Rules for Classification and Construction

V Analysis Techniques
1 Hull Structural Design Analyses

2 Guidelines for Fatigue Strength Analyses of Ship Structures

Edition 2004
 The following Guidelines come into force on August 1st, 2004

Alterations to the preceding Edition are marked by beams at the text margin.

Germanischer Lloyd Aktiengesellschaft

Head Office
Vorsetzen 35, 20459 Hamburg, Germany
Phone: +49 40 36149-0
Fax: +49 40 36149-200
[email protected]

www.gl-group.com

"General Terms and Conditions" of the respective latest edition will be applicable
(see Rules for Classification and Construction, I - Ship Technology, Part 0 - Classification and Surveys).

Reproduction by printing or photostatic means is only permissible with the consent of

Germanischer Lloyd Aktiengesellschaft.

Published by: Germanischer Lloyd Aktiengesellschaft, Hamburg

Printed by: Gebrüder Braasch GmbH, Hamburg
V - Part 1 Table of Contents Chapter 2
GL 2004 Page 3

Table of Contents

Section 1 General
A. Objective of the Guidelines ........................................................................................................ 1- 1
B. Objective of a Fatigue Strength Analysis ................................................................................... 1- 1
C. Analysis Methods ....................................................................................................................... 1- 1
D. Stress Types ............................................................................................................................... 1- 2

Section 2 Modelling of the Structure

A. General ....................................................................................................................................... 2- 1
B. Calculation Models for Global and Local Nominal Stresses ...................................................... 2- 1
C. Calculation Models for Structural Stresses at Welded Joints ..................................................... 2- 2
D. Calculation Models for Notch Stresses at Plate Edges ............................................................... 2- 3

Section 3 Definition of Load Cases and Calculation of the Stress Spectrum

A. General ....................................................................................................................................... 3- 1
B. Simplified Deterministic Analysis .............................................................................................. 3- 1
C. Spectral Method ......................................................................................................................... 3- 3
D. Simulation of the Stress History ................................................................................................. 3- 3

Section 4 Assessment of the Results

A. General ....................................................................................................................................... 4- 1
B. Assessment of the Fatigue Strength ............................................................................................ 4- 1
V - Part 1 Section 1 C General Chapter 2
GL 2004 Page 1–1

Section 1

General

Preamble B. Objective of a Fatigue Strength Analysis

The proof of sufficient fatigue strength, i. e. the 1. The objective of a fatigue strength analysis is
strength against crack initiation under dynamic loads normally:
during operation is useful for judging and reducing
– the identification of structural areas with high
the probability of crack imitiation of structural mem-
cyclic stresses
bers during the design stage.
– the analysis and assessment of critical structural
Due to the randomness of the load process, the details
spreading of material properties and fabrication fac- – the determination of structural alternatives
tors and to effects of ageing, crack initiation cannot be
completely excluded during later operation. Therefore
among other things perodical surveys are necessary. 2. The identification of structural areas with
high cyclic stresses requires the calculation of stress
ranges and mean stresses in the areas under considera-
tion for one of the stress types mentioned in D. to-
gether with a suitable representation of the results that
A. Objective of the Guidelines allows the identification of critical details.

3. The analysis and assessment of critical struc-

1. These guidelines give further information tural details and the determination of structural alter-
about the performance of fatigue strength analyses of natives is normally performed under consideration of
ship structures in addition to the Construction Rules the expected stress history as well as the local stress
for Seagoing Ships (references of this kind hereafter concentration or a corresponding detail category.
are related to Section 20 of Chapter 1, Hull Structures,
of the Construction Rules, I – Ship Technology, Part 1
– Seagoing Ships). In view of the great variety of ship
structural details, different load effects and available C. Analysis Methods
analysis methods, it is necessary to keep these guide-
lines relatively general.
1. Three methods are available for considering
the wave loads which dominate the fatigue behaviour
of ship structures:
2. In accordance with the Construction Rules for
Seagoing Ships, it is assumed that the fatigue strength – simplified deterministic method, see 2.
is assessed on the basis of design S-N curves and the – spectral method, see 3.
Palmgren-Miner's Rule for the consideration of vari-
able-amplitude loading. In cases, such as initiated – simulation of the stress history, see 4.
cracks or welds with partial penetration, however,
All three methods, which differ in their computation
other approaches, e.g. those based on fracture mechan-
effort, will be dealt with in Section 3 in more detail.
ics, can be suitable for the assessment. They can re-
quire other calculation models and analysis proce-
dures. 2. In the simplified deterministic analysis, se-
lected load situations are considered and from this,
stress spectra for the whole service life are derived,
3. In addition to the information regarding the adopting the assumptions in the Construction Rules
performance of fatigue strength analyses, which are for Seagoing Ships for the subsequent strength as-
illustrated by examples, attention should be given to sessment.
the detail design to aim at a fatigue-resistant structure,
an objective which is still of great importance because 3. When applying the spectral method, the
the complete assessment of all details under considera- structural response to stochastic load processes is
tion of all possible load effects is not yet practicable. calculated in the frequency domain, and the stress
Chapter 2 Section 1 D General V - Part 1
Page 1–2 GL 2004

spectrum for the whole service life is determined with from relatively coarse finite element models. The
statistical methods for the subsequent fatigue strength effective breadth has to be considered. Further guid-
assessment. ance for the computation of nominal stresses by finite
element analysis is contained in Chapter 1 – Guide-
4. Contrary to the spectral method, the simula- lines for Strength Analyses of Ship Structures with the
tion of the stress history is performed in the time do- Finite Element Method.
main, making it possible to consider the direct calcula-
tion of nonlinear effects during the load and damage 3. Welded structures which cannot be related to
process. a classified detail or where additional stresses occur
which are not or not fully considered by the cata-
5. The selection of the method depends primar- logued detail can be assessed with respect to fatigue at
ily on the question as to whether acceptable results can weld toes on the basis of the structural or hot-spot
be expected in view of the respective simplifications stress σs, which contains the stress increase due to the
and assumptions. In the case of the simplified deter- structural geometry, but not that due to the weld toe.
ministic method, which is at present mostly applied In some cases, structural stress concentration factors
for practical reasons, the main question is whether the Ks can be used for the determination of the structural
selected load situations consider sufficiently the rele- or hot-spot stress σs,
vant stresses and whether the shape and number of
cycles of the stress spectrum can be estimated with
satisfactory accuracy on the basis of experience and/or σs
similar computations. When applying the spectral Ks =
σn
method, it is necessary to check how far simplifica-
tions such as the neglect or approximate consideration
of nonlinear effects can be justified. In the case of a which result from parametric investigations. Here, the
simulation of the stress history, the length of simula- definition of the nominal stress and the validity ranges
tion time which allows extrapolation to the whole of the parametric formulae or diagrams have to be
service life with sufficient accuracy is of primary considered. The fatigue strength assessment is per-
interest. formed using specific S-N curves or detail categories
for structural stresses, as given in the Construction
Rules for Seagoing Ships.

D. Stress Types 4. Free plate edges are usually assessed on the

basis of the notch stress σk, which is computed for
1. Depending on the detail considered, the fa- ideally elastic material behaviour. In some cases, the
tigue strength analysis is normally based on one of the notch stress can be calculated with the aid of the theo-
following three types of stresses: retical stress concentration factor Kt resulting from
parametric investigations:
– nominal stress σn, see 2.

– structural or hot-spot stress σs at the toes of σk

welded joints, see 3. Kt =
σn
– notch stress σk at rounded plate edges, see 4.
Here, the definition of the nominal stress and the va-
2. In welded structures, the fatigue strength lidity ranges of the parametric formulae or diagrams
have to be considered. The fatigue strength assessment
analysis is normally based on the nominal stress σn in
is performed by using specific S-N curves or detail
the structural member at the location of the detail categories for free plate edges.
considered and an appropriate relation to a classified
detail according to the catalogue of details in the Con-
struction Rules for Seagoing Ships which also de- 5. As regards further types of stresses, e.g. the
scribes the direction of the relevant stress component. notch stress at weld toes, the elastic-plastic stress and
The nominal stress is usually calculated from integral strain in the notch root or the stress intensity at crack
load quantities and sectional properties (force or mo- tips or sharp notches, reference is made to the con-
ment per unit area or section modulus, respectively) or tinuative technical literature.
V - Part 1 Section 2 B Modelling of the Structure Chapter 2
GL 2004 Page 2–1

Section 2

Modelling of the Structure

A. General B. Calculation Models for Global and Local

Nominal Stresses

1. The modelling of the structure has to be per- 1. The calculation of nominal stresses is nor-
formed in such a way that, at the detail considered, the mally performed by using beam models of the ship’s
type of stress on which the fatigue assessment is to be hull girder, of the primary structural members as well
based can be calculated with sufficient accuracy. as secondary structural members, or by using rela-
tively coarse finite element models to represent these
structures. The latter method is recommended espe-
cially for more complex structures and load cases.
2. Two methods are normally applied in the
calculation: 2. It is common to distinguish between global
stresses in primary structural members and additional
– analytical methods, e.g. based on beam theory, local stresses in secondary structural members
which usually result in nominal stresses (frames, girders, stiffeners), see Fig. 2.1. The latter
include bending stresses in stiffeners and plates. In
– numerical methods, such as the finite element fatigue analyses, the combined effect of both compo-
method, which are particularly well suited for nents is to be taken into account.
the stress analysis of geometrically complex
structures as well as local stress increases
3. The extent of the model is primarily deter-
mined by the question of whether and how reasonable
boundary conditions can be formulated for partial
3. In the following special information is given models of the structure. Fig. 2.2 shows a typical over-
about the modelling of a structure in view of calculat- all model of the ship structure with a relatively coarse
ing the desired type of stress, see Section 1, D. Further mesh. As a result, global nominal stresses are obtained
information is contained in Chapter 1 – Guidelines for in the primary structural members; these have to be
Strength Analyses of Ship Structures with the Finite superimposed with local stresses as well as increased
Element Method. stresses due to reduced effective breadths.

Detail A: Detail B: Detail C:

Girder Stiffener Bottom Plate

Neutral
Axis

A B C

Global Stresses Local Stresses

Fig. 2.1 Global Stresses and Local Stresses

Chapter 2 Section 2 C Modelling of the Structure V - Part 1
Page 2–2 GL 2004

a)
b)
Type a)
Fig. 2.2 Overall model of a ship's structure a)

4. The resulting nominal stresses can

– be directly assessed in connection with a corre-

sponding detail category
Fig. 2.3 Types of weld toes
– be multiplied by stress concentration factors Kt
or Ks to obtain and assess locally increased 3. When applying the finite element method, it
notch stresses σk or structural (hot-spot) stresses is recommended in view of effects due to plate bend-
ing that models with plate or shell elements, or alter-
σs (see Section 1, D.3. and Section 1, D.4.) natively with solid elements, be used. It should be
observed that on the one hand the arrangement and
– serve as boundary conditions for local models of type of elements used have to allow the formation of
the structural details considered (alternatively, plate bending, and on the other hand only the linear
deformations are frequently prescribed at the stress distribution in the plate thickness direction has
model boundaries); see also C. and D. to be evaluated in accordance with the definition of
structural stress (i.e. neglect of the local stress increase
due to the weld toe). Fig. 2.4 shows the relationship
5. If crack initiation from the weld root is as-
sessed, it should be noted that the nominal stress has between different types of stresses at a weld toe of
type a) and type b), respectively.
to be related to the cross-sectional area of the weld;
see also the description of the respective detail catego-
ries in the Construction Rules for Seagoing Ships. sk = Reference Point
Bracket
Toe
ss Axial Stress at
Plate Surface
C. Calculation Models for Structural Stresses
sn
at Welded Joints
Weld sn

1. Structural or hot-spot stresses are normally

calculated by using local models of the ship structure. t Axial Stress Plate
The extent of the model has to be chosen such that in Plate
effects due to the boundaries on the structural area Weld toe type a): t/2 t
considered are sufficiently small and reasonable Weld toe type b): 5 10 mm
boundary conditions can be formulated.
Fig. 2.4 Types of stress at a weld toe of type a)
2. Basically, it is necessary to distinguish be- and type b), respectively
tween two types of weld toes at fillet or K-butt welds:
4. Recommended meshes with plate or solid
type a) weld toes on plate surfaces, e.g. at trans- elements are shown by the example in Fig. 2.5.
verse stiffeners or at the weld around the
toe of an attached plate When using thin plate or shell elements, which have to
be arranged in the mid-plane of the structural compo-
type b) weld toes at plate edges, e.g. at the termina- nents, the welds are usually not modelled. Exceptions
tion of a cut edge. are offsets between plates (e.g. doubling plates) or
small distances between adjacent welds, where the
Fig. 2.3 shows both types of weld toes by the example stiffness of the welds may be considered by inclined
of a welded structure. It should be noted that local plate elements, reinforced plate strips (Fig. 2.6) or by
plate bending may affect the structural stress, espe- introducing constrained equations (or rigid links) to
cially with type a). couple the node displacements and rotations. Gener-
V - Part 1 Section 2 D Modelling of the Structure Chapter 2
GL 2004 Page 2–3

ally, plate or shell elements having improved in-plane 5. For both types of modelling, the dimensions
deformation behaviour should be chosen. of the first two elements in front of the weld toe
should be chosen as follows:
If solid elements are used, the weld can be modelled
simply with prismatic elements, see Fig. 2.5. With – Weld toes of type a):
isoparametric 20-node elements, one element in plate The element length should correspond to the
thickness direction is sufficient due to the quadratic plate thickness t. In the transverse direction, the
shape function and the linear stress distribution. In plate thickness t may be chosen as well for the
connection with reduced integration order, the stresses breath of the plate elements. However, the total
can be extrapolated directly to the element surface and breath of the two shell elements or the solid
then to the weld toe. element in front of the weld toe should not ex-
ceed the attachment width w, where w is the
shell elements thickness of the attached plate above or below
(without welds) plus 2 times the weld leg length, see 2.5.
– Weld toes of type b):
At plate edges, the element length and depth
should be 10 mm. Due to the steep stress in-
crease in these areas, generally higher-order
elements with mid-side nodes or shell elements
with improved in-plane deformation behaviour
have to be used.

w 6. The stresses at the element surface or edge

are to be extrapolated in a linear way to the weld toe,
solid elements see Fig. 2.4. Fig. 2.5 shows the recommended evalua-
tion and extrapolation paths along the element edges
or surface centres. The stresses at the reference points
shown can usually be calculated by averaging the
element stresses at the adjacent two or four corner
nodes, respectively. A probably existing plate bending
stress at plate edges (type b) needs to be included to
only 50 %.

w If the weld is not modelled, the stresses have to be

extrapolated to the intersection line between the plates
modelled. If for a multi-axial stress state the stress
w = attachment width component perpendicular to the weld toe is smaller
than the largest principal stress in a sector of ± 60°
Fig. 2.5 Typical meshes and stress evaluation from this direction, the latter becomes the relevant
paths structural stress.
It has to be noted that the linear stress extrapolation is
coupled with the mesh fineness recommended in 5.,
plate thickness increased by where the stresses are slightly over-estimated at the
first reference point. If a finer mesh is chosen, the
reference points given by the International Institute of
Welding are recommended for stress extrapolation.

7. Probable cracks from the weld root may re-

+t/2 +t/2
quire the additional calculation and assessment of the
nominal stress in the weld throat area, see B.5.

D. Calculation Models for Notch Stresses at

thickness t Plate Edges

1. Notch stresses are normally calculated using

local models of the ship structure. The extent of the
model has to be chosen such that effects due to the
= weld leg length boundaries on the structural area considered are suffi-
ciently small and reasonable boundary conditions can
Abb. 2.6 Simplified modelling of fillet welds be formulated.
Chapter 2 Section 2 D Modelling of the Structure V - Part 1
Page 2–4 GL 2004

2. For the calculation of the notch stress σk at

plate edges (e.g. edges of cut-outs) assuming ideal-
elastic material behaviour, it is mainly the stresses in
the mid-plane of the plate that are significant. There-
fore, modelling of plate areas with membrane ele-
ments is sufficient in most cases, as far as the bending
stiffness does not significantly affect the load distribu-
tion.
Fig. 2.7 Typical mesh subdivision for the compu-
tation of notch stresses at edges of cut-
outs
3. In notched areas, the mesh fineness has to be
chosen such that the stresses at the plate edge can be 4. Stiffeners on primary structural members in
determined with sufficient accuracy. This means for the vicinity of the analysed notches are normally to be
example that at least eight elements with linear dis- considered in the model. A three-dimensional model is
placement function or five elements with quadratic recommended, with a realistic modelling of the termi-
displacement function should be arranged along the nations of the structural members, e.g. sniped stiffener
quarter of a circle. The subdivision in radial direction ends.
also has to be sufficiently fine in view of the stress
gradient. It is recommended that the element lengths 5. The evaluation of the results can be simpli-
in the radial direction be chosen no larger than in the fied if truss elements with negligible cross-sectional
tangential direction. The element arrangement for the areas (which do not affect the structural stiffness, but
stress analysis of a hole in the web of a girder is ex- yield directly the edge stress) are arranged at the plate
emplified in Fig. 2.7. edges.
V - Part 1 Section 3 B Definition of Load Cases and Calculation of the Stress Spectrum Chapter 2
GL 2004 Page 3–1

Section 3

Definition of Load Cases and Calculation of the Stress Spectrum

A. General B. Simplified Deterministic Analysis

1. In simplified deterministic analyses, unfa-

1. In the following, further information is given vourable situations in waves are selected for all load-
on the three analysis methods mentioned in Section 1, ing conditions, considering the maximum stress range
C. Here, only wave-induced loads and loads due to
Δσmax and associated mean stress σm.
different loading conditions are considered. Other load
effects, which may be relevant for certain ship types,
can be dealt with analogously. 2. The assumed wave loads and the load combi-
nation factors result from the assumptions in the Con-
struction Rules for Seagoing Ships. It has to be noted
2. Regarding the consideration of the loading here that the load components (e.g. horizontal bending
conditions, two approaches are possible, according to and torsional moments) have to be superimposed with
the Construction Rules for Seagoing Ships: their proper signs.

– assumption of the loading condition that is most 3. Generally, load combinations (cf. Fig. 3.1)
unfavourable for the fatigue strength of the de- with
tail considered, i.e. with respect to the mean
stress and the stress ranges – waves from ahead/astern (max. vertical bending
moments in the hull girder, correspondingly in-
– assumption of several representative loading creased or reduced side and bottom pressure in
conditions which on the one hand are typical for the forward, midship and aft part of the hull
the ship being considered and on the other in- girder), and
clude different cargo and ballast distributions – oblique head/stern waves (reduced vertical
and also different draughts. Frequently, 4 to 12 bending moments in the hull girder, max. hori-
loading conditions are sufficient, if they zontal bending and torsional moments, max.
transverse acceleration or tank pressures, with
– cause high hogging and sagging bending corresponding opposing pressure from outside)
moments in the hull girder,
should be considered. In the latter case, the considera-
tion of the upright and heeled ship position might be
– include the ballast and fully-laden condi-
necessary, depending on the structural detail under
tions, as well as an intermediate condition,
consideration. Unfavourable load combination may
if appropriate,
occur at port or starboard side.
– include cargo distributions which cause
high stresses in the bottom and in the bulk- 4. After calculating the highest stress range
heads. Δσmax and the associated mean stress σm for each
loading condition, further characteristics of the stress
Regarding consumables, consideration of the condi- range spectrum are estimated on the basis of the Con-
tion "50 % consumables" is normally sufficient. struction Rules for Seagoing Ships. For wave-induced
stresses, usually a straight-line spectrum in semi-
In extensive fatigue strength analyses for different logarithmic representation and a total number of load
structural areas, various loading conditions normally cycles nmax = 5 ⋅ 107 during the service life of about
have to be analysed, because unfavourable conditions 20 years are assumed, see also I – Ship Technology,
can result from different loading conditions for differ- Part 1 – Seagoing Ships, Chapter 1 – Hull Structures,
ent details. Section 20, A.2.4.
Chapter 2 Section 3 B Definition of Load Cases and Calculation of the Stress Spectrum V - Part 1
Page 3–2 GL 2004

1. Head Seas

pD
(+) (-)

az = 0 az = 0

T
z z

(-) (+)

max ps min ps

2. Beam Seas, upright Position

(+) (-)

(+) az = 0 (-) (-) az = 0 (+)

ay = 0 ay = 0

z z

(-) (+)

min ps max ps min ps

3. Beam Seas, heeled Condition

(+) (-)

az = 0 az = 0
(+) ay > 0 (-) (-) ay < 0 (+)
j > 0 (for tanks) j < 0 (for tanks)

z z

(-) j y (+) j y

min ps max ps min ps

Remarks:
- In cases 1. and 2., az ¹ 0 may be considered in connection with mean external pressure.
- The signs (+) and (-) apply to the longitudinal stresses due to vertical and horizontal wave bending of the hull
girder acc. to Construction Rules for Seagoing Ships.
- Torsion related stresses are to be included for ships with large deck openings.

Fig. 3.1 Typical Load Situations for the Calculation of the Highest Stress Range
V - Part 1 Section 3 D Definition of Load Cases and Calculation of the Stress Spectrum Chapter 2
GL 2004 Page 3–3

C. Spectral Method Relevant nonlinear effects regarding the damage

should be considered. This includes the pressure fluc-
1. When applying the spectral method, the fol- tuations at the ship's sides close to the still-water line
lowing calculation steps are performed for each load- which depend in a nonlinear way on the wave height,
ing condition: due to intermittent wetting of the structural area con-
sidered. Reference is made to the continuative techni-
1.1 Calculation of the structural response (stress) cal literature concerning methods to account for this
in the form of transfer functions (response amplitude effect.
operator RAO) for waves of different length. Various
angles of encounter are assumed for these waves. 4. For the long-term statistics of the sea states,
the data of the North Atlantic Ocean are usually used;
1.2 Calculation of the stress spectrum for all sea see Table 3.1. The angle of encounter is normally
states considered and determination of the stress range assumed to follow a uniform distribution, if more
distribution. accurate data is not available. The total service life is
usually set to 20 years. From these data, the total dura-
tion of the individual sea states is derived for the dif-
1.3 Establishment of the long-term distribution of ferent angles of encounter and loading conditions.
the stress ranges, considering the probability of occur- From this and from the mean period of the structural
rence of the individual sea states. response the number of cycles of the associated spec-
Some information on the necessary assumptions are trum of stress ranges follows. By summation of the
given in the following paragraphs. Further details are partial spectra, the long-term distribution of stress
given in the continuative technical literature. ranges can then be calculated.

2. The wave load cases in the first step (see 1.1) 5. It is recommended that the long-term distri-
should be calculated with a wave amplitude which lies bution of stress ranges at the structural detail consid-
within the range of the highest damage due to the ered be calculated separately for all different loading
number of load cycles (approx. 1 m). The ship’s speed conditions, because they are assessed with different
is usually assumed to be 2/3 of the service speed if no mean stresses. For each loading condition, the entire
service profile is specified. The wave lengths should service life of 20 years should be assumed in a first
be chosen such that the pressure fluctuations due to step. In this way, the results can be directly compared
shorter waves as well as the roll, heave and pitch reso- and it is possible to assess them for various combina-
nances of the ship are covered, and such that enough tions of loading conditions as well as for the most
values are available to allow for a sufficiently accurate critical loading condition.
representation of the transfer function between wave
amplitude and structural response (normally 15 – 25
wavelengths). The angle of encounter should be varied
in steps of 30 degrees, whereby waves running trans- D. Simulation of the Stress History
versely to the ship’s course usually require a special
consideration. If only angles of encounter from one 1. Different possibilities exist for the simulation
side of the ship are considered, the results for the of the stress history, inter alia:
corresponding detail on the other ship’s side have to
be included in the evaluation. For each wave situation, a) Simulation of the stress history on the basis of
two cases with a phase difference are considered in computed stress range spectra for the individual
order to obtain real and imaginary part of the har- sea states
monic load process.
b) Simulation of the complete motion and load
process for the whole ship structure.
3. In the second step (see 1.2), the stress spectra
are calculated for all sea states considered by the long- In any case, different sea states and angles of encoun-
term statistics, with all loading conditions and angles ter are considered for all loading conditions investi-
of encounter, by multiplying the quadratic transfer gated, from which the long-term distribution of stress
function with the individual wave spectra. For the ranges is evaluated in a similar way as that described
wave spectra, a two-parameter standard distribution in C.4.
(e.g. according to Pierson-Moskovitz) can be assumed.
The frequency distribution of the stress ranges results 2. Method a) is based on calculations of the
approximately from the spectral moments and the structural response, as described in C.2. – C.3. in con-
assumption of a narrow-banded Gaussian process nection with the spectral method, so that the prerequi-
(Rayleigh distribution of the maxima). A correction sites are the same. The advantage of simulation lies in
for the wide bandedness of the loading process can be the computation of the course of the stress history,
performed on the basis of published methods, whilst which allows a refined assessment of the damage
its neglect generally leads to slightly conservative process, e.g. by considering sequential effects or mate-
results. rial-related nonlinearities in corresponding concepts.
Chapter 2 Section 3 D Definition of Load Cases and Calculation of the Stress Spectrum V - Part 1
Page 3–4 GL 2004

3. In addition to that, method b) permits realistic such that significant statistical evaluations can be
consideration of nonlinearities in the load process, made with respect to the entire service life.
which may be significant for certain ship types.
5. From the stress history evaluated, the relevant
parameters for the assessment (e.g. largest stress range
Δσmax, shape of stress spectrum and associated mean
4. The loading conditions and sea states consid- stress σm) are to be determined using the rainflow
ered as well as the simulation time are to be chosen counting method.
 GL 2004
V - Part 1

hs \ tz 1.5 2.5 3.5 4.5 5.5 6.5 7.5 8.5 9.5 10.5 11.5 12.5 13.5 14.5 15.5 16.5 17.5 18.5 SUM
Section 3

0.5 0.0 0.0 1.3 133.7 865.6 1186.0 634.2 186.3 36.9 5.6 0.7 0.1 0.0 0.0 0.0 0.0 0.0 0.0 3050
1.5 0.0 0.0 0.0 29.3 986.0 4976.0 7738.0 5569.7 2375.7 703.5 160.7 30.5 5.1 0.8 0.1 0.0 0.0 0.0 22575
D

2.5 0.0 0.0 0.0 2.2 197.5 2158.8 6230.0 7449.5 4860.4 2066.0 644.5 160.2 33.7 6.3 1.1 0.2 0.0 0.0 23810
Table 3.1

3.5 0.0 0.0 0.0 0.2 34.9 695.5 3226.5 5675.0 5099.1 2838.0 1114.1 337.7 84.3 18.2 3.5 0.6 0.1 0.0 19128
4.5 0.0 0.0 0.0 0.0 6.0 196.1 1354.3 3288.5 3857.5 2685.5 1275.2 455.1 130.9 31.9 6.9 1.3 0.2 0.0 13289
5.5 0.0 0.0 0.0 0.0 1.0 51.0 498.4 1602.9 2372.7 2008.3 1126.0 463.6 150.9 41.0 9.7 2.1 0.4 0.1 8328
6.5 0.0 0.0 0.0 0.0 0.0 12.6 167.0 690.3 1257.9 1268.6 825.9 386.8 140.8 42.2 10.9 2.5 0.5 0.1 4806
7.5 0.0 0.0 0.0 0.0 0.0 3.0 52.1 270.1 594.4 703.2 524.9 276.7 111.7 36.7 10.2 2.5 0.6 0.1 2586
8.5 0.0 0.0 0.0 0.0 0.0 0.7 15.4 97.9 255.9 350.6 296.9 174.6 77.6 27.7 8.4 2.2 0.5 0.1 1309
9.5 0.0 0.0 0.0 0.0 0.0 0.2 4.3 33.2 101.9 159.9 152.2 99.2 48.3 17.7 6.1 1.7 0.4 0.1 626
10.5 0.0 0.0 0.0 0.0 0.0 0.0 1.2 10.7 37.9 67.5 71.7 51.5 27.3 11.4 4.0 1.2 0.3 0.1 285
11.5 0.0 0.0 0.0 0.0 0.0 0.0 0.1 3.3 13.3 26.6 31.4 24.7 14.2 6.4 2.4 0.7 0.2 0.1 124
12.5 0.0 0.0 0.0 0.0 0.0 0.0 0.0 1.0 4.4 9.9 12.8 11.0 6.8 3.3 1.3 0.4 0.1 0.0 51
13.5 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.3 1.4 3.5 5.0 4.6 3.1 1.6 0.7 0.2 0.1 0.0 21
14.5 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.1 0.4 1.2 1.8 1.8 1.3 0.7 0.3 0.1 0.0 0.0 8
15.5 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.1 0.4 0.6 0.7 0.5 0.3 0.1 0.1 0.0 0.0 3
Wave Scatter Diagram for the North Atlantic

16.5 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.1 0.2 0.2 0.2 0.1 0.1 0.0 0.0 0.0 1
SUM: 0 0 1 165 2091 9280 19922 24879 20870 12898 6245 2479 827 247 66 16 3 1 100000

Values stand for time fractions of sea states characterized by tz and hs

tz = zero upcrossing period [ s ]

Definition of Load Cases and Calculation of the Stress Spectrum

hs = significant wave height [ m ]

Page 3–5
Chapter 2
V - Part 1 Section 4 B Assessment of the Results Chapter 2
GL 2004 Page 4–1

Section 4

Assessment of the Results

A. General 2. The damage ratio computed using Palmgren-

Miner's Rule must generally not exceed the limit value
1. The calculation of stresses and their assess- D*:
ment with respect to fatigue strength should be per-
formed for as many details as possible, so that critical ⎛n ⎞
D = ∑ ⎜ i ⎟ ≤ D*
areas can be identified with certainty and effective ⎝ Ni ⎠
counter measures can be taken.
ni = number of stress cycles for the block i of the
2. An overview of the probable danger of fa- long-term spectrum, which has to be subdi-
tigue cracking is given by: vided into at least 20 blocks
– The presentation of maximum stress ranges in Ni = number of endurable stress cycles which
the structure analysed, whereby, in the case of results from the design S-N curve of the de-
nominal stresses, local stress increases or detail tail considered. The design S-N curve has to
categories of the structural details have to be ob- be corrected for several factors in accordance
served. with the Construction Rules for Seagoing
Ships - with respect to the mean stress,
– Existing experience for ships in service which amongst others.
indicates that primarily
D* = limit value defined in the Construction Rules
– longitudinal structural members in the upper for Seagoing Ships, see Section 1, A.1.
and lower flange of the hull girder due to hull
girder bending (except for the ship’s ends), The design lifetime results from
– longitudinal and transverse structural mem- L = 20 years D
bers at the ship’s sides due to local pressure
fluctuations, and
– structural components in the vicinity of the 3. If more than one loading condition is ana-
main propulsion plant and manoeuvring sys- lysed, the assessment can be performed in two ways,
tem as long as the damage ratio Dj has been computed for
each loading condition j for a service life of 20 years
as well as, in general, discontinuities in continuous (cf. Section 3, C.5.):
structural members are prone to fatigue.
a) Only the most critical loading condition is con-
The probability of fatigue failure increases if higher- sidered:
tensile steel is used due to the increased nominal stress
as long as no compensation is achieved by an im- ( )
Max D j ≤ D*
proved structural design and/or higher fabrication
quality.
b) The damage due to all loading conditions is
considered, taking into account their probability
of occurrence Pj (Σ Pj = 1),
B. Assessment of the Fatigue Strength
( )
∑ Pj ⋅ D j ≤ 0, 7 D*

1. The fatigue strength is assessed either

Here, a reduced limit value of the damage ratio is
– on the basis of a damage calculation using used, because less favourable probabilities of certain
Palmgren-Miner's Rule, or loading conditions may occur in reality compared to
the assumptions made, and because the additional
– in the case of standard stress spectra, using the damage due to the changes between the individual
permissible highest stress range Δσp loading conditions is not considered.
Chapter 2 Section 4 B Assessment of the Results V - Part 1
Page 4–2 GL 2004

Alternatively, the additional load cycles due to chang- 5. It is recommended that the results be assessed
ing loading condition may be included in the damage on the basis of the usage factor U with respect to stress
calculation, by taking the rainflow counting method ranges:
into account and making conservative assumptions for
the probability of occurrence of the individual loading
conditions. Δσmax
U =
Δσp
4. For standardized stress spectra, the criterion
for the assessment is When applying Palmgren-Miner's Rule, Δσp results
Δσ max ≤ Δσp from the condition that with an equivalent spectrum of
stress ranges, i.e. having the same shape and number
of load cycles, the limit damage ratio D* according to
Δσmax = highest stress range within the spectrum B.2. is reached. The application of the usage factor has
the advantage that - due to the relationship with
Δσp = permissible stress range
stresses - direct conclusions can be drawn with respect
The permissible stress range results directly from the to the necessary reduction of stresses or a possible
Construction Rules for Seagoing Ships, considering increase of the detail category. For example, in case of
the appropriate detail category or design S-N curve, as U = 1,2, the stress has to be reduced by 20 % or the
well as various other factors, such as the mean stress. detail category increased by 20 %.