct proportion to the
pressure drop [38] and can be determined using Eqs. (2)–(6). As is

jIex −Ij Table 3

δ¼  100% ð7Þ The reduction of pressure drop between the experimental data and numerical results.
Iex
qV m3/h 214 268 321 375 428 482 Average
Table 1 shows the relative deviations of exergy loss for each level of jΔpexp
A
−Δpexp
B
j
 100% 36.6 38.0 39.6 40.3 41.1 38.7 39.1
Δpexp
grids when the flow rate was set to 375 m3/h. As presented in Table 1, jΔpexp
A
−Δpexp j
B C
 100% 23.1 33.2 44.3 33.8 22.8 26.1 30.5
the relative deviations of exergy loss at the level 3 grids are generally Δpexp
B
jΔpexp −Δpexp j 62.5 63.0 63.5 59.4 55.0 56.0 59.9
less than those in the other four levels of grids. B
Δpexp
D
 100%
B
In addition, the grid independency was tested by comparing the jΔpCFD −Δp CFD j
36.6 34.6 38.8 43.5 40.0 39.7 38.8
A B
 100%
pressure drop between the grids of level i and i − 1 for each cyclone (i ΔpCFD
A
jΔpCFD −ΔpCFD j 32.9 33.8 31.2 28.6 29.6 31.9 31.3
ranged from 2 to 5). As is shown in Table 2, the relative deviations of B
ΔpCFD
C
 100%
B

pressure drop between the grids of level 3 and level 4 were less than jΔpCFD
B
−Δp CFD j
D
 100% 46.1 47.3 44.9 43.2 45.0 46.2 45.5
ΔpBCFD

Table 2 Table 4
The grid independency test for the four types of cyclones. Reduction of exergy loss for modified cyclone separators.
Grid level Type A Type B Type C Type D qV m3/h 214 268 321 375 428 482 Average
jΔP 1 −ΔP 2 j 6.4 10.4 7.8 6.2
ΔP 1  100/% jΔI CFD CFD
A −ΔI B j 33.1 30.8 36.0 40.8 36.5 37.4 35.8
ΔICFD
 100%
jΔP 2 −ΔP 3 j 4.6 7.0 7.0 3.4
ΔP 2  100/% A

jΔI CFD CFD

B −ΔI C j
 100% 35.7 34.6 33.1 30.4 31.8 32.1 33.0
jΔP 3 −ΔP 4 j 3.5 1.61 3.6 2.9
ΔP 3  100/% ΔICFD
B

jΔP 4 −ΔP 5 j 2.5 0.9 0.8 1.8 jΔI CFD CFD

B −ΔI D j
 100% 48.6 49.8 46.6 44.7 46.7 46.1 47.1
ΔP 4  100/% ΔICFD
B
196 L. Duan et al. / Powder Technology 303 (2016) 192–202
shown in Table 4, the relative reduction of exergy loss agrees well with
the relative reduction of pressure drop, indicating that the entropy gen-
eration analysis method can effectively evaluate the reduction of energy
consumption for modified cyclone separators.
3.2. Flow pattern
3.2.1. Velocity vector
In this section, the velocity vector was observed for a flow rate of
375 m3/h. As is shown in Fig. 3, three planes were studied in the gas-
outlet pipe (z = 200 mm) and near the vortex finder (z = −100 mm
and y = 0 mm). In Type A and Type B cyclones, both the outer down-
ward flow and the inner upward flow occur in the clockwise direction.
However, due to the gaps in the vortex finder, part of the flow in the
Type C and Type D cyclones are in the counter-clockwise direction espe-
cially in the near wall regions, as shown by the planes of z = 200 mm
and z = −100 mm. The flow field is even totally in the opposite direc-
tion in the gas-outlet pipe of the Type C cyclone. In the gas-outlet pipe of
the Type D cyclone, the inner flow is in the clockwise direction, but the
outer flow is in the counter-clockwise direction.
Through the gaps, a small portion of airflow enters the vortex
finder and destroys the boundary layer of the swirling flow in the
near-wall region [13]. Thus, the upward flow rate in the near-wall
(a)
(b)
(c)
Fig. 3. Velocity vector at various positions in the different types of cyclone separators. (a) z = 200 mm. (b) z = −100 mm. (c) y = 0 mm.
region of the vortex finder decreases dramatically in Type C and
Type D cyclones relative to that in Type A and Type B cyclones, as
shown in Fig. 3(c). As a result, the energy dissipation is significantly
reduced. The velocity vector in the vortex finders of Type A and Type
B cyclones indicates that the reflux flow is restrained effectively by
adding a reflux cone. In addition, the spiral gaps expand the flow
areas and increase the airflow through the vortex finder, resulting
in the reflux flow.
3.2.2. Velocity contours
For the four types of cyclones, the contours of the tangential and
axial velocities on the plane of y = 0 mm are depicted in Fig. 4. The
maximum tangential and axial velocities of the modified cyclones
all decrease, compared to those of the original cyclone separator.
Owing to the gaps, the maximum tangential velocity shifts from the
space under the vortex finder to the annular space outside the vortex
finder, and the maximum axial velocity moves from the under part to
the upper part of the vortex finder. The non-symmetry flow field is
also found to be ameliorated by adding gaps. In the gas-outlet pipe
of the Type C and Type D cyclones, the partial swirling flow turns
to the inverse direction compared with those in Type A and Type B
cyclones.
 L. Duan et al. / Powder Technology 303 (2016) 192–202
(a)
(b)
Fig. 4. Velocity contours on the plane of y = 0 mm. (a) Tangential velocity. (b) Axial
velocity.
3.2.3. Velocity profile
Fig. 5 shows the tangential and axial components of gas velocity at
various axial positions on the plane of y = 0 mm. There are four posi-
tions located in the gas-outlet pipe at z = 200 mm, among the vortex
finder at z = −100 mm, in the cylindrical part at z = − 300 mm and
in the conical part at z = − 500 mm, respectively. The magnitude of
the tangential velocity in Type C and Type D cyclones is generally less
than that in Type A and Type B cyclones, indicating that the swirling in-
tensity in Type C and Type D cyclones is weaker than that in Type A and
Type B cyclones. In the gas-outlet pipe of Type C cyclone, the tangential
197
velocity is below zero, indicating that the airflow direction is opposite to
that in Type A and Type B cyclones. The tangential velocity pattern
mainly shows the same behavior of the Rankine vortex.
In the space under the vortex finder, the magnitude of axial veloci-
ties in the near wall regions of Type C and Type D cyclones is larger
than that in Type A and Type B cyclones, allowing better separation ef-
ficiency, as shown by Xiong [13]. In the space under the vortex finder of
Type A and Type B cyclones, the axial velocity pattern shows an inverted
V profile, which would increase stress on the vortex finder and increase
vibration and noise [43]. In addition, in the vortex finder of Type A and
Type B cyclones, it shows a V profile. In the Type C and Type D cyclones,
the axial velocity patterns exhibit a small W profile in the vortex finder
and an inverted W profile in the space under the vortex finder. In addi-
tion, the symmetry of the flow field is improved by the gaps in the vor-
tex finder.
3.3. Distribution of entropy generation due to turbulent dissipation
As is presented in Fig. 6, the distribution of volumetric entropy gen-
eration rate due to turbulent dissipation was analyzed for a flow rate of
375 m3/h. In the space under the vortex finder, the value for the Type C
and Type D cyclones is about two orders smaller than that in Type A and
Type B cyclones due to the weak swirling intensity, resulting from the
escape of flow through the gaps. The value in the Type B cyclone is gen-
erally less than that in the Type A cyclone, indicating that the vortex
finder with a reflux cone leads to a decrease of energy consumption,
which was also reported by Lim et al. [9], Raoufi et al. [10], and Xiong
[13]. For these four types of cyclones, the maximum S‴gen, t is generally
obtained inside the vortex finders. It moves up from the entrance to
the upper part of the vortex finder by the addition of gaps. In addition,
the large value in Type A and Type C cyclones is located at the center
of the vortex finer, whereas it is achieved in the near-wall regions in
the Type B and Type D cyclones.
Fig. 7 also presents the volumetric entropy generation rate due to
turbulent dissipation at the same positions as Fig. 5. It is also ob-
served that the value in the space under the vortex finder of Type C
and Type D cyclones is less than that of Type A and Type B cyclones.
However, in the center of the gas-outlet pipe of Type C and Type D
cyclone, the value is slightly larger than that of the other two cy-
clones due to the interaction of the flows in the inverse direction.
Whereas, in the near-wall regions of the gas-outlet pipe, the value
in Type A and Type B cyclones is enormously less than it in Type C
and Type D cyclones. The tangential and axial velocities in the annu-
lar part outside the vortex finders of Type C and Type D cyclones are
larger than those of Type A and Type B cyclones. As a result, the en-
tropy generation rate for Type C and Type D cyclones is larger than
that for Type A and Type B cyclones.
3.4. Regional entropy generation
As described above, the distribution of S‴gen , t varies in different re-
gions for the four types of cyclones. Thus, each cyclone is divided into
four regions, including: the gas-outlet pipe, the inlet part, and the re-
gions around and under the vortex finder, as presented in Fig. 8. The
regional entropy generation due to turbulent dissipation and wall
friction was calculated by Eqs. (1)–(4) for each cyclone at flow rate
of 375 m3/h.
Fig. 9 shows the regional entropy generation for each cyclone sepa-
rator. It shows that the reflux cone and the gaps in the vortex finder re-
duce entropy generation in the regions around the vortex finder, under
the vortex finder, and in the gas-outlet pipe. The reflux cone restrains
reflux flow in the vortex finder. As a result, the entropy generation of
Type B cyclone decreases by 36.5% compared with that of the Type A cy-
clone, including 29.4% around the vortex finder, 7.3% under the vortex
finder, and 0.7% in the gas-outlet pipe. The non-symmetry of the flow
field was improved and the magnitude of the tangential and axial
198 L. Duan et al. / Powder Technology 303 (2016) 192–202

velocities decreased due to the gaps. Thus, for Type C cyclone, the entro-
py generation decreases by 33.4% compared with that of Type B cyclone,
including 27.1% under the vortex finder, 5.2% in the gas-outlet pipe, and
4.0% around the vortex finder. In addition, for the Type D cyclone, the
entropy generation reduces by 47.8% compared with that of the Type
B cyclone, with 30.2% under the vortex finder, 7.7% in the gas-outlet
pipe, and 12.9% around the vortex finder. These results indicate that spi-
ral gaps are superior to the straight gaps for the reduction of energy con-
sumption. However, the entropy generation in the inlet part of the three
modified cyclones increases by 0.9%, 3.0% and 3.0%, respectively, likely
due to the increase of the swirling intensity in the annular space outside
the vortex finder.

Fig. 5. Tangential and axial velocities at various axial positions on the plane through the Z-axis: P1—z = 200 mm; P2—z = −100 mm; P3—z = −300 mm; P4—z = −500 mm.
 L. Duan et al. / Powder Technology 303 (2016) 192–202 199

Fig. 5 (continued).

3.5. Grade separation efficiency 4. Conclusions

Fig. 10 shows the grade separation efficiency for the four cyclone
separators, as calculated by the Eulerian-Lagrangian multiphase model
at an inlet flow rate of 375 m3/h. The particle density is 2700 kg/m3
and the inlet particle concentration is 200 mg/m3. It is found that the
grade efficiency of Type B cyclone is generally less than that of the
Type A cyclone, which differs from the findings of Xiong's work [13].
However, compared with the grade efficiency of the Type A cyclone,
an improvement is observed for the Type C and Type D cyclones, in
agreement with the experimental data from Xiong's work [13]. This is
because that the particles did not access the gaps due to the inertia ef-
fect and the amount of the separated particles escaping from the vortex
finder entrance decreases due to the reflux cone. The Type D cyclone
shows better separation performance than the Type C cyclone.
The entropy generation analysis method combined with CFD meth-
od was successfully applied to investigate the flow field and the mech-
anism of the energy consumption in four types of cyclones. The cyclones
were considered in four parts: the gas-outlet pipe, the inlet part, the re-
gion around the vortex finder, and the region under the vortex finder.
The entropy generation of each part was then calculated and analyzed
at flow rate of 375 m3/h. The results showed that the reflux cone ame-
liorated the reflux flow in the vortex finder to effectively reduce the
overall energy consumption in the cyclone separator, as the entropy
generation of the Type B cyclone was decreased by 36.5% compared to
that of the Type A cyclone. The region around the vortex finder is the
key region for decreasing entropy generation for the Type B cyclone.
The gaps reduce the flow rate under the vortex finder and can

Fig. 6. Contours of the entropy generation rate on the plane of y = 0 mm and on the Z-axis.
200 L. Duan et al. / Powder Technology 303 (2016) 192–202

Fig. 7. S‴gen,t at various axial positions on the plan through the Z-coordinate: P1—z = 200 mm; P2—z = −100 mm; P3—z = −300 mm; P4—z = −500 mm.

effectively reduce energy consumption. The overall entropy generation Greek symbols
of the Type C and Type D cyclones decreased by 33.4% and 47.8%, respec- Δp cyclone pressure drop, Pa
tively, compared to the Type B cyclone. The region under the vortex ΔSgen,t entropy generation due to turbulent dissipation, W/K
finder is the key region that contributed to decrease entropy generation ΔSgen,w wall entropy generation, W/K
for Type C and Type D cyclones. In addition, the grade separation effi- ρ density, kg·m−3
ciency of Type C and Type D cyclones was slightly higher than that of μ dynamic viscosity, Pa·s
the Type A cyclone, suggesting that Type C and Type D vortex finders ε turbulent dissipation rate, W/kg
can improve cyclone performance. The Type D cyclone separator is the τw wall-shear stress, Pa
optimum one, exhibiting the lowest entropy generation and the highest Ω entire cyclone
separation efficiency.
Superscripts and subscripts
Notation ()''' volumetric rate
()1 variable at inlet
()2 variable at outlet

H enthalpy, J/s
I exergy loss by the entropy generation analysis, J/s Acknowledgements
Iex exergy loss by the exergy analysis, J/s
p pressure, Pa This work was supported by the general project “The cyclone
q_ m mass flow rate, kg/s separator performance investigation based on entropy generation
Re Reynold's number theory” sponsored by the National Natural Science Foundation of
T gas temperature, K China (No. 51474229), and the Special and Significant Projects of
T0 fluid temperature at inlet, K National Science and Technology “Safe operation technology
V fluid velocity, m/s optimization of ten billion level natural gas purification plant”
vp fluid velocity at the node in the first grid cell near the wall, (No. 2011ZX05017-005) sponsored by the Ministry of National
m/s Science and Technique.
 L. Duan et al. / Powder Technology 303 (2016) 192–202
Fig. 8. Regions inside the cyclone.
201
Fig. 9. Entropy generation for different regions in the four types of cyclones.
Fig. 10. Separation efficiency for the four types of cyclones.
References
[1] E.C. Clark, Caldwell, E.B. Schumacher et al., Cyclone separator for high temperature
operations, US, 3470678, 1969
[2] G.G. Sun, S.Q. Li, S.X. Yang, et al., Performance and application of a cyclone at high
temperature and high pressure, J. China Univ. Pet. 30 (6) (2006) 98–101.
[3] N.A. Tsochatzidis, K.E. Maroulis, Methods help remove black powder from gas pipe-
lines, Oil Gas J. 105 (10) (2007) 52–58.
[4] E.M. Moore, A.R. Mcfarland, Performance modeling of single- inlet aerosol sampling
cyclone, Environ. Sci. Technol. 27 (9) (1993) 1842–1848.
[5] K. Elsayed, C. Lacor, Optimization of the cyclone separator geometry for minimum
pressure drop using mathematical models and CFD simulations, Chem. Eng. Sci. 65
(22) (2010) 6048–6058.
[6] A.J. Hoekstra, Gas Flow Field and Collection Efficiency of Cyclone Separators(Ph.D.
thesis) Technical University Delft, 2000.
[7] K. Elsayed, C. Lacor, The effect of cyclone vortex finder dimensions on the flow pat-
tern and performance using LES, Comput. Fluids 71 (2013) 224–239.
[8] H.M. El-Batsh, Improving cyclone performance by proper selection of the exit pipe,
Appl. Math. Model. 37 (7) (2013) 5286–5303.
[9] K.S. Lim, H.S. Kim, K.W. Lee, Characteristics of the collection efficiency for a cyclone
with different vortex finder shape, J. Aerosol Sci. 35 (6) (2004) 743–754.
[10] A. Raoufi, M. Shams, M. Farzaneh, et al., Numerical simulation and optimization of
fluid flow in cyclone vortex finder, Chem. Eng. Process. 47 (1) (2008) 128–137.
[11] Y.H. Jin, C. Fan, M.X. Shi, Research and development of PSC-300 high efficiency cy-
clone tube with large gas capacity, Pet. Refin. Eng. 32 (2) (2002) 24–27.
[12] Y.H. Jin, J.J. Wang, H.W. Wang, et al., Experimental and numerical study of gas flow
in PSC type cyclone tube, J. China Univ. Pet. 30 (6) (2006) 71–82.
[13] Z.Y. Xiong, Z.L. Ji, X.L. Wu, Development of a cyclone separator with high efficiency
and low pressure drop in axial inlet cyclones, Powder Technol. 253 (2014) 644–649.
202 L. Duan et al. / Powder Technology 303 (2016) 192–202
[14] A.C. Hoffmann, L.E. Stein, Gas Cyclones and Swirl Tubes: Principles, Design and Op-
eration, 2nd ed. Springer, New York, 2002 41–55.
[15] T.G. Chuah, J. Gimbun, T.S.Y. Choong, A CFD study of the effect of cone dimensions on
sampling aerocyclones performance and hydrodynamics, Powder Technol. 162 (2-
1) (2006) 126–132.
[16] H. Shalaby, K. Pachdler, K. Wozniak, et al., Comparative study of the continuous
phase flow in a cyclone separator using different turbulence models, Int. J. Numer.
Methods Fluids 48 (11) (2005) 1175–1197.
[17] K.W. Chu, S.B. Kuang, A.B. Yu, et al., Particle scale modeling of the multiphase flow in
a dense medium cyclone: effect of fluctuation of solids flowrate, Miner. Eng. 33
(2012) 34–45.
[18] K.W. Chu, B. Wang, A.B. Yu, A. Vince, Particle scale modeling of the multiphase flow
in a dense medium cyclone: effect of vortex finder outlet pressure, Miner. Eng. 31
(2012) 46–58.
[19] H.F. Oztopa, K. AL-Salem, A review on entropy generation in natural and mixed con-
vection heat transfer for energy systems, Renew. Sust. Energ. Rev. 16 (2012)
911–920.
[20] A. Behzadmehr, Y. Mercadier, Numerical study of flow parameters and entropy gen-
eration on a centrifugal fan, Int. J. Exergy 6 (1) (2009) 80–92.
[21] T.A. Jankowski, Minimizing entropy generation in internal flows by adjusting the
shape of the cross-section, Int. J. Heat Mass Transf. 52 (15-16) (2009) 3439–3445.
[22] R. Ben-Mansour, A.Z. Sahin, Entropy generation in developing laminar fluid flow
through a circular pipe with variable properties, Heat Mass Transf. 42 (1) (2005)
1–11.
[23] E. Amani, M.R.H. Nobari, A numerical investigation of entropy generation in the en-
trance region of curved pipes at constant wall temperature, Energy 36 (8) (2011)
4909–4918.
[24] A. Biyikoglu, Entropy generation due to flow across the abrupt contraction of pipe
joints, Appl. Therm. Eng. 29 (5-6) (2009) 841–847.
[25] I. Kurbas, A. Durmus, H. Eren, et al., Effect of propeller type swirl generators on the
entropy generation and efficiency of heat exchangers, Int. J. Therm. Sci. 46 (3)
(2007) 300–307.
[26] A.J. Hoekstra, J.J. Derksen, H.E.A. VanDenAkker, An experimental and numerical
study of turbulent swirling flow in gas cyclones, Chem. Eng. Sci. 54 (13-14)
(1999) 2055–2065.
[27] M.D. Slack, R.O. Prasad, A. Bakker, et al., Advances in cyclone modeling using un-
structured grids, Chem. Eng. Res. Des. 78 (8) (2000) 1098–1104.
View publication stats
[28] C. Fredriksson, Exploratory Experimental and Theoretical Studies of Cyclone Gasifi-
cation of Wood Powder(Ph.D. thesis) Lulea University of Technology, Sweden, 2003.
[29] B. Wegner, A. Maltsev, C. Schneider, et al., Assessment of unsteady RANS in
predicting swirl flow instability based on LES and experiments, Int. J. Heat Fluid
Flow 25 (3) (2004) 528–536.
[30] M. Soos, R. Kaufmann, R. Winteler, et al., Determination of maximum turbulent en-
ergy dissipation rate generated by a Rushto impeller through large Eddy simulation,
AICHE J. 59 (10) (2013) 3642–3658.
[31] Fluent.Inc, Fluent6.3 User's Guide, 2006 12.37–12.47.
[32] S.A. Morsi, A.J. Alexander, An investigation of particle trajectories in two phase flow
systems, J. Fluid Mech. 55 (2) (1972) 193–208.
[33] F. Kock, H. Herwig, Local entropy production in turbulent shear flows: a high-
Reynolds number model with wall functions, Int. J. Heat Mass Transf. 47 (10−11)
(2004) 2205–2215.
[34] F. Kock, H. Herwig, Entropy production calculation for turbulent shear flows and
their implementation in CFD codes, Int. J. Heat Mass Transf. 26 (4) (2005) 672–680.
[35] F. Kock, H. Herwig, Direct and indirect methods of calculating entropy generation
rates in turbulent convective heat transfer problems, Int. J. Heat Mass Transf. 43
(3) (2007) 207–215.
[36] A. Bejan, Entropy Generation through Heat and Fluid Flow, Wiley, New York, 1982
48–105.
[37] L. Duan, X.L. Wu, Z.L. Ji, Application of entropy generation method for analyzing en-
ergy consumption of cyclone separator, J. CIESC 65 (2) (2014) 583–592.
[38] L. Duan, X.L. Wu, Z.L. Ji, et al., Entropy generation analysis on cyclone separators with
different exit pipe diameters and inlet dimensions, Chem. Eng. Sci. 138 (2015)
622–633.
[39] M.T. Landahl, E. Mollo-Christensen, Turbulence and Random Processes in Fluid Me-
chanics, 2nd ed. Cambridge University Press, 1992, ISBN 978-0521422130 10.
[40] B.E. Launder, W.C. Reynolds, Asymptotic near-wall stress dissipation rates in a tur-
bulent flow, Phys. Fluids 26 (1983) 1157–1158.
[41] S. Jakirlic, K. Hanjalic, A new approach to modeling near-wall turbulence energy and
stress dissipation, J. Fluid Mech. 459 (2002) 139–166.
[42] M. Hallback, J. Groth, A.V. Johansson, An algebraic model for nonisotropic turbulent
dissipation rate in Reynolds stress closures, Phys. Fluids 2 (10) (1990) 1859–1866.
[43] K. Elsayed, C. Lacor, The effect of cyclone inlet dimensions on the flow pattern and
performance, Appl. Math. Model. 35 (4) (2011) 1952–1968.