International Congress on Advances in Welding Science and Technology for

Construction, Energy and Transportation Systems (AWST - 2011)

24-25 October 2011, Antalya, Turkey
AWST-11/12

Improvement of Cold Crack Resistance by Controlling

the Weld Metal Microstructure
H. J. Kim
Korea Institute of Industrial Technology, 35-3, Hongchonri, Ibjungmyun, Seobuk-gu, Chonansi, 331-825, Korea
kimhj@kitech.re.kr

Abstract
High strength weld metal consists of various
microstructural constituents such as grain boundary
ferrite, acicular ferrite, ferrite side plate etc. Among
them, acicular ferrite has been well documented to be
beneficial for improving both strength and toughness but
others to be detrimental to them. However, the effect of
those constituents on the cold crack susceptibility has not
been clearly understood. This paper was prepared to
emphasize the importance of microstructural contribution
on the hydrogen-induced cold cracking (HICC) that
occurs in the weld metals. Two FCA welding wires were
prepared to have same level of tensile strength but to
have different weld microstructure by controlling the Ni
content in the wire. HICC susceptibility of those two
consumables was evaluated by multi-pass weld metal
cold cracking test. The diffusible hydrogen content was
varied by adding different amount of hydrogen gas in the
shielding gas. As a result of this study, it was
demonstrated that microstructural modification with the
decreased proportion of grain boundary ferrite and thus
the increase of acicular ferrite improved cold crack
resistance of weld metal significantly and resulted in the
accomplishment of preheat-free welding for 600 Mpa
strength level at the diffusible hydrogen level of 4.5
ml/100g. In the discussion, the beneficial effect of
acicular ferrite against HICC will be addressed based on
the microstructural characteristics of acicular ferrite.
Keywords: High Strength Weld Metal, Cold Cracking,
Weld Microstructure, Grain Boundary Ferrite, Acicular
Ferrite, Diffusible Hydrogen

1. Introduction
Due to the improved resistance to HAZ cracking
achieved in the modern high strength steels such as
TMCP steels and HSLA steels, their weld metals become
more prone to suffer from cold cracking problems than
the HAZ of base metal does. In particular, weld metal
cracking (WM cracking) has been reported to become a
major limiting factor in the preheat-free application of
advanced high strength steels [1]-[3]. Accordingly, the
cold cracking phenomena in high strength weld metals

has received more increasing attention recently [4]-[7].

Nevertheless, the microstructural factors relating with
WM cracking have not been well understood.
It has been generally accepted that cold cracking in HAZ
occurs given the co-existence of a sufficient quantity of
diffusible hydrogen (HD), a susceptible microstructure
and a tensile residual stress. To quantify the
susceptibility of HAZ microstructure, various carbon
equivalent (CE) formulae such as CE(IIW) and P cm have
been suggested and Pcm has been widely used for low
carbon steels. Considering all those three factors, the
minimum preheat temperature necessary to prevent HAZ
cracking (THAZ) has been derived from the result of ygroove cracking test [8] and then formulated as an
empirical equation (1) shown below.
THAZ(0C)=1440(Pcm+HD/60+ t/600) 392
(1)
where Pcm=C+Si/30+(Mn+Cu+Cr)/20+Ni/60+Mo/15+
V/10+5B, HD is the diffusible hydrogen content in
ml/100g, and t is the plate thickness in mm.
Hence, in above equation, the three major factors are
counted in with the terms that can be easily measured or
calculated. Of particular interest, however, was to
whether the Pcm value of weld deposit can used as the
one of the parameters for estimating the cold crack
susceptibility of multi-pass weld deposits.
Although WM cracking generally follows the same basic
rules as that of HAZ, the quantified parameter that can be
employed for evaluating microstructural effect on weld
metal cracking are relatively scare even though, using the
gapped bead-on-plate (G-BOP) test, some chemical
parameters has been proposed by many investigators [9][11]. It is presumably because the relation between weld
microstructure and WM cracking has not been fully
understood.
In 1980s, several researchers [12],[13] mostly in Japan
have studied on possible factors that would control WM
cracking in the multi-pass welds and, as for parameters,
they have selected the weld metal strength, HD content
and the height of weld metal. The preheat temperature
necessary to prevent WM cracking (T WM) was then
determined as a function of those parameters and the
following empirical equation (2) was given;

TWM(0C) = ARm + Blog(HD)+Chw+D

(2)

where Rm is the weld metal tensile strength in MPa, HD

is the diffusible hydrogen content in ml/100g, hw is the
weld metal height in mm, and A, B, C and D are
constants.
In equation (2), there are three parameters that are
controlling TWM. Of those three, the most relevant
parameter that can count for microstructural effect would
be the term of Rm as the weld metal strength strongly
depends on the weld microstructure.
From the above two equations, therefore, it can be easily
understood that the quantified parameter used for
microstructural effect is quite different between THAZ and
TWM. Furthermore, equation (2) suggests that, regardless
of weld metal microstructure, TWM of weld metals would
be the same if the weld metal strengths are identical. In
this respect, it was of interest to confirm if Rm is a
reliable measure in evaluating weld metal microstructure
with respect to WM cold crack susceptibility. If it turned
out not to be the case, a new parameter needs to be
proposed based on the more direct relationship between
weld microstructure and WM cracking.

Such a geometry of test specimen is similar to that has

been used by previous investigators [12,13], which was
specially designed to access the susceptibility to cold
cracking in multi-pass weld metal.
Table 1. Chemical composition of weld metals and base metal.

Materials
0Ni WM

C
.05

Si
.55

Mn
1.17

Ni
0.0

Cr
.03

1.5Ni WM

.05

.42

1.37

1.5

.03

BM

.08

.30

1.54

Steel plate of 80mm thick with a groove of 50mm deep

was also used to figure out the susceptible
microstructural constituent. The welding condition
selected for test weld beads was the one that typically
applied in actual fabrication with 1.4 mm FCAW wire
and with 100% CO2 gas shielding, i. e. 320 A - 32 V - 35
cm/min with a contact tube-to-work distance (CTWD) of
25 mm. The experimental variable was preheating
temperature which was preset as 40, 65, and 100 0C.
Those temperatures were maintained as inter-pass
temperature within a range of 50C.

Therefore, in the present study, it was further aimed to

identify and evaluate the effect of each constituent on
cold crack susceptibility of weld metals, and then to give
a basic guideline for designing new welding consumables
with improved resistance to WM clod cracking from the
microstructural point of view.

2. Experimental Procedure
For evaluating the microstructural aspect of weld metal
cold cracking in the 500 MPa strength level, two
different flux cored wires specified as AWS E71T-1 (1.4
mm in diameter) were employed. Both of them are
similar in composition except in Ni content. The FCAW
weld metal made with a commercial wire was called as
0%Ni-FCAW as it contains nil in Ni. The experimental
weld deposits made with new design was referred to as
1.5%Ni-FCAW as it contained about Ni of about 1.5%.
Table 1 shows the chemical compositions of their weld
metals deposited in the multi-pass weld metal cracking
test specimens described in the next section. Each set of
wires were fabricated to have various levels of HD by
changing the moisture content of the flux within the
wires; that is, by changing the sintering temperature of
some of the flux ingredients. After fabrication, HD
content of weld metal was measured by gas
chromatograph method following the procedure of AWS
specification [14].
Multi-pass weld metal cracking test was conducted using
specimens shown in Fig. 1. Steel plates of 50 mm
thickness were machined to have a 25 mm deep Vgroove with 60-deg included angle and then were
restraint-welded on the H-type jig made with 25 mm
thick plate in order to prevent any distortion of the plate.

Figure 1. Geometry of weld specimen for multi-pass weld metal

cracking test.

3. Results
3.1 Weld metal microstructure
As expected, microstructure of as-deposited FCAW weld
metals was composed with three major constituents such
as grain boundary ferrite (GF), acicular ferrite (AF) and
ferrite with second phase (FS) but the proportion of each
constituent is quite different between 0%Ni and 1.5%Ni
weld metals. Quantitative analysis preformed by point
counting method [15] showed that the weld
microstructure made by 0%Ni wire contains GF of 19%
and AF of 54%. With the addition of 1.5% Ni, the
proportion of GF dropped down to 6% and AF increased
to over 80%. The balance of the microstructure in each
case was found to be predominantly FS. Mechanical tests
have been conducted for both wires using all weld metal
test coupons and their results were quoted in Table 2.

International Congress on Advances in Welding Science and Technology for

Construction, Energy and Transportation Systems (AWST - 2011)
24-25 October 2011, Antalya, Turkey
AWST-11/12
Table 2. Results of mechanical tests.

Material

Hardness (HRB)

Yield St.
(MPa)

Tensile St.
(MPa)

0%Ni

95.3

579

611

1.5%Ni

97.7

610

628

Charpy Impact Energy (J)

Material

00C

-200C

-400C

0%Ni

58

57

27

1.5%Ni

94

97

90

magnification examination that the cold crack tends to

propagate through GF phase as shown in Fig. 4. Such
characteristics has been well documented by many
investigators for the cracks found mostly in weld metals
made by submerged arc welding (SAW) process, and has
been called as Chevron cracking [16]-[20]. All of the
defects examined in the other specimens showed the
same characteristics and thus it was concluded that the
GF is the one that is most susceptible to cold cracking in
the ferritic weld. This conclusion implies that the cold
crack resistance of weld metal can be improved by
decreasing the proportion of GF (or by increasing the
proportion of AF) in weld microstructure, as has been
made in the 1.5%Ni weld.

As can be expected from weld metal microstructure, the

1.5%Ni weld was slightly higher in hardness and tensile
strength, and showed substantially high impact toughness
compared with the 0%Ni deposit. High impact toughness
of 1.5%Ni weld is obviously due to the higher proportion
of AF and also to the higher Ni content than those in
0%Ni weld. As these two welds were significantly
different in weld microstructure, these welds were
confirmed to be appropriate for pursuing the purpose of
present study; i.e. evaluating microstructural effect on
cold cracking susceptibility of ferritic weld metals.

3.2. Susceptible microstructure

In the ultrasonic examination for the weld metal cracking
test specimen, it was found that 0Ni weld metal was
quite susceptible to cold cracking. Fig. 2 (a) shows the
macrostructure taken from transverse cross-section of the
weld that was deposited with 0%Ni wire in the 80mm
thick plate and Fig. 2(b) shows the two-dimensional
image of cold cracks revealed by scanning acoustic
microscope (SAM) in the same specimen. From these
results, it can be realized that this weld has many cold
cracks and the most of cracks are located in the regions
having columnar grain structure. This fact becomes
clearer in the examination of longitudinal section that is
presented in Fig. 3.
Fig. 3 shows the longitudinal cross section cut through
one of the SAM defects revealed in Fig. 2(b). As can be
seen in Fig. 3, the SAM defect turns out to be a crack
that is inclined about 45 deg to the welding direction and
propagates in stepwise manner. One thing to note here is
that this crack indeed confines in the region of asdeposited microstructure and appears to stop its
propagation when it reaches the region of grain-refined
microstructure developed by the subsequent pass. This
fact confirms the SAM result and implies that the asdeposited columnar grain structure is more susceptible to
cold cracking than the refined microstructure made by
subsequent passes. It was further noted in high

Figure 2. Result of cold cracking test performed in 80mm thick

plate with 0%Ni wire: (a) macrostructure and (b) cold
crack image revealed by SAM.

3.3 Result of weld metal cracking test

Using the seven wires fabricated to have different in Ni
and HD contents, multi-pass weld metal cracking test
was conducted with 50mm thick plate at various
preheating temperatures and the welded plates were
inspected by UT. Upon UT inspection, firstly defect-free
specimens were sorted out and then for the specimens
containing defects, the number of UT indications was
counted.

Welding direction

specimens were cracked regardless of HD content with

the number of UT indications tending to increase with
HD content. However, cold cracking could be prevented
with increasing the preheating temperature to 650C at HD
of 4.3 and 4.5 ml/100g but could not be avoided at the
higher level of HD content. At the highest level of HD
content, cold cracking occurs at 650C but the number of
UT indication decreases with preheating temperature
resulting in crack-free weld at 1000C. Therefore, the
border line that divides cracking (C) and no-cracking
(NC) conditions could be drawn as in this figure. This
C/NC boundary curve clearly shows the effect of HD
content on the preheating temperature being increased
with HD content over 5 ml/100g.

Figure 3. Weld metal cold crack shown in the longitudinal cross

section.

Figure 5. Result of multi-pass weld metal cracking test for 0%Ni

Wires

Figure 4. Optical microstructure showing the cold crack route

with respect to weld microstructure

In the specimen welded with 0 %Ni wire having HD of

4.3 ml/100g at preheating temperature of 400C, for
example, a total of 8 UT indications were recorded in the
full length of weld. As the preheating temperature
increases to 650C, no cold cracks were found. Other
wires of 0%Ni chemistry were also tested at various
preheating temperatures and their UT results are
summarized in Fig. 5. This figure presents the status of
cold cracking occurrence along with the number of UT
indications that can represent the comparative risk in
cold cracking at the given condition. In this figure,
cracked specimens are labeled as C and crack-free
specimens are as NC. As expected, both HD content
and preheating temperature play important roles in cold
cracking. At preheating temperature like 400C, all the

Figure 6. Result of multi-pass weld metal cracking test for 1.5%Ni

wires.

Similarly, based on the UT result obtained from 1.5%Ni

weld deposits, C/NC boundary line was also constructed
and is represented in Fig. 6. When HD content was 4.5
ml/100g, no-cracking was observed even with preheating
at 400C and this fact was confirmed by the duplicate test.
It implies that the minimum preheating temperature
necessary to prevent cold cracking is less than 40 0C for
1.5%Ni wire while it was 650C for 0%Ni at the same
level of HD content. For the HD content over 7 ml/100g,

International Congress on Advances in Welding Science and Technology for

Construction, Energy and Transportation Systems (AWST - 2011)
24-25 October 2011, Antalya, Turkey
AWST-11/12
0

cold cracks were found at 40 C but could be prevented

with preheating at 650C, which is substantially lower than
that for 0%Ni. Accordingly, it can be concluded that the
microstructural change made by Ni addition, i.e.
decreasing amount of GF associating with increasing
proportion of AF, leads to better resistance to WM cold
cracking.

4. Discussions
The purpose of this study was to identify the effect of
weld microstructure on the cold cracking behavior in
ferritic weld metals and then to propose a appropriate
parameter that can be employed for evaluating the
microstructural effect on WM cracking. In multi-pass
welds, it is well known that WM cracking can occur
either in perpendicular to the surface (vertical crack) or
at angle of 45 deg to welding direction (Chevron crack)
with a tendency to change from Chevron to vertical crack
as the weld metal strength increases [18]. In the present
study, all the cracks found in FCA weld metals studied
were found to be formed in Chevron-type and the
occurrence of this crack was depending on weld
microstructure, HD content and the preheating
temperature. At the same level of HD content, as 0%Ni
wire requires higher preheating temperature than
1.5%Ni, it could be concluded that 0%Ni weld is more
susceptible to cold cracking than 1.5%Ni deposit. This
conclusion is quite opposite to the expectation that has
been described in the introduction of this paper.
According to equations (1) and (2), 0%Ni deposit would
be less susceptible than 1.5%Ni as it has lower Pcm and
lower tensile strength (Table 1 and 2). However, this was
not the case in the present study. It means both equations
are no longer valid for explaining the present result and
the parameters like Pcm (or CE) or weld metal strength
cannot represent the microstructural effect on WM
cracking appropriately. This discrepancy arises
presumably because equation (1) did not care about the
characteristics of weld microstructure that is quite
different from that of HAZ and because equation (2)
mainly concerns about vertical cracks that generally
develops in the higher strength welds of tensile strength
over 700MPa. Hence, some other parameter must be
considered for proper explanation of the present result.
In the course of this study, it was demonstrated that
HICC found in present FCA welds were Chevron-type
cracks which have been known to be developed
following GBF. The mechanism by which Chevron crack
develops relating with GF has been suggested by many
workers [18]-[20]. They claimed that the high
longitudinal residual stresses developed in welded joint
led to plastic deformation in the softer region of grain
boundary ferrite and then the small incipient cracks
which form as a result interconnect under the shear
stresses acting at an angle of 45 deg. Therefore, if the
Chevron cracking is the main type of cracking, it is quite
reasonable to conclude that GF phase facilitates cold

cracking and thus deteriorates cold cracking resistance of

weld metal. Based on this conclusion, the difference in
cold cracking susceptibility between 0%Ni and 1.5%Ni
can be interpreted by the difference in %GF in their
microstructures, i.e. the lower resistance of 0%Ni deposit
to cold cracking is simply because of the higher
proportion of GF in its weld. This result has been
demonstrated even in the single pass weld performed by
G-BOP test using the present FCAW wires [21].
(a)

(b)

Figure 7. Effect of %GF and HD content on the occurrence of cold

cracking at preheating temperature of (a) 45 0C and (b) 650C.

In the above discussion, it was appreciated that, as far as

Chevron-type cracking concern, the proportion of GF
(%GF) would be more appropriate for estimating cold
cracking susceptibility of ferritic weld metal than the
ones like carbon equivalent or weld metal tensile
strength. At the similar strength level, therefore, the cold
cracking susceptibility can be considered as a function of
%GF and HD content. Such a consideration has been
attempted using the data shown in Figs. 5 and 6, and the
C/NC boundaries were re-constructed at the two different
preheating temperatures as shown in Fig. 7. The dotted
line is a boundary between the conditions under which
HICC will and will not occur in a multi-pass weld metal

cracking test. Even with the limited data available at this

moment, these two figures can clearly demonstrate the
effect of %GF on HICC susceptibility by showing that
the allowable %GF for no-HICC occurrence decreases
with increasing HD content between 4 and 8ml/100g. At
a preheating temperature like for example, the %GF for
crack-free weld decreases roughly from 20% to 5% with
increasing HD content from 4 to 6 ml/100g. In case of
preheating temperature of 65 0C, it decreases to a same
extent as the HD content increases from 6 to 8 ml/100g.

compared with 0%Ni deposit.

Such an effect of %GF can be used for a guideline in

designing new welding consumables with improved cold
cracking resistance through microstructural control.
More clearly, at least for welding consumables of
500~700 MPa strength level where GF is one of the
major microstructural constituents, cold cracking
resistance can be improved by decreasing the %GF in the
weld metal microstructure. If the weld metal strength
exceeds over 800 MPa, GF phase will not be present in
weld metal being the bainite more dominant resulting in
vertical-type cold cracks instead of Chevron-type cracks
[22]. Therefore, the effect of GF on HICC could be
appeared in the weld deposits that contain GF phase to
some extent.

References

Decreased proportion of GF can be derived by increasing

the content of AF which is well known to improve the
weld metal toughness. Therefore, weld metal toughness
might be a rough indicator that can be used for
estimating HICC resistance of ferritic weld metal; i. e.,
the tougher the weld metal the better resistance to HICC.
In fact, it has been stated by Davidson and his coworkers [23] that HICC resistance and toughness are
indirectly related through their mutual dependence on
microstructural features.

5. Conclusions
Multi-pass weld metal cracking tests were carried out to
evaluate the effect of weld metal microstructure on cold
cracking resistance of ferritic weld deposit using two sets
of FCAW wires which are different in Ni content (0%Ni
and 1.5%Ni) and thus are different in weld
microstructure, especially in %GF. From the
experimental results, following conclusions could be
made.
1) At the same level of HD content, the 1.5%Ni deposit
showed substantially better resistance to cold cracking
than 0%Ni and this difference was attributed mainly to
the variation in weld microstructure between these two
welds.
2) Microscopic analysis showed that all the cracks
observed were Chevron-type cold cracks in which GF
phase plays an important role as it provides with the
preferential route for crack propagation. This fact
justified the superior resistance of 1.5%Ni deposit as its
microstructure contained smaller amount of GF

3) Microstructural parameters based on weld metal

chemistry like carbon equivalent (CE or Pcm) or on weld
metal strength failed to explain the difference in cold
crack resistance of FCAW deposits studied in this
investigation. Therefore, a new parameter of %GF was
proposed that can be used for quantification of cold
crack susceptibility of weld metal microstructure.

[1]
[2]

[3]
[4]
[5]
[6]
[7]
[8]
[9]
[10]
[11]
[12]
[13]
[14]
[15]
[16]
[17]
[18]
[19]
[20]
[21]
[22]
[23]

T. W. Montemarano, J. of Ship Prod., 2, (1986), 145.

R. E. Dolby, Proc. of Int. Workshop on the Innovative Strucrtural
materials for Infrastructure in 21st Century, NIMS, Tsukuba,
Japan (2000), 131.
N. Yurioka, Proc. of First Int. Conf. on Adv. Structural Steels
(ICASS 2002), p. 119, NIMS, Tsukuba, Japan (2002).
E. Takahashi and K. Iwai, J. of JWS, 48, (1979), 885 (in
Japanese)
J. Vuik, IIW Doc. IX-1686-92, (1992).
R. J. Pargeter, TWI Report 461, (1992).
H. J. Kim, Int. J. of KWS, 2, (2002), 52.
Y. Ito and K. Bessyo, J. of JWS., 37, (1968), 683. (in Japanese)
M. McParlan and B. A. Graville, Welding J., 55, (1976), 95-s.
P. H. M. Hart, Welding J., 65, (1986), 14-s.
A. P. Chakaravarti and S. R. Bala, Welding J., 68, (1989), 1-s.
T. Yatake, N. Yurioka, R. Kataoka and E. Tsunerromi, J. of
JWS, 50, (1981), 291. (in Japanese)
N. Okuda, Welding J., 66, (1987), 141-s.
AWS A4.3-93: AWS, USA (1992).
D. J. Abson, A. Duncan and R. J. Pageter, IIW Doc. 1533-88
(1988).
S. S. Tuliani, Welding Res. Int., 6, (1976), 19.
V. S. Wright and I. T. Davison, Metal Const., 11, (1979),129.
J. M. F. Motta and R. L. Apps, Welding J. 61, (1982), 222-s
D. J. Allen, B. Chew and P. Harris, Welding J., 61, (1982), 212s.
G. Jinjin and W. Zuqian, Welding Int., 3, (1989), 984.
J. S. Seo, H. J. Kim and H. S. Ryoo, Met. and Mater. Int., 14,
(2008), 515
H. J. Kim and B. Y. Kang, ISIJ Int., 43, (2003), 706.
J. L. Davidson, S. P. Lynch and A. Majurndar, Pro. of the Conf.
on Hydrogen Management in Steel Weldments, Defense-Science
Technology Organization, Melbourne, Australia, (1996), 21.