Structural Strength of Work Boats and High Speed Crafts With Pre-Fabricated, Floating Panels in Aluminum

Norwegian University of Science and
Technology
Master’s Thesis
Structural strength of work boats and

high speed crafts with pre-fabricated,
floating panels in aluminum
Author: Supervisor:
Lars Johan Mong Prof. Jørgen Amdahl
June, 2011
Abstract
This thesis evaluates some relevant aspects related to structural integrity for work boats
and high speed vessels with floating frames. This structural design is expected to signifi-
cantly reduce the building cost, but it is also expected to reduce the structural strength of
the vessel. A model of a modified version of the JumboCat 60 (JC60) with floating frames
is used as a recurring example in the calculations.
As of date, the use of floating frames in not acknowledge by the classification societies. The
traditional version of the JC60 is however classified according to the DNV HSLC rules,
and it is shown that the scantlings for the floating frame version exceeds the minimum
requirements in the rules.
Finite element models of the traditional and the floating frame version of the JC60 has
been developed. These has been analysed for three different load conditions as defined in
the DNV Classification Notes 30.8, namely the symmetric bottom slamming, the trans-
verse split force, and the torsional/pitch-connecting moment load condition. These load
conditions are assumed to be the most critical. From the finite element analyses, it is seen
that the structural response for both models is quite similar.
Only the transverse split force load condition analyses showed stresses exceeding the al-
lowable stress levels given in the DNV HSLC rules. However, the results indicated that
the hull beam of the traditional model was slightly stiffer, so it is suggested that the plate
thickness for the shell plating in the floating frame version is increased at some critical
regions. An increase of the longitudinal stiffener shear area is also seen necessary to reduce
high shear stresses at critical areas for the floating frame version.
A fatigue assessment of the longitudinal stiffener-transverse floating frame connection has
been performed. Three different locations were identified as potentially critical in terms
of excessive fatigue damage. Those were located at the middle of the stiffener flange at
the weld toe, at the edge of the stiffener flange at the weld toe, and at the edge of the
frame bottom flange. The long term distribution of stresses was approximated by a two-
parameter Weibull distribution. The shape factor was set to 0.81 and the number of load
cycles for 20 years of service was set to 100 million. The maximum stress range in the load
history was assumed to occur for symmetric bottom slamming. Four finite element models
of the structural detail were developed and analysed for determination of the maximum
stress range. It was found by Miner summation that none of the locations considered
would experience a critical fatigue failure for 20 years of service.
Preface
This thesis is written as the last step to complete my Master of Science in Marine Technol-
ogy at Norwegian University of Science and Technology. It is a continuation of the project
work performed by myself during the fall semester 2011 and the Master’s Thesis’ written
by Jon Englund at KTH in december 2009 and Toralf Ervik at NTNU in june 2010.
Finite element modelling and analyses of those proved very time consuming, partly due to
technical problems and my own lacking experience. In order to obtain satisfying results,
some trial and error has been needed, but in the end I am satified with the results from the
finite element analyses. This has lead to that some parts of the scope are left unanswerred,
like the non-linear analysis of the longitudinal stiffener-floating frame connection. On the
other hand, a slamming analysis has been performed, although it was not included in
the scope. However, it was seen necessary to analyse in relation to the fatigue analysis,
and also to compare the structural response of the strengthened floating frame with the
floating frame as used in Toralf Erviks finite element models.
The cumulative fatigue damage was supposed to be calculated by both the nominal stress
range approach and the hot spot stress range approach. The latter proved very difficult, as
the hot spot stress range approach was designed for offshore steel structures. Consequently,
no relevant SN-curves corresponding to the structural detail analysed were obtained. A
possibility was to create my own SN-curve based on a series of assumptions to be used with
the hot spot stresses, but this was thought to give very inaccurate results. A calculation
of cumulative fatigue damange using the hot spot stress range approach was thus omitted
from the thesis. As only nominal stresses were considered, there was no point in including
a discussion about the stress concentration factors.
I would like to thank my supervisor Professor Jørge Amdahl at the Department of Marine
Technology at NTNU for his invaluable guidance and help related to all aspects of this
thesis. I would also like to express my gratitude to PhD Candidate Arswendy Arswendy
(NTNU) for useful help related to finite element modelling, Norwegian Sportsperson of
the Year 1970 Professor Stig Berge (NTNU) for providing relevant papers on fatigue, and
Stig Oma at Fjellstrand AS for valuable comments related to technical aspects.
Trondheim, June 2011

Contents
Abstract i
Nomenclature viii
Symbols . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . viii
Abbrevations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . x
1 Introduction 1
1.1 Scope . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.2 Background . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
1.3 JumboCat 60 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
1.4 Previous work . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
1.4.1 JumboCat 60 with floating frames . . . . . . . . . . . . . . . . . . . 9
1.4.2 The Floating Frame Principle . . . . . . . . . . . . . . . . . . . . . . 9
2 Design loads and strength requirements 11

2.1 Design codes in general . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
2.2 Design Loads . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
2.2.1 Slamming Loads . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
2.2.2 Sea Pressure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
2.2.3 Deck Loads . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
2.2.4 Transverse Bending Moment and Split Force . . . . . . . . . . . . . 13
2.2.5 Pitch connecting and twin hull torsional moment . . . . . . . . . . . 14
2.3 Strength requirements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
2.3.1 Hull girder strength . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
2.3.2 Plating and Stiffeners . . . . . . . . . . . . . . . . . . . . . . . . . . 16
2.3.3 Web frames and girders . . . . . . . . . . . . . . . . . . . . . . . . . 17
2.4 Direct calculation methods . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
2.4.1 Symmetric bottom slamming . . . . . . . . . . . . . . . . . . . . . . 19
2.4.2 Transverse split force . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
2.4.3 Torsion moment and pitch connecting moment . . . . . . . . . . . . 21
3 Finite Element Modelling 23

3.1 General . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
3.2 Torsion/pitch connecting model . . . . . . . . . . . . . . . . . . . . . . . . . 25
3.3 Transverse bending model . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
3.4 Slamming model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
4 Results from finite element analyses 31

4.1 Interpretation of results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
iii
4.2 Torsion/pitch connecting analysis . . . . . . . . . . . . . . . . . . . . . . . . 32
4.3 Transverse bending analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . 35
4.3.1 Transverse bulkheads . . . . . . . . . . . . . . . . . . . . . . . . . . 38
4.3.2 Transverse web frames . . . . . . . . . . . . . . . . . . . . . . . . . . 39
4.3.3 Plating and stiffeners . . . . . . . . . . . . . . . . . . . . . . . . . . 39
4.4 Slamming analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42
5 Fatigue assessment 45
5.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45
5.2 Development of the loading history . . . . . . . . . . . . . . . . . . . . . . . 48
5.3 Finite element models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53
5.4 Calculation of fatigue damage . . . . . . . . . . . . . . . . . . . . . . . . . . 57
5.4.1 HS A . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59
5.4.2 HS B . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61
5.4.3 HS C . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62
6 Conclusions 65
6.1 Recommendations for further work . . . . . . . . . . . . . . . . . . . . . . . 66
Bibliography 66
iv
List of Figures
1.1 The Floating Frame Principle . . . . . . . . . . . . . . . . . . . . . . . . . . 5

1.2 Typical work boat . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
1.3 Distribution of external lateral pressure in bottom structure . . . . . . . . . 7
1.4 Out of plane bending of stiffeners at ship side, floating frame . . . . . . . . 7
1.5 JumboCat 60 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
2.1 Still water transverse bending moment . . . . . . . . . . . . . . . . . . . . . 14

2.2 Combined torsional and pitch connecting twin hull moments . . . . . . . . . 15
2.3 Local load condition 2, symmetric bottom slamming [5] . . . . . . . . . . . 19
2.4 Global load condition 4, transverse split force, split outwards shown . . . . 20
2.5 Possible modelling of global load condition 5, torsion moment/pitch con-
necting moment, [5] . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
3.1 Web frames used in models . . . . . . . . . . . . . . . . . . . . . . . . . . . 24

3.2 Bulkheads used in models . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
3.3 Extent of models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
3.4 Outlines of models used for torsion/pitch connecting analyses . . . . . . . . 25
3.5 Idealised pressure distribution used in analyses . . . . . . . . . . . . . . . . 26
3.6 Boundary conditions, torsion/pitch connecting moment . . . . . . . . . . . 27
3.7 Loads and boundary conditions, transverse bending moment. Inward split
force shown . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
3.8 Pressure loads in slamming analyses . . . . . . . . . . . . . . . . . . . . . . 28
3.9 Modelling of passenger deck loads . . . . . . . . . . . . . . . . . . . . . . . . 29
3.10 Boundary conditions in slamming analyses . . . . . . . . . . . . . . . . . . . 29
4.1 Bending stress at frame #49 . . . . . . . . . . . . . . . . . . . . . . . . . . 33

4.2 Equivalent stress at frame #49 . . . . . . . . . . . . . . . . . . . . . . . . . 33
4.3 Comparison of torsional neutral axis . . . . . . . . . . . . . . . . . . . . . . 34
4.4 Comparison of warping deformation at frame #27 . . . . . . . . . . . . . . 35
4.5 Equivalent stress at frame #35, region 1 highlighted . . . . . . . . . . . . . 37
4.6 Stress at frame #35, floating frame model . . . . . . . . . . . . . . . . . . . 38
4.7 Extent of area with higher stress than allowable . . . . . . . . . . . . . . . . 38
4.8 Equivalent stress at highest loaded transverse web frame, region 2 highlighted 39
4.9 Bending stress at highest loaded transverse web frame . . . . . . . . . . . . 39
4.10 Local deformation at region 1, frame #35 floating frame model . . . . . . . 40
4.11 Comparison of plating equivalent stress . . . . . . . . . . . . . . . . . . . . 41
4.12 Comparison of plating bending stress . . . . . . . . . . . . . . . . . . . . . . 41
4.13 Comparison of stiffener web equivalent stress, frame #35 . . . . . . . . . . . 42
4.14 Comparison of stiffener web shear stress, frame #35 . . . . . . . . . . . . . 42
4.15 Comparison of web frame stiffness . . . . . . . . . . . . . . . . . . . . . . . 43
v
4.16 Vertical deformation at keel, bottom plating . . . . . . . . . . . . . . . . . . 44
5.1 Critical areas for fatigue consideration of high speed catamarans, [19]. . . . 46
5.2 Structural detail to be evaluated for fatigue strength . . . . . . . . . . . . . 46
5.3 Potential crack initiation sites . . . . . . . . . . . . . . . . . . . . . . . . . . 47
5.4 Relevant constructional details, Eurocode 9 . . . . . . . . . . . . . . . . . . 47
5.5 Flow chart for fatige analysis, [19]. . . . . . . . . . . . . . . . . . . . . . . . 48
5.6 Exceedances of stress ranges represented by the Weibull distribution with
different shape parameters . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51
5.7 Example of sub-division of load spectra to stress range blocks . . . . . . . . 52
5.8 Load spectras . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53
5.9 Equivalent stress for bottom stiffener-transverse web frame connection, slam-
ming analysis. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53
5.10 Concentrated forces applied to web frame to replicate bending moment. . . 54
5.11 Springs acting in vertical direction to carry slamming load . . . . . . . . . . 55
5.12 Model 1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56
5.13 Model 2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56
5.14 Model 3 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57
5.15 Model 4 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57
5.16 SN-curves to be used for the cumulative fatige damage calculation. . . . . . 58
5.17 Bending stress normal to weld along horizontal line at center of stiffener
flange . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59
5.18 Cumulative long term distribution of stresses and corresponding SN-curve . 60
5.19 Bending stress normal to weld along horizontal line at edge of stiffener flange 61
5.21 Bending stress parallell to weld along horizontal line at edge of web frame
bottom flange flange . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62
vi
List of Tables
2.1 Allowable bending stresses for plating and stiffeners . . . . . . . . . . . . . 16

2.2 Allowable stresses for web frames and girders . . . . . . . . . . . . . . . . . 18
2.3 Allowable stresses for web frame analysis . . . . . . . . . . . . . . . . . . . . 20
4.1 Allowable stresses, global analyses . . . . . . . . . . . . . . . . . . . . . . . 32

4.2 Stresses from torsion analyses . . . . . . . . . . . . . . . . . . . . . . . . . . 32
4.3 Torsional stiffness . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35
4.4 Allowable stresses and stresses from analyses . . . . . . . . . . . . . . . . . 36
4.5 Maximum stresses from analyses . . . . . . . . . . . . . . . . . . . . . . . . 37
4.6 Allowable stresses, web frame analysis . . . . . . . . . . . . . . . . . . . . . 42
4.7 Maximum stresses from slamming analyses . . . . . . . . . . . . . . . . . . 43
vii
Nomenclature
Symbols
acg Design vertical acceleration at LCG

av Design vertical acceleration
a∗ Number of stiffeners between consideres section and nearest support
ā Geometrical/material parameter in SN-curve
b Distance between hull centerlines
b∗ Breadth of load area
f1 Mechanical property factor
g0 Standard acceleration of gravity
h0 Vertical distance from the waterline to the load point
ka , kb Coefficients depending on element considered
kl Longitudinal slamming pressure distribution factor
kr Correction factor for curved plates
ks Sea load distribution factor
ksf Shear force factor
l Stiffener span
m See ā
m∗ Bending moment factor
n Number of hulls
n0 Total number of load cycles in the load history
ni Number of cycles in stress range block i
pcar Design pressure at car deck
pHS Hydrostatic pressure
psp Design sea pressure
psp,keel Design sea pressure at keel
psp,W L Design sea pressure at waterline
r∗ Average point load from stiffeners between considered section and nearest support
s Stiffener spacing
t Thickness
yb Distance from the center line to local center of one hull
z Height from base line to NA of cross structure
A Design load area
AS Minimum shear area for stiffener
AW Effective web area of girders
B Beam over all
Bhull Breadth of one hull
BM AX Maximum width of submerged part
viii
BW L Maximum width in the water line
CW Wave coefficient
D Cumulative fatigue damage
Di Fatigue for stress range block i
Fy Horizontal split force on immersed hull
H Stowage height
H1 Maximum significant wave height in which the vessel is allowed to operate
HS Significant waveheight
L Length of the craft
LBM AX Length where BM AX /BW L > 1
M Bending moment
MP Twin hull pitch connecting moment
MS Twin hull transverse bending moment
Msplit See MT W
MS0 Still water transverse bending moment
Mt Twin hull torsional moment
MT W Dynamic transverse bending moment
Ni Number of cycle to failure for stress range block i
Psl,b Design bottom slamming pressure
Psl,p Design pitching slamming pressure
S Stress range
S0 Maximum stress range in the load history
SG Girder span
Si Applied stress range in block i
T Fully loaded draught
T0 Draught at L/2 at normal operation
TL Lowest service speed draught at FP
Tz Zero crossing period
V Maximum speed
Z Section modulus
βx Deadrise angle at transverse section
βcg Deadrise angle at LCG
θ Heading
λ Weibull scale parameter
ξ Weibull shape parameter
ξW C Weibull shape parameter, worst case
ρ Density of liquid
σ Bending stress
σa Stress amplitude
σe Equivalent stress, Mises stress
σsl Allowable bending stress for slamming load
σx Bending stress in x-direction
σy Bending stress in y-direction
τ Shear stress
τsl Allowable shear stress for slamming load
τxy Shear stress in the xy-plane
ω Wave frequency
ωe Wave encounter frequency
∆ Fully loaded displacement
ix
Abbrevations
CN 30.8 Classification Notes No. 30.8
DNV Det norske Veritas
DOF Degree of freedom
FEA Finite element analysis
FP Forward perpendicual
HAZ Heat affected zone
HSLC High speed, light craft and naval surface craft
JC60 JumboCat 60
LCG Longitudinal center of gravity
LLC2 Local load condition 2
GLC4 Global load condition 4
GLC5 Global load condition 5
NA Neutral axis
WL Waterline
x
Chapter 1
Introduction
1.1 Scope
MASTER THESIS VÅREN 2011

for
Stud. Techn. Lars Johan Mong
Konstruksjonsstyrke til bruksbåter og hurtiggående båter med prefabrikerte,
utenpåliggende paneler i aluminium
Structural strength of work boats and high speed crafts with pre-fabricated, floating panels
in aluminium
Aluminium er et hyppig brukt materiale for bygging av bruksbåter til oppdrettsnæringen,
hurtiggående passasjerbåter og katamaranferger. Produksjon av båtskrog har tradisjonelt
krevd mye arbeidsinnsats, med tilpassing av plater, stivere, spant og andre elementer, og
et stort omfang av manuell sveising.
I prosjektet ALUBÅT er målsettingen å komme fram til mer kostnadseffektive måter for
å produsere aluminiumsskrog til bruksbåter. Det er utviklet konstruksjonsløsninger og
bygget mindre båter med utstrakt bruk av prefabrikkerte paneler, bestående av friksjon-
ssveiste ekstruderte aluminiumsprofiler. Panelene er brukt til båtdekk, -sider og -bunn.
Med paneler menes her hudplater med ferdige stivere.
En effektiv og økonomisk bruk av panelene i båtbygging forutsetter at panelene kan legges
utenpå skott og tverrammer (spant). Dette til forskjell fra tradisjonell bygging av stål-
og aluminiumbåter, hvor stivere påsveises platene. Videre føres stiverne normalt gjennom
spantene via utsparinger i disse. Stiverstegene sveises normalt til spantene, men det er
også eksempler på at de kun sveiste mot spantene i stiverens toppflens, evt. i kombinasjon
med brakett for skjæroverføring.
Bruk av prefabrikkerte paneler bestående av ekstruderte profiler sammenføyd med frik-
sjonssveis innfører ingen nye materialkvaliteter i båtbygging. Materialet som brukes er
vanligvis aluminiumlegeringene 6082 og 6005A, i behandling T6, som har godt doku-
menterte fasthetsegenskaper. Materialet svekkes ved sveising, men ikke vesensforskjellig
fra andre vanlige brukte aluminiumslegeringer og leveringstilstander.
Skrogkonstruksjon med bruk av paneler hvor stiverne legges utenpå skott og tverrammer
(spant) krever at det må regnes på mulig styrkereduksjon i kryssingspunktet mellom sti-
1
2 Introduction
vere og skott/tverrammer. Beregningsanvisning for dette er ikke beskrevet i de vanlige

dimensjoneringsregler for båter, hverken Nordisk Båtstandard av 1990 for yrkesbåter un-
der 15 m eller Det norske Veritas regler for hurtiggående båter (High Speed, Light Craft
and Naval Surface Crafts - HSLC). For hurtiggående båter aksepteres pr. dato ikke bruk
av utenpåliggende paneler.
Det er derfor ansett nødvendig å utvikle modifiserte dimensjoneringsregler for prefab-

rikkerte paneler. En forutsetning har vært at de modifiserte regler skal ivareta sikkerhet-
snivået som er implisitt i dagens regelverk.
Det var umiddelbart ikke klart hvilken dimensjoneringsfilosofi som ligger innbakt i Nordisk
Båtstandards krav til platetykkelse og motstandsmoment for stivere. Tilsvarende krav i
DnV’s Tentative Rules for Certification and Classification of Boats, 1997 og DnV Rules
for Classification of High Speed, Light Craft and Naval Surface Craft har derfor også blitt
vurdert . Dette arbeidet har bidratt til å avdekke grunnlaget for kravene i Nordisk Båt-
standard, og sammenligninger har vist at forskjellene mellom de tre regelverkene er mod-
erate. Når det gjelder yrkesbåter under 15 m har man derfor valgt å foreta modifikasjon
av Nordisk Båtstandard, som er den standarden aktuelle byggere av bruksbåter er kjent
med. Det har her spesielt blitt sett på kravene til motstansmoment av stivere/spant utsatt
for tverrbelastninger.
Styrkereduksjonen på grunn av opplagerkraft (konsentrert kraft) i kryssingspunkt mellom

stivere og spant er tatt hensyn til ved at kravet til motstandsmoment økes proporsjonalt
med den reduserende virkningen som forårsakes av den konsentrerte opplagerkraften. Det
er tatt utgangspunkt i en anerkjent dimensjoneringsprosedyre for kapasitet for plater med
konsentrert last på platerand gitt i Eurocode 9.
Et moment som kan ha betydning er evnen utenpåliggende paneler har til å oppta globale
skjærkrefter via lastinnføring fra tverrspant. Denne overføringen kan kun skje via sveisen
til panelets toppflens mot spant, i motsetning til over hele skipssiden ved tradisjonell
utførelse. Globale skrogbelastninger har generelt mindre betydning for (små) bruksbåter
og Nordisk Båtstandard inneholder derfor ikke krav til kontroll av global skjærkraft. HSLC
har derimot et eksplisitt krav til en slik kontroll ikke minst da tverrspant kan bli utsatt
for betydelige dynamiske laster ved høy fart i sjø.
Ved utenpåliggende panel vil ikke panelet bidra med effektiv flens til bøyning av tver-
rammene. Tverrammene må derfor dimensjoners for å bære lastene som innføres uten
medvirking fra huden. Dette gir noe økte dimensjoner på tverrammene, men for øvrig
synes det ikke å by på konstruktive utfordringer. Da tøyninger og spenninger i bunnen
av tverrammene ikke overføres til huden, vil imidlertid huden få en relativ forskyving i
forhold til bunnen av tverrammene. Det medfører en bøyning av stiverstegene i det pre-
fabrikkerte panelet ut av stivestegets plan. I et tidligere arbeid utført av Jon Englund
er disse spenningene funnet å være betydelige. Slike bøyespenninger er ikke til å un-
ngå med utenpåliggende panel; spørsmålet er hvor store spenninger som kan aksepteres.
HSLC reglene omhandler ikke direkte bruk av slike paneler og de spesielle tøynings- og
spenningstilstander som opptrer ved bruk av panel-løsningene, slik at nye vurderinger må
utføres.
Hensikten med denne oppgaven er å videreføre arbeidet til Jon Englund og Toralf Ervik.
Den vil spesielt fokusere på spenninger fra globale laster; tverrskips bøyemoment, torsjon-
smoment og "pitch-connecting" moment. Målet for oppgaven er å utarbeide grunnlag for
å søke Veritas godkjennelse av en aktuell, planlagt hurtiggående båt fra Fjellstrand med
Scope 3
utenpåliggende paneler.
Oppgaven foreslås gjennomført i følgende trinn:
1. Beskrive relevante globale laster og styrkekrav for skrogbjelken i henhold til DnV
HSLC. Gjennomgang av relevante anbefalte metoder for direkte styrkeberegninger
gitt i DnV Classification Note No. 30.8.
2. Gjennomgang og forbedring av modell av skrogbjelken for Fjellstands 60M Jumbo-
cat, utviklet i prosjektarbeidet høsten 2010. På grunnlag av denne modellen utføre
lineære analyser av skroget utsatt for tverrskips bøyemoment og torsjonsmoment.
Spenningsnivåene vurderes i forhold til akseptkriterier gitt i DnV HSLC og Class.
Note 30.8. Spenningskomponenter som eventuelt ikke dekkes av disse reglene kom-
menteres og diskuteres.
3. Vurdering av konstruksjonens godhet med hensyn på utmatting. Det er her ak-
tuelt å studere sveiseforbindelsene mellom panelene og tverrammene i skrogbunn
og i skipsside. En detaljert elementmodell lages for å bestemme relevante spen-
ningskonsentrasjonsfaktorer. Aktuelle levetidskurver for sveiseforbindelsene kart-
legges, for eksempel med utgangspunkt i Eurocode 9 for konstruksjoner i aluminium.
Spenningskonsentrasjonsfaktorer gitt i regelverk sammenlignes med resultat fra egne
beregninger.
4. På grunnlag av enkle vurderinger utarbeides anslag på langtidsfordelingen til spen-
ningsviddene, for eksempel basert på to-parameter Weibullfordeling. Utmattingsleve-
tiden til forbindelsene beregnes, og følsomhetsvurderinger foretas.
5. I den grad tiden tillater kan det også foretas analyser av forbindelsenes sammen-
bruddsstyrke ved hjelp av ulineær elementmetode (Abaqus).
6. Konklusjoner og forslag til videre arbeid.
Referanser:
Jon Englund: Structural strength of work boats and high speed crafts with floating frames,
MSc Thesis KTH, Centre for Naval Architecture, Stockholm, December 2009.
Jon Englund: Oppsummering av FE-analyser på Jumbocat 60 med flytande ramar, Oma
17.11.2009
Jon Englund: Finit element-analys av Knut-Johan, Oma 17.11.2009
Toral Ervik: Structural strength of work boats and high speed crafts ith pre-fabricated,
floating panels in aluminium, MSc Thesis NTNU, June 2010.
Literature studies of specific topics relevant to the thesis work may be included.
The work scope may prove to be larger than initially anticipated. Subject to approval
from the supervisors, topics may be deleted from the list above or reduced in extent.
In the thesis the candidate shall present his personal contribution to the resolution of
problems within the scope of the thesis work.
Theories and conclusions should be based on mathematical derivations and/or logic rea-
soning identifying the various steps in the deduction.
The candidate should utilise the existing possibilities for obtaining relevant literature.
4 Introduction
Thesis format
The thesis should be organised in a rational manner to give a clear exposition of results,
assessments, and conclusions. The text should be brief and to the point, with a clear
language. Telegraphic language should be avoided.
The thesis shall contain the following elements: A text defining the scope, preface, list
of contents, summary, main body of thesis, conclusions with recommendations for further
work, list of symbols and acronyms, references and (optional) appendices. All figures,
tables and equations shall be numerated.
The supervisors may require that the candidate, in an early stage of the work, presents a
written plan for the completion of the work. The plan should include a budget for the use
of computer and laboratory resources which will be charged to the department. Overruns
shall be reported to the supervisors.
The original contribution of the candidate and material taken from other sources shall be
clearly defined. Work from other sources shall be properly referenced using an acknowl-
edged referencing system.
The report shall be submitted in two copies:
• Signed by the candidate
• The text defining the scope included
• In bound volume(s)
• Drawing and/or computer prints which cannot be bound should be organised in a
separate folder.
Thesis supervisor
Prof. Jørgen Amdahl
Contact person at Fjellstrand: Stig Oma
Deadline: June 15 2011
Trondheim, January 15, 2011
Jørgen Amdahl
1.2 Background
When designing a marine vessel, the designer tries to achieve minimum life cycle costs
for the vessel while still fulfilling its requirements to serviceability and safety. For high
speed, light craft and naval surface craft, there exists a wide range of rules addressing
both serviceability and safety.
One of many important aspects of ship design is the weight of the hull. The payload will
usualy be defined by the customer, so a challenge for the designer is to obtain minimum
ship displacement while still fulfilling requirements set by both rules and customer. For
high speed vessels, the hull resistance is often dominated by friction, as opposed to wave
generation, which is the main contributor to hull resistance for most types of commer-
cial ships. Thus, in order to minimize life cycle costs, hull weight is in general of more
importance for high speed vessels than for other types of vessels.
Background 5
While steel is the fabrication material for most commercial ships, high speed and light
craft often use other, more lightweight materials. The mass density of aluminum is ap-
proximately one third of that of steel. Aluminum may be fabricated in practically the
same way as steel, thus making aluminum an attractive fabrication material for the ship-
builders. There are however a few drawbacks concerned with the use of aluminum versus
steel. Compared to steel, aluminum is more expensive, weaker, and has a significantly
reduced fatigue performance [22].
Aluminum hull design has traditionally been very much alike the design of steel hulls.
The hull girder strength in longitudinal and transverse direction has been provided by
stiffeners, web frames, bulkheads, etc., all welded to the outer plating and decks of the
vessel. The fabrication requires a lot of fitting and manual welding. It is desireable for
the shipbuilders that the manual labour in the fabrication is kept to a minimum as it is
generally less efficient and more expensive than automated fabrication.
The floating frame concept has been conceived as a measure to reduce both fabrication
time and costs. This is achieved by significantly reducing the need for manual labour. As
the concept name suggests, the transverse web frames of the vessel are floating, meaning
that they are not welded to the outer plating as in traditional hull girder design. Instead,
the web frames are placed directly on top of the longitudinal stiffeners, as illustrated in
figure 1.1.
Figure 1.1: Above: Traditional hull girder design. Below: The floating frame principle.
The use of floating frames has up to now been limited to relatively small crafts, mainly
work boats with lengths up to 15 meters. The seaworhiness of such crafts is not limited by
the same set of rules that apply to larger vessels, i.e. rules set by classification societies.
However, the trend of the latest years is that the displacement of work boats, for exam-
ple used in the fish farming industry, has increased due to more advanced and heavier
equipment installed on these work boats. The shipbuilders have complied to this industry
demand by designing wider, not longer, work boats to avoid the need for classification as
required for longer vessels.
6 Introduction
Figure 1.2: Typical work boat used in the fish farming industry.
As of date, the rules set by the classification societies, for example Det norske Veritas
(DNV), does not acknowledge the use of floating frames for vessels defined as High Speed,
Light Craft and Naval Surface Craft, nor any other types of vessels. Prosjekt Alubåt
was a research project launched and funded by the Norwegian Research Council with the
purpose of aiding norwegian aluminum shipbuilders. Among the project participants were
Norsk Industri, the Norwegian University of Science and Technology (NTNU), and several
shipbuilders. One of the topics examined in the project was the use of floating frames
for aluminum ships with lengths exceeding 15 meters. This would thus require a new set
of rules or a modification of the current rules set by the classification societies. Prosjekt
Alubåt ended in 2010 without any final conclusion in regards to the use of floating frames
for larger aluminum vessels.
From a structural point of view, the floating frame concept is inferior to traditional hull
girder design. By welding the tranverse web frames on top of the longitudinal stiffeners,
the frames no longer have the outer plating acting as their bottom flange. This may be
compensated for by adding a bottom flange to the transverse web frames and increasing
the main dimensions. This will increase the hull weight as more material will be used to
maintain the same strength as for the traditional design.
Another important issue is the distribution of stresses and possible stress concentrations.
External forces acting on the outer plating can not be transfered directly from the outer
plating to the transverse web frames, but must be transfered via the longitudinal stiffeners
as bending and shear stresses. The weld in the stiffener-flange connection may potentially
have to carry large loads. The strucure will be subjected to cyclic loading, so there may
be some fatigue issues at the weld connecting the stiffeners and web frames. Large out-
of-plane bending stresses in the stiffener webs may occur due to frame deflections.
Background 7
(a) Floating frame (b) Traditional frame
Figure 1.3: Distribution of external lateral pressure in bottom structure
Figure 1.4: Out of plane bending of stiffeners at ship side, floating frame
8 Introduction
1.3 JumboCat 60
The JumboCat 60 is a high speed catamaran introduced in 1996, and since produced
various versions. It has a loading capacity of approximately 60 vehicles and 450-600
passengers.
Figure 1.5: The JumboCat 60
It has the following main dimensions:

• Length over all, LOA, 60.0 meters
• Length between perpendiculars, LPP, 54.0 meters
• Breadth, moulded, B, 16.5 meters
• Depth, moulded, D, 5.85 meters
• Draught, fully loaded, T, 2.24 meters
• Displacement, fully loaded, ∆, 580 metric tonnes
• Design speed, V, 35 knots
A norwegian yard has designed a modified version of the JumboCat 60 (JC60) with floating
frames, and this design will be evaluated in the following sections. Specific details about
the design will however not be included due to confidentiality reasons.
Previous work 9
1.4 Previous work
1.4.1 JumboCat 60 with floating frames
As mentioned in 1.2, the possible use of floating frames was evaluated in Prosjekt Alubåt.
Included in this evaluation were the Master’s Thesis’ written in 2009 and 2010 by Jon
Englund at Kungliga Tekniska högskolan and Toralf Ervik at NTNU, respectively. Both
Englund and Ervik performed finite element analyses (FEA) of a midship compartment of
the JC60 with floating frames suscepted to what is defined as Local Load Condition 2, see
section 2.2. Their analyses revealed structural stresses exceeding the allowable as defined
by the strength requirements set by DNV, see section 2.3[12][11]. Ervik concluded that a
strengthening of the transverse web frames was necessary to comply with the classification
rules.
Furthermore, Ervik briefly discussed the fatigue performance of a selected structural detail,
and found that it would have a fatigue lifetime exceeding the assumed 20 year lifetime of
the vessel. The work of Englund and Ervik was continued by Lars Johan Mong. During the
fall of 2010, he performed FEA for a simplified model of the JC60 with both traditional
hull girder design and floating frames. At this point, the design of the floating frames
were modified and a bottom flange was added to the transverse web frames. Subjected
to a combination of torsional and pitch-connecting moment (see section 2.2), the models
showed comparable strength, both within the requirements set by the rules [21].
1.4.2 The Floating Frame Principle
The floating frame principle has been used for various smaller vessels. The rescue crafts
Knut Johan and at least four other rescue crafts of the Norwegian Rescue Company all
had floating frames. Knut Johan had a length over all of 13.80 meters, a displacement
of 20 metric tonnes, and a service speed of 20 knots. According to [1], Knut Johans
floating frame section modulus was significantly lower than the required by Nordic Boat
Standard. This was supported by Englund, who performed FEA on Knut Johan. He found
that the bending stresses in the floating frames exceeded the allowable stress level [10].
Despite of this, an inspection of Knut Johan in 2006 discovered no structural deficiencies
or weaknesses after 30 years of service [1]. This was also confirmed by the Norwegian
Rescue Company and the Knut Johans crew.
Aalberg indicated that the interaction between the frame and the outer skin might be the
reason why the structure proved strong enough. The conservativeness of the design loads
was also deemed a possible explanation. The outer skin-floating frame interaction was
evaluated by Ervik, and he showed that the outer plating and the longitudinal stiffeners
provided only a small increase in stiffness as compared to the frame itself [12]. It should
however be noted that Ervik studied the JC60 hull while Aalberg studied the hull of Knut
Johan, two different hulls in terms of hull shape and frame design. The experience from
Knut Johan can not be considered as completely compatible for other types of ships.
Latorre and Herrington of the University of New Orleans discussed the use of floating
frames aluminum high-speed craft. They proposed a hull girder design with alternating
fixed and floating transverse frames, and found that it had adequate strength for use in a 40
knot high speed craft with a length of approximately 40 meters [16]. This was confirmed
by both FEA and experimental tests. It was also suggested that a proper selection of
10 Introduction
panel components and available extrusions could provide a weight reduction of 15%. A
continuation of the study was published in a paper in 2000. They confirmed the results,
but did not provide any conclusions related to fatigue despite it being part of the study
[17].
Chapter 2
Design loads and strength

requirements
2.1 Design codes in general
A wide range of rules and regulations for seagoing vessels are intended to preserve the
safety of life and goods. Three sets of rules with possible applicability for a large high
speed vessel with floating frames has been evaluated in [2], namely
• Nordic Boat Standard,
• DNV Tentative Rules for Certification and Classification of Boats,
• DNV Rules for Classification of High Speed, Light Craft and Naval Surface Craft
(DNV HSLC).
All of these sets of rules are intended for vessels with a traditional hull girder design. The
use of them is hence problematic, and one needs to try to interpret the level of safety given
by them implicitly to make them applicable for the case at hand. A combination of the
three is impossible as the design loads are defined differently [2]. The most suitable and
advanced set of rules has been identified as DNV HSLC, and will in the following be used
as basis for determination of the structural integrity of the JC60 with floating frames.
2.2 Design Loads
The design loads to be considered are defined in DNV HSLC part 3, chapter 1. When
considering the loads that will act on the vessel, no differentiation between traditional hull
girder design and floating frame design is needed. In general, the loads are determined by
the vessels main characteristics, service restriction, hull shape, etc. None of these factors
differ when comparing traditional and floating frame design.
In the following sections, only those loads relevant for the load conditions to be examined
are presented. They may be categorised as slamming loads, sea pressure, deck loads, trans-
verse bending moments and loads, and pitch connecting and twin hull torsional moments.
11
12 Design loads and strength requirements
2.2.1 Slamming Loads
[7] separates between bottom slamming and pitching slamming. They are defined, repec-
tively, as
Sec. 2, C201 Bottom slamming
∆ 50 − βx
0.3 !
Psl,b = 1.3kl TO 0.7 acg (2.1)
nA 50 − βcg
Sec. 2, C203 Pitching slamming
21 20TL

Psl,p = ka kb CW 1 − (2.2)
tan(βx ) L
For frames, [12] found Psl,b = 82.3 kN/m2 and Psl,p = 80.7 kN/m2 .
It is somewhat unclear how DNV arrived at these formulas, and it is difficult to determine
the accuracy of the results. However it is assumed that the formulas provide conservative
estimates of the actual slamming loads.
2.2.2 Sea Pressure
The sea pressure acting on the craft’s bottom and side below design waterline is defined
in Sec.2 C501 [7]:
h0

psp = 10h0 + ks − 1.5 CW (2.3)
T
This gives pkeel = 48.5 kN/m2 and pW L = 32.6 kN/m2 . The values of the coefficients
are taken from [18]. By comparison, the hydrostatic pressure at the keel, meaning at a
depth of 2.24 meters, is pHS = 22.5 kN/m2 .
2.2.3 Deck Loads
For the deck loads, it is convenient to distinguish between the car deck and the passenger
deck. Section 2 C701 in [7] requires pressure on decks to be taken as:
p = ρH (g0 + 0.5av ) (2.4)
For accomodation decks, i.e. passenger decks, the load parameter is ρH = 0.35 t/m2 ,
yielding a passenger deck pressure of ppass = 5.35 kN/m2 [18].
In [18], Økland investigated the difference of using point loads for the cars versus a dis-
tributed pressure over the area of the cardeck. He applied point loads of 15.3 kN for
each wheel in one load case, and a distributed pressure of 4.0 kN/m2 for the other load
case. It was determined that the results from the two cases were practically identical. For
simplicity, pcar = 4.0 kN/m2 will thus be used.
Design Loads 13
2.2.4 Transverse Bending Moment and Split Force
[7] Sec.3 B202 states that the twin hull transverse bending moment shall be assumed to
be the greater of:
acg

MS = MS0 1 + (2.5)
g0
MS = MS0 + Fy (z − 0.5T ) (2.6)
In this case, where acg = g0 , this means the greater of double the still water transverse
bending moment and the sum of the the still water transverse bending moment and mo-
ment induced by the transverse horizontal split force. The horizontal split force on the
immersed hull is taken as:
V

Fy = 3.25 1 + 0.0172 √ L1.05 T 1.30 (0.5BW L )0.146 (2.7)
L
" #
LBM AX LBM AX BM AX 2.10

· 1− + H1 (2.8)
L L BW L
However, Økland used a different formula, taken from an older version of the DNV HSLC
rules:
6 70 V L

Fy = 0.1L 2
1.6 − √ 1 + 0.1 √ 53 − (2.9)
0.5BW L
1.5
L L L
T
Obviously, there has been a modification of the formula. Økland also investigated the
JC60. His version had a displacement of 560 metric tonnes and a draft of 2.15 meters.
One would anticipate a smaller force for a vessel with lower draft and displacement given
that the vessels were otherwise identical. This is however not the case. Using Øklands
input data and equation 2.9, a horizontal split force of Fy = 7204 kN is found. Using the
same input data and equation 2.8, we get Fy = 1791 kN , about one quarter of the force
found from equation 2.9.
The same goes for the case at hand. As expected, the force is slightly increased, giving
Fy = 7661 kN from equation 2.9 and Fy = 1889 kN from equation 2.8. Heggelund, using
the same input data as Økland, and referring to DNV HSLC rules, found the horizontal
split force to be Fy = 6304 kN . This value was also used by Økland in his analysis.
In Sec.3 B204, the still water transverse bending moment is given as

MS0 = 4.91∆ yb − 0.4B 0.88 (2.10)
This gives MS0 = 4800 kN m. By simple integration, [15] derived another formula for
determination of the still water transverse bending moment:
∆gB Bhull

MS0 = 1−2 (2.11)
8 B
This formula yields a still water transverse bending moment of MS0 = 6472 kN m. It was
however noted by Heggelund that equation 2.11 would yield conservative results as the
weight of the pontoons and vessel sides increases the weight towards the sides of the vessel
compared with the weight between the hulls.
Figure 2.1: Still water transverse bending moment
For consistency, the formulas given in [7] will be used. Equations 2.5 and 2.6 yield MS =
9599 kNm and MS = 13073 kNm, respectively. The value z, denoting the distance from
the baseline to the neutral axis for transverse bending, was set equal to 5.5 meters as in
[15].
The latter article reveals some of the changes that has been made to DNV HSLC. The
dynamic part of the transverse bending moment, i.e. the moment induced by the horizontal
split force is given as MT W = Fy (z − 0.75T ) as opposed to MT W = Fy (z − 0.5T ) in
equation 2.6. Consequently, the net horizontal split force has its centre of attack at 75%
of the draught for the old formula and at 50% for the new formula.
However, this increase in distance from neutral axis does not compensate for the reduction
of the horizontal split force. The dynamic transverse bending moment, MT W , is found
to have a value of 24552 kNm by Heggelund, while the current rules give a value of 8273
kNm.
Heggelund used VERES to determine the long term distribution of the transverse bending
moment. Using linear calculations and a uniform distribution of wave directions, he found
that for a probability of exceedence equal to 10−8 , the dynamic transverse bending moment
had a value of about 4500 kNm given a maximum allowable significant wave height of 2.5
meters. The maximum allowable significant wave height is calculated from a formula given
in section 2, B204, and has a value of 2.23 meters.
This indicates that conservatism is still preserved in the estimations of the design loads
despite a dramatic change in the rules.
2.2.5 Pitch connecting and twin hull torsional moment
The twin hull pitch connecting moment is given in section 3, B301 as
acg L∆
MP = (2.12)
8
Likewise, the twin hull torsional moment in section 3, B401, may be assumed to be
Strength requirements 15
acg b∆
Mt = (2.13)
4
Figure 2.2: Combined torsional and pitch connecting twin hull moments
The pitch connecting and torsion moment was found to be, respectively, MP = 38406 kNm
and Mt = 18207 kNm. Heggelunds VERES analysis found, as mentioned in section 2.2.4,
with a probability of exceedence of 10−8 , a torsional moment of about 4000 kNm given a
maximum allowable wave height of 2.5 meters. One has to account for nonlinearities and
other uncertaintis, but it is assumed that the calculated load is conservative in the present
case.
2.3 Strength requirements
The strenght requirements are taken as defined in DNV HSLC Part 3, Chapter 3 - Hull
structural design, aluminum alloy. These rules include requirements to manufacturing,
member dimensions, allowable stresses, etc. In terms of allowable stresses, these are to be
found using the design loads as defined in section 2.2. Those requirements relevant to the
present case are presented in the following sections.
2.3.1 Hull girder strength
The rules regarding hull girder strength, both longitudinal and transverse, are given in
section 4. A103 states that "For new designs (prototypes) of large and structurally compli-
cated craft (e.g. multi-hull types) a complete 3-dimensional global analysis of the transverse
strength, in combination with longitudinal stresses, is to be carried out".
The hull section modulus requirement is given in B101:
M
Z = (2.14)
σ
The shear strength of the hull girder is to be sufficient so that the shear stress does not
exceed, as given in C103
allowable bending stress

τ = √ (2.15)
3
It is to be evaluated for locations with doubtful shear area, which might be the case for
the stiffener-frame interaction for the floating frame design.
For twin hulls, the connecting structure between the hulls is to have adequate transverse
strength when subjected to transverse forces and moments as given in section 2.2. E201
defines the equivalent stress as
q
σe = σx2 + σy2 − σx σy + 3τxy
2 (2.16)
This is also known as the von Mises-stress. The allowable stresses, from E202, in the
connecting structure are to be taken as:
Bending stress, σ = 160f1
Shear stress, τ = 90f1
Bending stress, σe = 180f1
2.3.2 Plating and Stiffeners
Requirements to plating and stiffeners are defined in DNV HSLC part 3, chapter 3, section
5.
Maximum allowable bending stresses in plates and stiffeners, as in A301, are given in table
2.1.
Allowable bending stresses

Item Plate Stiffener
N/mm2
Bottom, slamming load 200 f1 180 f1
Bottom, sea load 180 f1 160 f1
Side 180 f1 160 f1
Deck 180 f1 160 f1
Flat cross structure, slamming load 200 f1 180 f1
Flat cross structure, sea load 180 f1 160 f1
Bulkhead, collision 180 f1 160 f1
Bulkhead, watertight 220 f1 200 f1
Table 2.1: Allowable bending stresses for plating and stiffeners
The thickness of the bottom plating is to be at least (B302):
√
22.4kr s Psl
t = √ (2.17)
σsl
By this formula, the thickness requirement is 5.74 millimeters for the traditional design.
For the floating frame design, one has to discuss whether the bottom plating should be
assumed to be a heat affected zone (HAZ). The bottom plating is to be extruded with
Strength requirements 17
longitudinal stiffeners and the extruded parts may be welded together so that the HAZ
are located in noncritical areas. However, the bottom plating may at some point require
repairs by welding. It is thus nonconservative to assume the bottom plating without HAZ.
The minimum thickness of the bottom plating for the floating frame is estimated to 6.41
millimeters. The design bottom plate thickness is 6 and 7 millimeters for the traditional
design and the floating frame design, respectively, thus exceeding the requirements.
From C201, the required section modulus of longitudinals supporting the bottom plating
m ∗ l2 sPsl
Z = (2.18)
σsl
This gives a minimum value of 27.7 cm3 for the traditional design and 34.6 cm3 for the
floating frame design. The traditonal design has bulb-100 profiles in its bottom structure
with a minimum section modulus of 30.1 cm3 . The floating frame has T110 profiles for its
bottom structure longitudinals with a minimum section modulus of 36.3 cm3 .
The required shear area of longitudinals, also given in C201
6.7(l − s)sPsl
AS = (2.19)
τsl
Equation 2.19 was evaluated by [12] and found unsuitable for the floating frame design.
He proposed the following formula
6.7lsPsl
AS = (2.20)
τsl
The minimum shear areas were found to be 3.34 cm2 and 5.46 cm2 for the traditional
design and floating frame design, respectively. The shear area is assumed to be equal to
the cross sectional area of the web of the profile. This area is 4.5 cm2 for the bulb-100
profile used in the traditional design and 5.0 cm2 for the T110 profile used in the floating
frame design. A design modification is thus necessary for the bottom stiffeners in the
floating frame design in order to meet the requirements. This can be done by choosing a
stiffener with a larger shear area or using another material in order to reduce the shear
area requirement.
2.3.3 Web frames and girders
The required strength of web frames and girders is given in DNV HSLC pt3. ch.3 Sec. 6.
The maximum allowable stresses is as given in A401:
Allowable stresses
Item Web frames and girders
Bending Shear stress Equivalent
stress (N/mm2 ) stress
(N/mm2 ) (N/mm2 )
Dynamic load 180 f1 90 f1 200 f1
Sea/static load 160 f1 90 f1 180 f1
Table 2.2: Allowable stresses for web frames and girders
B401 and B402 states that the required section modulus and effective web area for girders
subjected to lateral pressure is not to be less than, respectively:
2 b∗ p
mSG
Z = (2.21)
σ
10 (kst Sb∗ p − a∗ r∗ )
AW = (2.22)
τ
There is no distinction between the two designs in terms of requirements as the input data
for equation 2.21 and 2.22 are the same. These are Zmin = 365.62 cm3 and AW = 36.5
cm2 , respectively. The shear areas as of design are 18 cm2 and 21 cm2 for traditional and
floating frame design, respectively. The wide gap in requirement and actual value appears
inexplainable, but it is assumed that the shear area for both designs are sufficient as the
traditional design is DNV approved. The minimum section modulus’ for the frames are
438.77 cm3 for the traditional design and 561.10 cm3 for the floating frame design, both
larger than the minimum requirement.
2.4 Direct calculation methods
The direct calculation methods presented in this section are taken from DNV Classification
Notes No. 30.8, Strength analysis of hull structures in high speed and light craft. They
are compatible with the DNV HSLC rules [5].
CN 30.8 distinguishes between transverse web frame analyses and global strength analyses.
Web frame analyses deals with the structural response for a transverse web frame, normally
located midships in the middle of a compartment, suscepted to extreme local loads. Global
strength analyses addresses the global structural response when the vessel is subject to
extreme global wave loads and/or loading.
For web frame analysis, a 3-dimensional finite element analysis may be applied. [5] advises
that the model covers the length of one compartment in the midship are from base line to
upper deck, and extend from centre of one compartment to centre of the next compartment.
Furthermore, it suggests that plating, webs and flanges are modelled as separate elements
to ensure accuracy, and that 3 elements should be used over the height of the web of the
frame. Boundary conditions should be applied to best model the reality.
For global strength analysis, the finite element model should in general cover the complete
ship. CN 30.8 underlines the importance of including the geometrical hull shape, transverse
Direct calculation methods 19
bulkheads, decks, and any torsional boxes. In the case that simplified modelling is applied,
the simplifications has to be clearly identified and discussed.
Size and type of elements should be chosen to ensure that the effects of bending, shear and
torsion of the hull beam are accounted for. Using four noded elements, CN 30.8 suggests
using maximum three elements per frame spacing and three elements per tier. For FEM
in general, a length to breadth ratio exceeding 3 is not advised as ’long’ elements tend to
perform poorly.
CN 30.8 lists seven load conditions for the web frame analysis and seven load conditions for
the global strength analysis. Only three load conditions will be presented in the following
sections. They are local load condition 2 (LLC2) - symmetric bottom slamming, global
load condition 4 (GLC4) - transverse split force, and global load condition 5 (GLC5) -
torsion moment/pitch connecting momemt.
2.4.1 Symmetric bottom slamming
This load case may be decisive for the bottom structure, and investigates the effect of
slamming impact pressure on one frame. The slamming impact pressure is taken as found
in section 2.2.1. The design load area is to be taken as the frame spacing times the length
of the frame between the upper turn of bilge. It is also advised that, for a 3-dimensional
model, the slamming pressure is applied to one frame and sea pressure on the bottom
panels of the other frames.
Figure 2.3: Local load condition 2, symmetric bottom slamming [5]
For the web frame analysis, e.g. the symmetric bottom slamming analysis, the acceptable
stresses are taken as:
Design loads Plating Stiffeners and

girders
Dynamic loads (slamming)
- equivalent stress 220 f1 200 f1
- bending stress 200 f1 180 f1
Static loading (sea pressure)
- equivalent stress 200 f1 180 f1
- bending stress 180 f1 160 f1
Table 2.3: Allowable stresses for web frame analysis
Allowable shear stress: τ = 90f1
2.4.2 Transverse split force
The transverse split force load condition may be decisive for the structure between the
hulls, the side and the bulkheads. The load case is meant to represent the maximum
horizontal wave loads acting on the hull. This horizontal wave force is to be combined
with the transverse still water bending moment. This load condition is in fact two different
load cases, as one should consider both positive and negative split forces in combination
with the still water transverse bending moment.
The transverse split bending moment in CN 30.8 is defined as
3
Msplit = Fy · y + Ms,keel (2.23)
4
1
Ms,keel = [(x + y) − x̃] (2.24)
4
(a) Split load condition
(b) Geometrical definitions
Figure 2.4: Global load condition 4, transverse split force, split outwards shown
Direct calculation methods 21
It is noted that the sum of the horizontal forces is to act at 75% of the draught. This is
inconsistent with the rules, where Msplit = MT W = Fy (z − 0.5T ), as discussed in section
2.2.4. The is probably due to that CN 30.8 was published in August 1996, and is not been
updated according to the current rules.
2.4.3 Torsion moment and pitch connecting moment
This load case combines the twin hull torsional and pitch connecting moment. It may be
decisive for the cross structure. For the floating frame design, this load condition may
cause large deflections of the frame relative the outer skin. This may induce high stresses
in the longitudinal stiffeners.
As the figure shows, CN 30.8 suggests applying the moments as line loads in the deck of
the model.
Figure 2.5: Possible modelling of global load condition 5, torsion moment/pitch connecting
moment, [5]
Chapter 3
Finite Element Modelling
3.1 General
The finite element method is the common approach for structural analysis in general, and
especially for marine structures as they are often complex structures and subject to a
wide range of loads. Most structural problems are to complex to be solved by classical
analytical methods, thus it is convenient to apply the finite element method to solve these
problems. However, the results of a FEA are seldomly exact, but with experience and a
correct interpretation of the results, the results will be accurate enough for engineering
purposes.
The software used for the modelling was ABAQUS CAE, versions 6.9.2 and 6.10.2. The
two versions are essentially the same in terms of functionality. ABAQUS standard has
been used for processing the analyses, and ABAQUS viewer used for the post-processing.
The elements used are 4-node and 3-node thin shell elements and 2-node beam elements.
As modelling is a time consuming task, the same meshes were used for all analyses. This is
presented in more detail in sections 3.3 - 3.4. The models were built up by three different
main parts:
• The outer plating and car deck, longitudinally stiffened, and the longitudinal bulk-
head along the centerline. This part was extruded in the models length.
• The transverse web frames
• The transverse water tight bulkheads
23
24 Finite Element Modelling
(a) Floating frame (b) Traditional frame
Figure 3.1: Web frames used in models
(a) Bulkhead, floating frame model (b) Bulkhead, traditional model
Figure 3.2: Bulkheads used in models
As shown, no structural elements above the car deck were modelled. The upper decks
and sides were neglected because it was assumed that the potentially critical structural
response was to be found in the lower hull structure. However, the upper structure in-
deed contributes to some structural strength of the hull girder, thus implying that this
simplification contributed to conservativeness in the analyses.
Some structural elements were neglected or simplified in cases where it was assumed that
the simplification itself were of little importance for the present cases. Vertical stiffeners of
the web frames and bulkheads were neglected as their main function is to prevent buckling.
The same goes for the bracket flanges. These are however structural elements that would
increase the structural strength, so the neglection of these element also contribute to
conservativeness.
For all load cases, the same simplifications, loads, boundary conditions, etc., were practi-
cally equal for the floating frame model and the traditional hull girder model.
Torsion/pitch connecting model 25
Figure 3.3: Extent of models, part highlighted in red used for the web frame analyses,
part highlighted in green used for global analyses.
3.2 Torsion/pitch connecting model
The combined torsion/pitch connecting load condition is to represent the worst possible
wave load in terms of torsion, meaning that the vessel is a moving through a wave that has
a direction and wave period that gives the maximum torsional moment. In [14], Heggelund
found that, given a forward speed of 35 knots, the extreme torsional moment occured for
a wave direction of 66 degress (head sea = 0 degrees) and a wave period of 4.4 seconds.
The models used for the combined twin hull torsional and pitch connecting moment are
simplified models of the complete vessels. The model length is 45 meters, stretching from
frame #5 to #50 with a frame spacing of 1 meter, as specified in the drawings. The
length over all of the vessel is 60 meters. In other words, the stern and bow of the vessel
were not included in the model. This simplification was done to reduce the amount of
work associated with the modelling. It was assumed that both weight and buoyancy
contributions of the stern and bow was small, so the simplification should not represent a
significant inaccuracy.
(a) Complete model (b) Transverse members
Figure 3.4: Outlines of models used for torsion/pitch connecting analyses
Another simplification is the neglection of reduced cross sectional area towards the ends
of the model. The transverse web frame at frame #27 and transverse bulkhead at frame
#26 located in the midship section of the vessel has been used to represent all web frames
and bulkheads over the length of the model. Consequently, the cross section of the outer
plating and car deck is constant over the length of the model. This implies a neglection of
reinforcing structural elements and increased outer plating at the fore and aft sections of
the vessel. It is however assumed that this modelling represents the hull girder strength
in a satifactory fashion, and that the simplification itself is conservative.
As shown in section 2.4.3, it is suggested by DNV to apply the torsion and pitch connecting
moment as constant line loads in the deck over the length of the model. In [14], Heggelund
applied a linearly varying load between − L/4 and L/4 and constant otherwise. A better
way to represent the actual load condition was assumed to be by applying the load to the
immersed part of the hull. It also represented an oppurtunity to evaluate the stiffener-web
frame interaction when the stiffeners were subjected to a lateral pressure.
For simplicity, the load was applied to the bottom structure, i.e. between the upper turns
of the bilge, as an evenly distributed pressure. The net forces would thus act in the middle
of the quarter-models. This is however inconsistent with DNVs proposal, where the net
forces were to act in the middle of the quarter-vessels. By applying an evenly distributed
pressure of 0.0192 N/mm2 over the bottom structure and half length of the model, a pitch
connecting moment of 38406.2 kNm was achieved, as estimated in section 2.2.5. However,
this load yielded a twin hull torsional moment of 21848.8 kNm, 20% higher than the
design load estimated by the rules. As discussed in section 2.2.5, the torsional moment is
probably very conservative in the present case.
(a) Positive pressure (b) Negative pressure
Figure 3.5: Idealised pressure distribution used in analyses
The sum of forces and moments acting on the models are zero, but boundary conditions
were still needed to prevent rigid body movement that could be caused by minor defiencies
in the modelling. The longitudinal axis is z, the transverse z, and the vertical y. The
origin is located at the baseline, amidships at the centerline. By symmetry of forces,
the model should be in balance at the centerline, midships, i.e. at x=0, z=0. However,
fixing all degrees of freedom (DOFs) for all nodes at x=z=0 introduced numerical errors
in the analyses. Two nodes at x=z=0 had all translational DOFs fixed, which surpressed
potential rigid body translation and rigid body rotation about the z-, and y-axis. Springs
counteracting longitudinal displacement were applied at the ship side. Their function was
to prevent rigid body rotation about the vertical neutral axis.
Transverse bending model 27
(a) Boundary conditions at longitudinal bulkhead (b) Springs at ship sides
Figure 3.6: Boundary conditions, torsion/pitch connecting moment
3.3 Transverse bending model
The models used for the transverse bending analysis are essentially the same models as used
for the torsion/pitch connecting moment analysis, as explained in section 3.2. However,
due to symmetry, only half the vessel was evaluated. The implications of the modelling
simplifications are assumed to contribute to conservativeness and the models suitable to
determine the structural response of the cross structure between the hulls and otherwise.
As prescribed by DNV, the split forces representing the horizontal wave forces are com-
bined with the still water transverse bending moment. The still water transverse bending
moment is applied as an evenly distributed pressure on the car deck, representing the
weight of the vessel, and an evenly distributed pressure on the bottom outer plating, rep-
resenting the vessels buoyancy. The values of these pressures are 5.7 kN/m2 at the deck
and 12.7 kN/m2 at the bottom bilge, yielding a still water transverse bending moment of
4800 kNm about the neutral axis at y = 5.5 m, as estimated in section 2.2.4.
The splitforces were not applied as proposed in [5]. The reason for this, as discussed in
section 2.2.4, was that there has been a rule change after the last version of DNV CN 30.8
was published. It was interpreted from [6] that the sum of horizontal forces were to act at
50% of the draught, not at 75% of the draught as suggested in DNV CN 30.8.
The split force was separated into three sets of concentrated forces. Two sets, being one
quarter of the total split force each, were applied at both sides of the hull at the waterline.
The last set, one half of the total split force, was applied at the keel point. Distributing
the keel load and waterline loads over each node over the length of the models gave
nodal concentrated forces of Fsplit,keel = 2616.3 N/node and Fsplit,W L = 1308.2 N/node.
Positive and negative split forces were combined with the still water transverse bending
moment in two separate analyses.
Symmetry was provided by fixing all nodes along the centerline (x=0) for all DOFs. This
also prevented rigid body movement.
(a) 2-dimensional (b) 3-dimensional
Figure 3.7: Loads and boundary conditions, transverse bending moment. Inward split
force shown
3.4 Slamming model
The models used for the slamming analysis were similar to those used by Englund and
Ervik for their slamming analyses. One half of the midship compartment from frame #26
to #35 was modelled. The models are consisted by the same three main parts as presented
in section 3.1. As Ervik and Englunds analyses revealed structural stresses exceeding the
acceptable levels as defined by DNV, it was desireable to perform the same analysis with
the strengthened web frame compared to previous scantlings.
It is noted that both Englund and Ervik performed their slamming analysis with a more
refined mesh than what is the case for this analysis. This should be taken into account
when comparing the results, as a refined mesh normally indicates higher local stresses
than a coarse mesh.
The meshes are in good compliance with the DNV CN 30.8 recommendations. Three
elements over the web height of the frame should be used, the floating frame model has
two and the traditional model has three. It is assumed that the meshes represents the
true web frame structure in a satisfactory fashion.
The slamming load was applied to bottom plating between the the upper turns of the
bilges in an area spaning longitudinally over a frame spacing. For the rest of the bottom
structure, sea pressure was applied. The car deck pressure was applied as an evenly
distributed pressure over the area of the car deck.
(a) 2D (b) 3D, slamming load highlighted
Figure 3.8: Pressure loads in slamming analyses

Slamming model 29
The force from the passenger deck for this particular half compartment is estimated to
have a value of approximately 360 kN. It is assumed that about two thirds of this force
is transferred through the ship side and the transverse vertical girder between the car
deck and the passenger deck. The remaining third is carried by the centre pillars. For
simplicity, this load is applied to the models at every web frame in the deck as distributed
pressures over small areas, as shown in figure 3.9.
Figure 3.9: Modelling of passenger deck loads
The transverse bulkheads at the ends were not included in the model as the ends of the
compartment had all DOFs surpressed. Also, a symmetry boundary condition was applied
at the vertical plane along the centerline.
Figure 3.10: Boundary conditions in slamming analyses

Chapter 4
Results from finite element

analyses
4.1 Interpretation of results
A finite element structural analysis is of limited value unless a correct interpretation and
discussion of the results is undertaken. For complex structural problems, like the present
cases, this is often a challenging task. Several sources of inaccuracy has to be considered,
for example:
• Simplified geometry,
• simplified load modelling,
• poor element/mesh performance,
• numerical deficiencies,
• inaccurate boundary conditions
As discussed in chapter 3, the modelling is conservative, not only in a global sense, but also
locally. For example, curved plates in the real structure are obviously modelled as a series
of straight shell elements. As a consequence, the model surface is not as smooth as the real
surface. This may affect the flow of stresses, possibly introducing stress concentrations
that exceed the stresses in the real structure.
Stress concentrations may also be found where different structural members interact, e.g.
longitudinal and transverse members. In the real structure, brackets, cutouts, flanges,
etc., are applied to avoid high stress concentrations. Structural details like the above
mentioned are not included in the models, and stress concentrations at certain locations
should be expected and subsequently evaluated.
Deficiencies in the mesh may also introduce higher stresses for the models than what is
the case for the real strucutre. Elements with poor element shapes or elements that are
small compared to their surrounding elements, has a tendency of overestimating the stress
level.
31
32 Results from finite element analyses
4.2 Torsion/pitch connecting analysis
The analyses with the complete model gave the results given in table 4.2. The correspond-
ing allowable stresses are given in table 4.1.
Allowable stresses, global analyses, MPa

Plating and Girders
Stiffeners
Equivalent stress 86.4 108.0
Bending stress 76.8 96.0
Shear stress 43.2 54.0
Table 4.1: Allowable stresses, global analyses
Maximum stresses from torsion analyses, MPa

Floating frame Traditional
Plating and Girders Plating and Girders
stiffeners stiffeners
Equivalent 94.4 91.2 92.4 85.2
stress, Seq
Bending 87.9 69.9 92.4 72.5
stress, |S11 |
Bending 39.8 77.0 36.6 78.4
stress, |S22 |
Shear stress, 39.3 43.8 45.3 36.1
|S12 |
Ratio, maximum vs. allowable stresses
Equivalent 109.3% 84.4% 106.9% 78.9%
stress
Bending 114.5% 72.8% 120.3% 75.5%
stress 1
Bending 51.8% 80.2% 47.7% 81.7%
stress 2
Shear 91.0% 81.1% 104.9% 66.9%
stresses
Table 4.2: Stresses from torsion analyses
The maximum stresses for the floating frame model are found to exceed the allowable.
They are found at the longitudinal stiffeners flanges in the twin hull cross structure towards
the ends of the model, at the intersection with frame #49.
Torsion/pitch connecting analysis 33
Figure 4.1: Bending stress at frame #49
Figure 4.1 shows the location of the maximum bending stress for the floating frame model.
The stiffener flanges are modelled as beam elements, and ABAQUS viewer does not sup-
port rendered beam profiles for the contour plots. The location is thus manually high-
lighted with a red dot. The stress is concentrated in the flange, and high stresses are not
found in the surrounding structure.
The equivalent stresses at the same location are shown in figure 4.2. Again, the maximum
stress, as given in table 4.2, is located in the same stiffener flange, marked with a red dot.
Stress concentrations are seen in the same region for the stiffener web and the plating.
These stresses are however lower than the allowable stress.
Figure 4.2: Equivalent stress at frame #49
By not considering the longitudinal stiffeners flanges, the maximum stresses in the floating
frame model are within the allowable stresses. These stresses are found towards the ends
of the model, at frame #5-6 and #49-50. The stresses closest to the allowable stresses are
found in the region as shown in figure 4.2. For the plating, equivalent and bending stress
of about 90-95% of the allowable is found there. Also, the shear stress in the longitudinal
stiffener web shown in figure 4.1 is 90% of the allowable.
The results of the traditional model are very similar to that of the floating frame model.
The maximum equivalent and bending stress is found in a longitudinal stiffener flange in
the twin hull cross structure at frame #15, at the intersection with a transverse bulkhead.
As for the the floating frame model, the stress is local. Disregarding the longitudinal
stiffener flanges, maximum equivalent and bending stresses for the traditional frame model
are found to have values of about 86MPa and 80 MPa, respectively.
The models display slightly different torsional behaviour. Figure 4.3 shows a contour plot
of the displacement magnitude for frame #50. It is seen that the displacements are slightly
larger for the floating frame model, and that the torsional neutral axis is located higher
for the traditional design than for the floating frame design.
(a) Floating frame (b) Traditional design
Figure 4.3: Comparison of torsional neutral axis
A simplified geometrical analysis of the torsional stiffness is summarized in table 4.3.

Nodal displacements at the corner of the deck has been used for frame #50. Although
the analysis being simple, it still serves as an indication of the torsional stiffness of the
models. The displacements at the deck corner are larger for the floating frame model, and
it is seen that this is partly due to a larger vertical distance between the torsional neutral
axis and the deck for the floating frame model compared to the traditional model. The
simple geometrical analysis yields torsional angles of 0.401 and 0.387 degrees for the the
floating frame model and the traditional model, respectively.
Transverse bending analysis 35
Torsional deformation at deck corner, frame #50

Floating frame Traditional design Unit
Neutral axis height 4425.83 4726.24 mm
Height at deck corner 5850.00 5850.00 mm
Vertical distance above NA 1424.17 1123.76 mm
Horizontal distance from NA 8250.00 8250.00 mm
Undeformed angle, α0 0.1709 0.1354 rad
9.7943 7.7567 deg
Vertical disp. at corner 57.53 55.84 mm
Horizontal disp. at corner -11.01 -6.95 mm
Deformed angle, α1 0.1779 0.1421 rad
10.1951 8.1439 deg
Torsional angle, φ = α1 − α0 7.00 · 10−3 6.75 · 10−3 rad
0.4008 0.3872 deg
Ratio, torsional rotation 103.53% 100.00%
Table 4.3: Torsional stiffness
The combined pitch connecting and torsional moments induce warping deformations in
the cross sectional planes. This effect is at its most predominant at the midsections of the
vessel. Figure 4.4 shows combined undeformed/deformed contour plots of the longitudinal
deformations at frame #27 for both models. The deformations are scaled 100:1.
Figure 4.4: Comparison of warping deformation at frame #27
The longitudinal warping deformations are found to have maximum values of about 12
and 13 millimeters for the traditional design and floating frame, respectively. Relatively
large shear stresses are found in the plating in the same regions, 39.3 MPa for the floating
frame model and 34.9 MPa for the traditional model.
4.3 Transverse bending analysis
The transverse bending load condition is considered as two separate load conditions. Both
inward and outward split force is combined with the transverse still water bending moment.
When considering the port half the vessel, the still water transverse bending moment is
acting clockwise (looking from stern to bow) about the longitudinal neutral axis located
at the centerline in the twin hull cross structure.
The inward splitforce gives a transverse bending moment acting counterclockwise abouth
the neutral axis, thus counteracting the still water bending moment. This load condition
is labeled transverse bending analysis 1, and the results from the analyses are listed in
table 4.4.
Maximum stresses from tranverse bending analyses 1

Equivalent 49.1 41.7 53.6 44.9
stress, Seq
Bending 47.1 36.9 52.7 41.3
stress, |S11 |
Bending 18.8 28.8 19.0 33.6
stress, |S22 |
Shear stress, 25.3 20.3 29.2 21.1
|S12 |
Equivalent 56.8% 38.6% 62.0% 41.6%
stress
Bending 61.3% 38.4% 68.6% 43.0%
stress 1
Bending 24.5% 30.0% 24.7% 35.0%
stress 2
Shear 58.6% 37.6% 67.6% 39.1%
stresses
Table 4.4: Allowable stresses and stresses from analyses
As expected, tranverse bending analysis 1 does not yield high stresses. The maximum
stresses found are generally only 50-70% of the allowable for both analyses, and they are
found to be slightly higher for the traditional model than for the floating frame model.
For transverse bending analysis 2, the horizontal split force is acting outwards, thus giving
a clockwise transverse bending moment about the neutral axis. The maximum stresses
are listed in table 4.5.
Maximum stresses from tranverse bending analyses 2

Equivalent 132.6 115.6 137.0 108.0
stress, Seq
Bending 132.6 115.0 137.0 102.8
stress, |S11 |
Bending 57.4 85.8 50.8 100.3
stress, |S22 |
Shear stress, 73.0 41.0 62.6 38.1
|S12 |
Equivalent 153.5% 107.0% 158.6% 100.0%
stress
Bending 172.7% 119.8% 178.4% 107.1%
stress 1
Bending 74.7% 89.4% 66.1% 104.5%
stress 2
Shear 169.0% 75.9% 144.9% 70.6%
stresses
Table 4.5: Maximum stresses from analyses
Large stresses are found in both models. For the traditional model, all maximum stresses
are found to exceed the allowable, implying that the models overpredict local stress con-
centrations, and that at least some of these may be neglected. The maximum bending,
bending and shear stresses occur at region 1 at frame #35, as shown in figure 4.5. Extreme
values are located in the longitudinal stiffener flange. Due to the simplified modelling of
the stiffener flanges and their interactions with transverse bulkheads and transverse web
frames, the stresses at these are expected to be highly overpredicted.
Figure 4.5: Equivalent stress at frame #35, region 1 highlighted
In figure 4.5, the stress range is such that dark red corresponds to 108 MPa, the maximum
allowable equivalent stress for girders. It is seen that the stress level for the floating frame
model is somewhat higher at region 1. For this load condition, the transverse bulkheads
are carrying a significant part of the load as they are far stiffer than the transverse web
frames. A higher stress level at the bulkheads for the floating frame model than for the
traditional model bulkheads may indicate that the traditional frames are carrying more
of the load than the floating frames.
4.3.1 Transverse bulkheads
For the traditional model, the maximum equivalent and bending stresses found in region
1 for frames #5, #15, #26, #35 and #50 are in the vicinity of the allowable stress ranges.
These stresses are however assumed to be somewhat overpredicted as the actual structure
has a smoother turn than what is seen in the model. Also, the bottom elements in region
1 are quite small compared to the surrounding elements, thus making a overprediction of
the stress probable.
The stress levels for the floating frame model are higher, and stress concentrations ex-
ceeding the allowable stress level are found at all bulkheads at region 1. The stresses are
dominated by transverse bending stress, and the shear stress at region 1 is close to zero.
(a) Equivalent stress (b) Transverse bending stress
Figure 4.6: Stress at frame #35, floating frame model
For figures 4.6a and 4.6b, dark red colour corresponds to the maximum allowable stresses,
108 MPa and 96 MPa, respectively. Areas in light gray have stresses exceeding the al-
lowable. These areas are expanded in figure 4.7. The maximum equivalent stress is 110.7
MPa, and the maximum bending stress is 115.0 MPa, 20% above the allowable bending
stress.
(a) Equivalent stress (b) Transverse bending stress
Figure 4.7: Extent of area with higher stress than allowable
The exceedance of maximum allowable bending stress extends over several elements, and
may thus be critical for the structure. However, the modelling of the bulkheads is conser-
vative, so an overprediction of the stresses is highly probable. If the bulkheads were to be
strengthened, an increase of plate thickness in the areas around region 1 would probably
be sufficient to reduce the bending stresses to an acceptable level.
4.3.2 Transverse web frames
In figure 4.8 dark red correspond to maximum allowable stress, while dark red and dark
blue correspond to maximum allowable stress in figure 4.9. Higher stresses are found in the
traditional frame than for the floating frame. This may indicate that the traditional frames
carries more load than the floating frames for the present case. The maximum stresses
found for the floating frame are about 80-90% of the allowable. For the traditional frame,
a maximum equivalent stress of 107.9 MPa is found in region 2. This stress is dominated
by the vertical bending stress at the same location, with a value of 100.3 MPa, 4.5%
higher than the allowable stress. This stress concentration is however confined within one
element.
The stress concentrations seen in region 2 are assumed to be overpredicted. Firstly, the
models are simplified in the sense that rounded corners for the vessel are sharp corners in
the model. Secondly, the relative difference in size of neighbouring elements is large.
Figure 4.8: Equivalent stress at highest loaded transverse web frame, region 2 highlighted
Figure 4.9: Bending stress at highest loaded transverse web frame
4.3.3 Plating and stiffeners
Very high, local bending stresses were found in the longitudinal stiffener flanges. However,
they are neglected in this section as the stress prediction for them is assumed to be
exaggerated for the present case due to the modelling of the stiffener flanges. High stresses
are also found in the plating and the stiffener webs.
For the plating, transverse bending stresses exceeding or close to the allowable level are
found at region 1 for both models. Theses stress concentrations are located at frames #5,
#15, #26, #35, and #50, i.e. at all transverse bulkheads. As for the bulkheads, stress
levels for the floating frame model are somewhat higher than for the traditional model.
The maximum equivalent stress is 106.5 MPa at frame #35, about 14% higher than for
the traditional model and 23% higher than the allowable equivalent stress at that location.
The bulk of the equivalent stress is transverse bending stress, having a maximum value of
103.1 MPa at the same location. This is 25% higher than for the traditional model and
34% higher than the allowable bending stress.
Since also the traditional model also has equivalent stresses and bending stresses exceeding
the allowable stress levels, it is reasonable to assume that the analyses has overpredicted
the stresses in the plating, and that the actual plating stresses for the traditional model
are within the allowable stress levels.
The increase of stresses for the floating frame model is primarily due to reduced stiffness
in transverse bending compared to the traditional model. This yields larger global and
local deformations. The largest local deformations are obviously found in region 1, at the
transverse bulkheads. These local deformations are shown in figure 4.10, scaled 100:1.
The contour plot displays rotational deformations.
Figure 4.10: Local deformation at region 1, frame #35 floating frame model
A comparison of the plating equivalent and bending stresses at frame #35 for the two
model are shown in figures 4.11 and 4.12. Areas where the stresses exceed the allowable
stress level have light gray colour.
Figure 4.11: Comparison of plating equivalent stress
It is seen that the area where the maximum allowable stress level is exceeded is quite
large for the floating frame model. However, figure 4.12 reveals a rather untidy bending
stress distribution for the floating frame model. This may indicate that there are errors
in the modelling or in the finite element analysis itself, thus introducing inaccuracies in
the results.
Figure 4.12: Comparison of plating bending stress
In order to reduce the stresses in the plating structure for the floating frame model, the
best solution is probably to increase the plating thickness at the critical areas. This is
assumed to be sufficient as the stresses are dominated by transverse bending stress, and
that other stress components are relatively small.
Stresses in the longitudinal stiffener webs are found to have values exceeding the allowable
stress levels by wide margins. They occur at region 1, and are shear dominated. Due to
the simplicity in the modelling of the stiffener-bulkhead interaction, stresses are exagger-
ated in the analyses. As for the bulkhead and the plating, the stresses from the floating
frame analysis exceed those from the traditional design analysis. This is has two main
explanations. Firstly, the local deformations, as shown in figure 4.10, are larger for the
floating frame model. Secondly, the shear area of the T55-profiles used in the floating
frame model at region 1 is 27% lower than for the bulb-65 profiles used in the traditional
model.
The equivalent stress for the stiffener flange at region 1 has a maximum value of 128.6
MPa for the floating frame model and 115.2 MPa for the traditional model. These values
are 49% and 33% higher than the allowable equivalent stress. Once again, this indicates
that the stresses are highly overpredicted as the traditional design is DNV approved.
Figure 4.13: Comparison of stiffener web equivalent stress, frame #35
In figure 4.13 and 4.14, the light gray areas indicate equivalent and shear stresses exceeding
the allowable stress levels. The stiffener webs has maximum shear stress values of 73.0
MPa and 62.6 MPa for the floating frame model and the traditional model, respectively.
They exceed the allowable shear stress by 69% and 45%.
Figure 4.14: Comparison of stiffener web shear stress, frame #35
Although the modelling has been performed in a conservatice manner, it is probable that
the longitudinal stiffeners for the floating frame design has stresses that are too high. To
reduce the shear stress, and hence the equivalent stress, the stiffeners shear area needs to
be increased. This implies that the height or the thickness of the stiffener webs needs to be
increased, possibly both height and thickness. It should also be noted that a bulb-profile
has better shear properties than a T-profile. A bulb-profile may however prove unsuitable
for the floating frame design as the transverse web frame bottom flange is to be welded to
the longitudinal stiffener flange.
4.4 Slamming analysis
The maximum stresses found in the slamming analysis are shown in table 4.7. The corre-
sponding allowable stresses are shown in table 4.6.
Allowable stresses, web frame analysis, MPa

Plating Stiffeners Girders
Equivalent stress 105.6 96.0 120.0
Bending stress 96.9 86.4 108.0
Shear stress 43.2 43.2 54.0
Table 4.6: Allowable stresses, web frame analysis

Slamming analysis 43
Maximum stresses from slamming analysis, MPa

Floating frame Traditional design
Plating Stiffeners Girders Plating Stiffeners Girders
Equivalent 39.3 65.6 57.2 40.5 67.8 78.0
stress, Seq
Bending 36.9 65.6 57.2 41.2 67.8 72.8
stress, |S11 |
Bending 36.9 18.6 52.9 36.5 25.0 86.5
stress, |S22 |
Shear stress, 22.4 16.7 22.5 23.0 24.7 24.4
|S12 |
Equivalent 37.2% 68.3% 47.7% 38.4% 70.6% 65.0%
stress
Bending 38.4% 75.9% 53.0% 42.9% 78.5% 67.4%
stress 1
Bending 38.4% 21.5% 49.0% 38.0% 28.9% 80.1%
stress 2
Shear 51.9% 38.7% 41.7% 53.2% 57.2% 45.2%
stresses
Table 4.7: Maximum stresses from slamming analyses
It can be seen that all stresses found from the analyses, for floating frame and traditional
design, are well within the strength requirements. The general tendency is that the stresses
for the traditional design is slightly higher. By comparison, [18] found maximum stresses
of about 90% of the allowable stresses in the bottom region of the web frame for the
traditional frame design. In his model, a frame spacing of 1.2 meters was used, whereas
it is 1.0 meters for the present case, so a reduction of stresses in the present case was
expected. 4.15 shows the deflection of the frames suscepted to slamming pressure. The
deformations are scaled 100:1.
Figure 4.15: Comparison of web frame stiffness
It can be seen that the deformation is almost equal for both models, with the floating
frame being slightly stiffer than the tranditional frame. This is however to be expected, as
the floating frame has a section modulus which is 5.6% higher than that for the traditional
frame. The maximum deflections are found to be 6.07 mm and 6.59 mm for the floating
frame and the traditional frame, respectively. The deflection is thus reduced by 7.9% for
the floating frame compared to the traditional frame. This indicates that plating and
longitudinal stiffeners in the bottom structure of the floating frame model does contribute
to some extent to structural strength of the web frane when suscepted to lateral pressure.
Figure 4.16: Vertical deformation at keel, bottom plating
Figure 4.16 confirms that the floating frame model is stiffer, and that the deformations
are very similar, as expected. It should also be noted that the issue of out of plane
bending of the longitudinal stiffeners, as discussed in [12] and [11], appears to be resolved
by strengthening of the transverse web frames. The highest stresses for the longitudinal
stiffeners are found at the connection with the transverse bulkheads with a maximum
value as given in table 4.7.
Chapter 5
Fatigue assessment
5.1 Introduction
The DNV HSLC rules provides no specific requirements related to fatigue, thus implying
that the fatigue strength may be sufficient when the design loads and strength requirements
have been fulfilled. However, DNV HSLC part 3, chapter 9, section 6 states that a number
of areas, considered to be critical, are to be considered in a fatigue strength analysis. In
general, areas where the dynamic stress level is considered as high, the fatigue strength is
to be considered [9].
Furthermore, the fatigue analysis is to be based on a period of time equal to the planned
life and usage profile of the craft. The minimum period is 20 years. Cumulative damage
due to the stress history may be expressed by linear cumulative damage usage factor. For
areas where a potential local failure can be detected before it leads to a critical event, the
damage usage factor is not to exceed the value h = 1.0. The design criteria specified is
to be fulfilled based on S-N data for mean value minus 2 times the standard deviation.
Fatigue is due to cyclic stress variations. There are several sources to such load effects, but
those of most interest for the present case are the linear global wave loads and the local
pressure variations due to waves. The local wave loads are due to the local sea pressure
on the outer plating, and the global stress is due to the behaviour of the ship hull as a
beam [13].
In accordance with the DNV HSLC rules, the fatigue criterion, using the Miner-Palmgren
approach, can be expressed as:
X ni
D ≈ ; Ni · Sim = ā (5.1)
i
Ni
ni : number of cycles in stress range block i

Ni : number of cycles to failure for stress range block i
Si : applied stress range in block i
ā, m : geometrical/material parameters
As prescibed in DNV HSLC, the design SN-curve is normally taken as the mean minus
two standard deviations curve for relevant experimental fatigue data. This implies that
test values exceed the design SN-curve with a 97.7% probability. In general, the fatigue
45
46 Fatigue assessment
strength is significantly reduced for increased joint complexity. It is also known that the
fatigue performance of welded aluminum structures is reduced compared to unaffected
material. Experimental tests have shown that 6082-T6 aluminum welds have a tensile
strength of approximately 160-170 MPa compared to circa 270 MPa for the same alloy in
an unaffected state [20].
For a JC60 with the floating frame design, as presented in the previous chapters, several
areas may be critical in terms of fatigue. In the following sections, only one area will
be evaluated. This is the longitudinal stiffener/transverse web frame connection in the
bottom structure of the vessel. From chapter 4, relatively high stresses were found at this
location, especially for the slamming analysis. The flange-to-flange connection is a rather
complex joint, so high stress ranges may prove critical.
Figure 5.1: Critical areas for fatigue consideration of high speed catamarans, [19].
(a) Bottom structure, floating frame. Critical detail (b) Critical detail
highlighted
Figure 5.2: Structural detail to be evaluated for fatigue strength

Introduction 47
Due to the complexity of the structural detail to be examined, it is not immediately obvious
where a fatigue crack will have its origin. Three different potential crack initiation sites
were identified, and will be evaluated in the following sections. They are labeled HS A,
HS B, and HS C, as shown in figure 5.3.
Figure 5.3: Potential crack initiation sites
Eurocode 9 includes classifications of relevant structural details and corresponding SN-

curves. In Eurocode 9, HS A, HS B, and HS C corresponds to detail type 3.2, 3.4, and
3.5, respectively.
(a) HS A (b) HS B (c) HS C
Figure 5.4: Relevant constructional details, Eurocode 9
To determine the fatigue lifetime of a given structural detail, [19] proposed two different
methods named the spectral method and the alternative method. The spectral method
is based on analytical and computer analyses, while the alternative method is based on
experimental test. The flow chart of the methods is presented in figure 5.5.
Figure 5.5: Flow chart for fatige analysis, [19].
In the following sections, a simplified version of the spectral method will be presented.
5.2 Development of the loading history
The spectral method requires both hydrodynamic and structural response data to establish
the loading and stress transfer functions. The hydrodynamic response of high-speed craft
will be provided through a transfer function H as a function of heading θ and wave
frequency ω. It defines the loading experienced by the vessel in response to waves. Typical
transfer functions include:
Hv (ω, θ): Transfer function for vertical bending moment
Hh (ω, θ): Transfer function for horizontal bending moment
Ht (ω, θ): Transfer function for torsional bending moment
Hp (ω, θ): Transfer function for external pressure
A structural model will utilise the loading information from the hydrodynamic response
to develop the corresponding stress coefficients. This can be achieved by developing a
finite element model of the ship and subject it to the loading functions of interest. If
necessary, local models can be created to determine critical stress coefficients. Stress
transfer functions are developed on the basis of the results from the structural analyses.
They express the values of a certain stress component resulting from a unit load. The
stress coefficient are defined in association with the loading transfer functions:
Development of the loading history 49
Av : Stress per unit vertical bending moment

Ah : Stress per unit horizontal bending moment
At : Stress per unit torsional bending moment
Ap : Stress per unit external pressure
The stress history of a detail is normally a function of numerous load components, and
they can be linearly summed to develop a combined stress transfer function:
Hσ (ω, θ) = Av Hv (ω, θ) + Ah Hh (ω, θ) + At Ht (ω, θ) + Ap Hp (ω, θ) + . . . (5.2)
Equation 5.2 is used to develop the transfer function of one stress component acting at
one area of interest, meaning that a series of combined stress transfer functions is required
for a full fatigue analysis.
The short term response is developed by relating the stress transfer function to actual load
magnitudes. This is achieved by including the spectral information of the waves to account
for the load and stress variations over time. Several wave scatter diagrams and sea sprectra
are available. They are to be chosen to correspond to the service route of the vessel. A
sea spectrum provides a representation of the waves at a specific, stationary location over
time. The sea sprectra are defined by two or more parameters. They are adjusted to
best represent a sea state. A commonly used sea sprectrum is the two-parameter Pierson-
Moskowitz sea sprectra, defined as:
2π 1 2π
4 4 !
H2

−5 −4
Sη (ω | Hs , Tz ) = s ω exp − ω (5.3)
4π Tz π Tz
Sη is the sea state, ω the wave frequency, Hs the significant wave height, and Tz the zero
crossing period. The probability of occurence for a given sea state is defined in a scatter
diagram. The scatter diagram provides a probability pij for each significant wave height
given a zero crossing period and vice versa.
As the wave sprectra refer to a stationary point in the ocean, the given sea spectrum
needs to be modified to account for the forward speed of the vessel. The wave encounter
frequency, meaning the wave frequency experienced by the vessel, is defined as:
ωV

ωe = ω 1 − cos θ (5.4)
g
V and g are the vessel forward speed and gravitational acceleration, respectively. Accord-
ingly, the sea spectrum is modified:
1
Sη (ωe ) = Sη (ω) (5.5)
1− 2ωV
g cos θ
For the complete response spectrum, it is normal to account for ’short crestedness’ of the
seas. This can be achieved through the application of a cosine squared spreading function:
2
Sη (ωe , θ) = Sη (ωe ) · cos2 θ (5.6)
π
The response spectrum, for a given combination of Hs , Tz , and θ, is then given by:
Sσ (ω | Hs , Tz , θ) = |Hσ (ω | θ) |2 · Sη (ωe , θ | Hs , Tz ) (5.7)
The moments for the stress spectrum for the ith sea state and jth heading:
mkij = ω k Sσ (ω | Hs , Tz , θ) dω (5.8)
X
where mkij is the kth moment. The short term stress range distribution for the ith sea
state and jth heading is defined as:
" #
σ2
F∆σij (σ) = 1 − exp − (5.9)
8m0ij
The corresponding zero crossing frequency is given by:
1
s
m2ij
vij = (5.10)
2π m0ij
It is here assumed that the variation of stresses in a short-term sea state is a narrow
banded, random process. As fatigue and damage tolerance represent long-term events, it
is necessary to extrapolate the data to reflect long-term behaviour and response. The long-
term extreme response can be found by summation of the short-term responses over a given
period of time. This results in a probability density function for long-term distribution
defined by:
n∗ pi pj f∗ (σa )
P P
i j
f (σa ) = P P (5.11)
i j n∗ pi pj
σa = Stress amplitude
f∗ = Probability density function for short-term response
n∗ = Average number of responses per unit time, short-term response
pi = Weighting factor for the ith sea state
pj = Weighting factor for the jth heading
The total stress range distribution is finally given by:
F∆σ = rij F∆σij (σ)pij (5.12)

XX
i j
pij = The probability of occurence of a given sea state and head-

ing.
vij
rij = v0 The weighted function that consists of the crossing rate in
a given sea state and the average crossing rate.
v0 = pij vij The average crossing rate.
Development of the loading history 51
As no evaluation of the load spectra, as defined above, has been performed, an approxi-
mation of the long-term distribution of stress maxima is needed. [3] and [13] suggested to
model the long-term distribution as a two-parameter Weibull distribution:
1
log n

ξ
S = S0 1 − (5.13)
log n0
Where:
S0 = Maximum stress range in the load history.

n = Number of load cycles exceeding S.
n0 = Total number of load cycles in the load history.
ξ = Weibull shape parameter.
Figure 5.6: Exceedances of stress ranges represented by the Weibull distribution with
different shape parameters
For determination of the cumulative fatigue damage, the Miner sum can be calculated by
two different methods, depending on the SN-curve used. If the basic SN-curve is used,
ā = Nref · Sref
m , the cumulative fatigue damage can be calculated as [13]:
X ni n0 m

D = = · (2λ)m · Γ 1 + (5.14)
i
Ni ā ξ
Where λ is the Weibull scale parameter and Γ is the complete Gamma function. However,
if a SN-curve different from the basic SN-curve is to be applied, the Miner summation can
be performed by subdividing the load spectra into stress range blocks, as shown examplified
in figure 5.7.
Figure 5.7: Example of sub-division of load spectra to stress range blocks
Obviously, the choice of Weibull parameters is very important for the evaluation of fatigue
strength. It can be seen from equation 5.14 that the cumulative fatigue damage is sensitive
to changes of the Weibull distribution parameters. [12] estimated the shape parameter
ξ = 0.8. In [13], Heggelund performed a case study of a 120 meter catamaran. At the
location equivalent to the present case, he found the parameters to have different values
depending on the response levels considered. For the total range of response levels, he
found ξ = 0.90 and θ = 1.37, while ξ = 0.81 and θ = 1.19 were found for the response
levels most relevant for fatigue damage. The cumulative fatigue damage was found to be
13.5% higher for the data set with the most relevant response levels compared to the full
data set. For the present case, the following values will be used:
Shape parameter, ξ = 0.81
Worst case, ξW C = 0.90
Another important consideration is the total number of load cycles. For 20 years service,
which is the minimum period (see section 5.1), the total number of load cycles is often
set to n0 = 1 · 108 . In [13], it was assumed that a large passenger ferry would have 6700
operating hours per year, e.g. it would be in operation 76.5% of the time in its lifetime.
Subsequently, he found that n0 was in the range 1.6 − 1.8 · 108 . For the present case, this
appears overly conservative. The total number of load is set to n0 = 1 · 108 , which is still
considered a conservative estimate for the present case. The long term stress distribution
used for the following analyses is presented in figure 5.8.
Finite element models 53
Figure 5.8: Load spectras
Fatigue desing codes for aluminum structures using stress range and SN-curves can be
based on a nominal stress range approach or a hot spot stress range approach. Finite
element analysis will be used to determine the maximum stress range. For the present
case, this is assumed to occur when the vessel is subjected to symmetric bottom slamming
(see section 2.2.1). The relevant stresses is determined by applying the loads, stresses and
displacements found from the web frame analysis discussed in section 4.4 to a local model
as shown in figure 5.2b.
5.3 Finite element models
For the development of a local model, the boundary conditions need to be applied in a
manner such that the local model is compatible with the global model. For the bottom
longitudinal stiffener-transverse web frame connection, as illustrated in figure 5.2, the
web frame slamming analysis indicated symmetry about the vertical neutral axis. This is
shown in figure 5.9, displaying a contour plot of the equivalent stress at the given location.
Figure 5.9: Equivalent stress for bottom stiffener-transverse web frame connection, slam-
ming analysis.
In the real structure, the area of interest is slightly curved about the longitudinal neutral
axis. For simplicity, this curvature is neglected in the modelling. It is assumed that this
simplification has little impact on the local structural behaviour of interest at the potential
crack initiation sites, illustrated in figure 5.3.
The transverse web frame experiences in-plane bending due to the slamming load. From
the slamming analysis results, it is found that the bending stress for the transverse web
frame varies linearly over the height of the web frame. At the cross sectional neutral axis,
the bending stress is close to zero, indicating that the bending stress is almost exclusively
due to the slamming induced transverse bending moment and that axial forces in the web
frame at the given location are close to zero.
From the slamming analysis, bending stresses of -55,4 MPa (compression) and 38.0 Mpa
(tension) are found at the bottom and upper flange of the transverse web frame at the
given location. By Mbend = σbend Z, the bending moment is found to have a value of
Mbend = 26.2 kNm. In the modelling, this bending moment was added by applying six
sets of concentrated forces, three above the cross sectional neutral axis and three below,
as shown in figure 5.10. The concentrated forces were applied at locations approximately
1/3, 2/3, and 3/3 of the distance from the neutral axis to the top/bottom flange. Also
the magnitudes of the forces varied linearly with the height of the web frame, from -60.1
kN at the bottom flange to 40.3 kN at the top flange.
Figure 5.10: Concentrated forces applied to web frame to replicate bending moment.
The same slamming load as used for the transverse web frame slamming analysis is used
for the local model, i.e. an evenly distributed pressure acting on the bottom plating,
psl = 82.3 kN/m2 . This vertical load is mainly carried by the transverse web frame, so
springs acting in vertical direction were added to all nodes at the cross sectional faces of
the transverse web frame, shown as purple objects in figure 5.11.
Finite element models 55
Figure 5.11: Springs acting in vertical direction to carry slamming load

.
This modelling induces an additional bending moment to the transverse web frame. As-
suming that the web frame (in this case) can be considered a simply supported beam
and that the slamming load transferred via the longitudinal stiffener to the web frame
can be idealized as a concentrated force, the induced bending moment may be taken as
F L
M = slam 4 f rame = 0.96 kNm, merely 4% of the applied bending moment.
Although horizontal forces acting on the models, i.e. the concentrated forces acting on
the web frame, should be balanced, any imbalance will be absorbed by the symmetry
boundary conditions at the edges of the bottom plate. The contact in the area between
the welds has been modelled as a frictionless surface-to-surface contact, which is assumed
to resemble the actual contact in a satisfying manner.
In [8], DNV suggests two methods of modelling. The simplest way of modelling is offered by
shell elements arranged in the mid-plane of the structural components. 8-noded elements
are recommended in the case of steep stress gradients. A mesh size from txt up to 2tx2t
may be used. For the hot spot region, txt is preferred. The second method, more suitable
for complex cases, is offered by solid elements, preferably isoparametric 20-node elements.
Again, element length corresponding to plate thickness is recommended and no larger than
2 t. Furthermore, the fillet weld should be modelled to achieve proper local stiffness and
geometry.
Four different finite element models has been created in ABAQUS CAE, version 6.10.2.
The all have the same main characteristics and material properties, and loads and bound-
ary conditions are applied as stated above. What separates them is the elements used
in the modelling, the refinement of the mesh, and the weld modelling. They are labeled
models 1-4.
Model 1 is constructed by 8-node shell elements, reduced integration (S8R), in accordance
to [8]. Element sizes vary between 5.5mm x 5.5mm to 6.0mm x 6.0mm. The relevant
plate thicknesses are 10.0mm for the stiffener flange and 8.0mm for the web frame bottom
flange. This means that the mesh is slightly more refined than what is recommended. No
weld is modelled for this model, and since the shell elements are arranged in the mid-plane
of the structural components, a tie-constraint was needed to join the stiffener and the web
frame. For this constraint, it is necessary to establish a master-slave relationship for the
nodes. Generally, the nodes for the loaded structure are assigned as master nodes. For the
present case, where both structural elements are loaded, it was not obvious which set of
nodes that should be assigned as master nodes. However, analyses proved that the results
were practically unaffected by the choice of master nodes.
(a) Outline of model. (b) Tie-constraint at weld, relevant nodes high-

lighted
Figure 5.12: Model 1
Model 2 is contructed by 20-node solid elements with reduced integration, termed C3D20R
in ABAQUS. The element length vary between 5.0mm to 8.0 mm, where two elements is
modelled over the stiffener flange thickness and one element over the web frame bottom
flange thickness. For this model, no weld has been modelled, the stiffener and web frame
are thus connected through merging of nodes at the weld root, as illustrated in figure
5.13b. Model 2 is also in accordance with the recommendation given in [8], except for the
fillet weld, that has not been modelled.
(a) Outline of model. (b) Merged nodes at weld root
Model 3 is equal to Model 2, but this model has an idealised weld included. The weld
covers the full thickness of the web frame bottom flange thickness, i.e. a leg length of
8mm, while the horizontal leg length is 5mm. The element type used for the weld is ten
node tetrahedral elements, termed C3D10. In general, tetrahedral elements are inferior to
hexahedral elements in terms of accuracy for a finite element analysis, so the stresses at
the interface for these elements may be inaccurate. However, the load transfer should be
correct. No weld fusion has been modelled, nodes at the legs has simply been merged, as
illustrated in figure 5.14b.
Calculation of fatigue damage 57
(a) Weld fusion (b) Weld idealisation. Merged nodes at stiffener

flange and web frame bottom flange highlighted.
Model 4 is constructed exclusively by 4-node tetrahedral elements (C3D4). The weld is

idealised as a triangular prism where both leg lengths are 8mm. As for Model 3, no weld
fusion is included in the model, and the interaction between the structural elements is
equal for Model 3 and 4. However, Model 4 has a fine mesh at the hot spot region, with a
characteristic element length of approximately 2mm. For the rest of the model, an element
length of 8mm is applied.
(a) Model outline (b) Mesh at hot spot region
5.4 Calculation of fatigue damage
For determination of the cumulative fatigue damage, the nominal stress range approach
will be applied. For this approach, the design stress is the local nominal stress range, which
does not include stress raisers due to local weld geometry or local geometrical changes.
This implies that effects from structural discontinuity and local notch effects are implicitly
included in the fatigue strength curves, i.e. the SN-curves. The Eurocode 9 fatigue design
code is based on the nominal stress range approach, and will be used as basis for the
following calculations.
Three constructional details, shown in figure 5.4, has been identified as the equivalent for
the potential crack initiation sites, illustrated in figure 5.3. These contructional details are
categorised in detail category 23-3.4, 20-3.4, and 18-3.4 for HS A, B, and C, respectively.
The detail categories are labeled according to their characteristic stress range, Sref , which
is the stress range at Nref = 2 · 106 cycles. The corresponding SN-curves are shown in
figure 5.16.
Figure 5.16: SN-curves to be used for the cumulative fatige damage calculation.
The design stress range to be used with the SN-curves above is the nominal stress range
at the initiation site. It is stated in [4] that where the axial stress distribution is linear
across the member section about both axes, the stresses at the initiation point may be
used directly. This is not the case at the hot spot regions. The nominal stress needs to
be estimated by other means, as will be presented for HS A, B, and C in the following
sections.
5.4.1 HS A
Figure 5.17 shows the longitudinal bending stress (the stress component normal to the
weld) at the surface along the stiffener flange centerline for the four local models. It is
seen that the stresses are within close range of eachother from the end upto about 30mm
(3 times stiffener flange thickness) from the weld root. Close to the weld root, the stresses
differ significantly. This is due to the presence of weld and structural discontinuity and
how these structural details are modelled in the different models.
Figure 5.17: Bending stress normal to weld along horizontal line at center of stiffener
flange
In the area where the stress raisers at the weld do not contribute, it is seen that the bending
stresses increase close to linearly. It is thus assumed that the nominal stress at the HS A
may be found be linear extrapolation. At a distance 80mm from the weld root, a mean
stress of -12.5 MPa is found for the four models. Likewise, a mean stress at a distance
30mm from the weld root is found to have a value of -22.5 MPa, so linear extrapolation
yields a nominal stress of -28.5 MPa at the weld root. The stress at the weld root is used
instead of the stress at the weld toe to ensure conservativeness as the stress estimate may
be inaccurate.
It is assumed that the extreme stress range goes from zero to the extreme nominal stress,
so the maximum stress range in the load history is thus S0 = 28.5 MPa. With the Weibull
parameters as given in section 5.2, the long term cumulative load distribution is as shown
in figure 5.18.
Figure 5.18: Cumulative long term distribution of stresses and corresponding SN-curve
The relevant SN-curve for HS A, 23-3.4, has a cut-off limit at 10.1 MPa, meaning that
stresses below this level does not contribute to fatigue damage. It is seen that approx-
imately 4 · 104 load cycles will contribute to fatigue damage for the estimated Weibull
shape parameter, while the worst case Weibull shape parameter will have approximately
8 · 104 load cycles contributing to fatigue damage.
The Miner sum is found by subdividing the cumulative long term stress distribution into
a stress range blocks with load cycle intervals ranging from 102 to 103 for stress ranges
exceeding the cut-off limit. The number of load cycles in the stress range block is divided by
the corresponding number of load cycles to failure. By summation of all these contributions
to fatigue damage, the cumulative damage for HS A is:
D = 4.11 · 10−3
DW C = 5.22 · 10−3
Given a fatigue criterion of D = 1.0, it is reasonable to assume that HS A is not crit-
ical in terms of fatigue. The worst case Weibull shape parameter yields an increase of
approximately 27% for the cumulative fatigue damage.
5.4.2 HS B
To determine the nominal stress at HS B, the same method as in section 5.4.1 is applicable.
The longitudinal bending stress at the edge of the stiffener flange is shown in figure 5.19
Figure 5.19: Bending stress normal to weld along horizontal line at edge of stiffener flange
The stresses at the edge of the stiffener flange are similar to those at the center of the
flange, discussed in section 5.4.1. Accordingly, the bending stress at the weld root is
determined by linear extrapolation, yielding a extreme nominal stress of 29.3 MPa. In
this case, as for the case for HS A, the maximum stress range in the load history is thus
set to 29.3 MPa. The long term cumulative load distribution is as shown in figure 5.20.
The corresponding SN-curve is termed 20-3.4, with a cut-off limit of 8.8 MPa. For the
given cumulative long term stress distributions, approximately 1 − 2 · 105 load cycles
will contribute to fatigue damage. By Miner summation, as performed for HS A, the

cumulative fatigue damage is:
D = 7.92 · 10−3
DW C = 1.04 · 10−2
The results indicate that HS B is not critical. For the worst case Weibull shape parameter,
the cumulative fatigue damage is increased by 32% compared to the estimated Weibull
shape parameter. It is seen that the cumulative fatigue damage for HS B is almost doubled
compared to that for HS A.
5.4.3 HS C
The transverse bending stresses at the web frame bottom flange edge is found to vary
significantly for the different models, and the results from the analyses are inconsistent,
as seen in figure 5.21. A possible explanation is the modelling of the bending moment by
concentrated forces distributed over the cross section of the web frame. Very high local
stresses are found at the locations where the concentrated forces act, and these propagate
to the surrounding area, thus causing disturbances in the stress distribution.
Figure 5.21: Bending stress parallell to weld along horizontal line at edge of web frame
bottom flange flange
The transverse bending stress from the slamming analysis has a value of -55.4 MPa at the
same location. This is seen to correspond to the stresses found for models 2 and 3 about
16mm from the weld root, equal to two times the plate thickness for the web frame bottom
flange. The extreme nominal stress is taken as the stress from the slamming analysis, and
the corresponding maximum stress range in the load history is thus taken as S0 = 55.4
MPa. The cumulative long term distributions of stresses are subsequently as shown in
figure 5.22.
The corresponding SN-curve 18-3.4 for HS C has a maximum stress range of 43.4 MPa with
the correponding number of cycles to failure of N = 1·105 . It is assumed that the SN-curve
may be extrapolated upto a stress range of 85.5 MPa, where the corresponding number of
cycles to failure is N = 1 · 104 . By subdividing the cumulative stress distribution into a
finite number of stress blocks and performing Miner summation, the following cumulative
fatigue damages are found:
D = 3.52 · 10−1
DW C = 5.66 · 10−1
For a fatigue criterion of D = 1.0, the results indicate that excessive fatigue damage will
not occur at HS C. However, the cumulative fatigue damage for HS C is far higher than for
HS A and HS B, indicating that HS C is the most critical hot spot region. Although a series
of assumptions and simplifications has been made in the calculations, it is reasonable to
believe that they are conservative. Consequently, a critical fatigue failure at HS C appears
unlikely.
Chapter 6
Conclusions
A modified version of the JumboCat 60 with floating transverse web frames has been
evaluated from a strength perpective with focus on hull beam strength, local structural
response due to extreme loading, and fatigue of a structural detail in the bottom structure
of the vessel.
Using the DNV HSLC rules, it is found that the modified JC60 fulfils the minimum
requirements related to scantlings. Furthermore, the design loads are shown to be equal
for the floating frame version and the original version of the JC60.
The structural response for two global and one local load condition has been evaluated
by finite element analysis for the floating frame version and the traditional version of the
vessel. For the local slamming load condition, no stresses exceeding the allowable stress
levels defined by DNV HSLC rules are found. By comparing the floating frame model
and the traditional model, it is seen that the floating frame model is slightly stiffer for
this particular load condition. It was also shown that the high stres concentrations found
by Jon Englund and Toralf Ervik, especially at the longitudinal stiffener webs due to
out-of-plane bending, was significantly reduced by strengthening of the transverse floating
frame.
The global torsional/pitch connecting moment load condition also showed stress levels
within the allowable. It was indicated by the results that the traditional model had a
higher hull beam torsional stiffness than the floating frame model. The transverse bend-
ing moment load condition indicated that the original model hull beam was stronger in
transverse direction than the floating frame model. For this analysis, local stress concen-
trations were found to have stress levels exceeding the allowable. As the stresses for the
floating frame model were found to exceed those of the traditional model, it is suggested
that the plate thickness of the bottom plating at a specified area is increased for the float-
ing frame version. Also, the results from the transverse bending analysis indicated high
shear stresses were found for the longitudinal stiffeners at the transverse bulkheads for
the floating frame model. In order to reduce these stresses, it is suggested that at least
the stiffeners in the critical area have their dimensions increased in order to increase their
shear area.
The fatigue strength of the longitudinal stiffener/transverse web frame connection in the
bottom structure of the floating frame version was investigated. The cumulative fatigue
damage for three different hot spot areas was calculated by Miner summation. A Weibull
distributed cumulative long term distribution of stresses, using a shape parameter of ξ =
65
66 Conclusions
0.81 and a total number of load cycles of 1 · 108 for 20 years of service, was combined
with SN-curves given in Eurocode 9 in the Miner summation. The design stress was the
nominal stress at the crack initiation site. This was estimated using the results from
finite element analysis using four different local models of the structural detail, and the
local slamming finite element analysis. For the hot spot areas at the stiffener flange, the
cumulative fatigue damage was found to be in the region D = 4.1 · 10−3 − 1.0 · 10−2 , while
at the transverse web frame bottom flange, the cumulative fatigue damage was estimated
to be in the range D = 0.35 − 0.57. The fatigue criterion was assumed to be D = 1.0,
meaning that the structural detail evaluated is considered non-critical in terms of fatigue.
6.1 Recommendations for further work
As this thesis only studies the structural response for three load conditions, it is sug-
gested that the remaining load conditions, as defined in DNV Classification Notes 30.8,
are evaluated for the floating frame version of the JumboCat 60. Linear and non-linear
local analyses should be performed as seen appropriate, for example at the critical area
termed region 1 in the tranverse bending moment analysis.
A calculation of the hydrodynamic loads on the vessel would be useful to estimate the
actual global and local wave loads. These loads could in turn be used to determine the
global and local structural response more realistically than by the idealised loads applied
in the analyses presented in this thesis. A complete hydrodynamic analysis could also serve
as basis for calculation of the long term distribution of stresses used in fatigue calculations.
For a complete fatigue assessment, several other structural details other than the one
presented in this thesis need to be evaluated. Lab tests investigating the fatigue strength
of relevant structural details would be useful to develop SN-curves suited for the floating
frame design.
The increase of hull weight due to the use of floating frames should be calculated in order
to determine the life cycle costs for the floating frame desing compared to the original
design. It is in this case relevant to investigate if the reduced building cost compensates
for a potential increase in fuel consumption due to increased hull weight.
Bibliography
[1] A. Aalberg. Redningsselskapets skøyte ’knut johan’ - konstruert med stivere liggende
utenpå spantprofiler. Technical report, Norwegian University of Science and Technol-
ogy, NTNU, 2006.
[2] J. Amdahl. Vurdering av lokalkrav til plater og stivere i ulike standarder for småbåter
i aluminium. Technical report, Norwegian University of Science and Technology,
NTNU, 2006.
[3] S. Berge. Fatigue design of welded structures, September 2006.
[4] CEN. Eurocode 9: Design of aluminum structures - Part 1-3: Structures susceptible
to fatigue. European Commitee for Standardization, May 2007.
[5] DNV. Classification Notes No. 30.8, Strength Analysis of Hull Structures in High
Speed and Light Craft. Det norske Veritas, August 1996.
[6] DNV. Rules for Classification of High Speed, Light Craft and Naval Surface Craft,
Part 3, Chapter 3, Hull Structural Design, Aluminum Alloy. Det norske Veritas, July
2000.
Part 3, Chapter 1, Design Principles, Design Loads. Det norske Veritas, January
2005.
[8] DNV. Recommended Practice DNV-RP-C203, Fatigue Design of Offshore Steel Struc-
tures. Det norske Veritas, April 2010.
Part 3, Chapter 9, Direct Calculation Methods. Det norske Veritas, January 2011.
[10] J. Englund. Finit element-analys av knut johan. Unpublished, 2009.
[11] J. Englund. Structural strength of work boats and high speed crafts with floating
frames. Master’s thesis, Kungliga Tekniska högskolan, KTH, 2009.
[12] T. Ervik. Structural strength of work boats and high speed crafts with pre-fabricated,
floating panels in aluminium. Master’s thesis, Norwegian University of Science and
Technology, NTNU, 2010.
[13] B.W. Tveiten Heggelund, S.E. and T. Moan. Fatigue analysis of high speed aluminum
catamarans. In The Third International Forum on Aluminum Ships, 1998.
[14] S.E. Heggelund and T. Moan. Analysis of global load effects in catamarans. Journal
of Ship Research, 2001.
67
68 Bibliography
[15] T. Moan Heggelund, S.E. and S. Oma. Global structural analysis of large catamarans.
In Proc. Fifth International Conference on Fast Sea Transportation, FAST’99, 1999.
[16] P.D. Herrington and R.G. Latorre. Development of an aluminum hull panel for high-
speed craft. Marine Structures, 1998.
[17] P.D. Herrington and R.G. Latorre. Development of a high-speed aluminum ferry with
floating frames. Technical report, University of New Orleans, 2000.
[18] O.D. Økland. Analysis of transverse section of the 60m jumbo cat. Technical report,
Marintek, 1999.
[19] B. Rampolla Kramer, R.K. and A. Magnusson. Fatigue of aluminum structural weld-
ments. Technical report, Ship Structure Committee, May 2000.
[20] S. Missori and A. Sili. Mechanical behaviour of 6082-t6 aluminum alloy welds. Met-
allurgical Science and Technology, 1999.
[21] L.J. Mong. Konstruksjonsstyrke til bruksbåter og hurtiggående båter med prefab-
rikkerte, utenpåliggende paneler i aluminium. Technical report, Norwegian University
of Science and Technology, NTNU, 2010.
[22] USS. Steel vs. aluminum: The basics. http://xnet3.uss.com/auto/steelvsal/
basicfacts.htm, January 2011.

Structural Strength of Work Boats and High Speed Crafts With Pre-Fabricated, Floating Panels in Aluminum

Uploaded by

Document Informationclick to expand document information

Copyright:

Available Formats

Structural Strength of Work Boats and High Speed Crafts With Pre-Fabricated, Floating Panels in Aluminum

Uploaded by

Document Information

Original Description:

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Structural Strength of Work Boats and High Speed Crafts With Pre-Fabricated, Floating Panels in Aluminum

Uploaded by

Copyright:

Available Formats

Norwegian University of Science and

Structural strength of work boats and

Trondheim, June 2011

2 Design loads and strength requirements 11

3 Finite Element Modelling 23

4 Results from finite element analyses 31

1.1 The Floating Frame Principle . . . . . . . . . . . . . . . . . . . . . . . . . . 5

2.1 Still water transverse bending moment . . . . . . . . . . . . . . . . . . . . . 14

3.1 Web frames used in models . . . . . . . . . . . . . . . . . . . . . . . . . . . 24

4.1 Bending stress at frame #49 . . . . . . . . . . . . . . . . . . . . . . . . . . 33

2.1 Allowable bending stresses for plating and stiffeners . . . . . . . . . . . . . 16

4.1 Allowable stresses, global analyses . . . . . . . . . . . . . . . . . . . . . . . 32

acg Design vertical acceleration at LCG

MASTER THESIS VÅREN 2011

vere og skott/tverrammer. Beregningsanvisning for dette er ikke beskrevet i de vanlige

Det er derfor ansett nødvendig å utvikle modifiserte dimensjoneringsregler for prefab-

Styrkereduksjonen på grunn av opplagerkraft (konsentrert kraft) i kryssingspunkt mellom

(a) Floating frame (b) Traditional frame

Figure 1.3: Distribution of external lateral pressure in bottom structure

Figure 1.5: The JumboCat 60

It has the following main dimensions:

1.4 Previous work

1.4.1 JumboCat 60 with floating frames

1.4.2 The Floating Frame Principle

Design loads and strength

2.1 Design codes in general

• Nordic Boat Standard,

• DNV Tentative Rules for Certification and Classification of Boats,

2.2 Design Loads

2.2.1 Slamming Loads

Sec. 2, C203 Pitching slamming

2.2.2 Sea Pressure

2.2.3 Deck Loads

p = ρH (g0 + 0.5av ) (2.4)

2.2.4 Transverse Bending Moment and Split Force

MS = MS0 + Fy (z − 0.5T ) (2.6)

Figure 2.1: Still water transverse bending moment

2.2.5 Pitch connecting and twin hull torsional moment

The twin hull pitch connecting moment is given in section 3, B301 as

2.3 Strength requirements

2.3.1 Hull girder strength

allowable bending stress

2.3.2 Plating and Stiffeners

Allowable bending stresses

Table 2.1: Allowable bending stresses for plating and stiffeners

The thickness of the bottom plating is to be at least (B302):

The required shear area of longitudinals, also given in C201

2.3.3 Web frames and girders

Table 2.2: Allowable stresses for web frames and girders

2.4 Direct calculation methods

2.4.1 Symmetric bottom slamming

Figure 2.3: Local load condition 2, symmetric bottom slamming [5]

Design loads Plating Stiffeners and

Table 2.3: Allowable stresses for web frame analysis

Allowable shear stress: τ = 90f1

2.4.2 Transverse split force

(a) Split load condition