- Email: [email protected]
Development of a single cyclone separator with three stages for size-selective sampling of particles
Journal of Aerosol Science 89 (2015) 18–25
Contents lists available at ScienceDirect
Journal of Aerosol Science journal homepage: www.elsevier.com/locate/jaerosci
Development of a single cyclone separator with three stages for size-selective sampling of particles Chun-Woo Park, Dae-Hyun Song, Se-Jin Yook n School of Mechanical Engineering, Hanyang University, Seoul 133-791, Republic of Korea
a r t i c l e in f o
abstract
Article history: Received 27 December 2014 Received in revised form 23 May 2015 Accepted 1 July 2015 Available online 9 July 2015
In this study, a single cyclone separator composed of three cylinders of different diameters and one vortex finder was developed for size-selective sampling of particles. This cyclone separator had three particle traps, that is, dust bins, each of which was connected to each of the cylinders. A single cyclone separator with three stages was designed and manufactured. The numerically predicted sampling fractions agreed well with the experimental data. From both numerical and experimental results, three distinct sampling fraction curves with different geometric mean diameters (GMD) appeared and the sizeselective sampling of particles was possible using the single cyclone separator with three stages. Then, the flow rate of aerosol introduced to the cyclone separator was varied to change the GMD of particles collected on the bottom of each particle trap. & 2015 Elsevier Ltd. All rights reserved.
Keywords: Cyclone separator Size-selective sampling Sampling fraction GMD GSD
1. Introduction Size-selective sampling is essential for characterizing aerosol particles according to their size. Cascade impactors such as QCM impactor (Hering, 1987), MOUDI (Marple, Rubow, & Behm, 1991), and ELPI (Marjamäki, Keskinen, Chen, & Pui, 2000), are commonly used for size-selective sampling of particles. The cascade impactors, however, are usually operated at a relatively low flow rate and can have particle bounce and re-entrainment problem. If the impaction plates are coated with sticky material like grease to reduce particle bounce, the collected particles can be contaminated by the coating material. Therefore, it is needed to develop a device which is appropriate for size-selective sampling at a relatively high flow rate without using coating materials. Cyclone separators are widely used in many fields due to their simple geometry and high sampling flow rate. Griffiths and Boysan (1996) successfully modeled the performance of three types of cyclone samplers using a computational fluid dynamics (CFD) package. Lee, Yang, and Lee (2006) investigated the effect of the cylinder shape on the collection efficiency of a long-coned cyclone. Chuah, Gimbun, and Choong (2006) and Xiang, Park, and Lee (2001) evaluated the influence of the cone dimensions on the cyclone performance. Ray, Luning, Hoffmann, Plomp, and Beumer (1997) installed a cylindrical annular shell on top of the vortex finder to trap dusts escaping from an industrial cyclone. Hsiao, Huang, Hsu, Chen, and Chang (2015) investigated the influences of geometric configurations on cyclone performance and proposed the optimal ranges for the geometric dimensional ratios. Dietz (1982) and Lim, Lee, and Kuhlman (2001) employed the electrical force to improve the collection efficiency of cyclone separators. So far, a lot of studies have been conducted on the enhancement of
n
Corresponding author. Tel.: þ82 2 2220 0422; fax: þ 82 2 2220 2299. E-mail address: [email protected] (S.-J. Yook).
http://dx.doi.org/10.1016/j.jaerosci.2015.07.001 0021-8502/& 2015 Elsevier Ltd. All rights reserved.
C.-W. Park et al. / Journal of Aerosol Science 89 (2015) 18–25
19
Fig. 1. Schematic of the single cyclone separator with three stages, developed in this study: (a) perspective view and (b) front view.
the cyclone performance. As a result, it is possible to efficiently separate particles from air stream using cyclone separators. However, if a single cyclone separator is used, only the particles larger than a cut-off size are collected and the size-selective sampling cannot be achieved. Multiple cyclone separators with different cut-off sizes can be connected in series for size-selective sampling, similar to the cascade impactors. McFarland, Bertch, Fisher, and Prentice (1977) employed an apparatus consisting of a series of two cyclones and a centripeter, and fractionated fly ash from stack gas to have volume median diameters of 20 μm, 6.3 μm, 3.2 μm, and 2.2 μm with geometric standard deviations of about 1.8. Smith, Wilson, and Harris (1979) developed a five-stage cyclone system for in situ sampling with cut points of 5.4 μm, 2.1 μm, 1.4 μm, 0.65 μm, and 0.32 μm, by connecting five cyclone separators in series. Hsiao, Chen, Li, Greenberg, and Street (2010) developed a multi-stage cyclone system consisting
20
C.-W. Park et al. / Journal of Aerosol Science 89 (2015) 18–25
Fig. 2. Schematic of the experimental setup.
of an impaction inlet and five axial flow cyclone stages in series, and the cut-off sizes of the axial flow cyclones were 0.97 μm, 0.55 μm, 0.255 μm, 0.109 μm, and 0.04 μm. If multiple cyclone separators in series are used for size-selective sampling, the system can be complex and bulky, and additional flow control may be required to have a certain cut-off size depending on the operating flow rate of a cyclone separator. The purpose of this study is to develop a single cyclone separator with three stages for size-selective sampling. A numerical approach is established to estimate the fraction of particles collected in the particle trap of the cyclone separator. Experiments are conducted to validate the numerical approach. Finally, the effect of the flow rate of aerosol injected into the cyclone separator on the size distribution of the collected particles is numerically investigated.
2. Numerical Fig. 1 shows the schematic of the single cyclone separator with three stages, developed in this study. The cyclone separator basically consisted of three cylinders of different diameters and one vortex finder. The inner diameter of the 1st stage cylinder was denoted as D1, that of the 2nd stage cylinder as D2, and that of the 3rd stage cylinder as D3. The height of the cylinders was denoted as h. The inner diameter of the vortex finder was denoted as Dvf, and the height of the vortex finder hidden in the cyclone separator was set to be 2h. The diameter and the height of the cone bottom were denoted as Dcb and hcb, respectively. At the edge of each cylinder, there was a rectangular slit connected to each particle trap, that is, Trap 1, Trap 2, or Trap 3. For the slits at the 1st and the 2nd stages, the slit length was set to 5 mm and the slit width was denoted as t. The height and the width of the inlet were denoted as a and b, respectively. The dimensions of the cyclone separator normalized by D3(¼24.8 mm), were D1/D3 ¼4, D2/D3 ¼2, h/D3 ¼1, t/D3 ¼0.08, Dvf/D3 ¼0.5, Dcb/D3 ¼ 0.5, hcb/D3 ¼1.48, a/ D3 ¼0.4, and b/D3 ¼0.16. Airflow in the cyclone separator was simulated using a computational fluid dynamics (CFD) code, ANSYS FLUENT Release 14.0. The flow was assumed to be three-dimensional, steady, incompressible, and turbulent. The pressure at the outlet of the cyclone separator was set at 101.3 kPa. The temperature was assumed to be 20 oC. The conservation equations for turbulent flow in the cyclone separator were considered to be the continuity equation and the Reynolds averaged Navier–Stokes (RANS) equations. The RNG k–ε turbulence model was selected. The SIMPLEC algorithm was used for pressure–velocity coupling. The boundary conditions for flow analysis were velocity inlet at the cyclone inlet, pressure outlet at the cyclone outlet, and no-slip condition at the inner wall of the cyclone separator. The convergence criterion was set at 10 7 for iteratively solving the continuity, momentum, and energy equations. Grid dependency test was performed, and the number of cells was approximately 0.6 million. Particle trajectories were calculated in a Lagrangian reference frame using the Discrete Phase Models (DPM) supplied in the FLUENT. Particles were assumed to be spherical. The gravitational force, Brownian force, and Stokes drag force with slip correction were considered. Here, the calculation of the Brownian force was based on the Gaussian white noise random process (Li and Ahmadi, 1992). The particles were evenly spaced on the inlet plane and injected at the same velocity as the
C.-W. Park et al. / Journal of Aerosol Science 89 (2015) 18–25
21
airflow. The particles were assumed to be permanently trapped when they touched the bottom walls of the Trap 1, Trap 2, and Trap 3. The sampling fraction at each stage, fi, was calculated as fi ¼
Nt;i ; N t;1 þ N t;2 þ N t;3 þ Ne
ð1Þ
where i(¼ 1, 2, or 3) denoted the ith trap, Nt,i was the number of particles trapped on the bottom wall of the ith trap, and Ne was the number of particles that escaped the cyclone separator through the vortex finder. In numerical simulations, the flow rate of aerosol introduced to the cyclone separator, that is, Qa, was varied in the range between 40 L/min and 100 L/min. 3. Experimental Fig. 2 shows the experimental setup for measuring the sampling fraction of the single cyclone separator with three stages. The cyclone separator, of which design is shown in Fig. 1, was manufactured for experiments. Polystyrene latex (PSL, Thermo Scientific) spheres in liquid suspension and Micropearl SP (Sekisui Chemical) particles in dry form as powder were used. The PSL particles with the size of 1 μm were aerosolized using a Collison-type atomizer and dried in a diffusion-dryer. The Micropearl SP particles with the sizes of 3 μm, 5 μm, 8 μm, and 10 μm were well mixed and then aerosolized using a fluidized bed aerosol generator. A three-way valve was used to select the type of aerosol particles. Either the PSL aerosol or the Micropearl SP aerosol was mixed with clean air (make-up air), and the flow rate of the aerosol introduced to the cyclone separator was 60 L/min.
Fig. 3. SEM images of the PSL particles sampled on the TEM grids mounted on: (a) Trap 1; (b) Trap 2; (c) Trap 3; and (d) downstream filter.
22
C.-W. Park et al. / Journal of Aerosol Science 89 (2015) 18–25
Fig. 4. SEM images of the Micropearl SP particles sampled on the TEM grids mounted on: (a) Trap 1; (b) Trap 2; (c) Trap 3; and (d) downstream filter.
Fig. 5. Comparison of the sampling fraction of the cyclone separator developed in this study, between the simulation results and the experimental data.
Before the aerosol was introduced to the cyclone separator, TEM grids were mounted on the bottom of each trap (Trap 1, Trap 2, and Trap 3) and also on the filter located downstream of the cyclone separator. Then, the aerosol particles were
C.-W. Park et al. / Journal of Aerosol Science 89 (2015) 18–25
23
Fig. 6. Flow field in the cyclone separator developed in this study: (a) static pressure contour and (b) velocity contour.
Fig. 7. Tangential velocity distribution.
injected into the cyclone separator. After the sampling was completed, the TEM grids were unloaded from the cyclone separator and the downstream filter, and analyzed using a scanning electron microscope. Ten SEM images were taken from different positions on each TEM grid. By analyzing all SEM images, the number of particles collected on the bottom of each trap was counted according to particle size, and the sampling fraction at each trap was obtained using Eq. (1). The sampling experiment was repeated five times. Before each experimental run, the cyclone separator was cleaned and new TEM grids were installed on the bottom of each trap and the downstream filter. 4. Results Fig. 3 shows the SEM images of the PSL particles collected on the TEM grids mounted on the Trap 1, Trap 2, Trap 3, and downstream filter, and Fig. 4 displays the SEM images of the Micropearl SP particles. As shown in Fig. 3, in case of the PSL particles of 1-μm diameter, some of the particles were sampled only in the Trap 3, and most of them escaped the cyclone
24
C.-W. Park et al. / Journal of Aerosol Science 89 (2015) 18–25
separator and were collected on the downstream filter. As appeared in Fig. 4, in case of the mixture of the Micropearl SP particles having diameters of 3 μm, 5 μm, 8 μm, and 10 μm, no particles were collected on the downstream filter whereas the particles were sampled in each particle trap according to their size, that is, all of the 10-μm particles were sampled in the Trap 1, most of the 8-μm particles in the Trap 1, most of the 5-μm particles in the Trap 2, and most of the 3-μm particles in the Trap 3. From these SEM images, it is clearly shown that the cyclone separator developed in this study enabled the size-selective sampling. Fig. 5 shows the comparison of the sampling fraction at each trap between simulation results and experimental data. The aerosol flow rate was 60 L/min. The simulation results agreed well with the experimental data, validating the correctness of the simulation approach employed in this study. The particles larger than about 8 μm were estimated to be collected on the bottom of the Trap 1, showing a typical S-shaped sampling fraction curve for general cyclone separators. However, the sampling fraction curves for the Trap 2 and Trap 3 looked different from that for the Trap 1. In other words, the Trap 2 and Trap 3 showed the possibility to size-selectively collect the particles according to particle size. The geometric mean diameter (GMD) and geometric standard deviation (GSD) of the particles collected on the bottom of the Trap 2 were anticipated to be 5.43 μm and 1.23, respectively, and the peak value of the sampling fraction was calculated to be about 90%. The GMD and GSD of the particles collected on the bottom of the Trap 3 were predicted to be 2.92 μm and 1.39, respectively, and the peak value of the sampling fraction was estimated to be about 80%. As a result, the size-selective sampling of particles was
Fig. 8. Effect of the aerosol flow rate on the sampling fraction of the cyclone separator developed in this study: (a) Qa ¼40 L/min; (b) Qa ¼ 60 L/min; (c) Qa ¼ 80 L/min; and (d) Qa ¼ 100 L/min.
Table 1 Numerical results of the geometric mean diameter (GMD) and geometric standard deviation (GSD) of the particles sampled on the bottoms of the Trap 2 and Trap 3 of the cyclone separator at various aerosol flow rates. Aerosol flow rate (L/min)
40 60 80 100
Trap 2
Trap 3
GMD (μm)
GSD
GMD (μm)
GSD
7.93 5.43 4.09 3.40
1.23 1.23 1.25 1.21
4.30 2.92 2.15 1.82
1.46 1.39 1.40 1.37
C.-W. Park et al. / Journal of Aerosol Science 89 (2015) 18–25
25
possible by using the single cyclone separator with three stages, developed in this study. Fig. 6 shows the contours of static pressure and velocity on the XY-plane. The static pressure was highest at the 1st stage, decreased gradually at the 2nd and the 3rd stages, and became lowest at the entrance of the vortex finder. The velocity was generally higher near the cylindrical wall of each stage whereas the velocity was lower at the particle trap and cone bottom. Fig. 7 shows the tangential velocity distributions on A–A0 , B–B0 , C–C0 , and D–D0 planes. The peak velocity at the radial distance of about 0.05 m was due to the injection of aerosol through the inlet. The tangential velocities near the cylinder walls ranged approximately from 14 m/s to 16 m/s, that is, were similar. However, as the aerosol proceeded from the 1st stage to the 3rd stage, the centrifugal force became larger due to the decrease of the radius of the cylinder. Therefore, the size of the particles sampled in the Trap 3 was smaller than the size of the particles collected in the Trap 1. Fig. 8 shows the numerically predicted sampling fraction of the developed cyclone separator according to the aerosol flow rate (Qa). In addition, Table 1 lists the geometric mean diameter (GMD) and the geometric standard deviation (GSD) of the particles estimated to be sampled in the Trap 2 and Trap 3 at the tested aerosol flow rates. As Qa increased, the GMDs for both the Trap 2 and the Trap 3 decreased, and the cut-off size of the Trap 1 also became smaller. In the tested range of Qa, the GSD was kept at about 1.2–1.25 for the Trap 2 and it remained at about 1.4 for the Trap 3. For all tested aerosol flow rates, the reference cyclone separator showed three distinct sampling fraction curves for each of the Trap 1, Trap 2, and Trap 3, with the peak value of the sampling fraction of more than 75%, implying that the size-selective sampling of particles can be effectively achieved by varying the aerosol flow rate. 5. Conclusion In this study, a single cyclone separator with three stages, that is, a cyclone separator consisting of three cylinders of different diameters and one vortex finder, was developed for size-selective sampling of particles. The sampling fraction at each stage was investigated both numerically and experimentally, and the numerical results agreed well with the experimental data, validating the simulation approach employed in this study. Then, the effect of the aerosol flow rate on the sampling fraction of the developed cyclone separator was numerically investigated. By varying the aerosol flow rate in the range from 100 L/min to 40 L/min, the cut-off size for the Trap 1 could be determined in the particle size range between roughly 4.5 μm and 11 μm, the GMD of particles collected on the bottom of the Trap 2 in the size range between 3.40 μm and 7.93 μm, and the GMD of particles sampled on the bottom of the Trap 3 in the size range between 1.82 μm and 4.30 μm. The GSD of the collected particles was estimated to be about 1.2–1.25 for the Trap 2 and approximately 1.4 for the Trap 3. Acknowledgment This work was supported by the Global Frontier R&D Program on Center for Multiscale Energy System funded by the National Research Foundation under the Ministry of Science, ICT & Future Planning, Korea (NRF-2012M3A6A7054863). References Chuah, T.G., Gimbun, J., & Choong, T.S. (2006). A CFD study of the effect of cone dimensions on sampling aerocyclones performance and hydrodynamics. Powder Technology, 162, 126–132. Dietz, P.W. (1982). Electrostatically enhanced cyclone separators. Powder Technology, 31, 221–226. Griffiths, W.D., & Boysan, F. (1996). Computational fluid dynamics (CFD) and empirical modeling of the performance of a number of cyclone samplers. Journal of Aerosol Science, 27, 281–304. Hering, S.V. (1987). Calibration of the QCM impactor for stratospheric sampling. Aerosol Science and Technology, 7, 257–274. Hsiao, T.C., Chen, D.R., Li, L., Greenberg, P., & Street, K.W. (2010). Development of a multi-stage axial flow cyclone. Aerosol Science and Technology, 44, 253–261. Hsiao, T.C., Huang, S.H., Hsu, C.W., Chen, C.C., & Chang, P.K. (2015). Effects of the geometric configuration on cyclone performance. Journal of Aerosol Science, 86, 1–12. Lee, J.W., Yang, H.J., & Lee, D.Y. (2006). Effect of the cylinder shape of a long-coned cyclone on the stable flow-field establishment. Powder Technology, 165, 30–38. Li, A., & Ahmadi, G. (1992). Dispersion and deposition of spherical particles from point sources in a turbulent channel flow. Aerosol Science and Technology, 16, 209–226. Lim, K.S., Lee, K.W., & Kuhlman, M.R. (2001). An experimental study of the performance factors affecting particle collection efficiency of the electrocyclone. Aerosol Science and Technology, 35, 969–977. Marjamäki, M., Keskinen, J., Chen, D.R., & Pui, D.Y.H. (2000). Performance evaluation of the electrical low-pressure impactor (ELPI). Journal of Aerosol Science, 31, 249–261. Marple, V.A., Rubow, K.L., & Behm, S.M. (1991). A microorifice uniform deposit impactor (MOUDI): Description, calibration, and use. Aerosol Science and Technology, 14, 434–446. McFarland, A.R., Bertch, R.W., Fisher, G. L, & Prentice, B.A. (1977). Fractionator for size classification of aerosolized solid particulate matter. Environmental Science and Technology, 11, 781–784. Ray, M.B., Luning, P.E., Hoffmann, A.C., Plomp, A., & Beumer, M.I. (1997). Post cyclone (PoC): An innovative way to reduce the emission of fines from industrial cyclones. Industrial and Engineering Chemistry Research, 36, 2766–2774. Smith, W.B., Wilson, R.R., Jr., & Harris, D.B. (1979). A five-stage cyclone system for in situ sampling. Environmental Science and Technology, 13, 1387–1392. Xiang, R., Park, S.H., & Lee, K.W. (2001). Effects of cone dimension on cyclone performance. Journal of Aerosol Science, 32, 549–561.
Contents lists available at ScienceDirect
Journal of Aerosol Science journal homepage: www.elsevier.com/locate/jaerosci
Development of a single cyclone separator with three stages for size-selective sampling of particles Chun-Woo Park, Dae-Hyun Song, Se-Jin Yook n School of Mechanical Engineering, Hanyang University, Seoul 133-791, Republic of Korea
a r t i c l e in f o
abstract
Article history: Received 27 December 2014 Received in revised form 23 May 2015 Accepted 1 July 2015 Available online 9 July 2015
In this study, a single cyclone separator composed of three cylinders of different diameters and one vortex finder was developed for size-selective sampling of particles. This cyclone separator had three particle traps, that is, dust bins, each of which was connected to each of the cylinders. A single cyclone separator with three stages was designed and manufactured. The numerically predicted sampling fractions agreed well with the experimental data. From both numerical and experimental results, three distinct sampling fraction curves with different geometric mean diameters (GMD) appeared and the sizeselective sampling of particles was possible using the single cyclone separator with three stages. Then, the flow rate of aerosol introduced to the cyclone separator was varied to change the GMD of particles collected on the bottom of each particle trap. & 2015 Elsevier Ltd. All rights reserved.
Keywords: Cyclone separator Size-selective sampling Sampling fraction GMD GSD
1. Introduction Size-selective sampling is essential for characterizing aerosol particles according to their size. Cascade impactors such as QCM impactor (Hering, 1987), MOUDI (Marple, Rubow, & Behm, 1991), and ELPI (Marjamäki, Keskinen, Chen, & Pui, 2000), are commonly used for size-selective sampling of particles. The cascade impactors, however, are usually operated at a relatively low flow rate and can have particle bounce and re-entrainment problem. If the impaction plates are coated with sticky material like grease to reduce particle bounce, the collected particles can be contaminated by the coating material. Therefore, it is needed to develop a device which is appropriate for size-selective sampling at a relatively high flow rate without using coating materials. Cyclone separators are widely used in many fields due to their simple geometry and high sampling flow rate. Griffiths and Boysan (1996) successfully modeled the performance of three types of cyclone samplers using a computational fluid dynamics (CFD) package. Lee, Yang, and Lee (2006) investigated the effect of the cylinder shape on the collection efficiency of a long-coned cyclone. Chuah, Gimbun, and Choong (2006) and Xiang, Park, and Lee (2001) evaluated the influence of the cone dimensions on the cyclone performance. Ray, Luning, Hoffmann, Plomp, and Beumer (1997) installed a cylindrical annular shell on top of the vortex finder to trap dusts escaping from an industrial cyclone. Hsiao, Huang, Hsu, Chen, and Chang (2015) investigated the influences of geometric configurations on cyclone performance and proposed the optimal ranges for the geometric dimensional ratios. Dietz (1982) and Lim, Lee, and Kuhlman (2001) employed the electrical force to improve the collection efficiency of cyclone separators. So far, a lot of studies have been conducted on the enhancement of
n
Corresponding author. Tel.: þ82 2 2220 0422; fax: þ 82 2 2220 2299. E-mail address: [email protected] (S.-J. Yook).
http://dx.doi.org/10.1016/j.jaerosci.2015.07.001 0021-8502/& 2015 Elsevier Ltd. All rights reserved.
C.-W. Park et al. / Journal of Aerosol Science 89 (2015) 18–25
19
Fig. 1. Schematic of the single cyclone separator with three stages, developed in this study: (a) perspective view and (b) front view.
the cyclone performance. As a result, it is possible to efficiently separate particles from air stream using cyclone separators. However, if a single cyclone separator is used, only the particles larger than a cut-off size are collected and the size-selective sampling cannot be achieved. Multiple cyclone separators with different cut-off sizes can be connected in series for size-selective sampling, similar to the cascade impactors. McFarland, Bertch, Fisher, and Prentice (1977) employed an apparatus consisting of a series of two cyclones and a centripeter, and fractionated fly ash from stack gas to have volume median diameters of 20 μm, 6.3 μm, 3.2 μm, and 2.2 μm with geometric standard deviations of about 1.8. Smith, Wilson, and Harris (1979) developed a five-stage cyclone system for in situ sampling with cut points of 5.4 μm, 2.1 μm, 1.4 μm, 0.65 μm, and 0.32 μm, by connecting five cyclone separators in series. Hsiao, Chen, Li, Greenberg, and Street (2010) developed a multi-stage cyclone system consisting
20
C.-W. Park et al. / Journal of Aerosol Science 89 (2015) 18–25
Fig. 2. Schematic of the experimental setup.
of an impaction inlet and five axial flow cyclone stages in series, and the cut-off sizes of the axial flow cyclones were 0.97 μm, 0.55 μm, 0.255 μm, 0.109 μm, and 0.04 μm. If multiple cyclone separators in series are used for size-selective sampling, the system can be complex and bulky, and additional flow control may be required to have a certain cut-off size depending on the operating flow rate of a cyclone separator. The purpose of this study is to develop a single cyclone separator with three stages for size-selective sampling. A numerical approach is established to estimate the fraction of particles collected in the particle trap of the cyclone separator. Experiments are conducted to validate the numerical approach. Finally, the effect of the flow rate of aerosol injected into the cyclone separator on the size distribution of the collected particles is numerically investigated.
2. Numerical Fig. 1 shows the schematic of the single cyclone separator with three stages, developed in this study. The cyclone separator basically consisted of three cylinders of different diameters and one vortex finder. The inner diameter of the 1st stage cylinder was denoted as D1, that of the 2nd stage cylinder as D2, and that of the 3rd stage cylinder as D3. The height of the cylinders was denoted as h. The inner diameter of the vortex finder was denoted as Dvf, and the height of the vortex finder hidden in the cyclone separator was set to be 2h. The diameter and the height of the cone bottom were denoted as Dcb and hcb, respectively. At the edge of each cylinder, there was a rectangular slit connected to each particle trap, that is, Trap 1, Trap 2, or Trap 3. For the slits at the 1st and the 2nd stages, the slit length was set to 5 mm and the slit width was denoted as t. The height and the width of the inlet were denoted as a and b, respectively. The dimensions of the cyclone separator normalized by D3(¼24.8 mm), were D1/D3 ¼4, D2/D3 ¼2, h/D3 ¼1, t/D3 ¼0.08, Dvf/D3 ¼0.5, Dcb/D3 ¼ 0.5, hcb/D3 ¼1.48, a/ D3 ¼0.4, and b/D3 ¼0.16. Airflow in the cyclone separator was simulated using a computational fluid dynamics (CFD) code, ANSYS FLUENT Release 14.0. The flow was assumed to be three-dimensional, steady, incompressible, and turbulent. The pressure at the outlet of the cyclone separator was set at 101.3 kPa. The temperature was assumed to be 20 oC. The conservation equations for turbulent flow in the cyclone separator were considered to be the continuity equation and the Reynolds averaged Navier–Stokes (RANS) equations. The RNG k–ε turbulence model was selected. The SIMPLEC algorithm was used for pressure–velocity coupling. The boundary conditions for flow analysis were velocity inlet at the cyclone inlet, pressure outlet at the cyclone outlet, and no-slip condition at the inner wall of the cyclone separator. The convergence criterion was set at 10 7 for iteratively solving the continuity, momentum, and energy equations. Grid dependency test was performed, and the number of cells was approximately 0.6 million. Particle trajectories were calculated in a Lagrangian reference frame using the Discrete Phase Models (DPM) supplied in the FLUENT. Particles were assumed to be spherical. The gravitational force, Brownian force, and Stokes drag force with slip correction were considered. Here, the calculation of the Brownian force was based on the Gaussian white noise random process (Li and Ahmadi, 1992). The particles were evenly spaced on the inlet plane and injected at the same velocity as the
C.-W. Park et al. / Journal of Aerosol Science 89 (2015) 18–25
21
airflow. The particles were assumed to be permanently trapped when they touched the bottom walls of the Trap 1, Trap 2, and Trap 3. The sampling fraction at each stage, fi, was calculated as fi ¼
Nt;i ; N t;1 þ N t;2 þ N t;3 þ Ne
ð1Þ
where i(¼ 1, 2, or 3) denoted the ith trap, Nt,i was the number of particles trapped on the bottom wall of the ith trap, and Ne was the number of particles that escaped the cyclone separator through the vortex finder. In numerical simulations, the flow rate of aerosol introduced to the cyclone separator, that is, Qa, was varied in the range between 40 L/min and 100 L/min. 3. Experimental Fig. 2 shows the experimental setup for measuring the sampling fraction of the single cyclone separator with three stages. The cyclone separator, of which design is shown in Fig. 1, was manufactured for experiments. Polystyrene latex (PSL, Thermo Scientific) spheres in liquid suspension and Micropearl SP (Sekisui Chemical) particles in dry form as powder were used. The PSL particles with the size of 1 μm were aerosolized using a Collison-type atomizer and dried in a diffusion-dryer. The Micropearl SP particles with the sizes of 3 μm, 5 μm, 8 μm, and 10 μm were well mixed and then aerosolized using a fluidized bed aerosol generator. A three-way valve was used to select the type of aerosol particles. Either the PSL aerosol or the Micropearl SP aerosol was mixed with clean air (make-up air), and the flow rate of the aerosol introduced to the cyclone separator was 60 L/min.
Fig. 3. SEM images of the PSL particles sampled on the TEM grids mounted on: (a) Trap 1; (b) Trap 2; (c) Trap 3; and (d) downstream filter.
22
C.-W. Park et al. / Journal of Aerosol Science 89 (2015) 18–25
Fig. 4. SEM images of the Micropearl SP particles sampled on the TEM grids mounted on: (a) Trap 1; (b) Trap 2; (c) Trap 3; and (d) downstream filter.
Fig. 5. Comparison of the sampling fraction of the cyclone separator developed in this study, between the simulation results and the experimental data.
Before the aerosol was introduced to the cyclone separator, TEM grids were mounted on the bottom of each trap (Trap 1, Trap 2, and Trap 3) and also on the filter located downstream of the cyclone separator. Then, the aerosol particles were
C.-W. Park et al. / Journal of Aerosol Science 89 (2015) 18–25
23
Fig. 6. Flow field in the cyclone separator developed in this study: (a) static pressure contour and (b) velocity contour.
Fig. 7. Tangential velocity distribution.
injected into the cyclone separator. After the sampling was completed, the TEM grids were unloaded from the cyclone separator and the downstream filter, and analyzed using a scanning electron microscope. Ten SEM images were taken from different positions on each TEM grid. By analyzing all SEM images, the number of particles collected on the bottom of each trap was counted according to particle size, and the sampling fraction at each trap was obtained using Eq. (1). The sampling experiment was repeated five times. Before each experimental run, the cyclone separator was cleaned and new TEM grids were installed on the bottom of each trap and the downstream filter. 4. Results Fig. 3 shows the SEM images of the PSL particles collected on the TEM grids mounted on the Trap 1, Trap 2, Trap 3, and downstream filter, and Fig. 4 displays the SEM images of the Micropearl SP particles. As shown in Fig. 3, in case of the PSL particles of 1-μm diameter, some of the particles were sampled only in the Trap 3, and most of them escaped the cyclone
24
C.-W. Park et al. / Journal of Aerosol Science 89 (2015) 18–25
separator and were collected on the downstream filter. As appeared in Fig. 4, in case of the mixture of the Micropearl SP particles having diameters of 3 μm, 5 μm, 8 μm, and 10 μm, no particles were collected on the downstream filter whereas the particles were sampled in each particle trap according to their size, that is, all of the 10-μm particles were sampled in the Trap 1, most of the 8-μm particles in the Trap 1, most of the 5-μm particles in the Trap 2, and most of the 3-μm particles in the Trap 3. From these SEM images, it is clearly shown that the cyclone separator developed in this study enabled the size-selective sampling. Fig. 5 shows the comparison of the sampling fraction at each trap between simulation results and experimental data. The aerosol flow rate was 60 L/min. The simulation results agreed well with the experimental data, validating the correctness of the simulation approach employed in this study. The particles larger than about 8 μm were estimated to be collected on the bottom of the Trap 1, showing a typical S-shaped sampling fraction curve for general cyclone separators. However, the sampling fraction curves for the Trap 2 and Trap 3 looked different from that for the Trap 1. In other words, the Trap 2 and Trap 3 showed the possibility to size-selectively collect the particles according to particle size. The geometric mean diameter (GMD) and geometric standard deviation (GSD) of the particles collected on the bottom of the Trap 2 were anticipated to be 5.43 μm and 1.23, respectively, and the peak value of the sampling fraction was calculated to be about 90%. The GMD and GSD of the particles collected on the bottom of the Trap 3 were predicted to be 2.92 μm and 1.39, respectively, and the peak value of the sampling fraction was estimated to be about 80%. As a result, the size-selective sampling of particles was
Fig. 8. Effect of the aerosol flow rate on the sampling fraction of the cyclone separator developed in this study: (a) Qa ¼40 L/min; (b) Qa ¼ 60 L/min; (c) Qa ¼ 80 L/min; and (d) Qa ¼ 100 L/min.
Table 1 Numerical results of the geometric mean diameter (GMD) and geometric standard deviation (GSD) of the particles sampled on the bottoms of the Trap 2 and Trap 3 of the cyclone separator at various aerosol flow rates. Aerosol flow rate (L/min)
40 60 80 100
Trap 2
Trap 3
GMD (μm)
GSD
GMD (μm)
GSD
7.93 5.43 4.09 3.40
1.23 1.23 1.25 1.21
4.30 2.92 2.15 1.82
1.46 1.39 1.40 1.37
C.-W. Park et al. / Journal of Aerosol Science 89 (2015) 18–25
25
possible by using the single cyclone separator with three stages, developed in this study. Fig. 6 shows the contours of static pressure and velocity on the XY-plane. The static pressure was highest at the 1st stage, decreased gradually at the 2nd and the 3rd stages, and became lowest at the entrance of the vortex finder. The velocity was generally higher near the cylindrical wall of each stage whereas the velocity was lower at the particle trap and cone bottom. Fig. 7 shows the tangential velocity distributions on A–A0 , B–B0 , C–C0 , and D–D0 planes. The peak velocity at the radial distance of about 0.05 m was due to the injection of aerosol through the inlet. The tangential velocities near the cylinder walls ranged approximately from 14 m/s to 16 m/s, that is, were similar. However, as the aerosol proceeded from the 1st stage to the 3rd stage, the centrifugal force became larger due to the decrease of the radius of the cylinder. Therefore, the size of the particles sampled in the Trap 3 was smaller than the size of the particles collected in the Trap 1. Fig. 8 shows the numerically predicted sampling fraction of the developed cyclone separator according to the aerosol flow rate (Qa). In addition, Table 1 lists the geometric mean diameter (GMD) and the geometric standard deviation (GSD) of the particles estimated to be sampled in the Trap 2 and Trap 3 at the tested aerosol flow rates. As Qa increased, the GMDs for both the Trap 2 and the Trap 3 decreased, and the cut-off size of the Trap 1 also became smaller. In the tested range of Qa, the GSD was kept at about 1.2–1.25 for the Trap 2 and it remained at about 1.4 for the Trap 3. For all tested aerosol flow rates, the reference cyclone separator showed three distinct sampling fraction curves for each of the Trap 1, Trap 2, and Trap 3, with the peak value of the sampling fraction of more than 75%, implying that the size-selective sampling of particles can be effectively achieved by varying the aerosol flow rate. 5. Conclusion In this study, a single cyclone separator with three stages, that is, a cyclone separator consisting of three cylinders of different diameters and one vortex finder, was developed for size-selective sampling of particles. The sampling fraction at each stage was investigated both numerically and experimentally, and the numerical results agreed well with the experimental data, validating the simulation approach employed in this study. Then, the effect of the aerosol flow rate on the sampling fraction of the developed cyclone separator was numerically investigated. By varying the aerosol flow rate in the range from 100 L/min to 40 L/min, the cut-off size for the Trap 1 could be determined in the particle size range between roughly 4.5 μm and 11 μm, the GMD of particles collected on the bottom of the Trap 2 in the size range between 3.40 μm and 7.93 μm, and the GMD of particles sampled on the bottom of the Trap 3 in the size range between 1.82 μm and 4.30 μm. The GSD of the collected particles was estimated to be about 1.2–1.25 for the Trap 2 and approximately 1.4 for the Trap 3. Acknowledgment This work was supported by the Global Frontier R&D Program on Center for Multiscale Energy System funded by the National Research Foundation under the Ministry of Science, ICT & Future Planning, Korea (NRF-2012M3A6A7054863). References Chuah, T.G., Gimbun, J., & Choong, T.S. (2006). A CFD study of the effect of cone dimensions on sampling aerocyclones performance and hydrodynamics. Powder Technology, 162, 126–132. Dietz, P.W. (1982). Electrostatically enhanced cyclone separators. Powder Technology, 31, 221–226. Griffiths, W.D., & Boysan, F. (1996). Computational fluid dynamics (CFD) and empirical modeling of the performance of a number of cyclone samplers. Journal of Aerosol Science, 27, 281–304. Hering, S.V. (1987). Calibration of the QCM impactor for stratospheric sampling. Aerosol Science and Technology, 7, 257–274. Hsiao, T.C., Chen, D.R., Li, L., Greenberg, P., & Street, K.W. (2010). Development of a multi-stage axial flow cyclone. Aerosol Science and Technology, 44, 253–261. Hsiao, T.C., Huang, S.H., Hsu, C.W., Chen, C.C., & Chang, P.K. (2015). Effects of the geometric configuration on cyclone performance. Journal of Aerosol Science, 86, 1–12. Lee, J.W., Yang, H.J., & Lee, D.Y. (2006). Effect of the cylinder shape of a long-coned cyclone on the stable flow-field establishment. Powder Technology, 165, 30–38. Li, A., & Ahmadi, G. (1992). Dispersion and deposition of spherical particles from point sources in a turbulent channel flow. Aerosol Science and Technology, 16, 209–226. Lim, K.S., Lee, K.W., & Kuhlman, M.R. (2001). An experimental study of the performance factors affecting particle collection efficiency of the electrocyclone. Aerosol Science and Technology, 35, 969–977. Marjamäki, M., Keskinen, J., Chen, D.R., & Pui, D.Y.H. (2000). Performance evaluation of the electrical low-pressure impactor (ELPI). Journal of Aerosol Science, 31, 249–261. Marple, V.A., Rubow, K.L., & Behm, S.M. (1991). A microorifice uniform deposit impactor (MOUDI): Description, calibration, and use. Aerosol Science and Technology, 14, 434–446. McFarland, A.R., Bertch, R.W., Fisher, G. L, & Prentice, B.A. (1977). Fractionator for size classification of aerosolized solid particulate matter. Environmental Science and Technology, 11, 781–784. Ray, M.B., Luning, P.E., Hoffmann, A.C., Plomp, A., & Beumer, M.I. (1997). Post cyclone (PoC): An innovative way to reduce the emission of fines from industrial cyclones. Industrial and Engineering Chemistry Research, 36, 2766–2774. Smith, W.B., Wilson, R.R., Jr., & Harris, D.B. (1979). A five-stage cyclone system for in situ sampling. Environmental Science and Technology, 13, 1387–1392. Xiang, R., Park, S.H., & Lee, K.W. (2001). Effects of cone dimension on cyclone performance. Journal of Aerosol Science, 32, 549–561.