ISIJ International, Vol. 53 (2013), No. 2, pp.

213–220

Arc Behaviour and Cathode Melting Process during VAR:

an Experimental and Numerical Study

Alain JARDY,* Pierre CHAPELLE, Ashish MALIK, Jean-Pierre BELLOT, Hervé COMBEAU
and Bernard DUSSOUBS

Institut Jean Lamour – UMR 7198 CNRS/Université de Lorraine, Département SI2M Ecole des Mines, Parc de Saurupt, CS
14243, F-54042 Nancy Cedex, France.
(Received on February 15, 2012; accepted on November 20, 2012)

The present study aims to understand the melting of the consumable electrode in the VAR process and
gain some insight into the influence of an ensemble arc motion on the melting behaviour. In a previous
study, a 2D axisymmetric model of the heat transfer in the cathode had been developed. Using the oper-
ating parameters as model inputs, it enabled prediction of the melt rate and the evolution of the melting
area. Model results were successfully compared to melt rate measurements in an industrial VAR furnace.
In recent years, it has been claimed that the electric arc may not be considered as steady and axisymmet-
ric. Our experimental investigation of the luminosity recorded during an actual VAR heat confirms that a
transient 3D behaviour may take place. Therefore, a 3D version of the previous model was set up to pre-
dict the heat transfer and melting of the electrode, using the unknown ensemble arc motion as an input.
The arc is assimilated to a transient distribution of energy flux density. Results evidence that the influence
of the arc motion on the shape of the electrode tip can be very important. In industrial practice, the cath-
ode tip usually remains relatively flat during melting. The shapes of the computed electrode tips enable
us to propose some arc parameters which remain compatible with both the periodic behaviour of the light
emitted and the flatness of the electrode.

KEY WORDS: Vacuum Arc Remelting; cathode melting; ensemble arc motion; numerical simulation.

metals, such as zirconium and titanium.3) The strategic

1. Introduction
importance of these products and their very high added val-
The principle of the Vacuum Arc Remelting (VAR) pro- ue make it essential to acquire a detailed understanding of
cess, as illustrated in Fig. 1, consists in melting a consum- the melting process. The work presented here is part of a
able metallic electrode of the required grade under a high
vacuum, in order to obtain a sound ingot of good structural
quality. During remelting, an electric arc is maintained
between the tip of the consumable electrode (usually the
cathode) and the top of the secondary ingot (usually the
anode), in order to ensure melting of the electrode. The met-
al droplets fall through the arc plasma and progressively
build up the ingot, which solidifies in contact with the walls
of a water-cooled copper crucible. This operation ensures
both the purification of the metal, as it is gradually melted,
and controlled solidification of the ingot in terms of struc-
ture and chemical homogeneity. In order to stabilize the arc,
it can be confined with the aid of a magnetic field created
by an external induction coil. This field can also be used to
stir the liquid metal, the rotation induced being in the azi-
muthal direction. By reversing periodically the coil current,
stirring can be alternated.
Vacuum Arc Remelting is currently used to produce high
quality special steels and nickel-based superalloys.1,2) It also
represents the final stage in the melting cycle of reactive

* Corresponding author: E-mail: [email protected]

DOI: http://dx.doi.org/10.2355/isijinternational.53.213 Fig. 1. A schematic representation of the VAR process.

213 © 2013 ISIJ

 ISIJ International, Vol. 53 (2013), No. 2

program initiated twenty years ago at Institut Jean Lamour

(Nancy School of Mines) within the frame of industrial col-
laborations with several process users.4–6) The aim is to
develop a simulation software of the whole VAR operation,
and subsequently to help optimizing the process.
In this paper, we focus on the macroscopic behaviour of
the arc and the associated cathode melting process. Indeed,
a thorough understanding of the dynamics of the arc is
important for a proper control of the process performance
and ingot quality. The spatial and temporal distribution of
the arc determines the energy and electric current inputs to
Fig. 2. Example of a video image of an electric arc during vacuum
the ingot top surface and electrode tip, which control to a
arc remelting of a zirconium alloy, showing clusters of
great extent the ingot and electrode evolution. cathode spots.
Direct knowledge of the electric arc behaviour in the VAR
process is based on visualization studies performed first at
Sandia National Laboratories7,8) during the remelting of
steel or Ni-based superalloy electrodes. Similar experiments
have been carried out more recently by Chapelle et al. on
zirconium electrodes.9,10) One of the main conclusions from
these observations is that the behaviour of the arc is similar
to the diffuse mode of a vacuum arc created between cold
solid electrodes for the same range of current densities. The
arc consists of several dispersed clusters of cathode spots
Fig. 3. Schematics of the heat flux density distribution at the cath-
moving very rapidly over the whole surface of the cathode, ode base (uniform heating = 2D assumption).
as shown on Fig. 2. Each cluster of spots is the source of a
plasma jet expanding under vacuum in the electrode gap. No
luminous arc column is observed in the gap.
Such a behaviour seems to imply that the total energy
transferred from the arc to the cathode tip is distributed
more or less uniformly; in particular, no azimuthal direction
is privileged, so a 2D axisymmetric behaviour is expected
at the macroscopic scale. This is consistent with the fact that
the cathode tip remains relatively flat during full-scale melt-
Fig. 4. Schematics of the heat flux density distribution at the cath-
ing. Figure 3 represents schematically the cathode behav- ode base (uniform heating + rotating ‘arc-spot’).
iour in such a situation.
However, it has been recently reported that the arc often
does not behave axisymmetrically at the macroscopic scale. to simulate the cathode tip melting process in order to check
Based on measurements of the luminosity and magnetic the compatibility between the tip flatness and the existence
field created by the arc,11,12) Ward et al. suggested that most of an ensemble arc motion.
of the time, the centre of gravity of the arc was off-centered
and rotated in a time-averaged sense around the electrode
2. Observation of a Macroscopic Ensemble Motion of
axis with a constant speed (period equals typically 20 to 40
the Electric Arc
s when a superalloy electrode melts under nominally diffuse
conditions). Then it was assumed that the distribution of In this section, we report a visual investigation of the
current flow and heat input follows the distribution of this dynamic behaviour of the electric arc in an industrial VAR
location and a part of the arc was assimilated to a loosely furnace of coaxial type under various operating conditions.
focused rotating spot, radially located away from the ingot For this investigation, two synchronized video cameras
centreline. Such a behaviour might result (see Fig. 4) in a positioned in front of two diametrically opposite viewing
3D shape of the cathode tip. A 3D model of the ingot pool, glasses on top of the furnace chamber were used to film the
using this representation as a boundary condition,13,14) ena- annulus gap between the crucible wall and the electrode
bled several authors to conclude that the hydrodynamic (Fig. 5). This apparatus enabled us to observe only a portion
behaviour of the melt pool and ingot solidification process of the ingot top surface area adjacent to the crucible wall
can be greatly influenced by such an ensemble arc macro- and a luminous halo above it, which is generated by the arc.
scopic motion. The arc itself, which is essentially established at the elec-
However, the observation of an actual cathode tip, some- trode bottom surface, was not directly visible to the cameras.
times possible after a melt interruption, usually does not Video images were recorded during the melt of a zirconi-
show the presence of any localized excavation correspond- um alloy (Zy2) ingot. During this melt, five stirring
ing to the rotation of a spot. This apparent contradiction sequences, which are listed in Table 1, were used succes-
requires more research study to be achieved. From the sively in order to study the influence on the arc behaviour
experimental point of view, new measurements are needed of the presence of an external axial magnetic field (as stated
to characterize the arc behaviour. In parallel, it was decided in the introduction of this paper, such a field is employed

© 2013 ISIJ 214

 ISIJ International, Vol. 53 (2013), No. 2

iour, which was not visible on the unprocessed images. The

luminosity fluctuates quite regularly along both sequences,
with an alternation between very bright time periods and
other time periods during which the luminosity is notably
reduced. The bright time periods are generally shorter than
the darker ones. The fluctuations of luminosity in the two
diametrically opposed regions are essentially in phase oppo-
sition, i.e. when the luminosity passes through a maximum
in a region, it tends to be minimum in the opposite region.
A frequency analysis indicates that the fluctuations involve
several periods, with a dominating period of the order of 30
s, which is identical for the two cameras.
These fluctuations may be related to the arc behaviour.
Indeed, it can reasonably be considered that the luminosity
fluctuates as a consequence of the evolution of the cathode
spots forming the arc, due to their motion over the surface
of the electrode. Namely, a high luminosity indicates the
presence of an important number of spots on the electrode
surface close to this region, whereas a low luminosity
Fig. 5. VAR furnace equipped with two video cameras. denotes the presence of a small number of spots (or even the
absence of spots) nearby. On the basis of this assumption,
the fluctuations of the luminosity described above indicate
Table 1. Dominating periods of the luminosity fluctuations for dif-
ferent stirring sequences.
that the presence of the spots in a given region tends to vary
periodically with time. Moreover, the fact that the fluctua-
Stirring sequence camera 1 camera 2 tions on both sides are not in phase reveals that the cathode
#1 No stirring 28.1 s 29.9 s spots are not evenly spread over the cathode surface and that
the centre of gravity of their distribution (which can be
#2 Unidirectional (very low B) 29.9 s 23.9 s
regarded as the centre of gravity of the arc) is off-centre
#3 Alternated (low B, long period) 28.1 s 28.1 s with respect to the axis of symmetry of the electrode. The
#4 Alternated (high B, short period) 43.5 s 43.5 s combination of these two findings may be interpreted as an
#5 Alternated (low B, short period) 28.1 s 28.1 s evidence of the existence of an apparent macroscopic
motion of the centre of gravity of the arc across the elec-
trode surface (with a period of about 30 s). Namely, while
under usual practice to stir the metal in the liquid pool and the spots tend to cover more or less the whole electrode sur-
to confine the arc underneath the electrode). These condi- face in a time average sense, their spreading behaviour may
tions differ by the type (unidirectional or alternating with a be such that the centre of their distribution tends to move
periodic reversal of the field direction) and the magnitude of macroscopically over the surface with a given period. The
the magnetic field. Note that visualisations were also per- precise nature of this motion, which could be either a rota-
formed in the absence of any magnetic field. tional or a translation displacement, cannot be determined
To help interpreting the recorded films, a specific image from the present results.
processing procedure similar to that proposed by Ward et al.11) Note that an alternative scenario could lead to similar
was developed. The main steps of this procedure are sum- luminosity fluctuations as those shown in Fig. 7. In this sce-
marized below. First, the recorded film was split into a nario, the centre of the arc would be off-centre but station-
series of images. From these images, a new series of images ary, with spots arising at the centre location and expanding
was then generated by applying a 2 s moving average in radially outward towards the electrode periphery where they
order to suppress high frequency fluctuations related to indi- extinguish. However, this latter scenario must be rejected, as
vidual cathode spot behaviour and by reducing the sampling the time necessary for the spots to spread over the electrode
frequency to 5 frames/s. In a third step, a given region of radius (of the order of several milliseconds, given a typical
interest was extracted from each image and all the results velocity of 10 m/s15)) is much less than the time scale of the
were put side by side to build a temporal sequence (Fig. 6). luminosity fluctuations measured in our study.
Finally, the average luminosity in each extracted region was This phenomenon was observed most of the time for all
quantified and a Fourier analysis was performed to deter- the melt conditions tested. As shown in Table 1, the domi-
mine the frequencies of luminosity fluctuations along the nant period of the luminosity fluctuations is of the same
temporal sequence. order of magnitude (30 to 40 s) for all melt conditions
An example of two temporal sequences corresponding to (including the conditions without any stirring). It seems in
two diametrically opposite regions is illustrated in Fig. 7 for particular to be unconnected to the reversal period of the
the case of stirring conditions #3 (see Table 1). Each magnetic field. The reason for the larger value of the period
sequence spans a 4 min time interval. A plot of the temporal in the case of high field conditions (melt conditions #4)
evolution of the average luminosity for both sequences is remains unclear at this stage. Thus, an ensemble motion of
also presented on the figure. the arc seems to exist for various operating conditions and
The analysis of Fig. 7 revealed some unexpected behav- it appears to be relatively independent of the presence of an

215 © 2013 ISIJ

 ISIJ International, Vol. 53 (2013), No. 2

external axial magnetic field. Nevertheless, it should be ularly studied in the subsequent sections. The displaced centre
mentioned that in the absence of an external magnetic field of gravity of the electric arc is referred to as an ‘arc-spot’ in
(melt conditions #1), the fluctuations were less regular and the rest of this paper. Several questions clearly arise, namely:
the occurrence of relatively long periods with no significant • Does an arc-spot motion induce a non-axisymmetry in
fluctuations was more frequent. In contrast, the fluctuations the shape of the consumable electrode?
of the luminosity were the most regular in the case of a • What are the effects of various parameters of arc-spot
strong magnetic field with a short alternating period (melt like angular speed, power, distance from the centre, etc.
conditions #4). A possible explanation for this could be a on the cathode tip shape?
more efficient confinement of the cathode spots underneath • What causes this arc-spot motion?
the electrode due to a higher intensity of the magnetic field. • Does the evolution of the cathode tip shape affects the
This experimental part of our work enables us to confirm arc-spot motion, i.e. is there any back effect?
the conclusions reached by Ward et al.12) who reported a val- These are some of the points which need to be answered
ue of the time constant of the arc motion very similar as the for a better understanding of the VAR process. The aim of
one determined here. As discussed previously, the existence this paper is to give a clearer insight as far as the first two
of a slow motion of the arc centre (with a time period of points are considered.
around 30 s) could have important implications for the mod-
elling of the VAR process as computational models based on
a 2D axisymmetric geometry2,5) might not be sufficient to
describe the ingot and electrode behaviours (such models
use integration time steps smaller than 1 s).
The effects of a non-axisymmetric and time varying dis-
tribution of the arc on the electrode behaviour (in terms of
variations of the shape of the electrode tip) are more partic-

Fig. 6. Temporal sequence used to study the fluctuations of lumi- Fig. 7. Typical temporal sequences obtained for two diametrically
nosity above the ingot top. opposite regions.

Fig. 8. Computed temperature field (°C) in the cylindrical electrode – Zy4 remelting, uniform arc flux.

© 2013 ISIJ 216

 ISIJ International, Vol. 53 (2013), No. 2

arc motion.
3. Modeling of the Cathode Melting
The total power transferred to the cathode χ Ptot is the
This section briefly describes the numerical modelling of sum of a part distributed uniformly all over the bottom sur-
the transient thermal behaviour of the cathode. Basically, face of the cathode and a part concentrated in the small area
it is based on the 1D approach developed much earlier by of the arc-spot. The boundary condition is then, at every
Bertram and Zanner,16) as well as Jardy et al.17) However, an moment:
original feature of the present model is the representation of (1 − α ) χ Ptot (t ) α χ Ptot (t )
liquid metal removal at the electrode base, which allows its ϕ (r ,θ , t ) = + f (r ,θ , t )
π Relec
2
π Rspot2
use to predict the time evolution of the melt rate and the
.......................................... (3)
shape of the electrode tip for given operating parameters.
and the second term on the r.h.s. of Eq. (3) is only applied
3.1. Heat Transfer Equation if the considered location is concerned by the arc-spot (f =
Within the electrode, heat is transferred by conduction. 1 under the arc-spot, and f = 0 outside). The arc-spot focus
Heating of the electrode by Joule effect is neglected, Rspot , as well as the variable α, are input parameters of the
because of the very low resistivity of the metal to be remelt- model. Obviously, setting α = 0 degenerates the 3D model
ed. Under these conditions, the heat transport equation is into an axisymmetric one.
written in an enthalpic formulation:18)
3.3. Calculation Procedure
∂
( ρ h) = ∇ ⋅ (λ ∇ T ) ........................ (1) The heat transfer equation is discretized using a finite vol-
∂t ume method and fully implicit time scheme.19) During each
where h is the mass average enthalpy, T the temperature, ρ time step, we first calculate the temperature field in the elec-
the density and λ the thermal conductivity. The temperature trode from Eq. (1). For each mesh cell that has no bottom
dependence of all alloy thermophysical properties is taken neighbour, i.e. that is exposed to the arc, the energy flux
into account. received is simply calculated as:
In Eq. (1), assuming a constant and equal specific heat in
the solid and liquid phases, the enthalpy can be expressed as φbot (i ) = ∫∫ ϕ (r ,θ , t ) r dr dθ ................. (4)
S (i )
a function of the temperature and liquid fraction: h = CpT +
gl Hlat, where Cp is the specific heat, gl the liquid fraction and where index i refers to the cell in the finite volume mesh and
Hlat the latent heat of melting. Thus, solving Eq. (1) requires S(i) is the area of the bottom surface of the cell. At each time
knowledge of the solidification path, i.e. the evolution of the step, the new position of the arc-spot is calculated, and
liquid fraction with temperature. φ bot(i) is determined according to Eqs. (3) and (4).
Then, in order to simulate the consumption of the elec-
3.2. Boundary Conditions trode associated to the fall of liquid metal droplets formed
To obtain a full definition, boundary and initial conditions at the electrode tip, mesh cells whose temperature is greater
must be specified. Initially, the electrode temperature is con- than an “overheating temperature” Tliq + ΔTov are removed
sidered to be homogeneous and equal to the stub tempera- from the computational domain. After cell removal, the
ture. At the top of the electrode, a contact resistance boundary conditions are set at the new electrode tip for the
between the electrode and stub is taken into account while next time step. Calculations are performed until fully con-
at the electrode side, heat transfer is controlled by thermal sumption of the electrode.
radiation between the electrode and mould. The value of the It is important to notice that the melting of the electrode
emissivity is assumed to remain constant throughout the may induce some numerical irregularities in the shape of the
melting process. base. The simple approach presented here can therefore be
At the electrode base, the power brought by the electric improved by a careful distribution of the arc energy in such
arc (originating from localised Joule heating due to the high regions. In the 2D case, it has been shown20) that such a
current density or/and bombardment of the surface by the modification leads to only minor modifications in the cath-
plasma ions) is expressed by a condition of the Neumann ode tip behaviour. However, this remains to be done in the
type (imposed flux density). The total energy flux density 3D case.
supplied by the electric arc must obey the relation: Typical results of the model are the computed T(r, θ , z)
2π Relec
temperature map at any moment in the remaining electrode,
∫0 ∫0 ϕ (r ,θ , t ) r dr dθ = χ Ptot (t ) ............ (2) hence the instantaneous shape of the electrode base, as well
as the evolution of the electrode weight and overall melting
where Relec is the electrode radius and χ is the fraction of the rate with time. An additional interesting feature of the model
total power Ptot (t) = Uarc Iarc that is transferred to the elec- is its ability to account for the thermal effect of a crack
trode. The distribution of the energy flux density brought by located at the interface between two cells (e.g. the welding
the arc at the electrode tip is an input parameter of the mod- of electrode pieces) by means of a thermal resistance.21)
el. Indeed, Figs. 3 and 4 schematically present this distribu-
tion in two cases (2D and 3D versions). 3.4. Model Validation
The 3D model considers the existence of an arc-spot lia- The 3D model has been used (with α = 0) to simulate the
ble to move (as an example, a simple rotational motion is melting of a full-scale Zy4 electrode that had been presented
considered in Fig. 4) on the electrode base. This oversim- previously22) when only a 2D version of the model (implic-
plified representation enables us to account for an ensemble itly setting α to 0) was available. The thermophysical prop-

217 © 2013 ISIJ

 ISIJ International, Vol. 53 (2013), No. 2

Table 2. Thermophysical properties of the cathode material (zirco- Table 3. Operating parameters used for all simulations.
nium alloy Zy4).
Electrode radius Relec (m) 0.1
Thermal conductivity λ (W/m/K) 43.5
Melting current Iarc (A) 6.5×103
Density ρ (kg/m3) 6.24×103
Arc voltage Uarc (V) 32
Specific heat Cp (J/kg/K) 4.61×102
Fraction of power transmitted to the cathode χ 0.58
Latent heat of melting Hlat (J/kg) 2.19×105
Overheating temperature interval ΔTov (°C) 100
Emissivity ε 0.33
Phase change temperatures Tsol – Tliq (°C) 1 800 – 1 830 Table 4. Input data for the whole set of runs.

Arc-spot Arc-spot Radial Motion

Serial n° power focus location frequency
fraction α Rspot (m) dist (m) f (rpm)
1 (uniform
0 / / /
density)
2 (reference) 0.05 0.2 Relec 0.7 Relec 2
3 0.025 0.2 Relec 0.7 Relec 2
4 0.1 0.2 Relec 0.7 Relec 2
5 0.05 0.1 Relec 0.7 Relec 2
6 0.05 0.3 Relec 0.7 Relec 2
7 0.05 0.2 Relec 0.3 Relec 2
8 0.05 0.2 Relec 0.5 Relec 2

Fig. 9. Time evolution of both monitored and calculated melt rates – 9 0.05 0.2 Relec spiral 2
Zy4 remelting. 10 0.05 0.2 Relec 0.7 Relec 1
11 0.05 0.2 Relec 0.7 Relec 4
erties of the cathode material are listed in Table 2.
12 0.05 0.2 Relec 0.7 Relec 8
The 2D model predicted that the most important temper-
ature axial gradient remained confined at the bottom, close
to the electrode tip. Also, the concave shape of the base correspond to any actual melt. All runs were performed
resulted directly from the thermal losses by radiation along using an initial 3D meshing of 2 700 000 cells (100-90-300
the lateral surface of the electrode. All these conclusions are in r-θ-z directions).
exactly confirmed when the 3D model is used as the situa- As compared to the concave and almost flat electrode
tion degenerates to axisymmetry when no ensemble arc generally obtained in the case of uniform heating,21,22) the
motion is accounted for, so numerical results are identical. presence of an arc-spot results in a 3D shape of the electrode
Figure 8 shows the computed temperature field of the tip, even when a low value is set for α. As an example, Fig.
cathode at around half of the overall melting duration. The 10 shows the computed map of the electrode temperature at
insert zoom enables us to visualize the concave shape of the a given moment (the observation plane corresponds to the
electrode base. instantaneous arc-spot location).
Figure 9 depicts the temporal evolution of the predicted In that simulation, the arc-spot is assumed to carry α =
melt rate of the electrode as well as the melt rate monitored 5% of the total heat input. It rotates around the electrode
during the time interval throughout which the arc power was centreline with a 30 s period. Its radius Rspot is 20% of the
maximum. The variation in the overall melt rate, including electrode radius Relec and the location of its centre is dist =
an initial pre-heating stage followed by a sharp increase of 0.3 Relec. Such conditions correspond roughly to magnetic
the melt rate until reaching a quasi-stationary regime, is well measurements by Ward et al.12) The computed non-flatness
described by the model. The average calculated melt rate of the electrode tip seems very pronounced in this case.
was constant (equal to 5.9 kg/min) throughout most of the Indeed, if a “concavity” is defined as (Hmax – Hmin)/Relec,
melt. This value was found to be in excellent agreement where Hmax (resp. Hmin) is the maximum (resp. minimum)
with that monitored during the real melt (5.7 kg/min), which electrode height, its value reaches around 0.15. This result
validates the model. does not agree with usual industrial observation of the cath-
ode tip.
In order to study the influence of the arc behaviour on the
4. Influence of an Ensemble Arc Motion: Results and
melting process, a set of 12 runs has been achieved with var-
Discussion
ious values for the main arc-spot parameters, that are:
In this section, we study the influence of a non-zero value • α, the fraction of total power carried by the arc-spot
of the parameter α, which implies that part of the heat input • Rspot, the arc-spot focus
at the cathode tip is not uniformly distributed, but rather • dist, the radial location of the arc-spot centre
through an arc-spot moving on the electrode base. The main • f, the frequency of its rotational motion
operating parameters for all calculations discussed below Table 4 details the values that have been used for the
are reported in Table 3. Such operating parameters do not whole set of runs. Only the most significant results are

© 2013 ISIJ 218

 ISIJ International, Vol. 53 (2013), No. 2

Fig. 11. Influence of the spot centre location on the “concavity” (α =

0.05, Rspot/Relec = 0.2, f = 2 rpm).

Fig. 10. Example of electrode temperature map, showing the influ-

ence of arc-spot motion.

reported below.
As we can see on Fig. 11, the radial location of the arc-
spot has a pronounced effect on the computed shape of the
electrode tip.
The influence of the radial location of the arc-spot on the
shape of the electrode tip can be explained by the fact that
the time during which a given point is heated by the moving
Fig. 12. Influence of α on the “concavity” (dist/Relec = 0.7, Rspot/
spot decreases proportionally to the inverse of its radial Relec = 0.2, f = 2 rpm).
location. Thus for a given angular speed, the farther a point
from the centre, the lower the time during which it will be
heated by the spot and by the way the lower the deformation
of the electrode tip. Indeed, only a rotating spot located
close to the edge will not induce a non-planar and very
irregular shape. Therefore, the value dist = 0.7 Relec was cho-
sen to study the influence of α. Not surprisingly, Fig. 12
clearly indicates that a very low value (less than 5%) is
required if the tip is supposed to remain relatively flat.
If the arc-spot carries only a small fraction of the overall
arc energy, it is interesting to check if the measured lumi-
nosity fluctuations can be qualitatively confirmed by the
model. To this end, we have defined a numerical “luminos-
ity” in a very crude way, as:
φbot (i )
L=∑ ............................. (5)
i d (i, O )

where O is a fixed location on the cathode bottom edge (r = Fig. 13. Computed fluctuation of “luminosity” (α = 0.05, dist/Relec =
0.7, f = 2 rpm).
Relec, θ = 0) vertically facing a virtual video camera. Figure
13 gives an example of the variation of this computed ‘lumi-
nosity’ with time. the arc-spot location approaches O) are shorter than low-
Any change in Rspot has no influence, as the three curves luminosity ones. The comparison of Figs. 7 and 13 indicates
are almost identical. The periodic (30 s) variation is caused that the motion of an arc-spot carrying only a small percent-
by the arc-spot rotation. Such a qualitative result is system- age of the arc power explains qualitatively the experimental
atically obtained, as soon as α differs from 0. In all cases, observation.
it is interesting to notice that high-luminosity periods (when

219 © 2013 ISIJ

 ISIJ International, Vol. 53 (2013), No. 2

appearance at the through side viewing ports, as used in

5. Conclusions
some previous studies9,10) could help to provide more con-
An extensive study of the behaviour of the electric arc in clusive data.
the VAR process has been undertaken at both micro- and
macroscale. At the microscale, the arc appears as a collec- REFERENCES
tion of spot clusters moving at the cathode tip. Its behaviour 1) T. Quatravaux, S. Ryberon, S. Hans, A. Jardy, B. Lusson, P. E. Richy
and D. Ablitzer: J. Mater. Sci., 39 (2004), 7183.
corresponds to the diffuse operating mode. At the macros- 2) A. Jardy and D. Ablitzer: Mater. Sci. Technol., 25 (2009), 163.
cale, observation of the arc luminosity fluctuations confirms 3) L. Falk, A. Jardy, D. Ablitzer and P. Paillere: Rev. Métall., 87 (1990),
experimental findings by Ward et al.11) 209.
4) A. Jardy and D. Ablitzer: Rev. Métall., 87 (1990), 421.
The development of a 3D model of the heat transfer in the 5) A. Jardy and D. Ablitzer: Rare Metal Mat. Eng., 35 (2006), 119.
consumable electrode and associated melting process 6) P. Chapelle, A. Jardy, J. P. Bellot and M. Minvielle: J. Mater. Sci.,
43 (2008), 5734.
enables us to determine some effects of the electric arc 7) F. J. Zanner: Metall. Trans. B, 10B (1979), 133.
behaviour. In industrial practice, the cathode tip remains rel- 8) R. L. Williamson and F. J. Zanner: Proc. of 1991 Vacuum Metallurgy
Conf. on Melting and Processing of Speciality Materials, AVS, New
atively flat during melting. Yet, an arc which consists in a York, NY, (1992), 87.
uniform part and an arc-spot rotating around the axis of the 9) P. Chapelle, J. P. Bellot, A. Jardy, T. Czerwiec, X. Robbe, B.
electrode with a constant small radial position results in a Champin and D. Ablitzer: High Temp. Mater. Process, 4 (2000), 493.
10) P. Chapelle, T. Czerwiec, J. P. Bellot, A. Jardy, D. Lasalmonie, J.
concentration of part of the melting power in a given area. Senevat and D. Ablitzer: Plasma Source Sci. Tech., 11 (2002), 301.
This would lead to a preferential melting of this part of the 11) R. M. Ward and M. H. Jacobs: J. Mater. Sci., 39 (2004), 7135.
12) R. M. Ward, B. Daniel and R. J. Siddall: Proc. of Int. Symp. on
electrode, hence a non-flat electrode tip, in contradiction Liquid Metal Processing and Casting, ASM International, Materials
with the observation, unless the fraction of total power car- Park, OH, (2005), 49.
ried by the arc-spot remains very low (only a few%). In the 13) L. Yuan, G. Djambazov, P. D. Lee and K. Pericleous: Int. J. Mod.
Phys. B, 23 (2009), 1584.
latter case, both luminosity periodic fluctuations and elec- 14) V. Bojarevics, K. Pericleous, G. Djambazov and R. M. Ward: Proc. of
trode flatness can be obtained. Int. Symp. on Liquid Metal Processing and Casting, TMS, Warrendale,
PA, (2009), 21.
Other possible ensemble arc motions can be investigated 15) R. L. Boxman, P. J. Martin and D. M. Sanders: Handbook of Vacuum
using the model. Apart from the rotational motion discussed Arc Science and Technology, Noyes Publications, Park Ridge, NJ,
above, a spiral-like motion (#9 in Table 4) results in an elec- (1995), 257.
16) L. A. Bertram and F. J. Zanner: Proc. of Int. Conf. Modeling and
trode tip shape that remains flat enough, while still compat- Control of Casting and Welding Processes, TMS, Warrendale, PA,
ible with the periodic behaviour of the luminosity. (1986), 95.
17) L. Falk, A. Jardy and D. Ablitzer: Ironmaking Steelmaking, 19
In all cases, the reasons for such a macroscopic motion (1992), 226.
of the arc are to be understood yet. The most difficult to 18) V. de Felice, A. Jardy and H. Combeau: Proc. of Int. Symp. on Liquid
Metal Processing and Casting, TMS, Warrendale, PA, (2009), 97.
interpret and striking feature remains the rather large time 19) S. V. Patankar: Numerical Heat Transfer and Fluid Flow, Hemisphere
constant of this motion which is much greater than the time Pub. Corp., New York, (1980), 58.
scales characteristic of the behaviour of individual cathode 20) H. El Mir: PhD Thesis, Institut National Polytechnique de Lorraine,
Nancy, France, (2005), 108.
spots (in the range of 1 μ s to 1 ms) and does not correspond 21) H. El Mir, A. Jardy, J. P. Bellot, P. Chapelle, D. Lasalmonie and J.
either to any obvious time scales of the VAR process. Fur- Senevat: J. Mater. Process. Technol., 210 (2010), 564.
22) J. P. Bellot, P. Zabeti, H. El Mir, J. Jourdan, P. Chapelle and A. Jardy:
ther experiments coupling the optical diagnostics used in the Proc. of Int. Symp. on Liquid Metal Processing and Casting, TMS,
present study with direct visual observations of the arc Warrendale, PA, (2009), 3.

© 2013 ISIJ 220