Available online at www.sciencedirect.

com

ScienceDirect
Publishing Services by Elsevier
International Journal of Naval Architecture and Ocean Engineering 8 (2016) 360e374
http://www.journals.elsevier.com/international-journal-of-naval-architecture-and-ocean-engineering/

Ultimate torsional strength of cracked stiffened box girders with a large deck
opening
Lei Ao a,b, De-Yu Wang a,b,*
a
State Key Laboratory of Ocean Engineering, Shanghai Jiao Tong University, Shanghai, China
b
The Collaborative Innovation Center for Advanced Ship and Deep-Sea Exploration, Shanghai, China
Received 18 December 2015; revised 6 April 2016; accepted 15 April 2016
Available online 15 July 2016

Abstract

The present paper studies the ultimate torsional strength of stiffened box girders with large deck opening due to the influence of cracks. Three
types of hull girders with different spans are provided for comparison. Potential parameters which may have effects on the torsional strength
including the mesh refinement, initial deflection, material strain hardening, geometric properties of crack and stiffener are discussed. Two new
concepts that play an significant role in the ultimate strength research of damaged box girders are introduced, one of which is the effective
residual section (ERS), the other is the initial damage of the failure zone (IDFZ) for intact structures. New simple formulas for predicting the
residual ultimate torsional strength of cracked stiffened box girders are derived on the basis of the two new concepts.
Copyright © 2016 Production and hosting by Elsevier B.V. on behalf of Society of Naval Architects of Korea. This is an open access article
under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).

Keywords: Stiffened box girders; Residual ultimate strength; Initial damage; Nonlinear FEM

1. Introduction
The ultimate strength analysis of ship hull structures has
been regarded as a necessary safety index in ship design along
with the deepening of the research. For a common vessel, the
vertical bending moment is frequently evaluated as one pri-
mary component of the hull girder loads and the overall lon-
gitudinal bending state is considered as the most dangerous
condition. Several experiments of simple box girders under
pure bending moment have been carried out to investigate
their strength and trace the process of the global progressive
collapse during recent decades (Nishihara, 1984; Gordo and
Soares, 2009, 2014; Saad-Eldeen et al., 2012). The strength
of plates and stiffened plates have been also widely studied as
ship hull's typical structural elements subjected to axial
* Corresponding author.
E-mail address: dywang@sjtu.edu.cn (D.-Y. Wang).
Peer review under responsibility of Society of Naval Architects of Korea.
compression and tension (Ueda et al., 1995; Paik and Kim,
2002, 2008a, 2008b; Zhang and Khan, 2009; Khedmati
et al., 2010; Shanmugam et al., 2014; Tanaka et al., 2014;
Tekgoz et al., 2015). The buckling and post-buckling charac-
teristics of hull girders under bending moment and combined
loads have also been investigated to assess the hull girder
strength (Yao, 2003; Paik et al., 2008c; Kim and Yoo, 2008;
Benson et al., 2013; Saad-Eldeen et al., 2013). Paik and
Thayamballi (2006) presented the ultimate strength formulas
of structural components and global system structures based
on recent advances in the areas related to the ULS design of
ships and ship-shaped offshore structures.
However, it has been realized that the horizontal bending
moment and torque may approach or exceed the vertical
bending moment when severe roll motion of ships happens in
rough waves. For some types of ships such as container ships,
large bulk carriers and other engineering ships, they are much
more vulnerable to the torsional moment with a lower
torsional rigidity due to the impact of large deck openings.
Senjanovic et al. (2014) executed the global hydroelastic

http://dx.doi.org/10.1016/j.ijnaoe.2016.04.002
2092-6782/Copyright © 2016 Production and hosting by Elsevier B.V. on behalf of Society of Naval Architects of Korea. This is an open access article under the
CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).
 L. Ao, D.-Y. Wang / International Journal of Naval Architecture and Ocean Engineering 8 (2016) 360e374
loading and response of ultra-large container ships by using a
sophisticated beam model based on the advanced beam theory
considering shear influence on bending and torsion and a hy-
drodynamic model was proposed in a condensed form with a
consistent formulation of restoring stiffness. Shi and Wang
(2012) carried out an experiment of the ultimate hogging
strength of a container ship's hull structures and a test model
was designed on the basis of derived similar conversion re-
lationships to reflect the possible failure modes under hogging
moment. Few experiments about hull structures under
torsional loads have been completed except Sun's test of a ship
type girder with a large deck opening which reflects the
possible failure modes of large open deck ships under pure
torque (Sun and Soares, 2003). Tanaka et al. (2015) recently
performed the collapse tests and analysis of a container ship
hull girder model under combined bending and torsion. A
simplified method was proposed by applying Smith's method
to idealized beam elements considering the warping and shear
stresses obtained by the torsional analysis.
It is observed that the initial deflection due to fabrication
and welding of stiffened structures has a significant influence
on the ultimate strength of plate structures which has to be
taken into consideration in terms of ultimate strength analysis
for welded steel structures. Ozguc et al. (2006) adopted
simplified formulas to analyze hull girder ultimate strength on
four ISSC benchmark vessels considering initial imperfections
for both hogging and sagging cases subject to vertical bending
and pointed out the most crucial point for a simplified method
was the ability to simulate the actual collapse mode.
Vhanmane and Bhattacharya (2008) attempted to improve the
IACS CSR method by introducing the stressestrain relation-
ship considering initial imperfection including both initial
deflection and welding induced residual stresses in the plating
between stiffeners. The influence of the initial imperfections
was introduced through the effective width of the attached
plating in stiffener-plating combined element. The shape of
initial deflection in actual panels is normally complex, how-
ever the shape form of a Fourier series function is widely used
in numerical calculation (Paik and Lee, 2005; Fujikubo et al.,
2005).
A ship is normally designed to be operating at least 20
years under abominable sea states. During this period, certain
numbers of hull structures will suffer various damages
including the steel corrosion, fatigue cracks and mechanical
dents due to collision or object falling. These subsequent de-
fects may have great negative effects on the strength of
damaged structures. Therefore, an assessment of residual ul-
timate strength for structures in damaged conditions is indis-
pensable to determine the vessel's recoverability and
seaworthiness and what measures are appropriate to be applied
as repair or replacement. Brighenti (2005) carried out the
buckling analysis of variously cracked rectangular elastic thin
plates under tension and compression. the critical load
multiplier is determined for different relative crack length,
crack orientation and Poisson's coefficient of the plate's ma-
terial. Moreover a simple approximate theoretical model to
explain and predict the buckling phenomena in cracked plates
361
under tension. Qi and Cui (2006) proposed an advanced
analytical method to calculate the ultimate strength of intact
and damaged ship hulls coupled with an elasticeplastic
method and the analytical method was verified to be suitable
by comparing with model tests and nonlinear finite element
analyses.
The residual ultimate strength characteristics of steel plates
with longitudinal cracks under axial compression was exam-
ined through experimental and nonlinear finite element
methods (Paik et al., 2005; Paik, 2009). Major parameters
including the crack orientation, the crack location, the crack
size, the plate thickness, and the plate aspect ratio are
considered. The buckling failure can often precede strength
failure in thin plates when there are imperfections such as
openings and cracks. Thus as a result, Seifi and Khoda-yari
(2011) carried out the Experimental and numerical studies
on the critical buckling load of cracked plates subjected to
axial compression. Pan et al. (2013) proposed an improved
hybrid semi-analytical method for calculating elastic buckling
load of a thin plate with a central crack under axial
compression. Xu et al. (2014) analyzed the residual ultimate
strength of stiffened panels with locked cracks under axial
compressive loading by changing the geometrical character-
istics of cracks and panels. The further propagation of the
cracks was prevented and the transverse cracks were verified
to have much more significant affects than the longitudinal
ones. Alinia et al. (2007) investigate the influence of central
cracks on the residual strength and stiffness degradation of
shear panels using the finite element method. It was shown
that the length and angle of cracks can magnify each other
resulting in extensive loss in load-bearing capacity of shear
panels. Brighenti and Carpinteri (2011) analyzed the buckling
and fracture collapse mechanisms in an elastic rectangular thin
plate with a central straight crack under shear loading. It was
illustrated that fracture failure can occur replacing buckling in
cracked plates under shear only for very low fracture tough-
ness. Wang et al. (2015) carried out the ultimate shear strength
analysis of intact and cracked stiffened panels. It was shown
that the effect of the web height of the stiffeners on shear
strength can be ignored.
Most of the researches are focused on a plate or stiffened
plates with cracks, few authors have paid attention to the
reduction of the ultimate torsional strength for stiffened hull
girders with large deck openings due to crack damages, which
is an significant issue to concern. Series of analyses about the
ultimate torsional strength of stiffened box girders with large
deck opening have been carried out by using the Finite
Element Method to evaluate the influence of cracks. Three
different spans of hull girders of the developed finite element
model are considered to estimate the effects of the longitudinal
length. The influence of various factors including the mesh
refinement, initial deflection, material strain hardening, geo-
metric properties of crack and stiffener are discussed. Two
new concepts that play an significant role in the ultimate
strength analysis of damaged box girders are introduced, one
of which is the effective residual section (ERS), the other is
the initial damage of the failure zone (IDFZ) for intact girders.
362 L. Ao, D.-Y. Wang / International Journal of Naval Architecture and Ocean Engineering 8 (2016) 360e374
A new simple formula for predicting the residual ultimate
torsional strength of cracked stiffened box girders is developed
based on the new concepts.
2. Description of stiffened girder with cracks
In the view of ship designers, the vertical bending moment
is the most essential component of ship loads for most ship
types. However, it has been realized that the torsional moment
is equally important for ships with large deck openings. In the
present application, large deck openings are defined when one
condition is satisfied below according to the IACS Rules:
(1) Total width of hatch openings in one transverse section
exceeding 70% of the ship's breadth: b/B
0.7
(2) Total length of hatch opening exceeding 89% of hold
length: lH/lBH
0.89
(3) Total width of hatch openings in one transverse section
exceeding 60% of the ship's breadth and length of hatch
opening exceeding 70% of hold length: b/B
0.6 and lH/
lBH
0.7
A simplified ship hull stiffened girder with large deck
opening under crack damage is designed based on the
torsional experiment by Hai-Hong Sun and the IACS Rules.
The l/2 þ 3 þ l/2 bay model consists of three full bays plus
two half bays in the longitudinal direction for the consider-
ation of the influence of boundary condition and the opening
length. The model was strengthened by longitudinal flat-bar
stiffeners and transverse frame system. Table 1 gives the
main dimensions of the model. Fig. 1 and Fig. 2 give the
schematic of the stiffened large opening girder and the typical
sections. The deck opening is between two bulkhead with size
of 1140mm  600mm. The thickness of the model plates was
5 mm. The sizes of stiffeners and transverse frames were
20mm  5mm and 60mm  5mm respectively. The crack is
located beside the plates. The parameters s1 and s2 are pointed
out to indicate the influence of the distributed position of
cracks. It is assumed that cracks are holding perpendicular to
the box girder in the longitudinal direction. The cracks are also
assumed to be through the thickness. To have a better view of
the location of the cracks, the girder panels are divided into
several independent zones as shown in Fig. 3. The stiffened
girder is separated into A1eA3 and B1eB3 zones in the
longitudinal direction and E1eE3 zones for the transverse
direction.
Table 1
Main dimensions of the hull girder.
Description Value (mm)
Spacing of transverse frames, a 200,300,400
Girder breadth, b 800
Girder height, c 600
Deck opening length 1140
Deck opening width 600
Spacing of stiffeners 150
The premised or existing cracks can propagate as the
torsional loads are applied or increased, and subsequently
structural consequences can significantly differ from those
where the premised cracks are constant in size and geometry.
However, the unstable cracks can cause unstable propagation
at a certain stress level which means the crack propagation
path and the increased crack length due to crack growth are
indefinite. The crack propagation is another issue mainly
focusing on the proposition of reasonable crack propagation
criteria based on fracture mechanics or damage mechanics. In
this case, the cracks are considered as a variable model and
their evolvement rules are defined accurately for the time-
variant reliability assessment. This issue is much more
different from that discussed in this paper. Meanwhile, the
torsional buckling usually occurs prior to the crack propaga-
tion due to the thin thickness of the plates. In order to simplify
the analysis, the cracks are considered as non-propagating
cracks and the effects of cracks are analyzed as a function
of crack type, crack location and crack size.
3. Sensitivity analysis of finite element size
The element size should be chosen properly fine enough to
get the required numerical accuracy as well as spending less
time for calculations. The total element size was chosen
ranging between 5 mm, 10 mm, 20 mm for comparison shown
in Fig. 4. Fig. 5 indicates that the mesh size has little effect on
the calculated results when the size is less than 10 mm. Some
authors have indicated that the value of the ultimate strength is
very close with different number of elements near the crack
tips for cracked plates as long as the total mesh is fine enough.
However, it is proven the equivalent stress distributions at the
ultimate limit state near the crack tips are quite different for
different local element size. To figure out the influence of the
element number near the crack tips on the ultimate strength,
mesh size of 2 mm, 5 mm and 10 mm near the crack tips are
taken account for comparison by keeping the total element
size in 10 mm. The FEM mesh model was shown in Fig. 4. As
shown in Fig. 6 the percentage difference due to the element
size of crack tips is less than 3%. If the total size is fine enough
and the crack is short, no mesh refinement near the tips needs
to be considered to get the precise ultimate strength for box
girders. However, further study shows that the crack tip
refinement has to be counted for study of the stress distribution
near the tips when cracks grow larger and larger. In this paper,
the total element size was taken as 10 mm and no local mesh
refinement around the cracks was applied when cracks move
along the longitudinal direction but element size of 5 mm is
taken around the crack tips when changing the crack size.
4. Modeling of the initial imperfection
For stiffened panels, the initial deflection of a local plate
may be introduced by the fillet welding between panels and
stiffeners during the fabrication process. The most accurate
method is by experiment which is only available for the test
specimens. In most of the theoretical analyses of stiffened
 L. Ao, D.-Y. Wang / International Journal of Naval Architecture and Ocean Engineering 8 (2016) 360e374 363

Fig. 1. Schematic diagrams of stiffened large opening girder with cracks.

Fig. 2. Schematic diagrams of typical sections.

Fig. 3. Separated zones in the longitudinal and transverse direction for hull girders.

plates, a thin-horse mode of initial deflection assumed to have years. The so-called thin-horse mode which indicates the
the same shape as the buckling failure mode of panels is typical shape of initial deflection of stiffened structures has
widely employed in the numerical analysis during recent been proven by several authors. One example of initial
364 L. Ao, D.-Y. Wang / International Journal of Naval Architecture and Ocean Engineering 8 (2016) 360e374

Fig. 6. Comparison of torsional moment versus twisting angle curves for

different local mesh refinement.

initial deflection of plates in the longitudinal direction has a

multi-wave shape, which has a half wave shape in the trans-
verse direction (Paik, 2007).
Three types of initial deflection modes are accounted
respectively for the local plates between stiffeners, the inde-
pendent stiffeners and plate related stiffeners:
The thin-horse mode of initial deflection of local plates of
panel, wop

mpx
py
wop ¼ do sin sin ð1Þ
a b
Fig. 4. Diagram of variation of total and local mesh element size of cracked where do is a parameter accounted for the maximum amplitude
girder. of deflection varying from bay-to-bay and from span-to-span,
a is the length of local plate and b is the breadth. m is the half
wave number of the local plate taken as an integer corre-
sponding to a/b which must satisfy the following equation:
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
a=b  mðm þ 1Þ ð2Þ
Column-type initial deflection of plate related stiffeners,
wos
a
px
py
wos ¼ sin sin ; B ¼ 2b ð3Þ
1000 a B
Side-ways initial deflection of independent stiffeners due to
angular rotation about panel-stiffener, vos
a
px
vos ¼ $z$sin ð4Þ
1000hw a

where hw is the stiffener's web height.

The peak amplitude of local plate deflection do is taken as
Fig. 5. The torsional moment versus twisting angle curves of various total
the equation of the slenderness ratio b :
mesh sizes. 8
< 0:025b2 t for slight level
do ¼ 0:1b2 t for average level ð5Þ
deflection of the plate between stiffeners is the inner bottom : 2
0:3b t for severe level
plating of an existing handy-sized Bulk Carrier which has a
pffiffiffiffiffiffiffiffiffiffiffi
shape of wave and multi-wave, measured by some authors where the slenderness parameter: b ¼ ðb=tÞ sY =E , t is the
(Yao et al., 1998). Another experiment for welded aluminum thickness of plate, E is Young's modulus, and sY is the yield
plate structures has also shown that the fabrication-related stress of material.
 L. Ao, D.-Y. Wang / International Journal of Naval Architecture and Ocean Engineering 8 (2016) 360e374 365

On the contrary, it is much more inconvenient to consider 0.50

the real initial deflection patterns for complicated ship hull
structures. In this case, the initial deflections are further
simplified to the eigen buckling mode under corresponding 0.45
loading conditions. It is indicated that deflection shape of the
buckling mode under compression is much closer to the real

Tu/Tp
initial deflection pattern for stiffened plates. However, it has
0.40
not been proven suitable for more complicated stiffened
structures such as stiffened box girders. Tree types of initial
deflection are introduced respectively by the buckling mode no initial deflection
0.35 thin horse mode
method and the thin-horse method for different levels of
torsional buckling mode by eigen analysis
deflection shown in Fig. 7. Fig. 8 and Fig. 9 give the effect of compressional buckling mode by eigen analysis
the deflection amplitude on the strength value. It is declared
that the ultimate strength of stiffened girder is much more 0.30
0.0 0.5 1.0 1.5 2.0 2.5
sensitive to the initial deflection produced by the thin-horse
method. The deflection by the buckling mode method has Initial deflection amplitude(mm)
little influence on the strength value in contrast. However, this Fig. 8. Effect of initial deflection amplitude on the ultimate torsional moment
method makes it easier to achieve convergence for large using different methods.
complicated structures. As the amplitude increases, the
structural strength has a sharp reduction. The amplitude of the (
sY if εeq  εplat
initial deflection for further research in this paper is taken as seq ¼  n
2.27 mm without special instruction. K εeq  εo otherwise
ð6Þ

s 1=n
Y
5. Material property εo ¼ εplate 
K
To trace the material elasticeplastic process and ensure the where K, n, εplat are the constants based on the real material
accuracy of the collapse analysis, the material characteristics curve. It is used for Q345 in this study as follows:
must be adequately represented in the form of a true K ¼ 932MPa, n ¼ 0.2084, εplat ¼ 0.0085.
stressestrain curve in the FEM analysis. A modified power The modified power law gives the proper stressestrain
law is employed to accurately describe the stressestrain curve curve corresponding to the true material characteristics,
instead of the real material curve. Functions of the power law however, the real stressestrain curve must be acquired to get
are shown as below: the constants. In case of not knowing the true material curve,

Fig. 7. Comparison of initial deflection shapes introduced by different methods.

366 L. Ao, D.-Y. Wang / International Journal of Naval Architecture and Ocean Engineering 8 (2016) 360e374

0.5 0.5

0.4 0.4

0.3 0.3

Tu/Tp
Tu/Tp

0.2 0.2 no strain hardening

no initial deflection hardening rate 0.005
deflection amplitude 0.19mm hardening rate 0.01
deflection amplitude 0.76mm hardening rate 0.02
0.1 0.1
deflection amplitude 2.27mm hardening rate 0.03
stress-strain function

0.0 0.0
0.00 0.01 0.02 0.03 0.04 0.00 0.01 0.02 0.03 0.04
Twisting angle(rad) Twisting angle(rad)

Fig. 9. The torsional moment versus twisting angle curves for different initial Fig. 11. The torsional moment versus twisting angle curves obtained by using
deflection amplitudes. different material stressestrain curves.

another simplified bilinear material curve is introduced as simplified bilinear material characteristic is applied as 1% of
alternative. The bilinear stressestrain curve can be expressed the hardening rate is adopted for further study in this paper.
in following equation:
 
  1 1 6. Results of FEM
εeq ¼ seq  sY $  for seq
sY ð7Þ
nE E
In order to have a better understanding of the influence of
where n is the considered strain hardening rate. cracks on the collapse behavior of box girders with large deck
The material property is taken as: opening, the ultimate strength of an intact girder is taken into
consideration to have a previous view of the post-ultimate
E ¼ 205800MPa; sY ¼ 345MPa; sb ¼ 500MPa strength characteristic.
Fig. 10 gives a comparison of stressestrain curves obtained The boundary condition and loads are applied at the
by above-mentioned two methods by varying hardening rate. torsional centers as: front-end,
Fig. 11 shows that the ultimate strength of hull girders in- Ux ¼ Uy ¼ Uz ¼ 0,URz ¼ 0.06; back-end,
creases slowly as the strain hardening rate becomes larger. It is Ux ¼ Uy ¼ 0,URz ¼ 0.06. It can be seen that the torque
indicated that the calculated strength by bilinear material is implemented on the girder ends in the form of enforced
property has little different from that by the modified power displacement and other degrees of freedom are set free. The
law when the strain hardening rate is taken as n ¼ 1%. The relationship between torsional moment and the twisting curve
in solid line is illustrated in Fig. 12. It can be observed that the
ultimate strength of the torsional girder is approximately
3.93  108 N,mm. The torsional moment is indicated in a non-
550
dimensional value divided by the fully plastic torsional
500
moment Tp calculated by tY,2At. The load-bearing capacity
450 decreases sharply as well as the structural stiffness when the
400 max moment is reaching. It should be noted that when the
350 ultimate torsional state is reaching, the plastic failure and large
stress(MPa)

300 deformation region formed in the hull girder only concentrates

250
in local areas A1 and B3 and does not spread over the entire
baseline
model. The displayed collapse mode in the two regions nearby
200
formula the abrupt section is reasonable as the model is geometrical
150 hardening rate 1% symmetry restrained by the bulkheads and transverse frames
100 hardening rate 2% and the load is implemented in the form of enforced
hardening rate 3%
50 displacement of anti-symmetry at both ends of the model
0 shown in Fig. 13. Thus two concepts are introduced. One is the
0.00 0.03 0.06 0.09 definition of the effective residual section (ERS). Independent
strain sections E1 to E3 are marked out for cross section of the hull
Fig. 10. Comparison of material stressestrain curves obtained by different
girder. However, the collapse damage does not happen in all
methods. the sections during the ultimate limit state due to the disparity
 L. Ao, D.-Y. Wang / International Journal of Naval Architecture and Ocean Engineering 8 (2016) 360e374
0.5
8 stresses in other areas of the box girder. As shown in Fig. 13,
Tu=3.93x10 N mm
0.4
0.3
Tu/Tp
0.2
0.1
Intact girder without cracks
0.0
0.00 0.01 0.02 0.03
Twisting angle(rad)
Fig. 12. The torsional moment versus twisting angle curve for an intact girder
without cracks.
of the torsional stiffness for various sections. In this way, the
effective residual section is defined as sections that contain the
collapse failure regions. Another concept is the Initial Damage
of the Failure Zone (IDFZ). For a intact structure, the failure
mode is determined by the boundary condition and load forms
as well as its structure feature. However, the initial introduced
damage such as crack damage or reduction of thickness caused
by manufacture welding, structure fatigue or corrosion may
occur anywhere of the structure. The initial damage of failure
zone is defined when the initial damage happens in failure
zones of intact structure for example as region A1 and B3.
The torsional stiffness of a thin-walled box beam decreases
a lot due to the existence of the large deck opening. Thus, the
open section may twist much more under a given torsional
moment and large axial deformation occurs so that the cross
section will no longer remain flat. The surrounding of the cross
section cannot be assumed to be rigid in this case. When the
cross sections are free to warp, no warping stresses normal to
the cross section will be introduced. However, the warping
deformations will be partly restrained due to the cross
sectional discontinuities between cargo spaces. When warping
displacements are restrained, both warping and shear stresses
will be developed at the hull cross section during the torsional
collapse process. The self-equilibrium of the warping stresses
Fig. 13. Membrane stress distribution in a perfect girder at the ultimate limit state under torsional load around z direction:(a)Schematic diagrams of load applying
(b)membrane stress distribution of a perfect girder at the ultimate state under torque.
367
will introduce tensile stresses in some areas and compressive
two corners of the deck opening along the diagonal are under
compression and the other two are under tension when the hull
girder is under torsional load. The compressive corners reach
to local plastic buckling when the girder is under ultimate
torsional state and the warping stress reaches its maximum in
the corner region. During the post-buckling of the local panels,
the warping effects can become significant.
In most cases, cracks can be found at stress concentration
regions or areas with welding defect. For actual vessels, crack
damage usually appears at places such as the beginning or
ending of the welding bead, toes of brackets, the intersection
of welds, or cut-outs and some cracks are not visible due to
0.04
lack of cleanliness or difficulty of access. Paik (2009)
analyzed the ultimate strength of cracked steel plates as a
function of crack size, crack orientation and crack location and
pointed out that the reduction impact of cracks can be sig-
nificant different for various crack locations. The IACS has
summarized common crack types, damage cause and repair
recommendations in container carriers. Cracks can be found at
places such as the hatch side coaming, deck girder, side shell
plating or longitudinal bulkhead plating and bottom shell
plating or inner bottom plating along the drain hole for
container ships. In order to simplify the analysis, appropriate
presumptions should be made about crack types, crack loca-
tion and crack length for the box girder. Single and double
cracks are considered in this paper and it is assumed the crack
propagates to be a certain length at that location before the box
model collapses. It is indicated that the transverse crack has a
biggest effect and the longitudinal has a smallest. Thus, only
transverse crack is considered relative to local panels, the di-
rection of which is vertical to the longitudinal direction of the
hull girder. The cracks are also considered to be moving along
the transverse direction and the longitudinal direction of the
box beam.
6.1. Influence of a single edge crack
A single crack is assumed to be located in side section of
E1 and bottom section E3 separately and section E2 is not
taken into consideration due to geometrical symmetry of the
hull girder along the vertical direction at the center of the cross
368 L. Ao, D.-Y. Wang / International Journal of Naval Architecture and Ocean Engineering 8 (2016) 360e374

section. A series of ultimate strength analysis is undertaken 1.2

first on a stiffened girder with a horizontal crack located in
region C1 and C2 of section E2 respectively with varying 1.0
crack length and position. Fig. 14 shows the torsional moment
versus twisting angle curve of a center crack with various sizes 0.8
in region C1. It illustrates that the crack length has little effects

Tu/Tu0
on the torsional path as well as the ultimate value of the torque
0.6
while it is located in region C1. Fig. 15 shows the influence of
crack position with fixed size on the girder's ultimate strength
when it moves along the horizontal direction in regions of 0.4 Crack length: d/b=0.5
Baseline
bottom section E3. It can be seen that the ultimate torsional
Crack located in C1
moment changes little as the crack position varies for large 0.2
Crack located in C2
constant crack length. The largest relative difference is 3.7%
when cracks move close to the edges in region C1. In this case, 0.0
the crack breaks the constraint condition of the failure zone 0.0 0.1 0.2 0.3 0.4 0.5 0.6
Crack location s4/b
provided by bottom section E3. However, the impact of the
bottom cracks is still negligible. This phenomenon can be Fig. 15. Non-dimensional characteristics of the torsional moment versus crack
explained section E1 and E2 play the main role in restraining position for box girders with cracks located in regions C1 and C2 with fixed
the maximum load-carrying capacity because the bottom length d/b ¼ 0.5.
section has the largest torsional rigidity which is the hardest to
be destroyed. Cracks located in section E3 did not transform 0.5
the situation in spite of reducing the section stiffness. Sections
E1 and E2 are appointed as the effective residual section in
which the influence of cracks is considerable. Otherwise the 0.4
crack is not taken account of for reduction of the structure
ultimate strength. 0.3
Fig. 16 shows the relationship between the structure
Tu/Tp

strength and the crack length as the crack is situated in the

region of A1. It can be seen that there is a significant 0.2
decreasing tendency with the increase of crack sizes. The Crack length
d/c=0 d/c=0.1
crack size is one of the main parameters that have negative d/c=0.2 d/c=0.3
0.1
reduction effects on the structural strength corresponding to d/c=0.4 d/c=0.5
the effective residual sections. Fig. 17 illustrates the variation
of box girders' ultimate strength with cracks transforming 0.0
from region A1 to A3. It can be observed that the ultimate 0.00 0.01 0.02 0.03 0.04
Twisting angle (rad)
strength is most sensitive to the crack size when the crack
located in the failure region A1 other than regions A2 and A3. Fig. 16. The torsional moment versus twisting angle curves for hull girders
with cracks located in region A1 by varying crack lengths.

0.5
In the failure zone, the girder strength decreases synchro-
nously with the growth of the crack size. However, the
0.4 effective crack length diminishes to some extent for the non-
failure areas. When cracks are located in one area, the
discontinuity of girder components caused by cracks will lead
0.3
to local mutation of the torsional stiffness. The mutation of the
Tu/Tp

torsional stiffness and the stress concentration of the crack tips

0.2 enhance the impact of the warping stress and shear stress to
the most when cracks are located near the upper edges in the
Crack length
d/b=0.1 d/b=0.2
failure zone.
0.1
d/b=0.3 d/b=0.4 Fig. 18 shows the variation trend of the box torsional
d/b=0.5 strength when cracks moving along the vertical direction in
0.0 separate regions. It can be figured out that the anti-twisted
0.00 0.01 0.02 0.03 0.04 ability of the hull girder reduces to the minimum when
Twisting angle(rad) cracks are closest to the upper and lower edges of the panels.
Fig. 14. The torsional moment versus twisting angle curves for hull girders When cracks are located in the edge of the panels, boundary
with cracks located in region C1 by varying crack lengths. constraints imposed by surrounding plates which have large
 L. Ao, D.-Y. Wang / International Journal of Naval Architecture and Ocean Engineering 8 (2016) 360e374 369

0.5 0.5

0.4
0.4

0.3

Tu/Tp
Tu/Tp

0.3 0.2
Crack at region A2
Crack at region A3
Crack length d/c=0.5
Crack at region A1
s1=200 s1=400
0.1
s1=600 s1=800
0.2
s1=1000

0.0
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.00 0.01 0.02 0.03 0.04
Crack length Twisting angle(rad)

Fig. 19. The torsional moment versus twisting angle curves for hull girders
Fig. 17. Effect of crack length on the ultimate torsional for cracks located in
with edge cracks moving in the longitudinal direction, crack length is d/
different regions.
c ¼ 0.5.

0.50 8
< de ¼ ð1  aÞdreal
0.45 Aecrack ¼ de  tcrack ð9Þ
:
AERS ¼ Aeplate þ Aestiffener
0.40
where de is the effective crack length. Aeplate and Aestiffener
0.35
are areas of the effective plates and stiffeners that contained in
Tu/Tp

0.30 the ERS. a ¼ 0 when the crack is situated in the failure zone,
otherwise a is taken as a relative constant and a < 1.
0.25
When the thickness of all the girder plates is selected as a
Crack length d/c=0.5 same value, Aecrack and AERS can be replaced by de and
0.20 Crack at A1
Crack at A2 LERS. The effective crack length is considered as a full length
0.15 Crack at A3 when the crack is located in the failure zone for a single
0.10
crack, otherwise the effective length is a proper reduction of
0.0 0.1 0.2 0.3 0.4 0.5 0.6 the actual length. It has been figured out section E3 is not a
Crack location s2/c ERS that components of section E3 should not be taken into
account as part of the total effective area of the cross section.
Fig. 18. Variation of ultimate torsional strength of hull girder with cracks
moving up and down in different regions.
0.5

effect on the structure strength are destroyed. The ultimate

strength is much less reduced when cracks move up and down 0.4
in regions A2 and A3 than that located in A1.
Fig. 19 shows the torsional moment versus twisting angle
0.3
curves of edge cracks with length of b/c ¼ 0.5 moving along
(T0-Tu)/T0

the longitudinal direction. It can be seen that as long as the

crack deviates from region A1, it has much less influence on 0.2
reducing the girder's ultimate strength.
The reduction factor of ultimate strength for hull girders Baseline
with a single crack is given as: 0.1
FEM results
T0  Tu Aecrack
Rf ¼ ¼ ð8Þ 0.0
T0 AERS
T0 is the ultimate strength of a intact girder without cracks,
0.0 0.1 0.2 0.3 0.4 0.5
Tu is the ultimate strength of cracked girder, Aecrack is the
effective reduced area due to cracks, AERS is the total area of de/Le-total
the effective residual section. Fig. 20. Comparison of results obtained from FEM with that from Eq. (8).
370 L. Ao, D.-Y. Wang / International Journal of Naval Architecture and Ocean Engineering 8 (2016) 360e374
1.1
1.0
0.9
0.8
0.7
0.6
Tu/T0
0.5
0.4
formula
0.3
span a=200
0.2 span a=300
0.1 span a=400
0.0
0.0 0.1 0.2 0.3 0.4 0.5 0.6
Crack size d/c
Fig. 21. Comparison of results from FEM and Eq. (8) for girders with different
spans.
The comparison of the FEM results with the baseline of Eq.
(8) has been shown in Fig. 20. It can be observed that the
function of the reduction factor coincides well with the FEM
results. When crack length is larger than d/c ¼ 0.5, the
reduction factor is a little larger than the FEM results. This is
reasonable because section E3plays much more important
role as a boundary constraint imposed on section E1and E2
when crack size grows larger and larger. Fig. 21 shows the
ultimate torsional strength of cracked girders of different
frame spacing with increasing crack length. It can be seen
that results obtained by formula is slightly smaller than
actual values when the frame space is less than 200 mm. As
the longitudinal span decreases, the transverse support
members including the bulkheads and transverse frames have
much more effect on resisting the torsional deformation. It is
observed that Eq. (8) can be successfully used for the ulti-
mate strength prediction of the cracked girder with a single
edge crack.
6.2. Influence of double edge cracks
Fig. 22 shows the schematic diagram of stiffened girder
with double edge cracks. It is assumed that regions A1 and B1
are the failure zones of intact box girder under torsional load.
Sd1 and Sd2 are considered as the longitudinal distance be-
tween edge cracks and centers of the failure zones respec-
tively. Fig. 23 shows the torsion moment versus twisting angle
curves for double cracked girders of different frame spacing
by changing the longitudinal position of cracks. It is illustrated
that when cracks deviate from the failure zone of intact
girders, they have much less effects on the torsional strength of
stiffened box girders. The difference rate of torsional ultimate
strength reaches maximum value of 15%, 19%, 25% for
girders with various longitudinal spans respectively parallel to
the crack deviation distance from the failure zone. When both
edge cracks are located in the failure zones, the ultimate
Fig. 22. Schematic diagram of stiffened girder with double edge cracks.
torsional strength reduces to the minimum which is regarded
as the most vulnerable condition.
To have a better view of the influence of double edge cracks
on the girder's strength, the most vulnerable condition of both
cracks located in the failure zones is taken into consideration
independently. Fig. 24 gives the torsion moment versus
twisting angle curves for double cracked girders with frame
spacing of 400 mm by changing the crack size. It can be seen
that the ultimate torsional capacity reduces gradually as the
crack length increases. When the crack length is larger than d/
c ¼ 0.4, the reduction ratio becomes smaller.
To figure out whether the existence of bottom crack has
influence on the girder strength with double edge cracks, a
contrast calculation is shown in Fig. 25. It is indicated that the
bottom cracks has little effect which can be ignored.
Fig. 26 shows a comparison of the FEM results with that
obtained from Eq. (8) for different crack sizes. It is observed
that the formula suits well with the FEM results when crack
length is below d/c ¼ 0.4, otherwise the ultimate torsional
strength obtained by Eq. (8) is much smaller than the actual
value. It is shown that the error rate increases constantly along
with the growth of crack size. In this case, Aetotal has to be
calculated separately based on the crack size. Eq. (10) is given
as follow for hull girders with double edge cracks. Fig. 27
shows that the results obtained by Eq. (10) agree well with
the FEM results which is verified to be reasonable.
T0  Tu Aecrack
Rf ¼ ¼
T0 Aetotal
Aetotal ¼ AERS ; d=c  d1
ð10Þ
ðd  d1 Þ
Aetotal ¼ AERS þ $AnonERS ; d1  d=c  d2
d2  d1
Aetotal ¼ AERS þ AnonERS ; d=c
d2
d1 and d2 are two critical crack lengths associated with the
frame spacing and stiffness of the cross section which are
taken as d1/c ¼ 0.3 and d2/c ¼ 0.6 for double cracked girder
with span of 400 mm.
 L. Ao, D.-Y. Wang / International Journal of Naval Architecture and Ocean Engineering 8 (2016) 360e374 371

0.6 0.5

0.5
0.4

0.4
0.3
Tu/Tp

Tu/Tp
0.3

0.2
0.2
sd1=0, sd2=0 sd1=0, sd2=200 sd1=0, sd2=0 sd1=0, sd2=300
0.1 sd1=0, sd2=400 sd1=200,sd2=200 0.1 sd1=0, sd2=600 sd1=300,sd2=300
sd1=200,sd2=400 sd1=400,sd2=400 sd1=300,sd2=600 sd1=600,sd2=600
0.0 0.0
0.00 0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08 0.00 0.01 0.02 0.03 0.04
Twisting angle(rad) Twisting angle(rad)

Frame span a=200mm Frame span a=300mm

0.5

0.4

0.3
Tu/Tp

0.2

sd1=0, sd2=0 sd1=0, sd2=400

0.1 sd1=0, sd2=800 sd1=400,sd2=400
sd1=400,sd2=800 sd1=800,sd2=800
0.0
0.00 0.01 0.02 0.03 0.04 0.05
Twisting angle(rad)

Frame span a=400mm

Fig. 23. The torsional moment versus twisting angle curves for hull girders with double cracks moving in the longitudinal direction, crack length is d/c ¼ 0.5.

6.3. Influence of the stiffeners 7. Conclusions

Fig. 28 shows the torsional moment versus twisting angle
curves of cracked hull girders with various stiffener numbers.
It can be seen that stiffener number of left-right section plays a
much more important role than that of bottom section. In this
way, it is a more effective method to enhance the girder's anti-
twisted ability by aggrandize the quantity of stiffeners con-
tained in left and right sections. Fig. 29 gives the variation
tendency of the ultimate torsional moment with single and
double cracks versus stiffener thickness. It is obvious the
torsional moment of box girders with double cracks is lower
than that with single crack. The thicker the stiffener, the larger
the increase in the ultimate strength due to stiffener's effect is
attained.
The nonlinear finite element analyses (Nonlinear FEA)
have been performed to determine the ultimate torsional
strength of cracked box girders with large deck openings. By
comparing three types of hull girders with different spans,
potential parameters which may have effects on the torsional
strength are discussed.
The element size near the crack tips has little effect on the
ultimate strength of stiffened box girders under torsional loads
when the crack is short and the total mesh is fine enough, thus
precise results can be achieved without local mesh refinement
around the tips. However, when cracks grow larger and larger,
the stress near the crack tip is much more sensitive to the mesh
density which can't be ignored to get accurate results. It is
372 L. Ao, D.-Y. Wang / International Journal of Naval Architecture and Ocean Engineering 8 (2016) 360e374

0.4 1.1
1.0
0.9
0.3 0.8
0.7
Tu/Tp

Tu/T0
0.6
0.2
0.5
0.4

0.1 Crack length 0.3

d/c=0.1 d/c=0.2 d/c=0.3 0.2
d/c=0.4 d/c=0.5 FEM results
0.1 Eq.8
0.0 0.0
0.00 0.01 0.02 0.03 0.0 0.2 0.4 0.6 0.8 1.0
Twisting angle(rad) crack size d/c
Fig. 24. The torsional moment versus twisting angle curves for hull girders Fig. 26. Comparison of FEM results for double cracked girders with that from
with double cracks located in region A1 and B3 by varying crack lengths. Eq. (8).

0.4 1.1
FEM results
1.0
Eq.10
0.9
0.3 0.8
0.7
Tu/T0

0.6
Tu/Tp

0.2
0.5
0.4

0.1 0.3
0.2
double edge crack d/c=0.5
double edge crack d/c=0.5and bottom crack d/b=0.5 0.1
0.0 0.0
0.00 0.01 0.02 0.03 0.0 0.2 0.4 0.6 0.8 1.0
Twisting angle(rad) crack size d/c
Fig. 25. The effects of bottom crack on the ultimate strength of girders with Fig. 27. Comparison of FEM results for double cracked girders with that from
double cracks. Eq. (10).

observed that the stiffened girder is much more sensitive to the
initial deflection produced by the thin-horse method under
pure torque and the deflected shape introduced by the buckling
mode method has little influence on the ultimate strength value
of hull girders in contrast.
Two new concepts are presented for the ultimate strength
research of damaged box girders, one of which is the Effective
Residual Section (ERS), the other is the Initial Damage of the
Failure Zone (IDFZ) for intact hull girders. It is illustrated that
the existence of cracks has little effects on the torsional path as
well as the ultimate value of the torque while it is located in
the non-ERS. Results show that the ultimate torsional moment
changes slightly only when edge cracks break the constraint
condition of the failure zone provided by the non-ERS which
impact is still negligible.
The crack size is one of the main parameters that have
considerable reduction effects on the structural strength when
cracks are located in the effective residual region. The
torsional moment reduces to the least when cracks move
longitudinally to the failure zone. The girder strength de-
creases synchronously with the growth of the crack size when
located in the failure zone. However, the effective crack length
diminishes to some extent corresponding to the real size for
the non-failure areas. The difference rate of ultimate torsional
strength reaches maximum value of 15%, 19%, 25% for box
girders with various longitudinal spans respectively parallel to
 L. Ao, D.-Y. Wang / International Journal of Naval Architecture and Ocean Engineering 8 (2016) 360e374 373

0.5 0.5

0.4 0.4

0.3 0.3

Tu/Tp
Tu/Tp

number of stiffener
no stiffener number of stiffener
0.2 one at each section 0.2 no stiffener
one at left-right and five at bottom one at each section
two at left-right and three at bottom one at left-right and five at bottom
three at left-right and one at bottom two at left-right and three at bottom
three at left-right and five at bottom
0.1 0.1 three at left-right and one at bottom
three at left-right and five at bottom

0.0 0.0
0.00 0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08 0.00 0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08
Twisting angle(rad) Twisting angle(rad)

Intact girder without cracks Damaged girder with one crack

Fig. 28. The torsional moment versus twisting angle curves for hull girders by varying stiffener numbers.

0.6 References

0.5 Alinia, M.M., Hosseinzadeh, S.A.A., Habashi, H.R., 2007. Influence of central
cracks on buckling and post-buckling behaviour of shear panels. Thin-
Walled Struct. 45 (4), 422e431.
0.4 Benson, S., AbuBakar, A., Dow, R.S., 2013. A comparison of computational
methods to predict the progressive collapse behaviour of a damaged box
girder. Eng. Struct. 48 (0), 266e280.
Tu/Tp

0.3
0.2
0.1
0.0
2 4 6 8 10
Stiffener thickness(mm)
Fig. 29. Comparison of ultimate torsional strengths for girders with various
cracks by changing the stiffener thickness.
the crack deviation distance from the failure zone. When both
edge cracks are located in the failure zones, the ultimate
torsional strength reduces to the minimum which is regarded
as the most vulnerable condition. The thicker the stiffener, the
larger the increase in the ultimate strength due to stiffener's
effect is attained. Finally, simple predicted formulas are pro-
pose based on the new concepts which are verified to be
rational contrasted with the FEM results.
Acknowledgments
The present work is supported by the Chinese Government
Key Research Project KSHIP-II Project (Knowledge-based
Ship Design Hyper-Integrated Platform) No 201335. The au-
thors would like to acknowledge the Project support.
Brighenti, R., Carpinteri, A., 2011. Buckling and fracture behaviour of cracked
thin plates under shear loading. Mater. Des. 32 (3), 1347e1355.
Brighenti, R., 2005. Buckling of cracked thin-plates under tension or
compression. Thin-Walled Struct. 43 (2), 209e224.
no crack Fujikubo, M., et al., 2005. Estimation of ultimate strength of continuous
single crack stiffened panel under combined transverse thrust and lateral pressure part
double crack 2: continuous stiffened panel. Mar. Struct. 18 (5e6), 411e427.
Gordo, J.M., Guedes Soares, C., 2014. Experimental analysis of the effect of
frame spacing variation on the ultimate bending moment of box girders.
Mar. Struct. 37, 111e134.
12 Gordo, J.M., Soares, C.G., 2009. Tests on ultimate strength of hull box girders
made of high tensile steel. Mar. Struct. 22 (4), 770e790.
Khedmati, M.R., Zareei, M.R., Rigo, P., 2010. Empirical formulations for
estimation of ultimate strength of continuous stiffened aluminium plates
under combined in-plane compression and lateral pressure. Thin Walled
Struct. 48 (3), 274e289.
Kim, K., Yoo, C.H., 2008. Ultimate strengths of steel rectangular box beams
subjected to combined action of bending and torsion. Eng. Struct. 30 (6),
1677e1687.
Nishihara, S., 1984. Ultimate longitudinal strength of mid-ship cross section.
Nav. Archit. Ocean Eng. 22, 200e214.
Ozguc, O., Das, P.K., Barltrop, N., 2006. A proposed method to evaluate hull
girder ultimate strength. Ships Offshore Struct. 1 (4), 335e345.
Paik, J.K., Kim, B.J., Seo, J.K., 2008a. Methods for ultimate limit state
assessment of ships and ship-shaped offshore structures: part III hull
girders. Ocean. Eng. 35 (2), 281e286.
Paik, J.K., Kim, B.J., Seo, J.K., 2008b. Methods for ultimate limit state
assessment of ships and ship-shaped offshore structures: part
Idunstiffened plates. Ocean. Eng. 35 (2), 261e270.
Paik, J.K., Kim, B.J., 2002. Ultimate strength formulations for stiffened panels
under combined axial load, in-plane bending and lateral pressure: a
benchmark study. Thin Walled Struct. 40 (1), 45e83.
Paik, J.K., Lee, M.S., 2005. A semi-analytical method for the elastic-plastic
large deflection analysis of stiffened panels under combined biaxial
374 L. Ao, D.-Y. Wang / International Journal of Naval Architecture and Ocean Engineering 8 (2016) 360e374
compression/tension, biaxial in-plane bending, edge shear, and lateral
pressure loads. Thin Walled Struct. 43 (3), 375e410.
Paik, J.K., Thayamballi, A.K., 2006. Some recent developments on ultimate
limit state design technology for ships and offshore structures. Ships
Offshore Struct. 1 (2), 99e116.
Paik, J.K., 2007. Characteristics of welding induced initial deflections in
welded aluminum plates. Thin-Walled Struct. 45 (5), 493e501.
Paik, J.K., 2009. Residual ultimate strength of steel plates with longitudinal
cracks under axial compressiondnonlinear finite element method in-
vestigations. Ocean. Eng. 36 (3e4), 266e276.
Paik, J.K., Kim, B.J., Seo, J.K., 2008c. Methods for ultimate limit state
assessment of ships and ship-shaped offshore structures: part II stiffened
panels. Ocean. Eng. 35 (2), 271e280.
Paik, J.K., Satish Kumar, Y.V., Lee, J.M., 2005. Ultimate strength of cracked
plate elements under axial compression or tension. Thin Walled Struct. 43
(2), 237e272.
Pan, Z., Cheng, Y., Liu, J., 2013. A semi-analytical analysis of the elastic
buckling of cracked thin plates under axial compression using actual non-
uniform stress distribution. Thin Walled Struct. 73, 229e241.
Qi, E., Cui, W., 2006. Analytical method for ultimate strength calculations of
intact and damaged ship hulls. Ships Offshore Struct. 1 (1), 153e163.
Saad-Eldeen, S., Garbatov, Y., Guedes Soares, C., 2012. Analysis of plate
deflections during ultimate strength experiments of corroded box girders.
Thin Walled Struct. 54, 164e176.
Saad-Eldeen, S., Garbatov, Y., Guedes Soares, C., 2013. Effect of corrosion
severity on the ultimate strength of a steel box girder. Eng. Struct. 49,
560e571.
Seifi, R., Khoda-yari, N., 2011. Experimental and numerical studies on
buckling of cracked thin-plates under full and partial compression edge
loading. Thin Walled Struct. 49 (12), 1504e1516.
 2014. Some
Senjanovic, I., Vladimir, N., Tomic, M., Hadzic, N., Malenica, S.,
aspects of structural modelling and restoring stiffness in hydroelastic
analysis of large container ships. Ships Offshore Struct. 9 (2), 199e217.
Shanmugam, N.E., et al., 2014. Experimental studies on stiffened plates under
in-plane load and lateral pressure. Thin Walled Struct. 80, 22e31.
Shi, G.J., Wang, D.Y., 2012. Ultimate strength model experiment
regarding a container ship's hull structures. Ships Offshore Struct. 7
(2), 1e20.
Sun, H.-H., Soares, C.G., 2003. An experimental study of ultimate torsional
strength of a ship-type hull girder with a large deck opening. Mar. Struct.
16 (1), 51e67.
Tanaka, S., Yanagihara, D., Yasuoka, A., 2014. Evaluation of ultimate strength
of stiffened panels under longitudinal thrust. Mar. Struct. 36, 21e50.
Tanaka, Y., Ogawa, H., Tatsumi, A., Fujikubo, M., 2015. Analysis method of
ultimate hull girder strength under combined loads. Ships Offshore Struct.
10 (5), 587e598.
Tekgoz, M., Garbatov, Y., Guedes Soares, C., 2015. Ultimate strength
assessment of welded stiffened plates. Eng. Struct. 84, 325e339.
Ueda, Y., Rashed, S.M.H., Paik, J.K., 1995. Buckling and ultimate strength
interaction in plates and stiffened panels under combined in-plane biaxial
and shearing forces. Mar. Struct. 8 (1), 1e36.
Vhanmane, S., Bhattacharya, B., 2008. Estimation of ultimate hull girder
strength with initial imperfections. Ships Offshore Struct. 3 (3),
149e158.
Wang, F., et al., 2015. Ultimate shear strength of intact and cracked stiffened
panels. Thin Walled Struct. 88, 48e57.
Xu, M.C., Garbatov, Y., Soares, C.G., 2014. Thin-walled structures residual
ultimate strength assessment of stiffened panels with locked cracks. Thin
Walled Struct. 85, 398e410.
Yao, T., Fujikubo, M., Yanagihara, D., Varghese, B., 1998. Influences of
welding imperfections on buckling/ultimate strength of ship bottom plating
subjected to combined bi-axial & lateral pressure. Thin walled Struct.
425e432.
Yao, T., 2003. Hull girder strength. Mar. Struct. 16, 1e13.
Zhang, S., Khan, I., 2009. Buckling and ultimate capability of plates and
stiffened panels in axial compression. Mar. Struct. 22 (4), 791e808.