Professional Documents
Culture Documents
Park 2017 J. Phys. D Appl. Phys. 50 43LT03
Park 2017 J. Phys. D Appl. Phys. 50 43LT03
Letter
Abstract
A computational model of the mixing of multiple metal vapours, formed by vaporization of
the surface of an alloy workpiece, into the thermal arc plasma in gas tungsten arc welding
(GTAW) is presented. The model incorporates the combined diffusion coefficient method
extended to allow treatment of three gases, and is applied to treat the transport of both
chromium and iron vapour in the helium arc plasma. In contrast to previous models of
GTAW, which predict that metal vapours are swept away to the edge of the arc by the plasma
flow, it is found that the metal vapours penetrate strongly into the arc plasma, reaching the
cathode region. The predicted results are consistent with published measurements of the
intensity of atomic line radiation from the metal vapours. The concentration of chromium
vapour is predicted to be higher than that of iron vapour due to its larger vaporization rate.
An accumulation of chromium vapour is predicted to occur on the cathode at about 1.5 mm
from the cathode tip, in agreement with published measurements. The arc temperature is
predicted to be strongly reduced due to the strong radiative emission from the metal vapours.
The driving forces causing the diffusion of metal vapours into the helium arc are examined,
and it is found that diffusion due to the applied electric field (cataphoresis) is dominant. This
is explained in terms of large ionization energies and the small mass of helium compared to
those of the metal vapours.
Keywords: gas tungsten arc welding, computational modelling, multiple metal vapours,
combined diffusion coefficient method, cataphoresis
2
J. Phys. D: Appl. Phys. 50 (2017) 43LT03
Table 1. Composition of SUS 304 used in calculation, and iron and chromium were obtained using the composition of
properties of the component elements. SUS 304 given in [19], and the boiling temperatures and latent
Fe Cr heats of vaporization given in [21]. These data are given in
table 1. The presence of minority elements in SUS 304, in
Composition (mol %) 79.0 21.0 particular nickel and manganese, is neglected.
Boiling temperature (K) 3135 2755
The model assumes the plasma is in local thermodynamic
Enthalpy of vaporization (kJ mol−1) 349.6 344.3
equilibrium (LTE), and thermodynamic and transport proper-
ties were calculated using the methods presented by Murphy
The steady-state model solves the equations of mass conserva- [2]. The net radiative emission coefficients of iron vapour
tion, momentum conservation, energy conservation, and cur were taken from Menart and Malik [22], and those of helium
rent continuity. The transport of metal vapours by diffusion is were obtained from the work of Cressault et al [23], and those
treated using the combined diffusion coefficient method [11, of chromium vapour were obtained using the method of Cram
13], including the recent extension to three gases by Zhang [24]; a 1 mm absorption length was used, and the values for
et al [18], so that mixing of both iron and chromium vapours the gas mixtures were calculated using a mole-fraction aver-
with helium shielding gas can be considered. Two additional age, as recommended by Gleizes et al [25].
equations for the conservation of mass fraction of the iron and The heat transfer between the arc plasma and the tung-
chromium vapours are required: sten cathode and anode workpiece was calculated, taking
into account thermionic emission from the cathode, using the
∇ · ρvYFe = −∇ · JFe + SFe (1) method of Tanaka et al [20]. It was assumed that the weld-
pool surface remained flat, but flow in the weld pool was cal-
∇ · ρvYCr = −∇ · JCr + SCr (2) culated. A negative value of ∂γ/∂T , where γ is the surface
tension, was chosen, as is appropriate for low-sulfur stainless
where ρ is the mass density, v is the mass-average velocity, YI steel, in treating the surface tension gradient (Marangoni)
is the mass fraction, JI is the mass flux relative to the mass- force [10]. The thermophysical properties of tungsten were
average velocity and SI is the source term related to vaporiza- taken from [20], and those of stainless steel (SUS 304) proper-
tion of metal at the surface of the workpiece; subscripts Fe ties were obtained from [26, 27].
and Cr represent all species (neutral and singly- and multiply- For simplification of the sheath region, the mesh size for
ionized atoms) in the iron and chromium vapour respectively. the anode region is chosen as 0.1 mm and the mesh size for
The mass fraction of helium is determined after solving the the cathode region is chosen as 0.05 mm [28, 29]. The temper
two equations for the mass fraction for iron and chromium ature of all external boundaries is set to 300 K, including the
vapour using YHe = 1 − YFe + YCr . The mass flux of iron or workpiece due to the water cooling. The inflow rate of helium
chromium vapour is calculated using the combined diffusion shielding gas is 30 l min−1. Also, the metal vapours are not
coefficient method, and is given by allowed to enter through the computational domain. The max-
imum surface temperature of the workpiece is set to 2500 K;
n2
this is slightly lower than the results of a time-dependent
JFe = mFe mCr mHe DxFeCr ∇xCr + DxFeHe ∇xHe
ρ model [30], but was chosen to give comparable metal vapour
+ DPFe ∇ ln P + DEFe E − DTFe ∇ ln T (3) concentrations between the steady-state model and the meas-
urements [19] with which the results are compared.
n2
3. Results and discussion
JCr = mFe mCr mHe DxCrFe ∇xFe + DxCrHe ∇xHe
ρ
+ DPCr ∇ ln P + DECr E − DTCr ∇ ln T (4) We compare the predictions of our computational model with
experimental results presented by Tanaka and Tsujimura [19].
Both calculations and measurements used an arc current of
where n is the number density, mI is the average masses of
150 A, a lanthanated tungsten cathode of diameter 3.2 mm
the heavy species, xI is the sum of the mole fractions of the
with 60° angle tip, a SUS 304 stainless steel workpiece and
species, P is the arc pressure, E is the applied electric field,
an initial arc length of 3 mm. Note that in the experiment
T is the arc temperature, DxIJ is the combined ordinary dif-
there is some thermal expansion of the electrodes leading to a
fusion coefficient, describing diffusion due to mole fraction
decrease in arc length; this is neglected in the model.
gradients, DPI is the combined pressure diffusion coefficient, Figure 2 shows images of Cr I, Fe I and He I line radiation
describing diffusion due to pressure gradients, DEI is the com- 15 s after ignition of the arc, and predicted mole fraction dis-
bined electric field diffusion coefficient, describing diffusion tributions of chromium and iron vapours and helium shielding
due to externally-applied electric fields and DTI is the com- gas. In interpreting the images of the line radiation, it should
bined temperature diffusion coefficient, describing diffusion be noted that (a) the lines that are measured (Cr I: 520.8 nm,
due to temperature gradients. The source terms in (1) and Fe I: 538.3 nm; He I: 587.6 nm) are of different strengths, with
(2) are calculated using the Hertz–Knudsen–Langmuir equa- the Fe I line weaker than the other two, (b) the metal vapours
tion to take into account evaporation and condensation of iron ionize at much lower temperatures than helium, so that the
and chromium vapours [5]. The required vapour pressures of radiant intensities of the metal lines are peaked at much lower
3
J. Phys. D: Appl. Phys. 50 (2017) 43LT03
4
J. Phys. D: Appl. Phys. 50 (2017) 43LT03
The effect of the metal vapour has previously been consid- References
ered to be relatively weak in GTAW due to low vaporization
rate, unlike the dominant effect of metal vapour on the arc [1] Murphy A B 2015 A perspective on arc welding research:
in GMAW [2]. In the case of the argon shielding gas, little the importance of the arc, unresolved questions and future
metal vapour is formed (less than 1 mol%) from the surface of directions Plasma Chem. Plasma Process. 35 471–89
the workpiece. Metal vapour has an influence near the anode [2] Murphy A B 2010 The effects of metal vapour in arc welding
J. Phys. D: Appl. Phys. 43 434001
surface, but experiments and modelling both show that the
[3] Schnick M, Füssel U, Hertel M, Spille-Kohoff A and
upper region of arc plasma is not affected by metal vapour Murphy A B 2010 Metal vapour causes a central minimum
[35–37]. However, for helium shielding gas, the vaporization in arc temperature in gas–metal arc welding through
of the workpiece is increased, since the arc is constricted due increased radiative emission J. Phys. D: Appl. Phys.
to helium’s high specific heat and low electrical conductiv- 43 22001
ity at low temperature [38]. Further, cataphoresis is much [4] Schnick M, Fuessel U, Hertel M, Haessler M, Spille-Kohoff A
and Murphy A B 2010 Modelling of gas–metal arc welding
stronger for helium arcs, since the differences in ionization taking into account metal vapour J. Phys. D: Appl. Phys.
energy and mass between helium and metals is much larger 43 434008
than those between argon and metals. Therefore, we expect [5] Hertel M, Spille-Kohoff A, Fussel U and Schnick M 2013
that, even taking into account cataphoresis, the influence of Numerical simulation of droplet detachment in pulsed
metal vapour on argon GTAW arcs will remain small. gas–metal arc welding including the influence of metal
vapour J. Phys. D: Appl. Phys. 46 224003
[6] Murphy A B 2013 Influence of metal vapour on arc
temperatures in gas–metal arc welding: convection versus
4. Conclusions radiation J. Phys. D: Appl. Phys. 46 224004
[7] Zhao G Y, Dassanayake M and Etemadi K 1990 Numerical
We have used the combined diffusion coefficient method, simulation of a free-burning argon arc with copper
evaporation from the anode Plasma Chem. Plasma Process.
extended to a mixture of three gases, to treat diffusion of iron
10 87–98
and chromium vapours in helium shielding gas in GTAW of [8] Gonzalez J J, Gleizes A, Proulx P and Boulos M 1993
stainless steel. In contrast with the predictions of previous mod- Mathematical modeling of a free-burning arc in the
els, we show that the metal vapours produced from the anode presence of metal vapor J. Appl. Phys. 74 3065–70
workpiece are not fully swept away radially, but diffuse through [9] Lago F, Gonzalez J J, Freton P and Gleizes A 2004 A
the arc column to the surface of the cathode. This is consis- numerical modelling of an electric arc and its interaction
with the anode: part I. The two-dimensional model J. Phys.
tent with published emission spectroscopy measurements. D: Appl. Phys. 37 883–97
Further, differences in the predicted distributions of chromium [10] Tanaka M, Yamamoto K, Tashiro S, Nakata K, Yamamoto E,
and iron vapours explain the observed presence of chromium Yamazaki K, Suzuki K, Murphy A B and Lowke J J 2010
and the absence of iron on the tungsten cathode. The predicted Time-dependent calculations of molten pool formation
arc temperature is much lower than that obtained in previous and thermal plasma with metal vapour in gas tungsten arc
welding J. Phys. D: Appl. Phys. 43 434009
models, due to strong radiation from the chromium and iron [11] Murphy A B 1996 A comparison of treatments of diffusion in
vapours in the arc column. Cataphoresis, or diffusion due to the thermal plasmas J. Phys. D: Appl. Phys. 29 1922–32
applied electric field, is found to be the main mechanism for [12] Murphy A 1993 Diffusion in equilibrium mixtures of ionized
penetration of the metal vapours into the arc column; this has gases Phys. Rev. E 48 3594–603
been neglected in previous models of GTAW. Cataphoresis is [13] Murphy A B 2014 Calculation and application of combined
diffusion coefficients in thermal plasmas Sci. Rep. 4 4304
expected to be less important for argon shielding gas, since it is [14] Yamamoto K, Tanaka M, Tashiro S, Nakata K, Yamamoto E,
strongest when there are large differences in the ionization ener- Yamazaki K, Suzuki K, Murphy A B and Lowke J J 2009
gies and masses of the gases. The results obtained shows that Numerical simulation of diffusion of multiple metal
the extended combined diffusion coefficient method has the vapours in a TIG arc plasma for welding of stainless steel
capability to capture the mixing and demixing of three gases, Weld. World 53 R166–70
[15] Park H, Mudra M, Trautmann M and Murphy A B 2017 A
such as two different metal vapours in an arc plasma. coupled chemical kinetic and nucleation model of fume
formation in metal–inert-gas/metal–active-gas welding
Plasma Chem. Plasma Process. 37 805–23
Acknowledgments [16] Ogino Y, Hirata Y and Murphy A B 2016 Numerical
simulation of GMAW process using Ar and an Ar–CO2 gas
One of the authors (ABM) gratefully acknowledges support mixture Weld. World 60 345–53
from the JWRI International Joint Research Collaborator [17] Murphy A B 1997 Demixing in free-burning arcs Phys. Rev. E
program at the Joint Usage/Research Center on Joining and 55 7473–94
Welding, Osaka University. The authors also thank Drs John [18] Zhang X N, Murphy A B, Li H P and Xia W D 2014
Combined diffusion coefficients for a mixture of three
J Lowke and Eugene Tam of CSIRO for helpful discussions. ionized gases Plasma Sources Sci. Technol. 23 65044
[19] Tanaka M and Tsujimura Y 2012 Visualization of metal
vapor behavior in TIG welding Quart. J. Jpn. Weld. Soc.
ORCID iDs 30 164–70
[20] Tanaka M, Terasaki H, Ushio M and Lowke J J 2002 A
Hunkwan Park https://orcid.org/0000-0002-9824-6089 unified numerical modeling of stationary tungsten-inert-gas
Anthony B Murphy https://orcid.org/0000-0002-2820-2304 welding process Metall. Mater. Trans. A 33 2043–52
6
J. Phys. D: Appl. Phys. 50 (2017) 43LT03
[21] Zhang Y, Evans J R G and Yang S 2011 Corrected values for vapour behaviour in thermal plasma of gas tungsten arcs
boiling points and enthalpies of vaporization of elements in during welding Sci. Technol. Weld. Join. 13 566–72
handbooks J. Chem. Eng. Data 56 328–37 [31] Yamamoto K, Tanaka M, Tashiro S, Nakata K and
[22] Menart J and Malik S 2002 Net emission coefficients for argon- Murphy A B 2009 Metal vapor behavior in GTA welding
iron thermal plasmas J. Phys. D: Appl. Phys. 35 867–74 of a stainless steel considering the marangoni effect IEEJ
[23] Cressault Y, Rouffet M E, Gleizes A and Meillot E 2010 Net Trans. Electr. Electron. Eng. 4 497–503
emission of Ar–H2–He thermal plasmas at atmospheric [32] Tashiro S, Zeniya T, Yamamoto K, Tanaka M, Nakata K,
pressure J. Phys. D: Appl. Phys. 43 335204 Murphy A B, Yamamoto E, Yamazaki K and Suzuki K 2010
[24] Cram L E 1985 Statistical evaluation of radiative power losses Numerical analysis of fume formation mechanism in arc
from thermal plasmas due to spectral lines J. Phys. D: Appl. welding J. Phys. D: Appl. Phys. 43 434012
Phys. 18 401–11 [33] Mougenot J, Gonzalez J J, Freton P and Cressault Y 2013
[25] Gleizes A, Cressault Y and Teulet P 2010 Mixing rules for Argon and Arcal.37 plasma characteristics in a TIG
thermal plasma properties in mixtures of argon, air and configuration J. Phys. D: Appl. Phys. 46 495203
metallic vapours Plasma Sources Sci. Technol. 19 55013 [34] Murphy A B 1998 Cataphoresis in electric arcs J. Phys. D:
[26] Traidia A, Roger F and Guyot E 2010 Optimal parameters Appl. Phys. 31 3383–90
for pulsed gas tungsten arc welding in partially and fully [35] Etemadi K and Pfender E 1985 Impact of anode evaporation
penetrated weld pools Int. J. Therm. Sci. 49 1197–208 on the anode region of a high-intensity argon arc Plasma
[27] Kim W-H and Na S-J 1998 Heat and fluid flow in pulsed Chem. Plasma Process. 5 175–82
current GTA weld pool Int. J. Heat Mass Transf. [36] Gonzalez J J, Bouaziz M, Razafinimanana M and Gleizes A
41 3213–27 1997 The influence of iron vapour on an argon transferred
[28] Lowke J J and Tanaka M 2006 ‘LTE-diffusion approximation’ arc Plasma Sources Sci. Technol. 6 20–8
for arc calculations J. Phys. D: Appl. Phys. 39 3634–43 [37] Mougenot J, Gonzalez J-J, Freton P and Masquère M 2013
[29] Lohse M, Siewert E, Hertel M, Füssel U and Rose S 2015 Plasma–weld pool interaction in tungsten inert-gas
Modelling of the cathode sheath region in TIG welding configuration J. Phys. D: Appl. Phys. 46 135206
Weld. World 59 705–11 [38] Tanaka M and Lowke J J 2007 Predictions of weld pool
[30] Yamamoto K, Tanaka M, Tashiro S, Nakata K, Yamazaki K, profiles using plasma physics J. Phys. D: Appl. Phys.
Yamamoto E, Suzuki K and Murphy A B 2008 Metal 40 R1–23