Separation and Purification Technology 174 (2017) 22–28

Contents lists available at ScienceDirect

Separation and Purification Technology

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

CFD modeling of particle behavior in supersonic flows with strong swirls

for gas separation
Yan Yang a, Chuang Wen a,b,⇑
a
School of Petroleum Engineering, Changzhou University, Changzhou 213016, China
b
Section of Fluid Mechanics, Coastal and Maritime Engineering, Department of Mechanical Engineering, Technical University of Denmark, Nils Koppels Allé, 2800 Kgs.
Lyngby, Denmark

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

Article history: The supersonic separator is a novel technique to remove the condensable components from gas mixtures.
Received 12 June 2016 But the particle behavior is not well understood in this complex supersonic flow. The Discrete Particle
Received in revised form 19 September Method was used here to study the particle motion in supersonic flows with a strong swirl. The results
2016
showed that the gas flow was accelerated to supersonic velocity, and created the low pressure and tem-
Accepted 1 October 2016
Available online 3 October 2016
perature conditions for gas removal. Most of the particles collided with the walls or entered into the
liquid-collection space directly, while only a few particles escaped together with the gas flow from the
dry gas outlet. The separation efficiency reached over 80%, when the droplet diameter was more than
Keywords:
Particle motion
1.5 lm. The optimum length of the cyclonic separation section was approximate 16–20 times of the noz-
Supersonic flow zle throat diameter to obtain higher collection efficiency for the supersonic separator with a delta wing.
Swirl Ó 2016 Elsevier B.V. All rights reserved.
Gas separation

1. Introduction
The supersonic swirling separation is a new technology for the
gas processing. Its main structure includes a Laval nozzle, a vortex
generator, a diffuser, and so on [1–3]. This unique technique pos-
sesses a simple and compact tubular structure without any moving
parts to ensure a high reliability. It also does not use any chemicals
to reduce the emissions and protect the environment. Therefore,
the supersonic separator conforms to the development require-
ments of the gas processing including the safety, environmental
protection and energy saving.
The current research on the supersonic separation was mainly
focused on the single phase gas flow with and without a swirl,
and the condensation flow of water vapor. In the single gas flow,
Jassim et al. [4,5] used the numerical simulation method to study
the natural gas behavior in a high pressure supersonic nozzle. Kar-
imi et al. [6] discussed the effect of the flow parameters on the sin-
gle phase gas flow field in a Laval nozzle by the computational
model linked to MATLAB and HYSYS package. Wen et al. studied
the effect of the real gas model on the flow behavior and mass flow
rate through a supersonic separator [7,8]. The swirling flow was
not considered in the above mentioned studies. Malyshkina
⇑ Corresponding author at: School of Petroleum Engineering, Changzhou
University, Changzhou 213016, China.
E-mail address: chuang.wen@cczu.edu.cn (C. Wen).
[9,10] numerically studied the single phase gas flow with a swir-
ling flow in a supersonic separator with an Euler two-
dimensional model, and analyzed the influence of different Mach
number and temperature on the pressure recovery characteristic.
The artificial neural network was employed by Vaziri et al. [11]
to optimize the structure of a supersonic separator and obtained
the effect of the inlet parameters on gas phase flow field. In our
previous studies, the single phase gas flow was discussed in detail
through a supersonic separator with and without a swirl [12–14].
In the condensation flow of water vapor in a supersonic separa-
tor, Alfyorov et al. employed the CFX software to study the forma-
tion process of a liquid droplet in a supersonic flow, and analyzed
the distribution of the liquid droplets along radial and axial sec-
tions of the nozzle [15]. Schooshtari and Shahsavand [16] devel-
oped a new theoretical approach based on the mass transfer
rates to calculate the liquid droplet growth in supersonic condi-
tions for binary mixtures, and they [17] also analyzed the conden-
sation flow of the multi-components gas mixtures with the
nucleation theory. Castier [18] also carried out the numerical sim-
ulations of natural gas flow within a Laval nozzle both in consider-
ation of the single phase flow and the phase equilibrium. But the
swirl was not included in the numerical work as well. Ma et al.
[19] established an Euler two fluid flow model to investigate the
spontaneous condensation flow of water vapor. Then, they [20]
studied the effect of the location, where the external particles were
added, on the droplet condensation by using the heterogeneous

http://dx.doi.org/10.1016/j.seppur.2016.10.002
1383-5866/Ó 2016 Elsevier B.V. All rights reserved.
 Y. Yang, C. Wen / Separation and Purification Technology 174 (2017) 22–28 23

Nomenclature

CD drag coefficient qj heat flux

dij deformation tensor Re relative Reynolds number
dl particle diameter t time
E total energy ug gas velocity
F gas-particle interaction force ul particle velocity
Fa additional force Vl particle volume
Fd drag force dij Kronecker delta
Fs Saffman’s lift force lg gas viscosity
g acceleration of gravity vg gas dynamic viscosity
Ks constant qg gas density
l liquid phase ql particle density
ml particle mass s ¼ ij viscous stress
p pressure

nucleation theory. They proposed that it was reasonable to add In a supersonic separator, the centrifugal force plays a much
external nuclei to increase the size of the condensed droplets. more significant role than the gravity force, therefore, the gravity
The above mentioned studies focused on the single phase gas force can be neglected. The gas-particle interaction force F includes
flow or condensation flow in a supersonic separator, and very little the drag force Fd and the additional force Fa. That is:
attention was paid to the particle flows. Haghighi et al. [21]
F ¼ FD þ Fa ð5Þ
addressed the recent developments of the supersonic separation
technique, and pointed out that the liquid flow was an important The Fd is given as:
issue for this novel gas separation technology and further study
would improve the separation performance significantly. Liu 18lg C D Re
FD ¼ ðug
ul Þ ð6Þ
et al. tried to predict the particle flows in a supersonic separator ql d2l 24
using the Discrete Particle Method [22]. However, the particle
diameter used in their simulation was assumed from 10 lm to where ql and dl are the density and diameter of the liquid particle,
50 lm, which was extremely exaggerated from the typical particle respectively. ug is the fluid phase velocity, lg is the molecular vis-
size of 1.0 ⁄ 10
7–2.0 ⁄ 10
6 m (0.1–2 lm) [23] in the supersonic cosity of the fluid, CD is the drag coefficient. Re is the relative Rey-
separator. nolds number:
In this paper, we employ the Discrete Particle Method to predict qg dl jul
ug j
the particle flows at the supersonic speed. The separation effi- Re ¼ ð7Þ
ciency is evaluated at various particle diameters in our designed
lg
supersonic separator. The effect of cyclonic separation section on The strong shear and swirling flow exist in a supersonic separa-
collection efficiency is predicted using the Computational Fluid tor, and correspondingly the Saffman’s lift force significantly
Dynamics (CFD) technology. affects the particle motion. Hence, the Saffman’s lift force is
involved in our simulation as the additional force to predict the
2. Mathematical model droplet behavior. It can be described as:
pffiffiffiffiffi
2K s mg qg dij
2.1. Gas phase flow FS ¼ ðug
ul Þ ð8Þ
ql dl ðdlk dkl Þ0:25
The mass, momentum and energy conservation equations where FS is the Saffman’s lift force [24], Ks = 2.594, mg is the gas
are adopted for the numerical calculation of the gas phase as dynamic viscosity, dij is the deformation tensor.
follows:

@ 3. Numerical method
ðq ugi Þ ¼ 0 ð1Þ
@xi g
3.1. Geometry and mesh strategy
@
ðq ugi ugj þ pdij
sji Þ ¼ 0 ð2Þ The supersonic separator structure used in this simulation case
@xj g
includes a wet gas inlet, a Laval nozzle, a swirling device, a straight
@
 tube, a cyclonic separation part, a diffuser, a dry gas exit, and a liq-
qg ugj E þ ugj p þ qj
ui sij ¼ 0 ð3Þ uid exit, as shown in Fig. 1. The key sizes for the supersonic sepa-
@xj
rator include the nozzle throat and exit diameters. The nozzle
throat area determines the mass flow rate, while the expansion
2.2. Particle flow characteristics depend on the nozzle outlet area. The diameters
at the nozzle throat and outlet are 12.40 mm and 16.80 mm,
A particle motion in a supersonic separator is described as respectively. The entrance (nozzle inlet) and dry gas exit (diffuser
translation and rotation, and the governing equation for a discrete outlet) diameters of the supersonic separator are 80.40 mm and
liquid particle is: 40.00 mm, respectively. A straight tube is used to form a stable
flow located at the downstream of the Laval nozzle. The delta wing
dul is designed as a swirling device, which is installed between the
ml ¼ ml g
qg gV l þ F ð4Þ
dt straight tube and cyclonic separation part.
24 Y. Yang, C. Wen / Separation and Purification Technology 174 (2017) 22–28
Fig. 1. Structure model of a supersonic separator.
The mesh system is one of the major concerns for the numerical
simulations. To obtain the high-quality mesh, we employ the struc-
tured and unstructured grids for the supersonic separator in this
study. The tetrahedral elements are utilized for the delta wing area
as a result of the complicated geometry, and the hexahedral ele-
ments are performed for all the other parts of the supersonic sep-
arator. The mesh sensitivity test of the supersonic separator is
carried out to obtain the mesh independent results, and approxi-
mately 400,000 elements are performed for the numerical
simulations.
3.2. Numerical implementations
The governing equations are discretized with the finite volume
method, and the velocity and pressure fields are coupled by the
SIMPLE algorithm [25,26]. In a supersonic separator, the delta wing
generates a strong swirling flow at the supersonic speed. The fluid
correspondingly presents the violent turbulence characteristics.
Therefore, the Reynolds stress model is used to predict the super-
sonic flow with a strong swirl, because it can capture the charac-
teristics of anisotropic turbulence [27,28].
For the gas phase, the pressure inlet and pressure outlet condi-
tions are assigned to the entrance and exits of the supersonic sep-
arator. The no-slip and adiabatic conditions are utilized for all of
the solid walls. For the particle flow, the escape boundary condi-
tions are employed to the entrance and dry gas exit of the super-
sonic separator, which means that the particles will not return
the computational domain if they reach these boundaries. The trap
boundary conditions are implemented to the liquid exit of the
supersonic separator and the walls of the cyclonic separation part.
The reflect boundary conditions are used for all other walls.
4. Results and discussion
4.1. Model validation
In the developed computational model, there are two typical
flow structures, namely, the gas supersonic flow with a strong
swirl and the particles in a swirling flow. Consequently, the model
validation is conducted separately step by step. We firstly validate
our proposed model for the gas supersonic flow with a strong swirl,
and then the computational model is validated using the Discrete
Particle Method model for the particles in a swirling flow.
The first validation is performed in another kind of supersonic
separator [29], in which the swirling flow generator locates at
the upstream of the Laval nozzle, as shown in Fig. 2. Both of these
two separators work on the supersonic flows with a strong swirl,
although the specific structures are different. The comparison
between the numerical and experimental results is shown in
Fig. 3, which reflects the effect of the swirling flow on the gas flow
rates through the supersonic separator. In general, we can see that
the numerical results agree well with the experimental data. It
indicates that the proposed mathematical model accurately predict
the strong swirling flow at the supersonic speed.
The second validation is conducted in a cyclone separator, in
which the Discrete Particle Method model is usually employed to
predict the particle behavior in a swirling flow. The configuration
and dimensions of the cyclone separator are borrowed from [30],
as shown in Fig. 4. In the experiments, the working fluid was air
and the particle density was 3320 kg/m3. The inlet gas and particle
velocity were both 20 m/s. The exit tube was open to the air and
the gas pressure at the top of the vortex finder was 1 atm. We used
the same geometry and condition to compare the gas and particle
characteristics in the cyclone separator. Fig. 5 shows the tangential
velocity and collection efficiency from the simulation and experi-
ments. We can see that the computed values agree well with the
experimental data both for the gas and particle flows. It indicates
that the Discrete Particle Method model can be used to predict
the particle motion in a swirling flow.
Therefore, based on the above mentioned two cases, it can be
concluded that our proposed model can be used to predict the par-
ticle flow with a strong swirl in a supersonic separator for the gas
purification.
4.2. Gas flow field
The multi-components natural gas was employed in the case
studies. The detailed compositions can be found in Table 1. The
flow parameters of natural gas at the axis were shown in Fig. 6.
The gas expanded in the Laval nozzle and reached a supersonic
speed. In this simulation case, the gas Mach number was about
2.0 at the nozzle exit. Correspondingly, the gas static pressure
and temperature declined due to the gas expansion, which were
about 500,000 Pa and
80 °C at the nozzle exit, respectively. When
the gas flow reached the delta wing, the swirling motion was gen-
erated. The strong rotating flow was obtained because the change
of the velocity occurred in the supersonic speed. The maximum
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
u2x þ u2y velocity was about 340 m/s. In addition, the gas profiles
presented the sudden changes at z = 0.34 m. For instance, the Mach
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
number and u2x þ u2y velocity both decreased, while the static
pressure and temperature increased. This is because the natural
gas flow touches the surface of the delta wing in the central region.
4.3. Particle trajectory
The liquid particles were released and tracked at the nozzle out-
let in this study. The trajectories of the particle with different sizes
in the strong swirling flow at the supersonic velocity were
described in Fig. 7. The particle diameters, from 0.2 lm to
2.0 lm, were employed to perform the computational cases. We
could see that the particles represented the rotational motion
when they passed through the delta wing, where the strong swir-
ling flow was generated. Most of the particles were trapped by the
 Y. Yang, C. Wen / Separation and Purification Technology 174 (2017) 22–28 25

Fig. 2. Supersonic separators with a swirling device located at the upstream of the Laval nozzle.

Fig. 3. Effect of swirl strength on mass flow rates in a supersonic separator.

Fig. 5. Tangential velocity and collection efficiency in a cyclone separator.

Table 1
Mole composition of natural gas.

Gas composition Mole fraction (%) Gas composition Mole fraction (%)
CH4 91.36 i-C5H12 0.17
C2H6 3.63 n-C5H12 0.16
C3H8 1.44 H2O 0.03
i-C4H10 0.26 CO2 0.45
n-C4H10 0.46 N2 2.04

Fig. 4. Configuration and dimensions of the cyclone separator.
walls of the cyclonic separation part and the liquid exit of the
apparatus. However, a small number of particles still moved with
the gas phase from the dry gas exit.
Fig. 8 shows the single particle trajectory with different sizes in
detail. It could be seen that the particles had three representative
26 Y. Yang, C. Wen / Separation and Purification Technology 174 (2017) 22–28

Fig. 6. Gas flow field in the supersonic separator.

Fig. 7. Particle trajectories in the supersonic separator.

Fig. 8. Single particle trajectory in the supersonic separator.

moving state, that is (a) the particle colliding with the wall, (b) the it was centrifuged and collided with the walls, creating a thin
particle entering into the liquid-collection space, and (c) the particle liquid film, which finally went into the liquid-collection space
passing through the diffuser. For the first kind particle in Fig. 8(a), along the wall. Some particles entered into the liquid-collection
 Y. Yang, C. Wen / Separation and Purification Technology 174 (2017) 22–28 27

space directly without colliding with the wall. For other few parti-
cles, the drag force played a much more significant role than other
forces. Therefore, these particles passed through the diffuser
together with the gas flow, which caused the gas phase entraining
few liquids. In addition, some slip-gas flowed through the liquid
exit.

4.4. Collection efficiency

The collection efficiency is one of the most important technical

parameters for a supersonic separator. The Discrete Particle
Method was employed here to track the droplet motion and ana-
lyze the effect of the droplet size on the collection efficiency in
the different particle diameters. We assumed that the droplets
were trapped, when they reached the wall of the cyclone separa-
tion section. As shown in Fig. 9, the collection efficiency increased
with the increase of the droplet size. When the droplet diameters
were less than 1 lm, the collection efficiency was lower than
65%. It indicated that the smaller size droplets were easier to Fig. 10. Effect of cyclonic separation section on collection efficiency.
escape with gas phase rather than to be separated. The collection
efficiency was more than 80% with the droplet diameter of about
for the particle separation, when the dimensionless length was
1.5 lm, which demonstrated a good separation performance of
more than 16. Correspondingly, the collection efficiency nearly
our newly designed supersonic separator.
reached 90%. In addition, the wall friction could cause the increas-
ing temperature in the supersonic flow, which would decline the
4.5. Effect of cyclonic separation section on collection efficiency collection efficiency as a result of the re-evaporation of the con-
densed droplets. Consequently, the optimal dimensionless length
We employed the dimensionless length to represent the effect of the cyclone separation section was about 16–20 for the super-
of the length of the cyclonic separation section on the collection sonic separator with a delta wing.
efficiency in a supersonic separator. The dimensionless length
was defined as the ratio of the physical length of the cyclonic sep-
5. Conclusions
aration section to the throat diameter of the Laval nozzle. We can
see from Fig. 10 that the collection efficiency of the condensed dro-
Natural gas flow was accelerated to supersonic speed in a Laval
plets rose sharply with the increase of the length of the cyclonic
nozzle, which correspondingly resulted in the low pressure and
separation section while the dimensionless parameter was less
temperature of about 500,000 Pa and
80 °C at the nozzle exit,
16. It indicates that the cyclonic separation section is an important
respectively. The delta wing generated the violent swirling flow
influencing factor for the collection efficiency of the condensed qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
droplets. It not only determines the probability of collision in the high speed conditions, and the maximum u2x þ u2y velocity
between the droplets and the wall surfaces, but also affects the res- was about 340 m/s. The particle trajectory showed that most of the
idence time of the condensed droplets. For example, the collection particles collided with the walls or entered into the liquid-
efficiency was only about 30%, when the dimensionless length was collection space directly. It indicates that these particles can be
4. In this condition, the length of the cyclonic separation section successfully removed from the gas mixture. But there were still a
was very short, which resulted in a very small surface area and a few small particles escaping together with the dry gas flow, which
tiny residence time for the condensed droplets. On the contrary, could lead to the decline of the separation performance. The sepa-
the cyclonic separation section provided enough space and time ration efficiency reached over 80% with the droplet diameter of
about 1.5 lm. We proposed that the length of the cyclonic separa-
tion section was approximate 16–20 times of the throat diameter
of a Laval nozzle for the supersonic separator with a delta wing.

Acknowledgements

This work was supported by the Natural Science Foundation of

Jiangsu Province, China (BK20150270), the General Program of
Natural Science Research Project of Jiangsu Province Universities
and Colleges (15KJB440001), and the Scientific Research Project
of Changzhou (CJ20159034). C. Wen acknowledges the support of
the H.C. Ørsted Fellowship co-funded by Marie Curie Actions at
the Technical University of Denmark, DTU.

References

[1] P.B. Machado, J.G. Monteiro, J.L. Medeiros, H.D. Epsom, O.Q. Araujo, Supersonic
separation in onshore natural gas dew point plant, J. Nat. Gas Sci. Eng. 6 (2012)
43–49.
[2] C. Wen, X. Cao, Y. Yang, W. Li, An unconventional supersonic liquefied
technology for natural gas, Energy Educ. Sci. Technol. Part A Energy Sci. Res. 30
Fig. 9. Collection efficiency in the supersonic separator. (2012) 651–660.
28 Y. Yang, C. Wen / Separation and Purification Technology 174 (2017) 22–28
[3] Y. Yang, C. Wen, S. Wang, Y. Feng, Effect of inlet and outlet flow conditions on
natural gas parameters in supersonic separation process, PLoS ONE 9 (2014)
e110313.
[4] E. Jassim, M.A. Abdi, Y. Muzychka, Computational fluid dynamics study for flow
of natural gas through high-pressure supersonic nozzles: part 1. Real gas
effects and shockwave, Pet. Sci. Technol. 26 (2008) 1757–1772.
[5] E. Jassim, M.A. Abdi, Y. Muzychka, Computational fluid dynamics study for flow
of natural gas through high-pressure supersonic nozzles: part 2. Nozzle
geometry and vorticity, Pet. Sci. Technol. 26 (2008) 1773–1785.
[6] A. Karimi, M.A. Abdi, Selective dehydration of high-pressure natural gas using
supersonic nozzles, Chem. Eng. Process. 48 (2009) 560–568.
[7] Y. Yang, C. Wen, S. Wang, Y. Feng, Numerical simulation of real gas flows in
natural gas supersonic separation processing, J. Nat. Gas Sci. Eng. 21 (2014)
829–836.
[8] C. Wen, X. Cao, Y. Yang, Y. Feng, Prediction of mass flow rate in supersonic
natural gas processing, Oil Gas Sci. Technol. 70 (2015) 1101–1109.
[9] M. Malyshkina, The structure of gas dynamic flow in a supersonic separator of
natural gas, High Temp. 46 (2008) 69–76.
[10] M. Malyshkina, The procedure for investigation of the efficiency of purification
of natural gases in a supersonic separator, High Temp. 48 (2010) 244–250.
[11] B.M. Vaziri, A. Shahsavand, Analysis of supersonic separators geometry using
generalized radial basis function (GRBF) artificial neural networks, J. Nat. Gas
Sci. Eng. 13 (2013) 30–41.
[12] C. Wen, X. Cao, Y. Yang, J. Zhang, Supersonic swirling characteristics of natural
gas in convergent-divergent nozzles, Pet. Sci. 8 (2011) 114–119.
[13] Y. Yang, C. Wen, S. Wang, Y. Feng, P. Witt, The swirling flow structure in
supersonic separators for natural gas dehydration, RSC Adv. 4 (2014) 52967–
52972.
[14] Y. Yang, C. Wen, S. Wang, Y. Feng, Theoretical and numerical analysis on
pressure recovery of supersonic separators for natural gas dehydration, Appl.
Energy 132 (2014) 248–253.
[15] V. Alfyorov, L. Bagirov, L. Dmitriev, V. Feygin, S. Imayev, J.R. Lacey, Supersonic
nozzle efficiently separates natural gas components, Oil Gas J. 103 (2005) 53–
58.
[16] S.H.R. Shooshtari, A. Shahsavand, Reliable prediction of condensation rates for
purification of natural gas via supersonic separators, Sep. Purif. Technol. 116
(2013) 458–470.
[17] S.H.R. Shooshtari, A. Shahsavand, Predictions of wet natural gases
condensation rates via multi-component and multi-phase simulation of
supersonic separators, Korean J. Chem. Eng. 31 (2014) 1845–1858.
[18] M. Castier, Modeling and simulation of supersonic gas separations, J. Nat. Gas
Sci. Eng. 18 (2014) 304–311.
[19] Q.F. Ma, D.P. Hu, J.Z. Jiang, Z.H. Qiu, A turbulent Eulerian multi-fluid model for
homogeneous nucleation of water vapour in transonic flow, Int. J. Comput.
Fluid Dynam. 23 (2009) 221–231.
[20] Q.F. Ma, D.P. Hu, J.Z. Jiang, Z.H. Qiu, Numerical study of the spontaneous
nucleation of self-rotational moist gas in a converging-diverging nozzle, Int. J.
Comput. Fluid Dynam. 24 (2010) 29–36.
[21] M. Haghighi, K.A. Hawboldt, M.A. Abdi, Supersonic gas separators: review of
latest developments, J. Nat. Gas Sci. Eng. 27 (2015) 109–121.
[22] X. Liu, Z. Liu, Y. Li, Investigation on separation efficiency in supersonic
separator with gas-droplet flow based on DPM approach, Sep. Sci. Technol. 49
(2014) 2603–2612.
[23] B. Prast, B. Lammers, M. Betting, CFD for supersonic gas processing, in: Fifth
International Conference on CFD in the Process Industries, CSIRO, Melbourne,
Australia, 13–15 December 2006.
[24] P.G. Saffman, The lift on a small sphere in a slow shear flow, J. Fluid Mech. 22
(1965) 385–400.
[25] S.V. Patankar, D.B. Spalding, A calculation procedure for heat, mass and
momentum transfer in three-dimensional parabolic flows, Int. J. Heat Mass
Transf. 15 (1972) 1787–1806.
[26] M. Wasilewski, Analysis of the effects of temperature and the share of solid
and gas phases on the process of separation in a cyclone suspension preheater,
Sep. Purif. Technol. 168 (2016) 114–123.
[27] K. Elsayed, Design of a novel gas cyclone vortex finder using the adjoint
method, Sep. Purif. Technol. 142 (2015) 274–286.
[28] F.J. de Souza, R. de Vasconcelos Salvo, D. de Moro Martins, Effects of the gas
outlet duct length and shape on the performance of cyclone separators, Sep.
Purif. Technol. 142 (2015) 90–100.
[29] C. Wen, X. Cao, Y. Yang, J. Zhang, Swirling effects on the performance of
supersonic separators for natural gas separation, Chem. Eng. Technol. 34
(2011) 1575–1580.
[30] B. Wang, D.L. Xu, K.W. Chu, A.B. Yu, Numerical study of gas-solid flow in a
cyclone separator, Appl. Math. Model. 30 (2006) 1326–1342.