CRACK GROWTH BEHAVIOR IN THE INTERFACE OF REPAIR WELDED
1Cr-0.5Mo STEEL
Un Bong Baek, Kee Bong Yoon, Hea Moo Lee
Korea Research Institute of Standards and Science
1 Doryong-Dong, Yuseong-Gu, Daejeon 305-340, Korea
ABSTRACT
Creep-fatigue crack growth behavior was experimentally measured particularly when a crack was located in the heat affected
region of 1Cr-0.5Mo steel. Load hold times of the tests for trapezoidal fatigue waveshapes were varied among 0, 30, 300 and
3,600 seconds. Time-dependent crack growth rates were characterized by the Ct-parameter. It was found that the crack
growth rates were the highest when the crack path was located along the fine-grained heat affected zone(FGHAZ). Cracks
located in other heat affected regions had a tendency to change the crack path eventually to FGHAZ. Creep-fatigue crack
growth law of the studied case is suggested in terms of (da/dt)avg vs. (Ct)avg for residual life assessment.
1. Introduction
Fracture of the high-temperature operating facilities frequently occurs at welds due to crack growth. In details, nearly all
welding cracks initiate and grow at the fragile heat-affected zone(HAZ) between parent and weld metals. Thus, fracture
mechanical investigations on the weld interface crack are very important in evaluating remnant lifetime of the high-temperature
operating facilities.
The crack growth rate of creep or creep-fatigue damages has been characterized using a Ct parameter in the high temperature
fracture mechanics. However, growth of the weld interface crack between two different materials cannot be easily explained
with the Ct parameter developed for monolithic materials. Thus, instead of the conventional lifetime analysis in parent metal,
which can underestimate the crack growth rate in HAZ, several investigations for deriving fracture mechanical parameters for
the HAZ and weld metal have been performed for more exact explanations for the welds lifetime [1-3]. Excepting this static
loading condition above, studies for dynamic fatigue loadings are also needed for the creep-fatigue crack growth (CFCG). In
this study, welds including weld metal and HAZ was formed by adopting the repair welding procedure for 1Cr-0.5Mo steel as
one of the low alloy heat-resistance steels. After introducing a crack inside of the HAZ, its growth behavior was investigated
under creep-fatigue loading environments.
2. Experimental procedures
2.1 Uniaxial tensile tests
Specimens sampled from retired boiler header components were welded according to AWS E8016-B2 standards. The boiler
header was operated during 180,000 h and its chemical composition is written in Table 1. In order to characterize mechanical
properties of the specimen material, the uniaxial tensile tests were performed at room temperature and 538℃. The creep test
was also performed at 538℃ and the specimen length direction was controlled to coincide with the boiler header length
direction. The diameter of the marker region in the tensile specimen is 6.25 mm and two displacement-control testing
procedures at ambient and high temperatures are followed ASTM E8 and ASTM E21 standards, respectively. True stress-true
strain relationships were derived from the load-displacement records obtained as results of the uniaxial tensile tests. By
assuming a power-law relationship between the plastic strain
the equation form
ε p = Dσ
m
εp
and the flow stress σ, the obtained flow curve was fitted in to
The determined plastic constants D and m are summarized in Table 2.
2.2 Deadweight creep tests
Creep specimens having diameter 6.25 mm were tested according to ASTM E139 standards procedure using a deadweight
creep testing system. The specimens were heated to 538℃ and maintained 5 h inside of a coil-type electric furnace. An
extensometer was attached on the specimen and displacement data were recorded from the displacement gage directly
connected to the extensometer. After converting time-strain curve from the time-displacement data obtained as the raw creep
data, creep strain rate was determined by selecting the linear secondary creep regime at each stress condition. By assuming a
power-law relationship between the creep strain rate
form
ε&ss = Aσ
n
ε&ss
and the applied stress σ, the creep data were fitted in to the equation
-n
-1
[4]. The determined creep constants A and n were 1.235E-24 MPa ·hr and 9.18, respectively.
Table 1 Chemical composition of 1Cr-0.5Mo steel
Element
C
Si
Mn
P
S
Ni
Cr
Mo
Cu
Al
Fe
Wt. %
0.18
0.27
0.68
0.016
0.014
0.095
0.94
0.44
0.12
0.008
Bal.
Table 2 Tensile properties of 1Cr-0.5Mo steel at 24℃ and 538℃
Test Temp.
(℃)
24
538
Yield Stress
(0.2%) (MPa)
291
214
Tensile Strength
(MPa)
477
344
Elongation
(%)
35.6
32.9
Reduction
of Area (%)
72.8
79.0
Young's Modulus
(GPa)
188.0
127.9
D
(MPa-m)
4.48E-12
3.21E-17
m
3.80
5.96
Figure 1 Specimen location in welded block
2.3 Creep-fatigue crack growth tests
One C(T) specimen machined for the CFCG test is shown in Fig. 1; its crack is parallel to radius direction of the parent metal
and is controlled to grow along the HAZ. A pre-crack having its crack ratio 0.5 was introduced into the machined specimen and
exact location of the crack front inside of the HAZ was identified by mirror-like polishing and etching of the specimen side. Side
grooves were also machined on the specimen in order to confine the crack growth inside of the HAZ and to avoid the crack
tunneling phenomenon occurring by the CFCG test. The measurement of crack length was carried out using the DC voltage
drop method. Optimal positions of current input and voltage output were adopted from a previous research [5] and resulting
crack length was calculated by applying Johnson's equation [6]. When the crack length attained about 3-4 mm, the test was
terminated and the specimen was quenched into liquid nitrogen. The quenched specimen was fractured at low temperature
and its fractured surfaces were investigated using a traveling microscope. In order to measure crack growth amount at the final
point, the specimen thickness was divided into 8 equal sections and corresponding crack lengths were measured at 9 points.
Comparing the crack length measured with the predicted value, significant difference was modified [7]. In order to
In order to determine a period-dependent crack growth rate from the crack growth rate measured through a triangularwaveshaped fatigue test without load hold time, the fatigue tests were performed for 4 specimens by applying the triangular
waveshape having [loading time]/[unloading time] as 1/1 (unit: s). The fatigue crack growth results are summarized in Table 3.
Also in order to determine a load hold time-dependent crack growth rate, the fatigue crack growth tests were performed by
applying trapezoidal waveshapes having load hold times 30, 300 and 3600 s. The detailed testing conditions and results of the
CFCG are summarized in Table 4.
Table 3 Test conditions under triangular fatigue waveshapes
Spec.
no.
Rise time/
decay time
Max. load
(KN)
Min. load
(KN)
Total
cycles
A8
B1
B2
C1
1/1
1/1
1/1
1/1
5
5
5
7
0.5
0.5
0.5
0.7
4,035
5,450
5,400
19,270
Measured
Initial crack
(mm)
13.62
13.14
13.71
7.83
Measured
Final crack
(mm)
15.78
15.74
16.57
15.27
Predicted
Final crack
(mm)
15.51
15.55
16.39
15.46
Prediction
Error
(mm)
-0.27
-0.19
-0.18
0.19
Table 4 Test conditions under trapezoidal waveshapes
Rise time/
hold time/
decay time
1/30/1
1/30/1
1/30/1
1/300/1
1/300/1
1/300/1
1/3600/1
1/3600/1
1/3600/1
Spec.
no.
A0
A3
B7
A1
A6
C5
A7
B5
C11
Max. load
(KN)
Min. load
(KN)
Total
cycles
5
5
5
5
5
8
6.5
5.8
10
0.5
0.5
0.5
0.5
0.5
0.8
0.65
0.58
1
3,000
4,002
3,407
2,583
2,414
10,602
39
148
235
Measured
Initial crack
(mm)
13.15
13.10
13.42
13.17
12.83
6.79
13.50
13.14
8.53
Measured
Final crack
(mm)
15.33
16.45
16.23
15.88
15.26
14.23
14.79
15.26
12.53
Predicted
Final crack
(mm)
15.14
16.54
16.23
16.07
15.09
13.73
15.15
15.46
12.75
Prediction
Error
(mm)
-0.19
0.09
0.00
0.19
-0.17
-0.50
0.36
0.20
0.22
3. Results
3.1 crack growth under the triangular-waveshaped fatigue
The crack growth rate da/dN is plotted as a function of △K for the fatigue crack growth results from the triangular waveshape
loading in Fig. 2. Here the range of stress intensity factor △K is calculated using Eq. (1).
ΔK =
ΔP
BN
⎛a⎞
• F⎜ ⎟
W
⎝W ⎠
(1)
In the case of C(T) specimen, the crack shape function F(a/W) is expressed in the form of Eq. (2). a, W and BN is the crack
length, specimen width and effective specimen thickness, respectively. △P is the applied load range. When a/W is larger than
0.2, the testing is valid.
2
3
4
2 + a /W ⎡
⎛a⎞
⎛a⎞
⎛a⎞
⎛a⎞ ⎤
⎛a⎞
F⎜ ⎟ =
0.886 + 4.64⎜ ⎟ − 13.32⎜ ⎟ + 14.72⎜ ⎟ − 5.6⎜ ⎟ ⎥
1.5 ⎢
⎝ W ⎠ (1 − a / W ) ⎣⎢
⎝W ⎠
⎝W ⎠
⎝W ⎠
⎝ W ⎠ ⎦⎥
(2)
Through a linear regression of the data in Fig. 2, the period-dependent crack growth rate was estimated below (see solid line
in Fig. 2 and Eq. (3)).
da
= 0.27 × 10 −9 (ΔK )1.736
dN
(3)
Here units of da/dN and △K are m/cycle and MPa(m)1/2, respectively. Fatigue crack growth results obtained from the
triangular waveshape loading showed slight scattering behavior and followed the Paris law well.
3.2 crack growth under the trapezoidal-waveshaped fatigue
The crack growth rate da/dN versus △K data obtained from the trapezoidal waveshape loadings having load hold times 30,
300 and 3600 s are summarized and plotted in Fig. 3. Load hold time-dependent crack growth amount was assessed by
subtracting the triangular testing result (dotted line) from the trapezoidal testing result (solid line). In other words, the crack
growth difference between 1/3600/1 and 1/1 data becomes the time-dependent crack growth amount during 3600 s holding.
The time-dependent crack growth behaviors determined from Fig. 3 were summarized as the relationships between (da/dt)avg
and (Ct)avg and were plotted in Fig. 4. Fig. 4(a) displays the three testing results performed at the load hold time 30 s, Fig. 4(b)
and (c) display results obtained from two load hold times 300 and 3600 s, respectively. Here (da/dt)avg defined by Eq. (4) is
the crack growth per unit load hold time. (Ct)avg is defined by Eq (5).
1 ⎛ da ⎞
⎛ da ⎞
⎜ ⎟ = ⎜
⎟
⎝ dt ⎠ avg t h ⎝ dN ⎠ hold
(4)
t
(C t ) avg
1 h
= ∫ C t dt (5)
th 0
where th is the hold time at loaded state. The variable (Ct)avg was calculated from Eq. (6) by inputting the load, crack length
and material's creep constants. Eq. (6) is valid for materials displaying the elastic-secondary creep behavior.
Triangular Waveshape
1Cr-0.5Mo steel
at 538 oC
10-4
1Cr-0.5Mo steel
at 538 oC
Spec. No. C1
Spec. No. A8
Spec. No. B1
Spec. No. B2
da/dN (m/cycle)
da/dN (m/cycle)
10-5
10-6
Regression Line
1/30/1 waveshape
1/30/1 regression
1/300/1 waveshape
1/300/1 regression
1/3600/1 waveshape
1/3600/1 regression
10-5
Triangular waveshape
10-6
(da/dN) = 1.27x10-9(ΔK)1.736
-7
10
101
20
40
60
80
10-7
20
102
Δ K (MPa m1/2)
60
80
1/2
Figure 2 Regression results of fatigue crack growth rates
for 1Cr-0.5Mo steel at 538℃ under triangular waveshapes.
(a)
40
Δ K(MPa m )
Figure 3 Comparison of crack growth rates of 1Cr-0.5Mo
steel at 538℃ under trapezoidal and triangular waveshapes.
(b)
(c)
Figure 4 Time-dependent crack growth behavior for various load hold periods.
(a)30sec. hold (b) 300sec. hold (c) 3,600sec. hold
Figure 5 Time-dependent crack growth behavior of 1Cr-0.5Mo steel at 538℃
(a)
(b)
Figure. 6 Crack growth path (a) for C1 specimen under triangular fatigue waveshape (b) for C11 specimen under trapezoidal
waveshape with 3,600 sec hold time
(a)
(b)
(c)
Figure. 7 Crack damage at crack tip regions of various hold time specimens (a) 0 sec, (b) 300 sec (c) 3,600 sec
Figure. 8 Crack growth path when fatigue precrack tip is located exactly at FGHAZ region
(C t ) avg =
)
n −3
2
−
aαβ γ c (θ , n)
ΔK 4 F ′
(1 − ν 2 )
( EA) n −1 t h n −1 + C *
E
W F
(6)
where
2
n + 1 π (1 − ν 2 )
1 ⎡ (n + 1) 2 ⎤ n −1
)
n +1
,
, β = 1 / 3 , γ c (90 o , n) ≅ 0.38
=
α=
α
⎢
n
n +1 ⎥
n
In
2π ⎣ 2nα n ⎦
th is the load hold time and In is the non-dimensional constant depending on n [8]. C* is calculated from the already known Jintegral solution and can be formulated as Eq. (7) when the C(T) specimen satisfies the plane strain condition.
C* =
A
P ⎞
⎛ a ⎞⎛
⎟⎟
h
, n ⎟⎜⎜
n 1⎜
(W − a)
⎝ W ⎠⎝ 1.455ζB ⎠
n +1
(7)
1
⎡⎛ 2a ⎞ 2
⎛ 2a ⎞ ⎤ 2 ⎛ 2a ⎞
ζ = ⎢⎜
⎟ + 2⎜
⎟ + 2⎥ − ⎜
⎟ −1
⎝ W − a ⎠ ⎥⎦
⎝W − a ⎠
⎢⎣⎝ W − a ⎠
,
where,
P is the load value during the load hold time. Since h1
defined by the function of n was 9.18 for the 1Cr-0.5Mo steel, the h1 function identified from the J-Handbook could be
rearranged into Eq. (8). Here Eq. (8) is valid under this condition: 0.375≤(a/W)≤1.
h1 (
a
a
a
a
a
a
,9.18) = 12.86 − 99.65( ) + 311.04( ) 2 − 472.96( ) 3 + 353.31( ) 4 − 103.62( ) 5
W
W
W
W
W
W
(8)
Discussion
(da/dt)avg data measured from three load hold times 30, 300, and 3600 s could be clearly characterized with (Ct)avg values, as
shown in Fig. 4(a), (b) and (c), respectively. However, the values show slight differences among them in Fig. 5; a slope of
(da/dt)avg versus (Ct)avg curve obtained from the load hold time 30 s condition was 0.382, about one third of the result 1.103
from the load hold time 300 s condition. While the load hold time 3600 s condition yielded (Ct)avg value 0.791. Comparing with
the slope of (da/dt)avg versus (Ct)avg curve from the load hold time 30 s condition, two slopes from longer load hold times
showed significant difference. This means that mechanism change can be occur in the load hold time-dependent crack growth
between 30 and 300 s. In the case of the load hold time 30 s condition, creep-fatigue crack propagated straightly regardless of
location of the fatigue precrack. These results are similar with the results from the triangular waveshape loading. Fig. 6 (a)
shows the microstructure of C1 specimen side part, undergone the triangular waveshape test. It can be noticed that the
precrack front located at the fusion line initially passed the fusion line and propagated straightly into the HAZ. The load hold
time 30 s also resulted in similar straight crack shape because the time-dependent crack growth amount during short load
holding is negligible compared to the crack growth amount due to the unload/reload periods. In the cases of the long load hold
times 300 and 3600 s, creep-induced pores were observed near the crack fronts because the load hold time-dependent crack
growth occurred due to creep deformation and damage. Fig. 7 shows resulting crack front morphologies due to three load
holding conditions; while (a) no pore is identified in C1 specimen undertaken 30 s holding time, creep-induced pores are
observed in C5 and C11 specimens holded during 300 and 3600 s, respectively. This difference in the time-dependent crack
growth mechanism results in the slope difference in Fig. 5.
When the load hold time was 300 s, the crack grown due to creep damage had its path from the HAZ to FGHAZ. The crack
front attained at the FGHAZ grew along the FGHAZ. It means that this welded specimen has the FGHAZ as its weakest region.
However, three repeated testing results from the load hold time 300 s condition showed the crack propagations along the HAZ
and the crack fronts attained the FGHAZ nearly at the ends of testing. Thus, entire testing results display the CFCG rate in the
HAZ. When the load hold time was 3600 s, the crack path to FGHAZ was formed early stage and the crack growth in the
FGHAZ covered nearly whole specimen. Thus, entire testing results display the CFCG rate in the FGHAZ.
Since the FGHAZ has the lowest crack growth resistance at high temperature region, the (da/dt)avg versus (Ct)avg curve locates
at top of Fig. 5 and high crack growth rate is obtained even at same (Ct)avg value. Conclusively, in order to estimate lifetime of
a 1Cr-0.5Mo steel component implying cracks near the welding heat-affected zone and also undergoing high temperature
crack growth damage, the growth rate must be predicted using Eq (9), explaining the FGHAZ crack propagation or upper
bound cracking behavior.
(
⎛ da ⎞
−3
⎜ ⎟ = 6.132 ×10 (C t )avg
dt
⎝ ⎠ avg
)
0.791
(9)
If the specimen is loaded and maintained during 300 s and its precrack front locates in the FGHAZ, the resulting CFCG rate
may be expected to be higher than that in the HAZ center. In order to experimentally confirm this concept, additional
specimens having fatigue precracks at the FGHAZ (or exact boundary of the parent metal and HAZ) were machined and used
for measuring the crack growth rates under the condition of load hold time 300 s. As shown in Fig. 8, CFCG occurred from the
precrack front and then the crack propagated along the FGHAZ. The crack growth rate of this experiment was nearly same
with that of the testing condition of load hold time 3600 s. It means the mechanisms discussed above are correct.
Conclusions
CFCG behavior of 1Cr-0.5Mo steel welds including fatigue precracks at their HAZ was experimentally investigated in this study.
The major conclusions are:
(1) When a crack growth occurs due to mechanical fatigue loading, resulting crack path has no dependency on the metallic
microstructure.
(2) When a crack growth occurs due to creep damage, resulting crack path is determined by the FGHAZ regardless of the
location of the precrack front.
(3) The FGHAZ showed the highest CFCG rate than any other part in the HAZ.
(4) Thus, the lifetime estimation for welds, which is undergoing creep-fatigue damage, must be accomplished based on the
crack growth rate equation of the FGHAZ.
References
1 K.B. Yoon, K. W. Kim, U. B. Baek, 1998, "Modeling of Creep Crack Growth Behavior of Low-Alloy Steel Weld" Journal of
KSME A, Vol. 22, No 12, pp. 2153～2162.
(2) K. B. Yoon, H. W. Shim, Y. E. Shin, 1998, " A Computational Study on Ct Parameter for Weld Interface Crack Under High
Temperature Creep Condition" Journal of KSME A, Vol. 22, No 12, pp. 2239～2246.
(3) Saxena, A., Cretegny, L., Grover, P.S. and Norris, R.H., 1998, "Modeling of Fracture and Crack Growth in Weld Operating
at High Temperature," Proceedings of Int. HIDA Conference, Paris, France 15～17 April.
(4) U. B. Baek, H. M. Lee, 2000, Repot, 99 Development of Structural Integrity Measurement Technology Ch. 8, KRISS/IR
2000～014.
(5) Saxena, A., 1980, "Electrical Potential Technique for Monitoring Sub-critical Crack Growth at Elevated Temperature,"
Engineering Fracture Mechanics, No.13, pp. 741～750.
(6) Johnson. H. H., 1965, "Calibrating the Electrical Potential Method for Studying Slow Crack Growth," Material Research
and Standard, Vol.5, No.9, pp. 442～445.
(7) Saxena, A., and Han. J, 1986, "Evaluation of Crack Tip Parameters for Characterizing Crack Growth Behavior in Creeping
Material," ASTM, Task Group Report, Joint Task Group E24.08.07/E24.04.08.
(8) Shih C. F., 1983, Tables of Hutchinson-Rice -Rosengren Singular Field Quantities, Brown University.