Você está na página 1de 37

Ocean Engineering 28 (2001) 1097–1133

Ultimate strength of ship hulls under torsion


Jeom Kee Paik a,*, Anil K. Thayamballi b, P. Terndrup
Pedersen c, Young Il Park a
a
Pusan National University, Department of Naval Architecture and Ocean Engineering, 30
Changjeon-Dong, Kumjeong-Ku, Pusan 609-735, South Korea
b
Chevron Shipping Company, San Ramon, CA 94583-4289, USA
c
Technical University of Denmark, Department of Naval Architecture and Offshore Engineering,
Building 101E, DK-2800, Lyngby, Denmark

Received 17 May 1999; accepted 27 October 1999

Abstract

For a ship hull with large deck openings such as container vessels and some large bulk
carriers, the analysis of warping stresses and hatch opening deformations is an essential part
of ship structural analyses. It is thus of importance to better understand the ultimate torsional
strength characteristics of ships with large hatch openings. The primary aim of the present
study is to investigate the ultimate strength characteristics of ship hulls with large hatch open-
ings under torsion. Axial (warping) as well as shear stresses are normally developed for thin-
walled beams with open cross sections subjected to torsion. A procedure for calculating these
stresses is briefly described. As an illustrative example, the distribution and magnitude of
warping and shear stresses for a typical container vessel hull cross section under unit torsion
is calculated by the procedure. By theoretical and numerical analyses, it is shown that the
influence of torsion induced warping stresses on the ultimate hull girder bending strength is
small for ductile hull materials while torsion induced shear stresses will of course reduce the
ship hull ultimate bending moment.  2001 Published by Elsevier Science Ltd.

Keywords: Ultimate hull girder torsional strength; Container ship; Idealized structural unit method

* Corresponding author. Tel.: +82-51-510-2429; fax: +82-51-512-8836.


E-mail addresses: jeompaik@hyowon.pusan.ac.kr (J.K. Paik), akth@chevron.com (A.K.
Thayamballi), ptp@ish.dtu.dk (P.T. Pedersen).

0029-8018/01/$ - see front matter  2001 Published by Elsevier Science Ltd.


PII: S 0 0 2 9 - 8 0 1 8 ( 0 1 ) 0 0 0 1 5 - 4
1098 J.K. Paik et al. / Ocean Engineering 28 (2001) 1097–1133

Nomenclature
a the length of the plate unit
b the breadth of the unstiffened plate unit or plate part between
stiffeners for the stiffened plate unit
B the ship moulded breadth
Cb the block coefficient
co 0.14 or 0.459 when the unit of MTm is kN-m and tf-m or Lft-ft,
respectively
Cw the waterplane area coefficient at the scantling draft, which may be
approximated by 1.09Cb, if not available (Cw need not be taken
greater than 0.98 for typical bulk carriers)
D the ship moulded depth
df the scantling draft, which should not be taken as less than 12.5 m
(or 41 ft)
e distance to the effective shear centre of the hull girder within cargo
space, measured from the baseline of the vessel, positive upward
E the modulus of elasticity
FH the horizontal component of the sectional hull girder shear force
FV the vertical component of the sectional hull girder shear force
G E
the shear modulus which is taken as G=
2(1+n)
h the distance from the shear centre of the cross section to the
individual thin walled element
hc the distance to the tangent line at any point of the cross section
Ihh the central moment of inertia with respect to the shear center
It the St. Venant torsional constant
Iz the moment of inertia with respect to the z-axis which is taken as


Iz= y2tds
Iww the sectorial moment of inertia
k 2.7, 0.276 or 0.077 when the unit of MTm is kN-m or tf-m or Ltf-ft,
respectively
L the ship length between perpendiculars in m (or ft)
m the external torque moment per unit length
MH the horizontal hull girder bending moment
Ms the St. Venant torsional moment which is taken as Ms=GIty
MT the sectional torsional moment
MTm the maximum torsional moment
MTU the ultimate hull strength by torsion alone
MV the vertical hogging or sagging bending moment
MVU the ultimate hull strength by vertical hogging or sagging moment
alone
J.K. Paik et al. / Ocean Engineering 28 (2001) 1097–1133 1099

Mw the warping torsional moment which is taken as Mw=G(Ihh⫺


It)(y⫺c)
Mw the bimoment
s the co-ordinate indicating a point in the cross section through which
a thin walled element is passing
t the plate thickness (of cells)
wo the buckling mode initial deflection
x the location from aft perpendicular of the vessel
ys the distance from the shear center to the side shell
zs the distance from the shear centre to the upper deck
c the warping function
n the Poisson ratio
w(s) the sectorial co-ordinate (normalized warping function) of the cross
section element at the point s=s(y,x) which is taken as w(s)=兰s0hcds
so the yield stress of the material
srx the welding induced compressive residual stresses in x direction
sry the welding induced compressive residual stresses in y direction
sw the normal stress
ts the shear stress
y the amount of rotation of the cross section about the x-axis

1. Introduction

The marine industry has recently seen a revolution in container transport with a
very rapid growth of vessel sizes. At present, container vessels carrying 7000 TEU
have been built. Such vessels can have a length of 350 m, a breadth of 45 m, and
a draft of 13 m. With their large deck openings, these vessels will call for detailed
investigations of the hull girder response to torsion and to hull girder bending in
order to predict fatigue damage and ultimate hull girder capacity. To the authors’
knowledge, there have been no reported accidents where a container vessel or another
vessel with large hatch openings has been reported broken due to a combination of
torsional and hull girder bending loads. But torsion induced buckling damage in
deck structures of ships with large open decks is not rare, see for instance Hong et
al. (1987).
Most classification society criteria and procedures for ship structural design are
based on the first yield of hull structures together with buckling checks for structural
components (i.e., not for the whole hull structure). These methods have proved them-
selves to be effective for intact vessels in normal seas and loading conditions. How-
ever, their applicability to assess the survivability of vessels in accidental situations,
i.e., after collisions, grounding or overloading, is somewhat less certain. In these
cases, it is necessary to account more precisely for the interacting effects between
yielding, buckling and sometimes, rupture of local components and the related effects
on the global behaviour of the structural system.
1100 J.K. Paik et al. / Ocean Engineering 28 (2001) 1097–1133

While service proven, the traditional design criteria and associated linear elastic
stress calculations do not necessarily define the true ultimate limit state which is the
limiting condition beyond which a ship hull will fail to perform its function. Neither
do such procedures help understand the likely sequence of local failure prior to
reaching the ultimate limit state. It is of course important to determine the true ulti-
mate strength if one is to obtain consistent measures of safety which can form a
fairer basis for comparisons of vessels of different sizes and types. An ability to
better assess the true margin of safety should also inevitably lead to improvements
in regulations and design requirements.
From the point of view of a structural designer, it can be said with reasonable
certainty that useful methods are now available to predict the ultimate hull girder
strength under vertical bending moment alone. Admittedly, this is of fundamental
benefit, because the vertical bending moment is the most primary hull girder load
component.
However, it is realized that the horizontal bending moment may approach or
exceed the magnitude of the vertical bending moment when the vessel runs at an
oblique heading in high waves. This could mean that in some cases the effects of
horizontal sectional shear forces may not be insignificant. Also, in bulk carriers carry-
ing dense types of cargo such as iron ore, a non-homogeneous alternate hold loading
condition is common and, as a result, large vertical sectional shearing forces in
addition to longitudinal bending moment will be imposed on the hull structure. More-
over, torsional moments are normally considered to be important for vessels with
low torsional rigidity, for example in container vessels and some large bulk carriers
with large deck openings.
In order to obtain a complete picture of structural response, therefore, all hull
girder sectional load components mentioned above, i.e., vertical bending, horizontal
bending, vertical shear, horizontal shear and torsion, see Fig. 1, should be considered

Fig. 1. Hull girder sectional load components.


J.K. Paik et al. / Ocean Engineering 28 (2001) 1097–1133 1101

in a consistent manner in any first principles based procedure for hull girder
strength analysis.
The present study focuses on the ultimate strength of ship hulls under torsion.
Most previous studies on the ultimate hull strength have not dealt with the effects
of torsion. An important exception to this is the work of Ostapenko (Ostapenko and
Vaucher, 1980; Ostapenko, 1981; Ostapenko and Moore, 1982) who studied the
ultimate strength of ship hulls subjected to combined vertical bending, vertical sec-
tional shear and torsional moment, both theoretically and experimentally. Ostapen-
ko’s experimental models were made of very thin plate elements which could buckle.
The studies point out that when a ship hull with large deck opening is subjected to
torsion, warping effects can become significant after buckling of panels occurs so
that the original symmetrical hull cross section may potentially be transformed into
a structurally unsymmetrical one.
Studies of how the torsion induced warping affects the ultimate strength of ship
hulls, and of the ultimate hull strength interaction relationships between torsion and
other hull girder load components, are thus far not available. The authors have pre-
viously devoted much effort on the ultimate strength of ship hulls subject to com-
bined longitudinal bending and vertical sectional shear forces (Paik et al., 1996). In
this past work, the ultimate strength characteristics of various ship hulls under a
combination of vertical bending, horizontal bending and vertical sectional shear
forces were investigated numerically and theoretically. It was clear from these pre-
vious studies that the ultimate strength of ship hulls subject to bending moment alone
would be governed by collapse of the compression flanges and yielding of tension
flanges. On the other hand, for a ship hull under vertical sectional shear force, it is
the vertical members such as side shell and longitudinal bulkheads that sustain most
of the applied loads, and the hull could reach its ultimate limit state immediately
after these vertical members collapse.
These previous ultimate hull strength studies did not accommodate the effects of
torsion and horizontal sectional shear. In the present study, the previously developed
procedures are modified to take into account the effects of torsion and horizontal
sectional shear in addition to vertical bending, horizontal bending and vertical sec-
tional shear.
A primary aim of the present study is to investigate the ultimate strength character-
istics of ship hulls with large hatch openings under torsion. For thin-walled beams
with open sections, axial (warping) as well as shear stresses are developed by torsion
at the cross section. A procedure for calculating these stresses is briefly described.
As an illustrative example, warping and shear stress distributions for a typical con-
tainer vessel hull cross section under unit torsion are calculated using the theory. As
part of the study, a special purpose non-linear finite element software ALPS/HULL
for the efficient analysis of the ultimate strength of ship hulls subjected to a combi-
nation of five hull girder sectional load components (i.e., torsion, vertical shear,
horizontal shear, vertical bending and horizontal bending) and one local load compo-
nent (i.e., lateral pressure loads) has been developed based on the idealized structural
unit method (ISUM). Using the ALPS/HULL software, the effect of warping on the
ultimate hull girder strength is studied numerically. The influence of torsion on the
1102 J.K. Paik et al. / Ocean Engineering 28 (2001) 1097–1133

ultimate vertical bending moment of ship hulls is also studied, and an ultimate hull
strength interaction relationship between torsion and vertical bending is developed.

2. Hull girder torsional moments

Torsion is caused by forces which do not pass through the shear centre axis of a
ship hull cross section. Torsional moments acting on a ship hull may generally be
divided into two components, namely static (i.e., stillwater) torsion and dynamic
(i.e., wave induced) torsion. Stillwater torsion is due to non-symmetrical cargo load-
ing over port and starboard with the vessel as a whole remaining upright. In a seaway,
wave induced torsion is caused by non-symmetrical distribution of hydrostatic and
hydrodynamic forces over port and starboard. Mass acceleration forces resulting from
the ship’s motion also contribute to dynamic torsional moments. Finally, there may
also be some torsional (vibratory) contribution arising from propeller shaft torque
variations, vertical free forces from the propellers in twin screw vessels, and perhaps
free horizontal forces in the engine.
For practical purposes, dynamic torsion is dominant because cargo is usually
evenly loaded over port and starboard. The magnitude of (wave induced) torsional
moments depends not only on the hull form, the heading of the vessel to the wave
and wave height but also on the position of the centre of twist which is determined
by the structural arrangement of the hull cross section. An accurate direct calculation
of the torsional moments is an involved task, and so classification societies provide
simplified expressions for predicting the magnitude of torsional moments for design
purposes. For instance, ABS (1995) gives the following expression for the lifetime
maximum hull girder torsional moment:
MTm⫽k·L·B2·df{(Cw⫺0.5)2⫹0.1}·{0.13⫺(e/D)(co/df)1/2} (1)

The distribution of the torsional moment along the vessel length MT is specified
by ABS to be as follows:


5x·MTm when 0ⱕxⱕ0.2L
MT(x)⫽ MTm when 0.2L⬍x⬍0.8L (2)
5(1−x)MTm when 0.8LⱕxⱕL

The location of shear centre, i.e., the parameter “e” in Eq. (1), can be determined
under the condition that the net warping displacements must be zero (because other-
wise there would be an extension of the beam) and that the net first moments of
warping stresses, i.e., horizontal and vertical bending moments, about the centre of
twist must be zero (because otherwise related horizontal or vertical bending moments
J.K. Paik et al. / Ocean Engineering 28 (2001) 1097–1133 1103

will arise). The above three conditions can be given mathematically by (Pedersen
and Jensen, 1983; Hughes, 1988)

冕 w(s)tds⫽0 (3a)

冕 z·w(s)tds⫽0 (3b)

冕 y·w(s)tds⫽0 (3c)

where y and z are as defined in Fig. 1.


When the hull structure is symmetric with respect to the vessel centre line, it is
clear that the horizontal position of the shear centre is located along the vessel centre
line. The vertical location of the shear centre in a cross section without closed cells
can be determined by using the following equation (Pedersen and Jensen, 1983;
Hughes, 1988):

e⫽⫺
1

Iz
yw(s)tds (4)

This equation defines the location of the shear centre of the cross section below
the base line of the hull such that no twisting is produced. As illustrative examples,
the position of shear centre for the two types of idealized hull girder cross sections
in Fig. 2, nominally representing a container vessel and a bulk carrier, can be determ-
ined as follows (For various symbols used in Eqs. (5a) and (5b), Fig. 2 is referred to):
(1) Container vessel cross section (see Fig. 2a)
3D21(ts+tl)
e⫽ (5a)
B1(tb+ti)+6D1(ts+tl)

(2) Bulk carrier cross section (see Fig. 2b)


3B2D2ts+2D·c·(3B2−4c2)·td
e⫽ (5b)
B3tb+6B2Dts+2c·{3(B−c)2+c2}·td

3. Elastic response of a hull cross section under torsion

For a thin-walled beam with an open cross section, the torsional stiffness is much
less than that with a closed section. This implies that for a given torsional moment,
the open section may twist much more due to its low torsional rigidity. In contrast
1104 J.K. Paik et al. / Ocean Engineering 28 (2001) 1097–1133

Fig. 2. (a) Nomenclature: Simplified container vessel type cross section. (b). Nomenclature: Simplified
bulk carrier type cross section.

to the uniform (i.e., St. Venant) torsion of solid beams and a special class of warping-
free thin walled beams, non-uniform axial deformation (i.e., warping) occurs in the
case of thin-walled beams with open sections such that an initially plane cross section
will no longer remain plane. This would mean that torsional moments will develop
axial (warping) stresses as well as shear stresses when the warping displacements
are restrained, see Fig. 3.
In actual ship structures, warping displacements are normally partly restrained and
thus the analysis of warping stresses as well as hatch opening deformations is in
principle an essential part of the response analyses of ships.
J.K. Paik et al. / Ocean Engineering 28 (2001) 1097–1133 1105

Fig. 3. Warping displacements and stresses for an open cross section thin-walled beam under torsion
due to end restraints.

3.1. The nature of warping

When the cross sections are free to warp, warping stresses normal to the cross
section will not be introduced. However, at cross sectional discontinuities such as
at the transition between the cargo area and the engine room, or between the cargo
area and the fore body, and at cross deck beams, the warping deformations will be
restrained to varying degrees. The restraint at these locations induces warping
stresses, which for ships with large deck openings are significant, see Jensen et al.
(1992), and so the warping stresses, and the associated deformations, for instance
hatch opening distortions, must be accounted for in design.
In the linear elastic regime, the coupled torsional-horizontal bending response of
ship hulls can be determined by commercially available general purpose computer
software incorporating plane stress finite elements. However, in contrast to analysis
of vertical hull girder bending response, the analysis of the torsional response nor-
mally requires a model of the complete hull. The degree of effort needed leads to
the use of relatively coarse models where each element must represent the stiffness
of an assemblage of many real structural elements. Such models cannot be used
for direct detailed stress analyses. Besides, such analyses require exceptionally high
manpower from well-trained personnel, which is also costly. Even when finer finite
element modeling is used, the results obtained can be difficult to interpret since the
1106 J.K. Paik et al. / Ocean Engineering 28 (2001) 1097–1133

output sometimes contains clusters of local and erroneous information, which may
tend to obscure the more important global or overall structural behaviour.
For these reasons, there has been an effort to adopt a totally different method of
linear elastic analysis based on one-dimensional finite difference or finite element
methods, see Pittaluga (1978), Senjanovics and Grubisic (1979) and Pedersen (1983,
1985, 1991). These one-dimensional procedures are based on the assumption that
the beam cross sections are rigid in their own plane, even if they may warp out of
their planes. That is, a system of transverse frames and bulkheads are assumed to
exist in the real structure such that this assumption leads to a reasonable mathematical
model for the actual physical behaviour. Such one-dimensional analysis procedures
can predict overall linear elastic stress distributions with an accuracy that are compa-
rable to those found from comprehensive finite element models has been demon-
strated, for example, by Ostergaard et al. (1996).

3.2. Stress distribution at a hull cross section arising from torsion

Since beam models have been found to give a good overall representation of the
torsion induced warping stresses, we shall then in the present study use beam model
based stress distributions for purposes of a general discussion of the expected effect
of warping stresses on the ultimate hull girder strength.
Torsional equilibrium of a section of a beam with an arbitrary prismatic cross
section requires that
M⬘T⫹m⫽0 (6)
where ()⬘ denotes differentiation with respect to the longitudinal axis x of the hull
beam.
To describe the nature of warping stresses we shall first assume that the defor-
mation of the hull beam is linear elastic and that the shape of the cross section does
not change during deformation. Then the deformation u in the longitudinal x-direc-
tion can be written as:
u(x,s)⫽⫺c(x)w(s) (7)

As w(s) depends solely on the shape of the cross section and not on the defor-
mation, the deformation of the beam segment due to pure torsion can be described
by certain cross sectional constants, the rotation y=y(x) of the cross section about
the x-axis, and the warping function c(x).
These assumptions imply that the torsional moment MT can be expressed as the
sum of two components, i.e., the St. Venant torsional moment Ms plus the warping
torsional moment Mw, namely
MT⫽Ms⫹Mw (8)

The cross sectional constant It, called St. Venant’s torsional constant, is given as
J.K. Paik et al. / Ocean Engineering 28 (2001) 1097–1133 1107

It⫽I0x ⫹Ict (9)


0
where the first term I is due to contributions to the moment Ms from shear stresses
x
varying linearly over thickness ti of the straight thin walled elements in the cross
section. The associated maximum shear stresses in a thin-walled element with thick-
ness ti are given by tiGy⬘. The second term Ict is due to the constant shear flows
qa=tsti arising in each of the closed cells in a cross section. If the cross section
includes closed cells, their contribution Ict to It will usually be far the biggest.
For a unit St. Venant torsional moment, i.e., Ms=1 N-m, an example of the shear
stress distribution ts on a typical container vessel cross section is shown in Fig. 4.
The shear stress distribution ts is used when calculating the sectorial co-ordinate w.
The cross sectional constant Ihh used to calculate the warping torsional moment Mw
is called the central moment of inertia with respect to the shear centre and is
expressed by

Ihh⫽2 冘冕
i⫽1
h2tids (10)

Fig. 4. Distribution of shear stresses ts (N/m2) due to a unit St. Venant torsional moment Ms (N-m) in
at a typical container vessel hull cross section (Pedersen and Jensen, 1983).
1108 J.K. Paik et al. / Ocean Engineering 28 (2001) 1097–1133

As indicated in Section 2, the position of the shear centre e relative to the y-axis
can be determined from the condition that the normal stresses, i.e.,
sw⫽E∂u/∂x⫽⫺Ec⬘w (11)
due to the deformation in the x-direction do not result in axial forces or bending
moments around the axes passing through the shear centre.
The sectorial co-ordinate w is given by


w(s)⫽ (h⫺ts·It)ds (12)

where ts is the previously mentioned St. Venant torsional stresses in the cells (see
Fig. 4) for a unit moment, i.e., Ms=1 N-m. The co-ordinate s indicates a point in
the cross section through which a thin walled element is passing. As both the distance
h and the stress ts and thereby w and sw have no symmetry components with respect
to the symmetry axis z, the bending moment around any arbitrary axis drawn parallel
to the y-axis will be zero.
Using the principle of virtual work in connection with the internal and the external
shear deformation measures, the resultants of the normal stresses sw are obtained
for a cross section. This quantity is denoted the bimoment Mw and is expressed by
Mw⫽⫺EIwwc⬘ (13)
where the sectorial moment of inertia is given as

Iww⫽2冘冕 i
ti w2ds (14)

Using Eq. (13), Eq. (11) can be expressed as:


sw⫽wMw/Iww (15)
where Iww has the dimension [length6], while the bimoment Mw has the dimension
[force×length2].
Fig. 5 illustrates the variation of the torsion induced axial (warping) stresses sw
around the cross section, corresponding to a unit bimoment Mw. The shear stress tw,
the resultant of which is Mw in Eq. (8), is found in a manner analogous to the classical
procedure for determination of the shear stress due to a vertical sectional shear force
FV acting on the cross section. Fig. 6 shows the distribution of torsion induced shear
stresses on the typical container vessel hull cross section.
In this manner, all the stress contributions can be determined when the generalised
sectional forces Ms, Mw, and Mw are determined. These three generalised beam forces
can be expressed by the rotation y and the warping function c according to Eqs.
(8) and (13). The two displacement measures, i.e., y and c, can be determined by
solving the two coupled differential equations, namely
[GIt⫹G(Ihh⫺It)(y⬘⫺c)]⬘⫹m⫽0 (16)
J.K. Paik et al. / Ocean Engineering 28 (2001) 1097–1133 1109

Fig. 5. Distribution of axial (warping) stresses sw due to a unit bimoment Mw.

⫺(EIwwc⬘)⬘⫺G(Ihh⫺It)(y⬘⫺c)⫽0 (17)

The first Eq. (16) is the moment equilibrium Eq. (6) with MT inserted. The second
Eq. (17) expresses the axial force equilibrium (i.e., M⬘w=Mw). In order to solve Eqs.
(16) and (17), the boundary conditions at the ends of the prismatic beam segments
must be known.
It should be noted that Eqs. (16) and (17) describe pure torsion of a beam segment.
However, coupling between torsion and horizontal bending will occur if the cross
section is not symmetric about also the horizontal axis.
Pedersen (1983) suggests a method that can be used to analyze coupled horizontal–
torsional deformations of beam segments in a consistent manner. In that method, the
beam model is based on a Timoshenko–Vlasov theory for hull segments with slowly
varying sectional properties. As dependent variables, the method uses the horizontal
displacement of points along a longitudinal axis, the slope of the cross section due
to horizontal bending, the torsional rotation y(x), and the warping function c(x).
These four dependent variables are related among themselves through four coupled
ordinary differential equations.
Cross sectional discontinuities, which occur at the transitions between open and
closed parts of the hull and where deck beams or transverse girders are situated, are
modeled by discontinuity conditions in the differential equations. The effect of verti-
1110 J.K. Paik et al. / Ocean Engineering 28 (2001) 1097–1133

Fig. 6. Distribution of shear stress due to a unit warping torsional moment at a typical container vessel
hull cross section (Pedersen and Jensen, 1983).

cal or horizontal torsional boxes and transverse deck strips are taken into account
by introducing discrete bimoment springs specifying relations between the added
bimoment and the warping function c(x).
Fig. 7 shows a container vessel modeled by a number of prismatic cross sectional
elements, and involving three cross sectional discontinuities and twenty transverse
deck beams. Fig. 8 shows the calculated bimoment distribution along the hull girder,
and Fig. 9 shows how the applied torsional moment MT for this ship is found to be
distributed between the St. Venant torsional moment Ms and the warping moment
M w.
Pedersen (1991) has also studied the application of a beam theory in which the
warping deformation is described by a complete set of orthogonal functions instead
of just one function w. But the conclusion from that study was that the additional
complexity by introducing more deformation functions did not result in significantly
better accuracy. This conclusion is borne out by comparisons between results
obtained using beam theory with results from detailed finite element calculations by
Ostergaard et al. (1996).
J.K. Paik et al. / Ocean Engineering 28 (2001) 1097–1133 1111

Fig. 7. A one-dimensional structural model of a container vessel with 3 cross sectional discontinuities
and 20 transverse deck beams.

Fig. 8. Bimoment distribution for the hull depicted in Fig. 6.

Fig. 9. Applied torsional moment, St. Venant moment and warping moment for the hull depicted in
Fig. 6.

4. A hypothesis for the ultimate hull girder strength analysis considering


warping effects

The object of this section is to present a hypothesis on the analysis of ultimate


hull girder strength taking into account the warping effects. The validity of this
1112 J.K. Paik et al. / Ocean Engineering 28 (2001) 1097–1133

hypothesis will be examined in the following sections by numerical analyses. The


principal assumptions are that:

앫 The hull material is assumed to behave in a ductile manner, i.e., fast fracture
initiated by fatigue or material defects is not included.
앫 Potential deficiencies in material, repairs and construction can be excluded.
앫 Any deficiency in global or local structural alignment is negligible.
앫 Local failure with respect to material or welded connections does not occur.

From the description for the warping stresses sw as given in Section 3.2, it is
seen that these stresses are orthogonal to the distribution of axial stresses and to the
distribution of vertical and horizontal hull girder bending induced stresses, i.e.,


A

A

swdA⫽ sw·ydA⫽ sw·zdA⫽0
A
(18)

In this respect, the distribution of warping stresses resembles other self-equilibrat-


ing stress fields such as the following:

앫 thermal stresses
앫 shear lag effects, and
앫 residual stresses induced by welding.

Current simplified procedures for calculating the ultimate hull girder strength do
not take into account shear lag effects and thermal stresses, but many such procedures
do take into account residual stresses due to welding. But sensitivity analyses show
that the effect of welding induced residual stresses on the ultimate hull girder strength
can be small for actual ship structures (when the magnitude of compressive residual
stresses is small), see for instance Hansen (1996).
To describe the effect of the warping induced stresses on the collapse we shall first
consider the case where the ship hull cross section is subjected to a given torsional
deformation and on top of that an increasing, say, vertical bending moment.
Self-equilibrating longitudinal stress fields in the ship hull are in this case expected
to have negligible effect on the ultimate hull girder bending moment, since once
yielding has commenced in the most highly stressed regions further bending loading
will produce a redistribution of forces to the lower stressed zones. Provided there is
sufficient ductility, this process will continue until the entire tension flange has
yielded. The compressed part of the hull girder may also have the capacity to redis-
tribute forces across the whole width of the deck or bottom, provided it has stocky
proportions. The same behaviour is also predicted by the upper bound theorem of
the classical theory of plasticity, which says that for sturdy sections where buckling
is not the main failure mode, the collapse load will not be affected by self-equilibrat-
ing stress fields.
J.K. Paik et al. / Ocean Engineering 28 (2001) 1097–1133 1113

Therefore, if the scantlings of the compressed part of the hull girder cross section
are such that buckling does not occur before yielding, then the ultimate hull girder
bending moment cannot be expected to be influenced significantly by direct warping
stresses sw associated with a given torsional deformation.
Of course, the moment–curvature relation will be affected by direct stresses due
to warping since the stiffness of the hull will be reduced. For a discussion of the
effect of warping induced stresses on the stiffness we have to consider the four
stages, namely

앫 the linear elastic regime,


앫 the post-elastic region where the behaviour is influenced by yielding and buckling,
앫 the ultimate stage where the maximum load carrying capacity is reached, and
finally,
앫 the post-ultimate stages where the hull girder has negative stiffness.

In the linear elastic regime, the one-dimensional beam theory outlined in Section
3 and the plane stress finite element method provides excellent agreement with one
another and therefore the stiffness of the hull girder is expected to be well described
by these theories in this region. The effect of warping stresses will only be to reduce
the bending moments that bound this linear elastic regime.
In the post-elastic regime, the principal effect of warping stresses will be to reduce
the stiffness as yielding or buckling progresses from the most highly stressed regions
across the width of the deck or bottom. A result of this spread of yielding at a
relatively early stage will be a permanent deflection of the hull girder upon unloading.
This permanent deflection will depend on the magnitude of warping induced stresses.
Here it may be desirable to ensure that the process of redistribution does not advance
to a stage, which under service conditions leads to undesirable permanent defor-
mations of the hull girder.
As mentioned above, the warping induced stresses are not expected to have a
significant influence of the ultimate load. Provided severe buckling can be excluded,
the only reduction in the ultimate bending moment carrying capability will be caused
by the effect of the torsion induced shear stresses on the direct yield stresses.
The post-ultimate behaviour is governed by the slenderness of both the plate and
the stiffeners constituting the structural cross section. If the hull cross section is very
slender it must be expected that, due to the more advanced post-buckled stage for
cross-sections subjected to severe warping stresses, the decrease in load with increas-
ing deformation will be more pronounced.
If the loading situation is such that the ship hull is subjected to a relatively large
bending moment, and thereafter it is exposed to an increasing torsional moment,
then it is seen from Eqs. (16, 17) and (8) that due to yielding (or buckling) of the
highly stresses parts of the deck and bottom of the hull the warping torsional moment
Mw will be reduced. But that will not necessarily lead to collapse in a torsional mode.
The effect will merely be that a larger part of the torsional moment will have to be
carried as a St. Venant torsional moment.
The conclusion is that since the distribution of warping stresses described in Sec-
1114 J.K. Paik et al. / Ocean Engineering 28 (2001) 1097–1133

tion 3.2 fulfils the orthogonality condition (Eq. (18)), it is then to be expected that
for ship hulls subjected to bending moments that introduce compression in relatively
sturdy parts of the hull girder, the influence of warping stresses on the collapse
load will be insignificant. However, warping stresses will affect the moment versus
curvature relations as previously noted.

5. The ALLPS/HULL computer software

5.1. Benefits of ALPS/HULL

In the present study, the ALPS/HULL program (Paik, 1998a) is used to predict the
ultimate strength characteristics of ship hulls under torsion. ALPS/HULL employs a
special purpose non-linear finite element method designed to reduce modeling effort
and computing time.
By application of the conventional finite element method, non-linear structural
behavior has been analyzed, in several specific cases by a number of investigators,
reasonably well. However, a weak feature of the conventional finite element is that
it requires enormous modeling effort and computing time for large sized structures.
The most obvious way to reduce modelling efforts and computing time 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 non-linear
behaviour of large structural units. Ueda and Rashed (1974, 1984), who suggested
this idea, called it the Idealized Structural Unit Method (ISUM). By developing the
so-called deep girder ISUM unit (i.e., beam-column unit with a deep web), they
showed illustrative examples relating to analysis of the ultimate transverse strength
of ship frame structures efficiently and accurately.
For the ultimate longitudinal strength analysis of ship (plated) structures, Paik
(1987) developed the unstiffened and stiffened rectangular plate ISUM units which
were later applied to the ultimate strength analysis of general plated structures (Paik,
1996; Paik, 1998b). Based on this theory, the ALPS/HULL (Paik, 1998a) which is
a special purpose program for the ultimate strength analysis of ship hulls under
combined loads has been developed. In the ALPS/HULL analysis, a variety of hull
girder load components including vertical sectional shear, horizontal sectional shear,
vertical bending and horizontal bending as well as local panel pressure loads are
accommodated.
The ALPS/HULL program includes:

앫 ease of structural modeling,


앫 efficiency in computational effort, and
앫 acceptable accuracy in the resulting solution.

The program employs five types of the ISUM units which have three translational
degrees of freedom at each nodal point, namely
J.K. Paik et al. / Ocean Engineering 28 (2001) 1097–1133 1115

앫 the beam-column unit


앫 the unstiffened plate unit
앫 the stiffened plate unit
앫 the hard unit
앫 the virtual unit

Hard and virtual units never fail, e.g., they neither buckle nor yield. Virtual units
use negligibly small plate thickness. The ISUM units except for the hard and virtual
units take into account, either singly or in combination:

앫 buckling in compression and/or shear


앫 yielding in tension
앫 necking
앫 rupture due to excessive tension–deformation
앫 interaction effects between local and global system failures of structure
앫 combined bi-axial in-plane loads and shearing force
앫 lateral pressure
앫 fabrication related initial deflections
앫 post-weld residual stresses
앫 plate thickness reduction due to corrosion
앫 structure deterioration or ineffectiveness
앫 premised crack damage
앫 load or displacement control

As part of the present study, the procedures have been updated to deal with the
effects of hull girder torsion. The updated program is in the present study used to
numerically investigate the ultimate strength characteristics of ship hulls with large
deck opening under torsion, or combined torsion and vertical bending.

5.2. Structural modeling

The extent of the model for the ALPS/HULL ultimate hull strength analysis
depends on the condition of load application. When only the vertical bending moment
and/or vertical sectional shear forces are applied, a half hull module between two
neighbouring transverse frames or floors (called the single segment half hull model)
is sufficient because of the symmetry with respect to the vessel centre line. When
a horizontal bending moment or a horizontal sectional shear force is applied, how-
ever, a full hull module between two neighbouring transverse frames or floors (called
the single segment full hull model) is needed for the analysis.
To study warping effects induced by torsion, the complete hull should ideally be
modelled. But to gain insight in the effect of warping stresses, a full hull module
between two neighbouring transverse bulkheads (called the multiple segment full
hull model) is in principle is applied here. As will be discussed later, however, a
full hull module between two neighbouring transverse frames or floors is usually
1116 J.K. Paik et al. / Ocean Engineering 28 (2001) 1097–1133

suitable in practice, even for the load application including torsion. An added benefit
in this case is reduced computer cost.
For studying the warping effects of ship hulls under torsion, a full hull module
between two transverse bulkheads is first used. Fig. 10 shows example ALPS/HULL
models between two transverse bulkheads or between two transverse floors (for the
hypothetical 4300 TEU container vessel as will be studied later), Fig. 10a being
the multiple segment full hull model and Fig. 10b being the single segment full
hull model.
In order to smoothly transmit the torsional moment to the hull cross section, a
hard rectangular plate unit is added at the position of the transverse bulkhead for

Fig. 10. (a) An example of the multiple segment full hull model for the structure between two neighbor-
ing transverse bulkheads. (b) An example of the single segment full hull model for the structure between
two neighboring transverse floors.
J.K. Paik et al. / Ocean Engineering 28 (2001) 1097–1133 1117

the multiple segment full hull model or at the position of transverse floor for the
single segment full hull model.

5.3. Fabrication related initial imperfections

Initial deflections and welding induced residual stresses may reduce the ultimate
strength of the vessel, and should therefore for completeness be included in the
strength calculations. Extensive studies to assess post-weld imperfections in ship
plating have been undertaken by many researchers (for example, Smith et al., 1987).
In the ALPS/HULL calculations, the initial imperfections of individual plate and
beam-column units can be included in the form of initial deflections and welding
induced residual stresses as the analyst desires.
Distribution of welding induced residual stresses are idealized to be composed of
two residual stress blocks, namely a compressive residual stress block and a tensile
residual stress block. Since welding is normally performed along the transverse as
well as longitudinal edges of ship plating, both longitudinal and transverse residual
stresses will develop and affect the stiffness and strength of the structure. The magni-
tude of the residual stresses in the longer direction will be larger because the weld
length is longer. In the ALPS/HULL program, the transverse (plate breadth direction)
compressive residual stress is automatically approximated based on the value in the
other direction, as follows:
b
sry⫽ ·srx (19)
a

5.4. Boundary conditions

For the usual calculations of the ultimate hull girder strength, the degrees of free-
dom at the nodal points at one end of the hull module are restrained in all directions,
namely ship length, depth (vertical) and beam (transverse) directions. At the other
nodal points, all degrees of freedom are set to be free. The boundary conditions for
the single segment full hull model are similar to those for the multiple segment full
hull model. As will be indicated later, other types of the restrained conditions may
be introduced in some particular cases.

5.5. Load application

The hull girder loads are applied incrementally using displacement or load control
through the nodal points at the unrestrained end of the hull module. The magnitude
of lateral pressure loads acting on panels is at once applied in the beginning before
the hull girder loads are incrementally increased. The lateral deflection and stresses
developed by the lateral pressure loads which are internally calculated by an analyti-
cal formulation in the program are regarded as added initial deflection and initial
stresses of each plate unit, respectively.
1118 J.K. Paik et al. / Ocean Engineering 28 (2001) 1097–1133

The ultimate strength of ship hulls under a combination of various hull girder
sectional load components may be calculated numerically either by applying all load
components simultaneously until the ultimate capacity is reached, or by applying
one or multiple load components incrementally until the ultimate limit state is
reached with the rest of the load components at certain constant desired magnitude.
In the ALPS/HULL analysis undertaken here, the hull girder sectional load compo-
nents are incrementally applied in the following four steps, i.e.,

앫 1st step: The torsional moment is incrementally increased until a desired magni-
tude.
앫 2nd step: The vertical sectional shear force is incrementally increased until a
desired magnitude, with the torsional moment at a given fixed magnitude pre-
viously applied.
앫 3rd step: The horizontal sectional shear force is incrementally increased until a
desired magnitude, with the torsional moment and the vertical sectional shear
force each at a given fixed magnitude previously applied.
앫 4th step: The vertical and/or horizontal bending moments with a constant ratio of
both bending curvatures are incrementally increased until the ultimate strength is
reached, with the torsional moment, vertical sectional shear force and horizontal
sectional shear force each at a given fixed magnitude previously applied.

The internal stresses and displacements produced by the previous load applications
are accumulated, and so they would contribute to the change of stiffness and strength
of the structure under the subsequent load application. Fig. 11 shows the method of
load application in the ALPS/HULL analysis. In the following, the load generation
procedure is briefly explained. Note that bending stresses are applied according to
the beam theory but shear stresses are taken to be uniform over any specified struc-
tural members.

5.5.1. Generation of torsion


As indicated in Fig. 11a, the torsional moment is generated by applying a set of
nodal forces, i.e., FH and FV, at only four corner nodes such that the following
equilibrium condition is fulfilled:
FH·D⫽FV·B (20)

The magnitude of torsional moment applied to the hull module is then given by
MT⫽FV·B⫹FH·D (21)
where MT is the torsional moment.
Fig. 12 represents the resulting displacements, e.g., the warping displacement uT
in the x direction, the torsional displacements vT in the y direction and wT in the z
direction, of a thin-walled beam with open section under torsion. It is expected that
these torsional displacements will vary along the beam length. Therefore, the twisting
angle for the thin-walled beam can be calculated by
J.K. Paik et al. / Ocean Engineering 28 (2001) 1097–1133 1119

Fig. 11. (a) Generation of torsion by a couple of torque forces which are applied only at four hull
corners. (b) Generation of vertical sectional shear by uniform nodal displacements which are applied at
all nodal points of the unrestrained hull end section. (c) Generation of horizontal sectional shear by
uniform nodal displacements which are applied at all nodal points of the unrestrained hull end section.
(d) Generation of vertical bending by nodal displacements with respect to horizontal neutral axis which
are applied at all nodal points of the unrestrained hull end section. (e) Generation of vertical bending by
nodal displacements with respect to horizontal neutral axis which are applied at all nodal points of the
unrestrained hull end section.
1120 J.K. Paik et al. / Ocean Engineering 28 (2001) 1097–1133

Fig. 12. Nomenclature: Torsional displacements.

wT vT
q⫽ ⫽⫺ (22)
ys zs

5.5.2. Generation of vertical or horizontal sectional shear


To produce vertical or horizontal sectional shear forces, a set of uniform nodal
displacements are applied at all nodes of the unrestrained end section of the hull
module in the hull depth or beam direction, respectively, see Fig. 11b or c. Due to
the shear forces acting on the edge of the model, moments can be induced. To attain
a condition of pure shear, the moments induced by shear forces should in principle
be removed by applying corresponding moments in the opposite direction. Since the
moments induced by the shear force increments at a certain incremental loading step
are usually small, however, they are simply removed in the next incremental loading
step, i.e., without applying them in the opposite direction.

5.5.3. Generation of vertical or horizontal bending


The bending moment is automatically generated by applying linear axial displace-
ments with respect to the neutral axis at all nodes of the unrestrained hull end section,
see Fig. 11d or e. The position of neutral axis can vary at each incremental loading
step due to any member failure. Such variation of the neutral axis is automatically
calculated in the program, and the axial displacements at the next incremental loading
step are then applied using the updated neutral axis.

6. Ultimate strength characteristics of ship hulls under torsion

6.1. Object ship design

To investigate the ultimate strength characteristics of ship hulls under torsion and
the hypothesis in Section 4, ultimate strength of a hypothetical container vessel is
J.K. Paik et al. / Ocean Engineering 28 (2001) 1097–1133 1121

analyzed. The vessel has been designed to carry 4300 TEU containers. Fig. 13 shows
the midship section design of the vessel. The principal particulars of the vessel are:

앫 Length Between Perpendiculars (L)=283.2 m


앫 Length, scantling=278.7 m
앫 Breadth Moulded. (B)=32.2 m
앫 Depth Moulded. (D)=21.8 m
앫 Draught Moulded., design=12 m
앫 Draught Moulded., scantling (df)=13 m

Spacing of ordinary transverse floors is 3160 mm, but at mid-length of each hold
where a cell guide is positioned it is 1580 mm. When the hull cross section is
simplified as indicated in Fig. 2a, the position of shear centre is 8529.2 mm (i.e.,
0.391D) below the base line. The design torsional moment of the vessel as calculated
from Eqs. (1, 5a) and (5b) is 52,357 tonf-m.

Fig. 13. Midship design of a hypothetical 4300 TEU container vessel.


1122 J.K. Paik et al. / Ocean Engineering 28 (2001) 1097–1133

6.2. ALPS/HULL modelling

To investigate the effect of warping stresses on the ultimate strength of ship hulls
under torsion, a multiple segment full hull model between two neighbouring trans-
verse bulkheads is used as shown in Fig. 10a. Three cases are considered:

앫 Case I with fully free warping displacements


앫 Case II with partly restrained warping displacements
앫 Case III with entirely restrained warping displacements

For case I, the axial (longitudinal) displacement degrees of freedom at all nodes
except for one node at the unloaded end of the model on the outer bottom centre
line (for preventing the rigid body motion) are set to be free. At all nodes of the
unloaded end of the model, the y- and z-displacement degrees of freedom are fixed.
All degrees of freedom at the rest of nodal points are fully free. In this case, axial
(warping) stresses will not be developed by torsion since warping displacements
occur freely, but also the torsional rigidity will be very small because warping
restraints are not introduced at all.
For case II, all degrees of freedom at the nodes of the unloaded end of the model
are fixed and the rest are set to be fully free. In this case, since warping displacements
are partly restrained both axial (warping) and shear stresses are developed by torsion.
This case is more representative of real ship hull structures.
For case III, a model which is composed of a set of shear panels is considered
such that warping displacements are entirely prevented. All degrees of freedom at
the nodes of the unloaded end of the model are fixed, and also axial displacement
degrees of freedom at the nodes of all hull segment cross sections inside the model
are restrained. In this case axial stresses are not developed because axial strains of
individual plate units are zero and shear stresses are predominant. Due to the
restraints, the torsional rigidity will be very large in this case.
The progressive collapse behaviour of a ship hull is studied as pure torsion is
increased incrementally. To test the applicability of a smaller model, two kinds of
the structural models, namely the multiple segment full hull model (Fig. 10a) and
the single segment full hull model (Fig. 10b), will be used and the two sets of results
will be compared.
A series of ultimate hull strength analyses with varying hull girder load compo-
nents are carried out to investigate the ultimate strength interaction characteristics
of ship hulls under combined vertical bending and torsion. For this purpose, the use
of a smaller (i.e., shorter) hull model would be preferable because much computer
time can be saved. As will be discussed later, the single segment full hull model
(the smaller model) will be found to give reasonably accurate solutions of the ulti-
mate strength behavior of ship hulls under torsion, and therefore it is employed for
developing the ultimate hull girder strength interaction relationships.
In the present ALPS/HULL calculations, the rule minimum value of yield strength
of the material is used. The ultimate tensile strength of material was presumably
taken to be 1.4 times the yield strength. Strain hardening effects subsequent to yield-
J.K. Paik et al. / Ocean Engineering 28 (2001) 1097–1133 1123

ing were approximately accounted for through a small Young’s modulus, i.e., 1%
of Young’s modulus. The net scantlings after removing the corrosion margin were
used. No premised crack damage in the structure was considered. Also, the following
values of initial imperfections were assumed for plate units, i.e.,

앫 wo/t⫽0.15

앫 srx/so⫽⫺0.15

앫 b
sry⫽ ·srx
a

6.3. Calculated results and discussion

Fig. 14 shows the torsional moment versus twisting angle curves for the typical
4300 TEU container vessel hull, for all three above-mentioned cases (I, II, and III)

Fig. 14. Effects of warping on the torsional moment versus twisting angle curves of the hypothetical
4300 TEU container vessel hull under pure torsion.
1124 J.K. Paik et al. / Ocean Engineering 28 (2001) 1097–1133

until the ultimate strength is reached. The values of the torsional moment and the
twisting angle were calculated from Eqs. (20) and (21), respectively.
As would be expected, the torsional rigidity of the model for case I, i.e., with
fully free warping displacements, is very small. With increase in the degree of warp-
ing restraints, the torsional rigidity increases. As previously noted, warping restraints
are partly introduced at cross sectional discontinuities of real ship structures. There-
fore, the case II is more representative of real ship structures.
Figs. 15–17 show the distribution of torsion induced warping and shear stresses
in the hull cross section for each case immediately after the ultimate strength is
reached. The stresses shown were those observed for the hull segment of the
unloaded end of the model and they are mean values over the individual ISUM
plate units.
It is seen from Fig. 15 for case I that the values of axial (warping) stresses are
very small and shear stresses are dominant. Fig. 16 for case II with partly restrained
warping displacement shows that both axial (warping) and shear stresses are
developed. As previously discussed in Section 4, the warping stress distribution takes
a self-equilibrating pattern over the hull cross section, e.g., composed of compressive
stresses at some regions and tensile stresses at the other regions. The hull corner
areas, i.e., bilge and topside part, are “hot spot” (highly stressed) regions. At the
bottom plating around the vessel centre line and at the side shell around the mid-

Fig. 15. Stress distribution of the hypothetical 4300 TEU container vessel hull section at the ultimate
limit state under torsion for the case (I) with fully free warping displacements using the multiple segment
full hull model.
J.K. Paik et al. / Ocean Engineering 28 (2001) 1097–1133 1125

Fig. 16. Stress distribution of the hypothetical 4300 TEU container vessel hull section at the ultimate
limit state under torsion for the case (II) with partly restrained warping displacements using the multiple
segment full hull model.

height of the hull, the stresses become nearly zero. The stress distribution at the
ultimate limit state is similar to that in the linear elastic regime as indicated in Section
3. Fig. 17 for case III with entirely restrained warping displacements indicates that
shear stresses are dominant and the values of axial (warping) stresses are very small.
Returning to Fig. 14, we can compare case II (with the presence of both warping
and shear stresses) with case III (with only shear stresses) to see the effect of warping
stresses on the ultimate strength of ship hulls under torsion. The difference in the
ultimate strength between the two cases is 7.8%, even though the torsional rigidity
is quite different from each other. Therefore, it can be said that the effect of warping
stresses on the ultimate torsion strength of ship hulls is small, and the torsional
stiffness needed is ensured by the warping restraints, as will be discussed later.
Certainly, the cause of structural failure in case III is due to torsion induced shear
stresses, but the collapse occurs by the action of both axial and shear stresses in
case II. As discussed in Section 4, the self-equilibrating warping stress field should
not be expected to have significant effect on the ultimate strength of ship hulls under
torsion. This is consistent with the results in Fig. 14.
Fig. 18 shows the distribution of the three displacement components for case II
(i.e., uT, vT, wT as defined in Fig. 12) at the hull corner along the vessel length at
the ultimate limit state. It is seen from Fig. 18 that warping (axial) displacements
1126 J.K. Paik et al. / Ocean Engineering 28 (2001) 1097–1133

Fig. 17. Stress distribution of the hypothetical 4300 TEU container vessel hull section at the ultimate
limit state under torsion for the case (III) with entirely restrained warping displacements using the multiple
segment full hull model.

occur anti-symmetrically at the hull cross sections. Also, it is seen that the distri-
bution of torsional displacements along the vessel length is non-linear.
The ultimate strength based safety factor for the vessel, which is defined as the
ratio of the ultimate torsional moment (for the natural warping case) to the design
torsional moment is 4.05, which is sufficient enough to withstand the applied extreme
torsional moment. The first collapse at a local structural member occurred at 68,995
tonf-m, see Fig. 14. The first collapse based safety factor, which is defined as the
ratio of the strength corresponding to first collapse to the design torsional moment
is then 1.32. In summary, the effect of warping stresses on the ultimate strength of
ship hulls is small, and the torsional rigidity of ship hulls is ensured mainly by the
warping restraints in the structure.
Fig. 19 shows the progressive collapse behaviour of the container vessel under
torsion alone, until the ultimate strength is reached. In the calculations, the warping
displacements are partly restrained by applying the case II restrained condition men-
tioned above. The two sets of results, i.e., those obtained by using the multiple
segment full hull model and the single segment full hull model are compared in the
figure. It should be noted that the length of the single hull segment model is only
about 12% of the multiple hull segment model and thus the torsional rigidities of
both models can not be compared directly. It is seen that the difference of the ultimate
strength between the two models is small, about 2.3%.
J.K. Paik et al. / Ocean Engineering 28 (2001) 1097–1133 1127

Fig. 18. (a) Distribution of torsional displacements along the line a–a⬘ for the case (II) with partly
restrained warping displacements. (b) Distribution of torsional displacements along the line b–b⬘ for the
case (II) with partly restrained warping displacements. (c) Distribution of torsional displacements along
the line c–c⬘ for the case (II) with partly restrained warping displacements. (d) Distribution of torsional
displacements along the line d–d⬘ for the case (II) with partly restrained warping displacements.

Fig. 20 shows axial (warping) stress and the shear stress distribution for the single
segment full hull model at the ultimate limit state. As previously discussed, it is
seen from Fig. 20 that at the bilge and topside region, relatively higher levels of
stresses are developed, while at the bottom plating around the vessel centre line and
at the side shell around the mid- height of the hull, the stresses become nearly zero.
Since warping displacements of the single hull segment model are more restrained
than the multiple hull segment model, the magnitudes of warping stresses in the
single hull segment model are slightly larger than those of the multiple segment
model.
By comparing stress distributions of Fig. 20 with those of Fig. 16 (i.e., of the
multiple full hull segment model), and also referring back to Fig. 19, it is seen that
the single segment full hull model predicts the ultimate torsion strength reasonably
accurately. The single segment hull model can thus be used for the ultimate strength
1128 J.K. Paik et al. / Ocean Engineering 28 (2001) 1097–1133

Fig. 19. Comparison of the multiple segment full hull model with the single segment full hull model
results on the progressive collapse behavior of the hypothetical 4300 TEU container vessel hull under
pure torsion.

analysis of ship hulls, with concomitant savings in modeling effort and computer
time.
In summary, when pure torsion is applied, a ship hull would reach the ultimate
limit state if the hull corner regions (e.g., bilge and topside area) which are typically
the most highly stressed areas, collapse. Thus, scantlings of the hull corner region
should be sufficient for ship hulls with large deck openings in order to prevent hull
girder collapse due to torsion. Also, the single segment full hull model is capable
of giving a reasonably accurate prediction of the ultimate strength compared to the
multiple segment full hull model. Therefore, the single segment hull model should
be useful for ultimate strength analysis of ship hulls under a general set of hull girder
loads. To investigate the influence of torsion on the ultimate vertical bending moment
and also to develop the ultimate hull strength interaction relationship between torsion
and vertical bending, a series of ultimate strength analyses were carried out for the
typical 4300 TEU container vessel hull. The single segment full hull model was used
for this purpose. In the calculations, torsion is incrementally increased until a desired
magnitude, and then the vertical hogging or sagging bending moment is additionally
increased until the ship hull reaches the ultimate strength. To see the influence of
torsional moments on the ultimate vertical bending moment capacity, the magnitude
J.K. Paik et al. / Ocean Engineering 28 (2001) 1097–1133 1129

Fig. 20. Stress distribution of the hypothetical 4,300 TEU container vessel hull section at the ultimate
limit state under torsion, using the single segment full hull model.

of the torsional moment itself is varied. Fig. 21 shows the vertical bending moment
versus curvature curves for the typical vessel hull, varying the magnitude of torsion.
The magnitude of the vertical bending moment shown has been calculated by inte-
grating the existing axial stresses with respect to the horizontal neutral axis for the
entire hull cross section. It should be noted that the existing axial stresses are the sum
of torsion induced warping stresses and the vertical bending induced axial stresses. It
is seen from Fig. 21 that torsion is not a very sensitive load parameter affecting the
ultimate vertical bending moment, as long as its magnitude is not predominant.
Based on the calculated results, the ultimate hull strength interaction relationship
for the typical 4300 TEU container vessel is obtained as in Fig. 22. Shown in Fig.
22 is the ultimate hull strength under combined torsion and vertical bending, nor-
malized by the ultimate hull strength under torsion or bending alone. The following
formulations fit the ultimate hull strength interaction relationships between torsion
and vertical bending reasonably well:
For hogging:

冉 冊 冉 冊
MV
MVU
3.7

MT
MTU
3.7
⫽1 (23)

For sagging:
1130 J.K. Paik et al. / Ocean Engineering 28 (2001) 1097–1133

Fig. 21. Vertical bending versus curvature curves of the hypothetical 4300 TEU container vessel hull
varying the magnitude of torsion.

冉 冊 冉 冊
MV
MVU
3.1

MT
MTU
3.1
⫽1 (24)

7. Concluding remarks

The primary aim of the present study has been to investigate the ultimate strength
characteristics of ship hulls with large hatch openings under torsion. Warping may
to some extent play a role in the ultimate hull girder behaviour. Therefore, the effect
of warping stresses on the ultimate hull girder strength was studied by a numeri-
cal procedure.
As part of this study, the special purpose non-linear finite element program
ALPS/HULL for the efficient analysis of the ultimate strength of ship hulls subjected
to a combination of five hull girder sectional load components (torsion, vertical shear,
horizontal shear, vertical bending and horizontal bending) and one local load compo-
nent (lateral pressure loads) has been developed based on the idealized structural
unit method (ISUM). Using the ALPS/HULL program, the progressive collapse
behaviour of a hypothetical container vessel under torsion was analyzed. Three cases
ALPS/HULL structural analyses for the vessel are considered in investigating the
J.K. Paik et al. / Ocean Engineering 28 (2001) 1097–1133 1131

Fig. 22. Ultimate hull strength interaction relationship between torsion and vertical bending for the
hypothetical 4300 TEU container vessel.

effect of warping. In the first, termed the fully free warping case, the warping dis-
placements were freely allowed. In the second, termed the partly restrained warping
case, one end of the model was fixed. In the third, termed the entirely restrained
warping case, the model was made up of a set of shear panels. By comparing the
results from the three cases, the characteristics of torsional stiffness and the ultimate
torsion strength of the vessel are discussed as a function of the degree of warping
restraints. The influence of torsion on the ultimate vertical bending moment of the
typical container vessel hull was also studied, and ultimate hull girder strength inter-
action relationships between torsion and vertical bending were developed for the
same vessel.
From the present study, the following conclusions can be drawn:

1. When warping displacements occur freely, no axial (warping) stresses are


developed and shear stresses are predominant. However, since there are no warp-
ing restraints, the torsional rigidity is very small.
2. When warping displacements are partly restrained, both warping and shear stresses
are developed at the hull cross section. The distribution of warping stresses takes
a self-equilibrium pattern composed of tensile stresses in some areas and com-
pressive stresses in other areas.
3. When the hull module is modelled as an assembly of shear panels and torsional
load alone is carried, shear stresses are predominant. In this case the torsional
rigidity can be large.
1132 J.K. Paik et al. / Ocean Engineering 28 (2001) 1097–1133

4. For the typical 4300 TEU container ship hull considered in the present study, the
ultimate torsional moment for the case with partly restrained warping displace-
ments was 92.7% of the case with entirely restrained warping displacements. The
difference in the ultimate strength between the two cases is about 7%. It was
evident that the effect of warping stresses on the ultimate torsion strength was
small as long as the warping restraints are ensured and elastic buckling is pre-
vented. But it should be noted that the torsional stiffness may be significantly
affected by warping. Also, torsion induced shear stresses will of course affect the
ship hull ultimate bending strength.
5. By taking the single segment model between two neighbouring transverse frames
(instead of the multiple segment model between two transverse bulkheads) as the
extent of the ultimate hull strength analysis, a sufficiently accurate prediction of
the ultimate strength behavior could be obtained. The difference in the ultimate
torsion strength predictions between the two types of models was about 2% for
the typical container vessel.
6. Based on the ultimate hull strength interaction studies between torsion and vertical
bending undertaken in the present study, it is seen that torsion is not a very sensi-
tive load component affecting the ultimate vertical bending moment of ship hulls
as long as the magnitude of torsion is not predominant. However, it should also
be noted that the ultimate bending strength of ship hulls with low torsional rigidity
can be reduced significantly when torsional moments are large.

Acknowledgements

The present study has been undertaken with support from the American Bureau
of Shipping who is thanked for support. The authors are grateful to Prof. A. Ostap-
enko of the Lehigh University who kindly provided valuable data on the ultimate
strength of box girders under torsion. The second author was with the American
Bureau of Shipping when this work was carried out. The views expressed in this
paper are those of the authors and are not necessarily those of the institutions they
are affiliated with.

References

ABS, 1995. Guide for dynamic based design and evaluation of bulk carrier structures, The SafeHull
Project, American Bureau of Shipping, March.
Hansen, A.M., 1996. Strength of midship section. Marine Structures 9, 471–494.
Hong, D.P., Kim, O.H., Lee, Y.S., 1987. Analysis of structural damage of a large ore/coal carrier. In:
Proc. of Spring Meeting/STAR Symposium, The Society of Naval Architects and Marine Engineers,
Philadelphia, May.
Hughes, O.F., 1988. Ship structural design, A rationally-based, computer-aided optimization approach,
The Society of Naval Architects and Marine Engineers, New Jersey.
J.K. Paik et al. / Ocean Engineering 28 (2001) 1097–1133 1133

Jensen, J.J., Pedersen, P.T., Petersen, J.B., 1992. Stresses in containerships. Jahrbuch der Sciffbautechn-
ischen gesellscahft 86, 143–150.
Ostapenko, A., 1981. Strength of ship hull girders under moment, shear and torque. In: Proceedings of
Extreme Loads Response Symposium, SSC-SNAME, Arlington, VA, October.
Ostapenko, A., Moore, T.R., 1982. Maximum strength of ship hulls subjected to moment, torque and
shear. Report No. 479.5, Fritz Engineering Laboratory, Lehigh University, November.
Ostapenko, A., Vaucher, A., 1980. Ultimate strength of ship hull girders under moment, shear and torque.
Report No. 453.6, Fritz Engineering Laboratory, Lehigh University, July.
Ostergaard, C., Dogliani, M., Guedes Soares, C., Parmentier, G., Pedersen, P.T., 1996. Measures of model
uncertainty in the assessment of primary stresses in ship structures. Marine Structures 9, 427–447.
Paik, J.K., 1987. Ultimate Strength Analysis of Ship Structures by the Idealized Structural Unit Method.
PhD thesis, Osaka University, January.
Paik, J.K., 1996. The idealized structural unit method: Theory and practice. Pusan National University,
Department of Naval Architecture and Ocean Engineering, March.
Paik, J.K., 1998a. ALPS/HULL user’s manual (version 1998). Pusan National University, Department of
Naval Architecture and Ocean Engineering, August.
Paik, J.K., 1998b. ALPS/GENERL user’s manual (version 1998). Pusan National University, Department
of Naval Architecture and Ocean Engineering, August.
Paik, J.K., Thayamballi, A.K., Che, J.S., 1996. Ultimate strength of ship hulls under combined vertical
bending, horizontal bending and shearing forces. Trans. SNAME 104, 31–59.
Pedersen, P.T., 1983. A beam model for the torsional-bending response of ship hulls. Trans. RINA, The
Royal Institution of Naval Architects 125, 171–182.
Pedersen, P.T., 1985. Torsional response of container ships. J. of Ship Research 29 (3), 194–205.
Pedersen, P.T., 1991. Beam theories for torsional-bending response of ship hulls. J. of Ship Research 35
(3), 254–265.
Pedersen, P.T., Jensen, J.J., 1983. Strength of Marine Structures, Part 1 — Loads and global analysis (In
Danish). Den Private, IngeniØrfond.
Pittaluga, A., 1978. Recent developments in the theory of thin-walled beams. Computers and Structures
9, 69–79.
Senjanovics, I., Grubisic, R., 1979. Coupled horizontal and torsional vibration of ship hull with large
openings. In: Proc. of Euromech. Col. 122, Numerical Analysis of the Dynamics of Ship Structures,
ATMA, Paris.
Smith, C.S., Davidson, P.C., Chapman, J.C., Dowling, P.J., 1987. Strength and stiffness of ship’s plating
under in-plane compression and tension. Trans. RINA 130, 277–296.
Ueda, Y., Rashed, S.M.H., 1974. An ultimate transverse strength analysis of ship structures. J. of the
Society of Naval Architects of Japan 136, 309–324(in Japanese).
Ueda, Y., Rashed, S.M.H., 1984. The idealized structural unit method and its application to deep girder
structures. Computers and Structures 18 (2), 277–293.

Você também pode gostar