Journal of Electrostatics 69 (2011) 419e428

Contents lists available at ScienceDirect

Journal of Electrostatics
journal homepage: www.elsevier.com/locate/elstat

Three-dimensional analysis of electrohydrodynamic flow in a spiked

electrode-plate electrostatic precipitator
N. Farnoosh*, K. Adamiak, G.S.P. Castle
Department of Electrical and Computer Engineering, The University of Western Ontario, London, Ontario N6A 5B9, Canada

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

Article history: A three-dimensional numerical model has been created to evaluate the electrical and electro-
Received 12 October 2010 hydrodynamic characteristics of a single spiked wire-plate electrostatic precipitator. The hybrid Finite
Received in revised form Element e Flux Corrected Transport numerical technique is used for solving the Poisson and current
5 May 2011
continuity equations to estimate the electric potential and ion charge density distributions in the
Accepted 15 June 2011
precipitation channel. The fully three-dimensional turbulent airflow distribution is calculated using the
Available online 5 July 2011
commercial FLUENT software assuming a standard ke3 turbulence model. A non-uniform corona
discharge is assumed, as it is produced along the electrode in the form of a flat tape with some number of
Keywords:
Electrohydrodynamics
spikes. The EHD secondary flow pattern and its interaction with the main airflow in different planes
Spiked electrode along the precipitation channel are examined for different voltages applied to the corona spiked elec-
Electrostatic precipitator trode. The numerical results are compared with experimental data published in the literature.
Corona discharge Ó 2011 Elsevier B.V. All rights reserved.

1. Introduction of a homogeneous discharge along smooth wire electrodes and only

in a few publications more complicated electrode configurations
The electrostatic precipitator (ESP) has been used as a pollution were assumed, such as tuft or spiked-type corona electrode [1,2],
control device over many decades. In this device, the dust particles barbed plate [3] and barbed wire electrodes [4]. In these cases the
are charged by the ionic and electronic bombardment in the corona discharge is localized at some points along the corona elec-
precipitator channel, pushed toward the collecting plates by the trode and the ion charge density distribution near the corona elec-
electric forces and deposited on them. The electrostatic forces trode is strongly non-uniform. In practical applications with
exerted on the ionic space charge by the electric field (Coulomb a negative corona discharge isolated tufts with unstable intensity and
forces) induce the secondary electrohydrodynamic (EHD) flow, or location are generated along the corona electrode. Commercial
the ionic wind, which increases the airflow turbulence in the electrodes, such as barbed wire electrodes, plane strip electrodes and
channel. The complex processes in an ESP involve both electrical studded tube electrodes, are conventionally designed to predeter-
and mechanical effects, which have been examined by many mine the location of tufts and to achieve steady equally spaced corona
authors and many numerical techniques have been proposed in the tufts. The airflow patterns, electrical conditions and particle transport
literature for solving the governing equations. become essentially three-dimensional in these configurations.
Although the basic processes involved in electrostatic precipi- Vast majority of the published studies were limited to uniform
tation, such as particle charging, transport and collection are well current distribution along a smooth corona discharge electrode,
known, the fluid flow aspects of this process are still not completely a few authors considered negative corona discharge assuming tuft
understood, even though the interaction between the main and or spiked corona discharge electrode. The numerical study of
secondary EHD flows has been studied theoretically and experi- Yamamoto and his co-workers of multiple-tuft wire discharges
mentally during the past decades. indicated that each tuft point creates a pair of donut shaped rings of
Different model complexities have been reported in the literature airflow [1], which are more organized in a laminar airflow than in
for investigating the electrohydrodynamic effects in an ESP. The a turbulent one [5]. They also pointed out that a transition from the
geometry of the discharge electrodes plays a significant role on well organized spiral motion to the complete mixing regime
determining the electrical and aerodynamic characteristics of the depends on the tuft and wire-to-plate spacing. The tuft-corona
system. Almost all numerical studies considered the simplified case discharge also substantially increases the turbulence level. It was
shown in Ref. [1] that by increasing the applied voltage the number
* Corresponding author. of tufts along the wire increases, they become more stable and
E-mail address: [email protected] (N. Farnoosh). more uniformly distributed.

0304-3886/$ e see front matter Ó 2011 Elsevier B.V. All rights reserved.
doi:10.1016/j.elstat.2011.06.002
420 N. Farnoosh et al. / Journal of Electrostatics 69 (2011) 419e428

long-elliptic spiral flows in the direction of the airflow between

ground planes after adding the primary airflow into the model.
Many visual observations and velocity measurements have been
reported in the literature, which illustrate the EHD secondary flow in
ESPs [7e11]. The EHD flow pattern in a small scale barbed-wire
precipitator was examined in Ref. [12] using Laser Doppler
measurements. According to these results, each discharge point
along the corona wire creates a pair of highly structured recirculation
vortices and the barb spacing determines the EHD flow structure.
Davidson and McKinney [13] used hot-film anemometry
measurements to characterize the EHD flow in a barbed plate-to-
plate precipitator. They proposed the barbed plate electrode as
a useful design which effectively reduces the scale of corona
induced EHD flow and thus decreases particle mixing. However,
greater turbulence was generated in this precipitator than in
a conventional wire-plate precipitator.
More recently, 3-D Particle Image Velocimetry (PIV) measure-
Fig. 1. Configuration of a 3-D single spiked electrode-plate ESP model.
ments were carried out by Mizeraczyk group to study the EHD flow
patterns and particle trajectories in various ESP configurations
Yamamoto and his co-workers also did a three-dimensional assuming different channel widths [14], corona electrode orienta-
EHD simulation for modeling a typical industrial ESP consisting of tions [15] and smooth plate or flocking plane collecting electrodes
a spiked-type discharge electrode and convexeconcave-type col- (two stainless-steel plane meshes covered with nylon flocks) with
lecting electrode using Finite Difference approximation and the the corona wire located along the airflow, where the lower airflow
successive-over-relaxation (SOR) method [2,6]. In their model the velocity near the flocking plane electrodes prevented the re-
spikes were alternatively located along the corona electrode and entrainment of the deposited particles and increased the collec-
pointed toward the ground planes. They reported a zigezag motion tion efficiency of fine particles up to 94% [16]. In Ref. [17], they
generated in the direction of the main airflow, which changed to investigated the secondary EHD flow and particle collection
spiral rings for the higher applied voltages. According to their efficiency for submicron dust particles (0.25e1.5 mm) in a spike-
simulations, the secondary flow distribution without the external plate ESP under negative and positive DC voltage using 3-D
airflow consists of a pair of long-elliptic and circulatory cells PIV measurements. Their experimental results demonstrated
between spiked points along the wire, which changes to a pair of a complex turbulent airflow with a structure depending on the

Fig. 2. Schematic drawing of the spike-discharge electrode (top view).

Fig. 3. Electric potential contours in z ¼ 0 plane of ESP (number of contours 50, applied voltage 30 kV).
 N. Farnoosh et al. / Journal of Electrostatics 69 (2011) 419e428 421

Fig. 4. Ion charge density distribution contours in (a) y ¼ 0 and (b) z ¼ 0 planes (number of contours is 25, the maximum ion charge density on each spike is 485 mC/m3and the
applied voltage is 30 kV).

applied voltage level and the position of measuring plane with downstream parallel to the airflow and is 200 mm long, 1 mm thick,
respect to the spike tips. They concluded that the airflow in the and 30 mm tip to tip wide (Fig. 2). A negative high voltage supply is
spike-plate ESP has a strong three-dimensional variation. The connected to the spiked electrode, which creates gas ionization
velocity measurements showed that the turbulence intensity (corona discharge) at the spike tips generating ionic space charge into
generated when a negative voltage is applied to the corona spiked the channel. A flat velocity profile of 0.6 m/s at the channel inlet, zero
electrode is higher than for the case of positive voltage. However, gauge pressure at the outlet and zero velocity components at all
no comprehensive theoretical investigations on EHD secondary electrodes and walls are considered as the fluid boundary conditions.
flow field in ESPs with spike-discharge electrode could be found in The air dynamic viscosity m ¼ 1.57  105 kg/m$s, air density
literature. r ¼ 1.205 kg/m3 and ion mobility k ¼ 2.4  104 m2/V$s are chosen in
In this paper, the laboratory spike-plate ESP, experimentally calculations.
investigated in Ref. [17], is simulated, implementing a 3-D numer- The hybrid Finite Element Method (FEM) e Flux Corrected
ical technique. In Section 2, the 3-D computational model is fully Transport (FCT) technique has been implemented for solving the
described and the numerical techniques for solving the governing governing equations (Poisson and current continuity) to predict the
equations are briefly reviewed. Non-uniform velocity distribution electrical parameters (electric potential and ion charge density
patterns generated along and across the ESP channel due to the distributions) for different applied voltages. The full description of
induced EHD secondary flow are discussed in Section 3.

2. Model description and numerical algorithm

The investigated model of ESP consists of two parallel collecting

planes, which are electrically conducting and grounded, and
a spiked-type discharge electrode mounted in the middle of channel
as shown in Fig.1 [17]. The ESP channel is 600 mm long, 200 mm wide
and 100 mm high. The discharge electrode consists of a flat metal
tape having a series of spikes alternatively oriented upstream and

Fig. 6. Ion charge density distribution contours in the y ¼ 0 plane for non-uniform
Fig. 5. Ion charge density distribution on the collecting plates (number of contours 25 discharge from spike surfaces. (Number of contours is 20, the maximum ion charge
and maximum ion charge density 68.5 mC/m3). density on each spike tip is 565 mC/m3and the applied voltage is 30 kV).
422 N. Farnoosh et al. / Journal of Electrostatics 69 (2011) 419e428

Fig. 7. Airflow streamlines in plane A for negative applied voltage of (a) 30 kV and discharge current 350 mA (b) 23.5 kV and discharge current 170 mA (c) 19.2 kV and discharge
current 90 mA.

this method and the electrical boundary conditions can be found in 3. Numerical results and discussion
Ref. [18]. For ion charge density boundary condition on the spiked
electrode, zero ion charge density is assumed along the electrode, Fig. 3 shows the electric potential distribution in z ¼ 0 plane of
except on the spikes surfaces, where the corona discharges occur. the precipitation channel, assuming that the corona spiked elec-
Due to a non-uniform distribution of the ion charge density along trode is excited with a negative voltage of 30 kV and the glow
the corona electrode surface, the Kaptzov assumption suggested in corona discharge is generated uniformly along the entire length of
Ref. [18] is not practical in this simulation. Instead, the total corona the electrode. 50 contours are shown, which means that the
discharge current obtained from numerical simulations is potential difference between two adjacent lines is 600 V. The
compared with the experimental value obtained by Mizeraczyk contour density decreases from the area in vicinity of the corona
et al. [17]. In the iterative process the ion charge density on the spiked electrode to the grounded plates. Determining the 3-D
corona electrode surface is modified until the discharge current distribution of ion space charge density in the ESP channel for
obtained from simulation is in a close agreement with the experi- the spike-type discharge electrode is crucial, because ions are only
mental data. generated from areas close to the spikes tips and a noticeable
According to our investigations, the ion charge density distri- volume fraction of the ESP is charge free. The ion charge density at
bution pattern is only very slightly perturbed by the airflow; the spike tips has been evaluated to have the total average
therefore, the convection of ions is ignored and only the strong discharge current obtained from calculations equal to experimental
effect of electrostatic forces due to the presence of ionic space data presented in Ref. [17]. Fig. 4a shows the ion charge density
charge in electric fields on the airflow patterns is considered in this distribution contours in y ¼ 0 symmetry plane. When a negative
simulation. voltage of 30 kV is applied, the maximum ion charge density
 N. Farnoosh et al. / Journal of Electrostatics 69 (2011) 419e428 423

Fig. 8. Airflow streamlines in plane B for negative applied voltage of (a) 30 kV and discharge current 350 mA (b) 23.5 kV and discharge current 170 mA (c) 19.2 kV and discharge
current 90 mA.

of 485 mC/m3 is obtained close to the spike tips. The ions gener- significant changes were observed in the distribution of the ion
ated in the close proximity of each spike due to gas ionization space charge density farther from the discharge electrode. As
migrate to the collecting plates creating ionic space charge density, a result, all calculated velocity patterns are practically identical for
which decreases in density moving further away from the both cases.
discharge electrode toward the collecting plates and insulating The effect of the secondary EHD flow on the main airflow
walls as shown in the z ¼ 0 plane in Fig. 4b. The corresponding ionic pattern for different applied voltages is examined by comparing the
current densities on the ground plates are shown in Fig. 5 as well. airflow streamlines in three different planes (A, B and C) located
According to the time averaged currentevoltage characteristics along the channel length and perpendicular to the spiked electrode
curves the total discharge current at this voltage is 350 mA [17]. as shown in Fig. 2. Fig. 7aec shows the three-dimensional airflow
In this paper a uniform ion space charge density on the spike streamlines in plane A, which passes through the tip of the
surfaces was assumed, as depicted in Fig. 4a and b. All numerical upstream-directed central spike electrode for three negative
simulations were repeated considering non-uniform ion charge applied voltages of 30 kV, 23.5 kV and 19.2 kV, respectively.
density distribution close to the spikes, where the maximum These figures demonstrate a very complicated airflow structure
charge was generated in a small area close to the tip of each spike around the spikes tips.
(Fig. 6) to model a more realistic situation. The ion charge density As shown in Fig. 7a for the highest excitation voltage
distribution obtained in Fig. 6 is also in a good agreement with the (30 kV), due to the strong interaction between the secondary
numerical results presented by Brocilo [19]. Due to this change in EHD flow and the main airflow, a pair of vortices is generated in
the boundary conditions a small difference in the ion space charge plane A, located in the vicinity of the upstream-directed spike tip.
density close to the corona electrode can be noticed. However, no The two vortices block the main airflow in the upstream part of
424 N. Farnoosh et al. / Journal of Electrostatics 69 (2011) 419e428

Fig. 9. Airflow streamlines in plane C and negative applied voltage of (a) 30 kV and discharge current 350 mA (b) 23.5 kV and discharge current 170 mA (c) 19.2 kV and discharge
current 90 mA.

Fig. 10. Interaction of airflow streamlines in planes C and A for applied voltage Fig. 11. Equally spaced planes along the channel placed perpendicularly to the airflow
of 30 kV. direction.
 N. Farnoosh et al. / Journal of Electrostatics 69 (2011) 419e428 425

Fig. 12. Contours of x component of airflow velocity vector in planes DeH for the applied voltage of 19.2 kV.

the channel and drive the airflow toward the collecting plates. Fig. 9aec, shows no generated vortices. Since the secondary EHD
However, these two vortices get smaller and move closer toward flow from the spike tip in the plane C and the primary airflow are in
the spike electrode when the excitation voltage is decreased the same direction, their interaction is weaker and the vortices
to 19 kV (Fig. 7c.) The experimental results in Ref. [17] generated in this plane are not significant. Looking from a different
demonstrate another pair of vortices as well, which is formed angle to the airflow streamlines for 30 kV applied voltage (Fig. 10)
in the downstream of the channel very close to the collecting shows that some of the streamlines deflect toward the neighboring
planes preventing the main airflow from moving toward the spike tip located behind the plane C, which is in opposite direction
plates. However, this is not clearly visible in the simulation with respect to the main airflow, and then turn back toward the
results. The observed differences in the results could be due to downstream direction of the channel. Therefore, due to the strong
the resolution of the numerical method: the domain discretiza- interaction of the upstream-directed secondary EHD flow generated
tion close to the collecting plates is not sufficiently fine to capture from the upstream spike tip in the plane A with the oppositely
the vortices generated in these areas. On the other hand, the directed primary airflow, stronger vortices are generated in plane A
experimental airflow velocity patterns in planes A, B and C in Ref. rather than in plane C. Higher excitation voltages always induce
[17] are composed of three adjacent overlapping velocity fields stronger vortices.
and are obtained from averaging of 100 measurements using a 3- Figs. 12e15 show contours of the x component of airflow
D PIV method, in which fine TiO2 particles of less than 1 mm in velocity vector in planes DeH for different applied voltages. The
size were blown through the channel. Thus, in practice the planes are perpendicular to the ESP channel and spaced uniformly
velocity streamlines presented in their paper demonstrate the from 60 mm upstream the spike electrode (x ¼ 60 mm) to
time average trajectories of submicron particles, which mostly 60 mm downstream the spike electrode (x ¼ 60 mm) as shown in
follow the airflow streamlines. The numerical results in Fig. 7aec Fig. 11. Perturbed velocity patterns are generated in the channel
also show a slight deflection of airflow streamlines toward the due to the complex interaction between main airflow and EHD
center of the channel in the downstream direction after the flow. The velocity pattern in plane D is not much disturbed for the
airflow passes the spiked electrode. lowest applied voltage (19.2 kV), whereas quite a regular airflow
Fig. 8aec shows the airflow streamlines in plane B, which passes pattern originating from the spike tips is visible in planes EeH. By
between upstream and downstream spikes. For the same voltages, increasing the excitation voltage some airflow irregularities arise
the centers of the two vortices slightly move in the upstream in central part of the channel in plane D as shown in Fig. 14. For
direction and are smaller in size than the vortices generated in plane zero applied voltage to the corona electrode (no corona discharge),
A. The pattern of airflow streamlines in plane C, which passes Fig. 15 demonstrates the contours of x component of airflow
through the tip of the neighboring downstream-directed spike, velocity vector in planes DeH.

Fig. 13. Contours of x component of airflow velocity vector in planes DeH for applied voltage of 23.5 kV.

Fig. 14. Contours of x component of airflow velocity vector in planes DeH for applied voltage of 30 kV.
426 N. Farnoosh et al. / Journal of Electrostatics 69 (2011) 419e428

Fig. 15. Contours of x component of airflow velocity vector in planes DeH for zero applied voltage.

These results are in a good agreement with the experimental The 3-D EHD secondary flow distributions in the channel in
results obtained by Mizeraczyk and his co-workers [17]. The side the case of 30 kV applied voltage without the main airflow are
walls cause a dramatic airflow fluctuation around the spike tips shown in Fig. 18a and b from both side and top view. The EHD
nearest the walls. However, the airflow patterns around the central flow streamlines in Fig. 18a show two pairs of spiral vortices in
spikes tips are stable and very similar to each other as is also the vicinity of the spiked electrode: one downstream and the
reported in Ref. [17]. The results confirm the complex 3-D airflow other in the upstream direction of the channel. Fig. 18b shows
structures in spiked electrode configuration as well. the EHD flow streamlines from the top view. The asymmetry in
Figs. 16 and 17 show turbulence intensity contours in the y ¼ 0 this figure is due to the non-symmetric position of the spikes on
plane for the lowest (19.2 kV) and highest (30 kV) applied both sides of the strip electrode and the fact that the same value
voltages, respectively. The electric corona discharge is responsible of ion space charge density is assumed on the spike surfaces
for a strong increase of the turbulence level. Assuming 3% turbu- along the discharge electrode. The EHD flow has a swirling
lence intensity at the inlet and increasing the applied voltage motion in the channel and consists of one circular very small and
from 19.2 kV to 30 kV, the maximum turbulence intensity in the a bigger spiral vortex in the x direction and one long-elliptic
vicinity of the spiked electrode increases from 36% to 65%. The spiral vortex in the x direction. After repeating the simulation
larger density of contours in the downstream direction of the for a non-uniform ion charge density distribution on the spikes
channel in the case of 30 kV applied voltage also confirms larger surfaces along the electrode, i.e. the ion charge density on the
turbulence intensity generated in this area rather than for the lower spikes surfaces close to the walls is 75% of the ion charge density
excitation voltage. on spikes surfaces located in the middle, the EHD secondary flow

Fig. 16. Turbulence intensity contours in y ¼ 0 plane for applied voltage of 19.2 kV (number of contours is 20, the maximum and minimum turbulence intensities are 36% in
vicinity of spiked electrode and 3% at the inlet).

Fig. 17. Turbulence intensity contours in y ¼ 0 plane for applied voltage of 30 kV (number of contours is 20, the maximum and minimum turbulence intensities are 65% in vicinity
of spiked electrode and 3% at the inlet).
 N. Farnoosh et al. / Journal of Electrostatics 69 (2011) 419e428 427

Fig. 18. EHD secondary flow streamlines when the airflow velocity at the inlet is zero and applied voltage is 30 kV, assuming uniform ion charge density on spikes.

Fig. 19. EHD secondary flow streamlines when the airflow velocity at the inlet is zero and applied voltage is 30 kV, assuming non-uniform ion charge density on spikes.

streamlines in Fig. 19a and b demonstrate more symmetric show that the strength and dimension of the gas vortices depend
patterns as it was expected. not only on the value of the excitation voltage but also on the
mutual directions of the secondary EHD and the primary airflows.
4. Conclusions Most of the numerical results obtained in this simulation qualita-
tively agree with the experimental data presented in Ref. [17].
The developed 3-D numerical technique was applied to simulate Comparing the velocity streamlines obtained for the spiked-type
the electrical and electrohydrodynamic (EHD) characteristics of electrode geometry in this paper with our previous results,
a laboratory scale spiked-plate electrostatic precipitator. Due to assuming a smooth corona electrode in a single wire-ESP model
a high mobility of the ions produced by the corona discharge, the [18], emphasizes the significant role played by the discharge
effect of gas velocity fluctuations on the ion distribution was electrode structure, or discharge pattern, for determining the
neglected. The velocity patterns generated in the channel due to magnitude and structure of electrically induced flow, and turbu-
the strong interactions between the main and EHD flows were lence levels in the channel. It is suggested that this improved
demonstrated for different applied voltages. modeling and understanding of the secondary EHD flow charac-
The 3-D secondary EHD flow streamlines in different planes of teristics in the spiked-type precipitator design is crucial for an
the spike-plate ESP model obtained for various applied voltages accurate collection analysis of submicron particles in practical
428 N. Farnoosh et al. / Journal of Electrostatics 69 (2011) 419e428

applications. This numerical technique can also be applied for [7] T. Yamamoto, H.R. Velkoff, Electrohydrodynamics in an electrostatic precipi-
tator, J. Fluid Mech. 108 (1981) 1e18.
other types of electrode geometry and ESP arrangements.
[8] G.A. Kallio, D.E. Stock, Flow visualization inside a wire-plate electrostatic
precipitator, IEEE Trans. Ind. Appl. 26 (3) (1990) 503e514.
Acknowledgments [9] J.H. Davidson, P.J. McKinney, EHD flow visualization in the wire-plate and
barbed plate electrostatic precipitator, IEEE Trans. Ind. Appl. 27 (1) (1991)
154e160.
The authors acknowledge with thanks the financial support
[10] G.A. Kallio, D.E. Stock, Interaction of electrostatic and fluid dynamic fields
provided to N. Farnoosh by the Natural Sciences and Engineering in wire-plate electrostatic precipitator, J. Fluid Mech. 240 (7) (1992)
Research Council of Canada. This work was made possible by the 133e166.
[11] T. Ullum, P.S. Larsen, Swirling flow structures in electrostatic precipitator,
facilities of the Shared Hierarchical Academic Research Computing
J. Flow Turbul. Combust. 73 (3e4) (2004) 259e275.
Network (SHARCNET: www.sharcnet.ca) and Compute/Calcul [12] P.S. Larsen, E.M. Christensen, Secondary flow and turbulence in electrostatic
Canada. The authors also appreciate Prof. J.S. Chang and Prof. P. precipitators, Third Int. Symp. on Appl. of Laser Anemometry to Fluid Mech.,
Atten for their helpful discussions and comments. Lisbon, 1988, pp. 361e374.
[13] J.H. Davidson, P.J. McKinney, Turbulent mixing in a barbed plate-to-plate
electrostatic precipitator, Atmos. Environ. 23 (10) (1989) 2093e2107.
References [14] J. Podlin ski, R. Barbucha and J. Mizeraczyk, EHD flow patterns in narrow and
wide ESPs measured by 3D PIV method, in: Proc. 5eme Conf. Societe Francaise
[1] T. Yamamoto, M. Okuda, M. Okubo, Three dimensional ionic wind and elec- Electrostatique, Grenoble, 2006, pp. 53e56.
trohydrodynamics of tuft/point corona, IEEE Trans. Ind. Appl. 39 (6) (2003) [15] A. Niewulis, J. Podlinski, J. Mizeraczyk, Electrohydrodynamic flow patterns in
1602e1607. a narrow electrostatic precipitator with longitudinal or transverse wire
[2] H. Fujishima, Y. Ueda, K. Tomimatsu, T. Yamamoto, Electrohydrodynamics of electrode, J. Electrostat 67 (2009) 123e127.
spiked electrode electrostatic precipitators, J. Electrostat 62 (2004) 291e308. [16] J. Podlinski, M. Kocik, R. Barbucha, A. Niewulis, J. Mizeraczyk, 3D PIV meas-
[3] J.H. Davidson, P.J. McKinney, P. Linnebur, Three-dimensional (3-D) model of urementsof the EHD flow patterns in a narrow electrostatic precipitator with
electric field and space charge in the barbed plate-to-plate precipitator, IEEE wire-plate or wire-flocking electrodes, Czechoslovak J. Phys. 56 (2006)
Trans. Ind. Appl. 32 (4) (1996) 858e866. 1009e1016.
[4] L.M. Dumitran, P. Atten, D. Blanchard, Numerical simulation of fine particles [17] J. Podlinski, A. Niewulis, J. Mizeraczyk, Electrohydrodynamic flow and particle
charging and collection in an electrostatic precipitator with regular barbed collection efficiency of a spike-plate type electrostatic precipitator, J. Electro-
electrodes, Inst. Phys. Conf. Ser. 178 (2003) 199e205. stat 67 (2009) 99e104.
[5] T. Yamamoto, Y. Morita, H. Fujishima, M. Okubo, Three-dimensional EHD [18] N. Farnoosh, K. Adamiak, G.S.P. Castle, 3-D numerical analysis of EHD turbu-
simulation for point corona electrostatic precipitator based on laminar and lent flow and mono-disperse charged particle transport and collection in
turbulent models, J. Electrostat 64 (2006) 628e633. a wire-plate ESP, J. Electrostat 68 (6) (2010) 513e522.
[6] H. Fujishima, Y. Morita, M. Okubo, T. Yamamoto, Numerical simulation of [19] D. Brocilo, Electrode geometry effects on the collection efficiency of submi-
three-dimensional electro-hydrodynamics of spiked-electrode electrostatic cron and ultra fine dust particles in spike-plate electrostatic precipitators,
precipitator, IEEE Trans. Diel. Electr. Insul. 13 (1) (2006) 160e167. Ph.D. Thesis, McMaster University, 2003.