Electrode geometry optimization in wire-plate electrostatic precipitator and its impact on collection efficiency

Electrode geometry optimization in wire-plate electrostatic precipitator and its impact on collection efficiency

Journal of Electrostatics 80 (2016) 76e84 Contents lists available at ScienceDirect Journal of Electrostatics journal homepage: www.elsevier.com/loc...

4MB Sizes 0 Downloads 53 Views

Journal of Electrostatics 80 (2016) 76e84

Contents lists available at ScienceDirect

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

Electrode geometry optimization in wire-plate electrostatic precipitator and its impact on collection efficiency Zhiyuan Ning a, Janusz Podlinski b, Xinjun Shen a, Shuran Li a, Shilong Wang c, Ping Han c, Keping Yan a, * a

Industrial Ecology and Environment Research Institute, Zhejiang University, Tianmushan Road 148, Hangzhou 310007, China Centre for Plasma and Laser Engineering, The Szewalski Institute of Fluid Flow Machinery, Polish Academy of Sciences, ul. Fiszera 14, 80-231 Gdansk, Poland c Shenhua Guoneng Energy Group Corporation Limited, Beijing 100033, China b

a r t i c l e i n f o

a b s t r a c t

Article history: Received 3 December 2015 Received in revised form 26 January 2016 Accepted 6 February 2016 Available online xxx

Wide Electrostatic Precipitator (having plate-to-plate spacing 400 mm or more) is one of a promising ways to improve existing ESPs, yet its large-scale application has been limited because of potential collection efficiency reduction. This article focus on study the electrohydrodynamic flow inside the ESP and the particle collection efficiency by using Particle Image Velocimetry and Electrical Low Pressure Impactor respectively when several different electrode geometries were applied, in order to increase wide ESP efficiency. Results showed that geometry with juxtaposed high voltage wires provided best performance of the wide ESP (both in collection efficiency and running cost) among our tests. © 2016 Elsevier B.V. All rights reserved.

Keywords: Electrostatic precipitator Electrodes geometry Ionic wind Electrohydrodynamic flow Collection efficiency Electrical low pressure impactor Particle image velocimetry

1. Introduction Recently there are increasing concerns on particulate matter (PM) pollution in many developing countries. In fact, the increased levels of PM in the air can lead to anthropogenic particulate air pollution, which is consistently and independently related to the most serious effects, including lung cancer and other cardiopulmonary diseases. Some PM occurs naturally, but in large cities of China it is mainly the result of Human activities related to the vehicles, power plants and various industrial processes. Due to the highly toxic health effects of PM, most governments have created regulations for the particulates emission and concentration in the ambient air. Thus, PMs removal is important for the sake of environmental concerns.

Abbreviations: ESP, electrostatic precipitator; ELPI, electrical low pressure impactor; PIV, particle image velocimetry; TKE, turbulence kinetic energy; PM, particulate matter; HV, high voltage; EHD, electrohydrodynamic. * Corresponding author. E-mail address: [email protected] (K. Yan). http://dx.doi.org/10.1016/j.elstat.2016.02.001 0304-3886/© 2016 Elsevier B.V. All rights reserved.

Electrostatic precipitator (ESP) is a filtration device that removes PM from the treated gas. ESPs are widely used due to their advantages such as high total mass collection efficiency, low operational costs, high capacity and durability [1]. As European Union HVAC association suggests, a normal emission of ESP should be within range of 10e20 mg/Nm3. With proper maintenance and geometry 5 mg/Nm3 should be also achievable. Emission as low as 1 mg/Nm3 in Wet ESP should be guaranteed. Thus, most probably ESPs will stay in main stream of particle removal equipment, even with a stricter emission standards [2]. However, ESPs should be improved in many ways, especially the collection efficiency of submicron dust particles should be increased. Our previous work [3] showed that the gas flow inside a wireplate ESP is quite complex. Under actual operation conditions, ionic wind is generated by corona discharge. The direction of ionic wind in the corona discharge region (in the space between a high voltage (HV) and grounded electrodes) is from discharge electrode to collecting electrode [4], and it is usually directed perpendicularly to the original primary gas flow direction. Ionic wind can disrupt primary flow pattern, causing strong turbulences [5e7]. Normally, the level of turbulence increases along with the voltage increase.

Z. Ning et al. / Journal of Electrostatics 80 (2016) 76e84

Results indicate that resultant electrohydrodynamic (EHD) secondary flow pattern can alter significantly the primary flow [8] and impacts the particle collection efficiency (especially when particle diameter is at the range of 0.1e1 mm). There can be two reasons. Firstly, the gas flow velocity inside a turbulent area is typically higher than in a non-turbulent area, making particles inside this area harder to be collected. Secondly, normal ESP usually has symmetric wire geometry, making the generated flow pattern also symmetric [9,10]. The phenomena of the whole flow pattern can be analog to a flow pattern inside a Venturi tube. It creates a high speed area in the center, through which massive gas pass along with uncollected particles. Certainly, this could reduce efficiency significantly, mainly by reducing the general residence time and inhomogeneous distribution of Coulomb force. The significant decrease of efficiency can be anticipated if the big amount of particles passes through a central high speed area. It can be the case especially for wide wire-plate ESPs, which indeed reduce the investment costs effectively, but typically has lower collection efficiency. An understanding about how electricity and electrodes geometry influences the collection efficiency is critical during design and operation procedure of wide wire-plate ESPs. In order to optimize related parameters, modern techniques such as Particle Image Velocimetry (PIV) [11] and Electrical Low Pressure Impactor (ELPI) [12] are used to further understand the particle collecting process. Above methods have several key advantages, e.g.: non-contact measurement which can provide instantaneous or time-averaged results, measurement of the flow velocity field inside ESP (in HV discharge region) is possible. This study was conducted to improve the ESP with a wide wireplate geometry. One of approach of widening existing ESP geometry by removing every second collecting electrode was proposed and tested during this work. The loss of the ESP collection efficiency seems to be inevitable after such modification. Thus, in order to complement the efficiency loss caused by a fewer collecting plates, the geometries with additional grounded or HV wire electrodes were investigated. Several different electrode geometries were tested and their dust particle collection efficiency and EHD flow velocity fields were measured using ELPI and PIV respectively. The best electrodes geometries, both in collection and energy efficiency, were searched. 2. Experiment 2.1. Experimental setup The schematic diagram of the experimental apparatus is shown in Fig. 1. The apparatus consisted of a six major parts, i.e.: an ESP; a fan forcing the primary flow through the ESP; a dust generator; a negative polarity DC high voltage power supplier; an ELPI for collection efficiency measurement and a standard 2D-PIV equipment for EHD flow velocity field measurement. The ESP was made as a Perspex parallelepiped. The electrical electrode set consisted of a stainless-steel plate collecting electrodes and a stainless-steel wire electrodes (diameter of 0.09 mm). Several different electrodes geometries were investigated during this experiment and they will be described in the next subsection. An ambient air with a dust particles was introduced to the ESP inlet through a diffuser equipped with a mesh board, in order to distribute the primary flow uniformly along the ESP cross-section. The high voltage and the current on the DC power supplier output were measured with a resolution ±0.1 kV and ±0.1 mA respectively. A 1 MU current limiting resistor was inserted between the DC power supplier and the HV wire electrodes. The plate electrodes were grounded.

77

The dust particles suitable for our experiment should meet those key features: 1 - it won't hinder major gas flow; 2 e amount easily controllable and stable; 3 e good dispersion of laser light. All this features are fulfilled by moxibustion smoke particles. Thus, it has been chosen as a dust source for collection efficiency measurements with ELPI and as a tracer particles for PIV measurements. As it was measured by ELPI, almost 99% of moxibustion smoke particles are fine and submicron particles having diameters from 80 nm to 0.5 mm. The dust particles concentration was analyzed by ELPI at the ESP outlet. The ELPI was equipped with a 12 particle distributors, which can measure particle concentration for different diameters (0.02, 0.04, 0.08, 0.13, 0.21, 0.32, 0.50, 0.79, 1.26, 1.99, 3.13, 6.35 mm). For each electrodes geometry the dusty air sampling started at 0 kV and stepped every 10 kV, up to 50 kV. The collected data were averaged out by using software provided by the ELPI manufacturer. The ESP collection efficiency was calculated by using the following equation:

 hðrÞ ¼

1

 No ðrÞ $100% Ni ðrÞ

(1)

where, h(r) represents collection efficiency for each particle diameter (in %),No(r) represents outlet concentration of particles of size class r (in mg/m3), Ni(r) represents inlet concentration of particles of size class r (in mg/m3). The PIV measurements were carried out by using the second harmonic Nd:YAG laser as a light source. The light sheet produced from the laser beam was introduced along the ESP duct at the ESP symmetry plane, perpendicularly to the plate and wire electrodes. 200 images having resolution 2048  2048 pixels were captured for every single electrodes geometry and voltage, and then 100 instantaneous flow velocity fields were obtained by using a twoimages cross-correlation algorithm. Time-averaged flow velocity fields and apparent flow streamlines were calculated and presented afterwards. 2.2. Electrodes geometry The ESP duct was 600 mm long, 400 mm high and 200 mm wide. The original ESP geometry consisted of a three parallel plate collecting electrodes (on the top of the duct, bottom and in the middle between them). Thus, the distance between the plate electrodes was 200 mm. Four HV wire electrodes were mounted in the middle of each two plate electrodes (parallel to them); two wire electrodes in the top half, and other two in the bottom half of the ESP duct. The distance between the wire and plate electrodes was 100 mm, and the distance between the wire electrodes mounted in the same half of the ESP duct was 250 mm. The original ESP geometry hereafter is called to as ‘4hv3plates’. Widening of the ESP geometry by removing central plate collecting electrode (which corresponds to the removal of every second collecting electrode in a large industrial ESP) is proposed. After such modification the ESP with two plate electrodes (on the top and bottom of the ESP duct; 400 mm distance between them) and four HV wire electrodes in between them was obtained and hereafter is called to as ‘4hv’. Widening of the ESP just by removing plate collecting electrode inevitably decrease the collection efficiency. Therefore, in order to compliment the efficiency loss the geometries with additional grounded or HV wire electrodes were investigated. In order to dissipate the primary flow in the central area of the duct, an extra ionic wind was generated by lead-in the extra grounded or HV wire electrodes. Four different geometries with additional wire electrodes were tested.

78

Z. Ning et al. / Journal of Electrostatics 80 (2016) 76e84

Fig. 1. Schematic diagram of the experimental setup: 1 e laser power supply, 2 e synchronizer, 3 e laser head, 4 e optics, 5 e fan, 6 e computer, 7 e CCD camera, 8 e ELPI, 9 e ESP, 10 e dust generator, 11 e power supplier.

The first of them was composed by adding only one grounded wire electrode located at the geometric center of the ESP duct (this geometry hereafter called to as ‘4hv1g’). The distance from the additional grounded wire electrode to the plate electrodes was 200 mm, and to the HV wire electrodes 160 mm, as shown in Fig. 2a. The second geometry with additional wire electrodes was similar to the first one, but in this case the additional wire electrode (placed in the geometric center of the ESP duct) was supplied with a high voltage (same value as all other HV wire electrodes). This geometry hereafter is called to as ‘5hv’. The third geometry (hereafter called to as ‘4hv2g’) was composed by adding two grounded wire electrodes located at the center plane of the ESP duct. The distance between the additional grounded wires was 250 mm, as shown in Fig. 2b. The fourth geometry was similar to the third one (with two additional wires), but in this case all six wires were supplied with a high voltage, thus geometry hereafter is called to as ‘6hv’. The electrodes geometries with additional wire electrodes are presented in Fig. 2. During the ESP operation, the glow corona discharge (typical for positive corona) was observed on the surface of the additional grounded wire electrodes (for geometries 4hv1g and 4hv2g). In this case, corona discharge covered uniformly the entire surface of the

grounded wires. For the HV wire electrodes (supplied with a negative polarity) the spotted corona discharge with many tufts was observed. However, due to the distance to the plate electrodes, for geometries 5hv and 6hv the corona discharge from the additional wire electrodes was much weaker (only very dim lighting observed) than from the four surrounding HV wire electrodes. 3. Results 3.1. Electrical characteristics The currentevoltage characteristics for all investigated electrodes geometries are presented in Fig. 3. The corona onset voltage was consistently a little below 10 kV for all geometries. It can be explained by the same minimum distance (100 mm) between the HV wires and the grounded electrodes, what caused that the corona always started at the same voltage. As it can be seen in Fig. 3, for the original ESP geometry (4hv3plates) the corona current increased up to 626 mA at 50 kV. For the ESP with the central plate electrode removed (geometry 4hv) the corona current was almost exactly half that measured for 4hv3plates. One could expect that because the corona discharge from every HV wire electrode was directed to only one plate electrode instead of two (as for 4hv3plates). For geometries 5hv and

Fig. 2. Scheme of the electrodes geometries in the wide ESP with additional wire electrodes: (a) 4hv1g and 5hv, (b) 4hv2g and 6hv.

Z. Ning et al. / Journal of Electrostatics 80 (2016) 76e84

1600 4h hv3plates

1400

4h hv

Current [μA]

1200

4h hv1g

1000

4h hv2g

800

5h hv

600

6h hv

400 200 0 0

10

20 30 Voltagee [kV]

40

50 0

Fig. 3. Current-voltage characteristics for each electrodes geometry.

6hv the corona current was changed only slightly comparing to 4hv geometry because the additional HV wire electrodes were relatively far from the grounded plate electrodes. Moreover, the corona throttling phenomenon could occur. However, for geometries with additional grounded wire electrodes (4hv1g and 4hv2g) the current increased with a higher rate comparing to the other geometries. The reason is that additional grounded wires created extra paths for negative ions. It should be noted that for 4hv2g geometry the current increased fastest and at 50 kV reached 1373 mA. At this voltage the spark breakdown discharge occasionally happened and further increase of the voltage was not possible. 3.2. Particle collection efficiency In this section, the collection efficiency and energy consumption of the ESP having different geometries with additional wire electrodes were compared with the original ESP geometry (4hv3plates) and the 4hv benchmark wide ESP. The dust concentrations were measured by ELPI and on the basis of obtained results the collection efficiencies were calculated and are presented in Fig. 4. For geometries with grounded wire electrodes (4hv1g and 4hv2g) the collection efficiency was similar for low applied

79

voltages. At 20 kV the collection efficiency for both geometries with grounded wires was lower than for the 4hv benchmark, but with increasing voltage (30 kV and 40 kV) the collection efficiency gradually reached similar level as for the 4hv benchmark. However, at 50 kV the collection efficiency tendency for 4hv2g geometry stopped increasing rapidly, meeting its inflection point, what can be connected with a spark breakdown discharge occurring occasionally for this voltage. For the 4hv1g geometry the collection efficiency continually increased and for 50 kV surpassed the value obtained for the 4hv benchmark wide ESP. As for the collection efficiency as a function of the discharge power, in order to reach the same efficiency level the geometries with grounded wires consumed much more power than the original and benchmark geometries. For 4hv1g geometry, although the collection efficiency could surpass the 4hv benchmark at 50 kV, but it clearly consumed much more power to achieve the same efficiency level. More notably, despite the highest power consumption the 4hv2g geometry reached the lowest collection efficiency. For the same values of applied voltage, geometries with additional high voltage wire electrodes (5hv and 6hv) had higher collection efficiency than geometries with additional grounded wires and the 4hv benchmark geometry. Generally speaking, the collection efficiency for the 6hv geometry was closest to the original ESP (with the three plate collecting electrodes) and at 50 kV it was only 5% lower. Besides, the tendency of efficiency graphs for 5hv and 6hv geometries shows that increase of collection efficiency at higher applied voltages is probable because there is no inflection point even at 50 kV. Moreover, to obtain similar collection efficiency level the 5hv and 6hv geometries consumed almost the same amount of electrical power as the original ESP, and they were energetically more efficient than the 4hv benchmark geometry. All these features suggest that additional HV wire electrodes can be an alternative for replacing every second grounded plate. A widely accepted point is that the ESP collection efficiency will increase along with the voltage and the discharge power. However, our experiment showed that collection efficiency at some applied voltage can be increased just by adding HV wire electrodes, even if it will not significantly change the discharge power. One of the possible reasons is that the collection efficiency was changed because of the change of flow pattern inside the ESP. It suggests that the EHD flow pattern also plays an important role in particle removal process.

Fig. 4. Collection efficiency versus voltage and discharge power for each electrodes geometry.

80

Z. Ning et al. / Journal of Electrostatics 80 (2016) 76e84

3.3. Impact of flow pattern on collection efficiency 3.3.1. Flow velocity fields We speculate that differences in the ESP collection efficiency were caused by the different flow patterns induced for the different electrodes geometries. PIV method was used in order to quantitatively analyze how the different geometries influence the EHD flow pattern inside the ESP. The laser sheet introduced into the ESP duct and the position of the CCD camera defined the measurement area. In order to acquire a high quality frames we focused only on a lower side of the ESP duct, considering the flow pattern inside the ESP was symmetric. The results presented in the paper show the area extending from the lower collecting plate electrode to the 35 mm above the ESP symmetry plane. In Fig. 5 the contour maps of the velocity field and the apparent flow streamlines in the ESP with 4hv benchmark geometry are presented. Moreover, the foreground shows the electrodes geometry planes. The upper and lower planes are dedicated to the HV wires (always present). On the ESP symmetry plane the positions for HV and grounded wires (present or not, depending on the applied geometry) are marked with a dotted circles. The exact wires positions for each geometry can be found in Figs. 6 and 7. Other variables, including density of the trace particles, the turbine rotation speed, air temperature and humidity were all the same during measurements for the different geometries. When no voltage was applied the mean velocity of primary flow in the ESP duct was 0.2 m/s (Fig. 5a). After applying 10 kV (Fig. 5b) the corona discharge just started, was relatively weak and the flow velocity field in the 4hv benchmark ESP was changed very slightly (small changes in the streamlines paths behind the HV wires can be observed). For other electrodes geometries the flow velocity fields at 0 and 10 kV were very similar to these presented in Fig. 5 and are not presented in this paper. 3.3.1.1. 4hv1g geometry. Comparing to the 4hv benchmark geometry, the major modification of the EHD flow pattern for the 4hv1g geometry is that an extra ionic wind generated from the additional grounded wire electrode creates an inward force on entire gas flow, leading to squeeze the central flow. For 4hv1g setup, at the voltage 30 kV the intensity of the outlet vortex (near the second HV wire, at about x ¼ 350 mm) was much weaker comparing to the 4hv benchmark geometry, and a mass primary gas passing through with only little obstruction. Besides, in the central area, the additional grounded electrode generated the corona discharge of

opposite polarity than from the high voltage electrodes. So, part of the charged particles could become neutral in this area, what most probably decreased collection efficiency of these particles. However, for higher applied voltages (40 kV and 50 kV) the shape of the outlet vortex altered and the diameter was also expanded. Central vortex (between the first and the second HV wires, at about x ¼ 200 mm), which can become enough obstacle to the primary flow, was not only more intensive, but also form up earlier than for the 4hv geometry. This also suggests the reason why the collection efficiency was higher for the 4hv1g geometry than for the 4hv geometry at 40 kV and 50 kV. 3.3.1.2. 4hv2g geometry. For 4hv2g geometry the major modification in the EHD flow pattern comparing with the 4hv geometry was that two extra ionic winds were generated from the HV wire electrodes. It also leads to squeeze the central flow and the same “de-charge” issue remains as for 4hv1g geometry. Moreover, the distance between the HV and grounded wire electrodes decreased leading to the discharge current increase. Unfortunately, most of this energy was wasted because cannot charge particles efficiently. The ionic wind generated from the HV electrodes was no longer concentrated to the plates, but also pointing to the grounded wire electrodes (i.e. to the center of the ESP duct). Hence, migration velocity (velocity from the discharge to the collecting plate electrodes) significantly decreased. Besides, the electric potential of the additional grounded wire electrodes is positive comparing with the HV wires (supplied with negative voltage) so, the ionic wind generated from them has opposite orientation. At the outlet, the grounded electrodes are inducing part of a back flow. The central vortex cannot form even at 50 kV, causing the center part of the ESP duct becomes a free tunnel for many particles. The downside of this geometry was that the grounded wires were relatively close to the HV wire electrodes. The current increased rapidly and the breakdown possibility was greater than in other geometry cases. 3.3.1.3. 5hv geometry. For 5hv geometry the EHD flow patterns were very similar to that obtained for the 4hv benchmark. Some difference can be observed in the shape of the central vortex and the velocity in the outlet vortex. Comparing with the 4hv1g geometry, the major modification in the flow pattern is that the 5hv geometry generates ionic wind of opposite direction from the additional wire electrode. This dissipates the central vortex by acting against its revolving direction. Thus, only at 50 kV relatively weak central vortex was obtained for the 5hv geometry. However,

Fig. 5. Contour maps of velocity field and apparent flow streamlines in the ESP with 4 HV wire electrodes for 0 kV (a) and 10 kV (b). Positions for additional wires marked with dotted circles.

Z. Ning et al. / Journal of Electrostatics 80 (2016) 76e84

81

Fig. 6. Flow velocity fields measured in the ESP for different electrodes geometries and applied voltages.

adding an extra HV wire electrode in the center of the ESP duct makes the more particles can be charged effectively in this area. This geometry meets the requirement of increase particles charge, resulting in a good collection efficiency. 3.3.1.4. 6hv geometry. The EHD flow patterns generated in the ESP with 6hv geometry were similar to that obtained for the 4hv benchmark and 5hv. In the 6hv setup, no corona discharge was generated between the wire electrodes because they were supplied with the same voltage. Moreover, the additional HV wires were located in the same perpendicular plane as the base HV wires constituting 4hv setup, but the distance from additional wires to the plate electrodes was two times higher. Thus, the ionic wind generated from the additional HV wires was weak and had limited influence on the mean EHD flow pattern. However, for the 6hv geometry the intensity of each vortex increased and the migration velocity increased a little. The results showed that this geometry has highest collection efficiency among all tested wide ESP geometries. Addition of the HV electrodes in the center of the ESP more uniformly distributed the ionic wind and related negative ions in the duct, resulting in increase of the charge efficiency of the particles. So, even with similar mean EHD flow pattern, the collection efficiency could be increased. The obtained EHD flow results can be compared with the numerical simulation results presented in Refs. [13,14]. In the simulation a fixed voltage of 30 kV was applied and the primary flow varied from 0 to 1.0 m/s, which can be considered as an exceptional

supplement and comparison to our work. From the results obtained numerically a little similar flow patterns appeared when the primary flow velocity was 0.2 m/s. However, due to the non-centrally located HV wires in our wide ESP, the vortices generated between the HV wires and the nearest collecting plate showed a tendency to expanding volume because there was no obstruction from the vortices on the other side of the HV wires. Also the vortices in the downstream area were much harder to form, the shape and the location was also altered.

3.3.2. Quantification of electrohydrodynamic flow impact The ratio of the EHD number to the Reynolds number square (Re2) and the Turbulence Kinetic Energy (TKE) are used to quantify how the discharge will impact on the primary flow and the turbulence intensity. The equation and parameters for calculating the EHD number can be given as follows [15],

EHD ¼ I  L3

.  n2  r  mi  A

(2)

where: I represents discharge current (in mA); L represents characteristic length (¼ 0.1 m); n represents air kinematic viscosity, which is constant as 15  106 m2/s; r represents air density, ¼ 1.205 kg/m3; mi represents ion mobility, ¼ 2  104 m2/Vs;

82

Z. Ning et al. / Journal of Electrostatics 80 (2016) 76e84

Fig. 7. Apparent flow streamlines in the ESP for different electrodes geometries and applied voltage.

A represents discharge area, ¼ 0.12 m2. The Reynolds number can be obtained by the following equation,

Re ¼

d$u n

(3)

where: d represents characteristic length, ¼ 0.2 m; u represents primary flow mean velocity, ¼ 0.2 m/s; n represents air kinematic viscosity, which is constant as 15  106 m2/s. The EHD/Re2 ratio is applied to describe the intensity of the electrically induced secondary flow comparing to the primary flow. Table 1 EHD/Re2 value for different electrodes geometries. Voltage (kV)

10 20 30 40 50

If this ratio is greater than 1, it means that the secondary flow is the dominant factor, otherwise the primary flow dominates. The calculated results of this ratio are shown in Table 1. The results show that the value of the EHD/Re2 ratio increased with the voltage increase, indicating that electrically induced secondary flow gradually increases its influence on the primary flow. When EHD/Re2 is much lower than 1, the EHD force is weaker than viscosity resistance of the fluid. Thus, the results indicate that only direction of the primary flow was altered. However, when EHD/Re2 is approaching or greater than 1, both direction and flow velocity are altered. The geometries with the grounded wires show greater vulnerability than 5hv or 6hv geometries at the same applied voltage. Turbulence Kinetic Energy represents turbulence intensity. It can be obtained from the equation (4) [16],

  02 þ u02 k ¼ 0:5 u02 þ u z x y

(4)

where: k represents Turbulence Kinetic Energy;

EHD/Re2 4hv1g

4hv2g

5hv

6hv

0.02 0.93 4.12 9.55 18.88

0.04 1.50 7.01 17.84 29.80

0.04 0.39 1.61 3.58 6.60

0.20 0.91 2.43 4.84 8.25

u02 x represents impulse velocity in x direction; u02 y represents impulse velocity in y direction; u02 z represents impulse velocity in z direction. Since the velocity of original gas flow is parallel to the measurement plane, the anisotropic assumptions [17] can be used to

Z. Ning et al. / Journal of Electrostatics 80 (2016) 76e84

83

Table 2 Energy consumption for the ESP with different electrodes geometries.

U(kV) I(mA) P(W) Annual Q(kW.h)

4hv3plates original

4hv benchmark

4hv1g

4hv2g

5hv

6hv

35 257 9 78.84

54 425 23 201.48

49 816 40 350.4

e e e e

46 230 10.5 91.98

43 255 11 96.36

Table 3 Factors changed due to applying widen modification.

Before widening After widening

Collecting plate quantity

Total area of plate (m2)

Total mass of plates (kg)

Total mass of equipment (kg)

26 13

3743 1871.5

29681.61 14840.81

172742 157901

simplify the equation to:

  02 k ¼ 0:75 u02 þ u x y

(5)

Basing on the flow velocity fields measured by PIV, the TKE number was calculated for the entire flow field, for each geometry and at the highest applied voltage, i.e. at 50 kV. A noteworthy phenomenon is that the turbulence energy value shows direct correspondence to the particle collection efficiency at high voltage. The 6hv geometry has highest average TKE value (0.257) and it also has the highest collection efficiency. Accordingly is for the 4hv2g geometry which has the lowest TKE value (only 0.075) and the lowest efficiency. Other values for TKE are: 0.221 for the 4hv1g setup and 0.238 for 5hv setup. This result fits to the intention of the gas flow manipulation, i.e. to improve particle collection chance by alter the EHD flow. Very fine and submicron particles tend to follow the original flow at low TKE value. They can be carried out through a central high velocity tunnel, leading to a poor removal efficiency. However, with the higher value of TKE the particles have potential to move randomly, leading to less possibility to follow the original flow. In macro view, the removal efficiency increases correspondingly. 4. Economic analysis The economic analysis of the ESP power consumption and the primary investment cost was performed. For the power consumption analysis, the goal for collection efficiency of the fine and submicron particles (as used during our tests) was set at 75%. The comparison of the ESP with a different electrodes geometries in terms of energy usage is shown in Table 2. In order to achieve 75% collection efficiency, the ESP with the 4hv1g geometry needed 444% of power used by the ESP with the original 4hv3plates geometry, and 173% of that with the 4hv benchmark wide ESP. Even worse was for the 4hv2g geometry which failed to reach 75% efficiency (the maximum collection efficiency was 63%). The geometries with additional grounded wires clearly do not present enough potential to replace the central collecting plate. On the other hand, the 5hv and 6hv geometries needed relatively low energy usage to obtain 75% efficiency. They consumed much less energy than the 4hv benchmark setup, and only about 120% of energy used by the original 4hv3plates setup to achieve same collection efficiency. For the case of widening the exemplary existing SHWB60 ESP (cross section is 63.3 m2) [18] by removing every second collecting electrode, the primary investment costs of collecting plates were calculated and are shown in Table 3. Collecting plates are usually made of 304 stainless-steel (density 7.93  103 kg/m3). Given

current price is about 3700 USD per ton, thus, about 55000 USD can be saved if a normal ESP will be converted into a wide one. Considering that by removing every second collecting plate can save a lot of primary investment, the increased energy consumption (and related operational costs) obtained for the 5hv and 6hv geometries is not high. Moreover, the 5hv and 6hv geometries show good potential for further increasing collection efficiency because no inflection point after applying 50 kV. It suggests that the collection efficiency can significantly increase at higher applied voltages and potentially can reach the same values as for the original 4hv3plates geometry. The above advantages make the 5hv or 6hv geometry a good candidate to modify ESPs into a wide by removing every second collecting plate and adding HV wire electrodes. 5. Conclusions This article focus on investigations of the ESP with a wide wireplate geometry obtained by removing collecting electrode. Different electrodes geometries were compared during this work. The ESP collection efficiency and the related flow velocity field was measured by ELPI and PIV methods respectively. The results show that the wide ESP geometries with the additional HV wire electrodes meets relatively high collection efficiency and low energy consumption simultaneously. The analysis and the explanation of how the EHD flow inside the ESP will interfere with the particle removal was given. The EHD flow manipulation is one approach to improve the ESP collection efficiency. It was shown in this paper that the discharge character has significant impact on collection efficiency. Other approaches such as different shape of the collecting plates or applying different voltage to the central wire electrodes are theoretically possible. This could be the aim of future works. Acknowledgments This project was conducted at Industrial Ecology and Environment Research Institute of Zhejiang University, China and financially supported by the National High Technology Research and Development Program (2013AA065000), the Key Scientific and Technological Innovation Team Program of Zhejiang (2013TD07) and the National Science Centre, Poland (2014/13/D/ST8/03212). References [1] CAEPI, Development, Report on China electrical precipitation industry in 2012, China Environ. Prot. Ind. 6 (2013) 8e14. [2] CAEPI, Development, Report on electric precipitation industry of China in 2009, China Environ. Prot. Ind. 10 (2010) 17e20.

84

Z. Ning et al. / Journal of Electrostatics 80 (2016) 76e84

[3] X. Shen, Q. Zheng, Z. Ning, S. Wang, P. Han, K. Yan, PM 10 and PM 2.5 emission control by electrostatic precipitator (ESP) for coal-fired power plants IV: investigations on electrostatic precipitation by means of 2D PIV technique, Sci. Technol. Rev. 32 (2014) 43e50, http://dx.doi.org/10.3981/j.issn.10007857.2014.33.005.  ski, J. Mizeraczyk, Electrohydrodynamic flow [4] A. Niewulis, A. Berendt, J. Podlin patterns and collection efficiency in narrow wire-cylinder type electrostatic precipitator, J. Electrostat. 71 (2013) 808e814, http://dx.doi.org/10.1016/ j.elstat.2013.02.002. s pour optimiser les [5] K. Ulrich, E. Walter, G. Edgar, Des outils de calcul avance lectrofiltres, Rev. ABB 4 (1999) 33e42. performances des e [6] D. Brocilo, J.-S. Chang, R. Godard, A.A. Berezin, Numerical modelling of wirepipe electrostatic precipitator for control of fine particulates, J. Aerosol Sci. 30 (1999) S855eS856, http://dx.doi.org/10.1016/S0021-8502(99)80438-8. [7] P. Atten, F.M.J. McCluskey, A.C. Lahjomri, The electrohydrodynamic origin of turbulence in electrostatic precipitators, IEEE Trans. Ind. Appl. IA-23 (1987) 705e711, http://dx.doi.org/10.1109/TIA.1987.4504969. [8] J.S. Chang, J. Dekowski, J. Podlinski, D. Brocilo, K. Urashima, J. Mizeraczyk, Electrohydrodynamic gas flow regime map in a wire-plate electrostatic precipitator under positive coronas, in: CEIDP '05. 2005 Annu. Rep. Conf. Electr. Insul. Dielectr. Phenomena, 2005., IEEE, 2005, pp. 461e464, http://dx.doi.org/ 10.1109/CEIDP.2005.1560720. [9] A. Soldati, S. Banerjee, Turbulence modification by large-scale organized electrohydrodynamic flows, Phys. Fluids 10 (1998) 1742, http://dx.doi.org/ 10.1063/1.869691. [10] T. Yamamoto, M. Okuda, M. Okubo, Three-dimensional ionic wind and electrohydrodynamics of tuft/point corona electrostatic precipitator, IEEE Trans.

Ind. Appl. 39 (2003) 1602e1607, http://dx.doi.org/10.1109/TIA.2003.818983. [11] J. Podlinski, A. Niewulis, J. Mizeraczyk, Electrohydrodynamic flow in a wireplate non-thermal plasma reactor measured by 3D PIV method, Eur. Phys. J. D. 54 (2009) 153e158, http://dx.doi.org/10.1140/epjd/e2009-00040-x. [12] P.C.L. Kwok, S.J. Trietsch, M. Kumon, H.-K. Chan, Electrostatic charge characteristics of jet nebulized aerosols, J. Aerosol Med. Pulm. Drug Deliv. 23 (2010) 149e159, http://dx.doi.org/10.1089/jamp.2009.0795. [13] K. Adamiak, P. Atten, Numerical simulation of the 2-D gas flow modified by the action of charged fine particles in a single-wire ESP, IEEE Trans. Dielectr. Electr. Insul 16 (2009) 608e614, http://dx.doi.org/10.1109/ TDEI.2009.5128495. [14] N. Farnoosh, K. Adamiak, G.S.P. Castle, 3-D numerical analysis of EHD turbulent flow and mono-disperse charged particle transport and collection in a wire-plate ESP, J. Electrostat. 68 (2010) 513e522, http://dx.doi.org/10.1016/ j.elstat.2010.07.002. [15] IEEE-DEIS-EHD Technical Committee, Recommended international standard for dimensionless parameters used in electrohydrodynamics, IEEE Trans. Dielectr. Electr. Insul. 10 (2003) 3e6, http://dx.doi.org/10.1109/ TDEI.2003.1176545. [16] Z. Gao, J. Han, Y. Xu, Y. Bao, Z. Li, Particle image velocimetry (PIV) investigation of flow characteristics in confined impinging jet reactors, Ind. Eng. Chem. Res. 52 (2013) 11779e11786, http://dx.doi.org/10.1021/ie400891u. [17] Y. Liu, M.G. Olsen, R.O. Fox, Turbulence in a microscale planar confined impinging-jets reactor, Lab. Chip 9 (2009) 1110, http://dx.doi.org/10.1039/ b818617k. [18] D. Zhang, The Design Manual of Dust Removal Engineering, Chemical Industry Press, 2003.