Control Soot Emission Trubulent Diffusion Flame Ac Dc Corona

Control of Soot Emission of a Turbulent Diffusion Flame by DC or AC Corona Discharges HIROMICHI OHISA, ITSURO KIMURA, and HIDEYUKI HORISAWA*

Department of Aerospace Engineering (H. O., I. K.) and Department of Precision Mechanics (H. H.), Tokai University, Hiratsuka, Kanagawa, Japan The effects of DC or AC (14 kHz) corona discharges, formed between tips of opposed needle electrodes, on soot emission of a propane turbulent diffusion flame were investigated experimentally. It is shown that when a DC corona discharge (e.g., 3.6 W; 0.06% of the combustion energy released by the flame) or a discharge system composed of three AC coronas (e.g., 25.5 W in total; 0.43% of the combustion energy) is applied across the lower part of the flame, with a gap width such that the electrode tips are located outside the reaction zone, a marked reduction in soot emission is observed, without noticeable change in the shape of flame luminous region. When corona discharges are applied, increases of the density of charged species and/or charged soot particles are observed in the flame over the whole length downstream of the corona application. It is suggested that, in the case of DC corona application, additional air and inorganic charged species and electrons, produced in the air near the tip of the positive electrode, are carried into the flame mainly by corona winds, whereas in the case of the AC corona application the inorganic charged species and electrons are carried into the flame by diffusion processes. The charged species and electrons carried into the flame may influence the state of charging of incipient soot particles and also reduce the concentration of growing ions, i.e., soot precursors, which directly relate to the soot emission of the flame. In TEM photographs it was found that separate soot particles, or those forming chains in the flame, decrease in mean size with the application of corona discharges. Smaller size soot particles burn faster than larger size particles in the high-temperature oxidizing atmosphere at the flame top region. © 1998 by The Combustion Institute

INTRODUCTION Soot formation in flames is an important practical and fundamental problem that remains inadequately understood despite over a century of valuable research. Weinberg et al. demonstrated that the growth of soot particles in a flame is more or less related to charged species and it can be controlled by means of electric fields [1, 2]. For the formation of incipient soot particles, the mechanism proposed by Calcote starts with chemi-ions [3], while the mechanism based on neutral free radicals is also accepted [4]. It was shown experimentally by Kono et al. [5] that the sooting tendency of diffusion flames changes with the frequency of alternating electric field applied, and Weinberg [6] elucidated the mechanism of soot formation, growth, and oxidation, based on the results of Kono et al. [5]. It was also suggested by Mizutani and Nakahara [7] that the application of alternating electric field is of use for understanding the mechanism of soot emission in diffusion flames. *Corresponding author. Address: Department of Precision Mechanics, School of Engineering, Tokai University, 1117 Kitakaname, Hiratsuka, Kanagawa, 259-1292, Japan. E-mail: [email protected] COMBUSTION AND FLAME 116:653– 661 (1999) © 1998 by The Combustion Institute Published by Elsevier Science Inc.

The objective of the present work is to pursue effective soot control in a turbulent diffusion flame by application of DC or AC coronas in which electric power is negligibly small compared to the flame heat-evolution rate. In the case of DC coronas, the effect of aeration due to corona winds in decreasing soot emissions will be large, although the charged species and electrons produced by the discharges may have an effect. In the case of the AC coronas used (14 kHz), which do not produce noticeable corona winds, the effect of charged species and electrons will be dominant in the change of sooting tendency of the flame. In view of practical objective, in the present experiment, a fully developed turbulent diffusion flame with a relatively large fuel flow rate and a small electrode configuration (opposed needle electrodes) are used. As for the effects of corona discharges on flames, increases of heat transfer and the prevention of blow-off had been reported; these are caused by corona winds involving charged particles [8 –10]. It also has been shown that sooting of a diffusion flame is influenced by electrons emitted from a coated wire placed in the flame [11]. 0010-2180/99/$–see front matter PII S0010-2180(98)00054-6

654

H. OHISA ET AL.

Fig. 1. (a) Application of a DC corona discharge and soot collection. (b) Typical arrangement of DC corona electrodes at h 5 3 cm, and the reference plane for the evaluation of corona winds or equivalent pure air flows (plane figure).

EXPERIMENTAL Flame Description and Application of DC or AC Corona Discharges A fully developed turbulent diffusion flame, formed when propane (3.66 Nl/min) is injected into quiescent air from a 1.5-mm-diameter orifice, was used to evaluate the effect of DC or AC corona discharges on soot suppression. To prevent blow-off, a hydrogen sheath was formed at the burner orifice, flowing hydrogen at 0.95 Nl/min through an annular orifice with a width of 0.25 mm (see Fig. 1a). The soot emission of the flame (determined at 15 cm above the flame top) is 0.041 mg/min. DC or AC (14 kHz) corona discharges formed between opposed needle electrodes (1.0 mm in diameter) were applied to the flame at the height 3 cm or 25 cm from the fuel orifice (Fig. 1a). The outer diameter of the flame reaction zone is 0.7 cm or 4.0 cm at the height 3 cm or 25 cm, respectively. A section of the flame with a typical electrode arrangement for a DC

corona discharge is shown in Fig. 1b. In the case of AC coronas to show a marked soot-suppression effect, a discharge system was devised, composed of three pairs of opposed needle electrodes arranged in a horizontal plane (Fig. 2). In the electrode geometry used, where the electrode tips are in the air outside of the

Fig. 2. Typical arrangement of AC corona discharge system at h 5 3 cm (plane figure).

SOOT EMISSION OF TURBULENT DIFFUSION FLAME reaction zone of the flame, the flow of electric current starts from the tip of one electrode, splits into two paths along the reaction zone of the flame, joins at the other side of the reaction zone and arrives at the tip of the other electrode. The ionization of gas in the discharges will prevail in the vicinity of electrode tips, where electric field strength is large. Measurements of Corona Winds and Method to Generate Equivalent Air Flows The velocity (horizontal component) distribution of a corona wind, when a DC corona discharge is applied to the flame, was determined in the reference plane which is perpendicular to electrodes and locates at the periphery of the induced air flow (see Fig. 1b), using a pitot tube composed of thin quartz L-shaped tubing (inner diameter: 1.0 mm). It was found that the corona wind produced at the positive electrode is much larger than that at the negative electrode. In the case of a 0.5-mA DC corona (gap width between electrodes, 3.0 cm) applied at h 5 3 cm, on the reference plane for the positive electrode, the lateral width of the corona wind was 1.28 cm, the vertical width was 1.35 cm, and the maximum velocity was 2.88 m/sec. In the case of DC coronas, to evaluate independently the effect of charged species involved in corona winds, experiments were carried out in which pure air flows equivalent to the corona winds were applied to the flame. The equivalent air flows with the same flow rates and with velocity profiles similar to those of the corona winds were obtained by selecting the shape of the orifice and the mesh of the grid attached to the orifice from which the air flows out. In the experiment, in which the equivalent air flows are applied at h 5 3 cm, it was noticed that when the maximum velocities are less than 0.5 m/sec at the reference plane facing the flows, no noticeable decrease in soot emission was observed. In the scope of the present experiments it was confirmed that the maximum velocities observed at the reference plane facing the negative electrode in the case of DC corona application or those observed at the reference planes facing both electrodes in the case of AC corona application do not exceed 0.5 m/sec.

655

Evaluation of Cation Concentration in the Flame The variation of cation concentration in the flame during the application of coronas was evaluated using the electrostatic probe method [12, 13]. A double probe composed of two 0.3-mm-diameter Pt wires (gap of two wires: 0.7 mm; exposed length: 1 mm) was swept horizontally across the axis of the flame with a velocity of 2.0 m/sec, keeping the potential difference at 180 volt, and the trace of probe current was recorded by an oscilloscope. For a fixed probe geometry, the probe current is close to being proportional to the cation concentration. Evaluation of Soot Emission and Sooting Tendency of the Flame All the burnt gas from the flame flows through a glass filter placed at 15 cm above the flame top, for a fixed period of time (Fig. 1a). The soot emission of the flame with or without application of coronas is evaluated by measuring the mass of the soot collected on the glass filter. It is confirmed that the mass of soot collected on the filter increases linearly with the samplingtime, in the range of 0 –10 min for the normal flame with a considerable soot emission and 0 –100 min for a low soot emission flame applied with a DC corona discharge. The change of soot particle characteristics along the flame height during the application of coronas was investigated on TEM (transmission electron microscopy) photographs of soot samples, collected on carbon-supported copper grids swept across the flame at a velocity of 2.0 m/sec (thermophoretic sampling method) [14].

EXPERIMENTAL RESULTS Experiment with DC Corona Discharges The effect of DC corona discharges on the soot emission in the propane turbulent diffusion flame is shown in Fig. 3, taking soot emission normalized by that of the unperturbed flame (in

656

Fig. 3. The effect of DC corona discharges applied at lower part of the flame on soot emission.

the absence of corona discharge) as ordinate and corona current as abscissa. In this case, coronas formed between opposed needle electrodes are applied at the height 3 cm from the orifice of the burner tube (blue flame region). It is seen that the soot emission of the flame can be reduced markedly by the application of small-current (low-power) corona discharges, especially in the case of g 5 3.0 cm ( g: gap width between electrodes) and a discharge current .0.5 mA. In the case of the current, the discharge voltage takes the highest value (;7.1 kV) and the corona wind peaks. The electric power of the corona discharge (3.6 W) is very small compared to the heat evolution rate of the flame, 6.0 kW. In the experiment with DC corona discharges applied at h 5 3 cm, it is probable that hydrogen, forming a sheath around the lower part of the flame, is forced into the flame by the corona wind and affects the soot emission. In the present experiments, however, the effectiveness of corona discharge in soot suppression is confirmed for the case of corona application at h 5 25 cm which is far above the sheathed zone. In the present experimental configuration, where a corona discharge of low power is applied locally to a relatively large scale, turbulent diffusion flame, no appreciable change in the outline of flame luminous region is observed, and it is also confirmed that when the electrode surfaces are covered by quartz glass to suppress

H. OHISA ET AL.

Fig. 4. Comparison of the effects of DC coronas and equivalent air flows applied at h 5 3 cm, on soot emission.

electric current so that an electrostatic field is merely applied to the flame, no noticeable soot-suppression effect is observed. As for the causes of reduction of soot emission by DC corona discharges, the following mechanisms may be involved: aeration due to corona winds produced at the positive electrode and introduction of charged species and electrons produced in the vicinity of electrode tips into the flame. In Fig. 4, the data of normalized soot emission shown in Fig. 3 for g 5 3.0 cm and the flow rates of corona winds produced at the positive electrode, evaluated at the reference plane located on the periphery of induced air flow around the flame (see Fig. 1b), are plotted, taking discharge current in logarithmic scale (abscissa). It is seen that the aeration by corona wind plays an important role in the suppression of soot. In the figure, the values of normalized soot emission in the case of the application of equivalent pure air flows are also shown. The two soot-emission curves shown in the figure, for DC coronas and for equivalent pure air flows, indicate a marked difference in the low current region (,0.02 mA), suggesting that for the reduction of soot emission some electrical mechanism must be working, although it is masked by aeration effect in the region of discharge current larger than 0.02 mA. It is also seen in the figure that an electrical mechanism works to increase soot emission in the region of current larger than

SOOT EMISSION OF TURBULENT DIFFUSION FLAME

Fig. 5. The effects of AC corona discharges applied at lower part of the flame by the AC corona discharge system (Fig. 2), on soot emission.

5.0 mA, where the characteristics of discharges deviate from those of coronas (abrupt decrease in discharge voltage). Experiment with AC Corona Discharges To make clearer the existence of an electrical mechanism for soot-emission reduction by corona discharges, experiments with AC coronas (14 kHz) were performed in which the aeration by corona winds can be ignored. Although it was confirmed that a single AC corona discharge formed between opposed needle electrodes, as in the case of a DC corona, is effective for soot suppression, the results obtained using the discharge system with three identical AC coronas (see Fig. 2) are shown below. The effect of AC coronas on soot emission, in the same propane turbulent diffusion flame as that used in the case of DC coronas, is shown in Fig. 5, taking normalized soot emission in ordinate and corona current per electrode pair in abscissa. In this case the AC corona discharge system is applied at the height 3 cm. It is seen that in most of curves of Fig. 5, there is a limit in the magnitude of current where the discharge mode transfers from corona to arc and an abrupt decrease in discharge voltage appears. In the discharge system composed of three AC coronas, the limiting value of current was lowered to some extent compared to the case of the application of single AC corona.

657

It is seen in Fig. 5 that soot emission can be reduced also by the application of the AC corona discharge system. It is confirmed that the growth of corona winds, of which an aeration effect causes noticeable soot-emission reduction, is not observed in this experiment. In Fig. 5 it is noticed that the effect of soot-emission reduction is large when the gap width is small for a fixed discharge current, except the case of g 5 1.2 cm. This result suggests that the charged species and electrons produced mainly at the vicinity of electrode tips diffuse into the flame at the upper part of the corona application and the quantity of them introduced into the flame is large when the distance between the tips and the reaction zone is small. The discharge voltage of a AC corona composing the discharge system with g 5 2.4 cm is 5.0 kV at discharge current 1.7 mA and hence the total discharge power of the system is 25.5 W. The AC corona discharge system applied at h 5 25 cm was also effective for the suppression of soot emission, although the effectiveness was lower than that applied at h 5 3 cm. Variation of Cation Concentration in the Flame by the Application of Corona Discharges The traces of cation concentration obtained by a double probe swept across the flame at the heights, h 5 8 cm, 25 cm, and 40 cm are shown in Fig. 6 versus the radial position in the flame. Each figure involves five traces of cation concentration. Figure 6a is the case without discharge, Fig. 6b is the case with a DC corona of 0.5 mA applied at h 5 3 cm, and Fig. 6c is the case with 3 AC coronas of 1.7 mA per electrode pair applied at h 5 3 cm. At a height of h 5 8 cm in case Fig. 6a, five cation-concentration traces take nearly an identical shape and two peaks are observed clearly corresponding to the two reaction zones traversed by the probe, although in other figures the shape of the trace changes in each sweep due to the effect of aerodynamic and electrostatic turbulence. It is seen that the application of corona discharges at h 5 3 cm increases cation concentration about 500 times (DC corona) or 50 times (AC coronas) at h 5 8 cm and 25 cm, though the times of

658

Fig. 6. Cation concentration, (a) without corona discharge, (b) with a DC corona discharge (0.5 mA, 7.1 kV, g 5 3.0 cm) at h 5 3 cm, the sweep of probe is perpendicular to electrodes, (c) with the AC corona discharge system (1.7 mA per electrode pair, 5.0 kV, g 5 2.4 cm) at h 5 3 cm.

concentration increase are decreased at h 5 40 cm to some extent. In Figs. 4 and 5, curves of the maximum value of cation concentration observed at the flame sections at h 5 8 cm and 25 cm are added. DISCUSSION Mechanism of Soot Suppression by Corona Discharges The experimental results shown in Fig. 4, in which the effect of DC coronas is compared with that of the equivalent aeration, and the experimental results with AC coronas shown in Fig. 5, which are free from aeration effect, suggest that charged species and electrons produced by corona discharges work to suppress soot emission. It is probable in the present experimental situation that the charged species are produced mainly in the vicinity of electrode tips where electric field strength is large, and are 2 inorganic (e.g. N1, N1 2 , O2 , . . . ), because the tips of electrodes remain in the air around the flame. The charged species and electrons are

H. OHISA ET AL. carried into the flame mainly by corona winds in the case of DC coronas or by laminar and turbulent diffusions in the case of AC coronas, and they can influence the sooting tendency of the flame, as follows. The inorganic anions and free electrons carried into the pyrolysis region of a flame may reduce the concentration of growing ions, which are assumed as soot precursors in the ionic mechanism of soot formation [3]. Furthermore, the inorganic cations carried into the pyrolysis region may increase the percentage of charged incipient soot particles by charge transfer, and hence reduce their agglomeration rate. Smaller size soot particles burn faster than larger size particles in the hightemperature oxidizing atmosphere. For the case of DC corona application (Fig. 4), the marked soot decrease is observed for discharge currents of the order of 1022 mA, and for the case of AC corona application (Fig. 5), it is observed in much larger discharge currents (;1 mA). It is noticed, however, that the orders of cation concentration in the flame where the marked soot decrease appears are the same for the cases of DC and AC corona applications (;10 [AU]). For the soot reduction in the present experiments, two other mechanisms may play some role. One is the effect of electric fields due to space charge, suggested by one of the reviewers [15]; such electric fields may exist over the whole length of the flame, decreasing the residence time of charged soot particles in the pyrolysis zone and hence suppressing their growth. The other is the contribution of ozone generated by corona discharges, suggested by the above reviewer and Prof. T. Sano [15, 16]; ozone is effective in oxidizing soot particles in the flame. The corona discharges also generate radicals such as appear in combustion as intermediates, however, their influence on the sooting tendency of the flame will be small, since their quantity is small compared to that generated in the flame by chemical reactions. Transmission electron micrographs (TEM) of soot samples collected at h 5 8 cm, 18 cm, and 30 cm of the flame are shown in Figs. 7, 8, and 9, for the cases without corona application, with DC corona application at h 5 3 cm, and with AC corona application at h 5 3 cm. In all the figures it is seen that the soot takes granular

SOOT EMISSION OF TURBULENT DIFFUSION FLAME

659

Fig. 7. TEM photographs of soot samples, without corona discharge, (a) collected at h 5 8 cm, number of sweeps: 2400, (b) collected at h 5 18 cm, number of sweeps: 36, (c) collected at h 5 30 cm, number of sweeps: 36.

forms at h 5 8 cm and the joining of soot particles into chain form proceeds at h 5 18 cm and 30 cm. It is noticed also that in Fig. 8 with a DC corona accompanied by a corona wind, a marked decrease of soot, compared to the case of Fig. 7, is observed at h 5 8 cm, while in Fig.

9 with AC coronas, where aeration and introduction of charged species and electrons into the flame proceed only by diffusion processes, marked decreases of soot, compared to the case of Fig. 7, are not observed at h 5 8 cm and 18 cm. Furthermore, it was noticed generally that

Fig. 8. TEM photographs of soot samples, with application of a DC corona (0.5 mA, 7.1 kV, g 5 3.0 cm) at h 5 3 cm, (a) collected at h 5 8 cm, number of sweeps: 2400, (b) collected at h 5 18 cm, number of sweeps: 36, (c) collected at h 5 30 cm, number of sweeps: 36.

660

H. OHISA ET AL.

Fig. 9. TEM photographs of soot samples, with application of the AC corona discharge system (1.7 mA per electrode pair, 5.0 kV, g 5 2.4 cm) at h 5 3 cm, (a) collected at h 5 8 cm, number of sweeps: 2400, (b) collected at h 5 18 cm, number of sweeps: 36, (c) collected at h 5 30 cm, number of sweeps: 36.

soot particles in the samples at h 5 8 cm, or soot particles which form chains in the samples at h 5 18 cm and 30 cm decrease in mean size with the application of coronas. An example is

shown in Fig. 10. Smaller size soot particles burn faster than larger size particles in oxidizing atmosphere, e.g., the high-temperature region at the flame top.

Fig. 10. TEM photographs of soot samples, collected at h 5 30 cm, (a) without corona discharge, (b) with application of a DC corona (0.5 mA, 7.1 kV, g 5 3.0 cm) at h 5 3 cm, (c) with application of the AC corona discharge system (1.7 mA per electrode pair, 5.0 kV, g 5 2.4 cm) at h 5 3 cm.

SOOT EMISSION OF TURBULENT DIFFUSION FLAME Practical Use of the Present Soot Suppression Method The smallness of the electrode configuration compared to the size of flames and the low electric power required suggest the possibility of the present soot suppression method for practical use. For the best soot-emission reduction (about 1/50) in Fig. 3, the required electric power is 3.6 W, and for that (about 1/20) in Fig. 5, 25.5 W, which are very small compared to the heat evolution rate of the flame, ;6.0 kW. CONCLUDING REMARKS The effects of DC or AC (14 kHz) corona discharges, formed between tips of opposed needle electrodes, on soot emission of a propane turbulent diffusion flame were investigated experimentally. The results are as follows: 1. When a DC corona discharge (e.g., 3.6 W) or a discharge system composed of three AC coronas (e.g., 25.5 W in total) were applied across a lower part of the flame, with a gap width such that the electrode tips are located in the air outside the reaction zone, a marked reduction in soot emission was observed, without noticeable change in the shape of flame luminous zone. 2. It was confirmed by an electrostatic probe that when a corona discharge is applied, the density of charged species and/or charged soot particles in the flame is increased over the whole length downstream of the corona application. 3. In TEM photographs, it was found that the soot particles in the samples collected at a lower part of the flame, or the soot particles that form chains in the samples collected at middle and upper parts of the flame decrease in mean size with the application of corona discharges. 4. In the case of DC coronas, additional air and inorganic charged species and electrons, produced in the air at the vicinity of positive electrode tip are carried into the flame mainly by corona winds, and those produced at the vicinity of the tip of negative electrode,

661

by diffusion process. In the case of AC coronas without noticeable corona wind, the inorganic charged species and electrons are carried into the flame only by diffusion processes. Those carried into the flame may influence the state of charging of incipient soot particles and also reduce the concentration of growing ions, i.e., soot precursors, which relate to the soot emission of the flame. The smallness of the electrode configuration compared to the size of flames and the low electric power required suggest the possibility of the present soot suppression method for practical use.

REFERENCES 1. 2. 3. 4. 5.

6. 7. 8. 9. 10.

11. 12. 13.

14. 15. 16.

Place, E. R., and Weinberg, F. J., Proc. Roy. Soc. (London) A289:192–205 (1965). Mayo, P. J., and Weinberg, F. J., Proc. Roy. Soc. (London) A319:351–371 (1970). Calcote, H. F., and Keil, D. G., Pure Appl. Chem. 62:815 (1990). Frenklach, M., and Warnatz, J., Combust. Sci. Technol. 51:265 (1987). Kono, M., Iinuma, K., and Kumagai, S., Eighteenth Symposium (International) on Combustion, The Combustion Institute, Pittsburgh, 1981, pp. 1167–1174. Weinberg, F. J., NATO Conference Series VI: 7, Plenum, New York, 1981, p. 243. Mizutani, Y., and Nakahara, M., Trans. Jpn. Soc. Mech. Eng. B49:1530 –1538 (1983). Bradley, D., and Gupta, M. L., Combust. Flame 40: 47– 63 (1981). Bradley, D., and Nasser, S. H., Combust. Flame 55: 53–58 (1984). Calcote, H. F. and Berman, C. H., Am. Soc. Mech. Engr., Fossil Fuels Combustion Symposium PD-Vol. 25, 25–31, S. N. Singh Book No. H00453, 1989. Bowser, R. J., and Weinberg, F. J., Nature 249:339 –341 (1974). Clements, R. M., and Smy, P. R., Brit. J. Phys. (J. Phys. D) 2:1731–1737 (1969). Calcote, H. F., Ninth Symposium (International) on Combustion, Academic Press, New York, 1963, pp. 622– 637. Dobbins, R. A., and Megaridis, C. M., Langmuir 3:254 –259 (1987). One of the reviewers, Private communication. Sano, T., Tokai University, Private communication.

Received 4 February 1997; accepted 10 April 1998