University of Wollongong

Research Online
Faculty of Engineering - Papers (Archive) Faculty of Engineering and Information Sciences

2005

Evaluation of calculated arc energy heat input in

MMAW processes
Zoran Sterjovski
zoran@uow.edu.au

John Norrish
University of Wollongong, johnn@uow.edu.au

Rian Holdstock
rh083@uow.edu.au

Glen Sloan

David J. Nolan
University of Wollongong, dnolan@uow.edu.au

http://ro.uow.edu.au/engpapers/3426

Publication Details
Sterjovski, Z., Norrish, J., Holdstock, R., Nolan, D. J. & Sloan, G. (2005). Evaluation of calculated arc energy heat input in MMAW
processes. Australasian Welding Journal - Welding Research Supplement, 50 42-48.

Research Online is the open access institutional repository for the University of Wollongong. For further information contact the UOW Library:
research-pubs@uow.edu.au
 Evaluation of calculated arc energy heat input
in MMAW processes
Zoran Sterjovski, John Norrish, Rian Holdstock, Glen Sloan and David Nolan
The ability o f m anual m etal arc welders to reach target heat
input (HI) values between 1.3 kJ/m m a nd 2.9 kJ/m m by vary­
ing.w eld travel speed was examined. Weld process data was
collected during the dow nhand and overhead w elding o f 6
highly restrained quenched and tem pered (QT) steel plates and
the results show ed that the m easured average H I values were
closely m atched to the target HI. However, there w as consider­
able scatter, particularly at higher H I targets, which warranted
investigation. The frequency at which welders achieved their
target H I (or travel speed) decreased a t higher H I values, but
nonetheless the standard deviation (SD) fro m the average H I
was less than 10% fo r all target H I values, w hich is w ell within
the average 15% deviation allow ed fo r a qualified w eld in the
Australian Pressure Vessel Standard, A S3992-1998.
The effect o f localised variations in voltage on H I w as also
considered. Even though voltage fluctuated by as m uch as 40%
during the deposition o f an electrode, the overall im pact on
arc energy H I w as considered negligible because o f the short
duration a t these ‘extreme ’ voltages. Moreover, it w as fo u n d
th a t the levels o f diffusible hydrogen m arginally decreased as
H I increased, which lends support to the fa c t that H I is a con­
tributor to w eldm ent cooling rate. Additionally, differences in
m icrostructure and hardness were also observed betw een low
(1.3 kJ/m m ) and high (2.9 kJ/m m ) H I values in the diffusible
hydrogen test (bead on plate) sam ples.
Keywords
H eat input, m anual m etal arc w elding, diffusible hydrogen,
travel speed, hardness, quenched and tem pered steel.
Introduction
T he heat input (HI) equation is com m only used to define the
com binations of voltage, current and travel speed in arc welding
processes that w ill result in fit-for-purpose w elds. Studies of
heat flow in arc welding have found that for a given process,
and for travel speeds w ithin the usual range for typical w eld­
ing procedures, the arc energy per unit length o f w eld (or HI)
is a useful param eter to correlate w ith w eldability resu lts1. In
this paper, the sensitivity and reliability o f the H I param eter
for use in m anual m etal arc w elding (M M AW ) processes is
investigated.
The HI equation (Equation 1) calculates the total pow er o f
the arc m ade available per unit length o f w eld. It is not the total
H I absorbed by the metal, as often inferred, because consider­
able heat can be lost by radiation, convection, conduction and
sp atter1. It is reported that arc w elding processes are able to
capture the arc energy input as heat to the extent o f 20-85% C
The authors are respectively: Research Fellow, Professor,
PhD student at the University o f W ollongong, N S W 2522;
P rincipal Subm arine Technical A uthority a t A S C Pty Ltd,
G PO B ox 2472, Adelaide, SA 5001; a nd Senior L ecturer a t
the U niversity o f Wollongong, N S W 2522.
In MMAW, the arc efficiency is 70-85% , situated som ew here
betw een that for subm erged arc w elding (SAW) w here the arc
is covered by a flux, and gas tungsten arc w elding w here 20%
o f arc energy is lo st by radiation. In addition to w eld process
factors, jo in t preparation and m aterial type also have an effect
on arc efficiency.
w here H I is heat input in kJ/m m , V is voltage (V), I is current
(A) and S is travel speed (m m /m inute).
H I target values are typically specified in qualified weld
procedures for steel w ith the intention of producing a w eld
that satisfies m inim um m echanical property requirem ents and
is free o f defects; in particular hydrogen assisted cold cracking
(HACC). However, control o f H I alone does not guarantee weld
quality and is usually accom panied by m inim um preheat treat­
m ent and m axim um interpass tem perature in w eld qualification
procedures. As a result, a means of microstructural (cooling rate)
control is achieved, as is increased tim e at higher tem peratures
for diffusible hydrogen to diffuse/effuse from the w eldm ent.
The recom m ended H I ranges for the M M AW o f quenched
and tem pered (QT) grade steels, akin to the steel investigated
(BIS812 EM A), varies from 1.25 kJ/m m to 5.0 kJ/mm. Selection
o f the appropriate nom inal H I w ithin this range is dependent
on plate thickness, as are the selection of preheat treatm ent
and interpass tem perature ranges2. M inim um H I values are
typically specified as p art o f a H A CC risk m itigation strategy
and m axim um H I values are specified to ensure that strength
requirem ents are met.
C ooling rate, w hich is a function o f m aterial, geom etric and
process param eters, is m ore critical than H I alone in influencing
w eldm ent m icrostructure and, in turn, m echanical properties.
Therefore it w ould appear that a cooling rate (or heat output)
param eter w ould b e m o re beneficial than a H I param eter.
However, cooling rate is difficult to quantify w ithout directly
m easuring it. M ost attem pts to correlate w eld cooling rate with
H I have been unsuccessful, except for sm all ranges o f current,
voltage and travel speed in M M A W and SAW p ro cesses1.
D ifficulties in establishing a correlation betw een cooling rate
and H I exist because cooling rate is not only controlled by HI,
but also by m any oth er variables such as w eld geom etry, plate
thickness, m aterial type, preheat/interpass tem peratures and
w elding location. Furtherm ore, cooling rate is also dependent,
on the specific values o f I, V and S that produce a given HI
because they affect w eld bead shape and volum e3.
T h e m ain p roject, w hich is bein g u ndertaken u nder the
au sp ices o f th e C o o p erativ e R e sea rch C en tre fo r W elded
Structures, focuses on the follow ing issues w ith respect to
M M A W w eldm ents jo in in g BIS 812 E M A m aterial in subm a­
rine structures4:
1. Determ ination o f welding param eter tolerance levels against
the initiation o f H A CC in highly restrained w eldm ents and
the suitability o f m itigation strategies.
 2. Evaluation o f current non-destructive testing (NDT) methods Table 1: N um ber o f w eld runs and num ber o f electrodes
and their ability to detect HACC. required f o r P lates 1-6.
3. R e s e a rc h in to th e d e v e lo p m e n t o f a lte r n a tiv e N D T

Total Electrodes
No. Downhand

Total No. Runs

Run (Average)
Electrodes per
No. Overhead
methods.

(Estimate)
Target HI
(kJ/mm)
4. R esearch into the influence o f segregation in w eldm en t

Runs

Runs
m icrostructure to the initiation o f H A CC, and the variation Plate
of the m icrostructure in relation to w elding HI.
5. M easurem ent o f residual stresses during and after the w eld­
ing o f a highly restrained butt joint.
1 1.3 19 16 35 6 210
6. Variation in the toughness o f w eld m etal w ith respect to the
position of welding and investigation into correlation between 2 1.7 15 13 28 7 196
Charpy im pact test results and fracture toughness. 3 2.0 12 11 23 9 207
The aim o f this paper is to present results o f initial investiga­ 4 2.2 13 13 26 9 234
tions o f the suitability o f the H I approach and its relationship
w ith cooling rate in MMAW. This has been done by assess­ 5 2.5 10 10 20 12 240
in g data acquired from the M M A W o f 6 highly restrained 6 2.9 9 13 22 15 308
Q T plates, w hich are prim arily being used for the m ain body
o f the research w ork detailed above, coupled w ith diffusible Table 2: A verage m easured travel speed, voltage and current
hydrogen test results undertaken at the sam e heat inputs as the f o r each p la te in the 1G and 4G positions.
h ighly restrained plates. Consequently, results show ing the

Voltage (V)
1G C urrent

4G C urrent
relationships betw een HI, hardness, m icrostructure and levels

1G and 4G
(m m /m in)
4G Travel
(m m /m in)
1G Travel

Speed
Speed
o f diffusible hydrogen are presented and discussed.

(A)

(A)
Plate
Experimental method
Welding of te s t plates
1 176 167 132 122 24
A total o f 6 highly restrained test plates w ere w elded at various
h eat inputs betw een 1.3 and 2.9 kJ/m m in both the overhead 2 143 170 106 124 24
(4G ) and dow nhand (1G) positions, as show n in Table 1. Each 3 127 170 95 130 24
test panel consisted o f tw o 1600 m m x 500 m m plates w ith a
4 116 175 85 129 24
“2/3 1/3” w eld preparation along the 1600 m m length. A gril­
lage of transverse and longitudinal bracing w as interm ittently 5 95 169 71 123 24
w elded to the plates on the “ 1/3” preparation (or 1G) side for 6 84 166 60 122 24
longitudinal and transverse restraint. A dditional longitudinal
restraint was achieved by initially w elding 300m m lengths at
each end, w hich resulted in a final test w eld length o f 1000 boxes prior to use. Two w elders w orked on each plate and they
m m . The test plate and strongback m aterial used was 35 m m alternated after each electrode. Figure 1 shows photographs of
thick BIS 812 E M A QT steel. the test plate and w elders w elding in the 1G and 4G positions.
B etw een 5 and 15 electrodes (or rods) w ere required to Table 2 lists the average recorded travel speeds in the 1G and
com plete the 1000 m m w eld length, depending on the target HI. 4G positions for each plate.
T he w eld consum able, an E12018-M 2 low hydrogen electrode,
was 3.2 m m in diam eter for w elding in the 4G position and 4
Monitoring weld process param eters
m m in diam eter for w elding in the 1G position. T he m inim um T he A rcw atch™ w eld m onitoring system was used to collect
preheat treatm ent tem perature, 120°C, w as achieved by using voltage (V) and current (A) data for an interval o f 64 seconds
electric resistance heating pads, and the interpass tem perature for 1 to 3 electrodes in each w eld run. The data collected were
w as m a in ta in ed below 175°C . T he w eld p re p a ra tio n w as verified w ith a calibrated m ulti-m eter that m onitored voltage
ground to rem ove paint, oxidation and other potential sources and current throughout the entire w elding process. Voltage and
o f hydrogen. A ll consum able w ere baked and placed in hot current w as m easured at the w elding m achine and w eld lead

Figure 1. (a) O verhead a nd (b) dow nhand M M A W o f the Q T plate.

 losses w ere approxim ately 1-2 volts. Travel speed (m m /m in)
w as m easured from the run-out length (ROL) and run-out tim e
o f the electrode. The weave am plitude was also m onitored and
at all tim es less than 3 tim es the electrode diameter.
Diffusible hydrogen levels as a function of HI
T he levels o f diffusible hydrogen were m easured by the inert
gas h ot extraction m ethod on sam ples w elded in the 1G po si­
tion. Sam ples were prepared, and tested, as specified in AS/NZS
3752:1996, “W elding - M ethods for determ ination o f the dif­
fu sib le hydrogen content o f ferritic w eld m etal produced by
arc w elding”. Average diffusible hydrogen levels taken from 3
sam ples w ere plotted as a function o f average HI. It should be
noted that the diffusible hydrogen test is carried out on a bead
on p late sam ple and gives an indication o f the level o f diffus­
ible hydrogen (mL/lOOg) to com e from an electrode during the
d eposition o f a w eld bead at a particular HI.
M icrostructure and hardness as a function of HI
O ptical m icroscopy and Vickers hardness testing was carried
out on the w eld bead o f diffusible hydrogen tests sam ples
w elded at heat inputs o f 1.3 kJ/m m and 2.9 kJ/m m , in order to
determ ine the effect of H I on m icrostructure and hardness. The
results presented are prelim inary and sim ply from the diffusible
hydrogen test piece samples.
Results
Figures 2-7 show the scatter in HI resulting from variations
in w elding travel speed for Plates 1 to 6. F igure 8, w hich is
a p lo t o f target HI versus the calculated average HI, shows
that the calculated average H I closely m atched the target HI.
Figure 9 is a plot o f the average HI, standard deviation (SD)
and range for each test plate, and F igure 10 shows the H I
frequency distribution curves at each target input. The latter
figure reiterates the presence o f scatter in m easured H I values
despite w ell-m atched target and average H I values, and also
highlights other trends, w hich are discussed later, in relation
to an increasing target HI.
Figure 11 shows the typical voltage profile during the deposi­
tion o f an electrode and the im pact o f this profile on localised
HI. F or exam ple, at m axim um voltages the localised H I w ould
increase by as m uch as 40% (assum ing current and travel speed
rem ained constant). Figure 12, a plot o f m easured levels o f dif­
fusible hydrogen fo r a range o f heat inputs, shows the levels
o f diffusible hydrogen m easured in the w eld bead m arginally
decrease as H I increases. Figures 13, 14 and 15 respectively
show the effect o f H I on m acrostructure, m icrostructure and
hardness o f diffusible hydrogen w eld test sam ples.

2 .5
-T a rg e t HI (1 .3 k J /m m )
2 .3 2 .3 -
_ ■ H I (p e r e le c tro d e ) - 4G E 2.1 -
2.1
■ A HI (p e r e le c tro d e ) - 1G E
~3 1.9 -
44
1.7
P3
Q . 1 i ! = ■, l 1 -5 i
1.3 ' l . S i ' 1 ■1 ■■■ ■ A A
A
£ 1 .3 -
+-■
05 1.1 (8 1.1 -T a rg e t HI (1 .7 k J /m m )
CD 0
X 0 .9 ■ I 0 .9 H I (p e r e le c tro d e ) - 4G
0 .7 ■
0 . 7 -j H I (p e r e le c tro d e ) - 1G
0 .5 0.5
0 2 4 6 8 1012141618202224262830323436 0 2 4 6 8 10 12 14 16 18 20 22 24 26 28
Run No. Run No.
Figure 2. C alculated heat inputs in each w eld run f o r Figure 3. C alculated heat inputs in each w eld run fo r
P late 1. H I calculations are based on variations in travel P late 2. H I calculations are b a sed on variations in travel
speed, average voltage an d constant current. The target H I speed, average voltage and constant current. The target H I
w as 1.3 kJ/mm. was 1.7 kJ/mm.

3 .0 3.0
-T a rg e t H I (2 .0 k J /m m )
2.8 _ 2.8
E 2.6 HI (p e r e le c tro d e ) - 4 G
| 2.6 -I
E H I (p e r e le c tro d e ) - 1G
2 .4 2 .4
2.2 2.2 _ 4 i_ :
55
Q. 2.0 I 2.0- » i *i
_c 1.8 £ 1.8
•*—3
03 1.6 -T a rg e t H I (2 .2 k J /m m )
0 I 1 ‘6
X 1.4 X 1.4 H I (p e r e le c tro d e ) - 4G

1.2 1.2 -I H I (p e r e le c tro d e ) - 1G

1.0 1.0
0 2 4 6 10 12 14 16 18 20 2 2 24 0 2 4 6 8 10 12 14 16 18 20 22 24 26
Run No. Run No.

Figure 4. C alculated heat inputs in each w eld run f o r P late Figure 5. C alculated h ea t inputs in each w eld run fo r Plate
3. H I calculations are based on variations in travel speed, 4. H I calculations are based on variations in travel speed,
average voltage a nd constant current. The target H I w as 2.0 average voltage and constant current. The target H I was 2.2
kJ/m m . kJ/mm.
 4.0
3.8
3.6
“5
3.4
XT. 3.2
= 3.0
£ 2.8
15 2.6 ■
'CD
I 2.4 ■
2.2
— Target HI |2.9 kJ/mm) ■ HI (per electrode)-4G ^a HI (per electrode)-1G
2.0
4 6 8 10 12 14 16 18 22 24 2 4 6 8 10 12 14 16 18 20 22
R un No. R un No.
F igure 6. C alculated heat inputs in each w eld run fo r Figure 7. C alculated heat inputs in each w eld run fo r
P la te 5. H I calculations are based on variations in travel Plate 6. H I calculations are based on variations in travel
speed , average voltage and constant current. The target H I speed, average voltage a n d constant current. The target H I
w as 2.5 kJ/mm. was 2 .9 kJ/mm.

4.6
E 2.75 ^ 4.1 □ A v e ra g e
5-2.50 | 3.6 R ange

q.2.25 3 3.1
3 2.00 12.6
x 1.75 Z 2 -1
§,1.50 S 1.6
g 1.25 1.1
< 1.00 0.6
1.00 1.25 1.50 1.75 2.00 2.25 2.50 2.75 3.00 2 3 4
Target H eat Input (kJ/m m ) P late No.
Figure 8. A verage H I versus target H I fo r all o f the test Figure 9. Average HI, SD a n d range f o r P lates 1-6. H I
plates. Average H I calculations are based on variations in calculations are based on variations in travel speed, average
travel speed, average voltage and constant current. voltage a n d constant current.

0.50

0.45

0.40

1 °-35

1 0.25
CD
w
H 0.20
E
| 0.15

^ 0.10

0.05

HI (kJ/m m )

Figure 10. H I fre q u en c y distribution curves f o r Plates 1-6. Frequency is norm alised by dividing the num ber o f events in the
range m edian by the total num ber o f events, a nd the H I values (x axis) represent values w ithin ± 0.05 kJ/mm. H I calculations
are ba sed on variations in travel speed, average voltage a nd constant current.

WELDING RESEARCH SUPPLEMENT AUSTRALASIAN WELDING JO U R N AL-VO LU M E 5 0 ,THIRD QUARTER - 2 0 0 5 45

 34
32
30
28
26
24
22
20
18
16
14
12
0 5 10 15 20 25 30 35 40 45 50 55 60 1.00 1.50 2.00 2.50
T im e (sec) H eat Input (kJ/m m )

Figure 11. A typical p lo t o f voltage versus time fo r constant Figure 12. M easured diffusible hydrogen levels versus HI. Sample
current M M A W process fo r both the 1G and 4G positions. size f o r each H I was 3 a n d the error bars represent SD.

Figure 13. Weld m acrograph o f the diffusible hydrogen test sam ples f o r a H I o f (a) 1.3 kJ/m m a n d (b) 2 .9 kJ/mm. Travel speed
was the variable param eter. Sam ple w idth is 15 mm.

fig u re 14. P hotograph o f w eld m etal m icrostructure fro m the centre o f the w eld bead in diffusible hydrogen test sam ples fo r a
H o f (a) 1.3 kJ/m m and (b) 2.9 kJ/m m . Travel speed w as the variable parameter. E rror bars represent 10 microns.

16 AUSTRALASIAN WELDING J O U R N A L -V 0 II IMF Fn t u


Discussion
The H I param eter, together with preheat treatm ent and interpass
tem perature control, has been used successfully for a num ber
o f years as a m eans of controlling w eldability in steels. T hese
variables, w hen com bined, attem pt to avoid hardenability and
h eat affected zone H A CC by controlled cooling. O ther factors
w hich also influence w eldability are: consum able handling;
consum able baking procedures; clean w elding preparation;
and h eat treatm ents such as outgassing and p ostw eld heat
treatm ent.
Steel and w eld consum able m anufacturers direct end-users
toward quality w elding by recom m ending a broad range of
process param eters, usually current and voltage, or HI. W eld
qualification procedures are then undertaken to facilitate a
vveld that com plies w ith either a military, structural or pressure
/essel standard. A com pliant w eld has a m inim um acceptable
evel o f d isco n tin u ities and the rig h t blen d o f m ech an ical
troperties. H ow ever, an appreciable change in arc w elding
:onditions (current, voltage and travel speed) to those used
or qualification results in the need for re-qualification5. The
evel o f variation in current, voltage and travel speed allow ed
3 quantified in the A ustralian Structural and Pressure Vessel
Itandards6’7. A m axim um average deviation o f 15% fo r travel
peed, 15% fo r current and 15% for voltage from the qualified
alue is specified in AS 1554.4-2004 for M M AW processes6. In
te unlikely event that all three param eters are near the lim iting
alues this is an extrem ely liberal constraint. In contrast, the
taxim um deviation allow ed in the pressure vessel w elding
andard (A S3992-1998) is 15% from the average arc energy
>r H I) input that was qualified, w hich is considerably stricter6,
hese specifications outlined in the A ustralian S tandards do
Dwever provide a benchm ark for assessing the H I data from
e current project.
T he cu rrent project involved investigating the effect o f vari-
ions in arc energy H I (by varying w eld travel speed) on the
sceptibility to H A CC and various m echanical properties. A
11 range o f H I values (1.3-2.9 kJ/m m ) were selected to estab-
h w hether they w ere susceptible to H A CC under conditions
high restraint. N on-destructive testing by various m ethods,
fluding conventional ultrasonic testing, phased array ultra-
nic testing, acoustic em ission m onitoring and radiography,
er extended periods (up to 210 days) verified that the w elds
:re o f acceptable quality in term s o f defects. This suggests
it the heat inputs investigated were appropriate, although
: to be verified by destructive means. Similarly, it should be
ted that the m echanical properties o f the test plates are yet
be verified.
Table 1 illustrates the relationship between HI and the num ber
veld runs required in the 1G and 4G positions, and the aver-
; num ber o f electrodes required to com plete a w eld run. A s
increases the num ber o f runs required to com plete the w eld
reases due to the increase in deposition volum e associated
h a higher H I (Figure 13). A lso, the num ber o f electrodes
aired for w eld com pletion increased w ith an increase in
w hich is prim arily due to the w elder consum ing less o f the
:trode due to the physical effects o f excessive heat.
figures 2-7 show the H I values calculated from the recorded
age, current and travel speed data from the vast n um ber o f
trades used to weld Plates 1 to 6. The location of the majority
te data points, irrespective o f w elding position, lie w ithin
MMAW variation lim it (±15% ) considered acceptable for
n A S3992-19987 and travel speed in A S1554.4-20046. A s
suit, the average H I was close to the target HI (Figure 8).
greatest deviation from nom inal H I was generally recorded
ng the w elding o f the first root run (in the 4G position),
fly as result o f the differences in the root gap along the
N f ; R p q p A R T I - l Q IID D I c m c m t
300
□ 1.3 kJ/m m
° 275
> 2.9 kJ/m m
^ 250
w
CD
£
■§ 225
X
% 200
150
Figure 15. Average Vickers hardness a n d SD (error bars) fo r
the deposited W M fro m the diffusible hydrogen test sam ples
fo r a H I o f 1.3 kJ/m m and 2.9. kJ/mm. Travel speed w as the
only variable parameter.
length o f the w eld, w hich required uniform filling to ensure
successful deposition o f ensuing w eld runs. W elders w ere also
acclim atising to the required travel speed in the root run.
T he m agnitude o f digression from the target H I increased
w ith rising H I due to the increased sensitivity o f w eld travel
speed at shorter ran out lengths. T he A ustralian Structural and
Pressure Vessel Standards do cater for this by expressing the
allow able variation in voltage, current, travel speed or H I as a
percentage. Therefore w hen attem pting to establish HI related
trends or m anipulating w eldm ent properties a significant gap
(7>30% ) betw een potential H I target values is required. It is
also im portant to be accurate w ith pre-heat and interpass tem ­
peratures as they also influence w eldm ent cooling rate.
Figures 2-7 also show that the w elding position did not have
an effect on the am ount and m agnitude o f scatter in H I, and for
the 1.3 kJ/m m plate (Plate 1) m ost o f the m easured H I values
w ere slightly above the target H I suggesting that the w elders
w ere near their m axim um travel speed limit.
F igures 8-10 are ‘sta tistica l’ plots o f the data b ased on
m easured travel speeds, average voltage and constant current
recordings. Figure 8 shows that the average HI for each weld was
closely m atched to the target HI. These averages were calculated
from 9 8 ,7 3 ,4 4 ,2 3 2 ,2 4 8 , and 250 HI (voltage, current and travel
speed) m easurem ents for Plates 1-6, respectively. Furtherm ore,
F igure 9 shows the SD and range for the m easured H I values at
each target HI. A large range in m easured H I values is seen for
each plate and this is m ainly due to the H I deviations observed
in the root run for reasons previously discussed. Nevertheless,
the SD was less than 10% for target HI, e.g., the SD w as ±0.13
kJ/m m for Plate 1 and ±0.28 kJ/m m for Plate 6. It should be
noted how ever that the w elding was carried out in laboratory
conditions and a higher deviation from the target HI in fabrica­
tion w elds m ay be encountered, particularly w hen w elding in
confined space or w hen access is obstructed.
F igure 10 plots the H I frequency distribution curves for
each test plate. The frequency is norm alised to allow direct
com parisons betw een the target heat inputs. A general increase
in the w idth o f the distribution curve and a general decrease in
the norm alised frequency as H I increases are observed, w hich
highlights the difficulty experienced by w elders in achieving
the correct travel speeds at higher heat inputs. Even so, the
quality o f the w elds was not affected as verified by ultrasonic
and radiographic exam ination.
 The preceding discussion was w ith respect to HI variations
resulting from changes in travel speed. T he follow ing discus­
sion focuses on the effects o f voltage and travel speed on local­
ised HI, and the effect o f H I on cooling rate, m icrostructure
and hardness. Figure 11 shows a plot o f voltage against tim e
during electrode deposition, revealing localised H I variations
due to voltage fluctuations, w hich are caused by fluctuations
in electrode arc length and m achine characteristics (high speed
oscillation). The voltage recorded peaked at - 3 4 V and dipped
to - 1 6 V (disregarding short circuits), F igure 11. T he im plica­
tions of these voltage variations are a localised H I increase by
- 4 0 % and a decrease by -3 3 % from the target HI. The im pact
o f this is uncertain, but it is likely to be insignificant due to
th e short duration at these voltages and the w eaving technique
u sed by the w elders. However, local variations in travel speed
th at were observed, but could not be quantified, may be more
significant. To explain, the w elders m ark their ROL and a w eld­
ing supervisor keeps them aw are o f the tim e, so that they are
able to significantly adjust their travel speed during deposition.
Consequently, the travel speed (and H I) for significant portions
o f the electrode can differ. M oreover, w hen the w elders travel
speed was incorrect (too fast or too slow ) occasionally they
w ould over-correct w ith the ensuing electrode to ‘im prove’ the
average HI. This was particularly a problem at high heat inputs
(Figures 5-7). It should be stated that the w elders w ere highly
skilled and accustom ed to close control o f HI.
T he levels o f diffusible hydrogen w ere m easured by the hot
extraction m ethod at the sam e arc energy heat inputs as Plates
1-6. Figure 12, w hich is a plot o f diffusible hydrogen levels
versus m easured H I, shows a m arginal decrease in diffusible
hydrogen levels w ith increasing HI. This trend, how ever slight,
is expected as higher average heat inputs result in slower cooling
rates w hich aid the diffusion o f hydrogen aw ay from the w eld­
m ent. It should be noted that the diffusible hydrogen levels for
each H I are low and greatly influenced by consum able handling
procedures, w hich is identical to the procedures used by these
particular w elders in the fabrication o f the subm arine hull. T he
effect o f H I on m icrostructure is evident in F igure 14, w hich at
1.3 kJ/m m the ferrite is predom inately acicular (Figure 14(a))
and at 2.9 kJ/m m the m icrostructure consists o f ferrite grains
(Figure 14(b)). T he ferrite m orphology is dependent on intra-
g ranular particles present (oxides, sulfides etc) that can act as
nucleation sites for acicular ferrite and the cooling rate o f the
w eldm ent8. O bviously in this case, the higher H I (2.9 kJ/m m )
slow ed the cooling rate sufficiently to prom ote grain bound­
ary nucleation and growth o f ferrite at the expense o f acicular
ferrite the nucleation o f acicular ferrite. E ven so, the hardness
o f the low er HI w eld is only slightly higher (Figure 15), again
due to the faster cooling rate associated w ith low er heat inputs.
T hese observed trends are supported by R udolph9, w ho reported
that m icrostructural changes and hardness decreases as H I is
increased during the M M AW o f Fe alloys.
Conclusions
T he follow ing conclusions have been draw n for the range of
heat inputs (1.3-2.9 kJ/m m ) m easured during the M M A W of
35 m m thick Q T plate:
• T he num ber of w eld runs required decreased with increasing
HI.
• Scatter in m easured H I values increased as the target H I
increased.
• T h e SD of HI was less than 10% and the m ajority o f H I
m easurem ents fell w ithin the 15% deviation allow ed for
AQ
average H I in AS3992-1998, albeit using experienced w eld­
ers and close supervision.
• T he frequency at w hich w elders can achieve their target
travel speed (and HI) decreased at higher heat inputs.
• Variation in localised HI can be up to approxim ately ±35%
due to localised fluctuations in voltage, but the effects are
considered negligible due to the short durations o f these
variations and the therm al inertia o f the process.
A dditionally, the follow ing conclusions w ere draw n from
the investigation o f bead-on-plate diffusible hydrogen tests at
1.3 and 2.9 kJ/m m:
• From the resulting m icrostm cture and hardness it is evident
that HI has an effect on cooling rate. Low H I values promoted
a m icrostructure that was predom inantly acicular ferrite.
• Low H I values also increase the W M hardness and produce
m arginally higher levels o f diffusible hydrogen.
Future Work
The rem ainder o f the test program is currently being im ple­
m ented and w ill soon be finished. D estructive exam ination
o f the w elds by m illing and m agnetic particle inspection will
verify the accuracy o f the N D T experim ents. C haracterisation
o f the test w eld m icrostm cture (opposed to the bead on plate
m icrostructure) and m echanical test results will shed further
light on the sensitivity o f HI. T he results from this project will
also com plim ent results from other relevant projects outlined
by Sloan4.
Acknowledgements
T he current w ork was conducted as part o f projects sponsored
by the Cooperative R esearch Centre for W elded Structures and
the A ustralian Subm arine Corporation (ASC Pty Ltd). CRC-W S
was established and is supported by the A ustralian Governm ent
Cooperative Research Centres Program. The authors would also
like to acknowledge the in-kind support received from Bluescope
Steel, A N STO and DSTO. The authors w ould especially like to
thank Jennifer Kelly, Dave Anderson, Bob Miller, Abe Erasmus,
Chris M edcalf, Barry N ixon and M ick Fenby from A SC for the
m aterials, equipm ent and w elding expertise associated w ith the
m anufacture o f the test welds.
References
1. G .E . Linnert, W elding Metallurgy- Volume 1, American Welding Society,
4th Edition, GM L Publications, South Carolina, USA, 1994.
2. B isalloy Pty. Ltd., 2000, “W elding of Bisplate Q uench and Tempered
Steels”, Technical Brochure.
3. N. U. Ahm ed and B. L. Jarvis, “The effect o f welding conditions on the
thermal cycles in single electrode submerged arc welding”, Australasian
W elding Journal, 4 3 ,3 , 1998, pp 40-47.
4. G. W. Sloan, “Structural Integrity o f Submarine Pressure Welds”, Proceed­
ings of the 7th Interantional Conference o f W elding Research, 2005.
5. Standards A ustralia, “AS 1210:1989 - SA A Unfired Pressure Vessels
Code”, Standards Australia, Sydney, 1989.
6. Standards Australia, “AS/NZS 1554.4:2004 Structural steel welding - Weld­
ing of high strength quenched and tem pered steels” , Sydney, Australia,
2004.
7. Standards Australia, “AS/NZS 3992:1998 - Pressure Equipment: Welding
and Brazing” , Sydney, Australia, 1998.
8. J. E Lancaster, “M etallurgy o f Welding”, 6th Edition, Abington Publishing,
Cambridge, England, 1999.
9. G. Rudolph, “The influence o f weld energy input on microstructure of
Nicrofer 6020 hM o - and Nicrofer 3127 LC-shielded metal arc butt weld­
m ents”, Praktische M etallographie, 24(9), pp 416-430, 1987.