The Effect of Cyclone Shape and Dust Collector on Gas-Solid Flow and Performance

Transcription

1 Internatonal Journal of Mechancal and Aerospace Engneerng 6 01 The Effect of Cyclone Shape and Dust Collector on Gas-Sold Flow and Performance Kyoungwoo Park, Chol-Ho Hong, J-Won Han, Byeong-Sam Km, Cha-Sk Park, and Oh Kyung Kwon Abstract Numercal analyss of flow characterstcs and separaton effcency n a hgh-effcency cyclone has been performed. Several models based on the expermental observaton for a desgn purpose were proposed. However, the model s only estmated the cyclone's performance under the lmted envronments; t s dffcult to obtan a general model for all types of cyclones. The purpose of ths study s to fnd out the flow characterstcs and separaton effcency numercally. The Reynolds stress model (RSM) was employed nstead of a standard k-ε or a k-ω model whch was sutable for sotropc turbulence and t could predct the pressure drop and the Rankne vortex very well. For small partcles, there were three sgnfcant components (entrance of vortex fnder, cone, and dust collector) for the partcle separaton. In the present work, the partcle re-entranng phenomenon from the dust collector to the cyclone body was observed after consderable tme. Ths re-entranment degraded the separaton effcency and was one of the sgnfcant factors for the separaton effcency of the cyclone. Keywords CFD, Hgh-effcency cyclone, Pressure drop, Rankne vortex, Reynolds stress model (RSM), Separaton effcency. C I. INTRODUCTION YCOLNES, whch can separate the partcles from an ar stream, have been wdely used n many ndustral processes, such as ar polluton control and envronmental cleanng processes due to ther well adaptablty to harsh condtons, smplcty to desgn, and low costs to operate and mantan. The cyclone desgns are generally classfed nto straght-through, un-flow, and reverse-flow cyclones accordng to the purpose n use. Among them, t s known that the use of tangental nlet and reverse-flow s the most common way for cyclone desgn. Because of the above mentoned many merts of cyclone, much attenton have been pad on predctng the flow felds n cyclones both by expermental and numercal methods for last few decades [1-3]. The performance of a cyclone separator s generally characterzed by the collecton effcency of partcles and the pressure drop through the cyclone. Accordng to the many researches, the cyclone heght, dameter, and shape (.e., cylnder or rectangular), the shape and dameter of vortex fnder, and the nlet geometry can nfluence consderably the performance of the cyclones. Kyoungwoo Park s wth the Department of Mechancal Engneerng, Hoseo Unversty, Asan , Rep. of Korea (phone: ; fax: ; e-mal: kpark@hoseo.edu). Chol Ho Hong s wth the Department of System Control Engneerng, Hoseo Unversty, Asan , Rep. of Korea (e-mal: chhong@hoseo. edu). J Won Han s wth the Department of Mechancal Engneerng, Hoseo Unversty, Asan , Rep. of Korea (e-mal: jwhan@ hoseo. edu). Byeong Sam Km s wth the Department of Automotve Engneerng, Hoseo Unversty, Asan , Rep. of Korea (e-mal: kbs@ hoseo. edu). Oh Kyung Kwon s wth the KITECH, Cheonan , Rep. of Korea (e-mal: kwonok@ktech.re.kr) In 006, [4] analyzed the nfluence of the shape of the cyclone cylnder on the flow characterstcs and collecton performance by usng the commercal package, FLUENT. They observed that the long-cone cyclone has an unstable flow felds and these characterstcs results n the short crculatng flow at the vortex fnder openng and affects adversely the partcle collecton effcency. [5] evaluated the effect of vortex fnder shape and dameter on cyclone performance and flow feld numercally. In order to predct partcles trackng n the cyclone, the Euleran- Lagrangan procedure was used. They found that the tangental velocty and separaton effcency are decreased when the cyclone vortex fnder dameter s ncreased. Recently, [6] studed numercally the effect of cyclone nlet dmenson on the flow pattern and performance usng the Reynolds stress turbulence model (RSM) for fve cyclone separators. They showed that the tangental velocty n the cyclone decreases wth ncreasng the cyclone nlet dmensons. They also found that the effect of changng the nlet wdth s more sgnfcant than that of the nlet heght for the cut-off dameter and the optmal rato of nlet wdth to heght, b/a, s from 0.5 to 0.7. The nteracton of gas-sold plays an mportant role n the flow feld and performance n cyclones. There are two knds of methods to predct t, that s, one-way and two-way couplng approaches. An one-way couplng method s base on the assumpton that the presence of the partcles doses not affect the flow feld because the partcle loadng n a cyclone separator s small [7]. On the contrary, a two-way couplng effect [8] s consdered the effect of the partcles on the gas flow. In ths model, the partcle source-n cell (PSIC) model s generally used to solve the momentum equaton of a partcle n the two-phase flow. In the present study, the effects of the cyclone dmensons, the partcle sze, and the presence of dust collector on the flow characterstcs and performance of the cyclone are nvestgated numercally. The turbulent flow s analyzed by usng the Reynolds stress model (RSM) and the Euleran-Lagrangan approach s mplemented for predctng the partcle moton. A one-way couplng effect, whch the gas flow s not affected by the presence of the partcles, s employed to estmate the gas-sold nteracton. The computatonal model can predct the two-phase flow n cyclones accurately and provde the desgn concept for the presence of the dust collector. 37

2 Internatonal Journal of Mechancal and Aerospace Engneerng 6 01 b L Inlet Fg. 1 Schematcs of the cyclone and ts grd system II. PHYSICAL MODEL The schematc of the cyclone separator and the generated grd system consdered n the present work s gven n Fg. 1 and the geometrcal dmensons are depcted n Table 1. As can be seen n Table 1, all dmensons are normalzed by usng the dameter of cyclone body (D = 90 mm). Accordng to the cyclone s heght, t can be dvded nto three parts such as vortex fnder (annular space), separaton space and dust collecton part. The nlet ppe s mounted tangentally onto the sde of the cylndrcal part of the cyclone body and the workng flud (gas and partcles) s ncomng through ths secton wth the unform velocty such as v n = 5 m/s. The ext tube, called the vortex fnder, s fxed on the top of the cyclone. A. Governng Equatons III. THEORETICAL ANALYSIS The flow n cyclones s assumed to be turbulent swrl flow wth ncompressble flud and t can be reasonably predcted by the Reynolds stress model (RSM). The turbulent flow for gas can be descrbed by the Reynolds-Average Naver-Stokes (RANS) equaton and the equaton of contnuty for the mean moton. They are expressed n tensor notaton as follows: Dmenson L e a Ext (Vortex fnder) D e D e TABLE I GEOMETRICAL DIMENSIONS FOR THE CYCLONE Cyclone dameter (D) Gas outlet dameter (D e) Dust collector nlet dameter (D d) Dust collector dameter (D d) Ext length (L e) Vortex fnder length (S) Cylnder length (h) Cyclone length (H) Dust collector length (L d) Inlet wdth (a) Inlet heght (b) Inlet length (l) D D d D d Dust collector S h L d H Values/ D (mm) 1.0D (90 mm) 0.5D (145 mm) 3/8D (108 mm) 1.0D (90 mm).0d (580 mm) 0.5D (145mm) 0.5D (145 mm) 4.0D (1160 mm) 1.7D (493 mm) 0.D (58 mm) 0.5D (145 mm) 1.0D (90 mm) x ( U ) = 0 t x j ρ x xx j x j U 1 P U ( U ) + U j = + ν ( u u j ) Where U s the mean velocty, x the coordnate system, t the tme, P the mean pressure, ρ the gas densty, and ν the knematc vscosty. u u j (= R j ) represents the Reynolds stress tensor and u = u U s the -th flud fluctuaton velocty component. As shown n Eq.(), the Reynolds stresses should be modeled usng varous assumptons. In the present work, the Reynolds stress terms are drectly calculated by RSTM. B. Turbulent Modelng The accurate predcton of a strong turbulent swrl flow s generally dependent on the turbulent model used. In the present work, the Reynolds stress turbulent model (RSTM) whch solves the ndvdual Reynolds stress term ( u u j ) by usng dfferental transport equatons s used. The transport equatons of Reynolds stresses can be wrtten as ( ) ( ) u u j + U k u u j = D j + P j + Π j + E j t xk Where the ndvdual terms n the rght hand sde stand for stress dffuson, stress producton, pressure stran, and dsspaton terms, respectvely, and defned as follows; ( u u ) νt Dj = j xk σ k xk U j U 1 Pj = u uk + u juk, P = Pj xk xk ε j = C1 u u k 3 j k + C P j 3 j P k Π δ δ Ej = δjε (7) 3 Here ν t s the turbulent vscosty, P the fluctuaton knetc energy producton, k the turbulent knetc energy ( = (1/ ) u u j ), and ε the dsspaton rate of k, respectvely. The emprcal constants are σ k = 1, C 1 = 1.8, and C = 0.6 [9]. The transport equaton for the turbulent dsspaton rate ( ε ) s expressed as ε ε νt ε ε U U ε + j = ν + Cε1 ( u u j ) Cε t x j x j σε x j k x j k The values of constants are σ ε = 1.3, C ε = 1.9. C. Partcle Moton Equatons (1) () (3) (4) (5) (6) (8) C ε = and The basc assumptons employed n the present work to model the partcle moton are as follows; the sold (partcle) has a fully sphercal shape and dsperses dlutely n the gas phase so that the gas-sold nteractons and the nfluence of the dspersed partcle volume fracton on the gas phase are neglgbly small. 38

3 Internatonal Journal of Mechancal and Aerospace Engneerng 6 01 Generally, the partcle loadng n a cyclone s small (3-5%), and therefore, t can be assumed that the presence of the partcles does not affect the flow feld (.e., one-way couplng). Addtonally, collsons between partcles and the walls of the cyclone are assumed to be perfectly elastc and the nteracton between partcles s neglected because of dlute flow. In order to smulate the partcle moton n the cyclone, the dscrete phase model (DPM) was used by defnng the ntal poston, velocty and sze of ndvdual partcles. The trajectory of partcles was obtaned by ntegratng the force balance on the partcle. The equaton of moton of a small partcle, whch s ncluded the effects of nonlnear drag and gravtatonal forces, n terms of the Euleran-Lagrangan approach s gven by [10]. dup, 18 C Re D r = ( u up, ) + g 1 dt ρ 4 pd p dxp, dt u p, µ ρ ρ p (9) = (10) Here u and up, are the gas and partcle veloctes n -drecton, respectvely, xp, ts poston, ρ p the partcle densty, d p the partcle dameter, g the acceleraton of gravty n -drecton. ρ, µ are the densty and vscosty of gas, respectvely. Re r represents the relatve Reynolds number, whch s defned as ρd p up, u Rer = (11) µ The drag coeffcent, C D, for sphercal partcles s calculated by usng the correlaton proposed by Hnds[11] 4 C D = for Rer 1 Re < (1) r 4 1 /3 CD = 1+ Rer Rer 6 for 1< Rer < 400 (13) The frst term on the rght-hand sde (RHS) of Eq.(9) s the drag force per unt partcle mass, generally the domnatng force, due to the relatve slp between the partcle and the flud. D. Numercal Approaches The gas phase s treated as a contnuum by solvng the RNS equatons wth the turbulent model, whle the sold (or dsperse) phase s solved by trackng a large number of partcles through the calculated flow feld. The dspersed phase can exchange mass, momentum, and energy wth the flud phase. The numercal smulatons presented n ths work were done by means of STAR-CCM+ [1] whch s one of the general purpose commercal S/Ws. The pressure-velocty couplng problem s resolved by the SIMPLE algorthm [13]. The unsteady RSTM was used wth a tme step of s. Due to dffculty to reach the convergence n smulatons, the standard k ε model was used frstly to calculate the turbulent propertes, and then the RSTM was adopted to obtan the fnal results. The solutons are treated as converged ones when the 5 sum of normalzed resdual s less than1 10. The boundary condton for arflow veloctes at the cyclone nlet was assumed to be unform at dfferent nlet veloctes. The pressure boundary condton at ext was employed. At the wall, no-slp boundary condton was appled for velocty. IV. RESULTS AND DISCUSSION A. Valdaton of CFD Model n the Cyclone Fg. presents the pressure drop n the cyclone separator for varous nlet veloctes n order to valdate the present CFD-model by comparng the expermental data and the computatonal results [14]. As shown n ths fgure, the pressure drop calculated n the present work shows a good agreement wth the expermental data. It can be also seen that the pressure drop, whch s strongly related wth the pumpng power, ncreases wth the ncrease of the nlet velocty. B. Comparson of Turbulent Models Comparson of tangental velocty profles calculated at y = - 0.4m usng the RSTM and standard k ε model are shown n Fg. 3. As can be seen, the RSTM predcts much better compared to the other model. The hghly swrl flow, n general, generates a strong ansotropy n the turbulent structure and t causes the standard k ε model to provde naccurate predcton of both the locaton of maxmum velocty and the Rankne vortex whch s composed of a forced vortex(nner regon near the cyclone axs) and a free vortex(near the wall). It s clear that the RSTM model can predct the Rankn-type tangental velocty profles very well. C. Grd Dependency Grd refnement test has been performed n order to make sure that the solutons are grd ndependent. For ths, three grd systems (.e., the number of 10,000, 50,000 and 40,000) are generated and compared wth each of them for nondmensonal axal velocty (v a /v n ) n Fg. 4 and computatonal tme n Table II. The polyhedral mesh wth hgh-aspect layer method s used to generate the grd systems because the hgh-pressure gradent and strong double vortex flow (swrl Pressure drop (kpa) : Experment [14] : Computaton[14] : Ths Study Inlet velocty (m/s) Fg. Valdaton of computatonal model by comparson wth the expermental data [14] 39

4 Internatonal Journal of Mechancal and Aerospace Engneerng 6 01 Axal velocty (V a / V n ) Tangental velocty (m/s) Ths study [RSTM] Ths study [k-ε] Poston (m) Fg. 3 Comparson of turbulent models at v n = 5 m/s : Grd No. of 10,000 : 50,000 : 40,000 : Experment [3] Dstance (x/r) Fg. 4 Grd dependency for three cases of grd systems TABLE II COMPUTATIONAL TIME FOR THREE GRID SYSTEMS No. of cells Computatonal tme (hours) Base Refne #1 Refne # 10, , ,000.0 flow) are occurred n the cyclone. As shown n Fg. 4, the axal velocty profles have a good agreement wth the expermental data [3] except for the base grd (10,000). Table II shows the computatonal tme for three grd systems and Refne #3 (No. of 40,000) needs more tme to convergence. In the present work, all computatonal results are obtaned usng the 50,000 cells grd by consderng both the accuracy and the computatonal cost. D. Flow Characterstcs The performances of the cyclone are the separaton effcency and pressure drop. It s well known that the ncrease of nlet velocty of gas can ncrease both the pressure drop and separaton effcency, smultaneously. In the present work, the effects of nlet gas velocty and cyclone shape on the pressure drop are nvestgated frstly. Fg. 5 shows the pressure drop n the cyclone accordng to the nlet velocty for three cases. It can be seen n ths fgure that ncreasng the nlet velocty causes the Pressure drop (kpa) : Case I : Case II (Dd = 0.5D) : Case III (De = 0.3D) Inlet veloccty (m/s) Fg. 5 Pressure drop accordng to nlet velocty for varous cyclone specfcatons pressure drop to ncrease exponentonally. In addton, the vortex fnder dameter (CASE III, D e = 0.5D 0.3D) greatly affects the pressure drop compared to that of con tp dameter (CASE II, D d = 3/8D 0.5D) at the same nlet velocty as shown n Fg. 6. Ths means that the effect of D e on the cyclone performance (pressure drop) s much sgnfcant than that of D d. It s also found that the con tp dameter has almost no nfluence on pressure drop and the effect of vortex fnder dameter on the pressure drop becomes sgnfcantly as the nlet velocty s ncreased. From ths result, t s necessary to carry out the optmzaton for the cyclone shape. E. Separaton Effcency The separaton effcency (η), whch s one of the cyclone performances, s defned as follows [15], ρ p d pvt η = 1 exp t 9µ D res (14) where tres s the gas resdence tme n the cyclone and t ncludes followng two parts: t res = Vn 1 Vs Q Q (15) In the rght hand sde of Eq.(15), the frst and second terms represent the mean gas resdence tme n the entry regon and n the outer vortex, respectvely. V n s the volume of the entry regon below the centerlne of the nlet duct and Q s the volumetrc gas throughout. Especally, Vs s defned due to the fact that about 90% of the entre nlet gas volume flows from the outer vortex to the nlet vortex core along the cyclone wall. More detaled equatons for them are explaned n [16]. In the present work, the partcle resdence tme for three cases, whch s an mportant factor for predctng the separaton effcency, s calculated fort the nlet velocty of 5 m/s and they are lsted n Table III. 40

5 Internatonal Journal of Mechancal and Aerospace Engneerng 6 01 Fg. 6 presents the separaton effcency for varous nlet veloctes. As shown n Fg. 6, η s ncreased wth the nlet CASE I CASE II CASE III TABLE III PARTICLE RESIDENCE TIME IN EQ. (1) 3.16 (sec) 6.05 (sec) 7.78 (sec) velocty and ths s a general phenomenon n cyclones. It can be also seen from the fgure that the cut-off sze ( x 50 ), whch s one of the fundamental characterstcs of cyclones, s vared wth the partcle dameter. The meanng of x 50 s the dameter of partcle that has a 50% probablty of capture by the spn of the nner vortex. For low mass loadng, the cut-off sze can be defned as follows [17], x 50 = 18 µ (0.9 Q) ( ρ ) π ρ v ( H S) p t, cs t (16) where v t, cs s the tangental velocty of the gas at the nner core radus and H t represents the cyclone total heght. As can be seen n Fg. 6, the values of cut-off sze are predcted accordng to the nlet velocty, that s, the partcle szes of x 50 are.5µ m,1.8µ m, and 1.5µ m for v n = 10, 15, and 0 m/s, respectvely. It s also found that f the partcle dameter s over 4µ m ; all partcles are separated rrespectve of the nlet velocty. The partcles whch are entered at nlet are moved wth the gas phase and f the centrfugal force s strongly enough, they are pushed the outsde of CS (control surface, a vrtually extended surface of the vortex fnder) and fallng along the cyclone wall and then collected at the dust collector. However, small partcles, whch generally flow wth a gas because of a weak centrfugal force, enter the nsde of CS and then outflow through ext. Fgure 7 shows the exstence amount of sold partcle nsde the cyclone at a specfc tme. It can be found that the amount sold partcle s decreased wth pass by tme except for a heavy partcle such as a partcle dameter of 3µ m. Ths means that small partcles flow out wth a gas. In addton, Fgure 7 also shows that the fractonal effcency s abruptly decreased at a Separaton effcency (η ) Partcle dameter (µm ) : Vn=10m/s : Vn=15 : Vn=0 Fg. 6 Separaton effcency accordng to the partcle dameter for varous nlet veloctes Fractonal effcency µm 1.5µm 1.75µm.0µm 3.0µm Tme (sec) Fg. 7 Fractonal effcency accordng to the tme for varous partcle dameters certan tme. Ths s due to the fact that the partcle n the cyclone outflows through three dfferent flow passages durng ts downward movement : (1) some partcles leave the cyclone through the entrance of the vortex fnder as soon as the partcle enters nsde the cyclone, () the second passage s that the partcle outflows wth a gas flow from the end of con-tp after tme passes a lttle bt, (3) for the cases of d < 1.75µ m (.e., lght partcles), the fractonal effcency s dramatcally decreased wthn several seconds. The partcles wth small dameter flow nto the dust corrector and most of them are flow out from the corrector because of a small quantty of moment of gas n the dust corrector whle the large partcle s remaned at the corrector. Fg. 8 shows the separaton effcency for varous partcle dameters and cyclone shape. For the same partcle dameter, the separaton effcency s greater than that of base model (CASE I). In addton, when the partcle dameter s larger than 5 µm, there s no nfluence on the effcency rrespectve of the cyclone shape as shown n Fg. 8. Fgure 8 also shows that the dameter of the con tp (CASE II, D = 0.5 D(7.5 mm) ) and vortex fnder (CASE III, D = 0.3 D(57 mm) ) has a sgnfcant Separaton effcency ( η ) e d : Case I : Case II (Dd = 0.5D) : Case III (De = 0.3D) Partcle dameter ( µm ) Fg. 8 Separaton effcency accordng to the partcle dameter for varous cyclone specfcatons p 41

6 Internatonal Journal of Mechancal and Aerospace Engneerng 6 01 nfluence on the effcency compared to the base model (CASE I) whle the effect of t on the pressure drop s slght, as explaned n Fg. 5. It s also found from Fg. 5 and Fg. 8 that hgher tangental velocty results n larger centrfugal force, whch n turn leads to hgher separaton effcency. An nterestng fact s found between Fg. 5 and Fg. 8 that the pressure drop nsde the cyclone s lttle nfluenced by the con tp dameter (D d ), whereas t greatly affects on the cyclone effcency. Ths means the con tp dameter s most sgnfcant factor from the performance pont of vew n the cyclone desgn. V. CONCLUSIONS In order to obtan the optmal solutons of the desgn varables of the cyclone separator, the effect of dameters of vortex fnder and cone tp dameters on ts performance such as separaton effcency and pressure drop were numercally carred out. For ths, the cost effectve and hgh fdelty RANS solver for turbulent flow such as the commercal software (.e., STAR-CD) was used. Reynolds stress transport model (RSTM) was adopted to smulate the ansotropc turbulent flow n a cyclone and ths model had a good agreement compared wth the expermental data. As the results of the present work, the effect of vortex fnder dameter (D e ) on the performance was much greater than that of con tp dameter. It could be also found that the dameter of the con tp (B) has sgnfcant nfluence on the effcency whle t has almost no nfluence on the pressure drop. That s, from the performance of cyclone pont of vew, the con tp dameter s more sgnfcant parameter than the vortex fnder dameter. Due to the tradeoffs between the conflctng objectve functons (.e., pressure drop and separaton effcency), the shape optmzaton has to be carred out n our further study. ACKNOWLEDGMENT Ths work was supported by the New and Renewable Energy of the Korea Insttute of Energy Technology Evaluaton and Plannng (KETEP) grant funded by the Korea government Mnstry of Knowledge Economy (No. 010T ). k Turbulent knetc energy P j Turbulent producton term p Pressure drop NOMENCLATURE U Mean velocty component u Fluctuaton component of -drecton u u Reynolds stress term v t j Tangental velocty x Coordnate system ε Dsspaton rate of k η Separaton effcency n Eq.14 µ Vscosty ν t ρ Turbulent eddy vscosty Gas densty ρ p Partcle densty REFERENCES [1] D. Leth, W. Lcht, the Collecton Effcency of Cyclone Type Partcle Collectors-A New Theoretcal Approach, AIChE Symp. Ser. 68, pp , 197. [] A.J. Hoekstra, H.J Derksen, H.E.A. Van Den Akker, An Expermental and numercal study of turbulent swrlng flow n gas cyclones, Chem. Eng. Sc. Vol.54, pp , [3] A.J. Hoekstra, Gas flow feld and collecton effcency of cyclone separators, Ph.D. Thess, Techncal Unversty Delft, 000. [4] J.W. Lee, H.J. Yang, and D. Y. Lee, Effect of cylnder shape of a long-coned cyclone on the stable flow-feld establshment, Powder Technology, Vol.165, pp.30-38, 006. [5] A. Raouf, M. Shams, M. Farzaneh, and R. Ebrahm, Numercal smulaton and optmzaton of flud flow n a cyclone vortex fnder, Chemcal Engneerng and Processng, vol. 47pp , 008. [6] K. Elsayed and C. Lacor, The effect of cyclone nlet dmensons on the flow pattern and performance, Appled Mathematcal Modelng, vol.35, pp , 011. [7] J.J. Derksen, Separaton performance predctons of a Starmand hgh effcency cyclone, AIChE Journal, Vol.49, pp , 003. [8] G. Wan, G. Sun, X. Xue, and M. Sh, Sold concentraton smulaton of dfferent sze partcles n a cyclone separator, Powder Technology, vol. 183, pp , 008. [9] B.E. Launder, G.J. Reece, and W. Rod, Progress n the development of a Reynolds stress turbulent closure, J. of Flud Mechancs, vol.68, pp , [10] B. Zhao, Y. Su, and J, Zhang, Smulaton of gas flow pattern and separaton effcency n cyclone wth conventonal sngle and spral double nlet confguraton, Chemcal Engneerng Research and Desgn, vol.84, pp , 006. [11] W.C. Hnds, Aerosol Technology: Propertes Behavor and Measurement of Arborne Partcles, John Wley & Sons, New York, 198. [1] STAR-CCM+ v4.0 Methodology, 007. [13] S.V. Patankar, Numercal Heat Transfer and Flud Flow, Taylor & Francs, [14] B. Wang, L.L. Xu, K.W. Chu, and A.B. Yu, Numercal study of gassold Flow n a Cyclone Separator, Appled Mathematcal Modelng, vol.30, pp , 006. [15] R. Clft, M. Ghadr and A.C. Hoffman, A Crtque of two models for cyclone performance, AIChE Journal, vol.37, No., pp.85-89, [16] F. Qan, M. Zhang, An Extended model for determnng the separaton performance of a cyclone, Chem. Eng. Technol, vol.9, no.6, pp.74-78, 006. [17] A.C. Hoffman and L.E. Sten, Gas Cyclones and Swrl Tubes:Prncples, Desgn, and Operaton, nd Ed. Sprnger, 008. Kyoungwoo Park: Dr. Kyoungwoo Park s currently workng at Department of Mechancal Engneerng n Hoseo Unversty as a professor. As soon as he acqured hs Ph.D. degree at Hanyang Unversty, he entered the LG Industral Systems Co., Ltd., and he researched on the flow and thermal characterstcs around/n elevators. In 1999, he joned the Ralph Gref research team for two and half years, who s a professor of Unversty of Calforna at Berkeley and studed about convecton heat transfer wth CFD. After he came back to Korea n 001, he researched and taught students at IDOT of Hanyang Unversty untl Feb Hs man nterestng felds are heat transfer, mcro/macro thermo/ flud dynamcs, optmzaton, and CFD. Recently, hs researches are focus on the semconductor and dsplay manufacturng process and the aerodynamcs and optmzaton of Wng-In- Ground Effect Vehcles. 4