Computational Fluid Dynamic Modelling of Particle Charging and Collection in a Wire-to-Plate Type Single-Stage Electrostatic Precipitator

Transcription

1 Aerosol and Air Quality Research, 18: , 018 Coyright Taiwan Association for Aerosol Research ISSN: rint / online doi: /aaqr Comutational Fluid Dynamic Modelling of Particle Charging and Collection in a Wire-to-Plate Tye Single-Stage Electrostatic Preciitator Ji-Woon Park, Chul Kim, Jaehong Park, Jungho Hwang * School of Mechanical Engineering, Yonsei University, Seoul , Korea ABSTRACT Electrostatic reciitators (ESPs) have been widely used to control articulate ollutants, which adversely affect human health. In this study, a comutational fluid-dynamic model for turbulent flow, article trajectory, and article charging in ESPs is resented using a re-develoed corona discharge model (Kim et al., 010), wherein electric field and sace charge distributions in the lasma region are numerically calculated. The ESP under consideration is a wire-to-late singlestage ESP, which consists of a series of discharge wires and two collecting lates. Two different kinds of articulates are considered in this study; fly ash and sucrose articles. Fly ash was selected because many ESPs have been utilized in coalfired ower lants to cature fly ash articles generated from combustion. Sucrose was selected to comare our numerical calculation results with exerimental data found in literature. The electrical characteristics of the ESP, article trajectories, article charge numbers, and collection efficiencies under various oerating conditions are demonstrated. For fly ash, the overall collection efficiencies based on article mass are 61, 86, 95, and 99% at 45, 50, 55, and 60 kv, resectively, at a flow velocity of 1 m s 1. Keywords: Electrostatic reciitator; Corona discharge; Plasma region; Particle charge; PM.5. INTRODUCTION Removal of articles smaller than.5 µm in diameter (PM.5 ), which are considered as risk factors of diseases, has gained considerable attention (Wen et al., 015). Electrostatic reciitators (ESPs) have been widely used to remove airborne articles in common industrial articulate control system owing to its high efficiency under low ressure dro (Huang and Chen, 00; Lin and Tsai, 010; Ruttanachot et al., 011; Zhu et al., 01). The otential of this technique has been established by industries; however, the technique exhibits certain limitations, including a low collection efficiency of submicron articles (Gouri et al., 013). In a wire-to-late tye single-stage ESP, sace charges are formed by air ions generated as a result of corona discharge when a high voltage is alied between discharge wires and grounded collecting lates. The articles in the flowing gas are charged by the sace charges and then collected on the ground lates by an electrostatic field. Because exerimental investigation of ESPs is exensive (Adamiak, 013), numerical analysis has been widely used * Corresonding author. Tel.: , Fax: address: hwangjh@yonsei.ac.kr for design and erformance evaluation uroses. The electrostatic reciitation involves a comlex interactive hysical rocess among turbulent flow, electric field, sace charge distribution and article charging and motions. Modelling of corona discharge is comlicated and challenging even for a straightforward electrode configuration (Adamiak, 013). Talaie et al. (001b) develoed a corona discharge model based on the fact that increasing the alied voltage increases the lasma region, where ions are generated owing to electron-imact reactions around a discharge wire. Ion density values at the lasma boundary were calculated by emirical equations. Chen and Davidson (00) numerically investigated the one-dimensional corona lasma region around a discharge wire by considering Katzov s hyothesis. They reorted that the ion densities remained relatively constant in the lasma region. In a study of our grou (Kim et al., 010), a comutational methodology was roosed for calculating the lasma region thickness based on the study of Chen and Davison (00). The ion densities were obtained using an analytical method and were defined both at the lasma boundary and within the lasma region. Kim et al. (010) alied their methodology to wire-to-duct ESP, and the results were in reasonable agreement with exerimental data of revious studies (Penney and Matick, 1960; MacDonald et al., 1977; Lawless and Sark, 1980; Ohkubo et al., 1986) with various geometries. For analysis of article charging and motion in the ESP, the Lagrangian aroach has been widely used with corona

2 Park et al., Aerosol and Air Quality Research, 18: , discharge models (Goo and Lee, 1996; Varonos et al., 00; Talaie, 005; Lei et al., 008; Farnoosh et al., 011; Liu et al., 015; Wen et al., 016). Goo and Lee (1996) develoed a numerical model for an ESP using the Lagrangian aroach which uses the concet of time-series analysis for the article motion in a turbulent flow field. They reorted that the Lagrangian aroach was more effective than the Eulerian aroach to analyze article motion in an ESP because the effects of article inertia and non-uniform diffusivity on the article motion were omitted in the Eulerian aroach. Moreover, the Lagrangian aroach can be combined with dynamic charging to obtain article charge numbers, which are affected by article size, article ermittivity, electric field intensity and air ion concentration. Talaie (005) alied the work of Talaie et al. (001b) to simulate the erformance of wire-duct single-stage ESPs. Talaie et al. (001a) stated that the main advantage of the Lagrangian aroach in comarison with the Eulerian aroach is to consider the effect of article size. This article reorts a comutational study that is an exansion of the study of Kim et al. (010) that simulated the electric field and sace charge distributions of a wireto-late ESP by taking into account the lasma region. In this article, flow velocities and article trajectories were calculated and article collection efficiency in a wire-to-late ESP was redicted. The standard k ε turbulent model based on Finite Volume Method and the Lagrangian aroach with dynamic article charging were added to the corona discharge model of Kim et al. (010) to solve the turbulent flow and the motion and charging of articles in the ESP. The GAMBIT.0 code and FLUENT 14 with user-defined function (UDF) were used. METHODOLOGY The standard k ε turbulent model was used to solve gas flow in FLUENT 14. Particle trajectory was calculated by using discrete hase model (DPM). Poisson equation, charge conservation equation and a storing lace were defined by using three user-defined scalars (UDSs) as well as the userdefined function (UDF) of DEFINE_SOURCE. The effect of electrical body forces on the gas flow was observed by the UDF of DEFINE_SOURCE. Sace charge distribution was udated by using the UDF of DEFINE_EXECUTE_AT_ END. Electrical otentials and charge densities were converted into electric fields and current densities, resectively, by using the UDF DEFINE_ON_DEMAND. The UDF of DEFINE_SCALAR_UPDATE was used to udate article charge with a traezoidal rule. Two UDFs of DEFINE_DPM_BODY_FORCE and DEFINE_DPM_ DRAG were used to aly electrical body forces on the articles and take into consideration the article sli, resectively. Gas Flow Field For incomressible flow, the equations of continuity and momentum with electrical body force ( F E ) can be exressed as follows: u 0 T u u I T uu F F E qe where ρ is the density of air, u is the flow velocity, is the ressure, I is the unit vector, µ is the dynamic viscosity of air, µ T is the eddy viscosity defined as ρc μ k ε, k is the turbulent kinetic energy, ε is the dissiation rate, q is the charge density of gaseous ions and E is the electric field. In the standard k ε turbulent flow model, the equations governing the turbulent kinetic energy and dissiation rate are exressed as follows: uk T / k G u T / k C1 / kg C / k G where σ ε and σ k are the turbulent Prandtl numbers for k and, resectively, and G is the roduction rate of k (Launder and Salding, 1974). The values of the constants are as follows: C µ = 0.09, C 1 = 1.44, C = 1.9, σ ε = 1.0 and σ k = 1.3. Particle Trajectory and Charging Particles enter the inlet and migrate toward the outlet under the couled effect of the electric field, sace charge and flow field. The article trajectory is exressed using the following equations: du 18 u uf dt d C F c qe/ m where u is the article velocity, d is the article diameter, ρ is the mass density of a article, F is the Coulomb s force exerted on a article, m is the mass of a article and q is the electrical charge of a article (= n e; n is the charge number of a article and e is the charge of an electron). C c is the sli correction factor defined as follows (Allen and Raabe, 198): d C ex 0.39 c d where is the mean free ath of air (0.066 µm at 1 atm and 93 K) (Hinds, 1999). Particles in an ESP are charged by air ions generated by E (1) () (3) (4) (5) (6) (7) (8)

3 59 Park et al., Aerosol and Air Quality Research, 18: , 018 corona discharge. The article charge number (n ) is estimated using diffusion and field charging theories (Reist, 1993): n = n d + n f (9) dn dt dn dt d ndk Ee d ciniex 4 kbd T f n f KEeZiNins1 ns 3 Ed ns 4K e E (10) (11) (1) where n d is the number of charges on a article by diffusion charging, n f is the number of charges on a article by filed charging, n s is the saturation charge attained after sufficient time at a secified charging condition, k B is the Boltzmann constant, T is the absolute temerature, K E is the constant of roortionality = 1/4πε 0, ε 0 is the ermittivity of vacuum, c i is the mean thermal seed of air ions (40 m s 1 at 98 K and 1 atm), N i is the ion number concentration (assumed as N i = q/e), t is the residence time, Z i is the ion mobility (1.4 m V 1 s 1 of ositive ions) and ε is the relative ermittivity of the article. Electric Field and Sace Charge Distributions The distributions of electric otential (V) and ionic current density ( j) are governed by the Poisson equation and charge conservation equation as follows: q V (13) 0 j 0 (14) For a uniform temerature and by omitting the diffusion effect, the ionic current density is defined as follows: j q uz E qz E i i (15) where E V. The flow velocity u is omitted in this study as it is negligible (under 10 m s 1 ) comared to the ion velocity ( ZE i, in the order of 10 6 m s 1 ). Once j is obtained, the electric current can be calculated by integrating Eq. (15) over the area of the ground electrode. MODEL DESCRIPTION In order to validate the roosed CFD model, the electrodes configuration and exerimental data of Lawless and Sarks (1980) were used. They erformed exeriments on a high-voltage corona wire-to-late configuration in the absence of articles. The geometry configuration and comutational region of this study are illustrated in Fig. 1 (eight discharge wires and two collecting lates). The wire-to-late distance (S), wire-towire distance (D), wire radius (r w ) and duct length (L) were mm, 8.6 mm, 1.59 mm and m, resectively. The lasma boundary was located at distance r from the Fig. 1. Geometry configuration (D).

4 Park et al., Aerosol and Air Quality Research, 18: , center of a wire. The radius (r ) of the lasma region is defined such that the reduced electric field strength (E/N = Td) is higher than 10 Td (N is the neutral gas density, which can be obtained from the ideal gas law) (Grabarczyk, 013). The comutational region contained 37,39 meshes with a quadratic shae (See Fig. ). The mesh located in the roximity of a wire was substantially refined to 5 µm owing to the large gradient of the electric otential in the roximity of the wire. Boundary conditions used in this study are summarized in Table 1. u a and V a reresent the inlet gas flow velocity and alied voltage on a wire, resectively. q a is the charge density in the lasma region. The initial article charge number was assumed to be zero. The semi-imlicit method for ressure-linked equations (SIMPLE) algorism was used to secify ressure-velocity couling in the equations of continuity and momentum. In order to solve the couling of the Poisson equation and charge conservation equation for secified alied voltage, the discharge model resented by Kim et al. (010) was used. Two different articles were considered in this study; fly ash and sucrose articles. Fly ash was selected because many ESPs have been utilized in coal-ower lants to cature fly ash articles generated from combustion. Sucrose was selected to comare our numerical calculation results with exerimental data of Huang and Chen (00). Huang and Chen (00) carried out a controlled exerimental study as art of their health and toxicological studies to understand the behaviour ultrafine articles (diameter < 100 nm) in ESPs. Sucrose aerosols of about numbers cm 3 in total number concentration, 4 nm of geometric mean diameter (GMD), and 1.8 of geometric standard deviation were generated with an electrosray aerosol generator. Using these sucrose aerosol articles, Huang and Chen (00) Fig.. Comutational domain: (a) Meshes in total domain; (b) Meshes in a single wire domain. Table 1. Boundary conditions. Equations Inlet Outlet Symmetric Discharging wire Collecting late Continuity and Momentum u = u a = 1 atm y u = 0 u = 0 u = 0 Poisson Charge conservation x x x x Motion of article u = 0 x x x x x u = 0 y y y y V = V a q = q a (In the lasma region) V = V a q = q a (In the lasma region) y y y u = 0 u = 0 u = 0

5 594 Park et al., Aerosol and Air Quality Research, 18: , 018 carried out erformance tests of their Lab-made ESPs. The relative ermittivity (ε and the mass density of sucrose articles were assumed as 3.3 and 1.59 g cm 3, resectively. The geometric mean diameter of fly ash articles was assumed as 0.1 µm based on the exerimental results of Strand et al. (00). The relative ermittivity and the mass density of fly ash articles were assumed as 3.0 (Baoyi et al., 01) and g cm 3 (Ghosal and Self, 1995), resectively. The number distribution of articles of size d that enter the ESP was assumed to be a log-normal distribution; 1 ln d ln GMD dfn ex d ln d ln g ln g (16) where σ g is the geometric standard deviation, which was assumed to be.0 (Ghosal and Self, 1995). The number distribution was also converted into the mass distribution using the following equation: df mi m i dfn (17) i M where m i is the total mass of the articles in grou i, and M is the total mass of all the grous (= m i ). RESULTS AND DISCUSSION Numerical calculations were conducted for various alied voltages in the range of kv and various duct flow velocities in the range of 0 8 m s 1. The temerature and ressure were 93 K and 1 atm, resectively. For each flow velocity, the satial distributions of voltage and electric field between the eight discharging wires and a collecting late were obtained by solving the Poisson equation (Eq. (3)) as well as the Lalace equation ( in Eq. (3)). The satial distributions of voltage are illustrated in Figs. 3(a) and 3(b), resectively, for and q 0, when the alied voltage was 60 kv. Fig. 3(c) illustrates the electric field distribution along the Y-axis at X = 0 when. For each value of the alied voltage, the electric field was the maximum at the wire surface; however, it decreased remarkably with distance along the Y direction. The electric field at the discharging wire (Y = r w = 1.59 mm) increased from 5.79 MV m 1 to 7.7 MV m 1 as the alied voltage increased from 45 kv to 60 kv. However, when the Poisson equation was used for calculation (Fig. 3(d)), the electric field at the discharge wire was constant at 5.33 MV m 1 for various alied voltages. This henomenon can be exlained with Katzov s hyothesis that the electric field at the surface of a discharging wire remains constant at the corona onset voltage for all values of voltage above the onset voltage. The corona onset electric field at the discharge wire can be obtained using Peek s law as follows; E 0 = f (δ (δ/r w ) 0.5 ) [V m 1 ] (18) where f is a factor that accounts for wire roughness, δ is the air density at 1 atm and 5 C. For the geometry configuration used in this study, Eq. (18) results in 5.49 MV m 1, which is aroximately equal to the calculated value (5.33 MV m 1 ). It is noteworthy that in Fig. 3(d), the electric field decreases in the roximity of the discharge wire and increases in the roximity of the collecting late, when comared to Fig. 3(c). The henomenon occurs in the resence of sace charges as the net electric field formed at any location is equal to the vector sum of ± Y direction fields roduced by all the sace charges (Suerosition rincile). The results resented in Fig. 3(d) were in reasonable agreement with those of revious studies (Sekar and Stomberg, 1981; McLean, 1988). The charge density (q) distribution when the alied voltage was 60 kv is resented in Fig. 4(a). The variation of charge densities at X = 0 along the Y-axis under various voltages is illustrated in Fig. 4(b). The current densities at the collecting lates were also calculated for various alied voltages (see Fig. 5). The results were in agreement with the exerimental data of Lawless and Sarks (1980), which was also verified by Kim et al. (010). Fig. 6 illustrates trajectories of fly ash articles under various alied voltages when the flow velocity was 1 m s 1. Three articles each of size 0.1 µm entered the ESP at three ositions, namely osition #0 (Y = 0), osition #1 (Y = 8.6 mm) and osition # (Y = 57 mm), resectively. The trajectories of these three articles are reresented as P0, P1, and P, resectively. Each article gradually acquired sace charge as the article advanced with the flow along the X-direction and aroached the collecting electrode located at Y = mm (see Fig. ). When the alied voltage was 45 kv, all the three articles finally escaed from the ESP; however, all of those articles were catured when the alied voltage increased to 60 kv. Fig. 7 illustrates that these three articles gradually acquired sace charges when they advanced along the X-direction. The article migrating in the roximity of the discharging wire (injected at osition #0) carried a higher charge than the other two articles (injected at ositions #1 and #) owing to the stronger electric field (~5 MV m 1 ; see Fig. 3(d)) and higher charge density (see Fig. 4(b)). The charge numbers of articles injected at ositions #1 and # were aroximately equal as these two articles assed through regions which had similar electric fields; for examle, when the alied voltage was 60 kv, the electric field in 40 mm < Y < mm varied from 0.38 MV m 1 to 0.47 MV m 1 (see Fig. 3(d) and Eq. (1)). Fig. 8(a) resents the collection efficiency of ESP for 0.1 µm articles (fly ash) under various flow velocities and various alied voltages. The collection efficiency is defined as Nt % 100 (19) N i where N t and N i are the number of articles traed on the collecting late and the number of articles injected to the inlet, resectively. Two hundred articles were injected to calculate the collection efficiency. For any alied voltage, the collection efficiency decreased exonentially as the

6 Park et al., Aerosol and Air Quality Research, 18: , Fig. 3. Electrical characteristics: (a) Voltage distribution solved by Lalace equation at 60 kv of alied voltage; (b) Voltage distribution solved by Poisson equation at 60 kv of alied voltage; (c) Variations of electric field solved by Lalace equation along Y-axis for various alied voltages; (d) Variations of electric field solved by Poisson equation along Y-axis for various alied voltages. flow velocity increased. These results can be substantiated with the Deutsch-Anderson equation (Hinds, 1999), which is exressed as follows: WL 1ex US (0) where U is the flow velocity. The electrical migration velocity (W) is exressed as W = Z E; here, Z is the electrical mobility, which is defined as follows: Z nec c (1) 3d When the flow velocity increases, the residence time decreases, and therefore, the charge number decreases (see Eqs. (9) (1)). The decrease in charge number results in the decrease of electrical mobility, electrical migration velocity and collection efficiency. Fig. 8(b) resents the collection efficiencies of ESP for various fly ash article sizes and various alied voltages at flow velocity of 1 m s 1. The collection efficiency trend followed that in literatures (McLean, 1988; Yoo et al., 1997; Ylätalo and Hautanen, 1998; Huang and Chen, 00). The efficiency was the minimum for article sizes of µm regardless of the alied voltage. For examle, when the alied voltage was 60 kv, 0.1 µm articles and 0.5 µm articles had highly similar collection efficiencies, 7% and

7 596 Park et al., Aerosol and Air Quality Research, 18: , 018 Fig. 4. (a) Charge density distribution at 60 kv of alied voltage; (b) Variation of charge density along the Y-axis for various alied voltages. Fig. 5. (a) Current density distribution at 60 kv of alied voltage; (b) Average current density on the collecting lates for various alied voltages. 68%, resectively, notwithstanding the significantly dissimilar charge numbers, 4.17 and 44.4, resectively. However, it is interesting to note that the 0.1 µm articles and 0.5 µm articles had highly similar electrical mobilities (Z ) of 115 µm V 1 s 1 and 109 µm V 1 s 1, resectively. For articles smaller than 0. µm, diffusion charging is the redominant mechanism even in the resence of electrostatic fields. Moreover, the article charge (n ) is roortional to d ln(1 + αd ), where α is the constant related to ion-article collisions. Considering that the Cunningham correction factor can be aroximated as 3.69 (λ/d ) 1/ for the intermediate range of λ/d (Lee and Liu, 1980), the article

8 Park et al., Aerosol and Air Quality Research, 18: , Fig. 6. Trajectories of fly ash articles (0.1 µm) when the flow velocity was 1 m s 1 at various alied voltages: (a) 45 kv; (b) 50 kv; (c) 55 kv; (d) 60 kv. Fig. 7. Variation of charge number for fly ash articles (0.1 µm) when the flow velocity was 1 m s 1 at various alied voltages: (a) 45 kv; (b) 50 kv; (c) 55 kv; (d) 60 kv. mobility (Z ) is roortional to d 1/ for d < ~00 nm, and therefore, η decreases as d increases. For articles larger than aroximately 00 nm in diameter, field charging is the dominant mechanism, and n ~ d. Therefore, the article mobility is roortional to d 1/ for d > ~00 nm, and hence, η increases as d increases. As the alied voltage increases, the collection efficiencies also increase notwithstanding the sizes of articles because of the higher article charging. The correlations between the electrical mobility and collection efficiency at various alied voltages are illustrated in Fig. 9. A linear correlation was obtained for each alied voltage. For WL/US << 1 (~10 6 in this study), the Deutsch-Anderson equation (Eq. (0)) is exressed as follows, from the Tayler series; L L W Z E US US () Therefore, the collection efficiency increased linearly with electrical mobility under constant alied voltage. The rate of increase becomes higher with alied voltage. Fig. 10 illustrates the effect of ESP on fly ash article number and mass distributions under various alied voltages at flow velocity of 1 m s 1. The overall collection efficiencies based on article number were 30, 54, 75 and 90% at alied voltages of 45, 50, 55 and 60 kv, resectively. At the

9 598 Park et al., Aerosol and Air Quality Research, 18: , 018 Fig. 8. Collection efficiency at flow velocity 1 m s 1 ): (a) for articles of 0.1 µm under various alied voltages and flow velocities; (b) for various article diameters under various alied voltages. Fig. 9. Correlations between the electrical mobility and the collection efficiency for various article diameters and alied voltages at flow velocity 1 m s 1. downstream, the roortion of µm articles increased owing to their low collection efficiency. As the emission regulations of PM are tyically based on mass, the article number distribution was converted into mass distribution. Then, the mode diameter of the ustream articles was shifted to 3 µm. The overall collection efficiencies based on article mass were 61, 86, 95 and 99% at 45, 50, 55 and 60 kv, resectively. Similar lots could be reared for the various flow velocities. So far, our CFD model for turbulent flow, article trajectory, and article charging was alied to a wire-tolate tye single-stage ESP which consisted of a series of eight discharge wires and two collecting lates. With fly ash articles having a log-normal size distribution (GMD = 0.1 µm, geometric standard deviation =.0), the electrical characteristics of the ESP, article trajectories, article charge numbers, and collection efficiencies under various oerating conditions were demonstrated. Our CFD model was also alied to another wire-tolate tye single-stage ESP (three discharge wires and two collecting lates) which was used in an exerimental study of Huang and Chen (00). In their study, sucrose aerosols having a log-normal size distribution (GMD = 4 nm, geometric standard deviation = 1.8) were used. The wireto-late distance (S), wire-to-wire distance (D), wire radius (r w ) and length of collecting late (L) were 60 mm, 4 mm,

10 Park et al., Aerosol and Air Quality Research, 18: , Fig. 10. Particle distributions of fly ash articles under various alied voltages at ustream and downstream of the ESP at flow velocity 1 m s 1 : (a) Number; (b) Mass. Fig. 11. Calculated and exerimental collection efficiencies for different article diameters under different alied voltages. 0.3 mm and 300 mm, resectively. The air flow rate and the alied voltage were 100 L min 1 and 6.4 kv, resectively. Fig. 11 shows exerimental data of the ustream and downstream size distributions of sucrose aerosols as well as collection efficiencies for different article sizes. Fig. 11 also shows that our CFD calculation results were in good agreements with the exerimental data of Huang and Chen (00). It should be noted that the effect of humidity (water vaour concentration) on corona discharge was not considered in our CFD model. In a real situation, the corona onset voltage decreases with the increase of water vaour concentration. Therefore, a correction function of water vaour concentration should be added to a conventional Peek formula (Wang and You, 013). CONCLUSIONS A CFD model for turbulent flow, article trajectory, and article charging was develoed and alied to two different wire-to-late single-stage ESPs. Particle trajectories, article charging, and collection efficiencies were simulated for various flow velocities, article sizes, and alied voltages. The results were in reasonable agreement with those of revious studies. Two kinds of articulates were considered in this study; fly ash and sucrose, which ossess different surface roerties and carry different levels of charges. The differences in size, density, and relative ermittivity affected article charging, trajectories, and collection efficiencies. The roosed model can be used to design ESPs for removing or samling articulate materials in the air.

11 600 Park et al., Aerosol and Air Quality Research, 18: , 018 ACKNOWLEDGEMENT This research was suorted by Railroad Technology Research Program (17RTRP-B ) funded by Ministry of Land, Infrastructure and Transort of Korean Government and by Korea Ministry of Environment (MOE) as Advanced Technology Program for Environmental Industry. This research was also suorted by the Korea Institute of Energy Technology Evaluation and Planning (KETEP) and the Ministry of Trade, Industry & Energy (MOTIE) of the Reublic of Korea (No ). REFERENCES Adamiak, K. (013). Numerical models in simulating wirelate electrostatic reciitators: A review. J. Electrostat. 71: Allen, M.D. and Raabe, O.G. (198). Re-Evaluation of Millikan s oil dro data for the motion of small articles in air. J. Aerosol Sci. 6: Baoyi, L., Yuing, D. and Shunhua, L. (01). The electromagnetic characteristics of fly ash and absorbing roerties of cement-based comosites using fly ash as cement relacement. Constr. Build. Mater. 7: Chen, J. and Davidson, J. H. (00). Electron density and energy distributions in the ositive DC corona: Interretation for corona-enhanced chemical reactions. Plasma Chem. Plasma Process. : Farnoosh, N., Adamiak, K. and Castle, G.P. (011). Numerical calculations of submicron article removal in a sike-late electrostatic reciitator. IEEE Trans. Dielectr. Electr. Insul. 18: Ghosal, S. and Self, S.A. (1995). Particle size-density relation and cenoshere content of coal fly ash. Fuel 74: Goo, J.H. and Lee, J.W. (1996). Monte-Carlo simulation of turbulent deosition of charged articles in a late-late electrostatic reciitator. Aerosol Sci. Technol. 5: Gouri, R., Zouzou, N., Tilmatine, A. and Dascalescu, L. (013). Enhancement of submicron article electrostatic reciitation using dielectric barrier discharge in wireto-square tube configuration. J. Electrostat. 71: Grabarczyk, Z.J. (013). Laboratory ignition of hydrogen and carbon disulhide in the atmosheric air by ositive corona discharge. J. Electrostat. 71: Hinds, W.C. (1999). Aerosol technology: Proerties, behavior, and measurement of airborne articles, nd Edition. Wiley, New York. Huang, S.H. and Chen, C.C. (00). Ultrafine aerosol enetration through electrostatic reciitators. Environ. Sci. Technol. 36: Kim, C., Noh, K.C. and Hwang, J. (010). Numerical investigation of corona lasma region in negative wireto-duct corona discharge. Aerosol Air Qual. Res. 10: Launder, B.E. and Salding, D.B. (1974). The numerical comutation of turbulent flows. Comut. Comut. Methods Al. Math. 3: Lawless, P.A. and Sarks, L.E. (1980). A mathematical model for calculating effects of back corona in wire-duct reciitators. J. Al. Phys. 51: Lei, H., Wang, L.Z. and Wu, Z.N. (008). EHD turbulent flow and Monte-Carlo simulation for article charging and tracing in a wire-late electrostatic reciitator. J. Electrostat. 66: Lin, G.Y. and Tsai, C.J. (010). Numerical modelling of nanoarticle collection efficiency of single-stage wirein-late electrostatic reciitator. Aerosol Sci. Technol. 44: Liu, Q., Zhang, S.S. and Chen, J.P. (015). Numerical analysis of charged article collection in wire-late ESP. J. Electrostat. 74: McDonald, J.R., Smith, W.B., Sencer III, H.W. and Sarks, L.E. (1977). A mathematical model for calculating electrical conditions in wire-duct electrostatic reciitation devices. J. Al. Phys. 48: McLean, K.J. (1988). Electrostatic reciitators. IEE Proc. A 135: Ohkubo, T., Nomoto, Y., Adachi, T. and McLean, K.J. (1986). Electric field in a wire-duct tye electrostatic reciitator. J. Electrostat. 18: Penney, G.W. and Matick, R.E. (1960). Potentials in DC corona fields. Trans. Am. Inst. Electr. Eng. Part 1 79: Reist, P.C. (1993). Aerosol science and technology, nd Edition. McGraw-Hill, Singaore. Ruttanachot, C., Tirawanichakul, Y. and Tekasakul, P. (011). Alication of electrostatic reciitator in collection of smoke aerosol articles from wood combustion. Aerosol Air Qual. Res. 11: Sekar, S. and Stomberg, H. (1981). On the rediction of current-voltage characteristics for wirelate reciitators. J. Electrostat. 10: Strand, M., Pagels, J., Szila, A., Gudmundsson, A., Swietlicki, E., Bohgard, M. and Sanati, M. (00). Fly ash enetration through electrostatic reciitator and flue gas condenser in a 6 MW biomass fired boiler. Energy Fuel 16: Talaie, M.R., Fathikaljahi, J., Taheri, M. and Bahri, P. (001a). Mathematical modeling of double-stage electrostatic reciitators based on a modified Eulerian aroach. Aerosol Sci. Technol. 34: Talaie, M.R., Taheri, M. and Fathikaljahi, J. (001b). A new method to evaluate the voltage-current characteristics alicable for a single-stage electrostatic reciitator. J. Electrostat. 53: Talaie, M.R. (005). Mathematical modeling of wire-duct single-stage electrostatic reciitators. J. Hazard. Mater. 14: Varonos, A.A., Anagnostooulos, J.S. and Bergeles, G.C. (00). Prediction of the cleaning efficiency of an electrostatic reciitator. J. Electrostat. 55: Wang, X. and You, C. (013). Effect of humidity on negative corona discharge of electrostatic reciitators. IEEE Trans. Dielectr. Electr. Insul. 0: Wen, T.Y., Krichtafovitch, I. and Mamishev, A.V. (015). Reduction of aerosol articulates through the use of an

12 Park et al., Aerosol and Air Quality Research, 18: , electrostatic reciitator with guidance-late-covered collecting electrodes. J. Aerosol Sci. 79: Wen, T.Y., Krichtafovitch, I. and Mamishev, A.V. (016). Numerical study of electrostatic reciitators with novel article-traing mechanism. J. Aerosol Sci. 95: Ylätalo, S.I. and Hautanen, J. (1998). Electrostatic reciitator enetration function for ulverized coal combustion. Aerosol Sci. Technol. 9: Yoo, K.H., Lee, J.S. and Oh, M.D. (1997). Charging and collection of submicron articles in two-stage arallellate electrostatic reciitators. Aerosol Sci. Technol. 7: Zhu, J., Zhao, Q., Yao, Y., Luo, S., Guo, X., Zhang, X., Zeng, Y. and Yan, K. (01). Effects of high-voltage ower sources on fine article collection efficiency with an industrial electrostatic reciitator. J. Electrostat. 70: Received for review, May 1, 017 Revised, Setember 14, 017 Acceted, Setember 30, 017