You are on page 1of 8

Fatigue & Fracture of

Engineering Materials & Structures


doi: 10.1111/j.1460-2695.2010.01521.x

Fatigue crack propagation in API 5L X-70 pipeline steel longitudinal


welded joints under constant and variable amplitudes
M . A . N E V E S B E L T R Ã O 1,2 , E . M . C A S T R O D E Z A 1 a n d F . L . B A S T I A N 1
1 Laboratory of Fracture Mechanics, Department of Metallurgical and Materials Engineering, COPPE, Federal University of Rio de Janeiro,
21941-972, Rio de Janeiro, Brazil, 2 GKSS Forschungszentrum Geesthacht GmbH, Institute of Materials Research, Materials Mechanics, Solid-State
Joining Processes (WMP), Max-Planck-Strasse 1, 21502, Geesthacht, Germany

Received in final form 24 August 2010

A B S T R A C T Fatigue crack propagation (FCP) under constant and variable amplitude loading in base
metal (BM), weld metal (WM) and heat affected zone (HAZ) of longitudinal welded
joints of an API X-70 pipeline steel was investigated. Constant amplitude loading tests
were performed at R = 0.1 and 0.5, whereas for variable amplitude testing single peak
tensile overloads (OLs) alternating between 75 and 100% of maximum load were applied
at 2.5 mm intervals in crack growth. Results of SE(B) specimens tested under constant
and variable amplitude loading revealed that BM, WM and HAZ regions subjected to R =
0.5 and low K-values presented the highest crack growth rates. At higher K values
FCP rates in all the studied regions were similar and the R effect on FCP rate was no
more observed. Crack growth retardation due to OLs was observed at the three studied
regions, showing a decrease on the FCP delay with a decreasing on K.

Keywords alternate tensile overloads; API 5L X-70 pipeline steel; fatigue crack propa-
gation; longitudinal welded joints.

NOMENCLATURE a = crack length (mm)


BM = base metal
C = constant of Paris equation
da/dN = fatigue crack propagation rate (mm/cycle)
FCP = fatigue crack propagation
HAZ = heat-affected zone
1
K max = maximum stress intensity factor (MPa.m 2 )
1
K min = minimum stress intensity factor (MPa.m 2 )
1
K OL = stress intensity factor at overload (MPa.m 2 )
1
K op = crack opening stress intensity factor (MPa.m 2 )
n = constant of Paris equation
N = number of cycles
OL = single peak tensile overload
R = stress ratio
r OL = monotonic plastic zone size (mm)
RS = reversed shear
SAW = submerged arc welding
SE(B) = three-point bending specimen
WM = weld metal
WM-L = tensile test specimen (longitudinal direction of weld)

Correspondence: M. A. Neves Beltrão. E-mail: marcelo.beltrao@gkss.de


c 2010 Blackwell Publishing Ltd. Fatigue Fract Engng Mater Struct 34, 321–328 321
322 M . A . N E V E S B E L T R Ã O et al.

WM-T = tensile test specimen (transversal direction of weld)


α = 1/3π or 1/π for plane strain or plane stress state, respectively
1
K = stress intensity factor range (MPa.m 2 )
σ ys = tensile yield strength (MPa)
σ u = ultimate tensile strength (MPa)

INTRODUCTION reduced or increased. On the other hand, Tür and Var-


dar17 and Singh et al.18 reported that overloads applied
Pipelines for oil and gas transportation are sometimes too close lead to crack acceleration because crack jumps
exposed to cyclic loads that could cause their fracture. at each overload exceed the retardation in the subsequent
Pipeline welded joints are regions where the initiation and few baseline cycles. According to Celik et al.,19 the severity
propagation of fatigue cracks has been observed, mainly of retardation is a function of the overload peak proxim-
at stress concentration points like inclusions and weld ity: overloads applied closely and separately sufficiently
defects. are responsible for crack growth acceleration and retar-
A large number of parameters affect fatigue life of a dation, respectively. Several models have been proposed
structural component. An important one is the stress ratio to explain the FCP delay due the OLs, including models
(R) defined by Eq. (1) based on residual stresses,5,20 strain hardening,5 plasticity
K min induced crack closure20–21 and crack tip blunting.5,21
R= , (1) The aim of this paper is to study the effect of the stress
K max
ratio (R) under constant and variable amplitude loadings
where K min and K max are the minimum and maximum on base metal (BM), weld metal (WM) and heat affected
stress intensity factor, respectively. zone (HAZ) of longitudinal welded joints of API 5L X-70
The influence of the stress ratio on fatigue crack propa- pipeline steel. The fatigue behaviour of the welded joint
gation (FCP) is shown by the increase of crack propaga- regions subjected to a single peak tensile OL alternated
tion rate when R-value is increased. However, this effect is by two different levels was also evaluated for variable am-
more pronounced at regions I and III of the Paris curve. plitude loading.
Concepts like residual compressive stress and environ-
mental effects have been developed to explain the stress EXPERIMENTAL PROCEDURE
ratio effect on the FCP rate.1–3 However, after Elber’s
discovery, a great number of researches on stress ratio Materials
effect have been based on the concept of crack closure.4 The BM, WM and HAZ regions of longitudinal welded
The types of load to which structural materials like joints of an API 5L X-70 pipeline steel were studied.
pipelines are exposed during operation vary with time. In The welded joints were obtained by SAW process. The
this sense, it is important to consider variable amplitude main characteristics of the SAW process that was used are
loadings when studying fatigue behaviour of structural shown in Table 1. The chemical composition of the BM
materials because load variations can accelerate or delay is presented in Table 2.
the FCP rate.5–10 Consequently, an accurate prediction
of fatigue life in these conditions is necessary. One of the
simplest forms of studying material behaviour under load Table 1 Main characteristics of the longitudinal welding
variations is through tensile overloads (OLs) application
within constant amplitude fatigue loading. Electrode Welding conditions
It has been well documented in literature that an OL
Tack welding: MAG CO2 Heat input (kJ/cm) = 1.39–4.45
application during the crack propagation stage influences
Manufacturer: Belgo Bekaert
the subsequent crack growth behaviour because a crack Diameter (mm) = 3.2
propagation rate delay until a minimum da/dN value is Inside welding: SAW Number of wires: 3
observed. Several works on literature5,10–15 reported that Manufacturer: Bavaria Heat input (kJ/cm) = 33.0
the highest crack growth delays occurred when the highest Schweisstechknik
OL levels were applied. Diameter (mm) = 4.0
When periodic overloads are applied within constant Outside welding: SAW Number of wires: 4
Manufacturer: Bavaria Heat input (kJ/cm) = 31.0
amplitude loading intervals, OL interactions may occur.
Schweisstechknik
Yuen and Taheri16 mention that if multiple tensile OLs Diameter (mm) = 4.0
are applied too closely or frequently, the FCP delay can be


c 2010 Blackwell Publishing Ltd. Fatigue Fract Engng Mater Struct 34, 321–328
FATIGUE CRACK PROPAGATION IN WELDED JOINTS 323

Table 2 Chemical composition of API 5L X-70 BM (% of weight)

C S Al Si P Ti V Cr Mn Ni Cu Nb Mo N Ceq
0.08 0.002 0.028 0.25 0.017 0.027 0.045 0.02 1.73 0.194 0.016 0.064 0.003 0.0067 0.40

Table 3 Mechanical properties of longitudinal welded joints of


API X-70 steel

Region σ ys(0,2%) (MPa) σ u (MPa) Elongation (mm)

BM 492 583 7.3


WM-L 635 694 7.8
WM-T 557 627 6.3

Fig. 2 Orientation of the fatigue test specimens with respect to the


welded joint (out of scale).

mately) under load control using a sinusoidal waveform,


frequency of 30 Hz and two stress ratio values (R = 0.1
and 0.5).
Constant and variable amplitude loads were used. Two
different levels of single peak tensile OLs alternating be-
tween 75 and 100% of maximum load were applied ev-
ery 2.5 mm of crack growth, always at the same a value,
Fig. 1 Dimensions of SE(B) specimens for fatigue tests
(dimensions in mm).
in the variable amplitude tests. The interval of 2.5 mm
was used to avoid any interaction effect between OLs.
The last OL of 100% was not applied on the test speci-
The mechanical properties of BM and WM regions are mens subjected to R = 0.5, in order to avoid the specimen
presented in Table 3. Tensile tests in WM region were fracture.
performed in longitudinal and transversal directions of Reference parallel marks equally spaced of 0.5 mm were
the weld, WM-L and WM-T, respectively. introduced on the surface of the specimens to allow visual
crack growth measurements. The function described by
ASTM E399 standard24 for SE(B) specimens was used for
Fatigue tests
K-value calculations.
Three point bending SE(B) test specimens (X-groove de-
sign) of dimensions and geometry established at BS 6835
standard22 were used for the FCP tests, Fig. 1. These spec- RESULTS AND DISCUSSION
imens were chemically etched with Nital 2% solution for
Constant amplitude loading
weld revealing and notch positioning. In the case of HAZ,
the notch was positioned in accordance to weld positional Figure 3 shows the a versus N curves for the BM, WM and
criteria established at BS 7448 part 2 standard23 : 50% of HAZ regions of the welded joints subjected to constant
the notch was positioned on WM and 50% on HAZ. Fig- amplitude loading for both R-values, 0.1 and 0.5.
ure 2 presents a scheme showing the orientation of the The obtained results revealed clearly the effect of R on
fatigue test specimens with respect to the welded joint. the relation between a and N . All welded joint regions sub-
Fatigue propagation tests of BM, WM and HAZ regions jected to R = 0.5 presented an accelerated crack growth
were performed using a MTS model 810 servohydraulic for a small number of cycles in comparison with R = 0.1.
testing machine (100 kN load capacity) in accordance For R = 0.1, this accelerated growth only occurred after a
with BS 6835 standard recommendations.22 All specimens large number of cycles. This behaviour is reflected on the
were tested at ambient air temperature (24 ◦ C, approxi- da/dN versus K curves for BM, WM and HAZ regions,


c 2010 Blackwell Publishing Ltd. Fatigue Fract Engng Mater Struct 34, 321–328
324 M . A . N E V E S B E L T R Ã O et al.

Table 4 Values of K and da/dN of each welded joint region


where FCP rates tend to be similar for both R-values studied

Region K (MPa.m1/2 ) da/dN (mm/cycle)

BM ∼
= 34 ∼
=3.3 × 10−4
WM ∼
=47 ∼
=3.8 × 10−4
HAZ ∼
=36 ∼
=2.6 × 10−4

K op , the crack is fully opened. A small R-value represents


a small K min value for the same K and consequently a
larger K-value for the crack to be fully opened. On the
other hand, Lal25–27 presents a model based on mecha-
nistic approach to examine the influence of positive stress
ratio values (0 ≤ R ≤ 1) on FCP of ductile materials
Fig. 3 a versus N curves of BM, WM and HAZ regions subjected
to R = 0.1 and 0.5 under constant amplitude loading. independent of crack closure arguments. This approach
considers the presence of two micromechanisms during
crack growth: submicroscopic cleavage at the beginning
and RS (its contribution will increase gradually). Accord-
ing this approach, both mechanisms may be present dur-
ing crack propagation for R < 0.6. However, as the crack
growth rate is no longer affected by an increase in R-
value, RS is the dominant mechanism beyond this point.
In this sense, the similarity between crack growth rates ob-
served here in each region may indicate a RS-dominant
crack growth mechanism close to K-values presented in
Table 4.
The individual analysis of the welded joint regions
through their da/dN versus K curves in Fig. 4, shows
that the BM presents the highest crack growth rate at all
K ranges studied for R = 0.5. On the other hand, for
R = 0.1, this behaviour is observed for K values above
25 MPa.m1/2 , approximately. The experimental Paris law
coefficients of region II obtained in this work for the
Fig. 4 da/dN versus K curves of BM, WM and HAZ regions for three welded joint regions studied are shown in Table 5.
R = 0.1 and 0.5 under constant amplitude loading.
Other coefficients obtained by Shi et al.,28 Fernandes29
and Vosikovski and Rivard30 for X-52, X-60 and X-65
steels, respectively are also presented in Table 5 for com-
presented in Fig. 4, where all curves for both R-values are
parison.
presented.
Table 5 shows that the Paris equation coefficients ob-
Figure 4 shows that the welded joints tested at R =
tained in this work for all welded joint regions of X-70
0.5 presented, initially, the highest crack growth rates.
steel for R = 0.5 presented analogous values. On the other
It can be also seen that the FCP rates of each welded joint
hand, for R = 0.1, the different coefficient values obtained
region present similar behaviour at higher K-values in-
indicate that the crack growth rate in region II of prop-
dicating that the FCP rates are in this case almost indepen-
agation can be influenced by the microstructure of the
dent from the R-values. Table 4 presents the approximate
welded joint regions. In this sense, the results show that
da/dN and K values where the similarity between crack
the FCP in region II is not affected by the microstructure
growth rates was observed.
of the welded joints for R = 0.5.
Although no crack closure measurements were carried
out, the similarity between da/dN curves might be ac-
counted for the decrease of this phenomenon at high
Variable amplitude loading
K-values. This idea is reinforced by the fact that this
phenomenon is more pronounced for small values of K, All a versus N curves of the three welded joint regions
where K min is below or close to K op value and that above tested at variable amplitude loading for R = 0.1 and 0.5


c 2010 Blackwell Publishing Ltd. Fatigue Fract Engng Mater Struct 34, 321–328
FATIGUE CRACK PROPAGATION IN WELDED JOINTS 325

Table 5 Coefficients of Paris law for the three regions of the welded joints of X-70 steel obtained in this work and from the literature

Welded joint region

BM WM HAZ

Steel R Author C n C N C n

X-70 0.1 This work 2.06 × 10−11 4.59 1.60 × 10−9 3.22 4.03 × 10−10 3.66
0.5 3.73 × 10−9 3.28 3.07 × 10−9 3.16 3.18 × 10−9 3.20
X-52 0.1 Ref. [28] 6.11 × 10−14 4.42 2.46 × 10−14 4.19 – –
X-60 0.1 Ref. [29] 5.13 × 10−10 3.61 2.60 × 10−11 4.91 – –
0.7 4.35 × 10−9 4.74 4.74 × 10−10 4.09 – –
X-65 0.05 Ref. [30] 1.33 × 10−9 3.53 – – – –

The results of the a versus N curves presented in Fig. 5


for variable amplitude loading with OLs reveal the same
behaviour than for constant amplitude loading: BM, WM
and HAZ regions subjected to R = 0.5 show a more accel-
erated crack growth than R = 0.1, inducing a shorter fa-
tigue life. This behaviour is reflected on the da/dN versus
K curves (Figs 6–8): initially, like at constant amplitude
tests, a greater crack propagation rate results for R = 0.5.
On the other hand, similar FCP rates were observed for
high K-values. BM region exhibited the highest crack
propagation rates for both R-values, this effect being more
evident for R = 0.1, while WM and HAZ regions showed
similar behaviour, as can be seen in Fig. 9.
Figures 6–8 show a small drop of crack propagation
rate after OL application for all fatigue tests carried
out, characterizing a crack delay, as expected. How-
Fig. 5 a versus N curves for welded joints under variable amplitude
ever, when K values increase, it is perceptible that the
loading for R = 0.1 and 0.5.
drop in crack propagation rate promoted by the OLs
are presented in Fig. 5. The respective da/dN versus K decreases until a point where crack delay is no more
curves are presented in Figs 6–8. All these curves are also observed.
presented in Fig. 9 for comparison. The points of OL The size of the monotonic plastic zone promoted by the
application are indicated by arrows. OL can be estimated from10

Fig. 6 da/dN versus K curves of BM region for R = 0.1 and 0.5 under variable amplitude loading.


c 2010 Blackwell Publishing Ltd. Fatigue Fract Engng Mater Struct 34, 321–328
326 M . A . N E V E S B E L T R Ã O et al.

Fig. 7 da/dN versus K curves of WM region for R = 0.1 and 0.5 under variable amplitude loading.

Fig. 8 da/dN versus K curves of HAZ region for R = 0.1 and 0.5 under variable amplitude loading.

 2 values, crack growth due to OL application is more sig-


K OL
rOL = α , (2) nificant than the effect of FCP delay expected with K.
σys
In this work, the presence of secondary microcracks was
where K OL is the maximum stress intensity factor at the observed in all welded joint regions at high values of K
OL, σ ys is the yield strength of the material and α is 1/3π (Fig. 10a–c).
or 1/π for plane strain or plane stress state, respectively. As the final OLs were applied at high values of K, K OL
This plastic zone is responsible for a compressive resid- must be close to the fracture toughness of the material and
ual stress field formation. As the fatigue crack advances the amount of secondary microcracks will increase. Thus,
through this zone, a higher crack opening load is required the contribution to crack propagation of the increase of
for crack propagation promoting the observed FCP delay. microcracks is more effective than the expected delay.
From Eq. (2), higher K-values were expected to cre-
ate larger plastic zone sizes and consequently an increase
CONCLUSIONS
on the FCP delay effect. However, this behaviour was
not observed. Borrego et al.7 observed a similar trend in In this work, the stress ratio (R) effect on FCP of longi-
6082-T6 aluminium alloy. According them, at high K- tudinal welded joint regions of an API X-70 steel under


c 2010 Blackwell Publishing Ltd. Fatigue Fract Engng Mater Struct 34, 321–328
FATIGUE CRACK PROPAGATION IN WELDED JOINTS 327

constant amplitude loading and variable amplitude load-


ing under a single cycle peak tensile OL application was
investigated. The results obtained lead to the following
conclusions.
The longitudinal welded joint regions subjected to
R = 0.5 initially presented the highest FCP rates in
the constant amplitude tests. At higher K values all
welded joint regions present similar behaviour indicat-
ing that the FCP rates get more independent from
the R-values. The similarity between crack growth rates
at high K-values in each region may indicate a RS-
dominant crack growth mechanism or a decreasing of
crack closure phenomena. The discrepancy between n-
values from Paris constants for R = 0.1 may indicate an in-
fluence of the microstructure on FCP of the welded joints
investigated.
Single peak tensile OL application on variable ampli-
Fig. 9 da/dN versus K curves of BM, WM and HAZ regions for
R = 0.1 and 0.5 under variable amplitude loading.
tude tests promoted delays on FCP rates in all welded
joint regions for both R-values, as expected. However,
the propagation delay was not observed at elevated K

Fig. 10 SEM images of (a) BM, (b) WM and (c) HAZ regions at high K -values. The presence of secondary microcracks can be observed in
all regions.


c 2010 Blackwell Publishing Ltd. Fatigue Fract Engng Mater Struct 34, 321–328
328 M . A . N E V E S B E L T R Ã O et al.

values. The higher amount of secondary microcracks at growth under variable amplitude loading. Int. J. Fatigue 14,
elevated K values promoted by OL application can ex- 377–386.
plain this behaviour. 13 Verma, B. B. and Pandey, R. K. (1999) The effects of loading
variables on overload induced fatigue crack growth retardation
parameters. J. Mater. Sci. 34, 4867–4871.
Acknowledgements 14 Kumar, R., Kumar, A. and Kumar, S. (1994) Delay effects in
The authors would like to thank TenarisConfab (special fatigue crack propagation. Int. J. Pres. Ves. Piping 67, 1–5.
15 Sander, M. and Richard, H. A. (2006) Fatigue crack growth
thanks to Eng. S. Kojima) for the welded joint mate-
under variable amplitude loading. Part I: experimental
rials, the Materials Group of CTEx/Brazilian Army for investigations. Fatigue Fract. Engng. Mater. Struct. 29,
the laboratorial facilities, CNPq (CT-Petro Project num- 291–301.
ber 502084/03-0) and CAPES for the financial support, 16 Yuen, B. K. C. and Taheri, F. (2006) Proposed modifications
CAPES and CNPq for the scholarship of M.A.N. Beltrão to the Wheeler retardation model for multiple overloading
and CAPES for the fellowship of E.M. Castrodeza. fatigue life prediction. Int. J. Fatigue 28, 1803–1819.
17 Tür, Y. K. and Vardar, Ö. (1996) Periodic tensile overloads in
2024-T3 Al-alloy. Eng. Fract. Mech. 53, 69–77.
REFERENCES 18 Singh, K. D., Khor, K. H. and Sinclair, I. (2008) Finite
element and analytical modeling of crack closure due to
1 Kujawski, D. and Ellyin, F. (1987) A fatigue crack growth repeated overloads. Acta Mater. 56, 835–851.
model with load ratio effects. Eng. Fract. Mech. 28, 367– 19 Celik, C. E., Vardar, Ö. and Kalenderoglu, V. (2004)
378. Comparison of retardation behaviour of 2024-T3 and
2 Bulloch, J. H. (1994) Fatigue threshold in steels – mean stress 7075-T6 alloys. Fatigue Fract. Engng. Mater. Struct. 27,
and microstructure influences. Int. J. Ves. Piping 58, 103– 713–722.
127. 20 Mc Evily, A. J. and Ishihara, S. (2002) On the development of
3 Zhang, J., He, X. D. and Du, S. Y. (2003) A simple crack closure at high R levels an overload. Fatigue Fract.
engineering approach in the prediction of the effect of stress Engng. Mater. Struct. 25, 993–998.
ratio on fatigue threshold. Int. J. Fatigue 25, 935–938. 21 Mohanty, J. R., Verma, B. B. and Ray, P. K. (2009) Prediction
4 Kujawski, D. (2001) A fatigue crack driving force parameter of fatigue crack growth and residual life using an exponential
with load ratio effects. Int. J. Fatigue 23, S239–S246. model. Part II: mode-I overload induced retardation. Int. J.
5 Borrego, L. P., Ferreira, J. M., da Cruz, J. M. P. and Costa, J. Fatigue 31, 425–432.
M. (2003) Evaluation of overload effects on fatigue crack 22 BS 6835 Standard (1988) Determination of the Rate of
growth and closure. Eng. Fract. Mech. 70, 1379–1397. Fatigue Crack Growth in Metallic Materials.
6 Meggiolaro, M. A. and de Castro, J. T. P. (2003) On the 23 BS 7448 Part II Standard (1997) Fracture mechanics
dominant role of crack closure on fatigue crack growth toughness tests. Method for determination of K IC , critical
modeling. Int. J. Fatigue 25, 843–854. CTOD and critical J values of welds in metallic materials.
7 Borrego, L. P., Costa, J. M. and Ferreira, J. M. (2005) Fatigue 24 ASTM E399 Standard (1999) Standard Test Method for
crack growth in thin aluminium alloy sheets under loading Plane-Strain Fracture Toughness of Metallic Materials.
sequences with periodic overloads. Thin-Walled Struct. 43, 25 Lal, D. N. (1994) A mechanistic model for the influence of
772–788. stress ratio on the LEFM fatigue crack growth behavior of
8 Huang, H. L. and Ho, N. J. (2001) The observation and metals and alloys – I. crack ductile materials. Eng. Fract. Mech.
analysis of the dislocation morphology of fatigue crack tips at 49, 871–897.
steady state propagation rates subject to a single peak load. 26 Lal, D. N. (1996) A detailed physical analysis of the R-effect
Mater. Sci. Eng. A298, 251–261. on LEFM fatigue crack growth – I. On the combined roles of
9 Hammouda, M. M. I., Osman, H. G. and Sallam, H. E. M. critical zones, FEFM parameters and stress ratio. Eng. Fract.
(2004) Mode I notch fatigue crack growth behaviour under Mech. 55, 115–132.
constant amplitude loading and due to the application of a 27 Lal, D. N. (1996) A detailed physical analysis of the R-effect
single tensile overload. Int. J. Fatigue 26, 183–192. on LEFM fatigue crack growth – II. On the combined roles of
10 Rushton, P. A. and Taheri, F. (2003) Prediction of crack growth mechanisms and stress ratio. Eng. Fract. Mech. 55,
growth in 350WT steel subjected to constant amplitude with 289–312.
over-and under-loads using a modified Wheeler approach. 28 Shi, Y. N., Chen, B. Y. and Zhang, J. X. (1999) Effects of
Mar. Struct. 16, 517–539. welding residual stresses on fatigue crack growth behaviour in
11 Ramos, M. S., Pereira, M. V., Darwish, F. A., Motta, S. H. butt welds of a pipeline steel. Eng. Fract. Mech. 36, 893–902.
and Carneiro, M. A. (2003) Effect of single and multiple 29 Fernandes, J. L. (2002) Uma Metodologia para a Análise e
overloading on the residual fatigue life of a structural steel. Modelagem de Tensões Residuais. PhD Thesis, PUC, Rio de
Fatigue Fract. Engng. Mater. Fatigue Fract. Engng. Mater. Janeiro, RJ, Brazil (in Portuguese).
Struct. 26, 115–121. 30 Vosikovski, O. and Rivard, A. (1981) Growth of surface
12 Geary, W. (1992) A review of some aspects of fatigue crack fatigue cracks in a steel plate. Int. J. Fatigue 3, 111–115.


c 2010 Blackwell Publishing Ltd. Fatigue Fract Engng Mater Struct 34, 321–328

You might also like