You are on page 1of 11

International

Journalof
Fatigue

ELSEVIER

International

Journal of Fatigue 23 (2001) 103-113


www.elsevier.com/Iocate/ijfatigue

Sulfide stress corrosion cracking and fatigue crack growth of


welded TMCP API 5L X65 pipe-line steel
L.W. Tsay a,*, Y.C. Chen a, S.L.I. Chan b
a Institute of Materials
b Institute

Engineering, National Taiwan Ocean University, Zpei-Ning Road, 202 Keelung,


of Materials Science and Engineering, National Taiwan VniversiQ, Taipei, Taiwan,

Received

8 March

2000; received

in revised

form 9 June 2000; accepted

28 August

Taiwan,
ROC

ROC

2000

Abstract
Slow strain rate tensile test and fatigue crack growth (FCG) test were performed to evaluate the fracture behavior of API 5L
X65 steel weldments after hydrogen-charging.
Regardless of the testing environments,
tensile fracture of all welds was located at
the weld metal (WM), which had the lowest hardness as compared to that of the other parts of the weld. When hydrogen-charged
in an H,S-saturated
solution, all specimens suffered a small drop in tensile strength as compared to those tested in air; however,
the loss in ductility was more significant. The susceptibility to hydrogen embrittlement
of the welds could be reduced significantly
by subjecting the welds to 6OOTY2 h tempering treatment. With the tempering treatment, the number of surface cracks in the WM
after hydrogen-charging
decreased and the tensile properties improved. Results of FCG test demonstrated
that in a higher stress
intensity factor range (AK), tempered steel plates with a severely banding structure had the higher fatigue crack growth rates
(FCGRs) than those of as-received ones. The WM of an uncharged weld had the lower FCGRs as compared to the parent metal
within the same AK range. However, the enhancement
of crack growth in the WM was very pronounced after hydrogen-charging.
SEM fractographs
of tensile and fatigue-fractured
surfaces revealed a quasi-cleavage
fracture in the embrittled region. 0 2001
Elsevier Science Ltd. All rights reserved.
Keywords:

Slow strain rate tensile test; Fatigue crack growth rates; Hydrogen embrittlement; Quasi-cleavage

1. Introduction
The resistance of a pipe-line steel to sulfide stress corrosion cracking (SSCC) and hydrogen-induced cracking
(HIC) is very important for steel to be used in sour
oil/gas applications. Elongated MnS inclusions are the
most susceptible sites for HIC initiation [ 1,2] as hydrogen atoms can easily accumulate at the interface between
the steel matrix and non-metallic inclusions. The segregation of elements such as P, Mn, C in the steel enhances
the formation of hard bands on cooling [3,4]. The existante of the banding structure not only decreases the
resistance of the steels to HIC [3,4] but also reduces both
the notch impact toughness and the reduction of area in
the tensile test [5]. Especially in controlled rolling plates
[l], HIC tends to propagate along pearlite bands or low
temperature transformation microstructures in the steel.

* Corresponding
24625324.

author. Tel.: +886-2-24622192;

fax: +886-2-

Under the same strength level, microstructural changes


from ferrite/pearlite mixtures to tempered martensite in
the steel are effective in improving its SSCC resistance
[6]. It has also been noted [6] that low carbon bainite
(LCB) or LCB/ferrite mixtures can improve the SSCC
resistance of the pipe-line steels. In a banded structure,
the hydrogen transportation flux is larger along the longitudinal direction relative to the rolling direction (RD)
than that along the transverse direction relative to the
RD [7,8]. Tau et al. [9] studied the hydrogen-assisted
fatigue crack propagation of bainitic and tempered martensitic steel, it was found that the microstructure was a
dominant factor in determining the hydrogen-assisted
crack growth of the steel. However, no direct relationship between hydrogen permeation behaviour and hydrogen-assisted fatigue crack growth (FCG) has been
determined [9].
In general, the heat-affected zone (HAZ) of a weld
(especially at the region adjacent to the fusion line) is
hardened after welding. The hardened microstructures
are usually sensitive to hydrogen embrittlement (HE)

0142-l 123/01/$ - see front matter 0 2001 Elsevier Science Ltd. All rights reserved.
PII:SO142-1123(00)00081-5

104

L. W. Tsay et al. /International

Journal of Fatigue 23 (2001) 103-113

Nomenclature
BM
base metal
CGHAZ coarse-grained heat-affected zone
CT
compact tension
FB
fusion boundary
FCG
fatigue crack growth
FCGR fatigue crack growth rate
FGHAZ fine-grained heat-affected zone
HAZ
heat-affected zone
HE
hydrogen embrittlement
HIC
hydrogen-induced cracking
LCB
low carbon bainite
PWHT postweld heat treatment
RD
rolling direction
SEM
scanning electron microscope
SMAW shielded metal arc welding
SSCC sulfide stress corrosion cracking
SSRT slow strain rate tensile
TEM
transmission electron microscope
TMCP thermo-mechanical control process
WM
weld metal

and stress corrosion cracking. To avoid the formation of


such susceptible microstructures, the carbon equivalent
or the alloying element additions have to be limited [lo].
It is reported that lowering the carbon content to the
range of 0.01-0.05% markedly decreases the hardness
of the central segregation zone and reduces the HIC susceptibility of high-strength pipe-line steels [5]. Significant advances in the steel-making process have been
achieved by the combination of the controlled rolling
with on-line accelerated cooling process, which is generally known as thermo-mechanical
control process
(TMCP) [ 111. The microstructures of TMCP steels are
highly refined by rapid cooling, resulting in a significant
improvement on strength and toughness of the steels
[ 121. In addition, as the strength of TMCP steels is achieved by refined microstructures, an improved weldability
of the steels by reducing the alloying additions can be
expected [lo]. Furthermore, the HAZ of TMCP steel
welds possesses a higher resistance to cold cracking and
a better toughness as compared to that of the conventional steel welds [13,14].
Hydrogen in the steel reduces the grain boundary
cohesion strength [15] and decreases the fracture stress
[ 16,171. The fatigue crack growth rates (FCGRs) of the
steel are accelerated in the presence of hydrogen [9,1820]. In this work, the SSCC and fatigue crack growth
(FCG) of a TMCP API 5L X65 steel plate and weldments have been investigated. The resistance to SSCC
of pipe-line steels was usually determined by slow strain
rate tensile (SSRT) tests [21-231. The FCGRs of the
weldment were measured and the results were compared

to those of the parent plate. In addition, the FCGRs in


various regions of a weld were determined by aligning
the crack growth normal to welding direction. While the
welds were usually used in the as-welded condition, the
influence of postweld heat treatment (PWHT) on the
resistance to HE and FCG of the weld has also been
evaluated; by subjecting the welds to tempering at 600C
for 2 h. The fractography associated with the tensile and
fatigue-fractured specimen were examined, with special
emphasis on the location of fracture and the change of
fracture modes after hydrogen-charging.

2. Material and experimental procedures


The used material was a 15 mm thick API 5L X65
steel plate produced by the TMCP process. In this process, the extent of banding found in conventional steel
plates was effectively reduced by cooling with water
through the temperature range of 800-500C. The
chemical composition of the steel by weight percent was
O.lOC, 1.49Mn, 0.25Si, 0.25Ni, O.O24Nb, O.O24Cu,
O.O5V, O.O14P, 0.001s with the balance being Fe. The
shielded metal arc welding (SMAW) process was
applied to fill up the joint. The joint geometry is a singlebevel 25 groove, with a root gap of 10 mm. A total of
14 welding passes was needed to complete the weld. A
high cellulose type electrode with a diameter of 4.0 mm,
which was suitable for the welding of API X60 grade
steel, was used in this study. The electrode used satisfied
the AWS A5.5 E7010-G standard specification. The

LW. Tsay et al. /International

nominal composition of the deposit by weight percent


was O.O9C,057Mn, 0.2OSi, O.O14P, O.OllS, 0.52Mo and
the remaning Fe. The welding parameters employed in
this study were : welding current=120 A, welding voltage=25 V and welding speed=130 mm/mm.
Various specimens with different orientation with
respect to the RD sectioned from the welded plate are
shown in Fig. 1. Schematic diagrams showing the
dimensions of tensile, Charpy impact and FCG specimens employed in this work are shown schematically in
Fig. 2. SSRT tests were performed at room temperature
either with a strain rate of 5~10~~ s-l in air or 5~10~~
s-l in an H,S-saturated solution. Tensile specimens of
steel plate aligned either parallel or transverse to RD are
designated as L and T specimens, respectively. For the
welded tensile (W) specimen, the weld metal (WM) was
located at the centre of the 50 mm gauge length [Fig.
2(a)]. Furthermore, some tensile specimens were subjected to a PWHT at 600C for 2 h. The testing solution
was prepared according to the NACE standard TM-Ol77-86. To ensure the saturation of solution with H,S during the test, continuous purging of H,S bubbles was carried out throughout the test. Tensile testing results represent data from at least three specimens for each test.
Charpy impact and FCG tests were also performed at
room temperature in air. To measure the impact energy
of the weld, the V notch of the impact specimen was
located either at 1 mm away from the vertical fusion
boundary in the fine-grained HAZ (FGHAZ) labeled
HAZ specimen, or at the WM (i.e. WM specimen). The
impact values represent results from an average of at
least five specimens.
For compact tension (CT) specimens of the steel plate,
crack growth directions were aligned either parallel or
normal to the RD of the steel plate, which were designated as TL and LT specimens, respectively. The effect
of various microstructures in the weldment on the
FCGRs was investigated by making the crack growth

Fig. 1. Schematic representation


tioned from the welded plate.

showing

the testing

specimens

sec-

105

Journal of Fatigue 23 (2001) 103-113

!
I

$1

i I
I

IlO! 12

I
1

unit : mm

Fig. 2. Schematic diagrams showing the dimensions


of (a) tensile,
(b) Charpy impact and (c) compact tension (CT) specimens employed
in the test.

direction normal to the welding direction. Such specimens were named as the fatigue-weld (FW) specimens.
As shown in Fig. 2(c), the crack would propagate 10
mm long within the WM including a 2 mm precrack,
then across the vertical fusion boundary into the HAZ,
finally entering the base metal (BM). For the CT specimens (LT, TL and FW specimens) subjected to 6OOC/2
h tempering treatment, a notation of 6 behind the specified specimens is given, e.g. TL-6 stands for a tempered
TL specimen. In case of hydrogen-charged CT specimens, a capital H is added to the last symbol of designated specimens, thus, TL-H represents the hydrogencharged TL specimen. The specimens profile was cut
by an electrodischarge machine with a copper wire electrode, followed by grinding the specimen to the required
thickness. A computerized MTS hydraulic servo-controlled testing system was utilized to perform the fatigue
crack growth experiment. The loading frequency was 20
Hz with a constant amplitude sinusoidal waveform of
the applied load at a stress ratio of 0.1 throughout the
test. In order to restrict the crack growth direction along
the centreline of the specimen, a 90 V notch of 0.5 mm
depth was machined on the opposite surfaces of all CT
specimens. In order to study the effect of hydrogen on

106

L. W. Tsay et al. /International

the crack growth, CT specimens were hydrogen-charged


in the H&saturated
solution for 48 h to ensure that
specimens were fully charged with hydrogen. Then, they
were washed in a mixed acid to clean off the surface
corrodant, followed by rinsing with distilled water. After
hydrogen-charging, no detectable defects were found in
the specimens, which indicated the steel plate and weld
were resistant to HIC. To inhibit the trapped hydrogen
from diffusing out of the specimens during fatigue testing, the hydrogen-charged specimens were subjected to
an electroless copper coating. The time interval between
hydrogen-charging and fatigue testing was maintained at
a constant minimum period of time.
Microhardness
measurements
across the fusion
boundaries between the deposit and BM were taken
using a Vickers microhardness tester. A 3% nital solution was used to reveal the microstructures in various
regions of the weld. Fractographic observations of the
fractured specimens were made on a Hitachi S4100 scanning electron microscope (SEM). The thin foils examined by a JEOL-100 CX-II transmission electron microscope (TEM) were prepared by a standard jet-polisher
using a 5% perchloric acid, 25% glycerol and 70% ethanol solution.

3. Results and discussion


3.1. Microstructure observation
Metallographs revealing different microstructures in
various regions of the as-welded weldment are given in
Fig. 3. The microstructures of WM consisted of grain
boundary ferrite, Widmanstatten ferrite, acicular ferrite
and small amount of microphases [Fig. 3(a)]. Mainly
upper and lower bainite, together with minor amount of
Widmanstatten ferrite, were observed in the coarsegrained HAZ (CGHAZ) [Fig. 3(b)]. In the region
reheated to a temperature just above AC,, a significantly
refined grain size (fine-grained HAZ, FGHAZ) as compared to that of the BM was obtained [Fig. 3(c)]. The
FGHAZ as examined by the TEM comprised of finegrained ferrite and pearlite [Fig. 3(d)]. The steel plate
used in this study also revealed a mixture of fine ferrite
and pearlite [Fig. 3(e)]. It shows that the degree of banding in the steel plate was not as severe as that in conventional steel plates. However, segregation bands reappeared after tempering at 6OOC/2 h [Fig. 3(f)]. In
addition, no significant changes of microstructures in the
HAZ and WM between the as-welded and tempered
specimens have been found. It has also been observed
that the WM contained a large amount of oxide
inclusions.

Joumal of Fatigue 23 (2001) 103-113

3.2. Microhardness measurement


Microhardness distributions in various regions of the
weld are displayed in Fig. 4. The steel plate had a
Rockwell hardness of HR, 92.6 (or Hv205) in the asreceived condition. After tempering at 600C for 2 h, a
slight increase in hardness to HR, 94.5 (or Hv215) was
obtained. Regardless of the tempering treatment, the
most hardened region was located at the CGHAZ [Fig.
4(a)] and the WM was found to be slightly softer than
the BM. Fig. 4(b) shows the hardness distribution along
the centreline of the deposits from the bottom to top
weld sides. Understandably the top weld was harder than
the rest of the weld, due to a previous deposit tempered
by subsequent welding passes. The associated reheating
temperature below AC,, resulted in the decrease in hardness of previous deposit. However, the deviation in hardness was found to be narrow, given a certain fluctuation
in hardness. Thus, the tempering effect imposed by subsequent welding thermal cycles on the hardness of the
previous deposits seemed to be insignificant. This result
could be attributed to the inherent low alloy content and
low hardenability of the WM. However, the as-welded
deposit of the final welding pass, which was untempered
by the welding heat, had a slightly higher hardness than
that found in the former deposits, as shown in Fig. 4(b).
3.3. Charpy impact test
The room temperature impact energy of L and T
specimens sectioned from the steel plate were 206 and
188 J, respectively. This is in accordance with the fact
that the refined microstructures of TMCP steel plate have
a high impact toughness. The impact energy of the HAZ
was also as high as 214 J. However, the impact toughness of the WM was as low as 80 J and could be attributed to the coarse solidified microstructures and the
presence of large amount of inclusions in the WM.
3.4. Tensile properties
Table 1 lists the tensile properties of various specimens tested in air and in an H&saturated solution. For
the steel plates tested in air, tempered specimens (L6,
T6) have a slightly higher strength than that of untempered specimens (L, T). This may be due to the reappearance of the banded structure after tempering, which has
an anisotropic effect on the tensile properties. Consequently, the tensile properties changed when stressed
along the banding direction. Also, L and T specimens
had similar tensile properties. The fusion zone was
slightly softer than the other regions of the weld, and
the tensile fracture of welds was located at the WM
when tested in air. Furthermore, a decline in tensile
strength was observed for the tempered welds. As a
whole, all specimens tested in air revealed a higher

L. W. ?kuy et al. /lnternationnl

Fig. 3. Microstructures
and (f) tempered BM.

in various

regions

of the as-welded

Journal

weldment.

strength and excellent ductility. However, as the testing


environments
changed from air to H,S-saturated
solution, a significant degradation
in tensile properties was
found. A slight decrease in tensile strength and a remarkable drop in ductility occurred for the steel plate tested
in the H,S solution, in contrast to those tested in air.
Steel plates still had slightly better tensile properties than
the welds. It can be noted that tensile properties of tempered welds (W6) were superior to those of the aswelded weldment (W), suggesting that the susceptibility
of the welds to HE could be reduced by a PWHT at
6OOC/2 h. Meanwhile, with or without PWHT, the fracture site of the welds in the H,S-saturated
solution
always located at the WM.
Macrographs
showing the fractured tensile specimens
tested in an H,S-saturated
solutions are shown in Fig. 5.
By comparing Fig. 5(a and b), it can be seen that more
circumferential
(secondary)
cracks are present on the
surface of T specimens than on the L specimens surface.
In addition to these circumferential
cracks which were
perpendicular
to the loading direction,
small stepwise
cracks aligned parallel to the loading direction
of L
specimens had been found, as shown in Fig. 5(a). For
the W specimens,
cracks were mainly initiated
and
propagated
in the WM [Fig. 5(c)]. Furthermore,
the

oj

Fatigue23

(2001)

(a) WM, (b) CGHAZ,

103-113

(c) FGHAZ,

107

(d) TEM micrograph

of (c), (e) BM

CGHAZ
adjacent
to the fusion boundary
(FB) was
highly susceptible
to HE, as evidenced
by the short
cracks adjacent to fusion boundary.
The amount of
cracking was reduced clearly for the welds with PWHT
[Fig. 5(d)], noticeable necking was found in this specimen showing again the improved tensile properties for
the tempered weld. SEM fractographs
of tensile fractured specimens showed ductile dimple fracture for all
the specimens tested in air. For the specimens tested in
the H,S-saturated
solution,
quasi-cleavage
fracture
resulted as shown in Fig. 6.
3.5. Fatigue

crack propagation

Fig. 7 shows the FCGRs (da/dN) vs the stress intensity


factor range (AK) for the steel plates and welds tested
in air. The LT specimens had similar FCG characteristics
as those of the TL specimens,
as shown in Fig. 7(a).
Thus, the crack growth characteristics
of the steel plate
were not sensitive to the orientations
of the specimens.
The extent of banding in these two orientations
with
respect to the RD have been reduced by the TMCP treatment, thus, it is reasonable to have such results. However, the banded structure reappeared
after tempering
treatment, as already shown in Fig. 3(f). The FCGRs of

108

L. W. Tsay et al. /International

(a)
-7.

Journal of Fatigue 23 (2001) 103-113

--nqzzz~~

I I

WIV/:

BM
HAZ
--.-Tempered
-+-As-welded

0'

12

10

Location (mm)
Top Weld

EC
9

200

x
Y

7---150 - Bottom Weld

3
2

100

--Tempered

50
01
0

-D-As-welded
2

10

12

14

16

Location (mm)
Fig. 4. Microhardness
distribution in the weld, (a) across
boundary and (b) along the centreline of the deposits.

the fusion

tempered TL (i.e., TL-6) specimens were higher than


those of untempered ones, especially in the high AK
region. Hence, higher FCGRs in the TX-6 specimen
could be attributed to the enhanced crack growth along
the banded structure. Fig. 7(b) shows the result of crack
growth tests for the welds as compared to the steel
plates. For the as-welded weld (FW), as the crack propagated across the WM and HAZ into the BM, similar
crack growth behavior as that of the TL specimen was
found. The results also demonstrated that the FCGRs of
WM and HAZ in FW specimens were lower than those
of the BM within the same AK range. The WM and HAZ
have a significant higher resistance to crack growth than
Table 1
Tensile properties

of various

specimens
L
In air

Materials
Properties/conditions
UTS (MPa)
Elongation (x100%)
Reduction in area (x100%)
Fracture location
a UTS: Ultimate

tensile strength.

636
21
40
_

tested in air and in an H&saturated

In H2S
598
6
3
_

6*: subjected

that of the steel plate (or BM). Similar results were


reported for the EH36 steel weldments [24].
The crack growth behavior in various regions of an
as-welded weldment is not only affected by the microstructures but also by the presence of different residual
stresses in these regions. The distribution of residual
stresses in a welded plate has long been known [25]. The
longitudinal residual stress, which acts along the welding
direction, had a maximum value in the WM and changed
from tensile to compressive stresses as it moved away
from the fusion zone. Meanwhile, the transverse residual
stress varies symmetrically from compressive at the
edges to tensile at the centre of the plate. It is known
that the presence of compressive residual stress ahead of
the crack tip will cause crack closure [26,27], resulting
in reduced FCGRs. As the crack propagated across the
as-welded WM, the partial relief and redistribution of
residual stresses caused the crack face to bend or to
rotate [28], hence, an enhanced crack closure occurred.
Thus, decreased FCGRs in the WM could be expected.
When the crack grew further into the HAZ, the magnitude of longitudinal residual stresses gradually decreased
hence its effect on the FCGRs was found to be reduced.
Finally, the crack growth characteristics became similar
to that of the parent metal as the crack grew in the BM.
The fatigue crack growth behaviour of FW specimen
was in agreement with the model of Shi et al. [28], who
performed a detailed study on the effect of welding
residual stresses and the fatigue crack growth of pipeline steel.
Tempered welds (FW-6) had higher FCGRs in the
WM and HAZ than those in the as-welded weldments
(FW). No significant microstructural changes were
revealed between the as-welded and tempered welds in
different regions of the welds. Such changes of crack
growth behavior for tempered welds, thus could be attributed to the relief of residual stresses after tempering.
The two crack growth curves of IV-6 and TL-6 specimens intersected at a AK of about 20.6 MPa&. Lower
FCGRs in the WM than those in the tempered steel plate
were found at the low AK range. However, the situation
was reversed for the crack growth in the HAZ of FW6 specimens. The da/dIV values in the HAZ were higher
than those in the tempered steel plate at the same AK

solutions*

T
In air

In H,S

L6*
In air

621
23
40
-

573
5
3
-

646
20
40
_

to 6OOC/2 h tempering

treatment.

In H2S
624
7
3
_

T6*
In air
635
22
40
-

In H2S
602
7
3
-

W
In air
575
15
40
WM

In H,S
531
3
4
WM

W6*
In air
562
16
40
Wh4

In H2S
547
6
4
WM

109

Fig. 5. Macroscopic
photographs
showing the different fractured
(d) W6 specimens. Arrows indicating examples of circumferential
of the L specimen.

Fig. 6. SEM fractographs


the H,S-saturated
solution
(b) W specimens.

of the tensile fractured


showing quasi-cleavage

specimens
fracture

specimens tested in the H2S-saturated


surface cracks. SWC: stepwise crack

tested in
of (a) L,

range. As the crack passed through the HAZ, consistent


FCGRs were resulted for both FW-6 and TL-6 specimens.
The results of fatigue crack growth tests for steel
plates and welds after hydrogen-charging
are shown in
Fig. 8. For the hydrogen-charged
TL (i.e. TL-H) specimen, enhanced crack growth only occurred at the low
AK range [Fig. 8(a)]. Meanwhile, the effect of hydrogen

solution for the (a) L, (b) T, (c) W and


aligned parallel to the loading direction

on accelerating
the crack propagation
of TL-6-H specimen was more prominent.
Such a result may again be
attributed to the reappearance of banding structure in the
tempered specimen, owing to the banding structure being
sensitive to HIC. The hydrogen diffusivity was the highest along the longitudinal
direction of a banded structure
than that along the other directions [7,8], thus sufficient
hydrogen atoms diffused into the strained region ahead
of the crack tip and assisted crack propagation.
Consequently, the tempered steel plate after hydrogen-charging
had higher FCGRs than those of the untempered
one
within the low AK range. It also demonstrated
that the
FCGRs of TL-6-H and TL-H specimens were similar to
those of TL-6 and TL specimens respectively,
as the AK
was greater than about 20 MPadm. Such results could
be related to the relatively slow transportation
of hydrogen atoms to the embrittled region ahead of crack tip
during crack growth, when the fatigue crack propagated
at a faster rate. Thus, at higher AK values the influence
of hydrogen was found to be lesser.
For the hydrogen-charged
welds, the increase
in
FCGRs was very significant as compared to those of the
uncharged specimens. As shown in Fig. 8(b), the FCGRs
in the WM and HAZ of the charged specimens (i.e. FWH) were higher than those of the FW specimens.
The
results also showed that the degree of enhanced crack
growth was more pronounced
for the weld than for the
steel plate after hydrogen-charging.
When the crack
propagated through the HAZ into the BM, similarcrack
growth properties for the FW and FW-H specimens were
obtained, where the influence of hydrogen was lesser
owing to the higher crack growth rates relative to hydro-

L. W. Tsay et al. /International

Journal of Fatigue 23 (2001) 103-113

1o-3

(4

t(a)

LT
4

1o-5

TL-6-H

TL-H

TL-6

TL

30

20

AK (MPa&)

AK (MPadi)

(b)

TL

FW

;
lo-

40

IO

20

30

40

A.K (MPa,h)

Fig. 7. Fatigue crack growth behaviours of (a) steel plates and (b)
weldments.

gen diffusion. Significantly accelerated crack growth


was found in the WM of FW-6-H specimen. The resistance to FCG in the WM was not improved by tempering
treatment. But as the crack growth passed the WM into
the HAZ, the FCGRs significantly dropped, suggesting
the tempered HAZ with a higher resistance to crack
growth than the WM in the FW-6-H specimen.
Fig. 9 shows the macroscopic appearances of fatiguefractured surfaces of various specimens. As mentioned
earlier, the FW specimen has a higher resistance to crack
growth than the TL specimen at the low AK range without hydrogen-charging.
Both steel plates and welds
revealed a flat and smooth fracture surface, as shown in
Fig. 9((a) and (b)) respectively. Furthermore, no obvious
differences were found in the macro fatigue-fractured
appearance in various regions of the weld between the

AK (MPahi)
Fig. 8. Fatigue crack growth behaviours of the hydrogen-charged (a)
steel plates and (b) weldments.

as-welded and tempered welds. For the specimens after


hydrogen-charging, a change in fatigue fractography was
found. Accelerated crack growth of the TL-H and TL6-H specimen was associated with a shinny area ahead
of the notch tip [Fig. 9(c)]. This area has been shown to
be quasi-cleavage fracture in a later micrograph of Fig.
10(c). The extent of embrittled area was deeper in the
central portion of the specimen than in the free surfaces.
Such features could be due to a higher constraint
developed at the central portion of the specimen,
implying the detrimental effect of hydrogen in this
region. In an earlier work by Toribio [29,30], it was indicated that the diffusion of hydrogen was governed not
only by the concentration gradient but also by the hydrostatic stress field in the material. Both the FW-H and

L. W. T~say et al. / Internutional

Journul of Fatigue 23 (2001) 103-I 13

Ill

precrack

of crack growth

Direction

f
Fig. 9.

Macroscopic

appearances

of fatigue-fractured

surfaces

FW-6-H specimens showed similar macro fatigue-fractured features. The coarse columnar grains, which were
not refined by the subsequent
welding passes, revealed
brittle fracture features. It was also found that the HAZ
of FW-6-H specimen was less susceptible
to HE, and
revealed less brittle features (not shown here) as compared to that of the FW-H specimen.
SEM fractographs
of fatigue-fractured
specimens are
shown in Fig. 10. Without hydrogen-charging,
fatiguefractured appearance of the steel plate revealed a transgranular fatigue fracture with fatigue striations at high
AK values [Fig. 10(a)]. Although the WM and HAZ of
the FW specimen had a higher resistance to crack growth
as compared to the steel plate, fatigue fracture appearance was observed to be similar between them [Fig.
10(b)]. The only difference was the presence of a large
amount of oxide inclusions in the WM of the FW specimen. After hydrogen-charging,
the fatigue fracture surface of the shinny area in the TL-6-H specimen was
mainly quasi-cleavage
fracture in the low AK range [Fig.
10(c)]. It was reasonable that the enhanced crack growth
of hydrogen-charged
steel plates was associated with the
more brittle fracture appearance.
A remarkable
difference was found in fatigue fractography of the FW-H and
FW specimens. The portion of fracture surface consisted
of a large extent of quasi-cleavage
in the WM of FWH specimens [Fig. 10(d)]. In addition, noticeable quasicleavage fracture was also observed in the CGHAZ of
FW-H specimens [Fig. 10(e)]. These results imply that

of (a) TL, (b) FW, (c) TL-H and (d) FWH

specimens.

the WM and CGHAZ of the as-welded weldment are


highly sensitive to HE, hence, enhanced crack growth in
the WM and HAZ of FW-H specimen can be expected.
As mentioned
previously,
the FW-6-H specimen also
exhibited
high FCGRs in the WM and decreased
in
FCGRs as the crack grew into the HAZ. The observation
on fatigue-fractured
surfaces showed similar fracture
features in the WM of FW-H and FW-6-H specimens,
again confirming
that the PWHT had little effect in
improving the hydrogen-assisted
FCG behaviour of the
WM for this pipe-line
weld. However,
transgranular
fatigue fracture was found in the CGHAZ of FW-6-H
specimen [Fig. 10(f)]. The improved resistance to crack
growth in the CGHAZ of FW-6-H specimen, as shown
in [Fig. 8(b)], could be related to the effective reduction
of quasi-cleavage
fracture in this region.

4. Conclusions
The sulfide stress corrosion cracking and fatigue crack
growth of welded TMCP API 5L X65 pipe-line
steel
have been studied in this work. Following
conclusions
are drawn:
1. Ferrite and pearlite mixed structures were found in the
as-received
steel plate. Clearly
banded
structures
which would deteriorate the mechanical properties of
the steel plate reappeared after tempering at 6OOC/2

I12

Fig. IO. SEM fractogrdphs


CGHAZ of FW-6-H.

L. W. T.sq et (11./International

of fatigue-fractured

specimens

Journal of Fatigue 23 (2001) 103-l 13

(a) TL, (b) WM of FW, (c) TL-6-H.

h. The microstructure
of weld metal (WM) consisted
of grain boundary
ferrite, Widmanstatten
ferrite,
acicular ferrite and a few microphases. Bainite with a
small amount of Widmanstatten
ferrite were observed
in the coarse-grained
heat-affected
zone (CGHAZ).
The fine-grained
heat-affected
zone (FGHAZ) comprised of fine-grained
ferrite and pearlite.
2. The most hardened region was located at the CGHAZ,
and the WM had a slight lower hardness than the base
metal (BM). The relatively lower hardness in the WM
in essence could be attributed to the inherent low
alloy content and low hardenability,
resulting in the
tensile fracture of welds located at this region. Postweld heat treatment (PWHT) of the weld had only
a minor influence on the changes of microstructures
and hardness.
3. Tensile properties of steel plates were found to be
superior to welds tested either in air or in an H,Ssaturated solution. After tempering at 6OOC/2 h, the
improved tensile properties of the weld tested in an
H,S-saturated
solution
were
associated
with
a
reduction in the amount of surface cracks in the WM.

(4 WM of FW-H,

(e) CGHAZ

of N-H

and (f)

4. The WM and HAZ of the as-welded weldment possessed a higher resistance to crack growth than the
steel plate. The reappearance
of banded structure in
the steel plate after tempering
treatment resulted in
an enhancement
of crack growth as compared to the
as-received
steel plate, as AK was greater than 22.4
MPa.\im. The effect of hydrogen seemed to be diminished as AK increased. This can be attributed to the
relatively slow hydrogen diffusion to the crack tip as
compared to the fast FCGRs at high AK. In addition,
the acceleration
of fatigue
crack
growth
rates
(FCGRs)
was rather significant
for the hydrogencharged untempered
weld. Meanwhile, the resistance
to crack growth in the WM did not improve for the
hydrogen-charged
tempered weld. However, the tempered HAZ could retard the fatigue crack growth as
compared to the WM of a hydrogen-charged
tempered weld.
5. Fractographic
observations
of tensile fractured and
fatigue-fractured
surfaces by SEM revealed quasicleavage fracture in the embrittled region.

L.W. Tsay et al. /International

Acknowledgements
The authors gratefully acknowledge the financial support of the Republic of China National Science Council
(Contract No. NSC 87-2216-E-019-008).

References
111 Taira T, Tsukada K, Kobayashi Y, Inagaki H, Watanabe T. Corrosion 1981;37(1):.5.
VI Moore EM, Warga JJ. Mat Perf 1976;15(6):17.
H, Bardou
G, Charbonnier
JC. Corr Sci
131 Margot-Marette
1987;27(10/11):1009.
141 Matsumoto K, Kobayashi Y, Ume K, Murakami K, Taira K, Arikata K. Corrosion 1986;42(4):337.
151 Grange RG. Met Tram 1971;2(2):417.
Y, Ume
K, Hyodo
T, Taira
T. Corr
Sci
161 Kobayashi
1987;27(10/11):1117.
171 Lee HL, Chan SLI. Mats Sci Eng 1991;A142(2):193.
181 Tau L, Chan SLI. Mats Letters 1996;29(1-3):143.
191 Tau L, Chan SLI, Shin CS. Corr Sci 1996;38(10):1.
[lOI Yurioka N. Weld in the World 1996;35(6):375.
1111 Tsukada K, Ohkita T, Ouchi C, Nagamine T, Hirabe K, Yako K.
Nippon Kohan Tech. Report, Overseas no. 35, 1982;24.

Journal of Fatigue 23 (2001) 103-113

113

1121 Tamehiro H, Yamada N, Matsuda H. Tram ISIJ 1985;25(1):54.


1131 Tsukada K, Yamazaki Y, Kosuge S, Tokunaga T, Hirabe K. Nippon Kokan Tech. Report, Overseas no. 35, 1982;35.
[141 Watanabe I, Suzuki M, Tagawa H, Kunisada Y, Yamazaki Y,
Iwasaki
N. Nippon Kokan Tech. Report, Overseas
no. 47,
1986;52.
A, Stefan B. Metal1 Mats 1990;28(6):426.
1151 Vyrostkova
[161 Hardie D, Liu S. Corr Sci 1996;38(5):721.
1171 Tsay LW, Huang WB, Li YM, Chen C. J Mats Eng Perf
1997;6(2): 177.
[181 Hippeley CA, Lane CE. Mats Sci Tech 1990;6(8):735.
1191 Tsay LW, Huang WB, Chen C. Fat Fract Eng Mat Struct
1997;20(7):1033.
1201 Chuang JH, Tsay LW, Chen C. Int J Fat 1998;20(7):531.
[211 Turn Jr JC, Wilde BE, Toribio CA. Corrosion 1983;39(9):364.
1221 Ikeda A, Kaneko T, Ando Y. Corr Sci 1987;27(1&11):1099.
v31 Kindlein W Jr, Schilling PT, Schroeder RM, Miiller IL. Corr
Sci 1993;34(8):1243.
v41 Tsay LW, Chem TS, Gau CY, Yang JR. Int J Fat 1999;21(8):857.
~251 Residual stresses and their effect. The Welding Institute, 1981.
WI Kang KJ, Song JH, Earmme YY. Fat Fract Eng Mat Struct
1990;13(1):1.
1271 Beghini M, Bertini L, Vitale E. Fat Fract Eng Mat Struct
1994;17(12):1433.
WI Shi YW, Chen BY, Zhang JX. Eng Fract Mech 1990;36(6):893.
WI Toribio J. J Mats Sci 1993;28(9):2289.
[301 Toribio J. J Mats Sci Eng 1996;A129(1-2):180.

You might also like