Onset Voltage of Corona Discharge in Wire-Duct Electrostatic Precipitators

Transcription

1 36 International Journal of Plasma Environmental Science & Technology, ol.4, No.1, MACH 1 Onset oltage of Corona Discharge in Wire-Duct Electrostatic Preciitators H. Ziedan 1, A. Sayed 1, A. Mizuno, and A. Ahmed 1 1 Electrical Engineering Deartment, Assiut University, Egyt Deartment of Ecological Engineering, Toyohashi University of Technology, Jaan Abstract This aer is aimed at investigating how the onset voltage of corona in an electrostatic reciitator is influenced by the number of discharge wires, the wires' radius and the sacing between wires and the collecting lates. The initiation of the corona discharge on the wire surface is located for each wire. The onset voltage is measured for a laboratory model of a reciitator and calculated based on the criterion of self-recurrence of electron avalanches growing in the vicinity of the discharge wires. This calls for calculating the electric field in the vicinity of the discharge wires using the charge simulation method. The calculated onset voltage values agreed reasonably with those measured eerimentally. Keywords Onset voltage, corona discharge, electric field, electrostatic reciitators I. INTODUCTION In industrial countries, standards were adoted to imose some control over the emission for dark smoke into the atmoshere and, hence, to reduce atmosheric ollution. Electrostatic reciitation [1] is an efficient method of cleaning industrial gases from susended. The basic rinciles governing the oeration of electrostatic reciitators are relatively straightforward and are well described in the literature [1, ]. Conventional duct-tye reciitators are made u of a number of discharge wires hanged vertically between grounded collecting lates. A high negative voltage is alied to the discharge wires and negative ions are formed by corona discharge at the wires. These ions accelerate to the collecting lates and charge the articles in their way. When a charged article reaches the collecting late, the charge is neutralized and the article is collected. An accurate knowledge of the electrostatic field distribution and the onset voltage of corona from the discharge wires is of fundamental imortance in develoing good models of the different rocesses that take lace in reciitators [3]. Aroimate electric field values at the corona onset were calculated in duct-tye reciitators with an infinite number of discharge wires at a regular sacing using equations develoed before [4]. The voltage that would roduce the onset field at the surface of the discharge wires was considered the corona onset voltage and was eressed by an equation for an infinite number of discharge wires at a regular sacing [4]. A method was described [5] for calculating the onset voltage of corona from the discharge wires in duct-tye reciitators. The onset voltage was evaluated for the central and the outer wires of a 7-wire reciitator. No attention was forwarded to locate the initiation of the corona discharge on the wire surface as the corona is not Corresonding author: Hamdy Ziedan address: ziedan9@yahoo.com initiated at the same oint on the discharge wires. This aer is aimed at investigating how the onset voltage of corona in an electrostatic reciitator is influenced by the number of discharge wires, the wires' radius and the sacing between wires and the collecting lates. The initiation of the corona discharge on the wire surface is located for each wire. The onset voltage is measured for a laboratory model of a reciitator and calculated based on the criterion of self-recurrence of electron avalanches growing in the vicinity of the discharge wires. This calls for calculating the electric field in the vicinity of the discharge wires using the charge simulation method. The calculated onset voltage values agreed reasonably with those measured eerimentally. II. ELECTIC FIELD CALCILATION The well-known, charge-simulation technique [6, 7] is used to calculate the electric field for wire-duct ESP with m wires (m is odd). Fig. 1 shows a one-quadrant of the cross-section of the ESP in the X-Y lane. For simlicity, the reciitator is assumed infinitely long the Z-direction. The surface charge on each wire is simulated by (N 1 ) line charges located at radius rf, where r f = f r c, where f is a fraction, chosen.5 in the resent work and rc is the radius of discharge wire. The surface charge on each late of the ESP is simulated by a number (N ) line charges located outside the late at a distance from the late (a) equal to the distance between two adacent simulation charges (b) as shown in Fig. 1. Thus, the total number of unknown simulation charges is (mn 1 + N ). There is a symmetry around both - and y- aes, Fig. 1, reduces the number of unknowns to n (= (mn 1 + N )/4). To evaluate the unknown simulation charges Q, = 1,, 3,, n, a set of boundary oints equal to the simulation charges is chosen on the surface of discharge wires and collecting lates as shown in Fig. 1 to satisfy the boundary conditions: Φ = at discharge wires, Φ = at collecting lates eceived; December 15, 9

2 Ziedan et al. 37 a b N Simulated Charges Collecting late Boundary Points Y Ionization zone N 1 r f r c Fig. 1. Arrangement of simulation charges and boundary oints for one quarter of a wire duct ESP. X r i r c avalanche Collecting late The otential Φ i is at the ith boundary oint of coordinates ( i, y i ) is the sum of otential contributions due to all simulation charges. i n 1 Where, P i is the otential coefficient eressed as: P i ln 1/ where, P 1 i Q i = 1,, 3, 4,, n = 1,, 3, 4,, n i yi y i yi y i yi y y y i (, y ) are the coordinates of the th simulation charge. Satisfaction of the boundary conditions at the boundary oints formulates a set of equations relating the values of simulation charges to the otential values at the boundary oints. This is eressed in a matri form as: [P] [Q] = [ b ] (3) where [P] is the otential coefficient matri (with dimension n n), [Q] are unknown simulation charges (with dimension n 1), and [ b ] are the otential values of the boundary oints (with dimension n 1). Solution of the set of Eq. (3) determines the unknown simulation charges Q. To check the accuracy of the solution, a set of check oints is chosen (each check oint is located between two successive boundary oints). The otential value is to be checked against the alied value ( at the wire surface and zero at the collecting lates). Once the accuracy is checked and the simulation charges are known, the electric field intensity at any oint (, y ) can be determined: n E Q (4) n E y Q y y y y (5) The magnitude of the electric field intensity at oint is calculated as: E E E [/m] (6) y i 3 4 (1) () Fig.. The develoment of an avalanche. III. COONA ONSET OLTAGE CALCULATION With the wires of a wire-duct reciitator stressed by negative H suly, the electric field near a wire reaches the threshold value for ionization of gas molecules by electron collision. A rimary electron avalanche starts to develo along the direction of maimum field away from the wire. The avalanche grows through the so-called ionization zone whose boundary defines the zone where the ionization coefficient eceeds the coefficient of electron attachment, Fig.. In order to simlify the calculations, the following assumtions are made: 1) The ionizing-zone etends from the wire surface to the oint where almost all (i.e %) of the electrons get attached to neutral gas molecules and from negative ions. ) The electron avalanche grows under the influence of its own sace-charge field besides the alied electrostatic field. 3) The sace-charge-field of the avalanche is the same as if all its ositive ions were concentrated at a distance 1/ from its ti. This assumtion has been found satisfactory in similar work on uniform and nonuniform fields [8]. At the onset voltage, the avalanche should somehow rovide an initiating electron at the wire surface to start a successor avalanche, ossibly by hotoemission, ositive ion imact, metastable action or field emission. Field emission is ossible only at field strengths eceeding /m [9]. Electron emission by ositive ion imact is more than two orders of magnitude less robable than hotoemission [1]. Metastables have been reorted to have an effect aroimately equal to that of ositive ion imact [11]. Therefore, only the first mechanism (electron emission by hotons) was considered in determining the onset voltage. The condition for a new (successor) avalanche to develo [5, 9] is ri rc d 1 h e g e d (7)

3 38 International Journal of Plasma Environmental Science & Technology, ol.4, No.1, MACH 1 H rc (a) One wire of ESP L3 L L1 1 Central 3 H d d (b) Multi-wires of ESP Fig. 3. Configuration of wire-duct ESP. where Townsend s first ionization coefficient h : coefficient of electron emission by the action of hotons : electron attachment coefficient g () :geometry factor to account for the factor that some hotons will not be received by the discharge wires of ESP, [5, 9]. The corona onset voltage does not aear elicitly in last relation, however, the alied voltage affects the values of,,.. The onset voltage is the critical value which fulfills Eq. (7). Y X I. DISCHAGE PAAMETES In order to calculate the onset voltage of corona on each wire of the ESP at atmosheric ressure, the relation (7) was solved using the values available in the literature [1], for,, h and. The equations relating [m-1] and [m-1] at atmosheric ressure to the electric field E [/m] were eressed as: (8) E E e E Fig. 4. Electrostatic Field Distribution along y-ais for an alied voltage of 5 k. Y X (9) for (E) E E (1) for (E) The coefficient of hoton absortion was taken 5 m-1 [1]. The coefficient of electron emission by hotons h was taken [1] e 3. ESULTS AND DISCUSSION A. Accuracy of charge simulation technique The accuracy of charge simulation technique is checked by investigating how the boundary conditions Fig. 5. Electrostatic field distribution along y-ais for ESPs with one-, three-, five- and seven-wires at an alied voltage of 5 k. are satisfied in cases of one and multi-discharge wires (3, 5, 7 discharge wires), Fig. 3. It is satisfactory that the maimum ercentage error of the calculated surface otential of discharge wires did not eceed 1-6 and the maimum ercentage error of the calculated surface otential of collecting lates did not eceed (Theoretically, the otential at any oint along the

4 Ziedan et al One-wire ESP Three-wires ESP Five-wires ESP Seven-wires ESP Electrostatic Field, /m Distance along y ais, m Fig. 6. Electrostatic field distribution along y-ais near the central discharge wire for ESPs with one-, three-, five- and seven-wires at an alied voltage of 5 k. 4 Angle Ө around wire erihery, Degree Fig. 9. Electrostatic Field Distribution around the discharge wire for three-wire ESP (at 5 k) Electrostatic Field, /m One-wire ESP Three-wires ESP Five-wires ESP Seven-wires ESP Angle Ө around wire erihery, Degree Distance along y ais, m Fig. 7. Electrostatic field distribution along y-ais near the collecting late for ESPs with one-, three-, five- and seven-wires at an alied voltage of 5 k. Fig. 1. Electrostatic Field Distribution around the discharge wire for five-wire ESP (at 5 k). o 18o o 7o Angle Ө around wire erihery, Degree Fig. 8. Electrostatic Field Distribution around the discharge wire for one-wire ESP (at 5 k). collecting lates should be zero) irresective of the number of discharge wires and ESP geometry (wire radius rc, wire-to-wire sacing d and wire-to-late sacing H of ESP). The discharge wires of ESP (in case of multidischarge wires) are arranged in a row equidistant from the two collecting ground lates. All the wires have the same alied voltage and the same diameter. For comarison urose, the wire-duct ESP arameters as reorted in the literature [13], (rc =.85 mm, d = 7.6 cm, Angle Ө around wire erihery, Degree Fig. 11. Electrostatic Field Distribution around the discharge wire for seven-wire ESP (at 5 k). and H = 3.8 cm) are considered. The decrease of the electrostatic field intensity along the vertical ais of symmetry (y-ais) is shown in Fig. 4 for alied voltage of 5 k (less than the onset voltage). It is quit clear that all the resent calculations agreed well with those reorted before [13]. This confirms the accuracy of the resent calculations.

5 International Journal of Plasma Environmental Science & Technology, ol.4, No.1, MACH 1 Angle Өma, Degree 4 Fig. 15. ariation of corona onset voltage with the wire-to-wire sacing of three-wires ESP. Angle Өma, Degree Fig. 1. ariation of angle Өma with the wire-to-wire sacing of three-wires ESP. Fig. 16. ariation of corona onset voltage with the wire-to-wire sacing of five-wires ESP. Angle Өma, Degree Fig. 13. ariation of angle Өma with the wire-to-wire sacing of five-wires ESP. Fig. 14. ariation of angle Өma with the wire-to-wire sacing of seven-wires ESP. B. Electrostatic Field along ESP ais The electrostatic field intensity along the ESP ais (Y-ais) starting from the surface of the central discharge wire u to collecting late is shown in Fig. 5 for ESPs with one-, three-, five- and seven-discharge wires. It is clear that the field near the central discharge wire in case of one-discharge wire ESP is higher than that for ESP with multi-wires. This is simly elained by the shielding effect imosed on the central wire due to other wires, Figs. 5 and 6. On the other hand, the field near the collecting late for one wire ESP is smaller than that for ESP with multi-wires, Figs. 5 and 7. This conforms to the fact that the voltage alied to the discharge wires is the Fig. 17. ariation of corona onset voltage with the wire-to-wire sacing of seven-wires ESP. same, so the line integral of the field value from the central wire surface u to the collecting late should be the same whatever the number of discharge wires. C. Electrostatic Field around Discharge Wires Electrostatic field distributions around the dischargewires of different ESP configurations are shown in Figs In all ESP configurations, the electrostatic field around the outer wire is higher than that around the other (inner) wires. This is because the inside wires are shielded by the outer wires.

6 Ziedan et al. 41 TABLE I CALCULATED COONA ONSET OLTAGE ALUES FO DIFFEENT ESP CONFIGUATION Corona onset oltage, k Central-wire 1-wire -wire 3-wire 1-wire ESP wires ESP wires ESP wires ESP TABLE II AIATION OF ANGLE Ө ma AND ONSET OLTAGE OF COONA ON THE DISCHAGE WIES AS INFLUENCED BY THE WIE-TO-WIE SPACING OF THEE-WIE ESP wire-to-wire sacing, cm.5 d 1. d 1.5 d. d.5 d 3. d 3.5 d 4. d ight (outside) Central wire wire Ө ma onset, k Ө ma onset, k TABLE III AIATION OF ANGLE Ө ma AND ONSET OLTAGE OF COONA ON THE DISCHAGE WIES AS INFLUENCED BY THE WIE-TO-WIE SPACING OF FIE-WIE ESP wire-towire Central-wire 1-wire -wire Ө onset, ma Ө onset, ma Ө onset, ma sacing, cm.5 d 1. d 1.5 d. d.5 d 3. d 3.5 d 4. d k k k TABLE I AIATION OF ANGLE Ө ma AND ONSET OLTAGE OF COONA ON THE DISCHAGE WIES AS INFLUENCED BY THE WIE-TO-WIE SPACING OF SEEN-WIE ESP wire-to-wire Central wire 1-wire -wire 3-wire sacing, cm Ө ma onset, k Ө ma onset, k Ө ma onset, k Ө ma onset, k.5 d 1. d 1.5 d. d.5 d 3. d 3.5 d 4. d TABLE CALCULATED AND MEASUED COONA ONSET OLTAGE FO ONE-WIE ESP Wire radius, mm onset-calc, k Grounding circuit Average Standard ± ± ±. H circuit Average Standard ± ± ±.45 D. Corona Onset oltage Calculation The roosed criterion (7) is alied for different ESP configurations with 1, 3, 5 and 7 discharge wires. The calculated onset voltage values of corona on each wire are given in Table I. It is quite clear that the onset voltage is higher for the central wire and decreases for wires in the direction toward the end of the collecting lates. This is because the central wire is fully shielded by the other wires at stated above with a subsequent decrease of the field at its surface. At the central wire, the onset voltage increases

7 4 International Journal of Plasma Environmental Science & Technology, ol.4, No.1, MACH 1 (a) Contactor switch Auto- Transformer ectifier circuit Water resistance Faraday cage µa H-Transformer C µa Measuring resistance ESP (b) Contactor switch Auto- Transformer ectifier circuit Water resistance C µa Measuring resistance H-Transformer µa ESP Fig. 18. ariation of corona onset voltage with the wire-to-wire sacing of seven-wires ESP. (a) In the high circuit. (b) In the ground circuit. TABLE I CALCULATED AND MEASUED COONA ONSET OLTAGE FO THEE-WIE ESP. Central-wire ight-wire Left-wire r =.5 mm onset-calc, k Average Standard ± ± ±.31 r =.5 mm onset-calc, k Average Standard ± ± ±.36 Central-wire 1-wire L1-wire -wire L-wire TABLE II CALCULATED AND MEASUED COONA ONSET OLTAGE FO FIE-WIE ESP. onset-calc, k r =.5 mm Average Standard 8.66 ± ± ±.9.4 ± ±.4 onset-calc, k r =.5 mm Average Standard ± ± ± ± ±.5 with the increase of the number of discharge wires, Table I as the shielding effect becomes more ronounced with the increase of the number of discharge wires, Table I. The angle of maimum electrostatic field (Ө ma ) around the erihery of discharge wires deends on the ESP configuration. This angle defines where the rimary avalanche grows in the vicinity of the discharge wires. It is mainly influenced by the wire-to-wire sacing. The angle Ө ma is equal o for the central wire, o for the outer wire and has a value between o and o for other wires. With the increase of the wire-to-wire sacing, Ө ma of all wires tends to aroach o. This corresonds to the case where each wire behaves searately with no interaction among the wires as given in Tables II-I and shown in Figs

8 Ziedan et al. 43 The onset voltage values of corona on the discharge wires of ESP deend on the wire-to-wire sacing for the same wire radius and wire-to-late sacing. The onset voltage is higher for corona on the central wire when comared with other wires, Figs This is attributed to the decrease of the field in the vicinity of the central wire due to the above mentioned shielding effect by the other wires. With the increase of the wire-to-wire sacing, the onset voltage decreases because of the less shielding effect and associated decreasing of the field in the vicinity of discharge wires. With further increase of the wire-to-wire sacing, the onset voltage aroaches the same value for all discharge wires where each wire behaves searately with no interaction (shielding) among wires, Figs The difference between the corona onset voltage of the outer wire and that of the central wire decreases with the increase of the wire-to-wire sacing, Tables I-I and Figs However, the difference is so small that the effect of the sace charge from the coronating wire on changing the surface field of the non-coronating wires is negligible with resect to the electrostatic field due to the alied voltage. This is the only simlifying assumtion adoted in the resent study. I. EXPEIMENTAL ALIDATION ESULTS A. Eerimental Setu A wire-duct electrostatic reciitator (ESP) was setu in the H laboratory of Assiut University to comare the calculated and measured corona onset voltage values for different ESP configurations. The set-u consists of: 1) A regulating transformer with - ac inut voltage feeds a variable suly voltage to the highvoltage transformer through a contactor switch. ) H transformer stes the voltage u to the desired value in the range - 1 k for rectification. A half-wave H rectifier circuit comosed of two -ma, 14 k PI diodes and 1 nf, 14 k smoothing caacitor generates a variable dc voltage in the range - 14 k. The dc voltage is alied across a 8- Mresistance in series with a micro-ammeter for measuring the generated dc voltage. This voltage is alied to the investigated duct-wire ESP through a.5 M water-resistance. The water-resistance is to limit the current in case of a flash haens in ESP. 3) The discharge wires of the ESP are coer conductors with diameters of.5,.73 and 1. mm. The wires are hinged vertically between two collecting lates. Each wire is terminated at both ends by two smoothing coer sheres to avoid field concentration. The to ends of the wires are stressed through a smooth coer stri connection. Wire-to-wire sacing is maintained constant at 6.3 cm. 4) The collecting lates forming the duct of the ESP are made of steel and hanged vertically by a wooden suort. The dimensions of each late are 1-m height and 1-m length. Plate-to-late distance is maintained constant at 8.9 cm. 5) A Faraday s cage is used to shield a digital micro-meter for recording corona currents in discharge wires. B. Eerimental Technique To measure the onset voltage from each discharge wire, the shielded micro-ammeter is connected to the discharge wire and the corona current is recorded with the increase of the alied voltage. The onset voltage corresonds to the alied voltage when the microammeter starts to record a reading ust above the zero value. C. Calculated Onset voltages against Those Measured Eerimentally Table gives the calculated and measured values of the corona onset voltage in an ESP with one discharge wire as influenced by the wire radius. The measured values were recorded when the digital micro-ammeter was connected in the ground and high-voltage circuits, Fig. 18. It is quite clear that the calculated values agreed reasonably with those measured eerimentally. The measured values are almost the same irresective of the connection of the micro-ammeter in the ground and highvoltage circuits. The higher the wire radius the higher is the corona onset voltage. This is simly attributed to the decrease of the electric field at the wire surface and its vicinity with the increase of the wire radius. Tables I and II give the calculated and measured values of the corona onset voltage in ESP's with 3 and 5 discharge wires for two different values of the wire radius. Again, the calculated values agreed reasonably with those measured eerimentally. The deendency of the onset voltage on the wire radius is the same as discussed for the ESP with one discharge wire. The onset voltage is higher for corona on the central discharge wire when comared with other wires, Tables I and II. This is attributed to the decrease of the field in the vicinity of the central wire because of the shielding effect imosed by the other wires as stated above in section - D. The onset voltage decreases gradually starting from central wire reaching its minimum value for the outer discharge wires. This is because the above-mentioned shielding effect decreases gradually in the direction away from the central wires with a subsequent increase of the electric field and decrease of the onset voltage as deicted in Tables I and II. II. CONCLUSION (1) The field near the central discharge wire in case of one-discharge wire ESP is higher than that for ESP with multi-wires due to the shielding effect imosed on the central wire due to the other wires. () The calculated electric field along the ESP ais

9 44 International Journal of Plasma Environmental Science & Technology, ol.4, No.1, MACH 1 decreases away from the discharge wire in agreement with revious calculation. (3) In all ESP configurations, the electrostatic field around the outer wire is higher than that around the other (inner) wires. This is because the inside wires are shielded by the outer wires. (4) The angle of maimum electrostatic field (Ө ma ), which defines where the rimary avalanche grows, is influenced by the wire-to-wire sacing. The angle Ө ma is equal o for the central wire, o for the outer wire and has a value between o and o for other wires. (5) With the increase of the wire-to-wire sacing, Ө ma of all wires tends to aroach o. This corresonds to the case where each wire behaves searately with no interaction among the wires. (6) The onset voltage of corona in electrostatic reciitators with 1, 3 and 5 discharge wires is calculated and measured in the laboratory. The calculated values agreed reasonably with those measured eerimentally. (7) The onset voltage increases with the increase of the radius of discharge wires for the same wire-to-wire and collecting late-to-late sacings. (8) The onset voltage increases gradually in the direction away from the central discharge wire reaching its maimum value at the central wire and minimum value at the outer wires. [1] L. B. Loeb, Electrical Coronas: Their Basic Physical Mechanisms. Berkely, CA: University of California Press, 1965, [11] M. Abdel-Salam, Ionization and deionization rocesses in gases, in High oltage Engineering Theory and Practice. M. Abdel- Salam, H. Anis, A. E. Morshedy,. adwan, New York, NY: Marcel Dekker,, [1] M. P. Sarma and W. Janischewsky, "D.C. corona on smooth conductors in air. Steady-state analysis of the ionisation layer," Proceedings of the Institution of Electrical Engineers, vol. 116, , [13] A. A. Elmoursi and G. S. P. Castle, "Modeling of corona characteristics in a wire-duct reciitator using the charge simulation method," IEEE Transactions on Industry Alications, vol. 3,. 95-1, ACKNOWLEDGMENT The authors would like to thank Prof. Mazen Abdel- Salam of Assiut University, Egyt for his interest in this research work. EFEENCES [1] J. M. obinson, "Electrostatic reciitation," in Electrostatics and its Alications. A. D. Moore, New York, NY: Wiley, 1978, [] S. Oglesby and G. B. Nichols, Electrostatic Preciitation. New York, NY: Marcel Dekker, 1978, [3] K. J. McLean, "Electrostatic reciitators," IEE Proceedings A, Science, Measurement and Technology, vol. 135, , [4] S. Sekar and H. Stomberg, "On the rediction of current-voltage characteristics for wire late reciitators," Journal of Electrostatics, vol. 1, , [5] M. Abdel-Salam and D. Wiitanen, "Calculation of corona onset voltage for duct-tye reciitators," IEEE Transactions on Industry Alications, vol. 9,. 74-8, [6] M. Abdel-Salam, "Electric Fields," in High oltage Engineering Theory and Practice. M. Abdel-Salam, H. Anis, A. E. Morshedy,. adwan, New York, NY: Marcel Dekker,, [7] H. Singer, H. Steinbigler, and P. Weiss, "A Charge Simulation Method for the Calculation of High oltage Fields," IEEE Transactions on Power Aaratus and Systems, vol. PAS-93, , [8] K. Honda, "On a Streamer Breakdown Criterion of a Uniform Air Ga" (in Jaanese), The Journal of the Institute of Electrical Engineers of Jaan, vol. 85, , [9] M. Abdel-Salam and K. Stanek, "On the calculation of breakdown voltages for uniform electric fields in comressed air and SF6," IEEE Transactions of Industry Alications, vol. 4, , 1988.