j o u r n a l o f m a t e r i a l s p r o c e s s i n g t e c h n o l o g y 2 0 7 ( 2 0 0 8 ) 30–39

journal homepage: www.elsevier.com/locate/jmatprotec

Effect of weld thermal cycle on microstructure and fracture

toughness of simulated heat-affected zone for a 800 MPa
grade high strength low alloy steel

Yaowu Shi a,∗ , Zhunxiang Han b

a School of Materials Science & Engineering, Beijing University of Technology, 100 Ping Le Yuan,
Chaoyang District, Beijing 100022, PR China
b Dongfang Boiler Group Co. Ltd., Zigong, Sichuan 643001, PR China

a r t i c l e i n f o a b s t r a c t

Article history: In the present investigation, thermal simulated specimens were used to investigate the
Received 4 September 2006 effect of welding cooling time and peak temperature on characteristic fracture toughness
Received in revised form and microstructure feature of heat-affected zone (HAZ) for an 800 MPa grade high strength
18 October 2007 low alloy (HSLA) steel. It is found that the fracture toughness is the best for the simulated
Accepted 16 December 2007 coarse-grained HAZ, when the cooling time of t8/5 is 18 s. In addition, the size of prior austen-
ite grain, and the volume fraction of bainitic ferrite and M/A constituent increase with
increasing the cooling time. However, the volume fraction of martensite decreases with
Keywords: increasing the cooling time. Remarkable decrease of toughness is observed with increas-
HSLA steel ing the size of austenite grain and the volume fraction of M/A constituent. Moreover, there
Welded joint exists the effect of orientation on fracture toughness for the specimens subjected to weld
Heat-affected zone thermal cycle. Generally, the fracture toughness of simulated HAZ with L–T orientation is
Fracture toughness higher than that with T–L orientation. The reason may be related to that the strip structure
formed during rolling is remained after the thermal simulation. Furthermore, the investi-
gation shows that the toughness of coarse-grained zone is higher than that of fine-grained
zone for the simulated HAZ. The reason may be related with the microstructure evolution of
the HAZ during the complete thermal cycle used in the simulation. For the fine-grained HAZ,
the shorter cooling time of t8/5 may be not benefit for the self-tempering and decomposition
of M/A constituents, then the toughness of the fine-grained HAZ is lower. For the coarse-
grained HAZ, however, the longer cooling time of t8/5 , may promote the decomposition of M/A
constituents, then the toughness of coarse-grained HAZ is improved. The fracture tough-
ness deteriorated drastically for the partly phase transformed HAZ may be related with the
formation of mixture microstructure, in which the M/A constituent is distributed in shape
of network.
© 2007 Elsevier B.V. All rights reserved.

1. Introduction bulldozer, excavators, loaders, etc. are extensive requisite

for various construction works, mine production or even for
With the development of engineering constructions, large- disaster salvation. Moreover, it should be mentioned that
scale and heavy-duty construction equipments such as working condition of the construction equipments is rather

∗
Corresponding author. Tel.: +86 10 67392265; fax: +86 10 67396138.
E-mail address: shiyw@bjut.edu.cn (Y. Shi).
0924-0136/$ – see front matter © 2007 Elsevier B.V. All rights reserved.
doi:10.1016/j.jmatprotec.2007.12.049
 j o u r n a l o f m a t e r i a l s p r o c e s s i n g t e c h n o l o g y 2 0 7 ( 2 0 0 8 ) 30–39 31

severe. Besides complex impact and fatigue loading, the
equipments may be subject to low temperature in winter.
Thus, the strength matching and toughness of welded joints
are important except exact selection of welding consumables
and welding procedure.
Nowadays, high strength low alloy (HSLA) steel is widely
used in engineering structures. The reason is due to
many attractive properties of the steel, such as higher
strength/weight ration, formability, and weldability. Despite
these useful properties, the welding of this steel, when not
critically controlled, has often posed problems, particularly in
the shop condition.
Under thermal effect of welding, evident change of
microstructure and property occurs in the base metal adjacent
to the weld metal. Sometimes, the change of the microstruc-
ture may lead to local brittleness in heat-affected zone (HAZ).
In addition defects, stress concentration and higher residual
stresses are easy to coexist in the welded joint. Thus, frac-
ture failure of weldment cannot be completely prevented till
now.
HAZ of welded joint is very narrow in width, and the
HAZ consists of many fine regions having different structures.
When the fracture toughness is measured using true weld-
ment, the results may represent a global property of HAZ. The
measured results cannot be used to describe the property of
the fine region of HAZ. Thus, it is very difficult to analyze
the effect of a characteristic microstructure on fracture tough-
ness using the true weldment. Since the thermal simulation
technique has generated with representation of Gleeble, the
research on the relation between microstructure and property
for the welded joint becomes easy. Of course, the weld thermal
simulation technique has disadvantage like other simulation
techniques. For example, the grain size of simulated speci-
men is slightly larger than that of true welded joint when
experiencing the identical thermal profile. The reason is that
the thermal pinning is not considered in the thermal simula-
tion process. Even though the simulated result may not fully
reflect the true welded joint, the thermal simulation technique
becomes an important tool in the field of current weld physical
metallurgy and steel rolling.
According to the metallurgical characteristics of the HSLA
steel, heat input of welding process significantly affects the
mechanical property of heat-affected zone. In the present
study thermal simulated specimens were used to investigate
the effect of welding cooling time and peak temperature on
fracture toughness and microstructure feature of HAZ for an
800 MPa grade HSLA steel. The results will be available for
determining welding procedure and evaluating fracture safety
for large-scale construction equipments.
2. Materials and experimental procedure
2.1. Materials
The test material was a 800 MPa grade high strength low alloy
steel plate of 16 mm thickness produced by Wuhan Steel Com-
pany. The plate was heat-treated by water-cooling at 923 ◦ C at
2.5 min/mm, the tempering at 600 ◦ C at 3 min/mm. The chemi-
cal composition expressed in wt% was as follows: 0.11C, 0.23Si,
0.87Mn, 0.012P, 0.009S, 0.32Cu, 0.46Ni, 0.59Cr, 0.30Mo, 0.04V,
0.04Al, 0.003B, remainder ferrite. The mechanical properties
of the plate were: yield stress,  y = 845 MPa, tensile strength,
 u = 880 MPa, elongation = 15%, and average value of Charpy
Vee-notch tests were 115 J and 98 J at −20 ◦ C and −40 ◦ C,
respectively.
2.2. Specimens preparation
The thermomechanical simulation was conducted in a Glee-
ble 1500. Square bas specimens (11 mm × 11 mm × 105 mm)
were prepared and subjected to thermal cycles. The maximum
temperature, holding time and cooling rate of the thermal
cycle parameters were selected according to possible proce-
dures during metal arc welding. After the thermo-mechanical
process, the simulated specimens were cut and ground to
the standard Charpy size of 10 mm × 10 mm × 55 mm. The test
specimens were prepared in the L–T and T–L orientation,
in which the first letter designates the direction of load-
ing, while the second letter designates the direction of crack
propagation. L represents the longitudinal rolling direction
of the sheet, and T represents the long transverse direc-
tion.
The main parameters of the thermal cycles are given in
Tables 1 and 2, respectively, for simulating various cooling
rate and peak temperature. In the experiments the thermal
cycle parameters are selected on a basis of actual GMAW weld
experiments and swing a range. Actually, in this work the
peak temperature and cooling rates in the thermal cycles are
well designed to simulate possible heating history and possi-
ble HAZ microstructures occurred in real welding production.
The thermal cycling parameters given in Table 2 would pro-
duce microstructures evolution occurred in different zones of
the HAZ.

Table 1 – Time intervals for thermal simulation cycle parameters (time: s)

Heating time from Holding time at Cooling time from Cooling time from Cooling time from Heat input,
ambient to 1300 ◦ C 1300 ◦ C 1300 ◦ C to 800 ◦ C 800 ◦ C to 500 ◦ C 500 ◦ C to ambient Q (kJ/cm)

3 2 10 6 40 10.7
3 2 15 9 65 14.1
3 2 20 18 80 22.3
3 2 25 27 95 29.2
3 2 32 45 110 41.1
3 2 55 100 140 70.0
3 2 110 240 260 125.5
32 j o u r n a l o f m a t e r i a l s p r o c e s s i n g t e c h n o l o g y 2 0 7 ( 2 0 0 8 ) 30–39

Table 2 – Time intervals for simulating various peak temperatures (time: s)

Simulated peak Heating time from Holding temperature Cooling time from Cooling time Cooling time from
temperature, ◦ C ambient to simulated at peak temperature peak temperature from 800 ◦ C to 500 ◦ C to ambient
peak temperature to 800 ◦ C 500 ◦ C

1300 3 2 25 27 95
960 8 4 3 18 110
800 10 8 0 18 140
650 14 14 – – 180

2.3. Fracture toughness testing
The notch of the three-point bend specimens was spark-
machined first, then fatigue precracking was done to develop
an initial crack length a0 including the notch length. The ratio
of initial crack length to specimen width, a0 /W, was approxi-
mately equal to 0.45–0.55.
For evaluating fracture toughness, the elastic-plastic
parameter J-integral was used. The determination of J-integral
was conducted in accordance with British Standard Draft
BS7448-1997 (BSI, 1991). In the testing the single-specimen
technique was used to measure the toughness parameter.
The fracture toughness tests were carried out in a MTS 810
materials testing system with loading rate of 0.5 mm/min. Test
temperature was ambient temperature. During testing load
versus load-line displacement (P − ) was recorded continu-
ously, then the data was processing through a software origin.
The value of J-integral can be calculated for the three-point
bend specimen as follows:
J = Je + Jp
where Je is the elastic component of J and Jp is the plastic
component of J.
1 − 2

PS
 a 2
Je = f (2)
E BW 3/2 W were then sliced to a thickness of 500 ␮m, then the thick-
where P is the load at a characteristic point, B the thickness
of specimen, W the width of specimen, S the span, which is
equal to 4W, and a is the crack length. In the calculation a0
was used as a.  is the Poisson’s ratio, and it is 0.3 for steel. E
is the elastic modulus and it is 210,000 MPa for steel.
carried out using spark machine. The sectioned samples were
cold mounted, ground on SiC papers, polished, and chemically
etched by 4% Nital.
In order to reveal the boundaries of the prior austen-
ite grains, a solution of 3 g picric acid and 5 mL synthetic
detergent in 100 mL distilled water adding small amount of
H2 O2 was used. The etching temperature was about 50 ◦ C.
Linear secant technique was used to measure the size of
the prior austenite grains. A two-step metallographic etch-
ing technique was used to distinguish M/A constituents from
carbides (Ikawa et al., 1980). In the first step the electrolyt-
ically etching was conducted in CH3 COONH2 (5 g) and NaF
(0.5 g) in distilled water (100 mL) at room temperature and a
voltage of 3 V for 3 s. Then in the second step the electrolyt-
ically etching was conducted in NaOH (25 g) and picric acid
(5 g) in distilled water (100 mL) at room temperature and a
voltage of 6 V for 30 s. After the two-step etching, the M/A con-
stituents remained unetched, but carbides were recessed in
the ferrite matrix. The volume fraction of M/A constituents
was determined through metallographic point-counting tech-
(1) nique using an AMRAY-1000B scanning electron microscope
(SEM).
For detailed analysis of microstructure, thin foil trans-
mission electron microscopy (TEM) samples were prepared
from small 3 mm diameter cylinders removed from the
cross-section of the simulated specimens. The cylinders
ness was reduced to 80–100 ␮m by grinding. A solution of
10% perchloric acid in ethanol was employed for thinning
to perforation in a twinjet electropolisher. Thin foil sam-
ples were observed under a JEM 200CX TEM operated at
160 kV.

a  3(a0 /W)
1/2
[1.99 − (a0 /W)(1 − a0 /W)(2.15 − 3.93(a0 /W) + 2.7(a0 /W) )]
2
0
f = (3)
W 2(1 + 2a0 /W)(1 − a0 /W)
3/2

2Up
Jp = (4) 3. Results and discussion
Bb

where b is the ligament length, and it is equal to W − a0 . Up
is the plastic component of deformation energy. The value of
Up was obtained from the relevant area under the load versus
load-line displacement curve.
2.4. Metallographic observation
Microstructure of simulated HAZ was observed by a MCF3 opti-
cal microscopy. All sectioning of the welded specimens was
3.1. Effect of cooling time on toughness of HAZ
In the test all the characteristic values of J-integral were Jc ,
Ju , or Jm , respectively. Among them, Jc represents a measure
of fracture toughness at instability without significant stable
tearing crack extension. Ju is the toughness at fracture instabil-
ity after the onset of significant stable tearing crack extension.
Jm is the toughness at the first attainment of a maximum load
plateau for fully plastic behavior. Even the value of Ju and Jm
 j o u r n a l o f m a t e r i a l s p r o c e s s i n g t e c h n o l o g y 2 0 7 ( 2 0 0 8 ) 30–39
Table 3 – Effect of cooling time on fracture toughness of simulated HAZ for the 800 MPa grade HSLA steel
Cooling time TL orientation
t8/5 (s) Jm (kN/m)
6 55.7
6 67.8
6 62.2
9 61.5
9 70.7
9 66.1
9 –
18 68.8
18 77.8
18 70.7
27 – 146.0
27 – 146.9
27 – 145.0
45 65.3
45 58.8
45 60.1
100 54.7
100 60.6
100 48.8
240 60.9
240 48.3
240 52.7
are size-dependent and a function of test specimen geometry,
it may be useful to serve as a basis for material comparison
and technological selection. Effect of cooling time on fracture
toughness of HAZ is given in Table 3 and Fig. 1, in which t8/5 is
the cooling time from 800 ◦ C to 500 ◦ C.
It is indicated from the Fig. 1 that when t8/5 is 18 s, the frac-
ture toughness in the simulated HAZ is the highest. When the
value of t8/5 is 45 s or longer, however, the toughness is deteri-
orated. These results imply that the weld heat input should be
restricted to limit the cooling time t8/5 equal to about 18 s. This
tendency is especially clear for the specimens with L–T orien-
tation, in which fracture instability occurs, when the cooling
time t8/5 is longer than 45 s. The minimum toughness occurs at
t8/5 of 100 s, in which crack rapidly propagates without obvious
ductile crack growth.
It is well known that the mechanical properties of steels
are strongly dependent on the microstructure. Rapid heat-
ing and subsequent cooling of the weld thermal cycle lead to
dramatic evolution of microstructure in HAZ. Generally, the
phase transformation sequences may be described as follows:
in the temperature range above 800–1300 ◦ C, the austenite
grains grow rapidly. In the subsequent cooling process from
800 ◦ C to 300 ◦ C, the austenite decomposes to various ferrite
morphologies.
The austenite to ferrite decomposition starts with the for-
mation of allotriomorphic ferrite at prior austenite boundaries
and eventual coverage of these boundaries. With the con-
tinued cooling, the side plate ferrite may nucleate at the
ferrite/austenite boundaries and extends into the untrans-
formed austenite grain interiors. Further cooling to low
temperature, the bainitic ferrite or acicular ferrite may form.
When carbide-free bainitic ferrite is formed, the remained
austenite is enriched in carbon and becomes stable. The car-
bon content of the remaining austenite may reach 0.5–0.8 wt%.
33
LT orientation
Jm (kN/m) Ju (kN/m) Jc (kN/m)
– – –
– – –
– – –
166.6 – –
158.6 – –
143.1 – –
131.6 – –
171.4 – –
168.5 – –
188.4 – –
– –
– –
– –
151.3 – –
121.6 – –
– 73.8
– – 54.7
– 84.1 –
– – 46.6
– 80.7 –
– 90.7 –
– 103.4 –
With further cooling to temperature close to room temper-
ature, the remaining austenite may completely or partially
transform to martensite.
As the M/A constituent forms during bainite transfor-
mation, the carbon-enriched, untransformed regions will
partially transformed into martensite at low temperature. The
carbon-enriched austenite regions are formed by rejecting of
carbon from ferrite to austenite following the transformation
of bainite ferrite. The formation of M/A constituents leads
to the deterioration of toughness in the weld HAZ. In order
to observe M/A constituent, a two-step electrolytical etching
was conducted. The etching technique is different from the
chemical etching of Nital, which is usually used for reveal
traditional microstructure of steel. The M/A revealed by the
above two-step electrolytical etching is considered to be coex-
isting with bainite ferrite, as defined by Ohtani et al. (1990).
Bramfitt and Speer (1990) also suggest a similar definition of
M/A. Studies indicate that two types of martensite are involved
in the M/A constituent. They are lath martensite with higher
start temperature of martensite transformation and plate-
like martensite, in which twin can be identified (Hrivanak
et al., 1992). In addition, carbides in the M/A constituent
can be precipitated during the self-tempering of martensite
in subsequent cooling process, or carbides can transformed
directly from carbon-rich austenite. The decomposition of
M/A constituents causes the improvement of HAZ toughness
apparently. To avoid confusion in this study, bainite structures
are classified as lower bainite and upper bainite according to
the classical classification. BI is considered to be bainite ferrite
plus M/A.
Actually, it is found that there exist allotriomorphic ferrite,
bainitic ferrite and martensite for the simulated HAZ of the
test steel. On the basis of the thermal cycles given in Table 1,
the measured results on the size of the prior austenite grain,
34 j o u r n a l o f m a t e r i a l s p r o c e s s i n g t e c h n o l o g y 2 0 7 ( 2 0 0 8 ) 30–39

Table 5 – Effect of cooling time t8/5 on volume fraction of

M/A constituent
t8/5 (s) M/A constituent (%)

9 9.6
18 12.8
27 15.2
45 17.8
100 26.2
240 24.2

Fig. 1 – Effect of cooling time on fracture toughness: (a) T–L
orientation and (b) L–T orientation.
volume fraction of various structure constituents and cooling
time are listed in Table 4.
It is clear that with increasing cooling time t8/5 , i.e. with
increasing holding time at higher temperature, the size of
the austenite grain increases. At the same time the volume
fraction of martensite decreases and the volume fraction of
bainitic ferrite increases, with increasing the cooling time t8/5 .
If the cooling time is longer than 100 s, the allotriomorphic fer-
rite starts to appear. It should be mentioned that the austenite
grain size of the coarsened grain HAZ for the real welded joints
was slightly larger than that for the simulated HAZ, in which
t8/5 was 18 s. The main reason may be related to the peak
temperature of 1300 ◦ C selected in the thermally simulation
experiments. The peak temperature used in the simulation
was slightly lower than that in real coarsened grain HAZ.
It is well known that lath martensite exhibits good tough-
ness. However, the M/A gives a significant effect on toughness
of welds. Several publications have indicated that the M/A
island is the main microstructure controlling toughness (Kim
et al., 1991; Matsuda et al., 1991; Davis and King, 1993; Ale et
al., 1996; Liao et al., 1998; Chen et al., 2006). Because M/A con-
stituent is hard and brittle particle, and is usually regarded as
the sites of crack initiation (Toyoda, 1988). Table 5 gives the
results on effect of cooling time t8/5 on the volume fraction
of M/A constituent. With increasing the cooling time t8/5 , the
volume fraction of M/A constituent is increased. The max-
imum fraction of M/A constituent appears at t8/5 equal to
100 s. In fact, the higher the amount of M/A in the simulated
HAZ, the lower the toughness is. Moreover, with increasing
the cooling time t8/5 , the morphology of M/A constituent will
be transformed from bar to block. Typical morphology of M/A
constituent is shown in Fig. 2 with different cooling time.
From the results above, it is indicated that the size of prior
austenite grain, and the volume fraction of bainitic ferrite and
M/A constituent increase with increasing the cooling time.
The volume fraction of martensite decreases with increasing
the cooling time. Thus, it may be understood that the remark-
able decrease of toughness is observed with increasing the size
of austenite grain and the volume fraction of M/A constituent.
Fig. 3 shows the effect of cooling time on fracture surfaces
of the simulated specimens. When t8/5 is 18 s, the fractograph
is in dimples. However, when the value of t8/5 is 100 s, the frac-
tograph is in quasi-cleavage. It is indicated that better fracture
toughness of the coarsened HAZ can be obtained, when the
value of t8/5 is 18 s. When the value of t8/5 is 100 s, the tough-
ness is deteriorated. The fractographic observations submit a
powerful substantiation for the fracture testing results.

Table 4 – Effect of cooling time on the size of prior austenite grain and the volume fraction of microstructure constituents
t8/5 (s) Size of prior austenite grain (␮m) Allotriomorphic ferrite (vol.%) Bainitic ferrite (vol.%) Martensite (vol.%)

6 68 0 0 100
9 93 0 16.4 84.6
18 117 0 39.3 60.7
27 128 0 57.5 42.5
45 145 0 86.9 13.1
100 151 3.2 96.8 0
240 159 4.7 95.3 0
 j o u r n a l o f m a t e r i a l s p r o c e s s i n g t e c h n o l o g y 2 0 7 ( 2 0 0 8 ) 30–39
Fig. 2 – M/A constituent with different cooling time: (a) t8/5 = 18 s, (b) t8/5 = 45 s, (c) t8/5 = 100 s and (d) t8/5 = 240 s.
It should be mentioned that there exists the effect of ori-
entation on fracture toughness for the specimens subjected
to weld thermal cycle. It may be seen from Fig. 1 that the
fracture toughness of simulated HAZ with L–T orientation is
higher than that with T–L orientation except the toughness at
the cooling time t8/5 of 100 s. Fig. 4 shows the microstructure
of prior austenite grain for the coarse-grained HAZ simu-
lated. It is clear that strip-like microstructure produced by
rolling process is remained. Even though the high temperature
heating of 1300 ◦ C was subjected during the thermal simula-
tion, the alloy elements may not completely dissolved. The
reason is that the heating rate is very quick and the time
stayed in the high temperature is very short for the welding
simulating process. Thus, the rolling direction was remained
after the thermal simulation of welding. This is different
with the general heat treatment, in which the strip struc-
ture should disappear after heat treatment, as any elements
segregated might redistribute during heating to quenching
temperature.
3.2. Effect of peak temperature on toughness of HAZ
Fig. 5 shows the effect of peak temperature on fracture tough-
ness of simulated HAZ on the basis of the thermal cycles
given in Table 2. It is indicated that the fracture toughness
of specimens with L–T orientation is obviously higher than
that with T–L orientation at all the simulated peak temper-
atures. Moreover, it is felt that the lowest toughness occurs
35
at the peak temperature of 800 ◦ C and the averages of the
fracture toughness reach to 57.9 kN/m and 97.8 kN/m for the
specimens with T–L orientation and L–T orientation, respec-
tively.
It should be pointed out that the highest toughness occurs
at the peak temperature of 1300 ◦ C. The averages of fracture
toughness reach to 72.4 kN/m and 176.1 kN/m for T–L and
L–T specimens, respectively. That is, the fracture toughness
is sensitive to the orientation of the specimens. It should be
mentioned that the fracture toughness of base metal spec-
imen with L–T is that Jm = 146 kN/m and Ji = 63.2 kN/m in
previous investigation, where Ji is the fracture toughness at
crack initiation determined by a multi-specimen technique.
Thus, the fracture toughness of the coarse-grained HAZ is near
the fracture toughness of the base metal for the L–T speci-
men.
The results above shows that the toughness of coarse-
grained HAZ is higher than that of fine-grained HAZ subjected
to the thermal cycle with peak temperature of 960 ◦ C. More-
over, it is inferred that the partly transformed zone displays
the lowest toughness. This result is the same as previous
investigations (Davis and King, 1993). The fracture toughness
deteriorated drastically for the simulated specimen subjected
to thermal cycle with peak temperature of 800 ◦ C is because
the formation of mixture microstructure in the partly trans-
formed zone of HAZ. As shown in Fig. 6, the distribution of
M/A constituent of the partly transformed zone is in shape
of network. The mechanism of forming the M/A constituent
36 j o u r n a l o f m a t e r i a l s p r o c e s s i n g t e c h n o l o g y 2 0 7 ( 2 0 0 8 ) 30–39

Fig. 3 – Fracture surfaces ahead of crack tip for the thermally

simulated specimens: (a) t8/5 is 18 s and (b) t8/5 is 100 s.

Fig. 5 – Effect of peak temperature on fracture toughness of

simulated HAZ: (a) T–L orientation and (b) L–T orientation.

The change of fracture toughness in the simulated HAZ

subjected to different peak temperature is also proved by the
stretch zone width (SZW) measurement. The welded joint was
made by gas metal arc welding (GMAW) process. The size of
the test plate was 500 mm in length and 150 mm plus 150 mm

Fig. 4 – Simulated coarse-grained HAZ with strip-like

microstructure.

network may be as follows: heating to the peak temperature

of 800 ◦ C, the carbide distributed at grain boundaries may dis-
solve and enter into phase-transformed austenite. Due to the
lower heating temperature, diffusion of carbon is rather dif-
ficult. Thus, carbon is rich in boundaries. Then, the network
distribution of the M/A constituent was formed after cooling.
The network distribution of M/A constituent may also lead to
the further deterioration of toughness. Fig. 6 – M/A constituent of partly phase transformed HAZ.
 j o u r n a l o f m a t e r i a l s p r o c e s s i n g t e c h n o l o g y 2 0 7 ( 2 0 0 8 ) 30–39 37

in width. The shielding gas was mixture of 80 vol.% Ar plus

20 vol.% CO2 . Welding wire was OK Antrod 13.29 (Ni–Mo) pro-
duced by ESAB with diameter of 1.2 mm. Details of welding
variables: voltage 28 V with direct current electrode positive
(DCEP); current 240 A; speed 5 mm/s. The preheat and inter-
pass temperatures were both 120 ◦ C. The fracture toughness
test was carried out on three-point bend specimens of cross-
section W = 2B = 24 mm, where W was the width of specimen,
and B the thickness. The notch was located across three region
of base metal, HAZ and weld metal, as called cross-bond spec-
imen (Prasannakumar et al., 1983). The ratio of precracking
length to specimen width, a0 /W, was approximately equal to
0.45–0.55. The bending test was conducted at ambient temper-
ature. After the bending test, the specimen was broken in liq-
uid nitrogen temperature. The values of SZW were measured
on fractograph by SEM observation. The values of SZW repre-
sent the fracture toughness characterized by crack tip opening
Fig. 7 – Distribution of SZW measured in a cross-bond
displacement. The measurement results are shown in Fig. 7.
GMAW specimen.
It is clear that there is uneven distribution of fracture
toughness in front of fatigue precracking for the welded spec-
imen. Under the condition of the test, it is indicated that the
fracture toughness of weld metal is higher than that of base indicate that the fracture toughness of different zones of the
metal. The fracture toughness of coarse-grained HAZ is bet- real weld HAZ have the same order as the fracture toughness
ter, and the fracture toughness of the partly transformed zone obtained from the thermally simulated specimens subjected
is worse. The present results of the HAZ SZW measurements to different peak heating temperatures.

Fig. 8 – TEM observation: (a) lath martensite (50,000×), (b) BI type carbide-free bainitic ferrite (20,000×), (c) BII type bainitic
ferrite (20,000×) and (d) BIII type bainitic ferrite (20,000×).
38 j o u r n a l o f m a t e r i a l s p r o c e s s i n g t e c h n o l o g y 2 0 7 ( 2 0 0 8 ) 30–39
When the above real welded joint specimen was used to
measure the fracture toughness, the Jm value of the coars-
ened grain zone of the HAZ was equal to 298.4 kN/m. However,
the Jm value of the simulated specimen subjected to the same
cooling time and peak heating temperature, as the real weld
HAZ was equal to 176.1 kN/m. Thus, the fracture toughness
Jm obtained from the simulated specimen and real weld spec-
imen cannot be compared each other, even though the both
have similar microstructures. The main reason may be related
to that the loads for general yielding and fracture of the speci-
men were different for the simulated specimen and real weld
specimen, as indicted by Dolby’s work (Dolby, 1972).
It should be mentioned that the triaxiality at crack tip of
the small-size specimen is not matched to that of the real full-
thickness specimen. The triaxiality will give a strong effect on
fracture toughness values under elastic-plastic condition. In
addition, if we transform the SZW measurement to CTOD or
J integral fracture toughness, there are some undetermined
factors. The transformation from SZW or CTOD to J integral
may be influenced by constraint factor. In addition the weld
strength mismatching affect the constraint factor (Shi et al.,
1998). Furthermore, weld strength mismatching give influ-
ence on the testing result of elastic-plastic fracture toughness
(Burdekin et al., 1994).
Fig. 8 shows the lath martensite and bainitic ferrite based
on the TEM observation. Fig. 8(a) shows that the martensite
has clear lath character. As the steel contains more alloy ele-
ments, the steel exhibits better stability against tempering.
Moreover, carbide-free bainitic ferrite can be found in the sim-
ulated HAZ, as shown in Fig. 8(b). As indicated (Ikawa et al.,
1980), there are three kinds of bainitic ferrite. That is, BI , BII and
BIII bainitic ferrite. Among them, carbide-free BI type bainitic
ferrite is the main reason deteriorating toughness, as the BI
type bainitic ferrite coexists with M/A constituents.
It should be pointed out that the coarse-grained HAZ
mainly consists of low carbon martensite and BII and BIII type
bainitic ferrite, as shown in Fig. 8(a), (c) and (d), but carbide-free
BI type bainitic ferrite is a little. However, the fine-grained HAZ
mainly consists of martensite and carbide-free BI type bainitic
ferrite. It should be pointed out that when the effect of peak
temperature was studied in the present work, cooling time
was not maintained constant in the experiments, as given in
Table 2. The purpose is to follow the thermal cycles occurred
in the practical HAZ. This means that HAZ microstructure was
affected by both peak temperature and cooling time.
Experimental measurement shows that the average of the
grain size is equal to 128 ␮m in the coarse-grained HAZ, and
28 ␮m in the fine-grained HAZ, and 46 ␮m in base metal.
Although the toughness is dependent on the grain size in gen-
eral, actually in this investigation the toughness of the coarse
grain HAZ is higher than that of fine-grained HAZ. The reason
may be related to the microstructure evolution during the sim-
ulated thermal cycle. The HAZ toughness should be explained
based on the whole thermal cycle subjected to the HAZ. In this
investigation, the peak temperature of 960 ◦ C and short hold-
ing time of 3 s at high temperature lead to the formation of
the fine-grained HAZ. At the same time, the shorter cooling
time of t8/5 , as given in Table 2, may be not benefit for the self-
tempering and decomposition of M/A constituents, then the
toughness of the fine-grained HAZ is lower. However, the peak
temperature of 1300 ◦ C and longer holding time of 25 s at high
temperature lead to the formation of the coarse-grained HAZ.
In addition, the longer cooling time of t8/5 , as given in Table 2,
promotes the decomposition of M/A constituents, then the
toughness of coarse-grained HAZ is improved.
4. Conclusions
(1) When cooling time t8/5 is 18 s, the fracture toughness in
the simulated HAZ is the highest. When the value of t8/5
is 45 s or longer, however, the toughness is deteriorated.
These results imply that the weld heat input should be
restricted to limit the cooling time t8/5 equal to about 18 s.
(2) It is indicated that the size of prior austenite grain, and
the volume fraction of bainitic ferrite and M/A constituent
increase with increasing the cooling time. However, the
volume fraction of martensite decreases with increasing
the cooling time. Remarkable decrease of toughness is
observed with increasing the size of austenite grain and
the volume fraction of M/A constituent.
(3) There exists the effect of orientation on fracture toughness
for the specimens subjected to weld thermal cycle. Gen-
erally, the fracture toughness of simulated HAZ with L–T
orientation is higher than that with T–L orientation. The
reason may be related to that the strip structure formed
during rolling is remained after the thermal simulation.
(4) The investigation shows that the toughness of coarse-
grained zone is higher than that of fine-grained zone for
the simulated HAZ. The results may be explained by the
microstructure evolution during the simulated thermal
cycle. For the fine-grained HAZ, the shorter cooling time of
t8/5 may be not benefit for the self-tempering and decom-
position of M/A constituents, then the toughness of the
fine-grained HAZ is lower. For the coarse-grained HAZ,
however, the longer cooling time of t8/5 , may promote the
decomposition of M/A constituents, then the toughness of
coarse-grained HAZ is improved.
(5) The fracture toughness deteriorated drastically for the
partly phase transformed HAZ may be related to the
formation of mixture microstructure, in which the M/A
constituent is distributed in shape of network.
Acknowledgement
This research is supported by the Wuhan Iron and Steel Co.
and the funding project of PHR (IHLB), Beijing Education Com-
mittee.
references
Ale, R.M., Rebello, J.M.A., Charlier, J., 1996. A metallographic
technique for detecting martensite–austenite constituents in
the weld heat-affected zone of a micro-alloyed steel. Mater.
Charact. 37, 89–93.
Bramfitt, B.L., Speer, J.G., 1990. A perspective on the morphology
of bainite. Metall. Trans. A 22, 817–829.
BSI, 1991. BS4778-1: 1991, Fracture mechanics toughness tests,
Part 1: Method for determination of K1c , critical CTOD and
 j o u r n a l o f m a t e r i a l s p r o c e s s i n g t e c h n o l o g y 2 0 7 ( 2 0 0 8 ) 30–39
critical J values of metallic materials, British Standards
Institution, London, 1991.
Burdekin, F.M., Kocak, M., Schwalbe, K.H., Denys, R., 1994.
Significance of strength mis-match in welds—summary of a
round table discussion. In: Schwalbe, K.H. (Ed.), Mis-Matching
of Welds. ESIS Publication 17, Mechanical Engineering
Publications Limited, London, pp. 103–111.
Chen, Y.T., Guo, A.M., Wu, L.X., Zeng, J., Li, P.H., 2006.
Microstructure and mechanical property development in the
simulated heat affected zone of V-treated HSLA steels. Acta
Metall. Sinica 19, 57–67.
Davis, C.L., King, J.E., 1993. Effect of cooling rate on intercritically
reheated microstructure and toughness in high strength low
alloy steel. Mater. Sci. Technol. 9, 8–15.
Dolby, R.E., 1972. Fracture toughness comparison of weld HAZ
and thermally simulated microstructures. Metal Const. Br.
Weld. J. 4, 59–63.
Hrivanak, I., Matsuda, F., Ikeuchi, K., 1992. Investigation of M/A
constituent in high strength steel welds. Trans. JWRI 21, 9–31.
Ikawa, H., Oshige, H., Tanoue, T., 1980. Effect of
martensite-austenite constituent on HAZ toughness of a high
strength steel. Trans. Jpn. Weld. Soc. 11, 87–96.
39
Kim, B.C., Lee, S., Kim, N.J., Lee, D.Y., 1991. Microstructure and
local brittle zone phenomena in high-strength low alloy steel
welds. Metall. Trans. A 22, 139–149.
Liao, J.S., Ikeuchi, K., Matsuda, F., 1998. Toughness investigation
on simulated weld HAZs of SQV-2A pressure vessel steel.
Nucl. Eng. Des. 183, 9–20.
Matsuda, F., Li, Z.L., Bernasovsky, P., Ishihara, K., Okada, H.,
1991. Investigation on the behavior of the M–A
constituent in simulated HAZ of HSLA steel. Weld. World 29,
307–313.
Ohtani, H., Okaquchi, S., Fujishiro, Y., Ohmori, Y., 1990.
Morphology and properties of low-carbon bainite. Metall.
Trans. A 21, 877–888.
Prasannakumar, S., Ramachandran, E.G., Ruge, J., 1983. Fracture
toughness testing of weld composite specimens. Metal Const.
15, 142–146.
Shi, Y.W., Sun, S.Y., Murakawa, H., Ueda, Y., 1998. Finite element
analysis on relationships between the J-integral and CTOD for
stationary cracks in welded tensile specimens. Int. J. Pres. Ves.
Pip. 75, 197–202.
Toyoda, M., 1988. Fracture Toughness Evaluation of Steel Welds.
Osaka University.