Integrated parametric study of a hybrid-stabilized argonwater arc under subsonic, transonic

and supersonic plasma flow regimes

This article has been downloaded from IOPscience. Please scroll down to see the full text article.
2011 J. Phys. D: Appl. Phys. 44 435204
(http://iopscience.iop.org/0022-3727/44/43/435204)
Download details:
IP Address: 147.229.93.199
The article was downloaded on 14/10/2011 at 13:41
Please note that terms and conditions apply.
View the table of contents for this issue, or go to the journal homepage for more
Home Search Collections Journals About Contact us My IOPscience
IOP PUBLISHING JOURNAL OF PHYSICS D: APPLIED PHYSICS
J. Phys. D: Appl. Phys. 44 (2011) 435204 (20pp) doi:10.1088/0022-3727/44/43/435204
Integrated parametric study of a
hybrid-stabilized argonwater arc under
subsonic, transonic and supersonic
plasma ow regimes
J Jeni sta
1
, H Takana
2
, H Nishiyama
2
, M Bartlov a
3
, V Aubrecht
3
,
P K renek
1
, M Hrabovsk y
1
, T Kavka
1
, V Sember
1
and A Ma sl ani
1
1
Institute of Plasma Physics, AS CR, v.v.i., Thermal Plasma Department, Za Slovankou 3, 182 00
Praha 8, Czech Republic
2
Institute of Fluid Science, Tohoku University, 2-1-1 Katahira, Aoba-ku, Sendai, Miyagi 980-8577, Japan
3
Faculty of Electrical Engineering and Communication, Brno University of Technology, Technick a 8,
616 00 Brno, Czech Republic
E-mail: jenista@ipp.cas.cz
Received 4 May 2011, in nal form 1 September 2011
Published 14 October 2011
Online at stacks.iop.org/JPhysD/44/435204
Abstract
This paper presents a numerical investigation of characteristics and processes in the worldwide
unique type of thermal plasma generator with combined stabilization of arc by argon ow and
water vortex, the so-called hybrid-stabilized arc. The arc has been used for spraying of
ceramic or metallic particles and for pyrolysis of biomass. The net emission coefcients as
well as the partial characteristics methods for radiation losses from the argonwater arc are
employed. Calculations for 300600 A with 22.540 standard litres per minute (slm) of argon
reveal transition from a transonic plasma ow for 400 A to a supersonic one for 600 A with a
maximum Mach number of 1.6 near the exit nozzle of the plasma torch. A comparison with
available experimental data near the exit nozzle shows very good agreement for the radial
temperature proles. Radial velocity proles calculated 2 mm downstream of the nozzle exit
show good agreement with the proles determined from the combination of calculation and
experiment (the so-called integrated approach). A recent evaluation of the Mach number from
the experimental data for 500 and 600 A conrmed the existence of the supersonic ow regime.
(Some gures in this article are in colour only in the electronic version)
1. Introduction
Plasma generators with arc discharge stabilization by water
vortex exhibit higher input powers and lower mass ow rates
(200 kW, 0.3 g s
1
) than gas-stabilized (Ar, He) arcs (50 kW,
3 g s
1
). This indicates that they have special performance
characteristics, such as high outlet plasma velocities (up
to 7000 ms
1
), temperatures (30 000 K), plasma enthalpy
and powder throughput, compared with commonly used
gas-stabilized torches [1, 2]. In a water-stabilized arc, the
stabilizing wall is formed by the inner surface of the water
vortex which is created by tangential water injection under
high pressure (410 atm) into the arc chamber. Evaporation
of water is induced by the absorption of a fraction of Joule
power dissipated within the conducting arc core. Further
heating and ionization of the steam are the principal processes
which produce a water plasma. The continuous inow and
heating lead to an overpressure and the plasma is accelerated
towards the nozzle exit. The arc properties are thus controlled
by the radial energy transport from the arc core to the walls
and by the processes inuencing evaporation of the liquid
wall. Over the last 15 years, water-stabilized arcs have been
intensively investigated, basic processes in the arc column have
been described and models of arcs with water stabilization
have been formulated. Complex experimental data about the
properties of the arc and parameters of the generated plasma
were presented in [26], and a two-dimensional numerical
model of Gerdien arc was formulated in [7, 8].
0022-3727/11/435204+20$33.00 1 2011 IOP Publishing Ltd Printed in the UK & the USA
J. Phys. D: Appl. Phys. 44 (2011) 435204 J Jeni sta et al
Figure 1. Principle of hybrid plasma torch WSP

H with combined gas (Ar) and vortex (water) stabilizations. Water is injected tangentially
and creates a vortex in the chamber. The arc burns between the cathode, made of a small piece of zirconium pressed into a copper rod, and
the water-cooled anode rotating disc. The calculation domain is shown by a dashed line.
Figure 2. The plasma spraying torch WSP

H with hybrid stabilization (left), i.e. the combined stabilization of arc by axial gas ow (Ar or
N
2
) and water vortex. The external rotating disc anode is made of copper. Images of plasma jets produced by WSP

H (right) from the

mixture of steam and argon for different operational conditions: 300 A and 24 slm of argon (top), spraying of Cu particles at 500 A and
36 slm of argon (middle), supersonic jet at 300 A, 12 slm of argon at 10 kPa of surrounding atmosphere (bottom).
For some applications such as plasma spraying, where
plasma jets are used to heat and accelerate the particles injected
into the jet, the high-enthalpy jets generated in water torches
ensure a high level of efciency of particle heating and thus
high throughputs, while low plasma ow rates and density
result in lower efciency of particle acceleration [9]. Thus, it
is desirable tocreate a plasma torchwiththermal characteristics
typical of water torches but with increased plasma ow rates
and densities. A combination of gas and vortex stabilization
has been used in the so-called hybrid-stabilized electric arc; its
principle is shown in gure 1. In the hybrid H
2
OAr plasma
arc the discharge chamber is divided into a short cathode part
where the arc is stabilized by tangential argon ow in the axial
direction, and a longer part which is water-vortex-stabilized.
This arrangement not only provides additional stabilization of
the cathode region and protection of the cathode tip, but also
offers the possibility of controlling plasma jet characteristics
in a wider range than that of pure gas or liquid-stabilized
torches [10, 11]. The arc is attached to the external water-
cooled rotating disc anode a few mm downstream of the torch
orice. The characteristics of the hybrid-stabilized electric arc
were measured and the effect of gas properties and owrate on
plasma properties and gas-dynamic ow characteristics of the
plasma jet were studied. Experiments [10, 12] demonstrated
that plasma mass ow rate, velocity and momentum ux
in the jet can be controlled by changing mass ow rate in
the gas-stabilized section, whereas thermal characteristics are
determined by processes in the water-stabilized section. The
domain for numerical calculation is shown in gure 1 by a
dashed line and includes the discharge area between the outlet
nozzle for argon and the near-outlet region of the hybrid plasma
torch.
The water-stabilized arc as well as the hybrid arc have
been used at IPP AS CR, v.v.i., in the plasma spraying torches
WSP

and WSP

H(160 kW) for spraying metallic or ceramic

powders (TiO
2
, Al
2
O
3
, ZrSiO
4
,W-based, Ni-based alloys, Al,
steel) injected into the plasma jet (gure 2). Some of the
latest studies of physical and chemical processes in materials
after their interaction with plasma can be found, e.g., in
[1317]. Recently, an experimental plasmachemical reactor
PLASGAS(gure 3) equippedwiththe sprayingtorchWSP

H
has been employed for the innovative and environmentally
2
J. Phys. D: Appl. Phys. 44 (2011) 435204 J Jeni sta et al
Figure 3. Schematic diagram (left) and a picture (right) of the experimental reactor for plasma pyrolysis and gasication.
friendly plasma treatment of waste streams with a view to
their sustainable energetic and chemical valorization and to
a reduction of the emission of greenhouse gases [18, 19].
Pyrolysis of biomass was experimentally studied in the reactor
using crushed wood and sunower seeds as model substances.
A high content of combustible mixture of hydrogen and
carbon monoxide was produced (syngas). Production of
higher hydrocarbons is suppressed due to higher temperatures
and high level of ultraviolet radiation in the reactor volume.
Experiments with the PLASGAS reactor have currently been
focused on the produced gas composition in dependence on
the reactor atmosphere and jet characteristics for different
conditions including inow cold reaction gases such as, e.g.,
oxygen, carbon dioxide and air. The most recent application
of the waste plasma treatment is the gasication of rubber oil
produced by the industrial waste tyre pyrolysis.
This work deals with a numerical study of the properties
and processes in a hybrid arc for high currents (300600 A)
and argon mass ow rates (22.540 standard litres per minute,
slm). In contrast to our previous investigation [20, 21], special
attention is devoted to the ow structure and temperature eld
in the discharge when the local Mach number is higher than
one. Our former results indicated the possibility [20] and
also showed the existence [22, 23] of supersonic ow regime
for currents higher than or equal to 400 A. In addition, a
detailed comparison of the calculated results with the latest
experiments is presented in this study; rst of all, a comparison
of temperature and velocity proles near the nozzle exit.
Section 2 gives information about the model assumptions,
plasma properties, boundary conditions and the numerical
scheme. Section 3 reveals the most important ndings such as
thermal and uid dynamic characteristics of the plasma within
the discharge and in the near-outlet regions, along with power
losses fromthe arc and a comparison of calculated results with
experiments.
2. Physical model and numerical implementation
2.1. Assumptions and the set of equations
The following assumptions are applied for the model:
(1) the numerical model is two-dimensional with the
discharge axis as the axis of symmetry,
(2) plasma ow is laminar/turbulent and compressible in the
state of local thermodynamic equilibrium,
(3) argon and water create a uniform mixture in the arc
chamber,
(4) only self-generated magnetic eld by the arc itself is
considered,
(5) gravity effects are negligible,
(6) the partial characteristic and the net emission coefcient
methods for radiation losses from the arc are employed,
(7) all the transport, thermodynamic and radiation properties
are dependent on temperature, pressure and argon molar
content.
A few comments should be mentioned on these
assumptions:
(1) The cylindrical discharge chamber (gure 1) is divided
into several sections by the bafes with central holes. Water
is injected tangentially into the chamber by three sets of
three inlet holes (totally 9 holes) placed equidistantly along
the circumference at angles of 120

. The inner diameter

of the water vortex is determined by the diameter of the
holes in the bafes. Water is usually pumped under
pressures of 0.39 MPa (0.6 MPa) with owrates of 10 Lmin
1
(16 Lmin
1
). Higher pressures ensure better hydrodynamic
stability of the arc. Since water ows in a closed circuit, it is
also exhausted at two positions along the arc chamber.
In order to see the ow structure near the outlet, in our
calculation domain we also included the near-outlet region
which extends up to 20 mm from the exit nozzle. In the
experiment, the distance from the exit nozzle to the anode
can be changed from 5 to 20 mm. It can be expected that
regions close to the exit nozzle will remain undisturbed by the
presence of the anode, while the more distant regions (15
20 mm) will be inuenced by 3D effects (the anode jet and
anode processes), provided the anode is placed somewhere
20 mm from the nozzle exit.
It comes out from these considerations that the two-
dimensional assumption is valid in major part of the domain
due to(a) cylindrical symmetryof the discharge chamber setup,
(b) tangential injection of water through the holes along the
circumference and (c) the exible distance between the nozzle
exit and anode.
3
J. Phys. D: Appl. Phys. 44 (2011) 435204 J Jeni sta et al
(2) The assumption of laminar ow is based on experiments,
showing the laminar structure of the plasma owing out of the
discharge chamber in the space between the nozzle exit and
the anode. The laminar ow has been observed for currents up
to 600 A. It comes out from our previous calculation [7] that
Reynolds number based on the outlet diameter 6 mm reaches
13 000 at maximum in the axial region and decreases to 300
in arc fringes. The type of ow inside the discharge chamber
is questionable since no diagnostics is able to see inside the
chamber and it is not clear whether the laminar plasma stream
is a result of laminarization of the plasma ow at the outlet.
To check possible deviations from the laminar model, we have
employed large eddy simulation (LES) with the Smagorinsky
sub-grid scale model. It was found that simulations for laminar
and turbulent regimes give nearly the same results, so that the
plasma ow can be considered more or less laminar for the
operating conditions and the simplied discharge geometry in
this study. The maximum detected discrepancy between the
turbulent and laminar models is 7%for the relative temperature
difference at the arc axis 2 mm downstream of the nozzle exit
for 500 Aand 40 slmof argon. For reasons of generality, all the
results presented here were calculated using the LES turbulent
model.
(3) The assumption of complete (uniform) mixing is a
simplication of reality since, based on experiments, argon and
water species do not mix homogeneously in the hybrid torch,
especially at lower currents. This assumption was discussed in
more detail in [20] and it was concluded that this assumption
can underestimate the temperature and velocity in the axial
discharge region to some extent.
The complete set of conservation equations representing
the mass, electric charge, momentum and energy transport of
such plasma can be written in the cylindrical coordinates as
follows:
Continuity equation:

t
+
1
r

r
(vr) +

z
(u) = 0. (1)
Momentum equations:

t
(u) +

r
(uv) +

z
(uu) =
p
z
+ j
r
B

2
3

z
_

_
1
r

r
(rv) +
u
z
__
+

z
_
2
u
z
_
+
1
r

r
_
r
_
u
r
+
v
z
__
, (2)

t
(v) +

r
(vv) +

z
(uv) =
p
r
j
x
B

2
3

r
_

_
1
r

r
(rv) +
u
z
__
+
w
2
r
+
1
r

r
_
2r
v
r
_

2v
r
2
+

z
_

_
u
r
+
v
z
__
. (3)
Energy equation:
e
t
+
1
r

r
_
r
_
(e + p)v
T
r
__
+

z
_
(e + p)u
T
z
_
=
1
r

r
[r(
rr
v +
rz
u)] +

z
(
rz
v +
zz
u) + j
r
E
r
+j
z
E
z
R. (4)
Charge continuity equation:
1
r

r
_
r

r
_
+

z
_

z
_
= 0. (5)
Equation of state:
p = R
g
T. (6)
Here z and r are the axial and radial coordinates, u, v and w
are the axial, radial and tangential components of the velocity,
respectively, is the mass density, is the viscosity (in the case
of LES model, the turbulent contribution
turb
is also added)
p is the pressure, B

is the magnetic eld strength, e is the

density of energy (internal and kinetic) produced or dissipated
in the unit volume, T is the temperature, is the viscous stress
tensor, j
z
and j
r
are the axial and radial components of the
current density, E
z
and E
r
are the axial and radial components
of the electric eld strength, is the electric potential, R is
the source term for the radiation losses and R
g
is the molar gas
constant. The magnetic eld strength B

is calculated from
the BiotSavart law, the current density

j from the Ohms law

j =

E.
2.2. Properties of argonwater plasma mixture
The waterargon mixture can be described by the formula
(H
2
O)
(1q)
Ar
q
where the argon molar amounts q were chosen
from0 to 1 with a step of 0.1. In total, 35 chemical species were
considered [24]. For the temperature range 40020 000 K we
supposed the following decomposition products: e (electrons),
H, O, Ar, O
+
, O
2+
, O
3+
, O

, O
2
, O
+
2
, O

2
, O
3
, H
+
, H

, H
2
, H
+
2
,
H
+
3
, OH, OH
+
, OH

, HO
2
, HO

2
, H
2
O, H
2
O
+
, H
3
O
+
, H
2
O
2
,
Ar
+
, Ar
2+
and Ar
3+
. For the temperature range 2050 000 K
the set of products is somewhat different, including also
multiply charged ions: e (electrons), H, O, Ar, O
+
, O
2+
, O
3+
,
O
4+
, O
5+
, O
6+
, H
+
, Ar
+
, Ar
2+
, Ar
3+
, Ar
4+
, Ar
5+
and Ar
6+
.
The calculations were performed using the modied Newton
method for the solution of nonlinear equation system which
is composed of equations of Saha and mass action law type
expressing individual complex components by the help of basic
ones (e,H,O,Ar). The systemis completed by the usual particle
and charge balance assuming quasineutrality and equilibrium.
The thermodynamic properties and the transport coef-
cients of this gas mixture were calculated according to
the ChapmanEnskog method in the fourth approximation
described, e.g., in [25] for temperatures 40050 000 K [24]
and pressures 0.10.3 MPa in the local thermodynamic
equilibrium.
2.3. Radiative transfer methods
Two radiation methods are implemented in the energy equation
for energy losses from the argonwater plasma: (1) the net
emission coefcients for the required arc radius of 3.3 mm,
and (2) the partial characteristics method, both of them for
different molar fractions of argon and water plasma species in
dependence on temperature and pressure.
The principle of the method of partial characteristics is that
integrals over radiation frequencies in equations for radiation
quantities can be performed in advance to form functions
4
J. Phys. D: Appl. Phys. 44 (2011) 435204 J Jeni sta et al
Som and Sim, called partial characteristics [26, 27]. The
partial characteristics include all time demanding integrations
over frequencies. These functions are stored in data tables
according to several parameters, such as plasma pressure,
temperature and geometrical dimensions. The tables are then
used for 3D computations with spatial integrations only.
The partial characteristics Som and Sim are dened as
follows [28]:
Som(T
X
, T
R
, |R X|) =
_

0
B

(X)

(X)e

_
R
X

() d
d,
(7)
Sim(T
X
, T

, X) =
_

0
[B

(X) B

()]

(X)

()
exp
_

_
X

() d d
_
(8)
where B

is the Planck function,

is the absorption
coefcient, T is the plasma temperature, and , X and R are
the radial coordinates. This method has been successfully used
in some previous papers [29, 30] for prediction of radiation
intensities, radiation uxes and the divergence of radiation
uxes for various temperature proles of SF
6
arc plasmas.
Note that if isothermal temperature prole is assumed, the
function Somis equal to the isothermal net emission coefcient
introduced by Lowke [31] as

N
=
_

0
B

(T )

(T, p) exp[

(T, p)R] d (9)

where R represents the radius of the plasma column. The
isothermal net emission coefcient,
N
, corresponds to the
fraction of the total power per unit volume and unit solid angle
irradiated into a volume surrounding the axis of the arc and
escaping from the arc column after crossing a thickness R
of the isothermal plasma. Application of the net emission
coefcients to a prediction of energy balance gives good results
for central arc temperatures, but it cannot predict accurate
temperature proles at low temperatures near the edge of the
arc, because of the absorption of ultraviolet radiation emitted
at the centre of the arc at high temperatures (self-absorption
using this method can only be accounted for in a highly
approximate way by using negative emission coefcients at
low temperatures).
In the radiation model, continuous radiation due
to photorecombination and bremsstrahlung processes is
included as well as discrete radiation consisting of thousands
of spectral lines. Broadening mechanisms of atomic and ionic
spectral lines due to Doppler, resonance and Stark effects are
considered. The numbers of oxygen and argon lines included
inthe calculationare O(93lines), O
+
(296), O
+
2
(190), Ar (739),
Ar
+
(2781), Ar
2+
(403), Ar
3+
(73). In addition, molecular
bands of O
2
(SchumanRunge system), H
2
(Lyman and Verner
systems), OH (transition A
2

+
X
2

i
) and H
2
O (several
transitions) are also implemented [32]. Absorption coefcient
as a function of wavelength is calculated from infrared to far
ultraviolet regions and the tables of partial characteristics for
100035 000 K. The net emission coefcient method used here
is a special case of the partial characteristics method with a zero
partial sink (Sim = 0).
The net emission coefcient method takes into account
the reabsorption occurring in the hot emitting regions. The
partial characteristics method, in addition, takes into account
the reabsorption of radiation in the low-temperature regions of
the arc. In the following, when we mention the reabsorption
of radiation or the impact of the reabsorption of radiation, it
refers to the reabsorption of radiation in the low-temperature
region in the partial characteristics method.
One comment about the choice of the arc radius for the
net emission coefcient method: the arc radius of 3.3 mm
corresponds to the radius of the calculation domain and is used
as the radius of the absorbing region. This radius was assumed
in our calculation from the point of simplicity of numerical
implementation. Our calculation showed that there exists a
low-temperature region close to the water vapour boundary
and that the arc radii having temperatures higher than 10 kKare
2.72.8 mm for 300 A and 2.93.0 mm for 600 A. Taking
into account the effect of the divergence of radiation ux on
the temperature proles, we found that an increase in the arc
radius for a few tenth of mm will cause only a low decrease
in the divergence of radiation ux. Thus, deviation from the
correct arc radius of the order of tenth of mm will cause a
practically negligible increase in the temperature in the hot arc
regions.
2.4. Boundary conditions and numerical scheme
The calculation region and the corresponding boundary
conditions are presented in gure 4. The dimensions are
3.3 mm for the radius of the discharge region, 20 mm for the
radius of the outlet region and 78.32 mm for the total length.
These dimensions agree with the hybrid torch experimental
setup.
(a) Inlet boundary (AB) is represented by the nozzle exit for
argon. Along this boundary we assume the zero radial velocity
component, v = 0. Because of the lack of experimental
data, the temperature prole T (r, z = 0) and the electric
eld strength E
z
= /z = const. for a given current
are calculated at this boundary, before the start of the uid-
dynamic calculation itself, iteratively from the Elenbaas
Heller equation including the radiation losses fromthe arc (our
previous numerical experiments showed a weak dependence
of the form of the boundary temperature prole on the overall
solution). The inlet velocity prole u(r) for argon plasma for
the obtained temperature prole T (r, z = 0) is pre-calculated
from the axial momentum equation under the assumption of
fully developed ow.
(b) Axis of symmetry (BC). The zero radial velocity and
symmetry conditions for the temperature, axial velocity and
electric potential are specied here, i.e. T /r = u/r =
/r = 0, v = 0.
(c) Arc gas outlet plane (CD). The zero electric potential
= 0 (the reference value) and zero axial derivatives of the
temperature and radial velocity are dened at CD, T /z =
v/z = 0. Values of the axial velocity are interpolated from
the inner grid points.
(d) Arc gas outlet plane (DE). The zero radial velocity and zero
radial derivatives of the temperature, axial velocity and electric
5
J. Phys. D: Appl. Phys. 44 (2011) 435204 J Jeni sta et al
Figure 4. Discharge area geometry. Inlet boundary (AB) is represented by the nozzle exit for argon. The length of the discharge region is
58.32 mm, the length of the outlet is 20 mm. Along the line FE we specify the outlet nozzle and the wall of the hybrid plasma torch
equipment.
potential are dened here, T /r = u/r = /r = 0,
v = 0. Pressure is xed at 1 atmosphere, p = 1 atm.
(e) Outlet wall and the nozzle (EF). We specify no slip
conditions for velocities, u = v = 0, constant values of E
r
and E
z
(/z = /r = 0) and T (r, z) = 773 K (500

C)
for the temperature of the nozzle.
(f) Water vapour boundary (FA). Along this line we specify
the so-called effective water vapour boundary, named in
gure 4 as the water vapour boundary with a prescribed
temperature of water vapour T (R = 3.3 mm, z) = 773 K.
This is a numerical simplication of a more complex physical
reality assumed near the phase transition water vapour in the
discharge chamber. The shape of the phase transition between
water and vapour in the discharge chamber is experimentally
unknown and it is unclear so far whether the structure of
the transition is simple or very complicated, for example,
with a time-dependent form. Various irregularities in the
transition such as splitting of the phase transition or water
drops in the vapour phase can increase the complexity of the
transition. In [33] the iteration procedure for determination
of the mass ow rate and the radius of the water vapour
boundary for each current was proposed, based on comparison
with available experimental temperature and velocity at the
outlet and the electric potential drop in the chamber. It was
concluded that the best t between experiment and numerical
simulation for all currents exists for a mean arc radius of
3.3 mm. The corresponding values of water mass owrates are
0.228 g s
1
(300 A), 0.286 g s
1
(350 A), 0.315 g s
1
(400 A),
0.329 g s
1
(500 A) and 0.363 g s
1
(600 A). The magnitude of
the radial inow velocity is calculated from the denition of
mass ow rate
v(R) =
m
2R

z
(R, z)z
,
where (R, z) is a function of pressure and thus dependent
on the axial position z, z is the distance between the
neighbouring grid points. Because of practically zero current
density in the cold vapour region (no current goes outside of
the lateral domain edges), the radial component of the electric
eld strength is set to zero, i.e. E
r
= 0. The axial velocity
component is set to zero, u = 0.
Since we do not solve here the equation for tangential
velocity component w, the distribution of win the discharge for
the presented results was taken from our previous calculations
[7, 34] solved by the SIMPLERiteration procedure [35] which
enables calculation of w for the axisymmetric case, i.e. w
as a function of z and r coordinates. The velocity w acts
here only through the centrifugal force w
2
/r in the radial
momentum equation (3). The inuence of the tangential
velocity component w on the overall arc performance has
been studied in more detail in our previous papers [7, 34].
In section 2.1 it was mentioned that water is pumped into
the discharge chamber under pressures of 0.39 or 0.6 MPa
with ow rates of 10 or 16 Lmin
1
. The tangential velocity
component at the water vapour boundary corresponding to
these conditions is 78 ms
1
. It was found that the effect of
tangential velocity on the arc performance is small under such
conditionsthe maximum difference in the outlet parameters
is less than 2% regarding the case omitting the tangential
velocity component. The reason is that the centrifugal forces
w
2
/r are small compared with other dominant terms in
the radial velocity equation. Therefore, we adopted the
distribution of tangential velocity obtained from our previous
calculation for the numerical simulation.
For time integration of (1)(4), LU-SGS (LowerUpper
Symmetric Gauss-Seidel) algorithm [36, 37], belonging to
the group of the density-based methods [38], coupled with
the Newtonian iteration method is used for the integration
of discretized equations in time and space. To resolve the
compressible phenomena accurately, the Roe ux differential
method [39] coupled with the third-order MUSCL-type
(Monotone Upstream-centred Schemes for Conservation
Laws) TVD (Total Variation Diminishing) scheme [40, 41] is
used for the convective term. The electric potential from (5)
is solved in a separate subroutine by the TDMA (Tri-Diagonal
Matrix Algorithm) line-by-line method. From (1)(4) we
obtain, u, e and. Pressure is determinedfromthe pressure
dependence of the internal energy U(p, T ) = e(p, T )
0.5| u|
2
and temperature is calculated from the equation of
state (6) p/ = R
g
(p, T )T , using the pre-calculated values of
the product R
g
(p, T ) T as a function of temperature, pressure
and argon molar fraction in the mixture [24].
6
J. Phys. D: Appl. Phys. 44 (2011) 435204 J Jeni sta et al
The main reason why we consider the time transient
terms in our computation (equations (1)(4)), even though
we are interested in the steady states, is to capture the shock
wave structure that may occur in the ow. LU-SGS is one
of the time integration methods and only used when the
time transient terms are considered. Also, the use of time-
dependent formulation has a great advantage over the steady-
state formulation that the nal steady states are unambiguous.
If the time transient terms are not considered, the steady
simulation cannot predict the appearance of the shock wave
exactly.
The computer program is written in the FORTRAN
language. The task has been solved on an oblique structured
grid with nonequidistant spacing. The total number of grid
points was 38 553, with 543 and 71 points in the axial and
radial directions, respectively.
3. Results of calculation
3.1. Thermal, uid ow and electrical characteristics of the
plasma
Calculations were carried out for currents 300, 400, 500
and 600 A. Mass ow rate for the water-stabilized section of
the discharge was taken for each current between 300 and
600 A from our previously published work [33, 42], where it
was determined iteratively as a minimum difference between
numerical and experimental outlet quantities. The resulting
values are 0.228 g s
1
(300 A), 0.315 g s
1
(400 A), 0.329 g s
1
(500 A) and 0.363 g s
1
(600 A). Argon mass ow rate was
varied in agreement with experiments in the interval from 22.5
to 40 slm, namely 22.5, 27.5, 32.5 and 40 slm. It was found
in experiments [43] that part of argon is taken away before it
reaches the torchexit because argonis mixedwithvapour steam
and removed to the water system of the torch. The amount of
argon transferred in such a way from the discharge is around
50%for the currents studied. Since the present model does not
treat argon and water as separate gases and the mechanism of
argon removal is not included in the model, we consider in the
calculations that argon mass ow rate present in the discharge
equals one-half of argon mass ow rate at the torch inlet (i.e.
for 22.5 slm we take 50% of this amount, etc). Relatively high
values of argon mass ow rate, used also in experiment, were
chosen here to demonstrate compressible phenomena.
It can be argued that the addition of argon into the steam
plasma changes the energy balance within the arc to some
extent and can inuence evaporation of water. The former
measurements carried out at the plasma jet close to the exit
nozzle showed that argon and water species are not mixed
homogeneously within the arc columnwater species are
concentrated mainly at the outer arc regions while argon
species prevail in the axial region. The degree of non-
homogeneous mixing depends on the arc current and argon
ow rate. Our former experiments and calculations conrmed
that the arc temperature decreases only slightly with increasing
argon mass ow rate, i.e. addition of argon does not inuence
the energy balance of the hybrid-stabilized arc signicantly.
Since the numerical model does not include an evaporation
model and we are not able to measure evaporation of steam
directly, it can be expected from these considerations that
production of steam will be inuenced mainly by the presence
of water species close to the phase transition, so that the
estimate of steam mass ow rates made for the water-vortex-
stabilized arc would not be far from reality and would be
applicable also for the arc with hybrid stabilization. The
motivation for why we took mass ow rates for water-
stabilized section from our previous papers is as follows: (1)
to be consistent with our previous calculation and to compare
the impact of more precise plasma properties on the arc
performance (gures 1519) and (2) experiments for a hybrid-
stabilized arc have not been carried out for all the currents and
argon mass ow rates presented in this study.
Figure 5 shows the velocity, temperature, pressure and
the Mach number in the outlet nozzle and near-outlet regions
for 600 A, water mass ow rate of 0.363 g s
1
and 40 slm
of argon. The partial characteristics method for radiation
losses is employed. The results shown here demonstrate
the largest variations of velocity, temperature, pressure and
the Mach number just after the jet exhausts from the torch
nozzle among all the studied currents and argon mass ow
rates. Supersonic ow structure in the near-outlet region is
obvious with clearly distinguished shock diamonds with the
maximum Mach number about 1.6 with 10 500 ms
1
. The
corresponding velocity and the Mach number maxima overlap
with the temperature and pressure minima and vice versa.
Since the pressure decreases at the torch exit to a nearly
atmospheric pressure, the computed contours correspond to
an under-expanded atmospheric-pressure plasma jet.
The corresponding axial proles of the Mach number,
pressure, temperature and velocity along the arc axis
downstream from the nozzle orice (at the axial position
58.32 mm) for the same run are presented in gure 6. Several
successive wave crests and troughs along the axis for each of
the physical parameters is a typical feature of supersonic uid
ow. The variation of presented quantities is between 1.1 and
1.7 for the Mach number, 0.7 and 1.4 atm for pressure, 7200
and 10 000 ms
1
for velocity, and 18 000 and 23 500 K for
temperature.
Figure 7 displays the temperature and velocity elds in the
discharge and the near-outlet regions for (a) 500 and (b) 600 A
with water mass ow rates of 0.329 g s
1
(500 A), 0.363 g s
1
(600 A) [33] and argon mass ow rate of 0.554 g s
1
(one-half
of 40 slm). The net emission coefcient radiation method is
employed. The orientation of the calculation domain is the
same as in gures 1 and 4. Since the ratio of the axial to the
radial dimensions of the calculation domain is 24 the scaling
of the radial and axial coordinates is not proportional to make
the contours inside the discharge region clearly visible. Argon
ows axially into the domain, whereas water evaporates in the
radial direction from the water vapour boundary. Both the
results for 500 and 600 Aexhibit a supersonic under-expanded
plasma ow regime but a progression from weak to highly
pronounced shock diamond structure at 600 A is obvious. The
maximum velocities are 7200 ms
1
(500 A) and 9400 ms
1
(600 A) near the axial position of 60 mm. Further downstream
the velocity amplitudes decrease due to viscosity dissipation
7
J. Phys. D: Appl. Phys. 44 (2011) 435204 J Jeni sta et al
Figure 5. Velocity, temperature, pressure and the Mach number contours in the outlet nozzle and near-discharge regions for the 600 A arc
discharge. Water mass ow rate is 0.363 g s
1
, argon mass ow rate is 40 slm (standard litres per minute). The partial characteristics
radiation method is employed. Supersonic ow structure is obvious with clearly distinguished shock diamonds. The maximum Mach
number reaches 1.6. Contour increments are 500 ms
1
for velocity, 1000 K for temperature, 0.1 atm for pressure and 0.1 for the Mach
number.
(a)
(b)
Figure 6. Proles of the Mach number, pressure, temperature and velocity along the arc axis starting at the nozzle exit for (a) 500 A and (b)
600 A. Argon mass ow rate is 40 slm, the partial characteristics radiation method. Supersonic outlet is obvious with distinguished shock
diamonds, more pronounced for 600 A.
8
J. Phys. D: Appl. Phys. 44 (2011) 435204 J Jeni sta et al
Figure 7. Temperature and velocity contours for (a) 500 A and (b) 600 A arcs, the net emission coefcient method. Water mass ow rates
are 0.329 g s
1
(500 A) and 0.363 g s
1
(600 A); argon mass ow rate is 40 slm for both currents. Progression of a supersonic ow structure
at the outlet is clearly visible. Contour increments are 1000 K for temperature and 500 ms
1
for velocity.
and due to the reduction of the difference between the jet static
pressure and back pressure.
The impact of reabsorption of radiation on the distribution
of temperature and velocity within the discharge and the near-
outlet regions for 600 A is obvious in gure 8. The partial
characteristics method gives lower temperatures and higher
velocities in the outlet region. Similar results are found for all
currents and argon mass ow rates.
Figure 9 presents the radial proles of the Mach number
2 mm downstream of the nozzle exit with an argon mass ow
rate of 40 slm. It is clearly demonstrated that for currents
higher than 400 A, a supersonic rare plasma in the central parts
of the discharge is surrounded by a subsonic, much denser but
still hot plasma. Due to generally higher velocities and lower
temperatures near the outlet the partial characteristics method
gives higher values for the Mach number; the difference
regarding the net emission coefcient method is below 0.1 at
the arc axis.
Different ow structures for currents between 300 and
600 A and 32.5 slm of argon are visible in gure 10. Subsonic
plasma ow at 300 A (Mach 0.8) is converted to transonic
owat 400 A(Mach 1) at the outlet. The onset of supersonic
ow structure formation is visible at 500 A (Mach 1.15) and
the fully developed supersonic ow with shock diamonds is
formed for 600 A (Mach 1.4).
The structure of the plasma owing out of the discharge
chamber (the structure of plasma jet) was photographed by the
optical cameras Sensicam Fast Shutter and Canon 450D with
grey lter operating at the visible spectrum. The transition
from subsonic to supersonic regime was not observed visually
for the present conditions. The transition starts in the centre
of the jet and diamond structures are probably hardly visible
due to strong radiation of the hot surrounding plasma. One
can see in gures 5 and 6 that the maximum difference in the
neighbouring temperature crests and troughs at the arc axis is
only about 1500 Kfor 600 Aand 40 slmof argon. Temperature
contours for the radial positions higher than 1 mmfromthe arc
axis are nearly independent of the arc radius (not inuenced
by the supersonic structure at the axis) and temperature is still
high enough (20 kK) to produce strong radiation.
Values of temperature, voltage drop, the Mach number and
overpressure as functions of arc current (300600 A) and argon
mass ow rate (22.540 slm) for the partial characteristics
method are displayed in gure 11. Temperature and the Mach
number shown here are taken at the arc axis 2 mmdownstream
of the outlet nozzle, the voltage drop and overpressure refer
9
J. Phys. D: Appl. Phys. 44 (2011) 435204 J Jeni sta et al
Figure 8. Temperature and velocity contours for (a) the net emission and (b) the partial characteristics methods, 600 A, 27.5 slm of argon,
water mass ow rate 0.363 g s
1
. The partial characteristics model gives lower temperatures and higher velocities in the outlet region.
Contour increments are 1000 K for temperature and 500 ms
1
for velocity.
to the length of the discharge region of the hybrid-stabilized
electric arc (58.32 mm). The maximum values of velocity
(9500 ms
1
), overpressure (1.6 atm) and the Mach number
(1.5) occur for 600 A and 40 slm of argon. It is evident from
the slope of contours that overpressure and the Mach number
increase with argon mass owrate while the velocity increases
only slightly. The transition from subsonic to supersonic ow
occurs for currents around 400 A. The electric potential drop
in the discharge chamber increases with current and decreases
with increasing argon mass ow rate, the maximum value
reaches 164 V. The rise of the potential drop with current is
typical for arcs with relatively high energy losses from the arc,
in our case due to radiation and conduction.
Temperature contours for the net emission and partial
characteristics methods are shown in gure 12. Temperature
depicted is taken again at the arc axis 2 mm downstream of
the outlet nozzle. Reabsorption of radiation increases the
temperature in arc fringes and decreases it near the arc axis,
and as a result the net emission method provides higher axial
temperatures. The relatively small increase in velocity with
argon mass ow rate has thus an apparent explanation: on
the one hand, the increase in argon mass ow rate implies
Figure 9. Radial proles of the Mach number 2 mm downstream of
the nozzle exit, argon mass ow rate is 40 slm. The Mach numbers
in the axial region are higher than 1 for currents equal to or higher
than 400 A. Somewhat higher values of the Mach number are
provided by the partial characteristics method.
10
J. Phys. D: Appl. Phys. 44 (2011) 435204 J Jeni sta et al
Figure 10. Velocity contours in the outlet nozzle and near-discharge regions for 32.5 slm of argon. The partial characteristics radiation
method is employed. Water mass ow rates are 0.228 g s
1
(300 A), 0.315 g s
1
(400 A), 0.329 g s
1
(500 A), 0.363 g s
1
(600 A), contour
increments are 500 ms
1
. Supersonic ow structure is obvious for 600 A.
the corresponding increase in velocity. On the other hand, the
temperature decreases with argon mass owrate, thus lowering
the increase in plasma velocity.
3.2. Comparison of radial temperature and velocity proles
with experiments
A number of experiments have been carried out on the hybrid-
stabilized electric arc in the recent past for currents 300
600 A with 22.540 slm of argon. Temperature is one of the
fundamental plasma parameters, needed also for the evaluation
of other quantities.
In the experiment, the radial proles of temperature at the
nozzle exit were calculated fromoptical emission spectroscopy
measurements. The procedure is based on the ratio of emission
coefcients of hydrogen H

line and four argon ionic lines

using calculated LTE composition of the plasma for various
argon mole fractions as a function of temperature [24]. From
the calculated molar fractions of hydrogen and argon it is
easy to obtain the emission coefcients of H

and argon
lines. The temperature corresponding to an experimental
ratio of emission coefcients is then found by cubic spline
interpolation on the theoretical data.
Figures 13 and 14 compare the measured and calculated
radial temperature proles 2 mm downstream of the nozzle
exit for 300600 A and 22.5 and 40 slm of argon. Excellent
agreement is demonstrated for 300 and 600 A where the
measured proles nearly coincide with the two proles
calculated using the net emission and partial characteristics
radiation methods. For 400 and 500 A agreement is better for
the proles calculated by the net emission coefcients (black
colour).
A relative difference between the calculated and
experimental values of temperature at the arc axis 2 mm
downstream of the nozzle exit is evaluated for a broad range
of currents and argon mass ow rates. The relative difference
is dened here as
= 100 abs(T
num
T
exp
)/T
num
,
where T
num
(T
exp
) are the values of the calculated
(experimental) temperature. It was demonstrated that the
maximum relative difference between the calculated and
experimental temperature proles is lower than 10% for the
partial characteristics and 5% for the net emission radiation
method used in the present calculation, i.e. the net emission
radiation method gives better agreement with experiment as
regards axial temperatures.
Acomparison of the measured and calculated temperature
proles with our former calculations [23] is shown in gure 15
for 500 and 600 A. The set of proles is calculated/measured
again 2 mm downstream of the nozzle exit. The term new
model introduced here refers to the present model with the
assumptions described in sections 2.1 and 2.2, while the old
model means the former one with the following assumptions:
(a) the transport and thermodynamic properties of the argon
water plasma mixture are calculated using linear mixing
rules for non-reacting gases based either on mole or mass
fractions of argon and water species [23],
(b) all the transport and thermodynamic properties as well
as the radiation losses are dependent on temperature, and
argon molar content but NOT dependent on pressure,
(c) radiation transitions of H
2
O molecule are omitted.
In our present model, (1) all the transport and
thermodynamic properties are calculated according to the
ChapmanEnskog method in the fourth approximation, (2)
all the properties depend on pressure, (3) radiation transitions
of H
2
O molecule are considered. It is obvious that
improvements in the properties caused a better convergence
11
J. Phys. D: Appl. Phys. 44 (2011) 435204 J Jeni sta et al
Figure 11. Velocity (kms
1
), overpressure (atm), the Mach number and voltage drop (V) as functions of arc current (300600 A) and argon
mass ow rate (22.540 slm). Temperature and the Mach number are taken at the arc axis 2 mm downstream of the outlet nozzle, voltage
drop and overpressure refer to the length of the discharge region of the hybrid-stabilized electric arc. The partial characteristics radiation
method is employed.
Figure 12. Temperature contours (kK) in dependence of arc current (300600 A) and argon mass ow rate (22.540 slm) for the net
emission and partial characteristics methods. Temperature depicted is taken at the arc axis 2 mm downstream of the outlet nozzle.
Reabsorption of radiation slightly decreases axial temperatures.
12
J. Phys. D: Appl. Phys. 44 (2011) 435204 J Jeni sta et al
Figure 13. Experimental and calculated radial temperature proles 2 mm downstream of the nozzle exit for 300600 A and 22.5 slm of
argon.
Figure 14. Experimental and calculated radial temperature proles 2 mm downstream of the nozzle exit for 500 and 600 A and 40 slm of
argon.
between the experiment and calculation; radial temperature
proles obtained by our new model give a better comparison
with experimentslower temperatures and sharper proles
compared with our former calculation. Similar results were
also obtained for the net emission method.
A more comprehensive view on the closeness of the
calculated and experimental temperature proles is given in
gure 16. The dots in the plot denote the so-called average
relative temperature difference dened as

T
av
=
100
N

i=1
abs(T
i
num
T
i
exp
)/T
i
num
,
which estimates a kind of average relative difference along the
temperature prole, N is the number of available coincident
13
J. Phys. D: Appl. Phys. 44 (2011) 435204 J Jeni sta et al
Figure 15. Experimental and calculated radial temperature proles 2 mm downstream of the nozzle exit for 500 and 600 A with 27.5 and
32.5 slm of argon, the partial characteristics method. The so-called new model stands for the present model, the old model presents our
previous one with simplied plasma properties (see the text).
Figure 16. Average relative difference (see the text) between the calculated and experimental radial temperature proles, shown in %, at the
axial position 2 mm downstream of the nozzle exit, partial characteristics. The so-called new model stands for the present model,
exhibiting better agreement with experiments; the old model presents our previous one with simplied plasma properties (see the text).
14
J. Phys. D: Appl. Phys. 44 (2011) 435204 J Jeni sta et al
Figure 17. Velocity proles 2 mm downstream of the nozzle exit for 300600 A with 22.5 slm of argon. Calculationthe partial
characteristics method; re-calculated experimental prole is based on the experimental temperature prole and calculated Mach number (see
the text). The so-called new model stands for the present model, the old model presents our previous model with simplied plasma
properties (see the text). The re-calculated velocity proles show better agreement with the experiment.
numerical T
i
num
and experimental T
i
exp
values of temperature
along the radius. Our present new model apparently gives a
better comparison than the old model in all cases.
In addition to temperature proles, velocity proles at
the nozzle exit and mass and momentum uxes through the
torch nozzle are important indicators for characterization of
the plasma torch performance. In the experiment, the velocity
at the nozzle exit is determined fromthe measured temperature
prole and power balance assuming local thermodynamic
equilibrium [44]. First, the Mach number M is obtained from
the simplied energy equation integrated through the discharge
volume [7], second, the velocity prole is derived from the
measured temperature prole using the denition of the Mach
number
u(r) = M c{T (r)},
where c{T (r)} is the sonic velocity for the experimental
temperature prole estimated from the T&TWinner code [45].
The drawback of this method is the assumption of the constant
Mach number over the nozzle radius. Nevertheless the
existence of supersonic regime (i.e. the meanvalue of the Mach
number over the nozzle exit higher than 1) using this method
was found for 500 A and 40 slm of argon, as well as for 600 A
andfor argonmass owrates higher than27.5 slm[44]. Similar
results have also been reported in our previous work [22].
For more exact evaluation of velocity proles we
employed the so-called integrated approach, i.e. exploitation
of both experimental and numerical results: velocity proles
are determined as a product of the Mach number proles
obtained from the present numerical simulation and the sonic
velocity based on the experimental temperature proles. The
results for 300600 A with 22.5 and 40 slm of argon for the
partial characteristics method are displayed in gures 17 and
18. Each graph contains four curvesvelocity proles based
on the new and old models (see above), the experimental
velocity prole and the velocity prole obtained by the
integrated approach, we will call it re-calculated velocity
prole (re-calculated experiment in the graphs). It is clearly
visible that agreement of such re-calculated experimental
velocity proles with the numerical ones is much better
than between original experiments and calculation. A large
15
J. Phys. D: Appl. Phys. 44 (2011) 435204 J Jeni sta et al
Figure 18. Velocity proles 2 mm downstream of the nozzle exit for 500 and 600 A with 40 slm of argon. Notation is the same as in
gure 17.
Figure 19. Average relative difference (see the text) between the calculated and re-calculated (the experimental temperature prole and the
calculated Mach number) radial velocity proles, shown in % at the axial position 2 mm downstream of the nozzle exit, the partial
characteristics method. The so-called new model again denotes the present model and exhibits better agreement with experiments than the
old model.
discrepancy between the old and new velocity proles is
also apparent, especially for lower currents.
Figure 19 presents the same type of plot as is presented
in gure 16 but with an analogous denition of the average
relative velocity difference

u
av
=
100
M

i=1
abs(u
i
num
u
i
reexp
)/u
i
num
,
where u
i
num
and u
i
re-exp
are the numerical and re-calculated
velocities at the point i, M is the number of available points
at which the difference is evaluated. It is again evident that
the present new model gives generally a much lower relative
difference than the old model for all studied cases.
The main reason for a relatively large difference between
the experimental and re-calculated velocity proles presented
in gures 17 and 18 are the Mach numbers used in both
approaches. In order to obtain the Mach number from the
measured energy balance of the arc, it has to be assumed in
the simplied energy equation integrated through the discharge
volume that the Mach number is constant over the discharge
radius. This approach underestimates the Mach number in
the region near the arc axis and overestimates it in the arc
fringes. In our opinion this is the main reason for a bigger
difference between the original experimental and numerical
velocity proles. In the integrated approach the Mach
number from numerical simulation is a function of the radius
with the maximum value at the arc axis. We obtained quite
good agreement between the calculated and re-calculated
velocity prolesbelow 12% for 500600 A and somewhat
higher for 400 A (15.4%) and 300 A (24.4%), see gure 19,
and we believe that the re-calculated velocity proles are
more realistic.
16
J. Phys. D: Appl. Phys. 44 (2011) 435204 J Jeni sta et al
Figure 20. Temperature and velocity at the arc axis 2 mm downstream of the nozzle exit. Left: 40 slm of argon, right: 600 A. The
re-calculated velocities shown here are based on the Mach number proles for the partial characteristics method.
Figure 21. The arc efciency as a function of arc current for 40 slm of argon (left). It is dened as = 1 (power losses)/(U I), where
the power losses are due to radiation and radial conduction. Power losses in % is the ratio powerlosses/(U I), shown in dependence of
current and argon mass ow rate (right).
The last set of plots shown in gure 20 presents the values
of temperature and velocity at the arc axis 2 mmdownstreamof
the nozzle exit (at the axial position 60.32 mm) as a function of
arc current for 40 slm of argon (left) and as a function of argon
mass owrate for 600 A(right). It is apparent that reabsorption
of radiation emerging from the partial characteristics method
decreases the temperature in the region near the arc axis.
Temperatures obtained from experiments are closer to those
calculated by the net emission method, except for the case of
600 A and 40 slm of argon where it nearly overlaps with the
temperature obtained from the partial characteristics method.
The velocities obtained from calculation by the two radiation
methods are close to each other. The re-calculated velocities
shown here are based on the Mach number prole for the partial
characteristics method. Difference between the calculated
and re-calculated velocities is in most cases much lower than
between the calculated and original velocity proles estimated
from experiments. The so-called integrated approach is thus
useful as feedback in evaluation and interpretation of data from
experiments.
3.3. Power losses from the arc
Energy balance, responsible for the performance of the hybrid-
stabilized argonwater electric arc, is illustrated in the last set
of graphs. Figure 21 (left) demonstrates the arc efciency
from the arc discharge as a function of current for 40 slm
of argon. The arc efciency is dened here as = 1
(power losses)/(U I) with U being the electric potential
drop in the vortex-stabilized part of the discharge chamber
(along the BC line, from B to the end of the exit nozzle,
gure 4) and I the current. The power losses from the arc
stand for the conduction power lost from the arc in the radial
direction and the radiation power leaving the discharge, which
17
J. Phys. D: Appl. Phys. 44 (2011) 435204 J Jeni sta et al
Figure 22. Dependence of the arc voltage drop in the discharge
chamber on argon mass ow rate for various arc currents.
Experimental values of the arc voltage are taken from [12].
are considered to be the two principal processes responsible
for the power losses. The arc efciency is relatively high and
ranges between 77%and 82%for the net emission coefcients,
and 80% and 84% for the partial characteristics; the difference
of about 34% between the two methods is obviously caused
by the amount of radiation reabsorbed in the colder arc regions
in the partial characteristics. The ratio of the power losses to
the input power in the discharge chamber U I is indicated as
the power losses in a per cent scale, see gure 21 (right). The
power losses slightly increase with increasing argon mass ow
rate and with decreasing current. The reason for a decrease in
the power losses with increasing current is that the power spent
in the discharge region U I increases more steeply than the
radiation and conduction losses from the arc. This tendency is
attributed to the temperature and pressure dependence of the
divergence of radiation ux and of the other properties.
Figure 22 shows the dependence of the arc voltage
drop U on argon mass ow rate for various arc currents.
A comparison with experiments is available for 300500 A
and argon mass ow rates 22.5 and 27.5 slm [12]. In the
experiment, the voltage drop was measured between the
cathode and the exit nozzle. To obtain the voltage drop for
the vortex-stabilized discharge chamber, we have to subtract
the measured voltage drop on the cathode gas-stabilized part of
the arc column, which was in the range 715 V. The appropriate
error bars are displayed with vertical lines. The displayed
results show the highest relative difference of 18% for 300 A
and 27.5 slm of argon, the other cases exhibit lower difference
from 4% to 11%.
Figure 23 (left) displays the typical radial proles of
temperature, divergence of radiation ux and radiation ux
for 600 A and argon mass ow rate of 40 slm. The axial
position is 4 cm from the argon inlet nozzle, i.e. inside the
discharge chamber. Temperature reaches 24 700 K at the axis
and declines to 773 K at the edge of the calculation domain.
The radiation ux reaches 9.7 10
6
Wm
2
at the arc edge
with the maximum magnitude 3.1 10
7
Wm
2
at a radial
distance of 2.2 mm. The divergence of radiation ux becomes
negative at the radial distance over 2.6 mm, i.e. the radiation
is reabsorbed in this region. Even though the negative values
of the divergence of radiation ux in arc fringes are relatively
small compared with the positive ones in the axial region, the
fraction of the emitted radiation that is reabsorbed (dened
as the ratio of the negative and positive contributions of the
divergence of the radiative ux, see below) is 32.4%, because
the plasma volume increases with the third power of the radius.
Figure 23(right) shows the amount of reabsorbedradiation
(%) in the argonwater mixture plasma within the arc discharge
for currents 300600 A as a function of argon mass ow
rate. The negative and positive parts of the divergence of
radiation ux are integrated through the discharge volume.
Reabsorption dened here is the ratio of the negative and
positive contributions of the divergence of radiation uxit
ranges between 31%and 45%and increases for lower contents
of argon in the mixture.
Adirect comparison of the amount of reabsorbed radiation
with experiments is unavailable; however, the indirect sign of
validity of our results is the very good agreement between the
experimental and calculated radial temperature proles 2 mm
downstream of the outlet nozzle presented above.
4. Conclusions
A numerical model for an electric arc in a plasma torch with
the so-called hybrid stabilization, i.e. combined stabilization
of arc by gas and water vortex, is presented. To study
the possible compressible phenomena in the plasma jet,
calculations are carried out for the interval of currents 300
600 A and for relatively high argon mass ow rates between
22.5 and 40 slm. The partial characteristics as well as the net
emission coefcient methods for radiation losses from the arc
are employed. The results of the present calculation can be
summarized as follows:
(a) The numerical results showed that transition to the
supersonic regime starts around 400 A. The supersonic
structure with shock diamonds occurs in the central parts of the
discharge in the outlet region. The computed proles of axial
velocity, pressure and temperature correspond to an under-
expanded atmospheric-pressure plasma jet.
(b) The partial characteristics radiation method gives slightly
lower temperatures but in most cases higher outlet velocities
and the Mach numbers compared with the net emission
coefcient method. It was conrmed that the net emission
radiation method provides somewhat better agreement with
experiment compared with the partial characteristics as
regards radial proles of temperature 2 mm downstream from
the nozzle exitmaximum relative difference between the
calculated and experimental temperature proles is lower than
5% for the net emission coefcients, and below 10% for the
partial characteristics. It was also found that the average
relative difference between the calculated and re-calculated
velocity proles is a little better for the partial characteristics.
(c) Reabsorption of radiation ranges between 31% and 45%, it
decreases with current and also slightly decreases with argon
18
J. Phys. D: Appl. Phys. 44 (2011) 435204 J Jeni sta et al
Figure 23. Radial proles of temperature, divergence of radiation ux and radiation ux for 600 A and argon mass ow rate of 40 slm inside
the discharge chamber at the axial position of 4 cm (left); partial characteristics. Reabsorption of radiation occurs at 2.6 mm from the axis.
Reabsorption of radiation (right) for different currents and argon mass ow rates is dened as the ratio of the negative to the positive
contributions of the divergence of radiation uxit ranges between 30% and 45% and slightly decreases for higher argon mass ow rates.
mass ow rate. The arc efciency reaches up to 7784%, the
power losses fromthe arc due toradiationandradial conduction
are between 16% and 24%.
(d) Comparison with available experimental data demonstrated
very good agreement for temperaturethe maximum
relative difference between the calculated and experimental
temperature proles is lower than 10% for the partial
characteristics and5%for the net emissioncoefcient radiation
methods. The calculated radial velocity proles 2 mm
downstream of the nozzle exit show good agreement with
those evaluated from the combination of calculation and
experiment (integrated approach). Agreement between the
calculated radial velocity proles and the proles analysed
purely from experimental data is worse. Evaluation of the
Mach number from the experimental data for 500 and 600 A
gives values higher than that close to the exit nozzle, it
thus proves the existence of the supersonic ow regime.
The present numerical model also provides better agreement
with experiments than our previous model based on the
simplied transport, thermodynamic and radiation properties
of an argonwater plasma mixture.
The existing numerical model will be further extended to
study the effect of mixing of plasma species within the hybrid
arc discharge by the binary diffusion coefcients [46, 47] for
three specieshydrogen, argon and oxygen.
Acknowledgments
J Jeni sta is grateful for nancial support under the Fluid
Science International Collaborative Research Project (J11059,
J10049, J09016) from the Institute of Fluid Science, Tohoku
University, Sendai, Japan, and their computer facilities.
Financial support from the projects GA CR 205/11/2070 and
M100430901 from the Academy of Sciences AS CR, v.v.i.,
is gratefully acknowledged. Our appreciation goes also to
the Institute of Physics AS CR, v.v.i., for granting their
computational resources (Luna/Apollo grids). The access to
the METACentrum supercomputing facilities provided under
the research intent MSM6383917201 is highly appreciated.
References
[1] Chr aska P and Hrabovsky M 1992 An overview of water
stabilized plasma guns and their applications Proc. Int.
Thermal Spray Conf. 13 ITSC (Orlando, FL, 25 May5
June) ed C C Berndt (Materials Park, OH: ASM
International) pp 815
[2] Hrabovsk y M, Konr ad M, Kopeck y V and Sember V 1997
Processes and properties of electric arc stabilized by water
vortex IEEE Trans. Plasma Sci. 25 8339
[3] Hrabovsk y M, Kopeck y V and Sember V 1995 Water
stabilized arc as a source of thermal plasma Heat and Mass
Transfer under Plasma Conditions (Cesme, Turkey, 48 July
1994) ed P Fauchais (New York: Begell House) pp 918
[4] Hrabovsk y M 1998 Water-stabilized plasma generators Pure
Appl. Chem. 70 115762
[5] Hrabovsk y M, Konrad M, Kopecky V and Sember V 1999
Properties of electric arc stabilized by mixture of water
with ethanol Heat and Mass Transfer under Plasma
Conditions (Tekirova, Antalya, Turkey, 1823 April 1999)
ed P Fauchais et al (Annals of the New York Academy of
Sciences vol 891) (New York: New York Academy of
Sciences) pp 5763
[6] Hrabovsk y M 2009 Thermal plasma generators with water
stabilized arc Open Plasma Phys. J. 2 99104
[7] Jeni sta J 1999 Water-vortex stabilized electric arc: I.
Numerical model J. Phys. D: Appl. Phys. 32 276376
[8] Jeni sta J 1999 Water-vortex stabilized electric arc: II. Effect of
nonuniform evaporation of water J. Phys. D: Appl. Phys.
32 277784
[9] Hrabovsky M, Konrad M, Kopecky V and Sember V 2000
Properties of water stabilized plasma torches Thermal
Plasma Torches and Technologies vol 1 ed O P Solonenko
(Cambridge, MA: Cambridge International Science
Publishing) pp 24266
[10] Brezina V, Hrabovsk y M, Konr ad M, Kopeck y V and Sember
V 2001 New plasma spraying torch with combined
gasliquid stabilization of arc Proc. 15th Int. Symp. on
Plasma Chemistry ISPC 15 (Orleans, France, 913 July)
19
J. Phys. D: Appl. Phys. 44 (2011) 435204 J Jeni sta et al
ed A Bouchoule et al (Orleans: ISPC 15 Local Organizing
Committee) pp 10216
[11] Hrabovsk y M, Kopeck y V and Sember V 2003 Effect of gas
properties on characteristics of hybrid gas/water plasma
spraying torch Proc. 16th Int. Symp. on Plasma Chemistry
ISPC 16 (Taormina, Italy, 2227 June) ed R dAgostino
(Taormina: ISPC 16 Local Organizing Committee) on
CD-ROM
[12] Hrabovsk y M, Kopeck y V, Sember V, Kavka T, Chumak O and
Konr ad M 2006 Properties of hybrid water/gas DC arc
plasma torch IEEE Trans. Plasma Sci. 34 156675
[13] Chr aska T, Neufuss K, Dubsk y J, Ctibor P and Rohan P 2008
Fabrication of bulk nanocrystalline alumina-zirconia
materials Ceram. Int. 34 122936
[14] Chr aska T, Neufuss K, Dubsk y J, Ctibor P and Klementov a M
2008 Fabrication of bulk nanocrystalline ceramic materials
J. Thermal Spray Technol. 17 8727
[15] Chr aska T, Hostomsk y J, Klementov a M and Dubsk y J 2009
Crystallization kinetics of amorphous
aluminazirconiasilica ceramics J. Eur. Ceram. Soc.
29 315965
[16] Ctibor P, Ageorges H, Sedl a cek J and

Ctvrtlk R 2010
Structure and properties of plasma sprayed BaTiO
3
coatings
Ceram. Int. 36 215562
[17]

Stengl V, Ageorges H, Ctibor P and Murafa N 2009
Atmospheric plasma sprayed (APS) coatings of
Al
2
O
3
TiO
2
system for photocatalytic application
Photochem. Photobiol. Sci. 8 7338
[18] Van Oost G, Hrabovsk y M, Kopeck y V, Konr ad M, Hlna M,
Kavka T, Chumak O, Beeckman E and Verstraeten J 2006
Pyrolysis of waste using a hybrid argonwater stabilized
torch Vacuum 80 11327
[19] Van Oost G, Hrabovsk y M, Kopeck y V, Konr ad M, Hlna M
and Kavka T 2008 Pyrolysis/gasication of biomass for
synthetic fuel production using a hybrid gaswater
stabilized plasma torch Vacuum 83 20912
[20] Jeni sta J 2004 Numerical modeling of hybrid stabilized
electric arc with uniform mixing of gases IEEE Trans.
Plasma Sci. 32 46472
[21] Jeni sta J, Bartlov a M and Aubrecht V 2007 The impact of
molecular radiation processes in water plasma on
performance of water-vortex and hybrid-stabilized electric
arcs Proc. 34th IEEE Int. Conf. on Plasma Science and The
16th IEEE Int. Pulsed Power Conf. PPPS-2007
(Albuquerque, NM, 1722 June) (IEEE Catalog Number:
07CH37865) pp 142932
[22] Jeni sta J, Takana H, Hrabovsk y M and Nishiyama H 2008
Numerical investigation of supersonic hybrid
argonwater-stabilized arc for biomass gasication IEEE
Trans. Plasma Sci. 36 10601
[23] Jeni sta J, Takana H, Nishiyama H, Bartlova M, Aubrecht V
and Hrabovsk y M 2010 Parametric study of hybrid
argonwater stabilized arc under subsonic and
supersonic regimes J. High Temp. Mater. Process. 14 6376
[24] Krenek P 2008 Thermophysical properties of H
2
OAr plasmas
at temperatures 40050 000 K and pressure 0.1 MPa Plasma
Chem. Plasma Process. 28 10722
[25] Krenek P and N eni cka V 1984 Calculation of transport
coefcients of a gas mixture in the 4-th approximation of
EnskogChapman theory Acta Tech. CSAV 28 42033
[26] Soloukhin R I 1980 Radiative Heat Transfer in
High-Temperature Gases (London: Hemisphere)
[27] Sevastyanenko V G 1979 Radiation transfer in a real
spectrum. Integration over frequency J. Eng. Phys.
36 13848
[28] Aubrecht V, Bartlov a M, Urban F and Valenta J 2004 Partial
characteristics of radiation for thermal plasmas in H
2
O and
argon Proc. 15th Int. Conf. on Gas Discharges and their
Applications GD 2004 (Toulouse, France 510 September)
ed M C Bordage et al (Toulouse: GD 2004 Local
Organizing Committee) pp 1414
[29] Aubrecht V and Lowke J J 1994 Calculations of radiation
transfer in SF
6
plasmas using the method of partial
characteristics J. Phys. D: Appl. Phys. 27 206673
[30] Aubrecht V and Bartlov a M 1997 Calculations of radiative
heat transfer in SF
6
arc plasmas IEEE Trans. Plasma Sci.
25 81523
[31] Liebermann R W and Lowke J J 1976 Radiation emission
coefcients for sulfur hexauoride arc plasmas J. Quant.
Spectrosc. Radiat. Transfer 16 25364
[32] Bartlov a M and Aubrecht V 2006 Photoabsorption of diatomic
molecules Czech. J. Phys. 56 (Suppl. B) B6327
[33] Jeni sta J 2003 Water-vortex stabilized electric arc: III. Radial
energy transport, determination of water-vapour-boundary
and arc performance J. Phys. D: Appl. Phys. 36 29953006
[34] Jeni sta J, Kopeck y V and Hrabovsk y M 1999 Effect of vortex
motion of stabilizing liquid wall on properties of arc in
water plasma torch Heat and Mass Transfer under Plasma
Conditions (Tekirova, Antalya, Turkey, 1823 April 1999)
ed P Fauchais et al (Annals of the New York Academy of
Sciences vol 891) (New York: New York Academy of
Sciences) pp 6471
[35] Patankar S V 1980 Numerical Heat Transfer and Fluid Flow
(New York: McGraw-Hill)
[36] Jameson A and Yoon S 1987 Lowerupper implicit schemes
with multiple grids for the Euler equations AIAA J.
25 92935
[37] Yoon S and Jameson A 1988 Lowerupper
symmetric-GaussSeidel method for the Euler and
NavierStokes equations AIAA J. 26 10256
[38] Kwak D, Kiris C and Housman J 2011 Implicit methods for
viscous incompressible ows Computers Fluids 41 5164
[39] Roe P L 1981 Approximate Riemann solvers, parameter
vectors, and difference schemes J. Comput. Phys.
43 35772
[40] Leer van B 1979 Towards the ultimate conservative difference
scheme: V. A second-order sequel to Godunovs method J.
Comput. Phys. 32 10136
[41] Chung T J 2010 Computational Fluid Dynamics 2nd edn (New
York: Cambridge University Press)
[42] Jeni sta J 2003 The effect of different regimes of operation on
parameters of a water-vortex stabilized electric arc J. High
Temp. Mater. Process. 7 1116
[43] Kavka T, Chumak O, Sember V and Hrabovsk y M 2007
Processes in Gerdien arc generated by hybrid gaswater
torch Proc. 28th Int. Conf. on Phenomena in Ionized Gases
ICPIG 2007 (Prague, Czech Republic, 1520 July) ed J
Schmidt et al (Prague: Institute of Plasma Physics AS CR,
v.v.i.) pp 181922
[44] Kavka T, Maslani A, Chumak O and Hrabovsk y M 2008
Character of plasma ow at the exit of DC arc gaswater
torch Proc. 5th Int. Conf. on Flow Dynamics ICFD 5
(Sendai, Japan, 1719 November) ed S Maruyama (Sendai:
5th ICFD, Local Organizing Committee) pp OS8-11
[45] Pateyron B and Katsonis C 2009 T&TWinner, available from
http://ttwinner.free.fr/
[46] Murphy A B 1993 Diffusion in equilibrium mixtures of
ionized gases Phys. Rev. E 48 3594603
[47] Murphy A B 2001 Thermal plasmas in gas mixtures J. Phys.
D: Appl. Phys. 34 R15173
20