
Arc Efficiency for Gas Tungsten Arc Welding DCEN‐GTAW Hans Åström, Nils Stenbacka, Kjell Hurtig University West, Trollhättan, Sweden. ABSTRACT Published data in the literature between 1955 and up to 2011 shows a wide spread in arc efficiency data, between 0,36 up to 0,90. In the present study the experiments was planned using factorial design at two levels for each of the variables; current, arc length, gas type‐pure Ar and Ar+2H2. Each of the variable combinations was replicated five times. The substrate used was a water cooled Cu‐block. Inlet and outlet cooling water, flow and temperature to the GTAW torch and Cu‐
substrate was measured. Total energy input to the TIG torch, current x voltage, energy going to the substrate and the torch was determined. The results were evaluated using statistical software programs MINITAB® and MODDE®. This resulted in a model for the energy distribution and the arc efficiency for gas tungsten arc welding with DCEN. Key words: Gas Tungsten Arc Welding, arc efficiency, TIG welding, energy input, energy losses, process parameters. List of symbols and abbreviations: n Arc efficiency qt Power input to torch qs Power input to substrate U Arc voltage I Welding current t Measurement time ΔT Temperature difference for cooling water cp Specific heat v Cooling water volume AL Arc length DCEN Direct current electrode negative 1 1. Introductio
on Why is iit importantt to know th
he arc efficiiency of TIG
G welding m
more preciseely? GTAW is one of the mosst widely ussed arc weld
ding metho ds for welding of stainless steel. Itt is importaant to know ho
ow much off the energyy input to thhe torch thaat actually ttransfers too the base m
material. Arc efficciency playss an important role in m
many aspeccts of weldin
ng, for exam
mple in cooling rate o
calculattions, for du
uplex weld m
metal the coooling rate from 1200 C down to 800oC is critical for the form
mation of th
he correct p
phase balan ce between
n austenite and ferrite.. Arc efficciency (n) iss defined ass: n= qs/qtt qt qs Figure 11 Schematicc heat transfer, energy to substratte, losses to
o torch and radiation lo
osses. Where q
qs is power input transsferred to thhe substrate
e and qt is the total pow
wer input to the arc from
m the power source. In the case off DC arc we
elding used in the preseent study; q
qt= arc voltage x current x arc time. T
Two differennt approach
hes to deterrmine arc effficiency are
e used by different researrchers; one is based onn the use of calorimetric experimeents and the
e other d uses differrent heat flo
ow models calibrated w
with measured parameeters. The u
use of method
calorimetric experiiments is a d
direct approoach, whereas the modelling app roach is an indirect approacch. Examplees of calorim
metric approoaches are given in refference [1,22,3,4], mode
elling approacch examples are given I reference [5,6,7]. Re
eview of arcc efficiency sstudies between 1955 un
ntil 2011 is ggiven in refe
erence [8] ((also published in the o
open docum
ment databaase DIVA ‐ h
http://www
w.diva‐portaal.org). In thhe present sstudy a direct calorimeetric approach was used. 2 2. Experimenttal procedu
ure A fixed TIG torch was
w used an
nd the subs trate was a water cooled Cu-blockk, for the se
etup see
figure 2.
TIG torch
Cu substrate
C
Co
ooling water
Figure 22. Experimental setup Power ssource: Migatronic Com
mmander 40
00 AC/DC
TIG torcch: Binzel Tornado
T
WH
H0
Electrod
de: 2,4 mm WT20 98 % wolfram 2 % Thorium
m attached to the negaative pole, tip
t angle
o
60
Gas: Arr, Ar+2%H2 gas flow 10l/min
The Cu--substrate was
w cooled with fresh w
water and th
he temperature and floow rate was
s
measurred continuo
ously. Welding voltage was measu
ured at the torch and thhe welding current
was me
easured with
h a Hall elem
ment. The ccooling wate
er to the we
elding torch was also
measurred regardin
ng temperatture and flow
w rate.
All of the measurem
ment signals was colle
ected by a measuremen
m
nt module ffrom National
Instruments and prresented in a LabView--interface, the
t signals was
w also loggged to a file.
Samplin
ng frequenccy during the
e test 50 kH
Hz, total sam
mpling time after reachhing a stead
dy state
conditio
on; 120sek.
Energy input to the
e torch qt = U x I x t.
Energy input to sub
bstrate, qs, and torch lo
osses q = v x ∆T x cp
cp=speccific heat forr water 4,18
81[kJ/(kg·K))]
Experim
mental matriix for the fac
ctorial desig
gn:
Arc leng
gth [mm]
2 (giving
g~10,5V)
5 (giving
g~12,5V)
Current [a
amp]
75
150
Shielding ggas
Ar
Ar2H2
3 This dessign matrix gives total 8 experime
ents which were
w
replica
ated 5 timess.
3. Results
In table 1 the results of the 40
0 test runs a
are shown.
Table1. E (qt) total energy input to the TIG
G torch, Su
ubstr = the arc
a efficienccy, Gun= los
sses to
the cooling of the to
orch, losses
s= other lossses, for exa
ample, irrad
diation to thhe surroundings.
The ressults from th
he experime
ents were evvaluated us
sing the stattistical softw
ware progra
ams
MINITA
AB® and MO
ODDE®.
In figure
e 3 the norm
mal probability plot for tthe tree ene
ergy parts can
c be seenn.
Probability Plot of Energy to Substrrate=Arc efficiency
Prob
bability Plot of Energy to tor ch
Normal - 95% CI
Mean
StDev
N
AD
P-Value
95
90
Normal - 95% CI
99
0,8345
0,02540
40
0,392
0,364
Mean
StDev
N
AD
P-Value
95
90
99
0,09903
0,02366
40
1,373
<0,005
90
80
70
70
30
Percent
80
70
60
50
40
60
50
40
30
30
20
20
10
10
10
5
5
5
1
1
0,775
0,800
0,825
0,850
Substrate
0,875
5
0,900
0,925
0,02
0,04
0,06
0,08
0,10
Gun
0,12
0,14
0,16
0,18
0,06695
0,02747
40
0,320
0,520
60
50
40
20
0,750
Mean
StDev
N
AD
P-Value
95
80
Percent
Percent
Probab
bility Plot of Losses
Normal - 95% CI
99
1
0,000
0,025
0,050
0,075
L
Losses
0,100
0,125
0,15
50
Figure 3
3. Probabilitty plots.
The arcc efficiency and
a other lo
osses are n ormal distributed but lo
osses to thee torch are not
normal distributed at the 95% confidence
e level. This is probably
y due to thee closed coo
oling
system used for the
e torch.
4 3.1Arc e
efficiency:
Investigation: TIG (MLR)
Scaled
d & Centered Coeffic
cients for Arc efficien
ncy
0,010
0
0,000
0
-0,010
0
N=40
DF=33
R2=0,733
Q2=0,608
I*Gas(Ar2H)
I*Gas(Ar)
AL*Gas(Ar2H)
AL*Gas(Ar)
AL*I
Gas(Ar2H)
Gas(Ar)
I
AL
-0,020
0
R2 Adj.=0,685
RSD
D=0,0143
Conf. lev
v.=0,95
Figure 4
4. Factors affecting
a
the
e Arc Efficie
ency.
As can be seen in figure 4, inc
creasing the
e current an
nd arc length will reducce the arc efficiency
and the use of an Ar2H
A
ure compare
ed to pure Ar
A shielding gas gives a positive effect.
e
All
2 mixtu
of the in
nteraction effects fall ou
utside of the
e 95% conffidence interval. Regresssion coeffiicients
for the a
arc efficienccy:
5 Observe
ed versus model
m
prediction can be
e seen in fig
gure 5.
Figure 5
5. R2 a meassure of the fit for the mo del and Q2 a measure of how well thhe fitted mod
del will predict n
new experim
mental condittions.
Model m
mapping forr the differen
nt shielding
g gases used in the investigation ccan be seen
n in
figure 6.
Figure 6 Model ma
apping of arc efficiencyy for the diffe
erent shield
ding gases; Ar and Ar2H.
6 3.2 Ene
ergy losses in the torch:
Approximately 5-10
0% of the energy supp
plied to the torch
t
is lost due to torcch-cooling. As
A can
be seen
n in figure 7 torch losse
es are incre asing with higher
h
welding current,, as can be
expecte
ed. The weld
ding current is the onlyy parameterr significant at the 95%
% confidence
e
interval affecting to
orch losses.
Figure 7
7. Factors effecting
e
ene
ergy lossess in the torch
h.
The mo
odel fit for to
orch losses explains 69
9% of the re
esults (R2=0
0,691). The model map
pping of
torch lossses showss a quiet diffferent beha
avior for the two tested shielding ggases, see figure
f
8.
Figure 8
8. Mapping of torch los
sses
7 If we compare the model predictions for torch losses to the mean value of the 5 replicates for
each parameter setting we can see that the model fit is quite god, table 2.
Test series MV measured Mod. Value Diff Meas‐ModValue 2‐75‐Ar 0,064 0,066 ‐0,002 2‐150‐Ar 0,118 0,112 0,006 5‐75‐Ar 0,085 0,085 0,000 5‐150‐Ar 0,115 0,115 0,000 2‐75‐Ar2H 0,092 0,090 0,002 2‐150‐Ar2H 0,124 0,122 0,002 5‐75‐Ar2H 0,084 0,083 0,001 5‐150‐AR2H 0,109 0,112 ‐0,003 Table 2. Mean value measured – model value.
3.3 Other losses
The mean value for other losses is 7% with a rater wide spread 2-12%. The model for these
losses based on the experimental data shows a very low prediction power with an R2 value of
33% and a Q2 value of 3%.
5. Discussion
The raw data are normal distributed except for the energy losses in the torch. The reason for
this is that the closed loop cooling system is not powerful enough to give a constant
temperature difference between inlet- and outlet water temperature during the measurement
time. This can be seen in figure 9, showing two examples of the temperature difference
changes during the test cycle, at low welding current the temperature difference is going
down, the cooling is effective, but at high current the situation is changed.
5 mm arc length‐75 amp‐Ar
5 arc length‐150 amp‐Ar
1,2
∆T
∆T
2,9
1
0,8
2,7
2,5
0
100
200
Sek
300
0
100
200
300
Sek
Figure 9. Temperature difference cooling water out-in to the torch during test cycle.
In spite of this the measured torch losses compared to the model predictions are quite god.
8 A question not answered by this investigation is if the arc efficiency and torch losses are
influenced by different torch designs.
4. Conclusions
Increased arc length and current gives reduced arc efficiency for the GTAW-DCEN
welding process.
Ar2H shielding gas gives a positive effect on arc efficiency compared to a pure Ar gas,
but the effect is small, ~+1,5%.
GTAW-DCEN is an effective welding process with an arc efficiency value in the range
0,81-0,86.
5. Acknowledgments
Financial support for the study given by the KK-foundation is gratefully
acknowledged.
6. References
1. Rykalin, N.,Berechnung Def Värmevorgänge Beim Schweissen. VEB Verlag
Technick, Berlin 1957.
2. Apps, R.L.& Milner, D.R., Heat flow in argon-arc welding. British Welding Journal,
1955, 2(10),pp 475-485.
3. Wilkinson, J.B. &Milner, D.R., Heat transfer from arcs. British Welding Journal, 1960,
7(2), pp 115-128.
4. Smartt, H.B, Stewart, J.A.& Einerson, C.J., Heat transfer in gas tungsten arc welding.
ASM Metals/Materials Technology Series, No8511-011.
5. Dutta, P., Joshi, Y.& Franche,C., Determination of gas tungsten arc welding
efficiencies. Experimental Thermal and Fluid Science, Vol 1, Issue1, July 1994, pp8089.
6. Mishra, S. & DebRoy, T., A heat-transfer and fluid-flow-based model to obtain a
specific weld geometry using various combinations of welding variables. J. Applied
Physics 98, 2005,p.044902.
7. Bag, S. & De, A. Probing of Transport Phenomena Based Heat Transfer and Fluid
Flow Analysis in Autogeneous Fusion Welding Process. Metal & Materials
Transactions A, Vol 41A, Sept. 2010, pp. 2337-2347.
8. Stenbacka, N., Chouquet, I. & Hurtig, K. Review of Arc Efficiency Values for Gas
Tungsten Arc Welding, IIW Doc. XII-2070-12/212-1229-12.
9