Thermal Plasma Torches and Technologies - Vol.1 - Plasma Torches. Basic Studies and Design (O. P. Solonenko 2003)

THERMAL PLASMA
TORCHES
AND
TECHNOLOGIES
THERMAL PLASMA
TORCHES
AND
TECHNOLOGIES
Volume 1: Plasma Torches. Basic Studies and Design
edited by
O.P. Solonenko
Institute of Theoretical and Applied Mechanics, Siberian Branch of the
Russian Academy of Sciences, Novosibirsk
CAMBRIDGE INTERNATIONAL SCIENCE PUBLISHING
Published by
Cambridge International Science Publishing
7 Meadow Walk, Great Abington, Cambridge CB1 6AZ, UK
http://www.cisp-publishing.com
Published 2003
Cambridge International Science Publishing
Conditions of sale
All rights reserved. No part of this publication may be reproduced or
transmitted in any form or by any means, electronic or mechanical, including
photocopy, recording, or any information storage and retrieval system,
without permission in writing from the publisher
British Library Cataloguing in Publication Data
A catalogue record for this book is available from the British Library
ISBN 1 898326592
Contents
Electric Arc Generators of Thermal Plasma: Review (M.F. Zhukov) .............. 1
Trends in Thermal Plasma Technology (L.F. Pfender) ...................................... 20
Integrated Analysis of Induction Plasma Systems (M.I. Boulos) ...................... 42
Plasma Metallurgy: Current State, Problems and Prospects (Yu.V. Tsvetkov) 62
Atmosphere Plasma Spraying: Theory, Modelling, Diagnostics, Computer-
Aided Design and Some Applications (O.P. Solonenko) ............................... 80
Metallic and Ceramic Materials: Present and Future (V.A. Neronov) .......... 102
Extreme Technologies in Building Material Production (G.G. Volokitin,
V.E. Borzykh, N.K. Skripnikova) .................................................................... 112
Achievements of Plasma Tomography (V.V. Pickalov) ...................................... 123
New Vortex Method of Plasma Insulation and the Ranque Effect (A. Gutsol
and J.A. Bakken).............................................................................................. 133
Physical Phenomena in a Hollow Cathode and Interaction of Powder with
the Vacuum Arc (V.S. Cherednichenko, A.B. Zagorskii and L.K. Pavlenko) ..
.......................................................................................................................... 150
Experimental Investigation of Electric, Energetic and Optical Character-
istics of the Induction Transformer-Type Discharge (I.M. Ulanov and
S.N. Soldatov) .................................................................................................. 166
Mathematical Modelling of Transformer Discharge (E.B. Kulumbaev and
V.M. Lelevkin) .................................................................................................. 175
Integrated Method of Research of Processes in Thermal Plasma
Generators (O.Y. Novikov and V.F. Putko) .................................................... 191
Modelling of Electric Arc Plasma (A. Zhainakov, R. Urusov and A. Valeeva) 197
On the Problem of Turbulent Arc Modelling (O.I. Yas'ko) .............................. 206
Energy Characteristics of Electric Arc Heaters for Tetrafluoromethane (A.N.
Timoshevskii, B.A. Pozdnyakov and V.S. Ponkratov) ................................... 224
Numerical Investigation of the Characteristics of a Steady Curved Arc in
External Fields (G.A. Desyatkov, V.Ts. Gurovich and E.B. Kulumbaev) .... 231
Properties of Water-Stabilized Plasma Torches (M. Hrabovsky, M. Konrad, V.
Kopecky and V. Sember) .................................................................................. 242
Modeling of a D. C. Arc Plasma Torch with a Hydrogen-Argon Mixture
as the Working Gas (Xi Chen, Peng Han and Lan Yu) ................................ 267
The Modified Balance Method of Calculating Characteristics of Near-
Anode Processes (Yu.V. Bruevich, I.P. Nazarenko and I.G. Panevin) ......... 280
Application of Heat Diffusion Theory to Cold Electrode Erosion for Nonsta-
tionary Arc Spots (A. Marotta and L.I. Sharakhovsky) .............................. 291
Emission Current Density and Electronic Work Function of Metal to
Plasma (H.Ts. Zayatuev) ................................................................................. 310
Erosion of Multi-arc Cathodes (A.N. Timoshevkii and V.S. Ponkratov) .......... 319
Influence of Three-Phase Electric Arcs, Burning on Oxide Melts, on
Quality of Electric Power (U.B. Ashimov, E.A. Abdrachmanov, G.D.
Manapova and T.S. Maldybaeva) ................................................................... 326
Modelling of Flow Stabilization by the Swirl of a Peripheral Flow as
Applied to Plasma Reactors (E.P. Volchkov, V.P. Lebedev, V.I. Terekhov and
N.E. Shishkin) .................................................................................................. 335
Low-density Plasma Jets: Production and Investigation (L.I. Kuznetsov and
V.N. Yarygin) .................................................................................................... 352
Effect of Arc Current Modulation on Thermal Plasma Flow in Plasma Torches
(F.A. Salyanov)................................................................................................. 365
PREFACE
To The Memory of Mikhail F. Zhukov
Our original intention was to publish this two-volume book at the
occasion of the 80th birthday of Prof. Mikhail F. Zhukov, Academician
of the Russian Academy of Sciences, one of the leading Russian and
CIS scientists working in the area of plasma science and technol-
ogy. Unfortunately, Prof Zhukov died in December 1999 and we would
therefore like to devote this book to the life and scientific achievement
of this outstanding scientist, pioneer of thermal plasma.
He was born September 6, 1917 and graduated from the Moscow
State University (in mechanics) in 1941. He started working as an
engineer in the Zhukovsky Central Aerodynamics Institute. In 1946-
1959, he was the head of the department in the Central Institute of
Airplane Engines. Since 1960 until his death, Mikhail F. Zhukov worked
in the Siberian Branch of the Russian Academy of Sciences. Zhukovs
fundamental investigations in the fields of gas dynamics and elec-
tric arc thermal plasma are well known in Russia and abroad. The
scope of his scientific interests was very wide: supersonic compressors,
Mikhail F. Zhukov
new types of wind tunnels, near-electrode processes, electrode erosion,
heat transfer, arcing stability, new plasma technologies, etc.
Zhukovs scientific school on plasma dynamics is known to both
Russian and foreign investigators. M.F. Zhukov was a pioneer of
experimental investigations of thermal plasma generators. He and his
disciples were the first involved in the study of some fundamental
physical processes in electric arc plasma. He developed a universal
system of similarity criteria and estimated the significance of these
values under specific arcing conditions. He elaborated the arcing theory
in a laminar flow. His engineering methods for designing the elec-
tric and thermal characteristics of electric arc plasma torches are used
widely in the development of laboratory and industrial apparatuses.
Zhukovs team have carried out numerous investigations concerning
near-electrode processes, heat transfer through the arc spot, split-
ting of the closing radial part of the arc in linear scheme plasma
torches. They have also investigated the problem of stability of multi-
arc system operation without a ballast resistance in the electric circuit.
These studies allowed the development of highly effective linear plasma
generators and high-current cathode units with a long operation time
in different gas media. Twenty years ago, he put forward a hypothesis
about the recirculation of cathode materials atoms close to the surface.
This was discovered later in experiments. It was a specific case, but
it opened the way to solving the problem. A new class of high-
resource cathodes has been developed, including the non-erosion cathode
working in a carbon-containing medium.
Powerful high-enthalpy arc plasma generators with interelectrode
inserts were developed under the supervision of M.F. Zhukov. They
are widely used in plasma chemical technologies. There were sig-
nificant achievements in the production of ultrafine powder of
refractory compounds, synthesized in plasma chemical reactors. These
powders are also used in metallurgy for the improvement of the physical
and mechanical properties of ferrous and nonferrous metals.
M. Zhukovs intuition and his vision of future conditioned the
appearance in the 1980s of a new scientific and technological direction
plasma dynamics of dispersed systems. His achievements determined
the progress in plasma-jet spraying with powder coatings and com-
positions. His team has made a great contribution to the investigation
and practical realization of plasma (no fuel oil) igniters in the grid-
coal power stations.
The popularization of science achievements and their applications
received a lot of attention from M. Zhukov. He wrote and spoke about
different subjects: plasma for powder spraying, metal cutting and
processing, waste treatment, recirculation of toxic waste to a serviceable
product, etc. He often spoke at lectures to engineering public, sci-
entific-practical workshops, radio and TV. Thus, he promoted the whole
range of possible practical applications of electric-arc thermal plasma
in new technologies.
M.F. Zhukov is the author of many books and articles. With his
participation and under his editing, the whole series of famous books
with plasma torch construction has been published. The publication
of 20-volume series Low-Temperature Plasma will soon be com-
pleted. Many of these books have became essential reading for re-
searchers and specialists in the field of thermal and low-tempera-
ture plasma. These editions have became a theoretical tool for the
design and manufacturing of high-efficiency equipment for new plasma
technologies.
Mikhail Zhukov was a man of many interests. In addition to carrying
out extensive investigations, he was an organizer and participant of
conferences and workshops on gas-discharge plasma for almost forty
years.
From 1975 until 1980, M.F. Zhukov was the Chief Scientific Secretary
of the Presidium of the Siberian Branch of the USSR Academy of
Sciences. He was the Editor-in chief of the journal Communications
of SB AS USSR for 15 years, and a member of the Editorial board
of Contributions to Plasma Physics. Mikhail F. Zhukov was the Acad-
emician of the International Energy Academy, a Honorable Academician
of the Engineering Sciences Academy.
M.F. Zhukov was the Head of the Plasma Dynamics Department
of the Institute of Theoretical and Applied Mechanics of the Sibe-
rian Division of the Russian Academy of Sciences. He was absorbed
with new creative plans and aspirations.
This book is published as a result of international scientific and
technical discussions and collaboration which took place during the
preparation and running of 3rd International Workshop on Thermal
Plasma Torches and Technologies (TPPT97) (25-29 August, 1997,
Novosibirsk, Akademgorodok, Russia). The organisation of this Workshop
and preparation of this collection were sponsored by the Siberian Branch
of Russian Academy of Sciences, Samsung Heavy Industries Co. Ltd.,
Daeduk R&D Center (Taejeon, Korea), Gusinoozyerskaya GRES
(Gusinoozyersk, Russia), Novosibirsk Plant Khimkontsentrat (Nov-
osibirsk, Russia) and Production-Promotion Company Mercury-II
Ltd. (Tomsk, Russia).
Leading scientists presenting the scientific school of Prof. Mikhael
F. Zhukov and other leading scientists in plasma research and de-
velopment from Russia, CIS, USA, Canada, France, Japan, Czech
Republic, China, Brazil, Norway, Yugoslavia and Korea have con-
tributed this book.
Prof. Oleg P. Solonenko,
Editor
Contents
Electric Arc Generators of Thermal Plasma: Review (M.F. Zhukov) .............. 1
Trends in Thermal Plasma Technology (L.F. Pfender) ...................................... 20
Integrated Analysis of Induction Plasma Systems (M.I. Boulos) ...................... 42
Plasma Metallurgy: Current State, Problems and Prospects (Yu.V. Tsvetkov) 62
Atmosphere Plasma Spraying: Theory, Modelling, Diagnostics, Computer-
Aided Design and Some Applications (O.P. Solonenko) ............................... 80
Metallic and Ceramic Materials: Present and Future (V.A. Neronov) .......... 102
Extreme Technologies in Building Material Production (G.G. Volokitin,
V.E. Borzykh, N.K. Skripnikova) .................................................................... 112
Achievements of Plasma Tomography (V.V. Pickalov) ...................................... 123
New Vortex Method of Plasma Insulation and the Ranque Effect (A. Gutsol
and J.A. Bakken).............................................................................................. 133
Physical Phenomena in a Hollow Cathode and Interaction of Powder with
the Vacuum Arc (V.S. Cherednichenko, A.B. Zagorskii and L.K. Pavlenko) ..
.......................................................................................................................... 150
Experimental Investigation of Electric, Energetic and Optical Character-
istics of the Induction Transformer-Type Discharge (I.M. Ulanov and
S.N. Soldatov) .................................................................................................. 166
Mathematical Modelling of Transformer Discharge (E.B. Kulumbaev and
V.M. Lelevkin) .................................................................................................. 175
Integrated Method of Research of Processes in Thermal Plasma
Generators (O.Y. Novikov and V.F. Putko) .................................................... 191
Modelling of Electric Arc Plasma (A. Zhainakov, R. Urusov and A. Valeeva) 197
On the Problem of Turbulent Arc Modelling (O.I. Yas'ko) .............................. 206
Energy Characteristics of Electric Arc Heaters for Tetrafluoromethane (A.N.
Timoshevskii, B.A. Pozdnyakov and V.S. Ponkratov) ................................... 224
Numerical Investigation of the Characteristics of a Steady Curved Arc in
External Fields (G.A. Desyatkov, V.Ts. Gurovich and E.B. Kulumbaev) .... 231
Properties of Water-Stabilized Plasma Torches (M. Hrabovsky, M. Konrad, V.
Kopecky and V. Sember) .................................................................................. 242
Modeling of a D. C. Arc Plasma Torch with a Hydrogen-Argon Mixture
as the Working Gas (Xi Chen, Peng Han and Lan Yu) ................................ 267
The Modified Balance Method of Calculating Characteristics of Near-
Anode Processes (Yu.V. Bruevich, I.P. Nazarenko and I.G. Panevin) ......... 280
Application of Heat Diffusion Theory to Cold Electrode Erosion for Nonsta-
tionary Arc Spots (A. Marotta and L.I. Sharakhovsky) .............................. 291
Emission Current Density and Electronic Work Function of Metal to
Plasma (H.Ts. Zayatuev) ................................................................................. 310
Erosion of Multi-arc Cathodes (A.N. Timoshevkii and V.S. Ponkratov) .......... 319
Influence of Three-Phase Electric Arcs, Burning on Oxide Melts, on
Quality of Electric Power (U.B. Ashimov, E.A. Abdrachmanov, G.D.
Manapova and T.S. Maldybaeva) ................................................................... 326
Modelling of Flow Stabilization by the Swirl of a Peripheral Flow as
Applied to Plasma Reactors (E.P. Volchkov, V.P. Lebedev, V.I. Terekhov and
N.E. Shishkin) .................................................................................................. 335
Low-density Plasma Jets: Production and Investigation (L.I. Kuznetsov and
V.N. Yarygin) .................................................................................................... 352
Effect of Arc Current Modulation on Thermal Plasma Flow in Plasma Torches
(F.A. Salyanov)................................................................................................. 365
1
Electric Arc Generators of Thermal Plasma: Review
Electric Arc Generators of Thermal
Plasma: Review
M. F. Zhukov
Institute of Theoretical and Applied Mechanics, Siberian Branch, Russian Academy
of Sciences, Novosi bi rsk, Russi a
INTRODUCTION
The main electrical and physical processes in a discharge chamber of
a linear plasma torch are reviewed. The classification of linear plasma
torches is proposed. The diagrams of the plasma torches for technological
processes and plasma chemical reactors as well as their problems are
discussed.
1. MAIN ELECTRICAL AND PHYSICAL PROCESSES IN THE
DISCHARGE CHAMBER O. DC PLASMA TORCHES [1]
The dynamics of electric arc plasma includes a complex physical
phenomenon, occurring in the discharge chamber of the plasma torch,
which can be regarded as both thermal and an electric and physical device.
The electric arc is influenced by different factors: the gas flow, inner
and outer magnetic fields, construction elements of the chamber which
confine the arc, the elements being often under electric voltage, differing
from that of the arc. Thus, there is a strong need for deep understanding
of a wide spectrum of the processes taking place in the discharge chamber
in order to develop highly effective plasma torches.
1.1. Dynamics of the gas flow in the cylidrical channel and
structure of the arc column [2]
Figure 1 presents the principal scheme of a linear plasma torch with
a long discharge chamber and the qualitative distribution picture of its
energetic characteristics along the z axis, i.e. the specific energy generation
of arc EI and the full specific heat flux per unit length of the chamber
wall Q .
The arc burns between two electrodes, placed at the ends of
2
Thermal Plasma Torches and Technologies
the discharge chamber. Gas G is introduced into the chamber usually
with the circumferential component of velocity W. A cylindrical
chamber may be assembled from individual sections (disks) or it
can be solid. One can observe three typical zones in the structure
of the flow and the discharge. In the first zone, i. e. initial area,
the arc 2 is stabilised along the channel axis, the gas is heated
intensively along this axis, but inside the boundary of heat layer
3 with low conductivity the thickness of this layer increases in
the direction down the flow. This layer possesses high hydrodynamic
stability and it prevents heat transfer between the cold medium
4 and the zone with the highest temperature and electrical conductivity.
This fact explains a relatively low electric field strength in the
initial area of the channel. At the end of the initial area the tangential
velocity field starts rearranging and this leads to a decrease of
arc stability along the chamber axis. Turbulent boundary layer 1
starts to develop near the discharge chamber wall. There are almost
no transverse pulsations of the arc in the initial area.
As seen from Fig. 1 (sections III), the thermal and near-wall
boundary l ayers st art t o i nt eract very i nt ensi vel y. Due t o t hi s
interaction, the thermal layer is destructed and a large amount of
the hot gas is dispersed in the radial direction, as it is clearly seen
in Fig.2, a,b. Electric arc 5 starts to oscillate chaotically in time
Fig.1. Principal schema of a linear plasma torch and qualitative picture of energy
characteristics distribution along the chamber axis z. 1 boundary layer, 2 arc
zone, 3 thermal layer, 4 cold medium, EI is a specific energy generation, Q is
a specific heat flux to the chamber wall.
3
and space under the influence of turbulent pulsations of the gas
mass (see Fig. 1 and 2).
As is seen from Fig.1, the thermal and near-wall boundary layers
starts to interact very intensively. Due to this interaction the thermal
layer is distructing and a great amount of the hot gas is dispersing in
a radial direction, as it is clearly seen in Fig.2, a,b. Electric arc 5 starts
to oscillate chaotically in time and space under the influence of tur-
bulent pulsations of the gas mass (see Figs.1 and 2).
The initial zone of the third area is characterised by stability
of the arc field strength E (Fig.1). The heat flux on the chamber
wall is determined mainly by the arc and high-temperature gas radiation.
The first zone length (gasair, W0) is usually not longer than
12-15 diameters of the cylindrical discharge chamber. The transient
zone is characterised by increase of the technical voltage of the
electric field, defined by the ratio between the potential difference
of two diskprobes and the distance between mean cross-sections
of the disks. This increase is connected with heat transfer intensification
between the arc and the gas, non-correspondence of the real length
of the arc to the measurement region length due to curves and splitting
of the arc into several conducting channels, and due to other causes.
A radiant heat flux to the chamber wall starts to be accompanied
by a heat flux, caused by convective heat transfer, intensively growing
downflow.
Growing on the transient zone, the electric field voltage reaches
its maximum value E before the third area, i.e. the developed turbulent
flow region, and remains stable downflow (in case of a constant
flow rate along the channel). At a large length of the discharge channel
a b
Fig.2. Toeppler (a) and Schlieren (b) pictures of the arc operating in the air jet,
flowing through a nozzle into the submerged space: 1 the boundary of a jet nucleus,
2 electric arc column, 3 thermal gas layer, 4 flow core.
4
with the developed turbulent flow and at the lack of measures aimed
at decreasing the heat flux on the wall, the heat flux, starting in
some channel section, approaches the value of specific energy released
by the arc, i.e. local heat efficiency approaches zero.
1.2. Shunting [1]
Shunting, i.e. electric breakdown 2 between arc 1 and plasma torch
chamber wall (Fig.3a), is the most typical electric process when
an arc burns in a cylindrical chamber.
Let us investigate the qualitative picture of arc shunting in an exit
electrode of a single-chamber plasma torch. Let us assume that at time
moment t
1
the arc is occupying the position ABC. Under the influence
of hydrodynamic and electrodynamic forces, the radial section of the
arc AB drifts into the flow direction, and as a result of this the arc length
and voltage increase and they are connected together by the ratio
+ =
) (
0
) (
t l
e
dl l E V U
,
where U is the sum of near-electrode potential drops, E(l) is the electric
field strength, l(t) is the arc length at time t.
Fig.3. (a) - The scheme of arc shunting in dc plasma torch with a cylindrical exit
electrode: 1 - the arc, 2 - large-scale shunting, 3 and 4 - small-scale shunting;
(b) - qualitative picture of a break-down generation between the arc and the plasma
torch electrode.
5
In electric arc generators
V E l dl
e
l t
<<
( )
( )
0
. To explain the shunting
process let us investigate the simplest case, when E(l)=const.
We also assume that the exit electrode potential is equal to zero,
and the tip point (point C) of the face electrode forms the origin of
the co-ordinate z. Then the arc potential disribution along the z axis
for the time t
1
qualitatively corresponds to curve 1 (Fig.3, b). The
voltage U, necessary for a breakdown changes along the z axis in
correspondence to curve 3. The voltage decreases downflow due to
the increase of the mean mass temperature. There is a potential difference
between an arbitrarily chosen point M on the arc column with co-
ordinate z and the electrode surface
U(z)=U(t) E
.
z.
Under the influence of the applied potential difference between the
arc and the wall a breakdown may happen when
U(z) U*.
At time moment t
1
the necessary condition is not fulfilled in any channel
section. In some consequent time moment t
2
the arc may occupy po-
sition ABC, at which curve 2 for the arc potential distribution along
the axis and curve 3 contact at point K. In this channel section DE value
DU(z) = U*, and this means that a breakdown will run between the electrode
and the arc column and this leads to the formation of a radial discharge.
With the appearance of a new channel the old channel ABB disappears.
The newly formed radial arc region will be drifted by the flow and the
shunting process will be repeated.
Shunting leads to pulsations of the arc voltage, its length, temperature
and velocity of the gas flow, and other parameters of the arc and the
plasma torch; it causes the formation of a falling volt-ampere characteristics
of the arc. Small-scale shunting 4 between the near-wall arc and the
electrode surface (Fig.3, a) mainly determines the erosion rate for the
electrode material. Small-scale shunting includes also electric breakdown
arc-to-arc 3, which was formed in the arc loop and indirectly also influenced
the rate of material erosion. The small-scale shunting initiates additional
pulsations of the arc voltage, the amplitude and frequency of which
approximately differ by the factor of magnitude of the amplitude and
frequency, initiated by the large-scale shunting.
6
Figure 4 presents a classic oscillogram of a temporary change of the
arc voltage, caused by shunting, the arc burning in the linear plasma
torch with gas-vortex stabilisation. The oscillogram shows not only large-
scale voltage pulsations 1, but also smaller pulsations 2, determined by
small-scale shunting. Specific arc erosion (or the erosion rate) is de-
termined greatly by the time during which the arc spot is motionless
on the electrode surface. The detectable physical-mechanical changes
are going on in the metal during such a relatively short time period
(10
4
s) only if a large specific heat flux flows into the electrode body
in the spot area. The large temperature gradients in the cooling elec-
trode body initiate a strong thermoelastic stress, which leads to crack
formation in the metal which then leads to the formation of gas cavi-
ties, decreasing of heat conductivity; thus, the temperature of the metal
surface layer increases and reaches the melting point.
As mentioned above, a set of fundamental problems appears in this
case. They include:
(1) clarification of the physical processes which determine continuous
motion of the near-electrode arc region and accompanying motion of
the root spot on the electrode surface without melting of the spot;
(2) design and application of alloys and composite materials which
would be less sensitive to the formation of dislocations at large short-
term specific heat loads.
1.3. Classification of the linear scheme plasma torches [3]
Work on the development of devices for thermal plasma generation has
been continuing for a long time and information concerning the peculiarities
Fig.4. The classic oscillogram of arc voltage changes in a linear plasma torch.
7
of their work has been accumulated; from time to time, a more detailed
classification is needed. This helps to estimate objectively the existing
results and outline new ways of developing more effective systems applied
to new technologies.
Let us analyse the classification of only linear plasma torches, which
represent a wide range of plasma generators regading the consumed power,
variations of applied working gases and pressure ranges.
Knowledge of fundamental physical processes, taking place in the
discharge chamber of a plasma torch, helps to suggest a simple clas-
sification and to reduce multiple constructive solutions to three prin-
cipal classes.
The first class consists of the plasma torch with a tubular exit electrode
and self-aligning arc length. This type of torch is used widely in the
industry and scientific research work. The voltampere characteristics
(VAC) of these plasma torches are of the drooping nature (Fig.5, curve
1). The average arc length l is a function of current value, chamber
diameter, gas flow rate, pressure and it depends on the working gas and
on the polarity of the exit electrode. The large-scale shunting, taking
place in the transient zone of the discharge channel, forms the mechanism
for arc length changes.
The second class of the plasma torches is characterised by the fact
that the average arc length l is constant in a relatively wide range of
current changes (I
a
< I < I
1
), while the other parameters, mentioned above,
are constant, and this length is always less than that for the self-aligning
arc I
a
< I
sa
. The VAC is U-shaped (Fig.5, curve 2). Without taking special
measures, the value of U is limited by the contact point of two VAC
lines (point a). At I > I
a
, curve 1 merges with curve 2 into a single
VAC of the arc, because arc shunting in this case takes place in a channel
with diameter d
2
.
There are several technical solutions, providing the constant aver-
age arc length. One is a peculiarity of a broken gas flow behind a step,
formed by the step construction of the exit electrode, that consists of
two cylinders with different diameters, the diameter of the exit part of
the electrode d
3
, being larger than d
2
. This forms the basis of one of
technical solutions, which draws the greatest attention of the engineers.
We should mention that the VAC of the arc I is positioned below the
VAC of the arc with self-aligning length. These plasma torches are working
in a stable manner without additional resistance in the circuit on the
rising branch of the arc VAC.
Recently, special attention has been paid to the third type of linear
plasma torches, in which the average arc length in non-changeable, but
longer than that of the self-aligning arc [4]. The VAC of this arc (Fig.5,
8
curve 3) is positioned above the VACs of the two types of torch mentioned
previously. The increase of the average arc length is achieved by in-
troduction of the interelectrode insert (IEI) between the anode and the
cathode, the length of which is larger than l
sa
. The insert may be solid,
porous, with any gas injection through porous structure or sectional with/
without gas injection into intersection gaps.
The development of a block-module plasma torch with the IEI, which
maintains power from 1 to 35 MW at moderate arc current, was a good
engineering solution. Each plasma torch block is produced as an el-
ement of a separate sectional plasma torch.
A plasma torch with a gas-vortex IEI can be regarded as an inter-
esting and promising solution. In this plasma torch, a heat flux on the
discharge chamber wall is determined by the radial heat transfer from
the arc.
The proposed classification greatly simplifies the choice of an op-
timum circuit of a plasma torch for this or that plasma technological
process, helps to stress the attention on the choice of power supply, con-
structional materials, providing serviceability of plasma torch elements
in the conditions of large specific heat fluxes beforehand, assists in the
choice of a rational scheme for automatization of technological proc-
esses, etc.
Fig.5. The voltampere characteristics of the arc operating in linear plasma torches
of different types: 1 with a tubular exit electrode, 2 with a step electrode,
3 - plasma torch with an interelectrode insert.
9
2. TWO-JET PLASMA TORCH [3]
The rapid growth of the plasma torch technologies requires new, more
stringent demands for plasma torch parameters such as life-time, operationak
stability of the electric arc, the possibility to use a wide range of plasma-
forming (working) gases, relatively simple construct-ion of generators
and launching systems for electric arc initiation. There are technological
processes in which it is more expedient and energetically profitable to
treat surfaces directly by the electric arc. In this case, the intensity of
heat fluxes on treated surfaces, depending on conditions of arc interaction
with the surface, varies from 1 10
7
to 25 10
7
W/m
2
.
Among many types of electric arc plasma torches one should men-
tion two-jet ones. In these torches, the large part of the arc burns in
the open space. In the general case, the operating regime, including the
arc VAC, is determined not only by current values, gas flow rate and
its properties, but also by electrode disposition.
Some design peculiarities form the basis of this type of two-jet plasma
torches. They are connected with the necessity to exclude tungsten as
the cathode material, which limits ranges of working gases.
Figure 6 presents the scheme of an experimental set-up with a two-
jet plasma torch. Two identical blocks, i.e. the cathode and the anode
blocks, form the constructional basis of this plasma torch. Each of them
consists of a cylindrical copper electrode, a diaphragm, electrically isolated
from the electrode, and two vortex chambers for gas injection.
The arc voltampere characteristics (Fig.7a,b) are slightly falling in
a wide range of current changes for all studied types of nozzles with
different diameters d
2
and distances a between nozzles. The existence
Fig.6. A general scheme of an experimental set-up with a two-jet plasma torch and
a power supply system.
10
of multiple independent parameters, which influence the arc strength (I,
G, d
2
, internozzle distance a and the angle between electrode axes), simplifies
the choice of a power supply system and optimisation of electric char-
acteristics of the plasma torch. For the case in which a = 90
o
(Fig.7,
b), arc voltage U is calculated using formula:
U =2 10
3
[I
2
/(G
1
+G
2
)d
2
]
0.20
[(G
1
+G
2
)/d
2
]
0.25
(Pd
2
)
0.35
(2+2a/l
1
)(G
1
+G
2
)/d.
Here l
1
is the electrode length. Due to the fact that more than 2/3 of
the arc length is situated outside the cathode or anode block, the heat
efficiency of the blocks is sufficiently high (~0.9).
Fig.7. Voltampere characteristics of the arc: a d
1
= 30 mm, d
2
= 25 mm, a = 60
mm, G = 12 g/s; b d
1
= 70 mm, d
2
= 25 mm, a = 110 mm, G = 40 g/s.
a
c
b
11
Industrial tests of this plasma torch (lasting more than 60 hours) show
that at the axis scanning of arc spots with frequency of 3 times per minute
the specific erosion of the electrodes turns to be different. For the anode,
it is an order of magnitude less than that for the cathode, and is equal
to 610
11
kg/C at current I = 200 A. Accounting for the possibility of
changing the polarity of the electrodes, one may guarantee the life-time
for continuous work of a plasma torch as not less than 400 hours.
The above scheme of a two-jet plasma torch shows the high reliability
of its operation, simple arc ignition, and capacity control.
3. HEAT TRANS.ER IN A CYLINDRICAL ARC CHAMBER [2]
Let us investigate the heat fluxes into the wall of the electric discharge
chamber of the most promising plasma torch with an interelectrode insert
(IEI). In the initial region of the channel (z=z/d < 12) the heat fluxes
are mainly determined by radiation of the arc and gas (Fig.8, a).
The heat fluxes q and
Q
rapidly increase downwards the channel;
Fig.8. Heat losses distribution along the electric-arc chamber (a) and the measuring
section (b) on the developed turbulent area: I = 120 A, (15) m
p
= 0; 0.21; 0.63;
1.0; 1.51, correspondingly; 6 the level of radiant heat flux.
12
one can see the influence of an intersection gas g
f
injection on this
parameter. The increase of the heat flux on the wall is connected with
the growth of mean-mass gas temperature and the increase of intensity
of the convective heat exchange between the gas and the wall. Figure
8, b shows the heat flux distribution
Q
on the wall of the measured
section, maintained in the channel in the developed turbulent region,
at the different values of shielding gas injection m
p
=(u)
p
/(u)
o
through
a slot. Without injection of a shielding gas (m
p
= 0), the changed heat
flux on the wall coincides with the one calculated for turbulent heat
exchange (shaded line I). Injection of even a small amount of shielding
gas decreases heat fluxes (curves 25).
The efficiency of chamber wall protection by a gas screen is determined
by the dimensionless correlation
= (q
k
q
wp
)/q
k
,
which characterises the ratio of the convective heat flux on the wall,
captured by the gas screen (q
k
q
kp
) to the heat flux q
k
= St(U)
o
(h
o
-h
w
) at the lack of the protective screen [4]. Here h
o
is mean-mass
enthalpy in the investigated channel section; h
w
is the gas enthalpy at
the wall temperature.
Experimental results were generalised outside the zone of the absolute
screen (z
1
> z
1
1
). Figure 9 presents the experimental data, sufficiently well
generalised by the equation of the following type:
' = (1+0. 24K)
0. 8
(1+K
2
)
0.14
,
Fig.9. Generalisation of experimental data on the efficiency of the gas screen.
13
where K= (z
1
z
1
1
)Re
s
0.25
/m
p
S; Re
s
= (u)
s
S/
o
. Here
o
(measured in
Ns/m
2
) is the gas viscosity at the mean-mass temperature of the flow,
parameter m
p
was varied from 0.2 to 1.5; the slot width was changed
from 1.3 10
3
to 4.210
3
m. The mean square deviation curve (continuous
line) does not exceed 2%.
The first factor in the formula = f(K) determines the efficiency
of the gas screen on the plate [4] at the gas injection in relation to the
plate, situated at angle = 0. The shaded line in Fig.9 demonstrates
this dependence.
4. PLASMA TORCHES-REACTORS
4.1. Multijet reactor [1,5,6]
The use of plasma torches in chemical industry and metallurgy, in
hydrodynamic investigations and for other purposes reveals some unsolved
problems. They are closely connected with the task to create powerful
and economic electric arc reactors with long life-time, which meet the
demands, such as the uniform velocity and temperature fields along the
section. Although the power of a single plasma torch exceeds many thousands
kW, it is usually released in a short period of time when using high currents,
which greatly complicates the possibility of increasing the electrode durability
and, consequently, the service life.
It is possible to solve the problem of extending the life-time by the
development of a multijet reactor with a common flow mixing cham-
ber, in which one portion of the working gas travels through plasma
torches and the other portion, in the case of necessity, goes directly to
the reactor. Such kinematic scheme of gas supply will simplify the control
over all parameteres, including the increase of capacity due to the in-
crease of the number of plasma torches.
A three-jet reactor (Fig.10) with a total capacity of 300 kW, inner
diameter D = 0.115 m and length L = 0.23 and 0.46 m was investi-
gated. Cold air was injected through the base of the cylindrical chamber.
The reactor walls were cooled by water. The pressure in the reactor was
assumed to be atmospheric.
The temperature fields, shown in Fig.11, are a good illustration of
the mixing efficiency of cold and hot gases. The optimum chamber length
is evidently equal to L = L/D = 2.
In addition to the parameter of mixing efficiency, we should mention
another important characteristic, i.e. heat efficiency, determined by
the value of heat losses through the wall. The experiment shows that
heat efficiency is determined with satisfactory accuracy by the
approximation equation
14
h h h = - =
-
( ) / Re
. .
1 228
0 5 0 8
L
.
The agreement between the calculations, using this formula, and the
experiment is shown in Fig.12, where is the right side of this formula.
4.2. Design features of multijet reactors for producing ultrafine
powders [5]
In the group of the reactors described in [7], the multijet straight
reactors [8,9], in which the plasma torches are maintained at different
angles of their axes to the reactor axis, i.e. from 15 to 90 (Fig.13),
have the widest technological possibilities.
Fig.10. A three-jet reactor: 1 AC plasma torch; 2 mixing chamber; 3 input
of cold gas.
Fig.11. Gas temperature field at the exit of the reactor: L = 2, G
2
= 30 g/s; 1 -
G
1
= 60 g/s, T
2
/T
1
= 9.3; 2 G
1
= 60 g/s, T
2
/T
1
= 11.3; 3 G
1
= 3 g/s, T
2
/T
1
= 10.5.
15
Thus, the analysis of interaction of the slope angle of plasma jets
with the reactor characteristics is important when developing the rec-
ommendations for designing the reactors of this type. Some drawbacks
were revealed in studies of the reactors with the plasma torch slope angle
of 90
, i.e. with radial injection of plasma jets into the reactor. They
include:
1) upward gas flows (at the vertical location of the chamber axis)
above the collision area of the plasma jets lead to the formation of a
vault C from the caked initial material, which in time overlaps the
cross-section of the chamber and greatly reduces the duration of con-
tinuous work (Fig.13, a);
2) existence of upward flows in the reactor makes it practically impossible
to implement effective treatment of highly dispersed material with submicron
size particles and also low-melting point materials;
3) the life-time of the mixing chamber, made of stainless steel (type
Cr18Ni10Ti), is limited to 350400 hours due to erosion near the ex-
its of the plasma torch nozzles;
4) substitution of steel by copper increases the life-time of the chamber,
but also increases heat losses in it by 4050%.
The analysis of the influence of the plasma jet slope angle on the
main characteristics of the three-jet plasma torch shows that the op-
timum value is in the range from 30
to 45
[5].
Using the experimental results, the following dependence was sug-
gested for the calculation of convective heat exchange of high temperature
flows with the reactor walls:
St = A
fx
B
fx
Re Pr
. -0 05
.
Fig.12. Correspondence between experimentally defined efficiency and calculated
one.
16
5. SPECI.IC EROSION O. ELECTRODES
Electrode erosion is determined by near-electrode processes, the nature
of the working gas, electrode materials and their crystal structure,
development of dislocations, formed at thermal stresses in metals, and
other processes.
5.1. Recirculation of the cathode ions
This fundamental phenomenon was experimentally discovered in 1982
1985 [1,10] during investigations into a tungsten rod electrode. The essence
of it lays in the fact that the evaporated tungsten is ionised in the zone
of the contracting arc spot; tungsten ions are returned to the metal surface
by the electric field and partial reconstruction of the cathode takes place.
Calculations show that the optimum length of the cylindrical part of
the cathode l
c
(Fig.14) should be equal to zero.
where A and B are the coefficients accounting for the influence of the
reactor wall lining material. The values A and B are, correspondingly,
for a non-thermoinsulated channel 0.781, (0.412); and at lining: by
zirconium dioxide 0.692, (0.426); by chromium oxide 0.643, (0.424);
by silicon dioxide 0.703, (0.422); subscript f indicates that the conditions
were determined from the mean-mass temperature, subscript x defines
the parameters corresponding to the axis co-ordinate. The observed reactor
possesses sufficiently good kinematic and dynamic characteristics and
heat efficiency. Moreover, when powder materials are used, a vault from
caked material does not form in the reactor (Fig.13, b).
a b
Fig.13. Pictures of reactors: axes of plasma jets are perpendicular to the axis of
the reactor chamber (a) and have a slope of 30(b).
17
This principle forms a basis for design and effective work of highly
effective cathode blocks with tungsten rods with zero projection [1,10-
12]. They have very small values of specific erosion G , close to
110
13
kg/C at the atmospheric pressure in inert gase in respect to tungsten
(argon, nitrogen, hydrogen) and at the current value up to 1 kA (Fig.14).
The scatter of experimental data (shaded area) is wide due to many causes.
The main of them are: non-qualitative stuffing of tungsten into the cooled
copper holder and, as a consequence, poor thermal contact between copper
and tungsten; non-satisfactory water cooling of the cathode block, pulsations
of an arc spot along the surface of the tungsten electrode due to bad
arrangement of gas injection into the vortex chamber, non-coaxiality of
the tungsten rod and a gas vortex.
The same Fig.14 shows the experimental values of the specific erosion
of tungsten cylindrical rods at l
c
> 0 [12]. Curve 1 characterises de-
pendence G = f (I) for cases when several tungsten rods are pressed
into the body of a copper pipe, as it is shown in Fig.14, and the ra-
dial end of the arc splits into several current-conducting channels. Their
number depends on the current. The arc elements are rested on the basis
of the tungsten rod inserts by their ends. But even in this case specific
erosion of such electrode is higher than that in the case l
c
= 0.
If only one tungsten rod with l
c
> 0 is pressed into a face copper
plate, cooled by water, then its specific erosion will be much higher (curve
2). Both curves, as it is seen in the figure, lay higher than the shaded
zone and this is the evidence that the cathode block with l
c
= 0 is re-
ally optimum. At the present time, there is a task of further optimisation
of the cathode block geometry to decrease specific erosion.
Fig.14. Specific erosion of tungsten rod cathodes in inert gases versus arc current.
18
The significant role in solving the problem of the cathode life-time
was played by the discovery of the phenomenon of spontaneous arc splitting.
Investigations have led to an original solution, i.e. simulation of the
conditions, at which spontaneous splitting of the cathode closing region
of the arc and the cathode spot root to the thermoemission inserts in
the cylindrical copper electrode take place [6]. At the increase of to-
tal current, the current in single current-conducting channels remains
the same (and even decreases) after the splitting, independent of their
number. In this case, as the experiment shows, (Fig.15, curve 7), specific
erosion does not change with the increase of current for thermochemical
cathodes, or decreases a little. The operation of the elements of the radial
regions of the arc is stable without ballast resistors in the electric power
supply. Specific erosion of the thermochemical insert at total current
1000 A is not higher than 110
11
kg/C.
5.2. Diffusive anode arc attachment to the surface of a copper
exit electrode
First of all, we should mention the existence of such an arc attachment
to the inner surface of the copper exit pipe electrodeanode shielded
by argon, injected through a slot between the anode and the IEI, with
the latter placed maintained between the cathode and the anode.
Any technical gas may be the working gas. If we use air as the working
gas and argon as shielding gas (25% of total flow rate), specific erosion
of the copper anode becomes approximately equal to 610
12
kg/C. There
is no theory of the diffusive root of the arc. It is possible that the
Fig.15. Specific erosion of the cathode blocks.
19
formation of a non-contracted arc root is connected with uniform
microshunting of the arc at any point of the cylindrical surface. However,
this hypothesis needs a detailed study.
References
1. M.F. Zhukov, A.S. Koroteev and B.A. Uryukov, Applied Dynamics of
Thermal Plasma. Novosibirsk, Nauka Publishing, 1975 (in Russian).
2. M.F. Zhukov, A.S. Anshakov, I.M. Zasypkin, et al., Electric-Arc Generators
With Interelectrode Inserts. Novosibirsk, Nauka Publishing, 1985 (in
Russian).
3. Plasma Torches. Researches and Problems, Ed.: M.F.Zhukov. Novosibirsk,
Institute of Thermophysics SB RAS, 1995 (in Russian).
4. S.S. Kutateladze, Fundamentals of Heat Transfer Theory. Novosibirsk,
Nauka Publishing, 1970 (in Russian).
5. G.V. Galevskii, M.F. Zhukov, et al., Hydrodynamics and Thermotechnical
Characteristics of Three-Jet Direct-Flowing Reactor for the High-Tem-
perature Synthesis of Ultra-Disperse Materials. Novosibirsk, 1990. Preprint
USSR Academy of Sciences, Siberian Branch, Institute of Thermophysics,
No.226-90 (in Russian).
6. O.Ya. Novikov, P.I. Tamkivi, A.N. Timoshevsky, et al. Multiarc Systems.
Novosibirsk, Institute of Thermophysics, Siberian Branch, USSR Academy
of Sciences, 1988 (in Russian).
7. S.A. Panfilov and E.B. Grinshpun, Some Peculiarities of Apparatus
Arrangements of PlasmaProcesses of Ultradisperse Powder Materials.
Moscow. Academy of Sciences of USSR, 1981.
8. P.N. Tsybulev, V.A. Pop, V.D. Parkhomenko, et al., Plasma Processes
in Chemical Industry. Chernogolovka, Institute of Chemical Physics,
USSR Academy of Sciences, 1987 (in Russian).
9. A.V. Bolotov, A.N. Kolesnikov, et al., Plasma Processes in Chemical
Industry. Chernogolovka, Institute of Chemical Physics, USSR Academy
of Sciences, 1987 (in Russian).
10. M.F. Zhukov, A.V. Pustogarov, G.-N.B. Dandaron, et al., Thermochemical
Cathodes. Novosibirsk, Institute of Thermophysics, Siberian Branch,
USSR Academy of Sciences, 1985 (in Russian).
11. G.Yu. Dautov, V.L. Dzyuba and I.N. Karp, Plasma Torches With Stabilized
Eletric Arc. Kiev, Naukova Dumka, 1984 (in Russian)
12. A.V. Pustogarov, V.I. Zavidev, G.R. Zhienbekov, et al., Thermophysics
of High Temperatures, 1985, 23, No.5.
20
Trends in Thermal Plasma Technology
L F Pfender
Department of Mechanical Engineering and ERC for Plasma-Aided Manufacturing,
University of Minnesota, Minneapolis, MN 55455
1. INTRODUCTION
Thermal plasma technology has passed through a gradual transition
stage from primarily space-related activities in the sixties to a more
and more materials-oriented focus in the eighties and nineties. Space-
related needs provided a strong impetus for basic thermal plasma research
and developments as, for example, for the development of plasma
torches covering power levels ranging from 1 kW to more than 10
MW. Research specifically geared towards an understanding of plasma/
particulate interaction and the chemistry in thermal plasmas, however,
did not commence until the early eighties, although some successful
applications were already in existence at that time as, for example,
in the area of arc welding, arc cutting, and in some specific areas
of plasma synthesis of fine powders. At this point it should be emphasized
that the successful development of arc circuit breakers and of arc
lamps contributed immensely to our basic understanding of electric
arcs.
In addition to these more conventional applications, thermal plasma
technology covers today a wide spectrum of applications as well as
new developments which may be classified as (1) thermal plasma coating
techniques, including plasma spraying, wire arc spraying, and plasma
chemical vapor deposition (TPCVD); (2) thermal plasma synthesis
of fine powders, in particular, powders in the nanometer size range;
(3) thermal plasma waste destruction, in particular, of toxic waste
materials; (4) thermal plasma densification of powders; (5) thermal
plasma metallurgy, including melting and re-melting applications in
large furnaces; and (6) thermal plasma extractive metallurgy.
In this overview, an attempt will be made to assess trends in this
field and to speculate on the role which thermal plasma technology
might play in the much broader context of material science and en-
gineering as we approach the next century.
21
Because of space limitations, only the first two of the previously
mentioned thermal plasma technologies will be covered in this overview.
2. THERMAL PLASMA COATING TECHNOLOGIES
This section will include plasma spraying, wire arc spraying, and
thermal plasma chemical vapor deposition (TPCVD). It will primarily
consider d.c, arcs as the plasma source.
2.1. Plasma spraying
Over the past 35 years, plasma spraying has become a well-established
and widely used technology with applications ranging from corrosion,
temperature-, and abrasion-resistant coatings to the production of
monolithic and near-net shapes of metallic and ceramic parts. Powders
of glassy metals can be plasma sprayed without changing their
amorphous characteristics and, as demonstrated in recent years,
superconductive materials can be deposited by the plasma spray process.
Besides the most common atmospheric pressure plasma spray process
in ambient air (APS), other plasma spray processes have been de-
veloped, including spraying at low pressures (LPPS), at supersonic
velocities, under controlled ambient conditions (for example, in argon)
and even under water. The design of plasma spray torches for the
various plasma spray processes has been essentially the same, based
on producing a plasma jet by a d.c. arc operated between a stick-
type cathode and a nozzle-shaped anode as shown schematically in
Fig.1. Recent torch developments, however, have been exploring central
injection of the powder particles into the plasma, shrouding of the
plasma jet, and a combination of shrouding with anti-vortex flow in
order to improve deposition efficiency and quality of the coatings.
Among recent innovations, automation and robotics applied to the
plasma spray process are probably the most important new devel-
opment in this rapidly growing field.
Fig. 1: Schematic of the plasma spray process with d.c. plasma torch.
22
In spite of these impressive developments, some of the underly-
ing fundamentals of the plasma spray process are still poorly un-
derstood. This applies to the characteristics of the plasma jet as well
as to the interaction of powder particles with the plasma and also
to the formation of the coating on a substrate.
For the case of atmospheric plasma spraying (APS), the fluid dynamics
of the plasma jet leads to the development of turbulence associated
with strong entrainment of ambient gas into the plasma jet [1,2] as
shown schematically in Fig. 2. This behavior of a plasma jet as sketched
in Fig. 2 has been confirmed by shadowgraphs and CARS spectroscopy
[3], by conditional sampling experiments [4], and by probe sampling
of the plasma jet [5].
Superimposed to these fluid dynamic effects are a surging and whip-
ping motion commonly seen in time-resolved photographs of plasma
jets (Fig. 3). This motion is caused by axial and circumferential motion
of the anode arc root within the anode nozzle giving rise to arc in-
stabilities [6,7]. Correlations between these arc instabilities and arc
voltage acoustic, and light emission fluctuations [2] confirm that the
observed plasma jet fluctuations (Fig. 3) are to a large extent due
to arc instabilities.
Fig. 2: Main regions of a transitional plasma torch.
23
Recent studies of a commercial spray torch, considering both
vortex and straight flow of the plasma gas, revealed fluctuations
in the frequency range from 2 to 6 kHz and a strong dependence
of the arc behavior on the plasma gas composition (Ar/H
2
, Ar/
He, Ar/ N
2
mi xt ures) [8]. These fl uct uat i ons affect bot h t orch
performance (anode lifetime) and coating quality. Attempts have
been made to determine the motion of the arc root in the anode
nozzle by using magnetic probes [9] and also to calculate the anode
arc root position in the anode nozzle [10] using Steenbecks minimum
principle [11, 12]. Experimental results are in reasonable agree-
ment with analytical predictions.
The previously discussed fluctuations of the plasma jet may lead,
in extreme cases, to situations where the injected powder parti-
cles miss a substantial fraction of the hot plasma jet [13]. This
may lead to a severe degradation of the quality of the coatings
due to unmelted and only partially melted powder particles in the
coatings.
Attempts to model the plasma spray process require, as a first step,
a comprehensive model of the plasma jet. Unfortunately, modeling
of plasma jets faces several obstacles as indicated by experimental
observations. Diagnostics of plasma jets using emission spectroscopy,
laser scattering (Rayleigh and collective Thomsen scattering), laser
Fig. 3. Short time exposures (50 ns) of a d.c plasmajet.
24
Doppler anemometry, enthalpy probes, and mass spectrometry (for
references, see Ref. 14) indicates that strong deviations from Local
Thermodynamic Equilibrium (LTE) may prevail over almost the entire
plasma jet volume. This fact, combined with the previously mentioned
fluctuations of typical plasma spray jets imposes severe difficulties
on modeling attempts. A realistic model must also take the large-
scale entrainment of ambient gas into account. Conventional turbulence
models cannot predict the intermittency and the unmixing phenom-
enon of turbulent flows. These models neglect the spottiness

or
fragmentariness of real turbulent flows which have been observed
in many situations. Only a multiphase model for turbulent flow can
reproduce these effects [15].
The model which is based on Spaldings approach [15-17], treats
the plasma jet as a two-fluid mixture consisting of hot, out-moving
fragments and cold, in-moving fragments as shown schematically in
Fig. 4.
The governing equations include the transport equations for mass,
momentum, and energy for two different fluid parcels (in-moving parcels
and out-moving parcels). Auxiliary relations that govern the physical
phenomena of the interfluid mass, momentum, and energy exchange
are used together with a description of the mechanisms that control
the growth or iminution of the fragment size. The results may be
presented in conditional- and unconditional-averaged forms and compared
with experimental results from enthalpy-probe measurements [18,19].
Since the quality of a plasma sprayed coating depends heavily on
Fig. 4. Schematic of the two-fluid model and boundaries for the computational domain.
25
the properties of particulates, including velocity, temperature, degree
of melting, and their statistical distributions, considerable efforts have
been made to develop predictive capabilities for the particle behavior
in plasmas. An initial simplified approach in which particle effects
have been decoupled from the plasma flow [20] has now been ex-
tended to iterative techniques [21,22] to include the effects of the
particle cloud on the plasma flow, and to the stochastic particle spray
model incorporated into a computer code known as the LAVA code
[23]. The LAVA code allows to predict particle trajectories and thermal
histories including melting which is calculated simultaneously with
the motion of the gas in a fully self-consistent manner, including turbulent
dispersion of particles. Combined with transient, multicomponent, and
non-LTE capabilities, the stochastic particle spray model designates
LAVA as a unique comprehensive computational model.
2.1.1. Summary of present R&D efforts in plasma spraying
The focus of present research efforts seems to be on both diagnostics
and modeling with the goal of:
a.) improving and optimizing plasma torch performance. This includes
studies of plasma jet instabilities caused by the fluid dynamics and
the associated arc behavior in typical plasma spray torches. Closely
related to the arc behavior are problems of electrode erosion.
A new approach for modeling of turbulent plasma jets has been
proposed using a two-fluid description of the plasma and its sur-
roundings. Comparisons with experimental data have been ham-
pered by fluctuations of plasma jets which, so far, have not been
included in corresponding models. Recently, attempts have been
reported to incorporate fluctuations in such models [24].
b. )Another major objective of present research efforts is the control
of the particle spray pattern and the associated optimization of
the deposition efficiency. This aspect is of particular concern when
spraying of expensive powder materials is considered. Besides
deposition efficiency, the quality of coatings may also be strongly
affected by the spray pattern.
c.) Some of the most recent research efforts have been concentrating
on coating formation on a substrate which includes splat formation,
solidification of splats, splat microstructure, cohesion of neighboring
splats, adhesion of the coating to the substrate, and the control
of porosity and of unmelts in the coating [25]. It has been found
that the substrate temperature may be the governing parameter in
terms of coating adhesion [25]. Residual stresses which are in-
herent to this coating process may be the determining factor for
26
integrity or failure of a coating. Since complete control of the plasma
spray process is the ultimate goal of all R&D efforts in this field,
the establishment of comprehensive data bases and the selection
of primary control parameters attracted particular attention over
the past years. This does not only apply to the APS, but also to
other plasma spray processes.
Recent developments are primarily concerned with:
a. ) Sensor Development. Such sensors must be robust to toler-
ate the hostile plasma spray environment, but at the same time they
have to be affordable (cost-effective).
b.)Among various control strategies, feedback control of the plasma
jet behavior (enthalpy level and fluctuations of the jet) is consid-
ered to be a viable option.
c.) Control of temperature and velocity of powder particles injected
into the plasma is a challenging problem and many of the leading
plasma spray laboratories all over the world are engaged in soling
this problem.
d.)In-situ control of the coating thickness is another challeng-
ing problem which has not yet been solved.
e.) Substantial advances in the area of robotics have been already
demonstrated for both motion of the spray torch and the substrate
during the coating process.
Over the past years, applications of plasma spraying experienced
a slow, but consistent growth. As this technology further matures to
the level of complete, automated control, a more rapid growth of this
technology can be expected. Because the lack of efficient controls
has been the primary obstacle for the growth of this field.
2.2. Wire arc spraying
Wire arc spraying is an inexpensive coating process, usually restricted
to spraying of metals and alloys. A wide array of substrate materials
can be coated with this process, including ceramics, metals, and plastics.
The applications range from wear resistant coatings, to coatings for
corrosion protection of large scale structures (for example, bridges),
and to the restoration of worn metallic parts. Although similarities
with the plasma spray process exist, there are fewer parameters governing
this process.
Figure 5 shows a schematic of the dual wire arc spray arrange-
ment. The material to be deposited is introduced into the arc in the
form of two wires serving as consumable arc electrodes. A gas jet
across the arc removes molten droplets from the wire tips, atomizes
27
droplets and drives them to the substrate. The coating is formed by
the impact, deformation and rapid solidification of individual mol-
ten droplets on the substrate resulting in a coating structure consisting
of a series of overlapping lamellae. The adhesion of the coating depends
upon the interactions among individual lamellae and between lamellae
and the substrate. The bonding mechanisms of arc sprayed coatings
are still poorly understood and process parameters are still optimized
by empirical methods. Excellent adhesion of coatings is considered
to be one of the most important prerequisites for industrial ap-
plication [26-30].
If the interface adhesion is poor, coating detachment may occur
resulting in premature failure. The bond strength of a coating de-
pends on the extent of both physical and chemical interactions be-
tween the coating and the substrate material and on the microstructure
of the interfacial region. Poor adhesion may be attributed to poor
interfacial interlocking, low degree of metallurgical bonding and high
internal stresses.
Adhesion strength depends to a large degree on particle veloci-
ties. In conventional wire arc spraying, the velocities of particles are
subject to certain limitations, so the coating produced with conventional
primary gas atomization has relatively high porosity and relatively
low bond strength. Secondary gas atomization spraying is a newly
developed approach for achieving more uniform particle size distributions,
more focused spray patterns, higher particle velocities, and improved
coating properties [31].
Fig. 5. Schematic of the double~wire arc spray process.
28
It has been shown that secondary gas sprayed stainless steel coatings
on aluminum substrates reveal higher bond strength than those sprayed
with only primary gas. The following three kinds of bonding mechanisms
are operative in spraying with secondary gas atomization; (I) physical
bonding, (2) mechanical bonding, (3) metallurgical bonding. The improved
bonding is believed to be due to the higher temperatures of the liquid
metal droplets at the instant of impact on the substrate. These higher
temperatures are the result of higher droplet velocities and of reduced
entrainment of cold air, leading to an interdiffusion layer between
the substrate and the coating. The presence of this interdiffusion layer
as verified by elemental analysis of the coating cross section is associated
with the measurement of improved bond strength [31]. Secondary gas
flow can also reduce mixing of atomizing gas with the surrounding
air resulting in less oxide content and less chromium loss (for stainless
steel) in the coating [32]. For a further reduction of the oxide content
in the coating carbon dioxide has been used as the atomizing gas which,
also, reduced the porosity in the coatings [32].
One of the drawbacks of wire arc spraying is the generation of
substantial amounts of metal fumes which pose a human health hazard.
Studies of fume generation as a function of operational parameters
for the related process of arc welding have been reported in the literature
[33-35]. The control of fume generation at the source by modification
of the process would lead to a wider acceptance of wire arc spraying
for industrial applications. In recent experiments, using aluminum
as the wire material, fume generation has been quantified with com-
puterized image processing. The results show that metal evaporation
at the wire tips is the primary source of fumes, and oxidation seems
to enhance metal evaporation. There is a pronounced asymmetry of
fume generation, because of the current concentration (high current
density) at the cathode [36].
2.2.1. Summary of present R&D efforts in wire arc spraying
Present research activities are geared towards improvement of coating
quality by:
a. ) control of the arc,
b.)by optimization of the flow, and
c.) control of the droplet formation.
Coating quality refers primarily to coating adhesion and cohesion,
density (porosity), uniformity, oxidation, and thickness control. It
should be pointed out that the previously mentioned control func-
tions are not independent of each other. Arc and flow control, for
example, are directly coupled to the droplet formation
29
Control of the arc has to be done in conjunction with the power
source, because the characteristics of the power source will affect
the arc behavior. Arc fluctuations which manifest themselves by arc
voltage fluctuations should be minimized and arc extinction must be
avoided [37].
Optimization of the flow includes the primary or atomizing gas
flow (supersonic), the secondary or shrouding flow, the type of gas
being used, and its temperature. The effects of turbulence on air en-
trainment and its reduction by gas shrouding are active research
objectives. In this context, it should be mentioned that modeling of
the wire arc spray process, including flow effects, is another active
research topic.
As previously mentioned, there are substantially fewer parameters
which affect the wire arc spray process compared to the plasma spray
process. Therefore, developments which are already in progress and
which focus on computer control systems of the wire arc spray process,
have a high potential for early success.
2.3. Thermal plasma chemical vapor deposition (TPCVD)
Over the past years, anew technology has been initiated which makes
use of thermal plasmas for chemical vapor deposition of an array
of materials and this process gained prominence by demonstrating
the highest deposition rates for diamond coatings. This emerging
technology will fill the void between the high quality, but extremely
slow coating processes (sputtering, physical vapor deposition) and
Fig. 6 Schematic of diamond TPCVD with a dc plasma torch.
30
the rapid, but difficult to control thermal spray processes.
Because of its unique features, TPCVD may find numerous ap-
plications and some of them are already considered for industrial fab-
rication as, for example, diamond and dense ceramic or supercon-
ducting films. There are indications that TPCVD may play an im-
portant role in the fabrication of nanostructured films. As this emerging
technology further matures, there is no doubt that other applications
will come into the picture.
In the process of TPCVD, a high energy density plasma produces
high density vapor phase precursors for the deposition of relatively
thick films. A typical arrangement for TPCVD is shown in Fig. 6
In this case, a d.c. plasma torch generates a high temperature, high
velocity plasma jet which impinges on a cooled substrate. With tem-
peratures close to the torch nozzle exit exceeding 10
4
K, the precursor
material, which is injected into the plasma, is rapidly vaporized and
due to the high velocities of the plasma jet (in the order of 100 m/
s), accelerated towards the substrate. In front of the cooled substrate,
a boundary layer forms, characterized by steep gradients. Such boundary
layers in chemically reacting gases attracted strong interest in connection
with space flight and re-entry simulation and, therefore, sub boundary
Fig 7. Schematic of the triple-torch plasma reactor for TPCVD.
31
layers have been extensively analyzed [38].
Besides d.c. plasma jets, other types of plasma reactors have been
also utilized which make use of high frequency (r.f.), hybrid and mi-
crowave plasmas. For increasing the available plasma volume, multiple
torch arrangements have been developed. A typical reactor with three
torches mounted in a way that the three jets coalesce to form an extended
plasma region (Fig. 7), and with reactant injection through a wa-
ter-cooled probe into the region where the three jets merge has been
used for a parametric study of the influence of the process param-
eters on the film characteristics [39,40].
Typical power levels have been 36 kW total power, and deposi-
tion rates of 40 to 60 m/hr are achieved at reactor pressures of 270
Torr. This reactor has also been used for the homoepitaxial growth
of diamond films with growth rates of 100 to 200 m/hr [41,42].
As an example, Fig. B shows typical diamond films produced in
this reactor without (Fig. 8a) and with (Fig. 8b) secondary precursor
injection. Secondary precursor injection through a ring surrounding
the plasma plume (Fig. 7) provides more uniform precursor distri-
bution over the substrate surface which has a pronounced effect on
the surface morphology of the deposited films. Figure 8b indicates
an almost uniform film morphology over the entire substrate diameter
(30 mm).
Present research efforts dealing with this emerging technology are
Fig. 8a. Micrographs of diamond films deposited with the triple torch plasma reactor
without secondary precursors.
Fig. 8b. Micrographs of diamond films deposited with the triple-torch plasma reactor
with secondary precursors.
32
primarily concerned with establishing the knowledge base, necessary
for further developments. Both experiments and modeling efforts consider
the chemistry in the boundary layer in front of a substrate in order
to find the most important growth species and to assist in understanding
of the growth process. In the case of diamond CVD, growth rate and
crystal structure and orientation as a function of the governing parameters
are some of the most important research topics.
As an example, Fig. 9 shows an arrangement of the precursor species
in front of a substrate during diamond deposition from plasma. [4]
Similar arrangements may use one or several arcs as the plasma source.
These arrangements are very useful for manipulating the thickness
of the boundary layer in front of the substrate by varying the ar-
gon injection flow rate through the injection probe indicated in Fig.
9. By increasing the argon flow rate and keeping the other parameters
the same, the boundary layer thickness may be reduced and the
effect of this reduction on the boundary layer chemistry may be
determined. Corresponding modeling work shows that the importance
of hydrocarbon and carbon species for diamond deposition as the
thickness of the boundary layer changes as shown in Fig.10. Both
Fig. 9. R.F. plasma reactor for TPCVD of diamond films with attached Quadropole
mass spectrometer [43,44].
33
the growth chemistry and the growth rate are strongly affected by
the thickness of the chemically reacting boundary layer in front
of these substrates.
For relatively thick boundary layers (2 mm), the model predicts
diamond growth rates of 10 m/hr in accord with experimental data,
with the dominant growth species being methyl. For very thin boundary
layers (=0.1 mm) the model predicts diamond growth rates of several
hundred m/hr, with a transition to monatomic carbon as the dominant
growth species as shown in Fig. 10 [45, 46].
2.3.1. Summary of present R&D efforts in TPCVD
Present research efforts focus on three aspects;
a. ) Boundary and substrate surface chemistry As previously men-
tioned, both experiments and modeling efforts have been concentrating
on finding the primary growth species in front of the substrate in
conjunction with modeling of the boundary layer and surface chemistry.
This does not only apply to the deposition of diamond films, but also
to TPCVD of other materials.
b.) Control of film morphology and film quality are current re-
search topics in this field. Film quality is of particular concern for
optical and electronic applications of diamond films.
c.) Film growth rates and film adhesion. Although film deposi-
tion using thermal plasmas as a tool for generating growth species,
Fig. 10 Individual precursors contribution to diamond growth [45,46].
34
result in rather high growth rates, the quality of the deposited films
may or may not be sufficient for the intended application. Produc-
ing high quality films at high growth rates remains still a challenge.
Film adhesion Is one of the major problems in the field of diamond
deposition, especially for deposition on substrate materials which cannot
tolerate high temperatures (< 5000 C) and on a variety of metal-
lic substrates such as, for example, steel [47], by using intermedi-
ate layers of materials to which diamond tends to adhere well (for
example, Mo), or by using metallic binder materials [48,49], the adhesion
problem may be avoided.
Although there is no clear-cut line between research and devel-
opment , t hi s l i ne i s even more bl urred i n t he case of TPCVD.
Developments in this area have been particularly visible in diamond
film fabrication. Free-standing diamond wafers produced by TPCVD
are already commercially available with diameters up to 25 cm. Coating
of complex three-dimensional shapes is a difficult, hut not insurmountable
problem and corresponding efforts are already in progress.
As with most emerging technologies, economic considerations may
be the overriding concern for new developments. This has been the
case, for example, in the field of diamond film developments. The
initially predicted growth of the market in this field did not mate-
rialize.
Similar as in the case of plasma and wire arc spraying, complete
control of the TPCVD process is the ultimate goal of R&D efforts.
3 THERMAL PLASMA SYNTHESIS
Thermal plasmas which are a source of very high temperatures and
steep temperature gradients offer an attractive and chemically non-
specific route for the synthesis of fine powders down to the nanometer
size range.
The supersaturation of vapor species, which provides the driving
force for particle condensation, can be very large in thermal plasmas,
leading to the production of ultrafine particles by homogeneous nucleation.
Ceramic powders such as carbides, nitrides, oxides, and solid so-
lutions have been successfully synthesized in thermal plasma reac-
tors. Thermal plasmas suitable for the synthesis of fine powders are
primarily produced by means of high-intensity arcs (a.c. or d.c.) and
high-frequency discharges (r.f. and microwave).
The high temperatures in thermal plasma reactors lead to short
processing times which translates into relatively small reactors with
high throughput. In spite of this and other advantages of thermal plasma
reactors, the high processing costs compared to competing processes
35
must be offset by superior properties of the products. In general, only
high value-added products are economically viable.
Recent advances in novel reactor and process designs have enhanced
the quality of powders produced in thermal plasmas. These novel designs
include the r.fl-d.c. hybrid reactor [50], the reactive submerged arc
(RSA) [51], multiple plasma jets, and counter-flow liquid-injection
plasma reactors. They aim at maximizing the heating, the mixing,
and the residence time of materials in the plasma. Either the discharge
itself or the plasma flame downstream of the discharge may be used
for synthesizing the powders. In thermal plasma synthesis, the re-
actants may be gases, liquids, or solids before injection into the plasma.
However the availability of gas-phase precursors for metals is se-
verely limited. Therefore, the most commonly used reactants for plasma
synthesis have been solids. Several years ago, a liquid-injection method
Fig. 11. Schematic of the counter flow plasma reactor for injection of liquid precunsors.
36
was developed to overcome the problems associated with solid in-
jection and to capitalize on the benefits of gaseous reactants [52].
In this case, a conventional d.c. plasma torch produces a high-
temperature plasma jet, usually of an inert gas (Fig. 11). The at-
omized liquid-precursor material is injected from the end opposing
the plasma jet as a fine mist in counter flow into the plasma, which
gives rise to a recirculation vortex. The recirculation provides ex-
tended dwell time of the precursor in a relatively hot zone, assur-
ing complete vaporization and dissociation of the precursor mate-
rials. The rapid quench downstream of the hot zone induces super-
saturation of the dissociated precursor vapor, which in turn, leads
to the desired chemistry and to the formation of fine powder par-
ticles via homogeneous nucleation. The powder is collected in the
water-cooled collection chamber and in a filter attached to the outlet
of this chamber The same basic principle is used in other reactor
configurations and with other than liquid-precursor materials.
In the past, most of the oxide-synthesis work concentrated on
binary compounds. Large-scale production of TiO
2
(pigment) has
Fig.12. Schematic of plasma reactor for the synthesis on nanometer size particles
[61].
37
been pursued for many years by Tioxide in the United Kingdom
using MW-sized arc gas heaters.
With the advance of liquid-injection plasma synthesis within the
last eight years, ternary, quaternary, and higher component oxide solid
solutions including spinels of aluminates, ferrites, and chromites [52-
55], and the high-temperature oxide superconductors [52,56-59], have
been synthesized in both d.c. and r.f. plasmas. Besides these more
exotic oxide compounds, there has been little activity in thermal-
plasma synthesis of oxides.
In contrast to the synthesis of oxide powders, there are strong and
continuing efforts in thermal-plasma synthesis of nonoxide ceramic
powders. The strongest interest seems to be in nitrides, followed by
carbides and finally borides. The most common reactants for ther-
mal-plasma synthesis of nonoxide ceramics are solids and metal halides
[60].
Recently an experiment has been described in which a plasma con-
taining vapor-phase precursors was expanded through a nozzle with
a hot ceramic wall [61] as schematically shown in Fig. 12. This ar-
rangement approaches a configuration of one-dimensional flow with
one-dimensional temperature gradients in the direction of the flow
in the nozzle, leading to high uniformity of the quench rate. Furthermore,
a nozzle provides much higher quench rates than would be obtain-
able otherwise. Using this arrangement, ultrafine SiC particles have
been synthesized with SiCl
4
and CH
4
as reactants. The results show
that the particles produced in the process have a narrow size dis-
tribution peaking at a few nanometers. Both SiC and Si particles have
been produced with median diameters of 8.5 nm and with 90% of
the particles having diameters of less than 16 nm.
3.1. Summary of present R&D efforts in thermal plasma
synthesis
Present research efforts concentrate on non-oxide powders (nitrides,
carbides) and on new approaches for the synthesis of nanometer size
particles. The future growth of this technology will hinge to a large
extent on the development and implementation of effective controls.
This includes control of the particle size and size distribution as
well as of the morphology of the produced powders. Novel reactor
designs are of great interest as, for example, multiple torch arrangements,
hybrid reactors combining d.c. torches with r.f. torches and other
innovative designs which should not only improve the quality of the
product, but also enhance the efficiency of the process.
38
Recent developments are primarily concerned with the scale-up of
plasma reactors and with automatic control of the process.
4. CONCLUSIONS
The following conclusions are restricted to the areas which have been
covered by this overview, i.e., (2) thermal plasma coating technologies
and (2) thermal plasma synthesis of fine powders.
In general, the growth of thermal plasma technologies has been
hampered by the lack of controls. It is felt that synergistics between
basic research and development will be crucial for the implementation
of controls. This implementation will have a strong impact on the
growth of thermal plasma coating technologies.
Developments in plasma synthesis will also benefit from better
controls, and it is anticipated that the focus of research in this area
will be increasingly towards nanometer-structured materials.
5. ACKNOWLEDGEMENTS
A number of colleagues and present and former graduate students contributed
to the material presented in this overview. Their contributions, as well
as the support from NSF and DOE, are gratefully acknowledged.
6. References
1. R. Spores and F. Pfender, Flow Structure of a Turbulent Thermal Plasma
Jet, Surf Coat Technol., 37, pp.251-270 (1989).
2. R. Spores, Analysis of the Flow Structure of a Turbulent Thermal Plasma
Jet, Ph.D. Thesis, University of Minnesota-Minneapolis (1989).
3 F. Pfender, J. Fincke, and R. Spores, Entrainment of Cold Gas into
Thermal Plasma Jets, Plasma Chem. Plasma Process., 11(4), pp.529-
543 (1991).
4. R.W. Schefer, V. Hartmann, and R.W. Dibble, AIAA 1,25, p. 1318 (1987).
5. M. Brossa and F. Pfender,
t
Probe Measurements in Thermal Plasma
Jets, Plasma Chem. Plasma Process., 8(1), pp.75-90(1988).
6. SA. Wutzke, Contidions Governing the Symptomatic Behavior of an
Electric Arc in a Superimposed Flow Field, Ph.D. Thesis, University
of Minnesota-Minneapolis (1967).
7. S.A. Wutzke, E. Pfender and E.R.G. Eckert, Study of Electric-Arc Behavior
with Superimposed Flow
1
, AIAA J, 5(4), pp. 707-714 (1967).
8. M.P. Planche, Z. Duan, 0. Lagnoux, J. Heberlein, P. Fauchais, and E.
Pfender,
t
Study of Arc Fluctuations with Different Plasma Spray Torch
Configurations, Proceedings of the 13th International Symposium on
Plasma Chemistry, edited by C.K. Wu (Beijing, China) Vol.111, pp.1460-
1465 (1997).
9. M.P. Collares and F. Pfender, Magnetic Probe Measurements in Plasma
Spray Torches, Proceedings of the 13th International Symposium on
Plasma Chemistry, edited by C.K. Wu (Beijing, China) Vol.111, pp.1466-
1470(1997).
10. S. Paik, P.C. Huang, J. Heberlein, and E. Pfender, Determination of
the Arc-Root Position in a DC Plasma Torch, Plasma Chem. Plasma
39
Process., 13(3), pp.379-397 (1993).
11 W. Finkelnburg and H. Maecker, Electric Arcs and Thermal Plasmas,
Encyclopedia of Physics, vol. XXII, Springer, Berlin (1956).
12. Th. Peters, Z Phys, 144, p.6
1
2 (1956).
13. S. Malmberg, Analysis of the Plasma Jet Structure, Particle Motion,
and Coating Quality During dc Plasma Spraying, Ph.D. Thesis, University
14. P. Fauchais, I.F. Coudert, and M. Vardelle, Diagnostics in Thermal
Plasma Processing, Plasma Diagnostics, vol. I, edited by 0. Anciello
and D.L. Flamm, Academic Press, New York (1989).
15. D.B. Spalding, Two-Fluid Models of Turbulence, CFDU Report CFDI8S/
4, CPDU, Imperial College, London (1985).
16. D.B. Spalding, A Turbulence Model for Buoyant and Combusting Flows,
CFDU Report CFD/86/4, CFDU, Imperial College, London (1986).
17. D.B. Spading, Zeit. Physiochem. Hydrodynam., 4, p.323(1983).
18. P.C. Huang, J. Heberlein, and F. Pfender, A Two-Fluid Model of Turbulence
for a Thermal Plasma Jet, Plasma Chem. Plasma Process., 15(1),
pp.25-46(1995).
19. P.C. Huang, A Turbulent Swirling Arc Model and a Two-Fluid Turbulence
Model for Thermal Pl asma Sprays, Ph.D. Thesi s, Uni versi ty
ofMinnesota~Minneapolis (1993).
20. D. Apelian, D. Wei, and M. Paliwal, Thin Solid Films, 118, p.395 (1984).
21. Y.P. Chyou and F. Pfender,
'
Behavior of Particulates in Thermal Plasma
Flows, Plasma Chem. Plasma Process., 9(1), pp.45-71(1989).
22. O.P. Solonenko, Fulerian and Lagrangian Modelling of Dust-Laden Plasma
Jets, Proceedings of the 13th International Symposium on Plasma
Chemistry, edited by C.K. Wu (Beijing, China) Vol.111, pp.1471-1473(1.997).
23. C.H. Chang,
1
Numerical Simulation of Alumina Spraying in Argon-Ilelium
Plasma Jetit, Proceedings of the International Thermal Spray Conference
(Orlando, Florida) p.793 (1992).
24. J.H. Park, Z. Duan, J. Heberlein, E. Pfender, Y.C. Lau, and H.P. Wang,
Modeling of Fluctuations Experienced in N
2
and N
2
/H
2
Plasma Jets Issuing
Into Atmospheric Air, Proceedings of the 13th International Sympo-
sium on Plasma Chemistry, edited by C.K. Wu (Beijing, China) Vol.1,
pp.326-331 (1997).
25. P. Fauchais, AC. Liger, M. Vardelle, and A. Vardelle, Formation of
Plasma-Sprayed Oxide Coatings, Proceedings of the lulian Szekely
Memorial Symposium on Materials Processing and the TMS Fall Ex-
traction & Processing Conference, edited by H.Y. Sohn, J.W. Evans,
and D. Apelian, (Cambridge, Massachusetts) pp.571-592 (October 1997).
26. H.-D. Steffens, Z. Flabiak, and M. Wewel, IEEE Trans. Plasma Sci.,
18(6), p.974 (1989).
27. M.L. Thorpe, J. Mat Process., 134, p.69 (1988).
28. D.R. Marantz and D.R. Marantz, Proceedings of the 3rd National Thermal
Spray Conference, p. 113 (1990).
29. E. Sampson and L. Neider, Proceedings of the 5th National Thermal
Spray Conference, p.271(1993).
30. P.R. Sampson, Proceedings of the 5th National Thermal Spray Con-
ference, p. 257 (1993).
31. X. Wang, J. Hebericin, E. Pfendcr, and W. Gerberich,
1t
Effect of Gas
Velocity and Particle Velocity on Coating Adhesion in Wire Arc Spraying
1
,
40
Proceedings of the 9th National Thermal Spray Conference, Thermal
Spray: Practical Solutions for Engineering Problems, edited by C.C.
Berndt, published by ASM International, pp. 807-811(1996).
32. x. Wang, J. Heberlein, F. Pfender, and W. Gerberich, Effect of Shrouded
C0
2
Gas Atomization on Coating Properties in Wire Arc Spray, Pro-
ceedings of the 8th National Thermal Spray Conference (Houston, Texas),
pp.31-37(1995).
33. R.F. Heile and D.C Hill, Welding J., pp. 201-205 (1975).
34. P.J. Wewitt and A.A. Hirst, Ann. Occup. Hyg., 37, p.297 (1993).
35. M. Ushia, K. Nakata, M. Tanaka, H. Tong, and T. Mita, Trans JWRI,
23, p.21 (1994).
36. T. Watanabe, X. Wang, J. Heberlein, and F. Pfender, Fume Generation
Mechanism in Wire Arc Spraying, Proceedings of the 12th Interna-
tional Symposium on Plasma Chemistry, edited by J.V. Heberlein, D.W.
Ernie, and J.T. Roberts (Minneapolis, Minnesota) Vol.11, pp.889-894
(1995).
37. T. Watanabe, X. Wang, J. Heberlein, E. Pfender, and W. Herwig, Voltage
and Current Fluctuations in Wire Arc Spraying as Indications for Coating
Properties, Proceedings of the 9th National Thermal Spray Confer-
ence, Thermal Spray: Practical Solutions for Engineering Problems,
edited by C.C. Berndt, published by ASM International, pp.577-583 (1996).
38. E.R.G. Eckert and F. Pfender, Advances In Plasma Heat Transfer,
Advances in Heat Transfer, 4, pp.229-313 (1967).
39. Z.P. Lu, J. Heberlein, and P. Pfender, Process Study of Thermal Plasma
Chemical Vapor Deposition of Diamond, Part I. Substrate Material,
Temperature, and Methane Concentration
t
, Plasma Chem. Plasma Proc.,
12(1), pp. 35-52(1992).
40. Z.P. Lu, J. Heberlein, and E. Pfender, Process Study of Thermal Plasma
Chemical Vapor Deposition of Diamond, Fart II; Pressure Dependence
and Effect of Substrate PreTreatment, Plasma Chem. Plasma Proc.,
12(1), pp.55-69(1992).
41. KA. Snail, CM. Marks, Z.P. Lu, 1. Heberlein, and F. Pfender, High
Temperature, High Rate Homoepitaxial Synthesis of Diamond in a Thermal
Plasma Reactor, Mater. LetL, 12, pp.301-305 (1991).
42. Z.P. Lu, K. Snail, C. Marks, J. Heberlein, and B. Pfender, High Rate
Homoepitaxial Growth of Diamond in Thermal Plasma, Proceedings
of the 2nd Internati onal Symposi um on Di amond Materi al s (The
Electrochemical Society, Washington, D.C.) Vol.91-8, pp.99-106(1991).
43. P. Gruel, H. Yoon, D. Ernie, and J. Roberts, Map Res. Soc. Symp.
Proc., 334, p.141(1994).
44. P.O. Gretiel, J.T. Roberts, and D.W. Ernie, Mass Spectrometric Analysis
ofa Thermal Plasma Used for CVD of Diamond Films, Proceedings
of the 12th International Symposium on Plasma Chemistry, edited by
J.V. Heberlein, D.W. Ernie, and J.T. Roberts (Minneapolis, Minnesota)
Vol. IV, pp.2209-2214 (1995).
45. B.W. Yu, A Model for Chemical Vapor Deposition of Diamond in a
Radio-Frequency Induction Thermal Plasma, Ph.D. Thesis, University
46. B.W. Yu and S.L. Girshick, Atomic Carbon Vapor as a Diamond Growth
Precursor in Thermal Plasmas, J Appl. Phys., 75, pp.3914-3923 (1994).
47. M. Asmann, C.F.M. Borges, J. Heberlein, and F. Pfender, Thermal Plasma
41
Chemical Vapor Deposition of Diamond on Steel, Proceedings of the
13th International Symposium on Plasma Chemistry, edited by C.K.
Wu (Be4ing, China) Vol.111, pp. 1206-1211(1997).
48. C. Tsai, J. Nelson, W. Gerberich, I. Heberlein, and F. Pfendder, Metal
Reinforced Thermal Plasma Diamond Coatings, J. Mater. Res., 7, pp.1967-
1969(1992).
49. C. Tsai, J.C. Nelson, W.W. Gerberich, I. Heberlein, and F. Pfender,
Diamond-Metal Composite Coatings on Cemented Carbide Tools, Diamond
Related Mater., 2, pp.617-620(1993).
50. T. Yoshida, T. Tani, H. Nishimura, and K. Akashi, J Appl. Phys., 2,
p.640(1983).
51. A. Kumar and R. Roy, J Mater Res, 3(6), p.1373(1989).
52. P. Kong and F. Pfenden Proceedings of the 2nd Int. Conf. Ceram. Powder
Processing Sci. (Berchtesgaden, 1988).
53. P. Kong and B. Pfender, Plasma Synthesis of Fine Powders by Counter-
Flow Liquid Injection, Combustion and Plasma Synthesis of High Tem-
perature Materials, edited by Z.A Munir and J.B. Holt (VCH Publish-
ers, Inc., New York) pp.420-430 (1990).
54. P. Kong, T. Or, L. Stachowicz, and B. Pfender, Better Ceramics Through
Chemistry IV, Mater. Res. Soc. Symp. Proc., edited by B.J.J. Zelinski,
C.J. Brinker, D.E. Clark, and D.R. Ulrich (Pittsburgh, PensyIvania) 180,
p.849(1990).
55. T. Or, Z. Lu, L. Stachowicz, P. Kong, and E. Pfender, Plasma Processing
and Synthesis of Materials III, Mater. Res. Soc. Symp. Proc., edited
by D. Apelian and J. Szekely (Pittsburgh, Pennsylvania) 190, p.83 (1991).
56. H. Zhu, Y.C. Lau, and F. Pfender, RF Plasma Synthesis of YBa
2
Cu
3
O
7
Powders
' '
, J Supercond, 3, pp.171-175 (1990).
57. K. Terashima, T. Yoshida, and K. Akashi, Jpn. Symp. on Plasma Chemistry,
p. 57 (1988).
58. T. Ono, M. Kagawa, Y. Syono, M. Ikebe, and Y. Muto, Ullrafine Powders
Prepared by the Spray-ICP Technique
t
, Plasma Chem. Plasma Process.,
7(2), pp. 201-209 (1987).
59. H. Zhu, Y.C. Lau, and E. Pfender, Deposition of YBa
2
Cu
3
O
7-x
Thick Films
by the Spray-JCP Technique
t
, Proceedings of the 9th International
Symposium on Plasma Chemistry, edited by R. dAgostino (Pugnochdso,
Italy) Vol.3, pp.1497-1502(1989).
60. P.C. Kong and B. Pfender, Chapter 14: Plasma Processes, Carbide,
Nitride and Boride Materials Synthesis and Processing, edited by A.W.
Weirner, published by Chapman & Hall, London (1997).
61. N. Rao, S. Girshick, I. Heberlein, P. McMurry, S. Jones, D. Hansen,
and B. Micheel, 'Nanoparticle Formation Using a Plasma Expansion
Process', Plasma Chem. Plasma Process., 15(4), pp.581-606 (1995).
42
Integrated Analysis of Induction Plasma
Systems
Maher I. Boulos
Plasma Technology Research Centre (CRTP), Dept. of Chemical Engineering, Universit
de Sherbrooke, Sherbrooke, Qubec, J1K 2R1, CANADA
ABSTRACT
A review is presented of advances in induction plasma torch designs and system
analysis. Emphasis is placed on recent studies dealing with a turbulent, three
equation, k-e model formulation which was successfully used for the representation
of the flow, temperature and concentration fields in the discharge under high plasma
gas flow rates and power levels. When combined with a power supply model,
an integrated system analysis is described, which allows the plasma conditions
to be linked to the electrical parameters of the power supply. The proposed approach
provides a valuable tool for integrated induction plasma design and parameter
optimization.
1. INTRODUCTION
The radio frequency (r.f.) inductively coupled plasma source has been
the subject of numerous studies for more than half a century. Its early
development can be traced back to Babat [1] in 1947 and Reed [2]
in 1961. Since then, important developments have been made in the
following four principal areas:
Aerospace research for nuclear rocket simulation and the test-
ing of materials for thermal shields of space vehicles. Installations
in the MW power range were built for this purpose in the USA and
in the former Soviet Union in the 60s and 70s. These were mostly
operated under low pressure with Argon or Air as the plasma gas.
Analytical applications of the inductively coupled plasma (ICP)
for elemental analysis using ICP/Atomic Emission spectroscopy (AE)
or ICP/Mass Spectrometry (MS) techniques. The plasma sources
developed for these applications were generally of low power (1 to
5 kW), operated at a frequency of 27.3 MHz or higher. The plasma
gas was mostly argon, helium or nitrogen. Thousands of such installations
are presently in operation around the world.
43
Integrated Analysis of Induction Plasma Systems
Induction plasma processing of materials. This area was started
in the seventies with the development of novel techniques for the pro-
duction of high purity synthetic quartz for fiber optics. In the eighties
and nineties, attention was shifted to metal and ceramic processing
in general whether for powder spheroidization and densification, Induction
Plasma Vacuum Spraying (IPVS), Induction Plasma Reactive Deposition
(IPRD), Suspension Plasma Spraying (SPS), Plasma Flash Vapori-
zation and Deposition (PFVD) and the induction plasma synthesis
of ultra-fine powders (UFP).
Induction plasma treatment of waste materials. Technology has
rapidly developed in this area in the late eighties and nineties. These
were mostly dedicated to the destruction of military waste materi-
als, or the destruction of chloro-fluorocarbons, and more recently water-
based waste materials.
Successful applications of induction plasma technology in each
of these areas depended, however, on the availability of efficient and
reliable sources for plasma generation and a fundamental understanding
of the basic phenomena involved. A review is presented in this pa-
per of recent advances in induction plasma torch design and math-
ematical modeling with emphasis on a novel integrated approach for
the analysis of induction plasma systems as a whole taking into account
the interactions between the plasma torch and the power supply.
2. PLASMA TORCH DEVELOPMENT
The availability of a reliable plasma generating device is a key
requirement for any significant industrial process development. Plasma
torch development, on the other hand is often motivated by significant
potential for research and industrial applications. The induction plasma
has been no exception in this respect with early torch development
work mostly funded by NASA in the late sixties and early seventies
for its aerospace programs. M. Thorpe of TAFA Corporation in N.H.
USA, Reboux in France, and S. Dresvin of St. Petersburg Technical
University in Russia, have been of the leading researchers in this area
to whom many of the early developments are credited. The main features
of the early induction plasma torch designs include the use of a quartz
plasma confinement tube combined with a wide range of plasma
stabilization techniques schematically represented in Figure 1.
The specific design and the nature of the plasma confinement tube
and of the intermediate separator tube have varied over the years.
The simplest torch designs used quartz, air or water-cooled, for the
plasma confinement tube. Their main limitation was mostly in the
difficulty of scaling up to higher power levels in the hundreds kW,
44
Fig.1 Commonly used induction plasma stabilization techniques (after Dresvin [3]).
and their lack of reliability for long service operation. Alternate torch
designs were developed over the past twenty years along the following
lines:
Metal wall torches in which the quartz plasma confinement tube
is protected by a segmented, water-cooled, metal wall. Figure 2 shows
early designs which were developed in the seventies in the former
Soviet Union. Their energy efficiency is, however, lower than alternate
ceramic or quartz tube torches due to ohmic losses in the metal wall
segments. They are also more difficult to ignite and are less toler-
ant to variations of the chemistry of the plasma gases which can be
contaminated by the metal wall.
Ceramic wall torches in which the quartz plasma confinement tube
is replaced by an appropriate ceramic with a high thermal conduc-
tivity and thermal shock resistance. This patent protected design shown
in Figure 3, developed at the CRTP at the Universit de Sherbrooke
in the eighties also involves the moulding of the induction coil in
a polymer-matrix composite, which gives a rugged and reliable torch
design with high energy efficiency. Torches of this type are presently
designed and manufactured on a commercial scale by Tekna Plasma
Systems Inc. in Sherbrooke, Quebec for power rating varying between
30 kW to 400 kW. This design allows for the addition of different
torch nozzles to control the flow pattern in the emerging plasma jet.
Hybrid plasma torches. These have been developed by Profes-
45
Fig.2 Typical design of the metal wall induction plasma torch (after Dresvin [4]).
Fig.3 Ceramic wall induction plasma torch developed at the CRTP, Universit de
Sherbrooke
Powder + carrier gas
Central gas
Torch cooling
water IN
Intermediate
tube
Powder injection
probe
Plasma
confinement tube
Gas distributor head
Sheath gas
Torch cooling
water OUT
Torch body
Coil water OUT
Coil water IN
Induction coil
Exit nozzle
46
RF coil
Quartz tubes
DC power
supply
Out
In
Fig.4 D.c. / r.f. hybrid plasma torch (after Yoshida et al. [5]).
sor T. Yoshida at the University of Tokyo in Japan in the late sev-
enteen and early eighteen. The d.c./r.f. hybrid torch shown in Fig-
ure 4 after Yoshida et al. [5], features a d.c. torch operating in com-
bination with an r.f. induction plasma torch on which it is co-axi-
ally mounted. The power to the d.c. torch is generally limited to 5
or 10 % of the total plasma power. This design gives rise to higher
energy density in the center of the discharge at the expenses of in-
troduction of low level metallic contaminants in the flow due to electrode
erosion of the d.c. plasma torch. An r.f./r.f. hybrid plasma design
was developed by Kameyama et al. [6]. The d.c. torch in this case
is replaced by a second r.f. induction torch operated at a higher oscillator
frequency than that of the main discharge. Such a design eliminates
the source of contamination of the flow at the expense of a considerable
increase in complexity.
3. MATHEMATICAL MODELING AN INTEGRATED APPROACH
Mathematical modeling is an important tool in thermal plasma studies.
It provides means for the understanding of flow, temperature and
concentration fields in plasma flows and gas mixing phenomena. A
number of models were developed over the years for the inductively
coupled r.f. plasma source. These were generally based on a laminar,
two-dimensional, steady state representation of the plasma under local
thermodynamic equilibrium conditions [7,8]. Over the past ten years,
attention was also given to the development of turbulent flow models
[9-12] and the use of an integrated approach for the modeling of the
47
induction plasma system [12,14-16]. The latter involved the modeling
the magneto-hydrodynamics of the discharge and its interaction with
the electrical characteristics of the power supply circuit. In the following,
a brief description is given of recent advances in this area.
3.1 Plasma dynamic models
The principal difficulty in the development of a generalized mathematical
model of the inductively coupled plasma lies in the fact that while
laminar flow models are satisfactory at low plasma gas flow rates
and plasma power levels, they are generally inadequate at high flow
and power conditions. The difficulty is due to the fact that as the
plasma power is increased with an associated increase of the plasma
gas flow, laminar-to-turbulent transition, is observed in the low
temperature regions of the flow, with the conditions in the high
temperature regions of the discharge remaining predominantly laminar.
The simultaneous presence of these two distinct flow regions offers
an added complexity in terms of model convergence. In recent studies
by Chen et al. [10,11], Merkhouf et al. [12] and Ye et al. [13], a
three equation (3E) turbulence fluid model was proposed taking into
account the density fluctuations in the plasma. The model makes use
of the standard k-e turbulence formulation. The following assumptions
were made:
Steady state and isotropic turbulent flow;
Axi-symmetric two-dimensional system of coordinates;
Plasma is in local thermodynamic equilibrium (LTE) condition;
The plasma is optically thin;
The thermodynamic and transport properties have no fluctuations
except for the density;
The viscosity dissipation of the thermal energy is negligible.
3.1.1 Governing equations
The governing equations for the 3E fluid model are briefly described
as follows (in the equations, the dependent variables with a bar denote
the conventional time-averaged quantities, others are the density-weighted
averaged quantities, see Merkhouf et al. [12], and Ye et al. [13]):
(a) Continuity equation
( ) ( ) 0
1
v r
r r
u
z

(1)
(b) Momentum conservation equations
48
( ) ( )

,
_
,
_
r
u
r
r r z
u
z
vu r
r r
uu
z
eff eff

1 1
(2)
Lz eff eff
F
z
v
r
r r z
u
z z
p
+
,
_
,
_

1
( ) ( )
,
_
,
_
r
v
r
r r z
v
z
vv r
r r
uv
z
eff eff

1 1
(3)
Lr eff eff eff
F
r
w
r
v
r
v
r
r r r
u
z r
p
+ +
,
_
,
_
2
2
2
1

( ) ( )

,
_
+ +
,
_
,
_
r r
v
r
w
r
w
r
r r z
w
z
vw r
r r
uw
z
eff eff
eff eff

1 1
(4)
where u, v, w are the axial, radial and tangential components of the
flow velocity;
Lz
F
and
Lr
F
are axial and radial components of the
Lorentz force;
t l eff
+ is the effective viscosity of the plasma,
which is the sum of the molecular viscosity
l
and the turbulent viscosity
t
.
(c)Energy conservation equation
( ) ( )
R J eff eff
U Q
z
h
r
r r z
h
z
vh r
r r
uh
z
+
,
_
,
_
1 1

(5)
where h is the plasma enthalpy,
) Pr / ( ) / (
t t p eff
C + , is the
combined molecular and turbulent energy transport coefficient, Pr
t
is the turbulent Prandtl number.
J
Q and
R
U
are the Joule heating
rate and the volumetric radiation energy loss rate of the plasma.
(d) k-e equations
( ) ( ) +
,
_
,
_
G
r
k
r
r r z
k
z
vk r
r r
uk
z
k k
1 1
(6)
49
( ) ( )
k
C
k
G C
r
r
r r z z
v r
r r
u
z
2
2 1
1 1
,
_
,
_
(7)
where ) Pr / (
k t l k
+ , ) Pr / (

t l
+ , are t he combi ned
transport coefficients for the turbulent kinetic energy k and its dissipation
rate , respectively,

/
2
k C
t
; Pr
k
and Pr
e
are the corresponding
Prandtl numbers. G is the generation rate of the turbulent kinetic energy.
'
,
_
+
1
]
1
,
_
,
_
+
1
1
]
1
,
_
,
_
,
_
2
2
2 2 2 2
2
z
v
r
u
r
w
r
r
z
w
r
v
r
v
z
u
G
t
(8)
(e) Temperature variance equation
( ) ( )
2
2 2
2 2
"
" 1 "
"
1
" T
k
C G C
r
T
r
r r z
T
z
vT r
r r
uT
z
T T g eff eff

,
_
,
_
,
_

(9)
where
1
1
]
1
,
_
,
_
2 2
Pr r
T
z
T
G
t
t
T
(10)
is the generation rate for the temperature variance.
The constants in the preceding equations (1) (10) are as fol-
lows: C
= 0.9, C
1
= 1.44, C
2
= 1.92, C
g
= 2.0, C
T
= 6.0,
Pr
k
= 1.0, Pr
= 1.30, Pr
t
= 0.7.
The solution of the above listed fluid dynamic equations requires
a knowledge of the associated electromagnetic fields for the calcu-
lation of the Lorentz forces (
Lz
F and
Lr
F
), and Joule heating acting
on the plasma (
j
Q ). These were computed using the vector potential
formulation of the two-dimensional electromagnetic fields after
Mostaghimi and Boulos [8]. This model considers the interaction between
the applied magnetic field of the coil and induced magnetic field of
the plasma. The individual equations for the component for the vector
potential, A
, inside the r.f. discharge are written as:

50
0
1
0
2 2
2
+
,
_
I
R R R
A
r
A
r
A
r
r r z
A

(11)
0
1
0
2 2
2
+
,
_
R
I I I
A
r
A
r
A
r
r r z
A

(12)
where A
R
and A
I
are, respectively, the real and imaginary parts of
the vector potential, r & z are the coordinates in the radial and axial
directions, the electrical conductivity of the plasma,
0
, the magnetic
permeability of free space and = 2f, with, f, the oscillator frequency.
The vector potential on the wall and the free space surrounding
the plasma is determined by the superposition of the coil and the plasma
effects. In the numerical scheme used, the plasma torch is divided
into a number of control volumes in the form of cylindrical loops,
the components of the potential vector on the wall and free space
surrounding the discharge can be written as:

+
. .
1
,
1
,
) (
2
) (
2
V C
i
i i I i
i o
i
coil
i
c c o
i R
k G S A
r
r
k G
r
r I
A
(13)
( )

. .
1
,
0
,
2
V C
i
i i i R i
i c
i I
k G S A
r
r I
A
with
c
r r r
0
(14)
with
( ) ( )
i
i i i
i
k
k E k k k
k G
) ( 2 2
) (
2

(15)
and
K
i

4r
c
r
i
(r
c
+ r
i
)
2
+ (z
i
z
c
)
2
(16)
The first summation in Eq. (13) extends over the number of coil
turns and the second one extends over the current carrying regions
of the discharge, where I
c
is the coil current, z
c
is the axial position
of the current loop, K and E are the complete elliptic integrals, r
0
51
and r
c
are, respectively, the radius of the plasma confinement tube
and coil and r
i
and S
i
are the radius and cross section of the i
th
control
volume. The corresponding electrical field intensity, E
, the axial and

radial components of the magnetic field H
z
and H
r
, can be calculated
as follows:
E
j A
(17)
( )
rA
r r
H
z
1
0
and
( )
A
z
H
r

0
(18)
The r.f. discharge power dissipated in the plasma load, P
0
:
P E dv
v
0
2
1
2
=
z
s
q
(19)
3.1.2 Torch geometry and operation conditions
The rf plasma torch and the corresponding coordinates used by Ye
Fig.5 Torch geometry and system of coordinates (after Ye et al. [13]).
Q1 Q3 Q2
r3
r0
Zs
r
z
RF coil
Zp
Z1
Z2
Z3
Z = 0
r1
r2
rc
52
Table 1 Dimension and operation parameters of the r.f. induction plasma torch (after
Ye et al. [13]).
r
1
= 1.35 mm z
1
= 35.15 mm
r
2
= 4.80 z
2
= 75.15
r
3
= 13.0 mm z
3
= 113.0 mm
r
0
= 17.5 mm z
p
= 35.15 mm
r
C
= 22.0 mm z
S
= 60.0 mm
Q
1
= 10 slpm (Ar) P
0
= 1530 kW
Q
2
= 25 slpm (Ar) f = 3 MHz
Q
3
= 4080 slpm (Ar/H
2
, 10% vol. H
2
p = 200760 torr
slpm standard liter per minute
et al. [13] are schematically shown in Figure 5. Table 1 summarizes
the main torch dimensions and operation parameters. In order to obtain
a satisfactory cooling of the inner torch wall, a sheath tube r = r
3
is inserted into the torch, the sheath gas Q
3
is injected into the torch
by passing through the annular channel between the sheath tube and
the inner wall of the plasma confinement tube. Because the flow rate
of the sheath gas is relatively high, it could be an important source
of plasma turbulence. A volumetric percentage of 10 % hydrogen is
added to the argon sheath gas.
3.1.3 Typical results
The conservation equations of the 3E turbulent fluid model were solved
by using the same boundary conditions as used by Merkhouf et al.
[12]. The calculations domain for the rf plasma torch is z = 0113.0
mm, r = 017.5 mm with a non-uniform grid of 39 35 mesh points.
Highlights of the results are presented in terms of the isocontours
of the plasma temperature, the ratio of the turbulent to the laminar
viscosity, r
, and relative turbulence intensity, r

turb
. The latter two
parameters were calculated as follows:
(a)Relative turbulent viscosity
l
t
r

(20)
53
Fig. 6 Temperature field (left) and relative turbulence viscosity (right) for an atmospheric
0 5 10 15
r (mm)
0
20
40
60
80
100
z

(
m
m
)
1
0
0
0
0
K
8
0
0
0
K
5
0
0
0
K
2
0
0
0
K
0 5 10 15
r (mm)
0
20
40
60
80
100
z

(
m
m
)
1
3
5
1
3
20
(b)Relative turbulence intensity
2 2 2
int
3 / 2
w v u
k
r
turb
+ +
(21)
Figure 6 shows typical temperature isocontours (left) and the cor-
responding relative turbulent viscosity field, r
, (right) for an atmospheric

pressure pure argon plasma at a power level of 15 kW and a sheath
gas flow rate, Q
3
= 80 slpm (Ar). From a comparison of both sides
of the figure, it may be noted that the flow is highly turbulent in
the upstream cold end of the torch near the entrance of the central
gas, Q
2
. In the wall boundary layer in the vicinity of the plasma con-
finement tube, high turbulent viscosities are also observed. These are
in the range of 3 to 5 times the local molecular viscosity. Within the
discharge in the coil region, the plasma temperature and consequently,
its molecular viscosity, is sufficiently high that laminar flow pre-
dominates.
Local distribution of the turbulence intensities in the flow are given
in Figure 7 for two plasma pressure (101.3 kPa and 26.7 kPa). The
plasma power in this case was 15 kW and the sheath gas flow rate
= 40 slpm (Ar). High turbulence levels are observed on the upstream
cold side of the discharge and along the wall boundary layer of the
54
0 5 10 15
r (mm)
0
20
40
60
80
100
0.1
0.2
0.3
0.1
0 5 10 15
r (mm)
0
20
40
60
80
100
z

(
m
m
)
z

(
m
m
)
0.1
0.2
0.2
0.3
0.2
0.1
Fig.7 Relative turbulence intensity isocontours for an Argon plasma at 15 kW and
Q
3
= 40 slpm and (a) pressure = 101.3 kPa (b) pressure = 26.7 kPa (after Ye et al.
[13]).
plasma confinement tube. A slight increase in turbulence can also
be noted at the end of the central powder injection probe which is
responsible for the local cooling of the flow in this region. Higher
turbulence levels are also noted at atmospheric pressure compared
to that for low pressure operation.
3.2 Integrated model
The development of an integrated model of the r.f. induction plasma
aims specifically at predicting the electrodynamic interactions between
the plasma and the power supply. For this a detailed analysis of the
electrical characteristics of the discharge is required as well as dynamic
model of the power supply circuit. Recently, a number of models have
been proposed for the simulation of the integrated induction plasma
generator system [12,14-16]. With the exception of the study by
Kim et al. [15], these do not go through the full computation of the
flow and temperature fields in the discharge and accordingly do not
include the inherent changes in the electromagnetic characteristics
of the plasma which may result from changes in the electrical parameters
of the power supply circuit. The work of Kim et al. [15], on the other
hand, relies on a laminar flow model which does not allow for the
55
computation of the flow and temperature fields in the discharge under
realistic gas flow, and high power conditions. A comprehensive model
was proposed by Merkhouf and Boulos [16] based on the coupling
of the above described, k-e, turbulent plasma dynamic model with
a non-linear analytical model of the generator circuit. The overall
electrical circuit is analyzed in this case with power electronic concepts
with a simplified anode model for the triode. For a given triode type,
and a given value of the d.c. plate voltage, the model provides a means
of computing steady state output electrical signal for the generator
such as grid voltage, r.f. plate voltage and current, d.c. plate current,
overall efficiency of power transmission to the plasma and the plasma
impedance.
3.2.1 Plasma electrodynamic model
The plasma electrodynamic parameters were calculated based on the
solution of the above listed plasma dynamic model equations (Eqs.
1-10), with the corresponding two-dimensional electromagnetic field
equations (Eqs. 11-22). The reactive power supplied by the generator
to the coil, P
coil
, could then be calculated as follows:
( )
+
v
z r o coil
dv H H f P
2 2
2
1

(22)
where dv is the volume element.
The equivalent inductance of the plasma coil is calculated as follows:
L
eq

2P
coil
Ic
2
(23)
The equivalent resistance, R
eq
, represented by the plasma as seen
by the coil is calculated from the r.f. discharge power, P
0
, and the
square of the total equivalent induced current, I
ind
, as follows:
2
2
m
I
P
R
ind
o
eq

with
s
ind
ds E I

(24)
where m is the ratio of the total induced current to the coil current
(m = I
ind
/ I
c
).
56
The equivalent impedance of the plasma coil, Z
eq
, is finally de-
rived as:
( ) ( )
2 2
2
eq eq eq
fL R Z +
(25)
3.2.3 Power supply circuit model
Figure 8 shows a schematic of a typical r.f. power supply circuit.
The input power to the oscillator is supplied by the filtered dc current
supply via the choke coil, L
1
, which maintains a constant dc plate
voltage, V
p0
. The load of the generator is composed of the plasma
torch, tank coil, L
2
, with internal resistance R
2
and leads, which constitute
an equivalent resistance, R
l
, and an inductance, L
l
. A capacitor, C
2
,
in parallel the load draws up an oscillating circuit. The grid feedback
factor, k, is defined as the ratio of the alternating grid voltage, V
g
,
to the r.f. anode voltage, V
p
, which is adjusted mechanically by the
operator, to a value within the limits of (0.1 < k < 0.3)[17]. The
control factor, h, is used for fixing the negative d.c. grid voltage,
which depends in turn, on the value of the grid resistance, R
g
, and
capacitor, C
g
.
In the induction plasma installation, the oscillator circuit is tuned
to its natural resonant frequency which can be calculated as:
f
L L L C
eq
=
+ +
1
2
2 1 2
p
d i
(26)
Fig.8 Block diagram of the electrical circuit of the r.f. power supply (after Merkhouf
and Boulos [16]).
C1
Ip0
Vp
L1
V2
L2
I1 Ic
R2
L
l
R
l
Vp0
Triode
C2
Oscillator
Rg Cg
k, h
Grid command circuit
Solid state
rectifier

Plasma torch
57
At this frequency, the magnitude of the equivalent impedance of
the tank circuit is given by:
Z
L L L
R R R C
co
eq l
eq l
=
+ +
+ +
2
2 2
d i
(27)
The anode voltage, V
p
, is the sum of the d.c. rectifier supply voltage,
V
p0
, and the alternating voltage, V
2
, formed in the oscillator circuit.
V V V V V t
p po
= + = -
2 2 2
with ( ) cos( )
max
w
(28)
The grid voltage, V
g
, is the sum of the fixed bias (-kh(V
2
)
max
) and
the a.c. component (-V
2
) delivered from the oscillator circuit.

V k V kh V V V
g p
= - - =
2 2 2 0
( ) ( )
max max
with a
(29)
The average value of the plate current pulse is given as:
( ) ( ) ( ) [ ]
c c c
D
p
p
k
R
V
I
cos sin 1
0
0

(30)
where
c
is the cut-off angle of the oscillator tube, see Merkhouf and
Boulos [16].
The d.c. power input to the plate circuit, P
dc
, the fundamental a.c.
output oscillator power, P
os
, and the plate loss in the tube, P
d
, are
calculated as:
P V I
dc p p
=
0 0
(31)
co
M
os
Z
V
P
2
2
2
(32)
P
d
P
dc
P
os
(33)
The plate circuit efficiency, h
p
, torch coupling efficiency, h
t
, and
overall efficiency, h
0
are calculated respectively as follows:
58
Table 2 Specifications of the tank circuit components (after Merkhouf and Boulos
[16]).
C
2
[nF] [] [] L
2
[H] R
2
[] R
1
[] L
1
[H] k[]
1.5 20 700 0.1 0.06 0.01 0.05 0.3
Power supply
circuit model
RF induction plasma
fluid dynamic model
Yes
No
End
Convergence
Req,
Leq
f, P0
Triode characteristics
Plate resistance, Rp
Amplification factor,
Grid feedback, k
Coil dimension
Plasma gas
DC plate voltage Vp0
Initial P0 and f
Fig.9 Block diagram for the integrated power supply induction plasma model (after
Merkhouf and Boulos [16]).
( )
po po
co
dc
os
p
I V
Z V
P
P 2 /
2
max 2

(34)
( )
( ) [ ]
co
c os
os
o
t
Z V
I R R P
P
P
2 /
5 . 0
2
2
2
1 2
+

(35)
t p o
(36)
3.2.3 Typical results
To calculate the operating point of the r.f. plasma installation, for
a given value of the d.c. plate voltage V
p0
, both the fluid dynamic
and power supply circuit models have to be solved simultaneously.
These are coupled by the hierarchical algorithm schematically represented
in Figure 9. The solution procedure is as follows, first the fluid dynamic
model is solved for a given initial guess of the dissipated power (P
0
)
59
5 6 7 8 9 10 11
0.5
0.6
0.7
0.8
0.9
1.0
1.1
0.5
0.6
0.7
0.8
0.9
I
n
d
u
c
t
a
n
c
e

[

H

]
Plate voltage, Vp0 [kV]
R
e
s
i
s
t
a
n
c
e

[
O
h
m
]
Fig.10 Plasma resistance and inductance as function of the plate voltage (after Merkhouf
and Boulos [16]).
in the plasma and oscillator frequency (f). The equivalent plasma
impedance can then be calculated using Eq. 25. The power dissipation
(P
0
) in the plasma, and frequency (f) are then computed, and new
plasma impedance was recalculated using P
0
and f as the input to
the r.f. fluid dynamic discharge model. This procedure was repeated
until P
0
and f had a constant convergent values.
Results were given by Merkhouf and Boulos [16] for an atmos-
phere pressure of pure Argon plasma. The tank circuit components
specifications are given in Table 2. Figure 10 shows the variation
of the equivalent inductance and resistance of the plasma torch with
the plate voltage (V
p0
). It is noted that the inductance drops slowly
when the plate voltage is increased. This is due to the observed increase
of the high temperature region and its shifting closer to the wall of
the plasma confinement tube. This results in the increase of the cross
section of the plasma through which the magnetic flux penetrates and
the corresponding reduction of the space between the coil and the
r.f. discharge which is responsible for the flux leakage. The plasma
reactance is compensated automatically by the generators tank circuit
which adjust its frequency to mach the reactive load. The equiva-
lent resistance, on the other hand, increased with the increase of the
plate voltage. The maximum plasma volume is dependent on the size
of the torch. The increase in volume means higher resistance, while
a higher mean temperature has the opposite effect. The combined effect
of plasma temperature and plasma volume gives rise to the observed
slower rate of increase of the plasma resistance with the plate voltage
at high values of the plate voltage.
Figure 11 shows the variation of the oscillator efficiency and the
60
5 6 7 8 9 10 11
68
70
72
74
76
78
80
0
10
20
30
40
O
s
c
i
l
l
a
t
o
r

e
f
f
i
c
i
e
n
c
y
,

o
s
[
%
]
A
n
o
d
e

l
o
s
s
,

P
d
[
k
W
]
Fig.11 Oscillator coupling efficiency and anode loss as function of the plate voltage
(after Merkhouf and Boulos [16]).
5 6 7 8 9 10 11
86
87
88
89
90
91
92
93
94
0
20
40
60
80
100
T
o
r
c
h

c
o
u
p
l
i
n
g

e
f
f
i
c
i
e
n
c
y
,

t

[
%
]
O
s
c
i
l
l
a
t
o
r

p
o
w
e
r
,

P
o
s
[
k
W
]
anode loss as function of the plate voltage. As expected, the anode
loss in the triode increases rapidly with the increase of the plate voltage
which in turn reduce the oscillator efficiency. The latter varies between
60 % to 80 % depending of the triode tube characteristics and op-
erating conditions. Figures 12 and 13 show the variation of the torch
efficiency and overall coupling efficiency for the r.f. plasma system
as function of the plate voltage. The torch coupling efficiency in-
crease with the increase of the plate voltage. This is due to the increase
of the injected power which gives rise to a corresponding increase
of the volume of the discharge and, consequently, of the cross section
linked with the magnetic flux. Since the oscillator power (P
os
) is dissipated
respectively in the plasma and in the inductor resistance, the torch
Fig.12 Oscillator power and torch coupling efficiency as function of the plate voltage
61
5 6 7 8 9 10 11
68
69
70
71
72
73
74
0
20
40
60
80
100
120
O
v
e
r
a
l
l

c
o
u
p
l
i
n
g

e
f
f
i
c
i
e
n
c
y
,

o
[
%
]
I
n
p
u
t

p
o
w
e
r
,

P
d
c
[
k
W
]
coupling efficiency, which depends on their ratio, is improved when
this ratio is higher.
References
1 G.I. Babat, J Inst Elec Eng, 94, No.27, 27-37 (1947).
2 T.B. Reed, High Temp Sci, 32, No.5, 821-824 (1961).
3 S.V. Dresvin (ed), In: The Fundamentals of Theory and Design of HF
Plasma Generators, translated from Russian (1993).
4 S.V. Dresvin (ed), In: Physics and Technology of Low Temperature
Plasmas, Iowa State Univ Press (1977).
5 Y. Yoshida et al, J Appl Phys, 54, No.2, 640-646 (1983).
6 T. Kameyama, et al, J Mat Sci, 25, No.2A, 1058-1065 (1990).
7 J. Mostaghimi, et al, Plasma Chem Plasma Process, 4, No.3, 199-
217 (1984).
8 J. Mostaghimi and M.I. Boulos, Plasma Chem Plasma Process, 9, No.1,
25-44 (1989).
9 M. El-Hage, et al, J Appl Phys, 65, 4178-4185 (1989)..
10 K. Chen and M.I. Boulos, J Phys D: Appl Phys, 27, 946-952 (1994).
11 X. Chen, et al, In: Preliminary Study of the 3-Equation Turbulence
Model of an R.F. Plasma Torch, Proc of 3
rd
Asia-Pacific Conf. on Plasma
Science and Technology, Vol.1, Tokyo, Japan (1996), pp.71-76.
12 A. Merkhouf, et al, In: Mathematical Modeling of Plasma Systems, Pro-
ceedings of the Julian Szekely Memorial Symposium on Materials Process-
ing, Boston, USA (1997), pp.509-528.
13 R. Ye, et al, Int J Heat Mass Transfer, 42, No.9, 1585-1595 (1999).
14 J. Fouladgar, et al, IEEE Trans Magn, 29, No.6, 2479-2481 (1993).
15 J. Kim, et al, IEEE Trans Plasma Sci, 25, No.5, 1023-1028 (1997).
16 A. Merkhouf and M.I. Boulos, Plasma Sources Sci Tech, 7, 599-606
(1998).
Fig.13 Input power and overall coupling efficiency as function of the plate voltage
62
Plasma Metallurgy:
Current State, Problems and Prospects
Yu.V.Tsvetkov
A.A.Baikov Institute of Metallurgy, Russian Academy of Sciences, Leninskii Pros-
pect 49, GSP-1, Moscow 117911, Russia
Metallurgical processes, based on plasma technology, are capable of
satisfying a large number of requirements of modern metallurgy: devel-
opment of highly intensive processes, complex utilization of processed
raw materials and ecological compatibility with the environment,
preparation of materials with special properties.
We shall discuss some of the main achievements in the area of
plasma metallurgy [1,2,3].
The development of theory and mathematical approaches describing
the processes of generation of plasma in arc and electrode-free genera-
tors enabling engineering calculations of plasma generators with suf-
ficient power.
Design solutions (in hardware) of arc generators (to 10 MW), and
hf generators (to 1 MW) with possible further improvements.
Development of procedures for examining plasma metallurgical
processes, mainly the processes of plasma reduction and synthesis
based on computer calculations of high-temperature thermodynamics of
processes enabling the evaluation of the yield, temperature conditions
and energy consumption in the process; mathematical modelling of
physical-chemical processes, including the processes of heat and mass
exchange which are often limiting, using computers; direct experiments
with the calculated parameters; selection optimisation of design and
technological features of the process and pilot-plant and industrial
applications.
The development of the scientific concept of plasma metallurgy;
unified fundamental considerations regarding interaction of plasma with
the substance and optimisation of design and technological solutions to
63
Plasma Metallurgy: Current State, Problems and Prospects
develop highly efficient process and systems for metallurgy of ferrous
and nonferrous metals.
A number of successful examples of industrial application of plasma
processes [1, 2].
The contribution of Soviet (mainly Russian) scientists and engineers
to the establishment of plasma metallurgy as a new scientific and tech-
nical direction. They include the laboratory of the first technological
electric arc plasma generators, developed at the A.A.Baikov Institute
of Metallurgy and supervised by N.Rykalin, scientific and design ba-
sis for the development of a powerful electric arc plasma apparatus by
the Novosibirsk School of Academician M.Zhukov, significant contri-
bution to the development of theoretical and applied plasma chemistry
by Professor L.Polak and his colleagues, development of advanced
plasma equipment for plasma remelting and refining a number of met-
als and alloys at the E.O.Paton Electric Welding Institute in Kiev,
headed by Academician B.Paton, work of a team of metallurgists,
headed by Academician N.Lyakishev working on the development, con-
struction and industrial application of plasma remelting of high-grade
steels, and a number of scientific and technical organizations and ex-
perts.
Plasma metallurgy is based on thermodynamics and kinetics of high-
temperature processes [4, 5] using the assumptions of equilibrium and
nonequilibrium plasma chemistry. As a scientific and technical disci-
pline, plasma metallurgy has a number of special features determined
by both the continuous nature of the processes and systems used for
these processes and by the special role played by heat and mass ex-
change processes.
The most expensive theoretical examinations have been carried out
to examine the gas-phase processes which are similar to classic plasma
chemical processes, the processes of plasma reduction and synthesis in
reducing media [1, 2, 3], and also in the certain problems of plasma
refining and remelting [1, 6].
In most cases, high-temperature investigations are carried out using
computers for both thermodynamic calculations [7, 8] and for exam-
ining the kinetics and modelling plasma processes [3]. A number of
mathematical models have been developed for describing various
mechanisms of interaction of plasma with the substance [3, 9] in or-
der to examine and optimise the plasma processes.
In the majority of cases metallurgical processes take place outside
the plasma generator in plasma heat carriers which received, in the
plasma generator, the chemical composition and a specific degree of
excitation of atoms and molecules. They are also characterized by the
64
presence of a condensed phase as the process material and/or target
product.
Thermodynamic calculations are carried out to evaluate: the yield of
the product, energy consumption, essential data for determining the
transfer coefficient and the required volume of the reaction system.
From the procedure viewpoint, the problem of thermodynamic cal-
culations of the system with a plasma heat carrier is reduced to deter-
mining the equilibrium parameters of a multicomponent system includ-
ing components in different states of the substance. The required in-
formation is obtained in a solution using a computer of a system of
chemical equilibrium equations. This system of equations is compiled
using methods based on the first variance principle of chemical ther-
modynamics, the second variance principle of thermodynamics and the
application of the law of acting masses.
The degree of depth utilization in practice is determined by the pres-
ence of appropriate algorithms and programs. The approaches based on
variance principles are more universal and in particular, in cases out-
side the limits of classic thermodynamics they it possible to take into
account in a simple manner the Coulomb interaction at high degree of
ionization, formation of solutions, intermolecular interactions of differ-
ent type, etc. At the same time, the relatively simple nature, easy to
understand form and tradition cause that the main part of calculations
of high-temperature systems, which are of interest for metallurgy, has
been carried out using a procedure based on the law of acting masses.
Examples of calculating specific metallurgical systems were published
in the literature [3, 7, 8].
In analysing special features of high-temperature kinetics, it is con-
venient to examine gradually the homogeneous or gas-phase process
and processes in which condensed phases take part.
In most cases the kinetics of gas-phase chemical reaction in plasma
heat carriers is identical with conventional chemical kinetics. After all,
in a general case the main assumptions of classic kinetics the law
of action of masses, the theory of the activated complex, the Arrhen-
ius equation, based on the existence of the equilibrium Maxwell
Boltzmann distribution, are applicable only up to a specific tempera-
ture or, more accurately a specific degree of violation of distribution.
The possibility of using the MaxwellBoltzmann distribution is deter-
mined by the relationship between the relaxation time of the system and
the transformation time of the particles as a result of chemical reac-
tion. It is natural that the temperature boundary of applicability of
classic kinetics is linked with the activation and energy and increases
with an increase of this energy. According to the investigations, the
65
effect of violation of the internal degrees of freedom, caused by the
chemical reaction, on its rate and the effect of nonequilibrium popu-
lation of oscillatory levels on the reaction rate to E >10RT are small.
At E = 5RT deviations do not exceed 20%.
The requirement E > 5RT should also be accepted as the condition
of possible application of classic kinetics.
In most cases, gas-phase processes of plasma technology are carried
out in jets of a plasma heat carrier so that it is necessary to solve
jointly the equations of chemical kinetics and hydrodynamics to de-
scribe these processes.
Plasma metallurgical processes in which condensed compounds take
place may be subdivided into processes in which the initial compound
are in the compact state in the form of a solid material of sufficient
length or a metallurgical melt, processes in which the initial compound
is in the dispersed form, gas-phase processes with formation of the
product in the form of powder or compact form. The relationships
gathering the majority of processes in the last group are almost iden-
tical with those of the gas-phase processes because in most cases the
product forms outside the temperature zone, typical of the plasma
chemical processes.
If the compact is subjected to the effect of a plasma heat carrier and
the conditions of heat removal are such that there is no extensive
evaporation or sublimation of the process compound, the process takes
place in accordance with the standard laws of solid phase kinetics. In
interaction of the surface of the solid with the discharge, the
electrophysical and electrochemical effect of the discharge may exert
a significant influence on the crystallo-chemical act, the reaction ca-
pacity of the surface and adsorption-desorption processes in the sub-
surface layer.
The processes in which the melt takes place represent one of the
most promising directions of plasma metallurgy. However, the theory
of their interaction with chemically active plasma is still in the devel-
opment stage. Even for such a simple plasma process as the interac-
tion of a metallic melt with nitrogen plasma aimed at alloying the
former it was noted that the process is highly specific [6]. In our in-
vestigations of reducing plasma melting of oxides of the metals of the
iron group we detected a number of specific moments of the
macrokinetics of the plasma process in comparison with the reduction
process under the conditions of arc heating in conventional
electrothermal furnaces.
Transferring to examining the processes in which dispersed com-
pounds take place, it is interesting to discuss initially certain assump-
66
tions which we have made regarding the role of the aggregate state in
metallurgical processes. The aggregate state of the components of the
reaction has a controlling effect on the thermodynamic and kinetic pa-
rameters of the process, the relationship between these parameters, and
their possible correlation with the nature of the chemical bond in the
substance and its physical-chemical properties. The existence of struc-
ture-sensitive properties, typical of a real solid, leads to large differ-
ences in the nature of the variation of the kinetic and thermodynamic
parameters. The theory of the active complex indicates that to ensure
that the rate of the heterogeneous process is equal to or higher than
that of the homogeneous process, it is necessary to reduce greatly the
activation energy, i.e., catalysis. In other words the catalytic effect of
the surface is an essential condition for carrying out the process of
reduction of solid components with gases, especially in the low- tem-
perature range. With increasing temperature the difference in the rates
decreases and above some temperature, typical of the examined reac-
tion, the homogeneous reaction becomes more advantageous.
Examining the possibilities of intensifying the reduction processes
with increasing temperature, it should be mentioned that for the proc-
esses of the solid-gas type, they are greatly restricted both as a result
of a low activation energy and due to a possible reduction of the ad-
sorption and catalytic properties of the surface with temperature which
often leads to interruption of acceleration of the process. Heterogene-
ous processes may be greatly inhibited by the formation of a dense
layer of the solid product which blocks the surface and results in an
impedance of the reduction process. With increasing temperature the
heterogeneity of the process on the factor facilitating the interaction
becomes an obstacle for intensifying the process. Naturally, further
marked acceleration of the process can be achieved transferring the
reagents of the gas phase.
The above-mentioned motion logically leads to the generalizing as-
sumption made by us on the restriction on plasma chemical processes
with participation of condensed compounds by the rate of transfer of
components with the gas phase [3, 4]. This assumption is based on the
results of experimental examination of the kinetics of reduction of
tungsten oxides in the jet of plasma heat carrier when using, as a re-
ducing agent, hydrogen, carbon and the previously noted correlation of
the experimental determined dependence of the degree of reduction with
the data on the degree of transition of components of the gas phase
obtained in mathematical modelling of the process.
The assumptions on the restriction of plasma reduction of dispersed
starting material by the degree of homogenization has been confirmed
67
in examining the processes of reduction of oxides of molybdenum and
nickel and iron, ammonium paramolybdates and paratungstates and in
a number of independent investigations. It may be regarded as gener-
ally accepted.
These assumptions are also used as a basis for developing a number
of mathematical models for description, investigation, prediction and
control of plasma metallurgical processes [1, 2, 9].
Flexible automated systems represent a higher form of organization
of production in which the most labour-intensive functions are carried
out by computers and creative functions by designers, technologists and
organizers of production working in automated working areas.
Flexible systems include two main types of modules apparatus
(technological, transport, measuring, controlling, etc.) and program-
ming used to construct programme facilities for controlling the sys-
tems. The modular nature of construction enables formation of compo-
sition of technical and programming devices in accordance with the
actual requirements of production and ensures extensive application of
standard means of the apparatus-programming interface.
Experimental investigations carried out at the A.A.Baikov Institute
of Metallurgy, Tulachermet Holding Company and in number of other
organizations have shown that the application of plasma technology
enables application of flexible processing systems in metallurgy be-
cause the results show that the same apparatus solution of main plasma
systems can be utilized for processing various types of starting met-
allurgical materials.
The universal nature of plasma technology is very important for
metallurgy the possibility of processing materials in any aggregate
state (gaseous, liquid, solid, in the form of compacts and powder) with
the formation of the target products in the required form.
It is also promising to use plasma for processing complex ores and
various waste in metallurgical production [10]. In most cases, the de-
velopment of new technologies for more efficient utilization of raw
materials requires additional energy consumption per production unit
One of the most important problems in evaluating the prospects of
development of plasma technology is its link with the development of
power engineering. For example, possibilities have been examined of
developing a nuclearmetallurgical complex containing an aggregate
for direct reduction of iron with plasma production and heating of the
processing gas-reducing agent [11].
Electrothermal technology is one of the most advanced direction in
metallurgy. In comparison with conventional pyrometallurgical proc-
esses, it ensures a high concentration of energy and temperature at a
68
high thermal and electric efficiency, high specific productivity on small
areas, a small volume of gases, absence of effluents, and easy control
and automation.
Conventional electrothermal technology has a number of disadvan-
tages: the need to use carbon-graphite electrodes, high requirements to
their quality and their high consumption, dependence of the parameters
of the process on the properties of the charge, disrupting the process
and causing failure situations, and insufficient flexibility of the proc-
ess as regards to variation of the composition of the gas phase.
Modern ore electrothermal technology uses mainly a low-parameter
arc discharge (high current and low voltage, although energy at high
voltages and relatively low currents can be used more efficiently in
furnaces). Transition to higher voltages increases the electrical effi-
ciency of equipment.
Plasma technology eliminates the shortcomings of classic
electrothermal technology and enables systems to be produced with
high specific power and high energy efficiency.
On the whole, the plasma processes satisfy ecological requirements;
they are included in closed technological cycles in which each link
becomes initial for the subsequent link, and in the ideal situation the
links can be readjusted rapidly and without inertia.
Processes take place continuously at a matched rate.
Plasma metallurgy ensures optimum matching of the problem of
development of metallurgy in a new energy basis with the problems of
ensuring complex processing of raw materials.
There are a number of ecological problems associated with the con-
struction of thermal power stations using low-grade high-ash coal.
Construction of thermal power stations create the problem of transport-
ing electric energy associated with considerable losses. Therefore, the
concept of development of energy technological complexes, combining
an energy producing system with metallurgical production, is quite
urgent. The role of plasma technology as a universal means of tech-
nological utilization of electric energy may be quite considerable. The
composition of such a system should include standard plasma equip-
ment in which the following processes can be carried out: jet-plasma
processes of treatment or processing gaseous, liquid or disperse raw
materials, distributed in the plasma jet; processes in the furnace with
a liquid bath, plasma shaft processes.
The investigations carried out by A.V.Nikolaev [1, 12, 13] indicate
the directions of increasing the efficiency of utilizing energy in plasma
metallurgical systems of this kind and development, on the basis of
these systems, of continuous processes of reduction yielding dense
69
metals and powders. On the basis of the method of the system analy-
sis of the energy and material structure, proposed by A.V.Nikolaev,
investigations were carried to examine a number of plasma metallur-
gical processes and systems, and approaches were proposed with de-
veloping continuously acting modern technological systems. Plasma
metallurgical modules have been proposed for a number of cases and
testing on existing prototypes components of continuous metallur-
gical systems, including original plasma reactors with a vertical and
horizontal layer of the melt, rotation and adsorption, powder dispenser,
etc.
It is also interesting to discuss an EPP reactor with a rotating cath-
ode. In the reactors of this type the introduced particles move along a
complex trajectory in the diffusion discharge zone so that the target
process may be carried out already in the volume of the reactor fol-
lowed by phase separation in the liquid bath. A similar procedure is
used for extracting metals of the platinum group from sulphide, cop-
per-nickel ores in South Africa containing chromite.
The work carried out at the Institute of Metallurgy of the Russian
Academy of Sciences and other organizations [14] has created suitable
conditions for developing electrode-free and capacity-plasma generators
of the megawatt power with sufficiently high energy parameters.
Superhigh frequency technology is also highly promising. Electrode-
free plasma generators are characterized by a long operating life, the
generated plasma is clean, it is possible to generate nonequilibrium
plasma, i.e., increase the intensifying effect of plasma on chemical-met-
allurgical processes. They have been used successfully for plasma
chemical synthesis of relatively wide range of compounds and also in
chlorine metallurgy, including production of titanium pigment from ti-
tanium tetrachloride.
One of the promising directions of accelerating the application of
plasma metallurgy is the development of mini plants with plasma met-
allurgy.
The plasma hydrogen processes [15] are ecologically clean. As any
process of hydrogen reduction these processes can be carried out us-
ing a closed cycle in reduction of both oxides and sulphur.
The problem of utilizing hydrogen in plasma metallurgy combines
plasma processes with one of the most promising directions of devel-
opment of nuclear power engineering based on the use of hydrogen as
a secondary heat carrier [16]. In metallurgy, hydrogen is a chemical
reagent/reducing agent ensuring production of high-purity refractory
metals (tungsten, molybdenum, rhenium). Sponge iron, obtained in
hydrogen reduction, is clean in respect of phosphorous and sulphur.
70
The use of gaseous reducing agents corresponds to a tendency to
develop coke-free metallurgy. Low-temperature plasma is a powerful
intensifier of gas reduction processes. The advantages of hydrogen as
a medium and a reagent are based on its high chemical activity in the
ionized state, generation of a large amount of heat in recombination of
hydrogen atoms and the efficiency of heat and mass exchange.
The solution of the problem of producing cheap hydrogen together
with the development of promising, including plasma production meth-
ods of hydrogen from water, should be found by extracting hydrogen
fro the side gas and waste gases of metallurgical production and metal
vapour processes.
The currently used metallurgical equipment does not satisfy require-
ments of technical progress. For example, the most advanced metallur-
gical system electric furnace is highly sensitive to the proper-
ties of the charge (electrical conductivity, homogeneity of the compo-
sition), requires special carbon electron materials, and its upper power
limit, controlled mainly by varying the current, is restricted. Shaft fur-
naces, including blast furnaces, require special preparation of ore, use
coke which is in short supply, and combine two opposite processes
oxidation and reduction which complicates control and increases the
consumption of reagents and results in the formation of a large amount
of gases, and consequently, increased dust emission.
Plasma electrothermal technology is to a large degree free from
these shortcomings and enables the use of high voltages and relatively
low current thus increasing the efficiency of heat exchange and elec-
tric efficiency of equipment.
One of the simplest processes used in plasma metallurgy is plasma
remelting. In contrast to conventional arc and vacuum-arc melting, this
process is characterized by considerably high stability of the energy
parameters, the possibility of using various gas media and enable pro-
duction of a highly clean metal with the purity comparable with that
obtained in electron beam and electron slag remelting. The loses of
alloying components of the alloys at the atmospheric or increased pres-
sure are minimized. The reduction or neutral atmosphere may be used
in combination with refining slags thus increasing the refining effi-
ciency of the metal. In contrast to electro slag remelting, the require-
ment of high electrical conductivity is not applied to the slag.
In comparison with vacuum furnaces, in plasma-arc furnaces it is
possible to induce a slag and blow liquid metal. The plasma-arc fur-
naces have a number of labour hygiene and ecological advantages:
there is no emission of smoke in to the atmosphere, the noise level is
lower, and the effect on the operation of the energy system is reduced.
71
A furnace with a ceramic crucible is a system of mainly periodic
action which has one or several vertical plasma generators and a bot-
tom electrode.
The metal is refined by removing the metallic inclusions, gases dis-
solved in the metal and other harmful impurities using different met-
allurgical reactions, which is deoxidation with hydrogen ensuring a
high purity of the billet. Active deoxidizing agents in plasma treatment
are also hydrocarbons, it is possible to select the conditions under
which the degree of carburization is minimum.
Alloying the metals with an arc plasma flow [6, 1] is used mainly
as a nitriding process in production of nitrogen-containing steels; it
appears promising to use this method for nitride hardening of refrac-
tory metals.
The interaction of melts with the hydrocarbon containing plasma
suitable conditions are created for alloying these melts with carbon
directly from the gas phase. One of the advantages of gas-phase alloy-
ing is a more uniform distribution of the hardening phase over the vol-
ume of the billets [1].
On the whole, development of industrial technology equipment for
plasma remelting in furnaces with a ceramic crucible should be attrib-
uted to Soviet (Russian) plasma metallurgy. The results of these inves-
tigations were used to develop a plasma section equipped with furnaces
with melting plasma generators with a tungsten cathode and a hollow
graphite cathode with a unit power of 10 MW.
Approximately 150 grades of steels and alloys are melted, includ-
ing: steels and alloys with very low carbon content or the carbon con-
tent controlled over a narrow range, steels alloyed with nitrogen, com-
plexly alloyed steels containing easily oxidized expensive components,
complex steels and alloys where it is important to have a stable chemi-
cal composition in a very narrow range, and creep-resisting and pre-
cision alloys.
Plasma furnaces were constructed and developed in the former
USSR. These furnaces have been successfully introduced and operate
in a number of plants producing high-grade steel in Freital (formerly
GDR), including a 30 t plasma furnace producing 10~000 tonnes of
corrosion resisting and alloy steel (around 40 grades) [1, 2]. Conse-
quently, this experience and East German licence were used to con-
struct the most powerful plasma steel melting furnace by the Austrian
Company Voest Alpine.
Tests of ac plasma generators, carried out at the Chelyabinsk Met-
allurgical Plant, have shown a number of problems in their service:
unstable plasma jet; at large distances between the plasma generator
72
and the surface of the melt, the discharge is closed between the plasma
generators, thus disrupting their operating regime and causing rapid
failure; the amount of energy required for cooling is high; the gas con-
sumption rapidly increases, etc. It is therefore recommended to equip
plasma furnaces with mainly dc plasma generators.
The plasma arc as an energy source for melting may also be used
in combination with other sources.
Low-pressure plasma generators are promising for plasma remelting
processes. This is caused by the following factors: the highly
nonequilibrium nature of the plasma of this type enables intensification
of chemical reactions; the loses of energy as the result of convection
and radiation decrease because the transfer of energy to the processed
object is carried out mainly by electrons; low-pressure plasma flows
are easier to control using, for example, electric or magnetic fields.
Vacuum plasma-induction melting has been successfully developed
in the former USSR (Russia). This method includes the advantages of
plasma-induction (with special efforts to an inert atmosphere developed
by the Japanese Company Daido Steel Corporation) and vacuum melt-
ing.
In comparison with conventional vacuum-arc remelting, vacuum-
plasma refining remelting enables more flexible control of solidifica-
tion and refining.
It should be mentioned that the first Soviet experimental vacuum
plasma furnace for melting metallic materials was constructed by the
A.A.Baikov Institute of Metallurgy. Fundamental results were obtained
and it was shown that the vacuum plasma arc can be used efficiently
to process iron and refractory and active metals, and the energy struc-
ture of the process of equipment has been analysed.
In the group of plasma melting processes a special position is oc-
cupied by reduction melting in which the oxide raw material (ore con-
centrates) are subjected to plasma heating in the presence of the reduc-
ing agent (carbon) gaseous reducing agents (hydrogen, natural or con-
versed gas).
After developing the process using the system constructed by
Tetronix Company (1.4 MW), South African experts developed equip-
ment with a power of 10.5 MW and a productivity of 50000 t of fer-
rochrome per annum (it is possible to produce 45% of ferrochrome in
processing complex chromite ore with the degree of extraction of chro-
mium increased by 15-20%, a reduction of the phosphorus and sulphur
content, resulting in a high degree of extraction of platinoids, nickel
and cobalt). In cooperation with Quebec Ferrotitan Company, the most
powerful (107 MW) plasma six-electrode furnace with a liquid bath for
73
processing local fine-grained ilmenite was also developed in South
Africa.
The A.A.Baikov Institute of Metallurgy, Russian Academy of Sci-
ences, and Tsentroenergotsvetmet Production Association developed and
constructed on an industrial scale at the Yuzhuralnikel Concern a proc-
ess of reduction melting metals of the iron group (nickel, cobalt) us-
ing a plasma furnace with a ceramic crucible, gas, solid and combined
reducing agent with preferential use of a hollow graphite cathode [1,
2].
In comparison with the previously used technology and equipment,
the rate of reduction increases, the losses of expensive and difficulty
available metal with dust emission and slags are used, the consump-
tion of graphite material in the reducing agent is lower, the number of
processing cycles is also lower and the working conditions are im-
proved.
In the area of reduction plasma metallurgy of ferrous metals signifi-
cant results have been obtained by the company SKF Steel (Sweden)
where a number of processes were developed in the shaft variant.
In the former USSR, the development of reduction processes in
shaft-type system equipped with jet plasma generators, was on a con-
siderably smaller scale (total power up to 2 MW). In production of
ferroalloys it was shown possible to melt silico-manganese and carbon
low-phosphorus ferromanganese in a single stage.
An important role played in modern metallurgy by the methods of
ladle refining requires heating the metal in the ladle-furnace system.
The use in the ladle-furnace of conventional arc heating is restricted
by carburization of liquid metal and by the well-known shortcomings
of alternating current. The use of direct current and plasma heating
alleviates these problems and enables the temperature of metal to be
regulated during the process of correcting its chemical composition for
continuous casting.
Slurry containing a nonferrous metal is a specific waste of metal-
lurgical production. In particular, the metallurgical plants of the CIS
now recover every year several million tonnes of slurry with a high
zinc content and this is transferred to the blast furnace slurry from ore,
and steelmelting slurry from metal scrap. Secondary use of zinc-con-
taining slurry leads to the build up of zinc in the working space of
blast furnaces and this disrupts the operation and damages the lining
of the furnace mouth.
The plasma processes of the SKF Company, Plasmadust and
Plasmazink, realized at the plant in Landskrona, are well-known.
A group of investigators at the Gintsvetmet, Tulachermet, The In-
74
stitute of Metallurgy and Tsentroenergotsvetmet carried out calculations,
laboratory and pilot-plant experiments to test gas carbothermal reduction
in various systems with a liquid pool and plasma heating.
The conditions of complete (99.5\%) sublimation of zinc and reduc-
tion of iron with production of low-carbon steel were determined; the
parameters of these processes are not inferior to those of the Swedish
processes Plasmadust and Plasmazinc.
The processes in which the process material (usually, the dispersed
material or dispersed during introduction) is added to the plasma jet
belonging to the jet-plasma processes and as regards the nature of the
jet (inert or chemically active) can be subdivided into the processes of
plasma processing and plasma reduction and synthesis. The reactors
for plasma jet processes are closely linked, as regards their design,
with plasma generators and should ensure interaction of the material
with plasma-forming gas, the development of the reaction zone, the ef-
fective influence of quenching media, sampling of the product and also
removal of heat. As regards the nature of interaction of the jet with the
process material, the reactors are subdivided into direct flow, counter
flow (including those with a fluidized bed), with counter jets, and
multijet.
The investigations in the area of development of materials by
plasma arc dispersion have resulted in new processes of producing
spheroidised, modified and high-dispersion powders which are promis-
ing for spraying wear-resistant coatings and producing cermet and po-
rous components.
The processes of plasma spheroidizing have been tested and applied
on a small scale for a number of materials, for example, powders of
zirconia, tungsten carbide, titanium carbide and a number of other
metals, oxides, and carbides.
In plasma dispersion of dense materials in the form of rods or wire
and also in heating and melting dispersed compounds, it is possible to
obtain a wide range of spheroidized powders of pure metals, alloys and
compounds with the particle size varying from fractions of a micron
to several millimetres and, under the corresponding conditions, hollow
or porous particles, clad and composite powders.
The combination of heating and melting in plasma with simultane-
ous evaporation of some metal enables us to produce powders with
grains coated with metallic shells, formed during heterogeneous con-
densation.
Plasma can be used for a relatively simple evaporation of com-
pounds, including those with the highest melting point, and fair con-
densation. This is a basis of development of the processes of produc-
75
ing ultrafine powder with the particle size varying from tens to thou-
sands of angstroms [3, 18]. In pressing and sintering ultrafine powder
it is possible to obtain blanks and components with high density and
mechanical strength. The high reactivity of the ultra dispersed powders
greatly reduces the energy losses in producing dense components from
them.
One of the most developed processes is plasma spraying of coatings.
Plasma equipment is fabricated for depositing coatings to protect
against corrosion and increasing the wear resistance of components.
Plasma technology enables initial products, starting material and
semifinished products produced for fabricating composite material.
Examples of plasma jet chemically active processes may be the di-
rect synthesis of compounds from powders of metals in nitrogen
plasma, generated by high-frequency plasma generators, with produc-
tion of refractory nitride of titanium, aluminium and zirconium and
other metals; synthesis of carbides in nitrogen plasma with addition of
carbons; synthesis of compounds in shf plasma; synthesis of pigment
dioxide in oxidizing hf plasma from titanium tetrachloride [19].
Tests were carried out on various scales in the former USSR to
develop processes of producing highly dispersed powders of W, Mo,
Ta, Nb, carbides of Ta, Nb, Ti, Si, B, W, V, Zr, Hf, nitrides of Ti, Si,
Nb, Ta, B, V, carbonitrides, borides, silicides, and other compounds in
reduction of oxides, chlorides, and fluorides in contact with hydrogen,
natural gas, ammonia, nitrogen and when using arc, HFI, HFC and
SHF plasma generators [20, 21].
Plasma-hydrogen reduction in jet reactors has been described in
sufficient detail in [1, 2, 3]. The optimum area of application is the
industry of refractory metals based to a large extent on using hydro-
gen as a reducing agent and a medium for sintering and treatment proc-
esses.
On the whole, plasma chemistry and plasma-hydrogen processes of
reduction and synthesis may be regarded as a source of producing
highly dispersed powders having, in a number of cases, unique prop-
erties ensuring their extensive use in various areas of technology.
A considerable achievement of Soviet science and technology is the
development of the ecologically clean highly efficient process of plasma
hydrogen reduction. The A.A.Baikov Institute of Metallurgy of the
Russian Academy of Sciences, together with a number of organizations
(ChFVNIITS, UzKTZhM, Tsentroenergotsvetmet and VNIIETO) have
applied this process to the reduction of tungsten oxides to produce
highly dispersed powders at the Uzbek Concern of Refractory and
Creep resisting Metals.
76
Only in the stage of producing tungsten powder the specific labour
requirements are reduced by 2.6 times, and the consumption of hydro-
gen 1.5 times with a large increase of productivity in comparison with
standard technology. This demonstrates convincingly the efficiency of
energy and resources supply facilities with the optimum combination
of plasma technology in the metallurgical process. It should be men-
tioned that this relationship has also been reflected in a number of
other proposals by the A.A.Baikov Institute of Metallurgy applied on
the industrial or enlarged scale.
Unique properties of the highly dispersed powder which have not yet
been completely investigated, regardless of extensive investigations of
their properties [22], facilitate compacting (reduction of sintering tem-
perature, reduction of the number of rejects caused by insufficient
melting) and improvement of the service properties of components pro-
duced from it, including the wear resistance of hard alloys.
At present there are suitable conditions for extensive development
of plasma metallurgy: sufficiently high level of theoretical develop-
ments; the results of a large number of search investigations describ-
ing the main types of plasma-metallurgical processes and equipment for
their realization, efficiency, suitability and the most promising areas of
application of plasma technology.
An essential condition for development for plasma metallurgy is the
provision of a material base for pilot plant investigations and the de-
velopment of subsequent series production of standard metallurgical
equipment. Some of the plasma-metallurgical processes may be realized
with the minimum investment by re-equipping the existing unprofitable
metallurgical plants and systems for plasma heating.
Returning to the problems of metallurgy in the near future, it should
be mentioned that the harmful effect of industrial production on the
environment will be greatly changed by radical transformation of
mainly power engineering and metallurgy which provide the largest
negative contribution.
In ferrometallurgy, the most marked ecological detrimental effect,
equalling 25% of the total production cost of steel by the blast fur-
naceconverter system, is provided by coke chemical, and sinter pro-
duction.
The general trend to excluding these processing methods by devel-
oping various processes and systems of direct reduction (coke-free met-
allurgy), including plasma reduction is not accidental.
However, the author favours the concept developed by him in co-
operation with N.N.Rykalin, N.P.Lyakishev and other investigators
according to which the optimum solution of ecological and technologi-
77
cal problems of metallurgy should be obtained by developing the en-
ergy technological complex.
These concepts are in agreement with the ideology of nuclear and
metallurgical industry and atomic hydrogen power engineering. These
concepts were used as a basis for design studies for the Kola region
of Russia in constructing the Kolsk Nuclear Power Station and the
Olenegorsk iron ore deposits. It was planned to use the electric and
thermal energy of the nuclear power station to produce hydrogen an
agent for reducing the superconcentrate of the Olenegorsk enrichment
factory with production of high-grade metal, including iron powders.
It was proposed to use plasma technology for producing and heat-
ing hydrogen. The Chernobyl catastrophe resulted in cancellation of
this project.
Subsequently, the effort of teams of the Institute of Metallurgy of
the Russian Academy of Sciences and Tulachermet were directed at
finding possibilities of developing an experimental energy and techno-
logical complex in other regions of the country (Tula, Erevan,
Semipalatinsk) in the direction of possible replacement of the nuclear
power station by an ecologically pure clean thermal electric power sta-
tion [23, 24]. It is efficient to replace combustion of solid fuel by its
gasification ensuring a large reduction of harmful emissions.
One of the most promising systems of ecologically clean thermal
electric power stations is that based on the development of investiga-
tors at Novosibirsk (Institute of Electrical Engineering; Institute of
Thermophysics, Siberian Division of the Russian Academy of Sciences)
with a plasma thermal reactor [25]. As regards one of the variants of
development of the metallurgical plant of the future, joint investigations
were carried out by Sibgipromez and the Institute of Economics of the
Siberian Division of the Russian Academy of Sciences to evaluate the
possibilities of using the synthesis gas for direct production of iron and
in blast furnaces with promising technical and economic parameters.
The core of the complex is the ecologically clean thermal produc-
tive power station (and also possibly safe nuclear power station) which
produces from low-grade carbon fuel and biomass a synthesis gas used
for production of electric and thermal energy, and in secondary chemi-
cal production (together with synthetic coalproduct of carburization
of biomass) in ecologically clean plasma metallurgical production [26,
2].
The metallurgical plant of the future is an integrated or miniplant
in the composition of the energymetallurgical complex, and is fully
automated. This is also the result of the continuous nature involving
nature of technological processes. In comparison with the conventional
78
system, coke chemical, sinter production and blast furnace processing
are excluded. The main apparatus are technological; modules are re-
duction and meltingrefining processing. Variants of these systems are
a shaft furnace for direct reduction, a melting reduction furnace with
a hollow cathode, steel melting furnaces and a furnace/ladle with
plasma heating; in the near future it will be a continuous plasma sys-
tem: oresteelrolled stock combining the modules of prevalent reduc-
tion in the solid phase, additional reduction in refining in the liquid
phase followed by continuous casting.
CONCLUSIONS
Summing up these considerations, it should be mentioned that the work
of Russian investigators has contributed significantly to the develop-
ment of all scientific and technical prerequisities for extensive appli-
cation of plasma metallurgy: on a sufficiently high level of theoretical
investigation it was possible to develop fundamentals of the theory of
processes of plasma reduction and synthesis, a procedure has been
developed for investigating plasma processes, including calculations of
high-temperature thermodynamics, the kinetics of heat and mass ex-
change processes and condensation using mathematical modelling by
means of developed models of interaction of plasma with the substance;
the main types of plasmametallurgical processes and apparatus have
been developed, including those for system analysis; the efficiency,
suitability and most promising areas of application of plasma technol-
ogy have been evaluated and its results have been confirmed by a
number of examples of practical application of plasma metallurgical
processes ensuring an increase of productivity, energy and resources
saving, improvement of the working conditions and satisfactory com-
patibility with the environment. An essential condition of development
of plasma metallurgy is the provision of a material base for experimen-
tal and industrial investigations and development and serious produc-
tion of standard plasma apparatus.
Some of the plasma metallurgical processes may be realized with a
minimum capital investment by re-equipping the existing unprofitable
metallurgical production and aggregates for plasma heating.
However, the radical solution of the problem should be achieved by
developing by means of plasma technology the energy and technologi-
cal complexes combining the ecologically clean thermal power station
(or safe nuclear power station), producing electric energy, heat and
processing (reaction) gases used in plasmametallurgical production
(integrated or miniplant) which produces, from ores and industrial
79
waste, metals, alloys and compounds of the required form and quality
on a high ecological and technological level.
References
1. Yu.V.Tsvetkov, et al., Plasma metallurgy, Nauka, Novosibirsk, 1992.
2. Yu.V.Tsvetkov, i n: Thermal pl asma and new materi al technol ogy,
vol.2, Cambridge Interscience Publishing, 1995, pp.291-322.
3. Yu.V.Tsvetkov and S.A.Panfilov, Low-temperature plasma in reduction
processes, Nauka, Moscow, 1980.
4. Yu.V.Tsvetkov, i n: Physi cs and chemi stry of pl asma metal l urgi cal
processes, Nauka, Moscow, 1985, pp.5-18.
5. L.S.Polak, et al., Theoretical and applied plasma chemistry, Nauka,
Moscow, 1975.
6. A.A.Erokhin, Plasma-arc melting of metals and alloys. Physico-chemi-
cal processes, Nauka, Moscow, 1975.
7. A.L.Suris, Thermal dynamic analysis of high-temperature processes,
Moscow, 1978.
8. G.B.Sinyarev, et al., Using computers for thermal dynamic calcula-
tions of metallurgical processes, Nauka, Moscow, 1989.
9. A.I.Pustovoitenko, et al., in: Physics and chemistry of plasma met-
allurgical processes, Nauka, Moscow, 1985, pp.434-445.
10. Yu.V.Tsvetkov, i n: Effi ci ent uti l i zati on of ores and concentrates,
Nauka, Moscow, 1989, pp.27-37.
11. N.N.Rykalin, et al., in: Atomic hydrogen power engineering and tech-
nology, No.2, Atomizdat, Moscow, 1978, pp.16-31.
12. A.V.Nikolaev, in: Physics and chemistry of plasma metallurgical proc-
esses, Nauka, Moscow, 1985, pp.67-93.
13. A.V.Nikolaev, Stal, No.11, 14-18, 1993.
14. N.N.Rykal i n and L.M.Soroki n, Metal l urgi cal hi gh-frequency pl asma
generators, Nauka, Moscow, 1987.
15. Yu.V.Tsvetkov, in: Atomic hydrogen power engineering and technol-
ogy, Energoatomizdat, Moscow, 1988.
16. V.A.Legasov, Priroda, No.3, 14-23, 1977.
17. V.A.Frolov, et al., Metally, No.3, 1989.
18. R.A.Andrievskii and N.M.Nuzhdin, Poroshk. Metall., vol.2, Moscow,
1986.
19. I.V.Antipin, et al., Fiz. Khim. Obrab. mater., No.4, 146-148, 1968.
20. N.V.Alekseev, et al., Poroshk. Metall., No.8, 1-4, 1980.
21. V.D.Parkhomenko, et al., Plasma chemical technology, Nauka, Nov-
osibirsk, 1991.
22. R.U.Kalamazov, Yu.V.Tsvetkov, High-dispersion powders of tungsten
and molybdenum, Metallurgy, Moscow, 1988.
23. V.Frol ov, et al ., Journal of hi gh-temperature chemi cal processes,
vol.1, No.9, 1992.
24. V.Frol ov, et al ., Inst. J. Hydrogen energy, vol .18, No.8, 665-672,
1993.
25. G.Nozdrenko, et al., Izv. Vuz. Energetika, No.8, 68-74, 1988.
26. N.P.Lyakishev, Yu.V.Tsvetkov, et al., in: Proceedings of the interna-
tional conference, Ferrous metallurgy in Russia and CIS in the twenty
first century, Moscow, 1994.
80
Atmosphere Plasma Spraying: Theory,
Modelling, Diagnostics, Computer-Aided
Design and Some Applications
O.P. Solonenko
Inst. of Theoretical and Applied Mechanics, Siberian Branch, Russian Ac. Sci.
Novosibirsk, Russia
1. INTRODUCTION
Plasma spraying is one of the promising methods of producing the
coatings and composite materials. Because of a wide temperature
and dynamic range of plasma jets and the possibility of using vari-
ous plasma forming media (neutral, oxidising, reducing, etc.), it is
possible to combine in a single technological process both phase and
chemical transformations, ensuring the required modification of ini-
tial powders, and spraying materials with the required structural het-
erogeneity [1-11].
However, plasma spraying is a highly science-intense technology
the potential of which is far from completely utilised because the
plasma torchhigh-temperature technological flowsprayed mate-
rial system has been studied insufficiently. Advances in this area
can be made only by formulating detailed investigations which would
gradually include all links of the chain of formation of a stable final
product with the required properties guaranteed not only under labo-
ratory but under industrial conditions one.
To formulate these investigations, we have proposed the concept
of the complex experiment (CE) in plasma spraying [7,12,13].
Gradual application of this concept makes it possible to improve
greatly the reliability of the results of fundamental and applied in-
vestigations by a rational combination of the possibilities of physical
and computing experiments. For efficient application of CE it is es-
sential to make rapid advances in understanding the physical proc-
esses forming the base of this and of a number of related technolo-
gies (spheroidization, densification and evaporation of powders;
spray casting process; microatomization of powders, etc.), and it is
81
Atmosphere Plasma Spraying: Theory, Modelling, Diagnostics, ....
also necessary to develop and improve methods for their effective
investigation and optimization.
In the last 20 years, a number of reviews have been published in
the area of thermophysics of plasma spraying [4-6, 10-12], including
the review [13] of the author of this paper. They reflect the state of
investigations of a different problems forming the basis of this ex-
tremely important and promising technology.
In this paper, we shall focus our attention only on discussion of
the key problems connected with the entire chain of thermophysical
processes determining the chain plasma torchformation of dusted
technological flowsprayed coating in the context of the improve-
ment of conventional technologies of atmosphere plasma spraying
(APS). The main accent will be placed on the results obtained by us
after publication of the review [13].
2. FUNDAMENTAL PROBLEMS OF PLASMA SPRAYING
Since the deposited coating is a heterogeneous material, the main
aim of plasma spraying is to produce the homogeneous layers with
required structure over the spraying spot. For efficient design and
control of the structure of sprayed materials, it is important to make
further progress in solving the following principal problems [12,13]
(see Fig.1).
It is evident that the specific content of the program of investiga-
tions and developments in solving the problems described previously
should be determined within the framework of the specific concep-
tual approach forming the basis of the equipment development and
Fig.1 The basic chains of the plasma spray technology and main directions of our
investigations.
Thermal
plasma
torch
Powder
injection
Dusted
plasma
jet
Formation
of sprayed
material
Post
treatment
of sprayed
material
Powder
material
metals
ceramics
composites
DC torches
type 1
type 2
type 3
RF torches
one-tube
two-tube
multi-tube
radial-annualr
slit
modelling
diagnostics
theory
modelling
diagnostics
laser beam
electron beam
electric arc
plasma jet

82
technological process improvement. We believe that one of promis-
ing approaches have been proposed by us in [12,13]. This approach
is based on three principles:
(i) development, investigation and application of the plasma
torches with interelectrode inserts (IEE) (Fig.1, type 3) both with
laminar and turbulent jet outflow ensuring guaranteed distributed
(diffusion) attachment of the arc on an anode [1416]; this ena-
bles axisymmetric plasma flows to be generated;
(ii) improvement of the methods of powder materials injection
into the plasma flow by transferring to radial-circumferential injec-
tion of the powder to obtain an axisymmetric radially converging
flow of particles directly beyond the zone of anode attachment of
the arc;
(iii) application of replaceable output attachments, including the
steps and diffusers with hot walls, enabling the gas dynamic struc-
ture of the flow to be affected in order to create suitable conditions
for subsequent efficient treatment of the powder and its protection
against the detrimental effect of the environment (oxidation, disso-
ciation, gas saturation, etc.).
The purposeful development of this approach is promising for
practically all science-intensive high-temperature technologies taking
place in the presence of the condensed phase because, as the result
of axial symmetry in the entire generated gas-disperse processing
flow, it is possible to optimise this flow on the basis of combined
physical, computing and materials science experiments [4,5,9].
It should be mentioned that all these approaches require optimisa-
tion of the technology of powder plasma spraying by optimising the
equipment and thermal- and gasdynamic conditions of the process.
However, another method of optimising the technology, with other
conditions being equal, is optimising by designing new powder mate-
rials with specified unique properties and development of equipment
and technology which would enable the coatings inheriting the prop-
erties of the initial sprayed material (for example, see papers [17-
19]).
2.1. DC plasma torches
Investigation of the physical relationships governing burning of the
electric arc with an axial gas flow blown onto the arc, and examina-
tion of the possibilities of controlling the arc by attachment to the
output electrode/anode in order to design DC plasma torches ensur-
ing the required quality of the out-flowing jetstationary nature and
axisymmetric gasdynamic structure, and also the reproducibility of
83
the gasdynamic characteristics of the jet during their repeated acti-
vation.
It is pertinent to note that the main type of plasma generators
used in powder spraying are the DC plasma torches with
stabilisation of the arc by the channel wall and, in particular, the
simplest type of these plasma generators the plasma torches with
the self-setting mean arc length (type 1) and with the arc length
fixed by the direct or reverse step (type 2). These plasma torches
are used in the conventional equipment manufactured by companies
Metro, Plasmatechnik, Castolin, SNMI, MTS, Plasmadyne, Miller
Thermal Inc., Snecma, the spraying plasma torches used in Poland,
etc. When using the plasma torches with self-setting mean arc
length, this results in low-frequency large-scale shunting between
the arc column and the channel wall. Disadvantages of this type of
plasma generators with special reference to the spraying process
and related technologies of powders processing have been examined
by us in [13]. In our publications [14-16] we have presented the re-
sults devoted to design and study the DC plasma torch with an
interelectrode insertion (IEI) and distributed (diffusion) attachment
of the arc on the anode surface.
2.2. Methods of powder injection
The development and examination of the possibilities of the new
methods of powder materials injection into the plasma flow which
enable the degree of concentration of the dust-laden jet to be regu-
lated and ensure high uniformity of the distribution of the particle
velocity and temperature in its cross sections for the spraying sys-
tems, and also greatly increase the productivity and efficiency of
the technological process, including the increase a resulting from the
powder flow rate/transport gas flow rates ratio. According to our
publications [2022], the productivity and efficiency of the plasma-
jet powder processing may be increased essentially.
Modelling of powder injection productivity and efficiency of the
powder materials processing in plasma jets are mainly determined
by the gas particles flow structure inside the nozzle and at initial
sections of high-temperature dusted jet. In the majority of techno-
logical processes, based on the use of the plasma torches with the
internal electric arc, the powder is injected before the nozzle exit
section.
Several variants and their combinations are possible in this case,
such as: (i) cross injection into the nozzle behind the anode arc at-
tachment spot, (ii) powder injection into the plasma torch channel
84
together with a plasma-forming gas (this method may be modified so
that the powder is injected by means of an additional transporting
gas between the cathode and the anode), (iii) axial injection of parti-
cles by means of transporting gas through the cathode. A compara-
tive analysis of the advantages and disadvantages of the above
methods have been analysed in a review paper [13]. The main con-
clusions of this analysis are: (i) as far as efficiency is concerned,
the third method may be rather good, but there are some problems
if it is necessary to provide the maximum possible productivity, (ii)
from the productivity, efficiency and simultaneous simplicity of the
technological process viewpoints the first one, in our opinion, is most
promising, in the case of radial-annular injection of particles into the
plasma torch channel behind the arc attachment spot. These meth-
ods are promising for intensification of almost every high-tempera-
ture technology of disperse materials processing (thermal spraying,
spheroidization and evaporation of powders, etc.). Due to axisym-
metricity of the heterogeneous plasma flow, in the case of the diffu-
sive anode attachment of the arc spot, it is possible significantly
simplify the optimisation of the technologies by using the conception
of joint physical and computational experiments [7,12]. Table 1 rep-
resents the results of the calculations for maximum productivity
k
eff
=G
p
/G
f
and efficiency h
eff
at radial-annular injection of powder
into the diffuser with a fixed length L/D
c
.
Heat efficiency is the ratio of the particles enthalpy flow at
the nozzle exit cross-section and the inlet enthalpy flux of plasma
flow. Here G
p
and G
f
are the powder mass flow rate and flow rate
of plasma-forming gas. Calculations have been made for Al
2
O
3
-par-
ticles having diameter of D
p
= 50, 100 mm for three gas mass flow
G
f
, g/s 0.5 0.5 1.0 1.0 2.0 2.0
D
P
, m 50 100 50 100 50 100
V
o
p
, m/s 1.8 1.0 2.8 1.6 4.3 2.7
K
(1)
max
3.95 3.9 3.9 3.85 3.9 3.8
(1)
max
0.64 0.61 0.63 0.62 0.63 0.64
V
o
p
, m/s 1.9 1.1 3.1 1.9 4.8 3.0
k
(2)
max
3.1 3.0 3.1 2.9 3.05 2.7
(2)
max
0.71 0.69 0.71 0.67 0.70 0.62
Table 1 The estimates of maximum productivity and efficiency
85
rates G
f
(air, T
f
= 6000 K, D
c
= 10
2
m). Powder injection velocity
V
p
o
in each specific case was chosen provided the stream of parti-
cles at the nozzle-exit cross-section does not intersect the flow axis
and is 0.05D
c
away from it. Superscript indices 1,2 correspond to
max
and k
max
for those modes when it was required to melt the par-
ticles only or both to melt and overheat them up to temperature
T
p*
= (T
pm
+T
pb
)/2, where T
pm
and T
pb
are fusion and boiling tempera-
tures of the particle material. The analysis of the results presented
in Table 1 shows that using the radial-annular injection of the pow-
der one can attain high productivity and efficiency of process.
At the same time, the systems of powder injection directly into
the arc burning zone are rather problematic due to the possible in-
stability of the arc in the presence of the powder and its vapour,
and requires the additional investigations.
2.3. Formation of the dusted plasma technological flow
Modelling. The main problems consist of examination of the
interphase exchange of the momentum, heat and mass in high-tem-
perature heterogeneous jets of multicomponent gas and, in particular,
of jets impinging onto barriers, taking into account the flow colli-
sions in the jet, increased loading of the jet with the powder, its
polydisperse nature, prior history of the particles injection, their
complex aggregate state in the flow, and also the stochastic nature
of the local distribution of the velocity, temperature and size of par-
ticles in the cross sections of the jet, including the spraying spot.
The flows, arising at plasma spraying, are high-temperature two-
phase turbulent ones. To modelling these flows a method based on
the Eulerian description for a gas phase and the Lagrangian one for
the disperse one is widely used.
This approach was applied to modelling of particles heating in in-
duction plasma [23] and plasma plume [22,2427]. The model [28]
developed on the basis of this method was applied to numerical
analysis interphase momentum heat and mass transfer under evapo-
ration of Si-particles in nitrogen plasma [29].
In all these works the gradientless model of particle heating was
used for description of the disperse phase. Though a small particle
sizes justifies application of this model, the estimations show, that
the error can be significant for the case of widely used ceramics,
such as Al
2
O
3
. Since the accurate prediction of a particle thermal
state is the main purpose of the modelling, it is necessary to com-
pare these results with the ones obtained by a more exact method
taking into account the temperature gradient inside the particle.
86
In this connection, it is very important to formulate a new ap-
proaches for Eulerian and Lagrangian computing of the turbulent
dust-laden plasma jets and their combinations with the purpose to
improve the quality and reliability of the obtained numerical results,
including the accuracy of the description of complicated technologi-
cal flow.
To illustrate the possibilities of such approach the comparison
was made of the results obtained using two models of particle heat-
ing in plasma. The widely used model gradientless heating was ap-
plied for modelling of two-phase plasma jet temperature and velocity
fields. Calculation were conducted for the jet under dense loading
conditions. The modelling of particle thermal parameters was per-
formed by the method accounting for the gradient of temperature in-
side particles in the previously obtained numerical velocity and tem-
perature fields of the carrier gas flow, corresponding to the same
dusted plasma jet [22].
In this study, the method [7] was applied for modelling the parti-
cle heating which takes into account the temperature gradient inside
it. This approach uses the modification of the control physical vol-
ume (CFV) method to account for particle melting. The particle vol-
ume is divided into spherical layers in such a way that the front of
melting is the common boundary surface of two adjacent layers. It
is assumed that the temperature distribution inside any layer is para-
bolic. Using integration over CFV, the heat conduction equation is
reduced to the Cauchy problem for an ordinary differential equations
system that determines the temporary evolution of mean temperature
inside layers. To make this system equivalent to the initial boundary-
value problem, we used a procedure based on recovering parabolic
coefficients from the known mean values of temperature and conti-
nuity conditions for temperature and heat flux. The procedure also
uses the boundary conditions at the particle external surface and
centre and at the melting front. Full description of the method was
given in [7]. The results of the paper [22] were chosen for realisa-
tion of the aforesaid comparison. The mathematical model [22] in-
cludes widely used (k-)-model of turbulence for closing equations
of carrying flow, Lagrangian Stochastic Deterministic model [30] to
describe the particle motion in the high temperature turbulent flow,
Particle Source in Cell (PSI-cell) method [31] for computing parti-
cle-plasma interaction. The model [7,13] was used to take into con-
sideration the suppression of the turbulence energy by the disperse
phase. This model does not require any new empirical constants, so
the standard values of the (k-)-model constants were used:
87
Fig.2 The diagram of flow. Fig.3 The particles distribution by size.
C C C
k h 1 2
1 1 92 0 09 1 0 13 0 9 = = = = = = .43, . , . , . , . , . .
m e
s s s
The geometry of the flow under consideration is shown in Fig.2.
The heterogeneous flow modelled was calculated in two steps. At
first, the flow within the plasma torch channel was calculated. The
distributions of velocity, temperature and turbulence energy at the
initial section were uniform in the flow region behind the area of
axial symmetrical (diffusive) arc attachment. At the outlet from the
channel all variables had boundary conditions of the boundary layer
type. The particles were introduced into the flow normally to the x-
axis at the distance of 0.5D
c
behind the initial section. The distribu-
tion of initial particles co-ordinates along the injector slit were as-
sumed to be uniformly random, while their initial velocities had a
Gaussian distribution with the mean value V
p
=1.64 m/s.
The temperature gradient within the particle was not taken into
account when the thermal state of the disperse phase was modelled,
i.e. the surface temperature was assumed to be equal to the mean
volume temperature. The losses due to particle radiation were taken
into account in the heat balance of the disperse phase. The tem-
perature dependence of the coefficient of emissisivity of Al
2
O
3
was
approximated in accordance with data [32].
The resulting profiles of variables of the carrying flow and the
particles parameters were used as boundary conditions for the cal-
culation of the heterogeneous jet normally impinging on an obstacle.
The calculations were carried out using the following values of
geometrical and regime parameters: diameter of the plasma torch
nozzle D
c
= 10
2
m, its length L
c
= 410
2
m, the width of the in-
jector slit h = 10
3
m, spraying distance L = 0.1 m, plasma-forming
gas (air) flow rate G
f
= 10
3
kg/s, transport gas (air) flow rate G
tr
=
10
4
kg/s, Al
2
O
3
particles flow rate G
p
=(12)10
3
kg/s, initial tem-
88
perature of the plasma T
f
= 6000 K, initial intensity of turbulence
Ka=0.1. The thermodynamic and transport properties of air were
calculated with the help of the data tabulated in [33].
In all calculations, the temperature of the nozzle walls and the
obstacle was taken as constant T
b
= 600 K, the environment tem-
perature was taken equal to T
f ,
= 300 K. The particle size distribu-
tion function is shown in Fig.3. For the calculations it was approxi-
mated by dividing the particles into ten groups. The flow fields ob-
tained are shown in Figs.4 and 5 for two loading ratios (k = G
p
/
G
f
= 1 for the upper half of the figures, k = 2 for the lower ones).
Because the stochastic solution of this problem can be realised
only on a supercomputer, to perform comparison it was supposed,
that the initial particle velocity is equal to the average starting one
in the previous problem. Dynamic interaction and heat transfer be-
tween plasma and particles was modelled using the (same as in
[22]) empirical laws for the drag force [34] and heat transfer [35]:
C
d
p
p
p
f p
f
p
f f
f p f p
= + =
F
H
G
I
K
J
+
F
H
G
I
K
J

24
1 0 15 2 0 6
0 687
0 5 0 33
0 2
Re
Nu . Re , . Re Pr .
.
,
,
,
. . , ,
, ,
.
e j
l
l
r m
r m
Here Re /
p f p f p f
D U U = - r m
H H
is Reynolds number, symbol

corresponds to the parameters at the so-called film temperature

T T T
p f
= + ( ) / 2, the first subscripts f and p correspond to the pa-
rameters of the gas flow and particles, and additional subscripts
or p correspond to the parameters calculated at the local gas or
surface temperature of the particles. Calculations were performed
for the same values of the particle diameters as in base modelling.
The variations of the gas temperature along the ten particle tra-
jectories, corresponding to ten groups considered, are shown in
Fig.4 The isothachs of the carrier gas flow at two loading ratio; up-
per and lower half of picture correspond to k = 1 and 2, accord-
ingly.
89
Fig.6. It is seen that the large particles penetrate deeper into the
high temperature core (initial part of trajectory is not resolved in the
used scale). The sharp temperature fall due to intensive mixing at
the nozzle exit is more pronounced for the smaller particle flow
rate. The temperature level is smaller for the greater particle flow
rate due to intensive flow cooling by particles. The temperature his-
tory of the particles is presented in Fig.7. The dark isotherm corre-
sponds to the melting point of the particle material. It is necessary
to note that for the smaller loading ratio the large particles heating
is slower due to the higher velocities of the carrier flow. The in-
crease of the particle melting degree along the trajectory is shown
in Fig.8. These plots are in accordance with Fig.7 and also show
Fig.5 The isotherms of the carrier gas flow at two loading ratios; upper and lower
half of picture correspond to k = 1 and 2, accordingly
Fig.6 Variation of gas temperature along particle trajectories: a) k = 1, b) k = 2.
b a
90
more rapid melting of the large particles for the higher loading ratio.
The range of particle temperature before impact is 23002700 K
for the low loading ratio and 23802440 K for the higher one. The
temperature differences between these results and those obtained
with the use of the gradientless model are higher for the low loading
ratio and are of the order of 15%.
Hence, to obtain more realistic results while modelling the high-
temperature turbulent gas flow with admixtures of inertia particles,
it is desirable to combine simultaneously two approaches: (i) the
Lagrangian stochastic deterministic model, and (ii) the Lagrangian
modelling of the representative set of single particles with account
for their complex aggregate states in-flight.
Diagnostics. It is impossible to research and develop high-tem-
perature technologies based on powder processing without local si-
multaneous in-flight measuring the temperature, velocity and size of
single particles. During the last decade such complex apparatus has
a
b
Fig.7 Variation of particles temperature along particle trajectories: a) k = 1, b) k = 2.
Fig.8 Variation of particles melting degree along trajectories: a) k = 1, b) k = 2.
a
b
91
be realised on different laboratory levels and improved by a number
of scientific groups in Germany, France, Canada, USA, Japan and
Russia.
Two approaches realising such multiparameter apparatus for dis-
perse phase study and control are based on: (i) optical registration
of the light emitted by the single particle, (ii) combination of latter
and some laser-optical diagnostic method (time-of-flight or LDA).
The first approach at the multiparameter measurements is more
simple and reliable. It is based on mathematical processing of the
pyrometric signal, modulated by a special diaphragm (mask), corre-
sponding to the flying pass single particle.
The new realisation of this approach, proposed by us in [36,37],
appears to have considerable promise for the local simultaneous in-
flight measuring the vector of velocity, size and surface temperature
of single particles. Figure 9 illustrates the sector multislit diaphragm
placed in the object plane of our diagnostic apparatus. This complex
diaphragm includes the following elements: 1) circular slit with ra-
dius of r
0
and with centre O
1
used for starting up the two-channel
analog-digital converter (ADC) by scattered laser radiation at the
moment of its intersection by the particle; 2) the narrow straight slit
S
1
of width h
1
, intersecting the axis Ox under angle a at the dis-
tance l
1
from the point O
1
; 3) the wide slit S
W
, being part of the ring
Fig.9 The principal diagram of diagnostic apparatus (a), the multislit sector diaphragm
(b), and calibration device used (c).
92
with the centre at the point O
1
, with internal radius R
W
and width
h
W
designed for measuring the particle temperature and size; 4) the
narrow straight slit S
2
of width h
2
located at the distance of l
2
from
the point O
1
and perpendicular to the axis Ox; it is not necessary,
although it provides additional information.
As was shown in our publications [36,37], the following measur-
ing accuracy for the alumina particles can be provided: 23% for
the velocity, 58% for the temperature, and less than 10% for the
particle size.
2.4. Formation of sprayed materials
Examination of nonstationary conjugate conductiveconvective heat
transfer and phase transitions under interaction of the heterogeneous
flow and single melted particles with the sprayed surface, and also
further development of the physical fundamentals of coating forma-
tion enabling their laminated structure and service characteristics
(porosity, adhesion, cohesion, thermal conductivity, etc.) are the
main problems to be predicted.
According to the analysis fulfilled in [13], the following basic
variants of melted particlesubstrate interaction
( / ) t D u
d
p po
must
be considered and discussed systematically (Fig.10): (1) spreading
and simultaneous solidification of the droplet on a solid substrate,
(2) spreading, simultaneous solidification of the particle with partial
melting of the substrate, (3) spreading of the droplet over the par-
tially melting substrate, (4) spreading of the droplet on a solid
substrate.
Here and below: = T T
pm
/ is non-dimensional temperature; sub-
scripts p and b correspond to the particle and the substrate; the
additional subscript m characterises the parameter of correspond-
ing material at the melting point; z
p
, , are the instant co-ordi-
nates of the flattening droplet top and fronts of solidification and
melting;
c
is non-dimensional contact temperature.
In the framework of this direction, the creation of map of the
plasma sprayed splats has considerable promise for clear under-

bm c
> < 1
bm c
> 1
bm c
<1

bm c
1
Fig.10 Schematic presentation of different variants of droplet-substrate interaction.
93
a b
Fig.11 The photographs of plasma jets impinging on a flat barrier at different regimes
of outflowing: a - turbulent (Re=820); b - laminar (Re=580).
standing the ways of developing the new generation of plasma
sprayed materials. In this connection, our last publication [12,13,38-
42] can be useful at criterion generalisation, classification and inter-
pretation of the experimental data obtained for the case of low vis-
cosity - high thermal conductivity of melted particle material.
2.5. Post thermal treatment of sprayed material
Examination of a set of processes (heating, phase and structural
transformations in the coating material and the substrate, etc.), ac-
companying the heat treatment of plasma-sprayed materials and
coatings by the means of highly concentrated energy fluxes (electric
arc, plasma jet, electron beam, laser radiation) are now in the stage
of the intensive study in different scientific groups.
Among the different methods of post treatment of sprayed mate-
rial the use of the laminar plasma jet is rather perspective [13-16].
This method is rather simple in a realisation and effective, and, from
the point of view of the heat density flux is comparable with an arc.
Figure 11. shows the photographs of jets leaving the nozzle of the
plasma torch at turbulent and laminar regime of flow out corre-
sponding to two different flow rate of the plasma-forming gas
(nitrogen) at the fixed distance from the outlet of the nozzle to the
barrier (8 gauges). In the second case (Fig.11,b), the density of heat
flux from laminar plasma jet to the surface is of several times
higher in comparison with turbulent one (Fig.11,a).
3. COMPUTER-AIDED DESIGN OF PLASMA SPRAY
TECHNOLOGY
94
For further development of high plasma technologies for material
treatment and production under conditions of their superfast quench-
ing, including interaction with solid surface, it is necessary to carry
out the combined experiment [7,12], covering sequentially all the
links of the chain, defining the formation of a stable final product
with the preset properties. These properties must be guaranteed not
only under laboratory conditions but industrial ones as well. In the
first place, this is related to the plasma spray and surface hardening
technologies.
The problem-oriented core of the complex experiment in plasma
spraying is the dialogue programme complex (DPC) plasma torch
dusted jetcoating [43,44]. This is a multipurpose, constantly pro-
gressing complex of scientific, methodical, programme and informa-
tion means aimed at substantial increase in the efficiency of the
fundamental and technological investigations in plasma spraying and
related technologies. The creation of such complex, incorporating in
compressed form the knowledge base for appropriate subject-matter
fields will be useful for establishment of the inverse link between
the specific practical demands of technologists and scientific labora-
tories, and will stimulate the solution of some scanty investigated
problems, providing uniform accuracy of the calculation of either
phenomena. In this paper the brief survey of the results in the
above-mentioned field of activity is presented.
DPC plasma torchdusted jetcoating is a problem-oriented
complex of programs. It is intended for various categories of users.
On one hand, it should be rather simple to operate. On other hand,
it should be multipurpose, since it is necessary at least for: (i) study-
ing the interphase momentum, heat and mass transfer on the basis
of computer experiment, (ii) joint physical and computational experi-
ment, (iii) approbation and mutual comparison different physico-
mathematical models and experimental data, (iv) training of special-
ists engaged in the technologies of powder materials plasma
processing and, first of all, thermal spraying, (v) solution of specific
applied problems, including optimisation of the regime parameters.
In its turn, DPC orientation for a wide range of users and sim-
plicity of its operation require a thorough consideration of the fol-
lowing problems: (i) the development of simple and effective struc-
ture of the programme packages, (ii) the creation of internal data
structure, providing convenience in programming of the individual
subsystems responsible for simulation of either processes included in
the scope of investigation as well as allowing their broadening, (iii)
the choice and development, if necessary, of sufficiently reliable and
95
universal numerical methods and creation of software on their basis
enabling to synthesise different schemes of computational algo-
rithms, (iv) the development of input languages close to natural de-
scription of the classes considered problems, (v) the provision of the
possibility to operate with different data banks, (vi) the creation of
service modules set for representation and interpretation of the re-
sults obtained, (vii) the provision of error diagnostics throughout the
operation of the complex.
One of the important requirement imposed for DPC is to provide
the possibility to carry out the computational experiment using the
models with different degree of complexity and hence adequacy
while describing either processes constituting the chain of the phe-
nomena. Hence, it is expedient to simultaneous developing the
physico-mathematical models and appropriate software, constituting
the scope of DPC, in the following directions: (i) the development of
analytical methods providing the solution of certain model problems
in the final form, (ii) obtaining the approximate analytical engineer-
ing solutions, (iii) direct computational experiment using the model
with different degree of complexity and dimensionality. A rational
combination of these three approaches allows to promote the optimi-
sation of the regime parameters of the processes as it may be im-
plemented in two stages: (i) high-speed computing set of model
problems permiting the approximate analytical or rough numerical
solution to obtain extreme estimates as well as to investigate quali-
tative the behaviour of divider function and localisation of the opti-
misation region, (ii)subsequent refinement of the process behaviour
in the whole using the more complete models allowing to obtain a
detailed information (e.g. while investigating the melt droplet
substrate interaction, the first stage may consists of prediction of
the contact temperature, the dynamics of the particle solidification
at its deformation on the surface, and the thickness and diameter of
splat after solidification by using the theoretical solutions; the sec-
ond stage of interest may includes the detailed information about the
phase transition, the rate of supercooling in the separate particles
zones at interaction by using more complicated description of com-
plex problem).
The last version of the dialogue programme complex provides: (i)
engineering calculation of linear scheme plasma torches (volt-
ampere characteristics (VAC), thermal efficiency, mean-mass values
of velocity, enthalpy and temperature of plasma flow behind the an-
ode arc attachment spot, etc.), (ii) simulation of powder injection
into plasma torch nozzle behind the anode attachment arc spot (one-
96
sided injection, radial-annular one) and calculation of gasdynamic
parameters of the heterogeneous plasma flow for the nozzle exit
section, (iii) Eulerian and Lagrangian modelling (by using the differ-
ent physico-mathematical descriptions) of the heterogeneous high-
temperature subsonic jet (both laminar and turbulent) of complex
gas composition with polydisperse particles, including the interaction
with a normally placed substrate, (iv) simulation of unsteady state
conjugate conductive convective heat transfer between flattening
and solidifying melted particle and substrate, (v) prediction of the
regularities of sprayed material formation (lamellar structure, macro-
porosity, dynamic of coating growth) dependent on the statistical
distributions of particles velocity, temperature, aggregate state, and
size over the spraying spot.
4. SOME PRACTICAL APPLICATIONS OF THE RESULTS
4.1. Plasma spraying in preparation of metal supported catalysts
The development of new environmentally clean and energy safe
technologies is one of todays important tasks. Examples are gas
and oil processing, motor fuel production, purification of wastewater
containing organic compounds, etc. Catalysts to be used in these
processes must possess high thermal stability and wear resistance
properties.
Application of metal substrates as catalyst supports is becoming
more common. Metals have some advantages that make their appli-
cation in catalysis more attractive than ceramic supports. Their ad-
vantages include a high mechanical strength and heat conductivity
characteristics. Protection of catalytic coatings against mechanical
shock and chemical exposure to the reaction medium and the con-
servation of the catalytic activity at high temperatures are the main
problems when metal-supported catalysts are used. The preparation
of materials with strong contact between the catalytic layer and the
metal surface at high temperatures is also a complicated problem.
Plasma spray of washcoatings of alumina, titania, zirconia, etc.
will allow one to solve successfully the problem of the adhesion of
the active component to the metal surface and to protect the metal
support substrates from aggressive medium and improve their dura-
bility for high temperature use. Moreover, the plasma spray method
will give an opportunity to design composite materials consisting of
different combinations of plasma-sprayed catalytic and protective
layers.
It is important to note that there are no strict limitations on the
geometry of granulated or monolithic supports when plasma spray
97
methods are used. Both substrates of simple geometry (flat sur-
faces) and complex substrates (foam materials and honeycomb
monoliths) can be applied successfully.
Although plasma spray techniques are well developed and in
common use today, information on the application of these methods
in catalysis is rather scarce. Traditionally, plasma spraying is applied
when the formation of dense, non-porous protective layers is neces-
sary and this makes the direct use of this method in catalysis rather
complicated.
Our publications [45,46] are devoted to application of above-men-
tioned DC plasma torch for atmosphere plasma spraying of the
alumina coatings under different regimes of the jet outflow (laminar,
transient or turbulent) on the metal substrates of different
geometries (plates, foams) for subsequent synthesis of perovskite
catalysts.
The properties of the catalysts (Fig.12), developed with the use
of coating sprayed under laminar regime of plasma jet outflow, were
compared with those of catalysts having the same phase composition
and BET surface area but synthesized without the use of the
plasma-sprayed alumina layer. The catalysts calcined at tempera-
tures from 600 to 1000 C demonstrated the same level of activity
as catalysts synthesized using the plasma technique. A difference in
the activity between these types of catalysts appeared when the du-
ration of calcination at 1000 C was increased from 3 to 9 hours.
There was no decrease in the activity of the catalyst with a plasma-
sprayed alumina layer whereas the temperature to give 50% conver-
sion rose from 600 to 655 C for the catalyst prepared without a
plasma-sprayed layer (Fig.13). It appeared that a plasma-sprayed
washcoating layer consisting of alumina on a Ni foam support pre-
vented the nickel from becoming oxidized more successfully and in-
creased the thermal stability of the catalyst.
4.2. Piston of diesel engine with a thermal barrier coating
Zirconia stabilised by yttria powder coating [47], sprayed on a piston
head by the use of the developed plasma torch providing the laminar
plasma jet outflow (Fig.14), possesses an improved service charac-
teristics (adhesion, cohesion, porosity). The coating produced per-
mits: (i) completely eliminate the burnout of the piston head, (ii) to
decrease the operating temperature near the piston ring grooves, re-
sulting in an increase service time of the piston, (iii) to increase the
combustion temperature of fuel mixture, resulting in an increase the
efficiency of combustion process and ecological characteristics of
98
Fig.12 Micrograph of a cross section view of a catalyst consisting of LaCo on a Ni
foam support.
Fig.13 Temperature dependencies of methane conversion (1% CH
4
in air) on catalysts,
V = 400 h
1
: 1 - 10% LaCoO
3
Ni, plasm., T = 1000C, 3 h; 2 - 10% LaCoO
3
/Ni, chem.,
T = 1000C, 3 h; 3 - 10% LaCoO
3
/Ni, plasm., T=1000C, 15 h; 4 - 10% LaCoO
3
/Ni,
chem., T = 1000C, 15 h.
diesel, (iv) to increase the economic feasibility of diesel no less than
5%, and (v) to increase the diesel engine service time as a whole
no less than 20%.
5. CONCLUSIONS
In the review, the author has attempted to analyze the current state
of the thermophysics of plasma spraying, mainly in the context of
problems examined in the last decade in the Laboratory of Plasma
Dynamics of Disperse Systems of the Institute of Thermophysics
(and now of the Institute of Theoretical and Applied Mechanics) of
Siberian Division of the Russian Academy of Sciences.
The individual problems, arising during explanation, were dis-
cussed. The solution of these problems is of considerable impor-
tance for the efficient understanding of the physical processes form-
ing the basis of the plasma torchpowder material injectiondusted
plasma jet formationformation of the coating of single particles -
its subsequent heat treatment chain.
The obtained results of investigations have allowed us to create
the first version of the dialogue programme complex plasma torch
dusted plasma jetcoating for direct computational experiment in
plasma spray technology. The creation and further development of
such problem-oriented software incorporating in compressed form
the knowledge and data banks will be useful by far not only for fun-
damental investigations but for the technological applications as
well. We hope that it will be also useful for more close the feed-
99
back between the technological groups and research laboratories.
For more rapid further development and extension of its possibilities,
in our opinion, it is essential to establish closer co-operation
between different scientific groups, because plasma spraying is
multidiscipline technology. It is obviously that there are a lot of
common points between the plasma spray technology and other ther-
mal spray technologies, therefore from ideological and scientific
points of view this paper may be useful also for another branches of
thermal spray technology.
References
1 N.N.Rykalin, et al., In: Plasma Processes in the Metallurgy and in the
Technology of Nonorganic Materials (in Russian), Nauka, Moscow (1973),
p.66.
2 D.Apelian, et al, International Metals Review, 28, No.5, 271 (1983).
3 N.El-Kaddah, et al, Metallurgical Transactions B., 15B, No.59 (1984).
4 E.Pfender, Pure and Appl. Chem., 60, No.5, 591 (1988).
5 M.Ushio, Proc. of Jap. Symp. on Plasma Chem. Vol.1, (1988), p.187.
6 T.Yoshida, Materials Trans. JIM, 31, No.1, 1 (1990).
7 M.F.Zhukov and O.P.Solonenko, High-Temperature Dusted Jets in the
Powder Materials Processing (in Russian), Institute of Thermophysics
Novosibirsk, (1990).
8 D.Steffens and M.Mack, Pure and Appl. Chem., 62, No.9, 1801 (1990).
9 R.W.Smith and R.Novak, Intern. J. of Powder Metallurgy, 23, No.4, 231
(1991).
10 J.Szekely and R.C.Westhoff, In: Thermal Plasma Applications in Ma-
terials and Metallurgical Processing, Vol.55, N.El-Kaddah (ed), The Min-
erals, Metals & Materials Society (1992).
11 P.Fauchais and M.Vardelle, Pure and Appl. Chem., 66, No.6, 1247 (1994).
12 O.P.Solonenko, Pure and Appl. Chem., 62, No.9, 1783 (1990).
13 O.P.Solonenko, In: Thermal Plasma and New Materials Technology, Vol.2,
p.7, O.P.Solonenko and M.F.Zhukov (eds), Cambridge Interscience Pub-
lishing, England (1995).
14 V.I.Kuzmin, et al, Proc. of the 3
rd
European Congress on Thermal Plasma
Processes (1995), p.405.
th
Int. Thermal Spray Conf., Kobe, Japan
(1995), p.1091.
th
National Thermal Spray Conf., Houston,
USA, (1995), p.264.
17 V.E.Ovcharenko, et al., In: VDI-Gesellschaft Werkstofftechnik, 3
rd
Europ.
Congress on Thermal Plasma Processes, VDI-Verl., Dusseldorf (1995),
p.395.
18 V.E.Ovcharenko and O.P.Solonenko, Proc. of the 14
th
Int. Thermal Spray
Conf., Kobe, Japan (1995), p.1151.
19 V.E.Ovcharenko, et al, Proc. of 4
th
European Conf. on Thermal Plasma
Processes, Athens, Greece (1996).
20 O.P.Solonenko and A.L.Sorokin, In: Plasma Jets in the Development
of New Materials Technology, O.P.Solonenko and A.I.Fedorchenko (eds),
VSP Publishing, Utrecht (1990), p.243.
100
21 O.P.Solonenko and A.L.Sorokin, Izvestiya SO AN SSSR, Ser. Techni-
cal Sci. (in Russian), 4, 73 (1990).
22 O.P.Solonenko and A.L.Sorokin, In: VDI-Gesellschaft Werkstofftechnik,
3
rd
Europ. Congress on Thermal Plasma Processes, VDI-Verl., Dusseldorf
(1995), p.129.
23 P.Proulx, et al, Int. J. Heat Mass Transfer, 28, No.7, 1327 (1985).
24 P.Proulx., et al, Proc. of the 8th Intern. Symp. on Plasma Chem., 13
(1987).
25 A.Vardelle, et al, Proc. of the 7th Intern. Symp. on Plasma Chem., 898
(1985).
26 C.H.Chang and J.D.Ramshaw, Proc. of the 10th Intern. Symp. on Plasma
Chem., 4, No.1,1-6 (1991).
27 O.P.Solonenko and A.L.Sorokin, Proc. of the 10th Intern. Symp. on Plasma
Chem.,1, No.1, 1-2 1 (1991)
28 G.S. Pavlovic, et al, J. of High Temperature Chemical Processes, Suppl.,
1, No.3, 381 (1992).
29 P.Stefonovic, et al, Numerical analysis of heat transfer between nitrogen
plasma and injected Si particles in axisymmetric reactor (in print).
30 D.Milojevic, Proc of the 2nd Workshop on Two-Phase Flow Prediction,
Erlangen, Germany, 31 (1985).
31 C.T.Crowe, et al, Trans. of ASME, J. Fluids Engineering, 325 (1977).
32 M.Vardelle, Cycle Thesis. Univ. of Limoges, France (1980).
33 N.B.Vargaftik, Tables of the Thermophysical Properties of Liquids and
Gases, 2nd ed., Hemisphere, Washington D.C. (1975).
34 D.J.Carlson and R.F.Hoglund, AIAA Journal, 2, No.1 (1964).
35 I.V.Kalganova and V.S.Klubnikin, J. of High Temperature (in Russian),
14, 408 (1976).
36 A.A.Mihalchenko, et al, Proc. of the 8
th
National Thermal Spray Conf.,
Houston, USA (1995), p.741.
37 O.P.Solonenko, et al, Proc. of 3th Asia-Pacific Conf. on Plasma Sci-
ence and Technology, Tokyo, Japan (1996), p.241.
38 O.P.Solonenko, et al, Proc. of the 14
th
Int. Thermal Spray Conf., Kobe,
Japan (1995), p.359.
39 O.P.Solonenko and A.V.Smirnov, Proc. of 12
th
Int. Symp. on Plasma Chem-
istry, Minneapolis, USA (1995), p.874.
40 O.P.Solonenko, Proc. of the 8
th
National Thermal Spray Conf., Houston,
USA (1995), p.423.
41 O.P.Solonenko and A.V.Smirnov, Proc. of 3th Asia-Pacific Conf. on Plasma
Science and Technology, Tokyo, Japan (1996), p.247.
42 O.P.Solonenko, In: Werkstofftechnologie auf dem Weg in das Jahrhundert:
Vortrage und Veroffentlichungen des gleichnamigen 15. Dortmunder
Hochschul kol l oqui ums (1996), Dortmund/ Lehrstuhl fur
Werkstofftechnologie, Universitat Dortmund.-Aachen, Mainz (1996).
43 O.P.Solonenko, In: High-Temperature Dust-Laden Jets in Plasma Tech-
nology, O.P.Solonenko and A.I.Fedorchenko (eds), VSP Publishing House,
Utrecht, Netherlands (1990), p.59.
44 O.P.Solonenko, et al, Proc. of 6th National Thermal Spray Conf., Anaheim,
USA (1993).
45 Z.R.Ismagilov, et al., Proc. of 13th Intern. Symp. on Plasma Chemis-
try, Beijing, China (1997).
46 O.Yu.Pogyacheva, et al., In: Application of plasma spraying in prepa-
101
ration of metal supported catalysts, Catalysis Today (in print), (1998).
47 O.P.Solonenko, et al, Proc. of 5th Intern. Conf. Film and Coatings,
Saint-Peterburgh, Russia (1998).
102
Metallic and Ceramic Materials:
Present and Future
V.A. Neronov
Institute of Theoretical and Applied Mechanics, Siberian Branch, Russian Academy
of Sciences, 4/1 Institutskaya Str., Novosibirsk, 630090, Russia
The basis of scientific-technical progress is power engineering,
automatization, and materials. Traditional materials cannot satisfy
complex demands of high temperatures, velocities, loadings, and
aggressive media. It leads to elaboration of new high-temperature
materials on the basis of refractory compounds borides, carbides,
nitrides, silicides, and their compounds. On the basis of refractory
compounds, traditional materials are made considerably better; high-
effective materials of atomic power engineering, of electrotechnical
intention are manufactured, as well as instrumental, fire-proof,
space, structural, and other materials. One of the main directions of
present-day material studies is the elaboration of principles of
designing materials (including coatings) with a highly dispersed
structure. Plasma chemical synthesis of ultra-fine powders is of
primary importance in this matter.
Materials occupy a special place in this triad since the develop-
ment of power engineering and automatization depends on the level
of materials.
Todays development of materials follows seven research direc-
tions shown in Fig. 1. They are ceramic, composite, polymer, vit-
reous, and ultra-pure (10
7
10
10
% of impurities) materials, coat-
ings and new metallic materials being considered as individual ac-
tivities.
These seven research directions can be complemented by another
one: diagnostics and material quality control.
Traditional materials cannot satisfy complex demands of high
temperatures, velocities, loadings, and aggressive media. The capa-
bilities of traditional quasi-equilibrium processes have been ex-
hausted for further improvement of technology. Extreme values of
103
Metallic and Ceramic Materials: Present and Future
Fig. 1. Structure of material development.
pressures, velocities, temperatures, dimensions, and other param-
et ers achi eved i n pract i ce l ed t o qual i t at i vel y new, st rongl y
nonequilibrium processes. A number of principally relevant inven-
tions were made, such as plasma torches, superhigh pressure de-
vices, impulse devices, effective engines, etc. Such notions as the
ultra-fine state of the matter, superconductivity, superelasticity ap-
peared or received new meaning.
Strongly nonequilibrium processes in the field of substance and
material synthesis are attractive due to variety of nonequilibrium
states of the matter and possibilities of its transformation into this
or that state with required properties. These possibilities, however,
are not adequately used at the moment.
Extensive and continuously increasing production of metals is
one of the most remarkable features of the XXth century. Metals
acquire a more and more important role, despite the development
and utilization of a large amount of new constructive materials
nonmetallic ones.
During 20 years, from 1960 to 1980, the world production of
iron has increased by approximately 2.7 times, nickel 4 times, cop-
per 2.3 times, aluminum 4.7 times, zinc 2 times, and titanium 17
times [1].
Scientists predict that in the nearest future the metal output will
104
be increasing. Assuming that the mean yearly consumption of steel,
for example, approaches the level of developed countries (500 kg
per capita), then about 2.53.0 billion tons of steel should be pro-
duced on the Earth every year. Taking the metal output of 1960 for
unity (note that this value has been achieved at least during 45
millennia), the above idea can be illustrated by the following plot
(Fig. 2). We can see that in the accepted scale, the steel output
curve at the end of the second millennium goes vertically upwards,
like a rocket. The rate is impressive, is not it? Moreover, it is tre-
mendous! What does the production of these vast amounts of metal
mean?
Let us consider several examples. For yearly production of cast
iron in the Soviet Union in early 80s, the iron ore transportation
required a train that could encircle the Earth equator seven times
[1]. Three more circles, and this will be equal to the distance to the
Moon. Cosmic scales on the Earth!
In 1975, the ferrous metallurgy of the Soviet Union burnt the
amount of fuel equivalent to the biannual output of coal in Great
Britain, a traditionally coal-mining country. A tremendous consump-
tion of water and energy!
Nonferrous metallurgy requires even higher consumption per ton
of metal.
In 1986, the stationary sources, automobiles, agricultural and
military facilities, air, naval, and railway transport of the USSR
produced 350 kg of hazardous waste per capita, which means that
each Soviet citizen got a kilogram of waste every day.
In 1990, each Soviet citizen already got 450 kg of waste per
Fig. 2. Dynamics of normalized metal output.
105
year. This is an average, doctored figure. A particularly hard situ-
ation was observed at that time in 43 cities of Russia. The leader
in this list was Norilsk. Only the stationary sources of this city
produced yearly about 13 tons of waste per one Norilsk inhabitant
[2]. The situation with water basins is not less catastrophic.
The above examples are sufficient to repeat the famous saying
that the humankind rushes at full speed towards the edge of an
abyss. The definition of scientific-technical progress presented at
the beginning of this paper (power engineering, automatization, ma-
terials) is, therefore, incomplete and one-sided. Ecology must be the
primary reason to be taken into account.
A question arises here: what should we do next? Should we in-
crease the production or is there a rational alternative? The re-
searchers say there is one: improvement of the quality of materials
and the most rational use of properties imparted by nature. For
example, the theoretical strength of metals is higher than practically
achieved results by a factor of 10100.
Considerable progress has been made in this direction. Let us
come back to examples. In laboratory tests of copper wire 1.25
microns in diameter, a strength of 294 kgf/mm
2
was obtained in-
stead of 25 kgf/mm
2
for technical copper. The strength of a thin iron
wire 1.6 microns in diameter was 1310 kgf/mm
2
instead of 3040
kgf/mm
2
typical of commonly used steel [1]. Thus, the strength in-
creased 3040 times. In the ideal case, this means a 3040fold
economy of metal. Fibrous materials are used in practice but not yet
sufficiently.
There is another example of a material that was obtained in ex-
treme conditions but is frequently used now. We speak about the
research performed at t he Al l -Uni on Inst i t ut e of Li ght Al l oys
(AILA) under the guidance of Academician A.Belov [1].
The traditional method of obtaining the major part of metallic
semifinished items includes metal melting, melt crystallization into
an ingot, and pressure treatment for the ingot to acquire a neces-
sary shape. The weak link in this chain is crystallization.
The process of crystallization includes the changes in the chemi-
cal composition of the melt, redistribution of elements and admix-
tures segregation. In particularly adverse conditions of the forma-
tion of large ingots, the level of segregation can reach 300 % for
carbon, 600 % for phosphorus, 1000 % for sulphur [3, p.502]. This
means that different sorts of metal, up to complete spoilage, are ob-
tained in different parts of the ingot. Besides, the growth of grains
occurs during crystallization, which impairs the metal properties.
106

a
b
c
d
Fig. 3. b Structure of granules (a-c) and ingot (d) obtained from the same nickel
alloy with cooling rates of 10000 and 4C/s, respectively [1].
Obviously, the increasing of the crystallization rate is an effective
method for improvement of the metal structure and properties. The
usual rate of surface cooling of a solidifying ingot is about 1 de-
gree per second. Tremendous cooling rates have been achieved at
the AILA. about a million degrees per second. This superhigh cool-
ing rate could be achieved for small ingots that are granules 5
500 microns in diameter (they are commonly obtained by spraying
a metal melt by a jet of pure neutral gas or by liquid metal spray-
ing due to centrifugal forces).
Figure 3 shows the structure of an ingot cast by the traditional
107
method and the structure of granules obtained with a cooling rate
of 10000 C/s. A drastic change in the metal structure is seen. The
grain size decreases by a factor of 1000 and more, which is equiva-
lent to the formation of a billion of grains and more instead of one
grain. Besides, the material obtained is uniform.
The technology of preparing articles from granules (granular
metallurgy) includes the following basic stages: magnetic separation
and activation of granules, their separation into fractions, degassing
and capsule welding-up into a container, treatment at a temperature
of about 1200 C and pressure of about 2000 atm, thermal treat-
ment of articles, mechanical treatment, and quality control. The use
of the new technology increased the high-temperature strength of en-
gine disks made of nickel alloys by more than 1.5 times.
Granular metallurgy has great prospects. The treatment of cast
articles at high temperatures and pressures has a special signifi-
cance. For example, the treatment of cast turbine blades decreased
the spoilage percentage from 2530 % to 05 %, simultaneously in-
creasing the service life 22.5 times. Note that a complex-construc-
tion blade costs roughly as much as a car [1].
The development of metallic materials will make use of techni-
cal achievements of advanced technology. This means the creation
of strongly nonequilibrium structures. New metallic materials are
cool-resistant steels (elimination of adverse admixtures, doping),
radiation-resistant steels in atomic power engineering, steels with
high content of nitrogen (ten-fold supersaturation), alloys with
amorphous and microcrystalline structure (1.5-fold increase of ul-
timate strength), constructive materials made of ultrafine powders
(which also leads to better mechanical characteristics), high-tem-
perature steels, aluminum alloys with high strength and low density
(AlLi alloys), alloys with shape memory, etc.
Only 0. 0015 % of multi-species alloys containing more than
three components were studied in mid-80s, but the specialists pre-
dict that by 2000 the number of complex alloys will significantly
increase and amount to 250 000. Almost all metals of the periodic
system will be used in practice [4].
Great importance in the development of new metallic materials
will belong to oxygen-free refractory compounds: carbides, nitrides,
borides, silicides, etc., especially in the ultra-fine state the region
where plasma chemistry will obviously play a positive role.
Refractory oxygen-free compounds are widely used due to a
valuable complex of their properties: they have a high melting tem-
perature (up to 4000 C, tantalum and hafnium carbides) and a high
108
strength approaching that of the hardest material, diamond (cubic
boron nitride, etc.), they are extremely stable in metal melts (tita-
nium and zirconium diborides, boron, aluminum and silicon ni-
trides), they are not destroyed in the air when heated up to 1700 C
(molybdenum, tungsten, rhenium disilicides, silicon carbide) and
weakly evaporate in vacuum (niobium and hafnium carbides), they
have unique atomic physical properties (boron, uranium, thorium
carbides) and low electric conductivity (boron nitride), they are
high-temperature semiconductors (boron and silicon carbides). Re-
Objectives Material characteristics, examples of its usage
Refractory
materials
High resistance to melted materials, erosion resistance, heat resistance.
Vacuum metal spraying: metal coatings on ceramics, metals, glasses, plastics,
fabrics, and leather. Calorizing of glasses instead of silvering in mirror
industry, calorizing of black plate (instead of tin-plate) in food-canning
industry. Protective coatings for thermocouples for controlling the temperature
of steel, cast iron, and other alloys. Materials: TiB
2
, TiC, TiN, BN, AlN, SiC,
MoSi
2
, etc.
Electrotec
hnical
materials
High-temperature furnace heaters up to 1650 C (SiC, MoSi
2
). Heaters for
burners with neutral media up to 2800 C (NbC). Single crystal growing,
thermal treatment of specific alloys.
Materials
with
special
electrophy
sical
properties
Good thermoelectric properties, high heat resistance.
Thermoelectrodes of high-temperature thermocouples. Thermocouples with
MoSi
2
/WSi
2
thermoelectrodes in oxidizing media (high stability, sensitivity of
10 V/deg, measurements of temperature of melted glass, oxidizing gas media
in chamber and reverberatory furnaces). Thermocouples with C/ZrB
2
, C/TiC
thermoelectrodes in neutral and carbonizing media (high stability, sensitivity of
65 V/deg, linear dependence in the range of 6002500 C). Thermoelectrodes
of high-temperature thermogenerators, cathodes of powerful electronic devices
for melting of refractory highly pure metals and alloys, growing and
purification of single crystals, etc. Electrodes of MHD-generators, reusable
electrodes for resistance welding of carbon steel, stainless steel, etc.
Materials: Si
3
N
4
, MoSi
2
, SiC, complex carbides Zr
2
WC, TaWC, TaMoC,
NbWC, NbMoC, etc.
Constructi
ve
materials
Hard alloys for pumping aggressive media and pulps, compressing high-
temperature gases. Articles of silicon carbide: nozzles and vanes in sprayers for
superphosphate drying. Hydrocyclone bodies, jet mill lining, etc.
Instrumen
t materials
High hardness, strength, wear and chemical resistance. Abrasive and blade
instruments on the basis of tungsten carbide (hardness of up to 92 HRA,
elasticity modulus of up to 70000 kgf/mm
2
, compressive strength of 600
kgf/mm
2
, impact strength of up to 0.6 kgm/cm
2
) and tungsten-free hard alloys
on the basis of titanium carbide, complex titanium-niobium carbide, titanium
carbonitride, TiB
2
, ZrB
2
, TiC-TiB
2
, TiN-AlN. Carbide and nitride coatings
increase the resistance of cutting and hard-alloy instruments by 25 times.

The use of refractory compounds [5,6]
109
fractory compounds are used in metallurgy, space technology, and
other important areas [5].
The materials based on refractory compounds and their applica-
tion are briefly presented in the table bellow.
Special attention should be paid to a cutting superhard material
on the basis of boron nitride hexanite-R characterized by a high
cutting capability, high resistance to strong shock loads. This amaz-
ing result cannot be currently explained by any existing physical
theory. Propulsion! Ceramic armor! Fabulous prospects of ceramics!
Indeed, the further the experiment is from the theory, the closer it
is to the Nobel Prize! Ceramics is the thing that the humankind
started to master. Modern ceramics is the ceramics on a new heli-
cal turn. In fact, a kind of ceramic material study has appeared.
The main problem (it refers both to metallic and ceramic mate-
rials) is the obtaining of materials in an ultra-fine state in order to
combine strength and plasticity. The most important scientific direc-
Fig. 4. Scheme of research activity. Bold line our developments, thin lines - joint
research with other organisations, double lines - research activity without our participation
110
tion in the field of material study is, therefore, the study of com-
plex equilibrium and nonequilibrium states, and construction of dia-
grams of state.
The strategy of material development is to find and preserve the
optimum ratio between the fraction of traditional materials, the
fraction of materials of the nearest future, and the fraction of prom-
ising materials. the fraction of the latter should be at least 10-
15 %, as evaluated by specialists.
The studies that we conducted in the field of borides formed
theoretical and practical grounds for subsequent works in scientific
centers of the former Soviet Union and Europe. This involved the
training of qualified staff and creation of new materials with a valu-
abl e compl ex of propert i es, mai nl y on t he basi s of al umi num
dodecaboride (Fig. 4):
monocrystals;
polycrystals;
superhard materials for instruments, with high resistance and
three-fold cutting capability as compared with alloys on the basis
of tungsten carbide (in treatment of hard graphites);
hard materials with characteristics at the world level;
coatings of special purpose with an economic effect of 5060
thousand US dollars for one article, wear-resistant coatings;
thermoelectric materials with record-breaking characteristics at
1750 K that have not been surpassed yet (the priority of these re-
sults has been noted abroad).
Thus, the performed work laid the basis for a promising scien-
tific-technical direction in the field of borides.
We have analyzed the state-of-the-art of boride research as a
whole and described it in a number of monographs (e.g., [7, 8]).
Our investigations in the field of physics, chemistry, and technol-
ogy of refractory compounds favored the development of a princi-
pally new scientific direction of advanced material study Synthe-
sis, investigation and utilization of fine-grained systems in the study
of composite materials at the Siberian State Mining-Metallurgical
Academy (Novokuznetsk, Prof. G.V.Galevsky). Ultrafine borides
and other compounds were used for the first time in domestic prac-
tice.
A titanic contribution to material study, including refractory
compounds, was made by our famous scientist Grigory Samsonov,
a corresponding member of the Ukranian Academy of Sciences. He
brought up a lot of scientists 200 candidates of sciences, 15 doc-
tors of sciences, developed together with his followers 500 new sub-
111
stances, put into practice 150 technologies, wrote 40 monographs
and 2000 articles. He did it during a short time. As it always hap-
pens, envious people stood in his way. He died when he was 57, in
the prime of his life. He was in combat at the WW II. He founded
a scientific school recognized in his lifetime by the entire world. He
lives with us and will live for a long time because a person is alive
as long as somebody remembers about him. These are his words,
but he did not speak about himself, he spoke about those who do
their best to serve the Motherland and the Humankind in a spirit of
high responsibility. He was such a person. We should know and
remember it. Another such person we must say this because we
celebrate his jubilee is Mikhail Zhukov who, keeping up with the
time and ahead of it, made a great contribution to organization,
science and practice in the field of plasma dynamics and related
areas.
References
1. A.Belov, Metal: improvement of quality the way to economy, Nauka
i Zhizn (Science and Life), 1982, No. 2, P.2-9.
2. Argumenty i Fakty (Arguments and Facts), 1990, No. 34, P.7.
3. E.V.Abrosimov, I.I.Ansheles, V.A.Kudrin, Yu.V.Kryakovsky, V.N.Orlov,
Steel Metallurgy, Metallurgiya, Moscow, 1961, 680 p.
4. H.Waschul l , Praparati ve Metal l ographi e, VEB Duetscher Verl ag fur
Grundstoffindustrie, Leipzig, 1984.
5. P.S.Kisly, Obtaining articles made of refractory compounds by means
of powder metallurgy, Zhurnal Vsesoyuznogo Khimicheskogo Obshchestva
im. D.I.Mendeleeva (Journal of All-Union Mendeleevs Chemical So-
ciety), 1979, Vol. 24, No. 3, P.270-275.
6. Properties, synthesis and application of refractory compounds, Ref-
erence Book, T.Ya.Kosolapova (ed.), Metallurgiya, Moscow, 1986, 928
p.
7. G.V.Samsonov, T.I.Serebryakova, V.A.Neronov, Borides, Atomizdat, Moscow,
1975, 376 p.
8. T.I.Serebryakova, V.A.Neronov, P.D.Peshev, High-Temperature Borides,
Metallurgy, Chelyabinsk, 1991, 368 p.
112
Volokitin G.G., Borzykh V.E. and Skripnikova N.K.
Tomsk State Architectural-Building University, Tomsk, Russia
Extreme Technologies in Building
Material Production
Recently, there has been a marked tendency to use non-traditional
technologies for the material production and waste utilization of
different origin. It is very important because the possibility of
quasidisbalanced processes for the further improvement of existing
technologies appeared to be exhausted. The value of disbalanced
processes in the field of synthesis and obtaining new materials as well
as utilization of vital activity waste of persons is concluded in varied
disbalanced conditions of material and possibilities of it leading to one
or another condition with technology- useful characteristics. The local
concentration of energy is necessary for the realization of disbalanced
processes. The latest knowledge on disbalanced processes has allowed
the application of the most up-to-date technologies in the field of
material study and ecology, such as plasma, pulsed and high pressure.
Using the energy of plasma flows in industry has allowed us to
modify old and develop new technologies. In particular, this concerns
machine building and chemical industry. Building industry and ecology
has been on the side from these tendencies, however here shifts are
marked, which are the most significant in the production of building
materials and utilization of different type of waste.
Operated and controlled plasma heating allows us to obtain an ef-
ficient mode of the technological process, ensuring the maximum use-
ful output under minimum specific expenses of material and energy.
Besides, high-temperature heating plasma by flows creates conditions
of strong disbalance, when high-energy particles participate under the
moderate middle-mass temperature in the working process. New
knowledge of these processes has allowed the group of authors to
create a number of new technologies in the field of building industry.
These technologies have gained a new base in applied sciences, such
113
Extreme Technologies in Building Material Production
as plasma and chemistry, building materials.
The main essential contribution of the authors to the development
of building materials processes and plasma chemistry is:
theoretical genelization of research results in the field of plasma
chemistry and building material processes with the aim of processing
building products with the help of low-temperature plasma [13]
development of temperature field calculation strategies on the
depth of processing the product for the two-phase case in heating its
surface by plasma [4];
creation of generalized object models and a process of automated
plasming of building materials on the basis of different mineral astrin-
gent [5];
study of hydrodynamics and heat-exchange in the moving film of
a mineral substance melt, under the work of mass power and in con-
ditions of strong effects with reference to electro- and plasma tech-
nologies of obtaining mineral filaments [6].
PLASMA PROCESSING OF BUILDING MATERIALS SURFACE
The development of plasma technology has allowed the gas and
plasma processing of the building materials surface to be replaced by
plasma. The results of initial research into obtaining coatings on
building products by means of plasma devices by melting their surface
have shown that in comparison with gas and plasma the processing
speed of heavy concrete and brick has increased 2.53.0 times, with
the heating of basic material mass reduced greatly. It is estimated
that at a processing speed of 70 mm/sec with the average value of
coating voltage the take-off is 0.9
.
10
5
N/m
2
and the covering thickness
is 0.4 mm [7].
For the development and increasing efficiency it is useful to use
Fig. 1. Dependence of temperature on processing time: 1 on surfaces; 2 at the
depth 0.005; 3 at the depth 0.01; 4 at the depth 0.05.
T, C
t, s
114
thermal processing principal criteria of the optimal regime: technologi-
cal compatibility of plasma processing with the process of building
products; the final product reliability and processing productivity, not
reducing main processing, which in turn are connected with heating
and physical and chemical processes, running under the influence of
plasma on building materials at the moment of creation of glass type
coverings on their surface.
In spite of the fact that a number of works [8,9] have been de-
voted to the interaction of plasma with hard substances, there is no
information on the determination of temperature fields in the section
of building products processed by low-temperature plasma under dif-
ferent power plasma device data that is extremely important for study-
ing physical and chemical conversions which take place in the prod-
uct under the heat influence, which are responsible for adhesive tough-
ness, chemical stability, water absorbtion, and frost resistance.
Earlier in work [10] a simplified strategy of temperature field cal-
culations while heating the surface by a plasma flow was proposed;
however the processes occuring on the product surface, leading to the
new phase (evaporation, formation, melting) appearance under which
heat and physical factors , c, a and others are changed, were not
taken into consideration. To achieve reliable results of sharing tem-
perature fields, the univariate Stephan task for two-phase case was
solved. Numerical realization of the model under real significant ini-
tial data was carried out by the difference method on the tacit
scheme.
The results of calculations have been confirmed by experiments
(Fig. 1) and were then used in physical and chemical conversions oc-
curring in silicate products under the action of true temperatures.
Silicate product durability and, accordingly, under plasma process-
ing adhesive durability of covering, are determinated by the pres-
ence of fundamental calcium hydrosilicates such as tobermorite and
csonotolite, which firmly connect particle grains. However, according
to [12], in the temperature range 9731143 K the above mentioned
hydrosilicates change to -calcium silicate (wollastonite), changing un-
der 1398 K into -modification (pseudowollastonite) with lower dura-
bility. The analysis of the results shows that adhesive durability de-
pends not only on the depth of the destructive layer but also on other
factors, one of which is the granulometric filling.
Having taking as a basis the idea of anchor fastening of glass
covering with the product base, the authors of this article have pro-
posed a generalised process model of the shaping of glass coatings on
building products, based on the principle of ensuring reliable connec-
115
tion between the coating and the base and taking into account (on a
large scale) all physical and chemical processes taking place in the
formation of reliable glass type coverings for service
GENERALISED MODEL OF THE PROCESS OF FORMATION OF
GLASS-TYPE COVERING ON BUILDING PRODUCTS IN
PLASMA PROCESSING
The effect of plasma flows results in the melting of processing
material components with further forming of glass-type coverings on
the product surface. Physical and chemical processes, occuring in the
material under plasma processing and responsible for the durability of
the bond of the covering with the substrate, are very complex and
have been studied insufficiently. In this connection, this work is
devoted to the creation of a generalized model of glass-type covering
forming processed on building products based on principles of ensuring
a reliable bond between the covering and the base.
Figure 2 shows the generalized product model indicating the collec-
tion of different objects of building materials. In the base model of the
shaping of the covering there is the frame 2, safely connecting the
new formation 1, formed under the heat influence of plasma on the
materials of the base 3. During thermal decoration of building prod-
ucts the content of the frame cells is the frozen melt, and the frame
itself is created by the structure of the material not affected by the
effect of plasma heat. Naturally, this condition can be ensured under
the corresponding granulometric composition of the filling and the
physical and chemical properties of the processed material.
The mathematical frame description in the chosen generalized
model of the thermoprocessed product satisfies the Dirichlet condition
[13], namely: function h(t) is limited and has a number of extreme
values in the period T (here h(t) is the height of the considered spot
Fig. 2 Generalized product model, subject to plasma processing: 1 new formation;
2 frame; 3 basis.
116
of the frame at the moment of time t at the scan). Then h(t), as any
periodic function satisfying the Dirichlet condition can be presented in
addition to the Fourier condition.
It is necessary to note that the values of amplitudes, harmonic fre-
quencies and their initial phases with reference to investigated objects
cannot be determined correctly. In this case, a probable approach on
the basis of casual cases modeling of plasma and substance interac-
tion of non-homogeneous structures is possible. However, in the pres-
ence of a number of true factors, such as the condition of melt pen-
etration in the body of the product, the formation of connected lay-
ers, weakening of bonds of the covering with the product, we have
adopted as the most reliable method of real objects analysis, subjected
to thermal processing, a set of necessary statistics and generalization
of the results in the case of formation of strong bonds of the cover-
ing with the product. Within the framework of the generalized model
it is reasonable to consider as a special case the new formations proc-
ess and the frame shaping responsible for a strong bond between the
covering and the product base using an example of silicate brick
plasma processing.
From analysis [10], it is possible to note that after the thermal in-
fluence on the product material its structure contains the following
areas (Fig.3): melting 1, connecting layer 3, base 4; here melting 1,
interacting with the grain filling (sand) 2, serving as frame, accord-
ing to the accepted generalized model, melts with it, forming a glass-
type framework on cooling the basis of which is the product 4, which
is not subjected to thermal destruction and has saved its strong bond-
ing. This explanation fulfils the anchor function of the framework re-
Fig. 3 Scheme of interaction melting with the base of the silicate product: 1 melting;
2 grain filling; 3 bonded layer; 4 base of the product.
117
sponsible for bonding of the covering and the base. Simultaneously,
melting, interacting with the astringent and also in its micro- and
macrobody, casued due to heat exchange causes destruction of cal-
cium hydrosilicates, forming herewith the connected layer 3 reduced
bonding. The thickness of the connected layer depends directly on the
amount of the melt on the product surface and its capacity. To ensur-
ing strong bonding of the covering with the base, the following con-
dition must be met:
d
g c con
> + (1)
where d
g
is the size of the grain filling, mm;
c
is the thickness of the
covering, mm;
con
is the thickness of the connecting layer, mm. The
thickness of the covering
in relation to the power of the plasma

device and the processing rate is in Fig. 4. The thickness of the
connected layer depends on the depth of penetration of temperatures
at which material destruction occurs. When the total thickness of the
melt and the destructed connected layer is greater than the thickness
of the grain filling, the bonding strenth with the base will be low.
The research work carried out on industrial waste conversion
(ashes HES, slags) into mineral fillings has resulted in the construc-
tion of a plasma and chemical reactor [14] leading to a number of
fundamental problems on studying hydrodynamics and heat exchange
in the moving film of the melting mineral material substance under the
action of mass power. The proposed method is illustrated in Fig. 5.
The realization of the proposal requires the solution of the follow-
Fig. 4 Dependency of covering thickness on the plasma processing mode: values N,
kW: 38 (1), 36 (2), 24 (3), 56 (4).
v, m/s
118
T
ex
Fig. 5 Schema of heat exchange in the moving melting film on the wall of the RPCR:
angular velocity of reactor rotation; 1 fluid film; 2 skull; 3 steel wall, q
n
-
total density of heat flow; q
w
density of the heat flow in the skull layer; V
max
-
maximum axial velocity of the liquid in the film; T
p
softening temperature; T
n
and
T
ex
temperature on the internal and external surfaces of the reactor, respectively; -
thickness of the moving melting film; L
1
and L
2
the thickness of the steel wall and
thickness of the skull layer; r - radial coordinate, r
0
radial coordinate of phases of
the liquidsolid section.
ing problems: creating suitable conditions for efficient supply of heat
energy from plasma streams into the dispersed material, conditions for
the shaping of a film of the melting mineral substance in the
semiclosed volume and disintegration of the film in the filling in the
field of centrifugal power. The physical process model can be pre-
sented in the following type (Fig. 5): when the rotating homogenising
device receives the powder charge, being the pseudo-liquid, the charge
has the shape of a strongly stretched truncated paraboloid, on the sur-
face of which a skull layer forsm at the initial moment under insuf-
ficient heat energy. Under the action of centrifugal power melted
glassed masses moves on the surface of the skull layer as fine films.
It is assumed that at a high velocity of reactor rotation ( = 50...100
1/sec) the melting film is small in comparison with the radius; glass
formation took place at the bottom of the reactor; the skull layer due
119
to its low heat conduction coefficient plays the role of heat isolation
in passive heat protection of the walls of the reactor; on the border
between the fluid and the hard phase the temperature of the melting
raw materials is the softening temperature; the heat put into the sys-
tem partly goes to melting of the mixture of raw materials and is car-
ried away together with the moving melting film, but is partly trans-
ported through the skull layer in the airstrip of extraction to ventila-
tion; heat losses in bands of glass are not taken into account; the heat
balance system forms; the current of film melting is formed and lami-
nar; the axial velocity changes along the height of the reactor; tan-
gential stress at the liquidgas border is extremely little; all heat
physical characteristics of glass mass and charge are constant, except
the viscosity factor, which is a function of temperature; the
thermocapillary effect is not taken into account.
The mathematical problem is based on the given physical models
and is a system of equations of pulses conservation, energy and mass
conservation with the corresponding boundary conditions. The change
of pressure defines the movement of the film. It is assumed to pro-
portional to the normal pressure created by the centrifugal power in
the elementary volume of the film. The source system is defined by
considering the assumption on the absence of a correlation between
the axial velocity and the vertical coordinate and the initial middle
mass temperature and expressing the axial velocity through the den-
sity of irrigation. As a result, we get a common differential equation
of the third order for the determination of the melting temperature as
functions of the radius
d
d
k
k
q q e
w
3
3
0
2 1
( )( )
= =
C
T
T T
r r
p
p
1
2
0
( ),
q
C q
T

p
=

1
2
q
C q
T
w
w
p
=

1
2 (2)
where is the dimensionless temperature. For closed systems, we
shall consider heat transfer in three layers to the system (Fig. 5):
q
T T
L L
w
p c
=

+

( )
1
1
2
2
1
T T
q L
p
w
=
1
1
T T q
L L
p w
= +
1
1
2
2

(3)
120
The condition for the determination of the thickness of the layer is
a correlation for the heat balance of the system two-layer wallmov-
ing film for the higher part of the cylinder where there are conditions
of melting for following fibre creation. The differential equation (2)
was determined numerically. To find the beginning of stabilization of
the current area of the fluid film, the dependence of the Nusselt
number on the longitudal coordinate was examined. As a result, the
initial heat area is evaluated by the value x/Pe= 5.5, where Pe is the
number of heat intensity. The middle mass temperature change is de-
fined by the Nusselt number. Considering the dependence Nu
p
(k/k
0
)
on the automodel mode, for the stabilizing current area after approxi-
mations of numerical relationships we shall have:
Nu
q
q
p

4 07
0 643
,
,
k
k
A
r
w
0 0
0 6
0 0011
0 6
=

, exp
,
Pr
,
;
Y
q
q
q
q
0 98
016 0 8
.
. .
;

r
A
q
q
w

0
0 48 0 2
2
0 2
0 324 =

. Pr
Re
Re
. .
.
;
Pr
( )
; Pr
( )
; = = =

T
a
T
a
A
C
q r
w
p
1
0
;
Y
Nu Nu
p
=
4 4
.
(4)
Here A is the dimensionless parameter characterizing the heat
mode of the viscous flow of the film; is the experimental factor
taking into account consumption, dynamic and geometric features; k
is the dimensionless friction factor; k
0
is the friction factor for the
parabolic part of the axial velocity; Re- is the Reynolds number.
The correlations obtained for the surface in the four-parameter
space can be used for determining the mean time of melting in the
reactor at a specific temperature and allow the selection of the nec-
essary parameters of the device for ensuring complete boiling of the
121
glassmass; in essence, it is the starting point for experimental revision
of the working parameters of a real installation.
Experiments were carried out to check the results of calculations
using the mathematical model of film current melting on the walls of
the revolving reactor. The experiments were conducted with melting
of quartz, diopside, basalt and ash. Thickness measurements of the
melting film were carried out by the method of frozen layer melting
(freeze process). The analysis of the layer structure showed (traces
of spottinesses, directed vertically along the reactor wall) a specific
flow direction that confirms the existence of the effect of the tangen-
tial stress when defining the boundary conditions. The calculated mean
mass temperature was compared to the melt temperature, evaluated
using a pyrometer. The measured melt temperature and film thick-
nesses with the error not more than 20 % are co-ordinated by the
assumption that points to the aplicability of dependences obtained with
the limit of the specified inaccuracy for the calculation of technologi-
cal parameters of the device.
The results were realizedin the experimental stand Plasma-rotor for
the determinations of the main technological parameters of stable
functioning of all nodes of the installation, automatic systems and
readings of the measuring instruments. The optimum technological pa-
rameters of work of the RPCR with a productivity of 35 kg/hour of
the filling USING ash GRES-2 as the initial material (Tomsk). With
the adding of limestone fromthe Kamensk deposit, the parameters were
as follows: power of the plasma device 70 kW; the mass consuption
of raw materials 13 g/sec; angular velocity of rotation of the RPCR
400 1/sec; the radius of the cylindrical container 50 mm, height 200
mm; the factor of heating intensity under external heat transfer 800
W/m
2
K.
The filling obtained in plasma technologies has high heat stability,
large length of the filament and high tensile strength, so it can find its
use not only in the building industry but also when making objects of
for special applications.
Therefore, using plasma generators in the mineral filament produc-
tion allows the use of raw materials with a high content of high-tem-
perature oxides with a significant reduction of harmful emission into
the atmosphere and a reduction of the consumption of scarce sources
of energy (coke, natural gas, etc.); technology is simplified by short-
ening some sections of the traditional technological chain and opens
a real possibility of the automation of mineral filament production.
The results show that the use of high-concentrated sources of en-
ergy (low-temperature plasma) ensures the realization of technologi-
122
cal processes in extreme conditions, basing on high-speed (fraction of
a second) physical-chemical reactions, instant changes of the charac-
teristics of processing material and ensuring a one-stage technologi-
cal process.
In this connection, it is possible to say that the decision on the
problem of development and making of new materials with specific
characteristics, based on technological processes under extreme con-
ditions, is highly promising and ensures the maximum useful product
with minimum power expenditure.
References
1 G.G. Volokitin and B.P. Romanov, Study of phase transformations of
calcium hydrosilicates under plasma processing. Use of low temperature
plasma in technologies of inorganic materials and powder metallurgy,
Report All-Union. conf., Riga, Zinatne (1985), pp.128-135.
2 G.G. Volokitin, et al, In: Physical and chemical glass-type covering studies,
when processing the surface of silicale material by low-temperature
plasma. Use of low-temperature plasma in technologies of inorganic
materials and powder metallurgy. Report All-Union. conf., Riga, Zinatne
(1985), pp.136-143.
3 G.G. Volokitin, et al, Thermodynamic process studies, running on surfaces
of silicate bricks when processing it by low-temperature plasma, Tomsk
State University of Architecture and Building, Tomsk (1986).
4. G.G. Volokitin, et al, In: Calculation of heat fields in the area of an silicate
brick in processing by low-temperature plasma. Generators of low-tem-
perature plasma: Thes. Report. XI All-Union conf. generators of low-
temperature plasma, Novosibirsk (1989), pp.324-325.
5 G.G. Volokitin, Generalised process model of forming glass-type covering
on building products under plasma processing, Izv. Sib. Otd. AN SSSR,
Tekh. Nauk, (1990), p.99-102.
6 V.E. Borzych, In: Works of III Minsk International Forum on Heat Mass
Transfer, Minsk (1996.), ANK
7 M.E. Ermolaev, et al, Building materials, No.7, 21 (1976).
8 P.V. Pashtskiy and T.V. Kusina, Fiz. Khim. Obrab. Mater., No.3, 37-39
(1987).
9 A.B. Demidovich, et al, Eng. Phys. Journal, No.3, 461-466 (1984).
10 I.O. Bugs, et al, Influence of heating on structured conversions in silicat
products, Izv. AN SSSR, (1983), p.10.
11 R.C. Guger and B.V. Ovchinskiy, The Elements numerical analysis and
mathemati cal processi ng the resul ts experi ment.- 2 i ssue, Sci ence
Publishing (1970).
12 Ch.F. Taylor, Chemistry of Cements, Moscow, Building Publish (1969),
13 I.V. Kuzmin and V.A. Kedruev, Basis of theory information and coding,
Kiev High school (1977).
14 G.G. Volokitin, et al, Patent 2000977 Russia. Method of producing mineral
filament by plasma heating (Russia).- 93057718/33,
123
Achievements of Plasma Tomography
V.V. Pickalov
Recent years are characterized by the fast development of computational
physics, because the progress in modern computers has made it possible
to investigate more complicated physical phenomena in numerical
simulations. This is also valid for research into gas- and fluid dynamics,
in plasma physics. It is possible now to get three-dimensional space
and time characteristics of gas and plasma flows, with dynamic volume
vizualization of the numerical data.
To verify some theoretical models of physical phenomena, one needs
to compare theoretical calculations with experimental data. At this point,
a researcher has often difficulties, because he has to compare the multi-
dimensional modelling computations with the experimental diagnostic
data, which are only available at a set of points or lines, and very often
they are averaged in time.
Computerized tomography is a link between the experiment and the
theory, it gives a tool to the experimentalist and the theoretician for
an adequate comparison of the theoretical calculations with the experimental
measurements [47,49]. Tomography methods allow one to treat such
measurements with proper space and time resolution, and then to get
experimental data in such a form, which makes the comparison of the
theory and the experiment to be possible.
This paper gives a review of modern optical tomography techniques,
and corresponding two- and three-dimensional algorithms are also discussed.
Special emphasis is laid on the synthesis of tomography approaches
with the new image processing technique.
Some examples of the new methods of tomography for incomplete
data, local and pseudo-local tomography of high gradient structures are
discussed.
Institute of Theoretical and Applied Mechanics, Novosibirsk 630090, Russia
1. INTRODUCTION
124
2. TOMOGRAPHICAL PLASMA DIAGNOSTICS
Up to now tremendous effort have been applied to resolve spatial features
of low temperature plasma with the help of plasma tomography methods.
Well developed optical techniques for side observations of plasma
have been used in plasma tomography: emission [3,4,33,34,48,49], inter-
ferometry [7,29,30,55,61], schlieren technique [14], etc.
Up to now different experimental tomography systems have been applied
for diagnostics of different kinds of plasmas: thermal [32-38,49] [46,
50,53,57], low pressure [52,51], high frequency discharge [15], high
temperature plasmas [8,9,16,20,39], [40-45,54], and others.
It is known that in physical applications of tomography the following
problems complicated the diagnostics procedure:
- small number of angles of view K;
- small number of detectors for each angle N
p
;
- limited angle problem;
- limited region of impact parameters p;
- random noise in data;
- non-parallel arrangement of line integration system;
- finite width of the detectors;
- apparatus distortions of the signal measuring system;
Such problems sometimes make the interpretation and evaluation stages
of tomogram treatment very difficult.
Most of these problems can be illustrated by the Radon transform.
Indeed, in mathematics the Radon transform for two-dimensional functions
g(x,y) means the integrals over it along straight lines [11,17,22,44,49]:
f p g x y dl
p
( , ) ( , )
,
x
x
=
-
z
(1)
For simplicity, we will treat g(x,y) as a distribution of local emis-
sivity coefficients, and f(p,) as an integral emission of two-dimensional
plasma cross-section along the line with impact parameter p and an-
gle to the axis X (see Fig.1, [49]).
The Radon transform has an analytical inversion, namely (2D case):
g x y d
f p dp
p p
( , )
( , )
( )
, = -
-
-
z z
1
2
2
0
2
0
p
x
x
p
(2)
Thus, for reconstruction of the plasma image we need experimental
measurements of emission projections f (p,) for the angles inside the
region [0, ] and impact parameters p[R
0
, R
0
], where R
0
is a radius
125
of the circle which includes the whole plasma object inside. Though
usually Eq. (2) is not used for practical tomography, it shows all troubles
with our inverse problem. For it we need parallel geometry of data
acquisition, e.g. a set of parallel lines for detecting system. Then the
noise in data could be significantly enlarged after reconstruction, because
of singularity of the internal integral (which is called the convolution
or filtration integral).
Many of the problems mentioned above have some known solutions,
e.g. regularization procedures against noise [56], but others have to be
solved yet.
In particular, when information about plasma is space irregular it
is possible to extract more information for some regions than for others.
From the published literature ([11,44,49]) we know some attempts to
use Region of Interest (ROI) tomography approaches for the so-called
exterior Radon problem, when the data are measured for all angles x,
but only for those p which are outside some circle
0
: | p | >
0
. It
was found that the solution of this problem is unique, but very unstable.
In this case the measurements give information only about external regions
outside of the circle of radius
0
, and internal regions have no influence
on the resulting integrals.
On the opposite, the internal Radon problem means the reconstruction
of image inside a circle with radius
0
, when the limitations for measured
data are in the form | p | <
0
.

In this case there are much more un-
certainties in data, because external regions have input to any line which
comes through the internal region.
Possible approaches to solve the internal Radon problem are:
- to use the maximum of a priori information and to make some
Fig.1 The scheme of plasma emission measurements.
126
kind of extrapolation procedures to get projections of plasma emissions
in those parts, where such projections were not measured; the first steps
in this direction were made in our papers [19,20];
- to try to restore not exact values of local emissions, but to get
a so-called local tomogram which keeps main geometrical features of
the exact tomogram, and for such restoration those methods need in-
tegral measurements only for lines that come through a given point (x
0
,y
0
).
Thus, as the first step in the solution of low-informative tomography
problem one could suggest to use local tomograms, and first of all Vainberg
image [58] and summation image [2,49].
The following algorithms that can be used for the aims of plasma
flow visualization are based on wavelet decompositions [60] as a tool
for another localization type of tomograms, or pseudolocal tomograms
[23,24].
Local tomograms give us an approximate image of the plasma cross-
section, which shows the main geometrical features of the phenomena
under investigation. It gives us information about positioning of high
gradients in plasma, and shows the main flows in it.
But even getting such local tomograms involves some more diffi-
culties,
namely:
- possible low space resolution of detecting system, when the width
of detector is larger than the size of plasma structures of interest;
- limited angle covering, when detectors look into plasma only in
the region | | <
0
< .
The next section illustrates some items discussed above.
3 LOCAL TOMOGRAPHY METHODS
Usual tomography is global in the sense that Radon inversion at the
point (x
0
, y
0
) depends upon the all integrals f(p, ) of the object g(x,y)
(see Eq.(2). Such nonlocality forces the usage of all data in the standard
algorithm even if we are interested only in a small region of the object.
And local methods could reduce sometimes the data acquisition time
and the volume of experimental measurements.
In the papers of Faridani [12,13] a L-operator was introduced, which
gives the edge detection algorithm and yields a local tomogram. Much
earlier this operator was involved in practical tomographical devices
by E.Vainberg [58], and we call such tomogram a Vainberg image. Faridani
with co-workers really have used a combination of operator and its
inverse
1
(which is also local and occurs to be a summation image):
g g g
F
= +

1
.
(3)
127
Convolution in Eq.(2) is the source of nonlocality, and Vainberg has
changed this filtration with generalized function 1/p
2
just to the sec-
ond derivative of f:
g x y
d f p
dp
d
g g
V
p p
V
( , )
( )
,
,
( )
= -
=
z
=
1
2
2
2
0
2
2
0
p
x
l
p
x
x
(4)
where p
0
= x sin + y cos .
It differs from the real reconstruction first of all by such property
that the convolution in Radon inversion (2) is equivalent to the mul-
tiplication of in the Fourier domain by the function |n|, whereas the
second derivative gives us there ||
2
as a multiplier for
~
( ) f n
(or | | for
~
g (||)).
We introduce here such a definition of the Fourier transform:
F g v g v g x
R
xv
n
( )( )
~
( ) ( ) . =
z
-
x e d
i 2p
(5)
Then for given scalar , this defines the Riesz potential operator
[5,44]:
E( )( )
~
( ). I g g
a a
n n n =
-
(6)
Fig.2 Action of backprojection operator to produce the summation image.
128
For n-dimensional case, if < n, then the general inverse Radon transform
is [44]:
g I R I R g
n
=
- + -
1
2
1 a a
x
#
,
d i (7)
where R
g is the n-dimensional Radon transform, and R

#
is the back
projection operator.
For Laplacian operator , we have
I
=
2
2
1
4
,
and
Dg I g I R I R g
n
=- = -
- - - + -
4 2
2 2 2 2 1
p p
a a
x
#
.
e j
(8)
The operator I
=
1
1
2
is called the Lambda operator (it is also
known as Calderons operator, see [12]):
Lg I g I R I R g
n
= =
- - + - 1 1
1
2
a a
x
#
,
e j (9)
and taking = 1, we have
Lg R I R g
n
=
-
1
2
#
x
d i (10)
and its inverse
L
- -
=
1 2
1
2
g R I R g
n #
. x
d i (11)
For our case of two-dimensional tomography n = 2, and equations (10)
and (11) yield:
Lg R I R g =
-
1
2
2 #
,
x
e j (12)
=
1
1
2
g R R g
#
.
(13)
129
Now, we have the relations between -tomography, summation image
[59]:
g x y f p d
S
( , ) ( ( )) , =
z
1
2
0
0
p
x x
x
p
(14)
and the Vainberg image (4):
V g g
g
V
= = 2 , (15)
g g
S
=
1
2
1
.
(16)
Finally, for Faridani image g
F
we have:
g g
F V S
= +
1
2
2 .
(17)
It should be emphasized here that an exact reconstruction of g(x,y)
from only line integrals which pass through the region of interest is
impossible in the general case [44], so instead we put the problem of
approximate estimation of this function or of some special features, such
as locations of discontinuities, general geometrical characteristics, etc.
3.1. Local tomography in 3D
The 3D generalization of the Vainberg method is presented in the following
way
g P L f g
V
n
( ) ( ) ,
#
x =
2
b g
(18)
where L
2
is a 2D Laplace operator and P
#
is the operator of 3D
backprojection [44].
To get the Vainberg image one has to process each projection by
the Laplace operator and then to perform 3D backprojection. Using the
connection between the Laplace and Fourier operators one can rewrite
(18) as
g F P v F f
v
n
( ) | | ( ) .
#
x g =
-
1
4
3
1 2
2
p
d i (19)
130
where F
3
1
is the operator of 3D inverse Fourier transform. The Vainberg
image yields amplification of the high frequency components in comparision
with the exact tomogram.
Some results of numerical simulations are presented in Fig.3. Fig-
ure 3a is a cross-section of a three-dimensional phantom, and Fig.3b
shows the Vainberg image for this phantom reconstructed from 25 two-
dimensional projections (from [28]).
CONCLUSION
As was shown, local tomography can give promissing results for the
problems of gas and plasma flow vizualization. Further development
of reconstruction algorithms could be based in some of it parts on modern
wavelet analysis [1,10,21,31], which is already penetrating into tomography
methods [6,18,60].
Wavelet decomposition can be used as a tool for noise reduction through
the wavelet shrinkage approach [25,26] or for some contrasting filtering
of projections before their processing by A usual tomography filter.
This work was partially supported by RFBR (Russia), DFG and BMBF
(Germany).
References
1 A. Aldroubi and M. Unser (eds), Wavelets in medicine and biology, Boca
Raton, CRC Press (1996), p.616.
2 H.H. Barrett and W. Swindell, Analog. reconstruction methods for transaxial
tomography, Proc. IEEE, Vol.65, No.1 (1977), pp.89-107.
3 D.M. Benenson and A.J. Baker, AIAAJ., 9, No.8, 1441-1446 (1971).
4 D.M. Benenson and A.A. Cenkner, Trans. ASME. Ser. C.J. Heat Transfer,
92, No.2, 276-284 (1970).
5 C.A. Berenstein and D.F. Walnut, Wavelets and local tomography. Wavelets
in medicine and biology, A. Aldroubi and M. Unser (eds), Boca Raton,
CRC Press (1996), pp.231-261.
6 M. Bhatia, et al, IEEE Trans. Med. Imag., 15, No.1, 92-101 (1996).
Fig.3 Cross sections of 3D phantom and its Vainberg image.
131
7 S. Cha and C.M. Vest, Appl. Opt., 20, No.16, 2787-2794 (1981).
8 R.C. Chase, et al, Opt. Eng., 20, No.3, 486-492 (1981).
9 D.F. Da Cruz Jr. and A.J.H. Donn'e, Rev. Sci. Instrum., 61, No.10, 3067-
3069 (1990).
10 I. Daubechies, Ten lectures on wavelets, SIAM, Philadelphia (1992), p.357.
11 S.R. Dean, The Radon transformation and some of its applications, John
Wiley, New York (1983), p.289.
12 A. Faridani, Results, old and new, in computed tomography. Inverse
problems in wave propagation, G. Chavent et al (eds), The IMA vol-
umes in math. and appl., Vol.90, Springer Verlag, New York (1997), pp.167-
193.
13 A. Faridani, et al, J. Appl. Math., 52, 459-484 (1992).
14 G.W. Faris and R.L. Byer, Opt. Lett., 12, No.2, 72-74 (1987).
15 E.F. Gippius, et al, Tomography research of the transversal HF-discharge
plasma, ESCAMPIG-92, St Petersburg, EPS, Vol.16F (1992), pp.330-
331.
16 R.S. Granetz and P. Smeulders, Nucl. Fusion, 28, No.3, 457-476 (1988).
17 G.T. Herman, Image reconstruction from projections: The fundamentals
of computerised tomography, Academic Press, New York (1980), p.316.
18 M. Holschneider , Inverse Problems, 7, 853-861 (1991).
19 L.C. Ingesson, et al, Rev. Sci. Instrum., 66, No.1, 622-624 (1995).
20 L.C. Ingesson and V.V. Pickalov, J. Phys. D: Appl. Phys., 29, No.12,
3009-3016 (1996).
21 G. Kaiser, A friendly guide to wavelets, Birkhauser, Boston (1994), p.300.
22 A.C. Kak and M. Slaney, Principles of computerised tomographic imaging,
IEEE Press, New York (1988), p.329.
23 A.I. Katsevich and A.G. Ramm, Inverse Problems, 11, No.5, 1005-1023
(1995).
24 A.I. Katsevich and A.G. Ramm, J. Appl. Math., 56, No.1, 167-191 (1996).
25 E.D. Kolaczyk, J. Amer. Stat. Assoc., 91, No.9, 1079-1089 (1996).
26 E.D. Kolaczyk, An application of wavelet shrinkage to tomography. Wavelets
in medicine and biology, A. Aldroubi and M. Unser (eds), Boca Raton,
CRC Press (1996), pp.77-92.
27 A.V. Likhachov and V.V. Pickalov, Nucl. Instrum. Meth. Phys. Res. (A),
359, No.1-2, 370-375 (1995).
28 A.V. Likhachov and V.V. Pickalov, Nucl. Instrum. Meth. Phys. Res. (A),
(1998) (to appear).
29 I.H. Lira and C.M. Vest, Appl. Opt., 26, No.5, 774-776 (1987).
30 I.H. Lira and C.M. Vest, Appl. Opt., 26, No.18, 3919-3928 (1987).
31 A.K. Louis, et al, Wavelets: Theory and Applications, John Wiley, Chichester
(1997), p.324.
32 C.D. Maldonado and H.N. Olsen, JOSA, 56, No.10, 1305-1313 (1966).
33 T.S. Melnikova and V.V. Pickalov, Beitr. Plasmaphysik, 22, No.2, 171-
180 (1982).
34 T.S. Melnikova and V.V. Pickalov, Beitr. Plasmaphysik, 24, No.5, 431-
445 (1984).
35 T.S. Melnikova and V.V. Pickalov, The electric arc in a gas flow, Proc.
17th ICPIG, Budapest, Contr. papers, Vol.2 (1985), pp.789-791.
36 T.S. Melnikova and V.V. Pickalov, Soviet Journal of Applied Physics -
Scripta Technica, 1, No.6, 61-70 (1987).
37 T.S. Melnikova and V.V. Pickalov, Computer-aided plasma tomography.
132
High-temperature dust-laden jets in plasma technology, O.P. Solonenko
and A.I. Fedorchenko (eds), Proc. Intern. Workshop, Novos., USSR (1988),
VSP, Utrecht, The Netherlands (1990), pp.257-282.???
38 C.A. Monnig, et al, Spectrochimica Acta (B), 43, No.9-11, 1217-1233
(1988).
39 B.R. Myers and M.A. Levin, Rev. Sci. Instrum., 49, No.5, 610-616 (1978).
40 Y. Nagayama, Jpn. J. Appl. Phys., 24, No.6, 787-788 (1985).
41 Y. Nagayama, J. Appl. Phys., 62, No.7, 2702-2706 (1987).
42 Y. Nagayama and A.W. Edwards, Rev. Sci. Instrum., 63, No.10, 4757-
4759 (1992).
43 Y. Nagayama, et al, Phys. Rev. Lett., 61, No.16, 1839-1842 (1988).
44 F. Natterer, The mathematics of computerised tomography, Wiley, New
York (1986), p.222.
45 A.P. Navarro, et al, Rev. Sci. Instrum., 52, No.11, 1634-1643 (1981).
46 H.N. Olsen, et al, JQSRT, 8, 1419-1430 (1968).
47 V.V. Pickalov and N.G. Preobrazhenskii, Sov. Phys. Usp. (AIP), 26, No.11,
974-990 (1983).
48 V.V. Pickalov and T.S. Melnikova, Beitr. Plasmaphysik, 24, No.4, 417-
430 (1984).
49 V.V. Pickalov and T.S. Melnikova, Plasma tomography [in Russian], Nauka,
Novosibirsk (1995), p.229.
50 A. Plessl, Appl. Phys., 21, No.4, 377-389 (1980).
51 W. Schielke, et al, Surface and Coating Technology, 74-75, 552-561
(1995).
52 W. Schielke, Surface and Coatings Technology (1998) (to appear).
53 N. Sebald, Appl. Phys., 21, No.3, 221-236 (1980).
54 S. Sugimoto, et al, J. Appl. Phys., 66, No.11, 5228-5231 (1989).
55 D.W. Sweeney and C.M. Vest, Intern. J. Heat and Mass Transfer, 17,
No.12, 1443-1454 (1974).
56 A.N. Tikhonov and V.Y. Arsenin, Solutions of ill-posed problems, Winston,
New York (1977).
57 W. Tiller, Appl. Phys., 25, No.3, 317-328 (1981).
58 E.I. Vainberg, et al, Soviet J. Non-dest. Test, 17, 415-423 (1981).
59 G.N. Vishnyakov, et al, Opt. Spectrosc. (USSR), 62, No.6, 803-806 (1987).
60 D. Walnut, Local inversion of the Radon transformation in the plane using
wavelets. Wavelet Application in Signal and Image Processing, Proc.
SPIE, Vol.2034, San Diego, C.A. (1993), pp.84-91.
61 Yu.A. Zakharenkov, et al, Laser and Particle Beams, 8, No.1-2, 339-
342 (1990).
133
New Vortex Method of Plasma Insulation and the Ranque Effect
New Vortex Method of Plasma Insulation
and the Ranque Effect
A. Gutsol
1)
and J. A. Bakken
2)
1)
Institute of Chemistry and Technology, Kola Science Centre of Russian Academy of
Sciences, Apatity, Murmansk Region, 184200, Russia
2)
Department of Metallurgy, Norwegian University of Science and Technology,
N-7034 Trondheim, Norway
The efficiency of thermal insulation of microwave generated plasma
using reverse vortex flow was investigated experimentally and by numerical
simulations. Comparison was made with the conventional vortex method
of plasma insulation. Changing the location of the vortex inlet to the
exit end of the plasma torch leads to a significant decrease of the heat
loss to the wall: from 30% to 5%. This result is in contradiction with
the traditional explanation of the Ranque effect. A new simple explanation
of the Ranque effect of energy separation in the vortex tube is proposed.
Energy separation takes place due to radial motion of turbulent micro-
volumes with different tangential velocity in the strong centrifugal field.
The new model of energy separation explains such apparently mysterious
phenomena as counter-rotating of the central vortex flow layers observed
in some experiments and in numerical simulations. A new approach for
consideration of the confined vortex flows is defined.
1. INTRODUCTION
The vortex method of plasma stabilization and insulation is well known.
In this method the swirl generator is placed upstream relative to the
electric discharge and the outlet of the plasma jet is directed to the
opposite side. It is well known that in intensive vortex streams a central
recirculation zone of reverse flow occurs near the swirl generator. The
recirculation flow results in an upstream transfer of energy from the
center of the vortex stabilized plasma, and a significant part of this
energy arrives at the plasma torch walls and becomes lost. This well-
134
known phenomenon demands sufficient cooling of the plasma torch walls.
Apart from power losses the existence of a reverse flow along the
axis of intensive swirled flows can have other unpleasant consequences.
So, in our experiments on processing of fluoride solutions of rare re-
fractory metals salts [1, 2] the reverse flow in the microwave (MW)
plasma generator (Fig. 1a) brought a portion of solution sprayed in the
top part of the plasma chemical reactor 4, specifically in the connecting
cone 5, into the MW discharge zone 3. It resulted a considerable modified
discharge regime, that was observed as a change of its color, and a
change of the magnetron current, which is affected by the microwave
power reflected from the discharge. Besides, interaction of the fluo-
ride solution and its decomposition products with the quartz tube 1 of
the MW plasmatron reduced the service life of this tube and resulted
in pollution of the produced oxides by silicon.
To prevent penetration of the solution into the plasmatron we re-
placed the cylindrical discharge tube with a specially made tube (Fig.
2) with quartz ribs 2 welded on the outlet end. It was possible to observe
that the plasma flow from the standard cylindrical tube into ambient
air was divergent, while the tube with ribs formed a jet of plasma.
Though the reduction of length of the reverse flow zone suppressed
penetration of the solution into the plasmatron, it, probably, did not improve
a b
2
1
3
5
4
6
7
Fig. 1 Scheme of the MW plasma torch with supposed flow patterns of gas and plasma.
(a) - old scheme with flow patterns for conventional vortex plasma stabilization;
(b) - new scheme with flow patterns for reverse vortex plasma stabilization. 1 -
cylindrical quartz tube; 2 - original swirl generator; 3 - plasma; 4 - plasma chemical
reactor; 5 - connecting cone; 6 - water-cooled diaphragm; 7 - additional swirl generator.
135
significantly the power characteristics of the plasma generator, and the
thermal losses still were rather high as testified by intensive heating
(up to luminescence) of the outlet end of the modified quartz tube.
According to the new concept of plasma insulation [3, 4] based on
the idea of reverse vortex formation [5] the outlet of the plasma jet
is directed to the swirl generator side. In this case the plasma gas should
enter the discharge zone from all sides except the outlet side, and no
significant recirculation zone should be formed.
Flow pattern in the conventional vortex unit (Figure 1a) should be
similar to that in uniflow vortex tube, and in reverse vortex unit (Figure
1a) should be similar to that in the most effective counterflow vortex
tube [6, 7]. So, it seems reasonable to compare the results of inves-
tigation for efficiency of thermal insulation in two mentioned vortex
systems with a theory of the Ranque effect.
2. EXPERIMENTS
Experiments were made with a microwave (MW) plasma generator with
a MW power input up to 5 kW. This plasma torch is a part of an
experimental facility for treatment of inorganic salt solutions [1, 2]. A
Fig. 2 Scheme of the modified discharge tube for MW plasma torch. 1 - cylindrical
quartz tube; 2 - welded quartz ribs.
1 2
136
sketch of the MW plasma torch with supposed flow patterns of gas
and plasma is shown in Fig. 1 [3, 4]. The quartz discharge tube 1 (inner
diameter 44 mm, length about 140 mm) passes perpendicularly through
two wave-guides (9045 mm
2
, unshowed) which supply H
10
mode of
the MW energy (frequency 2.4 GHz) from two magnetrons. In the
conventional scheme (Fig. 1a) the plasma gas (air or nitrogen) enters
the discharge chamber through four inlet openings of the original tangential
swirl generator 2, resulting in stabilization of the plasma 3 on the axis
of the quartz tube 1 by the strong rotation of the gas. In the experimental
plasma-chemical set-up [1, 2] the MW plasma torch is joined to the
uncooled massive steel reactor 4 by an uncooled steel connecting cone
5.
For experiments with reverse vortex stabilization (Fig. 1b) an additional
vortex generator 7 with a water-cooled diaphragm 6 (diameter 26 mm)
was installed between the quartz tube and connecting coin. Calorimetric
and electrical measurements permitted to determine the MW power input
W
p
into the discharge and the heat losses W
t
to the water-cooled parts
of the plasma torch. (Unfortunately it is not possible to use water cooling
for the quartz tube of the MW plasma torch, but as this tube is sur-
rounded by the water-cooled parts of the plasma torch almost on all
sides, it was supposed practically all the heat from the quartz tube ends
up in the cooling system due to convection.)
The experimental results [3, 4] are presented in Fig. 3 (dots with
full curves) in dependence on J - the energy input into the discharge
per unit mass of plasma gas consumption. The power input was around
3.5 kW and varies a little due to the fact that changing the gas flow
conditions also influences the discharge conditions. The dots of curve
1 were obtained for the old scheme (Fig. 1a) without the diaphragm
and with the plasma-chemical reactor. Curve 2 corresponds to the same
scheme, but with the diaphragm. Curve 3 corresponds to the new
reverse vortex flow scheme (Fig.1b) with the reactor. As the heat flux
to the plasma torch walls from the reactor was significant, two ad-
ditional series of experiments were made. The reactor was removed,
the plasma torch was turned upside-down, and a hot plasma jet was
directed upwards into ambient air. As the heat losses in the old scheme
without the diaphragm were extremely large for all energy inputs, only
the new scheme (Fig. 1b) was used in the additional experiments.
Plasma gas might be supplied through the original vortex generator (2,
Fig.1) for realizing the conventional vortex stabilization scheme, or through
the additional vortex chamber (7, Fig. 1) for realizing the reverse vortex
scheme of plasma stabilization. Curves 2' and 3' (Fig. 3) correspond
to these two cases.
137

J [kJ/g]
0
10
20
30
40
50
2
W
t
/W
p
[%]
1 2 3 4
3
3
1
2
The experimental investigation showed that if the plasma was stabilized
by the conventional vortex flow the energy loss to the plasma torch
walls might exceed 30 %. With the reverse vortex flow the energy
loss was only about 5 %. The heat loss in a simple system of this type
corresponds to the low heat loss in plasma generators with a porous
discharge chamber, which, however, is very complex end expensive to
manufacture. More over, if the reverse vortex flow configuration is
used, almost all the plasma forming gas should pass through the dis-
charge zone. As the axial velocity in the top region of the reverse
vortex should be quite low, discharge stability problems should not occur.
As the flow direction should be constant throughout the axial region,
it seems possible to inject additional gas or particles into the top of
the reverse vortex. Tests with ZrO
2
+Y
2
O
3
powder were made in the
described MW facility. The introduction of the powder into the top
of plasma torch with the reverse vortex flow plasma stabilization ensures
melting and spheroidization of particles up to 100 mm.
So, the reverse vortex system seems very promising for different
plasma-chemical processes and other technical applications utilizing mi-
Fig.3 Heat losses in the microwave plasma generator. Full curves - experiments, broken
curves - numerical simulations. 1 - old scheme without the diaphragm and with the
plasma-chemical reactor; 2 - old scheme with the diaphragm and with the plasma-
chemical reactor; 2' - old scheme with the diaphragm and without the plasma-chemical
reactor (experiment and simulation); 3 - new scheme with the diaphragm and with
the plasma-chemical reactor; 3' - new scheme with the diaphragm and without the
plasma-chemical reactor (experiment and simulation).
138
crowave plasma devices as well as DC and AC plasmatrons, RF in-
duction plasma torches, and probably gas burners.
3. NUMERICAL SIMULATIONS
The numerical simulations of the MW plasma torch were made using
the fluid flow and heat transfer simulation program Fluent. In the 2D
axisymmetric geometry the conservation equations for mass, energy and
radial, axial and azimutal momentum were solved simultaneously. It is
worth while to make a few remarks on our experience with the FLUENT
program.
First, it should be noted that the simulation of rotating flows requires
an order of magnitude more computational time, than simulation of flows
without rotation in the same geometry.
The second remark concerns the till now widely used use k- model
of turbulence. The program allows to carry out simulation using various
modifications of this model, and with the Reynolds Stress Model (RSM)
for turbulence. The second model involves calculation of the individual
Reynolds stresses. The improved versions of the FLUENT program
allow to take into account the directional diffusivity of turbulence and
a non-equilibrium wall function. In the recommendations concerning ways
of simulating flows in mass force fields, and, in particular, of rotating
flows, it is noted that the k- model gives considerably less exact results,
than the Reynolds Stress Model. However, simulations using the k-
model require much less time. So, it is recommended to start a simulation
with the help of the k- model, and then to use the results obtained
as the first approximation for next simulations.
Our experience with trial simulations of elementary rotating flows
has shown, that in some cases the application of the k- model gives
completely unrealistic results, the use of which even as the first ap-
proximation can only complicate further simulation. As an example we
demonstrate in Fig. 4 the obtained stream lines in the axial plane for
the rotating flows of water. The water enters tangentially into the cylindrical
vessel (diameter 44 mm, length 100 mm) through a circular slit (width
2.5 mm) and leaves the vessel through the diaphragm (diameter 26 mm)
in one end. The incoming velocity of water is 1 m/s and the initial tangential
velocity is 10 m/s. The slit is located on the cylindrical surface (Fig.
4-1) next to the end with the diaphragm or on this end (Fig. 4-2 and
4-3). It is quite obvious that the flow patterns in all these cases should
be similar, but simulation using the k- model of turbulence for two
models gives completely different pictures (Fig. 4-1 and 4-2), which
differ not only by the size of a recirculation zone, but even by the flow
direction in this zone. At the same time, simulation with use of the RSM
139

1
2
3

Fig. 4 Stream lines in the axial plane for the rotating flows of water. The flow of water
enter tangentially into the cylindrical vessel (diameter 44 mm, length 100 mm) through
the circular slit (width 2.5 mm) and leave the vessel through the diaphragm (diameter
26 mm) in left end. The incoming velocity of water is 1 m/s and the initial tangential
velocity is 10 m/s. 1 - slit on the cylindrical surface, k- model; 2 - slit on the left end,
k- model; 3 - slit on the left end, RSM model.
model shows, that the flow pattern does not change so dramatically
with change of the water input location.
To account for turbulence under the experimental [3, 4] conditions,
the Reynolds Stress Model, as the more suitable for rotating flows, was
used. In the near-wall region the program used the logarithmic law for
velocity. The law-of-the-wall for temperature in FLUENT comprises
two different laws: a linear law for the thermal conduction sublayer
and a logarithmic law for the turbulent region. The minimum size of
the grid cell near the wall was about 0.3 mm. The heating of the plasma
was assumed to take place in an idealized uniform heat generation
zone in the central region of the rotating flow. The pre-described heating
zone length was 120 mm. Its diameter was varied, but was usually 26
140
mm. The metallic parts of the plasma generator were supposed to have
a constant temperature of 300 K. On the cylindrical wall convective
cooling with a heat transfer coefficient of 50 W m
-2
K
-1
and radiation
cooling with an external emissivity of 0.8 were assumed. Test calcu-
lations showed that the most realistic flow pattern in the inlet regions
was obtained in a 2D geometry when the discrete tangential gas in-
let jets were simulated by fixing the rotating velocity in the cells next
to the cylindrical wall and defining the appropriate mass sources for
the same cells.
Figure 5 shows on the axis plane the stream lines, profiles of axial
velocity and the temperature distribution for the conventional vortex
scheme, and Figure 6 - for the new reverse vortex flow scheme.
In these two modeling cases the heating zone (3.5 kW, length 0.14 m,
diameter 44 mm) was in the center of MW plasma torch quartz tube
(1, Figure 1). Gas (nitrogen) enters tangentially into discharge chamber
(conventional scheme) or into an additional vortex chamber (new reverse
vortex flow scheme). The initial velocity of the tangential gas jets were
estimated from the experimental data of the pressure drop and the gas
consumption. In two cases showed in Figures 5 and 6 the initial tan-
gential velocities (and the gas consumption) are equal 100 m/s (1 g/
s) and 225 m/s (1.5 g/s) accordingly. It is easy to see (Fig. 6) that
300 1130 1960 2790 3620 4440 5270 6100
Temperature [K]
Fig.5 Temperature distribution, stream lines and profiles of axial velocity for three
different cross-sections and for outlet of the MW plasma torch with conventional
vortex flow. Numerical simulations for the uniform heat source (power 3.5 kW; length
0.12 m; diameter 26 mm) in the center of the old MW plasma torch (length 0.14 m;
diameter 44 mm). Gas (nitrogen, 1 g/s) enters tangentially with velocity 100 m/s through
four narrow inlet openings into the discharge chamber. The recirculation flow results
in an upstream transfer of energy from the center of the vortex stabilized plasma.
141
the reverse vortex compresses the heat zone and protects the plasma
torch walls from overheating. As it was supposed [3-5], the main part
of the plasma gas passes through the high temperature discharge zone
and size of recirculation zones are considerably reduced. In the old
scheme (Fig. 5), on the contrary, the main part of the incoming gas
mixes with hot recirculated flow and moves along the cylindrical quartz
wall thus bypassing the discharge zone. The calculated energy losses
for the appropriate cases are shown by dots of the broken curves on
Fig. 3 (curve 2' - for the conventional vortex scheme, curve 3' - for
the new reverse vortex flow scheme). A discrepancy between ex-
perimental and calculated results may occur due to the oversimplified
description of the discharge zone, and probably also because of ex-
perimental errors. Nevertheless, this discrepancy is small enough to conclude
that the energy losses might be reliably predicted for other electric discharges
and in flames by employing numerical simulation models.
4. CONTRADICTION WITH THE TRADITIONAL RANQUE EFFECT
THEORY
The results of experiments and numerical simulations of conventional
Temperature [K]
285 833 1380 1930 2480 3030 3580 4120
Fig.6 Temperature distribution, stream lines and profiles of axial velocity for three
different cross-sections and for outlet of the MW plasma torch with reverse vortex
flow. Numerical simulations for the uniform heat source (power 3.5 kW; length 0.12
m; diameter 26 mm) in the center of the new MW plasma torch (length 0.14 m;
diameter 44 mm). Gas (nitrogen, 1.5 g/s) enters tangentially with velocity 225 m/s
through three narrow inlet openings into the additional vortex chamber (swirl generator).
The reverse vortex compresses the hot zone and protect the plasma torch walls
from overheating.
142
and reverse vortex flow systems contradict the traditional theory of
the Ranque effect [6, 7].
It is worth while noting that modern ideas about the properties and
possibilities of confined vortex flows are quiet contradictory. Confined
vortex flows are widely used in different devices with different, of-
ten opposite aims. So in vortex and cyclones combustors vortex flows
improve heat exchange inside the combustor and with its walls. At the
same time in vortex plasma torches swirling flows stabilize the plasma
fluid at the axis of the device and protect the torch walls from overheating.
It is well known that the vortex flows are broadly used for dust separation,
but at the same time there is an experimental fact that small particles
with a given size concentrate at a defined radius of the vortex flow.
The gas nuclear missile program, which was developed in USA at the
50 - 60-ies was partially based on this effect [7]. The typical high turbulence
level of vortex flows is generally accepted, but in a well known ex-
periment [8] a combustion flame was laminarized by rotation of a cylindrical
wire net screen.
All these apparently contradictory experimental data were obtained
for very similar flows, only the location and method of vortex formation
were different. The main part of relevant publications on vortex flows
concerns technical applications, and the authors of these publications
failed to notice that their data were in contradiction with data obtained
in other technical branches. Nevertheless, there is one effect which
is known from the early 30-ies as the Ranque (or vortex) effect in which
many of the contradictions mentioned above are focused. So it is impossible
not to notice contradictions when one tries to interpret this effect which
sometimes seems quite mysterious.
The substance of the Ranque effect is that in technically very simple
vortex devices without any moving parts (Fig. 7), an initial isothermal
gas stream, which enters into the tube 1 (length L, diameter D, L >>
D) through the tangential inlet 2, divides in two flows with different
temperatures: a cold axial flow 5 leaves the tube through the cen-
tral opening of the outlet 5 (diameter d
c
~ D/2) and a hot peripheral
one 4 regulated by the throttle 3. This effect was established by Ranque
in 1931 during an investigation of the temperature field inside a dust
separation cyclone. From that time until today the flow of papers at-
tempting to explain this effect is quite considerable. The main resent
publications were reviewed in [9]. It should be noted that the litera-
ture concerning the Ranque effect published in Russian is numerous,
but almost unknown for west scientists. The references on the main
early west publications may be found, for example, in [10]. Among this
flow of papers there are only few original publications with quiet se-
143
rious attempts of explanations. The main part of publications contain
different modifications of traditional theories. The authors of some others
papers trying to explain the Ranque effect forgot the basic physical
principles. For example, in [11] it is supposed that the centrifugal force
affect on the individual molecules, but according to the principles of
gas-kinetic theory the individual molecules move in a straight lines.
The predominant explanation [6, 7] is based on the concept of in-
tensive turbulent heat flux from the center of the vortex tube to the
periphery - i.e. from the cold zone to the hot one! - due to radial adiabatic
motion of fluid elements in the large pressure gradient. According to
this theory, inside high-speed vortex flows and, first of all, inside high-
speed reverse vortex flows the intensive radial heat flux should ex-
ist until the temperature distribution corresponds to the adiabatic one.
The temperature field inside the nonuniform pressure field is said to
have an adiabatic distribution when the temperature of the adiabatically
moving test gas volume is equal to the surrounding temperature. Ap-
proximately adiabatic temperature distribution is, for example, found
in the Earth atmosphere. So, according to the predominant theory of
the Ranque effect the heat transport is almost always directed from
the center of the vortex tube to the periphery irrespective of the temperature
A
L
D
d
c

2 A 3 4 5
2
5
L
d
c

D
A 3 1 4
A
A-A
A-A
a
b
1
Fig. 7 Scheme of the counterflow (a) and parallel-flow (b) Ranque vortex tubes. 1 -
cylindrical tube; 2 - tangential gas inlet; 3 - throttle; 4 - outlet of hot gas; 5 - outlet
of cold gas; L - length of the tube; D - diameter of the tube; d
c
- diameter of the cold
gas outlet.
144
distribution. But, to our mind, to suppose that strong turbulence pro-
motes such radial temperature distribution is the same as to expect that
intensive mixing may promote stratification of two different liquids.
Thus, according to the traditional explanation of the Ranque effect,
the heat flux from the central region of a reverse vortex flow stabi-
lized plasma torch should be larger than for the conventional vortex
one, because the flow pattern in reverse vortex unit is similar to that
in the most effective counterflow vortex tube.
5. NEW EXPLANATION OF THE RANQUE EFFECT
It was proposed [4, 5, 9] a new simple explanation of the Ranque effect
to be described in the following (Fig. 8): The initial isothermal gas stream
entering through the tangential inlet 2 with a non-uniform velocity distribution
3 becomes much more turbulent during the interaction with the cylindrical
wall of the tube 1 and with the main vortex flow. So, inside this turbulent
rotating bulk flow, microvolumes with different circumferential velocity,
but with equal temperature, appear. It is well known that in a centrifugal
field elements with low tangential velocity 6 move to the axis and elements
with high tangential velocity 7 move to the periphery. In the coordinate
system connected with the bulk rotating flow the resulting force F (Fig.
F
F
7
6
4
2
3
5
1
Fig. 8 Proposed scheme of the energy separation in the vortex tube.1 - cylindrical
wall of the tube; 2 - tangential gas inlet; 3 - profile of inlet velocity; 4 - turbulent
micro vortex formed near the cylindrical wall, direction of its rotation is opposite to
the main one; 5 - turbulent micro vortex formed in the area of mixing of the main
vortex flow with the incoming tangential stream, direction of its rotation is the same
as the main one; 6 - microvolume with low tangential velocity and the force F acting
on this volume; 7 - microvolume with high tangential velocity and the force F acting
on this volume.
145

T
c
[ K ]
- T
c
[ K ]
- 4 0
0
1 0
2 0
1
0 . 8 0 . 6 0 . 4
0 . 2
8) accelerates these elements in the opposite radial directions. In this
way a radial separation of elements with different kinetic energy takes
place. The subsequent adiabatic expansion of the central elements with
low kinetic energy in the radial pressure field of the vortex flow produces
the low-temperature flow. The deceleration of the peripheral high-kinetic
energy elements due to friction produces the high-temperature flow.
Numerical estimations based on such a simple model [9] are in reasonable
good agreement with the usual vortex tube characteristics (Fig. 9): the
difference between initial and cold gas temperatures T
c
and cold capacity
T
c
in dependence on the mass fraction of cold gas.
It is necessary to note, that separation of fluid elements with dif-
ferent tangential velocity and, as the consequence, with different ki-
netic energy in a of centrifugal force field was observed and earlier
in experiments and simulations, however the obtained results were not
analyzed satisfactory. The mentioned results were received at study
of a motion of a liquid in bends of pipelines. For a laminar flow of a
liquid the maximum velocity in a bend of a pipe does not exceed those
in a direct pipe [12, 13]. In a case of a turbulent flow [14] centrifu-
gal forces, arising at a movement liquids in a bend of the pipeline, cause
Fig. 9 Characteristics of the vortex tube. - mass fraction of cold outlet gas;
T
c
- temperature difference between incoming gas and cold outlet gas; T
c
- relative
heat capacity of cold outlet gas; solid line - experimental data, dotted line - theo-
retical estimations based on proposed model.
146
such redistribution of elements of a flow, that maximum velocity in a
bend exceeds those in a direct pipe. Thus maximum of speed, natu-
rally moved from an axis of turn of the pipeline. The occurrence of
a zone of a flow with velocity, exceeding maximum initial, means re-
distribution of energy on radius, that is the same process, which oc-
curs in a vortex tube. Thus, the bend of a pipe represents simplified
vortex tube.
It is necessary to understand, that process of spatial redistribution
of kinetic energy and, as a consequence, complete energy is very widespread
phenomenon. Perhaps simplest example is a laminar flow in a pipe. After
finishing the transient processes, when due to the thermal conductiv-
ity all elements of a flow have got an identical temperature, it appears,
that the kinetic and the complete energy for central elements is higher,
than for peripheral.
In a vortex tube after the process of the kinetic energy redistribution
the process of adiabatic expansion of the central elements with low
kinetic energy is followed, results in formation of a cold flow. The
deceleration and heating peripheral elements of a flow is the unessential
process, it is quite possible to imagine and, to some extent, to real-
ize a situation, when a peripheral flow will not be heated up signifi-
cantly (for example, in the case of intensive rotation of the energy separation
chamber (cylindrical tube 1 on Fig.7)).
This new model explains [9] the main features of all the available
experimental facts including such apparently mysterious phenomena as
counterrotating of the central vortex flow layers observed in some
experiments [15] or existence of unrotated central jets inside vortex
flows [16] . Fig. 8 shows two main regions of the turbulence forma-
tion inside the vortex tube: a boundary layer near the cylindrical wall
and an area of mixing of the bulk vortex flow with the incoming tangential
flow. Turbulent micro-vortexes 5 formed in the area of mixing rotate
in the same direction as the main vortex one. Turbulent micro-vortexes
4 formed near the cylindrical wall (in the coordinate system connected
with the bulk vortex) rotate in opposite direction in relation to the bulk
vortex If large amount of such micro-vortexes with low mean tan-
gential velocity and opposite direction of rotation arrive the central region
of the main vortex then this central region rotates in opposite direc-
tion.
The improved program (FLUENT with RSM and consideration of
directional diffusivity and non-equilibrium wall function) was used for
numerical simulation of the confined reverse vortex flow (Fig. 10). In
this modeling case air enters the cylindrical vessel (length 0.1 m; di-
ameter 44 mm) through the circular slit (2.5 mm) with tangential ve-
147
Fig.10 Stream lines and tangential velocity W profiles for confined reverse vortex flow.
Simulation of the air flow in the cylindrical vessel (length 0.1 m; diameter 44 mm).
Air enters through the circular slit (2.5 mm) with tangential velocity W
0
= 100 m/s and
radial velocity 3 m/s.
locity W
0
= 100 m/s and radial velocity 3 m/s. It is possible to see in
Fig. 10 the negative tangential velocity W near the axis of the vessel.
6. NEW MODEL FOR CONSIDERATION OF CONFINED VORTEX
FLOWS
Existence of the revised conceptual approach permits to examine, on
the basis of a unified model, all experimental results concerning confined
vortex flows. The main points of this model are as follows:
1. High turbulence level, which is a characteristic feature of vortex
flows, is not a product of these flows themselves. As usual, turbulence
may be created in a region of vortex formation, in a boundary layer
near the cylindrical wall or in an area of mixing of the vortex flow with
additional incoming flows. In ordinary non-rotating flows the thickness
of such turbulence production regions increases in the main flow di-
rection. In vortex flows, however, there is another transport mecha-
nism for turbulence. From the narrow regions of turbulence produc-
tion the low tangential velocity microvolumes move to the axis, and high-
velocity ones to the periphery. Thus, the presence of these radially moving
microvolumes with tangential velocities different from the average rotational
speed inside bulk vortex flow is the reason for the observed signifi-
cant level of turbulence in the whole system.
148
2. The main specific features of vortex flows in different devices
like heat exchange intensification or the apparently contradictory ef-
fect of laminarization of flames [8] are associated with the presence
of recirculation zones. The number and shape of such zones (for example,
see Figures 4, 5, 6 and 10) and the direction of gas circulation in them
depend critically on the inlet conditions, wall roughness and system geometry.
So, utilization of the usual integral characteristics of vortex flows such
as the swirl number is often not sufficient to describe the main features
of these flows, especially with counterflow geometry.
7. DISCUSSION
As it was shown changing the location of the vortex inlet to the exit
end of the plasma torch leads to a significant decrease of the heat losses.
A new method of reducing the harmful influence of the reaction zone
on the surrounding area and, first of all, on the device walls [5] is based
on such simple design modifications. The unwanted influence may, for
instance, be due to the high temperature or high corrosion activity of
the plasma, flame or reaction products formed in plasma generators
or in combustion chambers. It may also be necessary to prevent
contamination of high purity products caused by partial melting, evaporation
or dissolution of the chemical reactor walls. The basic principle of the
method [5] is to locate the reaction products outlet at the axis in the
same end of the reactor as the pipe shaped inlet of the reactants and,
i.e. on the inside of this pipe shaped inlet. When the flow direction of
the products is opposite to the initial flow of the reactants then no significant
recirculation zone is formed inside the reactor. This method is promising
for energy saving (for example in gas-fired metallurgical furnaces, domestic
gas heaters and plasma torches), for protecting the environment
(laminarization of combustion should lead to a decreased CO and NO
x
production), for designing new simplified devices (for example, combustors
of jet engines or plasma torches with uncooled walls). New approach
for consideration of the confined vortex flows may give new practical
results in increasing the efficiency of various cold devices: vortex
tubes for gas cooling and gas separation, dust cleaning cyclones, chemical
reactors, etc.
8. CONCLUSION
Simple design modification of vortex stabilization system leads to a significant
decrease of the heat flux to the walls of plasma torches (from 30%
to 5% for microwave plasma). This result is in contradiction with traditional
theory of the Ranque effect, but is in good agreement with the new
simple explanation of the energy separation in vortex tubes described
149
here. Advanced numerical simulation methods permit reliable predictions
of heat losses in plasma torches with conventional as well as reverse
vortex flows stabilization. Experimental investigations and numerical
simulations show that reverse vortex systems are very promising for
various plasma-chemical processes and other technical applications of
different types of plasma devices (non-transferred arcs, RF induction
and MW plasma torches) as well as for gas burners.
Acknowledgments
This cooperative work was made possible by financial support from The Research
Council of Norway in the frame of The Nordic Scholarship Scheme for The
Baltic Countries and Northwest Russia.
References
1 A.F. Gutsol, Khimiya vysokikh energiy (High Energy Chemistry), 29 No.5,
373-376 (1995).
2 A.F. Gutsol and V.T. Kalinnikov, Zhurnal prikladnoy khimii (Russian J.
of Appl. Chem.), 68, No 6, 885-891 (1995).
3 A.F. Gutsol, In: Using of the reverse flows for insulation of the inter-
action zones and qualitative interpretation of the Ranque effect. Chemistry
and technology of processing of complex raw materials of Kola Pe-
ninsula, Apatity, KSC of RAS (1996), pp.68-71.
4 V.T. Kalinnikov and A.F. Gutsol, Doklady Akademii Nauk. (Proceedings
of the Russian Academy of Sciences, Technical Physics), 353, No.4,
469-471 (1997).
5 A.F. Gutsol, Method for the flow system formation. Notice of Allowance
of 1 July 1996 on the Russian Patent Application No 95112323.
6 A.P. Merkulov, The Vortex Effect and Its Application in Engineering (in
Russian), Mashinostroenie, Moscow (1969), p.184 p.
7 A.K. Gupta, et al, In: Swirl flows, Abacus Press, England (1984).
8 N.A. Chigier, et al, Comb. and Flame, 14, 171-180 (1970).
9 A.F. Gutsol, Uspekhi fizicheskikh Nauk (Russian Physics - Uspekhi),
167 (6), 665-687 (1997).
10 J.M. Nash, In: Vortex Expansion Devices for High Temperature Cryo-
genics. Proceedings of the 26
th
Intersociety Energy Conversion Engineering
Conference (IECEC-91),Boston, Massachusetts (1991). Publ. By American
Nuclear Society, USA, Illinois, La Grange Park. Vol.4, pp.521-525.
11 L.M. Dyskin, In: Energetic separation in rotated gas flow, Izvestiya Vuzov.
Stroitelstvo (Proceedings of Higher School. Building), No.11 (1995),
pp.116-118.
12 S.V. Patankar, et al, J. Fluid Mech., 62 (3), 539-551 (1974).
13 M.P. Arnal, et al, Int. J. Heat and Fluid Flow, 13, No.4, 347-357.
14 S.V. Patankar, et al, J. Fluid Mech., 67(3), 583-595 (1974).
15 V.E. Finko V.E. Zhurnal tekhnicheskoy fiziki (Russ. Journal of Tech-
nical Physics), 53 (9), 1770-1176 (1983).
16 G.N. Abramovich and R.S. Trofimov, Inzhenerno-Fizicheskiy Zhurnal (Jour-
nal of Engineering Physics and Thermophysics), 53 (5), 751-757 (1987).
150
Physical Phenomena in a Hollow Cathode
and Interaction of Powder with the
Vacuum Arc
Cherednichenko V.S., Zagorskii A.B., Pavlenko L.K.
INTRODUCTION
Vacuum hollow cathode arc dischargers allow extremely high heat fluxes
to be produced on heated surfaces (up to 10
7
W/m
2
) at low working
voltages (30100 V). These arc discharges are favoured in appropriate
industries because of their high efficiency (up to 8090%) in a range
of pressure from 1 to 10
3
Pa.
The basic design of an arc discharge is a tube with the inner di-
ameter R and working gas being fed through it into the low pressure
chamber. The open end of the tube is directed to the anode. The length
of the cathode tube is several inner diameters of the tube. There are
four main working regimes of hollow arc discharge being distinguished
by the magnitudes of plasma gas flow rate G, pressure P in a vacuum
chamber and discharge current j : 1) normal regime 2) low flow rate
of plasma gas regime, 3) low current regime and 4) high pressure re-
gime. The normal regime is of the greatest interest for its applications
in plasma technologies so the latter one was thoroughly investigated in
the present paper.
While investigating the high current arc in a 500 kW facility, a discharge
was burnt between a hollow tantalum cathode and a flat anode. The latter
one represents itself a metal in a mould. At the normal regime at the
current of up to 300 A the plasma column in the space between the cathode
and the anode had the diffused bell-shaped form. The plasma column
becomes clear, sharp bounded with the visible diameter approximately
equal to the cavity diameter. The sharp space stabilization of the plasma
column shows the leading role of gas-dynamics features of the plasma
gas flow through the hollow cathode and the great influence of discharge
currents own magnetic field on the processes behavior of different kinds
151
Physical Phenomena in a Hollow Cathode and Interaction of Powder.
Fig.1. Hollow cathode arc discharge.
inside the cathode cavity, near the cathode spots and namely inside the
plasma column. Figure 1 shows the photograph of a discharge with a
hollow cathode at the normal high-current working regime. The normal
working regime of the hollow cathode arc discharge is realized under
the conditions of mean current density greater than 10
6
A/m
2
calculated
in the middle of the cathode cross section and the Reynolds number of
the plasma gas flow greater than 100. The normal working regimes are
easy recognized by the large cathode hot zone. The maximum cathode
temperature is established in some distance from the tube exit. The zone
of high ionized plasma, which provides current exchange in the diffu-
sion attachment cathode spot regime is observed inside the cathode cavity.
One of the main discharge main features is that powder and granulated
materials can be introduced and maintained in the plasma column in
the current-conducting zone as a result of electromagnetic and gas-dynamic
forces. The extended stable movement of dispersed materials in the region
of intensive power exchange results in significant advantages of plasma
heating technologies. Control of the particle transient time inside the
arc column allows us to achieve the high temperature needed for dif-
ferent technologies (refining, spheroidization, evaporation, dispersion
changes, production of material vapours, etc.).
To prodive information on the new technical concept and method of
heating powder and granulated materials, Fig.2 and 3 show the pho-
152
Fig.2. Photo of hollow cathode plasma arc discharge through blue optic filter with
transparent band of 350 nm to 450 nm.
Fig.3. Photograph of hollow cathode plasma arc discharge through red optic filter
with transparent bend of 550 nm to 650 nm.
tograph of a hollow cathode vacuum plasma discharge with the same
parameters as the one in Fig.1: current is 1000 A, voltage 35 V, plasma
gas is argon, working pressure is 1.73 Pa, consumption of powder (tantalum)
is 3 10
4
kg/s (1 kg/h). The photograph presented in Fig.2 was made
153
through a blue optical filter transparent in the light wavelength range
from 0.35 mm to 0.45 mm, and the photograph in Fig.3 was produced
using a red optical filter with a transmission band from 0.55 mm to
0.65 mm. The tantalum powder, fraction 20 to 60 m, was inserted into
the central part of the plasma column, is clearly shown there.
The powdered materials are held in stable manner inside the vacuum
plasma column and are heated up to the melting point and higher as
a result of interaction with hot plasma. The main well known advan-
tages of plasma heating can be fully realized in such devices. Special
design of vacuum arc plasma heaters with a hollow cathode, its high
power level (100600 kW) achieved now allow plasma advance tech-
nologies to be realised on a principally new ground in a wide range of
working parameters (pressure, temperature) in order to solve techno-
logical problems (maximum heat flux on particles surface reaches 10
7
-
10
8
W/m
2
). The interaction between powder and plasmas own magnetic
field prevents powder loses even for single particles so 100% processing
efficiency is reached.
Due to these unique features, it has become possible to carry out
on the industrial scale treatment of dispersed materials directly in the
hot plasma flow where they can be heated to the melting point, trans-
formed to the vapour phase, ionized and produce plasma flows of treated
materials (both metals and nonmetals). The developed heating system
allows us to perform not only refining and reducing but also spheroidization
and changes of the dispersion of powders, producing pure vapors and
plasma jet flows for the plasma-chemicals reactions where the treated
materials are efficiently transformed to commercial products or are used
for surface coating. Various gases can be used to organize processes
in dissociated or ionized media. The special structure of the plasma column
and the gases existing there in different phases create appropriate fa-
vourable conditions for unique chemical reactions to run in directions
not possible under the normal conditions. Thus, all the peculiarities of
plasma interactions with the treated materials should be taken into account
when examining the problems of treatment of powder materials in vacuum
plasma for any business plans concerning electro-technology processes.
THE PLASMA HOLLOW CATHODE
The plasma gas fed into the cathode tube supports the diffusion cathode
spot. The cathode temperature is self-adjusted by its interaction with
the highly ionized plasma. For the cathode system with the thermal electrons
emitted by the cathode, with the ion current and back electron current
being taken into account, the charge and energy conservation equations
can be written respectively as follows
154
j j j j
em i oe
+ (1)
j T j T q j
i e k oe e r em
+ + + + + ( ) ( ) 2 2 (2)
where q
r
is the surface radiation flux;
k
, are the absolute values of
the cathode sheath potential drop and emission energy respectively;
is the ionizing potential of the plasma gas; T
e
is the electron temperature
(eV). The current density and radiation flux are given by [1]:
j e n
e T
M
i
e

_
,
1
2
1
2
;
j k A T e
em k
e
kT
k

1
2
(3)
where n is the volume concentration of charged particles, is the ionizing
coefficient, T
k
is the temperature of the cathode surface, K. Since the
given analysis is of the qualitative nature, the Schottkys correction in
the second equation of system (3) is omitted. The coefficients k
1
, k
2
have
the order of unity and take into account the possible influence of the
outer cathode surface.
The energy conservation equation on the inner side of a hollow cathode
with the system of equations (3) taken into consideration can be
expressed in a following form:
j T T
M
m
e
k T k A T e
i e k e
e
T
k k
e
k T
k
e
k
+ + + +

_
,
[ ( ) ] 2 2
2
1
2
1
4
2
2

(4)
Numerical analysis of equation (4) enables the dependence the of cathode
sheath potential drop
k
on cathode temperature T
k
to be determined and
is presented in Fig.4. The state of point (
k
*
, T
*
) on the curve is de-
termined by the minimization of the left side of equation (4) with re-
spect to the potential :
If one takes argon as the plasma forming gas, the approximate value
of
k
*
will be
k
*
5.5T
e
. The temperature T
*
relates to the region
of the start of intense thermal emission. The temperature T
1
corresponds
to the maximum possible level of the back electron flux to the cath-
155
Fig.4. Variation of cathode sheath potential drop versus cathode temperature in an
active zone.
ode at
k
= 0. For the point (
k
*
, T
*
), equation (4) becomes:
1
2
3
1
2
4
2
2
1
n
eT
M
T k T k A T e
e
k k
e
k T
e k
k
_
,
+ + +

( )
*

(5)
with the next condition being satisfied
j
j T
T
oe
i e
e

+ 2
The net result of approximate qualitative solution of equation (4) indicates
that for every cathode temperature there are two regimes which essentially
differ by their cathode discharge mechanism. It is obvious that the upper
curve of solution (4) corresponds to the ascending (stable) branch of
the voltampere characteristic of the cathode sheath (d
k
/ dj > 0) and
that the lower curve of this solution corresponds to descending one (which
needs additional stabilization). In the steady state, the total heat flux
on the cathode surface is equal to zero (q

= 0). For the upper dependence
in Fig.4. the inequality dq

/ dT
k
< 0 holds and dq

/ dT
k
> 0 for the
156
lower one respectively. Therefore, the upper branch of solution (4) is
stable in the electrical and energy sense. When
k
becomes less than
k
*
, the discharge linkage in the hollow cathode transforms from its diffusion
form to contracted form so the cathode temperature approaches to T
1
(or the melting point of cathode material if it is lower than T
1
). From
the Steenbecks minimal voltage principle point of view, the stable discharge
will be situated closest to point (
k
*
, T
*
) so the voltage drop for the
considered hollow cathodes with the characteristic value of T
e
will be
of the order of 2540 V. This range of the working cathode voltage drop
is with observed in experiments. When low-current regimes with the voltage
approaching the above-mentioned minimum threshold are used in industrial
arc plasma torches with the hollow cathode, unstable voltage pulsat-
ing discharges occur (switching to the low characteristic regime in Fig.4).
The analogous discharge behavior is observed when the plasma gas flow
rate decreases for specific working current because of the decrease of
concentration n. Thus, it can be proposed that for every definite minimum
gas flow rate (i.e. for definite concentration of n) there is minimum cathode
temperature T
*
below which the stationary discharge changes to unstable
form. Numerical assessment of the current conservation law shows that
the ion current and back electron current are of the same order near
the critical point (
k
*
, T
*
).
The qualitative analysis of discharge processes inside the hollow cathode
presented above is in good agreement with the available experimental
data [2,3].
It should be noted that all the assessments being made depend only
slightly on the assumptions regarding the cathode temperature and emission
energy (or even specific emission mechanism).
HEATING OF POWDER MATERIALS
To solve the heat transfer and heat balance problem for the particles
in plasma with sharply differing temperatures of electrons, ions and neutral
components, the particle is usually considered as a probe whose potential
is determined from the condition of total current being equal to zero
(floating potential). Using this approach for high-pressure plasma, the
authors of Ref. 4 derived the governing equations for the heat flux on
the surface of a particle 1 m in size taking into account the energy
loss on in evaporation of particles material and radiation exchange between
plasma and the particle. Heating of submicron particles was studied in
[5] where it was found out that both the temperature of the particle and
plasma temperature strongly influence the absorption of plasma radiation
due to the effect of the particle size. These investigations were continued
in [6] where the heat flux on the particle surface was determined separately
157
for ions, electrons and neutral molecules. In all these papers, the energy
distribution function for all plasma components was assumed to be
Maxwellian, i.e. equilibrium, although it is well known that the particles
cause perturbation in plasma associated with the recombination of ions
on the surface and with the presence of negative charge on the particle.
If the negative potential is high enough, the electron distribution function
can be considered to be equilibrium to a first approximation since a
large part of electrons are reflected from the particle. However, the
distribution function of the ions in the vicinity of the particle can greatly
differ from Maxwellian one so the calculations carried out on this basis
may lead to large errors when calculating the heat flux. The nonequilibrium
ion distribution function can be derived from a set of appropriate kinetic
and electrodynamic equations [7]. It is a very complex problem to find
out the heat flux by such sophisticated approach; it is linked with solving
the Poisson equation whose right part in integral manner on the required
potential. In this connection, it is eseential to investigate initially the
problem in a simplified approach assuming the equilibrium electron
distribution function, as carried out by many authors [46]. In the present
paper, the results of electric probe theory, which considers the plasma
distortion caused by the particles, are also used during the calculation
of the ion flux on the particle.
The formulated problem will be investigated under the following initial
assumptions: plasma is assumed to be two-component and fully ionized
(it is proved in [8] that all neutral components are forced out from the
plasma column); it is a steady state problem without any oscillations
and relaxation processes; it is the problem of a single particle in uni-
form plasma. In the steady state regime, the particle has negative charge
because of the difference between thermal velocities of electrons and
ions (the total charge current to the particle is equal to zero).
It is shown by analysis of heating of a small spherical particle that
the thermoemission current comparable with the ion one appears at some
temperature threshold. This current has a strong influence on the to-
tal particle heat flux. So the problem of heating of the particle will be
solved in two stages: 1) the total heat flux on the spherical particle will
be calculated without thermal emission current taken into account (low
temperature regime); 2) the accuracy of the results received in the first
step will be improved by taking into account the thermal emission current.
This approach allows us to distinguish sharply the effect of different
factors on the process of heating the spherical particles in the plasma
flow.
Let us estimate individual contributions to the total heat flux with
special reference to heating of a spherical particle argon plasma of a
158
hollow cathode arc discharge with electron and ion temperatures being
equal approximately to T
e
~ 510 eV and T
i
~1 eV. The ionization potential
is taken to be equal to = 15.8 eV. Under the conditions accepted here,
in analysis of the ions their thermal energy can be neglected in com-
parison with the recombination energy. In the plasma with the parameters
mentioned above, the spherical particle acquires potential
p
which is
equal to 20 V. It is therefore necessary to consider the contribution
of the kinetic energy, equal to e V, where V =
p
, acquired by the
ion in the particle field. Thus, every ion and every electron bring re-
spectively the following energy to the particle
i
= + eV,
e
=
2kT
e
. Moreover, the energy equal to the work function of the electron
is generated during absorption of the electron by the particle surface.
The same quantity of energy is also spent during the recombination of
the ion on the particle surface.
In the model being analyzed, it is considered all the ions hitting particle
are recombined and the electrons are absorbed. This feature leads to
disappearance of the ions reflected from the spherical particle and,
consequently, the Maxwellian distribution function of ions is distorted
and a large difference appears. The high negative potential of the probe
particle which is raised up to 20 V causes a large part of the elec-
trons to reflect from it. Therefore, as accepted in the probe theory [9],
the energy distribution function of electrons will be assumed to be
Maxwellian. In such circumstances, only those electrons whose energy
is high enough to overcome the potential barrier will hit the surface of
the spherical probe particle. This key point condition allows us to calculate
the values of electron current and electron heat flux on the sphere re-
spectively:
j e n
k T
m
e
e
e
e
eV
kT
e

2
(6)
( )
q n
k T
m
k T e
e
e
e
e
eV
kT
e

+

2
2

(7)
To close this problem, it is necessary to obtain expressions identi-
cal with (6) and (7) for the ion component.
Since the ion energy distribution strongly differs from the Maxwellian
one, this problem is more complicated. There are many models for calculating
the current of charged particles attracted by the probe. A detailed re-
159
view can be found in [9]. All of these models include diffirent apriori
assumptions concerning the ion energy distribution and the potential of
the plasma area polarized by the probe. These models give similar values
of the charge particle current on the probe. Let us consider the Allen-
Bohm model [10] which is easier to understand.
It is known that the AllenBohm model is valid when R >> r
D
(R
is the particle radius, r
D
is Debyes radius). On the other hand, the condition
l
j
>> R should be fulfilled for collisionless plasma (l
j
is the mean free
path length). Consequently, the AllenBohm model is valid when
l
j
>> R >> r
D
. The analysis of [9,10] shows that this condition is ful-
filled for the 10...40 m diameter particles inside the vacuum plasma
of the hollow cathode arc. Therefore, the AllenBohm model may be
used to determine the ion current. The calculation of the ion flux to
the sphere (the probe) is carried under the following assumptions. Since
T
e
>> T
i
,

the ion temperature is neglegted; ions are in the state of rest
far from the particle; the ion movement is only radial due to the spherical
symmetry of the electric field; collisions and ionization are absent so
the total flow on the sphere does not change; plasma is quasineutral.
These suppositions allow us to receive the following expressions for ion
current and ion heat flux on the sphere:
j e n
k T
M
i
e

1
2
(8)
( ) q n
k T
M
e V
i
e

+
1
2
(9)
where M is the ion mass of the plasma gas.
In the expressions (6), (7), (9), the potential of the sphere is still
an unknown function. It may be calculated from the condition of to-
tal current on the sphere being equal to zero, so equating the expres-
sions (6) and (8), one can solve them in respect of the potential
V
k T
e
M
m
e

_
,

0
1
2
2
ln .
(10)
The total heat flux on a cold sphere of the probe particle will be
160
q q q n
k T
M
k T
M
m
k T
i e
e
e
e
e
+

+

_
,
1
2
1
2
2
2
ln
(11)
The two first terms in the parentheses of (11) represent the contri-
bution to the ion heat flux, the third one 2kT
e
is the electron com-
ponent of the total heat flux. Regardless of the relatively low electron
contribution to the total heat flux (in argon plasma, for example, with
T
e
~ 5 eV and the cold particle regime its part is approximately 25%)
the latter one depends strongly namely on its temperature. This is linked,
firstly, with the fact that electron temperature j
i
~ kT
e
in the Allen-
Bome model has a strong effect on ion current and, secondly, that the
electron temperature also affects the ion energy acquired by the ion in
the particle field. Such current and potential behavior was examined
in details, for example, in [1]. Thus, analysis of expression (11) shows
that the properties of the plasma gas (, M) and the temperature of electrons
T
e
are the main parameters which can be used to control the total heat
flux on the particle.
Expression (11) was derived on the basis of the assumption that electro-
dynamic and thermal processes which take place during sphere heat-
ing are not linked together. This kind of process separation is possi-
ble only when the thermoemission current from the sphere can be ne-
glected comparing to the ion current. This approach to examining the
heat processes is suitable for the treatment of metallic powders with
low melting points. On the other hand, when treating the high-melting
metals such approach will be wrong and it is impossible to neglect the
thermoemission current.
The presence or absence of the thermoemossion current from the
sphere changes the total heat flux through the value of the floating
potential. If thermoemission current is absent (cold sphere), the floating
potential is calculated from expression (10); if thermoemission current
is present (hot sphere) the value
p
obtained by solving the current
conservation equation on the sphere surface:
j j j
i em e
+ 0
(12)
where j
i
, j
em
, j
e
are the current densities of ion, thermoemission and
electrons, respectively.
In order to calculate the total heat flux on the sphere in the pres-
ence of thermoemission current, let us consider that all the electrons
emitted by the hot sphere are absorbed by the plasma. This suggestion
161
allows us to rewrite the expression (12) as follows:
1
2 2
0
2

+

e n
k T
M
AT e e n
k T
m
e
e
e
kT
e
e
eV
kT
e
(13)
where T is the temperature of the particle, A is the thermoemission
coefficient. When writing expression (13), it is assumed that the
thermoemission current is never switched off (potential distribution in
the close vicinity of particles surface is supposed to be monotonic).
The reasons which can cause the violation of this condition were analyzed
in detail in [12].
The relationship for the sphere particle potential with termoemission
current on its surface being taken into consideration is obtained out from
(13):
p
e e
e
kT
e
e
k T
e
m
M
A T e
e n
k T
m
_
,
ln .
2
2
2

(14)
Writing now the heat flux expression for the hot particle, disre-
garding the energy being taken away by thermoemission electrons (e
e
~
2kT ~ 0.3 eV), gives:
q n
k T
M
k T
M
m
k T
e
e
e
e

+

1
2
1
2
2
2
ln
+

_
,
+

1
]
1
1
1
1

k T
A T e
e n
k T
M
A T e
e n
k T
M
k T
e
e
kT
e
e
kT
e
e
ln 1 2 4
2 2

(15)
where the first three terms (upper line of a formula) coincide with (11).
The negative sign of the fourth term in the formula is linked with the
decrease of the particle potential and, consequently, with the decrease
of the ion kinetic energy acquired in the electric field of the particle.
This small logarithmic decrease of the heat flux is compensated by the
162
linear increase of the number of back electrons in the plasma colliding
with the sphere. The latter fact is explained the decrease of the potential
of the particle with increasing thermoemission current. With the accuracy
to the terms of the first order with respecto j
em
/j
i
, the expression (15)
may be rewritten in the form which clearly illustrates the influence of
thermoemission current:
q q AT e k T
e
kT
e
+
0
2

where q
0
is the heat flux value according to (11).
For clearer representation of the conditions under which the effect
of the thermoemission current on the floating potential must be taken
into account, let us rewrite (9) as follows:
p
e i
e
em
e
k T
e
j
j
j
j
_
,
ln ,
0 0
where j e n k T m
e e e
0
2 = / p is the electron current density on the particle
with zero potential. Transforming the last expression assuming that the
particles floating potential under the condition of zero thermoemission
current is determined only by the relationship between the ion and electron
currents:

p
e i
e
e em
i
kT
e
j
j
kT
e
j
j
+ +
_
,
+ ln ln ,
0 0
1
where
( )
0
0
k T e j j
e i e
/ ln /
is the particles potential without
thermoemission; =k T
e
/e ln (1 + j
em
/j
i
) is the correction which takes
into account the thermoemission current influence on the potential. Let
us estimate the relationship between the thermoemission current density
and ion current density for different materials during their heating up
to melting points in a low pressure plasma flow with the parameters
determined by experiments [8]: n

= 10
21
m
3
, T
e
= 4 eV (6.2410
19
J);
the plasma gas is argon. The calculated data presented in Table 1 show
that the thermoemission current should be taken into consideration if
refractory materials are treated.
163
Fig.5. Calculated distribution of heat flux and floating potential on tantalum particle
vs. its temperature.
Figure 5 shows the values of the floating potential and density of
the heat flux versus the temperature calculated for heating of tantalum
powder under the experimental parameters mentioned above. The strong
heat flux increasing on the particle in the vicinity of the melting point
causes the given influence on treatment of refractory metals. The reason
for the increas of the heat flux is the decrease of the floating poten-
tial on the particle and consequently, increase of the energy being brought
on it by the electrons.
The above-mentioned amplified heating effect of tantalum particles
in the plasma flow is observed in modelling experiments and is utilised
in industrial equipment.
While calculating the heat flux according to (15), one has to remember
that if the decrease of particles potential is too large, the appropriate
conditions used when deriving this formula are violated. The next point
that also must be taken into account is that not all thermoemission electrons
leave the particle during the large drop of the floating potential as supposed
Material , eV
10
4
,
/m
2

m
, K
j
e
(T
m
), A/ m
Tin 4.38 60 505 2.9810
33
Titanium 3.95 60 1998 260.6
Beryllium 3.92 60 1560 3.16
.
10
1

Zirconium 4.12 330 2133 2768.1
Niobium 3.96 57 2688
1.5410
5

Molybdenum 4.20 51 2923
2.4910
5

Tantalum 4.20 55 3223
1.5410
6

Tungsten 4.50 75 3689
7.2610
6

Table 1 Relationship of thermoemission and ion current densities on particles of
different materials at melting points
164
when deriving (15). Moreover, the cloud of emitted electrons around
the particle may greatly distort not only the field around but also cause
large changes in the distribution functions of ions and electrons which
were used in calculating the corresponding currents and heat fluxes. But
if the floating potential deviation is small enough, formula (15) depicts
the heat flux change with sufficient accuracy. The assessment of heating
of the finely dispersed tantalum powder, treated in the hollow cathode
arc discharge plasma, has shown that according to the heat flux level
determined from formula (15), the particles are heated to its melting
point 0.15 m from the end of the cathode. This relates to the largest
powder fraction (60 m) and the maximal velocity of the particles
(~ 50 m/s). Calculations were based on the experimental data obtained
in examination of the velocity of the particles in a low-pressure, low-
temperature plasma flow [14]. The particles with the fraction smaller
than 60 mm (2040 m) and minimum velocity (~ 20 m/s) begin to melt
0.020.04 m from cathodes end. More detailed investigation at high
thermoemission currents should be based on the exact solutions of kinetic
equations [7]. Such approach may not only clarify the boundaries of
the applicability of simple calculation equations but should also help
to reveal all the features of particle heat transfer in the plasma flow
under more intense thermal regimes.
References
1. Chen F. Plasma diagnostics . Ed. by R.Hadlstone and S.Leonard (in
Russia).- Mir, Moscow, 1967.
2. Cherednichenko V.S., Galkin S.G., Kosinov V.A. Arc plasma flow generation
(in Russia): Institute of Thermophysics, Novosibirsk, 1987, P.306-322.
3. Cherednichenko V.S., et al. Thermophysics and Aeromechanics, 1994,
Vol.1, No. 4.
4. Uglov A.A., Lokhov U.N., Gnezdovetz A.G. Dokl. Ak. Nauk SSSR, 1979,
V.224, N 2.
5. N.N. Rykalin, A.A.Uglov, U.N.Lokhov, A.G.Ghezdovetz (in Russian) Dokl.
Ak. Nauk SSSR, 1979, V.248, N6.
6. N.N. Rykalin, A.A.Uglov, U.N.Lokhov, A.G.Ghezdovetz (in Russian) High
temperature Physics, 1981, V.19, N3.
7. Cherdnichenko V.S., Zhikharskij A.B. Nauchn. Vestnik NGTU, Novosibirsk,
1995, No.1.
8. Cherednichenko V.S., Kosinov V.A. Izv. SO AN SSSR, Ser. Tekhn.
Nauk., 1980, N 13, V.3.
9. Kozlov O.V. Electric probe in a plasma (in Russia).- Atomizdat, Moscow,
1969.
10. Allen J.E., Boud R.L., Reynolds P. The Proc. of the Phys. Society. Section
B.- 1957, Vol.70, Pt.3, N 447B.
11. Alekseev B.V., Kotelnikov V.A. Probe method of plasma diagnostics
(in Russian), Energoatomizdat, Moscow, 1988.
12. Cherednichenko V.S., Zhikharskij A.B., Judin V.,S. XI Low tempera-
165
ture plasma arc dischargers, Novosibirsk, 1989, Part II.
13. Kotelnikov V.A. Inzh.- Fiz. Zh., 1984, V.47, N4.
14. Zirjanov S.A. Electric Processing Processes and Systems, Novosibirsk
State Technical University, 1995.
166
Experimental Investigation of Electric,
Energetic and Optical Characteristics of
the Induction Transformer-Type Discharge
I.M. Ulanov, S.N. Soldatov
Institute of Thermophysics, Siberian Branch of Russian Academy of Sciences, 630090,
Novosibirsk, Russia
INTRODUCTION
Investigations of the induction transformer-type discharge opens up new
opportunities for the design of new transformer-type plasma torches,
new types of lasers, powerful light sources in visible and ultraviolet
ranges of spectra. The absence of any electrodes and a long service life
enables the application of plasma torches in continuous large-scale
technologies. The discharge chamber of a plasma torch can be used not
only for gas heating but as a plasma chemical reactor also. The large
length of the closed induction discharge (about 34 meters) provides
prolonged contact between gaseous reactants. Thus, the yield of the product
can be magnified.
The use of transparent discharge chamber made from quartz opens
up new opportunities for creation of absolutely new light sources.
And, at last, the closed discharges of the transformer type may be
the base for the development of new plasma furnaces with low pressure,
large diameter and high useful volume; the temperature in heating is
easily variable. These furnaces can find numerous applications in mi-
croelectronics.
Unfortunately, the electrodeless induction discharges at frequencies
5020000 Hz with a magnetic core have been studied insufficiently. The
possibility of developing transformer-type plasma torches was proven
in [14].
The initiator of studies of induction discharge at frequencies of 60-
9600 Hz (with the aim to develop transformer-type plasma torches) was
the American scientist H. Eckert. In [1], he estimated the current fre-
quency and plasma diameter which correspond to the discharge expire
167
Experimental Investigation of Electric, Energetic and Optical Characteristics
during the current transition of the zero point. The author calculated
also the magnetic circuit cross-section which provides the highest strength
of the vortex electric field of discharge and performed experiments in
a chamber with a frequency of 9600 Hz. In [2], the same author ob-
tained the transformer-type discharge in argon at a frequency of 60 Hz.
But the discharge died out on the way to atmospheric pressure due to
some reasons; unstable discharge at a pressure of up to 4060 kPa, stagnant
gas, small diameter of the magnetic circuit. The experimental results
on low-frequency discharge in transformer-type plasma torches are presented
in [3]. Here the electric and spectral characteristic of discharge in argon
(sometimes in air) were studied for low pressure. And again, with the
same reason, they failed to obtain the discharge at the atmospheric pressure.
At last, the electrodless induction stable discharge of transformer type
was obtained in [4] for atmospheric pressure under the power 150 kW.
It able to exist as long as possible.
2. EXPERIMENTAL UNIT
During the design of experimental setup, the following problems must
be solved. The first one is the manufacturing of the. The second problem
was the design of plasma chamber, units for gas inlet and plasma outlet.
The development of the transformer for the stable discharge of transformer
type (with low energy losses in the magnetic core) was the main aim
Fig.1. The principal schema of experimental set-up under investigation.
Gas inlet
Ar, air,
He
to vacuum
pump
to vacuum
gauge
168
of the construction of this experimental unit.
As is well-known, the induction voltage on the secondary one turn
of the transformer (but the plasma turn is single always) is defined by
the formula:
U = 4.44 B S f, (1)
where B is the induction in the magnetic core (T), S is the cross-section
of the magnetic core (m
2
), and f is the current frequency (Hz).
This voltage must be higher than the arcing voltage. Since for dif-
ferent gases a quite different voltage is required for maintaining the discharge
under specific current and specific thermophysical parameters of the gas,
the level of U, calculated by formula (1), will also be different. For
example, the required voltage for air is 4 times higher than for argon
(under the same conditions). Thus, varying B for consatnt f, we can sustain
the induction discharge in different gases. However, with the increase
of B, the losses in the magnetic core increases also it is unfavorable
from the energetic point of view.
According to [5], the law of the losses on B is the power law with
a degree of 1.82.1. The losses, connected with the increment of the
current frequency, have a power law 1.11.5.
Thus, the choice of the material for the magnetic core is determined
by minimal losses and stable magnetic characteristics. Sometimes it is
more profitable to use higher frequecies than a high magnetic field induction.
The schematic diagram of the experimental set-up is shown in Fig.1.
Fig.2.
P, kV
U, V
169
The magnetic core 1 was made from twelve separate sections. Each section
had a primary six-turn winding 4. The use of a segmented magnetic
core allowed us, using the parallel connection of the sections, to pro-
duce up to 2000 V on the secondary turn at no-load, and provide up
to 1200 V with a load on the plasma winding (with the power source
voltage of 800 V).
The plasma generators primary winding was powered by a PPChV-
250-10 rotary frequency machine with a current frequency of 10 kHz.
Its power is 250 kW, and the maximum voltage is 800 V.
To manufacture the magnetic core, we used the steel of grade 3425
with a sheet thickness of 0.08 mm. The total area of the magnetic circuit
was 760 cm
2
and it weighed 500 kg. The dynamic characteristics of trans-
formers from elecrotechnical steel number 3425 and 1521 are depicted
in Fig. 2 (at high frequencies). It is obvious from the figure that the
cold-produced finished steel 3245 has a lower ohmic loss and a bet-
ter magnetic curve when the hot-produced steel 1521.
Water-cooled discharge chamber 2 consisted of electrically insulated
sections 3 with 10-cm inner diameter. The chamber was fitted with two
optical windows to observe the discharge stability and measurements
of the plasma column radiation flux. Electrodes 6 were tungsten rods
2 mm in diameter, intended for igniting the glow discharge and facilitating
breakdown.
We injected gas into the plasma chamber through vortex unit 5, thereby
insuring the stabilization of the discharge column under atmospheric pressure.
The plasma exited through the rear side of the chamber through a heat
exchanger and a vacuum pump when operating at low pressures or directly
into the atmosphere when operating at the atmosphereic pressure.
The following electric characteristics of discharge were measured:
I, the discharge current, using the Rogovskys coil, and U, the voltage,
using an auxilary turn of insulated wire laid on the chamber perimeter.
The shape of the current in the plasma was monitored using an oscillograph.
3. EXPERIMENTAL RESULTS
The investigation of the volt-ampere characteristics of induction discharge
demonstrated that the electric field strength in these discharges is less
than the strength for constant current discharges under the same conditions
(i.e., the diameter of the discharge chamber, discharge current, and gas
flowrate). This is, possibly, connected with the inductive type of the
discharge field and with certain plasma temperature nonequilibrium. The
electric field strength vs. pressure is shown in Fig. 3 for a discharge
in air. The same relationship is shown in Fig. 4 for argon for different
current levels.
170
Fig.3.
Fig.4.
It is obvious from these figures that the electric field strength in-
creases with increasing pressurr. At the atmospheric pressure and a current
of 85 A, this field strength is 4.5 V/m (in air). And for the current
of 100 A, in argon, it equals 1.1 V/m. A sharp decrease of electric
field strength in induction transformer-type discharge is observed at the
pressure less than 13 Pa. The field intensity in air or argon diminishes
10 times.
The discharge temperature, total voltage and emission yield as a function
of the pressure are shown in Fig. 5 for a closed induction transformer-
type discharge in argon. The temperature was measured with a tung-
stenrhenium thermocouple. There were no corrections for the own emission
of the thermocouple. As obvious from the Figure, a drastic increase of
the discharge voltage, temperature and emission is observed for the pressure
above 16.5 kPa. This point corresponds to the discharge contraction and
development of large-scale oscillations of the discharge column. The use
E, V/cm
U, V
Argon
P, Pa
E, V/cm
U, V
P, Pa
G = (11.5) g/s
171
Fig.5.
Fig.6.
of vortex stabilization allows a discharge to be generated under higher
pressures up to atmospheric. Without vortex stabilization the discharge
goes out at a pressure above 1520 kPa. The relationships for the discharge
voltage, gas temperature at the discharge axis and the emission percentage
as a function of current are plotted in Fig. 6 (carbon dioxide medium).
The volt-ampere characteristic for discharge in argon and air is shown
in Fig. 7. The temperature at the discharge chamber axis as a function
of current (argon and air) is plotted in Fig. 8.
The analysis of power losses vs. frequency revealed that best fre-
quency range is 2030 kHz. Further increase of frequency causes higher
power losses.
The measurement of the visible and low-ultraviolet emission yield
was performed with the optical power gauge IMO-2N. It has a spec-
P, Pa
I, A
172
tral sensitive region from 0.33 mm up to 10.6 mm. But for registra-
tion in ultraviolet spectra we used FMP-39 device with a region of 0.16-
0.6 mm. The visible spectra were cut by light filters BG23 and YB10,
and ultraviolet spectra were selected by an UF1 filter. The emission yield
vs. consumed electric power is shown in Fig. 9 for different gases.
Mercury additions into argon enhance the emission yield. When mercury
is added, the yield is 5 times higher for the ultraviolet range and 14
times higher for the visible range of spectra. For a consumed electric
power of 11.2 kW, the light yield both in visible and ultraviolet ranges
was 14%. That is, it was about 11 times higher than for a gas discharge
in pure argon.
The light yield for the described setup for a discharge in a mercury-
argon medium is about 140170 lm/W. The light flux was measured
by an U116 lux-meter. For comparison, we may note that the light yield
Fig.8.
Fig.7.
E, V/cm
40 Pa
7 Pa
400 Pa
33 Pa
air argon
173
in modern mercury luminescent lamps is up to 8590 lm/W, this pa-
rameter for white-light lamps (40 W) is 110120 lm/W. The light yield
for mercury lamps with high and ultrahigh pressure is 4555 lm/W [6].
Similar investigations were performed with antimony, sulphur, se-
lenium and cadmium. The most interesting results were obtained in ex-
periments with antimony in argon. Here we observed the lowest elec-
tric field strength (0.24 V/cm) and the highest emission yields in vis-
ible and ultraviolet regions of spectra. The magnification in the ultraviolet
region is 3 times, and in visible region 8 times (in comparison with pure
argon).
Specific light blue emission was observed during experiments with
sulphur in argon; this testifies the excitation of strong lines of sulphur
in the range 0.450.55 mm.
CONCLUSIONS
1. The investigation of electric and power characteristics of discharge
on the developed setup indicate that the induction transformer-type discharge
will be used successfully for the construction of high-power transformer-
type plasma torches.
2. Powerful stable induction discharge in air and argon was obtained
for the first time. It was achieved under the atmospheric pressure and
can be used for development of plasma chemical reactors, which uti-
lize this type of discharge.
3. The temperature characteristics of induction discharge can be useful
for the elaboration of plasma furnaces operating in aggressive media
because the temperature may be regulated over a wide range by adjusting
the discharge current.
Fig.9.
argon (P = 26.6 Pa)
carbon dioxide (P = 8.0 Pa)
nitrogen (P = 6.7 Pa)
helium (P = 13.3 Pa)
(measurement range 0.334.0 m)
W, kV
174
4. The investigation of the optical characteristic of discharge demonstrated
that new light sources can be created on the basis of this kind of dis-
charge.
References
1. Eckhert, H.U., Plasma generation using gas inductive heating with low-
frequency current, AIAA J., 1971, Vol. 9, P. 1452.
2. Eckert, H.U, IEEE Trans. Plasma Sci., 1974, Vol. 2, P. 308.
3. Goldfarb, V.M et al., High Temperatures, 1979, Vol. 17, P. 703.
4. Kogan, V.A. and Ulanov, I.M., The investigation of the possibility of
developing transformer-type plasma generators, High Temperatures,
Vol. 31, No. 1, 1993, P. 95-110.
5. Rusin, Yu. S., Glikman, I.Ya., and Gorsky, A.N., Electronmagnetic Elements
of Radioelectronic Apparatus. Reference Book, Moscow, Radio i Svyaz,
1991, P. 225. (In Russian).
6. Reference Book on Light Technique, ed. By Yu. Aizenberg, Moscow,
Energoatomizdat, 1983, P. 472. (In Russian).
175
Mathematical Modelling of Transformer Discharge
Mathematical Modelling of Transformer
Discharge
E.B. Kulumbaev and V.M. Levelkin
Kyrgyz-Russian Slavic University, Bishkek, 720000, Kyrgyz Republic
The theoretical investigations of the low-frequency discharge of a
transformer type without gas flow were carried out. The application
of the kinless approach with the electrodynamic description of the
discharge and the possibility of a steady approach during calculation
of the argon plasma characteristics in the frequency range 110 kHz
and at pressures of 10-100 kPa was shown. The existence of two regimes
of discharge burning was established: stable one with a high temperature
and unstable one with a low temperature. The influence of frequency
and voltage on the plasma coil, type and pressure of plasma gas on
the discharge characteristics was investigated. The results of modelling
are in agreement with experiments.
Designations:
T - temperature;
n - particle concentration;
p - pressure
multivalued scalar potential;
E, B vector of the electrical field intensity;
E.B vector of the magnetic field intensity;
magnetic flow;
= 2f a circular frequency;
S heat flow potential;
N
e
= k
1
n
e
n
a
k
r
n
2
e
n
i
rate of electron generation;
B = 3km
e
n
e
v
e
/m
a
- the factor of the elastic interaction between electrons
and heavy particles;
k
1
, k
r
ionisation and recombination factors;
m particle mass;
176
, , D
a
- coefficient of conductance, thermal conductivity and ambipolar
diffusion;
- emissivity
U
1
the ionisation potential;
T
R
= 300 K the wall temperature;
0
the magnetic constant;
k
B
the Boltzmanns constant;
indexes e, I, a electronic, ionic and atomic components.
1. INTRODUCTION
The inductive excitation of the low-frequency discharge using a magnetic
circuit (the transformer-type plasma torch) is the perspective way
of the electrodless plasma production with a power about 1 MW [1].
The transformer-type plasma torch represents a transformer, in which
the primary winding is fed from the generator at frequencies 110
kHz and electrodless plasma forming in the toroidal chamber forms
the secondary turn of the electrical circuit. The transformer-type plasma
torches combine the advantages of electrodless discharge (the maintenance
and generation of the spectropure plasma at the practically unlimited
resource of work), t he advant ages of t he si mpl e power suppl y
(commercially produced electromachines or transformer generators)
by comparison with sources of RF and microwave electrodless discharges
and provide the plasma production in large-volume discharge chambers
[1]. The feasibility of creation of the transformer-type plasma torch
is shown in [2-4], in which the idea of the transformer discharge
in the static mode at a low pressure of 4060 kPa and in the absence
of the stabilizing discharge gas blow-through caused in quenching
is developed. In research in [5], the problem of the transformer-type
plasma torch was solved and the steady burning of the discharge with
the gas blow-down up to atmospheric pressure has been achieved.
The purpose of this work was to develop the mathematical model
of the transformer-type discharge and to carry out research of the plasma
characteristics, frequency, the voltage on the plasma coil, and the type
and pressure of the gas.
2. MODEL. ESTIMATIONS AND APPROXIMATIONS
2.1 The Model
Lets consider the low-frequency discharge of the transformer type
in the static mode without gas blow-through. The physical model includes
processes of the Joules heat yield of closed currents induced by the
variable magnetic flux and the heat removal by means of heat conduction
and emission. The mathematical description of these processes is carried
177
out in the framework of the two-temperature approximation of plasma
[6] and the quasistationary balance equation for the energy of the
electronic gas:

t
n k T T j E B T T U n
e B e e e e e
3
2
1
F
H
G
I
K
J
= +

b g ( ) ,
(1)
the balance equation for the energy of heavy particles:
t
n n k T T B T T
a i B e
3
2
+
L
N
M
O
Q
P
= + b g b g b g,
(2)
the continuity equation for the electronic gas:
n
t
D n n
e
A e
e = +
b g ,
(3)
Maxwells equation:
E dl
d
dt
B j
m i

z
= + = b g
, .
0
(4)
The system (14) is supplemented by Daltons law, the quasineutrality
condition, the Ohms law and expressions for factors of the two-
temperature argon plasma [6].
2.2 The skinless approximation of the electrodynamic model
The presence of the current in the transformer discharge plasma is
connected to the rotational electrical field E, produced by the variable
magnetic flow
m
=
m0
exp(it) in the magnetic circuit and by the
variable magnetic flow
i
B d S

z
induced by plasma currents. By

neglecting the primary circuit active resistance for the given frequency,
the amplitude
m0
does not depend on the plasma current, whose
demagnetising action is compensated by the growth of the primary
circuit current. For the evaluation of the contribution of the component
of the rotational field E
i
created by the variable magnetic field induced
178
by the plasma current turn present in the form of the circular current
I with perimeter L
k
. The maximum induction of the magnetic field
will be in the ring centre B
0
=
0
I/L
k
. The upper estimate gives the
expression for the magnetic flow through the ring
i
=
0
IL
k
/4. Using
data [5] as ordinary values for the transformer discharge: I = 100
300
A, L
k
>> 2 m, = 210
4
Hz,
m0
=100
1000 V, one can obtain

the ratio of the electrical field components E
i
/E >> (
i
/L
k
)/(
m0
/
L
k
)=,
0
IL
k
/(4
m0
) >> 10
2
. At T ~ 10 hK the skin layer 2
0
/ ( )
~15 cm appreciably exceeds the discharge chamber radius R
k
~ 4 cm.
Therefore, the skinless approximation is applicable for the description
of the transformer discharge electrodynamics and the electrical field
circulation differs from zero only for the account of the magnetic
flow varied in time in the magnetic circuit. The applicability of the
assumption had been corroborated by the results of the study. So one
can suppose that in the discharge plasma occupying the two-coupling
area, the equations
=

E E 0,
are true.
Taking into account the Ohms law j E

= , from the electrical current
continuity equation we may write for the multiple-valued potential
= 0. (5a)
The boundary conditions on the surface the toroidal chamber
partition reduced the two-coupling area given by

+
= =
0 0 0
0, ,
(5b)
The potential jump
0
= i
m0
is determined by the speed of the
magnetic flux change in the magnetic circuit, so the circulation is
equal to
E dl i i t
m

z
=
0
exp( ). (5c)
It should be noted that in the framework of the above assumptions
179
the electrodynamical model of the transformer discharge is similar to
the model of the electrical arc with electrodes coincided with the sur-
face , where one side is the cathode, and the other one is the anode.
Consequently, the term the transformer arc can be used.
2.3 The stationary approximation of the transformator discharge
In the low-frequency discharge, the characteristic time of the establishment
of the electrons average energy or temperature t
e
is considerably longer
that the period of electric field variation 2/. In this field with the
amplitude of approximately A
e
[7],
A
= e
2
E
2
0
v
e
/2m
e
(
2
+v
2
e
) is the
set up speed of the energy by an electron in the variable electrical
field with the amplitude E
0
averaged over time; =1/
e
; v
e
the frequency
of electron collisions with heavy particles; the average relative
part of the energy transmitted from electrons to heavy particles by
collisions. According to the experiment data [5], for the transformer
arc in argon at atmospheric pressure E
0
~ 1 V/cm. Assuming T
e
0
kK, 10
4
, v
e
10
11
Hz, we receive the ratio of the electron energy
amplitude to its average value
A
e
/(3k
B
T
e
/2) 10
3
, i.e. the assumption
that the field influence on plasma parameters is small and the plasma
condition may be considered as close to equilibrium is valid.The product
of the specific energy yield n
e
A
into the discharge chamber volume
2R
k
L
k
= 910
3
m
3
gives the power dissipated in the discharge of
110 kW that agrees with the experimental value of ~100 kW. The
thermal sluggishness of the equilibrium plasma is determined through
the temperature conduction c with a specific time R
2
k
/c
10
1
10
2
s it is less than period of the electrical field variation. Then the average
values of plasma parameters approximately can be considered as
stationary.
Table 1 The calculated in frameworks of non-stationary (averages) and stationary
models the maximum values of discharge parameters
Model Non-stationary Stationary
U
2
,
V
F,
kHz
T
e
,

T ,
K
n
e
,
10
21
m
3

I ,
kA/m
T
e
,
K
T,
K
n
e
,
10
21
m
3
I,
kA/m
P = 100 kPa
10 8228 8223 2.92 5.29 8227 8222 2.92 5.28 200
1 8240 8235 2.97 5.41 8239 8234 2.97 5.41
10 9081 9081 6.74 15.48 9081 9081 6.74 15.48 300
1 9102 9101 6.99 16.33 9098 9098 7.00 16.30
P = 7 kPa
10 9740 9726 2.72 10.00 9665 9646 2.57 9.78 150
1 9873 9866 3.13 10.92 9920 9907 3.02 10.85
10 10620 10619 5.69 16.77 10609 10607 5.62 16.66 200
1 10729 10726 6.53 18.89 10800 10799 6.59 18.89

180
Fig.1 The steady-state temporary dependences of deviations of maximum values
of the electron temperature T
e
= T
e
(t) <T
e
>, 10
3
K (curve 1), the temperatures of
heavy particles T, 10
3
K (curve 2), the electron concentrations n
e
/<n
e
> (curve
3), the electrical current per unit length I=
Edr
r
r
1
2
z in the discharge I/<I> (curve
4) as functions of their average per the period of the electrical field intensity variation
(curve 5) at p = 7 kPa; f = 10 (a); 1 (b) kHz; U
2
= 150 V.
2.4 Calculated valuations
For corroboration of the applicability of the considered approximations
and strict numerical valuations, the one-dimensional discharge in the
double coaxial tube with internal radius r
i
and external radius r
2
with
cooled walls has been considered. It was assumed that the magnetic
flux with the amplitude
m0
was placed in the internal cavity of the
double tube.
The system of cylindrical symmetric equations (14) with the boundary
conditions R = r
1
; r
2
: T
e
/
r
= 0; n
e
/
r
= 0; T = T
R
was solved by the
numerical method on the basis of the implicit differential four-point procedure
having stability and monotonic properties.
The calculation of the argon transformer discharge characteristics
for the experimental [5] data has been carried out: p =7; 100 kPa;
f = 1; 10 kHz; r
1
= 0.33 m; r
2
= 0.40

m; and secondary voltage in plasma
U
2
=
m0
= 150; 200; 300 V.
Figure 1 shows the temporary dependences of deviations of the discharge
parameters from average values (Tab. 1). One can see that at p =
7 kPa the amplitude of discharge characteristics oscillations accounts
for less than 10% of their average values. At the atmospheric pressure
these deviations are even smaller, the electron temperature is quasistationary
and other parameters are practically frozen. The amplitude of the variation
of the discharge parameters increases with decrease of the frequency:
181
Fig.2 Electron temperature, heavy particle temperature, the electrical field intensity
and the magnetic field induction as functions of the distance in the argon plasma
at p = 100 kPa, f = 10 kHz, U
2
= 200 V.
at p = 100 kPa, f = 0.1 kHz, relative deviations reach ~40%. The average
maximum values of the parameters slightly differ from the values calculated
in the framework of the stationary model (see Tab.1).
The distributions of the discharge stationary characteristics at p =
100 kPa, f = 10 kHz, U
2
= 200 V are presented in Fig.2. Since the electric
field intensity is ~1/r, the scanning effect is not insignificant and it may
be neglected. Separation of the heavy particle temperature from the electron
temperature in the wall layers is appreciable. Temperature equilibrium
exists in the central zone.
3. ANALYTICAL RESEARCH
3.1 The equilibrium model
To determine the qualitative features of the transformer type discharge
we considered the one-dimensional discharge whose characteristics
may be described in the framework of the axially symmetric stationary
energy balance in the equilibrium plasma, and solutions of Maxwells
equations (4) by the skinless approximation for the amplitudes of the
azimuthal electrical field intensity and the axial magnetic field induction
may be written as:
=
= =
F
H
G
I
K
J
z
1 1
2
2
2
0
2
r
d
dr
r
dS
dr
i
r
Edr
E
E B
m
r
r

,
, .
(6)
182
The boundary conditions express the thermal equilibrium of the wall
layer gas with cold walls and the absence of the magnetic field out-
side the double tube (as for the infinitely long solenoid):
r = r
1
:S = 0; r = r
2
: S = 0; B = 0.
The formula for the amplitude of the azimuthal electric field intensity
follows also from the electrodynamic model (5) which potentially is the
linear function of the azimuthal angle in the cylindrical coordinates system.
By introducing a new variable x = ln(r/r
1
)/ln(r
2
/r
1
), the energy balance
equation and boundary conditions may be transformed to the form:
=
= = = =
F
H
G
I
K
J
F
H
G
I
K
J
d S
dx
r
r
r
r
r
u S S
x S x S
x
2
2
2
1
2
1
2
2
1
2
2
0 0 1 0
( ) ( )
: , : ,
ln
(7)
Here u =
m
r
r
0 2
1
4
ln .
3.2 The solution without reference to the emission
Ignoring emission ((S) = 0), the task (7) allows the analytical solution
at real transfer properties of plasma. The function S(x) has a maximum
in the area 0 < x <1 and satisfies the conditions x = x
m
: dS/dx =
0, S = S
m
.
With x
m
and S
m
being the subject to determination, the first integral
of equation (7) takes the form of the conservation law
1
2
2
2 2
2 2
dS
dx
u V S u V S
m
F
H
G
I
K
J
+ =

( ) ( ),
where 2u
2
V
(S)=2u
2

( ' ) ' S dS
S
0
z
is the potential energy being the
monotonic growing function of S. The second integral of equation
(7) may be written as
=
z
2u x x
m
m S
S
dS
V S V S
m
( )
'
( ) ( ' )
.

183
Here i s real i zed i n areas 0 < x < x
m
(dS/dx>0) and x
m
<x<1
(dS / dx < 0), respectively. The boundary conditions (7) locate an
extremum coordinate x
m
= 0.5 and the corresponding maximum value
S
m
can be calculated from the relation:
u
dS
V S V S
m
S
m
=
z

( ) ( )
.
0
The integral by additive allocation of the integral function singularity
in the top integration limit can be transformed to the form
u dS
S
S
V S V S S S S
m
m
m m m
S
m
= +
L
N
M
M
O
Q
P
P
z
2
1 1
0

( )
( ) ( ) ( )( )
.
(8)
Finally, the solution can be written as
dS
V S V S
ux
u x
m
S
x
x
'
( ) ( ' )
,
( ),
.
.

R
S
T
z
=

<
0
2
2 1
0 0 5
0 5 1
Using the found solution, the electrical characteristics of the discharge
per unit length can be determined: current, power, active resistance
I Edr i V S Q E rdr V S R
Q
I
V S
m m m a
m
m r
r
r
rl
= = = = = =
z z

4 2
0 5
4
2
2
2
1
2
1
( ), ( ),
.
( )
.
To analyse this solution, it is convenient to choose S
m
as an exter-
nal parameter with the values of V
(S
m
) and u at given dependence (S).
The parameter u, presenting the complex consisting of the values ,
mo
, r
1
, r
2
is the criterion of discharge similarity.
The characteristics of the transformer discharge plasma as functions
of the similarity parameter being proportional to secondary voltage
mo
were

calculated. The argon and air factors at the atmospheric pressure
were taken from the references in [6]. Calculations show (Fig.3) that
the contribution of the second term in the right part of (8) may be neglected
and it may be approximately considered that u 2 S S
m m
/ ( ) . This
184
dependence has the form typical for equilibrium discharges (see [7]).
There is the minimum threshold value of the secondary voltage at which
the maintenance of the equilibrium plasma of the transformer discharge
is possible. At the secondary voltage greater than the threshold, transformer
discharge plasma can form. At the secondary voltage greater than the
threshold voltage, two stationary modes of discharge burning can be
realized: with a low temperature (bottom branch) and with a high temperature
(top branch). The conditions on the bottom branch are unstable rela-
tively temperature fluctuations and any small temperature increase causes
the real voltage to increase to the value necessary for the maintenance
of the mode with a new temperature. As a result, the plasma heating
begins and continues to reach the steady-state conditions on the top branch.
The temperature maximum forms in the vectorially average radius rr
1 2
(x
m
= 0.5) and is shifted towards the radial flow of electromagnetic energy.
The discharge currentvoltage characteristic of the category (Fig.4) has
drooping (unstable mode) and growing (steady mode) branches.
3.3 The solution with emission taken into account
The fast increase of temperature as a result of an increase of the
secondary voltage in the steady mode is explained by energy losses
in emission, which, in the framework of the considered model, are
proportional to the area of the cross ring section. If emission is taken
into account, the right part of (7) becomes explicit so that its analytical
solution is not possible. However, this dependence is weaker and it
Fig. 4 The current-voltage characteristic of the argon and air discharge at atmospheric
pressure.
Fig. 3 The dependence of the maximum temperature of the argon and air discharges
at atmospheric pressure as function of the similarity parameter.
185
may be neglected in comparison with the dependence on S. The
assumption has been corroborated by results of the calculations. Denoting
a r
r
rl
r
r
r
r
r
r
r
dx r r
r
r
x
x
2
1
2 2 2 2
1
2
1
2 2 2
1
2
1
0
1
2
2
2
1
2 2
1
1
2 1
1
2
1
4
=
F
H
G
I
K
J
=
F
H
G
I
K
J
=
z
ln ln ln
d i (9)
It is convenient to express values necessary for calculation of r
1
and
r
2
through the perimeter and by analogy with (8), we receive
u S a S S
m m m
+ ( ( )) / ( ). 4
2

(10)
When emission is taken into account, the parameter u is no longer
the similarity criterion.
It is convenient to express values necessary for calculation r
1
and
r
2
through the perimeter L
k
and the radius R
k
of the discharge cham-
ber r
1
= L
k
/(2) = R
k
; r
2
= L
k
/(2) + R
k
.
The calculation results for argon at the atmospheric pressure using
the data from [5] at L
k
= 230 cm, R
k
= 3.65 cm, are shown in Figure
5. The account of the emission does not change qualitatively the de-
pendence of the maximum temperature on secondary voltage, increase
its threshold value and essentially decrease temperatures values in the
Fig. 5 The dependence of the maximum temperature of argon discharge at atmospheric
pressure versus secondary voltage (curve 1 without considering emission; 2 with
emission taken into account; 3 the numerical solution).
186
steady mode. When L
k

2R
k
and assuming that 4S
m
<<a
2
(S
m
) from
formulas (9,10) one can receive the approximate expression for the voltage
at the plasma turn as a function of maximum discharge temperature
L
k
2 ( ) / ( ) S S
m m
which corresponds to the local balance of Joules
energy and emission at the discharge chamber perimeter. The values calculated
from this formula agree with the top branch of curve 2 of Fig. 5.
3.4 The numerical solution
The results are shown in Fig. 5. The finite-difference technique permits
a numerical solution only in the steady mode (curve 3 in Fig.5). The
voltage threshold value at the argon plasma turn at the atmospheric
pressure is approximately equal to ~190 V. At smaller values of
m0
the simple solution T(r) = T
R
= 300 K is independent of the temperature
distribution defined in the first approximation. Radial distributions
of temperature, electric field intensity and magnetic field induction
at
m0
= 200V are similar to the distributions shown in Fig.2. It
should be noted that the equilibrium model slightly overestimates the
plasma temperature. The calculated values of the electric field intensity
E >> 0.9 V/cm, and the current I = 2R
k
Edr
r
r
1
2
320
z
A axially uniform
at the discharge chamber length for with heat conduction and emission
taken into account are about equal. At
m0
= 300 V, dominant losses
are the emission losses, which represent 80% of dissipated energy.
4. NUMERICAL RESEARCH
4.1 The three-dimensional model
The description of the transformer discharge is carried out in the
framework of the three-dimensional model. The transformer arc
characteristics are considered in the special orthogonal coordinates
system in which the flat curve r
0
(l) coincident with the discharge
chamber axis is chosen as the axis, here l is the length of the axis
line measured from any fixed point l = 0 (Fig.6).
The radius-vector of any point of the space is determined as (11).
Here
v
is the unit vector of the main normal to axis r
0
(l), is the
unit vector of the binormal; the cross coordinates x and y are distances
from the axis measured along the normal and binormal, respectively.
The third coordinate is the length of the axial line l.
For the determination of the curvilinear space metrics lets take the
increment of a radius-vector dr from the equation (11) by the account
of Frene-Serres formulas

187
Fig. 6 The diagram of the discharge chamber with the fragment of the cross section.
dr vdx dy kx dl
H H
H H
= + + ( ) , 1
Here k(l) is the curvature of the coordinate axis
H
r
0
(l),
H
(l)=
H
r
0
/
l is a unit vector of the tangent line. Hence the square of the length
element is equal
dr dr dx dy kx dl
H H
= + +
2 2 2 2
1 ( ) .
In the case of the circular section of the discharge chamber, it is expedient
to pass from coordinates x,y to polar coordinates , in the plane
( v , ) (Fig.6):
dr dr d d k dl
H H
= + +
2 2 2 2
1 ( cos ) .
Lames factors, corresponding to this metrics Lames factor are equal
H

= 1, h

= r, h
1
= 1k cos and permit

equations (1-3, 5a) to be
recorded for the stationary case in the chosen curvilinear coordinates
system in the following generalized form
1
0
h h h
h h
h
h h
h l
h h
h l
i
l l
l
S

F
H
G
I
K
J
+
F
H
G
I
K
J
+
F
H
G
I
K
J
L
N
M
M
O
Q
P
P
+ = .
(12)
188
Here
, S
are non-linear factors whose form depends on meaning

of variable (Table 2) and are established from correspondence to
equations (13, 5a).
For the solution of the system of differential elliptical equations (12)
the boundary conditions for dependent variables are set for the entire
calculated area limit which, due to the task symmetry, is limited by co-
ordinates 0 < l < L, 0 < < R
k
, 0 < < 2, where L = L
1
+ pR
0
/2+L
2
.
In input and output sections, symmetry conditions for = {T
e
, T, n
e
}
and the given potential values give
/
;
, , ); l
L L 1 0 1 0 0 1
0
= = =
= = =
At the lateral surface of the discharge chamber, the conditions of the
electron normal interface, the given values of heavy particle tempera-
ture, the absence of the normal component of the electric field inten-
Table 2 Non-linear factors of the generalized equation (12)
Fig. 7 The influence of current (a) and pressure (b) on electron and heavy particles
temperature distributions in toroidal argon discharge.

S

Equation
T
e
e
0.5 E
2
B(T
e
T)-
-U
1
n
e
(1)
T
B(T
e
T) (2)
n
e
D
A
n
e
(3)

0 (5a)

189
Fig. 8 The currentvoltage characteristic of argon at p = 7 kPa (xxx are experimental
data [5]).
Fig. 9 The dependence of the voltage on the argon plasma turn versus the pressure
at I = 300 A (xxx are experimental data [5]).
sity may be written as
=
= =
= = T n T T
e e R R R
k K
, ; . : /

l q
0
4.2 The toroidal discharge
For the toroidal transformer discharge L
1
= L
2
= 0; k(l)=R
1
0
= 2/
L
k
= const; = {T
e
, T, n
e
} do not depend on the coordinate l and
equation (12) for the potential has the simple solution = iU
2
(l/
L
k
). The electric field intensity is equal
E E E
l
k
iU
L k

= = =
0 0
2
1
, , ,
( cos )
The geometrical ratio r = R
0
(1 k cos) agrees with (6).
The calculation of two-dimensional characteristics of toroidal argon
discharge was carried out for f = 10 kHz; L
k
= 230 cm; R
k
= 3.65 cm
(experimental datae from [5]).
The calculation results are shown in Figs. 79. The influence of current
and pressure on the transformer discharge characteristics is similar to
patterns established for the electrical arc [6].
As the calculations have shown, the plasma in equilibrium decreases
by increasing a current and a pressure (Fig.7). At pressures less than
10 kPa, the separation of electron temperature from heavy particle tem-
190
perature is appreciable along the entire cross section of the discharge
chamber. At p = 90 kPa, a practically equilibrium discharge core forms
and equilibrium is observed in the wall layers. The electron tempera-
ture near walls remains at the level ~4
5 kK. By increasing the pressure

the tendency to contraction of the transformer arc is becomes more distinctive.
Comparison between the calculated and experimental discharge current-
voltage values at p = 7 kPa is shown in Fig. 8. Satisfactory quanti-
tative agreement is found between theory and the experiments. The maximum
deviations between the theoretical and the experimental values are at
U
2
< 50 V and U
2
> 200 V. The calculated dependence of the second-
ary voltage versus pressure at I = 300 A (Fig.9) only qualitatively repeats
the experimental one. The quantitative deviations may be explained by
the possible influence of the gas-dynamic characteristics which are not
taken into account in the considered model. At a pressure of p > 40
kPa the burning of the transformer arc is possible only under the condition
of the gas-dynamic stabilization of the vortex [5]. From the energy viewpoint,
it is obvious that the injecting of cold gas into the discharge chamber
causes cooling of the arc. Therefore, for the given current the required
secondary voltage (Fig.9) is higher than that required in the static case.
References
1 S.V. Dresvin, et al., RF and microwave plasmatrons, Nauka, Novosibrisk
(1992).
2 H.U. Eckert, AIAA J. 9, No.8, 1456 (1971).
3 N.N. Rikalin, et al., Fiz. Khim. Obrab. Mater., No.4, 1555 (1977).
4 V.M. Goldfarb, et al., Termofiz. Vysokikh Temper., 17, No.4, 698 (1979).
5 V.A. Kogan and I.M. Ulanov, Termofiz. Vysokikh Temper., 31, No.1,
105 (1993).
6 V.S. Engel'sht, et al., The theory of the electric arc column, Nauka,
Novosibirsk (1990).
7 Yu.P. Raizer, The physics of gas discharge, Nauka, Moscow (1987).
191
Integrated Method of Research of Processes in ........
Integrated Method of Research of
Processes in Thermal Plasma Generators
O.Y.Novikov and V.F.Putko
Samara State Technical University, 443098, Samara, Galaktionovskaya 141
The 3-rd International Working Meetings was used as a basis for
preparing this article as the review of the results of work of the
Samara group, representing one of branches of scientific school of
M.F. Zhukov, the Academician of the Russian Academy of Science,
published mainly in published in a series of monographs under the
edition of Prof Zhukov [14].
The strong effect of the power supply system of plasma units on
processes proceeding in them is now recognized. These systems are so
complex that the description and research of them are possible only by
an integrated method, realized in the problems of electrodynamics as
the so-called circuit or structural method. This situation is not only
retained but is more aggravated with further improvement of plasma
devices. The tendency to use of a uniform method of research of
plasma units has resulted in the development of integrated methods of
the analysis of processes in generators of low-temperature plasma,
similar to methods of the analysis of electric circuits [5].
It enables the use of an effective method of space of states, coor-
dinates of which are variables, describing the reserve of energy in vari-
ous power tanks of system.
Thus, an electrical arc in a plasma torch is considered as a
memristive multipole, realized as two structures: a nonlinear resistive
(dissipative) or a linear system with memory. The dynamic properties
of the linear system were investigated by spectral methods [9]. This
analysis has shown that the inertia of an arc is caused by the reserve
of energy in its thermal field. The order of the system depends on the
heterogeneity of thermal fields and in the first approach can be char-
acterized by the time constants the arc column, near-electrode zones,
and the halo. The infrafrequency part of the spectrum is caused by the
thermal inertia of devices, stabilizing its situation in space. The solu-
tion of an inverse problem has enabled, for example, the control of the
192
process of formation of a welding seam by the spectral characteristics
of the arc.
The analysis of the frequency response of the electric arc suggests
that, in many cases, the dynamic properties of an arc can be charac-
terized of a time constant and, hence, are represented by the dynamic
part of the first order, i.e. the decrease of the order of the linear dy-
namic system is correct.
The method of the space of states is especially convenient for the
qualitative research of the modes of operations of electric arc units and
has been used for the analysis of stability in the first linear approxi-
mation (in small), and also for determination of the area of attrac-
tion of steady equilibrum modes (in large).
The main result of the research of stability in small is the deter-
mination of essential small parameters, neglect of which results in
qualitatively incorrect results. In the Maxwel equations, included in the
system of magnetohydrodynamics, one such essential parameter is in-
ductance (in the integrated description). Thus, the neglect of the re-
serve of energy in magnetic fields, even if they are very small, can
result in faulty results. The significance of this parameter is determined
by the peculiarities of the dynamic properties of the arc.
The significance of small parameters limits the possibilities of the
reduction of the order of the differential equations, describing proc-
esses in the electric arc unit. In particular, this limits the use of the
widely known Kaufman criterion, the observance of which is a neces-
sary and sufficient condition of stability only for systems of the first
order.
The analysis of stability in large was carried out by the direct
Lyapunov method. This method finds the boundary of the region of
attraction of the burning mode of the arc, i.e. the reserve of stability
of the regime with respect to deviations which can be both casual and
regular, appearing, for example, at start-up of multiarc units. Research
of the influence of parameters of the electric arc unit on the position
of the boundary of the area of attraction, i.e. on the reserve of stability
was carried out.
To estimate the error of definition of the boundary, the latter was
determined using two directions from the area of burning, and from
the area of extinguishing the arc.
The complexity of the use of the parametrical Lyapunov vectors-
functions has forced us to select their structure in the elementary kind
of the square-law form of the variable space of states.
The direct Lyapunov method was used for the analysis of the
stability of the AC arc. Thus, the trajectory in the space of states was
193
composed from the trajectory of slow movement, characteristic of the
basic part of the half-cycle of burning and the trajectory of fast move-
ment during the transition of current through zero level. The general-
ized law of switching, taking into account the inertia of the thermal
field of the arc, was used for bonding.
The characteristic feature of the space of states of electric arc units,
determined by the traditions of generalization of the static voltage-
current characteristics of the arc, is that of one of the basic modes -
extinguishing mode fits in the region in which the static characteris-
tics are not determined.
This has required the introduction of nonlinear transformation of the
coordinates of the space of states, in particular for two-pole represen-
tation of the arc coordinates current and conductivity were used in-
stead of the traditional ones, i.e. voltage and current.
The efficiency of methods of research in small and in large is
illustrated on an example of multiarc units [10].
In research of the movement of an arc alongside with the use of the
traditional integrated description of movement of an arc, the results
were used to develop an integrated method of the analysis of movement
of the thermal condition in relation to the environment, the so-called
slip of the arc. This description is based on the apparatus of generali-
zation of forces and flows which by virtue of the linearity of the
Onzager ratio near the thermodynamic balance enables a linear model
of slip near to equilibrum modes of movement of the electrical arc to
be constructed. It results in the use of the uniform apparatus of the
analysis of the power condition and movement of the arc in the gen-
eralized space of states.
The qualitative research of the generalized space of states was also
carried out by the methods of research of stability in small and in
large.
Alongside the traditional ways of the joint solution of problems of
power and spatial stability based, as a rule, on the parametric consid-
erations of movement when analysing the power stability, the influence
of the power parameters of the electrical arc on its spatial stabilization
was investigated.
The latter is connected with the speed of slip when analysing the
movement of the arc. As shown in early works [11], the speed of slip
is determined by the thermal time constant of the arc, i.e. the dynamic
properties of the arc, being an element of the electrical circuit.Thus,
the power stability and spatial stabilization of the electrical arc are
interconnected. The dependence of both these parameters on the same
parameters of the electrical circuit is not new. So, if the dependence
194
of the power stability on parameters such as the EMF of a source, the
active and reactive impedance of a circuit of a power supply directly
follows from the analysis of the phase portrait of the space of states,
the dependence of the spatial stability on these parameters is indirect
through the current of the arc, determined by the same parameters.
The account of slip results in the formation of new connections, for
example, the dependence of the spatial stabilization on the ability of
the circuit of the power supply to accumulate energy in electrical
fields, i.e. on capacitor elements. The presence of capacitances, their
influence and the arrangement in the arc circuit determines the origi-
nal availability of power tanks and the possible speed of receving of
energy by the arc. This defines the dynamic characteristics of slip of
the arc, i.e. its spatial stabilization.
The advantages of the integrated method are especially evident in
the research of complex electric arc units. They are plasma units with
the use of dynamic magnetic fields for controlling the electrical arc, for
example, running or rotating [4,12].
The analysis of behaviour of the electrical arc in a rotating mag-
netic field has shown deep analogy of the proceeding processes to the
processes occurring in electrical machines, but also their essential dis-
tinctions.
When imposing the rotating magnetic field on the arc running be-
tween the electrodes located on the axis of this field, the arc begins to
rotate about the axis under action of magnetic viscosity forces. The
Lorenz forces displace the axis of the conducting channel from the axis
of the magnetic field, forcing it to make precession movement. The
conditional balance of the forces, acting on the arc, can be determined
from the resulting action, i.e. movement of the arc axis. The flow of
a pulse, transmitted to the environment, results in centrifugal move-
ment with double overrotation, with respect to both the arc and the
axis of the magnetic field. The slip of the arc is directed to the cen-
tre. Equilibrium precession movement of the arc is established as a
result.
At this stage, the processes are close to the process of asynchronous
start-up of a synchronous motor. The difference is that the radius of
precession of the arc varies depending on the arc current, the induc-
tion of the magnetic field and the speed of its rotation. Thus, this de-
vice is similar to the synchronous engine with the rotor winding geom-
etry changing depending on the listed parameters. The listed
dependences of the radius of a precession are also determined by the
spatial structure of the rotating magnetic field, i.e. number of its poles.
Analysis of the stability of this movement shows that it is
195
asimptotically steady as a whole and has no bifurcational parameters,
i.e. there is no phenomenon of the so-called overturn, characteristic for
synchronous engines.
Overturn of synchronous engines, i.e. the infringement of synchro-
nous rotation of its rotor, arises at when the moment of resistance ex-
ceeds the maximum rotating moment. In this case, the moment of
resistance is caused mainly by the aerodynamic resistance to movement
of the arc and overturn could come when the critical speed of rotation
of the the magnetic field is exceeded. It does not come via the asymp-
totic reduction of the radius of precession with increase of the
frequency of rotation and stabilization of the arc on the axis of the ro-
tating magnetic field.
Experience has shown that at precession movement of the arc, it ro-
tates about the axis with the speed greatly exceeding the speed of ro-
tation of the magnetic field. This is explained by that the arc in the ro-
tating magnetic field represents a rare example of the dynamic
stabilization of plasma in the non-uniform magnetic field.
In an arc, stabilized in a non-uniform transverse magnetic field
there is the rotating moment of Lorenz forces which causes rotation of
the arc around its axis. The speed of this rotation is determined by arc
current, the intensity and heterogeneity of the magnetic field and does
not depend on the speed of its rotation.
The heterogeneity of the magnetic field depends on the design of
inductor creating this field. The magnetic field of a two-pole inductor
is practically homogeneous. In multipole inductors, the intensity of the
magnetic field incerase in proportionally to the distance from its axis.
The arc in rotating magnetic fields created by both two- and multipole
inductors was experimentally investigated. It has enabled the influence
of the Magnuse effect on the process of stabilization of the arc by the
rotating magnetic field.
In summary, it is possible to make a few remarks concerning the so-
called retrograde motion of the arc, i.e. movement of the arc in the
direction opposite to the action of Lorenz forces.
The retrograde motion of the arc was found out when examining the
movement of an arc in a transverse magnetic field at reduced pressure.
The majority of researchers explain the retrograde motion of an arc by
the special mechanism of moving its electrode attachment spots.
However, the retrograde motion of the electrical arc can also be
explained by the processes taking place in the conducting channel, if
we take into account not only the movement of the arc under the ac-
tion of Lorenz forces but also its slip.
The retrograde motion can arise when the speeds of movement and
196
slip are directed in the opposite directions and speed of slip is more
than the speed of a movement. The considered example of stabilization
of the arc by the rotating magnetic field is exactly the case when ra-
dial parts of slip and movement are directed in the different directions.
A suitable example of the retrograde motion of the arc is its move-
ment in the elementary construction of horizontal arrangement of elec-
trodes. The electrical arc between horizontal electrodes is curved up-
wards by the convective flow and Lorenz forces and slides downwards,
in the zone of increased heating. This slip can be treated as the action
of the force, proportional to the gradient of the strength of the elec-
trical field.
The retrograde motion of the arc in the transverse magnetic field at
reduced pressure can be explained by its slip in the direction of the
gradient of magnetic induction.
The magnetic field starts to influence strongly the transport prop-
erties of plasma with decrease of pressure when the free path length
length becomes commensurable with the Larmor radius.
The movement of the cathode spots of an mercury arc of low pres-
sure in the direction of increasing induction of the magnetic field has
been specified for the first time by Kesaev [12].
References
1. M.F. Zhukov (editor), Theory of electric arc in the conditions of forced
heat exchange, Nauka, Novosibirsk (1977),115-211.
2. M.F. Zhukov and A.S. Koroteev (editors), Theory of the thermal electric
arc plasma, Nauka, Novosibirsk (1987), 5-78.
3. B.N. Devyatov and O.Y. Novikov (editors), Mathematical methods of
examining he dynamics and problems of control of low-temperature
plasma, Nauka, Novosibirsk (1991).
4. Thermal Pl asma and New Materi al s Technol ogy. Vol ume1 /Edi tors:
Solonenko O.P., Zhukov M.F., Cambridge Inter. Publishing (UK), 1994.
5. O.Y. Novikov and V.F. Putko, Theoretical fundamentals of electrical
engineering at the interface of sciences, in: IV International Symposium
Theoretische Elektrotechnik, Ilmenau,Germany (1987).
6. M.F. Zhukov, et al., Electric arc gas heaters, Nauka, Moscow (1973).
7. M.F. Zhukov, et al ., Appl i ed dynami cs of thermal pl asma, Nauka,
Novosibirsk (1975).-
8. M.F. Zhukov, Experi mental exami nati on of pl asma torches, Nauka,
Novosibirsk (1977).
9. U.P. Kamaev, et al, in: Stability of the electric arc, Nauka, Novosi-
birsk (1973).
10. M.F. Zhukov and O.Y. Novikov (editors), Multiarc systems, Nauka,
Novosibirsk (1975).
11. O.Y. Novikov, Modernisation of high-voltage switches and conductors
to them, Kuibyshev (1962).
12. I.G. Kesaev, Cathode processes in the mercury arc and problems of
stability of this arc, Gosenergoizdat, Leningrad (1961); Trudy VEI, No.
67.
197
Modelling of Electric Arc Plasma
A. Zhainakov, R. Urusov and A. Valeeva
Republic Centre of New Information Technologies, Ministry of Social Education and
Science of Kyrghyzstan, RCNIT, pr. Mira, 66, Bishkek, 720044, Kyrghyzstan
The wide used of electric arc equipment and plasma technologies
in science and technology demand the solution of some problems,
connect ed wi t h i ncreasi ng servi ce l i fe of t hi s equi pment s
exploitation and optimisation of working regime [1-3].
In the given operation, the outcome is the reduced solution of
the three-dimensional equations of energy and potential of the elec-
tric field for the stream of argon of in electric plasma at atmos-
pheric pressure. The field of speeds is considered given. Calcula-
tion will be carried out in the Cartesian coordinate system Y, X, Z.
The equation of energy has the following form:
div div grad r l s y = + - c T T
p
U E
@ E
= B
2
(1)
Where , c
p
, , , are t he coeffi ci ent s of densi t y, t hermal
capaci t y for const ant pressure, t hermal conduct i on, el ect ri c
conduction, radiation as functions of temperature. T, U (Ux, Uy,
Uz), E (Ex, Ey, Ez) are the temperature, vectors of speeds and
electric field strength, respectively. The vector of electric field
strength E has t hree component s t hat compl i cat e cal cul at i ons;
therefore, it is more convenient to enter the scalar variable, i.e.
the potential of electrical field , connected with E by the relation
E = grad f. Further, using Ohms law j =E = grad and
the condition of conservation of current div j = 0, we shall note the
equation for the potential:
198
div grad ; s f = B = 0 (2)
The solution of the differential equations (1-2) will be carried
out by the digitization method [4]. The essence of the method con-
sists in the following. The derivatives of the unknown quantities
are substituted by differences, i. e. the continuous values of the
function are substituted by discrete values in the final number of
points, called as nodes, the designed area. The designed area is
divided into the final number of not intersecting check sizes in
such a manner that each nodal point is one check size. This is fol-
lowed by triple integration of the differential equation with respect
to the check size. We obtain the discrete analog of differential
equations which include the values of the unknown function in sev-
eral nodes. Thus, the differential equation of the second order is
reduced to the system of the linear algebraic equations, which then
are solved by the iteration method. Integrating a set of equations
(1-2) with respect to the check size, we shall receive discrete ana-
logs:
the equation for temperature
a T a T a T a T a T a T a T S.
a D A P F a D A P F
a D A P F a D A P F
a D A P | F a D A P F
p p w w E E N N S S T T B B
w w w W E e e e
N n n n S s s s
T t t t B b b b
= + + + + + +
= + = + -
= + - = +
= + - = +
where (| | ) | ,0 ; | | | ,0| ;
| | | ,0| ; | | | ,0| ;
| | ,0| ; | | | ,0| ;
> C
> C > C
> C > C
The operator [|A,B|] defines the highest values of A and B.
S x y z x y z = + + - s f s f s f d / d d / d d / d ; = B = B = B
2 2 2
Y D D D
a a a a a a a
p E W N S B T
= + + + + +
199
F c U y z F c U y z F c V x z
F c U x y F c W y x F c W y x
D y z x D y z x
D x z y
D x z y D x y z
D
e p
e
w p
w
n p
s p
s
t p
b
b p
b
e e
e
w w
w
n n
n
s s
s
t t
t
= = =
= = =
= =
=
= =
r r r
r r r
l d l d
l d
l d l d
@ E @ E @ E
@ E @ E @ E
> C = B > C = B
> C
= B
> C = B > C = B
D D D D D D
D D D D D D
D D D D
D D
D D D D
; ; ;
; ; ;
/ ; / ;
/ ;
/ ; / ;
n
b b
b
x y z = l d D D > C = B / ;
P (the Peclet number) is defined as P
e
= F
e
/D
e
, etc.
To approximate the convective addends, the circuit with was
used by the degree law, and the A function (| P | ) has the follow-
ing form:
A P P | | |0, 1 0.1| | | ;
5
= B = B = -
the equation for the potential
where / ; / ;
/ ;
/ ; / ; / ;
a y z x a y z x
a x z y
a x z y a y x z a y x z
a a a a a a a
w s E e
N n
S s T t B b
P E W N S B T;
= =
=
= = =
= + + + + +
s d s d
s d
s d s d s d
D D D D
D D
D D D D D D
Fig.1 The circuit of the designed area.
S
J
X
Y
0
S0
Z
J
200
The obtained discrete analogs represent linear algebraic equa-
tions and also link the value of the unknown function to the cen-
tral point from it by the values in adjacent nodes. The numeri-
cal solution of the data in the equations will be carried out by the
iteration method. The Zaidel- Gauss method is used widely. How-
ever, the method has an essential disadvantage slow convergence.
In t he met hod proposed i n [4], an i t erat i on met hod, ensuri ng
faster convergence than the Zaidel-Gauss method is used.
It is especially important for the tasks requiring extensive cal-
culations. Figure 1 shows the circuit of the designed area, having
the form of a rectangular parallelepiped.
To define the boundary conditions, it is necessary that the
electrical current is applied normal to surfaces S0, S; the tempera-
ture on these surfaces is defined from the condition dT/ dx = 0; on
the side surface, temperature is equal to the surface temperature
of envi ronment , and t he current i s absent . The coeffi ci ent s of
transfer of argon are taken from practice [2]. To improve the sta-
bility of the iterative process, we used the relaxation method. The
criterion for completing calculations is the condition
max| |/ max| |
1
f f f
N N 1 N
- <
- -
e
Here f is the calculated function, N is the number of the iteration,
e = 10
5
.
Fig.2 Temperature T, T
0
and current of an electrical current.( experimental
data [5]).
201
Figure 2 shows the calculations of of temperature T carried out
for argon plasma at the atmospheric pressure, current 75 A in a
square channel, sizes X = 30 mm; Y = 26.5 mm; Z = 26.5 mm. The
components of speed Uy and Uz we equal to zero.
As the used method of solution follows from the outcome of
solving differential equations, it ensures a steady, converging itera-
tive process. The results of calculations are in satisfactory agree-
ment with the experimental data [5].
One of most important problems of electric arc welding in ad-
vanced engineering and metallurgy is the optimization of welding
conditions. It is necessary to carry out comprehensive research of
thermodynamic processes, running in welding pool, for producing
high-quality welds. The fluid flow of molten metal in the weld pool
in arc welds exerts a significant effect on the quality of the weld
seam and the depth of weld penetration in the workpiece. For this
purpose, numerical and analytical models of heat transfer and liq-
uid metal flow in welds in arc welding are required. This problems
was studied by Rozenthal [6] in an analytical model and numerical
models were developed in [7-11]. In this work, we propose a nu-
merical model of the weld pool, based on the simultaneous solution
of a complete system of magneto-hydrodynamic equations. It is
assumed that the plane is the surface of the welded product. The
radius of the weld pool is 24 mm under normal welding condi-
tions. It is also assumed that the welding arc is stationary and in
weld pool metal is melted and incompressible, processes are sta-
tionary, the liquid metal flow is laminar, crystallization processes
are not consi dered. The heat fl ow fl owi ng from t he arc t o t he
product and released in it due to Joule heating is distributed by
conductive and convective flows. The metal motion is caused by
electromagnetic and buoyancy forces, pressure and surface tension
gradients, viscous stress. The boundaries of the weld pool (liquid
-solid metal) are determined from the solution of the equations.
Fig.3 Schematic sketch of the weld pool.
202
The model is based on a complete system of magneto-hydrody-
namic equations
rot / grad * grad
/ / div div div( )
grad div / grad
V V V j B
V P V V
V h jE C h
p
r r
r h ht r
r l
- = -
+ + + =
= - +
2
2
2 2 3 2 0
? D
= B
@ E
Y
(3)
rotE = 0; j = rotH; divB = 0; divD = 0
The system is closed by the state equation and Ohms law:
D=ee
0
E; B=m
0
H; j=sE
Here , , , , C
p
are the thermodynamic functions and metal
transfer coefficients: density, heat capacity, viscosity, thermal and
electrical conductivity, respectively. The solution of the system (3)
is carried out in the cylindrical coordinate system in view of sym-
metry, and the boundary conditions are specified on the entire pe-
rimeter of the computation range. For the energy equation, we have
the heat flow distribution on the pool surface or temperature dis-
t ri but i on. The t emperat ure gradi ent on t he open surface of t he
welding pool T/z is determined by the heat flow from the elec-
tric arc to the welded product and heat losses into the ambient me-
dium. The heat losses through the open surface are negligible be-
cause of high melting point and high thermal conductivity of the
welded material. For the motion equation, the conditions are speci-
fied: the axial velocity component u is equal to zero. Shear stress
Fig.4 Streamline and temperature fields m V z jB / , = = 0 0 > C
.
203
Fig.5 Streamline and temperature fields m v z T r t jB / / * / , = - = 0
> C
.
Fig.6 Streamline and temperature fields.
(v/z) arises as the surface tension gradient on the welding
pool surface due to the significant radial temperature gradient as
depends on temperature. For Maxwells equation, we define the
current density distribution on the surface. The conditions of ad-
hering to the velocity are specified on the side and lower surfaces
of the welding pool, temperature is supposed to be equal to the
solid metal temperature. The symmetry condition is satisfied on the
axis. Thus, the boundary conditions are written as follows ( -
the heat boundary radius, b - boundary of metal solid phase):
204
z r T z Q r
r R T T j j r
z r R u v z r
R z z
= =
= =
= = = -
0 0
0 0 0
, : / ( )
: , ( )
, : , / /
a l
a
m g
(4)
0 0 0 0 0 0
0 0 0
0 0 0 0
2
= = = = =
= = = = =
= = = = =
z b r T u r v j r j
z b r R T T u v j I R
z b r R T T u v j z
z z
R
R z
, : / / , , / ,
, : , , , /
, : , , , /

p

? D
This gives the closed mathematical model, determining heat and
hydrodynamic flows in the welding pool. The numerical solution of
equation (3) with the boundary conditions (4) is made in variables
vorticity ()stream function (), enthalpy (h) and electric cur-
rent function () by the ZeidelGauss finite difference method of
according to a five-point scheme. Sheet aluminum was welded with
a stationary electric arc, I =150 A. The radius and length of the
welding pool do not exceed 4 mm and 6 mm, respectively. Ther-
modynamic functions, transport coefficients and the aluminium sur-
face t ensi on coeffi ci ent were t aken from [12]. It i s seen from
Fig.4, where streamline fields () are presented, that under the
action of the buoyancy force the melt metal rises up to the surface
from the central heated zone of the welding pool and colder liquid
metal sinks at the welding pool boundary. On the surface of the
weld workpiece, the melt flow is directed from the axis radially
outwards and the maximum velocity reaches ~ 20 cm/s. This is
characteristic for the flow regime; due to surface tension, the cold
metal draws off hot metal from the center, the maximum flow ve-
locity reaches ~ 1m/s at the welding pool boundary. As is seen
from Fig.5-6, liquid metal streamlines () change direction as a
result of the electromagnetic force action, the flow velocity on the
welding pool surface is directed to the axis of symmetry, the maxi-
mum value reaches 20 cm/s, 50cm/s at the axis of symmetry. In the
process of arc welding, all this three forces act simultaneously and
a complicated (Fig.6) pattern of flow is realized: two vortices form
in opposite directions. One of them is induced by the surface force,
the other one by electromagnetic ones. The averaged velocity at
the surface is higher (6 cm/s) than the velocity in the vortex, in-
duced by the electromagnetic force (50 cm/s). The temperature
fields show that the action of the electromagnetic force increases
weld metal penetration. The results are in satisfactory agreement
with the calculations [911]. With decrease of welding current (100
205
A) the maximum temperature, metal flow velocity and weld pen-
etration decrease.
References
1 M.F. Zhukov,Theory of thermal el ectri c arc pl asma. Sci ence WITH,
Novosibirsk (1987).
2 A. Zhainakov, et al, Electrical arc - generator of thermal plasma, Bishkek,
Ilim (1991).
3 A. Zhainakov and T. Urusova, Numerical solution of a three-dimensional
heat conduction equation, Bulletin KGNU, Bishkek (1996).
4 S.V. Patankar, Numerical methods of problem solving of heat exchange
and dynamics of a liquid, Energoatomizdat, Moscow (1984).
5 M.K. Asanaliev, et al, Measurement of coefficient airdynamic resist-
ance of a sphere in argon to plasma.
6 D. Rozental, Weld J.Res. Suppl. 20, 220 (1941).
7 J.A. Shercliff, J. Fluid Mech., 40, 241 (1970).
8 C. Sozou and W. Pickering, J.Fluid. Mech., 73, 641 (1970).
9 D.R. Atthey, J. Fluid. Mech., 98, 787 (1980).
10 S. Kou, Model Cast. And Weld Process, Proc. Sypm., Rindge (1982),
p.129.
11 S. Kou and D. Sun, J.Metallur. Transact. A.,16A, 203 (1985).
12 J.F. Lancaster, The Physics of welding, (1984), p.297.
206
On the Problem of Turbulent Arc
Modelling
O.I. Yasko
Heat & Mass Transfer Institute, Academy of Sciences of Republic Belarus, 15 P.
Brovky St. 220072 Minsk, Belarus
1. INTRODUCTION
Turbulence is a very intricate phenomenon. More than hundred years
have passed since Osborn Reynolds discovered his famous number, but
scientists have not managed yet to propose a method for pure theoretical
simulation of turbulent flows. A number of semiempirical methods are
developed for this purpose [1] that allow simulation of cold turbulent
flows, but high-temperature ones are much more difficult to model.
Especially difficult for the conventional methods is the simulation of
electric arc turbulence because of the nature of the processes depends
on the discharge conditions. Using empirical constants obtained in another
situation can lead to misleading results. Therefore, the simulation of
turbulent arcs is in urgent need of pure theoretical methods.
The basic hypothesis was proposed in [2] and verified in [3,4]. The
results of this verification are considered in this paper.
The hypothesis assumes the existence of pairs of charges, with the
product of their dimensions being equal to the Planck constant dimension.
There are four of them: electric and magnetic charges, impulse space,
energy time, mass moment of velocity. They are considered as real
charges possessing mutual interactions. The assumption that mass is a
charge such as electric one finds a confirmation in existing static gravitational
interaction whose law is similar to that of electrical charges. The universal
relationship is obtained in [2] for calculating the constants for differ-
ent charges static interaction. The theoretical value of the gravitational
constant has turned to be in perfect agreement with experimental magnitude.
But a moving electric charge creates a magnetic field. So the next assumption
may be made that a special field is produced by any charge during its
motion and that the Lorentz force is created due to interaction of the
207
On the Problem of Turbulent Arc Modelling
charge with the corresponding field. This suggestion complies with existence
of well known different vortices (tornadoes, cyclones in atmosphere,
turbulence etc.) since the Lorentz force bends the trajectory of a moving
charge producing a swirl. It is shown below that these assumptions
open the way for theoretical simulation of turbulent arcs.
2. GENERAL RELATIONS
The above hypothesis entails the assumption that the known laws of
the magnetic field may be applied to other ones created by moving different
charges. But knowing such a basic constant like the field permeability
is necessary for application of the laws. It can be attempted to find this
quantity from Nature Triangle and the simple model of the electron
structure likewise it was done in [2] for estimating constants of static
interactions.
For dynamic interaction, it is possible to suggest that the electron
consists of electrically charged subparticles rotating around the centre.
When the subparticles have different polarity of the charge q
i
its value
can be equal to the total one. The magnitude of subparticle mass can
be also assumed to be equal to the total electron mass m
e
. Though such
assumption involves negative mass, it complies with our main hypothesis
every charge can be positive or negative. The velocity of subparticle
rotation must be close to that of light c, so that this utmost value may
be taken for evaluation.
For the magnetic field, the above assumptions give perfect agreement
with experiment when the average length of field line L corresponds to
the characteristic size of the electron D h m c
e e
/ 2 (h is Plancks constant)
divided by 2
e
The last quantity can be derived from the Nature Tri-
angle [2]. For the magnetic field, its theoretical value is equal to
e
= 137.10953137. This
e
value leads to perfect agreement with ex-
periment for the theoretical constant of static interaction for electric
charges. For other fields,
i
can be also estimated from the Triangle.
For a vortex field,
v
= 7.058562810
9
. So we may define for the general
case
L
h
m c
i
e i
4
(1)
It is seen from Eq. (1) that L
v
for the electron is very short in comparison
with L
magn
.
The assumptions enable us to calculate separately the field induc-
tion in the electron B
i
and its strength H
i
. Their ratio gives the value
208
of the permeability constant of a field m
oi
= B
i
/H
i
. Expressions for the
field flux and induction follow from the hypothesis
i i
h q (2)
B
D
m c
hq
i
i
e
e
i

4 16
2
2 2
(3)
Field strength H
i
is determined as H
i
=I
i
/L
i
. Current I
i
can be found
from the circular rotation of the subparticle in the electron
I
q c
D
q c m
h
i
i
e
i e

2
2
(4)
The field strength is determined by the expression
H
q m c
h
i
i e i
8
2 3
2

(5)
The formula for permeability can be written as follows
i
i i
h
cq
0
2
2
(6)
For the vortex field (q
v
= m
e
), m
m
= 7.547410
8
m/kg
The energy of a field has the following form
W
B H H
i
i i i i

2 2
2
(7)
A charged particle moving in the corresponding field with veloc-
ity v
i
is affected by the Lorentz force
F v B
i i
= q
i
> C
(8)
Due to the trajectory bending by the Lorentz force, charged parti-
cles begin to rotate. The expressions for the cyclotron radius and frequency
209
take the form
r mv q B
i i i

/ (9)
f q B f
i i i m
/ 2
(10)
For the field induced by rotatable mass (q
v
=m), the induction cor-
responds to cyclotron frequency
B f
m m
2
(11)
At the low field energy, this kind of rotation can be considered as
a latent (or implicit) vortex which can change for explicit one at
the critical conditions. The appropriate assumption about such conditions
is the excess of the field energy over the translation one for particle
random movement W
f
> W
tr
= v
tr
2
/2 ( is density). Hence, the criti-
cal condition can be defined as
v H
tr i icr
2
0
2
2 2
(12)
H v
i cr tr
i
.

0
(13)
For ideal gas (P / = kT / m,
v
kT
m
tr

8
; P is pressure, T temperature,
k is the Boltzmann constant), Eq. (13) can be transformed to
H
P
i cr
i
.

8
0
(14)
For the vortex field (m
m.0
= 7.547410
8
m/kg) at atmospheric pressure,
H
cr
= 1.83710
2
0.02 kg/ms.
Equation (14) shows that for a given ideal gas the critical value of
field strength depends only on pressure, and the possibility of explicit
(free) vortex arising increases with the pressure fall. Equation (14)
is useful for estimating the critical conditions at local points.
Relations for the mass flow in a cylindrical canal (pipe) will be con-
sidered below. In accordance with similarity of different fields, mass-
flow density u (where u is the velocity of the flow) corresponds to
210
current density j for the electromagnetic field, and the total gas-flow
rate G r u r rdr
R
2
0
( ) ( ) matches to current I. Here R, r are the pipe
radius and the distance from its axis. Analogy between mass-vortex and
electro-magnetic fields gives the expression for the vortex-field strength
H r
r
r u r rdr
f
r
( ) ( ) ( )
1
0
(15)
In accordance with the considered model, onset of the turbulence is
impossible just at the pipe axis. The critical distance depends on the
intensity of the flow. The strength of the arised turbulent vortex self-
field increases with the distance from the initial point toward the pipe
periphery until it reaches the critical value. Then follows a zone of free
turbulence. The strength of the self-turbulent field in the centre of the
vortex is given by the expression
H
r
r y r
dy
y
v v
y
y
m
1
0
( , )
(16)
Here y
0
is the radius of a vortex central bore.
But near the pipe wall the size of the vortex is restricted. The vortex
flux v y y ( ) (where is the distance from the vortex centre) becomes
dependent on the moment of the flow velocity near the wall u r y
m
( ) .
These conditions enable the flow velocity profile in the vicinity of the
wall to be defined.
3. SOME VERIFICATIONS OF THE MODEL
3.1 Critical value of Reynolds number
An important verification of a hypothesis concerning turbulence origin
is the theoretical calculation of the critical value of Reynolds number
Re
cr.
If onset of turbulence is caused by the applied vortex field, the
critical conditions can be found from Eqs. (12) and (15).

v
R
r u r rdr
tr m
R
2
0
0
2
2 2
1
1
]
1
1
.
( ) ( )
(17)
211
For the average and U values
2
0
v
RU
tr
m

.
(18)
Substituting v
tr
=2 / l, where n and l are the kinematic viscosity and
mean free path [4,5], one can obtain
Re
.
cr
m
UD
l

8
0
(19)
The quantity (
m.0
)
0.5
is the characteristic length of the latent tur-
bulent field. The critical value of the Reynolds number depends on the
ratio of this quantity to mean free path of the medium.
Now we can compare the theoretical value of with its experimental
magnitude taking as an example the gas flow in a cylindrical pipe at
atmospheric pressure and t = 0 C. The values of l and are taken
from[5,6]. For nitrogen, we obtain
( ) .
.
,

m
m
0
0 5 5
3 26 10

, l = 5.88
10
8
m and Re
cr
4430 . The theoretical value of the critical Reynolds
number is twice as high as the experimental one. The results obtained
are close to the experimental value, but the difference by a factor of
2 can mean that the
m
value is incorrect or onset of turbulence is caused
by a phenomenon other than the flow field. Moreover, the assumption
u(r) = const is crude. In a cylindrical pipe, the profile of the laminar
flow which is converted into a turbulent one has the parabolic form:
u r r ( ) 1
2
,where u r u r U r r R ( ) ( ) / , / u U
av
05 . . (R is the ra-
dius of the pipe and U is axial velocity). The relative value of the local
Reynolds number for this condition can be described as follows
Re ( )
loc
loc
r
H
UR r
r rdr r r
1
1
1
2
1
4
2
0
3
(20)
Maximum of Re
loc
takes place at
v
r
m
(2/3)
0.5
, and Re
loc m
= 0.272
1/4. Since for the parabolic profile UR = u
av
D, the critical Reynolds
number is approximatelly four times as high as its maximum local value
at the moment of turbulence origination. Therefore, the theoretical magnitude
of Re
cr
is 8 times higher than Re
loc
.
Of course, the model of the electron internal structure used for
m
212
estimation is rather uncertain yet, and successful application of
m.0
for
turbulence modelling would be very useful for its clarification. Therefore,
another cause of the disagreement is likely to be examined.
For flow inside tubes, swirls are originated not only by the free vortex
field, but also by friction in the zone adjoining the tube wall. The wall
friction exists even in the laminar flow. Hence, only the friction near
the wall is responsible for primary turbulence. This problem can be examined
considering Eq. (16). It may be adopted for the cold flow: = const,
and for given r and ideal vortex the flux is also independent of r.
Under these assumptions, Eq. (16) takes the form
H y y
v m
= - rF ln ln
0
> C
(21)
Strength H
v
is induced by a vortex in its centre with the distance
approximately y
m
from the tube wall (neglecting the thickness of the
laminar sublayer). Quantity H
v
/ corresponds to the vortex forced by
wall friction at this point: H
v
/ =2y
m
u
m
= (ln y
m
ln y
0
). Factor 2 is
used because u corresponds to the centre of the vortex. This equation
specifies the velocity profile near the tube wall u
m
(y
m
) =H
v
/2y
m
=
( /2y
m
) ln (y
m
/y
0
)= (/ 2 y
m
) ln y
m
(/2y
m
) ln y
0
.

It is known as the
experimentally obtained universal velocity profile for the turbulent flow
when variables are presented in the dimensionless form: u
+
= u/(
w
/
)
0.5
; y
+
=y
m
(
w
)
0.5
/; where = is dynamic viscosity and
w
is
the shear stress at the wall [7]:
u y a b y
m m
+ + +
+ ( ) ln
(22)
Here a = b ln y
+
0
and b =
v
/2n y
+
m
=u
+
j
. These factors are constant
because they correspond to the velocities at the wall for the flow and
the vortex, respectively u
+
j
: = b = const, u
+
0
= a = const. The constant
value of a and b factors demonstrates that
v
has a linear dependence
on y
+
m
(
v
= 2bn y
+
m
) The function (22) is linear in semi-logarithmic
coordinates.
The universal profile consists of a laminar boundary layer u
+
= y
+
and two linear parts (Fig.1).
The strength of field H
v
increases with y
m
until it attains the criti-
cal value H
cr
. The further rise H
v
> H
cr
causes the formation of free vortices
which create their own field directed oposite to the initial one. The
development of free eddies takes course until their field neutralises the
whole excess of H
v
above H
cr
. Therefore, at H
v
> H
cr
the strength of
the free vortex field is equal to H
free
= H
v
H
cr
. To make available this
213
1 10 100
0
5
10
15
20
u
+
y
+
u
+
=a
1
+b
1
lny
+
u
+
=a
2
+b
2
lny
+
u
+
=y
+
Fig.1 Universal profile of velocity for wall turbulence
equality, the free vortex has to compensate a half of H
v
. Hence
H pv l y y pv y y y
p v y pv y
cr v m v m cr free
m cr
= - - - =
= - +
+ + + + +
+ + +
F F
F F
ln ln . (ln ln ln )
. ln ln . ln
0 0
0
0 5
0 5 2 0 5
? D
? D
This expression corresponds to the second linear part of the universal
velocity profile
u a b y
m 2 2 2
+ +
+ ln
, where a a b y
cr 2 1 1
0 5 +
+
. ln and b
2
=0.5b
1
.
The term ln y
free0
+
must be equal to zero since H
free
= 0 at H
v
=H
cr
.
The field strength produced by the vortex
v
is a constant quantity
in the region
H H
v cr
>
r
because it is in equilibrium with the total field
H H
total cr
.

In such a case, the rise of field strength is attained only
at the expense of ln( / ) y y
m
+ +
0
, and at the point H H
v cr
2 , the term
ln y
m
+
is two times higher than at H
cr
. The point ln ln
.
y y
m m cr
+ +
2 is the
end of the universal velocity profile as the free vortex becomes self-
sustaining at H H
free cr
> . This conclusion, as well as equality b
2
= 0.5b
1
,
are in fair agreement with experiment. The contradiction between the
rising flux
v m
b y
+
2
2
and the constant value of its field strength can

be explained by the variability of the relative vortex permeability. In-
duction of the field is constant over the considered zone

m cr m m m
H b y
. . 0 0 2
+
. It follows from this that

m cr m
H b y

/ 2
2
(23)
The form of the universal profile (Fig.1) can be derived theoretically.
For this purpose, the boundary conditions have to be imposed in ad-
dition to the general relations obtained above. The product of dimensionless
velocity and distance is equal to the Reynolds number. Hence, for a bending
point it gives the critical value of this number
214
u y H
cr cr cr
+ +
/ 2
(24)
At the point of the turbulent flow joining with the laminar sublayer
u y b
j j
+ +

1
(since for sublayer
u y
+ +
and du
+
/dy
+
= 1 while for turbulent
flow du
+
/dy
+
=b/y
+
). Factor a
2
is the vortex velocity at the wall a
2
u
+
j
.
Comparison of theoretical and experimental parameters of the univer-
sal profile is given in Table 1.
Some disagreement between the theoretical and experimental parameters
listed in Table1 for the universal velocity profile is caused by the nonideality
of vortices, neglecting the size of the boundary laminar sublayer, and
a smooth transition between the two linear parts of the profile.
The table demonstrates that the prediction of the critical value of
the Reynolds number for forced wall turbulence conforms well to ex-
periment: u y u y
cr cr cr cr
+ + + +
= =
@ E @ E
theor exp
, , 544 420 and it corresponds also to
the relative local value of the critical Reynolds number for turbulence
caused by the flow field Re Re / .
cr.loc cr
4 550
Analysis shows that the model provides quite correct critical magnitudes
of the Reynolds number. It is also revealed that there are two ways of
turbulence origination. At low Reynolds numbers (Re = 10
3
10
4
), turbulence
is created at the wall by friction. But at Re > 10
4
, the flow field ex-
ceeds the critical value of its strength that entails onset of free vorti-
ces. Unlike forced turbulence originated near the tube wall, the free one
arises in its central part.
3.2 Electric arc discharge
As distinct from a cold flow, where turbulent vortices are formed at
the expense of the stream kinetic energy, they are originated in electric
arcs due to the consumption of thermal power. The level of thermal energy
in the arc plasma is substantially higher than kinetic one. For example,
the enthalpy of air plasma at a pressure of 1 atmosphere and T = 10
4
K is approximately an order of magnitude higher than kinetic energy
Table 1 Comparison of theoretical and experimental parameters of universal velocity
profile for wall turbulence
Parameters a
1
a
2
b
1
b
2
y
+
cr
u
+
cr
Theoretical 4.47 5.84 5.84 2.92 34 16
Experimental 3.05 5.50 5.00 2.50 30 14
215
at the speed of sound (h = 4.7810
7
J/kg; a
2
/2 = 3.5610
6
J/kg). Using
higher energy enables thermal turbulence to produce more intensive
disturbance of the flow.
Modelling turbulence in the electric arc is a much more complicated
problem than in the cold flow. The processes of energy conversion and
transfer can predominantly affect the disturbance. Momentum and heat
transfer processes caused by turbulence depend on vortex dynamics: on
vortex formation, growth, and destruction. Vortex formation is controlled
by the field critical conditions and development is promoted by the thermal
nozzle effect which converts the thermal energy of arc plasma into kinetic
one of the vortices [6]. The growth of eddies is accompanied by heating
due to energy dissipation caused by friction. Effect of the last proc-
ess is rather appreciable even in comparison with Joule dissipation, especially
in the vicinity of the vortex central bore. Due to the heatreleased by
friction, temperature and pressure rise inside the vortex but thermal nozzle
acts in the opposite direction. This process can continue until the Lorentz
forces become incapable of preventing destruction of the vortex by the
rising pressure. Then it is destroyed by explosion. The maximal velocity
of exploded particles is defined by expression v h 2 , where h is
the enthalpy of vortex overheating.
The explosion of vortices can bring the main contribution into turbulent
heat transfer in nonuniform temperature fields. However, turbulent vortices
accumulate not only kinetic but also field energy. The latter also contributes
to heat transfer. In such a case, we may try to apply an effective enthalpy
increment h
ef
. If this increment could be determined as a function of
temperature and mass flow fields, we would be able to simulate theo-
retically turbulent discharges using an iteration procedure. The direct
but still rather uncertain way is the modelling of vortex dynamics. Another
possibility is accounting for some general relationships peculiar to the
eddies in the turbulent flow. To find some of these laws, we may take
into consideration that the arising vortices generate their own field directed
against the applied one but they can suppress only such a part of this
field which exceeds the critical value H
cr
.
There are four different zones along the radius of a turbulent arc
column. The first zone spreads from the arc axis to the point where the
strength of the applied field becomes equal to the critical value. The
turbulent eddies are originated at this point and their self-field reaches
H
cr
at H
appl
= 2H
cr
.The vortices become self-sustaining at this point which
is the border between the second and third zones. In these regions, the
swirling velocity of turbulent eddies can attain a high magnitude ow-
ing to power supply with the thermal nozzle effect but it is limited by
the sonic velocity. Unlike cold flows, the electric arc is an intensive source
216
of power and it is reliable to assume this limit to be attained in the centre
of a vortex: v
0.m
= a. The constant value of total field intensity allows
us also to proceed on the assumption that the free turbulent eddies have
the constant field energy complying with critical conditions:
H
v cr v.cr v.cr m.0 cr
2
tr
2
/ / / v / . B H 2 2 2 2 H B This assumption consid-
ers that at critical conditions ln(y
max
/ y
0
)=1 that corresponds to
(y
m
/y
0
) = e. The critical frequency of vortex central part rotation
f a y a a H
cr cr cr cr
= = = / / / /
.
2 2 2
0
2 2
p pF r p
@ E
(25)
This quantity relates to partial induction of the vortex central cir-
cle
B y y a y f
cr cr cr cr cr 0 0 0
2
0
2
. . .
( ) / /
(26)
The critical frequency is independent of the mass flow rate, being
only the function of temperature and pressure. Therefore, it is connected
with the critical value of the vortex field
B v H k a H k B k f
v cr tr cr cr cr cr . .
/ / = = = = r p r p p p
2 2 2 2
0
2
2 2 2
@ E
(27)
where k = v
tr
/ a.
Because the energy of the vortex field is specified only by H
cr
:
B
v
H
v
= k
2
a
2
, relation (27) between B
v
and f is valid not only for critical
conditions. In this case, we may determine the frequency and induction
of the vortex as function of its field strength
f a H v H
v cr

2
0
2
2 2 / /
(28)

B k a H
v v
2 2
/
(29)
It is seen that in the third zone at H
v
> H
cr
the vortex frequency and
field induction decrease with the increase of field strength. On the other
hand, increasing H
v
is related with the growth of the vortex size. Therefore,
the vortices become restricted at the border of the arc column form-
ing the fourth zone where the vortices cannot follow the general rela-
tions.
The decrease of B
v
with increasing H
v
is connected with the dependence
of vortex field permeability on its strength
217

v m v v v
B H k a H
.
/ /
0
2 2 2
+
(30)
Here
v
is the relative permeability and is the vortisability for the
vortex field (similarly to magnetisability). Thus,
v
m
v cr
v
k a H
H

_
,
2 2
0
2
.
.
(31)

k a
H
v
m m v
2 2
2
0 0
1
. .
( )
(32)
The vortisability is negative at H
v
> H
cr.
Since v
2
0
/ 2 h corresponds to enthalpy pulsations caused by thermal
turbulence, Eq .(28) shows that destabilisation of the arc plasma de-
creases with the rising vortex field at H
v
>H
cr
.

This conclusion can be
compared with experiment from [8] where temperature instability was
measured in a turbulent argon arc. Unfortunately, the velocity field was
not examined in the experiment and direct determination of the vortex
field is impossible. However, there is a possibility of calculating f

by
another method.
The energy for vortex development is taken from Joule dissipation
and divergence of heat fluxes caused by convection, conduction, and
radiation The total power consumed by turbulence per unit volume Q
turb
relates to its heat transfer rate
q h
turb ef
2
1 5
.
(33)
The relation may be defined as
h Q f
ef turb rise
/ (34)
Here 1 / f
rise
is the time needed for h
ef
development taking energy from
the source Q
turb.
Appropriate assumption is f
rise
= f. Equality of h
ef
from
Eqs. (33) and (34) gives
f
Q
q
turb
_
,
2
2 3 /
(35)
218
The value of H
v
for the region H
v
> H
cr
is derived by substituting f
into Eq. (28)
H
a
Q
q
v

_
,
2 2
2 3
2
2

/
(36)
The enthalpy pulsation can be determined from Eqs. (28) and (35)
h
v H Q
q
v
cr

_
,
0
2
2 3
2
2
/
(37)
This expression is invalid for the fourth zone, where the size of the
vortex is limited. To estimate the beginning of the restriction zone, the
size of y
m
can be examined
y y e
ae
f
a e
Q
q
m

_
,
0
2 3
2 2
2

/
(38)
Assuming that convective heat transfer can be neglected and elec-
tric field strength E is independent of radius, the quantities q
turb
, Q
turb
can be derived from the experimental profile of average temperature using
data on plasma properties as a function of temperature. The following
equations can also be used:
E
I
rdr
R
_
,
2
0
1

,
( )
q
r
E Q rdr
total rad
r

1
2
0
,
q
dT
dr
cond

, q
turb
= q
total
q Q E Q divq
cond turb rad cond
,; ,
2
where q
cond
, q
turb
and q
total
are radial
heat transfer rates: conductive, turbulent and total, while Q
rad
is the
volumetric power loss by radiation.
The experiments in [8] were carried out with a cascade arc burn-
ing in a tube with the inner diameter d = 10 mm at arc current I =
65 A and argon flow rate G = 7.7 g/sec. The flow regime corresponds
to cold Reynolds number (20
o
C) 1.0710
5
which complies with the
turbulent stream.
Comparison of the theoretical and experimental results is given in
Figs. 26. The distribution of different zones along the radius is specified
219
0,0 0,5 1,0 1,5 2,0 2,5 3,0
0
2
4
6
8
10
r,mm
H
v
x10
2
kg/ms
Fig.2 Profile of vortex field strength in argon arc, I = 65A, G = 7.7g/s, d = 10mm.
0,0 0,5 1,0 1,5 2,0 2,5 3,0
0
1
2
3
1
2
3
r, mm
qx10
-6
,
W / m
2
Fig.3 Heat transfer rates profiles: 1) q
S
; 2) q
cond
; 3) q
turb
.
by the profile of vortex field strength (Fig.2). It is seen that the free
vortex is initiated at r 0.25 mm. In accordance with the model, the
turbulent heat flux must also originate at this point, and it is confirmed
by Fig.3. Critical value H
v
= 1.83710
2
kg/ms corresponds to r 0.9
mm which is the end of the second zone. Theory predicts that enthalpy
pulsation begins to decrease from this point. This prediction is also in
good agreement with experiment that is exhibited by Fig.4. The restriction
of the vortex size begins from r 2.25 mm. It causes the bending of
220
0,0 0,5 1,0 1,5 2,0 2,5 3,0
0,2
0,4
0,6
0,8
1,0
1,2
1,4
1
2
3
r,mm
Hx10
-6
m
2
/s
2
Fig.4 Enthalpy pulsation in turbulent argon arc: 1 - experimental profile; 2 - theoretical
profile (correction: y
m
= const in the restriction zone); 3 - theoretical profile without
y
m
correction.
0,0 0,5 1,0 1,5 2,0 2,5 3,0
0
5
10
15
20
25
r,mm
y
m
x10
4
,
mm
Fig.5 Vortex size profile for a turbulent argon arc.
the H
v
curve at this point (Fig.2). The same effect is seen in Fig.5 that
demonstrates the plot of y
m
profile. Thus, a number of effects give evidence
in the favour of the hypothesis verified.
An additional evidence in the favour of our model follows from Fig.3
221
exhibiting the plots of heat transfer rates in the considered arc. It is
seen that at r 0.9 mm q
turb
q
cond
. In accordance with Fig.2, this point
corresponds approximately to H
v
= H
cr
, where vortices become self-sustaining.
Turbulent heat transfer is less than conductive one at r < 0.9 mm, but
it becomes dominant at r > 0.9 mm inspite of h

falling rapidly in this
region (see Fig.4). Since h relates to the maximal velocity of turbu-
lent vortex rotation, this result shows that the pulsation of kinetic energy
is unable of explaining the domination of turbulent heat transfer in this
region and only consideration of the field energy contained in destruction
of the vortices together with the kinetic describes the situation.
Another important confirmation of the verified model is also provided
by Fig.6, which compares the effective enthalpy increment obtained from
the average temperature profile
3 / 2
exp . ) 1 (
) 2 / (
turb ef
q H and that based
on experimental temperature instability accounting for theoretical re-
lations
exp
2
) 2 (
/ H H Q H
cr turb ef
. It is seen that agreement is rather
good with the exception of the restriction zone, and this plot also shows
r = 2.25mm as the border between the third and fourth zones.
4. ASSUMPTIONS ON TURBULENT ARCS MODELLING
The direct way for turbulent arcs modelling is connected with the
consideration of nonstationary vortex behaviour in nonuniform temperature
and velocity fields. But such a method is a very tedious one. It is expedient
to simplify the procedure by bringing in some basic relationships. The
expressions obtained above can be useful for this purpose. As it is shown
above, the key factor in turbulent arcs modelling is the effective enthalpy
H
ef
. It can be obtained from Eqs.(33)(35)
( )
1
2
1 5
r
d
dr
r H f H
ef ef

.
(39)
0.0 0.5 1.0 1.5 2.0 r,mm 3.0
0
2 0
4 0
80
60
- 1
- 2
h
ef,
kJ/kg
Fig.6 Radial distribution of effective enthalpy increment: 1) h
ef1
;

2) h
ef2
.
222
If initial temperature and velocity profiles are known, f can be found
from Eq. (35) that enables H
ef
to be derived from Eq. (39). Coeffi-
cients of turbulent thermal conductivity and kinetic viscosity can be obtained
from the following expressions

TURB ef
H
dT
dr
_
,
2
1 5 .
/
(40)
turb ef
H
du
dr
_
,
2 /
(41)
The profiles obtained assuming the laminar flow can be applied as
initial and they can then be numerically recalculated using turbulent
coefficients together with the molecular ones. Such an iteration procedure
will apparently provide the correct result. Some difficulties can arise
in the fourth zone of vortex size restriction, where disagreement between
the two forms of effective enthalpy is revealed (Fig.6), but this prob-
lem can be solved by applying appropriate corrections.
It is of interest that H
v
exhibits a linear dependence on radius (Fig.2).
In accordance with Eq. (15), this relation conforms to the uniform distribution
of the mass flow-rate across the arc column ru = const. For the con-
sidered experiment u = 67.5 kg/m
2
s or 70% of the average value.
If it is not a specific case but a general law representing, for exam-
ple, minimizing free energy, then this relationship will greatly facili-
tate turbulent arcs modeling.
Theoretical simulation of turbulent arcs would be also useful for modelling
cold flows.
5. CONCLUSIONS
A hypothesis of the existence a special vortex field created at mass movement
is verified. It has turned out to be quite useful, enabling us to derive
theoretically the critical value of the Reynolds number and the universal
velocity profile for the cold turbulent flow. Comparison of the theoretical
predictions with the experimental data for electric arc discharge also
demonstrates good agreement. It is shown that the instability of arc discharges
is governed by the vortex field laws. Heat and momentum transfer are
specified by turbulent eddies in uneven temperature and velocity fields.
The speed of vortex rotation increases up to sonic one at the expense
of thermal energy due to the thermal nozzle effect. The theory reveals
that four specific zones of turbulence can arrive in an electric arc. General
relationships for estimating parameters of turbulence are derived. Using
223
this approach, it may be possible to develop methods for pure theoretical
simulation of turbulent flows including plasma flows in electric arc
discharges.
ACKNOWLEDGMENTS
Grateful acknowledgments are made to the Fund of Fundamental Research of Belarus
for support of the study on thermal turbulence and to INTAS-94-2922 for support
of the research on electric arcs modelling.
NOMENCLATURE
a - sonic velocity; B - induction of field; c - light velocity; D - diameter;
D
e
- characteristic size of electron; E - electric field strength; e - basis of
natural logarithm; F - force; f - frequency; G - gas flow rate; H - strength
of field; h - enthalpy, Plank constant; h - enthalpy increment; I - current;
j - current density; k - Boltzmann constant; L - average length of field line;
l - mean free pass; m - mass; - P - pressure; Q - power per unit volume;
q - charge, heat transfer rate; R - radius of canal; r - radius; T - temperature;
u - velocity of flow; v - velocity of vortex; W - field energy per unit volume;
y - radius of vortex; - increment; - dynamic viscosity; - thermal conductivity;
- field permeability; - kinematic viscosity; - density; - electrical
conductivity; - shear stress; - field flux; - parameter of Nature Triangle;
- vortisability of eddy.
Subscripts: appl - applied av - average; cond - conductive; cr - critical; e
- electron; exp - experimental; i - varying; j - joining; lim - limited; loc -
local; m - mass, maximal; o - vacuum, central; rad - radiate; rise - rising;
theor - theoretical; tr - translation; v - vortex; w - wall.
Superscripts: + - nondimensional.
References
1 B.E. Launder and D.B.Spalding, Mathematical Models of Turbulence,
Academic Press, London and New York (1972).
2 O.I. Yasko, An elementary model of microworld, (Preprint N
0
19 / Luikov
Heat & Mass Transfer Institute , Russian), Minsk (1986), p.47.
3 O.I. Yasko, On the character of turbulence in electric arc,(Preprint N
o
2,
Heat and Mass Transfer Institute ) Minsk (1996), p.22.
4 O.I. Yasko, A new approach to turbulent arcs modelling, (Preprint N
0
2 / Heat & Mass Transfer Institute), Minsk (1997), p.24.
5 G.W. Kaye and T.H.Laby, Tables of Physical and Chemical Constants,
Longmans, Green & Co, London, New York and Toronto (1958).
6 S.S. Kutateladze and V.M.Borishansky, Reference Book for Heat Transfer,
(Russian), Gosenergoizdat, Leningrad and Moscow (1959).
7 E.R.G. Eckert and R.M.Drake, Jr., Heat and Mass Transfer (1959).
8 Y.K. Chien and D.M.Benenson, IEEE Trans. on Plasma Science, PS-
8, No.4, 411-417 (1980).
224
Energy Characteristics of Electric Arc
Heaters for Tetrafluoromethane
A.N. Timoshevskii, B.A. Pozdnyakov and V.S. Ponkratov
Institute of Theoretical and Applied Mechanics, SB RAS, Novosibirsk
Research into the use of low-temperature plasma in chemistry, metallurgy,
machinery and some other industries is being carried out in different
countries. Plasma processes cover heavy industries as well as industries
producing small amounts of special substances and materials. The most
developed installations for the generation of low-temperature plasma
are electric-arc plasma torches because of their reliability and stable
operation, a wide range of power used, and the ability to heat different
gases.
Plasma torches working on argon, hydrogen, air, methane and ni-
trogen have been described extensively in the literature [13], but there
is almost no data on compositions with fluorine. For example, a plasma
torch with tetrafluoromethane (CF
4
) as a working gas may be of con-
siderable interest for chemical industry. There is a small amount of data
in the literature on the plasma torch characteristics. As an example,
we can consider Ref. [4] describing a plasma torch with a self-alining
arc length; the plasma torch works on CF
4
(Fig. 1a).
A graphite bar worked as a cathode in the plasma torch; it was placed
in a copper water-cooled holder (with the help of a threaded connection).
The output electrode was made of copper; its length was 0.12 m, and
the anode diameter changed from 1 to 210
2
m. The experimental data
obtained using an electric current power of 100600 A were gener-
alised to derive an equation for calculating the voltampere charac-
teristic of this plasma torch:
U = 539 (Gdp/J)
0.29
[Volt],
where G is the gas flow rate [kg/s]; d the channel diameter [m],
225
Energy Characteristics of Electric Arc Heaters for Tetrafluoromethane
p the pressure [Pa], J the electric current power [A].
The heat characteristic has the form:
= 1.14 10
3
(I
2
/Gd)
0.94
(G/d)
0.12
(l/d)
0.62
As in Ref. [4], the pressure in the plasma torch channel was not varied
in these experiments. The anode diameter was varied in a narrow range.
Consequently it is not possible to evaluate the effect of the Knudsen
number, expressed as (p
.
d), on either the voltampere characteristic
of the arc or the specific heat losses in the plasma torch. Besides, plasma
torches with the self-alining arc length always have a drooping volt
Fig.1
226
ampere characteristic and, therefore, the real behaviour of the elec-
tric field of the arc in CF
4
is not known. From the equation of the plasma
torch heat characteristic, we conclude that the specific heat losses rise
with electric current as ~I
0.2
, when the relative length of the plasma
torch channel associated with the arc length is included in the equa-
tion ~(l/d)
0.62
. Thus, the increase of the single plasma torch power seems
to be realized more effectively by increasing the arc length, i.e. by using
plasma torches with a fixed length (with a step in the output electrode)
and plasma torches with interelectrode inserts.
Therefore, the following research problems were outlined:
a) to obtain data on the behaviour of an arc burning in CF
4
in a wide
range of parameters. The parameters are: the diameter of plasma torch
channel, the arc column length, output pressure, the plasma-forming
gas flow rate, electric current power, etc;
b) to determine the dependence of the electric field strength on the
channel diameter and electric current power;
c) to derive generalized equations of the electric and heat characteristics
of the plasma torch with a fixed arc length.
Several designs of plasma torches were studied (Fig.1). Figure 1b
shows a plasmatorch with the average arc length fixed with a step;
Fig. 1c shows a plasma torch with an interelectrode insert. In each
experiment, a bar made from dense graphite was used as the cath-
ode. It was pressed flush into a copper water-cooled holder. A cylindrical
anode made from copper worked as the output electrode.
Experiments were carried out in the following ranges of the oper-
ating parameters: electric current power of the arc was 1001000 A,
total flow rate of tetrafluoromethane in the plasma torch was (1.3
40)10
3
kg/s, the geometrical dimensions of the diaphragm were constant:
d
1
= 9 mm, l
1
= 10 mm, the length of the interelectrode insert was varied
from l
2
= 400720 mm, diameter d
2
= 435 mm. The anode dimen-
sions d
3,
l
3,
d
4
, l
4
were varied, but the ratios d
4
/d
3
and l
4
/l
3
were constant,
0.7 and 1.2, correspondingly. The pressure at the plasma torch exit in
most experiments was 10
5
Pa. A few control experiments were car-
ried out at pressure of 10
4
Pa.
Typical voltampere characteristics of the plasma torch arc for different
discharge channel are shown on Fig.2. Let us consider some of pre-
viously unknown facts:
very weak dependence of voltage on electric current;
small change of voltage with the plasma-forming gas (CF
4
) flow
rate;
voltampere characteristics have falling and rising areas;
transition of the falling branch into an increasing one occurs when
227
electric current is approximately 500 A; it does not depend on the controlling
diameter d
2
of the discharge channel;
change of gas pressure at the plasma torch exit from 10
5
Pa to
10
4
Pa (and as a result a change of the Knudsen criterion Kn~(pd
2
))
does not affect the voltage level in the arc.
These two last circumstances required a detailed study of the de-
pendence of the discharge electric field voltage on d
2
. This depend-
ence is given in Fig.3.
Experimental currents are averaged at arc current and gas flow rate
because of their weak effect. For the studied range of d
2
,we can find
two areas: when d
2
< 1 cm, E depends strongly on the discharge channel
diameter; when d
2
> 1 cm, this dependence is practically absent. For
example, the field voltage for d
2
= 25 cm and 3.5 cm is already the
same. The solid line in Fig. 4 shows the approximating function;
Fig.3 Dependence of the strength of the electric field of the arc on d
2
: o - experiment,
approximating dependence.
Fig.2 Voltampere characteristics of the arc: 1) G = 2.7510
3
; 2) 4.710
3
; 3) 8.5
10
3
; 4) 1.210
3
.
228
E = (4
.
10
5
/d
2
+ 1)V,
where is a function of the energy criterion.
Correlation between arc current and is shown in Fig.4 for all pa-
rameter ranges examined in these experiments.
For engineering calculations in current ranges above 300 A, this depend-
ence may be taken as constant. Consequently, the approximating de-
pendence on the energy criterion has a falling trend for simplifying the
generalized voltampere characteristic equation. The equation has the
form:
U = 3.5

10
3 .
(I
2
/Gd
2
)
0.05
(410
5
/d
2
2
+1) L
a
, Volt,
where L
a
is arc length.
The full arc length is calculated as
L = (l
2
+ +2.7d
2
), m.
Component is a design dimension from the cathode power to
the beginning of the interelectrode insert; the second component is the
arc length in anode capacity. Anode geometry and the average region
of the anode spot root are similar and are accurately determined as
dimension d
2
.
The averaged equation is accurate in the following ranges of pa-
rameters: I
2
/Gd
2
= (0.591.2)

10
9
, d
2
= (635) 10
3
m, l
2
= (674)
10
3
, p = 10
4
10
5
Pa.
Heat losses into the plasma torch elements were estimated in the
experiments. The generalized equation of efficiency is also derived from
results of measurements and has the form:
= 1/ = 4.3710
7
(I
2
/Gd
2
)
0.25
(G/d)
0.94
(pd)
0.68
L
1.15
Fig. 4. Generalized experimental data on in ranges: d = (635)

10
3
m, G = (1.340)
10
3
kg/s.
229
where L = l
n
/d
n
is the relative length of the discharge channel.
The accuracy of this formula is 15% in the following range of the
parameters:
(I
2
/Gd
2
) = (1.14.8) 10
5
[A
2
s

kg
1
m
1
], G/d = (0.2040.560)
[kg
.
s
1 .
m
1
], pd
2
= (0.63.5)
.
10
3
[Pa
.
m]
Agreement of the experimental data with the calculated results is
shown in Fig. 5.
If in the equation of efficiency we change the relative length of the
discharge channel L to the full length of the arc colomn L
a
and make
some transformations, we will have:
= 4.37
.
10
7
(I/d
2
)
0.5
(pd
2
)
0.68
(d
2
/G)
0.05
(L
a
/G)
1.15
Hence it follows that
the relative heat losses are practically independent of the Reynolds
number which is determined by d
2
dimension complex G/d
2
;
the relative heat losses depend strongly on the Reynolds number
which is determined by arc length ~ G/L
a
; these heat losses exceed
considerably the influence of the complex I/d (transformed energy criterion);
the working gas pressure has a strong effect on the plasma torch
section through the Knudsen criterion dimension complex pd
2
.
When designing plasma torches in practice, care should be taken
when selecting methods of obtaining the required power with minimum
energy losses.
Fig.5 Comparison of the experimental and calculated efficiency.
230
References
1 M.F. Zhukov, et al, Applied dynamics of thermal plasma, (in Russian),
Novosibirsk, Nauka (1975).
2 M.F. Zhukov, et al. Basis of calculation for plasma torches with the linear
scheme, (in Russian), Novosibirsk (1977).
3 M.F. Zhukov (ed), Plasma torches. Research. Problems, (in Russian),
Novosibirsk (1995).
4 V.S. Saidurov and G.N. Trunov, In: Electric-arc heater of tetrafluoromethane
with vortex arc stabilization (in Russian), Materials of VII Conference
of low-temperature plasma generators, Vol.1, Alma-Ata (1977).
231
Numerical Investigation of the Characteristics of a Steady Curved Arc
Numerical Investigation of the
Characteristics of a Steady Curved Arc in
External Fields
G.A. Desyatkov, V.Ts. Gurovich and E.B. Kulumbaev
Kyrgyz-Russian Slavonic University, Bishkek, Kyrgyzstan
1. INTRODUCTION
External gas-dynamic and magnetic fields are widely used to control
the geometry and characteristics of electric arcs in many plasma devices.
The interaction of an arc with external fields leads to a change of the
arc shape and generation of gas flows in plasma. As a result, an arc
curved under external actions is characterised by the absence of axial
symmetry and is essentially a three-dimensional formation. In a general
case, the problem of theoretical analysis of the properties of a curved
arc requires us to consider a set of the unsteady spatial equations of
magnetic gas-dynamics [1]. A solution of such a problem is connected
with significant difficulties, and up to the present time only the
approximate approaches based on various simplified theoretical
assumptions or requiring experimental data on local arc parameters
were developed.
Representations about the physical picture of an arc motion advanced
in the papers [2,3] allow us to determine its shape as the location of
the line of maximum temperature in space. On the basis of the simi-
lar approach, the model of spatial-temporary dynamics of the arc shape
in external gas-dynamic and magnetic fields of various geometry was
developed [4,5]. The basic problem of construction of models for finding
the arc shape is the determination of the plasma flow velocity in the
region of maximum arc temperature. Two opposite cases are usually
considered [6]: a gas flow around or gas flow through the near-axis
arc zone. Therefore, the choice of a regime of gas flow for the model
requires independent substantiation.
232
The realised approaches to numerical determination of the characteristics
of a curved arc are reduced to the analysis of a steady straight arc
in crossed magnetic and gas-dynamic fields [1]. In a number of cases,
the experimental temperature fields are used for the determination of
the gas-dynamic picture [7].
In this work, the self-consistent calculations of the characteristics
of a steady curved arc in its cross section were carried out and the
simple estimation of plasma velocity in the region of maximum arc
temperature is offered.
2. ANALYTICAL APPROACH
In [4,5], a simplified model has been developed which allows us to investigate
the spacetime evolution of low-current arcs in external gas-dynamic
and magnetic fields of different geometry. The analysis has been based
on the consideration of arc motion as displacement of its temperature
field due to the asymmetry of energy release and heat flux in the curved
arc. The energy equation describing the temperature distribution inside
an optically thin arc has the form
( )
1 1
2

S
t
U S S E + +
r r
,
where S is the heat flux potential determined by dS = (T) dT; T the
temperature;
r
U
the plasma velocity vector; E the electric field strength;
/ c
p
is thermal diffusivity; is mass density; is heat conductivity;
c
p
is the heat capacity; is electrical conductivity; is the volume
radiation density; and t is time.
Equation (1) is solved together with the Maxwell equation
rot
r r
E B t / ,
which determines the distribution of the electric field strength inside
the arc, where
H
B
is magnetic induction.
This model allowed us to obtain the equations for arc dynamics in
the form
( ) ( )
[ ]
( )
r
r
r r
V u k E
V u
p a
p a
+

0 0 0 0 0
2
1 4 1
0
/ ,
,
233
which determines the dynamics of the instantaneous local curvature
k
0
of the axial line of the arc as the function of the plasma velocity
V
p
in the region of the arc temperature maximum. Here u u
a a
,
are
the velocities of arc element motion in the normal and binormal directions;

0 0 0 0 0 0
/ , , c
p
are the plasma parameters at T=T
max
.
However this approach does not allow to determine the tempera-
ture and gas flow fields inside an arc.
3. STATEMENT OF PROBLEM AND BASIC EQUATIONS
We consider the problem of numerical modelling the curved arc under
the following formulation. Let in a result of action of a cross gas flow
(or mass forces), the arc burning between two motionless electrodes
has got a steady-state curved shape (Fig.1,a). We shall determine the
characteristics of plasma in the cross section of an arc located in a
plane of its symmetry (see shaded area on Fig.1) near to which a non-
uniformity of the characteristics in the direction perpendicular to it is
neglected. Such problem is stationary, spatially two-dimensional and can
be considered in the Cartesian system of coordinates x, y where the
x-axis is directed along the gas flow and the y-axis is perpendicular
to it (Fig.1,b).
It is supposed that the gas flow is laminar, plasma is in the local
thermodynamic equilibrium, radiation is volume, influence of the self
magnetic field on the arc characteristics is negligible. Then the set of
the Navier-Stokes equations can be written in the following form:
The equations of motion
u
u
x
v
u
y
F
p
x x
u
x y
u
y
v
x x
u
x
v
y
x
+ +

_
,
+ +
_
,
1
]
1
+
_
,
1
]
1
2
2
3
(1a)
Fig.1 The picture of arc burning in a cross gas flow (a) and the scheme of computational
region (b).
a
R
k
T
max
x
y
u
b
y
x
l
x
1
l
x
2
u
l
y
l
y
234
u
v
x
v
v
y
F
p
y x
u
y
v
x y
v
y y
u
x
v
y
y
+ + +
_
,
1
]
1
+

_
,
_
,
1
]
1
2
2
3
the equation of continuity
( ) ( )

x
u
y
v + 0
(1b)
the equation of energy balance
uC
T
x
vC
T
y x
T
x y
T
y
E
p p
+

_
,
_
,
+
2
(1c)
Here u, v are the velocity components on the axis x and y accordingly;
p is pressure; E is the modulus of the electrical field strength vector
perpendicular to the plane (x, y); F F
x y
,
are the components of mass
force density. The mass density , dynamic viscosity , thermal conductivity
, electrical conductivity and volume radiation density of the plasma
are the given functions of temperature. It is supposed that the change
of pressure inside the arc is insignificant in comparison with atmos-
pheric one and its influence on the plasma transport properties is neglected
[1].
The non-uniformity of the electrical field inside the arc along the
x-axis caused by the curvature of its column is found from the solu-
tion of the equation

r
E 0
. Accounting the chosen direction of the
coordinate axes it is approximately written as [4,5]
( ) ( )
E x E x R
k
+
0
1
(2)
where E
0
is the electric field strength on the line of the temperature
maximum having the curvature radius R
k
at the point (0,0) (Fig.1). The
values of E
0
and R
k
are considered to be given.
The boundary conditions for the system of the elliptical equations
(1) are set on the whole contour of the computational region which
due to a symmetry of the problem accordingly the x-axis is limited by
the coordinates l x l y l
x x y
1 2
0 ,
(see Fig.1,b). They have the form:
the conditions of the uniform cold flow on the entrance bound-
ary:
235
x l y l u u v T T
x y
<

1
0 0 , : , , ;
the normal gas outflow conditions from the computational region
on the exit boundary:
x l y l u x v T x p p
x y

2
0 0 0 0
2 2
, : ( ) / , , / , ;
the conditions of symmetry on the arc axis:
l x l y u y v T y p y
x x
1 2
0 0 0 0 0 , : / , , / , / ;
on the side boundary - the conditions of sticking for the gas flow
in the channel:
l x l y l u v T T
x x y
1 2
0

, : ,
or the conditions of the uniform cold gas flow in free space.
For the arc in an external mass field, the boundary conditions of
normal gas penetration through the boundary at the given pressure are
imposed on entrance, outflow and side boundaries.
The arc current is calculated from the formula
I Edxdy
l
l
l
y
x
x

2
0
1
2
(3)
4. TECHNIQUE OF NUMERICAL SOLUTION
For the numerical solution of the set of Eqs. (1) we use the method
of the finite differences on the rectangular non-uniform grid. Discretization
of the equations was made by the method of the control volume. The
pressure field was calculated with help of the SIMPLE procedure [8].
The obtained discretic analogues for all dependent variables were solved
by the iterative method.
The numerical realisation of the formulated model is rather com-
plicated as at the given arc parameters E
0
and R
k
, the velocity u
of
the cross flow is the own parameter of the problem. There is the unique
value of
u
at which the required stationary solution appropriate to a

converging numerical procedure is realised. As the iterative determi-
nation of the own parameter for considered two-dimensional problem
results in an obviously inefficient procedure of the solution, the spe-
cial technique of numerical solution was developed.
236
In its frame, the velocity u
is considered to be given, and the parameters

E
0
and R
k
are determined by the iterative procedure from the condi-
tions of fixed values of the temperatures T
0
and T
1
in the points (0,0)
and (x
1
,0) of the plane (x,y) accordingly. Then the discretic analogues
of the energy balance equation for control volumes with these knot points
become superfluous for determination of temperature. They deliver
the values of source members ensuring balances of energy in these
control volumes and calculated on current (on the given iteration) values
of dependent variables. Then, in conformity with the formula (2), the
source member in the point (0,0) defines E
0
, and in the point (x
1
,0) -
R
k
. If at such scheme of calculation of E
0
and R
k
in the common it-
erative procedure the numerical solution converges then the fixed values
of T
0
(0,0) and T
1
(x
1
,0) make sense of the intermediate conditions.
Note that the described technique of calculation can essentially simplify
a problem of numerical modelling of a movement of microwave and
optical discharges towards to a supporting parallel flux of electromagnetic
radiation [6]. Iterative determination of the normal velocity of the discharge
propagation is practically realised only in frameworks of one-dimen-
sional model (see, for example, [9]).
5. RESULTS AND ANALYSIS
Using the Eqs.(1)(3) and the above described technique, the characteristics
of the curved arc in the external cross gas flow, in the external magnetic
field and in the field of natural convection was numerically analysed.
The results obtained are shown in Figs.2-5.
Fig.2 The isotherms of the temperature field and the lines of gas streams in a cross
section of the arc burning in a cross gas flow (u
=65.7 cm/s).
-20 -10 0 10 20
-40
-20
0
20
40
60
G x y ( , ) const
0.1
0.2
0.3
0.4
0.5
0.6
mm -20 -10 0 10 20
-40
-20
0
20
40
60
-20 -10 0 10 20
T x y ( , ) const
1,
2
3
4
5
6
7
237
5.1. Arc in a cross gas flow
Calculation of the characteristics of the argon arc in a cross flow is
carried out at atmospheric pressure. The initial data for the arc at a
constant current of 60.3 A correspond to the experiment [10] in which
the influence of velocity and current on the distribution of temperature
in the cross section of the steady-state cross flow arc in the channel
of the rectangular section was investigated. They are as follows:
The results of calculations have shown that for all considered conditions
the sharp gas-dynamic and thermal fronts on the arc boundary from
a side of its concavity are characteristic (Figs.2 and 5). The cold gas
flow is decelerated on the forward arc front, promotes the formation
of its curvature, is heated up by the heat carried by thermal conduc-
tivity out from the high-temperature arc region, and extends. The main
quantity of the gas (90%) flows around the arc channel.
Downstream, gradual heating up of this gas and inflow of it in the
near-axis region from periphery occurs. In the field of the front, the
temperature and velocity of the inflowing gas are sharply increased,
and then rather slowly change inside the arc column (Fig.5). Behind
the arc front, the conditions of the quasi-uniform plasma flow are realised
(Fig.2). The narrow zone of release of the Joule energy is observed
and is displaced in the direction of the arc front (Fig.5).
The calculated temperature distribution is satisfactorily agreed with
experimental one (Fig.3). A qualitative difference is explained, first of
all, by the fact that the calculated results were obtained on the basis
of the two-dimensional model neglecting the non-uniformity of the arc
characteristics along its axis near the considered cross section. For the
arc with an interelectrode insert of 11.2 mm investigated in [10], this
non-uniformity leads to increase of the heat flux towards the cold gas.
This decreases the temperature gradient at the arc front and makes
the temperature profile more uniform. Besides, as pointed out in [10],
Computational
region
Experimental data [10] (
1
= 0.2 cm)
u
???
, cm/s T
0
, K T
1
, K
l
x1
= 4 cm 41.8 10333 9850
l
x2
= 6 cm 65.7 10402 9900
l
y
= 2.8 cm 94.7 10462 10050
127.0 10593 10300
u
, cm/s
238
-2 0 2 4 6
8.5
9.0
9.5
10.0
10.5
T, kK
x, mm
a
1 2 3 4
T, kK
8.5
9.0
9.5
10.0
10.5
y, mm
b
Fig.3 The temperature distributions (a) and (b) ( u
= 65.7 cm/s): the solid lines are

numerical results, and the experimental data [10].
for arcs with the current more than 20 A without the external cross
flow, the asymmetry of isotherms are observed due to the influence
of the electrode jets. The own impulse of the electrode jets may also
lead to expansion of the high-temperature zone in the arc cross sec-
tion. But the consideration of these effects does not qualitatively change
the flow picture.
The accuracy of the calculation results is also confirmed by agreement
of the theoretical values of the arc current obtained from Eq.(3) with
experimental value of 60.3 A: 85 A for velocity 41.8 cm/s; 67 A for
65.7 cm/s; 56 A for 94.7 cm/s; 60 A for 127 cm/s.
5.2. Arc in the mass force fields
The calculations of the characteristics of an argon arc in the external
mass force fields were carried out under the follows conditions:
a) the magnetic field
r
B B ( , ) 0
under the action of the Ampere force
F EB F
x y
, 0
where B is the induction of the field;
b) the gravity field
r
g g ( , ) 0 under the action of the Archimed force
F g F
x y

( ) , 0
where g is gravity acceleration.
The initial data for calculations were the following: l
x1
= 3 cm, l
x2
=
6 cm, l
y
= 3 cm, x
1
= 0.2 cm; T
0
= 9000 K, T
1
= 8000 K; B = 0.2 mTs.
As follows from the calculation results for these regimes, the common
gas-dynamic patterns in the vertical cross section of the horizontal arc
in the field of forces of Ampere and Archimedes are qualitatively the
same (Fig.4) but differ from the considered above one for the exter-
nal gas flow. In the field of mass forces, the radial gas inflow in the
discharge from periphery is formed which further moves in the direction
of curvature of the arc column.
239
-20 -10 0 10 20
1,
2
3
4
5
6
T x y ( , ) const
-20
0
20
40
-20 -10 0 10 20
G x y ( , ) const
0.1, g/s
1.1
2.1
mm
Fig.4 Isotherms of the temperature field and the gas flow lines in cross section of the
arc burning in an external magnetic field (B = 0.2 mTs).
Fig.5 Variations of the temperature T, velocity u, difference of pressure p and current
density j of argon arcs burning in various conditions: 1 - external gas flow: u

= 65.7
cm/s, I = 67 A, R
k
= 5.9 mm; 2 - external magnetic field: B = 0.2 mTs, I = 23 A, R
k
=
5 mm; 3 - natural convection: I = 17 A, R
k
= 5 mm.
Such flow pattern arises due to that the cross gas velocities are com-
parable to velocity of the thermal wave (i.e. the gas has enough time
to be heated up) and that the negative pressure difference p p p

inside the arc is appeared. However, as for the external flow, the pressure
difference is positive before the forward arc front that limits a quantity
of gas flowing into the discharge from a side of arc concavity. In the
field of the front, the temperature and velocity of inflowing gas are
also sharply increased. Further, already in the currentless region, a gas
continues to accelerate in the field of the Archimed force and begins
to decelerate in the field of the Ampere force.
6. ESTIMATION OF PLASMA VELOCITY IN THE REGION OF
MAXIMUM ARC TEMPERATURE
The simple approach based on the theory of shock waves allows to
-20 0 20 40 60
-0.2
0
0.2
0.4
0.6
0.8
1.0
1.2
p, Pa
1
3
2
x, mm
-4 -2 0 2 4 6
40
80
120
160
1
2
3
j , A /cm
2
x, mm
240
consider the problem of estimation of plasma velocity in a high-temperature
zone of the discharge.
Let us assume that the cold gas flow has at infinity the mass density
and velocity u
. The gas remains cold up to the front of the dis-

charge. Along the zero gas stream line, the Bernoulli equation is ful-
filled in the form
u p u
1
2
1
2
2 +

/ .
Here the index 1 corresponds to the parameters before the dis-
charge front;
p
1
is the pressure difference appearing in result of reducing
gas velocity at the approach to the front. At transition of the flow behind
the discharge front, heating and ionization of gas takes place, and the
region of the quasi-uniform flow is realised with the temperature close
to maximum one inside the discharge and with density
1
with the absence
of appreciable viscous forces and pressure difference.
The thickness of the thermal and gas-dynamic boundary layer where
reorganisation of the flow takes place is small in comparison with the
size of the area of the quasi-uniform flow. We consider it as a shock
surface on which the conditions of preservation of the mass flow densities

u u
1 0 0
, and components of flow momentum density
+ u p u
1
2
1 0 0
2
are fulfilled.
The above equations define the gas-dynamic parameters of the plasma
flow inside and before the discharge for the given parameters of the
cold gas flow. Taking into account that
0
<<

we have
u u u u p u
0 0
1 2
1 0
1 2
1
2
2 2 2

( / ) , ( / ) , /
/ /
(4)
Note that for similar estimations in [3] the formula u u
0 0
2

( / )
giving on the order exceeding values of plasma velocity inside an arc
was obtained.
The gas flow in the vertical cross section of a horizontal arc in the
field of gravity under action of the Archimedes force density f g

is differently formed. The uniform plasma flow of extention L is caused

now by joint action of the Archimedes force and negative pressure behind
the discharge front. Thus, similarly to the previous case, we have
u fL u fL p fL
0 0 1 0
2
1 0
2 2 1 2

/ , / , ( / ).
Comparison of the calculated gas velocities in the zone of maximum
241
isotherm of the arc with estimations of their values by the above mentioned
model shows the follows. For the regime with the external gas flow
with
u
=0.418 m/s and u
=1.27 m/s, formula (4) gives the values of

velocities 1.77 m/s and 5.49 m/s that well correlate with calculated values
2.78 m/s and 6.07 m/s accordingly. For the regime in which natural
convection is taken into account, the model estimation of the plasma
velocity about 1.93 m/s is close to the value of 1.97 m/s from numerical
calculation.
References
1 M.F.Zhukov, et al, Theory of Thermal Electric Arc Plasma, Vol.1, Methods
of mathematical investigation of plasmas, (in Russian), Novosibirsk,
Nauka (1987).
2 O.Ya.Novikov, Stability of electric arc, (in Russian), Publ. Energy, Leningrad
(1978).
3 H.Maecker, Proc. IEEE, 4, 4-14 (1971).
4 G.A.Desyatkov, et al, J. High Temperature Chem. Processes, 1, No.3,
291-298 (1992).
5 G.A.Desyatkov, et al, Proc. of 10th Intern. Symp. on Plasma Chemis-
try, Bochum, Germany, Vol.1 (1991) pp.1.1-24.
6 Yu.P.Raizer, Physics of gas discharge, (in Russian), Publ Nauka, Moscow
(1987).
7 N.Sebald, J. Appl. Phys, 21, 221-236 (1980).
8 S.Patankar, Numerical methods of solution of problems of heat exchange
and hydrodynamics, (in Russian), Energoatomizdat Moscow (1984).
9 J.P.Kneht and M.M.Michi, AST, No.1, 133-140 (1989).
10 D.M.Benenson and A.A.Tsenkner, Teploperedacha, (in Russian), No.2,
69-77 (1970).
242
Properties of Water-Stabilized Plasma
Torches
Milan Hrabovsk, Milo Konrd, Vladimr Kopeck and
Viktor Sember
Institute of Plasma Physics, Academy of Sciences of Czech Republic,
182 21 Prague 8, Czech Republic
Abstract: Plasma torches with water stabilized arcs provide an alternative
to commonly used sources of thermal plasmas based on gas-stabilized arcs
or RF discharges. Relatively long arc columns can be stabilized by water
walls under very low mass flow rates through the torch chamber. This leads
to high values of plasma temperature and enthalpy. In water-stabilized torches,
an oxygen-hydrogen plasma jet is produced with extremely high plasma enthalpy
and flow velocity. Plasma torches with the water-stabilized arc provide special
performance characteristics in some plasma processing applications like plasma
spraying or waste treatment. Physical processes which determine the properties
of generated plasma are discussed and basic characteristics of the water plasma
torch are presented.
1. INTRODUCTION
Thermal plasmas are commonly generated in inductively coupled discharges
or in electric arcs which are stabilized by the gas flow. Typically, inductively
coupled plasma torches are characterized by plasma temperatures from
6 000 K to 10 000 K. Averaged temperatures in torches with non-transferred
electric arcs with gas stabilization are somewhat higher, usually in the
range from 8 000 K to 14 000 K. Average plasma enthalpies, determined
as the ratio of the useful power of the plasma generator to the flow rate
of the plasma forming gas, vary from 1 to 100 MJ/kg. Further increase
of plasma temperatures and enthalpies is limited by the fact that the
flowing gas protects the arc chamber walls from thermal overloading
and a minimum possible gas flow rate thus exists for the given arc power.
243
Properties of Water-Stabilized Plasma Torches
Higher thermal loading is possible if the walls are created by water and
an arc is stabilized by wall evaporation. Thus, water-stabilized arcs can
be utilized as sources of thermal plasma with high temperatures and
enthalpies.
The electric arc with stabilization of the arc column by the water
vortex was first described more than seventy years ago by Gerdien and
Lotz [1] [2]. Basic experimental investigations of the water-stabilized
arcs were performed in the 1950s. Maecker et al. [3,4] measured basic
electric characteristics of the arc and studied the effect of the length
and diameter of the stabilizing channel. Several investigators measured
arc plasma temperatures using methods of emission spectroscopy [5]-
[7]. All authors reported very high plasma temperatures in the arc column
with maximum about 50 000 K in the centerline position. The princi-
ple of arc stabilization by the water vortex was utilized in the plasma
torch designed for plasma spraying and cutting [8]. Despite the pos-
sibility of achieving extreme performance characteristics, the development
of water-stabilized torches has been limited especially due to their more
complex structure and due to the lack of understanding of physical processes
in the arc which is necessary for improvement of the torch design.
The investigation of water-stabilized plasma torches started several
years ago with the aim of better understanding of the arc processes,
development of plasma torches and determination of the characteristics
of the generated plasma jet. Some basic results of this investigation are
presented. Differences between water- and gas-stabilized torches are discussed
and experimental characteristics of water-stabilized arc and of the generated
plasma jet are shown. The performance characteristics of water-stabilized
torch in applications like plasma spraying are presented.
2. EFFECT OF PLASMA GAS AND ARC CHAMBER DIMENSIONS
ON ARC CHARACTERISTICS
To estimate the effect of the material properties of the plasma gas as
well as of the effect of dimensions of the arc chamber on the arc
characteristics, we derive simple equations for the basic characteristics
of the arc column, which can be applied on both liquid- and gas-stabilized
arcs in a cylindrical chamber of the plasma torch. The integral energy
balance equation of a cylindrical arc column can be written in the form
[9]
A
r
T
k R E A R mh
z
A h v
n
R r
z

4 2 ) (
) (
2
+ =
=
(1)
where r is plasma density, v
z
axial velocity, h enthalpy, s electric conductivity,
244
k thermal conductivity, T temperature, e
n
net emission coefficient representing
power loss due to radiation and E electric field intensity. A = p R
2
is
the cross section of the arc chamber with radius R, m is the mass flow
rate from the chamber walls per unit length, r and z are the radial and
axial coordinates. The equation (1) was obtained from the energy balance
equation of a cylindrical arc column by integrating over the cross section
of the arc chamber. The quantities averaged over the cross section A
are defined by the equation
=
R
rXdr
R
X
0
2
2
1
(2)
A simple equation can be obtained if derivatives in equation (1) are
approximated as
L
A h v
z
A h v
z z

=
) (
,
R
S
R
S
r
T
k
R r
= =
(3)
where L is length of the arc chamber and heat flux potential S is defined
as
=
T
T
dT k S
0
(4)
Enthalpy h(R) of vapour flowing from the wall into the arc chamber
corresponds to the boiling temperature of water T
B
and can be put equal
to zero h(R) = h (T
B
) = 0. The integral energy balance equation can
be then written in the form

2
2
2 2 2 2
4 2
R
I
E R R S
L
h G
n
= = + +
(5)
where arc current I and total mass flow rate G are given by the equations
245
E R dr r E I
R

2
0
2 = =
(6)
z
R
z
v R dr v r G
2
0
2 = =
(7)
The following two equations for electric field intensity E and arc current
I can be then derived
n
R S
L
h G
R
E

2 2
4 2
1
+ + =
(8)
n
R S
L
h G
R I
2 2
4 2 + + = (9)
These equations can be applied to both gas-stabilized and water-stabilized
arcs.
In equations (8) and (9), material coefficients s, h, S and e
n
which
are dependent on plasma temperature, R and L are geometrical design
parameters and G is the total mass flow rate. Thus, the relation between
the electrical characteristics of the arc I, E and the averaged plasma
temperature can be found from the equations (8) and (9) for given values
of the ratio G/L of the gas flow rate to the arc length, and for given
arc chamber radius R.
Figures 1 and 2 show the curves defined by the equations for three
plasma media water, argon and nitrogen. LTE values of s, h and S in
dependence on temperature were determined using computer code ADEP
[10], values of net emission coefficient e
n
were taken from [11,12].
Stabilization of the arc by water leads to high electric field intensities
and thus high arc powers, and also to high plasma enthalpies. Curves
in Figs. 1 and 2 also represent the strong effect of the ratio of the mass
flow rate to the arc length G/L on the arc characteristics. It will be shown
in part 5 that the main difference between the gas-stabilized and wa-
ter-stabilized torches is given by different values of the ratio G/L. For
gas-stabilized torches, the mass flow rate G can be controlled independently,
246
8000 12000 16000 20000
plasma temperature [K]
0
4000
8000
12000
16000
E
l
e
c
t
r
i
c

f
i
e
l
d

i
n
t
e
n
s
i
t
y

[
V
/
m
]
0.001
0.01
G/L = 0.1 kg/s.m
Fig. 1 Electric field intensity in the arc column with radius R = 3 mm for different
ratios of gas flow rate G to the arc length L for arc in water (full line), nitrogen
(dash-and-dot line) and argon (dash line).
0 200 400 600 800 1000
arc current [A]
0E+0
1E+8
2E+8
3E+8
p
l
a
s
m
a

e
n
t
h
a
l
p
y

[
J
/
k
g
]
G/L =
0.001
0.01
0.1
Fig. 2 Dependence of mean plasma enthalpy on the arc current for different ratios
of the gas flow rate G to the arc length L for arc in water (full line), nitrogen (dash-
and-dot line) and argon (dash line).
but a minimum value G/L exists for the given arc power. This value
is related to the maximum possible heat flux to the walls of the arc chamber,
as the flow of plasma gas protects the walls against thermal overloading.
For the water-stabilized arcs, the mass flow rate G is determined by
the rate of evaporation of the stabilizing wall m. Thus, G = mL is a
247
parameter, which is given by the balance of radial heat transfer in the
arc chamber. Long arc columns at low mass flow rates can be stabi-
lized in water and thus very low values of G/L can be achieved in the
water stabilized arcs. In part 4, processes controlling the mass flow rate
will be analyzed.
3. PRINCIPLE OF STABILIZATION OF ARC BY WATER
A schematic picture of a water-stabilized arc is shown in Fig. 3. An
arc is ignited in the center of a vortex of water which is created in the
arc chamber by means of tangential water injection. Water flows over
the segment, which determines the inner diameter of the vortex, into the
exhaust slot. Evaporation from the inner surface of the vortex and heating
and ionization of vapour are principal mechanisms that produce arc plasma.
The energy, dissipated in the conducting arc core by Joule heating, is
transported radially to the inner surface of the vortex by radiation, heat
conduction and turbulent transfer. Evaporation rate m is determined by
a fraction of total power reaching the liquid. The other part of transferred
energy is absorbed in a vapour zone (between the arc column and the
surface of water) causing heating and ionization of vapour. A part of
the energy which is transferred into the liquid represents power loss Q.
Thus, the balance of radial heat transfer is decisive for the value of
m, which affects all characteristics of the arc. This balance is analyzed
in the following part on the basis of experimental results.
vapour zone
I E
m
cathode
liquid vortex
anode
Q
tangential
water inlet
water
outlet
exit
nozzle
arc
Fig. 3 Schematics of liquid stabilized arc.
248
d
Fig. 4. Schematic of the arc chamber.
4. EXPERIMENTAL INVESTIGATION OF THE WATER-
STABILIZED PLASMA TORCH
4.1. Experimental system
A schematic picture of the experimental arc chamber is shown in Fig.
4. The chamber is divided into several sections by the baffles with central
holes. Water is injected tangentially into the sections where the vortex
is created. The inner diameter of the vortex is determined by the diameter
of the holes in the baffles. Water is discharged at two positions along
the arc chamber. The cathode is created either by a graphite rod, which
is automatically moved into the chamber to compensate for erosion, or
by a small rod of zirconium, which is pressed into a copper sleeve rod.
An anode made of a copper disc with internal cooling is located outside
the arc chamber downstream of the nozzle exit. The anode disc rotates
to reduce strong electrode erosion in the atmosphere containing oxygen.
The length of the part of the arc column stabilized by the water vortex
inside the arc chamber is 55 mm, the inner diameter of the vortex is
7 mm, the diameter of the nozzle is 6 mm. The length of the part of
arc column outside the chamber between the exit nozzle and the anode
varies between 3 and 15 mm as the anode attachment moves along the
anode surface.
The scheme of the experimental system is shown in Fig. 5. Water
from the tank WT is pumped into the arc chamber and after leaving
the chamber due to the inertial force in the vortex it is exhausted us-
ing the second pump. The water flow rates through the system have to
be exactly adjusted and kept constant to ensure stable operation of the
249
. . . .
.
. .
R
B
R
B
VD
VD
VD VD
T
T
q
q
T
q
WT
C
A
S
PS
+
EN
P2 P1
PC OMA
lens
. .
_
Fig. 5. Experimental system. WT water tank, q flow meter, T thermometer,
R
B
bias resistance, VD voltage divider, S shunt, PS power source, EN -
exit nozzle, A anode, C cathode, P1, P2 electric probes.
torch. The flow rates and temperatures are measured both at the input
and at the output of the arc chamber and the anode. Arc voltage and
the potential of the exit nozzle were measured using voltage dividers
VD, for arc current measurement shunt S was connected in the cath-
ode part of the circuit. For measurement of the plasma potential at the
position of the exit nozzle, the high resistance divider has to be used
to measure the floating plasma potential.
Electric probes P1 and P2 were applied for the measurement of plasma
flow velocity. The probes were negatively biased by connecting to the
cathode through the bias resistance R
B
, i.e., they were in the ion col-
lecting regime. For low values of the bias resistance, the probe potential
was given by the ion saturation current and thus it was highly sensi-
tive to the plasma properties in the position of the probe tip. In this
regime, the probe signal exhibited intensive fluctuations as plasma dis-
turbances passed along the probe. From the correlation of signals of
two probes located in different axial positions along the plasma jet the
velocity of movement of plasma disturbances could be evaluated, which
is supposed to characterize the velocity of the plasma flow. To reduce
the probe erosion, the probes were moved across the plasma jet with
the velocity that was small compared to the plasma velocity. The probe
measurements are described more in detail in [13,14].
The temperature of the plasma was measured by optical emission
spectroscopy. The Jobin Yvon OMA system used for spectroscopic
diagnostics consists of a grating monochromator HR-320 (Czerny-Turner
250
L A BE PJ
L
S
C
Fig. 6. Schematic setup for Schlieren photography. LA-pulsed Nd laser, BE-beam
expander, PJ-plasma jet, L-lens f=1500 mm, S-Schlieren stop, C-camera.
configuration, f = 0.32 m) fitted with the linear photodiode array de-
tector RY/1024, a control device and a personal computer. The Image
of the plasma jet was transversely scanned by means of an optical stand
with an optical fibre, giving a spatial resolution about 0.1 mm. In this
way time-averaged radial intensity profiles were obtained [1517].
The structure of the flow field of the plasma jet was visualized by
schlieren photography. In the focus of the lens a circular stop was placed
so only deflected rays were imaged by the camera (Fig. 6).
4.2. Measured characteristics of the water stabilized torch
The voltampere characteristics of the arc are shown in Fig. 7. Arc voltage
U
a
and the potential difference between the cathode and the exit nozzle
U
e
are plotted against the arc current. The measurements were made
for several positions d of the anode with respect to the axis of the nozzle
(see Fig. 4). The averaged values of arc voltage are shown as well as
minimum values during the saw tooth-shaped fluctuations of arc voltage
Fig. 7. Voltampere characteristics of the arc for several distances d of anode from
the nozzle axis. U
a,min
- minimum value of arc voltage, U
a,av
mean value of arc
voltage, U
e
- potential of the exit nozzle.
350 400 450 500 550 600 650
160
200
240
280
320
360
U
e
[mm] =
6
5
4
3
U
a, mi n
U
a, a v
7

[
V
]
1 [A]
@
251
caused by movement of anode attachment [18]. It can be seen that the
total arc voltage depends on the position of anode as the length of the
arc column between the nozzle exit and the anode attachment is influenced
by d. The potential of the nozzle U
e
is not affected by d.
The power balance of the arc (Fig. 8) was evaluated from the electric
characteristics of the arc and from calorimetric measurements on the
stabilizing water system. Total power input W
T
is plotted as well as the
power input W
c
into the part of the arc column stabilized by the wa-
ter vortex. The other characteristics of the stabilized part of the arc
column are the power loss to water P
w
and total enthalpy flux F
e
through
the exit nozzle.
The temperature in the water plasma jet was determined from the
analysis of the results obtained by different spectroscopic methods, including
the Boltzmann plot of OII lines in the spectral region 370475 nm, different
ratios of the relative emission coefficients of OI lines in the IR spec-
tral region, hydrogen Balmer H
b
and some OII lines, and Stark broadening
of H
b
line. A typical sample of the radial distribution of emission spectrum
in the region around H
b
line is shown in Fig. 9. Evaluation and inter-
pretation of the spectroscopic data was based mostly on the assump-
tion of LTE and cylindrical symmetry. Nevertheless, deviations from LTE,
resulting from both the presence of steep temperature and density gradients,
and high velocity of the plasma flow, as well as from the cylindrical
symmetry due to anode arc root attachment occurring outside the main
plasma jet, were also investigated [16,17].
The distribution of the relative emission coefficients was obtained
with the help of asymmetrical Abel inversion based on the assumption
350 400 450 500 550 600 650
current [A]
0
40
80
120
160
200
p
o
w
e
r

[
k
W
]
W
t
W
c
F
e
P
w
P
a
d = 3 mm
d = 5 mm
d = 6 mm
Fig. 8. Power balance of the arc for several positions of anode. W
T
total power
input, W
c
power input into stabilized part of the arc column, P
w
power loss to
the water vortex, P
a
power loss to the anode, F
e
total enthalpy flux through the
exit nozzle.
252
[nm]
468 480 492 504
,
r
)

[
a
.
u
.
]
r

[
m
m
]
0
.
0
3
.
0
OII
H
OII

Fig. 9. Radial distribution of emission coefficient (,r) in the spectral region with
H
b
line (486.13 nm) and several OII lines.
that the plasma column keeps a cylindrical shape within which an asymmetric
distribution of radiation, and hence temperature and density, exists only
normal to the direction of observation [19]. Deviations from LTE were
examined using the close-to-LTE concept that describes the departure
of excited state populations form their LTE values by means of nonequilib-
rium factors which are defined as the ratio of the true state population
density to that given by equilibrium relations. The nonequilibrium factors
were obtained by comparing the measured ratio of line and underlying
continuum emission coefficients with that calculated using the upper level
Saha equation in case of an atomic line and the Boltzmann distribu-
tion in case of an ionic line. To take into account the reabsorption of
spectral lines, an iterative computational procedure based on the op-
tically thick integral equation for spectral intensities was applied to
simulate the measured line-to-continuum ratio [20].
It has turned out, among others, that a significant underpopulation
of atomic and ionic excited states can be expected, having a strong impact
on the energy balance in the plasma jet, which can explain, at least partially,
why measured temperatures do not agree sufficiently with the predic-
tions of the present numerical model of the water plasma jet [21] that
is likely to overestimate radiation losses [22].
Based on this analysis, the temperature distribution in the laminar
part of the plasma jet was derived. On the other hand, this approach
failed in fringes and the region near the arc anode attachment due to
the strong departure from LTE and quasi-cylindrical symmetry used for
an evaluation of the Abel inversion. The resulting radial profiles of tem-
perature at the position 2 mm downstream of the nozzle exit are shown
in Fig. 10 for several values of arc current [17].
253
0.0 0.5 1.0 1.5 2.0 2.5 3.0
8
12
16
20
24
28
r [mm]
T

[
1
0
3

K
]
I = 600 A
500 A
400 A
300 A

Fig. 10. Radial profiles of plasma velocity at the position 2 mm downstream of
the nozzle exit.
The generated plasma jet is characterized by the presence of large
scale structures with very high density gradients [18]. Two schlieren
images of plasma jet in Figs. 11 and 12, taken at the arc current of
500 A and exposure time of 35 ns, show the structure of the mixing zone
around the jet and flow field close to the torch exit. Intensive turbu-
lent mixing causes rapid decrease of temperature and velocity along the
jet. Isotherms of the plasma jet generated at an arc current of 400 A
Fig. 11. Schlieren photography of the jet for larger deflections (diameter of schlieren
stop = 5 mm).
254
are shown in Fig. 13, development of the centerline velocity along the
jet is shown in Fig. 14. The velocities were determined from the time
shift of signals of two probes corresponding to the time of flight of plasma
disturbances [13,14]. The high level of correlation of the signals was
found by the statistical analysis of measured signals for probe distances
up to 7 mm from the torch exit. Plasma flow velocities close to the nozzle
exit are higher than 1 km/s and are strongly influenced by the arc current.
Despite of their rapid decrease with increasing distance from the nozzle
exit, both the temperature and the velocity are high in the jet.
4.3. Analysis of power and mass balances in the arc chamber
The basic physical processes in the arc chamber are processes controlling
the radial and axial transport of energy dissipated in the arc column
by Joule heating. The energy and mass balances in the stabilizing channel
are shown schematically in Fig. 15. Part Q
R
of power IE dissipated in
the conducting arc core by Joule heating is transported radially to the
inner wall of the water vortex. Here it causes evaporation of mass m
of water. The rest of power Q
W
is absorbed either in the body of water
or in the walls of arc chamber. Principal mechanisms involved in the
radial transport of energy from the conducting core through the sheath
of vapour to the water wall are radiation, heat conduction and turbulent
transfer. Radiative heat transfer is the decisive mechanism for very high
plasma temperatures achieved in the water torch. The total radiative loss
from the unit length of the arc column can be expressed by the empirical
relationship [23]
Fig. 12. Schlieren photography of the jet region close to the torch exit for smaller
deflections (diameter of schlieren stop = 1 mm).
255
10 20 30 40
-2
-1
0
1
2
3
-3
22 21
20
19
18
17
16
15
6 = 14 x 10
3
K
13 12
z [mm]
r

[
m
m
]
Fig. 13. Isotherms in the water plasma jet generated in a water torch at an arc
current of 400 A and an arc power of 107 kW.
0 40 80 120 16
axial distance [mm]
0
2
4
6
8
v
e
l
o
c
i
t
y

[
k
m
/
s
]
I = 300 A
I = 400 A
I = 500 A
I = 600 A
Fig. 14. Development of centerline velocity along the plasma jet.
IE Q
R
= (10)
where I is the arc current, E the electric field intensity and a an empirical
coefficient. Fraction e of this power is absorbed in the vapour sheath
surrounding the arc core. The absorption coefficient e is dependent on
the mass flow rate m of vapour evaporated from the water surface per
unit length of the arc column, i.e., e = e(m). The remaining power reaches
the water surface, part of this radiation is absorbed in the boiling layer
of water on the surface from which the water is evaporated. The power
256
water
vortex
vapour
sheath
conducting
core
m
mass balance
energy balance
exit
nozzle
boiling
layer
I .E
.IE.{1-(m)}
Q
w
= .IE.{1-(m)}-m.Q
ev
Q
R
=IE
m L r vdr .
Re
=
2
0

= 2
0
r vhdr
Re
L IE Q m Q w ev ( . )
=
T = T
0
T = T
B
spent for evaporation is given by the relationship
)] ( [
0
T T C m mQ
B w ev
+ = (11)
where l is the specific latent heat of vaporization, C
w
the specific heat
capacity, T
B
boiling temperature and T
0
is the temperature of water in
the vortex. The rest of radiated power Q
w
enters the body of water and
is absorbed either in the water or in the walls of stabilizing chamber.
As can be seen from Fig. 15, the ratio of power which comes to the
water surface to the total Joule power is a function of the empirical
parameters a and e:
( ) ) ( 1 m
IE
mQ Q
ev w
=
+
(12)
The total energy and mass balances at the exit nozzle of the torch
are written in the right hand side of Fig. 15. The plasma velocity v in
the integrals can be expressed as v = cM, where c is the velocity of
sound and M is the Mach number. Assuming that the radial pressure
gradients are negligible and radial velocities are small compared to the
Fig. 15. Mass and energy balances in arc stabilized by water vortex.
257
axial velocity, the Mach number can be assumed to be independent of
the radial co-ordinate at the exit nozzle. This assumption is often ap-
plied in the description of high current arcs in nozzle flows [24]. The
following equations for Mach number M and evaporation rate m can
be then derived:
( )
+ +
=
E
E
R
R
B w
w
dr c r T T C dr ch r
Q IE L
M
0
0
0
2 )] ( [ 2
(13)
dr c r M mL
E
R

=
0
2
(14)
where r is plasma density, h enthalpy, L the length of the arc column
and R
E
the radius of the exit nozzle. The integrals in equations (13)
and (14) can be determined from the measured temperature profiles and
from the transport and thermodynamic coefficients of arc plasma assuming
the existence of LTE [25]. Thus, Mach number M and mass flow rate
m can be evaluated from measured temperature profiles at the nozzle
exit and from the power balance of the arc chamber. The values of the
Mach number and the mass flow rate evaluated from the results of
measurements are shown in Fig. 16 in dependence on arc current. From
the values of M, the ratio of stagnation pressure inside the arc chamber
to the atmospheric pressure p/p
0
can be evaluated. This ratio was derived
assuming that the temperature along the arc column is constant. For
the isothermal flow, the pressure ratio is given by the equation
( ) 2 exp
2
0
M p p = (15)
where is the adiabatic coefficient of water plasma. The dependence
of the pressure ratio on the arc current is also presented in Fig. 16.
Knowing the value of Mach number, the velocity profile can be derived
from the measured temperature profile using relation v(r) = Mc[T(r)]
where the velocity of sound c is the function of temperature for equi-
librium conditions in plasma. The pressure at the nozzle exit was supposed
to be equal to the atmospheric pressure. As the maximum of the pressure
ratio given by equation (15) was relatively low (see Fig. 16) and the
dependence of the sound velocity on pressure is weak, the possible error
258
Fig. 16. Mach number M, evaporation rate m and pressure ratio p/p
0
in dependence
on arc current.
0 .0 1. 0 2 .0 3 .0
0
2 000
4 000
6 000
8 000
I = 600 A
500 A
400 A
300 A
v
e
l
o
c
i
t
y

[
m

s
-
1
]
r [mm]
Fig. 17. Radial profiles of plasma velocity at the position 2 mm downstream of
the nozzle exit.
in estimation of pressure has no effect on the resulting values of ve-
locity. The evaluated velocity profiles are shown in Fig. 17. Plasma
velocities are high compared to the velocities commonly reported for
gas-stabilized plasma torches. The velocities at r = 0 can be compared
with the velocities of plasma disturbances determined from the probe
measurements shown in Fig. 14. Measured velocities at Fig. 14 are somewhat
lower than the plasma velocities determined from the temperature profiles,
with the exception of values for I = 600 A.
In Fig. 18 the values of the left hand side term of equation (12) are
plotted against m. The values of the quantities in (12) were determined
259
Fig. 18. Dependence of the fraction of power coming to the water surface on evaporation
rate.
Fig. 19. Balance of radial transfer of energy in the arc column. ( is the fraction
of total arc power transferred in the radial direction, the fraction of this radiant
power absorbed in the vapor sheath, the fraction of radiant power spent for evaporation
and is the fraction of radiant power absorbed in water.
from the measured data as Q
w
= P
w
/L, E = U
a
/L and Q
ev
was evaluated
from measured temperatures T
0
using equation (11). Assuming absorption
coefficient (m)0 for m0, the values of empirical factors and (m)
can be evaluated if the dependence in Fig. 18 is extrapolated to
m = 0. The resulting values of constant factor and the dependence
of absorption coefficient on arc current are given in Fig. 19 for extrapola-
tion of the dependence in Fig. 18 by linear function (curves 1) and by
the polynome of the second order (curves 2). Figure 19 also shows the
fraction of radiated power spent for evaporation = mQ
ev
/IE and the
3.00 4.00 5.00 6.00
evaporation rate m [kg/s.m]
0.35
0.40
0.45
0.50
0.55
(
Q
w

+
m
.
Q
e
v

)
/
I
.
E
300 400 500 600
arc current [A]
0.01
0.10
1.00
p
o
w
e
r

f
r
a
c
t
i
o
n
1 = 0.57
2 = 0.87
1
2
1
2
1
2
260
fraction of radiated power absorbed in the water = Q
w
/IE. Most of
the radiation is absorbed in the water and contributes to the power loss,
about 0.20.5 of radiation power is absorbed in the vapour sheath and
contributes to the heating of vapour, and only 0.010.02 of radiation
power is spent on evaporation from the water surface. The fraction of
the total power spent on evaporation is then 0.0070.01. This very
low fraction of total arc power spent on the production of the plasma
gas is the main cause of extremely high plasma temperatures and
velocities accompanied by the low mass flow rate through the exit nozzle.
5. COMPARISON OF WATER STABILIZED AND GAS STABILIZED
TORCHES
The basic characteristics of the plasma jet, evaluated from measurements
(see part 4) carried out on the water-stabilized plasma torch operating
at a power up to 176 kW, are summarized in Table 1. In addition, the
characteristic frequencies of the production of turbulent eddies due to
entrainment of cold gas into the jet are given. The frequencies were
determined from the power spectra of emitted light from the jet [26].
High characteristic frequencies are related to high plasma flow velocities.
The non-dimensional frequencies given by Strouhal number St = fD/v
m
are close to the values found in jets produced in gas stabilized torches
with substantially lower flow velocities.
The values given in Table 1 are related to the nozzle exit. As there
is a part of the arc further downstream of the nozzle (between the nozzle
and the external anode), some additional energy is dissipated into the
jet, so the total enthalpy is somewhat higher than the one in Table 1.
The comparison of the operation regimes of the water-stabilized torch
with the regimes of conventional gas-stabilized torches and the high power
torch PlazJet (produced by TAFA) is shown in Fig. 20. The values for
the conventional gas-stabilized torches were determined from the data
given in [27].
The water torch is characterized by a very low mass flow rate and
a high arc power. These characteristics are connected with the principle
of stabilization of the arc by the water wall. Comparison of the typi-
cal parameters of the water-stabilized torch with the characteristics of
two commercially available gas-stabilized plasma torches is given in
Table 2.
The values of the mean plasma enthalpy and mean temperature for
the gas-stabilized torches were estimated assuming torch efficiency equals
60%. The thermodynamic properties of plasma gases were calculated
on the basis of computer code and database ADEP [10]. The values for
261
the water-stabilized plasma torch were taken for the case of the total
arc length including the part between the nozzle and the external
anode and so the temperatures are somewhat higher than in Table 1.
The principal difference between the water- and gas-stabilized arcs
follows from the effect of ratio G/L on the arc characteristics (see part
2). Due to possibility of higher thermal loading of the wall in the water-
stabilized plasma torches, the arc can be stabilized with very low values
of the ratio G/L. For the water torches, values G/L of the order 0.001
kg/ms are typical while for gas-stabilized torches minimum possible
values G/L are more than one order higher in magnitude. Thus, very
high plasma enthalpy and temperature can be achieved in the water-stabilized
torches.
The effects of the torch parameters (for the cases given in Table 2)
on the operation characteristics in plasma processing applications are
illustrated on the influence of loading on transfer of heat and momentum
to treated material. Due to the injection of the treated substance into
the plasma jet, the temperature and velocity of plasma are reduced. Thus,
the heat and momentum transfer to the particles is dependent on the amount
of treated material. The effect of loading for the water- and gas-sta-
bilized torches is compared in Figs. 21 and 22.
Cooling of plasma due to the presence of particles results in a re-
duction of the heat flux to the particles. Curves in Fig. 21 represent
the dependence of heat flux potential S
Fig. 20. Comparison of operation regimes of different plasma torches - water torch,
conventional gas torches and high power torch PlazJet.
0.0 2.0 4.0 6.0
mass flow rate [g/s]
0
50
100
150
200

PlazJet
water torch
conventional gas torches
p
o
w
e
r

[
k
W
]
262
Table 1 Water-stabilized torch operating parameters and jet exit condtions
arc current (A) 300 400 500 600
power input (kW) 84 106.8 139 176
mass flow rate (g/s) 0.204 0.272 0.285 0.325
mean temperature (K) 13 750 14 500 15 400 16 200
centerline temperature (K) 19 000 23 000 26 200 27 200
mean enthalpy (MJ/kg) 157 185 230 272
mean velocity (m/s) 1 736 2 635 3 247 4 230
centerline velocity (m/s) 2 494 4 407 5 649 7 054
mean density (kg/m
3
) 4.15x10
-3
3.64x10
-3
3.1x10
-3
2.72x10
-3

centerline density (kg/m
3
) 1.92x10
-3
1.23x10
-3
0.98x10
-3
0.92x10
-3

mean density ratio S 0.0034 0.0030 0.0026 0.0023
characteristic frequency (kHz) 52 68 96 118
Strouhal number St 0.18 0.15 0.18 0.17
Reynolds number Re 473 786 1 140 1 770
Mach number M 0.317 0.445 0.505 0.617

0.0 0.2 0.4 0.6 0.8 1.0
fraction of jet power spent for powder heating
1E+3
1E+4
1E+5
h
e
a
t

f
l
u
x

p
o
t
e
n
t
i
a
l

[
W
/
m
]
water
Ar/H
2
(33/10 slm)
25 kW
Ar/H
2
(65/3 slm)
43.5 kW
N
2
/H
2
(235/94 slm)
200 kW
176 kW
84 kW
=
p
w
T
T
w
dT T k T S ) ( ) (
(16)
on the fraction of jet power available for particle heating. Here T
w
=
1000 K is the particle surface temperature, T
p
the plasma temperature
and k the thermal conductivity of plasma. The heat flux to particles is
given by the relation
) (
w p
T S d Nu Q = (17)
Fig. 21 Heat flux potential S in dependence on fraction of jet power available for
particle heating.
263
1 10 100
powder mass flow rate [kg/hour]
10
100
1000
10000
v
e
l
o
c
i
t
y

[
m
/
s
]
N
2
/H
2
(235/94 slm)
200 kW
water
176 kW
water
84 kW
Ar/H
2
(33/10 slm)
25 kW
Ar/H
2
(65/3 slm)
43.5 kW
where d
p
is the particle diameter and Nu the Nusselt number. The power
available for particle heating was determined as
) (
p
T h G P Q = (18)
where P is the torch power, h is its efficiency, G the mass flow rate
of the plasma gas and h(T
p
) the plasma enthalpy at temperature T
p
. The
curves were determined from the characteristics of different plasma torches
(see Table 2). The efficiency of utilizing the jet power for powder heating
is substantially higher for the water torch and very high heat fluxes to
the particles can be achieved. The effect of particle loading on plasma
velocity is illustrated in Fig. 22.
The velocity of plasma and, consequently, the velocity of particles
are reduced when the powder mass flow rate is increased. The load-
ing effect on the plasma velocity v
p
is represented by the curves de-
termined from the balance of the momentum flux in the plasma jet:
p exit
v M G Gv ) ( + = (19)
where G is the mass flow rate of plasma, v
exit
the mean plasma velocity
at the torch exit and M the powder mass flow rate. The curves were
determined from the characteristics of different plasma torches (see Table
2). The effect of particle loading on plasma velocity for water plasma
Fig. 22. The dependence of plasma velocity v
p
on powder mass flow rate.
264
TABLE 2. Typical parameters for plasma spraying torches. Values for gas mixture
Ar/H
2
for two ratios of gas mixture flow rates correspond to plasma torch Plasma-
Technik F 4. Values for N
2
/H
2
are given for torch PlazJet [27].
Plasma
medium
arc current
[A]
arc power
[kW]
mass flow rate
[g/s]
G/L
[kg/s.m]
enthalpy
[MJ/kg]
temperature
[K]
Ar/H
2
(65/3 slpm)

750

44

1.93

0.15

13.5

12 100
Ar/H
2
(33/10 slpm)

500

25

0.98

0.08

15.3

10 800
N
2
/H
2

(235/94 slpm)

500

200

5.0

0.1

24

6 200

water

300

84

0.20

0.004

252

15 800

water

600

176

0.33

0.006

320

17 500

torch is strong due to the very low plasma density. On the other hand,
this is partially compensated by the very high exit velocity of the plasma.
6. CONCLUSIONS
Principle of stabilization of arcs by the water wall can be utilized for
stabilization of long arc columns at very low flow rates of the plasma
gas. The torches based on this principle are sources of oxygenhydrogen
plasma with high temperatures and plasma flow velocities. The high content
of hydrogen results in high enthalpy and sound velocity of plasma. The
arc voltage and consequently the arc power are substantially higher compared
to gas-stabilized torches.
The basic parameter which determines arc and plasma characteris-
tics is the mass flow rate through the arc chamber. In the case of the
water plasma torch, this parameter cannot be controlled independently,
but is determined by the power balance of the radial transfer of energy
in the arc chamber. The part of energy spent for evaporation is sub-
stantially lower than the energy absorbed in the produced vapour and
this is the principle cause of low mass flow rates and high tempera-
tures found in experiments. High plasma velocities in combination with
low plasma densities lead to intensive interaction of the plasma jet with
ambient atmosphere and to a high level of turbulence. The mixing of
plasma with the ambient cold gas is characterized by very low time constants.
These properties of generated oxygenhydrogen plasma jet can be
utilized in plasma processing applications. The high efficiency of uti-
lizing plasma enthalpy for powder heating and high heat fluxes to the
particles enable large throughputs of powder, about one order higher
than for the common gas-stabilized torches [28]. Typical values of powder
265
throughput are 4050 kg/hour of ceramic powder and 80-100 kg/hour
of metal powder. Particle loading reduces substantially plasma veloc-
ity, but this disadvantage is compensated by very high plasma veloc-
ity. At present, the only industrial scale application is plasma spray-
ing, mainly for such tasks like large-area coatings, production of self-
supporting ceramics parts and powder processing. High level of turbulence,
short time constants of mixing and the chemical composition of gen-
erated plasma make the water-stabilized torch a potential candidate for
waste treatment [29,30].
The authors would like to thank to the Grant Agency of the Czech Re-
public for the support of this work under the projects No. 102/98/0813 and
No. 106/96/K245.
References
[1] H. Gerdien, and A. Lotz, Wiss. Veroffentlichungen Siemenswerk 2, 489
(1922).
[2] H. Gerdien, and A. Lotz, Z. Tech. Phys. 4, 157 (1923).
[3] H. Maecker, Z. f. Phys. 129, 108-122 (1951).
[4] F. Burnhorn, and H. Maecker, Z. f. Phys. 129, 369-376 (1951).
[5] R. W. Larentz, Z. f. Phys. 129, 343-364 (1951).
[6] R. Weiss, Z. f. Phys. 138, 170-182 (1954).
[7] F. Burnhorn, H. Maecker, and T. Peters, Z. f. Phys. 131, 28-40 (1951).
[8] B. Gross, B. Grycz, and K. Miklossy, Plasma Technology, Iliffe Books
Ltd., London (1968).
[9] B. W. Swanson, in Current Interrupti on of Hi gh-Vol tage Network
(K.Ragaller, ed.), pp. 137-179. Plenum Press, New York - London (1978).
[10] ADEP - Data bank and computer code, LMCTS URA 320 CNRS, Universit
de Limoges.
[11] H. Riad, J. J. Gonzales, and A. Gleizes, Proc. of ISPC 12, pp. 1731-
1736, Minneapolis, Aug. 21-25, 1995.
[12] V. Aubrecht, and B. Gross, J. Phys. D: Appl. Phys. 27, 95-100 (1994).
[13] M. Hrabovsk, M. Konrd, and V. Kopeck, in Heat and Mass Transfer
under Plasma Conditions (P. Fauchais, M. Boulos and J. van der Mullen,
eds.), Begell House, New York-Wallingford (U.K.), pp. 61 - 67 (1995).
[14] M. Hrabovsk, M. Konrd, and V. Kopeck, ESCAMPIG 96, Poprad,
Slovakia, Europhysics Conference Abstracts, Vol. 20 E, Part B, pp.
277-278 (1996).
[15] V. Sember, Proc. of 21
th
ICPIG, Vol. 2, pp.74-75, Bochum, Septem-
ber 19-24, 1993.
[16] V. Sember, in Heat and Mass Transfer under Plasma Conditions (P.
Fauchais, M. Boulos and J. van der Mullen, eds.), Begell House, New
York-Wallingford (U.K.), pp. 143-149 (1995).
[17] V. Sember, A spectroscopic Analysis of the Thermal Water Plasma Jet,
Ph.D. Thesis, Institute of Plasma Physics ASCR, Prague (1997).
[18] M. Hrabovsk, M. Konrd, V Kopeck, J. Kravrik, and P. Kubes, Proc.
of 17
th
Symp. on Plasma Physics and Technology, pp. 179-181, Prague,
June 13-16, 1995.
[19] Y. Yasutomo, K. Miyata, S. Himeno, T. Enoto and Y. Ozawa, IEEE Trans.
Plasma Sci. PS-9(1), 18-21 (1981).
266
[20] V. Sember, Proc. of 18
th
Symp. on Plasma Phys. and Tech., pp. 245-247,
Prague, June 17-20, 1997.
[21] J. Jenista, Proc. of 12
th
Conf. on Gas Disch. and Appl., 50-53, Greifswald,
Sept. 8-12, 1997.
[22] V Sember, Proc. of 14
th
ESCAMPIG, Dublin, August 26-29, 1998.
[23] G. R. Jones, and M. T. C. Fang, Rep. Prog. Phys., vol. 43, pp. 1415
- 1465 (1980).
[24] J. S. Lowke, and H. C. Ludwig, J. Appl. Phys., vol. 46, pp. 3352 -
3360 (1975).
[25] M. Hrabovsk, and P. Krenek, Proc. 11
th
Inter. Symp. on Plasma Chemistry,
Loughborough, (ed. J. Harry), vol. 1, pp. 315 - 320 (1993).
[26] M. Hrabovsk, M. Konrd, V Kopeck, and J. Hlna, Proc. of ISPC 12,
Minneapolis, Aug. 21-25, pp. 1627-1632 (1995).
[27] G. Irons. Proc. of Workshop on Industrial Applications of Plasma Chemistry,
Vol. B, pp. 53-65. Minneapolis, Aug. 25-26, 1995.
[28] P.Chrska, and M. Hrabovsk, Proc. of Int. Thermal Spray Conf. &
Exhib., pp. 81-85. Orlando, USA, May 28-June 5, 1992.
[29] V. Brozek, M. Hrabovsk, and V. Kopeck, Proc of ISPC 13, pp.1735-
1739. Beijing, August 18-22, 1997.
[30] V. Brozek, M. Hrabovsk, and V. Kopeck, Journ. of High Temp. Mat.
Process, in press.
267
Modeling of a D. C. Arc Plasma Torch with a Hydrogen-Argon Mixture....
Modeling of a D. C. Arc Plasma Torch
with a Hydrogen-Argon
Mixture as the Working Gas
Xi Chen, Peng Han and Lan Yu
Department of Engineering Mechanics, Tsinghua University, Beijing 100084, China
Modeling results are presented for the heat, mass and momentum transfer
in a D. C. arc plasma torch with 2% H
2
98% Ar (mole) as the working
gas. The demixing of the gas mixture is treated by using the combined-
diffusion-coefficient approach. Typical computed results of the flow,
temperature and concentration fields within the plasma torch are presented
for both the laminar and turbulent regimes. The modeling results show
that an appreciable demixing effect exists in the present plasma torch.
The hydrogen species is more concentrated in the high temperature region
located near the torch axis and downstream from the cathode tip.
1. INTRODUCTION
D. C. (direct current) arc plasma torches are involved in many applications,
such as the thermal plasma spraying, the thermal plasma chemical vapor
disposition of diamond or superconducting films, the thermal plasma
preparation of high-tech ceramic powders, the thermal plasma destruction
of wastes, and so on. A better understanding of the plasma torch
characteristics is required for the design and optimization of the related
processes. Numerous experimental and modeling studies were thus devoted
to the studies of the plasma torch. Especially, numerical modeling has
been widely employed to reveal the complicated phenomena associated
with the fluid flow and heat transfer within the arc plasma torch
[15].
Different gas mixtures are often chosen as the working gas instead
of a pure gas in order to improve the torch performance. For exam-
ple, it is well known that the addition of a small fraction of hydrogen
into argon as the working gas can remarkably improve the heat transfer
268
to injected particles for the thermal plasma spraying. Hence, it is in-
teresting to know what is the new special feature of the fluid flow and
heat transfer within a D.C. plasma torch when a gas mixture is used
as the working gas.
The usual practice in the modeling of a plasma torch with a mixed
gas (e.g. hydrogenargon mixture) as its working gas is that the mixing
ratio between different gases is regarded as everywhere uniform. Namely,
the mixed gas is treated as a pure-like gas, and thus the only effect
caused by the gas mixture is that its properties are different from those
for each of the pure gases forming the mixture. However, from the
view-point of the molecular theory of gases [6,7], there would appear
a demixing process between the two gas constituents because of the
existence of the great temperature gradients within the plasma torch.
Demixing is a phenomenon caused by the existence of concentration
and temperature gradients in the gas mixture, resulting in the partial
separation or the non-uniform spatial species-concentration distributions
of the different chemical species in the plasma. Generally the lighter
species (H
2
, H, H
+
etc. for the case of hydrogen-argon mixture) would
diffuse in the direction of the temperature gradient, while the heavier
species (Ar, Ar
+
, etc.) would diffuse in the opposite direction. This separation
process caused by the temperature gradient would proceed until its mass
flux is balanced by that of the ordinary diffusion deduced by the con-
centration gradient. Taking this demixing effect into account, the hy-
drogen species in a D. C. plasma torch would be more concentrated
in the region downstream from the cathode tip, where the plasma tem-
peratures are comparatively high. At the same time, there would be
a higher concentration of argon near the wall of the constrictor-tube
or the anode-nozzle, where the gas temperatures are lower. It is the
purpose of the present paper to study the demixing effect, i.e. to study
in what extent the species separation will be induced, and how the demixing
phenomenon will affect the torch performance. For this reason, two-
dimensional modeling is carried out to investigate the heat, mass and
momentum transfer within a D.C. plasma torch with a gas mixture of
98% (mole) argon and 2% hydrogen as the working gas. The modeling
is concerned with an arc plasma torch of a lower power level usually
employed in the laboratories in order to facilitate the comparison with
corresponding experimental data. The combined-diffusion-coefficient
approach proposed recently by Murphy [8-10] is employed to treat the
species diffusion in the gas mixture. This approach was shown to be
able to greatly simplify the treatment of diffusion in an ionized gas mixtures
in local chemical equilibrium (LCE) and to give fairly good predicted
results [10]. For the case of turbulent flow, the two-equation model
269
proposed by Launder and Spalding [11] for ordinary temperature flu-
ids is used.
2. COMPUTATIONAL DOMAIN
Figure 1 is a schematic sketch of the computational domain employed
in this study. Due to the irregular shape of a few boundaries of the
computational domain (electrode surfaces) of the plasma torch, the whole
domain is divided into two parts, i.e. the inlet region I and the arc region
II. The computation is conducted in these two regions, respectively,
at first and then a few iterations are performed between the computation
in the region I and that in the region II. In the region I, a uniformly
mixed argon-hydrogen entering tangentially into the torch as its working
gas. In the computation this oncoming gas flowrate is treated as a mass
source located at the region abcd in the z-r coordinate system. The
mixed gas from the entry then flows towards the arc region II. In the
region I, the gas is less heated so that the gas temperature is almost
constant except for the region near the cathode surface. The swirling
velocity at the torch inlet is calculated based on the mixed-gas flowrate
and the area of entry holes. From the modeling of the region I, the
velocity field within the whole region I can be obtained and thus the
upstream boundary conditions required for the modeling of the region
II are also known. The modeling for the region II is the kernel of the
present simulation. In this region, the computation is performed by solving
the simultaneous continuity, momentum, species, energy and electromagnetic
field equations (as well as the turbulent kinetic energy and its dissipation
rate equations for the turbulent flow regime).
III. GOVERNING EQUATIONS AND BOUNDARY CONDITIONS
The assumptions employed in this study include the axisymmetrical and
steady flow as well as the optically thin, LCE and LTE (Local
Fig. 1 The computation domain.
270
Thermodynamic Equilibrium) plasma. The inlet gas is an argon-hydrogen
uniform mixture. The following governing equations have been employed
in tmodeling:
The mass continuity equation
( ) ( )

r
r v
z
r v
r z
+ 0
(1)
The radial momentum conservation equation
v
v
r
v
v
z
P
r r r
r
v
r z
v
z
v
r
r
r
z
r r r z
+
_
,
_
,
+ +
_
,
1
]
1
2
(2)
The axial momentum conservation equation

v
v
r
v
v
z
P
z z
v
z r r
r
v
r
v
z
j B
r
z
z
z z z r
r
+
_
,
_
,
+ +
_
,
1
]
1
+ 2
1

(3)
The swirling momentum conservation equation

v
rv
r
v
rv
z z
rv
z r r
r
rv
r r
rv
r
r z
( ) ( ) ( ) ( ) ( )
+
1
]
1

1
]
1
+

1
]
1

1 2
(4)
The species conservation equation
v
f
r
v
f
z r r
r
f
r z
f
z
r
H
z
H
f
H
f
H
+
_
,
_
,
_
,
1
+

_
,
1
r r
r
M
r
M
_
,
z
M
z
M

_
,
_
,
1
r r
r
M
r z
M
z
Mh
H
Mh
H

(5)
_
,
_
,
1
r r
r
D
T
T
r z
D
T
T
z
ha
T
ha
T
The energy conservation equation

271

v
h
r
v
h
z r r
r
c
h
r z c
h
z
j j
U
r z
p p
r z
r
+
_
,
_
,
_
,
+
+
1
2 2
(6)
( ) ( )
+ +
_
,
1
]
1

1
]
1
5
2
1 k
eCp
j
h
z
j
h
r z
k
Cp
h h
f
z r r
r
k
Cp
h h
f
r
B
z r H Ar
H
H Ar
H
The electrical potential or the current continuity equation

1
0
r r
r
r z z
_
,
_
,

(7)
Here v v v
r z
, ,

are the radial (r-), axial (z-) and circumferential ()
components of the gas velocity vector. P and T are gas pressure and
temperature. , , , , k C
p
and h are the temperature- and concentration-
dependent density, viscosity, thermal conductivity, specific heat at constant
pressure, electric conductivity and specific enthalpy of the hydrogen-
argon mixture, respectively. U
r
in Eq. (6) is the radiation power per
unit volume of plasma. in Eq. (7) is the electric potential and is related
to the radial and axial components of the current density vector (j
r
and
j
z
) by ( ) j r
r
/ and ( ) j z
z
/ . B
, the component of
the magnetic induction intensity vector, is related with j
z
through the
following equation:
B
r
j d
o
z
r
0
(8)
in which
0
is the magnetic permeability.
In the foregoing equations, the species and energy conservation equations,
Eqs. (5) and (6), have been written by using the combined-diffusion-
coefficient approach [810]. If J
H
is used to express the diffusion flux
vector of hydrogen species, both the terms associated with the con-
centration gradient and the temperature gradient should be included in
the expression [810] of J
H
:
J
H X H
HA
T
X D T + (ln )
(9)
Where X
H
is the mole fraction of hydrogen in the argon-hydrogen mixture,
( )
_
,
5
2
k h h
eCp
f
r
j
f
z
j
B H Ar H
r
H
z
( ) [ ] ( ) [ ]

z
h h J
r r
r h h J
H Ar z H Ar r
1
272
while
X
A H
HA
x
m m
MM
D is the effective diffusion coefficient with respect
to the mole fraction gradient X
H
. m
A
and m
H
are the average molecular
weight of the heavy particles (without accounting for electrons) coming
respectively from argon and hydrogen, while M is the average molecular
weight of the whole gas mixture (with accounting for electrons). D
HA
x
and
DHA
T
are the combined ordinary-diffusion coefficient and the combined
temperature-diffusion coefficient (including thermal diffusion), respectively
[8-10]
. Since
X f M M
H H H
( / )
, Eq.(9) can be rewritten as
J
H f H M Mh H
HA
T
f M M D T + + (ln )
(10)
Where f
H
is the mass fraction of hydrogen species in the argon-hydrogen
mixture, while
M
H
is the average molecular weight of all the gas particles
coming from the original hydrogen species
f
,
M
and
Mh
are the
effective diffusion coefficients with respect to the gradients f
H
,
M
and
M
H
. It is easy to show that
( )

f X H
M M / ,
( )

M X H H
f M / and
( )

Mh X H H
f M M /
2
.
Correspondingly, the following terms associated with the species diffusion
have been added into the energy equation as its additional source
( ) [ ] ( )
( )
S h h
k
Cp
h h f
k h h
eCp
f
h H H Ar H Ar H
B H Ar
H

1
]
1

J j
5
2
(11)
besides the usual source terms concerning the Joule heating rate
(
( )
j j
r z
2 2
+ / ) and the radiation power per unit volume (U
r
). Where
h
H
and h
Ar
are the specific enthalpies (J/kg) of pure hydrogen and pure
argon. k
B
and e are the Boltzmann constant and the elementary charge.
The additional source term given in Eq. (11) is somewhat different from
273
that used by Murphy in his papers [8-10] in two aspects. The first is
that the negative sign in J
H
was missed, and the second is that the
last term of Eq.(11) was ignored by him.
The boundary conditions used in the modeling are as follows: Along
the axis (r = 0), symmetrical conditions are used. At the solid wall,
no velocity slip and no species flux are used. Zero potential and 500
K are given at the anode surface. At the upstream end of the cath-
ode in the region II, the current density is given by the total arc current
divided by the cross section area of the cathode. One-way boundary
conditions are used at the downstream boundaries. The surface tem-
perature of the cathode tip is restricted to be lower than 3000 K in
the modeling.
Of course, if a pure gas (e.g. argon) is concerned with in the modeling,
Eq.(5) and the additional source terms associated with species diffu-
sion in the energy equation (6) will disappear.
For the case of turbulent plasma flow, in order to calculate the turbulent
viscosity, two additional differential equations concerning, respectively,
the turbulent kinetic energy K and its dissipation rate are required
to be simultaneously solved [11], and the wall function method is used
to treat the boundary conditions at the solid surfaces. All the molecular
transport coefficients in the foregoing equations (2)(6) should be substituted
by the molecular plus turbulent transport coefficients. Namely,
+
t

h
p
t
t
k
C
+

Pr

f
A B
H
HA
x t
t
m m
MM
D
Sc
+

Where
T
is the turbulent viscosity, Pr
t
and Sc
t
are turbulent Prandtl
number and Schmidt number, and are chosen as 0.9 and 0.8, respectively.
The turbulent kinetic energy equation is as follows:
v
K
z
v
K
r z
K
z r r
r
K
r
G
z r K K
+
_
,
_
,
_
,
1
]
1
+
1
(12)
And the turbulent kinetic-energy dissipation-rate equation is:

v
z
v
r z z r r
r
r K
c G c
z r
+
_
,
_
,
_
,
1
]
1
+
1
1 2
( )
(13)
274
where
1
1
1
1
1
]
1
,
_
+ +
,
_
+
,
_
+
,
_
+
,
_
+
,
_
2
2 2 2 2 2
2 2 2
z
v
r
v
r
v
r
v
z
v
r
v
z
v
r
v
G
r z
r z r
t

(14)
and
K
t
K
+

+
t
The turbulent viscosity is determined by
t
c K
2
/ . The model
constants used in this turbulent model, c c c
k 1 2
, , ,

and
, are chosen
as 1.44, 1.92, 0.09, 1.0 and 1.3, respectively.
Special attention has to be paid to the treatment of the near-wall
region adjacent to the electrodes. Because the physical mechanisms
in the plasma sheath are very complicated and quite different from the
arc-column region, and because so far our understanding about the
phenomena inside the sheath is considerably incomplete, we employ a
relatively simple but rather powerful approach to calculate the elec-
tric conductivity of the near-wall region in our simulation. Since the
non-LTE state is expected to prevail in the near-wall region, we as-
sume that the two-temperature plasma model can be employed in the
region between a grid point C at the electrode surface and its neighbor
0 0.01 0.02 0.03 0.04
Axial Distance (m)
-0.004
-0.003
-0.002
-0.001
0
0.001
0.002
0.003
0.004
R
a
d
i
a
l
D
i
s
t
a
n
c
e
(
m
)
0 0.01 0.02 0.03 0.04
Axial Distance (m)
-0.004
-0.003
-0.002
-0.001
0
0.001
0.002
0.003
0.004
R
a
d
i
a
l
D
i
s
t
a
n
c
e
(
m
)
18 15
12
4
9
17 16 14 13
11
10
11
Fig. 3 The isotherms (10
3
K) for the case of laminar flow regime.
Fig. 2 The stream lines for the case of laminar flow regime.
275
grid point P located in the region outside the sheath. The electron temperature
at the point P is assumed to be equal to the heavy-particle tempera-
ture, while the electron temperature at the point C is assumed to be
a fixed value (e.g. 8500K). On the other hand, the heavy-particle temperature
is assumed to vary linearly from the point P to the point C, while the
temperature at the point C is not higher than 3000 K for the cathode
and equal to 500 K for the anode. The mean electric conductivity in
the near-wall region from the point C to the point P is calculated from
the current continuity condition by using the following equation
( )

x dx
x
/ / 1
0
(15)
in which x is the distance between the point C and the point P. Since
always assumes an infinite value, the currents can flow pass the sheath
region and the arc can be maintained.
4. RESULTS AND DISCUSSIONS
The finite difference method and the SIMPLE-like (SIMPLEC) algorithm
[12] have been employed to solve the set of governing equations. 122
(axial) X 22 (radial) grid points are used in the computation.
In the modeling, the mixture of 2% hydrogen (mole) and 98% ar-
gon is chosen as the working gas. The pure argon is also used in the
simulation in order to compare the modeling results with those obtained
for the argon-hydrogen mixture. The property databases ( , , , , k C
p
and
h) of the pure argon, the 98% argon (mole) 2% hydrogen and the
90% argon (mole) 10% hydrogen plasmas have been used in the modeling.
The temperature space of the property tables is 100 K. Linear inter-
polation procedure is employed for calculating the plasma properties
for any given temperature and hydrogen concentration. The arc cur-
rent is kept to be 221 A for all the cases presented here. Both the
laminar and turbulent flow regimes are studied in the modeling. For the
case of laminar flow regime, the flow rate is 0.53 m
3
/h, while 2.1 m
3
/
h is taken for the case of the turbulent flow regime.
Some typical calculated results are shown in Figs. 211. Figure 2-
4 show the computed stream lines, isotherms and hydrogen mass-fraction
distribution within the D. C. arc plasma torch for the case of lami-
nar flow regime. Corresponding computed results for the case of turbulent
flow regime are shown in Figs. 57. From these computed results it
is seen that although the computed results shown in Figs. 24 for the
case of laminar flow regime are generally similar to their counterparts
276
for the case of turbulent flow regime shown in Figs. 57, the details
of those contours are somewhat different. Due to the heating of the
gas in the arc region, the space between the stream lines near the torch
axis is much greater than that near the tube wall. The highest tem-
perature appears at the axis and locates at a short distance downstream
from the cathode tip. The maximum value of the hydrogen mass fraction
is 23 times as great as its inlet value, showing an appreciable demixing
effect. In comparison with the case of laminar flow, the stream lines
are a little more appreciably shifted towards the wall, the highest temperature
is about 2000 K higher and the location of the arc attachment at the
anode surface appears at a more downstream region, while the maximum
value of hydrogen mass fraction is a little greater for the case of the
0 0.01 0.02 0.03 0.04
Axial Distance (m)
-0.004
-0.003
-0.002
-0.001
0
0.001
0.002
0.003
0.004
R
a
d
i
a
l
D
i
s
t
a
n
c
e
(
m
)
2.6
2.2
1.8
1.4
1.0
0.6
0.2
0.6
1.0
0 0.01 0.02 0.03 0.04
Axial Distance (m)
-0.004
-0.003
-0.002
-0.001
0
0.001
0.002
0.003
0.004
R
a
d
i
a
l
D
i
s
t
a
n
c
e
(
m
)
Fig. 4 The mass fraction of hydrogen (10
3
K) for the case of laminar flow regime.
Inlet value: 1.0210
3
.
Fig. 5 The stream lines for the case of turbulent flow regime.
Fig. 7 The hydrogen mass-fraction () contours for the case of turbulent flow regime.
Inlet value: 1.02.
Fig. 6 The isotherms ( K) for the case of turbulent flow regimes.
0 0.01 0.02 0.03 0.04
Axial Distance (m)
-0.004
-0.003
-0.002
-0.001
0
0.001
0.002
0.003
0.004
R
a
d
i
a
l
D
i
s
t
a
n
c
e
(
m
)
20 18 16
14
12
10 4
0 0.01 0.02 0.03 0.04
Axial Distance (m)
-0.004
-0.003
-0.002
-0.001
0
0.001
0.002
0.003
0.004
R
a
d
i
a
l
D
i
s
t
a
n
c
e
(
m
)
0.6
1.0
1.4
1.8
2.2
2.6
3.0
1.0
0.6
0.2
277
0.01 0.02 0.03 0.04 0.05
100
150
200
250
300
350
400
450
Axial Velocity
A
x
i
a
l

V
e
l
o
c
i
t
y

(
m
/
s
)
Axial Distance (m)
12
13
14
15
16
17
18
19
T
e
m
p
e
r
a
t
u
r
e

(
1
0

3

K
)
Temperature
0.01 0.02 0.03 0.04 0.05
100
150
200
250
300
350
400
450
Axial Velocity
A
x
i
a
l

V
e
l
o
c
i
t
y

(
m
/
s
)
Axial Distance (m)
12
13
14
15
16
17
18
19
T
e
m
p
e
r
a
t
u
r
e

(
1
0

3

K
)
Temperature
turbulent flow.
Figure 8 and 9 plot the computed variations of the axial velocity and
the gas temperature along the torch axis for the laminar and turbu-
lent regimes, respectively. The axial variations of the gas temperature
are quite similar for both the laminar and the turbulent flow regimes,
but the maximum temperature is 2000 K higher for the turbulent re-
gime. Although the flowrate of the working gas for the turbulent flow
regime is about 4 times as great as that for the case of laminar flow
regime, the maximum axial velocity at the torch axis for the case of
the turbulent flow regime is only about 130% greater than that for the
case of laminar flow regime due to a fatter velocity profile for the former.
The highest axial velocity for the turbulent flow regime appears at a
somewhat more downstream location in comparison with that for the
laminar flow regime. The averaged value of the axial gas velocity at
the torch exit is much greater than that near the torch inlet due to the
intense gas heating. Hence the swirl number assumes a very small value
at the torch exit, although its value is rather high at the torch inlet.
Figure 10 compares the radial distributions of the hydrogen mass
fraction at 5 different axial locations for the case of laminar flow regime.
It is seen that significant demixing phenomenon appears within the plasma
torch, showing appreciable non-uniform radial hydrogen concentration
distributions after the working gas has been heated by the arc. The
hydrogen species is appreciably concentrated towards the high tem-
perature region near the torch axis. The maximum value of the hydrogen
mass fraction appears at the torch axis and at the location 0.02 m
downstream from the cathode tip.
Figure 11 compares the computed variations of the gas tempera-
ture along the torch axis for three different cases: (A) Taking the 2%
Fig. 8 The computed variations of the axial velocity and the temperature along the
torch axis for the case of laminar flow regime.
Fig. 9 The computed variations of the axial velocity and the temperature along the
torch axis for the case of turbulent flow regime.
278
0.01 0.02 0.03 0.04 0.05
12
13
14
15
16
17
18
19
2% Hydrogen With Demixing
2% Hydrogen Without Demixing
100% Argon
T
e
m
p
e
r
a
t
u
r
e

(
1
0

3

K
)
Axial Distance (m)
Fig. 10 The computed hydrogen mass fraction (10
3
) distributions at 5 different axial
locations. Laminar flow regime.
Fig. 11 Comparison of the computed temperature distributions along the torch axis
for three different cases. Laminar flow regime.
-0.004 -0.002 0.000 0.002 0.004
0.0
0.5
1.0
1.5
2.0
2.5
3.0
Inlet
0.013
0.017
0.035
x=0.02
M
a
s
s

F
r
a
c
t
i
o
n

(

1
0
-
3

)
Radial Distance (m)
hydrogen 98% argon as the working gas and taking the demixing effect
into account; (B) Taking the 2% hydrogen 98% argon as the working
gas but the demixing effect is ignored; and (C) Taking the pure ar-
gon as the working gas of the plasma torch. Examination of the computed
axial variations of the gas temperature in Fig. 11 reveals that the demixing
effect appreciably changes the temperature distribution and enhances
the value of the highest temperature. Since only small fraction of
hydrogen is added into the working gas, the properties of the
2% hydrogen 98% argon mixture is not quite different from those
of the pure argon. Hence, the computed temperature distribution for
the case (B) is almost identical to that for the case (C). However, if
the demixing effect is included, the energy transport and thus the temperature
distribution are appreciably affected.
In the present modeling, an important error source is that the plasma
flow has been assumed to be axisymmetrical, and thus all the work-
ing gas will be heated effectively by the arc. A three-dimensional effect
always exists in an actual plasma torch, and a part of the working gas
may flow in a roundabout way without passing the arc region. The
applicability of the two-equation turbulence model to the plasma torch
is questionable. The non-LTE effect may be important, especially in
the region near the torch wall. More sophisticated study is highly desired
to clarify the effects due to these complicated factors.
ACKNOWLEDGMENT
This work was supported by the National Natural Science Foundation of China (No.
59676011) and the Doctorate Thesis Fund of the National Education Department of
279
China. The authors would like to thank Dr. A. B. Murphy whose property database
for the argon-hydrogen mixture has been used in this study.
REFERENCES
1 A. H. Dilawari, etal, ISIJ International, 30, No.5, 381-389 (1990).
2 D. A. Scott, et al, Journal of Applied Physics, 66, No.11, 5232-5239
(1989).
3 R. Westhoff and J. Szekely, Journal of Applied Physics, 1991, 70 (7),
3455-3466.
4 C. George and E. Pfender, In: Numerical Simulation of a Multi-Com-
ponent Non-Transferred Constricted DC Arc, Proceedings of the 13th
International Symposium on Plasma Chemistry, Vol.1, Beijing University
Press, Beijing (1997), pp.314-319.
5 M. Suzuki, et al, Plasma Chemistry and Plasma Processing, 16, No.4,
399-415 (1996).
6 J. O. Hirschfelder, et al, Molecular Theory of gases and Liquids, Chapter
8, John Wiley & Sons Inc., New York (1954).
7 S. Chapman and T. G. Cowling, The Mathematical Theory of Non-Uniform
Gases, Chapter 14, 3rd Edition, Cambridge University Press, Cam-
bridge (1990).
8 A. B. Murphy and C. J. Arundell, Plasma Chemistry and Plasma Process-
ing, 14, No.4, 451-591 (1994).
9 S. C. Snyder, et al, Physical Review E, 52, No.3, 2999-3009 (1995).
10 A. B. Murphy, Journal of Physics D: Applied Physics, 29, No.3, 1922-
1932 (1996).
11 B. E. Launder and D. B. Spalding, Computer Methods in Applied Mechanics
and Engineering, 3, 269-289 (1974).
12 S.V. Patankar, Numerical Heat Transfer and Fluid Flow, Hemisphere,
Washington, D.C, (1980).
280
The Modified Balance Method of
Calculating Characteristics of Near-Anode
Processes
Yu.V. Bruevich, I.P.Nazarenko and I.G. Panevin
The lifetime of arcjet thrusters and MPD thrusters is determined in many
cases by the cathode efficiency which depends on the thermal condition
and the kind of plasma contact with its surface (spot or diffuse discharge).
In the case of the diffuse discharge, the thermal condition of the cathode
is connected with its geometry, the cooling rate and the plasma parameters
in the near-cathode region of the arc discharge. Therefore, the problem
of plasma parameters determination in the arc near-cathode region with
diffuse discharge is of considerable interest.
1. THE PHYSICAL MODEL OF ARC NEAR-CATHODE REGION
In most cases, the two-layer model of the arc near-cathode region [1-
3] is used for high pressure discharge (P ~10
5
Pa). This model, shown
in Fig. 1, includes the collisionlless layer I and the collisional layer II.

Fig. 1.
discharge
plasma
cathode
281
Modified Balance Method of Calculating Characteristics...
The collisionless layer (the space charge layer) contacts with the cathode
surface. Usually it is assumed that its thickness d
0
is less than the electron
free path length and there are no interactions between plasma and particles
within the layer I. Moreover, it is considered that in the layer I the positive
volume charge is concentrated in the main and a large part of the cathode
potential drop U
c
is realized. The electrical field which exists in the layer
I accelerates the emission electrons moving from the cathode surface
towards the discharge plasma, and also the plasma ions moving to the
cathode surface. But the electric field in this layer brakes the plasma
electrons which direct to the cathode surface, and only the fast electrons
whose energy is enough to get over the braking barrier U
c
are able to
reach the cathode surface. The total current density j=j
e
em
+j
i
+j
e
pl
and
its components (the plasma ion current density j
i
, the plasma electron
current density j
e
pl
and the emission electron current density j
e
em
) are
invariable within the collisionless layer. The emission electrons which
are accelerated in the layer I have on the boundary between the layers
I and II the energy (eU
c
+
5
/
2
kT
w
). Here T
w
- the cathode surface temperature,
e - the electron charge, k - the Boltzmanns constant. As for the arc
discharge in the arcjet eU
c
>>kT
w
, it is possible to consider the emission
electrons as a monoenergetic stream with energy eU
c
. This stream brings
in the collisional layer the impulse stream and the energy flux, where
n
e
em
and V
e
em
- the emission electron density and their velocity on the
boundary between the layers I and II, m
e
- electron mass. These impulse
stream and energy flux pass to the plasma particles in the layer II.
The collisional (ionization) layer thickness is much bigger than the
free path length of the plasma particles. Within the layer II the ion current
forms, the emission electron stream relaxes and the heavy particles
temperature changes from the temperature T
w
on the boundary with the
layer I to the temperature close to the electron temperature at the boundary
with discharge plasma T
e
. Because of the difference between the heavy
particles and electron temperatures and also because of the ionization
equilibrium displacement, the plasma composition in the layer II dif-
fers from the equilibrium composition.
The total current density j is also invariable within the collisional
layer. However, as it is followed from Fig. 1, the current densities j
e
em
and j
i
within the layer II change their values, and the current density
j
e
pl
changes the value and the sign: at the boundary with the layer I the
current density j
e
pl
is provided with the fast plasma electrons and is directed
to the cathode surface, on the boundary with discharge plasma the current
density j
e
pl
is caused by the motion of all plasma electrons and is di-
rected to the discharge plasma.
The energy losses with the streams of the plasma particles which leave
282
the layer II through the boundaries are compensated for by the energy
supply with the stream of emission electrons.
There two main calculation methods for determination of the near-
cathode region characteristics. The first method, which we shall name
the exact method, is based on the differential equation system solution
and allows us obtain information about the cathode potential drop U
c
,
the heat flux q
0
which passes from the near-cathode plasma to the cathode
surface and also the information on the distribution over the collisional
layer thickness of temperature and the concentrations of the particles,
the electric field strength E, plasma potential difference j, and so on.
The second method which is used for the estimation of U
c
and q
0
values proceeds from the integral balance correlations.
2. THE EXACT CALCULATION METHOD
As the collisional layer thickness in the diffuse discharge is much smaller
than the radial dimension of the discharge, the equation system for the
layer II is written in the one-dimensional approach. Besides, it is suggested
that the emission electron current density may be written with the help
of RichardsonDashmans equation with Schottkys correction.
The equation system which takes into account the processes of re-
laxation of emission electrons, the formaton of the ion current and the
ionization equilibrium displacement and so on, was suggested in [3]:
5
2
5
2
5
2
5
2
kn U
dT
dx
kn V
dT
dx
d
dx
dT
dx
j E n e kT
W kT
dj
dx
d j
dx
e
e
e e
e
e
e
e
e i e
e e
e
+ =

+ +
( )
(1)
5
2
k n U V n U V
dT
dx
d
dx
dT
dx
W j E
a a i i e e i
+ + + =
F
H
G
I
K
J
+ + b g b g
(2)
W
m
M
kn v v T T
e
e
i
e ea ei e
= + 3 b gb g
(3)
en E
dP
dx
m n V
k k
V V
m n V
k k
V V
e
e c e ea
add ea
e a
e e ei
add ea
e i

b g b g
= 0
(4)
283
en E
dP
dx
m n V
k k
V V
M
n v V V
e
i e e ei
add ei
e i i ia i a

b g b g
2
0 =
(5)
m
e
n
e
V
e
+ M
i
n
i
V
i
+ M
a
n
a
V
a
= 0 (6)
n
e
= n
i
; (n
e
em
<< n
e
) (7)
P = k(n
e
T
e
+ n
i
T + n
a
T) = const (8)
P = kn
e
T
e
(9)
P
i
= kn
i
T (10)
( ) ( ) n T n n T n
e e e a e e
=
3
(11)
( )
n
e
dj
dx
d n V
dx
e
i i i
= =
1
(12)
j
e
pl
= -en
e
V
e
(13)
j
i
= en
i
V
i
(14)
j
e
em
= -en
e
V
e
em
(15)
j = j
e
pl
+ j
i
+ j = const (16)
j j n Q n Q n Q dx
e
em
e
em
a ea i ei e ea
x
w
= + +
F
H
G
I
K
J
L
N
M
O
Q
P
z
exp
. . . 1 1 1 1 1 1
0
1
2
(17)
pl
= m
e
n
e
+ M
i
n
i
+ M
a
n
a
(18)
d U d m n V
pl
e e
em
e
em
=
2
2
c h
e j
(19)
E
d
dx
=

(20)
284
d
dx
kT n Q
e e ee
=
F
H
G
I
K
J
1
2
3
2
1 1
.
(21)
dV
dx
V V n Q V V n Q
e
em
e
em
e e ee e
em
i i ei
= +
1
2
1 1 1 1

. .
c h c h (22)
Here T
e
and T are the electron and heavy particles temperatures;
e
and
are the heat conductivities of electrons and heavy particles; n
e
, n
i
,
n
a
are the concentration of the electrons, ions and atoms; n
e
em
is the emission
electron density; V
e
, V
i
, V
a
are the directed velocities of the electron,
ion and atom motion; V
e
em
- the emission electron velocity;
=
m V
e e
em
2
- the kinetic energy of emission electrons; M
i
and M
a
- the atom and
ion masses; j - the plasma electric potential;
i
- the ionization potential;
U - the directed velocity of motion of plasma as whole in the layer II
due to the transfer of emission electrons impulse stream to the plasma
particles; n
e
- the net electron (ion) generation rate;
ea
,
ei
,
ia
, - the
frequencies of electron-atom, electron-ion and ion-atom elastic
collisions; Q Q Q
ea ei ee
1 1 11 1 1 . . .
, , - the average cross sections of impulse transfer
by elastic collisions of particles; k
ea
, k
ei
, k
add
are kinetic coefficients;
(T
e
), (T
e
) are the ionization and recombination constants; P
e
and P
i
are the partial pressures of electrons and ions.
The influence of the collisionlless layer on the equation system
(1)-(22) is taken into account in the boundary conditions: the cathode
potential drop is represented by the potential drop in the layer I
U
c
. Besides, because of the small thickness of the collisionlless layer
d
0
, the parameters at the boundary between the layers I and II and
the cathode surface are identified.
The boundary conditions for the system (1)(22) are written as
follows:
a) on the boundary between the layers I and II (x=0)
T = T
w
; j = U
c
; U = 0; e = eU
c
; j
i
= j
iw
;
dT
dx
dT
dx
dT
dx
j U
j j
eU
kT
j AT
e
kT
W
e
e
W
e
pl
W
c
e
pl
w
e ch
c
ew
e
em
w
ef
w
w
= =
=
F
H
G
I
K
J
=
F
H
G
I
K
J
; ;
exp ; exp
2
(23)
285
b) on the boundary with the discharge plasma (x = d)
j
e
em
= 0;
dT
dx
dT
dx
C
dn
dx
C
e e

= = =
1 2
;
; (24)
Here A - Richardsons constant,
ef
- the effective material work function,
indices w and - mean the conditions on the boundary between
layers I and II and on the boundary with the discharge plasma, accordingly.
The value of the heat flux q
0
which is passed into the cathode body
is determined from the energy balance on the cathode surface
j
kT
e
j U
dT
dx
q T j
kT
e
e
pl
W
ew
ef iw c i ef
w w e
em ew
ef
w
+
F
H
G
I
K
J
+ + + =
= + + +
F
H
G
I
K
J
5
2
5
2
0
4

b g
(25)
where e
w
- the blackness degree of the cathode surface, s - the Stephan-
Boltzmanns constant.
As shown in [3], the solution of the equation system (1)(22) is a
relatively labour-consuming procedure and requires the large expenditure
of the calculation time. Therefore, the set of the received solutions is
limited: we calculated the near-anode processes characteristics for the
arc discharge burning in argon (P = 10
5
Pa) on the tungsten cathode
for the different values of T
ew
when the values of T
w
or q
0
remain in-
variable.
3. THE BALANCE CALCULATION METHOD
The balance calculation method which is used usually is approximate
and does no take into account some of the processes taking place in
the collisional layer. In this method, it is assumed that the electron and
heavy particles temperatures are constant in the layer II. Besides, it is
suggested that the energy flux j
e
em
U
c
carried in the layer II by the emission
electrons is spent on the compensation for energy losses with the ion
current
j
kT
e
iw i
w
+
F
H
G
I
K
J
5
2
, with the fast plasma electrons current and with
total current on the boundary with the discharge plasma
j
kT
e
e
5
2
.
286
The equation system in the balance method which is used usually
looks as follows [4]:
j = j
e
em
+ j
i
+ j
ew
pl
(26)
j AT
e
kT
e
em
w
ef
w
=
2
exp

(27)

ef
=
w
- Dj (28)
=

3 75 10
4
1
2
, E
w
(29)
( ) E A U j
w i c i
= 5700
1
4
1
4
1
2
(30)
j j
eU
kT
en V eU
kT
e
em
e
c
e
e e c
e
w ch
w w
=
exp exp
4
(31)
j j
enV
i i
i i
ch
= =
4
(32)
P = k(n
e
T
e
+ n
i
T
i
+ n
a
T
a
) (33)
n
e
= n
i
(34)
b(T
e
)n
e
n
a
= a(T
e
)n
e
3
(35)
where A
i
- the atom weight, V
kT
m
e
e
e
=
8
,
V
kT
m
i
i
=
8
,
w
- the cath-
ode material work function.
The energy balance equations on the cathode surface and for the collisional
layer as a whole are solved in common with the equations (26)(35)
q T j
kT
e
j U j
kT
e
w w e
em w
ef
i c ef e
pl e
0
4
1
5
2
5
2
+ + +
F
H
G
I
K
J
=
= + +

b g
(36)
287
j U
kT
e
j
kT
e
j U
kT
e
j
kT
e
e
em
c
w
i i
e
pl
c
e e
+
F
H
G
I
K
J
= +
F
H
G
I
K
J
+
+
F
H
G
I
K
J
+
5
2
5
2
5
2
5
2
(37)
The equation system (26)(37) was solved for such conditions as the
system (1)(22). The comparison of the calculation results which was
obtained by the exact and balance methods in [3] showed that the cathode
potential drop values U
c
calculated with the help of the balance method
exceed by 12 V the values of the cathode potential drop which were
determined with the exact method. The difference of the values of the
heat flux q
0
which is passed into the cathode body is more essential.
The values q
0
which were determined by the balance method can be twice
as high as the values q
0
calculated by the exact method.
Since the balance method is essentially less labour-consuming than
the exact method, it is expedient to modify this method which would
make it possible to develop a relatively simple procedure and increase
the accuracy of determination of the values U
c
and q
0
.
4. THE MODIFIED BALANCE METHOD
The considerable difference in the values of q
0
calculated by the different
methods are associated with the fact that the conventional balance metjod
does not take into account some special features of the collision layer
appearing when calculating its characteristics by the exact calculation
method. For an arc discharge burning in argon (P ~ 10
5
Pa) on a tungsten
cathode, these peculiarities may be written as follows:
1. the electron temperature near the cathode surface T
ew
exceeds the
electron temperature at the discharge plasma boundary T
e
by ~2000
K;
2. the heavy particle temperature T near the cathode surface is equal
to the cathode surface temperature T
w
and at the discharge plasma boundary
T T
e
;
3. the heat flux
dT
dx
determined by the heavy particle heat con-
ductivity changes only slightly and is approximately equal to 410
3
W/
cm
2
;
4. the current density on the cathode surface j
iw
is determined by the
diffusion in the near-cathode layer and may be written as j
iw
= B(T
e
)j
ich
288
where B(T
e
) is the coefficient which depends on temperature T
ew
; here
j
ich
is the chaotic ion current density;
5. the near-cathode plasma composition is non-equilibrium, the maximum
possible electron generation rate values at the temperatures T
ew
and T
e
make up about 0.95 from the value
max
n
e
which is the maximum value
at the corresponding temperatures T and T
e
;
6. under the effect of the emission electrons, the plasma motion as
a whole in the collisional layer is directed towards the discharge plasma
and through the boundary with the discharge plasma the energy flux
U
5
/
2
kT
e
(n
e
+ n
i
+ n
a
) goes out.
Taking these peculiarities into account, the equation system for the
modified balance method has the following form:
j = j
e
em
+ j
iw
+ j
ew
pl
(38)
j AT
e
kT
e
em
w
ef
w
=
2
exp

(39)
j
ef
= j
w
- Dj (40)
=

3 75 10
4
1
4
, E
w
(41)
( ) E A U j
w i c i
= 5700
1
4
1
4
1
2
(42)
j j
eU
kT
en V eU
kT
e
pl
e
c
e
e e c
e
w ch
w w
=
exp exp
4
(43)
j B T j
enV
B T
i w e w i
i i
e
ch
= = ( ) ( )
4
(44)
P = k(n
e
T
e
+ n
i
T
i
+ n
a
T
a
)
w
= k(n
e
T
e
+ n
i
T
i
+ n
a
T
a
)
(45)
T
= T
e
(46)
n
e
= n
i
(47)
T
e
= T
ew
2000 (48)
n n n n
e e a e
=
3
(49)
289
,
max
n n
e e
= 0 95 (50)
( )
U
eU
m
m n
m n M n M n
c
e
e e
em
e e i i a a
=
+ +

2
(51)
The energy balance equations in the modified balance method may
be written as follows:
q T j
kT
e
dT
dx
j U
kT
e
j
kT
e
w w e
em w
ef
i w c i ef
w
e
pl e w
ef
w
0
4
5
2
5
2
5
2
+ + +
= +
+ + +
+ +

(52)
j U
kT
e
dT
dx
j U
kT
e
j
kT
e
j
kT
e
j U kT n n n
e
em
c
w
e
pl
c
ew
iw i
w
e
e
pl
c e i a
w
+
F
H
G
I
K
J
= + +
F
H
G
I
K
J
+ +
F
H
G
I
K
J
+
+ + + + +

5
2
5
2
5
2
5
2
5
2

b g
(53)
The results of calculations using the equation system (38)(53) for
the arc discharge burning in argon (P = 10
5
Pa) on the tungsten cath-
ode are presented in Fig. 2 and 3.
The modified balance method, characterised by a considerably lower
labour content in comparison with the exact method, was uased to calculate
the range of change of cathode surface temperature T
w
from 3500 to
4000 K, the near-cathode electron temperature T
ew
from 15000 to 23000
K. Figure 2 and 3 shows the dependences of the cathode potential drop
U
c
and the heat flux q
0
on the total current density j.
The graphs show the influence on the values U
c
and q
0
of param-
eters such as the temperature T
w
and T
ew
. Figure 2 and 3 also compare
the values U
c
and q
0
calculated by the exact method. The sufficient
correspondence of the results obtained by the exact and modified bal-
ance methods allows us to apply the modified balance method for calculating
the near-cathode region characteristics in argon.
The application of the modified balance method for other ranges of
P, T, T
e
in argon or the arc discharge burning in other gases requires
additional calculations by the exact method.
290
Fig.1
Fig.2
References
1. T.H. Lee, A.N.Greenwood and W.D.Breingan, A self consistent model
for the cathode region of a high pressure arc, Proc. 7th Intern. Conf.
Phenomena Ionized Gases. Belgrade, 1965, p.670-680.
2. K.C.Hsu and E.Pfender, Analysis of the cathode region on a free-burning
high intensity argon arc, J.Appl.Phys., vol.54,N7, 1983,p.1769-1777.
3. A.M.Zimin, I.P.Nazarenko, I.G.Panevin, V.I.Khvesyuk. Mathematical modeling
of cathode processes. Nauka, Novosibirsk (1993).
4. Near-electrode processes and erosion of plasmatron electrodes / M.F.Zhukov
(editor) Nauka, Novosibirsk (1997).
291
Application of Heat Diffusion Theory to Cold Electrode Erosion
Application of Heat Diffusion Theory to
Cold Electrode Erosion for Nonstationary
Arc Spots
A. Marotta and L. I. Sharakhovsky
1
Instituto de Fsica Gleb Wataghin, Universidade Estadual de Campinas,
Unicamp 13083-970, Campinas, So Paulo, Brazil
1. INTRODUCTION
The problem of erosion is most actual for electric arc heater (EAHs)
electrodes, especially the cathode, because of its higher erosion than
the anode. Operating in the regime of autothermoionic emission,
characterized by extremely high current densities and heat fluxes,
these electrodes can be used only under the condition of rapid
displacement of the arc spot. To avoid fast destruction of the electrode,
a magnetic field or a gas vortex is usually used.
Extensive efforts have been made by many authors to understand the
physics underlining the erosion of copper electrodes (see [1] and ref-
erences therein). In order to establish the dominant mechanism of erosion
in oxidized cathode surfaces, Guile et al [2] investigated different physical,
chemical, and electrochemical processes, using an Arrhenius-type equation
of the form [2]
d d m A G kNT
e
/ exp( / ), (1)
where dm/d is the mass erosion rate, A is the rate factor, G
e
is the
effective erosion activation energy, k is the Boltzmanns constant, N is
the Avogadros number and T is the electrode surface temperature. The
value of G
e
was found by Guile to be close to the activation energy
for the interatomic bond loosening (fusion) of the pure metal copper
electrode material, rather than to the activation energy of other processes,
such as oxide or metal vaporization, oxide decomposition, etc. The Guile
1
Permanent address: The Luikov Heat & Mass Transfer Institute, P. Brovki
street, 15, 220072, Minsk, Republic of Belarus
292
conclusion was that the fusion of the solid electrode metal plays the
dominant role in the erosion of oxidized copper cathodes.
Recently, a thermophysical macroscopic model for cold cathode erosion
has been presented in [1,35]. The model is based on the cathode fu-
sion mechanism proposed by Guile. The erosion model gives the spe-
cific mass erosion rate g (kg C
1
) as function of the operating parameters,
such as the current I, the arc displacement velocity v and the electrode
surface temperature T. Besides these, there are parameters character-
istic for the erosion process, the minimal erosion value (accounting for
the microerosion process) g
0
, and the effective erosion enthalpy h
ef
.
The parameter h
ef
accounts for all processes ending up in electrode erosion.
A parameter similar to h
ef
is also found in the technology of heat shielding
materials for space systems moving in the atmosphere with hypersonic
speed. There, the effective enthalpy is obtained as the ratio of the heat
flux, which enters the body, to the mass flow rate of the ablated material.
The effective enthalpy is a convenient approach to deal with the complex
conditions of unsteady and combined (thermal, electromagnetic,
hydrodynamic, etc.) effects that occur in cold electrode erosion phenomena.
The kinetics of these processes is very complex, and their precise calculation
is hardly feasible. Here, we will obtain the effective enthalpy of erosion
semi-empirically: the value of the erosion will be measured experimentally,
and the heat of erosion will be calculated theoretically in accordance
with the present thermophysical model [1].
New characteristic arc spot parameters, the volt equivalent of the
cathode arc spot heat flux U and the effective arc spot current density
j, have been introduced in the thermophysical theory in the place of the
usual, the near cathode voltage drop U and the arc spot current density
j, respectively. Electrical and optical methods are often employed to measure
U and j, for fixed cathode arc spot attachments [6]. In a cold cath-
ode, the arc spot has a complex unstable structure, consisting of an
agglomeration of many fast moving microspots, as demonstrated by
Rakhovsky [6]. For these conditions, the measurement of U, and es-
pecially of j, by the above mentioned methods, presents serious diffi-
culties. This explains the extremely high scattering of experimental data
on cold cathode arc spot current densities, obtained by many authors
(up to 5 orders of magnitude, see e.g., [6]).
In the thermophysical approach, U and j are measured by thermal
methods. This permits to obtain, in a straightforward manner, the value
of the arc spot heat flux density q
0
= jU to the cathode surface, as U
is thermally measured in such a way that all the near cathode processes
contributing to the heating of the cathode spot are included [3]. Ob-
viously, this procedure, at the same time excludes all heat fluxes originating
293
at other places than at the near cathode arc spot region. Here, we will
also obtain the effective arc spot current density j semi-empirically, using
erosion experimental data and the thermophysical model of erosion.
The thermophysical model, presented in [1], is based on the assumption
of constant velocity v of the arc spot. However, according to investi-
gations of Szente et al [7-9], the chemistry of the surface layer on the
electrode can greatly influence the character of motion of the arc spot.
On clean surfaces, in the medium of inert gases, it is characteristic the
presence of a surface drag, that leads to a reduction in the velocity
and periodic stops of the arc spot, making the arc spot to show a step-
wise motion. Molecular gases, such as air and nitrogen, on the other
hand, makes the arc appear to move in a manner more similar to the
continuous motion.
Below, in the two first sections, we present the thermophysical theoretical
model for the continuous and step-wise motions of the arc spot. Next,
briefly, we discuss the thermal methods for the measurements of U and
j. Then, we present the experimental data for the erosion and compare
them with theoretical results. These last data were taken from experi-
ments that models the behavior of the arc in EAHs, as well from ex-
periments in Electrical Discharge Machining (EDM) equipments.
We will show that the present thermophysical model gives serious
justification to the formulation of the problem of erosion in the form
of heat diffusion equations, with special initial and boundary conditions,
which consider only the integral energy characteristics of the arc spot,
the thermophysical properties and the regime of the electrode surface.
This approach strongly simplifies the problem of electrode erosion, making
it possible to avoid the formulation of complex and uncertain plasma
microscopic models.
2. THE CONTINUOUS EROSION MODEL
We assume that the Joule heating of the copper cathode under the arc
spot can be neglected. This is possible if the arc spot current density
does not exceed 10
8
A/cm
2
[10]. Then, the arc spot can be replaced by
an ideal circular heat source, with uniformly distributed heat flux density
q
0
= jU = const [10]. We study the heating of the electrode surface in
coordinates attached to and moving with the source with constant velocity
v (see Fig. 1). The maximum time of a given electrode point heat exposure
to the arc spot is equal to
max
= d/v, where d is the arc spot diameter.
Due to high arc velocities, the Fourier number Fo
d
= a
max
/d
2
<<1 [10],
where a is the electrode thermal diffusivity. This means higher gradients
of electrode temperature along direction z, perpendicular to the electrode
surface, than along coordinates x and y.
294
Then, for the solution of the three-dimensional heat conduction problem
we disregard the heat flow along the surface of the electrode and solve
the one-dimensional heat conduction problem in z, obtaining the solu-
tion for the temperature T(z, ) (see [1]). Here, the time is taken with
the meaning of the interval of time since the front edge of the arc spot
(line SRT in Fig. 1.) has passed over the point, i.e. = l/v (see Fig.
1). During time a given point on the electrode surface within the arc
spot is being heated by the arc. We can distinguish here three possi-
bilities: (a) there will be no points under the arc spot reaching the fusion
temperature T
f
; (b) after a certain time
0
, two zones inside the arc
spot will be found, the zone MRNPM, with temperature T < T
f
, and
Fig. 1. Schematic diagram of the arc spot heat source. MRNPM - zone with temperature
T<T
f
; MPNQM is the zone with temperature T=T
f
(the fusion zone).
the fusion zone MPNQM, with T = T
f
=const; and (c) the intermedi-
ate (or transition) case between (a) and (b). It is possible to show that
the time for a given point under the arc spot to reach the fusion tem-
perature T
0
is given by the expression [1]
,
) (
4
2
0
0 1
]
1
q
T T
a
f

(2)
295
where T and are, respectively, the initial temperature and the heat
conductivity of the electrode.
In order to characterize the fusion zone extent in the arc spot we
use the non-dimensional parameter f = l
0
/d = t
0
v/d (see Fig. 1, with points
in the border line MPN at l
0
= v
0
being the first points to reach temperature
T
f
. It is easy to see that f can be written in the following form [1]:
,
8
) (
2 5 . 1
2 2 5 . 1
U aj
T T s
f
f

(3)
where I v s / we call the normalized velocity. It is obvious that the
three possible domains of values of f, f > 1, f <1 and f = 1 are related,
respectively, to the three cases (a), (b) and (c) mentioned above.
It is obvious to assume that, for points in the zone MRNPM, the
heat removal Q
r
, through heat conduction into the electrode body, will
be equal to the arc spot heat input Q
0
. However, for the fusion zone,
we assume that the heat removal and heat input over this zone are related
as Q
r
< Q
0
, which difference, Q
er
=Q
0
Q
r
, is called the erosion heat.
Q
er
is the total heat spent in the erosion process, i.e., the heat spent
in transforming the electrode mass in the fusion zone from the solid state
into the plasma one. Further, we assume that the mass erosion rate G
(kgs
1
) is proportional to the erosion heat Q
er
, which yields
,
0
gI h G h Q Q Q
ef ef r er
(4)
where h
ef
is the erosion effective enthalpy and g is the specific erosion
rate (kg C
1
). Our aim is to calculate the integrals Q
0
and Q
r
throughout
the entire area of the fusion zone F
MPNQM
. Neglecting the thickness of
the liquid layer, and assuming T = T
f
at z = 0, the heat conduction problem
for the fusion zone can then be solved for >
0
. The solution of this
problem gives the instantaneous heat flux density, which is removed through
the fusion zone [1,11]:
. tan
2
0
0 1 0

q
q
r (5)
After calculating the integrals Q
0
and Q
r
over the fusion zone area,
we obtain [1]:
296
I U UWI Q
er er
(6)
where the non-dimensional function W of the parameter f is given by
[1]
, ) (
4
sin
2
2 1
1
1
]
1
+ +

w w f f W
(7)
where , w
1
and w
2
are functions of f, shown in [1]. Function U
er
=UW
we call the volt-equivalent of erosion heat. Equation (7) can be approximated
by a rather simple expression [1]
. 477 . 1
04 . 0
442 . 0
475 . 2
13 . 7
1
,
_
+
+
+

f f
f W W
(8)
The standard deviation of (8), with respect to (7), is equal to 0.011 (W
max
=1).
The domains of W and f are related as follows: W= 0 for f 1 and
0 < W < 1 for f < 1. From (6), we obtain the expression for the specific
erosion rate [1],
,
0
ef
h
UW
g g +
(9)
where the additional term g
0
(called microerosion) has been introduced
to take account of the experimental fact that, even for the case of
W = 0, there is observed a certain minimum value of erosion g =
g
0
~ const. Even when melting and erosion must vanish according to
the above obtained thermophysical result, there remains a certain minimum
value of erosion because of the deviation of the characteristics of the
real arc spot from the simple model assumed here of an uniform and
constant heat flux density throughout the entire area of the spot.
From the above results, we see that the thermophysical model for
continuous arc motion can be represented, in the simplest form, by the
system of three equations (3), (7) or (8) and (9).
297
3. THE STEP-WISE EROSION MODEL
The step-wise displacement of the arc spot implies the existence of the
spot only at certain discrete positions. We assume an infinitely short
time of displacement of the arc spot from one point to another. Thus,
the spot residence time at a given point on the electrode surface can
be written as
L / ,
(10)
where L = nd is the arc spot step length, v is the mean arc spot motion
velocity from spot to spot and n > 1. We assume that the process of
electrode surface heating within the arc spot during time t takes place
uniformly and simultaneously over the entire arc spot area. After a certain
time t
0
from the beginning of the heating process all the spot area should
reach the temperature of fusion of the electrode T
f
.
Up to time
0
the heat flux supplied to the electrode surface Q
0
is
equal to the heat removal Q
r
through heat conduction into the electrode
body. Starting at time
0
some heat should be spent in melting the electrode
surface. We call the time-average (over the period of time total erosion
heat flux (in W) Q
er
= Q
0
Q
r
, where Q
0
= q
0
F is the total average arc
spot heat input during time , Q
r
= Q
0
+Q is the total average heat removal
by heat conduction into the electrode body and F is the area of the spot.
Here, Q
0
= q
0
F
0
/ is the arc spot average heat input (and heat removal)
during time
0
and Q (F / ) q d
r
0
is the total average heat removal

during time >
0
. Calculating this integral using equation (5), we obtain
the total mean erosion heat flux during time
Q
er
=Q
0
Q
r
=IUW
s
, (11)
where W
s
is given by the non-dimensional function
,
_

) 1 (
1
tan
2
1
s s
s
s
s
f f
f
f
W
(12)
and f
s
=
0
/ is the characteristic parameter for the pulsed (or step-wise)
arc spot erosion model. Comparing f
s
with f in (3), it is possible to show
that these parameters are related by the relationship
.
n
f
f
s

(13)
298
Thus, the parameter f
s
for the step-wise motion of the arc spot is similar
to the parameter f for the continuous arc spot motion model. On the
other hand, function W, for the step-wise model is similar to function
W for the continuous model.
Similar to the previous section, using equation (11), and assuming
that the total average erosion heat Q
er
is proportional to the mass erosion
rate G (see equation (4)), we obtain the equation for the specific mass
erosion rate for the step-wise erosion model as
,
0
ef
s
h
UW
g g +
(14)
where g
0
was introduced for the same reason as in formula (9).
4. COMPARISON BETWEEN THE EROSION MODELS
Firstly, it is useful to compare both, the continuous and step-wise models,
called as thermophysical models, with the old Guile model, expressed
in the form of the Arrhenius equation (1). In Fig. 2a,b we show the
theoretically calculated curves of erosion g for these two models versus
the inverse value of the electrode temperature. Figure 2a relates to the
continuous model and 2b to the step-wise one. Therein, we plotted also
the curves, marked with A, calculated according to (1). In this equation
we used G
e
= 0.13 eV in accordance with [2]. The value of A in (1)
was not given by the authors and we have chosen it so as to bring in
Fig. 2. Specific erosion g vs 1/T: (a) continuous model for different values of s
(shown); (b) step-wise model for different values of n (shown) and for s =1.5; A -
eq. (1), C - continuous model, for s = 1.5.
1 2 3
1
10
(a)
s=0.25
A
s=1.5
s=1.0
s=0.5
g

(
1
0
-
8

k
g
/
C
)
1/T (10
-3
K
-1
)
1 2 3
1
10
(b) A
C
n=8
n=4
n=2
n=1
g

(
1
0
-
8

k
g
/
C
)
1/T (10
-3
K
-1
)
299
coincidence the initial points for equation (1) and for the thermophysical
models at T = T
f
.
For the continuous model we used four different values of the pa-
rameter s (0.25, 0.5, 1.0 and 1.5 ms
1
A
0.5
). For the step-wise model,
we used four different values of the non-dimensional step n (1, 2, 4 and
8) with constant value of s = 1.5. It is seen that the thermophysical
model gives qualitatively similar results to those obtained by (1), though
revealing a stratification with respect to both the parameters s and n
which is not taken into account in (1). One can see also a rapid increase
in the erosion with the increase of the step n or the decrease of the normalized
velocity s.
For the calculations, we took identical values of all parameters, entering
the dimensionless parameters f and f
ss
. The thermophysical parameters
of copper we took in accordance with [12,13]: l = 377 Wm
1
K
1
, and
a = 10
4
m
2
s
1
. We took the following values for the remaining parameters:
j = 1.35 10
9
A m
2
, U = 6.78 V and h
ef
= 66 MJ kg
1
in accord with
the corresponding mean values for our experiments (see below and in
[3,4]). Comparison between the thermophysical models can also be carried
out via the corresponding dimensionless functions W and W
s
, which are
Fig. 3. Comparison between W (dash-dot curve) and W
s
(solid curves) for different
values of n (shown). Dashed curves difference W
s
W. In the inset, W
s
versus
parameter n for different values of s (shown in in ms
1
A
0.5
).
0 1 2 3 4
0.0
0.2
0.4
0.6
0.8
1.0
4
2
1
W
,

W
s

a
n
d

W
s

-

W
f = f
s
n
2 4 6 8 10
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
n = L/d
1.5
1.0
0.5
0.25
W
s
300
proportional to the erosion g (see equations (9) and (14)). Figure 3 gives
the plots of W, W
s
, and W
s
W versus f for different values of n, taken
as parameter. We see that the difference W
s
W rapidly increases with
increasing n. The inset gives the dependence of W
s
on n for different
values of s, taken as parameter. It is evident that the erosion rapidly
increases with the increase in n and decrease of s. From Figs. 2 and
3 we see that, even the regime of n = 2 can substantially increase erosion
in comparison with the continuous regime or with n = 1. Therefore, for
the purpose of measuring the values of L or n, the diagnostics must be
carried out with the precision of the order of the spot diameter d; this
imposes stringent requirements on the space and time resolution of the
equipment.
5. EXPERIMENTAL INVESTIGATIONS
The volt-equivalent of the arc spot heat flux U was obtained from the
equation U = Q
0
/ I, by measuring Q
0
and I, where Q
0
is the total arc
spot heat flux. We used the nonstationary thermal method in an uncooled
experimental setup equipped with coaxial ring electrodes. The measurements
were carried out in a wide range of parameters: magnetic induction 0.13-
0.95 T, air pressure 0.14 MPa and current 0.0251 kA. The axial gas
velocity was in the range 1.35.3 ms
1
. The detailed description of the
method used for the measurement of U has been given elsewhere [3].
Let us simply note that it is based on the fact that the heat supply into
the electrode through the arc spot is localized in the limits of the area
Fig. 4. Scheme of the experimental setup for erosion investigations.
301
of the spot itself and does not depend on the total surface of the electrode,
in contrast to the radiant convective heat supply from the arc column.
Therefore, by changing the geometry of the electrodes or by installing
calorimeters in different distances from the arc, it is possible to isolate
from the total heat flux Q the constant component Q
0
, which does not
depend on the geometry. Q can simply be obtained from the slope dT/
dt of the recorded temperature of the uncooled electrode as function of
time [3]. From our measurements, we concluded that UU. We also
obtained the following expression of U versus B:
. 28 . 4 52 . 6 B U + (15)
where U is given in volts and B in Tesla [3].
For the measurement of the arc spot current density j we have used
both a steady and an unsteady method. The later used the same experimental
setup as for the measurement of U. The detailed description of the
measurement of j by the unsteady method was given elsewhere [3]. It
is observed in the record of electrode temperature T versus time t that
the derivative dT/dt is constant up to a certain point after which it begins
slowly to drop. This point is assigned to the onset of fusion in the arc
spot, corresponding to f = 1. Measuring I, v and T as dT/dt starts to
drop, and using equation (3) with (for the continuous motion) we obtain
the effective arc spot current density j. However, for the step-wise motion,
from equation (13), making , we conclude that it is not possible to obtain
j if we do not know n.
We did not measure the value of n in our experiments as it presents
great difficulties. We have simply considered the motion of the arc spot
to be a continuous one or one with n = 1. From our thermal experi-
mental data we obtained j = (0.931.27)10
9
Am
2
[3], which shows good
agreement to the measurements of Rakhovsky (j = (0.52)10
9
(Am
2
),
carried out with highly sensitive optical-electronic equipment. This evidences,
to a considerable degree, that it was possible to ensure the motion of
the spot in the regime, close to the continuous one or to the regime with
n = 1. We can also conclude that the thermal method can give quite
accurate results, when compared to optical methods.
A coaxial water cooled experimental setup, shown in Fig. 4, was used
for the cathode erosion measurements in magnetically driven arcs. The
system was equipped with commercial copper ring electrodes. Compressed
nondewatered atmospheric air has been used as the working gas. The
mass erosion rate G was measured by the weighting method. We measured
I, v, the integral heat flux Q supplied to the outer ring inner electrode
surface and the temperature T
r
of the ring at some radius r. The elec-
302
trode surface temperature T was calculated from these measurements.
Two values of the inner ring electrode diameter D
1
, 50 and 90 mm, were
used.
In Fig. 5, we present experimental data on the erosion of a copper
cathode in air plasma for the following range of parameters: current
I = 24.51000 A, arc velocity v = 19.3344 ms
1
, electrode surface tem-
perature T = 3001073 K and magnetic field strength B = 0.130.242
T. One can see a rather large scattering of experimental points, but also
an evidently sharp increase in the erosion, beginning from a certain current,
which depends on the diameter of the electrode. In Fig. 5, we marked
with big open squares or triangles two groups of points (points 2 and
3), which were obtained under constant conditions (water cooling, magnetic
field and air flow rate), with only current being changed. We call these
groups of points as highlighted. For the electrode with diameter of
90 mm an increase in the erosion begins with higher current, since because
of the larger diameter it has lower temperature.
Assuming that the onset of melting in the arc spot (f = 1) is the reason
for the sharp increase in the erosion (points inside the framework in
Fig. 5), it is then possible to calculate the effective arc spot current
density j. From these points we obtained the average value j = 1.3510
9
A m
2
.
0 200 400 600 800 1000
0
1
2
3
4
C3
C2
1
2
3
4
g

(
1
0
-
8

k
g
/
C
)
I (A)
Fig. 5. Specific cathode erosion rate g vs current I: 1,2 points for 2R
1
= 50 mm;
2 highlighted points with B = 0.133 T; 3 highlighted points for 2R
1
= 90
mm with B = 0.03T; 4 (g = g
0
); inside frames: points with 1 f ; C2 best fitting
to points 1 and 2, C3 best fitting to points 3; Solid curves step-wise model,
dashed curves - continuous model.
303
The two similar figures in Figs. 6a,b,c present the experimental values
of U
er
as function of g. The left-hand side figure represents the con-
tinuous model and the right-hand side one the step-wise model. We
observe, from equations (6) and (9), that U
er
can be written as a lin-
ear function of g, i.e.,
U
er
h g g
ef
.
0
b g
(16)
Thus, from the fitting of the experimental points to a straight line, it
is possible to obtain the values of the important erosion parameters h
ef
and g
0
. In Fig. 6a we have taken the average values of U and j, i.e.,
U = 6.78 V and j = 1.3510
9
A m
2
. The correlation coefficient of the
straight line fitted to the experimental points in Fig. 6a was 0.8 for continuous
model and 0.83 for the step-wise one. The value of U in Fig. 6b is taken
according to equation (15), while of j is the average value given above.
In this case, the correlation coefficient was 0.89 for both models. Finally,
Fig. 6c (correlation coefficient 0.942 for continuous model and 0.95 for
the step-wise one) uses U given by equation (15) and j given according
to the equation
Fig. 6. Volt equivalent of cathode erosion heat U
er
vs specific erosion rate g (left-
hand side continuous model, right-hand side step-wise model): (a) fitting with
average values U = 6.78 V and j =1.35 10
9
Am
2
; (b) fitting with U vs B (equation
(15)) and on average value j = 1.35 10
9
Am
2
; (c) fitting with U vs B (equation
(15)) and j vs B (equation (17); straight line: best fitting to points 1, 2, 3 and 4.
(see symbols in Fig. 5).
0 2 4 6
0
1
2
3
4
(c)
R=0.95
N=108
U
e
r

(
V
)
g (10
-8
kg/C)
0 2 4
0
1
2
R=0.832
N=108
(a)
0 2 4
0
1
2
3
R=0.893
N=108
(b)
0 2 4 6
0
1
2
3
4
(c)
R=0.942
N=108
U
e
r

(
V
)
g (10
-8
kg/C)
0 2 4
0
1
2 (a)
R=0.801
N=108
0 2 4
0
1
2 (b) R=0.887
N=108
304
j B + (1.282 2.6 ) 10 ,
9
(17)
where j is given in Am
2
and B in Tesla. We see a constant improvement
of the best fit (shown by the improvement of the correlation coefficients)
as we take into account the dependencies U and j on B. This evidences
about the correctness and importance for erosion calculations of taking
into account the functions U(B) and j(B). The theoretical curves C2 and
C3 in Fig. 5 (dash lines for the continuous model and solid lines for
the step-wise one) correspond to the best straight line shown in Fig.
6c. From this line we also obtain the best values h
ef
= 66 MJ kg
1
and
g
0
=3.1 10
9
kg C
1
for the continuous model, and 81 MJ kg
1
and
2.44 10
9
kg C
1
for the step-wise one, respectively.
We used above a process of successive approximations (not presented
here) in order to obtain the best theoretical curves fitted to experimental
points in Fig. 6c. As a step toward this objective, we used the best fitted
curves to the experimental data of v(I) and T(I), taken for the highlighted
points. Formula (17) was obtained from the best fitted curves g(I) shown
in Fig. 5, for the two diameters 50 and 90 mm (or two magnetic fields,
0.133 T and 0.033 T, respectively) [4].
During processing of experiments with the step-wise model, we assumed
the value of the step n as constant, since we did not measure it. With
such an approach, the value of n is eliminated in the calculations of
erosion in accordance with equations (12), (13) and (14) and do not
affect the results shown in Fig. 6. However, such approach leaves unknown
the value of j in our experiments, if there is a step-wise motion of the
spot. We presented formula (17) assuming continuous spot motion or
motion with step n = 1. We applied small interelectrode gap (3 mm)
and narrow electrode rings to improve the uniformity of the arc rota-
tion. As we mentioned above, sufficiently close agreement between our
data on effective current density assuming and the results of the most
elaborated measurements, carried out with a highly sensitive optical-
electronic equipment in [6], evidences that it was possible for us, to
a considerable degree, ensure the motion of the spot in the continuous
regime or with n = 1. The ignorance of the true value of the step, of
course, gives some additional error to the results of processing of ex-
periments.
The aforesaid attests to the fact that the knowledge of the true na-
ture of the motion of the arc is desirable for obtaining more precise
data, both about j, and about the values h
ef
and g
0
.
6. OTHER AUTHORS EROSION EXPERIMENTS
The arc velocity seems to have a relevant role in establishing the final
305
Fig. 7 (a) Theoretical specific erosion rate g vs arc velocity v for different values
of dT/dv (Ksm
1
) (shown) for I = 100 A, T
0
= 480 K, h
ef
= 130 MJ kg
1
and g
0
=
1.7 10
9
kg C
1
; (b) experimental points taken from [14] and best fitting curve
for dT/dv = 4 Ks m
1
, the remaining parameters the same as in (a).
value of electrode temperature. However, it is known from different authors,
that for magnetically driven arcs, an increase in arc velocity can both
decrease and increase erosion, depending on the operating conditions
of the electrodes [14,15]. A possible explanation for this behavior, based
on the thermophysical model, is the following. With the aid of the magnetic
field, easily are achieved velocities of motion of the arc up to several
hundreds and even few thousands of meters per second (see [16]). The
motion of the arc, at a high velocity relative to the surrounding gas,
is responsible for the strong turbulent perturbations and increased rate
of heat transfer between plasma and the walls of the electrode [17]. With
the increase in the velocity of the arc, the mean temperature of the surface
of the electrode in the zone of motion of the arc spot (or the temperature
of the surface of the electrode in front of the spot) increases with an
increase in the velocity of the arc. This behavior should lead to a value
dT/dv > 0.
The second reason consists in that, with an increase in the veloc-
ity of the arc, decreases the time interval between the repeated visits
of the arc of one and the same point of the surface of the electrode.
This is consequence of the motion of the arc along a closed trajectory
in the circumference of the electrode. With an increase in the frequency
of rotation, the surface of the electrode does not manage to cool suf-
ficiently for the following arrival of the arc spot. Then, for this rea-
son also we should have dT/dv > 0. As it is shown in [17], this effect
306
is substantial only at arc velocities v > 100 m/s, which are possible for
magnetically driven arcs.
Taking a linear dependence of the electrode temperature with arc
velocity, T = T
0
+ v(dT/dv), where T
0
is the initial electrode tem-
perature and dT/dv = const, then for a magnetically driven arc we
should have dT/dv > 0. The other case of dT/dv < 0 can be imple-
mented only with gas-dynamic arc rotation due to the supplementary
cooling of the electrode with vortex flow.
Figure 7a shows theoretical plots of the erosion behavior with in-
creasing arc velocity. For dT/dv > 0 there is a range of arc velocities
for which the specific erosion decreases, followed by some minimal level
g > g
0
, and then the erosion increases with increasing velocity. Figure
7b, with data taken from [14], illustrates a qualitative comparison of
the behavior of erosion versus arc velocity. The authors in [14] did not
report the temperature conditions of the copper cathode. They used in
their experiments electrolytic copper and pure argonnitrogen mixtures.
In our calculations we used the average values U = 6.7 V and j = 1.41
10
9
A m
2
for B = 0.05T, according to equations (15) and (17), and
g
0
= 1.7 10
9
kg C
1
, according to [14]. We assumed T = 480 K and
dT/dv = 4 Ks m
1
, so that f = 1 is obtained at the point of erosion transition
to the regime with g
0
= const = 1.710
9
kg C
1
. The best agreement
of the theoretical curve with the experimental data was obtained by making
h
ef
=130 MJ kg
1
. The difference between these g
0
and h
ef
values and
our data could be explained by the difference in materials and the presence
in their gas composition of rather high concentrations of argon, which
can cause U and j values substantially lower than for pure nitrogen or
Fig. 8. Comparison between the present step-wise erosion theory and an EDM experiment
(data taken from [18]).
0 40 80 120
0.0
0.5
1.0
1.5
2.0
2.5
(b)
4

g

(
1
0

-
8

k
g
/
C
)
v (m/s)
0 40 80 120 160
0.0
0.5
1.0
1.5
2.0
2.5
(a)
g

(
1
0
-
8

k
g
/
C
)
v (m/s)
7 5
4
0 -4
307
air. As one can see in Fig. 8b, with further increase in the velocity, one
can expect an increase in the erosion. Unfortunately, the authors in [14]
did not give experimental points in this region. Application of the step-
wise model for the processing of the same experiments gives somewhat
worse results. In the EAH the real state of motion of the arc is not either
ideally continuous or step-wise one. However, the Electrical Discharge
Machining (EDM) process can provide us with periodically repeating
electrical discharges with accurately adjustable pulse duration. Such ex-
periments are reported, for example, in [18]. We disregard the value
of g
0
due to its smallness.
Using the authors data in [18] for the thermophysical properties of
steel (in the range of temperatures from 298 K to 1808 K (melting point),
= 56.1 Wm
l
K
1
, c
p
= 575 J kg
1
K
1
and = 7545 kg m
3
) as the cathode,
for U = 4.575 V and assuming j = 2 10
9
A m
2
, we obtain in Fig.
8 very good agreement of the step-wise model with the experimental
data [18]. Here, c
p
and are the specific heat and mass density of steel,
respectively. The effective erosion enthalpy obtained for the EDM gives
the value h
ef
= 4.8 MJ kg
1
much lower than for the cathode in EAHs
(6680 MJ kg
1
). This shows that, under the conditions of EDM, there
is considerably a more effective mechanism of removal of the melt in
the spot than in the EAH. In contrast to EAHs, erosion plays a posi-
tive role during electroerosion machining of materials.
7. CONCLUSION
A simple macroscopic thermophysical model of erosion of the cold cathode
has been described. We have considered the continuous motion with constant
velocity of the arc and the step-wise (intermittent) pulsed arc discharge.
We have shown that the Guile model (in the form of the Arrhenius equation)
is qualitatively similar to the particular case of the thermophysical model
with fixed values of the normalized arc velocity s and the dimensionless
step of the arc spot n. We proposed the dimensionless parameter f to
characterize the continuous motion erosion and f
s
for the step-wise pulse
discharge erosion. We have introduced a new method, namely a thermal
method, for the measurement of the important parameters U and j. We
have measured these parameters for a commercial copper cathode in air,
i.e., the same conditions as we have used for the erosion measurements.
The obtained value of j shows rather good agreement with the best values
published in the literature and measured by sophisticated optoelectronic
methods.
We have shown that for a more accurate calculation of cold cath-
ode erosion in EAHs with magnetic displacement of the arc, it is im-
portant to consider the dependence of the volt-equivalent arc spot heat
308
flux U and the effective arc spot current density j on the magnetic field.
We also explained the unexpected behavior of erosion, observed by different
authors, when, increasing the arc velocity can either increase or decrease
erosion. We have also shown theoretically that the nature of the mo-
tion of the arc spot substantially influences the erosion: the step-wise
motion of the spot leads to the increase in the erosion in comparison
with the continuous motion, the greater it is, the greater the length of
the step L of the arc. A sufficiently good agreement is obtained between
the theory and our own and other authors experimental data, obtained
on EAHs and EDM equipments. As a result of processing of experi-
ments with the present model, we obtained the important parameters for
the calculation of erosion, the effective enthalpy of erosion h
ef
and the
specific microerosion g
0
. For the copper cathode of EAH in air they
are equal to approximately 70130 MJ kg
1
and 1.73.1 mg C
1
, de-
pending on the type of copper material of the cathode and chemical com-
position of plasma. For the steel cathode in the EDM process, the ef-
fective enthalpy gives about 5 MJ kg
1
. This demonstrates the presence
of a considerably stronger mechanism of erosion in EDM equipments
than in EAHs, which is expected.
8. ACKNOWLEDGMENTS
We thank Mr. A. A. B. do Prado and Mr. J. B. Pinheiro for their technical
assistance in this work. We acknowledge the financial support of CNPq, FAPESP
and FINEP of Brazil.
References
1. A. Marotta and L. I. Sharakhovsky, A theoretical and experimental in-
vestigation of copper electrode erosion in electric arc heaters. I: The
thermophysical model, J. of Phys. D: Appl. Phys., vol. 29, pp. 2395-2403,
1996.
2. E. Guile, A. H. Hitchcock, K. Dimoff, and A. K. Vijh, Physical implications
of an effective activation energy for arc erosion on oxidized cathodes,
J. of Phys. D: Appl. Phys., vol. 15, pp. 2341-2355, 1982.
3. L. I. Sharakhovsky, A. Marotta, and V. N. Borisyuk, A theoretical and
experimental investigation of copper electrode erosion in electric arc
heaters. II: The experimental determination of arc spot parameters,
J. of Phys. D: Appl. Phys., vol. 30, pp. 2018-2025, 1997.
4. L. I. Sharakhovsky, A. Marotta, and V. N. Borisyuk, A theoretical and
experimental investigation of copper electrode erosion in electric arc
heaters. III: Experimental validation and prediction of erosion, J. of
Phys. D: Appl. Phys., vol. 30, pp. 2421-2430, 1997.
5. A. Marotta and L. I. Sharakhovsky, Heat Transfer and Cold Cathode
Erosion in Electric Arc Heaters, IEEE Transactions on Plasma Sci-
ence, vol. 25, No. 5, pp. 905-912, October 1997. Special issue on high
pressure arcs and high frequency thermal plasmas.
309
6. V. I. Rakhovskii, Experimental study of the dynamics of cathode spots
development, IEEE Transactions on Plasma Science, vol. PS-4, no.
2, pp. 81-102, 1976.
7. R. N. Szente, R. J. Munz, and M. G. Drouet, The effect of low con-
centrations of a polyatomic gas in argon on erosion on copper cathodes
in a magnetically rotated arc, Plasma Chemistry and Plasma Processing,
vol. 7, no. 3. pp. 349- 364, 1987.
8. R. N. Szente, R. J. Munz, and M. G. Drouet, Cathode erosion in inert
gases the importance of electrode contamination, Plasma Chemis-
try and Plasma Processing, vol. 9. no. 1, pp. 121-132, 1989.
9. R. N. Szente, R. J. Munz, and M. G. Drouet, The influence of the cathode
surface on the movement of magnetically driven electric arc, J. of Phys.
D: Appl Phys., vol. 23, pp. 1193-1200, 1990.
10. A. V. Luikov, A. V. Borovchenko, V. I. Krylovich, V. V. Toropov, L. I.
Sharakhovsky, and A. S. Shaboltas, Heat transfer in near electrode
region of the electric arc blown, in Proc. of Int. Symp. JSME, (To-
kyo), pp. 113-119, 1967.
11. V. I. Krylovich and A. S. Shaboltas, Thermal regime of a near elec-
trode region of moving electric arc. part II Izvestiya of BSSR Acad-
emy of Sciences, Ser. Phys.-Energet. Science, vol. 1, pp. 93-98, 1973.
In Russian.
12. Y. S. Touloukian, R. W. Powell, C. Y. Ho, and P. G. Klements, Ther-
mal Conductivity of Metallic Elements and Alloys. Vol. 1, New York,
Washington: IFI/Plenum, 1970.
13. Y. S. Touloukian, R. W. Powell, C. Y. Ho, and P. G. Klements, Dif-
fusivity. Vol. 10, New York, Washington: IFI/Plenum, 1973.
14. R. N. Szente, R. J. Munz, and M. G. Drouet, Effect of the arc ve-
locity on the cathode erosion rate in argon nitrogen mixtures, J. of
Phys. D: Appl. Phys. Vol. 20, pp. 754-756, 1987.
15. A. S. Anshakov, A. N. Timoshevsky and E. A. Urbakh, Erosion of copper
cylindrical cathode in air, Izvestya of Siberian Departament of USSR
Acad. of Sci., Techn. Sci., vol. 2, no. 7, pp. 65-68, 1988. In Russian.
16. L. I. Sharakhovsky, Experimental investigation of an electric arc motion
in annular ventilated gap under the action of electromagnetic force,
J. of Eng. Phys., vol. 20, no. 2, pp. 306-313, 1971.
17. A. S. Koroteev, V. M. Mironov, and J. S. Svirchuk, Plasmatrons - design,
characteristics, calculations. Moscow: Mashinostroenie, 1993. In Russian.
18. D. D. DiBitonto, P. T. Eubank, M. R. Patel, and M. A. Barruffet, Theoretical
models of the electrical discharge machining process. I. A simple cathode
erosion model, J. Appl. Phys. Vol. 66, no. 9. pp. 4095-4103, 1989
310
Emission Current Density and Electronic
Work Function of Metal to Plasma
H.Ts. Zayatuev
Gusinoozyerskaya GRES, Gusinoozyersk, Russia
One of todays most important and challenging problems of physics of
near-electrode processe in gas discharges is known as the abnormal
emission phenomenon which is experimentally observed on thermal emission
arc discharge cathodes under inert gas conditions [17]. However, in
spite of a variety of papers, concerned with the explanation of this
phenomenon [611], it might be pointed out that there are still no theoretical
investigations the results of which would be in close agreement with
the experimental data. In [7], the authors analysed the literature and
the results of their own investigations and concluded that the best matched
hypothesis to explain the abnormal emission phenomenon is the hypothesis
of large electronic work function e changes by the RichardsonDashmann
formula [12]
j AT
e
kT
e

_
,
exp

, (1)
as applied to their emission from the metal to plasma. This paper indicates
that to provide the current density and energy balance on the active surface
of a tungsten cathode the effective electronic work function e
eff
at
P = 10
5
Pa is bound to lincrease linearly with a rise of electrode temperature.
In this case, the increase of the electronic work function over 1000 K
range has been more than 1 eV, i.e. a factor of 10
1
over the data known
in the literature [13].
Thus, the paper [7] gives a formal explanation of the abnormal emission
phenomenon implied in wide e
eff
changes with the cathode tempera-
ture, but in fact a new problem has appeared because the strong de-
pendence e
eff
= f (T
c
) is difficult to explain in the context of the classic
theory of electronic emission from the metal.
311
Emission Current Density and Electronic Work Function
Thus, the purpose of this paper is to develop the theoretical model
to provide an explanation for the linear dependence of the effective electronic
work function on cathode temperature and thereby the problem of abnormal
emission as a whole.
With this aim in view, an analysis of the formula (1) and its deduction
can be performed:
(i) the formula includes the constants A and which characterize
the properties only of a solid body and values of this costant are ex-
perimentally defined under high vacuum conditions. Analysis of a large
amount of information on the measurements of these constants [13] points
to their wide scatter even for pure metals. For example, different au-
thors present the measured values of the electronic work function for
single crystal tungsten in the range from 4 to 7 eV and the range of
Richardsons constant from 3010
4
to 60010
4
A/m
2
K
2
. Such strong
distinctions of the experimental data are difficult to explain especially
in cases when Richardsons constant is the combination of world constants
itself. The situation with compounds and alloys of metals may be thought
of as a catastrophic one because only the deviation of the measured data
differs of 10
4
10
5
times.
(ii) the derivation of the formula is based on the following assumptions:
the first is that the electron emission is from hot metal to vacuum, i.e.
under zero density condition, and the second is that electronic gas in
vacuum near the electrode surface, which consists of only emission electrons,
and the electronic gas in metal are in thermodynamic equilibrium.
In spite of the fact that the measured results of the electronic work
function and Richardsons constant vary within wide limits and the formula
is given only for electron emission from metal to vacuum, it is the only
way of describing different emission phenomena on gas discharge cathodes,
i.e. to describe the electronic emission from metal to a finite density
medium. It follows that emission metal-to-plasma contact processes, i.e.
in contact with the medium the density of which is greatly higher tha
that of vacuum, is bound to differ essentially from the processes be-
ing described by the RichardsonDashman theory. Because of this reason,
the experimental results may be differ for both A and measurements
and thermal emission current densities j
e
on gas discharge cathodes.
Indeed, in the context of the assumption of thermodynamic equilibrium
between electron gases inside and outside the metal, the electron emission
conditions can be changed in the presence of the plasma with its electron
component near the electrode surface.
In order to clarify the extent to which the plasma has an effect on
emission processes, we consider a system of two interacting phases: electron
gases of metal and plasma at the state of thermodynamic equilibrium.
312
The conditions of the equilibrium are the equality of the temperature
and chemical potential level of the interacting phases at the region of
their contact [14]:
T
m
=T
e
=T
c
, eE
F
=e
e
. (2)
Here T
m
, T
e
and T
c
are temperatures of electron gases of the metal, the
plasma and the contact layer between them; E
F
is the Fermi level of
electrons of the metal; m
e
is the level of plasmaelectron chemical potential.
As part of theoretical investigations of near-electrode processes, we
also proceed from the conditions of equality of chemical potentials of
metal and plasma at the state of thermodynamic equilibrium. However,
in so doing, the position of Fermi level E
F
about the zero level of vacuum
or vacuum electronic work function em
0
, are supposed to be constant
and in accordance with the level of the plasma-electron chemical po-
tential the potential difference is recorded in the near-electrode region
[15,16].
In contrast to earlier studies, this paper suggests a somewhat dif-
ferent method of treatment of the emission processes based on the following
proposition: the Fermi level of electrons of metal in the region of the
metal-to-plasma contact can change its position about the zero level of
vacuum in accordance with the level of the plasmaelectron chemical
potential. Below we present a metal-to-plasma-contact model and the
results obtained in this work.
By analogy with contact phenomena of solid bodies, to describe the
processes of charged-particle transfer in metal-to-plasma contact, we
shall draw on the theory of zone diagrams which is based on the analysis
of the energy spectrum of electrons in metal. So far as this theory has
found its vertification in description of the properties of amorphous materials
[16], and by virtue of the fact that plasma also has a disorded struc-
ture and in the electron characteristic it is a close anologue of semi-
conductors and dielectrics, let us suggest apriori that the concept of
a zone or energy diagram also applies to this theory. Thus, the zone
diagram of plasma can be thought of as alternate energetic zones such
as the valency zone with the ceiling E
V
, the free zone or conduction
zone with the bottom

E
c
and the forbidden zone which is between E
V
and E
c
. The width of the forbidden zone approximates the potential of
ionization of plasma-forming gas E
i
atoms. This process can be thought
of as the electron transition from the valency zone involved in the transition
of electric current through the plasma.
313
Figure 1 illustrates the idealised zone diagrams of the metal and the
plasma prior to their contact. Metal-to-plasma contact with the same
temperature inidcates the start of the electron transition from one phase,
with the higher chemical potential level, to the second phase. Figure
2 shows the diagram where the solid lines indicate the case in which
the level of the plasma electron chemical potential is above the Fermi
potential of the metal. In this case, the electrons will transfer from plasma
into metal and the transition will continue until the levels of both phases
become equal. As a result, external contact potential difference V
c
occurs
between the plasma and the metal or a potential difference takes place
between their surfaces in the surrounding space. The internal potential
difference in the immediate metal-to-plasma contact region is zero. The
broken lines are the case when the Fermi metal level is above the level
of the plasma electron gas chemical potential.
This special feature of the plasma zone structure, i.e. the presence
of forbidden zone and conduction zone, determines the value of the electronic
work function from metal to plasma e
e
that matches the position of
the electron gas chemical potential level of near-electrode plasma with
an accuracy of the width of the conduction zone E
c
(see Figs.2 and 3),
i.e.
e
e
e
eff
. (3)
This is caused by the fact that the electrons can transfer from metal
only into the plasma conduction zone since the portion of the energetic
0 0
c

o

i

F

V

Fig.1. Idealised energetic (zone) diagrams of metal and plasma before contact.
E
F
is the Fermi level of metal electrons; e
0
is electronic work function of
metal in vacuum (vacuum work function); m
e
is the chemical potential level
of the plasma electron gas; E
i
is the potential plasma-forming gas atom ionisation
or the width of the plasma forbidden zone; E
V
is the ceiling of plasma va-
lency zone; E
c
is the bottom of plasma conduction zone.
314

V
c
0 0
n
c
v
c
n
c
v
ec
V
c

eff

eff

diagram of the plasma from chemical potential level m
e
to the bottom
of the conduction zone E
c
is the forbidden zone for them. Thus, in accordance
with the relationship between the E
F
and m
e
levels, e
eff
can be both lower
and higher than the vacuum electronic work function e
0
. It follows that
the level of the plasma electron gas chemical potential at the point of
contact virtually determines the value of the electronic work function
from metal to plasma. The expression for the level of the plasma electron
chemical potential or the effective electronic work function in terms of
the conditions (2 and 3) takes the form [17]:
e e kT
m kT
h
n
e e c
e c
ec

_
,
1
]
1
1
ln /
/
2
2
2
3 2
, (4)
where T
c
is the contact layer temperature; n
ec
is the plasma electron
concentration in the contact layer; m
e
is electron mass; k is Boltzmanns
constant; h is Planks constant. Since the system is in thermodynamic
equilibrium at equal phase temperatures and the thermal velocities of
electrons escaping from the metal are correspondingly equal to thermal
Fig.2. The energetic diagram of metal to plasma contact after establishment
of thermodynamic equilibrium. Surface metal temperature T
m
is equal to the
temperature of plasma electron gas T
e
, e
eff
and are effective electronic work
functions from metal to plasma; V
c
and V
c
is the contact potential differ-

ence between the external surface of the metal and the plasma; n
ec
v
ec
are
the electron flows from metal to plasma and from plasma to metal in the
thermodynamic equilibrium state.
315
velocities of plasma electrons, n
ec
determines the concentration of both
the electron flow from plasma to metal and of the electron flow from
metal to plasma. Then, when substituting (4) into the RichardsonDashman
formula (1), the expression for the current density of the electron emission
from metal to plasma is:
j en
kT
m
en v
e
EM
ec
c
e
ec ec
_
,

2
1
4
1 2
/
. (5)
Here v
ec
is the mean thermal velocity of the emitted electrons which
agrees with the contact layer temperature or electrode surface temperature.
It follows that this expression is in line with the formula for the cur-
rent density of electrons in motion with the mean thermal velocity v
ec
which agrees with the cathode temperature.
In real systems, the plasma temperature is practically always higher
than the electrode temperature, and not only electroncs but also ions
take part in electric current transition in plasma. As a consequence of
the ambipolar diffusion of electrons and ions from plasma to metal, the
internal contact potential difference V
f
is formed between them. The
conditions of thermodynamic equilibrium in this case are written only
for the contact layer through which the exchange with particles takes
place itself. The concentration of electrons n
M
ec
escaping from the metal
is equal to the concentration of electrons n
ec
P
arriving from plasma to
the surface of the metal through the potential barrier V
f
. The difference
between the electron flows out of the metal and the plasma is caused
by different values of their thermal velocities is equalized by the flux
of thermal ions from the plasma. The concentration n
M
ec
= n
ec
P
= n
ec
is de-
termined by the formula:
n n
eV
kT
ec e
f
e

_
,
0
exp
(6)
Here n
e0
and T
e
are the concentration and temperature of the perturbed
near-electrode plasmaelectron gas; V
f
is the contact potential differ-
ence the value of which is deduced from the condition of equality of
the total current density to zero to the electrode surface:
j j j
e
em
iT eT
rev
+ 0. (7)
While writing this expression to solve it for current densities (6),
316
we shall present the equation (7) as:
1
4
8 8 1
4
0
0
1 2 1 2
0
1 2
en
kT
m
kT
m
eV
kT
en
kT
M
e
c
e
e
e
f
e
i
i

_
,

_
,
1
]
1
1
_
,
+

_
,

/ / /
exp
. (8)
Here M is the ion mass of the plasma-forming gas; T
i
and n
i0
are
the temperature and concentration of near-electrode plasma ions. Solving
(8) in relation to eV
f
, we obtain an expression for internal contact potential
difference
eV kT
MT
m T
n
n
T
T
f e
e i
e
i
c
e
e
_
,
_
,
ln
0
0
1
. (9)
The substitution of (9) into (4) and (5) gives the expression for the
electronic work function:
e kT
m T
h
MT
m T
T
T
n
e c
e c e
e i
c
e
io
_
,
_
,
ln
/
2
2
1
2
3 2
, (10)
and the formula for the current density of thermal emission of electrons
from metal to plasma:
j An T
T
T
T
T
e
em
i e
e
i
c
e

_
,
0
1 2
1
/
, (11)
where A=e(k/2pM)
1/2
is the plasma-forming gas emission constant.
Figures 3 and 4 give dimensionless values of electric work function
and thermal emission current density defined by the equations (10) and
(11). As illustrates by curves presented they are in agreement with data
of paper [7].
317
Fig.3. Dependence of dimensionless work function on the plasma and cathode
temperature.

j
c

10
2

10
1

10
0

0.2 0.4 0.6 0.8 T
c
/T
e

Fig.4. Dependence of relative current density of the electron thermal emission
on the cathode and plasma temperature.
REFERENCE
1. S.V.Lebedev and S.Z.Khaikin, J. of Experimental and Theoretics Physics,
Vol.26, No.6, pp.723-735 (1954) (in Russian).
2. V.K.Rokhadgi, Raketnaya Tekhnika i Kosmonavtika, Vol.4, No.10, pp.214-
216 (1966) (in Russian).
318
3. A.M.Dorodnov, N.P.Kozlov and Ya.A.Pomelov, Thermophysics of High
Temperatures, Vol.9, No.3, pp.483-487 (1971) (in Russian).
4. S.V.Lebedev, Thermophysics of High Temperatures, Vol.11, No.5, pp.986-
994 (1973) (in Russian).
5. V.N.Anikeyev, Izvestiya Sibirskogo Otdeleniya USSR Ac. Sci., Ser. of
Tech. Sci., No.3, Issue 3, pp.60-67 (1981) (in Russian).
6. A.A.Portnikov, V.A.Petrosov and I.N.Ostretsov, Phizika i primeneniye
plazmennykh uskoritelyei, Minsk, Nauka i Tekhnika, pp.239-260 (1974)
(in Russian).
7. S.P.Vaschenko, G.-N.B.Dandaron, M.F.Zhukok and Kh.Ts.Zayatuev, Sibirskii
Fiziko-Tekhnicheskii Zhurnal, No.1, pp.98-106 (1992) (in Russian).
8. B.Ya.Moizhes and B.A.Nemchinskii, Zhurnal Tekhnicheskoi Fiziki, Vol.44,
No.12, pp.2539-2547 (1974) (in Russian).
9. V.L.Ginsburg and V.P.Shabanskii, Doklady USSR Ac. Sci., Vol.100, No.3,
pp.445-448 (1981) (in Russian).
10. M.P.Zektser, Thermophysics of High Temperatures, Vol.13, No.3, pp.491-
496 (1975) (in Russian).
11. A.P.Nevskii, A.I.Sharakhovskii and B.B.Rodkevich, Vzaimodeistviye dugi
s elektrodami plazmotrona, Minsk, Nauka i Tekhnika (1982) (in Rus-
sian).
12. I.N.Ostretsov, V.A.Petrosov, A.A.Potrnikov and B.B.Rodkevich, Zhurnal
Prikladnoi Mekhaniki i Tekhnicheskoi Fiziki, No.1, pp.162-164 (1974)
(in Russian).
13. V.S.Fomenko and I.A.Podchernyaeva, Emissionnye i Adsobtsionnye Svoistva
Veschestv i Materialov, Moscow, Atomizdat (1975) (in Russian).
14. R.Kheibud, Termodinamika Ravnovesnykh Protsessov, Moscow, Mir Pub-
lishing (1983) (in Russian).
15. F.G.Bakht and V.G.Yurev, Zhurnal Tekhnicheskoi Fiziki, Vol.49, No.5,
pp.905-944 (1979) (in Russian).
16. A.Phelts, Amorphnye i Stekloobraznye Neorganicheskiye Tvyerdye Tela,
Moscow, Mir Publishing (1986) (in Russian).
17. V.L.Granovskii, Electricheskii Tok v Gaze. Ustanovivshiisya Tok, Moscow,
Nauka Publishing (1973) (In Russian).
3 1 9
Erosion of Multi-arc Cathodes
A.N. Timoshevskii and V.S. Ponkratov
Institute of Theoretical and Applied Mechanics SB RAS
Erosion of electrode materials is a thermal phenomenon and is
connected with powerful heat fluxes onto the electrode surface in
the zone of the arc spot. One of possible ways of increasing the
electrode lifetime is based on dividing the high-current arc column
into several independent arc parts in near-electrode areas; this
method results in a local decrease of the electrode heat load.
Research of phenomena such as self-division of the near-cathode
area of the arc column into several stationary arcs burning on a sin-
gle electrode in air is described in detail in [1], with examples of
practical realization of multiarc electrodes. The electrodecathode,
Fig.1, is an empty copper cylinder, with active parts made of zir-
conium or hafnium pressed into one of its sides.
The active parts play the role of thermochemical cathodes of the
divided arc and stabilise the process of discharge division (Fig. 2).
In this system, material erosion occurs simultaneously on several
electrodes; however, the effect of electric current power on each el-
ement is smaller in comparison with the electric current of the main
arc.
Fig.1. Diagram of a plasma torch with a multiarc cathode: 1) copper holder
of the cathode; 2) active Hf inserts; 3) nozzle; 4) vortex ring; 5) anode; 6)
electric arc.

3 2 0
So, it is now possible to reach high current load on thermo-
chemical compound electrodes and extend their lifetime at the same
time [2].
Transition to the stationary division of the arc and a uniform
electric current distribution in the active elements made of Zr and
Hf takes place over a period of time. It is connected with either the
dynamics of gas flows in the electrode cavity or with the heating up
of the oxidenitride layer (dielectric) which generates in air under
cathode spots on the insert surfaces. Transition processes occuring
at the moment of arc connection have an effect on the erosion char-
acteristics of even ordinary thermochemical cathodes; because of
this reason, it was essential to carry out experiments to examine
these regimes for multiarc electrodes.
The experiments were carried out in a plant using a plasma torch
for metal cutting (Fig.1). Six Hf inserts with a diameter of 2.5 mm
were pressed (uniformly around it) into a copper electrode with an
inner diameter of 12 mm. With these geometrical dimensions of the
cathode and the nozzle (diameter 7 mm), and at an air flow rate of
22.5 g/s, the dividing current was 120150 A; it determined the
number of discharge n stable zones for the given common current
I = nI
c
. During the experiments, in addition to recording the period
of changes of common current, high-speed filming was carried out
of the arc behaviour in the inner cathode cavity. Figure 3a, b shows
the curves of arc current changes with time and the corresonding
number of stable stationary discharge zone.. The current increase
t i me const ant was defi ned by t he power mai ns; bal l ast wat er
rheostats and inductivity were included there. Films were analyzed
to describe the mechanism of discharge development. The following
results were obtained: ignition of a discharge in the gap between the
cathode and the output anode; connection of the main discharge and
Fig.2. Arc division in the cathode cavity I = 680 A.
3 2 1
gradual movement of the cathode root along the inner electrode sur-
face to the zone containing the inserts; this results in the rotation
of the radial part of the arc. Nonstationary division occurs at this
moment; it is characterized by acts of unstable movement of arc
spots and is followed by a stationary regime. During the movement
of cathode spots along the holder material one can see erosion of
copper which is ejected from near-cathode areas with streams. It is
possible to conclude that the electric current intensity determines
the dynamics of arc self-dividsion. When I > 450500 A, connec-
tion of stationary roots takes place gradually (as a rule) during
quite a long period of time of ~ 0.6 s, though at least three roots
may exist after ~0.1 s (Fig. 3a). Faster transition to the division re-
gime occurs at lower arc current (Fig. 3b).
Fig.3. Change of electric current I and the number of divided arcs n in the
cathode with: a) I = 620 A, b) 400, c) 530 (step start).
Division into three roots is observed often; however, one of the
roots disappears and appears again only when the current provid-
ing stability is reached. This transition trend is likely to be ex-
plained by the influence of copper vapours upon gas-dynamics and
by the method of preparation of emission insert layers; increasing
starting current increases the amount of copper vapours generated.
As shown in [3], the surface layer of a hafnium insert consists
mainly of hafnium oxides and nitrides. Their emission features may
develop only with temperatures of about 2 000 K. According to the
experimentally obtained dependence of the heat flux into the cath-
ode on current, the heating time of the insert surface up to this tem-
perat ure i s i n t he range from 10 s ( I = 100 A) t o 0. 05 s
(I = 700 A); the duration of current rise up to the set value is 1
2 s. In this process, the cathode spots are usually retained in the
3 2 2
place of contact between the insert and the holder; this results in in-
creased erosi on of copper around t he i nsert s. A represent at i ve
exper-iment is shown in Fig. 3b. During the first four second, arc
current was 120 A; it was then rapidly increased to 530 A; as this
takes place, stationary division into the maximum number of roots
was set during (13)
.
10
1
s. So, when the surface layers of the
thermochemical insert are heated enough, the time of arc division in
the cathode cavity corresponds practically to the time of the follow-
ing current increase up to the value providing stable parallel dis-
charge burning. The cathodes which have a flush-mounted bar in the
butt end surface show very good results in the division dynamics.
The first start with miminum current provides the fast input of the
arc column and of the cathode spot inside the electrode; visible
marks of melting are absent. It also prepares the emission layers of
all inserts by a less powerful heat flux. Graduall heating of the
active insert was known before as a method of better utilization of
thermochemical cathodes. The invention described in [4] suggests
that it is efficient to mix some argon into air (at the start of opera-
tion). Argon decreases the initial heat flux to the insert and to the
between the holders. When the required regime is reached, argon is
removed. Erosion of hafnium inserts of multiarc cathodes was de-
termined from their mass loss: each startup of cathode operation for
determining erosion in long regimes lasted from 30 to 50 min. The
mass loss was determined from suspending and measuring the es-
cape of linear bars (deepening). Comparison of the data from these
two measurements is necessary because a part of holder material is
lost (this is noted especially at connection moments). Figure 4
presents the data on multiarc cathode erosion on long connections.
Maximum current to each insert with the highest current of the
common arc did not exceed 170 A (I ~ 150 A); thus, the degree of
erosion expected according to the data in [5] should have been
Fig.4. Cathode erosion in long regime I = 400 A. 1) cathode weight; 2) linear
deepening of the inserts.
3 2 3
about 10
11
kg/C. It is shown by the averaged results of measure-
ments of the linear escape of active elements (dotted line). The ir-
regular line in Fig.4 (measurement of specific erosion by electrode
mass losses) shows clearly moments of impulse burning-out of the
copper holder; this burning-out occurs in transition of arc connec-
tion and the establishment of stationary division regime.
Erosion of hafnium inserts of multiarc cathodes is more rapid in
the connection regime (Fig.5) in comparison with the available data.
The average level G in the connection regime is approximately
2.10
10
kg/C; it does not depend on the electric current. For an or-
dinary thermochemical cathode [3], specific erosion rises practically
exponentially with the increase of electric current; for example, ero-
sion is 7.10
10
kg/C at I = 400 A. The number of possible connec-
tions of the multiarc cathodes is 710 times higher than that of the
conventional cathodes. As indicated by the behaviour of the curves
1 and 2 (Fig.5), the dynamics of divided discharges influences in a
simple manner the erosion of the whole cathode: one can observe
regimes with an intensive loss of electrode material; these regimes
Fig.5. Cathode erosion in connection regime: 1) I = 650 A; 2) 800; 3) 750
(step start).
Fig.6. Influence of arc ignition methods and holder material upon cathode
erosion at I = 400 A; holder material is copper: 1) ordinary start; 2) step
start. Holder made of zirconium bronze: 3) step start.
3 2 4
are able to induce fast breakdown of the cathode. If step start of
the plasma torch is carried out (see Fig.3,b) (by doing so, the ox-
ide surface of thermoemission inserts is heated up in advance), elec-
trode erosion is absolutely even, the dotted curve in Fig.5.
The influence of the regimes of the determined current settings
is very strong in the initial 2050 connections. With the usual meth-
ods of arc ignition (Fig.6) erosion initially increases; its intensity
decreases i n subsequent connect i ons (curve 1). Because of t he
pulsed thermoelastic stresses (they appear in the cathode with cur-
rent fluctuations), operation becomes unstable in the regime of
multiple connections; even in the inital stage the near-surface in-
sertholder thermal contact is disrupted; higher erosion follows.
Deepening of the insert relatively to the surface of the inner holder
cavity makes the cooling regime more efficient by decreasing the
intensity of thermal vibrations of the holder material near the con-
tact edge. Step plasma torch start changes the cathode erosion dy-
namics (curve 2). In initial 4050 connections growth up to the set
value takes place.
In this start regime, erosion of the contact edges is reduced by
prior heating of the surfaces of the thermochemical bars before arc
division;
however, cyclic thermal loading of the insert breaks gradually the
copper around the bar and erosion intensifies. So, if the holder
material is able to restore the initial dimension in cooling, it is
possible to expect steady specific erosion not only in the first
moments of connection, but also in subsequent operation. Fireproof
bronze belongs to such materials [6].
Cathodes with the holder made of chromiumzirconium bronze
were studied. The results were satisfactory (straight line 3).
Using these procedures, it is possible to select optimum holder
materials for complicated thermochemical cathodes together the
methods of initial start and establishment of the regimes for the
plasma torches with multiarc cathodes. In the main operating pa-
rameters, the multiarc cathodes are much more efficient than the
ordinary thermochemical ones.
References
1. O.P. Solonenko and M.F. Zhukov (editors), Material erosion and dy-
namics of electric arc discharge in cylindrical electrodes, in: Thermal
plasma & new materials technology. V.1. Investigation & design of thermal
plasma generators, Cambridge Interscience Publishing, Cambridge (1994).
2. O.Ya Novikov., P.I., Tamkivi, A.N Timoshevskii., et al. Multiarc sys-
tems [in Russian] Nauka, Novosibirsk (1988).
3 2 5
3. M.F. Zhukov, A.V.Pustogarov, G.-N.B. Dandaron, A.N. Timoshevskii,
Thermochemical cathodes [in Russian], Institute of Thermophysics
SB RAS, Novosibirsk (1985).
4. Patent 847598 USSR, ICI B23K 9/16 Way of plasma treatment// Bykhovskii
D.G., Kunin V.S. et al. Zhukov M.F., Kozlov N.P., Pustogarov A.B. et
al. Near-electrode processes in arc discharges [in Russian] Nauka,
Novosibirsk (1982).
5. V.G. Osinzev, A.I. Grigoriev, Yu.V. Kurochkin, et al, Izvestiya SB RAS,
4, No.15, (1987).
3 2 6
Influence of Three-Phase Electric Arcs,
Burning on Oxide Melts, on Quality of
Electric Power
U.B. Ashimov., E.A. Abdrachmanov, G.D. Manapova and
T.S. Maldybaeva
Kazakh National Technical University, 480013, Almaty, Kazakhstan
The three-phase electric arc, generated by special electric arc installations
is used widely in a number of technological processes of metallurgical,
chemical and building industries, superseding traditional technologies
at the expense of increase of productivity, capacity, reduction of processes
stages, and other.
The characteristic of three-phase electric arc units is generation
by electric arcs of higher harmonic components of currents and voltage
in supply circuits which negatively influences the quality of electric
power [1].
The higher harmonics, generated by electric arcs, have a strong
effect on quality parameeters of electric power, such as the non-sinusoidal
wave of the form of the voltage curve in AC circuits, on technological
parameters of electroreceivers, connected to them, and also on the
work of electric arc units. So, for example, they result in large difficulties
in automatic control, as the measuring bodies of electronic regula-
tors react to the effect of current and voltage, with the latter greatly
deformed by higher harmonics [2].
Certain amount of research has been carried out the field of metallurgy
on higher harmonic components of current and voltage of electric
steelmelting furnaces. The dependences of levels and spectral structure
of higher harmonics of three-phase arcs on the electrical and tech-
nological parameters and work conditions of furnaces have been
established.
Recently, the three-phase arc has been considered widely in the
know-how of electromelted oxide refractory materials. The production
3 2 7
Influence of Three-Phase Electric Arcs Burning on Oxide Melts
of these materials comes true in three-phase electric-arc furnaces as
OKB-2126, 2130, RKZ-4, RKZ-6 and other ones, specially devel-
oped for these purposes.
As against metal, oxide refractory materials are characterized by
higher melting point, no conductivity in the solid state, and their melts
have low electric- and thermal conductivity, high viscosity [3]. These
specific properties stipulate special features of burning of three-phase
electric arcs on oxide refractory melts and generation of higher harmonics
of currents and voltage.
Therefore,, it is essential to carry out experimental research of
the spectral structure of higher harmonics of currents and voltage,
generated by three-phase electric arcs, burning on oxide melts, re-
veal the laws of changes of the structure and levels of higher har-
monics in relation to the electrical and technological parameters and
working conditions of mining and thermal furnaces.
Experimental research of higher harmonic components of currents
and voltage, generated by three-phase electric arcs, burning on oxide
melts, were conducted in an integrated laboratory three-phase four-
electrode furnace with a rectangular bath and a linear electrode block
(with a split phase A on outer electrodes), capacity 150 kVA and
on the industrial furnace RKZ-4, capacity 4 kVA.
During experimental melting, we recorded and measured the signals
of current and voltage with the help of the measuring-registration
circuit, developed on the basis of the analysis of systems existing
in metallurgy and electric power industry, the including pick-up unit
of current and voltage, multichannel magnetograph K-067, oscillographs
N8-13, six-channel self-recorder K-338, analyzer of harmonics. The
researched electrical signals, reduced with the help of dividers up
to 0.5-1 V, were written down on magnetic tape of seven-channel
magnetograph K-067 and paper tape of A-338 automatic recording
systems (Fig.1).
Processing the experimental data, the calculation of the harmonic
structure of current and voltage were conducted with the help of systems
specially developed for these purposes for processing the informa-
tion from magnetographs and a program of parameter calculation of
higher harmonics (Fig. 2).
Experimental melting in the three-phase four-electrode furnace was
conducted at a voltage of U = 100140 V, current I = 300450 A,
length of disintegration of electrodes L
d
= 85-120 mm, length of arc
L
a
= 10/30 mm. As a charge we used basalt rock.
Figures 3 and 4 show the characteristic oscillograms of phase current
and voltage, recorded during experimental melting. As indicated by
3 2 8
the oscillograms, the forms of the current curve is deformed insig-
nificantly and is close to sinusoidal, with the largest distortions shown
by the phase voltage curves.
The calculated results were used to plot dependences of the non-
sinusoidal wave factor K
ns
of phase current and voltage on melting
time t. In Fig. 5 and 6, the curve changes of the non-sinusoidal wave
factor of phase voltage and current in melting process, character-
istic for all investigated modes, are indicated.
The analysis of the data shows that K
ns
is affected strongly in the
initial period. The sharp and large fluctuations of higher harmon-
ics of phase currents and voltage at this stage are connected to the
effect on separate charge portions in the subelectrode spaces in zones
of arc burning, resulting in cooling, breakage and burning out. In
the main, such fluctuations are characteristic of split phase A, as
explained by the existence in one phase of two electric arcs, arrangement
of electrodes concerning the furnace bath (outer electrodes); owing
to electrical connection between these electrodes, the instability of
burning on one electrode influences work of the arc on the other electrode.
In the given furnace space, heating and increase of the size of the
liquid bath, the increase of temperature, natural and chemical uni-
formity improved conditions of the existence of arcs and stability
of their burning. The K
ns
curve of phase current and voltage in this
mode of operation is reduced up to minimum significance.

Fig.1 Combined measuring circuit.
3 2 9
Fig.2 System of information processing from magnetograph.
Fig.3 Oscillograms of voltage and current at I = 450 A, L
g
= 20 mm, L
d
= 100mm.
Fig.4 Oscillograms of voltage and current at I = 450 A, L
a
= 0 mm, L
d
= 100 mm.
3 3 0
Fig.5 Curve changes of non-sinusoidal wave factor of phase voltage.
Fig.6 Curve changes of non-sine wave factor of phase current.
With melting of the charge, the arcs gradually open, increase in
size and mobility. Intensive heat exchange of arcs with the environment,
their cooling and decrease of stability of arc burning explain some
increase of fluctuations of harmonics of current and voltage at this
stage of the melting process.
As shown by the results of research, the most significant term in
the harmonic spectrum of both current and voltage is the third harmonics.
The third harmonic of phase voltage depending on melting stages change
in the interval (7.959.8) %, and current in the interval (1.17.8)
%.
The increase of current in electrodes from 300 up to 450 A re-
duces the level of higher harmonic components of voltage with (20-
32)% up to (1226) %, and current with (2.811)% up to (1.09.5)
3 3 1
%, is connected with increase of temperature and improvement of the
stability of arc burning.
With increase of the length of electrode disintegration, irrespec-
tive of the size of the entered capacity, the contents of higher har-
monics, especially third harmonics, in the phase current and volt-
age grows in the initial melting moment. It is connected with increase
of the active resistance of the bath in the interelectrode space, which
hinders firing of the furnace and results in stable arc burning.
It is necessary to note that in all modes of operations of electric
arc installation the level K
ns
of phase voltage is above that of the
phase current. It is confirmed by the oscillograms of phase currents
and voltage.
The research of the laws of generation of higher harmonics by three-
phase electric arcs on an RKZ-4 industrial mining-thermal furnace
RKZ-4 was conducted at various electrical modes of its work: U =
249281 V, I = 910 kA.
Figure 7-12 show the characteristic recorded plots of higher harmonics
of current and voltage indicated by the work of the furnace trans-
former on 5 steps (U = 281 V, I = 910 kA).
the start of melting (the period of loading and after-loading mode).
This period is characterized by sharp and large fluctuations of higher
harmonics of current and voltage. Duration t = 3035 min, the non-
sinusiodal wave factor of current K
ns
= (4.57.01830) %, the non-
sinusoidal wave factor of voltage K
ns
= (2.03.56.010.0) %. The
duration of the mode depends on the quantity and mode of charge
loading;
the mode of quiet arc burning. It is characterized by minimum
fluctuations of higher harmonics. Duration t = 3540 min, K
ns
= (4.0-
5.0 11,0) %. The levels of higher harmonics grow up to the end
of the mode;
Fig.7 Recording of the 3rd, 5th harmonics of voltage at charge loading.
Fig.8 Recording of the 3rd, 5th harmonics of voltage in the quiet arc burning.
3 3 2
Fig.9 Recording of the 3rd, 5th harmonics of voltage in the open arc burning mode
Fig.10 Recording of the 3rd, 5th, 7th harnomics of current ar charge loading.
Fig.11 Recording of the 3rd,5th,7th harmonics of current in the quiet arc burning
mode.
Fig.12 Recording of the 3rd,5th,7th harmonics of current in the open arc mode.
- the open arc mode. It is characterized by large fluctuations of
higher harmonics. K
ns
= (5.57.0 12.013.0) %, K
ns
= (5.57.5 -
11.0) %. It is also necessary to note that in all modes the levels of
higher harmonics of current on the electrode are above the levels of
higher harmonics of voltage.
The similar character of change is observed on the recording of
of capacity (Fig.13).
3 3 3
The period of significant fluctuation of capacity at the beginning
of melting, stipulated by charge loading with heating of the bath and
melting of part of the charge, is replaced by the plot of rather quiet
allocation of capacity. With opening of an arc new fluctuations of
capacity begin.
In Fig. 14 and 15 the curve changes of the non-sinusoidal wave
factor of phase current and voltage of the REZ-4 furnace in the melting
process of corundum refractories are shown
Fig.13 Recording of capacity.
Fig.14 Curve changes of non-sinusoidal wave factor of phase current.
The comparative analysis of the dependence of levels and char-
acters of changes of the maximum harmonics of current and volt-
age in relation to the technological stages of the melting process in
the RKZ-4 industrial furnace and laboratory three-phase four-electrode
furnace shows that the changes of the non-sinusoidal wave factors
are similar and have general laws. The revealed laws of the change
of the spectral structure, levels and nature of fluctuation of higher
3 3 4
Fig.15 Curve changes of non-sinusoidal wave factor of phase voltage.
harmonics of currents and voltage in the melting process of oxide
materials are determined by physico-chemical processes, occurring
in the furnace bath and by the position of electrodes in relation to
the charge and the bath. It allows to identify the condition of the
furnace bath (form, sizes, natural and chemical uniformity, temperature)
under the characteristics of higher harmonics of current and volt-
age and use them as informative parameters, determining the indi-
vidual stages of the corundum melting process.
Thus, the research results are of interest not only for the devel-
opment of measures for decreasing the higher harmonics of current
and voltage in circuits with three-phase electric arcs, but also for
system engineering in the automatic control of melting processes of
oxide materials.
References
1. Smelyansky M.Ya., Mineev R.K. Influence of electric arc furnaces
on systems of electric supply. - Moscow, Energia, 1975.
2. Mineev R.K., Micheev .P., Ryzhnev Yu.L. Efficiency increase of electric
supply of electric furnaces. - Moscow, Energoatomizdat, 1986.
3. Strelov K.K., et al. Technology of refractory materials. - Moscow, Metallurgy,
1988.
335
Modelling of Flow Stabilization by the Swirl of a Peripheral Flow...
Modelling of Flow Stabilization by the
Swirl of a Peripheral Flow as Applied to
Plasma Reactors
E.P. Volchkov, V.P. Lebedev, V.I. Terekhov and N.E. Shishkin
Institute of Thermophysics, Siberian Branch, Russian Academy of Sciences,
Novosibirsk, Russia
1. INTRODUCTION
The gas-swirl stabilization of plasma jets is one of effective methods
of its retention in the near-axial area of channels in generators of
l ow-t emperat ure pl asma [1]. Except t he effect of gas-dynami c
compression, the peripheral swirl allows to solve another urgent problem
- to protect the reactor walls from the heat influence of the plasma
jet [2,3]. Swirl flows are also used for the flow structure formation
and control of the heat and gas-dynamic characteristics of different
power devices and apparatuses, using high-temperature working media:
in swirl furnaces and burners, in aviation engines, etc. [4,6]. Investigations
show that during swirl stabilization the gas-dynamic structure of the
flow influences significantly the spatial stability of the plasma column
and its characteristics.
The main difficulty, arising during the design of the devices with
swirl retention of plasma jets or flames, is poor knowledge of the
aerodynamics and heat and mass transfer processes in these devices.
The stabilization process is extremely sensitive to the change of regime
and geometrical parameters. Thus, the determination of the boundaries
of the stable localization of high-temperature jets is necessary for
the prediction of the optimal regime parameters for plasma torches
and plasma-chemical reactors.
The problem under consideration is very complicated. The three-
dimensional character of the flow is accompanied by the impact of
a strong field of centrifugal forces on the transfer processes. The
anisothermal nature of the flow can affect significantly the mixing
336
of flows. This is confirmed by investigations in this field [5,7-9].
The process of gas-dynamic stabilization of a jet in the near-axial
area can be explained using the concepts of the stability mechanism
in rotating flows. Available experimental data on the turbulence structure
in rotating flows [8,10-12] demonstrate the drastic decrease of ve-
locity pulsation in areas with damping of turbulent exchange. That
is why the jet localization process in the near-axial area will be affected
by the flow laminarization in a layer of mixing under the action of
the mass forces field.
In general, the formation conditions of these areas in swirled flows
are known now [4,8,13,14]. They follow from the principle of the
flow stability in the field of mass forces, formulated by Raleigh. Ac-
cording to this principle, the necessary conditions of turbulence damping
are circulation and density distribution over the radius /r > 0 and
/r>0.The intensification of turbulent transfer will occur with
reverse signs of circulation and density gradients.
We would like to notice one more important condition, influencing
the processes of jet mixing in swirl flows. The swirl of the peripheral
stabilizing jet leads to the gradient increase of the total averaged velocity.
According to the existing ideas concerning the mechanism of turbulent
energy generation [8,15], this factor must lead to intensification of
mixing, deteriorating the thermal insulation of the plasma jet from
chamber walls.
Thus, three main factors will influence the mechanism of plasma
retention: the damping of the turbulent transfer at the expense of cen-
trifugal forces, buoyancy forces, resulting from the density gradi-
ent over the radius, and mixing intensification due to the increase
of the total velocity gradient in the swirl flow as well. In [9,13-16]
experimental estimates of each factor contribution during stabilization
of jets with different density in vortex chambers were performed.
However, these data cannot be considered as exhaustive ones, because
they do not illustrate the variety of swirl flow interaction. The great
influence on mixing processes as a whole will be caused by the flow
boundedness, characteristic for the flow interaction in a vortex chamber.
Boundary layers, forming at the side and front walls, can lead to a
cardinal restructuring of the flow. According to experimental data
[17,18], the size relation of peripheral and near-axial jets and rim
thickness, separating these jets, affect the flow formation.
The creation of reliable methods for the calculation of jet development
with gas-swirl stabilization meets some principal difficulties. First
of all, at present there is no strict theory, describing the effect of
mass forces on turbulence. Available empirical ratios and simple theories
337
need experienced proof for the certain conditions under considera-
tion.
Experimental studies with direct injection of plasma jets into a
channel are complicated by methodological difficulties. The high level
of temperatures does not allow to perform fine parameter measurements
with good spatial resolution. Thus, the model constructions, illus-
trating the main features of the physical mechanism of the complex
studied process, are of great importance.
The present work is devoted to experimental modeling of swirl jets
mixing in channels. In particular, the parameters in the near-axial
flow area were studied thoroughly. Two characteristic cases of jets
mixing were investigated during experiments. They are most often
realized in generators of low-temperature plasma: mixing in long
cylindrical channels (Fig.1,a) and in swirl diaphragm chambers (Fig.1,b).
In the first case, the flow is modeled inside the discharge channel
of plasma torch and in the second - inside the discharge channel of
the plasma reactor for various plasma-chemical processes. The main
aim of the presented cycle of experimental investigation is to reveal
the total regularities of transfer process in these two kinds of swirl
flows as well as the individual peculiarities of structure formation
of the flow.
(a)
(b)
Fig. 1. Studied schemes of mixing of co-axial swirled jets; a) cylindrical channel;
b) diaphragm vortex chamber. 1 - swirler; 2 - separation rim; 3 - side wall;
4 - edge cap.
338
2. MIXING OF CO-AXIAL SWIRL JETS IN A TUBE
Experimental studies were carried out in a cylindrical channel with
an inner diameter D = 46 mm and length L/D = 9 gauges (Fig.1,a).
The near-wall jet was fed at the inlet to the working section through
a tangential ring slit with the height s = 2 mm. The secondary flow
was swirled inside the slit by spiral ribs with different outlet angles
relative to the axial direction.
In some experiments, we studied the influence of a rim, separating
the flows, on mixing of co-axial flows. The rim thickness t changed
at the expense of the variation of the axial jet diameter d, when its
relative value became t/s = 0.15, 1.5, 4 and 8.
Experiments with jets heat mixing were carried out under conditions
close to isothermal ones: one of the flows (central or peripheral one)
was heated by 60C in relation to another. The main air flow was
fed with velocity W
0
= 30300 m/s, the relative parameter of air injection
through the slit was varied in a wide range m =
s
W
s
/
0
/W
0
=0.28.
The measurement of the fields of velocity components, their pulsation
and temperature was performed by microprobes. The detailed description
of measurement methods and their errors is given in [19].
The mixing of the peripheral ring jet with the central flow in a
tube with high injection parameters has a jet-like character. Profiles
of longitudinal and tangential velocity and temperature are described
by well-known jet dependences. Generalizations of the measured radial
distributions of longitudinal velocity and temperature, presented in
Fig.2, confirm the above. The line in this figure corresponds to calculation
according to Wieghardts formula:
( )
[ ]
W V T y y
x

exp .
.
/
069
0 5
13 6
(1)
where
( ) ( ) W W W W W
x x xm

0 0
; V V V
m
, ( ) ( ) T T T T T
m

0 0
are dimensionless velocities and temperatures; y y
m
; and y
0 5 .
is
the coordinate of the half maximal velocity or temperature. The
universality of the ratio (1) was valid in all studied range of swirl
angles, distances from the inlet and injection parameters. The only
exception is the behaviour of the longitudinal velocity component at
low injection parameters (m < 1), when near-wall processes start the
governing influence on the flow. Also, under these conditions the flow
339
development in axial direction is affected by the swirl of a peripheral
jet, which can result in a significant deformation of the velocity profiles.
This problem is considered in detail in [19,20].
The radial distribution of a circulation channel =V
r flow swirl
angle ( = arctan (V
/W) and the moment of momentum flux are also

described by the jet regularities (1). This conclusion, obtained in [21]
is important for the development of theoretical models on jet-like flows
mixing.
The temperature change along the axial length is one of param-
eters, characterizing the intensity of flow mixing. The temperature
value on the axis is usually expressed in dimensionless form =
(T
0
T
mix
)/(T
01
T
mix
), where T
0
and T
01
are the flow temperatures on
t he axi s i n t he current cross-sect i on and i n t he channel i nl et ,
T
mix
=(G
0
C
p
0
T
01
+G
s
C
p
S
T
s
)/(G
0
+G
s
)C
p
mix
is the temperature of complete
mixing, G
0
, C
p
0
and G
s
C
p
S
are the flow rate and specific heat capacity
of the central and peripheral flows, correspondingly.
The experimental data on temperature changes along the channel
axis for different swirl angles of the peripheral flow with a thin separation
rim (t/s = 0.15) are presented in Fig.3. Dimensionless temperature,
equal to 1, corresponds to the initial region of the central jet, and
temperature, equal to 0, corresponds to the area of complete mix-
ing. It is obvious from the graph that the length of the initial re-
Fig. 2. Profiles of longitudinal velocity (a) and temperature (b) in the jet boundary
layer. Lines correspond to calculations by formula (1).
340
gion is small for the non-swirled peripheral jet (2 caliber for m =
0.5 and 1 caliber for m = 5). The temperature change at the main
region of the flow depends on the injection parameter and swirl angle.
The jets mixing is more intensive for m = 5 than for m = 0.5 as for
swirled so for non-swirled jets. The peripheral flow swirl leads to
a reduction of the intensity of the mixing processes. According to
Fig.3, it can be considerable.

Fig. 3. Influence of peripheral flow swirl on the temperature along a chan-
nel axis.
This character of T
0
change along the axis is explained by the effect
of centrifugal forces. The swirl of the peripheral flow creates a positive
circulation gradient over the radius, which leads to the damping of
velocity pulsation and flow stabilization in the near-axial area. This
is confirmed by the measurements of the turbulence degree along the
length in the near-axial area. The results of these measurements are
presented in Fig.4. The turbulence degree there is determined as
Tu W W
0 0
2
'
where W
0
and W
0
'
2
are the average and root-mean-square
velocity pulsation at the axis in the studied cross-section. It is clear
from Fig. 4 that the swirl of the peripheral flow decreases the tur-
bulence degree in the near-axial area. Nearly complete degeneration
of initial turbulence is observed for low injection parameters (m =
5) and swirl angles
s
= 5874.
The turbulence change along the channel axis has a complex character
for high injection parameters (m = 5). The turbulence maximum is
reached in the joining area of the boundary layers of the jets. There
341
Fig. 4. Turbulence degree of a jet with peripheral swirl in the near-axial area.
is no such extremum for the swirled peripheral flow within the length
of the region.
The thickness of the separation rim t/s is an important parameter,
characterizing the mixing process. At the first stage we studied the
influence of the rim size on the regularity of temperature alteration
along the axis without rotation. These data are presented in Fig.5.
An increase of the rim thickness, separating two non-swirled flows,
promotes their mixing and, correspondingly, the decrease of the length
of potential flow zone in the near-axis area. Thus, the decrease of
t/s from 0.15 to 8 reduces the relative length of potential core from
2.5 to 0.5, in so doing the temperature along the axis decreases drastically,
almost reaching the value of complete mixing.
A change of the separation rim thickness causes a significant alteration
of the pulsation pattern of the flow. This can be observed in Fig.
6, where the data on the turbulence intensity along the axis are presented
for different values of t/s. If for the thin rim the turbulence degree
stayed nearly constant and relatively low Tu 4%, then for the thick
rim (t/s = 8) it reached the value Tu 28%, and maximum in Tu
distribution shifted to the outlet cross-section with a growth of the
rim thickness.
342
Fig. 5. Influence of separation rim thickness on temperature alteration along
the axis without swirl.
Fig. 6. Influence of separation rim thickness on turbulence intensity, when
mixing non-swirled flows.
The intensification of mixing processes with a rim thickening is
caused by the formation of a swirl trace behind the separation rim,
which leads to the additional flow turbulization. This conclusion is
confirmed by experimental data of [22] on the mixing of non-swirled
bounded jets.
In the swirled flow a temperature decrease along the axis at the
expense of a rim thickening occurs not so drastically as in the non-
swirled flow (Fig.7). The swirl of the peripheral flow leads to the
intensive damping of the turbulent pulsation in the near-axial area.
This mixing peculiarity of co-axial flows with a swirl is shown in
343
Fig.8. There one can see the turbulence alteration along a channel
length for different swirl angles at the inlet and constant rim size
t/s = 4. As for the case with the sharp rim (Fig.4), the flow turbulence
degree decreases with increase of the swirl angle, and its maximum
shifts from the initial cross-section.
Thus, an increase of the separation rim thickness and flow swirl
has an alternative impact on the mixing processes and turbulence in
the near-axial area. Moreover, in some cases the influence of these
factors may be commensurable and mutually compensating. This
conclusion is confirmed by Fig.9, where experimental points for the
wide rim (t/s = 4) and swirl of the peripheral flow (
s
= 74) are
relatively close to the data for the sharp rim and swirl absence.
Fig. 7. Temperature distribution along the axis with the thick rim and the
peripheral flow swirl.
Fig. 8. Influence of the swirl angle on turbulence intensity in the near-
axial area with the thick separation rim.
344
Fig. 9. Temperature on channel axis with varying swirl angle and separa-
tion rim thickness for constant flow rate of near-axial and peripheral flows.
Fig. 10. Generalization of experimental data on temperature alteration along
the channel axis. 1 - calculation by formula (3); 2 - calculation by formula
(2).
To estimate gas temperature along the channel axis with the swirl
of the peripheral jet and different thickness of the separation rim,
the empirical dependence is suggested:
+

_
,
1
]
1
1
1 01
0
0
1 25
0 8
. Re Re ,
, ,
.
.
s x r
r
D
(2)
where ( )
Re ;
, s x s s s
W x x

0
( )( )
Re
, 0 0 0 0 0 0 r s
W r , x
0
i s t he
length of initial region of the central flow.
345
The comparison of calculation according to this formula with ex-
perimental data is presented in Fig.10. One can see good correspondence
between the experiment and calculations.
For the case of the non-swirled peripheral jet the temperature along
the axis of the central flow was satisfactorily described by the de-
pendence for the turbulent flow regime:
+

_
,
1
]
1
1
1 0 25
0
0
1 25
0 8
. Re Re .
, ,
.
.
s x r
r
D
(3)
3. JET STABILIZATION IN SWIRL CHAMBERS
The experimental study of stabilization of near-axial jets was carried
out in a vortex chamber with a diameter of D
k
= 100 mm and the
length of L
k
= 150 mm (Fig.1,b). The peripheral flow of cold gas
was swirled by means of guides, whose slits were situated under different
angles. The near-axial jet, heated up to 100 C, was injected with
different velocities along the vortex chamber axis. Air from the chamber
was ejected through a diaphragm edge cap at the chamber outlet. The
flow rate ratio of the peripheral flow and the near-axial jet varied
in maximally possible limits from 0 to . The tangential Reynolds
number at the periphery of the chamber was Re
k
= V
k
R
k
/v = 510
3
210
5
.
The detailed description of this installation and measurement methods
is presented in [17].
The influence of the near-axial jet injection on radial distributions
of tangential and axial velocities is shown in Fig.11. Let us note the
main peculiarities of the vortex chamber aerodynamics in the presence
of the near-axial jet. The presence of the injected flow affected only
an area bounded by the diaphragm size (r < r
0
). At the periphery,
the experimental data with or without injection nearly coincide for
both the circular and axial velocity components. The input of the non-
swirled jet to the near-axial area causes significant deceleration of
the rotation motion, and the longitudinal velocity component at the
axis increases with increasing injection.
Experimental studies showed that the flow in the vortex chamber
with injection can be considered as a quasi-1D, i.e., radial distri-
butions of tangential velocity do not change over the chamber height,
and the injection affects the longitudinal velocity only in the near-
axial area. Thus, the main features of the flow pattern in common
vortex chambers without injection maintain for the case of injection
[8].
346
Fig.11. Profiles of tangential and axial velocities in a vortex chamber with
and without near-axial injection.
Fig. 12. Temperature distribution over a radius of a vortex chamber. a) G
0
=
1 g/s, b) Re
k
= 2.75 10
4
.
Temperature profiles over the radius of the vortex chamber for
different flow rates of the peripheral jet are shown in Fig.12,a. Gas
injection along the axis without peripheral flow (G
k
/G
0
= 0) results
in a quick jet mixing in the chamber volume, and even in initial cross-
sections the temperature does not change over the radius. Temperature
347
gradients in the mixing zone increase with increase of tangential velocity,
and the near-axial jet localizes in the narrow near-axial area due to
the damping of heat transfer processes.
The flow rate increase of the near-axial jet with steady param-
eters of the peripheral flow leads to more intensive jet expansion
(Fig.12,b).
Let us analyze the temperature distribution along the axis of the
vortex chamber for various cases of near-axial jet interaction with
the swirled peripheral flow. Maximal temperature changes over the
chamber height for various tangential Reynolds numbers for the periphery
Re
k
and constant flow rate of the near-axial jet G
0
=10
3
kg/s are
presented in Fig.13. Experimental data were treated adjusting the
temperature of complete mixing for different flow rate ratios.
It is obvious from Fig.13,a that the values of relative tempera-
tures at the chamber axis decrease with an increase of rotation in-
tensity of the peripheral flow. At first sight, this contradicts to the
impact mechanism of centrifugal forces on the damping of the tur-
bulent transfer in rotating flows. However, analyzing Fig.13,a, one
can see that the main temperature alteration occurs near the cham-
ber inlet, where the flow in the edge boundary layer can have a con-
siderable effect. Actually, the higher tangential Reynolds number Re
k
,
the higher gas flow rate passing over the edge from the periphery
to the chamber axis in the radial direction. The flow rate of cold
gas in the edge boundary layer may be calculated using ratios from
[8].
As it follows from Fig.13a the temperature gradient over the length
of the chamber with rotation intensification decreases in the area beyond
the edge boundary layer (X/D
k
> 0.15). This confirms the weaken-
ing of mixing processes in the main volume of the chamber at the
expense of mass forces.
At the next stage the experimental data of Fig.13a were recalculated
to the initial jet temperature. It was equal to the temperature beyond
the edge boundary layer. The results of this treatment are presented
in Fig.13b. Here, the relative temperature is (T
m
T
mix
)/(T
m
*
T
mix
),
where T
m
*
is the gas temperature at the axis beyond the boundary layer.
This treatment allowed us to exclude the influence of dilution of the
jet by the edge gas flow, and analyze the impact of mass forces on
turbulent mixing in the main volume of the chamber. It is obvious
that, on the contrary to the data in Fig.13a, the gas increase in the
peripheral flow leads to the stabilization of the near-axial flow and
weakening of convective heat transfer in the vortex chamber.
The generalization of experimental data on jets mixing in vortex
348
Fig. 13. Maximal temperature alteration over the chamber height.
chambers was performed with the use of various criteria. One of the
widely used generalizing criteria is the integral parameter:
K I R M W r dr R R WV r dr
k k x
r
k k x
R
R
k s
k

0 0
2
0
2
2 2
0

, (4)
representing the ratio of the impulse flux of the near-axial jet to the
moment of momentum flux of the peripheral stabilizing flow [4].
Calculating parameter K according to the conditions at the chamber
inlet, we will obtain:
K
W F
W V F
G W
G V
in
x
k xk k k
x
k k

0 0
2
0 0 0
.
(5)
349
Experimental data on the heat parameter of the jet, calculated with
a use of parameter K
in
, are presented in Fig.14. There one can see
the data for different flow rate ratios and for the constant diameter
of the exhaust orifice, whose boundary is marked by the dotted line
in the figure. It is obvious that in this form the experimental results
can be easily generalized. Moreover, with the jet heat size approximately
equal to the exhaust diameter, the jet stabilization occurs, when parameter
K
in
< 0.3. This corresponds to the results of [23]. The experimen-
tal results of this work demonstrate that the regime of the minimal
jet mixing with the peripheral swirled flow in a non-diaphragm channel
occurs, when parameter K is in the range 0.6 K
in
0.15.
The experimentally determined critical value of the swirl parameter
K
in
0.3, when stabilization of the near-axial jet is achieved, is proved
by calculation analysis of [9]. When mixing highly non-isothermal
or density heterogeneous flows, the ratio of tangential velocity at the
boundary of the near-axial jet to the longitudinal velocity compo-
nent V
m
/W
x0
was the most suitable stabilization parameter. This is
confirmed by the correlation of experimental and calculation data on
the maximum expansion of the plasma jet in the vortex chamber, presented
in Fig.15. The jet stabilization is obtained there at V
m
/ W
x0
1.
According to [17], this regime under isothermal conditions is achieved
at V
m
/W
x0
5. This manifests a significant influence of jet heat-
ing on the damping process of turbulence mixing in the swirl flow.
The current work was fulfilled with a partial financial support of the
Russian Fund of Fundamental Investigations (grant No. 96-02-19122).
Fig. 14. Heat boundary of a jet in a vortex chamber.
350
Referenes
1. M.F. Zhukov, et al., Electric Arc Generators with Interelectrode Insertions
//Novosibirsk:Nauka. - 1981. (in Russian).
2. E.P. Volchkov, G.R. Baldinov, V.I. Terekhov, Yu.N Tkach. Experimental
Investigation of Jet Stabilization in Near - Axis Region Plasma Re-
actor Contrib. Plasma Physics, 1988, v. 28, No. 1, p. 27-39.
3. A.I. Leontiev, E.P. Volchkov, V.P. Lebedev, et al., Heat Protection of
Plasmatron Walls // No-vosibirsk: Institute of Thermophysics SB RAS.-
1995. - 336 p. (in Russian).
4. A.K. Gupta, D.G. Lilley, and N. Syred, Swirl Flows // Abacus Press-
1984. - 588 p.
5. Chigier N., Beer J., Gregov D., Bassindale K. Jet Flames in Rotat-
ing Flow Fields // Combustion and Flame. - 1970, v. 14, p. 171-179.
6. Syred N., Beer J. Combustion in Swirling Flows: A review // Combustion
and Flame, 1974, v. 23, No. 2, p. 143-202.
7. I.V. Barmin, B.I. Katorgin, and V.I. Solonin, Concentration Fields of
Heterogeneous Gases in a Vortex Chamber // Proceedings of Moscow
High Technical College. - 1971. - No. 144. - P. 77-80. (in Russian).
8. S.S. Kutateladze, E.P. Volchkov, and V.I. Terekhov, Aerodynamics and
Heat and Mass Transfer in Limited Swirl Flows // Novosibirsk. - In-
stitute of Thermophysics. - 1987. - 287 P. (in Rus-sian).
9. G.R. Baldinov, E.P. Volchkov, N.A. Dvornikov, Ma Tun Tce, and V.I.
Terekhov, Gas-Swirl Stabilization of a Jet in Near-Axial Area of Plasma
Reactor // J. of Engineering Physics. - 1993. - Vol. 64. - P. 131-140.
(in Russian).
10. J. Beer, N. Chigier, T. Davis, K. Bassindale. Laminarization of Turbulent
Flames in Rotating Environments // Combustion and Flame, 1971, v16,
No. 1, p. 39-45.
11. E.P. Volchkov, S.Yu. Spotar, and V.I. Terekhov, Swirled Near-Wall Jet
in a Cylindrical Channel // Preprint. - Institute of Thermophysics SB
of USSR Academy of Sciences. - Novosibirsk. - 1982. - 42 p. (in Russian).
12. Yu.B. Kolesnikov and E.P. Sukhovich, Experimental Study of Turbu-
lent Characteristics in Axis-Symmetrical Swirled Flow // Izvestiya [Acad.
Fig. 15. Expansion regularity of the near-axial plasma jet, T
0
= 5500 K.
351
of Sciences of Latvian SSR], Series of Phys. and Appl. Sciences. -
1983. - No. 4. - P. 72-78. (in Russian).
13. V.K. Tchukin, Heat Transfer and Hydrodynamics of Internal Flows in
Fields of Mass Forces. - Moscow. - Mashinostroenie. - 1980. (in Russian).
14. E.P. Sukhovich, Experimental Study of Jet Mixing in a Limited Swirled
Flow // Izvestiya [Acad. of Sciences of Latvian SSR], Series of Phys.
and Appl. Sciences. - 1975. - No. 2. - P. 57-63. (in Russian).
15. B.P. Ustimenko, Processes of Turbulent Transfer in Rotating Flows /
Alma-Ata. - 1978. (in Russian).
16. E.P. Volchkov, G.R. Baldinov, V.I. Terekhov, and Yu.N. Tkach, Investigation
of Development Regularities of a Jet in Swirled Gas Flow // in: Generation
of Electric Arc Plasma Flows. - No-vosibirsk. - 1987. - P. 184-199.
17. E.P. Volchkov, V.I. Terekhov, and Yu.N. Tkach, Experimental Study of
the Mixing of Near Axial Jet with Peripheral Flow in Vortex Chamber.
- Preprint./ Institute of Thermophysics of SB Acad. of Sceinces of USSR,
No. 124-85. - 1985. - 42 p. (in Russian).
18. V.P. Lebedev and N.E. Shishkin, Influence of a Thickness of an Outlet
Rim of Tangential Slit on the Efficiency of a Swirled Film Cooling //
in: Near-Wall Jet Flows. - Novosibirsk. - 1984. - P. 14-20. (in Rus-
sian).
19. E.P. Volchkov, V.P. Lebedev, and N.E. Shishkin, Experimental Study
of Gas Film Cooling in a Tube // Izvestiya [Acad. of Sciences USSR],
Technical Sciences. - 1983. - Vol. 1. - No. 3. - P. 25-99. (in Russian).
20. E.P. Volchkov, N.A. Dvornikov, and V.I. Terekhov, To Calculation of the
Swirled Gas Film Cooling in a Cylindrical Channel // J. of Appl. Mech.
and Theor. Phys. - 1986. - No. 5. - P. 59-68. (in Russian).
21. E.P. Volchkov, V.P. Lebedev, and N.E. Shishkin, The Efficiency of Swirled
Gas Film Cooling on Adiabatic Surface // Swirl Effect and Its Use in
Techniques. Proceedings of the 2 All Union Scientific Conference. -
Kuibyshev. - 1976. - P. 217-220. (in Russian).
22. S.C. Kacker, L.M. Whitelaw. An Experimental Investigation of the Influence
of Slot-lip-thickness on the imperious-wall Effectiveness of the Uni-
form-density. Two-dimensional Wall Jet // Int. J. Heat and Mass Transfer,
1969, vol. 12, 9. - p. 1196-1201.
23. R.P. Lyane and Yu.V. Ivanov, Development and Mixing of Co-Axial Jet
in Axial-Symmetrical Swirled Flow // Izvestiya [Acad. of Sciences of
Latvian SSR], Physics, Mathematics. - 1971. - Vol. 20. - 1. - P.58-
65. (in Russian).
352
Low-density Plasma Jets: Production and
Investigation
L.I. Kuznetsov and V.N. Yarygin
Institute of Thermophysics, Siberian Branch of Russian Academy of Sciences, Novosibirsk,
Russia
1. INTRODUCTION
Experimental studies of the gas-dynamic structure of low-density jets
beyond a sonic nozzle at high pressure drops [13] have shown that
in a wide range of parameters, including modes of rarefied flows, the
geometrical configuration of the jet and the distributions of the density
and total pressure are self-similar when the value of the complex
Re Re
*
L
N = is constant. In this expression N p p =
0
is the ratio of
the stagnation pressure p
0
to the pressure in ambient space p
.
When a gas escapes into a medium with a temperature different
from the stagnation temperature of the gas of the jet, the distribution
of the parameters in the mixing zone depends on the temperature drop
and on the temperature levels. In this connection, it is possible for the
temperature factor to affect the structure of the jet as a whole, es-
pecially at small values of Re
that correspond to the transition to the

rarefied flow. The authors do not know of any analysis of these questions.
It should be noted that no systematic studies of the gas-dynamic structure
of rarefied high-temperature jets have been performed at all. At present
there are only single papers [4,5] containing data on the measurement
of the total pressure and density in individual modes of flow, and studies
of the physical processes in gas streams expanding from a plasma source
have been conducted without the necessary attention to the gas-
dynamic structure of the jets [68].
In the present paper, we describe the results of a study of the density
distribution in jets beyond a sonic nozzle and analyse the effect of the
temperature factor on the structure of the jets in the transition from
353
Low-density Plasma Jets: Production and Investigation
the mode of continuous flow to the scattering mode. In order to ex-
clude from consideration the effects leading to a change in the physico-
chemical properties of the gas at high temperatures (chemical reac-
tions, dissociation, excitation of internal degrees of freedom with a marked
change in the ratio of heat capacities) the study was conducted at relatively
low temperatures on a monatomic gas, i.e. argon.
At T < 16000 K, the excitation of electron levels can be neglected,
since under these conditions separations of the outer electrons (ioni-
sation) occurs preferably [9]. Under typical conditions with the use of
plasma heaters the degree of ionisation is small as a rule and does not
exceed 1% [5]. The recombination coefficient is of the order of
10
10
10
12
cm
3
/s [10,11] and the relaxation rate is low [12]. This
gives the reason to assume that the ionisation in the jet is practically
frozen-in and, consequently, the adiabatic index along a streamline
corresponds to a monatomic gas.
2. INSTRUMENTATION AND X-RAY REGISTRATION SYSTEM
The measurements were carried out in a vacuum chamber [13] equipped
with an electron gun, an X-ray registration system and an arc heater
(Fig.1). A three-electrode gun (5) generates a beam (12) whose energy
is up to 30 keV and the current is up to 50 mA. The accelerating voltage
in these measurements was 20 kV (except those where the electron
energy was varied), and the beam current ranged from 1 to 3 mA as
a rule. For the beam to enter the working chamber, a two-chamber
system of differential pumping (6) was used, which generates a working
pressure in the thermal cathode area of about 10
3
Pa under a pressure
in the working chamber p
up to 100 Pa.
The electron beam passed through the vacuum chamber and arrived
at an aluminium collector (13) with an entrance window 14 mm in
diameter. The electron beam current was measured by a milliammeter
(14) included in the collector circuit.
The X-rays were detected with a proportional counter, operated with
an argonmethane detector gas (10). The location of the measurements
was determined by the diameter of the electron beam and by the width
of a collimator slit (9). The collimator was positioned in front of the
entrance window of the counter. The spatial resolution was about 2
mm in all planes.
The signal from the counter comes to a recording device comprising
an amplifier, a discriminator, an electronic counter, and an integrator.
The arc heater was used as a gas-dynamic source. It was mounted
on a three-component mechanically driven traverse. The design of the
arc heater is shown in Fig.2. The arc heater is of the single-
354
chamber design with vortex stabilisation of the arc. It consists of three
main units: cathode (1, 2), anode (3, 4), and stagnation chamber with
a replaceable sonic nozzle (5) with diameter d = 15 mm. The working
gas can neutral gases (argon, helium, nitrogen) or oxygen containing
gases (CO
2
, air). In the first case, the cathode (2) is made of tung-
sten and in the second case a zirconium cathode is used. The arc heater
steadily works in the range of the mass flow rate of the gas, G = 0.15-
1.5 g/s, stagnation pressure, 510
4
510
5
Pa and provides equilib-
rium heating of the gas up to temperature T
0
= 10000 K. Tempera-
ture was determined from the energy balance of the method of the equi-
librium sound speed [14]. Peculiarities of work at lowered pressure and
the methods of accounting for the modes of operations are presented
in more detail in [14].
3. ELECTRON-BEAM X-RAY METHOD
Today, the electron beam is one of the main methods for investigation
of rarefied gas flows allowing us to determine a series of local parameters
[15]. The main idea of the active electron-beam method is to determine
gas parameters based on the radiation spectrum excited by a fast electron
beam. The above-mentioned spectrum covers the range from the X-
ray region to the visible one. The relation between the intensities excited
by electron beam spectra and local gas parameters are provided either
theoretically or by a calibration curve.
The electron-beam technique is characterised by a good spatial localisation
in combination with small disturbances in the region under consideration.
Ultraviolet and visible beam-induced radiation is most generally em-
ployed for diagnostics [13,16,17]. However, the application of these tech-
Fig.1 Density measurement circuit. 1 rough vacuum pump, 2, 3 diffusion pumps,
4, 7 electromagnetic lens, 5 electron gun, 6 system of differential pumping, 8
deviative lens, 9 X-ray collimator, 10 X-ray detector, 11 vacuum chamber,
12 electron beam, 13 collector cup, 14 milliammeter, 15 arc heater.
Fig.2 (right) Arc heater. 1, 2 cathode, 3, 4 anode, 5 stagnation chamber with a
replaceable sonic nozzle.
355
niques for the investigation of high-temperature gas flows is associ-
ated with difficulties. One of them arises when the temperature of gas
under study increases by up to several thousand degrees. In this case,
there appears underground gas radiation in the optical region that masks
the electron-beam-induced radiation. The use of the X-ray spectrum
region [18-20] is most appropriate for overcoming the above difficulties
and allowed the capabilities of electron-beam diagnostics of rarefied
gas and plasma to be increased.
These methods are based on the interaction of fast beam electrons
(with energies amounting to several tens of keV) with the Coulomb
field of the atom nucleus [18] or with electrons in the outer shells [20],
which generates respectively the bremsstrahlung or characteristic
X-ray radiation, the intensity of which is proportional to the concen-
tration of atoms of the investigated gas.
The concept of the density measurement by the bremsstrahlung was
suggested in [18,19] and developed in [20,21]. The integral intensity
of X-ray radiation, I, of the gaseous target is directly proportional to
the density number of the gas and is expressed as follows:
I cinZ
V V
V
i
d
=

2
(1)
Here c is the coefficient depending on the interaction cross-section
and detector aperture, V is the acceleration voltage of the electron beam,
V
d
is the threshold of registering by the detector of -quanta in the
region of soft X-ray radiation, Z is the atomic number, n is the atom
concentration, i is the electron beam current.
The integral pulse numbers of -quanta, N
j
, registered by the X-ray
detector is expressed as follows:
N c inZ
V V
V
i
d
= '
ln( / )
2
(2)
The characteristic X-ray radiation intensity obtained with the use
of the Tompson model for the determination of the ionisation cross-
section of internal atom shells [20] is expressed by
I Bin
V V
V
N
j
j
j
=
( )
~
2
(3)
where V
j
is the potential of j-shell ionisation, B is the parameter depending
356
on a statistical weight of level, the probability of transition and frequency
of -quantum (B = idem for some definite kind of atoms and line of
spectra).
It follows from (1)(3) that the intensities (or N) of bremsstrahlung
and characteristic radiation are directly proportional with the concentration
of target atoms.
Figure 3 presents the spectrum of X-ray radiation from argon in the
220 keV range, which was taken while moving a one-channel-wide
window of the differential discriminator over the energy spectrum
of radiation. Also shown are the peaks of the characteristic radiation
from S, Fe, and Cu. The peaks were recorded to identify the argon
K-lines. In the case of bremsstrahlung, the discriminator was set in
the integrated mode of operation and the discrimination level was about
5 keV. For work with characteristic radiation from argon the window
of the differential discriminator consisted of 6 channels (dotted lines
in Fig.3).
The important peculiarity of the electron-beam technique is the possibility
to make measurements at any temperature of the object to be stud-
ied, since the inner shells do not influence the measurement data. The
gas flow velocity does not also influence the results, since the time
of -quanta radiation ranges from 10
8
to 10
13
s.
The investigations of bremsstrahlung of some gases confirm the theo-
retical assumptions and allow the universal dependence of radiation intensity
on the determining parameters to be plotted (Fig.4). The investigation
of the characteristic radiation intensity [20,21] dependence in Ar over
the range i = 0.0720 mA, V = 325 kV, n = 510
14
4.210
15
cm
3
confirm a linear dependence of the characteristic radiation intensity,
electron beam current and concentration (Fig.5).
Fig.3 X-ray spectrum from argon, exited by an electron beam with energy 20 keV.
Fig.4 (right) Universal calibration curve.
357
Fig.5 Calibration curve for characteristic radiation intensity.
Fig. 6 (right) Beam cross-section.
The estimates show that the radiation absorption between the de-
tector and the electron beam, as well as the effect of fluorescent radiation
and disturbances created by the electron beam in the gas do not in-
fluence significantly the accuracy of measurements in the X-ray
region. The upper limit of the method with respect to atom concen-
tration is determined mainly by electron beam scattering in the medium.
The lower limit is determined by the apparatus sensitivity, detector aperture
and beam current. Thus for A n
max
is about 3l0
15
cm
3
with V =
20 kV and 10 cm beam length, and n
min
is about 310
12
cm
3
with i=
1 mA, a spatial resolution of about 1 mm
3
, the relative solid angle of
the detector of 10
5
, and the signal integration time less or, approxi-
mately, equal 10 s.
It should be noted that the measurement technique for the local gas
density developed by characteristic radiation increases the possibilities
of electron-beam diagnostics, since it allows measurements of partial
densities in multicomponent gas mixtures and to make investigations
in the vicinity of walls.
Fig.7 Axial profiles of density for Re
L
= 100.
358
Besides it was established by special measurements that the use of
X-ray region leads to a better spatial localisation of measurements. A
typical beam size determined by the radiation intensity in X-ray region
is much less than that measured in an optic region under the same conditions
of beam formation (Fig.7). The matter is that the secondary electrons
do not effect the radiation excitation in the X-ray spectrum region, since
their energy level does not exceed tens electron-volt. These data were
obtained by using electric gages [22].
4. GAS-DYNAMIC STRUCTURE OF LOW-DENSITY PLASMA
JETS
The principal operating parameters (see Table) were varied in the following
ranges: T
0
=2905200 K, d = 1.15 mm; G = 0.0031.4 g/s; Re
*
=300-
40000; N = 30030000. The temperature of ambient gas T
is 290 K.
The given measurement results of density in
N x N /
or
N y N /
coordinates are shown in Figures 813, where
= = / /
0 0
n n is the ratio of density to that in the stagnation chamber,
x x d y y d x = = / , / , is the distance along jet axis from nozzle, y is the
distance from the jet axis.
The possibility of generalising the data on the density distribution
in the jet at fixed values of Re
L
and T T
0
/

in a wide range of val-
ues of Re
L
>100 and N > 100 is shown, first of all, on the basis of the
experiments.
As an example, the results of measurements of the axial density
distribution are presented in generalised form in Fig.7 (for Re
L
=100),
Fig. 8 Cross-section of density for Re
L
= 100.
359
Table The operation modes parameters under experimental investigations
No T
0
, K G, g/s p
???
,
tor
d, mm Re
*
N Re
L
1 290 1.14 534 2,125 40100 550 1010
2 290 1.14 185 2.125 40100 4480 600
3 290 0.74 104 5.0 10900 866 371
4 1510 1.39 1260 3.11 9900 860 341
5 290 0.68 85 3.135 6100 2080 312
6 290 0.56 08 2.125 19700 4150 307
7 290 0.423 41 3.115 10000 3520 168
8 2000 1.1 590 3.115 6600 1650 162
9 290 0.236 71 1.10 15750 9400 162
10 2030 0.666 970 3.135 3950 610 160
11 290 0.389 23 2.110 30100 9200 105
12 1500 1.030 146 2.116 10900 10800 105
13 1510 0.729 223 5.0 3220 950 105
14 290 0.222 30.8 5.0 31.30 925 103
15 2970 1.13 407 5.0 3360 1110 101
16 4970 0.92 1290 3.11 3180 1010 100
17 290 0.196 20.6 2.12 68.40 7050 81
18 290 0.118 34 1.10 7900 9400 81
19 3840 0.857 572 2.14 5000 3900 81
20 5170 0.898 920 3.21 2910 1300 81
21 5060 0.85 926 3.11 2890 1300 80
22 290 0.20 18.7 2.12 7020 7750 80
23 2000 0.666 239 3.13 3970 2460 80
24 2020 0.913 179 2.14 7900 9700 80
25 4910 0.295 413 3.12 1030 996 33
26 1510 0.156 83.8 1.10 3160 28500 19
27 290 0.0125 11.6 3.11 297 370 16
28 4710 0.302 41 3.11 1090 10000 11
29 4700 0.148 24.8 5.0 331 3230 5.6
30 290 0.005 2.7 1.10 332 5050 4.7
31 290 0.0027 1.9 2.12 100 1040 3.2
p
, t or
360
while transverse density profiles in three different cross sections of
the jet are presented in Fig.8 (the conditions are denoted by the cor-
responding mode number in Table ). As is seen, the data are well generalised
for fixed values of Re
L
and T
0
. Similar generalisation also occurs for
other values of Re
L
(in the range Re
L
=501000). On the basis of the
data obtained it can be concluded that in the indicated range of Re
L
the density distribution along the length of the jets studied is self-similar
with respect to N when the values of T
0
and Re
L
are fixed. This is
in agreement with the conclusions made on the basis of a study of nitrogen
jets at T
0
=T
=300 K [3].
The next stage of the studies was the analysis of the effect of the
temperature factor on the structure of low-density jets. The analysis
of the experimental data is considerably simplified by the self-similarity
with respect to N in the density distribution in the jets which was established
above. The results of measurements of the axial density distributions
with a fixed value Re
L
= 80 and different T
0
= 2905200 K are pre-
sented in Fig.9. As seen from the graph, agreement of the axial density
distributions for different T
0
is observed in the first cycle of the jet.
In the region behind the Mach disk the temperature factor begins to
affect the axial density distribution and the experimental data become
stratified. With an increase in the value of Re
L
the region in which the
temperature factor has no effect spreads down-stream, and when
Re
L
= 150 generalisation of the experimental data with respect to the
axial density distributions at different values of T
0
(T
0
= 2902000 K)
is observed in the entire length of the jets studied x N / 3 With a
decrease in Re
L
(Re
L
< 80) the temperature factor already begins to
affect the axial density distributions in the region of the initial section
of the jet (the first cycle).
Transverse density profiles in the cross section x N / =0.44 with
Fig.9 Axial profiles of density for Re
L
=80 and different stagnation gas temperature.
361
Fig.10 Density profiles in the cross-section x N / = 0.44.
Re
L
= 100 and T
0
= var. are presented in Fig.10,a. As can be seen, the
transverse density profiles for different T
0
are satisfactorily general-
ised only in the nonviscous zone of the core of the jet. In the region
of the mixing zone, which merges with the zone of the compressed layer,
the density profiles are stratified with respect to the temperature factor.
The transverse density profiles at large values of Re
L
have a similar
form. At smaller values of Re
L
, the effect of the temperature factor
on the generalised functions for the density already shows up in the
core of the jet and at its axis (Fig.10,b,c).
The reorganisation of the flow pattern with a change in Re
L
can be
Fig.11 Axial density profiles with a change value of Re
L .
362
traced on the basis of the results (Fig.11). At large values of Re
L
(Re
L
= 300), the density distribution in the region of the nonviscous core
of the jet is close to that calculated on the basis of a model of a nonviscous
gas. The position of the Mach disk agrees with the data for high-density
streams. In this case, the width of the shock-wave front is small and
the density ratio at the Mach disk is close to four, which is in accordance
with the limiting value for a direct shock wave in a monatomic gas with
M >> 1. The measurements of the transverse density profiles at Re
L
=370
and T
0
=290 K show (Fig.12) that an extended annular layer of com-
pressed gas is retained behind the Mach disk and the mixing zone does
not penetrate to the jet axis at least out to x N / = 2 . A similar pattern
is observed in high-temperature jets. The presence of the annular flow
behind the Mach disk (without passage through the speed of sound at
the axis) was also noted in the studies of the authors of [3].
The reorganisation of the described flow pattern occurs with a decrease
in Re
L
, the suspended shock waves, the Mach disk, and the mixing layer
gradually become thicker; the merging of the mixing zone at the jet
axis occurs behind the Mach disk and the region of merging moves
upstream. This reorganisation also leads to a change in the nature of
the flow behind the Mach disk: the annular viscous layer becomes thicker
and the merging of the mixing layer at the jet axis leads to an increase
in the density behind the Mach disk, which is clearly seen in Figs.12
and 13. This increase in density propagates upstream with a decrease
in Re
L
. At Re
L
= 80, the merging of the mixing layer occurs close behind
the Mach disk. The maximum density increase in the region of the Mach
disk is reached with Re
L
= 3080, which indicates the merging of the
viscous layer directly in the region of the Mach disk and the latter cannot
be considered as an isolated shock wave.
The thickening of the viscous layer strengthens the ejecting effect
on the flow behind the Mach disk and the characteristic cyclicity of
Fig.12 Cross-section of density behind the Mach disk.
363
the flow (barrels) appears. Such cyclicity is well seen, for example,
in Fig.12 at Re
L
= 80. With a further decrease in Re
L ,
the density of
shock waves degenerate, the flow becomes almost fully viscous (the
merging of the viscous layer already occurs near the nozzle cut), and
at Re
L
=10 the transition to the so-called scattering mode is observed.
The rise in density at Re
L
=10 and T
0
/T
>>1 is due to cooling of the

gas in the mixing zone.
The qualitative changes in the structure of a jet of the monatomic
gas beyond the sonic nozzle with variation in the rarefaction and the
temperature factor have been clarified as a result of the experiments
and generalisations performed. The quantitative data obtained on the
density distribution can be used for comparison with calculation in the
development of a model of a jet.
References
1 V.S.Avduevskii, et al, Effect of viscosity on the flow in the initial sec-
tion of a strongly underexpanded jet, Bulletin of the Academy of Sci-
ences of the USSR, No.1 (1971).
2 N.I.Kislyakov, A.K.Rebrov, and R.G.Sharafutdinov, J. of Applied Mechanics
and Technical Physics, No.1 (1973).
3 V.V.Volchkov, et al, J. of Applied Mechanics and Technical Physics, No.
2 (1973).
4 N.I.Yushchenkova, S.A.Lyzhnikova, and V.I.Nemchenko, In: The structure
of supersonic jets of gas and low-temperature plasma. Transfer Ef-
fects in a Low-Temperature Plasma (in Russian), Nauka i Tekhnika,
Minsk (1969).
5 R.B.Fraser, F.Robbin and S.Talbot, Phys. Fluids, 14 2317 (1971).
6 M.Sinclair, A.A.Sonin, and J.H.de Leeuw, Phys. Fluids, 10, No.4 (1967).
7 G.A.Lukyanov and G.V.Petukhov, High Temperature, 7, No.5 (1969).
8 V.M.Goldfarb and I.E.Kostygova, In: Optical study of a supersonic ni-
trogen jet. Study of Low-temperature Plasma Sources [in Russian], Vol.
466, Izd. LGPI, Leningrad (1971).
9 E.Resler, S.-C.Lin, and A.Kantrowitz, J. Appl. Phys., 23, 1390 (1952).
10 W.A.Clayden, Langmuir probe measurements in the R.A.R.D.E. plasma
jet, in: Rarefied Gas Dynamics. Proc. of 3rd International Symposium,
Vol.2, Academic Press (1963).
11 V.M.Goldfarb, et al, J. of Applied Mechanics and Technical Physics,
No.1 (1967).
12 R.H.Kirchhoff and L.Talbot, AIAA J., 9, No.6, 1098 (1971).
13 L.I.Kuznetsov and V.N.Yarygin, In: X-ray bremsstrahlung of gas targetsand
its use for measuring the local density of a gas. Apparatus and Methods
of X-Ray Analysis, Mashinostroenie, Leningrad, 11 (1972).
14 S.S.Kutateladze and V.N. Yarygin: Academician S.S. Kutateladze, Se-
lected Papers (in Russian), Novosibirsk, Nauka, (1989) 246.
15 L.A.Gochberg, AIAA Paper, 94-2635: 43 (1994).
16 E.P.Muntz and D.J.Mardsen, In: Electron Excitation Applied to the Ex-
perimental Investigation of Rarefied Gas Flows. Rarefied Gas Dynamics,
Acad. Press, N-Y-L, 2:495 (1963).
364
17 A.A.Bochkarev, et al, In: Measurement of Gas Flow Parameters Using
an Electron Beam, Experimental Methods in Rarefied Gas Dynamics
(in Russian), Inst. of Thermophysics, Novosibirsk (1974).
18 C.A.Ziegler, et al, RSI, 35:450 (1964).
19 A.N.Trohan, J. of Applied Mechanics and Technical Physics, No.3 (1964).
20 L.I.Kuznetsov, A.K.Rebrov, and V.N.Yarygin, Diagnostics of Ionized Gas
by Electron Beam in X-ray Spectrum Range, Proc. of 11th Intern. Conf.
on Phenomena in Ionized Gases, Prague (1973).
21 N.G.Zharkova, et al, High Temperature, No.1 (1976).
22 L.I.Kuznetsov, et al, In: Investigation of Diagnostic Electron Beam Plasma.
Diagnostic of Rarefied Gas Flows (in Russian), Inst. of Thermophysics,
Novosibirsk (1979).
3 6 5
Effect of Arc Current Modulation on Thermal Plasma Flow
Effect of Arc Current Modulation on
Thermal Plasma Flow in Plasma Torches
Farih A. Salyanov
ABSTRACT
In this paper, the theoretical investigation of the effect of arc current
modulation on the thermal plasma flow in plasma torches is described.
The basis of the nonstationary mathematical model is a set of differential
equations of energy, mass and momentum conservation, and the integral
form of Ohms law. Unlike the previous well-known nonstationary theories,
this boundary value problem takes into account axial acceleration of the
plasma flow, radius variability of the arc column and arbitrary law of the
alternating component of arc current in plasma torches.
As follows from the general solution of this problem, modulation of
arc current leads to the existence of the longitudinal waves of tempera-
ture and electric conductivity of the plasma, electric field strength and
other parameters of the thermal plasma flow. The influence of modula-
tion of the parameters on the processes in plasma torches and plasma flow
is studied in detail.
1. INTRODUCTION
Investigations of the thermal plasma flows and processes in plasma
torches are of considerable interest due to a wide range of applications
of high-heat flows. At the present time, plasma techniques can be
used in the treatment of materials, in metallurgy and energetics, plasma
spraying, plasma chemistry and other new branches of plasma technology.
The possibility of theoretical studies of the processes in arcs stabilized
in a tube was shown for the first time in the classic paper of H. Mekker
[1]. By now, the basic phenomena and processes in arc plasma torches
have been well investigated, the basis of their theories and basis of
the theory of thermal plasma flows have been created. The funda-
mental results of the thermal plasma problems and technologies were
Institute of Mechanical Engineering RAS
2/ 31 Lobachevsky Str., PO Box 99,Rus - 420503 Kazan,
3 6 6
obtained by M.F. Zhukov, E. Pfender, O.I.Yasko, I.G.Panevin and
their colleagues and many other famous scientists. These results are
published in many classic books and review papers, for example
[26].
It should be noted that most of the results on plasma problems
were obtained assuming the stationarity of processes. The peculiarities
of arc plasma torches with the alternating current are less investi-
gated. The well-known theories of the nonstationary arc current of
O.Mayr [7], A.M. Cassie [8] do not consider the influence of the
gas thermodynamical properties and the flows velocity. These theories
do not also consider the effect of the plasma torches construction
on the arc parameters. As us known the main difficulties in the theoretical
studies of these plasma torches were caused by the presence of tran-
sitional processes at the current pauses. In the works of O.M.Novikov
and his colleagues, Y.C.Svirchuk and some other scientists, aconsiderable
progress in this part of plasma physics was achieved, (see, for ex-
ample [911]).
An interesting direction in the area of plasma physics is associ-
ated with nonstationary problems considering the influence of small
perturbations on the direct-current arc. The investigations in this field
can have many applications in plasma technologies. In works [12,13]
and others, the influence of acoustic perturbation on the arc is ex-
perimentally investigated. It is established that the presence of acoustic
waves results in oscillations of current, electric field intensity, temperature,
velocity of the gas, and other parameters of plasma and plasma torches.
Our theoretical and experimental investigations also showed a strong
effect of the arc current modulation on the characteristics of plasma
torches and the properties of plasma flow [1417]. For example, it
was proved that the variation of the frequency and amplitude of al-
ternating component of the current changes active and reactive components
of the arc impedance, as well as the electrodynamic characteristics
and conditions of dynamic stability of arcs in plasma torches. The
present paper develops the theoretical ideas that were used in our
previous work.
2. MATHEMATICAL MODEL
The arc column stabilized in a tube with a gas flow is described by
a set of nonstationary nonlinear equations. The general set of these
integro-differential equations for the thermal plasma flow in a plasma
torch is derived and discussed in details in studies by M.F.Zhukov
and his colleagues [2,5 and other]. Therefore, it is not necessary to
give these equations completely in this paper.
3 6 7
It is well-known that, at present, it is not possible to get the general
solution of the general system of equations for the thermal plasma
flow in a plasma torch. However, having made a series of simplifying
assumptions, it is possible to compose some mathematical models for
the area of thermal plasma physics, and obtain the analytical solution
of their equations.
Analysing the phenomena in the thermal plasma flow and heat transfer
processes in a plasma torch, one can see that Joules dissipation, radial
conduction heat transfer, axial convective heat transfer, and radia-
tion flux play the basic role in the equation of energy conservation.
These processes of heat transfer make the fundamental contribution
in the balance of energy. The simplifying assumptions are physically
substantiated, discussed many times and used in the well-known
theoretical studies, for example [2,5,1417]. Thus, taking these as-
sumptions into account, the processes of heat and mass transfer in
the area of the arc column can be described by the following com-
bination of equations:

W
t
W
l z
R r
r
r
r
E +

_
,
+
1
2
2
(2.1)
( )

t l z
+
1
0
(2.2)
t l z
P
l z
+
(2.3)
I z t R E z t r z t rdr ( , ) ( , ) ( , , )
2
2
0

(2.4)
The following notation is used: r, z are the cylindrical coordinates
referred to R and l, respectively; R, l are the channel radius and its
length; t is the time multiplied by ; , , , W, , , P, are density,
the axial component of velocity, heat conduction, total energy per
unit mass, electric conductivity, emissivity, and gas pressure; is
the frequency of oscillations; I is current; E is the electric field intensity;

1
/R,
1
radius of the arc column.
It is seen that our mathematical model consists of gas dynamic
3 6 8
and thermodynamic problems. First, let us consider the gas dynamic
problem.
3. THE GAS DYNAMIC PROBLEM
In accordance with the adopted mathematical model, the laminar gas
flow under the plasma torch conditions can be described by the Euler
and continuity equations in the form:
( )

t l z
+
1
0
(3.1)

t l z
P
l z
+
(3.2)
Let us take advantage of the Rieman method [19] and assume that
( ), ( ) , ( , ) P P P P z t (3.3)
Taking into account expressions (3.3) allows one to write the set
of equations (3.1) and (3.2) as

P
t
P
P
l z
+ +
1
]
1
1
1
1
1
0
(3.4)
P
t
P
P
P
l z
+ +
1
]
1
1
1
1
0
(3.5)
Equations (3.4)(3. 5) will correlate under the indispensable condition
of
3 6 9
1

P
P
P
(3.6)

P P
_
,

2
(3.7)
Thus, for the (P) function one obtains

P
P
t
2
(3.8)
( )

P
P
d P t
2
(3.9)
The thermodynamic processes in the arc column can be described by
the polytropic equation in the form

P n
c c
c c
n
p
v
,
(3.10)
where n is the polytropic index, c the specific heat of the polytropic
process, A = const.
It is well known from the classical thermodynamics that the equation
of the polytropic process contains as its particular case the isothermal
process (at n = 1) and the adiabatic process (at n = 1/, where
is the adiabatic constand).
By integrating the right part of equation (3. 9) and taking into
account expression (3.10), one obtains
( ) P
n
A n
P const
n
t
2
1
1
2
.
(3.11)
3 7 0
We can see that equation (3.11) allows one to calculate the function
depending on gas pressure P. The constants in formula (3.11) can
be obtained from physical conditions of the given problem.
Equation (3.11) can be transformed for the conditions
( )
, , / , , P m n n
m m
1 1
(3.12)
and one obtains the expression for function ()
( )
t
Bm
m
const
m
2
1
1
2
.
(3.13)
Analysing equations (3. 11) and (3. 13), one can see that the condition
= const implies that the process is isothermal or isobaric. Hav-
ing multiplied equation (3. 13) by gas density , one can see that
the theoretical condition = const assumes the condition = const.
Thus, when modelling the thermal plasma flows in plasma torches,
it is necessary to take into account the axial acceleration of the plasma
flow and the variations in the gas expenditure in the channel.
4. THE THERMODYNAMIC PROBLEM
In accordance wi t h our mat hemat i cal model , t he nonst at i onary
thermodynamic problem on the thermal plasma flow in the area of
the arc column in the plasma torch can be described by the set of
equations:

W
t
W
l z
R r
r
r
r
E +

_
,
+
1
2
2
(4.1)
( )

t l z
+
1
0
(4.2)
I z t R E z t r z t rdr ( , ) ( , ) ( , , )

2
2
0

(4.3)
As is seen, we have rather complicated nonstationary and non-
linear set of the integro-differential equations. To get the general solution
of this set, the function of the thermal potential S
1
and total energy
3 7 1
per unit volume N, can be introduced,
where
S d T N W
o
T
1

,
(4.4)
Let us assume the linear approximations of the functions N, ,
depending on the function S, where S = S
1
S* is the value of the
function S
1
at r = .
In accordance with the adopted formulation of our thermal problem,
the set of equations (4.1)(4. 3) can be written as
S
t l
S
z
a
r r
r
S
r
c E S b S +

_
,
+
1
2
1 (4.5)
I z t R E z t S r z t r d r
s
( , ) ( , ) ( , , )
2
2
0

(4.6)
Here
a R N c a R c a R b c
l z
s s s
+
( ) , , ,
2 1
1
2
2
2
1 2

We can solve equations (4.5) and (4.6) for arbitrary initial and
boundary conditions in the form
S r t r t S r z r z S z t
S
r
r
( , , ) ( , ) , ( , , ) ( , ), ( , , ) , | , 0 0 0 0
1 2 0

+
0 0 0
1 0 ( ), ( ) ( ) , . z z k z k
It is seen that this nonstationary thermal problem takes into account
the axial acceleration of the thermal plasma flow, variability of the
expenditure and compressibility of the gas, radius of the arc column
and the arc current in the channel of the plasma torch.
The nonlinear parabolic equation (4.5) with variable coefficients
can be solved by means of transition to the new independent coor-
dinates,
(4.7)
3 7 2
r
r
x z t
b k
a z b
l
1
0 0
0
0
2

, , ( ), .
(4.8)
In this way, equation (4.5) can be rewritten as
b
S
x t
b
z
r S
r
a
r
r
r
S
r
c E x S b x S
_
,

_
,
+
1
1
2
1
1
1
1
1
2
1
( , ) ( )
(4.9)
One can obtain the general analytical solution of equation (4.9)
under the indispensable condition

2 2
1 1
t
b
z
k k c o n s t +
_
,
, .
(4.10)
Then we have
b
S
x
a
r r
r
S
r
k r S
r
c E S b S

2
1 1
1
1
1 1
1
1
2 2
1
2
2
_
,
+ +
(4.11)
We shall seek for the solution of the equation in the form
S r x
c
b
E x
x
d x U r x
x
( , , ) exp
( , )
( )
( , , )
1
1
0
2
0
1
_
,
(4.12)
From formulas (4.11) and (4.12) we get a linear separable vari-
ables equation, with respect to the function U r x ( , , )
1
,
b
U
x
b U
a
r r
r
U
r
k r U
r
2
1
2
1 1
1
1
1 1
1
2
+

_
,
+
(4.13)
Let us take advantage of the Fourier method
U r x X x V r ( , , ) ( , ) ( )
1 1
(4.14)
3 7 3
Now, from equation (4.13) we get the set of equations
d
d r
r
dV
d r
k r
a
d V
d r
r V
n
1
1
1
1 1
2
1
2
1
2
0
_
,
+ +
(4.15)
( )
b
x
X x b a
n

2
1
2 2
ln ( , +
(4.16)
The solution of equation (4.15) can be written in terms of degenerated
hypergeometrical functions,
V r
r
n
n
( ) , ;
1
2
1
2
1
4

_
,
(4.17)
where
n
(finite r = 0 ) are eigenfunctions of the equation describing
the arc column with the account of the molecular and convection heat
and mass transfer,
n
n
n n n n
m
n
r
m
m
r
2
1
2
2 2 2 2
2
1
1
2
1
4
1
2 1
4
, ;
( ) ( )...[ ( ) ]
( !)
;
_
,

+
+ + +
_
,
(4.18)

k
a
k
1
1
,
is determined by formula (4.10).
The proper values
n
of the functions may be obtained from the
equation
n
n
r
2
1
2
1
4
0 , ;
_
,

(4.19)
The general properties of the functions (
n
,,r
1
) were described
in [20], where 0 the functions
n
transfers in the Bessels func-
tion J
0
(
n
, r).
The solution of equation (4.16) is dependent on the form of the
function , which is determined by condition (4.10). In case of the
stationary arc current with modulated alternating component we can
3 7 4
assume that (z). This ad hoc assumption leads to

2
0
2
0
1 + k z ,

0
2 1
0 0
2
k
b k

k
a
1
0
2
0 0
2
b k
a
(4.19)
where
0
is the radius of the arc column at z= 0.
Having taken into account these conditions, we can write the solution
of equation (4.16) as
X x f
c x
b k
n
n
( , ) ( )exp
( ( ) )

_
,

2 1
2
0 0
2
(4.20)
where
n
n
a b k
b k
+ 2
2
2
0 0 0
2
0 0 0
2
Thus, the solution of equation (4.11) may be presented in accordance
with the obtained expressions and formulas. Coming back to the initial
coordinates, we can represent the solution of equation (4.5) in the
form
) (
S r z t
c
b
E x
x
d x
f
c
b k
z
r
z
n
n
n
n
n
( , , ) exp
( , )
( )
( ) exp ( ) , ; .
1
]
1
1
1
]
1

_
,
1
0
2
0
2
0 0
1
2
2 2
2
2
1 1
4
(4.21)
The functions
( ) f
n
, f f t
b k
k z
n n
( ) +
_
,
2
1
0 0
0
, may be ob-
tained from (4.21) and initial and boundary conditions (4.7). Thus,
we have
f A at
n n
( ) ( ), >0
(4.22)
3 7 5
f
B Y
b k
t
b k
k z
n
n
n
n
( )
( ) ( )
_
,
_
,
'
1 0 0
2
0 0
0
2
2
2
1 ,at < 0, (4.23)
where

( )
( )
b k
k
0 0
2
0
4
4
(4.24)
( )
( )
Y z
c
b R
I x
x
r x r d r d x
c
b k
s
z
( ) exp
( , )
( )
( , )
1
]
1
1

'
1
0
2
2
2
2
0
2
2
0 0
0 2
0 2
1

(4.25)
The functions ( ) ( ) ( )
n n
Y , , are obtained from A
n
(t),
B
n
(z) and Y (z) with the help of formulas (4.24) and (4.25). The
functions A
n
(t), B
n
(z) are the Fourier coefficients of the expansion
in degenerated hypergeometrical functions
n
n
r
2
1
2
1
4
, ;
_
,
of the
functions
( )
1
r t , and ( )
2
r z , , within the interval 0 1
1
r . The
coefficient ( )
n
0 at 0 , and ( )
n
z 0 at > 0.
Thus, one obtains the general analytical solution of equation (4.5)
S r z t
I z t
R E z t
s
( , , )
( , )
( , )

2
2 2

(4.26)
_
,
'

n n
n
n
n
n
n
n
n
r
( ) , ;
( )

2
2 2
2
1
1
2
1
4
3 7 6
( )
( )
+

_
,
_
,
_
,

n
n
n
n
n n
n
n
n
n
n
b k r
b k

2 0 0
1
2
2 2
2
2 0 0
2
1
2
1
4
2
, ;
where

n n
n
r r d r
_
,
0
1
2
1
2
1 1
1
4
, ;
(4.27)
Expressions (4.6) and (4.21) provide one with the nonlinear in-
tegral equation for the electric field intensity E(z,t):
( ) ( )
( )
( )
z t E z t
c
b
E x
x
d x
z
, , exp
,
_
,
1
0
2
0
(4.28)
where
( )
( )
( )
z t
I z t
R F z t
s n
,
,
,
2
2 2

(4.29)
( ) ( ) ( ) ( )
F z t f
c
b k
z
n n
n
n
n
, exp
_
,
1
2 2
0 0
2
1
(4.30)
Under condition I = const, equation (4.28) can easily be trans-
formed into the Bernoulli equation, the solution of which is known
very well. In this paper we assume I = I (z, t). Therefore, we shall
seek for the solution of eq. (4.28) by the integral method. Equation
(4.28) can be written as
( )
( )
( )
( )
( )
( )
2 2 2
1
0
2
0
1
0
2
0
1
0
2
0
c
b
x
x
d x
c
b
E x
x
c
b
E x
x
d x d x
z z z
, ,
exp
,
_
,
_
,
_
,
(4.31)
Integrating the right hand side of equation (4.31), one obtains
3 7 7
( )
( )
( )
( )
2 2
1
0
2
0
1
0
2
0
c
b
E x
x
c
b
E x
x
d x d x
z z
,
exp
,
_
,
_
,

(4.32)
( )
( )
_
,
exp
,
2
1
1
0
2
0
c
b
E x
x
d x
z
Taking into account the expressions obtained, we can write

( )
( )
( )
( )
exp
, ,
c
b
E x
x
d x
c
b
x
x
d x
z z
1
0
2
0
1
0
2
0
1
2
1
2
_
,
1
]
1
1
(4.33)
As a result, we have the general solution of equation (4. 28) in
the form
( )
( )
( )
( )
E z t
z t
c
b
x
k x
d x
z
,
,
,
+
+
_
,
1
2
1
1
0
2
0
0
(4.34)
Thus, we have obtained the general solution of the nonstationary
thermodynamic problem described by the set of equations (4.1)(4.3)
for the arbitrary initial and boundary conditions in form (4.7) and
accounting for the axial change of the arc current in the plasma torch.
5. DISCUSSION. SPECIAL CASES
It is seen from formulas (4.26) and (4.34) that the modulation of the
arc current leads to the existence of the longitudinal damped waves
of the thermal potential S(r,z,t), the total energy per unit volume N,
the electric field intensity E(z,t), the electric conductivity of the plasma
and other parameters of the thermal plasma flow and arc column in
the plasma torch. From solutions of the set of equations of our
mathematical model it can be seen that at > 0 the properties of
the arc column and the thermal plasma flow are independent of the
3 7 8
boundary conditions
2
(r,z). At < 0 these properties are independent
of the initial conditions -
1
(r,t). To study the solutions obtained
in detail, let us examine some special cases of our thermodynamic
problem.
5.1 Let us introduce a function ( )

1 1 1
1
2
2
2
1
4
r t A
r
, , ;
_
,
,
and assume the modulation law is
( ) ( ) I z t I i t , cos +
0
1 , where i
I
I
m
<
0
1,
I
m
is the amplitude of the alternating component of the arc current,
I
0
is the stationary current. Let us consider the cases when > 0.
Then, from formulas (4.26), (4.29), (4.30) and (4.34), we have
( )
( )
( )
S r z t
I t
R E z t
s
, ,
,

2
2 2
1

1
1
2
2
2
1
4
, ;
_
,
r
, (5.1)
( )
( ) ( )
( )
z t
I i x
c
b k
R A
c
b k
s
,
cos , exp
exp
_
,
_
,
0
2 1
2
0 0
2
0
2
1 1
2
0 0
1
2
2
2

, (5.2)
( )
( )
( ) ( )
( ) ( ) ( )
( )
( )
( )
E z t
E i t
z
c
b k
c E
b
c
b k
i x k x
x
c x
b k
d x
z
,
cos
exp
exp
cos ,
exp
,
_
,
_
,
+ +

_
,
_
,
1
]
1
1
1
1
1
0
2 2 2
0 0
1 0
2
0 0
4 2
0 0
2
0
2 1
2
0 0
0
1
2
1
2
1
1
2
4
1 1
4

(5.3)
E
I
R A
s
0
0
2
1 1
2
_
,
1
2
0 0 0
2
1
2
a
b k
( ) ( )
x k x + 1
0
3 7 9
Here E
0
is the electric field intensity in the section of the channel
at z = 0.
The integral expression in (5.3) can be written as
( ) ( ) ( ) ( )
( )
( ) z t i x
c
b k
x dx
k
i
d
z
o
z
, cos , exp +

_
,
_
,
'
1
4 2
1
2
2
2
0 0
0
4 3
0
2

(5.4)
( ) ( ) +

_
,

_
,

1
]
1
1
+

2
2 2
0 0 0 0
0 0
i
b k
d
b k
d
z z
cos cos sin sin

( ) ( ) +

_
,
_
,

1
]
1
1

i
b k
d
b k
d
z z
2
0 0
0
0 0
0
2
2
4
2
4
cos cos sin sin

Here
( ) ( )
( )

_
,

exp
4
2
0 0
4 2
c
b k
x .
The calculations of the integrals in the function
( )
z t , are not difficult.
Using the obtained formulas and the expressions, we may get the formulas
for the distributions of the electric field energy, the electric impedance,
electric conductivity and other characteristics of the arc column in
the thermal plasma flow. For example, for the electric resistance we
have
( )
( ) ( )
( )
R z t
E
I z
c
b k
c E z t
b
c
b k
,
exp
,
exp
_
,
_
,
1
]
1
1
1
1
1
0
2
0
2 2
0 0
1 0
2
0 0
4 2
0 0
1
2
2
1
1
2
4
(5.5)
It is seen from (5.5) and (5.4), that the impedance of the arc column
in case of modulated electric current contains the active component
and the reactive component depending on the frequency oscillation
of the current electricity. The existence of the reactive component
of the electric impedance in the nonstationary arc column to sug-
gests that, in the thermal plasma flow is distributed in the channel
with the internal inductance.
From the obtained expressions one can derive the formula for the
electric power:
3 8 0
( )
( )
( ) ( )
( )
g r z t
I i t
R z
c
b k
c E z t
b
c
b k
r
o
, ,
cos
exp
,
exp
, ; .
_
,
_
,
1
]
1
1
1
1
1

_
,
1
2
2
1
1
2
4
1
4
2
2
2 4 2
0 0
1
1 0
2
0 0
4 2
0 0
1
2
1
1
2 2
2
(5.6)
The mean value of the power during the period of oscillations of
the electric current is found by integrating (5.6) over time.
5.2. In this section, we explore the influence of the parameters
of the electric current modulation, physical properties and expenditure
of the gas on the dynamical characteristics of the arc column and
the thermal plasma flow.
In the general case, alternating components of the arc current and
electric field intensity are out of phase by . In the audio frequency
range at the bass , the function I(t) retards in phase with respect
to E(t). Analysing the given formulas, one can see that the increase
of the gas expenditure and that of the amplitude of the alternating
component of the arc current have the phase angles intermediate between
those of I(t) and E (t) increases.
It is seen from (5.1)(5.6) that the increase of the frequency os-
cillation of the electric current is due to the thermal inertia of the
plasma flow intensity. In this case, the value of the phase angle
is reduced. In the extreme case, for example, at the condition ,
for the function R(z,t) we have
( )
( ) ( )
( ) R z t
E
I z
c
b k
c E
i
b k
c
b k
d
z
,
exp exp
_
,
+
+
_
,
_
,
1
]
1
1
1
1
1
0
2
0
2 2
0 0
1 0
2
2
0 0 0
4 2
0 0
0
1
2
2
1
1
4 1
2
4
(5.7)
The electric resistance of the arc column does not depend on time.
It is seen that the influence of the internal inductance and reactive
component of the electrical resistance on the properties of the arc
column disappears. In these cases, the electrons concentration and
3 8 1
the carrier mobility in the plasma flow do not change during the period
of oscillations of current electricity. This result is consistent with
the conclusion of many well-known theories on the alternating arc
current [79, 1419, and other].
From the experimental investigations it is well known that, with
the increase of the channel length and the reduction of the gas ex-
penditure in plasma torches, the extreme characteristics of the arc
column and the plasma flow are rapidly stabilized. These results are
also confirmed in many theoretical investigations of the stationary
processes in the plasma torches with the continuous current. For the
plasma torches with the modulation of the arc current, these pecu-
liarities of the arc column and the plasma flow scantily explored.
From the formulas of our theory it can be seen, that at condition
l
_
,

, the functions S,E,R,g and other characteristics of the
arc column in the plasma flow became independent on the coordi-
nate z.
In the case, when the perturbation of the amplitude modulation
of the arc current can be disregarded ( at i~ 0), the results and formulas
of the presented theory coincides with the results of the stationary
theories [21].
6. THE RESULTS OF THE CALCULATIONS
In this section the results of calculations of the nonstationary electric
and thermal characteristics of the thermal plasma flow in the area
of arc column in the plasma torch are presented.
Figure 1 demonstrates the functions
( )
1 1 1
, r depending on . It
is seen, that at = 0, function
1
transfers in the Bessels function
( ) J r
0 1
, . The increase of the axial acceleration of the plasma flow
(), of the profile the radial distribution function grows narrow. The
dependences of the values
n
on are presented in the Fig.2
Figure 3 demonstrates the longitudinal waves of the electric field
intensity, damping along the channel. The dynamics EI character-
istics are presented in the Fig.4. It is seen, that with the increase
of the frequency oscillation , the phase angle reduces.
Figure 5 presents the results of calculations of the dependence of
the mean value the local temperature of the plasma on the frequency
oscillation. At f = 1000 Hz we have minimum T.
These results are of practical interest and are associated with effects
of the thermal inertia of the thermal plasma flow; they need special
investigations.
3 8 2
7. CONCLUSION
Assuming the series of the reasonable simplifying assumptions the
nonstationary mathematical model for the thermal plasma flow in the
channel of the plasma torches is composed. Unlike the previous well-
known nonstationary theories, this boundary value problem takes
additionally into account axial acceleration of the plasma flow, radius
variability of the arc column and arbitrary law of the alternating
component of arc current in the plasma torch.
The function of the electric current is adopted in the form I(z,t).
The dependence of function I on the coordinate z may be interest-
ing for the plasma torches modelling with the electric current dis-
tributed along the channel. The adopted mathematical model consists
of gas dynamic and thermodynamic problems.
The solution of the gas dynamic problem was obtained by the Rieman
method. Analysing this solution one can see that at the mathemati-
cal modelling of the processes in the plasma torches it is necessary
to take into account the axial acceleration of the thermal plasma flow
and the compressibility of the gas.
The nonstationary thermodynamic problem on the thermal plasma
flow in the area of the arc column is described by a set of nonlin-
ear integro-differential equations. The general solution of this problem
was obtained by the modification method of the DAlamber. The general
solution was derived in the form of superposition longitudinal waves,
damped along the channel. The formulas obtained allow calculations
of the nonstationary characteristics of the arc column and the thermal
plasma flow. These formulas also allow study in detail the influence
of the modulation parameters, physical properties and expenditure
Fig.2 Dependence
n
on . 1
1
; 2
2
; 3
3
.
Fig.1 Dependence function
1
(
1
, r). 1 = 0; 2 = 10; 3 = 30.
( )
1 1 1
, r
3 8 3
Fig.3 Longitudinal waves of electric field intensity in the channel: E (z,
0
) =
E (z,
0
) E
e
;

0
= 2; G = 1 g/s; i 0, 30; 1 = 2,5 kHz; 2 = 5 kHz.
Fig. 4 Dynamic E I characteristics: z = 0,2; G = 0,4 g/s; i = 0.05; 1 = 50 Hz;
2 = 300 Hz; 3 = 1 kHz.
Fig.5 The dependence of the local temperature of plasma on the frequency of oscillations:
z = 0.4; G = 1 g/s; 1 i = 0.1; i = 0.3.
of the gas flow, geometrical forms the channel on the characteris-
tics of the arc column and the plasma flow.
From the presented theory it can be seen that the reactive com-
ponent of the impedance of the nonstationary arc column depends
on the amplitude of the oscillation of current electricity property and
E, V/cm
E, V/cm
3 8 4
expenditure of gas and the geometrical form of the channel. For the
first time it has been ascertained that the thermal plasma flow in the
plasma torch with the modulation of the arc current has the inter-
nal inductance distributed along the channel.
The laws of modulation of electricity current and the change of
the plasma flow characteristics are displaced in phase. The phase
angle depends on the property and expenditure of the gas and geo-
metrical form the plasma torch. With the increase of the frequency
of oscillation this phase angle decrease and in the case , the
function of thermal potential S and thermal properties of the plasma
flow do not depend on the time.
The presented theoretical investigation generalizes the series of
the known theories about the thermal plasma flows in the plasma torch
with the arc current.
ACKNOWLEDGMENTS
Author would like to thank Academician RAS, Professor M. F. Zhukov and
Professor Dr. O. P. Solonenko for attention and displaying keen interest
in this problem.
References
1 H. Maecker, Z. Fur Physik, 157, No.1, 1 (1959).
2 M.F. Zhukov,et al, Applied dynamic of the thermal plasma (in Rus-
sian), Nauka, Novosibirsk (1975), p.298.
3 E. Pfender, Pure and Appl. Chem., 60, No.5 91-606 (1988).
4 E. Pfender, Plasma Chem. Plasma process, 9, No.1,167-194 (1989).
5 M.F.Zhukov, et al, The generators of the plasma arc current with the
sections between of the electrodes in the channels (In Russian), Nauka,
Novosibirsk (1981), p.222.
6 O.I. Yasko, Arc current in the plasmotron (In Russian), Minsk, Nauka
Tekhnika (1977), p.156.
7 O. Mayer, Archiv fur Elektrotechnik, 37, No.12, 588-608 (1943).
8 A.M.Cassie, New Theory of Arc Rupture and Circuit Severity, CIGRE,
Paper, N.102 (1939), p.1-14.
9 R.L. Phillips, 18, No.1, 65-78 (1967).
10 U.S. Svirchuk, In: Theory alternating arc current (In Russian): Theories
of the arc current in the conditions the heat and mass transfer, M.F.
Zhukov (ed), Nauka, Novosibirsk (1977), p.87-114.
11 O.Ya. Novikov, In: The generals methods analysis of the stability the
arc current: Theories of the arc current in the conditions the heat and
mass transfer (In Russian), M.F. Zhukov (ed), Nauka, Novosibirsk (1977),
p.115 -128.
12 G.A. Galeshian, et al, Plasma Physics, 18, No.8, 1084-1091 (1991).
13 M.A. Antinian, et al, Journal Technical Physics, 1992, v. 62, N 5, p.
164-168.
3 8 5
14 S.M. Kriganski and T.V. Krivoborskaia Journal of the Technical Physics
(In Russian) 51, 235-237 (1971).
15 G.U. Dautov, et al, Physics and chemistry in the treatment of ma-
terials (In Russian), No.3, 51-55 (1979).
16 F.A. Salyanov and M.M. Sharapov, Engineering Physical Journal (Minsk,
In Russian), XLII, No.4, 648-652 (1982).
17 F.A. Salyanov, Basis of physics of low temperature plasma, plasma
apparatus and technologies (In Russian), Nauka Phyzmatlit., Moscow
(1997), p.240
18 R.G. Phillips, Z. Fur. Physics, 211 (1968).
19 L.I. Sedov, Mechanics continuous medium (In Russian), Vol.1, Nauka
Physmatlit, Moscow (1970), p.568.
20 G. Beitmen and A. Erdane, Highers transcendentals functions, Vol.1,
Nauka, Phyzmatlit, Moscow (1970), p.294.
21 G.U.Dautov, In: Investigations arc column in the channel with the gas
flow, Informations Siberian Depatment RAS, Series Technical Sience,
vol.1, No.3 (1968).

Thermal Plasma Torches and Technologies - Vol.1 - Plasma Torches. Basic Studies and Design (O. P. Solonenko 2003)

Uploaded by

Document Information

Original Description:

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Thermal Plasma Torches and Technologies - Vol.1 - Plasma Torches. Basic Studies and Design (O. P. Solonenko 2003)

Uploaded by

Copyright:

Available Formats

THERMAL PLASMA

, inside the r.f. discharge are written as:

, the axial and

, and relative turbulence intensity, r

, (right) for an atmospheric

corresponds to the parameters at the so-called film temperature

in relation to the power of the plasma

g is the n-dimensional Radon transform, and R

induced by plasma currents. By

1000 V, one can obtain

are non-linear factors whose form depends on meaning

5 kK. By increasing the pressure

and the constant value of its field strength can

at which the required stationary solution appropriate to a

is considered to be given, and the parameters

= 65.7 cm/s): the solid lines are

. The gas remains cold up to the front of the dis-

is differently formed. The uniform plasma flow of extention L is caused

=0.418 m/s and u

=1.27 m/s, formula (4) gives the values of

The energy conservation equation

The electrical potential or the current continuity equation

is the total average heat removal

is the contact potential differ-

/W) and the moment of momentum flux are also

that correspond to the transition to the

>>1 is due to cooling of the

Taking into account the expressions obtained, we can write

You might also like