Vous êtes sur la page 1sur 25

Ship Structural Design

Course Documentation

Ship Structure Design

Contents
1

INTRODUCTION........................................................................ 3

MATERIALS .............................................................................. 4
2.1 MATERIALS USED IN SHIPBUILDING ..................................................... 4
2.2 STEEL IN SHIPBUILDING .................................................................. 4
2.3 STEEL IN MATERIAL SCIENCE ............................................................ 6

STRUCTURAL DESIGN ............................................................... 9


3.1
3.2
3.3
3.4
3.5

FAILURE MODES ........................................................................... 9


STILL-WATER BENDING MOMENT ...................................................... 10
DIMENSIONING ACCORDING TO LONGITUDINAL STRENGTH ........................ 12
OTHER LOADS ........................................................................... 13
MODELING ............................................................................... 15

BACKGROUND FOR CLASS RULES (BULKHEADS) .................... 17


4.1 GENERAL REMARKS ON BULKHEADS ................................................... 17
4.2 PLATES ................................................................................... 17
4.2.1 Stiffeners ......................................................................... 19

(RE-)DESIGN FOR FATIGUE .................................................... 22

REFERENCES .......................................................................... 25

2
DNV GL Rev.1.0

Ship Structure Design

Introduction

Ship structural design differs from design in mechanical engineering. Some characteristic
differences are:
Mechanical engineering
kinematic/kinetic
gears, bearings, screws etc
Screws, bolts, adhesives, etc.
2-d, 3-d, explosion drawings

Function
Elements
Connection
Plans

Ship structural design


static
plates, profiles, brackets
welding, adhesives
largely 2-d

Initial design affects detailed (structural) design. Ignoring structural design in the initial design
stage can lead to considerable problems in subsequent design stages, often connected with
increased steel mass and construction cost. E.g. unsuitable distribution of ballast may lead to high
bending moments requiring stronger (and heavier) scantlings.
Ship structural design and marine engineering need to work closely together. There are many
interfaces. Example: Structural support for main and auxiliary engines. Here geometrical
requirements (position of engine and bolts to structural support) and stiffness requirements must
be met. This often requires larger double bottom height in the engine room. Other constraints
come from the assorted large pipes running in the engine room (e.g. for cooling water).
For ship structural design, an engineer needs to bring together knowledge for assorted aspects of
his craft:

Knowledge
Knowledge
Knowledge
Knowledge

about
about
about
about

the material used


the production processes
regulations pertaining to structural design of ships
analysis methods as needed for first principle design

Design for production starts with understanding the production processes. In the following we will
look at the technological developments of ship production.
Structural design rules of classification societies and navies are based on physics often enhanced
by some empirical corrections, e.g. for material imperfections, environmental uncertainties and
other deviations from ideal assumptions. A fundamental understanding of general strength analysis
is essential for the structural designer. While the ship is a complex 3-d structure, fundamental
beam theory can already aid much of the initial dimensioning and helps in a qualitative
understanding of many of the rules aiding structural design in ships and offshore platforms. This
lecture requires a very concise (and also somewhat superficial) treatment of the topic. Basics of
technical strength analysis can be found in many textbooks. These are recommended for further
studies, particularly if the here employed concepts of strength analysis are missing or forgotten.

3
DNV GL Rev.1.0

Ship Structure Design

2.1

Materials

Materials used in shipbuilding

The following materials are used in shipbuilding:

Rolled plain steel for usual applications (plates and profiles for ship hull, foundations, etc.)
Rolled special steel (high-tensile steels, low-temperature steels e.g. for LNG tankers, corrosion
resistant steel for product tankers, non-magnetic steel for compass area)
Cast steel (parts of rudder, stern, stem)
Forged steel (parts of the equipment like anchors, chains, rudder shaft)
Light metals, particularly aluminum (compass area, superstructure, boats)
Plastic and wood (interior equipment, boats, pipes)

Steel is by far the most important material for shipbuilding. Aluminum, titanium, fiber-reinforced
plastics (FRP) and composite materials play a relatively small role. Typical material costs in the
year 2000 were:

500 /ton for shipbuilding steel


2900 /ton Aluminum
7500 /ton FRP
15000 /ton composites

Supply and demand introduce considerable fluctuations in raw material prices. However, the basic
fact remains that steel is much cheaper than other materials for large ships.
The higher-value materials are predominantly used for fast ships (V > 35 knots), yachts and navy
ships. In the former Soviet Union, titanium has been used as a shipbuilding material, particularly
for submarines.

2.2

Steel in shipbuilding

The steel used in shipbuilding comes as plates or profiles:

Plates
Plate dimensions differ from shipyard to shipyard. Actual plate dimensions follow from many
aspects: maximum possible dimensions, volume sections, necessary steps in forming plates,
scrap, etc. The maximum plate dimensions depend on external factors (e.g. railway limitations)
and internal factors (crane facilities, plate storage, etc.). Thickness of plates is increased in
steps of 0.5 mm. Classification rules come often in the form of formulae giving the required
thickness of plates. These are rounded down to steps of 0.5 mm if they exceed by 0.2 mm
maximum, otherwise rounded up. E.g. 10.67 mm becomes 10.5 mm, 10.73 mm becomes 11
mm. The steel mass of plates is generally taken as 8 kg/m2mm, accounting for margins in
rolling.
Standard plates are kept (always) in storage; typical dimensions are (dimensions always in
mm):
8000 2400 56.5
12000 2700 712
Standard plates have the following advantages and disadvantages:
+ simple and cheap storage, exchangeability and availability
more scrap
The alternative is commissioned material which is ordered specifically for each ship. The
percentage of standard plates and commissioned plates differs again from shipyard to
shipyard. The plates are delivered within margins of accuracy following EN 29, Table 2.I. The
plates can thus have considerable initial deformations.

4
DNV GL Rev.1.0

Ship Structure Design

Table 2.I: Admissible deviations from ideal plane for standard normal plane,
Euronorm EN 29; all dimensions in mm
Admissible deviation from plane
Nominal measured
1500 2000 2500
thickness length
<1500 <2000 <2500 <2750
3<4
1000
7
8
10
2000
14
18
22
4<6
1000
7
7
8
10
2000
14
16
21
25
6<8
1000
7
7
8
8
2000
14
14
16
17
8<10
1000
7
7
7
7
2000
13
13
14
15
10<20
1000
7
7
7
7
2000
12
12
12
13
20<50
1000
7
7
7
7
2000
11
11
12
12
50
1000
Individually agreed

for nominal plate width


2750 3000 3250
<3000 <3250 <3500
8
20
8
8
9
16
18
20
7
8
8
13
16
18
7
8
8
12
13
14

3500
<4000
9
20
9
16

General strength properties of all steels are:


Youngs modulus
Poisson coefficient
Shear modulus

2.1105 N/mm2
0.3
0.81105 N/mm2

Table 2.II: shows examples of plate consumption for some ship types.

tanker (280 000 tdw)


tanker (200 000 tdw)
Bulker (150 000 tdw)
Tanker (120 000 tdw)
containership (4500 TEU, 40 000 tdw)
LNG tanker (50 000 tdw)
containership (1000 TEU, 20 000 tdw)

Required
[t]
42000
28000
20000
17000
14000
11000
5000

plates Plate thickness [mm]


13-58
13-55
14-55
12-48
13-78
9-48
12-65

Profiles
Rolled or built profiles are used as stiffeners of plates in shipbuilding. The most popular profiles
are:
a)
b)
c)
d)

Holland profiles following EN 10067


Angular profiles following DIN 1028 / DIN 1029
Built profiles from flat steel welded using fillet welds
Profiles with a flange or corrugated plates

Profiles are ordered as plates following standard lengths or commissioned to a certain


application. Usually only one profile form (often Holland profiles) is kept on stock to simplify
storage management and assembly plans.
HP and L profiles are used for small and medium stiffeners being cheaper than built profiles. Lprofiles are less available in qualities required in shipbuilding and feature a stronger asymmetry
making them more susceptible to fold over, but they offer more section modulus per mass
(important e.g. for reefers). The relatively thin flange of the L-profiles allows easier longitudinal
butt welds than for HP-profiles. Built profiles from flat steel are employed for large stiffeners
(longitudinal deck and bottom stiffeners in large ships etc.).
High-tensile steels are increasingly used. However, care should be taken in using high-tensile
steels. Corrosion and fatigue strength are not improved, cost are higher, ductility and safety
margin between elastic yield and fracture reduced.
5
DNV GL Rev.1.0

Ship Structure Design

A 3500 TEU containership requires approximately 12000 t steel of different quality in plates and
profiles. From the base material, approximately 37000 steel plate parts and 28000 profile parts are
made. This requires approximately 130,000 cutting meters and 450,000 welding meters. Cutting
and welding processes depend on steel quality. The steel quality can be tailored towards certain
properties by adding other elements, by the thermo-mechanical treatment and by the rolling
process.

2.3

Steel in material science

"Steel" denotes iron-based material which is suitable for rolling. The essential difference between
ordinary steel and pure iron is the amount of carbon in steel which reduces the ductility but
increases the strength and susceptibility to hardening when rapidly cooled from elevated
temperature. With very few exceptions (containing high amount of chromium), steel contains less
than 2% carbon. The carbon content of cast iron is higher. Steel may be alloyed or not. Alloyed
steel contains further substances such as chromium, nickel, etc. beyond minimum threshold
values. Various micro-structures may be obtained by different heat-treatments.
Non-alloyed steels are important for shipbuilding as welding material. High-tensile steels are
generally alloyed quality steels. Steels are classified according to EU 156-80 norm in Europe.
Classification societies have adapted these norms and supply tables with classification of steels and
corresponding quality requirements concerning chemical composition and production method. For
shipbuilding steels, four characteristics are of prime interest:

strength

ductility (lack of brittleness)

suitability for welding

corrosion properties
The desired properties sometimes conflict and are not easy to obtain. The steel production process
generally follows several steps allowing enough ductility to improve the desired properties as far as
possible.
Metals form on a sub-microscopic level grid structures. We distinguish two elementary forms: facecentred cubic and body-centred cubic, Fig.2.1. Certain substances can exist in two or more
crystalline forms, e.g. graphite or diamonds are allotropic modifications of carbon. Allotropy is
characterized by a change in atomic structure which occurs at a definite transformation
temperature. Four changes occur in iron, denoted by , , , . Of these, , , and are bodycentred, and is face-centred. Iron has thus two allotropic modifications. The change in atomic
structure from austenite (body-centred) to ferrite (face-centred) in cooling down is marked by a
significant contraction.

Fig.2.1: body-centred cubic and face-centred cubic


The pure forms are an ideal. In real materials, the grids contain assorted errors which influence the
material properties (like strength). The sub-microscopic grids form crystallites or grains which are
separated by grain boundaries (unoccupied grid positions, discontinuities in orientation of grid etc).
The number and distribution of grain boundaries depends on thermo-mechanical production
process. Welding changes the grain structure near the weld. Grains range in size from several m
to 1 mm.

6
DNV GL Rev.1.0

Ship Structure Design

Fig.2.2: Iron-Carbon diagram: temperature and carbon content determine microscopic structure
The iron-carbon diagram (Fe-C diagram), Fig.2.2, shows the various forms of iron/steel with the
technical terms denoting the microscopic structure. Only the left part of the Fe-C diagram is
important for shipbuilding steels.

Austenite

Ferrite

Martensite

Perlite
Fig.2.3: Microscopic structures of iron/steel

7
DNV GL Rev.1.0

Ship Structure Design

The various forms differ in the arrangement of carbon in the iron matrix, the crystalline structure
and the extent of carbon between grains. The main forms are, Fig.2.3:
Ferrite

is body-centred cubic (-Fe), low strength, ductile and magnetic.

Austenite

is face-centred cubic (-Fe), highly ductile and not magnetic. Austenite at room
temperature is only possible in alloyed steels.

Cementite

is the crystallized iron carbide Fe3C. It is very hard and brittle, of high strength, but
not ductile.

Perlite

denotes lamellar deposits of iron carbide in ferrite crystals. In the perlite mutation,
carbon leaves the -grid before the sudden conversion to -grid

Martensite

is the instable conversion product of the austenite. Martensite is formed when the
cooling is too rapid to allow perlite formation. The carbon atom then is trapped
inside the -grid which induces high internal stresses. Tempering (hardening) of
steel is based on this process.

Bainite

is the structure form between Perlite and Martensite, consisting of ferrite needles
with embedded carbides

During heating and cooling, this structure changes. Rapid cooling creates e.g. Martensite which is
brittle. Reheating and slow cooling results then again in ductile ferrite structures.
The properties of a steel are determined by chemical composition and crystalline structure. These
are influenced by the metallurgical composition (alloy), rolling technique and heat treatment. Steel
is produced in several steps, starting with raw iron and recycled steel. Raw iron is useless as basic
material for construction as the high contents of carbon, sulphur and phosphor make it brittle and
unsuitable for welding. Thus the contents of elements making the material brittle (C, P, S, Si, Mn)
have to be reduced already in the raw iron. Further processes turn raw iron and recycled steel to
raw steel. Small and homogeneous grains are desirable for both strength and ductility. This can be
achieved e.g. by adding small quantities of aluminium during the desoxidation process of the steel
production.
Modern steel rolling equipment allows detailed control of temperature and rolling pressure. The
time history of pressure and temperature (together with the chemical composition of steels)
determines the structure of the steel. Heat treatment allows a finer, more homogeneous
redistribution of the carbon in the steel resulting in better strength and ductility. Rapid cooling
leads to brittle microscopic structures. This is a problem in welding, i.e. the plate will have good
strength properties, but the weld will introduce local transitions.

8
DNV GL Rev.1.0

Ship Structure Design

Structural Design

3.1

Failure modes

The whole structure should be designed sufficiently strong, but with a minimum of financial effort.
''Sufficiently strong'' encompasses various aspects depending on the function of a structural
member and how often failure may be acceptable:

Yield: Usually design rules require that the total stress in a structural member stays below
an elastic yield limit, i.e. failure is defined as plastic deformation.

Buckling: Structures under axial compression may suddenly buckle. This buckling leads to
plastic deformation as the load continues to act after the buckling initiates.

Rupture: For exceptional (maybe once-in-a-lifetime) loads, plastic deformation due to


bending, torsion or buckling is acceptable. If the stresses stay below the rupture limit, the
structure will then deform, but still continue to function. This ultimate strength philosophy
applies to ship collisions or shock impact for navy ships.

Fatigue: Ships and offshore structures are subject to time-dependent loads due to seaway,
vibrations from engine and propeller, and changing load conditions. Even if the individual
load cycles remain below the elastic strain limit, after millions of load cycles the
microscopic structure of the material changes. This is called fatigue of the material.
Microscopic cracks grow under continued load cycles typically with a speed of 10-8 to 10-3
mm per load cycle. Eventually the crack reduces the residual strength of the structural
member to an extent that the structure fails. Fatigue strength of structures depends on
many aspects like load collectives (amplitudes and frequencies of load cycles), stress
concentration, microscopic properties of the material etc. For practical design, a
compensation factor accounts for the fatigue strength reducing the permissible static
stress. This compensation factor depends on the structural detail and is high for sharp
corners. The factors are usually derived employing model tests with typical structural
details.

Vibrations: Vibrations (usually coming from propeller or engine) are not only a problem in
terms of fatigue strength, but can in themselves be unacceptable, e.g. in passenger ships,
for certain measuring equipment etc. Over the decades, the trend in ships has been
towards lighter structures and higher installed power. This has lead to increasing vibration
problems. Today, a structural designer must also consider excessive vibration as a failure
mode to be avoided.

Deflections: For some structures, excessive deflection means failure. This applies for
foundations for main engines, supports for propeller shafts etc. If these structural elements
deflect too much, the supported parts will not function, e.g. a propeller shaft will destroy its
bearing.

Usually for the dimensioning of the main structural elements, we assume a static load, even though
in reality the ship is subject to a time-dependent (dynamic) load due to the seaway. The static load
model simplifies the computation considerably and is sufficient for most cases.
The failure mode then specifies either upper limits for stress or strain (deformation) in the
structures. Loads on the ship are due to pressure from the water and weights from the ships own
weight and its cargo. These loads are distributed as local stresses through the whole structure of
the ship. Global bending and torsion stresses superimpose with local loads (e.g. weights from an
engine, a cargo item, local wave impact). The structural design must now ensure that everywhere
there is sufficient safety against all failure modes. Usually one or two failure modes dominate the
design.
Structural design follows a top-down approach. The loads are considered rather globally and the
ship is considered as a first approximation as a beam. This results e.g. in a required section
modulus W for the midship cross section to avoid longitudinal stresses exceeding the elastic stress
limit1. Buckling considerations require then longitudinal and transverse stiffeners. These may be
locally strengthened for additional local loads. Manufacturing technology and the need to have

1 Usually the upper deck is the most critical region in this respect requiring very thick plates for containerships.

9
DNV GL Rev.1.0

Ship Structure Design

pipes and cables running through the ship require assorted cut-outs in the structure and the
individually stiffeners and bulkheads need to be connected by smaller connecting plates assuring
sufficient strength and stiffness (against buckling). These connections are usually particularly
sensitive to fatigue strength and are designed in the end. Strength considerations and
manufacturing consideration need often compromises. Usually the manufacturing consideration
determine the general type of design, i.e. the general geometry of stiffening profiles, and the
strength considerations determine the actual thickness of the material employed.
The structural weight of the ship is a main weight item reducing the available payload which earns
the income of the ship. This is particularly critical for fast ships where the engine and fuel already
account for a considerable weight item. On the other hand, a very intelligent light-weight design,
balancing the stress distribution in all structural members as evenly as possible, will result in an
expensive ship in manufacturing. Reasonable compromises are found employing experience
(qualitative understanding of individual factors) and numerical simulations.

3.2

Still-water bending moment

In a first simple global consideration, the ship hull is considered as a beam subject to external and
internal forces (hydrostatic and hydrodynamic forces, weight). Following basic static principles, the
forces must be in equilibrium:
F=0

M=0

and

Systems are statically determined if the unknown internal forces (bending moment and transverse
force) can be determined from the equilibrium conditions. The freely floating ship hull is a statically
determined system. One distinguishes between still-water loads and additional loads in seaways.

Fig.3.1: Transverse force; C = cargo, ls = light ship, ER = engine room


The ship hull in still water can be considered as a free-free supported beam. Thus bending moment
and transverse force must both be zero at the forward and aft ends. The load on the ship hull
consists of weight and lift forces, Fig..3.1. Both are differently distributed over the length of the
ship. The sum (integral) of each force component must be the same, and centre of gravity and
buoyancy must have the same longitudinal position. Otherwise the ship would trim and sink until
this condition is fulfilled. The integration of the difference between weight w(x) and lift forces b(x)
from one ship end to a cross section gives the transverse force Q at that cross section:
x

Q( x) = ( w( x) b( x)) dx
0

The transverse force Q(x) is integrated to get the bending moment:


x

M ( x) = Q( x) dx
0

10
DNV GL Rev.1.0

Ship Structure Design

These computations are performed in practice by CAD programs, varying draft and trim
systematically until the residual forces and moments at the ship ends are zero.
A complete computation of the longitudinal strength includes also the deflection of the ship hull.
This computation requires the Youngs modulus E and the longitudinal distribution of the (area)
moment of inertia of the cross sections. The parts considered in this computation must at least run
over a major part of the ship length. This yields then the section modulus for deck WD and the
section modulus for the bottom WB, Fig.3.2.

Fig.3.2: Position of neutral fiber for cargo ship;


''Neutrale Faser''=neutral fiber, ''bei Trockenfrachtern''=for cargo ships
For a first estimate of effective shear areas, only the vertical ship hull and longitudinal bulkheads
are to be considered. The shear stresses are later computed more precisely in a finite element
analysis (FEA).
Continuous hatch coamings are considered as part of the longitudinal carrying structures.
Germanischer Lloyd (GL) takes the distance of the outer fiber as
eD' = z (0.9 + 0.2y/B)

eD' > e_D

z and y are the distance of the upper edge of the hatch coaming from the neutral fiber and centreplane, respectively. This determines the maximum longitudinal stress acting on the upper edge of
the ship hull. The superstructure is not considered as a supporting member in the longitudinal
strength analysis of the ship hull.

D =

M '
eD
I0

respectively

B =

M
eB
I0

The shear stress follows from:

Q S

I0 t

S = moment of the structurally effective cross section with respect to the neutral fiber, t = plate
thickness.
Assuming in first approximation I =const., we get the approximation for the deflection:

wB ( L / 2) = M ( L / 2)

L2
11.4 EI

The still-water bending moment can be influenced by shifting masses. This changes the trim while
keeping the displacement constant.
11
DNV GL Rev.1.0

Ship Structure Design

3.3

Dimensioning according to longitudinal strength

The dimensioning of civilian ships for longitudinal strength follows the rules of Classification
Societies. Navies usually pass their own rules which are usually stricter than the rules for cargo
ships, but increasingly cooperate with Classification Society. The rules are harmonized between the
individual Classification Societies to ensure a common safety standard for the most important
strength criteria. These rules specify e.g. maximum permissible stresses. The design equation for
longitudinal strength is:
MT perm Wmin
MT = MSW + MWV = total bending moment
MSW = still-water bending moment
MWV = (vertical) additional wave bending moment
The minimum section modulus Wmin and the permissible stress perm are specified. The additional
wave bending moment can be hardly influenced. This leaves the calm-water bending moment as
main design quantity. To meet the above condition, the design should be such that the calm-water
bending moment is small enough for the minimum section modulus Wmin to suffice. Otherwise, the
section modulus has to be increased.
The following condition is internationally agreed, Nitta et al. (1992):
The minimum section modulus in [cm3] is
Wmin =k C L2 B (CB + 0.7)
where L and B are in [m].
C is the wave parameter:

C = 10.75 [(300-L)/100]1.5
C = 10.75
C = 10.75 [(L-350-L)/150]1.5

k is a material constant:

k=1
k = 0.78
k = 0.72

for 90 L 300 m
for 300 < L 350 m
for
L > 350 m

for standard mild steel


for HT 315
for HT 355

The vertical wave bending moment in hogging condition is specified as:


MWV,hog = 0.19 M C L2 B CB [kNm]
The factor M describes the distribution over the ship length:
M = 2.5 x/L
M=1
M = (1-x/L)/0.35

for 0 x/L 0.4


for 0.4 < x/L 0.65
for 0.65 < x/L 1

The vertical wave bending moment in sagging condition is specified as:


MWV,sag = -0.11 M C L2 B (CB+0.7) [kNm]
Thus the ratio of hogging to sagging moments is:

M WV , hog
M WV , sag

0.19 C B
1.73
=
0.7
0.11(C B + 0.7)
1+
CB

The value for CB is determined as the maximum of 0.6 and the actual block coefficient. The
hogging moment is thus always smaller than the sagging moment.

12
DNV GL Rev.1.0

Ship Structure Design

Combine the requirement for a minimum section modulus with the additional wave bending
moment in sagging condition to get the dynamic stress in seaways:

=
For the hogging condition we get

M WV , sag
Wmin

= 110

= MWV,hog = 89 MPa
= MWV,hog = 110 MPa

[MPa]

for CB = 0.6
for CB = 1.0

The permissible maximum stress amidships for standard mild steel is perm = 175 MPa. Thus less
than half the permissible stress is left for the smooth-water bending moment. High-tensile steels
allow higher values depending on the yield stress: perm = 175/k.
Similar rules apply to transverse forces. Details are found in the rules of the Classification
Societies.
The formulas for the permissible additional wave bending moments are based on a probability of
10-8 over the life span of the ship, Nitta et al. (1992).
Bending moments and transverse forces in calm water are generally computed in a detailed
manner. They depend on load condition, amount and distribution of provisions (fuel, fresh water,
lubrication oil, etc.) and ballast water. Thus calm-water moments and forces change over the
voyage of the ship. Classification societies prescribe computations for begin and end of the voyage
as well as for intermediate conditions. The dimensioning is then based on the maximum load of
these conditions.
For most ship types, a suitable arrangement of the ballast compartments and load cases allows to
achieve permissible stresses with the minimum section modulus. For some ship types, e.g. ro-ro
ships, this target can often not be reached.

3.4

Other loads

Ship in a seaway: Natural seaway is irregular and described by spectra which can be
interpreted as a superposition of regular elementary waves of different amplitude, length
and direction. Basics of ship seakeeping and spectra can be found in Bertram (2000). Here
we consider for general understanding only the effect of a sinusoidal regular wave of same
length as the ship length ( = L). The largest bending moments occur near this wave
length. The computation is performed quasi-static. The ship is then put in the wave in such
a way that it is in equilibrium for trim and sinkage. The computation of the internal forces
follows in principle the same way as for still water. The section areas are taken from a
section area curve (or computed from a CAD description) for the appropriate local draft in
the wave. The difference in transverse forces and bending moments with respect to the
still-water case are computed from the difference lift distribution between still water and
wave integrating again over the ship length. These difference internal forces are then
superposed to the still-water internal forces.

Torsion: Apart from longitudinal bending, torsion plays a major role. Fig.3.3 explains
qualitatively why the hull is subject to torsion in oblique waves. The eccentric action of the
lift resultant force induces a torsion moment. (This consideration is simplified omitting
some further effects responsible for torsion moments such as mass distribution and radius
of inertia around the longitudinal axis.) Simulation tools for ship seakeeping (strip
methods) can compute the torsion moments in natural seaway considering also the ship
motions. But these computations require some effort and dedicated software. A simple
estimate for the maximum torsion moment [kNm] in seaways is:
MTor,wave = C 0.11 L B2
The torsional moment is particularly important for short, wide ships because the ship width
affects the moment quadratically. The absolute value of the wave torsional moment is in
general much smaller than the longitudinal bending moment. It causes significant stresses
only in particularly soft (with respect to torsion) ships.

13
DNV GL Rev.1.0

Ship Structure Design

Fig.3.3: Torsion moments in oblique waves; G = weight resultant force of the cross
section; B = lift resultant force of the cross section
Container ships with their quasi-open cross sections are particularly critical in this respect.
It is now much more difficult to specify an adequate cross section quantity (corresponding
to the midship section modulus) to determine the resulting stresses. The main problem is
that torsion deflection and bending deflection are always coupled in a container ship. Thus
under the action of torsion moment, there are both shear and normal stresses. Fig.3.4
shows a ship-like body under torsion with a typical pattern of deflection. Furthermore, the
computation of the stresses due to torsion is complicated as closed and open cross section
alternate and the ship hull is not prismatic (particularly short parallel midbody for most
container ships). Thus finite element analyses have to be employed early in the design to
give a realistic prediction for this case.

Fig.3.4: Deflection of ship hull due to torsion


For container ships, we can then no longer simply compare load moments, but must
compare resulting stresses. The difference between permissible total stresses and dynamic
total stresses yields the calm-water stresses. These are the results of the combination of
longitudinal stresses and torsional stresses. The dynamic total stresses depend on the draft
for these ship types. Thus for each draft (load case) a separate computation is required to
give a stress balance which then is compared to permissible stresses.

14
DNV GL Rev.1.0

Ship Structure Design

3.5

Modelling

The structural analyses of modern maritime structures employ predominantly the finite element
analyses (FEA). FEA can solve both static and dynamic (vibration) problems. The basic idea is to
discretize a complex structure into a finite number of simple elements, Fig.3.5. All elements are
described by nodal points which are connected by straight lines or curves to form the element's
geometry.
The deflections within the individual elements are described approximately as functions of the
deflections of the nodal points. The synthesis follows then from the condition of continuity, i.e.
shared nodal points between elements must have the same deflection. This results in a system of
linear equation for the nodal deflections. The size of the system of linear equation corresponds to
the number of nodes and degrees of freedom. The system is sparse and special algorithms exist for
fast solution. The solution of the system of equations yields the relation between loads and node
deflections. Then further information like stress distribution is easily derived.
In the past, classification societies gave simple rules for dimensioning plates and girders, e.g.
giving directly thickness or requiring section modulus. The trend is towards first-principle design,
i.e. design based on analysis rather than experience. Increasingly classification societies now
include a FEM stress analysis as alternative or even substitute old rules by a rule specifying only a
maximum stress level. This means more freedom for the designer, but also a more sophisticated
and expensive design process. Even if the stress analysis is optional, following the simple old
approach means over-dimensioned, non-competitive ships.

Fig.3.5: Types of finite elements; top row: truss, beam, plain-strain plate; bottom row: plate,
volume, shell
There are many books on finite element methods, outlining the general theory. There also are
several commercial multi-purpose finite element programs with associated documentation.
However, in addition considerable experience is required to generate suitable and efficient finite
elements grids with the associated input (moments of inertia, etc.) quickly. This experience is
difficult to maintain with only occasional analyses. Shipyards therefore often out-source finite
element analyses, predominantly directly to the classification society who is also in charge of
approving the analysis. The Rules of DNV GL give guidelines for global FEA for containerships and
local FEA for fatigue.
In making the FEA model, one needs to define first what information is required in what accuracy.
This determines the level of detail of the model and thus the required effort for grid generation and
computation. Consider the crane in Fig.3.6 as an example. Fig.3.7 shows as an elementary truss
model which allows to determine how the force p is transmitted versus hanger and boom as a
function of the boom angle 1. This task can still be solved manually without the help of a
computer. Fig.3.8 shows a refined model using beam elements for the crane column allowing also a
simple estimate of forces and bending moment in the foundation. Plate elements are required for a
detailed analysis of the stress in the foundation and the crane structure. Fig.3.9 shows a complete
ship hull model with the integrated crane.

15
DNV GL Rev.1.0

Ship Structure Design

Fig.3.6: Ship cross section with crane

Fig.3.7: Truss model of a crane boom

Fig.3.8: Beam model of crane boom

Fig.3.9: Complete FEA model with ship hull and


crane

16
DNV GL Rev.1.0

Ship Structure Design

4.1

Background for Class Rules (Bulkheads)

General remarks on bulkheads

The function of a bulkhead is compartmentation (for separation of cargo, fuel, ballast water, etc.
and for damage stability) and transverse strength.
Minimum number and arrangement of watertight bulkheads follows from various aspects:
1. Minimum requirements from classification society
A collision bulkhead has to be arranged in a distance d from the forward perpendicular (or a
corrected reference line for ships with bulbous bow). The collision bulkhead extends from
bottom to freeboard deck. For cargo ships the distance d follows from:
min(10 m, 0.05 L) d 0.08 L
There are special regulations for ships with bow doors (ferries). Details are given in the
regulations of classification societies.
A watertight bulkhead has to be arranged at each end of the engine room. A watertight
after-peak bulkhead has to be arranged where the propeller shaft leaves the ship. In
modern ships, the after-peak bulkhead coincides usually with the aft end bulkhead of the
engine room.
2. Damage stability requirements
Additional bulkheads may be necessary to meet damage stability requirements.
3. Arrangement of bulkheads
Watertight bulkheads should usually lie within one section plane (x=const). If this cannot
be realised the horizontal parts must also be made to be watertight. The bulkheads extend
usually up to the freeboard deck. For ships with a long forecastle, the collision bulkhead
extends up to the first deck above freeboard deck.
Load assumption for dimensioning bulkheads assumes water pressure on one side. The assumed
water height is 1 m above the upper edge of the collision bulkhead. The reason is possible draft
increase and heel angle in case of damage. The density of water is taken as 1000 kg/m3 for
convenience. (The regulations have their roots in a time when calculations were done manually.)
So the inherent safety of assuming larger pressure height is somewhat reduced in reality.
The dimensioning of bulkheads consisting of flat plat with stiffeners follows from the reasoning
given below.

4.2

Plates

The simplified mechanical model considers a plate strip of width 1 rigidly supported at both ends.
The length of the strip is the stiffener spacing a, the thickness is t. The strip is under constant
distributed load q = p1. (The pressure due to water height 1 m above upper edge of bulkhead.)

a
17
DNV GL Rev.1.0

Ship Structure Design

q a2
12

For linear-elastic material behaviour, the maximum moment appears at the ends:

M =

The section modulus for a rectangular cross section of height t and width 1 is:

W = t2/6

This yields a maximum stress:

max =

M
p a2 6 p a2
=
=
W
12 t 2
2t2

We employ the usual dimensions for the individual quantities: max [N/mm2], p [kN/m2], a [m], t
[mm]. If the quantities are taken in these dimensions, the formula is re-written as:

max = 10 3

p a2
2t2

This yields the minimum required thickness for a given maximum stress (all quantities in the
dimensions given above):

t = a 10 2

p
20 max

The classification rules give:


t [mm] = a [m] cp

tmin [mm] = 6.0

with

+ tk

f =

235
REH

the material number for static load

tk is the corrosion addition; so after a certain time with corresponding corrosion, the bulkhead still
has sufficient strength. The corrosion addition depends on thickness of the plate and what whether
the bulkhead limits a tank (accelerated corrosion).
If we neglect the corrosion addition we obtain within linear-elastic theory:

max,el =

500
C p2

GL gives

where max,el [N/mm2]

for collision bulkheads: Cp = 1.1

Cp = 0.9

for other bulkheads:


This then converts to:

500 R EH

= 1.76 R EH
1.12 235
500 R EH
=

= 2.63 R EH
0.9 2 235

max, el =

for collision bulkheads

max, el

for other bulkheads

Thus the classification society accepts exceeding the elastic yield limit in the outer layers of the
plate, i.e. partial plastic deformation.
Within an idealised theory of plastification, the linear stress distribution changes to a stress
distribution where the maximum elastic yield stress is reached in a step function over the thickness
of the structure:
18
DNV GL Rev.1.0

Ship Structure Design

stress distribution

linear-elastic regime

after plastification
layers

in

outer

complete plastification

In principle the structure will then support twice the bending moment of the linear-elastic limit
case, before complete plastification is reached. The safety margin of the classification society is
then 2.0/1.76=1.14 for collision bulkheads. The other bulkheads have a safety factor < 1 against
complete plastification. This considers that at larger deformations further structural reserves
appear due to membrane stresses and material hardening.

4.2.1

Stiffeners

The pressure height h is taken (following the rules of GL) from the middle of the unsupported
length l to 1 m above upper edge collision bulkhead or bulkhead deck. We distinguish two forms of
end support:

h
L/2

hp

0.6hp

simple support

fixed support

pressure height h

Fixed support is assumed if the stiffener ends are connected via brackets to stiff structures like
longitudinals or floor plates. The simplified mechanical model for a stiffener with fixed support is:

l/2

l/2

The approximation for the maximum bending moment in the beam (exact for triangular load) is:

M max

a p l2
10
19

DNV GL Rev.1.0

Ship Structure Design

where p is taken at half-length of the beam.


The required section modulus according to GL is (in typical dimensions):
W [cm3] = Cs a [m] l2 [m2] p [kN/m2]
Within linear-elastic theory, we then have:

max, el =

M
1
=
W 10 C s

kNm 100
2
cm 3 = C N / mm
s

For collision bulkheads we have:

Cs=0.33f

For other bulkheads we have:

Cs=0.265f

]
100 REH

= 1.3 REH
0.33 235
100 REH
=

= 1.6 REH
0.265 235

max, el =
max, el

Again, we have to consider the safety against complete plastification. Let us assume in slight
simplification that we have a triangular distributed load:
q(x) = 2 q (1-x/l)

where q=pa is the load at half-length of the beam

The transverse force is then:

Q ( x) =

2
qx 2
ql 2qx +
3
l

The location of the maximum field moment is where the transverse force is zero:

1
x0 = l1
= 0.423 l
3

The maximum field moment is:

M(x0) = 0.128ql2

The moment at the beam end is: MF = 0.064ql2 = ql2/15.6


The load where the elastic yield limit appears in the outer layer of the beam is (in typical
dimensions):

q el = R EH

10
W el
l2

The plastic limit load in the theoretical limit of complete plastification is:

q pl = R EH

15.6
W pl
l2

For typical profiles in shipbuilding, the plastified section modulus is 20% to 50% higher than the
elastic section modulus, thus on average Wpl = 1.35 Wel. This then yields a plastic limit load:

q pl =

15.6
1.35 q el = 2.1 q el
10

The safety against complete theoretical plastification is qpl /qperm . This gives a safety of
2.1/1.3=1.6 (or 60% margin) for collision bulkheads and 2.1/1.6=1.3 (or 30% safety margin) for
other bulkheads. A similar calculation gives 1.7 for collision bulkheads and 1.35 for other bulkheads
for simply supported stiffeners:

20
DNV GL Rev.1.0

Ship Structure Design

l/2

l/2

The two support types thus give similar safety against theoretical complete plastification.

21
DNV GL Rev.1.0

Ship Structure Design

(Re-)Design for Fatigue

Fatigue causes structural failure by repeated low-amplitude loading of material, resulting in


microscopic cracks forming and growing until the structure fails by rupture. Factors influencing
fatigue include: load (amplitudes, frequency, time history of load spectra), quality of detailed
design (avoiding stress concentration), quality of ship construction (alignment, welding,
deformations, etc), level of corrosion, material (steel, aluminium). In all cases, the crack initiation
is due to stress concentration in the structure. Solutions are:

Simple replacement without design change if the fatigue damage appears after many
years accepting repeated repair

Reduction of vibration amplitudes causing the fatigue

Redesign of local structure for better fatigue characteristics (e.g. increasing scantling
thickness, avoiding sharp corners and openings, rounding edges, eliminating hard spots)
and/or construction standards (e.g. heat treatment, grinding of welds)

Total redesign involving possibly different materials

Generally, the preferred option is today a redesign paying attention to fatigue aspects.

Stress concentrations appear at geometric discontinuities:

hard spots (rigid structure ending in the middle of a relatively soft structure)

ends of stiffeners and brackets; these zones are particularly prone to fatigue cracks and
thus examined with care by experienced surveyors. The solution consists often in
modifying a brackets shape rather than increasing scantling thickness, adding brackets to
soften discontinuities, or continuing structures to next strong structural zone.

significant change in cross section, Fig.5.1; probability of problems increases with


magnitude of difference in cross section and abruptness of transition between the two
cross sections. The solution consists in making the transition gentler, e.g. by introducing
brackets with suitable large radius of curvature.

significant change in thickness; probability of problems increases with magnitude of


difference in thickness and abruptness of transition between the two scantlings. The
solution consists in making the transition gentler, having at least a ratio of 1:3 in the
transition.

openings; they reduce the structural cross section and cause in addition stress
concentration. Solutions consist in inserting a plate of larger thickness, adding ring
stiffeners or orthogonal stiffeners around the opening, modifying the shape of the opening
to decrease stress concentration or increasing the spacing between openings. Sometimes
the opening can be omitted completely.

misalignments; already a misalignment by half a scantling thickness increases the stress


concentration considerably. The solution is to rectify the misalignment (and pay due
attention to proper alignment in new construction)

3-d intersections, Fig.5.2; complex 3-d intersections with hard points are sometimes not
avoidable. Here special care has to be taken to avoid excessive stress concentration.

Geometric discontinuities can also be small-scale:

pores

inclusions

effects of excessive heat

cracks

grindings

22
DNV GL Rev.1.0

Ship Structure Design

Fig.5.1: Crack at change of cross section

Fig.5.2: 3-d intersection with cracks in


transverse stiffener and horizontal plate

In the following, several case studies show typical damages and repair solutions:

Fig.5.3: Fatigue damage at bracket and repair solution


damage: crack in bracket

repair

Fig.5.4: Fatigue damage at bracket and repair solution

23
DNV GL Rev.1.0

Ship Structure Design

damage : cracks in both adjacent stiffeners

repair

Fig.5.5: Fatigue damages at brackets and repair solutions


damage: crack in supporting longitudinal

repair

Fig.5.6: Fatigue damage at stiffener end and repair solution


damage: crack in welding

repair

Fig.5.7: Fatigue damage in bracket at hatch coaming and repair solution


damage: crack in outer plating

repair

Fig.5.8: Fatigue damage at abrupt change in cross section and repair solution

24
DNV GL Rev.1.0

Ship Structure Design

damage: crack at abrupt transition

repair

Fig.5.9: Fatigue damage at transition from small to big stiffener and repair solution

damage: crack at hatch corner

repair: double plate

Fig.5.10: Fatigue damage at hatch corner and repair solution

Acknowledgement
Parts of these notes are based on the book Grundzge des Schiffbaus of Prof. Dr.mult. Eike
Lehmann and on lecture notes of Prof. H.J. Petershagen (TU Hamburg-Harburg).

References

BERTRAM, V. (2000), Practical Ship hydrodynamics, Butterworth & Heinemann, Oxford.


NITTA, A.; ARAI, H.; MAGAMO, A. (1992), Basis of IACS unified longitudinal strength standard,
Marine Structures, Vol.5, pp.1-21.

25
DNV GL Rev.1.0

Vous aimerez peut-être aussi