Tests On Ultimate Strength of Hull Box Girders Made of High Tensile Steel

Marine Structures 22 (2009) 770–790
Contents lists available at ScienceDirect
Marine Structures
journal homepage: www.elsevier.com/locate/
marstruc
Tests on ultimate strength of hull box girders made

of high tensile steel
J.M. Gordo, C. Guedes Soares*
Centre for Marine Technology and Engineering (CENTEC), Technical University of Lisbon, Instituto Superior Técnico, Av.
Rovisco Pais, 1049 001 Lisbon, Portugal
a r t i c l e i n f o a b s t r a c t
Article history: The experimental results of the collapse of three box girders
Received 15 February 2006 subjected to pure bending moment are presented. The structures
Received in revised form 15 February 2009 are made of high tensile steel of 690 MPa of nominal yield stress
Accepted 12 August 2009
reinforced with bar stiffeners of the same material. The moment
curvature curves are presented covering the pre- and post-collapse
Keywords:
regions. The modes of collapse for each box girder are discussed
Ultimate strength
considering the variation on the panel’s slenderness. The concept
Hull girder strength
Bending moment of efficiency of high tensile steel structures is introduced. The
Residual strength concept is very useful to identify the governing parameters
affecting the ultimate strength of 3D structures under predomi-
nant bending moment.
Ó 2009 Elsevier Ltd. All rights reserved.
1. Introduction
The evaluation of ultimate capacity of ships under bending moment is a very important issue of the
structural design. It is associated with a global failure of the hull and the final result is normally the loss
of the ship, its cargo and human lives.
In the last years several papers have been published on the subject, most of them on the evaluation
of the ultimate bending moment of ships made of normal mild steel. The existing calculation methods
may be divided into two groups: finite elements methods and simplified methods. In the two groups
there has been a great activity and comparison between the different methods is available in the
literature [1].
* Corresponding author. Tel.: þ351 218417607; fax.: þ351 218474015.

E-mail address: guedess@mar.ist.utl.pt (C. Guedes Soares).
0951-8339/$ – see front matter Ó 2009 Elsevier Ltd. All rights reserved.
doi:10.1016/j.marstruc.2009.07.002
J.M. Gordo, C. Guedes Soares / Marine Structures 22 (2009) 770–790 771
The authors have developed a method [2] that has been validated against data from a full scale accident
[3] where the loading conditions could be well established and against some small scale experiments of
models representing simplified typical sections of ships [4–8]. The results of these comparisons showed
that the method can be used confidently on typical hull configurations and for normal steel.
Within a European project a concept of fast monohull was studied in which the use of very high
strength steel was aimed at allowing a lighter structure to be developed as described by Janssen [9].
Such large departure from normal practice motivated studies on fatigue strength and on ultimate
strength of structures made with this material. It is within the scope of this project that the present
ultimate strength experimental study was conducted.
This was required because the validation referred earlier was made with models made of mild steel
or steel with much lower strength than the HTS 690 and they are very limited in number. The use of
steel of much different strength will induce collapse at different levels of plate and columns slen-
derness and this calls for new experimental results, covering the appropriate range of the governing
parameters of the plating.
For these reasons a limited series of experiments were conducted with moderate size specimens
made of HTS 690, as reported here, which should provide a good basis to assess the ultimate strength of
various hull configurations and to compare them with predictions of the existing computational tool [2].
2. Main parameters of the structural design
The main parameters affecting the structural design of ship hulls subjected to bending moment are
the plate and column slenderness, because they affect directly the effectiveness of the panels under
compression. These parameters depend directly from the geometry of the structural elements and
from the material properties, being defined as:
rffiffiffiffiffi
b so
Plate slenderness; b ¼
t E
rffiffiffiffiffi
a so
Column slenderness; l ¼
r E
The geometric characteristics of interest are the width (b) and the span (a) of the structural elements,
as well as their thickness (t) and the radii of gyration (r) of the cross section of the stiffener with an
appropriate associated plate. Other geometric characteristics may affect the behaviour of the stiffener
in special cases. This may occur when the stiffener is very weak or it has low torsional rigidity,
promoting a different mode of collapse known in the literature as tripping.
The material properties of interest are the yield stress (so) and the modulus of elasticity (E). The
shear modulus of elasticity (G) has some influence on the tripping stress of the stiffener. Also the nature
of the stress–strain curve of different steels may affect the elasto-plastic behaviour of the structural
elements under compression, especially concerning on having or not having a constant yielding stress.
The use of HTS 690 instead of other normal steels (NS, H32, etc.) for the fabrication of the ship’s hull
has three different effects on the ultimate bending moment:
– an increase of the global strength proportional to the ratio between the yield strength corre-
sponding to the types of steel under comparison, for the same geometry;
– a reduction on the effectiveness of the structural elements, plates and stiffened plates, due to an
increase of their slenderness. For example, the slenderness of a plate made of HTS 690 is higher by
1.7 than the one of normal mild steel (MS), as there are differences in the yield strength but the
modulus of elasticity is the same;
– a reduction of the thickness of the scantlings due to a greater yield stress of HTS 690 in comparison
to others steels, which leads to a further increase on the slenderness, reducing the effectiveness of
those structural elements.
772 J.M. Gordo, C. Guedes Soares / Marine Structures 22 (2009) 770–790
In view of these points the efficiency of HTS 690 should be considered and the applicability of
current methods to this type of structures must be evaluated based on experimental results.
In order to provide data for those comparisons a plan of experiments was developed for box girders
subjected to pure bending moments. These box girders may reproduce in a simple manner the
behaviour of the ship’s structure under bending, allowing the identification of the differences of using
HTS 690, widening the range of validity of the method and covering the behaviour of panels of high
column slenderness.
3. Hull strength evaluation
There are several methods available to evaluate the ultimate moment in sagging or hogging that
a hull may sustain. The authors have been working on a method [2] that is able to predict the overall
behaviour of the hull under bending moment. This method predicts not only the ultimate bending
moment but also the pre- and post-collapse behaviour. It considers all the modes of collapse of the
structure and it also includes an algorithm to deal with residual stresses and corrosion.
This method and the software that has been developed to implement it, proved to give good
prediction for normal steel made ships when compared to the tests and results available in the liter-
ature [6]. The method developed has been shown to be appropriate to deal with situations of combined
vertical and horizontal ultimate strength [10] and also with damaged structures [11]. However it has
not been tested against results obtained for hulls made with HTS 690.
3.1. Assessment of the hull girder strength
The ability of the hull girder to sustain an applied bending moment may be understood as the
summation of individual contributions of each stiffened plate element that one may subdivide the
entire cross section between two frames. This can be expressed as:
Z X
M ¼ ðz zn ÞsðzÞdA ¼ ðzi zn Þsi ðzi ÞAi (1)
A
where the average stress s on the stiffened panel is a function of the average strain 3 and the last one is
dependent on the location zi of the element and on location of the neutral axis zn:
sðzi Þ ¼ f ð3i Þ and 3i ¼ gðzi ; zn Þ (2)

The main difficulty of this approach is to know the relation between the stress and the strain over
a large range of strains including pre-collapse, collapse and post-collapse. The importance of the last
region comes from the buckling of some elements before the ultimate bending moment is achieved.
The relation mentioned above depends on many parameters including residual stresses due to
welding, geometric imperfections, transverse support due to frames rigidity, etc. Other effects to be
considered are 3D effects or the lack of support on the middle of the large panels. Because the relation
between stress and strain is far from being linear the position of the neutral axis of the whole section is
changing with the loading and must be computed step by step.
The stress–strain curves may be obtained from a data base [12] or by approximate methods [13,14]
based on the empirical formulas for the ultimate strength of panels under axial loading and extended
in order to cover the whole stress–strain curve including the post-collapse region. Normally the design
methods used for that purpose are: Faulkner’s method [15], Perry–Robertson method [16,17] and the
critical stress for use as serviceable limit.
3.2. Influence of HTS on the strength of stiffened plates
The evaluation of the average column strength is based on the strength of the effective stiffened
plate, fu.ef.col, reduced by a factor that is the ratio between the effective area and the total area of the
element. If As is the stiffener area and be is the effective width of plate then one has:
Table 1
Characteristics of the Box Girders.
Box Length (mm) Span (mm) Breadth (mm) Depth (mm) Stiffeners
BG-H200 100 þ 4*200 þ 100 200 800 600 B20 4 mm
BG-H300 100 þ 3*300 þ 100 300 800 600 B20 4 mm
BG-H400 100 þ 3*400 þ 100 400 800 600 B20 4 mm

As þ be t
fu:col ¼ fu:ef :col (3)
As þ bt
This formula considers the same yield stress for the stiffener and the plate. The effective width of the
plate affects both terms of the second member and depends very much on the slenderness of the
plating, residual stresses (sr) and initial deformations (d).
be ¼ b,f ðb; sr ; dÞ (4)

So, when the yield stress increases, the thickness may be lowered and one has a magnified increase
on the plate slenderness, which has an inverse influence on the effective width reducing it very much.
Thus one has a reduction of weight and on the plate buckling stress (effective width) that may originate
design corrections to satisfy the buckling criteria.
The increase on the yield stress of the material generates a reduction on the ultimate strength of the
effective column since it increases directly the column slenderness, l. But one may have a further
reduction due to the change on the effective radii of gyration that results from the decrease of the
effective area of the associate plate of column and the shift in the neutral axis of the set stiffener/plate.
This last effect conducts normally to a reduction of effectiveness of the panel, but in some particular
cases, it may be neutral or positive depending on the geometry properties of the stiffener and the
associated plate.
In conclusion, the average column strength is very much affected by the reduction of both terms
when the yield stress of the material increases.
One may define the concept of effectiveness as equivalent to the normalised strength of the column
and then say that the increase on the yield stress generates a double reduction on the effectiveness of
the panel.
The global efficiency of a panel made of HTS may be evaluated by the ratio between the effec-
tiveness of the panel of HTS and the effectiveness of the mild steel panel.
fHTS
u:col
Efficiency of HTS ¼ (5)
fMS
u:col
Frames
Span = 200, 300 or 400
L50x20x6
P
Support Support
span
R
3500 1000, 1100, 1400 3500
Fig. 1. Scheme of setup.

Fig. 2. Supports and model during manufacturing at LISNAVE shipyard.
The important design parameters to be controlled when using HTS are: b, t, a, and tk. The control
shall be made b/t, a2so and by the evaluation of the tripping stress in order to avoid large reduction on
the global efficiency.
In a similar way, the concept of global efficiency of an HTS structure under bending, GBEHTS, may be
introduced just by replacing the effectiveness of the panel by the ultimate bending moment of the box
girder, as follows:
HTS
Mult HTS
Mult M M
GBEHTS ¼ MS
¼ , MS , HTS (6)
Mult MHTS MMS MMS
ult
The first two terms of the last member are the structural geometric efficiency of the structure made
of HTS and of mild steel (inverse) and the last term is the material efficiency equal to the ratio between
the yield stresses of both materials.
Fig. 3. Cross section of the box girders.

Table 2
Geometry of the test specimens.
Model Length (mm) b (mm) t (mm) h (mm) th (mm) Area (mm2) No. of frames
H200 1000 150 4 20 4 680 5
H300 1100 150 4 20 4 680 4
H400 1400 150 4 20 4 680 4
MHTS and MMS are the first yielding bending moment defined by the product of the yield stress and
the section modulus, Z, as follows:
Mo ¼ so ,Z (7)
One may understand the two first terms mentioned before as the structural efficiency under bending
because it includes both the geometric component, also present in the first yield moment, and the
material properties. Thus, for the same cross section, Eq. (6) can be rewritten as:
SBEHTS sHTS
GBEHTS ¼ , (8)
SBEMS sMS
The structural bending efficiency of the box girder (or the hull) measures the tendency of the
structure to behave plastically or to collapse due to buckling of the panels under compression. In the
first case SBE is greater than 1 and in the last case tends to be less than 1. However, one has always
SBEUTS < SBEMS because a structure made of high tensile steel is more prone to collapse by buckling
than one mild steel structure.
Fig. 4. Test on the box girder H200. Test setup.

Table 3
Geometrical parameters and predicted collapse strength of the specimens.
Model b/t b lnom lef fp fs Ap/At

H200 37.5 2.20 0.97 0.85 0.702 0.604 0.88
H300 37.5 2.20 1.45 1.28 0.702 0.437 0.88
H400 37.5 2.20 1.93 1.70 0.702 0.255 0.88
4. Plan of experiments
4.1. General information
The material of the specimens is HTS 690 and the principal material properties for structural
analysis are the yield stress and the Young modulus which are taken as been 690 MPa and 200 GPa,
respectively.
The three models are three or four bays span in length. The one with shorter distance between
frames (200 mm) has four bays span and the two others with 300 and 400 mm space framing have 3
bays.
The main nominal dimensions of the boxes and the bar stiffeners are presented in Table 1.The
transverse frames are ‘L’ stiffeners L50 28 6 made of mild steel but the longitudinal stiffeners are
continuous through the model in order to avoid misalignments.
4.2. Type of experiment
The tests consist of a four point bending of a beam like box girder. The beam is divided into three
parts: two symmetric supporting parts and, in the middle, one has the box girder model to be tested,
Fig. 1. Details of the manufacturing process at LISNAVE shipyard are presented in Fig. 2.
The box girder is subjected to pure bending moment, inducing tension on the bottom and
compression on the top of the box.
4.3. Geometric properties of the models
All models are made of 4 mm thick plate. The spacing between longitudinal stiffeners is 150 mm on
the top side (Fig. 3) which will be subjected to compression leading to width to thickness ratio of 37.5.
The range of nominal column slenderness covered is from 0.97 to 1.93. The plate slenderness b is
constant and equal to 2.2 but with a b/t of 37.5. Using normal steel this ratio corresponds to a b < 1
-500
-450
-400
-350
-300
Load (kN)
-250
-200
-150
-100
-50
0 10 20 30 40 50 60 70
0
Vertical Displacement (mm)
Fig. 5. Load displacement curve in one actuator (half total load) for H200 box.
-400
-350
-300
Load (kN) -250
-200
-150
-100
-50
0 5 10 15 20 25 30 35 40 45
0
which means that the normal steel plating should have a plastic collapse but due to the very high yield
stress of HTS 690, the plate induced failure should be due to buckling of the plating.
The main geometric properties are summarised in Table 2 and the test setup is shown in Fig. 4,
where one may see one of the supporting structures on the rigth and the box girder to be tested on the
left side with the loading device.
The plating slenderness is the same for all models (b ¼ 2.2) with a corresponding effectiveness fp of
0.702, evaluated according to Faulkner’s formula [15]. The nominal column slenderness varies from
0.97 to 1.93, which corresponds to a variation of the average ultimate strength, fs [13], from 0.604 to
0.255, Table 3. The effective column slenderness, lef, which uses the effective plate width (fp$b) instead
of the nominal width of the plate, is lower than the nominal column slenderness, lnom, due to the very
high ratio between the plating area Ap and the total area the plating and stiffener At.
5. Experimental results
The experiments started using several cycles of loading in order to obtain some stress relief on the
panel in tension. The evaluation of the energy absorption in this region allows the indirect estimation
of the residual stresses level due to welding [7]. However, as may be seen from the graphical results of
the tests, the different curves are very close to each other in the different cycles. This is due to the
existence of only one stiffener in the panel under tension and thus the residual stress relief refers only
to the welding of this stiffener.
-350
-300
-250
Load (kN)
-200
-150
-100
-50
0 5 10 15 20 25 30 35 40 45
0
3.5
3.0
2.5
Efficiency
2.0 Material Efficiency
Structural Efficiency
1.5 Global Efficiency
Column effectiveness
1.0
0.5
0.0
0.50 1.00 1.50 2.00 2.50
Nominal column slenderness
Fig. 8. Dependence of the structural and global efficiency on the nominal column slenderness.
The initial imperfections were not measured. However, the panels were very flat, without any
visible local deformation. This ‘flatness’ was consequence of a very careful welding procedure
according the specifications of the steel manufacturer.
5.1. Load vertical displacement relationship
The load displacement curves of the three box girders are shown in Figs. 5–13. The box H200 shows
a sharp discharge immediately after collapsing at approximately 40 mm of absolute vertical
displacement (Fig. 5). Before the collapse and after the maximum point of loading of the last cycle the
slope of the curve reduces, which may mean that the plasticity is spreading in some parts of the box
where residual stresses are higher.
The maximum load achieved in each actuator was 459 kN at a global displacement of 39.5 mm. This
means that the total vertical load supported by the box was 918 kN.
The box H300 collapsed on an earlier stage, at 33.6 mm of vertical displacement and a load of
380 kN (Fig. 6) in each actuator or 760 kN in total. The behaviour of the H300 model is very similar to
the H200 box but the curvature at the maximum load is lower than before with an initial stage of
smooth discharge followed by a sudden drop of the load carrying capacity with a full development of
the collapse deformations. The drop on the load is of the same magnitude (w50 kN) but after that, the
load for the H200 continues to fall quickly while on the H300 box it seems to be more stabilized.
The box H400 shows an almost linear path until 288 kN of load on each actuator at 27.2 mm of
displacement and the top panel buckled at this stage with a small decrease on the loading (4 kN).
1800
Bending Moment (kN.m)
1600
1400
1200
1000
800
600
400
200
0
0 0.002 0.004 0.006 0.008 0.01 0.012
Curvature (rad/m)
Fig. 9. Bending moment average curvature relationship for box H200.

1800
1600

1400
1200
1000
800
600
400
200
0
0 0.002 0.004 0.006 0.008 0.01 0.012 0.014
Curvature (rad/m)
Fig. 10. Bending moment curvature relationship at opposite sides of box H200.
After that point it continues to be able to sustain more load but with a much smaller slope that
becomes virtually flat around the maximum loading point achieving a maximum load of 308 kN
(617 kN in total) at 36 mm of vertical displacement at the loading point (Fig. 7). The unloading was
smoother than the others because the structure was already very deformed at that stage.
5.2. Moment curvature relationship
The relationship seems to change linearly with the column slenderness in the range of variation.
However the column effectiveness’ reduction rate is higher than the decrease of the structural effi-
ciency of the box girder in bending. This means that 3D structures subjected to bending are less
sensitive in terms of strength to the increase on the column slenderness parameter than the strength of
panels that are under compression
Table 4 presents the estimated yield moment and compares it with the results of the tests of the box
girders. In the last two lines a measure of the efficiency of the structure due to structural instability
(called here structural efficiency) is calculated and the last line presents the global efficiency of the use
of HTS 690 as compared to mild steel of 240 MPa yield stress. The global efficiency varies from 1.72 to
2.56 assuming that the middle term of Eq. (6) is one (SBEMS ¼ 1). In fact, this term decreases with the
increase in the space framing and thus the real global efficiency, as defined in equation (6), is higher
than the one presented in the table and the difference to presented value is higher as the frame spacing
increases.
This variation is mainly due to the increase on the span between frames since the cross section
characteristics remains unchangeable. Fig. 8 shows the dependence of the ultimate moment in terms of
the efficiency of the box from the column slenderness.
1400
Bending Moment (KN.m)
1200
1000
800
600
400
200
0
0 0.002 0.004 0.006 0.008 0.01 0.012 0.014 0.016 0.018 0.02
Curvature (rad/m)
1400

1200
1000
800
600
400
200
0
0 0.005 0.01 0.015 0.02 0.025
Curvature (rad/m)
The relationships seem to change linearly with the column slenderness in the range of variation.
However the column effectiveness’ reduction rate is higher than the decrease of the structural effi-
ciency of the box girder in bending. This means that 3D structures subjected to bending are less
sensitive in terms of strength to the increase on the column slenderness parameter than the strength of
panels that are under compression.
Fig. 13. H300: general deformations at collapse load.

Table 4
Comparison prediction and test results. Efficiency of HTS 690.
Box Girder identification H200 H300 H400

HTS yield bending moment (kNm) – MHTS 1711
HTS ultimate bending moment (kNm) – Mult 1526 1269 1026
MS yield bending moment (kNm) – MMS 595
sHTS/sMS and MHTS/MMS (yield) 2.875
Structural efficiency: Mult/MHTS 0.892 0.742 0.600
Global efficiency of HTS: Mult/MMS 2.56 2.13 1.72
Fig. 9 presents the relationship between the applied bending moment and the average curvature
measured between the extremes of the box girder under testing. The average curvature is the mean
value of two measurements made on each side of the box. Individual relationship may be observed in
Fig. 10. The differences between the two graphics indicate a transverse rotation of the box during the
loading path.
One has to note that at collapse the two curvatures are identical indicating that the box girder is in
the upright position. The vertical lines result from exceeding the measuring limits of the transducers.
Figs. 11 and 12 show the relationship for the H300 model. The sudden discharges of load at collapse
mentioned before during the analysis of the load vertical displacement curves are not so evident on the
moment curvature curves. The main reason is that the local curvature at collapse varies very much
along the model due to the local nature of the collapse, as can be observed in Fig. 13.
Figs. 14 and 15 refer to the box girder H400 which is the slender one. The relationship between the
bending moment and the curvature is almost linear until the applied moment reaches 956 kNm at
a curvature of 0.00743 rad/m. At this point the top panel began to buckle developing large out of plane
deformations. Consequently the structural tangent modulus reduced very much but continued to have
positive values until the ultimate bending moment (1026 kNm) is achieved at a curvature of
0.01055 rad/m.
One has to note that the collapse was very smooth and the applied bending moment was almost
constant in a large range of curvature at this stage. The bending moment is greater than 1024 kNm in
the range of curvatures from 0.00987 to 0.01074 rad/m. But the out of plane deformations at this stage
increased largely.
The structural modulus during the discharge has an intermediate value between the pre buckled
(quasi linear elastic range) and the post-buckled stages.
1200
1000
800
600
400
200
0
0 0.002 0.004 0.006 0.008 0.01 0.012 0.014
Curvature (rad/m)
1200

1000
800
600
400
200
0
0 0.002 0.004 0.006 0.008 0.01 0.012 0.014 0.016
Curvature (rad/m)
5.3. Modes of collapse
The modes of collapse obtained for the three box girders present three different configurations
corresponding to different histories.
The H200 box had a sudden noisy collapse with a great discharge of load and large deformations
due to the formation of plastic hinges at the middle span of the stiffeners under large compressive
loading. Only the two central bays were affected: one with deformations toward the stiffener and the
other towards the plating as one may see in Figs. 16 and 17. The outer bays remained virtually flat after
discharge of load, meaning that there were not much plastic deformations under the collapse load.
The collapse process was initiated on the second bay from the left (Fig. 16) by failure of the stiff-
eners, inducing the collapse of the adjacent panel bay by plating induced failure (Fig. 18) and the
occurrence of large permanent out of plane deformations on the side panels (Fig. 19).
It has to be noted that plane sections do not remain plane during the plastic stage and one has very
large deformations and rotations on the frames involved in the collapse mechanism. These axial and
Fig. 16. H200: general deformations after the collapse load.

Fig. 17. H200: overall view of deformations after collapse.
rotational deformations are more intense in the middle of frames, both in the top and side panels, than
in the connections.
The formation of the plastic hinges in all stiffeners (Fig. 18) means that they were loaded at the
stresses close to the yield stress and after the local discharge of loading associated with formation of
the first plastic hinge the load was transferred to the nearest stiffeners promoting the simultaneous
collapse of the panel. That is a typical column failure mode expected for short span stiffened panels,
Fig. 19.
The failure of the compressed panel propagates to the side structure leading to the failure of its
plating. There are the formation of two plastic hinges at 45 relatively to the frame system, that join at
Fig. 18. H200: column collapse between frames.

Fig. 19. H200: Stiffener collapse on top panel and side deformation.
the middle of the side plating, continuing with a plastic line parallel to the transverse frame, as may be
observed in Fig. 20.
The H300 box showed a smoother collapse but presented a similar mode of deformations after
collapse. However the plastic hinges did not originate a neck in most of the stiffeners. That should be
the reason for the absence of noise during the collapse. So there was a plastic flow at the middle span of
the stiffeners without any out of plane deformations for most of them. Fig. 21 presents the residual
deformations after the test.
The box H400 is also a 3 bay model (Fig. 22) like the box H300 but it has presented a more complex
history of the out of plane deformations on the top panel (compression). Visible semi sinusoidal
Fig. 20. H200: inside view of collapse after cutting.

Fig. 21. H300: Buckling of the compression panel and residual deformations on the side panel.
deformations between frames might be observed during the loading path, inwards in the end spans
and outwards in the middle span (Fig. 23). This deformed mode generates high level of stress on the
extreme of the stiffeners but of different nature depending on the location, compression on the top of
the stiffeners between the lateral frames and tension on the top of the stiffeners at the middle bay. It
was a more marked and global collapse due to the large span between frames. The mode of collapse
shall be classified like the others as column induced failure.
5.4. Residual stresses
Gordo and Guedes Soares [7] presented a structural tangent modulus method to estimate indirectly
the level of residual stresses in experiments where one has an intermediate discharge of loading.
According to the method, the cross section area (Ap) of the residual stresses tensile strips belonging to
the panels under tension may be estimated in the points where one has a discharge in the previous
Fig. 22. H400: Overall view of the test before collapse.

Fig. 23. H400: residual deformations after collapse on top and side panels.
cycle of loading. The mechanical reason is the variation on the structural tangent modulus DEI at that
point of loading, when passing from loads already reached in previous cycles to loads above the last
maximum, leading to new plastic deformation on the tensile strips and consequent decrease of the
effective structural tangent modulus. That cross section area is given approximately by:
DEI
Ap ¼ (9)
E,z2p
where zp is the distance between the neutral axis and the panel under tension.
For the H200 box the calculated variation of the structural tangent modulus is 10 and 30 MN m2 at
the maximum load for the first and second cycle, respectively. This corresponds to having a tensile strip
area of 556 mm2 and a tensile strip width of 14 mm for the first case.
Considering that the box has the same welding in all connections and the spacing between stiff-
eners in the panel under compression is 150 mm, the residual stress level is 23% of the yield stress in
that panel.
H200 - Structural Tangent Modulus

200000
180000
160000
140000
EI (KN.m2)
120000 Load 1
Load 2
100000 Load 3
80000
60000
40000
20000
0
0 200 400 600 800 1000 1200 1400 1600 1800
Fig. 24. Structural modulus of H200 at different cycles of loading.


200000
Load 1
180000
Load 2
160000 Load 3
Load 4
140000
Load 5
EI (KN.m2)
120000
100000
80000
60000
40000
20000
0
0 200 400 600 800 1000 1200
The second value, related to the variation of the structural modulus on the maximum of the second
cycle (30 MN m2), is evaluated at a very high load and incorporates already some decrease on the
rigidity of the panel under compression due to higher out of plane deformations. Thus the method is
not applicable at this level of loading.
Fig. 24 presents the structural tangent modulus during each loading cycle. One should note that the
structural modulus is approximately constant until the maximum load of the previous cycle is reached
and decrease after that point.
For the H400 box girder, the residual stress level computed is 14% of the yield stress at the
maximum of 3rd cycle of loading. At initial cycles it was not detected any change in the structural
modulus, which means that there was a partial relief of residual stresses during transportation or
mounting process as it may be observed in Fig. 25. The scatter of structural modulus is relatively higher
than the one observed for the H200 box.

200000
180000
160000
140000
EI (KN.m2)
120000 Load 1
Load 2
100000 Load 3
80000
60000
40000
20000
0
0 200 400 600 800 1000 1200 1400

1400
H200
1300 1307
H300
1253 H400
1200 1205
Ultimate Moment (kNm)
1166
1135
1115
1100
1077
1021
1000
926
900
883
861
837
800 794
738 724
700
600
0 1 2 3 4 5 6 7
Fig. 27. Influence of residual stresses on the ultimate bending moment according to the numerical method.
Also one may observed that the structural modulus reduces very much to 1/3 of the previous value
after the premature failure at 940 KN m.
A similar behaviour was found for the box H300, without much difference on the cycles at early
stages of loading and measurable difference in the structural modulus after the second cycle, Fig. 26.
The difference at 800 KN m of the bending moment indicates a level of residual stresses similar to
H400. However this level of loading is very high and thus, the computed variation in structural
modulus may include already some decrease of rigidity in the panel under compression.
6. Comparison with numerical method
The numerical method [6] implemented in HULLCOL software was used to predict the ultimate
bending moment of the box girders and to compare with the experimental results. The method allows
incorporating the effect of the residual stresses in the load shortening curves of the stiffened plate
elements [13] for compression and tension. The influence of the residual stresses on the tensile
behavior was implemented in the code because it may influence very much the moment curvature
curve in heavy welded structures by reducing the initial stiffness and leading to higher values of the
curvature for the same bending moment. For the usual representation of the residual stresses
composed of two tensile zones with width ht each at yield stress and a compressed region in the
middle of the plate with residual stress sr, the initial slope of the average stress–strain curve of the
plate is given by a tangent modulus of:
Table 5
Comparison between tests and numerical results.
Model H200 H300 H400

Ultimate bending moment (kNm) – test, Mt 1526 1269 1026
Ultimate bending moment (kNm) – HULLCOL, Mc 1307 1115 861
Ratio – Mc/Mt 0.86 0.88 0.84
SBEHTS – test 0.89 0.74 0.60
SBEHTS – HULLCOL 0.76 0.65 0.50
b 2ht
Et ¼ E (10)
b
The full squash stress is achieved at elongation 3r higher than the material yield strain, 3o, and it can
be expressed by the formula:
b
3r ¼ 3o (11)
b 2ht
This behaviour softens the structure in tension and, for 3D structures subject to bending moment, it
reduces the structural modulus in the initial stages of loading, leading to higher collapse curvature than
the one achieved in a residual stress free structure.
Thus the impact of the residual stresses on the ultimate bending moment of the structure may
predict and the changes on the moment curvature relationship may be estimated.
However in these tests on box girders made of very high tensile, the influence of the residual
stresses is not expected to be high since one has very few welding in the tensile side and the buckling
stress of the panel under compression is very low in comparison to the yield stress, which means that
the panel tends to buckle before the middle part of plate elements may reach stresses close to the yield
stress.
The reduction of the ultimate bending moment with the increase of the residual stresses parameter
h for the three box girders predicted by HULLCOL (version 3.1) is presented in Fig. 27. Notice that the
tensile width parameter h equal to 1 corresponds to a compressive residual stress of 5.6% of the yield
stress and h ¼ 3 correspond to 19%, using the expression:
2ht
sr ¼ so (12)
b 2ht
The reduction on the ultimate bending moment due to the increase of residual stresses is small
compared to the reduction due to the increase of the frame spacing.
In Table 5 it is compared the prediction due to the approximate numerical method and the test
results. The approximate method shows to be conservative by 15% when predicting the ultimate
bending moment. One should note that the ratio between calculated and test result is very consistent
through the different frame spacing with a mean value of 0.86.
The main reason for these conservative values arises from the use of effective width in the formulas
for the prediction of column strength, Eq. (3). The collapse of these very slender panels is predomi-
nantly induced by stiffener’s failure at low average stress [18,19]. In that condition, the associated
plating shall be considered totally effective. The Faulkner method, as implemented in HULLCOL, is more
adequate to predict the ultimate strength of panels induced by plate failure or hybrid failure. This is an
extreme situation of very low stiffener to plate area ratio where Faulkner’s method seems to become
conservative.
7. Conclusions
The efficiency of the HTS 690 is very good when the steel is applied on box girders subjected to
bending moments.
The structural efficiency decreases as the span between frames increases. The increase of span
increases the column slenderness. The structural efficiency decreases linearly with column slenderness
in the range considered for this parameter.
The global efficiency, which includes structural and material efficiency, is higher than that expected
from the effectiveness of the plate and the column of the panel under compression affected by the
material efficiency.
The modes of collapse obtained are all classified as column mode of failure, as expected. However
the final deformed shape presents different configurations resulting from mainly the different column
slenderness originating different load-out of plane deformations histories on the panels under
compression.
The plate induced failure could not be achieved because of incompatibilities between plate thick-
ness available, setup of the test and maximum geometry of the models.
The range of the column slenderness is very high due to the high ratio between the plate and the
total area of the representative stiffened plate (Ap/At ¼ 0.88). This means that for actual design practice
one may obtain a global efficiency of the HTS 690 on the order of 2.5 taking as basis the normal mild
steel (MS) structure.
Acknowledgements
The work presented in this paper was done under the project ‘High Tensile Steel 690 in Fast Ship
Structures (FASDHTS)’, CEC BRITE/EURAM Programme (2000–2004), which was partially financed by
the European Union through the contract no. GRD1-1999-10558. The partners were: TNO Building and
Construction, Centre for Mechanical Engineering (NL), AF Industriteknik AB (S), AG der Dillinger
Hüttenwerke (D), Alstom Chantiers de l’Atlantique (F), Bureau Veritas (F), Chalmers University of
Techonology AB (S), Flensburger Schiffbau mbH&Co. KG (D), Germanischer Lloyd AG (D), Instituto
Superior Técnico (P), Lisnave Estaleiros Navais (P), Royal Schelde group BV (NL), Technische Universität
Hamburg-Harburg (D), and Van der Giessen-De Noord shipbuilding division BV (NL).
References
[1] Yao T, Astrup O.C, Caridis P., Chen Y.N., Cho S.-R., Dow R.S., et-al. Ultimate hull girder strength. ISSC special task committee
VI.2 In: Proceedings of the 14th international ship and offshore structures congress (ISSC), Japan, 2000. 3:321-91.
[2] Gordo JM, Guedes Soares C, Faulkner D. Approximate assessment of the ultimate longitudinal strength of the hull girder.
Journal of Ship Research 1996;40(1):60–9.
[3] Rutherford SE, Caldwell JB. Ultimate longitudinal strength of ships: a case study. Transaction of SNAME 1990;98:441–71.
[4] Dow R, Hugill R, Clark J, Smith C. Evaluation of ultimate ship hull strength. In: Extreme loads response symposium.
SNAME; 1981. p. 133–47.
[5] Faulkner JA, Clarke JD, Smith CS, Faulkner D. The loss of HMS cobra – a reassessment. Transactions of RINA 1984;127:
125–51.
[6] Gordo JM, Guedes Soares C. Approximate method to evaluate the hull girder collapse strength. Marine Structures 1996;
9(1):449–70.
[7] Gordo JM, Guedes Soares C. Experimental evaluation of the ultimate bending moment of a box girder. Marine Systems and
Ocean Technology 2004;1(1):33–46.
[8] Nishihara S. Ultimate longitudinal strength of mid-ship cross section. Naval Architecture and Ocean Engineering 1984;22:
200–14.
[9] Janssen GTM. Fatigue based design rules for the application of high tensile steel in ships. In: Proceedings of the seventh
international marine design conference; 2000. p. 317–28. Korea.
[10] Gordo JM, Guedes Soares C. Interaction equation for the collapse of tankers and containerships under combined bending
moments. Journal of Ship Research 1997;41(3):230–340.
[11] Gordo JM, Guedes Soares C. Residual strength of damaged ship hulls. In: Proceedings of IMAM; 2000. p. 79–86. Napoli,
Italy.
[12] Smith CS. Influence of local compressive failure on ultimate longitudinal strength of a ship’s hull. In: Proceedings of the
third international symposium on practical design in shipbuilding (PRADS); 1977. p. 73–9. Tokyo.
[13] Gordo JM, Guedes Soares C. Approximate load shortening curves for stiffened plates under uniaxial compression. In:
Faulkner D, Cowling MJ, Incecik A, Das. PK, editors. Integrity of offshore structures – 5. Glasgow, Warley, U.K.: EMAS; 1993.
p. 189–211.
[14] Yao T, Nikolov PI. Progressive collapse analysis of a ship’s hull under longitudinal bending. Journal of the Society of Naval
Architects of Japan 1991;170:449–61.
[15] Faulkner D. A review of effective plating for use in the analysis of stiffened plating in bending and compression. Journal of
Ship Research 1975;19:1–17.
[16] Ayrton WE, Perry J. On struts. The Engineer 1886;62:463–513.
[17] Robertson A. The strength of struts. In: Proceedings of the institution of civil engineers, 25; 1925.
[18] Gordo JM, Guedes Soares C. Compressive tests on short continuous panels. Marine Structures 2008;21(2–3):113–326.
[19] Gordo, J.M. and Guedes Soares C. Compressive tests on continuous long stiffened panels. In: Proceedings of the 27th
international conference on offshore mechanics and arctic engineering (OMAE 2008), June 15–20, 2008, Estoril, Portugal,
paper 57873.

Tests On Ultimate Strength of Hull Box Girders Made of High Tensile Steel

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Tests On Ultimate Strength of Hull Box Girders Made of High Tensile Steel

Uploaded by

Copyright:

Available Formats

Marine Structures 22 (2009) 770–790

Contents lists available at ScienceDirect

Tests on ultimate strength of hull box girders made

* Corresponding author. Tel.: þ351 218417607; fax.: þ351 218474015.

2. Main parameters of the structural design

3. Hull strength evaluation

3.1. Assessment of the hull girder strength

sðzi Þ ¼ f ð3i Þ and 3i ¼ gðzi ; zn Þ (2)

3.2. Inﬂuence of HTS on the strength of stiffened plates

be ¼ b,f ðb; sr ; dÞ (4)

3500 1000, 1100, 1400 3500

Fig. 1. Scheme of setup.

Fig. 2. Supports and model during manufacturing at LISNAVE shipyard.

Fig. 3. Cross section of the box girders.

Fig. 4. Test on the box girder H200. Test setup.

Model b/t b lnom lef fp fs Ap/At

4.1. General information

4.2. Type of experiment

4.3. Geometric properties of the models

Load (kN) -250

5.1. Load vertical displacement relationship

Fig. 9. Bending moment average curvature relationship for box H200.

Bending Moment (kN.m)

5.2. Moment curvature relationship

Bending Moment (kN.m)

Fig. 13. H300: general deformations at collapse load.

Box Girder identiﬁcation H200 H300 H400

Bending Moment (KN.m)

5.3. Modes of collapse

Fig. 16. H200: general deformations after the collapse load.

Fig. 17. H200: overall view of deformations after collapse.

Fig. 18. H200: column collapse between frames.

Fig. 20. H200: inside view of collapse after cutting.

5.4. Residual stresses

Fig. 22. H400: Overall view of the test before collapse.

H200 - Structural Tangent Modulus

Fig. 24. Structural modulus of H200 at different cycles of loading.

H400 - Structural Tangent Modulus

Fig. 25. Structural modulus of H400 at different cycles of loading.

H300 - Structural Tangent Modulus

Fig. 26. Structural modulus of H300 at different cycles of loading.

6. Comparison with numerical method

Model H200 H300 H400

You might also like