LLOYD'S REGISTER INTEGRATED FATIGUE DESIGN
ASSESSMENT SYSTEM
Abstract
This paper describes the multi-level fatigue design
assessment procedure developed by Lloyd's Register
to estimate the fatigue strength of ship structural
details. The method used to determine a representative fatigue strength capability for ship
structural details, and the supporting fatigue testing programme of large ship structure are reviewed. The traditional maximum lifetime load approach to
estimate the fatigue demand is discussed, and the spectral analysis procedure adopted in the subject
procedure is described. Examples of distribution of fatigue damage around the hull envelope are given to illustrate the influence of the load components, and the wave environment.
1. INTRODUCTION
The objectives of Ship Classification are to develop
and implement Rules and Regulations, which in conjunction with proper care and operation on the part of the shipowner and operator, will provide an
appropriate standard of structural strength during the service life
of the ship. On the provision
of anadequate level of maintenance associated with regular structural surveys and appropriate repair procedures for incidental structural damages, Classification Society Rules are intended to
safeguard the structural integrity of the hull structure
for a service
life periodof at
least 20 years.Computational structural analysis methods combined
with service experience of large and complex ship structures have shown that success or failure in a
structural sense, is directly affected by the
performance of structural details. The failure of structural details is generally associated with two interrelated cumulative damage processes, namely,
fatigue and corrosion, Violette (1994). As early as
1913, Lloyd's Register expressed its concern on the fatigue mode of failure of ship structural details in a Ab. No. 021
Franck L. M. Violette, DUT., M.ENG.
Advanced Studies & Rule Development Group
Technical Planning & Development Department
Lloyd's Register of Shipping - Ship Division
71, Fenchurch Street
London, EC3M 4BS, United Kingdom
paper presented to the Royal Institution of Naval
Architects, Lloyd's Register (1913). Over the last
decade, fatigue failures of structural details and their potential consequences have attracted an increased attention amongst the shipping industry. In view of the higher percentages of high tensile steel being used, the application of sophisticated techniques to peorm structural and fabrication optimisation, and the
implementation of strict environmental regulations, the occurrence of fatigue cracking cannot be considered any longer as a mere fact of life. Since fatigue cracks can be possible points of initiatión for the failure of the cargo containment barrier, and/or significant structural failures, it is essential that fatigue performance be given a more detailed consideration at the design,
and construction stages as well as during the ship's operational life.
In 1989, Lloyd's Register (LR) initiated a long term
research and development programme with the aim to develop and implement an integrated fatigue
design assessment procedure for ship structural details. In June 1994, ShipRight FDA (Fatigue Design Assessment), a multi-level integrated fatigue design procedure has been released to world-wide LR plan approval offices and to shipyards. This paper reviews the main components of the research project,
and outlines the level 2 ShipRight FDA integrated
fatigue design assessment procedure. Some of the more salient aspects of the procedure are illustrated
with examples from the fatigue
analysisof the
longitudinal members of a double hull tanker.
2. SHIP STRUCTURAL DETAIL FATIGUE DESIGN 2.1 GeneraI
Whilst for at least the past two decades, Classification
Societies have carried out fatigue calculations for
damaged or novel ship structural components, procedures or criteria for the design against fatigue
have not been explicitly or rationally considered in
1/17
TECHgßsc UN1VERSITJ.T
aboratorum 'oor
Mekelweg 2, 2628
CD Oft
lei: 015 768 Fax: 015 7aClassification Society's Rules until recently, except for
specific ship types such as LNG ships. However,
implicit fatigue requirements are incorporated for
many structural components through the use of
permissible stress levels determined from the application of simpUfied fatigue damage assessment techniques combined with service experience data. In fact, fatigue has been given due considerations at all stages of the Classification process through implicit requirements or procedures as follows
During the plan approval process, in addition to the use of permissible stress levels, the surveyor applies particular attention to the design of critical
structural details by using a set of experience
based detail design recommendations compiled
over the years. This s carried out in close co-operation with the ship designers in order to achieve the best solution in terms of both the
fatigue performance and fabrication;
During the construction stage, the field surveyor
assigns particular attention to the critical areas
which have been
identified during the planapproval process, in order to ensure that satisfactory levels of workmanship, alignment and fit up are achieved;
And, finally, during the life of the ship, the scope
and extent of the periodical surveys give due
attention to the critical structural details, and the
detection of any onset of cracking. Should a fatigue crack be detected, remedial action is taken to prevent its reoccurrence, and the survey data is transferred to the LR damage database for analysis. Noteworthy defects or increase in
fatigue failure trends of structural components are communicated to plan approval surveyors, and
Rule development engineers, and an implicit fatigue criteria such as a permissible stress level, or a detail design recommendation may be
formulated for inclusion in the Rules.
2.2 Limitations
of
an Experience BasedProcedure
However, it should be recognised that with new generation ships such as double hull tankers, the
value of service experience may not be directly
applicable. Whilst many of
the new
oil tanker structuraldetails can be related
to existing hullconfigurations, the increased proportion of high
tensile steels, the extent of structural optimisation, as well as changes in the loading patterns could change the significance of an experience based procedure,
Ferguson & Violette (1991). Significant changes in
terms of ship structural design and construction, and
the introduction of strict environmental regulations have called for the development of rational direct
calculation fatigue design procedures where the following aspects which are of particular concern need to be addressed:
The extrapolation of ship structural concepts may not be directly applicable to new designs, and the service experience of double hull tanker structural
details, especially large double hull tankers, is limited;
Recent experience with high tensile steel used on second generation single skin VLCC structures
has shown that when a significant degree of structural optimisation has been carried out on
the primary web structures, the additional loads created on the secondary members may increase the risk of fatigue damage;
With double hull tankers, cargo leakage as a
result of fatigue cracking would result in serious difficulties to clean and/or ventilate the double hull spaces due to their cellular arrangement. Moreover, leakage of cargo oil, or inert gas into
the ballast spaces would place the ship in a
hazardous situation with potential risks of explosion, loss of life, and environmental disaster; The ship structural performance and life could be
severely influenced by the ballast tank coating
performance. The initiation of fatigue cracks in the ballast spaces may precede coating breakdown. If undetected, or if no remedial action has been taken, the localised coating breakdown in way of a detail stress concentration will create a severe localised corrosion cell, which in turn will
accelerate the fatigue crack propagation process, Violette (1994);
The quality of the workmanship, and the
construction tolerances may influence the fatigue
performance of ship structural details. Fatigue cracking as a result of poor detail design may be difficult to cure without significant and expensive structural modifications.
3. ShipRight FDA FATIGUE DESIGN ASSESSMENT
lt can be appreciated that the evaluation of fatigue performance of ship structural details is a complex
process, and that fatigue failures may have dramatic consequences. To attain and maintain a satisfactory
fatigue performance, a realistic fatigue procedure
should give due consideration to the following stages in the life of a structural detail
The conceptual design of ship structural details;
The analysis of the fatigue performance by a
direct calculation method giving due consideration to the cumulative nature of the fatigue damage process, and accounting for fabrication and
workmanship factors;
The verification of the structural details
workmanship, alignment and fit-up tolerances, through adequate survey procedures and defects acceptance criteria during the ship construction; The in-service survey procedures, and the
monitoring of the critical structural detail items
during the ship lifetime.
For this purpose, ShipRight FDA (1994) has been
developed as a multi-level fatigue design assessment procedure (Level 1,2 and 3) to address each of the requirements highlighted above. The procedure is a
total approach to the prevention of fatigue failures
encompassing the design, construction and in-service performance of ship structural details. lt s supported by the ShipRight CM (1994) Construction Monitoring
procedure. and the ShipRight HCM (1994)
HullCondition Monitoring procedure. The main features of
the multi-level FDA procedure are reviewed in the
following sections.
LEVEL 1: STRUCTURAL DETAIL DESIGN GUIDE
The primary purpose of the Structural Detail Design
Guide is to promote good detail design at an early
stage of the design process, and to provide guidance for improvement of detail design. The Guide has been compiled from the world-wide detail design and the in-service expertise of plan approval, newbuilding and field surveyors. Therefore, it
is based on a vast
experience based knowledge database consideringaspects such as design and analysis, construction tolerances and fabrication issues, and in-service performance. In addition, extensive analytical, and Finite Element Analyses (FEA) have been performed for each recommended structural details to confirm and optimise the structural configuration to maximise
its fatigue performance. A typical Structural Detail Design Guide datasheet is shown in Figure 1. At
present, the Structural Detail Design Guide addresses double hull tanker, and bulk carrier critical areas. The
development of the Guide is considered to be a continuous process with regular updates to reflect
trends in service experience, design and construction
practice as well as to incorporate results from the
ongoing FDA and FEA studies, and LR fatigue testing programme.
SHIP STRUCTURAL DETAIL FATIGUE STRENGTH
For design purposes, the S-N curve approach to the estimation of the fatigue strength capability is considered to be the most common and convenient approach. However, the assignment of a S-N curve to a ship structural detail will invariably involve a certain amount of engineering judgement. To ensure that the FDA procedure could be applied consistently, it was considered essential
to automate the S-N curve
selection process. This task is performed by the FDA S-N Curve Expert procedure.
5.1 NominaI and Reference S-N Curve
The traditional nominal stress S-N curve approach s expressed as a function of the nominal stress range
AS for a given typical structural detail as follows:
(1)
Traditional nominal stress S-N curves are widelyavailable from design codes such as BS5400 (1980), UK.Den (1983), Eurocode (1985), 11W proposal
(1982), etc. In general, these standard are applicable to a limited set of typical standard details. In view of the large variety of ship structural details in terms of
both geometry and loading, the application of standard detail S-N curves has shown to be a difficult process. Comparative fatigue studies correlated with service experience and/or full scale measurements
have shown that standard S-N curves may yield
fatigue lives significantly different from the recorded values.
A more explicit method to apply the standard S-N curves to ship structural details is to derive,, the geometrical Stress Concentration Factor (SCF) of the standard detail geometry, as well as the SCF of the ship structural detail, in
order to obtain a more
representative S-N curve. The S-N curve may
therefore be rewritten as follows ship
N
1\ K,d =j
,fl' ASK,d '
The disadvantages of this approach is that geometrical information on the standard S-N Curve structural detail is often limited,
and FEA with
compatible mesh size for both the standard structural
detail, and the ship structural detail is required to
determine the SCF's. Moreover, it should be borne in mind that standard S-N curves tend to represent the lower limit of fatigue capability of the standard detail i.e. geometrical and scantling configuration leading to the worst fatigue strength within the scope of application of the detail. For example, the transition between BS5400 (1980) Class F and F2 is dictated
only by the attachment length (150 mm), and the
edge distance (10 mm), and no reference is made, for example, for the influence of the plate thicknesses.
To enable a more representative estimation of the fatigue strength for ship structural detail within the
Ab. No. 021 3/17
Fig. I
Structural Detail Design Guide Typical DatasheetLloyd's Register - ShipRight Strategic Research & Development Page 5
r
LOCATION: Connection of side and longitudinal bulkhead longirudinais to transverse webs in double side tanks
EXAÍVWLE
No. 1:
Asymmetrical face/higher tensile steel side longTh.idinal totransverse web flat-bar stifferters
GROUP
No. i
CRITICAL AREAS DETAll. DESIGN IMPROVEMENT
SHELL PLATING
______________
-BULKHEADx
NOTESOIT
SOFT//EEL TOE [1 SOFT ÁMD OR SR4CKETS DETAIL è4AX1 (M S)-A1A1ETRIC4L TOE h-MAX 15L2.Od
U UliLi
U 4cRmcALAREAS B,ac*et Á.th,nwmo.J
754
X0.5)
4
--MAX15
IThIckness Flat Sa,
Thickness 12Omm
-4-MAX15 Thickness I=6/18
dj
d180-300 Areaof stiffener accordancewith Rule requirementsCONSIDERATION ITEMS FOR IMPROVEMENT
Critical Location: Asymmetrical face / higher tensile steel side longitudinal face bar connections at the heel and the toe of the web stiffeners. Connections between the base line and O.8D above the
base line.
Fatigue Mechanism: Soft toe and heel detail or symmetrical soft toe brackets to reduce peak stresses under fatigue loading from dynamic seaway loads and ship motions.
Building Tolerances: Ensure alignment of the web stiffener, the back bracket and the web of the side longitudinal.
Welding Requirements: Fillet welding having mínimum weld factor of 0.44 (Web stiffeners to face bars of side longitudinals. Back brackets to face bars of side longitudinals).
A wrap around weld, free of undercut or notches, around the plate thickness.
FIGURE
i
DETAIL DESIGN GUIDELINES FOR DOUBLE HULLTANIR STRUCTURAL DETAILSframework of a design procedure, it was considered
Ka = Kg(R,tid7.la,d,c,th)Kje)K,i(ö)K,(e)
(5)
that:
The nominal stress should remain the reference stress used for FDA purposes, as it is a convenient stress measure for engineers involved in structural design assessments;
And, a large library of S-N curves based on
actual ship structural detail geometries should be made available.
To satisfy these requirements, the hot spot stress
approach was reformatted as follows
Thus,
=
N = IÇ(Ka AX)"
N=
nl, ShIBased on FEA of standard structural details, and the
available S-N curves data, a reference S-N curve
defined by K1, has been evaluated. The reference
S-N curve represent the fatigue strength
of the
welded material including the fillet weld geometrical stress concentration. The geometrical SCF of the ship structural detail then becomes the driving parameter to estimate the fatigue strength, and its derivation is reviewed in the following Section. For reference
purposes, the S-N curves derived by the FDA S-N
Curve Expert procedure are assigned a value which represent the stress range magnitude of the mean S-N curve at stress cycles.
5.2 Geometrical Stress Concentration Factor Based on systematic
linear elastic FEA of ship
structural details, parametric formulations
of the
geometrical SCF's have been
derived. The FE
modelling standard adopted to determine the SCF's is
based on a FE mesh of t x
tin way of stress
concentration areas, where t is the thickness of the plate containing the crack initiation site. The SCF has been determined using the centroidal stress normal to the expected crack plane for the element adjacent to the theoretical intersection edge. Taking into account that the weld factor for the connections underinvestigation is of the order of 0.44t, the SCF stress
gradient pick up point is at a minimum distance of
0.06t from the weld toe. A typical FE mesh for a soft toe - soft heel web stiffener connectìon to the flange of a longitudinal member is shown in Figure 2.
The parametric SCF formulation is defined as a combination of influence functions defined as follows for the end connection geometry shown in Figure 3.
(4)
A-Ati
A-A
nCrltical Location
th
(3)
Fig. 3. Stress Concentration Factor Parameters.
The axial loading SCF has been chosen as the
reference SCF for the computation of the structural
detail fatigue strength capability by the FDA S-N Curve Expert. To account for the influence of the
SCF's for modes of loading other than axial, loading mode bias factors are introduced in the computation of the total nominal stress applied to the detail. The loading mode bias factor for loading componen.t i is defined as follows
K
KB,
=-Ka
(6)
To illustrate the SCF parametric formulation, Figure 4
shows the variation of Ka
for a soft toe flat bar
stiffener connection with
d = loo
mm,
t1/t2 =1.07th =I5mm,Kw(0)K,,,(S)Ke(e)=l.
Ab. No. 021 5/17
100 150 200 250 300 350
SoftToe Radius -mm
Fig. 4. Ka versus Soft Toe Radius
ASh. K
ç ''
h, K 1.350 1.325 1.300 1.275 1.250Figure 5. illustrates the variation flat bar stiffener connection
R=I5Omm, th=l5mm,
Kw(0)K,,,(Et)Ke(e) = i
K 1.350 1.300 1.250 1.200 1.150 1.100 1.00Fig. 5. K versus t1 ¡t. Ratio
lt can be appreciated that the FDA S-N Curve Expert
approach permits to determine more realistic S-N
curves for ship structural details, since the S-N curve
is a direct function of the SCF, and thus the detail
geometrical parameters.
5.3 Fatigue Testing of Ship Structure Models
To increase the confidence level in fatigue strength predictions and to gain a better understanding of the
fatigue crack initiation and propagation process in ship structural details, Lloyd's Register initiated a
programme of fatigue testing of large ship structure models in 1992. The Krylov Shipbuilding Research
Institute in Russia was commissioned to carry out
these tests. There are many benefits associated with performing large scale fatigue tests of realistic ship structures which may be summarised as follows
Shipbuilding workmanship standards are used to fabricate the models;
Realistic ship loads are applied to the models resulting in a more realistic stress field , stress concentration levels, and cyclic stress patterns in way of the critical locations;
The kinematics
of crack
propagation at the potential crack initiation sites provide valuable information with regard to the relative severity of each potential crack initiation sites, their propagation rates, and the potential consequences when applied to real ship structures;The assessment of the redundancy level of the
structural system, and the load redistribution
of K for a soft toe
with
il = 200
mm,
= 300 mm,
andmechanisms which may affect the crack propagation rates at certain crack initiation sites.
Extensive post analysis using FEA can be
performed to confirm the strain / stress
measurements, to extend the results to the full scale structure, and to optimise the structural
configuration with respect to fatigue strength; And, finally, experimental data combined with FEA permits the calibration of the parametric
formulations of the SCF's associated with the
reference S-N curve which have been
implemented in the FDA S-N Curve Expert.
To date, the ongoing experimental research
programme has addressed the following ship
structural details
1/4 scale models
of VLCC hopper welded
connections to double side and double bottom;1/4 scale models
of VLCC hopper
flanged connections to double side and double bottom;1/3 scale models of VLCC longitudinal connecons
to web fiat bar stiffener with various end connections, and steel yield strength.
Photo 1-2 show the large scale ship structures models under testing conditiqns.
io
to 106 loadcycles are applied to the models which are
instrumented with uniaxial and rosettes strain gauges
to monitor the field and the hot spot stresses.
Acoustic monitoring has also been used to determine
the spatial location of the onset of cracking well
before the crack could be detected by visual inspection. The crack propagation rates have been recorded in order to monitor the crack growth in the structural components, and enable further analysis to be performed using fracture mechanics methods. 6. LONG TERM STRESS RANGE SPECTRUM
MODEL
Since the wave environment generates complex
loading patterns, it can be appreciated that the prediction of the long term stress spectrum for marine
structures remains a complex problem. Due to the
cumulative nature of the fatigue damage process, it is
essential that an adequate procedure be used to
predict the long term stress range distribution. Figure 6. illustrates the distribution of fatigue damage for
three typical long term stress range spectrum defined analytically as Weibull functions with shape factors of 0.8, 1.0, and 1.2. lt is shown that most of the fatigue damage is produced by the small to medium stress ranges i.e. the low to medium seastates, by virtue of their associated number of stress cycles ( iO5 to 108
Ab. No. 021 6/17
1.25 1.50 1 75
o o -c Q-N o o û-
N-Fig. 2 Typical Fine Mesh FE Model for Soft Toe Soft Heel Longitudinal End Connection
l.
LL
L I
T.1I I
LL L .L
L I_[_ J 11.11
i LL
-L -L_-L-L
.L LLLLLLL.LL
..-L_[:LLH
JL
...
.:...
.:J
L U
.-. U... :LI
t i t IIl
tf
Li
1l(lM
:.c&: .. 1 1 t 1.1 LI_t I .IllllL
j i_l I I Lt'I
L
ro:r
rr
:i
tif.
..
(t11llIll
LL
_L
LL_1LU_
L .JLL.1_l_.
ELL
-..L.LLLLLLJJ LJ._
i
ILl l.i
ULLI.rJ.LLL.t .i.L
i i.._..IL_[.LJJ.L.LHI!iH...
LLLLLLLLLL _L I
_L LL L
LLLL L
r
a
Long Term Weibull 0.80
- - - Long Term Weibull 1.20
Fatigue Damage - Weibull 1.00
Long Term Weibull 1.00 Fatigue Damage - Weibull 0.80
- - - Fatigue Damage - Weibull 1.20
Fig. 6. Typical Long Term Stress Spectrum and Fatigue Damage Distributions stress cycles). Therefore, in
order to achieve a
reliable prediction the structural detail fatigue life, the Ab. No. 021
mathematical model used to predict the long term stress distribution should give due attention to this
8/17
--/
--
I
1E+00 1E+01 1E+02 1E+03 1E+04 1E+05
1E06
1E+07 1E+08Number of Stress Cycles
5.00E-03 4.00E-03 cl) o) - 3.00E-03 E
o
a) - 2.00E-03 o) u-1.00E-03 0.00E+00 225.00 200.00 175.00 150.00 125.00 C) 100.00 g 75.00 cl) 50.00 25.00 0.00part of the spectrum. To determine the long term
stress spectrum, two approaches are available, namely, the maximum lifetime load approach, and the spectral approach. Both procedures are reviewed in the following sections, with emphasis on the spectral method of analysis used in the FDA procedure.
6.1 Maximum Lifetime Load Approach
The main advantage of the maximum lifetime load
approach is its inherent simplicity. Based on the readily available maximum lifetime loads parametric expressions used for strength assessment, this procedure is well suited for design purposes. Figure 7. illustrates the procedural steps involved in the prediction of the long term stress spectrum. However, since fatigue damage is proportional to the cube of the stress range, and the procedure emphasises on the 108 stress range which is remote from the area of interest with respect to fatigue, variations in the 108
stress range and/or the Weibull shape factor can
result in a significant variation in the resulting fatigue life as shown in Figure 8. In view of the sensitivity of the resulting fatigue life to the limited set of modelling variables, some of the modelling assumptions need to be reviewed:
For a given
structural detail, each individual applied maximum lifetime loads is calculated at a probability level of 108. Considerations to loads computed at a probability level of say, 10 , may be more appropriate provided the iO4 loads have not been extrapolated from the 108 base usinganassumed long term distribution;
20.00 10.00 5.00
PiJ11
100 in
220 280----'
0.80 400 460 Stress range - N/mm2The partial load factors or correlation coefficients
used in the load effect combination model are
often based on the dominant load parameter. and represent a snap shot seastate where the
dominant load component is maximised. This combination may not be representative of the less severe seastates represented by the lower part of
the spectrum, where other headings, wave
heights and periods, or load components may
dominate;
The number of stress cycles for a service period
of 20 years
isin general based on the 10
incident waves assumption used for the North
Atlantic longitudinal strength requirement. Where several load components are involved, the resulting number of stress cycles may differ from the number of incident waves;
The Weibull shape factor is often mapped on the
dominant load parameter shape factor for the
North Atlantic wave environment, and it is
questionable whether this is representative of the
resultant load effect arising from several load
components. Furthermore, since the shape of the stress spectrum is directly affected by the wave environment, the stress response in each of the seastate, the ship loading conditions, and the ship seakeeping performance, ¡t is disputable whether this assumption is applicable for a range of ship types, and characteristics.
w C-_C
cn2
o
DLL wFig. 8 Sensitivity of Fatigue Damage to Stress Range and Weibull Shape Factor
Ab. No. 021 9/17
Identification of Load Process Applied to Structural Detail
I
Computation of Maximum Lifetime Loads Magnitude
P=lo -
I
Computation of Structural Influence Coefficient
I
Computation of Maximum Lifetime Stress Magnitude
P=iO
-I
Identify Dominant Load Effect & Assign Partial Load
Effect Factors
I
Weibull Shape Factor based on Dominant Load Effect
I
% Lifetime Number of Incident Waves
I
Wave Induced Loads &
Motion Computation Structural Response Analysts Statistical Analysis Fatigue Strength Voyage Simulation Levei2 " Analytical / Parametric Formulation Level 3 - First Principle Ship Motions
& Load Software Level 2
- Analytical, FE Bean, Model SCF
Levei3
e
Global 3D FE Model Local Zoom FE ModelSeastate Wave Height - Wave Period
Ship Heading - Ship Speed
Structural Detail Fatigue Strength - S-N Curve
loo Al Fatigue Wave Environment Load Process I I-Pr@10
cl
V S1@ 10-e V L = LIC'? + ." Shape Parameter SpecFDA SoWerShort Term Stress Response in Irregular Waves
Deterministic Fatigue Life Probability of Failure
Regular Waves Amplitude & Phase Angle
Hull Girder Loads, External Wave Pressur Internal Cargo Pressure
Local Stress Influence
Coefficients
SSC Wave Energy Spectrum
Service Profile Probability Matrix
Wave Height - Wave Period Ship Heading - Ship Speed
Loading Condition
Safe Life Acceptance Criteria N V Pn@10 ' C l IO L V
Loads - Load Effects Stress Combination Short Term Fatigue Damage Rat. Service Experience 4 Fatigue Monitoring
Uncertainty Stress Model
Uncertainty Workmanship
Uncertainty Fatigue Model
Maximum Load Approach Lifetime Stress Spectrum
I
Fig. 7 Maximum Lifetime Load Procedure
Fig. 9 Spectral Fatigue Analysis Procedure
Ab. No. 021 10/17
. Load Process n
.
..
.
L
Calculate Total Stress
I
Calculate Stress Rangelt can be appreciated that this procedure is subject to
a number of assumptions, and modelling
simplifications. To achieve reliable fatigue life
estimates, calibration using service experience data has been shown to be essential. Whilst this procedure
cannot be rejected
as a
design tool to obtainefficiently an estimate of the fatigue strength, the
sensitivity of the fatigue process, and the dependence of the required calibration on the service experience
base make this procedure difficult to apply in a consistent manner. Furthermore, reduced confidence levels are introduced when the structural configuration or the loading patterns depart from the service experience base used for calibration.
6.2 FDA Spectral Analysis Procedure
To enhance the level of confidence in the
determination of the long term stress spectrum, and
address the issues of the maximum lifetime load
approach highlighted above, the FDA procedure uses
a first principles approach based on the spectral method of analysis. Two levels of analysis have been developed using the same theory, but different level of mathematical modelling. Level 2 is used for design purposes and uses simplified mathematical models. Leve! 3 uses sophisticated mathematical models, and is aimed at confirmation of the fatigue performance of novel structural details. The procedural steps of Level 2 and Level 3 are illustrated in Figure 9
The application of the level 2 procedure to longitudinal members of a double hull tanker can be summarised in the following steps
6.2.1 Computation of Wave Induced Loads
The wave induced primary load components considered are
External hydrodynamic wave pressure; Hull girder vertical wave bending moment; Hull girder horizontal wave bending moment; Water ballast/cargo inertia pressure.
The amplitude and phase angle of the above load
components are calculated in regular waves for the following ranges
of wave parameters
for each significant loading conditions as shown in Table 1. Table iAb. No. 021
FDA Level 2 computation of the wave induced loads is based on a parametric formulation of the ship six degrees of motions, and the global and local loads.
Using the basic concept of the single degree of
freedom vibration model, and the systematic analysis of the computation of the ship motions and loads of
over 250 representative hull forms, the Response
Amplitude Operator (RAO) for the motions and loads
has been decomposed into a series of influence
functions based on the work by KSRI (1992)
RAO(V, x,w)=
f (V,
,[a, ])f(V, x.
[a1 ]_) ( 'For example, for the vertical bending moment, the
RAO can be further decomposed to
isolate theinfluence parameters as follows:
RAO(V,,w)= fa'80(L,B,T,Fn,Cw,CB,k,j(h)
fa(X)fa
(x) L)f (x)
(8)
The amplitude function
f
180(.) determines the maximum response amplitude which for the verticalbending moment occurs at 180°. The amplitude function fa(X)describes the ratio the response to
the maximum response amplitude at 180° for a given
ship to wave angle as shown in
Figure 10. Theamplitude function
f (x)
describes the position of the maximum response amplitude on the wave frequency axis as shown in Figure 11.1.00 0.80 0.60
0.40
0.20
Parameter - Range Increment
7.
0-360°
20°0.2-1.2rad/s
0.O4rad/sV
0-Vs
25%VsLoading Conditions Fully loaded Ballast
0.00
o 30 60 90 120 150 180
Ship to Wave Heading - Degrees Fig. 10. Vertical Bending Moment fa(X)
The function
f, ()
describes the RAO basic shape as a function of the wave frequency, response naturalfrequency, the ship length, and the function f (x).
L (x) is
the function describing the longitudinal distribution of the bending moment.Expressions in a similar format have been derived for the horizontal, and torsional bending moments, the ship motion six degrees of freedom, and the external hydrodynamic wave pressure RAO's. Since the phase
angles do not lend themselves to a relatively simple description by analytical functions, the phase angle of
the RAOs has been
stored in a database forcomputation purposes pending further analysis.
1.40 1,30 1.20 1.10 1.00 0 30 60 90 120 150 180
Ship to Wave Heading - Degrees
Fig. 11. Vertical Bending Moment f(x) 6.2.2 Structura! Influence Coefficients
For each load components considered above, the
structural influence coefficient C, ( i.e. direct stress normal to the most likely plane of crack propagation
at the weld toe connection of web stiffener to the longitudinal flange ) due to the application of a unit
load ¡ is determined as follows
.
Vertical wave bending momentC1
= K8
(9)
.
Horizontal wave bending moment C2 = K13zzz I
Hydrodynamic wave pressure
C = K8
12ZLF
f(x)
Water ballasticargo inertia pressure
C4 = K8 12ZLF
f(x)
C5=KB
6E1 C6= K8
12 4 EIC7=K8
=S (w) dw
Assuming that the stress process is narrow banded,
(10)
the stress range distribution can be expressed ¡n terms of a Rayleigh distributions
s2 4: 2 exp 8-)
(20)
For the side shell, where the presence of the wave free surface creates a non-linear effect with a
truncation of the pressure load harmonics, a time
domain simulation procedure is performed to
calculate the short term stress statistics.
(12)
For the secondary load components arising from the deflection of the primary structure, the structural influence coefficient C due to the application of a unit deflection or rotation i is determined as follows
Deflection Side 1 Deflection Side 2 Rotation Side 1
4E1
Rotation Side 2Ç = K1
/(16)
For the secondary load component arising from the deflection of the primary structure, the deflections and rotations transfer functions are calculated from FEA
for a number of wave cases representative of the
array of regular wave conditions defined in Table 1.
6.2.3 Short Term Fatigue Damage
For a given ship to wave heading, wave frequency. ship speed and loading condition, the total stress can be expressed as follows
n
(17)
For the given stress check point location, ship loading
condition, ship speed, ship heading to waves, and
seastate expressed in terms of significant wave height H113, and mean zero crossing period T, the
short term stress
statistics are calculated. Thespectral function S (w) is calculated directly from
the wave spectral function S(w) (ISSC spectrum),
the transfer function H1 (w) of the ith load process,
and the complex conjugate of the H(co) of th jth load process as follows:
S(w)=S(w)C,C1H,(w)H(w)
(18)
The spectral moments required for calculation of the spectral bandwidth and zero crossing frequency are given as follows
Ab. No. 021 12/17
(19)
Using a closed form solution for the fatigue damage, the short term fatigue damage rate and associated stress cycle rate can be calculated. For a given stress
check point location, ship loading condition, ship speed, ship heading to waves, and seastate expressed ¡n terms of significant wave height H113, and mean zero crossing period T, the accumulated fatigue damage ¡s expressed as follows:
D=''
(21)
N(S,)
The accumulated fatigue damage in one seastate can be expressed as follows
D=
= Tu
and
K
+ i 1m,
Uo
=-:
The deterministic fatigue damage accumulated in a
given seastate can be obtained from the following
expressions TB " Q
D=
(25)
(22)
For a narrow banded process, the accumulatedfatigue damage in one seastate may be rewritten as follows:
B"
n
K I)
JS"p(S) dS t(m,m1,S0,B) (23)
j..i(m,m1
,S0, B)(2J)"
"F
Since the stress process is not a strictly narrow
banded process, a rainflow correction factor X(m,c) is introduced to remove the conservatism due to the
narrow band assumption, Wirshing (1977). The expected number of stress cycles is obtained from the stress process zero crossing frequency as follows
(24)
m
+1 (26)
2For each seastate, the short-term fatigue damage
accumulation rate, and stress cycle rate are computed to enable the computation of the long term fatigue damage.
6.2.4 Voyage Simulation
Since fatigue damage is a cumulative process, and the long term stress range distribution is a function of the long term wave environment, it is essential that due consideration
is given to the derivation of a
realistic wave environment. Using a concept similar to
100 Al longitudinal strength standard based on the
North Atlantic wave environment, the 100 Al fatigue wave environment standard has been formulated. lt is computed systematically for a combination of trading
routes for the ship type, and ship characteristics
subject to the FDA investigation. The trading routes are a direct function of the ship type, and they have
been determined from statistical analysis of
world-wide trading pattern. The Global Wave Statistics data,
BMT (1986), is used to determine a service profile
matrix giving the probabilities of occurrence of the seastates defined in terms of significant wave height, mean wave period, loading condition, ship to wave heading and service speed.
6.2.5 Computation of Long Term Fatigue Damage
The total lifetime accumulated fatigue damage Dt over
a specified service period Ts is computed from the
Q= X(n
).i(m,m ,B,S0)o(2-.h)"o'
probability matrix of occurrence of the short term
seatates as follows'
T,. B"Q
Dt=
(27)
K
where
Q, =
P,P,PkP/,
jkl and ¡s the. .k ,1
stress level parameter for a given seastate i,j,k,l
6.2.6 Fatigue Acceptance Criteria
The fatigue life results are given in two formats as
follows
The conventional deterministic fatigue life format calculated with an S-N curve with an associated
probability of survival of 97,5%, and a fatigue
damage factor of 1,0 for 20 years;
The probabilistic format based on a simple log-norma? format for multiplicative limit state functions yields a probability of failure and a safety index for a given number of service years Wirshing (1987). 7. DOUBLE HULL VLCC FDA APPLICATION To illustrate typical distributions of fatigue damage,
FDA computations have been performed foF an
idealised double hull VLCC ship with uniform typical
end connection design i.e. flat bar web stiffener
200x12 around the envelope.
7.1 Influence of Loading Components
Figures 12,13 illustrate the fatigue damage distribution due to the hydrodynamic wave pressure
alone in head seas, in a fully loaded and ballast
condition respectively for a given seastate (H113,T). lt
can be seen that the maximum fatigue damage
occurs below the waterline. In the fully loaded
condition, the fatigue damage in way of the outer
bottom is comparatively low due to both the
depthwise variation of the longitudinal scantlings, and the exponential decay of the hydrodynamic pressure. In ballast condition, the fatigue damage on the lower part of the side shell tends to be more uniform.
N
/
Fig. 12. Fatigue Damage Factor Fd Distribution
-Hydrodynamic wave pressure only - Full
Load - Head Seas - Envelope Fd= 6.0
Ab. No. 021 13/17
= nr
K
J-''I'll
Fig. 13. Fatigue Damage Factor Ed Distribution -Hydrodynamic wave pressure only - Ballast - Head Seas - Envelope Fd10.0
Figure 14. illustrates the fatigue damage distribution in beam seas. lt is shown that on the weather side of
the ship, the fatigue damage is significantly larger than on the lee side. The influence of the rolling
motion is also noticeable with the maximum fatigue damage zone extending over the waterline.
I
Ï
I
!
Fig. 14. Fatigue Damage Factor Fd Distribution
-Hydrodynamic wave pressure only - Fully
Loaded - Beam Seas - Envelope Fd=20.0 Figure 15. illustrates a typical distribution of fatigue damage due to the wave induced hull girder vertical bending moment.
Fig. 15 Fatigue Damage Factor Fd Distribution
Vertical Bending Moment- Fully Loaded Head Seas - Envelope Fd13.0
Figure 16 illustrates a typical distribution of fatigue damage due to
the wave induced
hull girder horizontal bending moment . The difference in fatigue damage on the side shell s due to the variation of S-N curves due to the difference in the SCF arising from the ratio of the flange thickness to flat bar thickness. Ab. No. 021!I. ï
Fig. 16 Fatigue Damage Factor Ed Distribution Horizontal Bending Moment Fully Loaded -Beam Seas - Envelope Fd=0.15
7.2 100 Al Fatigue Wave Environment
Figure 17 shows the probability distribution of the
wave direction for the subject ship based on the 100
Al fatigue wave environment for large crude
oiltankers. lt is shown that these distributions differ for
the fully loaded and ballast voyage. Due to a slight
dominance of the wave direction, the fatigue darage
is maximised on one side of the ship in the fully loaded voyage, and the other side in the ballast
voyage. 240
20'
180.--Fully Loaded
--- Ballast
a Fully Loaded & Ballast
Fig. 17 Probability Distribution of Wave Direction
Figures 18,19 illustrate the distribution of fat!gue damage at midship and at a forward frame next to the
forward bulkhead of tank No. 1 respectively. lt is
shown that the fatigue damage in way of the midship section is dominated by the vertical bending moment at the deck and bottom, whisit the side shell is subject
to higher fatigue loading in way of the fully loaded
waterline. For the forward section, due to the larger amplitude of wave pressure as a result of the relative motion of the ship, and the reduction in the vertical bending moment amplitude, the side shell and outer
14/17 280 260
3o414
340 320P4
40 60 111111 HhllilIllhli 11h11 Illbottom, especiaHy in way of the fully loaded waterline area is more prone to fatigue damage.
-n.
/=
Fig. 18 Fatigue Damage Factor Ed Distribution -Midship Section -Envelope Fd=1.0
Fig. 19 Fatigue Damage Factor Fd Distribution -Forward Section -Envelope Fd=1 .0
7.3 Alaska to Gulf of Mexico Wave Environment
o 340 320 300 280 260
iI
100 240 220 180 -.-- Fully Loaded-- Ballast
e-- Fully Loaded & Ballast
Fig. 20 Probability Distribution of Wave Direction
Figure 20 shows the probability distribution of the
wave direction for the subject ship based the Alaska to Gulf of Mexico trading pattern . The distributions for the fully loaded and ballast voyage exhibit some degree of symmetry about the quartering seas axis.
Due to the dominance of the quartering wave
direction, t is expected that the fatigue damage will be higher due to the combined action of the vertical and horizontal bending moment, as well as the lateral motions inducing higher hydrodynamic wave pressures.
I
llIllIllIIlllIllll!I. II IlillIllhl lii
/
Fig. 21 Fatigue Damage Factor Ed Distribution
-Midship Section -Envelope Fd=1.0
Fig. 22 Fatigue Damage Factor Fd Distribution -Forward Section - Envelope Fd1 .0 Figure 21,22 illustrate the distribution of fatigue damage at midship and at a forward frame next to the
forward bulkhead of tank No. 1 respectively. lt is
shown that due to
the increased probability ofoccurrence quartering seas, the fatigue damage due to the hydrodynamic pressure in way of the side shell
and the outer bottom is increased. lt should be
pointed out that this wave environment only applies to
a small percentage of the total
trading patternsencountered by this
ship type, and
is given to highlight the significance of the wave environment to fatigue damage.8. FATIGUE CONTROL PLAN AND CONSTRUCTION MONITORING
Since fatigue performance of ship structural details can be influenced by the workmanship standard, the fit-up and the alignment of the structural components,
t is essential that due attention is given during the
construction stages to ensure that the structure will be representative of the assumptions used in the fatigue
design assessment. To achieve this objective, a
Fatigue Control Plan ¡s developed at the plan
Ab. No. 021 15/17
approval stage, and the items to be considered are as follows
Identification of critical areas by the FEA Structural
Design Assessment procedure (SDA), and the
FDA procedure;
Marking of the critical areas and structural details on the ship plan;
Definition of the construction tolerances for the
critical structural details in terms of welding requirements, fit-up and misalignment tolerances. The erection sequence of the blocks is also to be specially considered iii order to minimise
misalignment, and locking of residual stresses
during assembly.
The fatigue control plan uses the fatigue life
results computed by the FDA procedure to
determine the level of inspection during construction.
During the newbuilding construction, the field
surveyor will draw particular attention to the critical areas highlighted by the Plan Approval Office, and the
Fatigue Control Plan. Enhanced levels of visual
inspection,
and NDEINDT may be
required at selected critical locations.9. SUMMARY & CONCLUSIONS
Throughout the development of the ShipRight FDA
procedure, the objectives have been to provide a
flexible, reliable and total approach to assess the
problem of fatigue damage of ship structural details. The main features of the first principle Fatigue Design
Assessment procedure have been reviewed. The
procedure based on the structural spectral method of analysis has been implemented into a user friendly Window TM integrated software for design and
assessment purposes. The disadvantages associated with the application of the traditional maximum load approach to the determination of the long term stress
range spectrum which have prompted the
development of the subject procedure have been
summarised. Novel direct calculation features such as
the S-N Curve Expert, the computation of wave
induced loads by parametric expressions, the use of a voyage simulation procedure have been outlined, and typical application examples have been given to
illustrate the application of the procedure.
In summary, in
order to
attain and maintain asatisfactory fatigue performance of ship structural
details, it is essential that a global approach giving
due consideration to the design, construction and ¡n
service stages of the life of structural details be applied. This has been achieved by
ensuring an adequate level of detail design is
performed by providing the Structural Detail
Design Guide;
ensuring that adequate structural design
concepts and sound analysis techniques are
used by providing a first principle fatigue design assessment tool supported by an ongoing research programme of theoretical and
experimental work;
ensuring that the workmanship, and construction standards are performed to a satisfactory level, and that the fatigue strength can be maintained by the provision of enhanced survey procedures, and the application of a hull condition monitoring system.
10. REFERENCES
British Maritime Technology, Global Wave Statistics,
1985.
BS 5400, Part 10, 1980. Code of Practice for Fatigue.
Steel, Concrete and Composite Bridges. British
Standard Institution.
ECCS - Technical Committee 6 - Fatigue Recommendations for Fatigue Design of Steel Structures, 1985
Ferguson J.M., Violette F.L.M, Some Effects on Ship Structural Design created by the Increased Application of Higher Tensile Steels, Proc. IMSDC 91, Kobe, Japan.
liS/lIW Doc. 700-82, Welding in the World, No 20.7/8,
1982
KSRI, Load Spectra for Ship Structure Fatigue
Evaluations, Restricted, 1992
Lloyd's Register, Paper presented at the Royal
Institution of Naval Architects, London, 1913
Lloyd's Register, ShipRight FDA - Structural Detail Design Guide, 1994
Lloyd's Register, ShipRight FDA - Fatigue Design Assessment - Procedures Manual, 1994
Lloyd's Register, ShipRight SDA - Structural Design Assessment - Procedures Manual, 1994
Lloyd's Register, ShipRight HCM - Hull Condition Monitoring - Procedures Manual, 1994
Lloyd's Register, ShipRight CM - Construction Monitoring - Procedures Manual, 1994
Lloyd's Register, Comparative Fatigue Damage Analysis of a 280,000 Dwt VLCC, September 1992 -Restricted
Offshore Installations : Guidance on Design and Construction. New Fatigue Design Guidance for Steel
Welded Joints in Offshore Structures. UK Den,
August 1983.
Violette ELM, The Effect of Corrosion on Structural Detail Design, Int. Conf. Marine Corrosion Prevention,
Royal Institution of Naval Architects, London, Oct.
1994
Wirshing P.H, Fatigue under wide band random
Stresses using the rainflow method, Journal of Engineering Materials and Technology, ASME, 1977 Wirshing P.H, Chen Y.N., Consideration of Probability - Based Fatigue Design for Marine Structures, Proc. Marine Structural reliability Symposium, Arlington, VA,
1987. 11. NOMENCLATURE
I
ZU.' zzz zYY s ¡f(x)
K8, C, P,(t) H/i) n p(S)a
so S, tS N(S) n(S) B p/mm I,SO,B) m K E T Ab. No. 021SecQnd moment of area Section modulus about flange Hull girder section modulus about ZZ Hull girder section modulus about YY Stiffener spacing
Effective span
Bending moment shape function at
critical location x from span point
Bias stress concentration factor loadcase i
Structural influence coefficient of the ith load process P,(t)
Load process i
Stress process ¡ complex form
Total number of load processes I ship
motions influence parameters Stress range probability function
Standard deviation of the stress process
S-N curve stress range at stress cycles
Nominal stress,stress range
Number of allowable stress cycles at
stress range S,
Number of stress cycles with stress range S for the given seastate
Expected number of stress cycles in the given seastate
Modelling bias for the stress prediction model
Correction factor for multi-linear S-N curve
Slope of selected S-N curve Intercept of selected S-N curve Spectral bandwidth
Seastate duration
Pk =p(kijl)
17/17
U Mean zero crossing frequency
P, = p(i
jkl)
probability ith loading conditionPI =p(i
iki)
probability jth ship to wave headingprobability kth ship speed
p, = p(l ¿1k)
probability Ith seastate (H1I3,TZ)SCF ship structural detail 50F standard detail
Knom Intercept nominal S-N curve
Kh, Intercept reference S-N curve
hv Hot spot stress range
m S-N curve inverse slope