Marine Structures 21 (2008) 327352

Contents lists available at ScienceDirect

Marine Structures
journal homepage: www.elsevier.com/locate/
marstruc

Ultimate strength analysis of a bulk carrier hull girder

under alternate hold loading condition A case study
Part 1: Nonlinear nite element modelling and ultimate
hull girder capacity
Hadi K.K. Amlashi a, *, Torgeir Moan b
a

Department of Marine Technology, Norwegian University of Science and Technology (NTNU) N-7491, Trondheim, Norway
Centre for Ships and Ocean Structures and Department of Marine Technology, Norwegian University of Science and Technology
(NTNU), N-7491, Trondheim, Norway
b

a r t i c l e i n f o

a b s t r a c t

Article history:
Received 4 September 2007
Received in revised form 18 December 2007
Accepted 19 December 2007

This is the rst of two companion papers dealing with nonlinear

nite element modelling and analysis of the ultimate strength of
a bulk carrier hull girder under alternate hold loading (AHL)
condition. The purpose is to contribute to establishing rational
ultimate longitudinal strength criteria for the hull girder under
combined loading. The focus is on the hogging condition. An
important issue is the signicant double bottom bending in empty
holds in AHL due to combined global hull girder bending moment
and local loads. The local loads may substantially reduce the
strength of the hull girder. Different AHL conditions, i.e. fully
loaded cargo and (partially) heavy cargo are considered. A critical
review of external and internal design pressures for different AHL
conditions is accomplished using both CSR-BC rules and DNV rules.
A methodology for nonlinear nite element modelling of hold
tanks of a bulk carrier under AHL is presented by use of ABAQUS.
A mesh convergence study is carried out in order to nd the appropriate mesh for the model. The implication of using different design pressures on the hull girder strength is assessed. The FE
results can be used as a basis for establishing simplied methods
applicable to practical design of ship hulls under combined loadings. This issue is discussed in the companion paper.
2008 Elsevier Ltd. All rights reserved.

Keywords:
Bulk carrier
Alternate hold loading
Double bottom bending
Stress redistribution
Nonlinear FE analysis
Ultimate hull girder strength

* Corresponding author. Tel.: 47 73 595565; fax: 47 73 595528.

E-mail address: hadi.k.amlashi@ntnu.no (H.K.K. Amlashi).
0951-8339/$ see front matter 2008 Elsevier Ltd. All rights reserved.
doi:10.1016/j.marstruc.2007.12.006

328

H.K.K. Amlashi, T. Moan / Marine Structures 21 (2008) 327352

1. Introduction
In the design of bulk carriers the alternate hold loading (AHL) condition is usually assumed to be the
dimensioning loading condition for the double bottom structure when the global hull girder is in hogging, see Fig. 1. In the AHL condition the middle cargo hold is empty and alternate holds are loaded with
maximum cargo deadweight at maximum draught. Lateral loading on the bottom, inner bottom and
lower sides of holds acts together with (typically) an overall hogging bending moment as shown in
Fig. 1. As a result, the global strength of the hull girder is reduced due to signicant double bottom
bending at the middle of the empty hold.
The ultimate hull girder capacity of any ship type can be estimated by a hierarchy of methods, such
as: (a) Full nonlinear FE analysis (rst attempted by Chen et al. [1], Kutt et al. [2] and Valsgrd et al. [3]);
(b) Simplied FE methods such as ISUM (initially proposed by Ueda and Rashed [4]); (c) Smith-type
simplied methods (proposed by Smith [5]) based on Naviers hypothesis and average stressaverage
strain relationship for individual panels; and (d) Simple closed-form formulations based on further
simplications as compared to Smith-type methods, e.g. without considering the progressive development of the collapse and load redistribution (initially suggested by Caldwell [6]). Signicant efforts
have been devoted to methods (b), (c) and (d) that include works done by Ueda et al. [7] for ISUM, Yao
[8,9] for Smith-type method and Mansour et al. [10] for Simple closed-form formulations. However, fewer studies are reported using method (a), because it requires huge computational resources.
An issue is that the present conventional Smith-type simplied methods cannot account for the effect of double bottom bending in bulk carriers and therefore seem to be on the non-conservative side as
reported, e.g. in Moan et al. [11]. On the other hand, a simplied linear approach could be quite conservative because it neglects the effect of stress redistribution, or reserve capacity beyond the collapse
of typical bottom panels [11].
Most of the nonlinear FE analyses published so far seem to be dealing with ships under pure bending, e.g. [13,12]. This type of loading allows for modelling and analysis of a single cargo tank or even
only a few frame spacings of a tank. Proper treatment of the double bottom bending in the AHL conditions requires that a larger part of the ship including both loaded and empty hold tanks be included
in the FE model.
It seems that, so far, only the study by stvold et al. [13] has been carried out to investigate the effect
of double bottom bending on the hull girder strength. The vessel was a Panamax bulk carrier with an
optimized bottom design for combined loading according to DNV rules. Therefore, further work using
different (local and global) load combinations by nonlinear nite element analysis is necessary. In particular, such an analysis is required when the new rule for bulk carriers, e.g. CSR-BC [14] is applied. It is
also important to appropriately deal with different AHL conditions.
The overall objectives of this paper and the companion paper [15] are as follows:
(a) To describe a methodology for the nite element modelling of (1/2 1 1/2) hold tanks of a bulk
carrier under AHL condition;
(b) To calculate loads for a bulk carrier under AHL condition according to different rules, e.g. CSR-BC
[14] and DNV [16] and to assess the ultimate strength of the hull girder under such loadings;
(c) To study the inuence of using different (local and global) load combinations on the ultimate hull
girder strength;

MCom.
MGlobal

MGlobal

MLocal

Tension

MGlobal

Compression

Fig. 1. Global and local load distribution of a bulk carrier in alternate hold loading condition.

H.K.K. Amlashi, T. Moan / Marine Structures 21 (2008) 327352

329

(d) To study the effect of double bottom bending in the empty hold tank on the global stress redistribution in that area; and
(e) To provide a basis for establishing simplied methods applicable to practical design of ship hulls
under combined loadings.
There is a substantial amount of technical materials that has to be covered in explaining the behaviour of the hull girder under AHL condition. Hence, this paper only addresses objectives (a)(c). A more
detailed assessment of the ultimate behaviour in the double bottom area is explored in the companion
paper which addresses the overall objectives (d) and (e).
2. Vessel particulars
The vessel selected as a base case for the present nite element analysis is a Capesize Single Side
(SS) bulk carrier. The hull cross-section of the vessel is shown in Fig. 2. It has been benchmarked in
the ISSC Committee VI.2 report [17] against a number of simplied methods. The main particulars of
the vessel are summarized in Table 1. The ship is longitudinally stiffened except for the side shell between the hopper tank and the wing tank which are transversely stiffened. The transverse frame spacing is 870 mm while the longitudinal stiffener spacing is 880 mm.
Due to lack of details, some assumptions are made based on the ISSC-hull cross-section geometry. It
is supposed that the bulk carrier has nine hold tanks. The number 1 hold tank is in the fore of the ship.
Based on this assumption, the hold No. 6 will be approximately located in the middle of the ship presuming that the length of the hold tank is 26,100 mm. The girders in the double bottom area are
denoted GR2640, GR5280, GR7920, GR10560, GR14080 and GR18480 for the Side Girders based on
their respective distances in millimeters from the centre girder, see Fig. 2.

Y(MPa)

No.

Dimensions

Type

390x27

Flat-bar

392

333x9+100x16

Tee-bar

352.8

283x9+100x14

Tee-bar

352.8

283x9+100x18

Tee-bar

352.8

333x9+100x17

Tee-bar

352.8

283x9+100x16

Tee-bar

352.8

180x32.5x9.5

Bulb-bar

235.2

283x9+100x17

Tee-bar

352.8

333x9+100x18

Tee-bar

352.8

10

333x9+100x19

Tee-bar

352.8

11

383x9+100x17

Tee-bar

352.8

12

383x10+100x18

Tee-bar

352.8

13

383x10+100x21

Tee-bar

352.8

14

300x27

Flat-bar

392

880
870
880

Fig. 2. Cross-sectional data for the ISSC-benchmarked bulk carrier [17].

330

H.K.K. Amlashi, T. Moan / Marine Structures 21 (2008) 327352

Table 1
Principal dimensions of the vessel
Length
(B.P.)

Length
(0.97 LLWL)

Breadth
(MLD)

Depth
(MLD)

Draft
(Design, MLD)

Draft
(Scantling, MLD)

285

281.3

50

26.7

19.808

19.83

3. Material models
Four kinds of steel are used in this vessel. The materials considered are summarized in Table 2. A
bilinear elasto-plastic material model with strain hardening is used for the nonlinear analyses, as
shown in Fig. 3. The strain hardening parameters are also given in Table 2.
The deck and upper 2.0 m part of the sides and wing tanks (including stiffeners) are constructed
from High Strength steel (HS-40) and the remaining parts of the structure are made of HS-36 and
HS-32, as shown in Fig. 2. Only bulb proles (Stiffener No. 7) have Mild Steel (MS) grade.
4. Finite element modelling aspects
The FE code used for the analysis is ABAQUS/Standard [18]. This is a widely used nite element code
for nonlinear structural analyses. The geometry drawings and meshing operations are all carried out
using ABAQUS/CAE. This is a consistent interface for creating, submitting, monitoring, and evaluating
results from ABAQUS/Standard simulations [19].
The primary concern in developing the nite element model is to generate a model that can give
best possible predictions of the ultimate strength of a bulk carrier under AHL condition with feasible
computation efforts. Two aspects are of importance in this connection. The rst is that the nite element mesh should be ne enough to capture accurately the behaviour induced by the combined local
and global loadings. The second aspect is that the boundary conditions must be such that the global
hull girder bending and local loading are correctly transmitted through the hull girder. Based on the
latter assumption, a reasonable model is chosen to extend from halfway hold No. 5 up to halfway
hold No. 7 lengthwise, i.e. (1/2 1 1/2) hold tanks. Symmetry about the centre-plane is also
considered.
The nite element modelling consists of choosing the appropriate element type and mesh size. Even
today, rapid increases in computer processing power and memory have not eliminated computational
cost and time constraints. This is due to the constant increase in the required mesh density to converge
to the most reliable solution. An increase in mesh density (ne mesh) through the model is theoretically possible but not in practice due to signicant efforts and computational costs. Therefore, a balance
between required accuracy and efforts is needed. A description of the modelling is discussed in the following sections.
4.1. Element type and mesh convergence
There are different kinds of shell elements available for modelling steel-plated ship structures. The
available shell elements are distinguished in reduced and full integration elements, either of triangular
or quadrilateral shape. The difference lies in the number of sampling points (over the area) used to perform the Gaussian or Simpson integration in order to construct the elements stiffness matrix. Reduced
Table 2
Material parameters for the material models used (Fig. 3)

MS
HS-32
HS-36
HS-40

Yield stress
(MPa)

Youngs modulus;
E (MPa)

Strain hardening
parameter; C (MPa)

Poisson
ratio

235
313.6
352.8
392

210,000
210,000
210,000
210,000

825
625
675
600

0.3
0.3
0.3
0.3

H.K.K. Amlashi, T. Moan / Marine Structures 21 (2008) 327352

331

Stress

C
Y

Strain

Fig. 3. Bilinear elasto-plastic material model.

integration four-node quadrilateral elements use one sampling point over the area whereas fully integrated elements use four sampling points. This makes the reduced integration elements computationally less expensive. However, when reduced integration elements are used, more elements may be
required to reproduce the deformation modes of a structural part. A major shear-locking problem
is associated with full integration elements in which transverse shear stresses are overestimated
that cause the element to reveal excessive stiffness. On the other hand, an hour-glass problem
may be associated with reduced integration elements. However, the general-purpose shell elements
that are integrated in most nite element codes provide accurate solutions in those circumstances
and are used for most thick and thin shell problems.
In cases involving elasto-plastic response, ve integration points are commonly used through the
thickness of the shell elements to provide adequate modelling of the progress of yielding through
the section. In problems involving signicant strain reversals (such as low-cycle fatigue studies),
more points would generally needed [18].
In general, mesh convergence tests can be conducted for particular cases of structures and element
types in order to nd out an optimal mesh size. Evidences seem to indicate that a proper shell model of
a single span stiffener and associated plate (simple stiffened plate) may require 1000 Degrees of Freedom (DoF) or more (see, e.g. [20]). A model of a compartment then may roughly involve 500,000 DoF
taking symmetry into account. For a (1/2 1 1/2) compartment model, this mesh density would imply more than 1,000,000 DoF. This poses a severe challenge for the FE nonlinear strength assessment of
ship hull girders.
As a result, mesh convergence analysis should determine the point where both computational costs
and accuracy can be justied. This should preferably be done on critical panels for instance the centre
panel in the bottom of the present vessel between centre girder and GR2640. The panel selected for the
convergence study is a long six-frame spacing panel shown in Fig. 3. The panel is 15,660 mm long, with
2640 mm width and plate thickness 18.5 mm. The stiffener characteristics are given in Fig. 2.
A convergence study is carried out for the critical panel using three mesh sizes. Both uniaxial loading and combined biaxial loading and lateral pressure are analyzed with three mesh sizes. The mesh
size is characterized by the number of elements in the plate between longitudinal stiffeners (see
Fig. 4). For instance, case 3-el implies three elements between stiffeners, nine between web frames,
three for the web of the stiffener and two for the ange. The total number of the elements for the
six-frame spacing panel is 864, 2160 and 4032, respectively, for 3-el, 5-el and 7-el.
An investigation of the effect of element type on the ultimate strength is carried out (see Fig. 5). The
element type S4 is a general shell element in ABAQUS which can be used for both thin and thick shells
as well as small and large strain applications. Element type S4R is suited for small-strain, large-rotation
analysis. Element type S4R5 uses 5 DoF but can only be used for small-strain applications. Fig. 5 shows
that very little difference is observed between elements S4, S4R and S4R5 for the combined loading.

332

H.K.K. Amlashi, T. Moan / Marine Structures 21 (2008) 327352

Fig. 4. Meshes considered in convergence study for the critical stiffened plate in the bottom: (a) 3-el; (b) 5-el; and (c) 7-el.

The element type S4 will be the element of choice for the modelling except a few situations (in linear
part) where the three-node triangular elements used to ensure their conformity with specic geometry constraints, such as web frames around the holes.
The ultimate strength for uniaxial and combined loading cases for the three mesh sizes is shown in
Fig. 6. It is seen that the ultimate strength of the panel under combined loading is 25% smaller than for
the uniaxial loading (see also Table 3). This is due to the high lateral pressure (0.24 MPa) and biaxial loading applied for this case. It is interesting to see that mesh convergence is faster for the uniaxial loading
than the combined loading (see Fig. 6 and Table 3). This is explained by the fact that high lateral pressure
and biaxial loading in the combined loading case increase the complexity of possible failure modes.
It is noted that the accuracy for the 5-el mesh is much better than for the 3-el case. The 5-el mesh is
used in the (1/2 1 1/2) hold tank model. It should be mentioned that the (1/2 1 1/2) hold tank
model is still quite CPU demanding with this mesh size.
4.2. FE model of the hull girder
Based on the above assessment, a combined (ne and coarse) mesh strategy is needed to keep the
number of DoF on a manageable level. Coarse meshes are used in areas far from the region of interest.
They are assumed to behave linearly elastic.
0.8
0.7
0.6
0.5
0.4
0.3
0.2
S4
S4R
S4R5

0.1
0

0.5

1.5

Fig. 5. Ultimate strength of the critical stiffened panel in the double bottom area under combined loading considering different element types; case 5-el.

H.K.K. Amlashi, T. Moan / Marine Structures 21 (2008) 327352

0.8

0.8

0.6

0.6

0.4

0.4
3-el
5-el
7-el

0.2
0

0.5

1.5

333

3-el
5-el
7-el

0.2

0.5

1.5

Fig. 6. Ultimate strength convergence (element type S4) of the critical stiffened panel in the double bottom area for (a) uniaxial
loading and (b) combined loading.

Therefore, the model is composed of two parts, one which accounts for linear behaviour and one
which accounts for nonlinear behaviour. The middle part of the empty hold tank, shown in Fig. 7, is
modelled with signicantly greater mesh density in most parts of the double bottom with full nonlinear capabilities to help studying the effect of inelastic behaviour of the bottom structure on the response of the hull girder. It should be noted that Fig. 8 only shows an outline view of the entire
model. Since the mesh is dense in nonlinear part, it is very difcult to show the entire model with
all element edges visible. The rest of the model is linear (Fig. 7). It includes adjacent holds, i.e. their
adjacent parts of the double bottom, side and deck of the middle hold as well as web frames.
The mesh density in the ne mesh region is illustrated in Fig. 8 and can be summarized as follows:
-

ve elements between longitudinal stiffeners;

15 elements between transverse web frames;
ve elements across the height of longitudinal stiffeners; and
two elements across the ange of longitudinal stiffeners.

It should be discussed whether or not two elements across the ange are sufcient with respect to
tripping failure of slender stiffeners. However, they seem to be adequate for the present vessel, as the
stiffeners in the double bottom area are stiff enough to prevent the tripping failure.
The remaining parts of the FE model, i.e. the linear part and the upper region of the nonlinear part
are based on a coarse mesh. This means that one element is used between stiffeners, three between
transverse web frames and one for the web and the ange of longitudinal stiffeners.
It should be emphasized that the FE model is developed following the cross-sectional data of the
ship provided in ISSC Committee report VI.2 which is not deemed sufcient for complete modelling
of the hold tanks. Due to lack of data, hypothesis corrugated bulkheads as well as lower (and upper)
stools are assumed. This might slightly affect the results obtained for a bulk carrier under pure bending
moment. Nevertheless, it is still possible to compare with the results obtained based on Smith-type
methods [17].

Table 3
Normalized ultimate strengths for the critical stiffened panel in the double bottom with three mesh sizes; element type S4 (see
Fig. 5)
Case study

Uniaxial loading

Combined loading

3-el
5-el
7-el

1.0
0.98
0.98

0.79
0.73
0.71

334
H.K.K. Amlashi, T. Moan / Marine Structures 21 (2008) 327352

Fig. 7. Extent of the nonlinear and linear parts in the FE model of (1/2 1 1/2) hold tanks.

H.K.K. Amlashi, T. Moan / Marine Structures 21 (2008) 327352

335

Fig. 8. A close-up view of the nonlinear part in double bottom of the empty hold tank.

Structural details such as manholes, brackets, tripping brackets and minor strengthening are not included in the model, where local plastications and stress concentrations may cause the analysis to stop
before nonlinearities occur at the areas of interest. The model consists of all the major parts, such as: double bottom, hopper side tank, side shell, side frame, top wing tank, deck and transverse bulkheads.
The total numbers of DoF in the nonlinear part is around 475,000, while the linear part has about
620,000 DoF.
5. Boundary conditions
To prescribe appropriate boundary conditions is a main challenge in modelling a part of a (ship)
structure by nite elements. For the particular model of the present Capesize bulk carrier, i.e.
(1/2 1 1/2) hold tank model, the following boundary conditions are introduced.
Since the cargo and sea pressure is assumed to be symmetric with reference to the centre-plane (see
Fig. 7) the symmetry condition can be applied at the centre-plane.
The major challenge is associated with the boundary conditions for the fore and aft end-sections.
These sections lie in the middle of holds No. 5 and 7, respectively (see Fig. 1). This means that for
each end-section, symmetry should be accounted for. Also, the nodes at the end-sections should in
general be able to translate freely in the sections plane. Therefore, the necessary constraints on the
nodes at the end-sections should be imposed. Such constraints can be implemented by limiting the
end plane nodes to follow the displacement of a reference node in the direction normal to the end
plane. The reference node, in this case, is the intersection of centreline and elastic neutral axis. For
this purpose, a feature available in ABAQUS (2006) namely surface-based kinematic coupling is applied
to impose the necessary constraints.
Another issue is to appropriately include the still-water shear force in the model. Fig. 9 shows a typical still-water shear force curve for a (typical) bulk carrier under AHL condition [21]. It is noted that
between two consecutive bulkheads there is a location (normally at the middle of each hold) where
the global shear force is approximately zero. It is assumed (for the present FE model) that the two
end (symmetric) sections have zero shear forces (see Fig. 10).
It should also be mentioned that the relative vertical displacement of the two transverse bulkheads
included in the model must be zero. To simulate this condition simply support boundary conditions at
the location of bulkheads could be applied, as shown in Fig. 10(a).
However, xing the vertical displacement of the transverse bulkheads would obviously inhibit hull
girder bending when combined lateral loads and global bending moment are applied. These constraints are then imposed by applying two point loads in the transverse bulkheads sufciently large

336

H.K.K. Amlashi, T. Moan / Marine Structures 21 (2008) 327352

Fig. 9. Typical distribution of shear force for a bulk carrier in AHL condition (IACS-Rec46, 1997; illustration purpose only).

to give a zero vertical displacement at the bulkheads when the model is loaded with lateral pressures
as shown in Fig. 10(b). The magnitude of the loads has been determined by running a linear elastic analysis with cargo and sea pressure loads, xed vertical displacement of the relevant nodes at the bulkheads and then retrieving the reaction forces at the xed nodes. It is observed that the required
xation forces are relatively small.
Since it is anticipated that the collapse of the hull girder will be progressively unstable due to local
buckling, the global bending is introduced by a rotation at each end of the model rather than imposing
end moments.

6. Loading conditions
The hull girder loads may be divided into still-water loads and wave-induced loads. The main global
loading is the vertical bending moment. The governing still-water loading condition for a bulk carrier
under hogging is alternate hold loading (AHL). The assessment of still-water loads is complex for bulk
carriers [11]. The primary concern is to assess the ultimate strength of the hull girder under the newly
issued CSR-BC rules [14]. We use DNV rules [16] as an alternative approach. Different AHL conditions
are considered, namely (partially) heavy cargo AHL and fully loaded cargo AHL. The lateral loads include both the still-water and wave loadings.

Pe

Ph

Ph

Ph

Y
X

Pe

Ph

Idealized multi-span beam

FS

FS

Modified multi-span beam

Fig. 10. Simple multi-span beam representation of the FE model indicating the loading conditions and boundary conditions: (a) idealized multi-span beam; (b) modied multi-span beam; Ph is the average difference between downward (cargo) and upward (sea)
pressures in loaded tank and Pe is the average upwards (sea) pressures in empty tank.

H.K.K. Amlashi, T. Moan / Marine Structures 21 (2008) 327352

337

6.1. Rule global bending moment

The design value for vertical bending moment is given by combining the longitudinal bending moment due to waves with the longitudinal bending moment in still-water caused by the most unfavourable loading condition.
The design value for wave-induced hogging moment is equal to 6.6 GN m according to CSR-BC rules
[14] and DNV rules [16]. The corresponding design still-water hogging moment is equal to 4.8 GN m in
both CSR-BC rules [14] and DNV rules [16]. However, the design wave-induced and still-water hogging
moments given in ISSC Committee report VI.2 [17] are, respectively, 6.74 GN m and 3.3 GN m.

6.2. Common structural rules for bulk carriers (CSR-BC [14])

There are two standard loading conditions for direct strength analysis of bulk carriers in AHL condition, namely (partially) heavy cargo AHL (CSR-LC-10) and fully loaded cargo AHL (CSR-LC-5). These
are shown in Figs. 11 and 12. The draught denoted by Ts is dened as scantling draught taken equal
to the maximum draught.
Different wave load cases are described in CSR-BC [14] based on the equivalent design wave
(EDW). The EDWs are specied by regular waves that generate response values equivalent to the
long-term response values of the load components considered to be dominant to the respective structural members. The corresponding hull girder loads and motions of the ships in each load case are
obtained by multiplying the reference value of each component by the relevant load combination factor (LCF). The relevant still-water bending moment is also to be added to the hull girder loads for each
load condition.
Regarding local loads, external (hydrostatic and hydrodynamic) pressures and internal (static and
inertial) pressures are considered as lateral loads in still-water and in waves. The internal static pressures are induced by the weights carried, while the internal inertial pressures are induced by the accelerations on the masses and calculated with relevant load combination factor (LCF).
Conservatively, no internal inertial forces are considered in CSR-BC rules [14] for a bulk carrier under hogging condition. The background for this is explained as follows. In hogging condition, the vertical acceleration of the cargo is downwards, which means that the inertial forces on the inner bottom
are upwards. Assuming that the downward internal static pressures of the cargo on the inner bottom
are positive, the inertial pressures will, therefore, be negative for this case. Therefore, prescribing zero
internal inertial forces will be a conservative approach for hogging condition. The calculated cargo and
sea pressures for heavy cargo AHL (CSR-LC-10) and fully loaded AHL (CSR-LC-5) based on CSR-BC rules
[14] are given in Table 4.

Fig. 11. Standard loading conditions for direct strength analysis of BC-A in AHL condition: top, the (partially) heavy cargo AHL (CSRLC-10); and bottom, fully loaded cargo AHL (CSR-LC-5); Excerpted from CSR-BC rules [14].

338

H.K.K. Amlashi, T. Moan / Marine Structures 21 (2008) 327352

Fig. 12. Denition of internal cargo pressure in hold tanks: (a) heavy cargo; and (b) fully loaded. Excerpted from CSR-BC rules [14].

6.3. DNV rules [16]

According to DNV classication notes 31.1 [22], see Fig. 13, the load condition DNV-LC-1 (alternate
hold loading with heavy cargo) corresponds to the CSR-LC-10 (heavy cargo AHL). Loading condition
DNV-LC-5 is also an AHL condition with full load, which is equivalent to CSR-LC-5.
DNV rules [16] dene design pressure loads for direct calculations that represent external and internal
pressures from sea, liquid in tanks, dry cargo, stores and equipment. Normally, the design sea pressure is
assumed to be acting on the outer shell at full draught. Moreover, the internal pressures are given for the
panel in question irrespective of possible simultaneous pressure from the opposite side. The gravity and
acceleration forces from cargo and equipment are also considered. This implies that the inertial forces
due to cargo were taken into account in DNV rules for both hogging and sagging conditions. In case of
hogging, this will be a more conservative approach than the CSR-BC approach. The calculated cargo
and sea pressures for the two mentioned loading conditions are given in Table 4.
6.4. Applied loads comparison
External and internal pressures calculated for the present vessel according to the CSR-BC rules and
DNV rules are all summarized in Table 4 and Fig. 14. Based on the results obtained, the following issues
can be identied:
(i) The design dynamic sea pressure specied by DNV and CSR-BC rules differ considerably; for
panels in bottom centreline 4 KPa in DNV rules, while 40 KPa in CSR-BC rules; for ship side panels
at maximum draught 55 KPa in DNV rules, while 120 KPa in CSR-BC.
(ii) The design cargo pressure calculated according to DNV rules at inboard-part of the inner bottom is
67% larger than the CSR-BC rule values for AHL condition with heavy cargo, see Fig. 14. This is due
to the higher cargo level and inclusion of vertical acceleration of cargo (g0 0.5av) in DNV rules, as
discussed previously. When a more relevant acceleration coefcient for hogging condition is used,
that is g0  0.5av, the cargo pressure according to DNV rules is signicantly reduced as a result of
the negative inertial forces (see Table 4). Still the cargo pressure in DNV rules (345 KPa) is higher
than that in CSR-BC (308 KPa), due to the higher cargo level used in DNV rules.
(iii) It is the difference between downward (cargo) pressures and upward (sea) pressures of the loaded
cargo hold which is of importance for the double bottom bending. For the present vessel, the difference between downward (cargo) pressure and upward (sea) pressure for the heavy cargo AHL
according to DNV rules is 4.6 times larger than that in the CSR-BC rules at inboard-part of the double bottom (see Table 4).
Based on the signicant differences in the calculated pressures according to different rules for the
present vessel, it is questionable whether or not the rule local pressures are relevant for hull girder
strength assessment, for instance in terms of a nite element hold tank analysis. As mentioned above,
the CSR-BC rule local loads are used as the main reference and DNV rule local loads are considered as an
alternative.

Table 4
Summaries of the calculation for pressure prole (CSR-BC [14] and DNV [16])
LC

Description

CSR-BC-A, applicable
to empty hold in
alternate condition
(mid-hold is empty
hold) (CSR-LC-10)

Pressure calculation
Midship draught in the considered loading condition, TLC, in m

rg
Hydrodynamic pressure (PW) for load case F2, PF2, in MPa

Hydrostatic pressure, PS, in MPa

Density of dry bulk cargo, rc, in t/m3
Assumed angle of repose, j, in degrees
hHPL, in m
h1, in m
h2, in m

II

CSR-BC-A applicable
to loaded hold in
alternate condition
(mid-hold is loaded
hold) (CSR-LC-5)

Inner bottom (4  2640)

Inner bottom (10,56018,480)
Hopper tank plate
Lower stool plate
Inner bottom (4  2640)
Dry bulk cargo pressure, PCS, in MPa
Inner bottom (10,56018,480)
Hopper tank
Lower stool, side shell & BHD
Midship draught in the considered loading condition, TLC, in m
Hydrodynamic pressure (PW) for load case F2, PF2, in MPa

Hydrostatic pressure, PS, in MPa

Density of dry bulk cargo, rc, in t/m3
Assumed angle of repose, j, in degrees
hHPU, in m
h0, in m
Dry bulk cargo pressure, PCS, in MPa

III

DNV-BC-A alternate
hold loading
condition
(DNV-LC-1)

Inner bottom
Hopper tank
Lower stool
Side shell & BHD

pL, in MPa
Hydrodynamic pressure, pdp, in MPa

Sea pressure for load point below waterline, p1, in MPa

IV

DNV-BC-A alternate
bulk cargo with lled
hold loading
condition (DNV-LC-5)

Sea pressure for load point above waterline, p2, in MPa

Density of dry bulk cargo, rc, in t/m3
a0
av
Assumed level of cargo surface in hold, hC, in m Inner bottom
Side shell
BHD
Inner bottom (4  2640)
Dry bulk cargo pressure, PC, in MPa
Inner bottom (10,56018480)
Hopper tank
Lower stool, side shell & BHD
Density of dry bulk cargo, rc, in t/m3
Inner bottom (4  2640)
Dry bulk cargo pressure, PC, in MPa
Inner bottom (10,56018,480)
Hopper tank
Lower stool
Side shell & BHD

The cargo pressures are based on g0 0.5av, while the values in parentheses correspond to g0  0.5av.

19.83
10.05
0.12
(2y B, z TLC)
0.08
(2y B,z 0)
0.04 (z y 0)
0.20
3.0
35
6.51
0
3.94
1.97
0.8
0.8
0.308
0.250
0.085
0.086
16.46
0.12
(2y B, z TLC)
0.08
(2y B, z 0)
0.04 (z y 0)
0.165
1.5
35
16.21
3.0
0.283
0.167
0.143
0.08
0.028
0.055
(2y B, z T)
0.031
(2y B, z 0)
0.004 (y z 0)
0.230
(2y B, z 0)
0.202 (y z 0)
0.025
3.0
0.45
3.86
14.6
7.6
11.1
0.514 (0.345)a
0.4 (0.268)a
0.073 (0.05)a
0.034 (0.023)a
1.5
0.444
0.421
0.232
0.172
0.08

340

H.K.K. Amlashi, T. Moan / Marine Structures 21 (2008) 327352

Fig. 13. Bulk cargo distribution lling part of cargo hold tank (heavy cargo). Excerpted from DNV-CN.31.1 [22]. (a) Transverse view;
(b) longitudinal view.

7. Sequence of load application

Two alternative load cases are considered, namely (1) pure global bending moment and (2) combined lateral loads and global bending moment. The loading histories for both cases are shown schematically in Fig. 15. The global bending is incrementally introduced by a rotation at each end of the
model up to 0.01 radian. The lateral loads are incremented up to their prescribed rule values during
the rst 10% of the total time and then held constant up to the end of the total time. Therefore, the
end rotation when the full lateral loads are applied is 1 103 radian.

8. Initial imperfections
Initial imperfections in the form of geometric imperfections (non-uniformity in shape, eccentricities and local indentations) and residual stresses are induced in marine structures during fabrication.
These imperfections can signicantly affect the ultimate strength of thin-plated structures and should
be accounted for in the ultimate strength evaluation. The imperfections are caused during a complex
fabrication process and are subject to signicant uncertainty related to the magnitude and spatial variation. In this study only geometric imperfections are considered.

0.0343(0.0343) MPa

0.05(0.05) MPa

CSR-LC-5
(FULL)

0.025(0.025) MPa

DNV-LC-5
(FULL)

0.12(0.12) MPa

0.025(0.025) MPa

(0.08) MPa

(0.08) MPa
DNV-LC-1
Heavy Cargo

CSR-LC-10
Heavy Cargo
0.085(0.167) MP a
0.308(0.283) MPa 0.250(0.283) MPa

0.073(0.232)
0.514(0.444) MP a 0.40(0.421) MPa

0.28(0.245) MPa
0.24(0.205) MPa

0.025(0.025) MPa

0.28(0.245) MPa

Pressure profile according to CSR-BC

0.23(0.23) MPa
0.202(0.202) MPa

0.23(0.23) MPa

Pressure profile according to DNV

Fig. 14. Pressure proles for alternate hold loading conditions according to CSR-BC (a) and DNV (b); the values outside the parenthesis are applicable for heavy cargo AHL, while those in parenthesis are relevant for fully loaded cargo AHL.

H.K.K. Amlashi, T. Moan / Marine Structures 21 (2008) 327352

Local loads

341

Total time

1.0

FRule

(2)
0.1

(1)

MRule

Global bending moment

Fig. 15. Loading histories: (1) pure bending; (2) combined loading.

Three types of initial out-of-straightness are accounted for plate and stiffener imperfections. In long
plates compressed in the longer direction the most signicant distortions for plates are those having
a half-wavelength approximately equal to the width of the plating [23]. Initial (out-of-straightness) deformations of the stiffeners are assumed to have the form of single half-waves between transverse
frames [24]. This kind of exural distortions may involve possible mode interaction with overall stiffener-induced failure. Stiffener web out-of-atness and ange out-of-straightness deformations are
also present. They may initiate torsional instability depending upon the stiffener geometry, but normally they are prevented in most design codes by appropriately dimensioning the stiffeners.
Obviously, the most accurate method is to use real measured imperfections. If the precise shape of
an imperfection is known, the geometrical imperfection forms can be directly specied. However, detailed information on the amplitude and form of imperfections in the relevant structure is not (always)
available. Therefore, an equivalent geometric imperfection shapes are used. Imperfections can be
implemented in most commercially available FE codes mainly in three different ways:
(1) A linear superposition of buckling eigenmodes obtained from an eigenvalue buckling analysis.
(2) The denition of the nodal coordinates.
(3) A solution obtained from a static analysis.
The third method is only applicable for very simple structures which are not sensitive to the imperfection shapes. In the second method, however, the node numbers and the relevant perturbations are
directly given to obtain the pre-deformed conguration. The relevant perturbations are usually generated from trigonometric functions or are extracted from measurements, if any. The methodology, however, requires that the user manually makes sure that the imperfection patterns are consistent across
the boundaries between the separately treated regions of a complex structure. This may be a cumbersome task to handle. The rst method, however, is widely used for the structures which are more sensitive to the imperfection forms. This method is adopted in this study.
As a crude assumption, the rst eigenmode of an FE buckling analysis is assumed to be the pattern
with the most unfavourable effect on the collapse strength. The imperfection shapes obtained, however, are only suitable for plates when the collapse is only caused by geometrical nonlinearity and
when the pre-buckling behaviour is almost linear. Moreover, for stiffened plates, stiffener web outof-atness and ange out-of-straightness deformations are coupled with other types of deformation
in higher eigenmodes. Consequently, for more complex structures, suitable initial imperfection shapes
have to be extracted from higher eigenmodes when the ordinary geometry properties are used.
An enhanced technique is to change the geometry properties such as to decouple those local deformations of interest from lower eigenmodes. This can be done by reasonably changing the thickness of
those parts of importance as well as removing unnecessary parts.

342

H.K.K. Amlashi, T. Moan / Marine Structures 21 (2008) 327352

The adopted range of values for the plate out-of-plane and stiffener lateral/exural imperfection
magnitudes correspond to values recommended by the DnV regulations [22]. The calculations are carried out with a plate out-of-plane and stiffener lateral/exural deections as being, respectively, 0.005b
and 0.001a/0.001a, where a and b are the length and width of the plate between the stiffeners and web
frames, respectively.
In the present study, initial imperfections are accounted for in the nonlinear part of the model by
performing relevant FE buckling analyses. They are all carried out in ABAQUS/Standard. This procedure
will be discussed in the following.
The nonlinear part of the model is sub-structured so that the combined initial imperfection pattern
for each individual part is generated. The procedure, as shown in Fig. 16, is started with stimulating
stiffener out-of-straightness deformation. This is fairly simple and is done by appropriately changing
the thicknesses of plate and stiffeners so as to force the panel to buckle as a whole. The same procedure
is applied in second and third steps, to capture the desired shapes for plate out-of-plane and stiffener
out-of-plane deformations. In each step the magnitude of initial imperfection is imposed to the panel
before starting the next step.
9. Analysis strategy-computing runs
In the course of the present study, several preliminary analyses were carried out to verify the modelling, load cases and boundary conditions both for the linear elastic buckling analyses and for the nonlinear analyses. The effect of constraints between the nonlinear and linear parts is found to be
negligible on the ultimate strength of the hull girder under combined loading. This is because those
constraints will only provide marginal stiffness increase in local areas far from the critical nonlinear
regions and, therefore, are not considered to affect the conclusion regarding the effect of local pressures
on the ultimate strength reduction. These constraints are not accounted for in the nonlinear analyses.
It should be mentioned that all the nonlinear analyses, for which the results are presented, include
the ultimate collapse as well as post-collapse ranges for the hull girder. The latter is more timeconsuming which indicates the presence of complex mechanisms caused by extensive plastication
or buckling of many structural members. Because of these local instabilities, global load control
methods such as the Riks method in ABAQUS/Standard are not appropriate. A strategy to stabilize
this class of problems is to apply volume proportional damping throughout the model in such a way
that the viscous forces introduced are sufciently large to prevent instantaneous buckling or collapse
but small enough not to affect the behaviour while the problem is stable. This feature is available in
ABAQUS (2006). The damping factor is chosen such that the viscous dissipated energy during the loading step is a small fraction of the change in strain energy.
A number of cases (nine in total) have been studied here. A summary of all analysis cases is shown in
Table 5.
In order to assess the effect of initial imperfections on the hull girder strength, the cases with no imperfections, i.e. cases C1, C3 and C8 are studied. Also, a modication to the hull conguration is done to
investigate the effect of reinforcement of the critical panel on the hull girder strength (cases C8 and C9).
The Modied hull conguration remains the same except for the critical panel close to the centreline
which is assumed to have a plate thickness of 25 mm instead of 18.5 mm from centreline to the GR2640.
CPU-time required on an AMD AthlonX2 86-64 Dual Core PC with 4 GB RAM was ranging from several minutes for the buckling analyses of the components up to several days for the nonlinear analyses.
10. Results and discussion
The global response of the vessel under pure global bending as well as combined lateral loads and
global bending is illustrated through moment-end rotation curves. They are compared with the result
presented by ISSC Committee VI-2 report [17] based on Smith type methods. Stress contour plots are
presented to provide an overview of the stress distribution in the structure, which may indicate how
the collapse of the larger parts of the structure (bottom and inner bottom) is developed. More extensive
results of the stress distribution in the double bottom area are shown by Amlashi [25] and Amlashi and
Moan [15].

H.K.K. Amlashi, T. Moan / Marine Structures 21 (2008) 327352

Fig. 16. Magnied illustration of different imperfection mode shapes.

343

344

H.K.K. Amlashi, T. Moan / Marine Structures 21 (2008) 327352

Table 5
Summary of analysis cases
Case no.

Description

Imperfection

Load case

C1
C2
C3
C4
C5
C6
C7
C8
C9

PB-No-Imp
CSR-LC-10-No-Imp (heavy cargo)
PB-Imp
CSR-LC-10-Imp (heavy cargo)
CSR-LC-5-Imp (full loaded cargo)
DNV-LC-1-Imp (heavy cargo)
DNV-LC-5-Imp (full loaded cargo)
Modied-CSR-LC-10-No-Imp (heavy cargo)
Modied-CSR-LC-10-Imp (heavy cargo)

No imperfection
No imperfection
Imperfection
Imperfection
Imperfection
Imperfection
Imperfection
No imperfection
Imperfection

Pure bending
CSR-LC-10
Pure bending
CSR-LC-10
CSR-LC-5
DNV-LC-1
DNV-LC-5
CSR-LC-10
CSR-LC-10

10.1. FE results against ISSC-Yaos result

The results of the cases C1 and C3 (PB-No-Imp and PB-Imp) are compared with ISSC-Yaos result in
Fig. 17. Yaos result is transformed from the moment-curvature curve into the present moment-end
rotation plot. The bending stiffness in Yaos result is somewhat larger while a major drop in the stiffness is observed close to the ultimate strength. This is due to the fact that the bending stiffness and
post-collapse behaviour depend on the extent of the model used. The stiffness reduction of the hull
girder for case C3 compared to C1, due to imperfections, is marginal.
It is seen that the post-collapse resistance dramatically decreases for the pure bending cases. This is
due to the fact that the stresses in the bottom shell are uniformly distributed up to the collapse of the
hull girder which means that the large part of the bottom is collapsed almost simultaneously. It is observed that the ultimate strength in case C3 is approximately 10% higher than ISSC-Yaos result. The
possible reasons are the interaction of multi-span panels in the FE models and the effect of residual
stresses accounted for in the Yaos model.

20

x 10

18
16

Bending Moment
(N.m)

14
12
10
8
6
C1 (PB-No-Imp)
C3(PB-Imp)
ISSC-Yao

4
2
0
0

End rotation

5
x 10-3

Fig. 17. Moment-end rotation response curves. Comparison of cases C1 and C3 with ISSC-Yaos result.

H.K.K. Amlashi, T. Moan / Marine Structures 21 (2008) 327352

345

Fig. 18. Von Mises stress contour plots for analysis case C1 (PB-No-Imp) at ultimate failure load level.

10.2. Overview of the FE results

Stress contour plots for cases C1C7 at ultimate load level are shown in Figs. 1925. The deection pattern is also indicated in an exaggerated scale. The following information can be extracted from the gures:
(i) The stress elds at the ultimate load level for cases C1 and C3 are practically similar. The location
of the collapse in the double bottom is also the same in these two cases, see Figs. 18 and 20.
(ii) The bending of the double bottom at ultimate load level is visible for cases C2 and C4 (CSR-LC-10No-Imp and CSR-LC-10-Imp), as shown in Figs. 19 and 21. The shear buckling in centre girder and
GR18480 is also observed for both cases. Although the collapse is concentrated in the middle of
the empty hold tank in both cases, the collapse in the case C4 involves the adjacent frame span
(Fig. 21). It is then expected that the ultimate strength of the hull girder is reduced substantially
as compared to the case C2. This is discussed subsequently in detail.
(iii) By comparing Fig. 21 for the case C4 (CSR-LC-10-Imp) with the case C5 (CSR-LC-5-Imp) in Fig. 22 at
ultimate load level it is seen that there is a clear difference in the sequence of collapse between the
two mentioned cases. The collapse load of the hull girder in the case C5 is attained when the

Fig. 19. Von Mises stress contour plots for analysis case C2 (CSR-LC-10-No-Imp) at ultimate failure load level.

346

H.K.K. Amlashi, T. Moan / Marine Structures 21 (2008) 327352

Fig. 20. Von Mises stress contour plots for analysis case C3 (PB-Imp) at ultimate failure load level.

critical panel in the outboard part of the bottom is failed, while the case C4 involves buckling of
two frame spans as mentioned above. Shear buckling in centre girder and GR18480 is observed for
the case C5 similar to that in the case C4.
(iv) A signicant double bottom bending is observed in cases C6 (DNV-LC-1-Imp; Fig. 23) and C7
(DNV-LC-5-Imp; Fig. 24). The gures indicate that the collapse is initiated from the centre critical
panel in the inboard-part of the bottom for these two loading conditions.
Hence Figs. 1824 show that the stress distribution in the structure changes as a result of combined
lateral loads and global bending compared to a pure bending loading. The double bottom bending due
to lateral loads may initiate large stress redistribution between bottom and inner bottom. This issue is
pursued in more detail in the companion paper [15].
10.3. Ultimate hull girder strength
The moment-end rotation response curves obtained for cases C1C7 together with ISSC-Yaos result
are shown in Fig. 25. A summary of the computed ultimate capacity and rst gross element failure moment is given in Table 6.

Fig. 21. Von Mises stress contour plots for analysis case C4 (CSR-LC-10-Imp) at ultimate failure load level.

H.K.K. Amlashi, T. Moan / Marine Structures 21 (2008) 327352

347

Fig. 22. Von Mises stress contour plots for analysis case C5 (CSR-LC-5-Imp) at ultimate failure load level.

It is emphasized that the ISSC-benchmarked vessel has been designed in accordance with neither
the CSR-BC [14] rules nor DNV rules [16]. The actual design loading condition is unknown to the authors. Therefore, it is very difcult to draw any clear comparison between them in view of design loading condition. The purpose here has been to investigate the effect of different AHL conditions on the
strength of the hull girder for a hypothetical yet realistic bulk carrier.
The rst gross element failure moment (Mf) is dened as the bending moment where the moment-end
rotation curve has its rst distinct change of slope. This value is not a measure of rst element failure in the
structure, but rather the failure of a large part of the bottom. Obviously, some uncertainty is associated
with measuring this value. The results from the nonlinear FE analyses are discussed in the following.
It is seen from Fig. 26 that the bending stiffness of the hull girder is smaller for the case C2 (CSR-LC10-No-Imp) than the case C1 (PB-No-Imp). As a general observation, increasing the local loads will decrease the bending stiffness of the hull girder. Whether or not this is a general property for a hull girder
under combined loading cannot be identied from these analyses.
A particular feature of the moment-end rotation curves is the double-maxima shape of the curves
for the combined loading cases. The reason for the secondary peak of the bending capacity is the stress
redistribution in the double bottom area.

Fig. 23. Von Mises stress contour plots for analysis case C6 (DNV-LC-1-Imp) at ultimate failure load level.

348

H.K.K. Amlashi, T. Moan / Marine Structures 21 (2008) 327352

Fig. 24. Von Mises stress contour plots for analysis case C7 (DNV-LC-5-Imp) at ultimate failure load level.

Comparison of curves C4C7 (combined loading cases) with C1 and C3 (pure bending cases) in
Fig. 25 shows that one of the consequences of local loads is to convert post-collapse behaviour from
unstable to a stable form and simultaneously a signicant reduction of the ultimate strength. The results in Fig. 25 and Table 6 conrm the previous nding by stvold et al. [13] that the ultimate strength
of a bulk carrier hull girder under AHL condition is likely to be signicantly reduced as a result of global
stress redistribution between bottom and inner bottom (signicant double bottom bending in the
empty hold tank).
It is further noted that the lateral loading does not only inuence the ultimate strength of the hull
girder, but also the progressive collapse behaviour (double-maxima shape).

x 109

18
16

Bending Moment
(N.m)

14
12
10
8
6
C1 (PB-No-Imp)
C2 (CSR-LC-10-No-Imp)
C3 (PB-Imp)
C4 (CSR-LC-10-Imp)
C5 (CSR-LC-5-Imp)
C6 (DNV-LC-1-Imp)
C7 (DNV-LC-5-Imp)
ISSC-Yao

4
2
0

End rotation
Fig. 25. Moment-end rotation response curves.

5
x 10-3

H.K.K. Amlashi, T. Moan / Marine Structures 21 (2008) 327352

349

Table 6
Summary of computed hogging moment capacity and rst gross element failure moments for all analysis cases
Case no.

Case id.

Mu (GN m)

Mf (GN m)

Mu/Mf

Mf/MC3
f

Mu/MC3
u

ISSC
C1
C2
C3
C4
C5
C6
C7

ISSC-Yao
PB-No-Imp
CSR-LC-10-No-Imp (heavy cargo)
PB-Imp
CSR-LC-10-Imp (heavy cargo)
CSR-LC-5-Imp (full loaded cargo)
DNV-LC-1-Imp (heavy cargo)
DNV-LC-5-Imp (full loaded cargo)

17.36
19.52
12.21
18.95
8.01
9.12
5.26
4.93

18.17
10.30
17.96
7.20
8.90
5.02
4.12

1.07
1.18
1.06
1.11
1.02
1.05
1.20

1.00
0.40
0.49
0.30
0.23

1.00
0.42
0.48
0.28
0.26

An issue here is the signicant reduction of the ultimate capacity due to lateral loading (see Table 6).
For this particular vessel under CSR-AHL conditions, the ultimate hull girder strength and rst gross
element failure moment are dropped approximately 50%. Further reduction in the ultimate strength
of the hull girder and rst gross element failure moment is observed for the vessel under DNV-AHL
conditions. Since the true design cargo and sea pressures for the present vessel were unknown, the assumed scantlings may not be consistent with the lateral pressures used here.
It is observed that the effect of assumed imperfections is more pronounced when the combined lateral loads and global bending are present. The reduction in the ultimate strength of the hull girder due
to the initial imperfections is 3% for the pure bending cases (C1 vs. C3), while the reduction is about 35%
for the combined loading cases (C2 vs. C4). The same tendency was found in the results obtained by
stvold et al. [13], however, with a strength reduction of about 14%.
The signicant reduction of ultimate strength due to interaction of lateral loads and imperfections
can be explained by the fact that the transverse compressions in way of inner bottom and bottom shell
in the empty hold tank together with high lateral sea pressure will signicantly decrease the stiffness
and strength of components in the double bottom area. This has previously been demonstrated for the
critical panels in the bottom shell (see Fig. 6).
The difference in the ultimate strength reduction (due to imperfections) between the present vessel
and stvolds vessel is because the present vessel under CSR-AHL condition is not well stiffened in way
of bottom shell for this particular loading, while the stvolds vessel had an optimized bottom design
for combined loading. It should be born in mind that the way in which the imperfections and local

x 109

No imperfections

x 109

12

12

10

10

Bending Moment
(N.m)

Bending Moment
(N.m)

End rotation

C2 (CSR-LC-10-No-Imp)
C8 (Modified-CSR-LC-10-No-Imp)

Imperfections

5
-3

x 10

C4 (CSR-LC-10-Imp)
C9 (Modified-CSR-LC-10-Imp)

End rotation

4
x 10-3

Fig. 26. Moment-end rotation response curves. (a) Comparison of cases C2 and C8 (effect of hull girder strengthening with no imperfections). (b) Comparison of cases C4 and C9 (effect of hull girder strengthening considering imperfections).

350

H.K.K. Amlashi, T. Moan / Marine Structures 21 (2008) 327352

loads interact depends upon the geometry of the hull girder and the stress redistribution due to the
double bottom bending and should be carefully exercised.
In order to illustrate the effect of imperfections on the ultimate hull girder strength in view of combined lateral loads and global bending, the two cases C8 (Modied-CSR-LC-10-No-Imp) and C9 (Modied-CSR-LC-10-Imp) are studied here. The results are shown in Fig. 26 and Table 7. The modication in
the Modied-Hull cases, i.e. cases C8 and C9 has a larger effect on the rst gross element failure moment than the ultimate strength of the hull girder.
For this particular geometry, the effect of the modication on the ultimate failure moment is marginal, while the rst gross element failure moment (due to this modication) is increased about 12% in
both cases, i.e. C8 and C9 as compared to corresponding cases C2 and C4, see Table 7. This is due to the
fact that the rst gross element failure moment is the moment where a large part of the bottom is collapsed and since the critical panels in the central part of the bottom are strengthened in cases C8 and
C9, the collapse of the bottom is to some extent delayed.
It is also interesting to see that the moments at rst gross element failure and at ultimate failure are
practically the same for the case C9 (Modied-CSR-LC-10-Imp). It means that there is no reserve capacity
beyond the rst gross element failure moment (Mu/Mf) for the case C9. However, the reserve capacity
beyond the rst gross element failure is 10% for the case C8 (Modied-CSR-LC-10-No-Imp), see Table 7.
The trend is also observed for the case C4 (CSR-LC-10-Imp) and the case C2 (CSR-LC-10-No-Imp),
being, respectively, 11% and 18%. It means that when the imperfections are present, the reserve capacity
beyond the rst gross element failure (Mu/Mf) is reduced. This can be explained by the fact that the initiation of buckling in critical (intact) panels of the bottom for the case C2 (CSR-LC-10-No-Imp) will force
the other (intact) panels to deform in mode shapes which are not compatible with the deformations
caused by the critical modes of buckling for those particular panels. This may essentially give an extra
strength for other panels because the imperfection shapes developed there are not the critical ones and
may have some stiffening effects. The accounted imperfections for the case C4 (CSR-LC-10-Imp), however, are based on the critical modes of buckling for the panels in the double bottom area. Hence the
strength of each individual panel might be smaller than that for the case C2 (CSR-LC-10-No-Imp).
Therefore, the reserve capacity beyond the rst gross element failure (Mu/Mf) for cases without imperfections will be larger than that for cases with imperfections (based on the critical modes of buckling). Further documentation of the interaction between local loads and imperfections is needed.
10.4. Effect of differences in local loads
Different pressure proles based on CSR-BC rules and DNV rules have been used in the FE ultimate
strength evaluation. They include both AHL with fully loaded cargo and AHL with heavy cargo. The results are shown in Fig. 25 and Table 6.
The variation in the ultimate strength as well as rst gross element failure moment for the cases C4
C7 (Table 8) seems to indicate that the relative difference between the average pressure in the cargo
hold and the average pressure in the empty hold (Ph/Pe) may play an important role in the ultimate
strength reduction in AHL conditions (see Fig. 10). It may roughly indicate the intensity of double bottom bending. It is seen, from Table 8, that the Ph/Pe is reduced for the case C5 (CSR-LC-5-Imp; fully
loaded AHL) as compared with the case C4 (CSR-LC-10-Imp; heavy cargo AHL). It shows that the double
bottom bending will be more severe in the case C4 than the case C5.
It is observed that the Ph/Pe for the case C7 (DNV-LC-5-Imp; fully loaded AHL) is larger than that for
the case C6 (DNV-LC-1-Imp; heavy cargo AHL). This can explain the results shown in the moment-end
rotation response curves (Fig. 25), where the ultimate hull girder strength of the case C7 is less than
Table 7
Calculated moment capacities and rst gross element failure moments for cases C8 and C9
Case no.

Case id.

Mu (GN m)

Mf (GN m)

Mu/Mf

Mmodied
/Mf
f

Mmodied
/Mu
u

C2
C8
C4
C9

CSR-LC-10-No-Imp
Modied-CSR-LC-10-No-Imp
CSR-LC-10-Imp
Modied-CSR-LC-10-Imp

12.21
12.54
8.01
8.04

10.30
11.44
7.20
8.04

1.18
1.10
1.11
1.00

1.11

1.12

1.03

1.00

H.K.K. Amlashi, T. Moan / Marine Structures 21 (2008) 327352

351

Table 8
Average downward and upward pressures for different pressure proles
Case no.

Case id.

Mu (GN m)

Mf (GN m)

Mu/Mf

Ph (N/mm2)

Pe (N/mm2)

Ph/Pe

C4
C5
C6
C7

CSR-LC-10-Imp
CSR-LC-5-Imp
DNV-LC-1-Imp
DNV-LC-5-Imp

8.01
9.12
5.26
4.93

7.20
8.90
5.02
4.12

1.11
1.02
1.05
1.20

0.046
0.014
0.109
0.146

0.26
0.23
0.22
0.22

0.18
0.06
0.50
0.66

Ph is the average difference between downward (cargo) and upward (sea) pressures in loaded tank and Pe is the average upward
(sea) pressures in the empty tank (see Fig. 10).

that of the case C6. It means that, for this particular vessel, the DNV-LC-5-Imp (fully loaded) AHL condition is more severe than DNV-LC-1-Imp (heavy cargo) AHL condition.
It is observed that the Ph/Pe according to DNV rules (cases C6 and C7) is much larger than that based
on CSR-BC rules (cases C4 and C5). This fact explains the more signicant reduction in the strength of
the hull girder for the former cases. Further works are needed to document the effect of differences in
local loads on the ultimate hull girder strength.
11. Conclusions
Extensive nonlinear FE strength analyses of a Capesize bulk carrier under alternate hold loading
(AHL) in hogging condition have been carried out. These analyses required a very signicant effort, covering a wide range of issues of importance in design loads assessment and in FE strength assessment. A
global FE (1/2 1 1/2) hold tanks model of the ISSC-benchmarked bulk carrier is established. A mesh
convergence study is carried out in order to nd the appropriate mesh for the model. The CSR-BC rule
local loads are used as the main reference and DNV rule local loads are considered as an alternative.
Both (partially) heavy cargo AHL and fully loaded cargo AHL are considered. The effect of imperfections
on the strength of the hull girder under AHL condition is assessed. The implication of using different
design pressures on the hull girder strength is investigated.
FE modelling aspects of a (1/2 1 1/2) hold tanks model under AHL condition are discussed and
the model is generated to accurately predict the ultimate strength of the hull girder. The model is composed of two parts, one which accounts for linear behaviour and one which accounts for nonlinear behaviour. A total numbers of 475,000 DoF and 620,000 DoF are included in the nonlinear part and the
linear part, respectively.
Regarding local pressures, the design cargo pressures calculated according to DNV rules are much
larger than those in CSR-BC for both (partially) heavy cargo AHL and fully loaded cargo AHL (almost
50% larger). This is because the inertial forces due to cargo are assumed to be zero (in hogging condition)
for the CSR-BC rules, while the inertial forces are included in the cargo pressure calculation for the DNV
rules. It might be discussed to what extent the local pressures are relevant for hull girder strength assessment, for instance in terms of a nite element hold tank analysis. The difference between design pressures
calculated according to fully loaded cargo AHL and (partially) heavy cargo AHL should also be noted.
It is found that the hull girder strength of a bulk carrier under AHL condition is signicantly reduced
as a result of global stress redistribution between bottom and inner bottom (signicant double bottom
bending in the empty hold tank). The global stress redistribution induces unique progressive collapse
behaviour, i.e. double-maxima shape in the bending moment-end rotation curves. The normal practice for bulk carrier design (in hogging condition) is such that the (partially) heavy cargo AHL is governing the design of double bottom. It is found that the fully loaded cargo AHL could also be the
governing loading condition depending upon the design pressures in the empty and cargo holds.
It is found that the interaction between the initial imperfections and combined lateral loads and
bending moment depends on the geometry of the hull girder and the stress redistribution due to
double bottom bending. The reserve capacity beyond the rst gross element failure moment for AHL
condition is found to be reduced due to initial imperfections.
It is found that the relative difference between the average pressure in the cargo hold and the
average pressure in the empty hold is a measure of bending exibility of the double bottom area.
The relative difference has a direct effect on the hull girder strength reduction.

352

H.K.K. Amlashi, T. Moan / Marine Structures 21 (2008) 327352

In this paper various issues relating to design loads assessment and nonlinear FE strength assessment of bulk carriers under AHL conditions are addressed. The crucial issue is the characterisation
of design sea and cargo pressures for different AHL conditions. Efforts are still needed to assess the uncertainty of bulk carrier hull girder strength under different AHL conditions in hogging. The relative
difference between the average pressure in the cargo hold and the average pressure in the empty
hold should be noted. The effect of the relative difference on the hull girder strength depends on
how the global stress redistribution is developed in the double bottom area. This issue is briey discussed in this paper and more efforts to assess its implication on the hull girder strength are needed.
More efforts are also needed to assess the level of interaction between the initial imperfections and
local loads.
Acknowledgments
The authors would like to thank Tom K. stvold and Eivind Steen, DNV for kindly providing information and valuable discussions. The help of M.T. Tavakoli in performing FE buckling analyses is also
appreciated. The support from Centre for Ships and Ocean Structures through providing a dual core
PC for the nonlinear analyses of the present study is also acknowledged.
References
[1] Chen YK, Kutt LM, Piaszczyk CM, Bieniek MP. Ultimate strength of ship structures. SNAME Trans 1983;91:14968.
[2] Kutt LM, Piaszczyk CM, Chen YK, Liu D. Evaluation of the longitudinal ultimate strength of various ship hull congurations.
SNAME Trans 1985;93:3353.
[3] Valsgrd S, Jorgensen L, Boe AA, Thorkildsen H, Ultimate hull girder strength margins and present class requirements.
SNAME symposium 91 on marine structural inspection, maintenance and monitoring, Arlington; March 1991. B 1-19.
[4] Ueda Y, Rashed SMH. The Idealized Structural Unit Method and Its Application to Deep Girder Structures. Comput Struct
1984;18(2):27793.
[5] Smith CS. Inuence of local compressive failure on ultimate longitudinal strength of a ships hull. Trans PRADS 1977:739.
[6] Caldwell JB. Ultimate longitudinal strength. Trans RINA 1965;107:41130.
[7] Ueda Y, Rashed SMH, Paik JK. Buckling and ultimate strength interaction in plates and stiffened panels under combined
inplane biaxial and shearing forces. Mar Struct 1995;8(1):136.
[8] Yao T, et al. Progressive collapse analysis of a ships hull under longitudinal bending. J Soc Naval Arch Jpn 1991;170:44961.
[9] Yao T, et al. Progressive collapse analysis of a ships hull under longitudinal bending (2nd report). J Soc Naval Arch Jpn
1992;172:43746.
[10] Mansour A, Thayamballi A, Ultimate strength of a Ships hull girder. Ship structure committee report, SSC-299; 1980.
[11] Moan T, Shu Z, Drummen I, Amlashi HKK, Comparative reliability analysis of ships considering different ship types and
the effect of ship operations on loads. Trans SNAME, Florida; 2006.
[12] Japan Shipbuilding Research Association (JSRA), Investigation into structural safety of aged ships. Report of no.74 rule and
regulation panel; 1999. p. 3550 [in Japanese].
[13] stvold TK, Steen E, Holtsmark G, Nonlinear strength analyses of a bulk carrier a case study. In: Proceedings of the ninth
PRADS, STG, Hamburg; 2004.
[14] CSR-BC. IACS common structural rules for bulk carriers. IACS Publications; 2006 [became effective on 1 April 2006].
[15] Amlashi HKK, Moan T, Ultimate strength analysis of a bulk carrier hull girder under alternate hold loading condition
a case study, Part 2: double bottom bending and stress redistribution. Mar Struct; submitted for publication.
[16] DNV, Rules for classication of ships; July 2006.
[17] Yao T, et al., Report committee VI.2: ultimate hull girder strength. In: Proceedings of the 14th international ship and
offshore structures congress, Nagasaki, Japan; 2000. p. 32191.
[18] ABAQUS/Standard. General purpose nite element analysis program standard module, version 6.6. USA: Hibbit, Karlson
& Sorensen, Inc.; 2006.
[19] ABAQUS. Theory manual and example manual, version 6.6. USA: Hibbit, Karlson & Sorensen, Inc.; 2006.
[20] Amlashi HKK, Moan T, On the strength assessment of pitted stiffened plates under biaxial compression loading. In: 24rd
international conference on offshore mechanics and arctic engineering, OMAE 2005, Halkidiki, Greece; June 2005.
[21] IACS-Rec46, Bulk carriers guidance and information on bulk cargo loading and discharging to reduce the likelihood of
over-stressing the hull structure. Guidelines and recommendations, No. 46; 1997.
[22] DNV, Strength analysis of hull structure in bulk carriers. Classication Notes No. 31.1; January 2002.
[23] Smith CS. Compressive strength of welded steel ship grillages. Trans RINA 1975;117.
[24] Faulkner D. A review of effective plating for use in the analysis of stiffened plating in bending and compression. J Ship Res
1975;19(1).
[25] Amlashi HKK, Ultimate strength and reliability analysis of ship hulls. Ph.D. thesis, Trondheim: Norwegian University of
Science and Technology (NTNU); 2008.