Ultimate Limit Stat Design Ship Hulls

ABS TECHNICAL PAPERS 2002
Ultimate Limit State Design of Ship Hulls

Jeom Kee Paik (M), Pusan National University, Ge Wang (M), American Bureau of Shipping,
Bong Ju Kim (StM), Pusan National University, and Anil Kumar Thayamballi (LM), Chevron
Shipping Company LLC
Originally presented in SNAME Transactions, Vol. 110, 2002. Reprinted with the permission of the Society of Naval
Architects and Marine Engineers (SNAME). Material originally appearing in SNAME publications cannot be reprinted
without written permission from the Society, 601 Pavonia, Ave., Jersey City, NJ 07306.
ABSTRACT
This paper is a logical sequel to the authors last two SNAME annual meeting papers (Paik et al. 2000, 2001) which
dealt with the ultimate limit state design of ship plating and stiffened panels. It aims to deal with the advanced ultimate
limit state design of ship hulls under vertical bending moments. Traditionally, design criteria and procedures were
primarily based on allowable stresses and buckling checks. It is now well recognized that the limit state approach is a
better basis for design, because it determines, in a more realistic way, the real safety margin of any economically
designed structure. While the limit state design for steel structures uses limit states classified into four types, namely
serviceability limit state, ultimate limit state, fatigue limit state and accidental limit state, the present paper is concerned
with the ultimate limit state of ship hulls.
In this paper, efficient and accurate methodology for the progressive collapse analysis of ship hulls is presented. The
characteristics of progressive collapse behavior of a total of 10 typical merchant ships under vertical bending are then
investigated using the analysis method presented. Effects of lateral pressure and horizontal moment on the hull girder
ultimate vertical moment are studied. Closed-form ultimate strength formulations for the ultimate strength of ships are
developed. Finally, the ultimate limit state design format for ships is addressed.
85

INTRODUCTION
During the last few decades, the emphasis in
structural design has been moving from the allowable
stress design to the limit state design, since the latter
approach has many more advantages. A limit state is
formally defined as a condition for which a particular
structural member or an entire structure fails to perform
the function that it has been designed for. From the
special viewpoint of a structural designer, four types of
limit states are considered, namely
Serviceability limit state (SLS)

Ultimate limit state (ULS)
Fatigue limit state (FLS)
Accidental limit state (ALS)
The structural design criteria against the ULS are

based on plastic collapse or ultimate strength. The design
of many types of structures including merchant ship
structures has in the past tended to rely on estimates of
the buckling strength of components, usually from their
elastic buckling strength adjusted by a simple plasticity
correction. This is represented by point A in Fig.1. In
such a scheme, the structural designer does not use
detailed information on the post-buckling behavior of
component members and their interactions. The true
ultimate strength represented by point B in Fig.1 is
typically higher although one can never be sure of this.
Design load level 1
SLS conventionally represents failures under normal

operations due to deterioration of less vital functions
such as
ULS (also called ultimate strength) represents the

collapse of the structure due to loss of structural stiffness
and strength related to
Loss of equilibrium in part or of entire structure,
considered as a rigid body (e.g., overturning or
capsizing)
Attainment of the maximum resistance capacity of
sections, members or connections by gross
yielding, rupture or fracture
Instability in part or of the entire structure
resulting from buckling or plastic collapse of
plating, stiffened panels and support members
FLS represents fatigue crack occurrence in
structural details due to stress concentration and crack
damage accumulation under the action of repeated
loading.
ALS represents excessive structural damage as
consequences of accidents, e.g., collisions, grounding,
explosion and fire, which affect the safety of the
structure and the environment.
It is important to note that in limit state design of a
structure, these various types of limit states may be
required to have different safety levels. The actual safety
level to be attained for a particular type of limit state is a
function of its perceived consequences and ease of
recovery to be incorporated in design.
86
Force
Local damage which may reduce the durability of

the structure or affect the efficiency of structural
or non-structural elements
Unacceptable deformations which affect the
efficient use of structural or non-structural
elements or the functioning of equipment
Excessive vibration or noise which causes
discomfort to people or affect non-structural
elements or the functioning of equipment
Deformations and deflections which may spoil the
aesthetic appearance of the structure
Linear
elastic
response
B
A
Ultimate strength
Buckling strength
Design load level 2
Proportional limit
Displacement
Figure 1. Structural design considerations based on the

ultimate limit state
In design, when the load level 2 shown in Fig.1 is
applied, the structure will be safe, but if the load level 1
is applied the structure will possibly collapse. Arguably
the ultimate strength is a better basis for design, but as
long as the strength level associated with point B
remains unknown (as it is with traditional allowable
stress design or linear elastic design methods), it is
difficult to determine the real safety margin. Hence more
recently, the design of structures such as those of navy
ships as well as offshore platforms and land-based
structures such as steel bridges has tended to be based on
the ultimate strength.
The safety margin of structures can be evaluated by
a comparison of its ultimate strength with the extreme
applied loads as depicted in Fig.1. To obtain a safe and
economic structure, the ultimate load-carrying capacity
as well as the design load must be assessed accurately.
The structural designer can perform a structural safety
assessment in the preliminary design stage if there are
simple expressions available for accurately predicting
the design loads, load combinations, and the ultimate
strength. A designer may even desire to do this not only
for the intact structure, but also for structures with
premised damage, in order to assess and categorize their
damage tolerance and survivability.
It is of importance to note that the structure must be
designed so that brittle fracture does not take place even
if ductile modes of structural failure may be allowed.
This is because brittle failure will lead to a sudden
collapse of the global structure while ductile failure

modes will allow the structure to redistribute internal
stresses so that a sudden drop in structural resistance can
be avoided. To be of adequate ductility, the following
items are normally required to be satisfied in structural
design, namely
Meet the material toughness requirements
Avoid a situation of combined high stress
concentration and undetected weld defects in the
structural details
Design the structural details so as to allow a
certain plastic deformation resulting in avoidance
of transverse weld failure
Make the scantlings of the strength members so
that a sudden decrease of structural resistance
should not take place in the post-ultimate strength
regime
The present paper is concerned with ULS design
methodologies for hull girders of merchant ships. The
progressive collapse characteristics of a total of 10
typical merchant ships under vertical moment are
investigated. Closed-form ultimate strength formulations
for the ultimate strength of ships are presented. Finally,
the paper addresses the ultimate limit state design format
for ships.
using the so-called deep girder unit. Several different

types of ISUM units such as the beam-column unit (also
called plate-stiffener combination unit), the rectangular
plate unit and the stiffened panel unit have so far been
developed.
In almost a parallel development to ISUM, Smith
(1977) suggested a similar approach to ISUM. In the
Smith approach, a ships hull is modelled as an assembly
of only plate-stiffener combination units for each of
which the load versus end deformation characteristics
need to be first obtained using nonlinear finite element
analyses. The Smith approach is also a type of the ISUM
in this regard.
The idealized structural unit method (ISUM) has
been recognized as an efficient and accurate
methodology for the progressive collapse analysis of
steel plated structures such as ships, offshore platforms
and box-girder bridges (Paik & Thayamballi 2002). Steel
plated structures are typically composed of several
different types of structural members such as support
members (or beam-columns), rectangular plates and
stiffened panels. In ISUM modeling, such members are
regarded as the ISUM units, as shown in Fig.2. It is
important to realize that an identical structure may be
modeled in somewhat different ways by different
analysts, but it is of course always the aim to model so
that the idealized structure behaves in (nearly) the same
way as the actual structure.
EFFICIENT AND ACCURATE METHODOLOGY

FOR THE PROGRESSIVE COLLAPSE ANALYSIS
OF SHIPS
The approaches for the progressive collapse analysis
of ships hull structures may be classified into two
groups, namely
Conventional nonlinear finite element method
Simplified (special purpose) nonlinear finite
element method
The conventional nonlinear finite element method
can of course be used to analyze the detailed nonlinear
response of ship structures which may involve both
geometric and material nonlinearities until and after the
overall hull girder collapse is reached. While the
application of the conventional nonlinear finite element
method to the progressive collapse analysis of ships
hulls is not impossible, it is usually impractical because
of the huge amount of computational cost involved,
specifically when a series of the analyses varying design
variables are required.
An alternative to nonlinear finite element analyses is
to reduce the number of degrees of freedom. Modelling
the object structure with very large sized structural units
is perhaps the best way to do that. Properly formulated
structural units in such an approach can then be used to
efficiently model the actual nonlinear behaviour of large
structural units. Ueda & Rashed (1974, 1984), who
suggested this idea, called it the idealized structural unit
method (ISUM) when they attempted to analyze the
ultimate strength of a ship transverse framed structure
(a) A typical steel plated structure
(b) Structural idealization as an assembly of platestiffener combination units
87

z
L
bf
~f
hw
b
tw
u 2, Rx2
u1, Rx1
v 1, Ry1
v2, Ry2
w2, Rz2
w1, Rz1
Figure 3(a). The ISUM beam-column unit with attached

plating ( : nodal points)
z
L
(c) Structural idealization as an assembly of platestiffener separation units
bf
~f
hw
u1, Rx1
tw
1
v 1, Ry1
u 2, Rx2
x
v2, Ry2
w2, Rz2
w1, Rz1
Figure 3(b). The ISUM beam-column unit without

attached plating ( : nodal points)
y
w3, Rz3
v2, R y2
v3, R y3
u2, Rx2
u3, R x3
(d) Structural idealization as an assembly of stiffened

panels
w2, R z2
Figure 2. Various types of idealizations for a steel plated

structure
The ISUM beam-column unit has two nodal points,
as shown in Fig.3, i.e., one at the left end and the other at
the right end. Each node is located where the beam is
connected to another member. The nonlinear behavior of
the beam-column unit is expressed by three translational
degrees of freedom at each nodal point. A rectangular
(unstiffened) plate can be modeled as one rectangular
plate unit as shown in Fig.4, while a stiffened panel can
be modeled as one stiffened panel unit, as shown in Fig.5.
The behavior of the rectangular plate or stiffened panel
unit is formulated in terms of three degrees of freedom at
each of the corner nodal points. A larger supporting
member such as a deep girder, in which local web
buckling can occur, may be modeled as an assembly of
the panel unit and the beam-column unit, where the web
is modeled as one plate (or panel) unit and the flange is
modeled as one beam-column unit. Depending on the
purpose of analysis, one may use different types of the
ISUM units, i.e., with different nonlinear behavior
characteristics. For ultimate strength analysis, the ISUM
units will need to take into account buckling and
yielding as shown in Fig.6.
w4, Rz4
~t
1
u1, Rx1
4
u4, R x4
w1, Rz1
v1, R y1
v4, R y4
Figure 4. The ISUM rectangular plate unit ( : nodal

points)
bfy
tfy
hwy
twy
y
w3, Rz3
v2, Ry2
v3, Ry3
u3, Rx3
u2, Rx2
2
w2, Rz2
u1, Rx1
w1, Rz1
twx
b
a
b
b
tfx
a
~t
bfx
w4, Rz4
4
u4, Rx4
v1, Ry1
hwx
v4, Ry4
Figure 5. The ISUM stiffened panel unit ( : nodal

points)
88
L = 18.0 m
B = 4.2 m
D = 2.8 m
Y = Yield strength
cr = Buckling strength
u = Ultimate strength
Tension
Compression
Imperfect
Perfect
Figure 7(a). Mid-ship section of the Dow frigate test ship
cr
u
u
Y
Figure 6. Idealized stress-strain behavior of the ISUM

plate or stiffened panel unit for the ultimate strength
analysis
Figure 7(b). ALPS/ISUM model for the Dow frigate test

hull
15
: Experiment (Dow 1991)

: HULLST (Yao et al. 2000)
with initial imperfections
at an average level
10
Vertical moment (MNm)
For ultimate strength analysis of ship structures

under extreme hull girder loads, the structural model
used can cover the extent of a cargo hold, or more
approximately a hull section between two adjacent
transverse frames. Rectangular plate or stiffened panel
units are employed for the purpose of modeling side
girders, transverse webs, and inner / outer shell plating
of the structure. The ISUM theory was automated within
ALPS/ISUM program which stands for nonlinear
Analysis of Large Plated Structures using the Idealized
Structural Unit Method. The validity of ALPS/ISUM
was checked by comparing the computed results with the
experimental results such as those obtained using large
scale ship hull models.
Figure 7 shows a selected ALPS/ISUM comparison
result for test models, which pertain to the experiment of
Dow (1991) who tested the 1/3 scale frigate hull model
in sagging. The ALPS/ISUM model in this case extends
between web frames. While it would be more relevant to
take the hull module between transverse bulkheads as
the extent of the analysis, the present simpler model
between web frames may also be appropriate as long as
the transverse frames are strong enough so that they
would not fail prior to the longitudinal members.
Figure 7(c) shows the progressive collapse behavior
of the Dow test structure under sagging or hogging
moment, as obtained by ALPS/ISUM. The Dow test
result for sagging is also plotted. In the ALPS/ISUM
computations, the magnitude of initial imperfections is
varied. Figure 7(c) also plots the results of Yao et al.
(2000) as obtained using the so-called Smith method
which models the structure as an assembly of only the
plate-stiffener combinations. It is seen from Fig.7(c) that
ALPS/ISUM provides quite accurate results when
compared with the experiment. Of interest, the
computing time used was 2 minutes for the ALPS/ISUM
analysis using a Pentium III personal computer.
3
2
5
4
0
5
-5
ALPS/ISUM:
-10
: w opl
: w opl
: w opl
: w opl
5 : w opl
6 : w opl
1
2
3
4
-15
-4
-3
-2
-1
= 0.025 2 t , rcx / Y = 0.0

= 0.025 2 t , rcx / Y = 0.05 (Slight )
= 0.1 2 t , rcx / Y = 0.0
= 0.1 2 t , rcx / Y = 0.15 (Average)
= 0.3 2 t , rcx / Y = 0.0
= 0.3 2 t , rcx / Y = 0.3 (Severe)
Curvature 10-7 (1/mm)
Figure 7(c). Comparison of ALPS/ISUM with the Dow

test results, varying the level of initial imperfections
As another example, a 105,000 DWT double hull
tanker hull with one center-longitudinal bulkhead is now
considered to see the influence of structural idealization
techniques on the progressive collapse behavior under
vertical bending. Based on the structural idealization
techniques noted above, six types of ALPS/ISUM
modeling methods are considered as shown in Fig.8. As
the extent of the analysis, the first five models (i.e.,
Models I to V) take a single hull segment between two
89

adjacent transverse frames or floors, while Model VI
takes one cargo hold between two transverse bulkheads.
Mode I models the structure by only the plate-stiffener
combination elements (beam-column units). In Model II,
the entire structure is idealized by the plate-stiffener
separation elements. While deck and bottom stiffened
panels in Models III and IV are modeled by the stiffened
panel units, Model III idealizes all vertical members
using the plate-stiffener separation elements, and Model
IV models the center-longitudinal bulkhead using the
plate-stiffener separation elements. In Models V and VI,
all members are modeled by the stiffened panel units. It
is supposed that individual ALPS/ISUM units have a
slight level of initial imperfections in the form of initial
deflection and residual stresses; buckling mode initial
deflection of plating = 5% of the plate thickness, residual
stress = 5% of the yield stress, column type initial
deflection of stiffeners (plate-stiffener combinations) =
0.15% of the member length.
Figure 8(c). ALPS/ISUM model III (No. of plate units =

126, No. of beam-column units = 111, No. of stiffened
panel units = 19)
Figure 8(a). ALPS/ISUM model I (No. of plate units = 0,

No. of beam-column units = 282, No. of stiffened panel
units = 0)
Figure 8(d). ALPS/ISUM model IV (No. of plate units =
panel units = 35)
Figure 8(b). ALPS/ISUM model II (No. of plate units =

panel units = 0)
Figure 8(e). ALPS/ISUM model V (No. of plate units = 0,
No. of beam-column units = 0, No. of stiffened panel
units = 36)
90

It should however be noted that the above statement
may generally not be true for transversely framed hulls,
including transversely framed barges. Also, greater
deviation in the results shown may be possible if lateral
pressure effects are considered. In such cases, model VI
between two transverse bulkheads may need to be
adopted.
ers
sv
an
Tr
h
ulk
eb
ea
ds
Figure 8(f). ALPS/ISUM model VI (No. of plate units =

84, No. of beam-column units = 0, No. of stiffened panel
units = 288)
10
Vertical moment103 (MNm)
Hog
0
Sag
1
-5
1
2
3
4
5
6
6
5
2 4
-10
-3
-2
-1
: Model I
: Model II
: Model III
: Model IV
: Model V
: Model VI
2
PROGRESSIVE COLLAPSE CHARACTERISTICS

OF TYPICAL MERCHANT SHIPS
In this section, the characteristics of progressive
collapse behavior of merchant ships under vertical
sagging or hogging are investigated using the ISUM
theory. A number of typical ship type designs (10) with
features judged to be typical of such structures are
studied, namely
- Single hull tanker
- Double hull tanker with one center-longitudinal
bulkhead
- Double hull tanker with two longitudinal
bulkheads
- Single sided bulk carrier
- Double sided bulk carrier
- 3,500 TEU container vessel
- FPSO
- Shuttle tanker
Curvature10-7 (1/mm)
Figure 9. Progressive collapse behavior of a 105,000

DWT double hull tanker hull with one centerlongitudinal bulkhead under vertical bending moment, as
obtained by the six types of modeling methods
Figure 9 compares the progressive collapse behavior
of the ship hull under vertical bending moment, as
obtained by the six models. It is observed that the results
obtained from the various types of structural modeling
considered are similar except Model I in sagging. In this
regard, the simpler model extending between two
adjacent transverse web frames may usually be
appropriate for the progressive collapse analysis of
ships hulls. The ultimate strength results obtained by
Model I in sagging tend to be pessimistic compared with
those of the other modeling methods. This may be due to
the fact that the plate-stiffener combination models are
not very relevant to represent the collapse behavior of
deck panels in compression, while they may be
appropriate to heavier bottom structures. It is apparent
that Model II more accurately represents the vertical
bending stress distribution at vertical members or
horizontal bending stress distribution at horizontal
members (i.e., deck or bottom panels).
L = 313.0 m
B = 48.2 m
D = 25.2 m
F.S. = 5.1 m
Figure 10(a). Schematic representation of mid-ship

section of a 254,000 DWT single hull tanker
L = 233.0 m
B = 42.0 m
D = 21.3 m
F.S. = 4.12 m
Figure 10(b). Schematic representation of mid-ship

section of a 105,000 DWT double hull tanker with one
center-longitudinal bulkhead
91
L = 315.0 m
B = 58.0 m
D = 30.3 m
F.S. = 5.12 m
Figure 10(c). Schematic representation of mid-ship

section of a 313,000 DWT double hull tanker with two
longitudinal bulkheads
L = 282.0 m
B = 50.0 m
D = 26.7 m
F.S.
Deck = 5.22 m
Side shell = 0.87 m
Bottom = 2.16 m
Figure 10(d). Schematic representation of mid-ship

section of a 170,000 DWT single sided bulk carrier
L = 273.0 m
B = 44.5 m
D = 23.0 m
F.S.
Deck = 5.16 m
Side shell = 0.86 m
Bottom = 2.58 m
Figure 10(e). Schematic representation of mid-ship

section of a 169,000 DWT double sided bulk carrier
L = 230.0 m
B = 32.2 m
D = 21.5 m
F.S. = 3.27 m
Figure 10(f). Schematic representation of mid-ship

section of a 3,500 TEU container vessel
92
L = 258.0 m
B = 40.0 m
D = 24.2 m
F.S. = 3.62 m
Figure 10(g). Schematic representation of mid-ship

L = 305.0 m
B = 45.3 m
D = 27.0 m
F.S. = 3.27 m
Figure 10(h). Schematic representation of mid-ship

L = 230.6 m
B = 41.8 m
D = 22.9 m
F.S. = 3.5 m
Figure 10(i). Schematic representation of mid-ship

section of a 113,000 DWT FPSO (floating, production,
storage and offloading unit)
L = 254.0 m
B = 46.0 m
D = 22.6 m
F.S. = 3.6 m
Figure 10(j). Schematic representation of mid-ship

section of a 165,000 DWT shuttle tanker
Table 1. Hull sectional properties of the 10 typical merchant ships
Notes: SHT = single hull tanker, DHT#1 = double hull tanker with one center-longitudinal bulkhead, DHT#2 = double
hull tanker with two longitudinal bulkheads, Bulk#1 = single sided bulk carrier, Bulk#2 = double sided bulk carrier,
Cont#1 = 3500 TEU container vessel, Cont#2 = 5500 TEU container vessel, Cont#3 = 9000 TEU container vessel, FPSO
= floating, production, storage and offloading system, Shuttle = shuttle tanker, I = moment of inertia, Z = section
modulus, Y = yield stress, M p = fully plastic bending moment.
Table 1 indicates the principal dimensions of the ten
ships. Figure 10 shows schematic representations of the
mid-ship sections of all ships considered. It is evident
that the ship structural characteristics vary significantly
depending on the cargo types or missions, among other
factors.
In the ALPS/ISUM calculations, some important
influential parameters on the ultimate strength of ships
under vertical moment are varied, namely level of initial
imperfections, lateral pressure and horizontal moment. It
is considered in the calculations that individual structural
units have fabrication related initial imperfections (weld
distortions and residual stresses). The longitudinal
stiffeners have initial imperfections which are considered
to be w osx = 0.0015a and rsx = 0.0 , where w osx =
maximum initial deflection of longitudinal stiffeners, a
= length of the stiffener, rsx = residual stress of the
stiffener. For plating between longitudinal stiffeners, the
level of initial imperfections is varied at the two types
(slight and average levels), suggested by Smith et al.
(1988) as follows
Slight level: w opl = 0.025 2 t , rcx = 0.05 Y
Average level: w opl = 0.1 2 t , rcx = 0.15 Y
In the ALPS/ISUM computations, deck or bottom

stiffened panels as well as vertical members (i.e., side
shells and longitudinal bulkheads) are modeled by the
plate-stiffener separation models as assemblies of the
ALPS/ISUM rectangular plate units and the ALPS/
ISUM beam-column units, the latter being used without
attached plating, as shown in Fig.3(b). This modeling
method more accurately represents the vertical bending
stress distribution at vertical members or horizontal
bending stress distribution at horizontal members (i.e.,
deck or bottom panels), while plating between
longitudinal support members in typical merchant ship
structures may normally not fail prior to longitudinal
support members.
Progressive collapse behavior under vertical moment
Figure 11 represents the progressive collapse
behavior of the considered ship hulls under vertical
hogging or sagging moment, varying the level of initial
imperfections. Some selected typical failure events are
represented in the figures.
As shown in Figs.11(a), (b) and (c), the collapse of
the compression flange of the tanker hulls takes place
prior to the yielding of the tension flange as in design of
usual ship structures. The initial imperfections
significantly affect the progressive collapse behavior of
the ship hulls. Also, there is still some residual strength
93
Vertical moment 103 (MNm)
20
10
-10
-20
For hogging:
1. Buckling collapse of lower longitudinal bulkhead
longl.*
2. Buckling collapse of lower side shell longl.*
3. Buckling collapse of bottom longl.
4. Buckling collapse of bottom girder longl.*
5. Buckling collapse of keel plates
6. Buckling collapse of center girder longl.*
7. Buckling collapse of lower side shell plate*,
bottom girder plates* & bottom girder longl.
8. Buckling collapse of bottom plates
9. Buckling collapse of lower longitudinal bulkhead
plates & bottom girder plates*
10. Ultimate limit state
Note: * denotes that the related failure event starts.
Mt=-13.618 103 MNm
-2
1918
Mt=13.618 103 MNm
1
2
2
1
Level of initial imperfections:
1 : Slight
2 : Average
For sagging:
11. Buckling collapse of upper longitudinal
bulkhead longl.*
12. Buckling collapse of upper side shell longl.*
13. Buckling collapse of deck girder longl.*
& center girder longl.*
14. Buckling collapse of deck girder longl.*
& center girder longl.
15. Buckling collapse of deck longl.*
16. Buckling collapse of deck longl.
17. Buckling collapse of center girder plates*,
deck plates*, yielding of bottom keel plates
& center girder longl.*
18. Buckling collapse of side shell plates*
15
16
17
-1
9 10
11
12
13
14
2
1
-3
Curvature
10-7
(1/mm)
Figure 11(a). Progressive collapse behavior of the 254,000 DWT single hull tanker under vertical moment varying the
level of initial imperfections, as obtained by ALPS/ISUM
even after buckling collapse of the compression flange.
This is due to a shift of the neutral axis towards the
tension flange, resulting from loss of effectiveness of the
collapsed compression flange as shown in Fig.12. Of
interest, as the bending moment increases, the neutral
axial position changes quickly and becomes stable, as
shown in Fig.12. This is because the neutral axis is
calculated for partially effective hull cross-section after
the bending moment is applied, while it is estimated for
fully effective cross-section before loading. This implies
that the section moduli calculated for fully effective hull
cross-section may not always be a real indication of the
ship hull sectional load resistive properties. The ultimate
hogging moment of the tanker hull is larger than the
ultimate sagging moment as usual.
In bulk carriers, the spacing of transverse frame (or
floor) at the bottom part is different from that at deck or
at side shells. Figs.11(d) and (e) represent the
94
progressive collapse behavior of the bulk carrier hulls

under vertical moments, varying the level of initial
imperfections. In contrast to the tanker hulls described
above, the tension flange (i.e., bottom plates) of the bulk
carrier hull under sagging moment yields prior to
buckling collapse of the compression flange (i.e., deck
plates). In hogging condition, however, buckling
collapse of the compression flange (i.e., bottom plates)
takes place prior to yielding of the tension flange (i.e.,
deck plates). This is because the deck panels of bulk
carrier structures are typically much sturdier than bottom
panels. Regardless of this, the section modulus at bottom
is of course much larger than that at deck because bulk
carriers have large deck openings. It is however less
consistent with the normally expected ultimate strength
characteristics of usual ship designs since the ultimate
hogging moment of bulk carriers is smaller than the
ultimate sagging moment.
10
For hogging:
1. Buckling collapse of outer bottom longl.*
& bilge keel
2. Buckling collapse of outer bottom longl.
& lower side longl.*
3. Buckling collapse of lower sloping longl.*
& inner bottom longl.
& lower side longl.
5. Buckling collapse of center girder longl.
& lower sloping longl.
6. Buckling collapse of bottom girder plates*
& outer bottom plates*
7. Buckling collapse of outer bottom plates
& yielding of deck longl.*
8. Buckling collapse of center girder plates*
& yielding of deck plates*
9. Buckling collapse of bottom girder plates
& yielding of deck longl.
10. Yielding of deck plates
11. Buckling collapse of center girder plates
& inner bottom plates ( Ultimate limit state)
-5
17 16
Mt=6.240 103 MNm

1 : Slight
2 : Average
For sagging:
12. Buckling collapse of side longl.
between deck and 2nd deck* & deck longl.*
14. Buckling collapse of side longl.
between deck and 2nd deck & deck plates*
15. Buckling collapse of side shell
between deck and 2nd deck* & deck plates
17. Buckling collapse of side shell
between deck and 2nd deck
14
-2
1
9 10 11
15
-3
Mt=-6.240 103 MNm
-10
12
13
-1
Curvature
10-7
(1/mm)
Figure 11(b). Progressive collapse behavior of the 105,000 DWT double hull tanker with one center-longitudinal
bulkhead under vertical moment varying the level of initial imperfections, as obtained by ALPS/ISUM
30
20
10
0
-10
For hogging:
& inner bottom longl.*
4. Buckling collapse of bottom girder longl.
& lower side shell longl.*
5. Buckling collapse of lower sloping tank longl.*
& lower longitudinal bulkhead longl.*
& yielding of deck longl.*
7. Buckling collapse of outer bottom plates*,
yielding of deck plates* & upper longitudinal
bulkhead plates*
& yielding of deck longl.
9. Buckling collapse of bottom girder plates,
yielding of deck plates & yielding of upper side
10. Buckling collapse of inner bottom plates*
( Ultimate limit state)
2
1
11
Mt=-16.489 103 MNm
-30
-3
-2
-1
1
2

1 : Slight
2 : Average
11. Buckling collapse of upper inner side shell
longl.*
12. Buckling collapse of lower longitudinal
bulkhead longl.*
13. Buckling collapse of deck girder longl.
14. Buckling collapse of upper outer shell longl.
15. Buckling collapse of upper inner side shell longl.
& deck plates*
16. Buckling collapse of deck plates & longitudinal
bulkhead plates*
17. Buckling collapse of deck plates & upper
inner/outer shell plates* (Ultimate limit state)
15
17 16
-20
Mt=16.489 103 MNm
For sagging:
12
13
14
2
1
4
3
9 10
78
Curvature
10-7
(1/mm)
Figure 11(c). Progressive collapse behavior of the 313,000 DWT double hull tanker with two longitudinal bulkheads
under vertical moment varying the level of initial imperfections, as obtained by ALPS/ISUM
95
20
For hogging:
& center/side girder plates
& lower side shell longl.*
7. Buckling collapse of lower side shell plates*
9. Buckling collapse of inner bottom plates
& lower side shell long.
10. Yielding of deck plates * ( Ultimate limit state)
10
2
1
11
15 14
Mt=-11.334
-20
-3
12
13
MNm
-2
For sagging:
11. Buckling collapse of upper sloping longl.*,
upper side longl.*
12. Buckling collapse of deck longl. *
14. Buckling collapse of upper side longl.,
deck plates*, upper sloping plates*
& yielding of bottom girder longl.*
15. Buckling collapse of deck plates
& upper side shell plates*
(Ultimate limit state)
103

1 : Slight
2 : Average
-10
4
3
67
5
Mt= 11.334 103 MNm

10
89
-1
Curvature
10-7
(1/mm)
Figure 11(d). Progressive collapse behavior of the 170,000 DWT single sided bulk carrier under vertical moment varying
the level of initial imperfections, as obtained by ALPS/ISUM
20
10
For hogging:
1. Buckling collapse of bilge keel
2. Buckling collapse of inner bottom longl.*
4. Buckling collapse of center girder plate,
bottom girder plates* & outer bottom longl.*
7. Buckling collapse of outer bottom plates*,
inner bottom plates, bottom girder plates*
& outer bottom longl.
8. Buckling collapse of outer bottom plates,
lower sloping plates*
2
1
Mt=9.467 103 MNm
7
56
4
2

1 : Slight
2 : Average
For sagging:
10. Buckling collapse of upper side shell longl.*
11. Buckling collapse of upper sloping longl.*
12. Buckling collapse of upper sloping plates*
& deck longl.*
13. Buckling collapse of deck plates*, deck longl.
& Yielding of bottom girder longl.*
14. Buckling collapse of deck plates,
upper horizontal stringers
15. Buckling collapse of upper sloping longl.
10
11
12
151413
Mt=-9.467 103 MNm
-3
-2
-1
1
2
-10
-20
89
Curvature
10-7
(1/mm)
Figure 11(e). Progressive collapse behavior of the 169,000 DWT double sided bulk carrier under vertical moment
varying the level of initial imperfections, as obtained by ALPS/ISUM
96
10
For hogging:
Mt=4.193 103 MNm

& outer bottom longl.*
& center girder longl.
under 4th deck
6. Buckling collapse of inner bottom longl.
& center girder plates*
8. Buckling collapse of bottom girder longl.,
Inner/outer bottom plates & lower side shell plates*
9. Buckling collapse of center girder plates,
4th deck longl., bottom girder plates
& yielding of upper deck longl./plates
10. Yielding of upper side shell plates*
23
10
1
2

1 : Slight
2 : Average
For sagging:
11. Buckling collapse of upper deck plates
& inner/outer side shell plates*
12. Buckling collapse of upper deck longl.
13. Buckling collapse of 2nd deck longl.
& inner/outer side shell longl.*
14. Buckling collapse of inner/outer side shell plates
between upper deck and 2nd deck
-5
11
12
14 13
Mt=-4.193 103 MNm
-3
89
2
1
-10
6 7
-2
-1
Figure 11(f). Progressive collapse behavior of the 3,500 TEU container vessel under vertical moment varying the level of
initial imperfections, as obtained by ALPS/ISUM
15
10
5
0
-5
For hogging:
Mt=6.399 103 MNm
1. Buckling collapse of outer bottom longl.*

& bottom girder plates*
6. Buckling collapse of inner bottom longl.,
yielding of upper deck longl. & side shell
longl./plates between upper deck and 2nd deck
7. Buckling collapse of inner bottom longl.
& yielding of side shell longl.
10. Yielding of side shell plates
10
1
2

1 : Slight
2 : Average
For sagging:
11. Buckling collapse of side shell longl.

between upper deck and 2nd deck*
& side shell plates between upper deck and 2nd deck*
14. Buckling collapse of upper deck plates
between upper deck and second deck
17. Buckling collapse of side shell plates
11
12
13
17 16 15 14
-10
Mt=-6.399 103 MNm
-3
89
2
1
-15
67
5
-2
-1
Curvature
10-7
(1/mm)
Figure 11(g). Progressive collapse behavior of the 5,500 TEU container vessel under vertical moment varying the level of
97
20
For hogging:
& yielding of side shell longl.
7. Yielding of upper deck longl.
8. Yielding of upper deck plates
10
Mt=10.705 103 MNm

6
13
12
Mt=-10.705 103 MNm
-20
-3
-2
1
2

1 : Slight
2 : Average
For sagging:
9. Buckling collapse of side shell plates
between upper deck and 2nd deck*,
upper deck plates & yielding of bottom girder
plates*
13. Yielding of outer bottom longl.
& outer bottom plates (Ultimate limit state)
14. Yielding of side shell plates
between upper deck and second deck
2
1
14
-10
7 8
9
10
11
-1
Curvature
10-7
(1/mm)
Figure 11(h). Progressive collapse behavior of the 9,000 TEU container vessel under vertical moment varying the level of
10
For hogging:
1. Buckling collapse of outer bottom longl.,
bilge keel & center girder longl.*
lower side longl.* & bottom girder longl.*
8. Buckling collapse of inner bottom plates,
center girder plates & yielding of deck longl.
3
2
5
67
89
Mt=6.056 103 MNm

1 : Slight
2 : Average
For sagging:
-5
17 16 15

& upper side longitudinal bulkhead*
12. Buckling collapse of upper side longl.*
13. Buckling collapse of upper side shell
bulkhead* & deck plates*
& upper side shell*
15. Yielding of bottom girder longl.*
17. Yielding of bottom girder longl.
10
11
12
13
14
Mt=-6.056 103 MNm
-10
-3
-2
-1
Curvature
10-7
(1/mm)
Figure 11(i). Progressive collapse behavior of the FPSO hull under vertical moment varying the level of initial
imperfections, as obtained by ALPS/ISUM
98
15
For hogging:
1. Buckling collapse of bilge keel & outer bottom
longl.
2. Buckling collapse of lower side longl.*
6. Buckling collapse of lower side longl.
7. Buckling collapse of outer bottom plates,
center girder plates* & bottom girder longl.
9. Yielding of deck longl.*
10. Yielding of deck plates*
10
9 10
78
4
3
Mt=8.300 103 MNm
1
2
2
1

1 : Slight
2 : Average
-5
2
1
-10
15 14
Mt=-8.300
-15
For sagging:
11
12
103
-3
-2

12. Buckling collapse of side longl. between deck
and second deck*
13. Buckling collapse of deck plates* & side longl.
between deck and second deck
14. Buckling collapse of deck plates & yielding of
bottom girder longl.*
13
MNm
-1
Curvature
10-7
(1/mm)
Figure 11(j). Progressive collapse behavior of the shuttle tanker hull under vertical moment varying the level of initial
imperfections, as obtained by ALPS/ISUM
Height to neutral axis from base line (mm)
24000
Figures 11(i) and (j) show the progressive collapse

behavior of the FPSO and shuttle tanker under vertical
moment. It is seen that the progressive collapse
characteristics of these ships are very similar to the usual
tanker hulls.

1 : Slight
2 : Average
20000
2
1
16000
12
12000
4 5 6
11 12 13 14
78
10
15 16
17
18 19
8000
1
2
4000
0
0
Figure 12. Variation of the neutral axis due to structural

failure for the single hull tanker, as obtained by
ALPS/ISUM (For explanation of numerals, see
Fig.11(a))
Figures 11(f), (g) and (h) represent the progressive
collapse behavior of the container vessel hulls under
vertical moments, varying the level of initial
imperfections. In contrast to the usual behavior of tanker
structures, it is observed that the deck panels under axial
compressive loads in sagging condition do not buckle
and reach the ultimate strength by gross yielding. This is
because the deck panels are very stocky with large plate
thickness.
Effect of lateral pressure

The effect of lateral pressure on the ultimate hull
girder strength is now investigated for selected ship hulls
under vertical moment. The ALPS/ISUM models
previously used for progressive hull girder collapse
analysis without lateral pressure are employed. The
external water pressure considered is calculated as a sum
of static and hydrodynamic pressure components for
head sea in the heavy ballast condition (ABS 2000). The
effects of water pressure inside the ballast tanks are not
considered for convenience of the structural modeling.
Figures 13(a) to (c) show hypothetical
representations of the applied water pressure
distributions for head sea state in the heavy ballast
condition of the 313,000 DWT double hull tanker with
two longitudinal bulkheads, the 170,000 DWT single
sided bulk carrier and the 9,000 TEU container vessel,
respectively.
In the present ALPS/ISUM calculations, the initial
imperfections for plating are fixed at the average level,
while those for stiffeners are the same as defined above.
It is assumed that the lateral pressure is applied to the
individual structural members in the same direction to
the initial deflections.
99
W.L.
30
Hogging
20
With water pressure
10
0
-10
Figure 13(a). Schematic of water pressure distribution

for the 313,000 DWT double hull tanker with two
longitudinal bulkheads, being a sum of static and
hydrodynamic pressure for head sea state
W.L.
Without water pressure
With water pressure
-20
Double hull tanker

with two longitudinal bulkheads
Sagging
-30
-3
-2
-1
Figure 14(a). Progressive collapse behavior of the

313,000 DWT double hull tanker with two longitudinal
bulkheads under vertical moment varying the magnitude
of water pressure, as obtained by ALPS/ISUM
20
Single sided bulk carrier
Hogging
Figure 13(b). Schematic of water pressure distribution

for the 170,000 DWT single sided bulk carrier, being a
sum of static and hydrodynamic pressure for head sea
state

10
With water pressure
0
With water pressure
-10
Sagging
-20
-3
W.L.
-2
-1
100
9,000 TEU container
Hogging
Figures 14(a) to (c) show the progressive collapse

behavior of the three ship hulls under vertical moments.
Figure 14(d) represents the variation of the ultimate hull
girder strengths as a function of the water pressure
magnitude. The effectiveness of plates under compression is normally further reduced by lateral pressure
loading. In the case of the hull girder, this is primarily
due to the shear lag effect. When the amount of lateral
pressure is large, therefore, the ultimate strength of
ships hulls can be possibly smaller than that without
lateral pressure. In calculating the ultimate capacity of
ship hulls, the lateral pressure related shear lag effect can
approximately be taken into account so that the ultimate
stress or the effectiveness (i.e., effective width/breadth)
of individual structural members is predicted considering
lateral pressure as another load component.
Figure 14(b). Progressive collapse behavior of the

170,000 DWT single sided bulk carrier under vertical
moment varying the magnitude of water pressure, as
obtained by ALPS/ISUM
20
Figure 13(c). Schematic of water pressure distribution

for the 9,000 TEU container vessel, being a sum of static
and hydrodynamic pressure for head sea state
10
With water pressure
With water pressure
-10
Sagging
-20
-3
-2
-1
Figure 14(c). Progressive collapse behavior of the 9,000

TEU container vessel under vertical moment varying the
magnitude of water pressure, as obtained by
ALPS/ISUM
1.1
30
V, H = angles of cross sectional plane
Hogging
1.0
Sagging
Hogging
0.9
Muo = Ultimate moment without lateral pressure
: Double hull tanker with

two side-longitudinal bulkheads
: Single sided bulk carrier
: 9,000 TEU container
0.7
0.6
0
50
Water pressure level (%)
Figure 14(d). Variation of the ultimate hull girder

strengths as a function of the magnitude of water
pressure, as obtained by ALPS/ISUM
The effect of lateral water pressure on the ultimate
hull girder capacity under hogging moment may not be
neglected specifically for double hull tankers and bulk
carriers, while that under sagging moment may normally
be small. This is because seawater is applied mainly to
ship bottom or lower side structures which are
compressed in hogging. For different loading conditions,
the distributions of internal pressure as well as external
pressure may vary so that their effects on the progressive
collapse behavior of ship hulls are of course different.
Again, it is noted that the effect of water pressure on the
ultimate hogging moments may be different for different
ship types.
Effect of horizontal moment
The effect of horizontal moment on the ultimate hull
girder strength is now investigated for selected ship hulls.
The initial imperfections for plating are fixed at the
average level, while those for stiffeners are the same as
defined above. The water pressure is not applied. Figures
15(a) to (c) show the progressive collapse behavior of
the 313,000 DWT double hull tanker with two
longitudinal bulkheads, the 170,000 DWT single sided
bulk carrier and the 9,000 TEU container vessel,
respectively. Figure 15(d) represents the ultimate hull
girder interaction relationship between vertical and
horizontal moments. It is evident that the effect of
horizontal moment on the ultimate hull girder strength is
of significance. It is noted that the horizontal moment is
typically not the maximum when the vertical moment is
the maximum and thus a relevant consideration for load
combination is necessary in performing a design check
using the results of Figure 15(d).
1
2
3
10
4
5
0
-10
H
=0.0
V
H
=0.25
V
H
=0.5
V
5
4
3
2
1
2
3
-20
-3
H
=0.8
V
H
=1.0
V
4
5
Double hull tanker

with two longitudinal bulkhead
Sagging
-30
100
-2
-1
Figure 15(a). Progressive collapse behavior of the

313,000 DWT double hull tanker with two longitudinal
bulkheads under combined vertical and horizontal
moments, as obtained by ALPS/ISUM
20
Hogging
for vertical or horizontal moments
0.8
20
1
2
3
10
2
1
-10
V
= 0.0
V
H
= 0.25
V
H
= 0.5
V
5
4
3
Sagging
-3
4
5
H
= 0.8
V
H
= 1.0
V
Single sided bulk carrier
-20
-2
-1
Figure 15(b). Progressive collapse behavior of the

170,000 DWT single sided bulk carrier under combined
vertical and horizontal moments, as obtained by
ALPS/ISUM
20
Hogging
Mu
Muo
1
2
10
3
4
5
4
H
H
= 0.0 4
= 0.8
V
V
H
H
2
= 0.25 5
= 1.0
V
V
H
3
= 0.5
V
-10
2
1
Sagging
-20
-3
9,000 TEU container

-2
-1
Figure 15(c). Progressive collapse behavior of the 9,000

TEU container vessel under combined vertical and
horizontal moments, as obtained by ALPS/ISUM
101

1.5
: Double hull tanker with

two longitudinal bulkheads
: 9,000 TEU container
1.0
MV / MVu
0.5
1.85
MV
M Vu
0.0
+
-
Tens.
Comp.
Ux
xY
D-gus
M
+ H =1
M Hu
-0.5
E
x
0.0
0.2
0.4
0.6
MH / MHu
0.8
+
gus
+
-
Tens.
Comp.
1.0
CLOSED-FORM ULTIMATE STRENGTH

FORMULATION FOR SHIP HULLS
Figure 16 shows examples of the variation of the
hull girder longitudinal stress distribution during the
progressive collapse. It is generally observed that the
ship reaches its ultimate limit state if both collapse of the
flange and side shell in compression and yielding of the
other flange in tension occur. In this situation, the side
shell in the vicinity of the neutral axis can still remain in
a linear elastic state. This is due to the fact that with an
increase in the bending moment the internal stress
(strength) of the collapsed parts is reduced even before
reaching the ultimate hull strength, while the yielded
parts can not sustain further increase of applied loads
and the stresses remain unchanged. It should be noted
that first yielding could in some cases occur after the
buckling of the compression flange, depending on
specifics of the design.
(a)
(b)
Figure 16. Variation of the longitudinal stress
distribution during the progressive collapse under
hogging moment (+: Tension, -: Compression), as
obtained by ALPS/ISUM (a) pre-ultimate limit regime
(b) ultimate limit state
xE
guh
Ux
xY
Figure 15(d). Ultimate hull girder strength interaction

relationships between vertical and horizontal moments,
as obtained by ALPS/ISUM
102
D-guh
Mean = 0.971
COV = 0.164
-1.5
xE
E
x
-1.0
(a) Sagging
(b) Hogging
Figure 17. Longitudinal stress distribution over a ships
cross-section at the overall collapse state as suggested by
Paik & Mansour (1995)
It has been recognized that although the overall
collapse of a ships hull under vertical bending moment
is initiated and governed by collapse of the compression
flange, there is still some reserve strength beyond
collapse of the compression flange. This is because after
buckling of the compression flange occurs the neutral
axis of the hull cross-section moves toward the tension
flange and a further increase of the applied bending
moment is normally sustained until the tension flange
yields. At later stages of this process, side shell platings
around the compression and the tension flanges will also
fail. Therefore, the pioneering suggestion of Caldwell
(1965) for the longitudinal stress distribution at the hull
girder ultimate limit state typically overestimates the
ultimate bending capacity of a ship hull.
Paik & Mansour (1995) have made a more refined
suggestion for the longitudinal stress distribution over a
ships cross-section at the state of overall collapse as that
shown in Fig.17. The longitudinal stress distribution
shown in Fig.17 resembles that of Fig.16(b). As may be
seen from Figs.16(b) and 17, the compression flange has
collapsed and the tension flange has yielded at the
moment the ultimate strength is reached, but the side
shell in the vicinity of the neutral axis is still intact
(linear elastic).
The longitudinal axial strain, xi , of the i th
longitudinal strength member in the effective hull crosssection, which is assumed to remain plane, may be given
by one of the following two equations which are in fact
identical, namely
xi =
xi =
z
1 i
E i g e
E i (D g e )
(zi g e )
(1a)
(1b)
where g e = neutral axis of the effective hull crosssection from the base line, B , D = axial stresses at
bottom or deck plating, E i , xi = Youngs modulus and
axial strain of the i th longitudinal strength member,
D = ship depth, z i = coordinate in the ship depth

direction from the base line to the central axis of the i th
member.
Once the axial strain is determined, the longitudinal
axial stress, xi , of the i th longitudinal strength
member may be obtained as follows
xi = E i xi for xi Y
M uh =
(2)
It is assumed that overall collapse of a ships hull

girder takes place when the axial stress at the tension
flange (i.e., either bottom plating in hogging or deck
plating in sagging) reaches the material yield stress,
while the region in compression has collapsed (or vice
versa). The longitudinal stress distribution at the ultimate
limit state may be divided into four regions, namely (1)
yielded region, (2) elastic tension region, (3) elastic
compression region, and (4) collapsed compression
region, as that shown in Fig.17.
In the sagging condition as shown in Fig.17(a), the
neutral axis, g us , above the base line at the ultimate
limit state can be calculated as follows
g us =
when D just reaches the equivalent yield stress of the

deck stiffened panels.
The ultimate hull girder hogging moment, M uh , can
then be calculated as follows
E
A ek z k + 4 xlU A el z l
1 xiY Ai zi +2 xjE A jz j +3 xk
E
A ek + 4 xlU A el
1 xiY Ai +2 xjE A j +3 xk
(3a)
M us =
Y
xi
E
xk
A (g z )
) + A (z g ) (3b)
A i (g us z i )
A ek (z k g us
E
xj
U
xl
us
el
us
where g us = as defined in Eq.(3a).

Similarly, in the hogging condition as shown in
Fig.17(b), the neutral axis position above the base line at
the ultimate limit state can be calculated as follows
g uh =
A z + A z +
A + A +
1
U
xi
ei i
U
xi
ei
E
xj
ej
E
xj
ej
E
xk
E
xk
A k z k + 4 Yxl A l z l
A k +4 Yxl A l
(4a)
where x is calculated from Eq.(2) together with Eq.1b
A ei (g uh z i )
E
xk
A k (z k g uh
E
xj
Y
xl
A ej g uh z j
A l (z l g uh ) (4b)

: Slight
: Average
0.8
(Mu / Mp)Design formula
summations for members in the regions 1, 2, 3 or 4,

respectively.
The ultimate hull girder sagging moment, M us , can
then be calculated, with the sagging moment or the
compressive stress is taken as negative while the
hogging moment or the tensile stress is taken as positive
as follows
) +
U
xi
where g uh = as defined in Eq.(4a).

In calculating Eqs.(4) or (5), the longitudinal axis
stress for individual structural elements (i.e., either
plating or stiffeners) must be taken as a value smaller
than either the ultimate stress or the material yield stress,
the former being predicted using the closed-form
expressions for stiffeners (with or without attached
plating) or those for plating between stiffeners (Paik &
Thayamballi 2002).
Figure 18 plots the correlation between ALPS/ISUM
results and the closed-form predictions of the ultimate
bending moments for 10 typical commercial ships
indicated in Figure 10. The mean and COV of the
present closed-form expression predictions against the
ALPS/ISUM progressive collapse analyses for ship hulls
with initial imperfections at an average level are 0.994
and 0.087, respectively.
1.0
where x is calculated from Eq.(2) together with

Eq.(1a) when B just reaches the equivalent yield stress
of bottom stiffened panels. The superscripts, Y , E and
U , represent the yield stress, elastic stress and ultimate
stress, respectively. The subscript, e , indicates the
effective section. 1 ( ) , 2 ( ) , 3 ( ) or 4 ( ) are
0.6
0.4
0.2
Slight level: Mean=0.947, COV=0.085
Average level: Mean=0.986, COV=0.059
All data: Mean=0.970, COV=0.072
0.0
0.0
0.2
0.4
0.6
(Mu / Mp)ALPS/ISUM
0.8
1.0
Figure 18. Correlation between ALPS/ISUM progressive

collapse analyses and the closed-form expression
predictions varying the level of initial imperfections
ULTIMATE LIMIT STATE DESIGN FORMAT
Steel plated structures are likely to be subjected to
various types of loads and deformations, which may in
some cases be extreme and accidental. The mission of a
structural designer is to design a particular structure that
can withstand such loads and deformations throughout
its expected service life.
In design, the structure is required to have an
103

adequate margin of safety because of possible overloading or accidental loading above the service loads,
which can arise from changing the planned use, and
because of uncertainties in evaluation and control of
loads or load effects (e.g., stress, deformation) and
structural resistance (capacity) as well as variations in
building procedures. The partial safety factor concept
based design criterion of a structure under multiple types
of loads is typically expressed as follows
Dd < Cd
where
Dd = o
(5a)
D (F
, fi )
design
capacity,
ki
ki
design
demand,
Cd =
Ck
M
D ki (Fki , fi )
characteristic measure of demand for load type i ,

calculated from the characteristic measures of loads, Fk ,
and weighed by the partial safety factor, f , taking
account of the uncertainties related to loads, o = partial
safety factor taking into account the degree of
seriousness of the particular limit state in regard to safety
and serviceability accounting of economical and social
consequences as well as any special circumstances (e.g.,
purpose of the ship, type of cargo, interaction of the limit
state considered with the others), C k = characteristic
measure of capacity, M = m c = capacity related
safety factor, m = partial safety factor taking account
of the uncertainties due to material properties, c =
partial safety factor taking account of the uncertainties in
the capacity of the structure, such as quality of the
construction, corrosion, method considered for
determination of the capacity.
Eq.(5a) may be rewritten as follows
=
Cd
>1
Dd
(5b)
where = safety measure.

In Eqs.(5), the demand is normally associated with
load effects (e.g., stress, deformation) or loss of kinetic
energy, while the capacity means the load-carrying
capacity or the energy absorption capability. A nominal
or characteristic measure of demand or capacity
approximately corresponds to a specified percentage of
the area below the probability curve of the random
variables. The partial safety factors are normally
determined by the most probable design points divided
by the nominal or characteristic values of the random
variables.
It is important to realize that the partial factors in
Eqs.(5) may depend on the design situation and the types
of the limit states. For useful guidelines in determining
the partial safety factors related to limit state design of
steel structure, ISO 2394 (1998), NTS (1998) and ENV
1993-1-1 of Eurocode 3 (1992) may be referred to.
104
To achieve the criterion of Eq.(5), two types of

structural design philosophies have been typically
employed, namely
Allowable stress design
Limit state design
In the allowable stress design, the focus is on

keeping the stresses resulting from the design loads
under a certain working stress level that is usually based
on successful similar past experience. While the
working stress is obtained by a linear elastic response
analysis of the structure, the allowable stress is defined
so as to satisfy the structural safety requirement that the
structure must not fail given normal levels of load,
strength and operational variability. Regulatory bodies or
classification societies usually specify the value of the
allowable stress as some fraction of the mechanical
properties of materials (e.g., uniaxial yield or ultimate
tensile strength).
In contrast to the allowable stress design, the limit
state design is based on the conditions that the structure
ceases to fulfill its intended function. For these
conditions, the applicable strength is estimated and used
in design as a limit for such behavior. For steel structures,
the limit state criterion of a structure is typically written
so that the structural resistance (capacity) must be
greater than the load effects (e.g., working stresses) with
relevant factors.
In limit state design, C k in Eqs.(5) is defined by the
level of loads which can be sustained until the structure
reaches the limit state, while D k is calculated by the
linear elastic response analysis similar to that of the
allowable stress design method. The load-carrying
capacity of a structure is normally evaluated using
simplified design formulations or more refined
computations such as nonlinear elastic-plastic large
deformation finite element analyses with appropriate
modeling related to geometric / material properties,
initial imperfections, boundary condition, load
application, finite element mesh sizes, and so on. The
structural capacity analysis will be undertaken based on
most unfavourable failure modes for the idealized
structure.
It is necessary to ensure that the structure has an
adequate degree of reliability against the ultimate limit
state. Two types of design formats are normally used,
namely (ISO 2394 1998)
Probabilistic design format
Partial factor format
The latter format is typically used for normal design

purposes, while the former is more relevant for specific
design problems or for calibration of the partial safety
factors.
For the ultimate limit state design of steel plated
structures, the basic variables which characterize load
effects, material properties and geometric parameters
should be identified first. Methodologies or simplified

models for computing the load effects and the loadcarrying capacities must be established. Once the two
models, i.e., for calculating both load effects and
ultimate strength are obtained, the ultimate limit state
function, G , can be given from Eq.(5a) as a function of
some basic variables, x1 , x 2 , x n , as follows
G (x1, x 2 ,..., x n ) = 0
(6)
When G 0 , the structure is in the desired state.

The models always have uncertainties due to many
reasons. A computation model is in fact a function of
random variables, namely
Ym = Y(x1, x 2 ,..., x n )
(7)
where Ym = value computed by the model, Y =

function of the model, x i = random variables.
As long as the random variables are uncertain, the
model function is not exact so that Ym may always have
some errors. This is typically due to lack of knowledge
or simplification in developing the model. The exact
solution, Yo , of the problem may be expressed by
Yo = Y* (x1 , x 2 ,..., x n , 1 , 2 ,..., m )
(8)
where i = random variables related to the model

uncertainties, Y* = exact function. In Eq.(8), the
statistical properties of i may normally be determined
from experiments or observations. For the ultimate
strength model, the mean of i is determined as the
average value which correctly predicts the test results.
ASSESSMENT OF SAFETY MEASURE
For the ultimate limit state design of ship hulls, the
characteristic measure of the design capacity, Ck , in
Eqs.(5) is the ultimate hull girder strength while D k is
the characteristic value of the total bending moment. To
approximately take into account the correlation between
still water and wave-induced bending moments, the
following type of equation can be used for calculating
the characteristic value of the total bending moment,
namely
Mt = kswMsw + k wMw
(9)
where k sw and k w are load combination factors for still

water moment, M sw , and wave-induced bending
moment, M w , respectively. These account for the nonsimultaneous occurrence of extreme still water and
wave-induced loads.
To consider dynamic load effects, the total bending
moment may be given by
M t = k sw M sw + k w (M w + k d M d )
(10)
where k d is the load combination factor related to the

dynamic bending moment, M d , arising from either
slamming or whipping. M d is taken as the extreme
dynamic bending moment in the same wave condition
(e.g., sea state) as the wave-induced bending moment
while the effect of ship hull flexibility may be accounted
for in the computation of M d . In very high sea states,
M d is normally ignored because the possibility of
whipping is low. For considering the hull girder effects
of slamming in oceangoing merchant ships, it has been
suggested that M d = 0.15M w may be used for tankers in
sagging, but M d = 0 in hogging (Mansour &
Thayamballi 1994). While external pressure loads
imposed on the ships hull in seaways can be calculated
in terms of sea water heads, the internal pressure loads
must be determined for each fully loaded cargo hold and
ballast tank, as caused by the dominating ship motions
(pitch and roll) and the resulting accelerations.
On the other hand, the traditional allowable stress
design approach may take the section modulus, Z , for
C d and the minimum required section modulus, Z min ,
(specified by IACS or classification societies) for D d .
The ultimate hull girder strength, M u , can be obtained
by the progressive collapse analyses (using
ALPS/ISUM) or from Eqs.(3) or (4). It is important to
note that aging ships may have suffered structural
damages due to corrosion, fatigue cracks or weld region
fracture. For assessment of safety measure of damaged
ships, the effects of such structural damages should of
course be taken into account in predicting M u .
Table 2 indicates safety measure calculations for the
10 typical merchant ships with average level of initial
imperfections, but without structural damage, considered
in the present study. The total bending moment is
calculated from Eq.(9) when k sw = k w = 1.0 , while
M sw and M w are determined from the IACS unified
formula. Z min is also computed from the IACS unified
formula. M u is the ultimate vertical moment of the ship
with average level of initial imperfections as obtained by
ALPS/ISUM. For safety measure calculations based on
the ultimate strength, o = M = 1.0 is adopted in this
regard. When Eqs.(3) or (4) are employed for ultimate
hull girder strength predictions, however, M = 1.15 is
recommended to be used.
As evident from Table 2, the safety measure based
on section modulus (allowable stress design approach)
has greater margin at bottom than at deck for all of the
ships considered. In some ships such as bulk carriers and
container ships, however, the safety measure based on
ultimate hull strength (ultimate limit state design
approach) has less margin in hogging than in sagging.
105

Table 2 Safety measure calculations for the 10 typical merchant ships
Item
Z
(m3)
Z min
(m3)
Z
Z min
M sw
(GNm)
Mw
(GNm)
Mt
(GNm)
Mu
(GNm)
Mu
Mt
Deck
Bottom
Deck
Bottom
SHT
66.301
70.950
60.699
60.699
DHT#1
29.679
39.126
27.814
27.814
DHT#2
77.236
103.773
73.494
73.494
Bulk#1
44.354
62.058
44.040
50.516
Bulk#2
39.274
50.544
38.950
42.196
Cont#1
18.334
27.228
17.252
18.689
Cont#2
26.635
42.894
26.327
28.521
Cont#3
44.376
58.785
44.042
47.712
FPSO
31.040
38.520
26.991
26.991
Shuttle
43.191
49.175
36.992
36.992
Deck
Bottom
1.092
1.169
1.067
1.407
1.051
1.412
1.007
1.228
1.008
1.198
1.063
1.457
1.012
1.504
1.008
1.232
1.150
1.427
1.168
1.329
Sag
Hog
-5.058
5.584
-2.318
2.559
-6.125
6.815
-4.210
4.673
-3.516
3.868
-1.557
1.943
-2.377
3.162
-3.976
5.107
-2.249
2.488
-3.083
3.409
Sag
Hog
-8.560
8.034
-3.923
3.682
-10.365
9.674
-7.124
6.661
-5.951
5.599
-2.636
2.250
-4.022
3.237
-6.729
5.597
-3.806
3.568
-5.217
4.891
Sag
Hog
-13.618
13.618
-6.240
6.240
-16.489
16.489
-11.334
11.334
-9.467
9.467
-4.193
4.193
-6.399
6.399
-10.705
10.705
-6.056
6.056
-8.300
8.300
Sag
Hog
-16.767
15.826
-6.899
8.485
-19.136
23.566
-14.281
14.434
-12.165
12.027
-6.800
5.953
-9.571
9.049
-16.599
13.075
-7.282
8.760
-11.280
11.404
Sag
Hog
1.231
1.162
1.106
1.360
1.161
1.429
1.260
1.274
1.285
1.270
1.622
1.420
1.496
1.414
1.551
1.221
1.202
1.446
1.359
1.374
Notes: Z min = minimum required section modulus specified by IACS, M t = M sw + M w , M u = ultimate vertical
moment of ship hulls with average level of initial imperfections, but without structural damage, as obtained by
ALPS/ISUM.
As previously discussed, this is because in bulk
carriers or container ships the ultimate hogging moment
is not greater than the ultimate sagging moment even if
the section modulus at bottom is larger than that at deck.
This is an unusual consequence of structural failure
event. In sagging, the tension flange (i.e., bottom panels)
yields prior to buckling collapse of the compression
flange (i.e., deck panels). In hogging, however, the
compression flange (i.e., bottom panels) collapses much
earlier than yielding of the tension flange (i.e., deck
panels).
This is in fact in contrast to the expectation of the
ship structural designers who would employ the
traditional design methodology based on the allowable
stress. As long as the section modulus at bottom is
greater than that at deck, they might have presumed that
the ultimate hull girder strength in hogging will be
greater than that in sagging. But this is not always true.
When the ultimate hogging moment is marginal or
overlooked at the preliminary design stage, the bulk
carrier is likely to collapse and sink if a forward cargo
hold is flooded so that the hogging moment is amplified.
This indicates the disadvantage of the traditional
structural design procedures for ships based on the
allowable stress and/or the sectional moduli. The
ultimate limit state design procedure can avoid such a
problem since it can easily determine the real safety
margin of any economically designed structure.
CONCLUDING REMARKS
The aims of the present study have been to
investigate the characteristics of progressive collapse
behavior of typical merchant ship hulls under vertical
sagging or hogging and also to develop the ultimate limit
state design format of ship hulls.
The progressive collapse characteristics for 10
106
typical merchant ships including a single hull tanker, a

double hull tanker with a centerline longitudinal
bulkhead, a double hull tanker with two longitudinal
bulkheads, a single sided bulk carrier, a double sided
bulk carrier, a 3,500 TEU container vessel, a 5,500 TEU
container vessel, a 9,000 TEU container vessel, a FPSO
and a shuttle tanker are studied applying the idealized
structural unit method. Some important influential
parameters such as initial imperfections, lateral pressure
and horizontal moment were varied when looking at the
ultimate hull girder strength characteristics under
vertical moments.
Based on the insights developed in the present study,
closed-form formulations for the ultimate strength of
ships under vertical moments are presented and validated
by a comparison with the ALPS/ISUM solutions. The
uncertainties of the developed closed-form expressions
are quantified in the form of bias and COV.
One important aspect of ultimate limit state design
for the ship hull girder is that a sequence of structural
failure events, some of them undesirable to varying
degrees, occur prior to reaching the ultimate limit state
of the hull girder. A format for the ultimate limit state
design of ships is proposed, and its use is described. The
role of ultimate limit state design procedure as a tool is
also likely to achieve a damage tolerant structural design,
i.e., taking into account age related structural
degradation such as corrosion and fatigue cracks.
Related to the ultimate limit state design of ships,
there are a few problem areas which still remain to be
resolved. On the other hand, the required tools and
technologies for ultimate strength assessment of ship
structures on the basis of hull girder ultimate strength
have now developed to a reasonable degree. The
development of procedures and criteria for ultimate limit
state design of various types of ship structures is one

area of future challenge. As seen by the results of the
present study, the outlook in this regard appears to be
quite promising.
ACKNOWLEDGEMENTS
The present study was undertaken with support from
the American Bureau of Shipping. The authors are
pleased to acknowledge Dr. D. Liu, Dr. J. Spencer and
Dr. Y.S. Shin for their comments. The views expressed
in this paper are those of the authors and are not
necessarily those of the institutions with which they are
affiliated.
REFERENCES
ABS (2000). Rules for building and classing steel
vessels. American Bureau of Shipping, Houston.
Caldwell, J.B. (1965). Ultimate longitudinal
strength. RINA Transactions, The Royal Institution of
Naval Architects, London, 107:411-430.
Dow, R.S. (1991). Testing and analysis of 1/3-scale
welded steel frigate model. Proceedings of the
International Conference on Advances in Marine
Structures, Dunfermline, Scotland, 749-773.
ENV 1993-1-1 (1992). Eurocode 3: Design of steel
structures, Part 1.1 General rules and rules for buildings.
British Standards Institute, London.
ISO 2394 (1998). General principles on reliability
for
structures.
Second
Edition,
International
Organization for Standardization, Geneva.
Mansour, A.E. and Thayamballi, A.K. (1994).
Probability based ship design; loads and load
combination. Ship Structures Committee, Report SSC373, US Coast Guard, Washington DC.
NTS (1998). Design of steel structures. N-004,
Norwegian Technology Standards Institution, Oslo.
Paik, J.K. and Mansour, A.E. (1995). A simple
formulation for predicting the ultimate strength of ships,
Journal of Marine Science and Technology, The Society
of Naval Architects of Japan, 1(1):52-62.
Paik, J.K. and Thayamballi, A.K. (2002). Ultimate
limit state design of steel plated structures, John Wiley &
Sons, Chichester, U.K.
Paik, J.K., Thayamballi, A.K., Wang, G. and Kim,

B.J. (2000). On advanced buckling and ultimate strength
design of ship plating, SNAME Transactions, 108:249290.
Paik, J.K., Thayamballi, A.K., Kim, B.J., Wang, G.,
Shin, Y.S. and Liu, D. (2001). Ultimate limit state design
of ship stiffened panels and grillages, SNAME
Transactions, 109.
Smith, C.S. (1977). Influence of local compressive
failure on ultimate longitudinal strength of ships hull.
Proceedings of the International Symposium on Practical
Design in Shipbuilding, Tokyo, 73-79.
Smith, C.S., Davidson, P.C., Chapman, J.C. and
Dowling, P.J. (1988). Strength and stiffness of ships
plating under in-plane compression and tension. RINA
Transactions, 130:277-296.
Ueda, Y. and Rashed, S.M.H. (1974). An ultimate
transverse strength analysis of ship structure. Journal of
the Society of Naval Architects of Japan, Tokyo,
136:309-324 (in Japanese).
Ueda, Y. and Rashed, S.M.H. (1984). The idealized
structural unit method and its application to deep girder
structures. Computers & Structures, 18(2):277-293.
Yao, T., Astrup, O.C., Caridis, P., Chen, Y.N., Cho,
S.R., Dow, R.S., Niho, O. and Rigo, P. (2000). Ultimate
hull girder strength, Report of Special Task Committee
VI.2, International Ship and Offshore Structures
Congress, Nagasaki, Japan, October, 2:321-391.
NOMENCLATURE
b = breadth of plating
E = elastic modulus
t = thickness of plating
w opl = buckling mode initial deflection of plating
b Y
= plate slenderness ratio
t E
rcx = compressive residual stress in the longitudinal
direction
Y = material yield stress
107
Discussion
P. Rigo, Member
I congratulate the authors for their interesting paper that is
the conclusion of several valuable papers published by the
same authors. At the opposite to the previous papers that
provided sophisticated formulations to better assess failure
modes, this paper concerns a revised practical methodology
which is definitively design oriented. Instead of proposing
advanced and complex numerical approaches, the authors
propose a simple and reliable method that can be easily
implemented by Classification Societies, shipyards and design
offices.
The previous Paik-Mansour (1995) method had as major
shortcomings to consider a single reference panel (element) for
each major component (deck, side shell, bottom and double
bottom). It was also difficult to use when several steel grades
are used. In addition, it was only suitable for merchant ship
having flat bottom and vertical side shells. It was thus difficult
to use for frigates and slender ships.
With this paper and the revised closed-form ultimate
strength bending moment formulation, the former
shortcomings do not remain anymore. The method seems now
ready to be applied to any type of ship hulls. Nevertheless
users must keep in mind the methods assumptions:
a) the collapse is assumed to occur between 2 web-frames
that are assumed to fail after the considered ultimate bending
moment;
b) the vertical distribution of the longitudinal strain is
assumed linear and it is obviously not the case for multi-deck
ships like passenger vessels having large side/deck openings;
c) the ultimate stage (stress distribution) is assumed (see
Figure 17).
Concerning this last assumption, have the authors
experience of ships for which the assumed ultimate stress
distribution is not valid? In order to avoid this assumption a
standard progressive collapse analysis seems in some case
more suitable and it does not require more computing time.
Positions of the neutral axis at the assumed ultimate limit
state (sagging and hogging) are given, respectively by Eqs. (3a)
and (4a). To use these equations, it is necessary to compute the
element stress using Eq.(2) and before each elements axial
strain using Eq.(1). Obviously Eq.(1) requires knowing the
neutral axis, which is obtained by Eq.(3). So it seems that an
iterative procedure must be used. Do they need to update the
stress in the elastic regions (i.e., 2 and 3) using the new
position ( g u ) of the neutral axis? Could the authors give
information about their recommended procedure?
E. Steen, T.K. stvold, E. Byklum and

S. Valsgrd, Det Norske Veritas
The authors address a very important issue, which is directly
related to the safety of ships against total loss and failure. The
subject has been on the agenda for some decades and it
deserves a continuous attention as the assessment of the safety
margins against ship hull girder collapse is of outmost
importance.
The authors approach to the problem is to use a progressive
hull girder model applying a coarse ISUM element modelling
technique. Since the ISUM elements are very central in the
method, though even not directly addressed in the present
paper, it is natural to raise some questions in this respect.
108
Do the ISUM elements need a predefined idealized stressstrain curve (Fig.6), describing both the pre- and post-collapse
characteristics of well defined panel areas or is this an
integrated part of the method? From our understanding the
actual loading situation locally on a panel will affect the
assumed idealized stress-strain behaviour. Since this is not
known before the internal load/stress response is found it will
be interesting to know on which assumptions these
characteristics are generated. We are particularly referring to
the problem of combined load effects (bi-axial and shear),
nominal stress gradients across panels, lateral pressure effects
and the three translational d.o.f. per node problem. Or is the
method only intended for pure uni-axial loading situations?
Moreover, it would be interesting to have more background on
the method for generating the stress-strain post-collapse
characteristics since they are thought to be of importance for
the total hull girder collapse behaviour.
Figure 7 shows comparisons with the present authors
approach and the Dow (1991) hull girder experiment. Added in
the same figure is also some results from Yao (2000) (based on
the Smith method) showing quite diverging results. It would be
interesting to have the authors comments on these large
differences bearing in mind the similarities of the two methods,
both relying on some predefined stress-strain curves for the
pre- and post-collapse characteristics of plate and stiffeners.
From Fig.7c the ultimate hull girder capacity is seen to be close
to 10 MNm for both hogging and sagging condition. It would
be interesting to know how far this is from the simple estimate
of M u = minimum section modulus of deck or bottom
multiplied by the yield stress (same yield stress in deck an
bottom?). The authors also show that the hull girder capacity
depends strongly on the initial imperfections assumed in the
model. A comment on the assumptions used for size and shape
of the model imperfections (and residual stresses) in plate and
stiffeners validated against the corresponding experimental
model values, would be helpful.
The authors also apply different modelling techniques both
locally and globally, the latter meaning a single inter frame
modelling versus a full cargo tank modelling. As we
understand the results presented they are all valid for a single
inter frame model for both sagging and hogging conditions. It
would be interesting to hear the authors opinion on the
relevance of the hogging results, neglecting local cargo load
effects and double bottom bending effects; i.e., how to assess
the effects from variable transverse bi-axial stresses and shear
stresses across the inner and outer bottom panels, double
bottom axial stresses and pure local lateral pressure effects on
panel strength (single span versus two-span effects).
Another question concerns the assumptions regarding
Naviers hypothesis for the simplified hull girder model and
longitudinal stress evaluation, i.e., whether this assumption is
adequate for such complex hull girder response or not? Some
comments on hull girder shear loads effects would also be
useful.
In the ISUM elements the post-collapse curve is defined (or
calculated?), but we cannot see if this effect is included in the
simplified hull girder model. Is it neglected all together, and if
so, can the authors explain the arguments for it?
More questions and comments to this interesting paper
would be appropriate, but due to space limitations we have to
save these for other occasions.

H.P. Cojeen, Member, and
W.S. Peters, Life Member
The views expressed here are those of the discussers, and not necessarily
those of the U.S. Coast Guard or the Department of Transportation.
It is usual for discussers to thank the authors for their good

work and efforts. These authors led by Prof. Paik are to be
especially commended for the series of exceptional papers (4 to
6 papers) that they have contributed to SNAME over the past
several years.
This paper, which culminates a decade of research, provides
some very interesting calculation results that deal with the
structural safety margin of various ship types. The successful
use of progressive collapse analysis for the assessment of
structural capacity of a design is particularly encouraging. The
calculation results that include the effects of initial
imperfections, lateral pressure and horizontal bending moment
give us pause, but surely make us re-consider classification
society standards, and when taken with the paper of Dr. Payer,
seem to point us in a logical direction. The methodology
described is very useful because the results of such an analysis
can reveal the process of hull girder failure and the associated
loads at which various structural components collapse or yield.
The idea that the authors suggest to use the ultimate limit state
procedure to develop structurally damage tolerant designs is
especially noteworthy.
We want to compliment the authors for their courage to
present the calculation results in Table 2 that show the safety
measure calculation results for ten different ships. We believe
that these show a number of interesting trends that should be
explored. In particular, when we look at the computed results,
we will assume that the ratio of ultimate vertical moment
( M u ) to the total bending moment ( M t ) is an acceptable
expression of the safety margin incorporated into the design.
We believe that the results shown that containerships, FPSOs
and the shuttle tanker seem to have a greater margin than
double hull tankers and the bulk carriers. Given the similarity
of the hull configuration, we are interested to learn why the
results shown for the shuttle tanker are so different than that
shown for the double hull tankers.
The results for the double hull tankers in the sagging
condition and for the bulk carriers in the hogging condition
cause us to ask some very basic questions. The results
presented take into account an average amount of
imperfections, but what happens to these results when greater
imperfections are considered? How would some minor contact
damage or higher plate panel deflection effect the results?
Further, it is unclear to what degree the effects of lateral
pressure and horizontal moment have been included.
The results from the paper suggest that these factors
contribute to degrade the ultimate hull girder strength. While
keeping that in mind, it may be difficult to consider the
presented results, calculated without any structural damage, as
indicating that there is sufficient safety margin for these vessels
when they may operate with some increased imperfections or
damage.
The authors pointed out the likelihood of the sinking of a
bulk carrier when a forward hold may be flooded, even
partially. We think that the results presented here suggest that
these types of ships, especially with the losses that continue to
be recorded, need to be built with a greater margin of the ratio
of M u to M t . We believe that, for these ship types, there is
ample experience which shows that the level of maintenance is
inferior and thus if the more correct level of imperfections
were used in your model then these ship types would have an
even smaller margin of the ratio of M u to M t as compared to

other ship types.
Simply stated, we believe that some serious consideration
should be given to re-looking at the IACS unified requirements
for hull girder strength. We feel that the design method
presented here can assist with that re-assessment.
Authors Closure
To start with, we thank all the discussers for their kind
remarks regarding our paper. We thank them even more for the
time they spent developing their various discussions.
Dr. P. Rigo: The Paik-Mansour formula for predicting the
ultimate hull girder strength of ships is refined and improved in
this paper. Though the assumption about the longitudinal stress
distribution over the cross section at the ultimate limit state
remains unchanged, a ship hull can be modeled more easily,
e.g., as an assembly of the plate-stiffener combination models,
see Fig.2(b), or combined support members and plating, the
latter being called the plate-stiffener separation models, see
Fig.2(c).
Y
N.
N.
A.
A.
u
Hull
module
Longitudinal
displacement
distribution
Strain distribution
Stress distribution Stress distribution

at ULS
in linear elastic state
Fig.A.1(a) Distributions of longitudinal strains and stresses for

a ship hull with the same transverse frame spacing at deck, side
and bottom (for hogging condition).
Y
N.
N.
A.
A.
u
Hull
module
Longitudinal
displacement
distribution
Strain distribution
Stress distribution Stress distribution

at ULS
in linear elastic state
Fig.A.1(b) Distributions of longitudinal strains and stresses for

a ship hull (e.g., bulk carrier hull) with different transverse
frame spacing at deck, side and bottom (for hogging condition).
One basic assumption of this simplified method is, as Dr.
Rigo points out, that the hull cross-section remains plane up to
the ultimate limit state under bending moments. In structural
modeling of ship hulls, a hull module between the two adjacent
transverse frames is taken as the extent of analysis. While the
transverse frame spacing in most ships is identical, that in bulk
carriers is different at deck, side and bottom. To handle the
109

condition that the hull cross-section remains plane, a
displacement control is usually applied so that any structural
member is subjected to longitudinal axial displacement which
is proportional to the associated member length or transverse
frame spacing as well as bending curvature. As a result, the
distribution of longitudinal strains over the hull cross-section is
linear for both identical and different transverse frame spacing
as those shown in Fig.A.1.
It has been proven that this assumption is valid for most
merchant cargo vessels, though some types of ships such as
multi-deck passenger vessels may be different. In the authors
opinion, however, the deviation might not be significant as

long as the transverse frame system is properly designed and
provides enough support to the longitudinal members.
The longitudinal stress distribution presumed at the ultimate
limit state compares reasonably with that obtained by the
progressive collapse analysis using ALPS/HULL. Samples of
the refined progressive analysis are shown in Fig.A.2 for
typical merchant cargo vessels considered in the present study.
It is however noted that in some cases the tension flange does
not yield, while the compression flange has collapsed, entering
the post-ultimate strength regime.
-0.8
1.2
N.
-0.6
0.8
x / Y
x / Y
1.0
x / Y = 1.123
y / Y = 0.336
/ = 0.001
Y
0.6
0.4
N.
0.2
x / Y = 0.628
y / Y = 0.085
/ = 0.0
Y
-0.2
0.0
0.0
0
x (10-3)
-0.8
-1
-2
-3
x (10-3)
1.2
1.0
x / Y
-0.6
x / Y
-0.4
-0.4
x / Y = 0.616
y / Y = 0.086
/ = 0.0
-0.2
A.
-1
-2
0.4
A.
x / Y = 0.859
y / Y = 0.266
/ = 0.0
Y
0.0
Hog
-3
x (10-3)
0.6
0.2
0.0
0
0.8
Sag
x (10-3)
(a) Double hull VLCC

1.2
-0.8
N.
-0.6
0.8
x / Y = 1.083
y / Y = 0.314
/ = 0.003
0.6
0.4
x / Y
x / Y
1.0
N.
0.2
x / Y = 0.709
y / Y = 0.059
/ Y = 0.002
-0.2
0.0
0.0
0
x (10-3)
-0.8
-1
-2
x (10-3)
-3
-4
1.2
1.0
-0.4
A.
x / Y = 0.532
y / Y = 0.035
/ = 0.0
Y
-0.2
x / Y
-0.6
x / Y
-0.4
-1
-2
x (10-3)
-3
0.6
A.
x / Y = 0.980
y / Y = 0.294
/ = 0.0
Y
0.4
0.2
0.0
0
0.8
0.0
Hog
-4
x (10-3)
Sag
(b) Capesize bulk carrier

1.2
-1.0
N.
-0.8
0.8
x / Y
x / Y
1.0
x / Y = 1.076
y / Y = 0.323
/ = 0.0
0.6
0.4
N.
0.2
-0.4
x / Y = 0.729
y / Y = 0.288
/ = 0.0
Y
-0.2
0.0
0.0
0
x (10-3)
-0.8
-1
-2
x (10-3)
-3
1.2
1.0
-0.4
x / Y
-0.6
x / Y
-0.6
A.
x / Y = 0.592
y / Y = 0.050
/ = 0.0
-0.2
0.0
0
-1
-2
x (10-3)
0.8
A.
0.6
x / Y = 1.126
y / Y = 0.338
/ = 0.0
Y
0.4
0.2
-3
Hog
0.0
0
x (10-3)
Sag
(c) 9,000 TEU container vessel

Fig.A.2 Longitudinal stress distributions of selected ship hulls at the ultimate limit state under hogging or sagging, as obtained by the
progressive collapse analysis using ALPS/HULL
110
1.2
-0.8
N.
-0.6
0.8
x / Y
x / Y
1.0
x / Y = 1.119
y / Y = 0.330
/ = 0.0
Y
0.6
0.4
N.
0.2
x / Y = 0.505
y / Y = 0.039
/ = 0.0
Y
-0.2
0.0
0.0
0
x (10-3)
-0.8
-1
-2
x (10-3)
-3
1.2
1.0
-0.4
A.
x / Y = 0.561
y / Y = 0.039
/ = 0.0
Y
-0.2
x / Y
-0.6
x / Y
-0.4
-1
-2
Hog
-3
x (10-3)
0.6
A.
x / Y = 0.931
y / Y = 0.279
/ = 0.0
Y
0.4
0.2
0.0
0
0.8
0.0
0
x (10-3)
Sag
(d) FPSO
Fig.A.2 Longitudinal stress distributions of selected ship hulls at the ultimate limit state under hogging or sagging, as obtained by the
progressive collapse analysis using ALPS/HULL (Continued)
Dr. Rigo asks if an iteration is needed to calculate the
neutral axis position at the ultimate limit state. As shown in
Fig.A.3, some iterations can certainly help get more accurate
position of the neutral axis. However, the deviation and its
influence on the resulting ultimate hull girder strength are
limited so that it may not be necessary to attempt iteration.
1.050
(a): guh (Neutral axis above the base line at ULS in hogging)
(b): Muh (The ultimate hull girder hogging moment)
(c): Mus (The ultimate hull girder sagging moment)
(d): gus (Neutral axis above the base line at ULS in sagging)
(a) 3.7 %
1.025
Error
(b) 1.6 %
(c) 0.6 %
(d) 2.1 %
0.975
2
10
12
14
16
18
20
Iteration number
Fig.A.3 Influence of iteration for calculating the neutral axis

position at the ultimate limit state
The following may help better understand the calculation
procedure for the ultimate hull girder strength of a ship:
a) Model a ship hull as an assembly of plate-stiffener
combination elements, i.e., stiffeners with associated plating (A
ship hull can also be modeled as an assembly of plate-stiffener
separation elements, refer to Fig. 2 of the paper);
b) Calculate the neutral axis position of the partially
effective hull cross section, or g e in Eq.(1). In our application
examples, the compressed plating takes some reduced
(effective) plate width:
b e 1.0 for 1
=
2
b 2 / 1 / for > 1
where =
b
t
(A.1b)
c) Calculate the ultimate compressive strength of individual

plate-stiffener combination elements using the closed-form
formula of Paik & Thayamballi (2002):
u
=
Yeq
1
2
2 2
0.995 + 0.936 + 0.170 + 0.188 0.067
1
2
(A.2)
where , = column and plate slenderness ratios for full (not
effective) section, Yeq = equivalent yield stress of the plate-
1.000
be
=1
b
(A.1a)
Y
, while the tensioned plating takes a full
E
width:
stiffener combination element when the yield stress of stiffener

is different from that of attached plating. Eq.(A.2) implicitly
includes the possible effects of local buckling or tripping as
well as initial imperfections (initial deflection and welding
residual stress) at an average level. Also, both column and
plate slenderness ratios used in Eq.(A.2) are calculated for full
section, i.e., without evaluating the effective width of the
attached plating. This may sometimes be of benefit when
evaluation of the plate effective width is difficult.
d) Calculate the applied longitudinal stresses of individual
elements using Eqs.(1a) and (2) with B = Y for sagging or
using Eqs.(1b) and (2) with D = Y for hogging.
e) Using the ultimate compressive strengths obtained from
the step c) and applied longitudinal stresses obtained from the
step d), calculate the neutral axis at the ultimate limit state
from Eq.(3a) for sagging or from Eq.(4a) for hogging. It is
noted that the calculated stresses should be smaller than or
equal the ultimate compressive stress in compressed part or the
yield stress in tensioned part.
f) Recalculate the applied longitudinal stresses of individual
elements following the step d) above, but with regard to the
updated neutral axis, i.e., obtained from the step e).
g) Compute the ultimate hull girder strength from Eq.(3b)
for sagging or from Eq.(4b) for hogging.
The authors fully agree with Dr. Rigos opinion that the
progressive collapse analysis is more suitable and
recommendable to calculate the ultimate hull girder strength
since a progressive analysis more precisely and properly
accounts for the buckling collapse behavior in the pre- and
post-collapse regime.
111

Dr. E. Steen et al.: It is always a pleasure to have valuable
discussions from the DNV team of Dr. Steen and Dr. Valsgard
and to share their wide and varied experiences with us.
They ask about the characteristics of the ISUM plate
element in terms of stress-strain relationship and combined
load effects. This element was developed by using an analytical
approach that considers the effects of combined loads (biaxial
loads, edge shear and lateral pressure) and initial imperfections
(initial deflection and welding residual stresses). The stressstrain relationships are expressed by closed-forms in both preand post-ultimate strength regimes. Details of the derivation of
this ISUM plate element can be found in Paik & Thayamballi
(2002).
The material of the Dow test structure is mild steel (with
Y = 245MPa) for the entire hull, and the section modulus at
deck ( Z D = 0.0439m 3 ) is similar to that at bottom
( ZB = 0.0429m3 ). The ultimate hull girder strength in hogging

is similar to that in sagging. In fact, a simple estimation for
M u , i.e., M u = ZB Y = 10.511 MNm for hogging and
M u = ZD Y = 10.756 MNm for sagging well correlates
with the ALPS/HULL solutions.
The initial imperfections in the Dow test structure were
reported to be between slight and average level, but closer
to the slight level. With the slight level of initial imperfections,
the ALPS/HULL predicts that the ultimate hull girder strength
is 10.400 MNm for hogging and 9.762 MNm for sagging.
Based on the comparison of the progressive collapse analyses,
the authors conclude that the characteristics of structural
elements used in the calculation affect the resulting ultimate

strength behavior, and that the structural modeling of a ship
hull as an assembly of plate-stiffener separation elements may
give more accurate results.
A ship hull module between two adjacent transverse frames
can be taken as the extent of the ultimate hull girder strength
analysis. One reason for that is that typically transverse frames
are designed not to fail prior to the stiffened panels between
them. However, if large lateral load due to extreme or
accidental situation is acted upon, transverse frames may
severely deflect or even collapse. The progressive collapse
behavior of the hull section will be different, and a cargo hold
between two transverse bulkheads may need to be analyzed in
this case. ALPS/HULL method applies the conventional finite
element technique, and employs large structural elements,
often referred to as the idealized structural units. As a result,
ALPS/HULL can be applied to analyzing a cargo hold in the
same manner, while most of the methods based on Smiths
methodology may not be able to. The formulations of these
idealized structural units involve biaxial stresses, edge shear
and out-of-plane pressure, the last being treated as a constant
load component. Some examples of the computed stress
components are also given in Fig.A.2.
Regarding the question on the Navier hypothesis for
developing the closed-form expression of the ultimate hull
girder strength, the authors reply to Dr. Rigo may be referred
to. In deriving the closed-form ultimate hull girder strength
formulation, the effect of the post-ultimate strength behavior is
not accounted for.
ULS based safety measure
2.0
1.5
Note:
Hollow: Sagging
Solid : Hogging
: Single hull tanker
: Double hull tanker
with one center-longitudinal bulkhead
: Double hull tanker
with two longitudinal bulkheads
: Double sided bulk carrier
: 3,500 TEU container vessel
: FPSO
: Shuttle tanker
1.0
0.5
0.5
1.0
1.5
2.0
Section modulus based safety measure
2.5
Fig.A.4 The section modulus based safety measure versus the ultimate strength based safety measure for
ships longitudinal strength amidships
112

Mr. H.P. Cojeen and Mr. W.S. Peters: Following the
suggestions of Mr. Cojeen and Mr. Peters, we compare the
section modulus based safety measure versus the ultimate hull
girder strength based safety measure as shown in Figure A.4.
Clearly these two safety measures do not always correlate. The
ultimate strength more realistically measures the strength
capacity of a ships hull, and the traditional section modulus
approach may not be adequate in determining the real safety
margin of longitudinal strength of some types of ships.
Mr. Cojeen and Mr. Peters raise a question why the ultimate
strength of shuttle tankers tends to be greater than that of
double hull tankers in sagging condition. It is interesting to
compare two different ship types, but any conclusion from such
comparisons can not be viewed as universally true before
extensive studies have found the same in many other ships. The
shuttle tanker in our analysis has much thicker deck plates than
the double hull tanker of similar size, as shown in Fig.A.5. This
causes a huge difference in the sagging condition. However,
not all shuttle tankers have such thick deck plates, and designs
of most shuttle tankers are similar to double hull tankers. We
do not expect substantial difference in hull girder ultimate
strength between trading tankers and shuttle tankers.
Double hull tanker #1

Deck: t=16mm, HT32
L = 233.0 m
B = 42.0 m
D = 21.3 m
F.S. = 4.12 m
Bottom: t=16.5mm, HT32

Shuttle tanker
Deck: t=22mm, HT32
L = 254.0 m
B = 46.0 m
D = 22.6 m
F.S. = 3.6 m
Bottom: t=16mm, HT32
Fig.A.5 Thickness of deck and bottom plating in double hull

tanker and shuttle tanker considered in the present study
10
For hogging:
1. Buckling collapse of outer bottom longl.,
bilge keel & center girder longl.*
lower side longl.* & bottom girder longl.*
8. Buckling collapse of inner bottom plates,
center girder plates & yielding of deck longl.
3
2
5
67
89
3
Mt=6.056 103 MNm

1 : Slight
2 : Average
3 : Severe
For sagging:
-5

& upper side longitudinal bulkhead*
12. Buckling collapse of upper side longl.*
13. Buckling collapse of upper side shell
bulkhead* & deck plates*
& upper side shell*
15. Yielding of bottom girder longl.*
17. Yielding of bottom girder longl.
10
11
17 16 15
12
13
14
2
1
Mt=-6.056 103 MNm
-10
-3
-2
-1
Curvature
10-7
(1/mm)
Fig.A.6 The influence of initial imperfections on the progressive collapse behaviour of a FPSO hull, as obtained by ALPS/HULL
113

Mr. Cojeen and Mr. Peters are very concerned with the
strength reduction due to initial imperfections. As shown in
Fig.A.6, fabrication related initial imperfections significantly
reduce the load carrying capacity of individual structural
members and also hull girder. It is interesting to note that the
first local buckling occurs at the bending moment close to the
IACS total design moment in this specific ship. The ultimate
hull girder strength is at least 20% higher than the IACS design
moment. With more and more measurements of initial
imperfections becomes available, we could define a more
realistic reference level of initial imperfections and thus we
will more precisely know the safety margin of the ships hull
using the procedure developed. We believe that our paper and
studies of other researchers in the literature will contribute to
the international efforts to improve the design requirements for
hull girder strength.
In some ships considered in the present paper, the ultimate
hogging moment is smaller than the ultimate sagging moment
even if the section modulus at bottom is greater than that at
deck. One of the reasons is that a higher level of initial
imperfections (i.e., an average level rather than a slight level)
was assumed. If the slight level of initial imperfections is
considered, the ultimate hogging moment is larger than the
ultimate sagging moment in most ships, as would be expected.
The present paper did not use any hard corner connection
in the progressive collapse analysis of ship hulls. A hard
corner fails by yielding in tension or compression, but it is not
allowed to buckle. For bulk carriers, structures that can be
treated as hard corners are: connections of side shell and
upper deck, hopper knuckles, and connections of side shell and
bottom shell (bilge), connections of bottom girders with inner
and outer bottom plates. Figure A.7 shows the influence of
hard corners on the progressive collapse behavior of a Capesize
bulk carrier hull, as obtained by ALPS/HULL. In this specific
type of the ship, the ultimate bending moment can be improved
by 7% if the effects of hard corners are taken into account.
The authors share the opinion of Mr. Cojeen and Mr. Peters
that any structural damage will reduce the load-carrying
capacity and the safety measure can be further decreased. To
keep the integrity of aging ships at a certain level, damage
tolerant design procedure in addition to relevant repair/
maintenance together with close-up survey must be applied.
The authors plan to present this issue in a separate paper in
near future.
In closing, the authors are indeed pleased to have the
benefit of the high quality discussions to their paper, which
significantly add value to the paper and serve to emphasize
various aspects of the ultimate limit state design approach.
20
Case D
Case C
Case B
: Hard corners
No hard corner
10
Case A
CL
Case B
Case A
CL
Case
A
B
C
D
-10
Muh (GNm) Increase (%)

14.434
100.0
14.550
100.8
14.665
101.6
15.520
107.5
: Hard corners
: Hard corners
CL
Case C
Case D
CL
-20
-3
-2
-1
Fig.A.7 Influence of hard corners on the progressive collapse behavior of a Capesize bulk carrier hull, as obtained by ALPS/HULL
114

Ultimate Limit Stat Design Ship Hulls

Transféré par

Informations du document

Copyright

Formats disponibles

Partager ce document

Partager ou intégrer le document

Options de partage

Avez-vous trouvé ce document utile ?

Ce contenu est-il inapproprié ?

Droits d'auteur :

Formats disponibles

Ultimate Limit Stat Design Ship Hulls

Transféré par

Droits d'auteur :

Formats disponibles

ABS TECHNICAL PAPERS 2002

Ultimate Limit State Design of Ship Hulls

Ultimate Limit State Design of Ship Hulls

ABS TECHNICAL PAPERS 2002

Serviceability limit state (SLS)

The structural design criteria against the ULS are

Design load level 1

SLS conventionally represents failures under normal

ULS (also called ultimate strength) represents the

Local damage which may reduce the durability of

Figure 1. Structural design considerations based on the

ABS TECHNICAL PAPERS 2002

using the so-called deep girder unit. Several different

EFFICIENT AND ACCURATE METHODOLOGY

Ultimate Limit State Design of Ship Hulls

(a) A typical steel plated structure

(b) Structural idealization as an assembly of platestiffener combination units

ABS TECHNICAL PAPERS 2002

Figure 3(a). The ISUM beam-column unit with attached

(c) Structural idealization as an assembly of platestiffener separation units

Figure 3(b). The ISUM beam-column unit without

(d) Structural idealization as an assembly of stiffened

Figure 2. Various types of idealizations for a steel plated

Figure 4. The ISUM rectangular plate unit ( : nodal

Figure 5. The ISUM stiffened panel unit ( : nodal

Ultimate Limit State Design of Ship Hulls

ABS TECHNICAL PAPERS 2002

Figure 7(a). Mid-ship section of the Dow frigate test ship

Figure 6. Idealized stress-strain behavior of the ISUM

Ultimate Limit State Design of Ship Hulls

Figure 7(b). ALPS/ISUM model for the Dow frigate test

: Experiment (Dow 1991)

Vertical moment (MNm)

For ultimate strength analysis of ship structures

= 0.025 2 t , rcx / Y = 0.0

Curvature 10-7 (1/mm)

Figure 7(c). Comparison of ALPS/ISUM with the Dow

ABS TECHNICAL PAPERS 2002

Figure 8(c). ALPS/ISUM model III (No. of plate units =

Figure 8(a). ALPS/ISUM model I (No. of plate units = 0,

Figure 8(b). ALPS/ISUM model II (No. of plate units =

Ultimate Limit State Design of Ship Hulls

ABS TECHNICAL PAPERS 2002

Figure 8(f). ALPS/ISUM model VI (No. of plate units =

Vertical moment103 (MNm)

PROGRESSIVE COLLAPSE CHARACTERISTICS

Figure 9. Progressive collapse behavior of a 105,000

Ultimate Limit State Design of Ship Hulls

Figure 10(a). Schematic representation of mid-ship

Figure 10(b). Schematic representation of mid-ship

ABS TECHNICAL PAPERS 2002

Figure 10(c). Schematic representation of mid-ship

Figure 10(d). Schematic representation of mid-ship

Figure 10(e). Schematic representation of mid-ship

Figure 10(f). Schematic representation of mid-ship

Figure 10(g). Schematic representation of mid-ship

Figure 10(h). Schematic representation of mid-ship

Figure 10(i). Schematic representation of mid-ship

Figure 10(j). Schematic representation of mid-ship

Ultimate Limit State Design of Ship Hulls

ABS TECHNICAL PAPERS 2002

Table 1. Hull sectional properties of the 10 typical merchant ships

Ultimate Limit State Design of Ship Hulls

In the ALPS/ISUM computations, deck or bottom