Ocean Engineering 113 (2016) 246–254

Contents lists available at ScienceDirect

Ocean Engineering
journal homepage: www.elsevier.com/locate/oceaneng

Prediction of fatigue crack growth in a ship detail under

wave-induced loading
Xiaoshun Yan a,b, Xiaoping Huang a,n, Yingcai Huang c, Weicheng Cui a
a
Collaborative Innovation Center for Advanced Ship and Deep-sea Exploration, Shanghai Jiao Tong University, Shanghai 200030, China
b
China Ship Development and Design Center, Wuhan 430064, China
c
School of Transportation, Wuhan university of Technology, Wuhan 430070, China

art ic l e i nf o a b s t r a c t

Article history: Fatigue life prediction based on fracture mechanics has become the focus of research on the strength of
Received 15 March 2014 ship structures; however, it is difficult to summarize a general formula for calculating stress intensity
Accepted 30 October 2015 factors (SIF) for cracks in ship details, and the application of fatigue crack propagation (FCP) theory is
Available online 20 January 2016
limited to simple structures and simple loads in some way. To address this problem, in the present study,
Keywords: the SIF of a crack in a ship detail was calculated by combining a PATRAN finite element model for the
Stress intensity factor whole ship with the advantages of ANSYS for SIF calculation, incorporating a macrocode written to
Fatigue crack propagation achieve the transformation. The method was validated by comparison with existing empirical formulas.
Unique curve model Also a method of generating the ship fatigue loading spectrum is demonstrated based on the design wave
Design wave approach
approach. Finally, an example is given of combining the unique curve model of FCP with the proposed SIF
Hatch corner
calculation method and the fatigue loading spectrum generation method, to predict the fatigue life of a
welded hatch corner joint in a vessel with large hatch openings. This serves as a reference for fatigue
assessment of other complex ship details.
& 2015 Elsevier Ltd. All rights reserved.

1. Introduction fundamental parameter of fracture mechanics, and many attempts

Since a ship is frequently subjected to complex loads during its
service life, fluctuating stresses always exist in structural mem-
bers. These may cause fatigue failure of local as well as global
structures, with catastrophic consequences. Therefore, fatigue
strength assessment is an important criterion in structural ship
design. Among the approaches to assessing fatigue strength,
fracture mechanics-based fatigue evaluation has the capability to
take initial crack size, load sequence, residual stress, etc. into
consideration (Cui, 2003; Fricke, 2003; Sumi and Inoue, 2011).
Much effort has been spent over the past several decades to
develop a fatigue crack propagation theory and put it into practice
(Doerk and Rörup, 2009; Doshi and Vhanmane, 2013; Fricke et al.,
2002; Okawa and Sumi, 2008; Sumi, 2013).
Despite the many studies of fracture mechanics-based fatigue
life prediction over the years (Cui et al., 2011; Huang et al., 2008;
Paris et al., 1961), some problems remain when applied to the
practice of marine structures. An important issue is the calculation
of stress intensity factors (SIFs) when complex loads are applied to
intricate engineering joints. Stress intensity factor (SIF) is a
n
Corresponding author. Tel.: þ 86 21 34207009.
E-mail address: xphuang@sjtu.edu.cn (X. Huang).
have been made to calculate it accurately. Newman and Raju
(1981) calculated the SIFs of semi-elliptical surface cracks of dif-
ferent sizes in plates. Empirical formulas developed by 3-D finite
element analysis were shown to be consistent with experimental
observation. Rhee et al. (1991) developed empirical formulas for
the SIFs of surface cracks in the weld toe of T-shaped tubular joints
subjected to brace tension and in-plane and out-of-plane bending
loads, for a wide range of joint dimensions in existing jackets.
Bowness and Lee (2000) used several empirical formulas to pre-
dict weld toe magnification factors for semi-elliptical cracks in T-
butt joints. Some relevant fatigue design rules such as BS7910
(British Standards Institution, 2005) etc. also recommend empiri-
cal formulas for determining SIFs in the weld toe of different kinds
of joints.
However, at least two problems arise when using any of these
empirical formulas in practical engineering situations: (1) All the
empirical formulas were derived from finite element simulations
or laboratory experiments, the models for which were simple and
were subjected to elementary tensile or bending stresses, and the
stress distribution in all cases was assumed to be uniform and
acting in a direction normal to the cross-section. However, such
circumstances are almost unachievable in actual ocean engineer-
ing structures, and the formulas would obviously be different
for realistic stress distributions and directions. (2) The stresses

http://dx.doi.org/10.1016/j.oceaneng.2015.10.056
0029-8018/& 2015 Elsevier Ltd. All rights reserved.
 X. Yan et al. / Ocean Engineering 113 (2016) 246–254 247

Fig. 1. A flat plate with crack in ship structures.

defined in empirical formulas are generally nominal (far-field)

stresses, but it is difficult to define the nominal stress in real
structures. For example, in Fig. 1 the stress on boundary A of the
plate is not uniformly distributed—so which node stress represents
the nominal stress? How far along y axis can reasonably be said to
be “far-field”? Obviously, the node that is chosen on the boundary
parallel to the crack and its distance from the crack both con-
tribute significantly to the SIF. In addition, authoritative empirical
formulas proposed are all, by far, about relatively uncomplicated
structures and cannot be applied to general marine structures such
as hatch corners in vessels with large hatch openings, aggregate
junctions in the bilge, tubular joints with stiffeners on platform
decks and so on, all of which are particularly vulnerable to fatigue
failure. Some joints are so intricate that it is unlikely that it would
be possible to propose a unified, reasonable empirical formula.
To sum up, the current simplified methods permit only a rough
evaluation, and limit the broad application of a fracture
mechanics-based fatigue strength assessment. In order to apply Fig. 2. MPC for connection of shell and solid connection ((a) before connection;
advanced fatigue crack propagation (FCP) theory to fatigue life (b) after connection).
prediction in general, a universal and accurate method of calcu-
lating SIFs of cracks in complex stress fields is required.
The present study mainly investigates the FCP method of fati- RSCCON Surf-Vol or Rigid meets the requirement; the former was
gue life prediction in complex ship structures subjected to wave- selected in the present work. The nodes on the shell elements
induced stresses. To make use of the pre-modeled whole-ship were selected as the dependent nodes, and the corresponding
finite element model in PATRAN (commercial FEM software widely nodes on the solid elements were selected as independent. In
used in ocean engineering) and take the advantages of ANSYS for Fig. 2(a), nodes A–E are dependent, while nodes A1–E1 and A2–E2
SIF calculation, a macroprogram was written as part of this study are independent. Fig. 2(b) shows the PATRAN model after the
to achieve the transformation from PATRAN to ANSYS. The SIFs for MPCs were created.
differently-sized cracks in the detail of a hatch corner subjected to When the SIF of the crack is calculated, the result is inaccurate
wave-induced loading were then calculated. The fatigue loading if the FEM mesh at the tip of the crack is too coarse. The ANSYS
spectrum for the hatch corner of a ship was generated using the software provides a specialized singular element and KSCON
design wave approach, and its fatigue life was predicted based on command for SIF calculations, and is recognized as an accurate
the unique curve model, which will be detailed later in this paper,
method of calculating SIFs. Conversely, PATRAN does not possess
of fatigue crack propagation.
this capability, and relatively little research using PATRAN to cal-
culate 3-D crack SIFs has been carried out. It was therefore
necessary to derive a suitable method of combining the cap-
2. SIF calculations for structures subjected to complex load
abilities of PATRAN and ANSYS.
In practice, the coordinates of PATRAN and ANSYS do not
2.1. Coupled degrees of freedom and node transformation
always coincide, and transformation and rotation are needed. In
During structural analysis, marine structures are widely mod- the present study, the macro patran2ansys.mac was written to
eled by the commercial FEM software PATRAN; the elements used achieve the following functions: (1) reading the PATRAN node
are mainly 2-D shell and 1-D beam. Therefore, in order to calculate coordinates and displacement data; (2) reading the ANSYS node
the SIFs of 3-D surface cracks, it is necessary to create solid models coordinates; (3) the transformation and rotation of the coordi-
and the connections which link the shell elements and solid nates; (4) displacement interpolation; (5) to add the loading at
elements. each node to the ANSYS boundary. To be clear, the stiffness of solid
The shell elements and solid elements are linked by multi- model in ANSYS differs due to surface crack growth. Since the
point constraints (MPCs). In PATRAN, either of the MPC types influence is small, this paper does not take it into consideration.
248 X. Yan et al. / Ocean Engineering 113 (2016) 246–254
Fig. 3. Schematic diagram of a flat plate with surface crack.
Fig. 4. Flat plate model in PATRAN ((a) Coarse FEM mesh; (b) model with shell and
solid elements).
2.2. Verification of the method
In order to demonstrate in detail and then verify the method
described above, SIF calculations of semi-elliptical surface cracks
in relatively simple structures subjected to uniform stresses are
displayed and the results have been compared with the results of
authoritative empirical formulas.
2.2.1. Flat plate subjected to tension and bending
In an FEM model, a flat plate measuring 5 m  2 m  20 mm
was subjected to a uniformly distributed tensile stress and a
simple bending stress (Fig. 3). The initial coarse mesh is shown in
Fig. 4(a). Then the crack area of the coarse model was refined and
swept to solid elements, and the outcome is shown as Fig. 4(b). A
simulated tensile stress of 110 MPa and a bending stress of 9 MPa
were then applied separately, and static analyses were carried out.
Then, a corresponding solid model with a semi-elliptical sur-
face crack was created in ANSYS. When creating the ANSYS solid
model, SOLID95 element (a kind of solid element with 20 nodes)
and a singular element (a kind of element dedicated to SIF cal-
culation, containing some collapsing nodes) were selected to
ensure accurate results. The mesh generated at the tip of the crack
is shown in Fig. 5. The nodes on the boundary of the ANSYS model
were constrained by displacements interpolated based on the
node coordinates, and displacements obtained by PATRAN stress
analysis, using the macro patran2ansys.mac.
Following static analysis in ANSYS, the SIFs for different crack
sizes were calculated and compared with the results of the
empirical formulas proposed by Newman and Raju (1981) (Fig. 6).
It is seen that the results of the two methods were in close
agreement.
2.2.2. T-butt joint subjected to tensile stress
A PATRAN model of one-half of the T-butt joint is shown as
Fig. 7. The geometric parameters were: L ¼35 mm, T ¼20 mm,
t¼20 mm, b¼200 mm, hatt ¼300 mm, h0 ¼400 mm, and θ ¼45°. A
tensile stress of 110 MPa was applied on the boundary. After
refining and sweeping the coarse model, the new model com-
bining shell elements and solid elements is shown in Figs. 8 and 9
show the corresponding model with a crack in ANSYS. The
patran2ansys.mac macro was then compiled to perform the
necessary transformation, interpolation and loading. The SIF
results for cracks of different sizes from ANSYS static analysis were
compared with the values by the empirical formula of Bowness
and Lee (2000). Fig. 10 shows close agreement between the two
methods. Its worth noting that the effect of stress singularity of
the weld toe is also taken into consideration by the linear
exploration method and the collapse elements provided by ANSYS.
3. FCP law and loading spectrum for ship structures
3.1. Unique curve model of FCP
The crack propagation model is the core of FCP theory and
consequently more research has been done on it. The most
representative formula is the Paris law (Paris et al., 1961), which
has been widely studied and used due to its simplicity, and the
formula constants C and m in Paris law have been frequently
tested for different materials. Since the Paris formula ignores the
effects of stress ratio and load sequence, in many subsequent
studies the original formula was modified and new ones proposed.
However, most of the new formulas added more parameters, or
changed the original parameters of Paris’s law, with the result that
the large body of experimental C and m data cannot be directly
used. To remedy this, Huang et al. (2008, 2009) proposed the
unique curve model, which takes advantage of the existing C and
m database and at the same time takes the stress ratio and loading
sequence into consideration. This model combines simplicity and
practicability, and has potential for future use. The basic expres-
sion of the unique curve model is:
da h
m
m i
¼ C ΔK eq0  ΔK th0 ; ð1Þ
dN
ΔK eq0 ¼ M R MP ΔK; ð2Þ
 X. Yan et al. / Ocean Engineering 113 (2016) 246–254 249

Fig. 5. ANSYS model of flat plate with surface crack and mesh of crack tip.

8
>  β1
< ð1  RÞ
> ð  5 rR o 0Þ
M R ¼ ð1  RÞ  β ð0 r R o 0:5Þ ð3Þ
>
>
: ð1:05 1:4R þ 0:6R2 Þ  β ð0:5 r R o 1Þ
8 n
< ry
a þ r y o aOL þ r OL  r Δ
aOL þ r OL  a  rΔ
MP ¼ ð4Þ
:1 a þ r y Z aOL þ r OL  r Δ

 2 !2 !2
K max K OL ΔK u
ry ¼ α ; r OL ¼ α max
; rΔ ¼ α ð5Þ
σy σy σ y

0:29
α ¼ 0:35  h  i2:15 ð6Þ
1 þ 1:08K 2max = t σ 2y

ΔK u ¼ K i1 i
min  K min ð7Þ

where a is the crack length; N is the number of applied cycles; C, m

are the Paris parameters; ΔKeq0 is the equivalent SIF range cor-
responding to stress ratio R¼0; ΔKth0 is the threshold of SIF range
corresponding to stress ratio R¼0;MR, MP are the correction fac-
tors for the effects of stress ratio and load sequence, respectively.
The values and relevant physical basis of MR and MP can refer to
the literature Huang et al. (2008, 2009). If MR ¼MP ¼ 1, the influ-
ences of stress ratio and load sequence are not considered and the
unique curve model reverts to Paris law, which reflects the uni-
versality of the unique curve model.

3.2. Fatigue loading spectrum generation based on design wave

approach

3.2.1. Fatigue load spectrum with sequence

Fig. 6. Comparison between proposed method and empirical formula ((a) tension;
(b) bending).
Constructing a fatigue load spectrum that closely resembles the
actual loading on the engineering structure is essential for accu-
rate prediction of its fatigue life. The physical phenomena implied
in fatigue are not chaotic but random; therefore the fluctuations
may be characterized by mathematical functions and by a limited
number of parameters (Hobbacher, 2009).
Wave-induced fluctuating stresses are stochastic in the short
term but are regularly distributed in the long term. In general, the
250 X. Yan et al. / Ocean Engineering 113 (2016) 246–254

Fig. 7. Schematic diagram of T-butt joint with semi-ellipse surface crack.

Fig. 8. Hybrid model of a T-butt joint, with shell and solid elements.

range of wave-induced stresses is considered to be consistent with where Δσi is the stress range of the i-th level; ni is the corre-
a two-parameter Weibull distribution whose probability density sponding number of loading cycles; and m is the exponent of the
function is: crack propagation law. If a two-parameter Weibull distribution is
  "   # assumed for the random stress ranges, Eq. (9) can be written as

 h Δσ h  1 Δσ h

f Δσ ¼ exp  0 r Δσ o þ 1 ; ð8Þ (Doshi and Vhanmane, 2013):
q q q
  1=m
where Δσ is the wave-induced stress range; h is known as the m
Δσ eq ¼ qm U Γ 1 þ ; ð10Þ
shape parameter; and q is known as the scale parameter. When h q
and q are determined, the wave-induced stress ranges for a certain
period is equivalent to a series of pseudo-random numbers gen- where Γ(.) is the gamma function.
erated by the program. For random loading, the equivalent stress However, when the random stress range is transformed into an
range (Sumi, 2013) is commonly used to estimate the fatigue life equivalent constant stress range, the effects of the load sequence
and defined by: are ignored, as are the cycle counting methods, such as zero-
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
X X crossing counting, peak counting, range-pair counting, rain flow
Δσ eq ¼ m ðΔσ m i Uni Þ= ni ; ð9Þ
counting and so on. Therefore, in order to take the loading
 X. Yan et al. / Ocean Engineering 113 (2016) 246–254
Fig. 9. Symmetrical FEM model in ANSYS.
Fig. 10. Comparison between proposed method and empirical formula ((a) deepest
point; (b) surface point).
251
sequence into consideration, the generated pseudo-random
numbers or the random stress ranges are not treated.
3.2.2. Determination of h and q
Most classification societies (GL, DNV, ABS, CCS etc.) recom-
mend conservative values, empirical formulas or spectral analysis
methods to calculate h, the value of which has been shown to have
significant effect on fatigue life prediction; and q is obtained from:
Δσ n0
q¼ 1=h
; ð11Þ
ðln n0 Þ
where Δσ n0 is the fatigue stress range at a probability level of
exceedance equal to 1/n0. The stress range for a given probability
of exceedance may be calculated by the design wave approach,
whose definition states that the stress range is double the stress
response at the design wave load.
In practice, some obstacles remain as to the selection of stress
response. As we know, three main approaches are available in the
definition of local stress during fatigue assessment based on
stress-life (Fricke, 2003). The stress that is relevant for calculating
the SIF is the nominal stress, that is, the stress sufficiently distant
from the crack; but it is not clear what this distance is, as discussed
above. When empirical formulas are used, the nominal stress that
is thought to act on a selected node is largely subjective, and may
vary from individual to individual, with the result that the corre-
sponding predicted SIF and fatigue life also varies considerably.
It is difficult to define the nominal stress when empirical formulas
for SIF calculation are applied. The present SIF calculation method
has no such problem. This is because, in linear systems theory, the
relationship between the wave loads and the induced stress is linear.
As we know, another linear relationship exists between nominal
stress and SIF. That is to say, there is a linear relationship between the
three, i.e. wave loads, nominal stress and SIF.
The stress distribution and SIFs under the design wave loading
can be obtained using the design wave approach and the proposed
SIF calculating method. Then the design wave loads, nominal
stress and SIF correspond to each other. Any node in the vicinity of
the fatigue hot spot may be selected as the reference point. The
stress response at this reference point is then inferred from the
252 X. Yan et al. / Ocean Engineering 113 (2016) 246–254

result of the stress field obtained above. The scale parameter q can Structural Rules (HCSR) (International Association of Classification
then be derived from Eq. (11) and the shape parameter recom- Societies, 2013) as:
mended by classification society rules.
ND ¼ 31:557  106 ðf 0 T D Þ=ð4log 10 LÞ; ð12Þ

where f 0 , generally taken to be equal to 0.85, is the proportion of

3.2.3. Number of cycles
the working time that accounts for the design life; TD is the design
The number of cycles of random alternating stress during the
life, year; and L is the rule length.
service time of the structure is given in the Harmonized Common

Table 1 3.3. Failure criterion

Principal parameters of the container ship.
Length between perpendiculars Lpp 320.0 m
Molded breadth B 42.8 m
Depth D 24.8 m
Draft T 13.0 m
Block coeff. (for L) Cb 0.6637
Speed (full load at T) U 25 kn
An upper limit to crack growth needs to be specified. Typically,
for ships and ocean engineering structures, the critical crack depth
has been variously taken to be (1) some fraction of the thickness of
the plate (Doshi and Vhanmane, 2013); or (2) the thickness of the
plate (Shetty and Baker, 1990); or (3) derived from a fracture cri-
terion (Righiniotis and Chryssanthopoulos, 2004). The first two
criteria are the most frequently used. The half-thickness of the

Fig. 11. PATRAN FEM model of the whole ship.

Fig. 12. Contours of the maximum principal stresses around the crack.
 X. Yan et al. / Ocean Engineering 113 (2016) 246–254
Fig. 13. Load sequence of random stress range (piece).
Fig. 14. PATRAN FEM model with shell and solid elements.
plate was selected as the critical crack depth in the present study
for safety considerations.
4. Example application of the method for a ship hatch corner
Since the main purpose of the paper is to introduce a procedure
of fatigue crack propagation prediction for complex structures and
loads, the location of the crack in the hatch corner is selected.
Hatch corner in ships with large hatch openings is one of critical
locations for fatigue assessment. Since the structures and the
stress field in this area are both very complex, it is difficult to carry
out fatigue assessment by traditional FCP-based methods, and very
few relevant reports on the subject appear in the literature.
The particular object selected in this study was a hatch corner
in the vicinity of the forward engine room bulkhead of a large
container ship. The relevant parameters of the ship are shown in
Table 1. The design life was set at 25 years. The PATRAN finite
element model of the whole ship and the mesh grids around the
hatch corner are shown in Fig. 11. The thickness of plate A in Fig. 11
is 20 mm; the thickness of plate B is 68 mm. Plate B consists of
high-strength steel, with yield stress 355 MPa. Plates A and B are
welded together. The length of the fillet weld is assumed as
10 mm; the angle is 45°. The stress field in plate B was determined
by static analysis of the whole ship. Coupled with the low
253
Fig. 15. SIF for deepest point (a) and surface point (b) of surface crack, for design
wave load.
resistance to brittle fracture of high-strength steel, a semi-elliptical
surface crack at the weld toe on plate B was assumed. By means of
comprehensive fracture mechanics investigations of the defects
given by relevant literature (Germanischer Lloyd, 2009; Doerk and
Rörup, 2009), the depth and length of the surface crack were set at
3 mm and 20 mm.
The refined FE mesh at the hatch corner is shown in the inset of
Fig. 11. The main load parameter in this location was the torsion
and bending moment. The scatter diagram for the North Atlantic
recommended by DNV (2003) was selected. The probability level
of exceedance was set at 10  4.
The stress response of a reference node around the crack for
the design wave loading was 61.7 MPa (Fig. 12). Therefore, the
corresponding stress range was determined for a probability level
of exceedance of 10  4 at 123.4 MPa. The HCSR provides for the
shape parameter h to be set at 1 in an area of this kind. Then the
scale parameter results in q ¼13.4 from Eq. (11). The number of
alternating stress cycles during the service time was given by Eq.
(12) as approximately 6.7  107; therefore 6.7  107 pseudo-
random numbers are generated as the random alternating stress
ranges over 25 years. A segment of the random alternating stress
ranges are shown in Fig. 13.
Using the method described above, the local fine mesh of shell
elements was swept to solid elements and MPCs connecting the
shell elements to the solid elements were created; see Fig. 14. Then
the corresponding solid model containing a surface crack was
created in ANSYS, and the macro patran2ansys.mac was applied to
achieve the necessary transformation, rotation, interpolation and
loading. The calculated SIFs for cracks of different sizes are shown
254
140
120
Critical crack length
100
Crack size (mm)
Critical crack depth
80
60 Crack length 2c
40 Crack depth a
20
0
0 5 10 15 20
Service time (years)
Fig. 16. Crack growth curve.
in Fig. 15, in which the red points are the scattered values of the
calculated SIFs, and the smooth surface is interpolated by
triangulation.
The unique curve model was applied using the Paris model
parameters given by the Hobbacher (2009): the Paris multiplicative
constant C¼1.5  10  11 and exponent m¼ 3. The threshold value of
pffiffiffiffiffi
the SIF range equivalent to R¼ 0 is ΔK th0 ¼ 6MPa m. The SIF in the
calculation was obtained by triangular interpolation from the
scattered SIF data. The residual stress and the load sequence were
taken into consideration. Another program was written to calculate
the cycle-by-cycle fatigue life, based on FCP. The growth curve of
the target crack at the hatch corner was then derived (Fig. 16). It is
concluded from the figure that the crack was predicted not to reach
the critical size within 25 years, and therefore the hatch corner
meets the design requirement. However, if the ship continues to be
used beyond the specified 25-year period, the crack will propagate
to its critical size in the 32nd year—that is, the fatigue life of the
hatch corner is 32 years. It is remarkable that the crack propagates
slowly in the first few years then more rapidly in the final years of
the design life of the vessel, and therefore it is necessary to provide
for it to be inspected after it has been in service for a few years.
5. Conclusions
In order to be able to apply fatigue crack propagation theory
more broadly in engineering practice than is now possible, this
work has attempted to settle the problems of SIF calculation and
fatigue loading spectrum generation.
Since the current empirical formulas for SIF determination of 3-
D cracks are only suitable for simple structures and loads, it is
necessary to find a general and accurate method to calculate SIF of
cracks in complex structures subjected to unevenly distributed
loading as exemplified by the forces generated by ocean waves on
a ship. The proposed method of SIF calculation was described in
detail and verified by comparison with several empirical formulas.
The method, which is suitable for complex structures and loads,
combines the PATRAN finite element model with the advantages
of ANSYS for SIF calculation.
In addition, a method of generating the fatigue loading spec-
trum for marine structures based on a design wave approach was
introduced in this study to solve the problem of the definition of
nominal stress. When the proposed SIF calculation method is used,
X. Yan et al. / Ocean Engineering 113 (2016) 246–254
the fatigue load spectrum can be derived from the stress response
of a reference point in the area prone to fatigue, by using a design
wave approach.
Finally, an example of the method was applied to a hatch cor-
ner in a large container ship. A combination of the fatigue load
spectrum generation method, SIF calculation method and the
unique curve FCP model has provided a procedure for fatigue life
prediction based on fracture mechanics for ship details and other
ocean engineering structures.
Acknowledgments
The research reported in this paper is supported by the
National Natural Science Foundation of China, Project no.
51279102. The support is gratefully acknowledged. The authors
25 30 35 40
would also like to express a word of thanks to Dr. Shifeng Ding and
Mr. Ying Hong for their help.
References
Bowness, D., Lee, M.M.K., 2000. Prediction of weld toe magnification factors for
semi-elliptical cracks in T-butt joints. Int. J. Fatigue 22 (5), 369–387.
British Standards Institution, 2005. Guide to Methods for Assessing the Accept-
ability of Flaws in Metallic Structures. (BS7910).
Cui, W.C., 2003. A feasible study of fatigue life prediction for marine structures
based on crack propagation analysis. Proc. Inst. Mech. Eng. Part M: J. Eng. Marit.
Environ. 217 (1), 11–23.
Cui, W., Wang, F., Huang, X., 2011. A unified fatigue life prediction method for
marine structures. Mar. Struct. 24 (2), 153–181.
Det Norske Veritas (DNV), 2003. Fatigue Assessment of Ship Structures.
Doerk, O., Rörup, J., 2009. Development of toughness and quality requirements for
YP47 steel welds based on fracture mechanics. In: Proceedings of the 19th
International Offshore and Polar Engineering Conference. ISOPE, 386–391.
Doshi, K., Vhanmane, S., 2013. Probabilistic fracture mechanics based fatigue eva-
luation of ship structural details. Ocean Eng. 61, 26–38.
Fricke, W., 2003. Fatigue analysis of welded joints: state of development. Mar.
Struct. 16 (3), 185–200.
Fricke, W., Cui, W., Kierkegaard, H., et al., 2002. Comparative fatigue strength
assessment of a structural detail in a containership using various approaches of
classification societies. Mar. Struct. 15 (1), 1–13.
Germanischer Lloyd, 2009. Chapter 5: Supplementary rules for application of steel
with yield strength of 460 N/mm2. Rules for Classification and Construction, I-
Ship Technology, Part 1 – Seagoing Ships. Self Publishing, Hamburg.
IACS, 2013. Harmonised Common Structural Rules.
Hobbacher, A., 2009. Recommendations for fatigue design of welded joints and
components. Welding Research Council, Shaker Heights, OH.
Huang, X., Moan, T., Cui, W., 2008. An engineering model of fatigue crack growth
under variable amplitude loading. Int. J. Fatigue 30 (1), 2–10.
Huang, X., Moan, T., Cui, W., 2009. A unique crack growth rate curve method for
fatigue life prediction of steel structures. Ships Offshore Struct. 4 (2), 165–173.
Newman Jr., J.C., Raju, I.S., 1981. An empirical stress-intensity factor equation for the
surface crack. Eng. Fract. Mech. 15 (1), 185–192.
Okawa, T., Sumi, Y., 2008. A computational approach for fatigue crack propagation
in ship structures under random sequence of clustered loading. J. Mar. Sci.
Technol. 13 (4), 416–427.
Paris, P.C., Gomez, M.P., Anderson, W.E., 1961. A rational analytic theory of fatigue.
Trend Eng. 13 (1), 9–14.
Rhee, H.C., Han, S., Gipson, G.S., 1991. Reliability of solution method and empirical
formulas of stress intensity factors for weld toe cracks of tubular joints. Off-
shore Mech. Arct. Eng. 3, 441–452.
Righiniotis, T.D., Chryssanthopoulos, M.K., 2004. Fatigue and fracture simulation of
welded bridge details through a bi-linear crack growth law. Struct. Saf. 26 (2),
141–158.
Shetty, N.K., Baker, M.J., 1990. fatigue reliability of tubular joints in offshore struc-
tures: crack propagation model. In: Proceedings of the 9th International Con-
ference on Offshore Mechanics and Artic Engineering, 2, 33–40.
Sumi, Y., 2013. Fatigue crack propagation in marine structures under seaway
loading. Int. J. Fatigue.
Sumi, Y., Inoue, T., 2011. Multi-scale modeling of fatigue crack propagation applied
to random sequence of clustered loading. Mar. Struct. 24 (2), 117–131.