Plasma Chemistry and Plasma Processing, Vol. 15, No. 2, 1995
Study of the Dynamic and Static Behavior of dc Vortex Plasma Torches: Part I: Button Type Cathode J.-F. Brilhac, ! B. Pateyron, I G. Delluc, I J.-F. C o u d e r t , ~ a n d P. F a u c h a i s 1 Received January 19, 1994; revised July 21, 1994
This work was devoted to the study of the dynamic and static behavior of a dc vortex plasma torch with a button type cathode (power level below 60 k W ) and a tubular anode. The influence o f the evolution o f arc current ~ gas flow rate G, and nature, and electrode polarities on the dynamic and static behaviors was investigated. To characterize the dynamic behavior o f the torch, a set-up was developed to measure the fluctuations o f arc voltage U and current, plasma jet radiation, and acoustic pressure..4 frequency peak (around 4000 Hz) was found #1 each signal and was characteristic o f the arc column-anode wall disruptions. The frequency change of this peak vs. I, anode internal diameter d, and G was studied and its variation represented by semiempirical relationships. To characterize the static behavior o f the torch, the variations of U and thermal efficiency 17 with I, G, gas nature, and anode design were studied and also represented by semiempirical relationships.
KEY WORDS: Vortex plasma torch; arc characteristic; dimensionless analysis; electrical, acoustical, and optical measurements; signal treatment.
NOMENCLATURE d, anode diameter (m) D, vortex chamber diameter (m) f, frequency (Hz) G, mass flow rate of plasma gas (kg/s) G', volume flow rate of plasma gas (slm) ho, mass enthalpy (J/kg) Hp, mass enthalpy of plasma gas (MJ/kg) I, arc current intensity (A) L, anode length (m) ~Equipe "Piasmas-Lasers-Mat~riaux," Laboratoire "Mat~riaux C~ramiques et Traitements de Surfaces," URA320 CNRS, 123, Av. A. Thomas, 87060 Limoges C~dex, France. 231 0272.4324/95/0600-0231507.50/0 ,~ 1995PlenumPublishingCorporation
232
Brtlhac et al.
m, mass (kg) p, gas pressure (Pa) P, electrical power (W) Pa, Pr, power losses in the cooling circuits of anode and cathode (W) Pc, total power losses (W) S = ( 1/3)/[( 1 - z3)/( 1 - z2) t'5] tan(0), swirl number t, time (s) T, temperature (K) To, reference temperature corresponding to 1% electron concentration in the plasma gas (K) tan(0) = Wom/Uom Uom, axial velocity in the nozzle (m/s) Worn, azimuthal velocity in the gas injection chamber (m/s)
z=d/D cro, electrical conductivity (mho/m) Po, viscosity (kg/m s) 17, thermal efficiency of the torch ~:0, mean thermal conductivity (W/m K) r, total delay of a record (s) re, delay between two recorded points (s) Af, frequency resolution (Hz) Ah, mean gas enthaipy variation (J/kg) At, time lag (s)
Dimensionless groups
Pr= (Po' ho)/(tCo" To), Prandtl number Re= G/pod, Reynolds number Sf=fd2~o/#o/l, frequency criteria Si = 12/Gdcroho,energy criteria Su = Udcro/l, electric field intensity criteria
1. INTRODUCTION Thermal plasmas are now widely used in various industrial processes: cutting, welding, spraying, heating, melting, remelting and purification of metals, extractive metallurgy, blast furnaces, gas synthesis, ultrafine powders production, waste destruction, etc. This is due to their unique properties: high power densities (up to 10SW/m3), high temperatures (3000 to 15,000 K), very low thermal inertia, tailored atmosphere, and high thermal conductivity.
Behavior of dc Vortex Plasma Torches
233
Considering the number of plasma devices used in the world, most of them are based on dc arcs, either transferred or blown. Actually, many de vortex plasma torches are used in industrial processes~: ladle heating (with Plasma Energy Corporation (PEC) torches), blast furnaces air temperature boosting (with Aerospatiale (AS) torches at SFPO and Uckange in France), cuppola heating (with AS torches at Peugeot in France, or Westinghouse torches at G.M. in the USA), smelting and reduction (SKF torches with their Swedchrome process), metallurgical wastes recycling (SKF torches with their Plasmadust process), toxic waste destruction (with Retech torches for Plasmox process in Switzerland, AS torch with Rh6ne-Poulenc in France), TiO2 pigments production (with Tioxide torches in Great Britain), C2H2 synthesis (with Hiils torches in Germany), etc. The power levels range from 0.5 to 7 MW. One of the very critical points for the industrial use of such torches, is the lifetime of the electrodes, which in certain applications now reaches up to 1000 h. Such results have been obtained by empirical design of the electrodes based on previous experience. However, any modification of the electrode design of such high-power plasma torches is rather expensive, and studies have been performed in the 1960's and 1970's, especially in Russia, to see how the design of small torches (below 100 kW) could be extrapolated to torches working at a few M W . t2-6) These studies were essentially devoted to the static behavior of the torches (voltage-current characteristic, thermal efficiency vs. arc current etc.). Yas'ko t2'3~ and Zhukovt4'5) were the first to introduce dimensionless analysis of the static behavior of their torches, such an analysis allowing one to check if the results obtained with low power levels (<100 kW) could be extrapolated to highpower-level plasma torches ( P > I MW). Though at that period the arc fluctuations were described, ~4"7~ no quantitative measurements of these fluctuations were performed. The reasons were probably that, on the one hand, there was a lack of powerful tools for signal treatments (via fast Fourier transform, for example), and on the other hand synchronized digital multichannel oscilloscopes were not available. However, it seems that these fluctuations are closely linked to the arc stability and electrode erosion and if numerous models have been developed for the arc. column or the extinguishing plasma jet, ta-t3~ very few of them are related to the arc behavior at the electrodes. The arc attachment at the surface of a hot cathode t~4-j~ seems now rather well understood. But, except for the phenomena at the anode of transferred arcs, t~6~9~ there is still a complete lack of reliable models accounting for the attachment of the arc at the surface of the anode or cold well type cathode of a dc plasma torch, and experiments to understand better the arc fluctuations, linked to the restrike mode at the electrode surface, are necessary.
234
Bfilhac etai.
This work is devoted to the study of the dynamic behavior of dc vortex plasma torches with power levels below 60 kW. Such power levels allow rather cheap manufacturing of different electrode designs and thus a systematic study of the effect of this design on the torch working conditions. A dimensionless analysis of our torches was also performed in order to see if their static behavior was comparable to that of high-power plasma torches such as those of A6rospatiale (AS) and Plasma Energy Corporation (PEC). These two torches were chosen in this investigation, because arc characteristics corresponding to their behavior were available, t2°) For simplicity, the study was started with a dc vortex plasma torch with a hot button type cathode and a cylindrical anode nozzle. In this case only the fluctuations of the arc root at the anode surface were to be taken into account, and this is the subject of Part 1 of the present paper. The second part is devoted to the same study but with the torch being equipped with a cold well type cathode. We will present successively in this first part a short review of the physical processes controlling the arc behavior and the dimensionless analysis,c2-4~the experimental devices used, and the data processing, the dynamic behavior of the torch, and finally its static behavior.
2. STATE OF THE ART IN THE VORTEX PLASMA TORCH DESIGN AND CHARACTERIZATION The first part of this study is focused on the behavior of a dc plasma torch with a hot button type cathode and a vortex injection of plasmaforming gas. Among the three types of anodes which can be coupled with this cathode, C4"6)an anode with a constant diameter has been chosen, because it represents well the type of torches used in industrial conditions. It corresponds to a self-stabilizing arc length resulting in arc characteristic (arc voltage evolution vs. arc current for different gas flow rates and nature and different torch designs) always decreasing. The influence of the vortex flow on the arc attachment is not yet clearly understood. Zhukov t4~ assumed that, while the rotational momentum of the gas in the anode is weakly affected by the plasma jet, the axial one increases strongly due to the high velocity of the plasma column, and the vortex efficiency diminishes rapidly. These results were confirmed by the calculations of Chyou and Pfender. ~2~) Theoretical and experimental investigationst4"7~have allowed us to make a simple model of the interactions between the arc and its surrounding gas in the anode channel and the resulting attachment mode of the arc column at the anode wall. As shown in Fig. 1, the arc terminus is pushed downstream by drag and Lorentz forces close to the arc column (however, the Lorentz
Behavior of dc Vortex Plasma Torches
235
Fig. 1. Scheme of arc shuntings in a dc plasma torch.
force at the anode wall acts in the opposite direction) with three characteristic arcings~4'6~: --large scale arc-to-wall shunting (1), --small scale arc-to-wall shunting (2), --arc-to-arc shunting (3). For the self-stabilized arcs, the first type of arcing (1) defines the mean length of the arc column and the erosion area and rate. As the arc terminus is forced downstream by the drag force increasing the arc length and voltage, a situation is reached where the potential difference between the arc column and anode wall allows shunting, cl?eating a new anode spot upstream of the previous one (see (1) in Fig. 1). As observed by Zhukov, c4"6)the arc column close to the arc terminus is not as stable as that observed closer to the cathode (see Fig. 1). These small instabilities or "turbulences," which can reduce the distance between the wall and the arc column, seem to favor the arc shunting. These results were confirmed by Wutzke et al. (7) This mechanism of disruption generates arc voltage fluctuations in a saw tooth shape (4"z2"z3~at a frequency between 1 and 15 kHz. From Zhukov, C4°6~the second type of shunting (2) corresponds to the radial displacement of the arc terminus. Recent works of J.-F. Coudert et al. in the laboratory (24~have shown that this type of shunting downstream of the previous anode spot is less stable than type (1) shunting because of the higher voltage between the arc column and the wail, leading to a higher probability for a new breakdown to occur coming back to an "upstream" case (type 1). Compared to type (1), this type (2) shunting corresponds to higher values of the minima of the voltage undulation. The last one (3), generating low-magnitude undulations of arc voltage (arc length change being small) at frequencies higher than 10 kHz, c4"6~increases the erosion rate of the anode (the arc spot being fixed at the anode wall as long as this type of shunting occurs). However, it has not been observed in our voltage signal. These studies showed clearly that the static characteristics of the arc are closely linked to its dynamic behavior, depending on the way the cold
236
Brilhac et aL
gas is injected. For example, the work of Coudert et aL (23) with an axial gas injection, a stick type cathode, and a constant-diameter anode (self-stabilizing arc length) has shown that mainly one typical frequency corresponding to type (1) and (2) arcing could be observed for a given arc current, gas flow rate, and nature, and nozzle diameter. The mean lifetime of the arc root (or the inverse of the voltage fluctuation frequency) as well as the mean value of the voltage jumps are very strongly connected to the arc current and nozzle diameter (for a given plasma gas nature) and not very sensitive to the gas flow (at least for those studied). In our case with arc currents below 350 A, the electrical radius of the arc calculated with the simple equation of Eienbaas-Heller is below 5 mm, and the arc constriction by the nozzle (7 < d mm < I0) is not very important, which explains the low sensitivity of the voltage jumps with gas flow rate. Such results were confirmed by recent measurements~21~on voltage jumps amplitude. On the contrary, with a vortex injection t25) three characteristic frequencies were observed: one attributed to the arc root movement due to the vortex at a frequency lower than that corresponding to axial displacement and a third one resulting from the coupling of both movements. These voltage fluctuations and those of the arc current (mainly the power source ripple) cause the fluctuations of the electrical power dissipated in the torch, resulting in fluctuations of light emission of the plasma jet at the same characteristic frequenciesC26'2s) and acoustical pressure ~22'27~linked to the time derivative of the power fluctuations. The simultaneous study of the four signals may yield interesting information on the arc behavior. According to the manufacturing cost of any change in the design of the electrodes of high-power torches, considerable attention was focused on the search for similarity criteria t2"4) in order to extend the electrical and thermal characteristics measured with low power level plasma torches to those of high power. These criteria were obtained by the reduction in dimensionless form of the equations describing the physical phenomena occurring in the discharge chamber: - - f r o m Ohm's law, an electric field intensity criterion Su is obtained which is characteristic of the arc resistance, - - f r o m the energy equation, the criterion Si is obtained, which represents the ratio between enthalpy flow and Joule's energy instant dissipation. - - f r o m the momentum equation, the Reynolds number Re characteristic of the gas flow is obtained:
Ud~o Su----,
I
12 Si-
-
-
Gdo~o'
G Re =--
12od
Behavior of dc Vortex Plasma Torches
237
Table !. Constants for the DifferentPlasma Gases
Nature of gas To (K) Oo (A2 s3/kg m3) /~o (kg/m s) I¢o(J/m K) ho (J/kg)
Nitrogen 8600 ! 200 0.00022 2 45,800,000
Argon 9400 2350 0.000261 0.487 5,200,000
where or0, ho, Po are, respectively, the electrical conductivity, enthaipy, and viscosity of the plasma gas at a mean temperature To corresponding generally to an electron concentration of 1% in the plasma (see the corresponding values in Table I), d is the diameter of the nozzle (anode), I the arc current (A), U the arc voltage (V), and G the mass flow rate of plasma gas (kg/s). Other dimensionless numbers such as the Mach number M, Knudsen number Kn = 1/p • d, where p is the pressure at the outflow of the torch, are also used, and when an external magnetic field is applied, a criterion related to magnetic interaction must also be defined. According to Z h u k o v , ~4"6) for a torch with a hot button type cathode and an anode of constant internal diameter (i.d.) without external magnetic field, the dimensionless number Su is related only to three criteria: Si, Re, Kn. The generalization of the variations of the experimental data can be achieved with a unique relationship S u = K - S i n. Re b. Kn c. In previous studies, t2-5'2°'29) this type of relationship was established to characterize the behavior of de and ac vortex plasmas torches, and they allowed one to recalculate the arc characteristics with a relatively small error (<5%).
3. E X P E R I M E N T A L S E T - U P 3.1. Design of the Torch
The torch used was developed in the laboratory and is shown in Fig. 2. The plasma-forming gas injected in the vortex chamber, with an internal diameter D of 45 ram, was either nitrogen or argon, but the presented results are related to nitrogen. The button type cathode was made of thoriated tungsten (W + 2wt.%Th, ~b= 7 mm) fitted in a copper watercooled holder which was t r u n c a t e d - c o n e - s h a p e d . A tubular anode was coupled with this cathode, and its diameter was constant (CD) ( d = 8 or 9 ram) for a length of 100 mm. The swirl number S was calculated only for the cold air flow rate in this torch with the following relationship, already used by Pateyron ¢29) to
238
B~lhac etal.
HotButtontypec ~
~
Anode(constantdiameter) i
Copperbuttonhol~e~as ir~ectlon¢harnb!r insulaUngmaterial Fig. 2. Schemeof the button-typecathode torch with two differentshaped anodes.
calculate the number S of the Plasma Energy Corporation torch (PEC) :
S=l_. (l-z 3)
3 ( l - z 2 ) I's" tan(0)
For this torch S=4.2, and this value is slightly higher than that calculated for the Aerospatiale torch ( S = 3.1) and lower than that calculated for the PEC torch ( S = 13). These relatively high values of S correspond to strong vortices. In order to use an automatic electrical starting of this torch instead of an auxiliary carbon electrode, some designing modifications were necessary. The interelectrode spacing was reduced, and both cathode and anode walls in the gas injection chamber were flattened (this cathode is calledflat-shaped below). Only one anode was used with this new design, and its constant diameter was equal to d = 7 mm. The torch power supply was a dc source (diode rectifier with an hexaphasis G R A E T Z bridge). This electrical supply, designed to work at power levels between 200 and 900 kW, exhibits unfortunately a rather high ripple at power levels below. 100 kW. According to the hexaphasis bridge, the frequency of this ripple is 300 Hz and its harmonics.
3.2. Characterization of the Dynamic Behavior
3.2.1. Experimental Set-up A scheme of the experimental set-up is shown in Fig. 3. Four fluctuation measurements were made: voltage, arc current, radiation, and acoustic pressure. The light measurement was performed with an optical bench collecting the radiation emitted at a given point of the plasma jet outside the nozzle. The image of the plasma jet was focused by a lens on the tip of an optical fiber (600/am in diameter). After the final test no filter or monochromator
Behavior of de Vortex Plasma Torches
239
Measurements of arc voltage and arc current
Vortex
torch
Opticalfiber
Photodiode Honeywell Monochromator HIOVIS
I
Measurements of plasma radiation
Mic~Fh~n~e
Measurements of acoustic pressure
Fig. 3. Set-upfor the characterizationof the dynamicbehaviorof the plasma torch.
was used to select a given wavelength range which was defined by the band pass of the lenses, optical fiber, and photodiode: roughly from 400 to 1200 nm. The detection was made by a photodiode (Honeywell Photo, Darlington, SD 5410-2). Frequencies up to 500 kHz could be recorded with this set-up. The acoustic pressure fluctuations were measured with a zI t t microphone cartridge (Bruel and Kjaer Type 4135). This microphone has a very large bandwidth (from 4 to 100 kHz). For the arc voltage and arc current measurements, two Hall effect probes were used (ABB Petercem EM010 for the voltage and Petercem TC015 for the current). For these probes, the maximum frequency recorded was up to 500 kHz. All these measurements were recorded with a digitizing oscilloscope (LeCroy 9314M) and were stored and analyzed with the help of a computer (HP 9000-217).
3.2.2. Data Processing The signals recorded were digitized with the four synchronized input channels of the oscilloscope. To achieve good precision of the measurements, 50,000 points were recorded on each oscilloscope channel, and then 10,000 of them were transferred to the computer for numerical treatment. In order to limit the number of recorded data and according to the signal frequencies (<10kHz), the sampling frequency was chosen as 100,000 samples per second, and for this frequency the delay between two recorded points was re = l0 #s. The total delay of a record was thus r=0.1 s. To shift from time domain to frequency domain, a fast Fourier transform (FFT) algorithm was
240
Brilhac et
al.
used. The FFT was calculated with waveforms at 2 ~°= 1024 points windowed by a Hamming function before processing. The corresponding frequency spectra contained 512 elementary frequencies. If all the 1024 points were taken successively, the observation delay was 1024 x 10/~s= 10.24 ms. The maximum frequency was 50 kHz, and the frequency resolution was Af= 97.6 Hz. To decrease the value of Af, the observation time must be increased, and for that a waveform was built by taking 1024 points among the 10,000 ones which are stored in the computer memory, the corresponding time step being ire, with i= 1 to 9. With the help of this method, the sampling frequency could be changed during the data processing, and was adjustable to the studied signals. 3.3. Characterization of the Static Behavior
Arc current I and arc voltage U were measured with the help of probes and averaged on five values with one second between two successive records. Gas flow rates G' were recorded with the help of rotameters, and the powers dissipated in the cooling circuits were calculated from the water flow rates and the temperatures recorded by thermocouples disposed at the input and output of the separated cathode and anode cooling circuits. These results allow us to calculate the mean gas enthalpy variation Ah and the thermal efficiency of the torch 17: Ah = U. I - P , / G and 17= 1 -Pc~U" I where Pe = P x + Po, P,, and Px being, respectively, the power losses in the cooling circuits of the anode and cathode. The incoming gases (N2, H2, 02, Ar) are stable species in the standard conditions (T= 298 K, p = 105 Pa), and by definition their enthalpy is zero. The plasma gas enthalpy at the torch exit is evaluated in comparison with the incoming gas enthalpy and corresponds to the gas enthalpy increase in the torch. 4. DYNAMIC BEHAVIOR 4.1. Characterization of the Gas Flow inside the Torch without Electric Arc
Without the electric arc, the measurements of the acoustical pressure was also performed on a sketch of the torch to characterize the cold gas flow inside the torch. Figure 4 shows the acoustical frequency spectrum obtained for an air flow rate G'=53 slm and with a nozzle of 10 mm i.d. Almost identical results were obtained with nitrogen. Characteristic frequencies appear: one of 1566 Hz and its first harmonic (3170 Hz). As shown in Fig. 5, this frequency increases with the rise of gas flow rate and decreases when the nozzle i.d. increases. This frequency peak, already found in previous studies, t3°'3j) seems to be characteristic of the vortex flow inside the
Behavior of dc Vortex Plasma Torches
241
Magnitude(a.u.) 117
1566
0
2000
4000 Frequency(Hz)
6000
Fig. 4. Acoustic frequency spectrum (arbitrary units) for d = 1 0 m m , L = 1 0 0 m m , and (7' = 53 sire.
outflow nozzle. A low frequency (117 Hz) is also isolated and remains constant with any change of flow rate and nozzle design. It is probably generated by the venting system.
4.2. Cone-Shaped Cathode To characterize the dynamic behavior of the torch, the arc voltage and arc current fluctuations were systematically measured for different values of /, G', and d. An example of the recorded signals is shown in Fig. 6. Strong fluctuations can be observed on each one, and two characteristic undulations appear: a low-frequency fluctuation (around 300 Hz), which is very strong on arc current wave, and another one at higher frequency (around 5000 Hz). This second one appears on each signal. The arc voltage wave has a sawtooth
f (KHz) 5
4' 3
~P ¢o o ~
o L=10Ommd=14mm D=45mm L=100mm d=10mm D=45mm 50
100
150
I
2°°G' ( s l m )
Fig. 5. Variation of the characteristic frequency of the vortex flow without electric arc G' and d.
242
Brilhac et al.
'
Voltage
'~
V..V
V
Ir"~ "~ ~"~VV ~v'ry,, ,,~tvwV,,", .
.
.
.
.
.
.
.
.
.
.
.
.
U=OV
c.,.~.t
t
/~la
~
.......j l
... \ .......... / . . . . . . .
",.. . . . . .
I = 3. la
I=OA t = 10 ms
Fig. 6. Shape of the voltage and corrent fluctuations for the anode with a constant diameter (I=256 A, G ' = 5 0 sire, d = S m m , U=61 V).
shape, and its magnitude is very large (equal to the mean arc voltage). When reducing the ripple amplitude by using a resistance in the torch circuit, no variation of the voltage frequency was observed. In Fig. 7, the corresponding frequency spectra obtained are shown. Harmonics of 300 Hz appear and remain constant with a change of the static parameters. They are characteristic of the power supply (rectifier with hexaphasis bridge), which generates, as emphasized in Section 3.1, an undulation at a frequency equal to 6 x 50 Hz = 300 Hz. A sharp peak at 4848 Hz is observed on each spectra, and its frequency changes with arc current and anode design. This frequency seems characteristic of the arc root displacement at the anode. VOLTAGE SIGNAL
I
CURRF..~~ N N .
it ill
-
"|
-
Fig. 7. Corresponding frequency spectra of the signals shown in Fig, 6.
Behavior of de Vortex Plasma Torches
243
f (Hz) 5200 5100 5000 4900
oG=55 51~
o~,~
= G=55 sireI
+ G=50 slrr[ 0=8 m m y ~ "
4800
o G=50 slm10=9 mrn
~G=40~sl~ ~
4700
&
•
~,~
Aa=40 sl~
A
46OO -4500 200
calculated frequencies ~ 250
= 300
, 350
I(A)
Fig. 8. Arcing frequency variation with I and d for the torch with cone-shaped cathode.
Such a behavior was observed in previous studies ~4'6'7'22'25~ and seemed characteristic of the arcings between the arc column and the anode wall. The following frequency ranges were found depending on the gas nature, arc current, and anode diameter: from 1 to 1 0 k H z , (4"6) from 500 to 2500 Hz, t7) around 5 kHz, ~22) and from 5 to 15 kHz. ~23) In fact, as explained in Section 2, under the action of the drag and electromagnetic forces acting on it, the arc terminus is pushed downstream as long as arcing does not occur at a shorter or longer distance. This results in a sawtooth shape of the voltage fluctuations at a characteristic frequency f. This frequency increases with a rise of the arc current, and decreases with a rise of the anode diameter, as shown in Fig. 8. These variations can he explained by the decrease of the gap between the internal anode wall and the arc column (whose diameter increases with arc current, t32> which leads to a rise of the arcing frequency between them (see Fig. 1) and a decrease of the voltage fluctuations. No significant change of frequency could be observed with a change of gas flow rate between 40 and 55 slm. For an i.d. equal to 10 mm [slightly bigger than the diameters of the anodes used for the results presented in Fig. 8 (8 or 9 mm)], the frequency linked to the vortex flow (without the arc) of the cold gas increased almost linearly from 1000 to 1600 Hz when the gas flow rate was raised from 40 to 60 slm (see Fig. 5). This characteristic frequency cannot be isolated in the frequency spectra when the electric arc was applied. Moreover, the frequency peak, characteristic of the arc root movement at the anode wall, appeared
244
Brfl~c et al.
at higher frequencies (between 4650 and 4850 Hz) almost independently of G'. However, the vortex flow could still be present although its characteristic frequency, of much lower magnitude compared to the arc frequencies, did not appear. So the effect of the vortex flow of the cold gas is strongly weakened by the electric gas as has already been shown (21) and, contrarily to the results of Russ et al., (25) seems to have a weak effect on the arc root displacement inside the anode compared to the arc column-anode wall disruptions. The expansion of the gas due to heating adds axial momentum but does not change the rotational inertia of the gas. The vortex flow may be pushed closer to the wall, increasing the velocity gradients which will increase the wall friction and thereby reduce the swirl level. Moreover, it has to be emphasized that in our vortex torch, the arc attachment occurs in the anode channel 30 mm downstream of the arc chamber, i.e., at a position where the plasma column is well developed and thus the swirl level strongly reduced, while in the Russ experiments the arc attachment seems to occur very close to the arc chamber where the vortex was still high. To characterize the frequency change of arc disruptions vs. I, G, and d, a correlation was calculated in a previous studyt33} between the members of Si, Re, and L/d (L = total length of the anode), and a number fd3/G proportional to the Strouhal number. This calculated relationship was good, but the Strouhal number did not seem very representative of the physical phenomena involved in the variation of the frequency. So a new member was sought using the method of dimensional analysis. This number was obtained by the change of the reference basis related to the independent dimensions [L (m)=length, T (K) = temperature, t (s)=time, I (A)=arc current, m (kg)= mass], to a new basis related to the following five parameters:/, L, ~c- mean thermal conductivity, p = viscosity, o. = electrical conductivity. From the basis change, a new dimensionless number was derived and chosen as the ratio between the frequency f and its dimensions in the new reference basis expressed in (L -2. p-0.s.o.-O.5, i,):
Sf - f d2"f~°/ ll°
(1)
I This number takes into account the two experimental parameters (I and d) which act upon the frequency when the previous correlation was related only to d and G. A new relationship was calculated with this number: Sf=0.031 Si -°'4j5 Re -°'4j3 -
(2)
It takes into account well the frequency change with variation in the experimental parameters as shown in Fig. 8 (mean standard deviation A = 0.64%).
Behavior of dc Vortex P l u m Torches
245
4.3. Reverse Polarity: Flat-Shaped Anode
The torch was tested in reverse polarity although this electrical configuration led to a strong erosion of the thoriated tungsten electrode (which was the anode in this polarity). Nitrogen was used as plasma gas (G' = 60 slm), and the arc current ranged between 225 and 270 A. In good agreement with the dynamic behavior of the same torch in direct polarity, strong fluctuations of the arc voltage with a sawtooth shape were observed. However, the frequency of this fluctuation was lower with this electrical configuration (from 1000 to 2500 Hz vs. 3000 to 4000 Hz in direct polarity). The stronger cathodic jet (compared to the anodic one), which reduces the arc root movement under the action of the drag force close to the cylindrical electrode wall, results in a continuous displacement instead of disruptions: 34~ The disruptions are probably only at the anode, which is now perpendicular to the plasma jet, and such a geometry might explain the lower frequency observed. Moreover, a displacement of the erosion zone near the exit of the nozzle was observed. This change was linked to an increase of the arc length, the erosion area being located at the first half part of the nozzle in the case of direct polarity connection. Such a behavior was already observed in Ref. 4, but no clear explanation was found. 4.4. Fluctuations of Acoustical Pressure and Plasma Jet Radiations
The fluctuations of acoustical pressure and plasma radiation were studied, in order to see if they were correlated to the undulations of arc voltage and current. In Fig. 9, the synchronized signals of arc voltage and current, resulting power level and plasma radiation, derivative of the power level, and acoustical pressure are depicted for the torch equipped with the cone-shaped-cathode and a CD anode of 6 mm i.d. (I= 260 A, nitrogen flow rate G' = 50 slm). The time lag due to the transport of plasma from the inside of the torch to the outside was very short (around 100/as, the plasma jet velocity being about 500 m/s with these experimental values~ss~ compared to the total delay of the record (10 ms). Thus, this time lag was not taken into account in Fig. 9. However, the time lag in acoustic pressure measurements due to the distance (d= 1 m) between the torch and the microphone was not negligible. In fact, the sound velocity at the mean temperature (303 K) of the surrounding atmosphere of the plasma jet is roughly 350 m/s. So the time lag in acoustic measurement was At~2.85 ms, and the corresponding signal is shifted to the right in Fig. 9. However, as no low-frequency component was detected on this signal ( f ~ 5000 Hz), this time lag does not have any influence on the measurements. A strong correlation between the electrical power fluctuations, where the two characteristic undulations of the arc voltage (5728 Hz) and of the
2,16
Brilhac et =!, 10 ms
,~
dPI (It la.u.)
Aco=etJc(w.)
2.85 ms >
Fig. 9. Shape of the fluctuations for the anode (6 mm i.d•) ( I = 260 A, G'-- 50 sire, U = 7"/V).
current signals (300 Hz) appeared, and the radiation signal can be observed. Similarly there is a strong correlation between the time derivative of the electrical power and the acoustical pressure wave. For these two signals, the low frequency undulation (at 300 Hz) disappeared. The frequency of 1500 Hz detected in the sound signal when the torch was not ignited, with frequency corresponding to the vortex flow, was not seen in any signal with the ignited plasma• Other measurements are necessary to get more detailed information on the correlations between sound signals and power fluctuations• Whatever be the experimental parameters, the radiation signals and the acoustic pressure waves were systematically correlated with the power signal and its derivative (undulations at the same frequencies). In Fig. 10, the frequency spectra corresponding to the signals shown in Fig. 9 are shown. Two characteristic frequencies are isolated: one at 5728 Hz characteristic of the arc column-anode wall disruption, and another at the lower frequency generated by the undulation of the power supply .(300 Hz and its harmonics). This low-frequency peak is very strong in electrical power and plasma radiation spectra, while it disappears practically in the spectra of the derivative of electrical power and acoustical waves. Thus, on the one hand, the source of sound of a plasma torch is strongly correlated to the derivative of the electrical power. Such a result was already obtained for a free burning arc. t27~ On the other hand, it is clear that the
Behavior of de Vortex Plasma Torches
247
.4 Arc Voltage
i|l
:1
i IL[IIlect~c~ij4.~ier
] Powor radlatlon
ILl
+ I Oedvative of electrical power
+
Acoustical prmmure
L
_
_ _ L . _
w.
Fig. 10. Corresponding frequency spectra (arbitrary units) of the signals shown in Fig. 9.
luminous fluctuations of the plasma jet are generated by the undulation of the electrical power, especially those due to the restrike mode of the arc at the anode. This specific behavior of the luminous fluctuations will be used to determine the jet velocity using the method developed by J.-F. Coudert et al. t28) and based on the propagation of these fluctuations. The study of the acoustical spectra will probably give information on the behavior of the arc attachment at the anode wall.
Brilhac ¢t ai.
u(v) 70 ~ o
65
~"'b
60
~
•"
~
50
200
=
55 slm(d=8 mm)~
9 ÷
- 45
80 slm(d=8 mm)o
mm)
~
• ~
calculated datas
50 slm ~ (d=8 mm)'
. 4 0 slrr(d=8 mm)
2~0
300
3~0
I(A)
Fig. II. Arc characteristics for the truncated-cone-shaped cathode for two anode i.d. (8 and 9 ram).
5. STATIC BEHAVIOR 5.1. Torch with a Truncated-Cone-Shaped Cathode With this type of cathode, two internal diameters (i.d.) of anode were studied ( d = 8 or 9 mm). The change of arc voltage with arc current for different nitrogen flow rates is shown in Fig. 11. The arc characteristics 2 are always decreasing, and rising values of G' lead to an incerase of U at constant arc current. These results are in good agreement with those given in Ref. 5, the arc length being self-adjusting (anode with constant diameter). A slight decrease of U with a rise of d is observed for given values of I and G'. This variation is probably due to the lower constriction of the arc column in the anode when its diameter increases. These variations are well represented by the relationship ( Udt~o _ 4.95
12
/-°'654( G / -0'327 (3)
with a mean deviation equal to 1.9%. The variations of U vs. I, G', and d are well represented as shown in Fig. I I where the calculated values of U are compared with the experimental ones in the graph (L U). For this torch with a hot button type cathode, the overall thermal losses, including radiation, occur mainly at the anode (70% of the total 2Thescalesof arc current are corrected related to those givenin Refs. 33 and 36 after calibration of the probe.
Behavior of dc Vortex Plasma Torches
249
Hp (MJ/kg) 20 19 18
17 16
15 14 13 12 11 10
40 slm 55 slm o
90 s l m .
~,
2t0
2~,0
'
2(~0
'
°
2~0
360 I (;A) 3:!0
Fig. 12. Torch enthalpy variation vs. arc current.
losses which represent about 5 kW, against 30% at the cathode). As shown in Fig. 12, the gas enthalpy Hp ranges between 9 and 20 MJ/kg and increases when the arc current rises and the gas flow rate decreases. The thermal efficiency varied from 55% for the lower gas flow rate to 70% for the higher one (Fig. 13). 5.2. Torch with a Flat-Shaped Cathode
Only one diameter of anode (d= 7 mm) was used with this cathode. Both nitrogen and argon were used as plasma gases in order to check the
Thermal
efficiency
0.7 90 s l m .
0.65 55 slm o
0.6 40 slm
Q A A M
0.55 L 200
,
220 '
2~0
'
260
'
2so
'
360 liAr20
Fig. 13. Variation o f the thermal efficiency of the torch vs. arc current.
Brillmeet ai.
250 u (v)
140
100 slm 80 60 50 40
120 100
slm slm slm slm
a == . = ~ i ~ =
8o
60
_ calculated values
~
100 slm A 40 • 80 sJm
~
o °
71slm
"
2(~00
"
200
4=
= = "~:::a
3()0
400
_.
] I(A)
Fig. 14. Arc characteristics for the fiat-shaped cathodes with nitrogen or argon as plasma gas (7 mm anode i.d.).
influence of the nature of the gas on the torch behavior. The torch was also tested in reverse polarity.
5. 2.1. Influence of the Nature of the Gas With nitrogen as plasma gas, a strong increase of the voltage was noticed compared to that obtained with an anode of 8 mm i.d. and with a truncated-cone-shape [80 to 150 V against 50 to 70 V for similar ranges of arc current (200 < I (A) <400) and gas flow rate (40 < G' (slm) < 100)]. This increase of U can be explained both by the anode diameter decrease (7 mm against 8 mm) and the lengthening of the anode channel with the flat shape injection chamber. An increase of G' led also to a rise of U (see Fig. 14). All these variations are well represented by the following relationship (standard deviation equal to 2.2%): 2
I
"1
\Gdtroho]
--0.643
--0.137
~pod]
Compared to Eq. (3), the exponent of Si is unchanged (-0.643 against -0.654) but the exponent of Re is smaller (-0.137 against -0.327). So, the increase of arc voltage with gas flow rate is more important for this anode compared to the one associated with a truncated-cone-shaped cathode. This might be due to a more efficient vortex with this shape of injection chamber. With argon as plasma gas (see Fig. 14), for the same ranges of gas flow rate and arc current, a strong decrease of U is noticed compared to the values obtained with nitrogen (35-50 V against 80-150 V). This voltage drop
Behavior of dc Vortex Plasma Torches
251
is due to the less constricted arc with a monoatomic gas (such as argon) compared to a diatomic gas (such as nitrogen whose mean integrated thermal conductivity is about three to ten times that of argonJ 37)) In fact, with nitrogen arcs, the heat transfer is higher than with argon arcs. Since the arc column constricts more as the radial heat losses increase, the electric field rises at constant arc current and thus the arc voltage increases. A relationship was also calculated to characterize the behavior of the torch with argon, and the following result was obtained (mean standard deviation = 3%): .
i2
~-0.565.
~-0.183
Udtr°-4.95(--] ( GI I \Gdcroho] \pod]
(5)
The exponent of Si is lower than that obtained with nitrogen (-0.565 against -0.643), and so the arc characteristics are flatter with argon (more stable behavior of the torch). To take into account the change in the nature of the gas, a new dimensionless number Pr (Prandtl number) must be introduced in this type of relationship. This number depends only on the nature of the gas: P r = I.to" ho/~Co" To, where !c0 is the mean thermal conductivity. With Pr, the following relationship was calculated with the experimental data obtained both for nitrogen and argon as plasma gases: /
-2
\-0"571
~
\-0"121
Ud(ro=2.04( '- } ( tr } I \Gdcroho} \ pod}
/
\-0.386
I p°n°} \tOoTo}
(6)
This correlation takes into account well the voltage change with all the experimental parameters obtained for the laboratory torch with a button type cathode, as shown in Fig. 14, where calculated and experimental values of U are compared (mean standard deviation equal to 2.4%). Of course such correlations hold only for the considered laboratory torch, and as we have no data related to other torches of this type, it is not possible to state that they are valid for other torches. However, as will be seen in Part 2, for torches with a well-type cathode, the correlations obtained with the laboratory torch are quite similar to those obtained for two other torches: Plasma Energy Corporation and Aerospatiale, whose design and power levels are very different.
5.2.2. Behavior in Reverse Polarity The behavior of the torch was also tested in reverse polarity. For nitrogen as plasma gas (G'=50 slm), the arc current ranged between 225 and 275 A. Compared to the direct polarity, a strong increase of arc voltage
252
Bdlhae etal.
was observed (150-170 V against 90-110 V). Quite similar variations were also observed c4) with a torch equipped with a "~utton type anode, and by Camacho t38"~9)with a cold well-type anode. This increase was correlated to the growth of the arc length, the mean position of the arc root at the anode wall being estimated from the erosion zone. This area extends on the first half part of the anode (from 15 to 40 mm) in direct polarity and near the exit of the channel (from 60 to 80 mm) for the reverse polarity. As already emphasized in Section 4.3, the interactions between arc root and gas flow are fairly different when the polarities are inverted. 6. CONCLUSION This work was devoted to the study of the dynamic and static behaviors of a dc vortex plasma torch with a button type cathode (power level below 60 kW working with N2 and Ar). In this configuration only the arc root at the anode was continuously moving (restrike mode) while the cathode spot was very stable, which was not the case with the well-type cathode studied in Part 2. To characterize the dynamic behavior of the torch, a set-up was developed to measure the fluctuations of arc voltage and current, plasma jet radiation, and acoustic pressure. Two characteristic frequenciesfwere found in each signal: a low-frequency one (300 Hz and its harmonics), characteristics of the torch power supply (current source) and another one at higher frequency (around 4000 Hz) characteristic of the arc column-anode wall disruptions. The corresponding undulation of the voltage signal had a sawtooth shape. Its magnitude was very large (about that of the mean arc voltage), and its frequency increased when the arc current was raised and/ or the anode internal diameter d decreased. No particular effect of the gas flow rate G on this frequency was observed for the considered flow rates (40 < G' (slm) < 55) and anode diameters (from 7 to 9 mm). This was probably due to the arc diameter, which for the considered arc currents below 350 A, was calculated from the Elenbaas-Heller equation to be less than 5 mm with nozzle i.d. between 7 and 9 mm. In such conditions as demonstrated by Coudert e t aLt4°) for spraying plasma torches, the gas flow rate had almost a negligible influence on the arc disruptions at the anode. Its influence was only sensitive for arc currents over 500 A for nozzle i.d. of 7 ram. The vortex injection of gas seemed to have no influence on the arc root disruptions (the characteristic frequency of the vortex cold gas flow in the anode disappeared with the electric arc). This is probably due to the arc attachment at the middle of the anode channel resulting in a well-developed plasma column where the swirl level is strongly reduced (high axial velocity of the plasma flow compared to the momentum of the vortex flow). A classical
Behavior of dc Vortex Plasma Torches
253
semiempirical relationship (Su =f(Si, Re, I/d)) was calculated between the dimensionless numbers: Su -
Udcro , I
12
Si - - -
Gdcroho'
G
Re = - -
~od
where O'o,/~o,and h0 are, respectively,the electricalconductivity, viscosity, and enthalpy of the plasma with a molar fraction of electrons at I%, and a new one Sf=f(Si, Re, I/d) characterizingthe frequency of voltage fluctuation through a new dimensionless number Sf=fd2~,,/b~o~o/l determinated by dimensional analysis.This correlation takes into account well the variations off with I and d. Both acoustical pressure fluctuations and plasma jet radiations are strongly linked to electrical power fluctuations. It is shown that the acoustical fluctuations are generated by the derivative of the electrical power of the torch, and that the radiation fluctuations are closely linked to the fluctuations of electrical power. These results have shown that the dynamic behavior of the torch is closely linked to the anode erosion through the lifetime of the anodic spot. A more detailed study of the influence of the macroscopic parameters on voltage, light, and acoustical fluctuations coupled with copper emission will probably help in the design of electrodes with a low erosion rate. Such studies are under completion. The analysis of the static behavior of a dc vortex plasma torch with a button type cathode and tubular anode of constant diameter showed that the arc characteristics were always decreasing. Compared to nitrogen whose thermal conductivity is high, the arc voltage decreased when argon was used as plasma gas. When the torch was connected in reverse polarity, the arc voltage strongly increased (150%) due to the lengthening of the plasma column. The variations of U with/, G, nature of the gas, and anode design were well represented by semiempirical relationships Su =f(Si, Re, Pr) with the Prandtl number defined as Pr = I.toho/xoTo. Unfortunately, due to the lack of data related to other torches with button type torches, it has not been possible to check the validity of our dimensionless relationships for the static behavior with other torches. However, as shown in Part 2 for a similar torch with a well-type cathode, the established dimensionless relationships fit rather well the data related to two more powerful torches (up to 2 MW) : the Aerospatiale torch and the Plasma Energy Corporation torch, both being equipped with well-type cathodes. ACKNOWLEDGMENTS This work has been supported by Electricite de France (D.E.R.A.D.E.I. Departement).
2,54
Brillmc et al.
REFERENCES I. P. Fauchais, J. High Temp. Chem. Process. 1, I (1992). 2. A. G. Shaskov and O. I. Yas'ko, "Application of approximate similarity for correlating arc characteristics," Institute of Heat and Mass Transfer of Academy of Belarus, Minsk, Belarus, July 2, 1973, p. 21. 3. O. I. Yas'ko, Br. J. Appl. Phys. 2, 733 (1969). 4. M. F. Zhukov, "Principes de calcul de plasmatrons ~. circuit lin~aire," Documents d 'informations op6rationnels de I'lnstitut de Thermophysique-Branche Sib~rienne de I'Acad~mie des Sciences, Novosibirsk (1975). 5. M. F. Zhukov, "Electrical arc plasmatrons," U.S.S.R. Academy of Sciences, 630090 Novosibirsk-90, Esperanza Engineering Group, p. I. 6. M. F. Zhukov, "Electric arc generators of low-temperature plasma," High-Temperature Dust Laden Jets, O. P. Solonenko and A. I. Fedorchenko, eds. (VSP, The Netherlands, 1989), p. I. 7. S. A. Wutzke, E. Pfender, and E. R. G. Eckert, AIAA J. 5, 707 (1967). 8. J. McKelliget, J. Szekely, M. Vardelle, and P. Fauchais, Plasma Chem. Plasma Process. 2, 317 (1982). 9. A. H. Dilawari and J. Szekely, Plasma Chem. Plasma Process. 7, 317 (1987). I0. O. P. Solonenko and A. L. Sorokin, "Numerical analysis of a heterogeneous turbulent plasma jet under radial-annular injection of powder materials," ISPC 10, Ute Ehlemann, H. G. Lergon, and K. Wiesemann, eds. (Univ. of Bochum, Germany, 1991), Vol. !. il. O. P. Solonenko, "The fundamental thermophysical problems of plasma-spraying," in Thermal Spray: Advances in Coating Technology (ASM Int., Metals Park, Ohio, 1992), p. 787. 12. C. H. Chang, "Numerical simulation of alumina spraying in an argon-helium plasma jet," in Thermal Spray: Advances in Coating Technology (ASM Int., Metals Park, Ohio, 1992), p. 793. 13. D. J. Varacalle, Jr. and W. L. Riggs, I1, "Analytically modelling the plasma spray deposition of Tribaloy 888," in Thermal Spray: Advances in Coating Technology (ASM Int., Metals Park, Ohio, 1991), p. 245. 14. X. Zhou, J. Heberlein, and E. Pfender, "Theoretical study of factors influencing arc erosion of cathode," accepted for publication in IEEE Trans. Components, Hybrids, Manuf. Technol. (1993). 15. C. Delalondre and O.'Simonin, "Numerical modelling of high-intensity arcs including a flow turbulence model," Proc. 12th Congress "International Union for Electroheat" (Les Editions de la Cheneli~re, Montreal, Canada, 1992). 16. N. Sanders, K. Etemadi, K. C. Hsu, and E. Pfender, J. Appl. Phys. 53, 4136 (1982). 17. D. M. Chen and E. Pfender, IEEEPIasma Sci. PS-9, 265 (1981). 18. K. C. Hsu and E. Pfender, J. Appl. Phys. 54, 3818 (1983). 19. K. C. Hsu and E. Pfender, J. Appl. Phys. 54, 4359 (1983). 20. L. Jestin, B. Pateyron, and P. Fauchais, "Dimensionless study of experimental pressure related behavior for two powerful vortex blown arcs and vortex-stabilized power plasma torches," High-Temp. Dust Laden Jets, O. P. Solonenko and A. I. Fedorchenko, eds. (VSP, The Netherlands, 1989), p. 21 I. 21. Y. P. Chyou and E. Pfender, Plasma Chem. Plasma Process. 9, 291 (1989). 22. D. Lemoine, "D~termination de rorigine de r~mission sonore des g~n~rateurs de plasma d courant continu de petite puissance," M~moire C.N.A.M., Paris (1983). 23. J.-F. Coudert, M.-P. Planche, O. Betoule, and P. Fauchais, "Study of the influence of the arc root fluctuations on a d.c. spray plasma torch," ISPC I I, J. Harry, ed. (University of Loughborough, 1993), Vol. 1, p. 198.
Behavior of dc Vortex Plasma Torches
255
24. J.-F. Coudert, M.-P. Planche, and P. Fauchais, "Influence of the anode arc attachment on the dynamic behavior of plasma jet produced by a d.c. plasma torch," to be published in proc. of "International symposium on heat transfer under plasma conditions," July 4-8, 1994, lzmir, Turkey (W. Begell, N. Y., 1994). 25. S. Russ, E. Pfender, and J. Heberlein, "Anode arc attachment control using boundary layers bleed holes," in Thermal Spray Research and Applications (ASM Int., Metals Park, Ohio, 1993), p. 97. 26. E. Pfender, W. L. T. Chen, and R. Spores, "A new look at the thermal and gas dynamic characteristics of a plasma jet," in Thermal Spray Research and Applications (ASM Int, Metals Park, Ohio, 1990), p. I. 27. M. Fitaire and T. D. Mant~i, Phys. Fluids 15, 464 (1976). 28. J.-F. Coudert, O. Betoule, and M.-P. Planche, "A new method of measurement of a plasma flow velocity close to the nozzle exit," ISPC I I, J. Harry, ed. (University of Loughborough, 1993), Vol. I, p. 204. 29. B. Pateyron, "Contribution it la rralisation et ~ la modrlisation de rracteurs plasmas soufflrs ou transfrr6 appliqurs/i la mrtallurgie et ~ la production de poudres ultrafines m~talliques ou crramiques," Thrse d'rtat n°21 (1987), Universit6 de Limoges, France. 30. J.-F. Brilhac, B. Pateyron, and P. Fauchais, "Acoustic diagnostics to characterize the behavior of plasma torches," 1SPC 10, Ute Ehlemann, H. G. Lergon, and K. Wiesemann, eds. (University of Bochum, Germany, 1991), Vol. 4, pp. 1.2-1 I. 31. L. Jestin and G. Schmitt, High Temp. Chem. Process. I, 511 (1992). 32. E. Pfender, "Plasma Generation," International Summer School on Plasma Chemistry, E. Pfender and K. Akaski, eds. (University of Tokyo, Japan, 1987), p. 65. 33. J.-F. Brilhac, B. Pateyron, J.-F. Coudert, P. Fauchais, A. Bouvier, P. Pasquin, and L. Jestin, J. High Temp. Chem. Process. I, 421 (1992). 34. P. Teste, Thesis, "Contribution fi I'(~tude de I'~rosion des 61ectrodes de torches ~ plasma, Incidence de la structure m~tallurgique," University of Paris VI, France, p. 22. 35. J.-F. Brilhac, B. Pateyron, J.-F. Coudert, and P. Fauchais, "Plasma jet velocity measurement of a d.c. vortex plasma torch." ISPC 1 I, J. Harry, ed. (University of Loughborough, 1993), Vol. I, p. 362. 36. J.-F. Brilhac, B. Pateyron, J.-F. Coudert, P. Fauchais, A. Bouvier, P. Pasquini, and L. Jestin, J. High Temp. Chem. Process. I, 557 (1992). 37. ADEP, Banque de donn~es de I'Universit8 et du CNRS. Ed. Direction des Biblioth~ques des MusSes et de I'Information Scientifique et Technique (1986); B. Pateyron, These de Doctorates Sciences Physiques, University of Limoges, France, Nb. 21, 1987. 38. S. L. Camacho, "Industry experience with the reverse-polarity plasma torch," ISPC-9, Workshop on industrial plasma application," M. I. Boulos and R. d'Agostino, eds. (University of Bari, Italy, 1989), Vol. 2, p. 9. 39. S. L. Camacho, "The reverse-polarity plasma torch," ISPC-8, K. Akashi and A. Kinhara, eds. (University of Tokyo, Japan, 1987), Vol. 1, p. 82. 40. J.-F. Coudert, M.-P. Planche, and P. Fauchais, "Plasma jet velocity measurements: Part I : d.c. spraying plasma torch," submitted to Plasma Chem. Plasma Process.