ANALYSIS OF ARC LIGHT MECHANISM AND ITS
APPLICATION IN SENSING OF GTAW PROCESS
P. J. Li and Y. M. Zhang
Welding Research and Development Laboratory
Center for Robotics and Manufacturing Systems and
Electrical Engineering Department
College of Engineering
University of Kentucky
Lexington, KY 40506
ABSTRACT
Sensing plays a key role in automating and controlling welding processes. Although arc
light sensing has advantages over through-the-arc sensing for arc length control and seam
tracking, the current technology relies on experimental data and lacks theoretical foundation. To
improve measurement accuracy, this work addresses the theoretical foundation for arc light
sensing. A theoretical model has been developed to correlate arc light radiation to welding
parameters. Distributions of different radiant sources in the arc column are studied. It is found
that the distributions of the ions of the shielding gas and the vapors of the base metal and
tungsten are not even, while that of the shielding gas atoms is. This suggests that the spectral
lines associated with the shielding gas atoms can be used to improve the accuracy of arc light
sensing. Hence, an arc light sensor has been developed to detect argon atom spectral lines in gas
tungsten arc welding. The relationship between the argon lines and the welding parameters has
been derived from the theoretical model. Seam tracking examples showed the effectiveness of
the developed method in improving the accuracy.
Key Words: arc light, sensing, spectral analysis, GTAW.
1. INTRODUCTION
Arc light radiation is a fundamental phenomenon associated with the arc welding process.
Early studies were primarily dedicated to investigating its hazardous effects [1-6]. Topics
included harm to human beings, correlation between welding process and arc light intensity, and
relationship between hazardous effects and the spectral bands of the arc light. Commercial
welding curtains and shades resulted from these studies [7].
Arc light is closely related to the plasma column of the arc. By means of spectral analysis
of arc light, temperature distribution and conductivity of the arc column [8-16], temperature and
emissivity of tungsten [17-20], droplet temperature [21,22], hydrogen density in the arc column
[23, 24], and temperature of the weld pool surface [25-28] were measured and analyzed.
Spectral analysis method has also been used to study the influence of minor elements on
weldability [29-31].
Early utilization of arc light in process sensing was based on its optical property. In the
work by Faber and Jackel, two symmetrically arranged light sensitive components were used to
achieve seam tracking in fillet welding by compensating their difference in the incident light
intensity [32]. Nomura used four CdS components implanted in a water-cooled copper plate
placed at the backside of the weld pass to monitor the keyhole [33, 34]. In the research by Salter,
vibration frequency of the weld pool in gas tungsten arc welding (GTAW) was monitored by
sensing the arc light reflected by the mirror-like surface of the weld pool [35-37].
A few further studies investigated the possibility of sensing welding behaviors such as arc
length variation and droplet transfer based on arc light. In particular, an U.S. patent [38]
introduced a method for arc length sensing in which, instead of using the arc voltage signal, the
integral intensity of arc light in the ultraviolet band was monitored to improve the accuracy. The
transfer mode of the droplets in GMAW was monitored and analyzed using a compact arc light
sensor with automatic sensitivity control [39-40].
Compared with prevalent sensing methods, such as arc electric signal sensing and arc sound
sensing [41-44], arc light sensing has obvious advantages including reliability and accuracy in
sensing arc behaviors. However, most work has been focused on detecting minor or harmful
elements in the arc. Efforts toward revealing the mechanism of arc light sensing, such as
investigating the sensitivity and stability of arc light in comparison with arc voltage sensing in
arc length measurement by Richardson and Edwards [45], are limited.
Arc light radiation is a complicated phenomenon involving several welding parameters.
Without understanding the mechanism, its utilization is based on merely subjective experience
and assumptions. To achieve high accuracy and reliability, the mechanism must be known.
Hence, this paper is dedicated to establishing an arc light radiation model in order to correlate the
arc light with the welding parameters. The dominant elements and their roles in determining the
radiation are analyzed. The most sensitive spectra of the arc light, with respect to different
welding behaviors, were analyzed and separated to improve the signal quality.
2. MODEL OF ARC LIGHT SENSING
During arc welding, the shielding gas in the plasma column can be assumed to consist of
three kinds of particles: electrons, ions, and atoms. Approximately, the plasma column can be
assumed in local thermodynamic equilibrium (LTE) condition [46,47], in which electron
collision plays an important role in excitation and ionization. The emission of arc light from a
plasma is characterized by the emission coefficient ε v , which is defined as the radiant energy
which is emitted by a unit volume of the radiating gas per unit of time, per unit of solid angle,
and per unit of frequency or wavelength. It can be expressed by the Kirchhoff’s Law [48, 49]:
2 ⋅ h ⋅ν 3
εν = k ' (v ) ⋅
2
c ⋅ (e
h⋅ν
k ⋅Te
(1)
− 1)
where Bv is radiation power cross the area of a unit solid angle in a unit frequency band,
k ' ( v ) is a frequency-dependent absorption coefficient,
c is velocity of light (2.9979×108 m⋅s-1),
Te is electron temperature in K,
ν is spectral frequency in Hz,
2
-23
-1
k is Boltzmann’s constant (1.3807×10 J⋅K ), and
h is Planck’s constant (6.6262×10-34 J⋅s).
It is known that the peak response wavelengths, 820 nm in this study, for commercial light
sensitive components fall into the infrared or ultraviolet band. Hence, the following equation,
the Rayleigh-Jeans approximation [48,49], can be used:
U ν = 8π
ν2
kTe
c3
(2)
Also, under the atmospheric pressure and normal welding current range, the temperature of
the electrons is very close to that of the arc [50]. Compared with the temperature of the arc (over
6,000 K), the difference is negligible. As a result,
v2
ε v = k ' (v ) ⋅ 2 2 kT
(2)
c
where T is the temperature of the arc in K. Assume that heat losses are all due to conduction
[49], then the Elenbass-Heller rule [46] holds for any unit component in the arc column:
1 rλdT
σ ⋅ E 2 + d(
) / dr = 0
r
dr
(3)
where E is voltage gradient of arc,
r is distance of the unit component to axis of arc column, m,
λ is thermal conductivity, W⋅m-1⋅K-1, and
σ is electric conductivity.
R
r
L
Fig. 1 A model of luminary of welding arc plasma
3
In the welding arc column, electric conductivity will decrease to zero abruptly beyond a
certain radius R [51]. It can be assumed that welding current passes evenly through the arc
plasma within a column of R in radius, as shown in Fig.1. Then the density of current jr is:
j r = I / ( πR 2 )
(4)
According to the differential form of Ohm’s law:
jr = σ ⋅ E
(5)
Within the radius R , by substituting jr into Equation (3), the following can be obtained:
1 rλdT
jr E = − d (
) / dr
r
dr
(6)
Assume λ is constant. Denote the temperature at r = R by T0 . Integration of Equation (6)
produces:
T=
IE
r2
(1 − 2 ) + T0
R
4πλ
(7)
By substituting T in Equation (2), one can obtain the spectral intensity of the arc light in the
column defined by r ≤ R , denoted by ε cv , which conducts welding current:
ε cν
r2
ν 2 IE
(1 − 2 ) + T0 )
= k ' ( v ) ⋅ 2k 2 ⋅ (
c
R
4πλ
(8)
When r > R , σ = 0 , Equation (3) becomes [51]:
IE
dr
2 πr
(9)
IE
r
ln
2πλ R
(10)
λdT = −
Its integration gives
T = T0 −
By substituting T in Equation (3), the spectral intensity of the plasma ( ε ncv ) in the region
defined by r > R which does not conduct the current can be obtained:
ε ncν = k ' (v ) ⋅ 2k
IE
r
ν2
(T0 −
ln )
2
c
2πλ R
(11)
The arc column can be assumed to be an optical lamina which does not absorb radiant
energy. The gradient of temperature along the axis of arc can be neglected [51]. By integrating
ε v over volume of arc column, one can obtain the radiant energy of the whole arc column in a
unit spectral band:
4
Biv = ∫∫∫ ε v dv
R
RL
0
R
= ∫ 2πr ⋅ L ⋅ ε cv dr + ∫ 2πr ⋅ L ⋅ ε ncv dr
R
RL
0
R
(12)
= L ∫ 2πr ⋅ ε cv dr + ∫ 2πr ⋅ ε ncv dr
where RL is the radius of arc plasma luminary and the spectral intensity of arc plasma column is
zero when r ≥ RL . From Equation (11), the upper limit of the integration, RL , can be calculated:
RL = R ⋅ e
2πλT0
IE
(13)
Hence,
kν 2
IE
Biv = k ' (v ) ⋅ 2 ⋅ L ⋅ R 2 ⋅
⋅ (e
c
2λ
4πλT0
IE
1
− )
2
(14)
Equation (14) accounts for the contribution of the atoms of the shielding gas to the arc
radiation. In addition to the shielding gas, other radiant sources such as the weld pool and the
electrode also contribute to the observed radiation of the arc. Hence, two additional terms, I 2
and constant, are added in Equation (14) to account for the contributions of heat input related
sources and other sources. As a result, a modified equation is achieved:
kν 2
I IE
Biv = k ' ( v ) ⋅ 2 ⋅ L ⋅
⋅
⋅ (e
π ⋅ j 2λ
c
4πλT0
IE
1
− ) + C1 ⋅ I 2 + C 2
2
(15)
where j is the average current density, and C1 and C 2 are constants. In gas tungsten arc
welding process, the current density remains constant in low welding current range and increases
slowly with the welding current in high current range [51]:
j = jc
I ≤ I*
(16)
j = jc + a I −150
I > I*
*
where a is a small positive constant, and I ranges from 120A to 150 A.
Equation (15) depicts the relationship between the radiation of the arc light and the welding
parameters including the current, the voltage gradient of the arc column, and the arc length. This
comprehensive model can play a fundamental role in understanding the phenomena associated
with arc radiation and arc light, as well as in applying the arc light sensing method to welding
process sensing and control. In the following discussion, we shall first verify the effectiveness of
the model through arc length sensing, and then improve the accuracy in arc length control and
seam tracking based on the model.
3. EXPERIMENTAL PROCEDURE
Experiments have been conducted toward three goals. The first one is to verify the arc light
sensing model based on arc length measurement. The second has been directed to spectral
analysis and accuracy improvement. Finally, arc light sensing method has been applied to seam
tracking.
5
Water-cooled copper plate, mild steel, low alloy steel A202M-93, and stainless steel 304
have been used in the experiments. Argon was used as shielding gas. The welding power source
was an inverter welding power source with output from 5A to 500A.
Torch
Arc Light
Sensor
Light
Water-cooled
Probe
(Diaphragm)
Amplifer
Filter
Monochrometer
A/D
Isolator
Host
Computer
Wavelength signal
A/D
PT
Amplifer
Current
Sensor
Amplifer
Filter
Microprocessor
A/D
Filter
A/D
Isolator
PT: Photomultiplier Tube
(a) Diagram
1
3
1
4
2
3
1-Torch
2-Arc
3-Specimen
5
2
4
4-Monochrometer
5-Light path (Diaphragm)
(b) Light path diagram
Fig. 2 Experimental set-up
The developed arc light sensor, 15 mm in diameter and 50 mm in length, was clamped on
the welding torch. The sensor works properly under practical conditions with protective design
and automatic sensitivity control. Its spectral response range is from 500 nm to 1000 nm. A
grating monochrometer, with 0.1 nm resolution and 67-600 nm/min scanning speed, was used in
our spectral analysis. The minimum view angle of the water-cooled probe (diaphragm) and the
spectrometer is 0.015 degree and the spectral response range of the system is 320 to 820 nm. An
automatically controlled moving weld table was employed so that the welding torch and
therefore the light path could remain stationary. The diaphragm was constructed on a threedimensional translation stage with 0.01 mm adjustments for precise positioning. A
microprocessor received the arc light signal, welding current signal, the spectral signal and
wavelength signal. It also controlled the welding equipment. Data was calibrated and then
transferred to a personal computer for further analysis. Fig. 2 shows the overall layout of the
experimental equipment. In the experiments the arc light from the integral arc column or part of
6
it was focused into the monochrometer for analysis. In partial arc spectral analysis, arc light
from a very thin layer or a very small part of arc column was monitored by diaphragm without
peripheral effects. The data was converted by Abel transformation with Pearce method [46]. The
purpose of this experiment was to determine the spatial distribution of element emission in the
arc column.
4. RESULTS
4.1. Arc Length Sensing
As a primary arc welding process for precise joining of metals, GTAW requires precise
control of its parameters. Arc length determines the distribution of the heat input and thus
requires control as precise as 0.2 mm. Although methods such as probe, optical, arc voltage
sensing, etc. have been available, the requirement of 0.2 mm has been difficult to realize. As
will be demonstrated, arc light sensing will be a promising method to achieve such high
accuracy.
Gas tungsten arc is relatively stable. It is reasonable to assume the radius of the electric
conducting column and the radius of arc plasma luminary as constants. Under normal welding
conditions, the voltage gradient of the arc column is also a constant. Thus, Equation (15) can be
converted to:
Biv = G1 L ⋅ I 2 (e
G2
I
1
− ) + G3 ⋅ I 2 + G4
2
(17)
where G1 , G2 , G3 , G4 are constants, L is arc length, and I is welding current. As can be seen
in Fig. 3, Equation (17) reveals that the intensity of the arc light in a unit spectral band linearly
changes with the arc length provided that the welding current remains constant.
Simplified model of arc length and Biv
Arc Light Intensity, arb. unit
6
120 A
5
4
90 A
3
70 A
60 A
2
1
0
0
1
2
3
4
5
Arc Length, mm
Fig. 3 Theoretical relationship between intensity of arc light in a unit band versus arc length
7
In order to verify the theoretical model, several experiments have been done to measure the
arc length based on arc light. Fig. 4 shows that the arc light signal, integral intensity of the arc
light in the band from 500 nm to 1000 nm, has a linear relationship with the arc length. This
result was obtained on water-cooled copper workpiece. The results on stainless steel plates are
given in Fig. 5. A similar linear relationship is observed when the arc length is 1 mm or larger
between the arc length and the integral intensity of the arc light over the band from 500 nm to
1000 nm. The experimental data in Figs. 4 and 5 can be described by the theoretical model with
following parameters
5
1
2
Vsignal = 0.0001 ⋅ L ⋅ I ( e I − ) + 0.000075 ⋅ I 2 + 0.0001
(18)
2
where Vsignal is arc light signal in volts. The units for the welding current I and the arc length L
are ampere and millimeter, respectively.
Arc light intensity signal, V
Of course, the arc light signal did not show a linear relationship with the arc length for short
arc as predicted in Fig. 3 by the theoretical model. Instead, the arc light intensity increases or
remains the same when arc length decreases below 1 mm. Similar phenomena were also
observed for mild steel and alloy steel.
5 Water-cooled
copper anode
60 A
4
70 A
3
90 A
120 A
2
1
0
1
2
3
4
5
Arc length, mm
Fig. 4 Arc light intensity versus arc length: water-cooled copper anode
Arc light intensity signal, V
5
Stainless steel 304
60 A
70 A
90 A
120 A
4
3
2
1
0
0
1
2
3
4
5
Arc length, mm
Fig. 5 Arc light intensity versus arc length: stainless steel anode
8
4.2. Improvement of Arc Light Sensing based on Spectral Analysis
0
350
360
370
380
390
400
420
ArI 433.35 nm
A rI 434.5 nm
A rI 430.0 nm
A rI 425.9 nm
Mn426.59 nm, A rI 426.6 nm
ArI 415.8, 416.4 nm
ArI 418.18 nm
A rI 419.1 nm
FeI 419.9 nm, A rI 420.0 nm
410
430
FeI 438.3 nm
FeI 440.4 nm
M nI 441.18 nm
A rI 442.6 nm
100
A rI 404.4 nm
MnI 406.4 nm
200
ArI 383.46 nm
FeI 385.7 nm
FeI 387.2 nm
A rI 389.4 nm
300
A rI 360.6 nm
A rI 363.2 nm
ArI 364.98 nm
ArI 367.06 nm , ArI 367.52 nm
ArI 369.08 nm
FeI 371.9 nm
A rI 372.9 nm
FeI 3743 nm , ArI 3743.3 nm
FeI 376.0 nm
A rI 377.0 nm
FeI 379.5 nm
Mildsteel, 50A
Arc length: 1mm
ArII 354.5 nm
FeI 358.1 nm, ArII 358.8 nm
Intensity, arb. unit (calibrated)
400
FeI 392.7 nm, Ar392.8 nm
A rI 394.9 nm
The inaccuracy of the theoretical model in short arc range can be explained and overcome
based on spectral analysis. It is known that the arc light consists of two parts: continuum and
discrete spectral lines (Fig. 6). When the model was established, the intensity of the arc light in a
unit spectral band was considered. The element emitting the light in this unit band is assumed to
be homogeneous in the whole arc column. If all the elements in arc plasma column are
distributed evenly in the column, the model would probably accord to the arc light intensity.
However, as will be shown below, such an assumption is inaccurate.
440
450
Wavelength, nm
Fig. 6 Spectral distribution of gas tungsten arc light
Using the diaphragm, spatial distribution of emission of elements in arc column was
examined. Two-dimensional emission distribution “maps” of argon atoms, argon ions and metal
vapors were produced. As expected, argon atomic lines are distributed throughout the arc column
(Fig. 7), and account for a major portion of the radiation emitted by welding arc. This
phenomenon has been observed for different workpiece materials such as stainless steel, watercooled copper plate, etc. as well as for different welding currents and arc lengths. Of course, the
spatial distribution of the intensity of argon atomic spectra varies and the intensity in general
increases toward the axis of the arc and the tip of the tungsten electrode (Fig. 7).
Distance from workpiece, mm
3.0
2.5
2.0
20
Tungsten
35 20 15 10 1
Current: 100 A
Arc length: 2 mm
Stainless steel 304
1.5
1.0
0.5
0.0
0.0
0.5
1.0
1.5
2.0
2.5
3.0
Distance from center of arc, mm
Fig. 7 Spatial distribution of Ar atomic spectra
9
Different from argon atoms, argon ions concentrate in a very small area immediately below
the tungsten electrode, as can be seen in Fig. 8. This is attributable to the high temperature
required for ionization.
Distance from workpiece, mm
5
Current: 100 A
Arc length: 3 mm
Stainless steel 304
4
3
Tungsten
13
1
2
1
0
0.0
0.5
1.0
1.5
2.0
2.5
Distance from center of arc, mm
Fig. 8 Spatial distribution of Ar ionic spectra
Distance from workpiece, mm
Spectrum of base metal vapors is distributed in a cone, concentrated just above the weld
pool as shown in Figure 9. Its area and height have an intimate relationship with welding
parameters. When the current is low, the height is large but the emission is relatively low. As the
welding current increases, the height decreases and the emission of metal vapor increases. It is
interesting that the metal vapor spectra are subject to severe fluctuations with random frequency.
Hence, the spectral components due to metal vapors should be filtered out to improve the
accuracy of arc length measurement.
5.5
5.0
4.5
4.0
3.5
3.0
2.5
2.0
1.5
1.0
0.5
0.0
Current: 100 A
Arc length: 3 mm
Stainless steel 304
Tungsten
1
7
0.0
0.5
1.0
1.5
2.0
Distance from center of arc, mm
2.5
Fig. 9 Spatial distribution of iron atomic spectra
It can be seen that, at the upper part of the arc column, the argon atomic and ionic lines
dominate the spectrum of the arc light. The intensity of metal lines, including the lines of
vaporized tungsten and base metal, is extremely weak due to the distance from the base metal
and the very high melting point of tungsten. When the distance toward the weld pool is reduced,
the spectra of base metal vapor become stronger but are subject to more severe fluctuations. As
shown in Fig. 10, argon atoms, argon ions, and metal atoms are the major emission sources in the
10
arc. As pointed out right after Equation (14), the equation only accounts for the contribution of
the atoms of the shielding gas. Although we have added two terms to modify the model to
account for the contributions of radiation sources other than the shielding gas, accurate
description of the fluctuated metal vapors is still difficult.
Argon ion
Argon atom
Argon ion
Argon atom
Metal atom
Metal atom
(A)
(B)
Fig. 10 Elements and their distributions in gas tungsten arc column. A-Long arc. B-Short arc.
Stainless Steel 304
60 A
70 A
90 A
120 A
Argon Atomic Line Intensity, V
6
5
4
3
2
1
0
0
1
2
3
4
5
Arc Length, mm
Fig. 11 Argon atomic line intensity versus arc length
The argon atomic line is distributed throughout the arc column, independent of workpiece
material. The distribution of argon ionic lines and metal vapor lines is not even in the arc
column and prone to be influenced by other related factors, and thus can not be used. Argon
atomic line of 696.5 nm or 750 nm or 394.8 nm was selected according to the above principles.
The signal shown in Fig. 11 was obtained using argon atomic line 696.5 nm. Two advantages
were resulted. The first is that the desired linear relationship between arc length and arc light
remains very well when arc length changes from 5 mm to almost short circuit, without being
influenced by workpiece materials. The second is that the noise in arc light signal decreases
significantly, due to the exclusion of metal vapor spectra. Experimental results show that the
relationship between argon atomic line intensity and arc length can be expressed as:
11
5
1
Vsignal = 0.00013 ⋅ L ⋅ I 2 ( e I − ) + 0.000025 ⋅ I 2 + 0.3
(19)
2
It is known that the second term in the equation was introduced to account for the influence
of heat input. Comparison of Equations (18) and (19) reveals that the influence of the heat input
on the line intensity has been significantly reduced from its level associated with the integral
radiation intensity. Because the first term in the equations corresponds to the theoretical
derivation, it implies that the line intensity is more theoretically predictable.
Hence, the
following equation has been derived from Equation (19) to measure the arc length in our arc
length control system:
Bsignal − N 4 − N 3 ⋅ I 2
L=
(20)
N2
1
2
I
N1 I (e − )
2
where Bsignal is argon line intensity signal received by the arc light sensor, and N 1 , N 2 , N 3 , and
N 4 are constants. It is evident that, to measure and control the arc length based on Equation
(20), the welding current I needs to be sampled during welding process.
Capability of arc light sensing in arc length feedback control has been measured within
normal welding arc length range from 0.5 mm to 5 mm. The accuracy was ±0.1 mm in spot
welding and ±0.2 mm in traveling welding with a step-shaped workpiece.
4.3. Weld Seam Sensing
The proposed arc light sensing method can be used to measure weld joint and its profile for
seam tracking or adaptive welding. To this end, the torch is oscillated across the weld joint as
shown in Fig. 12. The weaving motion causes changes in the arc light sensed at the joint
sidewalls. The resultant variation in the arc light is directly proportional to the fluctuation in the
distance between the surface of the weldment and the tip of the welding electrode. Fluctuations
in the arc light are monitored and the feedback signals are sent to the controller, which centers
the oscillating torch in the joint. Both V-type and square grooves were used in experiments, as
shown in Fig.12.
Fig. 13 illustrates the principle for detecting the deviation of the torch from the central line
of the weld seam in a V-groove. The deviation of the torch, denoted as D , can be determined by
the following equation:
T − T1 L
D= 2
(21)
⋅
T1 + T2 2
where L is the amplitude of oscillation, T1 and T2 are the time intervals between the peak and
the valleys in the waveform of the sensor output signal, as shown in Fig. 13. It can be shown
that:
B p − Bi
Ti
(22)
i = 1,2
α i = arctan(
)
⋅L⋅
N2
T1 + T2
1
2
N1 ⋅ I ⋅ (e I − )
2
12
where α i (i = 1, 2) are the angles of the bevel, N 1 and N 2 are constants as defined in Equation
(20), and B p and Bi (i = 1, 2) are outputs of the arc light signal at the center of the seam and the
end points of the oscillation, respectively. Fig. 14 shows the signal when the deviation D , the
leg of the groove, and the included angle α are 1 mm, 3 mm, and 90 ! , respectively. The
difference between T1 and T2 which indicates the deviation is clearly identifiable.
Scan
Scan
Scan
V-groove
1
2
3
Square groove
Travel Direction
① Welding Torch ② Base Metal ③ Weld Bead
Fig. 12 Arc light signal for detecting deviation of the welding torch
L
B
BP
B1
D
Left end of
oscillation P1
α
α1
α2
B2
Right end of
oscillation P2
T1 T2
B1: arc light signal corresponding to P1 B2: arc light signal corresponding to P2
BP: arc light signal corresponding to center of the seam
t
Fig. 13 Monitoring of torch deviation based on arc light signal
13
5
With 1 mm deviation
Arc light signal, V
4
3
2
Groove leg: 3 mm
Groove angle: 90o
Current: 70 A
1
0
0
2
4
6
8
10
12
Time, s
Fig. 14 Arc light signal for detecting deviation of the welding torch
Argon atomic line intensity signal, V
To further improve the accuracy, argon atomic lines have been used. Fig. 15 depicts the
atomic line signal in detecting a small V-groove (1 mm). Based on the signal, the groove can be
determined. Experiments showed that with argon atomic line intensity signal, 0.2 mm deviation
can be reliably detected for V-type groove with 1 mm or larger groove leg. For square butt
joints, the argon atomic line signals can successfully detect square butt grooves when the
minimum root opening is 0.5 mm for GTAW (Fig. 16) or 0.2 for needle-like PAW (Fig. 17).
However, if integral arc light signal is used, the minimum root opening increased to 1 mm for
GTAW and 0.5 mm for needle-like PAW.
5
4
Line intensity
3
2
Groove leg: 1 mm
1
0
1.0
Groove angle: 90o
1.5
2.0
2.5
3.0
Time, s
Fig. 15 Line intensity signal in detecting small V-groove
14
Argon atom line intensity signal, V
5
4
Clearance, 0.2 mm
Current, 3 A
3
2
1
0
0.0
0.4
0.8
1.2
1.6
2.0
Time, s
Argon atomic line intensity signal, V
Fig. 16 Line intensity signal in detecting square groove in GTAW
5
4
Clearance: 0.5 mm
Current: 50 A
3
2
1
0
0.0
0.4
0.8
1.2
Time, s
1.6
2.0
Fig. 17 Line intensity signal in detecting square groove in needle-like PAW
5. CONCLUSIONS
This study has been focused on the theoretical foundation for arc light sensing and its
application in accuracy improvement. The following are the major conclusions drawn from this
study.
•
A theoretical model of arc light radiation has been developed based on classical radiation
theory and arc physics. This gives an insight into the relationship between welding
parameters and arc light intensity and provides a theoretical foundation for improvement of
the accuracy.
15
•
Distributions of different radiant sources have been examined. It was found that the
distributions of argon ions and base metal vapors are not even in the arc column, but that of
argon atoms is.
•
The utilization of the argon spectral lines can significantly improve the accuracy of arc light
sensing, especially in short arc length range because of the even distribution of the argon
atoms in the arc column.
•
The relationship between the argon spectral lines and the welding parameters has been
obtained based on the theoretical model. An arc light sensor has been developed to measure
the argon lines. Hence, the argon spectral lines can be detected by the developed arc light
sensor and can be used to calculate the arc length based on the relationship.
•
Seam tracking examples have shown the effectiveness of the theoretical results and the
developed technology on accuracy improvement in arc light sensing.
REFERENCES
[1]
[2]
[3]
[4]
[5]
[6]
[7]
[8]
[9]
[10]
[11]
[12]
[13]
[14]
S.W.Horstman, E.A.Emmett and T.E.Kreichelt, Field Study of Potential Ultraviolet
Exposure from Arc Welding, Welding Journal, 1976, Vol.23(5): pp121s-126s
H.J.Hubner, J.R.Und, The measurement of radiation from arc welding carried out by
means of various welding processes, Welding and Cutting, 1972, Vol.24(8): pp290-293
D.H.Sliney, B.C.Freasier, Evaluation of Optical Radiation Hazards, Applied Optics, 1973,
Vol.12(1): pp1-24
H.E.Pattee, L.B.Myers, R.M.Evans and R.E.Monroe, Effects of Arc Radiation and Heat
on Welders, Welding Journal, 1973, Vol.20(5): pp297-308
A.G.Mazel and V.M.Pumpurs, The Emission Properties of Welding Materials in Arc
Welding, Automatic Welding, 1983, December: pp36-37
C.B.Andersen, Light Emission in Thermal Cutting, Welding in the World, 1986,
Vol.24(7/8): pp172-176
D.H.Sliney, C.E.Moss, C.G.Miller and J.B.Stephens, Transparent Welding Curtains,
Welding Journal, 1982, Vol.61(3): pp17-24
G.N.Haddad and A.J.D.Farmer, Temperature Measurements in Gas Tungsten Arcs,
Welding Journal, 1985, Vol.64(12): pp339s-342s
S.S.Glickstein, Temperature Measurements in a Free Burning Arc, Welding Journal,
1976, Vol.55(8): pp222s-229s
C.B.Shaw, Diagnostic Studies of the GTAW Arc (Part 1-Observational Studies), Welding
Journal, 1975, Vol.54(2): pp33s-44s
C.B.Shaw, Diagnostic Studies of the GTAW Arc (Part 2-Mathematical Model), Welding
Journal, 1975, Vol.54(3): pp81s-94s
Y.L.Song, Some Phenomena and Features of the Component Distributions in Welding
Arc, China Welding, 1995, Vol.4(1): pp65-69
Y.L.Song, Some Features of Welding Arc Radiation, China Welding, 1996, Vol.5(2):
pp96-101
K.Landes, G.Seeger and W.Tiller, Evaluation of the Mass Flow Field and the Electric
Current Density Field in a Free Burning TIG Similar Arc, Welding and Cutting, 1981,
Vol.33(4): ppE14-E16
16
[15]
[16]
[17]
[18]
[19]
[20]
[21]
[22]
[23]
[24]
[25]
[26]
[27]
[28]
[29]
[30]
[31]
[32]
[33]
[34]
G.M.Gordon, O.A.Cotter and E.R.Parker, Effect of Current and Atmospheres on Arc
Temperatures, Welding Journal, 1956, Vol.3(2): pp109s-112s
M.B.C.Quigley, Physics of the Welding Arc, Welding and Metal Fabrication, 1977,
Vol.45(10): pp619-626
F.Matsuda, M.Ushio and A.A.Sadek, Gas-Tungsten-Electrode(Report 4)-Measurement of
Cathode Temperature, Transactions of JWRI, 1989, Vol.18(1): pp24-27
F.Matsuda, M.Ushio and A.A.Sadek, Temperature and Work Function Measurements
with Different GTA Electrodes, Transactions of the Japan Welding Society, 1991,
Vol.22(1): pp7-13
F.Matsuda, M.Ushio and H.Fujii, Gas-Tungsten-Arc Electrode(Report 2)- Measurement
of cathode temperature, Transactions of JWRI, 1986, Vol.15(2): pp7-10
M.Ushio, D.Fan and M.Tanaka, Contribution of Arc Plasma Radiation Energy to
Electrodes, Transactions of JWRI, 1993, Vol.22(2): pp201-207
S.Kuroda, H.Fujimori, T.Fukushima, Measurement of Temperature and Velocity of
Thermally Sprayed Particles using Thermal Radiation, Transactions of the Japan Welding
Society, 1991, Vol.22(2): pp10-17
R.Kawase and M.Kureishi, Fused Metal Temperature in Arc Spraying-Study on Arc
Spraying(Report 3), Transactions of the Japan Welding Society, Vol.16, No.1, April,
1985: pp82-88
J.E.Shea and C.S.Gardner, Spectroscopic Measurement of Hydrogen Contamination in
Weld Arc Plasmas, Journal of Applied Physics, 1983, Vol.54(9): pp4928-4938
E.L.Grove, W.A.Loseke and E.S.Gordon, Development of a Portable Direct Reading
Spectrometer to Monitor Oxygen-Hydrogen Containing Contaminants in Gas TungstenArc Process Shields, Welding Journal, 1970, Vol.17(11): 538s-544s
H.G.Kraus, Experimental Measurement of Stationary SS 304, SS 316L and 8630 GTA
Weld Pool Surface Temperature, Welding Journal, 1989, Vol.68(7): pp269s-279s
W.H.Giedt, L.N.Tallerico and P.W.Fuerschbach, GTA Welding Efficiency: Calorimetric
and Temperature Field Measurements, Welding Journal, 1989, Vol.68(1): pp28s-32s
H.G.Kraus, Surface Temperature Measurements of GTA Weld Pools on Thin-Plate 304
Stainless Steel, Welding Journal, 1989, Vol.68(3): pp84s-91s
W.H.Giedt, X.-C.Wei, and S.-R.Wei, Effect of Surface Convection on Stationary GTA
Weld Zone Temperatures, Welding Journal, 1984, Vol.63(12): pp376s-383s
W.S.Bennett and G.S.Mills, GTA Weldability Studies on High Manganese Stainless
Steel, Welding Journal, 1974, Vol.21(12): pp548s-553s
G.S.Mills, Analysis of High Manganese Stainless Steel Weldability Problem, Welding
Journal, 1977, Vol. 24(6): pp186s-188s
J.C.Metcalfe and M.B.C.Quigley, Arc and Pool Instability in GTA Welding, Welding
Journal, 1977, Vol.24(5): pp133s-139s
W.Faber and B.Jackel, Practical Use of an Arc Sensor for Fillet Welding, Welding
International, 1989, Vol.3(1): pp27-30
H.Nomura, An Automated Control System of Weld Bead Formation in One-sided
Submerged Arc Welding, IIW Doc. XII-c-032-82
H.Nomura, Y.Satoh, K.Tohno, Y.Satoh and M.Kurotori, Arc Light Intensity Controls
Current in SA Welding System, Welding and Metal Fabrication, 1980, Vol.23(48):
pp457-465
17
[35]
[36]
[37]
[38]
[39]
[40]
[41]
[42]
[43]
[44]
[45]
[46]
[47]
[48]
[49]
[50]
[51]
R.J.Salter and R.T.Deam, A Practical Front Face Penetration Control System for TIG
Welding, Proceedings of and International Conference on Trends in Welding Research,
Gatlinburg, Tennessee, USA, 18-22, May, 1986, pp381- 392
C.Connelly, G.J.Fetzer, R.G.Gann and T.E.Auarand, Reliable Welding of HSLA Steels
by Square Wave Pulsing Using and Advanced Sensing (EDAP) Technique, Proceedings
of and International Conference on Trends in Welding Research, Gatlinburg, Tennessee,
USA, 18-22, May, 1986: pp421-423
R.T.Deam, Weld Pool Frequency: A New Way to Define a Weld Procedure, Proceedings
of and International Conference on Trends in Welding Research, Gatlinburg, Tennessee,
USA, 18-22, May, 1986: pp967-970
US Patent, 1966
Q.Wang, L.Zhang, Arc Light Sensing of Droplet Transfer in GMA Welding and the Realtime Control of One Pulse One Droplet, Proceedings of the 40th Annual Conference of
the Welding Technology Institute of Australia, Vol.1, 29 June-3 July, 1992: pp124-136
R.B.Madigan, T.P.Quinn and T.A.Siewert, Sensing Droplet Detachment and Electrode
Extension for Control of Gas Metal Arc Welding, Proceedings of the 2nd International
Conference on Trends in Welding Research, Gatlinburg, Tennessee, USA, 14-18, May,
1989: pp999-1002
J.A.Johnson, N.M.Carlson, H.B.Smartt, Detection of Metal-transfer Mode in GMAW.
Proc. 2nd International Conf. on Trends in Welding Res. Gatlinburg, Tennessee, USA,
14-18 May, 1989: pp377-381
J.H.Chen, D.Fan, Z.Q.He, J.Ye, Y.C.Luo, A Study of the Welding Mechanism for
Globular Metal Transfer from Covered Electrodes. Welding Journal, 1989, Vol.70(4):
pp145-s to 150-s
D.E.Clark, C.L.Buhrmaster, H.B.Smartt, Drop Transfer Mechanisms in GMAW. Proc.
2nd International Conf. on Trends in Welding Res. Gatlinburg, Tennessee, USA, 14-18
May, 1989:pp371-375
Q.L.Wang, The Study on the Relation Between Arc Sound and Arc Phenomena. Science
Report of Harbin Institute of Techology, 1981, Vol.78(9): pp124-129
R.W.Richardson, F.S.Edwards, Controlling GT Arc Length from Arc Light Emissions,
Proceedings of the 4th Trends in Welding Research, Gatlinburg, Tennessee, USA, 5-8
June 1995
Z.Y.Guo, W.H.Zhao, Arc and Thermal Plasma, First Edition, Science Press (China),
1980: pp22-23; pp260-319
Y.L.Song, J.Y.Li, Thermo-equilibrium in Welding Arc Plasma, Transactions of China
Welding Institution, 1995, Vol.15(2):pp138-145
G.Bekefi, Radiation processes in plasma, John Wiley and Sons, INC, New York, 1966
W. Lochte-Holtgreven, Plasma diagnostics, John Wiley and Sons, INC, New York, 1968
J.D.Cobine, Gaseous conductors, McGraw-Hill Book Company, 1958, P290
18