ARTICLE IN PRESS

0304-3886/$ - se

doi:10.1016/j.el

�CorrespondE-mail addr

Journal of Electrostatics 65 (2007) 228–232

www.elsevier.com/locate/elstat

The experimental research on electrode configuration and dischargecharacteristics of pulse discharge

Li Jie�, Shang Kefeng, Wu Yan, Wang Ninghui, Zhang Yi

Dalian University of Technology, Institute of Electrostatics & Specific Power, Dalian 116024, China

Received 9 November 2003; received in revised form 1 June 2004; accepted 15 August 2006

Available online 11 September 2006

Abstract

Wire–plate-type reactors with the pulse-induced plasma chemical process (PPCP) have been extensively investigated for flue gas

cleaning such as SOX and NOX, and the electrode configuration of reactors such as the plate-to-plate spacing and the wire-to-wire

spacing is directly related to the formation of discharge plasma and the utilization efficiency of discharge energy, so it is useful to study

the electrode configuration in order to provide a basis for PPCP applications. In this paper, the influence of these factors such as the

plate-to-plate spacing and the wire-to-wire spacing on the discharge characteristics was investigated using a positive, nanosecond pulse

generator. Some experimental results were obtained and discussed. It was found that when the wire-to-wire spacing (a) and the plate-to-

plate spacing (c) of reactors satisfied the relation below, a=c ¼ 0:4� 0:6, the primary streamer energy was more bigger.

r 2006 Elsevier B.V. All rights reserved.

Keywords: Wire–plate electrode; Electrode configuration; Discharge characteristics; Pulsed discharge plasma

1. Introduction

Since Masuda first utilized corona plasma produced by ananosecond pulse generator to remove NOX and SO2 fromflue gas in 1986 [1], scholars have extensively investigatedthe pulse-induced plasma chemical process (PPCP), andhave achieved some valuable results. Masuda et al. mainlyinvestigated the relationship between the denitrificationefficiency and these factors such as discharge energy, flowrate of flue gas, residence time and temperature of reactors[1,2]. Mizuno et al. [3,4] studied the effect of wet and semi-wet-type plasma reactors on desulfurization and denitrifi-cation efficiency, and concluded that pulse voltage withDC-bias voltage was more useful for desulfurization anddenitrification. Chang et al. [5] studied the reduction ofNOX by a corona radical injection method. Wu et al. [6]utilized the water vapor activation process for producingmore radicals to improve the entire desulfurizationefficiency of flue gas. Fujii and Rea [7] researched theeffect of circuit parameters on pulse discharge character-istics, and Wang et al. [8] investigated the matching

e front matter r 2006 Elsevier B.V. All rights reserved.

stat.2006.08.003

ing author. Tel.:/fax +86 411 84708576.

ess: lijie@dlut.edu.cn (L. Jie).

relationship between the pulse generator and the reactor,and achieved the primary parameters influencing pulse–voltage waveforms. Oda et al. [9] utilized nonthermalplasma with a combination of additives and catalyst totreat nitric oxide and achieved 90% NOX removalefficiency. Simultaneously, researchers started to carryout large-scale experimental investigations for industrialapplication. Dinelli et al. [10], Mok and Nam [11], and Wuet al. [12] carried out desulfurization and denitrificationresearch with a flow rate of 1000, 3228 and12,000–20,000Nm3/h, respectively.In this paper, the effect of electrode configuration

(including electrode spacing and corona wire distribution)on pulse discharge characteristics (including peak pulsevoltage, DC-bias voltage, energy of pulsed discharge andprimary streamer) was studied.

2. Experimental instruments

The schematic diagram of a wire–plate-type reactor isshown in Fig. 1. In our experiments, the frame (withdimensions of H�L: 85� 190 cm) made by steel pipes andcorona wires was used as anode, two parallel steel plates

ARTICLE IN PRESSL. Jie et al. / Journal of Electrostatics 65 (2007) 228–232 229

with dimensions of 100� 200 cm2 were used as groundelectrode, and rectangular steel wires with dimensions of4� 4mm2 were used as corona wires. The plate-to-platespacing (c) and the wire-to-wire spacing (a) were adjus-table. The pulse high-voltage was supplied by a positivenarrow pulse generator with a rotary spark-gap switch.The schematic of the circuit and experimental setup isshown in Fig. 2. The rise time of pulse voltage was less than100 ns, the width was in the range of 200–400 ns, and thepeak pulse voltage was adjustable in the range of 0–100 kV.

The voltage and current signals were sampled andrecorded by an HP54810A digital oscilloscope (HPCorporation) with different probes, namely voltage probeHV-100K (PEEC of Japan), and current probe (TektronixCorporation of USA), respectively. The discharge energywas simultaneously calculated and analyzed. The staticparameters (mainly the reactors’ capacitance and induc-tance when reactors are not supplied with pulse voltage) ofwire–plate discharge system were measured with PM6303Amulti-analyzer (FLUKE Corporation of USA).

3. Results and discussion

3.1. Relations between plate-to-plate spacing and discharge

characteristics

Table 1 shows the effects of the plate-to-plate spacing (c)on the peak pulse voltage, DC-bias voltage and single

ab

c

wire

plate

H

L

(a: Wire-to-wire spacing, b: Wire-to-plate spacing, c: Plate-to-plate spacing)

Wire electrode

Plate electrode

Fig. 1. Schematic of electrode configuration in a wire–plate-type reactor.

Fig. 2. Schematic of the circui

discharge energy. In our experiments, the plate-to-platespacings were 15, 20 and 25 cm, and the wire-to-wirespacings (a) were 6 and 9 cm, respectively. The value of thepulse-forming condenser was 2 nF and the output of theDC high-voltage generator was 85 kV. Analyzing the datain Table 1, we see that the peak pulse voltage (Vp) and theDC-bias voltage (Vb) increased with the augment of theplate-to-plate spacing (c) at the fixed wire-to-wire spacing(a) and output of the DC high-voltage generator; however,the single pulse energy (W) decreased with an increase inthe plate-to-plate spacing (c).When the number of the corona wires and the wire-to-

wire spacing are fixed, the wire-to-plate spacing isrelated to the static parameters (mainly the static capaci-tance and inductance) of reactors, the static capacitance ofreactors decreases correspondingly with an increase in thewire-to-plate spacing. We think the reasons for ourexperiments are as follows: the inception voltage ofstreamer corona increases with the augment of the wire-to-plate spacing, the wider the wire-to-plate spacing is, thehigher the inception voltage of streamer corona needs; butafter the streamer corona discharge starts, the appliedvoltage has little influence on the development ofstreamers. When the wire-to-plate spacing is narrow,streamer corona discharge of reactors starts at relativelylower applied voltage, and then the increase of appliedvoltage has little influence on the peak pulse voltage. Onthe other hand, streamer corona discharge of reactors withwide wire-to-plate spacing starts at higher applied voltage,so the peak pulse voltage is higher than that of the narrowwire-to-plate spacing; similarly, residual voltage rises withan increase in wire-to-plate spacing after streamer coronaquenches. But when wire-to-plate spacing is narrow, thedensity of the streamer corona and the primary streamerenergy are bigger.

3.2. Relations between wire-to-wire spacing and discharge

characteristics

In the experiments below, when the plate-to-platespacings were 15 and 20 cm, the numbers of discharge

t and experimental system.

ARTICLE IN PRESSL. Jie et al. / Journal of Electrostatics 65 (2007) 228–232230

wires were 15 and 17, and the invariable output voltages ofthe DC generator were 50 and 70 kV, respectively. Thechange relations of peak pulse voltage (Vp), DC-biasvoltage (Vb) and primary streamer energy (Wf) accordingto the wire-to-wire spacing (a) are shown in Fig. 3(1 and 2),Fig. 4(1 and 2) and Fig. 5(1 and 2), respectively. Typicalwaveforms of pulse voltage are shown in Fig. 6.

From the curves, we can see that a certain range of themaximal peak pulse voltage and primary streamer energyexisted while the wire-to-wire spacing changed; the DC-bias voltage decreased with an increase in the wire-to-wirespacing. We think the reason is that streamer corona willbe influenced by neighboring wires in the course of their

Table 1

Influence of plate-to-plate spacing on discharge characteristics

a (cm) c (cm) CR (nF) Vp (kV) Vb (kV) W (J/pulse)

6 15 0.39 67.0 15.5 2.500

20 0.35 73.8 19.3 1.780

25 0.31 83.0 22.5 1.285

9 15 0.34 65.5 14.6 2.360

20 0.32 69.8 17.4 1.735

25 0.29 75.0 18.0 1.095

2 4 6 8 10 1220

24

28

32

36

40

c=15cm

Vp/

kV

a/cm(1)

Fig. 3. Relations between peak pulse volt

2 6 10 120

2

4

6

8

10

12

14

c=15cm

Vb/

kV

a/cm

4 8

(1)

Fig. 4. Relations between DC-bias volta

transferring from corona wires to ground plates. When thewire-to-wire spacing is narrow, neighboring wires willinterfere in the transfer of streamer corona and reduce theelectric field intensity near corona wires [13]; on thecontrary, when the wire-to-wire spacing is wide, while thespacing increases, the sustaining role of streamers willweaken gradually. At an appropriate range, streamercorona produced by adjacent wires can support each otherand this facilitates the transfer of streamers. According toour experiments, an appropriate relation of the wire-to-wire spacing (a) and the plate-to-plate spacing (c) existed,namely a=c ¼ 0:420:6. When the relation was satisfied, theprimary streamer energy was bigger.

3.3. Relations between corona wire distribution and

discharge characteristics

In our experiments, the output voltage (50 kV) of the DChigh-voltage generator and the pulse discharge circuit wereinvariable. The plate-to-plate spacing was 15 cm and thelength of electric field was 160 cm. Because of the invariablelength of electric field, when the wire-to-wire spacingbecame narrower, more discharge wires were needed, e.g.the wire-to-wire spacing changed from 4.5 to 10.5 cm, thenumber of corona wires decreased from 34 to 15.

5 6 7 8 9 10 11 12 130

10

20

30

40

50

60

70

80

c=20cm

Vp/

kV

a/cm(2)

age (Vp) and wire-to-wire spacing (a).

5 6 7 8 9 10 11 12 130

5

10

15

20

25

c=20cm

Vb/

kV

a/cm(2)

ge (Vb) and wire-to-wire spacing (a).

ARTICLE IN PRESS

2 4 6 8 10 120.0

0.4

0.8

1.2

1.6

c=15cm

W/(

*k2)

J

a/cm5 6 7 8 9 10 11 12 13

0

1

2

3

4

5

6

c=20cm

W/(

*K1)

J

a/cm(1) (2)

Fig. 5. Relations between primary streamer energy (W) and wire-to-wire spacing (a).

Fig. 6. Waveforms of pulse voltage at different wire-to-wire spacing (a)

(the number of corona wires was 15).

0 5 10 150

10

20

30

40

50

60

Vp/

kV

a/cm

Fig. 7. Relations between peak pulse voltage (Vp) and wire-to-wire

spacing (a).

0 10 150

1

2

3

4

5

6

7

8

Vb/

kV

a/cm

5

Fig. 8. Relations between DC-bias voltage (Vb) and wire-to-wire spac-

ing (a).

0 5 10 150.0

0.5

1.0

1.5

2.0

W/(

*k3)

J

a/cm

Fig. 9. Relations between primary streamer energy (W) and wire-to-wire

spacing (a).

L. Jie et al. / Journal of Electrostatics 65 (2007) 228–232 231

The influence of the wire-to-wire spacing on peak pulsevoltage, DC-bias voltage and primary streamer energy isshown in Fig. 7, Fig. 8 and Fig. 9, respectively. From the

curves, we can see that peak pulse voltage, DC-bias voltageand primary streamer energy changed with an increase inwire-to-wire spacing. Fig. 10 shows the typical voltagewaveforms at different wire-to-wire spacing.

ARTICLE IN PRESS

Fig. 10. Waveforms of pulse voltage at different wire-to-wire spacing (a)

(the length of electric field is 160 cm).

L. Jie et al. / Journal of Electrostatics 65 (2007) 228–232232

4. Conclusions

When the number of corona wires, the wire-to-wirespacing and the output of the DC high-voltage generatorwere kept invariable, as the plate-to-plate spacing in-creased, the peak pulse voltage and the DC-bias voltagerose; the single pulse energy decreased.

When the number of corona wires, the plate-to-platespacing and the output of the DC high-voltage generatorwere kept invariable, as the wire-to-wire spacing changed, acertain range of maximal peak pulse voltage and primarystreamer energy existed. In our experiments, when the wire-to-wire spacing (a) and the plate-to-plate spacing (c)satisfied the relation below, a=c ¼ 0:420:6, the primarystreamer energy was injected into the reactor was more.

When the length of the electric field, the plate-to-platespacing and the output of the DC high-voltage generatorwere kept invariable, as the wire-to-wire spacing increased(namely the number of corona wires decreased), the peakpulse voltage, the DC-bias voltage and the primarystreamer energy decreased gradually. But when the numberof corona wires decreased to some extent, DC-bias voltagewas inclined to rise.

Acknowledgment

This research is financially supported by The Ministry ofScience and Technology of China.

References

[1] S. Masuda, H. Nakao, Control of NOx by positive and negative

pulsed corona discharges, in: Proceedings of the IEEE/IAS Annual

Meeting, Denver, USA, 1986, pp. 1173–1182.

[2] S. Masuda, Y Wu, Removal of NOx by corona discharge induced

by sharp rising nanosecond pulse voltage, in: Proceedings of

the International Conference on Electrostatics, Oxford, 1987,

pp. 249–254.

[3] A. Mizuno, J.S. Clements, R.H. Davis, A method for the removal of

sulfur dioxide from exhaust gas utilizing pulsed streamer corona for

electron energization, IEEE Trans. Ind. Appl. IA-22 (3) (1986)

516–521.

[4] A. Mizuno, K. Shimizu, T. Matsuoka, S. Furuta, Reactive

absorption of NOx using wet discharge plasma reactor,, IEEE Trans.

Ind. Appl. 31 (6) (1995) 1463–1468.

[5] J.S. Chang, P.C. Looy, J. Pevler, T. Yoshioka, K. Nagai, Reduction

NOx from a combustion flue gas by a corona radical injection

method, in: Proceedings of the IEEE/IAS, Annual Meeting, 1993,

pp. 1969–1975.

[6] Y. Wu, J. Li, N. Wang, G. Li, D. Xu, Study on increasing the SO2

removal efficiency with the radicals produced by H2O in pulse

discharge plasma process, Jpn. J. Appl. Phys, Part 2, No. 8A, 40

(2001) L838–L840.

[7] T. Fujii, M. Rea. Pulse corona characteristics, IEEE Trans. Ind.

Appl. 29 (1) (1993) 98–102.

[8] Wang Rongyi, Zhang Baoan, Bing Sun, Ninghui Wang, Yan Wu,

Weimi Gong, Shuhai Liu, Apparent energy yield of a high efficiency

pulse generator with respect to SO2 and NOx removal, J. Electrostat.

34 (1995) 355–366.

[9] T. Oda, T. Kato, T. Takahashi, K. Shimizu, Nitric oxide decom-

position in air by using nonthermal plasma processing with additives

and catalyst, IEEE Trans. Ind. Appl. 34 (2) (1998) 268–272.

[10] G. Dinelli, L. Civitano, M. Rea, Industrial experiments on pulse

simultaneous removal of NOx and SOx from flue gas, IEEE Trans.

Ind. Appl. 26 (3) (1990) 535–541.

[11] Young Sun Mok, In-Sik Nam, Positive pulsed corona discharge

process for simultaneous removal of SO2 and NOx from iron-ore

sintering flue gas, IEEE Trans. Plasma Sci. 27(4) (1999) 1188–1196.

[12] Yan Wu, Jie Li, Ninghui Wang, Guofeng Li, Industrial experiments

on desulfurization of flue gases by pulsed corona induced plasma

chemical process, J. Electrostat. 57(3–4) (2003) 233–241.

[13] Y.H. Kim, S.H. Hong, Two-dimensional simulation images of pulsed

corona discharges in a wire–plate reactor, IEEE Trans. Plasma Sci. 30

(1) (2002) 168–169.