ARTICLE IN PRESS

International Journal of Pressure Vessels and Piping 83 (2006) 826834

www.elsevier.com/locate/ijpvp

Assessment of a creep-fatigue crack initiation and crack propagation for

a welded cylindrical shell
Hyeong-Yeon Lee, Jong-Bum Kim, Seok-Hoon Kim, Jae-Han Lee
Mechanical Engineering Division, Korea Atomic Energy Research Institute, 150 Dukjin-dong, Yuseong-gu, Daejeon 305-353, Republic of Korea

Abstract
An assessment of the creep-fatigue crack bahaviour for a cylindrical structure with weldments has been carried out by using a
structural test and an evaluation with an assessment procedure. The structural specimen with a diameter of 600 mm and thickness of
7 mm is a welded cylindrical shell made of 316L stainless steel (SS) for one half of the cylinder and 304 SS for the other half. Eight
articial defects were machined and the defect behaviours were examined. In the creep-fatigue test, the hold time was 1 h at 600 1C and
the primary nominal stress was 45 MPa. The evaluation results for the creep-fatigue crack initiation and crack propagation were
compared with those of the observation images from the structural test. The assessment results for the creep-fatigue crack behaviour by
using the French A16 procedure showed that the A16 guide is reasonably conservative but overly conservative for the creep-fatigue crack
propagation in the case of a short hold time. It was shown that the crack initiation and propagation were dominated by a creep.
r 2006 Elsevier Ltd. All rights reserved.
Keywords: Creep-fatigue; Crack initiation; Crack propagation; Weld; Stress intensity factor

1. Introduction
An evaluation of the creep-fatigue crack behaviour is
one of the key factors in the design and assessment of a
high-temperature structure such as a liquid metal reactor
(LMR) subjected to a high temperature above 500 1C. The
design of a reactor structure under a creep-fatigue load can
be carried out by using high temperature design codes such
as ASME-NH [1], RCC-MR [2], and DDS [3], which
provide the linear damage summation rule on an evaluation of the creep-fatigue damage, mainly based on the
simplied inelastic analysis methods.
For a defect free structure operating at high temperature,
ASME-NH [1] provides the procedures of an elastic
analysis and an inelastic analysis for the creep-fatigue
damage limits. In the design by the analysis rule of RCCMR RB-3200 [2], the evaluation procedure for the zones
with geometrical discontinuities is provided for a fatigue
initiation but no procedure is given as yet for a creepfatigue crack initiation. Instead, the A16 [4] guide which
Corresponding author. Tel.: +82 42 868 2956; fax: +82 42 861 7697.

E-mail address: hylee@kaeri.re.kr (H.-Y. Lee).

0308-0161/$ - see front matter r 2006 Elsevier Ltd. All rights reserved.
doi:10.1016/j.ijpvp.2006.08.009

has been included in the RCC-MR 2002 [2] edition

provides the procedure for a creep-fatigue assessment for
a structure with a defect. In the present study, the A16
procedure was employed for an evaluation of the creepfatigue damage, creep-fatigue crack initiation and propagation for a cylindrical shell with defects. The measured
temperature data was used in the assessment. The A16
guide provides the defect assessment procedures based on
the fracture mechanics parameters KI, J, C* or the stresses
at a characteristic distance, d from the crack. In this study,
the conservatism of the A16 procedure for a creep-fatigue
crack initiation and propagation for the base metal and the
weld metal has been quantied for the present cylindrical
shell specimen. The creep-fatigue load was applied 500
times with a hold time of 1 h and the aspects of the creepfatigue damage and the creep-fatigue crack behaviour were
observed with an optical microscope periodically at every
100th cycle. The results from the assessment by using
the A16 procedure were compared with those from the
observation.
This study was carried out as part of the works to
develop the liquid metal reactor Korea Advanced LIquid
MEtal Reactor (KALIMER) [5].

ARTICLE IN PRESS
H.-Y. Lee et al. / International Journal of Pressure Vessels and Piping 83 (2006) 826834

827

of this type were used for the thermal ratchetting structural

tests [69]. The two plates of the 304 and 316L materials
were welded as shown in Fig. 2(b) with the welding
methods of a shielded metal arc welding (SMAW) and a
gas tungsten arc welding (GTAW).
For the specimen used in this study, there are similar
metal welds between the 304 and 304, and between the
316L and 316L while there are dissimilar metal welds
between the 304 and 316L at the mid-interface. The
behaviour of the two welding methods under a creepfatigue load is examined for the similar and the dissimilar
metal welds.
Eight defects were machined by an electrical discharge
machining (EDM) as shown in Fig. 2(b). In Fig. 2(b), each
defect has the dimensions of a height of 0.3 mm and a
length of 40 mm. Defect number (#) 1 and #5 in Fig. 2(b)

2. Structural test
2.1. Structural specimen
The creep-fatigue structural test facility used in present
study is shown in Fig. 1. The test facility is composed of a
hydraulic actuator of a 1MN capacity and a high frequency
induction heater with a capacity of 50 kW [6]. The
inductance coil has six turns as shown in Fig. 1(b), and
its outer diameter is about 650 mm with clearances of
around 15 mm between the test cylinder and the coil.
The structural test specimen has the dimensions of a
600 mm diameter, a 500 mm height and a 7 mm thickness.
The specimen is a welded cylindrical shell with one half a
304 stainless steel (SS) shell and the other half a 316L SS
one as shown in Fig. 2. Similar cylindrical shell specimens

Fig. 1. Creep-fatigue test facility.

90

SMAW

304

GTAW

GTAW

0
60

270

304

GTAW

304
5

SMAW
SMAW 8

GTAW 4

2
1

316L
SMAW

GTAW

316L

316L

(b)
Fig. 2. Creep-fatigue test model and its development gure.

500mm

500

200

304

(a)

360

ARTICLE IN PRESS
H.-Y. Lee et al. / International Journal of Pressure Vessels and Piping 83 (2006) 826834

828

are the surface crack type defects with a half penetration of

3.5 mm (thickness is 7 mm) uniformly along the circumferential direction from the inside surface and all the other
defects are the through-wall type. Defects of numbers #4
and #8 are machined along the mid weld interface line
between the 304 and 316L shell. The tips of each defect are
located at both the weld metal or base metal, no tips were
located at the heat-affected zone (HAZ). The specimen was
welded with the welding conditions of a current of
110120 A, voltage of 2630 V, heat input of 2.02.5 KJ/
mm, and a speed of 80150 mm/min. The ller metal of
ER316L SS was used as weld metal for all the welded
joints.

30 1C during a steady state at 600 1C. The measured

temperature data has been used in the thermal analysis.
As shown in Fig. 3, it took about 9 min to reach 600 1C
from 70 1C and 21 min to cool down again to 70 1C at a mid
part of the specimen. Mechanical load of 60 tons has been
applied in the axial direction and the axial nominal stress
is 45 MPa. So, one cycle with 1 h of a tensile hold time is
about 90 min as shown in Fig. 3.

2.2. Load conditions

The ABAQUS [10] nite element model for the test

specimen is a 3D half symmetric model as shown in Fig.
4(a) with four notches of defects #1#4 of Fig. 2(b). A total
of 53,760 eight-node linear brick elements and 73,228
nodes were used. The element size in front of the tip in
horizontal axis direction is selected to be 0.24 mm so that
the Gaussian point will be located exactly at 0.05 mm
( 50 mm) from the notch tip. The height of each notch
having a root radius of 0.15 mm is 0.3 mm and the semicircular shaped notch was modeled with four elements in
the vertical direction. The size of the tip element
(parallelogram shape) is 0.24 mm
0.015 mm. Thermal
loads shown in Fig. 3 and a mechanical stress of 45 MPa

The cylindrical specimen of Fig. 2(a) is subject to a cyclic

thermal load and a steady mechanical load. The specimen
is heated up to 600 1C and a hold time of one hour is
applied under a load control condition. The inductance coil
covers the mid 200 mm part of the specimen as shown in
Fig. 2(a). The gaps between the inductance coil and the
specimen along the circumferential direction were carefully
adjusted so that a uniform temperature could be obtained
at the central part of the specimen in the axial direction.
During an observation, the temperature variation over the
central part under heating was measured to be less than

3. Assessment of the creep-fatigue crack behaviour

according to A16
3.1. Finite element modeling

T(C)

600
HT=1hr

70
9

69

90

t (min)

1 cycle (90min)

Fig. 3. Thermal load conditions.

P=60ton
(n=45 MPa)

2
3

(a)

(b)
Fig. 4. Load and boundary condition of FE model, and principal stress distribution (300
).

ARTICLE IN PRESS
H.-Y. Lee et al. / International Journal of Pressure Vessels and Piping 83 (2006) 826834

were applied, and the bottom surface was xed as a

boundary condition as shown in Fig. 4(a). Among the four
defects, there is one surface defect (#1), one vertical defect
(#2) and the others (#3, #4) are horizontal through wall
defects. The distribution of principal stresses that are
Rankine equivalent stresses by the sd method at the end of
a 1-h hold time near the tip are shown in Fig. 4(b) whose
deformation scale factor is 300.
3.2. Assessment according to the A16 procedure
3.2.1. Creep-fatigue crack initiation
In the French RCC-MR code [2,4], the geometrical
discontinuities are assimilated with cracks for a creepfatigue estimation based on an elastic analysis. The design
by analysis rule in RB-3200 of RCC-MR provides a rule
for a fatigue in zones with geometrical discontinuities but it
does not provide one for a creep-fatigue crack initiation.
Instead, the A16 guide [4] which is a technical appendix of
the RCC-MR can be employed for an evaluation of the
creep-fatigue damage for a component with geometrical
discontinuities.
The creep-fatigue crack initiation of the A16 procedure
is based on the sigma-d(sd) method. The principal of this
procedure is to determine a stress and a strain at a
characteristic distance d from the crack tip and to
compare them with the material fatigue curve and the
creep strength data [4,11,12]. The distance d is specied as
0.05 mm for austenitic SSs [4] whose maximum specied
tensile strength at room temperature is less than 600 MPa.
Evaluation of the fatigue crack incubation usage factor
(A) is obtained by the ratio of the specied number of
cycles to the number of cycles prior to a fatigue initiation.
Evaluation of the creep crack incubation usage factor (W)
is obtained by the ratio of the specied duration of the
hold time t to the time prior to a creep initiation
T determined from the usual creep rupture property. If
the point of a coordinate for the fatigue damage and creep
damage lies inside the creep-fatigue interaction diagram,
it means that the crack does not initiate over the period t
under investigation. That is, the creep-fatigue damage
envelope in the A16 procedure means the crack
initiation line.
In this study, a ne mesh with the smallest element size
of 0.24 mm near the notch tip was used so that the
Gaussian point would be located at d ( 50 mmm) from
the notch tip in the plane of the crack. Then the elastically
calculated Rankine equivalent stress (the largest principal
stress) at distance d from the tip is determined by using an
elastically calculated stress, sde and the CreagerNeuber
rule [4] to take a plastic effect into consideration. The stress
eld near the notch tip is shown in Fig. 4(b).
The real strain range (D
) is obtained by adding the
strain range due to a plasticity (D
elpl ) and the strain range
due to a creep (D
fl ) as in Eq. (1) [4].
D
D
elpl D
fl .

(1)

829

Here the elasticplastic strain range (D
elpl ) is determined by adding the four terms according to the A16
procedure(1) strain range by an elastic analysis, (2)
plastic strain increase due to the primary stress range at the
point examined, (3) plastic strain increase according to the
Neuber rule, and (4) a plastic strain increase due to a
triaxiality.
The creep strain range (D
fl ) is the one due to a hold time
during one creep-fatigue load cycle. The creep strain (e)
formula for 304SS is given as the Blackburn type of Eq. (2)
and that for 316L is the BaileyNorton type of Eq. (3) [4].

fl
t 1  ert
x 1  est
m t,

(2)

fl c1 tc2 sn1 ,

(3)

where the 22 coefcients for 304 SS (in formula or constant

type) embedded in Eq. (2) given in technical Appendix A3
[12] of RCC-MR, and C1 2.966
1011, C2 5.129

101 and n1 4.125 for 316L in Eq. (3). The creep strain at
each tip of defect #3 due to one hour of a creep hold time
was calculated to be e 0.257 (%) for the base metal of
304 SS, and e 0.159 (%) for the weld metal of ER316L.
The real strain range (D
) in Eq. (1) was calculated as
0.00861 for the base metal notch (304 SS) and that for the
weld metal notch (ER316L) was 0.00750. The number of
cycles to a crack initiation has been determined from the
fatigue curve by using a strain range equal to (D
)/K at
600 1C where K is 1.5 according to the A16 procedure.
Then the cycles to a creep-fatigue crack initiation were
calculated as 572.5 for the base metal and 170.0 for the
weld metal for defect #3.
The sustained stress, skd at location d which causes a
creep crack initiation was calculated as 228.5 MPa for
defect #3. The creep rupture time for this stress was
calculated to be 13.2 h at the notch tip of the base metal
while it was 8.7 h at the notch tip of the weld metal.
Therefore, the creep-fatigue damage envelopes for the base
and weld metal are described as
n
t

,
(4)
A W 304 SS;base
572:5 13:2
n
t

.
(5)
170:0 8:7
It should be noted that an assessment according to the
A16 guide for a creep-fatigue damage provides less damage
for a base metal than a weld metal as shown in Eqs. (4) and
(5). The present creep-fatigue problem is creep dominant
and a crack initiation would occur mainly due to creep (W)
as shown at point P in Fig. 5.
For the weld metal part, the creep-fatigue damage
according to the A16 procedure is calculated as follows.
The allowable stress of the base material Sr is replaced by
JrSr where Jr is the coefcient of the weld rupture
characteristics. As for the fatigue, the curve is divided by
the ordinates D
of the reference curve for the base metal
by the coefcient of the weld fatigue characteristics, Jf.
Here Jr is 0.94 and Jf is 1.25. The fatigue endurance limit
A W 316L SS;weld

ARTICLE IN PRESS
H.-Y. Lee et al. / International Journal of Pressure Vessels and Piping 83 (2006) 826834

830

where q is the closure (Ro0) and mean stress (R40)

coefcient, E n is E for the plane stress and E/(1n2) for the
plane strain, and R is the minimum to maximum load ratio.
The increment of a fatigue propagation for each cycle type
i at a base metal defect #3 is calculated from the following
formula, which is commonly used for the base metal of
304SS and the weld metal of ER316L;

n
dafa i C DK eff i ,
(7)

25
Weld_200

W (creep damage)

20

Base_200
15
Weld_100

where (DKeff) is the effective SIF range for cycle type i,

C 6
108 and n 3.3 [4]. Then the increment of a
fatigue crack growth for the defect #3 at base metal
(304 SS) is given as Eq. (8).

10
Base_100
5
Creep-fatigue damage envelope
P

1
0
0

2
3
A (fatigue damage)

Fig. 5. Evaluation results for creep-fatigue damage for base and weld
metal.

and the rupture time for the weld metal were calculated as
Nd 170, Td 8.7, respectively. Thus the creep-fatigue
crack initiation envelope is given as in Eq. (5), which means
that a creep-fatigue crack would initiate in about eight
cycles and the initiation would be dominated by a creep
rather than a fatigue.
As the number of cycles is added, the calculated creepfatigue damage at the weld metal and base metal of defect
#3 are shown in the creep-fatigue damage envelope
in Fig. 5. The assessment results according to the A16
procedure shows a reasonable conservatism when compared with the observed results (described in next section)
between 100 and 200 cycles. For the present assessment
according to A16, it should be noted that the creep-fatigue
damage for the weld metal was larger due to the weld
strength reduction factors although the total strain range
was less.
3.2.2. Creep-fatigue crack propagation
A creep-fatigue crack propagation should be evaluated if
the number of cycles for the creep-fatigue crack initiation is
less than the actual load cycles. In the A16 procedure,
creep-fatigue crack propagation is calculated by adding the
crack increments due to a fatigue and a creep.
In the calculation of the fatigue crack growth (dafa), the
maximum effective stress intensity factor (SIF) range
should be determined to calculate the updated size of the
defect. The fatigue crack growth is estimated from the Paris
law with a SIF range of DKeff derived from a simplied
cyclic J value of DJ during the cycle based on the reference
stress concept and a factor q for the closure effects.
p
DK eff q E  DJ ,
(6)

dafa i;base 0:0027 mm;

(8)

dafa i;weld 0:0025 mm:

(9)

In the mean time the increment of a fatigue crack growth

for the defect #3 weld metal (ER316L) can be similarly
calculated by taking into account the difference of the
material properties such as the Youngs modulus. Then the
result for the weld metal was found to be slightly less than
that of the base metal as shown in Eq. (9). So the total
fatigue crack propagation (after initiation at 13 cycles) up
to 500 load cycles is calculated as 1.35 mm for the base
metal of 304 SS while it is 1.25 mm for the weld metal.
In the calculation of the creep crack growth (da), the
C n t should be determined during the hold time by taking
into account the updated time and by using the creep crack
propagation law. Then the updated size of the defect is
determined as a+dafa+da.
The amount of creep propagation during the hold time
tmi is calculated from the following formula:
Z ti tmi

q
dafl i
A C i t dt,
(10)
ti

C i t

is the C n integral whose formula is provided in

where
A16 for typical structures, A 8.05, q 0.81 [4] and ti is a
time to cycle type i and tmi is a holding time. The calculated
results of Eq. (10) shows that the amount of creep crack
growth per cycle for the base metal and weld metal are
calculated as follows.
dafl i;base 0:49 mm;

(11)

dafl i;weld 0:23 mm:

(12)

It is interesting to see that the increment of the creep

crack growth for the weld metal is less than half of the base
metal as shown in Eqs. (11) and (12). The differences
of the two values are mainly caused by the differences of
the creep properties between 304 and 316L as shown
in Eqs. (2) and (3).
However, it should be noted that this increment for the
weld metal is also overly conservative for the case of 304SS
for a hold time of 1 h. The measured temperature data were
used for the present assessment. The crack propagation
characteristics showing a greater crack growth for the base

ARTICLE IN PRESS
H.-Y. Lee et al. / International Journal of Pressure Vessels and Piping 83 (2006) 826834

metal than the weld metal are in agreement with the

observed images which is discussed in next section. Based
on the present assessment, it is concluded that the A16
procedure provides overly conservative results for a creepfatigue creep crack propagation in the case of a short hold
time like one hour in the present case.
4. Observation of a creep-fatigue damage
The creep-fatigue damage of the test model has been
inspected with an optical microscope intermittently after
every 100th cycle in a non-destructive way. Twelve points
near the notch tips of the 8 defects were observed. Various
creep-fatigue damage aspects at or near the notch tips were
investigated during the inspection. The behaviour of the
defect geometries (vertical and horizontal defects), and that
of the two welding methods (GTAW and SMAW) under a
creep-fatigue load have been examined.
An inspection by using the X-ray diffraction method was
carried out to see if there are any internal weld defects
along all the weld lines of the specimen. Tiny internal weld
defects were found at a few locations but the eight defects
were machined far enough away from the internal defects
so that their effects would be negligible.
The images for the vertical defects of #2 and #6 of Fig. 2
with a magnication factor of 350 after 200 cycles show
that a small amount of damage has occurred at the upper
part of the defects as shown in Fig. 6. When comparing the
two welding methods, the image by SMAW was slightly
more damaged than that by GTAW. Since the mechanical
stresses are acting in the axial direction, the damage for
the vertical defects is expected to be smaller than that
for the horizontal defects.
The damage aspects of the horizontal defects are
comparable. The images in Fig. 7 are for the part through
defects #1 (SMAW) and #5 (GTAW) after 300 cycles,
which show that there is no remarkable damage, and the

831

degree of damage for SMAW and GTAW is not

distinguishable.
The images of the horizontal defect #3 for 304 SS are
shown in Figs. 8 and 9. The yield strength of 304, 316L and
the ller metal made of ER316L at room temperature is
274, 231 and 258 MPa, respectively, which means that there
is a strength under-match between 304 and the weld metal
(ER316L) near defect #3. The strength data was measured
by using a plate type material specimen sampled from the
same structural material. It should be noted that the
surface is usually hardened due to cold work or a surface
machining so that its strength on the surface is usually
higher than the above values, and thus the fracture
toughness would be slightly reduced.
Defect #3 was the most damaged one among the eight
defects examined. The observed images for defect #3
showed that the crack propagated as much as 0.22 mm for
the weld metal (SMAW) and 0.71 mm for the base metal as
the number of cycles reached 500 cycles as shown in Figs.
810. The observed images have shown that the cracks

Fig. 8. Observed image of the horizontal defect #3 (weld metal, 350
).

Fig. 6. Observed image of the vertical defects (#2, #6; 200 cycles, 350
).

Fig. 7. Observed image of the surface defects (#1, #5; 300 cycles, 350
).

Fig. 9. Observed image of the horizontal defect (no. 3, 350
).

ARTICLE IN PRESS
H.-Y. Lee et al. / International Journal of Pressure Vessels and Piping 83 (2006) 826834

832

were mainly propagated near the surface without a

penetration.
Usually a welded joint is known to be weaker than a base
metal under a creep-fatigue loading. However, it is
interesting to see that the crack propagation length for
the base metal is larger than that of the ller metal as
shown in Fig. 10. The reasons for this nding are many
factors. First, the creep rupture strength of the weld metal
is higher than the base metal. Fig. 11 shows the minimum
stress-to-rupture curve for 304 and 316 in ASME-NH [1],
and 304LN and 316LN in RCC-MR [2]. The differences in
the creep strength become more distinguished as the hold
time increases. As the accumulated hold time is increased
from 1, 10, 100 and 1000 h, the differences in the rupture
stress becomes 19%, 28%, 34% and 38% for ASME-NH

while they become 17%, 36%, 59% and 84%, respectively

for RCC-MR. Therefore, it can be concluded that the main
reason was caused by the differences of the creep rupture
strength between the two materials. It is shown that the
weld metal can be stronger than the base metal for a creepfatigue loading if it has a higher creep rupture property.
Second, the fracture toughness of the ller metal is
higher than that of the base metal so that once a crack is
initiated at both notches, a crack for the base metal would
be propagated faster. The image of Fig. 12 observed at the
weld line of 304-316L after 100 cycles shows that the 304
SS is weaker than the ER316L. And it is important to note
that no internal or external weld defects near defects #3
were conrmed through the X-ray diffraction method
before the test started. When the images are magnied, it

0.8
base metal (304)

0.7

weld metal (316L)

0.5
0.4
0.3
0.2
0.1
0
100

200

300
400
No. of cycles

500

600
Fig. 12. Observed images of 304 and 316L steel near the weld defect #8
(after 100 cycles, 1050
).

Fig. 10. Aspects of crack propagation observed for defect #3.

400

400

350

creep rupture stress (MPa)

creep rupture stress (MPa)

Length of defect (mm)

0.6

304LN
316LN

300
250
200
150

304

300

316
250
200
150
100

100
0

200

400

600

800

1000

minimum time to rupture (hr)

(a)

350

RCC-MR[2]

200

400

600

800

minimum time to rupture (hr)

(b)

Fig. 11. Minimum stress-to-rupture curve.

ASME-NH[1]

1000

ARTICLE IN PRESS
H.-Y. Lee et al. / International Journal of Pressure Vessels and Piping 83 (2006) 826834

833

Fig. 13. Observed image of the horizontal defect (#8, 350
).

was shown that the micro-crack was propagating in the

intergranular mode. It should be noted that the base metal
is not always stronger than the weld metal.
The observed images for defect #8 which is located along
the weld interface line of 304 and the ER316L materials
show that the damage for a GTAW is less damaged than
that for a SMAW as shown in Fig. 13. It is also worth
noting that defect #8 which was machined along the weld
interface line is stronger than the base metal part of 304SS
at defect #3. In a dissimilar metal weld between 304 and
316L, the GTAW method was observed to be more
resistant to a creep-fatigue load than the SMAW method
as shown in Fig. 13.
5. Conclusions
In this study, a creep-fatigue structural test with a welded
cylindrical shell has been carried out and the test results
were compared with those by the French assessment
procedure of RCC-MR A16. The structural specimen has
the dimensions of a 600 mm outer diameter, a 500 mm
height and a 7 mm thickness. The specimen is a welded
cylindrical shell with one half a 304 SS shell and the other
half 316L SS by using the welding methods of a SMAW
and a GTAW.
The assessment results for the present creep-fatigue
problem according to the A16 procedure showed that the
creep-fatigue crack initiation occurred after about 13
creep-fatigue load cycles for the base metal of 304 SS and
8 cycles for the weld metal of 316L SS at defect #3.
As for the creep-fatigue crack propagation for the base
metal notch part of defect #3, the evaluation results
according to A16 procedure showed that the increment of a
fatigue crack growth per cycle was 0.0027 mm while that of
a creep crack growth under one hours hold time was
0.49 mm. As for the creep-fatigue crack propagation for the
weld metal, the fatigue crack growth increments was

0.0025 mm and creep crack growth increments was

0.23 mm. These results show that a crack propagation is
dominated by a creep crack growth. The crack propagation
for the base metal was larger than that for the weld metal.
This result is attributed to the superior creep behaviour of
316L to 304.
The creep-fatigue damage was observed with a portable
optical microscope intermittently at every 100th cycle nondestructively. The observed images showed that a creepfatigue crack initiation occurred at cycles between 100 and
200 for defect #3 of the base metal (304 SS). In addition,
the damage for the 304 SS part was measured to be more
severe than that of the weld metal (ER316L). The
horizontal defects were weaker than the vertical defects
because the primary stresses were acting axially. The
amount of creep-fatigue crack propagation of the crack
tip for the 304 SS base metal (defect #3) was measured to
be 0.7 mm after 500 load cycles. The present creep-fatigue
problem is creep dominant and a crack initiation and
propagation occurred mainly due to a creep. The images of
the damage at the tips of the notch showed that a crack
propagation occurred in an intergranular cracking mode. It
is shown that crack initiation and propagation is dominated by creep rather than fatigue.
The behaviour of the two welding methods of a
SMAW and a GTAW under a creep-fatigue load was
examined. From the creep-fatigue test upto 500 load
cycles, the GTAW method is observed to be slightly more
resistant to the creep-fatigue loads than the SMAW
method.
When the results from the assessment according to the
A16 procedure and structural test are compared, it is found
that the assessment procedure of A16 is reasonably
conservative for a creep-fatigue crack initiation while it
seems overly conservative for a creep-fatigue crack
propagation in the case of a short hold time like one hour
in the present study.

ARTICLE IN PRESS
834

H.-Y. Lee et al. / International Journal of Pressure Vessels and Piping 83 (2006) 826834

Acknowledgements
This work was performed under the long-term nuclear
R&D program sponsored by the Ministry of Science and
Technology of Korea.
References
[1] ASME Boiler and Pressure Vessel Code, Section III. Rules for
construction of nuclear power plant components, Div. 1, Subsection
NH, Class 1 Components in Elevated Temperature Service. New
York: ASME; 2004.
[2] Design and Construction Rules for Mechanical Components of FBR
Nuclear Islands, RCC-MR, 2002 Edition, AFCEN; 2002.
[3] DDS. Structural design guide for class 1 components of prototype
fast breeder reactor for elevated temperature service. Japan: JAPC;
1998.
[4] Technical Appendix A16 of RCC-MR. Guide for leak before break
analysis and defect assessment. AFCEN; 2002.

[5] Hahn DH, et al. KALIMER conceptual design report. KAERI/

TR-2204. Daejeon: Korea Atomic Energy Research Institute;
2002.
[6] Lee HY, Kim JB, Lee JH. Evaluation of progressive inelastic
deformation for the welded structure induced by spatial variation of
temperature. Int J Pressure Vessel Piping 2004;81(5):43341.
[7] Lee HY, Kim JB, Lee JH. Thermal ratcheting deformation for 316L
stainless steel cylindrical structure under moving temperature
distribution. Int J Pressure Vessel Piping 2003;80:418.
[8] Lee HY, Kim JB, Lee JH. Progressive inelastic deformation
characteristics of cylindrical structure with plate-to-shell junction
under moving temperature front. Korean Soc Mech Eng Int J
2003;17(3):40311.
[9] Kim JB, Lee HY, Park CG, Jeon GP, Lee JH. Creep-fatigue damage
evaluation of the 316SS Y-junction structures in a liquid metal
reactor. In: Proceedings of ICAPP 04, Paper 4323, Pittsuburgh, PA,
USA, June 1317, 2004.
[10] ABAQUS Users manual, Version 6.5, H.K.S. USA; 2005.
[11] Drubay B. A16: guide for defect assessment at elevated temperature.
Int J Pressure Vessel Piping 2003;80:499516.
[12] Subsection Z: Technical Appendix A3, RCC-MR, AFCEN; 2002.