- Email: [email protected]
Effect of ring shape attached on upper outlet pipe on fine particle classification of gas-cyclone
Separation and Purification Technology 141 (2015) 84–93
Contents lists available at ScienceDirect
Separation and Purification Technology journal homepage: www.elsevier.com/locate/seppur
Effect of ring shape attached on upper outlet pipe on fine particle classification of gas-cyclone Yuki Wakizono a, Tsuyoshi Maeda b, Kunihiro Fukui a, Hideto Yoshida a,⇑ a b
Department of Chemical Engineering, Hiroshima University, 1-4-1, Kagamiyama, Higashi-Hiroshima 739-8527, Japan Mitsubishi Electric Co., Ltd., Production Technology Development Group, 5-1-1, Oufuna, Kamakura city 247-8501, Japan
a r t i c l e
i n f o
Article history: Received 28 June 2014 Received in revised form 31 October 2014 Accepted 13 November 2014 Available online 3 December 2014 Keywords: Gas-cyclone Particle separation Cut size Particle size Sub-micron particles
a b s t r a c t The effect on particle collection efficiency of a ring attached to the upper part of the outlet pipe of a gas cyclone was examined by experiment and CFD simulation. From a visualization experiment using fine bubble foam and CFD simulation, the rotational recirculation region is at maximum near the upper plate and the region decreases to the lower position. For the tapered shape ring of type Dc, the cyclone pressure drop is minimum, because the rotational recirculation region is almost eliminated by use of this special ring. The 50% cut size of the tapered shape ring of type Dc shows a minimum value and maximum particle collection efficiency is obtained. For the tapered shape ring, the optimum wall thickness ratio of the upper part to lower part is 1.47 and maximum particle collection efficiency and minimum pressure drop is simultaneously realized by use of the type Dc ring. The average downward fluid velocity in the upper part of the cyclone for the type Dc ring is greater than that of the case without the ring. Particles tend to go toward the downward direction easily and high particle collection efficiency is achieved for this special ring. The experimental partial separation efficiency qualitatively agreed with the simulation results. Ó 2014 Elsevier B.V. All rights reserved.
1. Introduction Gas-cyclones are widely used in separation or size classification apparatus for gas–solid flows because of their simple structure and low cost. Recently, with improvements in the dimensions of various parts of cyclones, a reasonably high level of precision in the classification of sub-micron order particles has become possible [1–3]. Recent industrial standards for ceramics or metal powders require particle size distributions with a narrow standard deviation, because the physical properties of such classified particles are related to the performance of electrical, magnetic and chemical reactions. In particular, particles with a diameter of 0.1–1 lm are required for various powder handling processes. Armon et al. examined the flow field in the cyclone separator by use of Reynolds-averaged Navier–Stokes equation. The numerical results are compared with the experimental data and good agreement is observed [13]. Khairy et al. examined the effect of cone tip diameter on particle separation performance of cyclone by LES model. They reported cone tip diameter has an insignificant effect on separation performance of cut size of cyclone [14]. However, an accurate particle classification data in size range less than about 1 lm
⇑ Corresponding author. Tel.: +81 82 424 7853; fax: +81 82 424 5494. E-mail address: [email protected] (H. Yoshida). http://dx.doi.org/10.1016/j.seppur.2014.11.028 1383-5866/Ó 2014 Elsevier B.V. All rights reserved.
by cyclone separator is not always reported, then detailed study in sub-micron classification strongly required to improve high performance particulate products. In order to classify a particle cut size in the size range of 0.1–1 lm, conventional forced centrifugal classifiers are not always effective due to particle erosion problems induced by high speed particle impact on the walls, and the increased maintenance costs due to the high rotational movements. Because of the production of strong fluid turbulence induced by a high speed rotating blade, it is difficult to realize sub-micron classification using forced type centrifugal separators. While Iinoya et al. [4] found that it is possible to classify powders even in the 0.4 lm size range using special cyclones, it is generally difficult to change the cut size in a conventional cyclone separator. To solve this problem, Yoshida et al. [1,5] found that it is possible to change the cut size by using a moving circular guide plate at the cyclone inlet or by the use of an additional secondary flow injection method in the upper cylindrical part of the cyclone. In order to carry out high performance sub-micron classification, production of fluid turbulence should be small. Nobukiyo et al. [6] found the recirculation flow region is detected in the upper part of the outlet pipe. In order to reduce this recirculation flow region, the use of a special ring attached to the upper outlet pipe is effective in increasing separation performance. However, the optimum shape of the ring attached to the outlet pipe is not examined.
Y. Wakizono et al. / Separation and Purification Technology 141 (2015) 84–93
85
Nomenclature a, b ring width of upper and lower part (m) C Cunningham’s slip correction factor (–) CD drag coefficient of particle (–) Dp particle diameter (lm) Dp50 50% cut size of cyclone (lm) D cyclone diameter (m) E total collection efficiency of the cyclone (–) fc(Dp), fs(Dp) particle size distributions of coarse and fine sides respectively (–/lm) g gravity acceleration (m/s2) G inlet width ratio (–) K ring width ratio defined by Eq. (2) (–) L ring length (m) mc, ms mass of the collected particles for coarse and fine sides respectively (kg) n number of slit holes (–) p dimensionless pressure (–) Dp cyclone pressure drop (kPa) Q inlet flow rate (l/min) Re (=Du q/l) flow Reynolds number (–)
In this report, a new ring shape is found to improve high particle separation performance with a small circulation region. The fluid flow properties with the new optimum ring shape is examined by CFD simulation and flow visualization technique. Several new facts relevant to the effective application of the industrial gascyclone process are found in this report. 2. Experimental apparatus Fig. 1 shows a schematic diagram of the gas-cyclone used in this study. The cyclone diameter was set at 72 mm and each dimension was determined from experimental data that proved better particle separation performance [4]. In order to enhance rotational fluid velocity in the inlet part and to reduce recirculation flow near the upper outlet pipe, a special ring was attached to the upper outlet pipe. Without the special
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi Repð¼ Dp q ðuz v z Þ2 þ ður v r Þ2 þ ðu# v # Þ2 =l particle Reynolds number (–) S/ general function of production term (–) t time (s) t dimensionless time (–) r ; z dimensionless radial and axial coordinates (–) uz, ur, uh axial, radial and tangential fluid velocities (m/s) u, v, w dimensionless axial, radial and tangential fluid velocities (–) vz, vr, vh axial, radial and tangential particle velocities (m/s) Dg partial separation efficiency (–) / general function of conservation equation (–) h circumferential coordinate (–) C diffusion coefficient (–) m(=1/Re) dimensionless viscosity (–) l fluid viscosity (Pa s) q,qp fluid and particle density (kg/m3) s particle relaxation time (s)
ring, recirculation flow was detected from CFD simulation which is shown in the upper part in Fig. 2 [6]. The effect of ring length (H) and ring thickness (b) on particle separation performance was experimentally investigated by use of several different sized rings. Fig. 3 shows the various rings used in this study. Type S indicates the case without the ring and type A1 to A4 indicate the rings with different ring lengths. Type A to D indicate the rings with different ring thicknesses. The change of cyclone pressure drop by use of the ring was measured by the static pressure difference at the outlet pipe shown in Fig. 1. In order to separate very fine particles, the combined effect of inlet guide packing shown in Fig. 4 and the ring on particle separation efficiency is investigated in this report. The inlet fluid velocity is locally increased by use of the guide packing. In our previous report, the 50% cut size decreases to submicron range by use of the inlet guide packing [7]. However, the
Fig. 1. Experimental apparatus of test cyclone.
Y. Wakizono et al. / Separation and Purification Technology 141 (2015) 84–93
H
86
b
Fig. 2. Ring attached on upper part of outlet pipe.
The performance of the cyclones were examined by use of the partial separation efficiency Dg defined by the following equation, total collection efficiency and pressure drop.
Dg ¼
Fig. 3. Various types of rings attached on outlet pipe.
mc f c DDp mc f c DDp þ ms f s DDp
ð1Þ
In the above equation, mc and ms are the masses of the collected particles on the coarse and fine sides, respectively. The particle size frequency distributions for each size range are indicated by fc and fs. Particle size distributions were measured by the laser-light scattering method (Horiba Co., Ltd. LA-950). The experimental error in the partial separation efficiency was approximately 3%. A ring nozzletype particle disperser (Nisshin Eng. Co., Ltd.) was installed after the screw feeder. The test particle used was Kanto Loam (JIS, Z8901, No. 11) and particle size distribution is shown in Fig. 5. The mass median diameter was 1.8 lm and true density was 2900 kg/m3. The total flow rate, including the secondary flow rate of 100 l/min. was increased from 400 to 1170 l/min. and the powder flow rate was set at approximately 2.0 g/min. 3. Experimental results 3.1. Optimization of ring dimension
Fig. 4. Inlet guide plate and definition of inlet width ratio.
hybrid effect of both inlet guide packing and the ring on particle separation efficiency is expected to increase particle separation performance.
In order to find out the optimum ring dimension, the experiment was carried out by use of rings with various length and width. Fig. 6 shows the experimental results of 50% cut size and pressure drop by use of various ring length. The ring width was kept constant as 7 mm. It is found that 50% cut size and pressure drop indicate a minimum value when the ring length is equal to 40 mm. For the ring length smaller than 40 mm, the fluid recirculation region is not completely eliminated. For ring length larger than 40 mm, the cyclone pressure drop increases due to the increase of rotational fluid velocity and turbulence region. Due to the above reasons, the 50% cut size shows a minimum value with a ring length of 40 mm. Fig. 7 shows the effect of ring width on 50% cut size and pressure drop. In this case, the ring length was kept constant at 40 mm. The 50% cut size and pressure drop shows a minimum value for the ring width of 7 mm. For ring width
Y. Wakizono et al. / Separation and Purification Technology 141 (2015) 84–93
87
Fig. 5. Size distribution of test particles and experimental condition.
Fig. 6. Effect of ring length on 50% cut size and pressure drop.
Fig. 7. Effect of ring width on 50% cut size and pressure drop.
greater than approximately 7 mm, the fluid rotational velocity decreases due to the increased pressure drop, and the 50% cut size increases. From Figs. 6 and 7, the optimum dimensions of the ring is determined as ring length of 40 mm and width of 7 mm. Fig. 8 shows the effect of the guide packing and the ring on partial separation efficiency. For the standard case, 50% cut size is approximately 0.75 lm for the inlet flow rate of 800 l/min. The 50% cut size decreases to approximately 0.7 lm with the ring width of 7 mm. By use of the inlet guide packing and inlet width ratio of 0.6, the 50% cut size changes to 0.65 lm. However, the 50% cut size shows a minimum value of about 0.62 lm by use of the inlet guide packing and ring. Therefore, in order to increase high particle collection efficiency, the use of the ring and inlet packing is considered to be the most effective method.
1
0.8
0.6
0.4
0.2
0 0.3
0.5
0.7
0.9
2
Fig. 8. Effect of ring and guide plate on classification performance.
3.2. Flow visualization around inlet pipe In order to examine the recirculation region around the upper part of the outlet pipe, flow visualization around the outlet pipe
88
Y. Wakizono et al. / Separation and Purification Technology 141 (2015) 84–93
was carried out. Before the experiment, fine bubble foam was coated around the upper part of the outlet pipe as shown in Fig. 9. Next, airflow was introduced to the cyclone for approximately 30 min, then the photographs were taken from the four directions as shown in Fig. 9. It is considered that the region covered with foam indicates the fluid recirculation region. In the fluid recirculation region, the flow velocity is not strong and the foam will not change its position. The experimental results are shown in Fig. 10. The maximum foam region is observed in the photograph from direction 4. Near the upper plate, the foam region is maximum and this foam region decreases with the increase of downward direction. The foam region from directions 1 and 3 is relatively small compared to direction 4. From the visualization experiments of foam bubble, the maximum recirculation region is observed near the upper plate and the region decreases to the lower position. It is then expected that the tapered shape ring will be more effective to decrease the recirculation region. As a result, a new ring of tapered shape where the upper ring width is larger than that of the lower ring is newly considered. Next, the effect of the tapered shape ring on particle separation performance is examined.
Fig. 11. New types of taper rings.
Fig. 9. Experimental condition of flow visualization experiment.
Fig. 10. Fine bubble foam remained on upper outlet pipe after experiment.
Y. Wakizono et al. / Separation and Purification Technology 141 (2015) 84–93
Type S (-) Type Ac (K = 0.5) Type Bc (K = 1) Type Cc (K= 1.29) Type Dc (K= 1.47) Type Ec (K= 1.53)
50% cut size DP50 [μm]
1
0.9
0.8
0.7 0
1
2
3
Pressure drop
4
5
6
P [kPa]
Fig. 12. Relationship between pressure drop and 50% cut size.
89
3.3. New type of taper ring In order to control the fluid recirculation region, the tapered ring is expected to be more effective compared to the constant width ring. Fig. 11 shows the various types of tapered rings used in the experiment. The lower ring width is kept constant at 7 mm, but the upper ring width changes from 3.5 mm to 11 mm. The ring width of the upper and lower part for the type Bc ring is 7 mm. Type Bc is the same ring as type A3 in Fig. 3. Fig. 12 shows the effect of cyclone pressure drop on 50% cut size by use of the various kinds of tapered rings. The 50% cut size decreases with the increase of cyclone pressure drop. Type S, which is the case without the ring, shows maximum 50% cut size. The parameter K indicates the width ratio of the upper side to the lower side and is represented by the following equation.
K¼
b a
ð2Þ
The 50% cut size decreases with the increase of K, but with a K value greater than 1.47, the 50% cut size increases again. The increase of 50% cut size with the increase of as K value is due to the increase of cyclone pressure drop. Because the collision frequency of fluid flow to the upper ring part will be increased for the ring with a K value equal to 1.53, then the increase of 50% cut size is again found for the type Ec ring. Fig. 13 shows the effect of the parameter K on 50% cut size and cyclone pressure drop under a constant inlet flow rate of 800 l/min. The 50% cut size and cyclone pressure drop shows a minimum value for K equal to 1.47. Generally, cyclone pressure drop increases with the decrease of 50% cut size. However, the experimental data shown in Fig. 13 indicate very interesting results. The decrease of 50% cut size with the increase of K value is considered due to the following reasons:
Fig. 13. Effect of parameter K on 50% cut size and pressure drop.
(1) The average rotational fluid velocity in the upper part increases because of the decrease of effective fluid rotational area.
Fig. 14. Photographs of deposited particle on upper cylindrical wall (Q = 800 l/min).
90
Y. Wakizono et al. / Separation and Purification Technology 141 (2015) 84–93
Type S
Type S
Type Bc (K = 1)
Type Dc (K=1.43)
Type Bc (K = 1)
Type Dc (K=1.43)
Fig. 15. Bubble foam and particle remained on upper outlet pipe after experiment.
(2) The decrease of fluid recirculation region by use of the tapered ring. The decrease of cyclone pressure drop is due to reason 2. The decrease of 50% cut size is always accompanied by the increase of cyclone pressure drop. But, by use of the ring proposed in this paper, the 50% cut size decreases without the increase of cyclone pressure drop. The data is clearly shown in Figs. 12 and 13. Fig. 14 shows the particle deposition pattern on the upper cylindrical part for various types of rings. The average slope angle of particle deposition decreases with the increase of K. The average rotational fluid velocity increases with the increase of K, then the average slope angle of particle deposition decreases. However in the case of a K value equal to 1.57, the average slope angle of particle deposition increases again because of the increase of fluid turbulence created by the type Ec ring. Fig. 15 shows the particle deposition pattern for the three cases. After the separation experiment, photographs were taken on the surface of the upper part of the outlet pipe. It is found the minimum particle deposition is observed for the type Dc case. For the type S case without the ring, the amount of particle deposition indicates a maximum value. Results of flow visualization by use of bubble foam are also shown in the upper part. For the type Bc and Dc cases, the fine bubble area remaining after the experiment is very small compared to the case of type S. Therefore, in order to increase particle collection efficiency, the type Dc ring is the most suitable shape to reduce the fluid recirculation region.
4. Numerical simulation 4.1. Simulation method In order to compare particle separation performance with and without the ring on the upper outlet pipe, CFD simulation is effective in determining the difference of fluid flow and particle movement in the upper part of the cyclone. The three-dimensional Navier–Stokes equations and equations of particle motion were
Table 1 @ @ @ 1 @ ðr w/Þ ¼ Equations of fluid and particle motion ðr /Þ þ ðr u/Þ þ ðr v /Þ þ r @h @ z @r @t @ @/ @ @/ @ @/ r C r C C þ þ þ S/ . r @h @z @ z @r @r @h /
C
S/
u
m
r @p @z
v
m m
w
2 mv 2 @w r @p @r þ w r m r @h
mw 2 @v @p @h v w r þ m r @h
Particle’s Eq. of motion 2 2 C Re d r r dh ur ¼ D24 p 1s dv dt dt dt 2 2 C Re dr 2 dh þ r ddt2h ¼ D24 p 1s r dh uh dt dt dt 2 C Re d z ¼ D24 p 1s dz uz þ G dt dt 2 2
qp Dp s ¼ C18 l
G¼g
solved numerically. Table 1 summarizes the basic equations used in the simulation. The direct flow method to solve the Navier– Stokes equations directly was used to calculate the flow field in the cyclone [8]. To calculate the non-linear convection terms, an artificial viscosity coefficient of 1/3 was selected to obtain a stable flow field [9]. The control volume method was used in the simulation [10]. In order to calculate the partial separation efficiency for small particles, the radial grid spacing near the cylindrical, conical and the apex cone walls are designed to be as small as possible. The fluid boundary conditions on the wall, including the apex cone in the dust box, were set to zero. The boundary-fitted curvilinear coordinate system was used in the simulation. In the calculations, the equations of mono-sized particle motions were solved by the Runge–Kutta method [11]. It was assumed that particles are collected when they touch the wall surface and particle repulsion was not taken into consideration. The particle trajectories of mono-sized particles are calculated, however the calculated result does not depend on feed particle size distribution. In the experiment, inlet dust concentration was very small, so the interaction between particles and fluid was ignored in the simulation. Details of the calculation method are reported in our previous paper [12].
Y. Wakizono et al. / Separation and Purification Technology 141 (2015) 84–93
4.2. Simulation results of fluid flow Fig. 16-1 shows the simulation results of rotational fluid flow distribution in the upper part for the case without the ring. The rotational fluid distributions of the three different axial positions are shown and the fluid recirculation region is indicated as a shaded area. The fluid recirculation area is maximum at z/D equal to 0.015, and it decreases with the increase of the z/D value. The simulation results of the fluid recirculation region qualitatively agree with the visualization experiment shown in Fig. 10. Next, the effect of the ring on the fluid recirculation region is examined by CFD simulation. Fig. 16-2 shows the calculated rotational fluid flow distribution with and without the ring on the upper part of the outlet pipe. The fluid recirculation region is almost eliminated with the ring types Bc and Dc. For the type Bc ring, a small recirculation region still remains at z/D equal to 0.5, but the region
91
decreases for the type Dc ring. In order to decrease the fluid recirculation region, the use of the ring type Dc is effective. The simulation results qualitatively agree with the flow visualization experiment by use of small bubble foam in the upper part of the outlet pipe as shown in Fig. 15. Fig. 17-1 shows the axial and radial fluid velocity distribution at h equal to a 90 deg. plane for the three cases. For the type Bc ring, the inlet flow collided with the ring wall surface and the flow changes its direction rapidly. However for the type Dc ring case, the inlet flow changes its direction smoothly and downward velocity in the upper part is relatively large compared to the other cases. Fig. 17-2 shows the enlarged figure of fluid velocity distribution. For the type Dc ring, the downward fluid velocity is approximately two times greater than that of the case without the ring. As a result, particles entered into the cyclone are affected by strong centrifugal and downward force for the type Dc ring case. From CFD
Fig. 16-1. Calculated rotational fluid velocity distribution without ring.
Fig. 16-2. Calculated rotational fluid velocity distribution (Q = 800 l/min).
92
Y. Wakizono et al. / Separation and Purification Technology 141 (2015) 84–93
Fig. 17-1. Calculated fluid velocity distribution of the three types of cyclones.
Fig. 17-2. Inlet fluid velocity distribution of the two types of cyclones (Q = 800 l/min).
Fig. 18. Simulated particle trajectories of the three cases (Dp = 1.5 lm).
Y. Wakizono et al. / Separation and Purification Technology 141 (2015) 84–93
1
0.8
0.6
0.4
0.2
0 0.3
0.5
0.7
0.9
2
Fig. 19. Effect of rings on classification performance.
simulation, high particle collection efficiency is expected to the type Dc ring case.
5. Particle separation performance Fig. 18 shows particle trajectories of 1.5 lm particles for the three cases. Initial particle starting positions are the same for all three cases. Without the ring, particle movement is observed in the upper part of the cylindrical section, whereas for the type Bc and Dc cases, the particle movement in the lower part is detected compared to the type S. Because average downward fluid velocity of the type Dc is greater than that of the type S as shown in Fig. 17-2, particles tend to move in the downward direction easily. The maximum collection efficiency is obtained for the type Dc ring, and minimum collection efficiency of 31% is obtained for the type S. Fig. 19 shows the effect of the ring on classification performance. A 50% cut size of 0.67 lm is obtained for the type Dc ring, but the value changes to about 0.76 lm for the case without the ring. The 50% cut size of the type Bc is greater than that of the type Dc case. The calculated results of partial separation efficiency qualitatively agree with the experimental data. In order to increase particle collection efficiency, the type Dc ring is recommended. It is therefore expected to use the type Dc ring in dry-cyclones for practical processes.
6. Conclusion The effect on particle collection efficiency of a ring attachment on the upper part of the outlet pipe was examined by use of experimental and CFD studies and the following conclusions are obtained.
93
(1) The 50% cut size with the tapered shape ring of type Dc shows a minimum value and the maximum particle collection efficiency. (2) For the tapered shape ring of type Dc, the cyclone pressure drop is minimum, because the rotational recirculation region is almost eliminated in this special ring. (3) From the visualization experiment and CFD simulation, the fluid recirculation region is maximum near the upper plate and the region decreases to the lower position. (4) For the tapered shape ring, the optimum wall thickness ratio of the upper part to the lower part is 1.47 and maximum particle collection efficiency and minimum pressure drop is obtained. (5) Because average downward fluid velocity in the upper part of the type Dc ring is greater than that of the normal case without the ring, particles tend to go toward the downward direction easily and high particle collection efficiency is achieved for this special ring. (6) The 50% cut size for the type Dc ring shows the minimum value and experimental partial separation efficiency qualitatively agreed with the simulation results.
References [1] H. Yoshida, T. Saeki, T. Fujioka, T. Ueda, T. Fuyuki, Fine particle separation by revised type air-cyclone classifier, Kagaku Kogaku Ronbunshu 19 (1993) 476– 482. [2] H. Yoshida, T. Fujioka, K. Hashimoto, K. Nagahashi, Effect of blow-down on fluid flow and particle movement in cyclone classifier, Kagaku Kogaku Ronbunshu 21 (1995) 594–602. [3] H. Yoshida, T. Yamamoto, K. Okanishi, K. Iinoya, Elaborate classification of flyash particles by bench scale air cyclone, Kagaku Kogaku Ronbunshu 23 (1997) 363–370. [4] K. Iinoya, T. Fuyuki, Y. Yamada, H. Hisakuni, E. Sue, Dry sub-micron classification by a small blow-down cyclone, KONA, Powder and Particle 11 (1993) 223–227. [5] H. Yoshida, K. Ono, K. Fukui, The effect of a new method of fluid flow control on sub-micron particle classification in gas-cyclones, Powder Technol. 149 (2005) 139–147. [6] R. Nobukiyo, Y. Inada, K. Fukui, H. Yoshida, Improvement of particle collection performance of dry-cyclone by use of local fluid flow control method, Kagaku Kogaku Ronbunshu 33 (2) (2007) 92–100. [7] N. Takeda, T. Oshitari, K. Fukui, H. Yoshida, Effect of inlet clean air and guide plate on fine particle classification of gas-cyclone, J. Soc. Powder Technol., Japan 51 (2014) 614–622. [8] T. Kawamura, K. Kuwabara, Computation of high-Reynolds Number flow around a circular cylinder with surface roughness, AIAA 22nd Aerospace Sciences Meeting, 84-0340, Nevada, 1–11, 1984. [9] T. Kawamura, Numerical study of high Reynolds Number flow around a circular cylinder, Ph.D. Thesis, Tokyo Univ., Japan, 1980. [10] S.V. Patankar, Numerical Heat Transfer and Fluid Flow, Hemisphere Pub., Washington, USA, 1980, pp. 113–137. [11] R. Toei, Practice of chemical engineering programming, Soc. Chem. Eng., Japan Ed. 148–154 (1976). [12] H. Yoshida, T. Saeki, K. Hashimoto, T. Fuyuki, Size classification of sub-micron powder by air cyclone and three-dimensional analysis, J. Chem. Eng., Japan 24 (1991) 640–647. [13] R. Araman, S. Mehrzad, F. Meisam, E. Reza, Numerical simulation and optimization of fluid flow in cyclone vortex finder, Chem. Eng. Proc. 47 (2008) 128–137. [14] E. Khairy, L. Chris, Numerical modeling of flow field and performance in cyclones of different core-tip dimensions, Comput. Fluids 51 (2011) 48–59.
Contents lists available at ScienceDirect
Separation and Purification Technology journal homepage: www.elsevier.com/locate/seppur
Effect of ring shape attached on upper outlet pipe on fine particle classification of gas-cyclone Yuki Wakizono a, Tsuyoshi Maeda b, Kunihiro Fukui a, Hideto Yoshida a,⇑ a b
Department of Chemical Engineering, Hiroshima University, 1-4-1, Kagamiyama, Higashi-Hiroshima 739-8527, Japan Mitsubishi Electric Co., Ltd., Production Technology Development Group, 5-1-1, Oufuna, Kamakura city 247-8501, Japan
a r t i c l e
i n f o
Article history: Received 28 June 2014 Received in revised form 31 October 2014 Accepted 13 November 2014 Available online 3 December 2014 Keywords: Gas-cyclone Particle separation Cut size Particle size Sub-micron particles
a b s t r a c t The effect on particle collection efficiency of a ring attached to the upper part of the outlet pipe of a gas cyclone was examined by experiment and CFD simulation. From a visualization experiment using fine bubble foam and CFD simulation, the rotational recirculation region is at maximum near the upper plate and the region decreases to the lower position. For the tapered shape ring of type Dc, the cyclone pressure drop is minimum, because the rotational recirculation region is almost eliminated by use of this special ring. The 50% cut size of the tapered shape ring of type Dc shows a minimum value and maximum particle collection efficiency is obtained. For the tapered shape ring, the optimum wall thickness ratio of the upper part to lower part is 1.47 and maximum particle collection efficiency and minimum pressure drop is simultaneously realized by use of the type Dc ring. The average downward fluid velocity in the upper part of the cyclone for the type Dc ring is greater than that of the case without the ring. Particles tend to go toward the downward direction easily and high particle collection efficiency is achieved for this special ring. The experimental partial separation efficiency qualitatively agreed with the simulation results. Ó 2014 Elsevier B.V. All rights reserved.
1. Introduction Gas-cyclones are widely used in separation or size classification apparatus for gas–solid flows because of their simple structure and low cost. Recently, with improvements in the dimensions of various parts of cyclones, a reasonably high level of precision in the classification of sub-micron order particles has become possible [1–3]. Recent industrial standards for ceramics or metal powders require particle size distributions with a narrow standard deviation, because the physical properties of such classified particles are related to the performance of electrical, magnetic and chemical reactions. In particular, particles with a diameter of 0.1–1 lm are required for various powder handling processes. Armon et al. examined the flow field in the cyclone separator by use of Reynolds-averaged Navier–Stokes equation. The numerical results are compared with the experimental data and good agreement is observed [13]. Khairy et al. examined the effect of cone tip diameter on particle separation performance of cyclone by LES model. They reported cone tip diameter has an insignificant effect on separation performance of cut size of cyclone [14]. However, an accurate particle classification data in size range less than about 1 lm
⇑ Corresponding author. Tel.: +81 82 424 7853; fax: +81 82 424 5494. E-mail address: [email protected] (H. Yoshida). http://dx.doi.org/10.1016/j.seppur.2014.11.028 1383-5866/Ó 2014 Elsevier B.V. All rights reserved.
by cyclone separator is not always reported, then detailed study in sub-micron classification strongly required to improve high performance particulate products. In order to classify a particle cut size in the size range of 0.1–1 lm, conventional forced centrifugal classifiers are not always effective due to particle erosion problems induced by high speed particle impact on the walls, and the increased maintenance costs due to the high rotational movements. Because of the production of strong fluid turbulence induced by a high speed rotating blade, it is difficult to realize sub-micron classification using forced type centrifugal separators. While Iinoya et al. [4] found that it is possible to classify powders even in the 0.4 lm size range using special cyclones, it is generally difficult to change the cut size in a conventional cyclone separator. To solve this problem, Yoshida et al. [1,5] found that it is possible to change the cut size by using a moving circular guide plate at the cyclone inlet or by the use of an additional secondary flow injection method in the upper cylindrical part of the cyclone. In order to carry out high performance sub-micron classification, production of fluid turbulence should be small. Nobukiyo et al. [6] found the recirculation flow region is detected in the upper part of the outlet pipe. In order to reduce this recirculation flow region, the use of a special ring attached to the upper outlet pipe is effective in increasing separation performance. However, the optimum shape of the ring attached to the outlet pipe is not examined.
Y. Wakizono et al. / Separation and Purification Technology 141 (2015) 84–93
85
Nomenclature a, b ring width of upper and lower part (m) C Cunningham’s slip correction factor (–) CD drag coefficient of particle (–) Dp particle diameter (lm) Dp50 50% cut size of cyclone (lm) D cyclone diameter (m) E total collection efficiency of the cyclone (–) fc(Dp), fs(Dp) particle size distributions of coarse and fine sides respectively (–/lm) g gravity acceleration (m/s2) G inlet width ratio (–) K ring width ratio defined by Eq. (2) (–) L ring length (m) mc, ms mass of the collected particles for coarse and fine sides respectively (kg) n number of slit holes (–) p dimensionless pressure (–) Dp cyclone pressure drop (kPa) Q inlet flow rate (l/min) Re (=Du q/l) flow Reynolds number (–)
In this report, a new ring shape is found to improve high particle separation performance with a small circulation region. The fluid flow properties with the new optimum ring shape is examined by CFD simulation and flow visualization technique. Several new facts relevant to the effective application of the industrial gascyclone process are found in this report. 2. Experimental apparatus Fig. 1 shows a schematic diagram of the gas-cyclone used in this study. The cyclone diameter was set at 72 mm and each dimension was determined from experimental data that proved better particle separation performance [4]. In order to enhance rotational fluid velocity in the inlet part and to reduce recirculation flow near the upper outlet pipe, a special ring was attached to the upper outlet pipe. Without the special
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi Repð¼ Dp q ðuz v z Þ2 þ ður v r Þ2 þ ðu# v # Þ2 =l particle Reynolds number (–) S/ general function of production term (–) t time (s) t dimensionless time (–) r ; z dimensionless radial and axial coordinates (–) uz, ur, uh axial, radial and tangential fluid velocities (m/s) u, v, w dimensionless axial, radial and tangential fluid velocities (–) vz, vr, vh axial, radial and tangential particle velocities (m/s) Dg partial separation efficiency (–) / general function of conservation equation (–) h circumferential coordinate (–) C diffusion coefficient (–) m(=1/Re) dimensionless viscosity (–) l fluid viscosity (Pa s) q,qp fluid and particle density (kg/m3) s particle relaxation time (s)
ring, recirculation flow was detected from CFD simulation which is shown in the upper part in Fig. 2 [6]. The effect of ring length (H) and ring thickness (b) on particle separation performance was experimentally investigated by use of several different sized rings. Fig. 3 shows the various rings used in this study. Type S indicates the case without the ring and type A1 to A4 indicate the rings with different ring lengths. Type A to D indicate the rings with different ring thicknesses. The change of cyclone pressure drop by use of the ring was measured by the static pressure difference at the outlet pipe shown in Fig. 1. In order to separate very fine particles, the combined effect of inlet guide packing shown in Fig. 4 and the ring on particle separation efficiency is investigated in this report. The inlet fluid velocity is locally increased by use of the guide packing. In our previous report, the 50% cut size decreases to submicron range by use of the inlet guide packing [7]. However, the
Fig. 1. Experimental apparatus of test cyclone.
Y. Wakizono et al. / Separation and Purification Technology 141 (2015) 84–93
H
86
b
Fig. 2. Ring attached on upper part of outlet pipe.
The performance of the cyclones were examined by use of the partial separation efficiency Dg defined by the following equation, total collection efficiency and pressure drop.
Dg ¼
Fig. 3. Various types of rings attached on outlet pipe.
mc f c DDp mc f c DDp þ ms f s DDp
ð1Þ
In the above equation, mc and ms are the masses of the collected particles on the coarse and fine sides, respectively. The particle size frequency distributions for each size range are indicated by fc and fs. Particle size distributions were measured by the laser-light scattering method (Horiba Co., Ltd. LA-950). The experimental error in the partial separation efficiency was approximately 3%. A ring nozzletype particle disperser (Nisshin Eng. Co., Ltd.) was installed after the screw feeder. The test particle used was Kanto Loam (JIS, Z8901, No. 11) and particle size distribution is shown in Fig. 5. The mass median diameter was 1.8 lm and true density was 2900 kg/m3. The total flow rate, including the secondary flow rate of 100 l/min. was increased from 400 to 1170 l/min. and the powder flow rate was set at approximately 2.0 g/min. 3. Experimental results 3.1. Optimization of ring dimension
Fig. 4. Inlet guide plate and definition of inlet width ratio.
hybrid effect of both inlet guide packing and the ring on particle separation efficiency is expected to increase particle separation performance.
In order to find out the optimum ring dimension, the experiment was carried out by use of rings with various length and width. Fig. 6 shows the experimental results of 50% cut size and pressure drop by use of various ring length. The ring width was kept constant as 7 mm. It is found that 50% cut size and pressure drop indicate a minimum value when the ring length is equal to 40 mm. For the ring length smaller than 40 mm, the fluid recirculation region is not completely eliminated. For ring length larger than 40 mm, the cyclone pressure drop increases due to the increase of rotational fluid velocity and turbulence region. Due to the above reasons, the 50% cut size shows a minimum value with a ring length of 40 mm. Fig. 7 shows the effect of ring width on 50% cut size and pressure drop. In this case, the ring length was kept constant at 40 mm. The 50% cut size and pressure drop shows a minimum value for the ring width of 7 mm. For ring width
Y. Wakizono et al. / Separation and Purification Technology 141 (2015) 84–93
87
Fig. 5. Size distribution of test particles and experimental condition.
Fig. 6. Effect of ring length on 50% cut size and pressure drop.
Fig. 7. Effect of ring width on 50% cut size and pressure drop.
greater than approximately 7 mm, the fluid rotational velocity decreases due to the increased pressure drop, and the 50% cut size increases. From Figs. 6 and 7, the optimum dimensions of the ring is determined as ring length of 40 mm and width of 7 mm. Fig. 8 shows the effect of the guide packing and the ring on partial separation efficiency. For the standard case, 50% cut size is approximately 0.75 lm for the inlet flow rate of 800 l/min. The 50% cut size decreases to approximately 0.7 lm with the ring width of 7 mm. By use of the inlet guide packing and inlet width ratio of 0.6, the 50% cut size changes to 0.65 lm. However, the 50% cut size shows a minimum value of about 0.62 lm by use of the inlet guide packing and ring. Therefore, in order to increase high particle collection efficiency, the use of the ring and inlet packing is considered to be the most effective method.
1
0.8
0.6
0.4
0.2
0 0.3
0.5
0.7
0.9
2
Fig. 8. Effect of ring and guide plate on classification performance.
3.2. Flow visualization around inlet pipe In order to examine the recirculation region around the upper part of the outlet pipe, flow visualization around the outlet pipe
88
Y. Wakizono et al. / Separation and Purification Technology 141 (2015) 84–93
was carried out. Before the experiment, fine bubble foam was coated around the upper part of the outlet pipe as shown in Fig. 9. Next, airflow was introduced to the cyclone for approximately 30 min, then the photographs were taken from the four directions as shown in Fig. 9. It is considered that the region covered with foam indicates the fluid recirculation region. In the fluid recirculation region, the flow velocity is not strong and the foam will not change its position. The experimental results are shown in Fig. 10. The maximum foam region is observed in the photograph from direction 4. Near the upper plate, the foam region is maximum and this foam region decreases with the increase of downward direction. The foam region from directions 1 and 3 is relatively small compared to direction 4. From the visualization experiments of foam bubble, the maximum recirculation region is observed near the upper plate and the region decreases to the lower position. It is then expected that the tapered shape ring will be more effective to decrease the recirculation region. As a result, a new ring of tapered shape where the upper ring width is larger than that of the lower ring is newly considered. Next, the effect of the tapered shape ring on particle separation performance is examined.
Fig. 11. New types of taper rings.
Fig. 9. Experimental condition of flow visualization experiment.
Fig. 10. Fine bubble foam remained on upper outlet pipe after experiment.
Y. Wakizono et al. / Separation and Purification Technology 141 (2015) 84–93
Type S (-) Type Ac (K = 0.5) Type Bc (K = 1) Type Cc (K= 1.29) Type Dc (K= 1.47) Type Ec (K= 1.53)
50% cut size DP50 [μm]
1
0.9
0.8
0.7 0
1
2
3
Pressure drop
4
5
6
P [kPa]
Fig. 12. Relationship between pressure drop and 50% cut size.
89
3.3. New type of taper ring In order to control the fluid recirculation region, the tapered ring is expected to be more effective compared to the constant width ring. Fig. 11 shows the various types of tapered rings used in the experiment. The lower ring width is kept constant at 7 mm, but the upper ring width changes from 3.5 mm to 11 mm. The ring width of the upper and lower part for the type Bc ring is 7 mm. Type Bc is the same ring as type A3 in Fig. 3. Fig. 12 shows the effect of cyclone pressure drop on 50% cut size by use of the various kinds of tapered rings. The 50% cut size decreases with the increase of cyclone pressure drop. Type S, which is the case without the ring, shows maximum 50% cut size. The parameter K indicates the width ratio of the upper side to the lower side and is represented by the following equation.
K¼
b a
ð2Þ
The 50% cut size decreases with the increase of K, but with a K value greater than 1.47, the 50% cut size increases again. The increase of 50% cut size with the increase of as K value is due to the increase of cyclone pressure drop. Because the collision frequency of fluid flow to the upper ring part will be increased for the ring with a K value equal to 1.53, then the increase of 50% cut size is again found for the type Ec ring. Fig. 13 shows the effect of the parameter K on 50% cut size and cyclone pressure drop under a constant inlet flow rate of 800 l/min. The 50% cut size and cyclone pressure drop shows a minimum value for K equal to 1.47. Generally, cyclone pressure drop increases with the decrease of 50% cut size. However, the experimental data shown in Fig. 13 indicate very interesting results. The decrease of 50% cut size with the increase of K value is considered due to the following reasons:
Fig. 13. Effect of parameter K on 50% cut size and pressure drop.
(1) The average rotational fluid velocity in the upper part increases because of the decrease of effective fluid rotational area.
Fig. 14. Photographs of deposited particle on upper cylindrical wall (Q = 800 l/min).
90
Y. Wakizono et al. / Separation and Purification Technology 141 (2015) 84–93
Type S
Type S
Type Bc (K = 1)
Type Dc (K=1.43)
Type Bc (K = 1)
Type Dc (K=1.43)
Fig. 15. Bubble foam and particle remained on upper outlet pipe after experiment.
(2) The decrease of fluid recirculation region by use of the tapered ring. The decrease of cyclone pressure drop is due to reason 2. The decrease of 50% cut size is always accompanied by the increase of cyclone pressure drop. But, by use of the ring proposed in this paper, the 50% cut size decreases without the increase of cyclone pressure drop. The data is clearly shown in Figs. 12 and 13. Fig. 14 shows the particle deposition pattern on the upper cylindrical part for various types of rings. The average slope angle of particle deposition decreases with the increase of K. The average rotational fluid velocity increases with the increase of K, then the average slope angle of particle deposition decreases. However in the case of a K value equal to 1.57, the average slope angle of particle deposition increases again because of the increase of fluid turbulence created by the type Ec ring. Fig. 15 shows the particle deposition pattern for the three cases. After the separation experiment, photographs were taken on the surface of the upper part of the outlet pipe. It is found the minimum particle deposition is observed for the type Dc case. For the type S case without the ring, the amount of particle deposition indicates a maximum value. Results of flow visualization by use of bubble foam are also shown in the upper part. For the type Bc and Dc cases, the fine bubble area remaining after the experiment is very small compared to the case of type S. Therefore, in order to increase particle collection efficiency, the type Dc ring is the most suitable shape to reduce the fluid recirculation region.
4. Numerical simulation 4.1. Simulation method In order to compare particle separation performance with and without the ring on the upper outlet pipe, CFD simulation is effective in determining the difference of fluid flow and particle movement in the upper part of the cyclone. The three-dimensional Navier–Stokes equations and equations of particle motion were
Table 1 @ @ @ 1 @ ðr w/Þ ¼ Equations of fluid and particle motion ðr /Þ þ ðr u/Þ þ ðr v /Þ þ r @h @ z @r @t @ @/ @ @/ @ @/ r C r C C þ þ þ S/ . r @h @z @ z @r @r @h /
C
S/
u
m
r @p @z
v
m m
w
2 mv 2 @w r @p @r þ w r m r @h
mw 2 @v @p @h v w r þ m r @h
Particle’s Eq. of motion 2 2 C Re d r r dh ur ¼ D24 p 1s dv dt dt dt 2 2 C Re dr 2 dh þ r ddt2h ¼ D24 p 1s r dh uh dt dt dt 2 C Re d z ¼ D24 p 1s dz uz þ G dt dt 2 2
qp Dp s ¼ C18 l
G¼g
solved numerically. Table 1 summarizes the basic equations used in the simulation. The direct flow method to solve the Navier– Stokes equations directly was used to calculate the flow field in the cyclone [8]. To calculate the non-linear convection terms, an artificial viscosity coefficient of 1/3 was selected to obtain a stable flow field [9]. The control volume method was used in the simulation [10]. In order to calculate the partial separation efficiency for small particles, the radial grid spacing near the cylindrical, conical and the apex cone walls are designed to be as small as possible. The fluid boundary conditions on the wall, including the apex cone in the dust box, were set to zero. The boundary-fitted curvilinear coordinate system was used in the simulation. In the calculations, the equations of mono-sized particle motions were solved by the Runge–Kutta method [11]. It was assumed that particles are collected when they touch the wall surface and particle repulsion was not taken into consideration. The particle trajectories of mono-sized particles are calculated, however the calculated result does not depend on feed particle size distribution. In the experiment, inlet dust concentration was very small, so the interaction between particles and fluid was ignored in the simulation. Details of the calculation method are reported in our previous paper [12].
Y. Wakizono et al. / Separation and Purification Technology 141 (2015) 84–93
4.2. Simulation results of fluid flow Fig. 16-1 shows the simulation results of rotational fluid flow distribution in the upper part for the case without the ring. The rotational fluid distributions of the three different axial positions are shown and the fluid recirculation region is indicated as a shaded area. The fluid recirculation area is maximum at z/D equal to 0.015, and it decreases with the increase of the z/D value. The simulation results of the fluid recirculation region qualitatively agree with the visualization experiment shown in Fig. 10. Next, the effect of the ring on the fluid recirculation region is examined by CFD simulation. Fig. 16-2 shows the calculated rotational fluid flow distribution with and without the ring on the upper part of the outlet pipe. The fluid recirculation region is almost eliminated with the ring types Bc and Dc. For the type Bc ring, a small recirculation region still remains at z/D equal to 0.5, but the region
91
decreases for the type Dc ring. In order to decrease the fluid recirculation region, the use of the ring type Dc is effective. The simulation results qualitatively agree with the flow visualization experiment by use of small bubble foam in the upper part of the outlet pipe as shown in Fig. 15. Fig. 17-1 shows the axial and radial fluid velocity distribution at h equal to a 90 deg. plane for the three cases. For the type Bc ring, the inlet flow collided with the ring wall surface and the flow changes its direction rapidly. However for the type Dc ring case, the inlet flow changes its direction smoothly and downward velocity in the upper part is relatively large compared to the other cases. Fig. 17-2 shows the enlarged figure of fluid velocity distribution. For the type Dc ring, the downward fluid velocity is approximately two times greater than that of the case without the ring. As a result, particles entered into the cyclone are affected by strong centrifugal and downward force for the type Dc ring case. From CFD
Fig. 16-1. Calculated rotational fluid velocity distribution without ring.
Fig. 16-2. Calculated rotational fluid velocity distribution (Q = 800 l/min).
92
Y. Wakizono et al. / Separation and Purification Technology 141 (2015) 84–93
Fig. 17-1. Calculated fluid velocity distribution of the three types of cyclones.
Fig. 17-2. Inlet fluid velocity distribution of the two types of cyclones (Q = 800 l/min).
Fig. 18. Simulated particle trajectories of the three cases (Dp = 1.5 lm).
Y. Wakizono et al. / Separation and Purification Technology 141 (2015) 84–93
1
0.8
0.6
0.4
0.2
0 0.3
0.5
0.7
0.9
2
Fig. 19. Effect of rings on classification performance.
simulation, high particle collection efficiency is expected to the type Dc ring case.
5. Particle separation performance Fig. 18 shows particle trajectories of 1.5 lm particles for the three cases. Initial particle starting positions are the same for all three cases. Without the ring, particle movement is observed in the upper part of the cylindrical section, whereas for the type Bc and Dc cases, the particle movement in the lower part is detected compared to the type S. Because average downward fluid velocity of the type Dc is greater than that of the type S as shown in Fig. 17-2, particles tend to move in the downward direction easily. The maximum collection efficiency is obtained for the type Dc ring, and minimum collection efficiency of 31% is obtained for the type S. Fig. 19 shows the effect of the ring on classification performance. A 50% cut size of 0.67 lm is obtained for the type Dc ring, but the value changes to about 0.76 lm for the case without the ring. The 50% cut size of the type Bc is greater than that of the type Dc case. The calculated results of partial separation efficiency qualitatively agree with the experimental data. In order to increase particle collection efficiency, the type Dc ring is recommended. It is therefore expected to use the type Dc ring in dry-cyclones for practical processes.
6. Conclusion The effect on particle collection efficiency of a ring attachment on the upper part of the outlet pipe was examined by use of experimental and CFD studies and the following conclusions are obtained.
93
(1) The 50% cut size with the tapered shape ring of type Dc shows a minimum value and the maximum particle collection efficiency. (2) For the tapered shape ring of type Dc, the cyclone pressure drop is minimum, because the rotational recirculation region is almost eliminated in this special ring. (3) From the visualization experiment and CFD simulation, the fluid recirculation region is maximum near the upper plate and the region decreases to the lower position. (4) For the tapered shape ring, the optimum wall thickness ratio of the upper part to the lower part is 1.47 and maximum particle collection efficiency and minimum pressure drop is obtained. (5) Because average downward fluid velocity in the upper part of the type Dc ring is greater than that of the normal case without the ring, particles tend to go toward the downward direction easily and high particle collection efficiency is achieved for this special ring. (6) The 50% cut size for the type Dc ring shows the minimum value and experimental partial separation efficiency qualitatively agreed with the simulation results.
References [1] H. Yoshida, T. Saeki, T. Fujioka, T. Ueda, T. Fuyuki, Fine particle separation by revised type air-cyclone classifier, Kagaku Kogaku Ronbunshu 19 (1993) 476– 482. [2] H. Yoshida, T. Fujioka, K. Hashimoto, K. Nagahashi, Effect of blow-down on fluid flow and particle movement in cyclone classifier, Kagaku Kogaku Ronbunshu 21 (1995) 594–602. [3] H. Yoshida, T. Yamamoto, K. Okanishi, K. Iinoya, Elaborate classification of flyash particles by bench scale air cyclone, Kagaku Kogaku Ronbunshu 23 (1997) 363–370. [4] K. Iinoya, T. Fuyuki, Y. Yamada, H. Hisakuni, E. Sue, Dry sub-micron classification by a small blow-down cyclone, KONA, Powder and Particle 11 (1993) 223–227. [5] H. Yoshida, K. Ono, K. Fukui, The effect of a new method of fluid flow control on sub-micron particle classification in gas-cyclones, Powder Technol. 149 (2005) 139–147. [6] R. Nobukiyo, Y. Inada, K. Fukui, H. Yoshida, Improvement of particle collection performance of dry-cyclone by use of local fluid flow control method, Kagaku Kogaku Ronbunshu 33 (2) (2007) 92–100. [7] N. Takeda, T. Oshitari, K. Fukui, H. Yoshida, Effect of inlet clean air and guide plate on fine particle classification of gas-cyclone, J. Soc. Powder Technol., Japan 51 (2014) 614–622. [8] T. Kawamura, K. Kuwabara, Computation of high-Reynolds Number flow around a circular cylinder with surface roughness, AIAA 22nd Aerospace Sciences Meeting, 84-0340, Nevada, 1–11, 1984. [9] T. Kawamura, Numerical study of high Reynolds Number flow around a circular cylinder, Ph.D. Thesis, Tokyo Univ., Japan, 1980. [10] S.V. Patankar, Numerical Heat Transfer and Fluid Flow, Hemisphere Pub., Washington, USA, 1980, pp. 113–137. [11] R. Toei, Practice of chemical engineering programming, Soc. Chem. Eng., Japan Ed. 148–154 (1976). [12] H. Yoshida, T. Saeki, K. Hashimoto, T. Fuyuki, Size classification of sub-micron powder by air cyclone and three-dimensional analysis, J. Chem. Eng., Japan 24 (1991) 640–647. [13] R. Araman, S. Mehrzad, F. Meisam, E. Reza, Numerical simulation and optimization of fluid flow in cyclone vortex finder, Chem. Eng. Proc. 47 (2008) 128–137. [14] E. Khairy, L. Chris, Numerical modeling of flow field and performance in cyclones of different core-tip dimensions, Comput. Fluids 51 (2011) 48–59.