Welding

Research

Sponsored by the Welding Research Council

(WRC)

S U P P L E M E N T TO T H E W E L D I N G J O U R N A L , O C T O B E R

1972

A Simplified
Method
for Calculating
Cooling Rates in
Mild and Low
Alloy Steel
Weld Metals
BY E M I L G . S I G N E S
Inserting a thermocouple immediately behind the weld puddle during SMA W test run

Four steps are given for

predicting cooling rates
when welding by the
GMA W-argon,
GMAW-CO 2, SMAW
and SAW processes

E.G. SIGNES is Engineer. Alloy Development Section, Homer Research Laboratories, Bethlehem Steel Corp., Bethlehem, Pa.

ABSTRACT. A simple method has

been devised for predicting weld
metal cooling rates for mild and low
alloy steels. In this method the values
t o be plugged into an equation for
predicting the cooling rates are supplied by figures and tables w h i c h
summarize relationships based on
data obtained in our experimental
program in w h i c h cooling rates of
more than 3 0 0 bead-on-plate welds
were measured. In this program the
heat input was varied from 15 to 90
k J / i n . , the preheat temperature from
70 to 4 0 0 F. Beads w e r e deposited
on plates of f r o m 3 / 1 6 to 3 in. thickness w i t h the gas metal-arc welding

(GMAW) process, using both argon

and C 0 2 shielding gases, t w o different composition wires, and t w o
different composition base plates.
The submerged arc welding (SAW)
and manual shielded metal-arc w e l d ing (SMAW) processes were also
studied. Cooling rates w e r e also determined for each pass of t w o multipass welds.
W i t h i n the range studied, cooling
rates for given welding conditions
were not significantly changed by
variations in argon shielded G M A W
electrode
diameter,
base
plate
composition, or use of S M A W . However, cooling rates of G M A W - C 0 2

WELDING RESEARCH SUPPLEMENT!

473-s

C o n t r i b u t i o n s of
Studies

THICK PLATE (3D)

INTERMEDIATE-THICKNESS PLATE (2.5D)

^

-A

WELD

THIN PLATE,SURFACE TRANSFER INSIGNIFICANT (2D)

.

WELD
BEAD

VERY THIN PLATE,SURFACE TRANSFER SIGNIFICANT (2D+R)

I
m

M
II

WELD
BEAD

Fig. IFour types of heat

welds and of SAW welds were

significantly slower than the above
rates. The differences between the
bead-on-plate cooling
rates
and
those found for the multipass welds
were small.
Given the data and relationships
determined
in our
experimental
work, w e now have a method of
predicting the weld metal cooling
rate over the entire range of plate
thicknesses in mild and low alloy
steels for both argon and C 0 2
shielded G M A welding, and for submerged-arc and manual S M A w e l d ing. Furthermore, w h e n acceptable
welding conditions have been determined for one given thickness of
plate, w e can use another method,
also developed in this study, to find
either the heat input or preheat
temperature that would be required
to produce the same cooling rate in
some other thickness of plate. Taken
together, these methods comprise a
practical tool for welding engineers
concerned w i t h achieving desirable
weld properties by controlling the
w e l d metal cooling rate.
Introduction
It has long been known that w h e n
474-s | O C T O B E R 1 9 7 2

flow

low strength structural steels are

welded, both the w e l d metal and the
heat-affected zone are stronger than
the base metal through the entire
range of welding conditions. However, in the case of the higher
strength quenched and tempered
steels, the w e l d metal or heat-affected-zone, or both, may not be as
strong as the base metal if the
postweld cooling rate is too slow.
Furthermore, cracking may occur in
either zone if the cooling rate is too
fast. For these reasons, it has been
found necessary, w h e n
welding
these higher strength steels, to specify welding conditions that w i l l result in an acceptable rate of cooling.
Although many studies 1 - 1 3 have
dealt w i t h heat flow and the determination of codling rates in welding,
investigators have not as yet presented an empirically determined
method for calculating the w e l d
metal cooling rate over the entire
range of
plate thicknesses
for
G M A W , S M A W and S A W that would
require only the following input variables:
plate
thickness,
preheat
temperature, and heat input. The purpose of the present study was to develop such a method.

Previous

The effect of welding parameters

on w e l d metal cooling rates was
analyzed by Rosenthal 1 in 1 9 4 1 , and
other investigators have since then
made important contributions to this
f i e l d . 2 - 1 3 These studies established
that four different heat f l o w conditions can exist in the cooling weld,
depending on the relative plate thickness. These conditions, w h i c h are
depicted schematically in Fig. 1, are
summarized as follows:
1. Thick plates the heat is conducted approximately radially in all
directions into the plate. For brevity,
this pattern will be referred to as 3D
heat flow.
2. Intermediate-thickness
plates
given the effect of the bottom surface of the plate, the heat f l o w e n compasses a range of patterns
which
are
transitional
between
those of thin plates and thick plates.
These transitional patterns w i l l be referred to as 2.5D heat flow.
3. Thin plates the heat flow in
the plate is parallel to the plate surfaces. This pattern will be referred to
as 2D heat flow.
4. Very thin plates w h e n p!ates
are very thin, heat transfer by radiation or convection from the plates
can become a factor. For example,
w h e n very high heat inputs are used
in welding quite thin plate, surface
transfer is often significant. W e w i l l
refer to this pattern as 2D+R, i.e.,
2D+Radiation.
Rosenthal derived the following
equation for the 3D heat-flow condition:

27rK

(T-T0P

(D

= 3D cooling rate at temperature T

= thermal conductivity of the
metal
= temperature at w h i c h cooling rate is to be determined
To = preheat temperature
E = energy input to plate.
Adams 4 presented the
equation for 2D heat flow:

following

ferry0 , p2(T - T0>3

= 27rK c
ldTJ
? p " H p^ v / T

,2>

where:

TY
TT)

= 2D cooling rate at temperature T

= density of the base metal

C p = specific heat of the base

metal
p = thickness of the plate
3D RESION

Jhaveri, Moffatt, and Adams present a chart (redrawn in Fig. 2)

w h i c h indicates the cooling rate in
the 2.5D and 2D regions relative to
the cooling rate in the 3D region. Figure 3 is a chart w h i c h they developed to show the effect of radiation on cooling rate.
There are difficulties in using the
preceding analyses, however. First
of all, an essential term in the equation the energy input term is
subject to error. That is, the actual
value of this term depends on the efficiency w i t h w h i c h the energy is
transferred across the arc into the
plate, and published values for this
efficiency are not in agreement.
Thus, for the gas
tungsten-arc
(GTAW) process, arc
efficiencies
ranging from 21 to 4 8 % have been
reported; for the G M A W process and
the S M A W
process,
efficiencies
from 66 to 8 5 % have been reported;
and for submerged arc welding
(SAW) the reported range is 75 to

p Cp P 2 ( T ' T o ) ,RELATIVE

Fig. 2Relation between plate thickness and cooling rate (Ref. 5)

tions of preheat temperature

heat input in the 3D region,
determined B in the equation:

9 9 % . 7-10.14

Secondly, the physical properties

required for these equations and
charts do not lend themselves to
unequivocal
determination,
since
they vary w i t h temperature, and
during cooling the temperature in
the plate varies from T 0 to the weld
metal temperature. W h a t is more,
the variation w i t h temperature is not
a simple one, as witness the variation of specific heat w i t h temperature for a low carbon steel (Fig. 4).
In order to avoid these difficulties,
Dorschu 1 0 measured the w e l d metal
cooling rates for various combina-

THICKNESS

(T-To)

and
and

(3a)

for T = 1 0 0 0 F and 1 3 0 0 F, and found

that, respectively:
dj

= 3 7 7 x 1 0

-3

QOOpjoF

(3b)

'1000

(s)

^3(1300-T0)2
3.03x10

(3c)

\ h
where d T / d t is in deg(F)/sec. E is the

Fig. 3Relation between surface heat

transfer and cooling rate (Ref. 5)

heat input measured at the welding

head, not the actual heat input to the
plate, Ep, of Equation 1. The relationship between the t w o measures of
heat input is Ep = zE, w h e r e z = arc
efficiency. Dorschu used the G M A W
process, w i t h an 0.045 in. diam
E70S-3 electrode deposited w i t h an
a r g o n - 1 % oxygen shielding gas on
mild steel plate.
Coppolecchia
also
studied
3D
cooling in the shielded metal-arc
welding process. 14 He concluded that
arc efficiency in the S M A W process
was about 10% lower than in the
G M A W process. According to the
relationships discussed above, this
10% difference in efficiency means
that the manual w e l d would cool
about 10% faster than the gas metalarc weld.

Fig. 4Specific heat of iron vs.

tempera ture
WELDING RESEARCH SUPPLEMENT!

475-s

pass to find out whether there was

much variation from the first to the
last pass.

All-Weld-Metal Run
.045" A675 wire, argon + 1% 02 gas
Mild steel base plate
45 kji - 200F preheat/Interpass

1300F Cooling Rate

ass

dT/dt

1
2
3
4
5
6
7
8
9

4 3"
T

50.0

*
65.8
71.4
76.8
81.6
83.2
73.0

dT/dt

Pass

10
11
12
13
14
15
16
17
18

83.2
76.8
83.2
83.2
83.2
83.2
71.4
71.4
76.8

* Thermocouple melted

BACK-UP

off

Procedure Plate
.045" A632 wire, argon + 1% 0 2 gas
ASTM A517 Grade J base plate
60 kji - 70F preheat/interpass
Arc-Air Back Grooved

1300F Cooling Rate

Pass
1
2
3
4
5
6
7
8
9
10

dT/dt

83.8
76.8
71.4
76.8
76.8
83.2
76.8
69.2
66.6

Materials and Equipment

Mild steel was welded w i t h E70S3 (C-Mn-Si) electrodes. Both 0.045
and 0.062 in. diam wires were used;
shielding w a s done w i t h both argon1 % oxygen and C 0 2 gases. A proprietary (Ni-Mo-Cr-V) electrode wire
was used to weld ASTM A 5 1 7 Grade
M.
Manual
welding
(SMAW)
combinations used were E7018 electrodes deposited on mild steel and
E12018M on A 5 1 7 M. A submergedarc f l u x / f i l l e r metal combination F62EL12 was used w i t h mild steel base
plate.
Temperature measurements were
made w i t h 0.020 inch diam Pt/Pt10Rh thermocouples for all but the
submerged arc runs. In the latter
case, the fact that the slag had to be
penetrated by the
thermocouple
made the puddle diffcult to find, and
melting of the thermocouple was
common. Therefore, w e used 0.020
in. diam W - 5 R e / W - 2 6 R e thermocouples, w h i c h melt at temperatures
substantially higher than the P t / P t 10Rh couples. The cooling curves
were recorded on a Leeds and Northrup Speedomax H recorder.

* Thermocouple melted off

Fig. 5 All-weld-metal and procedure plate runs. Terms A675 and A632 wire represent
E70S-3 and a proprietary (Ni-Mo-Cr- V) bare wire electrode respectively; kji = kJ/in.

Test Program
Although these previous investigations had resulted in helpful equations, the equations as such did not
lend themselves to being integrated
into a simplified method of predicting weld metal cooling rates for a full
range of plate thicknesses. Restricting ourselves in the present test program to mild steel welds and low alloy steel welds minimized the problem of physical property variation
and therefore overcame one of the
problems in the way of developing
an equation capable of forming part
of a practical method of predicting
weld metal cooling rates.
As contrasted w i t h Dorschu and
Coppolecchia's programs at Airco,
our test program included plate thicknesses from 3 / 1 6 to 3 in. thick, thus
providing data for the various heat
flow conditions. Our testing also differed from Dorschu's study in that
the range of heat inputs was expanded
to
reflect
commercial
welding
conditions.
Although
Dorschu's work was generally restricted to heat inputs in the 10-30
k j / i n . range, heat inputs in actual
practice are usually much higher,
476-s I O C T O B E R

1 972

that is, up to 9 0 k J / i n . for G M A W

and S M A W process welds and even
higher for submerged arc welds.
Granted that for 3D heat flow the
relationships determined by Dorschu
could be expected to apply to the
entire range of heat inputs, w e designed our tests for the full range of
heat inputs not only to provide verification of previous 3D results but,
more important, to provide a realistic
data basis for the 2D and intermediate heat f l o w cases covering the
realistic range of heat inputs 15 to
90 k J / i n . In general, w e varied heat
input by changing the travel speed,
leaving current and voltage constant;
however, w e also studied the effects
of changing the heat input by varying
these settings instead of the travel
speed.
G M A W , S M A W and S A W processes were studied. In the case of
the G M A W process, the effect of
shielding gas and of electrode diameter was studied. The effect of different weld m e t a l / b a s e metal combinations was also studied for both the
G M A W and the S M A W processes.
For multipass welding, cooling
curves were determined for each

Welding and Recording Procedure

We had to make sure that the
plates were long enough and wide
enough that the plate edges did not
affect the heat flow. W e did not
determine a m i n i m u m permissible
length, but all welds w e r e made on
18 in. long plate, w h i c h is very much
longer than necessary. To determine
the m i n i m u m permissible w i d t h , w e
deposited beads on plates of various
thicknesses and widths, w i t h 9 0
k J / i n . heat input and 4 0 0 F preheat.
We found that, for plate under 1 in.
thick, a 9 in. w i d t h w a s necessary to
avoid edge effects in cooling as low
as 500 F, that is, a bead deposited
on any plate 9 in. wide or wider had
the same cooling rate for a given set
of welding parameters. For plates 1
in. thick and thicker, a 6 in. w i d t h
was sufficient. Therefore, all welds
on 3 / 1 6 , 1/4, 1 / 2 , and 3 / 4 in. plate
were made on 1 8 in. long x 9 in. wide
plate; all welds on 1, 1 V4, 11/2, 2, and
3 in. plate were made on 18 x 6 in.
plate.
All welds were made in the flat
position, in still air, on plates w i t h asrolled surfaces. Mostly single beadon-plate runs were made. As the arc
passed the mid-length of the w e l d , a
thermocouple was inserted into the
molten pool directly behind the arc.
The time-temperature cycle was recorded until the w e l d had cooled
below 500 F.

Table 1 Welding Conditions* and 1 300 F Cooling Rates for Welds in Base Program

dT/dt

Code
3 / 1 6 in . plate
|307

301
331
308
309
310
302
326
303
327
304
328
324
305
329

70
70
70
70
70
70
150
150
200
200
300
300
400
400
400

15
30
40
45
60
75
30
60
30
60
30
60
15
30
60

41 1

412

70
70
70
70
200
200
200
300
300
300
400
400
400

15
30
60
90
15
30
60
15
30
60
15
30
60

74.1
15.0
9,8
8.1
5,7
4.3
14.3
5.6
12.6
5.5
11.1
5.0
25.3
9.9
4.9

153.8
22.2
8.7
4.4
100.0
17.7
6.9
83.3
16.9
6.1
55.6
12.8
5.1

1/2 in. plate

202
201
214
203
204
205
206
207
216
208
209
217
210
211
215
212

70
70
70
70
70
200
200
200
200
300
300
300
400
400
400
400

15
30
45
60
90
15
30
60
90
30
60
90
15
30
45
60

285.7
111.1
45.5
30.8
11.4
250.0
80.0
20.8
8.1
69.0
16.0
6.3
166.7
55.6
21.1
12.8

3/4 in . plate

Q2
Q1

70
70

15
30

Code

333.3
142.9

Q10
Q3
Q4
Q5
Q6
Q7
Q12
Q8
Q13
Q9
Q11
Q14

70
70
70
200
200
200
200
300
300
400
400
400

45
60
90
15
30
60
90
30
90
30
45
90

95.2
60.6
32.3
285.7
1 17.6
45.5
26.3
95.2
18.9
87.0
47.6
1 5.6

1 in. plate
i

102
102R
101
1 16
103
104
105
105R
106
1 17
107
107R
121
1 15
108
1 18
109
120
110
11 OR
111
1 MR
119
1 12
11 2R

114

70
70
70
70
70
70
200
200
200
200
200
200
200
300
300
300
300
300
400
400
400
400
400
400
400
400

15
15
30
45
60
90
15
15
30
45
60
60
90
15
30
45
60
75
15
15
30
30
45
60
60
90

285.7
333.3
142.9
90.9
69.0
39.2
250.0
285.7
105.3

74.1
54.1
48.8
29.0
200.0
95.2
58.8
42.6
31.3
166.7
166.7
80.0
80.0
50.0
40.0
31.3
19.8

11/4 in. plate

1Q1
1Q2
1Q3
1Q4

70
70
70
70

30
45
60
90

142.9
111.1
66.7
35.1

V/z in. plate

H1
H12
H2
H3
H4

70
70
70
70
200

30
45
60
90
30

dT/dt
11/2 in.plate

3/4 in plate

1/4 in. plate

402
401
403
404
405
406
407
414
408
409
410

dT/dt

Code

153.8
87.0
66.7
45.5
125.0

H5
H15
H6
H7
H16
H8
H13
H9
H14
H11

200
200
300
300
300
400
400
400
400
400

60
90
30
60
90
30
45
60
75
90

52.6
28.6
95.2
44.4
25.0
80.0
48.8
40.8
28.6
18.2

2 in. plate

34
34R
1
1R
7
8
8R
48
35
35R
2
37
46
46R
9
3
10
38
38R
49
36
39
50
45R
11
5
43
40
40R
44
47

70
70
70
70
70
70
70
70
70
70
150
150
200
200
200
200
200
200
200
200
300
300
300
400
400
400
400
400
400
400
400

15
15
30
30
45
60
60
75
90
90
30
60
15
15
24
30
45
60
60
90
30
60
90
15
24
30
45
60
60
75
90

333.3
333.3
153.8
153.8
90.9
69.0
71.4
54.1
42.6
37,0
142.9
74.1
250.0
222.2
166,7
133,3
74.1
66.7
54.1
35.1
111.1
57.1
28.6
153.8
125.0
90.9
62.5
55 6
35.1
39.2
22.7

3 in . plate
T1
T10
T2
T7
T12
T8
T13
T9
T1 1
T4
T5

70
70
70
200
200
300
300
400
400
400
400

30
75
90
60
90
60
90
60
75
90
90

153.8
57.1
45.5
55.6
34.5
43 5
29.4
37.0
28.2
23.3
22.7

'T0= preheat temperature F. E heat input, kj in. riT dt 1 300 F cooling rate. deg(F) sec In all cases, 0 045 in E70S- 3
was deposited on mild steel, argor* * 1% oxygen gas: 27v, 250 amp, bead-on-plate welds

The temperature scale on the

chart paper was not linear; hence,
cooling rates could not be directly
determined by measuring the slope
at a given temperature. Rather, the
time for the weld to cool from 2 5 0 0
to 2000, 1500, 1400 and on d o w n to
500 F in 100 F increments was recorded; average cooling rates be-

tween various temperatures

then be determined.

could

Test Program in Detail

The first part of the program,
henceforth referred to as the base
program, consisted of empirically
determining, for mild steel, graphs
similar to those determined
by

Adams 5 (Fig. 2 and 3), w i t h only

thickness,
heat
input,
preheat
temperature, and cooling rate as
parameters. For this purpose, 145
beads were deposited, the following
ranges being used:
Plate t h i c k n e s s - 3 / 1 6 to 3 in.
Heat input-1 5 to 90 k J / i n .
Preheat t e m p e r a t u r e - 7 0 to 4 0 0 F.

WELDING RESEARCH SUPPLEMENT!

477-s

In a l l c a s e s , E 7 0 S - 3 w a s d e p o s i t e d
on mild steel. The G M A W process
w i t h a r g o n s h i e l d i n g gas w a s used,
and the heat input w a s varied by
changing the travel speed, the voltage and c u r r e n t being m a i n t a i n e d ,
at 2 7 v a n d 2 5 0 a m p , r e s p e c t i v e l y . A
d e s c r i p t i v e l i s t of t h e s e r u n s is g i v e n
in Table 1.

Table 2 Series T e s t e d in S e c o n d Phase of P r o g r a m

Series

Process

A
RB
SX
CO
70M
RBM
SAT

GMAW
GMAW
GMAW
GMAW
SMAW
SMAW
SAW

Flux or
shielding

Electrode
0.045 in.
0.045 in.
0 . 0 6 2 in.
0.045 i n .
E7018
E12018
EL12

diam
diam
diam
diam

Base metal

Argon
Argon
Argon
CO,

E70S-3
(Ni-Mo-Cr-V)
E70S-3
E70S-3

F62-EL12

Mild steel
A 5 1 7 Grade M
M i l d steel
M i l d steel
M i l d steel
A 5 1 7 Grade M
M i l d steel

T h e o b j e c t of t h e s e c o n d p h a s e of
this program w a s to determine the
e f f e c t s of p r o c e s s , e l e c t r o d e s i z e ,
shielding gas, and composition on
cooling rate. Table 2 p r e s e n t s t h e
v a r i o u s s e t s of c o n d i t i o n s t e s t e d .
Only beads w h i c h obeyed threedimensional heat transfer w e r e deposited. Each series, w i t h the except i o n of t h e s u b m e r g e d a r c s e r i e s ,
c o n s i s t e d of t h e f o l l o w i n g r u n s ;

Table 3 1 3 0 0 F C o o l i n g Rates f o r W e l d s in Series D e t a i l e d in Table 2 ( a .(b)

To

RB

SX

CO

70M

RBM

SAT

70
70
70
70
70
70
200
200
200
300
400

15
30
45
60
75
90
15
30
90
90
90

333.3
153.8
90.9
69.0
54.1
41.7
250.0
133.3
35.1
278
21.7

303.0
156.3
100.0
794
60.6
49.3
256.4
138.9
41.0
36.6
31.8

333.3
153.8
96 2
69.0
50.8
44.4
2778
123.5
39.2
29.9
26.2

285.7
122 0
89.3
52.6
41.7
33.7
232.6
91.7
290
23.9
19.4

277.8
133.3
100 0
74.1
64.5
64.5
285.7
113.5
41.5
34.5
308

285.7
1299
1250
909
71.4
606
285.7
133.3
55 6
360
309

Code
- 1
- 2
- 3
- 4
- 5
- 6
-21
-22
-26
-36
-46

Heat input.

870
52.6
42.0
32 5
25.5

800
248
16.0
14.2

(a| Bead-on-plate welds made at the following power settings Series A. RB and SX, ^ . ... ^250
v ..amp. Series CO,28 v.
230 amp; Series 70M and RBM, 24 v, 230 amp. Series SAT. 30 v. 500 amp See Table 2 for general conditions
(b}T0 = preheat temp, F; E heat input kj in dT dt 1 300 F cooling rale rieg(F) sec

Table 4 1 3 0 0 F C o o l i n g Rates f o r
W e l d s in w h i c h V o l t a g e and A m p erage w e r e Varied ( a ) l b |

Code

Tn

21
20
19
1
22
23

70
70
70
70
70
70

250

22.2
24.4
266
30.0
33.4
35.6

Amperage
varied,
ipm travel speed

amp

and

dT/dt

20
22
24
27
30
32

181.8
166.7
153.8
153.8
142.9
142.9

27

and

To

26
25
24
1
27
28
29

70
70
70
70
70
70
70

24.0
26.4
28.8
30.0
31.2
33.6
36.0

Amp

dT/dt

200
220
240
250
260
280
300

181.8
181.8
166.7
153.8
142.9
142.9
117.6

1 972

15
30
90

-36

90

.46

90

T h e travel speed in t h e 15 k J / i n .
runs w a s too fast for t h e t h e r m o couples to be i n s e r t e d in t h e s u b m e r g e d arc puddle, and for
this
reason t h e s u b m e r g e d arc
series
does not include these t w o runs.

Series of
Table 2

B, x 1 0

A
RB
SX
CO
70M
RBM
SAT

Standard
error of
estimate

2.95
3.16
294
2.34
2.87
3.18
1.74

781
5.24
4.94
5.62
790
14.4
4.63

Avg. % error
12.6
5.3
6.2
9.6
9.1
11.9
9.6

135

(a) To preheat temp. F. E heat input, kj i n . dT dt

1 300 F cooling rate. deg(F) sec
(b) In all cases bead-on-plate welds were made, using
0 0 4 5 in diam E70S-3 electrode on 2 in mild steel
plate with argon-1% oxygen shielding

478-s I O C T O B E R

-21
-22
-26

70
70
70
70
70
70
200
200
200
300
400

( B 0 forced to be zero)

dt.

dT 3D
Code

15
30
45
60
75
90

(1 3 0 0 - To )2

dT'

13 5

Volts

-1
-2
-3
-4
-5
-6

Preheat
temp, F

Table 5 Results of Regression A n a l y s i s (Part 1 a n d 2)

(Part 1]
Voltage varied,
ipm travel speed

kJ/iri.

Code

(Part 2)

Series of
Table 2
A
RB
SX
CO
70M
RBM
SAT

dt 1300

( 1 3 0 0 - T0)2
= B 0 ^ B,

(B 0 not forced

Bo

B, x 10

-10.8
-6.75
-629
-5.84
11.7
15.9
-3.53

3.29
3.18
3.13
2.53
2.50
2.68
1.85

Standard
error of
estimate

Avg. % error

5.93
5 59
3.98
509
5 50
12.8
4.55

5.9
5.4
6.5
7.4
7.2
12.5
7.5

In each case, the voltage and

amperage combination w h i c h gave
the best operating characteristics
was chosen; heat input w a s varied
by means of the travel speed. Cooling rates for the 11 runs of each of
the 7 series (except for Series SAT,
as explained above) are given in
Table 3.
In order to determine the effect of
changing the heat input by varying
the voltage and amperage, t w o short
series, detailed in Table 4, w e r e performed.
To find out w h e t h e r the cooling
rate in a multipass w e l d would vary
from pass to pass, w e determined
w e l d metal cooling curves of each
pass in both a double-vee grooved
procedure plate and a single-vee
grooved, all-weld-metal
procedure
plate. The joint configurations and
welding conditions are given in Fig.
5.
Results and Discussion
1300 F Cooling rate
The temperature scale on our
chart paper w a s not linear, so that
determination of cooling rates by
means of a tangent to the t i m e temperature curve was not possible.
The cooling rate w h i c h w e determined w a s the average cooling rate
between 1400 and 1 2 0 0 F, w h i c h is
the temperature range just above
the range in w h i c h mild steel w e l d
metal transforms on cooling. This
average cooling rate w i l l henceforth
be referred to as the 1 3 0 0 F cooling
rate. Even though they might not be
identical, the difference
between
these t w o measures of cooling rate
is quite small. This measure of cooling rate was chosen because it is
near the transformation temperature
and because it was found to correlate well w i t h weld metal strength.
The 1300 F cooling rates for each of
the series described above are given

in Tables 1, 3 and 4 for bead-on-plate

and Fig. 5 for multipass welds.
Determination of Heat Transfer
Conditions
Having determined the cooling
rate for each run, w e plotted a graph
similar to A d a m s ' first chart (Fig. 2)
in order to be able to separate 3D,
2.5D, and 2D cooling W e plotted
V, vs. H , where:

'

E (dT/dt) 13QO
(1300 - T 0 ) 2
( 1 3 0 0 - To)

and E is in k i l o j o u l e s / i n c h , T 0 in F,
d T / d t in deg(F)/second and p in
inches. Both V , and H , are n o n normalized versions of the d i m e n sionless parameters used by Adams
as, respectively, the ordinate and
abscissa in Fig. 2. All 145 runs of the
base program were represented in
this plot, w h i c h is s h o w n in Fig. 6.
This plot shows that there is no effect of thickness f o r H , > 4 0 . The 2D
cooling region should be represented
by a straight line passing through
the origin, and this is approximately
the case w h e n H , < 5. To determine
the region in w h i c h radiation is important, w e plotted Fig. 7, similar to
A d a m s ' chart s h o w n in Fig. 3. Here
w e plotted V 2 vs. H , w h e r e :
p;(1300-T0P
V, =

E* (dT/dt), 30o

all runs w h e r e H 2 <3.5 (and these

points have been omitted from Fig. 6,
since this figure represents only those
cases w h e r e radiation is insignificant).
Note that K, P, and C p are not included in these parameters, since
their values would not be expected to
vary from one grade to another w i t h in the range of steels included in this
study.
In summary, f r o m these curves w e
have determined the following:
For 3D cooling
2.5D cooling
2D
cooling
2D+R
cooling

H,
5
H,
H,

>40
<H, <40
<5, H;>3.5
< 5 , H, < 3 . 5

The 3 D / 2 . 5 D and 2D/2D+R crossover points have been used to construct Fig. 8. By using this figure, one
can rapidly determine w h i c h heat
flow condition exists for any given
set of parameters. This figure also
indicates that the cooling condition
is a function of the welding condition
in addition to thickness. This plot
does not distinguish between the 2D.
and 2.5D regions, but this distinction
is not necessary in the method, detailed instructions for w h i c h are
given in the section on Method for
Determining Cooling Rate.
Determination of 3D Equation
Base Program. Of the 145 runs in
the base program, 70 were found to
obey 3D cooling behavior at 1300 F.
Using the method of stepwise linear
regression, w e determined that the
coefficient B, in the following equation is 3.12 x 10- 3 :

p?(1200-T0)
E(p),/2
and V 2 and H2 are non-normalized
versions of the dimensionless parameters of Fig. 3. From this figure it is
evident that radiation is important for

( 1 3 0 0 ^

dT
dt/

'

1300

Comparison
of
Processes.
As
stated previously, only welding condi-

.005
,2.004

..003

> .001

10

20

30
40
50
H, = p 2 (l300-To) / E

60

Fig. 6 Representation of data for 1300F cooling.

Presentation form is similar to that in Fig. 2

Fig. 7 Representation of data for 1300F cooling.

Presentation form is similar to that in Fig. 3
WELDING RESEARCH SUPPLEMENT!

479-s

* *

where B 0 , theoretically zero, is the

value of (dT/dt) ,3oo for (1 300-T 0 )2 / E
= 0. To determine the practical validity of this assumption, w e also determined both B 0 and B, in Equation 6;
the results appear in Table 5. If w e
look at the columns labeled "Standard
Error of Estimate" and "Average %
Error," w e will notice that, except
for Series A, the latter analysis is only
slightly more accurate than the former. That is, the error to be expected
w h e n using Equation 5 w i l l generally
not be much more than that from
Equation 6. Consequently, w e are
justified in continuing to use Equation
5 w h i c h represents a great deal of
convenience at the sacrifice of only a
small amount of accuracy. From now
on, therefore, w h e n w e refer to the
coefficient B, for a given series w e
will be referring to that given in
Table 5 (Part 1).

^v *.. ^

.0
1.5
2.0
PLATE THICKNESS,INCHES
F/g, S A/eaf transfer condition for any combination of plate thickness
heat input, and preheat temperature

tions obeying 3D heat transfer were

used in this section of the study. This
restriction assumes that the relationships between the cooling rates of
the various series determined for 3D
heat transfer would also hold for the
other heat transfer modes. Furthermore in determining the 3D equations in this section, w e excluded the
runs for w h i c h E = 1 5 k j / i n . for the
following reasons:
1. The travel speed in the 15
k J / i n . runs was too fast for the
thermocouple to be inserted in the
submerged arc puddle. Thus, by
considering only those runs from 3 0
to 9 0 k J / i n . , w e could compare all
series on an equal basis.
2. As previously explained, the
1300 F cooling rate was determined
by measuring the time to cool from
1400 to 1 2 0 0 F. In the case of the
15 k J / i n . runs, this amount of time
480-s I O C T O B E R

1 972

was quite small, and an accurate

measurement was quite difficult.
These reasons justified eliminating
the 1 5 k J / i n . data, the remaining
range of 3 0 - 9 0 k j / i n . being in any
event fully representative of practical
welding conditions.
The coefficients B , in the equations
(1300-T0)2

(5)

were determined using the data in

Table 3. This was done by assuming
the value of zero for coefficient B 0 in
the equation

B 0 + B,

(6)

To determine whether there were

any significant differences between
the different series, w e employed a
statistical analysis known as Hald's
method. 1 5 This analysis
revealed
that, at the 9 5 % significance level,
there was no statistical difference
between the coefficients Bi determined for all of the G M A W - a r g o n
and S M A W series (A, RB, SX, 7 0 M ,
and RBM). On the other hand, the
BTs for the series CO and SAT
(GMAW-C02
and
SAW)
were
significantly different from these
B,'s and from each other. The standard error of the RBM series, however, was found to be statistically different from most of the standard errors of estimate in the other series.
This difference can be attributed
partly to the fact that the t w o biggest
deviations from the estimated curve
occurred in t w o of the runs w i t h fast
cooling rates (RBM 2 and RBM 3).
Although the percentage deviation
from the curve might be the same at
these high values, the residual itself
would be high, thus tending to
increase the standard error of estimate. Furthermore, w e would expect
the manual process to be less reproducible than either of the automatic
processes. For example, although
the standard error of estimate of
series 7 0 M is substantially lower
than that of series RBM, it is the second highest in the group.
Since the coefficients B, of series
A, RB, SX, 7 0 M , and RBM were not
found to differ significantly from one
another, w e lumped all the runs from
these series together and determined an average coefficient B,,
w h i c h turned out to be 3.02 x 10~ 3 .
The average % error w a s 9.9%.
Using the cooling rate data in Table 4, w e determined B, in Equation
1 to be 2.86 x 10~ 3 in the case
w h e r e the heat input w a s changed
by varying the voltage. However, in

p = thickness, inches
To= preheat temperature,F
E = heat input,kilo Joules / i n c h

20

p= thickness, inches
To= preheat temperature,F
E = heat input,kilo joules/inch
2
3
H2=p2(l300-To)/E/'p-

30

H, = p * ( 0 O O - T o ) / E

Fig. 9 Correction factor'P], vs. the parameter

f/,
valid for all cases where radiation is
insignificant

this case the heat input

range
studied w a s very small, 22.2 to 35.6
k J / i n . , and forcing the equation
through zero would artificially force
B, to be approximately equal to the
B,
previously
determined.
Here,
allowing B 0 to vary, as in Equation 6,
the slope B, w a s found to be 1.45 x
10" 3 , substantially different
from
that previously determined. This indicates that varying heat input by
changing the voltage has a much
smaller effect than varying it by
changing the travel speed. Low voltage runs w e r e found to have lower
cooling rates, w h i l e high voltage
runs had cooling rates higher than
those predicted by B, = 2.86 x 10~ 3 .
Since the electrode to plate distance
decreases as the voltage decreases,
w e would expect the arc efficiency to
increase as the voltage decreases.
This would explain the above behavior.
However, the voltage range, 2 0 - 3 2
v, used in this series, w a s far greater
than that w h i c h would be used in
practice. For our setup, for example,
good welding conditions could only
be obtained for voltage of about 2 5 28 v. Furthermore, the greatest variation of any cooling rate in this series
from that predicted by B, =3.02 x
1 0 " 3 (the average coefficient determined above) was 15%. For these
reasons, these results w i l l not prevent us from using this equation.
Using the cooling rate data in Table 4, w e determined B, in Equation
1 to be 3.03 x 1 0 " 3 in the case
w h e r e the heat input was changed
by varying the current. Furthermore,
w h e n w e allowed B 0 to vary, B, in
Equation 2 was found to be 3.06 x
10~ 3 . These results s h o w that varying the heat input by means of t h e
amperage gives results identical to
those obtained w h e n varying the
heat input by means of the travel
speed.

This figure

Fig. 10 Correction factor'P2. vs. the parameter Hi.

takes into account the effect of radiation on cooling

is

Determination of 2 D

Equation

Although the method to be described later does not require the use
of the 2D heat f l o w equation, the
coefficient B, in the equation
p2(1300E2

(7)

was calculated for the runs in the

base program and found to be 2.44 x
10- 4 .
Multipass
Welds. Since a hemispherical w e l d bead emitting heat
into the plate in all directions is
closely approximated only by a beadon-plate w e l d and since varying
geometry can be expected to affect
the cooling rates of the various
passes in multipass welds,
we
measured the cooling rate of each
pass in both a single-vee groove and
a double-vee groove w e l d , as previously described. Figure 5 gives the
1300 F cooling rate for each run.
We can see from these data that
in both the single and double vee
cases there is some difference between the maximum and the m i n imum cooling rate w i t h i n each plate.
In the case of the
double-vee
grooved
procedure
plate,
the
measured cooling rates varied from
67 to 83 F/sec, the average being
76 F/sec. For the single vee plate,
the cooling rates varied from 5 0 to
83 F/sec. However, this variation
includes the first.three passes w h i c h
were greatly influenced by the Vi in.
root opening, a gap that is rarely
used in anything but all-weld-metal
specimens. When
these
three
passes
are
eliminated
from
consideration, the range becomes
71 to 8 3 F/sec, w i t h an average of
78 F/sec.
In both cases the predicted beadon-plate cooling rate w o u l d have
been 72 F/sec, or s o m e w h a t slower

This

figure

than the measured 76 and 78 F/sec

averages. The slightly faster cooling
rates in the actual weld tests is attributable to the fact that there is
more volume into w h i c h the heat
can diffuse from the interior beads,
w h i c h appear to cool faster than the
exterior beads. For all but the t h i n nest plate, the actual average cooling rates in multipass butt welds w i l l
be a little higher than our predicted
rates. However, such
differences
are, as the examples cited show,
small enough to permit the practical
use of predictions by our method.
Therefore, only the bead-on-plate
cooling rate measurements w i l l be
used in w h a t follows.
Differences Between Our Results and
Those of Previous Investigators

Although results obtained in this

study agree in many respects w i t h
those published elsewhere, our f i n d ings show some significant differences.
Our values of B, in the equation

= B

(1300-T 0 ) 2

(8)

1300

are very close to Dorschu's experimental value of 3.03 x 10~ 3 for E70S3 on mild steel. However, the value of
B, determined by Coppolecchia for
S M A W (manual) is 3.42 x 10" 3 , from
w h i c h he concludes that the S M A W
process is less efficient than the
G M A W process, a result not substantiated by our results. One possible reason for his conclusion may
have to do w i t h the fact that he used
E14018 electrodes and HY100 base
plate, in contrast to the
lower
strength
materials
of
Dorschu's
work. In our work, although there
was no statistical difference bet w e e n cooling rates in low and high
strength steel w e l d metal, numerical

WELDING RESEARCH SUPPLEMENT!

481-s

are substantially too high. In fact,

assuming a submerged arc efficiency
of 1.0 would give z = 0.57 for the
GMAW-argon
and S M A W
processes, and z = 0.74 for G M A W - C 0 2 .

jo-

M e t h o d for D e t e r m i n i n g
Cooling Rate
Incorporating the choice of equations based on our experimental
results analyzed in the preceding section, w e developed the following
four-step method for determining the
weld metal cooling rate in any thickness plate of mild or l o w alloy steel
butt welded by the G M A W , S M A W
or S A W processes. The 1300 F cooling rate was chosen, as explained before, because it is near the transformation temperature and because
it was found to correlate well w i t h
weld metal strength.

ss

-.06

=04

p - thickness , inches
To = preheat temperature,F
E = heat input, kilo j o u l e s / i n c h

02-

-rr- cooling rate,F/second

15

25

20

H, P z ( 1 3 0 0 - T o )

30

35

I. Determine heat transfer

tion from Fig. 8.

40

/E

II. Determine the value of </>, (ratio of actual cooling rate to 3-D
cooling rate) and </>2(ratio of cooling rate without radiation to actual cooling rate) in accordance
w i t h the proper heat transfer condition.

Fig. 11 Plot for determining heat input. This curve

is valid for all cases where radiation is insignificant

differences
did
exist.
For
both
G M A W and S M A W , the
higher
strength weld metal/base
metal
combinations
cooled
somewhat
more rapidly than the lower strength
combinations. Both E14018
and
HY100 are more highly alloyed than
even E12018 and A 5 1 7 Grade M;
the presence of a higher alloy content may account for the higher
coefficient B,. If increased alloy content has such an effect, it may become
necessary
to
determine
composition
ranges
over
which
given equations can be applied w i t h
reasonable accuracy.
That the submerged arc process
exhibits slower cooling rates than
the other processes w a s expected
and has been reported by others. 8
Submergence of the arc leads to less
heat loss in the arc and therefore a
higher heat transfer efficiency. Part
of the reason for the slower cooling
in C 0 2 welding may also be the increased heat transfer efficiency in
this process. For example, it is
pointed out in the Welding Handbook 16 that "radiation loss of energy
may be over 2 0 % of the total input in
the case of argon welding arcs,
w h i l e in other welding gases the
radiation loss is not more than about
10%."
Dorschu and Coppolecchia both
estimated the arc efficiency (fraction
of input heat that reaches the plate)
by assuming a value of thermal
conductivity K = 0.068
cal/cmdeg(F)-sec in the modified Rosenthal
equation
482-s I O C T O B E R

1 972

condi-

(1300-T0)2
2TTK

ZE

Heat transfer
Condition

(9)

<t>2

"3D
~~
2.5 D, 2D

where z = arc efficiency. For G M A W

Dorschu's value for z was 0.84, and
for S M A W Coppolecchia obtained
the value of 0.74 for z. However,
using the same formula to derive submerged arc efficiency would give us
an efficiency of 1.45, w h i c h would
indicate that the values given above

From
Fig. 9
From
Fig. 9

2D + R a d i ation

1
From
Fig. 1 0

Note that Fig. 9 and 10 w e r e

constructed
by fitting
curves
to the data in Fig. 6 and 7 and
normalizing the ordinates.

/
/

r 2
xi I ~2

yS
y ^

a.' UJ

^ ^ p
^

^ * ^

= thickness, inches
To = preheat temperature,F
E * h e a t i n p u t , k i l o j o u l e s / inch

^ ^

T7 = cooling r a t e , F/second

^ ^ ^ ^

15

20

25

30

= p2 (1300-To) / E

Fig. 12 Plot for determining preheat temperature. This curve

is valid for all cases where radiation is insignificant

35

40

Specify
<t>3 (ratio of actual
cooling rate to cooling rate for
G M A W - a r g o n process) according to the particular welding
process under consideration.

0.5

0.4

a.
1.0
0.77
1.0
0.57

G M A W , argon shielded
G M A W , CO 2 shielded
S M A W (manual)
SAW

0.3

-0.2

IV. Calculate
follows:

the

cooling

rate

as
Q.JUJ
p = thickness , inches

= 3.02x10~ -

<M3

(1300-T0)

(10)

0.1-

To = preheat temperature,F
E=heat i n p u t , k i l o j o u l e s / i n c h

%- = cooling r a t e , F/second
at

IT'

where T 0 = preheat temperature, F; E=heat input, k j / i n ; and

the left side of the equation
equals the cooling rate at 1300 F
in deg (F)/sec.

Fig. 13Detail

II. Determine
H, from Fig. 1 1 .
If H , > 4 0 , go on to Step III. If
H<C40, go on to Step IV.

If acceptable welding conditions

have already been determined for
one given thickness of plate, the
data developed in this study can also
be used to determine either the heat
input or preheat temperature that
would be required to produce the
same cooling rate in some other
thickness of plate. To do so, one first
calculates the cooling rate for the
thickness of plate for w h i c h the w e l d ing conditions are k n o w n , and then
proceeds
with
the
methods
described below. These methods are
based on Fig. 11 and 12, w h i c h have
been generated, as w e r e earlier figures, by combining our data w i t h
Adams' figures (in this case, Fig. 2
and 3 of Reference 5).

III. For H, > 4 0 , calculate

input as follows:

To determine the heat input for a

given
plate
thickness,
preheat
temperature, and cooling rate:
I. Specify 4>3 (see Step III of section entitled M e t h o d for Determining Cooling Rate).

10

of Fig. 12 for VT <O.S

Methods for Determining

Heat Input and
Preheat Temperature

It should be noted that these methods are valid for only <2 = 1, that is,
where radiation is insignificant. As
s h o w n in Fig. 8, this is the case in
virtually all plates Vi in. thick and
thicker. If this method is used to
determine the heat input or preheat
temperature for thinner plate, the result should be checked w i t h Fig. 8 to
make sure it falls outside the "radiation significant" range. If it does not,
the actual heat input or preheat
temperature
will
be
somewhat
greater than that predicted.

H, = p 2 ( 1 3 0 0 - T o ) / E

E = 3.02x10-3

the

heat

(130-T>2
< d T / d <), 3 oo

IV. For H,<40, calculate the heat input as follows:

P2(1300-T0)
E

"

H,

where p = plate thickness, inches

To determine the preheat

temperature for a given plate thickness,
heat
input, and cooling rate:
I. Specify
$ 3 (see Step III of
section entitled Method for Determining Cooling Rate).
II. Determine
H,
from
Fig.
12, or, if V T (see Fig. 12) is less
than 0.5, f r o m Fig. 13. If H,>40,
go on to Step III. If H,<40, go on
to Step IV.
III. For H,>40, calculate the preheat
temperature as follows:

Conclusions
On the basis of an experimental
program to provide weld metal cooling rate data from tests w i t h both
mild and low alloy steels and a wide
range of thicknesses, heat inputs,
and preheat temperatures representative of commercial practice, w e
developed a simple method for predicting the w e l d metal cooling rate
from a given set of welding and heat
transfer conditions.
W i t h i n the range of conditions
studied, cooling rates for
given
welding conditions were not significantly changed by variations in
argon-shielded
GMAW
electrode
diameter, base metal composition, or
use of manual S M A W . However,
cooling rates of welds made by the
G M A W - C 0 2 and SAW
processes
were significantly slower.
A cknowledgments
The author wishes to thank P. Howe
for his help in developing the experimental technique and in performing the
experimental work, and G. M. Busch III
for his assistance in the manual welding
portion of the program. The editoral comments of B S. Mikofsky are greatly
appreciated
References

332E
1300

IV. For H,<40, calculate the preheat

temperature as follows:
Tn = 1 30 0

H.E

1. Rosenthal, D.,
"Mathematical
Theory of Heat Distribution During Welding and Cutting," Welding Journal, 20 (5),
May 1941, pp. 220s-225s.
2. Rykalin, N. N., Calculation of Heat
Flow in Welding, Hayah Academy,
U.S.S.R., 1951, Translated by Zvi Paley
and C. M. Adams, Jr
3. Wells, A. A., "Heat Flow in Welding," Welding Journal, 31 (5), May 1952,
pp, 263s-267s.

WELDING RESEARCH SUPPLEMENT!

483-s

4. Adams, C. M. Jr., "Cooling Rates

and Peak Temperatures in Fusion W e l d i n g , " Welding Journal, 37 (5), May 1958,
pp. 210s-215s.
5. Jhaveri, P., Moffatt, W. G., and
Adams, C. M, Jr., "The Effect of Plate
Thickness and Radiation on Heat Flow in
Welding and C u t t i n g , " Welding
Journal,
41 (1), January 1962, pp. 12s-16s.
6. Barry, J . M., Paley, Z., and A d a m s ,
C. M. Jr., "Heat Conduction from Moving
Arcs in W e l d i n g , " Welding Journal,
42
(3), March 1963, pp. 9 7 s - 1 0 4 s .
7. Paley, Z., Lynch, J . N., and A d a m s ,
C. M. Jr., "Heat Flow in Welding Heavy
Steel Plate," Welding
Journal.
43 (2),
Feb. 1964, pp. 71s-79s.
8. Christensen, N., Davies, V. de L. and
Gjermundsen,
K.,
"Distribution
of
Temperatures in Arc W e l d i n g , "
Brit.

Weld. J., Feb. 1 965, pp. 5 4 - 7 5 .

9. Myers, P. S., Uyehara, 0 . A., and
Borman, G L , "Fundamentals of Heat
Flow in W e l d i n g , " Welding
Research
Council Bulletin #123, July 1 967.
10. Dorschu, K. E., "Control of Cooling
Rates in Steel Weld M e t a l , "
We/ding
Journal, Al (2). January 1968, pp. 49s62s.
1 1 . LaFrance,
M.,
Prudhomme,
M.,
Murry, G., Constant, A., "Prevision de la
Durete sous Cordon d Aciers A 52 a Partir
de
I'Etude des
Transformations
de
TAustenite et en Fonction des Conditions
de Soudage M a n u e l , " (Prediction
of
Underbead Hardness of A 52 Steels on
the Basis of the Study of Austenite Transformations and As a Function of Manual
Welding Conditions), Rev. Met.,
June
1968, pp. 4 1 7 - 4 2 5 .

12. Bromage, K., " A r c Efficiency and

Heat Flow in Inert-Gas W e l d s , " Brit.
Weld. J., October 1968, pp. 4 9 3 - 5 0 0 .
13. Bradstreet, B. J., "Effect of Welding
Conditions on Cooling Rate and Hardness
in the
Heat-Affected-Zone,"
Welding
Journal.
48 (11), November 1969, pp.
499s-504s.
14. Coppolecchia, V. D., "Covered Electrode W e l d i n g , The Effects of Welding
Variables on Initial W e l d Metal Cooling
Rates," Airco Research and
Engineering
Department
Report RE-69,
106-CRE-44,
September 15, 1969.
15. Hald, A., Statistical
Theory With Engineering
Applications,
J o h n W i l e y and
Sons, New York, 1 9 5 2 , p. 5 7 1 .
16. A m e r i c a n Welding Society, We/ding
Handbook, Sixth Edition, Section One, p.
3.22

(1) Sensitivity of the Delta Test t o Steel

C o m p o s i t i o n s and Variables
by L. J. McGeady

The introduction and use of higher strength heat-treated steels have demonstrated the need for awareness of weldability and fracture problems in the total
composite of weld metal, heat-affected zone and plate material. Hence there has
developed need for an appropriate test specimen and procedure applicable to the
total composite weldment providing the opportunity for failure in any area. This
report describes a specimen applicable to this need and to present data to determine whether the proposed specimen, the Delta, allows failures to follow leastresistant paths because of specimen geometry and loading system. It is not the
purpose to recommend materials, welding procedures or processes, though it has
been necessary to study many of these in a wide variety of combinations to determine their influences on behavior of the specimen. The work reported in this
paper was sponsored by the Pressure Vessel Research Committee of the Welding
Research Council.

WRC
Bulletin
No. 172
May 1972

(2) Experimental Stress Analysis and Fracture

Behavior of Delta S p e c i m e n s
by J. M. Barsom

This investigation was undertaken to analyze the stress distribution in the

Delta specimen and to investigate the possible effect of the stress distribution on
the flow and fracture behavior of the Delta specimen.
The price of WRC Bulletin
should be sent to the American
33125. Orders for bulk lots, 10
Research Council, 34.5 East 47th
484-s | O C T O B E R

1 972

172 is $3.00 per copy. Orders for single copies

Welding Society, 2501 N.W. 7th St., Miami, Fla.
or more copies, should be sent to the Welding
Street, New York, N.Y. 10017.