Systematic Approach to Optimization of Submicron Particle Agglomeration Using Ionic-Wind-Assisted Pre-Charger

Electric agglomeration is the process in which particles are charged in electric fields and coagulate, which is enhanced by electric force or turbulence. The collection efficiency of submicron particles is improved, which can be a solution to submicron particle abatement in traditional electrostatic precipitators (ESPs), by using a pre-charger to increase median particle diameter and realizing particle pre-charging. In this study, a laboratory bipolar pre-charger with a perforated plate between discharging regions was designed to examine ionic-wind-assisted charge-induced agglomeration, and an ESP was arranged afterwards to collect the fine particles. Experiments were conducted to investigate submicron particle charging, agglomeration characteristics, and collection efficiency. Results indicated that a pre-charger with proper discharging voltage match and plate porosity can optimize particle agglomeration and improve collection efficiency by about 12% compared with the results obtained without a pre-charger. An optimal solution was achieved and a collecting efficiency of 96%–98% was obtained for all sizes by utilizing the turbulence caused by ionic wind and optimizing the experimental operation conditions.


INTRODUCTION
Particulate matter pollution is a serious global problem because fine particles can remain suspended in air and spread with atmospheric motion, causing permanent health and environmental damage (Kagawa and Ishizaka, 2014).Particles less than 1 µm in diameter can penetrate the alveoli where gas exchange occurs (Londahl et al., 2006;Cao et al., 2013).These particles behave similarly to gas molecules and affect gas exchange within the lungs, penetrating the lung into the blood stream and translocating further into the cell tissue and/or the circulatory system (Valavanidis et al., 2008), which lead to significant health problems such as asthma attacks, nonfatal heart attacks, and even cancer (Gorbunov et al., 2013;Crilley et al., 2014;Kim et al., 2015).Therefore, the problem of particulate matter pollution requires urgent resolution.
Electrostatic precipitator (ESP), which is common for particle abatement in power plants, is characterized by a high overall mass collection efficiency of above 99.9% with large fluxes and low pressure drops (Mizuno, 2000;Jaworek et al., 2007).However, conventional ESPs are not sufficiently effective in collecting submicron particles (Le et al., 2013;Chen et al., 2014;Wang et al., 2015).Thus, upgrading existing technologies to enhance performance in terms of removing submicron particles is of great importance.
Several ideas have been proposed to solve this problem by enhancing particle agglomeration to increase mean particle size, including acoustic coagulation to increase particle vibration and agglomeration (Noorpoor et al., 2013), electric agglomeration to increase particle electrostatic attraction and agglomeration (Kanazawa et al., 1993;Hautanen et al., 1995;Kildeso et al., 1995;Laitinen et al., 1996;Ji et al., 2004;Lin et al., 2011;Thonglek and Kiatsiriroat, 2014), and water vapor condensation to increase viscosity and agglomeration (Stairman, 1968), etc. Electric agglomeration is the process in which particles are charged in direct or alternate current electric fields, and collide and coagulate as driven by electric force.The agglomeration procedure is often strengthened by an external electric field or turbulators to increase particle collision in addition to that caused by Brownian motion.Given that particles with different polarities can attach together to form larger ones, the collecting efficiency of submicron particles improves through the pre-charger enhancement.
Considering its simple equipment construction, electric agglomeration has been studied and used to clean exhaust gas.Kanazawa et al. (1993) studied a two-stage ESP with a bipolar charging section, wherein the collection efficiency for submicron particles is about 80% under optimum condition.Maisels et al. (2004) used a developed Monte Carlo method to simulate the process of simultaneous charging and coagulation; he demonstrated that particle size distribution is clearly changed by the coagulation of high particle number concentrations.Xu et al. (2009) used a positive pulsed ESP to enhance particle charging and agglomeration to improve collection efficiency and achieved a PM 2.5 number reduction efficiency that exceeded 90%.Zhu et al. (2010) designed a closed-loop lab ESP with a bipolar pre-charger that yielded a grade collection efficiency of 95%-98%.However, further detailed investigations on submicron particle agglomeration technologies are still necessary, and the design of agglomeration equipment must be optimized.Ionic wind produced by electrohydrodynamic (EHD) flow is a noticeable phenomena from another aspect.In ESPs, the ions created in an inhomogeneous electric field near the discharge electrodes migrate toward the collecting electrodes and provide momentum to neutral molecules, which serve as the driving force to the fluid and cause the secondary flow, called "ionic wind" (Robinson, 1962).The ionic wind in ESPs can produce strong turbulence, which significantly affects the flow field and complicates particle movement (Mizeraczyk et al., 2003).
A laboratory bipolar pre-charger with a perforated plate between discharging regions was used in this study to examine ionic-wind-assisted charge-induced agglomeration.Afterwards, an ESP was arranged to collect fine particles.The effects of discharging voltage match and plate porosity on particle agglomeration and collection were evaluated by investigating the particle charging characteristics, median particle diameter, and collection efficiency.

Experimental Setup
A study on fine particle agglomeration and collection was conducted in a laboratory ESP combined with a bipolar pre-charger and a mixing region characterized by an acrylic duct 20 cm in width, 10 cm in height, and 100 cm in length.The experimental system (Fig. 1) can be divided into a precharging region, an agglomeration region, and an ESP.Through the system, particles are positively or negatively charged in the pre-charger, mixed and coagulated in the static mixer, and finally collected by a traditional ESP.
Aerosol particles were charged in two separate parallel corona charger chambers of the pre-charger.Each chamber consisted of one discharging electrode placed in the middle of each channel between grounding electrodes.The length, width, and height of each chamber were 20, 20, and 10 cm, respectively.The grounded stainless plate length was 6 cm, and the wire-plate distance was 2.5 cm.Flow velocity was 0.6 m s -1 , and particle residence time in the charger was 0.1 s, which ensured sufficient charging time and minimized particle collection in the pre-charger (Long and Yao, 2010).The discharging electrodes were ribbon electrodes made of stainless steel (Fig. 2).The power supplies used in the precharger and the ESP were high-frequency power supplies (Telsman, China).
In the agglomeration region, bi-polarly charged particles were fully mixed through the turbulence caused by the static mixer.Instead of traditional staggeredly arranged columnshaped pin fins, a set of specially designed inclined placed cubics served as turbolators to optimize flow mixture and to promote charged particle collision and agglomeration.The structure of the static mixer is shown in Fig. 2. The agglomerator was 20 cm in width, 10 cm in height, and 20 cm in length.The turbolators were 5 cm × 5 cm arylic cubes with 3 cm thick, and guide plates were added for flow uniformity.
The ESP, in which a set of four equidistant discharging electrodes was used to collect enlarged particles, was 20 cm in width, 10 cm in height, and 60 cm in length.The distance of adjacent discharging electrodes was 10 cm, and the wire-plate distance was 5 cm.
Furthermore, to utilize the ionic wind generated in the pre-charger, perforated plates were used between positive and negative discharging channels so that the ionic-windinduced turbulence could improve the mixture and collision of bi-polarly charged particles in the pre-charger.The plates were perforated by round holes in a consistent position with the diameters of 6, 8, and 10 mm, such that the porosities of perforated plates were 0%, 9.4%, 16.7%, and 26.2%.

Experiment Procedure
Particles used in the study were practically submicron and were generated through incense burning.The stick incense used in the study can burn steadily for more than 30 minutes with sufficient concentration and size distribution stability.The number of particle distribution is shown in Fig. 3. To investigate the effect of the pre-charger, power supplies were first turned on once the particle concentration was stabilized.The pre-charger performance could be evaluated by comparing the particle charge and the median diameter.Second, to investigate the combined effect of the pre-charger and the ESP, the power supply of the ESP was turned on to evaluate the collection efficiency of ESP with and without the pre-charger in order.All experiments were performed under ambient conditions.The flow velocity in the chamber was 0.6 m s -1 .
The I-V characteristics of the pre-charger and the ESP are shown in Fig. 4. In the pre-charger, the positive corona starts at the voltage 8 kV and breaks down at nearly 17 kV, whereas the negative corona starts at 6 kV and breaks down at about 20 kV.In the ESP, the corona starts at 9 kV and breaks down at about 25.5 kV.The average electric field strength in the pre-charger and the ESP rose to 6 kV cm -1 .Corona discharge current increased with increasing discharging voltage.Since the pre-charger in this study has the functions of both particle pre-charging and particle agglomeration.We chose the negative voltage to be as high as possible (usually 80% of range from onset voltage to the breakdown voltage) to ensure sufficient pre-charging and operational stability and the voltage of the positive channel was set from 9 kV to 15 kV (the operation region) to evaluate the electric agglomeration effect in different voltage matches.The discharging voltage of 25 kV was applied to the four ribbon electrodes of the ESP.

Evaluation Methods
To determine the agglomeration effect and final collection efficiency, ELPI+ (Electrical Low Pressure Impactor, Dekati Ltd., Finland), a real-time particle size spectrometer at scales of 6 nm to 10 µm with 14 stages, was used to measure grade particle concentration.In an ELPI+, particles were initially charged electrically by small ions produced in a unipolar positive polarity charger.Afterwards, particle size was classified in a low-pressure impactor based on their aerodynamic diameter.The impactor stages were insulated electrically and each stage was connected individually to an electrometer current amplifier.The diagnostic current value in each stage was proportional to the number of particles and was converted to a size distribution by using particle size dependent relations that describe the charger properties and the impactor stages.Therefore, when the initial concentration and outlet concentration were measured, collection efficiency could be calculated by Eq. (1).
where η is the grade collection efficiency, N inlet is the initial concentration of a certain diameter in (1 cm -3 ), and N outlet is the outlet concentration at the outlet of the ESP (1 cm -3 ).
To obtain in-depth insights on particle agglomeration, the characteristics of fine particle charging by the precharger were evaluated by supposing that the same amount of chargers were collected for same size particles.The analytical relation between the number of collected charges and the ELPI diagnostic current was derived early on as where n is the average charge units on particles, I is the diagnostic current of ELPI+ with the charger off (A), Q is the sample volume (cm 3 s -1 ), N is the number concentration of particles (1 cm -3 ), and e is the elementary electron charge (1.602 × 10 -19 C).However, particle concentration or collection efficiency failed to clearly reflect the agglomeration effects.Considering that the agglomeration procedure was frequently accompanied by increased median particle diameter (i.e., production of enlarged particles), a parameter called Agglomeration Enhancement Ratio was defined to demonstrate the enlargement ratio of median particle diameter: where R ae is the agglomeration enhancement ratio, D 0 is the median particle diameter before agglomeration procedure (µm), and D agg is the median particle diameter after agglomeration procedure (µm).
Median particle diameter D can be calculated by Eq. ( 4): ( ) where N i is the particle number of each grade measured by ELPI+, and D i is the mean particle diameter of each stage.
The agglomeration effects of the pre-charger could be further analyzed through considering the Agglomeration Enhancement Ratio.For example, R ae > 1 illustrates an increase of the main diameter and an achievement of effective agglomeration.

Particle Charging Characteristics
Ionic wind produced near discharging electrodes can cause strong turbulence toward grounding plates, which significantly affects flow field and particle migration.To utilize ionic wind for agglomeration enhancement, perforated plates were used between positive and negative discharging channels.In this manner, the ionic-wind-induced turbulence can drive the particles toward the grounding plates, move through the holes on the plate, and penetrate into the other channel with the opposite polarity.Thus, the chances of positively and negatively charged particles mixing and colliding are increased.
The discharging characteristics are shown in Fig. 5.Note that the porosity of perforated plates significantly affect the corona current.With rising porosity, the amount of positive ions that penetrate the negative channel increases, moving to the opposite direction and resulting in heightened current density.The current of the negative channel also increases with positive channel voltage, which indicates the increase in positive ions were produced and penetrated the negative channel.
To understand the particle charging procedure in the pre-charger, the characteristics of mean particle charge were investigated by the number of collected charges and ELPI+ diagnostic current, the analytical relationship of which was derived early through Eq. ( 2).Typical examples of the charging properties are shown in Fig. 6.Given the perforated plate porosity, the collected charges increases with the use of perforated plates.The particle charging procedure in the pre-charger is a complicated process which is affected by the electric field distribution, flow field, particle agglomeration, ion concentration, etc. Perforated plates enables the ions produced by positive and negative power supplies move into the other channel through the plate which would be collected by the imperforated grounding plate otherwise, providing a higher chance for uncharged particles getting charged.On one side, since the amount of negative ions and the current of negative channel is higher than those of positive channel, the mean particle charge increased comparing to smooth plate.On the other side, in agglomeration process, particles get the charge from smaller particles (the amount of negative ions is larger than positive ones) without change of diameter magnitude, i.e., in the same detection channel, therefore the calculated mean particle charge after agglomeration should be larger than theoretically the result without agglomeration.For example, the porosity 9.4% in this study was proved to be most effective for particle agglomeration, the mean particle charge unit increased most after agglomeration, which can be the reason of why the mean particle charge of porosity 9.4% is usually larger than that of 16.7%.In considering voltage matches, collected charges always decrease with increasing applied positive voltage, which is caused by the neutralization of bipolar particle agglomeration.

Agglomeration Characteristics Effect of Applied Voltage
To evaluate and obtain suitable voltage matches for the pre-charger, as well as ensure that particles were fully precharged, a negative voltage of -15 kV was applied to the negative channel.Positive voltage ranged from 9 kV to 15 kV to investigate the effect on agglomeration.The particle    7. The effect of agglomeration can be observed in the decrease of percentages in the first and second stages, as well as in the increase of larger particles in the percentage of other stages, which implies that smaller particles diminish and become larger particles.The agglomeration can be further proven by the R ae ratios, which were all above 1, indicating the improvement of average particle diameter.
The agglomeration enhancement ratio R ae was compared, as shown in Fig. 8, to compare and to obtain the appropriate voltage match.The increase of positive discharging voltage increased R ae when the positive voltage changed from 9 kV to 12 kV.This increase suggests that particles with high mean positive charges are produced with high discharging voltages, which mix better and agglomerate further with negatively charged particles under the effect of Coulomb force.However, when the positive discharging voltage was above 12 kV, R ae decreased from 1.33 to 1.10.This finding can be presumably explained through submicron particles in the positive channel, where multiple neutral particles can agglomerate on charged particles because of polarization.However, when overcharged, an excess of positive ions are produced and those parts of agglomeration are inhibited.

Effect of Perforated Plate
Four types of plates were used to evaluate the effect of perforated plate with various porosities.The voltage of the negative channel was -15 kV, whereas that of the positive channel was +12 kV.Particle percentage distributions under different perforated plate porosities are shown in Fig. 9.With the pre-charger on, the effect of agglomeration is evident across all porosities, which demonstrates improved median particle diameter.
To compare the effect of different porosities, the agglomeration enhancement ratio R ae under different porosities is shown in Fig. 10.For all cases, the highest R ae is under the voltage match (+12 kV, -15 kV), which agrees with previous voltage match results.Considering the porosities under varying voltage matches, generally perforated plates with porosities under 16.7% achieved better agglomeration results compared to regular smooth plates, while the maximum porosity under nearly every voltage match obtained the lowest agglomeration enhancement ratio.This finding indicates that perforated plates with proper porosities can promote flow exchange and particle collision in  discharging regions.Furthermore, high porosity is less effective probably because large holes in the plates cause inhomogeneity of the electric field, as well as disturbance and excessive neutralization with ions.To optimize particle agglomeration under this experimental situation, the voltage match (+12 kV, -15 kV) and perforated plate of porosity 9.4% can be chosen as an optimal solution.

Effects of Pre-Charger on ESP Performance Effects of Pre-Charger on ESP Corona Current
The discharging characteristics between positive channel voltage and ESP corona current are shown in Fig. 11.The diagnostic current of ESP falls from 380 µA to 367 µA with the increase in positive discharging voltage of the precharger.Given that ions and particles with positive charge can neutralize negative ions and negatively charged particles produced in the ESP, we can infer that higher positive channel voltages neutralize more negative ions and particles, which reduce the collected ions and particles of collecting electrodes and lead to decreased diagnostic ESP current.This phenomenon can similarly affect particle charge and capture processes.

Effects of Pre-Charger on ESP Collection Efficiency
The grade and total collection efficiencies of the ESP under different applied voltages without the pre-charger are shown in Fig. 12.The discharging voltage for ESP ranged from -9 kV to -25 kV.Results show that grade efficiency increases as discharging voltage rises.Total collection efficiency was 84.3% at a discharging voltage of -25 kV.
In contrast to the collection efficiency without the precharger application, Fig. 13 shows the typical effects of prechargers on grade collection efficiency.When considering the voltage matches applied on the pre-charger, the voltage of negative channel was controlled at -15 kV to ensure that fully negatively charged particles and positive channel voltages ranged from +9 kV to +15 kV.The discharging voltage for ESP was -25 kV to collect agglomerated particles.Generally, the bipolar corona-induced agglomeration can improve collection efficiency by 8%−12%.With increased positive discharging voltage, the grade collection efficiency increases to the highest at the match (+12 kV, -15 kV), and then drops down at the match (+15 kV, -15 kV), which is identical to the agglomeration results.The pre-charger also promotes particle charging in advance.Thus, negatively pre-charged particles are collected in the ESP faster, which indicates better collection efficiency compared to normal ESP.Furthermore, increased positive discharging voltage is adverse to particle pre-charging (as discussed in section 3.1), such that the discharging voltage match of (+15 kV, -15 kV) performed in a manner that was not as effective as the optimal solution.
Similar results were obtained in the case of porosity.When comparing the collection efficiency using perforated plates with different porosities at the match (+12 kV, -15 kV) proven to be the best match for particle agglomeration, the highest collection efficiency was achieved at a porosity of 9.4%, and a higher porosity was proven with less effective probability because of the inhomogeneity of the electric field, disturbance, and excessive neutralization with ions caused by large holes on the plate.Utilizing the turbulence caused by ionic wind and optimizing the experimental operation conditions resulted in an optimal solution a collecting efficiency that could be as high as 96%−98% for all size particles.The total mass collection efficiency was also at 96.8%.The improvement was nearly 12% compared with the condition without the pre-charger.
In addition, the wall loss in the experimental system was about 4% to 10% at different stages and 6% of mass loss in total with no DC voltage applied.To consider the mass balance in this study, 6% of particles were attached to the surface of reactor through diffusion, and the fine particle collection efficiency was improved to about 84%-96% with the enhancement of ESP and pre-charger, and the remaining 6% to 14% was released through off-gas.

CONCLUSIONS
According to the presented experimental investigations on fine particle charge, agglomeration, and collection, the   following conclusions are obtained.1) By utilizing ionic-wind-induced turbulence to improve collision of positively and negatively charged particles, the current of negative channel increases with the increase of positive channel positive, as well as the decrease of mean particle charge with the increasing of positive discharging voltage.2) With the use of perforated plates, the corona current in the negative channel increases with the heightened porosity of the perforated plates, leading to increased mean particle charges.
3) The positive and negative discharging voltage matches in the pre-charger can affect agglomeration and collection procedure.The positive electric field strength at 4.8 kV cm -1 (12 kV) can effectively improve particle agglomeration because it can increase the median particle diameter and ensure pre-charged particle collection.4) The porosity of perforated plates in the pre-charger with proper porosity can affect the discharging current and particle agglomeration procedures.The porosity of 9.4% is chosen as an optimal solution for the increased mean particle diameter by 39%.5) Given that the current of ESP is inhibited by positive discharging voltage, it can affect the collection efficiency in addition to the particle charge and agglomeration results.An optimal solution is achieved, and the collecting efficiency increases as high as 96%−98% for all size particles.Total mass collection efficiency is also improved by nearly 12%.

Fig. 1 .
Schematic diagram of the bipolar agglomeration system experimental setup.

Fig. 3 .
Fig. 3. Particle number distribution under a mass concentration of 1.5 mg Nm -3 and a flow velocity of 0.6 m s -1 .

Fig. 5 .
Fig. 5. Effect of positive channel voltage on the current of negative channel.

Fig. 11 .
Fig. 11.Effect of positive channel voltage on ESP corona current.

Fig. 12 .
Fig. 12. Collection efficiency of the ESP without pre-charger: (a) grade collection efficiency and (b) total mass collection efficiency under different discharging voltages.

Fig. 13 .
Fig. 13.Collection efficiency with pre-charger and ESP under (a) different voltage matches and (b) different plate porosities.